Specification of solar grade silicon: How impurities affect efficiency Bart Geerligs
Outline Objective and introduction Ingots and cells from artificially contaminated silicon feedstock: Titanium Aluminum Analysis of results Implications for feedstock specification 2
Objective Determine allowable concentrations of impurities in silicon feedstock for mc-si solar cells. Reasons: specific Silicon produced for Photovoltaics possibility of Low-cost and abundant Silicon feedstock from carbothermic reduction of quartz. SiO 2 + 2C Si + 2CO Fe, Ti, Al, and C are major impurities in silicon from carbothermic reduction. What are the target levels for these impurities? 3
Which specs are available? Si wafer manufacturers: we like to be on the safe side, the SEMI poly-si spec [<0.1 ppmw total metals] works for us For PV, there exist more specific earlier studies: J.R. Davis, et al., IEEE Trans El. Dev. ED-27, 677 (1980) Cz-growth also, Fally et al., Revue Phys. Appl. 22, 529 (1987) mc-si, but no info on Ti 4
Experimental procedure poly-si feedstock directional ingot with added impurity solidification & wafers furnace 5
Experimental procedure poly-si feedstock directional ingot with added impurity solidification & wafers furnace industrial in-line cell process SiNx:H coating 14.5 15% cell efficiency 6
Titanium 10 ppmw (parts-per-million by weight) of Ti were added to the feedstock 20 J sc V oc (W/cm 2 ) 18 16 14 15% rel 25% rel reference S10: Ti 0 20 40 60 80 100 position in ingot towards top (%) 7
Titanium 10 ppmw (parts-per-million by weight) of Ti were added to the feedstock 20 0 J sc V oc (W/cm 2 ) 18 16 14 reference S10: Ti 0 20 40 60 80 100 IQE (%) -50-100 ingot S10, position in ingot: 13% 19% 24% 54% 65% 86% 400 600 800 1000 1200 position in ingot towards top (%) wavelength (nm) Reduction of J sc due to strongly reduced red-response in IQE 8
Aluminum 5 ppmw of Al were added to the feedstock 20 J sc V oc (W/cm 2 ) 16 12 15% rel 25%rel reference S6: Al 0 20 40 60 80 100 position in ingot towards top (%) 9
Aluminum 5 ppmw of Al were added to the feedstock 20 0 J sc V oc (W/cm 2 ) 16 12 reference S6: Al IQE (%) -50 ingot S6, position in ingot: 14% 30% 53% 69% 0 20 40 60 80 100-100 400 600 800 1000 1200 position in ingot towards top (%) wavelength (nm) Reduction of J sc again due to reduced red-response in IQE 10
Analysis of results 15-25% reduction of J sc V oc due to 5 ppmw Al or 10 ppmw of Ti is too much to be acceptable. How can we determine the maximum allowable concentration? Can the cell efficiency for other concentrations be modeled and predicted? Is there experimental data to verify such a model? 11
Model for analysis segregation during ingot growth impurity concentration 1 C s 1 x C s (a.u.) 5 4 3 2 1 0.0 0 0.2 20 0.4 40 0.6 60 0.8 80 100 1.0 position position in ingot in towards ingot x top (%) 12
Model for analysis segregation during ingot growth impurity concentration If impurity dominates recombination 1 L 2 eff τ 1 eff 1 C s 1 x C s L eff in solar C s (a.u.) cell (a.u.) 15 4 3 2 top bottom 01 0.0 0 0.2 0.4 (1-x) 0.6 0.8 1.0 1 (square root of position in in ingot xtowards bottom) L eff 1 x L eff can be determined from the red-response of the IQE 13
Comparing L eff from IQE with model L eff follows expected decrease to top of ingot. (exception: bottom of S6) L eff (µm) from IQE 150 100 50 S10 (Ti) S6 (Al) top bottom 0 0.0 0.2 0.4 0.6 0.8 1.0 (1-x) (square root of position towards bottom of ingot) 14
Comparing L eff from IQE with model L eff follows expected decrease to top of ingot. (exception: bottom of S6) Conclusion: Relation between feedstock contamination and recombination is linear (no non-linear effects from precipitation, etc.). L eff (µm) from IQE 150 100 50 S10 (Ti) S6 (Al) top bottom 0 0.0 0.2 0.4 0.6 0.8 1.0 (1-x) (square root of position towards bottom of ingot) 15
Comparing L eff from IQE with model L eff follows expected decrease to top of ingot. (exception: bottom of S6) Conclusion: Relation between feedstock contamination and recombination is linear (no non-linear effects from precipitation, etc.). L eff (µm) from IQE 150 100 50 S10 (Ti) S6 (Al) top bottom 0 0.0 0.2 0.4 0.6 0.8 1.0 (1-x) (square root of position towards bottom of ingot) (at least for Al, Ti, for the used concentrations and probably lower) 16
Do-It-Yourself specification of solar grade silicon 1. Construct PC1D model for your cell process, and calculate cell efficiency versus L eff 2. Use 1/L eff2 C L (C L is impurity concentration in the feedstock) generic plot of cell efficiency versus C L (C L in a.u.). 3. One data point (impurity concentration and cell efficiency) to calibrate C L -scale. 4. Choose acceptance level of cell efficiency (cost analysis! e.g. 97% rel efficiency if feedstock 25% lower cost). 5. Read required impurity concentration from plot. 17
Graphical presentation of D-I-Y feedstock specification cell efficiency (% rel ) relative to high-purity feedstock 1.0 0.9 0.8 0.7 1 10 100 impurity concentration (a.u.) 14.5% cell techn. 17% cell techn. our results: 5 ppmw Al or 10 ppmw Ti 18
Graphical presentation of D-I-Y feedstock specification cell efficiency (% rel ) relative to high-purity feedstock 1.0 0.9 0.8 0.7 60x reduction 1 10 100 impurity concentration (a.u.) 14.5% cell techn. 17% cell techn. our results: 5 ppmw Al or 10 ppmw Ti 19
Graphical presentation of D-I-Y feedstock specification cell efficiency (% rel ) relative to high-purity feedstock 1.0 0.9 0.8 0.7 spec: 0.1 ppmw Al or 0.2 ppmw Ti 60x reduction 1 10 100 impurity concentration (a.u.) 14.5% cell techn. 17% cell techn. our results: 5 ppmw Al or 10 ppmw Ti 20
Conclusions Clear impact of Ti and Al at ppm level. Dependence of impact on position in ingot modeled according to segregation and linear relation between L eff -2 and C feedstock. Extrapolated feedstock specification based on 3% rel cell efficiency reduction: Al: 0.1 ppmw Ti: 0.2 ppmw See the paper for more details, also on carbon, mix of impurities, Fe, and modelling of economics! 21
Thank you for your attention Acknowledgements Oyvind Mjøs, NTNU Trondheim ScanArc, Scanwafer, HCT EC for contracts SOLSILC, SPURT, and SISI Coauthors: Petra Manshanden, Paul Wyers (ECN Solar Energy), Eivind Øvrelid, Ola Raaness, Aud Waernes (Sintef), Benno Wiersma (Sunergy) 22 22