Synthesis and shape control of CuInS2 nanoparticles

Article abstract of DOI:10.1021/ja103828f

Cu₂S-CuInS₂ hybrid nanostructures as well as pure CuInS₂ (CIS) nanocrystals were synthesized by methods of colloidal chemistry. The structure, the shape and the composition of these nanomaterials were investigated with transmission electron microscopy (TEM), powder X-ray diffraction (XRD) and energy dispersive X-ray analysis (EDX). By changing the reaction conditions, CuInS₂ nanorods with different aspect ratio, dimeric nanorods as well as hexagonal discs and P-shaped particles could be synthesized. Under our reaction conditions, CIS nanoparticles crystallize in the hexagonal wurtzite structure, as confirmed by Rietveld analysis of the X-ray diffraction patterns. The formation of Cu₂S-CuInS₂ hybrid nanostructures turned out to be an essential intermediate step in the growth of CIS nanoparticles, the copper sulphide part of the hybrid material playing an important role in the shape control of the CIS nanocrystals. By a treatment of Cu₂S-CuInS₂ with 1,10-phenanthroline, Cu₂S parts of the hybrid nanostructures could be removed, and pure CIS nanoparticles with shapes not accessible with other methods can be obtained. Our synthetic procedure turned out to be suitable to synthesize also other compounds, like CuInS₂-ZnS alloys, and to modify, in this way, the optical properties of the nanocrystals.

Full text of DOI:10.1021/ja103828f

Published on Web 10/19/2010
Synthesis and Shape Control of CuInS Nanoparticles
2
Marta Kruszynska, Holger Borchert, J u¨ rgen Parisi, and Joanna Kolny-Olesiak*
UniVersity of Oldenburg, Department of Physics, Energy and Semiconductor Research
Laboratory, Carl-Von-Ossietzky-Str. 9-11, 26129 Oldenburg, Germany
Received May 5, 2010; E-mail: joanna.kolny@uni-oldenburg.de
2 2 2
Abstract: Cu S-CuInS hybrid nanostructures as well as pure CuInS (CIS) nanocrystals were synthesized
by methods of colloidal chemistry. The structure, the shape and the composition of these nanomaterials
were investigated with transmission electron microscopy (TEM), powder X-ray diffraction (XRD) and energy
dispersive X-ray analysis (EDX). By changing the reaction conditions, CuInS nanorods with different aspect
2
ratio, dimeric nanorods as well as hexagonal discs and P-shaped particles could be synthesized. Under
our reaction conditions, CIS nanoparticles crystallize in the hexagonal wurtzite structure, as confirmed by
Rietveld analysis of the X-ray diffraction patterns. The formation of Cu
2
2
S-CuInS hybrid nanostructures
turned out to be an essential intermediate step in the growth of CIS nanoparticles, the copper sulphide
part of the hybrid material playing an important role in the shape control of the CIS nanocrystals. By a
treatment of Cu
2
S-CuInS
2 2
with 1,10-phenanthroline, Cu S parts of the hybrid nanostructures could be
removed, and pure CIS nanoparticles with shapes not accessible with other methods can be obtained.
Our synthetic procedure turned out to be suitable to synthesize also other compounds, like CuInS
alloys, and to modify, in this way, the optical properties of the nanocrystals.
2
-ZnS
1
3
1
. Introduction
material and an exciton radius of 4 nm. It has high radiation
5
-1
stability and an extinction coefficient of ∼10 cm (at 500
Semiconductor nanoparticles attract scientific attention due
14
15
nm), a value about 10× higher than that for CdTe. Therefore,
1
,2
to their size and shape dependent properties. Because of their
tunable absorption and emission in the visible and IR range of
the solar spectrum, semiconductor nanocrystals are considered
as an interesting absorber material for solar cells,
used in light emitting diodes or as fluorescence marker for
CIS is already widely used in thin film solar cells conventionally
16-18
prepared by sputtering or evaporation techniques.
In contrast, there are only few reports about chemical synthesis
of colloidal CIS nanoparticles with a narrow size distribution and
successful size or shape control. CIS nanocrystals were synthe-
sized by the reaction of copper and indium salts (chlorides,
3
-6
they are
7
19
8
-10
20-22
biomolecules.
However, most of the best studied materials
2
3
13,19,23
22
until now, like cadmium and lead chalcogenides, have a
drawback of containing highly toxic components that severely
restricts their possible applications.
iodides, and carboxylates
) with various sulfur sources
24
13,23
25
(
dithiocarbamates, thiourea, n-alkylthiols,
dithiols, el-
1
1,12
19,26
27
A possible alternative
emental sulfur,
and carbon disulfide ) as well as by thermal
material, being less toxic, but having a tunable absorption in
the visible range, is copper indium disulfide (CIS), being a direct
bandgap semiconductor with a bandgap of ∼1.5 eV in the bulk
(
(
13) Zhong, H.; Zhou, Y.; Ye, M.; He, Y.; Ye, J.; He, C.; Yang, C.; Li, Y.
Chem. Mater. 2008, 20, 6434.
14) Krunks, M.; Bijakina, O.; Varema, T.; Mikli, V.; Mellikov, E. Thin
Solid Films 1999, 338, 125.
(
(
1) Alivisatos, A. P. J. Phys. Chem. 1996, 100, 13226.
(15) Mitchell, K.; Fahrenbruch, A. L.; Bube, R. H. J. Appl. Phys. 1977,
48, 829.
2) Murray, C. B.; Kagan, C. R.; Bawendi, M. G. Annu. ReV. Mater. Sci.
2
000, 30, 545.
(16) Chopra, K. L.; Paulson, P. D.; Dutta, V. Prog. PhotoVolt.: Res. Appl.
2004, 12, 69.
(
(
(
3) Huynh, W. U.; Dittmer, J. J.; Alivisatos, A. P. Science 2002, 295,
2
425.
(17) Green, M. A. J. Mater. Sci. Mater. Electron. 2007, 18, S15.
(18) Powalla, M.; Bonnet, D. AdV. OptoElectron. 2007, 2007, . article ID
97454.
4) Arici, E.; Meissner, D.; Sch a¨ ffler, F.; Sariciftci, N. S. Int. J.
Photoenergy 2003, 5, 199.
5) Gur, I.; Fromer, N. A.; Geier, M. L.; Alivisatos, A. P. Science 2005,
(19) Xie, R.; Rutherford, M.; Peng, X. J. Am. Chem. Soc. 2009, 131, 5691.
(20) Qi, Y.; Liu, Q.; Tang, K.; Liang, Z.; Ren, Z.; Liu, X. J. Phys. Chem.
C 2009, 113, 3939.
3
10, 462.
(
(
6) Kumar, S.; Scholes, G. D. Microchim. Acta 2008, 160, 315.
7) Rogach, A. L.; Gaponik, N.; Lupton, J. M.; Bertoni, C.; Gallardo,
D. E.; Dunn, S.; Pira, N. L.; Paderi, M.; Repetto, P.; Romanov, S. G.;
O’Dwyer, C.; Torres, C. M. S.; Eychm u¨ ller, A. Angew. Chem. Int.
Ed. 2008, 47, 6538.
(21) Courtel, F. M.; Paynter, R. W.; Marsan, B.; Morin, M. Chem. Mater.
2009, 21, 3752.
(22) Koo, B.; Patel, R. N.; Korgel, B. A. Chem. Mater. 2009, 21, 1962.
(23) Li, L.; Daou, T. J.; Texier, I.; Kim Chi, T. T.; Liem, N. Q.; Reiss, P.
Chem. Mater. 2009, 21, 2422.
(
(
8) Niemeyer, C. M. Angew. Chem., Int. Ed. 2001, 40, 4128.
9) Chan, W. C. W.; Nie, S. Science 1998, 281, 2016.
(24) Pan, D.; An, L.; Sun, Z.; Hou, W.; Yang, Y.; Yang, Z.; Lu, Y. J. Am.
Chem. Soc. 2008, 130, 5620.
(
(
10) Bruchez, M. J.; Moronne, M.; Gin, P.; Weiss, S.; Alivisatos, A. P.
Science 1998, 281, 2013.
(25) Norako, M. E.; Franzman, M. A.; Brutchey, R. L. Chem. Mater. 2009,
21, 4299.
11) In Europe, the use of Cd and Pb is banned in electrical components
by the “Directive on the restriction of the use of certain hazardous
substances in electrical and electronic equipment 2002 95/EC”.
12) Peng, X. Nano Res. 2009, 2, 425.
(26) Panthani, M. G.; Akhavan, V.; Goodfellow, B.; Schmidtke, J. P.; Dunn,
L.; Dodabalapur, A.; Barbara, P. F.; Korgel, B. A. J. Am. Chem. Soc.
2008, 130, 16770.
(
1
5976 9 J. AM. CHEM. SOC. 2010, 132, 15976–15986
10.1021/ja103828f  2010 American Chemical Society

Synthesis and Shape Control of CuInS
2
Nanoparticles
A R T I C L E S
2
8
29
decomposition or irradiation of single molecule precursors.
However, these synthetic procedures involving Cu, In, and S do
not always result in the formation of monophasic CIS particles,
2.2. Synthesis of CuInS
CuAc (1 mmol), InAc
mixed with 10 mL of oleylamine in a three-neck flask and stirred under
vacuum at room temperature for 30 min. Next, the reaction was heated
to 240 °C under nitrogen flow. When the solution color changed from
turbid green to slightly yellow, the mixture of 0.25 mL 1-DDT and
2
Nanocrystals. For a typical synthesis,
(1 mmol) and TOPO (90%) (3.5 mmol) were
3
but also hybrid materials, e.g., composed of Cu
2
S-In
There is also one report about CIS nanorods growing
S nanodiscs, which can be converted to CIS in the end of
2 3
S can
30,31
occur.
on Cu
2
1
.75 mL t-DDT was rapidly injected, which resulted in an immediate
32
the reaction. To summarize, development of synthetic procedures
yielding pure CIS nanoparticles with well controllable size and
shape still remains a challenge.
color change of the reaction solution to brown. Small aliquots were
taken at different time intervals between 30 s and 1 h. Finally, the
reaction system was cooled to room temperature, and the particles were
precipitated and washed with ethanol to remove residual thiols, acetates
and TOPO. The purified precipitate was then redissolved in hexane.
It is anticipated in this place that CIS nanoparticles prepared by this
Concerning the crystal structure, it is noteworthy that bulk
CIS at room temperature has the chalcopyrite crystal structure,
whereas the zinc blend and wurtzite modifications are stable
only at high temperatures. In contrast, nanoparticles stable at
room temperature can be synthesized in all three crystal
method can contain an attached Cu
reaction conditions and growth time.
.3. Synthesis of CuInS -ZnS Nanocrystals. The first step of
the CuInS -ZnS nanocrystal synthesis was performed similarly to
2
S particle, depending on the exact
2
2
3
3,34
structures.
In the zinc blend and wurtzite structure, the
2
indium and copper atoms are randomly distributed over the
the reaction for CIS, i.e., indium and copper acetates and TOPO
were dissolved in oleylamine and heated to 240 °C. Next, the
mixture of thiols and, subsequently, zinc acetate (2 mmol) in 2
mL of OLAM were injected, and aliquots were taken at different
times. Afterward, a purification procedure with ethanol was used,
as described above for the CIS nanocrystals.
2
0
cation sites of the lattice, which allows a flexibility of
stoichiometry and, thus, tuning the Fermi energy over a wide
range. The latter feature makes CIS nanoparticles particularly
interesting for device fabrication.
Herein, we report a colloidal synthesis of wurtzite CIS
nanocrystals as well as Cu S-CIS nanohybrids, using simple
and commercially available chemicals as starting materials. Even
though our synthesis is a one pot reaction, the growth mecha-
nism of CIS nanocrystals involves several steps, their progres-
sion being determined by the starting conditions of the reaction.
As will be shown, the reaction starts by the formation of copper
sulphide nanoparticles, which serve as starting points for the
subsequent growth of CIS nanocrystals and play a major role
2
.4. Transformation of Biphasic Cu
2
S-CuInS
2
to Monopha-
2
sic CuInS Nanocrystals. In order to eliminate the Cu
2
2
S structural
constituent, biphasic nanocrystals were precipitated with methanol,
redispersed in a 0.4 M 1,10-phenanthroline solution in ethanol, and
stirred at room temperature for 17 or 24 h. The nanoparticles were
separated from the characteristically red colored solution, proving
the formation of copper-phenanthroline complex, by centrifugation,
and subsequently redissolved in hexane.
2.5. Characterization Methods.
in the shape control of the resulting CIS particles. Cu
disappears in the course of the reaction, so that monophasic
CIS nanoparticles are the final product. By stopping the reaction
2
S
2.5.1. UV-vis Absorption Spectroscopy. Absorption spectra
were measured on a Varian Cary 100 Scan spectrophotometer.
2.5.2. Transmission Electron Microscopy (TEM). Nanocrystal
size and morphology were studied with a Zeiss EM 902A
transmission electron microscope with an acceleration voltage of
0 kV. High resolution transmission microscopy observations as
at an intermediate stage, Cu
2
2
S-CuInS hybrid nanostructures
can be obtained. Upon changing the reaction conditions (ratio
between the precursors and reaction time), a variety of different
8
well as energy-filtered TEM measurements were performed on a
FEI Titan 80/300 kV microscope. Nanocrystals in hexane solution
were deposited onto the carbon-coated copper or nickel grids by
drop casting technique.
2.5.3. Energy dispersive X-ray Analysis. The integral stoichi-
ometry was obtained by the EDAX detector integrated into a FEI
Quanta 200 3D scanning electron microscope. Stoichiometry of
individual nanoparticles was analyzed with the EDX detector of
the FEI Titan 80/300 kV microscope.
shapes of CIS nanoparticles and Cu
2
2
S-CuInS nanohybrid
materials can be synthesized. In the present work, we describe
their formation mechanism and structure, based on TEM, XRD,
and EDX measurements.
2
. Experimental Section
.1. Materials. Copper(I) acetate (CuAc, 97%), zinc acetate
2
(
ZnAc 99.99%, metal basis), technical grade trioctylphosphine
2
2.5.4. Powder X-Ray Diffraction (XRD). Powder X-ray dif-
oxide (TOPO, 90%), 1-dodecanethiol (1-DDT, 98+%), tert-
dodecanethiol (t-DDT, 98.5%, mixture of isomers) and 1,10-
phenanthroline (g99%) were purchased from Aldrich. Indium(III)
fraction (XRD) was measured with a PANalytical X’Pert PRO MPD
diffractometer operating with Cu K radiation, Bragg-Brentano
R
θ-2θ geometry, a goniometer radius of 240 mm and variable slits.
acetate (InAc
3
, 99.99%, metal basis) and trioctylphosphine oxide
TOPO, 98%) were delivered from Alfa Aesar and oleylamine
OLAM, ∼C18-content 80-90%) from Acros Organics. All
The samples were measured on low background silicon sample
holders and prepared for the XRD measurements as follows:
Nanocrystals in hexane solution were dropped on the silicon
substrate and heated to 70 °C for 30 min, in order to remove the
solvent.
(
(
chemicals were used directly without any further purification.
The XRD data was analyzed using the X’Pert HighScore Plus
software in conjunction with the ICDD database, version 2.0902.
(
(
27) Du, W.; Qian, X.; Yin, J.; Gong, Q. Chem.sEur. J. 2007, 13, 8840.
28) Castro, S. L.; Bailey, S. G.; Raffaelle, R. P.; Banger, K. K.; Hepp,
A. F. J. Phys. Chem. B 2004, 108, 12429.
3
5
For Rietveld refinement, the program MAUD, version 2.14, was
used. In all cases, a polynomial background was used. The
parameters for instrumental line broadening were not refined, but
taken from the default instrument in the MAUD software. This is
justified, because the line broadening is dominated by size and strain
effects in the case of nanocrystalline samples. Texture effects were
(
29) Nairn, J. J.; Shapiro, P. J.; Twamley, B.; Pounds, T.; Wandruszka,
R. V.; Fletcher, T. R.; Williams, M.; Wang, C.; Norton, M. G. Nano
Lett. 2006, 6, 1218.
(
30) Han, W.; Yi, L.; Zhao, N.; Tang, A.; Gao, M.; Tang, Z. J. Am. Chem.
Soc. 2008, 130, 13152.
(
(
31) Choi, S.; Kim, E.; Hyeon, T. J. Am. Chem. Soc. 2006, 128, 2520.
32) Connor, S. T.; Hsu, C.; Weil, B. D.; Aloni, S.; Cui, Y. J. Am. Chem.
Soc. 2009, 131, 4962.
3
6
taken into account by the harmonic texture model implemented
into the MAUD program, using the parameter Lmax ) 4 which
(
(
33) Binsma, J. J. M.; Giling, L. J.; Bloem, J J. Cryst. Growth 1980, 50,
4
29.
34) Nose, K.; Soma, Y.; Omata, T.; Otsuka-Yao-Matsuo, S. Chem. Mater.
009, 21, 2607.
(35) Lutterotti, L.; Chateigner, D.; Ferrari, S.; Ricote, J. Thin Solid Films
2004, 450, 34–41.
2
J. AM. CHEM. SOC. 9 VOL. 132, NO. 45, 2010 15977

A R T I C L E S
Kruszynska et al.
However, to provide a good colloidal stability of the particles
and to obtain anisotropic shapes, TOPO has to be present in
the reaction mixture. TOPO and phosponic acids which are
3
8,39
present as impurities in commercially available TOPO
are
3+
hard Lewis bases and react preferentially with In , a hard Lewis
acid. The presence of TOPO in the reaction mixture decreases
and regulates the reactivity of the indium precursor.
We chose tert-dodecanethiol as a sulfur source, since thiols
1
3,23
are known to decompose at high temperatures.
Tert-
dodecanethiol has the advantage of being liquid, soluble in
organic solvents and to decompose at relatively low tempera-
tures, compared to n-alkyl thiols. The presence of 1-dode-
canethiol in the reaction mixture turned out to be necessary to
control the reactivity of the copper monomers. Thiols are soft
+
bases and react preferentially with soft acids like Cu . Tert-
dodecanothiol reacts with copper acetate to form copper sulphide
already at about 60 °C. In the presence of 1-dodecanothiol, this
reaction can be suppressed, and the formation of CIS nanopar-
ticles takes place. The synthesis of CIS nanoparticles can be
conducted by mixing the educts at room temperature and
subsequently heating them to 240 °C. However, in order to
improve the size distribution of the samples, we decided to start
the reaction at high temperature (240 °C) by the injection of
the thiols. In this way, the nucleation of the nanoparticles can
be controlled in a precise way.
In Figure 1a, a TEM image of a typical sample obtained with
our synthetic procedure is shown. The nanoparticles are uniform
in size and shape, having a mean width of 19.1 ( 1.4 nm and
a length of 44.8 ( 3.8 nm (Figure 1e). These values obtained
from TEM images (assuming an elliptic shape of the nanopar-
ticles for the computer-assisted image analysis) are in good
agreement with the values calculated from the powder X-ray
diffraction patterns (see below). The width of the nanoparticles
is not uniform. They consist of two cones with different angles,
one being truncated and their footings meeting at about the half
of the particle. This general structure could be found in most
of the samples obtained at different synthetic conditions and is
a direct consequence of the formation mechanism of the CIS
particles, as will be discussed later.
Figure 1. An overview TEM image of a typical CIS sample (a), SAED
pattern (b), HRTEM images of single CIS nanoparticles (c) and (d), and a
histogram of the width and the length of the nanocrystals (e).
corresponds to 2 variables for refinement. Furthermore, the aniso-
High resolution TEM images reveal that the particles are
single crystalline and the spacing of the lattice planes corre-
sponds to the wurtzite CIS structure. In the electron diffraction
image distances corresponding to (100), (002), (101), (102),
37
tropic size-strain model developed by Popa was used to take care
of the anisotropic crystallite shape, again using Lmax ) 4. This
corresponds to 3 variables for the anisotropic crystallite shape and
1
free variable for isotropic strain.
(
110), (103), and (112) could be identified (Figure 1b). An
analysis of the high resolution TEM (HRTEM) images of single
particles shows that the nanocrystals grow in the direction of
the c axis. In Figure 1c,d, examples of particles with (002) lattice
planes being perpendicular to the growth direction of the
nanocrystals are shown.
CIS is known to differ often from the exact 1:1:2 stoichi-
ometry. By the help of EDX measurements, we could confirm
the composition of our nanoparticles being close to 1:1:2,
however, with a small In excess (Cu:In ∼4:5). The EDX results
show also a small excess of sulfur, the atomic ratio S/(Cu+In)
being ∼1.1. This could be due to a sulfur rich surface, which is
a plausible assumption when using thiols as stabilizer. Even if
the particles do not have the perfect 1:1:2 stoichiometry, we
3
. Results and Discussion
3.1. Colloidal Synthesis of Uniform CuInS
2
Nanocrystals. In
this section, we present a synthesis route for pure CuInS
2
nanocrystals and give a detailed analysis of a typical sample.
Copper indium disulfide nanocrystals were obtained by fast
injection of a mixture of 1-dodecanethiol and tert-dodecanethiol
to a solution of copper and indium acetates in oleylamine and
TOPO. The reaction was carried out at 240 °C and yielded
elongated wurtzite CIS nanoparticles with a relatively narrow
size distribution (Figure 1). The choice of oleylamine as solvent
turned out to be essential to obtain a narrow size and shape
distribution. A control experiment with 1-octadecene as solvent
led to the formation of much smaller particles with a relatively
broad size distribution and a higher aggregation tendency. So,
OLAM not only plays the role of a solvent, but also regulates
the activity of the monomers and stabilizes the particles.
2
will refer to them as CuInS for the sake of simplicity. In order
to exclude the possibility that our nanorods are nanocomposites
of copper and indium sulphide, similar to particles described
(
37) Popa, N. C. J. Appl. Crystallogr. 1998, 31, 176.
(
36) Popa, N. C. J. Appl. Crystallogr. 1992, 25, 611.
5978 J. AM. CHEM. SOC. 9 VOL. 132, NO. 45, 2010
(38) Wang, F.; Tang, R.; Buhro, W. E. Nano Lett. 2008, 8, 3521.
1

Synthesis and Shape Control of CuInS
2
Nanoparticles
A R T I C L E S
Figure 2. STEM image of a single CIS nanoparticle (a) together with its
copper (b) and indium (c) map obtained by spatially resolved EDX
measurements. The distribution of sulfur (d) and copper (e) in the sample
was measured with energy filtered STEM.
in refs 30 and 31, rather than monophasic CIS nanoparticles,
we investigated also the composition of single particles by
energy filtered scanning TEM (STEM) and EDX. Figure 2a
shows a single particle. In Figure 2b,c, spatially resolved EDX
measurements of this particle can be seen, revealing the uniform
distribution of copper and indium within the particles. Energy
filtered dark field images in Figure 2d,e show that also sulfur
and copper are uniformly distributed. So, all three elements can
be found throughout the particles.
Figure 3a shows the powder X-ray diffraction pattern
measured for the sample analyzed above by TEM. The crystal
2
structure was identified to be hexagonal CuInS (see Supporting
Figure 3. (a) Diffraction pattern of the sample characterized by TEM in
Figures 1 and 2. The experimental data (dots) is plotted together with a
Rietveld fit (red line). (b) The crystallite shape resulting from refinement
according to the applied size-strain model. (c) Reconstructed pole figures
resulting from refinement according to the applied texture model. While
the (100) planes have a high probability to be oriented parallel to the sample
holder, the (002) planes are preferentially oriented perpendicular to the
sample holder.
Information for details on the structure). CIS prepared by
methods used in industrial processes, e.g., sputtering techniques,
1
6
typically has a tetragonal or cubic structure. However,
hexagonal structure was already observed for colloidally
2
0,22,24,32
prepared CIS nanocrystals by several other groups.
In
most cases, the hexagonal CIS phase was identified by compar-
ing the experimental diffraction to simulated stick patterns,
because there is no reference data for hexagonal CIS in the
be randomly oriented, but are more likely to lie flat on the
sample holder.
2
2,24,32
standard databases yet.
While the positions of the Bragg
reflections were found to match well, less attention was usually
paid to the relative intensities or line broadening. In some cases,
the match of intensity patterns is relatively poor. In the present
work, Rietveld refinement was used, in order to analyze also
the intensity distribution and anisotropic line broadening effects.
As will be shown in the following, this allows for obtaining
important information on the microstructure of the samples.
As obvious from the TEM images (Figure 1), the nanocrystals
have an elongated shape. If the direction of preferential growth
corresponds to the axis of highest symmetry, i.e., the c-axis in
the case of hexagonal structure, then the crystallite shape can
If the crystallite shape is assumed to be a sphere and
preferential orientation is neglected, then the positions of the
Bragg reflections are well matched, but the intensity distribution
is not (see Figure S1 of the Supporting Information). The fit in
Figure 3a was obtained taking into account the anisotropic
microstructure and preferential orientation. Figure 3b shows the
simulated crystallite shape resulting from refinement. Table 1
(column “fit 1”) sums up the most important parameters and
results. The analysis revealed that the nanocrystals are elongated
along the c-axis, and the values found for the size along the
c-axis (correlated with the broadening of the (002) reflection)
and perpendicular to the c-axis (e.g., correlated with the
broadening of the (100) reflection) are in good agreement with
the TEM results. Figure 3c shows reconstructed pole figures
deduced from the simultaneous texture analysis. The poles of
the (002) lattice planes are found preferentially in the circumfer-
ence of the projection plane, meaning that the elongated
nanocrystals preferentially lie flat on the sample holder. By
3
7
be simulated by a model developed by Popa that uses a
development into a series of symmetrized spherical harmonics.
Additionally, the Popa model describes microstrain, where the
coefficient <εhh> represents the relative changes of the distances
between the lattice planes in a crystallographic direction denoted
3
7
h. Closely related to anisotropic crystallite shapes, one has to
consider texture effects, because elongated crystallites will not
J. AM. CHEM. SOC. 9 VOL. 132, NO. 45, 2010 15979

A R T I C L E S
Kruszynska et al.
Table 1. Parameters and Results from Rietveld Refinement of the Diffraction Pattern of the Sample Shown in Figures 1-3
fit 1
3a
fit 2
S2
fit 3
fit 4
S3
fit 5
S3
fit 6
curve in Figure
S1,S3
not shown
shape
anisotropic
anisotropic
isotropic
isotropic
isotropic
anisotropic
size/strain refinement
texture
size refined before strain simultaneously simultaneously simultaneously simultaneously size refined before strain
yes
yes
no
no
no
yes
composition
occupancy Cu
occupancy In
occupancy S
unit cell par.
a, b [Å]
unit cell par.
c [Å]
atomic pos.
z (Cu,In)
Cu
0.5
0.5
1.0
3.9062
(0.0002
6.4251
(0.0010
0.3711
(0.0014
1
In
1
S
2
Cu
0.5
0.5
1.0
1
In
1
S
2
Cu
1
In
1
S
2
Cu0.5In1.5
0.25
0.75
1.25
3.9062
(fixed)
6.4251
(fixed)
0.3711
(fixed)
21.3
S
2.5
Cu1.5In0.5
0.75
0.25
0.75
3.9062
(fixed)
6.4251
(fixed)
0.3711
(fixed)
21.8
S
1.5
Cu0.81In1.01
0.404
0.505
1.0
3.9062
(0.0002
6.4252
(0.0010
0.3656
(0.0013
54.6
S
2
0.5
0.5
1.0
3.9062
(fixed)
6.4251
(fixed)
0.3711
(fixed)
21.4
3.9062
(0.0002
6.4240
(0.0009
0.3699
(0.0014
107.2
size along c axis [nm] 56.3
size ⊥ c-axis [nm]
21.3
0.16
4.09
5.51
23.09
0.22
4.03
5.35
21.4
0.13
10.4
18.2
21.3
0.14
10.5
18.4
21.8
0.14
10.3
18.1
21.1
0.16
3.94
5.24
strain <εhh>, in %
R
P
, in %
WP, in %
R
consequence, the (100) planes are preferentially oriented parallel
to the sample holder which results in a high density of poles in
the center of the corresponding pole figure.
It should be noted that parameters for size and strain were
not simultaneously varied to obtain the results shown in Figure
Figure 4) which results in a black color of the material. The
absorption onset reveals a bandgap of 1.5 eV. Because of the
small exciton radius in CIS (4 nm), no blue shift due to the
size quantization effect can be expected, since our particles have
sizes in the order of magnitude of tens of nanometers. Their
bandgap can, however, be influenced by doping the particles
with Zn, as will be discussed in the last section of the Results
and Discussion part.
3
. First, the parameters for anisotropic crystallite size were
refined neglecting strain (parameters set to zero). Afterward,
size parameters were fixed, and the strain parameters were
refined in addition. Refining all size and strain parameters
simultaneously slightly improved the fit, but the results were
suspicious. Larger values for the strain led to values for the
crystallite size in c-direction which considerably exceeded the
values found by TEM (see Table 1 (column “fit 2”) and Figure
S2 of the Supporting Information). The size effects seem to
dominate over strain effects, and fitting first the size and, in a
second step, the strain parameters appears to be more reliable
here. Considering both procedures, one can conclude that the
length of the crystallites along the c-axis is at least ∼56 nm
according to XRD. Together with the TEM data which revealed
a mean length of ∼45 nm, this is a good proof that most
nanocrystals of the ensemble are elongated along the c-axis and
consist of one single crystalline domain. Otherwise, the (002)
reflection would be broader.
2 2
3.2. Characterization of the Cu S-CuInS Hybrid Nanocryst-
als. Cu S-CuInS hybrid nanocrystals are an important inter-
2
2
mediate step in the growth of CIS nanoparticles. The formation
of a hybrid material in the beginning of the reaction can be
seen in the TEM images, where each particle is composed of
two parts of different contrast (see Figure 5a). EDX measure-
ments of these structures show a high copper excess (Cu:In ≈
4
:1), compared to the CIS stoichiometry. A further structural
characterization was done by the analysis of the diffraction
pattern of the hybrid nanoparticles. Figure 6a shows the
diffraction pattern of such a composite sample. By Rietveld
analysis, the diffraction pattern can be decomposed into two
contributions, one of them being a hexagonal CIS phase as in
the former section, the other being a hexagonal Cu
2
S phase.
Another remark is necessary with respect to elemental
2
4
The crystal structure of the Cu S phase was taken from a
2
composition. Pan et al. simulated Cu-In-S compounds with
various compositions in the case of cubic crystal structure. The
diffraction pattern turned out to be unaffected as long as the
valence stoichiometry parameter, defined as (2[S]/
2
4
(
[Cu]+3[In]))-1, is zero. Similar investigations were made
in the present study for the case of hexagonal crystal structure.
The intensities turned out to be only slightly dependent on the
Cu/In ratio, while keeping the valence stoichiometry parameter
at zero (see Figure S3 of the Supporting Information). By
consequence, XRD is not suitable to determine the exact
composition of the Cu-In-S compounds. According to the
EDX results, the valence stoichiometry parameter is close to
zero. Therefore, analyzing the XRD data with a stoichiometric
2
CuInS phase is reasonable. Using the stoichiometry derived
by the values of the EDX ratio and setting the occupancy of S
sites to 1 results only in a negligible improvement of the fits
(
see Table 1 (column “fit 6”)).
Turning now to optical properties, all particles synthesized
with our method absorb light in the whole visible range (see
Figure 4. Typical absorption spectra of pure CuInS
CuInS -ZnS nanocrystals.
2
and alloyed
2
1
5980 J. AM. CHEM. SOC. 9 VOL. 132, NO. 45, 2010

Synthesis and Shape Control of CuInS
2
Nanoparticles
A R T I C L E S
Figure 5. Cu
2
2 2
S-CuInS before (a) and CuInS nanoparticles after (b)
treatment with 1,10-phenanthroline. The sample in (a) was obtained after a
reaction time of 1 min.
database entry based on ref 39, and results for the most important
fitting parameters can be found in the Supporting Information
(
Table S2). Size-shape analysis (see also Figure 6b) revealed a
disk-like shape for the Cu
S part with ∼4 nm thickness and
13 nm diameter which is in reasonable agreement with the
2
∼
TEM image shown as inset in Figure 6a. The short axis
corresponds to the c-axis of the hexagonal structure. The lattice
parameters a and b being almost identical for both phases, the
Figure 6. (a) diffraction pattern of a typical Cu
TEM image). The experimental data (dots) is plotted together with a Rietveld
fit (red line) and the deconvolution into two phases, hexagonal CuInS (blue
2
S-CuInS
2
sample (inset:
2
CIS phase can epitaxially grow on top of the Cu S part. Sizes
2
determined by XRD for the CIS part are also in reasonable
agreement with the TEM images. Thus, the XRD analysis
clearly confirms that these particles are composite structures of
line) and hexagonal Cu S (pink line). (b) the crystallite shapes (arbitrarily
2
scaled) of the two phases resulting from refinement according to the applied
size-strain model. (c) reconstructed pole figures resulting from refinement
according to the applied texture model. Texture was constrained to be
identical for both phases.
two materials, CIS and Cu
,10-Phenanthroline reacts selectively with Cu ions, and so
it is possible to remove the copper sulphide part of the hybrid
2
S.
+
1
nanoparticles, since also the hybrid materials forming in the
beginning of the reaction can be converted to pure CIS
nanoparticles (Figure 5b).
3
0
particles. A comparison of TEM images of hybrid nanopar-
ticles before and after phenanthroline treatment (see Figure 5)
helps to identify the disk shaped part of the hybrid material
Furthermore, phenanthroline treatment influences the surface
composition of the resulting CIS particles. The latter results in
an increased aggregation tendency of the particles treated with
phenanthroline, as can be seen in the TEM images (Figure 5b).
Our finding can be explained taking into account the different
chemical properties of copper(I) and indium(III), one being a
soft and the other a hard Lewis acid. Surface copper atoms will
be preferentially stabilized with thiols. The thiols set free after
phenanthroline treatment, being soft Lewis bases, are not suitable
to stabilize an indium rich surface. This decreases the colloidal
stabilization of the sample. Restoring a good colloidal stability
turned out to be possible by an additional surface treatment with
a hard base, for example a carboxylic acid (oleic acid in our
experiment).
(
disappearing after the reaction with phenanthroline) as copper
sulphide, which is in accordance with the results of the analysis
of the diffraction pattern of these particles.
Control experiments with pure CIS nanoparticles treated with
phenanthroline show no change in the shape and the structure
of the material (see Figures S6 and S7 of the Supporting
Information). However, also in these experiments dissolution
of copper ions can be observed. Phenanthroline forms a red
colored complex with copper(I) ions. This red color appears
both in reactions with CIS nanoparticles and Cu S-CIS hybrid
2
materials. However, in the case of pure CIS, conservation of
the particle size and shape suggests that only surface copper
atoms are removed by the reaction with phenanthroline.
Removal of the copper(I) ions by phenanthroline treatment opens
up the possibility to obtain more differently shaped CIS
2
3.3. Formation Mechanism of the CuInS Nanocrystals. We
have studied the formation and growth of CIS nanocrystals by
taking samples at different time from the reaction vessel and
analyzing them with absorption spectroscopy, TEM, XRD, and
EDX. A large number of samples was evaluated for this purpose
(
39) Wang, F.; Tang, R.; Kao, J. L.; Dingman, S. D.; Buhro, W. E. J. Am.
Chem. Soc. 2009, 131, 4983.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 45, 2010 15981

A R T I C L E S
Kruszynska et al.
Scheme 1. Scheme of the Growth Mechanism of CIS
Nanoparticles
images taken at early stages of the reaction (Figure 7f,g and
Figure S8 of the Supporting Information). Further growth of
the CIS part of the hybrid material is accompanied by some
rearrangement of the copper sulphide phase from a spherical
to a disk shape (Figure 7b,c). This stage of the reaction gives
us the possibility to influence the final shape of CIS
nanoparticles.
The further growth of the CIS phase seems to take place only
on the interface between the two materials: When choosing
conditions for relatively fast growth of the CIS phase (e.g., by
using some In excess), the copper sulphide part of the nano-
particles is divided into two parts, situated at two ends of the
CIS phase, and further growth of the CIS phase in two directions
is observed, the diameters of both parts being controlled by the
size of both copper sulphide particles (Figure 7g-i, see in
particular the marked particles in Figure 7g). Apparently, no
perfect rearrangement of the copper sulphide part is possible
under these conditions of fast growth. Most of the particles are
divided into two parts of different sizes, which gives rise to the
formation of dimeric CIS nanorods with two different widths
and lengths (Figure 7h,i).
The results of the above reaction clearly show the role of
copper sulphide in the formation and shape control of CIS
nanoparticles. Both parts of the asymmetric dimers form
simultaneously under otherwise identical conditions (tempera-
ture, monomer supply). The only difference between them is
the size of the respective copper sulphide part of the hybrid
structure (e.g., particles 1 and 2 in Figure 7g). An analysis of
the TEM images demonstrates that smaller copper sulphide
particles lead to the formation of CIS nanorods with smaller
diameter and their conversion to CIS is faster than in the case
of bigger particles, which also influences the length of the CIS
nanorods generated. In slower reactions (Cu:In ratio 1:1), the
copper sulphide part of the hybrid material has enough time to
rearrange and to form only one disk (see Figure 7b,c). This
results in the growth of extremely uniform CIS particles (Figure
7d,e).
Figure 7. TEM images showing different growth stages of CuInS
2
nanocrystals. (a): Cu
Images b-e show particles obtained in a reaction with Cu:In 1:1 (b): 60 s,
c): 75 s, (d): 80 s, and (e): 90 s). In (f) an early stage of the CIS
nanoparticles growth is shown, the arrows depicting examples of particles,
where the growth of CIS inside of Cu S can be seen. Different growth stages
2
S particles forming in the beginning of the reaction.
(
2
of particles obtained under conditions of faster CIS growth (compared to
a-e) with a Cu:In ratio of 0.5:1 are shown in (g)-(i).
out of which only a small selection can be presented in detail
here. The particles forming at the beginning of the reaction
(
aliquots taken 30-60 s after injection of the sulfur source)
could be identified as copper sulphide. The copper-sulfur
system can form a variety of different crystalline structures, and,
according to the XRD measurements, a mixture of different
phases seems to be present (see Supporting Information).
Hexagonal Cu
makes an epitaxial growth of one of the material onto the other
possible. Also, a conversion of Cu S to CIS requires little lattice
distortion and does not have a high energy barrier. At the
S have a
2
S and CIS have the same anion lattices, which
2
reaction temperature (240 °C), the copper ions in Cu
2
On the basis of these observations, we propose that the
reaction has three steps, illustrated in Scheme 1. At the
beginning, copper sulphide nanoparticles form (phase I), in a
second step, these particles serve as starting point for anisotropic
growth of CIS. During this stage of the reaction, the diameter
of the copper sulphide disk increases, while its thickness
decreases, and CIS regions of the nanoparticles grow (phase
relatively high mobility, which can facilitate an exchange with
In ions and a chemical transformation to CIS.
In Figure 7, representatively selected nanoparticles obtained
at different stages of the reaction are shown. In the first step,
spherical copper sulphide particles are formed (Figure 7a). Then
a formation of a second phase can be observed, and
Cu
2
S-CuInS
2
nanoparticles are obtained (Figure 7b). Closer
2
II). Depending on the growth kinetics, the Cu S particles can
inspection suggests that the formation of a CIS nucleus takes
place inside the copper sulphide particles, yet not on their
surface, as can be seen in some particles from the TEM
remain one single part (slow regime) or be splitted into two
parts (fast regime). The growth of CIS goes on also in the third
step of the reaction (phase III, see also Figure 7d), when the
1
5982 J. AM. CHEM. SOC. 9 VOL. 132, NO. 45, 2010

Synthesis and Shape Control of CuInS
2
Nanoparticles
A R T I C L E S
copper sulphide parts of the nanohybrids get smaller and, finally,
disappear completely. Thus, as an intermediate stage, we obtain
hybrid nanoparticles composed of Cu S and CIS, the copper
2
sulphide part disappearing later in the course of the reaction.
Use of copper sulphide nanoparticles as a starting point for
the nucleation of another material has already been reported
3
0,31
32
for Cu
the case of Cu
2
S-In
2
S
2
3
hybrid materials
and CIS nanorods. In
S-In , uniform particles with different shapes
2 3
S
as well as core-shell structures could be obtained. The report
on CIS nanorods was the first one, where monophasic CIS
2
particles were synthesized using Cu S seeds. Despite the
similarity in the synthetic method, our approach gives an
additional possibility to control the shape of the nanoparticles
and yields samples with much narrower size distribution. In our
reaction, the diameter of the Cu
of the emerging CIS nanocrystal. The growth of CIS stops when
the complete Cu S part of the nanoparticle has been converted
2
S particle controls the width
2
to CIS and we do not observe any significant changes in the
shape of the particles after their growth has ended. Hence,
controlling both nucleation and growth of the copper sulphide
phase is the key to control the shape of CIS nanocrystals. In
the following, some examples will be given for the possible
shape control by our method. For a better comparison of the
results, all TEM images in the present and next section are
shown at the same magnification.
3
.4. Size and Shape Control. We have studied the formation
of CIS nanocrystals under various reaction conditions, system-
atically changing the ratio between the starting materials (CuAc,
InAc , TOPO, 1-DDT, t-DDT) and their overall concentration.
3
It turned out that the final shape of CIS nanoparticles mainly
depends on three parameters: the amount of TOPO (regulating
the activity of the In monomers), the ratio between 1-DDT and
CuAc (influencing the activity of Cu monomers) and the amount
of t-DDT (mainly responsible for the sulfur supply for the
reaction). Lowering the activity of one of the components will
stop the reaction at an earlier stage, which results in smaller
Figure 8. Different aspect ratio of CIS nanocrystals. Reaction conditions:
(a) Cu:In:TOPO 1:1:4.4, 0.25 mL 1-DDT, 1.75 mL t-DDT; (b) Cu:In:TOPO
1
:1:10, 0.25 mL 1-DDT, 1.75 mL t-DDT; (c) and (d) Cu:In:TOPO 1:1:3,
0.75 mL 1-DDT, 5.25 mL t-DDT, image (d) showing an intermediate stage
of the reaction; (e) and (f) Cu:In:TOPO 1:1:3.5, 0.5 mL 1-DDT, 1.5 mL
t-DDT, particles in image (f) were found in an aliquot taken in an early
stage of the reaction conducted with double overall concentration compared
to (e).
(
shorter) particles. The overall concentration of the reaction
mixture turned out to influence the shape of the particles to a
much smaller extent than the ratio between the educts.
Figure 8 shows different modifications of the elongated shape
obtained by changing the reaction conditions. The amount of
TOPO (90%) turned out to have some influence on the length
of the particles. Longer particles could be obtained in reactions
with a smaller amount of TOPO (Figure 8a), while shorter ones
grow when the amount of TOPO has increased (Figure 8b).
This can be due to the reduced activity of In in TOPO rich
reactions and the depletion of monomers through the growth
of nanoparticles. An influence of the amount of TOPO on the
nucleation step might also be a reason for obtaining longer
particles with less TOPO. However, this explanation seems less
likely, because absorption spectra taken at an early stage of
growth did not provide evidence for an influence of the TOPO
amount on the concentration of seeds.
Also, the variation of the amount of both thiols turned out to
have a strong influence on the shape of the elongated particles.
Increasing the amount of the sulfur source leads to the formation
of fish-shaped particles of up to 120 nm length and a maximum
width of 40 nm (Figure 8c). When stopping the reaction at an
2
earlier growth stage, bottle shaped Cu S-CIS hybrid particles
can be obtained (Figure 8d). An increased amount of sulfur in
the reaction mixture allows for growing larger particles. One
possible reason will be simply the availability of more material
for the growth of CIS. However, in all other reactions, the sulfur
source is also used in a more than 10-fold excess. So a more
probable reason for the influence of the sulfur amount on the
2
growth of CIS nanoparticles will be its impact on the Cu S part
Another possible explanation is the enhanced surface stabi-
lization by TOPO and the impurities present in technical grade
of the hybrid nanoparticles during the growth process. The
availability of more sulfur enhances the duration of the second
phase of the growth (see Scheme 1) and, probably, enables
3
8,39
TOPO,
which can also strongly bind to the surface of CIS
nanocrystals. Comparison of this shape to the wurtzite crystal
structure reveals that the tips can be terminated by cation rich
2
growth of the Cu S phase of the hybrid material, in parallel to
the growth of CIS, or, at least, delays it conversion to CIS. The
size distribution of these samples however, is broader than in
other cases, and the width of the particles much less uniform.
Samples uniform in width can be obtained by increasing the
amount of 1-DDT. Two examples of particles synthesized in
this way are shown in Figure 8e,f. Pencil-shaped particles in
(
(
1-11) facets. The angle between the [001] direction and
1-11) facets is 27.8°, and indeed, the tips of the nanocrystals
in Figure 8 b form a ∼30° angle with the c-axis. Addition of
TOPO can increase the stabilization of the In sites of these facets
and slow down further growth.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 45, 2010 15983

A R T I C L E S
Kruszynska et al.
Figure 9. Hexagonal CIS platelets obtained (a) by increasing the amount
of TOPO (90%): Cu:In:TOPO 1:1:50, and by addition of phosphonic acids:
(
b) methylphosphonic acid and (d) octylphosphonic acid. (c) Particles
synthesized by using chlorides instead of acetates, under otherwise the same
reaction conditions as described in the Experimental Section.
Figure 8e are pure CIS crystals, whereas Figure 8f shows an
intermediate step (of another reaction) with Cu S still present.
2
The shape of the tips of the pencil-shaped particles again gives
rise to the assumption that they are terminated by cation rich
(
1-11) facets (see the discussion above to Figure 8b). Probably
increased stabilization of Cu sites of these facets by the help of
-DDT gives rise to the formation of such a shape. The higher
concentration of 1-DDT decreases the activity of Cu monomers
and also influences the stabilization of the Cu S parts of the
hybrid nanostructures. So, the diameter of the Cu S nanodiscs
Figure 10. (a) Diffraction pattern of a typical sample similar to those in
Figure 9. The experimental data (dots) is plotted together with a Rietveld
fit (red line). (b) The crystallite shape resulting from refinement according
to the applied size-strain model. (c) Reconstructed pole figures resulting
from refinement according to the applied texture model.
1
2
2
does not change during growth of the CIS particles, which
results in their uniform width. It should also be noted, however,
that increasing the amount of 1-DDT has some influence on
the nucleation step of the reaction and larger amounts of 1-DDT
broaden the size and shape distribution of the particles by
slowing down the nucleation process.
Figure 10c) which appears reasonable for this shape. Table S1
in the Supporting Information reports more details on the fitting
parameters.
We repeated the above reaction with higher purity TOPO
(98%) and again obtained particles being elongated along the
c-axis (shapes similar to particles shown in Figure 1). So, not
TOPO, but the impurities in the technical grade TOPO must
have been responsible for the shape control in the first reaction.
Such findings motivated us to repeat reactions with 98% TOPO
and an addition of phosphonic acids (methylphosponic acid,
octylphosphonic acid, decylphosphonic acid, and tetradecylphos-
The samples described until now had elongated shapes with
a preferential growth direction along the c axis. Increasing the
amount of TOPO (90%) by about 10× leads to a structural
change from elongated to flattened, hexagonally shaped, faceted
particles. In Figure 9a, a typical overview TEM image of such
a sample is shown. Obviously, two different orientations of the
particles can be seen, with the c-axis oriented parallel or
perpendicular to the TEM grid, the rectangular shapes resulting
from the orientation of the c-axis parallel to the substrate. An
analysis of about 200 rectangular particles revealed that these
particles have a thickness (dimension parallel to the c-axis) of
phonic acid), being some of the impurities that can be found in
38,39
TOPO.
Indeed via adding of phosophonic acids, again
hexagonal disk shaped particles could be obtained. Figure 9b,d
shows two examples of shape control, obtained with meth-
ylphosphonic acid and octylphosphonic acid, respectively. Also
using chlorides instead of acetates as starting material for the
synthesis turned out to be a possibility to arrive at hexagonally
shaped particles (Figure 9c). Chloride and phosphonic acids are
strong ligands, which lower the activity of the monomers and
the chemical potential of the reaction mixture. This can explain
why thermodynamically more stable faceted crystals are formed
under these conditions.
2
0 ( 3 nm and a diameter of 26 ( 7 nm. Figure 10a gives the
XRD pattern of such a sample. Rietveld analysis also confirms
that these particles only contain pure CuInS . Size-shape analysis
2
confirms a crystallite shape of platelets with the c-axis being
the short axis (see Figure 10b). The texture analysis which has
been simultaneously performed revealed that the c-axis is
preferentially orientated perpendicular to the substrate (see
Interestingly, the conditions can so drastically change during
one reaction that particles having started to grow along the c-axis
1
5984 J. AM. CHEM. SOC. 9 VOL. 132, NO. 45, 2010

Synthesis and Shape Control of CuInS
2
Nanoparticles
A R T I C L E S
2
Figure 11. P-shaped nanoparticles of the Cu S-CIS hybrid material. (a)
Cu:In:TOPO 3:1:10, 70 s growth time, (b) Cu:In:TOPO 2:1:10, 4 min growth
time, (c) Cu:In:TOPO 1.5:1:10, 60 s growth time, and (d) Cu:In:TOPO
1
:1:50, 60 min growth time.
change the direction of their growth and continue growing
perpendicular to the c-axis. Some examples of such P-shaped
particles and the possibilities to control their exact shape are
shown in Figure 11. The experimental conditions for the growth
of P-shaped particles all involve reducing the activity of In
monomers, compared to that of Cu. The latter can be achieved
by increasing the Cu concentration (i.e., the Cu:In ratio) (Figure
Figure 12. TEM pictures of different growth stages of alloyed CIS-ZnS
1
1a-c), or by moderately lowering the In activity through the
2
nanoparticles: (a) directly after the injection of ZnAc ; (b) after 1 min of
addition of technical grade TOPO (Figure 11d). Under these
conditions, the reaction again starts with the formation of copper
growth; and (c) the XRD-pattern of the particles shown in (b).
sulphide and, subsequently, Cu
along the c-axis. However, the rearrangement of the Cu
regions to nanodiscs of uniform width (particle 1 in Figure 11a)
is followed by a further shape change. One part of the discs
grows, while the other becomes thinner (particle 2 and 3) and,
2
S-CIS nanohybrids growing
absorption spectra of pure CIS and CIS-ZnS alloyed nanopar-
ticles, obtained by addition of zinc acetate to the reaction
mixture. Both samples contain nanorods with similar widths
and length of several tens of nanometers. The absorption of the
sample containing zinc is clearly blue-shifted. This effect is due
to the alloy formation and, again, not a result of size quantization.
Figure 12c shows an XRD pattern of a Zn-containing sample.
Basically, the same reflections as in the case of CIS are
observed, but they are slightly shifted. This means that the
sample has the same crystal structure, but with slightly different
lattice parameters. The Rietveld fit was obtained with Cu, In,
and Zn ions, randomly distributed over the Zn sites of the
wurtzite ZnS structure and lattice parameters of a ) 3.874 Å
and c ) 6.349 Å. This is a good proof for the formation of a
2
2
S
2
finally, the whole Cu S part is located in one corner of the CIS
particles (particle 4), which subsequently grows in the direction
perpendicular to the c-axis. The growth only takes place at the
interface between the two phases. A closer look at the P-shaped
particles unveils that they consist of a hemisphere located on top
of an elongated, hexagonal platelet. The size of the hemisphere
and the platelet can be controlled by the reaction conditions, like
the ratio between Cu and In. The X-ray diffraction patterns of the
nanoparticles obtained in this way again correspond to CIS with a
wurtzite structure, a shape analysis of these complex structures is,
however, not possible by XRD.
CuInS -ZnS alloy.
3
.5. Synthesis of CuInZnS. Our synthetic procedure turned
out to be also suitable for synthesizing other compounds, like
CuInS -ZnS alloys. TEM images of (CuInS )-(ZnS) nano-
crystals (composition obtained by EDX) are presented in Figure
2a,b. As shown in Figure 4, the absorption can be tuned when
ZnAc is injected during the CIS synthesis. ZnS is a wide
bandgap semiconductor (3.7 eV), and it has already been
4. Conclusions
A novel synthetic procedure has been described, allowing
highly advanced size and shape control of CIS nanoparticles
2
2
2
and Cu S-CIS hybrid nanostructures. Upon reducing the
2
1
activity of the In precursor, we can avoid the homogeneous
nucleation of CIS under our reaction conditions. So, in the first
2
4
1
shown that it can form alloys with CIS, which results in a
blue shift of the bandgap of the resulting material, compared to
CIS. The alloy formation is possible because of a small lattice
(
(
40) Will, G.; Hinze, E.; Adbelrahman, A. R. M. Eur. J. Mineral. 2002,
1
4, 591.
41) Pan, D.; Weng, D.; Wang, X.; Xiao, Q.; Chen, W.; Xu, C.; Yang, Z.;
42
mismatch of only 2.2% between CIS and ZnS. Figure 4 shows
Lu, Y. Chem. Commun. 2009, 4221.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 45, 2010 15985

A R T I C L E S
Kruszynska et al.
step of the reaction, pure Cu
can only start growing inside the Cu
2
S nanoparticles grow, and CIS
S phase. Our investigations
demonstrate the importance of the formation of Cu S nanopar-
ticles and Cu S-CIS hybrid nanostructures for the shape control
of CIS nanoparticles forming at the end of the reaction. Growth
of CIS takes place only at the Cu S-CIS interface, and so the
diameter of Cu S nanodiscs, being part of Cu S-CIS hybrid
material controls the diameter of the growing CIS nanocrystal.
As a consequence, any change of the shape of the Cu S nanodisc
changing their size and shape should be expected. This
motivated us to use doping with Zn as a means to change the
absorption of the nanocrystals. ZnS is known to form alloys
with CIS, and our method turned out to be suitable to synthesize
such class of materials, too.
2
2
2
2
Acknowledgment. We would like to thank Erhard Rhiel
2
2
(
University of Oldenburg) and Heike Oetting (University of
Oldenburg) for assistance in obtaining TEM images, and Dr. Marco
Schowalter (University of Bremen, Germany) for taking the
HRTEM images as well as the elemental analysis of single particles.
We gratefully acknowledge funding by the Federal Ministry of
Education and Research (BMBF, Project No. 03SF0338C) and
funding of the EWE Research Group “Thin Film Photovoltaics”
by the EWE AG, Oldenburg.
2
during the reaction results in a width modulation of the growing
CIS nanorods. The latter can be tuned by changing the reaction
conditions such that nanorods with different aspect ratio and
shape come forth.
Treatment with 1,10-phenanthroline, a reagent strongly bind-
ing to copper(I) ions, turned out to be not only a possibility to
2 2
remove Cu S parts from Cu S-CIS hybrid nanostructures, which
Supporting Information Available: Additional details on the
results in new shapes of CIS nanoparticles, not accessible
otherwise, but is also a possibility to specifically remove surface
copper atoms. CIS nanocrystals with an indium rich surface can
be obtained in this way.
With nanocrystals synthesized by our method being in the
size range of tens of nanometers, no size quantization effects
and, consequently, no tuning of the absorption properties by
XRD analysis of CIS nanoparticles and Cu S-CIS nanocom-
2
posites, XRD analysis of Cu-S nanoparticles forming at the
beginning of the reaction, Study of CIS nanoparticles before
and after treatment with phenanthroline (XRD, TEM), TEM
image of Cu S-CIS nanocomposites at an early stage of the
2
reaction. This material is available free of charge via the Internet
at http://pubs.acs.org.
(
42) Parasyuk, O. V.; Voronyuk, S. V.; Gulay, L. D.; Davidyuk, G. Y.;
Halka, V. O J. Alloys Compd. 2003, 348, 57.
JA103828F
1
5986 J. AM. CHEM. SOC. 9 VOL. 132, NO. 45, 2010