- Email: [email protected]
Diffusion barriers for CIGS solar cells on metallic substrates
Thin Solid Films 431 – 432 (2003) 392–397
Diffusion barriers for CIGS solar cells on metallic substrates ¨ K. Herz*, A. Eicke, F. Kessler, R. Wachter, M. Powalla ¨ Sonnenenergie- und Wasserstoff-Forschung (ZSW), Hessbruehlstr. 21c, D-70565 Stuttgart, Germany Zentrum fur
Abstract Al2O3 layers of 1–3 mm were deposited as diffusion barriers by RF sputtering from a ceramic target on metal foils of Ti, Kovar䉸 and Cr steel. Cu(In,Ga)Se2 (CIGS)-based thin-film solar cells were deposited onto these substrates using a co-evaporation process for CIGS at T0550 8C. CIGS solar cells of 0.25 cm2 achieved efficiencies of approximately 10–11% without any Na doping. Without barriers, the cell efficiencies were limited to significantly lower values except for Ti substrates. The reduced efficiency values can be attributed mainly to a reduction in fill factors, and secondly, to reduced open-circuit voltages. The different solar cell efficiencies can be correlated with the amount of impurities entering the CIGS layer by diffusion from the substrates, as investigated by simultaneous secondary ion mass spectrometry and sputtered neutral mass spectrometry depth profiling. Without diffusion barriers, Fe and Cr concentrations of several hundred ppm were detected in CIGS layers on Cr steel. Fe, Ni, Co and Ti concentrations from Kovar䉸 and Ti substrates were much smaller, indicating a reduced diffusion. Using Al2O3 barriers, the concentrations of Fe and Cr in CIGS are reduced proportionally to the barrier thickness by up to a factor of 100 when compared to systems without barriers. 䊚 2003 Elsevier Science B.V. All rights reserved. Keywords: Metallic substrate; Solar cell; Cu(In,Ga)Se2; Diffusion barrier; SIMS; SNMS
1. Introduction Interesting and novel applications in photovoltaics are expected from thin and flexible solar modules, especially in the fields of space, aeronautic, and mobile applications. Within the past years the development of flexible and lightweight Cu(In,Ga)Se2 (CIGS) modules has intensified. These activities were encouraged by the relatively high small-area cell efficiencies obtained on polymer as well as on metallic substrates. The most interesting substrates are metal foils, since they can be coated in a roll-to-roll process at high temperatures of up to 600 8C and in a Se atmosphere. Especially stainless steel foils with a potential as low-cost substrates were tested w1,2x. Nevertheless, cell and module efficiencies were lower than on glass substrates. The preparation of highly efficient solar cells requires the deposition of a barrier layer to reduce the diffusion of impurities from the metal substrate into the solar cells w3–5x. If monolithical integration of the cells is desired to realise solar modules on electrically conducting substrates, the deposition of a dielectric barrier is necessary w4,5x. Thin Cr *Corresponding author. Tel.: q49-711-7870-216; fax: q49-7117870-230. E-mail address: [email protected] (K. Herz).
layers w3,8x as well as dielectric layers like Al2O3 w4,6x or SiO2 w6x deposited by sputtering or sol–gel-techniques have been used as diffusion barriers. In this study, the suitability of dielectric Al2O3 layers as diffusion barriers for three different metal substrates was investigated. Some results from secondary ion mass spectrometry (SIMS) and sputtered neutral mass spectrometry (SNMS) depth profiling as well as current– voltage (IyV) measurements on CIGS layers and solar cells are presented. The results are evaluated to demonstrate the influence of the barrier thickness and the substrate type on the amount of impurity diffusion and the solar cell behaviour. 2. Experimental 2.1. Substrates and barrier layers We used 0.1–0.2 mm thick metal foils of a ferritic Cr steel, Kovar䉸, and Ti as substrate materials. The chemical composition of these materials, as well as some technical data like density, coefficient of thermal expansion (CTE) and the maximum roughness Rt (tested by a Dektak profiler) are presented in Table 1. For comparison, these data are also given for the standard
0040-6090/03/$ - see front matter 䊚 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0040-6090(03)00259-1
K. Herz et al. / Thin Solid Films 431 – 432 (2003) 392–397
393
Table 1 Composition, density, CTE, and surface roughness Rt of SLG and metal foils Material
Composition (wt.%)
Density (gycm3)
CTE (ppmyK)
Rt (nm)
SLG Ti (Grade 1) Kovar䉸 Cr steel (ferritic)
SiO2, CaO, Na2O, K2O, Al2O3 Ti 99.5, F, C, N, O, H -0.2 (each) Ni 29, Co 17, Fe (rest) Cu, Mn, Si, C -0.5 (each) Cr 17, Fe (rest) Mn, Si, C, P, S -1 (each)
4.5 4.5 8.2 7.7
(9 8.6 5.9 10.5
10–15 1900 540 680
substrate material soda-lime glass (SLG). All commercially available metal foils showed a relatively high maximum roughness with Rts500–1900 nm when compared to glass substrates of Rts10–15 nm. The high roughness of unpolished substrates results from the rolling process during fabrication. Thin films of Al2O3 (1–3 mm) were deposited by RF magnetron sputtering on these substrates. The metal foils of 10=10 cm2 were sputter-coated from an Al2O3 target with a sputter power of 2000 W and an Argon pressure of 7 mbar. In order to achieve a good electrical insulation of Al2O3 layers, the sputter process was interrupted at half the deposition time to clean the sputtered surface from particles. A low defect density and spacially homogeneous high electrical insulation could be achieved by this procedure w5x. 2.2. Cell preparation Standard MoyCIGSyCdSyZnO test cells of 0.25 cm2 were deposited on Ti, Kovar䉸 and Cr steel sheets of 25=100 mm2 without a barrier layer and with Al2O3 barriers of 1, 2 and 3 mm thickness. A sputtered Mo back-contact layer of approximately 0.5 mm preceded the CIGS deposition (2 mm) in our standard inline machine at approximately 550 8C without any external Na doping. The high-ohmic CdS film (50 nm) and iZnO-film (50 nm) were deposited after CIGS preparation on the full area without masking. The Al-doped ZnO layer of 1.5 mm was sputtered from a ceramic target using a steel mask with circular holes. The sputtered ZnO-dots form the solar cells of 0.25 cm2 which were characterised by the standard current–voltage (IyV) measurements under AM1.5 equivalent illumination. 2.3. SIMS and SNMS analysis Samples for SIMS and SNMS analysis were prepared by using the same substrates as described above, but coated only with the Mo and CIGS layers (no CdS and ZnO). Simultaneous SIMSySNMS depth profiling was performed by a LEYBOLD SSM 200 System using 5 keV primary Arq ions within a scanned area of 2=2 mm2. A quantitative analysis from SNMS is possible down to a minimum concentration of approximately 0.04 at.% (400 ppm). The SIMS signal, on the other
hand, is sensitive to very low element concentrations, but a quantitative analysis is not straightforward. We, therefore, calibrated the SIMS signals of the metal impurities in CIGS by the quantitative SNMS analysis of samples with sufficiently high impurity concentrations (00.1 at.%). This procedure was especially used to achieve quantitative results of low concentrations of Fe and Cr diffused into CIGS from the substrate materials. 3. Results 3.1. Adhesion and morphology of barriers After sputter deposition, the Al2O3 barrier layers had a good adhesion to all metal substrates investigated here. No cracking was observed in the barrier layers after sputter deposition. The adhesion of Mo and CIGS on the barriers is usually sufficient to get satisfactory cell properties. Nevertheless, the mechanical stability of Al2O3 barriers, especially of the thicker ones, is sometimes poor on Kovar䉸 substrates after CIGS deposition. This problem could result from a phase transformation at T)400 8C, which has been observed in Kovar䉸 substrates w6x. 3.2. IyV characteristics of test cells The results of IyV-characterisation are given in Figs. 1–3 and Table 2. Cell efficiencies, open-circuit voltages and fill factors of seven test cells (0.25 cm2) with a 3 mm Al2O3 barrier and without a barrier are outlined for each substrate material. The main results are i. Cell efficiency values, voltages and fill factors of the seven best cells of ten show a distinct scatter that is typical for such small cell sizes. The scatter is caused by imperfections (shunts, cracks, etc.) which are inhomogeneously distributed among the cells. In spite of the relatively high substrate roughness of the Ti foils, this seems to have a minor influence on the cell function. ii. Cells on metal substrates with an Al2O3 barrier show higher efficiencies (9.6–10.6%) than cells without a barrier (5.0–9.9%). Still higher efficiencies could be achieved with specific Na doping (not shown here). iii. The reduction of cell efficiencies without a barrier film is relatively small for cells on Ti substrate
K. Herz et al. / Thin Solid Films 431 – 432 (2003) 392–397
394
Fig. 1. Efficiency h of test cells on metal substrates (Ti, Kovar䉸 and Cr steel) with and without Al2O3 barrier (3 mm).
(hmaxs9.9%) and rather large for cells on Kovar䉸 (hmaxs6.9%) and Cr steel substrate (hmaxs5.0%). iv. The cell efficiency losses in cells without diffusion barriers can be attributed (i) to reduced fill factors (Fig. 2, Table 2) caused by both smaller shunt resistances Rp and increased series resistances Rs as was analysed from IyV curves and (ii) to reduced open-circuit voltages Voc (Fig. 3, Table 2) and shortcircuit currents (not shown here). The reduction of Voc for substrates without a barrier is larger for Cr steel than for Kovar䉸. In the case of Ti, Voc-values of cells without a barrier are slightly higher than for cells with an Al2O3 barrier. 3.3. SIMS and SNMS profiling The results of SIMS and SNMS profiling are shown in Figs. 4 and 5. The concentrations of Cu, In, Ga, Se
and Mo from the CIGS and Mo layers, as well as Fe and Cr from the Cr steel substrate, are depicted vs. the sputter depth for (i) no barrier (Fig. 4) and (ii) 1 and 3 mm Al2O3 barriers (Fig. 5). In Fig. 5 Al and O from the barrier were also analysed. It can be clearly seen that there is a diffusion of Fe and Cr into the CIGS layer with an increasing slope towards the CIGS surface. The slope is most pronounced in the case of a 1 mm barrier layer. The concentrations of Fe and Cr in CIGS decrease with increasing barrier thickness, proving the function of the Al2O3 layer as a diffusion barrier. Films on Kovar䉸 and Ti substrates were analysed in the same manner. The results of all measurements including 2 mm barriers are summarised in Table 3. The element concentrations are represented by their SIMS intensities from depth profiles at 0.25 and 1.25 mm below the surface of the 2 mm thick CIGS layer. A quantification
Fig. 2. Open-circuit voltage Voc of test cells on metal substrates (Ti, Kovar䉸 and Cr steel) with and without Al2O3 barrier (3 mm).
K. Herz et al. / Thin Solid Films 431 – 432 (2003) 392–397
395
Fig. 3. Fill factor FF of test cells on metal substrates (Ti, Kovar䉸 and Cr steel) with and without Al2O3 barrier (3 mm).
Table 2 Solar cell properties of the best cells on each metal substrate (barrier: 3 mm Al2O3) Material
h (%) Voc (mV) jsc (mAycm2) FF (%)
Kovar䉸
Titanium
Cr steel
No barr.
Barrier
No barr.
Barrier
No barr.
Barrier
9.9 582 28.0 60.8
10.5 567 30.5 60.7
6.9 541 27.3 46.8
10.6 561 29.1 65.0
5.0 473 26.1 40.5
9.6 542 30.0 59
Fig. 4. Concentration profile of CIGSyMo on Cr steel by combined SIMS and SNMS investigations.
K. Herz et al. / Thin Solid Films 431 – 432 (2003) 392–397
396
of the SIMS intensities is given for the 56Feq and 52 Crq isotopes by their 100 cps concentration equivalent as described in Section 2.3. In Kovar䉸, the SIMS intensities of Feq were smaller by a factor of 5–10 as compared to Cr steel substrates. The 58Niq and 59Coq intensities of 1 and 5 cps in CIGS layers without a barrier were relatively small as compared to Feq and Crq intensities. On Kovar䉸 substrates with Al2O3 barriers, the signals from these elements were within the noise level ((0.5 cps). No Ti could be detected in the CIGS ((0.5 cps) with barrier-covered substrates. However, Ti intensities are very low even for films without barriers. 4. Discussion and conclusions The diffusion of metal atoms through the Mo back contact into the photovoltaic active layers may be a problem when using metal substrates for CIGS thin-film solar cells. The high deposition temperature of approximately 550 8C enhances diffusion processes as shown in this study for Kovar䉸 and Cr steel substrates by means of SIMSySNMS depth profiling. The diffusion velocities of Ti, Co and Ni within Mo and CIGS seem to be small when compared to Fe and Cr as the corresponding low count rates prove. The diffusion into CIGS could be decreased considerably by depositing Al2O3 barriers of 1, 2 and 3 mm onto the metal substrates by RF magnetron sputtering. The diffusion barrier effect increases with increasing barrier thickness. Due to higher impurity concentrations, CIGS solar cells deposited at T0550 8C on substrates without a barrier show a distinct efficiency degradation as com-
Table 3 SIMS intensities (cps) at a depth of 0.25y1.25 mm
No barrier 1 mm Al2O3 2 mm Al2O3 3 mm Al2O3
Ferritic Cr steel
Kovar䉸
56
56
Fe
q
400y300 120y30 40y20 7y4
52
Cr
q
230y200 210y35 20y9 9y3
Fe
q
45y35 22y6 4y1.5 3y1.5
Ti 58
Ni
q
1y1 (0.5 (0.5 (0.5
59
48
5y5 (0.5 (0.5 (0.5
1.0 (0.5 (0.5 (0.5
Coq
Tiq
Quantification: 56 Feq: 100 cpso 75 ppm; 52 Crq: 100 cpso 85 ppm.
pared to cells with an Al2O3 barrier. Corresponding to the amount of diffused elements, this effect is very small for Ti substrates, but rather significant for Cr steel substrates. In contradiction to this result, a relatively high efficiency of 10.8% was attained on a stainless steel substrate without a diffusion barrier by w2x. The reasons might be a lower deposition temperature (f500 8C), a thicker Mo layer (f1 mm) and another type of stainless steel (CrNi steel). The IyV-analysis indicates that the degradation with our Cr steel is dominated by a significant reduction of fill factor. The underlying changes in the series and shunt resistance values may be caused by increased front- and back-contact resistance (alloying effect) and reduced shunt resistance due to electrically conducting paths (metallic phases, etc.) in the CIGS layer. Such paths might form on grain boundaries as grain boundary diffusion probably dominates w8x. This effect also reduces the short-circuit current. In Ref. w7x the IyV-data from CIGS cells on stainless steel substrates without barrier at low forward voltages reflect
Fig. 5. Concentration profile of CIGSyMoyAl2O3 (1 and 3 mm) on Cr steel by combined SIMS and SNMS investigations.
K. Herz et al. / Thin Solid Films 431 – 432 (2003) 392–397
a tunnel-like transport mechanism which behaves like a low shunt resistance. A second reason for efficiency losses is a reduced Voc caused by metal impurities in the CIGS semiconductor, thereby increasing the recombination rates. As was evaluated in Ref. w3x, predominantly Fe impurities are presumed to be responsible for the deep level defects analysed by admittance measurements. On the basis of a statistical analysis about the correlation of efficiency and barrier thickness (not shown in this paper) and the SIMSySNMS analysis in this study, an Al2O3 barrier thickness of 02 mm for Cr steel and 01 mm for Kovar䉸 is recommended for CIGS deposition. Acknowledgments The authors acknowledge the support by the CIS group of ZSW. This work was supported by the European Commission under EC contract No. JOR3-CT98¨ Wissenschaft, Forschung und 0304, the Ministerium fur ¨ Kunst Baden-Wurttemberg (MWK BW) and the ‘Stif¨ tung Energieforschung Baden-Wurttemberg’.
397
References w1x J. Britt, S. Wiedemann, R. Wendt, S. Albright, Technical Report NRELySR-520-26840, 1999, p. 1. w2x T. Satoh, Y. Hashimoto, S. Shimakawa, S. Hayashi, T. Negami, Proceedings of the 12th Int. Phot. Sci. and Eng. Conf., Korea, 2001, p. 93. w3x M. Hartmann, M. Schmidt, A. Jasenek, H.-W. Schock, F. Kessler, K. Herz, M. Powalla, Proceedings of the 28th IEEE Phot. Spec. Conf., Anchorage, 2000, p. 638. w4x F. Kessler, K. Herz, M. Powalla, M. Hartmann, M. Schmidt, A. Jasanek, H.W. Schock, Proceeding Vol. 668 of the Mat. Res. Soc. Symp., San Francisco, 2001, p. H3.6.1. w5x K. Herz, F. Kessler, R. Wachter, ¨ M. Powalla, J. Schneider, A. Schulz, U. Schumacher, Thin Solid Films 403–404 (2002) 384. w6x G.S. Vicente, J. Herrero, A. Morales, C. Mafiotte, M.T. Gutierrez, M. Hartmann, A. Jasenek, H.W. Schock, Proceedings of the 17th Eur. Phot. Sol. Conf., Munich, 2001, p. 1098. w7x G.A. Naik, W.A. Anderson, Proceeding Vol. 668 of the Mat. Res. Soc. Symp., San Francisco, 2001, p. H3.10.1. w8x W.K. Batchelor, M.E. Beck, R. Huntington, I.L. Repins, A. Rockett, W.N. Shafarman, F.S. Hasoon, J.S. Britt, Proceedings of the 29th IEEE Phot. Spec. Conf., New Orleans, in press.
Diffusion barriers for CIGS solar cells on metallic substrates ¨ K. Herz*, A. Eicke, F. Kessler, R. Wachter, M. Powalla ¨ Sonnenenergie- und Wasserstoff-Forschung (ZSW), Hessbruehlstr. 21c, D-70565 Stuttgart, Germany Zentrum fur
Abstract Al2O3 layers of 1–3 mm were deposited as diffusion barriers by RF sputtering from a ceramic target on metal foils of Ti, Kovar䉸 and Cr steel. Cu(In,Ga)Se2 (CIGS)-based thin-film solar cells were deposited onto these substrates using a co-evaporation process for CIGS at T0550 8C. CIGS solar cells of 0.25 cm2 achieved efficiencies of approximately 10–11% without any Na doping. Without barriers, the cell efficiencies were limited to significantly lower values except for Ti substrates. The reduced efficiency values can be attributed mainly to a reduction in fill factors, and secondly, to reduced open-circuit voltages. The different solar cell efficiencies can be correlated with the amount of impurities entering the CIGS layer by diffusion from the substrates, as investigated by simultaneous secondary ion mass spectrometry and sputtered neutral mass spectrometry depth profiling. Without diffusion barriers, Fe and Cr concentrations of several hundred ppm were detected in CIGS layers on Cr steel. Fe, Ni, Co and Ti concentrations from Kovar䉸 and Ti substrates were much smaller, indicating a reduced diffusion. Using Al2O3 barriers, the concentrations of Fe and Cr in CIGS are reduced proportionally to the barrier thickness by up to a factor of 100 when compared to systems without barriers. 䊚 2003 Elsevier Science B.V. All rights reserved. Keywords: Metallic substrate; Solar cell; Cu(In,Ga)Se2; Diffusion barrier; SIMS; SNMS
1. Introduction Interesting and novel applications in photovoltaics are expected from thin and flexible solar modules, especially in the fields of space, aeronautic, and mobile applications. Within the past years the development of flexible and lightweight Cu(In,Ga)Se2 (CIGS) modules has intensified. These activities were encouraged by the relatively high small-area cell efficiencies obtained on polymer as well as on metallic substrates. The most interesting substrates are metal foils, since they can be coated in a roll-to-roll process at high temperatures of up to 600 8C and in a Se atmosphere. Especially stainless steel foils with a potential as low-cost substrates were tested w1,2x. Nevertheless, cell and module efficiencies were lower than on glass substrates. The preparation of highly efficient solar cells requires the deposition of a barrier layer to reduce the diffusion of impurities from the metal substrate into the solar cells w3–5x. If monolithical integration of the cells is desired to realise solar modules on electrically conducting substrates, the deposition of a dielectric barrier is necessary w4,5x. Thin Cr *Corresponding author. Tel.: q49-711-7870-216; fax: q49-7117870-230. E-mail address: [email protected] (K. Herz).
layers w3,8x as well as dielectric layers like Al2O3 w4,6x or SiO2 w6x deposited by sputtering or sol–gel-techniques have been used as diffusion barriers. In this study, the suitability of dielectric Al2O3 layers as diffusion barriers for three different metal substrates was investigated. Some results from secondary ion mass spectrometry (SIMS) and sputtered neutral mass spectrometry (SNMS) depth profiling as well as current– voltage (IyV) measurements on CIGS layers and solar cells are presented. The results are evaluated to demonstrate the influence of the barrier thickness and the substrate type on the amount of impurity diffusion and the solar cell behaviour. 2. Experimental 2.1. Substrates and barrier layers We used 0.1–0.2 mm thick metal foils of a ferritic Cr steel, Kovar䉸, and Ti as substrate materials. The chemical composition of these materials, as well as some technical data like density, coefficient of thermal expansion (CTE) and the maximum roughness Rt (tested by a Dektak profiler) are presented in Table 1. For comparison, these data are also given for the standard
0040-6090/03/$ - see front matter 䊚 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0040-6090(03)00259-1
K. Herz et al. / Thin Solid Films 431 – 432 (2003) 392–397
393
Table 1 Composition, density, CTE, and surface roughness Rt of SLG and metal foils Material
Composition (wt.%)
Density (gycm3)
CTE (ppmyK)
Rt (nm)
SLG Ti (Grade 1) Kovar䉸 Cr steel (ferritic)
SiO2, CaO, Na2O, K2O, Al2O3 Ti 99.5, F, C, N, O, H -0.2 (each) Ni 29, Co 17, Fe (rest) Cu, Mn, Si, C -0.5 (each) Cr 17, Fe (rest) Mn, Si, C, P, S -1 (each)
4.5 4.5 8.2 7.7
(9 8.6 5.9 10.5
10–15 1900 540 680
substrate material soda-lime glass (SLG). All commercially available metal foils showed a relatively high maximum roughness with Rts500–1900 nm when compared to glass substrates of Rts10–15 nm. The high roughness of unpolished substrates results from the rolling process during fabrication. Thin films of Al2O3 (1–3 mm) were deposited by RF magnetron sputtering on these substrates. The metal foils of 10=10 cm2 were sputter-coated from an Al2O3 target with a sputter power of 2000 W and an Argon pressure of 7 mbar. In order to achieve a good electrical insulation of Al2O3 layers, the sputter process was interrupted at half the deposition time to clean the sputtered surface from particles. A low defect density and spacially homogeneous high electrical insulation could be achieved by this procedure w5x. 2.2. Cell preparation Standard MoyCIGSyCdSyZnO test cells of 0.25 cm2 were deposited on Ti, Kovar䉸 and Cr steel sheets of 25=100 mm2 without a barrier layer and with Al2O3 barriers of 1, 2 and 3 mm thickness. A sputtered Mo back-contact layer of approximately 0.5 mm preceded the CIGS deposition (2 mm) in our standard inline machine at approximately 550 8C without any external Na doping. The high-ohmic CdS film (50 nm) and iZnO-film (50 nm) were deposited after CIGS preparation on the full area without masking. The Al-doped ZnO layer of 1.5 mm was sputtered from a ceramic target using a steel mask with circular holes. The sputtered ZnO-dots form the solar cells of 0.25 cm2 which were characterised by the standard current–voltage (IyV) measurements under AM1.5 equivalent illumination. 2.3. SIMS and SNMS analysis Samples for SIMS and SNMS analysis were prepared by using the same substrates as described above, but coated only with the Mo and CIGS layers (no CdS and ZnO). Simultaneous SIMSySNMS depth profiling was performed by a LEYBOLD SSM 200 System using 5 keV primary Arq ions within a scanned area of 2=2 mm2. A quantitative analysis from SNMS is possible down to a minimum concentration of approximately 0.04 at.% (400 ppm). The SIMS signal, on the other
hand, is sensitive to very low element concentrations, but a quantitative analysis is not straightforward. We, therefore, calibrated the SIMS signals of the metal impurities in CIGS by the quantitative SNMS analysis of samples with sufficiently high impurity concentrations (00.1 at.%). This procedure was especially used to achieve quantitative results of low concentrations of Fe and Cr diffused into CIGS from the substrate materials. 3. Results 3.1. Adhesion and morphology of barriers After sputter deposition, the Al2O3 barrier layers had a good adhesion to all metal substrates investigated here. No cracking was observed in the barrier layers after sputter deposition. The adhesion of Mo and CIGS on the barriers is usually sufficient to get satisfactory cell properties. Nevertheless, the mechanical stability of Al2O3 barriers, especially of the thicker ones, is sometimes poor on Kovar䉸 substrates after CIGS deposition. This problem could result from a phase transformation at T)400 8C, which has been observed in Kovar䉸 substrates w6x. 3.2. IyV characteristics of test cells The results of IyV-characterisation are given in Figs. 1–3 and Table 2. Cell efficiencies, open-circuit voltages and fill factors of seven test cells (0.25 cm2) with a 3 mm Al2O3 barrier and without a barrier are outlined for each substrate material. The main results are i. Cell efficiency values, voltages and fill factors of the seven best cells of ten show a distinct scatter that is typical for such small cell sizes. The scatter is caused by imperfections (shunts, cracks, etc.) which are inhomogeneously distributed among the cells. In spite of the relatively high substrate roughness of the Ti foils, this seems to have a minor influence on the cell function. ii. Cells on metal substrates with an Al2O3 barrier show higher efficiencies (9.6–10.6%) than cells without a barrier (5.0–9.9%). Still higher efficiencies could be achieved with specific Na doping (not shown here). iii. The reduction of cell efficiencies without a barrier film is relatively small for cells on Ti substrate
K. Herz et al. / Thin Solid Films 431 – 432 (2003) 392–397
394
Fig. 1. Efficiency h of test cells on metal substrates (Ti, Kovar䉸 and Cr steel) with and without Al2O3 barrier (3 mm).
(hmaxs9.9%) and rather large for cells on Kovar䉸 (hmaxs6.9%) and Cr steel substrate (hmaxs5.0%). iv. The cell efficiency losses in cells without diffusion barriers can be attributed (i) to reduced fill factors (Fig. 2, Table 2) caused by both smaller shunt resistances Rp and increased series resistances Rs as was analysed from IyV curves and (ii) to reduced open-circuit voltages Voc (Fig. 3, Table 2) and shortcircuit currents (not shown here). The reduction of Voc for substrates without a barrier is larger for Cr steel than for Kovar䉸. In the case of Ti, Voc-values of cells without a barrier are slightly higher than for cells with an Al2O3 barrier. 3.3. SIMS and SNMS profiling The results of SIMS and SNMS profiling are shown in Figs. 4 and 5. The concentrations of Cu, In, Ga, Se
and Mo from the CIGS and Mo layers, as well as Fe and Cr from the Cr steel substrate, are depicted vs. the sputter depth for (i) no barrier (Fig. 4) and (ii) 1 and 3 mm Al2O3 barriers (Fig. 5). In Fig. 5 Al and O from the barrier were also analysed. It can be clearly seen that there is a diffusion of Fe and Cr into the CIGS layer with an increasing slope towards the CIGS surface. The slope is most pronounced in the case of a 1 mm barrier layer. The concentrations of Fe and Cr in CIGS decrease with increasing barrier thickness, proving the function of the Al2O3 layer as a diffusion barrier. Films on Kovar䉸 and Ti substrates were analysed in the same manner. The results of all measurements including 2 mm barriers are summarised in Table 3. The element concentrations are represented by their SIMS intensities from depth profiles at 0.25 and 1.25 mm below the surface of the 2 mm thick CIGS layer. A quantification
Fig. 2. Open-circuit voltage Voc of test cells on metal substrates (Ti, Kovar䉸 and Cr steel) with and without Al2O3 barrier (3 mm).
K. Herz et al. / Thin Solid Films 431 – 432 (2003) 392–397
395
Fig. 3. Fill factor FF of test cells on metal substrates (Ti, Kovar䉸 and Cr steel) with and without Al2O3 barrier (3 mm).
Table 2 Solar cell properties of the best cells on each metal substrate (barrier: 3 mm Al2O3) Material
h (%) Voc (mV) jsc (mAycm2) FF (%)
Kovar䉸
Titanium
Cr steel
No barr.
Barrier
No barr.
Barrier
No barr.
Barrier
9.9 582 28.0 60.8
10.5 567 30.5 60.7
6.9 541 27.3 46.8
10.6 561 29.1 65.0
5.0 473 26.1 40.5
9.6 542 30.0 59
Fig. 4. Concentration profile of CIGSyMo on Cr steel by combined SIMS and SNMS investigations.
K. Herz et al. / Thin Solid Films 431 – 432 (2003) 392–397
396
of the SIMS intensities is given for the 56Feq and 52 Crq isotopes by their 100 cps concentration equivalent as described in Section 2.3. In Kovar䉸, the SIMS intensities of Feq were smaller by a factor of 5–10 as compared to Cr steel substrates. The 58Niq and 59Coq intensities of 1 and 5 cps in CIGS layers without a barrier were relatively small as compared to Feq and Crq intensities. On Kovar䉸 substrates with Al2O3 barriers, the signals from these elements were within the noise level ((0.5 cps). No Ti could be detected in the CIGS ((0.5 cps) with barrier-covered substrates. However, Ti intensities are very low even for films without barriers. 4. Discussion and conclusions The diffusion of metal atoms through the Mo back contact into the photovoltaic active layers may be a problem when using metal substrates for CIGS thin-film solar cells. The high deposition temperature of approximately 550 8C enhances diffusion processes as shown in this study for Kovar䉸 and Cr steel substrates by means of SIMSySNMS depth profiling. The diffusion velocities of Ti, Co and Ni within Mo and CIGS seem to be small when compared to Fe and Cr as the corresponding low count rates prove. The diffusion into CIGS could be decreased considerably by depositing Al2O3 barriers of 1, 2 and 3 mm onto the metal substrates by RF magnetron sputtering. The diffusion barrier effect increases with increasing barrier thickness. Due to higher impurity concentrations, CIGS solar cells deposited at T0550 8C on substrates without a barrier show a distinct efficiency degradation as com-
Table 3 SIMS intensities (cps) at a depth of 0.25y1.25 mm
No barrier 1 mm Al2O3 2 mm Al2O3 3 mm Al2O3
Ferritic Cr steel
Kovar䉸
56
56
Fe
q
400y300 120y30 40y20 7y4
52
Cr
q
230y200 210y35 20y9 9y3
Fe
q
45y35 22y6 4y1.5 3y1.5
Ti 58
Ni
q
1y1 (0.5 (0.5 (0.5
59
48
5y5 (0.5 (0.5 (0.5
1.0 (0.5 (0.5 (0.5
Coq
Tiq
Quantification: 56 Feq: 100 cpso 75 ppm; 52 Crq: 100 cpso 85 ppm.
pared to cells with an Al2O3 barrier. Corresponding to the amount of diffused elements, this effect is very small for Ti substrates, but rather significant for Cr steel substrates. In contradiction to this result, a relatively high efficiency of 10.8% was attained on a stainless steel substrate without a diffusion barrier by w2x. The reasons might be a lower deposition temperature (f500 8C), a thicker Mo layer (f1 mm) and another type of stainless steel (CrNi steel). The IyV-analysis indicates that the degradation with our Cr steel is dominated by a significant reduction of fill factor. The underlying changes in the series and shunt resistance values may be caused by increased front- and back-contact resistance (alloying effect) and reduced shunt resistance due to electrically conducting paths (metallic phases, etc.) in the CIGS layer. Such paths might form on grain boundaries as grain boundary diffusion probably dominates w8x. This effect also reduces the short-circuit current. In Ref. w7x the IyV-data from CIGS cells on stainless steel substrates without barrier at low forward voltages reflect
Fig. 5. Concentration profile of CIGSyMoyAl2O3 (1 and 3 mm) on Cr steel by combined SIMS and SNMS investigations.
K. Herz et al. / Thin Solid Films 431 – 432 (2003) 392–397
a tunnel-like transport mechanism which behaves like a low shunt resistance. A second reason for efficiency losses is a reduced Voc caused by metal impurities in the CIGS semiconductor, thereby increasing the recombination rates. As was evaluated in Ref. w3x, predominantly Fe impurities are presumed to be responsible for the deep level defects analysed by admittance measurements. On the basis of a statistical analysis about the correlation of efficiency and barrier thickness (not shown in this paper) and the SIMSySNMS analysis in this study, an Al2O3 barrier thickness of 02 mm for Cr steel and 01 mm for Kovar䉸 is recommended for CIGS deposition. Acknowledgments The authors acknowledge the support by the CIS group of ZSW. This work was supported by the European Commission under EC contract No. JOR3-CT98¨ Wissenschaft, Forschung und 0304, the Ministerium fur ¨ Kunst Baden-Wurttemberg (MWK BW) and the ‘Stif¨ tung Energieforschung Baden-Wurttemberg’.
397
References w1x J. Britt, S. Wiedemann, R. Wendt, S. Albright, Technical Report NRELySR-520-26840, 1999, p. 1. w2x T. Satoh, Y. Hashimoto, S. Shimakawa, S. Hayashi, T. Negami, Proceedings of the 12th Int. Phot. Sci. and Eng. Conf., Korea, 2001, p. 93. w3x M. Hartmann, M. Schmidt, A. Jasenek, H.-W. Schock, F. Kessler, K. Herz, M. Powalla, Proceedings of the 28th IEEE Phot. Spec. Conf., Anchorage, 2000, p. 638. w4x F. Kessler, K. Herz, M. Powalla, M. Hartmann, M. Schmidt, A. Jasanek, H.W. Schock, Proceeding Vol. 668 of the Mat. Res. Soc. Symp., San Francisco, 2001, p. H3.6.1. w5x K. Herz, F. Kessler, R. Wachter, ¨ M. Powalla, J. Schneider, A. Schulz, U. Schumacher, Thin Solid Films 403–404 (2002) 384. w6x G.S. Vicente, J. Herrero, A. Morales, C. Mafiotte, M.T. Gutierrez, M. Hartmann, A. Jasenek, H.W. Schock, Proceedings of the 17th Eur. Phot. Sol. Conf., Munich, 2001, p. 1098. w7x G.A. Naik, W.A. Anderson, Proceeding Vol. 668 of the Mat. Res. Soc. Symp., San Francisco, 2001, p. H3.10.1. w8x W.K. Batchelor, M.E. Beck, R. Huntington, I.L. Repins, A. Rockett, W.N. Shafarman, F.S. Hasoon, J.S. Britt, Proceedings of the 29th IEEE Phot. Spec. Conf., New Orleans, in press.