POLYMORPHOUS SILICON: TRANSPORT PROPERTIES AND SOLAR CELL APPLICATIONS

POLYMORPHOUS SILICON: TRANSPORT PROPERTIES AND SOLAR CELL APPLICATIONS C. Longeaud*, J. P. Kleider*, M. Gauthier*, R. Brtiggemann*, Y. Poissant**, and P. Roca i Cabarrocas** * Laboratoire de Genie Electrique de Paris (UMR 857 CNRS), Supdlec, Universitds Paris VI et XI, Plateau de Moulon, 91192 Gif sur Yvette Cedex, France, longeaud@lgep.supelec.fr ** Laboratoire de Physique des Interfaces et Couches Minces (UMR 7647 CNRS), Ecole Polytechnique, 91128 Palaiseau, France ABSTRACT Transport properties of hydrogenated polymorphous silicon layers (pm-si:h) deposited at 15 'C under various pressures in the range 8-293 Pa in sandwich (Schottky and p-i-n diodes) and coplanar structures have been compared to those of hydrogenated amorphous silicon (a-si:h) samples deposited at the same temperature in standard conditions. The layers have been studied as-deposited, annealed and after light-soaking. With increasing pressure up to 24 Pa: i) the density of states above the Fermi level decreases as determined by means of the modulated photocurrent technique, ii) the mobility-lifetime products of electrons and holes measured by means of steady-state photoconductivity and photocarrier grating techniques both increase. The highest values for the diffusion length of minority carriers exceed 2 nm. Capacitance measurements as a function of frequency and temperature show that the density of states at the Fermi level is lower in the pm-si:h than in the a-si:h films. After light-soaking the diffusion length of minority carriers in a-si:h is reduced by a factor oftwo whereas it is less reduced or not affected in the pm-si:h layers. Solar cells including this new material present an excellent stability. INTRODUCTION Exploring the space of parameters in radio frequency powered plasma enhanced chemical vapour deposition (RF-PECVD) we have shown that it is possible to deposit a new type of material exhibiting enhanced transport properties at least as far as the majority carriers are concerned [1, 2). In particular we have found that, deposited at a substrate temperature of 25 'C under high dilution of silane into hydrogen, helium or argon with a high pressure of the gas mixture and at high RF power, films of this new material present very high values of the mobility lifetime product for electrons (P-te) compared to films of hydrogenated amorphous silicon (a-si:h) deposited under standard conditions (no dilution of silane, low pressure and low RF power) [3]. Moreover, the behaviour of this new material under light-soaking makes it very promising for photovoltaic applications [4]. We believe that these peculiar properties are due to the microstructure of the material made of ordered nanoparticles of silicon embedded into an amorphous matrix as shown by high resolution transmission electron microscopy [5]. These nanoparticles are produced in the plasma and deposited onto the film since the glow discharge occurs in a regime close to powder formation [3,5]. Because of the presence of nano-sized ordered domains within the amorphous phase we have named this material polymorphous silicon (pm-si:h). While most of the materials studies have focused on layers deposited at 25 'C, the solar cells are usually prepared at 15 'C for example to avoid the reduction of the SnO2 substrate by the hydrogen of the discharge as well as to decrease the diffusion of impurities from the substrate into the intrinsic layer. Therefore we focus here on pm-si:h films prepared at a substrate temperature of 15 'C and show that this material is indeed an excellent candidate for photovoltaic applications. 51 Mat. Res. Soc. Symp. Proc. Vol. 557 1999 Materials Research Society

SAMPLES AND EXPERIMENTS The samples studied were deposited in a radio frequency powered (13.56 MHz) PECVD system [6] at a substrate temperature Ts of 15 'C. For the polymorphous films, the silane was highly diluted into hydrogen (3% of SiH 4 in 97 % of H 2 ) and the RF power Prf was of the order of 2 11 mw/cm. The total pressure P in the chamber was varied between 8 and 293 Pa. The transport properties of these samples were compared to those of an a-si:h sample prepared under standard conditions (T,=15 'C, Pif =5 mwcm- 2, P= 5 Pa). For both types of materials different types of samples were prepared: simple films deposited onto 759 Coming glass, Schottky diodes and p-i-n diodes, the i-layer being made of either pm-si:h or a-si:h. Two parallel ohmic aluminum electrodes 2 mm apart were thermally evaporated on the films deposited onto glass in order to perform the following experiments: conductivity and steady-state photoconductivity (SSPC) at different light intensities, modulated photocurrent (MPC) and steadystate photocarrier grating (SSPG). From these experiments we determined respectively the value of the dark conductivity and of the activation energy, the mobility-lifetime (latte) product of the majority carriers, here assumed to be the electrons, the shape of the density of states above the Fermi level, hereafter referred to as the MPC-DOS, and the mobility-lifetime (4th) product for the minority carriers (i.e. the holes). On the Schottky diodes we performed capacitance measurements as a function of frequency and temperature to determine the density of states at the Fermi level. We also performed time of flight measurements to determine the drift mobilities of the carriers. Finally, the I-V characteristics of the solar cells were measured at different light intensities to obtain the effective axt of carriers in the intrinsic layer of the pm-si:h cells by means of the variable irradiance method (VIM) [7]. The samples were studied in their as-deposited, annealed and light-soaked states. The light soaking was achieved under white light with a flux of the order of 4 mw/cm2 at a temperature of 8 'C during 7 hours for the films and 5 'C during 24 hours for the pin diodes. For the films the kinetics of the light soaking process was monitored by the evolution of the dark and photo conductivity measured at 2'C. RESULTS For all the studied samples the room temperature dark conductivities and activation energies range between 3x112 and 6x111 S.cmI and.7 and.9 ev, respectively. In Fig. 1 we present the evolution with pressure of the mobility-lifetime 17 - product for electrons and holes measured under red light (?,=67 nm) with a flux of 1 cm- s. These products were measured before and after light-soaking. The comparison with the results obtained on a standard a-si:h sample shows that in the as-deposited state the ltt products for holes and electrons are of the same order of magnitude or slightly lower than those measured on the standard sample. This is in contrast to what we obtain at a lower flux of 114 cm2 s where the gat products for electrons for the pm-si:h material are higher than for the a-si:h layer (see Figure 2). It can also be seen that as the pressure increases up to 238 Pa both gtte and 9ith increase simultaneously underlining a continuous improvement of the material quality. Finally it appears that in the range 2-26 Pa there is an optimum for the transport parameters and that in this range of pressure there is only a small influence of the light-soaking process on 9.tth. As a consequence, after light-soaking, the lath of the sample deposited at 238 Pa is three times higher than for the a-si:h sample. Moreover, the diffusion length of the minority carriers in the light-soaked state is higher than.2 gm for the pm- Si:H sample deposited at 238 Pa. We present in Figure 2 the evolution with pressure of the value of the MPC-DOS measured at.5 ev below the conduction band edge for samples in the as-deposited state. Also presented are the evolution of late measured under a dc flux of 114 cm-2s-1. Clearly, there is a good correlation between these two quantities: the lower the MPC-DOS the higher the lte. Note that the lgte 52

products were obtained here on as-deposited samples without any annealing. An annealing of the samples at the deposition temperature of 15 'C during 1-2 hours leads to an increase of these products by a factor of five to ten for the pm-si:h material and not for the a-si:h sample. g'le El ite LS lith * it'h LS 1-7 18 o C 1-7 1-8 oil l ii oi liii 1-9 5 1 15 P (Pa) 1-9 2 25 3 Figure 1 : Evolution vs pressure of the igt products of electrons and holes, measured with a dc flux of 117 cm2si before and after light-soaking (LS). Also shown are the ltg products of electrons and holes for a standard a-si:h sample prepared at the sam, temperature at P =5 Pa. 1-5 1111''li11l'l11111l11li'ili, As-deposited 11 -, 1-6 - /~o OL 116 I, C)' 1-7 1 1 11 i f I I I I [ I 1 1 5 1 15 2 25 3 P (Pa) Figure 2 : Evolution vs pressure of the MPC-DOS measured at.5 ev below the conduction band edge for samples in the as-deposited state. Also presented are the lite obtained at 14 2 1 Fd=14 cm s.these values can be compared to those obtained on a standard sample shown at P=5 Pa. Lines are guides to the eyes. 53

To understand this increase we have performed capacitance measurements versus temperature and frequency C(T,O) on a Schottky diode, the layer of which was deposited at 238 Pa, since it is around this pressure that the pm-si:h material seems to present the best transport parameters (see Fig. 1). The value of the DOS at the Fermi level we found was lower than 1"' cm 3 ev-', a value which is among the lowest ever found in intrinsic a-si:h samples. This lowering of the Fermi level DOS in pm-si layers compared to a-si:h layers can be considered as the main reason of the high values of the lite products. Indeed, time-of-flight measurements have shown that the drift mobility was of the order of 1.5 cm 2 V' s-1 comparable to the value found in standard a-si:h. The increase of the jitr products in pm-si:h material prepared at 15 C is thus mainly due to a lowering of the DOS instead of an increase of the mobility as already suggested for the pm-si:h prepared at 25 `C [2]. We present in Figure 3 the evolution with pressure of the value of the MPC-DOS measured at.5 ev below the conduction band edge for samples in the light-soaked state. Also presented are the evolution of it. measured under a dc flux of 114 cm-2 s-. Again, there is a good correlation between these two quantities: the lower the MPC-DOS the higher the tite. It can also be seen that the lower values of MPC-DOS in the light-soaked state are obtained for the samples deposited in the pressure range 2-26 Pa. 1-6 _ 119 - Light-Soaked - 117 1 1-1 1 C S 5 1 15 2 25 3 P (Pa) Figure 3 : Evolution vs pressure of the MPG-DOS measured at.5 ev below the conduction band edge for samples in the light-soaked state. Also presented are the i-te obtained at 14-2 -I FdC=1 cm s. These values can be compared to those obtained on a standard sample shown at P=5 Pa. Lines are guides to the eyes. Considering the results presented in Figs. 1-3 one expects the best solar cells to be produced in the pressure range of 2-26 Pa. This is confirmed by the results presented in Table I where we report the material and solar cells parameters for a series of p-i-n solar cells deposited at 15 C and having an i-layer thickness of the order of.4 jam. The p-layer for this series of cells consists of an a-sic:h alloy obtained from the dissociation of a gas mixture consisting of 6 sccm of silane, 5 sccm of methane, and 6 sccm of hydrogen. An undoped a-sic:h buffer layer was inserted between this p-layer and the intrinsic layer. As shown in Table I the effective li't product, deduced from the VIM measurements, presents a maximum at 21 Pa in agreement with the fill factor of the solar cell. We note that the values of fill factor and efficiencies reported in this table are lower than those previously reported for pm- 54

Si:H solar cells [4]. This is attributed to a non-optimization of the solar cells presented here since our purpose was mainly to correlate the material transport properties to the solar cell parameters. Table I: I(V) characteristics and (l-t)eff of pm-si:h cells deposited at different pressures. Pressure (i-t)eff FF Voc 'sc Eft. (Pa) (cm 2 V-1) (V) (ma/cm 2 ) (%) 132 1.31x1-8.64.79 8.8 4.3 158 1.53x1-8.66.82 9.9 5.3 21 3.71x1-8.67.85 1.9 6.2 238 1.44x1-8.58.81 1.9 5.2 263 1.42x1-8.62.82 11.2 5.6 Turning now to the stability of the solar cells, Fig. 4 shows the spectral response of a solar cell produced at 238 Pa. Indeed, at this pressure we expect a stable efficiency (see Figure 1) since the hole diffusion length is not altered by the light-soaking process. As shown in Figure 4, the spectral response of the solar cell does not change after 24 hours white-light soaking under 4 AM 1 at 5 'C except slightly at short k. As indicated above, the solar cells deposited at high pressure have not been optimized. Therefore the absolute values of the spectral response are low with respect to those obtained in solar cells with pm-si:h layers deposited at lower pressure and presented elsewhere [4]. c,.8.7.6.5.4 C.).3.2 -"-e- as-dep.1 P=238 Pa 4 45 5 55 6 65 7 75 X (nm) Figure 4: Spectral Response before (as-dep.) and after 24 h light-soaking (LS) of a pm-si:h cell deposited at 238 Pa. The cell thickness is.42 p.m. Lines are guides to the eyes. The importance of the optimization on the absolute efficiency of the solar cells is demonstrated by the results reported in Table II. These results have been obtained with a p-i-n solar cell in which we replaced the p(a-sic:h) layer by a p(gac-si:h) layer, the intrinsic layer being deposited at 238 Pa. As compared to the results of Table I we obtain higher values of the fill factor and open circuit voltage. Morevover the short circuit current of the solar cell remains almost constant, in agreement with the spectral response results (Fig. 4). The higher value of Jsc is due to 55

the smaller absorption in the p([c-si:h) layer. As previously reported [4], we can see in Table II that although the FF of the solar cell decreases upon light-soaking, the efficiency remains constant because of the increase in open circuit voltage. Table I: I(V) characteristics of a pm-si:h cell in its as-deposited state and after 24 hours lightsoaking at 5 'C under 4 AM I conditions. State FF Jsc Voc Efficiency (ma/cm 2 ) (V) (%) As-deposited.63 13.4.89 7.5 Light-soaked.6 13..94 7.3 CONCLUSION We have studied the transport properties of both minority and majority carriers as well as stability of polymorphous silicon films produced at different pressures at 15 'C. We observe a clear pressure range (2 < P _< 26 Pa) where the jit products for electrons and holes are maximum, corresponding to a minimum in the defect density. Moreover, for the best samples, the [t product of holes does not show any degradation. The optimum in the material properties has been correlated to the solar cell parameters, both in terms of efficiency and stability. ACKNOWLEDGEMENTS This work was supported by the Centre National de la Recherche Scientifique (CNRS) and the Agence de l'environnement et de la Maitrise de l'energie (A.D.E.M.E.) within the framework of their joint photovoltaic program (ECODEV). REFERENCES 1. A. R. Middya, S. Hazra, S. Ray, A. K. Barua and C. Longeaud, J. Non-Cryst. Solids 198-2, 167 (1996). 2. C. Longeaud, J.P. Kleider, P. Roca i Cabarrocas, S. Hamma, R. Meaudre and M. Meaudre, J. Non-Cryst. Solids 227-23, 96 (1998). 3. P. Roca i Cabarrocas, in Amorphous and Microcrystalline Silicon Technology, edited by R. Schropp, H. M. Branz, M. Hack, I. Shimizu, and S. Wagner (Mat. Res. Soc.. Proc. 57, San Fransisco, CA, 1998) 4. P. Roca i Cabarrocas, P. St'ahel and S. Hamma, Proc. of the 2nd World Conference on Photovoltaic Solar Energy Conversion, edited by J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann and E.D. Dunlop, Vienne, Austria, 355 (1998). 5. P. Roca i Cabarrocas, S. Hamma, S. N. Sharma, G. Viera, E. Bertran and J. Costa, J. Non- Cryst. Solids 227-23, 871 (1998). 6. P. Roca i Cabarrocas, J. B. Chevrier, J. Huc, A. Lloret, J. Y. Parey and J. P. M. Schmitt, J. Vac. Sci. Technol. A 9, 2331 (1991). 7. J. Merten, J.M. Asensi, C. Voz, A.V. Shah, R. Platz, and J. Andreu, IEEE Trans. on Electron Devices 45, 423 (1998). 56