Available online at www.sciencedirect.com Energy Procedia 00 (2009) 000 000 Energy Procedia 2 (2010) 59 64 Energy Procedia www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia E-MRS Spring meeting 2009, Symposium B Effects of the front surface field in n-type interdigitated back contact silicon heterojunctions solar cells D. Diouf a, *, J.P. Kleider a, T. Desrues b, P.-J. Ribeyron b a LGEP, CNRS UMR8507; SUPELEC; Univ Paris-Sud ; UPMC Univ Paris 6 ; 11 rue Joliot-Curie, Plateau de Moulon, F-91192 Gif-sur-Yvette Cedex, France b INES-CEA Grenoble 50 Avenue du Lac Léman, F-73370 Le Bourget du Lac, France Received 1 JuneElsevier 2009; received use only: in Received revisedate form here; 1revised December date here; 2009; accepted accepted date 20 here December 2009 Abstract Previous simulations of interdigitated back contact silicon heterojunction (IBC-SiHJ) solar cells have indicated that front surface passivation is a critical factor in the performance of such cells. This is why we here focus on the effect of a front surface field (FSF) layer by 2D numerical modelling. A FSF layer made of a highly doped thin crystalline silicon top layer makes the cell performance insensitive to the surface recombination velocity up to quite high values (5000 cm/s). It also reduces the lateral resistance losses due to the increased lateral current through the doped layer particularly in IBC-SiHJ solar cells with large pitches. A FSF layer can also be produced by doped hydrogenated amorphous silicon due to the induced accumulation layer at the crystalline silicon surface. The positive effect of such layer strongly depends on the a-si:h/c-si interface quality. c 2010Elsevier Published B.V. byall Elsevier rights reserved Ltd Keywords: hydrogenated amorphous silicon; silicon heterojunctions; rear contact, modelling. 1. Introduction The interdigitated back contact silicon heterojunction (IBC-SiHJ) solar cell is a promising structure to reach high efficiency [1, 2]. Previous simulations studies of IBC SiHJ solar cells [2, 3] carried out several limiting factors for this type of structure: front surface passivation, quality of a-si:h/c-si interfaces, quality of contacts (series resistance, metal coverage and metal work function). Here, we focus on the front side and the effect of introducing a front surface field (FSF) layer. Positive effects of introducing heavily doped thin crystalline silicon top layer as a FSF have been observed in silicon homojunction back contact solar cells [4, 5]: front surface passivation, UV- * Corresponding author. Tel.: +33-169-851-680; fax: +33-169-418-318. E-mail address: djicknoum.diouf@lgep.supelec.fr. 1876-6102 c 2010 Published by Elsevier Ltd doi:10.1016/j.egypro.2010.07.011
60 D. Diouf et al. / Energy Procedia 2 (2010) 59 64 2 Author name / Energy Procedia 00 (2010) 000 000 stability of front surface passivation and enhancement of fill factor and short circuit current. Here we discuss the interest of using a FSF layer in IBC-SiHJ solar cells. We study the effect of introducing either heavily doped crystalline silicon (c-si) or doped hydrogenated amorphous silicon (a-si:h) as a top layer. 2. Structure and physical models The reference design of the IBC-SiHJ solar cell used in our numerical simulations is presented in Figure 1. The geometrical and material parameters of the simulated device were chosen in agreement with a realistic fabrication process [6]. A 300 μm n-type polished crystalline silicon wafer is used as the substrate. The c-si resistivity is taken equal to 4.8 cm, corresponding to a doping density of 10 15 cm -3. An anti-reflective coating (ARC) corresponding to a standard hydrogenated silicon nitride layer (SiN x :H) is simulated at the front side choosing a 75 nm insulating layer with a wavelength independent refractive index n = 2.05. We defined two electrodes at the rear of the wafer: a p-type a-si:h emitter and an n-type a-si:h back surface field. The 50 μm wide region between the two doped layers (gap region) is modelled with an insulator layer. BSF and emitter are totally covered with aluminum metal contacts. ARC n = 2.05 ; 75nm FSF n-type crystalline Si 150 μm 50 μm 500 μm (n) a-si :H Emitter (p) a-si :H Figure1: Sketch of the reference IBC SiHJ solar cell. For a-si:h layers, the distribution of the density of states (DOS) is a combination of two exponentially decaying band tail states (donor-like valence band tail and acceptor-like conduction band tail) and two Gaussian distributions of deep defect states (one donor-like, the other acceptor-like) [7-9]. The DOS and doping concentrations of each a- Si:H layer were adjusted to set the Fermi level at room temperature at 0.2 ev below the conduction band edge in (n) a-si:h (BSF) and at 0.3 ev above the valence band edge in (p)a-si:h (emitter). Simulations were performed using the numerical device simulator ATLAS from Silvaco International. Doping dependence of carrier mobilities as well as Shockley-Read-Hall and Auger recombination mechanisms were taking into account [10]. The AM1.5G solar spectrum was used for the optical generation to simulate current-voltage, I(V), curves under standard one-sun illumination conditions at an intensity of 100 mw/cm 2. 3. Simulation results The front surface structure of n-type IBC homojunction solar cells traditionally consists in well-passivating antireflective coating made of thermal SiO 2 or SiN x :H layers [11, 12]. The insertion of a lightly doped n+ layer at the c-si front surface can be used on this cell structure [13] to create a so-called Front Surface Field (FSF) which shields the minority carriers from surface recombination [14]. The FSF also enhances lateral current transport in IBC-cells by decreasing the base resistivity [15]. This effect is of particular interest for industrial solar cells using large cell pitches and thinner substrates. For IBC cells which use the low temperature a-si:h/c-si heterojunctions (IBC-SiHJ cells) the front surface structure mostly consists in an a-si:h/arc stack. Here the thin a-si:h layer passivates the surface whereas a SiN x :H [6, 16] or TCO (Transparent Conductive Oxide) [17] layer is used for light trapping. Besides the thermal budget aspect, the main advantage of such low temperature front surface scheme is the possibility to fabricate it at any stage of the cell fabrication. However IBC-SiHJ cells face the same challenges about the base conductivity enhancement
D. Diouf et al. / Energy Procedia 2 (2010) 59 64 61 Author name / Energy Procedia 00 (2010) 000 000 3 than previously discussed for homojunction cells. Here our goal is to study the impact of the front surface structure on the efficiency of IBC Si-HJ solar cells. We compare the influence of three different front surface schemes: single ARC, n+ c-si diffused layer/arc named FSF1 and n-type a-si:h layer/arc named FSF2. The thickness and doping density of the c-si FSF layer in FSF1 are set at 500 nm and 5x10 18 cm -3, respectively. The thickness of the n- type a-si:h layer in FSF2 is set at 10 nm. 3.1. Passivation of the front surface To achieve high efficiency, the front surface passivation is very important in IBC SiHJ solar cell. As seen in Fig. 2, there is a strong dependence of cell performance with surface recombination velocity of carriers in classic IBC SiHJ solar cell (without FSF layer). Since most of the carriers are generated near the front side, while the pn heterojunction is at the back side, poor passivation means high surface recombination velocity of carriers (SRV), and this will cause important carrier recombination before they can reach the back side. The figure shows the positive effect of FSF layer in passivation of the front side. Indeed, there is no degradation in cell performance with increasing minority carrier surface recombination velocity up to 5000 cm/s. This result is very important in industrial conditions where it is difficult to obtain good passivation on large areas. 20 15 (%) 10 5 0 without FSF with FSF2 1 10 100 1000 10 4 10 5 10 6 SRV (cm/s) Figure 2: Influence of surface recombination velocities on the front side of IBC-SiHJ solar cell without and with FSF layers. No interface states were taken into account for FSF2. When using n-type a-si:h (FSF2), recombination can take place at the a-sih/c-si interface. To take account of such interface recombination, we introduced a 1nm thick defective c-si layer at the interface between a-si:h and c- Si. The defect distribution in this interface layer was assumed to be Gaussian, with donor-like defects located at 0.56 ev above the valence band maximum. In the following N SS (in cm 2 ) will denote the interface defect density, which is determined as the product d int *N it where d int is the thickness of the interface layer (1 nm) and N it (in cm 3 ) is the defect density in this layer, which is the integral over the band gap of the Gaussian interface DOS, g it (in cm 3 ev). Capture cross sections for both types of carriers were taken equal to 10-15 cm 2 for these interface states. The IBC SiHJ solar cell output parameters are strongly deteriorated by high interface defects density (N SS > 10 12 cm -2 ) due to recombination at the hetero-interface. As an example, Figure 3 shows the dependence of the solar cell efficiency upon N SS. The impact of this recombination is important because there is more carrier generation at the front surface. 20 18 16 14 (%) 12 10 8 6 4 10 10 10 11 10 12 10 13 10 14 N SS (cm -2 ) Figure 3: Impact on the efficiency of the density of interface defects at the front heterojunction in FSF2 case of IBC-SiHJ solar cells.
62 D. Diouf et al. / Energy Procedia 2 (2010) 59 64 4 Author name / Energy Procedia 00 (2010) 000 000 A correspondence can be made between the degradation induced by higher SRV (structure without FSF layer) and the degradation induced by higher interface defects density (structure with FSF2 layer). Simulations have been done for both structures by increasing SRV and interface defect densities. In Fig. 4, each point corresponds to a couple of values (SRV, N SS ) allowing the same short circuit current on both simulated structures. We found a linear relation between SRV and N SS : SRV = 13 + 6.7 10-11 N SS, SRV and N SS being expressed in cm/s and cm -2, respectively. This linear relation suggests that a perfect FSF2 layer (no interface defects at the a-si:h/c-si heterojunction) leads to an equivalent structure of IBC-SiHJ solar cell without FSF layer with SRV=13 cm/s. The equivalent SRV stays below 50 cm/s for interface defect densities up to 5.5 10 11 cm -2. Such values of interface defect densities are already achievable in actual heterojunction processes, as shown in the next section. 500 400 SRV (cm/s) 300 200 100 0 10 10 10 11 10 12 10 13 N SS (cm -2 ) Figure 4: Equivalence between surface recombination velocity in a structure without FSF and density of interface defects at the front heterojunction in FSF2 case of IBC-SiHJ solar cells. The line is a linear fit to the data. 3.2. Relation with QSSPC measurements N SS is not the best parameter to qualify interface passivation. The product N SS, being the capture cross section of minority carriers is a better choice since recombination also involves capture cross sections. However this is not often used, and it is even not sufficient since recombination also depends on the occupation of defect states, which in turn depends on illumination and bias conditions. The main and most popular way to evaluate the passivation quality of a-si:h/c-si interfaces consists in using the QSSPC (Quasi-Steady State PhotoConductance) technique [18,19] on symmetrical a-si:h/c-si/a-si:h structures. So, we used AFORS HET [20] to calculate QSSPC curves on symmetric FSF (10 nm)/c-si (250 μm)/ FSF (10nm) structures to estimate the effective lifetimes corresponding to the precedingly introduced interface defect densities. 10 4 1sun QSSPC lifetime (µs) 1000 100 10 10 10 10 11 10 12 10 13 10 14 N SS (cm -2 ) Figure 5: Effective lifetime at 1sun obtained from the QSSPC simulation of a symmetrical FSF2/c-Si/FSF2 structure. Fig. 5 shows the decrease of effective lifetime calculated at 1 sun associated to the increase in interface defect density. With such symmetric structures, lifetimes up to 1ms have already been measured by several groups [19, 21]. If we now compare Figures 4 and 5, we conclude that a-si:h FSF layers having equivalent SRV values well
D. Diouf et al. / Energy Procedia 2 (2010) 59 64 63 Author name / Energy Procedia 00 (2010) 000 000 5 below 50 cm/s can be processed. This suggests that very good front surface passivation can be achieved with the use of n-type a-si:h FSF layers. 3.3. Impact on the short circuit current and on the fill factor The fabrication of IBC-SiHJ solar cells using low cost technologies (screen printing, LASER scribing or metallic mask) induces high pitch values. The pitch value of an IBC-SiHJ solar cell, defined as the sum of the BSF width, the gap width and the emitter width, is large, in the millimeter range. It is interesting to know the influence of the FSF layer in the performance of cells with different pitches. So, we varied the pitch value in our simulations. The BSF and gap widths are set at 150 μm and 50 μm, respectively, and we varied the emitter width from 200 μm to 2000 μm. The optimum pitch corresponding to the maximum efficiency is in the range 600 μm to 900 μm. It represents the best trade-off between emitter coverage on the rear side and series resistance losses due to the increased lateral distances. 36 34 J SC (ma/cm 2 ) 32 30 28 W ith FSF2 W ithout FSF 400 800 1200 1600 2000 Pitch (µm) Figure 6: Variations of the short circuit current of IBC-SiHJ solar cells as a function of the pitch value. SRV=50 cm/ for the cell without FSF layer. N SS=5 10 10 cm -2 for FSF2. When we increased the pitch of the cell, emitter coverage on the rear side is increased and BSF coverage reduced. With large pitch (high emitter coverage), majority carriers (electrons) have to flow lateral distances in the millimeter range before reaching the BSF contact which influences the fill factor of the cell. Minority carriers (holes) have a shorter distance to travel before reaching the emitter, resulting in higher short circuit current as seen in Figure 6. 82 80 a/ 84 83,5 b/ FF (%) 78 76 74 72 with FSF2 without FSF 70 400 800 1200 1600 2000 Pitch (µm ) FF (%) 83 82,5 82 with FSF2 w ithout F S F 81,5 400 800 1200 1600 2000 Pitch (µm) Figure 7: Variations of the fill factor of IBC-SiHJ solar cells as a function of the pitch distance. In (a) the c-si wafer resistivity is 4.8 cm, while in (b) it is 0.58 cm. SRV=50 cm/ for cell without FSF layer. N SS=5 10 10 cm -2 for FSF2. With n+ crystalline silicon FSF layer, an electron in the first few micrometers of the c-si can diffuse in the FSF layer where there is negligible lateral resistance. With an (n) a-si:h FSF layer, the conduction band mismatch induces an accumulation layer of electrons in c-si at the hetero-interface. Electrons can thus flow in this accumulation layer with low resistive losses. The FSF layer reduces the lateral resistance losses due to base resistivity and thus enhances the fill factor of IBC-SiHJ solar cells. This effect of the FSF layer is more important in
64 D. Diouf et al. / Energy Procedia 2 (2010) 59 64 6 Author name / Energy Procedia 00 (2010) 000 000 structures with large pitches and high wafer resistivity, as was shown in silicon homojunction back contact solar cells [4]. We also performed simulations of the structure with low base resistivity (0.58.cm). As can be seen in Fig. 7b, no significant fill factor enhancement is found in this case by the introduction of either FSF structure. The lateral resistance induced by the c-si substrate is then too low to impact the fill factor. So, the FSF layer enhances the fill factor for IBC-SiHJ solar cells with large pitches or with high c-si wafer resistivity. 4. Conclusion The influence of front surface field layers has been discussed by 2D numerical simulations. Front surface passivation of IBC SiHJ solar cell needs to be very good to achieve high efficiency. The presence of the FSF layer significantly improves all photovoltaic output parameters: open-circuit voltage, short-circuit current, and fill factor. The FSF layer makes the cell performance insensitive to the surface recombination velocity up to quite high values (5000 cm/s). The FSF layer also reduces the lateral resistance losses due to the increased lateral current through the doped layer. This effect is more important in IBC-SiHJ solar cells with large pitches and high base resistivity. If the doped front layer is made of hydrogenated amorphous silicon, we observe that the positive effect of such a layer is decreased when increasing the density of states at the a-si:h/c-si interface. With interface qualities already achievable with today s technology an a-si:h FSF can provide equivalent front surface recombination velocities well below 50 cm/s. AKNOWLEDGEMENTS This work was partly supported by Agence Nationale de la Recherche (ANR) in the framework of the QCPASSI project, and by European Community's Seventh Framework Programme (FP7/2007-20013) under grant agreement no. 211821 (HETSI project). References [1] R.M. Swanson, Proc. 31st IEEE PVSC (2005) 889. [2] D. Diouf, J.P. Kleider, T. Desrues and P.J. Ribeyron, Mat. Sci. Eng. B (2009) 291. [3] R. Stangl, M. Bivour, E. Conrad, I. Didschuns, L. Korte, K. Lips and M. Schmidt, Proc. of the 22nd EUPVSEC (2007)870. [4] F. Granek, M. Hermle, D.M. Huljic, O. Schultz-Wittmann and S. W. Glunz, Progress in photovoltaics: reseach and applications, 17 (2008)47. [5] M. Hermle, F. Glanek, O. Schultz and S. Glunz, J. Appl. Phys.103 (2008) 054507. [6] T. Desrues P-J. Ribeyron, A. Vandeneynde, A.-S. Ozanne, F. Souche, Y. Veschetti, A. Bettinelli, P. Roca i Cabarrocas, M. Labrune, D. Diouf, J.-P. Kleider, M. Lemiti. Proc. 23rd EPVSEC (2008) 1673. [7] M.H. Cohen, H. Fritsche and S.R. Ovshinsky, Physical Review Letters, 22 (1969) 1065. [8] A.S. Gudovskikh, S. Ibrahim, J.-P. Kleider, J. Damon-Lacoste, P. Roca i Cabarrocas, Y. Veschetti, P.-J. Ribeyron, Thin Solid Films, 515 (2007) 7481. [9] A. S. Gudovskikh, J. P. Kleider, and R. Stangl, J. Non-Cryst. Solids 352 (2006) 1213. [10] User s manual for ATLAS from Silvaco International, version 5. 12.1. [11] R. M. Swanson, S. K. Beckwith, R. A. Crane, W. D. Eaides, Y. O.Wark, R. A. Sinton, and S. E. Swiiwiun, IEEE Transactions on Electron Devices, Vol. Ed-31 (1984) 661. [12] P. Engelhart, N.-P. Harder, A. Merkle, R. Grischke, R. Meyer, and R. Brendel, Photovoltaic Energy Conversion, Conference Record of the 2006 IEEE 4th World Conference,Vol.1, (2006) 900. [13] W.P. Mulligan, D.H. Rose, M.J. Cudzinovic, D.M. De Ceuster, K.R. McIntosh, D.D. Smith, R.M. Swanson, Proc. of the 19th EPVSEC (2004) 387. [14] P.E. Gruenbaum, R.R. King, and R.M. Swanson, J.Appl. Phys. 12 (1989) 66. [15] D.M. De Ceuster, P. Cousins, D. Rose, D. Vicente, P. Tipones and W. Milligan, Proc. of the 22nd EPVSEC (2007) 816. [16] M. Tucci, L. Serenelli, E. Salza, L. Pirozzi, G. de Cesare, D. Caputo, M. Ceccarelli, P. Martufi, S. De Iuliis, L. J. Geerligs, Proc. of the 23rd EPVSEC (2008) 1749. [17] M. Lu, S. Bowden, U. Das, R. Birkmire, Appl. Phys. Lett. 91 (2007) 063507. [18] R.A. Sinton, A. Cuevas, Appl. Phys. Lett. 69 (1996) 2510. [19] S. Olibet, E. Vallat-Sauvain, C. Ballif, Phys. Rev B 76 (2007) 035326. [20] R. Stangl, M. Kriegel and M. Schmidt, Proceeding 4th World Conference on Photovoltaic Energy Conversion, (2006)1350. [21] T. Mueller, S. Schwertheim, M. Scherff and W.R. Fahrner, Appl. Phys. Lett. 92 (2008) 033504.