An innovative approach for fabrication of Cu 2 ZnSnSe 4 absorber layers using solutions of elemental metal powders

Aninnovative approach hasbeendemonstratedfor thedeposition of Cu 2 ZnSnSe 4 (CZTSe)absorberlayers.Using a unique,safe solvent combination, movingaway from hydrazine, elemental Cu, Zn, Sn and S/Secan be fully dis-solvedatambientconditions,withthecompositioneasilycontrolled.Thepreparationandthespraydepositionof these solutions are performed in air, allowing a quick, easy and inexpensive process. Upon selenisation, large crystalsareobservedwhichareidenti ﬁ edastheCZTSekesteritephaseusingX-raydiffractionandRaman,thelat-ter showing no distinctive signs of any binary or ternary secondary phases. Using this method, CZTSe absorber layers have been prepared for use in thin- ﬁ lm solar cells reaching ef ﬁ ciency of 3.2%. With further device optimisation, improved device performance will be achieved. © Authors.


Introduction
Cu 2 ZnSn(S,Se) 4 (CZTSSe) compounds have emerged as an ideal replacement for current thin-film technology (such as CdTe or CIGS), due to their suitable band gap and high absorption coefficient, in addition to incorporating less toxic and more earth-abundant materials. However, improvements in efficiency are required before CZTSSe becomes a viable replacement.
There are numerous methods to prepare CZTSSe thin-films, including but not limited to: sputtering, co-evaporation, electrodeposition and several solution-processed methods [1]. This study will look into the method reported by Mitzi et al. [2], where the required elements are directly dissolved, without the use of salts which may incorporate detrimental impurities in the film. After depositing, the absorber layers are intermediately dried and the solvents decompose cleanly, forming a CZTSSe layer that is free from impurities. Direct solution processing is an ideal way to prepare CZTSSe, as the composition can be controlled easily, allowing band gap control. CZTSSe thin films can be deposited by many techniques including doctor blading, spin coating and spray coating. These techniques can be used to deposit the samples quickly, without the requirement of high vacuum equipment. However, solution processing of elemental metals and chalcogenides is typically challenging, due to their insolubility in most common solvents.
There have only been a few solvents that were implemented in the direct solution processing of CZTS absorber layers using elemental metals or chalcogenides. Mitzi et al. dissolved metal chalcogenides in hydrazine, for depositing CIGS [3] and later CZTSSe [2]. The solutions were deposited by spin coating and hydrazine was removed with a short drying step, leaving a uniform layer of CIGS and CZTS. Whilst the method proposed by Mitzi et al. [2,3] is effective, hydrazine is highly toxic and explosive, making the process hard to scale up. Therefore, an ideal approach should follow the main principles described by Mitzi [2,3], but using an alternative solvent to hydrazine. The alternative solvent needs to be more environmentally friendly than hydrazine and also should be able to dissolve metal chalcogenides and/or metals and then decompose easily forming a clean absorber layer.
A combination of 1,2-ethanedithiol and ethylenediamine was shown by Webber and Brutchey to dissolve a series of metal sulphides including As 2 S 3, Sb 2 S 3 and Bi 2 S 3 along with their selenide and telluride counterparts [4]. It was shown that phase-pure materials could be obtained after an annealing step at 350°C. The same group has also developed other methods which could be used for fabrication of thin-film photovoltaics. These include the deposition of SnS [5,6], dissolution of selenium and tellurium [7], amongst others [8].
Previous work by our group demonstrated that Cu 2 S, In 2 S 3 and their selenide counterparts can be dissolved in the 1,2-ethanedithiol/

Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / t s f ethylenediamine solvent combination [9]. The 1,2-ethanedithiol/ ethylenediamine solvent combination was later used for the deposition of CuInSe 2 solar cells reaching an efficiency of 8.0% [10]. The solvent combination was later shown by Zhao et al. [11] and Guo et al. [12] to dissolve elemental powders of metals for CIGS and CZTSSe deposition, reaching efficiencies of 9.5% and 6.4%, respectively. Furthermore, a move away from these two solvents was started by Yang et al. [13] using a mixture of thioglycolic acid and ethanolamine to prepare CZTSe precursor solutions, reaching an efficiency of 8.0%. Despite being a better alternative than hydrazine, ethylenediamine and 1,2ethanedithiol are still relatively toxic solvents, with the dithiol having a strong odor. As these solvents are still not ideal for large scale, there is still a need to find an alternative solvent combination.
In this work, we report an alternative solvent combination of cysteamine and ethanolamine, which has shown to be a suitable replacement. Here, cysteamine is used as the thiol source, and as a solid, it is easier and safer to handle, as well as being significantly less odorous than liquid thiol sources. Cu, Zn, Sn and Se powders have been dissolved in this solvent combination, and the solutions were deposited by a simple chromatography atomiser. After selenising in a graphite box, a crystalline layer of CZTSe was formed. With this method, an efficiency of 3.2% has been reached. Promising results were obtained from the first selenisation trials and show that improvements can be made through device processing optimisation.

Preparation of CZTSe precursor solution
CZTSe solutions were prepared using elemental metal powders. To prepare a solution with ratios of Cu/(Zn + Sn) = 0.8 and Zn/Sn = 1.2, 1.76 mmol of Cu, 1.00 mmol of Sn, 1.20 mmol of Zn and 4.00 mmol of Se were added to a vial containing 1 g of cysteamine and 10 ml of ethanolamine, which corresponds to an~10:1 amine/thiol ratio by weight. The mixture was magnetically stirred at room temperature for at least 24 h to allow the metals to fully dissolve and then diluted with dimethyl sulphoxide (DMSO) in a 1:1 ratio before deposition. The metals have been shown to be better soluble in an~10:1 mixture of amine source to thiol source [4] and therefore this ratio was utilised in the current study. Different thiol/amine ratios were found to show poor dissolution of the metals.

Absorber formation
The solution was sprayed using a chromatography atomiser onto a preheated molybdenum-coated soda lime glass (SLG) substrate at 310°C. After each deposited layer, the film was left for 2 min to dry at the same temperature. The temperature of 310°C was chosen as a balance between removing the solvents and preventing oxidation of the underlying Mo substrate. This method was repeated until the desired thickness was reached, with 6 layers giving an average thickness of 2.0 μm.
After the sample was deposited and dried, a selenisation step was carried out to promote grain growth within the absorber layer. The samples were placed inside a graphite box with~160 mg of selenium pellets, which was placed in a tube furnace and heated to~570°C. Although the temperature set point was 570°C, the maximum temperature reached, measured by a thermocouple placed under the sample, was 555°C. The mass of the Se used was determined through preliminary experiments, as too much Se lead to excessive delamination of the film, whilst using less lead to insufficient recrystallisation. The tube furnace was sealed and the graphite box was kept under a nitrogen atmosphere with an initial pressure of 40 kPa, and varied selenisation duration.

Fabrication of CZTSe device
After selenisation, an~70 nm cadmium sulphide (CdS) layer was deposited by chemical bath deposition, followed by an~80 nm intrinsic zinc oxide (i-ZnO) and~500 nm aluminium doped zinc oxide (AZO) layer, deposited by RF sputtering. Finally, the samples were scribed to give an active area of~0.25 cm 2 .

Characterisation
The thickness of the absorber layer was measured using a surface profilometer by Ambios (model XP-2). The compositional analyses of the CZTSe thin films was performed using a Bruker Quantax 70 energy dispersive X-ray spectroscopy (EDS), built into a Hitachi TM3030 table top scanning electron microscope (SEM). To analyse the crystallinity of the deposited CZTSe absorber layer, X-ray diffraction (XRD) patterns were taken, using a D2 phaser X-ray diffractometer by Bruker. The X-ray diffractometer was equipped with a Cu-Kα X-ray source and a lynxeye detector. The anti-scatter slit was set at 3 mm and the divergence slit was set at 1 mm. The sample was rotated at 15 rpm. A LabRam HR by Horiba Jobin-Yvon was used for Raman spectroscopy analysis of the samples, using an excitation laser with a wavelength of 633 nm. The high resolution surface and cross-sectional images were obtained using a JEOL 7800F field emission gun scanning electron microscope (FEGSEM). The aperture size was 30 μm with an operating voltage of 5 kV. Photocurrent densityvoltage (J-V) curves were recorded under the standard AM1.5 illumination (100 mW cm −2 ) with a Keithley 2440 5 A source meter.

Characterisation of the CZTSe absorber layer
The following results were obtained using the starting solution with a Cu/(Zn + Sn) of 0.8 and Zn/Sn ratio of 1.2. The composition can be easily changed to control the final stoichiometry of the CZTSe absorber layer by simply altering the starting concentration of elemental metals. The solutions prepared are optically transparent and remain stable for months (Fig. 1). From EDS analysis, the as-deposited samples gave composition ratios of Cu/(Zn + Sn) = 0.77-0.88 and Zn/Sn = 1.15-1.29. Notably there is S present in the as-deposited film with a Se/S ratio of 3.5. The remaining S that is present in the solution comes from the solid cysteamine and the diluting solvent DMSO, showing that S is introduced via the solvent and is still present in the film after drying at 310°C. However, after selenisation the Se/S ratio increased to~36.5, showing that during the selenisation process S is quite significantly displaced by Se.
To identify the growth process of the CZTSe films, XRD measurements were performed for films selenised for 10 to 50 min (Fig. 2 A). The whole process took 26 min to reach the maximum substrate temperature of 555°C, therefore selenising for 30, 40, and 50 min corresponds to a 4, 14 and 24 min dwell time at the desired temperature, respectively. In the as-deposited films (hence, 0 min selenisation), three broad peaks can be observed, that are associated with the CZTSe structure ( Raman spectroscopy (Fig. 2B) was performed to distinguish between potential binary and ternary secondary phases, such as Cu x Se, ZnSe and Cu 2 SnSe 3 , which may be present when preparing CZTSe. Three distinctive peaks can be observed at wavelengths 170 cm − 1 , 192 cm −1 and 232 cm −1 which can be assigned to the CZTSe kesterite phase. There are no observed peaks at 180 cm − 1 , 253 cm −1 and 260 cm −1 , respectively for Cu 2 SnSe 3 , ZnSe and CuSe [14].

Film morphology
The film morphology was investigated using SEM imaging (Fig. 3). From the surface image, crystals cover the entire surface with no visible cracks or voids. The surface is, however, wavy causing slight variations  in thickness throughout the absorber layer, which can be explained by the volume contraction of the solvent upon evaporation during deposition. The deposition process needs further investigation to improve the surface morphology. This may be achieved by optimising the drying temperature, drying time and absorber layer thickness.
Cross-sectional images show three distinctive layers for a 50 min selenisation step, from bottom to top: (1) MoSe 2 with a thickness of 2.0 μm; (2) a "sandwiched" structured absorber layer of~2.0 μm consisting of~0.5 μm crystals growing from the bottom and~1.2 μm growing typically top down, sandwiching~0.3 μm of uncrystallised absorber layer; (3) CdS, i-ZnO and AZO to complete the device. After 20 min of selenisation the top down crystals were~0.7 μm with a 1.0 μm of uncrystallised absorber layer. The thick layer of MoSe 2 has been shown to have a significant effect on device performance, with increasing thickness of MoSe 2 increasing the series resistance of the device [15]. The MoSe 2 thickness can be controlled by altering the selenisation conditions or by introducing a barrier layer between the absorber layer and the Mo [15]. Furthermore, it has been reported previously that the "sandwiched" absorber layer is likely formed due to the high porosity of the absorber layer. This allows Se vapour to rapidly diffuse to the bottom and allows crystal growth from the bottom-up [16]. This could also explain why the Mo selenises significantly at a similar time to the crystal growth from the top, as shown by XRD. To solve this problem, a solution soaking step in a cation precursor solution was implemented before selenisation to reduce the porosity of the film [16]. However, with these underlying problems, reasonable efficiencies are still producible, as discussed in Section 3.3.

J-V analysis
Devices have been prepared and measured with a notable improvement in performance with shorter selenisation times. From comparing the average device parameters prepared after different selenisation times (Table 1), a noticeable drop in all parameters can be observed with increasing selenisation time. Although better crystal growth and a thinner uncrystallised layer are obtained after longer selenisation times, the shortest selenisation time (20 min) resulted in the cell with the highest efficiency of 3.2% (Fig. 4). After 20 min, the selenisation temperature reached only 550°C for a very short period, before the furnace was turned off and allowed to cool naturally. This improvement with a relatively short selenisation time may be related to thinner MoSe 2 forming during shorter selenisations, or excess Sn loss during the prolonged annealing process. Sn loss can lead to formation of Cu and Zn binary phases, causing a deterioration in the series resistance and in turn the efficiency [17]. Further work will be carried out to clarify the reasons for this loss in performance.

Conclusion
A safe and simple approach has been demonstrated to dissolve elemental powders of metals at room temperature which have several months of shelf life. The metal composition can easily be altered to control the final composition of the film. Spray coating was chosen as a quick method to deposit a thin film of CZTSe precursor, and after selenisation a highly crystalline CZTSe absorber layer was formed, with a device efficiency of 3.2%. Optimisation of individual processes within this method can lead to further improvements in device performance. Improvements in the Mo-coated SLG, deposition and selenisation process can lead to a thinner MoSe 2 layer, for example by using a barrier layer [15]. Crystallisation may be improved by further optimisation of the selenisation parameters which include temperature, pressure and Se mass. Finally, the band gap for the current devices was previously calculated to be~1.0 eV. The incorporation of sulphur into the initial solution and the absorber layer is anticipated to lead to a more favourable band gap.