ARTICLE

www.rsc.org/materials
Chemistry
Materials
Journal of
Understanding the factors that govern the deposition and
morphology of thin films of ZnO from aqueous solution{
Kuveshni Govender, David S. Boyle, Peter B. Kenway and Paul O’Brien*

The Manchester Materials Science Centre and Department of Chemistry, University of

Manchester, Oxford Road, Manchester, UK M13 9PL. E-mail: paul.obrien@man.ac.uk;
Fax: 0161-2754616; Tel: 0161-2754652

Received 31st March 2004, Accepted 19th May 2004

First published as an Advance Article on the web 15th June 2004

The influence of the choice of complexing ligand, zinc counter-ion, pH, ionic strength, supersaturation, deposition
time and substrate on the nature of ZnO films grown from chemical baths (CBD) are discussed. There are
significant differences between CBD and similar routes such as hydrothermal methods for ZnO films. Modelling of
speciation and experimental results suggest that acicular ZnO morphologies are best obtained by limiting the
concentration of one of either Zn21 or OH2 in the presence of a large excess of the other. The presence of a prior
ZnO layer can facilitate nucleation at lower levels of supersaturation and enable size tailoring of ZnO columns.
The point at which the substrate is introduced into the bath is crucial and can lead to a significant difference in
both the width of the rods and optical transparency of the films. HR-TEM has yielded important structural
information and a growth mechanism for single crystalline ZnO rods by CBD is described for the first time.

1. Introduction ultimately oxoanions are formed. In order to form the poly-

The deposition from aqueous solution or CBD (chemical bath
deposition) of metal oxide thin films involves controlled pre-
cipitation on a substrate via hydrolysis and/or condensation
reactions of metal ions and/or complexes from aqueous solu-
tion.1 In CBD, the crystal morphology is strongly influenced by
experimental conditions including chemical speciation in the
solution (ligand, pH, metal counter-ion, ionic strength are all
important), the level of supersaturation, the temperature and
the nature of the substrate. Control of the size, shape and
orientation of ZnO crystallites on the substrates is a pre-
requisite for creation of high surface area materials for use in
many types of devices including photovoltaic and optoelec-
tronic devices.2 The fundamental optical physics of ZnO
nanostructures as a function of crystallite dimensionality have
only recently been experimentally verified.3 In this paper, we
have modelled the initial composition of deposition baths and
developed strategies for morphological control during CBD of
ZnO, leading to reproducible routes to polycrystalline and
single crystal array thin films. The mechanisms of growth of
thin films of ZnO rod arrays have been studied using HR-TEM
with FIB sample preparation techniques.
In order to understand the important physiochemical processes
involved in the CBD of ZnO, it is useful to summarise its solution
chemistry. In an aqueous solution, metal cations Mz1 are solvated
by water giving rise to aquo-ions, typically [M(OH2)n]z1. The
M–OH2 bond is polarised which facilitatesdeprotonation of the
coordinated water. In dilute solutions, a range of monomeric
species exist such as ([M(OH2)n2p(OH)p](z2p)1 and other hydroxy
species [M(OH)n] (it is often customary to omit the water);
{ Electronic supplementary information (ESI) available: Fig. A,B: film
thickness profiles for Zn–en and Zn–TEA systems; Fig. C: SEM images
DOI: 10.1039/b404784b
nuclear species, which subsequently develop into metal oxide
particles, reactions involving condensation reactions must occur.
Two important processes have been recognized, olation is the
formation of an ‘‘ol’’ bridge by reaction of a hydroxo- and aquo-
species as follows:
M–OH 1 M–OH2 A M–OH–M 1 H2O (1)
Oxolation leads to an ‘‘oxo’’ bridge by the dehydration of
hydroxo-species:
M2–(OH)2 A M–O–M 1 H2O (2)
Zinc hydroxide is amphoteric and complexation by OH2
can lead to soluble species such as ‘‘[Zn(OH)3]2’’ and
‘‘[Zn(OH)4]22’’ and hence ‘‘zinc hydroxide’’ is more soluble
in basic solution than a simple consideration of the solubility
product (y10216) would suggest.4
In the CBD of ZnO, ligands are employed to keep the free
zinc ion concentration low. Raising the bath temperature pro-
motes some dissociation of the zinc complex, leading to
controlled supersaturation of the free metal ion. Zinc is a labile
metal ion in aqueous solution and equilibria within stirred
solutions are generally attained quickly. Thermodynamic model-
ling of solutions representing the initial states of deposition baths
is a useful aid to understanding these systems.5,6
Morphology of ZnO crystallites
The thermodynamically stable phase for ZnO is zincite with
the wurtzite (hcp) structure although two cubic phases, a
metastable sphaleritic (zincblende)7 and high pressure rocksalt
type8 have been reported. The elongated hexagonal zinc oxide
crystal has both polar and non-polar faces. The former are

of ZnO films on ZnO template layers from HMT baths; Fig. D: XRD
patterns of ZnO microcolumns grown on ZnO template layers on
TO(F) glass, and on Au/TO(F) glass; Fig. E: SEM images showing
effect of increasing ionic strength on ZnO film growth; Fig. F: SEM
images and grain size distributions of ZnO films from HMT baths;
Fig. G: micrographs of ZnO thin film samples (Methods 1 and 2);
Table A: thermodynamic data; Table B: rod dimensions for different
deposition times. See http://www.rsc.org/suppdata/jm/b4/b404784b/
either Zn-terminated (the (0001) face) or O-terminated (the
(0001̄) face). The latter include (112̄0) and (101̄0) faces. The
morphology of a particular crystal is determined by the slowest
growing faces. Polar faces with surface dipoles are thermody-
namically less stable than non-polar faces, often undergo
rearrangement to minimise their surface energy and also tend to
grow more rapidly. Therefore it is necessary to distinguish between

This journal is ß The Royal Society of Chemistry 2004 J. Mater. Chem., 2004, 14, 2575–2591 2575
 equilibrium and kinetic growth morphologies. The former corres-
ponds to a minimum in the surface free energy. However, ther-
modynamic models that consider morphological development as a
function of internal structure are poor at describing the solution
growth of ZnO at moderate temperatures.9
The morphology of films or powders needs to be optimised
for each application. The morphology has been shown to affect
catalytic and photocatalytic activity.10 In ZnO powders, the
non-polar faces typically account for y80% of the total surface
and the (101̄0) faces can determine the adsorption of com-
plexes.11 A wide variety of ZnO crystallite morphologies are
observed for both precipitates and thin films including
columnar grains,12–14 rods,2,15 stars5 and spherical habits.2,16
Despite numerous studies, there is little understanding of the
mechanisms and factors that govern the observed morphology.
Two growth mechanisms are usually discussed for the growth
of single crystals: growth-dissolution-recrystallization pheno-
mena or nanoparticle aggregation.17
The morphology of ZnO is greatly complicated by its pro-
pensity to twinning. Twinned ZnO crystals have been investi-
gated and shown to have a fourling structure, in which three
‘‘legs’’ are related to the fourth (spine) by twinning. The spines
are related by twinning along the (112̄2) planes with a slight
distortion at the twinned faces, such that two angles between
pairs of spines are 97.6u and four are 116.1u.18,19 In the second
phase of fourling crystal growth, three sheets lying in the three
planes determined by the leg and the spine grow out from the
spine. Many deviations from this idealised structure can occur.
Fourlings of ZnO have been reported during forced hydrolysis of
zinc nitrate using HMT (hexamethylenetetramine),20 en (ethyl-
enediamine)21,22 and NH4OH23 using hydrothermal methods
and by the decomposition of zinc hydroxide carbonate.24 Under
hydrothermal conditions, bipyramidal twinning occurs via the
(0001̄) plane whereas dumbbell-like examples are formed via the
(0001) plane.25 The twinned relations of the crystallites were
found to be influenced by additives in the reaction medium.
Solution growth of ZnO rod array films
There is interest in highly structured metal oxide thin films as
high surface area substrates for solar cells (e.g. in the Grätzel and
eta configurations).26,27 One problem can be a need to anneal
as-deposited films in order to effect more complete crystal-
lization. A low-temperature synthetic procedure that requires no
post-deposition annealing step is potentially very useful.28 Few
systematic studies have been reported although recently Imai has
shown that using different substrates with identical baths yields
different morphologies.29 We have investigated solution-based
approaches with a defined range of experimental conditions for
growing single crystal arrays of ZnO by CBD.2
The literature describing ZnO precipitates is widely dis-
persed. The formation of high-aspect ratio ZnO crystallites has
attracted interest.30 Earlier studies of ZnO single crystal rods
have not led to a universally accepted mechanism for the crystal
growth. There is some evidence from ageing experiments that
the formation of ‘‘Zn(OH)2‘‘ is a prerequisite for the controlled
growth of needle-like ZnO crystallites.31 This observation is
consistent with a dissolution–reprecipitation mechanism in
which Zn(OH)2 acts as a reservoir of zinc. The Zn21 concen-
tration is held below the level where undesirable secondary
nucleation processes, which can lead to twinning, occur.
The use of hexamethylenetetramine (HMT) is reported
widely in the growth of acicular ZnO precipitates and thin
films. An early report in the patent literature describes heating
aqueous solutions of zinc chloride (ZnCl2) and an organic
compound that form a base (such as HMT) or an acid (such as
ethylene chlorohydrin) by hydrolysis.32 The control of ZnO
crystallite morphology using HMT has been described by
Fujita et al.,33 where precipitates derived from zinc nitrate were
globular or rod-shaped whereas those from zinc chloride were
2576 J. Mater. Chem., 2004, 14, 2575–2591
both globular and acicular. Formation of ZnO in acidified
baths containing ZnCl2 and HMT was suggested via two
routes; direct formation or via a Zn5Cl2(OH)8 intermediate.34
The globular form was predominant at high concentrations of
HMT. Andrés-Vergés and co-workers also used aqueous baths
containing zinc nitrate and HMT to form ‘‘rod-like’’ pre-
cipitates.20 Individual microcrystals were formed within a
narrow concentration range of zinc salt and HMT depending
on the salt used, whereas aggregates were formed when the
solution pH was not lowered. A decrease in the pH of the
solution resulted in the formation of individual needles or
prismatic microcrystals. These workers first suggested the idea
of orientated attachment to explain the acicular growth of ZnO
crystallites from solution. Spherical particles formed in solu-
tion tended to aggregate along their polarized anisotropic
c-axes to yield primary ‘‘rod-like’’ zinc oxide crystallites. Similar
observations have been made more recently for quasi-spherical
ZnO nanoparticles formed in basic methanolic solutions.35
The deposition of thin films of rod-like ZnO arrays on
fluorine doped tin oxide (TO(F)) glass substrates has been
reported by Vayssieres, using zinc nitrate–HMT solutions
similar to the work of Andrés-Vergés’ but in closed vessels.15
The success of the method depends on separating homo-
geneous and heterogeneous nucleation by controlling the inter-
facial tension. We have grown ZnO rods on ZnO template
layers on TO(F)-glass using carboxylate precursors in open
baths2 and demonstrated room temperature lasing from such
arrays.36 There is an important distinction to be made between
aqueous solution deposition of ZnO (thin films or precipitates)
from open and closed baths; the latter conditions may appro-
ach hydrothermal growth. In general, the solubility of ZnO is
greater under hydrothermal conditions and growth is typically
slower. Imai has also demonstrated the importance of substrate
and shown that upright ZnO rods could be grown only on
crystalline ZnO templates.37
HMT is commonly used in solution routes to acicular ZnO,
however it is not essential and its role is often misunderstood.
The primary role of HMT in aqueous deposition baths is to act
as a pH buffer. The kinetics of decomposition of HMT have
been investigated.38 The conversion of HMT to formaldehyde
(and ammonia) was found to be pH dependent in buffers of
constant ionic strength, with the reaction half-life decreasing
from 13.8 h at pH 5.8 to 1.6 h at pH 2.0 In a typical CBD
experiment, HMT provides a slow controlled supply of OH2
and is a poor ligand for zinc. Significant formation of the
Zn(OH)2 phase cannot be discounted, therefore a growth
mechanism based on dissolution–reprecipitation of preliminary
hydroxy-zinc clusters and zinc oxide is possible in heated
deposition baths containing HMT.31
2. Results and discussion
Forced hydrolysis of zinc carboxylates
The forced hydrolysis described by Matijevic is the simplest
precipitative process for metal oxides and involves ageing
of aqueous metal salt solutions at elevated temperatures (80–
100 uC).23 The pH decreases during the course of the reaction
due to the deprotonation of the hydrated cation.
DH
½M(H2 O)x
nz DA ½M(H2 O)x{1 (OH)
(n{1)z
(3)
??M(OH)n or MOy
Studies of the speciation of Zn21 in chemical baths allow
determination of the initial degree of supersaturation within
the system. Speciation calculations were carried out for baths
containing only zinc acetate in order to determine the point at
which the solution becomes supersaturated with respect to the
Zn(OH)2 (Fig. 1a). At 25 uC this critical value of pH is 6.68, in
general agreement with experimental results.

Fig. 1 a. Speciation diagram illustrating the hydrolysis of zinc acetate
(0.025 mol dm23) at 25 uC. Zn(OH)2 (s), represented by AH-2(s),
is predicted to precipitate at pH 6.68. A ~ Zn21; B ~ CH3COO2.
b. SEM image of ZnO precipitates on TO(F) glass formed from forced
hydrolysis of zinc acetate (0.025 mol dm23) at 90 uC.
Preliminary experiments involved attempts to deposit
ZnO films using CBD methods (open baths) on TO(F) glass
by forced hydrolysis of zinc salts. Due to the volatility of zinc
carboxylates as compared to nitrate, chloride and sulfate salts,
deposition of material only occurred for the former. Zinc
acetate and zinc formate solutions (0.025 mol dm23) containing
the TO(F) substrates were heated at 90 uC for 2 hours.
There are some unique characteristics of ZnO–carboxylate
systems, for example, Li et al. have demonstrated that forma-
tion of ZnO from basic solutions of zinc acetate by forced
hydrolysis is only possible under conditions which allow the
release of volatile by-products and is hence facilitated by use of
open baths or the periodic release of pressure in hydrothermal
systems.9 Carboxylates are often selectively adsorbed on the
surface sites on ZnO. Surface defects, on nominally Zn-free
O-terminated faces, in the form of steps with {101̄0} orientated
faces that expose zinc cations, have been evidenced by
exposure to formic acid with subsequent formation of
Zn-bound formate species.39 There is indirect evidence from
work involving etching of ZnO single crystals that indicates
carboxylates inhibit the rates of dissolution one hundred fold
and tend to aid ageing and growth of crystallites.40,41 Dis-
solution processes are controlled by Zn–O hydrolysis reactions
at Zn21 kink sites and catalysed by both H1 and ligands for
zinc.
Precipitates were visible in the deposition bath with zinc
acetate, but not zinc formate. For the former, the homogeneous
process was dominant and precipitation occurred almost
immediately upon heating of the solution. The films obtained
were powdery and non-adherent and scanning electron micro-
scopy of the films revealed poor surface coverage with rod-
like particles, often with dumbbell morphology, randomly
orientated on the substrate with little evidence for perpendi-
cular growth (Fig. 1b). The particle size distribution was large
and there is also evidence of twinning.
Fig. 2 Relationship between supersaturation in deposition baths, rate
of crystallite growth and morphology.
Development of crystal morphology
The production of a supersaturated (SS) solution is a pre-
requisite for crystallization and the level of SS is important in
determining the final crystal morphology, as exemplified by
CBD ZnO and shown in Fig. 2. We can broadly identify three
main growth regions, defined by the level of supersaturation.
. for relatively low levels of supersaturation in the region
below SS*, heterogeneous nucleation dominates. Crystal faces
grow via the outward displacement of a growth spiral originat-
ing from screw dislocations (the so-called BCF mechanism)42 in
the central region of the face. Well-formed polyhedral crystals
are generally produced under these conditions.
. in the intermediate supersaturation range SS* v SS v
SS**, growth occurs by a mixture of slow spiral growth and
2-dimensional nucleation of clusters on to crystal faces from
solution (polynuclear processes). In this region, 2D growth
generally predominates. Since the probability of occurrence of
2D nucleation is far greater near to edges and corners of faces,
growth steps will advance from them to yield a crenellated face
and often ‘‘hollowed’’ rod morphology.
. for high supersaturation SS w SS**, homogeneous nuclea-
tion is more important. The crystal faces become rough and
continuous linear growth occurs. In this region, ZnO crystals
adopt a dendritic to spherulitic morphology and with
increasing supersaturation, growth kinetics are controlled by
bulk diffusion as the solution becomes more viscous.
The acicular rod morphologies, typically observed for CBD-
ZnO films using baths containing HMT and a zinc salt at
moderately acidic pH 5,2 can be explained in terms of a
kinetically-controlled reaction involving relatively high con-
centrations of one (i.e. Zn21) over the other (i.e. OH2 ) com-
ponent. Similar arguments apply to growth of columnar ZnO
films using basic baths and ethylenediamine ligand, where the
high–low concentration relationship is reversed.
Influence of ligand on crystallite morphology
For cations that are not easily hydrolysed, such as Zn21,
precipitation of the metal hydroxide can be promoted by
raising the pH with a base. Lewis bases, such as ethylenedia-
mine (en) and triethanolamine (TEA), can also form complexes
with the metal. The complexes can also act as a reservoir for
metal, by buffering the free concentration to below the pre-
cipitation point.
In a ZnO CBD system where both ZnO and ‘‘Zn(OH)2’’ may
be formed, the less stable phase in contact with the media will in
general precipitate first. Under conditions such that the
hydroxide precipitates first, changing the temperature or pH
can result in the formation of the oxide. The oxide can thereby
form via a dissolution/reprecipitation process or by a phase
J. Mater. Chem., 2004, 14, 2575–2591 2577
 transformation. A lattice rearrangement, however, is more
complicated and depends on the degree of rearrangement and
dissolution of hydroxide to provide Zn21. The uniformity of
the initial hydroxide particles may be key in controlling
uniformity of the oxide particles.
There are several reports in the literature on the effect of
ligands in the deposition bath on the morphology of
ZnO.20,21,30 The hydrolysis of the cation can be controlled by
a slow release of hydroxide into the metal salt solution. At
elevated temperatures both HMT and urea undergo slow
chemical decomposition to generate OH2 ions.
Deposition of ZnO films on TO(F) glass using en as complexing
agent
We have previously demonstrated that good quality ZnO thin
films may be deposited from baths containing zinc acetate
(0.018 mol dm23) and en (0.042 mol dm23, at a final pH of 11).5
Further experiments were conducted using similar bath
conditions but with different zinc salts. Films were formed
from baths containing en and either zinc acetate, formate,
sulfate or chloride. No films could be deposited from baths
containing zinc nitrate. Speciation calculations were performed
for baths containing zinc acetate and en (at 25 uC), the
distribution plots indicated that baths are supersaturated with
respect to the hydroxide by pH ~ 6.92 (Fig. 3a and ESI:{ Table
A for binding constants). Small decreases in stability constant
values at higher temperatures did not produce significant
changes to speciation diagrams. It was found that good quality
films were produced at pH 11 (at 70 uC, 2 h), such baths are
supersaturated and the concentration of free zinc is very low,
orders of magnitude lower than [OH2].
All ZnO films were of the zincite phase (JCPDS 36-1451) as
determined by XRD (Fig. 3b). The crystallinity was slightly
improved upon annealing in air at 400 uC. In general, films were
thin (y1 mm) and peaks originating from the TO(F) substrate
were evident in the XRD. Infrared spectra (ATR-FTIR) were
Fig. 3 a. Speciation diagram for Zn–en system at 25 uC. Dashed
line represents zinc hydroxide precipitation point. A ~ [Zn21] ~
0.006 mol dm23 and B ~ [en] ~ 0.042 mol dm23, [CH3COO2] ~
0.012 mol dm23. AH-x represents the soluble zinc-hydroxy species.
b. X-Ray diffractograms obtained for as-deposited ZnO films from
i. en-baths (70 uC). ii. TEA-baths and iii. HMT-baths on TO(F) glass.
* Reflections assigned to SnO2–glass substrate (JCPDS 41-1445).
2578 J. Mater. Chem., 2004, 14, 2575–2591
recorded to determine whether material such as zinc hydroxide
(crystalline or amorphous) was present. No peaks indicative of
zinc hydroxide were observed.
Scanning electron micrographs of ZnO films grown from
baths containing en and zinc (as the acetate, formate, chloride
or sulfate) provided useful information and, with the exception
of zinc sulfate, films comprised twinned crystallites that formed
as starlike particles with needle-like spines (Fig. 4a: i, ii, iii).
McBride has postulated that these star-like morphologies grow
by a modified LaMer mechanism.31 Stars are formed from
twinning along the (112̄2) plane of the hexagonal lattice. Cross-
sectional SEM indicated that the film thickness in all cases was
of the order y1 mm. Films formed from sulfate-containing
baths were less crystalline and appeared to be a mixture of
small flowerlike particles and larger flat platelets (Fig. 4a: iv).
More detailed examination of films deposited from baths con-
taining en and zinc acetate appeared to show that the indivi-
dual arms of the starlike ZnO crystallites possessed some
helical structure (Fig. 4b). Similar morphology has been
reported for ZnO nanorods grown on nanocrystalline ZnO
templates, using baths containing HMT and zinc nitrate
modified by addition of sodium citrate.43 It has been suggested
that the presence of citrate generates the helical morphology.
Moreover, the structure closely resembles that observed for
nacreous calcium carbonate in red abalone (gastropod Haliotis
rufescens) and thus may offer insights into biomineralisation
processes. It appears that in the presence of acetate and en,
ZnO may also be deposited with similar helical structures.
Growth kinetics were monitored using the QCM technique,
typically an induction period was followed by a growth and a
short termination phase. The QCM technique is not directly
Fig. 4 a. SEM images (top) of ZnO films from en-baths (70 uC, pH 11)
containing: i. zinc acetate; ii. zinc chloride; iii. zinc formate and iv. zinc
sulfate. [Zn] ~ 0.018 mol dm23, [en] ~ 0.042 mol dm23. b. SEM image
(below) of ZnO film deposited from zinc acetate–en bath showing indivi-
dual crystallite arms possessing helical structures as described elsewhere.43
comparable to growth on TO(F)-glass substrates as the micro-
balance uses an Au-coated crystal. However useful com-
parisons can be made for different baths, assuming all other
factors (temperature, stirring rate, etc.) are equal. For the
en-system (ESI:{ Figure A) two distinct linear growth phases
following the induction period could be identified; a short and
rapid growth period over y250 s (rate ~ 0.12 nm s21) and a
slower growth phase from y1500–3000 s (rate ~ 0.01 nm s21).
Deposition of ZnO films on TO(F) glass using TEA as
complexing agent
In order to investigate the sensitivity of particle morphologies
to changing bath chemistry, our existing literature preparation
was followed with minor modifications (pH 12 vs. pH 11 and
TEA : metal ratio 5 : 1 vs. 4 : 1 and a similar metal ion con-
centration). Experiments were conducted at 70 uC with TEA
(0.072 mol dm23) using different zinc salts (0.018 mol dm23) on
TO(F) glass substrates as for the Zn–en experiments. The final
pH of the bath was 11. In initial experiments films were only
obtained from baths containing zinc nitrate.
Speciation calculations were carried out on baths containing
zinc nitrate and TEA. Supersaturation with respect to the
hydroxide occurred at pH 6.44 similar to the en system (Fig. 5a).
Moreover, the best quality films were produced at pH 11, where
the bath is again supersaturated and the concentration of the free
zinc is orders of magnitude lower than [OH2], similar to the
en-system. However in contrast to the en-system, in which the
zinc is mainly in the form of Zn–en complexes at pH 11, for TEA
the zinc speciation is defined by zinc-hydroxy complexes.
Comparison of micrographs of thin films from the present
study (Fig. 5b) with those obtained for ZnO powders,21 under
similar bath chemistry, leads to the following conclusions.
Firstly, the particle morphologies are virtually identical in both
cases. However, there is a substrate effect, as the widths of
particles comprising films are narrower than those formed in
solution (Fig. 5b: i and iii). The morphology of the films can be
described as spherical aggregates.
For both ‘‘as-deposited’’ and annealed films, the formation
of crystalline zincite was evident from XRD (Fig. 3b: ii).
However, unlike films deposited from en-containing baths, no
changes in the diffraction pattern were observed upon annealing
Fig. 5 a. Speciation diagram of Zn–TEA system (above) at 25 uC with
A ~ [Zn21] ~ 0.018 mol dm23 and B ~ [TEA] ~ 0.072 mol dm23.
Dashed line represents zinc hydroxide precipitation point. b. (below)
i. Top view and ii. cross-section SEM images of annealed ZnO films
from TEA-baths (70 uC, pH 11). [Zn] ~ 0.018 mol dm23; [TEA] ~
0.072 mol dm23. iii. Precipitates obtained from solution.
in air. Again no evidence for zinc hydroxide in the films was
obtained from IR spectroscopy. Results obtained from QCM
measurements were different from those with en-containing
baths (ESI:{ Figure B). The induction period (0–500 s)
was characterised by oscillations, followed sequentially by a
relatively linear period of rapid growth (y500–1500 s; average
rate ~ 0.09 nm s21) and then slower deposition (y1500–2500 s;
average rate ~ 0.02 nm s21). In comparison to the previously
described en-system, in the TEA system the transition from the
‘‘fast’’ to ‘‘slow’’ growth was relatively abrupt.
Deposition of ZnO films on TO(F) glass using HMT
Further studies of the effect of the chelating ligand on ZnO
particle morphology were carried out with HMT, which
decomposes slowly in heated aqueous solutions to yield
ammonia and formaldehyde. The concentration of HMT has
been reported to influence the rate of ZnO formation.34
Speciation calculations for the HMT system indicate that the
pH at which the bath becomes supersaturated at 25 uC with
respect to the hydroxide is y6.8 (Fig. 6a). All other factors
being equal (i.e. ligand and metal concentration), at values of
pH w 6, supersaturation in terms of ‘‘zinc hydroxide’’ will
be greater for HMT than en. Thin film formation requires
production of a supersaturated system hence we would expect
ZnO film formation to occur at lower pH values for HMT than
en. HMT decomposes under CBD conditions, which com-
promises any simple thermodynamic speciation calculations.
However, under typical experimental conditions, ZnO thin
films will be deposited from supersaturated bath solutions
containing relatively high concentrations of free zinc and low
Fig. 6 a. Speciation diagram (top) of Zn–HMT system at 25 uC.
Dashed line represents ‘‘zinc hydroxide’’ precipitation point. A ~
[Zn21] ~ 0.0083 mol dm23 and B ~ [CH3COO2] ~ 0.016 mol dm23
and C ~ [HMT] ~ 0.025 mol dm23. b. SEM images (below) of
ZnO films deposited from HMT-baths (0.1 mol dm23 Zn(NO3)2 and
0.1 mol dm23 HMT) in stoppered flasks in oven: i. pH ~ 6.8, TO(F)
glass, precipitation initiated at 50 uC; ii. pH ~ 6.8, ZnO template
layers, precipitation initiated at 50 uC; iii. pH ~ 5.0, TO(F) glass,
precipitation initiated at 95 uC; iv. pH ~ 5.0, ZnO template layers,
precipitation initiated at 95 uC.
J. Mater. Chem., 2004, 14, 2575–2591 2579
 concentrations of hydroxide. In essence this represents a
reversal of the situation encountered with the en-system.
Several workers have reported the homogeneous precipita-
tion of ZnO in HMT baths but Andrés-Vergés and co-workers
demonstrated that a decrease in solution pH resulted in the
formation of individual needles or prismatic microcrystals from
baths containing zinc nitrate and HMT.20 We have reasoned
that the presence of a pre-existing template layer (deliberate or
otherwise) would promote heterogeneous growth of ZnO films.
Moreover, as precipitation occurred rapidly in baths in which
no pH adjustment was made, by using both ZnO templates and
lower solution pH, good quality films can be deposited. SEM
images were recorded for films deposited on ZnO template
layers at an unadjusted pH of 6.8 (Fig. 6b: ii) and the lower pH
of 5 (Fig. 6b: iv), using bath conditions described above. It is
clearly seen that films deposited at both pH values are dense,
importantly however, those films deposited at lower values of
pH have well defined faces. The phenomenon has been
rationalised in terms of slower growth kinetics.20
The above observations have led to a new two-step approach
for the solution growth of highly orientated ZnO microcolumn
arrays. The procedure involves deposition of ZnO template
layers of the desired morphology and subsequent overgrowth
of ZnO microcolumns on the templates.2 Similar strategies
have been employed subsequently by others.44–46
Acicular morphologies encountered for solution grown ZnO
are generally assumed to result from one of two mechanisms
based on dissolution–recrystallization phenomena or alter-
natively nanoparticle-oriented aggregation. Results obtained by
Andrés-Vergés et al. are more consistent with the latter process.
However, the processes by which the particles couple at a certain
size are not clear. The spherical particle size was found to be
critical; these fuse to form embryonic rod-like microcrystals. It
was postulated that the oriented coupling through the particle
c-axis may be related to the net polarisation that occurs in this
direction in ZnO microcrystals. The particles subsequently
develop by a ripening mechanism. In this latter step, the particles
increase in length but not in width until they reach the required
dimensions. The formation of the well-developed faces is the final
step that takes place under slower kinetic conditions.20 Recently
the group of Weller have reported similar observations for ZnO
nanoparticle assemblies.35
The overall pattern of results indicates the sensitivity of the
morphology of generated ZnO particles to the nature of the
complexing ligand. The presence of en in the baths typically
leads to star-like morphologies, those with TEA produce spheri-
cal aggregates and HMT leads to formation of rods/needles.
Influence of counter-ion on growth of ZnO crystallites
All other factors being equal, changing the counter-ion for zinc
often yields different crystallite morphology. Changes in
morphology may derive from the effects of species that act
as promoters or inhibitors for nucleation and growth processes.
For example, by impeding growth on one or more faces, the
crystal develops only in certain directions and hence the final
morphology is altered. The process is often termed ‘‘crystal
growth inhibition’’. For example, studies of the interaction of
diblock copolymers with ZnO crystal faces have shown that
these substances influence both morphology and particle size
distribution of growing crystallites by adsorbing onto specific
faces and retarding growth perpendicular to these faces.47 An
additional factor associated with the counter-ion of the zinc
salt is volatility of the component or derived species (especially
for use of zinc carboxylates).
Adsorption of ions from solution to substrate may arise simply
as a consequence of simple electrostatic attraction of species on
the charged surface of ZnO (as a consequence of the solution pH
and amphoteric surface hydroxyl groups). However specific
adsorption effects may occur and lead to concentration
2580 J. Mater. Chem., 2004, 14, 2575–2591
dependent effects. With increasing concentration of a given
specifically adsorbed ion (SAI), the point of zero charge (PZC)
may shift and in addition, the sign of the surface charge as given
by the zeta (f)-potential may reverse. The first effect occurs
because the SAI may interfere with the adsorption of potential-
determining ions (i.e. H1 and OH2 for ZnO in contact with
aqueous solution). The second effect arises from adsorption of
the SAI into the inner layer of the solid–solution interface, which
in turn can reverse the charge of the outer diffuse double-layer.
In the present study, the influence of zinc salt counter-ion on
crystallite morphology has been investigated. As rod-like ZnO
morphologies were of particular interest in the present study,
initial experiments were conducted using TO(F) glass substrates
and baths containing HMT and zinc nitrate, chloride, per-
chlorate, acetate or sulfate. Growth morphology is highly
sensitive to the level of supersaturation. Speciation calculations
indicate that there are small, but significant, differences in levels
of ‘‘zinc hydroxide’’ supersaturation and ZnO precipitation
points for baths containing different counter-ions. In terms of
thermodynamic stability of complexes, the calculations suggest
that precipitation should follow the order sulfate, nitrate y
chloride y formate and acetate. Empirical observations followed
the order sulfate, perchlorate, chloride, nitrate, formate and
acetate anion. The induction period for baths containing acetate
was significantly longer than for baths containing the other salts.
Baths containing sulfate ion did not produce films on TO(F) glass
in this study. Images obtained by SEM of films deposited using
zinc chloride (Fig. 7b), perchlorate (Fig. 7c) and nitrate (Fig. 7d)
on TO(F) glass indicated that films were not dense and the
acicular crystallites poorly orientated. Although films deposited
using zinc acetate were dense the rod-like crystallites were not
well orientated (Fig. 7a). The particle dimensions also varied
substantially for each counter-ion used.
In subsequent experiments, ZnO template layers were used as
substrates (prepared using TEA baths) deposited on TO(F)
coated glass. It was observed that good quality films, comprising
discrete upright ZnO rods, on template layers, could only be
formed when substrates were immersed just before or as visible
precipitation occurred. For comparison attempts were also made
to deposit films using zinc formate. Scanning electron micro-
graphs (ESI:{ Figure C) revealed that films deposited on ZnO
templates were well orientated. The dimensions of crystallites
varied significantly for each precursor used. In agreement with
the earlier results, no films were obtained using zinc sulfate.
From the results obtained in the present study, it is clear
that use of different counter-ions in baths containing growing
crystallites has a noticeable effect on final film morphology.
Differences in kinetics (growth rates, adsorption of counter-
ions on growing crystal faces, etc.) are the likely origin. The
Fig. 7 SEM images of ZnO films on TO(F) glass from HMT baths
(0.025 mol dm23 zinc salt and 0.025 mol dm23 HMT, pH 5, 90 uC ,
1 hour); A. Zn(CH3COO)2; B. ZnCl2; C. Zn(ClO4)2; D. Zn(NO3)2.
effect is more pronounced in precipitates removed from the
baths used to deposit thin films and characterised by SEM.
Material obtained from baths containing acetate (Fig. 8a: i),
formate (Fig. 8a: ii) and chloride baths (Fig. 8a: iii) were mainly
rod-like whilst crystallites obtained from baths containing
nitrate (Fig. 8a: iv) and perchlorate (Fig. 8a: v) anions were
acicular. Those using sulfate were composed of flat hexagonal
platelets (Fig. 8a: vi). Twinning was observed in ZnO particles
obtained from perchlorate baths.
A comparison of the dimensions of the crystalline pre-
cipitates from the bath with crystals comprising films (ESI:{
Figure C) gives an indication of the effect of the template layer.
The ZnO precipitates obtained from acetate and formate baths
differ significantly in their typical dimensions. By contrast, the
Fig. 8 a. SEM images (top) of ZnO precipitates obtained from HMT
baths (0.025 mol dm23 zinc salt and 0.025 mol dm23 HMT, pH 5, 90 uC,
1 hour); i. Zn(CH3COO)2; ii. Zn(HCOO)2; iii. ZnCl2; iv. Zn(NO3)2;
v. Zn(ClO4)2; vi. ZnSO4. b. TEM images and SAED (below) of ZnO
precipitates obtained from zinc acetate–HMT baths.
crystallites comprising films do not exhibit significant difference
in width. The platelets arising from sulfate baths (no films could
be deposited) are flat and are y30 mm wide, much larger than any
of the particles formed from other systems. The most likely
explanation for the formation of these platelets is considerable
crystal growth inhibition of the polar c faces by sorption of the
sulfate anions.48,49 Films obtained from chloride baths were
homogeneous and again show the effect of the template layer on
morphology. The powders, however, are composed of both
agglomerated rods and much larger flakes. Nitrate- and
perchlorate-containing baths produced less homogeneous films.
The driving force behind the directional and ordered aggre-
gation has been studied by Ocaña et al., who observed similar
growth patterns using TEM for the production of ellipsoidal
a-Fe2O3 particles with phosphate anions.50 Matijevic et al. also
studied the effect of anionic surfactants on the size and shape of
ZnO particles. They found that the addition of a fluorinated
polyether carboxylate to a system, which typically produced
intertwined ZnO crystallites, results in the formation of separ-
ated rod-like particles.30
The clear differences in CBD of the ZnO crystallite mor-
phology as a function of counter-ion are interesting, as they
contrast with previous results obtained by our group using
chemically similar systems and hydrothermal conditions.21 In
the first instance, it is clear that the behaviour of different
counter-ions cannot be adequately described as simple adsorp-
tion of different ions on the ZnO surfaces. As described earlier, it
is known that carboxylate species are often selectively adsorbed
on a variety of surface sites on ZnO. Studies in the vapour phase
have shown that formate species adsorb on the ZnO (0001) polar
(Zn terminated) and ZnO (101̄0) non-polar faces while the
pristine (0001̄) surface was unreactive.51 The difference in reac-
tivity was accounted for by the lack of acid–base pairs on the
oxygen polar face. Interaction of formate with the (0001̄) face can
occur at defect sites, i.e. at exposed Zn21 at steps or oxygen
vacancies. Three structures have been proposed for the formate
anion sorbing onto metal oxide surfaces, viz. bridging, bidentate
and unidentate.52 Formate was found to coordinate to the metal
in a bidentate fashion on the (101̄0) face and in a monodentate
manner on the (0001) face or (0001̄) face at defect sites.
TEM images and SAED of crystallites removed from ZnO
films deposited from HMT/acetate-containing baths onto
template layers are shown in Fig. 8b: i and ii. Precipitates
obtained from these baths were similarly characterised (Fig. 8b:
iii and iv). Although the sample preparation resulted in some
contamination of ZnO rods by organic material, it was clear
that many of the rods were composed of discrete subunits,
orientated and connected along the c-axis of the growing
crystallites. The subunits are of the order y1 mm in length, for
both ZnO rods formed in solution and as thin films. The
mechanism of growth of ZnO rods is clearly complex and may
involve a number of processes. More detailed studies were
conducted using TEM and focussed ion beam (FIB) sample
preparation techniques and are discussed later.
Influence of substrate
The effect of the substrate on the nature of the ZnO films formed
was studied. Deposition experiments were performed using
identical bath conditions (0.025 mol dm23 Zn(CH3COO)2 and
0.025 mol dm23 HMT, pH 5, 90 uC, 1 h) but different
substrates: gold-coated TO(F) glass, ZnO template layers on
TO(F) glass, TO(F) glass and single crystalline (0001) sapphire.
Micrographs for the first three are shown in Fig. 9a
(identical magnification). There are marked differences in
average diameter of the crystallites deposited on the three
substrates. By contrast, the lengths of rods (as-determined by
cross-sectional SEM) were similar, in the range 2–2.5 mm. The
films deposited on both Au-coated TO(F) substrates and the ZnO
template layers were well-aligned. For the microcolumns grown
J. Mater. Chem., 2004, 14, 2575–2591 2581
 relationships between (112̄0) ZnO and (011̄2) Al2O3 and (0001)
ZnO and (01̄11) Al2O3.56 Deposition experiments were also
conducted on substrates that were used to produce the films
shown in Fig. 10a, after rinsing in nitric acid to remove deposits
and then using the standard cleaning procedure. Arrays of ZnO
rods (rather than the previously observed rotational twins)
were found to grow in troughs or scratches, formed presumably
through the cleaning process (Fig. 10b).

Influence of solution pH
In the present study, the influence of bath pH on as-deposited
films of ZnO, grown on ZnO template layers, using the zinc
acetate/HMT system was investigated. As for zinc nitrate/HMT
systems, homogeneous precipitation occurred more rapidly in
zinc acetate/HMT systems where no pH adjustment was made as
compared to those where the pH was lowered. The SEM
micrographs (Fig. 11) reveal that films from such baths (pH ~
6.8) comprised crystallites of different sizes with poorly
developed crystal faces. In subsequent experiments, baths were
adjusted to pH ~ 5 using HCl or acetic acid. In these cases, films
were composed of rods with well-defined faces and in addition,
the individual particles had a smaller average diameter than those
obtained from the former. The non-chloride containing baths
also showed the slowest homogeneous precipitation, i.e.
possessed the longest induction time before reaction occurred.

Effect of changing ionic strength

For sparingly soluble metal compounds, such as ZnO or Zn(OH)2
in the present study, increasing the ionic strength tends to increase
the solubility of the solid phase because the activity of the ions
decreases. In order to investigate the effects of ionic strength on
Fig. 9 a. SEM images (top) of ZnO films from HMT baths (0.025 mol dm23 the nature of ZnO films, experiments were conducted using the Zn
Zn(CH3COO)2 and 0.025 mol dm23 HMT, pH 5, 90 uC , 1 hour) on acetate/HMT system on ZnO template layers.
i. Au-coated TO(F) glass; ii. ZnO template layers on TO(F) glass; This work involved the variation of the ionic strength in the
iii. TO(F) glass. b. Crystallite size distribution (below) determined from Zn–HMT system using KCl as a background electrolyte. The
SEM images of ZnO nanocolumns deposited on Au-TO(F) glass. Radii ionic strength of the system was varied in the range [KCl] ~
of ZnO microcolumns reach terminal values within few minutes of
growth, leading to development of a stable size range that could be
0.025–0.4 mol dm23. The precise ionic strength of the bath
fitted to a normal Gaussian distribution (400 data sets, R ~ 0.984). could not be determined as solution pH was adjusted by
dropwise addition of glacial acetic acid. However, useful trends
directly on Au/TO(F) glass there are no signals arising from could be identified. It was observed that as [KCl] increased,
polycrystalline ZnO template layers (ESI:{ Figure D). Crystallite
size distributions were determined from micrographs of ZnO
nanocolumns deposited on Au-TO(F) glass (Fig. 9b) in order to
quantify the polydispersity and homogeneity of films. Using an
image analysis routine,53 average (mean) radii values of 266 nm
(standard deviation 54 nm) were determined.
Growth of ZnO films on single crystal (0001) sapphire from
HMT-containing baths was attempted. Preliminary experi-
ments using similar material from co-workers appeared to
indicate that unusual 60u rotational twinned structures could
be deposited (Fig. 10a). Previously a great deal of research
has focussed on ZnO films grown on (0001) sapphire
substrates.54,55 Other workers have demonstrated epitaxial

Fig. 10 SEM images of ZnO films deposited on single crystal sapphire

(0001) from baths containing zinc acetate (0.025 mol dm23) and HMT Fig. 11 SEM images of ZnO films from HMT baths (0.025 mol dm23
(0.025 mol dm23) at 90 uC. a. Starlike crystallites obtained, some 60u Zn(CH3COO)2 and 0.025 mol dm23 HMT) deposited on ZnO template
rotational twins are observed; b. rod arrays grown in troughs on layers at A. pH 6.8; B. pH 5 (adjusted with HCl); C. pH 5 (adjusted with
cleaned substrate previously used to deposit films shown in a. CH3COOH).

2582 J. Mater. Chem., 2004, 14, 2575–2591

homogeneous precipitation occurred more rapidly. It was
apparent that no ZnO rods were deposited from baths with
[KCl] w 0.2 mol dm23. The effect (if any) of zinc-chloro
complex formation, which would tend to increase stability to
precipitation with increasing [KCl], appears to be insignificant.
As shown in Fig. 12, SEM images reveal that within the
range [KCl] ~ 0.025–0.15 mol dm23 the width of individual
crystallites was decreased from those formed from baths
containing no KCl. The observation could be rationalised in
terms of specific ion adsorption e.g. on kink sites on the a-faces
of the ZnO crystal, which could indirectly promote growth in
the c direction. However, no measurements were done to
determine if particles also decreased in length. Ionic strength
effects have been demonstrated for solution growth of calcite
crystals, at low ionic strength, aggregation and alignment
phenomena are present, no such mechanisms operate at higher
ionic strength.57 For the range [KCl] w 0.15 mol dm23 parti-
cles changed morphology from rods to discs. Similar results
were obtained using zinc formate rather than acetate as a
precursor (see ESI:{ Figure E).
The influence of ionic strength on ZnO film growth is clearly
complex. From CBD experiments employing zinc acetate or
formate and HMT, it appears that there is a limiting ionic strength
whereby film growth is possible, beyond which only homogeneous
precipitates may be formed. Changing ionic strength also has
morphological effects, as evidenced from decreasing ZnO rod
crystallite width with increasing ionic strength.
Fig. 12 SEM images showing the effect of ionic strength on ZnO films
(on template layers) obtained from HMT baths (0.025 mol dm23
Zn(CH3COO)2 and 0.025 mol dm23 HMT, pH 5, 1 h). The ionic strength
was varied using different concentrations of KCl: A. 0 mol dm23;
B. 0.025 mol dm23; C. 0.05 mol dm23; D. 0.075 mol dm23; E. 0.1 mol dm23;
F. 0.125 mol dm23; G. 0.3 mol dm23; H. 0.4 mol dm23.
Effect of varying growth times
Andrés-Vergés et al. found that for ZnO grown under hydro-
thermal conditions, an increase in bath temperature for
solutions containing ZnCl2 and HMT resulted in a change in
morphology from rods to needle morphology.20 In these
studies, the length and final shape of the rod were shown to be
very dependent on the growth time. It is also reported that
ageing time also has a major influence on final crystal
morphology, ZnO habit changing from flaky aggregates to
flower-like morphology with increasing growth time.58
The development of ZnO rod array films (using zinc acetate)
on ZnO template layers as a function of deposition time was
investigated. Films were removed from baths after different
periods of time and characterised by SEM. The micrographs
(Fig. 13) show that crystallites possess an average diameter of
y300 nm after 2 minutes and y1 mm after 1 h. The cross-
sectional SEM images allow the length of the rods to be
determined. However it was only possible to determine the
average rod length for those grown for times greater than
10 min. At shorter growth times, it was difficult to distinguish
the template layer from the ZnO rod growth (ESI:{ Table B).
Evidence for oriented attachment processes was obtained for
homogeneous precipitates lying on films (Fig. 13D).
Summary of preceding survey of factors that influence CBD of ZnO
In general, the most likely crystal morphology for ZnO under
kinetically controlled growth conditions and at low super-
saturation is as acicular rods, whereas under thermodynamic
control, prismatic structures are expected.17 Hence by limiting
the supply of one reactant employing a large excess of the
other, anisotropic growth is facilitated along the polar c-axis of
the zincite structure. Several factors exert a profound effect on
the morphology of ZnO thin films grown by CBD and there are
important differences between CBD and the chemically similar
hydrothermal routes to ZnO powders or films. Under hydro-
thermal conditions, Trindade et al. have reported that neither
Fig. 13 SEM images of ZnO films from HMT baths (0.025 mol dm23
Zn(CH3COO)2 and 0.025 mol dm23 HMT) deposited on ZnO template
layers deposited for: A. 2 min; B. 5 min; C. 10 min; D. 20 min; E. 30 min;
F. 1 h. For D, a large dumbbell precipitate formed homogeneously in the
bath appears to indicate formation by orientated attachment.
J. Mater. Chem., 2004, 14, 2575–2591 2583
 solution pH, choice of counter ion and ionic strength, bath
temperature or reaction time exert any significant influence on
the characteristics of resultant ZnO powders. The choice of
ligand and nature of hydrothermal treatment were found to be
important,21 results that are clearly different from those
obtained for CBD films.
The results obtained in the present study suggest that nuclea-
tion and growth promotion/inhibition are significant and vary
considerably with small changes in bath chemistry. For example,
only large homogeneous precipitates can be obtained from
deposition baths in the presence of sulfate, which appears to
completely impede nucleation and growth of ZnO thin films, on
both TO(F) glass and ZnO template layers. Moreover, the
morphology of these precipitates (flat hexagonal y30 mm wide
platelets) is very different to those obtained using other zinc salts
(i.e. acetate, formate, nitrate, chloride and perchlorate). Flattened
crystallites were also observed in experiments involving increasing
ionic strength (using KCl) and HMT/zinc acetate systems.
In the aqueous CBD of ZnO, the influence of counter-ion on
crystallite morphology may be through weak electrostatic
forces (e.g. specific adsorption of ions resulting in change in
p.z.c. or f-potential, double layer compression, etc.) or more
significant physi- or chemisorption behaviour (e.g. at attractive
kink sites on crystal faces). However, increasing ionic strength
also tends to increase the solubility of the solid phase being
deposited but decrease the values of thermodynamic stability
constants of the metal–ligand complexes in deposition baths.
These facts highlight the complexity of CBD systems and
illustrate the difficulty in unambiguously assigning direct
causal relationships between the many interrelated parameters
(e.g. crystal morphology and choice of counter-ion).
Dependence of film morphology on point of introduction of
substrate in deposition bath
Empirically it has been found that the size of the crystallites
comprising films depends largely on when the ZnO template
substrate is immersed in the bath. If the substrate is immersed in
the bath from the point at which the pH of the solution is adjusted
to 5 at ambient temperature, the nanocrystalline template
dissolves as the bath is heated to 90 uC and no film is deposited.
If the substrate coated with a nanocrystalline ZnO template is
inserted in the bath about 15 min after the desired temperature of
90 uC is reached (TA), the bath produces no homogeneous
precipitation during the 2 h reaction period and a dense white film
is obtained. Images obtained by SEM reveal that the particles are
y 250 nm wide and y1.2 mm long (Fig. 14C,D). However, if the
substrate is placed in the heated deposition bath at the point of
onset of visible turbidity (ca. 30 min after reaching 90 uC) (TB), a
transparent film is obtained and some homogeneous precipita-
tion is present in the bath (Fig. 14A,B). The ZnO rods comprising
Fig. 14 SEM images of ZnO films deposited on nanocrystalline ZnO
templates introduced in baths from time: A,B TB and C,D TA.
2584 J. Mater. Chem., 2004, 14, 2575–2591
such films are y90 nm wide and y600 nm long. The aspect ratio
of the latter (y6.7), grown from time TB, is clearly better than the
former (y4.8) grown from time TA.
Use of nanocrystalline template layers for overgrowth of ZnO
films by forced hydrolysis of zinc carboxylates and using different
chelating ligands
The pattern of results in the preceding sections indicated that
the simplest and most promising synthetic route towards
growth of ZnO nanorod array films would involve deposition
on to nanocrystalline ZnO template layers. Preliminary
experiments involved the forced hydrolysis of zinc acetate
and zinc formate using nanocrystalline ZnO templates on glass
substrates. It was very apparent that very low levels of
homogeneous precipitation occurred using baths containing
zinc acetate, moreover no visible homogeneous precipitation
could be detected using zinc formate solutions. Transparent
films were obtained with both precursors. Subsequent experi-
ments involved the use of these template layers for the
overgrowth of ZnO films deposited using different chelating
ligands viz. en, TEA and HMT. Reactions were similar to those
reported earlier for each ligand on TO(F) glass substrates. The
transparent films were characterised by XRD and SEM.
X-Ray diffractograms obtained for nanocrystalline ZnO tem-
plates and overlying films are shown in Fig. 15. For films pro-
duced by forced hydrolysis of zinc acetate, using en and HMT,
there is strong enhancement of the (002) peak. For films deposited
from baths containing TEA ligand the effect is not as pronounced.
As indicated earlier, attempts to deposit ZnO films by forced
hydrolysis of zinc carboxylates with no added ligand were
unsuccessful using TO(F) glass substrates. However, SEM
images provide evidence that films are produced using identical
baths (0.025 mol dm23 zinc acetate or zinc formate) on
nanocrystalline ZnO templates (Fig. 16A,B and C,D respec-
tively). The films were homogeneous and composed of rod-like
array structures (ca. 140 nm wide and 650 nm long) when zinc
acetate was used (Fig. 16B), in contrast to the more nodular
particles (ca. 140 nm wide and 300 nm long) obtained when zinc
formate was used (Fig. 16D).
Films obtained from baths using en as a ligand for zinc were
composed of fused columnar crystals (ca. 110 nm wide and
300 nm long), with well-defined hexagonal end facets (c-faces).
Hence the presence of a template layer appears to direct growth
towards acicular, rather than starlike, morphologies. The
crystallite formed homogeneously in the bath and laying on top
of the film possesses a star-like morphology (Fig. 16E, inset)
and resembles those comprising films deposited over TO(F)
glass (Fig. 4a). Films deposited using the TEA ligand have a
Fig. 15 XRD patterns of ZnO films from: a. HMT-system ([ZnAc] ~
0.025 mol dm23, [HMT] ~ 0.025 mol dm23, pH 5, 90 uC, 2 h); b. en-
system ([ZnAc] ~ 0.018 mol dm23, [en] ~ 0.042 mol dm23, pH 11,
70 uC, 2 h); c. forced hydrolysis of zinc acetate (0.025 mol dm23, 90 uC, 2 h);
d. TEA-system ([Zn(NO3)2] ~ 0.018 mol dm23, [TEA] ~ 0.042 mol dm23,
pH 11, 70 uC, 2 h) deposited on nanocrystalline ZnO templates.
Nanocrystalline ZnO template shown in e. ZnAc ~ zinc acetate.
Fig. 16 SEM images of ZnO films deposited on nanocrystalline ZnO templates from: A,B. forced hydrolysis of zinc acetate; C,D. forced hydrolysis
of zinc formate; E,F. en-system; G,H. TEA-system.
more globular appearance and are less defined than others in
these investigations.
Based on the success of deposition baths containing HMT
and using microcrystalline ZnO template layers, further work
was conducted on the influence of the nanocrystalline template
layer on the subsequent growth of films, using HMT as the
ligand. The results were unambiguous, micrographs obtained
from films provide clear evidence that all other factors being
equal, the width of ZnO rods grown on nanocrystalline
(Fig. 17D, E and F) rather than microcrystalline (Fig. 17A, B
and C) template layers is far smaller.
Influence of choice of counter-ion on growth of ZnO rods on
nanocrystalline ZnO template layers
The influence of the zinc counter-ion on the ZnO rod dimen-
sions was investigated using similar strategies to those
employed for microcrystalline template layers. Micrographs
obtained from films (Fig. 18) suggest that films grown from
baths containing zinc nitrate and HMT possess crystallites with
the largest aspect ratio (ca. 10).
Growth of ZnO rods on nanocrystalline ZnO templates from
baths containing the products of decomposition of HMT
(formaldehyde and ammonia)
The amine HMT decomposes slowly in heated aqueous solu-
tions to yield ammonia and formaldehyde as initial reaction
products. Attempts were made to grow ZnO rods on nano-
crystalline ZnO templates directly from aqueous baths (90 uC)
containing zinc acetate, ammonia and formaldehyde. The
concentrations of the latter components were chosen assuming
stoichiometric decomposition of HMT. Reactions were carried
out at the initial solution pH of 6.9 and at pH 5, which was the
pH employed earlier in this study for the successful CBD of
ordered arrays of ZnO rods. No precipitation occurred from
J. Mater. Chem., 2004, 14, 2575–2591 2585
 Fig. 17 A. CBD-ZnO microcrystalline templates on TO(F) glass comprising nodular grains (deposited from TEA-systems); B,C. ZnO microcolumns
grown on CBD-ZnO templates from HMT-system ([Zn] ~ 0.025 mol dm23, [HMT] ~ 0.025 mol dm23, pH 5, 90 uC); D. nanocrystalline ZnO templates
on glass microscope slides (prepared by sol gel (SG) method); E,F. ZnO nanocolumns grown on SG-ZnO templates from HMT-system.

baths held at pH 5 in which dissolution of the nanocrystalline
ZnO template layer occurred. Experiments were also conducted
without formaldehyde i.e. only zinc acetate and ammonia pre-
sent. Speciation calculations were performed in order to assess
the initial degree of supersaturation and identify the important
components in deposition baths (Fig. 19a). Baths containing
either ammonia or ammonia and formaldehyde were found be
supersaturated with respect to zinc hydroxide at pH 6.8. These
results are in general agreement with experimental observations.
Similar thermodynamic calculations were performed for baths
containing different concentrations of HMT. Speciation calcula-
tions indicate that with an increase in HMT concentrations, there
are small differences in levels of ‘‘zinc hydroxide’’ super-
saturation and precipitation points (Fig. 19b). The hydroxide
precipitation points for the baths containing Zn:HMT in the
ratios, 2:1, 1:1, 1:2, 1:4, 1:8 occur at pH values of 6.8, 6.8, 6.9.
7.04 and 7.16 respectively. The calculated pH, at which the bath
is supersaturated with respect to the hydroxide containing
Zn:HMT in a 1:1 ratio, is exactly the same as the bath con-
taining the stoichiometric amount of ammonia and formalde-
hyde (or ammonia alone). However, these small differences in
supersaturation at a given pH do appear to be associated with
morphological changes in films, as SEM images for the films
deposited using different HMT concentrations show that as the
HMT concentration increases, the width of the rods decreases.
The aspect ratio for the rods, however, does not change

Fig. 19 a. Speciation diagram of Zn–NH3–HCHO system at 25 uC with

Fig. 18 SEM images of ZnO films on nanocrystalline ZnO templates
from HMT baths (0.025 mol dm23 zinc salt and 0.025 mol dm23 HMT,
pH 5, 90 uC, 2 hours); A,B. Zn(CH3COO)2; C,D. Zn(HCOO)2;
E,F. ZnCl2; G,H Zn(NO3)2.
A ~ [Zn21] ~ 0.0083 mol dm23, B ~ [CH3COO2] ~ 0.016 mol dm23,
C ~ [NH3] ~ 0.0083 mol dm23 and [HCHO] ~ 0.016 mol dm23. Dashed
line represents zinc hydroxide precipitation point. b. Speciation diagram
of Zn–HMT system at 25 uC showing [Zn21]free and zinc hydroxide
precipitation points for increasing [HMT].

2586 J. Mater. Chem., 2004, 14, 2575–2591

significantly i.e. narrower rods tend to reach smaller terminal
lengths.
For films deposited from aqueous baths (90 uC) at initial
pH ~ 6.9, containing zinc acetate, ammonia and formaldehyde
on nanocrystalline ZnO template layers, XRD and SEM
measurements were recorded. X-Ray diffraction patterns were
very similar to those obtained for ZnO rod films deposited from
baths containing HMT and zinc acetate on ZnO template layers
and exhibited enhancement of the (002) reflection. In agreement
with previous results, no evidence for incorporation of hydroxide
was obtained from ATR-FTIR analysis. SEM micrographs were
also very similar to those obtained for analogous HMT systems,
as films comprised ordered rod arrays of perpendicularly
orientated ZnO rods on nanocrystalline template layers
(Fig. 20A,B). The dimensions of the rods were y85 nm in
width and y600 nm long, which were smaller than those
obtained for films deposited from HMT/zinc acetate systems and
grown on nanocrystalline ZnO templates (Fig. 18A). For baths
containing only zinc acetate and ammonia, the films comprised
rods with globular caps (Fig. 20D) in contrast to the rods formed
when formaldehyde was present (Fig. 20A,B).
Effect of growth time
The XRD patterns for the films, deposited for different times,
are shown in Fig. 21. All patterns are consistent with zincite
phase. As the growth time increases, the (002) reflection
increases in relative intensity, consistent with formation of ZnO
rod crystallites. The (100) and (101) reflections remain broad
during deposition in contrast to the (002) peak, which becomes
noticeably sharper with increasing deposition time. This
observation is consistent with the growth of ZnO rods
crystallites, along the c-axis.
Scanning electron micrographs were recorded and grain size
distributions determined for films deposited for different growth
periods (ESI:{ Figure F). Grain size profiles were unimodal,
generally asymmetric and typical for abnormal grain growth
where grain size distributions are directed by different substrate–
grain relationships. The average grain diameter increased linearly
from y82 nm after 5 min to y136 nm after 2 h, corresponding to
an average growth rate of y0.23 nm min21. From cross sectional
SEM images it was clear that growing ZnO rods tend to broaden
slightly with deposition time, hence the definition of ‘‘average
grain size’’ is not precise and a direct comparison is only sensible
for rods of similar length.
Fig. 21 XRD patterns of ZnO films deposited on ZnO nanocrystalline
template layers from HMT baths (0.025 mol dm23 Zn(CH3COO)2
and 0.025 mol dm23 HMT, pH 5, 90 uC) for different periods of time.
a. 2 min. b. 5 min. c. 10 min. d. 20 min. e. 30 min. f. 1 h. g. 2 h.
In attempts to improve the aspect ratio of ZnO rods, as-
deposited ZnO films were immersed into fresh baths. An
additional motivation for this work was to gain further insights
into the actual growth mechanism of ZnO single crystals. In
preliminary experiments, a direct comparison was made of films
prepared using baths (HMT–zinc acetate system) identical to
those employed previously but using three different reaction
sequences. In the first group, ZnO rods were deposited on
nanocrystalline ZnO templates for 2 h (Method 1). In the second
group, ZnO rods were deposited on nanocrystalline ZnO for 2 h
and subsequently transferred into fresh heated baths, at the point
where visible turbidity became apparent, for a period of 1 h
(Method 2). The final experiments followed the second procedure
and then films were removed and placed in another fresh bath at
the critical point for an additional 1 h (Method 3).
Films were characterised by SEM and it was apparent that
although the length of the ZnO rods increases upon exposure to
fresh CBD solution, the width of rods also increases and hence
no improvement in aspect ratio is observed (Table 1 and Fig. 22).
In addition, crystallites comprising the films become very densely
packed, which is not desirable for applications that require post-
deposition processing such as infiltration of dye molecules onto
the rods. No conclusive evidence for film growth via orientated
attachment processes could be gained. The resolution of the
SEM is insufficient for detailed interpretation of growth of
nanostructured films. However, it was clear that in common with
the microcrystalline systems described earlier (see Fig. 13D),
growth of homogeneous precipitates may occur through
orientated attachment processes (Fig. 22G).
Attempts were made to gain more detailed structural informa-
tion, by use of HR-TEM and focussed ion beam (FIB) sample
preparation techniques. Samples were prepared from ZnO thin
films deposited by Method 1 (2 h deposition) and Method 2 (2 1
1 h deposition) as described previously (ESI:{ Figure G). The FIB
technique allows thin (y50 nm) specimens for HR-TEM to be

Table 1 Rod dimensions and aspect ratio of the films deposited for
the various times

time diameter/nm length/nm aspect ratio

a
2 min y77 ¡ 4 —
a
5 min y83 ¡ 6 —
a
10 min y85 ¡ 10 —
20 min y102 ¡ 9 220 ¡ 10 2.1
30 min y108 ¡ 11 530 ¡ 25 4.9
1h y122 ¡ 10 1000 ¡ 50 8.2
2h y137 ¡ 11 1500 ¡ 75 10.9
Fig. 20 SEM images of ZnO rod arrays deposited on nanocrystalline 2h11h y320 ¡ 11 2000 ¡ 100 6.3
ZnO templates layers from baths containing: A,B. zinc acetate 2h11h11h y560 ¡ 12 2500 ¡ 125 4.5
(0.025 mol dm23), HCHO (0.016 mol dm23) and NH3 (0.0083 mol dm23) a
The presence of the template layer compromised accurate length
(pH 6.9, 90 uC, 2 h) and C,D. zinc acetate (0.025 mol dm23) and NH3 measurements.
(0.0083 mol dm23) (pH 6.9, 90 uC, 2 h).

J. Mater. Chem., 2004, 14, 2575–2591 2587


Fig. 22 SEM images of ZnO rod arrays deposited on nanocrystalline
ZnO template layers for: A,B. 2 h (Method 1); C,D. 2 h 1 1 h
(Method 2) and E,F,G. 2 h 1 1 h 1 1 h (Method 3).
obtained with minimal damage in preparation. The integrity of
the original structure is maintained and each interface is clearly
represented (Fig. 23). Moreover, the single crystalline nature of
individual ZnO rods is clear from SAED.
The crystalline rods of ZnO formed comprise oriented nano-
crystallites, in which the lattice planes of individual particles are
perfectly aligned to form a single crystalline entity (Fig. 24). It is
also evident from TEM that the outer layers of ZnO rods (ca.
y5 nm) are composed of randomly orientated nanoparticles.
The atomic structure is clearly resolved in Fig. 25. Nucleation
will involve conditions of low to moderate supersaturation in
Fig. 24 HR-TEM of single ZnO rod (mag. 310k) grown on nanocrystalline ZnO template layer (Method 1). a. The outer porous layer is composed
of randomly orientated nanoparticles. b. Area indicated in a is shown in greater detail. The extended lattice structure, comprising individual ZnO
nanocrystals, is clearly evident.
2588 J. Mater. Chem., 2004, 14, 2575–2591
Fig. 23 HR-TEM of cross-section of ZnO thin film sample prepared by
Method 1 (mag. 627.5k) and SAED recorded for single crystalline ZnO
rod along the (100) zone axis. For reference the first series of reflections
are indexed. Double diffraction phenomena occurs leading to the
presence of forbidden reflections (e.g. (001)) and are denoted by X.
conjunction with an active interface. Under typical dynamic
bath conditions, it is more likely that these external aggregates
of nanoparticles will be subject to dissolution–reprecipitation
processes. The predominance of one process over the other will
determine the velocity of growth in a given direction. Sorption
of ions or anisotropic nanoparticles with favourable orienta-
tion approaching the polar faces of wurtzite ZnO (in contrast
to the non-polar faces), is expected to be effective and prohibit
desorption and rearrangement of growth units.
Evidence for orientated attachment of large ZnO crystallites
to existing ZnO rods grown on templates has been obtained,
for both films grown by Method 1 (2 h deposition on ZnO
templates) and Method 2 (2 h deposition followed by 1 h
growth in fresh baths). The pattern of results obtained from the
TEM studies appears to show that the process is not as
significant as the mechanism involving ZnO nanocrystal
aggregation. An example of the phenomena for films grown
by Method 1 is shown in Fig. 26, where the rod–rod interface of
a composite unit is shown. Although visually there appears to

Fig. 25 HR FEG-TEM image of ZnO rod showing lattice structure.
be considerable misalignment of the rods, at higher magnifica-
tion (Fig. 26B) the lattice configuration between the two
separate rods is evidently good. Similar structures were found
for ZnO rod arrays grown by Method 2. Hence it appears that
under the appropriate conditions, ZnO rods can develop by a
combination of different growth processes.
The primary aim of the HR-TEM investigation was to better
understand the growth mechanism of ZnO rods on nano-
crystalline templates. If we assume that deposition occurs from
solutions of low to intermediate levels of supersaturation,
growth would be expected to proceed via dislocation-based
growth spiral or 2D-nucleation based mechanisms. For all
samples in the present study (both in preliminary work and the
later samples) no evidence for the presence of significant
concentrations of defects such as dislocations or stacking faults
Fig. 26 (Left) HR-TEM of ZnO rods grown by Method 1 showing orientated attachment of two smaller rods. (Right) Inset is shown at higher
magnification to demonstrate the good lattice alignment of the two ZnO single crystals.
in ZnO rods could be obtained, from HR-TEM or SAED. The
pattern of results obtained in this study for the deposition of
ZnO rods on nanocrystalline ZnO template layers do not
support a mechanism based exclusively on a growth spiral.
However, our observations do not preclude the possibility that
such a mechanism may operate in early stages of growth.
A likely sequence of events would be 2D-nucleation on the
template layers, via the orientated aggregation of polar nanopar-
ticles and the subsequent development of single crystalline material
by dissolution-reprecipitation phenomena. As supersaturation in
baths decreases due to consumption of reagents, 2D-nucleation
and growth processes on surface terraces decrease rapidly in
significance. Crystal faces develop a smoother appearance through
aging via dissolution-recrystallisation processes, whereas adsorp-
tion and incorporation of growth units is limited by diminishing
concentrations of growth sites such as kinks on crystal faces.
3. Conclusions
The influence of reaction conditions including ligand, counter-
ion, pH, ionic strength, supersaturation, deposition time and
substrate on the nature of ZnO films grown by CBD has been
examined and discussed, the most important of which are
summarised in Fig. 27.
Use of different ligands leads to deposition of ZnO films with
different morphologies on TO(F) glass, e.g. ethylenediamine
usually gave rise to starlike crystals, TEA produced nodules
and HMT usually produced rods. Similarly, using identical
ligands but different zinc salts often gives rise to different
morphologies, with sulfate usually producing the most drama-
tic change. Speciation studies and empirical evidence suggest
that growth of acicular ZnO morphologies by CBD are best
obtained by limiting the concentration of one reactant (i.e.
either Zn21 or OH2) in the presence of an excess of the second
component, promoting the kinetically controlled form of the
final crystal. The presence of an existing ZnO substrate can
influence the morphology of subsequent ZnO growth. A
striking example is that ZnO films could not be deposited on
J. Mater. Chem., 2004, 14, 2575–2591 2589
 Fig. 27 Schematic overview of main results obtained in this study.
TO(F) glass by the forced hydrolysis of zinc carboxylates, but
rod-like arrays of ZnO were produced from chemically
equivalent baths on nanocrystalline ZnO templates. Templates
of ZnO can direct the deposited film e.g. in the en system, which
typically produced starlike crystallites, columnar structures
were deposited. High surface area arrays of upright ZnO
crystallites on nanocrystalline ZnO templates have been
deposited using baths containing HMT. The point at which
the substrate is introduced into the bath is crucial and leads to a
significant difference in the width of the rods and optical
transparency of the films. Thinner rods are produced when the
substrate is immersed in the bath at the point of visible
turbidity i.e. at the onset of precipitation. Zinc nitrate pre-
cursors tended to lead to formation of rod arrays on
nanocrystalline templates that possessed the highest aspect
ratios. It was also found that deposition baths containing zinc
acetate and the decomposition products of HMT (initial pH ~
6.9) produced similar rods arrays of ZnO to the standard HMT
system (initial pH ~ 5). It was found that increasing HMT
leads to slightly thinner rods, however, there was no change in
the aspect ratio as crystallites were shorter in length.
Attempts have been made to improve the aspect ratio of ZnO
rods, by growing ZnO rods using fresh solutions on preformed
ZnO rods arrays. The procedure was unsuccessful. The final
ZnO films comprised dense rod arrays that were increased in
length and in width with some fused columnar structures.
HR-TEM has yielded important structural information and
an insight into the growth of ZnO rods by CBD. The most
likely sequence of events involves 2D-nucleation on ZnO
template layers, growth via orientated aggregation of polar
nanoparticles and subsequent development of single crystalline
material by dissolution–reprecipitation.
Experimental
Film growth
Films of ZnO were grown on substrates immersed in aqueous
deposition solutions using forced hydrolysis and CBD
2590 J. Mater. Chem., 2004, 14, 2575–2591
methods. Substrates (microscope slides, TO(F) glass, micro-
or nanocrystalline ZnO templates, Au-coated TO(F) glass,
single crystal sapphire (0001)) were cleaned by a standard
procedure. A Mettler Toledo MA 235 pH/ion analyser and
InLab 413 electrode were used to record solution pH.
Modelling of solutions was performed using SPECIES
(AcadSoft Ltd.). The program calculates and displays specia-
tion curves as a function of pH. The input parameters for
SPECIES are the stability constants for the homogeneous
equilibria being modelled and their stoichiometric coefficients.
Film thickness measurements were performed using the QCM
technique, using a Maxtek PM-700 Series Plating Monitor and
probe employing a quartz crystal oscillator (5 MHz, AT-cut,
gold crystal, unpolished). The QCM apparatus was controlled
and experimental data recorded by use of a PC.
Forced hydrolysis was performed using solutions of zinc
acetate (0.025 mol dm23) or zinc formate (0.025 mol dm23) on
TO(F) glass. Final pH values for the zinc acetate and formate
baths were 6.9 and 6.4 respectively. Solutions were heated at
95 uC for a period of two hours. For experiments involving
nanocrystalline ZnO template layers on glass microscope slides,
substrates were immersed in solutions at ambient temperature
prior to heating of baths using the temperature-controlled
waterbath.
Deposition of ZnO thin films by CBD methods using ethyl-
enediamine (en) as ligand was achieved using aqueous solutions
of zinc acetate (0.018 mol dm23) and en (0.042 mol dm23),
adjusted to a pH of 11 with aqueous sodium hydroxide
(5 mol dm23). Films were grown at the desired temperature
(50–80 uC) for 1 hour. For TEA systems, solutions contain-
ing zinc nitrate (0.018 mol dm23) and triethanolamine
(0.072 mol dm23), adjusted to a pH of 11 with aqueous
sodium hydroxide (5 mol dm23), were used. Films were grown
over a range of temperatures (50–80 uC) for 1 hour. Unless
otherwise stated, experiments employing HMT were conducted
using bath solutions containing zinc acetate (0.025 mol dm23)
and HMT (0.025 mol dm23), adjusted to pH 5 with aqueous
acetic acid (5 mol dm23), for deposition times of 1 h.
 Microcrystalline template layers were deposited using the
TEA method described above. Nanocrystalline templates were
grown by sol–gel methods,59 involving hydrolysis and con-
densation of an alcoholic Zn-precursor under basic conditions,
in order to produce a homogeneous sol for subsequent dip-
coating on glass microscope slides. A suspension of zinc acetate
dihydrate (17.8 g, 0.08 mol) in n-propanol (120 cm3) was heated
under reflux conditions (130 uC, 20 min). The mixture was
allowed to cool to ambient temperature before rapid addition
of tetramethylammonium hydroxide (TMAH; 25% in MeOH,
36 cm3) to yield a transparent nanoparticulate ZnO coating sol
(y0.5 mol dm23), which could be concentrated further by
removal of solvent using a rotary evaporator. Films were
prepared by dip-coating substrates in the sol and sintering in a
furnace at 400 uC for 2 min.
Characterisation studies
X-Ray diffraction studies were performed using secondary
graphite monochromated Cu Ka radiation (40 kV) on either a
Philips X’Pert Materials Diffractometer (APD) or Bruker AXS
D8 diffractometer. Measurements on the former were taken
using a glancing angle incidence detector at an angle of 3u, for
2h values over 10–95u in steps of 0.04u with a count time of 2 s.
For the latter, scans were done over 2h values of 5u–90u and a
step size 0.01u or 0.02u. Scanning electron microscopy (SEM)
and energy dispersive X-ray analysis (EDAX) on carbon-
coated films was performed using a Philips Excel 30 FEG SEM
instrument or Philips 525 SEM instrument with an EDAX
DX4 EDS unit. Samples were carbon-coated (using an
Edwards Coating System E306A) for EDAX or otherwise
gold coated (Edwards Sputter Coater S150B). The focused ion
beam (FIB) thinning technique was used to prepare cross-
sectional ZnO thin film samples for characterisation by high
resolution transmission electron microscope (HR-TEM). This
work was conducted at FEI Bristol using a FIB 200 instrument
(Ga1 beam). In order to minimise ion beam induced artefacts
(i.e. amorphous layer and Ga implantation), a Pt buffer layer
was deposited on the area of interest on the specimen by ion-
beam assisted CVD, prior to thinning. TEM electron micro-
scopy was accomplished using either a Philips CM200 (200 kV)
microscope or a FEI Tecnai F30 FEGTEM (300 kV) instru-
ment. TEM specimens were mounted on a carbon-coated
copper TEM grid. Electronic absorption spectra were obtained
using a Helios Beta Thermospectronic spectrophotometer.
Infrared spectra were recorded using a Specac single reflectance
ATR instrument (4000–400 cm21, resolution 4 cm21).
Acknowledgements
POB, DSB and KG thank the EPSRC and Royal Society
(United Kingdom) and NRF (South Africa) for financial
support. The authors thank Dr Chengee Jiao from FEI
Company for FIB-TEM sample preparation.
References
1 P. O’Brien and J. McAleese, J. Mater. Chem., 1998, 8, 2309.
2 D. S. Boyle, K. Govender and P. O’Brien, Chem. Commun., 2002, 80.
3 J. C. Johnson, H. Yan, P. Yang and R. J. Saykally, J. Phys. Chem.
B, 2003, 107, 8816.
4 G. Bohnsack, Ber. Bunsen-Ges. Phys. Chem., 1988, 92, 803.
5 P. O’Brien, T. Saeed and J. Knowles, J. Mater. Chem., 1996, 6, 1135.
6 A. Degen and M. Kosec, J. Eur. Ceram. Soc., 2000, 20, 667.
7 W. L. Bragg and J. A. Darbyshire, Trans. Faraday. Soc., 1932, 28, 522.
8 Structure Reports, vol. 27, eds. W. B. Pearson, I. D. Brown and
A. Mc. L. Mathieson, D. Reidel Publishing Co., 1962, p. 475.
9 W. J. Li, E. W. Shi, W. Z. Zhong and Z. W. Yin, J. Cryst. Growth,
1999, 203, 186.
10 D. Li and H. Haneda, Chemosphere, 2003, 51, 129.
11 D. Scarano, G. Spoto, S. Bordiga, A. Zecchina and C. Lamberti,
Surf. Sci., 1992, 276, 281.
12 M. Izaki and T. Ohmi, J. Electrochem. Soc., 1997, 144, L3.
13 M. Izaki and J. Katayama, J. Electrochem. Soc., 2000, 147, 210.
14 K. Ito and K. Nakamura, Thin Solid Films, 1996, 286, 35.
15 L. Vayssieres, K. Keis, S. T. Lindquist and A. Hagfeldt, J. Phys.
Chem., 2001, 105, 3350.
16 N. Saito, H. Haneda, W. S. Seo and K. Koumoto, Langmuir, 2001,
17, 1461.
17 A. P. A. Oliveira, J. F. Hochepied, F. Grillon and M. H. Berger,
Chem. Mater., 2003, 15, 3207.
18 M. L. Fuller, J. Appl. Phys., 1944, 15, 164.
19 J. M. Cowley, A. L. G. Rees and J. A. Spink, Proc. Phys. Soc,
1951, 64, 639.
20 M. Andrés-Vergés, A. Mifsud and C. J. Serna, J. Chem. Soc.,
Faraday Trans., 1990, 86, 959.
21 T. Trindade, J. D. Pedrosa de Jesus and P. O’Brien, J. Mater.
Chem., 1994, 4, 1611.
22 E. Matijevic, Faraday Discuss., 1991, 92, 229.
23 E. Matijevic, Chem. Mater., 1993, 5, 412.
24 F. A. Sigoli, M. R. Davolos and M. Jafelicci, Jr., J. Alloys Compd.,
1997, 262–263, 292.
25 B. G. Wang, E. W. Shi and W. Z. Zhong, Cryst. Res. Technol.,
1998, 33, 937.
26 B. O’Regan and M. Grätzel, Nature, 1991, 353, 737.
27 C. Rost, I. Sieber, K. Ernst, S. Siebentritt, M. C. Lux-Steiner and
R. Könenkamp, Appl. Phys. Lett., 1999, 75, 692.
28 F. F. Lange, Science, 1996, 273, 903.
29 S. Yamabi and H. Imai, J. Mater. Chem., 2002, 12, 3773.
30 A. Chittofrati and E. Matijevic, Colloids Surf., 1990, 48, 65.
31 R. A. McBride, J. M. Kelly and D. E. McCormack, J. Mater.
Chem., 2003, 13, 1196.
32 F. Kazumi and T. Daigaku, Jpn. Pat. JP 57205320, 1982.
33 K. Fujita, K. Murata, T. Nakazawa and I. Kayama, Yogyo
Kyokaishi, 1984, 92, 227.
34 K. Fugita and K. Matsuda, Bull. Chem. Soc. Jpn., 1992, 65, 2270.
35 C. Pacholski, A. Kornowski and H. Weller, Angew. Chem., Int.
Ed., 2002, 41, 1188.
36 K. Govender, D. S. Boyle, P. O’Brien, D. Binks, D. West and
D. Coleman, Adv. Mater., 2002, 14, 1221.
37 S. Yamabi and H. Imai, J. Mater. Chem., 2002, 12, 3773.
38 J. G. Strom, Jr. and H. W. Jun, J. Pharm. Sci., 1980, 69, 1261.
39 R. Lindsay, E. Michelangeli, B. G. Daniels, T. V. Ashworth,
A. J. Limb, G. Thornton, A. Gutiérrez-Sosa, A. Baraldi,
R. Larciprete and S. Lizzit, J. Am. Chem. Soc., 2002, 124, 7117.
40 H. Gerischer and N. Sorg, Electrochim. Acta, 1992, 37, 827.
41 E. A. Meulenkamp, J. Phys. Chem., 1998, 102, 5566.
42 W. K. Burton, N. Cabrera and F. C. Frank, Philos. Trans. R. Soc.
London, 1951, A243, 299.
43 Z. R. Tian, J. A. Voigt, B. McKenzie and M. J. McDermott, J. Am.
Chem. Soc., 2002, 124, 12954.
44 L. E. Greene, M. Law, J. Goldberger, F. Kim, J. C. Johnson,
Y. Zhang, R. J. Saykally and P. Yang, Angew. Chem., Int. Ed.,
2003, 42, 3031.
45 M. Guo, P. Diao, W. Yan-Jie, B. Wang and S. M. Cai, Wuli
Huaxue Xuebao, 2003, 19, 478.
46 C. H. Hung and W. T. Whang, Mater. Chem. Phys., 2003, 82, 705.
47 M. Öner, J. Norwig, W. H. Meyer and G. Wegner, Chem. Mater.,
1998, 10, 460.
48 A. Degen and M. Kosec, J. Eur. Ceram. Soc., 2000, 20, 667.
49 M. Castellano and E. Matijevic, Chem. Mater., 1988, 1, 78.
50 M. Ocana, M. P. Morales and C. J. Serna, J. Colloid Interface Sci.,
1994, 171, 85.
51 A. Gutiérrez-Sosa, T. M. Evans, A. P. Woodhead, R. Lindsay,
C. A. Muryn, G. Thornton, J. Yoshihara, S. C. Parker,
C. T. Campbell and R. J. Oldman, Surf. Sci., 2001, 477, 1.
52 S. Crook, H. Dhariwal and G. Thornton, Surf. Sci., 1997, 382, 19.
53 M. A. Cousins and K. Durose, Thin Solid Films, 2000, 361–362, 253.
54 D. M. Bagnall, Y. F. Chen, Z. Zhu, T. Yao, S. Koyama,
M. Y. Shen and T. Goto, Appl. Phys. Lett., 1997, 70, 2230.
55 P. Zhu, Z. K. Tang, G. K. L. Wong, M. Kawasaki, A. Ohtomo,
H. Koinuma and Y. Segawa, J. Solid State Commun., 1997, 103, 459.
56 H. Shen, M. Wraback, J. Pamulapati, S. Liang, C. Gorla and
Y. Lu, MRS Internet J. Nitride Semicond. Res., 1999, 4S1, G3.60.
57 A. P. Collier, C. J. D. Hetherington and M. J. Hounslow, J. Cryst.
Growth, 2000, 208, 513.
58 J. Zhang, L. Sun, J. Yin, H. Su, C. Liao and C. Yan, Chem.
Mater., 2002, 14, 4172.
59 M. Hilgendorff, L. Spanhel, Ch. Rothenhausler and G. Müller,
J. Electrochem. Soc., 1998, 145, 3632.
J. Mater. Chem., 2004, 14, 2575–2591 2591