Preparation of stoichiometric CuInS2 surfaces—an XPS and UPS study

Preparation of stoichiometric CuInS2 surfaces—an XPS and UPS study

Thin Solid Films 431 – 432 (2003) 312–316 Preparation of stoichiometric CuInS2 surfaces—an XPS and UPS study ¨ K. Muller*, S. Milko, D. Schmeißer ¨ C...

130KB Sizes 1 Downloads 61 Views

Thin Solid Films 431 – 432 (2003) 312–316

Preparation of stoichiometric CuInS2 surfaces—an XPS and UPS study ¨ K. Muller*, S. Milko, D. Schmeißer ¨ Cottbus, Angewandte Physik-Sensorik, P.O. Box 101344, 03013 Cottbus, Germany Brandenburgische Technische Universitat

Abstract CuInS2 films are investigated with photoelectron spectroscopy. We find significant deviation from the stoichiometric composition depending on the preparation on the films. For Cu-rich films (start composition) we show that by Ar sputtering and annealing the surface composition varies along a binary cut of the Cu–In–S ternary phase diagram. We also present a novel technique which brings the surface composition close to the stoichiometric stability range of the CuInS2 phase. The technique used the evaporation of Cu on the sputter cleaned surface. 䊚 2003 Elsevier Science B.V. All rights reserved. Keywords: Stoichiometric CuInS2-films; Annealing; XPS; UPS

2. Experimental

1. Introduction CuInS2-films tend to be inhomogeneous in composition and stoichiometry due to the high mobility of Cu, even at room temperature. The range of stability of the phase CuInS2 itself is relatively large caused by low formation enthalpies of vacancies, interstitials or antisites w1x. However, this development leads to differences from the ideal elemental composition of 1:1:2. Furthermore, the high mobility of Cu promotes the chemical diffusion of Cu and can change the composition of CuInS2 and the type of conductivity (p–n) at temperatures below 100 8C w2x. In addition, films of Cu–In–S tends to build phase segregation like CuS (mineral name: covellite) at the surfaces and in between of grains w3x or in the bulk of the crystallites w4x. The formation of binary phases is extremely critical for solar cells because of their metallic conductivity w5x. These phases can be removed by KCN etching or by an optimised electrochemical process w6,7x. We have investigated the effect of an in situ heat treatment of samples prepared in the Cu-rich regime with additional amounts of Cu, deposited on the surface of the CuInS2 films and succeeded to prepare surfaces with almost ideal elemental composition. *Corresponding author. Tel.: q49-355694067; fax: 355694068. ¨ E-mail address: [email protected] (K. Muller).

q49-

We used CuInS2 layers deposited on a Mo film on top of a soda-lime-glass sample, prepared by sputter deposition of a CuyIn alloy and following reactive annealing in S2-atmosphere at 500 8C, according to Ref. w8x. These samples were mounted in a Ta holder with electrical contacts and the Mo layer was used as resistive heater. The temperature was measured with a pyrometer. With this set-up, in situ measurements of surface reactions were possible. Our experiments were performed as follows: the surface of the films was cleaned in situ by argonbombardment to minimise the oxidised surface and the C impurities as well as to reduce the content of segregated phases (CuS, Cu2S). Cu or In was evaporated onto as prepared substrates. We used calibrated Knudsen-cells to obtain a surface coverage ranging between 1 and 10 monolayer (ML) equivalent. After the metal deposition the samples were annealed at elevated temperatures (up to 550 8C) for 10 min. At every step (argon-bombardment, evaporation, heating) the spectroscopic characterisation by XPS and UPS was performed at room temperature in an Omicron-UHV-System with hemispherical analyser w9x. Spectra are taken with Al Ka (1486.6 eV) excitation. Elemental ratios were calculated from the XPS intensities, modified with the specific atomic sensitivity factors w10x. We start to describe the variations of the elemental ratios of CuInS2 samples, prepared by different methods upon argon-bombardment. These CuInS2 samples were:

0040-6090/03/$ - see front matter 䊚 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0040-6090(03)00150-0

¨ K. Muller et al. / Thin Solid Films 431 – 432 (2003) 312–316

313

Fig. 1. Ratio of the elemental concentrations of Cu and In as determined from Cu 2pyIn 3d-intensities of three samples, prepared by (a) electrochemical deposition w11x; (b) argon-bombardment w8x; (c) coevaporation w12x, see text.

Fig. 3. (a) CuyIn-ratio of a copper-rich sample as determined from the XPS data (Cu 2pyIn 3d-intensities) vs. annealing cycle and preparation step. After 15 min Ar bombardment (step 0) the annealingsteps 150–250–300–400–550 8C follows for 10 min, respectively. The cycles in detail: Cycle 1: 1 ML Cu, annealing (Steps 1–6); Cycle 2: 1 ML Cu, annealing (Steps 6–11); Cycle 3: 1 ML Cu, annealing (Steps 11–13, 400–550 8C). (b) HeII–UPS spectra of the sample after the evaporation of Cu and after the total annealing process (preparation step 13).

A. fabricated on Cu tape by electrochemical deposition of In and w11x B. prepared by sputtering of precursors of CuyIn and a following reactive annealing in a S2-atmosphere w8x C. prepared by co-evaporation w12x. 3. Results

Fig. 2. (a) Elemental-ratios of a copper-rich sample of charge B as determined from the XPS intensities (Cu 2p, In 3d, S 2p) after the individual annealing step, without additional evaporated Cu or In; (b) HeII excited valence band spectra of CuInS2, before and after preparation.

Fig. 1 shows the elemental ratio of Cu and In, calculated from the XPS intensities of the Cu 2d and In 3d lines as a function of the duration of argon-bombardment without additional Cu or In. We notice that the samples, prepared like b appear to be Cu-rich while a and c have an In rich surface composition. It is remarkable that the observed surface compositions are way off

314

¨ K. Muller et al. / Thin Solid Films 431 – 432 (2003) 312–316

Fig. 4. HeI- and XPS-spectra of a stoichiometric CuInS2 surface (after preparation step 13).

from the ideal values and that stoichiometric ratios were never observed on as deposited samples. The following experiments were done only with samples prepared like b, i.e. in the Cu-rich regime with a CuyIn-ratio of approximately 5 in average. These samples were annealed in several steps. In Fig. 2a we notice that the CuyIn-ratio is reduced from the high value of approximately 10 before sputtering to reach a value of approximately 5 after the annealing steps. That value could not be reduced by prolonged annealing, higher temperatures are avoided because of a damaging of the sample. A comparison of UPS (HeII)-spectra before and after annealing is shown in Fig. 2b. We notice an increase of the intensity of the In 4d line due to the annealing process. The spectrum reaches up to the Fermi level which indicates the existence of a metallic CuS phase at the surface. A rather unexpected finding is observed after the deposition of Cu on such surfaces. After cleaning the surfaces by soft Ar sputtering we have deposited Cu and annealed these surfaces to temperatures up to 550 8C. In Fig. 3a we summarise the progress of an annealing

process after evaporation of Cu in several steps. Here we find that the elemental ratios develop into almost stoichiometric values. After Cu deposition and annealing we finally find the following ratios of CuyIn, CuyS and InyS (in brackets, the theoretical value): 1.32 (1); 0.6 (0.5) and 0.45 (0.5), respectively. We notice that the surface reaction of the additional amount of Cu promotes a reaction which leads to a surface composition close to stoichiometry. In the UPS HeII spectra (Fig. 3b) this leads to the development of the In emission which is underrepresented before the annealing process. At the same time the metallic emission at the Fermi energy disappears and we find the valence band maximum of approximately 0.5 eV below the Fermi level. However, the VBM does not show a sharp onset but a rather soft tailing off, which indicates a rest of CuS (covellite) to remain at the surface of the films w13x. In Fig. 4 we have compiled the characteristic data obtained for the stoichiometric CuInS2 film after our preparation. The HeI spectrum shows the characteristic features w13,14x for CuInS2 surfaces marked as the valence band features A, B, C, D at y1.5, y3, y5 and

¨ K. Muller et al. / Thin Solid Films 431 – 432 (2003) 312–316

Fig. 5. (a) Elemental ratios Cu 2pyIn 3d, Cu 2pyS 2p, In 3dyS 2p as determined from the Al Ka–XPS data of the above mentioned annealing process (Fig. 4). The lines give the predicted ratios at the binary cut Cu2S–In2S3. The hatched area is the range of homogeneity for CuInS2 thin-films w8x. (b) The ternary phase diagram of Cu–In–S, after w9x. Indicated are the binary cuts of Cu2S–In2S3, CuS–InS and our elemental ratios.

y8 eV, respectively. In addition, we show the photoelectron spectra covering the range around the Cu 2p, the In 3d, as well as the range between y280 eV and the valence band region for the surface after the total preparation procedure (step 13). 4. Discussion First we like to stress that the observed elemental ratios are really within the stability range of the chalcopyrite phase. If we assume that approximately 10% of Cu is segregated as covellite at the surface, the XPSelemental ratios (Cu 2p, In 3d, S 2p) can be corrected for that contribution. This is done in Fig. 5a. Here we also plotted the range of stability of the chalcopyrite phase as taken out for thin films (hatched area) w8x. Our experiments indicated that this range is really reached after repeatedly exposing the surface to Cu and subsequent annealing steps.

315

We now discuss the variations in the elemental ratios upon sputtering and upon annealing. We find that these variations are not arbitrarily but there is a systematic dependence. This becomes evident when we plot the observed experimental ratios vs. the mole fraction x. This quantity represents a binary cut within the ternary phase diagram shown in Fig. 5b. If we plot our experimental ratios vs. the mole fraction of Cu2S–In2S3 we notice that there is an excellent agreement between our data and the expected CuyIn variations along this binary cut of the ternary phase diagram w15x. This behaviour is unexpected as the sputtering should not favour crystallographic preferences but is rather expected to cause an amorphous surface layer with a random composition. Another surprise comes from the fact that this agreement is only achieved when the Cu2S–In2S3 binary cut is used. The experimental data show significant deviations from the expected curves for other binary cuts. We always find significant deviations from at least one of the experimentally determined intensity ratios (Cu 2pyIn 3d, Cu 2pyS 2p, In 4dyS 2p). However, there is one exception. A reasonable agreement is also found for the CuS–InS line which is indicated in the ternary phase diagram in Fig. 5b, too. In particular we have to state that a cut along the Cu2S–InS line must be excluded due to significant differences between predicted and measured elemental ratios. It should be noted that this binary cut along the Cu2S–InS line (not shown in Fig. 5b) is discussed as a possible reaction path for epitaxial films w16x. This is in contrast to our findings for the polycrystalline CuInS2 sample described above. The comparison between the predicted ratios based on the cuts and the experimental values shows that the development of composition can only be described with the cuts Cu2S–In2S3 or CuS– InS. 5. Summary Our XPS–UPS analyses of CuInS2 thin films showed that starting with a Cu-rich polycrystalline sample we are able to prepare a stoichiometric surface. The procedure involves the evaporation of Cu in the ML range followed by a controlled annealing. We find that the variations in the elemental composition proceed from the Cu-rich side via the Cu2S–In2S3 binary cut of the ternary phase diagram. Acknowledgments We thank R. Scheer (HMI) and O. Tober (Odersun) for providing CuInS2 samples and H.J. Lewerenz for helpful discussions. The experimental assistance of J. Burkov, P. Hoffmann, and G. Beuckert as well as the help of the BESSY staff is acknowledged. This work was supported by DFG under grant no. GEP-SCHM 745y3.

316

¨ K. Muller et al. / Thin Solid Films 431 – 432 (2003) 312–316

References w1x H. Moller, ¨ Semiconductors for Solar Cells, Artech House, Norwood, 1993. w2x M. Kleinfeld, H.D. Wiemhofer, ¨ B. Bunsenges, Phys. Chem. 90 (1986) 711. w3x R. Scheer, H.J. Lewerenz, J. Vac. Sci. Technol. A 13 (1995) 1924. w4x V. Nadenau, D. Hariskos, H.W. Schock, M. Kreijci, F. Haug, A. Tiwari, H. Zogg, G. Kostoroz, J. Appl. Phys. 85 (1999) 534. w5x B. Blachnik, A. Muller, ¨ Thermochim. Acta 361 (2000) 31. w6x M. Aggour, H.J. Lewerenz, J. Klaer, U. Storkel, ¨ Electrochem. Solid-State Lett. 3 (2000) 399. w7x M. Aggour, U. Storkel, ¨ C. Murrell, S.A. Campbell, H. Jungblut, P. Hoffmann, R. Mikalo, D. Schmeißer, H.J. Lewerenz, Thin Solid Films 403–404 (2002) 57.

w8x C. Dziong, H. Metzner, S. Hessler, H.E. Mahnke, Thin Solid Films 299 (1997) 38. w9x P. Hoffmann, R.P. Mikalo, D. Schmeißer, Solid-State Electron. 44 (2000) 837. w10x C. Wagner, W. Riggs, L. Davies, J. Moulder, G. Mullenberg (Eds.), Handbook of X-ray Photoelectron Spectroscopy, Perkin Elmer Corporation, Physical Electronics Division, Minnesota, 1978. w11x I. Konovalov, O. Tober, M. Winkler, K. Otte, Sol. Energy Mater. Sol. Cells 67 (2001) 49. w12x R. Scheer, M. Alt, I. Luck, H.J. Lewerenz, Sol. Energy Mater. Sol. Cells 49 (1997) 423. w13x H.J. Lewerenz, H. Jungblut, Photovoltaik, Springer, 1995, p. 185. w14x R. Hunger, C. Pettenkofer, R. Scheer, Surf. Sci. 477 (2001) 76–93. w15x J. Binsma, L. Giling, J. Bloem, J. Cryst. Growth 50 (1980) 127. w16x W. Calvet, Ph.D. Thesis, BTU Cottbus, 2002.