0% found this document useful (0 votes)
9 views

Banerjee 2000

Nano physics

Uploaded by

aklibenaziez07
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
9 views

Banerjee 2000

Nano physics

Uploaded by

aklibenaziez07
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

Home Search Collections Journals About Contact us My IOPscience

Effect of the size-induced structural transformation on the band gap in CdS nanoparticles

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2000 J. Phys.: Condens. Matter 12 10647

(http://iopscience.iop.org/0953-8984/12/50/325)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 128.104.1.219
The article was downloaded on 05/08/2013 at 06:04

Please note that terms and conditions apply.


J. Phys.: Condens. Matter 12 (2000) 10647–10654. Printed in the UK PII: S0953-8984(00)16664-0

Effect of the size-induced structural transformation on the


band gap in CdS nanoparticles

R Banerjee, R Jayakrishnan† and P Ayyub


Department of Condensed Matter Physics and Materials Science,
Tata Institute of Fundamental Research, Mumbai 400005, India

Received 24 August 2000, in final form 25 October 2000

Abstract. The interrelation between particle size, crystal structure and optical properties in
semiconductor quantum dots has elicited widespread interest. We report the first attempt at relating
the size-induced transformation from a hexagonal to a cubic structure in CdS nanoparticles to a
change in the band gap. CdS nanoparticles with particle size in the 0.7–10 nm range were prepared
by chemical precipitation using thiophenol as a capping agent. Whereas the band gap for bulk
hexagonal CdS is about 2.5 eV, that for 1 nm cubic CdS nanoparticles was found to be almost
3.9 eV. We also suggest a simple mechanism (based on the periodic insertion of stacking faults) for
the transformation from the cubic zinc blende structure to the hexagonal wurtzite structure.

1. Introduction

Currently, there is great interest in the optical and transport properties of nanometre-sized
semiconductor particles or quantum dots [1]. These show significant departures from bulk
optical and electronic properties when the scale of confinement approaches the excitonic
Bohr radius (rB ), which sets the length scale for optical processes [2]. Quantum dots of
II–VI semiconductors have attracted particular attention, because they are relatively easy to
synthesize in the size range required for quantum confinement. These materials are of potential
interest in non-linear optics and in fast optical switching [3]. Nanocrystalline CdS, CdTe, CdSe,
ZnSe and PbS have been synthesized by a variety of methods including precipitation, sputtering,
electrochemical deposition and inverse micelles. A reduction in the particle size strongly
influences the crystallinity, melting point and structural stability. We need to understand how
these aspects affect optoelectronic properties.
Cadmium sulphide is a wide band gap semiconductor with Eg ≈ 2.5 eV [4]. It is used
in photodetectors and for solar cell applications as n-type window layers in heterojunction
devices [5]. The optical properties of CdS have been extensively studied [6, 7]. Quantum size
effects are quite pronounced because CdS has a rather large rB (≈3 nm) [1]. CdS nanoparticles
are attractive candidates for optoelectronic applications as it is possible to engineer the band
gap over a wide spectral range (visible to UV).
Bulk CdS has a hexagonal wurtzite-type (W) crystal structure with a = 0.4160 nm and
c = 0.6756 nm [8]. Two other structures (see table 1) are observed only in nanocrystalline
CdS [8]: (a) a cubic zinc blende (Z) phase under ambient conditions, and (b) a high-pressure
rocksalt phase [9]. The wurtzite to rocksalt transformation involves not only a change in
symmetry (hexagonal to cubic) but also a change in the nearest-neighbour atomic coordination
† Current address: Department of Materials and Interfaces, Weizmann Institute of Science, Rehovot, Israel.

0953-8984/00/5010647+08$30.00 © 2000 IOP Publishing Ltd 10647


10648 R Banerjee et al
Table 1. Structure data for the three known crystallographic phases of CdS [8].

Crystal Space Lattice


Structure type system group parameters (nm) Atom positions
Wurtzite (2H) Hexagonal P 63 mc a = 0.4160 Cd (0.333, 0.667, 0)
c = 0.6756 S (0.333, 0.667, 0.375)
Zinc blende (3C) Cubic F 43m a = 0.5832 Cd (0, 0, 0)
S (0.25, 0.25, 0.25)
Rocksalt Cubic F m3m a = 0.5420 Cd (0, 0, 0)
S (0.5, 0.5, 0.5)

(from four to six), whereas the wurtzite to zinc blende transformation involves only a change
in symmetry. The two types of transition are affected by particle size in different ways. High-
pressure studies show that as the particle size decreases, the critical pressure at which CdS
and CdSe transform from the wurtzite to the rocksalt structure increases [9]. Again, as the
size of CdS nanoparticles decreases, the equilibrium crystal structure tends to change from the
hexagonal W type to the cubic Z type. The size dependence of the structural stability in CdS
is one of the subjects of the present study.
The cubic (Z) phase has been observed [3] in chemically synthesized 3 nm CdS
nanoparticles and the structure became hexagonal (W) above 300 ◦ C. The authors suggest
that the cubic phase is either: (i) a non-equilibrium (metastable) phase that transforms to the
stable hexagonal phase on annealing, or (ii) an equilibrium phase for CdS nanoparticles. The
latter implies that the cubic to hexagonal transition on heating is a size-driven one caused
by sintering and grain growth. Experimentally, it is difficult to distinguish between the two
models since annealing cubic CdS nanoparticles would result both in coarsening as well as
thermally induced structural relaxation. Also, the authors observed the hexagonal phase only
on heating and not in any of their as-precipitated samples.
In x-ray diffraction (XRD) experiments, it may be difficult to detect the size-driven
structural transition due to the Scherrer broadening [10] in nanoparticles. The other
complicating factor is the considerable overlap between the XRD lines of the hexagonal and
cubic structures. Nanda et al [11] observed the hexagonal phase in 4.5 nm CdS particles, but
could not conclusively identify the structure of the 2.5 nm particles, and suggested a possible
mixture of hexagonal and cubic phases. In contrast, Herron et al [12] have indexed similar-
looking XRD patterns from 1.5–3.5 nm CdS nanoparticles to a single cubic phase with two
broad peaks.
Using high-resolution electron microscopy (HREM), Ricolleau et al [13, 14] studied the
structure of CdS nanoparticles prepared by the inverse micelle technique. They observed the
cubic (Z) structure (a = 0.582 nm) when d < 4 nm, and the hexagonal (W) structure for
d > 6 nm. For intermediate sizes, they observed either the Z-type phase with stacking faults
and twin boundaries which locally gave rise to the W-type phase, or a two-phase mixture
within the same particle with both cubic and hexagonal domains sharing their close-packed
planes at the boundary [14]. Most of the above results indicate that a size induced hexagonal
to cubic transformation occurs in CdS at ≈4 nm. Note that this is also the size regime in
which quantum confinement is expected. However, the possibility of a change in the band
gap accompanying the size-induced structural transition in CdS does not appear to have been
addressed so far. Here we report a detailed study of the mechanism of the structural transition
in CdS nanoparticles as a function of particle size, as well as the influence of the change in
crystal structure on the band gap.
Size-induced structural transformation in CdS nanoparticles 10649

2. Experimental procedure

CdS nanoparticles were synthesized via precipitation from an aqueous solution containing
cadmium chloride, thiourea and ammonia, by a process used for the fabrication of solar cells
comprising of CdTe and CuInSe2 thin films [15]. Measured quantities of CdCl2 and NH4 Cl
were added to an aqueous bath maintained at 65 ◦ C and dissolved by continuous magnetic
stirring. The capping agent, thiophenol was then added to the solution. The pH of the mixture
was raised to 7.5 by adding NH4 OH. Precipitation was achieved by adding a measured quantity
of thiourea and the reaction was allowed to proceed for ∼30 min. The presence of thiophenol
arrests the agglomeration of the CdS particles during precipitation. Further, since thiophenol
forms a coating on the surface of the CdS nanoparticles, it effectively reduces the heterogeneous
nucleation of new particles on the surface of existing ones, thus reducing the precipitation rate.
The precipitate was separated from the aqueous solution by centrifuging at 8000 rpm. The
separated nanoparticles were repeatedly washed in methanol to remove traces of thiophenol
and unreacted chemicals. After drying, some portion of the CdS sample was dispersed in
DMSO (dimethyl sulphoxide) to form a stable suspension for spectroscopic studies. The
dispersion was ultrasonically agitated to break up any CdS agglomerates, which might have
formed during centrifuging and drying. For XRD studies, a part of the sample was dispersed
on a glass slide with collodion. The chemical reaction that results in the precipitation of CdS
nanoparticles can be represented as:

Cd(NH3 )2+
4 + SC(NH2 )2 + 2OH → CdS + H2 CN2 + 4NH3 + 2H2 O (1)
The quantities of the reactants used are as follows: (1) M/50 CdCl2 and M/15 NH4 Cl added
to 100 ml distilled water, (2) different quantities of thiophenol solution (1% v/v in methanol)
were used to obtain particles of different average size, (3) M/7 thiourea was added for the
precipitation of CdS nanoparticles.
X-ray diffraction studies were carried out in a Siemens D500 diffractometer, using Cu Kα
radiation. The mean particle size (coherently diffracting domain size) was estimated from the
FWHM of the diffraction lines using the Scherrer formula [10] after removing the contribution
of the Kα2 component by the Rachinger method [16]. The slit widths used were 1◦ (scatter)
and 0.05◦ at the detector. The scan step size was 0.02◦ and the collection time was 4 s/step.
The optical absorption was recorded in a Shimadzu spectrophotometer in the wavelength range
200–800 nm, measurements being carried out at intervals of 0.5 nm. The path length in the
quartz cells was 1 cm.

3. Results

3.1. Structural transformation


Here we discuss XRD results from four of the nanoparticle CdS samples (CdS-I, CdS-II,
CdS-III and CdS-IV) synthesized using the above technique. The samples differ only in the
amounts of thiophenol used: from 0.5 ml in the case of CdS-I to 1.5 ml in the case of CdS-IV.
Expectedly, larger quantities of capping agent lead to smaller particle sizes.
The XRD pattern from CdS-I (see figure 1) can be consistently indexed on the basis of
the hexagonal, W-type structure, in which the six prominent lines correspond to the reflections
(1010), (0002), (1011), (1120), (1013) and (1122). The weak (1012) peak was also observed.
The estimated x-ray size for this sample was ∼10 nm based on the FWHM of the (1120)
peak. The XRD spectrum from CdS-II apparently exhibits only two broad peaks, centred at
2θ ≈ 27◦ and 2θ ≈ 47◦ , but a more careful study of the pattern suggests that these two peaks
are overlaps of multiple peaks. This is also indicated by the marked asymmetry of the peak
10650 R Banerjee et al

Figure 1. X-ray diffraction spectra from CdS samples with different average particle size (Scherrer
size). The XRD pattern from CdS-I is indexed on the basis of the hexagonal (2H ) wurtzite structure,
while that of CdS-IV has been indexed with respect to the cubic (3C) zinc blende structure.

Table 2. A summary of the crystal structure and optical absorption data for CdS nanoparticles.
Band gaps calculated from absorption edges for the four samples are listed. Note that there are two
band gaps for the CdS-II sample, resulting from a mixture of hexagonal and cubic phases present
in the sample.

Band gap (eV) Size (nm) Size (nm) Crystal


Sample Eg from Eg from x-ray structure
CdS I 2.59 9.0 10 hexagonal
CdS II 2.76 5.2 (h) hexagonal
3.54 4.5 (c) + cubic
CdS III 3.57 2.0 2 cubic
CdS IV 3.85 0.7 0.7 cubic

centered at 27◦ , resulting from the overlap of the (1010), (0002), and (1011) peaks of the
hexagonal (W) structure. The increased overlap in CdS-II is clearly a result of line broadening
due to the smaller particle size in this sample as compared to CdS-I. Note, however that the two
most prominent peaks for cubic CdS with a Z-type structure also occur close to 27◦ (111) and
44◦ (220). Thus, the presence of a cubic phase in CdS-II cannot be ruled out based on XRD
data, but the majority phase in this sample is certainly the hexagonal phase (as indicated by the
asymmetric peak broadening). Estimation of the mean x-ray size is not possible for CdS II due
to the peak overlaps as well as the possible presence of a mixture of cubic and hexagonal phases.
We now consider the XRD pattern from CdS-III. Again there are only two broad peaks
centered at 2θ ≈ 27◦ and 2θ ≈ 47◦ (figure 1), but there are two important differences between
the diffraction patterns of CdS-II and CdS-III. In CdS-III the width of the peak at ≈27◦ is
significantly smaller, and it is much more symmetric. These observations indicate that the 27◦
Size-induced structural transformation in CdS nanoparticles 10651

peak in CdS-III is a single line and not an overlap of multiple lines. It can be identified as the
(111) line (relative intensity, I /I0 = 100%) of the cubic Z phase. The reduction in the width
of this peak despite a decrease in particle size is a definite indicator of the change in crystal
structure from hexagonal to cubic. The second broad peak in CdS-III is an overlap of the cubic
(220) peak at ≈44◦ (I /I0 = 80%) and the cubic (311) peak at ≈52◦ (I /I0 = 60%). The mean
x-ray size for the CdS III sample, calculated from the FWHM of the (111) peak, is ∼2 nm.
The nature of the XRD pattern for CdS IV is similar to that of CdS III, but the width of both
lines is significantly larger. The mean particle size for this sample is ∼0.7 nm. Table 2 lists
the crystal structures and mean particle sizes derived from the x-ray diffraction data, for the
different samples of CdS nanoparticles.

3.2. Optical absorption


The optical absorption data for CdS-I, CdS-II, CdS-III and CdS-IV are shown in figure 2. The
absorption edge corresponding to the band gap in bulk hexagonal W-type CdS is indicated
by an arrow. It is evident from figure 2 that all four samples exhibit absorption edges which
are blue-shifted with respect to the bulk CdS band gap, arising from quantum confinement
effects in the nanoparticles. Note that there are two distinct absorption edges in CdS-II: one
at λ ≈ 450 nm and the other at λ ≈ 350 nm. This is a unique feature of this sample, as all
others exhibit a single absorption edge. The band gaps corresponding to the absorption edges
for all four CdS samples are listed in table 2. The band gap of bulk hexagonal W-type CdS is
≈2.5 eV. The blue-shift in the band gap in nanoparticles due to quantum confinement has the
quantitative form [17]:
h̄2 π 2
Eg ≡ Egnano − Egbulk = (2)
2MR 2
where R is the radius of the particle and M is the effective mass of the system. For hexagonal
CdS, M = 1.919×10−31 kg. Since CdS-I exhibits a hexagonal structure, equation (2) could be
used to estimate the particle size, using the observed blue-shift in the band gap, Eg = 0.09 eV.
The corresponding particle diameter is 9 nm, which is in very good agreement with the x-ray
size (10 nm).
The bulk band gap and the effective mass are unknown for the cubic Z phase of CdS. So,
the optical absorption data for CdS-III and CdS-IV, both of which consist of single phase cubic
CdS, were fitted to a simplified form of equation (2):
A
Eg = 2 (3)
R
where A is a constant which depends on the effective mass (M). The resulting value of the
band gap for the bulk cubic Z phase of CdS is Egcubic = 3.53 eV. This is only a notional value
since the cubic phase may never exist in the bulk.
The situation is a little more complex in CdS II, which is a mixture of the cubic and
hexagonal phases and consequently exhibits two distinct absorption edges. The band gap
corresponding to each absorption edge is given in table 2. It is important to assign the band
gaps correctly to the two phases and check for overall self-consistency. Since the band gap of
the bulk hexagonal phase (∼2.5 eV) is smaller than the band gap for the ‘bulk’ cubic phase
(∼3.5 eV), the smaller of the two band gaps in CdS II was assigned to the hexagonal phase and
the larger to the cubic phase. The corresponding particle diameters, obtained from equation (2),
for the hexagonal and cubic phases are 5.2 nm and 4.5 nm, respectively. Thus, the particle
size distribution in CdS-II is quite small, which is consistent with the fact that the other CdS
samples (I, III and IV) prepared by the same technique, consisted of a single phase, suggesting
a relatively small size distribution.
10652 R Banerjee et al

Figure 2. Optical absorption data for CdS-I, CdS-II, CdS-III and CdS-IV samples. All absorption
edges are blue-shifted with respect to the band gap of bulk, hexagonal CdS. Note that CdS-II shows
two absorption edges corresponding to the cubic and hexagonal phases.

Our data therefore confirm that a size induced hexagonal↔cubic transformation occurs in
CdS nanoparticles at about 4.5 nm, which is in excellent agreement with earlier TEM/HREM
observations [13]. Assigning the smaller band gap to the cubic phase and the larger band gap
to the hexagonal phase is unphysical since it would result in a red-shift of the cubic band gap
at smaller particle sizes. The final possibility is that both band gaps in CdS-II correspond
to the hexagonal phase with a bimodal size distribution. This requires that CdS-II consist of
hexagonal phase particles with sizes of 5.2 nm and 2.6 nm, and implies an unnaturally large
size distribution. Further, the presence of 2.6 nm CdS particles in the hexagonal phase would
contradict the observation of a hexagonal↔cubic transition at ≈4.5 nm. Thus, our assignment
of the smaller band gap in CdS-II to the hexagonal (W) phase and the larger to the cubic (Z)
phase is the only self-consistent one. It is also possible for individual CdS particles to consist
of mixed phases—as observed by Ricolleau et al [14]. This could either arise due to twinning
or faulting within a single cubic particle or due to domains of the cubic and hexagonal phase
coexisting within the same particle, sharing their respective close-packed planes at the domain
boundaries. However, in the present case, the observation of two distinct band gaps (CdS-II)
probably implies that the sample consists of distinct particles with the slightly smaller ones
having a cubic structure and the larger ones having a hexagonal structure.

4. Discussion

The hexagonal wurtzite and the cubic zinc blende phases of CdS are quite closely related
structures and the close-packed planes of these two structures ({0002} in the hexagonal case
and {111} in the cubic case) are identical. Both planes exhibit a two-dimensionally projected
sixfold symmetry and have the same projected in-plane density of atoms. Both structures
consist of periodic stacks of close packed Cd planes with S atoms occupying the fourfold
Size-induced structural transformation in CdS nanoparticles 10653

Figure 3. A schematic demonstration of a mechanism involving the introduction of deformation


stacking faults on alternate close-packed planes of the parent 3C structure, which can transform it
completely to the 2H structure.

coordinated tetrahedral interstitial sites. The difference between the two structures is only in
the stacking sequence of the close-packed planes. The stacking sequence is ABABABA. . .
(2H ) in the W structure and ABCABCABCA. . . (3C) for the Z structure.
Since the 3C and 2H structures are polytypic, it is relatively easy to transform one to the
other by locally changing the stacking sequence by introducing a series of twins or stacking
faults. A single twinned fault introduced into a 3C structure results in three layers adopting the
2H structure. An intrinsic stacking fault results in four layers taking on the 2H structure while
an extrinsic stacking fault results in three layers adopting the 2H structure [18]. Sebastian et al
[19] discuss in detail how different stacking fault configurations can occur in ZnS crystals and
transform the structure from 2H to 3C. They show that it is the deformation fault configuration
that is primarily responsible for the structural transition. This is produced by slipping one part
of the crystal past another, parallel to the close-packed basal planes of the hexagonal unit cell.
The partial slip or displacement vector is 1/3 1010 for a parent 2H structure, and 1/6 112 for
a parent 3C structure. The partial slip vector moves atoms from an A to a B position, from a B to
a C position and from a C to an A position. Deformation stacking faults introduced on alternate
close-packed planes of the 3C structure can transform it completely to the 2H structure. This
mechanism is schematically depicted in figure 3. The ease of introducing such stacking faults
depends on the stacking fault energy (SFE) of the compound. From a study of the dissociation
of dislocations by weak beam TEM, the SFE in CdS was found to be ∼8.7 mJ m−2 [20]. This
is quite low compared to typical semiconductors and metals which have SFEs in the range of
50–150 mJ m−2 . The low SFE in CdS implies that the transformation between the cubic and
hexagonal phases is energetically feasible. Further, the transformation mechanism involves no
distortion in the nearest-neighbour bonds and in the atomic coordination. The mechanism of
the wurtzite to rocksalt transformation, proposed by Tolbert and Alivisatos [21] is much more
complex and justifies the extreme conditions of high pressure required to induce it.
Note that we observed both the hexagonal and cubic phases in differently sized CdS
nanoparticles synthesized under similar conditions. This implies that the formation of cubic
CdS is a size effect and not due to a non-equilibrium synthesis process. Note also that the
finer-sized cubic CdS nanoparticles were obtained by adding larger quantities of surfactant,
which slows down the precipitation kinetics. The coarser CdS particles, precipitated at higher
rates than the finer particles, should—if at all—be expected to show greater non-equilibrium
(metastability) effects. But the bulk equilibrium hexagonal phase actually forms in the coarser
particles and the cubic phase in the finer ones. This negates the first of the two possible
explanations for the formation of cubic CdS (section 1) proposed in [3].
10654 R Banerjee et al

The nucleation and growth of CdS particles from solution is primarily governed by the
competition between the surface and volume free energies of the particle. During nucleation,
the surface energy contribution dominates, so that reductions in the free surface area and
surface energy would be favoured. The free surface area can be minimized by the nucleus
adopting a spherical shape (minimal surface/volume). This can be achieved by adopting a
crystal structure amenable to the formation of a spherelike nucleus by exposing close-packed
low-energy planes as facets on the surface of the nucleus. Symmetry considerations favour
the cubic crystal structure over the hexagonal one for the formation of a spheroidal nucleus
with a large number of {111} facets on the surface. We therefore expect the CdS nanoparticles
to adopt a cubic structure at the initial stage of nucleation and transform to the bulk stable
hexagonal structure during growth.

5. Conclusions

With a reduction in the particle size in CdS nanoparticles, there is a transformation from the
bulk hexagonal wurtzite-type structure to the cubic zinc-blende-type structure. The critical
transformation size is around 4–5 nm. We have been able to distinguish the effect of the size-
induced Scherrer broadening from the structural-transformation-induced changes in the x-ray
diffraction patterns. Our results indicate that the optical properties of nanocrystalline CdS
are governed not only by quantum confinement effects but also by the size-induced structural
phase transition. This is also the first attempt to obtain a value for the band gap of the cubic
zinc blende structure of CdS, based on the optical and x-ray diffraction data. The observation
of two distinct absorption edges in CdS nanoparticles is shown to result from the presence of
coexisting cubic and hexagonal phases. We have also presented a simple mechanism for the
transformation from the zinc blende to the wurtzite structure, based on periodic insertion of
stacking faults.

References

[1] Alivisatos A P 1996 J. Phys. Chem. 100 13 226


[2] Brus L E 1984 Chem. Phys. 80 4403
[3] Bandaranayake R J, Wen G W, Lin J Y, Jiang H K and Sorensen C M 1995 Appl. Phys. Lett. 67 831
[4] Zou B S, Little R B, Wang J P and El-Sayed M A 1999 Int. J. Quantum Chem. 72 439
[5] Danaher W I, Lyons L E and Morris G C 1985 Solar Energy Mater. 12 137
[6] Colvin V L, Schlamp M C and Alivisatos A P 1994 Nature 370 354
[7] Vossmeyer T, Katsikas L, Giersig M, Popovic I G and Weller H 1994 J. Phys. Chem. 98 7665
[8] Villars P and Calvert L D (eds) 1985 Pearson’s Handbook of Crystallographic Data for Intermetallic Phases
vol 2 (Metals Park, OH: ASM International)
[9] Chen C C, Herhold A B, Johnson C S and Alivisatos A P 1997 Science 276 398
[10] Guinier A 1963 X-Ray Diffraction (San Francisco: Freeman)
[11] Nanda J, Kuruvilla B A and Sarma D D 1999 Phys. Rev. B 59 7473
[12] Herron N, Wang Y and Eckert H 1990 J. Am. Chem. Soc. 112 1322
[13] Ricolleau C, Audinet L, Gandais M and Gacoin T 1998 Thin Solid Films 336 213
[14] Ricolleau C, Audinet L, Gandais M, Gacoin T, Boilot J P and Chamarro M 1996 J. Cryst. Growth 159 861
[15] Abd-Lefdil S, Messaoudi C, Abd-Lefdil M and Sayah D 1998 Phys. Status Solidi a 168 417
[16] Rachinger W A 1948 J. Sci. Instrum. 25 254
[17] Kayanuma Y 1988 Phys. Rev. B 38 9797
Mandal S K, Chaudhuri S and Pal A K 1999 Thin Solid Films 350 209
[18] Hirth J P and Lothe J 1968 Theory of Dislocations (New York: McGraw-Hill) p 288
[19] Sebastian M T, Pandey D and Krishna P 1982 Phys. Status Solidi a 71 633
[20] Takeuchi S and Suzuki K 1999 Phys. Status Solidi a 171 99
[21] Tolbert S H and Alivisatos A P 1995 J. Chem. Phys. 102 4642

You might also like