- Email: [email protected]
Control of Dislocation Cluster Formation and Development in Silicon Block Casting
Vailable online at www.sciencedirect.com
Energy Procedia 27 (2012) 66 – 69
SiliconPV: April 03-05, 2012, Leuven, Belgium
Control of dislocation cluster formation and development in silicon block casting D. Oriwola*, M. Hollatza, M. Reineckea a
SolarWorld Innovations GmbH, Berthelsdorfer Str. 111a, Freiberg D-09599, Germany
Abstract Dislocation clusters in multicrystalline silicon are known to be detrimental for photovoltaic cell efficiency. They reduce minority carrier lifetime and are able to build shunts. To get control of them is one of the main purposes of recent developments in crystal growth. For reducing dislocation generation and multiplication there are several issues to focus on during silicon block casting like crystallization condition, grain structure evolution, history of temperature field and phase boundary shape. Different industrial standard ingots were investigated by optical photography. It is shown, that dislocation cluster reduction can be achieved by using suitable conditions during ingot growth.
© Selection andand peer-review under responsibility of theofscientifi c committee of © 2012 2012Published Publishedby byElsevier ElsevierLtd. Ltd. Selection peer-review under responsibility the scientific the SiliconPV conference. committee of 2012 the SiliconPV 2012 conference Keywords: dislocation cluster; multicrystalline silicon; block casting, photovoltaic
1. Introduction Multicrystalline silicon for photovoltaic application has a market share up to 40%. Hence there are serious attempts to improve crystal quality and reduce crystal defects like impurities and dislocations. Dislocations are one dimensional defects in the crystal structure which are known to be heterogeneous distributed over wafer surface [1-2]. They build up so called dislocation clusters and can reach high dislocation densities up to 107cm-2. These areas crucially decrease minority carrier lifetime and so the cell efficiency [3-4], also they are able to introduce shunts [5]. The aim of the industrial crystal growth development is to optimize the block cast process and to improve crystal quality.
1876-6102 © 2012 Published by Elsevier Ltd. Selection and peer-review under responsibility of the scientific committee of the SiliconPV 2012 conference. doi:10.1016/j.egypro.2012.07.030
D. Oriwol et al. / Energy Procedia 27 (2012) 66 – 69
2. Experimental Multicrystalline silicon ingots were produced by industrial block cast technique, using three different growth processes with changes in crystallization conditions. The ingots were cut into bricks and they were sliced to wafer. About 40 bricks from different ingots were used. For every brick a selection of wafers distributed over the ingot height were chosen for further analysis. They were defect etched and then photographed under special light conditions. This gives a flip-book over ingot height where both dislocation etch-pits and grain structure are visible. Additional the ingot height for every wafer is known.
3. Results & discussion 3.1. Optical photography Fig. 1 shows a typical wafer with its grain structure and etch-pits distribution (dark contrast). Like already mentioned in the reference [1-2] the dislocation distribution is very inhomogeneous over wafer surface. The dislocations build up areas of high concentration, which are referred to be dislocation clusters.
Fig. 1. Optical photography of a defect etched wafer revealing grain structure and etch-pits distribution (image length: 75 mm)
3.2. Dislocation cluster seeds The development of a dislocation cluster is demonstrated in Fig 2. The pictures show the generation and further development of a dislocation cluster at a grain boundary from bottom to top. The site of the first arise of etch pits and dark contrast, respectively, is said to be a dislocation cluster seed. To each cluster seed an ingot height and relation to the grain structure can be allocated (sketched in Fig. 4 left and center).
Fig. 2. Three pictures from bottom (left) to top (right) of a small area within a silicon brick showing the generation (center) and development of a dislocation cluster. The height difference between the pictures is about 3 mm (image length: 15 mm)
67
68
D. Oriwol et al. / Energy Procedia 27 (2012) 66 – 69
25
15 10 5
Re
lat
ive
In
0-10-20 10 -30 20 -40 30 -50 go 40 -60 50 -70 th 60 -80 eig 70 -90 ht 80 00 [% 90-1 ]
All
0
Occurence
20
of cluster seeds [arb. units ]
35 30
C B A
Fig. 3. Occurrence of dislocation cluster seeds over ingot height. Blue: all investigated bricks. Green, orange and magenta: groups of growth processes showing different behaviors in cluster seed generation
Fig. 4. Left: sketch of a dislocation cluster in a side view on a brick. The red circle marks the cluster seed. Center: sketch of a top view on a wafer with three grains. The coloured lines are examples for cluster seeds related to grain boundary (yellow), triple point (green) or any site within a grain (blue). Right: Diagramm showing the relative frequency where cluster seeds occur regarding to grain strucutre
Fig. 3 contains a diagram displaying the occurrence of cluster seeds over ingot height. Over all 40 bricks (blue bars) there is a clear evidence that cluster seeds mainly occur in the low region of ingot height. Over 70 % of cluster seeds are generated in the lower 20 % of ingot height. The amounts of new cluster seeds strongly decrease in higher regions. If compared all three process types (referred to be A, B and C) a change in the behavior is remarkable considering the first two classes of ingot height from 0 % to 10 % and from 10 % to 20 %. Type A process is most conservative according to dislocation cluster generation. The occurrence of cluster seeds in the first 10 % is less than in the following 10 % to 20 % of ingot height. On the contrary type B produced the most dislocation cluster in bottom of the ingot. Type C is an intermediate one with a nearly balanced generation of cluster seeds within the first both classes. There is also a relation between the occurrence of cluster seeds and the grain structure. A cluster is generated at a grain boundary or a triple point between three grains if the first sign is connected to a grain boundary or a triple point, respectively. Is there no connection the dislocation cluster is defined to be generated within a grain. Fig. 4 (right) shows the result. Only less than 3 % of the cluster seeds appear to be generated within the grains. About 5.6 % have a connection to triple points and the by far highest
D. Oriwol et al. / Energy Procedia 27 (2012) 66 – 69
fraction of dislocation cluster seeds has their origin at grain boundaries with an amount of about 92 %. If putting together the last two fractions it is remarkable that over 97 % of all cluster seeds have a connection to a grain boundary.
3.3. Discussion The flip-books give a phenomenological and macroscopic view on the dislocation structure in multicrystalline silicon. By defect-etching the wafer only a part of the dislocations can be revealed. The fracture of dislocations, which are more or less parallel to wafer surface is invisible. It is also to consider, that the here given view is a frozen picture after a multistage process containing generation, multiplication and recovery of dislocations. Nevertheless it is obvious that dislocation clusters are generated in the lower regions of silicon ingots. They are mostly related to grain boundaries, which seem to be one of the significant dislocation sources. Certainly there can be other sources like impurities or condensations of point defects. The main driving force for dislocation generation are mechanical stresses during the block cast process, which can be distributed very inhomogeneous due to random grain orientation [6]. Thus the main issue is to reduce mechanical stresses during the block cast process. 4. Conclusion Optical photography of defect etched multicrystalline silicon wafers reveal the generation and development of dislocation clusters. There are cluster seeds marked by the first appearance of dislocation etch pits. Statistical observations reveal that the cluster seeds are generated mostly in the lower 20 % of a silicon ingot. It is to mention, that this observation was made phenomenological with a macroscopically kind of view. Three types of growth processes have been investigated with different behavior of the cluster seed generation. Thus it is possible to reduce dislocation cluster generation and so to increase the solar cell efficiency by choosing suitable growth conditions. Acknowledgements Special thanks to the SolarWorld analytical team: Juliane Walter, Michael Wolf and Alexander Fülle. References [1] Würzner S, Kaden T, Kreßner-Kiel D, Funke C, Möller H.J, Proceedings of the 24th European PV Solar Energy Conference and Exhibition, Hamburg, Germany, 2009, pp. 2133 – 2137. [2] Ryningen B, Stokkan G, Modanese C, Lohne O, Proceedings of the 23th European Photovoltaic Solar Energy Conference and Exhibition, Valencia, Spain, 2008, pp. 1253-1256. [3] Arafune K., Sasaki T, Wakabayashi F, Terada Y, Ohshita Y, Yamaguchi M, Physica B (276-377), 236-239 (2006). [4] I. Tarasov, S. Ostapenko, V. Feifer, S. McHugo, S. V. Koveshnikov, J. Weber, C. Haessler, E.-U. Reisner, Physica B (273274), 549-552 (1999). [5] Sopori B, Budhraja V, Rupnowski P, Johnston S, Call N, Moutinho H, Al-Jassim M, Proceedings of the 34th IEEE Photovoltaic Specialist Conference, Golden, USA, 2009, pp. 1969-1974 [6] Behnken H, Proceedings 24th European Photovoltaic Solar Energy Conference and Exhibition, Hamburg, Germany, 2009, pp. 1281 - 1285.
69
Energy Procedia 27 (2012) 66 – 69
SiliconPV: April 03-05, 2012, Leuven, Belgium
Control of dislocation cluster formation and development in silicon block casting D. Oriwola*, M. Hollatza, M. Reineckea a
SolarWorld Innovations GmbH, Berthelsdorfer Str. 111a, Freiberg D-09599, Germany
Abstract Dislocation clusters in multicrystalline silicon are known to be detrimental for photovoltaic cell efficiency. They reduce minority carrier lifetime and are able to build shunts. To get control of them is one of the main purposes of recent developments in crystal growth. For reducing dislocation generation and multiplication there are several issues to focus on during silicon block casting like crystallization condition, grain structure evolution, history of temperature field and phase boundary shape. Different industrial standard ingots were investigated by optical photography. It is shown, that dislocation cluster reduction can be achieved by using suitable conditions during ingot growth.
© Selection andand peer-review under responsibility of theofscientifi c committee of © 2012 2012Published Publishedby byElsevier ElsevierLtd. Ltd. Selection peer-review under responsibility the scientific the SiliconPV conference. committee of 2012 the SiliconPV 2012 conference Keywords: dislocation cluster; multicrystalline silicon; block casting, photovoltaic
1. Introduction Multicrystalline silicon for photovoltaic application has a market share up to 40%. Hence there are serious attempts to improve crystal quality and reduce crystal defects like impurities and dislocations. Dislocations are one dimensional defects in the crystal structure which are known to be heterogeneous distributed over wafer surface [1-2]. They build up so called dislocation clusters and can reach high dislocation densities up to 107cm-2. These areas crucially decrease minority carrier lifetime and so the cell efficiency [3-4], also they are able to introduce shunts [5]. The aim of the industrial crystal growth development is to optimize the block cast process and to improve crystal quality.
1876-6102 © 2012 Published by Elsevier Ltd. Selection and peer-review under responsibility of the scientific committee of the SiliconPV 2012 conference. doi:10.1016/j.egypro.2012.07.030
D. Oriwol et al. / Energy Procedia 27 (2012) 66 – 69
2. Experimental Multicrystalline silicon ingots were produced by industrial block cast technique, using three different growth processes with changes in crystallization conditions. The ingots were cut into bricks and they were sliced to wafer. About 40 bricks from different ingots were used. For every brick a selection of wafers distributed over the ingot height were chosen for further analysis. They were defect etched and then photographed under special light conditions. This gives a flip-book over ingot height where both dislocation etch-pits and grain structure are visible. Additional the ingot height for every wafer is known.
3. Results & discussion 3.1. Optical photography Fig. 1 shows a typical wafer with its grain structure and etch-pits distribution (dark contrast). Like already mentioned in the reference [1-2] the dislocation distribution is very inhomogeneous over wafer surface. The dislocations build up areas of high concentration, which are referred to be dislocation clusters.
Fig. 1. Optical photography of a defect etched wafer revealing grain structure and etch-pits distribution (image length: 75 mm)
3.2. Dislocation cluster seeds The development of a dislocation cluster is demonstrated in Fig 2. The pictures show the generation and further development of a dislocation cluster at a grain boundary from bottom to top. The site of the first arise of etch pits and dark contrast, respectively, is said to be a dislocation cluster seed. To each cluster seed an ingot height and relation to the grain structure can be allocated (sketched in Fig. 4 left and center).
Fig. 2. Three pictures from bottom (left) to top (right) of a small area within a silicon brick showing the generation (center) and development of a dislocation cluster. The height difference between the pictures is about 3 mm (image length: 15 mm)
67
68
D. Oriwol et al. / Energy Procedia 27 (2012) 66 – 69
25
15 10 5
Re
lat
ive
In
0-10-20 10 -30 20 -40 30 -50 go 40 -60 50 -70 th 60 -80 eig 70 -90 ht 80 00 [% 90-1 ]
All
0
Occurence
20
of cluster seeds [arb. units ]
35 30
C B A
Fig. 3. Occurrence of dislocation cluster seeds over ingot height. Blue: all investigated bricks. Green, orange and magenta: groups of growth processes showing different behaviors in cluster seed generation
Fig. 4. Left: sketch of a dislocation cluster in a side view on a brick. The red circle marks the cluster seed. Center: sketch of a top view on a wafer with three grains. The coloured lines are examples for cluster seeds related to grain boundary (yellow), triple point (green) or any site within a grain (blue). Right: Diagramm showing the relative frequency where cluster seeds occur regarding to grain strucutre
Fig. 3 contains a diagram displaying the occurrence of cluster seeds over ingot height. Over all 40 bricks (blue bars) there is a clear evidence that cluster seeds mainly occur in the low region of ingot height. Over 70 % of cluster seeds are generated in the lower 20 % of ingot height. The amounts of new cluster seeds strongly decrease in higher regions. If compared all three process types (referred to be A, B and C) a change in the behavior is remarkable considering the first two classes of ingot height from 0 % to 10 % and from 10 % to 20 %. Type A process is most conservative according to dislocation cluster generation. The occurrence of cluster seeds in the first 10 % is less than in the following 10 % to 20 % of ingot height. On the contrary type B produced the most dislocation cluster in bottom of the ingot. Type C is an intermediate one with a nearly balanced generation of cluster seeds within the first both classes. There is also a relation between the occurrence of cluster seeds and the grain structure. A cluster is generated at a grain boundary or a triple point between three grains if the first sign is connected to a grain boundary or a triple point, respectively. Is there no connection the dislocation cluster is defined to be generated within a grain. Fig. 4 (right) shows the result. Only less than 3 % of the cluster seeds appear to be generated within the grains. About 5.6 % have a connection to triple points and the by far highest
D. Oriwol et al. / Energy Procedia 27 (2012) 66 – 69
fraction of dislocation cluster seeds has their origin at grain boundaries with an amount of about 92 %. If putting together the last two fractions it is remarkable that over 97 % of all cluster seeds have a connection to a grain boundary.
3.3. Discussion The flip-books give a phenomenological and macroscopic view on the dislocation structure in multicrystalline silicon. By defect-etching the wafer only a part of the dislocations can be revealed. The fracture of dislocations, which are more or less parallel to wafer surface is invisible. It is also to consider, that the here given view is a frozen picture after a multistage process containing generation, multiplication and recovery of dislocations. Nevertheless it is obvious that dislocation clusters are generated in the lower regions of silicon ingots. They are mostly related to grain boundaries, which seem to be one of the significant dislocation sources. Certainly there can be other sources like impurities or condensations of point defects. The main driving force for dislocation generation are mechanical stresses during the block cast process, which can be distributed very inhomogeneous due to random grain orientation [6]. Thus the main issue is to reduce mechanical stresses during the block cast process. 4. Conclusion Optical photography of defect etched multicrystalline silicon wafers reveal the generation and development of dislocation clusters. There are cluster seeds marked by the first appearance of dislocation etch pits. Statistical observations reveal that the cluster seeds are generated mostly in the lower 20 % of a silicon ingot. It is to mention, that this observation was made phenomenological with a macroscopically kind of view. Three types of growth processes have been investigated with different behavior of the cluster seed generation. Thus it is possible to reduce dislocation cluster generation and so to increase the solar cell efficiency by choosing suitable growth conditions. Acknowledgements Special thanks to the SolarWorld analytical team: Juliane Walter, Michael Wolf and Alexander Fülle. References [1] Würzner S, Kaden T, Kreßner-Kiel D, Funke C, Möller H.J, Proceedings of the 24th European PV Solar Energy Conference and Exhibition, Hamburg, Germany, 2009, pp. 2133 – 2137. [2] Ryningen B, Stokkan G, Modanese C, Lohne O, Proceedings of the 23th European Photovoltaic Solar Energy Conference and Exhibition, Valencia, Spain, 2008, pp. 1253-1256. [3] Arafune K., Sasaki T, Wakabayashi F, Terada Y, Ohshita Y, Yamaguchi M, Physica B (276-377), 236-239 (2006). [4] I. Tarasov, S. Ostapenko, V. Feifer, S. McHugo, S. V. Koveshnikov, J. Weber, C. Haessler, E.-U. Reisner, Physica B (273274), 549-552 (1999). [5] Sopori B, Budhraja V, Rupnowski P, Johnston S, Call N, Moutinho H, Al-Jassim M, Proceedings of the 34th IEEE Photovoltaic Specialist Conference, Golden, USA, 2009, pp. 1969-1974 [6] Behnken H, Proceedings 24th European Photovoltaic Solar Energy Conference and Exhibition, Hamburg, Germany, 2009, pp. 1281 - 1285.
69