Open Access
Issue
EPJ Photovolt.
Volume 7, 2016
Article Number 70303
Number of page(s) 5
Section Semiconductor Thin Films
DOI https://doi.org/10.1051/epjpv/2016001
Published online 26 February 2016

© Wang et al., published by EDP Sciences, 2016

Licence Creative Commons
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1 Introduction

While the industrial production of chalcopyrite solar modules relies on vacuum-based deposition for most of the layers, there are in addition efforts to implement alternative vacuum-free methods. Among them, printing technologies are deemed attractive because of superior utilization of raw materials and the potential for high throughput roll-to-roll fabrication. Developments in this area have been concentrating mainly on the chalcopyrite absorber [1, 2]. The buffer layer, either CdS or, preferably, a Cd-free material such as Zn(O,S) [3, 4] or In2S3 [5, 6, 7, 8, 9, 10], can be prepared by dry as well as solution-based processes. The latter are typically implemented by chemical bath deposition and could be combined with a printed absorber to implement vacuum-free manufacturing of the core components of the cell. Nevertheless, printing the buffer layer could offer additional advantages in terms of material usage, in-line integration and with respect to the amount of (liquid) waste generated. In addition, certain printing technologies (such as inkjet printing [11]) provide highly localized deposition, direct patterning (without lithography) of materials in atmospheric environment and excellent raw material utilization. This is very attractive for the implementation of advanced cell concepts such as micro-concentrator cells [12].

In this work, precursor inks (molecular inks) were developed and combined with drying and annealing steps for the fabrication of In2S3 buffers. By tuning the rheological properties of the inks, both, spin-coating as well as inkjet printing could be implemented. By optimizing the composition of the precursor ink as well as the processing conditions, cells with more than 12% active area efficiency were achieved for both process variants.

thumbnail Fig. 1

Schematic diagram of In2S3 thin film formation process.

2 Experimental

The precursor-based processing of indium sulphide thin films consists of four steps (see Fig. 1): (i) formulation of metal salt precursor inks, (ii) spin coating/inkjet printing, (iii) pre-heating of deposited inks and (iv) annealing in H2S/Ar atmosphere. Firstly, a precursor ink was formulated by adding 1 mmol In(NO (Sigma-Aldrich, 99.99%) and 1.5 mmol thiourea (SC(NH, Merck, 99%) to an ethanol-based solvent (containing 9 mL ethanol and 1 mL ethylene glycol). Assuming that each thiourea molecule results in one free sulphur atom, the In/S ratio was then in nominal accordance with the stoichiometry of the In2S3 compound. However, different ratios were used in the experiments to study the influence on the cell performance. The ink was subjected to continuous stirring until it became transparent. Secondly, the precursor ink was deposited by either spin coating or inkjet printing onto substrates. For inkjet printing, a PiXDRO LP50 inkjet printer (Roth & Rau B.V.) and an industrial-grade Trident 256JetTM printhead (Trident, ITW) were used. The Trident print head has 256 piezoelectric nozzles, each with a diameter of 50 μm and a nominal drop volume of 1520 pL. The applied voltage and pulse duration of each nozzle can be tuned to obtain a well-jetted string of droplets. By increasing the resolution (dots per inch, dpi) of the printed pattern, deposited droplets can merge into a continuous film. The resolution used in our experiment was 500 dpi, in both X and Y directions. Then the sample was pre-heated in air at 150 °C for 1 min to evaporate the solvents and dry the film. Finally the pre-heated dry film was annealed in a quartz-tube furnace at 225 °C for 10 min under 5% H2S in Ar atmosphere to drive the formation of an In2S3 thin film. The quartz tube was evacuated and filled with argon three times before starting annealing.

The structure of In2S3 thin films deposited onto glass substrates was characterized by X-ray diffraction (XRD) operated in a 2θ range of 1070° on a Bruker D8-Advance X-ray diffractometer with Cu 1 radiation at an incident angle of 0.5° (grazing incidence mode), using a step size of 0.02° and step time of 5 s. The thickness and morphology of buffer layers were analyzed in a LEO 1530 GEMINI scanning electron microscope (SEM) of Zeiss. The SEM images were recorded at an acceleration voltage of 10 kV.

Test solar cells were fabricated by depositing the In2S3 buffers onto soda lime glass/Mo/absorber thin film stacks (cut down to a sample size of 2.5 × 2.5cm2) from our in-house large-area baseline. The sequential processing of Cu(In,Ga)Se2 absorbers relies on the chalcogenization of a multi-layered sputtered metal CuGa/In precursor, which is performed in nitrogen at atmospheric pressure using elemental Se [13]. Reference cells with chemical bath deposited CdS buffers were made with absorbers from the same batch. All devices were completed with an i-ZnO/ZnO:Al window layer and Ni/Al grids. Mechanical scribing was applied on each sample to define 8 solar cells with an area of 0.5 cm2. Solar cells were measured under simulated AM 1.5 illumination and standard conditions. The cells were not annealed or light-soaked before the measurements. The quantum efficiency was recorded using Xe arc and halogen lamps and a monochromator and referenced to calibrated Si and Ge solar cells. The short-circuit current densities measured with the sun simulator were slightly too high due to spectral mismatch. Unless stated otherwise, the values given below were calculated from the measured quantum efficiency (EQE) measurement using a tabulated AM 1.5 reference spectrum and refer to the active area of the devices [14].

thumbnail Fig. 2

XRD patterns of indium sulfide films prepared from precursor inks drop-cast onto glass substrates, (a) thiourea-containing precursor annealed in air, (b) thiourea-free precursor annealed in H2S and (c) thiourea-containing precursor annealed in H2S. The annealing time was 30 min for all samples.

3 Results

To confirm that the formation of In2S3 is possible in principle, precursor ink was drop-cast onto rinsed glass substrates followed by a pre-heating step in air at 150 °C for 1 min to remove residual solvents. XRD patterns of this pre-heated film showed no peak and further annealing in air for 30 min led to broad peaks with low intensity (see Fig. 2a). Instead, when sulfurized in Ar/H2S at 225 °C for 30 min, the film showed pronounced XRD peaks (Fig. 2c) corresponding to β-In2S3(JCPDS 00-025-0390) phases, which indicates the formation of crystalline film at this moderate annealing temperature. To identify the influence of thiourea, as a sulphur source, on the formation of In2S3, precursor ink without thiourea was also used for structural characterization. However, the XRD pattern of the obtained film shown in Figure 2b indicates inferior crystallinity. Compared to Figure 2c, the intensities of (311) and (400) reflections were decreased and embedded in a background presumably stemming from an amorphous phase.

thumbnail Fig. 3

SEM images of spin-coated films on glass/Mo substrates: (a) precursor film after pre-heating and (b) after annealing.

Table 1

Device parameters (best cells) of CIGSe solar cells with standard CdS buffer, spin-coated (C-03, C-04, C-05 and C-06) and inkjet-printed (C-72, C-77) In2S3 buffers. Sample C-03 was prepared from the thiourea-free precursor, while C-04, C-05 and C-06 were spin-coated from precursors with varied In-S ratio 1:1, 1:2 and 1:3 respectively. C-72 was prepared with the buffer annealed only in argon atmosphere (without H2S), while C-77 was annealed in H2S-containing atmosphere.

By adjusting the spin coating parameters, a buffer layer with a thickness of ca. 30 nm (after annealing) could be achieved. The cross-sections of spin-coated films on a glass/Mo substrate before and after annealing are shown in Figure 3. The thickness of the film after annealing was slightly reduced (from 34 nm to 28 nm in the given SEM images). Inkjet-printed films were similar in morphology on a microscopic level but slightly thinner (25 nm before and 20 nm after annealing, respectively). In terms of material utilization, this drop-on-demand technique is superior to spin coating. For instance, only 4 μL ink is needed to inkjet print a buffer layer on a standardized sample, in comparison with 100 μL ink required for spin coating.

Photovoltaic parameters of solar cells from a typical test run (best cells) are summarized in Table 1. To achieve the optimal solar cell efficiency, variations of the In/S ratio (1:1, 1:2 and 1:3) in the precursor were applied. The highest efficiency (12.8%) was obtained with the In/S ratio of 1:2 (C-05). The sample fabricated from the thiourea-free precursor (C-03) showed lower open-circuit voltage and fill factor compared to samples produced with thiourea-containing precursors. The transfer of the processing route from spin-coating to inkjet printing was successfully implemented, with the best cell efficiency of 12.2% (C-77).

thumbnail Fig. 4

Average (16 cells) and best efficiencies of solar cells buffered with CdS, spin-coated In2S3 and inkjet-printed In2S3. The error bars represent the standard deviation.

Solar cells with a buffer annealed in pure argon (C-72) showed only slightly lower efficiency. For each experimental condition, two 2.5 × 2.5 cm2 samples have been prepared identically, each containing 8 cells. The average and best efficiencies of the cells are depicted in Figure 4. The trend of the average of cell efficiencies upon different processing conditions corresponds well with the best values and all the data are within a reasonable deviation range.

The current-voltage (J-V) characteristics and external quantum efficiencies (EQE) of selected best cells from Table 1 are presented in Figure 5. The J-V curves of both cells with solution-processed In2S3 demonstrate diode properties (shunt and series resistance, diode quality factor) leading to fill factors that are even slightly higher than that of the CdS reference. The EQE measurements show a good photocurrent collection with a maximum of about 90% at a wavelength between 650 nm and 680 nm.

thumbnail Fig. 5

Comparison of dark and illuminated current-voltage (J-V) characteristics (left, current densities as measured, without correction for spectral mismatch) and external quantum efficiencies (right) of three best solar cells buffered with standard CdS, spin-coated In2S3 and inkjet-printed In2S3, respectively.

4 Discussion

The best results were achieved by annealing in a sulphur containing atmosphere but reasonable performance was also achieved using an inert, sulphur-free atmosphere (sample C-72 in Tab. 1). In contrast, adding a source of sulphur (preferably with S / In = 2) to the ink appears to be crucial, as judged from the XRD patterns as well as cell data. The influence of thiourea on crystallization, phases, and final structure of the In2S3 buffer needs to be investigated in more detail, also taking into account the possible out diffusion of sodium and copper from the absorber. A universally applicable ink should be non-corrosive (pH) and stable, i.e. should not produce precipitations even with prolonged storage. Besides the potential formation of Inx(OH), the required sulphur content may lead to Inx(OH,S)y precipitation. Certain solutions can indeed be used to grow layers (chemical bath deposition of Inx(OH,S)y[15]. It is therefore not surprising that initial efforts to formulate stable and non-corrosive water-based inks were not successful. Different metal salts and organic solvents have then been tested and lead to the simple approach used here, i.e. In(NO and thiourea in ethanol. Addition of ethylene glycol allows tuning the viscosity of the ink according to the requirements of the deposition method. This solvent system was observed to be stable under ambient temperature over months. A single coating step is sufficient to achieve a thickness suitable for buffer layers. This is different from the deposition of absorbers which requires multi-coatings with intermediate preheating steps in order to reach a micrometer-scale thickness [11]. Therefore the as-deposited buffer layer can be annealed directly without the pre-heating/drying step (confirmed in experiments not shown here). Besides annealing in inert (pure argon) atmosphere, this is another possible simplification of the process that could lower production costs. In this context, note that the annealing step used here to synthesize the compound appears to eliminate the need for post-deposition annealing of the completed device typically required for evaporated In2S3 buffers [16]. Higher annealing temperatures than those used in our work are assumed to further increase the crystallinity of the films. However, it is known from In2S3 buffers prepared by other technologies that the junction properties are degrading when the annealing temperature is too high [17] or the annealing time too long, due to Cu, Ga and In interdiffusion between the absorber and the buffer. A possible disadvantage of the process is contamination (C, O, N) stemming from the organic solvents, drying in air, decomposition of thiourea, or the In-salt. Also the use of toxic H2S, if not avoidable, may counter-balance the reduced amount of liquid waste.

Our In2S3 buffer, even in preliminary experiments, showed almost identical results to the CdS buffer in terms of homogeneity and cell parameters, suggesting that this route is stable with a wide process window. The potential with respect to different types of absorbers (including those capable of higher efficiency) needs to be verified in further work. Judging from the red response the current transport properties of the absorber are not adversely affected by the buffer layer preparation. Based on the experience with other deposition methods, the blue response with a thin In2S3 buffer is often slightly higher than that of CdS-buffered devices [10, 18]. This is not clearly observed here, partly because the blue response of the reference cell is already quite high (thin CdS). The blue response is influenced not only by the buffer layer transparency, but also by interference fringes and the collection probability of charge carriers generated close to the interface. The latter can be reduced when the there is an extended inversion zone (low absorber doping) [19]. A comparison of the optical absorption of films deposited on glass and the quantum efficiency of the cells suggests that this may play a role here but more experiments are needed to judge the blue response achievable with our printed buffers.

5 Summary

A precursor ink based process for preparing In2S3 buffer layers was proposed. By tuning the solution compositions, stable inks with suitable rheological properties were formulated. In2S3 thin films could be obtained by annealing the precursor films in a H2S-containing atmosphere at 225 °C as verified by XRD. The addition of thiourea to the precursor ink was found to improve the crystallinity of the films as well as cell performance. Successful implementation of ink deposition by spin coating or inkjet printing on CIGSe absorbers led to working solar cells with 12.8% and 12.2% efficiency respectively, which is comparable to CdS-buffered solar cells prepared from the same batch of absorbers.

Acknowledgments

The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement no. 609788 (CHEETAH project). The authors thank C. Kelch, M. Kirsch, M. Hartig for completion of the devices.

References

  1. S.M. McLeod, C.J. Hages, N.J. Carter, R. Agrawal, Prog. Photovolt.: Res. Appl. 23, 1550 (2015) [CrossRef] (In the text)
  2. T.K. Todorov, O. Gunawan, T. Gokmen, D.B. Mitzi, Prog. Photovolt.: Res. Appl. 21, 82 (2013) [CrossRef] (In the text)
  3. R. Klenk, A. Steigert, T. Rissom, D. Greiner, C.A. Kaufmann, T. Unold, M.C. Lux-Steiner, Prog. Photovolt: Res. Appl. 22, 161 (2014) [CrossRef] (In the text)
  4. T.M. Friedlmeier, P. Jackson, A. Bauer, D. Hariskos, O. Kiowski, R. Wuerz, M. Powalla, IEEE J. Photovolt. 5, 1487 (2015) [CrossRef] (In the text)
  5. C. Hönes, J. Hackenberg, S. Zweigart, A. Wachau, F. Hergert, S. Siebentritt, J. Appl. Phys. 117, 094503 (2015) [CrossRef] (In the text)
  6. R. Sáez-Araoz, J. Krammer, S. Harndt, T. Koehler, M. Krueger, P. Pistor, A. Jasenek, F. Hergert, M.C. Lux-Steiner, C.-H. Fischer, Prog. Photovolt: Res. Appl. 20, 855 (2012) [CrossRef] (In the text)
  7. T. Todorov, J. Carda, P. Escribano, A. Grimm, J. Klaer, R. Klenk, Sol. Energy Mater. Sol. Cells 92, 1274 (2008) [CrossRef] (In the text)
  8. M. Bär, N. Allsop, I. Lauermann, C.H. Fischer, Appl. Phys. Lett. 90, 132118 (2007) [CrossRef] (In the text)
  9. B. Yahmadi, N. Kamoun, R. Bennaceur, M. Mnari, M. Dachraoui, K. Abdelkrim, Thin Solid Films 473, 201 (2005) [CrossRef] (In the text)
  10. N.A. Allsop, A. Schönmann, H.J. Muffler, M. Bär, M.C. Lux-Steiner, C.H. Fischer, Prog. Photovolt: Res. Appl. 13, 607 (2005) [CrossRef] (In the text)
  11. X. Lin, J. Kavalakkatt, M.C. Lux-Steiner, A. Ennaoui, Adv. Sci. 2, 1500028 (2015) (In the text)
  12. M. Paire, L. Lombez, J.-F.O. Guillemoles, D. Lincot, J. Appl. Phys. 108, 034907 (2010) [CrossRef] (In the text)
  13. B. Rau, F. Friedrich, N. Papathanasiou, C. Schultz, B. Stannowski, B. Szyszka, R. Schlatmann, Photovolt. Int. 17, 99 (2012) (In the text)
  14. ASTM G173-03 (2012), Standard Tables for Reference Solar Spectral Irradiances: Direct Normal and Hemispherical on 37° Tilted Surface, (W.C. ASTM International, PA, 2012), www.astm.org (In the text)
  15. D. Hariskos, M. Ruckh, U. Rühle, T. Walter, H.W. Schock, J. Hedström, L. Stolt, Sol. Energy Mater. Sol. Cells 41-42, 345 (1996) [CrossRef] (In the text)
  16. P. Pistor, N. Allsop, W. Braun, R. Caballero, C. Camus, C.H. Fischer, M. Gorgoi, A. Grimm, B. Johnson, T. Kropp, I. Lauermann, S. Lehmann, H. Mönig, S. Schorr, A. Weber, R. Klenk, Phys. Stat. Sol. A 206, 1059 (2009) [CrossRef] (In the text)
  17. D. Abou-Ras, G. Kostorz, D. Hariskos, R. Menner, M. Powalla, S. Schorr, A.N. Tiwari, Thin Solid Films 517, 2792 (2009) [CrossRef] (In the text)
  18. P. Pistor, A. Grimm, D. Kieven, F. Hergert, A. Jasenek, R. Klenk, in: Proceedings 37th IEEE Photovoltaic Specialists Conference (PVSC), Seattle, 2011, p. 002808 (In the text)
  19. R. Klenk, H.W. Schock, in: Proceedings 12th European Photovoltaic Solar Energy Conference (EU PVSEC), Amsterdam, 1994, p. 1588 (In the text)

Cite this article as: Lan Wang, Xianzhong Lin, Ahmed Ennaoui, Christian Wolf, Martha Ch. Lux-Steiner, Reiner Klenk, Solution-processed In2S3 buffer layer for chalcopyrite thin film solar cells, EPJ Photovoltaics 7, 70303 (2016).

All Tables

Table 1

Device parameters (best cells) of CIGSe solar cells with standard CdS buffer, spin-coated (C-03, C-04, C-05 and C-06) and inkjet-printed (C-72, C-77) In2S3 buffers. Sample C-03 was prepared from the thiourea-free precursor, while C-04, C-05 and C-06 were spin-coated from precursors with varied In-S ratio 1:1, 1:2 and 1:3 respectively. C-72 was prepared with the buffer annealed only in argon atmosphere (without H2S), while C-77 was annealed in H2S-containing atmosphere.

All Figures

thumbnail Fig. 1

Schematic diagram of In2S3 thin film formation process.

In the text
thumbnail Fig. 2

XRD patterns of indium sulfide films prepared from precursor inks drop-cast onto glass substrates, (a) thiourea-containing precursor annealed in air, (b) thiourea-free precursor annealed in H2S and (c) thiourea-containing precursor annealed in H2S. The annealing time was 30 min for all samples.

In the text
thumbnail Fig. 3

SEM images of spin-coated films on glass/Mo substrates: (a) precursor film after pre-heating and (b) after annealing.

In the text
thumbnail Fig. 4

Average (16 cells) and best efficiencies of solar cells buffered with CdS, spin-coated In2S3 and inkjet-printed In2S3. The error bars represent the standard deviation.

In the text
thumbnail Fig. 5

Comparison of dark and illuminated current-voltage (J-V) characteristics (left, current densities as measured, without correction for spectral mismatch) and external quantum efficiencies (right) of three best solar cells buffered with standard CdS, spin-coated In2S3 and inkjet-printed In2S3, respectively.

In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.