Solar Energy 230 (2021) 13–58

Contents lists available at ScienceDirect

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

Review article on the lattice defect and interface loss mechanisms in

kesterite materials and their impact on solar cell performance
Meenakshi Sahu a, b, 1, Vasudeva Reddy Minnam Reddy b, 1, Chinho Park c, *, Pratibha Sharma a, *
a
Department of Energy Science and Engineering, Indian Institute of Technology Bombay, Powai, Mumbai 400076, India
b
School of Chemical Engineering, Yeungnam University, Gyeongsan 38541, Republic of Korea
c
Korea Institute of Energy Technology (KENTECH), 200 Hyukshin-ro, Naju, Jeollanam-do 58330, Republic of Korea

A R T I C L E I N F O A B S T R A C T

Abbreviations: PV, Photovoltaic; Si, Silicon; Cu2ZnSnS4 or CZTS, Copper-zinc-tin-sulfide; Cu2ZnSnSe4 or CZTSe, Copper-zinc-tin-selenide; Cu2ZnSn (SxSe1-x)4 or
CZTSSe, Copper-zinc-tin-sulfo-selenide; Cu(In, Ga)Se2; CIGSe, Copper-indium-gallium-selenide; Cu2CdSnS4 (CCTS), Copper-cadmium-tin-sulfide; Ag2CdSnS4, Silver-
cadmium-tin-sulfide; Ag2ZnSn(S, Se)4 , Silver-zinc-tin-sulfo-selenide; Ag2ZnSnS4, Silver-zinc-tin-sulfide; Ag2ZnSnSe4, Silver-zinc-tin-selenide; (Ag, Cu)2ZnSnSe4,
(Silver-Copper)-zinc-tin-sulfide; CdTe, Cadmium telluride; GaAs, Gallium arsenide; CuInS2, Copper-indium-sulfide; CuInSe2, Copper-indium-selenium; CuGaSe2,
Copper-gallium-selenide; Cu2(Zn1-xCdx)SnS4 (CZCTS), Copper-zinc-cadmium-tin-sulfide; Cu2Zn(Sn1-xGex)Se4, Copper-zinc-tin-gallium-selenide; (LixCu2x)ZnSn(SxSe1-
x)4, Lithium doped Copper-zinc-tin-sulfo-selenide; In, Indium; Ga, Gallium; Te, Tellurium; As, Arsenic; Cd, Cadmium; Mo, Molybdenum; Cu, Copper; Zn, Zinc; Ni,
Nickel; Ag, Silver; O, Oxygen; S, Sulfur; Se, Selenium; Al, Aluminum; Au, Gold; Na, Sodium; Li, Lithium; Rb, Rubidium; ZnS, Zinc sulfide; ZnSe, Zinc Selenide; Cu2S,
Copper sulfide; CuS, Copper(I) sulfide; Cu2SnS3, Copper tin sulfide; SnS, Tin(II) sulfide; SnS2, Tin sulfide; Cu2Se, Copper (II) selenide; CuSe, Copper selenide;
Cu2SnSe3, Copper tin selenide; SnSe, Tin selenide; SnSe2, Tin (IV) selenide; MoS2, Molybdenum disulfide; MoSe2, Molybdenum diselenide; NiO, Nickel oxide; P3HT:
PCBM, Poly(3-hexylthiophene) (P3HT) and [6,6]-phenyl C61-butyric acid methyl ester; TiO2, Titanium dioxide; Zn(O,S), Zinc oxysulfide; ZnO, Zinc oxide; (Mg, Zn)
O, Zinc oxide magnesium; Ga2O3, Gallium (III) oxide; CdS, Cadmium sulphide; ZnSnO, Zinc tin oxide; MoOx, Molybdenum Oxide; Zn1-xMgxO, Magnesium doped zinc
oxide; Zn1-xSnxOy, Tin doped zinc oxide; Zn1-xCdxS, Cadmium doped zinc sulfide; Mo(SxSe2-x), Molybdenum sulfoselenide; In2S3, Indium sulfide; AZO, Al-doped zinc
oxide; ITO, Indium tin oxide; SnO:F or FTO, Fluorine doped tin oxide; i-ZnO, Intrinsic zinc oxide; NaF, Sodium fluoride; TCO, Transparent conducting oxide; KCN,
Potassium cyanide; KOH, potassium hydroxide; HCl, Hydrochloric acid; Cu2O, Cpuros oxide; Zn(OH)2, Zinc hydroxide; Cd(OH)2, Cadmium hydroxide; NH4OH,
Ammonium hydroxide; TiN, Tin nitride; TiB2, Titanium diboride; Al2O3, Aluminium oxide; CeO2, Cerium(IV) oxide; SiO, Silicon monoxide; SiO2, Silicon dioxide;
SiCx, Silicon carbide; SiNx, Silicon nitride; WO3, Tungsten(VI) oxide; MoO3, Molybdenum trioxide; H2O2, Hydrogen peroxide; CuSO4, Copper (II) sulfate; Na2S,
Sodium sulfide; KMnO4, Potassium permanganate; K2Cr2O7, Potassium dichromate ; (NH4)2S, Ammonium sulfide; Br2-MeOH, Bromine with methanol; TU, Thiourea;
LiOH, Lithium hydroxide; VIn, Indium vacancy; VCu, Copper vacancy; VZn, Zinc vacancy; VSn, Tin vacancy; VS/VSe, Sulfur vacancy/ Selenium vacancy; CuZn, Copper
on zinc antisite; ZnCu, Zinc on copper antisite; CuSn, Copper on tin antisite; SnCu, Tin on copper antisite; ZnSn, Zinc on tin antisite; SnZn, Tin on zinc antisite; Cui,
Copper interstitial vacancy; Zni, Zinc interstitial vacancy; AgZn, Sliver on zinc antisite; ZnAg, Zinc on sliver antisite; CuCd, Copper on cadmium antisite; CdCu, Cadmium
on copper antisite; CdZn, Cadmium on zinc antisite; CuGe, Copper on germanium antisite; GeCu, Germanium on copper antisite; OSe, oxygen antisite to selenium; NNR,
Non-radiative recombination; NNRCs, Non-radiative recombination centers; VB, Valence bond; CB, Conduction band; VBM, Valence bond maximum; CBM, Con­
duction band minimum; Erel, Lattice relaxation energy; px, Polycrystalline; x, Single-crystalline; CBO, Conduction band offset; VBO, Valence band offset; S-Q,
Shockley-Queisser; Voc, Open-circuit voltage; FF, Fill factor; Jsc, Short circuit current; Rs, Sheet resistance; Rsh, Shunt resistance; q, Electron charge; Eg, Optical
Bandgap; A, Ideality factor; EA and Ea1 or Ea2, Activation energy; JL, Photocurrent current; J00, Reverse saturation current; T, Temperature; σn, p, Carrier capture
cross-section; γ, Material constant; τ, Maximum lifetime of the minority carrier; vth, Thermal velocity; Nt, Concentration of defects; ω, Frequency; ΔQk, Coordination
difference between the initial and final states; I0, Intensity at the lowest temperature; A1 and A2, Fitting parameters; Eb, Non-radiative electronic state transition; EU,
Urbach energy; μ, Chemical-potential; Tc, Critical temperature; Å, Angstrom; eV, Electron volt; meV, Milli electron volt; V, Volt; mV, Milli volt; μm, Micron; cm,
Centimeter; nm, Nanometer; s, Second; K, Kelvin; ◦ C, Degree Celsius; ΔG, Gibbs’s free energy; HWF, High work function; XRD, X-ray diffraction; LT-PL, Low-
temperature photoluminescence; PL, Photoluminescence; IQE, Internal quantum efficiency; EQE, External quantum efficiency; TRTS, Time-resolved THz spectros­
copy; TAS, Thermal admittance spectroscopy; TPC, Transient photo-capacitance spectroscopy; DLTS, Deep level transient spectroscopy; TPL, Temperature-dependent
photoluminescence spectroscopy; XPS, X-ray photoelectron spectroscopy; UPS, Ultraviolet photoelectron spectroscopy; IPES, Inverse photoelectron spectroscopy
technique; RTA, Rapid thermal annealing; CV, Capacitance-voltage; DLCP, Drive-level capacitance profiling; UV− Vis− NIR spectrometry, Ultraviolet–visible-near
infrared spectrometry; XRF, X-ray fluorescence; TEM, Transmission electron microscopy; KPFM, Kelvin probe force microscopy; DFT, Density Function Theory; HSE
and HSE06, Heyd-Scuseria-Ernzerhof; HF, Hartree-Fock; GGA, Semi local generalized gradient approximation; PBE, Semi local Perdew-Burke-Ernzerhof; GW, One-
body Green’s function; SILAR, Successive ionic layer adsorption and reaction; ALD, Atomic layer deposition.
* Corresponding authors.
E-mail addresses: chpark@kentech.ac.kr (C. Park), pratibha_sharma@iitb.ac.in (P. Sharma).
1
Equal contribution.

https://doi.org/10.1016/j.solener.2021.10.005
Received 22 June 2021; Received in revised form 2 October 2021; Accepted 4 October 2021
Available online 12 October 2021
M. Sahu et al. Solar Energy 230 (2021) 13–58

Keywords: Kesterite Cu2ZnSnS4 (CZTS), Cu2ZnSnSe4 (CZTSe), and its alloys Cu2ZnSn(Sx,Se1− x)4 (CZTSSe, 0 < x < 1), are
Kesterite solar cells cost-effective thin-film photovoltaic materials to meet future energy demands. The presence of more elements in
Defects these quaternary materials than binary or ternary materials provides flexibility in the fundamental physical and
Interface
chemical properties. The huge Voc losses have limited the performance of devices. On the other hand, progressive
Voltage loss
Passivation
studies for obtaining high efficiency kesterite-based Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(Sx,Se1− x)4 devices
have been ongoing over the past few years, but a concrete understanding is lacking. In this review, the lattice
defect and loss mechanisms in kesterite materials and their impact on solar cell performance are analyzed sys­
tematically. The origin and the cause of atomic defects, deep disorders and interface properties, and their in­
fluence on the solar cell performance are discussed. In addition, a comprehensive correlation between the
experimental and simulation methods has been attempted to determine the loss mechanisms in kesterite mate­
rials. This review also focuses on some approaches to overcome the loss mechanisms. These findings are expected
to provide a path for better device performance.

1. Introduction 9.15% to 11.3% for Cu2ZnSnS4 and Cu2ZnSnSe4, respectively, using as

an active layer material. Nevertheless, currently, the kesterite-based
Solar energy has always been an important renewable energy source solar cell performance is well below the Shockley-Queisser (S-Q) range
to reduce the dependence on fossil fuels, environmental concerns, and (Siebentritt, 2013). The efficiency of Cu2ZnSn(Sx,Se1− x)4 is still far lower
the growing global energy requirement. The well-established silicon- than Cu(In,Ga)Se2-based solar cells. The main performance difference
wafer-based photovoltaic (PV) technology has more than 90% share of between kesterite and chalcopyrite thin-film technology is the high
the total PV market. For single-crystal silicon (Si) solar cells, the highest open-circuit voltage (Voc) losses, denoted as Eg/q-Voc, where Voc, q, and
or record efficiency is 26.7% (Yoshikawa et al., 2017; Green et al., Eg, are the open-circuit voltage, charge of an electron, and optical
2019). On the other hand, certain challenges are associated with silicon bandgap. In an ideal case, the solar cell Voc should be equivalent to the
solar cells, such as an indirect bandgap, low absorption coefficient, bandgap of the absorber material. Thus, Voc losses are an important
wafer thickness up to 100 μm, (Xiong and Zhu, 2009; Luque and factor for highly efficient kesterite photovoltaic devices. Several
Hegedus, 2003) and an expensive fabrication process. Intensive research research groups have contributed their understanding of various aspects
is underway to develop alternative thin-film-based technology owing to that affect Voc in kesterite material-based solar cells in recent years
the limitations of Si-wafer-based technology. Thin-film technologies (Mitzi et al., 2011; Yang et al., 2015; Zhou et al., 2013a; Azimi et al.,
have potential benefits in terms of manufacturing costs, material use, 2014; D. Wang et al., 2018). These include the following features. (1)
and thicknesses up to nano-meters, and films can be deposited on flex­ Obtaining pure phase kesterite materials is much more complex because
ible substrates. Currently, thin-film techniques include copper-indium- of the presence of a large number of component elements complicating
gallium-selenide [Cu(In,Ga)Se2; CIGSe], cadmium telluride (CdTe), the properties of quaternary semiconductors compared to binary and
gallium arsenide (GaAs), and copper-zinc-tin-sulfur (Cu2ZnSnS4; CZTS). ternary semiconductors. Another reason is that it has a very narrow
Commercialized GaAs, CdTe, and Cu(In,Ga)Se2 thin film-based solar growth region. These complexities promote secondary phases (ZnS,
cells have been reported to have record efficiencies of 29.1%, 22.1%, Cu2S, Cu2SnS3, and SnS2) during growth and cell formation in pure
and 23.35%, respectively (Green et al., 2019; NREL, 2019). The main phase kesterite, affecting the solar cell performance (Katagiri et al.,
disadvantage related to these thin films is the presence of rare [indium 2009; Chen et al., 2009b; Kumar et al., 2015a; Siebentritt and Schorr,
(In), gallium (Ga), and tellurium (Te)] and toxic [arsenic (As) and cad­ 2012; Yin et al., 2015; W. Wang et al., 2018; Babichuk et al., 2019). (2)
mium (Cd)] elements, which limit mass production and the economic Lattice defects are pivotal and decisive for applications of semi­
application of these cell technologies (Fthenakis et al., 2009). conductors in solar cells because they affect the generation, separation,
In terms of the high efficiency and cost-effectiveness of PVs, semi­ and recombination (radiative or non-radiative) of the charge carriers
conductor kesterites copper-zinc-tin-sulfide (Cu2ZnSnS4 or CZTS) and directly. Kesterite Cu2ZnSnS4 and Cu2ZnSnSe4 and its alloys Cu2ZnSn
copper-zinc-tin-selenide (Cu2ZnSnSe4 or CZTSe), and their alloys (Sx,Se1− x)4 contain a variety of lattice defects, such as vacancies, e.g.,
copper-zinc-tin-sulfo-selenide [(Cu2ZnSn(Sx,Se1− x)4 or CZTSSe, 0 < x < VZn, VSn, and VCu, antisite defects, e.g., CuZn, CuSn, and SnZn and defect
1)] have attracted considerable attention as an absorber layer. Kesterites complexes or clusters, e.g., [ZnSn + 2ZnCu] [VZn + SnZn], and [VCu +
[Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(Sx,Se1− x)4] materials are a po­ ZnCu], defect levels that correlate with binary and ternary phases. These
tential candidate among thin-film technologies because of their defects introduce band tails that lead to non-radiative recombination in
remarkable properties, such as earth-abundant raw materials, non-toxic photovoltaic devices (Kumar et al., 2015a; Okano et al., 2015; Gokmen
elements, ideal direct bandgap (1.0 eV–1.5 eV), high optical absorption et al., 2013b; Walsh et al., 2012; Chen et al., 2013; Chen et al., 2010b;
coefficient (104–105 cm− 1) (Jiang and Yan, 2013; Chen et al., 2010b; Liu et al., 2020). (3) In the context of charge separation, band alignment
Yang et al., 2019; Kamoun et al., 2007; Yan et al., 2018; Scragg et al., is important at various interfaces, particularly the kesterite [Cu2ZnSnS4,
2008b; Kumar et al., 2015a; Siebentritt and Schorr, 2012; Park et al., Cu2ZnSnSe4, and Cu2ZnSn(Sx,Se1− x)4]/buffer (CdS) interface. Mismatch
2011; Riha et al., 2011; Ramasamy et al., 2012; Zhang et al., 2016a, of the band alignment can result in high-carrier losses by producing a
2016b, 2016c) and desirable electrical properties, making them poten­ large barrier to the photogenerated charge carrier or causing trap-
tial candidates in thin-film technologies. Research on Cu2ZnSnS4 and assisted recombination (Kaur et al., 2017; Yin et al., 2015; Liu et al.,
Cu2ZnSnSe4 began in the late 1950s when Goodman and Pamplin 2020; Courel et al., 2016; Rondiya et al., 2018). (4) The formation of a
designed quaternary I2-II-IV-VI4 semiconductor materials through the thick MoS2 or MoSe2 layer at the rare or back contact interface impedes
cross-replacement of I-III-VI2 chalcopyrite (Goodman, 1958; Pamplin, the flow of charge carriers (holes) from the absorber thin film towards
1964). First, Ito and Nakazawa (in 1988) examined the photovoltaic the back interface (Scragg et al., 2012b; Scragg et al., 2013; Dalapati
properties of Cu2ZnSnS4, and by the 1990s, (Ito and Nakazawa, 1988), et al., 2017; J. Kim et al., 2016).
the efficiency of photovoltaic devices was increased by 2.3% (Friedl­ All of these factors adversely affect the open-circuit voltage (Voc) in
meier et al., 1997) to 2.63% (Katagiri et al., 2001). Research on solar kesterite photovoltaic devices. Theoretical and experimental studies
cells based on kesterite materials has attracted considerable attention have shown that lattice defects and their associated band tails are the
since 2005. Various research groups have reported efficiencies ranging leading cause of the large open-circuit voltage (Voc) losses (Mitzi et al.,
from 5.9% to 12.6% for Cu2ZnSn(Sx,Se1− x)4, and 6.77% to 9.4% and 2013; Gokmen et al., 2013b; Gunawan et al., 2014; Chen et al., 2013;

14
M. Sahu et al. Solar Energy 230 (2021) 13–58

Chen et al., 2010b; Chen et al., 2010a). First-principles calculations of diffusion of elements (Kohara et al., 2001; Wada et al., 2001; Assmann
crystallographic defects for Cu2ZnSnS4 and Cu2ZnSnSe4 showed that et al., 2005; Orgassa et al., 2003). Unlike Cu(In,Ga)Se2 solar cells, MoSe2
both the non-stoichiometric and stoichiometric compounds have com­ formation at the kesterite absorber layer [Cu2ZnSnS4, Cu2ZnSnSe4, and
plex defect properties (Chen et al., 2013). Low ionization energy defects Cu2ZnSn(Sx,Se1− x)4]/Mo interface has a detrimental effect on the device
cause an increase in p-type conductivity that is responsible for performance. Because of the unstable nature of the kesterite absorber
enhancing the band tail levels. Kesterite Cu2ZnSn(Sx,Se1− x)4 materials layer [Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(Sx,Se1− x)4]/Mo interface,
have higher band tail levels than chalcopyrite Cu(In,Ga)Se2 materials voids sometimes appear at the interface (Kaur et al., 2017; Scragg et al.,
(Gokmen et al., 2013b). The leading cause of the limited efficiency of 2013). Therefore, the electrical properties of non-ideal top and bottom
kesterite material-based solar cells was believed to be the Cu/Zn antisite, interfaces may partly explain the decreasing cell performance. On the
and various procedures have been applied to minimize this defect; other hand, considerable efforts have been made to control and optimize
nevertheless, this claim has recently been questioned (Kim and Shin, the interface (Bär et al., 2011; Courel et al., 2014; Mitzi et al., 2011;
2017; Kumar et al., 2018; Moore et al., 2016; Bourdais et al., 2016). Kohara et al., 2001), but sufficient success and progress have been
Some defects, such as VS, SnZn, and [CuZn + SnZn], are found deep inside difficult.
the mid-gap state, making these states non-radiation recombination trap Fig. 2 shows the photovoltaic performance parameters of the pub­
centers for photogenerated carrier pairs (Li et al., 2019; Kim et al., lished kesterite material-based solar cells, namely short-circuit current
2018a, 2018b). density (Jsc), open-circuit voltage (Voc), fill factor (FF), and efficiency
Thus, non-radiative recombination centers (NNRCs) and band tail (η), as the function of the bandgap (Wang et al., 2014; J. Kim et al.,
levels deteriorate the performance of photovoltaic devices (Bourdais 2014; Xin et al., 2015; Haass et al., 2015; Todorov et al., 2013; Duan
et al., 2016). The physics of defects for kesterite materials has been et al., 2020; Y.-F. Qi et al., 2017; Haass et al., 2018; Cabas-Vidani et al.,
studied both theoretically and experimentally. Therefore, there is an 2018; Lee et al., 2015; S.-Y. Kim et al., 2019; Hsieh et al., 2016; Yan
urgent need to review the recent findings methodically, comprehend the et al., 2018; Sun et al., 2018; Yan et al., 2017; Son et al., 2019; S. Kim
charge loss mechanisms related to atomic defects, and gather valuable et al., 2016b; Giraldo et al., 2018; Green et al., 2019). Cu2ZnSn(Sx,
information from these findings. Significant progress has been made in Se1− x)4 absorber-based solar cells exhibit the highest efficiency because
response to these questions for kesterite materials. Several new obser­ of the low sulfur content (Eg ≈ 1.1 eV) and the presence of various de­
vations of their defect chemistry and physics are being made. fects of Cu2ZnSnS4 and Cu2ZnSnSe4. Compared to Cu2ZnSnSe4, the point
In addition to the defects, the non-ideal energy level alignment of the defects energy states in Cu2ZnSnS4 are located much deeper, which is
kesterite material-based solar cells is also detrimental to the perfor­ one of the reason for non-radiative recombination (NNR) loss in the
mance of devices. Thus, the interfaces that are the top (buffer/absorber) absorber (Chen et al., 2013). High efficiency was rarely observed in the
and bottom (absorber/rare) contact must be studied. Fig. 1 shows the bandgap range of 1.2 eV–1.3 eV because Cu2ZnSn(Sx,Se1− 4x)4 materials
conventional device structure of Cu2ZnSn(Sx,Se1− x)4. Cu2ZnSn(Sx, with a moderate S/(S + Se) ratio typically have high formation energies
Se1− x)4 solar cells have four interfaces: (1) rare contact (Mo)/kesterite and high phase separation (Chen et al., 2011). Thus, different S/(S + Se)
absorber layer [Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(Sx,Se1− x)4] ratios can influence the solar cell efficiency by changing the conduction
interface, which acts as a hole transport layer; (2) kesterite absorber band offset (CBO) at the buffer/absorber interface (Kim et al., 2018a,
layer [Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(Sx,Se1− x)4]/buffer or front 2018b). Furthermore, the reduction in efficiency is mainly inherent in
layer (CdS) interface, which acts as an electron separation and transport Voc and FF compared to the experimentally recorded performance pa­
layer; (3) buffer or front layer (CdS)/window layers (i-ZnO/AZO or ITO) rameters with the theoretical value of the S-Q range (Shockley and
interface, which acts as electron charge separation and transport layer; Queisser, 1961). Specifically, the open-circuit voltage is less than 60% of
and (4) window layers (i-ZnO/AZO)/metal layer (Ni-Al or Ag) interface, the S-Q range. Cu(In,Ga)Se2 devices (>0.40 V) show less Voc losses than
which acts as a collection of separated and transported charge carriers. Cu2ZnSn(Sx,Se1− x)4 devices (>0.57 V) (Kim and Shin, 2017). The Voc of
The interfaces at the top (buffer/absorber) and bottom (absorber/rare) the device reflects the voltage difference produced by charge carriers on
contact are essential for separating, collecting, and recombining (radi­ the front and rare electrodes.
ative or non-radiative) charge carriers. The majority carriers (i.e., holes) Several attempts have been made to clarify the leading cause of the
are collected by molybdenum (Mo) at the rare contact. In contrast, mi­ significant Voc losses, resulting in different explanations, and discussions
nority carriers (i.e., electrons) are separated and transported far before are still underway. Many researchers believed that high-density shallow
collection at the metal or end electrodes. Therefore, further improve­ defects such as CuZn and ZnCu were the origin of band tail levels and the
ment is needed for the collection of minority charge carriers. For the Cu decreased bandgap of the bulk material (Bourdais et al., 2016). Lately,
(In,Ga)Se2 cell, molybdenum (Mo) rare contact is favorable because it some research findings indicate that deep defects, such as SnZn2+ and Vs,
provides many benefits, such as good ohmic contact, excellent adhesion are the actual cause of the Voc losses (Li et al., 2019; Kim et al., 2018a,
with the substrate, and MoSe2 formation, which hinders the inter- 2018b). Thus, it is essential to analyze the effects of different defects on

Fig. 1. Conventional photovoltaic device structure of kesterite-based [Cu2ZnSnS4, Cu2ZnSnSe4 and Cu2ZnSn(Sx,Se1− x)4] material with different layers. The inset
shows the taming of the efficiency of the device.

15
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 2. Performance parameters of the single-junction solar cell: (a) Jsc, (b) Voc, (c) FF, (d) η are presented as a function of the bandgap as of the Shockley-Queisser
limit (solid line) and experimentally measured values for η > 11%. The red dots represent solar cells made by vacuum methods, and the green dots are for solar cells
made by non-vacuum methods. (e) Graphical representation of energy band alignment and photo-generated charge carrier transport mechanism at
different interfaces.

the solar cell performance parameters, which are reviewed and dis­ recombination) mechanisms, followed by their detrimental impacts on
cussed in detail below. the performance of photovoltaic devices in the fourth section. The fifth
Many experimental and theoretical studies on the origin of defects, section reviews the simulation models and experimental characteriza­
various interface losses, and their effects in the case of kesterite have tion measurement for defect identification and loss management on the
been discussed and published. These studies are crucial for interpreting interfacial recombination process. In this section, the earlier published
the voltage losses, the existence of band tails, shorter carrier lifetimes, theoretical and experimental results are discussed. In addition, infor­
mobilities, and low efficiency. Hence, this review is divided into six mation on the charge recombination loss mechanism and an under­
sections. The first section briefly summarizes the development history of standing of the identification of defects is then provided. The sixth part
the kesterite materials-based solar cells and briefly introduces the focuses on the approaches to reduce point and complex defects and
leading cause of low efficiency. In the second section, the structural, improve the different interfaces. Finally, a conclusion with challenges
optical, and electrical properties of kesterite materials are summarized. and perspectives to improve the performance of kesterite materials-
The third section provides detailed information on various defects (point based solar cells is presented.
and complex) and different interfaces losses (front and rare

16
M. Sahu et al. Solar Energy 230 (2021) 13–58

2. Kesterite materials and their general properties Table 1

Difference in the binding energy of stannite with respect to kesterite phase
2.1. Crystal structure calculated by many authors and the values in meV/atom.
Stannite phase Ref
The performance of solar cells of semiconductor materials depends Sulfide Selenide
strongly on the properties of crystal structures and elemental composi­
Chen groups − 2.9 − 3.8 (Chen et al., 2009a, 2009b)
tions. In the literature, solar cells with high power conversion efficiency
Person − 1.3 − 3.3 (Paier et al., 2009)
are made from polycrystalline materials. A crystal is a periodic Paier groups − 2.9…3.4 – (Persson, 2010)
arrangement of similar building blocks. Two important concepts, i.e.,
lattice and base, have been introduced to define the crystal structure. A
lattice is a set of regularly spaced points, and a building block of atoms is calculations revealed the kesterite crystal structure to have a few milli-
called a basis that yields the physical crystal structure attached to each electron volts per atom (meV/atom) higher binding energy than the
lattice point. This lattice contributes to a well-ordered structure that stannite crystal structure (Siebentritt and Schorr, 2012).
allows the more effective conversion of light into electricity. Thus, it is An approximately − 10.0 meV difference has been reported (Raulot
important to understand the crystal structure of kesterite materials for et al., 2005), but it is unclear if this variation is due to per atom or unit
high-efficiency solar cells. This section discusses the crystal structure of cell. Nevertheless, all the calculations predict kesterite to have a very
kesterite semiconductor materials. stable crystal structure for sulfide and selenide. On the other hand, in
Typically, quaternary chalcogenide consists of two types of crystal CuInS2, the stannite phase energy difference is marginally larger than
structures: kesterite with space type I4 and stannite with space group Cu-Au ordering. Hence, the stannite phase likely co-exists with the
I42m. (Fig. 3). A comparison of the X-ray powder diffraction (XRD) data kesterite phase, which explains the disorder amid the copper and zinc
of Cu2ZnSnS4 and Cu2ZnSnSe4 shows that the sulfide material charac­ lattice sites reported experimentally (Schorr, 2011). The predicted
teristic occurs in the kesterite structure, while the selenide material bandgap of the kesterite phase is larger than the stannite phase, which
characteristic occurs in the stannite structure. Because of the iso-elec­ might explain the comparatively low open-circuit voltages (Voc) found
tronic nature of Cu+ and Zn2+, it is difficult to differentiate between the in the kesterite material-based solar device compared to their bandgap
kesterite and stannite crystal structure by XRD because of an interaction (Mitzi et al., 2011; Redinger et al., 2011a). In the kesterite system, the
between the electron shell of the atoms and the X-rays. On the other stability of the Cu-Au order has been studied. Generally, the energy
hand, the iso-electronic atoms, copper (Cu), and zinc (Zn) can be difference in the reference of the kesterite phase is equivalent to the
distinguished by neutron diffraction because of the interaction between stannite energy difference, as for kesterite (Chen et al., 2009a). In
the neutrons and nuclei. An extensive study of neutron diffraction of contrast, Paier et al. reported a significantly lower energy difference for
different Cu2ZnSn(Sx,Se1− x)4 materials showed that both sulfide and Cu-Au ordering than stannite (Paier et al., 2009).
selenide occur in the kesterite crystal structure, not in the stannite Tetragonal distortion is another important structural parameter that
crystal structure, and some disorder exists amid the copper and zinc sites is also found in chalcopyrites. Owing to tetragonal distortion in the
(Schorr, 2011). chalcopyrites, the long-distance c-axial bond was observed compared
The chalcopyrite structure has been reported in various polymorphs: with the in-between axis bond. This deviation was directed to the crystal
CuInS2 detected in the chalcopyrite phase with the Cu-Au order field and non-degenerate valence bond maximum (VBM) (JH, 1975).
(Alvarez-Garcıa et al., 2002; Su et al., 1998), and Cu(In,Ga)Se2 observed The crystal field sign is essential for the symmetry of the highest valence
in the chalcopyrite structure. This can be explained by looking at the band (VB) (JH, 1975; Alonso et al., 2002; Hönes et al., 2008), an
variances in the binding energies, which can be calculated by density important condition for the anisotropy of the effective mass (Persson,
functional theory (DFT). The binding energy of chalcopyrite compared 2008). Thus, it is vital to view the c/2a values reported theoretically and
with the Cu-Au phase is 2.0 meV/atom higher for CuInS2 and CuInSe2, experimentally.
while it is 9.0 meV/atom higher for CuGaSe2 (Wei et al., 1999). An experimental combination of neutron and XRD studies, as well as
Therefore, Cu(In,Ga)Se2 exhibits the chalcopyrite phase after the addi­ theoretical calculations, have been reported for the structures (Schorr
tion of gallium (Ga) atoms, while Cu(In,Ga)S2 exhibits polymorphism, et al., 2007). Minor tetragonal distortion was observed in all calculations
which adversely affects the device efficiency (Neisser et al., 2001). This (Raulot et al., 2005; Chen et al., 2009a; Paier et al., 2009; Persson,
adverse effect could be partly due to the smaller bandgap of the Cu-Au
phase (Wei et al., 1999). Table 2
Various authors theoretically calculated and found variation in Lattice parameters and bandgap for Cu2ZnSnS4 and Cu2ZnSnSe4.
binding energy of kesterite with respect to stannite in the Cu2ZnSn(Sx, a = b (Å) c (Å) c/2a Bandgap (eV)
Se1− x)4 absorber material by DFT (Raulot et al., 2005; Chen et al.,
Cu2ZnSnS4 5.419 10.854 1.0015 1.5
2009a; Paier et al., 2009; Persson, 2010) or hybrid functional theory
Cu2ZnSnSe4 5.695 11.345 0.9960 1.0
(Paier et al., 2009). Their values are tabulated in Table 1. These

Fig. 3. Crystal structure of the kesterite-phase (left-side) and stannite-phase (right-side). (Reprinted with permission from Ref. Schorr, 2011).

17
M. Sahu et al. Solar Energy 230 (2021) 13–58

2010): 0.998 ≤ c/2a ≥ 1.006. Table 2 lists the most credible equilibrium 2000; Wibowo et al., 2007; Adhi Wibowo et al., 2007; Altosaar et al.,
lattice parameter values extracted from these perusals (Siebentritt and 2008; Todorov et al., 2010). Nevertheless, the bandgaps of Cu2ZnSnSe4
Schorr, 2012). The table shows that the lattice parameters of the sele­ solar cells have been reported to be 1.0 eV in almost all measurements
nide are slightly larger than those of the sulfide. The trends amid sulfide (Friedlmeier et al., 1997; Altosaar et al., 2008; Todorov et al., 2010).
and selenide and kesterite and stannite are independent of the different Recently, Ahn et al. resolved this difference (Ahn et al., 2010) (Fig. 4).
methods (Siebentritt and Schorr, 2012). The high bandgaps of Cu2ZnSnSe4 were observed by transmission
The issues related to the crystal structure of kesterite materials were measurements due to the coexistence of a ZnSe secondary phase. The
discussed. Single crystals, powders, and thin films of Cu2ZnSnS4, values are listed in Table 2. (Siebentritt and Schorr, 2012).
Cu2ZnSnSe4, and Cu2ZnSn(Sx,Se1− x)4 crystallized in the kesterite-type Thus, for Cu2ZnSn(Sx,Se1− x)4, a tunable bandgap between 1.0 eV and
structure. On the other hand, similar tetrahedral metal co-ordination, 1.5 eV has been estimated, as shown in Fig. 4(d) and 4(e) (He et al.,
i.e., two copper element, one II-group element, and one IV- group 2012). On the other hand, various theoretically methods have been used
element around each chalcogen [sulfur (S) or selenium (Se)] atom, is to calculate the bandgap of kesterite materials: First-principles study,
possible in both the I4 and I42m space groups. For that, different DFT (in different approximations), GW (one-body Green’s function)
experimental and simulation methods have been used to determine the calculations, and the hybrid functional theory, such as the HSE (Heyd-
crystal structure. Nevertheless, a clear conclusion can only depend on a Scuseria-Ernzerhof), the screening constant, and the Hartree-Fock (HF).
careful, detailed analysis of the structure of both materials (Ito, 2014). Some of them are listed in Table 3 (Siebentritt and Schorr, 2012). The
table showed good agreement of the theoretical values for the kesterite
2.2. Optical properties with their experimental results, while the bandgaps of the stannite
structure showed a lower value of approximately 100 meV.
The optical properties of the semiconductor material used as an For both materials, two different peak energies were recorded by
absorber layer in the device are the key of interest for solar cell appli­ low-temperature photoluminescence (LTPL) measurements. The
cations either in terms of the PV architecture (single or multi-junction
cell) or to set the theoretical limit (Shockley-Queisser). Furthermore,
studying the bandgap, density states, doping behavior, and transport Table 3
Calculated bandgaps for Cu2ZnSnS4 and Cu2ZnSnSe4 in the (eV).
property are crucial to attaining high-performance solar cells. The
following sections discuss the optical properties of kesterite absorber Kesterite Stannite Ref
materials. Sulfide Selenide Sulfide Selenide
The bandgap of a material is calculated and experimentally recorded
Botti 1.64 1.02 1.33 0.87 (Botti et al., 2011)*
by UV − Vis − NIR spectrometry (Wibowo et al., 2007), electro- groups
reflectance measurements at room temperature (Adhi Wibowo et al., Person 1.56 1.05 1.42 0.89 (Persson, 2010)
2007), and quantum efficiency measurements. Detailed studies showed Paier 1.49 – 1.30 – (Paier et al., 2009)**
groups
that the bandgap of Cu2ZnSnS4 reached an optimal value of 1.50 ± 0.01
Chen 1.50 0.96 1.38 0.82 (Chen et al., 2009a,
eV (Friedlmeier et al., 1997; Seol et al., 2003; Tanaka et al., 2005; groups 2009b)
Moriya et al., 2006; Scragg et al., 2008a; Kamoun et al., 2007; Sekiguchi
et al., 2006; Zhang et al., 2016a, 2016b, 2016c) predicted by Shockley * Here, the GW results are cited, which are supposed to give the most accurate
description of the electronic structure.
Queisser for high power conversion efficiency. In contrast, other studies
**Here, the hybrid functional results are cited, which are expected to give a
reported that the bandgap of the selenide compound varies from 0.8 eV
better description of the electronic structure.
to 1.65 eV (Friedlmeier et al., 1997; Babu et al., 2008; Matsushita et al.,

Fig. 4. Eg values of Cu2ZnSnSe4 thin films deposited using a co-evaporation technique at Tsub of 370 ◦ C, estimated by various methods; (a) (αhv)2 curve obtained by
the transmission measurement (inset), (b) [hv X ln (1 − EQE)]2 W2 vs. hv curve obtained by the EQE measurements (inset), and (c) PL spectra measured at 10 K.
(Reprinted with permission from Ref. Ahn et al., 2010), (d) Optical bandgap determination from the transmission spectra of Cu2ZnSn(SxSe1− x)4 thin films with
varying × values, (e) optical bandgaps of Cu2ZnSn(SxSe1− x)4 thin films and Cu2ZnSn(SxSe1− x)4 bulk. (Reprinted with permission from Ref. He et al., 2012).

18
M. Sahu et al. Solar Energy 230 (2021) 13–58

Table 4
Types of defects and their impact on kesterite devices.
Types of Description of Defects Ionization levels Formation Impact on the kesterite devices
Defects energy (eV)

Point defects Vacancies VCu One Acceptor 0.02 Ionized from (0/0) to − 1, the transition level ε(-/0) from the shallow states
(–) (Favorable)
VZn Two Acceptor (2 – Deep levels, Can contribute to luminescence and act as recombination centers
–) (Not Favorable)
VSn Four Acceptor – Deep levels, Can contribute to luminescence and act as recombination centers
(4 –) (Not Favorable)
VS/VSe One Donor (–) – Mid-gap state (Not Favorable)
Anti-sites CuZn One Acceptor ≥0.02 Deeper than that of VCu, High concentration can produce a significant amount of
(–) hole carriers and result in good p-type conductivity (Favorable)
ZnCu One Donor (–) – The shallow level near the conduction band (Not Favorable)
CuSn Three Acceptor – Deep and multi-level, act as recombination centers (Not Favorable)
(3 + )
SnCu Two Donor (2 –) – Deep level and mid-gap state, act as electron-hole recombination centers (Not
Favorable)
ZnSn Two Acceptor – Deep level, act as recombination centers (Not Favorable)
(2 + )
SnZn Two Donor (2 –) – Mid-gap state, act as electron-hole recombination centers (Not Favorable)
Interstitials Cui One Donor (–) – The shallow level near the conduction band (Not Favorable)
Zni Two Donor (2 –) – Mid-gap state, act as electron-hole recombination centers (Not Favorable)
Complex Stoichiometry (CuZn + – ~ 0.2 Impact on the electronic structure and optical properties are weak, act as benign
defects ZnCu) defects.
(CuSn + – >0.2 Impact on the electronic structure and optical properties are weak, act as benign
SnCu) defects.
ZnSn + SnZn) – More significant impact on the electronic structure decreases the bandgap by 0.3
eV
Non–Stoichiometry (VCu+ZnCu) – 0.3–0.6 Significant impact and responsible for local variation in chemical potential and
(VZn + SnZn) stoichiometric inhomogeneity
(ZnSn + –
2ZnCu)
(2CuZn + –
SnZn)

bandgap and peak energies cannot be compared directly because the 2.3. Electrical properties
bandgap is greater than at room temperature, whereas photo­
luminescence (PL) measurements are recorded at lower temperatures. The electrical properties, such as charge carrier flow, carrier type,
Moreover, all the PL peaks are broad and asymmetric, with large blue carrier mobility, and density, are fundamental properties for all opto­
shifts with the excitation energy density; this indicates the fluctuating electronic devices. In photovoltaic devices, the cell architecture is
potentials and moves the emission peak to lower energies. Nevertheless, determined by the type of carrier (N or P). The recombination lifetime
these PL signals can still be used as an indicator of the bandgap. These and the decreasing width are affected by the majority carrier density. In
broad PL signals have been observed in sulfide materials at 1.2 eV contrast, the mobility, especially the mobility of minority carriers, af­
(Miyamoto et al., 2008; Unold et al., 2011; Leitao et al., 2011) and 1.3 fects the diffusion length that significantly influences the device current
eV (Altosaar et al., 2008) and in selenide materials around 0.85 eV collection. For kesterite-based photovoltaic devices, a range of pub­
(Altosaar et al., 2008) and 0.95 eV (Grossberg et al., 2009; Redinger lished charge carrier properties can be caused by doping effects, fabri­
et al., 2011b). cation processes, changes in stoichiometry, and sample quality, but the
Those samples exhibiting the higher energy peaks are dominated by choice of the measurement technique can also be a reason.
the kesterite structure, while samples with a lower energy peak are Fig. 5(a) presents the charge carrier effects as a function of the Cu/
composed mainly of the stannite phase. Because of the existence of a (Zn + Sn) ratio measured by both the Hall Effect and capaci­
stannite structure with the lower bandgap, a strong tailing has been tance–voltage profiling. The carrier density increases exponentially as
reported in the spectra of quantum efficiency, even in the best photo­ the ratio of Cu/(Zn + Sn) increases linearly. Hall measurements revealed
voltaic device (Redinger et al., 2011a; Schubert et al., 2011; Todorov kesterite materials to exhibit p-type behavior. The relationship between
et al., 2010; Katagiri et al., 2008; Guo et al., 2010; Repins et al., 2011). the copper content and carrier concentration is often due to the domi­
The co-existence of other phases of different bandgaps is detrimental to nance of the CuZn antisite defects, which have been estimated to provide
the solar cell conversion efficiency (Rau and Werner, 2004). the lowest formation energy and shallowest acceptor states of any defect
Various theoretical and experimental procedures used to estimate (Chen et al., 2013). Compensated defect clusters, such as [VCu + ZnCu],
the optical bandgap of kesterite materials have been discussed briefly. In with extremely high levels, complicate a defect picture and allow sig­
the literature, the bandgap energies for the stoichiometric Cu2ZnSnS4 nificant changes in stoichiometry without the same levels of free car­
and Cu2ZnSnS4 samples were 1.5 eV and 1.0 eV, respectively. In riers. Carrier densities of approximately 1 × 1019 cm− 3 are found for the
contrast, the bandgap of the Cu2ZnSn(Sx,Se1− x)4 samples varies with the stoichiometric compound. Considerable deviations can be found for
S/Se ratio from 1.0 eV to 1.5 eV. Consequently, a linear relationship individual compositions, such as the charge carrier densities for IBM
exists between the bandgap and S/Se ratio in the Cu2ZnSn(Sx,Se1− x)4 device recorded with Cu/(Zn + Sn) = 0.85 and S/(S + Se) = 0.25 are less
sample. In addition, the effects of various crystal structures on the than 1 × 1015 cm− 3 (Wang et al., 2014). In contrast, analyses of other
bandgap were observed. On the other hand, a clear conclusion can only samples revealed 1 × 1017 cm− 3 with a similar composition. The data
draw upon a carefully detailed analysis of various measurements. were recorded by Hall Effect or capacitance–voltage profiling, as shown
in Fig. 5, which should be done with different sample arrangements:
first, only absorber films and later, the complete cell structure (Gross­
berg et al., 2019). Despite the various boundary conditions, the overall

19
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 5. Carrier density from the Hall Effect and

capacitance–voltage on single-crystal films (x) and
polycrystalline films (px) as a function of the Cu/
(Zn + Sn) ratio (a), and as a function of the S/(S +
Se) ratio (b). Note that not all of the films have the
same Zn/Sn and Cu/Sn ratios in (a), and the single-
crystal samples in (b) have a higher Cu/(Zn + Sn)
ratio than the px samples (0.85–0.96). (Reprinted
with permission from Ref. Grossberg et al., 2019),
Mobilities from Hall and Thz Absorption (TRTS) for
a single crystal (x) and polycrystalline films (px), as
a function of S/(S + Se) (c) and (d) Cu/(Zn + Sn).
(Reprinted with permission from Ref. Grossberg
et al., 2019).

agreement is reasonable. The typical samples shown in Fig. 5 can differ estimating the charge carrier density. The best photovoltaic solar cells
not only in the copper content but also in the Zn/Sn ratio and sulfur are typically cupper-poor and contain approximately 20%–30% sulfur,
content, and obviously sample processing, sodium (Na) content, and for which Fig. 5 shows the estimated carrier density (<1016 cm− 3) re­
possible addition. An important fact to consider is that structural anal­ ported for IBM’s champion device (Wang et al., 2014). The presence of
ysis is carried out using different methods in different laboratories, such sodium can play an important role in increasing the carrier density in Cu
as energy dispersive X-ray spectroscopy, inductively coupled plasma (In,Ga)Se2 materials, which is either by the diffusion of soda-lime glass
mass spectrometry, and X-ray fluorescence (XRF). Apart from the cali­ or by sodium fluoride (NaF) precursor layers or by the subsequent
bration challenges and varying depth bias, these composition analyses treatment of NaF (Rudmann et al., 2005; Caballero et al., 2009).
generally do not differentiate between the kesterite and secondary and The incorporation of sodium in kesterite devices has been examined
ternary phases. Therefore, the same sample compositions having various in several reports on thin films and single crystals (Nagaoka et al.,
secondary phases can show remarkably different compositions of the 2014a). In these reports, the sodium content increases the carrier den­
kesterite phase, as described by Just and co-authors (Just et al., 2016a; sity, as recorded by Hall measurements, whereas a recent study reported
Just et al., 2016b). a systematic decrease in carrier density with increasing sodium content
For copper-poor samples, Fig. 5(b) shows the carrier densities for (Tampo et al., 2017). In this report, a NaF layer was fabricated and
different samples as a function of the S/(S + Se) ratio. The variation in spread to the Cu2ZnSnSe4 absorber layer as the film developed.
copper content has a negligible effect on the carrier density, but for For photovoltaic applications, the transport of charge carriers plays
polycrystalline films, increasing the sulfur content decreases the carrier an essential role in moving from the electrode and into and within the
density, while it increases slightly for single crystals. Single crystals active semiconductor layer. Transport of a carrier can occur either by
behave differently from polycrystalline films. Their values have higher diffusion or by drift due to electric fields. For both events, the carrier
orders of magnitude, possibility related to the crystals of polycrystalline mobility plays a decisive role depending on the semiconductor type (P or
films being closer to stoichiometric than the Cu-poor structure. The N) and, based on the technique used, can extend to a very broad range of
carrier densities measured by the Hall measurements and capacitance values from below 1.0 cm2 V− 1 s− 1 for amorphous semiconductors to
(room temperature) measurements reflect the overall good agreement several thousand cm2 V− 1 s− 1 for epitaxial gallium arsenide (GaAs) thin
(in the same figure), meaning that both techniques are robust to films.

20
M. Sahu et al. Solar Energy 230 (2021) 13–58

The charge carrier mobility is generally determined by the funda­ many studies. Recently, TRTS measurements were used to examine the
mental properties of the semiconductor material, such as the bandgap, impact of disorder on the carrier mobility present in Cu2ZnSnSe4 films
the effective masses of the valence band maximum (VBM) and conduc­ (Hempel et al., 2018a). There was no change in the charge carrier
tion band minimum (CBM), carrier impurity, scattering of carrier- mobility in samples with varying degrees of Cu-Zn cation order, indi­
phonon, scattering at the grain boundaries, and possible localized de­ cating that the kesterite materials do not have a cation-order limiting
fects state (deep or shallow levels) trapping. Various experimental event for the charge carrier mobility.
measurement methods can be used to obtain the carrier mobility, such as The electrical properties of Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn
Hall measurement (Nagaoka et al., 2013; Nagaoka et al., 2014b; (Sx,Se1− x)4 have been studied. Despite the variation of the composition
Gunawan et al., 2018; Gunawan et al., 2015; Tai et al., 2016), voltage- ratios that are influenced by CuZn defects, kesterite samples showed p-
dependent internal quantum efficiency (IQE) measurements (Hages type conductivity. The above discussion shows that the presence of a
et al., 2016a; Gokmen et al., 2013a), or time-resolved THz spectroscopy dopant, stoichiometric, and growth temperature plays an important role
(TRTS) (Hempel et al., 2018b; Guglietta et al., 2014; Hempel et al., that affects the electrical properties of the kesterite samples.
2016; Hempel et al., 2017; Hempel et al., 2018a). Different required
sample geometries often give very different results for the carrier 3. Various defects (point and complex) and different interfaces
mobility, rather than being in good agreement with different measure­ loss (front and back recombination) mechanisms
ment methods to determine the carrier density and mobility. The Hall
measurements measure the majority of carriers, while the TRTS mea­ To understand the various loss mechanisms in chalcogenide com­
surements capture the mobility of aggregates for the minority and ma­ pounds, it is vital to perceive the formation of the various defects and
jority carriers. On the other hand, in photovoltaic devices, carrier flow is interface mismatch. Defects might appear as single defect levels with
usually dominated by the minority carriers (Hempel et al., 2018b). distinct energy states inside the bandgap or multiple defect levels and
Furthermore, both Hall and TRTS measurements are usually performed the defect distribution with a quasi-continuous state density inside the
on samples without back contact, but recent TRTS measurements yiel­ bandgap. Similar radii of various ionic species assist their formation in
ded good results for samples deposited on the back contact substrate chalcogenide compounds. These various defects involve vacancies, anti-
(Hempel et al., 2017). Fig. 5 shows the mobilities measured from the sites, interstitials, and complexes (Table 5) (Chen et al., 2013; Nagoya
Hall and THz absorption measurements for polycrystalline (px) and et al., 2010; Chen et al., 2010a; Chen et al., 2010b; Chen et al., 2009a).
single-crystalline (x) films.
Fig. 5(c), exhibits the mobility as a function of the S/(S + Se) ma­ 3.1. Point and complex defects
terial, there is a large range of values from 0.5 cm2 V− 1 s− 1 to greater
than 100 cm2 V− 1 s− 1 for the carrier mobility. An extensive study These point and complex defects construct the shallow acceptor
showed that the Hall mobility for single crystals coincides with the levels and donor levels, mid-gap states, and deep trap states inside the
analysis of TRTS on single crystals and thin films. In contrast, Hall bandgaps of the absorbing materials (Chen et al., 2010b). Furthermore,
analysis of polycrystalline films is typically much smaller than 1.0 cm2 complex defects can cause local bandgap fluctuations because of the
V− 1 s− 1. uneven structure or chemical stoichiometry and local chemical potential
This specifies that scattering at the grain boundary plays a significant fluctuations and fluctuations in the band structure caused by point de­
role in the flow of the carrier for extended periods because TRTS analysis fects. The ensuing sections provide a detailed description of various loss
examines the intra-grain transport on a scale of approximately 30 nm mechanisms because of the different defect types and interface
and does not have grain boundaries in single crystals per definition. mismatches.
Thus, the TRTS and Hall analysis for single crystals reflects a decrease in Among the tabulated defects, copper vacancies form a shallow
the charge carrier mobility from pure Cu2ZnSnSe4 with 140 cm2 V− 1 s− 1 acceptor level (~0.02 eV) upside the valance band maxima (VBM) and
to 10–40 cm2 V− 1 s− 1 for the pure sulfide. In this case, the higher values help enhance the p-type conductivity of the kesterite absorber material
from TRTS analysis than Hall measurements can be attributed to the Hall and solar cell efficiency. In contrast, other vacancies, such as VZn, VSn,
analysis examining the mobility of the majority (holes) and TRTS VS, and VSe, are unfavorable (Katagiri et al., 2001; Wibowo et al., 2007;
measuring the hole and electron mobility. The accurate value for the Tanaka et al., 2005). Moreover, VS and VSe vacancies produce mid-gap
mobility of the minority carrier can be calculated by subtracting the x- states, whereas VZn and VSn vacancies produce multiple states with
Hall values from the TRTS-px values, which are slightly smaller than the both midgap states and deep levels between the bandgap of Cu2ZnSnS4
TRTS values shown in Fig. 5(c). A new photo-Hall technique was used to (Chen et al., 2013; Biswas et al., 2010). Other acceptor vacancies play an
measure the majority and minority carrier mobilities (Guglietta et al., insignificant role in enhancing the p-type conductivity of kesterite
2014), together with intensity-dependent and excitation wavelength- materials.
dependent TRTS measurements (Hempel et al., 2018b). The TRTS Elements with similar radii and valences, such as copper and zinc,
measurement showed that the electron mobility for a pure Cu2ZnSnSe4 are liable for the antisite defects in the Cu2ZnSnS4. The formation en­
polycrystalline layer could be calculated at 140 cm2 V− 1 s− 1, whereas ergies of the defect and its transition levels play a crucial role in antisite
the mobility of the hole was less than approximately 10 cm2 V− 1 s− 1. defects to form either donor or acceptor levels in Cu2ZnSnS4. CuZn acts as
Scattering at the grain boundary is expected to affect the carrier a shallow acceptor level, but the formation energy is greater than for
mobility, especially under the condition of non-vanishing grain bound­ copper vacancies (~0.02 eV), whereas ZnCu acts as the shallow donor
ary potentials. From Fig. 5(d), a comparison of the hole mobilities level. A further antisite defect (SnCu) forms multi-levels between the
measured by Hall analysis for single crystals and polycrystalline thin bandgap. In contrast, a copper atom at the tin antisite (CuSn) defect
films shows that to clarify the difference between these analyses, the produces three transition levels. Similarly, when a zinc atom is replaced
grain boundary potentials in kesterites have a greater impact than kT. by a tin atom in the unit cell of Cu2ZnSnS4, the SnZn antisite defect forms,
Whether transport across grain boundaries or intra-grains is preventive which produces two mid-gap states. (Chen et al., 2013). The two
in photovoltaic devices is strongly dependent on the cell architecture generally detected interstitial defects, Zni and Cui, under copper-rich or
and sample morphology i.e. If the grain size is sufficiently large (~500 stoichiometric conditions are undesirable because they form mid-gap
nm), then the carrier absorbed in a single grain may not need to trans­ states and shallow donor levels, respectively. Hence, copper-poor and
port through the grain boundaries to reach the p-n junction. In contrast, zinc-rich conditions are favorable for high-efficiency solar cells (Tanaka
in a material with small grain sizes, the grain boundaries can hinder the et al., 2009). On the other hand, non-stoichiometry of one element in the
transport processes in solar cells. absorber material supports the formation of additional unenviable de­
The impact of Cu-Zn ordering on devices has been the subject of fects and defect complexes. In addition to vacancies, antisite and cluster

21
M. Sahu et al. Solar Energy 230 (2021) 13–58

Table 5
Device characteristics of the recent champion Cu2ZnSn(Sx,Se1− x)4 and Cu(In,Ga)Se2 or Cu(In1− x,Gax)(Sy,Se1− y)2 photovoltaic devices.
Device η FF % Voc (mV) Jsc (mA cm− 2) Eg (eV) Eg/q-Voc (V) Ref
%

Cu2ZnSn(Sx,Se1− x)4 12.6 69.8 513.4 35.2 1.13 0.617 (Wang et al., 2014)
11.1 69.7 459.8 34.5 1.13 0.670
10.1 63.7 517.0 30.8 1.15 0.633 (Barkhouse et al., 2012a)
Cu(In,Ga)Se2 19.5 79.9 692.0 35.2 1.14 0.448
Cu(In1¡xGax)(Sy,Se1¡y)2 21.7 79.3 746.0 36.6 1.13 0.384 (Jackson et al., 2015)
23.35 80.4 734.0 39.6 1.08 0.350 (Nakamura et al., 2019)

defects are the other dominating defects observed in Cu2ZnSnS4. Fig. 6 3.2. Interfaces loss (front and back recombination) mechanisms
(c) and 6(d) present the crystal structure and probable valance states of
Cu2ZnSnS4. All the solid circles are out of the plane, and the dotted Figs. 1 and 2e present the basic solar cell structure, energy band
circles are in the plane (Chen et al., 2013). alignments, and electrical charge carrier transport through various
The defect complexes are formed due to energetically self- interface layers of the kesterite device. The variances in the crystal
compensation of donor–acceptor between the vacancies, interstitials, structure, lattice parameter, work function, electron affinity, and me­
and anti-sites defects in the Cu2ZnSnS4 material because the formation chanical stress at the device interfaces stimulate various defects states,
energy of complexes anti-site defect is lower than the formation energy interface states, and recombination of photo-generated electrical charge
of individual defects. These defect complexes contain [CuZn + ZnCu], carriers (Meril and Sudha Kartha, 2009). Furthermore, the inter-
[SnZn + ZnSn], [CuSn + SnCu], [VCu + ZnCu], [VZn + SnZn], [ZnSn + diffusion of elements within the interfaces also forms various second­
2ZnCu], and [2CuZn + SnSn]. They can be categorized further into two ary phases. An obstruction of carrier extraction through the interface is
parts: stoichiometric conserving complex defects, including [AB + BA], due to the segregation of the binary and ternary compounds at the
[CuZn + ZnCu], [SnZn + ZnSn], and [CuSn + SnCu], where element B is interface. During the annealing process, elements inter-diffuse, such as
replaced by element A in one anti-site and element A is replaced by cadmium from the CdS buffer layer diffusing into absorber material
element B in another anti-site; and non-stoichiometric conserving while zinc and S/Se out-diffuse from the absorber layer. This may
complex defects, such as [VCu + ZnCu], [VZn + SnZn], [ZnSn + 2ZnCu], modify the band alignment at the interface of CdS/Cu2ZnSnS4 and
and [2CuZn + SnZn] (Chen et al., 2010a; Chen et al., 2010b; Wei and obstruct carrier extraction through the layers (Kato et al., 2013).
Zhang, 2005). The defect formation stoichiometric conserving clusters is In the same way, the formation of secondary compounds, such as
not dependent on the stoichiometric discrepancies in Cu2ZnSnS4, and Cu2S at the CdS/Cu2ZnSnS4 interface, and ZnS and SnS2 at the
their formation is omnipresent under copper-poor and copper-rich Cu2ZnSnS4/Mo interface and their segregation degrade the solar cell
conditions. These defects do not have any influence on the stoichio­ performance (Muhunthan et al., 2014; Kato et al., 2012; Wätjen et al.,
metric discrepancies in Cu2ZnSnS4. On the other hand, non- 2013). Another crucial issue for the device efficiency is the formation of
stoichiometric conserving complex defects, such as [2CuZn + SnZn], a MoS2 or MoSe2 layer at the interface of the absorber/Mo layer during
are responsible for local fluctuations in stoichiometry. The defect com­ annealing (sulfurization or selenization). At the interface, the formation
plex [2CuZn + SnZn] is formed when two anti-sites, zinc replaced by of a MoS2 or MoSe2 layer influences Voc and modifies the energy band
copper and zinc replaced by tin, combine and form a recompense defect alignment between the Cu2ZnSn(Sx,Se1− x)4 absorber and Mo rare con­
complex. tact (Gunawan et al., 2010; Shin et al., 2012; Shin et al., 2013a).
A CdS buffer or front layer is usually deposited by chemical bath
deposition (CBD) on an absorber layer of p-type kesterite to produce a
p–n junction. The energy band alignment at the interface of the buffer

Fig. 6. (a) Type I hetero-interface structure. (b)

Type II hetero-interface structure. Spike and cliff are
barriers for photo-generated and injection (Reprin­
ted with permission from Ref. Kaur et al., 2017), (c)
Crystal structure and (d) valence states of
Cu2ZnSnS4. The dotted circles are in the plane, and
the solid circles are out-of-plane. (Reprinted with
permission from Ref. Chen et al., 2013), (e) Sche­
matic band diagram for Cu(In,Ga)Se2 (Cu-poor) and
Cu2ZnSnS4 (Cu-poor and Zn-rich). The Cu(In,Ga)Se2
only forms the band tail while Cu2ZnSnS4 forms
both band tail and midgap states. The transitions BT
and BB correspond to the band–tail and band–band
transition, respectively. The CB and VB are the
conduction band and valance band, respectively.
(Reprinted with permission from Ref. Romero et al.,
2011).

22
M. Sahu et al. Solar Energy 230 (2021) 13–58

(CdS)/kesterite [Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(Sx,Se1− x)4] is a et al., 2012; Takahashi et al., 1998). Scragg et al. reported that the MoS2
vital area of study because of its essential role in charge carrier transfer. thickness increases with increasing annealing time (Scragg et al., 2013).
This can help comprehend the significant influence of band offset on the
transportation and recombination due to maximum carrier loss through 4. Effect and impact of defects and interface loss on the kesterite
and near the interface. The major question that must be answered: what photovoltaic performance
kind of interface is formed in between the buffer layer (CdS) and kes­
terite [Cu2ZnSnS4, Cu2ZnSnSe4 and Cu2ZnSn(Sx,Se1− x)4] absorber layer, 4.1. Impact of defects
Type I or Type II? various first-principles calculations and modeling
have been done to answer this question. Both the Type I and Type II Different defects, such as points and complexes, in kesterite materials
interfaces can be identified based on their conduction band alignment accompanied by carrier recombination loss at various interfaces influ­
[Fig. 6(a) and 6(b)] (Kaur et al., 2017). In Type I, the conduction band ence the solar cell performance. This might be the leading cause of the
minimum (CBM) of the front layer lies higher than the absorber layer, as gap in device performance among kesterite [Cu2ZnSnS4, Cu2ZnSnSe4,
shown in Fig. 6(a). As a result, a large potential barrier (spike) formed and Cu2ZnSn(Sx,Se1− x)4] and chalcopyrite [Cu(In,Ga)Se2] solar cells
that could inhibit the transport of electrons from the absorber to the (Wang et al., 2010; Chen et al., 2013; Ahmed et al., 2012). The con­
front layer through the interface. duction band minimum and valance band maximum can differ locally
On the other hand, in Type II, the conduction band minimum (CBM) because of dependence on the ionization energy of point and anti-site
of the front layer lies lower than the absorber layer, as shown in Fig. 6 defects in kesterite absorber materials.
(b). Consequently, a cliff is formed, and electron flow becomes easier Cu(In,Ga)Se2 has only the band tail states, whereas the band tail
from the absorber to the front layer through the interface (Scheer and states in Cu2ZnSnS4 are present along with a mid-gap, as shown in Fig. 6
Schock, 2011). Types I and II are assumed to be preferable for high open- (e) (Romero et al., 2015). The existence of mid-gap states in Cu2ZnSnS4
circuit voltage (Voc) and high short circuit current (Jsc), respectively. might explain the enhanced charge carrier trapping and the losses of the
The mismatch of the conduction band minimum of the two different photo-generated carrier. The Cu2ZnSnS4 emission band will closely be
materials is called the conduction band offset (CBO). Similarly, the correlated with the band tail states. Hence, the slight variation in the
valence band maximum mismatch is called the valence band offset spectra will be related to the potential fluctuations, explaining the poor
(VBO). Theoretical findings have reported that such offsets in the band solar cell power conversion efficiency. The zinc on the copper anti-site
could depend on connecting anions existing at the supercell interface, (ZnCu) ionizes and acts as a shallow donner. Thus, these defects can
the interface electric dipole role, and the lattice hindrance condition assemble and act as an electron sink and hole barrier. In the Cu2ZnSn(Sx,
(Bao and Ichimura, 2012a). Se1− x)4 device, these defects enhance the recombination of electrons and
For high-efficiency solar cells, high conductivity materials, such as prompt open-circuit voltage because of the structural non-homogeneity.
intrinsic zinc oxide (i-ZnO)/Al-doped ZnO (AZO), ZnSnO, and In2S3, are The maximal open circuit voltage stated for the Cu2ZnSnS4 device is
used for the window layer. These materials are highly conductive and within the scope of 500–600 mV, which is much less than their bandgap,
have a board optical bandgap that is vital for high transparency in the ~1.45 eV (Ahmed et al., 2012). In the Cu2ZnSn(Sx,Se1− x)4 device, the
desired spectrum for the front contact in the Cu(In,Ga)Se2 and defect activity limits the open-circuit voltage (Voc). Other liable factors
Cu2ZnSnS4 device. The wide optical bandgap, low cost, and moderate for Voc losses are recombination centers on a deep level within the bulk,
optoelectronic property of ZnO make it suitable as a window layer in at the interface, and grain boundaries in Cu2ZnSnS4 (Todorov et al.,
photovoltaic applications (Janotti and Van de Walle, 2009). The part of 2013; Polizzotti et al., 2013). The presence of these mid-gap states is
AZO is cleared because of its high conductivity and offers a suitable responsible for the non-radiative recombination and successive losses in
wide-bandgap for high transparency, but the precise part of i-ZnO is the open-circuit voltage. Nevertheless, the defect complex [VCu + ZnCu],
unclear (Weinhardt et al., 2004). The window layer of i-ZnO is suitable which is formed by self-compensation of copper vacancy and zinc on the
for a photovoltaic device with its beneficial effects in decreasing the copper anti-site, is beneficial for charge transfer in the Cu2ZnSn(Sx,
electronic inhomogeneities at the surface of absorber material (Rau and Se1− x)4 device (Redinger et al., 2013). This helps to enhance the overall
Schmidt, 2001). This eliminates the oxides and inhomogeneities at the solar cell performance. Complex defects [VCu + ZnCu] are energetically
CdS surface formed during deposition (Weinhardt et al., 2004) and is favorable, which helps to lower the valence and conduction band edges.
beneficial for enhancing the open-circuit voltage (Voc) (Rau and This defect resists the holes (majority charge carriers), and collects the
Schmidt, 2001). The part of i-ZnO can be understood in terms of the electron (minority charge carriers) by the copper-poor and zinc-rich
band alignment of ZnO with CdS. If the band alignment between the regions to the n-type buffer layer (Chen et al., 2010a). In contrast, the
buffer and window layer is unsuitable for the flow of the charge carrier, point defects and defect clusters listed in Table 4 also play a crucial part
the external quantum efficiency (EQE) of the solar cell decreases even if in the final solar cell efficiency. The formation of different states, such as
the internal quantum efficiency (IQE) of the absorber is high. Fig. 25 (a) mid-gap states, deep level states, and donor levels acts as recombination
shows the band alignment among ZnO and CdS is shown in (Yang et al., and trap centers for collecting photo-generated charge carriers at the
2016). In the literature, this type of CBO does not affect electron flow end electrode. Furthermore, non-stoichiometric conserving defect clus­
through the ZnO and CdS interface; hence, highly efficient Cu(In,Ga)Se2 ter formation affects the local stoichiometry variations and fluctuations
solar cells have been reported. Moreover, the microscopic interface in the conduction band, and valance band edges in the Cu2ZnSnS4 (Chen
properties, such as local charge distribution because of the formation of et al., 2013; Yin et al., 2015). The formation of defects of shallow and
interface chemical bonds, do not play a vital part in the band offset. donor effectively reduces the optical-bandgap in the Cu2ZnSnS4 and
Given its valuable effects in Cu(In,Ga)Se2 devices, the Mo layer is affects the absorption of light in the device.
used as a rare contact electrode in the kesterite device. The Mo rare Not all defects formed in Cu2ZnSn(Sx,Se1− x)4 are detrimental. Few
contact has a beneficial role for the Cu(In, Ga)Se2 solar cells in terms of defects segregate, particularly at or near the grain boundaries, to
good adhesion properties, ohmic contact, and MoSe2 formation, which decrease the recombination of the charge carrier (Persson and Zunger,
behave as a diffusion barrier (Assmann et al., 2005; Scragg et al., 2003; Yan et al., 2007). Defect segregation, such as ZnSn, Nai, and OSe on
2012b). On the other hand, its impact on kesterite devices is still under the grain boundaries, is helpful because of the following: (1) bonds
research, and the Mo layer had some deleterious effects on kesterite breakage or weakening results in the deep levels into the Cu2ZnSn(Sx,
devices if not customized (Kato et al., 2013). MoS2 or MoSe2 formed Se1− x)4, and (2) hinder the hole by forming a barrier and facile trans­
during annealing (sulfurization or selenization) of the absorber layer, portation of an electron via the grain boundaries (W.J. Yin et al., 2014).
which affected the Voc and band arrangement at the interface of kesterite Yan et al. reported that the defects on the grain boundaries are benign
and Mo, and had harmful effects on the solar cell performance (Kato electrically and do not form high recombination positions (Yan et al.,

23
M. Sahu et al. Solar Energy 230 (2021) 13–58

2007). In the Cu2ZnSnS4-based devices, the grain boundaries allow the Cu2ZnSn(Sx,Se1− x)4 based solar cells to more than 12.0%. Interface en­
effective transportation of electrons towards the end electrode. In other gineering is a critical and crucial issue for the Cu2ZnSn(Sx,Se1− x)4-based
photovoltaic devices, such as polymer solar cells and hydro-generated solar cell as well as in other generation photovoltaics. At the latest, the
silicon solar cells, the grain boundaries act as traps or recombination 22.5% efficiency of silicon-based photovoltaic was obtained using MoOx
sites (Dutta et al., 2013). as a hole collection layer in the place of a p-type amorphous silicon
layer. The optical absorbance improved compared to the reference
4.2. Impact of interface loss photovoltaic cell, and the fill factor increased from 75.4% to 77.1%. The
increased light absorption was attributed to the enhanced oxygen va­
The phase impurities (formation of secondary or ternary phases) and cancy density inside the MoOx layer (Graetzel et al., 2012).
poor interface quality (due to band offsets) are primarily responsible for A decrease in the efficiency of planar perovskite solar cells was
the lower kesterite device performance (Muhunthan et al., 2014). observed because of the screening of photo-generated charge carriers by
Recombination at the interface is one of the main reasons for the Voc ions accumulated at the interface (Tress et al., 2015). A recent study
losses (Gunawan et al., 2010). Table 5 compares the solar cell charac­ used a low-temperature NiO layer as a hole transport layer in a planar
teristics of the champion copper indium gallium sulfo-selenide [Cu perovskite heterojunction device. The layer made an approximately loss-
(In1− xGax)(Sy,Se1− y)2] and Cu2ZnSn(Sx,Se1− x)4 device. A comparison of free interface because it efficiently passivated recombination at the
the characteristics of the Cu2ZnSn(Sx,Se1− x)4 and [Cu(In1− xGax)(Sy, surface (Hou et al., 2016; Geissbühler et al., 2015). NiO may also be
Se1− y)2] devices showed that the absorber materials with the highest helpful as a hole transport layer in Cu2ZnSn(Sx,Se1− x)4 solar cells
record preformation efficiency have a similar bandgap but varies in the because it fulfills the necessity of a high work function material with an
Eg - qVoc values, where Voc is the open-circuit voltage, Eg is the optical appropriate valence band maximum at higher energy regarding the
bandgap, and q is the electron charge. valence band maximum of Cu2ZnSnS4. In perovskite solar cells, interface
The Voc losses play a pivotal role in the solar cell performance. The engineering has been studied by a careful choice of interface material.
Barkhouse group reported the best Cu2ZnSnS4 based solar cell with Without any light management, 19.3% efficiency was obtained by
10.1% efficiency and compared it with Cu(In, Ga)Se2-based solar cell suppressing the charge carrier recombination in the absorber layer
efficiency of 19.5%. The Voc losses were evident in terms of the Voc value (Dubey et al., 2016; Zhou et al., 2014). Recently, Kumar and co-authors
comparable to the bandgap (Barkhouse et al., 2012a). To clarify the examined P3HT:PCBM bulk heterojunction solar cells and reported the
decrease in Voc and the dominant mechanics for recombination of the great significance of the P3HT:Al interface to reduce the recombination
photo-generated charge carrier, the temperature-dependent Voc was of charge carriers and enhance the device efficiency (Adhikari et al.,
examined by plotting Voc as a function of T, as shown in Eq. (1): 2015).
J00 Recently, on a Cu(In,Ga)Se2 device prepared by a non-vacuum pro­
VOC = EqA − AkT
q ln JL (1) where EA, A, k, q, JL, and J00 are the recom­ cess, titanium dioxide (TiO2) was used as a buffer layer for electron
bination activation energy, ideality factor, Boltzmann constant, electron transport. An inverted surface formed because of elemental inter-
charge, photocurrent current, and reverse saturation, respectively. The diffusion at the TiO2/Cu(In,Ga)Se2 interface, which decreased inter­
Voc vs. T data must be a straight line. An extrapolation will give the face recombination and enhanced the Voc. TiO2 has a larger bandgap
activation energy EA/q at T = 0 K. Fig. 7(a) shows that for both the (3.4 eV) than CdS (2.4 eV). Hence, it has high absorption in the ultra­
Cu2ZnSn(Sx,Se1− x)4 devices, the intercept has a value lower than the violet (UV) region of the solar spectrum due to remarkable gain in
bandgap highlighting the dominating charge carrier recombination at photocurrent (Hsu et al., 2015). The performance of cuprous oxide solar
the interface. This may occur because of electrical defects at the inter­ cells has been limited due to conduction band offset into cuprous oxide
face of CdS/Cu2ZnSn(Sx,Se1− x)4 or cliff-like band alignment at the CdS/ (Cu2O) and buffer (n-type) material. This limits the Voc because of the
Cu2ZnSn(Sx,Se1− x)4 interface, a negative band offset, as shown in Fig. 6 negative conduction band offsets in regards to Cu2O. A 1D simulation
(b). In contrast, the Voc vs. T graph of the Cu(In,Ga)Se2 device showed an was used to examine the impact of various n-type materials on the open-
intercept that matched its bandgap well (1.14 eV) (Gunawan et al., circuit voltage. Between Zn(O,S), ZnO, ZnS, TiO2, (Mg,Zn)O, Ga2O3, and
2010). Hence, the efficiency of carrier collection is affected mainly by CdS, slightly positive conduction band offsets were observed in the Zn
the band alignment and interface defects at the CdS/Cu2ZnSn(Sx,Se1− x)4 (O,S) and Ga2O3, as shown in Fig. 8, and a high Voc was noted (Brandt
heterojunction interface, which adversely affects the device efficiency et al., 2014). Well studied and optimized interface (charge separating)
(Wang et al., 2014). Fig. 7(b) presents the higher carrier lifetimes in the energetics can help enhance the efficiency of different photovoltaic
case of the Cu(In,Ga)Se2 device compared to the Cu2ZnSn(Sx,Se1− x)4 devices (Kumar et al., 2015b). As mentioned above, considerable
device. Understanding the various interface recombination mechanisms research is still being conducted on the interface engineering of various
is essential for increasing the Voc and improving the efficiency of

Fig. 7. (a) Open circuit voltage as a function of temperature for Cu(In,Ga)Se2 and Cu2ZnSn(Sx,Se1− x)4 device. The activation energy for the recombination in
Cu2ZnSn(Sx,Se1− x)4 devices was smaller than their bandgap. (b) TRPL measurements show a shorter carrier lifetime for Cu2ZnSn(Sx,Se1− x)4 than Cu(In,Ga)Se2.
(Reprinted with permission from Ref. Gunawan et al., 2010).

24
M. Sahu et al. Solar Energy 230 (2021) 13–58

(DLTS), temperature-dependent photoluminescence spectroscopy (TPL),

and carrier lifetime measurement techniques help to understand the
physics of defects and defect clusters, their density, and energy states in
the bandgap of an absorber material.
Deep level transient spectroscopy (DLTS) is a widely used technique
to identify semiconductor deep level defects by measuring the concen­
tration, activation energy, and electron or hole capture cross-section for
each defect (Lang, 1974; Mulvey, 1964). This process works on the p-n
junction capacitance transients’ concepts because of the fixed temper­
atures filling and emptying of defects. Whether the deep defect is a trap
for minority or majority carriers can be determined by identifying the
sign of the capacitance-mode DLTS (C-DLTS) signal peak. The defect
concentration is directly proportional to the DLTS signal peak height,
Fig. 8. Band alignment of Cu2O (p-type) material with different n-type mate­
while the thermal emission rate of the defect helps determine the peak
rials with the possible error bars. (Reprinted with permission from Ref. Brandt
et al., 2014). position. Ln (en,p/T2) vs. 1/T is an Arrhenius curve of the corresponding
DLTS peak, where en,p is the defect carrier emission rate. The carrier
capture cross-section (σn,p) and activation energy (Ea) of the defect can
photovoltaic devices. The success of optimizing the interface quality in
be calculated from the intercept and slope of the Arrhenius plot,
the various photovoltaics is the real inspiration for the extensive dis­
respectively, as expressed in Eq. (2)
cussion of the various interfaces and their part in the solar cell perfor­ ( )
Eu
mance of Cu2ZnSn(Sx,Se1− x)4 devices. en,p (T) = γen,p T2 exp kT .(2) where γ is a constant of the material.
Das et al. used the current-mode DLTS (I-DLTS) method to examine the
5. Different simulation models and various characterization
two Cu2ZnSn(Sx,Se1− x)4 photovoltaic devices (Das et al., 2014). A
measurements for defects identification and loss management on
Cu2ZnSn(Sx,Se1− x)4 solar cell (S/Se = 1.4) with 4.1% efficiency
interfacial recombination process
exhibited two dominant deep acceptor states at Ev + 0.12 eV and Ev +
0.32 eV, which were assigned to the CuZn and CuSn antisites. In contrast,
5.1. Experimental and computational approached for detection of defects
the pure Cu2ZnSnSe4 device with 7.1% efficiency, exhibited a shallow
defect at Ev + 0.03 eV corresponding to the VCu level. The outcome of the
Kesterite chalcogenide materials have several types of defects that
pure Cu2ZnSnSe4 device showed that although the formation energy
have various impact mechanisms on the photovoltaic performance,
CuZn was lower, VCu may be the principal lattice defect in kesterite.
highlighting the need to characterize and control particular defects,
Nevertheless, the Cu2ZnSnSe4 device efficiency is still low, with only
specifically deep or killer defects. Various simulation methods and
benign defects. Qi et al. (Y. Qi et al., 2017) used a conventional C-DLTS
experimental characterization techniques have been used to identify the
method to identify CuZn and CuSn defects with Ea = 175 meV and 366
defects and interface loss in kesterite [Cu2ZnSnS4, Cu2ZnSnSe4, and
meV, respectively. Campbell et al. reported an Ea1 value of 86 ± 7 meV
Cu2ZnSn(Sx,Se1− x)4] device. Table 6 provides information on the defects
based on a broad DLTS signal at 115 K. They believed that the observed
of kesterite chalcogenide materials obtained by experimental methods.
peak might be contained in the close level distribution in energy, and
The influence of these points and complex defects on the device
attempts were made to isolate those states through Laplace analysis, but
performance must be correlated with the photo-generated carrier dy­
they were unsuccessful. Therefore, this peak was roughly recognized as
namics. Thermal admittance spectroscopy (TAS), transient photo-
a CuZn antisite. They also noted an energy state at Ea2 = 167 ± 10 meV
capacitance spectroscopy (TPC), deep level transient spectroscopy

Table 6
Defect states in Cu2ZnSn(Sx,Se1− x)4 materials obtained using different techniques.
Technique Activation energy Ea Concentration of defect Nt Carrier capture cross-section Critical temperature Τc Defect detected Ref
(meV) (cm− 3) σ (cm2) (μs)

DLTS 30 ± 10 6.46 × 1015 2.52 × 10− 20 166 VCu (Das et al., 2014)
120 ± 40 6.17 × 1014 1.31 × 10− 20 25 CuZn
320 ± 60 6.73 × 1014 2.04 × 10− 18 – CuSn
175 6.69 × 1014 1.06 × 10− 21 6.98 CuZn (Y. Qi et al., 2017)
366 9.46 × 1014 4.10 × 10− 18 – CuSn
86 ± 7 (1.70 ± 0.22) × 1014 (1.08 ± 0.95) × 10− 20
10,690 CuZn (Campbell et al.,
167 ± 10 (3.17 ± 0.32) × 1013 (2.20 ± 1.99) × 10− 20
– ZnSn 2019)
TAS 44.7 ± 0.7 – – – VCu (Fernandes et al.,
112.7 ± 3.5 – – – CuZn 2012)
130 − 200a – – – CuZn (Gunawan et al.,
2012)
134 4.96 × 1015 5.20 × 1017b 0.1 CuZn (Duan et al., 2013)
163 5.24 × 1015 1.56 × 1017b 0.47 CuZn
183 1.14 × 1016 8.00 × 1018b 0.42 CuZn
120 − 170c – – – CuZn (Larramona et al.,
600 ± 100 – – – VSn or CuSn 2015)
TPL 25 − 30c – – – Potential
fluctuation
120 − 30c – – – Donor
7±2 – – – Donor (Luckert et al., 2011)
27 ± 3 – – – Acceptor
277 ± 6 – – – CuSn or ZnSn (Grossberg et al.,
289 ± 7 – – – CuSn or ZnSn 2012)
38.3 ± 3.8 2.1 × 1018d – – VCu (Campbell et al.,
3.2 ± 0.5 – – – SnCu 2019)

25
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 9. (a) Arrhenius plots related to the signal from the I-DLTS spectra of device 1 for S/Se = 1.4 (b) and device 2 for pure Cu2ZnSnSe4 (c). (Reprinted with
permission from Ref. Das et al., 2014) (d) Arrhenius plots related to the signal from (e) C-DLTS spectra. (Reprinted with permission from Ref. Campbell et al., 2019)
(f) The lifetime of minority carriers and mobility as a function of the minority carrier diffusion length. (Reprinted with permission from Ref. Gokmen et al., 2013a).

[Fig. 9(d) and 9(e)], which was attributed to a tin-related defect. Cu2ZnSn(Sx,Se1− x)4 absorber materials await discovery. Thus far, more
Overall, the ZnSn was recognized according to the copper-poor and zinc- dominant non-radiative recombination centers (NRRCs) have not been
rich circulation state. Furthermore, additional shallower defects outside detected for some causes. Hence, newer and more independent methods
the range of DLTS have been measured by TPL, which will be com­ should be developed and exploited to examine the defect properties of
mented on below. Campbell et al. (Campbell et al., 2019) measured a Cu2ZnSn(Sx,Se1− x)4 absorber and establish a close relationship between
lifetime of 0.02 ns and a diffusion length of 203 nm of the minority the defects and non-radiation recombination.
carrier of a Cu2ZnSn(Sx,Se1− x)4 thin film through photoelectric mea­ TAS is also used to extensively estimate deep-level defects for
surements and analysis. These findings show an insufficient under­ Cu2ZnSn(Sx,Se1− x)4 (Kask et al., 2013; Yang et al., 2017). The Arrhenius
standing of the defects because there is a need for experimental correlation of TAS is between ln (ω0/T2) and 1/T (Walter et al., 1996),
identification of the defects. In many studies, similar energy states where ω denotes a characteristic frequency of the capacitance transition.
measured using the same technique on samples with the same homo­ Fernandes et al. fitted the complex admittance plane with an expanded
geneous bandgap have been shown to correspond to different defects. equivalent circuit [Fig. 10(a) and 10(b)] assuming two recombination
According to the stated data, as the density and carrier occupy a cross- centers (C1 - R1 & C2 - R2), where C1 - R1 and C2 - R2 are capacitor-
section of defects, recorded by DLTS, the shortest carrier lifetime of resistor pairs. The characteristic frequency is expressed as ω0 = 1/
Cu2ZnSn(Sx,Se1− x)4 can be calculated using Eq. (4), where τ, σn, vth, and (RiCi), i = 1, 2. Two defect states were measured using temperature-
Nt are the maximum lifetime of the minority carrier, electron capture dependent measurements, i.e., VCu at Ea1 = 44.7 ± 0.7 meV, and CuZn
cross-section, thermal velocity and concentration of defects, respec­ at Ea2 = 112.7 ± 3.5 meV [Fig. 10(c)]. With the help of TAS analysis,
tively. Table 6 lists the estimated carrier lifetimes τc (Miller et al., 2012; Gunawan et al. (Gunawan et al., 2012) and Duan et al. (Duan et al.,
Islam et al., 2015; Larramona et al., 2015; Das et al., 2014; Y. Qi et al., 2013) reported that in Cu2ZnSn(Sx,Se1− x)4, the Ea values (130– 200
2017; Campbell et al., 2019; Fernandes et al., 2012; Gunawan et al., meV) of CuZn defects also increases with increasing S/(S + Se) ratio
2012; Duan et al., 2013; Luckert et al., 2011; Grossberg et al., 2012). [Fig. 11(a) and 11(b)]. Gunawan et al. (Gunawan et al., 2012) assumed
Surprisingly, the values of τc obtained by DLTS were in the order of that the cause might be the concurrent reduction in the dielectric con­
microseconds or even milliseconds, which are several orders of magni­ stant of the crystal lattice constant while increasing anion replacement
tude higher than the recorded data (τm). From Fig. 9(f), the τm values from selenium to sulfur [Fig. 15 (c)]. Duan et al. (Duan et al., 2013)
obtained by time-resolved photoluminescence (TRPL) were mostly in assumed that the VBM decreases with increasing sulfur content, possibly
the range of ~1 ns to ~10 ns for the Cu2ZnSn(Sx,Se1− x)4 device having yielding higher activation energies defects. Recently, Enkhbait et al.
9%–11% efficiencies (Gokmen et al., 2013a; Gunawan et al., 2014). (Enkhbat et al., 2019) reported that the detected activation energy in­
Such a large difference suggests that some elemental properties of creases from 59.0 meV to 367.0 meV in a sulfur alloy. This shows that

Fig. 10. (a) AC response equivalent circuit of the Cu2ZnSnS4 PV device (b) Complex admittance plane of the Cu2ZnSnS4 PV device determined at 293 K and the fit for
the AC response equivalent circuits. (c) Arrhenius plot for estimating the activation energies EA1 and EA2 for defects. (Reprinted with permission from Ref. Fernandes
et al., 2012).

26
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 11. (a) Admittance signal of solar cells with varying sulfur contents: (S/(S + Se) = 0.35) for sample A, (S/(S + Se) = 0.8) for sample B, and (S/(S + Se) = 1.0) for
sample C, (b) Defect spectra estimated from (a) admittance signal. (Reprinted with permission from Ref. Duan et al., 2013) (c) The acceptor levels (dashed line) and
their dielectric constants are estimated from the admittance data. The blue-filled circles represent the theoretical data of the relative dielectric constants, and the star
points represent the experimental values. (Reprinted with permission from Ref. Gunawan et al., 2012).

the dominant defect level of Cu2ZnSn(Sx,Se1− x)4 is deepened at higher 100 meV and attributed it to CuSn or VSn. The smallest carrier recom­
sulfur contents and may clarify that high-efficiency Cu2ZnSn(Sx,Se1− x)4 bination lifetimes (Table 6), which are even higher than the τm values
devices are infrequently based on sulfur-rich material. Larramona et al. recorded by the TRPL, can also be evaluated from the published TAS
(Larramona et al., 2015) reported an ultra-deep defect at Ea = 600 ± data. Moreover, shallow defects, such as the CuZn, show a rapid charge

Fig. 12. (a) Temperature-dependent PL spectrum of

a Cu2ZnSn(Sx,Se1− x)4 thin film fabricated with 2%
Sn addition. (b) Arrhenius plots of the PL intensity
for Cu2ZnSn(Sx,Se1− x)4 thin films with different
levels of Sn addition, (Reprinted with permission
from Ref. Larramona et al., 2015) (c) LTPL spectra
of Cu2ZnSnS4 poly-crystals together with the fitting
result. (d) Arrhenius plots derived from the tem­
perature dependence of the PL spectrum of
Cu2ZnSnS4 poly-crystals. (Reprinted with permis­
sion from Ref. Grossberg et al., 2012).

27
M. Sahu et al. Solar Energy 230 (2021) 13–58

accumulation and must not be effective NRRCs because they cannot be different bandgaps. Remarkably, Grossberg et al. (Grossberg et al.,
activated simultaneously for both electrons and holes. On the other 2012) did not consider the activation energy of the low-temperature
hand, these defects can play an important role in the synergy process for system, but other reports found this value to be in the meV range. In a
charge recombination. low temperature system, the PL intensity measurement showed a rough
The defect levels were also detected using TPL. An association exists T− 2 dependence, similar to the relationship between the capture cross-
between temperature and steady-state PL intensity (Lee et al., 1995), as section of the defect and temperature (Krustok et al., 1997). Hence,
follows: the activation energy attained by TPL is not an accurate defect energy
[ ( ) ( )]
state.
IP,L (T) = I0 / 1 +A1 exp − EkTa1 +A1 exp − EkTa1 (3) where I0 is the
Different bandgaps and defect state distributions have been exam­
intensity at the lowest temperature; A1 and A2 are fitting parameters. In ined using TCP in Cu2ZnSn(Sx,Se1− x)4 (Cohen et al., 1992). Miller et al.
the original concept, Ea1 or Ea2 in Eq. (3) signifies the activation energy studied five Cu2ZnSn(Sx,Se1− x)4 devices with various (S/S + Se) ratios
of PL quenching due to a thermally activated non-radiation charge [Fig. 13(a) and 13(b)] (Miller et al., 2012).
transition similar to a non-radiative electronic state transition (Eb). In The TPC spectra show that for bandgaps smaller than 1.2 eV, the
some studies, these activation energies were attributed to the defect Urbach energy (EU) was within 18 meV, whereas for the higher bandgap
energy level. Whether this procedure is appropriate is a question. samples, the EU of band tail states was within range of 25 meV − 30
Nevertheless, some studies are reviewed. Laramona et al. (Larramona meV. In addition, most of the deep defects in the samples were centered
et al., 2015) detected two defect states using TPL, which were not at EV + 800 meV. Islam et al. (Islam et al., 2015) examined two samples
identified by TAS [Fig. 12(a) and 12(b)]. One was considered a potential [Fig. 13(c) and 13(d)]; their EU values were 54 meV for pure Cu2ZnSnS4
fluctuation with a low activation energy (25 meV–30 meV), while the and 69 meV for Cu2ZnSn(Sx,Se1− x)4 (S/(S + Se) = 0.15). In the
other was classified as a donor defect with a high energy state (120 Cu2ZnSnS4 sample, a deep defect centered at approximately 1.0 eV was
meV–130 meV). observed above the VBM. Currently, experimental measurements of
Campbell et al. (Campbell et al., 2019) measured the shallow donor defects in Cu2ZnSn(Sx,Se1− x)4 suggest that the Urbach energy or acti­
at 3.2 ± 0.5 meV and the shallow acceptor at 38.3 meV for SnCu and VCu, vation energy of shallow defects increases with increasing bandgap,
respectively. None of them were measured using the DLTS technique. whereas the location of the deep defect remains unchanged, as shown in
Furthermore, Campbell et al. (Campbell et al., 2019) reported a donor Fig. 14 (Levcenko et al., 2016).
acceptor-pair (SnCu and VCu) density value of 2.1 × 1018 cm3 from the a) The different Ea values in this range are for Cu2ZnSn(Sx,Se1− x)4
blue-shift of excitation-dependent PL spectra. Luckert et al. reported two films with various bandgaps; b) the σ values are estimated from the
Ea1 = 7 ± 2 meV and Ea2 = 27 ± 3 meV for the pure Cu2ZnSnSe4 (Luckert Arrhenius curves; c) the different Ea values in these ranges are for
et al., 2011). Later, Grossberg et al. (Grossberg et al., 2012) first detected Cu2ZnSn(Sx,Se1− x)4 films with various levels of Sn addition; d) the total
deep defects in pure Cu2ZnSnS4 by TPL. The two PL signals [Fig. 12(c)] density of the two shallow defects was calculated from the PL peak blue-
had comparable activation energies of 287 ± 7 meV and 277 ± 6 meV in shift value in excitation-dependent PL spectra.
their study [Fig. 12(d)]. They assumed that both of these peaks appeared The activation energy obtained was smaller than 200 meV in most
due to the impurity recombination process of bands of similar deep studies, and very few deep defects were observed. From the authors’
acceptor defects (CuSn or ZnSn) but different Cu2ZnSnS4 phases with point of view, these results on point and complexes defects are

Fig. 13. (a, b) Transient photo-capacitance

(TPC) spectrum of the Cu2ZnSn(Sx,Se1− x)4
samples. The bandgaps of samples 1 to 5 are
1.07, 1.15, 1.20, 1.47, and 1.51, respectively.
(Reprinted with permission from Ref. Miller
et al., 2012) TPC spectra of Cu2ZnSnS4 (c)
and Cu2ZnSn(Sx,Se1− x)4 [S/(S + Se) = 0.15]
(d) PV device. The solid black line was fitted
using the Gaussian function of the defect
band, OH1. Insets in figures show the sub-
bandgap optical transitions in the films.
(Reprinted with permission from Ref. Islam
et al., 2015).

28
M. Sahu et al. Solar Energy 230 (2021) 13–58

quaternary chalcogenide system. Chen et al. investigated the correlation

between defects (intrinsic point and complex) and electronic properties
of ternary (CuInSe2) and quaternary (Cu2ZnSnS4 and Cu2ZnSnSe4)
compounds comprehensively using first-principles and DFT calcula­
tions. For the quaternary compounds, Cu2ZnSnS4 and Cu2ZnSnSe4, the
chemical potential region boundary are more complex and narrower (M
- N - P - Q), as shown in Fig. 15(a) and 15(b), where a pure phase can be
obtained for the quaternary material in the absence of secondary phases
(Chen et al., 2013).
Under copper-poor conditions for the Cu2ZnSnS4 material, the
theoretically predicted chemical potential area was narrow compared to
the ternary CuInSe2 material: μCu = 0 to − 0.4 eV (Chen et al., 2010b).
The calculated chemical potential for Cu2ZnSnS4 agrees with the re­
ported experimental studies of secondary phase competition in
Cu2ZnSnS4 bulk material and thin films (Just et al., 2011; Fontané et al.,
2011; Scragg et al., 2012b). On the other hand, the stable chemical-
potential region covers copper-poor chemical-potential zones to fabri­
Fig. 14. Schematic band diagram of the defect levels for Cu2ZnSn(Sx,Se1− x)4
cate a pure phase absorber material with p-type properties (Chen et al.,
solar cells. (Reprinted with permission from Ref. Levcenko et al., 2016).
2013; Chen et al., 2010b).
The quaternary material can accommodate a minimum of 14
insufficient to elucidate the lifetime of ultra-short-term short-lived car­ intrinsic point defects with respect to the computation defect formation
riers in Cu2ZnSn(Sx,Se1− x)4 absorber materials. Some hidden informa­ energy. In these intrinsic point defects in the electronic structure,
tion may be important for understanding the mechanism for the Voc negatively charged point defects, i.e., VCu, VZn, VSn, CuZn, ZnSn, and
losses and efficiency in the Cu2ZnSn(Sx,Se1− x)4-based solar devices. CuSn, have acceptor transition energy states. In contrast, positively
Very little theoretical investigation has been done to examine the charged point defects, i.e., ZnCu, SnZn, SnCu, Cui, Zni, Sni, and VS or VSe
formation of point and complex defects and their effects on kesterite have donor transition energy states. Fig. 15 (c) and 15(d) and Fig. 16
materials and the device performance owing to the complexity of the show that in a copper-poor composition range, acceptor-like defects

Fig. 15. Estimated stable chemical potential

spaces (black area) of Cu2ZnSnS4 (left-side)
and Cu2ZnSnSe4 (right-side) in the planes of
(μZn, μSn) with various μCu in the (μCu, μZn,
μSn) chemical potential region. (Reprinted
with permission from Ref. Chen et al., 2013),
Change in the defect formation energy versus
Fermi energy at the chemical potential point
P in Cu2ZnSnS4 (c) and Cu2ZnSnSe4 (d). Only
the most stable charge state is plotted for a
similar Fermi energy. The charge state
changes at the circles (open for acceptors and
filled for donors), indicating the transition
energy levels. (Reprinted with permission
from Ref. Chen et al., 2013).

29
M. Sahu et al. Solar Energy 230 (2021) 13–58

absorber compounds. The theoretical (-/0) transition energy level of VCu

(ionized from neutral to − 1 charged) for Cu2ZnSn(Sx,Se1− x)4 is 30 meV
shallower, which is the same as CuGaSe2 and CuInSe2. For Cu2ZnSnS4,
the value is 10 meV above the valence band maximum (VBM). The (-/0)
level of CuZn is 0.11 eV for Cu2ZnSnSe4 and 0.15 eV for Cu2ZnSnS4
(according to 128-atom supercell calculation). Both values are deeper
compared to VCu. As shown in Fig. 17, other acceptor defects, such as
CuSn, ZnSn, VZn, and VSn have a larger formation energy and ionize into
higher valence states. The transition between the acceptor levels related
to them, specifically between high-charge states, such as (4-/3-) and (3-/
2-), is deeper. Therefore, their influence on the observed p-type behavior
may be insignificant in pure-phase kesterite absorber materials. On the
other hand, they can help with luminescence and act as a trap or
recombination center. As expected, in support of the donor defect, the
shallow ZnCu anti-site has a lower formation energy because of the much
smaller size and the electronic difference between Cu+ and Zn2+, as
shown in Fig. 17. In contrast, all other donor defects (SnCu, SnZn, Zni, VS,
and VSe) are found at deeper states in the bandgap of Cu2ZnSnS4 and
Cu2ZnSnSe4 kesterite material. These defects act as recombination or
trap centers, limiting the performance of kesterite (Cu2ZnSnS4 and
Cu2ZnSnSe4)-based solar devices. Thus, the Sn-abundant chemical po­
tential zone must avoid reducing the formation of deeper defects of SnCu
and SnZn. The formation of VS in Cu2ZnSnS4 was attributed to the vol­
atile nature of elemental sulfur during the high-temperature process.
Nevertheless, VS is highly unstable in its charged states, and the validity
of the tin atom decreased due to weak hybridization between VS and the
adjacent tin atoms (Zhang et al., 2016a, 2016b, 2016c). In addition, the
Fig. 16. Ionization states of defects in the bandgaps of Cu2ZnSnS4 (top) and
formation energy of positively charged VS2+ in Cu2ZnSnS4 was less than
Cu2ZnSnSe4 (bottom). The red bars represent the acceptor states, and the blue
bars represent the donor states, with their primary and final charge states 0.8 eV in the p-type samples. Hence, the Cu2ZnSnS4 absorber material
labelled in brackets. The estimated experimental bandgaps were validated by has a sufficient quantity of deep VS donor level when the ratio of cation/
DFT values of 1.5 eV and 1.0 eV, respectively. (Reprinted with permission from anion is greater than unity. This suggests that optimization of the sul­
Ref. Chen et al., 2013). furization process may be important for improving the performance of
the Cu2ZnSnS4 device (He et al., 2013). In contrast, a similar situation
(VZn, CuSn, and ZnSn) show a comparatively larger formation energy exists for selenium vacancy in Cu2ZnSnSe4 as for CuGaSe2 and CuInSe2
(Chen et al., 2013; Chen et al., 2010b). (Lany and Zunger, 2005; Shirakawa and Kukimoto, 1980). The calcu­
On the other hand, VCu and CuZn showed smaller formation energy, lated levels of selenium vacancy in CuGaSe2 and CuInSe2 are both deep,
reflecting the dominating nature of the VCu and CuZn defects, which but their formation energy (whether the charged or neutral state) is high
determine the p-type conductivity in the Cu2ZnSnS4 and Cu2ZnSnSe4 enough. Therefore, their involvement can be ignored. A comparison of

Fig. 17. Formation energy of lower energy defects in Cu2ZnSnS4 (a) and Cu2ZnSnSe4 (b) as a function of the chemical potential with the stable region at the
APQMNPG lines. The Fermi energy is assumed to be on top of the valence band (p-type nature), and the donor defect is completely ionized. (Reprinted with
permission from Ref. Chen et al., 2013), Estimated valence band and conduction band shifts due to different defect complexes in Cu2ZnSnS4 (c) and Cu2ZnSnSe4 (d).
The defect concentration is one defect complex in a 128-atom supercell. The valence and conduction band positions are represented as red and green lines,
respectively. (Reprinted with permission from Ref. Chen et al., 2013).

30
M. Sahu et al. Solar Energy 230 (2021) 13–58

the formation energies of the two-deep donor defects, such as charged binary phase diagram of the Cu2Se–In2Se3 system, with a chalcopyrite
SnZn antisite and anion vacancies in Cu2ZnSnSe4 and Cu2ZnSnS4, structure. Both Cu2ZnSnS4 and Cu2ZnSnSe4 with a kesterite structure
showed that they could be less than 1.0 eV in Cu2ZnSnS4 [Fig. 17(a)], have a solid solution area but smaller than CuInSe2 (T. Maeda et al.,
but much larger in Cu2ZnSnSe4 [Fig. 17(b)]. This may explain why the 2011). First-principles theory was used to calculate the formation en­
efficiencies of Cu2ZnSnSe4 and Cu2ZnSn(Sx,Se1− x)4 alloy photovoltaic ergy, whereas the hybrid density functional (HSE06) was used to esti­
devices have been higher than the Cu2ZnSnS4 device. In Cu2ZnSnS4 and mate the localized donor state (eigenvalue) in Cu2ZnSnS4 and
Cu2ZnSnSe4 [Fig. 17(a) and 17(b)] different self-compensated and self- Cu2ZnSnSe4. These results show that compensation between the
passivated a deep-level defects (also called defect complexes) can be acceptor and donor defects, such CuZn and SnZn, could reduce the for­
formed, such as [2VCu++SnZn2+], [CuZn− +ZnCu+], [VCu− +ZnCu+], mation energies of complex defects, such as [CuZn + SnZn] and [2CuZn +
[2CuZn− +SnZn2+], [ZnSn2− +SnZn2+], and [ZnSn2− +2ZnCu+]. Fig. 17 SnZn]. Even in stoichiometric samples, Cu/(Zn + Sn) = 1.0 and Zn/Sn =
shows the defect complexes, whose total formation energy is less than 1.0 have a higher concentration of passivated defect complexes. In
1.5 eV, and Fig. 15(a) and 15(b) presents their formation energy as a Cu2ZnSnS4, partially passivated defect complexes, such as CuZn + SnZn,
function of the chemical potential (μ) along the A - P - Q - M - N - P - G form a deep donor state while reducing the bandgap due to a fully
lines surrounding the stable region. In Cu2ZnSnS4 and Cu2ZnSnSe4, both passivated defect complex, such as [2CuZn + SnZn]. On the other hand,
defects CuZn and ZnCu are the lowest-energy acceptor and donor defects, the detrimental effects of [CuZn + SnZn] and [2CuZn + SnZn] complexes
respectively. Their passivation forms the antisite pair [CuZn + ZnCu], are weaker in Cu2ZnSnSe4 material. In stable chemical potential defects
which has a much lower formation energy of 0.2 eV. While the forma­ clusters [CuZn + SnZn] and [2CuZn + SnZn] with a low formation energy
tion energy of the [VCu + ZnCu] defect in Cu2ZnSnS4 and Cu2ZnSnSe4 is have population in the order of 107–1015 cm− 3 and 1010–1018 cm− 3 in
0.4 eV and 0.3 eV, respectively. Both anti-site defect complexes in the Cu2ZnSnS4, respectively. In Cu2ZnSnSe4, however, the population of
chemical potential range can have a high population of 1016–1018 cm− 3. [CuZn + SnZn] and [2CuZn + SnZn] were in the order of 105–109cm− 3 and
The influence of the CuZn/VCu acceptor and ZnCu donor on the 1015–1019cm− 3, respectively. Copper and tin-poor and zinc-rich condi­
electronic structure and optical properties is weak, despite the high tions are required to decrease their population in the sample. The SnZn
population of both antisite defects in the sample, DFT calculations (128- made donor defect has a localized donor state (eigenvalue) of approxi­
atom supercell) predicted a slight band edge shift [Fig. 17(c) and 17(d)]; mately 0.35 eV, 0.23 eV, and 0.12 eV below the CBM for SnZn, [CuZn +
hence, they can be considered as benign defects. On the other hand, SnZn], and [2CuZn + SnZn] complexes, respectively. In contrast, in the
another antisite defect pair, such as [CuSn + SnCu], [ZnSn + 2ZnCu], and Cu2ZnSnS4 localized donor state (eigenvalue), it is approximately 1.03
[ZnSn + SnZn], are not benign and have a more important effect on the eV, 0.63 eV, and 0.20 eV for SnZn, [CuZn + SnZn] and [2CuZn + SnZn]
electronic structure, e.g., reducing the [ZnSn + SnZn] bandgap to 0.3 eV complexes, respectively (Fig. 18).
in Cu2ZnSnS4 and 0.1 eV in Cu2ZnSnSe4. Fortunately, their formation This result suggests that in Cu2ZnSnSe4, the effects of both passivated
energy is higher. As a result, the population can be ignored (Fig. 17). In defects, [CuZn + SnZn] and [2CuZn + SnZn] complexes, are weaker than
particular, the formation energy of the [2CuZn + SnZn] pairs can be in Cu2ZnSnS4. This explains why higher efficiency is recorded when the
reduced from 0.6 eV to 0.2 eV under specific chemical potential fields selenium element composition is higher. These complex defects have
(Q–M− N− P lines in Fig. 17), which corresponds to defect pair density more adverse effects in Cu2ZnSnS4 that limit the efficiency in the
ranging from 1011 to 1018 cm− 3 (Chen et al., 2013; Chen et al., 2010b). Cu2ZnSn(Sx,Se1− x)4 alloy, even though with a higher sulfur composition,
First-principles electronic structure (DFT) calculations have been used to the bandgap increases to near the optimal range (1.4 eV–1.5 eV) ac­
examine the properties of defects in the Cu2ZnSnS4 semiconductor ma­ cording to the Shockley-Queisser model, but the efficiency decreased
terial. The following observations were made using first-principles cal­ with increasing sulfur content because of the formation of deep trap
culations: control of chemical potential is crucial for the growth of a levels that limit the carrier lifetime. All calculations were done using
good quality crystal without the presence of other phases and low defect density functional theory [semi local generalized gradient approxima­
density, CuZn antisite, responsible for p-type conductivity; n-type doping tions (GGA) with the semi-local Perdew-Burke-Ernzerhof (PBE) and
is very difficult because of low formation energies of acceptor defects; hybrid Heyd-Scuseria-Ernzerhof (HSE06) functional]. The results that
the formation of electrically neutral defect complexes is easy because of VCu is the dominant vacancy with a low formation energy in Cu2ZnSnSe4
their low formation energies, and they passivate deep donor levels in the compared to Cu2ZnSnS4. The formation energy of VCu for Cu2ZnSnS4
bandgap (Chen et al., 2010a). was calculated to be 0.74 eV and 1.70 eV, respectively, of PBE and
The first-principles pseudopotential was calculated using plane-wave HSE06, while for Cu2ZnSnSe4 is estimated by 0.46 eV and 1.37 eV PBE
basis functions. Under copper-poor and zinc-rich conditions, the energy and HSE06, respectively. The defect studies show that cation vacancies
formation of neutral vacancies (VCu = +0.37, VZn = +0.49, VSn = +2.98 have higher formation energies in Cu2ZnSnS4 than Cu2ZnSnSe4 (Kumar
for Cu2ZnSnS4, VCu = +0.19, VZn = +0.27, VSn = +2.51 for Cu2ZnSnSe4 and Persson, 2013). A canonical approach was used to estimate the
and VCu = -0.81, VIn = +0.37 for CuInSe2) was estimated as a function of concentrations of defects (neutral and charged) in Cu2ZnSnS4 and
the chemical potential of the constituent elements. The following ob­ Cu2ZnSnSe4 at fixed temperatures as a function of the stoichiometry and
servations were made. The Cu vacancy has a lower formation energy chemical potentials based on the model. Zinc-rich and copper-poor
than zinc and tin vacancies in both Cu2ZnSnS4 and Cu2ZnSnSe4. Copper, conditions are required for a stable Cu2ZnSnS4 compound with off-
zinc, and tin vacancies have higher formation energies in Cu2ZnSnS4 stoichiometric compositions, while the conformational space for a sta­
than Cu2ZnSnSe4, while the formation energy of sulfur vacancies is ble Cu2ZnSnSe4 material is broader than that of Cu2ZnSnS4. Four de­
lower than selenium vacancies. In CuInSe2, however, VCu has a lower fects, which are neutral CuZn and negatively charged VCu defects,
formation energy than Cu2ZnSnS4 and Cu2ZnSnSe4. These results passivated defects clusters [VCu + ZnCu], and [ZnSn + 2ZnCu], dominate
explain why it is difficult to form VCu in CuInSe2 compared to Cu2ZnSnS4 in stable Cu2ZnSnS4 and Cu2ZnSnSe4. A separate set of formation en­
and Cu2ZnSnSe4. The displacement of the surrounding sulfur or sele­ ergies for the DFT calculation was used, which leads to a high concen­
nium after vacancy formation showed that in Cu2ZnSnS4, |Δl| is higher tration of positively charged ZnCu defects (Mutter and Dunham, 2015).
when VZn = -0.063 Å and VSn = -0.089 Å compared to |Δl| of VCu = Li et al. compared the electron capture abilities of the three tin induce
-0.016 Å. In Cu2ZnSnSe4 also |Δl| of VZn − 0.036 Å and VSn − 0.096 Å is defects (i.e., SnZn2+, SnZn+, and [CuZn − SnZn]+) in Cu2ZnSnS4 theoret­
larger than |Δl| of VCu − 0.035 Å. Hence, the Cu2ZnSnS4 (Cu2ZnSnSe4) ically (Li et al., 2019). They reported that the bond strength of Sn-S plays
crystal structure is established by Sn–S (Sn–Se) chemical bonding. The an important role in the carrier capture cross-section around the defect.
displacement |Δl| − 0.006 is smaller in CuInSe2 because the formation At 300 K, the electron capture cross-sections for SnZn2+, SnZn+, and
energy of VCu is lower than Cu2ZnSnS4 (Cu2ZnSnSe4). Therefore, the [CuZn + SnZn]+ were predictable at 8.8 × 10− 14, 4.7 × 10− 18, and 4.5 ×
CuInSe2 has a high copper-poor solid solution region in the pseudo- 10− 14 cm2, respectively. SnZn2+ and [CuZn + SnZn]+ defects can be

31
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 18. (a) Electron density iso-surface of the

[CuZn + SnZn] donor state (iso-value 25% of the
maximum value). The green, blue, gray, and yellow-
filled circles represent copper, zinc, tin, and sulfur,
respectively, and the black circles indicate the CuZn
and SnZn defect sites.; (b) Estimated DOS of
Cu2ZnSnS4 (upper) and Cu2ZnSnSe4 (bottom) with a
[CuZn + SnZn] defect complex in a 128-atom
supercell. The total DOS (blue lines) and the pro­
jected DOS (red line) on tin 5 s orbital at the SnZn
site (multiplied by 128 for clarity). The energy is
relative to the VBM of Cu2ZnSnS4 (upper) and
Cu2ZnSnSe4. (Reprinted with permission from
Ref. Li et al., 2019).

evaluated as effective NRRCs. Furthermore, studies have shown that of SnZn+, the Sn–S bond around the defect becomes longer and softer.
SnZn2+ (2+/+) and [CuZn + SnZn]+ (+/0) have relatively high lattice Hence, its electron capture cross-section becomes much smaller. This
relaxation energies (Erel, 0.50 eV and 0.52 eV, respectively) because of a study presents a theoretical approach to analyze whether a deep defect is
small electronic state transition (Eb, 0.01 eV, 0.02 eV), allowing easier an effective NRRC. On the other hand, it is still difficult to conduct this
electron capture. In comparison, SnZn+ (+/0) exhibits a relatively small evaluation in complex situations where multiple defects can interact (Li
Erel (SnZn+) of 0.23 eV and thus a higher Eb (SnZn+) of 0.44 eV. Coor­ et al., 2019). S. Kim et al. (2019) suggested that the rapid recombination
dinated diagrams of SnZn2+ (2+/+) and SnZn+ (+/ 0) were assigned, as of charge carriers through NRRCs is associated with the (5 s2) lone pair
shown in Fig. 19 (a-c). Li et al. (Li et al., 2019) explored these results in electronic state on tin. First, the vacancy of sulfur (VS) could be an active
depth. The lattice near the antisite defects was distorted because of the NRRC with the support of an intermediate excited state, where tin can
difference in the ionic size and valency between the original and the occupy the electron (Kim et al., 2018a, 2018b). When neutral VS and Sn
substitutional atoms. In these three antisite defects, each tin atom co­ (IV) are close to each other, the VS 0 capture electrons and become VS+,
ordinates with four sulfur atoms. Lattice disorder arises when sulfur while Sn (IV) forms Sn (III), a metastable state after the electron is lost.
atoms rest toward or away from the defects. As shown in Fig. 19 (d-f), for The vacancy level of VS+ is near the VBM of Cu2ZnSnS4. Therefore, it can
SnZn2+ and [CuZn + SnZn]+, the Sn-S bonds become smaller and stronger occupy a hole under illumination and transform into VS2+. Hole
as the sulfur atom corresponds to the tin atom corresponding to the capturing stimulates the energy level of VS2+ near the CBM of
higher-frequency vibrational phonons. From Eqs. (5) and (6), it can be Cu2ZnSnS4. VS2+ can effectively occupy one electron, return to the VS1+
estimated that they have lower Eb (non-radiative electronic state tran­ charge state, and complete the non-radiative electron-hole recombina­
sition) and higher Erel (lattice relaxation energy) (where ωk is the fre­ tion process. In addition, under sulfur-poor conditions VS, [VS + CuZn],
quency of the defect correlated phonon modes and ΔQk is the SnZn defect have been identified, whereas the SnZn defect has been
coordination difference between the initial and final states). In the case studied under sulfur-rich conditions. They are active NRRCs and limiting

Fig. 19. (a) Schematic diagram of NNR prompted

by a killer or deep defect. Configuration coordinate
diagram (under the harmonic approximation) of the
non-radiative electronic state transition for (b)
SnZn2+ state capture an electron and convert to
SnZn+ state, (c) SnZn+ state capture an electron and
convert to neutral SnZn0. The non-radiative elec­
tronic transition state commonly arises at the cross
point of electronic energy curves. A system needs to
exceed the energy barrier Eb with the help of
phonon. ΔE = Ei0 − Ef0 is the energy difference at
the equilibrium states, and ΔQ correlates with the
difference in the equilibrium coordinates between
the initial and final states. The lattice relaxation
energy, Erel, arises from the configuration coordi­
nate of equilibrium between the initial and final
states. The defects are responsible for the phonon
modes and the electronic states. (d) SnZn2+, (e)
SnZn+, and (f) [CuZn − SnZn]+. The green iso-surfaces
show squared norms of wave functions of the tran­
sition energy levels of defects. In contrast, the vi­
bration of the sulfur element (in yellow) is denoted
by the red arrows near the SnZn anti-site (in blue)
correlated with the defect-induce phonon modes.
(Reprinted with permission from Ref. Li et al.,
2019).

32
M. Sahu et al. Solar Energy 230 (2021) 13–58

the lifetime of the minority charge carrier, and all of them are related to Se1− x)4 cells exhibited a low series resistance of 0.4–0.6 Ω cm2. The
the 5 s2 lone-pair state of tin. In Cu2ZnSnS4, the oxidation state of tin was absorber-Mo, back interface requirements may be more involved in
reduced from +IV to +II by forming an inert lone pair (5 s2). The ionic Cu2ZnSn(Sx,Se1− x)4 than Cu(In1− x,Gax)(Sy,Se1− y)2 devices (Polizzotti
radii of Sn(II) are larger than Sn(IV). Therefore, these ionic radii dif­ et al., 2013).
ferences make large lattice deformation around the lone-pair defect
centers. The lone pair stability suppresses the ionization of some 5.2.1. Optimization of the band alignment at the interface of buffer/
intrinsic defects and forms a deep carrier transition level in the bandgap. absorber
Thus, the lone-pair effect is related to tin, which results from deep defect
levels and large lattice distortions in Cu2ZnSnS4, and the rapid electron- 5.2.1.1. CdS/Kesterite [Cu2ZnSnS4, Cu2ZnSnSe4 and Cu2ZnSn(Sx,
hole non-radiative recombination (Kim et al., 2018a, 2018b; S. Kim Se1− x)4] interface. Using the chemical bath deposition method, CdS is
et al., 2019). generally used as an n-type buffer layer and is deposited on the p-type
τn = vth σ1n Nt (4) Cu2ZnSn(Sx,Se1− x)4 absorber layer to make a p-n junction, as in the Cu
(ΔE− Erel )2 (In,Ga)Se2 device. The world record Cu2ZnSn(Sx,Se1− x)4 device reported
Eb = 4E (5)
∑ rel an efficiency of 12.6% using CdS as the buffer layer (Wang et al., 2014).
Erel = k 12w2k ΔQ2k (6) To explain why the Voc losses of Cu2ZnSn(Sx,Se1− x)4 solar cells are so
large, and to improve the efficiency, it is important to understand band
5.2. Experimental and computational approach to the identification of alignment at the CdS/Cu2ZnSn(Sx,Se1− x)4 interface. Table 5 lists the
loss management on interfacial recombination: CBO at CdS/Cu2ZnSnS4 and CdS/Cu2ZnSnSe4 interface. Bär et al.
examined the interface at the CdS/Cu2ZnSnS4 heterojunction by X-ray
The mismatches in the chemical and physical properties of two photoelectron spectroscopy (XPS), ultraviolet photoelectron spectros­
materials can affect the charge carrier transport across the interfaces copy (UPS), and inverse photoelectron spectroscopy (IPES) (Bär et al.,
directly and reduce the device efficiency. The band alignments can be 2011). They reported − 0.33 eV and − 1.2 eV for CBO and VBO,
divided into two kinds, where the conduction bands of the buffer layer respectively, at the CdS/Cu2ZnSnS4 interface. Therefore, the CBM
are lower than those of the edge absorber layer, the potential difference measurement for Cu2ZnSnS4 and CdS confirmed the cliff-like CBO.
arising between the quasi-Fermi levels of the buffer layer and the Haight et al. reported a spike-like band offset for Cu2ZnSn(Sx,Se1− x)4
absorber layer under illumination will be minimized. Furthermore, using ultraviolet photoelectron spectroscopy (UPS). The band alignment
under the forward bias, the cliff barrier will impede the flow of injected was examined as a function of the S/(S + Se) ratio between Cu2ZnSn(Sx,
electrons and lead to an increase in recombination between the majority Se1− x)4 and CdS [Fig. 20(a)]. Their findings showed that the valence
charge carriers through defects at the interface of buffer/absorber. As a band (VB) edge decreased with increasing sulfur content in the S/(S +
result, the Voc and FF will decrease. On the other hand, spike-like band Se) ratio. In contrast, there was a very slight dependence of conduction
alignment, where the conduction band of the buffer layer is above the band minimum (CBM) on the sulfur content. As a result, the VBO at the
edge absorber layer, is less damaging to the Voc, and the Voc is almost interface of CdS/Cu2ZnSn(Sx,Se1− x)4 decreased with increasing sulfur
stable despite the increase in CBO. Furthermore, the simulations sug­ content from 0.97 eV to 0.54 eV, whereas CBO was within the range of
gested that Jsc is almost stable when CBO is <0.4 eV. Although the CBO 0.4 eV–0.5 eV, which is spike-like and slightly larger than the optimal
is very large (>0.4 eV), the photo-generated electron carriers cannot value (0–0.4 eV) (Haight et al., 2011).
cross the spike barrier when the Jsc is reduced suddenly. Thus, good Later, Tajima et al. measured VBO by X-ray photoelectron spec­
band alignment at the buffer/absorber interface is critical for highly troscopy (HAXPES) at the CdS/Cu2ZnSnS4 interface and reported that
efficient devices, and the optimal range of CBO is 0.0–0.4 eV (Huang
et al., 2014; Minemoto et al., 2001; Gao et al., 2018a). In this paper, the
major focus was on identifying the route for high-efficiency (>18%)
solar cells, which must include optimal conduction band alignment and
associated lower interface recombination currents. Thus, valence band
alignment is the critical subject for such solar cells instead of the con­
duction band. Cu2ZnSn(Sx,Se1− x)4 devices should have to optimize the
front and rear contacts interface to maximize the efficiency by reducing
the various interface losses, such as resistive, recombination, and
parasitic absorption.
In addition, the formation of a secondary phase and its role in the
rear contact is still unclear. At the back contact, the growth of secondary
phases, such as SnS2, ZnS, ZnSe, and Cu2SnS3, have been reported
(Repins et al., 2011; Wang et al., 2011; Scragg et al., 2012b; Ennaoui
et al., 2009), which may have inferred device resistance, film develop­
ment, or stability. It is unclear if the high series resistance (>3–4 Ω cm2)
observed in higher-bandgap (~1.3 eV or higher) Cu2ZnSn(Sx,Se1− x)4
devices (Mitzi et al., 2011; Shin et al., 2013b) are associated with back
contact or bulk of the film. In addition, during annealing, the formation
of MoSe2 (the layer thickness for MoSx is >100 nm and for MoSex >
1000 nm has been reported) in Cu2ZnSnS4 and Cu2ZnSnSe4 solar cells
have been reported (Wang et al., 2011; Gunawan et al., 2010; Shin et al.,
2012), which affect the device efficiency and Voc (Shin et al., 2012). For
the Cu(In,Ga)Se2 device, a MoSe2 layer thickness of approximately 100
nm was observed by secondary-ion mass spectrometry (SIMS) depth
profiling. Good ohmic contact was noted at the interface of the Fig. 20. (a) Spike-like conduction band offset with varying S/(S + Se) ratios.
absorber/back contact. Thus, the world-record Cu(In1− x,Gax)(Sy,Se1− y)2 (Reprinted with permission from Ref. Haight et al., 2011), (b) Band alignment
cells exhibited a series resistance less than ~0.5 Ω cm2 (Jackson et al., at the CdS/Cu2ZnSnS4 interface (Reprinted with permission from Ref. Tajima
2011; Repins et al., 2008). In contrast, the world-record Cu2ZnSn(Sx, et al., 2013).

33
M. Sahu et al. Solar Energy 230 (2021) 13–58

CBO decreased to 0.0 eV when the VBO value was ca. 1.0 eV [Fig. 20(b)] large barrier and impedes the flow of electrons. Hence, the photo-
(Tajima et al., 2013). Li et al. examined the CBO of Cu2ZnSnS4 and current approaches zero. Although the CBO at the ZnO/Cu2ZnSn(Sx,
Cu2ZnSnSe4 interfaces and observed − 0.06 eV for CdS/Cu2ZnSnS4 and Se1− x)4 interface was cliff-like [Fig. 21(b)], the photo-current with the
+0.034 eV for CdS/Cu2ZnSnSe4 (Li et al., 2012; J. Li et al., 2013). ZnS buffer layer was much higher, and the Voc and FF were much lower
Santoni et al. measured the VBO at the interface of CdS/Cu2ZnSnS4 by than the CdS buffer layer. The band structure of Zn(O,S) can be
XPS and estimated the CBO to be − 0.3 eV (Santoni et al., 2013) Udaka controlled by the O/(S + O) ratio, as shown in [Fig. 21(c)] (Sharbati and
et al. also studied the band alignment at the interface of CdS/Cu2ZnSn Sites, 2014; Sharbati et al., 2015).
(Sx,Se1− x)4 (x = 0, 0.3, and 1) by XPS, UPS, and IPES (Udaka et al., Therefore, it is vital to optimize the band alignment at the Zn(O,S)
2017). In contrast to Haight et al. (Haight et al., 2011), they reported and Cu2ZnSn(Sx, Se1− x)4 interface by changing the O/(S + O) ratio.
that the CBM of Cu2ZnSn(Sx,Se1− x)4 was increased markedly with sulfur Neuschitzer et al. investigated the impact of thiourea during the
substitution. The CBM of Cu2ZnSnS4 was 0.40 eV–0.45 eV higher than chemical bath deposition procedure on the features of Zn(O,S) films and
Cu2ZnSnSe4, while the VBM decreased to below 0.07 eV. Consequently, their influence on the performance of the Cu2ZnSnSe4-based devices
the conduction band offset (CBO) at the CdS/Cu2ZnSn(Sx,Se1− x)4 (Neuschitzer et al., 2016a). The J-V curve (Fig. 22) was lump-like in
interface decreased from +0.55 eV to − 0.14 eV with sulfur replacement shape, and the deformations of the J-V curves were less evident with a
of selenium. At x = 0, the CBO of +0.55 eV at the CdS/Cu2ZnSnSe4 decrease in thiourea content. The deformation in the J-V curves may be
interface was a somewhat high spike, which will hinder the flow of because of the larger conduction band spikes, which act as a barrier for
photo-generated charge carriers (electron), leading to a heavy loss of electron flow (Neuschitzer et al., 2016a). In the Zn(O,S) buffer layer, the
current. At x = 1, however, the CBO of − 0.14 eV at the CdS/Cu2ZnSnS4 O/(S + O) ratio increased with decreasing sulfur content in the CBO
interface will cause severe voltage loss. The CBO can be optimized by solution. Hence, the barrier height was lowered and even reversed into a
varying the S/(S + Se) ratio; the optimal ratio range is 0.3–0.4. The cliff-like shape for the sample with a lower thiourea (TU) concentration
Cu2ZnSn(Sx,Se1− x)4 device has an efficiency of 7.2%, 11.4%, and 7%, due to a deficiency. The deformity also ends (Fig. 22). The deformation
respectively, with x = 0, 0.3, and 1.0, respectively (Udaka et al., 2017). can be reduced, and after absorbing light, the Jsc, Voc, and FF increased
The first principle was also used to estimate the CBO. Bao et al. observed remarkably (Fig. 23), which might be caused by the decrease in the
two unequal interfaces amid CdS and Cu2ZnSnS4, the Cu-Zn layer or Cu- spike-like barrier due to the photo-doping of Zn(O,S) (Neuschitzer et al.,
Sn layer adjoining the interface, respectively. The CBO was estimated by 2016a) Grenet et al. also stated that Jsc is almost zero with Zn(O,S) as a
first principles to be − 0.2 eV and − 0.7 eV, respectively. At − 0.3 eV, the buffer layer because of the large spike barrier. Moreover, the efficiency
CBO was average (Bao and Ichimura, 2012b). Palsgaard et al. estimated increased remarkably to 5.8% after light absorption due to the decrease
the CBO at the interface of CdS/Cu2ZnSnSe4 by first principles, which in the spike barrier (Grenet et al., 2014).
was + 0.6 eV like the spike (Palsgaard et al., 2016).
An optional buffer layer has been studied for kesterite (Cu2ZnSnS4
and Cu2ZnSnSe4) and chalcopyrite [Cu(In,Ga)Se2] devices to remove the
parasitic absorption in the CdS layer because of its low bandgap and
toxicity. In addition to parasitic absorption, the CBO of CdS/ Cu2ZnSn
(Sx,Se1− x)4 is influenced easily by the S/(S + Se) ratio. Therefore,
optional buffer layers, such as Zn(O,S), (Zn,Cd)S, Zn1− xSnxOy,
Zn1− xMgxO, and In2S3, need to be studied to improve the band align­
ment at the interface of the buffer/absorber. Among them, Zn(O,S) was
tested as an optional buffer layer because of its benefits, such as low
toxicity, high transparency, abundance, and high and easily tunable
bandgaps (2.6 eV–3.8 eV) (Chua et al., 2016; Meyer et al., 2004; Li et al.,
2017). In Cu(In,Ga)Se2 devices, Zn(O,S) layer has been used as a buffer
layer. The world record efficiency of the Cu(In1− x,Gax)(SySe1− y)2 device
with CBD-Zn(O,S) buffer layer was 22.8%, which was achieved by Solar
Frontier K K, and is comparable to CdS reference devices (Kamada et al.,
2016).

Fig. 22. Light and dark J–V curves of Cu2ZnSnSe4 device with a Zn(O,S) buffer
5.2.1.2. Zn(O,S)/Kesterite [Cu2ZnSnS4, Cu2ZnSnSe4 and Cu2ZnSn(Sx,
layer prepared with varying TU concentrations in the CBD method. The CdS
Se1− x)4] interface. CBO is spike-like at the ZnS/Cu2ZnSn(Sx,Se1− x)4 layer was used as a reference buffer layer. (Reprinted with permission from
interface, which is as much as 1.1 eV [Fig. 21(a)]. High CBO produces a Ref. Neuschitzer et al., 2016a, 2016b).

Fig. 21. Schematic diagram of the band alignment at the interface of (a) ZnS/Cu2ZnSn(Sx,Se1− x)4 and (b) ZnO/Cu2ZnSn(Sx,Se1− x)4 (Reprinted with permission from
Ref. Barkhouse et al., 2012a, 2012b), (c) Band energy diagram of Zn(O,S) versus O/(S + O). (Reprinted with permission from Ref. Sharbati and Sites, 2014)
(Reprinted with permission from Ref. Sharbati et al., 2015).

34
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 23. Light J–V curves of Cu2ZnSnSe4 device with Zn(O,S) buffer layer deposited with varying concentrations of TU in the CBD method before and after light
soaking for 260 min. (Reprinted with permission from Ref. Neuschitzer et al., 2016a, 2016b).

In addition to changing the O/(S + O) ratio to Zn(O,S), the interface that there is a defect state near the Fermi level, which acts as a shortcut
of Zn(O,S)/Cu2ZnSn(Sx,Se1− x)4 can be improved by changing the surface to transport photo-generated electrons through the Zn(O,S) layer, as
of the Cu2ZnSn(Sx,Se1− x)4 absorber layer. Sakai et al. reported that the shown in Fig. 24(b) (Li et al., 2017).
conversion efficiency of Cu2ZnSnS4 devices with the Zn(O,S) buffer layer
decreases with increasing Zn/Sn ratio (Sakai et al., 2011). Thus, before 5.2.1.3. Zn1− xCdxS/Kesterite [Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(Sx,
depositing the Zn(O,S) buffer layer using the CBD method, they reduced Se1− x)4] interface. The CBO at the CdS/Cu2ZnSnS4 and ZnS/Cu2ZnSnS4
the Zn/Sn surface ratio from 0.96 to 1.1 by a chemical treatment on tin- interfaces is –0.3 eV and +1.0 eV, respectively. Therefore, Zn1− xCdxS
based (TCT). Thus, a high-power conversion efficiency of 5.85% was can be used as a buffer layer, which is a mixture of ZnS and CdS, and the
achieved for the 5.0 × 5.0 cm sub-model of Cu2ZnSnS4 (Hiroi et al., CBO can be optimized by varying the Cd/(Zn + Cd) ratio between
2012). The study outcomes suggest that the surface properties of the Zn1− xCdxS and Cu2ZnSnS4. Bao and Qiu estimated the band offsets be­
absorber layer are essential for the properties of the buffer/absorber tween Zn1− xCdxS and Cu2ZnSnS4 using first-principles calculations.
interface, which govern the solar cell performance. They reported that the CBO at the Zn1− xCdxS/Cu2ZnSnS4 interface was
On the other hand, Zhang et al. at Nankai University China reported spike-like and even lower than 0.3 eV when the Cd/(Zn + Cd) ratio was
the coexistence of secondary phases, such as ZnO and Zn(OH)2, in the Zn 0.625–0.75 (Bao and Qiu, 2016). Sun et al. used the successive ionic
(O,S) buffer layer, which will introduce band fluctuations because of the layer adsorption and reaction (SILAR) process to deposit a Zn1− xCdxS
enormous differences in the conduction band edge between ZnO and Zn buffer layer for the Cu2ZnSnS4 device (Sun et al., 2016). They suggested
(O,S), which is harmful to the device performance, as shown in Fig. 24 that the device performance of Cu2ZnSnS4 device with a Zn0.35Cd0.65S
(a). They offered an etching process using concentrated ammonium and buffer layer was enhanced compared to the CdS buffer layer, and 9.2%
used soft annealing to remove secondary phases, ZnO and Zn(OH)2, in maximum power conversion efficiency was recorded with a Voc value of
the Zn(O,S) buffer layer. As a result, the efficiency of the Cu2ZnSnSe4 747.8 mV, which was improved remarkably around 100 mV compared
device with a Zn(O,S) buffer layer increased from 1.17% to 7.2%. Ac­ to the CdS-references (Sun et al., 2016). This increase was attributed to a
cording to their investigation, the CBO of the etched and annealed Zn(O, spike-like CBO (0.37 eV) at the buffer/absorber interface, which may
S)/Cu2ZnSnSe4 interface was spike-like, with a very large 1.22 eV–1.28 play a role in reducing interface recombination. Messaoud et al. used the
eV, which would theoretically impede the flow of photo-generated CBD method to deposit Zn1− xCdxS as a buffer layer for the Cu2ZnSnSe4
electrons. After examining low-temperature transport, they suggested device (Messaoud et al., 2017). They reported that an increase in zinc
content in the Zn1− xCdxS buffer is accompanied by a decrease in Voc and
FF due to Zn(OH)2 and ZnO, which results in a CBO deficiency. There­
fore, a cadmium partial electrolyte (Cd PE) process used before
Zn1− xCdxS deposition can be used to avoid the adsorption of negatively
charged Zn(O, OH)x particles to form a thin layer of Cd(OH)2 ~2–3 nm.
As a result, interface recombination decreases (Messaoud et al., 2017).
Thus, all performance parameters of the Cu2ZnSnSe4 device with the Cd
PE/Zn0.02Cd0.98 buffer were enhanced compared to the CdS references,
and a maximum efficiency of 8.3% was obtained.

5.2.1.4. (Zn,M)O/Kesterite [Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(Sx,

Fig. 24. Schematic diagrams of the band alignments at the interface Zn(O,S)/ Se1− x)4] interface (M = Sn, Mg). Zn1− xSnxOy (ZTO) can be an optional
Cu2ZnSnS4 for (a) the untreated and (b) treated (the etched & annealed) sam­ buffer layer for Cu2ZnSn(Sx,Se1− x)4 devices. The conduction band
ples. (Reprinted with permission from Ref. Zhang et al., 2017). alignment at the Zn1− xSnxOy/Cu2ZnSn(Sx,Se1− x)4 interface can be

35
M. Sahu et al. Solar Energy 230 (2021) 13–58

optimized by modifying the Sn/(Zn + Sn) ratio. Platzer-Bazarkman et al. The mobility of the Cu2ZnSn(Sx,Se1− x)4 absorber was very low
deposited a Zn1− xSnxOy buffer layer on top of a Cu2ZnSnS4 absorber compared to the Cu(In,Ga)Se2 absorber, increasing the carrier concen­
layer by atomic layer deposition (ALD). They examined the band tration of the absorber to improve Voc. Kim et al. used In2S3/CdS as a
alignment at the Zn1− xSnxOy/Cu2ZnSnS4 interface at different deposi­ dual buffer by depositing a thin In2S3 layer onto CdS followed by rapid
tion temperatures (Platzer-Björkman et al., 2015; Ericson et al., 2017). thermal annealing (RTA) (J. Kim et al., 2014). Indium diffused into the
The maximum efficiency of the Cu2ZnSnS4 device with the Zn1− xSnxOy CdS and Cu2ZnSn(Sx,Se1− x)4 layers after the RTA process. The n-type
buffer layer was 9.0%, which is more than the CdS-references (Ericson doping defects in CdS lead to the formation of InCd, and the p-type
et al., 2017). Tajima et al. accumulated ZTO as a buffer layer for doping defect in Cu2ZnSn(Sx,Se1− x)4, giving rise to InSn. As a result, the
Cu2ZnSnS4 devices and achieved a maximum efficiency of 5.7%, which carrier concentration improved in both layers CdS and Cu2ZnSn(Sx,
is smaller than the CdS-reference (Tajima et al., 2016). They attributed Se1− x)4. As a result, a record efficiency of 12.7% was achieved using
the decrease in efficiency to the substantial lattice mismatch between In2S3/CdS as a dual buffer layer in the Cu2ZnSn(Sx,Se1− x)4 device (J.
the Zn1− xSnxOy and Cu2ZnSnS4 layers, which initiated the recombina­ Kim et al., 2014). Hiroi et al. synthesized a hybrid buffer layer by
tion at the interface. A double buffer layer of Zn1− xSnxOy/CdS was depositing In2S3 at the CdS layer. They achieved 9.2% efficiency for the
deposited to minimize the recombination at the interface (Tajima et al., Cu2ZnSnS4 device. Carbon and oxygen diffusion and cadmium and in­
2016). Before depositing the Zn1− xSnxOy layer, a thin layer of CdS ~ 10 dium diffusion were attributed to the improved device performance and
nm was deposited using the CBD process, followed by annealing under less carrier recombination at the i-ZnO/buffer interfaces for a lower
an N2 environment. A high Voc of 0.81 eV was obtained, and the effi­ VBM at the interface (Hiroi et al., 2013). In contrast to Kim et al. and
ciency was improved. Hiroi et al. (Hiroi et al., 2012; J. Kim et al., 2014), Yan et al. designed a
Zn1− xMgxO (ZMO) is a promising material for a buffer layer because CdS/In2S3 hybrid buffer with In2S3 between the CdS and Cu2ZnSnS4
of its high and tunable bandgap (3.3 eV–7.7 eV). Furthermore, the layers, followed by RTP annealing and the resulting enhancement in Voc
conduction band edge at the Zn1− xMgxO/Cu2ZnSn(Sx,Se1− x)4 interface and efficiency (5.5% and 6.6%). Indium can diffuse into CdS and
can be optimized by changing the Mg/(Zn + Mg) ratio. Hironiwa et al. Cu2ZnSnS4 layers simultaneously and increase the carrier concentra­
deposited Zn1− xMgxO by the co-sputtering of MgO and ZnO for a tions in both the buffers and absorbers. Moreover, the carrier order of
Cu2ZnSn(Sx,Se1− x)4 device (Hironiwa et al., 2014; Hironiwa et al., Cu2ZnSnS4 is increased by an order of magnitude compared to the CdS
2015a; Hironiwa et al., 2015b). The efficiency of the Cu2ZnSn(Sx,Se1− x)4 references, thereby improving the Voc of the device (Yan et al., 2016).
device with the Zn1− xMgxO buffer layer is remarkably small compared In summary, the arrangement of the band at the interface of CdS/
to the CdS references, which may be due to damage to the Cu2ZnSn(Sx, Cu2ZnSn(Sx,Se1− x)4 is not ideal, and it is essential to improve the
Se1− x)4 absorber during sputtering. A double buffer layer of Zn1− xMgxO/ arrangement of the band at the interface of buffer/Cu2ZnSn(Sx,Se1− x)4.
CdS was deposited to prevent sputtering damage (Hironiwa et al., 2014; First, CBO can be adapted by modifying the structure of the Cu2ZnSn(Sx,
Hironiwa et al., 2015a; Hironiwa et al., 2015b). The device efficiency Se1− x)4 absorber material, specifically the composition and surface
was increased by the double buffer layer and was similar to the CdS morphology. The CBD process for CdS deposition must be optimized to
references. Furthermore, the efficiency of the Cu(In,Ga)Se2 devices minimize interfacial diffusion and defects. On the other hand, different
reached a maximum compared to the CdS references with a double alternative buffer layers, such as (Zn,Cd)S, Zn(O,S), Zn1− xMgxO,
buffer layer. Zn1− xSnxOy, and In2S3, can be used to improve the arrangement of the
band at the interface of the buffer/absorber materials. Nevertheless, the
5.2.1.5. In2S3/Kesterite [Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(Sx, performances of Cu2ZnSn(Sx,Se1− x)4 photovoltaic devices with other or
Se1− x)4] interface. In2S3, Cu2ZnSn(Sx,Se1− x)4 can be a possible material alternative buffer layers were generally not ideal and even worse than
as a buffer layer for Cu2ZnSn(Sx,Se1− x)4 devices. The In2S3 semi­ the CdS reference. Overall, extensive investigation, and optimization of
conductor has an indirect bandgap (2.1 eV), which is slightly lower than the compositions and the synthesis and deposition processes of the
CdS (2.4 eV) (Barkhouse et al., 2012b; Sandoval-Paz et al., 2005). On the optional buffer layers, will be needed to improve the performance of
other hand, the transparency of In2S3 is higher than CdS because of the Cu2ZnSn(Sx,Se1− x)4 devices with optional buffers.
indirect bandgap of In2S3, which helps to enhance Jsc (Barkhouse et al.,
2012b). Barkhouse et al. reported that CBO is spike-like at the In2S3/ 5.2.2. Optimization of the band alignment at the interfaces of the buffer/
Cu2ZnSn(Sx,Se1− x)4 interface, with an S/(S + Se) ratio of ~0.4 eV in the front
optimal range of 0.15 ± 0.1 eV (Barkhouse et al., 2012b) The Cu2ZnSn
(Sx,Se1− x)4 device containing In2S3 as a buffer layer achieved a 5.2.2.1. Buffer/front (CdS/ZnO) interface. In addition to the high con­
maximum efficiency of 7.6% as compared to the reference CdS. Yan ductivity, the front contact in solar cells requires a wide optical bandgap
et al. also deposited In2S3 as a buffer layer for the Cu2ZnSnS4 device and for high transparency in the desired spectrum. Hence, the window layer
concluded that the In2S3/Cu2ZnSnS4 interface has a spike-like CBO 0.41 containing an intrinsic zinc oxide layer (i-ZnO) and Al-doped ZnO layer
eV ± 0.1 eV, which is somewhat larger than the optimal value. Thus, the (AZO) is used for the chalcopyrite and kesterite devices. Moderate op­
large CBO would inhibit the transport of photo-generated electrons (Yan toelectronic properties, low cost, and broad optical bandgap make ZnO
et al., 2014). Moreover, the Cu2ZnSnS4 device containing In2S3 as a an appropriate window layer for photovoltaic applications (Janotti and
buffer layer has a greater Voc and smaller Jsc and FF than CdS-references. Van de Walle, 2009). The role of AZO can be understood because it has
Yu et al. deposited In2S3 by thermal evaporation at various temperatures high conductivity and provides a suitable broad bandgap for trans­
(30 ◦ C–200 ◦ C). They reported that the CBO values at the In2S3/ parency. On the other hand, the precise role of i-ZnO is not fully un­
Cu2ZnSnS4 interface depend on the deposition temperature of In2S3 and derstood (Weinhardt et al., 2004). The i-ZnO window layer is suitable
vary in the range of 0.01 eV–0.41 eV (Yu et al., 2017). In addition, for photovoltaic devices with a beneficial effect in reducing electronic
sprayed and co-evaporated In2S3 was used as a buffer layer in the inhomogeneities on the surface of the absorbing layer. This removes the
Cu2ZnSnSe4 device, and the device efficiency was approximately 5.7% oxides and other in-homogeneities formed during CdS deposition on the
(Buffière et al., 2015a; Khadka et al., 2015). absorber surface and helps improve the Voc (Rau and Schmidt, 2001).
Some studies reported a double layer of In2S3 and CdS to maximize ZnO band alignment towards CdS is a crucial parameter to under­
the efficiency of kesterite material-based solar cells. Kim et al. suggested stand the key part of i-ZnO. If the band alignment between the front or
that the leading cause of Voc losses in the Cu2ZnSn(Sx,Se1− x)4 device may window film and the buffer film is unsuitable for the flow of charge
be the small majority carrier density and mobility product, i.e., low carriers, the EQE of the solar cell is reduced, even though the absorber
conductivity in the Cu2ZnSn(Sx,Se1− x)4 absorber (J. Kim et al., 2014). material has a high IQE. The ZnO/CdS interface was studied for Cu(In,

36
M. Sahu et al. Solar Energy 230 (2021) 13–58

Ga)Se2 devices in the early 1990s using XPS and UPS. The core-level but not a Schottky rare contact.
binding energy and Auger structures were investigated carefully to The construction of the kesterite device is inherited from the chal­
examine any interface response. The band offset at the interface of ZnO/ copyrite device. Therefore, molybdenum layers are used broadly as end
CdS was analyzed by XPS and UPS measurements, where CBO of 0.3 eV electrodes in kesterite devices. Compared to Cu(In,Ga)Se2 films, the
and VBO of 1.2 eV were calculated. Fig. 25 (a) presents a schematic kesterite films decompose into binary or ternary phases at high tem­
diagram of the band offset at the ZnO/CdS interface. It was reported that peratures (Scragg et al., 2012a; Scragg et al., 2013).
for photovoltaic applications, the flow of electrons across the ZnO/CdS An alternative decomposition process is easily initiated at the kes­
interface with this CBO was unaffected and resulted in higher perfor­ terite/Mo interface at higher temperatures via the following reaction
mance of Cu(In,Ga)Se2 solar cells. In addition, microscopic interface (Scragg et al., 2012b):
properties, such as local charge distributions resulting from the forma­
tion of interface charge chemical bonds, do not play a major role in band 2Cu2ZnSnS(e)4 + Mo2 ↔ Cu2S(e) + 2ZnS(e) + 2SnS(e) + MoS(e)2 (1)
offset (Ruckh et al., 1994). The change in Gibbs energy of reaction (1) at 550 C for Cu2ZnSnS4
◦
In another study, the band offset at the interface of ZnO/CdS was and Cu2ZnSnSe4 were − 150 kJ and − 100 kJ, respectively, indicating
examined by XPS and UPS. Fig. 25(b) presents a schematic diagram of that it is thermodynamically favorable for the decomposition of the
the band offset. A flat CBO of approximately 0.10 ± 0.15 eV and a VBO kesterite/Mo interface. Hence, the MoS2 or MoSe2 film forms easily at
of − 0.96 ± 0.15 eV were observed (Weinhardt et al., 2004). Both reports the interface of kesterite absorber and molybdenum rare contact (Scragg
published a diverse CBO for the interface of ZnO/CdS, but none et al., 2012a; Scragg et al., 2012b). Shin et al. adopted a modified
considered the effect of the band offset on the solar cell efficiency. On selenizing temperature to reduce the formation of MoSe2. They reported
the other hand, it was reported that the ZnO/CdS interface does not that the thickness of MoSe2 decreases with decreasing selenization
resist the flow of charge carriers (Ruckh et al., 1994). temperature, while the average grain size of the Cu2ZnSnS4 absorber
also decreases (Shin et al., 2012). Another way to reduce MoSe2 for­
5.2.3. Optimization of the band alignment at the interfaces of the Absorber/ mation is selenizing precursors at low selenium partial pressures. Yao
Back et al. reported an 8.2% Cu2ZnSnSe4 photovoltaic device with a very thin
MoSe2 film by selenizing an electrodeposited Cu/Sn/Zn precursor at low
5.2.3.1. Kesterite [Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(Sx,Se1− x)4]/ selenium pressures (Yao et al., 2017). On the other hand, deep defects,
Mo interface. Molybdenum (Mo) is commonly used in Cu(In,Ga)Se2 such as Se deficiencies, occur when the partial pressure of selenium is
devices because of its series of beneficial merits as an end electrode. low during selenization at high temperatures (Shin et al., 2012).
These properties involve good adhesion, stability at higher tempera­ The kesterite absorber material is typically synthesized under high
tures, and low resistivity between the Cu(In,Ga)Se2 absorber material sulfur or selenium and SnS(e) pressures at high temperatures to improve
and soda-lime glass substrate (Su et al., 2012; Li et al., 2011; Pethe et al., grain growth, reduce decomposition, and avoid deep defects (such as
2012; Salomé et al., 2010; Wu et al., 2012). Fig. 26 presents a schematic selenium deficiency) (Scragg et al., 2012b; Shin et al., 2012; Shin et al.,
diagram of the band alignment of the Cu(In,Ga)Se2 device. The interface 2015; Shin et al., 2013a; Xiao et al., 2016). Consequently, an extremely
of Cu(In,Ga)Se2/Mo is a Schottky contact that will increase the series thick MoS2 or MoSe2 layer is easily formed at the interface according to
resistance RS and give rise to a rollover event in the I-V curve of the reaction (1), which enhances the contact resistance among the kesterite
photovoltaic device (Spies et al., 2009). Typically, MoSe2 is found in the absorber and molybdenum rare contact (Shin et al., 2012; Shin et al.,
Cu(In,Ga)Se2/Mo interface, which can form during annealing. Fig. 26(b) 2013a). Therefore, one of the challenges to boost the power conversion
shows that the interface of Cu(In, Ga)Se2/MoSe2 is ohmic contact, while capability is to reduce the thickness of the MoSe2 interfacial film. After
the interface of MoSe2/Mo is still Schottky contact due to the bending of investigating the kinetics of MoSe2 interfacial film formation during
a large downward band at the MoSe2/Mo interface. The MoSe2 inter­ annealing of the Cu2ZnSnSe4 absorber layer under a selenium environ­
facial layer is generally very thin, ~10–100 nm. Tunneling will occur ment, Shin et al. reported that MoSe2 is formed through a three-step
about the MoSe2/Mo interface if the MoSe2 layer is very thin. Therefore, process, as shown in Fig. 27: diffusion of selenium via Cu2ZnSnSe4
a very thin MoSe2 interfacial film can act as a buffer layer, converting film, diffusion of selenium via pre-made MoSe2, and a reaction between
the Schottky-like contact to a quasi-ohmic (Spies et al., 2009; Wada molybdenum and selenium (Shin et al., 2013a). This suggests that
et al., 2001; Zhang et al., 2017). On the other hand, the transport of inhibiting selenium diffusion is the key to suppressing the formation of
holes will be suppressed when the MoSe2 interfacial layer is too thick. MoSe2. Thus, there are several ways to reduce the MoSe2 formation.
Zhang’s group at Nankai University China reported that the rare contact As selenium vapor first diffuses through the kesterite absorber layer,
barrier height in Cu2ZnSnSe4 devices is 135 meV because MoSe2 is too an intermediate layer, such as TiN, TiB2, MeO2, and ZnO, is introduced
thick, whereas it is only 24 meV with a thin MoSe2 layer <10 nm at the between the absorber and back contact, which acts as a barrier for sulfur
Cu2ZnSnSe4/Mo interface (Lee et al., 2015). Thus, the thickness of the or selenium diffusion and inhibits the formation of MoS2 or MoSe2
MoSe2 film must be well controlled to form a quasi-ohmic rare contact (Scragg et al., 2013; Shin et al., 2012; F. Liu et al., 2014; Schnabel and

Fig. 25. (a) Schematic band diagram with an offset at the CdS/ZnO interface. (Reprinted with permission from Ref. Ruckh et al., 1994), (b) Band alignment at the
interface (middle) after in the view of band bending. (Reprinted with permission from Ref. Weinhardt et al., 2004).

37
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 26. Schematic diagram of the band diagram of the Cu(In,Ga)Se2 device: (a) without the MoSe2 layer, (b) with the layer of MoSe2. (Reprinted with permission
from Ref. Gao et al., 2018a, 2018b).

Fig. 28. Schematic representations of the different rare contact alignments.

(Reprinted with permission from Ref. Schnabel and Ahlswede, 2017).

intermediate film in Cu2ZnSnS4 photovoltaic devices (F. Liu et al.,

Fig. 27. Schematic diagram of surplus selenium concentration profile across 2014). Consequently, the series resistance was decreased significantly.
the Cu2ZnSnS4/MoSe2/Mo interface during annealing process in a steady-state. Hence, the efficiency of photovoltaic devices was improved from 3.06%
(Reprinted with permission from Ref. Shin et al., 2013a).
to 4.40%, and the Jsc and FF were also increased. Nevertheless, TiB2
intermediate film will damage the crystallinity of the absorber, which
Ahlswede, 2017). will deteriorate the performance of the solar cells, specifically Voc (F. Liu
TiN can be used as the barrier material to block the diffusion of sulfur et al., 2014).
or selenium for its stability and slight diffusion at higher temperatures. In addition, Duchatelet et al. oxidized the molybdenum-coated soda-
Shin et al. reduced the thickness of the MoSe2 film from 1300 nm to 220 lime glass substrate to form a MoO2 film to reduce MoSe2 formation.
nm using a thin TiN layer as a diffusion inhibiter (~20 nm). The They reported that the ΔG of the MoO2 reaction with selenium was
resulting series resistance decreased from 3.42 cm2 to 1.8 cm2, and the +294 kJ/mol; hence, the reaction is not thermodynamically favorable.
power conversion efficiency increased from 2.95% to 8.9% (Shin et al., Consequently, MoO2 can inactivate molybdenum against selenization,
2012). In contrast, Scragg et al. reported that the TiN intermediate film leading to a substantial decrease in MoSe2 layer thickness. MoO2 shows a
would induce a higher series resistance, leading to a lack of fill factor. metallic electrical conductivity (Nina et al., 2008; Werfel and Minni,
Thus, in contrast to Shin et al. (Shin et al., 2012), the Cu2ZnSnS4 device 1983), and hence the MoO2 layer with a MoSe2 film influences the sheet
efficiency decreased even when a TiN intermediate film was grown, resistance of the end electrode (Duchatelet et al., 2013). Lopez-Marino
while MoS2 was out of growth (Scragg et al., 2013). Schnabel et al. et al. reported that MoO2 modifies the favorable orientation of MoSe2
examined the effects of different back contact arrangements of Mo and from (1 0 0) to (0 0 2), thereby allowing the Se-Mo-Se layers to change
TiN (Fig. 28) on Mo(Sx,Se2− x) formation, absorber crystallinity, and the orientation from perpendicular to parallel to the Mo substrate [Fig. 29
performance of photovoltaic devices (Schnabel and Ahlswede, 2017). (a) and 29(c)]. MoSe2 plays a role in the molybdenum substrate sele­
They reported that a thin TiN intermediate film could prevent Mo(Sx, nium barrier with parallel Mo-Se and reduces the MoSe2 thickness
Se2− x) formation but was unfavorable for the crystallinity of the (López-Marino et al., 2016). On the other hand, MoSe2 exhibits aniso­
absorber material. The photovoltaic performance of the device showed tropic conductivity. The electrical conductivity with Se-Mo-Se layers is
that the presence of Mo(Sx,Se2− x) is essential for achieving a high Voc. two orders of magnitude higher than that perpendicular to the Se-Mo-Se
Therefore, the best performing devices were the use of a combined Mo/ films. Hence, the electrical conductivity of MoSe2 with the molybdenum
TiN/Mo stack as a final electrode. Thus, the thickness of the Mo(Sx, substrate is much worse with parallel Se-Mo-Se (Würz et al., 2003).
Se2− x) film can be controlled easily by the thickness of the upper Mo A thin MoSe2 layer must be present to make molybdenum have
film. ohmic contact with the absorber. Hence, a cap layer of Mo ~ 20–70 nm
In addition to TiN, TiB2 is also used as the intermediate film. Liu et al. as a sacrificial layer was fabricated on the top of the MoO2 film (López-
prevented the formation of the MoS2 layer using the TiB2 layer as an Marino et al., 2016). Thus, the MoSe2 thickness can be increased by

38
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 29. MoSe2 Crystal structures with different directions, (designed by Crystal Maker software): (a) (1 0 0) and (b) (1 1 0) preferred orientation with Se–Mo–Se
sheets, which are perpendicular to the substrate, (c) (0 0 2) and (d) (1 0 3) preferred orientation with Se–Mo–Se sheets which are parallel and tilted to the substrate,
respectively. (Reprinted with permission from Ref. Gao et al., 2018a, 2018b).

altering the thickness of the top sacrificial cap layer of Mo and molybdenum interface. On the other hand, during the selenization
enhancing the solar cell performance. In addition, the MoO2 layer could procedure at higher temperatures under high sulfur or selenium pres­
effectively suppress the thickness of MoSe2 and improve grain growth sure, an excessively thick MoS2 or MoSe2 layer can form easily, which
and shunt resistance (Rsh) (López-Marino et al., 2016). will impede the flow of holes. Thus, it is vital to take adequate measures
A ZnO intermediate layer on molybdenum rare contact can also to reduce the MoS2 or MoSe2 thickness. MoS2 or MoSe2 formation can be
inhibit the interaction between the kesterite and molybdenum layer blocked using a temporally dense precursor, a thin (1 0 3) peak preferred
during the annealing process at high temperatures and inhibit the Mo(S,Se)2 film along with inclined Se-Mo-Se sheets, or an intermediate
decomposition of kesterite materials. López-Marino et al. grew a very barrier layer, containing TiN. TiB2 and MoO2.
thin ZnO layer (~10 nm) to act as an intermediated layer and reduce the
decomposition reaction at the interface of Cu2ZnSnSe4 absorber and 6. Approaches to reducing defects and defect-complexes and
molybdenum back contact. As a result, the binary phases of CuxSe were improving the different interface losses
removed, and the density of the voids was decreased. Thus, the series
resistance decreased significantly, and the efficiency improved from For high-efficiency solar cells based on kesterite materials, there are
2.5% to 6.0% (López-Marino et al., 2013a, 2013b). Li et al. and Liu et al. several simulation models and experimental approaches to reduce the
also suppressed MoS2 formation by fabricating an ultrathin of ZnO in­ formation of defects and defect clusters and improve the loss mecha­
termediate layer (10 nm) (Li et al., 2014; X. Liu et al., 2014). nisms between the heterojunction interfaces, and considerable research
The pathway to reducing MoS2 or MoSe2 formation after inserting is ongoing. These approaches include the following: a post-annealing
the intermediate film can be somewhat reasonable. Nevertheless, this treatment at low and high temperatures, defect passivation, suppress­
will complicate the procedure, and the series resistance of photovoltaic ing secondary phases by controlling the stoichiometry during the growth
devices cannot be reduced to a low level. Zhang et al. presented a two- process, passivation or modifying of the front and back contact,
way approach for reducing MoSe2 formation without an intermediate isoelectronic cation substitution, and doping of alkali or other metal
film to solve the complexity of the process to reduce MoSe2 formation elements. These approaches help minimize the losses of photogenerated
and the series resistance of photovoltaic devices (Lee et al., 2015; Gao charge carriers and improve the Voc losses. As discussed earlier, high
et al., 2018b). One method was to anneal the metal precursors in an Ar losses of the open-circuit voltage are detrimental to Cu2ZnSn(Sx,Se1− x)4
atmosphere at 300 ◦ C to form a temporary but dense metal alloy film. solar cells. This section discusses the recent improvements in the inter­
Thus, during high-temperature annealing, there was a barrier to sele­ face along with the formation of defects of Cu2ZnSn(Sx,Se1− x)4 based
nium diffusion. Therefore, the MoS2 thickness was decreased from 1200 solar cells.
nm to less than 10 nm. The decrease in series resistance was 3.4 cm2 to
0.61 cm2, but the efficiency was increased from 5.6% to 8.7% (Lee et al.,
2015). A thin (1 0 3) peak for a Mo(SSe)2 film with an inclined Se-Mo-Se 6.1. Reduce phases, defects, and defect-complexes
layer was fabricated by altering the surface topography of the Mo layers
(Gao et al., 2018b). The inclined Se-Mo-Se layer acts as a barrier to block 6.1.1. Post-annealing treatment at low and high temperature
Mo(S,Se)2 film formation and provides good electrical conductivity with The unfavorable bandgap and possible fluctuations in the kesterite
the Se-Mo-Se layer, as shown in Fig. 29(d) (Gao et al., 2018b). Conse­ absorber layer, which is influenced by the annealing parameters, such as
quently, Mo(S,Se)2 thickness decreased sharply from 1500 nm to 200 nm temperature, annealing time, pressure, annealing environment, and
with the change in molybdenum surface morphology. The series resis­ cooling time, has been studied extensively (Scragg et al., 2014; Rey
tance of Cu2ZnSn(Sx,Se1− x)4 devices decreased from 2.94 Ω cm2 to 0.49 et al., 2014; Krämmer et al., 2014; Scragg et al., 2016). If Cu2ZnSn(Sx,
Ω cm2, and the efficiency increased from 6.98% to 9.04%. Se1− x)4 band trailing is observed because of cationic disorder, there must
In summary, a very thin MoS2 or MoSe2 layer can act as a buffer layer be a change in the effective bandgap of the Cu2ZnSn(Sx,Se1− x)4 absorber
to convert Schottky contact to a quasi-ohmic contact at the kesterite/ layer because of annealing/quenching from various temperatures. Ray
et al. examined this by UV–Vis–NIR spectroscopy. A 110 meV decrease

39
M. Sahu et al. Solar Energy 230 (2021) 13–58

in the bandgap of Cu2ZnSnSe4 was observed when the annealing tem­ thin films exhibited predominant band tail state PL emission at ~330
perature was increased from 75 ◦ C to 300 ◦ C, and band-gap fluctuations meV, which is related to the deep states. A slight decrease in the open-
similar to the order–disorder kinetics in the Vineyard model were noted circuit losses was noted despite a change in the bandgap after anneal­
(Vineyard, 1956), as shown in Fig. 30 (Rey et al., 2014). Because ing at lower temperatures (Gershon et al., 2015). Future studies will
cationic order–disorder in Cu2ZnSn(Sx,Se1− x)4 show a similar tempera­ need to address the lack of association between sub-bandgap defect
ture dependence to the Vineyard model [Fig. 30(b)], the fluctuations in reduction and open-circuit voltage losses in Cu2ZnSn(Sx,Se1− x)4 solar
the bandgap of Cu2ZnSn(Sx,Se1− x)4 can indicate the order–disorder in cells.
the Cu2ZnSn(Sx,Se1− x)4 lattice (considered as stoichiometric). Thus, a Copper-poor (Cu/(Zn + Sn) = 0.8) and zinc-rich (Zn/Sn = 1.2) is the
decrease in lattice disorder could be possible by post-annealing most favorable condition for Cu2ZnSn(Sx,Se1− x)4-based solar cells to
optimization. achieve high-efficiency (Todorov et al., 2013). During post-annealing,
From the Vineyard model, there is a specific critical temperature unfavorable conditions cause the formation of other phases, point de­
(Tc). Above this Tc value, a completely disordered state has been seen fects, and complex defects in Cu2ZnSn(Sx,Se1− x)4 due to very narrow
(Vineyard, 1956). At moderate temperatures less than Tc, the extent of regions to grow pure phase. Therefore, it is essential to control the
disorder decreases with increasing annealing time. The time reaches stoichiometry during post-annealing, especially at high temperatures.
infinity as soon as it approaches the equilibrium state for a given tem­ Yang et al. examined the interrelation between the composition distri­
perature (Scragg et al., 2014). When the temperature is very close to the bution and defect formation with the post-annealing process in the
lower value, a higher equilibrium order parameter can be supposed Cu2ZnSnS4 thin films (Yang et al., 2015). Sulfurization at various tem­
according to the fundamental theory, but the kinetics to reach this perature control the formation of defects and other binary and ternary
condition may require an unreasonably long time. Therefore, the post- phases by improving the distribution of the metal element in Cu2ZnSnS4
annealing parameters, such as temperature and time, should be opti­ thin films. Fig. 31 shows the elemental distribution and cross-section
mized to achieve a uniform distribution of cations in the Cu2ZnSn(Sx, transmission electron microscopy (TEM) image at a sulfurization tem­
Se1− x)4 lattice. Hence, a suitable route is required to enhance the lattice perature of 570 ◦ C. Homogeneous distributions of copper, zinc, tin, and
order in the Cu2ZnSn(Sx,Se1− x)4, including post-annealing at low tem­ sulfur were observed at different locations from the composition per­
peratures, with Tc for the lattice order in the range of 200 ◦ C–250 ◦ C centage and compositional ratio in the optimized annealed Cu2ZnSnS4
(Scragg et al., 2014; Rey et al., 2014; Krämmer et al., 2014; Scragg et al., layer. Fig. 31(c-f) presents 2-D mapping of copper, zinc, tin, and sulfur.
2016). In regards to the holding period, long annealing times are needed Composition differences across the line [Fig. 31(a–d)] were observed, as
to produce a better order structure. From a practical point of view, shown in Fig. 31(g) and 31(h), respectively.
however, one – four hours appears sufficient to approach equilibrium Furthermore, the partial pressure of chalcogen (sulfur or selenium)
(Scragg et al., 2014; Rey et al., 2014). When the dwell time was longer also plays a vital role in the elemental distribution in the absorber layer
than one hour for 200 ◦ C–250 ◦ C, the bandgap of Cu2ZnSn(Sx,Se1− x)4 during post-annealing. The partial pressure of sulfur influences the
increased to 50 meV–100 meV compared to completely disordered distribution of elements in the absorber thin film and the solar cell
condition. This favors the theory that a longer and lighter post-annealing performance (electrical short devices to 6.4%.). The optimal condition
may improve the cation order in Cu2ZnSn(Sx,Se1− x)4 (Rey et al., 2014; to obtain a balanced elemental distribution at long-term temperatures
Krämmer et al., 2014). If there is substantial improvement in the was a sulfur pressure of 10 kPa. Fig. 32 presents the surface reaction and
cationic order through proper post-annealing, this will reduce the Voc element distribution during post-annealing. Thus, the optimal sulfuri­
losses. zation conditions, i.e., optimal partial pressure and temperature, offer
Gershon et al. performed intensity-dependent PL measurements at an even distribution of copper-poor and prevent the decomposition of
low temperatures to examine the effects of low-temperature annealing Cu2ZnSn(Sx,Se1− x)4 by 3, 4, and 5 reactions, and produces benign defect
on the band tail state defect and the solar cell performance. They re­ states (Zhong et al., 2014).
ported that a shift to band tail states toward higher energy photons
occurred when a Cu2ZnSn(Sx,Se1− x)4 thin film was annealed for one day 6.1.2. Elemental composition control under a favorable chemical potential
at 150 ◦ C. On the other hand, annealing films for more than one day The concentration of defects in Cu2ZnSnS4 and Cu2ZnSnSe4 changed
produced similar PL results to one-day annealing. This indicates that up with the chemical potential from poor to rich. When the chemical po­
to a specific upper limit, an ordered structure can be obtained using this tential of copper increases, the populations of the most prominent
process. Moreover, despite variation in the annealing conditions, all the acceptor defects, CuZn and CuZn, increase rapidly. Hence, a lower Fermi

Fig. 30. (a) Bandgap determination of Cu2ZnSnSe4 from transmittance and total reflection for samples that have undergone annealing and quenching at different
annealing temperatures (inset: Tauc plots), (b) plot of the order parameter, sulfur, estimated from the Vineyard model (inset: bandgap measured by UV–Vis-NIR
spectrophotometer) as a function of annealing temperature during annealing and quenching. (Reprinted with permission from Ref. Rey et al., 2014).

40
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 31. (a) Cross-section TEM image with marked

points (1 to 7) and lines (a–d) for EDX measure­
ments. (b) Compositional percentage and the
compositional ratio of different elements at posi­
tions 1 to 7. The 2-D elemental mapping in a marked
rectangular region for (c) Cu, (d), Zn, (e) Sn, and (f)
S. The compositional variation of different elements
across the marked line (g) from position a to posi­
tion b and (h) from position c to position d. The
Cu2ZnSnSe4 film was annealed at an optimal tem­
perature of 570 ◦ C in a sulfur ambient. (Reprinted
with permission from Ref. Yang et al., 2015).

conditions, VCu is more predominant than CuZn. Under the copper-

poor situation, a deviation of the ideal copper and zinc ratio was
observed because the population of [VCu + ZnCu] and [ZnSn + 2ZnCu]
defect complexes increases rapidly. The effects of these two defects in
the band structure of kesterite material are minimal. Hence, a high
concentration has no adverse influence on the device performance.
Downshifts in the valence band caused by [VCu + ZnCu] groups are
favorable because they exhibit increased charge carrier separation
(electron-hole) in the light absorber thin film.
Under zinc-rich and copper-poor conditions, the concentration of the
[2CuZn + SnZn] defect complex is decreased. Thus, a high concentration
of holes (~1015–1016 cm− 3) is obtained by the ionized VCu− when the
Cu/(Zn + Sn) ratio is 0.8. Experimentally, the hole concentration in
Cu2ZnSnS4 and Cu2ZnSnSe4 devices was ~5 × 1015 cm− 3 (Barkhouse
et al., 2012a; Wang et al., 2010). On the other hand, when the Cu/(Zn +
Fig. 32. Surface reaction and possible diffusion of elements in the Cu2ZnSn(Sx, Sn) ratio was higher ~1.0, the concentration of the holes was very high,
Se1− x)4 absorbing layer during sulfurization at high temperatures. (Reprinted with scattered values of 1017 to 1019 cm− 3 (Shin et al., 2011; Wibowo
with permission from Ref. Zhong et al., 2014).
et al., 2007; Tanaka et al., 2005; Leitao et al., 2011; Zhou et al., 2010;
Zhang et al., 2016a, 2016b, 2016c; Adhi Wibowo et al., 2007). When the
energy, a high population of holes, and good p-type conductivity are Cu/(Zn + Sn) ratio was ~1.0, the population of CuZn− defect and holes
observed. Despite the good p-type conductivity, a low solar cell effi­ were approximately 1016–1018 cm− 3; both match the experimental re­
ciency was observed because of the formation of other non-desirable sults. Despite this, poor device performance was observed due to the
defect complexes, such as [2CuZn + SnZn]. population of [2CuZn + SnZn] (1016–1018 cm− 3). This explains the low
On the other hand, when the chemical potential of copper decreased, efficiency in the Cu2ZnSnS4 solar cells with a Cu/(Zn + Sn) ratio around
the population of CuZn and CuZn− and the hole concentration decreased. 1.0 (Ennaoui et al., 2009; Tanaka et al., 2011; Zhou et al., 2010; Riha
After falling below a certain value (− 0.4 eV for Cu2ZnSnS4 and − 0.3 eV et al., 2009; K. Maeda et al., 2011; Araki et al., 2009).
for Cu2ZnSnSe4), other acceptor defects, such as VCu and VCu− , started to The defect and hole concentrations change as a function of the
increase. Thus, after reaching the lowest value, the hole concentration chemical potentials of zinc (μZn) and tin (μSn). The hole concentration
increased, and p-type conductivity was improved. Under copper-poor increases with ionization of the CuZn antisites, but the hole

41
M. Sahu et al. Solar Energy 230 (2021) 13–58

concentration decreases rapidly with µZn when zinc is rich at a fixed copper and zinc, and (2) a lower valence band (VB) edge position in
copper chemical potential (µCu). Changes in µSn, i.e., tin richness, have a Ag2ZnSnS4 (0.74 eV) than Cu2ZnSnS4 (Yuan et al., 2015). To supporting
very weak effect on the hole population. Three defect clusters [VCu + the AgZn anti-site defect quenching, the authors stated that the forma­
ZnCu], [2CuZn + SnZn], and [ZnSn + 2ZnCu] participate in the non- tion energies of defect complexes, such as [AgZn− +ZnAg+] (0.52 eV/
stoichiometry, i.e., either zinc becomes very rich or Sn becomes very pair) and [2AgZn− +SnZn2+] (>3eV/pair) in Ag2ZnSnS4 material, are
poor. [ZnSn + 2ZnCu] and [VCu + ZnCu] may have a very large popula­ larger than those of [CuZn− +ZnCu+] (0.20 eV/pair) and [VCu− +ZnCu+]
tion. For example, when the Zn/Sn ratio was ~1.05, the hole concen­ (0.4 eV/pair) in the Cu2ZnSnS4 material (Yuan et al., 2015). Thus, this
tration was more than 1020 cm− 3. In contrast, the hole concentration chemical design concept can efficiently reduce the distorted cation
was high regardless of whether zinc was very poor or Sn was very rich. arrangement and sub-bandgap defect in Ag2ZnSnS4.
This explains why a copper-poor and zinc-rich condition is important for Based on simulation results and chemical design principles, many
high-efficiency kesterite-based solar cells. studies focused on experimental examinations of Ag2ZnSnS4 and
Ag2ZnSnSe4 materials. The prepared Ag2ZnSnS4 and Ag2ZnSnSe4 ma­
6.1.3. Cationic substitution with small or large ionic radii terials showed a kesterite crystal structure with a bandgap of 2.01 eV
One possible way to control defect formation is to control the dis­ and 1.34 eV, respectively. Gershon et al. reported that effective red-
torted cation arrangement in the kesterite absorber material. Therefore, shifting in the PL spectra was used to identify the status of the sub-
distorted cation arrangements can be minimized by incorporating larger bandgap defect in kesterite, which was not found in Ag2ZnSnSe4.
cations. Thus, it will form a cationic order in kesterite samples. Never­ [Fig. 33(c)] (Gershon et al., 2016a) Hages et al. confirmed that, as the
theless, the replacement of cations may alter the crystal structure, e.g., [Ag]/[Ag + Cu] ratio was enhanced by 0.5 in the mixed (Ag,
from kesterite in Cu2ZnSnS4 (CZTS) to stannite in Cu2CdSnS4 (CCTS). Cu)2ZnSnSe4 compound, the Urbach energy was decreased by 42 mV for
With this concept, silver (Ag, 1.14 Å) to replace copper (Cu, 0.74 Å), pure Cu2ZnSnSe4 to 22 meV in the 50% silver-substituted solar cell
cadmium (Cd, 0.92 Å) to replace zinc (Zn, 0.74 Å), and gallium (Ge, (Hages et al., 2016b). These experimental results suggest that cationic
0.53 Å) to replace tin (Sn, 0.69 Å), may decrease cationic antisite defect disorder-induced band tailing is decreased significantly for silver-
formation in the crystal lattice. This part focus on the defect engineering substituted materials, supporting the theoretical prediction (Chen
of kesterite materials by cationic replacement (Shin et al., 2017). et al., 2010b).
Many studies have reported that pure Ag2ZnSnS4 and Ag2ZnSnSe4
6.1.3.1. Silver (Ag) replacement for copper (Cu) in materials based on show n-type nature, as seen in Fig. 33(d) (Gershon et al., 2016a; Wei and
kesterite. Because silver and copper are isoelectronic elements, silver is Nolas, 2015; Sasamura et al., 2012; Yeh and Cheng, 2014; Tsuji et al.,
an ideal competitor for Cu substitution; the radius of the silver is much 2010). From the defect calculations, except for the M point, in all
greater than copper and zinc. DFT simulation for Ag2ZnSn(Sx,Se1− x)4 chemical potentials, the dominant defects are VS, SnZn, and ZnAg donor
shows promising results to favor this design principle. The formation defects, and AgZn is an acceptor defect. When all the defects are ionized,
energy for the [AgZn + ZnAg] antisite defects in kesterite materials is 3.7 the VS defect shows a lower formation energy in Ag2ZnSnS4. The Fermi
times higher than that of [CuZn + ZnCu] antisite defects (Chagarov et al., level (1.0 eV) of Ag2ZnSnS4 was in the middle of the bandgap because of
2016). Independent simulation of Ag2ZnSnS4 by Yuan et al. supports charge recompensation between the acceptor and donor defects. In tin-
these results. [Fig. 33(a) and 33(b)] (Yuan et al., 2015). Yuan et al. re­ rich, zinc-rich, and sulfur-poor situations, the Fermi level is located (0.8
ported that the AgZn anti-site formation energy for Ag2ZnSnS4 is 1.1 eV eV) below the CBM to promote n-type nature in Ag2ZnSnS4. Conse­
greater than the formation energy of CuZn anti-site for Cu2ZnSnS4. Yuan quently, Ag2ZnSnS4 shows poor electrical conductivity and a low con­
et al. provided two supporting elements for the larger formation energy centration of charge carriers; the measured electron concentration of
of AgZn: (1) higher size discrepancy between silver and zinc compared to <1013 cm− 3 in the literature also supports this (Gershon et al., 2016a;

Fig. 33. a) Formation energies of intrinsic point

defects and defect pairs as a function of the chemical
potential along a path surrounding the stable
chemical potential region of a single-phase
Ag2ZnSnS4 and b) transition-energy levels of
intrinsic defects in the electronic bandgap of
Ag2ZnSnS4. (Reprinted with permission from
Ref. Yuan et al., 2015) c) Bandgap (derived from
Tauc plot; filled circles), PL peak position (measured
at room temperature; open squares), and degree of
red-shifting (energetic difference between bandgap
and PL peak position; red) and d) carrier density as a
function of [Ag]/[Ag + Cu] in (Ag,Cu)2ZnSnSe4.
(Reprinted with permission from Ref. Gershon et al.,
2016a).

42
M. Sahu et al. Solar Energy 230 (2021) 13–58

Yuan et al., 2015). Thus, using Ag2ZnSnS4 as absorber material in solar et al. obtained 9.24% efficiency for a Cu2(Zn1− x,Cdx)SnS4 device with
cell applications, a new solar cell architecture is required, supporting the 40% Cd-replaced, as shown in Fig. 34(a) (Su et al., 2015). The Cu2(Zn1−
n-type nature of (Ag,Cu)2ZnSnSe4. Gershon et al. reported a SnO:F/ x,Cdx)SnS4 solar cell showed 100 meV less Voc losses than the Cu2ZnSnS4
AZTSe/MoO3/ITO solar cell structure based on Ag2ZnSnSe4 material solar cell. This improvement in Voc losses and efficiency might be due to
and showed an efficiency of 5.2% (Gershon et al., 2016b). One inves­ the good morphology, absence of other detrimental phases, and less
tigation reported that 10% partial Ag-substitution in Cu2ZnSn(Sx, cationic distortion. On the other hand, the concentration of Cd element
Se1− x)4, resulted in 10.2% efficiency for (Ag,Cu)2ZnSnSe4 solar cells of more than 60% had an adverse effect on the efficiency. The pure
with an open-circuit voltage of 580 meV–590 meV, which are lower than Cu2CdSnS4 solar cell showed 1.53% efficiency (Su et al., 2015). Never­
the Voc losses 640 meV–660 meV for Cu2ZnSn(Sx,Se1− x)4-based solar theless, optimized cadmium replacement in Cu2ZnSnS4 offers a suitable
cells (Gershon et al., 2016a). This improvement in the Voc losses can be route to improve the Voc losses and solar cells performance, as shown in
explained by a decrease in band tailing, as explained in Fig. 33(c) Fig. 34(a). Moreover, single-crystal Cu2CdSnS4 solar cells showed a PCE
(Gershon et al., 2016a). On the other hand, more than 10% silver- of 2% with Voc losses of 0.5 V (Wagner and Bridenbaugh, 1977).
substitution promoted the decrease in solar cell performance by
decreasing the density of the majority carrier (Gershon et al., 2016a). 6.1.3.3. Germanium (Ge) replacement for tin (Sn) in materials based on
kesterite. The main purpose of germanium doping in kesterite Cu2ZnSn
6.1.3.2. Cadmium (Cd) replacement for zinc (Zn) in materials based on (Sx,Se1− x)4 is to increase the bandgap of the parent absorber compound.
kesterite. For single-junction PV applications, Cu2CdSnS4 (CCTS) mate­ After replacing tin with germanium in Cu2Zn(Sn1− xGex)S4, the bandgap
rial can be used as a suitable absorber, as shown in a tetragonal stannite- increased from 1.0 eV (1.5 eV) for x = 0 to 1.4 eV (1.94 eV) for x = 1.0
type structure with the space group I-42m with a bandgap of 1.4 eV (Su (S. Kim et al., 2016a; Ford et al., 2011). Hages et al. reported a 9.4%
et al., 2015; Pilvet et al., 2015; Zhao et al., 2015). DFT calculations have efficiency for Cu2Zn(Ge,Sn)(S,Se)4 material-based solar cells. Despite
been used to examine the defect properties of Cu2CdSnS4 (Yuan et al., the high efficiency, open circuit losses from 690 meV to 730 meV were
2015). Anti-site defects, such as CuCd and CdCu, may have a high for­ recorded (Hages et al., 2015). Bag et al. reported that with increasing
mation energy because of the larger difference in ionic radii and bond doping percentage of germanium (40%), the efficiency of Cu2Zn
length in Cu2CdSnS4 than Ag2CdSnS4. On the other hand, the formation (Sn1− xGex)Se4 solar cell increased with a high open-circuit voltage, but
energy of CuCd in Cu2CdSnS4 was slightly higher than CuZn in the Voc losses of the solar cell deteriorated at higher levels of germanium
Cu2ZnSnS4. Nevertheless, the formation energy of CuCd was compara­ doping (Bag et al., 2012).
tively low. As a result, in all the chemical potential points with the Admittance spectroscopy has been used to identify the presence of
lowest formation energy (0.3 eV–0.6 eV), the dominant acceptor defect deep levels defects in thin films with a higher germanium concentration
is the CuCd anti-site. Similar to Cu2ZnSnS4 (CuZn and VCu), the formation (Neuschitzer et al., 2016b). For more than 40% germanium-substitution,
energy of VCu (0.8 eV–1.3 eV) is larger than that of CuCd in Cu2CdSnS4. a sufficient red-shift was observed in the peak position of the PL spectra
These findings are somewhat contrary to the assumptions based on from 1.05 eV [for pure Cu2ZnSn(Sx,Se1− x)4 content] to 0.8 eV, as shown
cation size mismatch alone (Yuan et al., 2015). Thus, the authors in Fig. 35(a), despite the higher bandgap (1.4 eV) (Collord and Hill­
speculate that a small cation (copper) has substituted for a larger cation house, 2016). The deep defect levels of the bandgap in Cu2Zn(Sn1− x,Gex)
(cadmium) in CuCd. In contrast, for the AgZn defect, a larger cation Se4 may be responsible for this low-energy PL peak. The transition-
(silver) has substituted for a smaller cation (zinc) in Ag2ZnSnS4. The energy levels of the anti-sites of CuSn and SnCu are found near the
dominant CuCd anti-site acceptor defect was responsible for the p-type mid-bandgap (Fig. 16). Deep acceptor and donor defects may also occur
conductivity in Cu2CdSnS4. Although the [CuCd + CdCu] defect pair (0.2 in the electronic band structure of CuGe and GeCu. An increase in the
eV) has lower formation energy than the isolated CuCd anti-site defect, a energy of the trap level center was examined using a theoretical
similar value was found for the [CuZn + ZnCu] complex defect in approach with the germanium content: above the VBM for pure Cu2ZnSn
Cu2ZnSnS4. This suggests that a distorted cations arrangement is also (Sx,Se1− x)4 and from 106 meV to 170 meV for the 30% germanium-
dominant in the Cu2CdSnS4 lattice (Yuan et al., 2015). Moreover, the substituted material (Moore et al., 2012). Thus, a higher germanium
[2CuZn + SnZn] complex in Cu2ZnSnS4, the [2CuCd + SnCd] has low concentration can influence the band structure and form deep defect
≈0.43 eV–0.65 eV formation energy. Thus, sub-band defects may be levels near the mid bandgap, as observed by PL. An examination of the
observed in Cu2CdSnS4 because of the presence of a high population of impact of higher germanium concentrations (>5%) on Cu2ZnSn(Sx,
defect complexes. Pilvet et al. reported a redshift in the PL peak, which Se1− x)4 showed that an adverse effect on the Voc and finally reduced the
identifies as a sub-band defect in the Cu2ZnSnS4 and Cu2CdSnS4 at 0.25 efficiency. Collord and Hillhouse used an air-aging method to incorpo­
eV and 0.26 eV, respectively (Pilvet et al., 2015). rate a very small amount of germanium into Cu2ZnSn(Sx,Se1− x)4 thin
Surprisingly, a larger proportion of cadmium replacement in films. They recorded a low Voc loss of 548 meV [Fig. 35(b)] (Collord and
Cu2(Zn1− x,Cdx)SnS4 (CZCTS) positively impacted the Voc losses. Joel Hillhouse, 2016). Giraldo et al. also showed that a very small amount of

Fig. 34. (a) J–V characteristics of Cu2(Zn1− xCdx)SnS4 devices. (Reprinted with permission from Ref. Su et al., 2015) (b) Formation energies of the intrinsic and
complex defects as a function of the chemical potential with the stable region at the APQMNPG lines. (Reprinted with permission from Ref. Yuan et al., 2015).

43
M. Sahu et al. Solar Energy 230 (2021) 13–58

lifetime.
Li et al. examined the effects of Na on a Cu2ZnSn(Sx,Se1− x)4-based
device and studied the performance parameter. Na incorporation
reduced the density of deep-level traps state responsible for recombi­
nation and enhanced the Hall mobility and carrier density. By applying a
non-solution process for defects, passivation improved the carrier con­
centration but reduced the lifetime of the carrier (Rockett, 2005). By
examining the effects of various soda-lime glass thicknesses on the
passage of defects, Ishizuka et al. conducted similar research on Cu(In,
Ga)Se2 devices (Ishizuka et al., 2009). The change in the peak of the PL
spectra of soda-lime glass thin film thickness (Fig. 38) might be due to
the presence of oxygen antisites to selenium (OSe), leading to an increase
in p-type conductivity and efficiency in the Cu(In,Ga)Se2 device. The
open-circuit voltage and fill factor in the devices are affected mainly by
the recombination of charge carriers and dark current. Thus, the
Fig. 35. (a) PL spectra of the Cu2Zn(Sn1− xGex)S4 layer with varying germa­ passivation of defects by modifying the solution growth process is
nium concentrations, b) change in the open-circuit voltage (top), and calculated
important for achieving a high-efficiency device.
open-circuit voltage loss values (bottom) of aged solar cells. The ratio of the
actual open-circuit voltage to the maximum theoretical open-circuit voltage SQ
was also calculated to quantify the open-circuit voltage loss [Voc SQ (V) = 6.2. Improving the different interface recombinations
0.932 × Eg (eV) − 0.167]. The solar cells were aged in air for one month.
(Reprinted with permission from Ref. Collord and Hillhouse, 2016). 6.2.1. Passivation of the front interface
ALD-coated dielectric films help to reduce the defect density on the
germanium (<1.6% germanium-substitution) suppressed the Voc losses surface of materials, such as Cu(In,Ga)Se2 and silicon and increase
from 632 meV (for a pure Cu2ZnSnSe4 device) to 587 meV and increased carrier lifetime (Hsu et al., 2012; Vermang et al., 2013; Zhu et al., 2012).
the efficiency of the Cu2Zn(Sn1− x,Gex)Se4 solar cell to 10.1% (Giraldo ALD-coated films with a passivated interface show minimized inter-
et al., 2015). Neuschitzer et al. reported that a few at%-germanium- diffusion at the interface (Herz et al., 2003; Chambers et al., 2014).
substituted solar cells could achieve a 580 meV decrease in Voc, which is An extremely thin dielectric film, such as TiO2 and Al2O3, was coated
equivalent to the currently reported device (Neuschitzer et al., 2016b). between the CdS/Cu2ZnSnSe4 interface to passivate the interface and
minimize the interface recombination (Kim et al., 2017; Wu et al., 2014;
6.1.4. By passivation of defects Ranjbar et al., 2017b; Lee et al., 2016; Erkan et al., 2016). Wu et al.
Several studies have reported the passivation of these defects using deposited an ultrathin TiO2 layer between CdS and Cu2ZnSn(Sx,Se1− x)4
either high-temperature treatments or doping (Kronik et al., 1998; by atomic layer deposition (ALD) to reduce interface recombination (Wu
Rockett, 2005; J.V. Li et al., 2013; Ishizuka et al., 2009), which helps to et al., 2014). With the TiO2 passivation layer, the activation energy (EA)
enhance the solar cell performance by improving Voc and the minority of recombination was approximately 40 meV larger than that without
carrier lifetime. These various defects are caused by the disordered TiO2 passivation. This suggests that the TiO2 passivation layer decreases
arrangement of ion/elements or surface dangling bonds. Sodium (Na) is interfacial recombination (Wu et al., 2014). CV (capacitance–voltage)
used as a dopant to reduce or passivate the defect effects in Cu2ZnSnS4. and DLCP (drive-level capacitance profiling) analysis indicated a
Na is incorporated homogenously into the surface of the Cu2ZnSnS4 decrease in defect density across the interface of Cd/Cu2ZnSn(Sx,Se1− x)4
material. with the TiO2 passivation layer (Wu et al., 2014).
Zhou et al. examined the extent of defect passivation in the Cu2ZnSn Consequently, Cu2ZnSn(Sx,Se1− x)4-based solar cells with TiO2 film
(Sx,Se1− x)4 thin films using Na as a dopant (Zhou et al., 2013b). First, improved all device performance parameters, except for Jsc. The
Cu2ZnSnS4 nanocrystals were synthesized using a thermolysis process. decrease in Jsc may be due to a decrease in charge transport; this is
Na was then incorporated on the Cu2ZnSnS4 nanocrystal surface. Fig. 36 probably because of the insulation effect of the TiO2 layer. Hence, it is
presents a graphical representation of defect passivation in the kesterite- necessary to improve charge transport by optimizing the passive layer
based photovoltaic device. The carrier concentration and minority car­ thickness. Compared to the other thin film deposition methods, thick­
rier lifetime were increased by incorporating Na ions, thus improving ness control is more convenient using the ALD process. Ranjbar et al.
the solar cell performance. Different measurements, such as external used a solution-based method to deposit TiO2 film and passivate the
quantum efficiency (EQE) [Fig. 37(c)], lifetime measurement, and car­ front interface (Ranjbar et al., 2017b). The Voc of the device improved,
rier concentration measurement [Fig. 37(d) and 37(e)] carried out to but the FF and efficiency decreased. This may be due to the formation of
examine and understand solar cell performance after Na passivation. a high barrier at the conduction band after TiO2 layer deposition, and
After passivation by Na, the performance parameters of the solar cell hence a decrease in charge transport (Ranjbar et al., 2017b). After
increased, with efficiencies ranging from 3.89% to 6.15%, Voc from depositing the TiO2 layer, a potassium cyanide (KCN) treatment was
0.316 V to 0.361 V, and FF from 39.75% to 50.95%. Na did not affect used to promote charge transport; the Jsc and FF improved, and the ef­
grain growth, but it influenced the grain boundaries and minority carrier ficiency increased from 5.4% to 6.9% (Ranjbar et al., 2017b).
An Al2O3 layer is also a promising candidate for interface passivation
to improve the performance of Cu2ZnSn(Sx,Se1− x)4 based solar cells. Lee
et al. deposited an ultrathin Al2O3 layer for passivation by ALD. In the
Cu2ZnSn(Sx,Se1− x)4 film, the Al2O3 film enters the pinhole and grain
boundaries and passivates these portions (Lee et al., 2016). Generally,
sputtering damages the CdS surface during deposition of the TCO layer.
The Voc improved by ~5% when the Al2O3 layer replaced the i-ZnO film,
and the Cu2ZnSn(Sx,Se1− x)4-based solar cell showed 11.5% efficiency
(Lee et al., 2016). During CdS deposition, the Al2O3 film may be etched
by NH4OH, while the TiO2 film is stable in the CBD process (Ranjbar
Fig. 36. Schematic diagram of passivation of defects in the Cu2ZnSn(Sx,Se1− x)4 et al., 2017b). Hence, the series resistance was almost constant regard­
device. (Reprinted with permission from Ref. Zhou et al., 2013b). less of the Al2O3 film thickness at the Cu2ZnSnS4/CdS interface. In

44
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 37. (a) FE-SEM cross-section images of the (Mo/Cu2ZnSnS4:Na/Cu2ZnSnS4/Ag nanowires/ITO) solar cell, (b) J–V characteristics, (c) spectrum of EQE, (d)
spectra of TRPL and (e) C-V measurements of the carrier density. (Reprinted with permission from Ref. Zhou et al., 2013b).

donor defect CdCu formed at the Cu(In,Ga)Se2/CdS interface, resulting in

higher band bending. This converts the Cu(In,Ga)Se2 surface from p-
type to n-type, which helps separate the light-induced charge carriers
(Liao and Rockett, 2003a; Liao and Rockett, 2003b). In Cu2ZnSn(Sx,
Se1− x)4 devices, however, during CdS deposition, the neutral defect
complex [CdCu + VCu] can easily form. With a low concentration of
donor defect, CdCu and neutral defect CdZn appeared (Maeda et al.,
2012). Therefore, it is difficult to make an n-type surface on Cu2ZnSn(Sx,
Se1− x)4 films; thus, the advantage of intermixing copper and cadmium at
the interface CdS/Cu2ZnSn(Sx,Se1− x)4 is unclear. Kim et al. deposited a
5 nm thick Al2O3 layer between Cu2ZnSnSe4/CdS, which hindered the
intermixing of copper and cadmium [shown in Fig. 39(b)] (Kim et al.,
2017). The intensity of the PL spectra increased, showing a decrease in
non-radiative recombination for the device with the Al2O3 passivation
film [shown in Fig. 39(d)]. Finally, an improvement in Voc and FF was
observed (Kim et al., 2017).
The interface defect densities can be minimized using the epitaxial
process for the buffer layer on the Cu2ZnSn(Sx,Se1− x)4 absorber. Liu
et al. achieved CdS epitaxial growth on Cu2ZnSnS4 by sulfurization
under a combination of SnS and sulfur atmospheres, or with surface
defects annealed under a sulfur atmosphere, then passivated air
annealing (Liu et al., 2016). CdS epitaxial growth decreased the
recombination rate by reducing the interfacial defect density. Ulti­
mately, the solar cell efficiency is improved by 8.76% by increasing the
Fig. 38. PL spectra of a Cu(In,Ga)Se2 absorber film fabricated by varying the
lifetime of the minority carriers (Liu et al., 2016).
soda-lime glass thin film (SLGTF) thickness. The two luminescence peaks are On the other hand, the epitaxial growth approach reduces the
related to the donor–acceptor pair. (Reprinted with permission from interfacial defect density, but the lattice misalignment is ~7% in
Ref. Ishizuka et al., 2009). Cu2ZnSnS4/CdS interfaces, which is higher than that in Cu2ZnSnSe4
(~2.5%) and Cu(In,Ga)Se2 (~1.5%). Thus, the epitaxial growth of CdS is
contrast, the series resistance increased with increasing Al2O3 film detrimental to the Cu2ZnSnS4 device (Crovetto et al., 2016). Approxi­
thickness at the CdS/TCO interface (Lee et al., 2016). Erkan et al. mately 0.4% lattice misalignment of CeO2 with Cu2ZnSnS4 was
deposited an Al2O3 layer at the Cu2ZnSnS4/CdS interface by ALD and observed, which effectively reduced the interface defects. Furthermore,
reported that the acceptor defect densities decreased with a concomitant CBO at the Cu2ZnSnS4/CeO2 interface is − 0.120 ± 20 eV, which is
increase in depletion width. Consequently, Voc was improved (Erkan slightly less than the optimized CBO value and more beneficial than the
et al., 2016). conduction band offset of CdS/Cu2ZnSnS4. Crovetto et al. introduced a
Kim et al. observed copper accumulation and a cadmium deficiency CeO2 film between the Cu2ZnSnS4/CdS to decrease the interface
at the CdS/Cu2ZnSnSe4 interface, i.e., intensive intermixing of cadmium recombination (Crovetto et al., 2016). The mobility of electrons was
and copper occurred at the Cu2ZnSnSe4/CdS interface during CdS small due to the larger effective electron mass of CeO2, and the Voc
deposition. [Fig. 39(a)] (Kim et al., 2017). In Cu(In,Ga)Se2 devices, the showed improvements (Crovetto et al., 2016), but Jsc decreased because

45
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 39. HAADF-STEM images and corre­

sponding EDS maps of Cu, Cd, and S in
Cu2ZnSnSe4 solar cells (a) without and (b)
with an Al2O3 passivation layer at the
Cu2ZnSnSe4/CdS interface. (c) Changes in Voc
and efficiency as a function of the Al2O3
thickness (solid lines a guide to the eye and a
dashed line indicates a respective reference).
(d) 10 K PL spectra of Cu2ZnSnSe4 solar cells
with (purple) and without (green) a top Al2O3
passivation layer at the Cu2ZnSnSe4/CdS
interface. (Reprinted with permission from
Ref. Oh and Shin et al., 2017).

of the repressed photocurrent transport. Therefore, the CeO2 films (Zhao et al., 1999; Saint-Cast et al., 2010; Martın et al., 2001), SiCx
should be ultrathin and can be obtained by ALD. (Schmidt et al., 2001), SiNx (Saint-Cast et al., 2010), Al2O3 (Martın et al.,
In summary, the presence of defects at different interfaces promotes 2001), or Al2O3(SiO)/SiNx (Dullweber et al., 2012), have been employed
various interfacial recombinations. Thus, interface passivation is as the passivating layer for silicon photovoltaic devices. Vermang et al.
essential for high-efficiency Cu2ZnSn(Sx,Se1− x)4 solar cells. The inter­ deposited an ultrathin Al2O3 layer on the back contact of the Cu2ZnSnS4
face can be passivated either by an ultrathin layer of ultrathin di­ device by ALD. They reported a decrease in secondary phase formation
electrics, such as TiO2, CeO2, Al2O3, and SiO2, or epitaxial growth of the and carrier recombination. As a result, the performance parameters of
buffer. More research will be needed to determine if interface passiv­ the passivated device were superior to those of a device without rear
ation enhances the flow of charge carriers. passivation (Vermang et al., 2015). Kim et al. also deposited a layer of
dielectric material (Al2O3 or SiO2) to passivate the Mo/Cu2ZnSnSe4
6.2.2. Passivation of rear or back interface interface. They used nanosphere lithography to fabricate a passive layer
The semiconductor metal interface quality also plays a vital role in with nano-patterned local contacts (as shown in Fig. 40) (Kim et al.,
improving the performance parameter of PV devices. Therefore, the Voc 2017). Remarkable improvement has been observed in all performance
can be improved by minimizing recombination at the rare/absorber parameters. The luminescence intensity of the LT-PL spectra in the
interface, and Jsc can be improved by increasing the reflection at the Cu2ZnSnSe4 device was higher in the device with the dielectric layer
backside. Hence, the passivity of the semiconductor metal interface can (Al2O3 or SiO2), showing that Al2O3 or SiO2 effectively passivated the
be achieved using dielectric layers. Many dielectric layers, e.g., SiO2 back contact interface (Kim et al., 2017). In contrast to Vermang et al.

Fig. 40. (a) Graphical representation of the deposition of a passivation thin film with nano-patterned local contacts. (b) FE-SEM images of the nano-patterned contact
holes on Mo after lithography. (c) FE-TEM cross-sectional image of an Al2O3 passivation film (bottom, white dotted square) with a local contact between Cu2ZnSnSe4
and MoSe2. The insets are Fourier transform images in the area near the surface (yellow dotted square) and bulk (red dotted square), respectively. (d) J–V char­
acteristics of the Cu2ZnSnSe4 device with and without a passivation film at the interface (d) Al2O3 and (e) SiO2. (f) 10 K photoluminescence spectrum of Cu2ZnSnSe4
device with and without a passivation film at the interface (Al2O3). (Reprinted with permission from Ref. Oh and Shin et al., 2017).

46
M. Sahu et al. Solar Energy 230 (2021) 13–58

(Vermang et al., 2015) and Kim et al. (Kim et al., 2017), Liu et al. used a eV to 0.0 eV) for kesterite materials is slightly smaller than CuInSe2 (−
different fabrication process. They deposited an intermediate Al2O3 0.5 eV to 0.0 eV), whereas, the chemical potential range for zinc (~0.2
layer on the top of the molybdenum film that was converted to a self- eV) and tin (~0.6 eV) is much smaller than indium (1.0 eV) in CuInSe2
organized nanopattern, with nanosize pores between Cu2ZnSnS4/MoS2 (Chen et al., 2013). Consequently, numerous secondary phases, such as
interface for their interactions (Liu et al., 2017). During the initial point CuS(e), Cu2S(e), Cu2SnS(e)3, ZnS(e), SnS(e), and SnS(e)2, coexist easily
of sulfurization, the intermediate Al2O3 layer hinders the reaction at the with Cu2ZnSn(Sx,Se1− x)4. A zinc deficiency promotes the segregation of
interface of absorber/rare contact. As a result, MoS2 thickness Cu2SnS(e)3, SnS(e), and SnS(e)2 secondary phases (Chen et al., 2013).
decreased, and other phases, such as ZnS and voids, disappeared. Both Cu2SnS(e)3 and SnS(e) phases have a smaller bandgap than
Furthermore, the Cu2ZnSnS4 crystallinity increased with an improved Cu2ZnSn(Sx,Se1− x)4; thus, Voc losses increase. In contrast, in SnS(e)2, the
lifetime of minority carriers. The efficiency of the Cu2ZnSnS4 device conduction band edge is much smaller than Cu2ZnSn(Sx,Se1− x)4 (Schlaf
improved from 7.34% to 8.65% (Liu et al., 2017). et al., 1999; Chen et al., 2011). This is due to the recombination of
Recombination at the absorber/rare contact interface can be minority charge carriers and the production of a shunt path. Voc and FF
improved by producing an electrostatic field at the rare interface. This of Cu2ZnSnSe4 solar cells are decreased (Temgoua et al., 2015; Vauche
can be achieved by fabricating a material with a high work function et al., 2016; Vora et al., 2012), whereas excess zinc induces the ZnS(e)
(HWF), such as MoO3 and WO3, between the Cu2ZnSn(Sx,Se1− x)4 and phase. A less detrimental effect of the ZnS(e) phase was observed on the
molybdenum layers because the MoO3 work function (6.5 eV) is higher rare contact interface. On the front interface, however, charge carrier
than that of Cu2ZnSn(Sx,Se1− x)4 (5.2 eV) (Antunez et al., 2017). transport was blocked by forming a large spike-like CBO between the
Therefore, to equilibrate the Fermi level, the electrons begin to flow ZnS(e)/Cu2ZnSn(Sx,Se1− x)4 interface, causing a decrease in Jsc but an
from the lower work function Cu2ZnSn(Sx,Se1− x)4 to the higher work increase in Rs (Schwarz et al., 2013; Xie et al., 2015; Timo Wätjen et al.,
function MoO3 when Cu2ZnSn(Sx,Se1− x)4 is placed in contact with 2012; Hsu et al., 2013; Nagoya et al., 2011). Furthermore, the existence
MoO3; hence, an electrostatic field is formed. Holes will be driven by of the CuxS phase induced a shunt path, which increased the conduc­
rare contact, while electrons will flow toward the front contact, which tivity (Vora et al., 2012; Abdullaev et al., 1967; Fuertes Marrón et al.,
reduces electron-hole recombination at the absorber/rare interface 2004; Hsieh et al., 2011). This suggests that the aggregation of sec­
(Fig. 41). Ranjbar et al. inserted a very thin MoO3 layer between the ondary phases either on the front or rare interfaces has detrimental ef­
Cu2ZnSnSe4 and molybdenum layers and observed an improved mi­ fects on Cu2ZnSn(Sx,Se1− x)4 solar cells by interface recombination
nority lifetime and Voc because of the reduction of rare contact recom­ (Fairbrother et al., 2013; X. Yin et al., 2014). Therefore, structure con­
bination (Ranjbar et al., 2017a). trol, the avoidance of secondary phases, and the elimination of sec­
Zhou et al. inserted a very thin layer of carbon between the ondary phases are important for highly efficient kesterite solar cells. The
Cu2ZnSnS4 absorber and molybdenum rare contact, which reduced/ secondary phases from the rare contact interface can be removed by
removed the obstructions to carrier transportation through voids (Zhou impeding the reaction between the Mo back contact and Cu2ZnSn(Sx,
et al., 2015). This suggests that carbon attaches to the inner wall of Se1− x)4 absorber by introducing an intermediate layer, such as Al2O3,
voids, making contact between the Cu2ZnSnS4 and MoS2 in the void TiN, and ZnO, or modifying the post-annealing process (Scragg et al.,
portion after post-annealing. This enhanced the free carrier trans­ 2013; López-Marino et al., 2013a, 2013b; Li et al., 2014; X. Liu et al.,
portation significantly and increased Jsc (Zhou et al., 2015). Voc and FF 2014; Vermang et al., 2015; Nam et al., 2016). A chemical etching
were relatively constant because the intermediate carbon film did not process was used to eliminate the secondary phases from the front
influence the MoSe2 thickness (Zhou et al., 2015). interface of the kesterite film.
The potassium cyanide (KCN) etchant is generally used to eliminate
6.2.3. Etching of secondary phases the CuxS(e) secondary phases (Tanaka et al., 2012; G.Y. Kim et al., 2014;
Theoretically, the thermodynamic stabilities of both Cu2ZnSnS4 and Nguyen et al., 2015; Buffiere et al., 2015). Raman spectroscopy and
CuInSe2 have been estimated. Because of many competing secondary Kelvin probe force microscopy (KPFM) confirmed the successful elimi­
phases, there is a very narrow region for the formation of Cu2ZnSnS4 and nation of CuxS(e) secondary phases by the KCN etchant (Tanaka et al.,
Cu2ZnSnSe4 in the pure phase compared to CuInSe2, as shown in Fig. 42 2012; G.Y. Kim et al., 2014; Nguyen et al., 2015). As the KCN etchant is
(Chen et al., 2013). On the other hand, the range of stable chemical hazardous and harmful to health, it should use in very small amounts.
potentials for the constituents in kesterite Cu2ZnSnS4 and Cu2ZnSnSe4 is Buffiere et al. used a 5% aqueous solution of KCN etchant and 0.5 wt% of
smaller than CuInSe2. The range of chemical potentials of copper (− 0.4 potassium hydroxide (KOH) was added for safety purposes (Buffiere
et al., 2015). After 30 s of etching, the FF of the Cu2ZnSnSe4 device
improved. On the other hand, Zn, Sn, and Se were eliminated prefer­
entially by KOH from the Cu2ZnSn(Sx,Se1− x)4 surface. As a result, the
copper composition on the Cu2ZnSn(Sx,Se1− x)4 surface increased after a
long etching period (Buffiere et al., 2015). The richness of copper at the
surface increased the recombination rate at the CdS/Cu2ZnSnSe4 inter­
face, and the FF of the Cu2ZnSnSe4 device decreased. Therefore, to
improve the Cu2ZnSnSe4 device efficiency, the KCN etching time must
be less than 30sec, and the KOH concentration must be reduced (Buffiere
et al., 2015). On the other hand, KCN etching eliminates the CuxS phase
and improves the Cu2ZnSn(Sx,Se1− x)4 device performance. Bär et al.
reported a change in the surface composition after KCN etching.
Therefore, a cliff-like conduction band offset was observed due to the
increase in surface bandgap from 1.53 ± 0.15 eV to 1.91 ± 0.15 eV (Bär
et al., 2011). The energetic barrier for recombination of charge carrier
and Voc increased (Bär et al., 2011).
In addition to KCN etching eliminating the surface overlayer, such as
native oxides and airborne contamination, contamination-induced sur­
Fig. 41. Schematic band alignment at the interface of Cu2ZnSn(Sx,Se1− x)4/Mo face defects and interface recombination decrease (Erkan et al., 2015).
with a layer of a high work function material. (Reprinted with permission from Furthermore, Pinto et al. successfully eliminate the Cu2− x(S,Se) sec­
Ref. Gao et al., 2018a, 2018b). ondary phase using ethylenediamine and 2-mercaptoethanol chemical

47
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 42. Theoretical estimated phase stable region for Cu2SnSe8 (left-side), Cu2ZnSnS4 (middle-side), and Cu2ZnSnSe4 (right-side). (Reprinted with permission from
Ref. Chen et al., 2013).

mixture, which is safer (Pinto et al., 2016). This suggests that CuxS and PL spectroscopy. Ren et al. reported that for the front interface of
phases can be eliminated successfully using H2O2 according to the Cu2ZnSnS4 based solar cell, the SnS phases are detrimental, while the
following reaction. (Chavda et al., 2016). SnS phases passivated the rare contact by the formation of spike-like
CBO at the Cu2ZnSnS4/SnS interface, which produced a barrier for
Cu2 + S2− + 4H2O2 CuSO4 + 4H2O (2)
electron transportation to the rare contact, as shown in Fig. 43 (Ren
Cu + 2S 2−
+ 4H2O2 Cu2SO4 + 4H2O (3) et al., 2017). Hence, the location of the SnS phase formation needs to be
controlled. Bromine with methanol (Br2-MeOH) can be used as an
A hydrochloric acid (HCl) solution can be used for the selective etchant for Cu2SnS(e)3 secondary phases (Mousel et al., 2013; Timmo
removal of ZnS secondary phases (Fairbrother et al., 2012). Fairbrother et al., 2010).
et al. observed that after HCl etching, the efficiency increased from 2.7% In summary, the formation of secondary phases is easy because the
to 5.2% for the Cu2ZnSnS4 device (Fairbrother et al., 2012). Mousel et al. pure phase-stable region is too narrow to form kesterite Cu2ZnSnS4 and
used the HCl etching process to remove ZnSe phases, and the device Cu2ZnSnSe4. Secondary phases are harmful to the performance of the
efficiency was increased from 4% to 5% (Mousel et al., 2013). ZnSe Cu2ZnSn(Sx,Se1− x)4 device. Chemical etching can be an efficient way to
phases are usually eliminated by oxidizing processes, such as potassium eliminate these secondary phases and improve the performance of the
permanganate/sulfuric acid (KMnO4/H2SO4) solution, followed by a Cu2ZnSn(Sx,Se1− x)4-based devices.
sodium sulfide (Na2S) etching (López-Marino et al., 2013a, 2013b).
First, the Se0 phase forms through the oxidation of ZnSe. Subse­ 6.2.4. Impact of alkali and other metal dopants
quently, Na2S is used to remove the Se0 phases. López-Marino et al. used Several doping strategies of alkali metals have been used to improve
three types of oxidizing agents, H2O2, KMnO4, and potassium dichro­ the performance of solar cells. Doping with an alkali metal, such as Na,
mate (K2Cr2O7), to etch the ZnSe phase. KMnO4 was a better etchant had favorable effects on Cu(In,Ga)Se2 and Cu2ZnSn(Sx,Se1− x)4-based
than the other two etchants (López-Marino et al., 2013a, 2013b). photovoltaic devices (Hsieh et al., 2016; Nagaoka et al., 2014a). Na
Remarkable improvement in the solar cell performance was observed doping also improved the electrical conductivity and grain size (Zhou
after removing the ZnSe phase. Jsc and Rs improved due to the reduced et al., 2015; X. Liu et al., 2014; López-Marino et al., 2013a, 2013b; Zeng
hindrance of charge carrier transportation because of the ZnSe phases. et al., 2015). Recently, remarkable enhancement on the Cu2ZnSnS4
The Voc, FF, and Rsh improvement can be attributed to the surface device was achieved using several other alkali metals as dopants, such as
passivation of the Cu2ZnSnSe4 absorber material by the chemical Li, K, and Rb (Yang et al., 2016; Hsieh et al., 2016; Nagaoka et al.,
method. Ammonium sulfide [(NH4)2S] etchant can be used to eliminate 2014a).
SnS(e) and SnS(e)2 from the surface of the Cu2ZnSn(Sx,Se1− x)4 absorber Solar cells based on the Cu2ZnSn(Sx,Se1− x)4 absorber material with
(Vauche et al., 2016; Xie et al., 2015; Xie et al., 2014; Buffière et al., the Mo/ Cu2ZnSn(Sx,Se1− x)4/CdS/i-ZnO/Al:ZnO/Ni/Al cell structure
2017; Buffière et al., 2015b). XRD and Raman spectroscopy before and were fabricated, and their efficiencies were measured. The
after etching detected the elimination of Sn–Se phases by comparing the
Sn composition (Xie et al., 2014). Xie et al. stated that the (NH4)2S
etchant might passivate the Cu2ZnSnSe4 surface by eliminating the
surface oxides and sulfur-passivated content other than etching the Sn-
Se phases. Hence, all the device performance parameters improved,
and the efficiency increased 20–60% after etching with (NH4)2S (Xie
et al., 2014). On the other hand, the solar cell performance deteriorated
when the etching time was increased due to preferential etching of Se,
In, and Zn, leading to an increase in the copper composition at the buffer Fig. 43. (a) Band edges at Cu2ZnSnS4/SnS interface of and (b) charge carrier
CdS/absorber Cu2ZnSnSe4 interface (Buffière et al., 2017). Wei et al. transport paths close to the interface of the Cu2ZnSnS4/SnS. (Reprinted with
used Na2S to eliminate the SnS phase and analyzed it by Raman mapping permission from Ref. Ren et al., 2017).

48
M. Sahu et al. Solar Energy 230 (2021) 13–58

incorporation of Na, Li, and Rb to Cu2ZnSn(Sx,Se1− x)4-based solar cells 7. Conclusion and perspective
was beneficial and increased the solar cell efficiency by 5.5% (Altamura
et al., 2016). The theoretical approach showed that Na, Li, and Rb Any solar absorber technology must provide a promising high effi­
doping enhanced the solar cell performance by increasing the grain size ciency, have low or no toxicity, and be manufactured economically with
and suppressing the formation of secondary phases (Collord et al., simple integrations with existing manufacturing infrastructure.
2014). To include the experimental alkali elements, Cu2ZnSnS4 was Kesterite-based technologies, particularly Cu2ZnSn(Sx,Se1− x)4, have
exposed to the dopants in the precursor solution before post-deposition attracted considerable attention. The main cause of this may be the
annealing. Altamura et al. used HCl and thiourea (HCl + TU) with alkali improvement in the performance parameter of the device on the labo­
metals doping to improve the efficiency. Without the HCl + TU treat­ ratory scale and the promising properties of this absorbing material,
ment, the efficiency was 4.8%, 5.7%, and 5.5% for Na, Li, and Rb, which is already a promising candidate to take the place of commercial
respectively (Altamura et al., 2016; Granath et al., 2000). In contrast, Cu(In,Ga)Se2 absorbers. The main objective of this statement is the
after the HCl + TU treatment, the efficiency increased to 6.0%, 6.2%, abundance, low cost of raw materials, such as zinc and tin, and the
and 6.35% for Na, Li, and Rb, respectively, as shown in Fig. 44(a). Tong adoption of new low-cost technologies for the fabrication of Cu2ZnSn(Sx,
et al. used potassium as a dopant for Cu2ZnSnS4 solar cells. The grain Se1− x)4 absorber layers for PV devices. This review summarized the
size increased and the ZnS secondary phases decreased after K doping, as multi-component quaternary chalcogenide solar absorber material,
shown in Fig. 44(b) and 44(c). Jsc increased and Voc decreased, indi­ especially the theoretical and experimental results in various aspects of
cating a decrease in Rs of the device, as shown in Fig. 45, but the device defects and interface, their origins, and effects on the PV devices. Sig­
efficiency improved (Johnson et al., 2014; Tong et al., 2014). nificant improvements in the Cu2ZnSn(Sx,Se1− x)4 solar cell performance
Many studies suggested that Li increases the hole carrier density by have been published, from the first device of 0.66% PCE to the world
passivating the defects, thereby improving the efficiency (Hsieh et al., record device of 12.6%. Despite this, the efficiencies achieved and other
2016; Xin et al., 2015). Recently, the use of Li metal as a dopant and its performance parameters in the Cu2ZnSn(Sx,Se1− x)4-based solar cell are
influence on the CdS/Cu2ZnSnS4 interface was reported (Yang et al., still too low to meet commercial viability requirements. In terms of the
2016). The LiOH was used as the Li source to fabricate the (LixCu2x)ZnSn solar cell architecture, both Cu(In,Ga)Se2 and Cu2ZnSnS4 devices have
(SxSe1− x)4 device. EQE analysis revealed a blue shift, indicating an in­ similarities. On the other hand, the Cu(In,Ga)Se2 champion PV device a
crease in bandgap by Li incorporation, as shown in Fig. 46. (LixCu2x) boasts an efficiency of 23.35% along with Voc = 730 mV, Jsc = 39.6 mA
ZnSn(SxSe1− x)4 absorber bandgaps were measured by plotting [E ln(1 - cm− 2, and fill factor (FF) = 80.4% (Nakamura et al., 2019). The Cu2ZnSn
EQE)]2 vs. E, as shown in Fig. 46(b). The bandgap of the material (Sx,Se1− x)4-based PV device has shown a stagnant efficiency of 12.6%
increased with increasing Li/Cu ratio [Fig. 46(c)]. The VBM and CBM of with Voc = 510 mV, Jsc = 35.2 mA cm− 2, and fill factor (FF) = 69.8% for
(LixCu2x)ZnSn(SxSe1− x)4 with different Li/Cu ratios were measured by more than a half-decade (Wang et al., 2014). Comparing the data on the
UPS, as shown in Fig. 46(d). Fig. 46(d) indicates that the level of the properties of the two technologies reveals a large Voc difference. The
VBM and CBM with Li /Cu ratio shifted toward a lower energy. Cu2ZnSn(Sx,Se1− x)4 technique has an approximately 62% lower Voc than
Compared to VBM, the shift in CBM was small. Thus, the bandgap Cu(In,Ga)Se2 (35%) concerning their respective band gaps. These facts
increased, and the Voc was improved. A spike-like band offset arrange­ highlight the need for a properly systematic technical and simulation
ment was found at the CdS/LixCu2xZnSn(Sx,Se1− x)4 interface. The approach to find the main reason behind the large Voc losses.
incorporation of Li in Cu2ZnSn(Sx,Se1− x)4 makes a tunable bandgap with The main chemistry and physics mechanisms for these huge Voc
a change in CBO. Hence, the Voc and efficiency of the Cu2ZnSn(Sx, losses are still unclear. The huge Voc losses were attributed to atomic-
Se1− x)4 device improved (Yang et al., 2016). level disorders that induce sub-band distortion. Thus, researchers have
The above discussion suggests that doping with alkali metals can be a focused on the Cu/Zn disorder to perceive and resolve this problem. The
promising and beneficial path for the fabrication of high-efficiency population of VCu defects compared to CuZn defects in stoichiometric
Cu2ZnSn(Sx,Se1− x)4-based solar cells. samples is much lower. On the other hand, the VCu defect becomes

Fig. 44. (a) J–V characteristics of the Mo/Cu2ZnSn(Sx,Se1− x)4/CdS/TCO/Ni/Al device (best performing) sample doped with alkali elements (A = Li, B = Rb, and C =
Na) after a thiourea treatment. (Reprinted with permission from Ref. Altamura et al., 2016), (b) Surface and (c) cross-section images of Cu2ZnSnS4 thin films with
different K-doping concentrations (Mo glass/ Cu2ZnSnS4/CdS/ZnO/ITO). (Reprinted with permission from Ref. Tong et al., 2014).

49
M. Sahu et al. Solar Energy 230 (2021) 13–58

Fig. 45. (a) J–V characteristics curve and (b) EQE spectra of Cu2ZnSnS4 device with different K-doped Cu2ZnSnS4 absorber layer. (Reprinted with permission from
Ref. Tong et al., 2014).

Fig. 46. Effect of the Li/Cu ratio on (a) EQE analysis (b, c) Optical bandgap and (d) change in the conduction band and valence band edges with respect to CdS in the
(LixCu2− x)ZnSn(Sx,Se1− x)4 layer devices. (Reprinted with permission from Ref. Yang et al., 2016).

dominant under the Cu/(Zn + Sn) ≈ 0.8 and Zn/Sn ≈ 1.2 conditions and the interface of CdS and Cu2ZnSnS4 or Cu2ZnSnSe4 or Cu2ZnSn(Sx,
contributes to p-type conductivity. This shows the relationship of a real Se1− x)4 is cliff-like or a slightly spike-like barrier, respectively, which is
PV device with a low Cu/(Zn + Sn) ratio and high efficiency. The self- unfavorable for the Voc or Jsc, respectively. Second, a thick MoS2 or
compensated defect complexes, such as [VCu + ZnCu], [ZnSn + 2ZnCu], MoSe2 layer easily forms at the back interface during the post-deposition
and [2CuZn + SnZn], easily form because of non-stoichiometry in the process at high temperatures. Therefore, the MoS2 or MoSe2 film
kesterite material. [2CuZn + SnZn] clusters induce electron-entrapping thickness at the Cu2ZnSnS4/Mo or Cu2ZnSnSe4/Mo interface must be
levels in the absorbing materials and are harmful to the solar cell per­ controlled. Finally, binary and ternary phases form easily at the in­
formance. [2CuZn + SnZn] defect complex produces electron-trapping terfaces owing to the availability of narrow regions for stable Cu2ZnSn
states in Cu2ZnSnS4 and Cu2ZnSnSe4. On the other hand, the effect of (Sx,Se1− x)4.
electron-trapping induced by this defect is much larger in Cu2ZnSnS4 The integration of simulations with the experimental data in terms of
and Cu2ZnSnSe4. This may explain the high efficiency of the Cu2ZnSn the defect and interface loss is important for overcoming these issues
(Sx,Se1− x)4 device when the sulfur content is low. Another possibility related to Voc losses and high efficiency and can confer a deep insight
may be the concentration of isolated deep donor defects, such as SnZn beyond the limited boundaries. Therefore, the systematic development
and VSe in Cu2ZnSnSe4. of the association of the properties of various scales to estimate the
In addition to the effects of defects, the different interface chemistry desired results in Cu2ZnSn(Sx,Se1− x)4 technology was presented. Lately,
is another aspect influencing the high Voc losses. First, the conduction the reports of various measurements are dependent on the analysis re­
band alignment at the front interface of the buffer CdS/absorber ports. For example, XRD, EDX, and Raman spectroscopy are impres­
Cu2ZnSn(Sx,Se1− x)4 is not optimal. The conduction band offset found at sively efficient tools for evaluating kesterite-type materials. Information

50
M. Sahu et al. Solar Energy 230 (2021) 13–58

on the composition of the kesterites solid solution is included to obtain recombination. Lastly, binary and ternary phases easily form due to the
qualitative results for the formation of quaternary and other phases and availability of narrow regions for the stable Cu2ZnSn(Sx,Se1− x)4. The
the crystal structure of the pure phase. X-ray absorption spectroscopy removal of these secondary phases by chemical etching is an efficient
and synchrotron-based high-resolution X-ray fluorescence analysis offer way. On the other hand, several approaches have been made to remove/
unique and powerful contributions to address the lack of compositional reduce detrimental defects and interface loss in Cu2ZnSn(Sx,Se1− x)4 but
and structural homogeneity, which influence the optical and electrical without significant success.
properties of the material. The structural origin of point defects of non- In conclusion, this paper reviewed various aspects of the defect and
stoichiometric kesterites is revealed by neutron diffraction. FE-SEM, FE- defect complex and the interface and its detrimental influence on PV
TEM, and AFM are the most widely used techniques. FE-SEM is used for devices. The importance of theoretical and characterization techniques
assessing structural morphology, void detection, grain size monitoring, and their role in achieving high-efficiency solar cells is critical. Studies
elemental mapping, and composition analysis, while FE-TEM techniques of the chemistry of defects and the interface have revealed some
are used for defect studies, interplanar spacing, mono or polycrystalline, remarkable outcomes that can improve the interface properties and have
amorphous phases, and atomic structure of heterointerfaces. AFM pro­ beneficial defects. Furthermore, these studies can help reduce the
vides information on the morphology in 3D mode, quick elemental detrimental defects and interface recombination. Nevertheless, it is
composition. Measurements of the J-V characteristics can be used to essential to correlate the outcomes of theoretical and experimental
determine the formation of a p–n junction and performance parameters; studies to remove the obstacle to manufacturing high-efficiency solar
they also provide information on the electric field, depletion width, cells.
defect energy level, and defect density. DLTS is a powerful technique for
identifying defect trapping centers, capture cross-section, and defect
Declaration of Competing Interest
activation energy. KPFM provides information on carrier transport in
kesterites and other solar semiconductor materials. Contact potential
The authors declare that they have no known competing financial
difference (CPD) analysis can measure the work function separate from
interests or personal relationships that could have appeared to influence
the distribution of varied phases. LTPL measurements are valuable for
the work reported in this paper.
locating potential fluctuation, donors defects, and acceptor defects, such
as CuSn or ZnSn VCu SnCu, in a semiconducting material. Interface defects
are more harmful to the solar cell stability than bulk defects, as deter­ Acknowledgments
mined by in situ admittance spectroscopy.
Despite the different experimental techniques to measure the Author P. Sharma would like to acknowledge the IITB-ISRO space
composition, the carrier type and concentration, and the solar cell per­ technology cell at IIT Bombay India for funding the research work vide
formance, it is still difficult to investigate the isolated and complex de­ Project code 15ISROC002. Author C. Park would like to thanks the
fects in the prepared materials. These investigations entail high-quality KENTECH Research Grant (KRG2021-01-008). The authors are thankful
crystalline materials, which are rare for quaternary kesterite Cu2ZnSnS4 to the anonymous reviewers for their constructive suggestion.
and Cu2ZnSnSe4 materials. Optical luminescence or deep-level transient
spectroscopy, capacitance, and admittance spectroscopy can provide Authors contribution
essential information on the defect transition energy states in the
bandgap and carrier concentration. On the other hand, it is essential to Author M. Sahu conceived the present idea and wrote the manuscript
determine the cause of those states and what are the dominant defects in with support from V. R. Minnam Reddy. Author P. Sharma and C. Park
the materials. Both first-principles computation and density functional discussed and provided critical feedback. All the author contributed to
theory (DFT) are the most widely used theoretical approaches, providing the final version of the manuscript.
more atomic level and electronic structural information on kesterite
materials. First-principles calculations have been used to evaluate the
References
role of copper vacancies and interstitial diffusion on the formation of
p–n junctions and Fermi level pinning. Equilibrium defect concentra­ Abdullaev, G., Aliyarova, Z., Asadov, G., 1967. Preparation of Cu2se single crystals and
tions and supercell model based on the first-principles approach using investigation of their electrical properties. Phys. Stat. Sol. (b) 21 (2), 461–464.
density functional theory (DFT) provide information on the formation of Adhi Wibowo, R., Soo Lee, E., Munir, B., Ho Kim, K., 2007. Pulsed laser deposition of
quaternary Cu2ZnSnSe4 thin films. Phys. Stat. Sol. (a) 204 (10), 3373–3379.
intrinsic/point defects by estimating its formation energy. In the Adhikari, N., Dubey, A., Khatiwada, D., Mitul, A.F., Wang, Q., Venkatesan, S.,
supercell model, it was assumed that defects are located in very large Iefanova, A., Zai, J., Qian, X., Kumar, M., 2015. Interfacial study to suppress charge
supercells and that the interactions of defects in the neighboring cells are carrier recombination for high efficiency perovskite solar cells. ACS Appl. Mater.
Interfaces 7 (48), 26445–26454.
negligible. In Cu2ZnSn(Sx,Se1− x)4, the Cu-Zn anti-site defect is the most Ahmed, S., Reuter, K.B., Gunawan, O., Guo, L., Romankiw, L.T., Deligianni, H., 2012.
common because of the similar ionic radii and chemical properties of A high efficiency electrodeposited Cu2ZnSnS4 solar cell. Adv. Energy Mater. 2 (2),
Cu+ and Zn2+; for them, DFT calculations estimated low formation 253–259.
Ahn, S., Jung, S., Gwak, J., Cho, A., Shin, K., Yoon, K., Park, D., Cheong, H., Yun, J.H.,
energies. 2010. Determination of band gap energy (E g) of Cu 2 ZnSnSe 4 thin films: on the
In addition, several approaches have been made through experi­ discrepancies of reported band gap values. Appl. Phys. Lett. 97 (2), 021905.
ments and simulation to decrease the detrimental effect on PV devices. Alonso, M.I., Garriga, M., Rincón, C.D., Hernández, E., León, M., 2002. Optical functions
of chalcopyrite CuGa x In 1–x Se 2 alloys. Appl. Phys. A 74 (5), 659–664.
To modify the disorder by longer post-deposition, Raman spectroscopy
Altamura, G., Wang, M., Choy, K.-L., 2016. Influence of alkali metals (Na, Li, Rb) on the
can trace the disorder level and analyze the fluctuations in the bandgap, performance of electrostatic spray-assisted vapor deposited Cu 2 ZnSn (S, Se) 4 solar
which are affected by the Cu/Zn disorder. The conduction band offset cells. Sci. Rep. 6 (1), 1–9.
Altosaar, M., Raudoja, J., Timmo, K., Danilson, M., Grossberg, M., Krustok, J.,
can be adapted by modifying the structure of Cu2ZnSn(Sx,Se1− x)4 ma­
Mellikov, E., 2008. Cu2Zn1–x Cdx Sn (Se1–y Sy) 4 solid solutions as absorber
terial or by modifying alternative buffer layers, such as (Zn,Cd)S, Zn(O, materials for solar cells. Phys. Stat. Sol. (a) 205 (1), 167–170.
S), Zn1− xMgxO, Zn1− xSnxOy, and In2S3. MoS2 or MoSe2 film formation Alvarez-Garcıa, J., Pérez-Rodrıguez, A., Barcones, B., Romano-Rodrıguez, A.,
can be suppressed using intermediate layers, such as TiN, TiB2, and Morante, J., Janotti, A., Wei, S.-H., Scheer, R., 2002. Polymorphism in CuInS 2
epilayers: origin of additional Raman modes. Appl. Phys. Lett. 80 (4), 562–564.
MoO2 on the molybdenum back contact. Third, the front and rare Antunez, P.D., Bishop, D.M., Lee, Y.S., Gokmen, T., Gunawan, O., Gershon, T.S.,
interface defects can be passivated using different dielectric layers, such Todorov, T.K., Singh, S., Haight, R., 2017. Back contact engineering for increased
as Al2O3, TiO2, and SiO2; these interfacial defects promote recombina­ performance in kesterite solar cells. Adv. Energy Mater. 7 (15), 1602585.
Araki, H., Kubo, Y., Jimbo, K., Maw, W.S., Katagiri, H., Yamazaki, M., Oishi, K.,
tion at the interface. A high work function material (HWF), such as Takeuchi, A., 2009. Preparation of Cu2ZnSnS4 thin films by sulfurization of co-
MoO3 and WO3, can be used to suppress the back contact interface electroplated Cu-Zn-Sn precursors. Phys. Stat. Sol. C 6 (5), 1266–1268.

51
M. Sahu et al. Solar Energy 230 (2021) 13–58

Assmann, L., Bernede, J., Drici, A., Amory, C., Halgand, E., Morsli, M., 2005. Study of the Chen, S., Walsh, A., Gong, X.G., Wei, S.H., 2013. Classification of lattice defects in the
Mo thin films and Mo/CIGS interface properties. Appl. Surf. Sci. 246 (1–3), 159–166. kesterite Cu2ZnSnS4 and Cu2ZnSnSe4 earth-abundant solar cell absorbers. Adv.
Azimi, H., Hou, Y., Brabec, C.J., 2014. Towards low-cost, environmentally friendly Mater. 25 (11), 1522–1539.
printed chalcopyrite and kesterite solar cells. Energy Environ. Sci. 7 (6), 1829–1849. Chen, S., Walsh, A., Yang, J.-H., Gong, X.-G., Sun, L., Yang, P.-X., Chu, J.-H., Wei, S.-H.,
Babichuk, I., Semenenko, M., Golovynskyi, S., Caballero, R., Datsenko, O., Babichuk, I., 2011. Compositional dependence of structural and electronic properties of Cu 2 ZnSn
Li, J., Xu, G., Qiu, R., Huang, C., 2019. Control of secondary phases and disorder (S, Se) 4 alloys for thin film solar cells. Phys. Rev. B 83 (12), 125201.
degree in Cu2ZnSnS4 films by sulfurization at varied subatmospheric pressures. Sol. Chen, S., Yang, J.-H., Gong, X.-G., Walsh, A., Wei, S.-H., 2010b. Intrinsic point defects
Energy Mater. Sol. Cells 200, 109915. and complexes in the quaternary kesterite semiconductor Cu 2 ZnSnS 4. Phys. Rev. B
Babu, G.S., Kumar, Y.K., Bhaskar, P.U., Raja, V.S., 2008. Effect of post-deposition 81 (24), 245204.
annealing on the growth of Cu2ZnSnSe4 thin films for a solar cell absorber layer. Chua, R.H., Li, X., Walter, T., Teh, L.K., Hahn, T., Hergert, F., Mhaisalkar, S., Wong, L.H.,
Semicond. Sci. Technol. 23 (8), 085023. 2016. An experimentally supported model for the origin of charge transport barrier
Bag, S., Gunawan, O., Gokmen, T., Zhu, Y., Mitzi, D.B., 2012. Hydrazine-processed Ge- in Zn (O, S)/CIGSSe solar cells. Appl. Phys. Lett. 108 (4), 043505.
substituted CZTSe solar cells. Chem. Mater. 24 (23), 4588–4593. Cohen, J.D., Unold, T., Gelatos, A.V., Fortmann, C.M., 1992. Deep defect structure and
Bao, W., Ichimura, M., 2012a. First-principles study on influences of crystal structure and carrier dynamics in amorphous silicon and silicon-germanium alloys determined by
orientation on band offsets at the CdS/Cu2ZnSnS4 interface. Int. J. Photoenergy transient photocapacitance methods. J. Non-Cryst. Solids 141, 142–154.
2012. Collord, A., Hillhouse, H., 2016. Germanium alloyed kesterite solar cells with low
Bao, W., Ichimura, M., 2012. Prediction of the band offsets at the CdS/Cu2ZnSnS4 voltage deficits. Chem. Mater. 28 (7), 2067–2073.
interface based on the first-principles calculation. Jpn. J. Appl. Phys. 51 (10S), Collord, A., Xin, H., Hillhouse, H., 2014. Combinatorial Exploration of the Effects of
10NC31. Intrinsic and Extrinsic Defects in Cu $ _ {\bf 2} $ ZnSn (S, Se) $ _ {\bf 4} $. IEEE J.
Bao, W., Qiu, F.-Y., 2016. Band offsets engineering at CdxZn1− xS/Cu2ZnSnS4 Photovoltaics 5 (1), 288–298.
heterointerface. Chin. Phys. B 25 (12), 127102. Courel, M., Andrade-Arvizu, J., Vigil-Galán, O., 2014. Towards a CdS/Cu2ZnSnS4 solar
Bär, M., Schubert, B.-A., Marsen, B., Wilks, R.G., Pookpanratana, S., Blum, M., Krause, S., cell efficiency improvement: a theoretical approach. Appl. Phys. Lett. 105 (23),
Unold, T., Yang, W., Weinhardt, L., 2011. Cliff-like conduction band offset and KCN- 233501.
induced recombination barrier enhancement at the CdS/Cu2ZnSnS4 thin-film solar Courel, M., Andrade-Arvizu, J., Vigil-Galán, O., 2016. The role of buffer/kesterite
cell heterojunction. Appl. Phys. Lett. 99 (22), 222105. interface recombination and minority carrier lifetime on kesterite thin film solar
Barkhouse, D.A.R., Gunawan, O., Gokmen, T., Todorov, T.K., Mitzi, D.B., 2012a. Device cells. Mater. Res. Express 3 (9), 095501.
characteristics of a 10.1% hydrazine-processed Cu2ZnSn (Se, S) 4 solar cell. Prog. Crovetto, A., Yan, C., Iandolo, B., Zhou, F., Stride, J., Schou, J., Hao, X., Hansen, O.,
Photovoltaics Res. Appl. 20 (1), 6–11. 2016. Lattice-matched Cu2ZnSnS4/CeO2 solar cell with open circuit voltage boost.
Barkhouse, D.A.R., Haight, R., Sakai, N., Hiroi, H., Sugimoto, H., Mitzi, D.B., 2012b. Cd- Appl. Phys. Lett. 109 (23), 233904.
free buffer layer materials on Cu2ZnSn (SxSe1− x) 4: Band alignments with ZnO, Dalapati, G.K., Zhuk, S., Masudy-Panah, S., Kushwaha, A., Seng, H.L., Chellappan, V.,
ZnS, and In2S3. Appl. Phys. Lett. 100 (19), 193904. Suresh, V., Su, Z., Batabyal, S.K., Tan, C.C., 2017. Impact of molybdenum out
Biswas, K., Lany, S., Zunger, A., 2010. The electronic consequences of multivalent diffusion and interface quality on the performance of sputter grown CZTS based solar
elements in inorganic solar absorbers: Multivalency of Sn in Cu 2 ZnSnS 4. Appl. cells. Sci. Rep. 7 (1), 1–12.
Phys. Lett. 96 (20), 201902. Das, S., Chaudhuri, S.K., Bhattacharya, R.N., Mandal, K.C., 2014. Defect levels in
Botti, S., Kammerlander, D., Marques, M.A., 2011. Band structures of Cu 2 ZnSnS 4 and Cu2ZnSn (SxSe1− x) 4 solar cells probed by current-mode deep level transient
Cu 2 ZnSnSe 4 from many-body methods. Appl. Phys. Lett. 98 (24), 241915. spectroscopy. Appl. Phys. Lett. 104 (19), 192106.
Bourdais, S., Choné, C., Delatouche, B., Jacob, A., Larramona, G., Moisan, C., Lafond, A., Duan, B., Guo, L., Yu, Q., Shi, J., Wu, H., Luo, Y., Li, D., Wu, S., Zheng, Z., Meng, Q.,
Donatini, F., Rey, G., Siebentritt, S., 2016. Is the Cu/Zn disorder the main culprit for 2020. Highly efficient solution-processed CZTSSe solar cells based on a convenient
the voltage deficit in kesterite solar cells? Adv. Energy Mater. 6 (12), 1502276. sodium-incorporated post-treatment method. Journal of Energy Chemistry 40,
Brandt, R.E., Young, M., Park, H.H., Dameron, A., Chua, D., Lee, Y.S., Teeter, G., 196–203.
Gordon, R.G., Buonassisi, T., 2014. Band offsets of n-type electron-selective contacts Duan, H.S., Yang, W., Bob, B., Hsu, C.J., Lei, B., Yang, Y., 2013. The role of sulfur in
on cuprous oxide (Cu2O) for photovoltaics. Appl. Phys. Lett. 105 (26), 263901. solution-processed Cu2ZnSn (S, Se) 4 and its effect on defect properties. Adv. Funct.
Buffière, M., Barreau, N., Brammertz, G., Sahayaraj, S., Meuris, M., Poortmans, J., 2015a. Mater. 23 (11), 1466–1471.
Development of co-evaporated In2S3 buffer layer for Cu2ZnSnSe4 thin film solar Dubey, A., Adhikari, N., Venkatesan, S., Gu, S., Khatiwada, D., Wang, Q., Mohammad, L.,
cells, 2015 IEEE 42nd Photovoltaic Specialist Conference (PVSC). IEEE 1–4. Kumar, M., Qiao, Q., 2016. Solution processed pristine PDPP3T polymer as hole
Buffière, M., Brammertz, G., El Mel, A.-A., Barreau, N., Meuris, M., Poortmans, J., 2017. transport layer for efficient perovskite solar cells with slower degradation. Sol.
Effect of ammonium sulfide treatments on the surface properties of Cu2ZnSnSe4 thin Energy Mater. Sol. Cells 145, 193–199.
films. Thin Solid Films 633, 135–140. Duchatelet, A., Savidand, G., Vannier, R., Lincot, D., 2013. Optimization of MoSe2
Buffiere, M., Brammertz, G., Sahayaraj, S., Batuk, M., Khelifi, S., Mangin, D., El Mel, A.- formation for Cu (In, Ga) Se2-based solar cells by using thin superficial molybdenum
A., Arzel, L., Hadermann, J., Meuris, M., 2015. KCN chemical etch for interface oxide barrier layers. Thin Solid Films 545, 94–99.
engineering in Cu2ZnSnSe4 solar cells. ACS Appl. Mater. Interfaces 7 (27), Dullweber, T., Gatz, S., Hannebauer, H., Falcon, T., Hesse, R., Schmidt, J., Brendel, R.,
14690–14698. 2012. Towards 20% efficient large-area screen-printed rear-passivated silicon solar
Buffière, M., Mel, A.A.E., Lenaers, N., Brammertz, G., Zaghi, A.E., Meuris, M., cells. Prog. Photovoltaics Res. Appl. 20 (6), 630–638.
Poortmans, J., 2015b. Surface cleaning and passivation using (NH4) 2S treatment for Dutta, P., Kumar, M., Rathi, M., Ahrenkiel, S., Paul, S., Galipeau, D., Bommisetty, V.,
Cu (In, Ga) Se2 solar cells: a safe alternative to KCN. Adv. Energy Mater. 5 (6), 2013. Mechanism of the enhancement of electrical conductivity of nanocrystalline
1401689. silicon due to hydrogen plasma treatment. J. Nanosci. Nanotechnol. 13 (10),
Caballero, R., Kaufmann, C., Eisenbarth, T., Cancela, M., Hesse, R., Unold, T., Eicke, A., 6711–6720.
Klenk, R., Schock, H., 2009. The influence of Na on low temperature growth of CIGS Enkhbat, T., Kim, S., Kim, J., 2019. Device characteristics of band gap tailored 10.04%
thin film solar cells on polyimide substrates. Thin Solid Films 517 (7), 2187–2190. efficient CZTSSe solar cells sprayed from water-based solution. ACS Appl. Mater.
Cabas-Vidani, A., Haass, S.G., Andres, C., Caballero, R., Figi, R., Schreiner, C., Interfaces 11 (40), 36735–36741.
Márquez, J.A., Hages, C., Unold, T., Bleiner, D., 2018. High-Efficiency (LixCu1− x) Ennaoui, A., Lux-Steiner, M., Weber, A., Abou-Ras, D., Kötschau, I., Schock, H.-W.,
2ZnSn (S, Se) 4 Kesterite Solar Cells with Lithium Alloying. Adv. Energy Mater. 8 Schurr, R., Hölzing, A., Jost, S., Hock, R., 2009. Cu2ZnSnS4 thin film solar cells from
(34), 1801191. electroplated precursors: Novel low-cost perspective. Thin Solid Films 517 (7),
Campbell, S., Qu, Y., Major, J., Lagarde, D., Labbé, C., Maiello, P., Barrioz, V., Beattie, N., 2511–2514.
Zoppi, G., 2019. Direct evidence of causality between chemical purity and band-edge Ericson, T., Larsson, F., Törndahl, T., Frisk, C., Larsen, J., Kosyak, V., Hägglund, C., Li, S.,
potential fluctuations in nanoparticle ink-based Cu2ZnSn (S, Se) 4 solar cells. Platzer-Björkman, C., 2017. Zinc-Tin-Oxide Buffer Layer and Low Temperature Post
J. Phys. D Appl. Phys. 52 (13), 135102. Annealing Resulting in a 9.0% Efficient Cd-Free Cu2ZnSnS4 Solar Cell. Solar RRL 1
Chagarov, E., Sardashti, K., Kummel, A.C., Lee, Y.S., Haight, R., Gershon, T.S., 2016. (5), 1700001.
Ag2ZnSn (S, Se) 4: A highly promising absorber for thin film photovoltaics. J. Chem. Erkan, M.E., Chawla, V., Repins, I., Scarpulla, M.A., 2015. Interplay between surface
Phys. 144 (10), 104704. preparation and device performance in CZTSSe solar cells: Effects of KCN and
Chambers, B.A., MacDonald, B.I., Ionescu, M., Deslandes, A., Quinton, J.S., Jasieniak, J. NH4OH etching. Sol. Energy Mater. Sol. Cells 136, 78–85.
J., Andersson, G.G., 2014. Examining the role of ultra-thin atomic layer deposited Erkan, M.E., Chawla, V., Scarpulla, M.A., 2016. Reduced defect density at the CZTSSe/
metal oxide barrier layers on CdTe/ITO interface stability during the fabrication of CdS interface by atomic layer deposition of Al2O3. J. Appl. Phys. 119 (19), 194504.
solution processed nanocrystalline solar cells. Sol. Energy Mater. Sol. Cells 125, Fairbrother, A., Fontané, X., Izquierdo-Roca, V., Espíndola-Rodríguez, M., López-
164–169. Marino, S., Placidi, M., Calvo-Barrio, L., Pérez-Rodríguez, A., Saucedo, E., 2013. On
Chavda, A., Patel, M., Mukhopadhyay, I., Ray, A., 2016. Facile, noncyanide based the formation mechanisms of Zn-rich Cu2ZnSnS4 films prepared by sulfurization of
etching of spray deposited Cu2ZnSnS4 thin films for secondary phase removal. ACS metallic stacks. Sol. Energy Mater. Sol. Cells 112, 97–105.
Sustain. Chem. Eng. 4 (4), 2302–2308. Fairbrother, A., García-Hemme, E., Izquierdo-Roca, V., Fontané, X., Pulgarín-Agudelo, F.
Chen, S., Gong, X., Walsh, A., Wei, S.-H., 2009a. Crystal and electronic band structure of A., Vigil-Galán, O., Pérez-Rodríguez, A., Saucedo, E., 2012. Development of a
Cu 2 ZnSn X 4 (X= S and Se) photovoltaic absorbers: First-principles insights. Appl. selective chemical etch to improve the conversion efficiency of Zn-rich Cu2ZnSnS4
Phys. Lett. 94 (4), 041903. solar cells. J. Am. Chem. Soc. 134 (19), 8018–8021.
Chen, S., Gong, X., Walsh, A., Wei, S.-H., 2009b. Electronic structure and stability of Fernandes, P., Sartori, A., Salomé, P., Malaquias, J., Da Cunha, A., Graça, M.,
quaternary chalcogenide semiconductors derived from cation cross-substitution of II- Gonzalez, J., 2012. Admittance spectroscopy of Cu2ZnSnS4 based thin film solar
VI and I-III-VI 2 compounds. Phys. Rev. B 79 (16), 165211. cells. Appl. Phys. Lett. 100 (23), 233504.
Chen, S., Gong, X., Walsh, A., Wei, S.-H., 2010a. Defect physics of the kesterite thin-film Fontané, X., Calvo-Barrio, L., Izquierdo-Roca, V., Saucedo, E., Pérez-Rodriguez, A.,
solar cell absorber Cu 2 ZnSnS 4. Appl. Phys. Lett. 96 (2), 021902. Morante, J., Berg, D., Dale, P., Siebentritt, S., 2011. In-depth resolved Raman

52
M. Sahu et al. Solar Energy 230 (2021) 13–58

scattering analysis for the identification of secondary phases: Characterization of Cu Guo, Q., Ford, G.M., Yang, W.-C., Walker, B.C., Stach, E.A., Hillhouse, H.W., Agrawal, R.,
2 ZnSnS 4 layers for solar cell applications. Appl. Phys. Lett. 98 (18), 181905. 2010. Fabrication of 7.2% efficient CZTSSe solar cells using CZTS nanocrystals.
Ford, G.M., Guo, Q., Agrawal, R., Hillhouse, H.W., 2011. Earth abundant element Cu2Zn J. Am. Chem. Soc. 132 (49), 17384–17386.
(Sn1− x Ge x) S4 nanocrystals for tunable band gap solar cells: 6.8% efficient device Haass, S.G., Andres, C., Figi, R., Schreiner, C., Bürki, M., Romanyuk, Y.E., Tiwari, A.N.,
fabrication. Chem. Mater. 23 (10), 2626–2629. 2018. Complex interplay between absorber composition and alkali doping in high-
Friedlmeier, T.M., Wieser, N., Walter, T., Dittrich, H., Schock, H., 1997. Heterojunctions efficiency kesterite solar cells. Adv. Energy Mater. 8 (4), 1701760.
based on Cu2ZnSnS4 and Cu2ZnSnSe4 thin films. Proceedings of the 14th European Haass, S.G., Diethelm, M., Werner, M., Bissig, B., Romanyuk, Y.E., Tiwari, A.N., 2015.
Conference of Photovoltaic Science and Engineering and Exhibition. 11.2% efficient solution processed kesterite solar cell with a low voltage deficit. Adv.
Fthenakis, V., Wang, W., Kim, H.C., 2009. Life cycle inventory analysis of the production Energy Mater. 5 (18), 1500712.
of metals used in photovoltaics. Renew. Sustain. Energy Rev. 13 (3), 493–517. Hages, C.J., Carter, N.J., Agrawal, R., 2016a. Generalized quantum efficiency analysis for
Fuertes Marrón, D., Glatzel, T., Meeder, A., Schedel-Niedrig, T., Sadewasser, S., Lux- non-ideal solar cells: Case of Cu2ZnSnSe4. J. Appl. Phys. 119 (1), 014505.
Steiner, M.C., 2004. Electronic structure of secondary phases in Cu-rich Cu Ga Se 2 Hages, C.J., Koeper, M.J., Agrawal, R., 2016b. Optoelectronic and material properties of
solar cell devices. Appl. Phys. Lett. 85 (17), 3755–3757. nanocrystal-based CZTSe absorbers with Ag-alloying. Sol. Energy Mater. Sol. Cells
Gao, S., Jiang, Z., Wu, L., Ao, J., Zeng, Y., Sun, Y., Zhang, Y., 2018a. Interfaces of high- 145, 342–348.
efficiency kesterite Cu2ZnSnS (e) 4 thin film solar cells. Chin. Phys. B 27 (1), Hages, C.J., Levcenco, S., Miskin, C.K., Alsmeier, J.H., Abou-Ras, D., Wilks, R.G., Bär, M.,
018803. Unold, T., Agrawal, R., 2015. Improved performance of Ge-alloyed CZTGeSSe thin-
Gao, S., Zhang, Y., Ao, J., Lin, S., Zhang, Z., Li, X., Wang, D., Zhou, Z., Sun, G., Liu, F., film solar cells through control of elemental losses. Prog. Photovoltaics Res. Appl. 23
2018b. Tailoring Mo (S, Se) 2 structure for high efficient Cu2ZnSn (S, Se) 4 solar (3), 376–384.
cells. Sol. Energy Mater. Sol. Cells 176, 302–309. Haight, R., Barkhouse, A., Gunawan, O., Shin, B., Copel, M., Hopstaken, M., Mitzi, D.B.,
Geissbühler, J., Werner, J., Martin de Nicolas, S., Barraud, L., Hessler-Wyser, A., 2011. Band alignment at the Cu 2 ZnSn (S x Se 1–x) 4/CdS interface. Appl. Phys.
Despeisse, M., Nicolay, S., Tomasi, A., Niesen, B., De Wolf, S., 2015. 22.5% efficient Lett. 98 (25), 253502.
silicon heterojunction solar cell with molybdenum oxide hole collector. Appl. Phys. He, J., Sun, L., Ding, N., Kong, H., Zuo, S., Chen, S., Chen, Y., Yang, P., Chu, J., 2012.
Lett. 107 (8), 081601. Single-step preparation and characterization of Cu2ZnSn (SxSe1− x) 4 thin films
Gershon, T., Bishop, D., McCandless, B., Wang, W., Haight, R., 2015. Capturing the effect deposited by pulsed laser deposition method. J. Alloy. Compd. 529, 34–37.
of long low-temperature anneals on the sub-bandgap defect structure of CZTSSe, He, J., Sun, L., Zhang, K., Wang, W., Jiang, J., Chen, Y., Yang, P., Chu, J., 2013. Effect of
Carbon Nanotubes, Graphene, and Emerging 2D Materials for Electronic and post-sulfurization on the composition, structure and optical properties of Cu2ZnSnS4
Photonic Devices VIII. International Society for Optics and Photonics, pp. 955209. thin films deposited by sputtering from a single quaternary target. Appl. Surf. Sci.
Gershon, T., Lee, Y.S., Antunez, P., Mankad, R., Singh, S., Bishop, D., Gunawan, O., 264, 133–138.
Hopstaken, M., Haight, R., 2016a. Photovoltaic materials and devices based on the Hempel, H., Eichberger, R., Repins, I., Unold, T., 2018a. The effect of Cu-Zn disorder on
alloyed kesterite absorber (AgxCu1–x) 2ZnSnSe4. Adv. Energy Mater. 6 (10), charge carrier mobility and lifetime in Cu2ZnSnSe4. Thin Solid Films 666, 40–43.
1502468. Hempel, H., Hages, C.J., Eichberger, R., Repins, I., Unold, T., 2018b. Minority and
Gershon, T., Sardashti, K., Gunawan, O., Mankad, R., Singh, S., Lee, Y.S., Ott, J.A., Majority Charge Carrier Mobility in Cu 2 ZnSnSe 4 revealed by Terahertz
Kummel, A., Haight, R., 2016b. Photovoltaic device with over 5% efficiency based Spectroscopy. Sci. Rep. 8 (1), 1–9.
on an n-type Ag2ZnSnSe4 absorber. Adv. Energy Mater. 6 (22), 1601182. Hempel, H., Redinger, A., Repins, I., Moisan, C., Larramona, G., Dennler, G.,
Giraldo, S., Neuschitzer, M., Thersleff, T., López-Marino, S., Sánchez, Y., Xie, H., Handwerg, M., Fischer, S.F., Eichberger, R., Unold, T., 2016. Intragrain charge
Colina, M., Placidi, M., Pistor, P., Izquierdo-Roca, V., 2015. Large efficiency transport in kesterite thin films—Limits arising from carrier localization. J. Appl.
improvement in Cu2ZnSnSe4 solar cells by introducing a superficial Ge nanolayer. Phys. 120 (17), 175302.
Adv. Energy Mater. 5 (21), 1501070. Hempel, H., Unold, T., Eichberger, R., 2017. Measurement of charge carrier mobilities in
Giraldo, S., Saucedo, E., Neuschitzer, M., Oliva, F., Placidi, M., Alcobé, X., Izquierdo- thin films on metal substrates by reflection time resolved terahertz spectroscopy.
Roca, V., Kim, S., Tampo, H., Shibata, H., 2018. How small amounts of Ge modify the Opt. Express 25 (15), 17227–17236.
formation pathways and crystallization of kesterites. Energy Environ. Sci. 11 (3), Herz, K., Eicke, A., Kessler, F., Wächter, R., Powalla, M., 2003. Diffusion barriers for CIGS
582–593. solar cells on metallic substrates. Thin Solid Films 431, 392–397.
Gokmen, T., Gunawan, O., Mitzi, D.B., 2013a. Minority carrier diffusion length Hiroi, H., Sakai, N., Kato, T., Sugimoto, H., 2013. High voltage Cu 2 ZnSnS 4 submodules
extraction in Cu2ZnSn (Se, S) 4 solar cells. J. Appl. Phys. 114 (11), 114511. by hybrid buffer layer, 2013 IEEE 39th Photovoltaic Specialists Conference (PVSC).
Gokmen, T., Gunawan, O., Todorov, T.K., Mitzi, D.B., 2013b. Band tailing and efficiency IEEE 0863–0866.
limitation in kesterite solar cells. Appl. Phys. Lett. 103 (10), 103506. Hiroi, H., Sakai, N., Muraoka, S., Katou, T., Sugimoto, H., 2012. Development of high
Goodman, C., 1958. The prediction of semiconducting properties in inorganic efficiency Cu 2 ZnSnS 4 submodule with Cd-free buffer layer, 2012 38th IEEE
compounds. J. Phys. Chem. Solids 6 (4), 305–314. Photovoltaic Specialists Conference. IEEE 001811–001814.
Graetzel, M., Janssen, R.A., Mitzi, D.B., Sargent, E.H., 2012. Materials interface Hironiwa, D., Chantana, J., Sakai, N., Kato, T., Sugimoto, H., Minemoto, T., 2015a.
engineering for solution-processed photovoltaics. Nature 488 (7411), 304–312. Application of multi-buffer layer of (Zn, Mg) O/CdS in Cu2ZnSn (S, Se) 4 solar cells.
Granath, K., Bodegård, M., Stolt, L., 2000. The effect of NaF on Cu (In, Ga) Se2 thin film Curr. Appl Phys. 15 (3), 383–388.
solar cells. Sol. Energy Mater. Sol. Cells 60 (3), 279–293. Hironiwa, D., Matsuo, N., Chantana, J., Sakai, N., Kato, T., Sugimoto, H., Minemoto, T.,
Green, M.A., Dunlop, E.D., Levi, D.H., Hohl-Ebinger, J., Yoshita, M., Ho-Baillie, A.W.Y., 2015b. Annealing effect on Cu2ZnSn (S, Se) 4 solar cell with Zn1–xMgxO buffer
2019. Solar cell efficiency tables (version 54). Prog. Photovoltaics Res. Appl. 27 (7), layer. Phys. Stat. Sol. (a) 212 (12), 2766–2771.
565–575. Hironiwa, D., Matsuo, N., Sakai, N., Katou, T., Sugimoto, H., Chantana, J., Tang, Z.,
Grenet, L., Grondin, P., Coumert, K., Karst, N., Emieux, F., Roux, F., Fillon, R., Minemoto, T., 2014. Sputtered (Zn, Mg) O buffer layer for band offset control in
Altamura, G., Fournier, H., Faucherand, P., 2014. Experimental evidence of light Cu2ZnSn (S, Se) 4 solar cells. Jpn. J. Appl. Phys. 53 (10), 106502.
soaking effect in Cd-free Cu2ZnSn (S, Se) 4-based solar cells. Thin Solid Films 564, Hönes, K., Eickenberg, M., Siebentritt, S., Persson, C., 2008. Polarization of defect related
375–378. optical transitions in chalcopyrites. Appl. Phys. Lett. 93 (9), 092102.
Grossberg, M., Krustok, J., Raudoja, J., Raadik, T., 2012. The role of structural properties Hou, Y., Chen, W., Baran, D., Stubhan, T., Luechinger, N.A., Hartmeier, B., Richter, M.,
on deep defect states in Cu2ZnSnS4 studied by photoluminescence spectroscopy. Min, J., Chen, S., Quiroz, C.O.R., 2016. Overcoming the interface losses in planar
Appl. Phys. Lett. 101 (10), 102102. heterojunction perovskite-based solar cells. Adv. Mater. 28 (25), 5112–5120.
Grossberg, Maarja, Krustok, Jüri, Hages, Charles J., Bishop, Douglas M., Gunawan, Oki, Hsieh, T.-P., Chuang, C.-C., Wu, C.-S., Chang, J.-C., Guo, J.-W., Chen, W.-C., 2011. Effects
Scheer, Roland, Lyam, Samantha M., Hempel, Hannes, Levcenco, Sergiu, Unold, of residual copper selenide on CuInGaSe2 solar cells. Solid-State Electron. 56 (1),
Thomas, 2019. The electrical and optical properties of kesterites. J. Phys.: Energy 1 175–178.
(no. 4). Hsieh, Y.T., Han, Q., Jiang, C., Song, T.B., Chen, H., Meng, L., Zhou, H., Yang, Y., 2016.
Grossberg, M., Krustok, J., Timmo, K., Altosaar, M., 2009. Radiative recombination in Efficiency enhancement of Cu2ZnSn (S, Se) 4 solar cells via alkali metals doping.
Cu2ZnSnSe4 monograins studied by photoluminescence spectroscopy. Thin Solid Adv. Energy Mater. 6 (7), 1502386.
Films 517 (7), 2489–2492. Hsu, W.-C., Repins, I., Beall, C., DeHart, C., Teeter, G., To, B., Yang, Y., Noufi, R., 2013.
Guglietta, G.W., Choudhury, K.R., Caspar, J.V., Baxter, J.B., 2014. Employing time- The effect of Zn excess on kesterite solar cells. Sol. Energy Mater. Sol. Cells 113,
resolved terahertz spectroscopy to analyze carrier dynamics in thin-film Cu2ZnSn (S, 160–164.
Se) 4 absorber layers. Appl. Phys. Lett. 104 (25), 253901. Hsu, W.-W., Chen, J., Cheng, T.-H., Lu, S., Ho, W.-S., Chen, Y.-Y., Chien, Y.-J., Liu, C.,
Gunawan, O., Gokmen, T., Mitzi, D.B., 2014. Suns-VOC characteristics of high 2012. Surface passivation of Cu (In, Ga) Se2 using atomic layer deposited Al2O3.
performance kesterite solar cells. J. Appl. Phys. 116 (8), 084504. Appl. Phys. Lett. 100 (2), 023508.
Gunawan, O., Gokmen, T., Warren, C.W., Cohen, J.D., Todorov, T.K., Barkhouse, D.A.R., Hsu, W., Sutter-Fella, C.M., Hettick, M., Cheng, L., Chan, S., Chen, Y., Zeng, Y.,
Bag, S., Tang, J., Shin, B., Mitzi, D.B., 2012. Electronic properties of the Cu2ZnSn Zheng, M., Wang, H.-P., Chiang, C.-C., 2015. Electron-Selective TiO 2 Contact for Cu
(Se, S) 4 absorber layer in solar cells as revealed by admittance spectroscopy and (In, Ga) Se 2 Solar Cells. Sci. Rep. 5 (1), 1–7.
related methods. Appl. Phys. Lett. 100 (25), 253905. Huang, T.J., Yin, X., Qi, G., Gong, H., 2014. CZTS-based materials and interfaces and
Gunawan, O., Pae, S.R., Bishop, D.M., Lee, Y.S., Virgus, Y., Jeon, N.J., Noh, J.H., Shao, their effects on the performance of thin film solar cells. Phys. Stat. Sol. (RRL)–Rap.
X., Todorov, T., Mitzi, D.B., 2018. Carrier-resolved photo hall measurement in Res. Lett. 8 (09), 735–762.
world-record-quality perovskite and kesterite solar absorbers. arXiv preprint arXiv: Ishizuka, S., Yamada, A., Islam, M.M., Shibata, H., Fons, P., Sakurai, T., Akimoto, K.,
1802.07910. Niki, S., 2009. Na-induced variations in the structural, optical, and electrical
Gunawan, O., Todorov, T.K., Mitzi, D.B., 2010. Loss mechanisms in hydrazine-processed properties of Cu (In, Ga) Se 2 thin films. J. Appl. Phys. 106 (3), 034908.
Cu 2 ZnSn (Se, S) 4 solar cells. Appl. Phys. Lett. 97 (23), 233506. Islam, M.M., Halim, M.A., Sakurai, T., Sakai, N., Kato, T., Sugimoto, H., Tampo, H.,
Gunawan, O., Virgus, Y., Tai, K.F., 2015. A parallel dipole line system. Appl. Phys. Lett. Shibata, H., Niki, S., Akimoto, K., 2015. Determination of deep-level defects in
106 (6), 062407.

53
M. Sahu et al. Solar Energy 230 (2021) 13–58

Cu2ZnSn (S, Se) 4 thin-films using photocapacitance method. Appl. Phys. Lett. 106 Kim, S., Park, J.-S., Walsh, A., 2018a. Identification of killer defects in kesterite thin-film
(24), 243905. solar cells. ACS Energy Lett. 3 (2), 496–500.
Ito, K., 2014. Copper zinc tin sulfide-based thin film solar cells. Wiley Online Library. Kim, S., Rana, T.R., Kim, J., Son, D.-H., Yang, K.-J., Kang, J.-K., Kim, D.-H., 2018b.
Ito, K., Nakazawa, T., 1988. Electrical and optical properties of stannite-type quaternary Limiting effects of conduction band offset and defect states on high efficiency
semiconductor thin films. Jpn. J. Appl. Phys. 27 (11R), 2094. CZTSSe solar cell. Nano Energy 45, 75–83.
Jackson, P., Hariskos, D., Lotter, E., Paetel, S., Wuerz, R., Menner, R., Wischmann, W., Kohara, N., Nishiwaki, S., Hashimoto, Y., Negami, T., Wada, T., 2001. Electrical
Powalla, M., 2011. New world record efficiency for Cu (In, Ga) Se2 thin-film solar properties of the cu (in, ga) se2/MoSe2/mo structure. Sol. Energy Mater. Sol. Cells
cells beyond 20%. Prog. Photovoltaics Res. Appl. 19 (7), 894–897. 67 (1–4), 209–215.
Jackson, P., Hariskos, D., Wuerz, R., Kiowski, O., Bauer, A., Friedlmeier, T.M., Krämmer, C., Huber, C., Zimmermann, C., Lang, M., Schnabel, T., Abzieher, T.,
Powalla, M., 2015. Properties of Cu (In, Ga) Se2 solar cells with new record Ahlswede, E., Kalt, H., Hetterich, M., 2014. Reversible order-disorder related band
efficiencies up to 21.7%. Phys. Stat. Sol. (RRL)–Rap. Res. Lett. 9 (1), 28–31. gap changes in Cu2ZnSn (S, Se) 4 via post-annealing of solar cells measured by
Janotti, A., Van de Walle, C.G., 2009. Fundamentals of zinc oxide as a semiconductor. electroreflectance. Appl. Phys. Lett. 105 (26), 262104.
Rep. Prog. Phys. 72 (12), 126501. Kronik, L., Cahen, D., Schock, H.W., 1998. Effects of sodium on polycrystalline Cu (In,
Jiang, M., Yan, X., 2013. Cu2ZnSnS4 thin film solar cells: present status and future Ga) Se2 and its solar cell performance. Adv. Mater. 10 (1), 31–36.
prospects. In: Solar Cells—Research and Application Perspectives. Krustok, J., Collan, H., Hjelt, K., 1997. Does the low-temperature Arrhenius plot of the
JH, S.J.W., 1975. Ternary Chalcopyrite Semiconductors: Growth, Electronic Properties, photoluminescence intensity in CdTe point towards an erroneous activation energy?
and Application. Pergamon Press, Oxford. J. Appl. Phys. 81 (3), 1442–1445.
Johnson, M., Baryshev, S., Thimsen, E., Manno, M., Zhang, X., Veryovkin, I., Kumar, M., Dubey, A., Adhikari, N., Venkatesan, S., Qiao, Q., 2015a. Strategic review of
Leighton, C., Aydil, E., 2014. Alkali-metal-enhanced grain growth in Cu 2 ZnSnS 4 secondary phases, defects and defect-complexes in kesterite CZTS–Se solar cells.
thin films. Energy Environ. Sci. 7 (6), 1931–1938. Energy Environ. Sci. 8 (11), 3134–3159.
Just, J., Lützenkirchen-Hecht, D., Frahm, R., Schorr, S., Unold, T., 2011. Determination Kumar, M., Dubey, A., Reza, K.M., Adhikari, N., Qiao, Q., Bommisetty, V., 2015b. Origin
of secondary phases in kesterite Cu2ZnSnS4 thin films by x-ray absorption near edge of photogenerated carrier recombination at the metal–active layer interface in
structure analysis. Appl. Phys. Lett. 99 (26), 262105. polymer solar cells. PCCP 17 (41), 27690–27697.
Just, J., Nichterwitz, M., Lützenkirchen-Hecht, D., Frahm, R., Unold, T., 2016a. Kumar, M., Persson, C., 2013. Cu2ZnSnS4 and Cu2ZnSnSe4 as potential earth-abundant
Compositional dependence of charge carrier transport in kesterite Cu2ZnSnS4 solar thin-film absorber materials: a density functional theory study. Int. J. Theor. Appl.
cells. J. Appl. Phys. 120 (22), 225703. Sci 5 (1), 1–8.
Just, J., Sutter-Fella, C.M., Lützenkirchen-Hecht, D., Frahm, R., Schorr, S., Unold, T., Kumar, M.S., Madhusudanan, S.P., Batabyal, S.K., 2018. Substitution of Zn in Earth-
2016b. Secondary phases and their influence on the composition of the kesterite Abundant Cu2ZnSn (S, Se) 4 based thin film solar cells–A status review. Sol. Energy
phase in CZTS and CZTSe thin films. PCCP 18 (23), 15988–15994. Mater. Sol. Cells 185, 287–299.
Kamada, R., Yagioka, T., Adachi, S., Handa, A., Tai, K.F., Kato, T., Sugimoto, H., 2016. Lang, D., 1974. Deep-level transient spectroscopy: A new method to characterize traps in
New world record Cu (In, Ga)(Se, S) 2 thin film solar cell efficiency beyond 22%, semiconductors. J. Appl. Phys. 45 (7), 3023–3032.
2016 IEEE 43rd Photovoltaic Specialists Conference (PVSC). IEEE 1287–1291. Lany, S., Zunger, A., 2005. Anion vacancies as a source of persistent photoconductivity in
Kamoun, N., Bouzouita, H., Rezig, B., 2007. Fabrication and characterization of II-VI and chalcopyrite semiconductors. Phys. Rev. B 72 (3), 035215.
Cu2ZnSnS4 thin films deposited by spray pyrolysis technique. Thin Solid Films 515 Larramona, G., Levcenko, S., Bourdais, S., Jacob, A., Choné, C., Delatouche, B.,
(15), 5949–5952. Moisan, C., Just, J., Unold, T., Dennler, G., 2015. Fine-tuning the Sn content in
Kask, E., Grossberg, M., Josepson, R., Salu, P., Timmo, K., Krustok, J., 2013. Defect CZTSSe thin films to achieve 10.8% solar cell efficiency from spray-deposited
studies in Cu2ZnSnSe4 and Cu2ZnSn (Se0. 75S0. 25) 4 by admittance and water–ethanol-based colloidal inks. Adv. Energy Mater. 5 (24), 1501404.
photoluminescence spectroscopy. Mater. Sci. Semicond. Process. 16 (3), 992–996. Lee, J., Giles, N., Rajavel, D., Summers, C., 1995. Donor-acceptor pair luminescence
Katagiri, H., Jimbo, K., Maw, W.S., Oishi, K., Yamazaki, M., Araki, H., Takeuchi, A., involving the iodine A center in CdTe. J. Appl. Phys. 78 (9), 5669–5674.
2009. Development of CZTS-based thin film solar cells. Thin Solid Films 517 (7), Lee, Y.S., Gershon, T., Gunawan, O., Todorov, T.K., Gokmen, T., Virgus, Y., Guha, S.,
2455–2460. 2015. Cu2ZnSnSe4 thin-film solar cells by thermal co-evaporation with 11.6%
Katagiri, H., Jimbo, K., Yamada, S., Kamimura, T., Maw, W.S., Fukano, T., Ito, T., efficiency and improved minority carrier diffusion length. Adv. Energy Mater. 5 (7),
Motohiro, T., 2008. Enhanced conversion efficiencies of Cu2ZnSnS4-based thin film 1401372.
solar cells by using preferential etching technique. Appl. Phys Express 1 (4), 041201. Lee, Y.S., Gershon, T., Todorov, T.K., Wang, W., Winkler, M.T., Hopstaken, M.,
Katagiri, H., Saitoh, K., Washio, T., Shinohara, H., Kurumadani, T., Miyajima, S., 2001. Gunawan, O., Kim, J., 2016. Atomic layer deposited aluminum oxide for interface
Development of thin film solar cell based on Cu2ZnSnS4 thin films. Sol. Energy passivation of Cu2ZnSn (S, Se) 4 thin-film solar cells. Adv. Energy Mater. 6 (12),
Mater. Sol. Cells 65 (1–4), 141–148. 1600198.
Kato, T., Hiroi, H., Sakai, N., Muraoka, S., Sugimoto, H., 2012. Eur. Photovoltaic Sol. Leitao, J., Santos, N.M., Fernandes, P., Salomé, P., Da Cunha, A., González, J.,
Energy Conf., Proc. Int. Conf. 2236–2239. Ribeiro, G., Matinaga, F., 2011. Photoluminescence and electrical study of
Kato, T., Hiroi, H., Sakai, N., Sugimoto, H., 2013. Buffer/absorber interface study on fluctuating potentials in Cu 2 ZnSnS 4-based thin films. Phys. Rev. B 84 (2), 024120.
Cu2ZnSnS4 and Cu2ZnSnSe4 based solar cells: band alignment and its impact on the Levcenko, S., Just, J., Redinger, A., Larramona, G., Bourdais, S., Dennler, G., Jacob, A.,
solar cell performance, 28th European Photovoltaic Solar Energy Conference and Unold, T., 2016. Deep defects in Cu 2 ZnSn (S, Se) 4 solar cells with varying Se
Exhibition. EU PVSEC 2125–2127. content. Phys. Rev. Appl 5 (2), 024004.
Kaur, K., Kumar, N., Kumar, M., 2017. Strategic review of interface carrier Li, J., Du, Q., Liu, W., Jiang, G., Feng, X., Zhang, W., Zhu, J., Zhu, C., 2012. The band
recombination in earth abundant Cu–Zn–Sn–S–Se solar cells: current challenges and offset at CdS/Cu 2 ZnSnS 4 heterojunction interface. Electron. Mater. Lett. 8 (4),
future prospects. J. Mater. Chem. A 5 (7), 3069–3090. 365–367.
Khadka, D.B., Kim, S., Kim, J., 2015. A nonvacuum approach for fabrication of Li, J., Liu, X., Liu, W., Wu, L., Ge, B., Lin, S., Gao, S., Zhou, Z., Liu, F., Sun, Y., 2017.
Cu2ZnSnSe4/In2S3 thin film solar cell and optoelectronic characterization. J. Phys. Restraining the Band Fluctuation of CBD-Zn (O, S) Layer: Modifying the Hetero-
ChemC 119 (22), 12226–12235. Junction Interface for High Performance Cu2ZnSnSe4 Solar Cells With Cd-Free
Kim, G.Y., Kim, J.R., Jo, W., Lee, K.D., Kim, J.Y., Nguyen, T.T.T., Yoon, S., 2014a. Effects Buffer Layer. Solar RRL 1 (10), 1700075.
of Cu2− xS phase removal on surface potential of Cu2ZnSnS4 thin-films grown by Li, J., Wei, M., Du, Q., Liu, W., Jiang, G., Zhu, C., 2013a. The band alignment at CdS/
electroplating. Curr. Appl Phys. 14 (12), 1665–1668. Cu2ZnSnSe4 heterojunction interface. Surf. Interface Anal. 45 (2), 682–684.
Kim, J., Hiroi, H., Todorov, T.K., Gunawan, O., Kuwahara, M., Gokmen, T., Nair, D., Li, J., Yuan, Z.-K., Chen, S., Gong, X.-G., Wei, S.-H., 2019. Effective and noneffective
Hopstaken, M., Shin, B., Lee, Y.S., 2014b. High efficiency Cu2ZnSn (S, Se) 4 solar recombination center defects in Cu2ZnSnS4: Significant difference in carrier capture
cells by applying a double In2S3/CdS emitter. Adv. Mater. 26 (44), 7427–7431. cross sections. Chem. Mater. 31 (3), 826–833.
Kim, J., Kim, G.Y., Jo, W., Yang, K.-J., Sim, J.-H., Kim, D.-H., Kang, J.-K., 2016a. Effects Li, J.V., Kuciauskas, D., Young, M.R., Repins, I.L., 2013b. Effects of sodium incorporation
of Mo back-contact annealing on surface potential and carrier transport in Cu 2 in Co-evaporated Cu2ZnSnSe4 thin-film solar cells. Appl. Phys. Lett. 102 (16),
ZnSnS 4 thin film solar cells. RSC Adv. 6 (105), 103337–103345. 163905.
Kim, J., Park, S., Ryu, S., Oh, J., Shin, B., 2017. Improving the open-circuit voltage of Li, W., Chen, J., Cui, H., Liu, F., Hao, X., 2014. Inhibiting MoS2 formation by introducing
Cu2ZnSnSe4 thin film solar cells via interface passivation. Prog. Photovoltaics Res. a ZnO intermediate layer for Cu2ZnSnS4 solar cells. Mater. Lett. 130, 87–90.
Appl. 25 (4), 308–317. Li, Z.-H., Cho, E.-S., Kwon, S.J., 2011. Molybdenum thin film deposited by in-line DC
Kim, J., Shin, B., 2017. Strategies to reduce the open-circuit voltage deficit in Cu 2 ZnSn magnetron sputtering as a back contact for Cu (In, Ga) Se2 solar cells. Appl. Surf. Sci.
(S, Se) 4 thin film solar cells. Electron. Mater. Lett. 13 (5), 373–392. 257 (22), 9682–9688.
Kim, S.-Y., Son, D.-H., Kim, Y.-I., Kim, S.-H., Kim, S., Ahn, K., Sung, S.-J., Hwang, D.-K., Liao, D., Rockett, A., 2003a. Cd doping at the CuInSe 2/CdS heterojunction. J. Appl.
Yang, K.-J., Kang, J.-K., 2019a. Void and secondary phase formation mechanisms of Phys. 93 (11), 9380–9382.
CZTSSe using Sn/Cu/Zn/Mo stacked elemental precursors. Nano Energy 59, Liao, D., Rockett, A., 2003b. Cu depletion at the CuInSe 2 surface. Appl. Phys. Lett. 82
399–411. (17), 2829–2831.
Kim, S., Kim, K.M., Tampo, H., Shibata, H., Matsubara, K., Niki, S., 2016b. Ge- Liu, F., Huang, J., Sun, K., Yan, C., Shen, Y., Park, J., Pu, A., Zhou, F., Liu, X., Stride, J.A.,
incorporated Cu2ZnSnSe4 thin-film solar cells with efficiency greater than 10%. Sol. 2017. Beyond 8% ultrathin kesterite Cu 2 ZnSnS 4 solar cells by interface reaction
Energy Mater. Sol. Cells 144, 488–492. route controlling and self-organized nanopattern at the back contact. NPG Asia
Kim, S., Kim, K.M., Tampo, H., Shibata, H., Niki, S., 2016c. Improvement of voltage Mater. 9 (7), e401.
deficit of Ge-incorporated kesterite solar cell with 12.3% conversion efficiency. Appl. Liu, F., Sun, K., Li, W., Yan, C., Cui, H., Jiang, L., Hao, X., Green, M.A., 2014a. Enhancing
Phys Express 9 (10), 102301. the Cu2ZnSnS4 solar cell efficiency by back contact modification: Inserting a thin
Kim, S., Park, J.-S., Hood, S.N., Walsh, A., 2019b. Lone-pair effect on carrier capture in TiB2 intermediate layer at Cu2ZnSnS4/Mo interface. Appl. Phys. Lett. 104 (5),
Cu 2 ZnSnS 4 solar cells. J. Mater. Chem. A 7 (6), 2686–2693. 051105.

54
M. Sahu et al. Solar Energy 230 (2021) 13–58

Liu, F., Wu, S., Zhang, Y., Hao, X., Ding, L., 2020. Advances in kesterite Cu2ZnSn (S, Se) 4 Nagaoka, A., Miyake, H., Taniyama, T., Kakimoto, K., Nose, Y., Scarpulla, M.A.,
solar cells. Science Bulletin 65 (9), 698–701. Yoshino, K., 2014a. Effects of sodium on electrical properties in Cu2ZnSnS4 single
Liu, F., Yan, C., Huang, J., Sun, K., Zhou, F., Stride, J.A., Green, M.A., Hao, X., 2016. crystal. Appl. Phys. Lett. 104 (15), 152101.
Nanoscale microstructure and chemistry of Cu2ZnSnS4/CdS interface in kesterite Nagaoka, A., Miyake, H., Taniyama, T., Kakimoto, K., Yoshino, K., 2013. Correlation
Cu2ZnSnS4 solar cells. Adv. Energy Mater. 6 (15), 1600706. between intrinsic defects and electrical properties in the high-quality Cu2ZnSnS4
Liu, X., Cui, H., Li, W., Song, N., Liu, F., Conibeer, G., Hao, X., 2014b. Improving single crystal. Appl. Phys. Lett. 103 (11), 112107.
Cu2ZnSnS4 (CZTS) solar cell performance by an ultrathin ZnO intermediate layer Nagaoka, A., Yoshino, K., Taniguchi, H., Taniyama, T., Kakimoto, K., Miyake, H., 2014b.
between CZTS absorber and Mo back contact. Phys. Stat. Sol. (RRL)–Rap. Res. Lett. 8 Growth and characterization of Cu2ZnSn (SxSe1− x) 4 alloys grown by the melting
(12), 966–970. method. J. Cryst. Growth 386, 204–207.
López-Marino, S., Espíndola-Rodríguez, M., Sánchez, Y., Alcobé, X., Oliva, F., Xie, H., Nagoya, A., Asahi, R., Kresse, G., 2011. First-principles study of Cu2ZnSnS4 and the
Neuschitzer, M., Giraldo, S., Placidi, M., Caballero, R., 2016. The importance of back related band offsets for photovoltaic applications. J. Phys.: Condens. Matter 23 (40),
contact modification in Cu2ZnSnSe4 solar cells: the role of a thin MoO2 layer. Nano 404203.
Energy 26, 708–721. Nagoya, A., Asahi, R., Wahl, R., Kresse, G., 2010. Defect formation and phase stability of
López-Marino, S., Placidi, M., Pérez-Tomás, A., Llobet, J., Izquierdo-Roca, V., Cu 2 ZnSnS 4 photovoltaic material. Phys. Rev. B 81 (11), 113202.
Fontané, X., Fairbrother, A., Espíndola-Rodríguez, M., Sylla, D., Pérez-Rodríguez, A., Nakamura, M., Yamaguchi, K., Kimoto, Y., Yasaki, Y., Kato, T., Sugimoto, H., 2019. Cd-
2013a. Inhibiting the absorber/Mo-back contact decomposition reaction in Cu 2 free Cu (In, Ga)(Se, S) 2 thin-film solar cell with record efficiency of 23.35%. IEEE J.
ZnSnSe 4 solar cells: the role of a ZnO intermediate nanolayer. J. Mater. Chem. A 1 Photovoltaics 9 (6), 1863–1867.
(29), 8338–8343. Nam, D., Cho, S., Sim, J.-H., Yang, K.-J., Son, D.-H., Kim, D.-H., Kang, J.-K., Kwon, M.-S.,
López-Marino, S., Sánchez, Y., Placidi, M., Fairbrother, A., Espindola-Rodríguez, M., Jeon, C.-W., Cheong, H., 2016. Solar conversion efficiency and distribution of ZnS
Fontané, X., Izquierdo-Roca, V., López-García, J., Calvo-Barrio, L., Pérez- secondary phase in Cu2ZnSnS4 solar cells. Sol. Energy Mater. Sol. Cells 149,
Rodríguez, A., 2013b. ZnSe etching of Zn-rich Cu2ZnSnSe4: an oxidation route for 226–231.
improved solar-cell efficiency. Chemistry–A. European Journal 19 (44), Neisser, A., Hengel, I., Klenk, R., Matthes, T.W., Alvarez-Garcıa, J., Pérez-Rodrıguez, A.,
14814–14822. Romano-Rodrıguez, A., Lux-Steiner, M.-C., 2001. Effect of Ga incorporation in
Luckert, F., Hamilton, D.I., Yakushev, M., Beattie, N., Zoppi, G., Moynihan, M., Forbes, I., sequentially prepared CuInS2 thin film absorbers. Sol. Energy Mater. Sol. Cells 67
Karotki, A., Mudryi, A., Grossberg, M., 2011. Optical properties of high quality (1–4), 97–104.
Cu2ZnSnSe4 thin films. Appl. Phys. Lett. 99 (6), 062104. Neuschitzer, M., Lienau, K., Guc, M., Barrio, L.C., Haass, S., Prieto, J.M., Sanchez, Y.,
Luque, A., Hegedus, S., 2003. Photovoltaic science and engineering. Wiley Online Espindola-Rodriguez, M., Romanyuk, Y., Perez-Rodriguez, A., 2016a. Towards high
Library. performance Cd-free CZTSe solar cells with a ZnS (O, OH) buffer layer: the influence
Maeda, K., Tanaka, K., Fukui, Y., Uchiki, H., 2011a. Influence of H2S concentration on of thiourea concentration on chemical bath deposition. J. Phys. D Appl. Phys. 49
the properties of Cu2ZnSnS4 thin films and solar cells prepared by sol–gel (12), 125602.
sulfurization. Sol. Energy Mater. Sol. Cells 95 (10), 2855–2860. Neuschitzer, M., Marquez, J., Giraldo, S., Dimitrievska, M., Placidi, M., Forbes, I.,
Maeda, T., Nakamura, S., Wada, T., 2011b. First principles calculations of defect Izquierdo-Roca, V., Pérez-Rodriguez, A., Saucedo, E., 2016b. V OC Boosting and
formation in in-free photovoltaic semiconductors Cu2ZnSnS4 and Cu2ZnSnSe4. Jpn. Grain Growth Enhancing Ge-Doping Strategy for Cu2ZnSnSe4 Photovoltaic
J. Appl. Phys. 50 (4S), 04DP07. Absorbers. J. Phys. ChemC 120 (18), 9661–9670.
Maeda, T., Nakamura, S., Wada, T., 2012. First-principles study on Cd doping in Nguyen, T.T.T., Shin, H.-Y., Kim, G.Y., Kim, J.R., Jo, W., Yoon, S., Lee, K.D., Kim, J.Y.,
Cu2ZnSnS4 and Cu2ZnSnSe4. Jpn. J. Appl. Phys. 51 (10S), 10NC11. 2015. Raman scattering studies of Cu 2 ZnSnS 4 thin films: Local distribution of the
Martın, I., Vetter, M., Orpella, A., Puigdollers, J., Cuevas, A., Alcubilla, R., 2001. Surface secondary phase Cu 2–x S and the effect of KCN etching on Cu 2–x S. J. Korean Phys.
passivation of p-type crystalline Si by plasma enhanced chemical vapor deposited Soc. 66 (1), 117–122.
amorphous SiC x: H films. Appl. Phys. Lett. 79 (14), 2199–2201. Nina, K., Kimura, Y., Yokoyama, K., Kido, O., Binyo, G., Kaito, C., 2008. Direct
Matsushita, H., Maeda, T., Katsui, A., Takizawa, T., 2000. Thermal analysis and synthesis observation of MoO2 crystal growth from amorphous MoO3 film. Physica E 40 (9),
from the melts of Cu-based quaternary compounds Cu–III–IV–VI4 and Cu2–II–IV–VI4 2995–2998.
(II= Zn, Cd; III= Ga, In; IV= Ge, Sn; VI= Se). J. Cryst. Growth 208 (1–4), 416–422. NREL. Best Research-Cell Efficiencies. accessed 6 November, 2019. https://www.nrel.
Meril, M., Sudha Kartha, C., 2009. Engineering the properties of indium sulfide for thin gov/pv/cell-efficiency.html.
film solar cells by doping. Cochin University of Science & Technology. Okano, M., Phuong, L.Q., Kanemitsu, Y., 2015. Photocarrier dynamics in CIGS, CZTS,
Messaoud, K.B., Buffière, M., Brammertz, G., Lenaers, N., Boyen, H.-G., Sahayaraj, S., and related materials revealed by ultrafast optical spectroscopy. Phys. Stat. Sol. (b)
Meuris, M., Amlouk, M., Poortmans, J., 2017. Synthesis and characterization of (Cd, 252 (6), 1219–1224.
Zn) S buffer layer for Cu2ZnSnSe4 solar cells. J. Phys. D Appl. Phys. 50 (28), 285501. Orgassa, K., Schock, H.W., Werner, J., 2003. Alternative back contact materials for thin
Meyer, B., Polity, A., Farangis, B., He, Y., Hasselkamp, D., Krämer, T., Wang, C., 2004. film Cu (In, Ga) Se2 solar cells. Thin Solid Films 431, 387–391.
Structural properties and bandgap bowing of ZnO 1–x S x thin films deposited by Paier, J., Asahi, R., Nagoya, A., Kresse, G., 2009. Cu 2 ZnSnS 4 as a potential photovoltaic
reactive sputtering. Appl. Phys. Lett. 85 (21), 4929–4931. material: a hybrid Hartree-Fock density functional theory study. Phys. Rev. B 79
Miller, D.W., Warren, C.W., Gunawan, O., Gokmen, T., Mitzi, D.B., Cohen, J.D., 2012. (11), 115126.
Electronically active defects in the Cu2ZnSn (Se, S) 4 alloys as revealed by transient Palsgaard, M.L., Crovetto, A., Gunst, T., Markussen, T., Hansen, O., Stokbro, K.,
photocapacitance spectroscopy. Appl. Phys. Lett. 101 (14), 142106. Brandbyge, M., 2016. Semiconductor band alignment from first principles: A new
Minemoto, T., Matsui, T., Takakura, H., Hamakawa, Y., Negami, T., Hashimoto, Y., nonequilibrium Green’s function method applied to the CZTSe/CdS interface for
Uenoyama, T., Kitagawa, M., 2001. Theoretical analysis of the effect of conduction photovoltaics. In: 2016 International Conference on Simulation of Semiconductor
band offset of window/CIS layers on performance of CIS solar cells using device Processes and Devices (SISPAD). IEEE, pp. 377–380.
simulation. Sol. Energy Mater. Sol. Cells 67 (1–4), 83–88. Pamplin, B., 1964. A systematic method of deriving new semiconducting compounds by
Mitzi, D.B., Gunawan, O., Todorov, T.K., Barkhouse, D.A.R., 2013. Prospects and structural analogy. J. Phys. Chem. Solids 25 (7), 675–684.
performance limitations for Cu–Zn–Sn–S–Se photovoltaic technology. Philosoph. Park, D., Nam, D., Jung, S., An, S., Gwak, J., Yoon, K., Yun, J.H., Cheong, H., 2011.
Trans. Roy. Soc. A: Math. Phys. Eng. Sci. 371 (1996), 20110432. Optical characterization of Cu2ZnSnSe4 grown by thermal co-evaporation. Thin
Mitzi, D.B., Gunawan, O., Todorov, T.K., Wang, K., Guha, S., 2011. The path towards a Solid Films 519 (21), 7386–7389.
high-performance solution-processed kesterite solar cell. Sol. Energy Mater. Sol. Persson, C., 2008. Anisotropic hole-mass tensor of CuIn 1–x Ga x (S, Se) 2: Presence of
Cells 95 (6), 1421–1436. free carriers narrows the energy gap. Appl. Phys. Lett. 93 (7), 072106.
Miyamoto, Y., Tanaka, K., Oonuki, M., Moritake, N., Uchiki, H., 2008. Optical properties Persson, C., 2010. Electronic and optical properties of Cu 2 ZnSnS 4 and Cu 2 ZnSnSe 4.
of Cu2ZnSnS4 thin films prepared by sol–gel and sulfurization method. Jpn. J. Appl. J. Appl. Phys. 107 (5), 053710.
Phys. 47 (1S), 596. Persson, C., Zunger, A., 2003. Anomalous Grain Boundary Physics in Polycrystalline C u I
Moore, J., Hages, C., Lundstrom, M., Agrawal, R., 2012. Influence of Ge doping on defect n S e 2: The Existence of a Hole Barrier. Phys. Rev. Lett. 91 (26), 266401.
distributions of Cu 2 Zn (Sn x Ge 1–x)(S y Se 1− y) fabricated by nanocrystal ink Pethe, S.A., Takahashi, E., Kaul, A., Dhere, N.G., 2012. Effect of sputtering process
deposition with selenization. In: 2012 38th IEEE Photovoltaic Specialists parameters on film properties of molybdenum back contact. Sol. Energy Mater. Sol.
Conference. IEEE, pp. 001475–001480. Cells 100, 1–5.
Moore, J.E., Hages, C.J., Agrawal, R., Lundstrom, M.S., Gray, J.L., 2016. The importance Pilvet, M., Kauk-Kuusik, M., Altosaar, M., Grossberg, M., Danilson, M., Timmo, K.,
of band tail recombination on current collection and open-circuit voltage in CZTSSe Mere, A., Mikli, V., 2015. Compositionally tunable structure and optical properties of
solar cells. Appl. Phys. Lett. 109 (2), 021102. Cu1. 85 (CdxZn1− x) 1.1 SnS4. 1 (0≤ x≤ 1) monograin powders. Thin Solid Films
Moriya, K., Watabe, J., Tanaka, K., Uchiki, H., 2006. Characterization of Cu2ZnSnS4 thin 582, 180–183.
films prepared by photo-chemical deposition. Phys. Stat. Sol. C 3 (8), 2848–2852. Pinto, A.H., Shin, S.W., Aydil, E.S., Penn, R.L., 2016. Selective removal of Cu 2–x (S, Se)
Mousel, M., Redinger, A., Djemour, R., Arasimowicz, M., Valle, N., Dale, P., phases from Cu 2 ZnSn (S, Se) 4 thin films. Green Chem. 18 (21), 5814–5821.
Siebentritt, S., 2013. HCl and Br 2-MeOH etching of Cu2ZnSnSe4 polycrystalline Platzer-Björkman, C., Frisk, C., Larsen, J., Ericson, T., Li, S.-Y., Scragg, J., Keller, J.,
absorbers. Thin Solid Films 535, 83–87. Larsson, F., Törndahl, T., 2015. Reduced interface recombination in Cu2ZnSnS4
Muhunthan, N., Singh, O.P., Thakur, M., Karthikeyan, P., Singh, D., Saravanan, M., solar cells with atomic layer deposition Zn1− x Sn x O y buffer layers. Appl. Phys.
Singh, V., 2014. Interfacial properties of CZTS thin film solar cell. J. Solar Energy Lett. 107 (24), 243904.
2014, 1–2. Polizzotti, A., Repins, I.L., Noufi, R., Wei, S.-H., Mitzi, D.B., 2013. The state and future
Mulvey, T., 1964. Materials, testing and analysis. J. Sci. Instrum. 41 (5), 248. prospects of kesterite photovoltaics. Energy Environ. Sci. 6 (11), 3171–3182.
Mutter, D., Dunham, S.T., 2015. Calculation of defect concentrations and phase stability Qi, Y.-F., Kou, D.-X., Zhou, W.-H., Zhou, Z.-J., Tian, Q.-W., Meng, Y.-N., Liu, X.-S., Du, Z.-
in Cu $ _2 $ ZnSnS $ _4 $ and Cu $ _2 $ ZnSnSe $ _4 $ from stoichiometry. IEEE J. L., Wu, S.-X., 2017a. Engineering of interface band bending and defects elimination
Photovoltaics 5 (4), 1188–1196. via a Ag-graded active layer for efficient (Cu, Ag) 2 ZnSn (S, Se) 4 solar cells. Energy
Environ. Sci. 10 (11), 2401–2410.

55
M. Sahu et al. Solar Energy 230 (2021) 13–58

Qi, Y., Tian, Q., Meng, Y., Kou, D., Zhou, Z., Zhou, W., Wu, S., 2017b. Elemental Schmidt, J., Kerr, M., Cuevas, A., 2001. Surface passivation of silicon solar cells using
Precursor Solution Processed (Cu1–x Ag x) 2ZnSn (S, Se) 4 Photovoltaic Devices with plasma-enhanced chemical-vapour-deposited SiN films and thin thermal SiO2/
over 10% Efficiency. ACS Appl. Mater. Interfaces 9 (25), 21243–21250. plasma SiN stacks. Semicond. Sci. Technol. 16 (3), 164.
Ramasamy, K., Malik, M.A., O’Brien, P., 2012. Routes to copper zinc tin sulfide Cu 2 Schnabel, T., Ahlswede, E., 2017. On the interface between kesterite absorber and Mo
ZnSnS 4 a potential material for solar cells. Chem. Commun. 48 (46), 5703–5714. back contact and its impact on solution-processed thin-film solar cells. Sol. Energy
Ranjbar, S., Brammertz, G., Vermang, B., Hadipour, A., Cong, S., Suganuma, K., Mater. Sol. Cells 159, 290–295.
Schnabel, T., Meuris, M., Da Cunha, A., Poortmans, J., 2017a. Improvement of Schorr, S., 2011. The crystal structure of kesterite type compounds: A neutron and X-ray
kesterite solar cell performance by solution synthesized MoO3 interfacial layer. diffraction study. Sol. Energy Mater. Sol. Cells 95 (6), 1482–1488.
Phys. Stat. Sol. (a) 214 (1), 1600534. Schorr, S., Hoebler, H.-J., Tovar, M., 2007. A neutron diffraction study of the stannite-
Ranjbar, S., Hadipour, A., Vermang, B., Batuk, M., Hadermann, J., Garud, S., kesterite solid solution series. Eur. J. Mineral. 19 (1), 65–73.
Sahayaraj, S., Meuris, M., Brammertz, G., Da Cunha, A.F., 2017b. P-N Junction Schubert, B.A., Marsen, B., Cinque, S., Unold, T., Klenk, R., Schorr, S., Schock, H.W.,
Passivation in Kesterite Solar Cells by Use of Solution-Processed TiO2 Layer. IEEE J. 2011. Cu2ZnSnS4 thin film solar cells by fast coevaporation. Prog. Photovoltaics Res.
Photovoltaics 7 (4), 1130–1135. Appl. 19 (1), 93–96.
Rau, U., Schmidt, M., 2001. Electronic properties of ZnO/CdS/Cu (In, Ga) Se2 solar Schwarz, T., Cojocaru-Mirédin, O., Choi, P., Mousel, M., Redinger, A., Siebentritt, S.,
cells—aspects of heterojunction formation. Thin Solid Films 387 (1–2), 141–146. Raabe, D., 2013. Atom probe study of Cu2ZnSnSe4 thin-films prepared by co-
Rau, U., Werner, J., 2004. Radiative efficiency limits of solar cells with lateral band-gap evaporation and post-deposition annealing. Appl. Phys. Lett. 102 (4), 042101.
fluctuations. Appl. Phys. Lett. 84 (19), 3735–3737. Scragg, J.J., Choubrac, L., Lafond, A., Ericson, T., Platzer-Björkman, C., 2014. A low-
Raulot, J., Domain, C., Guillemoles, J., 2005. Ab initio investigation of potential indium temperature order-disorder transition in Cu2ZnSnS4 thin films. Appl. Phys. Lett. 104
and gallium free chalcopyrite compounds for photovoltaic application. J. Phys. (4), 041911.
Chem. Solids 66 (11), 2019–2023. Scragg, J.J., Dale, P.J., Colombara, D., Peter, L.M., 2012a. Thermodynamic aspects of the
Redinger, A., Berg, D.M., Dale, P.J., Djemour, R., Gütay, L., Eisenbarth, T., Valle, N., synthesis of thin-film materials for solar cells. ChemPhysChem 13 (12), 3035–3046.
Siebentritt, S., 2011a. Route Toward High-Efficiency Single-Phase Cu $ _ {\bf 2} $ Scragg, J.J., Dale, P.J., Peter, L.M., 2008a. Towards sustainable materials for solar energy
ZnSn (S, Se) $ _ {\bf 4} $ Thin-Film Solar Cells: Model Experiments and Literature conversion: Preparation and photoelectrochemical characterization of Cu2ZnSnS4.
Review. IEEE J. Photovoltaics 1 (2), 200–206. Electrochem. Commun. 10 (4), 639–642.
Redinger, A., Hönes, K., Fontané, X., Izquierdo-Roca, V., Saucedo, E., Valle, N., Pérez- Scragg, J.J., Dale, P.J., Peter, L.M., Zoppi, G., Forbes, I., 2008b. New routes to
Rodríguez, A., Siebentritt, S., 2011b. Detection of a ZnSe secondary phase in sustainable photovoltaics: evaluation of Cu2ZnSnS4 as an alternative absorber
coevaporated Cu 2 ZnSnSe 4 thin films. Appl. Phys. Lett. 98 (10), 101907. material. Phys. Stat. Sol. (b) 245 (9), 1772–1778.
Redinger, A., Mousel, M., Wolter, M.H., Valle, N., Siebentritt, S., 2013. Influence of S/Se Scragg, J.J., Kubart, T., Wätjen, J.T., Ericson, T., Linnarsson, M.K., Platzer-Björkman, C.,
ratio on series resistance and on dominant recombination pathway in Cu2ZnSn (SSe) 2013. Effects of back contact instability on Cu2ZnSnS4 devices and processes. Chem.
4 thin film solar cells. Thin Solid Films 535, 291–295. Mater. 25 (15), 3162–3171.
Ren, Y., Richter, M., Keller, J., Redinger, A., Unold, T., Donzel-Gargand, O., Scragg, J.J., Scragg, J.J., Larsen, J.K., Kumar, M., Persson, C., Sendler, J., Siebentritt, S., Platzer
Platzer Björkman, C., 2017. Investigation of the SnS/Cu2ZnSnS4 interfaces in Björkman, C., 2016. Cu–Zn disorder and band gap fluctuations in Cu2ZnSn (S, Se) 4:
kesterite thin-film solar cells. ACS Energy Lett. 2 (5), 976–981. Theoretical and experimental investigations. Phys. Stat. Sol. (b) 253 (2), 247–254.
Repins, I., Contreras, M.A., Egaas, B., DeHart, C., Scharf, J., Perkins, C.L., To, B., Scragg, J.J., Watjen, J.T., Edoff, M., Ericson, T., Kubart, T., Platzer-Björkman, C., 2012b.
Noufi, R., 2008. 19⋅ 9%-efficient ZnO/CdS/CuInGaSe2 solar cell with 81⋅ 2% fill A detrimental reaction at the molybdenum back contact in Cu2ZnSn (S, Se) 4 thin-
factor. Prog. Photovoltaics Res. Appl. 16 (3), 235–239. film solar cells. J. Am. Chem. Soc. 134 (47), 19330–19333.
Repins, I., Vora, N., Beall, C., Wei, S.-H., Yan, Y., Romero, M., Teeter, G., Du, H., To, B., Sekiguchi, K., Tanaka, K., Moriya, K., Uchiki, H., 2006. Epitaxial growth of Cu2ZnSnS4
Young, M., 2011. Kesterites and chalcopyrites: a comparison of close cousins. thin films by pulsed laser deposition. Phys. Stat. Sol. c 3 (8), 2618–2621.
National Renewable Energy Lab. (NREL), Golden, CO (United States). Seol, J.-S., Lee, S.-Y., Lee, J.-C., Nam, H.-D., Kim, K.-H., 2003. Electrical and optical
Rey, G., Redinger, A., Sendler, J., Weiss, T.P., Thevenin, M., Guennou, M., El Adib, B., properties of Cu2ZnSnS4 thin films prepared by rf magnetron sputtering process. Sol.
Siebentritt, S., 2014. The band gap of Cu2ZnSnSe4: Effect of order-disorder. Appl. Energy Mater. Sol. Cells 75 (1–2), 155–162.
Phys. Lett. 105 (11), 112106. Sharbati, S., Keshmiri, S.H., McGoffin, J.T., Geisthardt, R., 2015. Improvement of CIGS
Riha, S.C., Parkinson, B.A., Prieto, A.L., 2009. Solution-based synthesis and thin-film solar cell performance by optimization of Zn (O, S) buffer layer parameters.
characterization of Cu2ZnSnS4 nanocrystals. J. Am. Chem. Soc. 131 (34), Appl. Phys. A 118 (4), 1259–1265.
12054–12055. Sharbati, S., Sites, J.R., 2014. Impact of the Band Offset for n-Zn (O, S)/p-Cu (In, Ga) Se $
Riha, S.C., Parkinson, B.A., Prieto, A.L., 2011. Compositionally Tunable Cu2ZnSn (S1–x _ 2 $ Solar Cells. IEEE J. Photovoltaics 4 (2), 697–702.
Se x) 4 Nanocrystals: Probing the Effect of Se-Inclusion in Mixed Chalcogenide Thin Shin, B., Bojarczuk, N.A., Guha, S., 2013a. On the kinetics of MoSe2 interfacial layer
Films. J. Am. Chem. Soc. 133 (39), 15272–15275. formation in chalcogen-based thin film solar cells with a molybdenum back contact.
Rockett, A., 2005. The effect of Na in polycrystalline and epitaxial single-crystal CuIn1− Appl. Phys. Lett. 102 (9), 091907.
xGaxSe2. Thin Solid Films 480, 2–7. Shin, B., Gunawan, O., Zhu, Y., Bojarczuk, N.A., Chey, S.J., Guha, S., 2013b. Thin film
Romero, M.J., Du, H., Teeter, G., Yan, Y., Al-Jassim, M.M., 2011. Comparative study of solar cell with 8.4% power conversion efficiency using an earth-abundant
the luminescence and intrinsic point defects in the kesterite Cu2ZnSnS4 and Cu2ZnSnS4 absorber. Prog. Photovoltaics Res. Appl. 21 (1), 72–76.
chalcopyrite Cu (In, Ga) Se2 thin films used in photovoltaic applications. Phys. Rev. Shin, B., Zhu, Y., Bojarczuk, N.A., Jay Chey, S., Guha, S., 2012. Control of an interfacial
B 84 (16), 165324. MoSe2 layer in Cu2ZnSnSe4 thin film solar cells: 8.9% power conversion efficiency
Rondiya, S., Rokade, A., Sharma, P., Chaudhary, M., Funde, A., Jadhav, Y., Haram, S., with a TiN diffusion barrier. Appl. Phys. Lett. 101 (5), 053903.
Pathan, H., Jadkar, S., 2018. CZTS/CdS: interface properties and band alignment Shin, D., Saparov, B., Mitzi, D.B., 2017. Defect engineering in multinary earth-abundant
study towards photovoltaic applications. J. Mater. Sci.: Mater. Electron. 29 (5), chalcogenide photovoltaic materials. Adv. Energy Mater. 7 (11), 1602366.
4201–4210. Shin, S.W., Gurav, K., Hong, C.W., Gwak, J., Choi, H.R., Vanalakar, S., Yun, J.H., Lee, J.
Ruckh, M., Schmid, D., Schock, H., 1994. Photoemission studies of the ZnO/CdS Y., Moon, J.H., Kim, J.H., 2015. Phase segregations and thickness of the Mo (S, Se) 2
interface. J. Appl. Phys. 76 (10), 5945–5948. layer in Cu2ZnSn (S, Se) 4 solar cells at different sulfurization temperatures. Sol.
Rudmann, D., Brémaud, D., Zogg, H., Tiwari, A., 2005. Na incorporation into Cu (In, Ga) Energy Mater. Sol. Cells 143, 480–487.
Se 2 for high-efficiency flexible solar cells on polymer foils. J. Appl. Phys. 97 (8), Shin, S.W., Pawar, S., Park, C.Y., Yun, J.H., Moon, J.-H., Kim, J.H., Lee, J.Y., 2011.
084903. Studies on Cu2ZnSnS4 (CZTS) absorber layer using different stacking orders in
Saint-Cast, P., Benick, J., Kania, D., Weiss, L., Hofmann, M., Rentsch, J., Preu, R., precursor thin films. Sol. Energy Mater. Sol. Cells 95 (12), 3202–3206.
Glunz, S.W., 2010. High-efficiency c-Si solar cells passivated with ALD and PECVD Shirakawa, Y., Kukimoto, H., 1980. The electron trap associated with an anion vacancy
aluminum oxide. IEEE Electron Device Lett. 31 (7), 695–697. in ZnSe and ZnSxSe1− x. Solid State Commun. 34 (5), 359–361.
Sakai, N., Hiroi, H., Sugimoto, H., 2011. 37th IEEE Photovol. Special. Conf. (PVSC) Shockley, W., Queisser, H.J., 1961. Detailed balance limit of efficiency of p-n junction
19–24. solar cells. J. Appl. Phys. 32 (3), 510–519.
Salomé, P., Malaquias, J., Fernandes, P., Da Cunha, A., 2010. Mo bilayer for thin film Siebentritt, S., 2013. Why are kesterite solar cells not 20% efficient? Thin Solid Films
photovoltaics revisited. J. Phys. D Appl. Phys. 43 (34), 345501. 535, 1–4.
Sandoval-Paz, M., Sotelo-Lerma, M., Valenzuela-Jauregui, J., Flores-Acosta, M., Ramırez- Siebentritt, S., Schorr, S., 2012. Kesterites—a challenging material for solar cells. Prog.
Bon, R., 2005. Structural and optical studies on thermal-annealed In2S3 films Photovoltaics Res. Appl. 20 (5), 512–519.
prepared by the chemical bath deposition technique. Thin Solid Films 472 (1–2), Son, D.-H., Kim, S.-H., Kim, S.-Y., Kim, Y.-I., Sim, J.-H., Park, S.-N., Jeon, D.-H.,
5–10. Hwang, D.-K., Sung, S.-J., Kang, J.-K., 2019. Effect of solid-H 2 S gas reactions on
Santoni, A., Biccari, F., Malerba, C., Valentini, M., Chierchia, R., Mittiga, A., 2013. CZTSSe thin film growth and photovoltaic properties of a 12.62% efficiency device.
Valence band offset at the CdS/Cu2ZnSnS4 interface probed by x-ray photoelectron J. Mater. Chem. A 7 (44), 25279–25289.
spectroscopy. J. Phys. D Appl. Phys. 46 (17), 175101. Spies, J., Schafer, R., Wager, J., Hersh, P., Platt, H., Keszler, D., Schneider, G.,
Sasamura, T., Osaki, T., Kameyama, T., Shibayama, T., Kudo, A., Kuwabata, S., Kykyneshi, R., Tate, J., Liu, X., 2009. Pin double-heterojunction thin-film solar cell
Torimoto, T., 2012. Solution-phase synthesis of stannite-type Ag2ZnSnS4 p-layer assessment. Sol. Energy Mater. Sol. Cells 93 (8), 1296–1308.
nanoparticles for application to photoelectrode materials. Chem. Lett. 41 (9), Su, C.-Y., Liao, K.-H., Pan, C.-T., Peng, P.-W., 2012. The effect of deposition parameters
1009–1011. and post treatment on the electrical properties of Mo thin films. Thin Solid Films 520
Scheer, R., Schock, H.-W., 2011. Chalcogenide photovoltaics: physics, technologies, and (18), 5936–5939.
thin film devices. John Wiley & Sons. Su, D., Neumann, W., Hunger, R., Schubert-Bischoff, P., Giersig, M., Lewerenz, H.,
Schlaf, R., Pettenkofer, C., Jaegermann, W., 1999. Band lineup of a SnS 2/SnSe 2/SnS 2 Scheer, R., Zeitler, E., 1998. CuAu-type ordering in epitaxial CuInS 2 films. Appl.
semiconductor quantum well structure prepared by van der Waals epitaxy. J. Appl. Phys. Lett. 73 (6), 785–787.
Phys. 85 (9), 6550–6556.

56
M. Sahu et al. Solar Energy 230 (2021) 13–58

Su, Z., Tan, J.M.R., Li, X., Zeng, X., Batabyal, S.K., Wong, L.H., 2015. Cation substitution Walsh, A., Chen, S., Wei, S.H., Gong, X.G., 2012. Kesterite thin-film solar cells: advances
of solution-processed Cu2ZnSnS4 thin film solar cell with over 9% efficiency. Adv. in materials modelling of Cu2ZnSnS4. Adv. Energy Mater. 2 (4), 400–409.
Energy Mater. 5 (19), 1500682. Walter, T., Herberholz, R., Müller, C., Schock, H., 1996. Determination of defect
Sun, K., Yan, C., Liu, F., Huang, J., Zhou, F., Stride, J.A., Green, M., Hao, X., 2016. Over distributions from admittance measurements and application to Cu (In, Ga) Se2
9% efficient kesterite Cu2ZnSnS4 solar cell fabricated by using Zn1–xCdxS buffer based heterojunctions. J. Appl. Phys. 80 (8), 4411–4420.
layer. Adv. Energy Mater. 6 (12), 1600046. Wang, D., Zhao, W., Zhang, Y., Liu, S.F., 2018a. Path towards high-efficient kesterite
Sun, R., Zhuang, D., Zhao, M., Gong, Q., Scarpulla, M., Wei, Y., Ren, G., Wu, Y., 2018. solar cells. J. Energy Chem. 27 (4), 1040–1053.
Beyond 11% efficient Cu2ZnSn (Se, S) 4 thin film solar cells by cadmium alloying. Wang, K., Gunawan, O., Todorov, T., Shin, B., Chey, S., Bojarczuk, N., Mitzi, D., Guha, S.,
Sol. Energy Mater. Sol. Cells 174, 494–498. 2010. Thermally evaporated Cu 2 ZnSnS 4 solar cells. Appl. Phys. Lett. 97 (14),
Tai, K.F., Gunawan, O., Kuwahara, M., Chen, S., Mhaisalkar, S.G., Huan, C.H.A., Mitzi, D. 143508.
B., 2016. Fill factor losses in Cu2ZnSn (SxSe1− x) 4 solar cells: insights from physical Wang, K., Shin, B., Reuter, K.B., Todorov, T., Mitzi, D.B., Guha, S., 2011. Structural and
and electrical characterization of devices and exfoliated films. Adv. Energy Mater. 6 elemental characterization of high efficiency Cu 2 ZnSnS 4 solar cells. Appl. Phys.
(3), 1501609. Lett. 98 (5), 051912.
Tajima, S., Kataoka, K., Takahashi, N., Kimoto, Y., Fukano, T., Hasegawa, M., Wang, W., Chen, G., Cai, H., Chen, B., Yao, L., Yang, M., Chen, S., Huang, Z., 2018b. The
Hazama, H., 2013. Direct measurement of band offset at the interface between CdS effects of SnS 2 secondary phases on Cu 2 ZnSnS 4 solar cells: a promising
and Cu2ZnSnS4 using hard X-ray photoelectron spectroscopy. Appl. Phys. Lett. 103 mechanical exfoliation method for its removal. J. Mater. Chem. A 6 (7), 2995–3004.
(24), 243906. Wang, W., Winkler, M.T., Gunawan, O., Gokmen, T., Todorov, T.K., Zhu, Y., Mitzi, D.B.,
Tajima, S., Umehara, M., Mise, T., 2016. Photovoltaic properties of Cu2ZnSnS4 cells 2014. Device characteristics of CZTSSe thin-film solar cells with 12.6% efficiency.
fabricated using ZnSnO and ZnSnO/CdS buffer layers. Jpn. J. Appl. Phys. 55 (11), Adv. Energy Mater. 4 (7), 1301465.
112302. Wätjen, J.T., Scragg, J.J., Ericson, T., Edoff, M., Platzer-Björkman, C., 2013. Secondary
Takahashi, M., Fujiwara, T., Ikushima, A.J., 1998. Defect Formation in GeO2–SiO2 Glass compound formation revealed by transmission electron microscopy at the
by UV-excited Poling. Jpn. J. Appl. Phys. 37 (S1), 71. Cu2ZnSnS4/Mo interface. Thin Solid Films 535, 31–34.
Tampo, H., Kim, K.M., Kim, S., Shibata, H., Niki, S., 2017. Improvement of minority Wei, K., Nolas, G.S., 2015. Synthesis and characterization of nanostructured stannite
carrier lifetime and conversion efficiency by Na incorporation in Cu2ZnSnSe4 solar Cu2ZnSnSe4 and Ag2ZnSnSe4 for thermoelectric applications. ACS Appl. Mater.
cells. J. Appl. Phys. 122 (2), 023106. Interfaces 7 (18), 9752–9757.
Tanaka, K., Fukui, Y., Moritake, N., Uchiki, H., 2011. Chemical composition dependence Wei, S.-H., Zhang, S., 2005. Defect properties of CuInSe2 and CuGaSe2. J. Phys. Chem.
of morphological and optical properties of Cu2ZnSnS4 thin films deposited by Solids 66 (11), 1994–1999.
sol–gel sulfurization and Cu2ZnSnS4 thin film solar cell efficiency. Sol. Energy Wei, S.-H., Zhang, S., Zunger, A., 1999. Band structure and stability of zinc-blende-based
Mater. Sol. Cells 95 (3), 838–842. semiconductor polytypes. Phys. Rev. B 59 (4), R2478.
Tanaka, K., Oonuki, M., Moritake, N., Uchiki, H., 2009. Cu2ZnSnS4 thin film solar cells Weinhardt, L., Heske, C., Umbach, E., Niesen, T., Visbeck, S., Karg, F., 2004. Band
prepared by non-vacuum processing. Sol. Energy Mater. Sol. Cells 93 (5), 583–587. alignment at the i-ZnO/CdS interface in Cu (In, Ga)(S, Se) 2 thin-film solar cells.
Tanaka, T., Nagatomo, T., Kawasaki, D., Nishio, M., Guo, Q., Wakahara, A., Yoshida, A., Appl. Phys. Lett. 84 (16), 3175–3177.
Ogawa, H., 2005. Preparation of Cu2ZnSnS4 thin films by hybrid sputtering. J. Phys. Werfel, F., Minni, E., 1983. Photoemission study of the electronic structure of Mo and Mo
Chem. Solids 66 (11), 1978–1981. oxides. J. Phys. C: Solid State Phys. 16 (31), 6091.
Tanaka, T., Sueishi, T., Saito, K., Guo, Q., Nishio, M., Yu, K.M., Walukiewicz, W., 2012. Wibowo, R.A., Kim, W.S., Lee, E.S., Munir, B., Kim, K.H., 2007. Single step preparation of
Existence and removal of Cu2Se second phase in coevaporated Cu2ZnSnSe4 thin quaternary Cu2ZnSnSe4 thin films by RF magnetron sputtering from binary
films. J. Appl. Phys. 111 (5), 053522. chalcogenide targets. J. Phys. Chem. Solids 68 (10), 1908–1913.
Temgoua, S., Bodeux, R., Naghavi, N., Delbos, S., 2015. Effects of SnSe2 secondary Wu, H.-M., Liang, S.-C., Lin, Y.-L., Ni, C.-Y., Bor, H.-Y., Tsai, D.-C., Shieu, F.-S., 2012.
phases on the efficiency of Cu2ZnSn (Sx, Se1− x) 4 based solar cells. Thin Solid Films Structure and electrical properties of Mo back contact for Cu (In, Ga) Se2 solar cells.
582, 215–219. Vacuum 86 (12), 1916–1919.
Timmo, K., Altosaar, M., Raudoja, J., Grossberg, M., Danilson, M., Volobujeva, O., Wu, W., Cao, Y., Caspar, J.V., Guo, Q., Johnson, L.K., Mclean, R.S., Malajovich, I.,
Mellikov, E., 2010. Chemical etching of Cu 2 ZnSn (S, Se) 4 monograin powder. In: Choudhury, K.R., 2014. Characterization of CZTSSe photovoltaic device with an
2010 35th IEEE Photovoltaic Specialists Conference. IEEE, pp. 001982–001985. atomic layer-deposited passivation layer. Appl. Phys. Lett. 105 (4), 042108.
Timo Wätjen, J., Engman, J., Edoff, M., Platzer-Björkman, C., 2012. Direct evidence of Würz, R., Marrón, D.F., Meeder, A., Rumberg, A., Babu, S.M., Schedel-Niedrig, T.,
current blocking by ZnSe in Cu2ZnSnSe4 solar cells. Appl. Phys. Lett. 100 (17), Bloeck, U., Schubert-Bischoff, P., Lux-Steiner, M.C., 2003. Formation of an
173510. interfacial MoSe2 layer in CVD grown CuGaSe2 based thin film solar cells. Thin Solid
Todorov, T.K., Reuter, K.B., Mitzi, D.B., 2010. High-efficiency solar cell with earth- Films 431, 398–402.
abundant liquid-processed absorber. Adv. Mater. 22 (20), E156–E159. Xiao, Z.-Y., Yao, B., Li, Y.-F., Ding, Z.-H., Gao, Z.-M., Zhao, H.-F., Zhang, L.-G., Zhang, Z.-
Todorov, T.K., Tang, J., Bag, S., Gunawan, O., Gokmen, T., Zhu, Y., Mitzi, D.B., 2013. Z., Sui, Y.-R., Wang, G., 2016. Influencing mechanism of the selenization
Beyond 11% efficiency: characteristics of state-of-the-art Cu2ZnSn (S, Se) 4 solar temperature and time on the power conversion efficiency of Cu2ZnSn (S, Se) 4-based
cells. Adv. Energy Mater. 3 (1), 34–38. solar cells. ACS Appl. Mater. Interfaces 8 (27), 17334–17342.
Tong, Z., Yan, C., Su, Z., Zeng, F., Yang, J., Li, Y., Jiang, L., Lai, Y., Liu, F., 2014. Effects of Xie, H., Dimitrievska, M., Fontané, X., Sánchez, Y., López-Marino, S., Izquierdo-Roca, V.,
potassium doping on solution processed kesterite Cu2ZnSnS4 thin film solar cells. Bermúdez, V., Pérez-Rodríguez, A., Saucedo, E., 2015. Formation and impact of
Appl. Phys. Lett. 105 (22), 223903. secondary phases in Cu-poor Zn-rich Cu2ZnSn (S1− ySey) 4 (0≤ y≤ 1) based solar
Tress, W., Marinova, N., Moehl, T., Zakeeruddin, S.M., Nazeeruddin, M.K., Grätzel, M., cells. Sol. Energy Mater. Sol. Cells 140, 289–298.
2015. Understanding the rate-dependent J-V hysteresis, slow time component, and Xie, H., Sánchez, Y., López-Marino, S., Espíndola-Rodríguez, M., Neuschitzer, M.,
aging in CH 3 NH 3 PbI 3 perovskite solar cells: the role of a compensated electric Sylla, D., Fairbrother, A., Izquierdo-Roca, V., Pérez-Rodríguez, A., Saucedo, E., 2014.
field. Energy Environ. Sci. 8 (3), 995–1004. Impact of Sn (S, Se) secondary phases in Cu2ZnSn (S, Se) 4 solar cells: a chemical
Tsuji, I., Shimodaira, Y., Kato, H., Kobayashi, H., Kudo, A., 2010. Novel stannite-type route for their selective removal and absorber surface passivation. ACS Appl. Mater.
complex sulfide photocatalysts AI2-Zn-AIV-S4 (AI= Cu and Ag; AIV= Sn and Ge) for Interfaces 6 (15), 12744–12751.
hydrogen evolution under visible-light irradiation. Chem. Mater. 22 (4), 1402–1409. Xin, H., Vorpahl, S., Collord, A., Braly, I., Uhl, A., Krueger, B., Ginger, D., Hillhouse, H.,
Udaka, Y., Takaki, S.I., Isowaki, K., Nagai, T., Kim, K.M., Kim, S., Tampo, H., Shibata, H., 2015. Lithium-doping inverts the nanoscale electric field at the grain boundaries in
Matsubara, K., Niki, S., 2017. Electronic structure of Cu2ZnSn (SxSe1− x) 4 surface Cu 2 ZnSn (S, Se) 4 and increases photovoltaic efficiency. PCCP 17 (37),
and CdS/Cu2ZnSn (SxSe1− x) 4 interface. Phys. Stat. Sol. c 14 (6), 1600178. 23859–23866.
Unold, T., Kretzschmar, S., Just, J., Zander, O., Schubert, B., Marsen, B., Schock, H., Xiong, S., Zhu, M., 2009. Fundamentals and Applications of solar cells. Science Press,
2011. In 37th IEEE Photovoltaic Specialist Conference. IEEE, Seattle. Beijing.
Vauche, L., Risch, L., Arasimowicz, M., Sánchez, Y., Saucedo, E., Pasquinelli, M., Goislard Yan, C., Huang, J., Sun, K., Johnston, S., Zhang, Y., Sun, H., Pu, A., He, M., Liu, F.,
de Monsabert, T., Grand, P.-P., Jaime-Ferrer, S., 2016. Detrimental effect of Sn-rich Eder, K., 2018. Cu 2 ZnSnS 4 solar cells with over 10% power conversion efficiency
secondary phases on Cu2ZnSnSe4 based solar cells. J. Renew. Sustain. Energy 8 (3), enabled by heterojunction heat treatment. Nat. Energy 3 (9), 764–772.
033502. Yan, C., Liu, F., Song, N., Ng, B.K., Stride, J.A., Tadich, A., Hao, X., 2014. Band
Vermang, B., Fjällström, V., Pettersson, J., Salomé, P., Edoff, M., 2013. Development of alignments of different buffer layers (CdS, Zn (O, S), and In2S3) on Cu2ZnSnS4.
rear surface passivated Cu (In, Ga) Se2 thin film solar cells with nano-sized local rear Appl. Phys. Lett. 104 (17), 173901.
point contacts. Sol. Energy Mater. Sol. Cells 117, 505–511. Yan, C., Liu, F., Sun, K., Song, N., Stride, J.A., Zhou, F., Hao, X., Green, M., 2016.
Vermang, B., Ren, Y., Donzel-Gargand, O., Frisk, C., Joel, J., Salome, P., Borme, J., Boosting the efficiency of pure sulfide CZTS solar cells using the In/Cd-based hybrid
Sadewasser, S., Platzer-Björkman, C., Edoff, M., 2015. Rear surface optimization of buffers. Sol. Energy Mater. Sol. Cells 144, 700–706.
CZTS solar cells by use of a passivation layer with nanosized point openings. IEEE J. Yan, C., Sun, K., Huang, J., Johnston, S., Liu, F., Veettil, B.P., Sun, K., Pu, A., Zhou, F.,
Photovoltaics 6 (1), 332–336. Stride, J.A., 2017. Beyond 11% Efficient Sulfide Kesterite Cu2Zn x Cd1–x SnS4 Solar
Vineyard, G.H., 1956. Theory of order-disorder kinetics. Phys. Rev. 102 (4), 981. Cell: Effects of Cadmium Alloying. ACS Energy Lett. 2 (4), 930–936.
Vora, N., Blackburn, J., Repins, I., Beall, C., To, B., Pankow, J., Teeter, G., Young, M., Yan, Y., Jiang, C.-S., Noufi, R., Wei, S.-H., Moutinho, H., Al-Jassim, M., 2007. Electrically
Noufi, R., 2012. Phase identification and control of thin films deposited by co- benign behavior of grain boundaries in polycrystalline CuInSe 2 films. Phys. Rev.
evaporation of elemental Cu, Zn, Sn, and Se. J. Vacuum Sci. Technol. A: Vacuum Lett. 99 (23), 235504.
Surf. Films 30 (5), 051201. Yang, K.J., Sim, J.H., Son, D.H., Jeon, D.H., Hwang, D.K., Nam, D., Cheong, H., Kim, S.Y.,
Wada, T., Kohara, N., Nishiwaki, S., Negami, T., 2001. Characterization of the Cu (In, Ga) Kim, J.H., Kim, D.H., Kang, J.K., 2017. Comparison of chalcopyrite and kesterite
Se2/Mo interface in CIGS solar cells. Thin Solid Films 387 (1–2), 118–122. thin-film solar cells. J. Ind. Eng. Chem. 45, 78–84.
Wagner, S., Bridenbaugh, P., 1977. Multicomponent tetrahedral compounds for solar
cells. J. Cryst. Growth 39 (1), 151–159.

57
M. Sahu et al. Solar Energy 230 (2021) 13–58

Yang, K.-J., Kim, S., Kim, S.-Y., Ahn, K., Son, D.-H., Kim, S.-H., Lee, S.-J., Kim, Y.-I., efficiency improvement in Cu2ZnSn (S, Se) 4 solar cell. Phys. Stat. Sol. (RRL)–Rap.
Park, S.-N., Sung, S.-J., 2019. Flexible Cu 2 ZnSn (S, Se) 4 solar cells with over 10% Res. Lett. 9 (12), 687–691.
efficiency and methods of enlarging the cell area. Nat. Commun. 10 (1), 1–10. Zhang, J., Lexi, S., Yujun, F., Erqing, X., 2006. Cu2ZnSnS4 thin films prepared by
Yang, K.J., Sim, J.H., Son, D.H., Kim, D.H., Kim, G.Y., Jo, W., Song, S., Kim, J., Nam, D., sulfurization of ion beam sputtered precursor and their electrical and optical
Cheong, H., 2015. Effects of the compositional ratio distribution with sulfurization properties. Rare Met. 25 (6), 315–319.
temperatures in the absorber layer on the defect and surface electrical characteristics Zhang, X., Han, M., Zeng, Z., Lin, H.Q., 2016a. The instability of S vacancies in Cu 2
of Cu2ZnSnS4 solar cells. Prog. Photovoltaics Res. Appl. 23 (12), 1771–1784. ZnSnS 4. RSC Adv. 6 (19), 15424–15429.
Yang, Y., Kang, X., Huang, L., Pan, D., 2016. Tuning the band gap of Cu2ZnSn (S, Se) 4 Zhang, X., Han, Y., Chai, S.-Z., Hu, N.-T., Yang, Z., Geng, H.-J., Wei, H., 2016b. Advances
thin films via lithium alloying. ACS Appl. Mater. Interfaces 8 (8), 5308–5313. in Cu2ZnSn (S, Se) 4 thin film solar cells. Acta Phys. Chim. Sin. 32 (6), 1330–1346.
Yao, L., Ao, J., Jeng, M.-J., Bi, J., Gao, S., Sun, G., He, Q., Zhou, Z., Sun, Y., Chang, L.-B., Zhang, X., Kobayashi, M., Yamada, A., 2017. Comparison of Ag (In, Ga) Se2/Mo and Cu
2017. A CZTSe solar cell with 8.2% power conversion efficiency fabricated using (In, Ga) Se2/Mo Interfaces in Solar Cells. ACS Appl. Mater. Interfaces 9 (19),
electrodeposited Cu/Sn/Zn precursor and a three-step selenization process at low Se 16215–16220.
pressure. Sol. Energy Mater. Sol. Cells 159, 318–324. Zhao, J., Wang, A., Green, M.A., 1999. 24⋅ 5% Efficiency silicon PERT cells on MCZ
Yeh, L.-Y., Cheng, K.-W., 2014. Preparation of the Ag–Zn–Sn–S quaternary substrates and 24⋅ 7% efficiency PERL cells on FZ substrates. Prog. Photovoltaics
photoelectrodes using chemical bath deposition for photoelectrochemical Res. Appl. 7 (6), 471–474.
applications. Thin Solid Films 558, 289–293. Zhao, W., Wang, G., Tian, Q., Huang, L., Gao, S., Pan, D., 2015. Solution-processed
Yin, L., Cheng, G., Feng, Y., Li, Z., Yang, C., Xiao, X., 2015. Limitation factors for the Cu2CdSn (S, Se) 4 thin film solar cells. Sol. Energy Mater. Sol. Cells 133, 15–20.
performance of kesterite Cu 2 ZnSnS 4 thin film solar cells studied by defect Zhong, J., Xia, Z., Luo, M., Zhao, J., Chen, J., Wang, L., Liu, X., Xue, D.-J., Cheng, Y.-B.,
characterization. RSC Adv. 5 (50), 40369–40374. Song, H., 2014. Sulfurization induced surface constitution and its correlation to the
Yin, W.J., Wu, Y., Wei, S.H., Noufi, R., Al-Jassim, M.M., Yan, Y., 2014a. Engineering performance of solution-processed Cu 2 ZnSn (S, Se) 4 solar cells. Sci. Rep. 4 (1),
grain boundaries in Cu2ZnSnSe4 for better cell performance: a first-principle study. 1–9.
Adv. Energy Mater. 4 (1), 1300712. Zhou, F., Zeng, F., Liu, X., Liu, F., Song, N., Yan, C., Pu, A., Park, J., Sun, K., Hao, X.,
Yin, X., Tang, C., Sun, L., Shen, Z., Gong, H., 2014b. Study on phase formation 2015. Improvement of J sc in a Cu2ZnSnS4 Solar Cell by using a thin carbon
mechanism of non-and near-stoichiometric Cu2ZnSn (S, Se) 4 film prepared by intermediate layer at the Cu2ZnSnS4/Mo Interface. ACS Appl. Mater. Interfaces 7
selenization of Cu–Sn–Zn–S precursors. Chem. Mater. 26 (6), 2005–2014. (41), 22868–22873.
Yoshikawa, K., Kawasaki, H., Yoshida, W., Irie, T., Konishi, K., Nakano, K., Uto, T., Zhou, H., Chen, Q., Li, G., Luo, S., Song, T.-B., Duan, H.-S., Hong, Z., You, J., Liu, Y.,
Adachi, D., Kanematsu, M., Uzu, H., 2017. Silicon heterojunction solar cell with Yang, Y., 2014. Interface engineering of highly efficient perovskite solar cells.
interdigitated back contacts for a photoconversion efficiency over 26%. Nat. Energy Science 345 (6196), 542–546.
2 (5), 1–8. Zhou, H., Hsu, W.-C., Duan, H.-S., Bob, B., Yang, W., Song, T.-B., Hsu, C.-J., Yang, Y.,
Yu, J., Zheng, Z., Dong, L., Cheng, S., Lai, Y., Zheng, Q., Zhou, H., Jia, H., Zhang, H., 2013a. CZTS nanocrystals: a promising approach for next generation thin film
2017. Effect of deposited temperatures of the buffer layer on the band offset of photovoltaics. Energy Environ. Sci. 6 (10), 2822–2838.
CZTS/In 2 S 3 heterostructure and its solar cell performance. Chin. Phys. B 26 (4), Zhou, H., Song, T.-B., Hsu, W.-C., Luo, S., Ye, S., Duan, H.-S., Hsu, C.-J., Yang, W.,
046802. Yang, Y., 2013b. Rational defect passivation of Cu2ZnSn (S, Se) 4 photovoltaics with
Yuan, Z.K., Chen, S., Xiang, H., Gong, X.G., Walsh, A., Park, J.S., Repins, I., Wei, S.H., solution-processed Cu2ZnSnS4: Na nanocrystals. J. Am. Chem. Soc. 135 (43),
2015. Engineering solar cell absorbers by exploring the band alignment and defect 15998–16001.
disparity: the case of Cu-and Ag-based kesterite compounds. Adv. Funct. Mater. 25 Zhou, Z., Wang, Y., Xu, D., Zhang, Y., 2010. Fabrication of Cu2ZnSnS4 screen printed
(43), 6733–6743. layers for solar cells. Sol. Energy Mater. Sol. Cells 94 (12), 2042–2045.
Zeng, F., Sun, K., Gong, L., Jiang, L., Liu, F., Lai, Y., Li, J., 2015. Back contact–absorber Zhu, L.Q., Li, X., Yan, Z.H., Zhang, H.L., Wan, Q., 2012. Dual Function of Antireflectance
interface modification by inserting carbon intermediate layer and conversion and Surface Passivation of Atomic-Layer-Deposited $\hbox Al _ {2}\hbox O _ 3 $
Films. IEEE Electron Device Lett. 33 (12), 1753–1755.

58