Professional Documents
Culture Documents
Computational Chemistry
Shubin Liu, Ph.D.
Research Computing Center
University of North Carolina at Chapel Hill
Outline
Introduction
Methods in Computational Chemistry
Ab Initio
Semi-Empirical
Density Functional Theory
New Developments (QM/MM)
Hands-on Exercises
The PPT format of this presentation is available here:
http://its2.unc.edu/divisions/rc/training/scientific/
/afs/isis/depts/its/public_html/divisions/rc/training/scientific/short_courses/
its.unc.edu
About Us
ITS Information Technology Services
its.unc.edu
http://its.unc.edu
http://help.unc.edu
Physical locations:
401 West Franklin St.
211 Manning Drive
10 Divisions/Departments
Information Security
Research Computing Center
User Support and Engagement
Communication Technologies
Enterprise Applications
Research Computing
Where and who are we and what do we do?
Groups
Infrastructure -- Hardware
User Support -- Software
Engagement -- Collaboration
its.unc.edu
About Myself
Ph.D. from Chemistry, UNC-CH
Currently Senior Computational Scientist @ Research Computing Center, UNC-CH
Responsibilities:
Support Computational Chemistry/Physics/Material Science software
Support Programming (FORTRAN/C/C++) tools, code porting, parallel computing, etc.
Offer short courses on scientific computing and computational chemistry
Conduct research and engagement projects in Computational Chemistry
Development of DFT theory and concept tools
Applications in biological and material science systems
its.unc.edu
About You
Name, department, research interest?
Any experience before with high
performance computing?
its.unc.edu
Think BIG!!!
What is not chemistry?
From microscopic world, to nanotechnology, to daily life, to
environmental problems
From life science, to human disease, to drug design
Only our mind limits its boundary
its.unc.edu
Central Theme of
Computational Chemistry
STRUCTURE
DYNAMICS
REACTIVITY
its.unc.edu
SEQUENCE
STRUCTURE
DYNAMICS
FUNCTION
EVALUTION
Multiscale Hierarchy of
Modeling
its.unc.edu
What is Computational
Chemistry?
Application of computational methods and
algorithms in chemistry
Quantum Mechanical
i.e., via Schrdinger Equation
also called Quantum Chemistry
Molecular Mechanical
i.e., via Newtons law F=ma
also Molecular Dynamics
Empirical/Statistical
ii
H
H
tt
Focus
FocusToday
Today
its.unc.edu
10
its.unc.edu
11
ii
H
H
tt
H E
N
Z
h2
2
H i -
2m
i 1
1 ri
n
j i
i 1
1
rij
1
H h i
i 1
ji i 1 rij
its.unc.edu
12
Equation to Solve in
ab initio Theory
H E
Known exactly:
3N spatial variables
(N # of electrons)
its.unc.edu
To be approximated:
1. variationally
2. perturbationally
13
nuclei
electrons
nuclei 22
electronsnuclei
nuclei e 22Z
2
2
e ZAA
2
2
H
i
A
H
i
A
2
m
2
m
riA
e
A
2
m
2
m
riA
ii
AA
ii
AA
e
A
electrons
nuclei
electronse 22
nucleie 22Z Z
e ZAA ZBB
e
i i j j
rijr A B RRAB
ij
AB
A B
its.unc.edu
14
AbInitio Methods
Accurate treatment of the electronic distribution using the
its.unc.edu
15
Three Approximations
Born-Oppenheimer approximation
its.unc.edu
16
Born-Oppenheimer
Approximation
the nuclei are much heavier than the electrons and move more slowly than the electrons
freeze the nuclear positions (nuclear kinetic energy is zero in the electronic
Hamiltonian)
electrons
22 22 electrons
i
22mme i
ii
electrons
electrons
H
Helel
ii
2
2
electronse 2 nuclei
nucleie 2Z Z
ee22ZZA electrons
e
e ZAA ZBB
A
ij A
rriA
rrAB
AA
i i j j rr
ABB
iA
ij
AB
nuclei
nuclei
**
el
elelH
eleldd
el
E , E
H
Helel elel Eelel , E
**
el el dd
el the
el nuclear-electron attraction and nuclear E depends on the nuclear positions through
its.unc.edu
17
Approximate Wavefunctions
Construction of one-electron functions (molecular orbitals,
MOs) as linear combinations of one-electron atomic basis
functions (AOs) MO-LCAO approach.
NN
l qqi
uukl k rri ii ;;
l
i
kl k
i
1
k
k 1
--up)
(spin
(spin
up)
ii
(spin
(spin--down)
down)
its.unc.edu
18
a 1 b 2 c 3 z n
1n1!
b 2 2c 3 3
z n n a 2 2b 11c 3 3
z n n
a 1
a
b
c
z
a
b
c
z
n!
a 11
a
b 11
b
1n1! c 11
c
n!
z 11
z
a 2 2
a
b 2 2
b
c 2 2
c
z 2 2
z
a 3 3
a
b 3 3
b
c 3 3
c
z 3 3
z
a n n
a
b n n
b
c n n
c
z n n
z
a 11b 2 2c 3 3
z n n abc
z
a
b
c
z
a b c
z
its.unc.edu
19
1 2 3 N
HF
HF 11 122 2 33 3 NN N
There are N MOs and each MO is a linear combination of N AOs.
Thus, there are nN coefficients ukl, which are determined by
making stationary the functional:
N
HF H HF
EE
ij
HF
HF HF H HF
ij
i , j 1
i , j 1
uu SS uu ij
ij
k ,l 1
k ,l 1
N
N
*
*
ki kl lj
ki kl lj
20
CICI
cc; ;
1
1
There are
EE
CICI CICI HH CICI
2N
2 N
n n
cc SS cc
, 1
, 1
2N
9
Davidsons method can be used to find one
9
40 10
10
10
or a few eigenvalues of a matrix of rank 109.
10
its.unc.edu
21
1 2 3 N
HF
HF 11 122 2 33 3 NN N
*
*S u
EE
u
HFHF HFHF H HFHF i
ijij
ukiki Sklkl uljlj ijij
, j 1
k ,l 1
i , j 1 k ,l 1
N
N
NN
NN
N
N
NN
E P h 11 P mk nl ; P u **u
HFHF HH
HFHF Enucnuc Pmnmn hmnmn 22 Pklkl mk nl ; Pklkl ukikiulili
mm, n, n11
k k,l,l 11
i i 11
E HF 0
uuki E HF 0
ki
its.unc.edu
HartreeFock equations
22
Overlap integral
F H P
S |
| 1 |
2
occ
Density Matrix
its.unc.edu
P 2 ci ci
i
23
Self-Consistent-Field (SCF)
c F c S
its.unc.edu
24
Ab Initio Methods
Semi-empirical
Semi-empiricalmethods
methods
(MNDO,
(MNDO,AM1,
AM1,PM3,
PM3,etc.)
etc.)
excitation
excitationhierarchy
hierarchy
(CIS,CISD,CISDT,...)
(CIS,CISD,CISDT,...)
(CCS,
(CCS,CCSD,
CCSD,CCSDT,...)
CCSDT,...)
Hartree-Fock
Hartree-Fock
(HF-SCF)
(HF-SCF)
perturbational
perturbationalhierarchy
hierarchy
(MP2,
MP3,
MP4,
(MP2, MP3, MP4,)
)
excitation
excitationhierarchy
hierarchy
(MR-CISD)
(MR-CISD)
perturbational
perturbationalhierarchy
hierarchy
(CASPT2,
CASPT3)
(CASPT2, CASPT3)
Multiconfigurational
MulticonfigurationalHF
HF
(MCSCF,
(MCSCF,CASSCF)
CASSCF)
Full
FullCI
CI
its.unc.edu
25
Whos Who
its.unc.edu
26
Size vs Accuracy
Full CI
Accuracy (kcal/mol)
0.1
Coupled-cluster,
Multireference
Nonlocal density functional,
Perturbation theory
10
1
its.unc.edu
10
100
Number of atoms
1000
27
AN EXAMPLE
Equilibrium structure of (H2O)2
W.K., J.G.C.M. van Duijneveldt-van de Rijdt, and
F.B. van Duijneveldt, Phys. Chem. Chem. Phys. 2, 2227 (2000).
95.7 pm
symmetry:
RCOO,e
= 291.2
s
96.4 pm
95.8 pm
pm
its.unc.edu
28
SCF
SCF
G2
G2
DFT
DFT
CH
CH44
CH22++H
H22
CH
544(2)
544(2)
542
542
492
492
534
534
543
543
CC22H
H44
H22++H
H22
CC22H
203(2)
203(2)
204
204
214
214
202
202
208
208
H
H22CO
CO
CO++H
H22
CO
21(1)
21(1)
22
22
3
3
17
17
34
34
22NH
NH33
N22++33H
H22
N
164(1)
164(1)
162
162
149
149
147
147
166
166
22H
H22O
O
H22O
O22++H
H22
H
365(2)
365(2)
365
365
391
391
360
360
346
346
22HF
HF
FF22++H
H22
563(1)
563(1)
562
562
619
619
564
564
540
540
Gaussian2(G2)methodofPopleandcoworkersisacombinationofMP2andQCISD(T)
its.unc.edu
29
cc
its.unc.edu
30
Basis Functions
Slaters (STO)
Gaussians (GTO)
22
xxllyymmzznn **exp
r
exp r
its.unc.edu
31
its.unc.edu
Each
Eachatom
atomoptimized
optimizedSTO
STOisisfit
fitwith
withnn
GTOs
GTOs
Minimum
Minimumnumber
numberof
ofAOs
AOsneeded
needed
Contracted
ContractedGTOs
GTOsoptimized
optimizedper
peratom
atom
Doubling
Doublingof
ofthe
thenumber
numberof
ofvalence
valenceAOs
AOs
32
Polarization /
Diffuse Functions
Polarization: Add AO with higher angular
momentum (L) to give more flexibility
its.unc.edu
33
Correlation-Consistent
Basis Functions
its.unc.edu
34
Pseudopotentials,
Effective Core Potentials
its.unc.edu
35
Correlation Energy
HF does not include correlations anti-parallel electrons
Eexact EHF = Ecorrelation
Post HF Methods:
its.unc.edu
36
Configuration-Interaction (CI)
In Hartree-Fock theory, the n-electron wavefunction is approximated by one single
Slater-determinant, denoted as:
HF
CI HF c
its.unc.edu
37
HF i 11j 2 2k 3 3
NN NN
HF
i
j
k
its.unc.edu
ijababa 11b 2 2k 3 3
NN NN , ,
ij
a
b
k
etc.
etc.
38
Truncated
Configuration Interaction
Number of linear variational parameters
in truncated CI for n = 10 and 2N = 40.
Levelof
excitation
Numberof
parameters
Example
CIS
n(2Nn)
300
CISD
+[n(2Nn)]2
78,600
CISDT
+[n(2Nn)]3
18106
FullCI
2N
n
109
its.unc.edu
39
Multi-Configuration
Self-Consistent Field (MCSCF)
The MCSCF wavefunctions consists of a few selected determinants or CSFs. In the
MCSCF method, not only the linear weights of the determinants are variationally
optimized, but also the orbital coefficients.
prescribed active orbitals (complete active space, CAS). This is the CASSCF method.
The CASSCF wavefunction contains all determinants that can be constructed from a
given set of orbitals with the constraint that some specified pairs of - and -spinorbitals must occur in all determinants (these are the inactive doubly occupied
spatial orbitals).
Internally-contracted MRCI:
MR - CISD (ck C1k C 2 k ) k
k
its.unc.edu
40
Coupled-Cluster Theory
System of equations is solved iteratively (the convergence is
its.unc.edu
41
Mller-Plesset Perturbation
Theory
The Hartree-Fock function is an eigenfunction of the
n-electron operator
F.
its.unc.edu
42
Semi-Empirical Methods
These methods are derived from the HarteeFock model, that is, they
its.unc.edu
43
Semi-Empirical Methods
Number 2-electron integrals () is n4/8, n = number of basis functions
Treat only valence electrons explicit
Neglect large number of 2-electron integrals
Replace others by empirical parameters
Models:
Complete Neglect of Differential Overlap (CNDO)
Intermediate Neglect of Differential Overlap (INDO/MINDO)
Neglect of Diatomic Differential Overlap (NDDO/MNDO, AM1, PM3)
its.unc.edu
44
Approximations of 1-e
integrals
H
U
H
VVABAB
U
BBAA
H
H 00
H
ABSS
H
AB
11 A B
AB
AB
B
22 A
One parameter per element
its.unc.edu
45
Popular DFT
Noble prize in Chemistry, 1998
In 1999, 3 of top 5 most cited journal
articles in chemistry (1st, 2nd, & 4th)
46
its.unc.edu
47
its.unc.edu
48
Energy E
The energy density functional is variational.
E E 0 if 0
its.unc.edu
49
E nn [ ]
nuclei
A B
Z AZ B
, E ne [ ] r Vext r dr, J [ ]
rAB
1
2
r r
r r drdr ,
But what about the kinetic energy TS[] and exchange-correlation energy
Exc[]?
its.unc.edu
50
Kohn-Sham Scheme
H nk nk nk ,
exact density.
its.unc.edu
unknown is to calculate
where
H K Vne (r ) Vee ( r ) Vxc (r )
and
K
1 2
,
2
The Only
, Unknown
Za
a | r Ra |
(r ) 3
Vee (r )
d r,
| r r |
E xc
Vxc ( r )
,
Vne (r )
( r ) f nk | nk (r ) |2 ,
nk
51
XC
its.unc.edu
52
LDA Functionals
Thomas-Fermi formula (Kinetic) 1
parameter
TTF C F
5/3
3
r dr, CF 3 2
10
2/3
4/3
3 1/ 3 2 / 3 1/ 3
r dr , C X 3 4
8
53
2
B
LDA
4/3
EX EX
,
1 2
4/3
1
1 2
1
2 / 3
5/3
2
t
W
W
F
1 d 1/ 3
9
18
c
e
1 / 3
dr
B3LYP
xc
aE E
B
x
LYP
c
m , n 1
k ,l 1
b Pmn Pkl km nl
54
VWN5
BLYP
HCTH
BP86
TPSS
M06-L
B3LYP
B97/2
MPW1K
MPWB1K
M06
its.unc.edu
Density Functionals
LDA
local density
GGA
gradient corrected
Meta-GGA
kinetic energy density
included
Hybrid
exact HF exchange
component
Hybrid-meta-GGA
Meta-GGA is bleeding
edge and therefore
largely untested (but
better in theory)
Hybrid makes bigger
difference in cost and
accuracy
Look at literature if
somebody
has compared
functionals for
systems similar to
yours!
55
Density Functionals
S.F. Sousa, P.A. Fernandes and M.J. Ramos, J. Phys. Chem. A 10.1021/jp0734474 S1089-5639(07)03447-0 56
its.unc.edu
57
DFT Developments
Theoretical
Extensions to excited states, etc.
Better functionals (mega-GGA), etc
Understanding functional properties, etc.
Conceptual
More concepts proposed, like electrophilicity, philicity, spin
Computational
Linear scaling methods
QM/MM related issues
Applications
its.unc.edu
58
its.unc.edu
59
60
2!
r dr
N
N 2
N 2
N
2E
N r dr
2E
2 r
61
Conceptual DFT
Basic assumptions
E E [N, (r)] E []
Chemical processes, responses, and changes
expressible via Taylor expansion
its.unc.edu
62
Hardness / Softness
HOMO LUMO
1 2E
LUMO HOMO
, S 1 /
Maximum Hardness
(MHP)
2Principle
2 N
2
HSAB (hard and Soft Acid and Base) Principle
its.unc.edu
63
function
f r
Nucleophilic attack
f r N 1 r N r
Electrophilic attack
r N r N 1 r
its.unc.edu
64
Electrophilicity Index
2
2
Physical meaning: suppose an electrophile is immersed in an
electron sea
1
E N N 2
2
The maximal electron flow and accompanying energy decrease
are
N max
E min
its.unc.edu
65
Prez, P. J. Org. Chem. 2003, 68, 5886. Prez, P.; Aizman, A.; Contreras, R. J. Phys. Chem. A 2002, 106, 3964.
its.unc.edu
66
-0.093
-0.093
LA
Aa
-0.091
0.085
Bb
-0.089
-0.172
-0.088
-0.089
-0.171
0.246
-0.083
-0.090
-0.087
-0.0842
0.087
-0.230
-0.220
0.247
Bb
-0.087
0.088
0.084
Aa
0.0892
-0.226
-0.2518
0.218
-0.247
-0.233
0.221
0.2161
-0.226
-0.2506
0.2157
its.unc.edu
67
Yue
Xia, Dulin Yin, Chunying Rong, Qiong Xu, Donghong Yin , and Shubin Liu, J. Phys. Chem. A, 2008, 112, 9970.
Nucleophilicity
Much harder to quantify, because it related to local
hardness, which is ambiguous in definition.
= -N - S()2
Minimizing in Eq. (14) with respect to ,
one has
=-N and = - N2.
Making use of the following relation
A B
A B
Jaramillo, P.; Perez, P.; Contreras, R.; Tiznado, W.; Fuentealba, P. J. Phys. Chem. A 2006, 110, 8181.
its.unc.edu
68
( )2
Ee I
2
Ayers, P.W.; Anderson, J.S.M.; Rodriguez, J.I.; Jawed, Z. Phys. Chem. Chem. Phys. 2005, 7,
1918.
Ayers, P.W.; Anderson, J S.M.; Bartolotti, L.J. Int. J. Quantum Chem. 2005, 101, 520.
its.unc.edu
69
Dual Descriptors
3rd-order cross-term derivatives
2
f r
2 N r
f 2 r dr 0
f 2 r f
2 E
r 2 N
r f r
f r
f 2 r LUMO r HOMO r
its.unc.edu
Ayers, P.W.; Morell, C., De Proft, D.; Geerlings, P. Chem. Eur. J., 2007, 13, 8240
70
Geerling, P. De Proft F. Phys. Chem. Chem. Phys., 2008, 10, 3028
Steric Effect
one of the most widely used concepts
in chemistry
its.unc.edu
71
we have
1
TW
8
dr
72
its.unc.edu
73
QM/MM Example:
Triosephosphate Isomerase (TIM)
DHAP
H2O
GAP
494 Residues, 4033 Atoms, PDB ID: 7TIM
Function: DHAP (dihydroxyacetone phosphate) GAP (glyceraldehyde 3-phosphate)
its.unc.edu
74
DHAP
GAP
its.unc.edu
75
its.unc.edu
76
=====================
Energy Barrier (kcal/mol)
-----------------------------------QM/MM
21.9
Experiment
14.0
=====================
its.unc.edu
77
diagonalization ~N3
orthonormalization ~N3
matrix element evaluation ~N2-N4
100
90
80
matrix or orbitals
Strategy:
CPUsecondsper CGstep
60
Diagonalization
50
40
matrix
OLMO
NOLMO
20
10
methods
100
200
300
400
500
600
700
800
900
Atoms
78
What Else ?
Solvent effect
its.unc.edu
Relativity effect
Transition state
Excited states
Temperature and pressure
Solid states (periodic boundary condition)
Dynamics (time-dependent)
79
its.unc.edu
80
Popular QM codes
Gaussian (Ab Initio, Semi-empirical, DFT)
Gamess-US/UK
Spartan
(Semi-Empirical)
DMol3/CASTEP
(DFT)
Molpro
(Ab initio)
ADF (DFT)
ORCA
its.unc.edu
(DFT)
81
Reference Books
its.unc.edu
82
http://its2.unc.edu/divisions/rc/training/scientific/
/afs/isis/depts/its/public_html/divisions/rc/training/scientific/short_courses/
its.unc.edu
83
Hands-on: Part I
Purpose: to get to know the available ab
initio and semi-empirical methods in the
Gaussian 03 / GaussView package
ab initio methods
Hartree-Fock
MP2
CCSD
Semiempirical methods
AM1
The WORD .doc format of this hands-on exercises is available here:
http://its2.unc.edu/divisions/rc/training/scientific/
/afs/isis/depts/its/public_html/divisions/rc/training/scientific/short_courses/labDirections_compchem_2009.doc
its.unc.edu
84
Hands-on: Part II
Purpose: To use LDA and GGA DFT methods to
calculate IR/Raman spectra in vacuum and in
solvent. To build QM/MM models and then use
DFT methods to calculate IR/Raman spectra
DFT
LDA (SVWN)
GGA (B3LYP)
QM/MM
its.unc.edu
85