You are on page 1of 9

Solar Energy Materials & Solar Cells 119 (2013) 181189

Contents lists available at ScienceDirect

Solar Energy Materials & Solar Cells


journal homepage: www.elsevier.com/locate/solmat

Recent trends in direct solution coating of kesterite absorber layers


in solar cells
Y.E. Romanyuk a,n, C.M. Fella a, A.R. Uhl a, M. Werner a, A.N. Tiwari a,
T. Schnabel b, E. Ahlswede b
a
b

Laboratory for Thin Films and Photovoltaics, EmpaSwiss Federal Laboratories for Materials Science and Technology, 8600 Dbendorf, Switzerland
Zentrum fr Sonnenenergie- und Wasserstoff-Forschung Baden-Wrttemberg (ZSW), 70565 Stuttgart, Germany

art ic l e i nf o

a b s t r a c t

Available online 11 July 2013

This review describes a specic group of non-vacuum methods for depositing kesterite solar cell
absorbers, which we refer to as direct solution coating (DSC). These methods work by coating a layer of
precursor-containing solution on the surface of a substrate followed by appropriate thermal processing to
yield the desired kesterite Cu2ZnSn(S,Se)4 phase. General aspects, challenges and future prospects of
the DSC are discussed in detail. Two research highlights are presented in which kesterite layers are
processed from solutions of metal salts in simple non-toxic organic solvents to yield up to 7.5% efcient
solar cells.
& 2013 Elsevier B.V. All rights reserved.

Keywords:
Kesterite
Cu2ZnSn(S,Se)4
CZTS
Non-vacuum deposition
Direct solution coating
Thin lm solar cell

1. Introduction
Kesterites the family of compounds comprising Cu2ZnSnS4
(CZTS), Cu2ZnSnSe4 (CZTSe) and associated alloys are certainly
one of the most intriguing absorber materials for thin-lm solar
cells today. They may offer low-cost manufacturing starting from
abundant and non-toxic materials, provided that the optoelectronic characteristics of kesterite solar cells are better understood and can be further improved [1,2]. Kesterite thin lms can
be prepared by a variety of deposition methods resulting in
different material qualities and solar cell efciencies. The current
record efciency of 11.1% is reported for a Cu2ZnSn(S,Se)4 (CZTSSe)
absorber processed with a mixed solution-particle approach in
hydrazine solvent [3]. Other notable results include 9.3% efciency
for CZTSSe by co-sputtering with post-annealing in SnS and S [4],
9.15% for CZTSe by co-evaporation [5], 8.4% for CZTGeSSe sintered
from nanoparticle inks [6], 7.3% for CZTS by electroplating metal
precursors with post-sulfurization [7], and recently, 7.5% for
CZTSSe processed from a metal salt solution in dimethyl sulfoxide
(DMSO) followed by selenization [8]. Several recent reviews
summarize the state of the art in non-vacuum processing of
kesterites [1,9], the current understanding of the defect physics
[10], as well as general aspects of the emerging kesterite solar cell
technology [1,2,11,13].

Corresponding author. Tel.: +41 58 765 4169; fax: +41 58 765 1122.
E-mail addresses: yaroslav.romanyuk@empa.ch,
yaroslav.romanyuk@gmail.com (Y.E. Romanyuk).
0927-0248/$ - see front matter & 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.solmat.2013.06.038

CZTSSe can only be a reasonable alternative to the established


Cu(In,Ga)Se2 (CIGS) technology if the cell efciencies reach nearly
the same values, i.e. in the range of 20% in the laboratory. Lower
efciencies are acceptable only if the overall cost per watt is
signicantly lower. As the cost of the absorber material contributes
only a minor part in today's vacuum-based CIGS solar module
manufacturing and system costs, the replacement of the expensive
indium and gallium must also be accompanied by a cost-effective
and simple production route. This means low cost precursor
materials and high throughput processing methods have to be
combined for development of high efciency CZTSSe solar module
production. It is hence very encouraging that the rapidly progressing CZTSSe technology already achieves similar efciencies for
devices processed with both traditional vacuum and for nonvacuum approaches.
In this paper we focus on the specic group of non-vacuum
deposition methods that work by coating a layer of precursorcontaining solution on the surface of a substrate followed by
appropriate thermal processing to yield the desired kesterite
phase. We refer to this group of methods as direct solution coating
(DSC) in analogy with the expression direct liquid coating seen in
publications from the IBM group e.g. [9], or the often used
expression chemical solution deposition (CSD) [14]. There are
several reasons why we would like to limit our considerations only
to the direct coating of solutions:
This approach offers all advantages of non-vacuum deposition
such as the possibility to use low-cost and high-throughput
equipment, low wastage of raw materials, high uniformity of

Y.E. Romanyuk et al. / Solar Energy Materials & Solar Cells 119 (2013) 181189

layer properties over large area and nally, potential deployment of large-scale production with low capital investment.
In contrast to the other popular solution-based techniques for
kesterites such as chemical bath deposition (CBD) [15,16],
electrodeposition (ED) [7], successive ionic layer absorption
and reaction (SILAR) [17], and spray pyrolysis [18,21] which
imply a slow growth of the kesterite phase during the actual
deposition step, the DSC techniques such as spin-, spray-, dip-,
knife-, and curtain-coating, as well as printing techniques like
ink-jet, exography, and gravure allow fast coating of the wet
layer with low wastage of the precursor solution.
As compared to nanoparticle-based inks, solutions offer the
following benets: (i) no need for pre-synthesized nanocrystals; (ii) readily available, cheap metal salts with high purity
and long storage time; and (iii) reproducible processing with a
homogeneous intermixing of constituent elements on the
molecular level.
It is important to note that the nucleation of nanoparticles can
also occur in an initially homogeneous precursor solution leading
to the formation of a colloid sol, especially if the solution
constituents favor condensation and redox reactions. This fact
blurs out the clear division between true solution and colloidal
dispersion approaches. In this paper we will focus on homogeneous precursor solutions prepared without any pre-synthesized
nanoparticles, not excluding the possibility of the in-situ formation of colloidal aggregates.
The DSC approach obviously has its own challenges and
limitations. The most critical issues are: (i) the absolute need for
a high-temperature annealing step, preferably in a chalcogencontaining atmosphere; (ii) the adequate chemical coordination of
metal species in coated precursor layers to prevent undesirable
metal segregations and metal losses during a subsequent thermal
treatment; (iii) unwanted effects associated with the drying of a
wet lm such as increased layer porosity, cracking and delamination; (iv) possible contamination with solvent elements and
unwanted residuals of the initial precursor solution. Here we
present a critical assessment of the most important aspects of
the DSC and describe two research highlights in which kesterite
layers are processed from solutions of metal salts in simple nontoxic organic solvents to yield up to 7.5% efcient solar cells.

2. General aspects of DSC of kesterites


The DSC of kesterites involves several steps starting from
solution formulation and nishing with high-temperature annealing to yield the desired crystalline CZTS(Se) layer. Fig. 1 represents
a typical DSC procedure that includes the preparation of precursor
solution, coating of a wet lm, drying, and a nal annealing step
for chemical reaction and crystallization. It is important to note
that coating and drying steps can be repeated to obtain multilayer
structures. This is often required to build up layer thickness, but
also can be used for improving layer homogeneity and creating
intentional compositional gradients in the multilayer precursor
coating. During each step, a variety of parameters have to be
controlled as discussed below in the individual sections.
2.1. Solution formulation
2.1.1. Solvent
The solvent is one of the most critical components and is often
used to name a particular DSC approach in the scientic community, e.g. hydrazine route or DMSO route. Two primary requirements on the solvents are: (i) to dissolve Cu-, Zn-, Sn-, and, if
possible, chalcogen-containing compounds, and (ii) not to leave

Precursor solution

coating
multilayer coating

182

Wet film

drying
Dry film
annealing / crystallization
Crystalline CZTS(Se) layer

Fig. 1. Processing steps in direct solution coating of a kesterite layer, including a


possibility of multilayer coating.

any detrimental impurity elements in the CZTS(Se) layer after the


annealing step. There are many secondary factors that should be
considered such as surface tension, volatility, reactivity, toxicity,
and cost. Only polar solvents have been used for the DSC of
kesterite due to the ionic nature of the metal compounds to be
dissolved. An ideal solvent for DSC should be able to coordinate
metal cations in the precursor solution by forming soluble complexes which, however, should fully decompose during the heat
treatment without leaving unwanted impurities.
Water is certainly the cheapest, safest and most environmentally friendly solvent. There is, however, only one report of water
being employed as the solvent for spin-coating water-ethanol
solution of metal salts and thiourea to obtain CZTS layers on glass
substrates [22]. Many salts of Cu and Sn readily hydrolyze in water
and thus induce high solution acidity (pH o2) that can corrode not
only the molybdenum-coated substrate but also certain metallic
parts of the coating equipment. Water molecules coordinate metal
cations only weakly, so that hydrated salts of Cu, Zn, or Sn release
water upon the drying step well below 300 1C. Handling of dried
metal salt precursors under ambient conditions is problematic,
because anhydrous salts of Cu, Zn, and especially Sn are hygroscopic and therefore intensively intake the moisture from air
resulting in re-crystallization and undesired segregations. In addition, salts of Sn(IV) can easily evaporate upon annealing or even
during drying (e.g. SnCl4 has a low boiling point of 114 1C), which
leads to a decrease of Sn content if no additional stabilizers are
used. Two other issues with water as the solvent are the high
surface tension, often leading to poor wetting, and its tendency to
introduce oxygen impurities.
Hydrazine (N2H4) is the carbon- and oxygen-free compound
that can be considered as an ideal solvent because it dissolves
metal chalcogenides and elemental chalcogens (sulfur and selenium) and excludes the introduction of any detrimental impurity
elements by decomposing cleanly into N2, NH3, and H2. Another
important feature of hydrazine is its strong reducing action that
prevents any metal oxidation [9]. The hydrazine route was developed by the IBM group [23] to achieve the current highest
efciency of 11.1% for a CZTSSe absorber [3]. The IBM hybrid route
foresees, however, the formation of ZnS(Se) nanoparticles because
of the low solubility of ZnS(Se) in N2H4. Since this hybrid approach
is beyond the scope of this review we refer the interested reader to

Y.E. Romanyuk et al. / Solar Energy Materials & Solar Cells 119 (2013) 181189

other dedicated reviews [1,9,24]. An example of the true solution


where all metals including Zn were incorporated into a N2H4based solution was reported recently [25,26]. The hydrazinocarboxylic acid (NH2NHCOOH) was used to fully dissolve zinc powder
in hydrazine, and an 8.08% efcient CZTSSe cell was achieved [25].
Hydrazine is highly toxic and explosive, and all of its processing
requires inert atmosphere and handling inside a glove box. The
transportation and use of hydrazine is restricted in certain
countries, which obviously stimulates the development of nonhydrazine approaches.
Numerous organic solvents have been tested for the DSC of
kesterites, including mixtures of alcohols [27], ethylene glycol [28],
2-methoxyethanol [2932], pyridine [3335], N,N-dimethylformamide
(DMF) [36] and dimethyl sulfoxide (DMSO) [8,37]. Ethanol (boiling
point 78.4 1C) easily dissolves the metal salts, whereas propylene
glycol (1,2-propanediol, b.p. 188.2 1C) can be added to coordinate metal
cations in the solution as demonstrated for CIGS [38]. Metal salt
solutions in alcohols and diols tend to leave carbon residues after
drying and annealing steps. 2-methoxyethanol (b.p. 124 1C) is a solvent
with mixed polar and non-polar character, and can be used to dissolve
reagents with different ionicity/covalancy, e.g. ionic metal salts
together with organic additives like amines [2932]. Pyridine (b.p.
115.2 1C) and DMSO (b.p. 189 1C) are polar aprotic solvents that can
dissolve signicant quantities of thiourea or thioacetamide together
with necessary metal salts. Pyridine was used to obtain CZTS layers
[33,35] and to demonstrate a 1.44% efcient CZTS cell [34], although
the low ash point of 17 1C and unpleasant sh-like odor do not
encourage its use in laboratory without adequate protective measures.
DMF solutions of metal salts with elemental sulfur were sprayed and
post-annealed to achieve 0.51% solar cell [36]. DMSO dissolves a wide
range of organic and inorganic substances and is miscible with most
common organic solvents. DMSO functions as a Lewis base and bonds
with a metal either through the oxygen atom or sulfur atom to form
metal-organic complexes. Furthermore, it contains sulfur which is,
however, too inert to serve as a suitable sulfur source during the
sulfurization step. Thiourea is therefore added to ensure the presence
of reactive chalcogen in the precursor solution [8,37].

2.1.2. Metal and chalcogen compounds


Salts of Cu, Zn, and Sn, such as chlorides, nitrates, acetates, or
iodides, are readily available in high purity and widely used
because of their solubility in a wide range of solvents. Since many
salts are hygroscopic, hydrates are often employed to simplify the
processing and to avoid uncertainties in weight. Copper(II) salts
are preferred over copper(I) compounds because the latter are less
soluble and can undergo oxidation or disproportionation, yielding
insoluble precipitation. Regardless of the initial copper oxidation
state, it takes the Cu(I) state in the kesterite phase after the nal
annealing step with chalcogen. Sn(II) and (IV) chlorides were
chosen in all previous reports with only a few exceptions (e.g.
SnS2 was dissolved in hydrazine [25]). In addition, metal dithiocarbamate precursors were explored as metal precursors because
of their low thermal stability [35]. Concentrations of Cu, Zn, and Sn
salts are typically chosen in the approximate molar ratio 1.8: 1.2:
1.0 aiming at the Cu-poor and Zn-rich non-stoichiometric composition of the annealed CZTS(Se) layer for which highest conversion
efciencies are observed [10,39]. Loss of metals, e.g. Zn and Sn
[32], may occur during the annealing step and therefore the initial
precursor concentrations have to be adjusted adequately.
The presence of a source of chalcogen (S or Se) in the initial
precursor solution is desired in order to induce the formation of
kesterite during the annealing step. Either sulfur [25], thiourea
[8,34,37], or thioacetamide [28,33] has been employed as the
chalcogen source. Elemental sulfur or selenium is certainly the
simplest choice but their limited solubility in polar solvents, with

183

the exception of hydrazine [25] and DMF [36], somewhat limits


their usage. Thiourea is a primary choice, being soluble in numerous polar solvents and able to coordinate metal ions in solution
[40]. The chalcogen source can be added in molar excess to the
metals since its residuals are volatile at the annealing temperature.
2.1.3. Other solution components
Generally, no other solute components beside metal compounds
are required in the initial solution and they are avoided whenever
possible because of the risk of additional carbon or oxygen contamination. In practice, however, additives such as complexing agents,
binders, pH stabilizers, antioxidants, as well as surfactants and other
rheology-adjustment agents are added in order to obtain good
coating properties. Monoethanolamine (MEA) [2932,41] as a complexing agent was needed for obtaining stable solutions, in particular
when the solvent could not coordinate effectively metal cations.
MEA and other amines decompose cleanly and can also act as
antioxidants. Organic binder ethyl cellulose can be helpful when
doctor-blading precursor solutions in alcohols while preventing the
oxidation of metal species [42]. However, a carbon-rich layer was
found between the molybdenum back contact and CZTSe crust after
selenization [27]. Ammonium acetate and de-ionized water were
added to a solution in 2-methoxyethanol to increase pH and to
suppress the corrosion of the Mo-coated substrate [30].
2.2. Film coating and drying
2.2.1. Substrate
Soda-lime glass (SLG) coated with a sputtered molybdenum layer
of 0.51.0 m thickness serves as a suitable substrate for the DSC of
kesterites, allowing not only the characterization of but also the
subsequent processing and testing of solar cell devices. In analogy to
CIGS solar cells, molybdenum is exclusively employed as the back
contact in CZTS(Se) solar cells due to the benecial formation of an
interfacial layer of MoS(Se)2 that ensures a quasi-Ohmic contact
between Mo and CZTS(Se). The MoS(Se)2 layer can be quite thick,
several hundred nanometers, without deteriorating the performance
of high-efciency cells [3,8]. Recent studies indicate that the Mo/CZTSe
interface is thermodynamically unstable, which leads to the appearance of binary selenides together with MoSe2 [43]. No special surface
treatment is required for as-deposited molybdenum layers before
liquid coating. Prolonged storage, contact with water or overheating
can induce the formation of the undesired surface layer of watersoluble molybdenum oxide, which can cause reduced wetting, absorber delamination, and additional resistive losses in solar cells.
For developing new solution chemistry, bare SLG can be employed
[22,28,35], although the possible corrosion of metal contact with very
acidic [30] or basic solutions should be considered in advance. It is
interesting to note that despite the fact that almost all published
papers on non-vacuum techniques claim a high potential for roll-toroll fabrication, kesterite layers on exible substrates have only
scarcely been investigated for nanoparticle dispersions [44,45] but
not for true solutions yet.
2.2.2. Film coating
There is a large selection of coating techniques compatible with
a wide range of solutions with various rheological properties. In
other words, it should not be problematic to select an appropriate
coating technique, provided that the selected solution can be
converted into a high-quality kesterite material after the nal
thermal treatment. Relevant coating techniques already used in
laboratory research of kesterites are:
Spin-coating is the most popular laboratory technique for the
DSC of kesterites [25,29,32,37]. Spin coating is suitable for runny

184

Y.E. Romanyuk et al. / Solar Energy Materials & Solar Cells 119 (2013) 181189

solutions with viscosity of up to 1000 cp. Thin layers of typically


50300 nm can be deposited at once and repeated coatings
with intermediate drying are applied to build the necessary
layer thickness. The spin-coating of multilayers enables the
creation of compositional gradients, although there have not
been any literature reports on that matter yet. Spin coating is
not roll-to-roll compatible and it is used solely in laboratory
research because of its high material consumption, long processing times and the restriction to small-area, at substrates.
Knife coating or doctor blading is another laboratory-scale technique [8,27] that can be applied even for very viscous solutions
above 105 cp. Wet lm thicknesses of 1050 m can be obtained
by adjusting the blade-to-substrate distance. Wetting with the
precursor solution is improved when coating on heated substrates and multilayer coatings are applied to cover uncoated
spots and to obtain more homogeneous dry layers.
Spray coating is different from spray pyrolysis because it does
not aim at the deposition of the nal kesterite phase but serves
only for the coating of a wet or dry lm that is converted into
the kesterite phase during the subsequent annealing. Spray
coating at a substrate temperature of 170 1C was used to
remove the solvent and stabilizer during a fast coating of
CuZnSnS precursor but no devices were processed [41],
whereas the spray coating at 160 1C with post-annealing
yielded up to 0.51% cell [36].
Dip-coating is perhaps the simplest technique to explore new
solution chemistry [28]. This technique, however, is not preferred because both sides of the substrate get coated and the
wet layer is usually thin, on the order of 10100 nm, thus
requiring multiple coatings.
Curtain, slide, and slit coating have not been reported for
kesterite deposition yet, although these methods are suitable
for large-area coating either on rigid or exible substrates.
Their use for research purposes is not justied, because the
minimum volume of initial solution required per one coating
run or solution recipe is often substantial, on the order of 10
50 ml even for cm-size substrates, leading to undesirable
material wastage. These methods will certainly attract attention once a promising solution chemistry is identied which
leads to industrially-relevant solar cell efciencies.
Ink jet, exography and gravure are fast printing techniques and
fully compatible with roll-to-roll deposition on exible substrates. A unique feature of these methods is the possibility of
patterning. While the most popular patterning methods today
are mechanical scribing and laser ablation, patterning by
printing can reduce the number of processing steps. Printing
has already been employed for CZTSSe nanoparticles [46] and
could soon be extended to the DSC of kesterites.

2.2.3. Drying
Even though drying can be considered as a primitive, standard step
for solvent evaporation, it is found to be decisive for the layer
morphology, uniformity, and the presence of impurity elements and
phases in the nal CZTS(Se) layer. Drying starts already when the wet
lm is being coated, but it is typically performed as a separate step.
Wet layers are often dried on a hot plate for several minutes at
temperatures from 200 1C to 350 1C. On the one hand, the drying
temperature is selected to be as high as possible to maximize the
evaporation of solvent and remove carbon and, on the other hand, it is
restricted by the oxidation of the precursor material and also the
molybdenum substrate. Inert atmosphere, e.g. N2, allows increasing
the temperature up to 580 1C [37]. Additional top heating with IR
lamps can improve the layer homogeneity [27,32].
It is important to ensure a good wetting of the substrate,
e.g. Mo-coated glass, with the precursor solution in order to obtain a

homogeneous wet layer without uncoated spots and also to avoid the
appearance of thicker, inhomogeneous areas at the edgesthe socalled coffee-ring effect [47]. For achieving controlled drying, highboiling point solvents that dry slowly, such as propylene glycol (b.p.
188.2 1C) or DMSO (b.p. 189 1C), are often preferred. This implies that a
relatively high drying temperature, e.g. 300 1C for DMSO, has to be
applied to ensure complete solvent evaporation. A small amount of
surfactants for reducing surface tension, or coating on a pre-heated
substrate could also be helpful for suppressing the coffee-ring effect.
The tensile stress generated during the removal of the solvent
and organic components can give rise to lm cracking. Microscopic
cracks in the dried layer which propagate through the whole dry
layer thickness can be fatal for the solar cell absorber because they
can short the top and back contacts. This phenomenon is very
pronounced if the lm thickness exceeds a certain critical thickness which is inversely proportional to the tensile stress and the
Young's modulus of the lm [14]. The multi-layering approach
demonstrated for spin-coating e.g. [25,32,37] and knife-blading
[8,27] is an effective way to avoid microscopic cracks, and also to
improve the layer uniformity. This is because each layer is thinner
than the critical thickness, and an intermediate drying between
coatings helps to fully evaporate/decompose solvent species. If a
dried precursor undergoes a volume expansion during the last
chalcogenation step, compressive stress can even be observed in
the nal absorber layer, resulting not in cracks but in bulges [32].
One important question rarely addressed in publications on the
DCS of kesterites is the chemical state and coordination of metal
ions in the dried lm that is often called amorphous precursor
lm. It is rather impossible to obtain an amorphous precursor lm
that consists of a molecularly homogeneous mixture of initial
metal salts without any additives. Instead, a carbon-rich matrix
with coordinated metal ions, mixed metal oxides or chalcogenides,
or ideally the CZTS(Se) phase is obtained depending on the initial
solution and drying conditions. The chemical coordination of
metal cations in the dry lm should prevent possible oxidation,
hydrolysis and segregations, and should at the same time be weak
enough to decompose cleanly during the nal annealing step.
2.3. Annealing and crystallization
Annealing is the last processing step (Fig. 1) that can be
performed in either inert or reactive atmosphere. There are only
a few investigations of the DSC of kesterites which employ an inert
atmosphere, including annealing under nitrogen atmosphere [35]
or in vacuum [33,41]. These studies, however, do not prove the
quality of the obtained kesterite materials by fabricating solar
cells, but are rather focused on exploring new solution chemistries
to obtain the kesterite phase. All DSC-processed kesterite cells
reported to date were fabricated with an annealing step in sulfuror selenium-containing atmosphere, clearly indicating the necessity of the reactive atmosphere even if the initial precursor
solution already contains a chalcogen source.
The effect of the reactive chalcogen-containing atmosphere is
manifold. First, the reactive annealing can compensate the decit of
chalcogen in the dried precursor layer by increasing the S(Se)/Me ratio
to 1:1. This condition is imperative for a phase-pure material with
suitable electronic properties, since anion vacancies (VS and VSe) can
induce detrimental deep donor levels [10]. Second, the reaction with
chalcogen can lead to a volume expansion and densication during
annealing, thus eliminating the layer porosity and increasing the grain
size. Finally, the chalcogen vapor overpressure is essential for preventing a possible decomposition of the kesterite phase into binary copper
and zinc chalcogenides, and volatile SnS(Se) and S(Se) [48].
Elemental selenium [8,25,27,32,37], sulfur [34] or H2S [2931]
have been employed to date. H2S has the highest reactivity and can
be diluted with inert nitrogen down to 3% without deteriorating the

Y.E. Romanyuk et al. / Solar Energy Materials & Solar Cells 119 (2013) 181189

sulfurization performance [30], however its use is not favored due to


the highly toxic nature of the gas. Elemental sulfur and selenium are
much safer in processing although they are less reactive as they exist
in vapor as polymeric species, S212 and Se28. Since the chalcogen
vapor is supplied to the dry precursor lm from the atmosphere,
the crystallization and grain formation are more pronounced on the
surface. If the precursor layer is porous and, more importantly,
the chalcogen overpressure is signicant, the reactive chalcogen
can penetrate deep through the precursor layer and react with the
molybdenum back contact to form MoSe2 [8]. It is worth noting that
the addition of extra SnS(Se) into the annealing atmosphere, which is
claimed to be an indispensable condition for achieving stable CZTS
surfaces [48], has not been used or disclosed in any of the abovementioned publications.
The maximum annealing temperature is typically in the range of
470550 1C. The formation of the kesterite phase, however, starts at a
lower temperature and can proceed via intermediate phases [31,49].
For CZTS thin lms fabricated by a solgel sulfurization of spin-coated
metal salt precursors, CuxS is formed rst and then is removed by
sulfurization at more than 450 1C [31]. For CZTSe lms from doctorbladed metal salt precursors, the selenization starts with the fast
formation of copper selenides and proceeds with the subsequent
incorporation of Sn and Zn to yield the CZTSe phase above 280 1C [49].
These two examples indicate that the formation path strongly depends
on relative reactivity of each metal, so that Cu precursors exhibit
typically higher reaction rates than those of Sn and Zn. The reaction
pathway is thus governed by kinetic factors, such as the fast reaction
of copper with selenium at low temperatures, indicating a low
activation energy for the initial reaction and a high diffusion of copper
atoms from the precursor layer.
The reactive annealing can be performed in open or closed
reactors. The open reactor, e.g. a two-zone tube furnace with an inert
carrier gas [27,32], allows a controllable and continuous supply of
chalcogen to the precursor layer. In this case it is difcult to achieve
overpressure of volatile species, so the decomposition reactions and
the loss of metals, in particular Sn, are possible at elevated temperature [32]. Closed reactors comprise boxes made of graphite, quartz,
etc, in which the precursor layer with the source of chalcogen are
placed e.g. [8,37]. They can retain the overpressure of chalcogen and
metal-containing species, and thus enable annealing close to equilibrium conditions. As compared to conventional resistive substrate
heating, the rapid thermal processing (RTP) based on a rapid substrate
heating with IR lamps can offer additional benets such as low
thermal budget, better temperature uniformity and, importantly, the
possibility to omit or promote the formation of intermediate phases,
e.g. binary copper chalcogenides. The RTP has successfully been tested
for annealing of DSC-prepared CZTS [31] and CZTSSe [8].
3. Research highlights
We highlight two different DSC routes focusing on doctor
blading (i.e. knife-coating) to demonstrate a scalable route with
a low material consumption. The rst example is based on alcohol
solutions for pure selenide CZTSe and uses a binder for optimal
coating rheology. The second example describes mixed sulfoselenide CZTSSe based on DMSO solutions that can avoid an
additional binder.

185

ratios in the precursor solution Cu/(Zn+Sn) 0.65 and Zn/Sn 1.48


for sample A, and Cu/(Zn+Sn) 0.58 and Zn/Sn 1.78 for sample B,
respectively. A second, more viscous solution with 10 wt% ethyl
cellulose (4 mPa s, Sigma-Aldrich) dissolved in 1-pentanol (99+%,
Alfa Aesar) was prepared and mixed with the metal-containing
solution to obtain a suitable rheology for knife coating. Knife
coating of the nal solution was done on a molybdenum-coated
1 mm thick SLG. The metal precursor lm was dried under a lamp
and on a hot plate at a maximum temperature of 230 1C. The
precursor was converted into crystalline CZTSe layers in a twozone furnace in elemental selenium atmosphere at a substrate
temperature of 470 1C. Fig. 2 shows the cross-section scanning
electron microscopy (SEM) images of a dried metal precursor layer
(left) and a selenized layer with a carbon-rich interface layer
between Mo and the CZTSe absorber (right). The presence of the
carbon-rich layer is certainly undesirable, yet it does not prohibit
the solar cell operation. It was shown that the carbon-rich layer
contains signicant amounts of metals [32] and does not induce
any blocking diodes or an excessive series resistance. In contrary,
in the case of thin CIGS absorbers such a layer has even been
shown to have a benecial effect as compared to the reference Mo
back contact because it may reduce the back contact recombination and thus improve the open-circuit voltage [50].
Crack-free absorber layers with Cu-poor and Zn-rich composition
were further processed into solar cells. The currentvoltage characteristics of CZTSe solar cells (cell area 0.09 cm2) obtained from
precursor solutions A and B exhibit similar conversion efciencies,
Fig. 3. The observed cross over of dark and light JV curves and the
violation of the shifting approximation can be caused by multiple
factors known for the chalcogenide CIGS solar cells [51], such as the
photodoping of the CdS buffer layer upon illumination, the reduction
of an electron barrier at the absorber/buffer interface due to the
lling of surface acceptor states upon illumination, an illuminationdependent shunt path, or the voltage-dependent carrier collection
phenomena [51]. Sample B with a high Zn/Sn ratio (EDX of crystalline
absorber) features a higher Voc, Fig. 3. Comparing the PV parameters
of CZTSe device B with the vacuum-processed record selenium
device reported in the literature [5], all device parameters are lower:
Voc 362 vs. 377 mV (4.0%), Jsc 22.2 vs. 37.4 mA/cm2 (40.6%),
FF49.6 vs. 64.9% (23.6%) and 4.28 vs. 9.15%, where the
percentage in parentheses indicates the difference to the values of
NREL's vacuum deposited cell. Losses in the short-circuit current
density are the most striking and can be explained by an overall low
quantum efciency (QE) with increased losses for high wavelength
photons [27]. A high doping and consequently small space charge
region width and poor electronic quality could be responsible for low
carrier collection. The room temperature time-resolved photoluminescence (TR-PL) measurement was performed on a photoluminescence lifetime spectrometer Hamamatsu C12132-12 with 532 nm
excitation wavelength. The TR-PL near the band edge emission
wavelength 1310 nm (0.95 eV) of the nished solar cell device B
reveals a short minority carrier lifetime of only 0.44 ns (determined

3.1. CZTSe absorbers processed from solutions in alcohols with


organic binder
Precursor solutions were prepared by dissolving copper(II)
nitrate hemipentahydrate (99.99+ %, Sigma-Aldrich), zinc(II)
nitrate hexahydrate (98%, Sigma-Aldrich) and tin(IV) chloride
hydrate (98%, Alfa Aesar) in mixed ethanol and 1,2-propanediol.
The weight of metal salts was calculated to yield nominal metal

Fig. 2. Cross-section SEM images showing the conversion of a dry precursor lm


(left) processed from a 1,2-propanediol/ethanol solution with ethyl cellulose binder
into a crystalline CZTSe layer (right) upon annealing in selenium atmosphere. The
residual carbon-rich layer due to the presence of the organic binder is visible below
the crystalline CZTSe layer.

186

Y.E. Romanyuk et al. / Solar Energy Materials & Solar Cells 119 (2013) 181189

Fig. 3. Currentvoltage (JV) characteristics in the dark (dashed lines) and under
AM1.5 illumination (solid lines) for two CZTSe absorbers with different Zn/Sn
composition ratios as determined by EDX. The solar cell with higher Zn content
(B) features a higher Voc, whereas both cells exhibit similar conversion efciency.

Fig. 5. Cross-section SEM images showing (a) a dry precursor lm processed from a
DMSO solution and (b) a nished CZTSSe solar cell. The selenized CZTSSe absorber
exhibits a three-layer morphology with a crystalline top and bottom crust, and a
porous middle part.

Fig. 4. Time-resolved photoluminescence (TR-PL) of the nished solar cell B


described in Fig. 3. Inset shows the PL spectrum and the blue vertical line indicates
the emission energy of 0.95 eV, which is monitored during the time-resolved
measurement. The system response (gray) is also drawn. (For interpretation of the
references to color in this gure legend, the reader is referred to the web version of
this article.)

by a single exponential t to the spectrum (red curve) in Fig. 4). The


inset of Fig. 4 shows the PL signal with emission at 0.95 eV, which is
consistent with the band gap of the CZTSe absorber extracted from
the QE data [27]. A small minority carrier lifetime in combination
with a narrow space charge region width could therefore be the
limiting factors of device B.
3.2. CZTSSe absorbers processed from DMSO solutions without
binder
The second route uses a binder-free DMSO solution and is
based on the approach reported by Ki et al. that has achieved a
power conversion efciency of 4.1% for mixed sulfo-selenide
CZTSSe [37]. Owing to low carbon residues and only slight Sn
loss, high efciencies well above 7% (cell area 0.24 cm2) could be
demonstrated for a simplied process [8]. Copper(II) acetate
monohydrate (99.99%), zinc(II) chloride anhydride (99.995%), tin
(II) chloride dihydrate (99.995%) and thiourea (99%) were dissolved in DMSO (99.9%, all chemicals purchased from Sigma
Aldrich) by magnetic stirring. Prior to coating, the solution was
diluted with two parts of acetone to improve the wetting behavior.
The deposition on a SLG/Mo substrate was performed via several

Fig. 6. Raman spectra of the selenized samples measured after scratching the
sample in order to remove material to the different depths as listed. The relevant
Raman modes are indicated.

Y.E. Romanyuk et al. / Solar Energy Materials & Solar Cells 119 (2013) 181189

187

Table 1
Highest reported solar efciencies for kesterite absorbers processed by the direct solution coating. The currently highest efciency for the hybrid nanoparticle-solution
approach with hydrazine solvent is also given as a reference.
Compound

CZTS
CZTSe
CZTSSe

Solvent/stabilizer

2-Methoxyethanol/monoethanolamine
Ethanol+1,2-propanediol / ethyl cellulose
2-Methoxyethanol / monoethanolamine
DMSO
DMSO
Pyridine
Hydrazine
Hydrazinea

Coating technique

Spin-coating
Knife-coating
Spin-coating
Knife-coating
Spin-coating
Spin-coating
Spin-coating
Spin-coating

Crystallization
Atmosphere

Temperature (1C)

N2+H2S (5%)
Se in N2
Se in N2
Se in N2
Se in Ar
S and Se in vacuum
Se in inert atmosphere
S

500
470
560
540
500
500
500
4 500

Solar cell efciency (%)

Refe-rence

2.23
4.28
2.76
7.5
4.1
1.44
8.08
11.1

[30]
[27]
[32]
[8]
[37]
[34]
[25]
[3]

Hybrid solution-particle approach.

subsequent steps of doctor blading and drying at 300 1C in


nitrogen atmosphere. Afterwards, the sample was annealed in a
selenium-containing nitrogen atmosphere inside a graphite box.
The substrate was heated to 540 1C using a home-built halogenlamp-heated furnace.
The morphology of the precursor and the complete solar cell
are displayed in Fig. 5. The selenized absorber exhibits a tri-layer
structure with two layers of larger grains (top and bottom layer of
the absorber, about 550 nm and 250 nm thick, respectively) and
one layer in the middle (400 nm) with smaller grains leading to a
total layer thickness of about 1.2 m. Optimizing the morphology
to a single crystalline layer with larger grains might be the crucial
development for further improvement of the cell performance.
Since no organic stabilizers were used, there was no carbon-rich
layer as compared to Section 3.1. More precisely, the carbon
content is below the detection limit of EDX, but is determined to
be about 3 at% by secondary ion mass spectrometry. The reason for
the tri-layer structure is not yet fully understood, but could also be
related to the inuence of Mo on the balance between the
formation and decomposition of CZTS(e) as discussed previously.
Similar layered structures have also been reported both for
nanoparticle-based CZTS [52,53] and CIGS solar cells with low
carbon content [54,55], but its origin and especially the inuence
of carbon residues is under discussion.
Raman spectra were obtained on a WITec CRM 200 system with
a 532 nm laser source after scratching the sample with a scalpel in
order to remove material to different depths (Fig. 6). For depths of
0.3 m, 0.4 m and 1.2 m, the kesterite phase CZTS (A1-mode at
329 cm1) and CZTSe (A1-mode at 196 cm1, additional modes at
233 cm1 and a shoulder at 173 cm1) could be identied. Since the
mode at 233 cm1 is relatively broad, it could also superpose with
MoSe2 and ZnSe, so their existence as secondary phases in the
absorber cannot be excluded. However, after scratching through the
entire absorber layer, all expected modes corresponding to MoSe2
are visible (170 cm1, 238 cm1, 288 cm1 and less intense secondorder-modes at about 350 cm1 and 450 cm1). Furthermore, the
ZnSe phase (252 cm1) could be clearly identied. Its presence is
consistent with ndings on kesterite decomposition at the interface
with Mo [43]. Even though some optimization could help to avoid
huge amounts of MoSe2, its presence does not deteriorate the solar
cell performance with an open-circuit voltage of 404 mV, a shortcircuit current density of 29.5 mA/cm2, a ll factor of 62.8% and a
power conversion efciency of 7.5% for a cell area of 0.24 cm2. In
comparison to the results from Section 3.1, all device parameters
are higher, the largest differences are in Jsc (+32%) and FF (+27%),
which also results in a higher (+75%); the difference in Voc (+11%)
is comparatively small. However, the Voc of the alcohol approach in
Section 3.1 is remarkably high for a pure selenide device. For a

direct comparison, either an additional sulphur source for the


alcohol approach or the use of a sulphur-free precursor for the
DMSO approach would be interesting.

4. Conclusions and outlook


The DSC is a simple, solution-based method for obtaining solargrade kesterite absorbers. There are numerous options for preparing
the initial precursor solutions which can be applied on a substrate
with various coating techniques, crystallized into the desired kesterite
phase, and nally processed into thin-lm solar cells. Table 1 summarizes solar cell efciencies for the DSC-processed kesterite absorbers reported to date, primarily during the last two years. There is a
large efciency spread depending on the chosen route, and the highest
efciencies of 7.5% and 8.08% have been demonstrated for the DMSO
[8] and the hydrazine [25] routes, respectively.
It is highly anticipated that further improvements in conversion
efciency will be possible in the near future. The emphasis is expected
to be on non-hydrazine-based solvents, where novel chemistries
involving chalcogen-containing and thermally labile ligands will be
explored. Some concepts that have been successfully tested for CIGS,
for example degradable molecular precursor solutions [56], could be
transferred to the CZTS(Se) absorbers. Multilayer DSC for improving
sintering behavior and tailoring compositional gradients like in the
case for CIGS [55] should be explored.
Although the DCS and other solution techniques such as CBD, ED,
and spray pyrolysis are expected to offer lower manufacturing costs
than vacuum-based methods, this assumption still needs to be
proven. Several other factors such as nal module efciency, process
robustness, energy consumption, price of raw materials, and safety
considerations will certainly have a decisive impact on the technology deployment and nal cost. In this respect, we do not intend to
present the DSC as the most suitable method for immediate
industrial implementation at this moment, but would like to
promote it as a facile and versatile tool for kesterite research. Using
DSC-processed kesterite absorbers, critical parameters such as optimum back contact, buffer layer, absorber doping, and others can be
explored to further bolster the efciency of kesterite solar cells.

Acknowledgements
Y.E.R. acknowledges the nancial support from the Swiss
National Science Foundation, project no. PZ00P2_126435/1 under
Ambizione program. T.S. acknowledges nancial support from a
Karlsruhe School of Optics and Photonics PhD grant. Hamamatsu
(Switzerland) is acknowledged for enabling TR-PL measurements.

188

Y.E. Romanyuk et al. / Solar Energy Materials & Solar Cells 119 (2013) 181189

The authors would like to thank Theresa Magorian Friedlmeier,


Manuel Lw and Michael Rawlence for their critical comments and
revisions.

[28]

References

[29]

[1] D.B. Mitzi, O. Gunawan, T.K. Todorov, K. Wang, S. Guha, The path towards a
high-performance solution-processed kesterite solar cell, Solar Energy
Materials and Solar Cells 95 (2011) 14211436.
[2] S. Siebentritt, Why are kesterite solar cells not 20% efcient, Thin Solid Films
535 (2013) 14.
[3] T.K. Todorov, J. Tang, S. Bag, O. Gunawan, T. Gokmen, Y. Zhu, D.B. Mitzi, Beyond
11% efciency: characteristics of state-of-the-art Cu2ZnSn(S,Se)4 solar cells,
Advanced Energy Materials 3 (2012) 3438.
[4] V. Chawla, B. Clemens, Effect of composition on high-efciency CZTSSe devices
fabricated by co-sputtering of compound targets, in: 38th IEEE Photovoltaic
Specialists Conference (PVSC38), Austin, Texas, USA, 2012, pp. 29902992.
[5] I. Repins, C. Beall, N. Vora, C. DeHart, D. Kuciauskas, P. Dippo, B. To, J. Mann, W.
C. Hsu, A. Goodrich, R. Nou, Co-evaporated Cu2ZnSnSe4 lms and devices,
Solar Energy Materials and Solar Cells 101 (2012) 154159.
[6] Q. Guo, G.M. Ford, H.W. Hillhouse, R. Agrawal, A generalized and robust
method for efcient thin lm photovoltaic devices from multinary sulde
nanocrystal inks, in: Proceeding of IEEE Photovoltaic Specialists Conference 37
(2011) 35223526.
[7] S. Ahmed, K.B. Reuter, O. Gunawan, L. Guo, L.T. Romankiw, H. Deligianni, A
high efciency electrodeposited Cu2ZnSnS4 solar cell, Advanced Energy
Materials 2 (2012) 253259.
[8] Schnabel, Loew and Ahlswede, Vacuum-free fabrication of 7.5% efcient
Cu2ZnSn(S,Se)4 solar cells based on metal salt precursors, submitted to
Solar Energy Materials & Solar Cells 2013 http://dx.doi.org/10.1016/j.solmat.
2013.06.021 (PII:S0927024813003061) accepted for publication.
[9] T.K. Todorov, D.B. Mitzi, Direct liquid coating of chalcopyrite light-absorbing
layers for photovoltaic devices, European Journal of Inorganic Chemistry 1
(2010) 1728.
[10] S. Chen, A. Walsh, X.-G. Gong, S.-H. Wei, Classication of lattice defects in the
kesterite Cu2ZnSnS4 and Cu2ZnSnSe4 earth-abundant solar sell absorbers,
Advanced Materials 25 (2013) 15221539.
[11] S. Delbos, Ksterite thin lms for photovoltaics: a review, EPJ Photovoltaics 3
(2012) 35004.
[12] K. Ramasamy, M.A. Malik, P. O'Brien, Routes to copper zinc tin sulde
Cu2ZnSnS4 a potential material for solar cells, Chemical Communications 48
(2012) 57035714.
[13] M.P. Suryawanshi, G.L. Agawane, S.M. Bhosale, S.W. Shin, P.S. Patil, J.H. Kim, A.
V. Moholkar, CZTS based thin lm solar cells: a status review, Materials
Technology 28 (2013) 98109.
[14] R.W. Schwartz, M. Narayanan, Chemical solution deposition-basic principles,
in: D.B. Mitzi (Ed.), (Hrsg.), Solution Processing of Inorganic Materials,
Wiley-Interscience, 2009, pp. 3376.
[15] A. Wangperawong, J.S. King, S.M. Herron, B.P. Tran, K. Pangan-Okimoto, S.
F. Bent, Aqueous bath process for deposition of Cu2ZnSnS4 photovoltaic
absorbers, Thin Solid Films 519 (2011) 24882492.
[16] N.M. Shinde, C.D. Lokhande, J.H. Kim, J.H. Moon, Low cost and large area novel
chemical synthesis of Cu2ZnSnS4 (CZTS) thin lms, Journal of Photochemistry
and Photobiology A: Chemistry 235 (2012) 1420.
[17] S.S. Mali, P.S. Shinde, C.A. Betty, P.N. Bhosale, Y. Woo Oh, P.S. Patil, Synthesis
and characterization of Cu2ZnSnS4 thin lms by SILAR method, Journal of
Physics and Chemistry of Solids 73 (2012) 735740.
[18] N. Nakayama, K. Ito, Sprayed lms of stannite Cu2ZnSnS4, Applied Surface
Science 92 (1996) 171175.
[19] N. Kamoun, H. Bouzouita, B. Rezig, Fabrication and characterization of
Cu2ZnSnS4 thin lms deposited by spray pyrolysis technique, Thin Solid Films
515 (2007) 59495952.
[20] V.G. Rajeshmon, C. Sudha Kartha, K.P. Vijayakumar, C. Sanjeeviraja, T. Abe,
Y. Kashiwaba, Role of precursor solution in controlling the opto-electronic properties of spray pyrolysed Cu2ZnSnS4 thin lms, Solar Energy 85 (2011) 249255.
[21] S. Das, K.C. Mandal, Low-cost Cu2ZnSnS4 thin lms for large-area highefciency heterojunction solar cells, in: 38th IEEE Photovoltaic Specialists
Conference (PVSC38), Austin, Texas, USA, 2012, pp. 26682673.
[22] M.Y. Yeh, C.C. Lee, D.S. Wuu, Inuences of synthesizing temperatures on the
properties of Cu2ZnSnS4 prepared by solgel spin-coated deposition, Journal of
SolGel Science and Technology 52 (2009) 6568.
[23] D.B. Mitzi, L.L. Kosbar, C.E. Murray, M. Copel, A. Afzal, High-mobility ultrathin
semiconducting lms prepared by spin coating, Nature 428 (2004) 299303.
[24] W.-C. Hsu, B. Bob, W. Yang, C.-H. Chung, Y. Yang, Reaction pathways for the
formation of Cu2ZnSn(S,Se)4 absorber materials from liquid-phase hydrazinebased precursor inks, Energy & Environmental Science 5 (2012) 85648571.
[25] W. Yang, H.-S. Duan, B. Bob, H. Zhou, B. Lei, C.-H. Chung, S.H. Li, W.W. Hou,
Y. Yang, Novel solution processing of high-efciency earth-abundant Cu2ZnSn
(S,Se)4 solar cells, Advanced Materials 24 (2012) 63236329.
[26] H.-S. Duan, W. Yang, B. Bob, C.-J. Hsu, B. Lei, Y. Yang, The role of sulfur in
solution-processed Cu2ZnSn(S,Se)4 and its effect on defect properties,
Advanced Functional Materials 23 (2013) 14661471.
[27] C.M. Fella, A.R. Uhl, Y.E. Romanyuk, A.N. Tiwari, Cu2ZnSnSe4 absorbers
processed from solution deposited metal salt precursors under different

[30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]
[52]

selenization conditions, Physica Status Solidi A Applications and Material


Science 209 (2012) 10431048.
Y. Sun, K. Zong, H. Zheng, H. Wanga, J. Liu, H. Yan, M. Zhu, Ethylene glycolbased dipcoating route for the synthesis of Cu2ZnSnS4 thin lm, Materials
Letters 92 (2013) 195197.
K. Tanaka, M. Oonuki, N. Moritake, H. Uchiki, Cu2ZnSnS4 thin lm solar cells
prepared by non-vacuum processing, Solar Energy Materials and Solar Cells 93
(2009) 583587.
K. Maeda, K. Tanaka, Y. Fukui, H. Uchiki, Inuence of H2S concentration
on the properties of Cu2ZnSnS4 thin lms and solar cells prepared by solgel
sulfurization, Solar Energy Materials and Solar Cells 95 (2011)
28552860.
K. Maeda, K. Tanaka, Y. Nakano, H. Uchiki, Dependence on annealing
temperature of properties of Cu2ZnSnS4 thin lms prepared by solgel
sulfurization method, Japanese Journal of Applied Physics 50 (2011)
01BE1001BE15.
G.M. Ilari, C.M. Fella, A.R. Uhl, Y.E. Romanyuk, A.N. Tiwari, Cu2ZnSnSe4 solar
cell absorbers spin-coated from amine-containing ether solutions, Solar
Energy Materials and Solar Cells 104 (2012) 125130.
A. Fischereder, T. Rath, W. Haas, H. Amenitsch, J. Albering, D. Meischler,
S. Larissegger, M. Edler, R. Saf, F. Hofer, G Trimmel, Investigation of Cu2ZnSnS4
formation from metal salts and thioacetamide, Chemistry of Materials 22
(2010) 33993406.
H. Zhang, Z.Q. Li, Y.R. Chen, J.J. Li, Z. Sun, Z. Yang, S.M. Huang, Growth of
Cu2ZnSn(S,Se)4 thin lms by a simple eco-friendly solution route method,
Surface Review and Letters 19 (1250034) (2012) 17.
M. Edler, T. Rath, A. Schenk, A. Fischereder, W. Haas, M. Edler, B. Chernev,
B. Kunert, F. Hofer, R. Resel, G. Trimmel, Copper zinc tin sulde layers prepared
from solution processable metal dithiocarbamate precursors, Materials Chemistry and Physics 136 (2012) 582588.
M. Kurokawa, K. Tanaka, K. Moriya, H. Uchiki, Fabrication of threedimensional-structure solar cell with Cu2ZnSnS4, Japanese Journal of Applied
Physics 51 (2012) 10NC33.
W. Ki, H.W. Hillhouse, Earth-abundant element photovoltaics directly from
soluble precursors with high yield using a non-toxic solvent, Advanced Energy
Materials 1 (2011) 732735.
A.R. Uhl, C. Fella, A. Chirila, M.R. Kaelin, L. Karvonen, A. Weidenkaff, C.N. Borca,
D. Grolimund, Y.E. Romanyuk, A.N. Tiwari, Non-vacuum deposition of Cu(In,
Ga)Se2 absorber layers from binder free, alcohol solutions, Progress in
Photovoltaics: Research and Applications 20 (2012) 526533.
H. Katagiri, K. Jimbo, W.S. Maw, K. Oishi, M. Yamazaki, H. Araki, A. Takeuchi,
Development of CZTS-based thin lm solar cells, Thin Solid Films 517 (2009)
24552460.
J. Madarasz, P. Bombicz, M. Okuya, S. Kaneko, Thermal decomposition of
thiourea complexes of Cu(I), Zn(II),and Sn(II) chlorides as precursors for the
spray pyrolysis deposition of sulde thin lms, Solid State Ionics 141142
(2001) 439446.
H.T. Kim, D. Kim, C. Park, Temperature effects on Cu2ZnSnS4 (CZTS) lms
deposited by spraying method, Molecular Crystals & Liquid Crystals 564
(2012) 155161.
M. Kaelin, D. Rudmann, F. Kurdesau, H. Zogg, T. Meyer, A.N. Tiwari, Low-cost
CIGS solar cells by paste coating and selenization, Thin Solid Films 460 (2005)
486490.
J.J. Scragg, J.T. Wtjen, M. Edoff, T. Ericson, T. Kubart, C. Platzer-Bjrkman,
A detrimental reaction at the molybdenum back contact in Cu2ZnSn(S,Se)4
thin-lm solar cells, Journal of the American Chemical Society 134 (2012)
1933019333.
Q. Tian, X. Xu, L. Han, M. Tang, R. Zou, Z. Chen, M. Yu, J. Yang, J. Hu, Hydrophilic
Cu2ZnSnS4 nanocrystals for printing exible, low-cost and environmentally
friendly solar cells, CrystEngComm 14 (2012) 38473850.
J.V. Caspar, Y. Cao, M.S. Denny Jr., W.E. Farneth, Q. Guo, A.S. Ionkin, L.K.
Johnson, M. Lu, I. Malajovich, D. Radu, H.D. Rosenfeld, K.R. Choudhury, W. Wu,
Solution Chemical Routes to High Efciency Cu2ZnSn(S,Se)4 Thin-Film Solar
Cells, E-MRS Spring Meeting, Strasbourg, France, 2012, presentation O32.
F. Gao, S. Yamazoe, T. Maeda, K. Nakanishi, T. Wada, Structural and optical
properties of In-free Cu2ZnSn(S,Se)4 solar cell materials, Japanese Journal of
Applied Physics 51 (2012) 10NC29.
R.D. Deegan1, O. Bakajin, T.F. Dupont, G. Huber, S.R. Nagel, T.A. Witten,
Capillary ow as the cause of ring stains from dried liquid drops, Nature
389 (1997) 827829.
J.J. Scragg, T. Ericson, T. Kubart, M. Edoff, C. Platzer-Bjrkman, Chemical
insights into the instability of Cu2ZnSnS4 lms during annealing, Chemistry
of Materials 23 (2011) 46254633.
C.M. Fella, A.R. Uhl, C. Hammond, I. Hermans, Y.E. Romanyuk, A.N. Tiwari,
Formation mechanism of Cu2ZnSnSe4 absorber layers during selenization of
solution deposited metal precursors, Journal of Alloys and Compounds 567
(2013) 102106.
V. Haug, A. Quintilla, I. Klugius, E. Ahlswede, Inuence of an additional carbon
layer at the back contactabsorber interface in Cu(In,Ga)Se2 thin lm solar
cells, Thin Solid Films 519 (2011) 74647467.
R. Scheer, H.-W. Schock, Chalcogenide Photovoltaics: Physics, Technologies,
and Thin Films Devices, Wiley-VCH Verlag & Co. KGaA, Weinheim, 2011.
Y. Cao, M.S. Denny Jr., J.V. Caspar, W.E. Farneth, Q. Guo, A.S. Ionkin, L.
K. Johnson, M. Lu, I. Malajovich, D. Radu, H.D. Rosenfeld, K.R. Choudhury,
W. Wu, High-efciency solution-processed Cu2ZnSn(S,Se)4 thin-lm solar cells

Y.E. Romanyuk et al. / Solar Energy Materials & Solar Cells 119 (2013) 181189

prepared from binary and ternary nanoparticles, Journal of the American


Chemical Society 134 (2012) 1564415647.
[53] Q. Guo, G.M. Ford, W.C. Yang, H.W. Hillhouse, R. Agrawal, Enhancing the
performance of CZTSSe solar cells with Ge alloying, Solar Energy Materials and
Solar Cells 105 (2012) 132136.
[54] C. Kind, C. Feldmann, A. Quintilla, E. Ahlswede, Citrate-capped Cu11In9
nanoparticles and its use for thin-lm manufacturing of CIS solar cells,
Chemistry of Materials 23 (2011) 52695274.

189

[55] A.R. Uhl, M. Koller, A.S. Wallerand, C.M. Fella, L. Kranz, H. Hagendorfer, Y.
E. Romanyuk, A.N. Tiwari, S. Yoon, A. Weidenkaff, T.M.Friedlmeier,
E. Ahlswede, D. VanGenechten, F. Stassin, Cu(In,Ga)Se2 absorbers from stacked
nanoparticle precursor layers, Thin Solid Films 353 (2013) 138142.
[56] G. Wang, S. Wang, Y. Cui, D. Pan, A novel and versatile strategy to prepare
metal-organic molecular precursor solutions and its application in Cu(In,Ga)
(S,Se)2 solar cells, Chemistry of Materials 24 (2012) 39933997.