SCREEN PRINTED N-TYPE SILICON SOLAR CELLS FOR INDUSTRIAL APPLICATION Valentin D. Mihailetchi1, Johann Jourdan1, Alexander Edler1, Radovan Kopecek1, Rudolf Harney1, Daniel Stichtenoth2, Jan Lossen2, Tim S. Böscke2, Hans-Joachim Krokoszinski2 1 International Solar Energy Research Center (ISC), Rudolf-Diesel-Str. 15, D-78467 Konstanz, Germany, Phone: +49 7531 3618348; Fax: +49 7531 3618311; Email: valentin.mihailetchi@isc-konstanz.de. 2 Bosch Solar Energy AG, Wilhelm-Wolff-Str. 23, D-99099 Erfurt, Germany ABSTRACT: The rapid fall in module prices demands researchers to come up with substantial efficiency improvement and cost reductions in the solar cell and module processes. This paper presents a solar cell development on n-type Cz-Si substrates with homogeneous diffusion of boron emitter and phosphorous backsurface-field. The resulting solar cell is bifacial and it is fabricated using only industrial compatible techniques. The best achieved solar cell efficiency, using a screen printed and firing through metallization, on 241 cm2 (total area) wafers is 18.6%. Moreover, we present a new method to passivate boron emitters which substantially reduces surface recombination resulting in improved VOC of the cells. We show that internal quantum efficiencies of these n-type cells when illuminated from the front- and from the back side are identical when good quality CzSi substrates are used in the fabrication. 1 INTRODUCTION The rapid fall in module prices leads researchers to intensify their research on efficiency improvement of solar cells and to further reduce the costs of the fabrication process. Additionally, new cell concepts are currently preparing to enter into production. One of these cell concepts that recently receive much attention is based on n-type silicon substrates that feature a front boron emitter, a rear phosphorous back-surface-field (BSF), as well as front- and rear screen printed metallization grid. Efficiencies up to 18.5% have been reported on large area for such n-type solar cells fabricated using industrial applicable techniques [1], thus exceeding the typical efficiencies currently achieved by a similar process based on p-type substrates. N-type silicon has long been proven to have higher tolerance to common transition metal impurities, potentially resulting in higher minority carrier diffusion lengths compared to p-type substrates [2,3]. Additionally, the minority carrier lifetime does not suffer from light induced degradation due to the Boron-Oxygen related defect which is commonly present in p-type Cz-Si [4]. In spite of these fundamental material advantages, n-type cells comprising homogeneous front emitter and rear BSF are not readily produced industrially due to insufficient understanding and the lack of a cost-effective production process allowing for high conversion efficiencies. It is crucial to improve the boron diffusion to form the emitter, passivation of boron emitters, and metallization. This paper presents a solar cell process based on ntype monocrystalline wafers with boron front emitter, and low cost fabrication processes ready for industrial use (such as screen printing and microwave plasma enhanced chemical vapour deposition (PECVD) of hydrogenated silicon nitride (SiNx)). Using this process we demonstrate efficiencies of 18.6% on 156×156 mm2 monocrystalline Cz-Si wafers. Additionally a new method to passivate boron emitters is presented which results in an excellent passivation quality. This new method is simpler and more cost-effective compared to well-established methods, such as those based on atomic layer deposition of Al2O3 [5] and thermally or chemically grown SiO2 layers [6-8]. We also show that internal quantum efficiencies of these n-type cells when illuminated from the front- and from the back side are identical, if the process conditions are adjusted with the aim to obtain bifacial cells. 2 EXPERIMENTAL DESIGN We developed our n-type process on 156×156 mm2 phosphorous-doped monocrystalline Czochralski (Cz) Si wafers with base resistivity ranging between 1 and 10 Ωcm. The n-type Cz crystals are grown and sliced into wafers at Bosch Solar Energy AG in Arnstadt, Germany. The crystal growth and wafering is performed on industrial equipment. Figure 1 shows a schematic crosssection of our complete fabricated n-type solar cell. The solar cell process comprises an alkaline texture on the front side and a polished back side. On the front side (light receiving side), a p+ emitter with a sheet resistance of typical 60 Ω/square was diffused from a boron tribromide (BBr3) source in a quartz tube furnace of an industrial scale. The BSF on the back side is formed in a separate diffusion step by diffusing phosphorous in a similar quartz-tube furnace using a POCl3 source. The passivation of the BSF diffusion region is achieved by a PECVD SiNx deposition while for the front boron emitter the passivation is achieved using an in-house developed method consisting in a passivating layer and a SiNx antireflection coating stack. The fabrication process includes also the necessary cleaning steps using baths containing HCl, HF, H2O2/H2SO4, and rising water. The metallization on the front and back side was applied by screen printing Figure 1: Schematic cross-section of the fabricated ntype solar cells. and firing through of metal pastes using an H-pattern screen design. The cell area is 241 cm2 which equals to the area of our pseudo squared 156×156 mm2 monocrystalline wafer with a diagonal of 205 mm. Hence no edge isolation by snapping the edges or similar tricks was applied. 3 RESULTS AND DISCUSSION 3.1 Passivation of boron emitters One of the drawbacks in the development of an industrially compatible n-type cell process is the passivation of the boron emitter. As PECVD-SiNx, which is commonly used to passivate phosphorous emitters in the standard p-type process, does not passivate boron emitters, a considerable amount of effort has been put to find new methods to passivate boron emitters. Currently the most successful methods used are stacks in which the top layer is SiNx used as antireflection coating layer and as hydrogen source for the passivating layer(s) underneath. Among the passivation layers used in combination with SiNx are: a thin atomic layer deposition of Al2O3 [5], amorphous silicon deposition [9], thermally grown SiO2 [6,7], chemically grown SiO2 [8]. However, some of these methods are quite costly, other yield only layers of limited quality. In our quest to continuously improve the passivation quality of boron emitter and to reduce the costs, we have developed a new method for passivation. The method consists of a thin passivating layer and a SiNx anti-reflection coating stack. Details about the properties of the passivating layer and its application method will be the subject of a later publication. Herein we have designed an experiment to test the passivation quality of our new method compared with other known methods used to passivate industrialtype boron diffused emitters. The test structure consists of a polished n-type monocrystalline Cz substrate symmetrically diffused with 60 Ω/square boron emitter on both sides. After the necessary cleaning steps the passivation layer(s) are applied on p+ diffused regions. The samples were subsequently subjected to a firing step followed by a QSSPC measurement. From this the effective recombination lifetime τeff , the implied VOC at 1 sun intensity, and the emitter saturation current density JoE at high injection level (∆n=2×1016 cm-3) were extracted. A schematic cross-section of such a test structure is shown in figure 2. Figure 3 compares the τeff and JoE of our passivation method against several other passivation methods that includes a wet thermal SiO2/SiNx stack, a chemical SiO2/SiNx stack (NAOS method), and just a SiNx layer. The results in figure 3 clearly demonstrate the excellent passivation potential of our developed method, with τeff =517 µs, implied VOC=683 mV, and JoE = 14 fA/cm2 per side. Figure 2: Schematic cross-section of the fabricated test structures to investigate the passivation of boron emitters. Figure 3: Effective minority carrier lifetime τeff measured on a p+np+ test structure at 1 sun illumination and emitter saturation current density JoE measured at an injection level of 2×1016 cm-3 as a function of different methods to passivate the p+ (boron emitter) region. The passivation methods tested in this study are: a silicon nitride layer (SiNx), a stack of chemically grown SiO2 in a nitric acid solution (NAOS) and SiNx, a stack of wet thermally grown SiO2 (20 nm) in a quartz tube furnace and SiNx, and our new developed method consisting in a stack of a passivating layer and SiNx. These results are comparable to the best achieved passivation of p+ surfaces up to now, namely the atomic layer deposition of Al2O3 and SiNx stack [5]. Although thermal SiO2/SiNx stack produce very good passivation as well, as can be seen from the results of figure 3, it requires, however, an extra high temperature step in the solar cell process. 3.1 Solar cell results and discussion The excellent passivation results obtained on boron emitters were then applied to fabricate n-type cells using the process described above and schematically shown in figure 1. On the front boron emitter we have applied our passivation method while on the phosphorous BSF we deposited the commonly used SiNx passivating layer. The metallization was applied on both sides of the cells by screen printing metal pastes. The cells went through a firing step to complete the contact formation. Table 1 shows the solar cell parameters of our best fabricated cell and a solar cell with the best bifacial efficiencies ratio. The best achieved efficiency is 18.6%, and it is certified by Fraunhofer ISE CalLab in Germany. It should be noted that all solar cells that we have processed so far have a polished rear side and a random pyramid texture on the front side. This results in different short-circuit current densities (JSC) under front and back side illumination. However, this effect could be eliminated by comparing the internal quantum efficiencies (IQE) measured under front and back side illumination conditions. Table I: Comparison of the best solar cell parameters measured under standard test conditions (AM1.5G, 100 mW/cm2, 25 o C). The solar cells were fabricated on 241 cm2 monocrystalline n-type Cz Si with base resistivity of 8 Ωcm. The result for the best efficiency of a bifacial fabricated cell is also shown. The implied VOC value was measured using QSSPC on a similar cell but without the metal contacts on front and back side. Substrate, Surface Jsc implied Voc FF pseudo-FF Voc η illumination texture [mA/cm2] [mV] [mV] [%] [%] [%] Cz, best, front Rand. pyramids 38.3* 637* 664 76.5* 82.5 18.6* Cz, front Rand. pyramids 38.5 634 75.2 18.3 Cz, back alkaline polished 34.2 631 75.9 16.4 *values measured by ISE CalLab. Figure 4 shows the IQE measurement of our best cell (black circles) illuminated from the front side. The response observed at short wavelengths (below 0.5 µm) confirms the excellent passivation of boron emitter (60 Ω/square) demonstrated in figure 3. Moreover, figure 4 also shows the IQE data of an optimized bifacial cell, which have different BSF sheet resistance and metallization grid as compared with cell optimized only for the front side. The IQE results of the bifacial cell show that independent from which side the cell is illuminated the response of the cell is the same. This demonstrates on one hand the outstanding passivation of the diffused surfaces and on the other hand the very high lifetime of the material, which obviously does not suffer substantially during the process. A symmetrical IQE from both sides is an excellent characteristic because it allows us to fabricate bifacial cells with the same power conversion efficiencies from the front- or back side illumination, providing that the back surface is also textured. Based on the experimental data shown in table 1 and figure 4 the major difference between front side and back side performance of the cells is in the JSC. Therefore work is under way to adapt our cell process to allow for both sides to be textured in order to match the JSC. An advantage of realizing cells with equal performance from the front- or back side (fully bifacial) can be taken in the module interconnection. If such a performance match is desired for certain applications, then equal performance match will allow for the introduction and fabrication of planar interconnected ntype cells in a module. A schematic representation of our proposed Planar Interconnected N-type Cells (PINC) in a module is shown in figure 5. Figure 4: Internal quantum efficiency (IQE) of our best fabricated cell (confirmed by ISE CalLab), and of a bifacial cell illuminated from the front- and back side. Figure 5: Schematic representation of our proposed Planar Interconnected N-type Cells (PINC) in a module of our fully bifacial cells. In the PINC module the boron emitter side of a cell is planar connected with the BSF side of an adjacent cell. 4 CONCLUSSION We have presented a simple industrial process to fabricate n-type solar cells with boron front emitter and phosphorous back-surface-field which leads to an efficiency of 18.6% for large area (241 cm2) monocrystalline Cz-Si substrate. We demonstrated that this cell process results in identical internal quantum efficiencies when a cell is illuminated from the emitter side or from the back-surface-field side. This feature allows for the fabrication of fully bifacial cells, with identical efficiencies from both sides, and to introduce an innovative cells interconnection in the module. Additionally, we introduce a new method to passivate boron emitters for industrial application with excellent passivation quality. ACKNOWLEDGEMENTS This work is financially supported by German government (BMU) within EnSol project, contract number 0325120A. REFERENCES [1] A. Weeber et al., Proc. of the 24th European Photovoltaic Solar Energy Conf., p. 891 (2009). [2] D. MacDonald, L. J. Geerligs, Appl. Phys. Lett. 85, 4061 (2004). [3] J. E. Cotter et. al., 15th Workshop on Crystalline Silicon Solar Cells & Modules: Materials and Processes, p. 3 (2005). [4] K. Bothe, R. Sinton, and J. Schmidt, Prog. Photovoltaics 13, 287 (2005). [5] J. Benick, et. al., Proc. of the 24th European Photovoltaic Solar Energy Conf., p. 863 (2009). [6] J. Libal et al., Proc. of the 20th European Photovoltaic Solar Energy Conf., p. 793 (2005). [7] M. J. Kerr, Ph.D. thesis, Australian National University (2002). [8] V. D. Mihailetchi, Y. Komatsu, L. J. Geerligs, Appl. Phys. Lett. 92, 063510 (2008). [9] P. P. Altermatt et al., Proc. of the 21st European Photovoltaic Solar Energy Conf., p. 647 (2006).