Water in Biological and Chemical Process

WATER IN BIOLOGICAL AND CHEMICAL PROCESSES
Building up from microscopic basics to observed complex functions, this insightful

monograph explains and describes how the unique molecular properties of water
give rise to its structural and dynamical behavior, which in turn translates into its
role in biological and chemical processes.
The discussion of the biological functions of water details not only the stabilizing
effect of water in proteins and DNA, but also the direct role that water molecules
themselves play in biochemical processes, such as enzyme kinetics, protein synth-
esis, and drugDNA interaction. The overview of the behavior of water in chemical
systems discusses hydrophilic, hydrophobic, and amphiphilic effects, as well as the
interactions of water with micelles, reverse micelles, microemulsions, and carbon
nanotubes.
Supported by extensive experimental and computer simulation data, highlighting
many of the recent advances in the study of water in complex systems, this is an
ideal resource for anyone studying water at the molecular level.
biman bagchi is a Professor at the Indian Institute of Science, Bangalore. He is a
Fellow of the Indian National Science Academy, the Indian Academy of Sciences,
The National Academy of Sciences, India, and TWAS, The Academy of Sciences
for the Developing World, Italy.
Cambridge Molecular Science
As we move further into the twenty-rst century, chemistry is positioning itself as
the central science. Its subject matter, atoms and the bonds between them, is now
central to so many of the life sciences on the one hand, as biological chemistry
brings the subject to the atomic level, and to condensed matter and molecular
physics on the other. Developments in quantum chemistry and in statistical
mechanics have also created a fruitful overlap with mathematics and theoretical
physics. Consequently, boundaries between chemistry and other traditional sciences
are fading and the term Molecular Science now describes this vibrant area of
research.
Molecular science has made giant strides in recent years. Bolstered by both
instrumental and theoretical developments, it covers the temporal scale down to
femtoseconds, a timescale sufcient to dene atomic dynamics with precision, and
the spatial scale down to a small fraction of an angstrom. This has led to a very
sophisticated level of understanding of the properties of small molecule systems, but
there has also been a remarkable series of developments in more complex systems.
These include protein engineering, surfaces and interfaces, polymers, colloids, and
biophysical chemistry. This series provides a vehicle for the publication of advanced
textbooks and monographs introducing and reviewing these exciting developments.
Series editors
Professor Richard Saykally
University of California, Berkeley
Professor Ahmed Zewail
California Institute of Technology
Professor David King
University of Cambridge
WATER IN BIOLOGICAL
AND CHEMICAL PROCESSES
From Structure and Dynamics to Function
BIMAN BAGCHI
Indian Institute of Science, Bangalore
University Printing House, Cambridge CB2 8BS, United Kingdom
Published in the United States of America by Cambridge University Press, New York
Cambridge University Press is part of the University of Cambridge.
It furthers the Universitys mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.
www.cambridge.org
Information on this title: www.cambridge.org/9781107037298
Biman Bagchi 2013
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2013
Printed in the United Kingdom by TJ International Ltd. Padstow Cornwall
A catalog record for this publication is available from the British Library
Library of Congress Cataloging in Publication data
Bagchi, B. (Biman)
Water in biological and chemical processes : from structure and dynamics to
function / Biman Bagchi, Indian Institute of Science, Bangalore.
pages cm. (Cambridge molecular science)
Includes bibliographical references.
ISBN 978-1-107-03729-8
1. Water in the body. 2. Water chemistry. I. Title
QP535.H1B34 2013
612
/
.01522dc23
2013013114
ISBN 978-1-107-03729-8 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
To my mother and my father, Abha and Binay K. Bagchi. They taught me, from an
early age, to love poetry and science, opening doors to the wonders of Nature.
Understanding the role of water as the ubiquitous solvent for the chemical biology
and throughout molecular science remains one of the most active areas of current
scientic research. The puzzling issues that arise throughout this eld require a
unied understanding of structure, dynamics and thermodynamics. This book
provides a valuable resource in relating microscopic properties to complex phenom-
enology, connecting diverse topics of contemporary interest.
David J. Wales, University of Cambridge
This book by Biman Bagchi covers an extremely broad range of topics on water,
written with an eye to relating theory and experiment and by someone who has
insight into both. Its use of recent references in the eld is a helpful attribute. The
author emphasizes that our understanding is not a closed subject and so there will be
further room for developments, debate on interpretation, and discussions. For
teachers of topics in equilibrium and nonequlibrium statistical mechanics there is
also, I believe, much useful material on interesting applications.
Rudy A. Marcus, California Institute of Technology
Water continues to both fascinate and confound those who study its properties and
its vital roles in lifes structures and dynamical processes. In this unique book Biman
Bagchi has brought together an extraordinary range of experimental data and the
results of both theory and simulation studies at a level generally accessible to readers
with a background in chemistry at the rst year university level. He illuminates how
the remarkable properties of water are key to a multitude of chemical and biological
processes and in doing so provides both insight and the springboard for new
investigations of this endlessly fascinating liquid.
Graham R. Fleming, University of California, Berkeley
Contents
Preface page xv
Acknowledgements xviii
Part I Bulk water 1
1. Uniqueness of water 3
1.1 Introduction 3
1.2 Molecular structure 4
1.3 Six unique features 7
1.4 Modeling of water 9
1.5 Conclusion 10
2. Anomalies of water 13
2.1 Anomalous properties 13
2.1.1 Density maximum 13
2.1.2 Isobaric specic heat (C
P
) 15
2.1.3 Isothermal compressibility (
T
) 15
2.1.4 Coefcient of thermal expansion (
P
) 16
2.1.5 Dynamic anomalies present at low temperature 17
2.2 Translational and orientational order 19
2.3 Temperaturedensity range of water anomalies 21
2.4 Conclusion 22
Appendix 2.A Microscopic expressions of specic heat, isothermal
compressibility, and coefcient of thermal expansion 23
Appendix 2.B Quantication of spatial order in water 24
3. Dynamics of water: molecular motions and hydrogen-bond-breaking
kinetics 27
3.1 Introduction 27
3.2 Timescales of translational and rotational motion 28
vii
3.3 Jump reorientation motion in water 30
3.4 Effects of temperature on water motion 33
3.5 Translational diffusion 35
3.6 Hydrogen-bond lifetime dynamics 36
3.7 Vibrational dynamics of the OH bond 39
3.8 Dielectric relaxation 40
3.9 Solvation dynamics 42
3.10 Ionic conductivity of rigid ions in water 45
3.11 Electron transfer reactions in water 47
3.12 Motion becomes collective at low temperature 49
3.13 Conclusion 50
Appendix 3.A Rotational time correlation functions 51
Appendix 3.B Quantication of hydrogen-bond
lifetime dynamics 58
4. Inherent structures of liquid water 61
4.1 Introduction 61
4.2 Transition between inherent structures of water 66
4.3 Connected water cluster moves during transition 67
4.4 HB network restructuring 67
4.5 Coordination number uctuation in inherent structure and
corresponding dynamics in parent liquid 68
4.6 Low-energy excitations in liquid water 69
4.7 Conclusion 69
5. The pH of water 71
5.1 Introduction 71
5.2 Temperature and pressure dependence of pH 73
5.3 Mechanism of autoionization 74
5.4 pH of blood 75
5.5 Food and blood pH 76
5.6 pH of seawater 77
5.7 Conclusion 77
Part II Water in biology 79
6. Biological water 81
6.1 Introduction 81
6.2 Relaxation measurements 83
6.3 Unique characteristics of biological water 83
6.4 Phenomenological models and simple theories 84
6.5 Proteinglass transition and hydration-layer dynamics 88
viii Contents
6.6 Protein aggregation and biological water 90
6.7 Conclusion 90
Appendix 6.A The dynamic exchange model 91
7. An essential chemical for life processes: water in biological functions 97
7.1 Introduction 97
7.2 Role of water in enzyme kinetics 99
7.3 Role of water in drugDNA intercalation 101
7.4 Role of water in the biological function of RNA 105
7.5 Water-mediated molecular recognition 107
7.6 Protein folding and protein association: role of biological water 109
7.7 Role of water in beta-amyloid aggregation in Alzheimer disease 109
7.7.1 Role of water in the early stages of oligomer formation 110
7.7.2 Role of water in the late stages of bril growth 111
7.8 Role of water in photosynthesis 112
7.9 Conclusion 114
8. Hydration of proteins 117
8.1 Introduction 117
8.2 What is the thickness of the hydration shell? 118
8.3 How structured is the water in the hydration shell of a protein? 121
8.4 Orientational arrangement of water molecules at the surface 123
8.5 Dynamics of the protein hydration shell: experimental studies 124
8.5.1 Dielectric spectrum 124
8.5.2 Nuclear magnetic resonance studies 126
8.5.3 Quasi-elastic neutron-scattering experiments 127
8.5.4 Vibrational spectroscopy 128
8.5.5 Solvation dynamics 129
8.6 Conclusion 131
Appendix 8.A Orientation of water molecules in the hydration layer 132
9. Understanding the protein hydration layer: lessons from
computer simulations 135
9.2 Molecular motion in the hydration layer 136
9.4 Computer simulation of solvation dynamics 142
9.6 Explanation of anomalous dynamics in the hydration layer 144
9.7 Proteinglass transition at 200 K: role of water dynamics 144
Contents ix
9.8 Free-energy barrier for escape of water molecules from
protein hydration layer 146
9.9 Conclusion 146
10. Water in and around DNA and RNA 151
10.1 Introduction: the unique role of water in stabilizing
DNA and RNA 151
10.2 Hydration of different constituents 152
10.3 Groove structure and water dynamics 153
10.4 Translational and rotational dynamics of water molecules in the
grooves 153
10.6 Entropy of groove water and dynamics 156
10.7 Correlation between diffusion and entropy: AdamGibbs
relation 157
10.8 Sequence dependence of DNA hydration: spine of hydration
in AT minor groove 159
10.9 Effects of nanoconnement and surface-specic interactions 161
10.10 Water around RNA 161
10.10.1 Structure of water around RNA 162
10.10.2 Dynamics of water around RNA 162
10.11 Conclusion 162
Appendix 10.A Hydrogen-bonding pattern around DNA 163
11. ProteinDNA interaction: the role of water as a facilitator 167
11.2 Structural analysis of proteinDNA complex: classication of
hydration water 168
11.3 Dynamics of water around a proteinDNA complex 169
11.4 Role of water in thermodynamics of proteinDNA interactions 170
11.5 Protein diffusion along DNA 174
11.6 Conclusion 174
12. Water surrounding lipid bilayers: its role as a lubricant 177
12.2 Hydration of different constituents: phospholipids and buried
proteins 179
12.3 Rugged energy landscape for water motion 179
12.4 Translational and rotational dynamics of water 180
12.6 Transport of small molecules across the bilayer 182
x Contents
12.7 Transport of large molecules across the bilayer 184
12.8 Electrostatic potential across the membrane 184
12.9 Conclusion 185
13. The role of water in biochemical selection and protein synthesis 187
13.2 Role of water in kinetic proofreading 188
13.2.1 Brief analysis of the HopeeldNinio approach
to kinetic proofreading 190
13.2.2 Analysis of experimental results in the light of the
HopeldNinio formulation 190
13.2.3 Aminoacylation of tRNA during protein synthesis 192
13.2.4 tRNA selection in ribosome 194
13.2.5 DNA replication 196
13.3 Water as a lubricant of life 196
13.4 Conclusion 197
Part III Water in complex chemical systems 199
14. The hydrophilic effect 201
14.2 Water near ions 202
14.3 Water near an extended hydrophilic surface 204
14.4 Aqueous hydrophilic binary mixtures 207
14.4.1 Waterurea binary mixture 208
14.4.2 Waterguanidinium hydrochloride
binary mixture 209
14.5 Aqueous salt solutions 209
14.5.1 Ionic conductivity 209
14.5.2 Viscosity 211
14.6 Conclusion 212
15. The hydrophobic effect 215
15.2 Hydrophobic hydration 217
15.3 Temperature dependence of hydrophobicity: enthalpy
versus entropy stabilizations 219
15.4 Hydropathy scale 220
15.5 Pair hydrophobicity and potential of mean force between two
hydrophobic solutes 221
15.6 Biological applications of potential of mean force 223
15.6.1 Protein folding 224
Contents xi
15.6.2 Hydrophobic association 227
15.6.3 Pattern formation in chiral molecules 227
15.7 Hydrophobic collapse of polymers 227
15.7.1 The FloryHuggins theory 228
15.8 Molecular-level understanding of hydrophobic interaction 230
15.9 Hydrophobic force law 234
15.10 Hydrophobicity at different length scales 234
Appendix 15.A PrattChandler theory 236
15.A.1 Cavity distribution functions 237
15.A.2 Theory for AWand AA pair correlations 239
16. The amphiphilic effect: the diverse but intimate world of aqueous
binary mixtures 243
16.1 Introduction: the role of aqueous mixtures in chemistry
and biology 243
16.2 Non-ideality of amphiphilic binary mixtures 245
16.3 WaterDMSO binary mixture 245
16.4 Wateralcohol binary mixture 249
16.4.1 Aqueous methanol solution 250
16.4.2 Aqueous ethanol solution 250
16.4.3 Watertertiary butyl alcohol 250
16.5 Wateracetone binary mixture 252
16.6 Waterdioxane binary mixture 252
16.7 Liquidliquid structural transformation in aqueous
binary mixtures: a generic phenomenon for amphiphilic solutes 253
16.8 Theoretical development 254
16.9 Biological applications 256
17. Water in and around micelles, reverse micelles, and microemulsions 261
17.1 Introduction: different self-assemblies in water 261
17.2 Structure of micelles and reverse micelles 262
17.2.1 Micelles 262
17.2.2 Reverse micelles 263
17.3 Dynamics of water surrounding micelles 265
17.4 Free-energy landscape of hydrogen-bond arrangements at the
surface 266
17.5 Reverse micelles and microemulsions: dynamics of water 268
17.6 Orientational dynamics 269
17.7 Coreshell model 270
xii Contents
17.8 Distance-dependent relaxation near the core of the reverse
micelle: propagation of surface-induced frustration 273
17.9 Ising model description of the dynamics 273
18. Water in a carbon nanotube: nature abhors a vacuum 277
18.2 Type and structures of carbon nanotubes 277
18.3 Structure of water inside a carbon nanotube 278
18.4 Dynamics and transport of water 279
18.4.1 Translational motion of water inside a CNT 279
18.4.2 Rotation of water molecules within a CNT 280
18.5 Nanotubes as a ltration device 282
18.6 Conclusion 283
Part IV Advanced topics on water 285
19. The entropy of water 287
19.2 Relation between entropy and diffusion 291
19.2.1 Diffusionentropy scaling relation:
the Rosenfeld relation 291
19.2.2 The AdamGibbs relation 293
19.3 Calculation of the entropy of water 295
19.3.1 From structure 296
19.3.2 From dynamics 297
19.4 Entropy from cell theory 298
19.5 Entropy of water in conned systems (reverse micelles, carbon
nanotubes, grooves of DNA) 299
19.6 Conclusion 300
Appendix 19.A Entropy for translational degree of freedom of
an ideal gas (SackurTerode equation) 301
Appendix 19.B Entropy for vibrational degree of freedom 302
Appendix 19.C Entropy for rotational degree of freedom 303
20. The freezing of water into ice 305
20.2 Phase diagram of water and ice 306
20.3 Ice formation in micro-droplets 307
20.4 A lesson from the freezing of interacting spheres and
the difference from water 308
20.5 The freezing of water 308
Contents xiii
20.6 Nucleation of an embryo 309
20.7 The freezing of water in computer simulations 310
20.8 Mechanism of ice formation 311
20.9 Freezing inside nanotubes 314
21. Supercritical water 317
21.2 Inhomogeneous density uctuation in supercritical uids 318
21.3 Crossing the Widom line 320
21.4 Spectroscopic studies of supercritical uids 320
21.5 Conclusion 322
22. Approaches to understand water anomalies 323
22.2 Reason for density maximum 327
22.3 Reason for large isobaric specic heat of water 327
22.4 Percolation model of water 327
22.5 Hydrogen-bond network rearrangement dynamics 330
22.5.1 Energy landscape view of hydrogen-bond
rearrangement dynamics 331
22.5.2 Depolarized Raman scattering prole 333
22.6 Low-temperature anomalies 334
22.7 Conclusion 341
Epilog 345
Index 349
The color plates will be found between pages 78 and 79.
xiv Contents
Preface
This book attempts to summarize the large body of experimental and simulation
data gathered recently on the structural and dynamic aspects of water in complex
chemical and biological systems. In the process we try to present a unied view of
this emerging eld. While most discussions on water focus on its role in complex
systems (like the role of water as a polar solvent stabilizing the native state of a
protein), I thought it would be equally, if not more, appropriate to study and if
possible explain why water has so many unique properties and how it is able to
play important parts in so many diverse settings. For example, water molecules
themselves need to change and adjust to the surface. In enzyme catalysis, they
participate actively and get consumed as a chemical not act just as a good solvent
facilitating the catalysis a fact not often appreciated.
Many important aspects of water have been discovered only in the last two
decades or so. For example, we came to know about the astonishingly fast rate of
solvation of a polar solute by water only around the mid-nineteen nineties! The
detailed role of water in chemical reactions, such as in electron transfer, has also
become clearer around the same time. It is therefore not surprising that it is only now
that we have turned our attention towards understanding molecular aspects of
waters role in biology. The specic role of water in most of the biological processes
is far from well understood even today.
Studies of unique properties of water have often followed two disjointed paths.
On the one side, detailed microscopic properties of water molecules, both in the bulk
and in and around biomolecules, have been studied in vitro, such as water structure
and arrangement around proteins and DNA. These studies have often remained
conned to their own domains of choice/focus, with hardly any attempt to connect it
with other properties and functions of water. The second line of studies has focused
on the utilitarian aspects of water. Here the approach is largely qualitative and
focused on the role of water in various aspects of life and nature. The latter have
xv
been popular since antiquity. Neither of these two approaches addresses the explicit
(especially dynamic) role of water molecules in biological functions.
Water that is present in biological cells, in the grooves of DNA, on the surfaces of
proteins is found to be quite different fromwater in the bulk, the water that we drink.
The term biological water was coined to highlight this difference. In nature, water
is also found within rocks and conned systems, such as in tree leaves. Such
conned water also exhibits properties quite distinct from those in the bulk. The
main modication that occurs from the bulk state of water is the partial or even full
loss of the hydrogen-bond network that so uniquely denes water. In biological and
many natural systems, water faces a multitude of interactions from the surface.
However, water seems to retain sufcient resources of its own to adjust to new
environments and continue to perform its wide-ranging roles.
We have placed special emphasis on properties that have been observed in bio-
molecules, such as proteins, DNA, and RNA, and in other complex systems such as
micelles, reverse micelles, and carbon nanotubes. As observed above, we tried to
see what happens to water due to the proximity to a foreign surface. Second, we
attempted to provide a coherent explanation of properties observed from a modern,
molecular, often dynamic, perspective. The latter relies heavily on recent advances
in the eld, often driven by computer simulations. Third, we spend considerable
effort to discuss biological functions of water. By biological function we do not
imply only the stabilizing effect of water in proteins and DNA, but the direct role
that water molecules themselves play in biochemical processes, such as in enzyme
kinetics and protein synthesis, that are essential for life. Thus, the third purpose of
this book is to articulate such biological and chemical functions in the light of our
current understanding of molecular aspects of water although, as stated above, the
development in this area is largely incomplete.
Throughout the monograph, we have attempted to avoid using mathematical
expressions and minute details of sophisticated theories in order to make the content
accessible to a larger number of students and interested readers who are not
professional researchers in the area. We believe that the properties of water are so
interesting, especially given the uniqueness of the liquid, that many scientically
inclined people will nd the subject fascinating. Although in some places detailed
discussions have been included to give a avor of the subject, we have attempted to
keep them at a minimum. We also address, towards the end of the book, certain
advanced topics of current research in water. They are not disjoint from the earlier
chapters and substantiate our efforts to explain the uniqueness of water. But readers,
if not interested in advanced topics, can avoid these chapters without much loss to
the completeness.
Our focus on molecular explanations of the observed properties distinguishes the
present monograph from the others existing in the literature. At the same time, this
xvi Preface
approach also limits the range of topics that we could address here. But there are
many excellent books/monographs on water which can supplement this lacuna.
Da Vinci called water Natural vehicle of change. We attempt to show here that
the detailed role of water in biological and chemical change can be fascinating and
elusive at the same time. We hope this book (despite many lacunae) will be
welcomed by students and scientists at large, especially because it documents
some of the signicant progress that has been made in the last few decades.
It is tting to end the preface of this book on water with the following well-known
quote of Mark Twain. My books are like water; those of great geniuses are like
wine. (Fortunately) everybody drinks water. I hope this book on water qualies as
Mark Twains water.
Preface xvii
Acknowledgements
Many people, particularly my students, present and past, have helped during the
writing of this book. Without their support, this project would never have been
complete. I am particularly thankful to Dr. Biman Jana, who started on this project
with me and contributed signicantly to the initial stages of development. Ms.
Susmita Roy helped enormously in preparing the gures, reading the manuscript
and correcting many errors, even adding paragraphs when needed. Mr. Saikat
Banerjee and Dr. Mantu Santra helped in the writing of the hydrophobicity chapter.
Mr. Rajib Biswas, Mr. Rakesh S. Singh, Ms. Sarmistha Sarkar, Mr. Milan Hazra,
Mr. Rajesh Dutta, Ms. Rikhia Ghosh, Mr. Jonathan Furtado, Mr. Arpan Kundu,
and Dr. Mantu Santra also read several of the chapters and offered corrections
and modications. Ms. Naina Vinayak helped in reproducing many gures. I am
grateful to Professor Kankan Bhattacharyya for many discussions, suggestions,
and encouragement. Professor Iwao Ohmine and Professor Graham Fleming have
always been sources of encouragement in this long endeavor. Professor Shinji Saito
has been an incredible source of information and strength he helped with many
gures that he generated from his own simulation data. Kaushik Bagchi offered
valuable suggestions at critical stages and Kushal Bagchi read many pages. I am
grateful to my wife, Ms. Sukla Das, for support and encouragement. I also thank my
many students and collaborators who helped fashion my ideas and concepts in this
rapidly developing subject.
xviii
Part I
Bulk water
1
Uniqueness of water
What makes water so unique? People have asked this question for
centuries. Yet the answer seems elusive. Despite all of its complex proper-
ties, water has an amazing, deceptively simple chemical composition
just two hydrogen and one oxygen atoms! Yet it exhibits highly unusual
and puzzling physical properties. In this chapter we make a list of six
unusual molecular properties that together could be responsible for the
unusual characteristics of water.
1.1 Introduction
It is fair to state that in the biologically relevant temperature and pressure range,
none of the properties of water are like those observed in other common liquids
(such as ethanol, benzene or acetonitrile). To begin with, water has remarkably
high boiling (100 degree Celsius) and melting (0 degree Celsius) temperatures that
are unusual for a liquid consisting of molecules that are so small in size and so
light in molecular weight. Additionally, it exhibits a high critical temperature
(374C), compared to the liquids with similar or comparable molecular structure
that are mostly in the gaseous state at room temperature and pressure (such as
hydrogen sulde (H
2
S) and carbon dioxide (CO
2
)). It has large specic heat and
exhibits many other thermodynamic anomalies to be discussed later in Chapter 2.
Understanding the origin of the anomalous properties of water has turned out to
be an extraordinarily difcult task a task that is only partly completed.
Nevertheless, we need to make a beginning, with whatever understanding we
have acquired of bulk water [15], in our attempt to understand the diverse (and
myriad) roles that water plays in many complex environments, including biology.
We discuss below six unique features that can be held responsible for many of the
properties of water. But rst we present a few of the essential details about a water
molecule so that those features can be understood and appreciated.
3
1.2 Molecular structure
A water molecule is made of two hydrogen atoms that are attached to an oxygen
atom via covalent bonds, making it look V-shaped (as shown in Figure 1.1). The
OH bond length is about 0.1 nm (1 ) and the HOH bond angle is 104.51 (in the
gas phase). The oxygen atom has two unused lone pairs of electrons on it and they
contribute in no small measure to the unusual properties of water.
Apart from the covalent bonds between oxygen and hydrogen atoms, a water
molecule also has the ability to form hydrogen bonds with four other neighboring
water molecules. Two hydrogen atoms pair up with two oxygen atoms of two
different water molecules while the oxygen atom pairs up with two different
hydrogen atoms of two different water molecules. That is, one water molecule can
form hydrogen bonds with four different water molecules, and the three-
dimensional arrangement is tetrahedral, as shown in Figure 1.2.
A typical hydrogen-bond (OHO) distance, that is, the distance between the
oxygen atom of one molecule and the hydrogen atom of the participating second
water molecule, is 0.25 nm (2.5 ).
It is worth noting here that Linus Pauling was the rst to mention the hydrogen
bond, in 1912. In 1935, he rst advanced the theory of hydrogen bonds between
water molecules. Using quantum mechanics and chemical bonds he evaluated that
both covalent bonds and other electrostatic forces hydrogen bonds were
commencing in water. According to Pauling the covalency in a typical OHO
hydrogen bond is about 5% [6].
Although the arrangement of oxygen and hydrogen atoms is V-shaped with
positive charges localized on the hydrogen atoms and the negative charge on the
oxygen atom, the situation is known to be more complex, with the distribution of the
electrons of the two lone pairs of electrons which have been recently described as
smeared between two tetrahedral lobes. As a result, in an approximate sense, the
local arrangement of water molecules in liquid water at low temperature can be
Figure 1.1. Molecular structure of water. The dark and light balls represent oxygen
and hydrogen atoms, respectively. The sticks between the oxygen and the
hydrogen atoms represent the OH chemical bond.
4 Uniqueness of water
regarded as a distorted tetrahedral, as shown in Figure 1.2, with each water
molecule, on average, forming four hydrogen bonds, as mentioned above.
In fact, the delocalized nature of the electron density between the two lobes facilitates
a fth neighbor to approach a tagged water molecule with a hydrogen atom of the
incoming molecule pointed towards the oxygen atom of the tagged molecule. This
allows formation of a bifurcated hydrogen bond and the existence of a water molecule
with ve neighbors. The ve-coordinate water molecule so formed plays an important
role in the thermodynamics and dynamics of liquid water. In Figure 1.3 we present a
schematic representation to show how two 4-coordinated water molecules can be
converted to one 3-coordinated and one 5-coordinated water molecule.
In Figure 1.3 we show the structure of a 5-coordinated species. Note that the
formation of a H bond with a fth neighbor is already possible even with localized
charges on the two tetrahedrally placed lobes of the oxygen atom, and has been
observed in many simulations with classical models of localized charges, but the
delocalized nature may further facilitate its formation.
The quantum nature of electrons makes a water molecule more responsive and
discriminative to external perturbation than possible in the classical world. The
Figure 1.2. Typical molecular arrangement in the HB network around a central
water molecule in liquid water. Four other water molecules form HBs (two donor
(upper) and two acceptors (lower)) with the central water molecule. The
environment around the central water molecule is tetrahedral. (Adapted with
permission from http://mi-bitacora-diaria.blogspot.in/2009_02_01_archive.html.)
1.2 Molecular structure 5
lengths of the hydrogen bonds are also exible and they vary considerably at a given
time even among the three or four hydrogen bonds formed by the same molecule.
To summarize, the perfect picture given in Figure 1.2 represents an ideal
situation. The tetrahedron around the central water molecule is often distorted,
except in ice, where collective effects reinforce a tetrahedral geometry. We shall
discuss later quantitative descriptions which do more justice to local arrangement of
water molecules in the liquid state. But right now we ignore the detailed corrections
of the picture depicted above in Figure 1.2 and proceed with it.
In order to form an extended (that is, percolating) network that connects a large
fraction of molecules of the entire system, there should be three or more hydrogen
bonds per water molecule (unless molecules form large disconnected linear
chains, which are unlikely and not seen in liquid water). Since each water
molecule can easily form four hydrogen bonds, it can support such a network.
Indeed this very ability to form a hydrogen-bond network has always been
hypothesized to be the main reason for many anomalies exhibited by water (as
shall be discussed later) [16].
Although bulk and conned water has been relentlessly studied both theoretically
and experimentally for many decades [7], quantitative progress towards under-
standing the mysteries of water began only after we could study the structure and
dynamics of about a thousand molecules via computer simulations. The landmark
papers by Stillinger and Rahman [3] in the 1970s marked a turning point in our
4
3
4
5
Figure 1.3. Pictorial representation of the conversion scheme from 4- to 5- and
3-coordinated water. Here oxygen atoms are light gray and hydrogen atoms are
black.
understanding of liquid water. We note that these simulations employ classical
mechanics and use a rigid model of water where charges are xed at different
sites to mimic the charge distribution of water.
Nevertheless, these simulations led to many important results. First, they conrm
the view that water contains an extensive hydrogen-bond network where the two
hydrogen atoms participate in one hydrogen bond each while the lone oxygen atom
forms two or three hydrogen bonds with the hydrogen atoms of nearest-neighbor
water molecules (as shown in Figures 1.2 and 1.3). The average number of hydro-
gen bonds per water molecule in liquid water under ambient conditions was found to
be about 3.5. As we have already mentioned, we need at least three hydrogen bonds
by a water molecule to ensure a connected percolating network; these computer
simulation studies established that liquid water is a giant gel consisting of water
molecules connected by hydrogen bonds. Doubly hydrogen-bonded water mole-
cules are also present in the network, as they connect two extended networks. Some
water molecules can be even singly hydrogen bonded and are called dangling
bonds. But the fraction of doubly and singly hydrogen-bonded water molecules
is small.
However, compared to chemical bonds (like the ones between hydrogen and
oxygen atoms in water) these hydrogen bonds are weak, with dissociation energy
comparable to thermal energy. As a result, these bonds continuously formand break
in liquid water. Hence the lifetime of a hydrogen bond is quite short, of the order of
two to three picoseconds (ps) where 1 ps = 10
12
s.
Thus, the extended network of H
2
O molecules in liquid water is a uctuating
network [4]. This uctuation lets water be responsive to foreign solutes because it
allows water to easily rearrange and solvate a large variety of solutes. This feature
partly allows water to act as a unique solvent.
1.3 Six unique features
As mentioned above, we can make a list of six unique features of water that are
responsible for many of its abilities and properties. We now list these properties.
(i) Awater molecule is small in size and low in molecular weight. The rst allows
it to occupy even relatively conned spaces, such as the grooves of DNA, or
the active sites of enzymes. In the latter case, often the presence of a single
water molecule plays an important role.
(ii) Due to the large electro-negativity difference between oxygen and hydrogen
atoms there is partial charge separation along the bond giving rise to (approxi-
mately) 0.84e charge on the oxygen atom and (again, approximately) +0.42e
charge on each hydrogen atom. This distribution of positive and negative
1.3 Six unique features 7
charges promotes its hydrogen-bonding ability. Since water is made up of two
hydrogen atoms that can form two H bonds and one oxygen atom that can
further hold two to three H bonds, each water molecule can form on the whole
four, and even ve, hydrogen bonds.
In addition, water can act as both a donor and an acceptor of a hydrogen
bond. Thus, it can stabilize both a positively and a negatively charged atom/
group. This property comes in handy at a protein surface. As mentioned earlier,
the electron charge in the lone pairs might be smeared between the oxygen
atoms, allowing water to react quite differently to an anionic or cationic
ligand.
(iii) An additional aspect of the waterwater hydrogen bond not captured in the
classical models is the transfer of electron density from the oxygen atom of the
acceptor molecule to the hydrogen atom of the donor molecule [8]. See
Figure 1.4 for an illustration of this phenomenon.
This electron transfer gives the hydrogen bond a small measure of covalent
character, estimated to be about 5% [9].
Note further that the OH stretching frequency of the donor molecule
decreases proportionally to the strength of the hydrogen bond. This purely
quantum effect allows additional cooperativity in hydrogen bonds.
(iv) Awater molecule is characterized by a large dipole moment which is reected
in the large dielectric constant of liquid water, about 80 at room temperature.
This large dielectric constant is extremely useful in many chemical processes.
Related to this feature, the two lone pairs of electrons on the oxygen atom
render a water molecule polarizable to electric elds from other molecules or
from a charged surface. Thus, water molecules can respond to the changing of
charge distribution in an external solute (or surface) to lower the energy of an
HB Donor
e
+
+
HB Acceptor
Figure 1.4. Electron cloud migration from the hydrogen-bond (HB) acceptor to the
HB donor water molecule. In a water dimer the hydrogen atoms (shown in white)
of a better HBacceptor become more positive, while the oxygen (shown in gray) of
the HB donor becomes more negative.
assembly or molecular arrangement. However, a water molecule is not highly
polarizable, and is quite low in polarizability.
In neat liquid water, the polarizability effectively increases the dipole
moment of each individual water molecule, and is partly responsible for the
large dielectric constant of water.
(v) A collection of water molecules can form many structures of nearly equal
energy. This is most evident from a study of water clusters [10]. Also, ice is
known to have many polymorphs. As many of these structures are of similar
energy this makes a collection of a small number of water molecules highly
adaptive to various complex environments [5]. For example, when a layer of
water molecules faces a non-polar surface, the spatial arrangement and orien-
tation of water molecules are quite different from those when the layer faces a
polar or charged surface. This is because water molecules can adopt many
different structures.
(vi) The remarkable ability of water to sustain a uctuating extended hydrogen-
bond network allows facilitation of many dynamic processes that would
otherwise be impossible. The marginally stable nature of the hydrogen-bond
network arrangement makes it rather easy to initiate the molecular rearrange-
ment. The hydrogen-bond network can easily be distorted. In addition,
hydrogen-bond energy in water spans a wide range of energy (~39 k
B
T for
liquid water at 25C temperature) [4].
We have made the above list such that each property operates at least somewhat
independently. There are obviously correlations among these. Actually the ability of
water to sustain multiple timescales is a unique feature of this liquid. Water seems to
be able to respond according to the speed of perturbation. It responds slowly to slow
perturbation and rapidly to fast perturbation.
In the subsequent chapters we shall try to rationalize the properties of water in
complex systems by using these six features.
1.4 Modeling of water
Unfortunately, it has turned out to be exceedingly difcult to accommodate all the
unique features of water within any given, classical model. This is reected in the
absence of satisfactory agreement between experiments and any given model [2].
This is a bit unusual (and of course frustrating) because when one usually models a
given molecule, such as methane, it is adequate to use a simple functional formsuch
as a Lennard-Jones potential that incorporates a measure of size and a measure of
interaction energy at an optimal separation between two molecules. In the case of
water molecules, such a simple procedure does not work. Here we have to account
1.4 Modeling of water 9
for at least two length scales one for the molecular size and the other for hydrogen-
bond length. Second, one needs to take into account the charge distribution. And as
we have already discussed, quantum effects (electron transfer) give rise to a
cooperativity in hydrogen-bonding that is hard to mimic within a classical model.
Thus, the unique features discussed above prove to be particularly hard to model.
As a result of these complexities, although more than 100 different potential
functions have been employed, no fully satisfactory model has yet been developed.
But many of these models have been able to explain many of the experimental
observables, such as density maximum, values of viscosity and self-diffusion coef-
cient, specic heat and compressibility, and dynamics of electron transfer reactions.
Thus, it is also not fair to state (as the statement is often made) that we do not
understand water. Although clearly perspectives differ, one should not lose sight of
the successes that have been achieved.
1.5 Conclusion
Liquid water is different from other liquids. Unique (and often termed anomalous)
properties of water originate ultimately from the unique molecular features of water.
We have made a list of six such features which combine to give rise to the unusual
properties of water. The list itself may not be unique or exhaustive but we think that
it provides a starting point to rationalize the properties of water.
In fact, an attempt to rationalize the diverse properties in terms of a few basic
features is a reductionist view which has a lot of advantages. Most importantly, it is
possible to get back to basics when one faces difculty in explaining the experi-
mentally observed properties. For example, a lot of our difculties in understanding
or describing the behavior of water molecules at the surface of proteins or charged
surfaces arise from our difculty in handling the polarizability anisotropy of water
molecules.
In the next chapter, we discuss a fewof the well-known anomalies of liquid water.
We shall return to the discussion of those anomalies again in the penultimate chapter
of the book. In the intervening chapters we shall discuss various properties of water
in diverse systems, with a close connection between theory and simulations.
References
1. J. H. Gibbs, C. Cohen, P. D. Fleming, and H. Porosoff, Toward a model for liquid water.
J. Solution Chem., 2 (1973), 277; P. D. Fleming and J. H. Gibbs, An adaptation of the
lattice gas to the water problem. J. Stat. Phys., 10 (1974) 157.
2. F. H. Stillinger, Effective pair interactions in liquid water. J. Phys. Chem., 74 (1970),
3677; F. H. Stillinger, Theory and molecular models for water. Adv. Chem. Phys., 31
(1975), 1.
3. F. H. Stillinger and A. Rahman, Improved simulation of liquid water by molecular
dynamics. J. Chem. Phys., 60 (1974), 1545; A. Rahman and F. H. Stillinger, Hydrogen-
bond patterns in liquid water. J. Am. Chem. Soc., 95:24 (1973), 79437948.
4. I. Ohmine and H. Tanaka, Fluctuation, relaxations and hydration in liquid water.
Hydrogen-bond rearrangement dynamics. Chem. Rev., 93:7 (1993), 25452566;
M. Matsumoto, S. Saito, and I. Ohmine, Molecular dynamics simulation of the ice
nucleation and growth process leading to water freezing. Nature, 416 (2002), 409.
5. Y. K. Cheng and P. J. Rossky, Surface topography dependence of bimolecular hydro-
phobic hydration. Nature, 392 (1998), 696.
6. L. Pauling, B. Kamb, et al., Linus Pauling: Selected Scientic Papers, World Scientic
series in 20th century chemistry, vol. 10 (River Edge, NJ: World Scientic, 2001).
7. B. Bagchi, Molecular Relaxation in Liquids (New York: Oxford University Press,
2012).
8. R. J. Gillespie and P. L. A. Popelier, Chemical Bonding and Molecular Geometry: From
Lewis to Electron Densities (New York: Oxford University Press, 2001).
9. E. Arunan, G. R. Desiraju, R. A Klein, et al., Denition of the hydrogen bond. Pure
Appl. Chem., 83:8 (2011), 16371641.
10. F. N. Keutsch and R. J. Saykally, Water clusters: untangling the mysteries of the liquid,
one molecule at a time. Proc. Natl. Acad. Sci. USA, 98 (2001), 1053310540.
References 11
2
Anomalies of water
The unique features of individual water molecules (discussed in the
preceding chapter) give rise to many anomalous properties of liquid
water. Commonly attributed to the presence of an extensive hydrogen-
bond network, these anomalies teach us a lot more about water itself.
Anomalies are observed in many properties, ranging from a density
maximum at 4C, the temperature dependence of isobaric specic heat
and isothermal compressibility to a host of dynamic properties. Here we
discuss some of them, with the emphasis on collective properties that are
relevant to our study of complex systems discussed later. Understanding
these anomalies is still the subject of considerable research activity.
2.1 Anomalous properties
Water is most anomalous at low temperatures, with the remarkable density max-
imum at 4C. When water is supercooled below its freezing/melting temperature of
0C at ambient pressure, most thermodynamic properties exhibit strong anomalies.
However, water exhibits many weak to strong anomalies even at room temperature,
particularly in its interactions with solute molecules. The existence of a large
number of anomalous properties makes water one of the most puzzling substances
known to mankind. We next discuss some of these anomalies. The understanding of
these anomalies is not only challenging and has bafed scientists for generations but
it also holds the key to evolving a unied understanding of this liquid.
2.1.1 Density maximum
The density anomaly is one of the oldest known and one of the most quoted puzzles
in the behavior of water [1]. Unlike other simple liquids, which expand upon heating
(density decreases), water contracts on heating above 277 K (4C), at atmospheric
pressure. The density prole of liquid water is shown in Figure 2.1 as a function of
13
temperature at various pressures. The temperature of maximum density (popularly
known as TMD) moves to lower temperature as the pressure increases, as it is more
difcult to expand at higher pressures. This TMD serves as a good measure of the
order in the liquid.
As is well known, solid ice is of lower density than liquid water. It is because of
this anomaly that ice can oat on water and sh can survive in the warmliquid water
below a layer of ice, at temperatures well below 0C. Freezing of water and melting
of ice are still not well understood.
There is a relatively simple explanation of the density maximumat 4C, in terms of
the average coordination number of water molecules. As we discussed in the rst
chapter, a given water can form different numbers of hydrogen bonds, ranging from
zero to six, with the most probable number being close to three at ambient conditions.
We can relate the volume of a given water molecule to the number of hydrogen bonds
it forms. Thus, a given water molecule with two hydrogen bonds has a larger volume
than one having six hydrogen bonds (both examples are relatively rare).
As temperature is decreased fromabove (say, from20C) towards 0C, the number
of hydrogen bonds per water molecule increases as the 2- and 3-coordinated water
molecules get replaced predominantly by more stable 4-coordinated water molecules.
In the process some 5-coordinated water molecules also form. Computer simulation
studies showthat at 10C, about 70%of the molecules are 4-coordinated while 3- and
5-coordinated are nearly equally populated at about 14% each. As mentioned above,
conversion of 2- and 3-coordinated water molecules to 4-coordinated ones is the
main reason for the increase in density on lowering the temperature of
water. However, as we approach 4C, energetic reasons now favor 4-coordinated
water molecules over 5- or 6-coordinated water molecules. These higher coordinated
0 5 10 15 20 25 30
Temperature (C)
7
5
5
0
2
5
1

b
a
r
1004
1002
1000
998
996
p
.
d
e
n
s
i
t
y

(
k
g
/
m
3
)
Figure 2.1. Temperature-dependent density of liquid water at various pressures.
Note the shifting of TMD to the lower temperature as pressure is increased.
(Adapted with permission from http://www.engineeringtoolbox.com/uid-
density-temperature-pressure-d_309.html.)
14 Anomalies of water
water molecules begin to get replaced by energetically more stable 4-coordinated
water molecules, leading to a fall in density. Thus, the density maximum is an
interplay between the natural strength of hydrogen bonds (which arise from charge
distribution in the water molecule) and the thermal energy.
Detailed numerical calculations indeed justify the above simple logic. Thus, the
maximum density of water at 4C is not too much of a mystery. However, explana-
tions of other anomalies are not that simple.
2.1.2 Isobaric specic heat (C
P
)
The specic heat of a substance provides a quantitative measure of the amount of heat
necessary to increase the temperature of the system by 1C. We now discuss how this
amount is related to the number of congurations (distinct molecular arrangements) that
are available to the system, within a small range of energy around a given energy. As we
provide heat energy to the system, the energy gets divided into all the microscopic states
of the system. All the microscopic states must get the energy. This energy is actually the
enthalpy (H), which includes both the internal and the mechanical energy (in the formof
PV, where P is the pressure and V is the volume of the system).
Since the enthalpy of the system is a sum of contributions of many molecules, the
probability distribution of enthalpy H is a Gaussian function, with the width of
the distribution naturally given by the mean-square deviation of enthalpy, or the
uctuation of enthalpy. The exact relationship between specic heat and enthalpy
uctuation is given in Appendix 2.A.
Now the number of congurations available to a system at a given energy is
measured by the entropy of the system. As discussed in Appendix 2.A, C
P
is thus
directly proportional to the entropy uctuation in the system.
Since the thermal uctuation should generally decrease with decreasing tempera-
ture, one would expect that C
P
should decrease with decreasing temperature. This is
the scenario for most simple liquids. However, in the case of water, it increases as
the temperature is decreased belowT = 320 Kand at temperature belowthe freezing/
melting temperature, the specic heat appears to diverge at a singular temperature
with a power law (as shown in Figure 2.2) [2].
2.1.3 Isothermal compressibility (
T
)
The isothermal compressibility of liquids gives a measure of the change in volume
of the system due the change in pressure applied at a constant temperature. A
microscopic expression of isothermal compressibility is given in Appendix 2.A.
The expression shows that compressibility is related to the natural uctuations in the
total volume of the system.
For most simple liquids,
T
decreases with decrease in temperature as the volume
uctuation decreases. However, in the case of water, it increases like C
P
below a
certain temperature and appears to diverge with lowering temperature, as shown in
Figure 2.3 [3].
2.1.4 Coefcient of thermal expansion (
P
)
The coefcient of thermal expansion
P
provides us with a measure of volume
change of the system due to a change in temperature at constant pressure.
Temperature (C) T
Cp
typical liquid
Tm 35 C
H
2
O
S
p
e
c
i
c

H
e
a
t

C
a
p
a
c
i
t
y
Figure 2.2. Temperature dependence of the isobaric heat capacity (C
P
) in liquid
water. The dashed line represents the behavior of typical liquids. Note the turn-
around and divergence-like behavior for water at the melting temperature (T
m
). The
gure is reproduced from the thesis of Dr. Pradeep Kumar. http://polymer.bu.edu/
~hes/water/thesis-kumar.pdf.
Tm
46 C
typical liquid
H
2
O
T
I
s
o
t
h
e
r
m
a
l

C
o
m
p
r
e
s
s
i
b
i
l
i
t
y
Temperature (C) T
Figure 2.3. Temperature dependence of isothermal compressibility (
T
) in liquid
water. The dashed line represents the behavior of typical liquids. Note the turn-
around and divergence-like behavior for water. The gure is reproduced from the
thesis of Dr. Pradeep Kumar. http://polymer.bu.edu/~hes/water/thesis-kumar.pdf.
For simple liquids, the volume of the system increases with temperature
and thus
P
is always positive. Also
P
decreases with decrease in temperature
as the volume and entropy uctuations in the system decrease. However, in the
case of water it becomes zero at the temperature where density is maximum
(TMD) and then becomes negative with further decrease in temperature. This
suggests that below TMD the entropy increases with decrease in volume. Like C
P
and
T
,
P
also seems to diverge with a power law at low temperature as shown in
Figure 2.4 [2].
Since the experiments on bulk liquids including water cannot be performed below
the homogeneous nucleation temperature (T
H
; for bulk water T
H
= 38C), where
crystal formation is found to become inevitable, it is not possible to test whether the
apparent divergences of the above three quantities at low temperature are indeed
divergences or something else. However, experiments on nano-conned water and
extensive computer simulation studies (which have been possible since the forma-
tion of crystals is difcult in such systems and we can study the liquid water well
below its homogeneous nucleation temperature) nd that these quantities do not
diverge but rather have a maximum at low temperature.
2.1.5 Dynamic anomalies present at low temperature
For simple liquids, the temperature dependence of dynamics is usually given by a
form that is known as the Arrhenius equation. For relaxation time, the Arrhenius
equation is given by =
0
exp
A
k
B
T
_ _
, where is the measured relaxation time and A
is the activation energy, which is usually weakly temperature-dependent, and
0
is a
tting parameter which can be regarded as a reference relaxation time. The
P
Tm
4 C
(c)
T
typical liquid
H
2
O
T
h
e
r
m
a
l

e
x
p
a
n
s
i
o
n

c
o
e
f
c
i
e
n
t
Temperature (C)
Figure 2.4. Temperature dependence of coefcient of thermal expansion (
P
) in
liquid water. The dashed line represents the behavior of typical liquids. Note the
unusual behavior of liquid water below the melting temperature (T
m
). The gure is
reproduced from the thesis of Dr. Pradeep Kumar. http://polymer.bu.edu/~hes/
water/thesis-kumar.pdf.
Arrhenius form is valid at high temperatures (above its freezing/melting tempera-
ture) and its origin is attributed to the presence of energy barriers that restrict the
motion of molecules. However, the temperature dependence of relaxation time
becomes non-Arrhenius at low temperature, below the freezing/melting tempera-
ture. The reason for such crossover in dynamic behavior has been a subject of
intense discussions, and it is usually attributed to the emergence of a situation where
motions of distinct molecules are correlated. Such correlated motions appear in cold
liquids below their freezing/melting temperature, where they are termed super-
cooled liquids. Here the liquids become increasingly more viscous and ultimately
some liquids transform into glass if cooled sufciently fast.
However, water behaves differently. In the case of water, correlated motions
appear even above the freezing temperature of 0C. Belowthe freezing temperature,
the motion of molecules becomes increasingly slower. In addition to the rapid
growth of specic heat and other response functions, the relaxation rates of water
show anomalous non-Arrhenius temperature-dependence.
There have been several explanations of this behavior, in terms of an impend-
ing rst-order phase transition, or the existence of a second critical temperature
owing to a liquidliquid transition at lower pressure. Unfortunately, these
suggestions cannot be veried in pure water as the liquid cannot be cooled
below 40C.
Inability to look at bulk water at low temperatures, say 3040C below the
freezing temperature, has motivated a different approach where water has been
studied in conned small systems that do not seem to freeze to ice easily. The
dynamics of water conned in nanopores and water surrounding biomolecules
(these water molecules can be cooled below 40C) are found to change rather
sharply from non-Arrhenius at high temperature to Arrhenius at low temperatures.
The logarithm of density relaxation time () of water conned in MCM-41-S pores
(the MCM-41-S nanoporous silica matrix has 1D cylindrical pores arranged in 2D
hexagonal arrays, with pore diameters characterized by a narrow distribution) is a
function of 1/T. It has been shown that log () has a distinct crossover from non-
Arrhenius (increasing activation energy with decrease in temperature) at high T to
Arrhenius (constant activation energy) at low T [4]. Analysis of molecular arrange-
ment shows that the low-temperature phase is a low-density liquid/amorphous
phase.
We must alert the reader that it is not clear to what extent the above experiments
on narrow pores with a rather small number of water molecules can be used to
explain, or can be related to, the anomalous properties of bulk water at low
temperatures. However, it does establish the existence of a low-density liquid
phase with a free energy perhaps not too different from the high-density liquid
phase (the normal liquid at room temperature).
2.2 Translational and orientational order
We now shift our attention towards more microscopic structural aspects of water.
These aspects are discussed by quantication of local order inside a liquid.
Translational or spatial order in a liquid provides information about the local
arrangement of water molecules as a function of distance. This can be ascertained
by looking at the pair correlation function (g(r)) of the liquid. The pair correlation
function (also known as radial distribution function) gives the probability of nding
a pair of molecules separated by distance r. They can be any pair. Since all the water
molecules are identical and the liquid is homogeneous, we can x our attention on
any water molecule and look for the arrangement around it. The function g(r) then
gives the probability of nding other water molecules around our central molecule,
at a separation r. If the value of g(r) is more than 1, then it simply means that the
probability of nding the particles is more than what one would expect according to
the density of the liquid and vice versa.
A typical g(r) of liquid water is shown below in Figure 2.5. Note that the radial
distribution function between oxygen atoms of two different water molecules give a
peak at about the hydrogen-bonding distance (approximately 0.3 nm) as the neigh-
boring two water molecules are hydrogen bonded through the hydrogen atomof one
of the water molecules, as shown in Figure 1.2 of Chapter 1.
The rst of the two approximately equal height peaks of g
OH
(r) corresponds to the
hydrogen involved directly in the hydrogen (O- -H) bond (the smaller distance)
while the second one at longer distance corresponds to the non-hydrogen-bonded
second hydrogen of the molecule involved in the hydrogen bond with the central
water molecule.
3
2
1
0
0 5 10
g
OO
g
OH
r ()
g
(
r
)
Figure 2.5. Typical radial distribution functions (g
OO
(r) and g
OH
(r)) in liquid water.
The modulations at small separation distances indicate the short-range local order
in the liquid. Adapted with permission from Frontiers in Bioscience, 14 (2009),
35363549. Copyright (2009) Frontiers in Bioscience.
2.2 Translational and orientational order 19
The initial modulation and the rst peak in g(r) are due to the formation of local
structure in the liquid. For completely uncorrelated systems, g(r) = 1, and thus the
order parameter is zero. For a system with long-range order, the modulation in g(r)
persists over large distances, causing the translational order to grow.
In order to further quantify the local translational (or, spatial) order around
molecules, one introduces a second quantity, t
O
, which is obtained by averaging g
(r) over separation r. Thus, t
O
is a number which is a function of temperature and
density (or pressure). The precise denition is given in Appendix 2.Bof this chapter.
A parameter such as t
O
can describe the variation of order when the temperature or
pressure of the liquid is changed.
Figure 2.6 depicts an interesting variation of this translational order parameter as
both temperature and density are varied but at constant pressure. As shown in
Figure 2.6, t
O
shows both a maximum and a minimum at low temperatures as the
density is lowered from a high value of 1.3 gcm
3
. Such a combined presence of
maximumand minimumis not observed at higher temperatures and seems to appear
for the rst time close to the freezing temperature. The initial increase in spatial
order is found to be due to the formation of an increasing number of 4-coordinated
(that is, hydrogen-bonded) water molecules. During this range the density of the
liquid also decreases, facilitating the formation of an open 4-coordinated network.
While the pair correlation function (or radial distribution function, g(r)) provides
information about two particle arrangements, it does not provide information about
the relative arrangement of three or four or more water molecules. Knowledge about
1.4
1.2
1
0.8
0.6
0.8 0.9 1 1.1 1.2 1.3
T=240K
T=280K
T=320K
(g/cm
3
)
tmax
qmax
T
r
a
n
s
l
a
t
i
o
n
a
l

o
r
d
e
r

t

T
e

t
r
a
h
e
d
r
a
l

o
r
d
e
r

q

tmin
Figure 2.6. Density-dependent translational (t
O
) and tetrahedral order (q) of liquid
water at various temperatures. The maximum of translational order coincides with
the maximum of tetrahedral order. Note that the plot of translational order also has
a minimum. Adapted with permission from Phys. Rev. E, 76 (2007), 051201.
Copyright (2007) American Physical Society.
such higher-order structural arrangements can provide information that is essential
to understand the properties of water. As discussed earlier, water lacks the perfect
tetrahedral local structure (of ice) in the liquid state. There is a lot of disorder in the
local arrangement of water. Description of such order and disorder requires con-
sideration of angles between bonds formed by nearest-neighbor molecules. That is,
one considers the relative arrangements of three water molecules. Such an arrange-
ment involving three water molecules can be described by the angle made by the
two bonds that connect one molecule with two others, as shown below.
The molecular arrangement involving three water molecules can be described by
using the trigonometric function of the angle between three molecules (see
Figure 2.7). As discussed in Appendix 2.B, one usually denes a function q by
averaging over all three-particle neighboring molecules. The temperature depen-
dence of q is shown in Figure 2.6 (where we have also shown the same for spatial
order parameter, t
O
). The parameter q tracks the behavior exhibited by t
O
.
There is, however, an interesting aspect to this temperature dependence. For
simple liquids, q and t
O
increase with increasing density of the system. However,
for liquid water both q and t
O
increase with decrease in density and go through a
maximum at a certain temperature. Maxima of q and t
O
seem to coincide with each
other for a given temperature (as shown in Figure 2.6).
2.3 Temperaturedensity range of water anomalies
It helps if we categorize the anomalies discussed above into three different types:
(1) thermodynamic anomalies (for example, in density, C
P
,
T
and
P
), (2) dynamic
anomalies (relaxation time or diffusion, dynamic crossover), and (3) structural
anomalies (in translational and orientational order).
However, interestingly, these anomalies do not persist over the entire temperature
and density (or pressure) range. Thus it is also important to know in which range
Figure 2.7. Molecular arrangement of three water molecules, with linear HBs. The
angle between three oxygen atoms (OOO) is indicated as .
2.3 Temperaturedensity range of water anomalies 21
these anomalies exist. The range of anomalies for three different types is shown in
Figure 2.8 [5].
This interesting gure shows that the region of thermodynamic anomalies is
bounded inside the region of dynamic anomalies which in turn is bounded inside the
region of structural anomalies. Thus, as a preliminary guess, it can be inferred that
thermodynamic and dynamic anomalies can be understood in terms of structural
anomalies [5].
2.4 Conclusion
The rapid variations (rise or fall) in the value of the thermodynamic response functions,
namely the specic heat, the isothermal compressibility (both increase), and the
coefcient of thermal expansion (which decreases with temperature when the latter is
lowered below the freezing point), are some of the known spectacular anomalies of
liquid water. These variations have till now eluded a fully satisfactory understanding
[6]. Many computer simulation studies have been done and several theoretical
approaches have been developed but they are still not universally accepted.
Figure 2.8. Here three shaded regions in the densitytemperature plane show
different types of water anomalies. The structurally anomalous region is bounded
by the loci of q (orientational order parameter) maxima (upward-pointing triangles)
and t
O
(translational order parameter) minima (downward-pointing triangles).
Inside this region, water becomes more disordered when compressed, as t
O
and q
decrease with increasing density. The loci of diffusivity (inverse of relaxation time)
minima (circles) and maxima (diamonds) dene the region of diffusion (D)
anomalies, where D increases with density. The thermodynamically anomalous
region is dened by the temperature of maximum densities, TMD (squares), inside
which the density increases when water is heated at constant pressure. Adapted
with permission from Nature, 409 (2001), 318321. Copyright (2001) Nature
Publishing Group.
Denitions of these response functions in terms of the mean-square uctuations
or correlations among appropriate thermodynamic quantities are given in Appendix
2.A. Thus, the increase of specic heat and compressibility is related to a rather
sudden increase in these uctuations as temperature is lowered below the freezing/
melting temperature of water/ice. Also, the increase in mean-square uctuations in
entropy and volume is accompanied by a decrease in correlations between these two
quantities. The latter could happen if there is some degree of anti-correlation
between the two uctuations. That is, increase in volume leads to decrease in
entropy and vice versa.
An age-old and qualitative explanation of these anomalies is provided by assum-
ing a two-state model of water. In this model, a large region of the liquid (much
larger than the size of an individual water molecule) can exist either in a high-
density liquid (which is the normal liquid, say at 10C) and a low-density liquid
which consists mostly of randomly connected mostly 4-coordinated (by hydrogen
bonds, of course) water molecules with a density only slightly higher than that of ice
but less than that of the high-density liquid dened above. However, over a range of
temperature, say between 275 K and 240 K, these two states have similar free
energy, with the low-density liquid (LDL) gaining stability over the high-density
liquid (HDL) as the temperature is lowered. However, these two regions can inter-
convert. This can give rise to large uctuations in enthalpy, entropy, and volume.
The HDL has higher entropy but lower enthalpy as it is a mixture of 3-, 4-, and
5-coordinated water molecules, with a signicant fraction in each. Thus, when an
HDL region converts to an LDL region, the volume increases but the entropy
decreases. Repeated conversions such as this lead to large uctuations.
While there is general agreement up to this point among different views of low-
temperature water anomalies, there is considerable disagreement about the progres-
sion of the system when temperature is further lowered, say below 240 K. We
shall refrain from discussing these different approaches, as we shall hardly need to
dwell on temperatures below even 250 K in the book. However, we shall use this
two-state picture, which is a generally accepted explanation of the anomalies below
240 K. This is consistent with our view that water molecules can form many nearly
isoenergetic structures among themselves.
APPENDI X 2. A MI CROSCOPI C EXPRESSI ONS OF
SPECI FI C HEAT, I SOTHERMAL COMPRESSI BI LI TY,
AND COEFFI CI ENT OF THERMAL EXPANSI ON
All the above three quantities (specic heat, isothermal compressibility, and coef-
cient of thermal expansion) provide the response of the system to external perturba-
tion of different kinds (clear from the names) and are called response functions of
2.4 Conclusion 23
the system. Statistical mechanics provides useful expressions for them in terms of
uctuations, which are given below. These expressions also provide insight into the
state of the system, as also discussed below.
Specic heat at constant pressure, C
P
, is related to the microscopic properties of
the system by the following relations,
C
P
=
@H
@T
_ _
P
= T
@S
@T
_ _
P
=
DS ( )
2
_
k
B
=
DH ( )
2
_
k
B
T
2
(2:A:1)
where H is the enthalpy, S is the entropy and k
B
is the Boltzmann constant. Here
DS ( )
2
) denotes mean-square entropy uctuations and DH ( )
2
) corresponds to
mean-square enthalpy uctuations. A way to understand the above expression is
as follows. At a constant temperature and pressure, the entropy of the system can
uctuate (within a bound) because of inow and outow of heat from the reservoir
(bath). The mean-square entropy uctuation is a measure of the heat that a system
can naturally absorb due to its inherent capacity when temperature is raised. For
most liquids these uctuations decrease with lowering temperature, as expected.
However, for water it is found to increase. Thus, entropy uctuations increase, for
water, on lowering the temperature.
In a similar vein, the isothermal compressibility,
T
, can be written as,
T
=
1
V
@V
@P
_ _
T
=
DV ( )
2
_
k
B
TV
(2:A:2)
The last formula shows that isothermal compressibility is related to the uctuation in
the total volume (V) of the system. Again, the amplitude of this uctuation in
volume, measured by the second moment, is a measure of the volumes accessible to
the system. Nevertheless, Eq. (2.A.2) is really an elegant expression.
The case of thermal expansion coefcient,
P
, is different. This is related to the
uctuation of volume and entropy of the system by the following relation,
P
=
1
V
@V
@T
_ _
P
=
1
k
B
TV
DVDS) (2:A:3)
APPENDI X 2. B QUANTI FI CATI ON OF SPATI AL
ORDER I N WATER
Awater molecule is spatially correlated with its nearest neighbors due to hydrogen-
bonding. That is, we expect to nd a number of water molecules at a nearly xed
distance and also at a relatively xed orientation, as discussed in Chapter 1. In order
to quantify this spatial order one often denes a translational order parameter as,
t
O
= 4
_
r
c
0
g r ( ) 1
r
2
dr (2:B:1)
Here r
c
is the separation where g(r) reaches the value unity after the peak, that is
before any other modulation due to higher separation. In most liquids, the tetra-
hedral parameter increases with lowering of temperature but in water it reaches a
maximum and then drops, as discussed in the text.
The local tetrahedral arrangement of neighbors around a water molecule in liquid
water is best captured by the orientational order parameter q, which is dened as,
q = 1
3
8
3
j=1
4
k=j1
cos
jk
1
3
_ _2
(2:B:2)
Here
jk
is the OOO angle formed by the two nearest-neighbor oxygen atoms
(j and k) with the central oxygen atom for which the local order parameter q is
being calculated. Here (j,k) indices are arranged so that the OOO bond angles are
picked up properly, without over-count. We then average the value of q over all the
oxygen atoms of the liquid. For a perfect tetrahedral local structure (as in ice), the
value of q is 1, while q is equal to zero for random rotational arrangements (such as
in the gas phase). In liquid water at room temperature, this value is close to 0.5,
which is pretty high.
References
1. R. Waller, Essayes of Natural Experiments (original in Italian by the Secretary of the
Academie del Cimento): (1964) Facsimile of English translation. Johnson Reprint, New
York (1964).
2. C. A. Angell, J. Shuppert, and J. C. Tucker, Anomalous properties of supercooled water.
Heat capacity, expansivity and proton magnetic resonance chemical shift from 038%.
J. Phys. Chem., 77 (1973), 3092.
3. R. J. Speedy and C. A. Angell, Isothermal compressibility of supercooled water and
evidence for a thermodynamic singularity at 45C. J. Chem. Phys., 65 (1976), 851858.
4. L. Liu, S.-H. Chen, A. Faraone, C.-W. Yen, and C. Y. Mou, Pressure dependence on
fragile to strong transition and a possible second critical point in supercooled conned
water. Phys. Rev. Lett., 95 (2005), 117802.
5. J. R. Errington and P. G. Debenedetti, Relationship between structural order and the
anomalies of liquid water. Nature, 409 (2001), 318321.
6. P. G. Debenedetti and H. E. Stanley, Supercooled and glassy water. Phys. Today, 56:6
(2003), 4046.
References 25
3
Dynamics of water: molecular motions
and hydrogen-bond-breaking kinetics
Molecular motions (rotation, translation, and vibration) of a water
molecule also turn out to be quite different from those of other common
liquids. Here all the six unique features of an individual water molecule
outlined in Chapter 1 manifest themselves in diverse ways. As we discuss
below, not only is the mechanism of displacements of individual water
molecules different, but the collective dynamics and dynamical response
of bulk water are also different. For example, the rotational motion of an
individual water molecule contains a surprising jump component and
vibrational energy relaxation of the OH mode involves a cascading
effect mediated by anharmonicity of the bond. These motions are reected
in many important processes such as electrical conductivity, solvation
dynamics, and chemical reactions in aqueous medium.
3.1 Introduction
It is natural to expect that the extensive hydrogen-bond network present in water
may substantially alter the nature of the molecular motion of individual water
molecules from those found in normal liquids where such a network is absent. In
those non-hydrogen-bonded liquids individual molecules usually move by small
steps. One such small step is mostly uncorrelated with the next one. Such a motion
by random steps is called Brownian motion. Brownian motion is the erratic and tiny
movement of small particles (often observable under optical microscope) when
large particles are suspended in a uid or gas of small particles. For example, if you
sprinkle tiny grains of dust into water, and then look at the dust particles under a
microscope, the dust particles appear to dance around, continuously and quite
randomly. This zigzag random motion happens regardless of how still the surface
of the water is kept.
This interesting phenomenon of random motion of particles in a liquid was
discovered in 1827 by the British botanist Robert Brown [1]. He was investigating
27
pollen grains in water, and noticed that they would not remain still under his
microscope. At rst he thought that the pollen was moving because it was alive.
But even hundred-year-old pollen grains danced around, so he knew there had to be
some other explanation. Later it was found that Brownian motion is exhibited not
only by particles suspended in liquids but also by the atoms and molecules con-
stituting a liquid itself, at least for most liquids. This mode of motion holds for most
liquids, starting from simple liquids such as argon (which can be approximated by
spheres), as shown below in Figure 3.1. In the same gure, we also show the
position displacements with time of several tagged molecules.
The reason for the small-step random motion of atoms and molecules in liquids is
the incessant collisions that they suffer with each other. In a beautiful series of
papers, Einstein showed that this Brownian motion is a consequence of the natural
motion of the molecules controlled by the temperature of the system [2]. Therefore,
Brownian motion is also called thermal motion of the molecules. Obviously, the
molecules of the liquid move faster when the liquid is heated, causing more agitated
Brownian movement of the big particles. Similarly, if you make the liquid less
viscous, the molecules can move more easily, also resulting in faster particle motion.
Turning now to water, it was indeed believed for a long time that a similar small-
step Brownian motion is the primary mode of displacement of water molecules in
the liquid too. However, recent studies have shown that the situation can be quite
different. In addition to small-amplitude motion, a water molecule often rotates by
large-amplitude jumps! A part of this chapter is devoted to telling the story of
anomalous water motion. However, even for large-amplitude motion surprisingly
the Einstein relation between diffusion and viscosity remains valid.
3.2 Timescales of translational and rotational motion
Translational and rotational diffusion coefcient of a molecule in a liquid provides a
quantitative measure of the dynamic timescales in the liquid. These coefcients are
related to viscosity by the StokesEinstein [2] and the DebyeStokesEinstein rela-
tion [3], respectively. Using the denition of diffusion coefcient in terms of mean-
square displacement [2] and the StokesEinstein relation, we can estimate the time
needed by a water molecule to translate a distance equal to its molecular diameter
trans
-

2
6D
(3:1)
Putting the value of molecular diameter ( = 2.75) and diffusion coefcient (D =
2.5 10
5
cm
2
/s) of water, one gets an estimate of timescale
trans
as ~5 ps (ps
denotes picosecond and 1 ps = 10
12
s). That is, a water molecule in the liquid state
moves one molecular diameter in 5 ps. This is quite fast!
28 Dynamics of water: molecular motions and hydrogen-bond-breaking kinetics
(a)
2
1
0.5
0
0.5
1
1.5
2
2.5
3
1
0
1
3
2
4
5
6
0
1
3
4
5
6
7
8
9
10
(b)
X-axis
Y-axis
Z-axis
Figure 3.1. (a) Simulation box of argon atoms (interacting with each other via the
LennardJones potential) extracted fromtheir trajectories. The position of an argon
atom is depicted as it executes its natural motion in the liquid state at temp = 183 K.
(b) This shows the trajectory of one tagged argon atom.
3.2 Timescales of translational and rotational motion 29
We now discuss the speed of rotational motion of a single water molecule. At
25C, the time constant of rotation of water is 2.5 ps. This is the time that a water
molecule takes to forget its initial state of rotation (determined by the angle it makes
with a laboratory frame). That is, water rotates also very quickly!
As we strive here to understand motion of water molecules at a microscopic level,
we need to use a certain formalism developed in the area of statistical mechanics.
This formalism is broadly known as time correlation formalism (TCF). While many
specialized texts exist in the literature on this important topic, we have included in
Appendix 3.A a brief discussion on the time correlation functions necessary to
understand the dynamics of rotation of a molecule in liquid.
As discussed in Appendix 3.A, we employ two kinds of time correlation func-
tions to describe rotational motion. They employ single particle and collective
quantities. While they can be quite different in some cases, usually they both
measure similar dynamics. Most of the experiments measure the collective response
of the liquid. It is, however, important to know the difference.
In the case of rotational motion of water molecules, the dynamic quantity is
naturally the angle that the water molecule makes with a coordinate axis, usually
with the z-axis in a space or laboratory xed frame. However, one cannot directly
measure the angle. Instead, one measures the cosine of the angle because in experi-
ments one sets a direction or axis by an external means, such as the electric eld of a
light. Light, being an electromagnetic wave, interacts with the dipole moment or the
polarizability of the water molecule. Thus, the light incident on the medium at time
t = 0 serves two purposes. First, it creates a direction or reference of measurement of
rotation. Second, it creates a disturbance or perturbation in the system. That is, the
electric eld of the light forces a rotation of some of the water molecules of the liquid.
After the light passes through the medium, the disturbed water molecules start their
natural Brownian motion and rotate back to equilibrium. This process of restoring the
isotropy can be measured optically by different techniques. One then constructs the
time correlation function and obtains the rate of rotation.
At the present time one can measure fast rotation by using ultrafast laser pulses.
Laser pulses are now available with width less than even 1 femtosecond (fs) where
1 fs = 10
15
s. However, there are many technical difculties in using short pulses.
But reliable experiments can be done with pulses of the order of 100 fs or so. As a
result one can measure the fast rotation of water.
3.3 Jump reorientation motion in water
Now we turn to a detailed discussion of rotational motion in water. As already
mentioned, the Debye rotational diffusive model was initially widely employed to
describe water reorientation. As explained above, it describes the reorientation as an
angular Brownian motion, that is, a sequence of uncorrelated small-amplitude,
angular steps [3]. Such a rotational Brownian motion picture is not a priori implau-
sible if one considers that a water molecule interacts strongly with its neighbors via
hydrogen bonds and that water rst has to break at least one H bond to reorient; the
resulting dangling OH then performs a random search for a new H-bond acceptor,
during which it reorients.
Unfortunately, even sophisticated experiments cannot unravel the detailed micro-
scopic nature or mechanism of the rotational motion of a small molecule in the
liquid state. Fortunately, however, one can examine this microscopic aspect of the
reorientation of water by employing computer simulations. This method allows one
to tag individual water molecules and follow their rotational and translational
motion over a period of time by solving Newtons equations of motion. One
needs to follow both the orientational and the translational (that is, the linear)
motion of the water molecules. When plotted against time, the motion of a molecule
over time is called a time trajectory.
Inspection of many such trajectories for the time-dependent orientation of the
individual water O-H bond shows that water molecules under normal conditions of
temperature and pressure move mainly by large-amplitude jumps [4,5]. This is
shown in Figure 3.2. There is of course a constant motion of water molecules by
small steps, but superimposed on such continuous motion are these large-amplitude
jumps that are absent in most liquids. This is a relatively new insight, developed
primarily by Damien Laage and James Hynes, in the year 2006 [5]!
Figure 3.2. Fluctuation of the direction of the dipole moment vector of a tagged
water molecule in the course of its motion through the liquid. The places where
large-scale uctuations occur are indicated by arrows. Adapted with permission
from J. Phys. Chem. B., 112 (2008), 14230. Copyright (2008) American Chemical
Society.
3.3 Jump reorientation motion in water 31
Laage and Hynes carried out detailed analysis and found the mechanism of such
jumps. As stated earlier, in order to reorient the tagged water molecule needs to
break at least one hydrogen bond. After reorientation, this water molecule again
forms an H bond to compensate for the loss of energy. Thus the mechanism of the
reorientation of a water OH bond is a natural dynamic process where the average
number of hydrogen bonds is locally conserved and is a simple consequence of the
trading of H-bond acceptors receiving an H bond from the tagged water OH.
Laage and Hynes analyzed molecular dynamics trajectories and recorded the
rotational dynamics of a water O*H* that was initially H-bonded to a water oxygen
O
a
but became H-bonded to a different oxygen O
b
(see Figure 3.3). For each of these
switching events, they examined the sequence preceding it and the sequence
following it, as long as no other H-bond exchanges occurred. They monitored the
oxygenoxygen distances, R
O*O
a
and R
O*O
b
, together with the angle between
the projection of the O*H* vector on the O*O
a
O
b
plane and the O
a
O*O
b
angle
bisector (Figure 3.3). When the bond breaks and re-forms, the angle rotates. It is
rather easy to understand the basics of the process. When = 0, then H* is
equidistant from O
a
and O
b
, and that describes the transition state of the switching
events.
Figure 3.4 shows the exchange of distances between the three oxygen atoms
during the hydrogen-bond exchange process. The distance between Q* and O
a
undergoes a rather sharp increase while that between O
*
and O
b
undergoes a sharp
decrease [5].
R
o
a
o
b
R
o
*
o
b
o
b
R
o
*
o
b
H
*
O
*
Figure 3.3. We show the key microscopic quantities used to determine the jump
reorientation motion in water. Notations O*, O
a
, and O
b
are dened in the text. The
HB involving hydrogen atom H* and oxygen atoms O* and O
a
breaks (in the
example) and gets replaced by the one between O* and O
b
. Adapted with
permission from J. Phys. Chem. B., 112 (2008), 1423014242. Copyright (2008)
American Chemical Society.
3.4 Effects of temperature on water motion
To be historically fair, other people did observe the existence of jump motions in the
rotation of water molecules in the liquid state but detailed analysis of the dynamics
of an individual event was not carried out before. Given that perspective, the Laage
Hynes mechanism of water rotation by large-amplitude jumps is indeed a departure
from conventional and prevailing wisdom that water rotation is Brownian; that is, it
occurs differently in water from in other liquids where motion by small steps
dominates. Experimental verication of the jump diffusion model came from a
beautiful study of the temperature-dependent rate of water rotation. However, both
the experiments and the interpretation of results are quite involved. We shall discuss
the results as simply as possible.
Investigation of the effect of temperatures was carried out by Fayer and
co-workers of Stanford University by studying the time variation of the frequency
of the bond between oxygen and deuterium atoms (that is, the OD bond) of a HOD
molecule [6]. HOD is water with one of the hydrogen atoms of water replaced by a
deuterium atom. The advantage of this replacement is that now the frequency of the
OD bond becomes signicantly different from that of the OH bond and can be
studied separately. The frequency of the OD stretch is a sensitive function of the
environment, in particular of the hydrogen bond that the deuteriumatommakes with
an oxygen atom of the neighboring water molecule.
This important study shows that the amplitude of the inertial component (extent
of inertial angular displacement) depends strongly on the stretching frequency of the
Figure 3.4. Time variation of key quantities across an HB breaking event. Here we
show the variation in the oxygenoxygen distance and the angle (as indicated in
Figure 3.3). The sharp variations in these two quantities are the signatures of a
bond-breaking event. Adapted with permission fromJ. Phys. Chem. B, 112 (2008),
14230. Copyright (2008) American Chemical Society.
3.4 Effects of temperature on water motion 33
OD oscillator. The amplitude becomes smaller as the OD stretching frequency
becomes smaller. This is because when the OD stretching frequency is small, the
hydrogen bond involving the OD bond is of higher strength and thus it is difcult to
perform the wobbling motion in a cone of larger amplitude.
However, the inertial component becomes frequency independent at lower
temperatures! At a high temperature there is a correlation between the amplitude
of the inertial decay and the strength of the ODO hydrogen bond, but at low
temperatures the correlation disappears, showing that a single hydrogen bond is no
longer a signicant determinant of the inertial angular motion. It is suggested that
the loss of correlation at lower temperature is caused by the increased importance of
collective effects of the extended hydrogen-bond network. The temperature
dependence of the experimentally measured orientational correlation function at t
= 100 fs as a function of OD stretching frequency is shown in Figure 3.5.
As discussed above, the long-time part of the reorientation involves a jump of the
O*H* bond from the direction of the old acceptor (O
a
) to that of the new acceptor
(O
b
) (the LaageHynes mechanism). This is a complicated process and it fully
randomizes the reorientation. It is expected that the timescale of this process should
depend on the availability of the new acceptor and not on the strength of the
hydrogen bond formed with the old acceptor. This suggests that the long-time part
(>100 fs) of the rotational anisotropy need not depend on the frequency of the OD
stretch.
Figure 3.5. The plot of temperature dependence of the experimentally measured
orientational correlation function at t = 100 fs as a function of OD stretching
frequency. Note the near dependence of the frequency of the OD stretch at low
temperature, which is an indication of the collective nature of the dynamics.
Adapted from Proc. Natl. Acad. Sci. USA, 105 (2008), 5295. Copyright (2008)
Proc. Natl. Acad. Sci. USA.
This has indeed been observed in recent experiments. The results are shown in
Figure 3.6. The rotational anisotropy functions for all the frequencies decay with a
single time constant of 2.6 ps.
3.5 Translational diffusion
The translational diffusion coefcient of a liquid provides information about the
mobility of particles in the system. For simple liquids, an increase in density (say,
via increase in pressure) of the systemmakes the motion of the particles slower, thus
making the diffusion slower. However, water molecules move faster with increasing
pressure, reaching a maximum as shown in Figure 3.7. This effect becomes more
and more prominent in the supercooled liquid region. This anomalous increase of
the diffusion with increasing pressure is attributed to the breaking of hydrogen
bonds.
It can be considered remarkable that even though the density decreases with
lowering temperature (below 4C), molecular motion slows down rapidly. This can
be rationalized from Figure 3.8. As the temperature is lowered (below 4C), density
decreases since water forms more and more 4-coordinated water molecules. This
leads to an increase in local order as discussed in Chapter 2. A pictorial demonstra-
tion of the strong temperature dependence of the Brownian motion of liquid water is
shown in Figure 3.8. Here we show molecular displacements at two different
temperatures (300 K and 250 K).
Note the vastly different dynamic behavior at the two different temperatures.
0
top
bottom
2509 cm
1
2519 cm
1
2519 cm
-1
2539 cm
1
2558 cm
1
2578 cm
1
2599 cm
1
r
(
t
)

-

a
n
i
s
o
t
r
o
p
y
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
1 2 3
t (ps)
4 5
5 4 3
t (ps)
r
= 2.6 0.1
2 1 0
0.05
0.10
r
(
t
)

-

a
n
i
s
o
t
r
o
p
y
0.15
0.20
0.25
0.30
0.35
Figure 3.6. Plot showing the long-time rotational anisotropy decay curves for
various frequencies at 25C. (Inset) A decay curve on a semi-log plot with single
exponential t. Adapted with permission from Proc. Natl. Acad. Sci. USA, 105
(2008), 5295. Copyright (2008) Proc. Natl. Acad. Sci. USA.
3.5 Translational diffusion 35
3.6 Hydrogen-bond lifetime dynamics
Several criteria to dene a hydrogen bond have been suggested in the literature,
based on either geometric conditions, energetic considerations, or orbital occu-
pancy. A widely employed geometric denition is based on two distances and an
angle. A hydrogen bond is supposed to exist between two water molecules if the
following set of three criteria is satised [7]:
(i) the distance R
OO
between the two oxygen atoms needs to be less than 3.5 (or
0.35 nm);
(ii) the angle
HOO
between the OO bond vector and the OH bond vector (OH
is the bond involved in the hydrogen bond) must be less than 30;
(iii) and lastly, the distance R
OH
should be less than 2.45 .
We illustrate the hydrogen-bond geometry in Figure 3.9.
Both rotation and translation of water molecules are at least partly determined by
the rates of breaking and forming of hydrogen bonds. These rates are therefore of great
interest. In order to describe the lifetime of a hydrogen bond (HB), we need to dene
two time correlation functions, denoted by S
HB
(t) and C
HB
(t). We describe the
expressions of these two correlation functions in the Appendix 3.B to this chapter.
S
HB
(t) is a special kind of function which quanties the time needed to break the bond
for the rst time it does not allow re-formation once you are dead, you are dead!
The second function, C
HB
(), on the other hand, includes re-formation in the lifetime.
The two functions clearly give different information, with the latter one more relevant.
Both of these functions of course depend on several criteria needed to quantify HB
forming/breaking, although the numbers obtained from different criteria are not too
different. The lifetime of S
HB
(t) in water is only about 0.5 ps at T = 300 K, while that of
C
HB
(t) is about 6.5 ps. That is, HBs are quite short-lived. This fact has important
Figure 3.7. Translational diffusion coefcient as a function of density. Note the
maximum at intermediate density. The gure is reproduced from the thesis of
Dr. Pradeep Kumar. http://polymer.bu.edu/~hes/water/thesis-kumar.pdf.
consequences. The difference in rate of decay of S
HB
(t) and C
HB
(t) can be correlated to
the different dynamic aspects of water they probe, as discussed later.
The decay of these two functions is shown in Figure 3.10. The much longer
lifetime given by C
HB
(t) is due to the re-formation of the bond after it is broken.
7 8 9 10 11 12 13 14 15 16 17 18
X ()
7
8
9
10
11
12
13
14
(b)
Y

(
)
Temperature = 250K
Figure 3.8. Brownian motion of a molecule in liquid water captured by computer
simulation at two different temperatures: (a) 300 K and (b) 250 K. This gure
shows a continuous time trajectory of the zigzag uncorrelated motion of a molecule
through the liquid. Note the increased localization of the waters motion.
9 10 11 12 13 14 15 16 17 18 19 20
X ()
17
18
19
20
21
22
23
24
(a)
Y

(
)
Temperature = 300K
Because C
HB
(t) allows for a long sojourn after breaking, C
HB
(t) may lead to an
overestimate of the value for the HB lifetime [8].
More insight into the decay dynamics than revealed by C
HB
(t) can be obtained by
looking into the time-dependent rate constant k(t), also dened in Appendix 3.B. The
time-dependent rate gives an instantaneous rate of decay of the function. As seen in
Figure 3.11, the post-transient relaxation of k(t) is clearly different. Beyond the transient
relaxation region, the slope of log k(t) decreases monotonically with time t and thus
cannot be tted to a power law either. We discuss this anomalous behavior below.
The complex behavior shown in Figure 3.11 originates from the coupling
between the hydrogen-bond population and the diffusion of water molecules. This
coupling makes the relaxation behavior non-exponential. Two bonded molecules
can diffuse apart only if the HB between them breaks. Again, a broken HB can
re-form if a molecule reverses its direction and diffuses back to its partner. This
correlation gives rise to non-exponential kinetics of hydrogen-bond rearrangement
in liquid water.
Figure 3.9. Geometrical denition of HB in water in terms of distances and angles
between two water molecules.
Figure 3.10. HB lifetime correlation function, C
HB
(t) (dened by Eq. (3A.1)) for a
pair of water molecules in the bulk. The inset shows the same for S
HB
(t) function
(dened by Eq. (3A.2)) Adapted from Phys. Rev. Lett., 89 (2002), 115505.
Copyright (2002) American Physical Society.
Near biological surfaces, one observes a marked non-exponential character of the
hydrogen-bond dynamics. The detailed quantication of the H-bond lifetime is
discussed in Appendix 3.B.
3.7 Vibrational dynamics of the OH bond
In vibrational relaxation we study the relaxation of a chemical bond. Because
vibrational relaxation is a sensitive probe of the water environment and dynamics,
considerable efforts have been devoted to understanding both the vibrational phase
and energy relaxations in bulk water. The focus of the present section is on the rapid
progress made in the past decade, both in experimental and in theoretical studies.
A water molecule is characterized by three intramolecular vibrational modes,
namely, the symmetric and the antisymmetric OH stretches and the HOH bend.
In the liquid state, the frequencies of these intramolecular vibrational modes get
shifted from their gas-phase values. Figure 3.12 shows the three (symmetric,
antisymmetric, and bending) normal modes of water and heavy water.
Vibrational dynamics in water can be categorized into two parts, vibrational
energy transfer and the dephasing (features associated with modulation of the OH
stretching frequency). Deak et al. [10b] used ultrafast IR-Raman spectroscopy to
study vibrational energy relaxation (VER) in water and heavy water [10]. They
found that the lifetime of the OH stretch in water and HDO is 1 ps whereas that of
the ODstretch in D
2
Ois 2 ps. It was Rey and Hynes who rst pointed out that VER
of the OH (and the OD) stretch could occur via the off-diagonal anharmonic
coupling with the overtone of the bending mode. Such off-diagonal anharmonic
Figure 3.11. The rate function k(t), computed from the molecular dynamics
trajectory (thin line). Note the non-exponential behavior at longer time. In the
inset the logarithmic scale of k(t) is shown with respect to time (t). Adapted with
permission from Nature, 379 (1996), 55. Copyright (1996) Nature Publishing
Group.
3.7 Vibrational dynamics of the OH bond 39
coupling can be efcient because the fundamental frequency of the OH stretch is
off-resonance with the rst overtone of the HOD bend by 530 cm
1
, which is
close to the frequency of the librational mode. On the basis of the seminal work of
Rey and Hynes [10c], Lawrence and Skinner [10d] have shown that the ultrafast
vibrational phase relaxation of the OH stretch can be understood from the
conventional KuboOxtoby theory of the frequency modulation time correlation
function [9]. Recently Nibbering and Elsaesser extensively reviewed the experi-
mental and theoretical investigations and discussed the potential of nonlinear
vibrational spectroscopy for microscopic understanding of HB dynamics in the
liquid state [10].
Both intra- and intermolecular vibrational dynamics of water are expected to be
modied signicantly near a heterogeneous surface. We shall discuss later the fact
that this indeed happens. This modied vibrational spectrumand dynamics can then
be used to extract microscopic information about the hydration layer.
3.8 Dielectric relaxation
The dielectric relaxation (DR) spectrum of pure water has been investigated in
considerable detail by different experimental techniques, such as dielectric loss and,
more recently, terahertz spectroscopy. The DR of water has also been investigated
by computer simulations. However, the computational efforts have been relatively
less successful because of the difculty of simulating polarizable water molecules.
Figure 3.12. Normal modes of water HDO with their frequencies in wave numbers.
Adapted with permission from J. Phys. Chem. A, 104 (2000), 4856. Copyright
(2000) American Chemical Society.
Nevertheless, steady progress has been made in recent years in incorporating the
effects of polarizability, but at a large increase in the cost of computation.
The complex dielectric function () is usually decomposed into the real and
imaginary parts,
( ) =
/
( )i
//
( ) (3:2)
where
/
() and
//
() are the real and imaginary parts of the dielectric function,
respectively. At room temperature, the real part
/
() (the permittivity factor) of
pure water is nearly 80 at a few MHz and about 1.8 at 10 000 GHz. The imaginary
part
//
() corresponds to absorption (dielectric loss) and exhibits a peak at a certain
characteristic frequency
m
. The DR time
D
is equal to 2/
m
. The dielectric
spectrum of pure water in the low-frequency region consists of two relaxing
components, with time constants of 8.3 ps and 1 ps, respectively; the former is
responsible for about 90% of the low-frequency relaxation. The 8.3 ps time com-
ponent is believed to be related to the rotational correlation time of 23 ps.
The increase in the value of
D
over the rotational correlation time can be under-
stood quantitatively in terms of the micromacro relation, which provides a relation-
ship between the orientational correlation time (a microscopic, single-particle
property) and the DR (a collective phenomenon). Simple continuum model argu-
ments give the following relation between the two relaxation times [11],
R
=
2
0
3
0
D
g
K
(3:3)
where
0
and
are the static and the innite-frequency dielectric constants of the

solvent (here, water), respectively, and g
K
is the well-known Kirkwood g factor,
with a value equal to 2.8. Kirkwoods g factor accounts for the orientational
correlation present in the liquid that enhances the value of DR time.
For water,
0
= 78.5,
= 4.86, and
D
= 8.3 ps at 300 K. Therefore, one gets
R
=
2 ps, which is close to the actual value. That is, the macromicro relation predicts the
Debye relaxation time to be signicantly longer than the single-particle relaxation time.
This is an interesting correlation which is yet to be fully understood from a physical
perspective. Nevertheless, the success of a relation like Eq. (3.3) is impressive.
Many high-frequency modes contribute to the dielectric spectrum of water
beyond the Debye relaxation regime. As already discussed, this spectral region is
extensively investigated by far-infrared spectroscopic techniques and simulations.
In addition to the 200 cm
1
band due to the intermolecular OO stretching and the
650 cm
1
band due to libration, there are a few high-frequency IR bands, which are
of relatively less weight as compared to the former two. These high-frequency
modes are under-damped and, therefore, are described differently.
3.9 Solvation dynamics
Solvation dynamics (SD) measures the time-dependent response of water to a
newly, instantaneously (mostly optically) created polar species within the liquid.
The quantity measured is the solvation energy of the polar solute probe due to
interactions with the dipole moments of the solvent (here water). This energy
derives contribution from the surrounding dipolar molecules of the solvent and is,
therefore, regarded as a collective quantity. However, the energy contribution froma
given solvent molecule also depends on the spatial separation r between the solute
probe and the chosen solvent molecule. This dependence varies r
2
for ion-dipole
interactions when we are calculating the energy of the ionic solute. Thus, in this
problem, the time-dependent contribution depends also on the length scale. This
makes SD a useful technique. In general, the ultrafast component (sub-100 fs in
water and acetonitrile) is dominated by the fully collective response, while the
slowest component derives a signicant contribution from the nearest-neighbor
water molecules.
The temporal evolution of solvent polarization relaxation may be described by
the non-equilibrium function S(t), dened by [12]
S t ( ) =
t ( ) 0 ( )
( ) 0 ( )
(3:4)
where (t) is the frequency denoting the position of the emission spectrumwhose time
dependence describes the red shift of the spectrum after excitation. Here, (t) is
determined either by taking the maximum of the spectrum (if the spectrum is sym-
metric) or by the average over the spectrum, that is, v(t) =
_
dI (; t)=
_
dI(; t),
where I(,t) is the time- and frequency-dependent emission spectrumof the uorescent
probe solute. S(t), as dened in Eq. (3.4), varies fromunity at time t = 0 to zero as time
goes to innity. Note that S(t) may contain a contribution fromsolute coordinates also,
particularly from solute self-motion, such as rotation and translation, which may
accelerate the rate of solvation.
The solvation time correlation function is often equated to the auto time correla-
tion function of energy uctuation. This is usually termed C(t) to distinguish it from
S(t). Thus, C(t) is dened as
C t ( ) =
E 0 ( )E t ( ))
E 0 ( )E 0 ( ))
(3:5)
where E(t) is the uctuation in solvation energy fromits average, equilibriumvalue
at time t. Within the linear response approximation, S(t) C(t). Therefore, we have
made no distinction between the two here.
The rst theoretical estimate of solvation time was obtained by generalizing the
continuum model of Born and Onsager by representing the dynamic properties of
the solvent through a frequency-dependent dielectric constant, (), which is
sometimes approximated by the simple Debye formula
( ) =
1 i
D
(3:6)
where
0
and
are the zero and innite-frequency values of the dielectric

constant, respectively, and
D
is the Debye relaxation time [13]. With the above
expression of the dielectric function, the continuum model predicts that SD of a
newly created ion and that of a newly created dipole is exponential, with time
constants given by
ion
L
=

0
_ _
D
(3:7)
dipole
L
=
2
C
2
0
C
_ _
D
(3:8)
where
C
is the dielectric constant of the solute probe. For water,
0
= 78.5,
= 4.86,
c
is typically between 4 and 5, and
D
= 8.3 ps at 300 K. Thus, the value of the
longitudinal relaxation time is
L
ion
0.5 ps. That is, even the continuum model
predicts an extremely fast solvation in water! Clearly, the small value of the
solvation time in liquid water is due to the large value of its static dielectric constant.
Interestingly, it can be shown that the force constant of polarization relaxation of
bulk water increases with the static dielectric constant. Therefore, SD becomes
faster as the static dielectric constant increases.
A rather simple experimental technique involving measurement of the time-
dependent uorescence Stokes shift (TDFSS) after an initial excitation has been
applied to measure SD in a large number of liquids. TDFSS occurs due to dipolar
solvation of the excited probe and thus gives an estimate of the solvation timescales.
In an important paper, Jimenez et al. reported the results of SDof the excited state of
the dye coumarin 343 (C343) in liquid water [14]. Their result is shown in
Figure 3.13. The initial part of the solvent response of water was found to be
extremely fast (few tens of femtoseconds) and it constituted more than 60% of the
total solvation energy relaxation. The subsequent relaxation was found to occur in
the picosecond timescale. The decay of the solvation time correlation function, S(t),
was tted to a function of the following form
S t ( ) = A
G
e
t
2
=
2
G
Bcos t ( )e
t=
1
Ce
t=
2
De
t=
3
(3:9)
where A
G
, C, and D (all tted) give relative weights of the initial Gaussian and the
subsequent exponential decay processes, and
G
,
1
,
2
, and
3
are the corresponding
relaxation time constants. The second term in Eq. (3.9) takes into account the
oscillatory features of the S(t) observed beyond the Gaussian decay in theoretical
calculations and simulations. The early simulation studies also predicted a fast
initial component with a Gaussian time constant less than 20 fs. Jimenez et al.
experimentally detected a Gaussian component (with a time constant of 28 fs, 48%
of the total amplitude) and a slower bi-exponential decay with time constants of 126
(20%) and 880 (35%) fs, respectively. Several other experimental and simulation
studies on SD of large dye molecules as well as electrons in water have demon-
strated that the dynamics of solvation in water is indeed ultrafast and occurs in tens
of femtosecond.
More recently, higher-order nonlinear optical measurements such as three-pulse
photon echo peak shift measurements have been carried out to study the SD.
Fleming and co-workers studied such three-pulse photon echo from the dye mole-
cule eosin in water. They found that a substantial amplitude (about 60%) of aqueous
solvation occurs within 30 fs. Athree-exponential t (up to 100 ps) to the data of eosin
in water yields time constants of 17 fs (73%), 330 fs (15%), and 3 ps (12%). Analysis
N O O
O
O
C 343
-
P P
P
expt
S

(
t
)
,

C
(
t
)
S
0
Time (ps)

0 0.5 1.0
1.0
0.5
0
Figure 3.13. Comparison of solvation time correlation function S(t) and C(t) for
dye C343 in water. The dashed line shows the experimental result (labeled as expt).
The MDsimulation result is labeled q. Also shown is a simulation for solvation of
a neutral atomic solute with the LennardJones parameters of the water oxygen
atom (S
0
). The experimental data were tted to Eq. (3.9) (using the constraint that
the long-time spectrum matched the steady-state uorescence spectrum) as a
Gaussian component (frequency 38.5 ps
1
, 48% of total amplitude) and a sum of
two exponential components: 126 (20%) and 880 (35%) fs. Adapted with
permission from Nature, 369 (1994), 471. Copyright(1994) Nature Publishing
Group.
of the experimental data led Song and Chandler [15] and Nandi and Bagchi [16] to
attribute this ultrafast solvation to the high-frequency intermolecular vibrational/libra-
tional modes of water, namely the hindered translational band at 180 cm
1
due to the
HB network and the 600 cm
1
band due to libration.
To summarize, the main result of this chapter is that liquid water exhibits a
surprisingly fast polarization response to polar perturbations. This ultrafast response
was not anticipated. This ultrafast response facilitates ionic conductivity and charge
transfer reactions in aqueous solutions, as discussed below.
3.10 Ionic conductivity of rigid ions in water
The ionic conductivity () is an important transport property of electrolyte solution.
This conductivity depends on the concentration of electrolyte in a nontrivial
way. The simplest expression for this dependence is given by the celebrated
DebyeHuckelOnsager law, which is given as follows:
L(c) = L
0
(A BL
0
)
c
_
(3:10)
where A and B depend on the molecular properties and
0
is the conductivity at
innite dilution and is called limiting ionic conductivity (
0
).
The limiting ionic conductivity (
0
) of rigid alkali cations exhibits fascinating
properties. In Figure 3.14 we showthe limiting ionic conductivity of alkali cations
against the inverse of the crystallographic radii at room temperature. Note the
cusp-like maximum near Cs
+
[17] which has a size similar to that of a water
molecule.
The limiting ionic conductivity of a rigid ion is inversely proportional to the
self-diffusion coefcient of the ions. This dependence goes by the name of
Nernsts law of electrochemistry. The Einstein relation relates the diffusion
coefcient to the friction coefcient of the ion (
ion
). In simple terms we have
the following relations.
0
~
1
Dion
(3:11)
D
ion
=
k
B
T
ion
(3:12)
The analysis of
ion
reveals interesting aspects of ion transport in water and provides
an explanation of the cusp-like dependence of limiting ion conductivity on the
inverse ion radius, as detailed below.
It has been observed that the friction on an ion can be approximately described as
a sum of two contributions: a collisional contribution that can be accounted for by
3.10 Ionic conductivity of rigid ions in water 45
the liquid viscosity and a polar (or dielectric) part that arises from interaction
between dipole moments of water and the charge of the ion.
For small ions the polar part of friction can become larger than the viscous
friction. This is responsible for the fall of the ion conductivity with r
1
ion
. There is,
however, a notable difference for water. In many other liquids this part is indeed
large. However, in water this contribution remains modest due to the ultrafast
relaxation in water, as demonstrated by SD experiments. That is, the solvent relaxes
too fast to offer signicant friction from its polarization uctuations.
The above elucidation provides a rather unusual illustration of the control of slow
phenomena (such as ion diffusion) by an ultrafast process such as SD [18].
The story of diffusion of small ions in water is, however, still not complete and the
picture given above is over-simplied. For example, theory would predict a rather
similar size dependence of conductivity for alkali cations and halide anions. In
reality, however, the limiting conductivities of positive (alkali) and negative (halide)
ions lie on different curves when plotted against the inverse radius (or radius).
The peak appears at a larger radius for the anions. In order to explain this result one
needs to consider an accurate interaction potential that differentiates between a
cation and an anion of equal size. Such a study was carried out by Rasaiah and
Lynden-Bell [18], who indeed found that solvation structures around cations and
anions are markedly different for ions of the same size (such as K
=
and Cl
ions).
Figure 3.14. The comparison of the values of the limiting ionic conductivity (
0
) in
water of rigid monopositive ions with the prediction of the molecular theory that
takes into account the ultrafast sub-50 fs SD in liquid water. Here (
0
) is plotted
as a function of the inverse ionic radius (r
1
ion
) in water at 298 K. The solid line
represents the predictions of the microscopic theory. The open circles denote the
experimental results. Here, the tetra-alkylammonium ions are represented by C
1
C
4
,
where C
n
= (C
n
H
2n+1
)4N
+
, n being 1, 2, 3, or 4. Adapted with permission from Acc.
Chem. Res., 31(1998), 181. Copyright (1998) American Chemical Society.
3.11 Electron transfer reactions in water
Because of the large dielectric constant of water, electron transfer reactions in this
liquid are strongly coupled to solvent polarization modes. The equilibrium solvent
effects are well accounted for within the celebrated Marcus theory of electron
transfer reactions [19]. The dynamic effects of electron transfer reactions have
been the subject of many interesting discussions in the scientic literature and
revealed some nice aspects of chemical kinetics in general, as articulated below.
Study of the dynamics of electron transfer uses the results obtained in SD.
For simplicity we consider the transfer of an electron between identical mole-
cules, as in self-exchange reactions, as depicted below
M
+n
+ M
+m
M
+(n+1)
+ M
+(m1)
.
Familiar examples involve self-exchange reactions in metal coordination com-
plexes, such as Co
+2
/Co
+3
or Fe
+2
/Fe
+3
. For such systems, the Marcus theory
shows that the reaction coordinate is the difference in solvation energy between
the reactant (M
+n
, M
+m
pair) and the product (M
+(n+1)
, M
+(m1)
pair) [19]. The
reaction can be envisaged to occur on a potential-energy surface that is characterized
by two minima (one for the reactant and one for the product), separated by a
maximum, like the one shown below. In general, m and n can differ but the main
conclusion discussed below remains unchanged.
Since the electron is coupled to the polarization modes of the solvent molecule,
the rate of the reaction can be affected signicantly by the dynamics experienced
along the reaction coordinate. The dynamics can even be rate-determining if it is
much slower than the time it takes to cross the barrier in the absence of solvent
dynamics effects.
In case of water, as we discussed above, SD demonstrated that the rate of
relaxation of polarization experienced by a charge is extremely fast. An electron
transfer is never as fast as this solvation. Therefore, solvent polarization can follow
the motion of the electron. Thus, in the case of electron transfer in water, dynamic
forces do not retard the reaction, and in fact can accelerate it. We now describe a
theoretical analysis to establish the above, rather surprising, conclusion.
Here we only briey describe the analysis of solvent effects on the electron
transfer reaction, with emphasis on the physical picture. In the study of solvent
dynamic effects on the rate of any chemical reaction (here the rate of an electron
transfer reaction), the usual approach is to invoke a phenomenological description
where the solvent dynamic effect is included through a friction coefcient, (s) [20].
So, the theoretical study involves two steps. First, we derive an expression for the
rate in the presence of solvent forces. Second, we derive an expression for the
friction.
3.11 Electron transfer reactions in water 47
For the expression of the rate in the presence of the frictional forces, we use the
expression derived by Grote and Hynes to calculate the rate of electron transfer [21].
According to GroteHynes theory the rate constant of electron transfer, k
ET
, is given
as follows:
k
ET
=
ET
e
E
k
B
T
(3:13)
where E is the activation energy and
ET
is the barrier-crossing frequency, which is
expressed as
ET
=

r
R
2
(3:14)
where
b
and
R
are two frequencies determined by the reaction potential-energy
surface,
b
is the barrier frequency,
R
is the longitudinal frequency of the solution
governing oscillations in the reactant well, and
r
is the reactive frequency
determined by the self-consistent relation,
r
=

2
b
r
(
r
)
(3:15)
We next need to nd the frequency-dependent friction (s) that we use in Eq. (3.16),
to obtain the barrier frequency. This is obtained from the solvation time correlation
function [22,23].
Figure 3.15. The double-well potential of the electron transfer reaction as depicted
above, from reactant (M
+n
, N
+m
) pair to the product (M
+(n+1)
, M
+(m1)
pair). Here
TS refers to the transition state.
D(s) =
s (s)
2
R
s[s (s)[
(3:16)
where (s) is the reaction coordinate time correlation function in the frequency
domain.
Let us summarize the steps quickly. First, we use the Marcus theory to obtain the
reaction free-energy surface. Second, we adopt the GroteHynes theory to obtain the
reaction rate. The latter needs frequency-dependent friction on the reactive motion,
which is the solvent polarization. Third, we use the solvation time correlation function
to obtain the frequency-dependent friction.
Since the solvation time correlation function is known both fromexperiments and
from computer simulations, we can easily carry out the above exercise. When this is
done, the theory predicts a lack of, or weak, dependence of the electron transfer rate
on solvent dynamics, for weakly adiabatic reactions; the reason being the
dominance of the ultrafast component in SD of water, so the solvent moves too
fast to offer any retardation!
This has been veried in experiments that show a lack of dependence of the rate
on the solvent relaxation time.
What is presented above is really a rather nice and physically appealing picture of
the lack of solvent dynamics effects on a large class of electron transfer reactions in
water where the Marcus theory is accurate.
3.12 Motion becomes collective at low temperature
Molecular motion in dense liquids is restricted because the tagged molecule is
temporarily caged by its surrounding molecules. It often collides with the same
molecules at nearly regular intervals. As the local structure in a liquid has a nite
lifetime which is much longer than the time between two successive collisions,
our tagged molecule remembers its immediate past. As the structural relaxa-
tion time becomes longer as the temperature is lowered, the memory becomes
longer. At the same time, more molecules get involved in the relaxation process
of a local structure. The motion becomes collective.
In liquid water, due to the presence of the HBs between water molecules,
relaxation is collective even at room temperature, which can be taken as between
10 and 30C. Actually, the property of water changes signicantly even across this
range. There is now sufcient experimental and simulation evidence that water
motion becomes signicantly collective near and below 10C. This feature can
have profound biological signicance because the solubility of biopolymers and
many organic molecules depends on temperature through the hydrogen-bond
network of water.
3.12 Motion becomes collective at low temperature 49
In fact, the existence of the density maximum at 4C at ambient pressure is
attributed to the increasing role of correlation between water molecules. As the
temperature is lowered, water molecules try to maximize the strength of the HBs.
As we discussed before, the HB is strongest when the bond is linear. One single
molecule can form four such linear hydrogen bonds and they also form a
hexagonal ring-like structure. These changes in molecular arrangements can
be quantitatively described in terms of two order parameters, namely, the
tetrahedral order parameter t
h
and the orientational order parameter, Q
6
. We
have discussed these parameters in Chapter 2. The tetrahedral order parameter
provides information about the coordination number of each water molecule,
while Q
6
provides information about the local hexagonal order. These order
parameters tell us how closely the local structural arrangement of water
molecules resembles that of ice. As the temperature of water is lowered towards
and below the freezing/melting temperature of water/ice, the larger the values of
these order parameters become. As discussed in Chapter 2, when the temperature
is lowered, 2- and 3-coordinated water molecules are progressively replaced by
the 4-coordinated water molecules. However, below 4C, the 5-coordinated
ones also get converted to 4-coordinated ones, giving rise to the density
maximum. These changes are quantitatively reected in the increase in the
value of the order parameters.
This transformation in water structure shows up as an alteration in the
molecular motion. Thus, molecular motion in low-temperature water becomes
sluggish. There have been intense studies of the dynamics of water below the
freezing temperature, even below 260 K. At this temperature the relaxation times
slow down considerably. When the temperature dependence of various relaxation
times is tted to empirical functional forms, a dynamic transition is predicted
around a temperature of 220 K. There is, however, no convergence on the possible
reason for this transition. Note that experimentally it has not been possible to
supercool water below 230 K.
3.13 Conclusion
The range of dynamic anomalies shown by liquid water at room temperature and
pressure is truly amazing. It is not just the extended hydrogen-bond network that is
responsible for this but also the small lifetime of an HB that allows large-scale
uctuations in liquid water (as discussed in Chapter 1).
While several earlier studies did nd indications of jump motions of water
molecules in the liquid phase, a rather complete dominance of these jump motions
was not expected and came as a surprise. The reason for the jump motion of course
lies in the extended hydrogen-bond network of liquid water. Fortunately in this case
the origin and the mechanism of the anomalous dynamics have been understood.
It is now expected that molecular motions of water in different restricted systems
may show signatures of the above anomaly. This has been veried already for the
protein hydration layer. But further studies are required in other systems.
APPENDI X 3. A ROTATI ONAL TI ME CORRELATI ON
FUNCTI ONS
The dynamic response functions can be expressed as integrals over time correlation
functions of the relevant quantities. The time correlation function of a dynamic
variable is dened in the following way. Let A(t) be the value of a dynamic variable
at time t. Then the time correlation function C
AA
(t) is dened by the following
expression
C
AA
(t) =
A(0)A(t)
_
= lim
T~
1
T
_
T
0
dsA(s)A(t s) (3:A:1)
The above expression denes a time average. The average is taken over a time
trajectory of the system. Properties of time correlations functions have been
discussed by Berne and Pecora [25].
In a liquid, as a result of inter-molecular interactions, molecules are continuously
rotating and translating. Thus, if we could tag a molecule and study its detailed
motion, we would nd it executing a randomBrownian motion not only in the three-
dimensional positional space, but also a similar motion in the three-dimensional
orientational space. For simplicity, let us rst consider the motion of a tagged
prolate-shaped molecule in a solvent of spherical molecules, as shown in
Figure 3.A.1 below.
We shall refer to this simple model as an ellipsoid in a sea of spheres (EISS). In
this model, both the position and the orientation of the tagged molecule undergo
continual change due to the interactions with the surrounding solvent molecules,
For a molecule of ellipsoidal symmetry, we need two angles ( and ) to denote the
orientation in the space xed frame. We shall denote the two angles by an orienta-
tion vector in this chapter. Both the position vector r and the orientation vector
execute small-amplitude Brownian motions in the respective conguration space.
Quantitative knowledge about the rate of orientation of anisotropic molecules in
liquid is essential to understand many relaxation processes and chemical reactions.
A large number of experimental techniques have been developed and employed to
understand the details of the molecular orientational process in liquids. As expected,
3.13 Conclusion 51
there are features which are common to both translational and rotational motions.
There are also many features which are quite distinct. In general, orientational time
correlation functions decay at a timescale faster than the corresponding time
correlation functions of the linear motion (like dynamic structure factors). For
example, the time constant of decay of the orientational time correlation function
of a tagged water molecule is 2.6 ps, during which a water molecule diffuses only
20% of its molecular diameter, which is quite meager. However, we shall describe
later howtranslational motions can deeply affect the rates of orientational relaxation
in dense dipolar liquids, in one of the most interesting aspects of orientational
relaxation.
One can gather a considerable amount of insight by studying the orientational
motion of molecules in computer simulations by monitoring both the spatial and the
orientational trajectories of the tagged molecule in simple model systems, like our
model of the isolated ellipsoid in the sea of spheres (EISS) discussed above. At
long times, the motions are diffusive and the rates of displacement in both the spaces
can be each described by a diffusion equation, with respective diffusion coefcients,
@
S
(r; t)
@t
= D
T
\
2
s
(r; t) (3:A:2)
@
S
(; t)
@t
= D
R
\
2
s
(; t) (3:A:3)
Figure 3.A.1. Motion of a prolate-shaped molecule in a solvent of spherical
molecules.
where
s
(r,t) and
s
(,t) are position- and orientation-dependent single particle
densities, respectively, dened as
s
(r; t) = rr
s
(t) ( )) (3:A:4)
s
(; t) =
s
(t) ( )) (3:A:5)
D
T
and D
R
are the rotational and translational diffusion coefcients, respectively.
The analogy between translational and rotational motion can be extended further.
We can dene not only position (r) dependent isotropic collective number density,
(r,t) but also position (r) and orientation () dependent collective orientational
density, (r,,t). These collective quantities are different from tagged (single)
particle densities, as they count not just the tagged but all the molecules in the
system present in a volume element around r and .
(r; ; t) =
i
r r
i
(t) ( )
i
(t) ( )
_ _
(3:A:6)
We would like to emphasize, at the very outset, the difference between the single-
particle and collective quantities because (i) different experiments measure different
quantities, (ii) they can have vastly different dynamics, and (iii) this fact is some-
times overlooked.
Orientational relaxation plays a key role in many relaxation processes, such as
polarization and DR, SD and quadrupolar relaxation. It profoundly inuences the
dynamics of the many important chemical reactions, such as the electron and proton
transfer reactions in a polar liquid. The orientational relaxation of an anisotropic
molecule in liquid clearly depends on the density and the temperature of the liquid
and also on the nature of the anisotropy of interaction potential. It may also be coupled
to the translational modes of the liquid and the internal modes of the molecule.
Orientational relaxation in a liquid can show rich and diverse dynamic behavior.
However, orientational motions are studied directly and indirectly by many
different experimental techniques. There are several issues that arise in the study
of orientational relaxation. For example, the relationship between the orientational
relaxation of a single molecule and the collective relaxation, where the orientation of
many molecules is probed, remains a frontier problem in the area of orientational
relaxation. Many experiments, such as NMR relaxation, incoherent neutron scattering
and Raman line shapes, are sensitive to the single-particle orientational motion. The
other limit of collective orientational motion involving all the molecules of the systemis
probed by experiments such as DR, depolarized light scattering, and coherent neutron
scattering. It has not yet been possible to successfully and directly study collective
relaxation involving a limited number of molecules (say, of the order of 10100).
3.13 Conclusion 53
However, indirect information about this intermediate limit can be obtained from such
linear spectroscopic techniques as TDFSS measurement of newly created ions or
dipoles or such nonlinear optical techniques as Kerr relaxation. In these measurements,
some moments of the inverse distances of the solvent molecules from the probe over
time and space-dependent solvent polarization are studied. If the orientational
relaxation of the nearest-neighbor molecule is signicantly different from those that
are far off in the bulk, then these indirect methods can provide reliable information on
collective orientational relaxation in the intermediate regime.
However, orientational time correlation functions are dened for the single
particle as
C
s
l
(t) =
P
l
(ê
i
(0):ê
i
(t)))
P
l
(ê
i
(0):ê
i
(0)))
(3:A:7)
And similarly for collective quantitity,
C
s
l
(t) =
j
P
l
(ê
i
(0):ê
j
(t)))
j
P
l
(ê
i
(0):ê
j
(0)))
(3:A:8)
where
i
is the unit orientation vector along the major axis of the ith molecule, and P
l
is the Legendre polynomial of order l.
The collective orientational relaxation in the long-wave length limit involving all
the molecules of the system was studied initially by borrowing concepts from the
continuum description of electrostatics. In these theories, the dipolar molecule is
replaced by a cavity of some simple shape and with a point dipole at its center and
the liquid is replaced by a frequency-dependent dielectric continuum. The interaction
between the rotating dipoles and the bulk of the liquid is included through a time-
dependent reaction eld, which arises from the electric polarization of the solvent by
the rotating dipole. The reaction eld is obtained by a quasi-stationary boundary value
calculation. This eld gives rise to a dielectric friction, which retards the rotation of the
dipole and gives rise to a non-exponential decay of the dipolar correlation function.
The classic example of such a theory is the work of Nee and Zwanzig [24].
The advantage of the continuum model theories [24] is that they provide simple
expressions for the orientational correlation functions that can be tested against
experiments. This is especially true for complex liquids where a microscopic theory
is bound to be complicated. The main drawback of the continuum model theories is
that they ignore the intermolecular correlations that are present in a dense dipolar
liquid. If these correlations are important in a relaxation process, then the continuum
model is obviously inadequate.
Dynamic light scattering: probe of density uctuation at long length scales
Among several experimental procedures light-scattering experiments provide
important information about many properties of the liquid. Light scattering occurs
due to local uctuations in the dielectric constant of the medium. These uctuations
arise due to the fact that the small molecules in solutions undergo Brownian motion
and so the distance between them in the solution is constantly varying with time. As
a result, the number of molecules in a small volume element uctuates in time and
also with location. Such uctuation in density gives rise to uctuations in the
dielectric constant which can be measured by determining the scattering of light
or neutrons, by choosing the appropriate experimental method.
Let us consider a medium with uctuating local dielectric constant
(r; t) =
0
I (r; t) (3:A:9)
where
0
is the average static dielectric constant of the liquid and (r,t) is the
dielectric constant uctuation tensor at a position r and time t, and I is the second-
rank unit tensor.
An incident electric eld can be given as
E
i
(r; t) = n
i
E
0
exp i k
i
:r
i
t ( ) [ [ (3:A:10)
where n
i
is a unit vector in the direction of the incident electric eld, E
0
is the electric
eld amplitude, k
i
is the wave vector and
i
is the angular frequency of the incident
electric eld. The corresponding scattered electric eld of wave vector k
f
and
angular frequency
f
, at a large distance R from the scattering volume, due to
interaction of the electric eld and local dielectric constant (r,t), is given by [1]
E
s
(R; t) =
E
0
4R
0
e
ik
f
R
_
v
d
3
r exp i q:r
i
t ( ) n
f
: k
f
k
f
(r; t):n
i
( ) ( ) [ [ [ [
(3:A:11)
where q = k
i
k
f
. The integration in the above expression is over the scattering
volume.
In light-scattering experiments one measures the spectral density of the electric
eld autocorrelation function of the scattered light wave, I
E
(), given as
I
E
(R; ) =
1
2
_
dE
s
(R; t)E
s
(R; t ))e
i
(3:A:12)
The spectral density of the light scattered by the medium is obtained by using
Eq. (3.A.11) and Eq. (3.A.12) is given by
3.13 Conclusion 55
I
if
q;
f
; R ( ) =
I
0
k
4
f
16
2
R
2
0
2
_ _
1
2
_
dt
if
(q; 0)
if
(q; t))exp i
f

i
( )t ( )
(3:A:13)
where I
0
|E
0
|
2
;
if
(q,t) = n
f
. (q,t).n
i
is the component of the dielectric constant
uctuation tensor along initial and nal electric eld directions. As is clear from
Eq. (3.A.13) the light scattering event that produces the wave vector change q and
frequency shift is entirely due to dielectric constant uctuations.
A light-scattering experiment measures the long-wavelength (k 0 limit)
response of a liquid and therefore it measures the collective density relaxation;
that is, the scattering cross-section derives contributions from a large number of
molecules. The spectrum obtained by dynamic light-scattering experiments at long
wavelengths and low frequencies is known as the RayleighBrillouin spectrum. It
provides a wealth of information about many properties of the liquid.
Like the spatial counterpart, orientational density uctuations also lead to light
scattering. Orientational uctuations are best studied by depolarized (VH) Rayleigh
scattering, because pure transverse uctuations are observed in this geometry and
the spectrum is less complicated. In general, the evaluation of the correlation
function is complicated. Since light scattering involves only small wave-vector
processes, the translational contribution to orientational relaxation can be ignored
and the error that is involved in assuming statistical independence of molecular
rotation and translation may not be signicant. If we further assume that interactions
between different rotating molecules are negligible, then the correlation function
that is measured by the light-scattering experiments corresponds to only single-
particle motion. The relevant correlation functions for several different models have
been given by Berne and Pecora [25].
The assumption of no interaction between the rotating molecules is not correct
in a dense dipolar liquid where the molecules may possess a dipole moment in
addition to being polarizable. In such a situation, dynamic light-scattering
experiments measure the collective property. The polarizability can be expanded
in terms of spherical harmonics and, for a molecule of ellipsoidal or cylindrical
symmetry, the only terms that appear are Y
2m
. So the dynamic quantity is Y
2m
(k, t)
and the correlation functions, C
2m
(k, t), are dened as
C
2m
k;t ( ) = Y
2m
k; 0 ( )Y
2m
k;t ( )): (3:A:14)
In the long-wavelength limit, the quantities of interest are C
2m
(k = 0, t), which can
be signicantly different from the single-particle correlation function dened as
S
2m
t ( ) = Y
2m
0 ( ) ( )Y
2m
t ( ) ( )) (3:A:15)
where (t) is the orientation of the dipole at time t.
An excellent introduction to the theory of dynamic light scattering and the
relevant correlation functions is given by Berne and Pecora [25].
Magnetic resonance experiments: probe of single-particle dynamics
Nuclear magnetic relaxation due to magnetic dipoledipole interactions in a liquid
can provide direct information about both the orientational relaxation time of a
rotating molecule and also the translational diffusion coefcient, as discussed
below. In nuclear magnetic resonance experiments, usually single-particle relaxa-
tion is measured. The spin-lattice relaxation time and the spin-spin relaxation time
are essentially the time integrals over various components of S
2m
(t). There is an
intriguing point about NMR relaxation that deserves special attention. Although
NMRexperiments measure time constants in the second or millisecond range, NMR
can provide accurate information about rotation time constants of molecules that are
in the picosecond range. We elucidate the reason below.
In NMR, the dipolar correlation function for the case of two interacting nuclear
spins separated by distance r
12
is dened as [26]
G(t) =
3
5
0
4
_ _
2
h
2
2
1
2
2
r
12
6
S
2m
t ( ) (3:A:16)
where S
2m
(t) is the single-particle correlation function dened in Eq. (3.A.15),
1
and
2
are the gyromagnetic ratios of the nuclei and r
12
is the distance between them.
For a simple Debye model of rotational diffusion Eq. (3.A.16) reduces to the
following simple equation [26]:
G(t) =
3
20
0
4
_ _
2
h
2
2
1
2
2
r
12
6
exp t=
c
( ) (3:A:17)
with the rotational correlation time
c
c
= a
2
=6D
R
; (3:A:18)
where a is the radius of the molecule and D
R
is the rotational diffusion constant,
which is related to the viscosity, , by the StokesEinstein relation as
D
R
=
k
B
T
8a
3
: (3:A:19)
For water at room temperature
c
is around 3 ps. This is a very short correlation time
and in such cases the relaxation time is given by
3.13 Conclusion 57
1
T
1
=
3
2
0
4
_ _
2
h
2
2
1
2
2
r
12
6

c
: (3:A:20)
For water this relaxation (relaxation from the intramolecular interaction in the
water) time is 4.78 s. The above equation shows that even though the time observed
in NMR experiments could be in the second or millisecond range, the microscopic
correlation time
c
can be in the picosecond range.
In the above we have discussed only two experimental procedures for the
evaluation of orientational relaxation time. However, there are other experimental
tools also present to determine the orientational dynamics of a liquid. A detailed
discussion is presented in ref. [20].
APPENDI X 3. B QUANTI FI CATI ON OF HYDROGEN- BOND
LI FETI ME DYNAMI CS
The lifetime of an HB is usually described in terms of two HB lifetime correlation
functions, denoted by C
HB
(t) and S
HB
(t), which are dened by the following
expressions [4]:
C
HB
t ( ) =
h 0 ( )h t ( ))
h)
(3:B:1)
S
HB
t ( ) =
h 0 ( )H t ( ))
h)
(3:B:2)
where h(t) is called the HBlifetime function, which is unity if the HBbetween a pair
of water molecules is intact at time t and zero if it is broken. On the other hand, h(t) is
unity only if the tagged bond has remained continuously intact from time t = 0 to the
present time t. This process of breaking and making of HBs in liquid water can be
probed indirectly by a variety of experimental techniques, and quantitative informa-
tion can be obtained from computer simulations.
From the above lifetime correlation functions, one can dene the time-dependent
rate of the HB breaking events from their computer simulations using the following
equation [6]
k(t) = h(0)[1 h(t)[H(t))=h) (3:B:3)
where, h 0 ( )= dh=dt ( )
t=0
Here k(t) is the average rate of change of HB population for those trajectories
where the bond is broken at time t later. Figure 3.11 shows k(t) determined fromtheir
simulation. At short times, k(t) manifests an assortment of motions leading to HB
breaking. The most important of these motions are librations, at timescale of less
than 0.1 ps, and inter-oxygen vibrations on the timescale of 0.10.2 ps. Beyond
this transient period, k(t) decays monotonically. To the extent that each HB
acts independently of other HBs and other processes of similar timescales, the
long-time decay of the k(t) would follow rst-order kinetics. That is, one would
expect k(t) v exp(vt), where 1/v would be the average HB lifetime.
References
1. R. Brown, A brief account of microscopical observations made on the particles con-
tained in the pollen of plants. Phil. Mag., 4 (1828), 161173.
2. A. Einstein, ber die von der molekularkinetischen Theorie der Wrme geforderte
Bewegung von in ruhenden Flssigkeiten suspendierten Teilchen. Annalen der Physik,
322 (1905), 549560.
3. P. Debye, Zur Theorie der anomalen Dispersion im Gebiete der langwelligen elektrischen
Strahlung. Berichte der deutschen Physikalischen Gesellschaft, 15 (1913), 777793.
4. B. Bagchi, Water dynamics in the hydration layer around proteins and micelles. Chem.
Rev., 105 (2005), 31973219.
5. D. Laage and J. T. Hynes, Amolecular jump mechanismof water reorientation. Science,
311 (2006), 832835.
6. D. E. Moilanen, E. E. Fenn, Y. Lin, J. L. Skinner, B. Bagchi, and M. D. Fayer, Water
inertial reorientation: hydrogen bond strength and the angular potential. Proc. Natl.
Acad. Sci. USA, 105 (2008), 52955300.
7. A. Luzar and D. Chandler, Hydrogen-bond kinetics in liquid water. Nature, 379 (1996),
5557.
8. S. Balasubramanian, S. Pal, and B. Bagchi, Hydrogen-bond dynamics near a micellar
surface: origin of the universal slow relaxation at complex aqueous interfaces. Phys.
Rev. Lett., 89 (2002), 115505.
9. D. W. Oxtoby, Dephasing of molecular vibrations in liquids. Adv. Chem. Phys., 40
(1979), 1; R. J. Kubo, Stochastic Liouville equations. Math. Phys., 4 (1963), 174.
10. E. T. J. Nibbering and T. Elsaesser, Ultrafast vibrational dynamics of hydrogen bonds in
the condensed phase. Chem. Rev., 104 (2004), 18871914. (b) J. C. Deak, S. T. Rhea,
L. K. Iwaki, and D. D. Dlott, Vibrational energy relaxation and spectral diffusion in
water and deuterized water. J. Phys. Chem. A, 104 (2000), 4066. (c) R. Rey and
J. T. Hynes, Vibrational energy relaxation of HOD in liquid D2O. J. Chem. Phys., 104
(1996), 2356. (d) C.P. Lawrence and J.L. Skinner, Vibrational spectroscopy of HOD in
liquid D
2
O. III. Spectral diffusion, hydrogen bonding and rotational dynamics. J. Chem.
Phys., 118 (2003) 254.
11. A. Chandra and B. Bagchi, Relationship between microscopic and macroscopic orien-
tational relaxation times in liquids. J. Phys. Chem., 94 (1990), 31523156.
12. B. Bagchi, Water solvation dynamics in the bulk and in the hydration layer of proteins
and self-assemblies. Annu. Rep. Prog. Chem., Sect. C, 99 (2003), 127175.
13. B. Bagchi, D. W. Oxtoby, and G. R. Fleming, Theory of the time development of the
Stokes shift in polar media. Chem. Phys., 86 (1984), 257267.
14. R. Jimenez, G. R. Fleming, P. V. Kumar, and M. Maroncelli, Femtosecond solvation
dynamics of water. Nature, 369 (1994), 471473.
15. X. Song and D. Chandler, Dielectric solvation dynamics of molecules of arbitrary shape
and charge distribution. J. Chem. Phys., 108 (1998), 2594.
References 59
16. N. Nandi and B. Bagchi, Anomalous dielectric relaxation of aqueous protein solutions.
J. Phys. Chem. A, 102 (1998), 8217.
17. B. Bagchi and R. Biswas, Ionic mobility and ultrafast solvation: control of slow
phenomena by fast dynamics. Acc. Chem. Res., 31 (1998), 181187.
18. J. C. Rasaiah and R. M. Lynden-Bell, Computer simulation studies of the structure and
dynamics of ions and non-polar solutes in water. Philos. Trans. R. Soc. Lond., 359
(2001), 15451574.
19. R. A. Marcus, On the theory of electron-transfer reactions VI. Unied treatment for
homogeneous and electrode reactions. J. Chem. Phys., 43 (1965), 679701.
20. B. Bagchi, Molecular Relaxation in Liquids (New York: Oxford University Press,
2012).
21. R.E Grote and J.T. Hynes, The stable states picture of chemical reactions II. Rate
constants for condensed and gas phase reaction models. J. Chem. Phys., 73 (1980),
2715.
22. A. Chandra, Electron transfer reactions in electrolyte solutions: effects of ion atmo-
sphere and solvent relaxation. Chem. Phys. Lett., 253 (1996), 456462.
23. D. A. Zichi, G. Ciccotti, J. T. Hynes, and M. Ferrario, Molecular dynamics simulations
of electron-transfer reaction in solution. J. Phys. Chem., 93 (1989), 62616265.
24. T. Nee and R. Zwanzig, Theory of dielectric relaxation in polar liquids. J. Chem. Phys.,
52 (1970) 6353.
25. B. Berne and R. Pecora, Dynamic Light Scattering (New York: John Wiley, 1976).
26. B. Cowan, Nuclear Magnetic Resonance and Relaxation (Cambridge: Cambridge
University Press, 1997).
4
Inherent structures of liquid water
Inherent structures of a liquid are static molecular arrangements that
are obtained through computer simulations of a parent high-
temperature liquid by removing the kinetic energy of the molecules in
such a fashion that the resulting structures each correspond to local
potential-energy minima. Structures that are similar to each other are
said to belong to the same basin. At low temperatures, the observed
properties of the parent liquid may be described in terms of the excita-
tion of these inherent structures. Thus, inherent structures provide a
useful platform to describe the structure and dynamics of complex
liquids, particularly at low temperatures. For water, distinct inherent
structures involve different hydrogen-bond connectivity among mole-
cules. Inherent structures provide valuable information about the pre-
sence of such defects as 3- and 5-coordinated water molecules in a
network of 4-coordinated water molecules. Inherent structures can
provide information about collective excitations of the liquid, and
temperature-dependent exploration of the energy landscape of the
liquid.
4.1 Introduction
In a dense liquid, although there is no long-range order, molecules are locally
ordered because they are required to pack to a high density. However, there are
many possible molecular arrangements which are of slightly different energies, and
a few arrangements of signicantly smaller energies. These packings are also of
different entropy. At low temperature, low-energy and low-entropy congurations
are more inuential in determining the equilibrium and dynamic properties of a
liquid. Molecular arrangements in these low-energy structures can be studied by
inherent structure analysis and this computer-intensive method now provides a
unique way to capture the local structure or molecular arrangement present in a
disordered liquid [1].
61
In liquid water, the local molecular arrangement can be understood from the
hydrogen-bond connectivity of water molecules. The number and nature (distance,
energy, angles between two bonds) of these bonds provide a more detailed descrip-
tion of local structure than the radial distribution function, as already described in
Chapter 2. In a normal, high-temperature liquid, these bonds are continually break-
ing and forming. The transient nature of these bonds and also inter-molecular
vibrations obscure the stable molecular arrangements. Inherent structures capture
(at least partially) the static, quasi-stable, molecular packing that can form the
backbone of a liquid structure. Of course there are a large number of such
arrangements.
Spatially different local arrangements or packing can be identied and character-
ized by studying relative molecular arrangements or packings that are close to each
other. One way to do so is to nd the nearest potential-energy minimum of the
respective structures. The structures/arrangements that are close to each other may
belong to the same potential-energy minimum. This unique ground-state structure
gives rise to different structures on excitation of the inherent structure (IS).
Aliquid at a given pressure and temperature can have many such ISs and together
they dene a potential-energy landscape to provide a global viewof liquid structure.
Transitions between ISs (that is, potential-energy minima) can be used to understand
dynamics in liquid or other complex systems.
It is important to understand further the advantages of IS. When we compute or
measure the radial distribution function of a liquid, we get an average measure of the
short-range order but no detailed information about molecular packings present in
the liquid. Different packings, on the other hand, can have different energies that can
be used to characterize or group them. Low-energy packing dominates properties as
temperature is lowered. That is, exploration of the energy landscape is temperature-
dependent. The potential minima corresponding to the ISs can be found in computer
simulations by using fairly sophisticated numerical techniques.
This technique of studying liquid structures was introduced by Stillinger and
Weber in the early 1980s [1,2]. Soon this method was used to study supercooled
liquids and glasses, protein folding, and liquid crystals, to name a few applications.
A schematic representation of the underlying potential-energy landscape of liquid is
shown in Figure 4.1.
If the system has several moderately stable congurations along with the global
one, the landscape is categorized as rugged. Then the motion of the system search-
ing for the global minimum becomes relatively slower according to the following
expression of diffusion derived by Zwanzig [3]:
D
+
= D exp ( k
B
T)
2
_ _
(4:1)
62 Inherent structures of liquid water
Here, D* is the effective diffusion coefcient, D is the diffusion coefcient with no
ruggedness in the potential-energy surface, and is the measure of ruggedness.
Actually, is the standard deviation of a Gaussian distribution of the energy of the
system. This motion gets even slower as the temperature is reduced because then it
feels more obstructions (barrier height) in transforming from one minimum to the
other.
Equation (4.1) is particularly useful to understand the slow dynamics of
liquids at low temperatures. Here relaxation is envisaged to occur by transitions
between different minima (see Figure 4.2) corresponding to different ISs.
Relaxation is slow due to both barrier crossing and diffusion in this rugged
landscape.
In Figure 4.3, we show representative snapshots of the ISs of liquid water at four
different temperatures obtained through computer simulation. Note the vast differ-
ence in the molecular arrangement between 300 K and 210 K. At high temperature
the hexagonal arrangement of ice is infrequent. However, at 210 K the low-density
arrangement is clearly evident with the presence of a large number of cavities. We
should point out that these are representative snaps, as a liquid passes through many
ISs during its time trajectory.
Figure 4.1. The potential-energy landscape of a liquid showing the energy minima,
maxima (barriers), basins, and the transition states of a liquid. In this gure, the
x-axis denotes schematically the conguration of the system and the y-axis the
energy of the IS. A distinct conguration of the system is given by the coordinates
of the atoms and molecules of the liquid whose IS is being plotted here. A basin
(just as in the case of a river basin) is a shallow minimum into which nearby states
would fall if we remove the thermal energy of all the molecules. This can also be
called the basin of attraction. Some basins are deeper than others. Adapted with
permission from Nature, 410 (2001), 259. Copyright (2001) Nature publishing
group.
4.1 Introduction 63
Figure 4.2. One-dimensional schematic representation of the sequential quenching
along a trajectory. Instantaneous congurations are quenched into the
corresponding local energy minima (ISs) on the total potential-energy surface.
Inherent structures are extracted from the trajectory at time t = t
1
, t
2
, t
3
, t
4
, etc.
Figure adapted with permission from Chem. Rev., 93 (1993), 2545. Copyright
Figure 4.3. Snapshots of ISs of liquid water at (a) 210 K, (b) 230 K, (c) 250 K, and
(d) 300 K, obtained via computer simulations. The gure is adapted from a
personal communication with Professor Shinji Saito.
We can envisage the similarities between different ISs by analyzing the distance
from the distance matrix between two ISs at time t
n
and t
n
:
R(t
n
; t
/
n
)
2
= [Q(t
n
) Q(t
/
n
)[
2
(4:2)
where Q(t
n
) denotes the positions of all the atoms of the liquid at time t
n
. R(t
n
, t
n
)
gives a measure of the distance of the system at times t
n
and t
n
. Thus, Eq. (4.2)
provides a symbolic representation, meant to capture the motion of the entire system
in a two-dimensional representation. As the difference between t
n
and t
n
increases,
R(t
n
, t
n
) varies. When the systemresides in a mimimum, R shows the motion typical
of a bound state, moving around the same state point. The region between two
minima is called a neck; the reason for the nomenclature is evident from
Figure 4.4.
A gure like Figure 4.4 is particularly useful to study ISs as the distance matrix
reveals the trapping of the system as the system oscillates around different ISs.
Much of the high-frequency change in the total energy of the system is due to
thermal motion and can thus be eliminated by IS study. In Figure 4.5 we compare the
potential-energy change along the dynamic trajectory for both the parent structure
and the corresponding IS.
The large uctuation in IS energy indicates the transition between different ISs.
Such structural changes become shrouded due to the thermal noise in the parent
Figure 4.4. The distance matrix from Eq. (4.2) for ISs of liquid water at 230 K. The
matrix is marked with a shaded square in the gure to show the distance between
the ISs Q(t
1
) and Q(t
2
). Each square is shaded with a different degree of darkness
denoting the number of times the same value R(t
1
, t
2
) was traced. That is, the darker
is the shade, the deeper is the minimum. Figure adapted from a personal
communication with Professor Shinji Saito.
4.1 Introduction 65
structure. Note the longer frequency variations along the time trajectory in the IS.
The latter is more useful than the former in understanding the slow dynamics of the
liquid.
4.2 Transition between inherent structures of water
As shown in the above picture, there are many ISs possible for a liquid and so for
water. Water can make transitions between these ISs and the dynamics in water can
be correlated with these transitions [5].
Note that it is impossible to identify the transitions unless the quenching time step
is much less than 1 ps. This is in distinct contrast to the dynamics observed in a
20
10
0
10
20
0 5 10 15 20
T
o
t
a
l

P
o
t
e
n
t
i
a
l

E
n
e
r
g
y

/

k
c
a
l
/
m
o
l
Times / ps
(a)
20
10
0
10
20
0 5 10 15 20
T
o
t
a
l

P
o
t
e
n
t
i
a
l

E
n
e
r
g
y

/

k
c
a
l
/
m
o
l
Times / ps
(b)
Figure 4.5. Total potential-energy uctuation of the system evaluated from a 28-ps
trajectory from a system containing 64 water molecules: (a) the instantaneous
structure energies (real energies along the trajectory) averaged over some time
intervals with t = 10 fs; (b) the corresponding IS energies with 10-fs intervals.
Figure adapted with permission from Chem. Rev., 93 (1993), 2545. Copyright
dense LennardJones liquid, where such discrete discontinuous changes are not
present [6]. These discrete changes along a trajectory reect the breaking and
formation of HBs and capture the hydrogen-bond rearrangement dynamics of liquid
water.
4.3 Connected water cluster moves during transition
At a relatively low temperature, it was found that the set of water molecules that
undergo largest displacements in a given time interval usually form a cluster of
hydrogen-bonded molecules. The observed clustering phenomenon characterizes
the IS transitions in water and can be interpreted as the analog of the string-like
motion observed in simple atomistic liquids and connected to the presence of
dynamic heterogeneities [7]. Similar results were found by Ohmine and Saito
using the TIP4P and SPC models for water [4].
4.4 HB network restructuring
Inherent structure analysis provides useful insight into the correlated hydrogen-
bond dynamics of water. Many experiments suggest that this tetrahedral HB net-
work has defects, such as an extra (fth) molecule in the rst coordination shell, or
3-coordinated species. Indeed, such 5-coordinated molecules have been directly
identied in simulations and the defects were found to be a catalyst for initiating
motion in the system, making an obvious possible connection between network
defects promoting diffusion and the basin transitions that give rise to diffusive
motion of that system [8].
To obtain a physical picture of the IS transitions for water and hopefully to better
understand the source of these transitions, let us now focus on the details of small
changes in the HB network along a minimum-to-minimum trajectory.
Analysis of the HB network based on IS congurations has shown that local PES
minima contain both linear bonds (LBs) and bifurcated bonds (BBs) whose fraction
is both temperature- and density-dependent. The HB network tends to form a
random but nearly tetrahedral network (no bifurcated bonds) on cooling, or on
lowering the pressure.
To quantify the changes in bonding characteristics, it is necessary to employ a
reasonable denition of HBformation, as discussed before. Previous work indicated
that several denitions provide physically reasonable and compatible results, so
precise denition is not important for the present discussion. The most widely used
denition is that two molecules are bonded if their oxygenoxygen distance is less
than 3.5 and their mutual potential energy is negative.
4.4 HB network restructuring 67
Distribution of bifurcated hydrogen-bond (BB) energies in ISs of water is found
to be bimodal with peaks at roughly 6 kJ/mol and 22.5 kJ/mol, while the LB
energy distribution is unimodal with a peak at roughly 24 kJ/mol. Therefore, the
energy associated with a change in the HB network for losing one LB and creating
two BBs ranges roughly from 21 kJ/mol to 12 kJ/mol, depending on which of the
two possible BBs are created.
The mechanismof such change in liquid water was explored by Laage and Hynes
as already discussed in detail in Chapter 3 [9]. The relative intensities of the peaks of
the BBenergy distribution suggest that such a mechanismwould more likely lead to
an increase in the overall energy. It has been hypothesized that the observed changes
involving an increase in E
IS
found during the IS trajectory are due to processes
LBBB, while changes associated with a decrease in E
IS
are due to the processes
BBLB.
One can of course calculate the distributions of LBs and BBs of ISs just before
and just after positive and negative jumps in the potential energy and study the
changes associated with energy changes greater than 9 kJ/mol to conrm or reject
the above hypothesis (that large jumps in the IS energy are due to bond-breaking
events). During an increase in potential energy, the average number of LBs
decreases while the number of BBs increases; the opposite situation occurs when
the potential energy decreases.
The distribution for BBs exhibits peaks at even numbers of BBs, which we expect
since for each LBlost, two BBs appear. This also implies that the distribution should
be zero for odd numbers of BBs. The anticorrelation of the BB and LB changes is a
strong indicator that there is a mechanism whereby the system accesses higher-
energy states via LBBB transitions. This mechanism is an important aspect of the
hydrogen-bond dynamics of water.
4.5 Coordination number uctuation in inherent structure
and corresponding dynamics in parent liquid
To reinforce the hypothesis that the basin change is associated with a restructur-
ing of the local connectivity, it was shown that the number of molecules with a
coordination number equal to 3, 4, or 5 is a function of time for a characteristic
time interval and these data were compared with the time dependence of <r
2
(t)>.
A clear anticorrelation is observed between the time dependence of the number
of threefold and vefold-coordinated molecules and the time dependence of
the fourfold-coordinated molecules, supporting the proposed interchange
mechanism. The fact that increases in the fraction of BBs coincide with
changes in <r
2
(t)> supports the expectation that motion is a result of network
imperfections.
4.6 Low-energy excitations in liquid water
Extended hydrogen-bond networks sustain certain low-energy (low-frequency)
excitations in liquid water not present in other, non-hydrogen-bonded liquids.
These excitations play an important role in chemical reactions in liquid water,
such as electron and proton transfer reactions. At roomtemperature these excitations
are short-lived because they are strongly coupled to other modes of liquid water.
Information about excitations in the condensed phases is obtained by diagonaliz-
ing the force constant matrix. This force matrix is obtained by taking the derivative
of the interaction energy of each molecule with respect to displacements of atomic
and/or molecular positions of a central molecule. Thus, for an atomic liquid such as
argon consisting of Natoms, the force matrix is a 3 N matrix. If the low-frequency
excitations can be approximated as harmonic, like the normal modes of a solid, then
the eigenvalues and eigenfunctions obtained by diagonalizing the force matrix
provide a useful description of the excitations of the system being studied.
For liquid water, some of the low-frequency excitations are sufciently long-
lived to merit attention. These are librations of the hydrogen atoms around HBs,
centered on the 600 cm
1
, O O oscillatory motion where the distance between
two oxygen atoms is modulated in a nearly harmonic fashion. This is centered on
200 cm
1
. The third one is a collective bending mode formed by three oxygen atoms
of molecules that are hydrogen-bonded. This is shown in Figure 2.8. The average
frequency of this mode is 50 cm
1
.
These normal modes can be identied by quenching liquid water (that is, remov-
ing the kinetic energy of individual water molecules in a sensible fashion), and
thereby removing the kinetic energy of all the molecules. The modes so obtained are
called quenched normal modes. Therefore, these modes can also be considered as
excitations of the ISs.
4.7 Conclusion
The emergence of the method of IS analysis has provided us with a valuable tool to
study the structure and dynamics of a liquid at low temperature. In the case of water,
which is structured even at room temperature, IS analysis provides valuable insight.
In particular, the existence of the 5- and 3-coordinated long-lived defects and its role
in relaxation processes are made clear through the IS analysis.
Unfortunately, however, it has not been fully possible to use the information from
ISs to explain the dynamics of a liquid. For example, attempts to obtain the diffusion
coefcient of the parent liquid from its IS have not succeeded, for reasons whose
detailed discussion is beyond the scope of this monograph. In a nutshell, ISs do not
contain information about dynamic correlations among molecules that control
4.7 Conclusion 69
diffusion in a liquid, particularly at lowtemperatures. This has somewhat limited the
utility of the IS approach.
References
1. F. H. Stillinger and T. A. Weber, Dynamics of structural transitions in liquids. Phys. Rev.
A, 28 (1983), 2408.
2. F. H. Stillinger and T. A. Weber, Point defects in bcc crystals: structures, transition
kinetics, and melting implications. Science, 225 (1984), 983; J. Chem. Phys., 81
(1984), 5095.
3. R. Zwanzig, Diffusion in a rough potential. Proc. Natl. Acad. Sci. USA, 85 (1988), 2029.
4. I. Ohmine and S. Saito,Water dynamics: uctuation, relaxation, and chemical reactions
in hydrogen bond network rearrangement. Acc. Chem. Res., 32 (1999), 741.
5. N. Giovambattista, F. W. Starr, F. Sciortino, S. V. Buldyrev, and H. E. Stanley, Transitions
between inherent structures in water. Phys. Rev. E, 65 (2002), 41502.
6. T. B. Schrder, S. Sastry, J. C. Dyre, and S. C. Glotzer, Crossover to potential energy
landscape dominated dynamics in a model glass-forming liquid. J. Chem. Phys., 112
(2000), 9834.
7. C. Donati, J. F. Douglas, W. Kob, S. J. Plimpton, P. H. Poole, and S. C. Glotzer, Stringlike
cooperative motion in a supercooled liquid. Phys. Rev. Lett., 80 (1998), 2338.
8. F. Sciortino, A. Geiger, and H. E. Stanley, Isochoric differential scattering functions in
liquid water: the fth neighbor as a network defect. Phys. Rev. Lett., 65 (1990), 3452;
Effect of defects on molecular mobility in liquid water. Nature, 354 (1991), 218;
Network defects and molecular mobility in liquid water. J. Chem. Phys., 96 (1992), 3857.
9. D. Laage and J. T. Hynes, A molecular jump mechanism of water reorientation. Science,
311 (2006), 832.
5
The pH of water
The pH of a liquid is a quantitative measure of the acidity of the medium
and is determined quantitatively by the free hydrogen ion concentration.
pH controls many biological and chemical activities in water. We know
that the pH of neutral water is 7.0. A near constancy of pH is extremely
important for life processes. The pH of sea water is estimated to vary
between 7.6 and 8. Another thing we need to know is that the pH of our
blood is 7.4. Recent theoretical advances have shed light on the unique
value of pH and also its temperature dependence. In this chapter we
introduce fundamental new developments in the area that have augmen-
ted and deepened our understanding of the acidity of water from a
molecular perspective.
5.1 Introduction
In water each hydrogen atom is involved in a covalent bond with an oxygen atom
of the same molecule, and in an HB with another oxygen atom of a neighboring
water molecule. According to quantum mechanics, it is also possible, at an energy
cost, to break the covalent bond in such a way that the proton of the OH covalent
bond becomes associated with this other water molecule, giving rise to the
formation of one H
3
O
+
ion and one OH
ion in the system. This phenomenon is

called autoionization and should be regarded as a chemical reaction in which
two water molecules react to produce a hydronium ion (H
3
O
+
) and a hydroxide
ion (OH
):
2 H
2
O 1 ( ) H
3
O
aq ( ) OH
(aq)
Water, however pure, is not a simple collection of H
2
O molecules. Even in pure
water, sensitive equipment can detect an electrical conductivity of 0.055 Scm
1
,
thereby revealing the presence of a small amount of ions that are intrinsic to pure
water, devoid of foreign bodies.
71
The preceding reaction has a chemical equilibrium constant of K
eq
= ([H
3
O
+
]
[OH
])/ [H
2
O]
2
= 3.23 10
18
. So the acidity constant, which is K
a
= K
eq
[H
2
O] =
([H
3
O
+
] [OH
])/ [H
2
O], has a value equal to 1.8 10
16
. For reactions in water (or
diluted aqueous solutions), the molarity (a unit of concentration) of water is
practically constant and is omitted from the acidity constant expression by
convention.
The resulting equilibrium constant is called the ionization constant, or dissocia-
tion constant, or self-ionization constant, or ion product of water, and is symbolized
by K
w
.
K
w
= K
a
[H
2
O[ = K
eq
[H
2
O[
2
= [H
3
O
[[OH
[; (5:1)
where [H
3
O
+
] = molarity of hydrogen or hydronium ion, and [OH
] = molarity of
hydroxide ion.
At standard ambient temperature and pressure, denoted by SATP (which is one
atmosphere pressure and 25C (298 K)), K
w
= [H
3
O
+
][OH
] = 1.0 10
14
. Pure
water ionizes or dissociates into equal amounts of H
3
O
+
and OH
, so their molarities
are equal:
[H
3
O
[ = [OH
[ (5:2)
At SATP, the concentrations of both hydroxide and hydronium ions are very low,
equal to 1.0 10
7
mol/L, and the ions are rarely produced: a randomly selected
water molecule dissociates approximately once in 10 hours. Since the concentration
of water molecules in water is largely unaffected by dissociation and [H
2
O] is
approximately equal to 56 mol/L, it follows that for every 5.6 10
8
water mole-
cules, one pair exists as ions.
Any solution in which the H
3
O
+
and OH
concentrations equal each other is

considered as a neutral solution. Absolutely pure water is neutral, although even
trace amounts of impurities could affect these ion concentrations and then water
may no longer be neutral. K
w
is sensitive to both pressure and temperature; it
increases when either increases.
Now, the self-dissociation constant of water is expressed in terms of pK
w
, which
is dened as,
pK
w
= log
10
K
w
(5:3)
As discussed earlier, the pK
w
of water under normal conditions is 14.
We can now introduce the pH of a solvent or solution. It determines the acidity of
an aqueous solution and is dened as
pH=log
10
[H
3
O
[ (5:4)
Thus, for water under normal conditions, the pH is 7.0 [1].
72 The pH of water
5.2 Temperature and pressure dependence of pH
It is important to know and understand the temperature dependence of the pH of
water. Not only does the temperature dependence describe the change with
temperature of liquid waters solvating properties and its behavior as a reaction
medium, it is also an immensely important quantity for biological purposes. The
pH of water shows interesting variation with temperature, depicted in Figure 5.1.
On increasing temperature from an ambient condition of 298 K, the pH rst
decreases. At 100C, the pH of water is about 6.1. It continues to decrease till
about 650 K. After ~650 K, the pH starts to increase again. Note that 647 K is the
critical temperature of liquid water. As the temperature is lowered below 298 K,
the pH continues to increase, reaching 7.4 at 0C. Thus, pH varies between 7.4
and 6.1 in the whole liquid range.
In an interesting study Ohmine and co-workers pointed out that when the
temperature rises from the normal to the supercritical region, the pK
w
of water
experimentally exhibits complex, non-monotonic temperature dependence, that
is, it rst decreases from14 and then increases rapidly (see Figure 5.1) [2]. Using a
quantummechanical study they provided a molecular-level picture of this peculiar
temperature dependence. The imbalance between the ionwater and the water
water molecular interaction strengths and the concomitant water density enhance-
ment in the hydration shell, observed (and well known) in the supercritical
liquids, combines to produce this temperature dependence of the pK
w
value. It
has been found that a large charge transfer from H
+
and OH
to the surrounding
water molecules plays an important role in determining the temperature
Figure 5.1. Temperature dependence of pK
w
. The solid line is for the present
theoretical result. The dotted line is for the experimental result at 32 MPa. Figure
adapted with permission from J. Chem. Phys., 122 (2005), 144504. Copyright
(2005) American Institute of Physics.
5.2 Temperature and pressure dependence of pH 73
dependence. In such transfers, water molecules not only from the neighboring
hydration shell but also from the outer distant hydration shell were found to play a
signicant role [2].
The hydration shell structure and the local density around H
3
O
+
and OH
show
interesting features with temperature variation. Though the rst-shell structures
around the solute ions remain more or less intact over temperature changes, the
second hydration shell and beyond become loose, and even disappear as tempera-
ture increases. It is found that the coordination number of water molecules around an
OH
is 4.5 at 300 K, which decreases slowly with increasing temperature. For

example, it becomes 4 at 800 K.
As stated earlier, self-ionization is the process that determines the pH of water.
Since the concentration of hydroniumat SATP (25C) is 1.0 10
7
mol/L, the pHof
pure liquid water at this temperature is 7. Since K
w
increases as temperature
increases, hot water has a higher concentration of hydronium ion than cold water,
but this does not mean it is more acidic, as the hydroxide concentration is also higher
by the same amount. Thus, pure water is always neutral with varying pH depending
on the value of pK
w
.
The pressure dependence of the pH of water also decreases with increase in
temperature but shows no non-monotonicity.
5.3 Mechanism of autoionization
In another interesting study Geissler et al. determined the origin and nature of the
forces that can act on the oxygenhydrogen bond in liquid water to cause molecular
dissociation [4]. As we previously noted, such events are extremely rare, certainly
on the femtosecond or picoseconds timescale. However, when it happens, the
system generates ions via rapidly crossing a transition state region. In this novel
study, transition path sampling and ab initio molecular dynamics trajectories assist
in revealing the autoionization mechanism. The following sequence of events
appears to take place in the process.
Let us assume that the system begins in a neutral state. Random high-energy
uctuations in molecular motions occasionally (about once in every 10 hours per
water molecule) produce an electric eld strong enough to break an oxygen
hydrogen bond, resulting in a hydroxide (OH
) and a hydronium ion (H

3
O
+
).
Once such a strong uctuation indeed occurs and the OH bond dissociates, the
proton of the hydronium ion gets delocalized as it travels along water molecules by
hopping from one molecule to another (like the diffusion of a defect in a solid), here
aided by the orientation of the H
3
O
+
ion and surrounding water molecules. This
migration causes a change in the hydrogen-bond network in the solvent. This
isolates the two ions, which are both stabilized by solvation.
74 The pH of water
However, the electric potential gradient that now exists due to the oppositely
charged ions triggers another change within 1 ps or so. This second reorganization
of the hydrogen-bond network allows rapid proton transfer down the electric
potential difference and subsequent recombination of the ions [4]. This timescale
is consistent with the time it takes for HBs to reorient themselves in water [5].
Thus, not only is the dissociation of the oxygenhydrogen bond rare, but only a
fraction of the ions generated escape recombination. These dynamic processes need
to be consistent with the low concentration of protons in pure water, which in turn is
determined by the competition between the bond enthalpy of an OH bond and the
solvation energy of the hydroxide and hydronium ions.
5.4 pH of blood
In vertebrates, blood is composed of blood cells suspended in plasma. Blood plasma
contains mostly water, i.e. 92% by volume. Other than water it contains proteins,
glucose, mineral ions, hormones, carbon dioxide, and of course the blood cells. The
pH of our blood is strictly controlled; it is maintained by a number of buffers within
a range of 7.35 to 7.45, more alkaline than pure water. Note also that the pH of the
ocean is typically between 7.6 and 8.4, with acidity seeming to be increasing due to
human impact and pH decreasing with time.
Note that pHis a logarithmof the concentration of hydrogen ions (protons). Thus,
a variation of 0.1 in pH is equivalent to a variation of 25% in the hydrogen ion
concentration in blood, which is not a small amount, and can have a disastrous
consequence. The following are the serious health consequences if the pH of our
blood moves out of this range.
If the pH falls below 7.35, a condition called acidosis, then malfunction of the
central nervous system causes depression. Severe acidosis where blood pH falls
even below 7.00 can lead to coma and even death.
If the pH of blood somehow moves above 7.45, the state is called alkalosis.
Severe alkalosis can also lead to death, but through a different mechanism
alkalosis causes all of the nerves in the body to become hypersensitive and over-
excitable, often resulting in muscle spasms, nervousness, and convulsions; its
usually the convulsions that cause death in severe cases.
A simple way to understand the importance of maintaining the pH of our blood
within a narrow range is to look at the pK
a
of the amino acids that constitute the
proteins. The ionic state of the protein is maintained by this value of pK
a
relative to
the pH of the surrounding uid. As an amino acid usually has two or three pK
a
values, it is often more useful to discuss it in terms of the isoelectronic point,
denoted by pI, which is the pH at which the amino acid is neutral, i.e. the zwitterion
form is dominant if the amino acid is stable, and the amino acid does not migrate in
5.4 pH of blood 75
an electric eld. There are several important amino acids (such as histidine, proline,
isoleucine) that have pI values in the range 6 to 8, and therefore the ionic state of these
acids is sensitive to the pH of the solution. As we know, in order to be biologically
active and efcient, proteins must be maintained in their folded native state with a
denite geometry and packing, with hydrophobic groups in the core and hydrophilic
groups outside exposed to the water environment. The stability of the folded state
depends on the charged state of the amino acids. As discussed earlier, the charged state
of the amino acids is determined by the medium pH or acidity of the medium.
Acids in the human body are generated by two means: (a) through regular
metabolic activities and (b) through the daily intake of food.
5.5 Food and blood pH
There are also several mechanisms by which our body maintains the pH around 7.4.
Some of these mechanisms use simple standard chemistry, some are more complex.
These mechanisms are: (i) the carbonic acidbicarbonate buffer system, (ii) the
protein buffer system, and (iii) the phosphate buffer system. Apart from these
buffers, the pH of our body is also maintained by exhalation of carbon dioxide,
elimination of hydrogen ions via the kidneys, etc.
In many medicinal systems, such as in the ancient Indian system, Ayurvedic
doctors advise control of food intake to lower the acidic nature of the body systemas
a treatment for many diseases. Fresh vegetables and fruits are particularly recom-
mended to lower this acidity. Among the fruits the following is a partial list of fruits
and vegetables that promote alkalinity of the body: gs and nuts, mango (yes! but
ripe ones only), papaya, watermelon, asparagus, kiwi, pears, pineapple, raisins,
vegetable juices, apples, apricots, alfalfa sprouts, avocados, bananas, garlic, lemons,
sweet seedless grapes, ginger, peaches, nectarines, grapefruit, most herbs, peas,
lettuce, broccoli, cauliower, etc.
On the other hand, the following is a partial list of foods and drinks that make
your body system acidic: fried foods, carbonated drinks (pops), coffee, white and
rened sugar, rened salt, articial sweeteners, antibiotics (and most drugs), white
our products (including pasta).
However, one need not be unduly alarmed because in an active person, the buffer
systems of the body are sufciently efcient to maintain the pH in the 7.357.45
range. However, at the same time, the system should not be over worked by
excessive intake of harmful foods and drinks. Long use of certain drugs, such as
steroids, can also severely compromise the functioning of our natural buffers.
Unfortunately, our understanding of the effects of food and drugs on the natural
buffers is still at its infancy. This is because many chemical processes are connected
within a biological cell and the reactions involved are very complex.
76 The pH of water
5.6 pH of seawater
It is important to note that the pH of seawater ranges between 7.6 and 8.4. The
alkanity of the pH of seawater is a little surprising at rst sight. The control starts
with carbon dioxide present in the atmosphere that gets dissolved in seawater.
Dissolved CO
2
reacts with seawater to form carbonic acid. Carbonic acid
(H
2
CO
3
) itself is unstable and dissociates to release a bicarbonate ion (HCO
3
)
and a carbonate ion CO
3
. They together form the buffer that assists in maintaining

the sea pH level. The carbonate ion can react with calcium ions (Ca), which are in
excess in seawater, leading to the formation of calcium carbonate (CaCO
3
), the
material out of which the shells of mussels and the bodies of corals are made.
5.7 Conclusion
The pHof water is an important property as it determines many reactions in aqueous
medium. However, as described here, understanding the unique pH of water is a
non-trivial problem. There are several issues that need to be considered. First, we
need to understand the relative stability of separated H
+
and OH
compared to
undissociated water molecules. Since the H
+
ion is of small size and of small weight
(just a proton), the consideration of this stability requires quantum mechanics. The
second issue is the rate of dissociation of an OH bond. This requires very strong
rare uctuations in the electric eld with proper polarization to act on the OHbond.
The third issue is the temperature dependence of pH. The last one is important for
biological reasons.
References
1. S. K. Lower, Acid-Base equilibria and calculations. Chem1 Virtual Textbook.
2. T. Yagasaki, K. Iwahashi, S. Saito, and I. Ohmine, A theoretical study on anomalous
temperature dependence of pKw of water. J. Chem. Phys., 122 (2005), 144504.
3. R. E. Mesmer, W. L. Marshall, D. A. Palmer, J. M. Simonson, and H. F. Holmes,
Thermodynamics of aqueous solution and ionization reactions at high temperatures
and pressures. J. Solution Chem., 17 (1988), 699.
4. P. L. Geissler, C. Dellago, D. Chandler, J. Hutter, and M. Parrinello, Autoionization in
liquid water. Science, 291 (2001), 21212124.
5. F. H. Stillinger, Theory and Molecular Models for Water. In Advances in Chemical
Physics: Non-Simple Liquids, Volume 31, eds I. Prigogine and S. A. Rice (John Wiley &
Sons, Inc., Hoboken, NJ, USA, 1975).
References 77
(a)
Figure 7.3. (a) X-ray crystal structures of adenylate kinase (ADK) showing the
open (denoted by O) and closed (denoted by C) states of the enzyme.
18
( )
X(A)
F
r
e
e

E
n
e
r
g
y

(
k
c
a
l
/
m
o
l
)
Minor Groovebound State
Intercalated State
Critical Region (Gate) Separated State
2
4
6
8
10
12
14
16
40
80
120
160
12
10
8
6
4
2
0
2
14
12
10
8
6
4
2
0
2
Figure 7.6. Free-energy landscape in the umbrella sampling coordinate X
(separation between DNA intercalation site and the drug) and the angle (see
Figure 7.5). The intercalated, minor groove-bound, and separated states are
specied, as is the critical region. The green arrowed dashed lines (and their red
projections onto the X- plane) are schematic guides to illustrate the most probable
path from separated minor groove-bound intercalated state through the
critical region. Adapted with permission from J. Am. Chem. Soc., 130, (2008),
(a)
(b)
Figure 9.6. Two-dimensional free-energy surface along with its contour map for
(a) strongly hydrogen-bonded quasi-bound water and (b) interfacial free water.
The color code of the free-energy landscape has been so chosen that the closely
spaced regions can be distinguished clearly. The presence of two minima in
(a) corresponds to two HB breaking events whereas in (b) the existence of a
single minimum indicates only one weak HB rupture. For both cases the escape
along the Z direction is evident from the contour. Adapted with permission from
J. Phys. Chem. B, 116 (2012), 29582968. Copyright (2012) American Chemical
Society.
Figure 11.1. Snapshot of a proteinDNA complex showing the helix-turn-helix
(HTH) binding motif of a protein associating with the target bases via the side-
chains. The common water molecules are highlighted here. The color code is as
follows: the HTH binding motif of the protein is indicated in blue, the remaining
part of the protein in colored red, and the DNA molecule is green. The water
molecules that are simultaneously hydrogen-bonded to the protein and DNA in the
complex are colored gray and the remaining water molecules are shown in
magenta. Adapted with permission from J. Chem. Phys., 135 (2011), 135101.
Copyright (2011) American Institute of Physics.
(b)
Figure 12.1. (b) Structure of bilayer by aggregation of these lipids and formation of
cell walls across which transport proteins carry chemicals in and out of the cell.
Figure 13.1. Crystal structure of Staphylococcus aureus isoleucinyl tRNA
synthetase (IRS) in complex with tRNA (Protein Data Bank code 1QU2(6)).
tRNA is shown in green, the CP1 domain in orange, the Rossmann fold in blue,
and rest of the protein is represented in silver.
k=0.0
k=1.0
Figure 14.4. Front snapshot of the rst layer of water molecules (layer thickness =
0.25 nm) at the interface with a hydrophobic apolar (k = 0.0, upper panel) and a
hydrophilic surface (k = 1.0, lower panel), showing the presence of hexagonal
structures (in yellow circles) on the apolar surface. T = 300 K. This result is taken
fromthe work by Castrilln et al. [1]. Figure adapted with permission fromJ. Phys.
Chem. B., 113 (2009), 1438. Copyright (2009) American Chemical Society.
Figure 16.5. Snapshots of the simulation of different binary mixtures water
DMSO in the top panel, waterethanol in the middle, and waterTBA in the
bottom panel. Water molecules are shown in silver. Co-solvents (DMSO,
ethanol, and TBA) are represented in blue. The snapshot is shown at two
different concentrations one before the onset of percolation to show the
microheterogeneity in the system, and one after the onset of percolation to
show the spanning cluster of the cosolvent. Figure adapted with permission
from J. Phys. Chem. B, 115 (2011), 685. Copyright (2011) American Chemical
Society.
Figure 20.3. Different stages of ice nucleation. Note the region indicated by the
yellow circle (actually a sphere in three-dimensional space), which indicates
the rst incipient ice formation. The time intervals are indicated on the upper
right corner of each panel [7]. Adapted with permission from Nature, 416 (2002),
409413. Copyright (2002) Nature Publishing Group.
Part II
Water in biology
6
Biological water
To understand the function of water in life processes at a microscopic
level, we need to know and understand the physical state of water in
biological systems. Water at the surface of proteins and DNA, within
tissues, cells, and cytoplasm behaves differently from bulk water. The
former has been termed biological water to distinguish it from bulk
water. However, the same unique features of water molecules discussed in
Chapter 1 manifest themselves differently in the new environments and, in
more than one way, help sustain and foster biological activity. Here we
point out the common characteristics of biological water and attempt to
offer a unied description of the same from a molecular perspective. We
discuss a simple model, termed the dynamic exchange model, that can be
invoked to explain the emergence of new timescales in the relaxation of
water in the biological environment.
6.1 Introduction
Remarkable properties that characterize bulk water are modied in biological
systems, where water molecules face a multitude of additional interactions. Water
molecules that populate the surfaces of proteins [1], inhabit the grooves of DNA[2],
reside (however temporarily) at the surfaces of lipid bilayers [3], or in tissues and
cells can exhibit properties that are quite distinct from those found in the bulk. Due
to its unique characteristics and certain common features observed in different
biological systems, this water has been termed biological water to distinguish it
from bulk water. Because of the additional interactions that a water molecule faces
at biological surfaces [4], the extended hydrogen-bond network of water in the bulk
becomes compromised. In some cases, the extended HB can be lost, either fully or
partially.
Nevertheless, the unique features of an individual water molecule that we dis-
cussed in Chapter 1 manifest themselves in the newer environments. The small size,
81
large dipole moment and the ability to form a multitude of stable structures all come
into play in biological environments and allow water to perform a multitude of
functions. As mentioned earlier, water molecules not only stabilize the native state
of a protein and the double-helix form of DNA, but actively participate in biological
processes (enzyme kinetics, biochemical synthesis). In tissues and cells, water
molecules, although by far the predominant species, still have to share space with
other biomolecules and organized structures. In most cases the water layer can be
only a few monolayers thick. The extended hydrogen-bond network that dominates
the properties of bulk water is lost either partially or almost entirely in such
circumstances.
The structure and dynamics of such layers, which are determined by hydrophobic
and hydrophilic interactions, among other inuences, are important for biological
(for example, immunological) responses. For example, it has been suggested that
the process of protein-ligand recognition is at least partly determined by the
energetics and dynamics of water during ligand binding.
A clear picture of the difference in dynamics between bulk water and biological
water can be seen in Figure 6.1. Here we plot the running trajectory of the angular
displacement of a randomly chosen and tagged water molecule, separately for one in
bulk water and one on the surface of a protein. The tagged water molecule is at
the surface of a protein lysozyme. In bulk, a water molecule undergoes frequent
Figure 6.1. Projection of cos of the angle of a tagged water dipole (in the bulk of
the simulated water) with the z-axis of a space xed frame (upper gure). In
the lower panel, the same has been shown for a doubly hydrogen-bonded
bound water (bound to the surface of protein lysozyme) molecule. The free water
molecule frequently executes large-amplitude jumps while a bound water molecule
stays mostly immobile. Adapted with permission from Chem. Rev., 105 (2005),
82 Biological water
large-amplitude jumps, while on the surface of a lysozyme the bound (or quasi-
bound) water hardly moves. We have chosen a case that perhaps provides an
exaggerated picture of the difference but it makes the point clearly.
6.2 Relaxation measurements
It is very difcult to directly observe water molecules in vivo at the surfaces of
proteins or lipid bilayers or in the grooves of DNA. X-ray studies of protein crystals
cannot provide full information as only a fraction of water molecules remain on the
surface and even then in restricted positions. Also, the hydration layer is mobile in
solution. One would like to know about the structural and dynamical characteristics
of these water molecules in the active state, ideally within biological cells. Such
detailed information is still not available in most cases. Much of our current under-
standing of water in biological systems has come fromstudy of proteins and DNAin
aqueous solution.
Initial information about the protein hydration layer came fromrelaxation studies.
Dielectric relaxation (DR) and NMR studies were the rst to reveal the existence of
water molecules in the restricted environments. Dielectric relaxation measurements
show the existence of an additional dispersion in protein solutions with time
constants in the 4050 ps time range (to be contrasted with 8 ps for bulk water),
while NMR estimates have varied from system to system, with claims ranging from
slow (with lifetimes in excess of 300 ps) to fast (with lifetimes 25 ps). The general
consensus now appears to be consistent with the DR data.
The dielectric relaxation spectra of various DNA solutions are also quite complex,
actually more complex than those observed for protein solutions. One of the reasons
for such complex relaxation behavior of DNA solutions is the presence of free and
bound counter ions in solution. Water molecules in the major and minor grooves of
water behave differently from those in the bulk. The presence of the positively
charged counter ions in turn inuences the response of many other water molecules.
Negatively charged phosphate ions also inuence the surrounding water molecules.
Several newer techniques, such as solvation dynamics (SD), two-dimensional
infrared (IR) spectroscopy and terahertz spectroscopy, have been employed to study
water at biological surfaces. We shall discuss the results in the designated chapters,
along with computer simulations. In the following we gather together a few common
characteristics of biological water that may help provide a general viewof the problem.
6.3 Unique characteristics of biological water
Below we list the common characteristics of water observed in/at constrained
environments, such as on the surfaces of biomolecules or within cells and tissues.
6.3 Unique characteristics of biological water 83
(i) Biological water might not sustain the extended three-dimensional hydrogen-
bond network that is present in bulk water. There could be a disrupted two-
dimensional network in some cases, such as in the protein hydration layer, or a
modied network as in the spine of hydration in the DNA double helix in
solution.
(ii) Water molecules may be locally organized and exhibit different structural
patterns. This is where the fth unique feature listed in Chapter 1 can be really
effective. Many of these structures represent local free-energy minima in the
conguration space of water, although not necessarily the deepest minimum.
(iii) Many water molecules can have less than four H
2
O as neighbors, or might be
devoid of any H
2
O neighbor. In many cases of the enzyme catalysis of
biochemical processes, one or two water molecules facilitate the reaction.
(iv) Water molecules can be broadly categorized into two types. A few water
molecules will almost always be strongly hydrogen-bonded to charged/polar
groups on the surface of biomolecules. They provide enthalpic stabilization to
the preferred state of the biomolecule. Then, there will also be water mole-
cules, more numerous than the former kind, that are relatively free. These
water molecules may face non-polar surfaces that do not have a hydrogen-
bond-forming ability. We shall call these two types bound and free. They
represent two limiting cases. There are many other water molecules on the
surface that are weakly bonded and partially free. The different kinds of
molecules serve different purposes.
(v) The dynamics of water molecules may span a much wider time range in the
form of biological water than in bulk water. There can be very fast dynamics
(as in bulk water) but there can be much slower dynamics due to the inuence
of the bound water molecules.
(vi) Relaxation functions that characterize the dynamics of biological water are
non-exponential in general, with the slow components providing a measure of
the interaction energy, as quantied below.
(vii) The density of biological water can be different in different regions, depending
on the surface. This point is related to point (ii) mentioned above. The
hydration layer is highly heterogeneous at small length and timescales.
All of the above characteristics need not be uniformly observed in all kinds of
biological water but some of them should manifest themselves in biological functions.
6.4 Phenomenological models and simple theories
The above characteristics have led to the formulation of a simple model of biolo-
gical water in the hope of capturing some of the general features. In Figure 6.2 a
84 Biological water
schematic illustration of the model is presented. We show the elementary processes
(rotation and translation) of water molecules in the hydration layer, and also the
escape from and entrance to the layer. These processes could be involved in
biological reactions and/or transformations, such as enzyme kinetics and protein
folding.
In sharp contrast to the large number of experimental and computer simulation
studies of the structure and dynamics of water in hydration layers reported in the
literature and to be discussed in later chapters in this book, there have been only a
few purely analytical (or model-independent) studies on the dynamics of the
hydration layer found around biomolecules, within tissues and cells, and in and
around self-assemblies. Some of the early theoretical studies invoked a simple
Figure 6.2. Schematic representation of the dynamic exchange model, showing the
dynamic equilibrium between the bound and free water molecules at the surface of
a protein in aqueous solution. At the surface, there are basically three types of water
molecules. One type is relatively immobile as these water molecules are doubly
hydrogen-bonded to a polar or charged group at the surface. The second type is free
and not hydrogen-bonded to any side-chain or protein atom. The third type is
singly hydrogen-bonded. However, not all hydrogen bonds (HBs) with protein are
strong. Some are quite weak and therefore cannot be regarded as being bound to
protein. We have also indicated the dynamic equilibrium in terms of a crossing and
recrossing of a double-well potential. Water molecules diffuse in and out of the thin
hydration layer, as also indicated on the gure. Adapted with permission from J.
Phys. Chem. B, 107 (2003), 1321813228. Copyright (2003) American Chemical
Society.
reactiondiffusion model which was used to understand the NMR results. In such a
model, the bound state was assumed to be close to the protein surface and the free
state away from it. Thus, the transition from bound to free was dependent on the
location of water relative to the chosen reaction surface and the distance from the
surface was the reaction coordinate. Thus, one could represent the dynamics of
the transition by a reactiondiffusion reaction. This model was ideal for treating the
nuclear Overhauser effect (NOE) results, which are sensitive to the distance
between the protein proton and the water proton.
Asomewhat different model was proposed more recently by Nandi and Bagchi [4].
This model was initially proposed to explain the anomalous DR of aqueous
protein solutions, but since then has been used to explain protein and DNA SD.
This is termed the dynamic exchange model as it envisaged the exchange of water
between the surface and the bulk water as a distinct dynamic process, as shown in
Figure 6.2.
The dynamic exchange model (illustrated in Figure 6.2) is based on the assump-
tion that the water molecules at the surface of proteins can be categorized into
distinct species as bound and free depending on the nature of their hydrogen-
bonding to the biomolecular surface (Figure 6.2). This equilibrium can be symbo-
lically written as [4]
Bound water free water (6:1)
This is a dynamic equilibrium. Bound water, however, does not remain bound for a
very long time. Also, there is a distribution of the energies of binding of water
molecules to the protein surface, so there is a distribution of binding energy that
needs to be included in the theoretical description. For simplicity, the model
considered below includes only the two state conditions of water molecules,
bound and free. A more detailed description, with some amount of derivation, is
given in Appendix 6.A of this chapter.
In the implementation of the model, it is further assumed that as the bound water
molecule is immobilized by the protein surface, it cannot rotate or translate. Thus, it
must become free to move. The bound to free (and free to bound) transition is
described as a chemical reaction. The free water molecules, on the other hand, are
assumed to behave as molecules in bulk water, although their rotation and transla-
tion diffusion are generally modied due to their interaction with the protein. This
surface layer of bound and free water is coupled to the bulk water outside the layer.
Although this is a key feature, it is ignored in many other models. In addition, we
allow the possibility of the bound water having a preferred orientation due to its
interaction with the protein.
Dynamic exchange between the bound and the free water molecules is described
by a reactiondiffusion equation, with both rotational and translational motions
86 Biological water
included. Because of the simplicity of the model, it can be solved analytically.
Details of the solution of the model are presented in Appendix 6.A to this chapter.
Note that this model is quite general. If we describe the bound and free water
molecules as two distinct minima in a bistable potential, it can describe hydrogen-
bond-breaking dynamics in many situations.
We summarize the main results of the dynamic exchange model, as follows.
(i) The coupled reactiondiffusion equation can be solved to obtain the two rate
constants, k
,
given by
k
= 0:5 B
B
2
4D
R
k
bf
_
_ _
with B = 2D
R
k
bf
k
fb
;
(6:2)
where D
R
is the rotational diffusion coefcient of the free water molecules in
the interface, k
bf
is the rate of bound-to-free transition, and k
fb
is the rate
constant of the reverse process. Typically, the rate constant of the bound-to-
free reaction, k
bf
, is smaller than that for the reverse process, k
fb
.
(ii) Because of the presence of a rather broad distribution of energy of the HB of
water molecules with various amino acid side-chains and also peptide back-
bone atoms at protein surfaces, the measured relaxation is expected to be
non-exponential in general. That is, the dynamics of each bond, under the
approximation of the two-state model, gives a bi-exponential decay for the
relaxation function, as envisaged above.
(iii) However, one can still group the relaxation times into two classes. Most of the
relaxation times are close to those in bulk water. Then a second, slower, relaxa-
tion time, given essentially by the rate of bound-to-free transition, comes from
the limiting situation when the rate of conversion from bound to free becomes
very small. In such a case, the above expression further simplies and the two
rate constants are given by 2D
R
and k
bf
. Thus, while one time constant remains
fast, of the order of 24 ps
1
, the other can slow down appreciably, even to the
extent of tens or hundreds of picoseconds, as mentioned above. In the dynamic
exchange model, the rate constant k
bf
is of course determined by the binding
energy. For the majority of sites, the time constant may range between 10 and
100 ps or so because the binding energies are often small. However, for a few
(but not rare or uncommon) molecules, the second time constant can be quite
long. For such slow water molecules, the two time constants are given by
f ast
-
bulk
s
(6:3)
slow
- k
1
bf
(6:4)
(iv) Note that in the same limit of large activation energies separating the bound
state from the free one, the residence time of the bound water molecules is
given essentially by k
bf
1
. This also gives the residence time (
res
) of strongly
bound water, that is,
slow

res
. This is an important result.
However, there could be only a small percentage (less than 10% for most
proteins) of water molecules on protein surfaces strictly obeying the condition
that satises Eq. (6.4).
The above dynamic exchange model applies mainly to the rotational relaxation of
interfacial water molecules. It has also been extended by Bhattacharyya et al. [6] to
treat wave-number-dependent relaxation, which includes translational diffusion
more realistically. As stated, the model is simple and phenomenological but captures
several essential aspects of water dynamics at the biological interface, in addition to
being analytically tractable.
Clearly, the dynamic exchange model is phenomenological and does not address
in detail the microscopic dynamics in the hydration layer. However, it is simple and
largely analytical, and provides an appealingly simple picture of the altered
dynamics at protein and DNA surfaces. In the appendix, a more detailed discussion
of the model is presented.
6.5 Proteinglass transition and hydration-layer dynamics
One way to understand the protein hydration layer is to go to the low-temperature
limit so that the dynamics slows down and one can hopefully discern different types
of motion. Such studies have been carried out, both by experiments and by simula-
tions. Neutron-scattering experiments show a sharp change in the molecular
motions of a hydrated protein around 220 K, which has been attributed to a glass
transition in the protein. The change is measured in the mean-square displacements
of the protein atoms. The mean-square displacement shows a rapid increase as the
temperature is increased beyond 220 K (see Figure 6.3)[7].
Biological activity is found to be restored after the proteinglass transition [8].
The proteinglass transition appears to be a general phenomenon at lowtemperature
in hydrated proteins.
Several studies have explored the relationship between a possible dynamic
transition in supercooled water and a transition in glass. The issue, however, is
complicated. In bulk water, indirect evidence (such as extrapolation of experimental
results at 260 Kand simulations with various force elds) suggests the occurrence of
a structural transformation in the supercooled liquid around 220 K.
Computer simulations identify such a transformation with a liquidliquid transi-
tion where the usual liquid, now called high-density liquid (HDL), at high
88 Biological water
temperature becomes transformed to a low-density liquid (LDL). While HDL
contains a random mixture of 5-coordinated, 4-coordinated, and 3-coordinated
water molecules and at ambient pressure is characterized by an average density of
about 0.98 gcm
3
at 230 K, LDL, on the other hand, is primarily 4-coordinated and
is characterized by a density of about 0.94 gcm
3
. The liquidliquid transition has
only been observed in computer simulations and in small conned systems, but in
the bulk is believed to be preempted by crystallization around 232 K.
Bulk liquid water is also believed to undergo a glass transition in the temperature
range 160180 K, but this is also not clearly established.
We have addressed some of these issues in Chapter 2 (the water anomaly chapter)
and shall return to them again later in Chapter 22 to discuss possible reasons for the
anomalies.
As remarked, hydrated protein samples exhibit a sudden onset of motions of the
protein side-chains. Protein backbone atoms also show the dynamic transition and
as expected the amplitude of uctuations is less here compared to those of side-
chain atoms. This onset of motion could be related to either of the two phenomena
mentioned above but could also be the result of a new phenomenon arising from the
coupling between the protein and water.
Clearly understanding of this interesting phenomenon requires understanding of
the hydrated water, in particular the coupling between the layer and the protein. As
temperature is lowered, the dynamics of both the sub-systems slow down. It is
believed that the dynamics (or, rather motion of amino acid side-chains) of a protein
Figure 6.3. Mean-square displacements of ns-ps motions in maltose binding
protein (H-MBP-D
2
O sample; gray circles) and in its hydration water (D-MBP-
H
2
O sample; black diamonds). Dynamic transitions (changes in the slope of
temperature-dependent mean-square displacements) in the protein and in its
hydration water take place at similar temperatures (~220 K). Adapted with
permission from J. Am. Chem. Soc., 130 (2008), 45864587. Copyright (2008)
6.5 Proteinglass transition and hydration-layer dynamics 89
can be described as diffusion in a rugged energy landscape where the water
molecules in the layer may retard the motion by offering frictional resistance. On
the other hand, the motion of the water molecules on the surface of the protein is
retarded by the presence of the protein, as discussed above. It is clear that the
slowing down in the motion of any one inuences the other. This synergy of motion
may explain the observed results, although a quantitative understanding of the
problem is yet to be achieved.
6.6 Protein aggregation and biological water
We shall discuss some aspects of protein aggregation and association in the next
chapter. But here we point out that the quasi-bound and free water molecules in the
protein hydration layer play an important role in directing protein association. The
hydrophobic patches on the surface of biomolecules that are surrounded by free
water molecules may help mediating the hydrophobic attraction between hydro-
phobic patches. These water molecules can also be easily displaced. However, the
directional role of hydrophobic attraction is perhaps more important. It is not yet
clear how long-range this hydrophobic attraction between two hydrophobic patches
could be and also not clear is the precise physical origin of this force. If sufciently
long-range, then this hydrophobic attraction can indeed provide the necessary
direction for proteins to move towards each other with proper orientation. Bound
water molecules may be enthalpically too stable to usher in attraction of any kind
between two biomolecules.
6.7 Conclusion
In this chapter we have introduced the concept that water in biological systems is
quite different from bulk water that we are so familiar with and that has been studied
extensively over many decades. In contrast, the study of water in biological systems
has just begun. The reason for such a late start is not any lack of interest. Rather,
experimental systems (water in and around proteins, DNAhydration later, etc.) were
not amenable to study even in vitro. In vivo study of water in important systems is
still largely an unexplored area.
Recent studies in different systems seem to reveal certain commonalities that
have motivated the use of the term biological water to distinguish water in
biological systems. First and foremost, the extended hydrogen-bond network pre-
sent in bulk water is mostly lost near a biomolecular surface. Second, water can exist
in different states. These molecules differ in their coordination with other water
molecules and with the surface of proteins or DNA or tissues. The term biological
water serves to emphasize this difference between water in biology and the one we
90 Biological water
routinely use. In fact, biological functions (discussed in the next chapter) require
some of the molecules to be free and some to be bound to the surface. In a certain
sense, it is the presence of an extended surface with varying polarity in all the
systems mentioned above that gives biological water unifying characteristics. In that
sense, the term may be regarded as pedagogical.
The attempt to classify the behavior of water in biological systems under one
common umbrella is an effort not devoid of ambiguity. Unlike bulk water, here
properties can vary substantially from system to system, as we shall see in the
chapters to follow.
APPENDI X 6. A THE DYNAMI C EXCHANGE MODEL
The dynamic exchange model (DEM) of dynamics in protein and the DNA hydra-
tion layer was originally proposed in 1997 and was subsequently further developed
in several other studies [4]. It is a simple phenomenological model, based on
arguments common in chemical kinetics, but serves as a starting point to address
the inuence of protein or DNA (or lipid, micelles) on the surrounding water
molecules.
It was assumed in the NandiBagchi model [4] that water molecules at the protein
surface can be categorized into three different (transient) species: (i) those bound to
protein polar/charged groups by two HBs, termed IBW2, (ii) those bound to protein
groups by one HB, called IBW1, and (iii) lastly those that are not bound but free,
termed IFW. It is also assumed that these three kinds of water molecules are in
dynamic equilibrium.
Figure 6.A.1 illustrates the dynamic equilibrium between different species and
also between interfacial free water (IFW) and free water in the bulk (BFW). We
make the distinction between IFWand BFWon the basis of location.
Note that here we have used a different notation as the number of species
considered is greater here and the description is more elaborate than what was
described in the main text.
Figure 6.A.1. Schematic illustration of the dynamic exchange between different
types of bound water and free water species at a protein surface. Here IBW1 and
IBW2 refer to interfacial water molecules bound to the biomolecule by one and two
HBs, respectively. IFW denotes interfacial free water (not hydrogen-bonded to the
biomolecule) and BFW is bulk free water.
6.7 Conclusion 91
The equilibrium constant for the above dynamic equilibrium between such
different types of water molecules can be written as follows:
K
12
=
IBW2 [ [
IBW1 [ [
=
k
12
k
21
(6:A:1)
K
1F
=
IFW [ [
IBW1 [ [
=
k
1F
k
F1
Here k
12
, k
21
, k
1F
, k
F1
are the rate constants of the exchange from one water species
to another. In this way we can write the other exchange processes in the hydration
layer in terms of their equilibrium constants. The equilibrium constants themselves
are related to the free-energy difference between the species, by the usual thermo-
dynamic relation, G = RT ln K.
Thus, the concentrations of different species in the hydration layer depend on the
nature of the surface. It is expected that the properties of the hydration layer vary
with the said concentrations.
Apictorial illustration of the model was already given in Figure 6.2. If the average
density of water at the surface (
S
) is different from that in the bulk (
B
), that should
be included in the conservation law that applies on average of each of the species
described above.
5
IBW1
> 5
IBW2
> 5
IFW
>=
S
The number or density of each of these species is also conserved, determined by
details of the waterprotein interaction.
An additional constraint imposed is that the doubly H-bonded water (IBW2)
molecules (in this transient state) are immobilized by the protein surface and thus
cannot rotate or translate. The interfacial free water (IFW) molecules essentially
behave almost as bulk water, although their rotation and translation diffusion are
generally modied due to their interaction with the protein. We thus envisage the
molecules IBW1, IBW2, and IFW in dynamic equilibrium among themselves,
which is further modulated by the presence of a large reservoir of bulk water just
outside the layer. Thus, the three species together form a grand canonical ensemble,
with each of the three species characterized by its own chemical potential.
Despite considerable discussions, it is not clear what the width of the hydration layer
in many of the surfaces should be. It thus becomes a parameter in the theory. However,
this width cannot be more than two layers for the assumptions of the theory to be valid.
The dynamic exchange model employs a hydrodynamic approach wherein the
dynamics of the three species in the surface layer is described by a reaction
diffusion equation and the bulk water dynamics is described by a simple diffusion
equation. Therefore, in this approach, the interactions are not considered explicitly
92 Biological water
but are understood to be reected in (i) average concentrations of the three species,
(ii) the values of the diffusion coefcients, and (iii) the rates of inter-conversion
among the species.
In principle the time evolution of the position- and orientation-dependent density
of the free and bound water
f
(r,,t) and
b
(r,,t), respectively, are dened as the
function of position r and orientation , at time t, for the respective species. The
position dependence of a tagged water molecule can change by diffusion. Diffusion
can accelerate relaxation of
f
(r,,t). However,
b
(r,,t) cannot change by transla-
tion. The role of translational diffusion in hydration-layer dynamics was considered
in ref. [6]. Here we shall omit the translational motion and consider the dynamics in
the layer to occur by rotation, which is a much faster process and is probed by DR
and SD. Additionally, we shall consider only a two-state model of water in the
hydration layer bound and free.
If we now restrict ourselves only to rotational motion, then the time evolution for
the free water density is given by [4]
@
@t

f
(
f
; t) = D
W
R
\
f
2
f
(
f
; t)
f
(
f
; t)
_
d
b
k
1

f

b
( )
_
d
b
b
(
b
; t)k
2

b

f
( )
(6:A:2)
and that for the bound water by
@
@t

b
(
b
; t) =
b
(
b
; t)
_
d
f
k
2

b

f
( )
_
d
f
f
(
f
; t)k
1

f

b
( )D
B
R
\
2
b
(
b
; t)
(6:A:3)
The right-hand-side of Eq. (6.A.2) and Eq. (6.A.3) describe the change in
f
and
b
,
respectively, due to rotational diffusion, where D
W
R
is the rotational diffusion con-
stant of the free water molecule and D
B
R
is the rotational diffusion constant of the
biomolecule. Here it is important to note that free and bound refer to the two
constituents of the biological water.
Now we assume that the orientation of the molecule does not change during the
free bound conversion. That is, we assume that the rate constants k
1
(
f
b
)
and k
2
(
b

f
) are local and can thus be approximated by the respective
functions.
k
1

f

b
( ) =
f

b
( )
k
2

b

f
( ) =
b
f
( )
(6:A:4)
6.7 Conclusion 93
Further we consider that the shape of the biomolecule is spherical and D
B
R
is
isotropic. So the above equations can be simplied as,
@
@t

f
(
f
; t) = D
W
R
\
f
2
f
(
f
; t) k
1
f
(
f
; t) k
2
b
(
b
; t) (6:A:5)
@
@t

b
(
b
; t) = k
2
b
(
b
; t) k
1
f
(
f
; t)D
B
R
\
2
b
(
b
; t) (6:A:6)
We next expand the angle-dependent densities in terms of spherical harmonics in the
space xed frame (may be placed at the center of the protein)
f
(
f
; t) =
lm
a
f
lm
(t)Y
lm
(
f
) (6:A:8)
b
(
b
; t) =
lm
a
b
lm
(t)Y
lm
(
b
) (6:A:9)
where a
f
lm
(t) are the expansion coefcients and Y
lm
(
f
) are the spherical harmonics
of ran l and projection m.
From the above equations we can easily derive the two differential equations for
the free and bound water molecules in the following way:
@
2
@t
2
a
f
lm
l(l 1)D
W
R
k
1
k
2
l(l 1)D
B
R
_
@
@t
a
f
lm
l(l 1)k
2
D
W
R
l
2
(l 1)
2
D
W
R
D
B
R
k
1
l(l 1)D
B
R
_ _
a
f
lm
= 0
(6:A:10)
@
2
@t
2
a
b
lm
k
1
k
2
l(l 1)D
B
R
_
@
@t
a
b
lm
l(l 1)k
2
D
B
R
_
a
b
lm
= 0 (6:A:11)
Thus, we have a solution of the following form where C
1
and C
2
are the coefcients
of two relaxation modes.
a
f
lm
(t) = C
1
e
n
/
1
t
C
2
e
n
/
2
t
_ _
(6:A:12)
n
/
1
and n
/
2
have the following forms:
n
/
1
=
X
X
2
4Y
_
2
(6:A:13)
n
/
2
=
X
X
2
4Y
_
2
(6:A:14)
94 Biological water
where
X = l(l 1)D
W
R
k
1
k
2
l(l 1)D
B
R
_
(6:A:15)
and
Y = l(l 1)k
2
D
W
R
l
2
(l 1)
2
D
W
R
D
B
R
k
1
l(l 1)D
B
R
_ _
(6:A:16)
Similarly, for the bound water species, we can obtain a solution of the form
a
b
lm
(t) = C
3
e
n
/
3
t
C
4
e
n
/
4
t
_ _
(6:A:17)
where C
3
and C
4
are the coefcients of the two relaxation modes. n
/
3
and n
/
4
have the
following forms:
n
/
3
=
M
M
2
4N
_
2
(6:A:18)
n
/
4
=
M
M
2
4N
_
2
(6:A:19)
where
M = k
1
k
2
l(l 1)D
B
R
_
(6:A:20)
and
N = l(l 1)k
2
D
B
R
_
(6:A:21)
According to the model, only the free water molecules contribute to relaxation.
Therefore, Eq. (6.A.10) is the important equation. If we further assume that D
B
R
is
negligible, then we recover the simple results quoted in the text.
References
1. R. Pethig, Protein-water interactions determined by dielectric methods. Annu. Rev. Phys.
Chem., 43 (1992), 177205.
2. R. B., Gregory, Ed. Protein Solvent Interactions (New York: Marcel Dekker, 1995).
3. J. X. Cheng, S. Pautot, D. A. Weitz, and X. Sunney Xie, Ordering of water molecules
between phospholipid bilayers visualized by coherent anti-Stokes Raman scattering
microscopy. Proc. Natl. Acad. Sci. USA, 100 (2003), 98269830.
4. N. Nandi and B. Bagchi, Anomalous dielectric relaxation of aqueous protein solutions.
J. Phys. Chem. A, 102 (1998), 82178221; N. Nandi and B. Bagchi, Dielectric relaxation
of biological water, J. Phys. Chem. B, 101 (1997), 1095410961.
References 95
5. S. K. Pal, J. Peon, B. Bagchi, and A.H. Zewail, Biological water: femtosecond dynamics
of macromolecular hydration. J. Phys. Chem. B, 106 (2002), 1237612395.
6. S. M. Bhattacharyya, Z.-G. Wang, and A. H. Zewail, Dynamics of water near a protein
surface. J. Phys. Chem. B, 107 (2003), 1321813228.
7. K. Wood, A. Frolich, A. Paciaroni, et al., Coincidence of dynamical transitions in a
soluble protein and its hydration water: direct measurements by neutron scattering and
md simulations. J. Am. Chem. Soc., 130 (2008), 45864587.
8. B. F. Rasmussen, A. M. Stock, D. Ringe, and G. A. Petsko, Crystalline ribonuclease A
loses function below the dynamical transition at 220 K. Nature, 357 (1992), 423424.
96 Biological water
7
An essential chemical for life processes: water
in biological functions
Water molecules play diverse roles in facilitating biological processes
that are aided by their unique features mentioned in Chapter 1. The
dynamic and spatial heterogeneity in the hydration shell of proteins
facilitates many important processes, such as enzyme functions, molecu-
lar recognition, and protein association, to name a few. Water plays
another role in facilitating drugDNA intercalation, as discussed
below. While some of these actions have been discussed extensively in
the literature, the direct participation of water molecules where one of the
oxygenhydrogen bonds is broken and the resulting proton and hydroxyl
anion are used as chemicals seems to have escaped much articulation,
even though these steps can inuence the kinetics of the catalysis pro-
cesses profoundly. In fact this is really impressive chemistry because, as
we discussed in Chapter 4, the breaking of the oxygenhydrogen covalent
bond to create a hydroxyl anion and a proton is a rare process in bulk
water, but this occurs apparently in an effortless manner in enzyme
catalysis. We cannot stop without mentioning its inevitable role in photo-
synthesis. In this chapter we try to congregate a few specic examples
where water plays a role as a direct collaborator in the functioning of the
biological world.
7.1 Introduction
We all knowthat water controls chemical and physical changes in the biological world
in both direct and indirect fashions. This was recognized long ago by Leonardo da
Vinci, who termed water a vehicle of natural changes. It is also termed, albeit less
poetically, a lubricant of life. What, however, was probably not recognized earlier is
that water actively and directly participates in (that is, not just mediates or lubricates) a
large number of biological processes, making their occurrence possible in the rst
place. In this process water often disintegrates into H and OH and combines with
reaction intermediates to take the reaction to completion.
97
In this chapter we shall discuss a few of the examples where this biological
function has been recently established. By biological function one usually
expresses the stabilizing effect of water in proteins and DNA. But, as mentioned
earlier, the direct participation of water molecules in many biochemical processes,
such as in enzyme kinetics and protein synthesis, is often overlooked. As dis-
cussed below, water is indeed an active contributor in many processes where
water is consumed as a chemical and the oxygen-HB is broken to stabilize the
product. As mentioned above, this is really impressive chemistry because, as we
discussed in Chapter 5, the breaking of oxygenhydrogen covalent bonds to create
a hydroxyl anion and a proton is a rare process in bulk water (happens only once
in 10 hours!) but this occurs apparently in an effortless manner in enzyme
catalysis. Enzymes seem to use their own electric elds to rupture this strong
covalent bond.
In the previous chapters, we have discussed water around biomolecules
(proteins, DNA) and lipid bilayers. We have discussed the role of water in
stabilizing these structures either directly through hydrogen-bonding or indir-
ectly through hydrophobic interactions. We have also discussed how water
surrounding these biomolecules facilitates easy conformational uctuations
(for proteins and DNA) and the large-amplitude motions that are often required
for biological function, such as intercalation of an anti-tumor drug into DNA. In
order to carry out such functions, water molecules often act in large numbers,
even in a collective manner. They seem to efciently use the enthalpyentropy
balance to minimize the free-energy barrier for such processes. Without waters
contribution (at a molecular level), most of the cellular processes would be
impossible [1].
As the human body constantly consumes a large amount of water every day, we
are constantly becoming dehydrated. No other uid can substitute for water.
Specically, it initiates the digestion of proteins, fats, and carbohydrates through
hydrolysis. Water and enzymes work together to maintain optimum digestion,
nutrient absorption, and health.
The scope of this chapter, in principle, is enormous. It spans from individual
biological processes (whose number is almost countless) to a general view of
life as articulated in Darwins theory of evolution. In an individual biological
process, water can actively participate (as in many enzymatic reactions dis-
cussed below) or facilitate the change, as in the intercalation of a drug
into DNA. Recent studies have revealed the role of water in both DNA
transcription and translation through a process known as kinetic proofreading,
which is a term to describe lack of error in protein synthesis. In the following
we discuss these topics from a molecular perspective that has been beginning
to emerge in recent years.
98 An essential chemical for life processes: water in biological functions
7.2 Role of water in enzyme kinetics
Enzymes are well recognized as biological catalysts. According to the general kind
of reaction they catalyze, enzymes are categorized into various classes, such as
oxidoreductase, transferase, hydrolase, lyase, and isomerase, to name a few. Among
them probably the best-known enzymes are hydrolases. They catalyze the breaking
of bonds with the direct assistance/participation of water molecules. The reaction
can be generally written in the following form
AB H
2
O AOH BH
In many cases, water turns into a proton and a hydroxyl anion and the AB bond
shown above breaks to make A
+
and B
. Therefore, a pair of electrons needs to be

moved around, and electrostatics of the medium must play an important role. We
discuss specic examples below.
The digestion of carbohydrates and proteins not only requires these hydrolases
but a sufcient amount of water is also necessary to execute the bodys digestive
mechanism. To show how these enzymes function in the body, we take the example
of sucrase. Sucrase inhabits the surface of the microvilli on the intestinal mucosal
membrane. It catalyzes the hydrolysis of sucrose, which is a disaccharide, into two
monosaccharide units: glucose and fructose [2]. During the catalysis process, when
sucrose binds to the active site of the enzyme (sucrase), the enzyme conguration is
changed such that the oxygen bridge between the two monosaccharides is exposed
to water molecules.
Our second example is provided by proteases, which are another type of hydro-
lases that catalyze the hydrolysis of peptide bonds in proteins. A general reaction
scheme for proteases is shown in Reaction 7.1.
Interestingly, there are many such examples of proteases that exist in nature, such
as, trypsin, chymotrypsin, elastase, thrombin, subtilisin, plasmin, pepsin, chymosin,
cathepsin D, renin, and HIV-1 protease, etc. To illustrate the direct participation of
water in the digestive mechanism we show another example of proteolysis
Reaction 7.1
7.2 Role of water in enzyme kinetics 99
performed by chymotrypsin. Chymotrypsin is preferentially responsible for the
cleavage of peptide amide bonds where the carboxyl side of the amide bond is a
tyrosine, tryptophan, or phenylalanine residue [3]. The reactants and products of this
amide bond-cleavage process are shown in Figure 7.1.
In addition to its direct participation in the bond-cleavage process, the localized
water molecule present in the catalytic active site can cause signicant stabilization
of the transition state formed in the middle course of a reaction.
Another interesting example is provided by the cleavage of the carbonoxygen
bond of a polysaccharide in the bacterial cell wall by the enzyme lysozyme [4]. The
bond-breaking event is proposed to proceed as follows (see Figure 7.2). First, the
carboxylic acid side-chain of Glu35 donates a proton to the oxygen of the unstable
CO bond. This forms an intermediate with a positively charged (unstable) carbon
H
2
C
CH
2
CH
2
Asp
102
His
57
Ser
195
C
O
O
O
H
H
N 1
3
N:
R`
O
H
H
H
Water
Active enzyme
Active enzyme
New C-terminus
of cleaved polypeptide
chain
H
O
R
R
C
C
O
O
N
CH
2
CH
2
Asp
102
Substrate
polypeptide
His
57
Ser
195
C
O
O
O
H
H
+
+
+ +
N 1
N:
H
2
C
R` NH
2
Figure 7.1. Biochemical reaction scheme of chymotrypsin. The amide bond
cleavage takes place in the presence of water.
Figure 7.2. Schematic illustration of the reaction scheme in the active site of
lysozyme. The glycosidic CO bond is cleaved in the rst phase of the reaction
process in the presence of a glutamic acidic hydrogen atom forming a positively
charged intermediate carbocation. The localized water molecule stabilizes the
intermediate carbocation by donating an OH
. The additional H
+
ion then
combines with the COO
of Glu35 to neutralize its negative charge.

atom. Asp52 acts to stabilize this intermediate either through electrostatic interactions
or by covalently binding to the carbon. The localized water molecule in the binding
pocket of the lysozyme now gets into the act. It stabilizes the positively charged
intermediate by donating an OH
, which attacks the carbon to yield the hydrolyzed

end product. The extra H
+
ion then combines with the COO
of Glu35, replenishing
the proton that was lost earlier. In this case a single water molecule stabilizes the high-
energy doubly charged separated state by donating one proton to negatively charged
glutamate-35 and the negatively charged hydroxyl group to stabilize the positively
charged cleaved part of the polysaccharide.
In their multiple direct roles, water molecules can lower the activation energy by
stabilizing the transition state by hydrogen-bonding and then providing a proton and
a hydroxyl ion. In a more indirect fashion, they can capture and direct the substrate
towards the active site for catalysis.
Aremarkable example of the latter is provided by adenylate kinase (ADK), where
a study has suggested that water molecules help stabilize a half-open-half-closed
(HOHC) state that facilitates substrate capture subsequent to product release (see
Figure 7.3). Here two reaction coordinates are necessary to describe the entire
process. These coordinates are the amplitude of the opening of the LID domain
(through which an ATP molecule enters) and the amplitude of opening of the NMP
domain (through which an AMP molecule enters). See Figure 7.3(a) for an illustra-
tion. So, these two coordinates cycle through large and small values as the enzyme
goes through its cycle during conversion of an ATP and an AMP to two ADP
molecules. As mentioned, water can stabilize an intermediate state that requires less
conformational movement and lower activation free energy for the continuous
catalytic cycle (see Figure 7.3(b)) [5].
The above constitute only a few examples of the extensive role of water in
enzyme catalysis. Water probably controls many aspects of enzyme catalysis that
continuously occur within biological cells and the number of such instances is
enormous. However, our understanding of these phenomena remains poor in most
cases, although failure of any enzymatic action can lead to serious consequences.
7.3 Role of water in drugDNA intercalation
Although a DNA molecule is heavily solvated by water (and ions), a detailed study
of the role of water in various biological functions of DNA has yet to be carried out.
However, some progress has recently been made towards understanding the role of
water in the intercalation of anti-tumor drugs (such as daunomycin, actinomycin,
and cisplatin; see Figure 7.4) into DNA. This is an important problem from a
medical point of view and at the same time the microscopic aspects turned out to
be quite interesting.
Studies have suggested the following sequence of events. Key DNA structural
changes involve the opening of the prospective intercalation site base pairs toward the
minor groove, followed by an increase in the vertical separation between the base pairs
involved, accompanied by hydrogen-bonding changes of the minor groove water
(a)
M
i
l
l
i
s
e
c
o
n
d
C
o
n
f
o
r
m
a
t
i
o
n
a
l

d
y
n
a
m
i
c
s
HOHC
Catalytic cycle
C
20.5
(b)
1
8
.
3
1
9
.
0
2
1
.
0
26.0 29.5
R
CM
()
LID-CORE
R
C
M
(
)
N
M
P
-
C
O
R
E
Figure 7.3. (a) X-ray crystal structures of adenylate kinase (ADK) showing the open
(denoted by O) and closed (denoted by C) states of the enzyme. See plate section
for color version. (b) Schematic free-energy contour diagram presented for the
different conformational states of ADK. The reaction takes place in the state where
both the LID and the NMP domains are closed, with the two substrates (ATP and
AMP) inside. After the phosphate transfer, the product molecules (two ADP) get
released by opening of the domains. The proposed catalytic cycle involves closed (C)
and half-open-half-closed (HOHC) states, and not the full open state. The HOHC
state is stabilized by water. The millisecond conformational uctuation observed in
the experiment may involve uctuation between HOHC and the fully open state.
Adapted with permission fromJ. Phys. Chem. A, 115 (2011), 36913697. Copyright
molecules. See Figure 7.5 for an illustration of the structural changes involved. In the
recent past, osmotic stress experiments on the intercalation of various drugs have
indicated an uptake of a rather large number of water molecules. In the case of
daunomycin, it has been estimated that DNA takes up to 18 water molecules [6].
Figure 7.6 depicts the free-energy landscape recently predicted for the intercalation
reaction. In this case, the drug rst settles down near the minor groove and waits for a
Figure 7.4. Pictorial representation of the structure of a few anti-tumor drugs:
daunomycin, actinomycin, and cisplatin.
Figure 7.5. (a) Schematic presentation for the angle for (a) the minor groove-
bound state, (b) the critical region, and (c) the intercalated state. The DNA is
represented as a stack of bars and the drug is shown as a combination of two
ellipses. is the angle for the dot product of unit vectors bt and pt. Adapted with
permission from J. Am. Chem. Soc., 130, (2008), 97479755. Copyright (2008)
high-energy uctuation for the gate to open for intercalation (see Figure 7.6). Water
molecules play an important role in lowering the energy cost of this uctuation [7].
Experimental study has indicated that the net change in entropy in the process of
intercalation of daunomycin into DNA is almost negligible, only 1.1 kcal/mole,
which is indeed surprisingly small for such a complex reaction [8]. Subsequent
theoretical analyses showed that this small net change is due to a combination of
steps which contribute to the observed total change (see Table 7.1). The loss of
Table 7.1. Estimation of overall entropy change for the intercalation process.
Adapted with permission from J. Phys. Chem. Lett., 2 (2011), 30213026.
Copyright (2011) American Chemical Society.
TS (kcal/mol) Intercalated state Minor groove-bound state Experimental
Daunomycin-DNA
(vibrational)
14.4 15.8
Water (vibrational) 12.7 4.3
Translation 14.6 14.6
Rotation 11.4 11.4
Total entropy change 1.1 37.5 1.1
18
( )
X(A)
F
r
e
e

E
n
e
r
g
y

(
k
c
a
l
/
m
o
l
)
Minor Groovebound State
Intercalated State
Critical Region (Gate) Separated State
2
4
6
8
10
12
14
16
40
80
120
160
12
10
8
6
4
2
0
2
14
12
10
8
6
4
2
0
2
Figure 7.6. Free-energy landscape in the umbrella sampling coordinate X(separation
between DNA intercalation site and the drug) and the angle (see Figure 7.5). The
intercalated, minor groove-bound, and separated states are specied, as is the critical
region. The green arrowed dashed lines (and their red projections onto the X- plane)
are schematic guides to illustrate the most probable path from separated minor
groove-bound intercalated state through the critical region. Adapted with
permission from J. Am. Chem. Soc., 130, (2008), 97479755. Copyright (2008)
American Chemical Society. See plate section for color version.
entropy of the drug (which is negative) is largely compensated by the change in
entropy of water, which is positive due to the release to the bulk of some of the water
molecules that surround the drug (see Figure 7.7). Such a situation has been
envisaged before but it has now been demonstrated quantitatively. In fact, water
also plays a role in facilitating the formation of the transition state by solvating it in
an appropriate fashion [9].
7.4 Role of water in the biological function of RNA
Several experimental studies show that trapped water molecules inside RNA have a
lot of biological importance. In 2006, Walter and co-workers showed that informa-
tion about structural change in a distant part of the RNA travels to the active site of
the molecule. Such an exchange of information takes place through a hydrogen-
bonding network formed by entrapped water molecules inside the RNA [10].
By performing molecular dynamics simulation studies, Sykes and Levitt
observed that the stability of RNA base pairs largely depends on the degree of
solvation of RNA (see Figure 7.8) [11]. If the number of water molecules is
inadequate for solvation then, due to insufcient inter-water HBs, water starts to
form H bonds with WatsonCrick bases of the DNA, decreasing the stability of the
base pairs by disrupting inter-base HBs. However, in the presence of a sufcient
amount of water, the number of HBs found to disrupt these base pairs becomes less
(see Figure 7.9). As a consequence, the base pairs exist in a stability minimum of
20100 water molecules, the upper limit of which corresponds to the approximate
number of water molecules contained in the rst hydration shell.
S = S
Drug
+

S
DNA
+ S
Water
Figure 7.7. Schematic representationof the drugDNAintercalationprocess toexplore
the entropyenthalpy balance in such a process. Here the double-stranded DNA is
represented in gray, the drug molecule is represented in black and the small gray circles
are the surrounding water molecules. Adapted with permission from J. Phys. Chem.
Lett., 2 (2011), 30213026. Copyright (2011) American Chemical Society.
7.4 Role of water in the biological function of RNA 105
The aminoacylation reaction of t-RNA is a key step in the translation of genetic
information. The general molecular mechanism reported by Moras and co-workers
highlights the role of water molecules in the recognition of t-RNA, which entails
water-mediated hydrogen-bond interactions [12]. They showed that more than 20
Figure 7.8. The average number of available HBs to disrupt the base pairs for
different sizes of nano-droplets and the AU base pair. This correlates well to the
stability of the base pairs. Adapted with permission from Proc. Natl. Acad. Sci.
USA, 104 (2007), 1233612340. Copyright (2007) Proc. Natl. Acad. Sci. USA.
Figure 7.9. Total amount of time spent in base-paired conformation over the entire
simulation normalized by the number of HBs. Adapted with permission fromProc.
Natl. Acad. Sci. USA, 104 (2007), 1233612340. Copyright (2007) Proc. Natl.
Acad. Sci. USA.
water molecules build a shell at the interface between the insertion domain and the
tRNA through a network of HBs, among which only three are direct proteinwater
RNA contacts. Here the layer of water molecules favors a dynamic recognition by
the formation of a network of nonspecic and versatile water-mediated interactions.
7.5 Water-mediated molecular recognition
Recognition of the binding sites of proteins by ligands, inhibitors, substrates, and
other proteins is crucial for the biological activity of proteins. As a ligand
approaches a protein, an important step toward binding may occur in a short time
(in a few tens of picoseconds) when the interfacial water molecules may determine
the course, the rate, and even the outcome of binding. For example, interaction
between the binding site of protein and ligand molecules depends upon the release
of the localized water molecules.
A microscopic theory of molecular recognition would need to discuss the free-
energy barrier (or rather, the free-energy landscape) that is experienced by the
incoming ligand at the surface. Experiments by Pal and Zewail have provided an
indication of the need for such a molecular-level description [13]. In their study of
molecular recognition by a protein mimic, the cobalt picket fence porphyrin, Pal and
Zewail found the need to assume an energy landscape which involved two barriers.
The rst step was the absorption of O
2
in the hydration layer of the protein, which
was followed by the subsequent binding. This may be a common mechanism in
many other cases. A more microscopic treatment of such phenomena requires
inclusion of hydration dynamics at the interface.
In the study of DNA recognition of the drug, the timescale of the twisting and
bending of DNA and dynamics of water in the hydration layer determine the
efciency of recognition for a stable double-strand structure in bulk water. The
diffusion of the binding ligand is slow, but at the interface the dynamics of the water
molecules is ultrafast in order to optimize the entropic and enthalpic balance (as we
have encountered in the previous section for the drugDNA intercalation process).
To gain an in-depth understanding of the role of water, the residence time of
ordered water molecules needs to be compared to other time constants of DNA that
are important for the recognition. First, it is necessary to consider the timescale of
making and breaking bonds of the dynamically ordered water (
DOW
) relative to that
of conformational changes (
CC
). The latter involves bending and twisting. But the
last process again involves the structurally ordered water molecules!
If
DOW
is much smaller than
CC
, then recognition become an efcient procedure
with structural integrity. The loss of order in the picosecond timescale is signicant
in changing the entropy, and this contribution to the free energy is possibly
controlled by the change in the rotations of water molecules.
7.5 Water-mediated molecular recognition 107
It is necessary to compare the residence time of weakly bound water (
DOW
) with
the timescale of the breaking or making of HBs in bulk water (
HB
). With a barrier of a
few kilocalories per mole, kinetically
HB
is of the order of a fewpicoseconds, and for
an efcient recognition
DOW
should not be orders of magnitude larger than the value
of
HB
, so that the efciency becomes favorable. If the ratio
DOW
/
HB
becomes
almost 1, then the degree of order is that of the bulk. Last, we need consider the
timescale for the motion of the drug in the groove (by orientational diffusion),
OD
,
relative to
DOW
. For the drug considered here,
OD
is much longer than
DOW
for both
types of DNA studied (dodecamer and calf thymus; see Figure 7.10)
Studies using densimetric and ultrasonic measurements have shown that
depending on the base sequence, a minor groove-binding drug, netropsin, dis-
places a signicant number of water molecules upon complexation with a DNA
duplex. From a structural point of view, the drug studied here binds to the minor
groove, facing the sequence AATTT of the central A-tract, with the piperazine
group close to one of the GC regions. It involves two three-centered HBs from
the nitrogen atoms of benzimidazole rings to the N (A18) and O (T7, T8, T19)
atoms of the DNA bases. This hydrogen-bonding (and electrostatic/dispersion
interactions) is facilitated by the existence of ordered water around the drug
molecule. If ordered water is involved in direct binding of the drug, then
Time Delay (ps)
DNA, Calf Thymus
DNA, Dodecamer
Time Scales
BW
DOW
CC
SW
1 ps
20 ps
450 ps
ns or longer
H
y
d
r
a
t
i
o
n

C
o
r
r
e
l
a
t
i
o
n

F
u
n
c
t
i
o
n
,

C
(
t
)
O
O
O
0.0
0.2
0.4
0.6
0.8
1.0
Bulk Water (Buffer)
0 20 40 60 80 100
Figure 7.10. Comparison of the hydration correlation functions for the drugDNA
(for dodecamer and calf thymus) complexes in aqueous buffer solutions. The result
for the drug in bulk water (buffer) has also been included for comparison. Here the
structured water molecules (represented as white circles) with the drug in the minor
groove are shown in the inset. In the inset, residence times are represented by
BW
for the bulk water,
DOW
for the dynamically ordered water,
CC
for conformational
change water, and
SW
for structural water. Adapted with permission from Chem.
Rev., 104 (2004), 20992123. Copyright (2004) American Chemical Society.
enthalpic contributions must also be considered. Hydrogen-bonding is also
possible for the drug with water close to the interface of the groove that is
more of a bulk type.
7.6 Protein folding and protein association: role of biological water
The dynamics of water around an extended, unfolded protein certainly plays an
important role in determining the rate of protein folding. For example, hydrophobic
collapse involves movement of water molecules away from the region between two
hydrophobic amino acid residues that form pair contact. Similarly, bends (an
important secondary structure of protein) also involve water mediation. In both of
the examples, the water molecules in close proximity to the protein amino acids are
expected to play a critical role through a subtle balance between enthalpic and
entropic forces.
The hydrophobic effect is considered to have a major control in the folding
process of globular proteins (we discuss the hydrophobic effect later in Chapter 15).
In consequence we nd the burial of the hydrophobic amino acid side-chains in
the core of the protein. Water tends to form ordered cages around non-polar groups,
which is called hydrophobic hydration, which leads to a reduction in the entropy of
the system. These surrounding water molecules regain entropy when they are
released after hydrophobic groups come in contact with each other. This contributes
to the free energy of stabilization of the protein [14]. However, we have discussed
several aspects of the hydrophobic effect in Chapter 5.
Water molecules in the protein hydration layer have a nite residence time. No
single water molecule stays in the layer forever, as it makes sojourn between the
layer and the bulk. This residence time can play a critical role in protein
association because it offers a quantitative estimate of the rigidity of the layer.
The nal act of association of two proteins may require partial desolvation
around the necessary amino acid residue sites. This is only possible if the
residence time of water around these sites is sufciently short. The residence
time is determined by the dynamics in the hydration layer. This correlation
between hydration layer and protein association is an important problem that
deserves careful attention.
7.7 Role of water in beta-amyloid aggregation in Alzheimer disease
Alzheimer disease is a severe neurodegenerative disorder, which is associated
with the premature death of neuronal cells. The disease gives rise to acute memory
loss, as a result of critical damage of brain tissue, and almost always leads to the
death of the patient in the process. Huge numbers of theoretical and experimental
studies have been carried out throughout the last few decades to nd out the
reason for the disease. In fact, investigations at molecular level have provided
interesting insights about the disease. The appearance of lesions or plaque for-
mation inside the brain has been found in the post-mortem brains of people
identied as having Alzheimer disease. These plaques or aggregations are speci-
cally composed of beta-amyloid peptides. Beta-amyloid peptides consist of
4042 amino acids, originated by the unusual proteolytic cleavage of amyloid
precursor protein, which is a human transmembrane protein reportedly needed for
memory.
The beta-amyloid peptide is an amphipathic molecule, and contains a central
hydrophobic cluster as well as charged residues. The sequence of amino acids in the
peptide is given as H
2
N-DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLM
VGGVVIAL-COOH.
The hydrophobic core is composed of residues lysine-16, leucine-17, valine-18,
phenylalanine-19, phenylalanine-20, and the italicized residues signify the hydro-
philic region, which, as discussed below, helps in the formation of a crucial bond
called the salt bridge.
Several studies have conrmed that beta-amyloid peptides undergo structural
transitions to form mobile oligomers that are composed of a particular
aggregation-prone conformation of the peptide. Once the oligomers exceed cri-
tical size, they nucleate to form protolaments which nally transform to cross-
beta sheets or brils that are responsible for the formation of extracellular amyloid
plaques. The tendency to form beta-strands is due to its ability to stabilize the
beta-turn by a salt bridge between residues aspartic acid-23 and lysine-28 and the
hydrophobic region.
It has been found in recent studies that water plays a major role in the process of
beta-amyloid aggregation [15] in the early and late stages of bril formation and
interactions with water greatly affect the folding energy landscapes of beta-amyloid
monomers [15,16].
7.7.1 Role of water in the early stages of oligomer formation
The structures of monomeric beta-amyloids in the brils satisfy the amyloid self-
organization principle, according to which bril stability is enhanced by maximiz-
ing the number of hydrophobic contacts and favorable electrostatic interactions
(formation of salt bridges and HBs) [17]. For A
1035
this principle suggests that the
formation of an intramolecular salt bridge between Asp23 and Lys28 (D23-K28)
may play an important role in the early stage of self-association of beta-amyloid
proteins.
By molecular dynamics simulation in explicit water, the chances of the sponta-
neous formation of the D23-K28 interaction in the isolated A
1035
monomer have
recently been explored [18]. The simulation results reveal that the folding landscape
of A
1035
can be partitioned into four basins that are separated by free-energy
barriers ranging from 0.3 kcalmol
1
to 2.7 kcalmol
1
. The disrupted salt bridge is
the most stable state. It is also found that burying K28 in the interior of the peptide is
an unfavorable process.
Fromthese results, it is clear that a stable intramolecular salt bridge can only form
if the intervening water molecules can be dragged out, which involves overcoming a
large desolvation barrier. Owing to this desolvation penalty, the structural motif with
a stable turn composed of the residues V
24
GSN
27
and a preformed D23-K28 contact
is a minor component among the simulated structures.
In fact, the extent of solvation in the four free-energy basins is vastly different. Thus
the results suggest that removal of water molecules facilitates the formation of the
intramolecular D23-K28 salt bridge and it is an early event in the oligomerization
process.
7.7.2 Role of water in the late stages of bril growth
The role of water in facilitating or inhibiting the growth of amyloid bril is rather
poorly understood. It has been speculated that the expulsion of water leading to
increase in solvent entropy might be a major driving force for bril formation in A
peptides. In this case, it is important to consider the sequence effect as well as to
account for conformational changes that occur. For example, in the case of A
1622
,
the small-sized random-coil structure expands to form -strand (larger size), which
is an unfavorable process as it involves exposure of the hydrophobic residues to the
solvent along with reduction in conformational entropy.
There are two major routes to assembly of -sheets. In the rst, a spontaneously
formed almost perfectly ordered one-dimensional water structure from the pore is
released into the bulk, thereby forming -sheets. In the alternative case, when
uctuations guide misalignment in the orientation of the -sheets, water release
occurs by leakage through the sides.
Computer simulation studies [19] have explored the association of -sheets in
A
1622
and found that in some of the trajectories water was released at an early
stage before assembly. However, in some of the trajectories, the two processes
mentioned above were found to be coincident. The predominant interactions that
facilitate protolament formation were essentially hydrophobic in nature. The
simulations also agreed with experimental results [20], which showed that the rate
of bril formation increases signicantly on reducing the hydration of aggregating
peptide molecules. It is found that the aggregation rate of A
1622
is maximumwhen
the structure is stabilized in reverse AOT micelles containing the least amount of
free water molecules.
7.8 Role of water in photosynthesis
Photosynthesis is perhaps the most important biochemical pathway known ever.
Almost all life depends on it. Higher plants, algae, some bacteria, and a few protists
and organisms are mostly responsible for this photosynthesis process and are
collectively referred to as photoautotrophs [21]. In this process water coupled
with carbon dioxide plays a central role in the synthesis of sugar from light, with
oxygen as a side product.
6CO
2
6H2O light C
6
H
12
O
6
6O
2
Carbon dioxide Water Light energy Glucose Oxygen
Photosynthesis is an extremely complicated process containing many coordinated
biochemical reactions. Principally, it occurs in two phases. In the rst phase photo-
synthetic reactions or light-dependent reactions arrest the energy of light and exploit
it to make high-energy molecules. During the second phase, the light-independent
reactions or the dark reactions use those high-energy molecules to capture carbon
dioxide (CO
2
) and make the precursors of glucose.
In the light-phase reaction light energy is converted to chemical energy in the
form of ATP (adenosine triphosphate) and NADPH (nicotinamide adenine dinu-
cleotide phosphate), which is utilized for synthetic reactions in photoautotrophs.
In the rst step, one molecule of the pigment chlorophyll absorbs one photon and
loses one electron. This electron is passed through an electron transport chain
(specically it is called Z-scheme, shown in Figure 7.11) that leads to the ultimate
reduction of NADP to NADPH. The chlorophyll molecule regains the lost
electron from a water molecule through a process called water photolysis,
which releases an oxygen molecule. So, the source of electrons in green-plant
and cyanobacterial photosynthesis is water. The overall equation for the light-
dependent reactions under the conditions of non-cyclic electron ow in green
plants is
2 H
2
O 2 NADP
3ADP 3P
i
light 2NADPH 2H
3ATP O
2
Two water molecules are oxidized by four consecutive charge-separation reac-
tions through photosystem II to form a molecule of diatomic oxygen and
four hydrogen ions. The outcoming electron in each step is transferred to
a redox-active tyrosine residue followed by the reduction of a photoxidized
paired-chlorophyll called P680 that acts as the primary (photon-driven) electron
donor in the photosystem II reaction center. The oxidation of water is catalyzed in
photosystem II by a redox-active composition that involves four manganese ions
and a calcium ion. This oxygen-evolving complex binds two water molecules and
stores the four oxidizing equivalents that are essential to drive the water-oxidizing
reaction. Photosystem II is the only known biological process that involves this
oxidation of water. The hydrogen ions contribute to the transmembrane chemi-
osmotic potential that causes ATP synthesis. Though oxygen is a side product of
the light-phase process, the majority of organisms on Earth, including photosyn-
thetic organisms, utilize oxygen for cellular respiration.
In the light-independent or dark reactions the enzyme RuBisCO (ribulose-1,5-
bisphosphate carboxylase oxygenase) consumes CO
2
from the atmosphere and in a
complex process called the CalvinBenson cycle releases three-carbon sugars that
are subsequently combined to form glucose.
3 CO
2
9 ATP 6 NADPH 6 H
C
3
H
6
O
3
-phosphate 9 ADP 8 P
i
6 NADP
3 H
2
O
Figure 7.11. A simplied view of the light reaction photosynthetic Z-scheme shown
in the diagram. In the rst steps of the Z-scheme, an external source of electrons is
required to reduce its oxidized chlorophyll molecules. The source of electrons is
water. Water is oxidized as a result of the light reaction of photosystem II. From
photosystem II, electrons pass through the electron transport chain (redox chain) and
energy released along this part allows the formation of ATP. Another light reaction at
photosystemI when activated transfers electrons to NADP+, where the protons from
water splitting are used to formNADPH. Adapted with permission fromNature, 445
(2007), 610612. Copyright (2007) Nature Publishing Group.
7.8 Role of water in photosynthesis 113
7.9 Conclusion
Despite the explicit role of water in many biological and cellular functions, we still
have a rather limited understanding of the details of the role of water in life
processes. Difculty clearly arises fromthe enormous complexity of these processes
and difculty of carrying out in vivo experiments. Fortunately, signicant progress
has been made in the last decade. In particular, the emergence of single-molecule
spectroscopic techniques has proven to be highly useful in our attempt to study
complex systems.
An aspect we have repeatedly emphasized is the consumption, as a chemical, of
water molecules in biological processes. This is important as we know that lack of
continuous input of water into our body can cause many serious illnesses. This
consumption of water largely occurs in enzymatic reactions, in protein synthesis,
and many other crucial processes. This is different from other stabilizing or lubri-
cating effects of water.
In the next fewchapters we shall discuss the role of water in biopolymers and self-
organized assemblies.
References
1. F. Despa, Biological water: its vital role in macromolecular structure and function. Ann
NY Acad Sci, 1066 (2005), 111.
2. H. Schiweck, M. Clarke, and G. Pollach, Sugar in Ullmanns Encyclopedia of
Industrial Chemistry (Wiley-VCH, Weinheim, 2007), doi:10.1002/14356007.
a25_345.pub2.
3. W. W. Appel, Chymotrypsin: molecular and catalytic properties. Clin. Biochem., 19
(1986), 31722.
4. D. C. Phillips, The hen egg-white lysozyme molecule. Proc. Natl. Acad. Sci. USA, 57
(1967), 484495.
5. Bharat V. Adkar, B. Jana, and B. Bagchi, Role of water in the enzymatic catalysis: study
of ATP + AMP 2ADP conversion by adenylate kinase. J. Phys. Chem. A, 115 (2011),
36913697.
6. X. Qu and J. B. Chaires, Contrasting hydration changes for ethidium and daunomycin
binding to DNA. J. Am. Chem. Soc., 121 (1999), 26492650.
7. A. Mukherjee, R. Lavery, B. Bagchi, and J. T. Hynes, On the molecular mechanism of
drug intercalation into DNA: a simulation study of the intercalation pathway, free
energy, and DNA structural changes. J. Am. Chem. Soc., 130 (2008), 97479755
8. J. B. Chaires, Athermodynamic signature for drug-DNAbinding mode. Arch. Biochem.
Biophys., 453 (2006), 2631.
9. A. Mukherjee, Entropy balance in the intercalation process of an anti-cancer drug
Daunomycin. J. Phys. Chem. Lett., 2 (2011), 30213026.
10. M. M. Rhodes, K. Rblov, J. Spooner, and N. G. Walter, Trapped water molecules are
essential to structural dynamics and function of a ribozyme. Proc. Natl. Acad. Sci. USA,
103 (2006), 13380
11. M. T. Sykes and M. Levitt, Simulations of RNA base pairs in a nanodroplet reveal
solvation-dependent stability. Proc. Natl. Acad. Sci. USA, 104 (2007), 1233612340.
12. S. Eiler, A. C. Dock-Bregeon, L. Moulinier, J. C. Thierry, and D. Moras, Synthesis of
aspartyl-tRNA (Asp) in Escherichia coli a snapshot of the second step. EMBO J., 18
(1999), 65326541.
13 S. K. Pal and A. H. Zewail, Dynamics of water in biological recognition. Chem. Rev. 104
(2004), 20992123.
14. A. Ben-Naim, The role of hydrogen bonds in protein folding and protein association.
J. Phys. Chem., 95 (1991), 14371444.
15. E. J. Straub, and D. Thirumalai, Toward a molecular theory of early and late events in
monomer to amyloid bril formation. Annu. Rev. Phys. Chem., 62 (2011), 43763.
16. D. Thirumalai, G. Reddy, and E. J. Straub, Role of water in protein aggregation and
amyloid polymorphism. Acc. Chem. Res., 45 (2012), 8392.
17. B. Tarus, J. E. Straub, and D. Thirumalai, Dynamics of Asp23-Lys28 salt-bridge
formation in A 1035 monomers. J. Am. Chem. Soc., 128 (2006), 1615968.
18. B. Tarus, J. E. Straub, and D. Thirumalai, Probing the initial stage of aggregation of the
A1035-protein: assessing the propensity for peptide dimerizatio. J. Mol. Biol., 345
(2005), 114156.
19. M. G. Krone, L. Hua, P. Soto, R. Zhou, B. J. Berne, and J. E. Shea, Role of water in
mediating the assembly of Alzheimer amyloid-beta Abeta1622 protolaments. J. Am.
Chem. Soc., 130 (2008), 1106611072.
20. S. Mukherjee, P. Chowdhury, and F. Gai, Effect of dehydration on the aggregation
kinetics of two amyloid peptides. J. Phys. Chem. B, 113 (2009), 531535.
21. A. L. Smith, Photosynthesis the synthesis by organisms of organic chemical com-
pounds, esp. carbohydrates, from carbon dioxide using energy obtained from light
rather than the oxidation of chemical compounds. Oxford Dictionary of Biochemistry
and Molecular Biology (Oxford: Oxford University Press, 1997), p. 508.
References 115
8
Hydration of proteins
In the preceding chapter we discussed only a few of the amazing array of
phenomena known where water molecules assist and even control the
biological functions of biopolymers. In this chapter we shall discuss
some exclusive features of the protein hydration layer with an emphasis
on the dynamics of the local structure. The unique features of water
discussed in Chapter 1 (section 1.3) manifest themselves in various
ways in the protein hydration layer. Their small size, ability to act
simultaneously as electron donor and acceptor, large dipole moment
and high mobility allow interfacial water molecules to perform and
also assist biomolecules to perform in wide-ranging functions. Not
only are the water molecules responsible for benecial functions,
they are also involved in protein association which could sometimes be
quite harmful (like association of beta amyloids in Alzheimer disease).
The study of the protein hydration layer has therefore been a subject of
many discussions in the past. However, a molecular-level quantitative
understanding has begun to emerge only recently in the last decade or so.
The emphasis of this chapter is on the dynamics of water molecules in
the layer.
8.1 Introduction
Water in and around protein and DNA, and within cells and tissues, has been termed
biological water [1]. Unlike bulk water, here water molecules are either conned or
spatially restricted and subjected to strong external inuence that modies many of
the properties of water. The movement of water molecules in the bulk has only a
transient connection to its nearest neighbors that gets broken after a short while and
neighbors get changed. In contrast, the movement of water in the vicinity of a
protein is more coordinated and the bond with the closest partner is maintained for a
longer time. Also, the extended hydrogen-bond network of water omnipresent in the
bulk becomes compromised on the surface. Now a water molecule can reside and
117
exchange only between several different states. The structure and also the dynamics
become heterogeneous. As discussed in the previous chapter, this heterogeneity is
an essential ingredient of waters role in biological functions. Water near hydro-
phobic and hydrophilic regions behaves quite differently, and performs quite dif-
ferent functions.
In this chapter, we shall attempt to present a molecular view of the interesting
characteristics of water present in the hydration layer that surrounds a protein. As
often discussed, water not only renders stability to the unique native state of a
protein, but the hydration layer surrounding it also controls all its biological activity
(see Figure 8.1 for a schematic representation of an aqueous protein solution). To
put the role of water in proper perspective, we can try to imagine an enzyme
performing its catalytic activity without water molecules. We recognize at once
that it is entirely impossible. As 70% of a biological cell in the human body consists
of water (the rest is ions and other biomolecules), the function of a protein has been
ne-tuned over many years of evolution to maximize its stability (and activity) with
the help of water. That is, water has been a partner all along.
It is, however, still not possible to study proteins directly within the cell and this
remains a serious limitation. Much of our knowledge of water in and around
proteins comes from studies of the latter in neat water (in vitro), sometimes in the
presence of cosolvent and ions, and is, therefore, insufcient to understand water
within biological cells.
However, even in such an over-simplied in vitro situation, our understanding of
the hydration shell has been slow to develop. For example, even the issue of the
width of the hydration shell has remained controversial. The structure and dynamics
of water in the hydration shell are dependent on its width and these quantities are
intimately related to functions such as ligand binding, molecular recognition,
enzyme kinetics, and protein association.
In the older (now discarded) view the hydration layer was considered as a rigid
iceberg surrounding the protein. This picture is depicted in Figure 8.1.
However, this iceberg model was discarded when many experiments showed
that while water slowed down somewhat at the surface it never slows down to the
extent that qualied it to be called an iceberg.
8.2 What is the thickness of the hydration shell?
The thickness of a protein hydration shell has long been an object of study, and also
a topic of lively debate. However, even this apparently simple question cannot be
answered in a straightforward way. Estimates of the thickness of the hydration shell
vary from one to two monolayers to 10 monolayers, depending on the experimental
probe used.
118 Hydration of proteins
The thickness of the hydration layer should be related clearly to the size of the
protein itself, with a shallow layer for a small protein (such as chicken villin
head-piece, known as HP36) and a thicker layer for a large protein (such as
adenylate kinase, known as ADK, and hemoglobin). If we denote this thickness
by l
H
, then we can quantitatively discuss the thickness in terms of the ratio R
th
dened by
R
th
= l
H
=R
P
( ) (8:1)
where R
P
is the radius of the protein. For small proteins, this ratio is small, less than
unity. For large proteins one expects the ratio to reach a constant value which is
again smaller than unity. Structural studies seem to provide a smaller value of R
th
than the dynamic studies. It is not prudent to discuss l
H
without considering R
P
, and
also the nature of the protein.
Protein
Hydration
Layer
Bulk
Figure 8.1. Schematic representation of the old view of a protein molecule in
aqueous environment, with a layer of strongly associated water (the hydration
layer is an iceberg), suspended in aqueous solution. The hydration layer moves
with the protein molecule (as proposed by the iceberg model), and beyond this
layer the water molecules adapt to the normal tetrahedral geometry. Adapted
with permission from Chem. Rev., 104 (2004), 20992123. Copyright (2004)
8.2 What is the thickness of the hydration shell? 119
Let us rst discuss estimates from DR measurements that provide several
important pieces of information. These experiments measure the frequency-
dependent dielectric constant and provide a measure of a liquids polarization
response at different frequencies. In bulk water, we have two dominant regions.
The low-frequency dispersion gives us the well-known Debye relaxation time,
D
,
which is equal to 8.3ps. There is a second prominent dispersion in the high-
frequency side with relaxation time constant less than 1ps which contains
combined contributions from low-frequency intermolecular vibrations and libra-
tion. Aqueous protein solutions exhibit at least two more dispersions. (i) A new
dispersion at intermediate frequencies, called, dispersion, which appears at a
timescale of about 50 ps in the dielectric spectrum, seems to be present in most
protein solutions. This additional dispersion is attributed to water in the hydration
layer. (ii) Another dispersion is present at very low frequencies and is attributed to
the rotation of the protein.
Dielectric relaxation can thus provide important information about the rotational
time constant of proteins which in turn provides valuable information about the
thickness of the protein hydration layer, as discussed below.
Experimentally the rotational time constant of the protein is found to vary from a
few microseconds to a few nanoseconds, depending on the size of the protein. The
rotational time constant is proportional to the rotational friction. Now the rotational
friction (
R
) of protein in solution can be estimated from the hydrodynamics that
relates this friction to the radius of the protein and the viscosity of the medium. This
hydrodynamic expression is given by
R
= 8R
3
H
(8:2)
where R
H
is the hydrodynamic radius of the protein. Now, if one uses the crystal-
lographic radius of the protein in the above equation, one nds that the correspond-
ing friction is less than the experimental value. Better agreement with the
experiment is achieved by adding a hydration layer of 34 thick to the crystal-
lographic radius of the protein. In terms of protein thickness l
H
, the hydrodynamic
radius should be given by
R
H
= R
P
l
H
(8:3)
So, most studies of rotational friction provide an estimate of l
H
34 . This would
be a lower estimate because the water layer is assumed here to be fully rigid and a
part of the protein molecule.
The above procedure provided historically the rst estimate of the width of the
protein hydration layer. The estimate (34 ) so obtained was believed to be fairly
accurate for a long time, till newer time-dependent studies and computer simulations
became available.
So, the story is a bit more elaborate than presented above. The total rotational
friction is a sum of two contributions a hydrodynamic friction due to viscosity and
a dielectric friction due to the charges on the protein surface and the polarity of the
water molecules. The rotational dielectric friction on the protein (
DF
) was calcu-
lated by using a generalized arbitrary charge-distribution model (where the charges
were obtained from quantum chemical calculation). The hydrodynamic friction
(with stick boundary condition) was obtained by solving the NavierStokes equation
by using the sophisticated theoretical technique known as the tri-axial ellipsoidal
method (
TR
) [2]. The calculation of hydrodynamic friction was carried out with
only the dry volume of the protein (no hydration layer). Now the total friction
obtained by summing up
DF
and
TR
gave quantitative agreement with the experi-
mental results, if the crystallographic radius was augmented by 3 . But such an
estimate needed the inclusion of the dielectric friction! This implies that the layer
could be thicker than 34 .
Computer simulation studies show that the hydration layer of smaller proteins
like HP36 extends only up to the rst layer of solvent. For larger proteins, as
emphasized earlier, researchers have concluded from their simulations that the
hydration layer around the protein has a thickness of more than 10, which amounts
to more than a three-monolayer thickness.
Recent studies by terahertz spectroscopy demonstrated clearly for the rst time
that proteins inuence the rate of movement of the surrounding water network over
a broader area [3]. The researchers took advantage of the fact that the vibrations of
water networks change not only as a result of the temperature, but also in response to
the proximity of proteins, as the water in the vicinity of proteins permits less
permeation of THz radiation. This phenomenon makes it possible to directly
observe the effects of proteins on water. The researchers can decide on the
state of water from the amount of the absorbed radiation. These measurements
demonstrated that proteins affect the rapid movements of the water network over a
broad area. Around 1000 water molecules were estimated to be inuenced by one
protein [3]. Such a far-reaching effect, extending up to a distance of 15 to 20 , had
earlier been predicted in simulations.
8.3 How structured is the water in the hydration shell of a protein?
The molecular arrangement of bulk water undergoes change near the protein sur-
face. The surface of a protein contains both hydrophilic and hydrophobic residues.
While the hydrophilic residues consist of mostly charged and polar atoms, hydro-
phobic residues consist of non-polar atoms. Water can form strong HBs with the
polar or charged surface atoms. Hydrophobic surface atoms are not capable of
forming such hydrogen-bonding. Another important factor of the protein surface
8.3 How structured is the water in the hydration shell of a protein? 121
is that it has a heterogeneous complex topology. This means the local geometry of
the surface can be different in different parts of protein. Thus, the heterogeneity of
the charge distribution and the surface topology are expected to have a crucial
inuence on the local structure of water molecules at the surface.
A water molecule around a hydrophilic surface can form an HB using both its
oxygen and hydrogen atoms. If the surface atom is negatively charged, it forms
hydrogen-bonding with the hydrogen atom of the water molecule. The water
molecule forms a hydrogen-bond through an oxygen atom if the surface atom is
positively charged.
Studies have shown that the proteinwater hydrogen-bonding strength is different
for backbone polar atoms from that formed with the side-chain polar atoms. To
understand the structural stability of HBs, the distributions of the electrostatic
energy for the HBs involving backbone polar atoms (BBO-W) and side-chain
polar atoms (SCO-W) were calculated for lysozyme (Figure 8.2). While the dis-
tribution for SCOWbonds was found to have a two-peak structure, the distribution
of electrostatic energy for BBOW bonds was single-peaked and also in the lower-
energy (less stable HB) region. Interestingly the lower-energy peak position of
SCOWand the peak position of BBOW were found to be same.
This character of the electrostatic energy distribution of SCOW and BBOW
HBs was found to be quite universal. The lower-energy peak was assigned to the
HBs where side-chain oxygen was involved in intramolecular hydrogen-bonding
Figure 8.2. Distribution of electrostatic energy of water molecules (which are
hydrogen-bonded with backbone oxygen and side-chain oxygen atoms) for
lysozyme. Note the bimodal character of the distribution for side-chain oxygen
atoms, indicating the presence of both strongly and weakly hydrogen-bonded water.
Adapted from J. Chem. Sci., 124 (2012), 317325. Copyright (2012) Springer Link.
with the protein hydrogen atoms to maintain its 3D structure. As the intramolecular
HBs were generally found to be strong, the HB with the water molecule was less
stable. The higher-energy peak was assigned to the HBs where side-chain atoms
were not involved in intramolecular hydrogen-bonding with the protein hydrogen
atoms. On the other hand, backbone oxygen atoms were mostly involved in strong
intramolecular HBs to maintain the rigid backbone structure of the protein and as a
result the hydrogen-bonding with water became weaker.
8.4 Orientational arrangement of water molecules at the surface
It has become clear over the years that the orientation of water molecules at the
protein surface determines and reects many of the properties of interfacial water.
There are certainly a few general rules. For example, this orientation is partly
determined by a simple, albeit approximate principle: water conserves its HB.
Therefore, near a hydrophobic surface water points its hydrogen-bonding sites
away from the surface, while near a hydrophilic surface the opposite is true.
Additionally, the orientation created near an extended surface gets propagated
inwards. Because of the strong correlation already present inside bulk water, the
surface-induced perturbation can be signicant, structurally speaking, over a rela-
tively large distance from the surface, even up to 610 molecular diameters.
Many biomolecules are characterized by surfaces containing extended polar
regions and also extended non-polar regions. A well-known example is provided
by beta-amyloid the well-known Alzheimer protein. It has extended hydrophobic
regions separated by hydrophilic regions, as discussed in Chapter 7. The hydration
of extended non-polar planar surfaces may involve novel structures that are orien-
tationally inverted relative to clathrate-like hydration shells, where unsatised HBs
are directed towards the hydrophobic surface. We have discussed these two geo-
metric arrangements in the appendix to this chapter (Appendix 8.A).
The structure of water molecules around the extended hydrophobic surface of a
polypeptide, melittin, was studied by extensive computer simulations. It was found
that interaction between the polypeptide melittin and water leads to two different
hydration structures near the biomolecular surface [4]. The two structures are
distinguished by a substantial difference in the waterwater interaction enthalpy,
and their relative contributions depend strongly on the surface topography of the
melittin molecule: clathrate-like structures dominate near convex hydrophobic sur-
face patches, whereas the hydration shell near at surfaces uctuates between
clathrate-like and less-ordered or inverted structures. The study underlines the
strong inuence of surface topography on the structure and free energy of hydro-
phobic hydration. Here again the ability of a cluster of water molecules to adopt
different polymorphic arrangements comes into play. Many biomolecules contain
8.4 Orientational arrangement of water molecules at the surface 123
surface regions that can be categorized as convex patches, deep or shallow concave
grooves, and roughly planar areas. A specic molecular arrangement of water
molecules could be particularly relevant depending on the topography of the
surface.
Research has also shown that water molecules forma connected two-dimensional
hydrogen-bond network around a protein. It is important to know how stable this
network is near the protein surface. The HB network longevity around hydrophilic
and hydrophobic residues can be explored through the rate of HB breaking. The
hydrogen-bond partner change is much less frequent for the water molecules that are
hydrogen-bonded to a hydrophilic residue, indicating that the HB network is long-
lived near the hydrophilic surface. The partner change is more frequent for the water
molecule that is not hydrogen-bonded to a protein and is around a hydrophobic
residue, indicating that the longevity of the HB network is shorter near a hydro-
phobic surface. The origin of this enhanced longevity of the HB network near a
hydrophilic surface is of course the increased electrostatic interactions between the
polar amino acid residues and the water molecules [5].
8.5 Dynamics of the protein hydration shell: experimental studies
The dynamics of water in the hydration layer that surrounds a protein surface
exhibits both fast and slow dynamics, but quantitative characterization of the
dynamics has proven to be quite difcult. This area has also seen many divergent
views.
As mentioned earlier, the rst signature of the inuence of the protein surface on
the dynamics of water came from the measurements of the rotational and transla-
tional diffusion coefcients of water in aqueous protein solutions. Analysis based on
hydrodynamic formulas (such as StokesEinstein and DebyeStokesEinstein
(DSE)) showed that an explanation of the observed values required a larger than
actual radius of the protein to be used in the Stokes expression of the friction (from
hydrodynamics). This indicated the presence of a substantially rigid water layer
around the protein surface. However, the story turned out to be more complex. We
have already discussed some of these aspects we now turn to a more detailed
discussion of several experimental results.
8.5.1 Dielectric spectrum
Although the frequency dependence of the dielectric spectrum contains a (mostly
orientational) response from all of the molecules (water, biomolecules, and ions) in
the system, assignment to the orientational relaxation of individual species is
possible when they are well separated in the frequency (or time) scales.
The dielectric spectra of aqueous protein solutions exhibit anomalous dielectric
increments where the value of the static dielectric constant of the solution is
signicantly larger than that of pure water. A typical experimental result illustrat-
ing the dielectric increment is shown in Figure 8.3, where the real part of the
frequency-dependent dielectric constant of myoglobin is evident. Both the incre-
ment at zero frequency and the overall shape of this curve have drawn a lot of
attention.
There are three universal features in the DR spectra of aqueous protein
solutions. Two distinct loss peaks are observed, near 510 MHz and 10100
GHz regions [6]. These two well-separated peaks (in the imaginary part of the
frequency-dependent friction) correspond to the protein and the bulk water
orientational relaxations, respectively. The additional high-frequency dispersions,
observed within the range 100 MHz to 10 GHz, are often referred to as
dispersion. While the two peaks near 510 MHz and 100 GHz are quite distinct,
the dispersion that occurs in the plateau region of the dielectric spectra has
relatively less weight. It was originally suggested that the dispersion was partly
due to the contributions from bound water in the hydration layer and also from
the internal motions of myoglobin. Similar results were obtained for other
proteins.
Figure 8.3. The real part of the complex frequency-dependent dielectric function
[()] of aqueous myoglobin solution for different concentrations. Concentrations
are (from top to bottom) 161, 99, and 77 mg/mL at 293.15 K. The symbols denote
experimental results while the solid line is a t to the theory of dynamics exchange
model developed by Nandi and Bagchi. Adapted with permission from J. Phys.
Chem. A, 102 (1998), 82178221. Copyright (1998) American Chemical Society.
8.5.2 Nuclear magnetic resonance studies
Nuclear magnetic resonance (NMR) is a versatile technique, which has been widely
used to study not only the structure but also the rotational (single particle) and
translational dynamics of water molecules, both in the bulk and near a surface. One
way to study the dynamics is to measure the magnetic eld dependence of water
proton spin-lattice relaxation rates with increasing eld strength. The results
obtained in these studies are explained by considering a chemical exchange model
in which the relaxation is assumed to be the weighted average of the relaxation rates
of water molecules free in solution and of those that are presumed to be bound to the
protein and rotated with the rotational correlation time of the protein (the time
constant of protein rotation is signicantly slower compared to bulk water).
A novel way to characterize the motion of the water molecules at the protein
surface is to place an external probe spin label with a large electron spin magnetic
moment, such as nitroxide. The magnetic eld from the unpaired electron is
approximately 1000 times larger than that from the protons, so that it is possible
to isolate the paramagnetic contribution to the water proton relaxation easily. In this
case, the relaxation mechanism is an interaction between the electron magnetic
dipole and the water proton dipole. Because the electron spin relaxation time of the
nitroxide radical is long, the correlation time for the intermolecular coupling is
determined by the relative translational motion of the protonelectron pair. Because
the large protein molecule moves much more slowly than water, the effective
correlation time for the coupling is the translational displacement correlation time
for the water near the nitroxide ion on the protein surface. Measurements of
the water proton relaxation rate over a wide range of magnetic eld strengths permit
extraction of the translational diffusion constant of the water molecules residing
near the nitroxide.
The diffusion constant of the surface water obtained from such studies was found
to be lower than that of the bulk value by a factor of 5. Nevertheless, it remains
orders of magnitude faster than motions in a rigid ice lattice, even in samples
hydrated to levels well below what is generally thought to be the full hydration
level of the protein.
Nuclear Overhauser effect (NOE) is another technique used to study the dynamics
of water near a heterogeneous surface. NOE intensities are modulated by dipole
dipole interactions between protons of protein and water in the hydration layer. This
interaction varies as R
6
, where R is the separation between the two protons.
Measurements of magnetization transfer using NOE have been used to obtain the
residence time of the hydration water. The residence time of water molecules in the
hydration layer immediate to the protein is not easily available by other techniques and
is valuable information in quantifying the rigidity of the layer.
NOE studies predict long residence times, of the order of 300500 ps, for water
molecules in the hydration layer. Such long residence times can be appropriate only
for water molecules strongly bound to the cavity of a protein. In fact, these initial
estimates from NOE have not been properly explained even today. It was pointed
out recently by Halle that all earlier NOE measurements derived signicant con-
tributions from distant water molecules as well, because the number of contributing
water molecules increases as R
2
and the characteristic time for orientational mod-
ulation of the internuclear vector R also increases as R
2
. Thus, earlier estimates from
NOE might not be reliable for the residence time of the water molecules.
More recent nuclear magnetic relaxation dispersion (NMRD) studies involving
water oxygen-17 nd a much shorter residence time, in the range 1050 ps [7].
These results are in better agreement with recent studies that seemed to rule out the
existence of slower dynamics.
However, it is important to note that the MRD experiments do not measure an
explicit time correlation function that could characterize water dynamics occurring
at different timescales. Thus it extracts only the average relaxation time of the
system. It is interesting to note how these recent developments (particularly results
from the NMRD technique and computer simulations) have changed our perception
about the dynamics of the hydration layer, from a rigid ice-like layer to a dynami-
cally mobile, somewhat slower than bulk but still active region.
8.5.3 Quasi-elastic neutron-scattering experiments
Inelastic neutron scattering is a technique that has been widely used both in the
liquid and in the solid states to measure the structure and dynamics at small (that is,
molecular) length scales. In an incoherent inelastic neutron-scattering experiment,
the measured quantity is the self-dynamic structure factor S
s
(Q, ), which gives
information, as in the liquid state, of the self-diffusion coefcient of the water
molecules. S
s
(Q, ) is the Fourier transform of the intermediate self-scattering
function F
s
(Q, t), which is dened by
F
s
(Q; t) =
1
N
N
i=1
e
iQ r
i
t ( )r
i
0 ( ) [ [
_
(8:4)
where r
i
(t) is the position of the ith scatterer at time t and the sum is over all of the
scatterers. The symbol < > denotes averaging over the initial distribution of the
particles.
The dynamics of hydration water was investigated by quasi-elastic neutron-
scattering (QENS) experiments in a completely deuterated penta-alanine peptide
at different levels of hydration (7, 30, 50, and 90%) and of dried powder. The last
contained only one structural water molecule. Even this lonely water molecule was
found to move rather fast, with an orientational correlation time of 2.2 ps only
comparable to that of bulk water in the ambient conditions. Thus, there was no really
slow water molecule in this system [8]!
The absence of slow dynamics in this system can be attributed to the fact that the
penta-alanine peptide does not have any polar side-chain atom which can form a
strong HB with water. With a higher level of hydration, the rotational dynamics of
water approached that of bulk water, again as expected. A QENS study of protein
dynamics was also carried out on the picosecond timescale of a protein, lysozyme
solvated in glycerol at different water contents, h (g water/g lysozyme). For all h, a
well-visible low-frequency vibrational bump was observed. The quasi-elastic scat-
tering can be decomposed into two Lorentzian components, corresponding to
motions with characteristic time constants of 15 ps and 0.8 ps. The 15 ps component
is the slow component, which is in the same range observed in many other experi-
mental studies.
Several other studies have indicated that at the protein interface, water behaves
like a bulk supercooled liquid in the sense that F
s
(Q, t) of water shows a two-step
(fast and slow) relaxation with a plateau in between. The fast relaxation is over in
less than a picosecond while the slow relaxation is approximately a stretched
exponential having a relaxation time in the hundreds of picoseconds or even in
nanoseconds. Water forms a somewhat rigid network structure around the protein as
discussed earlier.
Quasi-elastic neutron-scattering experiments and molecular dynamics simulations
of the hydration water dynamics of N-acetyl-leucine-methylamide (NALMA) solu-
tions as a function of concentration and temperature shownon-Arrhenius translational
dynamics over the temperature range 3 to +37C for a wide range of concentrations
[8]. The hydration dynamics at higher concentrations exhibits good correspondence
with the same signatures of non-Arrhenius behavior and non-exponential dynamics as
those observed for supercooled water well below 20C. This indicates that the
underlying potential-energy surface is rough at high concentrations. These studies
provide useful information about the connection between supercooled liquids and
glasses and its biological importance for proteinwater systems.
8.5.4 Vibrational spectroscopy
Vibrational spectroscopy of water probes the effects of the environment on the OH
(or the OD) stretching mode of water molecules and therefore exhibits exceptional
sensitivity towards hydrogen-bonding. This technique has been used for probing
solutesolvent interactions, for example, the inuence of Na
+
and Cl
ions on
solvent structure in electrolyte solutions. IR and near-IR vibrational spectroscopy
was applied to aqueous or almost dry protein samples to study the dynamics of the
protein hydration layer. Both native and thermally unfolded bovine serum albumin
(BSA) have been studied. BSA dry lms exhibit two kinds of intense and a broad
OH (and OD) sub-bands centered at 3260 and 2840 cm
1
(for OD, at 2350 and
2050 cm
1
, respectively) [9]. The rst of these two bands has been assigned to the
OH (OD) stretch of the water molecules where the H (D) of OH (OD) is
involved in strong hydrogen-bonding with other interfacial water molecules. It has
been suggested that these water molecules form a connected two-dimensional net-
work on the surface. The second band has been assigned to the OH (OD) stretch
where the hydrogen of OH (OD) is involved in hydrogen-bonding with the polar
groups of the protein. It was found that the rst band is much wider than the second
band.
8.5.5 Solvation dynamics
At a protein surface, the time dependence of the solvation energy of a newly created
probe derives contributions from many sources, not only from the surface and the
bulk water molecules but also from the amino acid side-chains and from ions (as
they always tend to be present in experimental systems). This makes the analysis of
the SD of a protein solution difcult.
Fleming and co-workers have combined three-pulse photon echo, time dependent
uorescence Stokes shift (TDFSS), and molecular dynamics simulation to obtain a
global t of the SD of eosin in bulk water that encompasses six orders of timescale
from femtosecond to nanosecond [10]. This global t successfully reproduces the
earlier results of SDwith three broad ranges of timescale: (1) a sub-50 fs component
(~6070%), (2) a fewhundred fs (~2030 %), and (3) near 10 ps (~10 %). When the
same eosin was used as a probe for the SD at the surface of lysozyme and a similar
global t was carried out, several interesting new results were observed. In addition
to the ultrafast (sub-100fs) component observed in bulk water, several slow
components emerged, which were found to depend on the timescale of observation.
Nevertheless, two slow components, one in the 100 ps range and another around
500 ps, were observed in this study. The experimental result of the solvation time
correlation function of eosin in lysozyme is shown in Figure 8.4.
There have been two different interpretations of the slow dynamics observed in
the SD of the lysozyme hydration layer. The rst attributes the intermediate time-
scales (3040 ps) to slow water. Bagchi and co-workers employed the dynamic
exchange model to relate the observed slow dynamics to the timescale of the
uctuation of water in the hydration layer [11]. In an alternative interpretation,
Song et al. used the formulation developed by Song and Marcus that relates the
solvation time correlation function to the DR of the medium. They attributed the
observed slow solvation dynamics to protein side-chain motion, and not to slow
water [12].
In a series of important studies, Zewail and co-workers examined the SD of
excited tryptophan as a natural probe in several proteins by using the TDFSS
technique [13]. The advantage of using tryptophan as a probe was twofold. First,
it was a natural probe, so the conformation of the protein was not disturbed and the
solvation of the native state was explored. Second, one could study proteins where
the tryptophan is partly or fully exposed to water, and so SD studies allowed one to
directly probe the response of biological water. They found a slowcomponent in the
solvation time correlation function, which was in the range 2040 ps. This was more
than an order of magnitude slower than the bulk response.
The solvation dynamics results of Zewail and co-workers are shown in Figure 8.5
for the protein Subtilisin Carlsberg (SC). The inset in the same gure shows faster
solvation when the probe was dansyl-bonded and placed at a distance of 67 from
the protein surface. The 2040 ps component was interpreted in terms of the bound
free dynamic equilibrium proposed in a dynamic exchange model of the hydra-
tion layer.
In a series of important studies Bhattacharyya and co-workers studied the SD
studies of a covalently bound probe to protein glutaminyl-tRNA synthetase
Figure 8.4. SD study of the dye eosin in water by third-order photon echo
spectroscopy. The peak shift data of eosin in water (solid circles) and lysozyme
complex in water (open triangles) are shown. The inset shows the lysozyme data
(open triangles) with ts (solid line). Adapted with permission from J. Phys.
Chem. B, 103 (1999), 7995. Copyright (1999) American Chemical Society.
(GlnRS), both in its native and in its molten globule state. They also observed two
slow components, one major component at 40 ps and a minor at 580 ps [14].
Some earlier studies had reported that SD in protein environments was non-
exponential with a long component with a long time constant of the order of 10 ns.
Such a slow timescale component appears to be due to the ultraslow motion of the
large solute probe or due to the slow conformational uctuation of proteins.
To summarize, all the above studies clearly indicate the existence of multiple
timescales in the hydration-layer dynamics. While a large fraction of hydration-
layer water remains almost as fast as its bulk counterpart, a sizable fraction is slow. It
is conceivable that the slow water molecules reside near the hydrophilic residues
that provide stability to the enzymes, while the fast water molecules participate in
the biological activities. For example, in adenylate kinase catalysis, one nds that
water molecules play an important functional role, which has been discussed earlier
in Chapter 7, section 7.2.
8.6 Conclusion
It is rather fascinating to note that the dynamic properties of the protein hydration
layer have been studied by so many different techniques. Initially, there was
controversy about the accuracy of the different techniques employed. The situation
became cleared when proper care was taken to isolate and interpret the results. For
example, DR and SD are mostly sensitive to the rotational motion of the water
molecules and the protein side-chain motions, while NOE is sensitive to the relative
translational motion between the protein and the water molecules. Naturally they
Figure 8.5. Solvation time correlation function for a tryptophan probe in the surface
of the protein SC. The inset shows the same for dansyl-bonded SCwhere the probe
is 67 away from the surface. Adapted with permission from J. Phys. Chem. B,
106 (2002), 12376. Copyright (2002) American Chemical Society.
8.6 Conclusion 131
provided different timescales. Another reason for the apparent discrepancy was that
different groups studied different proteins. Hydration-layer dynamics is expected to
be different for different proteins.
The basic understanding that has emerged from a large number of studies can be
summarized as follows. First, unique characteristics of water manifest themselves
by adapting to the heterogeneous environment at the protein surface to serve a
multitude of purposes and to give rise to non-exponential dynamics. Some water
molecules remain relatively immobile as they serve to stabilize certain protein
conformations while others remain mobile (or free) to help in other functions. In
the next chapter, we discuss microscopic studies that have enhanced our under-
standing of this important problem.
APPENDI X 8. A ORI ENTATI ON OF WATER MOLECULES
I N THE HYDRATI ON LAYER
Here we discuss in more quantitative detail the role of surface topography in
controlling the arrangement of water molecules near hydrophobic and hydrophilic
surfaces. The basic idea is fairly simple. Near a hydrophobic surface, water mole-
cules point their hydrogen-bond-forming sites away from the surface while the
opposite is true near a hydrophilic surface as water molecules form HBs with the
charged and polar groups on the protein surface. One can devise a simple geometric
scheme to capture these different orientations.
The orientation of each proximal water molecule with respect to the local surface
normal can be characterized by studying the orientation of its hydrogen-bond
vectors. Each water molecule has four such vectors, shows by arrows in
Figure 8.A.1. We next dene as the angle between each of the HB vectors (see
Figure 8.A.1) and the outward radial direction pointing from the carbon nucleus
associated with the surface towards the water oxygen atom.
One generally nds two types of structure near the hydrophobic region of the
protein. Schematic descriptions of these two types of structures are shown in
Figure 8.A.1. The rst is the clathrate structure. This corresponds to the geometry
where three of the four HB vectors of the tagged water molecule are oriented nearly
tangentially to its proximal surface atom, with one vector pointing away. The other is
the inverted hydration, in which water molecules proximal to the surface mirror the
structure of the clathrate-like one, with one HBvector pointing directly into the surface.
Classic clathrate-like hydration was evidenced around the Val8a residue, which
belonged to the convex region. The classic clathrate-like structure of the water
molecule was found around a hydrophobic surface.
In contrast, the inverted hydration structure was found for the melittin interface;
water molecules proximal to the residues belonging to the at central region showed
orientational uctuations among a strongly clathrate-like distribution, a weak
clathrate-like form, an inverted structure, and a mixed behavior. It was further
found that the prevalence of any one structural type typically persists for approxi-
mately 1020 ps.
The hydration structure of the central hydrophobic region could not be character-
ized by either a classic, small-molecule, clathrate-like structure or an orientationally
inverted structure characteristic of truly extended surfaces, but rather by a uctuat-
ing distribution over both.
To characterize the degree to which the structural uctuations derive contribu-
tions from the water molecules, we can consider a quantity f
in
(the ratio of the
number of proximal water molecules with any one of its HB vectors pointing
radially inward towards the solute atom, divided by the total number of proximal
water molecules of this solute atom) in each conguration. Simulations indicate a
high probability of clathrate-like structures (f
in
= 0) dominating near the convex
surface patch of Val8a, and that the probability of encountering inverted structures
(f
in
= 1) increases markedly near the relatively at region.
The proximity of polar or charged groups to relatively at or concave hydrophobic
regions can have large effects on the structure and energetics of hydration.
Figure 8.A.1. Orientation of a proximal water molecule relative to the solute
surface normal. Two molecules, each with four equivalent HB vectors, are
shown schematically with clathrate-like and inverted orientations, respectively.
Of the clathrate-like case, three out of four of the angles are tetrahedral (
t
) and
the remaining angle is 0. Such orientation leads to probabilistic distribution of
cos maximizing at 0.336 and 1. In contrast, inverted orientation would lead to a
cos distribution mirroring that of the former and maximizing at 1 and 0.336.
Adapted with permission from Biophys. J., 76 (1999), 17341743. Copyright
(1999) Elsevier.
8.6 Conclusion 133
References
1. R. Pethig, Protein-water interactions determined by dielectric methods. Annu. Rev.
Phys. Chem., 43 (1992), 177205; E. H. Grant, Nature, 196 (1962), 1194; N. Nandi,
K. Bhattacharyya, and B. Bagchi, Dielectric relaxation and solvation dynamics of water
in complex chemical and biological systems. Chem. Rev., 100 (2000), 2013; B. Bagchi,
Water dynamics in the hydration layer around proteins and micelles. Chem. Rev., 105
(2005), 3197.
2. A. Mukherjee and B. Bagchi, Rotational friction on globular proteins combining
dielectric and hydrodynamic effects. Chem. Phys. Lett., 404 (2005) 409413.
3. S. Ebbinghaus, S. J. Kim, M. Heyden, et al., An extended dynamical solvation shell
around proteins. Proc. Natl. Acad. Sci. USA, 104 (2007), 20749.
4. Y. -K. Cheng and P. J. Rossky, Surface topography dependence of biomolecular hydro-
phobic hydration. Nature, 392 (1998), 696699.
5. B. Jana, S. Pal, and B. Bagchi, Hydrogen bond breaking mechanism and water
reorientational dynamics in the hydration layer of lysozyme. J. Phys. Chem. B, 112
(2008), 9112.
6. E. Dachwitz, F. Parak, and M. Stockhausen, On the dielectric relaxation of aqueous
myoglobin solutions. Ber. Bunsenges. Phys. Chem., 93 (1989), 1454; S. Boresch,
P. Hochtl, and O. Steinhauser, Studying the dielectric properties of a protein solution
by computer simulation. J. Phys. Chem. B, 104 (2000), 87438752.
7. B. Halle, Protein hydration dynamics in solution: a critical survey. Phil. Trans. R. Soc.
Lond. Ser. B, 359 (2004), 1207; B. Halle and M. Davidovic, Biomolecular hydration:
From water dynamics to hydrodynamics. Proc. Natl. Acad. Sci. USA, 100 (2003),
1213512140.
8. D. Russo, R. K. Murarka, G. Hura, E. Verschell, J. R. D. Copley, and T. Head-Gordon,
Evidence for anomalous hydration dynamics near a model hydrophobic peptide.
J. Phys. Chem. B, 108 (2004), 1988519893.
9. D. E. Khoshtariya, E. Hansen, R. Leecharoen, and G. C. Walker, Probing protein hydration
by the difference OH (OD) vibrational spectroscopy: Interfacial percolation network
involving highly polarizable water-water hydrogen bonds. J. Mol. Liq., 105 (2003), 1336.
10. X. J. Jordanides, M. J. Lang, X. Song, and G. R. Fleming, Solvation dynamics in protein
environments studied by photon echo spectroscopy. J. Phys. Chem. B, 103 (1999),
79958005.
11. (a) N. Nandi and B. Bagchi, Dielectric relaxation of biological water. J. Phys. Chem. B,
101 (1997), 1095410961. (b) S. K. Pal, J. Peon, B. Bagchi, and A. H. Zewail,
Biological water: femtosecond dynamics of macromolecular hydration. J. Phys.
Chem. B, 106 (2002), 1237612395.
12. X. Song, D. Chandler, and R. A. Marcus, Gaussian eld model of dielectric solvation
dynamics. J. Phys. Chem., 100 (1996), 1195411959.
13. (a) S. K. Pal, J. Peon, and A. H. Zewail, Biological water at the protein surface:
dynamical solvation probed directly with femtosecond resolution. Proc. Natl. Acad.
Sci. USA, 99 (2002), 17631768. (b) J. Peon, S. K. Pal, and A. H. Zewail, Hydration at
the surface of the protein monellin: dynamics with femtosecond resolution. Proc. Natl.
Acad. Sci. USA, 99 (2002), 1096410969.
14. (a) D. Mandal, S. Sen, D. Sukul, et al., Solvation dynamics of a probe covalently bound
to a protein and in an AOT Microemulsion: 4-(N-Bromoacetylamino)-Phthalimide.
J. Phys. Chem. B, 106 (2002), 1074110747. (b) S. Guha, K. Sahu, D. Roy,
S. K. Mondal, S. Roy, and K. Bhattacharyya, Slow solvation dynamics at the active
site of an enzyme: implications for catalysis. Biochemistry, 44 (2005), 89408947.
9
Understanding the protein hydration
layer: lessons from computer simulations
Several questions have been asked repeatedly about the protein hydration
layer: (i) what is the width of the layer? (ii) Howfast is the water molecular
motion? And (iii) how different are the water structure and dynamics near
the hydrophobic and hydrophilic patches of a protein? Although initial
understanding of the protein hydration layer was fraught with difculties
and misunderstanding, considerable advances have been made in recent
years through the use of computer simulations. It is fair to state that
whatever we now know about the structure and dynamics of the protein
hydration layer has been guided to a great extent by simulations. This is
because the protein hydration layer is just a few monolayers thick, which
makes experimental determination very difcult, and studies provide indir-
ect information about the hydration layer. One nds that the unique proper-
ties of water molecules again manifest themselves in varieties of ways and
allow the water molecules to respond differently to different regions of the
protein surface. Here we discuss the molecular reasons for this diversity in
the properties of water in the protein hydration layer.
9.1 Introduction
The protein hydration layer is not only quite thin, compared to the size of the
protein, but it is also heterogeneous on a scale that is comparable to, may be slightly
larger than, the size of an individual water molecule. Because of this molecular-
length-scale heterogeneity, neither theory nor experiments could provide a clear and
consistent picture of the structure and dynamics of interfacial water. In particular,
different experimental techniques applied to different protein hydration layers can
provide apparently conicting results. In such a situation, computer simulations
have proved to be a valuable tool. Because the dynamic behavior manifested is rich
and diverse, a large number of computer simulation studies have been carried out. In
the following, we discuss the results obtained by molecular dynamics (MD) simula-
tions on protein hydration layers.
135
9.2 Molecular motion in the hydration layer
Early simulations were carried out to mimic powder samples of hydrated proteins,
that is, not in solution. The simulation study of the hydration of protein ribonuclease
A revealed that at room temperature and at high hydration, signicant translational
and rotational motions continue to occur in the layer (the hydration layer is not as
rigid as ice). Both translational and rotational diffusion coefcients of water mole-
cules in the layer are found to be correlated with the residence time because the latter
is a direct measure of the rigidity of the layer. The residence time of water molecules
in the hydration layer of myoglobin in aqueous solution was found to have a
distribution between somewhat less than 30 ps and more than 80 ps [1]. The
water molecules with much longer residence times were those that were either
buried inside protein cavities or in the clefts or had multiple interactions with the
protein and had higher (than average) binding energies. Water molecules with long
residence time exhibit slow orientational relaxations [1]. The trajectory analysis of
individual water molecules clearly showed two entirely different behaviors: one for
the bound state and the other for the freely moving state. Rapid exchange between
the two states was also observed, suggesting the existence of a dynamic equilibrium
between the two states.
An important dynamic measure of a hydration layer is provided by its survival
correlation time, which is dened in the following way. First, we make a list of all
the molecules present in the layer at the initial time. We can now proceed in two
ways. First, we can assign decay to the hydration layer whenever any single
water molecule leaves or enters the layer. We then average over all the molecules
in the layer. Let us denote this function by S
s
(t). This can be dened in the
following way
S
s
t ( ) = 5
1
N
L
0 ( )
N
L
0 ( )
i=1
h
i
t ( ) > (9:1)
where h
i
(t) is a function that is unity as long as the ith molecule is within the layer
and zero as soon as it leaves the layer. N
L
(0) is the total number of molecules in the
layer at time t = 0. This function goes to zero when all the molecules initially present
in the layer leave the layer. Some of the molecules leave and return after a while.
These molecules interfere with the observation of the genuinely slow or bound
molecules which take a long time to leave. In order to study the slow ones
specically one can dene another correlation function, C
L
(t), such that once a
molecule leaves the layer it is no longer counted in the sum:
C
L
t ( ) = 5
1
N
L
0 ( )
N
L
0 ( )
i=1
H
i
t ( ) > (9:2)
136 Understanding the protein hydration layer: lessons from computer simulations
where H
i
(t) is a function which is zero as soon as the ith molecule leaves the layer.
Here care must be taken to exclude the crossing and re-crossing due to short-time
intermolecular vibrational motions as they arise from different dynamic behavior.
The above denitions are particularly suitable for investigations in computer
simulations. They can be applied also to a ctitious layer in the bulk liquid, except
that in the latter case decay can happen by molecules crossing across the region
that is, by penetrating the sphere not allowed in the case of protein. In the case of the
protein hydration layer, these survival correlation functions decay slowly for the
hydration layer.
The same method can be used to study density uctuations by considering all the
water molecules in the layer. Let us denote this function by S
NN
(t). This is dened by
S
NN
t ( ) = 5
1
N
L
0 ( )
N
L
0 ( )
i=1
h
i
t ( )
1
N
L
t ( )
N
L
t ( )
j=1
h
j
t ( ) > (9:3)
This function measures the density uctuation in the layer and related to the rigidity
of the layer.
The average (over the water molecules in the layer) orientational time correlation
function also shows markedly non-exponential decay with a 37-times slower time
constant than that of the bulk for lysozyme [2]. Simulation studies show that the
slow molecules belong to those water molecules that had longer residence times
near the lysozyme. Thus, residence time correlates with the rotational correlation
time.
In addition, the average translational motion was also found to be subdiffusive.
Because the water molecules studied were initially constrained to be in the hydration
shell surrounding the protein (such as lysozyme), one of course expects that the
water molecules would exhibit slower initial displacements than those in the bulk. A
simple interpretation of the subdiffusive behavior has been presented in terms of a
theoretical model that employs a double-well potential near the surface [3]. The
subdiffusive motion originates from the recrossing of the tagged molecules into the
bound state. Thus, the slowing down in diffusion is not just due to the constraint of
being initially in the hydration layer but is also due to interconversion between the
two states.
The mechanism and the rate of hydrogen-bond breaking in the hydration layer
surrounding an aqueous protein have been studied by using a time correlation
function technique to understand these aspects in the hydration layer of lysozyme.
Water molecules in the layer are found to exhibit three distinct bond-breaking
mechanisms compared to bulk reorientation [4]. The reorientation processes are
associated with the hydrogen-bond breaking/switching events (HBSE). Three impor-
tant characteristics that are common for almost all of the reorientation processes
in bulk water are (Figure 9.1(a)): (i) a jump (60) in the angular direction of the
rotating OH bond (donor water molecule, O*H
2
) [5]; (ii) the new acceptor molecule
(O
n
H
2
) comes from the second coordination shell (D
O*O
n
> 4.1 ) to the rst
coordination shell (D
O*O
n
< 3.5 ) of the donor water (O*H
2
); and (iii) the old
acceptor molecule (O
o
H
2
) goes out from the rst coordination shell to the second
coordination shell of the donor water diffusively. The events accompanying the
reorientation process in the hydration layer are as follows:
(i) Both of the acceptor molecules are initially within the rst coordination shell of
the donor before reorientation and remain there even after the process. This
mechanism is the most prevalent (80% of all the HBSEs) in the hydration
layer (Figure 9.1(b)). This mechanism is clearly a consequence of the two-
dimensional network of water around the protein surface. The motion of the
incoming and the outgoing acceptor molecules is not diffusive prior to or even
after the HB-breaking event. We nd a sharp jump in the angular direction of
the rotating OH bond across the HBSE of this mechanism.
The constrained motion of the old acceptor, which is not allowed to go out
from the rst coordination shell of the donor, after the HB-breaking incident,
bears the signature of the presence of a connected network of water at the
lysozyme surface. If the old acceptor were to go out from the rst coordination
shell to the second coordination shell of the donor, connectivity would need to
be rearranged, which is both energetically and entropically demanding.
Similarly, the network prevents the new acceptor molecule entering the second
coordination shell.
(ii) In the second type of mechanism (present in 10% of all HBSEs), the new
acceptor molecule comes from the second coordination shell to the rst
coordination shell of the donor (Figure 9.1(c)). However, the old acceptor
molecule remains in the hydration shell before and even after the HBSE and
does not leave from the rst coordination shell of the donor after the breaking
event. We again nd a large angular jump for the donor water molecules across
the HBSE [4].
(iii) In the third kind of mechanism (present in 10% of all HBSEs), both of the
acceptor molecules are initially in the rst coordination shell of the donor,
but nally the old acceptor moves out of the rst coordination shell after the
bond-breaking (Figure 9.1(d)). The left-hand part of this mechanism pro-
vides the signature of the lower mobility of water molecules inside the
hydration layer. Here also we nd a sharp change in the angular direction
of the rotating OH bond across the HBSE. The last two mechanisms also
provide an indication of the dynamic exchange between the hydration layer
and the bulk.
b
Figure 9.1. (a) The evolution of the characteristic parameters across the HBSE in
bulk water. The lower panel displays the evolution of the distances between the
acceptor (O
o
H
2
is the old acceptor molecule, and O
n
H
2
is the newacceptor molecule)
and the donor (O*H
2
) water molecules across the switch. The upper panel shows the
evolution of the angular direction of the rotating O*H* bond across the switch. Note
the diffusive nature of the water molecules and the large angle jump here. (b) The
evolution of the characteristic parameters across the HBSE of the rst mechanism
in the hydration layer. For a detailed description of the upper and lower panels, see
the caption of (a). Note the constrained translational motion on both sides of
the switch and the large angle jump. (c) The evolution of the characteristic
parameters across the HBSE of the second mechanism in the hydration layer. For a
detailed description of the upper and lower panels, see the caption of (a). Note the
constrained translational motion in the right side of the switch and the large angle
jump. (d) The evolution of the characteristic parameters across the HBSE of the third
mechanism in the hydration layer. For a detailed description of the upper and lower
panels, see the caption of (a). Note the constrained translational motion in the left side
of the switch and the large angle jump. Adapted with permission fromJ. Phys. Chem.
B, 112 (2008), 91129117. Copyright (2008) American Chemical Society.
9.3 Hydrogen-bond lifetime dynamics
Related to the above discussion of the hydrogen-bond-breaking mechanism and
also the survival time of a layer, the study of HB lifetime dynamics has proven to
be a useful tool to understand water dynamics arising from the extended HB
network.
In the protein hydration layer water can form two types of HBs: one with water
itself (waterwater HB, and other one with protein atoms (proteinwater HB).
In the last subsection, the mechanism of the waterwater hydrogen-bond break-
ing in the hydration layer was described. One can now dene a correlation function,
C
L
t ( ) = 1 h
a
0 ( )h
b
t ( )) ( ) (9:4)
Figure 9.1. (cont.)
c
where h
a
is 1 when OH(donor) is hydrogen-bonded to an acceptor and 0 otherwise,
and h
b
is 1 when the same OH is hydrogen-bonded to another acceptor after bond
breaking and 0 otherwise. By discarding the contribution from back-reaction (re-
formation), the rate constant extracted from this correlation function gives the
forward rate of hydrogen-bond breaking.
Decay of the HB lifetime correlation function in the bulk water is nearly single
exponential with the characteristic time constant
0
= 1.8 ps. The characteristic time
constant of the correlation function obtained in the hydration layer is
0
= 2.6 ps [4]
(Figure 9.2). Thus, the lifetime of a waterwater HB in the hydration layer is on
average higher than in the bulk. However, this stability of the HBs in the hydration
layer is not homogeneous. While the lifetime of an HB around a hydrophilic surface
is 24-times longer than in the bulk, the increase in only 20% near a hydrophobic
surface. In fact, these average estimates hide a lot of important information which
can be gained by looking at the distribution, as discussed earlier.
Let us now concentrate on the proteinwater HBs. On average, proteinwater
HBs are more stable than bulk water. As we have seen earlier, the dynamics is
heterogeneous in the hydration layer, so one can ask a simple question: how
different is the HB involving the backbone atom from those involving the side-
chain atom? How different are they compared to the HB in bulk water. To answer
these questions, one can dene a similar correlation function where n
a
is 1 if a
particular pair (protein atom and water atom) is hydrogen-bonded and 0 otherwise,
Figure 9.2. Decay of the waterwater HB correlation function in a semilog plot in
the bulk and the hydration layer. The slope of this correlation function provides
information about the HB switching rate. Note the slow decay of the correlation
function in the hydration layer as compared to bulk water, indicating a slower
switching rate in the hydration layer. Adapted with permission from J. Phys.
and n
b
is 1 when that particular pair is broken and 0 otherwise. The decay of this
correlation function for bulk water is mono-exponential with a characteristic time
constant
0
~ 1.8 ps. The decays of the correlation function for the HBs involving
backbone atoms and those involving side-chain atoms are bi-exponential with
characteristic average times (<
0
>) 1.58 ps and 5.37 ps, respectively (Figure 9.3).
This result is really interesting as the HBs involving backbone atoms are less stable
than the bulk water. As we have discussed earlier in this chapter, the process of
reorientation is closely associated with HB breaking. Thus, these water molecules
hydrogen-bonded with backbone atoms rotate as fast as bulk water. This indicates
the presence of fast water in the hydration layer. The formation and breaking of the
protein-water HB might play an important role in determining the functionality of
the protein.
9.4 Computer simulation of solvation dynamics
As discussed in Chapter 3, SD provides information on molecular motions (primar-
ily rotation) by optically studying the energy uctuations in a solute probe. In the
experimental SD studies of the hydration layer of proteins, we need to either place
an external probe in the layer, or use a natural probe such as tryptophan, which is a
natural amino acid residue. An additional constraint is that the probe must be at least
partly exposed to the solvent. However, in computer simulation studies we have the
0 2 4 6 8 10
time (ps)
-2.5
-2
-1.5
-1
-0.5
0
l
o
g
1
0
(
1
-
<
n
a
(
0
)
n
b
(
t
)
>
)
Bulk water
Backbone oxygen
Sidechain oxygen
Figure 9.3. Decay of the HB correlation function in a semi-log plot for backbone
water, sidechain water, and in bulk water. The slope of this correlation function
provides information about the HB switching rate. Note the slow decay of the
correlation function for sidechain water. Adapted with permission from J. Chem.
Sci., 124 (2012), 317325. Copyright (2012) Springer Link.
advantage that we can use solvent-exposed polar amino acid residues to monitor
energy uctuation. We now discuss some of the recent interesting developments in
the eld.
Simulation studies of the SD of the polar amino acid residues in each of the three
helical segments of the protein HP36 reveal the presence of a small-amplitude slow
component in the SD, which is an order of magnitude slower than that of the bulk.
The correlation between the exposure of polar probe residues and the SDof different
secondary structures of a protein molecule was also established. The more exposed
the probe, the faster the SD is [6].
Constrained MD simulations with either frozen protein or frozen water
revealed the molecular mechanism of slow hydration processes and elucidated
the role of protein uctuations. Slow water dynamics in MD simulations requires
protein exibility. However, there still remains the controversy on the origin of
the slow component: whether the slow component results from the water or the
protein contribution. The initial dynamics in a few picoseconds represents fast
local motions such as reorientations and translations of hydrating water mole-
cules, followed by slow relaxation involving strongly coupled waterprotein
motions [7].
The role of protein side-chain motion in the slowdynamics of water around the
protein surface is investigated by calculating the HB lifetime correlation func-
tion, S(t), for two different conditions: (i) when the side-chain protein motion is
not constrained and (ii) when it is constrained. S(t) showed a long-time tail in
its natural condition; the function initially decays slowly in its constrained
condition compared to its natural condition and then decays to zero over a
long time.
9.5 Dielectric relaxation
Dielectric relaxation results are proven to be the most denitive to infer the
distinctly different dynamic behavior of the hydration layer compared to bulk
water. However, it is also important to understand the contributions that give rise
to such an anomalous spectrum in the protein hydration layer, and in this context
MD simulation has proven to be useful. The calculated frequency-dependent
dielectric properties of an ubiquitin solution showed a signicant dielectric incre-
ment for the static dielectric constant at low frequencies but a decrement at high
frequencies [8]. When the overall dielectric response was decomposed into protein
protein, waterwater, and waterprotein cross-terms, the most important contribu-
tion was found to arise from the self-term of water. The simulations beautifully
captured the bimodal shape of the dielectric response function, as often observed in
experiments.
9.6 Explanation of anomalous dynamics in the hydration layer
There are several explanations of water anomalies. In the water prole we have seen
there are some slow and some fast water molecules. Even the slow water molecules
are also transient in nature. Thus, so far now we have been convinced that the water
dynamics in the hydration layer is quite anomalous and slow compared to bulk
water. To explain the origin of the anomalies at the solutesolvent interface, two
different proteinwater systems have been considered [9]. The rst consists of a
frozen protein surrounded by water molecules thermalized at 300 K. In the second
system, the protein matrix is still kept frozen and, additionally, the electrostatic
interactions between the protein and water are eliminated. Results from these
systems are compared to a solution where proteinwater interactions are included
in full, the dynamics is unconstrained, and the entire system is thermalized at
300 K. The obtained results allow the contributions from geometrical and energetic
disorder and from protein motion to be considered separately. Around a static
protein both types of disorder (charge distribution and surface topology) act on
the water translational diffusion, contributing both to the average retardation and
also to the glassy-like anomalous translational diffusion of water at the protein
surface. The protein surface offers electrostatic HB pinning sites to water molecules
and the hydration water forms a percolating HB cluster that surrounds the whole
protein and hinders the water dynamics. Again when the electrostatic interactions
between the protein and the solvent are eliminated, water only forms relatively small
H-bonded clusters and supercial water diffusion is enhanced.
The rotational dynamics on the protein surface is basically shaped by electrostatic
interactions alone and the HBs formed by water with the protein surface break the
quasi-isotropic nature of the dipolar rotation that is found in the bulk. Also, for the
fully thermalized protein, a ratio between the characteristic times of the rst and
the second dipoledipole correlation function, =
1
R
=
2
R
, of about 5 is at variance
with the isotropic assumption, = 3, used in NMRD estimates of the translational
residence time.
In addition, protein motion reduces the retardation of the water dynamics, because
the dimension of the water translational space is increased and at the same time the
decay of the orientational correlation is accelerated. In spite of this accelerated
dynamics, hydration water diffusion remains anomalous for a thermalized protein.
9.7 Proteinglass transition at 200 K: role of water dynamics
As discussed in Chapter 8, many proteins seem to undergo a glass transition
around 200 K [10]. Experiments and simulations also show that below this tem-
perature, the dynamic behavior of proteins changes from anharmonic to harmonic
(Figure 9.4(a)). It has been expected that below this temperature, proteins form a
glassy state. Note that for most proteins, the enzymatic activity ceases below
220 K. It has been argued that water dynamics may hold the key to understanding
this unusual behavior of proteins. Note that water itself is believed to have a glass
transition around 165 K. It has also been suggested that water also has an additional
transition at a temperature below 228 K (Figure 9.4(b)). Below this temperature,
water behaves like a strong liquid while above this temperature it behaves like a
fragile liquid [11].
The proximity of this liquidliquid transition to the proteinglass transition tem-
perature is suggestive. Clearly, at temperatures below 220 K or so, the dynamics of
water and protein are highly coupled. A recent computer simulation study has shown
that the structural relaxation of protein requires relaxation of the water HB network
and translational displacement of interfacial water molecules. It is, therefore, clear that
the dynamics of water at the interface can play an important role. This is an interesting
problem that deserves further investigation.
(a)
(b)
Figure 9.4. (a) Mean-square atomic uctuation of protein. This shows a transition
around T = 220 K, the mean-square uctuation becomes low and varies weakly
with temperature below T = 220 K. (b) Water dynamics also changes its functional
dependence at the same temperature. This is termed dynamic transition. At high
temperature the dynamics follows non-Arrhenius behavior (fragile liquid) and at
high temperature it follows Arrhenius behavior (strong liquid). Adapted with
permission from Proc. Natl. Acad. Sci. USA, 103 (2006), 9012. Copyright
(2006) Proc. Natl. Acad. Sci. USA.
9.7 Proteinglass transition at 200 K: role of water dynamics 145
9.8 Free-energy barrier for escape of water molecules
from protein hydration layer
Recent calculations of free-energy barriers that separate the interfacial water
molecules at the surface of the protein chicken villin head-piece (HP36) from
bulk support the above picture. Only a few water molecules were found to be
strongly hydrogen-bonded and could therefore be termed slow (see Figure 9.5).
The free-energy calculations reveal a strong sensitivity of the barriers to the
secondary structure. In particular, it was found that there exists a region near
the junction of the rst and the second helix that contains a cluster of water
molecules (see Figure 9.6) which are slow in motion, characterized by long
residence times (of the order of 100 ps or more) and separated by a large free-
energy barrier from the bulk water [12].
9.9 Conclusion
The enormous importance of the protein hydration layer is the reason for the
continued study and discussions of this topic over the last half century. It has also
led to controversy and confusion, to an extent that one is reminded of the story of the
Strongly bound
water cluster
Helix-2
Helix-1
Helix-3
Figure 9.5. Representative snap extracting from the simulation trajectory, showing
the location of some water molecules in the rst hydration layer that are doubly
hydrogen-bonded to the protein residues. The helices are drawn as a ribbon. The
quasi-bound water molecules are drawn using the licorice model. Adapted with
permission from J. Phys. Chem. B, 116 (2012), 29582968. Copyright (2012)
six blind mens attempt to describe the shape of an elephant. The metaphor is quite
appropriate as different experimental techniques provide information about differ-
ent aspects of the hydration shell of different proteins. If one is a bit circumspect,
then there would remain no reason for such confusion and controversy. Below we
give an example with an explanation of such a controversy which will make our
point clear.
As discussed in Chapter 8, several experimental studies have detected an ultra-
slow (ranging from a few hundred picoseconds to few tens of nanoseconds)
component in the SD of an external probe, both in proteins and in self-assembly
(discussed later). Theoretical and computer simulation studies of HB lifetime
(a)
(b)
Figure 9.6. Two-dimensional free-energy surface along with its contour map for
(a) strongly hydrogen-bonded quasi-bound water and (b) interfacial free water.
The color code of the free-energy landscape has been so chosen that the closely
spaced regions can be distinguished clearly. The presence of two minima in
(a) corresponds to two HB breaking events whereas in (b) the existence of a
single minimum indicates only one weak HB rupture. For both cases the escape
along the Z direction is evident from the contour. Adapted with permission from
Society. See plate section for color version.
9.9 Conclusion 147
dynamics showthat the usually fast HBbreaking/re-formation process in bulk water
can slowdown noticeably at the surface of proteins and micelles. This may be due to
a combination of the rigidity of the water network at the surface and the strong HBs
between the polar head groups at the surface and the water molecules. Here, two
somewhat opposing factors arise. While the strength makes the bond-breaking
process slow, the rigidity of the interfacial water often forces re-formation of the
bond in short time. However, the experimentally observed ultraslow solvation
component does not appear to be coupled to the dynamics of such slow water
because its contribution to total solvation energy relaxation is found to be rather
small. This view is further substantiated by the fact that such a slow component is
totally absent in protein hydration dynamics when a natural probe located near the
surface is used. In such cases, the slowest time observed is just about 2040 ps. The
ultraslow component has also not been observed in the computer simulation studies
of SD in micelles and reverse micelles. In the latter cases, the slowest component is
again less than 100 ps. The orientational relaxation exhibits slower relaxation, as
remarked earlier.
In the present chapter we discussed quantitative aspects of protein hydration-
layer dynamics. Because of the complexity of the problem and perhaps due to a
certain lack of concerted effort, there are still many issues remained to be settled.
References
1. W. Gu and B. P. Schoenborn, Molecular dynamics simulation of hydration in myoglobin.
Prot. Struct. Funct. Genet., 22 (1995), 20.
2. M. Marchi, F. Sterpone, and M. J. Ceccarelli, Water rotational relaxation and diffusion in
hydrated lysozyme. J. Am. Chem. Soc., 124 (2002), 6787.
3. A. Mukherjee and B. Bagchi, Origin of the sub-diffusive behavior and crossover from
sub-diffusive to super-diffusive dynamics near a biological surface. Phys. Chem.
Commun., 6 (2003), 28.
4. B. Jana, S. Pal, and B. Bagchi, Hydrogen bond breaking mechanism and water reor-
ientational dynamics in the hydration layer of lysozyme. J. Phys. Chem. B, 112 (2008),
91129117.
5. D. Lagge and J. T. Hynes, A molecular jump mechanism of water reorientation. Science,
311 (2006), 832.
6. S. Bandyopadhyay, S. Chakraborty, S. Balasubramanian, and B. Bagchi, Sensitivity of
polar solvation dynamics to the secondary structures of aqueous proteins and the role of
surface exposure of the probe. J. Am. Chem. Soc., 127 (2005), 4071.
7. T. Li, A. A. Hassanali, Y. T. Kao, D. Zhong, and S. J. Singer, Hydration dynamics
and time scales of coupled waterprotein uctuations. J. Am. Chem. Soc., 129
(2007), 3376.
8. S. Boresch, P. Hchtl, and O. Steinhauser, Studying the dielectric properties of a protein
solution by computer simulation. J. Phys. Chem. B, 104 (2000), 8743.
9. F. Pizzitutti, M. Marchi, F. Sterpone, and P. J. Rossky, How protein surfaces induce
anomalous dynamics of hydration water. J. Phys. Chem. B, 111 (2007), 7584.
10. M. Tarek and D. J. Tobias, Role of proteinwater hydrogen bond dynamics in the
protein dynamical transition. Phys. Rev. Lett., 88 (2002), 138101; M. M. Teeter,
A. Yamano, B. Stec, and U. Mohanty, On the nature of a glassy state of matter in a
hydrated protein: relation to protein function. Proc. Natl. Acad. Sci. USA, 98 (2001),
1124211247.
11. K. Ito, C. T. Moynihan, and C. A. Angell, Thermodynamic determination of fragility in
liquids and a fragile-to-strong liquid transition in water. Nature, 398 (1999), 492495;
C. A. Angell, Formation of glasses from liquids and biopolymers. Science, 267 (1995),
19241935.
12. S. Roy and B. Bagchi, Free energy barriers for escape of water molecules from protein
hydration layer. J. Phys. Chem. B, 116 (2012), 29582968.
References 149
10
Water in and around DNA and RNA
Within our cells, DNA molecules exist in the double-helix form. Although
RNA is a single strand, it is highly organized. Water molecules in the well-
dened grooves of DNA exhibit properties that are different fromthe bulk.
The dynamics of the hydration layer around DNA is extremely complex,
due to the diverse arrangement of base sequences and because of the
presence of major and minor grooves. The negatively charged phosphate
back-bone and positively charged sodium ions also add to the complexity
of the problem. In experiments one nds unusual power law decay in the
dynamic response of the liquid and in simulations one nds correlated
motion of water molecules between the grooves. Although the dynamics
of water molecules around RNA has been relatively less studied, it also
exhibits rich dynamics. In this chapter, we discuss both systems, with an
emphasis on DNA. Water molecules unique ability to form various local
quasi-stable structures seems to play an important role in the grooves of
DNA, especially in the formation of the much-discussed spine of hydra-
tion around certain minor grooves. While a large number of experiments
have been devoted to this problem, signicant progress has occurred only
after long computer simulations were performed to disentangle the var-
ious contributions that control the hydration dynamics of a DNA mole-
cule. A clear picture of hydration has just begun to emerge.
10.1 Introduction: the unique role of water
in stabilizing DNA and RNA
In our cells, DNAis one of the most important molecules as it contains all our genetic
information. The central dogma of biology starts with DNA[1]. Certain characteristic
features of DNA are worth remembering. DNA is a rigid molecule on a small length
scale (35 or 10 base pairs) but quite a exible molecule on the larger length scale
(500 or 150 base pairs). These numbers depend (not too strongly) on the nature of
the solvent, such as pH, ion concentration, etc. The remarkable properties of DNA
depend on this rigidity at small length scales and the exibility at large length scales.
151
DNA hydration is crucially important for its conformation and utility, as
already noted by Watson and Crick [2]. The strength of the aqueous interaction
here is far greater than those in protein as DNAis a highly charged species due to the
presence of phosphate ions. To understand the hydration of DNA, one needs to
understand the hydration of phosphate ions, sugar, and the nucleotide bases (A, T,
G, and C), separately. Without water, the double-helix structure would not
have been stable, due to the repulsion between the negatively charged phosphates
of the different strands and, to a lesser extent, of the same strand. Water screens
these charges to make such an entangled structure stable. However, water
molecules surrounding the DNA molecules are also affected by the charges present
in the DNA.
The double-helix structure can take up a number of conformations (A-DNA,
B-DNA, C-DNA, Z-DNA, etc.) depending upon the level of hydration and ionic
strength. In nature, B-DNA is predominantly a double-helical structure which has a
wide major groove and a narrow minor groove. Both the grooves are fully hydrated.
Moreover, for several biological processes small molecules or large protein mole-
cules need to bind to the DNA segment at the right position to make the processes
errorless. This process is called recognition. This important process is largely
mediated by the water molecules around the DNA segment. Thus to understand
the structure and function related to the DNAmolecule, it is important to understand
the structure and dynamics of the water around it.
On the other hand, ribonucleic acid (RNA) is a linear polymer made up of four
bases, like DNA, except the base thymine of DNAis replaced by urasil and the sugar
oxyribose is replaced by ribose. It has the phosphate-sugar back-bone as well.
Unlike DNA, RNA is a single-strand polymer. This gives RNA markedly different
properties. Since RNA is a single-strand polymer with polar groups exposed to
water, the latter solvates it more extensively than DNA. RNA also exists in a
collapsed state to shield the hydrophobic core of the bases from water. However,
the resulting conguration of RNA in solution is different from that of DNA,
primarily because of the two free ends.
The dynamics of the hydration layer around a DNA molecule is extremely
complicated, with effects from the connement due to major and minor grooves,
electric eld effects from the negatively charged phosphate backbone and counter-
ions, and the ability of the nucleic acid groups (G, C, A, T) to formrather strong HBs
with water molecules, in addition to forming HBs among themselves.
10.2 Hydration of different constituents
Different base pairs contain different atoms that are of different polarity. The
environment faced by water molecules varies from groove to groove and from
152 Water in and around DNA and RNA
major groove to minor groove. As a consequence, water molecules surrounding the
DNA interface experience a highly heterogeneous polar environment. Therefore, a
probe placed near DNA reports non-exponential dynamics which is often hard to
interpret.
As in the case of protein hydration, water molecules again play a dual role. First
and foremost, they provide stability to the observed nucleic acid conformation.
Second, they stabilize, often for a long duration, complexes with diverse proteins
and intercalators. In the latter cases, localized ordered water molecules facilitate
formation of specic nucleic acid structures around external molecules/groups [3].
The detailed structural morphology of the DNA hydration structure is illustrated in
Appendix 10.A.
As already emphasized, study of the hydration dynamics of DNA is complex. We
need to understand both the group-specic and the base pair sequence-specic
dynamics of water.
10.3 Groove structure and water dynamics
In nature, the B-DNA double helix occurs predominantly over other helical struc-
tures. This B-DNA is a right-handed double helix with wide and shallow major
grooves and narrow and deep minor grooves. A schematic representation of the
groove structure is shown in Figure 10.1. On average, major grooves have a width of
1012 and that of minor groove is only 56 . Thus, B-DNA provides quite
different environments for water molecules in the two groove regions. Naturally,
water molecules exhibit heterogeneous dynamics that depend on the location of a
given water molecule along the double helix.
10.4 Translational and rotational dynamics of water molecules
in the grooves
The altered nature of water in the DNA grooves is reected in their dynamics, as
measured by several experiments and computer simulations. Translational
dynamics is usually measured in terms of mean-square displacement (MSD) of
water molecules in the respective groove regions. MSD can be measured by
neutron-scattering experiments. They can also be obtained in a straightforward
fashion from computer simulations. Such a simulation gave the following values
of the diffusion coefcients: 2.2 10
5
cm
2
/s for minor groove water, 3.4 10
5
cm
2
/s for major groove water and 5.2 10
5
cm
2
/s for bulk water molecules [4].
These values exemplify the large difference in the dynamics in different regions.
Note that these values are for a chosen interaction force-eld (here the GROMACS
10.4 Translational and rotational dynamics of water molecules in the grooves 153
force-eld with the TIP5P model of water interaction). So, the precise value of the
diffusion coefcient need not be taken too seriously but the trend remains the
same across different interaction potentials.
Another measure of the differing dynamics of water molecules in different
regions is provided by rotational dynamics. Rotational dynamics can be measured
in several ways, for example by following the dipoledipole orientational time
correlation function of groove water molecules. Again, computer simulations pro-
vide a straightforward approach to such a function. The correlation function is non-
exponential, with a slow component in the long time. The average time constants
obtained from the t to the function are found to be 28.5 ps, 6.2 ps, and 2.1 ps for
minor groove, major groove, and bulk water, respectively [4].
The rotational motion of a small but strongly interacting molecule such as water is
a good measure of the mobility (or uidity) of a small region because the relaxation
occurs in a relatively small time window(compared to translational diffusion, which
Figure 10.1. DNA double helix showing the major and minor grooves.
is a much slower process). Therefore, we can regard the above quoted time constants
as a fairly accurate quantitative measure of the relative mobility of water molecules
in the three regions of a DNA solution.
It is also important to note that the timescale for the slow component is much
longer for minor groove water than for major groove. The slow component is
found to be absent for bulk water. Both the translational and rotational dynamics
of water molecules are found to be signicantly slower in the minor groove region
compared to those in the major groove. The dynamics in the major groove region
is also found be signicantly slower than that in the bulk water.
To recapitulate, the study of SD involves placing a suitable uorescent probe inside
or outside the DNA. It can be placed inside by chemical modication of the DNAby
which a base or base pair is replaced by the probe. A probe can also be placed
outside when a dye intercalates into the DNA so that a part of it goes inside the
duplex. When optically excited, such a probe gives emissions that bear the signature
of the dynamics of the surrounding molecules, including water, base pairs, and
counterions. TDFSS has been a popular tool to investigate water dynamics in
various constrained and conned environments.
In this section we discuss such SD studies in the aqueous DNA system. Direct
probes of the dynamics of hydration with different time resolutions have been
carried out by placing a probe at the surface of small pieces of DNA, such as
dodecamer (or hexadecamer) duplex [5,6]. Unfortunately, however, in the case of
such small DNA molecules, the native structure can undergo signicant distortion
due to the presence of the solute. This can complicate the interpretation of the
results, as one would rst need to understand the hydration dynamics around an
undistorted DNA molecule. We shall proceed to discuss the results of SD with this
caveat in mind.
Aprobe that is known to intercalate into the minor groove of the DNAduplex and
whose native structure remains relatively unperturbed subsequent to minor groove
binding is the drug bisbenzimide (Hoechst 33258). By following the temporal
evolution of uorescence, two well-separated hydration times, 1.4 and 19 ps,
were observed. In bulk water the same drug is hydrated with time constants of 0.2
and 1.2 ps. In a rare study, the hydration dynamics of calf thymus DNA was
investigated and it was found that this DNA exhibits hydration dynamics with
similar timescales.
Relatively small timescales of hydration at the surface of the groove seem to
suggest that solvation is a dynamic process with two general types of trajectories,
the slowest of them (~20 ps) arising from water at the surface of the DNA.
Several other experimental studies have revealed, surprisingly, the existence of
much slower timescale of the order of a few 100 ps to 10s of ns. Here time-resolved
Stokes shifts in a dye-containing oligonucleotide have been observed over the entire
time range from 40 fs to 40 ns [6]. The dynamics could be t to a power law with a
small exponent of 0.15 (Figure 10.2). The origin of such a slow component in SD
has not been understood yet, but it may be due to the correlated motion of the ions
and water present in the solution. Simulation studies fail to nd such slow dynamics
in either water molecules or ions. It is also not clear whether structural relaxation of
a small DNA polymer can give rise to such slow power law decay.
The origin of the power law decay is still not clearly understood. Possible origins
include (i) a contribution from the ion atmosphere due to the counter ions and (ii)
correlated motion of the water molecules along the grooves. The rst contribution
could be related to the well-known DebyeFalkenhagen effect which arises from
correlated ion motion. The second contribution can arise from correlated motion of
water molecules between grooves.
10.6 Entropy of groove water and dynamics
One useful discriminator of structure and dynamics in liquid is obtained through
entropy. However, experimental or theoretical estimation of the entropy of a liquid
conned to a local region is quite hard. This has hampered our understanding of the
order/disorder transition in the local region at the mesoscopic length scale. One such
rare study concentrated on the estimation of the entropy of water molecules in the
groove region and correlated with the observed dynamics. Note that calculation of
Figure 10.2. Solvation time correlation function of aqueous DNA solution. Note
power law decay over the six decades of the timescale (40 ps to 40 ns). Adapted
with permission from J. Am. Chem. Soc., 127 (2005), 72707271. Copyright
the entropy of a liquid is itself quite difcult (as discussed later in Chapter 19) and
here connement adds additional complications. A two-phase thermodynamics
(2PT) method where liquid is partitioned between the harmonic solids and hard-
sphere gas has proven to be quite successful in producing an estimate of liquid
entropy. The same method is also used to calculate the entropy of water in the
groove region. The average values of the entropy of water at 300 K in both of the
grooves of DNA (the TS value in the major groove is 6.71 kcal/mol and that in
the minor groove is 6.41 kcal/mol) are found to be signicantly lower than that in
bulk water (the TS value is 7.27 kcal/mol). This estimation suggests that water
molecules are most constrained (lowest entropy) in the minor groove region,
leading to the slowest dynamics. Also the lower entropy for the major groove
region compared to bulk water can explain the observed slower dynamics of major
groove water compared to in the bulk [7].
Now, the entropic contribution to the free-energy change (TS) of transferring
a minor groove water molecule to the bulk is found to be equal to 0.86 kcal/mol
and that of transferring a major groove water to the bulk is 0.56 kcal/mol at 300 K.
These values can be compared with the value 1.44 kcal/mol for the melting of ice
at 273 K. This result suggests an ice-like structure in the minor groove of DNA [7].
This is also consistent with the slow diffusion and rotational relaxation of minor
groove water.
The enhanced ordering of water molecules in the minor groove of DNA was
found a long time ago in X-ray studies and was termed spine of hydration [8]. Thus,
the experimentally observed spine of hydration in the crystal structure can be
understood in terms of the lowest entropy of water molecules in the minor groove
region in the aqueous solution.
10.7 Correlation between diffusion and entropy: AdamGibbs relation
In liquids, one of the most celebrated equations that connects the microscopic
dynamics of molecules with thermodynamics is the AdamGibbs relation. This
relates the translational diffusivity of the system to congurational entropy as
follows:
D
T
= A exp
C
TS
C
_ _
(10:1)
Here D
T
is the translational diffusivity, S
C
is the congurational entropy, and A
and C are temperature-independent constants. This beautiful relation between
dynamics and thermodynamics has been used extensively to understand the
10.7 Correlation between diffusion and entropy: AdamGibbs relation 157
reduced diffusion in restricted environments or in supercooled liquids. The valid-
ity of the AdamGibbs relation is usually considered to be a sign of the collective
nature of the relaxation.
As mentioned earlier, it is difcult to obtain a quantitative measure of entropy. By
using the 2PT method (the method will be described later in Chapter 19), one can
obtain the entropy of water molecules in both major and minor grooves of DNA.
One can also get a measure of the translational diffusivity of those water molecules
from the mean-square displacement or velocity autocorrelation function all these
are fortunately easily available with computer simulations.
The results obtained from such a calculation were quite interesting. It was found
that the logarithm of diffusivity exhibits a linear dependence with 1/TS
C
when these
two quantities are obtained for different region of aqueous DNA. This suggests that
while the dynamics and thermodynamics of the water molecules in the diverse
regions of DNA are quite different from each other, they obey the AdamGibbs
relation quite nicely (Figure 10.3). It was also found that AdamGibbs plots at
different temperatures collapse on each other, indicating the temperature-
independent nature of the entropic barrier (C).
This result provides a strong microscopic basis for the observed spine of hydra-
tion along the AT minor grooves as demonstrated earlier by X-ray crystallography.
Figure 10.3. Correlations between diffusion coefcient, congurational entropy, and
tetrahedral order parameter (t
h
). Note that the left side of the y-axis represents the
logarithm of diffusivity and the right side of the y-axis represents (t
h
). The straight
line tting of the data validates the AdamGibbs relation between entropy and the
diffusion coefcient, as discussed in the text. The dashed line shows the correlation
between (t
h
) and congurational entropy. Adapted with permission from J. Phys.
Another interesting aspect of this observation relates to another celebrated equa-
tion, the StokesEinstein relation between diffusivity and viscosity. It tells us that
one can now dene the viscosity in a nano-conned region (grooves of DNA),
which is termed the microviscosity of that particular region [7]. The StokesEinstein
relation, where the translational diffusion constant and viscosity are inversely
related, provides a remarkable correlation between microviscosity and congura-
tional entropy.
In the DNA grooves, the molecular arrangement of water molecules gets modied
from that in the bulk. As discussed in Chapter 2, a useful characterization of the local
order in liquid water is provided by the tetrahedrality parameter, t
h
. The value of this
parameter in liquid water increases with lowering temperature as the liquid becomes
more tetrahedrally arranged. In a conned region, the effect of connement on this
tetrahedral order is not clear. Thus, it came as a bit of a surprise that the tetrahedral
order parameter increases with connement (see Figure 10.3). Additionally, the
conguational entropy of water in different regions is also found to be well correlated
with the tetrahedrality parameter, as also shown in Figure 10.3. It is clear from the
gure that order increases with decreasing congurational entropy [9].
Here also, the dependence of the translational diffusivity of water molecules in
different regions of both sequences on the congurational entropy of the respective
regions is examined.
10.8 Sequence dependence of DNA hydration: spine of hydration
in AT minor groove
So far, we have discussed the dynamics and thermodynamics of the water molecules
in the groove region of an aqueous DNAwithout paying attention to the sensitivity
of the dynamics on the DNA sequence. According to WatsonCrick base pairing,
two kinds of paring are possible and those are AT and GC. B-DNA made of these
two kinds of pairing gives rise to a different groove structure in terms of conne-
ment. The major groove is wider for both sequences. A large difference exists
between the minor groove structures of the two sequences. The minor groove of
the AT sequence is narrower (width 3.5 ) and deeper than that of the poly-GC
sequence (width 6 ). This implies that water molecules present in the minor groove
of the AT sequence are more conned compared to other types of grooves.
These differences in connement in groove structure for these two types of
sequences have severe effects on the surrounding water dynamics. The OOO
angle distribution of the water molecules in the major and minor grooves can
provide information about ordering. The distribution usually has two characteristic
peaks. The peak at 100 in water directly probes the amount of tetrahedrality present
10.8 Sequence dependence of DNA hydration 159
in the system. However, the peak at 60 probes the amount of interstitial water
present in the system.
The plot of the OOOangle distribution of water molecules present in the minor
grooves of the poly-AT sequence shows an enhancement of the peak at 100 (the
peak is shifted to 120) for the AT minor groove water compared to bulk water
distribution. The shift is due to the formation of an ice-like structure, which is a
trademark of the spine of hydration. This provides clear evidence of the presence of
a strong tetrahedral arrangement of water molecules in the AT minor grooves.
However, we nd a reduction of the peak value for the interstitial water molecules
(60) in the AT minor groove as compared to that for bulk water. This enhancement
of the tetrahedral ordering in the minor grooves of the poly-AT sequence can be
understood as a consequence of strong connement [9].
The plot of the OOO distribution of water molecules present in the minor
grooves of the poly-GC sequence shows that no such enhancement of the peak at
~100 as compared to the AT minor groove is observed, indicating that the water
molecules in the GC minor groove cannot form a strong tetrahedral network.
These results can be understood as follows. The presence of a strong inter-
base interaction in the GC strand (three HBs between G and C) makes the
groove structure quite rigid, and hence the interaction of water with the DNA
atoms is frustrated. The DNA atoms do not undergo deformations in
position to accommodate the preferred orientation of water to form a stable
tetrahedral structure. The opposite is true for the AT minor groove (two HBs
between A and T), where the intrabase interaction is less compared to that for the
GC strand. Here the groove structure can be deformed in such a way that water
molecules can attain their preferred orientation to form a stable tetrahedral
structure.
On the other hand, the OOO distribution for water molecules present in the
major grooves of both sequences shows hardly any difference between them. This
signies that the water molecules are structurally quite similar in the major grooves
of both sequences.
As shown in Figure 10.3, the values of the tetrahedral order parameter are found
to be 0.41, 0.47, 0.48, 0.52, and 0.57 for bulk water, the major groove water of poly-
GC, major groove water of poly-AT, the minor groove water of poly-GC, and the
minor groove water of poly-AT, respectively. Thus the tetrahedral ordering in the
groove of DNA increases with increasing connement.
This difference in the structure of water molecules present in the different
regions of the poly-AT and poly-GC sequences provides a microscopic
explanation of the observed thermodynamics and dynamics of water molecules in
these regions.
10.9 Effects of nanoconnement and surface-specic interactions
The total energy of the WatsonCrick HBs of the GC pair (three HBs) is 16.8 kcal/
mol and that of the AT pair (two HBs) is 7.0 kcal/mol. Thus, the average single
hydrogen-bonding energy of the GC pair is 5.6 kcal/mol and the same for the AT
pair is 3.5 kcal/mol. This indicates that the DNA atoms in the AT grooves are more
amenable to move (bend or distort) to ensure the interaction with the water mole-
cules, and thus the interaction energy between groove water molecules and groove
atoms may become higher for AT grooves than for GC grooves. The similar
dynamic behavior of water molecules in the major grooves of both poly-AT and
poly-GC sequences demonstrates that such interaction effects are not important for
the major grooves. The major grooves of both sequences have similar widths and
also the values of the tetrahedral order parameters in the major grooves are similar.
This result clearly implies the greater effect of connement, which is directly
correlated with the tetrahedral ordering of water in the grooves, rather than surface-
specic interactions on the water dynamics.
The slower dynamics of water molecules in the AT minor grooves compared to
the dynamics in the GC minor groove is observed. This result is in accordance with
the facts that (1) water molecules have higher interaction energies in the AT minor
groove rather than in GC and (2) water molecules in the AT minor groove experi-
ence greater connement than in the GC minor grooves. However, comparison of
the dynamics in the major grooves reveals that connement has the greater effect
and surface-specic interactions have little effect on the dynamics. To repeat, the
slowest dynamics of water molecules in the AT minor groove is primarily due to the
greater connement, which enhances the tetrahedral ordering in the AT minor
groove, rather than the stronger surface-specic interactions.
10.10 Water around RNA
RNA is the same as DNA except in two aspects. (a) The base in RNA is the sugar
ribose, while that in DNA is the sugar deoxyribose (hence the name DNA), and
(b) RNA contains the nucleotide base uracil while DNA contains thymine. RNA
molecules play diverse biological functions within our body, including transcription
of the gene through mRNA and translation in the ribosome through t-RNA. Some
RNAs also serve as enzymes. As RNAmolecules are single-stranded, they can bend
and fold locally to form diverse three-dimensional structures, and exhibit base
stacking that is accompanied by local bending of the RNA chain.
Several studies have explored the possible role of water in the biological activities
of RNA. Local structures of nucleic acids result from interplay between the solvent
and nucleic acid molecules so that both together constitute a functional and structural
10.10 Water around RNA 161
entity [10], although the dynamic role of water is just beginning to emerge. In helical
duplexes of RNA, due to the periodicity of contacts between water and each repeating
unit, explicit hydration patterns have often been observed. Here we discuss a few
aspects of the structure, dynamics, and biological function of water around RNA.
10.10.1 Structure of water around RNA
Cavities in the tertiary structure of RNA, formed by the polynucleotide folding,
contain trapped water molecules [10]. The crystal structure showed that the catalytic
core of wild-type (WT) hairpin RNA has a 350
3
volume inter-domain cavity. In
this context we should mention the MD simulation studies performed by Rhodes
and co-workers. They started the simulation by taking the crystal structure of the
RNA without any crystal water molecule as the initial structure and found ve to
seven water molecules entering that cavity during the course of the simulation [11].
They also observed that those water molecules form interconnected strings with a
total of 16 HB connections and RNA functional groups forming a coupled
hydrogen-bonding network in the catalytic core.
10.10.2 Dynamics of water around RNA
Trapped water molecules inside the RNA exhibit different dynamics depending on
the nature of the hydrophilic site they are attached to and the number of water
molecules encapsulated inside. Water molecules attached to atoms of the phosphate
group generally have a long residence time (as long as 700800 ps) whereas water
bound with a shallow groove has an especially shorter residence time (~100 ps)
[12]. Other water molecules with residence times in the 500 ps range are mainly
located in the vicinity of deep groove atoms. These ndings illustrate the large
dynamic heterogeneity present in the solvation layer of RNA. Moreover, the long-
lived water-mediated interactions and the extensive hydration are found to partici-
pate in the stabilization of RNA tertiary motifs and helices.
10.11 Conclusion
The structures and dynamics of water molecules in the DNA grooves play an
important role in the biological function of DNA. As discussed later, the intercala-
tion of the anti-cancer drug daunomycin into DNA is indeed facilitated by water
molecules in the grooves. While the role of water molecules in stabilizing a DNA
duplex is well known (water molecules screen the electrostatic repulsion between
the negatively charged phosphate ions, formHBs with the polar atoms of the nucleic
acid and indirectly promote the HB interaction between the aromatic rings of the
nucleic acids), its dynamic role is relatively less explored and less known. As water
occupies the major and minor grooves of the DNAdouble helix, it plays a myriad of
roles.
Hydration of the grooves is particularly relevant for AT minor grooves because
the groove is both narrow and deep. Water molecules are known to form a spine of
hydration along the minor grooves. Recent computer simulation studies have
quantied the nature of the spine of hydration by using concepts from bulk water.
Thus, one has used the tetrahedrality order parameter to demonstrate that the water
molecules are indeed partly ice-like. The entropy calculation has further substan-
tiated these observations.
Being a single strand, RNA is more open to the interaction of water.
Unfortunately there exist fewer studies of water dynamics around RNA. Here the
structure is relatively less rigid and also more open, allowing water molecules to
sustain stronger interactions with the nucleic acids.
APPENDI X10. A HYDROGEN- BONDI NG PATTERN
AROUND DNA
Hydration is greater and water is more strongly held around the phosphate groups
that run along the inner edges of the major grooves than anywhere else in the DNA.
This greater hydration around the charged phosphate groups plays an important role
in stabilizing the double-helix structure by screening the negative charges of the
phosphate groups of the two strands. In DNA, the bases themselves are involved in
pairing through hydrogen-bonding. However, even these groups, except for the
hydrogen-bonded ring nitrogen atoms (pyrimidine N3 and purine N1), are capable
of forming more HBlinks to water within the major or minor grooves in B-DNA(as
shown in Figure 10.A.1). Theoretical considerations and MD simulations indicate
that both grooves are equally hydrated with hydration roughly in the following
order: C
N4
/G
N2
/T
O2
> A
N6
/C
O2
/G
O6
> A
N3
/G
N3
/G
N7
/T
O4
>> A
N7
. In B-DNA,
guanine forms an HB to a water molecule from both the minor groove 2-amino
and major groove 6-keto groups with further single hydration on the free ring
nitrogen atoms (minor groove N3 and major groove N7). Cytosine can form an
HB to a water molecule with the major groove 4-amino and minor groove 2-keto
groups. Adenine forms an HB to a water molecule from the major groove 6-amino
group with further single hydration on the free ring nitrogen atoms (minor groove
N3 and major groove N7). Thymine can form an HB to a water molecule from both
the minor groove 2-keto and major groove 4-keto groups. Phosphate hydration in
the major groove is thermodynamically stronger but exchanges faster. There are six
hydration sites per phosphate, not including hydration of the linking oxygen atoms
to the deoxyribose or ribose residues. The deoxyribose oxygen atoms (O3 phos-
phoester, ring O4 and O5 phosphoester) all hydrogen bond to one water molecule,
whereas the free 2-OH in ribose is much more capable of hydration and may hold
on to about 2.5 water molecules. It should be noted that about 2% of the hydrating
water molecule sites may be transiently replaced by cations.
References
1. J. M. Berg, J. L. Tymoczko, and L. Stryer, Biochemistry, 5th edn. (New York: W H
Freeman, 2002).
2. J. D. Watson and F. H. C. Crick, Molecular structure of nucleic acids: a structure for
deoxyribose nucleic acid. Nature, 171 (1953), 737738.
3. M. Feig and B. M. Pettitt, A molecular simulation picture of DNA hydration around
A- and B-DNA. Biopolymers, 48 (1998), 199209.
4. S. Pal, P. K. Maiti, and B. Bagchi, Exploring DNA groove water dynamics through
hydrogen bond lifetime and orientational relaxation. J. Chem. Phys., 125 (2006),
234903; S. Pal, P.K. Maiti, and B. Bagchi, Anisotropic and sub-diffusive water motion
at the surface of DNA and of an anionic micelle CsPFO. J. Phys. Cond. Matt., 17 (2005),
S4317.
5. S. K. Pal, L. Zhao, and A. H. Zewail, Water at DNAsurfaces: ultrafast dynamics in minor
groove recognition. Proc. Natl. Acad. Sci. USA, 100 (2003), 81138118.
6. D. Andreatta, et al., Power-law solvation dynamics in DNA over six decades in time.
J. Am. Chem. Soc., 127 (2005), 72707271.
Figure 10.A.1. Possibility of HBs with water molecules and the base pairs in the
major and minor grooves of DNA (schematic). Adapted with permission from
J. Chem. Phys., 125 (2006), 234903. Copyright (2006) American Institute of
Physics.
7. S. Pal, P. K. Maiti, B. Bagchi, and J. T. Hynes, Multiple time scales in solvation
dynamics of DNA in aqueous solution: the role of water, counterions, and cross-
correlations. J. Phys. Chem. B, 110 (2006), 2639626402.
8. B. Jana, S. Pal, P. K. Maiti, S. Lin, J. T. Hynes, and B. Bagchi, Entropy of water in the
hydration layer of major and minor grooves of DNA. J. Phys. Chem. B, 110 (2006),
1961119618.
9. R. E. Dickerson, in Structure and Methods: DNA and RNA (R. H. Sarma and
M. H. Sarma, eds) (Schenectady, NY: Adenine, 1990), vol. 3, pp. 138.
10. E. Westhof, Water: an integral part of nucleic acid structure. Annu. Rev. Biophys. Chem.
17 (1988), 125144.
11. M. M. Rhodes, K. Rblov, J. Spooner, and N. G. Walter, Trapped water molecules are
essential to structural dynamics and function of a ribozyme. Proc. Natl. Acad. Sci. USA,
103 (2006), 1338013385.
12. P. Aufnger and E. Westhof, RNA solvation: a molecular dynamics simulation per-
spective. Biopolymers, 56 (2000), 266274.
References 165
11
ProteinDNA interaction: the role of water
as a facilitator
ProteinDNAinteraction is one of the most fascinating yet least understood
subjects of molecular biology, despite its paramount importance in a large
number of biological processes. It is hard to quantitatively understand the
nature and stability of the proteinDNA complex, given that the complex
can be delocalized, with the protein moving along the DNA chain, and
can even be transient. Water molecules wet the interface between them to
allowsuch a complex formation. It is natural to expect that water molecules
in the grooves of the DNA play an important role in proteinDNA interac-
tion. Proteins themselves are solvated by ions and water. This interaction
therefore involves displacement of a large number of water molecules
which nevertheless must form many HBs and facilitate hydrophobic bond-
ing between amino acid residues and base pairs. Thus, water molecules at
the interface play a multitude of roles which are beginning to be under-
stood. Just as in other problems of biology, each system behaves differently
and many studies are required before a quantitative picture can emerge.
11.1 Introduction
ProteinDNA interactions are essential for activating several cellular machines in
living cells, especially for the transcription, replication, and recombination pro-
cesses of genes. Needless to say, water plays an important role in each of these
processes. The quantitative understanding of the role of water in proteinDNA
interactions is still largely incomplete. We have already discussed in Chapter 7 the
role of water molecules in drugDNA intercalation and shall discuss in Chapter 13
the role of water in the kinetic proofreading in transcription. The ultimate goal is to
understand such processes in molecular detail.
Fortunately, macromolecular crystallographic studies can now be used to explore
the role of water in controlling both the specicity and the afnity of proteinDNA
interactions [1]. Such analyses of proteinDNAcrystal structures revealed that several
distinct contributing factors are involved in such complex formation, such as HBs,
167
electrostatic interactions, direct and indirect contacts between amino acids and phos-
phate groups, sugars and bases, water-mediated contacts, hydrophobic effects, ion
release, etc. [2]. Water-mediated interactions of course depend on acceptordonor HB
relationships with the atoms and residues of the macromolecules, on the electrostatic
elds present near the complex, and undoubtedly on packing density [3].
The crystal structures of proteinDNA complexes reveal the presence of several
ordered water molecules at proteinDNA interfaces. Such water molecules may
reside in the solvation shells of the protein before their binding to DNA, and they
may serve to ll all the gaps arising from imperfect matches of protein and DNA
surfaces to maintain a suitable packing density for the system, or they may act as
mediators of the proteinDNA recognition process. Some water molecules are also
found in the interior cavities of such macromolecules.
Surface water molecules can undergo fast exchange with the bulk solvent, as they
are comparatively less ordered than those water molecules located in the interior
of biomolecules. Interior water molecules are in much slower exchange with the
bulk solvent and are considerably well-ordered as they are strongly bound by the
biomolecules [4]. Such inexibility of the buried water molecules results in a
signicant entropic cost, at signicant enthalpic gain. In this chapter we highlight
topical advances in current understanding of the biological role of water in protein
DNA interactions.
11.2 Structural analysis of proteinDNA complex: classication
of hydration water
Water molecules participate in different hydration patterns in the solvation shell of
both protein and nucleic acids through hydrogen-bonding networks where water
molecules mainly connect side-chain and main-chain atoms with the functional groups
on the bases, and the anionic oxygen atoms of the phosphodiester backbone. Different
locations of water molecules in the proteinDNA complex serve in the recognition
process [5]. Depending on the direct or indirect links of water in the proteinDNA
complex, the hydration water molecules are categorized into four classes:
(i) those that make contact with both the protein and the DNA concurrently
(bridged water molecules) and participate in recognition directly; these water
molecules are clearly important, although they could be small in number;
(ii) those that are in contact with either the protein or the DNA solely;
(iii) water molecules that are proximal to hydrophobic atoms of either protein or
DNA; these water molecules are mobile and can easily be replaced;
(iv) those that interact entirely with other water molecules; these can be character-
ized as free water in the nomenclature of Chapters 5 and 7.
168 ProteinDNA interaction: the role of water as a facilitator
Analysis of a large number of proteinDNA complexes leads to the following
quantication of the above classication. Only about 510% of surrounding
water molecules belong to class (i) and 7080%belong to class (ii). The rest belong
to classes (iii) and (iv) [3].
Both crystallographic as well as MD simulation studies have pointed out that class
(ii) water molecules are involved in facilitating the proteinDNA binding process
by screening unfavorable electrostatic interactions. Very few water molecules (less
than 2%) mediate an HB between donor atoms of protein and acceptor atoms of
DNA. When protein atoms cannot reach the DNA due to some structural restriction,
these water molecules provide an extension to the side-chain atoms through
hydrogen-bonding. Thus they act as bridges to make the proteinDNA complex
formation feasible. However, these bridge water molecules are a minor fraction
relative to the total number of hydration-layer water molecules. Nevertheless, they
play an important role in recognition.
For example, in the trp repressorDNA complex there are only a few direct
contacts between the protein and the base-pairs [6]. It is interesting to note here that
three ordered water molecules are found at the proteinDNA interface that are
hydrogen-bonded with both the base pairs and the protein side-chain atoms. These
water-mediated bases are among the most important parts in controlling the repres-
sors afnity for the operator sequence in the DNA.
Another example is obtained in a structure of the 434 repressor that binds to a
DNA site with a lower afnity. This is similar to the trp repressorDNA complex
where glutamine residues have been observed to interact only indirectly with DNA
bases via water bridges [7]. The importance of such water molecules facilitating
the binding process and recognition is further highlighted by the crystallographic
studies. Molecular dynamics simulation studies of the paired homeodomainDNA
complex and estrogen receptorDNA complex have also elucidated the role of such
water molecules [8]. Aconsiderable number of ordered water molecules were found
to direct the specicity to nuclear hormone receptors, to mediate the recognition
process, and to enhance the stability of the solute.
11.3 Dynamics of water around a proteinDNA complex
As discussed earlier, the usual ultrafast dynamics of water can become relatively
slower in conned systems (such as reverse micelles and nanotubes) and also on
interacting surfaces such as silica, protein surfaces, and in the grooves of DNA.
Among different experimental methods, NMR techniques are more sensitive to the
residence times of interior water molecules within internal polar cavities than to
the surface water. Surface water molecules are rarely detected in NMRas they are in
fast exchange with the bulk solvent, having residence times less than 500 ps. Interior
11.3 Dynamics of water around a proteinDNA complex 169
water molecules within cavities are rather slow in exchange with the bulk. As for
example, in BPTI, residence times range from ~15 ns to ~200 ns.
Formation of a stable proteinDNAcomplex involves the rearrangement of water
molecules and release of counter ions and water molecules to the bulk. Zewail and
co-workers have used the time-resolved uorescence up-conversion technique to
explore the dynamics of the histoneDNA complex formation and the participation
of hydration water in the stability and specicity of the recognition process [9]. This
important study established the contribution from the entropic gain due to the
release of hydration water (often termed dynamically ordered water) to the bulk.
Molecular dynamics simulations were employed to explore the dynamic hetero-
geneity in water motion around a proteinDNA complex. The complex chosen
was the DNA-binding domain of human TRF1 protein and a telomeric DNA (see
Figure 11.1). It was observed that the slow water dynamics correlated with the
relaxation time of HBs of the water molecules connected with the protein and the
DNA. The restricted motion of such water molecules led to sluggish rotational and
translational dynamics of the hydration layer near the protein and DNAmolecules in
their complexed forms as well as in their free forms (see Figure 11.2).
Hydration water present near the major groove differs from that near the minor
grooves of DNA molecule in its complexed form. This is evident from the transla-
tional motion. We have discussed previously that when a DNAis in a free condition,
water molecules close to the major groove are relatively more mobile than those
near the minor groove [10,11]. In contrast, when the DNA molecule adapts the
complexed form, both the rotational and the translation dynamics of water mole-
cules behave in the opposite manner. With HB dynamics, rather frequent breaking
and re-formation of interfacial waterwater HBs is found to modulate the exibility
of the proteinDNA hydration layer. Moreover, the transverse and longitudinal
degrees of freedom of interfacial water molecules are greatly inuenced by the
formation of a proteinDNA complex in an inhomogeneous manner.
11.4 Role of water in thermodynamics of proteinDNA interactions
While the importance of ordered water molecules is obvious for the specicity of
proteinDNA interactions, these molecules also have a vital role in determining the
afnity or thermodynamics of such complexes. The formation of a stable protein
DNA complex of course needs a negative Gibbs free energy (G) of formation. The
change in free energy during complex formation depends upon the change in both
entropy (S) and enthalpy (H) as G = H T S), where T symbolizes the
temperature.
The enthalpy term is modulated with the change in noncovalent interactions
(electrostatic interactions and HBs) during complex formation. While enthalpy is
associated with molecular interactions, entropy is involved with multiple protein
and DNAconformations, variations in the structure of water molecules and counter-
ions, and other components. It is difcult to quantify the individual contribution of
each HB and ionic interaction. The individual contribution from G, H, and S
has been studied by using highly sensitive microcalorimetry experiments during
proteinDNA complex formation.
Figure 11.1. Snapshot of a proteinDNA complex showing the helix-turn-helix
(HTH) binding motif of a protein associating with the target bases via the side-
chains. The common water molecules are highlighted here. The color code is as
follows: the HTH binding motif of the protein is indicated in blue, the remaining
part of the protein in colored red, and the DNA molecule is green. The water
molecules that are simultaneously hydrogen-bonded to the protein and DNA in the
complex are colored gray and the remaining water molecules are shown in magenta.
Adapted with permission from J. Chem. Phys., 135 (2011), 135101. Copyright
(2011) American Institute of Physics. See plate section for color version.
The Cro repressor, trp repressor, met repressor, and glucocorticoid receptor DNA-
binding domain (GRDBD) protein were mainly selected for complexation with their
respective DNA targets that allow such quantitative determination [1214]. The
results of the estimated values for three proteins are shown in Table 11.1.
Interestingly, formation of the proteinDNA complex usually causes a large nega-
tive change in heat capacity. This implies loss of uctuations and formation of a rigid
structure. This has been attributed to the release of water molecules from the surface
of the complex. Water itself has a large heat capacity due to hydrogen-bonding. Such a
Figure 11.2. Translational and rotational dynamics of water molecules determined
by computer simulations in complexed form (solid line) with protein and DNA as
well as in the free component (dashed line). (a) Mean-square displacements (MSDs)
of water molecules residing in the rst hydration shell, water molecules in the major
and minor groove regions of the DNA, and water molecules present in the common
region of the complex are calculated and shown in the gure for both the complexed
form and the free form. (b) The reorientational time correlation function, C (t),
derived for the same water molecules located as mentioned above both in complexed
and free forms of protein and DNA. The comparison with the pure bulk state is also
highlighted in both gures. Adapted with permission fromNature Struct. Mol. Biol.,
16 (2009), 1224. Copyright (2009) Nature Publishing Group.
release can only partially explain the loss of heat capacity. In part, it may be due to the
presence of ordered water molecules and their large residence time.
Recent studies using osmotic pressure have allowed determination of the number
of water molecules that are displaced in specic and nonspecic binding of protein
to DNA. Such a method has been used to study the interaction of the gal repressor
with DNA [15]. Depending upon the nature of the osmolyte, between 100 and 180
water molecules were displaced during the above proteinDNAcomplex formation.
This appears to be an unusually large number and needs further conrmation.
It is interesting to note that shifting the repressor from a nonspecic complex to a
specic complex, the same number of water molecules are released as in the original
case with the specic complex formation. This is a bit surprising. This indicates
that no additional water molecules are released in the formation of a nonspecic
complex. A similar result has been found for the nonspecic complex between the
highly hydrated mutant GRDBD and a noncognate DNA. However, it is hard to
connect the actual number of water molecules with those found in complex struc-
tures. The binding of the gal repressor to a suboptimal DNA target has also been
studied extensively. In this case, the binding afnity is 2-fold lower than that of an
optimal DNA target, and it correlates with the release of six fewer water molecules.
In the case of the estrogen receptor DNA-binding domain protein bound to a DNA
target a similar number of water molecules are released at the interface, but the
binding afnity is 10-fold lower. Hence there may be other effects that can explain
the loss of binding afnity.
Table 11.1. Thermodynamic estimation calculated for met repressor, trp repressor,
and glucocorticoid receptor DNA-binding domain interactions with their DNA
targets [4]. Adapted with permission from Curr. Opin. Struct. Biol., 7 (1997), 126.
Copyright (1997) Elsevier.
Complex Temperature (C) G (kcalmo1
1
) H (kcalmo1
1)
-TS (kcalmo1
1
)
Met repressor 10.7 8.1 0.9 7.1
25.6 7.6 4.1 3.6
35.8 7.4 9.5 2.2
Trp repressor 10 11.9 3.9 8.0
13 12.0 5.8 6.2
16 12.1 8.0 4.1
20 12.1 12.8 0.6
40 12.4 33.6 21.3
GRDBD 10 15.2 9.5 24.7
18 15.5 8.2 23.7
23 15.8 7.3 23.1
34 16.4 3.8 20.2
11.5 Protein diffusion along DNA
For various biological functions within the cell, diffusion of different proteins along
segments of DNA chains is essential. This diffusion takes place in the search for
predetermined specic binding sites. Surprisingly such a search is rather efcient,
with search times observed to be about two orders of magnitude lower than theoretical
estimates based on a three-dimensional random walk in space. This particular obser-
vation has given rise to many discussions in the literature, with suggestions on the
mode of transport of the protein. It has been proposed that a protein combines a one-
dimensional sliding motion along a DNA duplex with three-dimensional hops to
minimize the search time. It has also been proposed that the sliding motion involves
spinning around the DNA chain. In Figure 11.3 we show a schematic illustration of
a possible scenario [16].
An important issue here is the free-energy landscape of the proteins motion
along the DNA. Theoretical studies seem to suggest that the barrier towards the
sliding motion is rather small, of the order of 1 kcal/mole or so. This small barrier
means that the proteinDNA nonspecically bound complex remains hydrated by
water throughout the motion, and water facilitates such transport in a large measure,
acting as a lubricant along the diffusion pathway.
The free-energy landscape faced by the protein as it moves around DNA is hard
to calculate precisely because so many different interactions are involved. An
important point to note is that a protein is much larger than a base pair of DNA
and thus can be in contact with a large number of bases. As mentioned earlier,
typically helical segments of the protein are inserted into the major grooves of the
DNA. The landscape is geometrically rugged as the protein faces different base pairs
along different segments of the DNA chain.
One way to understand the lack of any signicant barrier along the diffusion path
is the presence of the large number of interactions. Thus, the energy of the protein
along the diffusion pathway can be a Gaussian function of energy, according to the
famous central limit theorem of probability. As many interactions are involved, the
width of the distribution could be narrow.
This of course does not address the issue of the interactions responsible for
specic binding of the protein to the required binding site.
11.6 Conclusion
The orientation of the protein with respect to the DNA strand is an important issue
in the proteinDNA interaction. As evident from Figure 11.2, this orientation is
specic for each proteinDNA complex. Clearly, water molecules that are strongly
hydrogen-bonded to the charged or polar amino acid residues remain bonded to
those groups and additionally form the bridge between the DNA atoms. Probably
the energy contribution from water molecules at the proteinDNA interface signi-
cantly affects the specicity and stability of the proteinDNA complex, particu-
larly those involved in bridging the water molecules. It is easy to see that these
bridging water molecules are initially hydrogen-bonded to the proteins side-chain
residues and thus act as glue to join the protein with the target DNA [10].
Almost all proteins displace surface water molecules on binding to DNA. This
provides a favorable entropic cost to the change in free energy.
(a) (b)
(c)
Figure 11.3. Schematic illustration of the motion of a nonspecically bound protein
along rigid DNA chain segments. The motion involves the spinning of the protein
along the DNAhelix. Panel (a) approximates the protein as a sphere while (b) shows
the spinning motion. Panel (c) shows the small free-energy barriers involved. The
barriers are random, separated by minima, thus the landscape can be approximated
as rugged, allowing the use of the well-known Zwanzig expression to mimic the
retarding effects of the minima and the maxima on the diffusive sliding of the protein
along the chain.
11.6 Conclusion 175
References
1. E. Westhof, Structural water bridges in nucleic acids. In Water and Biological
Macromolecules (Westhof, E., ed.) (Boca Raton, FL: CRC Press, 1993), pp. 226243.
2. R. G. Brennan, S. L. Roderick, Y. Takeda, and B. W. Matthews, Protein-DNA confor-
mational changes in the crystal structure of a lambda Cro-operator complex. Proc. Natl.
Acad. Sci. USA, 87 (1990), 8165.
3. C. K. Reddy, A. Das, and B. Jayaram, Do water molecules mediate protein-DNA
recognition? J. Mol. Biol., 314 (2001), 619.
4. J. W. Schwabe, The role of water in proteinDNAinteractions. Curr. Opin. Struct. Biol.,
7 (1997), 126.
5. F. Spyrakis, P. Cozzini, C. Bertoli, A. Marabotti, G. E. Kellogg, and A. Mozzarelli,
Energetics of the protein-DNA-water interaction. BMC Struct. Biol., 7 (2007), 4.
6. Z. Shakked, G. Guzikevich-Guerstein, F. Frollow, D. Rabinovich, A. Joachimiak, and
P. B. Sigler, Determinants of repressor/operator recognition from the structure of the trp
operator binding site. Nature, 368 (1994), 469.
7. D. W. Rodgers and S. C. Harrison, The complex between phage 434 repressor DNA-
binding domain and operator site OR3: structural differences between consensus and
non-consensus half-sites. Structure, 1 (1993), 227.
8. D. Kosztin, T. C Bishop, and K. Schulten, Unbinding of retinoic acid from its receptor
studied by steered molecular dynamics. Biophys. J., 76 (1999), 188.
9. D. Zhong, S. K. Pal, and A. H. Zewail, Femtosecond studies of proteinDNA binding
and dynamics: histone I. Chem. Phys. Chem., 2 (2001), 219.
10. S. K. Sinha and S. Bandyopadhyay, Dynamic properties of water around a protein
DNA complex from molecular dynamics simulations. J. Chem. Phys., 135 (2011),
135101.
11. B. Jana, S. Pal, and B. Bagchi, Enhanced tetrahedral ordering of water molecules in
minor grooves of DNA: relative role of DNA rigidity, nanoconnement, and surface
specic interactions. J. Phys. Chem. B, 114 (2010), 3633.
12. D. E. Hyre and L. D. Spicer, Thermodynamic evaluation of binding interactions in the
methionine repressor system of Escherichia coli using isothermal titration calorimetry.
Biochemistry, 34 (1995), 3212.
13. J. E. Ladbury, J. G. Wright, I. M. Sturtevant, and P. B. Sigler, Athermodynamic study of
the trp repressoroperator interaction. J. Mol. Biol., 238 (1994), 669.
14. T. Lundbck and T. Hrd, Sequence-specic DNA-binding dominated by dehydration.
Proc. Natl. Acad. Sci. USA, 93 (1996), 4754.
15. M. M. Garner and D. C. Rau, Water release associated with specic binding of gal
repressor. EMBO. J., 14 (1995), 12571263.
16. P. C. Blainey, G. Luo, S. C. Kou, et al., Nonspecically bound proteins spin while
diffusing along DNA. Nature Struct. Mol. Biol., 16 (2009), 1224.
12
Water surrounding lipid bilayers: its
role as a lubricant
The lipid bilayer is one of natures amazing devices that serves a multi-
tude of purposes in sustaining life forms. Water is an integral part of a
lipid bilayer. Water molecules surround lipid bilayers and effectively
stabilize them from both the outside and the inside to support the cell
walls. Unlike protein surfaces, here the surfaces that water molecules
face are nearly homogeneous. Another major difference between the two
systems is that the lipid surfaces are more mobile. The mobility is required
as many chemicals need to be transported to the inside of the biological
cells that are covered and nurtured within these bilayers. Water is clearly
involved in the entire process of transfer, starting from the mobility of the
walls. Here we discuss the properties of the water that surrounds protein
bilayers.
12.1 Introduction
Lipids are amphiphilic surcants characterized by one head group connected to two
long hydrocarbon tails. In Figure 12.1 we show the molecular structure of a
phospholipid and also a bilayer formed by aggregation of these lipids. As can be
seen from the gure, each layer of the lipid bilayers is formed such that the
hydrophilic polar head groups face water while the hydrophobic hydrocarbon tails
pack together to form the interior.
The most important bilayers are formed by the phospholipids, shown in
Figure 12.1(a).They form the membrane that covers the surface of our blood cells.
Bilayer structures are stabilized by solvation of the head groups by water and by the
nearly total avoidance of water by the hydrophobic hydrocarbon tails. We observe
the same pattern in many self-assemblies found in nature, such as micelles, reverse
micelles, and microemulsions, discussed later.
Lipid bilayers are, however, not rigid, as they serve as the gateway to the
interior of the blood cells. All the matter needs to be transported back and forth
177
between the interior and the exterior of the cell. The transport is often accom-
plished by proteins which remain submerged in the bilayers, as shown in
Figure 12.1(b).
In order to serve their biological purpose, the lipid bilayers need to be quite
mobile and elastic to allow easy distortion. This is accomplished by the presence of
(b)
Figure 12.1. (a) One phospholipid, showing the head and the tails. Adapted with
permission from http://captain-nitrogen.tumblr.com/post/3201048590/hey-fatty.
(b) Structure of bilayer by aggregation of these lipids and formation of cell walls
across which transport proteins carry chemicals in and out of the cell. See plate
section for color version of (b).
(a)
178 Water surrounding lipid bilayers: its role as a lubricant
water around the head groups and a small number of water molecules inside. Thus,
water acts as a lubricant to facilitate the function of the membrane.
The cellular membrane contains many other molecules. As already mentioned,
it includes trans-membrane proteins that transport important nutrients and chemi-
cals to the interior of the cell across the membrane. The membrane also includes
other lipids, such as cholesterol etc., that serve to strengthen the phospholipid
bilayers.
In this chapter we are concerned with the structure and dynamics of water around
lipid bilayers. This area has seen a good deal of studies in recent times. Both
experiments and simulation studies have been carried out to understand the correla-
tion between water and lipid motions.
12.2 Hydration of different constituents: phospholipids
and buried proteins
In phospholipid bilayers the charged phosphate groups are located in the hydrated
region, approximately 0.5 nm outside the hydrophobic core. In some cases, the
hydrated region can extend much further into the bulk, for instance in lipids with a
large protein or a long sugar chain grafted to the head. Thus, almost nearly 1 nm of
the water layer is strongly inuenced by bilayers.
Next to the hydrated region is an intermediate region that is only partially
hydrated. This boundary layer is approximately 0.3 nm thick. Within this short
distance, the water concentration drops from 2 M on the headgroup side to nearly
zero on the tail (core) side [1]. However, in some cases water is known to access
pretty deep into the buried region.
12.3 Rugged energy landscape for water motion
Water molecules at the surface of the bilayers can exist in two states. They can be
strongly hydrogen-bonded to the charged or polar head groups or they can be in the
layer but not bonded. We again use the terms bound and free to denote these two
types of water molecules. Of course, no water molecule remains bound for ever and
there is a dynamic exchange between the two groups as mentioned earlier in
Chapter 6. Thus, what we essentially mean by bound water is a quasi-bound
water molecule, and bound is used to describe the instantaneous state of the
system.
It is hard to use thermodynamics to describe states of water molecules in hetero-
geneous environments such as the surface of lipid bilayers. Clearly, the bound water
molecules are stabilized by enthalpy while free molecules are stabilized by entropy
factors. One can construct a local free-energy surface that captures some of these
12.3 Rugged energy landscape for water motion 179
basic aspects of a lipid surface. The free-energy calculation is a bit complicated
because we need to include the contribution fromlarge uctuations in the position of
the lipid surface. Several lattice model calculations have been carried out that reveal
the existence of more than one phase transition in lipid bilayers, which further
reveals the complex free-energy landscape of the system [2].
12.4 Translational and rotational dynamics of water
It is clear that the motions (translational and rotational) of water molecules near a
lipid bilayer membrane are restricted. Nevertheless, they exhibit rich dynamic
behavior [3]. Much of the information has come recently from computer simula-
tions, which, as mentioned before, allow detailed follow-up of motion of individual
water molecules.
Rog and co-workers carried out a comprehensive study of a completely hydrated
1-palmitoyl-2-oleoyl-phosphatidylcholine (POPC) bilayer in the liquid-crystalline
state, and elucidated many general aspects of the nature of hydrated water at the lipid
surface. Their simulation study showed, not surprisingly, that the translational
motion of water near a lipid membrane is slower than in bulk water. However,
they found somewhat surprisingly that water molecules hydrogen-bonded to the
phosphate group move faster than those hydrogen-bonded with the carbonyl group
of the lipid. In order to characterize water molecules at the lipid surface, they
classied them into three different groups based on distance criteria [3].
(i) Water molecules within a distance of 4 from any membrane atom were
termed neighboring water (NW);
(ii) the water molecules that remained within a layer between 4 and 12 from
any membrane atom were characterized as intermediate water (IW); and
(iii) water molecules that were not closer than 12 to any membrane atom were in
the far water (FW) group. This is the same as bulk water according to our
earlier discussions.
A further classication was made on the basis of HB partner. Thus, the water
molecules hydrogen-bonded to phosphate oxygen are termed Op water, those to
carbonyl oxygen atoms are termed Oc water, and nally those clathrating choline
groups are termed Choline water.
Simulations were then used to calculate the mean-square displacement (MSD) of
these different types of water as a function of time, and from MSD Rog et al.
calculated the diffusion constants of different types of water. Figure 12.2 shows the
MSDs of different types of water as a function of time.
Molecular dynamics simulation and also spectroscopic evidence (discussed later)
suggest that the rotational dynamics of water around a lipid bilayer depends on the
degree of hydration of the lipid. As the degree of hydration decreases the bulk-like
motion of the water becomes less pronounced. The relative proportion of both
rapidly rotating and slowly rotating water increases [4].
As discussed earlier, the SD of an excited dye molecule (such as coumarin, Nile
Red) can provide useful information about the collective local dynamics of water. In
a series of investigations, Pal and co-workers studied the SD of 4-(dicyanomethy-
lene)-2-methyl-6-(p-dimethylaminostyryl)-4H-pyran (DCM) in dimyristoylpho-
sphatidylcholine (DMPC) vesicles in water by using the picosecond time-resolved
Stokes shift technique [5]. The experimental results are shown in Figure 12.3. They
found that the solvent relaxation time of the water molecules conned by the lipid
bilayer is 1 ns, which is much slower than the bulk water. The decay of the response
function was found to be bi-exponential, with two components of 230 ps (40%) and
1.6 ns (60%). These are long timescales, implying a slowmotion of the mediumand/
or of the water molecules. Unfortunately, the precise origin of such long times is not
clear yet.
Figure 12.2. Computed mean-square displacement (MSD) of (a, c, e) neighboring
(black line), intermediate (gray line), far (dotted line), and bulk (black-bold line)
water; (b, d, f) choline (black line), Op (dotted line), and Oc (gray line) water; (a, b)
in three dimensions, (c, d) in the membrane plane, and (e, f) along the bilayer
normal. Figure adapted with permission from Chem. Phys. Lett., 352 (2002), 323.
Copyright (2002) Elsevier.
12.6 Transport of small molecules across the bilayer
A lipid bilayer has to be always active as many molecules and ions are continuously
being transported across the layer. The layer is not too thin on a molecular scale,
almost 5 nm thick, which is much wider than the scale of most molecules (not
proteins though). The radius of a water molecule is only 0.14 nm and that of K
+
is
0.15 nm. The thickness of the layer is small on a macroscopic-length scale and
molecular details are important. That is, hydrodynamic arguments are not useful or
meaningful. In fact, transport of molecules across phospholipid bilayers has given
rise to a host of anomalies which are yet to be understood at a molecular level. We
list a few in the following.
(i) Hydrophilic solutes nd it harder to cross the bilayer than hydrophobic solutes.
This is attributed to the hydrophobic core of the lipid bilayer.
(ii) Ions have even lower permeability than neutral polar molecules, like dipolar
molecules. Among the ions, cations have a harder time crossing the bilayer
than anions.
(iii) Surprisingly, water has a rather large permeability, especially in comparison
with ions, for reasons yet to be fully understood.
Figure 12.3. Decay of solvation time correlation response function, C(t) of DCMin
1 mM DMPC lipid. The points denote the actual values of C(t) and the solid line
denotes the best t to a biexponential decay. The decay of the initial portion is
given in the inset. Figure adapted with permission from J Phys. Chem. B.,104
(2000), 4529. Copyright (2000) American Chemical Society.
(iv) Small neutral solutes seem to pass through the layer pretty fast, with a rate
much higher than that of water.
Some of the above apparent anomalies can be understood in the following fashion.
If
cross
is the time required by a solute molecule to cross the width, then we can
dene a diffusion constant for any given solute by
D
X
=
w
2
lipid
2
cross
(12:1)
where D
X
denotes diffusion coefcient of species X inside the lipid membrane. For
species of the same size, we then have the following series
D
NP
> D
water
> D
anion
> D
cation
(12:2)
We shall call this the lipid bilayer diffusion (LBD) series. In the above, D
NP
denotes
the diffusion coefcient of a non-polar molecule of the same size as water. The
above series is of importance because many of the species involved are of similar
sizes.
A physical explanation of the LBD series can be obtained if one considers the
free-energy barrier that a molecule of a given species needs to surmount to enter the
lipid bilayer. The free-energy barrier is determined largely by the polar character at
the lipidwater interface. In addition to negatively charged phosphate ions, we have
positively charged counterions which are mostly sodium ions. In such a charged
atmosphere, the least-stable species at the interface is the non-polar solute mole-
cule. Additionally, it would nd a free-energy minimum within the hydrophobic
environment of the bilayer. Thus, the free-energy barrier for entrance into the
hydrophobic layer can be modest.
In contrast, the negatively charged phosphate ions play a role in stabilizing the
cations at the interface. This stabilization energy is signicantly more than that of
anions, which are repelled by the phosphate ions. Thus, the anions face a lower free-
energy barrier than the cations. Anions nevertheless would acquire some degree of
stabilization from the positively charged counterions, mostly Na
+
, to maintain a
concentration at the interface. However, the self-diffusion coefcient itself will have
weak concentration-dependence. To summarize, cations should have a higher
entrance free-energy barrier and a lower diffusion coefcient than anions of the
same size.
Water molecules and other small non-charged polar molecules certainly would
nd a lower stabilizing atmosphere than the cations but higher than the non-polar
solutes of the same size. The non-polar solutes should face the lowest free-energy
barrier, followed by water molecules, when they are of the same size (such as water
and K
+
ions). The anions should face the largest barrier.
12.6 Transport of small molecules across the bilayer 183
The above argument seems to explain, albeit qualitatively, the above LBD series.
It is fair to assume that the interior of the lipid bilayer provides similar resistance to
all same-sized species as the free-energy surface should be at and should not
provide systematic force to retard motion signicantly differently from molecule to
molecule.
12.7 Transport of large molecules across the bilayer
The story is different for large solutes. Large molecules, particularly the hydrophilic
ones, nd it too difcult to pass through the lipid bilayer. To circumvent this difculty,
nature invokes a novel mechanism that keeps the solvation of these hydrophilic
moieties nearly intact but transports them to the inside of the cell, within a sack
formed by spontaneous uctuation, to be metaphorical. In reality, the molecule to be
transported is enclosed by a part of the membrane which then forms a vesicle that can
move through a part of the bilayer and then opens up inside the cell. Areverse process
is used to transport large molecules fromthe inside of the cell to the exterior of the cell.
The essence of such a transport is that the hydrophilic large molecules remain
solvated in water. The surface energy to create a vesicle by uctuation (which
involves a local dent in the layer and pinching off from the rest) is not too high.
Both are possible due to the abundance of water at the surface of the bilayer. The
vesicle formation is almost like the formation of a reverse micelle and the delivery
system is also quite simple.
12.8 Electrostatic potential across the membrane
Because of the presence of a large number of charge groups at both surfaces of the
phospholipid bilayer, there is a large electrostatic potential across the membrane.
Not only are the head groups of the phospholipids charged, there are alkali cations
(Na
+
, K
+
), Cl
, ATPase (which are negatively charged) close to the surface on both

sides. In addition, it now seems well established that charged (anionic) lipids
distribute asymmetrically across the native plasma membranes. This asymmetric
distribution is also believed to give rise to an electrostatic potential gradient along
the membrane surfaces, between the two leaets, which is local.
The potential arises from the difference in surface charges on the two sides of the
two leaets. A popular theory often employed is the GuoyChapman theory, which
is based on a continuum model description of the solvent and employs the Poisson
Boltzmann equation.
As the environment inside the lipid bilayer is hydrophobic, it has a low static
dielectric constant and therefore the electrostatic eld across the membrane does not
become screened as in bulk water.
12.9 Conclusion
As should be clear from the preceding discussions, lipid bilayers use the unique
characteristics of water for their biological functions. The charged interface between
the layer and the water outside plays a multitude of roles, including discrimination
between different chemical species and facilitating transport of chemicals to and
from the cell interior. Water molecules must also play an important role in the
functioning of ion channels and ion pumps that is yet to be understood in molecular
detail, although efforts are going on. The continuum solvent description for water
ignores the detailed role of water molecules, which could be interesting because
the length scales involved are not too great to justify a pure continuum model
description.
In the same vein, the variation of the diffusion coefcient of species across the
lipid membrane cannot be explained by employing hydrodynamic expressions, such
as the StokesEinstein relation. Here one would need to consider the free-energy
barrier for entrance into the layer for each species, charged (positive or negative) and
neutral; the free-energy barrier is expected to be different even for same-sized
species. The lipid bilayer diffusion series (LBDS) given by Eq. (12.2) is a manifes-
tation of such microscopic effects.
The role of water molecule as a lubricant of life is at its best in the lipid bilayers.
The large-scale motions needed for the functioning of the lipids such as the transport
of large molecules would not be possible without the small size and fast motion of
water molecules. Even when slow, water moves faster than all other species.
References
1. D. Marsh, Membrane water-penetration proles from spin labels. Eur Biophys J., 31
(2002), 559562.
2. Z. Zhang, J. Tobochnik, M. J. Zuckermann, and J. Silvius, Lattice model for hydrogen
bonding and hydration in pure lipid bilayers. Phys. Rev. E, 47 (1993), 37213729.
3. T. Rog, K. Murzyn, and M. Pasenkiewicz-Gierula, The dynamics of water at the
phospholipid bilayer surface: a molecular dynamics simulation study. Chem. Phys.
Lett., 352 (2002), 323.
4. K. J. Tielrooij, D. Paparo, L. Piatkowski, H. J. Bakker, and M. Bonn, Dielectric relaxa-
tion dynamics of water in model membranes probed by terahertz spectroscopy. Biophys.
J., 97 (2009), 2484.
5. S. K. Pal, D. Sukul, D. Mandal, and K. Bhattacharyya, Solvation dynamics of DCM in
lipid. J Phys. Chem. B., 104 (2000), 4529.
References 185
13
The role of water in biochemical selection
and protein synthesis
Richard Dawkins, in his celebrated work entitled The Blind
Watchmaker, observed that it is the blind force of nature that drives
natural selection and the evolutionary process as envisaged in Darwins
theory. While Darwins theory (generalized by Spencer) of the survival of
the ttest seems to work awlessly in the macroscopic living world, the
understanding of the same in the molecular world is incomplete, although
evolutionary ideas are routinely applied to explain the stability and occur-
rence of certain proteins and DNA. In this chapter we develop the idea that
in many biochemical processes in the cellular world, water molecules at
least partly control the outcome in the synthesis of proteins, eliminate
errors, and may even control enzymatic catalysis. The blind force of nature
is manifested and executed through water molecules.
13.1 Introduction
As discussed in the previous chapters, water plays a pivotal role in the structure,
dynamics, and function of biomolecules. They cannot function without water. The
role of water in the evolutionary process is another intriguing aspect that has drawn
some attention, particularly in the context of the prebiotic soup, which refers to the
aqueous solution of various constituents that ultimately form proteins, DNA, etc.
There are clearly two issues at hand here. The rst is the origin of life, which took
place in water. The second issue is the sustenance of life, which also needs water. In
the second context, the issue runs deep. Here water not only participates in biolo-
gical reactions and biosynthesis, but also eliminates errors and facilitates formation
for certain products. Thus, water plays the role of natural selector; the products
selected must go through a series of screenings by water.
In other words, the products formed in biosynthesis are pre-selected by water to
be able to survive in an aqueous environment.
In the following we discuss a few such processes.
187
13.2 Role of water in kinetic proofreading
The study of evolutionary trees seems to indicate that almost no natural selection
was invoked in the choice of proteins [1]. The selection appears to be random. One
thus tends to think the selection was guided by the stability of the protein and the
efciency in the active site region. A different view has emerged recently. The
synthesis of proteins after all went through a selection process, but one implemented
at the molecular level, and the tness was determined by none other than water. We
describe and develop this view briey here. The process of selection by elimination
of error in protein synthesis is known as kinetic proofreading (KPR), which is
applied generally to the selectivity of enzymes towards substrate absorption and
conversion to product. However, more specically, it applies to the avoidance of
error in protein synthesis. Here we shall rst discuss KPR from a general point of
view, with application to protein synthesis and DNA replication.
In biological synthesis, enzymes not only enhance the rate of reaction, but also
selectively choose the correct substrate leading to the desired product. Many
biochemical reactions, such as protein synthesis or DNA replication, exhibit high
specicity towards the selection of the correct substrates in the presence of many
other structurally or chemically analogous substrates. Due to the similar binding
energy of the substrate and the analog to the enzyme, the error rate (the ratio of the
rate of wrong product formation and that of the desired product formation) is
expected to be high. In contrast, the error rate is extremely low in the selection of
amino acids in protein synthesis (10
4
) [2] and DNA replication (10
9
) [3,4].
In 1972, Hopeld proposed a mechanism to explain the reason behind such high
selectivity of enzymes [5]. The mechanism has become widely known as kinetic
proofreading (KPR). At the same time Ninio independently proposed a similar
mechanism based on a different theoretical formulation [6]. Many experimental
studies have supported KPR, with fewexceptions that have surfaced in recent years.
Recently, Fersht proposed two mechanisms of KPR in tRNA-aminoacylation which
are likely to be equally valid based on the kinetic experimental results [7]. Fersht
found evidence that the most important discriminating step in KPR could be the
hydrolysis of a high-energy enzyme-substrate complex. Thus, a simple chemical
reaction with water plays a critical role! AMichealisMenten-like scheme has been
proposed recently where hydrolysis as a side reaction helps an enzyme to discrimi-
nate between two analogous substrates in spite of having similar binding energies.
The elementary steps of enzyme catalysis involve formation of the Michaelis
Menten complex, followed by product formation.
To illustrate the explicit role of water here we consider three biochemical
processes: (a) aminoacylation of tRNA in the presence of aminoacyl-tRNA
synthetase, (b) translation in the ribosome, and (c) DNA replication. In the case
188 The role of water in biochemical selection and protein synthesis
of tRNA-aminoacylation we consider isoleucyl-tRNA
Ile
synthetase (IRS) as an
example, where isoleucine is the correct and valine is considered as the wrong
substrate. This system has been extensively studied in several experiments and
there are many interesting ndings [710].
The structural analysis of IRS reveals that there are mainly two sites which are
involved in the aminoacylation reaction: the activation site that activates the amino
acid in the presence of ATP and the editing site (CP1 domain) 30 apart from the
active site that is responsible for deacylation of wrong products (see Figure 13.1) [11
14]. The role of the editingsite is tohydrolyze the wrong product, leadingto a lowering
of the error rate. The presence of this editing site makes IRS discriminate valine more
efciently than is observed in non-editing tRNA-aminoacylation enzymes [15].
We rst discuss the HopeldNinio KPR scheme, which is followed by a critical
analysis of the scheme in the light of the recent experimental results of
Figure 13.1. Crystal structure of Staphylococcus aureus isoleucinyl tRNA
synthetase (IRS) in complex with tRNA (Protein Data Bank code 1QU2(6)).
tRNA is shown in green, the CP1 domain in orange, the Rossmann fold in blue,
and rest of the protein is represented in silver. See plate section for color version.
aminoacylation of tRNA
Ile
in the presence of IRS. We then present a scheme
proposed by Fersht about the role of water and the editing domain on the selection
of isoleucine and valine.
13.2.1 Brief analysis of the HopeeldNinio approach
to kinetic proofreading
The well-known HopeldNinio scheme is usually given as the following sequence
of reactions between enzyme and substrate leading to the formation of the product
E + S
E + S
ES ES
*
k
l
c
k
m w
E + P
where the terms have their usual meaning. The letters on top of the arrows give the
respective rate constants. This, when applied to amino-acylation of tRNA, reads as:
E + ATP
E + aa.AMP
aa + E.ATP aa ( E.ATP ) ( aa.AMP ) E product
tRNA
k
k
m
w
l
c
The salient features of this reaction with regard to KPR can be summarized as
follows. The enzymesubstrate complex formation step involves a small amount
of discrepancy (f
0
), where f
0
is the ratio of wrong and correct product formed by
the enzyme. The rejection step also involves discrepancy by same amount (f
0
).
The ATP hydrolysis rate has to be irreversible so that maximum amount of
discrimination is obtained. In fact experimentally it has been proved that this
assumption is true. To obtain an error fraction smaller than 1/f
0
, more than one
high-energy intermediate needs to form. The amino acid transfer rate to the tRNA,
i.e., the last step, has to be very slow compared to the rejection step in the case of a
wrong substrate.
13.2.2 Analysis of experimental results in the light of the
HopeldNinio formulation
Although the isoleucyl-tRNA synthetase (IRS) from Escherichia coli does not
catalyze the overall mischarging of tRNA
Ile
with valine, it does undergo the rst
step of the reaction, the formation of an IRSValAMP complex. It suggests that the
discrimination is associated with a later step of the reaction path, indeed as proposed
by Hopeld.
The turnover number of the ATP pyrophosphatase reaction is almost the same
as the rate of transfer of isoleucine from IRSIleAMP to tRNA
Ile
over a wide
range of temperature and pH, indicating negligible hydrolysis of the intermediates
formed during the reaction, as is expected in accordance with the original KPR
theory.
We note here that mischarged ValtRNA
Ile
is hydrolyzed by IRS with a turnover
number of 10 s
1
at pH 7.78 and 25
C, compared with a value of 1.2 s

1
for the
transfer of isoleucine from IRSIleAMP to tRNA
Ile
or for the ATP pyrophos-
phatase reaction.
Nowwe discuss a deciency of the HopeldNinio scheme. Over a wide range of
conditions, the turnover number for the hydrolysis of ATP to AMP catalyzed by IRS
in the presence of valine and tRNA
Ile
is identical to that for the isoleucylation of
tRNA
Ile
catalyzed by the enzyme when the valine is replaced by [
14
C]Ile. This is
possible only when the discrimination through hydrolysis occurs after the rate-
limiting step. To the contrary, the KPR scheme proposed by Hopeld considers that
the discrimination step will be before the rate-limiting product formation step.
Experimentally, the amount of [
14
C]ValtRNA
Ile
formed is less than 0.8%. The
ratio of k
h
/k would have to be greater than 125 to account for this; i.e., k
h
must be
greater than 150 s
1
.
IRS:tRNA
ATP
IRS:tRNA:Val:AMP
k
IRS:Val tRNA
k
h
IRS Val tRNA
_
Val tRNA
Ile
SS
=
k
k
h
[IRS:tRNA:Val AMP[
This higher rate of hydrolysis is a unique feature of class I enzymes due to the
presence of an editing site other than the activation site.
Even if we consider the hydrolysis as the discrimination step in the KPR
mechanism, it alone cannot explain such an enhancement of discrimination and
rate of hydrolysis if we do not consider in the evaluation of discrimination the role of
the editing site.
The directly measured hydrolysis rate of ValtRNA
Ile
is only 10 s
1
! This is to be
compared with the required (by the Hopeld scheme) rate of 125 s
1
. Note also that
the rate constant for the hydrolysis of Ile-tRNA
Ile
catalyzed by IRS at saturating
concentrations is 0.014 s
1
at 25
C and pH 7.78. That is, the rate of hydrolysis of

valine is 70 times higher than that of Ile.
Both IRSIleAMP and IRSValAMP complexes hydrolyze with a rst-order
rate constant of 7 10
4
s
1
, indicating practically no editing at the stage of
enzymeaminoacylAMP formation. Moreover, these complexes are more stable
than the corresponding free aminoacyl adenylates, which hydrolyze with rates of
2.9 10
3
s
1
and 3.7 10
3
s
1
, respectively.
Whereas the addition of isoleucine (500 M) increases the hydrolysis rate of IRS
[
14
C]IleAMP to only 10 10
4
s
1
and the addition of isoleucine and ATP to
17 10
4
s
1
, the addition of isoleucine to IRS[
14
C]ValAMP causes 80% of the
ValAMP bound to the enzyme to be lost within a minute. The remainder hydro-
lyzes with a rate constant of 12.5 10
4
s
1
.
Based on the above results, Fersht proposed an alternative mechanism that
involved hydrolysis and is distinct from Hopelds mechanism. We next discuss
the scheme proposed by Fersht.
13.2.3 Aminoacylation of tRNA during protein synthesis
There are 20 aminoacyl-tRNA synthetases corresponding to 20 natural amino acids
that catalyze the aminoacylation reaction for a particular amino acid. Depending
upon the crystal structures and mode of binding to tRNAand ATP these enzymes are
classied into two major classes: class I and class II. All the class II enzymes and
non-editing class I enzymes follow the Hopeld mechanism as described above
where the dissociation of a high-energy enzymesubstrate complex involves hydro-
lysis. On the other hand, the mechanism of class I editing enzymes (with an extra
editing domain as shown in Figure 13.1) follows a more complex mechanism.
Below we describe the mechanism of aminoacylation and KPR for such enzymes
in detail.
As mentioned above, the experimental results do not seem to support the
picture of repeated activation as the mechanism of KPR. In an alternative scheme
to the HopeldNinio proposal, Fersht proposed that the repeated activation may
not be required if the critical step involves an enhanced rate of hydrolysis of the
wrong substrate. Fersht described two mechanisms as follows:
Mechanism I
IRS
150s
1
1.2s
1
IRS
*
.Val.AMP .tRNA IRS
*
.Val.tRNA IRS

+Val.tRNA
IRS

+Val+tRNA
Mechanism II
IRS
150s
1
10s
1
1.2s
1
Slow
IRS .Val.AMP .tRNA IRS .Val.tRNA IRS
*
.Val.AMP.tRNA IRS

+Val.tRNA
IRS + Val + AMP + tRNA
According to mechanism I, in the case of valine the enzyme goes to a highly
hydrolyzable state during the formation of the enzymesubstrate complex. After
the rate-determining step, i.e., the transfer of valine to tRNA
Ile
, the complex is
rapidly hydrolyzed, leading to the free enzyme and the original substrate. Thus the
hydrolysis increases the discrimination from 10 to ~180. In an alternative mechan-
ism (mechanism II) Fersht proposed that there is a possibility that the enzyme
undergoes a large conformational uctuation and goes to a highly hydrolyzable
conformation before the transfer of valine to tRNA
Ile
and the conformational
uctuation is the rate-limiting step. After the conformational uctuation it rapidly
hydrolyzes the ValAMPtRNA
Ile
complex.
Water again strikes as there is another hydrolysis step involved with the product
(ValtRNA) that is formed by mistake.
On the basis of the experimental result, Fersht concluded that both of these two
mechanisms are equally probable. However, it is possible to distinguish the validity
of these two mechanisms based on the experimental results that exist in the
literature. Let us discuss the drawbacks of these two mechanisms.
In both cases it is assumed that the enzyme undergoes a conformation change to a
highly hydrolyzable state. Structural analysis of IRS shows that it has an editing site
situated in the CP1 domain, which is ~30 apart from the active site, and the valine
is translocated to this site after transfer to the tRNA
Ile
and is hydrolyzed easily. It
suggests that the enzyme does not undergo any conformational uctuation: rather
there is a conformational site within the enzyme and the substrate goes to that site
and is hydrolyzed [11,1622]. As this translocation occurs after formation of the
valinetRNA
Ile
bond the hydrolysis step is involved after the formation of Val
tRNA
Ile
as in mechanism I. This point argues against mechanism II.
Now, according to mechanismI, the enzyme goes to the hydrolyzable state during
the formation of the IRSValAMPtRNA
Ile
complex and it gets hydrolyzed after
formation of the ValtRNA
Ile
bond. It is unlikely that the IRSValAMPtRNA
Ile
complex does not undergo hydrolysis in spite of the enzyme being in the hydrolyz-
able state. As we mentioned above, the substrate goes to the hydrolyzing site of the
enzyme after the transfer of valine to tRNA
Ile
. Thus, it cannot go to this conforma-
tion at the time of formation of the IRSValAMPtRNA
Ile
complex. In fact, none
of the above mechanisms take care of the hydrolysis properly. Moreover, no detailed
quantitative analysis yet exists in the literature to understand the contribution of the
enhanced hydrolysis to the overall discrimination and ATP hydrolysis.
Given the multitude of processes involved in aminoacylation of t-RNA, a quan-
titative understanding of KPR is not easy. The existence of many rate constants
makes accurate experimental elucidation of the proofreading mechanism also dif-
cult. Thus, the goal is not just to differentiate between Hopeld and Fersht schemes,
but also to rationalize the rate constants observed experimentally.
Advances in single-molecule spectroscopy have offered hope as one can obtain
the waiting time distribution, which contains more information than just the steady-
state rates. Also, there is the possibility that we can observe individual processes
separately and individually.
The enhanced hydrolysis of the wrong substrate is clearly a spontaneous and
natural process, particularly due to the abundance of water in biological cells. There
are still many critical steps involved in protein synthesis such as the KPR involved
in the selection of tRNA-aminoacylate in the ribosome and the selection of the
correct base pair in DNA replication. A difculty in the formulation of a correct
kinetic scheme for these problems is the non-availability of the rate constants for
several critical steps. Thus, even a correct scheme cannot be conrmed. Further
experimental studies of the individual kinetic steps are required.
Nevertheless, the current understanding emphasizes the role of water in protein
synthesis. It is fair to say that water acts as a primary material that selects, if not
controls, the sequence of the protein synthesized within biological cells.
13.2.4 tRNA selection in ribosome
After the charging of tRNAwith amino acid the charged tRNA (aa-tRNA) moves to
the ribosome, where the protein is synthesized through translation. Translation is the
key step of the entire protein synthesis process. The properly coded mRNA comes
out of the nucleus and sits at the ribosome. On the other hand, the charged tRNA
forms a ternary complex with the GTPase elongation factor-Tu (EF-Tu) and is
delivered to the aminoacyl (A) site of the ribosome. The aa-tRNA then moves to
the adjacent vacant peptidyl (P) site of the ribosome. Anewaa-tRNAcomes and sits
at the vacant A site and the amino acid attached to it forms a peptide bond to the
amino acid at the P site. The tRNA at the P site shifts to the exit (E) site and nally
leaves the ribosome. The peptide containing tRNA shifts from the A to the P site. A
newaa-tRNAcomes and sits at the Asite and in a similar way a newpeptide bond is
formed between the peptide chain and the newly arriving amino acid. This process
continues until the last amino acid is inserted into the peptide chain.
The selection of the correct aa-tRNA at the ribosome is vital since a wrong
selection leads to a wrong protein. If a wrong protein is synthesized it imposes a
huge impact on the cell in terms of diseases and survival of the cell. The common
consequence of wrong protein synthesis is that it goes to a mis-folded or completely
unfolded state, which can cause many diseases, for example Alzheimer disease. So,
it is extremely important for the ribosome to select the correct aa-tRNAand formthe
desired protein with the proper sequence of amino acids.
The selection of aa-tRNA at the A site of the ribosome is determined by the
specic interaction between the codon of mRNA and the anti-codon of tRNA. A
particular codon (a sequence of three nucleotide bases) specically chooses an
anticodon of tRNA through HB formation. Thus, the delity of protein synthesis
mainly depends on the specic selection of anticodon by a particular codon. The
typical HB energy of codon-anticodon triplets is ~5 kcal/mole and the expected
error fraction of wrong tRNA selection is ~0.2. In contrast, the observed error
fraction in living cells is observed to be ~10
4
! The obvious question is how
such a small error level is reached in spite of such a small energy difference in
codonanticodon triplet pairs for wrong and correct tRNAs. Kinetic proofreading
(the mechanism of error rectication) is successful in explaining such a small error.
Several uorescence and biochemical experiments reveal the detailed mechanism
of translation and the contribution of KPR towards the delity of protein synthesis.
The incorporation of an amino acid into the peptide is composed of two consecutive
processes: initial selection of tRNA at the A site of the ribosome followed by KPR
[23]. Various factors affect the initial selection of tRNA such as the HB energy
between the codonanticodon base pairs, the specic interactions between the large
subunit of the ribosome and aa-tRNA, etc. The contribution of the initial selection
step to the overall error fraction for Escherichia coli is observed to be ~1/6,
compared to the overall error fraction ~7 10
3
for cognate and near-cognate
anticodons. As a consequence the contribution of KPR is expected to be 1/24; i.e.
~80% of the observed delity comes from the KPR. The translation process occurs
through the following mechanism (see Figure 13.2).
At rst the EF-Tu-GTP-aa-tRNA complex comes to the A site of the ribosome.
Depending on the codonanticodon pairing and tRNAribosomal subunit interac-
tions, the tRNAis bound to the Asite. This is known as initial tRNAselection. Once
the tRNA is selected the GTP is activated and is hydrolyzed, followed by formation
of GDP and PPi. The hydrolysis of GTP makes the tRNAmore easily removed from
E P A
k
1
k
1
k
2
k
2
E P A
k
3
k
3
E P A
GTP
k
4
E P A
GTP
k
5
E P A
GTP GTP
k
6
E P A
GTP
k
7
E P A
GTP
E P
0 1 2 3 4 5 6 7
0.0 FRET 0.35 FRET 0.5 FRET 0.5 FRET 0.5 FRET 0.75 FRET 0.75 FRET
A
GTP
GTP
hydrolysis
G
T
P
k
1
Initial selection
k
6
Proofreading
Figure 13.2. Schematic diagram of the mechanism of translation and contribution
of KPR towards the delity of protein synthesis.
the ribosome for a near-cognate anticodon. Thus, water plays the key role in
enhancement of selectivity through hydrolysis.
13.2.5 DNA replication
The above analysis describes the role of hydrolysis in the natural selection of the
amino acid sequence in protein synthesis at the level of the translational process.
The other essential ingredient in the biochemical chain is DNA replication, by
which DNA is synthesized within the cell. This process consists of the following
critical steps:
(i) the double helix is opened by DNA helicase, leading to the formation of single
helical chain;
(ii) a new strand is formed around the single helices by adding the complementary
base pair sequence which forms the original strand by matching bases with the
help of DNA polymerase.
Natural selection of bases occurs at the second stage where KPR is implemented.
Here again water molecules play an important role by hydrolyzing the triphosphate
of wrong nucleoside bases, leading to the dissociation of the base pair.
13.3 Water as a lubricant of life
The term lubricant of life was perhaps rst used (in the sense being used here) by
Barron et al. [24]. These authors observed by Raman spectroscopy fast dynamic
events (in the picosecond timescale) where individual amino acid residues icker
between different secondary structure states. This ickering was attributed to fast
dynamic events in bulk water and mediated through HBs of the amino acid residues
with surrounding water molecules. These events driven by water could be the
guiding force in many functions of proteins (folding and unfolding, enzyme
kinetics, proteinDNA interaction), and hence water was termed the lubricant of
life.
As mentioned repeatedly, while we know a lot about the way in which proteins
fold, comparatively little is known about the detailed role that water plays in these
processes. The predominant interaction of water with proteins is through the
formation of HBs. As discussed in previous chapters, the structure of the water
surrounding a protein is continually changing, as HBs are broken and re-formed
at a very rapid rate. This leads protein secondary structures such as alpha-helices
and beta-sheets to inter-convert (or icker) among themselves on picosecond
(10
12
s) timescales. It is the same dynamics that is reected in SD at the protein
surface.
This ickering makes possible changes in protein structure that occur during
enzyme kinetics or even folding. The ever mobile water molecules reduce the
frictional forces that otherwise would retard the relevant motions of protein side-
chains. This effect has been called the lubricant of life.
Study of the low-frequency Raman spectrumof poly-l-lysine on both sides of the
helix-to-coil transition using ultrafast spectroscopy reveals the existence of a broad
band that can be assigned to HB stretching vibrations between the solvent and the
peptide. This band is found to shift depending upon the local structure of the peptide
and reects changes in the strength of the interactions between water and the
biopolymer. The frequency of the band shows that these dynamics occur on the
scale of 1530 ps.
Another important problem where water acts as a lubricant is the motion of
nonspecically bound protein along a DNAchain. Proteins travel to nd the specic
binding site. Such a motion is greatly facilitated by water molecules in the hydration
layer.
13.4 Conclusion
Biological systems are created in the presence and often with the help of
water molecules. And they need to be stable in the presence of water. Thus, the rules
of the game are dictated by water. The selection rules are essentially the rules of
compatibility with water. In this sense, water molecules are very much part of the
blind force of nature that has shaped life on our planet from the beginning.
We have already discussed in Chapter 6 the biological function of water. Here we
discussed a somewhat (although not completely) different issue. Water selects the
sequence of amino acids in proteins and of nucleic acids in DNA. This is truly an
amazing feat, even though not fully understood at present.
References
1. O. B. Ptitsyn and M. V. Volkenstein, Protein structures and neutral theory of evolution.
J. Biomol. Struct. Dyn., 4 (1986), 137156.
2. A. L. Lehninger, Biochemistry (New York: Worth, 1970).
3. J. D. Watson, Molecular Biology of the Gene, 2nd edn. (New York: W. A. Benjamin,
1970).
4. A. R. Fersht, Enzymic editing mechanisms and the genetic code. Proc. R. Soc. B, 212
(1981), 351379.
5. J. J. Hopeld, Kinetic proofreading: a new mechanismfor reducing errors in biosynthetic
processes requiring high specicity. Proc. Natl. Acad. Sci. USA, 71 (1974), 41354139.
6. J. Ninio, Kinetic amplication of enzyme discrimination. Biochimie, 57 (1975),
587595.
References 197
7. A. R. Fersht, Editing mechanisms in protein synthesis: rejection of valine by the
isoleucyl-tRNA synthetase. Biochemistry, 16:5 (1977), 10251030.
8. J. J. Hopeld, T. Yamane, V. Yue, and S. M. Coutts, Direct experimental evidence for
kinetic proofreading in amino acylation of tRNAIle. Proc. Natl. Acad. Sci. USA, 73
(1976), 11641168.
9. S. P. Hale and P. Schimmel, Protein synthesis editing by a DNA aptamer. Proc. Natl.
Acad. Sci. USA, 93 (1996), 27552758.
10. A. C. Bishop, K. Beebe, and P. R. Schimmel, Interstice mutations that block site-to-site
translocation of a misactivated amino acid bound to a class I tRNA synthetase. Proc.
Natl. Acad. Sci. USA, 100 (2003), 490494.
11. S. Fukai, O. Nureki, S. Sekine, et al., Structural basis for double-sieve discrimination of
L-valine from L-isoleucine and L-threonine by the complex of tRNAVal and Valyl-
tRNA synthetase. Cell, 103 (2000), 793803.
12. B. Zhu, P. Yao, M. Tan, G. Eriani, and E. Wang, tRNA-independent pretransfer editing
by class I leucyl-tRNA synthetase. J. Biol. Chem., 284 (2009), 34183424.
13. L. F. Silvian, J. Wang, and T. A. Steitz, Insights into editing from an Ile-tRNA synthe-
tase structure with tRNAIle and mupirocin. Science, 285 (1999), 10741077.
14. S. Cusack, A. Yaremchuk, and M. Tukalo, The 2 crystal structure of leucyl-tRNA
synthetase and its complex with a leucyl-adenylate analogue. EMBO J., 19 (2000),
23512361.
15. I. Gruic-Sovulj, N. Uter, T. Bullock, and J. J. Perona, Protein synthesis, post-translation
modication, and degradation. J. Biol. Chem., 280 (2005), 2397823986.
16. K. E. Splan, M. E. Ignatov, and K. Musier-Forsyth, Transfer RNAmodulates the editing
mechanism used by class II prolyl-tRNA synthetase. J. Biol. Chem., 283 (2008),
71287134.
17. I. Apostol, J. Levine, J. Lippincott, et al., Incorporation of norvaline at leucine positions
in recombinant human hemoglobin expressed in Escherichia coli. J. Biol. Chem., 272
(1997), 2898028988.
18. M. W. Zhao, B. Zhu, R. Hao, M. G. Xu, G. Eriani, and E. D. Wang, Leucyl-tRNA
synthetase from the ancestral bacterium Aquifex aeolicus contains relics of synthetase
evolution. EMBO J., 24 (2005), 14301439.
19. A. Fersht, Enzyme Structure and Mechanism, 2nd edn. (New York: W. H. Freeman,
1985).
20. G. Eriani, M. Delarue, O. Poch, J. Gangloff, and D. Moras, Partition of tRNA synthe-
tases into two classes based on mutually exclusive sets of sequence motifs. Nature, 347
(1990), 203206.
21. T. K. Nomanbhoy, T. L. Hendrickson, and P. Schimmel, Transfer RNA-dependent
translocation of misactivated amino acids to prevent errors in protein synthesis. Mol.
Cell, 4 (1999), 519528.
22. T. K. Nomanbhoy and P. R. Schimmel, Misactivated amino acids translocate at similar
rates across surface of a tRNA synthetase. Proc. Natl. Acad. Sci. USA, 97 (2000),
51195122.
23. S. C. Blanchard, R. L. Gonzalez Jr, H. D. Kim, S. Chu, and J. D. Puglisi, tRNAselection
and kinetic proofreading in translation. Nat. Struct. Mol. Biol., 11 (2004), 10081014.
24. L. D. Barron, L. Hecht, and G. Wilson, The lubricant of life: a proposal that solvent
water promotes extremely fast conformational uctuations in mobile heteropolypeptide
structure. Biochemistry, 36:43 (1997), 1314313147
Part III
Water in complex chemical systems
14
The hydrophilic effect
As the name suggests, a hydrophilic molecule (or an object) likes water
molecules and interacts with them strongly and directly. The interactions
are mainly electrostatic in nature as the hydrophilic objects are polar/
charged and water molecules can easily form HBs with such species. As a
result, the inuence of such objects on water structure and dynamics can
be signicant. Sometimes hydrophilic objects are small, such as alkali
cations or halide anions, but many times they are extended objects such
as silica or mica surfaces. Water near such extended hydrophilic objects
can be profoundly affected. In such a case the HB pattern gets distorted
near the surface as the water molecules form strong HBs with the polar
surface that frustrate its own HB network and this effect can propagate
well into the bulk. Of particular interest is the case when water is conned
between two hydrophilic surfaces. In addition, electrolyte solutions are of
course well known for their uses and properties. In this case the unique
ability of water molecules to form many isoenergetic structural arrange-
ments (or polymorphs) comes into play, as discussed below.
14.1 Introduction
In this chapter, we discuss why and how water structure changes near hydro-
philic objects. We follow this with a chapter on the other extreme: the hydro-
phobic effect. Familiar examples of hydrophilic objects are charged ions such as
Na
+
and Cl
. Solvation of such ions is important in many areas of chemistry and

biology.
Hydrophilic interaction is operative not only in solvating small rigid ions, but also
in stabilizing extended structures such as DNA, proteins, and inorganic extended
surfaces such as silica, mica, and zeolites. In the rst two examples (proteins and
DNA), both hydrophobic and hydrophilic interactions operate synergistically to
stabilize the structure. In the case of several common extended objects such as silica
and mica surfaces [1,2], it is primarily the hydrophilic interaction that dominates.
201
Here the surface atoms can form strong HBs with water molecules and exert
inuence on the extended HB network of water. Another important class of systems
is aqueous binary mixtures, where both hydrophobic and hydrophilic interactions
together determine many of the unusual properties exhibited by these systems. The
hydrophilic effect is partly responsible for water being such a good solvent for a
large number of polar molecules. Hydrophilic interaction also nds great use in
industry, as in hydrophilic chromatography.
Water conned between two hydrophilic objects forms an interesting and impor-
tant system with unique properties. Sometimes the amount of water molecules
conned between two hydrophilic surfaces is rather small, consisting only of a
few tens (or even fewer) of water layers. In such cases, the structural arrangement of
water molecules can become quite distorted compared with the bulk structure. Also,
the order imposed by each hydrophilic surface can propagate inwards and counter
each other. Waters ability to form many isomorphic structures is tested in such a
situation, as we discuss below.
In this chapter, we shall discuss different systems where hydrophilic interaction
plays important roles.
14.2 Water near ions
It has been observed that for not-too-small, singly charged ions, such as Cs
+
, K
+
, I
,
Cl
, NO
3
, NH
4
+
, SCN
, H
2
PO
4
, HSO
4
, HCO
3
, (CH
3
)
4
N
+
(tetramethylammo-
nium), and (NH
2
)
3
C
+
(guanidinium), have weaker ionwater interactions than
waterwater. When the size of an ion is comparable to that of a water molecule,
the hydrogen-bonding network of the surrounding water does not get signicantly
affected. In such a situation, the ions are minimally affected. In contrast, small, and
mostly multiply charged ions with high charge density such as, SO
4
2
, HPO
4
2
,
Mg
2+
, Ca
2+
, Li
+
, Na
+
, H
+
, OH
and HPO
4
2
, have stronger interactions with
surrounding water molecules and can interrupt the waterwater HB network. The
motion of these ions is also retarded.
As discussed, water easily solvates small positively charged alkaline (Li
+
, Na
+
,
K
+
) and negative halide (F
, Cl
, Br
, I
) ions. These are important phenomena

because of their common occurrence in nature as well as in chemistry and biology.
The free energy of solvation is large and negative as the entropic loss is more than
compensated by enthalpic gain due to the interaction between the charge and water
dipoles. Because oxygen is negatively charged and hydrogen is positively charged
and because the charge on the oxygen atom is about twice that on each hydrogen
atom, the effects of positively charged and negatively charged ions, even if they are
of the same size (such as K
+
and Cl
), can be quite different. In each case, however,

the structural arrangement of water molecules around ions differs from that in the
202 The hydrophilic effect
bulk (see Figure 14.1). Such hydrophilic interactions can lead to change (mostly
decrease) in entropy and lowering of enthalpy. Thus, unlike in the case of hydrophobic
interaction, here usually a large enthalpic gain can and does offset the entropic loss.
Depending on their relative abilities to induce structural changes in water, ions
have often been classied as structure-makers, termed kosmotropes, and structure-
breakers, termed chaotropes. These are of course qualitative or pictorial terms and
need to be quantied. As in other non-ideal solutions, this is done through the
concentration dependence of the viscosity of the aqueous solution, which in the case
of electrolytes can be described by the following expansion in concentration c,
c ( ) = c = 0 ( ) Ac
1=2
Bc (14:1)
where the concentration-independent constants A and B contain important informa-
tion about ionion and ionwater interactions. The coefcient A is the Debye
Waller coefcient determined by ionion interactions, screened by water. This does
not directly contain ionwater interaction effects. The coefcient B is called Jones
Dole coefcient and this is determined by ionwater interaction [3]. It is positive for
structure-makers (kosmotopes) and negative for structure-breakers (chaotopes). The
relationship of the sign with structure can be easily understood. Sometimes the value
of the coefcient B is very small, which implies that the ions do not perturb the water
structure signicantly. An example is KCl. Small monovalent ions and other
divalent ions give a positive B and are structure-makers. Large monovalent ions
give a negative B and are structure-breakers.
In some sense, therefore, small monovalent and divalent ions can be regarded as
hydrophilic while large monovalent ions are hydrophobic. For large monovalent ions,
the enthalpic gain is less than the entropic loss of free energy, as discussed below.
The concentration dependence of ionic conductivity needs to be understood from a
molecular viewpoint [4]. In this approach, a time correlation function representation of
viscosity derived long ago by Green and Kubo was used along with a molecular-level
description of the equilibrium correlation between the positions of ions and water
Figure 14.1. Aschematic illustration of the arrangement of water molecules around
a small ion. The gure is reproduced from http://www.porous-35.com/
electrochemistry-semiconductors-4.html
14.2 Water near ions 203
molecules. Such an analysis provides a microscopic expression for the JonesDole
coefcient B. However, a detailed quantitative study has yet to be carried out.
As indicated earlier, when the size of the monovalent ion becomes much larger
than the size of a water molecule, then the ion starts to interfere with the HBnetwork
of liquid water. In addition, the solvation energy due to ionsolvent interaction
decreases. According to Borns expression, this energy decreases with size r
ion
as
1/r
ion
. Thus, beyond a certain size, the entropic loss to the system due to the size of a
large ion becomes greater than the enthalpic stabilization due to the ionsolvent
interaction. Thus, the ion can behave as a hydrophobic solute.
14.3 Water near an extended hydrophilic surface
In nature, water is often found to occur in contact with interacting surfaces which are
larger than a water molecule in size. Examples include protein and DNA molecules,
silica and mica surfaces, and water in zeolite pores, to name a few. Recently studies
have been initiated to understand the nature of such water from a microscopic
perspective. We have already discussed proteins and DNA in detail in previous
chapters. In this section, we shall cover a few interesting non-biological susbstances.
By using computer simulations one can examine the structure and dynamics of
water conned between parallel silica surfaces [1]. The bulk structure of silica
contains tetrahedral silicate (SiO
4
) units. The surface contains exposed silanol
(-Si-OH) units which interact and form HBs with surface water (see Figure 14.2).
The effects of the difference in hydrophobicity/hydrophilicity on water structure
was captured in an elegant thought, or rather computer, experiment where the partial
charges of surface atoms were tuned from zero to the normal charge distribution.
When the partial charge is zero (q = 0), then the surface behaves like an apolar/
hydrophobic surface. On the other hand, when the charge distribution is normal
(q = 1), the surface is hydrophilic. In both scenarios, the water dynamics in the
adjacent layer slows down but for quite different reasons [5]. We will discuss this
important result in more detail.
Figure 14.2. Molecular representation of a silica surface. The hydroxyl (-OH) groups
are exposed in the surface and hydrogen-bond with the water molecules. This makes
the surface hydrophilic. The gure is reproduced from http://commons.wikimedia.
org/wiki/File:Schematic_silica_gel_surface.png?uselang=fr?uselang=fr.
In the case of the hydrophilic silica surface, it forms strong HBs with water
molecules and the local density of water becomes high.
This strong bonding and the high local density make the surface water rigid,
resulting in slow dynamics. In Figure 14.3 we show how the local density ((z))
changes as a function of distance from the surfaces, which reects the extent of
structural inhomogeneities introduced by connement. For the apolar surface,
however, water molecules form a more ordered hexagonal ice-like structure (see
Figure 14.4) which inuences the slow dynamics. Thus, while near a hydrophobic
surface, water forms an ice-like ordered structure which is of lowdensity, and which
becomes denser but disordered as the surface becomes hydrophilic. This is similar to
the analysis by Rossky et al. of protein hydration, as discussed earlier in Chapter 8.
It is important to note the difference that two conning surfaces makes to the
structural arrangement of water molecules, even pretty far into the liquid, such as
68 layers. Such a long-range effect of the surface is not seen when there is only one
hydrophilic surface. Clearly frustration is reinforced when the liquid is conned
between the two interacting layers.
Like silica, a mica surface also acts as a hydrophilic surface. Mica is a potassium
salt of alumino-silicate. If one replaces some of the Si atoms from silica by Al, then
because of the charge mis-match, the structure becomes negatively charged and the
extra negative charge is compensated by K
+
ions. Like the silica surface, here also
one nds that water molecules form hydrogen-bonding with the surface polar atoms
and some of the K
+
ions ll the interstitial position in the water structure. The
density of water also increases near the surface.
If one connes water molecules between the two mica surfaces, the water
molecules form a layer-by-layer structure from the surface to the center. It is intuitive
that the number of layers decreases with decreasing inter-surface separation and the
2.0
1.5
1.0
0.5
(
z
)

[
g

c
m
3
]
0.0
0.0 0.2 0.4
z [nm]
Bulk
k = 1.0
k = 0.6
k = 0.4
k = 0.0
d = 1.6 nm
0.6 0.8
Figure 14.3. Density prole, (z) (local density in 0.041-nm-thick slabs parallel to
the silica surfaces) for different surface polarities (k). T = 300 K, = 1.0 g cm
3
,
d = 1.6 nm. Figure adapted with permission from J. Phys. Chem. B., 113 (2009),
14.3 Water near an extended hydrophilic surface 205
layers become more and more ordered as a consequence of the connement. When the
two surfaces are close, with separation less than ve to six water diameters (approxi-
mately speaking), then there could be interference between the effects of the two
surfaces, leading to an increase/decrease of ordering in the layers.
However, one should note that the formation of such layered water molecules
needs a uniform surface structure, for example the mica surfaces discussed above.
Water can also absorb on the amorphous silica surface, which is not uniform. In such
a scenario water does not form a layer-by-layer structure, but a quite random
arrangement is found. Thus, not only the charge distribution of the hydrophilic
surface, but also the topology of the surface has signicant inuence on the structure
of water near the surface. The polymorphism of the water structure discussed in
Chapter 1 plays an important role in this context.
One can learn about the structure of a system from following its dynamics. The
orientational relaxation dynamics of water conned between mica surfaces has been
investigated by MD simulations. The presence of wide heterogeneity in the
dynamics of water adjacent to a strongly hydrophilic mica surface has been
observed. By analyzing the survival probabilities, a 10-fold increase in the survival
times for water that is directly in contact with the mica surface and a non-monotonic
variation in the survival times moving away from the mica surface to the bulk-like
k=0.0
k=1.0
Figure 14.4. Front snapshot of the rst layer of water molecules (layer thickness =
0.25 nm) at the interface with a hydrophobic apolar (k = 0.0, upper panel) and a
hydrophilic surface (k = 1.0, lower panel), showing the presence of hexagonal
structures (in yellow circles) on the apolar surface. T = 300 K. This result is taken
fromthe work by Castrilln et al. [1]. Figure adapted with permission fromJ. Phys.
Chem. B., 113 (2009), 1438. Copyright (2009) American Chemical Society. See
plate section for color version.
interior have been found. Moreover the orientational relaxation time becomes
longest for the connected water layer, and then decreases monotonically away
from the surface.
A pictorial representation of the effect of a hydrophilic surface on the orientation
of water molecules is depicted in Figure 14.5.
Note that the extensive HB network is compromised near both the hydrophilic
and the hydrophobic surfaces, but differently. In the case of the hydrophilic surface,
the enthalpic gain from the watersurface interaction compensates for the twin
losses of enthalpy and the entropy of water arising from the molecular rearrange-
ment imposed by the surface. However, for a hydrophobic surface, such a compen-
sation is not present. Therefore, the chemical potential of a water molecule near a
hydrophobic surface is higher than that in a bulk.
Another crucial issue is the distance to which the effect of the surface can be felt.
This is connected to the correlation length of the unperturbed liquid. The correlation
length increases with lowering temperature. Both density and orientational correla-
tion are involved.
14.4 Aqueous hydrophilic binary mixtures
Many aqueous binary mixtures owe their peculiar properties to the combined effects
of hydrophobic and hydrophilic interactions. Examples are abundant in chemistry.
Dimethyl sulfoxide, dioxane, ethanol and methanol, N-ethyl acetamide, many
hormones, steroids, and vitamin solutes exhibit this combined hydrophobichydro-
philic interaction in aqueous solution. Such hydrophilic-hydrophobic combined
effects are specically termed amphiphilic effects and are extensively exploited in
cosolvent chemistry. In Chapter 16 we extensively discuss the amphiphilic effects,
the corresponding properties, and applications with explicit examples. According to
the nature of an aqueous binary mixture they are often called non-ionic kosmotropes
or chaotropes. Let us take the example of urea (H
2
NCONH
2
).
Figure 14.5. Ordering of water molecules near a periodic hydrophilic surface.
14.4 Aqueous hydrophilic binary mixtures 207
14.4.1 Waterurea binary mixture
There are some small organic molecules with very simple chemical formulas that
have proved to be important for biological applications. Urea or carbamide is such
an organic compound, with the chemical formula (NH
2
)
2
CO. The molecule has two
amine (-NH
2
) groups joined by a carbonyl (C=O) functional group.
Urea serves an important role in the metabolism of nitrogen-containing com-
pounds by animals and is the main nitrogen-containing substance in the urine of
mammals. It is solid, colorless, and odorless (although the ammonia which it gives
off in the presence of water, including water vapor in the air, has a strong odor). It is
highly soluble in water and non-toxic. Dissolved in water it is neither acidic nor
alkaline. The body uses it in many processes, most notably nitrogen excretion. Urea
is widely used in fertilizers as a convenient source of nitrogen. Urea is also an
important raw material for the chemical industry.
Urea is an important cosolvent generally used in the study of denaturation
processes in several proteins. Inspired by the need to understand the microscopic
basis of the mechanism of protein denaturation, researchers have performed MD
simulations to assess the effects of urea on the hydrophobic interaction between two
methane molecules.
The urea molecules preferentially adsorb onto the charged hydrophilic residues
on the surface. This adsorption leads to a repulsion between the residues on the
surface of proteins and gives rise to a swelling of the protein, which exposes the
hydrophobic residues. The onset of water into the interior leads to a destabilization
of the native state resulting in denaturation. The outsidein action of urea in
denaturation also suggests that, in the presence of large amounts of denaturants, the
effective driving force for compact structure formation in proteins is decreased, as is
the hydrophobic interaction. The driving force has been argued to be a subtle
balance between hydrophobic interactions and interfacial free energies, both being
altered by urea. It also follows from this work that, because urea readily dissolves in
water without disruption of the water structure, one requires an excess amount of
urea (typically 68 M) before adsorption onto the surface residues of proteins
becomes effective [6].
There are some counterintuitive but interesting results obtained on the rate of
catalysis in aqueous urea solution. The rate increases at a very low concentration of
urea. This has been understood as follows: addition of urea can lead to denaturation
of the protein. However, to get a complete denaturation one needs to have a
sufcient concentration of urea. At low concentration, protein becomes partially
denaturated, making it more exible than in water while staying closer to the
native structure. Now, in most catalytic reactions, the change in conformation occurs
during the cycle. At a low concentration limit such cycling becomes faster as
exibility increases. Such a counterintuitive phenomenon is substantiated theoreti-
cally by Miyashita etal. in their cracking model of enzyme catalysis [7].
Many studies reveal that urea is either a poor kosmotrope or a chaotrope. However,
urea is a popular denaturant. The ureawater HBstrength is slightly weaker than water
to itself and this seems to increase the waterwater interactions favorably. It is preferred
over water for binding to the protein backbone, leading to protein denaturation.
14.4.2 Waterguanidinium hydrochloride binary mixture
Guanidinium hydrochloride (GdmCl), generally referred to as guanidine hydro-
chloride or guanidiniumchloride, is another interesting cosolvent. The hydration
of the counterion is important to the action of guanidinium with chloride ions.
Eventually they become weakly hydrated and that allows the water molecules to
easily rearrange around the protein surface and to interrupt the denaturation process.
However, some studies reported that in the case of GdmCl the interaction between
positively charged guanidinium ions with hydrophobic surfaces is the principal
driving force for denaturation [8]. Both urea and guanidiniummainly lead to protein
swelling and destruction by sliding between hydrophobic sites and consequently
dragging in hydrogen-bound water to complete the denaturation by dehydrating the
protein surface. However, the molecular mechanism underlying such cosolvent-
induced protein denaturation still needs more comprehensive understanding.
14.5 Aqueous salt solutions
Aqueous salt solutions are common in nature. The oceans are a vast reservoir of
NaCl, and our cells contain both NaCl and KCl salts, to name just two. These salts
are commonly called electrolytes. Electrolytes are divided into two types: (1) strong
electrolytes and (2) weak electrolytes. In aqueous solution, strong electrolytes
generally get fully dissociated into solvated cations and anions. Weak electrolytes
generally get partially dissociated while maintaining an equilibrium constant which
is characteristic of that particular salt. In such a situation, depending on the size of
the cation or anion and also the quantity of charge that it carries, it can have very
different mobility in aqueous solution. This can give rise to strong concentration-
dependence of the properties such as ionic conductivity, viscosity, etc., which can be
exploited in several applications.
14.5.1 Ionic conductivity
The conductivity (or specic conductance) of an electrolyte solution is a measure of
its ability to conduct electricity. Conductivity measurements are used routinely in
many industrial and environmental applications as a fast, inexpensive, and reliable
way of measuring the ionic content of a solution. For example, the measurement of
product conductivity is a typical way to monitor the performance of water-
purication systems. The conductivity of a solution containing one electrolyte
depends on the concentration of the electrolyte. Therefore, it is convenient to divide
the specic conductivity by concentration. This quotient is termed molar conduc-
tivity, and is denoted by
m
, which is dened as
m
= /c. Here is the specic
conductivity of the solution and c is the concentration of the electrolyte. As stated
earlier, strong electrolytes are believed to dissociate completely in solution. The
conductivity of a solution of a strong electrolyte at low concentration follows
Kohlrauschs law [9]:
L
m
= L
0
m
K

c
_
(14:2)
where L
0
m
is known as the limiting molar conductivity (limiting here means at the
limit of the innite dilution), and Kis an empirical constant. Moreover, Kohlrausch
also found that the limiting conductivity of anions and cations is additive; that is, the
conductivity of a solution of a salt is equal to the sum of conductivity contributions
from the constituent cations and anions.
A theoretical interpretation of these results was provided by the DebyeHckel
Onsager (DHO) equation [10],
L
m
= L
0
m
A BL
0
m
_ _
c
_
(14:3)
where A and B are constants that depend only on known quantities such as
temperature, the charges on the ions, and the dielectric constant and viscosity of
the solvent. As the name suggests, this is an extension of the DebyeHckel theory,
due to Onsager. As is widely known, DebyeHuckelOnsager theory of electrolyte
conductivity is limited to very low concentrations, less than 0.1 M solution. There
have been many attempts to extend the theory, and also understand the reason for the
breakdown. A useful approach to the problem utilizes the mode-coupling theory of
liquids that considers (i) charge density and (ii) current density as the two slow
variables. Charge density relaxation gives rise to the ion atmosphere effect while
current density relaxation gives rise to the electrophoretic effect.
In this approach, one employs the mode-coupling theory formalism to calculate
the friction of an ion. The resulting expressions involve chargecharge spatial
correlation functions that are concentration-dependent. In the limit of very low
concentration, the resulting expression for conductivity reduces to the Debye
HuckelOnsager expression. However, at higher concentrations the theory provides
a better description than the DHO limiting expression. This mode-coupling theory
approach appears to be valid upto 1 M solution.
There has been a steady interest in the development of understanding of the
concentration dependence of conductivity at high concentrations. It is well-
understood that the correlation between ions plays an important role in determining
concentration dependence. At high concentrations, we need the correlation at
smaller inter-ion separations. Due to the presence of water, it is non-trivial to
understand these correlations. This is still largely an unsolved problem.
Aweak electrolyte is one that is not fully dissociated. Typical weak electrolytes
are weak acids and weak bases. The concentration of ions in a solution of a weak
electrolyte is much less than the concentration of the electrolyte itself. For acids and
bases, the concentration of ions can be calculated when the value of the acid
dissociation constant is known. An explicit expression for the conductivity as a
function of concentration, c, known as Ostwalds dilution law, and is given by
1
L
m
=
1
L
0
m
L
m
c
L
0
m
_ _
2
K
a
(14:4)
where K
a
is the dissociation constant.
Both Kohlrauschs lawand the DebyeHckelOnsager equations break down as
the concentration of the electrolyte increases above a certain value. As already
mentioned, the reason for this breakdown is that as concentration increases the
average separation between cation and anion decreases, so that there is more inter-
ionic interaction.
14.5.2 Viscosity
Viscosity is an important transport property that determines the mobility of other
particles inside the uid. Studies of the viscosity of electrolyte solutions were
among the earliest experiments in the eld of solution chemistry and have strongly
inuenced the development of our understanding of the solvation processes. The
viscosity of an electrolyte solution is also of importance in engineering applications
and of research interest because the long-range electrostatic interactions presented
cause difculty in describing such systems. There has been a long history of
investigating the viscosity of electrolyte solutions. In 1905 Grneisen observed
experimentally that at very low concentrations the viscosity of electrolyte solutions
increased nonlinearly with concentration, regardless of the type of electrolyte
solution. This effect, named after him, is generally correlated as [11]
0
= 1 A

c
_
(14:5)
where and
0
are the viscosities of an electrolyte solution and pure solvent,
respectively, A is a positive constant, and c is the electrolyte molarity concentration.
Later, Falkenhagen and co-workers and Onsager and Fuoss established a method of
calculating parameter A starting from the DebyeHckel theory. However, the
above equation is only valid for concentrations up to about 0.01 mol/L.
According to the above equation the relative viscosity should always increase
with concentration. However, experiments show non-monotonic behavior for sev-
eral electrolytes such as most of the potassium halides, and several rubidium and
cesium halides [12].
In 1929 Jones and Dole proposed an empirical formula [13]
0
= 1 A

c
_
Bc (14:6)
for the viscosity of electrolyte solutions. A is related to inter-ion interaction and
the mobilities of solute ions. B, on the other hand, is the result of interactions
between solvent molecules and ions. The dominant effect is, in general, the latter
one. In Eq. (14.6), A is always positive but B may be either positive or negative
depending on the degree of solvent structuring introduced by the ions. Usually a
positive value of B is associated with structure-making (ordering) ions, whereas a
negative value of B is associated with structure-breaking (disordering) ions.
Given values of A and B, the JonesDole equation can reasonably well describe
experimentally observed viscosity behavior, but it is usually valid only for con-
centrations less than 0.1 mol/L.
There are several other empirical formulas that describe the concentration depen-
dence of the viscosity of an electrolyte solution. However, there exists no satisfac-
tory microscopic theory for the same. In particular, we do not yet fully understand
the concentration dependence of the viscosity and conductivity of KCl. This
remains an important lacuna.
14.6 Conclusion
Because of the enormous amount of work done in the last few decades, our under-
standing of both phenomena is now considerably advanced.
An especially interesting aspect important to biology and the whole of natural
science is the ordering of water molecules around molecules that contain both a
hydrophobic and a hydrophilic group in the same molecule. The ordering and
rearrangement of water molecules in such cases give rise to stable (often exotic)
structures that then perform a large number of chemical and biological functions.
This interaction remains a fascinating subject which has still kept a large number
of scientists engaged in the study of water.
In the case of water surrounding small hydrophilic molecules or extended
hydrophilic surfaces, the issues are again somewhat different from those in the
case of hydrophobic surfaces. In the present case of hydrophilic surfaces, the
orientational ordering of water molecules is quite different from that in the case of
hydrophobic surfaces (as seen from Figure 14.4).
Interestingly, water faces hydrophilic surfaces many times in a conned state,
such as in the major and minor grooves of DNA and reverse micelles, to give
popular examples. In such situations, the ability of water molecules to form many
different nearly isoenergetic structures (polymorphs) becomes really useful.
We have already discussed water in the grooves of DNAin Chapter 9 and we shall
discuss water in reverse micelles in Chapter 17.
References
1. S. R. V. Castrilln, N. Giovambattista, I. A. Aksay, and P. G. Debenedetti, Effect of
surface polarity on the structure and dynamics of water in nanoscale connement.
J. Phys. Chem. B., 113 (2009), 1438.
2. A. Malani and K. G. Ayappa, Relaxation and jump dynamics of water at the mica
interface. J. Chem. Phys. 136 (2012), 194701; A. Malani and K. G. Ayappa, Adsorption
isotherms of water on mica: redistribution and lm growth. J. Phys. Chem. B, 113
(2009), 1058.
3. B. Hribar, N. T. Southall, V. Vlachy, and K. A. Dill, How ions affect the structure of
water. J. Am. Chem. Soc., 124 (2002), 12302.
4. A. Chandra and B. Bagchi, Ionic contribution to the viscosity of dilute electrolyte
solutions: towards a microscopic theory. J. Chem. Phys., 113 (2000), 3226.
5. J. C. Rasaiah and R. M. Lynden-Bell, Computer simulation studies of the structure and
dynamics of ions and non-polar solutes in water. Phil. Trans. R. Soc. Lond. A, 359
(2001), 1545.
6. A. Wallqvist, D. G. Covell, and D. Thirumalai, Hydrophobic interactions in aqueous
urea solutions with implications for the mechanism of protein denaturation. J. Am.
Chem. Soc., 120 (1998), 427.
7. O. Miyashita, J. N. Onuchic, and P. G. Wolynes, Nonlinear elasticity, proteinquakes,
and the energy landscapes of functional transitions in proteins. Proc. Natl. Acad. Sci.
USA, 100 (2003), 12570.
8. P. E. Mason, G. W. Neilson, J. E. Enderby, et al., The structure of aqueous guanidinium
chloride solutions. J. Am. Chem. Soc., 126 (2004), 1146211470.
9. W. Wien, Obituary: Friedrich Kohlrausch. Annalen der Physik, 336:3 (1910), 449454.
10. M. R. Wright, An Introduction to Aqueous Electrolyte Solutions (New York: Wiley,
2007).
11. E. Gruneisen, Innere Reibung wasserigen Salzlosungen und ihren Zusammenhang mit
der Elektrolytischen Leitung. Wiss. Abh. phys. Reichanst., 4 (1905), 237.
12. V. M. M. Lobo, Handbook of Electrolyte Solutions: Part A & B (New York: Elsevier,
1989), p. 41.
13. G. Jones and M. Dole, The viscosity of aqueous solutions of strong electrolytes with
special reference to barium chloride. J. Am. Chem. Soc., 51 (1929), 2950.
References 213
15
The hydrophobic effect
Popular perception of the hydrophobic effect is exemplied in the oft-
quoted statement that oil and water do not mix. As we discuss here,
the hydrophobic effect is a manifestation of an effective attraction between
two non-polar foreign molecules or surfaces and is mediated through the
HB network of water molecules in solution. A related phenomenon, known
as hydrophobic hydration, describes the interaction between one non-polar
solute and the surrounding water molecules. The hydrophobic effect is
multifaceted and is a complex phenomenon, and is not fully understood yet.
The hydrophobic effect is temperature-dependent. It is entropic in a narrow
temperature range around room temperature (25 C), but becomes enthal-
pic at temperatures higher than 60C. The entropic origin arises from the
change in the HB pattern induced by the non-polar solute or surface that
frustrates the HBnetwork of water. The hydrophobic effect also depends on
the size of the non-polar solute. There seems to be a crossover in the
structure and properties of the water layer adjacent to the external solute
as the size is increased from small (comparable to the size of a water
molecule) to very large. The attractive force between two hydrophobic
surfaces is found to be long-ranged. Quantitative understanding of the
scope of this force and its applicability to biomolecular systems has
remained somewhat controversial.
15.1 Introduction
Walter Kauzmann was perhaps the rst to introduce the term hydrophobic bond-
ing to describe the observed tendency of oils in water to aggregate together to form
a separate entity. The term hydrophobic effect was popularized by Charles
Tanford through his inuential book The Hydrophobic Effect. This term literally
means water-fearing, and, as discussed below, arises from a naive, pictorial
(and wrong) explanation of the phenomenon in terms of an apparent repulsion
between water and hydrocarbons. The often-quoted example of hydrophobic effect
215
is the de-mixing of oil and water [14] and formation of near-spherical droplets on
lotus leaves. Nevertheless the oilwater interface has received a great deal of interest
in recent years. A striking example of the inuence of the altered structural
arrangement of water molecules near an oil-on-water surface is provided by an
anomalously high increase in the rate of a class of organic reactions. The free
dangling OH group emerging from the structural arrangement of the oilwater
phase boundary can enhance the rate of reaction by stabilizing the transition state
(TS) through the formation of stronger HBs to the TS than the reactant. It has been
found from experiments that for a particular reaction this rate enhancement can be
more than ve orders of magnitude [5].
The hydrophobic effect plays an important role in chemistry. It fosters the
formation of micelles and reverse micelles and many other structures and gives
rise to the unique solvation properties of aqueous binary mixtures (such as water
urea, waterDMSO, waterethanol, to name just a few). The hydrophobic effect is
also centrally important in biological systems. It is partly responsible for protein
folding, micellar aggregation, lipid bilayer formation, cell membrane formation, the
assembly of proteins into functional complexes, etc.
Very recently the adsorption and aggregation of a -amyloid fragment at the air/
water interface has been investigated by the combination of second harmonic
generation (SHG) spectroscopy, Brewster angle microscopy (BAM), and MD
simulation studies. It was found that in -amyloid the hydrophobic residue-rich
amino acid sequence 116 not only induces aggregation, but also exhibits a strong
preference for the airwater interface relative to the bulk. [6].
The hydrophobic effect is relatively easy to understand, at least semi-
quantitatively. It arises because water, at room temperature, makes sufciently
strong HBs among its molecules that are energetically favorable. So, water reorga-
nizes itself around a non-polar solute to maintain its HB network and this costs
entropy. This physical picture changes at higher temperature, as we discuss below.
At room temperature (around 25C), the enthalpy of solvation of a non-polar solute
Figure 15.1. A schematic representation of a reaction at an oilwater interface. The
transition state (AB
) is stabilized through the formation of strong HBs offered

from the free dangling OH groups. Here the transition state becomes more
stabilized than the reactants. Adapted with permission from J. Am. Chem. Soc.,
216 The hydrophobic effect
is usually negative, but the entropy cost overwhelms the enthalpic gain to make
solvation of such a solute thermodynamically unfavorable, resulting in very low
solubility.
Historically, the rst published observations pertaining to the hydrophobic effect
were made by Benjamin Franklin in 1891 when he poured oil into a pond and found
the oil to spread and make a thin layer of oil on the water. The same experiment was
repeated by Lord Rayleigh, who used it to determine, for the rst time, the size of a
molecule. Lord Rayleigh assumed that oil forms a monolayer on the water surface.
As he knew the volume of the oil poured and the area of the water surface, he knew
the size of each surfactant molecule of the oil. Both experiments used the logic that
oil does not mix with water.
The physical or theoretical interpretation of the hydrophobic effect also has an
interesting history and goes back far in time. In the early 1940s, Frank and Evans
proposed the famous iceberg model which assumed that water molecules form a
clathret-like cage surrounding the non-polar solute. That is, a dissolved solute
molecule modies the water structure in the direction of greater crystallinity.
The formation of such an open iceberg structure costs entropy, as already dis-
cussed above. However, this entropy-cost picture of Frank and Evans does not
capture the full picture, as described below.
Interestingly, nature seems to use a combination of hydrophobic and hydrophilic
interactions to performmany functions, such as enzyme kinetics, micelle formation,
transport of materials across biological cells, and the formation of the double-helix
structure of DNA, to name a few. Many chemically important molecules, such as
DMSO, phenol, ethanol, and dioxane, contain both hydrophobic and hydrophilic
groups and this combination is responsible for the unique properties of these
solvents. It is thus tting to study these two effects together.
15.2 Hydrophobic hydration
The observed hydrophobic effect is to be understood at two levels, as follows. First,
we need to understand the hydrophobic interaction between one isolated non-polar
solute molecule and the surrounding water molecules. Second, we need to under-
stand the interaction between two non-polar solute molecules, as a function of
distance mediated by intervening water molecules. Both are important and need to
be understood together. There are also independent attributes that need to be studied
separately.
The term hydrophobic hydration means the rst of the above, that is, the
interaction between one solute molecule and the surrounding water molecules.
Hydrophobic hydration can be quantied by measuring the free energy of transfer
of a non-polar molecule from its neat liquid state to water. It is straightforward to
15.2 Hydrophobic hydration 217
obtain the value of free energy of transfer by the following procedure. Let us
consider the equilibrium between the organic liquid and the aqueous solution
containing X
w
mole fraction of the non-polar solute. Since the chemical potential
of the solute must be the same in both the phases, we have the following relation
between the chemical potentials of the solute in the two phases
np
0
RT ln X
0
=
w
0
RT ln X
w
(15:1)
where
np
0
and
w
0
are the chemical potential of the non-polar solute in the pure
liquid and in water, respectively, and X
0
and X
w
are the mole fractions of the same in
the two cases. The above relation can be rearranged to obtain
np
0

w
0
= RT ln X
w
=X
0
( ) (15:2)
The left-hand side of the above expression gives us the change of chemical potential
on transferring one mole of the solute from its pure liquid state to water.
One nds that this change in chemical potential is large and negative for non-
polar, organic solutes. For example, the experimental results for normal alkanes can
be tted to the following simple form
np
0

w
0
= 2436 88n
C
(15:3)
where n
C
is the number of carbon atoms in the n-alkane chain. Here the chemical
potential is measured in cal/mol. The linear dependence on the number of carbon
atoms means that size matters in controlling hydrophobicity.
It is good to remember some numbers. For n-butane D is 5.9 kcal/mol. For
n-pentane, D is 6.86 kcal/mol, while for n-hexane it is 7.74 kcal/mol. For
cyclopentane and cyclohexane, the values are 6.0 and 6.73 kcal/mol. These are
large values, indicating the very low solubility of saturated hydrocarbons in water.
For aromatic hydrocarbons, the change in free energy is a bit smaller. For benzene
D = 4.62 kcal/mol and for tolune it is 5.43 kcal/mol. These are less hydrophobic
but nevertheless the values are sufciently large and negative to preclude any
signicant solubility.
Thermodynamic analysis shows clearly that at room temperature the bulk of the
free-energy cost of transfer of a non-polar solute to water is borne by entropy. For
example, for n-butane, the enthalpy change is H= 0.85 kcal/mol, while the entropy
change is S = 22.3 cal/mol/K. They combine to give the value = 5.9 kcal/mol
quoted above. The above values are at roomtemperature of 298 K. Also, note that the
enthalpy of solvation (that is, the enthalpy of transfer) of a non-polar solute is negative,
although small. Thus, the enthalpy change is indeed favorable.
The entropy of water in its own liquid state is ~17 cal/mol/K (at room tempera-
ture). Thus, insertion of each n-butane costs the solution about one molar entropy
equivalent of pure water. This is a formidable cost.
The relative contributions of enthalpy and entropy depend strongly on temperature,
and the picture changes with increase of temperature. We shall discuss this important
point in detail later (see below).
As we discussed above, the rst model that attempts to explain this entropy loss
was that of Frank and Evans, who proposed that water molecules in the rst layer of
the hydration shell form a cage-like structure by forming HBs around the non-polar
solute in a fence-like manner so as not to waste HBs by pointing them towards the
solute. This ordering clearly costs entropy. This iceberg model has sometimes been
taken too literally, for example in understanding the hydration shell of proteins. The
shell would certainly retain a certain dynamic character, as it would be in dynamic
equilibrium with the rest of the bulk. In fact, computer simulation studies indeed
show that water molecules around methane or ethane have a residence time of a few
tens of picoseconds at most, so the iceberg model indeed has a limited validity.
Although models have been proposed from time to time to explain the hydro-
phobicity of non-polar solutes, and they form interesting reading, we shall restrict
ourselves to the statistical mechanical model developed by Pratt and Chandler,
which we discuss later.
15.3 Temperature dependence of hydrophobicity: enthalpy
versus entropy stabilizations
The temperature dependence of hydrophobicity is interesting and has helped in
understanding the phenomenon. With an increase in temperature of the solution
from 0C, the solubility of a hydrophobic solute rst decreases, reaches a minimum
around 25C, and then continues to increase up to 100C. Thus, the hydrophobicity
becomes maximum near room temperature. This non-monotonous temperature
dependence of hydrophobicity might appear paradoxical and rightly so, as it has a
fairly deep physical origin. In Figure 15.2 we show the anomalous temperature
dependence of hydrophobicity observed in water.
The non-monotonic temperature dependence of hydrophobicity is a reection of
the fact that while hydrophobicity is entropic at room temperature, it is enthalpic at
high temperature when the HB network breaks down and the entropic cost sharply
decreases.
Another interesting thermodynamic effect associated with hydrophobicity is the
anomalous temperature dependence of the partial molar heat capacity of a hydro-
phobic solute [7]. The partial molar heat capacity is large and positive in contrast to
that of a non-hydrophobic solute.
The enthalpyentropy decomposition of the free energy of transfer of a non-polar
solute from its liquid to water is shown in Figure 15.3 [8,9]. The gure illustrates
schematically the strong temperature dependence of the entropy and enthalpy
15.3 Temperature dependence of hydrophobicity 219
of transfer. The dependence at the level of the free-energy change, however, is
rather weak.
It is important to assert the importance of Figure 15.3 in the general framework of
our understanding of hydrophobicity. This gure also shows that at room tempera-
ture the watersolute interaction is favorable, and the poor solubility of the hydro-
phobic solute is due to the entropic effect. At high temperatures, above 55C or so,
both the enthalpy and the entropy of solvation are unfavorable. Thus, this can be
considered a crossover temperature.
15.4 Hydropathy scale
In an attempt to quantify the degree of hydrophobicity, several authors, notably
Tanford, introduced a scale constructed on the basis of the free energy of transfer of
the solute from liquid hexane to liquid water. Although there is considerable
divergence among values given in different scales, these scales have proven parti-
cularly useful to categorize amino acid residues because these residues can have
both hydrophobic and hydrophilic groups. While some of the residues, such as
arginine and methionine, are hydrophilic because of the presence of charged groups,
alanine, isoleucine, leucine, and phenylalanine are all hydrophobic. We list one of
the scales in Table 15.1. In this scale, the hydrophobic amino acid residues are
positive while the hydrophilic residues are negative.
There are several other hydropathy scales (like the electro-negativity scales!).
Such scales have been used extensively to construct empirical force elds used in
the study of protein folding with the implicit solvent model (where solvent mole-
cules are not present at all).
280
I
n

(
s
o
l
u
b
i
l
i
t
y
)
300 320
generic case
hydrophobic
interaction
temp (in K)
Figure 15.2. Non-monotonic temperature dependence of solubility of a hydrophobic
solute in water. The dashed line shows schematically the expected behavior
observed in most solvents. Note that solubility is lowest near room temperature.
Adapted with permission from J. Am. Chem. Soc., 129 (2007), 54925502.
Such a characterization has been immensely useful because it allows us to realize
that the core of a protein is almost always lled with these hydrophobic residues,
hence the origin of the termhydrophobic core. Most of the hydrophilic amino acid
residues are usually found on the surface and they interact with the water molecules
to stabilize the protein.
15.5 Pair hydrophobicity and potential of mean force
between two hydrophobic solutes
While insertion of a non-polar solute into water is entropically unfavorable (at room
temperature) and can now be understood nearly quantitatively, the effective inter-
action between a pair of non-polar solute molecules is more difcult to understand.
This effective interaction is termed the potential of mean force, often just called
(a)
(b)
Figure 15.3. Temperature dependence of the enthalpy and entropy of transfer of a
typical hydrophobic solute from its own liquid to water. (a) The transfer of
neopentane from its pure phase to water, and (b) a regular solution: the transfer
of neopentane from the gas phase into a pure phase of neopentane. T
s
indicates the
temperature where the entropy of transfer is zero, T
h
indicates where the enthalpy
of transfer is zero. Adapted with permission from J. Phys. Chem. B, 106 (2002),
15.5 Pair hydrophobicity and potential of mean force 221
PMF. Understanding PMF is of great importance in biology and chemistry, in such
phenomena as aggregation, cluster formation, protein folding, and proteinDNA
interaction, to name a few. When large hydrophobic solutes are present in water,
then they introduce a distortion in the HB network around them. Therefore, two
hydrophobic solutes can interact with each other even when they are far apart, at the
scale of solvent molecular diameter. Such large-scale distortion of the water struc-
ture is not present for small solute molecules, such as methane, which can be easily
accommodated within the water structure.
As already mentioned, it is hard to obtain a quantitative measure of PMF between
two hydrophobic solutes, say for example, between two phenylalanine residues in a
protein. Theoretical studies have often modeled this process by studying the inter-
action between two spheres as a function of the distance between them, as shown in
Figure 15.4.
Table 15.1: Hydropathy index (or hydropathy scale) of different
amino acids. This scale determines how more hydrophobic a
particular amino acid is compared to others. The more positive is
the number, the more hydrophobic it is and vice versa. (Table 15.1
has been adapted with permission from J. Mol. Biol,. 157 (1982),
105132. Copyright (1982) Elsevier.)
Side-chain Hydropathy index
Isoleucine 4.5
Valine 4.2
Leucine 3.8
Phenylalanine 2.8
Cysteine/cystine 2.5
Methionine 1.9
Alanine 1.8
Glycine 0.4
Threonine 0.7
Tryptophan 0.9
Serine 0.8
Tyrosine 1.3
Proline 1.6
Histidine 3.2
Glutamic acid 3.5
Glutamine 3.5
Aspartic acid 35
Asparagine 3.5
Lysine 3.9
Arginine 4.5
In the same gure we also show the distance dependence of the effective
interaction energy between the two spheres. The results have been obtained by
computer simulations where the spheres are methane molecules. Note the
pronounced minimum between the two spheres at contact and then the max-
imum at intermediate distances. The interaction energy falls off to zero as the
two spheres move away. In some cases, one nds a second minimum at a
distance beyond the maximum [10]. Such a minimum at a larger separation is
referred to as a solvent separated pair and arises due to the structuring around
the hydrophobic spheres.
Figure 15.5 shows such an arrangement of water molecules around two methyl
molecules (represented as spheres) separated by a small distance. Note how the
water molecules move away from the methane molecules and form a ring to
maintain the HB network.
15.6 Biological applications of potential of mean force
Hydrophobic interaction between non-polar groups is involved in wide-ranging
phenomena in biology, starting from the association of proteins to proteinDNA
interaction and molecular recognition, to name a few. In the following we shall
discuss a few of them.
Figure 15.4. Potential of mean force (PMF) between two methane molecules in
water. This shows a rst deeper minimumcorresponding to the contact geometry of
the two methane molecules. Another second (less deep) minimum is also observed
in the PMF, corresponding to the solvent separated minimum. Adapted from thesis
entitled Molecular dynamics simulations of hydrophobic solutes in liquid water
by Andy Hsu, Institute of Atomic and Molecular Sciences, Academia Sinica.
(http://w3.iams.sinica.edu.tw/lab/jlli/thesis_andy/%5d.)
15.6.1 Protein folding
The rst case, as may be expected, involves the role of PMF in protein folding. Here
two hydrophobic amino acids need to come together to form the hydrophobic core. In
order to understand this process and also to investigate the process using computer
simulation, one needs to know the interactions between amino acids. While the
hydropathy scale provides a qualitative idea about this pair interaction, a more
quantitative estimate is required. Thus, the minimalistic implicit solvent models that
have been used for proteins along with a force eld based on the hydropathy scale and
helix propensities of the amino acids, although successful to some extent for small
proteins, fail to provide an accurate description for large proteins [11].
Recently a more detailed, orientation-dependent PMF the potential that depends
on the relative orientation of the two amino acid chains (see Figure 15.6) has been
proposed. This potential between any two amino acid residues was derived from the
statistical analysis of the experimental native structures deposited in the Protein
Data Bank (PDB). In this model potential amino acid side-chains are represented by
a single ellipsoidal of revolution. The PMF between two ellipsoidals of revolution
was obtained by considering all the interacting sites in them. The sitesite potentials
were then calculated from the statistics of their distance of separation obtained from
the crystal structures available in the PDB [11]. These sitesite potentials were then
summed up to obtain the distance and orientation-dependent potential between all
amino acid residues. The PMFs so obtained show many interesting features. Two of
the PMFs are shown in Figure 15.7.
Figure 15.5. Arrangement of water molecules around two methane molecules.
Water molecules around the methane molecules are oriented to maximize the
number of HBs among the water molecules.
The rst point to note is that the PMF obtained by this more elaborate scheme is in
general good agreement with the simple hydropathy-scale-based potential proposed
earlier. Second, this PMF (from PDB) correctly reproduces the effective attractive
interaction between two hydrophobic residues, such as phenylalanine, and effective
repulsive interaction between two hydrophilic residues, such as lysine, as shown in
Figure 15.7.
Orientation-dependent PMF gives valuable insight into the nature of the
orientation-dependent interaction between any two amino acid residues. The
orientation-dependent PMF also reveals many unexpected pair interactions which
defy the trend given by the hydropathy scale. An example is provided by the Arg
Arg pair interaction, which is found to be surprisingly attractive at short separation,
even though it is one of the most hydrophilic residues.
The reason was found to be the presence of an HBthat forms a bridge between the
two arginine residues. Such specic many-body effects cannot be captured in a
hydropathy scale.
As already noted, Figure 15.7(a) shows that the PMF between two phenylala-
nines is attractive at short distances. The repulsive potential at very short distance is
a reection of the electronic overlap. Since this potential is constructed from the
PDBby including data for a large number of proteins, the presence of many features
H
2
N
H
H
C
C
C
C
C
C H
O side chain
R
side residue
Principal Axis
r
ij
O
O
OH
N
H
N
H
Figure 15.6. Schematic representation of a minimalistic model of protein. Here
side-chains are represented by ellipsoids of revolution. Interaction between the two
side-chains is modeled by the distance distribution between them calculated using
the Protein Data Bank (PDB). Adapted with permission from Biochemistry, 45
(2006), 5129. Copyright (2006) American Chemical Society.
distance ()
distance ()
4 5 6 7 8 9 10 11 12 13 14 15 3
0
2
4
6
8
10
12
14
16
18
20
(a)
(b)
E
n
e
r
g
y

(
k
B
T
)
E
n
e
r
g
y

(
k
B
T
)
2 3 4 5 6 7 8 9 10 11 12 13 14 15
20
16
12
8
4
0
4
8
12
16
20
24
28
Figure 15.7. The interaction between the pair of side residues, each of which is
modeled as an ellipsoid, plotted as a function of distance for different orientations
of the ellipsoids: phenylalaninephenylalanine (hydrophobic residue: (a)) and
lysinelysine (hydrophilic residue: (b)). Note the attractive interaction at small
distances for the hydrophobic residue (Phe) and repulsive interaction for
hydrophilic residue (Lys). Arrows in the gure indicate the direction of the major
axis indicating the orientation of the side-chains. Adapted with permission from
Biochemistry, 45 (2006), 5129. Copyright (2006) American Chemical Society.
that are intuitively understandable but quantitatively not easily available is truly
satisfying.
Yet another interesting outcome of this recent study was the discovery of an
anomalous strong effective attractive interaction between two histidine residues and
between two tryptophan residues. This could be ascribed to the interaction mediated
by metals, as these are residues that form coordination complexes with metals such
as cobalt or iron.
15.6.2 Hydrophobic association
Hydrophobic interaction between non-polar groups is involved in wide-ranging
phenomena in biology, starting from the association of proteins to proteinDNA
interaction and molecular recognition, to name a few. This is essentially the same
process by which oils aggregate together in solution. This is closely related to the
pair hydrophobicity and PMF that we discussed earlier. An important point to
note is that pair hydrophobicity itself might not be enough in many cases.
However, no systematic study of higher-order hydrophobicity seems to have
been carried out.
15.6.3 Pattern formation in chiral molecules
An elegant example of hydrophobic force is provided by the role of hydrophobicity
in the pattern formation of chiral molecules. A study has been done by employing
a PMF developed by Ben-Amotz and Herschbach [12] using some of the ideas of
Bondi [13]. It has been pointed out how such an application can explain the stability
of a at interface in a racemic mixture but form a helical pattern in a solution made
of pure dextro or levo molecules.
15.7 Hydrophobic collapse of polymers
We have discussed the occurrence of hydrophobic collapse in the folding of
proteins. This is a general phenomenon and was treated quite early by Flory in
his well-known study of polymer conformations across theta temperatures in
polymer solutions. When a polymer chain is dissolved in water or the temperature
of the polymer solution is lowered, it often undergoes a transition from extended
state to collapsed state. This is reected in the sharp decrease in the size of the
polymer, measured by its radius of gyration. This phenomenon is known as
collapse transition. Though the collapse transition is primarily initiated by the
change of temperature, it may also be introduced by a change in the quality of the
solvent, such as changing the pH of the solvent, or adding a cosolvent, etc. In this
context, it should be noted that in a particular solvent, if the polymer chain
remains dissolved in the extended state, then the solvent is known as a good
solvent, whereas, if the polymer chain acquires the collapsed state, then the
solvent is called a bad solvent.
As mentioned earlier, a beautiful theory of polymer collapse in a solvent was
provided long ago by Paul Flory in a theory currently known as the FloryHuggins
theory. The present understanding of this phenomenon (including that of protein
folding) has been developed around this theory by modifying the same. We now
discuss the theory.
15.7.1 The FloryHuggins theory
In order to describe the collapse of a long-chain polymer in a poor solvent, Flory
developed a nice and simple theory in terms of entropy and enthalpy of a solution
of the polymer in water [14]. In order to obtain these two competing thermo-
dynamic functions, he employed a lattice model which can be justied by the
much larger size of the polymer than the solvent molecules. The polymer chains
are represented as random walks on a lattice, each site being occupied either by
one chain monomer or by a solvent molecule, as shown in Figure 15.8. The
fraction of sites occupied by monomers of the polymer can be denoted as ,
which is related to the concentration c, i.e., the number of monomers per cm
3
by
= ca
3
, where a
3
is the volume of the unit cell in the cubic lattice. Though the
lattice model is rather abstract, the essential features of the problem are largely
preserved here. This theory provides a convenient framework to describe solu-
tions of all concentrations.
The free energy F for this model has two components: an entropy term
describing the number of arrangements of the chains that can exist on the lattice
for a given , and an energy term describing the interactions between adjacent
molecules.
The entropy S of the solution can be obtained fromthe Boltzmann lawS = k
B
ln
where is the number of ways distinct solutesolvent congurations can be
generated on the lattice. This is obtained by using combinatorial analysis. The
nal form of the entropy has a surprisingly simple structure:
S[
site
=
N
ln
N
(1 )ln(1 ) (15:4)
The rst term is related to the translational entropy of the chain (
N
is the chain
concentration in dimensionless units, where N is the number of segments in the
polymer chain). The second term is similarly visualized as the translational entropy
of the solvent molecules (of volume fraction 1 ).
In fact, instead of considering the full entropy S(), it is more convenient to focus
on the entropy of mixing S
mix
. This is dened as the difference between S() and the
weighted average of the entropies of pure polymer [S(1)] and pure solvent [S(0)].
S
mix
() = S() S(1) (1 )S(0) (15:5)
The point to be noted here is that, due to this subtraction, all contributions to S()
which are independent of drop out of S
mix
.
For the particular formin Eqn. (15.5) the only change achieved by going fromS to
S
mix
is to eliminate the term

N
ln
1
N
, which is linear in all the non-linear terms
remain intact
S
mix
=
N
ln (1 ) ln (1 ) (15:6)
On the other hand, the energy term E contains, in general, three terms that describe
three different types of interactions present in a polymer solution
Monomermonomer interactions:
T
2

MM
2
Monomersolvent interactions: T
MS
(1 )
Solventsolvent interactions:
T
2

SS
(1 )
2
(15:7)
Figure 15.8. A schematic illustration of the mixing of a polymer chain in a solvent.
Each monomer unit (black) of the polymer or one solvent molecule (gray) occupies
one lattice site. The gure in general illustrates polymer conformation in solution.
However, all the three constants are not needed because all the terms in the
free energy per site which are independent of , or linear in , drop out when we
consider the energy change due to mixing, E
mix
. Thus, we write the nal expression
for E
mix
as,
1
T
E
mix=site
= (1 ) constant terms linear in (15:8)
with the following denition for
=
MS
1
2
(
MM

SS
) (15:9)
is called the Flory interaction parameter. This parameter is dimensionless, and
depends on the nature of the solvent and also temperature, pressure, and salt
concentration. Since the nature of the solvent itself depends on temperature and
pressure, is a complex quantity. Good solvents have a lowvalue of , whereas poor
solvents have a high value of . The case of = 0 corresponds to a solvent that is similar
to the monomer. In the case of this lattice model, the free energy comes entirely from
the entropy associated with various chain patterns on the lattice. In such a case,
temperature has no effect on structure, and the solvent is said to be athermal.
The overall free energy of mixing can now be obtained by adding the all
interaction parts
1
T
F
mix
[
site
=
N
ln(1 )ln(1 ) (1 ) mean field) ( (15:10)
Depending on the Flory parameter , there is a particular special temperature T =
at = 1/2, which corresponds to an exact cancellation between steric repulsion and
van der Waals attraction between monomers, and thus the chains are nearly ideal.
This temperature is known as the collapse temperature or theta temperature [15].
Equation (15.10) for the free energy of mixing is an expression that nds wide use in
physical chemistry. A quantitative understanding of hydrophobic collapse is
required to understand the initial stage of protein folding, as proteins are often a
nite chain consisting of a 50300 amino acid residue linear chain, which in many
aspects resembles a heteropolymer.
15.8 Molecular-level understanding of hydrophobic interaction
The difculty with the lattice models of hydrophobicity is that they do not do full
justice to the HB network of water. However, a molecular-level theory is also rather
difcult because the hydrophobic effect is a complex collective phenomenon
involving many water molecules. The hydrophobic solute perturbs the HB network
of water, resulting in a change of both entropy and enthalpy of the solutesolvent
system. Both the density and orientation of water molecules around a non-polar
solute can be perturbed. It is very hard to include in an analytical theory or
description the distortion in the relative orientation and packing of water molecules
around a non-polar solute.
As we discussed earlier, hydrophobicity is considered at two levels. First is the
hydration of a single non-polar solute and the second is pair hydrophobicity, where a
water-mediated interaction between two non-polar solutes is articulated. The former
is often referred to as hydrophobic hydration.
Understanding hydrophobic hydration requires an estimate of the chemical
potential of the non-polar solute in water. Actually, one measures the change in
chemical potential as the non-polar solute is transferred fromits own liquid to water.
This quantity is related to the hydropathy scale discussed earlier in the context of
protein folding.
A useful model, employed by Stillinger a long time ago, is to describe the
chemical potential in terms of the energy , W(), required to create a cavity of size
within the liquid [16]. Stillinger used the well-known scaled particle theory to nd
W(). It was noted that it is straightforward to nd W() when is less than the
diameter of the liquid (or solvent) being considered. When the cavity size is less
than the diameter of the solvent molecules, the cavity does not cause any signicant
distortion to the liquid, particularly if the liquid does not contain an extensive HB
network. However, in the case of water, the situation is a bit more complicated. Note
further that creating a cavity within the solvent is equivalent to placing a hard-sphere
solute as the cavity presents a hard surface to the solvent molecules.
While there is no general expression for the arbitrary value of the cavity radius ,
one can use macroscopic considerations to obtain the energy function W() for
much greater than the diameter of the solvent molecules. In this case the solvent
molecules near the solute see a hard wall. In the case of water, Stillinger showed that
a vapor-like very-low-density state of the solvent will be present near the surface
and the density will increase to bulk density as we move away from the wall. The
density prole may look like that of a gasliquid interface. Indeed the energy
function W() involves the vaporliquid surface tension term, and is given by [16]
W ( ) = 4pa
3
=3
_ _
3
4
a
2
_ _
2
16
a ( ) O 1 ( ) (15:11)
Here p is the pressure, a is solvent diameter,
is the surface tension in the planar

interface limit, and O(1) provides a correction for the curvature dependence of the
surface [16]. O(1) is also known as Tolmans length.
When is less than the diameter of the solvent, the energy is given by
W ( ) = kTln 1 4a
3
=3
_ _
3
_
0 _ _ 0:5 ( ) (15:12)
The above relations are for the cavity-creation energy function W() by assuming a
hard-sphere interaction between cavity and solvent molecules. Stillinger used the
above expressions to estimate the properties of the watervapor interface, but with
limited success. Nevertheless, W() for a hard-sphere cavity is a central quantity in
the description of hydrophobic hydration.
An elegant theory to go beyond the hard-sphere cavity was presented by
Pratt and Chandler [17], where the attractive part of the solutewater interac-
tion was treated perturbatively (in the spirit of the WeeksChandlerAndersen
(WCA) theory [18]). The central quantities in the PrattChandler theory are
two radial distribution functions, g
AW
(r) and g
AA
(r), that give, respectively, the
solutewater and the solutesolute two-particle correlation. [17]. In Appendix
15.A we describe the essential features of the PrattChandler theory, especially
the relations between the various pair-correlation functions needed in the
theory.
The PrattChandler theory provides an expression for the modication in the
chemical potential of the non-polar solute due to non-hard-sphere interaction
A
= [
(0)
A
[
HS
W
_
dr g
(0)
AW
(r)
_ _
HS
u
(1)
AW
(r) (15:13)
Here u
(1)
AW
(r) is the perturbation over the hard-sphere part of the potential,
u
AW
(r) = u
HS
(r) u
(1)
AW
(r). In the PrattChandler theory, u
AW
(r) considers the
interaction between the oxygen atom of water and the non-polar solute.
Application of Eq. (15.13) requires two quantities, (i) [
(0)
A
[
HS
and (ii) g
(0)
AW
(r)
_ _
HS
.
The former is obtained through a relation with the cavity pair distribution function,
[y
(0)
AA
[
HS
, between two non-polar solutes, assuming hard-sphere interaction among
them and also between solute and water molecules. We discuss the evaluation of
cavity distribution functions in Appendix 15.A.
In the PrattChandler theory, the distortion of the liquid structure and the work
done when two neutral hard-sphere solutes are brought from innity to a certain
distance gives a measure of the hydrophobic effect.
The pair correlation function g
(0)
AW
(r)
_ _
HS
is obtained through use of the Ornstein
Zernike equation and is briey described in Appendix 15.A. g
(0)
AW
(r)
_ _
HS
requires as
input the experimentally determined oxygenoxygen pair correlation function in
pure liquid water. The energy of the cavity formation is largely due to the decrease in
entropy of the water molecules at the surface of the cavity. However, it is hard to
obtain a quantitative estimate of the entropy change either from theory or from
experiment.
As pointed out, Stillinger also recognized the difference in the nature of hydration
between a small and a large non-polar solute and the existence of a vapor-like low-
density state near a large non-polar solute.
Physically, an external repulsive solute sphere that does not formany HBwith the
surrounding water molecules does not offer any enthalpic stabilization at all.
Therefore, the surrounding water molecules reorder themselves to preserve as
many HBs as possible. Therefore, the entropy of water decreases due to this
ordering forced on it by the solute.
Numerical evaluation of W() by using methods of statistical mechanics shows a
weak dependence at small but an exponential growth when is comparable to or
larger than the diameter of a water molecule [19].
The calculation of pair hydrophobicity is a bit more complex. This is dened by
the work done to bring two hydrated non-polar solute molecules from innite
separation (R ) to a separation R. When R is closer than the length of the
spatial correlations that exist in pure water, the solvents begin to reorganize further.
For each pair separation, R, the excluded volume D R; ( ) depends on R.
Following Stillinger, the reversible work (or potential of mean force),
W
2
R; ( ), required to bring the spheres from R to a distance R can be related
to the reversible work to create two spherical cavities separated by R,
W
2
R; ( ) = D
ex
R; ( ) D
ex
R ; ( ). An expression of reversible work for
pair hydrophobicity is provided by Chatterjee et al. [20]:
W
2
R; ( ) = kTln 1 8a
3
=3
_ _
3
_
R=a 2 _ 1 ( ) (15:14)
Note that W
2
R; ( ) is same as W ( ) for hydrophobic hydration except for a factor
of 2.
It is observed that W
2
R; ( ) decreases as the solute molecules come closer and
exhibits a minimum when the solute spheres are in contact. The change in entropy
plays an important role in pair hydrophobicity. When the solute spheres are distantly
separated, each of them orders neighboring water molecules around them. As the
pairs approach each other, the number of ordered water molecules decreases,
causing an increase of entropy. Below a certain distance the water molecules in
between two solute spheres come out and the spheres collapse.
The PrattChandler theory has been extended to consider complex molecules.
For example, the hard-sphere model of n-butane may have an excluded volume
D ; ( ), which is a function of the torsion angle and depends on the exclusion
radius of the methylene spheres. Then the part of the PMF (the potential of mean
force) arising from the solutesolvent interaction can be related to the reversible
work required to create a cavity with the shape and excluded volume D ; ( ) of the
n-butane molecule.
15.9 Hydrophobic force law
Experiments have repeatedly observed the presence of a long-range unexpectedly
strong attractive force between two hydrophobic surfaces suspended in bulk water.
Experiments show that this attractive force is present even at a separation of 200 .
The force increases almost exponentially down to 1020 . This hydrophobic force
between two hydrophobic surfaces has been termed hydrophobic force law.
Despite a large amount of work, the precise origin of the hydrophobic force has
remained unclear and is a subject of lively debate. Below we describe a theoretical
work which offers an explanation, although quantitative agreement appears to be
lacking.
The difculty in understanding the hydrophobic force law at such large distances
is the absence of any reliable theory to describe the propagation of joint density and
orientational correlation among water molecules. It is possible that orientational
correlation is longer-ranged than spatial correlation.
15.10 Hydrophobicity at different length scales
Let us nowconsider another interesting aspect of hydrophobic interaction. Consider
one hydrophobic object in water. It does not form HBs with the surrounding water
molecules. Thus it creates an excluded-volume region around it (as discussed above
in the context of Stillingers scaled particle theory) where the density of water nearly
vanishes. When these objects are small enough, water can reorganize near and
around them without sacricing too many HBs. The entropic cost of this structural
change leads to lowsolubility for small apolar species in water and also an attractive
PMF among a pair of non-polar solutes.
There is, however, no strong inducement for a small number of small hydrophobic
groups to associate in water. It is more likely that water can separate such species rather
than drive themtogether. Small-length-scale hydrophobic interaction can be understood
in terms of Stillingers scaled particle theory and the integral equation theory of Pratt
and Chandler. Of course, association/coagulation of non-polar solutes occurs when
concentration of the solute is increased beyond a critical concentration. The critical
micellar concentration is an example of such an association.
On the other hand, close to a large hydrophobic object, the persistence of an HB
network is geometrically impossible. Both the entropic and enthalpic cost may
become unfavorable and the resulting energetic effect can induce drying.
Furthermore, this drying can lead to strong attractions between two large hydro-
phobic objects at a distance much greater than the diameter of an individual water
molecule. This feature has been termed hydrophobicity at long length scales [21].
This has certain similarities with the hydrophobic force law discussed above.
The situation is depicted in Figure 15.9. A small solute is accommodated within
the water structure but the loss of HBs near the two extended hydrophobic
surfaces causes water to move away from those surfaces, producing a thin vapor
layer next to them. Fluctuations in the interfaces formed in this way can destabi-
lize and expel the remaining liquid contained between two such surfaces when
they are brought close to each other. The resulting pressure imbalance can cause
the surfaces to attract. In particular, if the liquid is close to the coexistence line
with the vapor phase, this attraction between surfaces can occur even when the
surfaces are widely separated. This kind of large-length-scale hydrophobic inter-
action is relevant to the solvation of macromolecules such as proteins.
Researchers have showed that the crossover between these two scenarios occurs
on nanometer length scales.
15.11 Conclusion
The hydrophobic effect is important in both biology and chemistry and is a clear
manifestation of the extensive HB network present in water. At room temperature
this effect is essentially entropic in origin. However, the enthalpic stabilizing
Figure 15.9. (a) Schematic view of local water structure near a small hydrophobic
sphere. The dashed lines indicate HBs. (b) Schematic view of water structure near
large parallel hydrophobic plates. The shaded area indicates regions where water
density is essentially that of the bulk liquid; vacant regions indicate where water
density is essentially that of the bulk vapor. Adapted with permission from J. Phys.
contribution increases with temperature, while the entropic destabilizing force
decreases, towards solvation of non-polar solutes. Although it was introduced at a
phenomenological level, the molecular-level understanding of the hydrophobicity
of individual non-polar molecules is now fairly well established.
However, the hydrophobic force between two non-polar molecules, often termed
pair hydrophobicity, is yet to be fully understood. Also, the origin of an unusually
strong hydrophobic force between two hydrophobic surfaces (with water in
between), at distances 100150 apart, has not been explained.
In the next chapter we discuss the water structure and dynamics around amphi-
philic molecules, where one part of the molecule attracts while another repels.
APPENDI X 15. A PRATTCHANDLER THEORY
The celebrated PrattChandler (PC) theory is usually the starting point of any
discussion on the hydrophobic effect. This theory can be regarded as an application
of the WeeksChandlerAndersen (WCA) perturbative theory of liquids to the
solvation of one and a pair of non-polar solute molecules. While Stillinger discussed
the chemical potential involved in creating a hard-sphere cavity in water using the
scaled particle theory, the PrattChandler theory used an integral equation descrip-
tion and showed how to properly discuss the effect within a general statistical
mechanical theory.
We discussed in section 15.8 the essential ingredient of the PC theory. In
particular, we discussed how Eq. (15.13) can be used to obtain the chemical
potential of a non-polar solute in water. This expression requires the solutesolvent
pair correlation function g
(0)
AW
(r)
_ _
HS
. PC theory shows how to obtain this function
and also the solutesolute pair correlation function. However, there is no closed-
form expression for these quantities and they need to be obtained by solving a set of
integral equations self-consistently, as discussed below. The theory is a bit involved
but quite rewarding in understanding.
In order to investigate both the water structure in the vicinity of a non-polar solute
(hydrophobic hydration) and the correlation between two solute molecules in water
(pair hydrophobicity), one calculates the change in local density of water molecules
at r due to the introduction of a solute molecule at the origin. According to the linear
response theory, the response (that is, the change) of the densities is dependent on
the structure of water in the absence of the disturbance. As a result, the equations for
solute-pair correlations become dependent on the waterwater correlation functions
for pure water.
The PC theory uses integral equations of the liquid state to obtain the pair
correlation functions associated with spherical non-polar solutes in water. It
requires as input the experimentally determined oxygenoxygen pair correlation
function for pure liquid water. As mentioned already, Stillinger had earlier used
the scaled particle theory for similar systems that use pure water properties as
input.
15.A.1 Cavity distribution functions
As discussed earlier, the quantity [
(0)
A
[
HS
is obtained through the cavity pair
correlation function, [y
(0)
AA
[
HS
, by [
(0)
A
[
HS
= k
B
T ln [y
(0)
AA
(0)[
HS
. Here [y
(0)
AA
(0)[
HS
is a
non-trivial quantity that describes the correlation between two cavities at the origin.
Pratt and Chandler described a scheme for calculating the hard sphere AA and AW
cavity distributions, which we discuss next.
There are three different types of interactions in a solution of a non-polar solute
(we label a spherical non-polar solute as A) and water (labeled as W) namely, the
AA, AW, and WW interactions. In their semi-empirical approach Pratt and
Chandler did not make any approximation for the waterwater (WW) interaction.
As we discuss here, the required pair correlation function among water molecules is
obtained from experiments using the oxygenoxygen correlation function of pure
water. However, they have considered a LennardJones potential for the AA
interaction.
u
AA
r ( ) = 4
A

A
=r ( )
12

A
=r ( )
6
_ _
(15:A:1)
where
A
and
A
are the generally accepted LennardJones parameters for molecules
of type A. Further, the AW interactions are modeled as a radial potential,
u
AW
r ( ) =
AW
r=
AW
( ) (15:A:2)
where f(x) is a dimensionless function of dimensionless variable x. Note that if we
take the LennardJones form, then x ( ) = 4 x
12
x
6
( ).
Both u
AA
r ( ) and u
AW
r ( ) have attractive and repulsive parts. Using the WCA
model, both these interaction potentials can be separated into a short-range repulsive
portion u
0 ( )
MM
/
r ( ) and a longer-range attractive portion u
1 ( )
MM
/
r ( ):
u
0 ( )
MM
/
r ( ) =
MM
/ u
MM
/ r ( ); r5r
0 ( )
MM
/
;
= 0; r > r
0 ( )
MM
/
;
u
1 ( )
MM
/
r ( ) =
MM
/ ; r5r
0 ( )
MM
/
;
= u
MM
/ r ( ); r > r
0 ( )
MM
/
(15:A:3)
15.A.1 Cavity distribution functions 237
Here r
0 ( )
MM
/
is the position of the minimumin the potential of u
MM
/ . Note that for ease
of representation we have used the labels M and M
to identify A and W (or we

would have had to write two separate equations for u
AA
r ( ) and u
AW
r ( )).
For most liquids, the short-ranged repulsion is the only interaction that is a
quickly varying function of the interparticle separation. Since particles are very
close together at liquid densities, the energetics and consequently the interparticle
structural changes involved in displacing the particles are dominated by the quickly
varying repulsive forces. Hence for most liquids, we can write the radial distribution
function as g
MM
/ r ( ) - g
0 ( )
MM
/
r ( ), where g
0 ( )
MM
/
r ( ) is the distribution function for the
hypothetical uid in which u
MM
/ r ( ) = u
0 ( )
MM
/
r ( ) (which means u
1 ( )
MM
/
r ( ) = 0).
However, liquid water cannot be modeled this way since the interactions in water
are far too complex, and there has been no satisfactory theoretical model for water
interactions yet. Pratt and Chandler bypassed the problem by considering that the
AA and AW interactions can be approximated by short-range interactions,
whereas the WW interaction is obtained from experiment.
g
AA
r; u
AA
; u
AW
; u
WW
( ) - g
AA
r; u
0 ( )
AA
; u
0 ( )
AW
; u
WW
_ _
= g
0 ( )
AA
r ( )
g
AW
r; u
AA
; u
AW
; u
WW
( ) - g
AW
r; u
0 ( )
AA
; u
0 ( )
AW
; u
WW
_ _
= g
0 ( )
AW
r ( )
(15:A:4)
The reference system, labeled with superscript 0 in the above equation, is the uid
in which attractions are ignored for AA and AW interactions, but interactions
between water molecules are treated exactly. Hence, g
0 ( )
WW
r ( ) is the oxygenoxygen
radial distribution function between two real water molecules when the hypothetical
A particles (hypothetical, because their long-range interactions are ignored) are
dissolved in water. Again, the solubility of non-polar solutes is very low in water,
hence g
0 ( )
WW
r ( ) is approximated as the oxygenoxygen radial distribution function of
pure water. The main contribution of the PC theory is the evaluation of g
0 ( )
AA
r ( ) and
g
0 ( )
AW
r ( ). One then uses Eq. (15.13) to obtain the chemical potential of the non-polar
solute in water.
Now the problem reduces to the description of repulsive spherical particles in
liquid water. We can get closer to the solution by recognizing that the problem is
closely related to the statistical mechanics of hard spheres dissolved in water. One
can now use the cavity distribution function dened by
y
AA
r ( ) = g
AA
r ( )exp
u
AA
r ( )
k
B
T
_ _
(15:A:5)
to approximate the two radial distribution functions as
g
0 ( )
AA
r ( ) = y
0 ( )
AA
r ( )
_ _
HS
exp
u
0 ( )
AA
r ( )
k
B
T
_ _
g
0 ( )
AW
r ( ) = y
0 ( )
AW
r ( )
_ _
HS
exp
u
0 ( )
AW
r ( )
k
B
T
_ _ (15:A:6)
where y
0 ( )
AA
r ( )
_ _
HS
and y
0 ( )
AW
r ( )
_ _
HS
are reference hard-sphere cavity distributions.
Note that we have already approximated that g
AA
r ( ) - g
0 ( )
AA
r ( ) and g
AW
r ( ) = g
0 ( )
AW
r ( ).
We may now combine Eqs. (15.A.3) and (15.A.4) to nd
y
AA
r ( ) - y
0 ( )
AA
r ( )exp
u
1 ( )
AA
r ( )
k
B
T
_ _
(15:A:7)
According to Eq. (15.A.6), y
0 ( )
AA
r ( ) - y
0 ( )
AA
r ( )
_ _
HS
, so that Eq. (15.A.7) now
becomes
y
AA
r ( ) - y
0 ( )
AA
r ( )
_ _
HS
exp
u
1 ( )
AA
r ( )
k
B
T
_ _
(15:A:8)
Similarly, one can nd
y
AW
r ( ) - y
0 ( )
AW
r ( )
_ _
HS
exp
u
1 ( )
AW
r ( )
k
B
T
_ _
(15:A:9)
Thus, the difference between the cavity distributions y
AA
r ( ), y
AW
r ( ), and those
appropriate to hard spheres dissolved in water is a simple mean eld Boltzmann
factor.
15.A.2 Theory for AW and AA pair correlations
In order to describe hydrophobicity at innite dilution, we need two pair correlation
functions dened as
h
AW
r ( ) = g
0 ( )
AW
r ( )
_ _
HS
1
h
AA
r ( ) = g
0 ( )
AA
r ( )
_ _
HS
1
(15:A:10)
15.A.2 Theory for AWand AA pair correlations 239
The subscripts A and W denote the non-polar solute and water, respectively. In the
above equations, the ideal gas contributions have been subtracted from the radial
distribution functions to dene the radial pair correlation functions. Next, we dene
the direct correlation functions c
AW
r ( ) and c
AA
r ( ) through the following Ornstein
Zernike-like equations:
h
AW
r ( ) = c
AW
r ( )
W
_
dr
/
c
AW
[r r
/
[ ( )h
WW
r
/
( )
h
AA
r ( ) = c
AA
r ( )
W
_
dr
/
c
AW
[r r
/
[ ( )h
WA
r
/
( )
(15:A:11)
where h
WW
r ( ) = g
0 ( )
WW
r ( ) 1 is the radial pair correlation function for pure liquid
water.
Due to the convolution nature of the integral in the above equations, it is
convenient to take a Fourier transform because the Fourier transform of the con-
volution of two functions is the product of the Fourier transform of the individual
functions. The Fourier transforms of Eq. (15.A.11) yield
^
h
AW
k ( ) = ^c
AW
k ( ) 1
W
^
h
WW
k ( )
_ _
^
h
AA
k ( ) = ^c
AA
k ( )
W
^c
AW
k ( ) [ [
2
1
W
^
h
WW
k ( )
_ _
(15:A:12)
where the caret indicates the Fourier transform. The term 1
W
^
h
WW
k ( )
_ _
comes
from the oxygenoxygen structure factor of pure water. It is necessarily nite and
positive. Hence Eq. (15.A.12) can be inverted to solve for ^c
AW
k ( ) and thus c
AW
r ( ) in
terms of well-behaved quantities. Similarly, c
AA
r ( ) is also well dened.
The very idea of introducing direct correlation functions is that they are short-
ranged, essentially zero beyond distances larger than the range of the interaction pair
potential. This suggests the following approximations for the direct correlation
functions when they are at distances greater than contact separations
c
AW
r ( ) = 0; r >
A
=2 ( ) r
W
= d
AW
c
AA
r ( ) = 0; r >
A
= d
AA
(15:A:13)
These approximate equations can be combined with the following exact boundary
conditions
g
0 ( )
AW
r ( )
_ _
HS
= h
AW
r ( ) 1 = 0; r5d
AW
g
0 ( )
AA
r ( )
_ _
HS
= h
AA
r ( ) 1 = 0; r5d
AA
(15:A:14)
to close the OrnsteinZernike-like integral equations (15.A.11). To solve these
equations, one must determine the c
AW
r ( ) function which, when inserted into Eq.
(15.A.11), satises Eq. (15.A.14). Once determined, Eq. (15.A.11) can be integrated
to yield h
AW
r ( ) for r >
A
=2 ( ) r
W
.
Equation (15.A.11) can be solved with the help of the closure relations (15.A.13)
and (15.A.14) this is the main idea behind the PrattChandler theory.
To repeat, once we get g
(0)
AW
(r)
_ _
HS
we can obtain the chemical potential of the
non-polar solute from Eq. (15.A.14). Similarly, g
(0)
AA
(r)
_ _
HS
gives the pair
hydrophobicity.
References
1. C. Tanford, The Hydrophobic Effect (New York: Wiley, 1973).
2. C. Tanford, The hydrophobic effect and the organization of living matter. Science, 200
(1978), 1012.
3. L. R. Pratt, Theory of hydrophobic effects. Annu. Rev. Phys. Chem., 36 (1985), 433.
4. K. A. Dill and S. Bromberg, Molecular Driving Forces: Statistical Thermodynamics in
Chemistry and Biology (New York: Garland Science, 2003).
5. Y. Jung and R. A. Marcus, On the theory of organic catalysis on water. J. Am. Chem.
Soc., 129 (2007), 54925502.
6. A. E. Miller, P. B. Peterson, C. W. Hollars, R. J. Saykally, J. Hyeda, and P. Jungwirth,
Behavior of -amyloid 116 at the air-water interface at varying pHby nonlinear spectro-
scopy and molecular dynamics simulation. J. Phys. Chem. A, 115 (2011), 58735880.
7. P. Bhimalapuram and B. Widom, The hydrophobic heat-capacity anomaly. Physica A,
298 (2001), 229.
8. N. T. Southall , K. A. Dill, and A. D. J. Haymet, A view of the hydrophobic effect.
J. Phys. Chem. B, 106 (2002), 521533.
9. B. Lee, The physical origin of the low solubility of nonpolar solutes in water.
Biopolymers, 24 (1985), 813.
10. L. X. Dang, Potential of mean force for the methanemethane pair in water. J. Chem.
Phys., 100 (1994), 9032.
11. A. Mukherjee, P. Bhimalapuram, and B. Bagchi, Orientation-dependent potential of
mean force for protein folding. J. Chem. Phys., 123 (2005), 014901.
12. D. Ben-Amotz and D. R. Herschbach, Estimation of effective diameters for molecular
uids. J. Phys. Chem., 94 (1990), 10381047.
13. A. Bondi, van der Waals volumes and radii. J. Phys. Chem., 68 (1964), 441.
14. P. J. Flory, Principles of Polymer Chemistry (Ithaca, NY: Cornell University Press,
1953).
15. P. M. Chaikin and T. C. Lubensky, Principles of Condensed Matter Physics (Cambridge:
Cambridge University Press, 1995)
16. F. H. Stillinger, Structure in aqueous solutions of nonpolar solutes from the standpoint
of scaled particle theory. J. Solution Chem., 2 (1973) 141.
17. L. R. Pratt and D. Chandler, Theory of the hydrophobic effect. J. Chem. Phys., 67
(1977), 3683.
18. J.D. Weeks, D. Chandler, and H. Andersen, Role of repulsive forces in determining the
equilibrium structure of simple liquids. J. Chem. Phys., 54 (1971) 5237.
References 241
19. G. Hummer, S. Garde, A. E. Garcia, A. Pohorile, and L. R. Pratt, An information theory
model of hydrophobic interaction, Proc. Natl. Acad. Sci. USA, 93 (1996) 8951.
20. S. Chatterjee, P. Debenedetti, and F. H. Stillinger, Scaled particle theory for hard sphere
pairs. II. Numerical analysis. J. Chem. Phys., 125 (2006) 204505.
21. K. Lum, D. Chandler, and J. D. Weeks, Hydrophobicity at small and large length scales.
J. Phys. Chem. B, 103 (1999), 4570.
16
The amphiphilic effect: the diverse but intimate
world of aqueous binary mixtures
Aqueous binary mixtures with amphiphilic solutes exhibit an amazing
range of interesting properties that have been somehow neglected theo-
retically and have begun to be understood at a molecular level only
recently. These binary mixtures are of great importance in biology and
chemistry. Important amphiphilic solutes such as dimethyl sulfoxide,
methanol, ethanol, dioxane, and tertiary butyl alcohol have a combina-
tion of hydrophobic and hydrophilic groups that allow them to interact
with biopolymers and large molecules in many different ways. Sometimes
these solutes can form aggregates in water while remaining soluble, and
provide a surface that is heterogeneous on the length scale of the mole-
cular diameter of water. Here we discuss some of the fascinating proper-
ties of these aqueous mixtures which have long been known, although not
often properly explained or understood at a molecular level.
16.1 Introduction: the role of aqueous mixtures in chemistry and biology
Aqueous mixtures constitute an important class of solvents for chemical and
biological applications; pure water hardly nds use as a solvent. As discussed in
the last two chapters, different aqueous solutions with a hydrophilic or hydrophobic
cosolvent have drawn enormous attention in the last three to four decades. On the
other hand, only limited theoretical or even experimental efforts have been devoted
to the study of those aqueous solvents where the solute is amphiphilic in nature. This
is indeed surprising because aqueous mixtures with small amphiphilic solutes are in
great demand in industry and also exhibit an array of unusual properties that have
remained largely unexplained, although the situation has begun to change in the last
decade.
Examples of important amphiphilic binary mixtures include water and dimethyl
sulfoxide (DMSO), watermethanol, waterethanol, watertertiary butyl alcohol
(TBA), water and glycerol, water and dioxane, to name a few. These amphiphilic
243
cosolvents form a miscible binary mixture that remains soluble over the entire
composition range, under ambient conditions. However, the chemical composition
of such binary solutions can locally be different, as the free energy can be rather
insensitive to local composition uctuations. Thus in combination with water
(which itself is capable of generating many structures), these amphiphilic cosolvents
are found to show many unique properties that are of immense importance in
various elds. The molecular structures of some of these cosolvents are shown in
Figure 16.1.
As these cosolvents contain both hydrophilic and hydrophobic groups, the same
molecule can induce opposite effects in water. The hydrophilic part can interact with
water to form strong HBs, while the hydrophobic part may induce cooperative
ordering in the systemby a hydrophobic hydration effect. These two effects combine
together to regulate the extensive HB network of water in their aqueous binary
mixtures that is reected in strong, often anomalous non-ideal behavior in many
physical properties such as viscosity, density, dielectric constant, excess mixing
volume, surface tension, heat of formation, etc.
Moreover, as the composition can be varied easily, a range of properties can be
accessed or tuned by varying the composition. For example, we can control the
dielectric constant of the solvent and the solubility of the solute, among other
(a)
(b)
(d)
(f)
Methanol
Tert-butyl alcohol
1,4-dioxane
(c)
Dimethyl sulfoxide
Ethanol
Acetone
(e)
(0.459)
(0.7)
(0.3)
(0.5723)
(0.466)
(0.611)
(0.565)
(+0.139)
(+0.16)
(+0.403)
(+0.208)
(+0.565)
(0.000) (0.000)
(0.000)
(+0.16)
(+0.4)
(+0.1010)
(+0.0428)
(+0.0320)
(+0.1582)
(+0.1010)
(+0.3278)
Figure 16.1. Molecular structure of some amphiphilic solutes. (a) Dimethyl
sulphoxide (DMSO), (b) methanol (MeOH), (c) ethanol (EtOH), (d) tert-butyl
alcohol (TBA), (e) acetone. For all these solutes the partial charges are indicated on
the corresponding atoms according to the GROMOS-96 force eld. (f) Molecular
structure of 1,4-dioxane. For 1,4-dioxane the partial charges are indicated on the
respective atoms according to J. Am. Chem. Soc., 127 (2005), 1101911028.
244 The amphiphilic effect: diverse but intimate world of aqueous binary mixtures
properties. For practical utility, the properties of the two components of the binary
mixture should be different from each other, so that their mixture can give rise to
special behavior. These composition-dependent properties are exploited to tune
relevant properties of proteins, polymers, and bio-polymers. Hence, these cosol-
vents have gained immense practical importance in the past decade, and their scope
of application is ever-widening, ranging from cryoprotection to enhancement of
enzymatic activity.
Despite their importance, a satisfactory microscopic understanding of such exotic
properties is not available even now because of the complexity in the nature of their
interactions. We have seen in the last two chapters that description of either hydro-
phobic or hydrophilic effects itself is quite difcult. Not surprisingly therefore the
development of a theory of amphiphilic binary mixtures has been left for posterity.
16.2 Non-ideality of amphiphilic binary mixtures
If the nature of the molecules constituting the binary mixture is not too different
from each other, such as a mixture of methane and ethane, then the mixture behaves
ideally and follows Raoults law. According to Raoults law, a property P of a
mixture is given by the addition of weighted contributions [1]
P
id
= x
1
P
1
x
2
P
2
(16:1)
In general, however, this simple Raoults law is found to be inadequate for many
practical applications. Interestingly, the most notable deviation is provided by those
binary solutions that we discuss in the current chapter, that is, where the solute is
amphiphilic in nature! These deviations reveal information about the intermolecular
interactions and also the molecular arrangement in these mixtures. The deviations
can be of various kinds, positive, negative and complex, showing both a maximum
and a minimum.
16.3 WaterDMSO binary mixture
Dimethyl sulfoxide (DMSO) is the simplest organosulfur compound, with the
formula (CH
3
)
2
SO (see Figure 16.1(a)) [2]. This colorless liquid is an important
polar aprotic solvent that dissolves both polar and non-polar compounds and is
miscible in a wide range of organic solvents as well as water. It penetrates the skin
very readily, giving it the unusual property of being secreted onto the surface of the
tongue after contact with the skin and causing a garlic-like taste in the mouth. It can
play a role as a protein stabilizer, an activator, a denaturant, or an inhibitor, and also
as a cryoprotector. In the drug discovery processes, DMSO is commonly utilized to
prepare the standard solvent for preparing stock solutions of compounds.
In aqueous DMSO solution, there surprisingly exist two ranges of compositions
where the concentration dependence of many properties exhibits non-monotonic
changes in their behavior. One of this occurs in the ~3540% DMSO concentration
(mole percentage) range and the other takes place at a lower DMSO concentration
range, ~1015%.
The anomalies near the high concentration range have been investigated by
extensive computer simulation studies to understand the molecular origin of the
anomalies observed in the thermodynamic, dielectric, and dynamic properties in the
said concentration range [3,4]. For example, both the shear viscosity and the thermal
conductivity of the mixture have been analyzed and compared with available
experimental data under ambient conditions. The enhancement of shear viscosity
at mole fraction x
w
= 0.65 (x
DMSO
= 0.35) has been quantitatively reproduced in
simulations [5].
Computer simulation studies have also found that one DMSO molecule forms
two rather strong HBs with two different water molecules to form a 1DMSO2H
2
O
complex in solution. Such a complex is of course not a stable chemical species in the
sense of a chemical molecule. Note that if every DMSO molecule is hydrogen-
bonded to two different water molecules, then the bonding condition is satised for
all and sundry at a 1/3 or 33.3%mole fraction of DMSO. If water molecules can still
form HBs among themselves (as each needs to form about four HBs) and methyl
groups can group together due to hydrophobic attraction, then such an arrangement
can cause signicant slowing down of the dynamics near the 3540% DMSO
concentration range, giving rise to anomalies [3].
Although, traditionally, more attention has been focused on the 3040% DMSO
concentration range, it is not until recently that the non-ideal properties of DMSO at
a lowconcentration range (1015%mole percentage) have been fully explored. The
non-ideality here has also been reected in a number of physical properties such as
viscosity, density, dielectric constant, translational and rotational diffusion con-
stants, excess mixing volume, surface tension, and heat of formation, to mention a
few [68]. Among several such properties, anomalous behavior in the diffusion
coefcient of DMSO and in the average local composition uctuation of water,
mean-square deviation of total dipole moment and the orientational correlation
function of the OH bond of water is demonstrated in the gures below [9].
The anomalies in the thermodynamic properties in the low DMSO concentration
range are further supported by a mass spectroscopic study which shows that with
varying water mole fraction (x
w
) the cluster structure changes nonlinearly upon
increasing x
w
, as shown in Figure 16.3. With decreasing water concentrations in
these mixtures, the relative intensity of water clusters decreases slightly until the
value of the water mole fraction, x
w
= 0.93 is reached. At x
w
~ 0.93, the relative
intensity of the water cluster starts to drop suddenly and decreases down to 0.2 at x
w
(b)
Figure 16.2. Anomalous properties of aqueous DMSO solution. (a) Diffusion
coefcient of DMSO molecules at different concentrations of the binary mixture.
(b) Local concentration uctuation of water within a sphere of 0.7 nm radius in the
simulation box. (c) Mean-square deviation of total dipole moment of the binary
mixture at different concentrations. (d) Orientational correlation function of the
OH bond of water. Note that in all the cases the deviation in the concentration
range x
DMSO
0.100.15 becomes evident. Figures adapted with permission from
J. Phys. Chem. B, 114 (2010), 12875 and J. Phys. Chem. B, 115 (2011), 685.
D
D
M
S
O

(

1
0
5

c
m
2
/
s
e
c
)
x
DMSO
0.05 0.10 0.20 0.25 0.30 0.15
(a)
0.8
0.7
0.6
0.5
0.4
~ 0.91. Thereafter, it begins to decrease again slowly and eventually becomes nearly
zero beyond x
w
~ 0.8. The variation in the size of the DMSO clusters is opposite to
that of the water clusters. This result is evidence that the microscopic structure in
DMSOwater mixtures can vary nonlinearly with the solvent composition, exhibit-
ing the existence of a critical value of the mixing ratios, at x
w
~ 0.910.93, where
(c)
(d)
Figure 16.2. (cont.)
drastic changes in the microscopic structure occur [10]. We are not aware of detailed
simulations for cluster size distribution of a waterDMSO binary mixture in the gas
phase under conditions similar to those employed in the mass spectroscopic studies.
Actually, low-temperature studies of cluster formation between water and DMSO
molecules can provide valuable information and is also a good test of computer
models.
Theoretical results for the solution phase are somewhat different from the experi-
mental results in the gas phase. In the solution phase we do have extensive clustering
among DMSO molecules but that starts around 1415% of DMSO.
16.4 Wateralcohol binary mixture
Most common wateralcohol mixtures are watermethanol and waterethanol
solutions. Mixtures of water with other alcohols higher in the series (such as
propanol and butanol) are also common. Waterphenol mixtures are also routinely
used in laboratories and surgical procedures as a disinfectant.
Wateralcohol mixtures act as good solvents because they can solvate organic
solutes which are not usually soluble in water. They can also solvate ions which are
sometimes soluble in alcohols.
These solvents are now found to have peculiar properties in low alcohol con-
centration limits. Here in this section we discuss three such well-known alcohols:
methanol, ethanol, and tertiary butanol (TBA). Figures 16.1(b), (c), and (d) illustrate
their structures, respectively.
Figure 16.3. Observed variation of the intensity ratio for DMSO clusters (open
symbols and dotted lines) and for water clusters (closed symbols and solid lines), as
a function of the water mole fraction (X
w
), within the mass range 80500 amu.
Figure adapted with permission from J. Phys. Chem. B, 105 (2001), 67596762.
16.4.1 Aqueous methanol solution
Watermethanol mixtures are important solvent media in both chemistry and bio-
chemistry. It was found that watermethanol mixtures utilize the phenomenon of
preferential solvation of ions and hydrophobic solutes. It is worth mentioning here
that the dipole moment of methanol is slightly lower than that of water. Moreover the
presence of the methyl group not only prevents the strong electrostatic interaction
with the other species but also makes methanol molecules considerably more bulky as
compared with water molecules. Apart fromthe structural particulars, the dynamics of
this cosolvent along with water is rather complex in watermethanol mixtures.
The diffusion coefcients of ionic solutes show nonideal behavior with variation
of composition of the solvent mixture in watermethanol binary mixtures. The
degree of non-ideality of the solute diffusion is found to be similar to the non-
ideality that is observed for the diffusion of water and methanol molecules in these
mixtures and is attributed to the enhanced stability of the HBs and formation of
interspecies complexes in the mixtures. The diffusion coefcient of water is found
to be minimum at x
w
= 0.5 and that of methanol shows the minimum at x
w
= 0.7.
However, the observed deviation from linear behavior with composition is found to
be a bit weaker than that found in simulations of waterDMSO mixtures [11,12].
16.4.2 Aqueous ethanol solution
A waterethanol binary mixture exhibits anomalies in physicochemical properties
over a wide range of composition. Waterethanol is perhaps the most-studied binary
mixture in the literature, starting fromthe days of Mendeleev. Partial molar volumes
indicate that the solute apparently contracts up to an ethanol mole fraction of x
eth
~
0.08. Dielectric measurements, permittivity spectra, and mass spectrometric studies
[13] all seem to suggest an underlying aggregation phenomenon in the low con-
centration range, although the precise value of the critical composition seems to
depend on the nature of the experiments used. It is interesting to note that most
alcoholic beverages fall in this composition range. Properties such as the local
composition uctuation and the radial distribution function and dynamic properties
such as the diffusion coefcient and the orientational correlation function all showa
dramatic non-monotonic behavior around that particular mole fraction range. Here
one representative anomalous property, the local composition uctuation, is dem-
onstrated in Figure 16.4 [14].
16.4.3 Watertertiary butyl alcohol
An aqueous solution of TBA represents yet another extreme where the effective
hydrophobic interaction dominates the hydrophilic interaction, shifting the
anomalous range to a much lower concentration, typically in the range x
TBA
~0:05.
For instance, waterTBA mixtures show a sharp minimum of the partial molar
volume and a maximum of the partial molar specic heat at TBA molar fraction
x
TBA
~0:04 [15]. Small-angle X-ray diffraction and neutron-scattering experiments
show abrupt density uctuations of the TBA solution in the mole fraction range
x
TBA
~0:06 0:16. The aggregation phenomenon in TBA is pretty well known, but
it has not been classied as a general phenomenon in terms of percolation in binary
mixtures [16,17].
In a series of recent experimental studies, it has been found that for water
methanol, waterethanol, and waterTBA binary mixtures, striking dynamic
anomalies occur at a low solute concentration range. The anomalies can be captured
by spectroscopic techniques. It has been observed that the rotational anisotropy in
such systems has a fast component which becomes faster and a slow component
which becomes slower with increasing solute concentration. Temperature depen-
dence of the concentration uctuations N(DX)
2
)
_ _
was observed for waterTBA
at various concentrations by the use of light-scattering spectra. In TBAwater
mixtures a sudden jump in values of N(DX)
2
) due to the temperature increase in
the range 17.563C was observed for x
TBA
> 0.05. This result is attributed to the
Figure 16.4. Anomalous properties of water-ethanol mixture. The gure shows the
local composition uctuation of ethanol (mole fraction x
i
) within a sphere of radius
r
c
, for three different values of r
c
. The bulk composition of the system is taken as x,
which is the same as x
eth
. It is interesting to note here that there is a sharp rise in the
average local composition uctuation in the system after x
eth
= 0.10, and that the
uctuation is maximum for r
c
= 0.6 nm, indicating that this anomalous
composition uctuation is a local phenomenon. Figure adapted with permission
from J. Phys. Chem. B, 116 (2012), 3713. Copyright (2012) American Institute of
Physics.
breakdown of the clathrate hydrate-like structure and the growth of the water part of
the local structure which is composed of water and TBA molecules. This interpreta-
tion is further supported by the mixing enthalpy observed for TBAwater mixtures,
which reects the change of the intermolecular interaction on mixing [18].
16.5 Wateracetone binary mixture
A wateracetone binary mixture is commonly used to separate non water-soluble
materials in a chromatography lab. Acetone can dissolve a large number of organic
molecules, with the exception of very greasy molecules. Though acetone and
DMSO are structurally quite similar (with the carbon of the carbonyl group of
acetone replaced by a sulfur atom in DMSO) (see Figure 16.1(e)), their hydrogen-
bonding abilities are found to be rather dissimilar. While the DMSOwater HBs are
found to be signicantly stronger with longer lifetimes than waterwater HBs,
wateracetone HBs are found to be much weaker and have much shorter lifetimes
than waterwater HB. Despite the existence of such diversity, wateracetone mix-
tures also show dramatic non-ideal behavior in their equilibrium and dynamic
properties with variation of composition.
Among several dynamic properties of this aqueous mixture, simulations found a
striking anomalous concentration region where the self-diffusion coefcient and the
orientational relaxation times of water and acetone molecules deviate from their
ideal behavior. The diffusion coefcient of water shows a minimumat x
w
= 0.75 and
that of acetone shows a minimumat x
w
= 0.90. The orientational relaxation times for
both water and acetone have been calculated and are observed to deviate in the same
concentration range [19].
16.6 Waterdioxane binary mixture
Dioxane is a cyclic diether forming a six-membered ring [20]. Thus it is a nearly non-
polar symmetric molecule. 1,4-Dioxane is an extraordinary solvent, capable of
solubilizing most organic compounds, and water in all proportions, and many inor-
ganic compounds. The self-diffusion coefcient of dioxane is 1.1 10
5
cm
2
/s, about
half that of a water molecule. The effective diameter of dioxane is 5.5 about twice
that of a water molecule. One should not forget that a waterdioxane mixture narrowly
avoids a lower critical consolute point. However, the effects of criticality are reected
in the values of the mutual diffusion coefcient and viscosity. Note that binary
mixtures are often chosen so that they are mixable (do not phase separate). Thus,
the two components interact attractively and strongly.
Along with the above structural view the low-frequency Raman spectra of dioxane
aqueous solutions have been analyzed fromthe dynamic aspect of the water structure.
The reduced Raman spectra of dioxane aqueous solutions are well distinguished by a
superposition of three characteristic modes of water and one Gaussian mode of
dioxane. The concentration dependence of the spectra indicates that the mode of
water disappears belowabout 0.8 molar fraction of water molecules; in fact this molar
fraction corresponds to a ratio of four water molecules to one dioxane. This result
suggests that the well-dened tetrahedral structure consisting of ve water molecules
is largely disrupted by the penetration of dioxane molecules above a 0.2 mole fraction
of dioxane, according to the Raman spectral experiments [21].
There are plenty of examples where different dynamics and thermodynamic
properties show remarkable anomalies in such an amphiphilic binary mixture.
Nevertheless it is very difcult to gather them all together in one chapter in a
book like this. However, for a waterdioxane mixture we discuss one such
property to point out the anomalous region that can be further followed by the
interested reader.
16.7 Liquidliquid structural transformation in aqueous binary
mixtures: a generic phenomenon for amphiphilic solutes
The experimentally as well as theoretically observed anomalies have been attributed
to a structural transformation that seems to be driven by a continuum percolation-
like transition at particular cosolvent concentration ranges. As an example, for
DMSO the anomalous mole fraction range is x
D
~ 0.120.15. The largest cluster
size of CH
3
CH
3
aggregation clearly indicates the formation of such percolating
clusters. As a result, markedly unusual behaviors in local composition uctuation,
diffusion constant and mean-square uctuation of total dipole moment are observed
that again suggest a structural transition around the same concentration range. As
mentioned, this transformation seems to be triggered by a clustering or percolation
transition where the hydrophobic groups cluster in a zigzag chain-like state. Water
plays a key role in facilitating this aggregation. However, this clustering does not
lead to any phase separation, as the system remains homogeneous on a macroscopic
scale, although local composition heterogeneities exist.
In essence, the percolation-based explanation may be made more accurate by
detecting the percolation threshold. By fractal dimension analysis and evaluation of
maximum cluster size, a percolation threshold (which appears at 15% DMSO mole
fraction) has been detected in aqueous DMSO mixture.
Such a percolating network structure was also observed in the case of water
methanol and waterethanol binary mixtures and is probably a general feature for
these types of cosolvents which have both hydrophobic and hydrophilic moieties.
Surprisingly, all these dynamic and thermodynamic anomalies are manifested in a
specic, narrow, low solute concentration range, depending on the nature of the
16.7 Liquidliquid structural transformation in aqueous binary mixtures 253
solute. Here we discuss the possibility that these anomalies arise from a general
aggregation phenomenon that involves rst a micro-micellization that is followed
by a percolation-transition-aided structural transformation in the solution phase
itself.
That this could be a general scenario is suggested by results of extensive MD
simulations of three different amphiphilic solutes DMSO, ethanol, and TBA in
respective aqueous binary mixtures. Theoretical analysis of large-scale computer
simulation results leads to the following molecular picture of the anomalous
concentration range. At extremely low concentration (less than 25% in most
cases) these solutes form micelle-like islands which are segregated. At somewhat
higher concentration these islands start aggregating through hydrophobic interac-
tion, and beyond a critical concentration they form a spanning cluster (see
Figure 16.5), via a percolation transition. The percolation phenomenon seems to
be quite general for such aqueous amphiphilic solutions where competetitive
hydrophobic and hydrophilic interactions determine the structural and dynamic
ordering of the whole system [9].
16.8 Theoretical development
As already emphasized, theoretical development in the area of aqueous binary
mixtures has been comparatively slow and to date no satisfactory molecular theory
exists that can describe the complex physical chemistry of a binary solution. The
reason is the complexity of the intermolecular potential. While binary mixtures have
often been studied by using a cell or lattice theory (as we discussed in the description
of a polymer solution in the Hydrophobic effects chapter), even such a description
is hard here because of the amphiphilic nature of the solute. It is really hard to
develop a quantitative theory that includes the two different types of local hetero-
geneity at two sides of a given solute molecule.
An initial study of a mixture of two spherical dipolar molecules already shows the
richness and the difculty of the problem. We briey describe such a study, which
brings out the role of intermolecular orientational correlations [22].
The molecular theory considers a dipolar liquid where the two constituents are
LennardJones spheres each with an embedded dipole moment at the center. The
LennardJones parameters (sizes, interaction strength parameters) and also values
of the dipole moments are different for the two species. The theory properly includes
the differing inter- and intramolecular correlations that are present in a binary
mixture. As a result, the theory can explain several important aspects of the non-
ideality of equilibrium solvation energy (broadly known as preferential solvation)
observed in experiments. The non-ideality of solvation is found to depend on both
the molecular sizes and the magnitude of the dipole moments of the solvent
Figure 16.5. Snapshots of the simulation of different binary mixtures water
DMSO in the top panel, waterethanol in the middle, and waterTBA in the
bottom panel. Water molecules are shown in silver. Co-solvents (DMSO,
ethanol, and TBA) are represented in blue. The snapshot is shown at two
different concentrations one before the onset of percolation to show the
microheterogeneity in the system, and one after the onset of percolation to
show the spanning cluster of the cosolvent. Figure adapted with permission
from J. Phys. Chem. B, 115 (2011), 685. Copyright (2011) American Chemical
Society. See plate section for color version.
16.8 Theoretical development 255
molecules. The interactions between the solvent molecules play an important role in
determining the extent of this non-ideality. The dynamic calculations are based on a
generalized Smoluchowski equation which has been used extensively for studies in
one-component liquids [22]. For a binary liquid, this study reveals rich and diverse
behavior such as dependencies on the sizes, the transport coefcients, and the polar
properties of the components. The theory also offers a detailed picture of the
dependence of the SD on the composition of the mixture. It is predicted that the
dynamics of solvation in a binary liquid is, in general, non-exponential and that
the details of the dynamics can be quite different from those in a one-component
liquid. In particular, the continuum model is found to be grossly inaccurate in
describing the SD in binary mixtures and rather extreme conditions are needed to
recover the predictions of the continuum model which can be attributed to the non-
ideality of the solvation.
16.9 Biological applications
The interest in most of these aqueous binary mixtures is not only due to their varied
and tunable properties as solvents and as reaction media but also due to their unique
biological properties.
It has been established that in some cases protein structure and function can be
greatly inuenced upon addition of a small amount of certain cosolvents/denatur-
ants in the aqueous medium. For this reason, the effects of DMSO, methanol,
ethanol, TBA, acetone, dioxane, and triuoroethanol (TFE) on protein structure
and function are increasingly being exploited in protein chemistry.
Among the cosolvents, the effects of DMSO on proteins are particularly inter-
esting and diverse. It can play a role as a stabilizer, an activator, a denaturant, an
inhibitor, and also as a cryoprotector. Denaturation of proteins induced by DMSOas
well as other cosolvents occurs at threshold concentrations where the tertiary and
even the secondary structures of proteins are highly disrupted [23]. UV circular
dichroismspectral study shows that from2025%(v/v) (~6%mole fraction) DMSO
concentration, lysozyme proceeds gradually from its native to the partially unfolded
state. This indicates a broad structural transition which is essentially completed by
50% (v/v) (~18% mole fraction) DMSO concentration [24].
Recent molecular dynamic simulations of lysozyme in waterDMSO reveal the
following sequence of changes on increasing DMSO concentration. (i) At the initial
stage (around 5% DMSO concentration) the proteins conformational exibility is
markedly suppressed (see Figure 16.6(a)). From study of the radial distribution
functions, we attribute this to the preferential solvation of exposed protein hydro-
phobic residues by the methyl groups of DMSO. (ii) In the next stage (1015%
DMSO concentration range) lysozyme partially unfolds, accompanied by an
increase both in uctuation and in exposed protein surface area (see Figure 16.6(a)
and (b)). (iii) Between 15% and 20% concentration ranges, both conformational
uctuation and solvent-accessible protein surface area suddenly decrease again,
indicating the formation of an intermediate collapse state. These results are in good
0 5 10 15 20 25 30 35 40
Mole Fraction of DMSO (%)
1100
1150
1200
1250
1300
1350
N
o
n
p
o
l
a
r

S
A
S
A

(
2
)

P
a
r
t
i
a
l

u
n
f
o
l
d
i
n
g

r
e
g
i
o
n
P
e
r
c
o
l
a
t
i
o
n

i
n
d
u
c
e
d

r
e
g
i
o
n
(b)
Figure 16.6. Conformational exibility and partial unfolding of lysozyme. (a) C
root mean square displacement (RMSD) and (b) non-polar solvent accessible surface
area (SASA) at several compositions of waterDMSO binary mixture. Note that the
partial unfolding region (515% mole fraction of DMSO) and the percolation-
induced region (1520% mole fraction of DMSO) together bear the signature of
non-monotonic concentration-dependent conformational uctuation around the
active site of the lysozyme. Figure adapted with permission from J. Chem. Phys.,
136 (2012), 115103. Copyright (2012) American Institute of Physics.
0 5 10 15 20 25 30 35 40
Mole Fraction of DMSO (%)
1.6
1.8
2
2.2
2.4
2.6
2.8
3
R
M
S
D

o
f

C

A
t
o
m
s

(
)
P
a
r
t
i
a
l

u
n
f
o
l
d
i
n
g

r
e
g
i
o
n
P
e
r
c
o
l
a
t
i
o
n

i
n
d
u
c
e
d

r
e
g
i
o
n
(a)
16.9. Biological applications 257
agreement with near-UV circular dichroism (CD) and uorescence studies [24].
Explanation of this apparently surprising behavior in terms of a structural transfor-
mation involves clustering among the methyl groups of DMSO. (iv) Beyond a 20%
concentration of DMSO, the protein starts its nal sojourn towards the unfolding
state with a further increase in conformational uctuation. Most importantly,
uctuation near the active site reveals that both partial unfolding and
conformational uctuations are centered mostly on the hydrophobic core of the
active site of the lysozyme [23]. These results could offer a general explanation and
a universal picture of the anomalous behavior of protein structure and function
observed in the presence of cosolvents (DMSO, ethanol, TBA, dioxane) at low
concentrations.
16.10 Conclusion
Interestingly, mostly those aqueous binary mixtures that contain a marked amphi-
philic character towards water are found to be particularly important in chemistry
and biology. This is clearly manifest in all useful aqueous binary mixtures such as
DMSO, methanol, ethanol, TBA, acetone, and dioxane to name a few. Some of
these solvents (DMSO, EtOH) are used at high concentration as effective denatur-
ants of protein. At lowconcentration they can exhibit a reversal of role and serve as a
promoter of stability. In this low-concentration regime the binary mixtures also
promote the catalytic activity of enzymes, as discussed above. Recent experimental,
theoretical, and simulation results exhibit the phase-transition-like scenarios dis-
cussed in this chapter. These systems need to be studied in great detail as much
remains to be understood.
References
1. P. Tzias, C. Treiner, and M. Chemla, Applicability of Raoults law in nonideal mixed
solvents. J. Solution Chem., 6:6 (1977), 393402.
2. S. W. Jacobs, E. E. Rosenbaum, and D. C. Wood, Dimethyl Sulfoxide (New York: Marcel
Dekker, 1971).
3. A. Luzar and D. Chandler, Structure and hydrogen bond dynamics of waterdimethyl
sulfoxide mixtures by computer simulations. J. Chem. Phys. 98 (1993),
81608173; A. K. Soper and A. Luzar, A neutron diffraction study of dimethyl sulph-
oxidewater mixtures. J. Chem. Phys., 97 (1992), 13201331; I. A. Borin and
M. S. Skaf, Molecular association between water and dimethyl sulfoxide in solution: a
molecular dynamics simulation study. J. Chem. Phys., 110 (1999), 64126420.
4. S. A. Schichman and R. L. Amey, Viscosity and local liquid structure in dimethyl
sulfoxide-water mixtures. J. Phys. Chem., 75 (1971), 98102.
5. C. N. Draghi, J. B. Avalos, and B. Rousseau, Transport properties of dimethyl sulfoxide
aqueous solutions. J. Chem. Phys., 119 (2003), 4782.
6. M. S. Skaff, Molecular dynamics simulations of dielectric properties of dimethyl sulf-
oxide: comparison between available potentials. J. Chem. Phys., 107 (1997), 7996.
7. C. C. Li, Thermal conductivity of liquid mixtures. Am. Inst. Chem. Eng. J., 22 (1976),
927930.
8. J. Mazurkiewicz and P. Tomasik, Viscosity and dielectric properties of liquid binary
mixtures. J. Phys. Org. Chem., 3 (1990), 493502.
9. S. Roy, S. Banerjee, N. Biyani, B. Jana, and B. Bagchi, Theoretical and computational
analysis of static and dynamic anomalies in waterDMSO binary mixture at low DMSO
concentrations. J. Phys. Chem. B, 115 (2011), 685692; S. Banerjee, S. Roy, and
B. Bagchi, Enhanced pair hydrophobicity in the waterdimethylsulfoxide (DMSO) binary
mixture at low DMSO concentrations. J. Phys. Chem. B, 114 (2010), 1287512882.
10. D. N. Shin, J. W. Wijnen, J. B. F. N. Engberts, and A. Wakisaka, On the origin of
microheterogeneity: a mass spectrometric study of dimethyl sulfoxidewater binary
mixture. J. Phys. Chem. B, 105 (2001), 67596762.
11. S. Chowdhuri and A. Chandra, Dynamics of ionic and hydrophobic solutes in water-
methanol mixtures of varying composition. J. Chem. Phys., 123 (2005), 234501.
12. L. A. Woolf, Insights into solute-solute-solvent interactions from transport property
measurements with particular reference to methanol-water mixtures and their constitu-
ents. Pure Appl. Chem., 57 (1985), 1083.
13. A. Wakisaka and K. Matsuura, Microheterogeneity of ethanolwater binary mixtures
observed at the cluster level. J. Mol. Liq., 129 (2006), 2532.
14. S. Banerjee, R. Ghosh, and B. Bagchi, Structural transformations, composition anoma-
lies and a dramatic collapse of linear polymer chains in dilute ethanolwater mixtures.
J. Phys. Chem. B, 116 (2012), 37133722.
15. T. Pradhan, P. Ghoshal, and R. Biswas, Structural transition in alcoholwater binary
mixtures: a spectroscopic study. J. Chem. Sci., 120 (2008), 275287.
16. K. Nishikawa, H. Hayashi, and T. Iijima, Temperature dependence of the concentration
uctuation, the Kirkwood-Buff parameters, and the correlation length of tert-butyl
alcohol and water mixtures studied by small-angle x-ray scattering. J. Phys. Chem.,
93 (1989), 65596565; K. Nishikawa, Y. Kodera, and T. Iijima, Fluctuations in the
particle number and concentration and the Kirkwood-Buff parameters of tert-butyl
alcohol and water mixtures studied by small-angle x-ray scattering. J. Phys. Chem.,
91 (1987),36943699.
17. G. DArrigo and J. Teixeira, Small-angle neutron scattering study of D
2
Oalcohol
solutions. J. Chem. Soc., Faraday Trans., 86 (1990), 15031509.
18. K. Iwasaki and T. Fujiyama, Light-scattering study of clathrate hydrate formation in
binary mixtures of tert-butyl alcohol and water. 2. Temperature effect. J. Phys. Chem.,
83 (1979), 463468.
19. R. Gupta and A. Chandra, Nonideality in diffusion of ionic and hydrophobic solutes and
pair dynamics in water-acetone mixtures of varying composition. J. Chem. Phys., 127
(2007), 024503.
20. F. C. Grozema , M. Swart , R. W. J. Zijlstra , J. J. Piet , L. D. A. Siebbeles, and P. T. van
Duijnen, QM/MM study of the role of the solvent in the formation of the charge
separated excited state in 9,9-bianthryl. J. Am. Chem. Soc., 127 (2005), 1101911028.
21. Y. Tominaga and S. M. Takeuch, Dynamical structure of water in dioxane aqueous
solution by low-frequency Raman scattering. J. Chem. Phys., 104 (1996), 73777381.
22. A. Chandra and B. Bagchi, Molecular theory of solvation and solvation dynamics in a
binary dipolar liquid. J. Chem. Phys. 94 , 83678377. A. Chandra and B. Bagchi,
A molecular theory of collective orientational relaxation in pure and binary dipolar
liquids. J. Chem. Phys., 91 (1989), 1829.
References 259
23. S. Roy, B. Jana, and B. Bagchi, Dimethyl sulfoxide induced structural transformations
and non-monotonic concentration dependence of conformational uctuation around
active site of lysozyme. J. Chem. Phys., 136 (2012), 115103.
24. S. Bhattacharjya and P. Balaram, Effects of organic solvents on protein structures:
observation of a structured helical core in hen egg-white lysozyme in aqueous dimethyl-
sulfoxide. Proteins, 29 (1997), 492.
17
Water in and around micelles, reverse micelles,
and microemulsions
In nature, water sometimes appears to have a mind of its own. It interacts
with complex molecules that possess both hydrophobic and hydrophilic
groups to facilitate the formation of exotic structures. Among many such
structures, micelles, reverse micelles, and microemulsions occupy special
places because of the nearly symmetric structures they form, their fre-
quent occurrence in nature, the conned and restricted environments
they provide, and the practical applications and utility they offer.
The extensive HB network present in bulk water gets frustrated in and
around these systems. The distinct structure and dynamics of water
molecules in these systems arise from the altered nature of hydrogen-
bonding and are reminiscent of those observed around biomolecules. As
in other cases, a large degree of understanding has come from computer
simulations. In this chapter we discuss some of the interesting properties
of these systems.
17.1 Introduction: different self-assemblies in water
The most well-known example of a self-organized systemis provided by proteins that
fold in aqueous solution into a compact native structure where most of the hydro-
phobic residues tend to reside in the core (popularly called the hydrophobic core)
while the hydrophilic residues are mostly on the surface. Such a self-organization is
possible because of the simultaneous presence of hydrophilic and hydrophobic amino
acid residues along the linear peptide chain. In fact, such self-organization of mole-
cules is fairly common in nature. In this chapter, we shall discuss how water leads to
the formations of exotic structures known as micelles, reverse micelles, and micro-
emulsions. These structures are formed by molecules known as surcants, which are
long-chain molecules and amphiphilic in nature, meaning that they contain two
distinct individual parts that like water (because the part is polar) and dislike
water (as it consists of hydrocarbons). These two opposite parts are usually located at
the two ends of the surfactant molecule, named head and tail.
261
17.2 Structure of micelles and reverse micelles
17.2.1 Micelles
The arrangement of surfactant molecules in micelles is relatively simple (compared
to proteins). On one end (the head group), surfactant molecules contain the polar
charged groups which are hydrophilic and the other end (the tail) usually contains a
non-polar long chain, consisting primarily of hydrocarbons that are hydrophobic.
When such amphiphilic surfactant molecules are dissolved in water, they associate
to form spherical or nearly spherical aggregates such that the hydrocarbon tails are
packed together to form a hydrophobic core and the hydrophilic groups face water.
Such self-organized structures are called micelles. This happens above a certain
critical concentration, known as the critical micellar concentration (CMC).
The size of the micellar aggregates is usually between 1 and 10 nm, and the
aggregation number, i.e., the number of surfactant molecules per micelle, ranges
from 20 to 200. Like proteins, the core of a micelle is essentially dry and consists of
the hydrocarbon chains. The polar charged head groups project outward into the bulk
water. Surrounding the core there is a layer composed of the ionic or polar headgroups,
bound counterions, and water molecules. This layer is called the Stern layer.
In between the Stern layer and the bulk water, there is another diffuse layer which
is sometimes termed the GuoyChapman (GC) layer [1]. The GClayer contains free
counterions and water molecules.
The structure of a micelle is dynamic and involves continuous exchange of
monomers between the aggregates and those in bulk solution. But this exchange
is slow, occurring on the timescale of several nanoseconds. A micelle also under-
goes slow uctuations in shape.
Micelles nd a wide variety of practical uses. They are used in detergents as
emulsiers. They are used in the extraction industry. Micelles form within the
human body to absorb vitamins. We discuss them below along with their structures.
Small-angle X-ray and neutron-scattering (SANS) studies have been used to
obtain detailed information on the structure of a variety of micelles. According to
these studies, the thickness of the Stern layer is 69 for cationic cetyl trimethy-
lammonium bromide (CTAB) micelles and anionic sodium dodecyl sulfate (SDS)
micelles. For nonionic micelles, the hydrocarbon core is surrounded by a palisade
layer, which consists of the polyoxyethylene groups hydrogen-bonded to water
molecules. The palisade layer is about 20 thick for neutral Triton X-100 (TX-100)
micelles. The radius of the dry, hydrophobic core of TX-100 is typically 2527
[2]. Thus the overall radius of the TX-100 micelle is about 51 and that of the SDS
micelle is about 30 . So, these structures are much bigger in size than a small
molecule. The radius of a water molecule is just about 1.5 . This is why micelles
are called nanoscopic materials. We show the structure of micelles in Figure 17.1.
262 Water in and around micelles, reverse micelles, and microemulsions
Many of the water molecules at the surface of a reverse micelle are restricted by
the HBs that they form with polar or charged groups at the surface of the micelle.
Thus, the micellar surface disrupts the extended HB network in more than one way.
17.2.2 Reverse micelles
Let us now discuss the structure of a reverse micelle. As the name suggests it has a
structural arrangement exactly opposite to that of a micelle. Reverse micelles
generally refer to aggregates of surcants (e.g., dioctyl sulfosuccinate, AOT)
formed in a non-polar solvent. In this situation, the polar headgroups of the
surcants point inward (core) and the hydrocarbon chains project outward into the
non-polar solvent [3]. The solvent one uses for reverse micelle formation is usually
liquid hydrocarbons. Recently the formation of reverse micelles in supercritical
uids such as ethane, propane, and carbon dioxide has been observed.
In contrast to micelles, the core region of a reverse micelle can encapsulate a fairly
large amount of water to form what is known as a microemulsion. Up to 50 water
molecules per surfactant molecule can be incorporated inside AOTreverse micelles.
Such a surfactant-coated nanometer-sized water droplet dispersed in a non-polar
liquid is called a water pool. The radius (r
w
) of the water pool varies linearly with
the water-to-surfactant mole ratio, W
0
. In n-heptane, r
w
(in ) is approximately
equal to 2W
0
. Thus, one can formreverse micelles with varying water pools and tune
the extent of connement by varying W
0
[4].
Importantly, apart from water, connement of other polar solvents such as
acetonitrile, alcohol, and formamide has been reported in such reverse micelles or
microemulsions. Below we illustrate the structural arrangement of reverse micelles
in Figure 17.2.
Figure 17.1. Schematic diagram of an aqueous micellar solution. The dry core,
Stern layer, headgroups, and bulk water are indicated in the gure. Figure adapted
with permission from Chem. Rev., 100 (2000), 2013. Copyright (2000) American
Chemical Society.
17.2 Structure of micelles and reverse micelles 263
(a)
(b)
Figure 17.2. (a) Snapshot from the MD trajectory of a model aqueous reverse-
micellar solution. Waterwater HBs are indicated by dashed lines. The layer-wise
decomposition of water molecules is shown in the snapshot. For this small reverse
micelle the number of intra- and interlayer HBs in the surface water (layer 3) is
greatly reduced. For water molecules in layer 2, in addition to intralayer HBs,
interlayer HBs are mostly with the inner layer 1. Figure adapted with permission
from J. Chem. Phys., 137 (2012), 01451519. Copyright (2012) American
Institute of Physics. (b) Schematic diagram of a reverse micelle in hydrocarbon.
Connement of water as a water pool is also highlighted here. Figure adapted
with permission from Chem. Rev., 100 (2000), 2013. Copyright (2000) American
Chemical Society. The water molecules in the water pool of the reverse micelle are
constrained by the charged groups at the interior surface. This restriction frustrates
the HB network near the interior surface. While this frustration decreases as one
travels inside the pool, it can persist well into a large pool. We shall discuss the
consequences in later sections.
17.3 Dynamics of water surrounding micelles
Water around a micelle exhibits distinct dynamic properties. Initially, the relaxa-
tion properties of micelles were studied using ultrasonic absorption in the 100
kHz2 GHz frequency range. These measurements indicated the existence of
multiple relaxation timescales. The longest relaxation time obtained was in
microseconds, which is really long if we consider that water dynamics occurs
on the picosecond timescale. In the intermediate time range, relaxation was found
to occur with a time constant of about 10 ns, and the fastest timescale measured
was in the 0.10.3 ns range. The longest relaxation time was attributed to the
exchange of surfactant monomers between bulk and micelles, and the shortest to
the rotation of the alkyl chains of the surcants in the core of the micelle. The
intermediate relaxation time was not assigned to any particular motion at that
time but is expected to be due to the relaxation of water molecules in the Stern
layer [5].
Solvation dynamics (SD) studies of micellar solutions have reported a timescale
which does not match with the dynamics either when the probe is inside the bulk
water or when it is inside the bulk hydrocarbon (core). This indicates that the probe
used resided neither in the bulk water nor in the dry core region, but was located in
the Stern layer. Bhattacharyya and co-workers have studied SD in several micelles
and found that the SD in the Stern layer of the micelles is three orders of magnitude
slower than that in bulk water (in bulk the relaxation time is on the sub-picoseond
timescale) [6]. The components that could cause solvation in the Stern layer of
micelles are the polar or ionic headgroups of the surcants, the counterions, and the
water molecules. In such an environment water motion could be severely restricted,
giving rise to the slow component of SD.
Computer simulation studies have explored translational and rotational dynamics
in micellar solutions and shown that both translational and rotational dynamics in
the hydration layer of micelles (Stern layer) are signicantly slower than that in the
bulk. The dipoledipole time correlation function (which measures the rotational
dynamics) shows the appearance of a long-time tail of the time constant in the
100 ps range or above. The dependence of the rotational dynamics on the probe
location has also been investigated and it was found that the dynamics becomes
faster as the probe moves away from the surface [7].
The translational diffusion of water molecules at the micellar surface also slows
down, but only by about 20%. Thus, the translational motion is less affected than the
rotational motion. This is due to the fact that the average mean-square displacement
(which measures the translational dynamics) is dominated by the fast-moving free
water molecules in the layer, while the long time slow decay of orientational
relaxation is found to be dominated by the bound water molecules.
17.3 Dynamics of water surrounding micelles 265
Hydrogen-bond lifetime analysis revealed that HBs between the polar head
groups of the micelle and the water molecules are much stronger than those between
two water molecules in bulk water and thus exhibit much slower dynamics almost
13 times slower than that of bulk water. This result indicates the presence of quasi-
bound water molecules on the surface.
More detailed analysis has looked into the fraction of differently hydrogen-
bonded species in the layer. About 6070% of the interfacial water molecules are
singly hydrogen-bonded with the polar groups of the micelle, while 2030% form
two such HBs [8]. A small fraction of the molecules do not form any HB with the
micelle and remain relatively free, although still in the hydration layer.
17.4 Free-energy landscape of hydrogen-bond arrangements
at the surface
As discussed above, the hydration layer of a micelle may contain two kinds of water
molecules: ones that are hydrogen-bonded to a polar head group (PHG) at the
micellar surface, through the PHGs oxygen, and others that are not. This hetero-
geneity allows the system to respond with multiple timescales, as required.
As discussed in Chapter 6, these two types of water can be termed bound,
denoted by IBW, and free, denoted by IFW. The bound water molecules can be
further distinguished based on the number of HBs that they make with the polar head
groups of the surfactant. The number of H-bonds can either be one or two. Therefore,
all interfacial water molecules can be classied into three types; we denote them IFW,
IBW1, or IBW2, depending on whether the interfacial water molecule has zero, one,
or two HBs with the head-group oxygens, respectively. Aschematic representation of
the three types of interfacial water molecules is shown in Figure 17.3.
Interestingly, the HBs that IBW2 forms with the polar head groups of micelles are
found to be not equivalent one is stronger than the other. Hydrogen-bonding with the
polar head groups provides extra stability for the interfacial water molecule. Thus the
stability of these three species is in the following order: IBW2 > IBW1 > IFW [8].
Simulation studies offer the following picture of the dynamics of water in the
hydration layer of micelles. Within the interfacial region, there is a constant
exchange of water molecules between the three states, IBW2, IBW1, and IFW.
The microscopic reactions between IBW2 and IBW1 on the one hand, and IBW1
and IFW on the other, are reversible, and are described by four distinct rate
constants, as described below [8].
IBW2 IBW1 I FW (17:1)
Determining these rates in such a complex system(which is also open to bulk water)
is a challenging task, although the basic theoretical formalism exists for it. For the
micelle CsPFO(the surfactant is cesium pentauorooctanoate, with bromide ions as
the counter ions), the average concentration of these species is IFW:IBW1:IBW2 =
1.1:8.0:0.9. These ratios can be different for different micelles, depending on the
microscopic structure on the micellar surface.
The ratio of the populations (IFW:IBW1:IBW2) can be used to get the free-
energy differences between the three species. A semi-quantitative free-energy
diagram is shown in Figure 17.4. Despite possessing two strong waterPHG
bonds, the concentration of the IBW2 species is rather low. This observation
indicates the signicant role of entropy in determining the free-energy differences
and thus the concentration of the species [8].
Fromthe free-energy and monomer-energy data, one nds that the entropy loss of
the IBW2 species over the IBW1 species to be 10 calmole
1
K
1
. This loss of
entropy for the IBW2 type of molecules is evidently related to the constraint due to
bonding. The IBW2 species favors more linear HBs with the PHG, and has a
signicantly different relative orientation of PHG groups in its rst coordination
shell. Additionally, water molecules surrounding an IBW2 species prefer to form a
tetrahedral geometry rather than an IBW1 species. These constraints decrease the
exibility of the IBW2 water species, thus reducing its entropy.
Figure 17.3. (a) Schematic representation of the surfactant molecule with atom
labels. (b) Schematic representation of the bonding pattern of the three types of
interfacial water molecule (namely IFW, IBW1, and IBW2). IBW1 and IBW2
types of water molecule form HB(s) with the polar headgroup of the surfactant
molecule(s). IFWmolecules, although present in the interfacial region, do not form
any HB with the surfactant and, instead, are bonded purely to other water
molecules in the vicinity. Figure adapted with permission from J. Phys. Chem. B,
17.4 Free-energy landscape of hydrogen-bond arrangements at the surface 267
Thus, to summarize, the water dynamics around micelles shows multiple timescale
relaxation behaviors with a wide range of timescales. The slowest components (with
time constants above 1 ns) are mostly due to the structural rearrangement in micelles.
The intermediate timescales (with time constants in the range 100 ps to 1 ns) arise from
relaxations due to interfacial water. The water dynamics is thus much slower than in
the protein hydration layer. This is due to the fact that water molecules are strongly
bound to the micellar head groups, which are highly polar. Computer simulation
studies have explored the hydrogen-bonded structure of the interfacial water mole-
cules, which helps one to understand the emergence of slow dynamics.
17.5 Reverse micelles and microemulsions: dynamics of water
The water dynamics inside the reverse micelles is expected to be different from bulk
water as here water molecules face connement (nanometer-sized pool) and also
specic interactions with the head groups of the surcants, depending on their nature.
Dielectric relaxation studies of water in reverse micelles provide information on the
mobility of the water molecules in the nanometer-sized pools. They show that the
amplitude of the DRin the water pool is substantially smaller than that in bulk water. It
is suggested that connement, rather than the local structure of the hydrogen-bonded
network, is responsible for the suppression of the relaxation rate and that, within the
water pool, the free water is not structurally equivalent to the bulk water. DAngelo
et al. studied the DRof AOTwatercarbon tetrachloride (CCl
4
) microemulsion in the
0.023 GHz frequency range as a function of the water-to-AOT molar ratio (0.2 < W
0
< 10). They detected a single relaxation time (about 7 ns at the lowest water content,
W
0
= 0.2) that becomes shorter with an increase in W
0
. It was proposed that for small
Figure 17.4. Schematic description of the free-energy prole of the interfacial
water species. The species are in dynamic equilibrium with themselves and with
water present in the bulk region of the micellar solution. The reaction coordinate is
arbitrary and does not imply any distance. Barrier heights are also arbitrary. Figure
adapted with permission from J. Phys. Chem. B, 107 (2003), 51945202.
reverse micelles, the slowest relaxation time represents the reorientational time of the
micelle itself. The polar head groups progressively become more hydrated with
increasing W
0
and their mobilities gradually increase. The relaxation at the highest
W
0
value is interpreted in terms of the rotational relaxation of the completely hydrated
AOT ion pairs at the head group region [9].
The slowdynamics in the water pool of reverse micelles was also explored by using
1
H-NMR spectroscopy. It was found that the rotational correlation time of water
lengthened by two orders of magnitude in the small water pools of heptaneAOT
water reverse micelles. Computer simulation studies explored the water motion in
reverse micelles of aerosol OT. They calculated the self-intermediate scattering
function F
S
(Q,t) for water hydrogens and compared the time Fourier transform of
the same with the QENS dynamic structure factor S(Q, ) and found good agreement
between the simulation and the experiment. They calculated the separate intermediate
scattering functions for rotational and translational motion. It was found that the decay
of the translational scattering function was non-exponential, indicating that this
behavior is due to lower water mobility close to the interface and also to connement-
induced restrictions on the range of translational displacements.
Several groups conducted SD experiments on the reverse micelle system. They
all found a slow component in the SD in the nanosecond timescale. The timescale
was found to depend on the radius of the water pool r
w
and molar ratio of water to
surfactant W
0
. Sarkar and co-workers studied the SD of C480 in AOTn-heptane
water microemulsions. They observed a distinct rise in the nanosecond timescale at
the red end of the emission spectra. They observed that in a small water pool (W
0
=
4, r
w
= 8 ) the solvation time was 8 ns, while for a large water pool (W
0
= 32, r
w
=
64 ) the response was bimodal with a fast component of 1.7 ns and a slower
component of 12 ns. Obviously, these studies missed all of the ultrafast solvation,
which occurs in the picosecond (or faster) timescale [10].
17.6 Orientational dynamics
Orientational dynamics inside a reverse micelle nanopool is explored using pump-
probe spectroscopy. The initial experiments have had a lower time resolution and
thus cannot measure rotational anisotropy for t < 200 fs. Thus mainly these experi-
ments provide information about the dynamics in the longer timescale. The main
results of these experiments are as follows: (1) the rotational anisotropies for the
larger reverse micelles (W
0
20) are single exponential with a decay time the
same as bulk water. (2) The rotational anisotropies for the smaller reverse micelles
(W
0
10) show an increasing slowing down of the dynamics as W
0
decreases
(Figure 17.5). These results have been understood in the framework of a two-
component core-shell model [11].
17.6 Orientational dynamics 269
17.7 Coreshell model
In reverse micelles, the water molecules in the internal pool can be divided into two
subensembles. One of the ensemble consists of those water molecules which are
near the charged head groups and involved in stronger polar interaction with head
groups. This subensemble is termed the shell as these water molecules form the
outer shell of the nano-pool water. The other subensemble contains water molecules
in the core of the reverse micelle and has dynamical character similar to those found
in bulk water. This subensemble is termed the core (see Figure 17.6).
In the presence of more than one space-separated set of molecules in the system and
alsowhenthe exchange betweenthe twosubensembles is slow, the contributions of each
(a)
(b)
Figure 17.5. Orientational anisotropy decays of the OD stretch mode in the various
samples. (a) Bulk water, W
0
= 60, 40, and 20, shows little variability in the anisotropies
of these samples. The decays are single exponential. (b) Bulk water, W
0
= 10, 5, and 2,
anisotropies reveal the progressively longer orientational relaxation timescales in the
smaller reverse micelles. The anisotropy decays of the small reverse micelles are
biexponential. Figure adapted with permission from. J. Phys. Chem. A, 110 (2006),
grouptothe signal canbe additive at a giventime andthe populationdecaycanbe simply
given by the weighted sum of the population relaxation of the two components. [12]
P(t) = a
1
P
1
(t) (1 a
1
)P
2
(t) (17:2)
Where, P
i
(t) is the population relaxation of component i and a is a weighting factor.
Clearly, the weighting factor a depends on the concentration of the chromophores, at
the given frequency range (assuming that transition dipole moment is independent
of frequency). In order to t the data, it is therefore not necessary to make any further
assumptions about the relative transition dipole moments of the ensembles, and, in
the discussion below, a can be treated as a weighting factor, signifying relative
populations in the core and shell regions.
With the above approximation, the anisotropy decay is given below
r(t) =
a I
1
|
I
1
l
_ _
(1 a) I
2
|
I
2
l
_ _
a I
1
|
2I
1
l
_ _
(1 a) I
2
|
2I
2
l
_ _
= 0:4
aP
1
(t)C
1
2
(t) (1 a)P
2
(t)C
2
2
(t)
aP
1
(t) (1 a)P
2
(t)
(17:3)
Here, the parallel and perpendicular pumpprobe signals due to component i are
denoted by I
i
|
and I
i
l
, respectively, and the respective orientational correlation func-
tions are given by C
i
2
(t). When the two components display a large separation of
timescales for both their vibrational lifetimes and orientational dynamics, then the
long-time anisotropy decay can be an accurate representation of the anisotropy of the
slowcomponent. However, both of the two components can be mixed at intermediate
times, making interpretation complicated. In such a case, a model like Eq. (17.3) can
be invoked to extract information about the orientational dynamics.
Surfactant nonpolar
tail group
Nonpolar
phase
Core
Interfacial
region (shell)
Polar phase
Surfactant polar
head group
Counterion (with ionic
head groups only)
Figure 17.6. Schematic representation of the different regions in a reverse micelle
and pictorial description of core and shell regions. Schematic gure has been
reproduced from Annu. Rev. Phys. Chem., 60 (2009), 385406.
17.7 Coreshell model 271
Equations (17.2) and (17.3) are the assumption that exchange between the core
and the shell is slow compared to the vibrational lifetime and orientational time.
Figure 17.7 depicts three simulated anisotropy decay curves with all parameters
except the weighting term, a, xed. The parameters were so chosen that they
represent bulk water and interfacial water. The vibrational lifetime and orientational
relaxation time of component 2 were set to the bulk-water values, T
2
= 1.8 ps and
r
2
= 2.6 ps. The vibrational lifetime and orientational time of component 1 were set
at T
1
= 4.5 ps and
r
1
= 20 ps.
From Figure 17.7 it is apparent that at long times the curves coalesce to a single
decay, showing that after the fast vibrational lifetime component has decayed the
anisotropy decay simply represents the orientational dynamics of the slowcomponent.
However, at intermediate times, the anisotropy can behave strangely, appearing to
reach a plateau value for a = 0.3 and even turning up for a = 0.05. We discuss this
interesting fact in more detail below as it has a signicant (and general) bearing on
interpretation of anisotropy decays measured by IR spectroscopy.
The signal-to-noise ratio of the anisotropy data begins to deteriorate after 8 ps (the
position of the dashed vertical line in Figure 17.7). This is despite the relatively
longer vibrational lifetime of the OD stretch of interfacial water molecules, and
particularly so when a signicant amount of the water in a reverse micelle is bulk-
like. Without the long-time decay (>14 ps) to indicate otherwise, the data could be
interpreted (wrongly) as reecting a signicant portion of water molecules (greater
a
a
a
Figure 17.7. Plots of anisotropy decays of a two-component system with calculated
time constants T
1
= 4.5 ps,
r1
= 20 ps, T
2
= 1.8 ps, and
r2
= 2.6 ps. The three plots
correspond to three different weighting terms, a
i
, as shown. While the long-time
anisotropy decays coalesce, the behavior of the curves at intermediate times is not
intuitively clear. The dashed vertical line shows the usual maximum time for
experimental data collection. When the anisotropy decays are only analyzed to this
point, it appears as though there is a component that does not undergo orientational
relaxation but that is incorrect. Figure adapted with permission from J. Phys. Chem.
B, 113 (2009), 85608568. Copyright (2009) American Chemical Society.
than 50% for a = 0.3) as nonrotating water molecules, even when 70% of the water
has bulk characteristics. As highlighted in Figure 17.5, this interesting fact brings out
a limitation of infrared spectroscopy when two systems contain two subensembles of
molecules that have different vibrational lifetimes and orientational dynamics in
such a case, the anisotropy is not a reliable probe of orientational dynamics.
17.8 Distance-dependent relaxation near the core of the reverse micelle:
propagation of surface-induced frustration
Connement brings novel features into the dynamics of water in reverse micelles.
Markedly non-exponential decay exists due to two different behaviors, near the
surface and inside the pool, at least for intermediate to large reverse micelles
(W
0
> 4). For larger sizes, relaxation changes from the slow behavior near the
surface to the fast relaxation at the pool [5].
Because of the presence of strong interactions of water molecules with the polar/
charged surface groups located inside the reverse micelles, the relaxation of water
molecules is slow near the surface (see Figure 17.8). As we move towards the center,
the relaxation of water naturally becomes faster. However, in some cases of intermedi-
ate size there is a possibility that water relaxation can be faster than in the bulk. Such a
situation has also been observed in water relaxation at the surface of lipid bilayers.
In the case of reverse micelles there is no conclusive signature of such an effect.
Computer simulation, however, does indeed show an initial relaxation of water that
is faster than in the bulk.
17.9 Ising model description of the dynamics
It has been proposed recently that the faster than bulk water relaxation observed is due
to frustration induced by the propagation of opposite correlations fromthe interior of the
micellar surface towards the center of the water pool. This can be easily understood by
employing a variant of the kinetic Ising model that was introduced recently in order to
model this effect of nano-connement on the orientational dynamics of water inside the
reverse micelles. The model assumed that the two spins at the two ends of the one-
dimensional chain remained xed in opposite directions. This mimics the orientation of
water molecules xed at diametrically opposite positions in the interior of reverse
micelles. This can be made clear by Figure 17.9(a) [13].
The one-dimensional Ising model also revealed the emergence of multiple time-
scales in the orientational dynamics as the chain length increases. For small to
intermediate-sized chains, the orientational dynamics of spins at the center acquired
a decay component which is faster than in the bulk (see Figure 17.9(b)). This rapid
decay component is a result of the cancellation of the polarization caging
17.9 Ising model description of the dynamics 273
propagating from the two ends of the chain. This result provided a simple explana-
tion of the acceleration of orientational decay observed for water molecules in the
central pool of the reverse micelle by Fayer and co-workers.
17.10 Conclusion
Water in conned systems (discussed here and also in the next chapter) exhibits
features markedly different from those observed in bulk water. While such studies
are important in their own right, they also are relevant for biological systems
where water molecules are always restricted and/or conned. Reverse micelles in
particular provide a good model to study the effects of such connement.
Note that the study of molecular aspects in conned systems has become possible
only in recent years. These studies require the development of time domain techni-
ques such as quasi-electron neutron scattering (QNS) and laser spectroscopy, on one
hand, and the potential of computer simulation on the other. As we articulated in this
chapter, the synergy between theories and experiments played a key role in the
development of our understanding of this area.
Figure 17.8. Semilog plot of normalized reorientational time-correlation function of
the unit vector along the OHbond of the water molecules in different layers. Layers
are represented as follows: bulk water, double dotted dashed line; central layer, single
dotted dashed line; intermediate layer, dashed line; surface layer, solid line. The
faster than bulk relaxation of the surface and intermediate layers is evident in the
reverse micelle. At intermediate times a crossover is observed and the surface layer
has a pronounced long-time decay component. The faster long-time decay for the
intermediate layer compared with the central water layer is also observed. Figure
adapted with permission from J. Chem. Phys., 137 (2012), 01451519. Copyright
(2012) American Institute of Physics.
References
1. M. Gehlan and F. C. DeSchryver, Time-resolved uorescence quenching in micellar
assemblies. Chem. Rev., 93 (1993), 199221.
2. H. H. Paradies, Shape and size of a nonionic surfactant micelle. Triton X-100 in aqueous
solution. J. Phys. Chem., 84 (1980), 599607.
3. N. Nandi, K. Bhattacharyya, and B. Bagchi, Dielectric relaxation and solvation dynamics
of water in complex chemical and biological systems. Chem. Rev., 100:6 (2000),
20132046.
4. T. Telgmann and U. Kaatze, On the kinetics of the formation of small micelles. 1.
Broadband ultrasonic absorption spectrometry. J. Phys. Chem. B, 101 (1997),
77587765.
5. R. Biswas, T. Chakraborti, Biman Bagchi, and K. G. Ayappa, Non-monotonic, distance-
dependent relaxation of water in reverse micelles: propagation of surface induced
frustration along hydrogen bond networks. J. Chem. Phys., 137 (2012), 01451519.
6. N. Sarkar, A. Datta, S. Das, and K. Bhattacharyya, Solvation dynamics of coumarin 480
in micelles. J. Phys. Chem., 100 (1996), 1548315486.
7. C. D. Bruce, S. Senapati, M. L. Berkowitz, L. Perera, and M. D. E. Forbes, Molecular
dynamics simulations of sodium dodecyl sulfate micelle in water: the behavior of water.
J. Phys. Chem. B, 106 (2002), 1090210907.
(a)
(b)
Figure 17.9. (a) A schematic illustration of the one-dimensional Ising chain model
with the spins at the boundary having xed orientation in opposite directions.
(b) The semilog plot of q
i
(t) vs. time for various spins with the given notations: the
central spin (the middle one) for different chain lengths. Figure adapted with
permission from J. Chem. Phys., 133 (2010), 08450917. Copyright (2010)
American Institute of Physics
References 275
8. S. Pal, S. Balasubramanian, and B. Bagchi, Identity, energy, and environment of
interfacial water molecules in a micellar solution. J. Phys. Chem. B, 107 (2003),
51945202.
9. M. DAngelo, D. Fioretto, G. Onori, L. Palmieri, and A. Santucci, High-frequency
dielectric properties of aerosol sodium bis-2-ethyl-hexylsulfosuccinate (AOT)H
2
O
CCl
4
systems in the reversed micellar phase. Phys. Rev. E, 52 (1995), R4620R4623.
10. S. Das, A. Datta, and K. Bhattacharyya, Deuterium isotope effect on 4-
aminophthalimide in neat water and reverse micelles. J. Phys. Chem. A, 101 (1997),
32993304.
11. I. R. Piletic, D. E. Moilanen, D. B. Spry, N. E. Levinger, and M. D. Fayer, Testing the
core/shell model of nanoconned water in reverse micelles using linear and nonlinear
IR spectroscopy. J. Phys. Chem. A, 110 (2006), 49854999.
12. D. E. Moilanen, E. E. Fenn, D. Wong, and M. D. Fayer, Water dynamics at the interface
in AOT reverse micelles. J. Phys. Chem. B, 113 (2009), 85608568.
13. R. Biswas and B. Bagchi, A kinetic Ising model study of dynamical correlations in
conned uids: emergence of both fast and slow time scales. J. Chem. Phys., 133
(2010), 084509-1-7.
18
Water in a carbon nanotube: nature abhors a vacuum
Water inside a carbon nanotube (CNT) shows another set of unusual
features which have been the subjects of great interest in recent years and
are currently being studied extensively by experiments and simulations. A
remarkable aspect unearthed is the ability of a CNT to act as water
transporter and lter. The properties of water inside a CNT depend on
the diameter d of the CNT, which is analogous to the parameter W
0
of reverse micelles discussed in the preceding chapter. However, water
within a single-wall CNT shows unusual features, such as single-le
diffusion, the theory of which was developed in the past but a proper
model system was lacking. The orientational motion of individual water
molecules exhibits slow dynamics, quite different from those in the bulk.
18.1 Introduction
The cavity inside a CNTis strongly hydrophobic. Yet when a CNTis kept immersed
inside water, water molecules easily pass through the inner channel of the CNT. In a
narrow single-wall CNT (SWCNT), the motion of water molecules occurs as a
single le. Actually, the single-le diffusion (SFD) model was proposed many years
ago but nds new applications in this problem. Water within SWCNTs exhibits
amazingly fast transport the origin of which is beginning to be understood. There is
hope that these properties can be harnessed for practical use in the near future. In the
following we discuss some of the properties of water inside a CNT but we shall not
address other properties (such as optical) of CNTs themselves.
18.2 Type and structures of carbon nanotubes
There are two types of CNT, single-walled and multi-walled, abbreviated as
SWCNT and MWCNT respectively. SWCNTs can be created by wrapping a one-
atom-thick layer of graphite, called graphene, into a seamless cylinder whereas
277
MWCNTconsists of multiply rolled graphene sheets. The way the graphene sheet is
wrapped is represented by two integers n and m and abbreviated as (n,m). These
integers represent the number of unit vectors along two directions in the honeycomb
crystal lattice of graphene. If m = 0 then it is called zigzag CNT, whereas if m = n
it is called armchair CNT, and others are called chiral CNT. The diameter of an
SWCNT can be obtained by the following formula:
d =
a
n
2
nm m
2
( )
_
(18:1)
where a = 0.246 nm.
18.3 Structure of water inside a carbon nanotube
Both experiments and computer simulation studies show that nanoconned water has
an interesting structure where molecules are ordered as in a one-dimensional solid (see
Figure18.1(a)). Simulations by Rasaiah and co-workers showed that the strongly
hydrophobic cavity of CNT, initially empty when immersed in water, becomes lled
rapidly with water [1]. This computer simulation study used a single-walled CNT that
was 13.4 long with a diameter of 8.1 . Each water molecule was found to lose two
HBs on average with respect to bulk water. The remaining HBs were highly oriented.
Only 15% of the remaining HBs sustain an HOH angle more than 30. This is
considerably lower than the percentage observed in bulk water (37%). The chain-like
network of water inside the CNT is essentially one-dimensional. Ab initio simulations
(a) (b)
Figure 18.1. Snapshot of the water arrangement from two different visual angles
inside the CNT. (a) Structure of the hydrogen-bonded water network inside the CNT
(b) Preferential alignment of encapsulated water wire was obtained from ab initio
simulation. Figure 18.1(b) is reproduced from Phys. Rev. Lett., 90 (2003), 195503.
278 Water in a carbon nanotube: nature abhors a vacuum
based on density functional theory also conrmed the formation of water wire inside the
CNT such that the direction of the dipole of each water molecule is directed along the
nanotube axis (see Figure 18.1(b)) [2].
As the diameter of the CNT increases, the number of water molecules that can be
incorporated inside also naturally increases and the arrangement no longer remains
like that of a single chain of water. Jun and Yan found that the average number of
water molecules inside an SWCNT(6,6) is 5.0 and they forma single le [3] and that
for an SWCNT(7,7) is 7.7. But SWCNT(8,8) and (9,9) (see Figure 18.2), which
have larger diameters than the previous ones, have 17.0 and 23.1 water molecules
inside the CNT on average and they do not form a single le. The distribution of
water inside a CNT is wavelike with minimal values at their openings and the
pattern remains unchanged with increasing diameter of the tube.
Water inside the SWCNT has also been characterized experimentally
using
1
H-NMR [4]. Water trapped inside a CNT can be identied by a relatively
broad peak which is slightly shifted up fromthe bulk water value because of the lesser
number of HBs in conned water. The spin-spin relaxation time of conned water is
signicantly less than that of bulk water, causing a broad
1
H-NMR spectrum.
18.4 Dynamics and transport of water
18.4.1 Translational motion of water inside a CNT
It was found both experimentally and from simulations that CNTs are fast transpor-
ters of water [1]. Rasaiah and co-workers showed from their MD simulations that
Figure 18.2. Distribution of water along the nanotube length axis (Z) for different
radii. Figure adapted with permission from Chin. Phys. B, 2007, 16 (2007),
335339. Copyright (2007) IOP Science.
during a 66 ns run, as many as 1119 water molecules enter the nanotube from one
end and leave from the other. This amounts to 17 water molecules per nanosecond
passing through that hydrophobic nano-channel on average. This is equivalent to a
value 0.53 10
6
cm
2
s
1
of diffusion constant, by assuming one-dimensional
diffusion along the rod. Although this value is much smaller than the self-diffusion
constant of water, this rate of water transport is several orders of magnitude faster
than that expected from conventional uid owtheory. As mentioned, experimental
evidence also supports such fast transport [5,6].
Conduction of water through a CNT was found to occur in pulses. While the
reason for such behavior is not fully understood, it can be attributed to the chain-like
water arrangement inside a CNT. As the CNT wall is strongly hydrophobic, water
molecules move smoothly without any hindrance from the wall, in a single le.
Such one-dimensional motion of particles through a narrow pore such that two
particles cannot cross each other is actually well-known in the literature and is termed
single-le diffusion. In the case of normal diffusion the mean-square displacement of
particles is proportional to time, whereas in single-le diffusion it is proportional to the
square root of time. The single-le diffusion mode was recently conrmed experi-
mentally by an NMRstudy of water conned in a CNTof diameter 1.4 . It was found
that the mean-square displacement is indeed proportional to the square root of time
from the pulse-eld gradient (PFG) NMR measurements (see Figure 18.3).
Similar results were obtained from simulation of water molecules conned in
narrow (6,6) carbon nanorings. Simulation also shows that at least two or more
water chains should be present inside the nanotube or nanoring to exhibit single-le
diffusion [7,8].
As already mentioned, single-le diffusion as a model was proposed a long time
ago to describe the diffusion of interacting particles (atoms, electrons) in a one-
dimensional system [9]. Detailed theoretical and experimental studies on model
systems show intermittent displacements and time-dependent diffusion [10]. Many
of the results can be explained by using a theoretical method called mode-coupling
theory. The theory shows that a more microscopic understanding can be achieved by
studying the velocity time correlation function of the particles. The latter is found (in
theory and simulations) to show surprising system-size dependence. This should be
possible to check by studying water molecules in CNTs. Another interesting
theoretical prediction involves the dramatic effects of a background noise on the
diffusion process in a one-dimensional system.
18.4.2 Rotation of water molecules within a CNT
Computer simulation studies reveal that the reorientational dynamics of water inside
a CNT is anisotropic [11] and the average orientational relaxation time of the dipole
moment of a water molecule is three orders of magnitude longer than that of bulk
water. Such relaxation happens due to diffusion of hydrogen-bonding defects across
the single-le water chain inside the nanotube. Relaxation time increases with the
length of the nanotube [11].
The standard reorientational correlation function of the HHvector of nano-conned
water was calculated by MD simulations. It was found that the short-time and long-
time correlation functions are of the orders of a few hundreds of nanoseconds in both
nanoring and nanotube whereas that of bulk water is 2 ps (see Figure 18.4). Fast HH
relaxation in nano-conned water is a result of the fact that the hydrogen of water
which is not participating in the hydrogen-bonding network of the single-le chain
inside the CNT is free to rotate without disturbing the existing HBs, which is not
possible in bulk water [12]. Such single-le arrangement of water molecules is
illustrated in Figure 18.5.
Due to the tetrahedral arrangement of water its rotational freedom is limited.
Also such bound hydrogen and free hydrogen can exchange their positions by the
well-known large-amplitude rotational jump mechanism with a time interval of
100 fs [13]. In bulk water the dipole, OH, and HH vector relaxation timescales
are 2.05, 2.3, and 2 ps, respectively. But in conned water such relaxation times
10 100 1000 10000
t (ps)
0.01
0.1
1
10
100
<

2
(
t
)

>

(
r
a
d
2
)
t
0.5
t
80 molecules
40 molecules
1 molecule
Figure 18.3. MSDof conned water molecules in a nanoring where two oppositely
polarized chains are stable. Both systems of 40 molecules and 80 molecules show
SFD whereas a system of one water shows normal diffusion. Figure adapted with
permission from ACS Nano, 4 (2010), 985991. Copyright (2010) American
Chemical Society.
change dramatically with the dipole relaxing extremely slowly with a timescale
of several nanoseconds, whereas the HH vector relaxes relatively fast.
Relaxation of the OH vector shows intermediate characteristics with a timescale
of 30 ps.
18.5 Nanotubes as a ltration device
As CNTs are found to be fast transporters of water, it has been proposed that these
nanotubes may be used as a water ltration device. This is indeed an interesting
possibility because CNTs can be produced easily and rather cheaply.
Figure 18.5. Single-le arrangement of water molecules inside a CNT. One of the
hydrogens of each water molecule is not participating in the H-bond network and is
therefore free to move.
Figure 18.4. Anisotropy of the reorientational relaxation of dipole, OH, and HH
vector for nano-conned water. Such anisotropy though present in bulk water is
very mild. Figure adapted fromACS Nano, 2 (2008), 11891196. Copyright (2008)
Maniwa and co-workers have shown that waterSWCNTs can be used as a
new type of molecular nanovalve [14]. They carried out a systematic investiga-
tion of waterSWCNTs in different gaseous environment below 0.1 MPa using
electrical resistance, X-ray diffraction, NMR measurements, and MD simula-
tions. They found that the resistivity of waterSWCNTs exhibits a signicant
increase in gas atmospheres below a critical temperature at which a particular
type of gas molecule can enter the SWCNT in an on-and-off fashion.
18.6 Conclusion
Unusual features observed inside CNTs may be considered one of the most recent
examples where the anomalous properties of water again surprised us. However,
some of the properties (such as the one-dimensional quasi-ordered arrangement) are
not hard to understand from the unique features of water molecules listed in
Chapter 1. It is somewhat harder (but not terribly hard) to understand the transport
of water across the nanotube. To an extent this is a manifestation of the well-known
adage nature abhors a vacuum. Continuous pressure uctuations of water push
water molecules into the nanotubes. However, a quantitative explanation does not
appear to exist yet.
References
1. G. Hummer, J. C. Rasaiah, and J. P. Noworyta, Water conduction through the hydro-
phobic channel of a carbon nanotube. Nature, 414 (2001), 188190.
2. D. J. Mann and M. D. Halls, Water alignment and proton conduction inside carbon
nanotubes. Phys. Rev. Lett., 90 (2003), 195503.
3. L. Hang-Jun and Z. Xiao-Yan, The structure and dynamics of water inside armchair
carbon nanotube. Chin. Phys. B, 16 (2007), 335339.
4. Q. Chen, J. L. Herberg, G. Mogilevsky, et al., Identication of endohedral water in
single-walled carbon nanotubes by 1H NMR. Nano Lett., 8 (2008), 19021905.
5. J. K. Holt, H. G. Park, Y. Wang, et al., Fast mass transport through sub-2-nanometer
carbon nanotubes. Science, 312 (2006), 10341037.
6. M. Majumdar, N. Chopra, R. Andrews, and B. Hinds, Nanoscale hydrodynamics:
enhanced ow in carbon nanotubes. Nature, 438 (2005), 44.
7. A. Das, S. Jayanthi, H. S. M. V. Deepak, et al., Single-le diffusion of conned water
inside SWNTs: an NMR study. ACS Nano, 4 (2010), 16871695
8. B. Mukherjee, P. K. Maiti, C. Dasgupta, and A. K. Sood, Single-le diffusion of water
inside narrow carbon nanorings. ACS Nano, 4 (2010), 985991.
9. J. L. Lebowitz and J. K. Percus, Kinetic equations and density expansions: exactly
solvable one-dimensional system. Phys. Rev., 155 (1967), 122138.
10. S. Pal, G. Srinivas, S. Bhattacharyya, and B. Bagchi, Intermittency, current ows, and
short time diffusion in interacting nite sized one-dimensional uids. J. Chem. Phys.
116 (2002), 5941.
References 283
11. B. Mukherjee, P. K. Maity, C. Dasgupta, and A. K. Sood, Strongly anisotropic orienta-
tional relaxation of water molecules in narrow carbon nanotubes and nanorings. ACS
Nano, 2 (2008), 11891196.
12. H. Kumar, B. Mukherjee, S. T Lin, C. Dasgupta, A. K. Sood, and P. K. Maiti,
Thermodynamics of water entry in hydrophobic channels of carbon nanotubes.
J. Chem. Phys., 134 (2011), 124105.
13. D. Laage and J. T. Hynes, Amolecular jump mechanismof water reorientation. Science,
311(2006), 832835.
14. Y. Maniwa, K. Matsuda, H. Kyakuno, et al., Water-lled single-wall carbon nanotubes
as molecular nanovalves. Nat. Mater., 6 (2007), 135141.
Part IV
Advanced topics on water
19
The entropy of water
The entropy of water under ambient conditions ( T = 298K and
atmospheric pressure) is equal to 16.8 calK
1
mol
1
. This amount
includes contributions from translational, rotational, and vibrational
degrees of freedom. The contribution of the vibrational degrees of free-
dom is negligibly small, and that of the translational degrees of freedom
is approximately three times that of the rotational degree of
freedom. These degrees of freedom of water molecules get quenched
differently in the protein hydration layer and in the grooves of
DNA. Unfortunately, accurate determination of the entropy of a water
molecule is often hard, although it is required in many applications such
as drug and ligand binding to proteins and DNA and complex structure
formation in aqueous solution. In complex systems, entropy acquires a
dynamic character. First, the diffusion or mobility of molecules is
often related to the entropy of the region. An example is provided by
the celebrated AdamGibbs relation, which provides an intriguing
relationship between diffusion (or viscosity) and the congurational
entropy of the liquid. Another example is provided by the
Rosenfeld scaling relation. Second, entropy may vary from region to
region, making the residence time of a molecule in that region a para-
meter of the entropic description. We have discussed this scenario
in the DNA hydration chapter. The issue then boils down to the
conguration space that is accessible to water molecules in a given
time. In this chapter, we shall discuss some of the aspects of entropy
of water.
19.1 Introduction
Entropy is an important thermodynamic property of a system and we depend on
entropy for much of our understanding of physical, chemical, and biological
phenomena. However, it is hard to calculate, even in computer simulations. The
entropy of a systemis highest in the gas phase and lowest in the solid phase. Entropy
287
has been fundamentally understood as a measure of disorder of the system, a
concept that originated from the Boltzmann law, which relates entropy (S) to the
number of microscopic states () of the system [1].
S = k
B
ln (19:1)
where k
B
is the Boltzmann constant, which is a universal constant with a value equal
to 1.38066 10
23
J/K = 8.617385 10
5
eV/K.
Recently, however, there has been an argument in favor of dening entropy in
terms of dispersal of energy at a given temperature [2]. This newer approach
provides a measure of entropy change in terms of distribution or spreading out of
the energy of a thermodynamic system, divided by its temperature. Although there
could be different views on the interpretation of entropy, we still have to rely on
equilibrium statistical mechanics to calculate the entropy of any system where
molecules interact strongly among each other. Unfortunately, this is also highly
non-trivial.
Despite the difculty, there are certain denite results that we can count on.
Statistical mechanics gives us an exact expression for translational entropy of an
ideal gas, which is known as the SackurTetrode equation, which shows that
even for an ideal gas [3] this expression is non-trivial, as discussed in Appendix
19.A. At the limit of high density of a liquid, the hard sphere provides a good
model for the structure. The entropy of a system of hard spheres (or billiard balls)
is much less than that of an ideal gas. This is due to excluded-volume effects,
as there is a limited amount of space free for a given atom to move around at
high density. It would be very difcult to perform a direct calculation of the
entropy of an ideal gas. Fortunately, we know how to calculate analytically
the entropy of a system of hard spheres, as we know the free energy by solving
an integral equation, called the PercusYevick equation, for the structure of
the liquid.
When atoms or molecules contain an attractive part in an intermolecular inter-
action, the magnitude of the potential energy due to interaction between the
particles increases with an increase in density or lowering of temperature, and
the entropy of the system, at xed volume, decreases. While conceptually easy to
understand, it is difcult to treat the attractive and repulsive parts together to
obtain both enthalpy and entropy at the same level of approximation. Ironically, it
is often easier to obtain the free energy of the systemby using the newtechnique of
statistical mechanics.
A signicant starting point is provided by the third law of thermodynamics,
which states that the entropy of a perfectly crystalline solid is zero at the absolute
zero of temperature. The third law allows one to estimate entropy by using the
following thermodynamic relation
288 The entropy of water
S(T) = S(T
0
)
_
T
T
0
C
P
T
dT (19:2)
Implementation of this exact relation depends on the accurate measurement of
specic heat over the entire temperature range. The absolute value of entropy
requires use of the third law of thermodynamics and the building up of entropy
from the absolute zero of temperature. Important insight can be obtained if we
compare Eq. (19.1) with Eq. (19.2). The latter denes specic heat in terms of
uctuation in entropy, at constant temperature and pressure.
In the case of liquid water, due to its extensive hydrogen-bonding, there are
two intriguing aspects that need to be considered. First, the specic heat of
liquid water is high, making the value of the entropy large, according to
Eq. (19.2). On the other hand, there is substantially more positional and orienta-
tional local order in water than in many other liquids (which do not sustain such
a hydrogen-bonding network), which could lead to a lower value of entropy,
according to Eq. (19.1).
The reason for the high specic heat of water is the existence of low-frequency
vibrational modes such as hindered translation (centered on 50 cm
1
), intermol-
ecular OO HB excitation (positioned around 190 cm
1
), and librations (centered
on 685 cm
1
). Note that the average thermal energy (measured in terms of transla-
tion modes) is given by (3/2)k
B
T, which amounts to (in the same dimension)
about 300 cm
1
. This is interesting! The intermolecular vibrational modes of
water (sometimes called collective modes) are close to and therefore accessible to
the system due to its own thermal energy! These low-frequency modes that are
essentially due to the HBnetwork all contribute to give rise to high values of entropy
and specic heat.
The entropy of water is a sum of all its vibrations (intramolecular and intermol-
ecular) and also rotations, represented as
S = S
trans
S
rot
S
vib
S
elec
(19:3)
Another important point here is that a state can contribute to entropy only when it
can be accessed within the time span of the experimental measurement of entropy.
This point needs a moment of pause and thought. Entropy is obtained by combining
the third law of thermodynamics and the specic heat of the liquid (given by Eq.
(19.2) above). The value of entropy at a given temperature is then obtained by
adding together the incremental increase with temperature by using the temperature-
dependent specic heat data.
The entropy of water in the ideal gas state is about 31 calK
1
mol
1
, under
ambient conditions (T = 298 K and atmospheric pressure). On the other hand, in the
liquid state it is equal to 16.8 calK
1
mol
1
. Both amounts quoted above include
contributions from translational, rotational, and vibrational degrees of freedom. The
contribution of the vibrational degrees of freedom is negligibly small. In the ideal
gas state, the contribution of the translational degrees of freedom is about twice that
of the liquid state. But in the liquid, the contribution of the translational degrees of
freedom is approximately three times that of rotational degree of freedom. This is
because extensive hydrogen-bonding in liquid water quenches the rotational
degrees of freedom more than the translational degrees of freedom. These degrees
of freedomof water molecules become quenched differently in the protein hydration
layer and in the grooves of DNA. In addition, the residence time of a water molecule
in those surfaces may also affect our description of the entropy in these regions.
Entropy and time can be intimately related in more than one way!
Let us say that it took 10 minutes to measure the specic heat at a
given temperature of 1 kg of water. Since we obtain entropy from specic heat, it
is the same amount of time that is available for the system to access all the states.
Here enters an important concept of statistical mechanics, which is the principle of
equal a priori probability. Equal a priori probability states that the probability
of the system being in any of the congurations of equal energy is equally
likely. The number of energy states is huge and it grows with the number of
particles in the system as 10
aN
where N is the number of particles in the
system and a is a constant with a value of the order of unity. Now, at normal
temperature, it may take 1 picosecond (1 picosecond = 10
12
s) to sample distinct
congurations. Then in 10 minutes, the system samples only 6 10
14
distinct
congurations. This is far less than the total number of states the system has for
1 mole of the substance.
Fortunately, however, the states that are sampled are the most probable states of
the systemwhich make the maximumcontribution to the thermodynamic quantities.
Thus, we obtain the average values (entropy, specic heat) of the thermodynamic
quantities with virtually no error. However, to achieve that kind of accuracy, our
systemstill needs to sample a large number of representative states. The sampling of
such a large number of states takes a longer time (due to slow relaxation dynamics)
at low temperatures, and the utmost care is needed.
Water in a restricted space, such as on the surface of micelles, or in the grooves of
DNA or inside nanotubes, is stabilized by energy. The entropy of water in such
systems is lower than that of free water. This gain of energetic stabilization at the
cost of losing the entropy of the system is known as the energy-entropy balance.
This phenomenon of energy-entropy balance is all-pervasive in science [4], but in
complex systems time becomes a relevant coordinate because the entropy and
enthalpy of a molecule might be realized at a different rate. This is clearly a tricky
concept to deal with.
19.2 Relation between entropy and diffusion
The diffusion constant of a liquid is a fundamental dynamic property which not only
is a measure of the uidity of the system but also is involved in many theoretical
descriptions that involve molecular processes such as chemical reactions. An intri-
guing relationship between diffusion constant and entropy has become a subject of
increasing scientic discussions in recent times. This has the potential for offering
insight into the nature of both, especially that of a correlation between the structure
and the dynamics.
While Boltzmanns law related entropy to the number of microscopic states, this
law is strictly valid only at constant energy where the ergodicity of the system
ensures that given sufcient time the system visits all the microscopic states.
Experimental systems are, however, not at constant energy or volume, so
Boltzmanns law, however elegant, is not directly useful to us.
There are two well-known expressions that are used for relating the entropy of a
liquid to the diffusion constant and both have been used extensively for water. The
rst was pioneered by Rosenfeld [5] and is known as the diffusionentropy scaling
relation, while the second is the celebrated AdamGibbs relation [6], which was
derived in 1965 to describe the apparent singularity in the temperature dependence
of the viscosity of a supercooled glass-forming liquid as the glass transition tem-
perature is approached from higher T. As mentioned above, both expressions have
been employed to describe diffusion in water.
19.2.1 Diffusionentropy scaling relation: the Rosenfeld relation
The Rosenfeld relation between scaled diffusion and excess entropy is dened
below [5]:
D
+
= a:e
bS
ex
(19:4)
where the dimensionless diffusion constant D
*
is dened by
D
+
= D

1=3
k
B
T=m ( )
1=2
(19:5)
S
ex
is the excess entropy (per particle) and is dened by
S
ex
= S S
id
(19:6)
S is the entropy per particle of the system; a and b are constants that depend weakly
on the material (here liquid) studied. They are often considered as universal
constants. In Eq. (19.5) m is the mass of a single water molecule (or whatever
species is being considered).
Although there does not appear to exist any derivation of this relation from
rst principles, the relation has been found remarkably successful in correlating
the diffusion constant with the entropy, and one often nds a master plot when D
*
is plotted against excess entropy. The excess entropy is usually calculated by
integrating the equation of state using the following thermodynamic relation
S
ex
= (U F
ex
)=T; (19:7)
where we have used the fact that the internal energy U
id
of an ideal gas is zero. The
excess free energy is obtained from the equation of state as discussed above
F
ex
=
_
d
P(; T) k
B
T
2
(19:8)
Evidence from computer simulations shows the correlation between D
*
and excess
entropy, obtained by Goel et al. [7]. The excellent agreement demonstrates the
validity of the Rosenfeld scaling relation.
The microscopic origin of the Rosenfeld relation between D
*
and S
ex
can be
understood in the following way. Let us assume k
m
is a microscopic relaxation rate
between two adjacent microscopic states in the phase space of the system. During
the experimental time
obs
, the system makes k
m
obs
transitions and therefore visits
that many microscopic states. So, we can set k
m
obs
proportional to the total number
of microscopic states so that the states are uniformly distributed. In the next step
we assume that each transition results in a displacement a of the particle. So, we
have the following relation between diffusion and the rate of transitions as,
D k
m
a
2
(19:9)
We now need to determine k
m
. Note that in a large system the number of micro-
scopic states is very large. So, it is unlikely that the system visits the same micro-
scopic state more than once. When the observation time is sufciently large, then the
product
obs
k
m
is proportional to the density of states of the system which in turn is
proportional to the total number of states for a conservative system. Thus, the rate k
m
of transition between the states in an average description can be equated to the
number of congurations in the phase space available for the system,
k
m
(19:10)
Now, according to the Boltzmann denition of entropy,
S = k
B
ln
= k
B
ln Bk
m
( )
= k
B
ln B=C ( )D=A
2
_
= k
B
ln (D+)
(19:11)
Here D
+
= B=C ( )D=A
2
is the scaled diffusivity; B and C are the constants of propor-
tionalities. This provides us with a relation between entropy and scaled diffusivity,
D+ = exp S=k
B
( ) (19:12)
which leads to the Rosenfeld relation. This is, however, a purely heuristic deriva-
tion, meant to show how the exponential dependence of diffusion on entropy can
appear from a simple random-walk exploration of the density of states.
A comparison of Eq. (19.2) with the Rosenfeld expression gives the values of the
Rosenfeld parameters as b = 1 and a = (B/CA
2
). Now, tting the Rosenfeld relation
to experimental results indeed gives values of b close to unity and also nearly
invariant from system to system.
We discussed the above derivation to bring out the essence of the Rosenfeld
scaling relation, which is valid when a system is ergodic with fast transitions
between the congurational states of the system, so that the diffusion coefcient
increases with entropy.
19.2.2 The AdamGibbs relation
This celebrated relation is given by
D = Ae
B=TS
c
(19:13)
where S
c
is the congurational entropy (per particle) of the liquid [6]. This formwas
derived in a heuristic way by Adam and Gibbs. Later a different, more microscopic
derivation was offered by Xia and Wolynes [8].
We need to state rst that Eq. (19.3) was derived to describe relaxation in glassy
polymers. A rapid increase in viscosity (and concomitant increase in relaxation
time) is envisaged to occur due to non-availability of congurational states that a
given state can make a transition to and thereby give rise to relaxation of stress. As a
result of this dearth of congurational states, relaxation involves rearrangement of
larger and larger regions as the temperature is decreased towards the glass transition
region.
In the derivation of Adam and Gibbs, the concept of cooperatively rearranging
region (CRR) was introduced and made use of. CRR is the minimum volume that
must participate or change for a relaxation to occur. In the AdamGibbs derivation,
this was taken as the volume that contains at least two conguration states so that
transition from one to the other can occur to give rise to relaxation. Thus, as the
conguration entropy of the system decreases, the length of this cooperatively rearran-
ging region increases. Adam and Gibbs showed that the transition probability varies as
exp zD=k
B
T ( ), where z is the size of the CRR and is the chemical potential
difference between the regions where relaxation can occur from the overall chemical
potential of the system. One needs a critical size z
*
for relaxation to occur because we
need at least two states. Adam and Gibbs were able to show
z
+
= N
A
S
+
c
=S
C
; (19:14)
where N
A
is Avagadros number, S
C
is molar entropy, and S
+
c
is a small critical
entropy of CRR whose value was left undened.
The derivation of the AdamGibbs expression by Xia and Wolynes [8] is rather
interesting and we briey discuss it here as it is quite different. In the XiaWolynes
treatment, one nds a nucleation barrier for the glass-to-liquid transition by using
the classic nucleation theory, which writes the free energy of a nucleus as a sum of a
bulk term and a surface term. The difference is that in the glass-to-liquid transition,
the surface tension can become a function of the size of the nucleus. This is a result
that follows from the random Ising model where the boundary between two phases
of the same symmetry can be wetted by states of intermediate values of the order
parameter, thus lowering the surface tension. It can be shown that the surface tension
decreases with radius R of the nucleus as

R
_
. If one assumes that the entropy of
glass is zero and that the enthalpy of the liquid and the glass are the same, then the
free-energy difference between the two phases is given by TS
C
. Application of the
standard nucleation theory then gives the following expression for the free-energy
cost of creating a nucleus of size R:
DG(R) =
4
3
R
3
TS
C
4R
2
o
=
R
_
(19:15)
Here
0
is a constant associated with the surface tension. If we now follow the
standard procedure of nding the size of the critical nucleus, which sets the
derivative
dDG(R)
dR
to zero, we recover the nucleation free barrier proportional to
(TS
C
)
1.
The main feature of the above derivation is the assumption that the surface
tension of a liquid nucleus decreases with radius as

R
_
.
It is interesting to note that the two expressions (Rosenfeld and AdamGibbs) for
diffusion give quite different dependences on the entropy. Both have been used
extensively and over a temperature range and seem to provide good agreement with
simulation results.
In Figure 19.1, tting of the simulation results of the diffusion of water to both
the Rosenfeld scaling form and the AdamGibbs form are shown [9,10]. Both the
forms give, surprisingly, equal ts to the simulated results over a large temperature
range.
Therefore, the ability of the entropy of water to provide a measure of diffusion is
well established, although the precise reason remains to be understood.
19.3 Calculation of the entropy of water
As already mentioned, quantitative estimation of the entropy of a liquid is astonish-
ingly difcult. For the solid state, one can evaluate the entropy by assuming a
harmonic oscillator model for each mode of vibration of the atoms/molecules. The
-12 -11 -10 -9 -8 -7 -6
10
-1
10
-2
10
-3
10
-4
10
-5
D
*
S
e
/ (Nk
B
)
(C) SPC/E
a=0.904
0.9
1.0
1.1
1.2
1.3
1.4
(a)
(b)
Figure 19.1. (a) Correlation plots of Rosenfeld-scaled diffusivities with excess
entropy, S
ex
, of the SPC/E water model. Data points lying along the highest and
lowest isochors are joined with smooth lines. Figure adapted with permission from
Society. (b) Semi-log plot of the diffusion constant D versus (TS
conf
)
1
for six
isochors. The double arrow denotes the range of D values where the relationship
log D versus (TS
conf
)
1
has been experimentally tested in bulk water. The lines are
provided as a guide to the eye. Figure adapted with permission from Nature, 406
(2000), 166169. Copyright (2000) Nature Publishing Group.
entropy (S) of a single harmonic oscillator can be calculated by using methods of
statistical mechanics outlined in Appendix 19.B. The nal expression is given by
S = k
B
ln 1 e
h
_ _
h
T
e
h
1 e
h
_ _
; (19:16)
This is the entropy of each mode of frequency . We need to consider all the modes.
In a solid, the total number of modes of vibration consisting of N atoms is 3N 6,
giving rise to an energy spectrum. The total harmonic entropy of the solid is
obtained by integration of Eq. (19.6) over this spectrum.
However, the entropy calculation of a liquid is not straightforward. Several
attempts have been made to calculate the entropy of water in its liquid state. Two
of them are discussed here.
19.3.1 From structure
The rst method describes the calculation of entropy from the structural order. As
we have discussed earlier in this chapter, the more ordered the system is, the less is
the entropy. This order of the system can be calculated from the MD simulation by
calculating the n-particle correlation function. According to Greens concept the
total entropy of a system can be written as a sum of two entropy terms. One of them
is the ideal gas entropy term and the other is termed the excess entropy term.
The rst is quite trivial as the ideal gas entropy is known for a state point.
The excess entropy of the system again can be written as the contribution from
the n-particle (n = 2, 3, N) entropy term which can be determined from the
n-particle correlation in the liquid structure. The calculation of the n-particle
correlation function for n > 2 is almost impossible and thus only the contribution
for n =2 (S
(2)
) is calculated, which needs the pair correlation function as input, by
using the following expression [11],
S
2 ( )
=

2
2!
2
_
g
2 ( )
ln g
2 ( )
g
2 ( )
1
_ _
dr
1
dr
2
d
1
d
2
(19:17)
where = N/V is the number density of the atoms/molecule, r
i
and
i
are the
coordinates of the center of mass and the Euler angle, determining the position
and the spatial orientation, respectively; g
(2)
is a two-particle correlation function
that depends on the positions (r
1
and r
2
) and orientations (
1
and
2
) of both
particles 12 variables in summary. If the investigated system, however, is homo-
geneous and isotropic, then the number of variables reduces, and g
(2)
becomes
a function of only six variables: the distance between particles r = r
2
r
1
, and
ve angles determining their mutual orientation. Fortunately the contribution
from the entropy term with n > 2 is negligibly small and one can ignore such
terms in the calculation. For liquid water one can calculate the pair correlation
function for position and orientation for water molecules in the system. Once
the pair correlation function is known, the simple integral equation is used to
calculate the excess entropy of the system for both positional and orientational
order.
Another important point in the entropy calculation is the time of the MD simula-
tion. It has been discussed earlier in this chapter that the entropy of the system
depends on the number of states visited by the system. The accuracy of the pair
correlation function and thus the accuracy of the entropy increase with increase in
the simulation time. One can plot the calculated entropy for different simulation
times and nd that the value of the entropy asymptotically saturates at longer times.
Thus it is important to report the total simulation time along with other details while
reporting the value of the entropy.
19.3.2 From dynamics
From the MD simulation trajectory, we can calculate the mass weighted velocity
time autocorrelation function of the center of mass translational motion. By taking
the Fourier transform of this autocorrelation function, the density of states (DOS) of
the water translational motion can be calculated. The zero-frequency DOS is
proportional to the diffusion coefcient of the liquid and it is thus nite.
However, the entropy weight function of the harmonic oscillator at zero frequency
has a singularity. Thus the harmonic approximation is not applicable here. The
newly developed 2PT method circumvents this difculty [12]. It assumes that the
DOS of liquid is a superposition of the solid and gas states. Next, one uses harmonic
approximation for the solid-state DOS to get the entropy due to the same. For the gas
part of the DOS, the hard-sphere approximation is used to get the other part of the
entropy. Addition of these two parts of the entropy gives the total translational
entropy of the liquid. Depending on the state of the system the partitioning can
be different where the fraction of the system in the gas/solid phase changes. Here
the singularity is taken care of by the gas part of the DOS where the hard-sphere
approximation is used (the hard-sphere entropy weight function is frequency-
independent and has no singularity at zero frequency). This method is quite success-
ful for the entropy calculation and it allows us to calculate other thermodynamic
properties of the liquid state which are otherwise quite difcult to obtain.
Note that the combination of solid-like and gas-like components is an alternative
way to obtain the reduction in the entropy value from that in the gas state. In the cell
theory of liquids, one uses an excluded volume to capture the constraint of the
molecules. Here the reduction is obtained by giving a lower-than-unity value to the
gas component.
The issue of the simulation time is also applicable here. One nds that this method
is relatively more efcient in terms of the simulation time. However, so far there is no
similar method to calculate rotational entropy.
The method of entropy calculation using both the structural and the dynamic
information of the systemsuggests the close relationship between the two in liquids.
Generally, when the structural order is larger, the dynamics of the system is slower
and vice versa. Thus the structure and dynamics of liquid water are coupled to each
other.
19.4 Entropy from cell theory
Recently an approach to obtain the entropy of water fromthe well-known cell theory
of liquids has been developed by Henchman [13] and provides not only quantitative
accuracy but also valuable insight into the origin of the entropy of liquid water. We
shall briey discuss the theory in the following.
He has replaced the interaction potential as a spherical harmonic of the form
u = u
c
+ (1/2)k
c
r
2
. Here u
c
is the zeroth energy, r is the distance from the equilibrium
position, and k
c
is the force constant. The communal entropy term has been omitted
as single cell occupancy is allowed.
Here the partition function Q is written as the product of N partition functions q,
as Q = q
N
. Pictorially, we have each particle moving in a spherically symmetric
harmonic potential. Thus the congurational part, Z, of the partition function can be
evaluated by integrating the Boltzmann factor of the potential in spherical polar
coordinates as
Z =
_
0
_
0
_
2
0
exp u
c
1
2
k
c
r
2
_ _
=k
B
T
_ _
r
2
sin dddr =
2k
B
T
k
c
_ _
3=2
e
u
c
=k
B
T
(19:18)
k
c
can be calculated fromthe ensemble average of the magnitude of total force acting
on each particle. The partition function gives us the free energy and then the entropy
is obtained by taking the temperature derivative of the free energy.
In the case of water, the situation is complicated because of the anisotropic nature
of the potential. Thus, we have effective harmonic potential for translation, rotation,
and librational motions. Each is characterized by a force constant and contributes to
the partition function, free energy, and entropy. Furthermore, a water molecule can
be categorized by the number of HBs it forms. Since these quantities can be
considered as thermodynamic, they make a contribution as the entropy of mixing,
also known as the cratic contribution.
Implementation of the above scheme required determination of the force con-
stants, which are obtained from MD simulations.
Below we gather all the necessary information and the results in a tabular form.
19.5 Entropy of water in conned systems (reverse micelles,
carbon nanotubes, grooves of DNA)
Study of the entropy of water in conned systems brings out several interesting
issues. Resolution of all these issues is of great importance in understanding these
systems. This problem is also of considerable theoretical interest because of the
effects of the nite volume of the conned systems. We can start the effect by
looking at the SackurTetrode equation which gives the entropy of a monatomic
ideal gas at volume Vand temperature T [3]:
Table 19.1. (a) We have reported the entropies (JK
1
mol
1
) and enthalpies
(kJmol
1
) for different acceptor types and also the contribution of different modes
to the total entropy. (b) Theoretical (including the contribution of different modes)
and experimental data for entropy of water are provided. (The table has been
adapted from a private communication with Prof. R. H. Henchman.)
(a)
Entropy and enthalpy of water
Entropy type
(JK
1
mol
1
)
Single
acceptor
Double
acceptor
Triple
acceptor Broken HB Average
Cage vibrational 47.3 43.9 43.1 45.7 44.3
Librational 17.7 15.0 13.1 18.6 15.0
Cratic 16.0 2.7 17.0 34.4 7.4
Orientational 2.4 3.4 3.9 3.4 3.2
Total entropy 83.4 65.0 77.1 98.7 70.1 (69.9experiment)
Total enthalpy
(kJmol
1
)
27.3 36.5 41.2 27.4 35.7
(b)
Entropy of water (JK
1
mol
1
)
Vibration Libration Cratic Orientational Average
Theory 44.3 15.0 7.4 3.2 70.1
Experiment 69.9
19.5 Entropy of water in conned systems 299
S
id
(T; V)
N
= k
B
5
2
ln
2mk
B
T
h
2
_ _
3=2
ln
V
N
_ _
_ _
(19:19)
Note the dependence on temperature T and volume per particle V/N terms. In
conned systems, often the density could be less. Second, interaction with the
surface groups can be structure-making (hydrophilic) or structure-breaking (hydro-
phobic). If the interaction is strongly hydrophilic, the entropy of the conned liquid
is strongly compromised and the uidity of the liquid becomes smaller than the
bulk. Precisely the opposite is expected to be true for hydrophobic surfaces.
However, around a hydrophobic surface water molecules form a classic clathrate-
like structure (as discussed in the chapter on the protein hydration layer) among
themselves, which is a structure of higher order. Thus they lose entropy upon such
an ordering. Thus, the uidity entropy of a water molecule around a hydrophobic
surface decreases by a certain amount because of the formation of the clathrate
structure.
Thus, entropy continues to provide a correlation of dynamics with structure and
molecular arrangement, even in conned systems.
19.6 Conclusion
As mentioned earlier, the entropy of water under ambient conditions (T = 298 K and
atmospheric pressure) is equal to 16.8 calK
1
mol
1
. Valuable insight can be
obtained if we compare this amount with the entropy of water in the gas phase.
Under ideal gas approximation, we obtain translational entropy from the well-known
SackurTetrode equation, Eq. (19.19), equal to 20.3 calK
1
mol
1
(e.u.), we obtain
rotational entropy equal to 10.5 calK
1
mol
1
, and vibrational entropy equal to
0.1 calK
1
mol
1
. These are in ambient conditions. Adding all these, we obtain the
value 30.9 calK
1
mol
1
, which is larger than the experimental liquid state value by
more than 14 calK
1
mol
1
.
While accurate calculation of the entropy of pure water turned out to be
a formidable task, it is, however, not hard to rationalize why the entropy of
liquid water is much smaller than that in the ideal gas limit. First, the translational
entropy is lower because of the excluded-volume effect. The volume available
to a water molecule is dened by its neighbors and the specic volume in
the SackurTetrode equation is to be replaced by the free volume available to
individual water molecules. When this is taken into account, we obtain a contribution
of 11.8 calK
1
mol
1
from translational entropy. Second, the rotational contribution
also gets reduced because of the restriction that rotational motion experiences
in liquid water due to hydrogen-bonding. The reduced value of the entropy
can be calculated by considering the heat of fusion of ice and also subtracting the
contribution of the communal entropy. This procedure gives a value of
4 calK
1
mol
1
for the rotational entropy. Adding these we obtain
15.9 calK
1
mol
1
, which is quite close to the experimental value.
This method of estimation of the entropy of water is rather illuminating and
useful [14,15]. For example, we need such a partitioning of entropy in the case of
ligand binding which displaces water from the protein surface or drugDNA
intercalation.
In this chapter we have discussed several more detailed aspects of the entropy
of water. This is an exciting area of research and much remains to be understood or
accomplished, at the level of both theoretical calculation of entropy and experi-
mental determination of its value. We have focused on two oft-used expressions,
one due to Rosenfeld and the other due to Adam and Gibbs, and tried to provide a bit
of understanding of both. We discuss several recent schemes proposed to discuss
the entropy of water. These new schemes provide fairly accurate values of entropy
for bulk water and are expected to be semi-quantitatively accurate for water in
complex systems.
APPENDI X 19. A ENTROPY FOR TRANSLATI ONAL
DEGREE OF FREEDOM OF AN I DEAL GAS
( SACKURTERODE EQUATI ON)
From basic expressions of statistical mechanics, one obtains the following expres-
sion for the entropy of any system:
S = k
B
ln Q ( ) k
B
@
@
ln Q ( ) (19:A:1)
where k
B
is the Boltzmanns constant, = (1/k
B
T) and Q() is the partition function.
The partition function of a system of ideal gas of N molecules is given by
Q() =
1
N!
V
3
_ _
N
; (19:A:2)
which leads to the following expression for entropy
S=N = k
B
ln
v
3
_ _
5
2
k
B
(19:A:3)
where v = V/N is the specic volume. This is the SackurTetrode equation, referred
to in the text.
19.6 Conclusion 301
APPENDI X 19. B ENTROPY FOR VI BRATI ONAL
DEGREE OF FREEDOM
The partition function of a single harmonic oscillator is given by
Q ( ) =
e
h=2
1 e
h
(19:B:1)
where is the reduced Planck constant and is the frequency of the harmonic
oscillator. Nowif we put this expression of Q() into the above equation for entropy,
we obtain for each mode of frequency
S = k
B
ln 1 e
h
_ _
h
T
e
h
1 e
h
_ _
(19:B:2)
The total number of modes of vibration in a water molecule is three. We have
already discussed this before. Each mode will contribute its own share of entropy.
The total partition function is the product of the entropies of each mode and the
entropy is the sum of the entropy contribution from each mode.
The entropy of the molecular vibrational degree of freedom is usually small
because the vibrational frequencies are pretty large. For a water molecule, these
vibrational frequencies are more than 15k
B
T, so the exponential terms in Eq. (19.B.2)
above are all very small.
In liquid water there are a number of collective inter-molecular vibrational
modes, such as the HB excitation around 200 cm
1
, which is like a breathing
mode involving displacement of many molecules. Another example is the libration
mode at around 600 cm
1
, which is a restricted rotation, due to hydrogen-bonding.
These low-frequency modes contribute to the entropy and specic heat of liquid
water. But these modes are highly anharmonic, with a short lifetime, so the above
description based on a harmonic oscillator cannot be used to describe them.
This difculty does not arise in solid, where we have an energy spectrum
corresponding to its 3N 6 modes of vibration. If this spectrum is discrete, then
the addition of the weight functions for each of these 3N 6 modes of vibration
gives the entropy of a solid (Einsteins theory of heat capacity).
Such a discrete frequency spectrum can be obtained from the diagonalization of
the dynamic matrix. We shall not go into the details here except to mention that the
dynamic matrix is a 3N 3N matrix formed by the mass weighted position
uctuations along the three spatial coordinates of each individual atom. However,
if the spectrum is continuous, then the addition can be replaced by integration and a
function describing the probability distribution of the frequencies of the spectrum
appears. This newfunction is called the density of state (DOS) and the integration of
this function over the whole frequency range is the total number of modes of
vibration, 3N 6.
APPENDI X 19. C ENTROPY FOR ROTATI ONAL
DEGREE OF FREEDOM
Here no closed-form expression is available except at the limits of very high and very
low temperatures. Another complication is that water has three principal moments of
inertia, which are 1.09, 1.91, and 3.0 10
40
gcm
2
. The rotational partition function of
water is a product of three partition functions with the respective moment of inertia.
The partition function of each rotational mode can be obtained by summing the
Boltzmann term over the rotational quantum number J, as given below
Q() =
2J 1 ( )exp(E
J
=k
B
T) (19:C:1)
Here the factor (2J+1) is the degeneracy factor associated with the rotational
quantum number, J. If we approximate the water molecule as a rigid rotator, then
the energy level of each rotational mode is determined by its moment of inertia and
is given by
E
(i)
J
=
h
2
8
2
I
i
J(J 1) (19:C:2)
where I
i
is the moment of inertia along the ith principal axis.
However, the calculation needs to be carried out numerically, as mentioned
earlier. That is, we need to go through the process of calculating the rotational
partition function Q() and then use Eq. (19.A.1) to obtain the entropy.
The above description of rotational entropy is somewhat approximate because the
rotational energies depend on the vibrational state of the molecule through
vibrationrotation coupling, as the molecule is not fully rigid. At ambient condi-
tions, this correction is non-negligible.
In the classic limit and under rigid rotator approximation, one can easily derive
the following expression for the partition function for one rotator with moment of
inertia given by I
i
Q(I
i
) =
8
2
k
B
T
h
2
I
i
2
(19:C:3)
For water, we have three such moments of inertia, and the partition function of one
water molecule is the product of all the three. For N non-interacting particles (the
ideal gas limit), the total partition function of the system is (Q
RW
)
N
.
References
1. L. Boltzmann, Lectures on Gas Theory (Berkely, CA: University of California
Press, 1964).
2. P. W. Atkins, Physical Chemistry for the Life Sciences, 1st edn. (Oxford: Oxford
University Press, 2005).
References 303
3. H. B. Callen, Thermodynamics and an Introduction to Thermostatistics, 2nd edn. (New
York: John Wiley, 1985).
4. A. Cornish-Bowden, Enthalpyentropy compensation: a phantom phenomenon.
J. Biosci., 27 (2002), 121126.
5. Y. Rosenfeld, Relation between the transport coefcients and the internal entropy of
simple systems. Phys. Rev. A 15 (1977), 25452549; A quasi-universal scaling law for
atomic transport in simple uids. J. Phys.: Condens. Matter, 11 (1999), 54155427.
6. G. Adam and J. H. Gibbs, On the temperature dependence of cooperative relaxation
properties in glass-forming liquids. J. Chem. Phys., 43 (1958), 139146.
7. T. Goel, C. N. Patra, T. Mukherjee, and C. Chakravarty, Excess entropy scaling of
transport properties of Lennard-Jones chains. J. Chem. Phys., 129 (2008),
164904164909.
8. X. Xia and P. G. Wolynes, Fragilities of liquids predicted from the random rst order
transition theory of glasses. Proc. Natl. Acad. Sci. USA, 97 (2000),
29902994; Microscopic theory of heterogeneity and nonexponential relaxations in
supercooled liquids. Phys. Rev. Lett., 86 (2001), 55265529.
9. M. Agarwal, M. Singh, R. Sharma, M. P. Alam, and C. Chakravarty, Relationship
between structure, entropy, and diffusivity in water and water-like liquids. J. Phys.
Chem. B, 114 (2010), 69957001.
10. A. Scala, F. W. Starr, E. L. Nave, F. Sciortino, and H. E. Stanley, Congurational
entropy and diffusivity of supercooled water. Nature, 406 (2000), 166169.
11. A. Baranyai and D. J. Evans, Direct entropy calculation from computer simulation of
liquids. Phys. Rev. A, 40 (1989), 3817.
12. S. T. Lin, M. Blanco, and W. A. Goddard, The two-phase model for calculating thermo-
dynamic properties of liquids from molecular dynamics: validation for the phase
diagram of Lennard-Jones uids. J. Chem. Phys., 119 (2003), 11792805.
13. R. H. Henchman, Free energy of liquid water from a computer simulation via cell
theory. J. Chem. Phys., 126 (2007), 064504064508.
14. L. M. Amzel, Loss of translational entropy in binding, folding, and catalysis. Proteins:
Struct. Funct. Genet., 28 (1997), 144149.
15. J. D. Dunitz. Entropic cost of bound water in crystals and biomolecules. Science, 264
(1994) 670.
20
The freezing of water into ice
The freezing of water into crystalline ice presents another highly inter-
esting problem. Pure water, when broken into small droplets, remains in
the liquid state till ~ 40C, when crystallization nally converts liquid
water into ice. As many dynamic events in water occur on an ultrashort
timescale, this continuation of the liquid state much below its thermo-
dynamic transition temperature is hard to understand. Essentially the
same phenomenon is encountered in computer simulations where one
again nds that water is hard to crystallize unless the density is kept
slightly below the normal density of 1 gcm
3
or the potential is modied
to promote crystallization. It has been observed that the secret to this
resistance to crystallization lies in the energy landscape of liquid water,
which is complex, with a large amount of entropy arising from many
different HB arrangements. This makes the search for the pathway to
crystallization hard. In this chapter we discuss many aspects of the
freezing of water using recent computer simulation results. One nds a
rich physics in the freezing of water that makes a connection with many
areas of condensed-matter physics.
20.1 Introduction
Inside the deep freeze of a refrigerator, water easily transforms itself into ice without
any difculty. So it does in the upper atmosphere where the beautiful snowakes
form. The upper layers of water in the lakes in the arctic region freeze into ice sheets
in winter. When the temperature is raised, the same ice transforms back into water
without any difculty. The freezing of water into ice and the reverse process of the
melting of ice into water are common processes that we see are happening around us
all the time. Ice formation is also known to occur in interstellar space.
However, in experiments one nds that water droplets remain unfrozen till ~40 C.
Water has proven to be notoriously difcult to freeze in computer simulations,
where long runs and either slightly lower density or modied potential models
305
are needed to observe crystallization. So, what is the mystery? Why is the freezing
of water into ice so difcult to understand? In fact, the detailed mechanism of the
transformation of completely pure water into ice is still not fully understood. An
additional complexity is the existence of ice in many crystalline forms. Our
common ice is actually stable only over a narrow range of temperature and
pressure centered on ambient conditions.
It is well known that the volume of ice is greater than that of water by about 8%. For
most liquids the density increases on transforming the liquid to ice, as the solid is
usually denser than the liquid. This clearly shows that ice is more structured and
has more open space in its molecular arrangement. The volume expansion of water
upon freezing is an anomalous behavior because volume decreases upon freezing
for other simple liquids. It is this very anomaly by which sh can survive in low-
temperature regions because ice oats on the upper layer of the lake and the lower
layer of the lake still contains liquid water which is of higher density.
20.2 Phase diagram of water and ice
Let us rst discuss the phase diagram of water. The phase diagrams show the
thermodynamically stable physical states of matter at different temperatures and
pressures. Within each phase, the material is uniform with respect to its chemical
composition and physical state. At typical temperatures and pressures on Earth
( T =298 K(25C) and P =1 atm) water is a liquid, but it becomes solid (that is, ice)
if its temperature is lowered below 273 K (0C) and gaseous (that is, water vapor) if
its temperature is raised above 373 K (100C), at the same pressure. Each line (phase
coexistence line) on a phase diagram represents a phase boundary and gives the
conditions when two phases may stably coexist (in any relative proportions) [1].
At the phase boundary, a slight change in temperature or pressure may cause the
phases to abruptly change fromone physical state to the other. The point where three
phase coexistence lines join is called a triple point. Here three phases coexist in a
stable manner but may abruptly and totally change into each other given a slight
change in temperature or pressure. There thus exist such singular conditions of
temperature and pressure where liquid water, gaseous water, and hexagonal ice
stably coexist. It may seemsurprising, but at a triple point both the boiling point of
water and the melting point of ice are equal. A critical point occurs at the end of a
phase line where the properties of the two phases become indistinguishable from
each other. Here under singular conditions of temperature and pressure, liquid water
is hot enough and gaseous water is under sufcient pressure that their densities are
identical. Critical points are usually found at the high temperature end of the liquid
gas phase line. For water, the critical temperature is 374C and the critical pressure
306 The freezing of water into ice
is 217.7 atm. In the states above the critical point, the steam is called supercritical.
Supercritical water also shows anomalous properties, and is discussed later.
The complexity of liquid water is continued into its solid phase also.
Experimentalists and theoreticians have obtained/predicted many different forms
of ice at different temperatures and pressures [2]. The phase diagram displaying the
stability regions of different forms of ice is shown in Figure 20.1.
20.3 Ice formation in micro-droplets
The rst indication of the complexity that one may encounter in understanding the
freezing of ice came from a set of experiments reported in 1970s by Speedy and
Angell [3]. These scientists studied the freezing of micro-droplets of ultra-pure
water the droplets were of a few micrometer (m) diameter. The reason for
choosing such a small size was to avoid the presence of impurities in the sample.
Impurities act as seed for crystal nucleation and growth, and have to be avoided if
we want to understand the crystallization of pure water. Since it is practically
Figure 20.1. Phase diagram of solid ice. The ice we see every day is called Ih ice. In
ambient conditions, all other ice phases are thus metastable, meaning that they have
higher free energy. However, some of these other ice forms are found in interstellar
space. Adapted with permission from J. Phys. Condens. Matter, 24 (2012), 155102.
Copyright (2012) Iopscience.
20.3 Ice formation in micro-droplets 307
impossible to get rid of the impurities (ions, salts) in water, even after processing for
purity, the aim of these experiments was to obtain at least a few droplets which are
truly pure.
The experiments of Speedy and Angell revealed quite a startling result. A few
of the samples (which were free of impurities) remained in the liquid state till
about 40C, where they underwent spontaneous crystallization. That is, pure
water droplets can exist in the liquid state till about 40C but not any lower!
Why? Nobody really knows the answer for sure yet, but progress is being made.
20.4 A lesson from the freezing of interacting spheres
and the difference from water
The freezing of a liquid can be studied fruitfully by computer simulations. In the
simplest model, one studies a system of spheres that interact with each other via a
short-range interaction potential, such as the LennardJones.
Under the inuence of such a potential, the spheres easily settle into a crystalline
form when the temperature is lowered or the pressure is raised. The solid phase can
be either a face-centered cubic (fcc) lattice or a body-centered cubic (bcc) lattice,
depending on the detailed nature of the potential. In the case of argon, the crystal
freezes into an fcc lattice, while liquid sodium freezes into a bcc lattice. The
transformation of simple spherical molecules (such as argon and sodium) into the
crystalline solid phase is nowrather well understood, largely because of the extensive
use of computer simulation studies, accompanied by theoretical analysis using meth-
ods of statistical mechanics.
However, it was not possible, till very recently, to learn about the mechanism of
the freezing of water into ice from computer simulations because water just would
not freeze into ice in the computer! That is, the freezing of liquid water is not as
simple as one would imagine! This completely agrees with the early experiments of
Speedy and Angell. But the reason is not clearly understood even today! That is,
even the apparently simple process of forming ice eludes us!
20.5 The freezing of water
Let us now look into the basic characteristics of the freezing/melting process of
water. Under normal conditions (and probably in the presence of impurities), bulk
water freezes into a hexagonal lattice with a release of 1.44 kcal/mol of latent heat.
This is accompanied by an increase of volume, as mentioned earlier, of 8%. Water
also has a variety where two hydrogen atoms can be replaced by deuterium. This is
called deuterium oxide or heavy water (D
2
O). It has physical properties similar to
normal water with some modication due to the isotope effect. In the case of heavy
water, the freezing point is raised to 3.8C and also the latent heat released due to its
freezing is 1.52 kcal/mol. These changes indicate that the freezing process is quite
sensitive to small changes in the molecular details.
Let us now spend a few moments understanding the essence of any freezing
process. A liquid lacks the order of a crystalline solid. If we consider molecules
which can be modeled as spheres (such as the noble gases, such as argon or neon),
then a liquid is a state where all these spheres are randomly arranged in space. At a
rst glance, it may appear that the transformation of this random arrangement to a
periodic solid may require a large degree of rearrangement and displacement of
atoms. That is not true. Nature takes a different, more prudent path of less energy
and less resistance, the path of nucleation.
20.6 Nucleation of an embryo
As mentioned above, freezing does not proceed as a sudden total transformation of
the whole liquid into solid ice; instead, a small embryo (or nucleus) of the solid rst
forms by random motions of atoms and molecules in the liquid [4,5]. Such a
coordinated random motion (somewhat of a contradiction in terms!) is termed a
uctuation. However, most of these embryos formed by uctuation within liquid
water are of higher energy than the liquid at the same volume. Therefore, they are
not stable and melt back into liquid more easily than they are formed.
But how then does freezing happen at all? Even though energetically unfavor-
able, once in a while, as a rare process, a chance uctuation can occur where the size
of the ice embryo can become larger than a certain critical size. The critical size is
dened as the size beyond which the ice nucleus is more stable than the liquid of the
same volume. Now there is no barrier and the crystal nucleus grows into the liquid.
Thus, growth is like a propagation of a crystalline front into the liquid. The scenario
is shown in Figure 20.2.
In the case of water, however, the formation of such an embryo is not easy, due to
the extensive hydrogen-bonding among its molecules. As the temperature of water
is lowered, the HBs also become stronger, stabilizing the existing random arrange-
ment of molecules. Thus, while ice becomes increasingly more thermodynamically
stable below the freezing temperature, the formation of the critical nucleus becomes
more difcult!
We have already discussed in Chapter 2 that water exhibits a large number of
anomalies at low temperature, below its freezing temperature. The properties of
liquid water at low temperature may be explained qualitatively in terms of the
existence of two disordered forms the high-density liquid (HDL) and the low-
density liquid (LDL). Our common water consists mostly of HDL. The freezing of
ice seems to be intimately connected with these two forms. However, scientists are
20.6 Nucleation of an embryo 309
still divided over these issues and we must alert the reader to the controversies that
have existed in this problem for a long time, and no consensus has yet been reached.
Recently, an interesting experiment was carried out on the freezing of water
within carbon nanotubes. When water is conned in nanotubes, it can be cooled
down to a very low temperature, 100C below the freezing temperature, without
encountering crystallization. This provides an important clue the size of the critical
nucleus (which is formed by a rare uctuation) could be larger in water than in
common liquids. Also, since the nucleus of ice is of lower density, one needs the
space for expansion, which might not be available in nanotubes. The conning
surface can also hinder the growth of the crystalline ice. All these factors conspire to
keep water in the liquid phase in nanotubes. Interestingly, at low temperature, the
HDL undergoes a transformation into LDL, providing the rst experimental ver-
ication of the existence of the LDL phase.
We now discuss the rst successful simulation of the freezing of ice, which
revealed a wealth of information about the freezing process, and also threw light
on the difculty of freezing water into ice.
20.7 The freezing of water in computer simulations
Perhaps the rst demonstration of the freezing of water into ice in computer
simulation was observed by Matsumoto et al. [6,7]. These authors simulated many
Figure 20.2. Free-energy cost for the formation of an embryo. The gray bold dotted
line shows the energy cost due to the formation of the surface (as there is a
mismatch at the surface of the two phases) of the embryo, while the gray faded
dashed line is the gain in energy because the solid is thermodynamically more
stable than the liquid. The black solid line shows the sum of the two energies. It
shows a maximum which poses a barrier for the growth of the embryo. This is
called the critical cluster. (prufe.mit.edu/~ccarter/3.21/Lecture_24/.)
different systemsizes, some for a very long time, and could capture and analyze many
aspects of the freezing of water, which unraveled (at least partly) the mystery of water
freezing. These seminal studies, together with several other computer-based studies
[810], have played an important role in todays understanding of the freezing of
water.
Matsumoto et al. found that the freezing of pure bulk water in a computer
simulation was next to impossible. This can be understood as follows. When we
are at a temperature well below 0C, the stable structure of the system is ice.
However, the system still remains in the liquid phase where the arrangement of
the water molecules is quite random. Such a state of the liquid is called a metastable
state. As shown in Figure 20.2, an ice nucleus of a size larger than the critical size
needs to formto transformthe liquid into ice. In the metastable supercooled state the
structure of the liquid changes continuously in the search for this elusive ice
nucleus. This is a difcult and rare process as it requires the breaking and formation
of an extended HBnetwork on the way. However, we need to nd the distance of a
local water structure from that of the ice. This is a hard process because the water
structure is continuously moving. An interesting new way to learn about the multi-
tude of non-crystalline liquid structures is by studying the potential-energy land-
scape of water, which we describe below.
20.8 Mechanism of ice formation
As we have discussed earlier, because of the presence of the HB network structure
water has an enormous number of stable or quasi-stable energy states (reected in its
high heat capacity). This is also reected in the difculty of ice formation in
computer simulations. The system needs to search for a long time to nd the global
minimumfor ice structure. The systemexplores the potential-energy landscape for a
considerable amount of time before the start of the ice formation. This scenario
agrees well with the nucleation picture discussed earlier in this chapter.
Matsumoto et al. [7] also studied the detailed mechanism of ice nucleation at
different stages of the search (see Figure 20.3). The hydrogen-bonded network
structure of the system in the different stages of the MD simulation revealed an
interesting mechanism of the ice formation. In the initial stage, they observed the
formation of ve-, six-, and seven-member rings of water molecules and those rings
were destroyed and re-formed continuously. Hydrogen bonds with a relatively
longer lifetime (more than 2 ns) are called long-lasting HBs. These long-lasting
HBs appeared occasionally and randomly at various locations in the systeminitially.
At low temperature, while 90% of the water molecules formed these long-lasting
HBs, the probability that all the HBs of a polyhedral structure (ring) were long-
lasting was next to impossible. Thus the probability of the formation of stable rings
in the systemwas also quite small. The systemstudied by Matsumoto et al. [7] went on
uctuating in such a situation for quite a long time.
After a sufciently long time, one rare uctuation appeared in the system and a
polyhedral structure (ring) was formed with all the long-lasting HBs and the ring
became quite stable. This polyhedron changed its position and shape by altering the
HBs with the surrounding water molecules, grew slowly and nally anchored at
the position it occupied. By this rare process, an ice nucleus formed that had a size
larger than the critical size mentioned above.
Once the nucleus of sufciently large size formed, it then expanded rapidly, by
transforming its HB network elements into mainly six-member rings, which perco-
lated through the entire three-dimensional space of the system. At the same time, the
system decreased its total potential energy rapidly. At the end of this period of rapid
Figure 20.3. Different stages of ice nucleation. Note the region indicated by the
yellow circle (actually a sphere in three-dimensional space), which indicates
the rst incipient ice formation. The time intervals are indicated on the upper
right corner of each panel [7]. Adapted with permission from Nature, 416 (2002),
409413. Copyright (2002) Nature Publishing Group. See plate section for color
version.
growth, a stacked honeycomb structure, consisting of six-membered rings, was
established throughout most of the system.
The above process of the formation of the long-lasting hydrogen bonds as a
function of time is depicted in Figure 20.4. There is an interesting intermittency in
this number of long-lasting hydrogen bonds. This number of long-lasting hydrogen
bonds uctuates in the initial nucleation period and rapidly increases afterwards. It
is important to note here that, during this initial stage, most of the long-lasting bonds
appear intermittently, reminiscent of the intermittent collective motion of water
molecules that occurs as a result of extensive hydrogen-bond network rearrange-
ment dynamics of liquid-state water. This suggests that nucleation of ice occurs out
of rare but natural uctuations of water.
The time dependence of uctuation of the number of long-lasting hydrogens
bonds during the initial stage yields a 1/f-type power spectrum with the same slope
as the structural uctuation associated with collective motion mentioned above. It is
known that 1/f power spectrum is a signature of intermittent dynamics. Such
intermittent dynamics is often a characteristic of so-called frustrated systems, such
as liquid water, where system oscillates between different energy minima.
Matsumoto, Saito and Ohmine. [7] studied freezing of water under constant
volume and constant temperature conditions. This does not allow volume uctua-
tion in the system. Liquid water at low temperature exhibits large-scale density
uctuations, which have been interpreted to mean that there are regions of low-
density liquid phase that facilitate initial nucleus formation. These low-density
Figure 20.4. The number of water molecules having long-lasting HBs is indicated
by the dark line. Note the rapid growth at around 250 ns [7]. Adapted with permission
from Nature, 416 (2002), 409413. Copyright (2002) Nature Publishing Group.
regions induce wetting of the nucleus. The small difference in density and other
order parameters (like tetrahedrality) reduce the surface tension (or surface energy)
of the crystalliquid interface thus facilitating the formation and growth of the
incipient crystal. This effect was partly included in a separate calculation that used a
slightly reduced water density of 0.96 gcm
3
where the freezing process becomes
faster. Therefore, reduction in the occurrence of such large-scale low-density
uctuations should lower the rate of crystallization rate and thereby may lead to
the formation of amorphous ice under ambient conditions. The low-density uctua-
tions may have the character of LDL discussed in Chapter 2.
Subsequent to the work of Matsumoto et al., further exploration of ice formation
in pure water has been carried out by Radhakrishnan and Trout [8], who studied the
details of order formation by studying the emergence of tetrahedral and hexagonal
order.
20.9 Freezing inside nanotubes
Water in well-characterized pores is a system of general interest because it serves as
model system for the non-bulk or inhomogeneous water that is ubiquitous in
biological and geological systems, as well as in nano-structured materials. Often
conned or interfacial water is highly relevant to the properties and functions of
entire systems, e.g., those in ion channels and clay minerals. X-ray diffraction
studies show that water can ll the inner space of open-ended single-walled carbon
nanotubes (SWCNTs) under ambient conditions and freezes into crystalline solids.
These are often referred to as ice nanotubes. Ice structures in conned systems are
characterized as stacked n-membered rings or equivalently as a rolled square-net
sheet. The formation of the ice nanotubes in CNTs has also been observed by NMR,
neutron diffraction, and vibrational spectroscopy studies.
Interestingly, the prediction of spontaneous ice formation in CNTs was rst made
in a molecular dynamics (MD) simulation study. It was shown that the conned
water freezes into square, pentagonal, hexagonal, and heptagonal ice nanotubes, and
unexpectedly it does so either continuously (unlike any bulk substances, including
bulk water) or discontinuously (despite the fact that it is essentially in one dimen-
sion). The nature of the transition depends on the diameter of the CNTs or the
applied pressure. Recent simulation studies predicted spontaneous formation of
octagonal ice nanotubes, ice nanotubes with hydrophobic guest molecules, single-
layer helical ice sheets, and multi-walled ice helices and ice nanotubes [9]. The
versatility of ice we know for bulk water seems to survive in the nano connement.
Among the properties of water in well-dened nanopores, a global picture of the
phase behavior is not yet available. We do not accurately know the pore-size
dependence of the melting point in the nanometer scale or the conditions for gradual
and abrupt freezing. Previous results with other substances imply that the melting
point and the freezing behavior depend strongly and non-monotonically on the
diameter of the nanopores.
Researchers fromCornel University performed extensive MD simulations of water
in wide ranges of nanotubes diameters (917 ) and temperatures (100300 K) to
explore the phase behavior of water in the temperaturediameter plane [8]. The
pressure was xed at atmospheric pressure. Spontaneous formation of crystalline
ice from liquid water under atmospheric pressure was monitored in the process of
decreasing T from 300 K for systems with D 12 and from lower temperatures for
systems with D < 12 . They observed nine one-dimensional ice structures sponta-
neously form in the CNTs at atmospheric pressure. The ice that forms in the smallest
diameter range has a ladder-like structure in which each molecule has only three
hydrogen-bonding neighbors. This ice structure is referred to as ladder-like ice. In the
next range of diameters, water freezes into a helical ice nanotube. In the third to sixth
ranges of diameter there is the formation of square, pentagonal, hexagonal, and
heptagonal structure ice nanotubes, respectively. In the seventh range of diameter
the conned water freezes into an octagonal ice nanotube. The inner space of the
octagonal ice nanotube may have additional water molecules. In the eighth and ninth
ranges of the tube diameter, liquid water freezes into double- and triple-layer (DL and
TL) ice. [9] Unlike ice nanotubes, the outermost layer of DL and TL ice is a hexagonal
HB network and so the number of HBs per molecule within the outermost layer is
three; another hydrogen-bonding neighbor is a molecule in the inner layer.
The appearances of such diverse ice structures in CNTs reveal the unique ability
of water molecules to form stable or quasi-stable arrangements through HBs. Water
is like the mythological demon that can take many shapes as and when required.
20.10 Conclusion
It should be clear from the above that because of sustained efforts over many
decades, signicant progress has now been achieved in the understanding of the
freezing of water into ice. However, there still remain many unsolved problems in
this area. For example, we do not yet have a quantitative theory of the nucleation of
ice in supercooled water. The molecular models we use in simulations are perhaps
too primitive, as most of them do not include the polarizability of water molecules.
The polarizability of water is large due to the two lone pairs of electrons on the lone
oxygen atom. Perhaps one would need to consider quantum simulations to fully
understand the freezing of ice.
The consequence of the existence of large-scale density uctuations due to the
existence of the two forms of water (HDL and LDL) in the freezing of water is yet to
be understood. So is the case for the freezing (or rather the lack of it) in CNTs.
Many of the outstanding problems in this eld are of great importance to natural
science. For example, the formation of ice in interstellar space remains to be
explored. Certain bacteria that we nd in plants have the unique ability to initiate
ice nucleation at as high a temperature as 1C. How they perform this task is also
shrouded in mystery. It is really interesting to nd the existence of such a large
number of fascinating problems in the area of water freezing.
References
1. L. Glasser, Water, water, everywhere. J. Chem. Edu., 81 (2004), 414418.
2. P. V. Hobbs, Ice Physics (New York: Oxford University Press, 1975).
3. R. J. Speedy and C. A. Angell, Isothermal compressibility of supercooled water and
evidence for a thermodynamic singularity at 45C. J. Chem. Phys., 65 (1976),
851858.
4. P. G. Debenedetti, Metastable Liquids: Concepts and Principles (Princeton, NJ:
Princeton University Press, 1996).
5. A. C. Zettlemoyer, Nucleation (New York: Dekker, 1969).
6. I. Ohmine and S. Saito, Water dynamics: uctuation, relaxation, and chemical reactions
7. M. Matsumoto, S. Saito, and I. Ohmine, Molecular dynamics simulation of the ice
nucleation and growth process leading to water freezing. Nature, 416 (2002), 409413.
8. R. Radhakrishnan and B. L. Trout, A new approach for studying nucleation
phenomena using molecular simulations: application to CO
2
hydrate clathrates.
J. Chem. Phys. 117 (2002), 17861796; Nucleation of crystalline phases of water in
homogeneous and inhomogeneous environments. Phys. Rev. Lett., 90 (2003),
158301158304; Nucleation of hexagonal ice (Ih) in liquid water. J. Am. Chem. Soc.,
125 (2003), 77437747.
9. D. Takaiwa, I. Hatano, K. Koga, and H. Tanaka, Phase diagram of water in carbon
nanotubes. Proc. Nat. Acad. Sci. USA, 105 (2008), 3943.
10. E. B. Moore and V. Molinero, Structural transformation in supercooled water controls
the crystallization rate of ice. Nature, 479 (2011), 506508.
21
Supercritical water
The uniqueness of water is not restricted to the liquid state. When liquid
water is heated to a temperature above its gasliquid critical temperature
of 647 K (374C) at high pressure, the state is called supercritical water.
Because of the high pressure, the density of this gaseous water is still
high. Supercritical water exhibits an amazing variation of properties.
Because of the breakdown of the extensive HB network of liquid water, it
can now solvate organic solutes like an organic solvent. This useful
property can be exploited in many ways, discussed in this chapter. In
addition, due to the proximity to the critical temperature, a large varia-
tion in density can be achieved by a small variation of pressure.
Theoretical and computer simulation studies suggest that spontaneous
large-amplitude density uctuations (due to the systems proximity to the
critical temperature) are partly responsible for the unusual properties of
supercritical water. We discuss the concept of the Widom line, introduced
recently to describe the marked enhancement observed in isobaric spe-
cic heat and isothermal compressibility when the temperature of water
vapor is varied, at a constant pressure above the critical pressure, across
the critical temperature.
21.1 Introduction
Supercritical uids are highly compressed gases at temperature and pressure con-
ditions such that the temperature is a few degrees above the critical point and the
pressure ranges from the critical pressure to above. For water, the critical tempera-
ture is 647 K and the pressure is ~218 atm(~22.1 MPa). Because of the proximity to
the critical point, supercritical uids combine the properties of gases and liquids in
an intriguing manner. Due to the high compressibility of supercritical uids, small
changes in pressure can produce substantial changes in the density of a supercritical
uid which, in turn, affects the diffusivity, viscosity, dielectric, and solvation
properties of these uids. Dramatic changes in the properties of these uids have
317
been observed in the supercritical state. There has been growing interest in the
inuence of supercritical uids on the kinetics and mechanisms of chemical reac-
tions [15].
The interesting properties of supercritical uids can be appreciated by studying the
case of water, which, of course, holds a special place in the literature of supercritical
uids. As already noted, the critical point of water is located at a pressure (P
c
) of
22.1 MPa (218 atm) and temperature (T
c
) of 647 K. Under such a high pressure, the
density is pretty high,
c
= 0.32 g cm
3
. By supercritical water (SCW) one usually
means water at a higher-than-critical temperature (that is, above 647 K) and at
relatively high density. At such high temperature, the extended HB network of liquid
water becomes essentially non-existent and water shows certain remarkable
properties. The dielectric constant of SCWis about 68, making it similar to organic
liquids in many respects [1]. As a result, high-temperature water behaves like an
organic solvent. Organic compounds enjoy high solubilities in near-critical water and
complete miscibility with SCW. Moreover, gases are also miscible in SCW.
These properties allow SCW to provide an environment and an opportunity to
conduct chemistry in a single uid phase that would otherwise occur in a multiphase
system under more conventional conditions. The advantages of a single
supercritical-phase reaction medium are that (i) higher concentrations of reactants
can often be attained and (ii) there are no inter-phase mass transport processes to
hinder reaction rates.
21.2 Inhomogeneous density uctuation in supercritical uids
Solvation of a foreign solute in dense liquid is usually a local phenomenon. However, a
large and growing body of evidence from the past decade suggests that when a solvent
has a large macroscopic compressibility, , there are consequences at the microscopic
level as well and this can affect solute solvation, dynamics, and reactivity of the solute.
It is known fromstudies of critical phenomena that the magnitude of the compressibility
is directly related to the correlation length, , which gives the distance over which
microscopic uctuations in uid density are spatially correlated. The divergence in the
compressibility as T T
c
from above (along the Widom line, see Chapter 2) corre-
sponds to a concurrent divergence of this correlation length , that implies the extension
of these density uctuations over macroscopic dimensions. Clearly, as we move away
froma solvents critical point to higher temperatures, the compressibility decreases and
the correlation length also decreases. Yet, locally there could be regions of large
compressibility where the correlation length may still extend over many molecular
diameters (say, 10 or 20), i.e., it may be of mesoscopic length. Therefore, if we take a
snapshot of each uid, we would nd that the instantaneous picture of an SCF in
its compressible regime resembles that of an inhomogeneous medium with high- and
318 Supercritical water
low-density regions extending over lengths of the order of the correlation length and
smaller. [2] Such density inhomogeneities are the hallmark of a supercritical uid and
evident in the molecular dynamics snapshot of a two-dimensional LennardJones uid
shown in Figure 21.1.
There is a simple way to rationalize the existence of these regions of high and low
density in equilibrium, homogeneous over a long-time and large length scale,
single-phase uid. [2] A large compressibility means that there is little free-energy
cost to compression of the uid. At a microscopic level, then, one expects that the
free energy cost of local density uctuations is also low. This of course implies that
surface tension of creating a gasliquid interface is small, which is indeed the case.
In other words, the entropic cost of moving an isolated molecule onto a high-density
region is nearly balanced by the energetic gain resulting from the increased number
of favorable intermolecular interactions [2]. Near the critical temperature, the gas
liquid surface tension varies as (
p
)
4
where
p
is the density difference between the
high and low density regions. Small surface tension near the critical temperature
facilitates creation of large-scale density inhomogeneity due to the force-eld of the
solute, giving rise to unusual solvation characteristics observed in experiments. [2]
Figure 21.1. Snapshot of instantaneous conguration a pure two-dimensional
LennardJones SCF at T
r
1.17 and
r
0.86. Figure granted with permission
fromChem. Rev., 99 (1999) 391418. Copyright (1999) American Chemical Society.
21.2 Inhomogeneous density uctuation in supercritical uids 319
21.3 Crossing the Widom line
The Widom line is a line in the PT phase diagram that starts from the critical point
and rises above the critical pressure while remaining at the critical temperature [3].
It has been observed that when one approaches this line at constant pressure above
the critical pressure by varying the temperature (all above T
C
), the system experi-
ences large density uctuations, as reected in the increase of compressibility.
Specic heat also increases. Thus, the anomalous properties in the supercritical
state can be correlated with the proximity to the Widom line.
The presence of this Widom line has been proposed to explain low-temperature
thermodynamic anomalies. As is evident fromthe above, the Widomline starts from
the critical point where density uctuation is highest and the uctuation becomes
weaker as the state point goes away from the critical point.
The concept of the Widom line is also applicable to the density uctuation and
inhomogeneity observed for different systems beyond critical point in the super-
critical region.
21.4 Spectroscopic studies of supercritical uids
The short-range or local structure in neat supercritical uid consisting of polar
molecules was investigated by the study of the Raman spectrum of uoroform
(CHF
3
) at reduced temperature T
r
= T/T
c
= 1.02. Raman scattering was from the
CH mode. The spectra were examined using the SchweizerChandler model of
vibrational phase relaxation, which attempts to explain the broadening of the
vibrational lineshape by decomposing intermolecular interactions between two
molecules into attractive and repulsive components [4]. Thus, the spectral shift
was also decomposed into contributions from attractive and repulsive components
as a function of density, which was varied by a factor of 50. The attractive
component showed a larger shift than anticipated from the uniform molecular
distribution. A local density enhancement due to attractive intermolecular interac-
tions was denitely observed in the neat supercritical uid [1].
Vibrational relaxation is a sensitive probe of local structure and dynamics [5].
Vibrational lifetimes and absorption spectra of the asymmetric CO stretching mode
(~1990 cm
1
) of W(CO)
6
in supercritical CO
2
are reported as functions of solvent
density and temperature [6]. Close to the critical temperature, the observables are
density-independent over a 2-fold range of density. A cluster model can explain the
data if small xed-size solutesolvent clusters are formed in the range of densities
around the critical density. If the size, and therefore the properties, e.g., local density
and spectrum of uctuations, are density-independent then the observables also
become density-independent. Such a structure may form if there is a liquid-like
condensation about the solute. Although the bulk system is slightly above the
critical temperature, an attractive solutesolvent interaction term greater than that
between solvent molecules changes the local properties of the system. A loca-
lized phase transition could occur forming a nano-droplet composed of only two
or three solvent shells. The nano-droplet would have a small xed size because the
range of the solutesolvent attractive interaction falls off relatively rapidly, e.g., as
1/r
6
, so beyond a few solvent shells condensation is no longer favorable. The
result could be the formation of a stable structure that does not change over a range
of densities. When a sufciently high pressure is applied, there is an overall
favorable change in the local free energy, the structure grows, and the vibrational
observables again become density-dependent. At higher temperature (~50C) the
increased solutesolvent interaction is insufcient to form the stable nano-droplet
structure. Clustering may still occur, but the cluster properties change continu-
ously with density [5].
An important point to note here is the separation of timescales between vibra-
tional relaxation and density relaxation. Vibrational relaxation is expected to be
faster than density relaxation [6]. Therefore, vibrational relaxation is a good probe
of the density inhomogeneity in supercritical water that is present on a short time-
scale short compared to density relaxation. The latter could occur in nanosecond
timescales when close to the critical temperature.
Simulation studies of solvation dynamics (SD) in SCW were reported for the rst
time by Rey and Laria [7]. Their studies indicated a biphasic decay of solvation energy,
with an ultrafast decay, rather similar to the one observed for bulk water. This is rather
surprising because here density is low and the extended HB network is non-existent,
thereby eliminating the contributions from the libration and intermolecular vibration
modes. Their results were subsequently corroborated by theory, which shows that the
ultrafast component arises here from the fast rotational motion of small water
molecules [8].
Recent simulation studies nd that the SDin SCF, CHF
3
, and CO
2
is also biphasic
in nature. The fast component of the total solvation energy here decays with a time
constant of about a picosecond. The other component relaxes at a rate with time
constant in the tens of picosecond regime. A set of recent experimental studies
employing a time-correlated single-photon counting technique has, however, indi-
cated that the slowcomponent has a time constant of about 5070 ps, which is much
slower than that observed in the above simulation, and probably was missed in later
studies [9].
Some of the anomalies of supercritical uids can be understood by using the idea of
the Widom line. One can then relate, for example, the width of a Raman line to the
temperature- and density-dependent correlation length of the uid. As we cross the
Widomline at constant density, we would expect a sharp rise in the width of the Raman
21.4 Spectroscopic studies of supercritical uids 321
line. This has been seen in the Raman width of the NN bond across the critical
temperature [10,11].
Supercritical water also exhibits ultrafast SD. Ultrafast solvation of supercritical
water has been explained by noting the fact that here single-particle rotation is fast
and drives SD. Thus, the physical origin is quite different from that in liquid water,
where the collective polarization response gives rise to ultrafast SD.
21.5 Conclusion
It is interesting to note that the useful properties of supercritical water arise from the
breakdown of the extensive HBnetwork that is at least partly responsible for many of
the anomalies of liquid water. We have discussed how the use of the idea inherent in
the Widom line helps in understanding the large-scale uctuations observed in super-
critical water. Because of the large separation of timescales between vibrational
relaxation and density relaxation, the vibrational line widths are inuenced signi-
cantly by the transient density inhomogeneity present near the critical temperature.
References
1. See the articles in Chem. Rev. (Special Issue: Supercritical Fluids), 99 (1999).
2. S. C. Tucker, Solvent density inhomogeneities in supercritical uids. Chem. Rev., 99
(1999) 391418.
3. G. G. Simeoni, T. Bryk, F. A. Gorelli, et al., The Widom line as the crossover between
liquid-like and gas-like behaviour in supercritical uids. Nat. Phys., 6 (2010), 503507.
4. K. I. Saitow, K. Otake, H. Nakayama, K. Ishii, and K. Nishikawa, Local density
enhancement in neat supercritical uid due to attractive intermolecular interactions.
Chem. Phys. Lett., 368 (2003), 209214.
5. D. J. Myers, M. Shigeiwa, M. D. Fayer, and B. J. Cherayil, Vibrational lifetimes and
spectral shifts in supercritical uids as a function of density: experiments and theory.
J. Phys. Chem. B, 104 (2000), 24022414.
6. B. Bagchi, Molecular Relaxation in Liquids, 1st edn. (New York: Oxford University
Press, 2012).
7. M. Rey and D. Laria, Dynamics of solvation of supercritical water. J. Phys. Chem., 101
(1997), 1049410505.
8. R. Biswas and B. Bagchi, Ion solvation dynamics in supercritical water. Chem. Phys.
Lett., 290 (1998), 223228.
9. F. Ingrosso, B. M. Ladanyi, B. Mennucci, and G. Scalmani, Solvation of Coumarin 153
in supercritical uoroform. J. Phys. Chem. B, 110 (2006), 49534962.
10. M. Musso, F. Matthai, D. Keutel, and K. L. Oehme, Critical Raman line shape behavior
of uid nitrogen. Pure Appl. Chem., 76 (2004), 147155.
11. N. Gayathri and B. Bagchi, Subquadratic quantum number dependence and other
anomalies of vibrational dephasing in liquid nitrogen: molecular dynamics simulation
study from the triple point to the critical point and beyond. Phys. Rev. Lett., 82 (1999),
48514854.
22
Approaches to understand water anomalies
The anomalies of water, however puzzling, hold the secret to understand-
ing this unique liquid [1]. Fortunately many aspects of the anomalous
behavior are now understood at a molecular level, particularly the
density maximum of water at 4C, the unique solvation properties, the
reason for large specic heat, and the unique properties of supercritical
water. However, there are still a fairly large number of anomalies that are
not yet well understood, such as the anomalies in supercooled water. The
latter have given rise to a large number of studies, and exotic nomencla-
ture such as no-mans-land, which is meant to say that a broad
temperature zone between 155 K and 232 K is not accessible experimen-
tally and hence cannot be studied directly. The reasons for the existence of
such a temperature zone and the state of water, if it were to be present
there, have captured the attention of researchers in recent times. However,
no consensus in the understanding has yet been reached. Despite this
lacuna, many other features are understood and we shall discuss some of
the above features in this chapter. While everything is not understood
quantitatively, a lot has also been achieved in the last few decades,
particularly after the computational infrastructure became available.
22.1 Introduction
Water has such a large number of anomalies that it is often futile to give a complete
list. On some internet sites, a list of more than hundred anomalies is given. Broadly
speaking, we can divide the anomalies into two classes: those above the freezing
temperature of 0C covering the liquid state till the boiling point, and those below
the freezing point up to its spontaneous crystallization temperature of 232 K.
The fact is that pure water, broken down into small droplets of micrometer size,
can be supercooled to 232 K. This in itself is a puzzle because water is a liquid with a
fast response and therefore not expected to fail to crystallize and reach the state of
minimum free energy. Nevertheless, liquid water can be supercooled by as much as
323
41C. Not only in experiments, but even computer simulations, nd that liquid
water does not crystallize easily.
We have already discussed many of the unusual properties of water at room
temperature, such as the temperature dependence of the pH of water, its hydro-
phobicity and hydrophilicity with respect to foreign solutes, its role in controlling
biological functions, etc. Here we shall discuss a fewmore such issues not discussed
previously.
We have already articulated the uniqueness of water in Chapter 1 in terms of a few
unusual molecular features. Some aspects of the properties of water in natural and
biological processes are not hard to understand at a simple level in terms of those
molecular properties, such as its small size and ability to form four HBs. In liquid
water, the average number of hydrogen-bonded neighbors of a single water mole-
cule at room temperature is about 3.5. As one needs at least a coordination number
of three per molecule to ensure the existence of an extended and connected network
of HBs percolating throughout the system, liquid water can be considered as a giant
gel. That is, in this network each water molecule is correlated to another, even to
distant ones, through an HB pathway. As these HBs themselves are short-lived, this
spatial correlation is also short-lived. Nevertheless, the important role of this HB
network becomes apparent when water molecules face an extended hydrophobic
surface, as discussed in previous chapters.
As mentioned in the preceding paragraph and also discussed in the rst chapter, the
HBenergy is such that an individual HBcan break and re-formagain about every few
picoseconds (1 ps = 10
12
s). That is, the lifetime of an HB is rather short and in the
ultra-fast timescale range. Thus at room temperature these HBs are continuously
breaking and forming, giving the entire network a uctuating character. The re-
arrangement of the HB network is an important dynamic event that controls many of
the anomalous properties of liquid water.
One needs to remember that different HBs in water have different energies due to
the ever-changing position and orientation of water molecules. This distribution of
HB energies is intimately connected to the uctuating nature of the HB network.
During its lifetime an HB can go from weak to strong and again to weak. This
uctuating HB network dominates the properties of water at room and low tem-
perature and makes its behavior quite different froma simple liquid, the prototype of
which is liquid argon, which can be modeled as a collection of spheres interacting
with a spherically symmetric potential.
Such a billiard-ball model of a simple liquid, initiated by J.D. Bernal, has
dominated theoretical studies/approaches towards the liquid state of matter for
half a century. Such a billiard-ball model, and its generalization to include non-
spherical shapes, works not only for liquid argon and krypton, but also for many
liquids such as methane, ethane, and carbon tetrachloride, to name a few. However,
324 Approaches to understand water anomalies
such a simple model fails completely for liquid water. It is sometimes stated that
liquid water at room temperature behaves like a supercooled simple liquid in the
sense that liquid water is dynamically correlated over an intermediate length scale
(meaning 1015 water molecular diameters) even at room temperature. However,
such a statement is hardly valid because of the uctuating nature of the HB network
and things are more complex, as recent research work has demonstrated.
The percolating HB network model was developed by many authors, notably
Gibbs et al. [2], Stillinger and Rahman [3], Stanley and Teixeira [4], Ohmine and
Saito [5], and several others. In the following we shall describe some of these
historical developments that led to the present-day understanding of the complex
properties of water.
The uctuating HBnetwork model of water outlined above and described in more
detail below needs to be quantied in order to explain the peculiar properties of
water. Attempts have been made to develop a two-state model of liquid water [6].
The idea behind such a model is that the HB network of water can exist in several
different forms. Two extreme limits are the high-density and the low-density forms
of liquid water. The high-density liquid consists of a signicant fraction of
5-coordinated (and even a few 6-coordinated) molecules (and also 3-coordinated
ones) while the low-density liquid water consists primarily of 4-coordinated mole-
cules (as in ice) [7]. The densities of the phases could differ by as much as 6%, if
they were to be present at ambient conditions. While the high-density liquid is the
one we observe under ambient conditions, it would be appropriate to discuss the
low-density, 4-coordinated phase as a liquid form only if it is at least a free-energy
minimum and therefore can be regarded as a metastable state. While both high-
density and low-density water forms have indeed been observed experimentally in
the amorphous state and are called HDA and LDA, respectively, the same has not
been observed for the liquid state, and it is still a conjecture. However, the basic idea
is that while the liquid state above 232 K, but below 277 K, is a homogeneous uid,
it locally undergoes transformations by uctuations between a high-density and a
low-density form. The question is: does such a large uctuation really exist in
supercooled water? If so, what is the origin?
In a proposed scenario, the two forms might represent two minima on the free-
energy surface of water when plotted against the conguration coordinates of water
molecules. The uctuation of the HB network can be considered as transitions
between these two forms. If such a description is valid then one can hope to achieve
an explanation of the anomalies without invoking a fully microscopic description,
which could be extremely difcult.
One can make the analysis a bit more quantitative by using the well-known
Landau theory of phase transition where one expands the free energy F in terms of
uctuation in the order parameter as
F = 5F >
1
2
A(t
h
5t
h
>)
2
1
3
B(t
h
5t
h
>)
3
1
4
C(t
h
5t
h
>)
4
(22:1)
where <t
h
> is the average value of the order parameter in water at a given
temperature. We have used the tetrahedrality parameter as a suitable order parameter
in the following discussion but we can also use density and/or the translational order
parameter q. In fact, one needs at least two (or even more) order parameters in this
discussion on low-temperature water. Low density may correspond to a high value
of t
h
and vice versa.
In the equation, the rst term of the Taylor expansion is zero because the free
energy is a minimumat the average value of the order parameter in the parent phase.
According to the Landau theory, near a continuous transition, the coefcient A goes
to zero as
A = a(T T
C
) (22:2)
A form like Eq. (22.1), with A given by Eq. (22.1), ensures large uctuations in the
system because the free-energy cost becomes small as T approaches T
C
. In the case
of the gasliquid critical point, the density is the order parameter and the term A in
Eq. (22.1) above goes to zero as the compressibility diverges. In the case of the
solidliquid phase transition, the term A does not go to zero and the term B is
important. Fluctuations are not too important in the case of a rst-order transition,
such as liquidcrystal.
Is there any evidence of large uctuations in the tetrahedral order parameter t
h
in
supercooled water? While experiments and simulations showan increase in isobaric
specic heat and compressibility on lowering the temperature, there is hardly any
convincing evidence for an increase in the uctuation of the tetrahedrality para-
meter. This is paradoxical because, as mentioned above, the two-state model in
some form or other has been used for a long time to explain the properties of water.
It is always tempting to introduce the concept of a critical point to explain the
increase of uctuations as in the Landau theory. But we lack a clear physical picture
at this point in time. It is certain that the free-energy surface becomes atter with
respect to local density uctuations as the temperature is lowered, which is opposite
to the occurrence in normal liquids. Enhanced uctuations lead to the nucleation of
the crystalline ice forms [8].
Understandably, many models have been proposed over the years to describe the
properties of water. These models together form the language or the folklore among
the scientists and engineers involved in the study of water. In the following we discuss
a few of them that have had the most inuence in unraveling the mysteries of water.
22.2 Reason for density maximum
Because of the popularity of the density maximum at 4C we discuss here a fairly
well-known molecular explanation of these anomalies. This explanation evolves
around the temperature dependence of the fraction of 3-, 4-, and 5-coordinated water
molecules. The fraction of these three shows interesting variation below 10C. As
we approach 10C, the fraction of 3-coordinated water molecules decreases rather
sharply and 4-coordinated ones increase. The number of 5-coordinated water
molecules shows only a slight variation in this temperature range. Such replacement
from 3- to 4-coordinated ones gives rise to an increase in density on lowering of the
temperature. However, near 4C the fraction of 5-coordinated water molecules
starts to drop in favor of 4-coordinated ones as ice-like local arrangements become
energetically more favored. This now leads to a decrease in density. These two
effects together give rise to the density maximum at 4C.
22.3 Reason for large isobaric specic heat of water
The specic heat of water (4.2 kJkg
1
K
1
) is typically two to three times greater than
the specic heat of other common liquids such as acetonitrile (2.23 kJkg
1
K
1
) or
ethanol (2.44 kJkg
1
K
1
), even when the water molecule is much smaller, with
fewer degrees of freedom. The large value of specic heat can be attributed to the
existence of the local quasi-stable low-frequency oscillatory modes. Examples
include hindered translation around 50 cm
1
, intermolecular vibration around
200 cm
1
, and librational modes around 585 cm
1
. In addition, HB breaking and
re-formation also contribute to the specic heat as all of themcontribute to uctuation
in the enthalpy. Note that
C
P
= 5(DH)
2
=k
B
T
2
> = 5(DS)
2
=k
B
> (22:3)
where H and S are uctuations in the enthalpy and entropy of the system.
Exchange of energy due to transitions in these localized oscillatory modes gives
rise to uctuations not present in other liquids such as acetonitrile and ethanol.
22.4 Percolation model of water
More detailed explanation of the thermodynamic and structural anomalies requires
the formulation of theories and models with predictive power. This model was
proposed for liquid water in 1970s by Julian Gibbs and co-workers [2]. It was
obtained by consideration of the two transitions (melting and boiling) which dene
the liquid phase. These transitions were discussed with the aid of two analogies to
well-known phenomena in polymer physical chemistry. In analogy to the helixcoil
transition in polypeptides and polynucleotides, the melting of ice was viewed as a
process consisting essentially of the destruction of the orderly interconnected small
(six-membered) rings of HBs characteristic of the crystal. The fact that the breakup
of interconnected small rings is cooperative, even when unaccompanied by the
breaking of bonds which are not parts of rings, is clearly seen by inspection of the
theory for the putatively analogous helixcoil transition.
The condensation of water vapor was, on the other hand, viewed in analogy to the
solgel transition in reversibly polymerizing systems, an analogy which explains its
cooperativity. Taken together, these interpretations of the phase transitions indicate
that the liquid can be viewed as an innitely and randomly branched gel of (rapidly
interchanging) HBs in which closures of rings (primarily large rings) occur at random
but in which there is no signicant preference for an ordered array of small rings.
These concepts also lead naturally to an interpretation of the triple point and
sublimation. This randomgel model is seen to be consistent with most of the known
properties of liquid water, in particular the radial distribution function, infrared and
Raman spectra, dielectric properties, density maximum, and anomalous properties
in the supercooled region. The difculty of such analogies is the quantication, as
the order parameters are all collective many-body quantities which are not always
easy to measure, even in simulations.
Stanley and Teixeira have introduced a new polychromatic correlated-site perco-
lation model [4], which has the novel feature that the partitioning of the sites into
different species arises from a purely random process that of random bond
occupancy. By polychromatic one thus means that each lattice site is differently
colored according to bond occupancy.
The rst hypothesis of this model assumes that the formation of bonds is
completely random and only temperature-dependent. This means that for a given
molecule, the probability of formation of each of the possible four bonds is
completely independent of the number of other bonds formed by the molecule
that is occupying the lattice point. The central point of the percolation theory is
that it builds a distribution of the cluster of molecules, connected by HBs. In spite of
the randomness of the HB formation, correlation among molecules enters through
energy stabilization due to HBs. To understand how the correlation arises in such a
random model, note that if a molecule forms four bonds with its neighbors, then
these neighbors all have at least one bond already formed. We now group molecules
in terms of the HBs they form and denote them as family, denoted by f
n
. That is, f
0
denotes isolated water molecules with no HBs, while f
1
denotes water molecules with
one HB, and so on. It is clearly impossible to have a molecule belonging to the
family f
0
next to another belonging to the family f
4
. Thus, there is a tendency for
molecules with a high number of HBs to group together in an almost segregative
process. One may more or less identify the presence of clusters of, say, f
4
molecules.
This picture looks very much like a mixture model we use to describe multi-
component liquids. However, note that instead of clusters, it is more meaningful to
speak about regions with a high density of HBs, or, even more exactly, that there is a
correlation length for the distribution of highly bonded molecules that increases
with p (the average bonding probability). Since p increases with decreasing tem-
perature, the correlation length also increases.
The percolation model next assumes that molecules with a larger number of HBs
occupy a larger volume than those with a smaller number of bonds. In other words,
one may associate a molecular volume related to the family f
i
to which the molecule
belongs. The volume increases with i. This rather strong hypothesis is conrmed by
molecular simulations and by X-ray experiments.
An important consequence of the above assumption is the presence of density
uctuations with a non-zero correlation length. That is because a molecule with a
larger than average number of HBs is more likely to be surrounded by other
molecules also with a larger than average number of HBs. In this way, it is possible
to justify the anomalous increase of compressibility with decreasing temperature. At
low temperatures, the number of bonds increases and the density uctuations
increase as well. These correlated uctuations are superimposed on the normal
thermally driven density uctuations present in other non-associated liquids. The
combination of the two competing behaviors yields the compressibility minimumof
the temperature dependence of isothermal compressibility.
The same argument can explain the increase of heat capacity with decreasing
temperature, observed at very low temperatures. Heat capacity is related to entropy
uctuations. Regions of high density of bonded molecules are certainly more
ordered, or have fewer distorted bonds, or have lower entropy than regions where
the presence of free or one-bond molecules dominates.
The maximum of the density at 4C is a consequence of the existence of regions
of molecules with a large number of bonds, which correspond to large molecular
volumes, high enthalpic stabilization due to stable HBs, and low entropy. This is an
unusual situation essentially due to the high directionality of the HBs and to the
tetrahedral structure. In normal liquids, such low-density regions with low entropy
can hardly exist, as entropy increases with volume.
The presence of a second component an impurity, ion, or a dissolved material
breaks the HB network locally. One can then understand why relatively small
amounts of another substance (except heavy water) added to liquid water at low
temperature normally reduce or even suppress the water thermodynamic anomalies.
This is true even for substances forming HBs with water, such as alcohols.
The above percolation model can also explain the effects of pressure, which
turn out to be more subtle. Pressure does not substantially change the number of
bonds, but modies the OOO angles and, as a consequence, the tetrahedral
geometry of the percolating network decreases with increase in pressure. If one
assumes that the probability function p(T), instead of counting the number of
intact bonds, measures the probability of forming a bond with the right tetrahedral
symmetry, then it is a decreasing function of pressure. Pressure, indeed, reduces
the water anomalies.
Rahman, Stillinger, and co-workers have discussed many aspects of the per-
colation process for HB networks in water [3,9]. Sets of congurations selected
from three MD simulations for liquid water have been analyzed for the distribu-
tion of HB clusters. Two simulations correspond to water at 1 gcm
3
, while the
third corresponds to highly compressed water at 1.346 gcm
3
. An energy
criterion was adopted to ascertain the existence of an HB between two molecules.
Thus one can dene a quantity n
HB
as the average number of HBs one water
molecule can make. As the cutoff value for bonding increases (becomes more
permissive), a bond percolation threshold is encountered at which initially dis-
connected clusters suddenly produce a large space-lling random network. For
the model studied, any chemically reasonable denition of hydrogen bond
leads to this globally connected structure through a few disconnected fragments
inhabit its interior.
These pioneering simulation studies clearly demonstrate the presence of the
percolation threshold for water under ordinary conditions. Although some poly-
gonal closures can exist, the critical percolation threshold is apparently well pre-
dicted by Florys theory of the gel point.
The numerical results obtained via an MD/percolation simulation are also shown
to be in excellent agreement with those obtained via the corresponding gelation
simulation of the random polycondensation of tetrafunctional units with intramole-
cular reaction allowed [10].
22.5 Hydrogen-bond network rearrangement dynamics
In the preceding section, we discussed the equilibrium aspects of the percolating
network of extended hydrogen-bonding in liquid water. We now discuss the
dynamics of the rearrangement of this network.
At room temperature, these HBs can break and re-form quite rapidly, leading to
the rearrangement of the network. At the microscopic level, we need to understand
the lifetime and bond-breaking mechanism of individual HBs between two neigh-
boring water molecules. In the case of a water molecule, such bond-breaking events
lead to the spatial/rotational displacements that in turn lead to the translational and
rotational diffusion of water molecules. Therefore, the basic mechanism of the
microscopic dynamics in water is expected to be different from a simple liquid
made of nearly spherical units, like in the case of argon.
22.5.1 Energy landscape view of hydrogen-bond rearrangement dynamics
The energy needed to break an HB in water is 24 kcal/mol and the lifetime is about
24 ps. The length of an HB (oxygenoxygen distance) is about 3.5 , as discussed
in Chapter 1. These bonds break and form at high speed. However, individual bond-
breaking and formation dynamics does not provide an overall description of the
major dynamic events of water. Phenomenologically, liquid water may be approxi-
mately considered as a mixture of ordered, low-density ice-like regions where the
coordination number per water molecule is close to 4 and the tetrahedrality para-
meter is large, and disordered, high-density regions where the coordination number
varies between 3 and 5 and the tetrahedrality parameter is low. Such a description
involves collective coordinates, involving more than one water molecule, such as
the tetrahedrality parameter and also the coordination number of water molecules in
a region. The advantage of employing a collective coordinate involving many
molecules is that one can also employ energy or free energy as a function of such
coordinates to describe transitions between different HB arrangements.
At low temperature, the most stable structure of water is of course crystalline ice,
consisting of six-membered rings of water molecules as basic units, forming an
extended three-dimensional HB network where each water molecule is coordinated
to four other water molecules by HBs. Upon melting, water absorbs a latent heat of
80 cal/g (1.4 kcal/mol), which is equivalent to breaking about 10% of its HBs, and
the system becomes frustrated, with local patches of tetra-coordinated, ice-like
structure. Water molecules can form HBs with neighboring molecules and form
many different local arrangements [11]. These different arrangements are quasi-
stable and correspond to local minima. Water thus has a rugged potential-energy
surface with various energy minima with different HB network structures. Water
undergoes sluggish dynamics on this potential-energy surface. On a short timescale
liquid water is thus amorphous gel-like, while on a much longer timescale it exhibits
diffusional motion as an ordinary liquid. Between these timescales, the HB network
rearrangement occurs intermittently and locally in space, involving the local col-
lective motions of tens of water molecules accompanied by large energy uctuation.
It is, however, not easy to quantify the nature and importance of such
intermediate-length-scale collective uctuations involving may be a few tens of
water molecules. As isobaric specic heat increases with the lowering of tempera-
tures below 273 K, these structures must have different entropy, although similar
free energy. Thus, the candidates are those who are different in enthalpy and entropy
such that one set of structures have low enthalpy and low entropy (like connected
4-coordinated water molecules) and another set have high entropy and high
enthalpy (like a disordered mix of 3-, 4-, and 5-coordinated water molecules). As
already mentioned, neither the distribution of the tetrahedrality parameter nor that of
volume measured by the Vornoi polyhedral shows the existence of a double peak or
shoulder to enable characterization of such differing structures, except near the
280300 K range where one sees a broad distribution. This is seen in different water
models, with slightly varying temperature, which is mostly around 280 K.
To identify the intermittent collective molecular motions associated with the HB
network rearrangement in water, IS analysis is employed, as discussed in Chapter 4.
Inherent structure analysis is used to identify the local minima (called the ISs, which
are considered as quasi-stable structures) of the potential-energy wells, which are
sequentially visited by the system along the trajectory. In the trajectory, the system
undergoes well-to-well transitions (IS transitions) after exhibiting vibrational
motions in individual wells. The sum is over all the water molecules.
Such IS analysis has proved to be quite useful in providing a semi-quantitative
description of water dynamics. From these ISs, one can construct a timedistance
matrix, dened by [5]
D t
m
; t
n
( ) =
N
i=1
[r
i
t
m
( ) r
i
t
n
( )[
2
_ _
1=2
(22:4)
where r
i
(t
n
) is the position of the center of mass of the ith water molecule in the IS at
time t
n
, t
n
= nDt (Dt is a time interval, 10 fs). From this two-dimensional map of t
m
versus t
n
one can nd the basins because D remains localized around a diagonal.
We can obtain certain sets of sequential ISs from island structures (basins); they
are mutually separated with small energy barriers and small structure differences.
The transitions between ISs within a basin (minor IS transitions) occur easily and
frequently [5].
Such an analysis shows that among the ISs visited by the system, there are islands
(basins) of various sizes, distributed randomly in time, and thus the transitions
among the islands (called the basinbasin transitions) occur intermittently. The ISs
of water are different from other liquids. As the temperature is lowered, the system
spends longer and longer times in a few basins. The basinbasin transitions com-
prise large IS changes but may often involve relatively small energy barriers [5].
For water, the intermittent collective motions are associated with the HB network
rearrangement in water, and an analysis employing a Hamming matrix of the graph
theory is employed. A Hamming matrix can quantify the creation and annihilation
of the HBs along an MD trajectory. The HB network can be represented by a
directed graph (digraph). The rearrangement of the HB network then corresponds
to the connection and disconnection of the edges of a digraph. The HB network of a
system consisting of N molecules can be expressed by a N N adjacency matrix,
whose elements a
ij
are equal to unity when the HB is donated from molecule i to
molecule j (one of the hydrogen atoms of molecule i forms the HB with the oxygen
atom of the water molecule j), and zero otherwise.
The Hamming distance between two ISs at times t
m
and t
n
is dened as [5]
D
H
t
m
; t
n
( ) =
ij
a
ij
t
m
( ) a
ij
t
n
( ) (22:5)
where the summation is performed over all different pairs of sites (i,j).
One speculates that in a large system, many intermittent collective motions do
occur independently at a time at different places in the system, and their total sum
yields a band structure in both the Euclidean and the Hamming distance matrices,
seemingly equivalent to the trace of diffusional motion. Therefore, to observe the
intermittent collective motions, we need to detect the structure changes in a local
area of the system, of the order of 10 or so in diameter.
It is then of interest to determine how one can experimentally observe the
intermittent collective motions associated with the HB network rearrangement,
which are buried under vibrations.
A limitation of IS analysis is that off-hand it does not give us information about
the local dynamics. Nevertheless, the ISs obtained from trajactories at different
temperatures reveal considerable variations.
22.5.2 Depolarized Raman scattering prole
There is an indirect way to detect intermittent local collective motions. In the case of
depolarized Raman scattering, the depolarization ratio is sensitive to low-frequency
uctuations in water. Depolarization is the scattering of the polarization of the
electric eld of light in a direction perpendicular to the original direction of
polarization. Each uctuating state has a distinct depolarization ratio. The inter-
mittent character of the dynamics is known to appear as a so-called 1/f frequency (f )
dependence in a power spectrum. The power spectrum is obtained by Fourier
transforming a time correlation function.
Now, in ordinary cases, the power spectrum exhibits a dependence on frequency
that is stronger than 1/f. That is, the intensity in the intermediate frequency range
falls off faster than the inverse of frequency. However, the dependence can become
weaker if the decay of correlation becomes slower than exponential due to repeated
revisits to the same state by the system. The existence of a 1/f spectrum is an
indicator of the existence of intermittent relaxation dynamics. For example, a
Fourier transform of a series of intermittent pulses, the simplest example of it, is
easily shown to be a type of 1/f spectrum.
The spectrum of the depolarized Raman scattering calculated from the trajectory
is plotted in Figure 22.1. We can see that liquid water has a wide Raman scattering
spectrum, which can be divided into three parts, a power law region above
520 cm
1
, a near white noise (that is, independent of frequency) region below
3 cm
1
, and the in-between transition region [5]. Such a baseline prole of Raman
scattering of liquid water is quite different from those of unassociated liquids, which
yield Lorentzian dependence in the low-frequency region. The power lawpart indeed
corresponds to the structural rearrangement dynamics of the HB network structure.
Detailed analysis shows that the power lawpart obtained fromIS analysis yields a
1/f

frequency dependence with an exponent of = 1.3, which is in good agreement
with experimental results. The signals of inter- and intramolecular vibrational
motions, discussed in the previous section, are seen as two shoulders at ~200 and
500 cm
1
, superimposed on this power law prole.
22.6 Low-temperature anomalies
As clear from discussions given above and also in early chapters of the book, the
area of low-temperature anomalies of water has drawn tremendous attention from
scientists of all spheres, particularly from theoreticians and simulation experts.
Despite all these efforts, this is one area which has remained controversial and a
consensus about the origin of the anomalies remains elusive. One of the reasons
perhaps is the presence of a no-mans-land, which is meant to mean the tempera-
ture zone between 155 K and 232 K. When cooled below 232 K, water always
Figure 22.1. Power spectrum of the Raman scattering intensity of liquid water
calculated from IS analysis. The solid line is for the instantaneous structures and
the dashed line is for the ISs along a trajectory. The dashed-dotted straight line
indicates 1/f

with = 1.3 (f is frequency). The system contains 216 water
molecules. Adapted with permission from Acc. Chem. Res., 32 (1999), 741.
crystallizes. Similarly, when heated above 155 K, the amorphous ice again crystal-
lizes. Nevertheless, investigations into the origin of the anomalous behavior of low-
temperature water have revealed much interesting information about water that is
proving useful in the understanding of even high-temperature water.
While there is a lack of consensus among various explanations, it is still mean-
ingful to discuss some of the explanations put forward, as they are interesting and
often quite important contributions. An appealing explanation is based on the
proposal of the existence of a liquidliquid transition in low-temperature super-
cooled water. This assumes the existence of two states of liquid water, analogous to
the two amorphous forms of the ice, namely the high-density amorphous (HDA) and
low-density amorphous (LDA) states (see Figure 22.2). In the liquid state these are
called the high-density liquid (HDL) and the low-density liquid (LDL) states [6].
The phase diagram is shown in Figure 22.2.
The inter-conversion between these two amorphous solid states is achieved by
changing the applied pressure through a rst-order phase transition. This has been
observed experimentally.
Now, it is conjectured that under appropriate conditions this rst-order phase
transition (or coexistence) line between the two solids in the glassy state continues
to extend into the deeply supercooled state of liquid water. In the supercooled state
this rst-order phase transition line could separate the two forms of the liquid water.
Figure 22.2. Phase diagram of water at low temperature. Note the no-mans-land
between T
H
and T
X
. This region can only be accessed by restricting the
crystallization. An example of such a system is water in a nanotube. Note the
extended HDA/LDA rst-order phase transition line into the no-mans-land
region which ends in a critical point denoted by the dot at the end of the line. At
atmospheric pressure if one decreases the temperature, then the region beyond the
critical point is accessed. The gure is reproduced from the thesis of Dr. Pradeep
Kumar. http://polymer.bu.edu/~hes/water/thesis-kumar.pdf.
As mentioned earlier, the form at higher density and lower tetrahedral order is
termed high density liquid (HDL, mostly 4- and 5-coordinated). Another form is of
lower density and higher tetrahedral order and is termed low density liquid (LDL,
mostly 4-coordinated).
The hypothesis is that this extended rst-order phase transition line in the (P,T)
plane could end in a critical point which is called the second critical point of water.
This critical point has been detected in some of the computer simulations for
model water systems. While there exist different values of the critical temperature
and pressure with different theoretical models, TIP5P water model (this model
provides some properties which are in good agreement with experiments, but
discrepancies remain in some others) gives the following values of the critical
parameters: T
C
= 220 K and P
C
= 200 MPa.
In order to understand the utility of this picture, let us nowconsider the gasliquid
critical point in the (P,T) plane. While there exists only a uid state beyond the
critical point, one can still nd that across the line that extends straight above the T
C
the density uctuation in the system shows a maximum when this line is crossed at
different pressures. Thus across this line, the response functions show a maximum.
This line is widely known as the Widom line. Now if one assumes that there exists a
similar Widom line in the supercooled region corresponding to the second critical
point, then one expects the maximum in response function across the line (see
Figure 22.3) [6].
Figure 22.3. Schematic representation of the HDLLDL critical point and Widom
line in low-temperature liquid water. If one follows path , then it crosses the
coexistence line but not the spinodal line, thus it stays in metastable region and no
appreciable changes occur in the thermodynamic response function. However, if one
follows the path , then it crosses the Widom line, which ensures a maximum in the
thermodynamic response function. Adapted with permission fromProc. Natl. Acad.
Sci. USA, 102 (2005), 1655816562. Copyright (2005) Proc. Natl. Acad. Sci. USA.
Experimental verication of all the above hypotheses has not been possible
because, as mentioned above, liquid water cannot be supercooled without crystal-
lization below232 K, while amorphous ice cannot be superheated above ~155 K. So,
nobody has been able to nd the LDL, except in a conned medium. However, the
properties of water are expected to differ in a conned mediumfromthose in the bulk.
Actually, it is even hard to characterize separately and uniquely the two forms that
are often talked about. It might be useful to dene water molecules in terms of HB
coordination number or the specic volume of each molecule. If one categorizes
water molecules according to coordination number (4-coordination for LDL and
3- and 5-coordinated for HDL), then some studies nd that the population of the two
forms of water uctuates with supercooling and the uctuation becomes maximum
at a temperature which may mark the crossing of the Widomline (for bulk water this
temperature is predicted to be ~230 K at 1 atm pressure) of the liquidliquid critical
point. Population uctuation of the 4-coordinated (LDL) and 5-coordinated (HDL)
species in supercooled water is presented in Figure 22.4 [12]. At high (T = 280 K)
and low (T = 230 K) temperatures, the population of both species uctuates quite
randomly. However, near the Widom line (T = 250 K and P = 1 atm for the TIP5P
model of water), the population uctuates between the two values and the uctua-
tion between the two species is anti-correlated with time. This suggests the corre-
lated interconversion between the two states of the system with time. This
phenomenon is called intermittency in the population uctuation. A simple two-
state model can describe this intermittency in the population uctuation quite well.
Such intermittent uctuations also explain the low-temperature thermodynamic
anomalies in supercooled liquid water.
A recently developed model to explain this intermittency observed in the popula-
tion uctuation assumes a bistable potential whose two minima are the HDL and
LDL [12]. By denition, these two forms differ by their specic volumes; HDL has
relatively lower specic volume than LDL. Now, natural total volume uctuations
of the systemat ambient pressure modify the underlying bistable potential as a larger
volume stabilizes the LDL state and vice versa. If one assumes that volume
uctuation is periodic in nature, then the effective potential of the system also
modulates periodically with the same frequency as volume uctuation.
As discussed in the introductory chapter, volume uctuation is directly propor-
tional to the isothermal compressibility (
T
) of the system and therefore behaves
anomalously in supercooled water. The
T
starts increasing with a decrease in
temperature below T = 320 K and has a maximum on crossing the Widom line
(observed in experiments on nano-conned systems).
Simulations, however, fail to provide any denitive signature of the presence of
high-density and low-density liquids. There is no evidence of structural and density
uctuations on a large scale across the Widom line. Though the local uctuations can
be attributed to anomalies such as specic heat, the isothermal compressibility which
is associated with the density uctuation on a large scale is not expected to show a
maximum along the Widom line on the basis of local uctuations. We show in
Figure 22.5 an analysis of the distributions of local molecular volume, the tetrahedr-
ality parameter, and also the structure factor over a wide range of temperatures.
Computer simulations reveal another interesting aspect of supercooled water,
which is the limited propagation of the coordination number [7]. As discussed
earlier, bulk water consists of species with different coordination numbers.
Supercooled water mostly consists of 4-cordinated species along with a small
percentage of 5- and 3-coordinated species. As the tetrahedral network structure is
the stable conguration of the system, 5- and 3-coordinate species are defects in the
system. These defects are of course not static, they propagate in the system.
0 500 1000 1500 2000
t (ps)
N
4
C
/
5
C

(
a
r
b
.

u
n
i
t
)
N
5C
(24 %)
N
4C
(65 %)
(b) T = 250 K
(a)
(b)
Figure 22.4. (a) Temporal variation of the numbers of 4- and 5-coordinated species
in supercooled water at T = 250 K. Note the strong anti-correlated uctuations
between N
4C
(t) and N
5C
(t) and intermittency in the uctuations of both species at
T = 250 K. (b) Corresponding power spectrum of the population uctuation time
correlation function. Note the 1/f dependence of the intensity which conrms the
presence of intermittency at T = 250 K. Adapted with permission from J. Phys.
Chem. B, (Lett.), 113 (2009), 22212224. Copyright (2009) American Chemical
Society.
The mechanism of conversion of a 5-coordinated water molecular species to a
4-coordinated molecular species has already been described in the molecular motion
chapter, and is popularly referred to as the LaageHynes mechanism. This involves
a large-amplitude angular jump of ~ 60. During this process, a 4-coordinated water
molecule gains an extra neighbor from the second solvation shell and becomes
5-coordinated. The 5-coordinated water molecule, however, is short-lived and has
Figure 22.5. Temperature dependence of distributions of (a) local volume, (b) the
tetrahedrality parameter, and (c) the static structure factor. The gure has been
reproduced from private communication with Prof. Shinji Saito.
on average a lifetime of less than 1 ps or so. The decay of the 5-coordinated water
molecule occurs by the departure of a water molecule from the rst solvation
shell; the departing water molecule then enters the rst solvation shell of a new
partner (the second solvation shell water molecule with respect to the decaying
5-coordinated water) to make it 5-coordinated. This propagation of the coordina-
tion is accompanied by the propagation of rotational jumps as each conversion is
associated with an angular jump of ~ 60. The propagation ends at a point where
the propagating 5-coordination number meets a 3-coordinated water molecule
which acts as sink. The mechanism implies that when the 5- and 3-coordinated
water molecules meet each other, two 4-coordinated water molecules are gener-
ated [7].
The above scenario suggests that the number uctuations of 3- and 5-coordinated
water molecules could indeed occur in phase and are anti-correlated with the
4-coordinated species. The transformational events become increasingly correlated
with each other as the temperature is lowered in the supercooled region. This in turn
may indicate uctuations between liquid-like and ice-like domains.
According to the kinetic theory of gases, the string length (l
st
) is the average mean
free path for the collision between 5- and 3-coordinated species, and goes as (l
st
~
3
1
);
3
is the number density of the 3-coordinated species. It is apparent that as
temperature decreases the population of the longer strings increases. Thus, the
string-like motion that connects the events of the inter-conversion between 4- and
5-coordinated species provides a measure of the dynamic correlation length in the
supercooled liquid water [7].
It seems plausible that the enhanced uctuations observed in many simulations
and in some experiments are nothing but a precursor to the liquidliquid transition
that inevitably occurs at 232 K. The role of such pre-freezing uctuations has often
been discussed in many contexts. Recently the same view has been supported by
extensive free-energy calculations.
Therefore one can envisage two different explanations or scenarios: one invokes
the presence of an LDL minimum in the order parameter space (in addition to the
liquid and ice minima), and the second explains the anomalies with only two free-
energy minima (HDL and ice), without the LDL minimum.
A variant of the two-state model of liquid water was used in a recent work by
Patey and co-workers that shows that if we categorize a water molecule in the liquid
by using the tetrahedral order parameter t
h
into liquid-like (lowt
h
) and ice-like (high
t
h
) molecules, and then treat the liquid as a binary mixture, such a binary-mixture
model can indeed reproduce most of the anomalies of liquid water [13] .Thus, there
does not seem to be any need to invoke the existence of the LDL state.
However, there remains the denite possibility of a metastable minimum that
corresponds to the LDL state. Such a possibility is supported by at least two
observations: (i) the existence of amorphous ice at low temperature; (ii) the exis-
tence of LDL in a conned medium.
In fact, a random network model (RNM) of water was proposed a long time ago
by Rice and coworkers [14] who demonstrated via computer simulations and
theoretical analyses of spectroscopic data from different experiments that the low
density amorphous (LDA) phase of ice can contain a signicant number of defects
in the formof (5,7) ring pair. They suggested that the inter-conversion between (5,7)
and (6,6) ring pairs can constitute an elementary excitation of LDA.
Note that irrespective of the models proposed, the enhancement of uctuations in
supercooled water is a reality. It is thus certain that the free-energy surface becomes
relatively at with respect to uctuations in density, in bond orientational and
tetrahedral order parameters, and also in the coordination number of the water
molecules. Such uctuations eventually lead to crystallization.
However, the true reason for the anomalies (resulting fromenhanced uctuations)
observed in low-temperature water is yet to be gured out. Correlation of these
uctuations increases sufciently in range to enable crystallization. Thus, these
uctuations are local but at the same time may involve 3050 water molecules that
is, involve second- or third-nearest neighbors. The question remains as to the cause
of such an increase in uctuations as the temperature is lowered.
22.7 Conclusion
Many of the anomalies of liquid water may be (or rather, are expected to be)
understood in terms of the two length scales and the two energy scales that are
needed to describe the interaction between any two water molecules. These two
length scales are the diameter of the individual water molecule and the length of the
HB. The two energy scales are the energy of non-hydrogen-bonded interaction
between water molecules and the energy of the HBs. Another important parameter is
the maximum number of stable HBs that a water molecule can form. At low
temperature where ice is the predominantly stable form in ambient conditions, the
structure is determined by the HB energy and the HB length. In ice these two
parameters along with the number of stable HBs can explain the stability of ice.
While in the liquid phase such a regular arrangement is disrupted by the entropic
disorder that favors the presence of molecules with many different coordination num-
bers, the inuence of ice-like 4-coordinated structures remains signicant at tempera-
tures between 242 and 320 K. As discussed earlier, the density anomaly can be
understood by the decrease of 3-coordinated water molecules in favor of 4-coordinated
ones till 4C and the decrease of 5-coordinated ones in favor of 4-coordinated below
4C. As discussed in this chapter, many such anomalies can be understood by using the
simple phenomenological approach, although it is not quantatively rigorous.
22.7 Conclusion 341
The above statements are indeed true and can be seen in a two-dimensional water
model created by Ben-Naim [15] and developed further by Dill et al. [16]. This
model is popularly known as the Mercedes Benz (MB) model, where water
molecules are represented as circles with three equivalent arms pointing outwards
fromthe center 120 apart. Strongest hydrogen bonds formwhen two arms fromtwo
different spheres point towards each other and are separated by a certain distance
which is larger than the sum of the radii of the two spheres participating in the HB.
Thus, HBs in the MB model have the same characteristics as the HB in three-
dimensional real water molecules. This model satises the criteria of two length
scales and two energy scales and a xed number of HBs, i.e., three. It has been
shown that the MB model displays many of the water anomalies in the liquid state,
for example, the hydrophobic/hydrophilic effect, the density maximum, etc. At the
strong HB limit and at low temperature it crystallizes to form a honeycomb lattice.
The analogies with the MB model establish the ability of a simple model to
capture many of these anomalous features. Nevertheless it is still more complex than
a simple liquid such as for example liquid argon, which popularly used by theore-
ticians as it involves at least ve parameters.
Interestingly at the low HB energy limit the MB model crystallizes to a triangular
lattice just as under a two-dimensional argon-like potential.
In this context we should mention the work of Lynden-Bell, who showed that if
the strength of the HB interaction is decreased relative to the shorter-ranger
LennardJones interaction, then the water-like anomalies disappear in three-
dimensional water [17]. In this case one nds that lowering the HBenergy gradually
removes the peak in the oxygenoxygen radial distribution function and makes the
microscopic structures look like a normal liquid.
However, many other anomalies of water can be explained in terms of models of
water that employ essentially the four parameters mentioned above. In some cases
the quantum nature of hydrogen and the quantum nature of the lone pair of the
oxygen atom must be taken into account. This is clearly apparent in the interaction
of water with charged solutes and also in the determination of the pH of water, as
discussed in Chapter 5. These are difcult problems to understand and remain very
much in the realm of recent research activities.
We have discussed the difculties we faced in understanding water between
273 and 232 K, although considerable progress has been made in recent years.
There remain many interesting features of such liquid unexplored in the role of
chemistry and biology [18,19]. We have discussed in Chapter 15 the elegant
explanation by Marcus and co-workers of the effect of hydrogen atoms of water
in enhancing the rates of enzymatic catalysis by ve to six orders of magnitude.
Water might play such vital roles in many biological and chemical processes
that are still unexplored.
References
1. W. Kauzmann, in Water, a Comprehensive Treatise (F. Franks, ed.), vol. 7 (New York:
Plenum, 19721982).
2. J. H. Gibbs, C. Cohen, P. D. Fleming, and H. Porosoff, Toward a model for liquid water.
J. Solution Chem., 2 (1973), 277.
3. F. H. Stillinger and A. Rahman, Molecular dynamics study of liquid water under high
compression. J. Chem. Phys., 61 (1974), 4973.
4. H.E. Stanley and J. Teixeira, Interpretation of the unusual behavior of H
2
O and D
2
O at
low temperatures: tests of a percolation model. J. Chem. Phys., 73 (1980), 3404.
5. I. Ohmine and S. Saito, Water dynamics: uctuation, relaxation, and chemical reactions
6. P. H. Poole, F. Sciortino, U. Essmann, and H. E. Stanley, Phase behavior of metastable
water. Nature, 360 (1992), 324328; L. Xu, P. Kumar, S. V. Buldyrev, et al., Relation
between the Widom line and the dynamic crossover in systems with a liquidliquid
phase transition. Proc. Natl. Acad. Sci. USA, 102 (2005), 1655816562.
7. B. Jana, R. S. Singh and B. Bagchi, String-like propagation of the 5-coordinated defect
state in supercooled water: molecular origin of dynamic and thermodynamic anomalies.
Phys. Chem. Chem. Phys. 13 (2011), 1622016226.
8. C. A. Angell, Highs and lows in the density of water, Nature Nano (News & Views), 2
(2007), 14.
9. F. H. Stillinger, Theory and molecular models for water. Adv. Chem. Phys., 31 (1975),
1101.
10. B. Bagchi, E. Donoghue, and J. Gibbs, Agreement between the gelation and molecular
dynamics models of the hydrogen-bond network in water. Chem. Phys. Lett., 94 (1983),
253258.
11. C. A. Angell, Formation of glasses from liquids and biopolymers. Science, 267 (1995),
19241935.
12. B. Jana and B. Bagchi, Intermittent dynamics, stochastic resonance and dynamical
heterogeneity in supercooled liquid water. J. Phys. Chem. B (Lett.), 113 (2009),
22212224.
13. S. D. Overduin and G. N. Patey, The interaction of patterned solutes in binary solvent
mixtures. J. Chem. Phys., 124 (2006), 94901.
14. M. G. Sceats, M. Stavola, and S. A. Rice, A zeroth order random network model
of liquid water, J. Chem. Phys. 70 (1979), 39273938; M. G. Sceats and S. A. Rice,
A random network model calculation oF the free energy of liquid water. J. Chem. Phys.
72 (1980), 61836192.
15. A. Ben-Naim, Statistical mechanics of waterlike particles in two dimensions.
I. Physical model and application of the PercusYevick equation. J. Chem. Phys., 54
(1971), 368295.
16. K. A. Dill, T. M. Truskett, V. Vlachy, and B. Hribar-Lee, Modeling water, the hydro-
phobic effect, and ion solvation. Annu. Rev. Biophys. Biomol. Struct., 34 (2005),
17399.
17. R. M. Lynden-Bell, Towards understanding water: simulation of modied water mod-
els. J. Phys.: Condens. Matter, 22 (2010), 284107.
18. J. Frenkel, Kinetic Theory of Liquids (New York: Dover, 1955).
19. D. T. Limmer and D. Chandler, The putative liquidliquid transition is a liquidsolid
transition in atomistic models of water. J. Chem. Phys., 135 (2011), 134503.
References 343
Epilog
In this book we have attempted to bring together some of the recent developments
on the role of water in diverse biological and chemical processes, emphasizing all
through a molecular approach. Due to the complexity of the systems, progress in
this area has been slowand discussion often incomplete. Nevertheless, we have tried
to describe how water as a small dipolar molecule is a part and parcel of complex
systems and often dictates the proceedings in the microworld, although this role is
often not obvious from the outcome.
We have examined the molecular features that are responsible for the unique, and
often termed anomalous, properties of water. We summarized the thermodynamic
and dynamic properties of bulk water, including the temperature dependence of the
pH of water. Many aspects of water, such as the ultrafast SD of charged species in
water, have been discovered only in the last one or two decades. In order to identify
the effects of solute molecules on water structure and dynamics we provided a brief
summary of the timescales of motion of water molecules in the bulk. In particular,
we noted the ultrafast timescale exhibited by water on many occasions. These
timescales undergo change in the surface of biopolymers.
Of great interest in both chemistry and biology is the role of water in the
biological functions of proteins, DNA, and lipids, etc. This is a hard problem. In
many of the early studies on proteinwater interactions, water was approximated as
a continuumsolvent. For example, in protein rotation, one endowed the protein with
a rigid boundary layer of water with the combined unit rotating in a continuum
solvent. In the case of DNA, one employed concepts from electrostatics such as the
double layer and employed the PoissonBoltzmann equation. Such an approach was
bound to be inadequate. Nevertheless, it was pursued for quite some time. Such a
macroscopic approach fails to provide dynamic information at a molecular level.
On the other hand, any sophisticated theoretical approach to treat water around
biomolecules faced several non-trivial problems. First, the properties of biomolecules,
such as the charge character and ruggedness (like the length of amino acid residues),
345
vary over a small length scale. The biomolecules themselves are of intermediate or
mesoscopic scale compared to the size of water. Because of the complex, short-
range, sharply varying nature of the interaction potential that water molecules face
near the biomolecular surface, an analytical approach to the problem is virtually
impossible. Only simple phenomenological models, such as the dynamic exchange
model developed by Nandi and Bagchi, seem to offer (limited) success in providing
a physical picture.
Under such difcult conditions, computer simulations have provided a highly
effective tool to investigate the coupling between water dynamics and biomolecular
motion and function. Molecular dynamics simulations have been performed with
increasingly accurate potential models and force-elds and longer timescales to
address detailed questions. For example, it is now quite routine to perform 100 ns
simulations of intermediate-sized proteins and tens of thousands of water molecules.
As a result, a large number of studies are being performed by many groups around
the world.
While considerable progress has been made in simulating the dynamics of water
molecules around proteins and DNA, relatively less success has been achieved in
the study of the biomolecules themselves as they move on much smaller timescales.
Thus, the study of the detailed role of water in biological functions is still at its
infancy.
We have also addressed many aspects of water in contact with different types of
solute molecules. Here the systems are a bit simpler to deal with both theoretically
and computationally. Nevertheless, many fascinating discoveries have been made
only recently, such as the anomalous composition-dependence of aqueous binary
mixtures, such as water and DMSOand water and ethanol. In addition to anomalous
bulk properties, these mixtures show strongly composition-dependent solvation
properties for biomolecules. The enzymatic properties of several proteins also
change with composition in a way that is not easy to rationalize.
We often use concepts developed for macroscopic systems to describe processes
where such concepts might not be applied. One such case is the use of entropy in
biological systems. Nevertheless, such concepts can provide a semi-quantitative
rationalization, as discussed here.
We have also described several advanced topics devoted to neat bulk water, such
as the freezing of water and also supercritical water. Both have attracted consider-
able attention in recent times. The low-temperature anomalies of water are slowly
being understood, although the eld remains the subject of lively debate.
A fact that has retarded progress in many of the areas discussed here is the
absence of experimental investigations. For example, study of the hydration layer
is hard as the layer is often very thin and the inuence of the bulk phases on
experimental observables cannot be ignored. In fact, different experimental
346 Epilog
techniques give somewhat different results. Thus, we need to keep in mind the
technique employed.
Among the new experimental techniques that have provided valuable new data
and new insight are single-molecule spectroscopy and non-linear optical techni-
ques. Single-molecule spectroscopy coupled with laser spectroscopy has given
information about enzyme kinetics, protein folding, and proteinDNA interaction,
among many other subjects. Note that single-molecule spectroscopy provides
information on a timescale which is typically milliseconds or longer. Therefore,
this technique provides a window that is not accessible by computer simulations.
Non-linear optical techniques are increasingly being applied to study water in
complex systems as this technique provides information about the structure and
dynamics of the local environment.
The role of water in promoting health and controlling diseases has been discussed
from antiquity. Studying the structure and dynamics in the bio-world might provide
an explanation and give the much-needed pathway of development of diseases such
as cancer and Alzheimer. There is clearly a war that is going on within our body
24 hours a day, and water is very much an active participant. But we do not yet know
the detailed role of water, which cannot be ignored if molecular-level understanding
is required. This molecular-level approach is now pursued vigorously and we can
look forward to many exciting developments in the future. We hope that the present
monograph has captured a part of this process.
Epilog 347
Index
Figure, table, and reaction locations are indicated in bold typeface
acetone
as amphiphilic binary mixture, 252
molecular structure, 244
acidosis, 75
AdamGibbs relation, 157158, 158, 293294, 295
adenylate kinase (ADK) (enzyme), 101, 101
Alzheimers disease, 109110, 123
amphiphilic effects, 207209, 243258, 244, 247, 249,
251, 255, 257, 342. See also surcants,
hydrophobic effects, hydrophilic effects
Angel, C. A., 308
anomalies. See also Widom line
amphiphilic binary mixtures, 245253, 247, 249, 251
bulk water, 1321, 14, 16, 17, 19, 20, 21, 22
explained by computer simulation for hydration
layer, 144
glass transition, 8890, 89
ice formation, 306, 308
large number of in water, 323326
structural. See local order
of supercooled water, 310
of supercooling, 324
thermodynamic. See specic heat. See temperature of
maximum density (TMD), isothermal
compressibility (Kt), coefcient of thermal
expansion (/f1ar/f0)
and two-stage water model, 2223
aqueous salt solutions. See electrolytes
Arrhenius equation, 1718
association
hydrophobic, 227
residence time of water molecules in proteins, 109
Bagchi, B., 86, 130
Barron, L. D., 196
Bell, L., 342
Ben-Amotz, D., 227
bending mode (hydrogen bond), 39, 69
Bernal, J. D., 324
Berne, B., 56, 57
Bhattacharyya, S. M., 88, 130, 265
bifurcated hydrogen bond, 6768
billiard ball model (of liquids), 324325
biological water. See also bulk water
DNA, 83
functions in, 97113, 99, 100, 101, 103, 104, 105,
106, 108, 113
inside a carbon nanotube, 277283, 278, 279, 281,
282
molecular characteristics of, 8384
molecular differences with bulk water, 8183, 82
natural selection in biomolecules, 187192, 189
protein hydration layer, 82, 83, 8890, 89
and protein synthesis, 192197, 195
theoretical studies of, 8488, 9195, 91
blood
lipid bilayers in, 178, 178
pH, 7576
Boltzmann law, 288, 291, 298, 303
Bondi, A., 227
bound water molecule (biological water). See also free
water molecules (biological water)
denition, 84
in dynamic exchange model, 8688
in lipid surface, 180
in micelles, 266268, 267
bovine serum albumin (BSA), 128129
Brown, Robert, 27
Brownian motion, 2728, 29, 37, 51, 52
bulk water. See also heavy water, biological water
anomalies, 1321, 14, 16, 19, 20, 21, 22
characteristics of, 3, 79, 8
computer simulations of, 67
freezing of, 305315, 310, 312, 313
inherent structures in, 6170, 63, 64, 65, 66
modeling, 910
molecular differences with biological water,
8183, 82
molecular structure of, 47, 4, 5, 324
pH, 7175, 73
349
carbamide. See urea
cell theory, 298299, 299
chaotropes. See ionization
chromatography, 202
clathrate hydration molecular structure, 124, 132133
clusters. See also percolating network
amphiphilic effects and, 253254, 255
DMSO, 246249, 249
hydrogen bond, 67, 330
coefcient of thermal expansion (ar), 1617, 17
collapse, 229
computer simulation. See also spectroscopy, nuclear
magnetic resonance experiments (NMR), inherent
structures (IS), experiments
and protein hydration layer thickness, 121, 124
dielectric relaxation, 143
DMSO, 245249, 247, 249
DNA hydration, 158, 171
to explain water anomalies, 144
hydration layer, 140142, 141, 142, 146, 146
of ice formation, 308, 310314, 312, 313
lipid bilayer, 180181, 181
of molecular motion, 31, 136138
molecular motion in hydration layer, 139
of polarizable water molecules, 41
of protein glass transition phase, 144145, 145
solvation dynamics (SD), 142143
in surcants, 266, 269
of water conned between silica surfaces, 204205,
204, 205, 206
of water density, 15
water glass transition, 89, 144145, 145
concentration
dependence in DMSO, 245249, 247
dependence of conductivity in electrolytes, 211
dependence of ions solutions and water dynamics,
203204
conductivity. See polarization
continuum model (of collective orientational
relaxation), 54
coreshell model, 270273, 270, 271, 272
covalent bonds. See also hydrogen bonds
rarity of breaking in bulk water, 9798
transfer of electron density in, 8, 8
of water creating V shape, 4, 4
Crick, F. H. C., 152
crystallographic experiments, 167169
da Vinci, Leonardo, 97
DAngelo, M., 268
Darwin, Charles, 187
Dawkins, Richard, 187
DebyeHuckelOnsager law, 45, 46, 210211
DebyeWaller coefcient, 203
density uctuations. See also temperature of maximum
density (TMD)
of protein hydration layer, 137
in supercritical water, 318319, 319
of water on silica surface, 204205, 205, 206
density maximum. See temperature of maximum
density (TMD)
dielectric constant
being useful in chemical processes, 8
and light scattering, 5557
in lipid bilayer, 184
polarization increases, 9
dielectric relaxation (DR). See also relaxation time,
electrolytes
computer simulation, 143
DNA, 83
hydrogen bond breaking kinetics, 4041
and protein hydration layer, 83, 120, 124125,
125
in reverse micelles, 268269
diffusion. See also viscosity, relaxation time
AdamGibbs relation, 157158, 158, 293294, 295
along DNA 174, 175
and inherent structure not containing
information on, 70
of ions in bulk water, 4546, 46
of ions in methanol, 250
Rosenfeld relation, 291293, 295
single le, 277, 280, 281
Dill, K. A., 342
dioxane
DMSO
as amphiphilic binary mixture, 245249, 247, 249
biological applications of, 256258, 257
clustering in, 253254, 255
DNA. See also protein synthesis
and drug recognition, 107109, 108
effects of nanoconnement in, 161
entropy and diffusion in, 156159, 158
groove structure in, 153155, 154
hydration of constituents in, 152153, 163164, 164
intercalaion of drugs into, 101105, 101, 103,
104, 105
and protein hydration layer dynamics, 167175, 171,
175
replication sequencing, 188, 196
solvation dynamics in, 155156, 156
spine of hydration in, 159160
stabilizing effect of water in, 151152
drinking water. See bulk water
dynamic equilibrium
of bound and free biological water
molecules, 86, 136
in dynamic exchange model, 9195, 91
dynamic exchange model, 85, 85, 8688, 9195, 91
Einstein, Albert, 28
Einstein relation law, 45
electrolytes. See also polarization, ionization, dielectric
relaxation (DR)
conductivity in, 209211
electron transfer in, 46, 4749, 48
ionic conductivity in, 4546, 46
350 Index
polarization, 30
viscosity in, 211212
electrons. See molecular structure
ellipsoid in a sea of spheres model (EISS), 5152
Elsaesser, T., 40
energy
entropy balance, 290
microscopic states of, 15
similarities between water density at different
temperatures, 18
water molecules can form many structures, 9
enthalpy (H). See also thermodynamics, entropy
in bulk water, 2324
of DNA protein hydration, 168
in hydrophobic effects, 215, 217, 221
stability of bound water molecule, 84
of water molecules near ions, 203
entropy. See also thermodynamics, enthalpy (H)
calculation of, 295300, 299
denition of, 287290
and diffusion, 291294, 295
of DNA hydration interaction, 156157
of DNA protein hydration, 168
during incalation, 104105, 105
in hydrophobic effects, 215, 217, 221
in ice formation, 305
in inherent structures, 61
in micelles, 267, 268, 268
and molecular congurations, 15, 2324
and vibrational molecular motion, 289290, 296, 302
of water molecules near ions, 203
enzyme catalysis
in aqueous urea solution, 208209
covalent bond in, 9798
role of water in, 99101, 99, 100, 101
enzyme kinetics. See enzyme catalysis
ethanol
as amphiphilic binary mixtures, 250, 251
Evans, D. J., 217, 219
experiments. See also spectroscopy, nuclear magnetic
resonance experiments (NMR), computer
simulation
crystallographic, 167169
light scattering, 5557
NALMA, 128
NMR, 5758, 126127, 170
QENS, 127128
extended network. See percolating network
Fayer, M. D., 33, 274
Fersht, A. R., 188, 192194
bril growth, 111112
Fleming, G. R., 129, 325
Flory, Paul, 227, 228, 330
FloryHuggins theory, 228230, 229
uctuation. See nucleation
uorescence up-conversion technique, 170
food, 76
force constant matrix, 69
force law (hydrophobic), 234
Frank, H. S., 217, 219
free energy
barriers, 146, 146, 180, 182184, 185
change in DNA hydration, 157
surcants and, 266268, 267, 268
free water molecules (biological water). See also bound
water molecule (biological water)
denition, 84
in dynamic exchange model, 8688, 9195
in micelles, 266268, 268
freezing 307
biological water, 310, 314315
bulk water, 305309, 310, 310, 312, 313
Fuoss, R. M., 212
Geissler, L., 74
Gibbs, J., 325, 327
glass transition phase (water), 144145, 145
grooves (DNA)
entropy in, 299300
molecular motion in, 154155
structure, 153155, 154, 159
Grote, R. E., 48
GroteHynes theory, 4849
Grneisen, E., 211
guanidinium hydrochloride, 209
GuoyChapman layer 262, 263
Halle, B., 127
Hamming matrix, 332333
Hang-Jun, L., 279
Hansen, E., 51
heavy water. See also bulk water. See also biological
water
effect of temperature on, 3335, 34, 35
freezing of, 309
supercooling, 3335
and vibrational spectroscopy, 128129
Henchman, R. H., 298
Herschbach, D. R., 227
heterogeneous surface topology
in DNA protein 170, 171
of protein hydration layer, 135
in proteins, 122
in RNA, 161162
Hopeld, J. J., 188
HopeldNinio scheme, 190192
hydration layer. See also Stern layer, protein hydration
layer
computer simulations of, 136138, 139
surface topology, 124
hydrodynamic friction, 121
hydrogen bond breaking
breaking in hydration layer, 139
in hydration layer, 137138
kinetics, 3649, 38, 39, 40, 44, 46, 48
hydrogen bond defects. See orientational order
molecular structure
Index 351
hydrogen bond lifetime
and anharmonic coupling, 3940
geometric denition, 36, 38
in micelles, 266
quantication, 50, 5859, 324
and time correlation functions, 3639, 38, 39
hydrogen bond network
being percolating, 67
uctuations in, 324, 330334
impossibility near large hydrophobic object,
234235, 235
low energy excitations in liquid water, 69
on mica surface, 207
micelle disruption, 263
not sustained in biological water in three dimensions,
84
protein, 124
in supercritical water, 318, 321
hydrogen bond types
in biology, 6768, 8183
in DNA, 163164, 164
in micelles, 266268, 267, 268
RNA interactions, 105107, 106
of silica surfaces, 204, 205
strength difference in protein backbone and side
chain atoms, 121123, 122
in urea water, 209
hydrogen bonds. See also inherent structures (IS),
covalent bonds
bifurcated, 5, 6
and difculty in ice creation, 310
diversity of in water, 8
uctuating molecular networks in water, 9
lifetime of, 7, 140142, 141, 142
long-lasting, 311314, 312, 313
orientational order in, 50
potential energy, 324, 331333
tetrahedral structure, 5, 50, 7172
hydrolases (enzyme), 99101, 99, 100
hydrolysis, 187196, 189, 195
hydropathy scale, 220221, 222
hydrophilic effects. See also hydrophobic effects,
amphiphilic effects
on electrolytes, 209212
in ion solvation, 203
on mica surface, 207
on parallel silica surfaces, 204205, 205
on protein surface, 122, 123, 124, 132133, 133
hydrophobic effects. See also hydrophilic effects,
amphiphilic effects
collapse, 227230, 229
at different length scales, 234235, 235
environment in lipid bilayer, 184
force law, 234
history of, 215217
and hydrophobic hydration, 217220, 220,
221, 222
in iceberg model, 217, 219
ice-like water structures on silica, 205, 206
of ions due to enthalpy and solvation energy, 203
molecular interactions in, 230233, 236241
in nanotubes, 280
and pair hydrophobicity, 221227, 223, 224, 225,
226
on protein surface, 122, 123, 132133, 133
Hynes, J. T., 3132, 40, 48
ice. See also supercooling, nucleation
density of, 14
formation, 305306, 308, 309, 310, 312, 313
formation in carbon nanotubes, 310, 314315
from micro-droplets, 308
phase diagram of water into, 306307, 307
polymorphs in, 9
tetrahedral molecular of, 6
iceberg model
hydrophobic effects, 217, 219
of protein hydration layer, 118, 119, 124
inherent structures (IS). See also hydrogen bonds,
computer simulation
bond transition in, 6768
temperature and, 6266, 63, 64, 65, 66
intercalation, 101105, 101, 103, 104, 105
interfacial water. See protein hydration layer
inverted hydration molecular structure, 124, 132133
ionization. See also electrolytes
auto, 7172, 74
conductivity, 4546, 46
and water, 4749, 48, 202204, 203
Ising model, 273274, 275, 294
isobaric specic heat (Cp). See specic heat (Cp)
isoenergetic structural arrangements. See polymorphs
isothermal compressibility (Kt), 1516, 16, 2324
Jimenez, R., 43
JonesDole coefcient, 203204
JonesDole equation, 212
jump motion. See rotational molecular motion
Kauzmann, Walter, 215
kinetic proofreading (KPR), 187196, 189, 195
Kohlrauschs law, 210
kosmotropes. See ionization
Kubo, R. J., 203
Kubo-Oxtoby theory of frequency modulation, 40
Laage, Damien, 3132
LaageHynes mechanism, 339
Landau theory, 325326
Lang, M. J., 45
Laria, D., 321
Levitt, M., 105
light scattering experiments, 5557
linear molecular motion. See translational molecular
motion
lipid bilayer. See also protein hydration layer
hydration of constituents in, 179
molecular structure of, 177179, 178
molecular transport in, 182184
potential energy in, 180, 184
solvation dynamics (SD), 181, 182
water dynamics in, 180181, 181
352 Index
lipid bilayer diffusion series (LPD), 182184
local density. See density uctuations
local order, 1921, 19, 20, 21
low temperature. See supercooling
lubricant
ickering phenomena, 196197
water as, 179, 185
Lynden-Bell, R. M., 46
lysozyme (enzyme), 100101, 101, 128, 129, 130, 137,
256258, 257
magnetic relaxation dispersion (NMRD), 127
Maniwa, Y., 283
Marcus theory (of electron transfer), 4749, 48
Marcus, R. A., 130, 342
Matsumoto, M., 310
mean square displacement (lipid bilayer), 180, 181
Mendeleev, D., 250
Mercedes Benz model, 342
metastable state, 310311, 310, 341
methanol
mica, 207
micelles, 261263, 263, 342 See also reverse micelles
microemulsion. See reverse micelles
model
billiard ball, 324325
continuum, 54
coreshell, 270273, 270, 271, 272
dynamic exchange, 85, 8688, 9195, 91
EISS, 5152
iceberg, 119, 217, 219
Ising, 273274, 275, 294
reactiondiffusion model, 86
two-stage water, 2223, 310, 325, 335341, 335, 338
WeeksChandlerAndersen, 236, 338
molecular motion
Brownian, 2728, 29, 37, 51, 52
of bulk water, 2735, 29, 31, 33, 34, 35, 36, 4950
rotational, 27, 2832, 31, 32, 5153, 54, 85, 8688,
9395, 154155, 180181, 181, 265, 280282,
282, 303
surcants, 265
translational, 27, 28, 35, 36, 85, 85, 88, 137, 154,
180181, 181, 265, 279280, 281
vibrational, 3940, 40, 128129, 289290, 296, 302,
320321
molecular structure. See also polymorphs
amphiphilic binary mixtures, 243245, 244
of DNA, 151152, 153155, 154, 159
and entropy of liquid water, 296297
in hydrophobic effects, 230233
lipid bilayer, 177179, 178, 182184
and local order, 1921, 19, 20, 21
and potential energy, 6266, 63, 65, 66
of protein hydration layer, 121124, 122, 132133,
133
of RNA, 152
surcants, 261263, 263, 264
of water around ions, 203
Moras, D., 106
myoglobin, 124125, 125, 136
N-acetyl-leucine-methylamide experiments
(NALMA), 128
Naim, Ben, 342
Nandi, N., 45, 86
nanotubes (carbon), 18
entropy in, 299300
freezing of water in, 310, 314315
molecular structure of, 278
molecular structure of water in, 278279, 278, 279
relaxation time, 1718
rotational molecular motion of water in, 280282,
282
translational molecular motion of water in, 279280,
281
types of, 277, 278, 282
natural selection (in biomolecules), 187192, 189
Nee, T., 54
Nernsts law of electrochemistry, 45
Nibbering, T. J., 40
nuclear magnetic resonance experiments (NMR). See
also spectroscopy, experiments, computer
simulation
of DNA hydration interaction, 170
and protein hydration layer dynamics, 126127
and relaxation time, 5758
nuclear overheusser effect (NOE), 126127
nucleation, 1617, 309, 310, 311314, 312, 313, 342
See also ice
Ohmine, I., 67, 310, 325
oligomerization, 110111
Onsager, L., 43, 212
Onuchic, J. N., 209
orientational order molecular structure 6
causing ve-sided shape, 5, 6
dependence in hydrophobic effects, 224227, 225,
226
hydrogen bonds, 50
on mica surface, 206207, 207
promoting diffusion, 6768
relaxation in, 5354
reverse micelles, 269273, 270, 271, 272
Ostwalds dilution law, 211
pair correlation function g(r), 1921, 19, 20, 21
pair hydrophobicity, 221227, 223, 224, 225, 226, 233
Patey, G. N., 340
Pauling, Linus, 4
Pecora, R., 56, 57
percolating network. See also clusters
allowing many dynamic processes, 9
development of, 325
history of, 327330
as reason for many anomalies, 67
PercusYevick equation, 288
pH
blood pH, 7576
of bulk water, 7175, 73
Index 353
pH (cont.)
seawater, 77
phase. See also supercritical water, supercooling
diagram (waterice), 306307, 307
glass transition, 8890, 89, 144145, 145, 307
metastable state, 310311, 310, 341
phospholipids, 177179, 178, 179
photosynthesis, 112113, 113
polar perturbations. See solvation dynamics (SD)
polarization. See also electrolytes
of DNA, 151152, 163164, 164
speed of, 44, 45
and water molecule arrangement, 9, 10
polymorphs, 202, 206, 342. See also molecular
structure
potential energy 63
bond transition in inherent structures, 6768
in DNA, 161
of hydrogen bonds, 324, 331
and molecular structures, 6266, 63, 64, 65, 66
potential energy minima. See inherent structures (IS)
potential of mean force (PMF), 221227, 223, 224, 225,
226
PrattChandler theory (PC), 232, 233, 236241
protein folding
ickering phenomena, 196197
hydrophobic effects, 220, 224227, 225, 226
water dynamics in, 109
protein hydration layer. See also surface topology,
hydration layer
association in, 90
binding sites, 107109, 108
and DNA, 167175, 172, 175
glass transition and, 8890, 89
molecular structure, 119, 122, 125, 130,
131, 133
and water residence time, 109, 136, 170
protein surface
inverted molecular structure, 124, 132133
topology, 8, 10
water behavior in, 51
protein synthesis. See also DNA
ADK, 101, 101
enzyme catalysis, 97101, 99, 100, 101,
208209
evolution of, 187192, 189
kinetic proofreading, 187196, 189, 195
proteins
amphiphilic effects on, 245
data bank, 224227, 225, 226
denaturization, 208
effects of DMSO on, 256258, 257
pH of amino acids, 7576
pump-probe spectroscopy, 269, 270
quantication of spatial order (to), 2021, 20, 21,
2425
quantum nature 6
of hydrogen bonds, 8, 7172
of temperature dependence in water bonds, 74
of water creating V shape, 6, 7172
quasi elastic neutron scattering experiments (QENS),
127128
quenched normal mode, 69
Radhakrishnan, R., 314
radical distribution function. See pair correlation
function g(r)
Rahman, A., 6, 330
Raoults law, 245
Rasaiah, J. C., 46, 278, 279
rate of decay, 3638, 38, 39
rate of dissociation, 72
RayleighBrillouin light spectrum, 55
reactiondiffusion model, 86
recognition (DNA), 152, 168169
relaxation time. See also diffusion, dielectric
relaxation (DR)
being collective at low temperatures, 4950
in DNA protein, 170
in inherent structures, 6263, 64
nanopores, 1718
non-exponential in biological water, 84, 87
and nuclear magnetic resonance, 5758
surcants, 264, 268269, 273274, 274, 275
of water between mica surfaces, 206
residence time (in proteins), 109, 136, 170
reverse micelles. See also micelles
dielectric relaxation, 268269
entropy in, 299300
molecular structure of, 263
orientational order molecular structure, 269273,
270, 271, 272
relaxation time, 273274, 274, 275
solvation dynamics (SD), 269
Rey, M., 40, 321
ribonuclease-A, 136
RNA
trapped water molecules in, 105107, 106
water dynamics around, 161162
Rog, T., 180181
Rosenfeld relation, 291293, 295
rotational molecular motion 27
in bulk water, 27, 2832, 31, 33
in DNA grooves, 154155
in dynamic exchange model, 85, 85, 8688, 9395
entropy for, 303
and jumping, 3032, 31, 32, 51
in lipid bilayer, 180181, 181
in surcants, 265
and time correlation functions, 30, 5153, 52, 54
of water inside a carbon nanotube, 280282,
282
rugged landscape, 174, 180
SackurTetrode equation, 288, 299, 300, 301
Saito, S., 310
scaled particle theory (of hydrophobic hydration),
231232
seawater (pH), 77
single le diffusion, 280, 281
Skinner, J. L., 40
354 Index
solvation dynamics (SD)
of bulk water, 13, 4245, 44
computer simulation, 142143
in DNA, 155156, 156
lipid bilayer, 181, 182
and protein hydration layer, 129131, 130, 131
of reverse micelles, 269
of supercritical water, 321322
solvents
acetone, 244, 252
dioxane, 244, 252253
DMSO, 244, 253254, 255, 256258, 257
ethanol, 244, 250, 251, 253254, 255
methanol, 244, 250
tertiary butyl alcohol (TBA), 244, 250252,
253254, 255
Song, X., 130
specic heat, 15, 16, 2324, 289290, 320, 327
spectroscopy. See also nuclear magnetic resonance
experiments (NMR). See also experiments. See
also computer simulation.
to detect local collective motion, 333334, 334
uorescence up-conversion technique, 170
light scattering, 5557
and protein hydration layer dynamics, 128129
pump-probe, 269, 270
RayleighBrillouin light spectrum, 55
of supercritical water, 320321
terahertz, 121
speed
difference between biological and bulk water
dynamics, 84
in DNA groove water, 155
of electron transfer, 47
of perturbation, 9
of polarization response, 44, 45
of rotational motion, 2832
of translational motion, 28
of water hydration dynamics, 127128, 131
of water inside a carbon nanotube, 279280
of water molecules with increasing pressure, 35, 36
Speedy, J., 308
spine of hydration (DNA), 157, 159160
standard ambient temperature and pressure (SATP), 72,
74
Stanley, H. E., 325, 328
statistical mechanics, 288, 290, 301, 308
Stern layer, 262, 263, 265, 342. See also hydration layer
Stillinger, F. H., 6, 62, 231, 233, 236, 325, 330
StokesEinstein relation
in DNA, 159
subtilisin Carlsberg (protein), 130, 131
supercooling. See also supercritical water, phase, ice
anomalies in, 334341, 335, 336, 338, 339
and coefcient of thermal expansion, 1618, 17
and hydrogen bond of heavy water, 3335
and inherent structures, 61, 69
and local order, 2021, 20, 21
motion becoming collective during, 4950
in NALMA experiments, 128
and protein hydration layer, 8890, 89
and specic heat, 15
and translational diffusion, 35
and two-stage model, 325326
supercritical water. See also supercooling, phase
denition of, 307
density uctuations in, 318319, 319
properties of, 317318
spectroscopic studies, 320321
vibrational molecular motion, 320321
Widom line in, 320
surface topology. See also protein surface, protein
hydration layer
heterogeneous, 122, 135, 161162, 170, 171
and inuence on water structure, 206207, 207
lipid, 180, 182184
mica, 206207, 207
protein, 122, 123, 124, 132133, 133
in rugged landscape, 90, 174
silica, 204205
surcants. See also amphiphilic effects
free energy landscape, 266268, 267, 268
molecular motion, 265
molecular structure of, 261263, 263, 264, 269273,
270, 271, 272
relaxation time of, 265, 268269, 273274, 274, 275
solvation dynamics, 265, 269
Sykes, M. T., 105
Tanford, Charles, 215, 220
temperature
and Brownian motion, 28
and coefcient of thermal expansion, 1617, 17
dependence in hydrophobic effects, 215, 219220,
220, 221
dependence on amphiphilic effects, 250252
and inherent structures, 6266, 63, 64, 65, 66
and isothermal compressibility, 1516, 16
and maximum density, 1315, 14
and pH, 7374, 73
and water motion, 34, 35
temperature of maximum density (TMD). See also
density uctuations
and coefcient of thermal expansion, 17
reason for, 327, 329
temperature, 1315, 14
terahertz spectroscopy, 121
tertiary butyl alcohol (TBA), 244
tetrahedral molecular structure
distorted, 5, 5, 2021, 21
in DNA, 153160, 158, 159
hydrogen bonds 4, 7172
ice, 6
when cooling, 67, 331332
theory
Cell, 298299, 299
FloryHuggins, 228230, 229
GroteHynes, 4849
Index 355
theory (cont.)
Kubo-Oxtoby, 40
Landau, 325326
Marcus, 4749, 48
PrattChandler, 232, 233, 236241
scaled particle, 231232, 233
thermal motion. See Brownian motion
thermodynamics. See also entropy, enthalpy (H)
bifurcated hydrogen bonds helping, 5
in hydrophobic hydration, 218219
of protein-hydration interactions, 170173, 172
at supercooled temperatures, 13
third law of, 288289
thickness (protein hydration layer), 118121, 124
time correlation functions. See rotational molecular
motion
time trajectory, 3132, 31
time-dependent uorescence Stokes shift (TDFSS), 44,
44
translational molecular motion
in bulk water, 27, 28, 35, 36
in computer simulations, 137
in DNA grooves, 154
in dynamic exchange model, 85, 85, 88
in lipid bilayer, 180181, 181
in surcants, 265
of water inside a carbon nanotube, 279280, 281
triple point. See phase
Trout, B. L., 314
two-stage water model, 2223, 118, 119, 310,
325326, 335341, 335, 338
urea, 208209
vibrational molecular motion
of bulk water, 27, 3940, 40
and entropy, 289290, 296, 302
and protein hydration layer dynamics,
128129
in supercritical water, 320321
viscosity. See also diffusion
DMSO, 246
in electrolytes, 211212
volume
expansion upon freezing, 306
uctuations in, 337338, 339
and isothermal compressibility, 1516, 16
Walter, N. G., 105
water pool, 263, 269
water-fearing. See hydrophobic effects
Watson, J. D., 152
Weber, T. A., 62
WeeksChandlerAndersen model (WCA), 236
Widom line 317, 320, 336337, 336, 342 See also
anomalies
Wolynes, P. G., 209
Xiao-Yan, Z., 279
Xia-Wolynes treatment, 294
Zewail, A. H., 88, 107, 130, 170
Zwanzig, R., 54, 62
356 Index

Water in Biological and Chemical Process

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Water in Biological and Chemical Process

Uploaded by

Copyright:

Available Formats

WATER IN BIOLOGICAL AND CHEMICAL PROCESSES

Building up from microscopic basics to observed complex functions, this insightful

are the static and the innite-frequency dielectric constants of the

are the zero and innite-frequency values of the dielectric

ion in the system. This phenomenon is

concentrations equal each other is

is 4.5 at 300 K, which decreases slowly with increasing temperature. For

) and a hydronium ion (H

. They together form the buffer that assists in maintaining

. Therefore, a pair of electrons needs to be

of Glu35 to neutralize its negative charge.

, which attacks the carbon to yield the hydrolyzed

, ATPase (which are negatively charged) close to the surface on both

C, compared with a value of 1.2 s

C and pH 7.78. That is, the rate of hydrolysis of

. Solvation of such ions is important in many areas of chemistry and

) ions. These are important phenomena

), can be quite different. In each case, however,

) is stabilized through the formation of strong HBs offered

is the surface tension in the planar

to identify A and W (or we

You might also like