- Email: [email protected]
Towards 20.5% efficiency PERC Cells by improved understanding through simulation
Available online at www.sciencedirect.com
Energy EnergyProcedia Procedia86(2011) (2011)78–81 1–5
SiliconPV: 17-20 April 2011, Freiburg, Germany
Towards 20.5% efficiency PERC Cells by improved understanding through simulation K. Van Wichelen a, L. Tous, A. Tiefenauer, C. Allebé, T. Janssens, P. Choulat, JL. Hernàndez b, E. Cornagliotti, M. Debucquoy, A. Ruocco, J. John, P. Verlinden c, F. Dross, K. Baert IMEC, Kapeldreef 75, 3001 Heverlee, Belgium b now at Kaneka c Amrock Pty Ltd, McLaren Vale, Australia
Abstract
This paper investigates by means of finite-elements and SPICE simulation the further improvement of solar cell performance of PERC-type cells by introduction of industry-compatible solutions, such as a 120 sq. phosphorous emitter with a lower surface concentration than that of current 60 sq. standard emitters and the replacement of the silver screen-printed front-side metallization by narrow, dense electro-plated copper lines on lowly resistive Ti or NiSi contacts. © under responsibility of of SiliconPV 2011. © 2011 2011 Published Publishedby byElsevier ElsevierLtd. Ltd.Selection Selectionand/or and/orpeer-review peer-review under responsibility SiliconPV 2011. Keywords: Solar Cell; Simulation; PERC; advanced emitter; Cu metallization
1. Description Mono-crystalline Cz full aluminium metal back surface field (full Al-BSF) solar cells with screen-printed silver contacts (Ag SP) can reach an efficiency of nearly 18%. The base is contacted at the rear side over the whole area with a screen-printed aluminium layer. For more advanced “Passivated Emitter and Rear locally Contacted” type architectures (PERC) with local Al contacts at the rear, efficiencies up to 19.2%
a
Corresponding author. Tel.: +32-16 -281385
E-mail address: [email protected]
1876–6102 © 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of SiliconPV 2011. doi:10.1016/j.egypro.2011.06.105
79
K. Van Wichelen et al. / Energy Procedia 8 (2011) 78–81
have previously been reported [1]. In these PERC cells, the phosphorous emitter originates from a standard POCl3 diffusion process resulting in a sheet resistance of about 60 sq. and a relatively high surface concentration of about 5x1020 cm-3. The rear local contact holes are defined in a SiO2/SiNx dielectric passivation stack that effectively reduces the surface recombination velocity as compared to full aluminium metal back surface field cells [2]. The front-side of the PERC cells is identical to full-Al BSF solar cells and consists of Ag screen-printed paste fired through a SiN anti-reflection coating (ARC). This paper investigates by means of finite-elements and SPICE simulation further improvement of solar cell performance by: (i) introducing a 120 sq. phosphorous emitter with lower surface concentration in order to reduce Auger recombination, parasitic absorption and the surface recombination velocity at the front-side (all three participating to the “dead layer” effect [3]) and (ii) replacement of the relatively wide screen-printed front-side finger metallization (80-125 m) by narrower, denser and better conducting electro-plated copper lines (40 m wide) on lowly resistive Ti or NiSi contacts to reduce shadowing and resistive losses.
2. Results 2.1. Details of the reference cell The reference PERC cells (size 12.5x12.5 cm2, semi-square) use Cz p-type substrates of 1 – 3 cm, corresponding to a doping level in the range of 5x1015 – 1.5x1016 cm-3. The bulk lifetime is around 500 s. The silicon thickness is 160 m. Random pyramid texturing and 80nm SiN ARC layers are present at the front-side. The front-side metallization consists of 125 m wide Ag screen-printed lines with an maximum height of 30 m, average height of 20 m and a pitch of 2.2 mm. The rear is passivated with a SiO2/ SiNx stack with local contact holes of 35 m in diameter and 600 m equidistantly spaced. A local Al BSF is formed in the openings by firing the Al layer. The performance of this reference cell was measured and reported [1] and is summarized in table 1 below:
TABLE 1. Reference PERC cell performance Cell type Measured (mean ± stdev. of set of 10 samples) Measured (best cell of set) Simulated
Voc (mV)
Jsc (mA/cm2)
FF (%)
(%)
634 ±3 637 637
37.9 ±0.4 38.2 37.9
78.0±1 79.1 79.5
18.8±0.4 19.2 19.19
For the optical simulation, a raytracer tool in Sentaurus TM software (SYNOPSYS) [4] is used to calculate a 1D charge generation profile for a 1 sun illumination condition, taking into account the texture and ARC layer at the cell’s front side. A mixed-mode simulation platform in the SentaurusTM software is used for the electrical simulations in this work that combines finite-element simulations of 2D micro-cells with SPICE network simulations to take into account the resistive losses of fingers, rear side local contacts and shunt resistor [5][6]. This simulation platform has been built and configured based on the specifications and characteristics of this reference PERC cell, following a bottom-up approach. The assumptions and estimated values which the simulations are based upon, are explained in table 2. The 3D current-crowding
80
K. Van Wichelen et al. / Energy Procedia 8 (2011) 78–81
effect at the localized (point) rear contacts is not present in the 2D finite-element simulation, where pointcontacts are replaced by line-contacts. The size of the rear contact openings in the 2D micro-cell has been scaled in order to match the contacted/passivated area ratio of the real cell. The minority carrier surface recombination velocity (SRV) parameter at the front-side is surface doping concentration dependent, according to Kerr [7][8], extrapolated to actual concentration, measured by SIMS (after correction for measured sheetresistance). The optical shadow effect of fingers and bus bars has been accounted for by normalization of the electrical output current to the total area of the cell, including the bus bar area. Floating junction recombination or non-ideal recombination effects (cell edge recombination, local series resistance problem originating from high contact resistance for firing Ag and local non-ideal diode recombination originating from Ag firing contamination, scratches, etc.. giving a non-ideality factor n2> 2, were not yet taken into consideration and would affect the fill factor (FF) output [9][10][11]. The simulation results of the cell are mentioned in table 1.
TABLE 2. Simulation models and parameter values for reference PERC cell max. lifetime (Scharfetter) base doping ( ~2 cm) front passivation SRV Ag paste resistivity,
value
Units
500
s cm-3
8x1015
surface conc. dependent[6] 5x10-6 cm
Ag/ Si contact resistivity
5x10-3
rear passivation SRV contact SRV shunt resistance (from dark-iv)
1x102 2x106 1x104
finger length finger pitch
12.1 2.2
cm2 cm s-1 cm s-1 cm2 cm mm
phosphorous concentration [at./cm3]
Parameter
10
21
10
20
10
19
10
18
10
17
10
16
120 Ohm/sq adv. emitter 60 Ohm/sq std. emitter (SIMS) 60 Ohm/sq std. emitter (active)
depth [ m]
0
0.5
1
FIG. 1. Experimental Phos. emitter profiles of 60 /sq ref. and 120 /sq advanced emitter
2.2. Predicted Performance improvements Using the reference cell described above as a base-line, several modifications are being investigated in order to bring up the performance. Firstly, the phosphorous emitter profile with a sheet resistance of 60 sq. is replaced by a 120 sq. emitter. The surface concentration is significantly lower in this case, resulting in better surface recombination velocity at the front-side, reduced Auger recombination losses and less bandgap narrowing in the emitter. The benefit is demonstrated by an increase of 12mV to 17mV in Voc , depending in the design of the front metallization, as shown in table 3. The SIMS profiles of both emitters are shown in figure 1. Secondly, the screen-printed silver lines are replaced by lower resistivity, narrower and larger aspect ratio Cu lines on top of Ti or NiSi which reduce the metal line resistance as well as the contact resistance, simultaneously reducing the shadow effect. The short-circuit current is increased by 2% to 4% depending on the line width. In order to compensate for the reduction in fill factor, a smaller pitch of the metal lines is, despite a moderate drop in Jsc, beneficial for the performance.
81
K. Van Wichelen et al. / Energy Procedia 8 (2011) 78–81
TABLE 3. Simulated PERC cell performance improvements Phos. emitter type
Front side metallization (finger width)
finger pitch (mm)
Voc (mV)
Jsc (mA/cm2)
FF (%)
(%)
125 m Ag SP (std.)
2.2
637
37.9
79.5
19.19
60 /sq emitter
80 m Cu-plated
2.2
639
38.7
80.5
19.93
60 /sq emitter
80 m Cu-plated
2.0
639
38.6
80.8
19.93
120 /sq emitter (adv.)
80 m Cu-plated
2.2
651
38.9
79.3
20.05
120 /sq emitter
40 m Cu-plated
2.2
654
39.6
78.6
20.35
120 /sq emitter
40 m Cu-plated
1.5
654
39.3
80.3
20.59
60 /sq emitter (std.)
3. Conclusions From a simulation point of view, the optimization of shadow losses by using narrow, high aspect ratio, good conducting Cu lines combined with a reduced emitter saturation current Jo from a 120 sq. emitter can result in power efficiency above 20.5%.
Acknowledgements The authors acknowledge A. Ristow from Photovoltech NV and G. Letay from Synopsys Inc. for helpful and constructive discussions.
References [1] Prajapati, V.; John, J.; Yang, X.; Mischke, W. and Hong, J. Silane free high-efficiency industrial silicon solar cells using dielectric passivation and local BSF. In: 25th European Photovoltaic Solar Energy Conference and Exhibition - European Photovoltaic Solar Energy Conference EUPVSEC 2010. pp.1125-1129; (6-10 September 2010; Valencia, Spain.) [2] Allebe, C.; Schlenker, E.; Hernandez, J.; Aleman, M.; Prajapati, V.; Van Nieuwenhuysen, K.; Janssens, T.; Lorenz, A.; Vermang, B.; Uruena De Castro, A.; Rosseel, E.; John, J.; Baert, K. and Poortmans, J. Process integration towards i-PERL structure. In: 25th European Photovoltaic Solar Energy Conference and Exhibition - EPVSEC. 2010. pp.1469-1474; (6-10 September 2010; Valencia, Spain.) [3] P. Choulat and F. Duerinckx, “Minimizing the dead layer of industrial phosphorus emitter formed by POCI3 diffusion,” in Proceedings of the 15th International Photovoltaic Science & Engineering Conference PVSEC, 2005, Shanghai, China. [4] Sentaurus-Device. Synopsys Inc., Mountain View, CA. URL: www.synopsys.com/products/tcad/tcad.html. [5] A. Zekry, A. Y. Al-Mazroo. A distributed SPICE-Model of a Solar Cell. IEEE TED Vol. 43, N.5 May 1996, p.691 [6] P.P. Altermatt, S. Steingrube, Y. Yang, C. Sprodowski, T. Dezhdar, S. Koc, B. Veith, S. Herrman, R. Bock, K. Bothe, J. Schmidt, R. Brendel, Highly predictive modelling of entire Si Solar cells for industrial applications. European Photovoltaic Solar Energy Conference 2009, Hamburg [7] M. J. Kerr. Surface, emitter and bulk recombination in Silicon and Development of Silicon Nitride passivated solar cells. PhD thesis, 2002, Australian National University [8] P.P. Altermatt, J.O. Schumacher, A. Cuevas, M. J. Kerr, S.W. Glunz, R.R. King, and G. Heiser. Numerical modeling of highly doped Si:P emitters based on Fermi–Dirac statistics and selfconsistent material parameters. J. Appl. Phys. 2002;92:3187. [9] O. Breitenstein, P.P. Altermatt, K Ramspeck, M. A. Green, Jianhua Zhao, A. Schenk. Interpretation of the commonly observed IV of c-Si having ideality factor larger than two. 4th World Conference on Photovoltaic Energy Conversion 2006, Hawaii [10] O. Breitenstein, P. Altermatt, K. Ramspeck, and A. Schenk. The origin of ideality factors n > 2 of shunts and surfaces in the dark I– V curves of Si solar cells. 21st European Photovoltaic Solar Energy Conf, Dresden, Sept. 4–8, 2006, pp. 625–628. [11] A. Schenk, U. Krumbein. Coupled defect-level recombination: Theory and application to anomalous diode characteristics. J. Appl. Phys. 78 (1995), pp.3185-3192.
Energy EnergyProcedia Procedia86(2011) (2011)78–81 1–5
SiliconPV: 17-20 April 2011, Freiburg, Germany
Towards 20.5% efficiency PERC Cells by improved understanding through simulation K. Van Wichelen a, L. Tous, A. Tiefenauer, C. Allebé, T. Janssens, P. Choulat, JL. Hernàndez b, E. Cornagliotti, M. Debucquoy, A. Ruocco, J. John, P. Verlinden c, F. Dross, K. Baert IMEC, Kapeldreef 75, 3001 Heverlee, Belgium b now at Kaneka c Amrock Pty Ltd, McLaren Vale, Australia
Abstract
This paper investigates by means of finite-elements and SPICE simulation the further improvement of solar cell performance of PERC-type cells by introduction of industry-compatible solutions, such as a 120 sq. phosphorous emitter with a lower surface concentration than that of current 60 sq. standard emitters and the replacement of the silver screen-printed front-side metallization by narrow, dense electro-plated copper lines on lowly resistive Ti or NiSi contacts. © under responsibility of of SiliconPV 2011. © 2011 2011 Published Publishedby byElsevier ElsevierLtd. Ltd.Selection Selectionand/or and/orpeer-review peer-review under responsibility SiliconPV 2011. Keywords: Solar Cell; Simulation; PERC; advanced emitter; Cu metallization
1. Description Mono-crystalline Cz full aluminium metal back surface field (full Al-BSF) solar cells with screen-printed silver contacts (Ag SP) can reach an efficiency of nearly 18%. The base is contacted at the rear side over the whole area with a screen-printed aluminium layer. For more advanced “Passivated Emitter and Rear locally Contacted” type architectures (PERC) with local Al contacts at the rear, efficiencies up to 19.2%
a
Corresponding author. Tel.: +32-16 -281385
E-mail address: [email protected]
1876–6102 © 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of SiliconPV 2011. doi:10.1016/j.egypro.2011.06.105
79
K. Van Wichelen et al. / Energy Procedia 8 (2011) 78–81
have previously been reported [1]. In these PERC cells, the phosphorous emitter originates from a standard POCl3 diffusion process resulting in a sheet resistance of about 60 sq. and a relatively high surface concentration of about 5x1020 cm-3. The rear local contact holes are defined in a SiO2/SiNx dielectric passivation stack that effectively reduces the surface recombination velocity as compared to full aluminium metal back surface field cells [2]. The front-side of the PERC cells is identical to full-Al BSF solar cells and consists of Ag screen-printed paste fired through a SiN anti-reflection coating (ARC). This paper investigates by means of finite-elements and SPICE simulation further improvement of solar cell performance by: (i) introducing a 120 sq. phosphorous emitter with lower surface concentration in order to reduce Auger recombination, parasitic absorption and the surface recombination velocity at the front-side (all three participating to the “dead layer” effect [3]) and (ii) replacement of the relatively wide screen-printed front-side finger metallization (80-125 m) by narrower, denser and better conducting electro-plated copper lines (40 m wide) on lowly resistive Ti or NiSi contacts to reduce shadowing and resistive losses.
2. Results 2.1. Details of the reference cell The reference PERC cells (size 12.5x12.5 cm2, semi-square) use Cz p-type substrates of 1 – 3 cm, corresponding to a doping level in the range of 5x1015 – 1.5x1016 cm-3. The bulk lifetime is around 500 s. The silicon thickness is 160 m. Random pyramid texturing and 80nm SiN ARC layers are present at the front-side. The front-side metallization consists of 125 m wide Ag screen-printed lines with an maximum height of 30 m, average height of 20 m and a pitch of 2.2 mm. The rear is passivated with a SiO2/ SiNx stack with local contact holes of 35 m in diameter and 600 m equidistantly spaced. A local Al BSF is formed in the openings by firing the Al layer. The performance of this reference cell was measured and reported [1] and is summarized in table 1 below:
TABLE 1. Reference PERC cell performance Cell type Measured (mean ± stdev. of set of 10 samples) Measured (best cell of set) Simulated
Voc (mV)
Jsc (mA/cm2)
FF (%)
(%)
634 ±3 637 637
37.9 ±0.4 38.2 37.9
78.0±1 79.1 79.5
18.8±0.4 19.2 19.19
For the optical simulation, a raytracer tool in Sentaurus TM software (SYNOPSYS) [4] is used to calculate a 1D charge generation profile for a 1 sun illumination condition, taking into account the texture and ARC layer at the cell’s front side. A mixed-mode simulation platform in the SentaurusTM software is used for the electrical simulations in this work that combines finite-element simulations of 2D micro-cells with SPICE network simulations to take into account the resistive losses of fingers, rear side local contacts and shunt resistor [5][6]. This simulation platform has been built and configured based on the specifications and characteristics of this reference PERC cell, following a bottom-up approach. The assumptions and estimated values which the simulations are based upon, are explained in table 2. The 3D current-crowding
80
K. Van Wichelen et al. / Energy Procedia 8 (2011) 78–81
effect at the localized (point) rear contacts is not present in the 2D finite-element simulation, where pointcontacts are replaced by line-contacts. The size of the rear contact openings in the 2D micro-cell has been scaled in order to match the contacted/passivated area ratio of the real cell. The minority carrier surface recombination velocity (SRV) parameter at the front-side is surface doping concentration dependent, according to Kerr [7][8], extrapolated to actual concentration, measured by SIMS (after correction for measured sheetresistance). The optical shadow effect of fingers and bus bars has been accounted for by normalization of the electrical output current to the total area of the cell, including the bus bar area. Floating junction recombination or non-ideal recombination effects (cell edge recombination, local series resistance problem originating from high contact resistance for firing Ag and local non-ideal diode recombination originating from Ag firing contamination, scratches, etc.. giving a non-ideality factor n2> 2, were not yet taken into consideration and would affect the fill factor (FF) output [9][10][11]. The simulation results of the cell are mentioned in table 1.
TABLE 2. Simulation models and parameter values for reference PERC cell max. lifetime (Scharfetter) base doping ( ~2 cm) front passivation SRV Ag paste resistivity,
value
Units
500
s cm-3
8x1015
surface conc. dependent[6] 5x10-6 cm
Ag/ Si contact resistivity
5x10-3
rear passivation SRV contact SRV shunt resistance (from dark-iv)
1x102 2x106 1x104
finger length finger pitch
12.1 2.2
cm2 cm s-1 cm s-1 cm2 cm mm
phosphorous concentration [at./cm3]
Parameter
10
21
10
20
10
19
10
18
10
17
10
16
120 Ohm/sq adv. emitter 60 Ohm/sq std. emitter (SIMS) 60 Ohm/sq std. emitter (active)
depth [ m]
0
0.5
1
FIG. 1. Experimental Phos. emitter profiles of 60 /sq ref. and 120 /sq advanced emitter
2.2. Predicted Performance improvements Using the reference cell described above as a base-line, several modifications are being investigated in order to bring up the performance. Firstly, the phosphorous emitter profile with a sheet resistance of 60 sq. is replaced by a 120 sq. emitter. The surface concentration is significantly lower in this case, resulting in better surface recombination velocity at the front-side, reduced Auger recombination losses and less bandgap narrowing in the emitter. The benefit is demonstrated by an increase of 12mV to 17mV in Voc , depending in the design of the front metallization, as shown in table 3. The SIMS profiles of both emitters are shown in figure 1. Secondly, the screen-printed silver lines are replaced by lower resistivity, narrower and larger aspect ratio Cu lines on top of Ti or NiSi which reduce the metal line resistance as well as the contact resistance, simultaneously reducing the shadow effect. The short-circuit current is increased by 2% to 4% depending on the line width. In order to compensate for the reduction in fill factor, a smaller pitch of the metal lines is, despite a moderate drop in Jsc, beneficial for the performance.
81
K. Van Wichelen et al. / Energy Procedia 8 (2011) 78–81
TABLE 3. Simulated PERC cell performance improvements Phos. emitter type
Front side metallization (finger width)
finger pitch (mm)
Voc (mV)
Jsc (mA/cm2)
FF (%)
(%)
125 m Ag SP (std.)
2.2
637
37.9
79.5
19.19
60 /sq emitter
80 m Cu-plated
2.2
639
38.7
80.5
19.93
60 /sq emitter
80 m Cu-plated
2.0
639
38.6
80.8
19.93
120 /sq emitter (adv.)
80 m Cu-plated
2.2
651
38.9
79.3
20.05
120 /sq emitter
40 m Cu-plated
2.2
654
39.6
78.6
20.35
120 /sq emitter
40 m Cu-plated
1.5
654
39.3
80.3
20.59
60 /sq emitter (std.)
3. Conclusions From a simulation point of view, the optimization of shadow losses by using narrow, high aspect ratio, good conducting Cu lines combined with a reduced emitter saturation current Jo from a 120 sq. emitter can result in power efficiency above 20.5%.
Acknowledgements The authors acknowledge A. Ristow from Photovoltech NV and G. Letay from Synopsys Inc. for helpful and constructive discussions.
References [1] Prajapati, V.; John, J.; Yang, X.; Mischke, W. and Hong, J. Silane free high-efficiency industrial silicon solar cells using dielectric passivation and local BSF. In: 25th European Photovoltaic Solar Energy Conference and Exhibition - European Photovoltaic Solar Energy Conference EUPVSEC 2010. pp.1125-1129; (6-10 September 2010; Valencia, Spain.) [2] Allebe, C.; Schlenker, E.; Hernandez, J.; Aleman, M.; Prajapati, V.; Van Nieuwenhuysen, K.; Janssens, T.; Lorenz, A.; Vermang, B.; Uruena De Castro, A.; Rosseel, E.; John, J.; Baert, K. and Poortmans, J. Process integration towards i-PERL structure. In: 25th European Photovoltaic Solar Energy Conference and Exhibition - EPVSEC. 2010. pp.1469-1474; (6-10 September 2010; Valencia, Spain.) [3] P. Choulat and F. Duerinckx, “Minimizing the dead layer of industrial phosphorus emitter formed by POCI3 diffusion,” in Proceedings of the 15th International Photovoltaic Science & Engineering Conference PVSEC, 2005, Shanghai, China. [4] Sentaurus-Device. Synopsys Inc., Mountain View, CA. URL: www.synopsys.com/products/tcad/tcad.html. [5] A. Zekry, A. Y. Al-Mazroo. A distributed SPICE-Model of a Solar Cell. IEEE TED Vol. 43, N.5 May 1996, p.691 [6] P.P. Altermatt, S. Steingrube, Y. Yang, C. Sprodowski, T. Dezhdar, S. Koc, B. Veith, S. Herrman, R. Bock, K. Bothe, J. Schmidt, R. Brendel, Highly predictive modelling of entire Si Solar cells for industrial applications. European Photovoltaic Solar Energy Conference 2009, Hamburg [7] M. J. Kerr. Surface, emitter and bulk recombination in Silicon and Development of Silicon Nitride passivated solar cells. PhD thesis, 2002, Australian National University [8] P.P. Altermatt, J.O. Schumacher, A. Cuevas, M. J. Kerr, S.W. Glunz, R.R. King, and G. Heiser. Numerical modeling of highly doped Si:P emitters based on Fermi–Dirac statistics and selfconsistent material parameters. J. Appl. Phys. 2002;92:3187. [9] O. Breitenstein, P.P. Altermatt, K Ramspeck, M. A. Green, Jianhua Zhao, A. Schenk. Interpretation of the commonly observed IV of c-Si having ideality factor larger than two. 4th World Conference on Photovoltaic Energy Conversion 2006, Hawaii [10] O. Breitenstein, P. Altermatt, K. Ramspeck, and A. Schenk. The origin of ideality factors n > 2 of shunts and surfaces in the dark I– V curves of Si solar cells. 21st European Photovoltaic Solar Energy Conf, Dresden, Sept. 4–8, 2006, pp. 625–628. [11] A. Schenk, U. Krumbein. Coupled defect-level recombination: Theory and application to anomalous diode characteristics. J. Appl. Phys. 78 (1995), pp.3185-3192.