Hahn- Meitner-Institut, Solare Energetik
Glienicker Str. 100, 14109 Berlin
About 20 years ago pyrite (FeS2) was proposed as a promising candidate for perspective usage as photovoltaic absorber material for thin film solar cells [1]. Among its physical properties -the very high absorption coefficient and a suitable energy bandgap (Eg≈0.95 eV) for photovoltaic energy conversion- also its nontoxicity and its composition from abundant elements were considered as particular advantages of pyrite.
However, the promisses could not be fulfilled. Though the quantum efficiencies and the photocurrents at AM1.5 were reasonably high for single crystalline pyrite samples, the open circuit voltages never exceeded about 200 mV at room temperature [2, 3], much to low compared to the band gap of pyrite. The highest efficiency reported so far is about 2.8 % at AM1.5 [2].
Recently, a numerical simulation with the program PC-1D was performed by Altermatt et al.[4] that used conservative measured parameters of pyrite (Eg=0.8 eV, µe≤10 cm2/Vs). These simulations (see Fig.1) confirmed earlier expectations concerning a pyrite solar cell. In a diffusion cell configuration, efficiencies between 10 and 20 % should be possible, provided that the cell is governed by Shockley-Read-Hall (SHR) recombination.
| |
In the present paper a series of recent results about the homogeneity range of pyrite, the doping of thin films and some surface properties are summarized. Furthermore, a model is discussed that could explain the difficulties to prepare a good heterojunction on a pyrite surface.
In order to explain the low photovoltages of pyrite already in 1989 Alonso-Vante et al. [5] proposed a significant sulpur deficiency in pyrite, i.e., a wide homogeneity range in the atomic% region, as a possible explanation. Theoretical arguments and a critical review of stoichiometry measurements reported in literature [6] have shown that pyrite should have a homogeneity range that is very narrow («1‰). With respect to intrinsic defects pyrite behaves like other compound semiconductors (GaAs, CdS or CdTe) that are used in device structures. To test the theoretical conclusions mentioned above stoichiometry measurements on MOCVD pyrite thin films have been performed in our group [7]. We could show, that, whenever a stoichiometry x<2.0 in FeSx was detected by Rutherford backscattering (RBS), a phase mixture of pyrite and pyrrhotite (Fe1-xS) existed in the films. While RBS is accurate within ±0.5 at%, measurements of the magnetic moments of the MOCVD films are even more sensitive for low concentrations of the ferrimagnetic pyrrhotite phase. These experiments showed that pyrite is a stoichiometric compound with a homogeneity range far below 0.5at%. Furthermore, we could show that secondary phases (especially pyrrhotite, but also marcasite-the metastable orthorhombic modification of FeS2) play an important role when preparing pyrite films.
Another problem is the doping of pyrite thin films. All pyrite films reported in literature were p-type. We prepared n-type pyrite films by cobalt doping with carrier concentrations of 9 x1019 to 4 x1020 cm-3 in a MOCVD process [8](Fig.2). From the change of the sign of the Seebeck coefficient an intrinsic hole carrier concentration of the p-type films of about 1 x1019 cm-3 was estimated. By fitting the temperature dependent Hall mobilities, the grain barrier heights were calculated to be between 37 (9 x1019 cm-3) to 7 (4 x1020-3) meV, which are comparable to the values reported for Si. However, the trap density at the grain barriers in pyrite is very high (about 2 x1013 cm-2, one order of magnitude higher than in polycrystalline silicon.
 |  |
Perhaps the key problem of pyrite is its surface structure, which seems to be responsible for the low photovoltages achieved up to now. Therefore, cobalt-doped pyrite films with different carrier concentrations were investigated by XPS and UPS (Fig.3). While the Fermi energy, derived from the core level spectra shifts with the carrier concentration as expected from theory, the Fermi level at the surface (UPS results) is pinned. This points to a high density of surface states, which was reported earlier by others [9].
 |
Fig.3 Shift of the Fermi energy relative to the energy of undoped pyrite films versus the charge carrier concentration of cobalt-doped films. The shift of the Fermi energy was derived from the core level spectra both of iron (< ) as well as sulfur (=). For comparison, the Fermi level, derived from the valence band spectra (®) have been added, which display a pinning of the Fermi level at the surface. |
Conclusions
• Pyrite is a stoichiometric semiconductor with a homogeneity range far below the atom% range. Therefore, volume defects should not limit the photovoltaic properties of pyrite, which is in agreement with the high quantum efficiencies, reported.
• Pyrite thin films can be doped n-type by cobalt during MOCVD deposition. The trap density at the grains of the polycrystalline films is about 1 order of magnitude higher compared to silicon. The doping of pyrite offers the possibility to prepare pn-junctions with pyrite single crystals.
• The Fermi level of pyrite thin films is pinned at the surface, independent on the electron concentration. This points to a high defect density at the surface (perhaps broken bonds of sulfur dumbells [10]) that decrease the open circuit voltages of pyrite solar cells. This behavior is somewhat similar to the situation in another sulfide: CuInS2. Solar cells prepared with this absorber material also suffer from a photovoltage that is too low compared to the band gap. The charge carrier transport in these cells can be described only by taking into account multistep tunneling via defects in the depletion region, implying also a high density of defects.
[1] W. Jaegermann and H. Tributsch, J. Appl. Electrochem., 13 (1983) 743.
[2] A. Ennaoui, S. Fiechter, G. Smestad and H. Tributsch, Proc. 1st World Renewable Energy Congr., Reading, UK, 1990, Pergamon Press, City, p. 458.
[3] K. Büker, N. Alonso-Vante and H. Tributsch, J. Appl. Phys., 72 (1992) 5721.
[4] P.P. Altermatt, T. Kiesewetter, K. Ellmer and H. Tributsch, in preparation, (2000) .
[5] N. Alonso-Vante, G. Chatzitheodorou, S. Fiechter, N. Mgoduka, I. Poulios and H. Tributsch, Solar Energy Mat., 18 (1988) 9.
[6] K. Ellmer and C. Höpfner, Phil. Mag. A, 75 (1997) 1129.
[7] B. Thomas, T. Cibik, C. Höpfner, K. Diesner, G. Ehlers, S. Fiechter and K. Ellmer, J. Mat. Sci.: Mat. Electronics, 9 (1998) 61.
[8] J. Oertel, K. Ellmer, W. Bohne, J. Röhrich and H. Tributsch, J. Crystal Growth, 198/199 (1999) 1205.
[9] M. Bronold, C. Pettenkofer and W. Jaegermann, J. Appl. Phys., 76 (1994) 5800.
[10] A.G. Schaufuß, H.W. Nesbitt, I. Kartio, K. Laajalehto, G.M. Bancroft and R. Szargan, Surf. Sci., 411 (1998) 321.
