Solar Energy 188 (2019) 1209–1220

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

Phase transition behavior and defect analysis of CuSbS2 thin films for T
photovoltaic application prepared by hybrid inks
⁎
Shahara Banua,b, Yunae Choa, Kihwan Kima, Seung Kyu Ahna, Jihye Gwaka, Ara Choa,b,
a
New and Renewable Energy Research Division, Photovoltaic Laboratory, Korea Institute of Energy Research (KIER), 152 Gajeong-ro, Yuseong-gu, Daejeon 305-343,
South Korea
b
University of Science and Technology (UST), 217 Gajeong-ro, Yuseong-gu, Daejeon, South Korea

A R T I C LE I N FO A B S T R A C T

Keywords: CuSbS2 (CAS) thin films were deposited via a non-vacuum hybrid ink method onto a Mo/soda lime glass (SLG)
CuSbS2 solar cell substrate. To fabricate the CAS films, Cu-Sb precursors were spin coated and then annealed with sulfur powder.
Hybrid ink During annealing, different amounts of sulfur powder were used to control the crystal orientation of the films. X-
S-flux ray diffraction (XRD) patterns were employed to examine the crystal orientation by calculating the texture co-
Phase transition behavior
efficient and Lotgering factor. It was found that the S-flux amount during sulfurization was a critical parameter
Recombination
Defects
for controlling the crystal orientation and phase transition of the CAS structure. Originally, CAS has an or-
thorhombic structure; however, if the S-flux was higher than the optimum, a pseudo-phase transition from
orthorhombic to cubic was observed. In addition, the electrical characteristics and defect properties were con-
ducted for the solar cells prepared with various S-flux to understand the difference in the photovoltaic perfor-
mances affected by the structural change. Admittance spectroscopy revealed that the defect levels were shal-
lower in the CAS solar cell with an orthorhombic structure, which could have contributed to the better
photovoltaic performance than that of the cubic structure. The CAS solar cell deposited with low S-flux exhibited
dominant VS2+ defects; however, for excessive S-flux, CuSb2− became prominent.

1. Introduction concentrations in the range of 1016–1018 cm−3 (Banu et al., 2016;

Over the decades, earth-abundant, cost-efficient, and en-
vironmentally friendly constituent absorber materials have attracted
significant interest as substitutes of commercialized chalcogenide based
Cu(InGa)Se2 (CIGS) and CdTe thin-film solar cells. For the design of
earth-abundant photovoltaics (PV) with environmentally friendly and
inexpensive constituents, a wide variety of research has been carried
out on absorber materials, such as CuSbS2 (Cho et al., 2017), Fe2S
(Berry et al., 2012; Moon et al., 2014), Sb2S3 (Chen et al., 2017), SnS
(Banu et al., 2017; Sinsermsuksakul et al., 2014), and Cu2SnS3 (Tiwari
et al., 2017), that contain earth-abundant and low-toxicity constituent
materials with simpler structures. Among these, CuSbS2 (CAS) has at-
tracted attention as a promising solar cell absorber material because it
is environmentally friendly, cost efficient, has earth abundant materials
and suitable optoelectronic properties for thin-film solar cells. It fea-
tures an appropriate band gap of 1.38–1.65 eV (Banu et al., 2016;
Welch et al., 2015a,b; Yang et al., 2014) for photovoltaic solar energy
conversion, a high absorption coefficient larger than 105 cm−1 in the
visible region (Welch et al., 2015a,b; Yu et al., 2013), tunable hole
Willian de Souza Lucas et al., 2017; Yang et al., 2014), and a hole
mobility of 0.1–1.0 cm2 V−1 s−1 (Welch et al., 2015a,b). It also shows
p-type conductivity owing to the dominant copper vacancies (VCu)
(Willian de Souza Lucas et al., 2017; Yang et al., 2014), low tempera-
ture grain growth (Yang et al., 2014), and high spectroscopic-limited
maximum efficiency (SLME) of 22.9% (Yu et al., 2013).
From the inspiration of its favorable fundamental optoelectronic
properties and promising applications, the CAS-based solar cell was
constructed first in 2005 (Rodriguez-Lazcano et al., 2005). Since then,
CAS solar cells have been demonstrated successfully via various ap-
proaches. However, so far it has been struggling with low device effi-
ciency mostly less than or near 1% (Macías et al., 2017; Wan et al.,
2016; Willian de Souza Lucas et al., 2016; Yang et al., 2014). Minimal
studies have reported greater than 2% efficiency, such as 2.55%, with a
device made by magnetron sputtering followed by a sulfurization pro-
cess (Zhang et al., 2018). In addition, a 3.13% device was prepared with
electrochemical deposition of metal stacks followed by subsequent
sulfurization (Septina et al., 2014). In previous study, we reported
3.22% conversion efficiency for a CAS device using a hybrid ink method

⁎
Corresponding author at: Photovoltaic Laboratory, Korea Institute of Energy Research (KIER), 152 Gajeong-ro, Yuseong-gu, Daejeon 305-343, South Korea.
E-mail address: icemua@kier.re.kr (A. Cho).

https://doi.org/10.1016/j.solener.2019.07.019
Received 22 April 2019; Received in revised form 10 June 2019; Accepted 6 July 2019
0038-092X/ © 2019 International Solar Energy Society. Published by Elsevier Ltd. All rights reserved.
S. Banu, et al.
in 2016, which is the highest reported efficiency until now using this
absorber (Banu et al., 2016). However, the efficiency of CAS solar cell is
still low after over a decade of research. Therefore, further optimization
of device fabrication and characterization are required for the im-
provement of device performance.
To understand the behavior of this absorber material extensively,
we systematically investigated the influence of Cu/Sb ratio to fabricate
phase pure CAS solar cell and its subsequent effects on photovoltaic
performance in another study (Banu et al., 2019). For the CIGS solar
cells, Cu/(Ga + In) (CGI) is one of the most critical and important
fabrication condition for making the absorber layer as a p-type because
the free carriers in CIGS-based materials are generated from the Cu
vacancies (VCu) (Lee et al., 2016). Similarly, the p-type conductivity of
the CAS solar cell is responsible for the free carriers induced by VCu
(Yang et al., 2014). Conversely, it is very delicate to make pure CAS
phase owing to the presence of other stable binary and ternary phases
of Cu-Sb-S compounds, such as Sb2S3, Cu3SbS3, Cu3SbS4, and
Cu12Sb4S13 (Banu et al., 2019; Cho et al., 2017; Yu et al., 2013).
Nevertheless, CuSbS2 single phase was fabricated successfully by con-
trolling the Cu/Sb ratio without the formation of other unexpected
phases (Banu et al., 2019).
Traditionally, characterization or analysis can help to understand
the limitation of solar cell performance and provide further information
for methods to improve the efficiency of the cell. Hence, the Cu/Sb ratio
was first optimized to create an appropriate amount of VCu, and then,
the recombination mechanism and defect properties were investigated
for the optimum Cu/Sb-ratio device. From these results, deep defects in
the absorber at 0.23 eV above the valance band maximum were ob-
served, which were responsible for the CuSb antisite defects. It was
suggested that the efficiency was still low due to the CuSb defects even
though VCu was optimized. Therefore, for further improvement of the
cell performance, the CuSb defects need to be scaled down.
To reduce the CuSb defects, different other conditions were tried.
Based on the reports of the CZTS solar cell, self p-type doping in CZTS is
related to VCu with CuZn antisite defects (Banu et al., 2019), and the
energy level of CuZn can be changed by the S concentration owing to the
strong hybridization between S 3p and Cu 3d (Duan et al., 2013).
Therefore, it is predicted that the deep defects level could be controlled
by changing the S-flux during the preparation of the CAS solar cells.
Thus, studying the role of sulfur flux on the CAS solar cells could be
fruitful to fabricate the device with fewer defects.
In this regard, we report a comparative study of CAS solar cells
prepared with various S-flux using hybrid ink with an optimized Cu/Sb
ratio. However, while we were trying to change S-flux, surprisingly a
phase transition behavior was observed in the CAS solar cells. This
pseudo-phase transition could be related to the preferred orientation of
CIGS and will be discussed in detail. In addition, the cell characteristics
deduced from the current density-voltage and external quantum effi-
ciency measurements and electrical defect properties from the capaci-
tance based measurements, including capacitance-voltage (C-V), ad-
mittance spectroscopy (AS), and drive-level capacitance profiling
(DLCP), were investigated to correlate with the phase transition beha-
vior.
2. Experimental section
2.1. Reagents
Copper (I) acetate, C2H3CuO2 (Cu(I)Ac, 97%), antimony (III)
acetate, C6H9O6Sb (Sb(Ac)3, 99.99%), monoethanolamine,
NH2CH2CH2OH (MEA, 99.0%), methanol, CH3OH (MeOH, 99.8%), and
sulfur powder (99.98%,) were purchased from Sigma-Aldrich. All che-
micals were used as received without further purification and stored in
a nitrogen-filled glove box to restrain from the degradation caused by
humidity or air. Mo-coated soda lime glasses were fabricated by a DC
magnetron sputtering system with a thickness of approximately 1 μm
1210
Solar Energy 188 (2019) 1209–1220
and sheet resistance of 0.18 Ω/□.
2.2. Preparation of the hybrid inks
The hybrid ink was prepared using Cu(I)Ac, Sb(Ac)3, MeOH, and
MEA. In a glove box, Cu(I)Ac and Sb(Ac)3 each were mixed with MEA
and MeOH in 10-mL vials separately and were sealed with the lids.
Then, both mixtures were taken out from the glove box and sonicated
separately in an ultrasonicator (Branson 2510) for approximately
30 min at room temperature to obtain complete dissolution. The Sb and
Cu precursor solution colors were clear and greenish-blue, respectively.
After the dissolution, both vials were returned to the glove box and
blended together. The molar ratio of Cu(I)Ac to Sb(Ac)3 was fixed at
1:1.2, which optimized in our previous study to make phase pure
CuSbS2 (Banu et al., 2019). The molar ratio of both Cu(I)Ac and Sb(Ac)3
with MEA was 1:6 to achieve adequate coordination between the MEA
and precursors (Cu and Sb). The weight ratio of Cu(I)Ac and Sb(Ac)3
with MeOH was 1:5 to adjust the viscosity of the precursor ink.
2.3. Thin film deposition
The precursor inks were coated onto Mo-coated (1 μm) soda lime
glass substrates by spin coating (ACE-200, Dong Ah Tech., Korea) at
2000 rpm for 20 s. Then, the samples were dried immediately on a hot
plate at 80 °C, 120 °C, and then 200 °C for 5 min each in air to evaporate
the solvent and induce thermal decomposition of MEA. The spin coating
and drying processes were repeated four more times to obtain the re-
quired thickness. The film deposition procedures were performed in
ambient air.
2.4. Sulfurization
The precursor films were subsequently sulfurized in a rapid thermal
annealing (RTA) system in which a graphite container was inserted
inside. The sulfurization chamber was equipped with a temperature,
time, and pressure controller. The precursor samples were placed inside
the graphite container along with the sulfur powder, which was then
placed inside the RTA heat treatment chamber. The setup of the RTA
sulfurization equipment is shown in reference Cho et al. (2013a). In this
study, the S amount (0.1–0.5 g) was chosen as main experimental
variables to control the S-flux inside the graphite container. For all
cases, sulfurization was performed at 450 °C for 30 min, which has
previously reported optimum growth conditions (Banu et al., 2016).
After loading the samples, the chamber was initially evacuated to a low
vacuum pressure of 10−3 Torr using a rotary pump. Then, the back-
ground pressure was regulated to 1.02 atm (1 atm + 1/3 psi) by in-
jecting nitrogen gas, and the time used to increase the temperature
between the set points was fixed to 5 min.
2.5. Device fabrication
The solar cells were fabricated according to the conventional Mo/
CuSbS2/CdS/i-ZnO/ITO/Al structure. A 60-nm CdS buffer layer was
deposited onto the sulfurized films via chemical bath deposition (CBD),
as reported previously (Lee et al., 2015). After depositing the buffer
layer, intrinsic ZnO (~50 nm) and indium tin oxide (~150 nm) window
layers were deposited by radio frequency (RF) sputtering. Finally, an Al
grid of a thickness of approximately 800 nm was deposited as a current
collector using thermal evaporation. The active area of the completed
cells was 0.2225 cm2.
2.6. Characterization and analysis
The chemical composition of the sulfurized films was analyzed using
energy dispersive X-ray spectroscopy (EDS; EDAX Genesis apex, accel-
erating voltage of 30 kV, collection time of 100 s). The roughness and
S. Banu, et al.
morphology of the surface were investigated using atomic force mi-
croscopy (AFM, Park System XE-100, 0.6 Hz scan rate) and high-re-
solution scanning electron microscopy (HRSEM; XL30SFEG Phillips Co.,
Holland at 10 kV), respectively. The crystal structure of the films was
identified using X-ray diffraction spectroscopy (XRD; Rigaku Japan, D/
MAX-2500) with a Cu Kα radiation source (λ = 1.5406 Å). The current-
voltage characteristics of the devices were measured using a source
meter (K2450, Keithley) under irradiation of simulated AM 1.5 sunlight
(K201, Mcscience) with a calibration standard (PVM 524, PV
Measurement Inc.). The external quantum efficiency (EQE) was de-
termined using an incident photon-to-electron conversion efficiency
(IPCE) measurement unit (PV Measurements, Inc., USA). The capaci-
tance-voltage (C-V) measurements, admittance spectroscopy (AS), and
drive-level capacitance profile (DLCP) measurements of the fabricated
cells were performed with an LCR meter (Agilent 4284A, Keysight). The
temperature of the cells was controlled by cooling (by flowing liquid
N2) or heating using a temperature controller (LTSE350-P, Linkam).
3. Results and discussion
Single phase CAS thin-film solar cells were prepared via a non-va-
cuum hybrid ink method. The hybrid ink consisted of metal precursors
(Cu(I)Ac and Sb(Ac)3), chelating agents (MEA), and a solvent (MeOH).
In the hybrid ink, the chelating agents and metal precursors make a
metal-chelate complex that controls the reaction rate for the formation
of dense thin layers (Cho et al., 2014, 2013a,b,c, 2012). Using this
advantage of the hybrid ink, dense CAS thin films were successfully
prepared, and the influences of the S-flux on the formation of CAS thin
film were investigated.
As mentioned before, in this study, the main experimental variable
is the S-flux, which is controlled by various S-amounts during the sul-
furizations (thermal treatment). To date, numerous studies have been
reported regarding the thermal treatment under different conditions,
such as H2S (Septina et al., 2014) gas and Sb2S3 vapor (Chen et al.,
2015; Saragih et al., 2017; Willian de Souza Lucas et al., 2016) to im-
prove the film quality. However, it is notable that thermal treatments
were operated in vacuum condition for all the above-mentioned studies.
It is worth mentioning that during sulfurization, S-flux and S-pressure
both are important to convert Cu and Sb precursors to CAS thin films
using our method. Hence, the main factor to change the S-flux was the
variation of the S-amounts with ambient S-pressure during the thermal
heat treatment (sulfurizations).
3.1. Morphology and structure
To determine the compositions of the films, an EDS analysis of the
sulfurized films was performed. It revealed that there was not a notable
change of the atomic ratios of Cu/Sb (0.76–0.83) with the variation of
the S-flux. An AFM analysis was performed to examine the surface
texture and roughness (Ra) of the sulfurized CAS thin films prepared
with various S-flux and is shown in Fig. 1. It was inspected from the
surface texture shown in Fig. 1(a)–(e), that the grain size increased and
expanded with the addition of S-flux. The films grown with 0.1–0.4 g S-
flux showed a similar surface texture in terms of grains regardless of the
size, whereas, after an additional increase of S-flux, no distinct grains
were observed due to the decomposition of the grains, as the grains
become larger coalescence of the grains was occurred. This decom-
position of grains for the films prepared with 0.5 g S-flux (Fig. 1(f))
could be explained by the excess S-flux which might affect the structure
of the absorber during the sulfurization processes. Moreover, this
structural change could initiate voids on the surface of the films. From
Fig. 1(f), it was observed that the Ra value was approximately the same
or slightly increased up to 0.3 g S-flux. However, a significant increase
of the Ra value occurred for the 0.4 g S-flux films and afterwards it
becomes steady. This indicated an increase of the surface roughness due
to the bigger grains and larger variation of grain heights. Although the
1211
Solar Energy 188 (2019) 1209–1220
Ra values for the 0.4 and 0.5 g films were similar, the AFM images of
them (Fig. 1(d) and (e)) were quite different. Therefore, other surface
analytical methods were performed to distinguish this difference.
Along with AFM, surface SEM images were also measured for the
better understanding of the films morphology. Fig. 2 shows the surface
morphologies of thin films before and after sulfurization with different
amounts of S-flux. As-deposited film (Fig. 2(a)) exhibited very dense
and compact amorphous morphology without voids and cracks. In ad-
dition, after sulfurization, all the films prepared in different S-flux en-
vironments showed micron-size grains, and as the S-flux increased, the
grains became bigger due to the acceleration of S diffusion into the
precursor films. From the SEM images, it was observed that the film
prepared with 0.1 g S-flux (Fig. 2(b)) showed pinholes, whereas upon
increasing the S-flux from 0.2 to 0.4 g, the general morphological
properties of the films such as the compactness and grain size of the
films, were found to be improved without having any pinholes or voids
(Fig. 2(c)–(e)). However, an excessive addition of S-flux (0.5 g) dis-
rupted the formation of dense film, resulting in a film with a notable
amount of voids (Fig. 2(f)), which was consistent with the corre-
sponding AFM results. Therefore, the SEM analysis also showed a si-
milar trend with the AFM images of the films prepared with 0.4 and
0.5 g S-flux.
To investigate the effect of S-flux on the crystalline behavior of the
CAS thin films, X-ray diffraction was carried out, and the corresponding
patters are shown in Fig. 3. XRD patterns of all sulfurized films showed
only the orthorhombic CuSbS2 (JCPDS card no. 44-1417) phase. Among
the XRD peaks of orthorhombic CuSbS2 phase, the signature peaks
known as (1 1 1), (4 1 0), and (3 0 1) are located at approximately
28–30°, shown as a color shade in Fig. 3. For the CAS thin films pre-
pared with an S-flux of approximately 0.1–0.4 g, the conventional
(1 1 1), (4 1 0), and (3 0 1) preferred orientations were detected. How-
ever, as the S-flux increased from 0.4 g to 0.5 g, the intensities of typical
(1 1 1), (4 1 0), and (3 0 1) diffraction peaks decreased significantly, and
the intensities of the other orientation peaks, such as (2 0 0), (4 0 0),
and (8 0 0) became more prominent. These XRD results also showed
good agreement with the AFM topography and SEM micrographs be-
cause the surface roughness and morphology for 0.5 g (Figs. 1 and 2)
were different from the others. For simplicity, the (1 1 1), (4 1 0), and
(3 0 1) preferred orientations were denoted as (111/301) and the or-
ientations of (2 0 0), (4 0 0) and (8 0 0) were denoted as (2 0 0) in this
study.
For CuInSe2 (CIS) and CIGS, two different preferred orientations,
(1 1 2) and (2 2 0)/(2 0 4), have been already reported (Cho et al.,
2013b; Kim et al., 2018; Rudmann et al., 2003), and the preferred or-
ientation in CIS and CIGS was controlled by regulating the Se-flux
(Chaisitsak et al., 2002; Gwak et al., 2015; Hanna et al., 2003). These
reports relevant with our study, as illustrated in Fig. 3. Two commonly
used calculation methods of the preferred orientation, such as the
Lotgering factor (LF) and texture co-efficient (TC), were used to quan-
tify the degree of preferred orientation.
To calculate LF and TC, XRD intensities of (hkl) plane with 2θ/θ
mode were used, and the representative graphs are shown in Fig. 4. The
LF and TC are defined by the following equations.
dV AkT ⎡ 1 ⎤
= Rs +
dJ q ⎢⎣ J − JSC ⎥
⎦ (1)
where P(hkl) is the ratio of integrated intensities of (hkl) reflection to
the sum of the reflections in the scanned range, P0(hkl) represents the
corresponding ratio for a randomly orientated CAS phase (JCPDS card
no. 44-1417) (Lotgering, 1959), and
I(hkl)
I0(hkl)
TC(hkl) =
1 n I(hkl)
n
∑i = 1 I0(hkl) (2)
where I(hkl) is the measured intensity of (hkl) plane, I0(hkl) is the
S. Banu, et al.
Fig. 1. AFM images of the sulfurized thin films prepared with different S-flux: (a) 0.1 g, (b) 0.2 g, (c) 0.3 g, (d) 0.4 g, and (e) 0.5 g, and (f) variations of surface average
roughness, Ra with different S-flux.
Fig. 2. SEM images of the surface morphology of CAS thin films (a) as-deposited and grown with different S-flux: (b) 0.1 g, (c) 0.2 g, (d) 0.3 g, (e) 0.4 g, and (f) 0.5 g.
intensity of standard CAS phase, and n is the number of (hkl) diffraction
peaks (Wang et al., 2014a,b). In this study, six influential peaks of
(2 0 0), (4 0 0), (1 1 1), (4 1 0), (3 0 1), and (8 0 0) were used for the
calculations. Fig. 4(a) and (b) shows the LF and CF values of the CAS
thin film as a function of S-flux. From Fig. 4(a) and (b), the LF and TC
for (1 1 1) and (3 0 1) decreased; however, that for (2 0 0) increased
with the addition of S-flux. From Eqs. (1) and (2), it is inferred that
when LF and TC are equal to 0 and 1, respectively, the films are entirely
non-oriented with the standard data. Conversely, the films with a
higher deviation from 0 and 1 means highly oriented to other planes
than the standard one. Therefore, Fig. 4(a) and (b) illustrate that the
film prepared with 0.4 g S-flux results the best match with the standard
data having (111/301) preferred orientations. On the other hand, the
film prepared with a 0.5 g S-flux was highly oriented to the (2 0 0)
plane.
1212
Solar Energy 188 (2019) 1209–1220
Based on these calculations and the morphology from AFM and
SEM, this difference could affect to the structure of CAS; therefore, the
lattice parameters of the films were calculated using the three strongest
peaks for each film XRD and presented in Fig. 4(c). The CAS thin films
prepared with S-flux up to 0.4 g showed the orthorhombic crystal
structure as usual. However, with the addition of more S (> 0.4 g),
surprisingly a phase transition behavior was observed, which seemed to
be a cubic structure from orthorhombic. Hence, this phase transition
behavior was defined as a “pseudo”-phase transition because the phase
was not entirely transformed to cubic and still there were small sig-
nature peaks from the orthorhombic phase. This pseudo-phase transi-
tion occurred because of the volume expansion of the unit cell resulting
from the allocation of the excess S atom.
Therefore, from the above discussion, it is attributed that a single-
phase CAS could be formed for an optimum ink composition and
S. Banu, et al.
Fig. 3. X-ray diffraction pattern of CAS thin films grown with different S-flux. A
reference pattern of orthorhombic CuSbS2 (PDF#44-1417) is included at the
bottom.
sulfurized condition, such as temperature and pressure, regardless of
the S-flux. However, the preferred orientation of CAS could be regu-
lated by the S-flux, similar to CIS/CIGS system, where Se-flux plays
critical role to control the orientations. It is worth noting that the effect
Fig. 4. (a) LF, (b) CF for (1 1 1), (3 0 1), and (2 0 0) planes of the CAS thin films as a function of S-flux, and (c) evolution of the unit cell parameters with S-flux in CAS
thin films during the sulfurizations.
1213
Solar Energy 188 (2019) 1209–1220
of orientation in CAS was more severe than that in CIS/CIGS because
the phase transition behavior was observed in excess S-flux condition.
Briefly, to make pure orthorhombic CAS phase, 0.4 g S-flux was the
optimum and a pseudo-phase transition can be occurred when the S-
flux exceed this limit.
3.2. Device performance and temperature-dependent PV characterizations
CAS devices were fabricated by depositing CdS/i-ZnO/ITO/Al layers
on the CuSbS2 absorbers grown with various S-flux. Fig. 5 shows the
illuminated current density-voltage (JV) and external quantum effi-
ciency (EQE) curves for the devices with different amounts of S-flux,
with the calculated band gap (Eg) from the corresponding EQE shown in
the inset. The device parameters extracted from the JV measurements
are summarized in Table 1. Upon increasing the S-flux, the device
parameters, such as JSC, VOC, and η, were increased and the FF was
almost constant up to the optimum amount of S-flux (to make orthor-
hombic CAS phase) due to the improved carrier collection, structure,
and morphology. However, when the amount of S-flux exceeded the
optimum level, excessive S-flux negatively affected the device perfor-
mance owing to the disrupted morphology of the film. In particular, the
variation of JSC was more prominent in the JV curves, as shown in
Table 1. The change of JSC was further analyzed by EQE curves shown
in Fig. 5(b). EQE also followed the same trend as the device
S. Banu, et al. Solar Energy 188 (2019) 1209–1220

Fig. 5. (a) Light JV and (b) EQE curves of CAS devices with various S-flux. The inset shows the calculated band gap from the corresponding EQE.

Table 1 performance, i.e., was improved until an optimum S-flux of 0.4 g which
Device parameters for CAS solar cells at room temperature with various S-flux. reflected a better carrier collection through the CAS absorber layer, and
S-flux (g) VOC (V) JSC (mA/cm2) FF (%) η (%) when the S-flux was greater than the optimum, the EQE decreased
drastically. In addition, the optical band gaps of the absorber layers
0.1 0.27 2.33 41.71 0.26 were also deduced from the long-wavelength cutoff of the EQE curves
0.2 0.35 3.45 42.42 0.51
for all devices, as shown in inset of Fig. 5(b). The band gaps of all
0.3 0.35 5.38 39.62 0.74
0.4 0.38 9.86 40.85 1.53
devices were calculated from [hν × ln (1 − EQE)]2 vs. hν plots and
0.5 0.29 3.59 34.61 0.36 presented in Table 2. The calculated band gaps for all devices were
approximately the same and perfectly matched with the band gap of
CuSbS2 (Banu et al., 2019, 2016). To verify these influences, electrical
Table 2 characteristics and defect properties of the absorber layer were eval-
Calculated band gap, Eg from EQE, barrier height, Фb from the fitting of ex- uated in the next section.
ponential (RsT vs. 1000/T) plot, hopping energy, Δ by fitting the exponential To obtain additional insight into the origin of the difference in the
temperature dependent JSC curve, and activation energy, Ea at 0 K from VOC vs. device performance of CAS thin-film solar cells with the variation of the
T curve, of CAS solar cells with different S-flux. S-flux, the carrier transport mechanisms of the devices were analyzed.
S-flux (g) Eg (eV) Фb (meV) Δ (meV) Ea (eV) In this regard, temperature dependent current density-voltage (JV-T)
measurements were performed for all devices in the temperature range
0.1 1.56 209 102.61 0.61
of T = 100–300 K.
0.2 1.55 155 103.03 0.84
0.3 1.54 147 129.12 0.86 It was discussed in the previous section that the variation of η values
0.4 1.53 143 165.45 0.87 for the solar cells made with different S-flux was mainly observed due
0.5 1.57 147 142.65 0.66 to the larger variation of JSC, which gave the insight on the carrier
collection problem. To further elucidate this issue, temperature-de-
pendent JSC and η values were determined from the JV-T measurements
and are shown in Fig. 6. It was noticed from Fig. 6 that JSC and η values

Fig. 6. (a) JSC and (b) η values obtained from the CAS solar cells for various S-flux under different temperatures.

1214
S. Banu, et al.
Fig. 7. Temperature dependence of the dark series resistance, RS, for the sulfurized thin films prepared with different S-flux: (a) 0.1 g, (b) 0.2 g, (c) 0.3 g, (d) 0.4 g,
and (e) 0.5 g. The inset shows the corresponding back-contact barrier height fitting.
of all studied devices were collapsing toward zero at lower temperature.
This rapid declination of JSC as well as η, particularly in the low tem-
perature region, could be explained by the fast recombination of the
photo-generated charges before they were collected (Banu et al., 2019;
Lee et al., 2016).
Generally, the limitations of JSC due to fast recombination are
negligible for device containing shallow defects states (close to the
valence or conduction band) because of easy capture and release of
carriers by defect sites. However, if the device contains deep defect
states located in the middle of the valence and conduction band, higher
1215
Solar Energy 188 (2019) 1209–1220
temperatures are required to activate the defects, and these high-energy
defects can act as recombination centers (Lee et al., 2016). Therefore,
deep defects can limit JSC owing to the fast recombination rate. This
result will be further supported by AS analysis in the next section.
On the other hand, using a standard diode analysis of a thin film
solar cell, series resistance, RS was extracted from the y-intercept of the
dV/dJ vs. 1/J plot as a function of temperature (Fig. 7),
dV AkT 1
dJ
= Rs + ⎡ ⎤
q ⎣ J − JSC ⎦
(3)where q is the charge of an electron, V is
the applied bias, J is the current density, A is the diode ideality factor, k
S. Banu, et al.
is Boltzmanńs constant and T is temperature (Hegedus and Shafarman,
2004; Sites and Mauk, 1989). From Fig. 7, increasing behaviors of series
resistance toward a low temperature were observed. This increasing
behavior of series resistance toward low temperature could be ex-
plained by the back contact barrier, which could block the carriers to
transport at the interface of Mo and CAS. Hence, back contact barrier
height, ɸB, were extracted from the temperature dependent dark series
resistance using the following equation:
k ϕ
Rs = R 0 + exp ⎛ B ⎞
⎜ ⎟
qA* T kT
⎝ ⎠
where A* is the effective Richardson constant and R0 is the background
series resistance resulting from the contact grid and TCO (Gunawan
et al., 2010; Wang et al., 2010). Corresponding fitting curve for ɸB are
shown in the inset of Fig. 7, and the values are summarized in Table 2.
When the S-flux increased up to an optimum level, ɸB decreased due to
the improved morphology and crystallinity. However, when the S-flux
increased more than the optimum level, it decreased again, which could
be the results of structural change in the film due to excessive amount
of S. Around room temperature, the effect of ɸB at the interface of Mo
and CAS was not noticeable. However, as the temperature gradually
decreased, the carrier transport was more affected by ɸB. Thus, it can be
asserted that the dramatic change of RS depending on temperature was
another reason for rapid declination of JSC with temperature. At room
temperature, there was no problem regarding solar cell operation be-
cause carrier could overcome that barrier even though S-flux increased.
However, as temperature decreased, recombination rate was influenced
by the S-flux; therefore, ɸB was different for various S-flux.
Furthermore, such a strong temperature dependence of JSC also can
be explained by hopping-type charge-carrier transport in the device
(Banu et al., 2019). Hopping energy, Δ is the energy difference between
the localized defect states. Therefore, high value of Δ represents fewer
defect states in the bulk of the device, resulting in a lower re-
combination. In contrast, a lower value of Δ indicates larger defect
states in the absorber layer, which could accelerate the recombination
rate. Thus, to determine the source of the recombination loss as a result
of the higher variation of JSC, Δ was deduced for all devices made with
different S-flux by fitting the exponential temperature-dependent JSC
curve shown in Fig. 8(a) using the following equation:
Δ with different S-flux was found, and the values were in the range of the
JSC\; = \;J00 \;exp ⎛−\; ⎞
⎝ kT ⎠
where Joo is the saturation current density prefactor. The values of Δ are
presented in Table 2. It is notable that with the addition of S-flux up to
0.4 g, i.e., until the optimum level for this deposition method, Δ
Fig. 8. (a) JSC vs. 1000/T curves to calculate the hopping energy, Δ, and (b) VOC vs. T curves and its linear extrapolation to 0 K indicating the activation energy, Ea of
the recombination process in CAS solar cells with different S-flux.
Solar Energy 188 (2019) 1209–1220
increased gradually; however, for a further increase of S-flux, Δ started
to decrease. Therefore, the hopping energy of the device made with
0.4 g S-flux was significantly higher than that of the other devices. This
results highlighted the existence of fewer defect states and less re-
combination loss for the 0.4 g S-flux device, resulting in a higher JSC
than that of the other S-flux.
To elucidate the effect of S-flux on recombination behavior, tem-
perature-dependent VOC of the CAS solar cells for various S-flux were
performed and compared (Fig. 8(b)). Usually, the activation energy, Ea
extracted from the temperature-dependent VOC provides information on
(4) the dominant recombination process in the device. Ea was determined
by extrapolating a plot of VOC vs. T at T = 0 K using the following
equation and presented in Table 2.
Ea AkT ⎛ J00 ⎞
VOC = − ln ⎜ ⎟
q q ⎝ JL ⎠ (6)
where JL is the photo-generated current density (Cao et al., 2011). For
all devices, the Ea values (Table 2) were significantly lower than their
respective bandgap values, suggesting that the recombination me-
chanism in these CAS cells was strongly influenced by interface re-
combination (Nadenau et al., 2000; Riedel et al., 2004; Wang et al.,
2014a,b), which could also cause the low VOC of the devices at room
temperature.
3.3. Defect properties
To evaluate the quality of the formed p-n junctions in the fabricated
CAS devices, a room temperature C-V analysis was performed. Fig. 9
shows the Mott-Schottky plots of the CAS devices with different S-flux,
exhibiting common p-type characteristics (negative values of the
slopes) of the films. The net carrier density, NC-V of the devices was
calculated from the slopes of the 1/C2–V plots using the following
equation:
C 3 ⎛ dC ⎞−1
NC − V =
qεε0 A2 ⎝ dV ⎠ (7)
where ε is the dielectric constant of CAS, ε0 is the dielectric constant of
vacuum, A is the sample area, and C is the capacitance. Corresponding
data listed in Table 3. No notable difference of NC-V among the devices
previously reported value (Banu et al., 2019). In general, the room
(5)
temperature C-V analysis provides information regarding the overall
carrier density, including interface and bulk. Therefore, it is unable to
distinguish between the interface and bulk defects. However, it is al-
ready verified from the Ea values of temperature dependent VOC
1216
S. Banu, et al. Solar Energy 188 (2019) 1209–1220

temperature range of 100–300 K, as shown in Fig. 10(a)–(e). The ca-

pacitance-frequency (C-F) spectra of the devices were obtained in the
dark at a zero bias from 100 Hz to 1 MHz. Usually, in AS, the low fre-
quency capacitance provides information for the deep traps and free
carriers densities, while the high-frequency capacitance is responsible
only for the free carrier density. From Fig. 10(a)–(e), it is noticed that
the capacitance of the device with optimum S-flux (Fig. (d)) showed a
smaller deviation as a function of frequency than those of the others,
indicating smaller trap densities in the absorber.
Fig. 10(f) shows an Arrhenius plot for the inflection point ω0 of the
C-F plot (Fig. 10(a)–(e)), where, ω0 was determined from the peak of
the – dC/d(ln(ω)) vs. ω plots. The defect activation energy, Et can be
calculated from the Arrhenius plot of ln(ω0/T2), using the following
equation (Duan et al., 2013; Kim et al., 2018; Chen et al., 2011):

E
ω0 = 2πυ0 T 2 exp ⎛− t ⎞
⎝ kT ⎠ (8)

where υ0 is the emission factor. The activation energy, Et, for the de-
Fig. 9. 1/C2 versus V curve for various S-flux.
vices prepared with 0.1, 0.2, 0.3, 0.4, and 0.5 g of S-flux were estimated
as 0.187, 0.186, 0.195, 0.202, and 0.237 eV, respectively, confirming
Table 3 the presence of deep defects. Comparing the Et values, it was observed
Carrier densities derived from C-V and DLCP for the CAS solar cells with dif- that the value increased with the addition of S-flux. A similar trend also
ferent S-flux.
was observed in CZTSSe solar cells, where it was reported that the
S-flux (g) NC-V (cm−3) NB (cm−3) NIF (cm−3) defect ionization energies increased as the S-concentration increased,
resulting in the down shifting of the valance band (Chen et al., 2011;
0.1 2.69 × 1017 7.97 × 1016 1.89 × 1017
0.2 2.10 × 1017 3.88 × 1016 1.71 × 1017
Duan et al., 2013). On the other hand, Willian de Souza Lucas et al.
0.3 1.73 × 1017 4.00 × 1016 1.33 × 1017 reported two deep defect states in CAS absorber at 0.17 eV below the
0.4 1.66 × 1017 2.94 × 1016 1.37 × 1017 conduction band minimum (CBM) and 0.24 eV above the valance band
0.5 1.82 × 1017 5.81 × 1016 1.29 × 1017 maximum (VBM), which were related to the VS donors and CuSb ac-
ceptors, respectively (Willian de Souza Lucas et al., 2017). Further-
more, according to Zhang et al. and Wei et al., the activation energy of
(Fig. 8(b)) showed that the interface defects were dominant. Thus, in
0.202 eV could be responsible for the VSb defects above the VBM or the
order to identify the defect types, an additional analytical method, such
Cui defects below the CBM (Wei et al., 1998; Zhang et al., 1998).
as DLCP, was required and will be discussed in next section (Heath
However, EDS value proved that all the films were Cu-poor; therefore,
et al., 2004, 2003).
the Cui defects were excluded in this study. Thus, it can be declared that
An AS analysis of all CAS devices was performed to estimate the
0.202 eV for optimum S-flux device was related to the VSb defects.
defect energy level inside the band gap of solar cells over the
Usually, defect levels below the CBM are referred to as n-type defects,

Fig. 10. Admittance spectra (AS) of CAS solar cells with different S-flux: (a) 0.1 g, (b) 0.2 g, (c) 0.3 g, (d) 0.4 g, and (e) 0.5 g measured in the temperature range of
300–100 K, (f) corresponding Arrhenius plot of the inflection point of the capacitance function calculated from the derivative of the AS.

1217
S. Banu, et al. Solar Energy 188 (2019) 1209–1220

Fig. 11. Deep level response of CAS solar cell device with different S-flux: (a) 0.1 g, (b) 0.2 g, (c) 0.3 g, (d) 0.4 g, and (e) 0.5 g measured at the temperature range
140–300 K.

and those above the VBM are referred to as p-type defects. Concisely,
from the AS results, it was estimated that the low S-flux mainly con-
tained n-type (VS) defects, and with the increment of S-flux up to the
optimum (0.4 g), n-type (VS) defects were gradually suppressed by S-
flux. However, as the S-flux increased more than the optimum, another
p-type defects CuSb, which were deeper than VSb, were prominent.
Therefore, although the films up to 0.4 g (optimum) contained an or-
thorhombic structure and a similar morphology regardless of the grain
size, the JSC and η values increased owing to the reduction of the n-type
defects VS to p-type defects VSb. Nevertheless, for an additional incre-
ment of the S-flux, the JSC and η values decreased due to the structural
change and existence of deeper defects. Moreover, it is also confirmed
that the recombination rate was affected by the existing deep defects in
the devices, and, as a results of that, JSC decreased with the decline of
temperature.
To identify the different types of defects (bulk/interface defects, free
carriers, and defect densities), DLCP measurements were performed
using the technique introduced by Heath et al. (2004). DLCP profile of
all the CAS devices using a fixed frequency of 10 kHz as a function of
temperature are shown in Fig. 11 and presented in Table 3. In order to
distinguish the bulk and interface defects, the results of room tem-
perature DLCP profile were compared with C-V data for all the devices.
As it is mentioned that C-V data provides the overall information for the
devices, including the interface and bulk states. However, DLCP is more
sensitive to bulk states. Thus, the defect densities at the interface (NIF)
were calculated by subtracting the DLCP data (NB) from the C-V data
(NC-V) at room temperature, as listed in Table 3. The NIF values of all the
devices were an order of magnitude larger than those of NB. Therefore,
it is demonstrated that interface recombination was dominant for all
CAS solar cells, which was in good agreement with the Ea values from
the VOC vs. T plot.
Besides, to further elucidate the effect of S-flux on the bulk prop-
erties of the films, free carrier (P) and defect densities (NT) of the de-
vices were derived (Table 4). Traditionally, the carrier density (NDL)
extracted from a low temperature DLCP profile represents the free
carrier density (P), whereas a high temperature NDL indicates the
Table 4
Free carrier densities (P) and defect densities (NT) derived from the DLCP of
CAS solar cells with different S-flux.
S-flux (g) P (cm−3) NT (cm−3)
0.1 5.03 × 1015 7.47 × 1016
0.2 4.45 × 1015 3.43 × 1016
0.3 3.86 × 1015 3.61 × 1016
0.4 6.73 × 1015 2.26 × 1016
0.5 8.65 × 1015 4.94 × 1016
response from both free carriers and deep defects (NB). Thus, the defect
density (NT) in the absorber could be extracted by subtracting the NDL
values at a high temperature (NB) from a low temperature (P) (Duan
et al., 2013; Heath et al., 2004). As presented in Table 4, the influence
of S-flux on free carrier densities (P) was insignificant. However, the
defect densities (NT) of the devices, which were derived from the dif-
ference between the NDL values at T = 300 K (NB in Table 3) and 140 K
(P in Table 4), were found to decrease with the increment of S-flux up to
0.4 g (optimum). Moreover, as the S-flux increased more (> optimum),
NT increased. Briefly, the device with optimum S-flux contained the
lowest NT in the bulk than those of the other devices, which showed
good agreement with the hopping energy, C-F plot (Fig. 10(a)–(e)), and
device performance.
From the above discussion, it can be interpreted that sufficient S-
flux is beneficial to reduce the n-type defects. However, an excessive S-
flux could cause structural change in the films, which could negatively
affect the device performance. Therefore, the optimum S-flux is needed
to fabricate orthorhombic CAS thin films with improved device per-
formance. However, the optimum S-flux device also showed a dominant
interface recombination. Thus, reducing the interface recombination
could be future steps to further improve the CAS device performance.
4. Conclusions
CAS solar cells were fabricated for various S-flux via a non-vacuum

1218
S. Banu, et al.
hybrid ink method and subsequent sulfurization process using the op-
timum ink composition and sulfurization condition (temperature, time,
and pressure). The effect of S-flux on the optoelectronic and defect
properties of CAS solar cells was investigated by the aim to reduce the
defect energy levels. However, with the change of S-flux, surprisingly a
phase transition behavior was observed. The films prepared until 0.4 g
(optimum) S-flux showed an orthorhombic structure with improved
grains, whereas a pseudo-phase transition from an orthorhombic to a
cubic structure was observed for the films prepared with excessive S-
flux. In addition, below the optimum S-flux, photovoltaic properties,
specifically JSC, were limited owing to the morphology with smaller
grains. However, for an excessive S-flux, JSC and other device para-
meters were mainly affected by the pseudo-phase transition. From the
results of the recombination mechanism and defect properties, it was
confirmed that all devices showed a dominant interface recombination.
Moreover, the CAS solar cell deposited with a low S-flux exhibited
dominant n-type VS defects, whereas the device containing excessive S-
flux showed a deeper defect energy level with dominant CuSb defects.
Therefore, controlling the S-flux is important for fabricating orthor-
hombic CAS thin films as well as improving the photovoltaic perfor-
mance by reducing the defects in the bulk state.
Acknowledgements
This study was conducted under the framework of the Research and
Development Program of the Korea Institute of Energy Research (KIER)
(B9-2411-01). This work was also supported by the Korea Institute of
Energy Technology Evaluation and Planning (KETEP), granded fi-
nancial resource from the Ministry of Trade, Industry & Energy
(MOTIE) of the Republic of Korea (No. 20173010012980,
20163010012430).
Appendix A. Supplementary material
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.solener.2019.07.019.
References
Banu, S., Ahn, S.J., Ahn, S.K., Yoon, K., Cho, A., 2016. Fabrication and characterization of
cost-efficient CuSbS2 thin film solar cells using hybrid inks. Sol. Energy Mater. Sol.
Cells 151, 14–23.
Banu, S., Ahn, S.J., Eo, Y.J., Gwak, J., Cho, A., 2017. Tin monosulfide (SnS) thin films
grown by liquid-phase deposition. Sol. Energy 145, 33–41.
Banu, S., Cho, Y., Kim, K., Kyu, S., Sik, J., Gwak, J., 2019. Solar energy materials and solar
cells effect of Cu content in CuSbS2 thin films using hybrid inks: their photovoltaic
properties and defect characteristics. Sol. Energy Mater. 189, 214–223.
Berry, N., Cheng, M., Perkins, C.L., Limpinsel, M., Hemminger, J.C., Law, M., 2012.
Atmospheric-pressure chemical vapor deposition of iron pyrite thin films. Adv.
Energy Mater. 2, 1124–1135.
Cao, Q., Gunawan, O., Copel, M., Reuter, K.B., Chey, S.J., Deline, V.R., Mitzi, D.B., 2011.
Defects in Cu(In, Ga)Se2 chalcopyrite semiconductors: a comparative study of ma-
terial properties, defect states, and photovoltaic performance. Adv. Energy Mater. 1,
845–853.
Chaisitsak, S., Yamada, A., Konagai, M., 2002. Preferred orientation control of Cu(In1-
xGax)Se2 (x≈ 0.28) thin films and its influence on solar cell characteristics. Jpn. J.
Appl. Phys. 41, 507.
Chen, S., Walsh, A., Yang, J., Gong, X.G., Sun, L., Yang, P., Chu, J., Wei, S., 2011.
Compositional dependence of structural and electronic properties of Cu2ZnSn(S, Se)4
alloys for thin film solar cells. Phys. Rev. B 83, 125201.
Chen, W., Kuo, D., Tran, T.H.I., Tuan, A.N.H., 2015. Preparation of CuSbS2 thin films by
co-sputtering and solar cell devices with band gap-adjustable n-type InGaN as a
substitute of ZnO. J. Electron. Mater. 45, 688–694.
Chen, X., Li, Z., Zhu, H., Wang, Y., Liang, B., Chen, J., Xu, Y., Mai, Y., 2017. CdS/Sb2S3
heterojunction thin film solar cells with thermally-evaporated absorber. J. Mater.
Chem. C 5, 9421–9428.
Cho, A., Ahn, S., Yun, J.H., Gwak, J., Song, H., Yoon, K., 2012. A hybrid ink of binary
copper sulfide nanoparticles and indium precursor solution for a dense CuInSe2 ab-
sorber thin film and its photovoltaic performance. J. Mater. Chem. 22, 17893.
Cho, A., Ahn, S., Ho Yun, J., Gwak, J., Kyu Ahn, S., Shin, K., Song, H., Hoon Yoon, K.,
2013a. Non-vacuum processed CuInSe2 thin films fabricated with a hybrid ink. Sol.
Energy Mater. Sol. Cells 109, 17–25.
Cho, A., Ahn, S., Yun, J.H., Eo, Y.J., Song, H., Yoon, K., 2013b. Carbon layer reduction via
a hybrid ink of binary nanoparticles in non-vacuum-processed CuInSe2 thin films. Sol.
1219
Solar Energy 188 (2019) 1209–1220
Energy Mater. Sol. Cells 110, 126–132.
Cho, A., Ahn, S.J., Yun, J.H., Gwak, J., Ahn, S.K., Shin, K., Yoo, J., Song, H., Yoon, K.,
2013c. The growth of Cu2-xSe thin films using nanoparticles. Thin Solid Films 546,
299–307.
Cho, A., Song, H., Gwak, J., Eo, Y.J., Yun, J.H., Yoon, K., Ahn, S.J., 2014. A chelating
effect in hybrid inks for non-vacuum-processed CuInSe2 thin films. J. Mater. Chem. A
2, 5087–5094.
Cho, A., Banu, S., Kim, K., Park, J.H., Yun, J.H., Cho, J.S., Yoo, J.S., 2017. Selective thin
film synthesis of copper-antimony-sulfide using hybrid ink. Sol. Energy 145, 42–51.
Duan, H.S., Yang, W., Bob, B., Hsu, C.J., Lei, B., Yang, Y., 2013. The role of sulfur in
solution-processed Cu2ZnSn(S, Se)4 and its effect on defect properties. Adv. Funct.
Mater. 23, 1466–1471.
Gunawan, O., Todorov, T.K., Mitzi, D.B., 2010. Loss mechanisms in hydrazine-processed
Cu2ZnSn(Se, S)4 solar cells. Appl. Phys. Lett. 97, 233506.
Gwak, J., Lee, M., Yun, H., Ahn, S., Cho, A., Ahn, S., Hulme, J.P., 2015. Selenium flux
effect on Cu(In, Ga)Se2 thin films grown by a 3-stage co-evaporation process. Isr. J.
Chem. 55, 1115–1122.
Hanna, G., Mattheis, J., Laptev, V., Yamamoto, Y., Rau, U., Schock, H.W., 2003. Influence
of the selenium flux on the growth of Cu(In, Ga)Se2 thin films. Thin Solid Films
431–432, 31–36.
Heath, J.T., Cohen, J.D., Shafarman, W.N., 2004. Bulk and metastable defects in CuIn1-
xGaxSe2 thin films using drive-level capacitance profiling. J. Appl. Phys. 95,
1000–1010.
Heath, J.T., Cohen, J.D., Shafarman, W.N., 2003. Distinguishing metastable changes in
bulk CIGS defect densities from interface effects. Thin Solid Films 431, 426–430.
Hegedus, S.S., Shafarman, W.N., 2004. Thin-film solar cells: device measurements and
analysis. Prog. Photovoltaics Res. Appl. 12, 155–176.
Kim, S., Yoo, H., Rana, T.R., Enkhbat, T., Han, G., Kim, J., Song, S., Kim, K., Gwak, J., Eo,
Y., Yun, J.H., 2018. Effect of crystal orientation and conduction band grading of
absorber on efficiency of Cu(In, Ga)Se2 solar cells grown on flexible polyimide foil at
low temperature. Adv. Energy Mater. 8, 1801501.
Lee, D., Yang, J.Y., Kim, Y.S., Mo, C.B., Park, S., Kim, B.J., Kim, D., Nam, J., Kang, Y.,
2016. Effects of the Cu/(Ga+In) ratio on the bulk and interface properties of Cu
(InGa)(SSe)2 solar cells. Sol. Energy Mater. Sol. Cells 149, 195–203.
Lee, S., Lee, E.S., Kim, T.Y., Cho, J.S., Eo, Y.J., Yun, J.H., Cho, A., 2015. Effect of an-
nealing treatment on CdS/CIGS thin film solar cells depending on different CdS de-
position temperatures. Sol. Energy Mater. Sol. Cells 141, 299–308.
Lotgering, F.K., 1959. Topotactical reactions with ferrimagnetic oxides having hexagonal
crystal structures-I. J. Inorg. Nucl. Chem. 9, 113–123.
Macías, C., Lugo, S., Benítez, Á., López, I., Kharissov, B., Vázquez, A., Peña, Y., 2017. Thin
film solar cell based on CuSbS2 absorber prepared by chemical bath deposition (CBD).
Mater. Res. Bull. 87, 161–166.
Moon, D.G., Cho, A., Park, J.H., Ahn, S., Kwon, H., Cho, Y.S., Ahn, S., 2014. Iron pyrite
thin films deposited via non-vacuum direct coating of iron-salt/ethanol-based pre-
cursor solutions. J. Mater. Chem. A 2, 17779–17786.
Nadenau, V., Rau, U., Jasenek, A., Schock, H.W., Introduction, I., Cugase, G., Rau, U.,
Nadenau, V., Schock, H.W., Introduction, I., Cugase, G., 2000. Electronic properties
of CuGaSe2-based heterojunction solar cells. Part I. Transport analysis. J. Appl. Phys.
87, 584–593.
Riedel, I., Parisi, J., Dyakonov, V., Lutsen, L., Vanderzande, D., Hummelen, J.C., 2004.
Effect of temperature and illumination on the electrical characteristics of polymer-
fullerene bulk-heterojunction solar cells. Adv. Funct. Mater. 14, 38–44.
Rodriguez-Lazcano, Y., Nair, M.T.S., Nair, P.K., 2005. Photovoltaic p-i-n structure of
Sb2S3 and CuSbS2 absorber films obtained via chemical bath deposition. J.
Electrochem. Soc. 152, 635–638.
Rudmann, D., Bilger, G., Kaelin, M., Haug, F., Zogg, H., Tiwari, A.N., 2003. Effects of NaF
coevaporation on structural properties of Cu(In, Ga)Se2 thin films. Thin Solid Films
431–432, 37–40.
Saragih, A.D., Daniel, A., Kuo, S.D., 2017. Thin film solar cell based on p-CuSbS2 together
with Cd-free GaN/InGaN bilayer. J. Mater. Sci. Mater. Electron. 28, 2996–3003.
Septina, W., Ikeda, S., Iga, Y., Harada, T., Matsumura, M., 2014. Thin film solar cell based
on CuSbS2 absorber fabricated from an electrochemically deposited metal stack. Thin
Solid Films 550, 700–704.
Sinsermsuksakul, P., Sun, L., Lee, S.W., Park, H.H., Kim, S.B., Yang, C., Gordon, R.G.,
2014. Overcoming efficiency limitations of SnS-based solar cells. Adv. Energy Mater.
4, 1–7.
Sites, J.R., Mauk, P.H., 1989. Diode quality factor determination for thin-film solar cells.
Sol. cells 27, 411–417.
Tiwari, D., Koehler, T., Klenk, R., Fermin, D.J., 2017. Solution processed single-phase
Cu2SnS3 films: structure and photovoltaic performance. Sustain. Energy Fuels 1,
899–906.
Wan, L., Ma, C., Hu, K., Zhou, R., Mao, X., Pan, S., Helena, L., Xu, J., 2016. Two-stage co-
evaporated CuSbS2 thin films for solar cells. J Alloys Compd. 680, 182–190.
Wang, J.T., Shi, X.L., Liu, W.W., Zhong, X.H., Wang, J.N., Pyrah, L., Sanderson, K.D.,
Ramsey, P.M., Hirata, M., Tsuri, K., 2014a. Influence of preferred orientation on the
electrical conductivity of fluorine-doped tin oxide films. Sci. Rep. 4, 3679.
Wang, K., Gunawan, O., Todorov, T., Shin, B., Chey, S.J., Bojarczuk, N.A., Mitzi, D., Guha,
S., 2010. Thermally evaporated Cu2ZnSnS4 solar cells. Appl. Phys. Lett. 97, 143508.
Wang, W., Winkler, M.T., Gunawan, O., Gokmen, T., Todorov, T.K., Zhu, Y., Mitzi, D.B.,
2014b. Device characteristics of CZTSSe thin-film solar cells with 12.6% efficiency.
Adv. Energy Mater. 4, 1–5.
Wei, S., Zhang, S.B., Zunger, A., 1998. Effects of Ga addition to CuInSe2 on its electronic,
structural, and defect properties. Appl. Phys. Lett. 72, 3199–3201.
Welch, A.W., Baranowski, L.L., Zawadzki, P., Dehart, C., Johnston, S., Lany, S., Wolden,
C.A., Zakutayev, A., 2015a. Accelerated development of CuSbS2 thin film photo-
voltaic device prototypes. Prog. Photovolt.: Res. Appl. 24, 1–29.
S. Banu, et al.
Welch, A.W., Zawadzki, P.P., Lany, S., Wolden, C.A., Zakutayev, A., 2015b. Solar energy
materials & solar cells self-regulated growth and tunable properties of CuSbS2 solar
absorbers. Sol. Energy Mater. Sol. Cells 132, 499–506.
Willian de Souza Lucas, F., Lucas, D.S., Welch, A.W., Baranowski, L.L., Dippo, P.C.,
Hempel, H., Unold, T., Eichberger, R., Blank, B., Rau, U., Mascaro, L.H., Zakutayev,
A., 2016. Effects of thermochemical treatment on CuSbS2 photovoltaic absorber
quality and solar cell reproducibility. J. Phys. Chem. C 3, 18377–18385.
Willian de Souza, F., Lucas, Peng, H., Johnston, S., Dippo, P.C., Lany, S., Mascaro, L.H.,
Zakutayev, A., 2017. Characterization of defects in copper antimony disulfide. J.
Mater. Chem. A 21986–21993.
Yang, B., Wang, L., Han, J., Zhou, Y., Song, H., Chen, S., Zhong, J., Lv, L., Niu, D., Tang,
1220
Solar Energy 188 (2019) 1209–1220
J., 2014. CuSbS2 as a promising earth-abundant photovoltaic absorber material: a
combined theoretical and experimental study. Chem. Mater. 26, 3135–3143.
Yu, L., Kokenyesi, R.S., Keszler, D.A., Zunger, A., 2013. Inverse design of high absorption
thin-film photovoltaic materials. Adv. Energy Mater. 3, 43–48.
Zhang, S.B., Wei, S., Zunger, A., 1998. Defect physics of the CuInSe2 chalcopyrite semi-
conductor. Phys. Rev. B 57, 9642.
Zhang, Y., Huang, J., Yan, C., Sun, K., Cui, X., Liu, F., Liu, Z., Zhang, X., Liu, X., Stride,
J.A., Green, M.A., Hao, X., 2018. High open-circuit voltage CuSbS2 solar cells
achieved through the formation of epitaxial growth of CdS/CuSbS2 hetero-interface
by post-annealing treatment. Prog. Photovolt.: Res. Appl. 1–7.