Journal of
Emerging Investigators
Fabrication of CuSbS2 Solar Cells by Sulfurization of Thermally
Evaporated Metal Stacks
Mariam Abou-Dahech1, Nicholas Franzer2, Rajendra Khanal2, Adam Phillips2, and Yanfa Yan2
1
Toledo Early College High School, Toledo, OH
2
University of Toledo, Toledo, OH
Summary
We report on the deposition of thin films of chalcostibite
(CuSbS2) created by sulfurization of thermally evaporated
metal stacks. We found by X-ray diffraction that
thermally evaporated copper and antimony metal films,
which were annealed for 1.5 hours at 400°C followed by
rapid sulfurization at 400°C for 15 minutes, exhibit the
CuSbS2 crystalline phase. The films have high optical
absorption with a band gap of ~1.6 eV. Our CuSbS2based thin film solar cells had low conversion efficiency,
due to the formation of a layer of molybdenum disulfide
(MoS2) and/or secondary phases at the molybdenum
back contact layer–CuSbS2 interface. To negate this
problem, we applied a layer of single-walled carbon
nanotubes (SWCNT) on the molybdenum back contact.
This arrangement allowed for a better junction between
the molybdenum and CuSbS2 layers. With the addition of
SWCNT, our best cell under AM 1.5 illumination exhibited
an open circuit voltage (Voc) of 321 mV, a short circuit
current density (Jsc) of 3.76 mA/cm2, and a fill factor of
27%, resulting in a conversion efficiency of 0.33%. This
work shows that for effective solar devices, it is critical
that device layers are electrically aligned to allow for
better electron passage. In the case presented here,
we increased the efficiency by creating a passageway
for electrons through the device by eliminating the
formation of MoS2. This can be achieved by inserting
single-walled carbon nanotubes (SWCNTs) that allow a
better junction between the molybdenum and CuSbS2
layers. An alternative may include changing sulfur
quantity, sulfurization time, or temperature.
Received: September 21, 2015; Accepted: March 24,
2016; Published: July 19, 2016
Copyright: (C) 2016 Abou-Dahech, Franzer, Khanal,
Phillips, and Yan. All JEI articles are distributed under the
attribution non-commercial, no derivative license (http://
creativecommons.org/licenses/by-nc-nd/3.0/). This
means that anyone is free to share, copy and distribute
an unaltered article for non-commercial purposes
provided the original author and source is credited.
Introduction
Photovoltaics produce power by converting the sun’s
energy into electrical power. Visible light is absorbed by
material with a lower band gap than the light’s energy.
The light travels through the other layers in the device
with higher band gaps until it reaches the lower band
Journal of Emerging Investigators
July 19, 2016
gap layer. Once light with energy greater than the band
gap of the semiconductor is absorbed, an electron-hole
pair is created (Figure 1a). An electric field produced at
the junction of the N-type and P-type layers drives the
electron and hole to opposite ends of the device where
the conductors are located. The electron-hole pair then
can be extracted to power a load (1). A high resistivity
transparent layer is included to prevent direct contact
of the two conductors present in the devices in case of
holes in any layers.
Figure 1. (a) Illustration of the parts of a photovoltaic device.
(b) A depiction of the individual layers that a CuSbS2 device
is comprised of.
Currently, silicon and thin film, such as cadmium
telluride (CdTe), panels dominate the photovoltaic market.
Silicon solar panels are made from wafers that must be
highly purified, causing their production to be expensive
(2), and thin films use rare or expensive elements
(3). Therefore, investigating emerging photovoltaic
technologies using nontoxic and economically viable
elements is warranted.
In this study, we used chalcostibite (CuSbS2) as an
absorbing material in a photovoltaic (PV) device because
of its feasibility and relatively abundant nature (4). With
a band gap similar to that of CdTe, CuSbS2 is expected
to absorb a similarly large portion of the solar spectrum.
Additionally, it is less expensive due to more earthabundant source materials. The CuSbS2 was fabricated
by thermally evaporating copper and antimony, followed
by annealing and sulfurization; this technique generated
large grain CuSbS2. For device fabrication, CuSbS2 was
deposited on molybdenum-coated soda lime glass in
substrate configuration (Figure 1b). The CuSbS2 device
1
Journal of
Emerging Investigators
Figure 2. Graphical representation of the (αhυ)2 vs hυ for
the three temperatures. By extrapolating the data to the
x-intercept, it is clear that the band gap is ~1.6 eV for all
annealing temperatures. This is near the ideal band gap for
solar conversion.
was completed with 100 nm of cadmium sulfide (CdS),
500 nm of intrinsic zinc oxide (i-ZnO), and 1 micron of
aluminum zinc oxide (AZO). The completed devices
exhibited low efficiency due to the formation of a layer
consisting MoS2 and secondary phases (compounds
that consist of Cu, Sb, and/or S that are not CuSbS2, e.g.
CuS2 or CuS) at the back contact during sulfurization
similar to what others have observed with a related
materials system (5). Inserting single-walled carbon
nanotubes (SWCNTs) between the Mo and CuSbS2
layers limited the MoS2 formation (6) and resulted in
increased efficiency.
Results and Discussion
Film Characterization
After the fabrication of our devices, several tests were
needed in order to characterize the finished solar cells.
We first determined the optical and crystal structure
properties of our materials.
The absorption coefficient was determined, and the
results were used to generate Figure 2, which shows
the square of the photon energy multiplied by the optical
absorption across wavelengths. Because CuSbS2 is a
direct band gap semiconductor, the x-intercept of the
linear portion is its band gap. This absorption spectrum
ultimately illustrates a high optical absorption by the
material produced by all three annealing temperatures.
A straight line was added to see what band gap our
extrapolation hits. We determined that the band gap
of the CuSbS2 layer is ~1.6 eV, which is a near ideal
band gap for our absorbing material, CuSbS2. Next, we
utilized the X-ray diffraction (XRD) instrument in order to
characterize the structure of the synthesized material.
Figure 3 shows the measured XRD data on the top and
the reference signals of crystalline Mo and CuSbS2 (the
material we deposited on and the material we created,
respectively) on the bottom. Since our peak locations of
Journal of Emerging Investigators
July 19, 2016
Figure 3. X-ray diffraction spectrum of metal stacks annealed
and rapidly sulfurized. Each peak characterizes a crystal
plane within the material’s overall crystal lattice in the film.
the measured data matched the standards, with roughly
the same height and position, we confirmed that we
had crystalline Mo (with a very strong peak at ~41°) and
CuSbS2 (with pairs of peaks at ~28° and 30°), which is
the material we were trying to make.
Solar Cell Performance
Knowing that we created CuSbS2 with a 1.6 eV band
gap, we investigated how a device made with this material
operated under direct sunlight. Complete devices were
made using 400°C annealing conditions. The dark and
illuminated current densities-voltage curves (J-V) were
measured to determine the conversion efficiency of the
solar cells (Figure 4a). Ideally, each photon above the
band gap should result in the generation of one electron.
By integrating all the photons in the solar spectrum at the
surface of the earth (AM 1.5) with energy greater than
the band gap of our material, 1.6 eV, we would expect 22
mA/cm2 (7). Our device resulted in a Jsc (the y-intercept)
of 0.758, which is extremely low. Many of the photons
did not yield electrons in our device. If the solar cell is
efficient the results would follow a classic diode curve,
in which the current density (shown on the y-axis) would
be nearly constant before sharply decreasing. The
measured J-V response of this device, though, is linear.
This could be due to high series resistance caused by
a resistive layer preventing electron movement or by a
Figure 4. J-V curve of the best CuSbS2-based thin-film
solar cell (a) without SWCNTs and (b) with SWCNTs at the
molybdenum and CuSbS2 junction.
2
Journal of
Emerging Investigators
Figure 5. Cross-sectional SEM (a) and backscattering electron (b) images indicate an insulating layer (shown
by black arrow) between the CuSbS2 and molybdenum. This is likely MoS2 formed during sulfurization.
Backscattering electron image of the materials stack with single-wall carbon nanotubes (c) does not show the
insulating layer.
very low shunt resistance caused by low quality material,
which both result in a low fill factor.
In order to determine what the problem was with
our device, we used cross-sectional scanning electron
microscopy (SEM) (shown in Figure 5a-b). Each layer of
the device (see Figure 1b for reference) is shown in this
SEM image (Figure 5a), except the n-layer because it is
thin and partially diffused into the absorber layer. In this
image, the CuSbS2 is the large layer in the center, and the
smaller one towards the bottom is the molybdenum. For
our device to work, the CuCbS2 and Mo layers must be in
good electrical contact. According to the backscattering
image (Figure 5b), there is a small dark layer (which
indicates a poor conductivity layer) between the CuSbS2
and Mo layers, likely composed of MoS2 and other Cu-,
Sb-, Mo-, and S-containing secondary phases, which
prevents contact between the conductor (Mo) and the
P-type layer (CuSbS2), similar to what others have
observed (5). Without this connection, the electrons
cannot successfully pass through opposite ends of the
device to be extracted and to power a load. It should be
noted that the XRD peak for MoS2 (~12°) is below the
measured range, so the MoS2 was not verified.
To eliminate the problem of unsuccessful electron
passage through the device, a layer of single-wall
carbon nanotubes (SWCNTs) was deposited on the
molybdenum prior to CuSbS2 deposition. SWCNTs have
recently been used to form good electrical contacts to
other solar cell materials even though the simple band
alignment predictions would suggest otherwise (8 - 9).
Here, they could increase the efficiency by either forming
good electrical contacts or by limiting the extent to which
S interacts with Mo (6). In either case, these tubes serve
as a passageway for electrons so that they can get to the
CuSbS2 layer of the device. Unlike the cross sectional
backscattering image of the device without SWCNTs
(Figure 5b), the cross sectional backscattering image
of the device with SWCNTs (Figure 5c) does not show
the dark line indicating an insulating layer between the
CuSbS2 and the Mo layer. We interpret this to mean that
the SWCNTs prevented MoS2 formation between the
Journal of Emerging Investigators
July 19, 2016
molybdenum and CuSbS2.
With no MoS2 layer or secondary phases evident,
we expected the device with the SWCNT layer to
perform better, providing that the SWCNT layer does not
adversely affect charge transport. In the J-V curve of the
device with SWCNTs, shown in Figure 4b, the current
density line (y-axis) better mirrors the theoretical graph
of an efficient solar cell; the graph is no longer linear.
Comparing the results of these devices in Table 1, we see
that the carbon nanotubes had a positive effect on the
solar cells’ efficiency. With this information, we conclude
that our solar cells with SWCNTs are significantly more
efficient than the solar cells without nanotubes. However,
since our material has the capability of reaching a 22
mA/cm2 current density, and based on the 3.764 mA/cm2
density our device actually reached, we also believe that
the efficiency of our device is lower than it could be.
Conclusion
CuSbS2 thin film devices created by thermal
Performance Parameters
Without SWCNTs
With SWCNTs
VOC (V)
0.086
0.321
JSC (mA/cm2)
0.758
3.764
Fill Factor (%)
24.8
27.0
Efficiency (%)
0.016
0.326
Table 1. Performance parameters of the best CuSbS2based thin-film solar cell with and without SWCNT at the
molybdenum and CuSbS2 junction.
evaporation have been characterized by their optical
and structural properties. CuSbS2 films demonstrate
high optical absorption, with a band gap of ~1.6 eV and
a 400°C anneal having stronger absorption. During the
fabrication of our solar devices, a layer of MoS2 formed
at the molybdenum/CuSbS2 junction, preventing a
significant PV response. Application of SWCNTs on the
molybdenum reduced the formation of MoS2 and allowed
for a better electrical junction between the molybdenum
and CuSbS2. This enabled us to get a solar cell efficiency
of 0.33% under AM 1.5 illumination. Further optimization
3
Journal of
Emerging Investigators
of the device is needed so that the layer of MoS2 will
not form and interfere with cell performance. Optimizing
the molybdenum and CuSbS2 junction by varying the
thickness of the carbon nanotubes (9) or by finding an
alternative layer that prevents formation of MoS2 are
two ways this can be achieved. Changing the amount
of sulfur used or the sulfurization time and temperature
may also result in better device performance.
Methods
Material Characterization
CuSbS2 was fabricated on soda lime glass substrates.
Prior to deposition, the soda lime glass was ultrasonically
cleaned using an ammonium-based solution. Copper
and antimony layers were thermally evaporated in a
three-layer metal stack sequence. Previous experiments
indicated that 75 nm of copper, 250 nm of antimony, and
25 nm of copper result in the best film. The stacks were
then annealed to induce mixing of the metal stack. The
samples were annealed in a argon purged furnace. The
temperature ramp time was 1.5 hours, and the dwell
time was 30 minutes. Three different temperatures, 400,
450, and 500°C, were tested to determine the optimum
annealing conditions. The samples were allowed to
cool before sulfurization. Sulfurization was completed
in a graphite box containing 0.15 g of sulfur powder
and heated to 400°C for 15 minutes in argon ambient
using a rapid thermal annealer. This caused the films
to crystallize into a 1-μm thick CuSbS2 film. The films
were characterized by X-ray Diffraction (XRD) and a
UV-Vis Spectrophotometer. The XRD was set to focus
beam mode with a sweep rate of 1°/min. The data was
obtained in air at room temperature.
Optical absorption determination
When light enters the device, absorption takes
place. By calculating this absorption, we confirmed our
band gap. The band gap from our absorbing material,
CuSbS2, determines the minimum amount of energy a
photon must have in order to be absorbed. The optical
absorption (α) was used to determine the band gap of
the material produced by each annealing temperature
using equation 1 (10).
Where d is the film thickness, and R and T are the
reflectance and transmission, respectively.
Solar Cell Fabrication
An 800-nm molybdenum back contact was sputtered
Journal of Emerging Investigators
July 19, 2016
on the SLG using Radio Frequency (R.F.) magnetron
sputtering. The CuSbS2 was then deposited following
the process noted above. A 100-nm thick CdS window
layer was deposited on the CuSbS2 using a chemical
bath deposition (CBD) technique (11), and the device
was completed by R.F. sputtering of 500 nm i-ZnO and
1µm AZO. Dot cells with an area of 0.09 cm2 were then
prepared by mechanically scribing. The finished devices
were characterized by imaging the cross-sections of
devices that were cracked in half. The SEM samples
were prepared by scribing the glass on the opposite
side of the device materials and cracking with a glass
cracking tool. The samples were mounted onto a
vertical holder that pointed the sample edge towards the
electron beam using copper tape to provide the required
conductivity. To acquire the SEM image of the surface,
the accelerating potential and working distance were set
to 2 kV and 11 mm, respectively. For the backscattering
measurements, the accelerating potential was increased
to 20 kV and the working distance was set to 25 mm. The
J-V response was measured under a solar simulator with
an AM 1.5 illumination, which is equivalent to that at the
surface of the earth.
Acknowledgements
This work was supported by the Research In Science
& Engineering (RISE) program at the University of
Toledo, with financial assistance from UT’s College of
Natural Science and Mathematics, the National Science
Foundation (No. 1230246), and discretionary funds from
Profs. Ellingson, Heben, and Yan.
References
1. Luque, Antonio. Handbook of Photovoltaic Science
and Engineering. Chichester: Wiley, 2003. Print.
2. Saga, Tatsuo. “Advances in Crystalline Silicon Solar
Cell Technology for Industrial Mass Production.” NPG
Asia Materials 2.3 (2010): 96-102. Print.
3. “National Center for Photovoltaics.” NREL. N.p., 5
June 2015. Web. 01 Aug. 2015.
4. Yang, Bo, Liang Wang, Jun Han, Ying Zhou, Huaibing
Song, Shiyou Chen, Jie Zhong, Lu Lv, Dongmei
Niu, and Jiang Tang. “CuSbS2 as a Promising
Earth-Abundant Photovoltaic Absorber Material:
A Combined Theoretical and Experimental Study.”
Chemistry of Materials (2014): 3135-143. Print.
5. Scragg, Jonathan J., J. Timo Wätjen, Marika
Edoff, Tove Ericson, Tomas Kubart, and Charlotte
Platzer-Björkman. “A Detrimental Reaction at the
Molybdenum Back Contact in Cu2ZnSn(S,Se) 4 ThinFilm Solar Cells.” Journal of the American Chemical
Society 134.47 (2012): 19330-9333. Print.
6. X. L. Liu, H. T. Cui, W. Li, N. Song, F. Y. Liu, G.
Conibeer and X. J. Hao Physica Status Solidi-Rapid
4
Journal of
Emerging Investigators
Research Letters 8, (2014).
7. “Spectrum Library.” PV Lighthouse. N.p., n.d. Web. 17
Sept. 2015.
8. Phillips, Adam B., Rajendra R. Khanal, Zhaoning
Song, Rosa M. Zartman, Jonathan L. Dewitt, Jon
M. Stone, Paul J. Roland, Victor V. Plotnikov, Chad
W. Carter, John M. Stayancho, Randall J. Ellingson,
Alvin D. Compaan, and Michael J. Heben. “Wiringup Carbon Single Wall Nanotubes to Polycrystalline
Inorganic Semiconductor Thin Films: Low-Barrier,
Copper-Free Back Contact to CdTe Solar Cells.”
Nano Letters 13.11 (2013): 5224-232. Print.
9. Schulz, Philip, Anne-Marie Dowgiallo, Mengjin
Yang, Kai Zhu, Jeffrey L. Blackburn, and Joseph
J. Berry. “Charge Transfer Dynamics between
Carbon Nanotubes and Hybrid Organic Metal Halide
Perovskite Films.” The Journal of Physical Chemistry
Letters 7.3 (2016): 418-25. Print.
10.Kharroubi, B., R. Baghdad, A. Abdiche, M. Bousmaha,
M. Bousquet, A. Zeinert, M. El Marssi, K. Zellama,
and S. Hamzaoui, “Mn Doping Effect on the Structural
Properties of ZnO-nanostructured Films Deposited by
the Ultrasonic Spray Pyrolysis Method.” Physica
Scripta 86.1 (2012): 015805. Web.
11.Chu, T. L. “Solution-Grown Cadmium Sulfide Films for
Photovoltaic Devices.” Journal of The Electrochemical
Society 139.9 (1992): 2443-446. Print.
Journal of Emerging Investigators
July 19, 2016
5