CRYSTALLINE Si SOLAR CELLS FROM COMPENSATED MATERIAL: BEHAVIOUR OF LIGHT INDUCED DEGRADATION Radovan Kopecek, Jayaprasad Arumughan and Kristian Peter International Solar Energy Research Center - ISC Konstanz, Rudolf-Diesel Str.15 , D-78467 Konstanz, Germany e-mail: radovan.kopecek@isc-konstanz.de Ethan A. Good Dow Corning Corporation, 2200 W. Salzburg Road, Midland, Michigan, United States Joris Libal, Maurizio Acciarri and Simona Binetti Department of Material Science, Università di Milano-Bicocca, via Cozzi 53, I-20125 Milano, Italy ABSTRACT: In this paper we focus on the detailed investigation of light induced degradation behaviour of cells from so called Solar Grade silicon (SoG-Si) material. Different types of materials were studied at various compensation levels. Large area industrial solar cells without surface texture with high efficiencies on mc-Si (>15%) as well as Cz-Si (>16%) substrates were fabricated and exposed to long time illumination at 1 sun. We observed that on one hand the light induced degradation (LID) was comparable to reference substrates and on the other not dependent on the absolute but on the net boron content only. One possible explanation for this unforeseen finding is that the phosphorus atoms in the SoG-Si material interact with boron centers reducing the recombination activity of e.g. B-O complexes. Keywords: crystalline Si solar cells, SoG-Si, light induced degradation INTRODUCTION Presently many silicon purification technologies based on low-cost approaches are being developed by different companies leading to so called Solar Grade silicon (SoG-Si) material [1, 2, 3]. Such Si materials may have a higher content of metallic impurities as well as of dopants (e.g. boron, phosphorus) as compared to the electronic grade Si wafers from highly purified and chemical vapour deposited poly-Si. However, the so called Siemens process is currently constrained by capital expansion costs. The commonly applied industrial solar cell process based on P-diffusion and firing through PECVD SiNx by metal pastes includes various effective gettering and passivation steps itself which additionally purify the Si bulk from highly diffusive impurities. These processes lead to solar cells with high efficiencies without the need of extremely pure Si base material. A threshold concentration exists whereby increased dopant compensation might have a negative impact on cell performance as well as on the long term stability compared to p-type devices with a negligible content of phosphorus. 2 MATERIAL USED For our studies directionally solidified mc-Si material as well as mono-crystalline Cz-Si ingots were used. 2.1 mc-Si ingots Two ingots were fabricated with a similar range in resistivity as can be seen in Figure 1. For the mc-Si reference highly pure poly-Si material was used whereas, for the second ingot, 10% of SoG-Si material was blended with poly-Si. 2.2 Cz-Si ingot As for the mc SoG-Si ingot 10% SoG-Si feedsock from a different supplier was admixed to pure poly-Si material for Cz-growth. In this case additional phosphorus was added which resulted in higher substrate resistivities as depicted in Figure 1. specific resistivity [Ωcm] 1 Cz SoG-Si mc SoG-Si mc reference 10 1 0,1 0 20 40 60 80 100 fraction solidified after caps removal [%] Figure 1: Specific substrate resistivity as a function of fraction solidified. The top and bottom parts have been removed prior to wafering. The red curve decreases towards the right which is the part that solidifies the last in the crystallisation process. For both compensated materials, due to different segregation coefficients of boron as compared to phosphorus, the resistivity increases. The lastly solidified part has the highest compensation level. I case of the CzSi ingot the compensation ratio Rc=(Na+Nd)/( Na-Nd) is increased from 1.2 at the top of the ingot (left side of Figure 1) to almost 23.5 at ingot bottom. The calculated hole carrier concentration (electrically active B-atoms) increase from 9x1015cm-3 to 1.4x1016cm-3. More details can be found in [5]. SOLAR CELL PROCESS AND RESULTS Large area wafers from all three ingots were used for the processing of industrial solar cells. 3.1 Solar cell process The process flow chart shown in Figure 2 was used for both – mc- and Cz-Si solar cells. saw damage removal and wafer cleaning 17 solar cell efficiency [%] 3 16 15 14 13 Cz SoG-Si mc SoG-Si mc reference 12 11 10 POCl3 emitter-diffusion 0 20 40 60 80 100 fraction solidified after removal of caps [%] PECVD SiNx Figure 3: Efficiencies of untextured mc- and Cz-Si solar cells as a function of solidified fraction. screen printing of front and rear, drying co-firing 4 edge isolation Figure 2: Solar cell process flow chart (top) and large area mc-Si (left) and Cz-Si (right) solar cells processed with these steps (bottom). The process starts with alkaline saw damage etching and wafer cleaning. Texturing was not included as for our investigations it was not a priority to reach highest efficiencies possible. The P-emitter was formed on both sides in an open tube furnace and subsequently one side was covered by a PECVD SiNx. The front and rear contacts were screen-printed and co-fired in an metal belt furnace. At the end the pn-junction was isolated by a dicing saw. In the bottom part of Figure 2 two pictures demonstrate the finished mc-Si and Cz-Si solar cells. 3.2 Results Wafers selected through the entire ingot were processed and characterised. Figure 3 summarises the results of solar cell efficiencies as a function of ingot position. Even without textured front surface high efficiencies such as >15% for mc-Si substrates and >16% for Cz-Si substrates were reached. In the case of the mcSi solar cells the influence of the bottom part of the ingot is visible through an efficiency drop. However the drop is more pronounced for the mc-Si reference suggesting that the drop is due to the in-diffusion of impurities from the crucible. LIGHT INDUCED DEGRADATION STUDIES In the last paragraph we have shown that high efficiencies can be achieved for solar cells on SoG-Si material on wafers from all parts of the ingot. However there is some concern that such cells could behave differently compared to “classical ones” (from pure polySi) when illuminated for long time. Therefore we have performed light induced degradation studies for the three types of solar cells and from different ingot positions. 4.1 Experimental set up An array of six 500W lamps was illuminating the cells from a distance of about 40 cm (Figure 4) exposing them to a light intensity about 1 sun and heating them to around 50°C (typical temperature in a module). Figure 4: Setup for the measurement of light induced degradation. Six 500W lamps are mounted in a distance of about 40 cm from the cells illuminating them with about 1sun and heating to 50°C. 4.2 Measurements on solar cells The solar cells were measured before and after such long term illumination and the difference of the cell parameters was calculated. Two measuring set-ups were used for characterisation: a sun simulator measuring all relevant cell parameters and a sunsVoc set-up from Sinton measuring the Voc drop of the cells. 10 illumination: one week at 1sun measurement: sun-simulator ∆Voc [mV] 2 ∆Jsc [mA/cm ] ∆η [%] ingot. As the interstitial oxygen concentration is comparable for both ingots and the boron concentration higher for the SoG-Si ingot the formation of B-O complexes must be somehow suppressed. 4.4 Results for mono Cz-Si cells Additional studies on mono c-Si solar cells from SoG Cz-Si wafers have confirmed our previous findings on mc-Si cells. Five Cz-Si cells from different ingot positions were selected and LID was measured as a function of time. The resulted decay in Voc is depicted in Figure 6. 0 position from top 100 (first solidified) 800 900 1000 1100 (last solidified) T=50°C -5 ∆ Voc [mV] 4.3 Results for mc-Si cells In our first experiments we measured the drop in efficiency of cells from the mc SoG-Si ingot in order to gain information about all relevant cell parameters. Figure 5 (on top) depicts the average drop in Voc, Jsc and in the cell efficiency for 3 ingot positions. A clear trend, namely a decrease in degradation, of all cell parameters is observed toward the top of the ingot. Such a trend implies that the degradation is dominated by the formation of B-O complexes as the interstitial oxygen concentration decreases towards the ingot top. After temperature treatment of the cells at 200°C for 10 minutes the degradation was regenerated completely which underlines our assumption. It is much more accurate to measure the Voc as it is not dependent on light intensity as Jsc thus in our following measurements the sunsVoc set-up was used for comparative studies. high boron content high compensation (∆Voc=8 mV) -10 -15 degradation 1 -20 -25 0,1 low boron content low compensation (∆Voc=20mV) 0 10 20 time [h] 0,01 bottom middle illumination: 2 days at >1sun 6 ∆Voc [mV] 5 Figure 6: Light induced degradation curves as a function of time for solar cells from different positions of the CzSi ingot. The plotted lines are guides to the eyes only. top sunsVoc reference mc SoG-Si sun simulator mc SoG-Si 4 3 2 1 0 bottom middle top Figure 5: Degradation in Voc, Jsc and efficiency for SoG-Si cell (top) and in Voc (bottom) as a function of fraction solidified for reference and SoG-Si solar cells. Figure 5 (at bottom) shows the summarised results of degradation measurements on the mc-Si cells from bottom, middle and top ingot position. The Voc from the top graph is included as well (blue dashed bars). It is visible that the measurements with sunsVoc apparatus and sun-simulator are in a very good agreement. In addition, what is quite surprising, the degradation in Voc compared to the reference mc-Si solar cells is identical for bottom positions and even much lower for cells from the upper part of the It is striking that the degradation in Voc is decreasing toward ingot bottom (in contrast to the mc-Si ingot in CzSi growth the bottom is the latest solidified part). A remarkable difference of 12 mV was observed after a LID after 20 hours illumination. Note that the oxygen content was identical (18ppma) for top and bottom ingot position as shown in reference [5] which is not typical for Cz-Si ingots. The segregation coefficient of boron is lower than 1 thus the overall boron concentration in the bottom part is the highest. But, as the segregation coefficient for phosphorus is lower than that of boron, the compensation rate is also higher in the bottom part leading to a lower net B-content and thus higher resistivities. Not only that P is compensating the free charge carriers leading to lower conductance of the substrates. By its presence it also somehow reduces the LID. Our hypothesis is that it forms a complex with B-atoms shielding them from interstitial oxygen and therefore preventing from the formation of B-O complexes. An indication for this was found in our photoluminescence spectra, where an additional peak was observed with an intensity dependent on the degree of compensation. Experiments are underway to prove our findings. In addition we have studied the regeneration process on our Cz-Si solar cells finding slower regeneration behaviour as compared to A. Herguth [6]. This difference can be explained by different O-, P- and B-concentrations (and possibly other species) in our solar cells. 5 CONCLUSIONS In the present study we have focused on the light induced degradation behaviour of industrially processed cells on mc- as well as on Cz-Si wafers from SoG-Si originating from different metallurgical purification techniques. Efficiencies above 15% have been reached for mc-Si cells and above 16% for Cz-Si cells. Solar cells from selected ingot positions have been illuminated at 1 sun and the degradation monitored in comparison with reference cells on substrates from pure poly-Si material. As one of the most prominent degradation mechanisms is based on the formation of recombinative B-O complexes, the compensated material with a higher B content was expected to show increased degradation behaviour. However this was not confirmed as the degradation magnitude was found to be dependent on the resistivity, which means on the net B-content only. We assume that the formation of B-O recombination centers during illumination can be reduced by the appearance of P-B pairs which was already suggested by Kruehler in the late 80s [4]. The P-atoms seem to form a bond with the Batoms shielding them from interstitial oxygen. An indication for this was found in our photoluminescence spectra, where an additional peak was observed with intensity dependent on the degree of compensation. We believe that this peak corresponds to P-B pairs which reduce the formation of B-O complexes under illumination and thus the light induced degradation based on this mechanism. ACKNOWLEDGEMENTS This work was supported by the European Commission within the FoXy project under the contract number 019811(SES6). REFERENCES [1] N. Yuge et al., Purification of metallurgical silicon up to solar grade, Sol. Energy Mater. Sol. Cells, 34 (1994) 243 [2] K. Peter et al., Investigation of Multicrystalline Silicon Solar Cells from Solar Grade Silicon Feedstock, Proc. Of the 20th EU PVSCEC, Barcelona (2005), 615 [3] K. Peter et. al., Future Potential for SoG-Si Feedstock from Metallurgical Process Route, these proceedings [4] W. Kruehler et al., Effect of PhosphorousCompensation on the Electronic Properties of SolarGrade Silicon, Proc. 8th EU PVSEC (1988), 1181 [5] J. Libal et al., Effect of Compensation and of metallic impurities on the electrical properties of Cz-grown solar grade silicon, Journal of Applied Physics, in press [6] A. Herguth et al., Further Investigations on the Avoidance of Boron-Oxygen Related Degradation by Means of Regeneration, Proc. Of the 22nd EU PVSCEC, Milan (2007)