Chemical synthesis of Cu(In) metal inks to prepare CuInS2 thin films and solar cells

Article abstract of DOI:10.1016/j.jallcom.2010.07.190

We report a chemical method to prepare Cu/In inks or Cu-In inks, from which CuInS₂ films are formed. Cu, In and Cu-In nanoparticles are initially synthesized by a polyol method, and then dispersed in solvents to form inks. Metal films are subsequently obtained by drop-casting the Cu inks and In inks in sequence or Cu-In inks directly. The sulfurization of the metal films is carried out in a tube with high sulfur vapors. The obtained CuInS₂ films are compact with large grains and pure phases. The band gap of the CuInS₂ film is determined to be ～1.45 eV, which is near the optimal band gap of the solar cell materials. Tentative CuInS₂ solar cell with an efficiency of 0.7% is fabricated.

Full text of DOI:10.1016/j.jallcom.2010.07.190

Journal of Alloys and Compounds 507 (2010) 317–321
Contents lists available at ScienceDirect
Journal of Alloys and Compounds
journal homepage: www.elsevier.com/locate/jallcom
Chemical synthesis of Cu(In) metal inks to prepare CuInS thin
2
films and solar cells
∗
Guanbi Chen, Lei Wang, Xia Sheng, Hongjuan Liu, Xiaodong Pi, Deren Yang
State Key Laboratory of Silicon Materials and Department of Materials Science and Engineering, Zhejiang University, Hangzhou 310027, People’s Republic of China
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 7 May 2010
Received in revised form 23 July 2010
Accepted 27 July 2010
Available online 5 August 2010
We report a chemical method to prepare Cu/In inks or Cu–In inks, from which CuInS2 films are formed. Cu,
In and Cu–In nanoparticles are initially synthesized by a polyol method, and then dispersed in solvents
to form inks. Metal films are subsequently obtained by drop-casting the Cu inks and In inks in sequence
or Cu–In inks directly. The sulfurization of the metal films is carried out in a tube with high sulfur vapors.
The obtained CuInS2 films are compact with large grains and pure phases. The band gap of the CuInS2 film
is determined to be ∼1.45 eV, which is near the optimal band gap of the solar cell materials. Tentative
CuInS2 solar cell with an efficiency of 0.7% is fabricated.
Keywords:
Nanoparticles
Metal inks
©
2010 Elsevier B.V. All rights reserved.
CuInS2 thin film
Solar cells
1
. Introduction
CuInS (CIS) thinfilms withopticalbandgap of1.5 eVare favored
by a physical top-down method (MA or gas atomization) [13]. In
this case, again, the MA cannot lead to a narrow size distribution
of Cu–In particles. It is usually difficult to handle gas atomization,
especially given the fact that the boiling point of In is low [14].
Therefore, simple and cost-effective methods to prepare particles
related to CIS films still need to be developed.
In this paper, we present a novel low-cost method to prepare
Cu, In or Cu–In nanoparticles, which then are used to fabricate CIS
films. Modified polyol method is successfully employed to synthe-
sis metal nanoparticles. The metal nanoparticles are then dispersed
in solvents to form inks. Later on, metal films can be subsequently
obtained by drop-casting Cu inks and In inks in sequence or Cu–In
inks, analogous to the subsequent or simultaneous evaporation in
industry, respectively. Finally, the sulfurization of the metal films
leads to the formation of CIS films, which are compact with large
grains and pure phases. CIS solar cell is also prepared and investi-
gated.
2
for the production of high-efficiency single-cell systems with large
open circuit voltage and of top cells in tandem systems [1]. How-
ever, expensive vacuum equipments are employed in industrial
production of CIS films, usually by subsequent evaporation of ele-
mental layers (i.e. Cu, In, etc.), or simultaneous co-evaporation of
these elements [2,3]. To reduce the cost of the thin-film solar cells
non-vacuum techniques are intensively studied [4–6]. Among a
variety of non-vacuum techniques, nanoparticle-based ink printing
is attractive, given the convenience for large-scale production and
sufficient utilization of materials [7–9]. Thereinto, nanoparticles
play an important role in the preparation of inks.
Up to now, three methods have mainly been reported for the
preparation of nanoparticles related to CIS films. The first is that CIS
particles are directly synthesized by sol–gel or mechanical alloying
(
MA). However, the sol–gel process leads to porous CIS aggregates,
which are readily oxidized [10]. Moreover, large distribution of par-
ticle sizes is also a serious problem in the MA [11]. The second is that
metal precursors such as metal oxides or metal salts are initially
prepared. After reduction and selenization/sulfurization, separate
particles are sintered together to form integrated films. However,
since metal-oxide impurities may remain in the films to initiate
cracks and shrinkages, an extra reduction process must be carried
out [12]. The third is that fine Cu–In alloy particles are produced
2
.
Experimental
Analytical grade reagents were used in this work. They were indium trichlo-
ride, InCl3·4H2O; copper dichloride, CuCl2·2H2O; poly(vinyl pyrrolidone) (PVP, K30);
sodium borohydride (NaBH4); diethylene glycol (DEG); tetraethylene glycol (TEG);
ethylene glycol (EG), 2-methoxyethanol (MeEtOH), and methyl alcohol (MeOH);
sulfur (S). The nanoparticles were synthesized by an approach similar with the com-
monly used polyol method [15]. 1 mmol metal salt (InCl ·4H O or CuCl ·2H O) and
3
2
2
2
◦
0.2 g PVP were firstly dissolved into 25 ml DEG under N2 at 140 C. Some amount
of freshly made NaBH4/TEG solution (0.2 g in 8 ml TEG) was added. The particles
continued to grow for some time (3 min for In particles and 30 min for Cu parti-
cles). As for the synthesis of Cu–In nanoparticles, the same experimental procedure
was used, in which equal amount of Cu and In salts were added. The products were
separated by centrifugation and washed several times by ethanol.
∗ Corresponding author. Tel.: +86 571 87952752; fax: +86 571 87952322.
E-mail address: mseyang@zju.edu.cn (D. Yang).
0
925-8388/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2010.07.190

3
18
G. Chen et al. / Journal of Alloys and Compounds 507 (2010) 317–321
The synthesized nanoparticles were mixed with
a
stock solvent
(
EG/MeEtOH/MeOH = 5:4:1 by weight). In the mixtures the nanoparticles
were 20% by weight. Inks were formed by ultrasonicating or grinding the mixtures.
Glass substrates were ultrasonically cleaned in a water–detergent solution and
then washed with acetone and ethanol. They were finally dried in a furnace at
◦
8
0
C. We used two methods to prepare metal films on glass substrates denoted as
stacking and mixing methods. In the stacking method the films were prepared by
drop-casting Cu inks and In inks in sequence (stacking method 1) or In inks and Cu
inks in sequence (stacking method 2). In the mixing method the films were formed
◦
by drop-casting Cu–In inks. After each drop-casting, a drying process at 80 C for
3
0 min was performed. The metal films were loaded into a quartz–tube furnace in
sulfur atmosphere under the protection of nitrogen gas. The furnace temperature
◦
was ramped from room temperature to 250 C in 0.5 h and stayed for 20 min to
◦
eliminate organic residuals. The temperature was then ramped to 550 C in 1 h and
held for another 1 h to complete the sulfurization. The CIS solar cell was prepared
with a structure of Ag/ITO/i-ZnO/CdS/CIS/Mo following the custom procedures.
X-ray diffraction (XRD) was carried out to study the crystalline structures of
all the samples by using an X’Pert PRO (PANalytical) diffractometer equipped with
◦
a Cu K␣ radiation source. Data was collected by step-scanning of 2ꢀ from 20 to
◦
◦
7
0 with a step of 0.02 and counting time of 1 s per step. Morphology of the sam-
ples was investigated by scanning electron microscopy (SEM, Hitachi S4800). The
optical transmittances (T) of the finished films were measured by using a UV–vis
spectrophotometer (Hitachi Model U-4 100). ABET solar simulator was used as the
light source and certificated with a monocrystal Si solar cell. I–V characteristic of
the device was carried on an ammeter of Keithley 2400.
Fig. 1. XRD diffraction patterns of (a) Cu nanoparticles, (b) In nanoparticles and (c)
Cu–In nanoparticles.
3
. Results and discussion
In the experiments, Cu, In or Cu–In nanoparticles are firstly syn-
−
B(OH) /BH4 can make a slower nucleation, leading to a multi-
3
thesized, respectively. The XRD pattern of the Cu nanoparticles
shown in Fig. 1(a) indicates that the Cu is face-centered-cubic (fcc)
crystal structure, the same as that in Ref. [16]. Their SEM image
shown in Fig. 2(a) displays that the Cu nanoparticles are uniform
with particle size of about 50 nm. Moreover, In nanoparticles with
tetragonal phase are synthesized, as indicated from the XRD pattern
in Fig. 1(b). In the pattern, no preferred orientation is found, indi-
cating that the In products follows an isotropic growth process. The
SEM image shown in Fig. 2(b) illustrates that the In nanoparticles
are sphere and poly-sized with a wide range of particle sizes from
nucleation. In the growth period, the nuclei will grow bigger at the
base of their initial sizes and distributions. Therefore, Cu nanoparti-
cles from nuclei with similar sizes will follow a same growth period
and display an approximate particle sizes. Whereas, In nanoparti-
cles from nuclei with varied sizes have a much wider range of size
distribution. Moreover, the higher diffusion rates of In in the sol-
vent makes them easier to ripen, also leading to a much wider range
of size distribution. Thus, the above-mentioned reaction mecha-
nisms lead to the morphological difference between the Cu and In
nanoparticles.
1
0 nm to 200 nm. The big difference of the particle size and distribu-
tion between the Cu and In nanoparticles can be explained from the
reaction dynamics, as following. Basically, the synthesis reaction
The XRD pattern shown in Fig. 1(c) elucidates that the Cu–In
nanoparticles contain the two main phases of In (tetragonal,
JCPDS-65-9682) and Cu11In9 (monoclinic, JCPDS-65-4963), which
is consistent with the previous report that stable component phases
of Cu and In nanoparticles follows a simple oxide/reduction cou-
ple rule, in which Cu2+ or In3+ are reduced by the strong reductive
◦
of Cu–In alloy above 150 C are mainly In and Cu11In9 [17]. In fact,
agent NaBH . The potential gap between the couple controls the
4
we obtain the two component phases at a relative lower tempera-
reaction potential. Comparing with the standard reduction poten-
◦
O
ture of 140 C. The obtained Cu–In nanoparticles are bigger in sizes
tial (E
= −0.338 eV vs SHE (standard hydrogen electrode),
3
+
In /In
due to the inter-diffusion of Cu and In nanoparticles [18], as indi-
cated from the SEM image of a Cu–In metal film in Fig 4(a). It is
considered that the polyol reaction might actualize some reactions
which are difficult to happen in solid reactions [18]. However, we
found that if the reaction temperature was over 150 C the Cu–In
nanoparticles would heavily agglomerate.
O
O
E
= 0.345 eV vs SHE, E
= −0.481 eV vs SHE), it is
2+
−
Cu /Cu
B(OH)3/BH4
clear that Cu2+ has a higher reaction potential to be reduced by
3
+
NaBH4 than In . Thus, at the very beginning of the reaction, it is
much quicker to form Cu nuclei, leading to a single nucleation event
◦
[
16]. However, the much closer potential gap between In3+/In and
Fig. 2. SEM images of (a) Cu nanoparticles and (b) In nanoparticles by polyol method.

G. Chen et al. / Journal of Alloys and Compounds 507 (2010) 317–321
319
nanoparticles could be decomposed to Cu16In and liquid In at high
9
temperatures, which may be beneficial for the sulfurization [19].
Thus, no distinct impurity phases are observed in the CIS film fabri-
cated by the mixing method. In the mixing methods, although the
inter-diffusion of Cu and In to form Cu11In phase dominates, some
9
Cu and In nanoparticles in stacking layers may also be sulfurized
ahead. In element may be evaporated in the form of indium sulfide
(
e.g. In S) at high temperature, leaving behind the impurity phase
2
of CuS in the sulfurized samples [19,20], which plays an active role
in the growth of CIS thin films [21]. Therefore, the impurity phase of
CuS could be observed in the X-ray patterns of the films fabricated
by the stacking methods. Moreover, it can be seen from Fig. 3 that
no oxide phases can be detected in all the CIS films, indicating that
during the nanoparticle synthesis, the PVP could slow down the
oxidation rate of the synthesized Cu, In and Cu–In nanoparticles, as
reported in Ref. [16,22].
The SEM image of the Cu–In metal film prepared by the mixing
method before sulfurization is shown in Fig. 4(a). It can be seen
that the Cu–In metal film is loose. After sulfurization, the CIS film
is formed from the metal film, which is a compact film with large
grain sizes, as shown in Fig. 4(b). Moreover, the cross-section SEM
image of the films in Fig. 5(a) can also verify this. Compact films
with thickness of over 3 ␮m can be obtained. The morphologies of
the films prepared by the stacking methods are similar to that by
the mixing method, which are not introduced. In the past decades,
there have been many reports concentrating on the sulfurization
of sputtering In/Cu stacking layer by either ex- or real-time tech-
niques [19,20,23]. It has been well accepted that such sulfurization
proceeds at the expense of Cu11In9 and is actually a surface reac-
tion. The Cu and In atoms from Cu11In9 have to diffuse through
the pre-formed CIS to react with S vapor. This can be verified from
the cross-section SEM image of our atypical “extra-thick” Cu–In
Fig. 3. XRD diffraction patterns of CIS films by (a) mixing method and (b) stacking
method 2 (In/Cu layers) and (c) stacking method 1 (Cu/In layers).
After the synthesis of Cu, In and Cu–In nanoparticles, the corre-
sponded inks are fabricated by dispersing the particles in solvents,
which is mentioned in the section of experiment. And then, CIS
films are formed by stacking and mixing methods and by sulfuriza-
tion in sequence. The XRD patterns of the CIS films fabricated by
three methods are shown in Fig. 3. It can be found that the main
diffraction peaks of all the samples conform to the tetragonal phase
of CIS (JCPDS-65-1572). Only single phase is observed in the film
prepared by the mixing method, whereas some impurity phases
of CuS can be found in the CIS films prepared by the two stack-
ing methods. In the mixing method, the Cu11In phase in the Cu–In
9
Fig. 4. SEM images of Cu–In metal films (a) before and (b) after sulfurization.
Fig. 5. Typical (a) and atypical (b) SEM cross-section images of the CIS films fabricated by the mixing method.

3
20
G. Chen et al. / Journal of Alloys and Compounds 507 (2010) 317–321
Fig. 6. Plot of absorbance against wavelength and (˛hꢁ)2 against photoenergy (hꢁ)
of CIS films from different preparation methods of metal films.
Fig. 7. Plot of current density vs voltage of the CIS solar cell.
nanoparticle films after sulfurization (Fig. 5(b)). In our experiments,
the sulfurized films have two distinct layers. The top layer of about
ative poor performance to the micro-cracks or pores due to the
stresses in the sulfurization process. These cracks on one hand are
beneficial for the sulfurization, but on the other hand may result
in more leakage current paths and high contact resistance which
are detrimental for the cell performances. In addition, the long sul-
furization time we employed may also result in thick MoS2 layer
which can increase the series resistance of our devices. Therefore,
careful control of the sulfurization is very important to improve the
performances.
3
␮m thick is compact with large grain sizes, whereas the lower
layer is loose with relatively small grains. In the extra-thick films,
Cu and In atoms must march through a longer diffusion distances
to be sulfurized on the surface, which is much more difficult. There-
fore, sulfur must diffuse downward to the bottom to sulfur these
atoms. The much lower sulfur pressure and shackled reaction place
can result in the much smaller CIS grains. From this point of view,
appropriate control of metal film thickness and sulfur pressure are
very important to prepare qualified CIS films for solar cells.
The optical transmittances (T) of the obtained CIS films are
measured by using a UV–vis spectrophotometer. The absorption
coefficient (˛) can be calculated from the relationship of transmit-
tance (T) following the formula below [24]:
4. Conclusions
A simple chemical method to prepare Cu/In or Cu–In nanoparti-
cles inks is developed. These metallic inks are applied to prepare CIS
films. Both stacking and mixing methods are employed to prepare
the metal films, corresponding to the two main states of art meth-
ods in industry. It is proven that all the methods based on our metal
ln(1/T)
˛
=
(1)
d
For simplicity, the direct characterization data of transmit-
inks are feasible to prepare CIS films. Moreover, the Cu11In phase in
9
tance are demonstrated in the term of absorbance (ln(1/T)). The
absorbance spectrums of the CIS films fabricated by different meth-
ods are shown in Fig. 6. Moreover, the fundamental absorption,
which manifests itself by a rapid rise in absorption, can be used to
determine the energy gap (Eg) of a semiconductor. It is well known
the mixing method is beneficial for the sulfurization, exhibiting lit-
tle impurity phases. CIS films by different methods similarly exhibit
high absorption coefficient and appropriate EG of 1.45 eV. CIS thin
film solar cell by the mixing method is fabricated and obtained an
efficiency of 0.7%.
m
that ˛hꢁ is proportional to (hꢁ − Eg) , where hꢁ is the photon
energy. Values of m for allowed direct and indirect, and forbidden
direct and indirect optical transitions are 1/2, 2, 3/2 and 3, respec-
tively. Thus, the band gaps are obtained by extrapolating the linear
Acknowledgements
portion of the plots of (˛hꢁ)1 versus hꢁ to ˛
/m
1/m
The authors would like to appreciate the financial supports from
= 0. For an allowed
direct transition (m = 1/2), ˛ can be expressed as follows [25]:
9
73 Project (No. 2007CB613403). We thank Jieru Wang and Guo-
ꢀ
ꢁ
liang Xu for their kind help for characterization and Jipeng Cheng
for his deep discussion in experiments.
A1
1/2
˛
=
(hꢁ − Eg)
(2)
hꢁ
where A1 is a constant and Eg is the energy gap. The plot of (˛hꢁ)2
versus hꢁ is also drawn (Fig. 6) for the allowed direct band gap.
The band gap EG of 1.45 eV is determined from the extrapolated
intercept with the Energy (hꢁ) axis. It can be seen that EG is nearly
the same for all the films fabricated by different methods, which
agrees with that of the CIS films synthesized with a similar method
References
[1] N. Guezimir, T. Ben Nasrallah, K. Boubaker, M. Amlouk, S. Belgacem, J. Alloys
Compd. 481 (2009) 543–548.
[
2] C. Calderon, P. Bartolo-Perez, J. Clavijo, J.S. Oyola, G. Gordillo, Sol. Energy Mater.
Sol. Cells 94 (2010) 17–21.
[3] A. Bollero, J.F. Trigo, J. Herrero, M.T. Gutierrez, Thin Solid Films 517 (2009)
167–2170.
4] F.M. Cui, L. Wang, Z.Q. Xi, Y. Sun, D.R. Yang, J. Mater. Sci.-Mater. Electron. 20
2009) 609–613.
[5] N.K. Allouche, N. Jebbari, C. Guasch, N.K. Turki, J. Alloys Compd. 501 (2010)
5–88.
6] S.J. Peng, F.Y. Cheng, J. Liang, Z.L. Tao, J. Chen, J. Alloys Compd. 481 (2009)
86–791.
[7] Z.D. Wang, X.L. Mo, J. Li, D.L. Sun, G.R. Chen, J. Alloys Compd. 487 (2009) L1–L4.
2
[
26].
A typical CIS solar cell by the mixing method is fabricated
[
(
with the conventional structure. The image in Fig. 7 shows the
current–voltage characteristic of the solar cell, indicating that it
8
[
2
exhibits a performance parameters of Voc = 0.37 V, Isc = 6 mA/cm ,
7
fill factor (F.F.) = 30% and efficiency (ꢂ) = 0.7%. We attribute the rel-

G. Chen et al. / Journal of Alloys and Compounds 507 (2010) 317–321
321
[
[
8] L. Li, N. Coates, D. Moses, J. Am. Chem. Soc. 132 (2010) 22–23.
9] H.W. Hillhouse, M.C. Beard, Curr. Opin. Colloid Interface Sci. 14 (2009) 245–259.
[19] A. Joswig, M. Gossla, H. Metzner, U. Reislohner, T. Hahn, W. Witthuhn, Thin
Solid Films 515 (2007) 5921–5924.
[
[
[
[
[
10] S.Y. Lee, K.H. Kim, B.O. Park, Thin Solid Films 516 (2008) 4709–4712.
11] T. Wada, H. Kinoshita, Thin Solid Films 480–481 (2005) 92–94.
12] M. Kaelin, D. Rudmann, A.N. Tiwari, Sol. Energy 77 (2004) 749–756.
13] B.M. Basol, Thin Solid Films 361–362 (2000) 514–519.
14] G. Norsworthy, C.R. Leidholm, A. Halani, V.K. Kapur, R. Roe, B.M. Basol, R. Mat-
son, Sol. Energy Mater. Sol. Cells 60 (2000) 127–134.
[20] M. Gossla, H.E. Mahnke, H. Metzner, Thin Solid Films 361–362 (2000) 56–60.
[21] J. Klaer, J. Bruns, R. Henninger, K. Seimer, R. Klenk, K. Ellmer, D. Braunig, Semi-
cond. Sci. Technol. 13 (1998) 1456–1458.
[22] S. Jeong, K. Woo, D. Kim, S. Lim, J.S. Kim, H. Shin, Y.N. Xia, J. Moo, Adv. Funct.
Mater. 18 (2008) 679–686.
[23] E. Rudigier, J. Djordjevic, C. von Klopmann, B. Barcones, A. Perez-Rodriguez, R.
Scheer, J. Phys. Chem. Solids 66 (2005) 1954–1960.
[24] H.X. Tang, M. Yan, H. Zhang, X.Y. Ma, L. Wang, D.R. Yang, Chem. Res. Chinese U
21 (2005) 236–239.
[
[
15] Y.G. Sun, Y.N. Xia, Adv. Mater. 14 (2002) 833–837.
16] B.K. Park, S. Jeong, D. Kim, J. Moon, S. Lim, J.S. Kim, J. Colloid Interface Sci. 311
(
2007) 417–424.
[
[
17] M. Gossla, H. Metzner, H.E. Mahnke, J. Appl. Phys. 86 (1999) 3624–3632.
18] N.H. Chou, R.E. Schaak, J. Am. Chem. Soc. 129 (2007) 7339–7345.
[25] N. Kavcar, Sol. Energy Mater. Sol. Cells 52 (1998) 183–195.
[26] C.P. Liu, B.H. Tseng, Thin Solid Films 480–481 (2005) 50–54.