- Email: [email protected]
Status of Industrial Back Junction n-type Si Solar Cell Development
Available online at www.sciencedirect.com
ScienceDirect Energy Procedia 92 (2016) 678 – 683
6th International Conference on Silicon Photovoltaics, SiliconPV 2016
Status of industrial back junction n-type Si solar cell development Stefan Bordihna, Verena Mertensa, Janko Cieslaka, Gregor Zimmermanna, Ronny Lantzscha, Jessica Scharfa, Steffen Geißlera, Stefan Hörnleina, Markus Neubera, Steffen Laubea, Alexandra Dietricha, Barbara Szramowskia, Sascha Esefeldera, Tabitha Ballmanna, Björn Eisenhawera, Andreas Mohra, Martin Schapera, Kai Pettera, Jörg W. Müllera, Daniel J. W. Jeonga, Ruiying Haob, T. S. Ravib a
Hanwha Q Cells GmbH, Sonnenallee 17-21, 06766 Bitterfeld-Wolfen, Germany b Crystal Solar Inc., 3050 Coronado Drive, Santa Clara, CA 95054, USA
Abstract Here we present the latest results of our double side contacted, all screen printed n-type mono silicon solar cell development. Ntype Czochralski (n-Cz) silicon solar cell results are compared to those of standard p-type Cz (p-Cz) silicon and n-type epitaxial (np+ epi) wafers, produced by Crystal Solar, having an integrated epitaxial boron doped p+-silicon layer. The np+ epi and p-Cz wafers are processed applying the Hanwha Q CELLS Q.ANTUM technology process flow to make PERC cells in our production line including process adaptations to mono wafers while the n-Cz wafers are processed with an extended Q.ANTUM sequence including additional processing steps like cleaning steps and a BBr3 tube furnace diffusion to create the rear side boron p+-silicon layer. We achieve conversion efficiencies up to 21.8 % for the n-type Cz silicon back junction solar cell with open circuit voltage values of 671 mV. The p-type Cz silicon solar cell shows non-stabilized efficiencies up to 21.2 %. The n-type epitaxial solar cells have efficiencies up to 21.7 % with fill factor values of up to 82.1% due to the high rear side conductivity of the integrated epitaxial boron doped p+-silicon layer. The latter solar cell results demonstrate a new path to industrial solar cells with efficiencies >22 % by combination of simple and robust solar cell processing and epitaxial wafer growth with built-in doping layers. © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 2016 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peerreview reviewbyby scientific conference committee of SiliconPV 2016 responsibility under responsibility of PSE AG. Peer thethe scientific conference committee of SiliconPV 2016 under of PSE AG. Keywords: n-type Si; back junction solar cell
1876-6102 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer review by the scientific conference committee of SiliconPV 2016 under responsibility of PSE AG. doi:10.1016/j.egypro.2016.07.042
Stefan Bordihn et al. / Energy Procedia 92 (2016) 678 – 683
1. Introduction Over the past years the so-called PERC technology, i.e. solar cells featuring a metal contact and a dielectric passivation on front and rear side, are introduced into mass production of solar cells. Mainly p-type (boron doped) multicrystalline (mc-Si) or Czochralski (p-Cz) are used but either for mc-Si [1, 2] or p-Cz [3] degradation losses of the wafer material is reported. Even if the p-Cz specific BO-related recombination is deactivated the material quality (in terms of minority carrier lifetime) is below the values reported for n-Cz, see [4] and [5, 6]. Hence, an improvement of the efficiency of PERC solar cells is possible by using phosphorous doped Cz Si material (n-Cz). It was shown that the higher segregation coefficient of phosphorous is not detrimental to the performance of the solar cell when keeping the base resistivity above 6 Ohm cm [7]. The highest processing synergy with established p-Cz PERC production sequences is achieved with a back junction concept which is also called n-PERT. This concept became more attractive during the last years and several research groups reported about efficiency results above 21 % [8–10] with record efficiencies of 22.5 % [11]. However, due to the low market share of n-Cz material its price level is still higher than those of p-Cz [12] although the productions costs are expected to be comparable. As alternative epitaxially (epi) grown n-type Si material has been introduced which has low cost potential [13]. By using epi Si material we presented in previous work independently confirmed efficiency values of 21.4 % [8]. In this article we present the latest results of our double side contacted, all screen printed n-Cz Si solar cell development. We compare n-Cz to standard p-Cz and n-type epitaxial (np+ epi) based solar cells and focus at first on the fill factor (FF)-loss analysis to evaluate the impact of the rear p+ layer. As second a short circuit current density (Jsc)-loss analysis is presented because the Jsc-value of back junction solar cells is known to be more sensitive to the front side recombination [14] than p-Cz front junction solar cells. We achieve conversion efficiencies of 21.8 % for the n-Cz solar cell concept with open circuit voltage values of 671 mV. The p-Cz reference shows nonstabilized efficiencies up to 21.2 % and the np+ epi solar cells have efficiencies up to 21.7 %. 2. Experimental In this work we compare n-Cz silicon solar cell results to those of standard p-Cz silicon and n-type epitaxially grown kerfless wafers having an integrated epitaxial boron doped p+-silicon layer. The cross sectional schematic view of the investigated three solar cell structures is shown in Fig. 1. The solar cell processing is based on an industrial p-type PERC process with all screen printed contacts. The ntype epitaxial wafers are produced by Crystal Solar and have a wafer thickness of 170 μm, a resistivity of 6 Ohm cm and an area of 243.4 cm². The np+ epi wafers and p-Cz Si wafers (resistivity 1 Ohm cm, thickness 180 μm, area 243.4 cm²) are processed applying the Hanwha Q CELLS Q.ANTUM technology process flow to make PERC solar cells [15] including adaptions to mono wafers (texturing process). The n-Cz Si wafers (resistivity ~10 Ohm cm, thickness 180 μm, area 243.4 cm²) were processed with an extended Q.ANTUM sequence including additional processing steps like cleaning processes and a BBr3 tube furnace diffusion to create the rear side boron p+-silicon layer. The rear dielectric passivation layer consists of AlOx and SiNx. Front and rear contacts of all three types of solar cells are screen printed using commercially available printing pastes and four bus bar layouts.
Fig 1. Cross sectional schematic view of the three different solar cell types under investigation: (left) n-Cz solar cell with additional rear side boron p+-silicon layer, (center) p-Cz solar cell with dielectric rear side passivation and local contacts and (right) np+ epi solar cell, made from epitaxial n-type Si wafers with built-in boron doped rear side epitaxial p+-Si layer, also with dielectric rear side passivation and local contacts.
679
680
Stefan Bordihn et al. / Energy Procedia 92 (2016) 678 – 683
3. Results and discussion Table 1. J–V data for the champion n-Cz, p-Cz and np+ epitaxial Si solar cells (in-house measurement, class AAA DC solar simulator calibrated to PTB standards). Cell Type
Jsc (mA/cm²)
Voc (mV)
FF (%)
Eta (%)
n-type Cz
40.4
671.0
80.2
21.8
p-type Cz
40.4
662.5
79.2
21.2
np+ epi
40.0
661.0
82.1
21.7
The champion solar cells of the n-Cz, p-Cz and np+ epi solar concept reflect best their technological differences. The J–V results of the champion cells are summarized in Table 1. The Jsc-values of n-Cz and p-Cz are similar whereas the np+ epi solar cell have a lower Jsc-value by 0.4 mA/cm². We achieve conversion efficiencies up to 21.8 % for the n-Cz solar cell with an open circuit voltage of 671 mV and therefore, 10 mV higher compared to pCz. The p-Cz solar cell shows non-stabilized efficiencies up to 21.2 % and the np+ epi solar cells efficiencies up to 21.7 % with fill factor values of up to 82.1 % which is ~1.9 %abs higher than for n-Cz and 2.9 %abs higher than for p-Cz. In Fig. 2 the illuminated J–V data (shifted by Jsc) and the SunsVoc data are shown. The high fill factor value of the np+ epi solar cell can be likely related to a low series resistance loss as the illuminated J–V data (shifted by Jsc) and the SunsVoc data overlaps in the voltage range of 550 mV – 650 mV.
Fig 2. J–V and SunsVoc data for the champion solar cells using (a) n-Cz, (b) p-Cz and (c) np+ epi Si wafer material.
681
Stefan Bordihn et al. / Energy Procedia 92 (2016) 678 – 683
3.1. Open circuit voltage differences As mentioned above the n-Cz solar cell shows an open circuit voltage (Voc) value of 671 mV which is ~10 mV higher than for p-Cz (662 mV) and np+ epi (661 mV). This difference corresponds to an efficiency gain of 0.33 %abs. To evaluate whether local recombination center causes the differences of the Voc, photoluminescence images are taken in Voc mode (excitation wavelength 808 nm). In Fig. 3 it is shown that no local spots with higher recombination can be observed. Hence, the higher Voc-values of the n-Cz solar cells can be related to a higher quality over the entire wafer area.
Fig 3. Photoluminescence images taken in Voc mode (scale 0–60k counts).
3.2. Fill factor analysis The differences in fill factor of the n-Cz, p-Cz and the np+ epi solar cells are summarized in Table 2. The fill factor analysis is based on the method proposed in Ref. [16]. The np+ epi solar cell gives the highest fill factor value of 82.1 % which is 1.9 %abs above the level of the n-Cz solar cell (FF = 80.2 %) and 2.9 %abs above the level of the p-Cz solar cell (FF = 79.2 %). The high fill factor level of the np+ epi solar cell can be attributed to a low series resistance of 0.1 Ohm cm² which corresponds to a fill factor loss of 0.55 %abs. The reason for the low series resistance is the high conductance of the integrated epitaxial boron doped p+-silicon layer. The series resistance of the n-Cz solar cell with a boron diffused rear p+-silicon layer has been 0.58 Ohm cm² which is slightly lower than the value obtained for the p-Cz solar cell (0.66 Ohm cm²). The pseudo fill factor (pFF) level achieved for the n-Cz solar cell is 83.3 % which is about 0.5 %abs above the level of the p-Cz and 0.7 %abs above the level of the np+ epi solar cells. These differences are also reflected by the fill factor loss caused by non-ideal recombination mechanisms. In the case of the np+ epi solar cells the high nonideal recombination indicates that its limits mainly the overall fill factor level of 82.1 %. Hence, FF-values of 82.6 % would be possible when the pFF of the np+ epi solar cells would be similar to those of the n-Cz solar cell which would increase the conversion efficiency by 0.13 %abs. Table 2. Fill factor loss caused by series resistance losses (FF-loss Rs), calculated series resistance (Rs), pseudo fill factor (pFF) and fill factor loss caused by non-ideal recombination (FF-loss J2) using I–V and SunsVoc data according to the method proposed in Ref. [16]. FF-loss Rs (%)
Rs (Ohm cm²)
pFF (%)
FF-loss J2 (%)
n-Cz
Cell Type
3.15
0.58
83.31
0.85
p-Cz
3.57
0.66
82.76
1.25
np+ epi
0.55
0.10
82.60
1.30
682
Stefan Bordihn et al. / Energy Procedia 92 (2016) 678 – 683
3.3. Short circuit current density analysis The short circuit current (Jsc) losses are analyzed using the method proposed in Ref. 17 and 18. The model uses the internal quantum efficiency (IQE) data, which are shown for the investigated solar cells in Fig 4-a. The model separates the short circuit current loss into the contributions: (a) front surface reflection, (b) front surface escape, (c) grid shading and (d) non-perfect IQE, see Fig. 4-b. The calculated loss contributions are listed in Table 3.
Fig 4. (a) Internal quantum efficiency data of the best n-Cz, p-Cz and np+ epi solar cells. (b) Schematic illustration of short circuit current (Jsc) loss contributions according to the method proposed in Ref. 17 and Ref. 18.
Table 3. Contributions of short circuit current losses (mA/cm²) for the best n-Cz, p-Cz and np+ epi solar cells. Cell Type
Front surface escape
Front surface reflectance
Grid shading
Non-perfect IQE
n-Cz
1.5
0.2
1.6
2.7
p-Cz
1.2
0.3
1.6
2.9
np+ epi
1.1
0.2
1.6
3.4
The main Jsc-loss for the n-Cz solar cell can be attributed to non-absorbed light that escapes at the front side after having reflected at the rear (called front surface escape). The Jsc-loss caused by the front surface escape is for n-Cz 0.3 mA/cm² higher than for p-Cz. This is likely related to the different processing of the n-Cz wafers which feature a smoother rear surface created by the cleaning steps prior to the BBr3 diffusion process. The Jsc of the np+ epi solar cell (Jsc = 40.0 mA/cm²) is lower by 0.6 mA/cm² due to a higher front side recombination compared to n-Cz solar cell (Jsc = 40.4 mA/cm²). The reason for this loss is presumably the lower base resistivity of the np+ epi wafer, ~6 Ohm cm, in comparison to the value of ~10 Ohm cm in the case of the n-Cz wafer. However, the loss of the Jsc due to a lower base resistivity is to some extend balanced by a slightly higher fill factor of the solar cells. Hence, the overall efficiency loss can be not directly extracted from the determined Jsc-loss. The impact of the base resistivity on the J–V data has been shown in previous work [7].
Stefan Bordihn et al. / Energy Procedia 92 (2016) 678 – 683
4. Summary In this article, we report on our work on solar cells made from n-type Czochralski silicon in comparison to results of standard p-type Cz silicon and epitaxially grown n-type silicon wafers with built-in boron rear side emitter. Using an extended version of the Q.ANTUM process flow we achieved efficiencies up to 21.8 % (in-house measurement) with standard screen printing metallization on front and rear and four bus bar design. A gain of approximately 0.6 % compared to p-type Cz silicon solar cells processed in parallel with the Q.ANTUM technology was shown. The solar cell made from n-type epitaxially grown silicon gives an efficiency of 21.7 % and a high fill factor value of 82.1 %. The high fill factor is attributed to the high conductivity of the integrated epitaxial boron doped p+-silicon layer. Additionally, the overall fill factor level can be increased further by decreasing the non-ideal recombination losses which have caused a reduction in the pseudo fill factor. The high efficiency obtained with np+ epi wafers using robust cell processing gives the chance to eliminate the need of a complex boron diffusion used in the conventional n-type Cz silicon process flow.
Acknowledgements Special thanks to the staff of the Reiner Lemoine Research Center and pilot line at Hanwha Q CELLS.
References [1] F. Kersten, P. Engelhart, H.-C. Ploigt, A. Stekolnikov, T. Lindner, F. Stenzel, M. Bartzsch, A. Szpeth, K. Petter, J. Heitmann and J. W. Müller, Sol. Energy Mater. Sol. Cells 142 (2015) pp. 83–86. [2] D. N. R. Payne, C. E. Chan, B. J. Hallam, B. Hoex, M. D. Abbott, S. R. Wenham, D. M. Bagnall, RRL Phys. Stat. Solidi 2016 doi: 10.1002/pssr.201510437. [3] K. Bothe, J. Schmidt, J. Appl. Phys. 99, 013701 (2006). [4] D. C. Walter, B. Lim, and J. Schmidt, Prog. Photovolt: Res. Appl., (2016) doi: 10.1002/pip.2731. [5] A. Cuevas, C. Samundsett, M. J. Kerr, D. H. Macdonald, H. Mäckel and P. P. Altermatt, in Proc 3rd WCPEC Osaka, Japan 2003 pp. 963–966. [6] A. Richter, S. W. Glunz, F. Werner, J. Schmidt, A. Cuevas, Phys. Rev. B 86, 165202 (2012). [7] V. Mertens, T. Ballmann, J. Y. Lee, M. Junghänel, F. Stenzel, L. Brandt, A. Schulze, P. Engelhart, J. W. Müller, P. Wawer, and K. H. Küsters, Energy Procedia 27 (2012) pp. 53–58. [8] V. Mertens, S. Bordihn, A. Mohr, K. Petter, J. W. Müller, D. J. W. Jeong, R. Hao, T. S. Ravi, “21.4% Efficient Fully Screen Printed n-Type Solar Cell on Epitaxially Grown Silicon Wafers With Built-In Boron Rear Side Emitter“, in Proc. 31st EUPVSEC, Hamburg, Germany 2015, pp. 1000–1002. [9] J. Cho, H.-N.-R. Shin, J. Lee, Y. Choi, J. Lee, H. Oh, T. Kim, M. Hwang, E.-C. Cho, Energy Procedia 77 (2015) pp. 279–285. [10] B. Lim, T. Brendemühl, T. Dullweber, R. Brendel, IEEE J. Photovolt. 6, 2 (2016) pp. 447–453. [11] http://www2.imec.be/be_en/press/imec-news/imec-22-5-percent-npert-solar-cel.html (last access 04.03.2016) [12] R. Lantzsch, V. Mertens, A. Mohr, S. Bordihn, S. Hörnlein, J. Cieslak, L. Niebergall, P. Kowalzik, T. Ballmann, M. Schaper, K. Petter, J. W. Müller, to be presented at nPV Workshop Chambéry, France 2016. [13] R. Hao, T. S. Ravi, V. Siva, J. Vatus, D. Miller, J. Custodio, K. Moyers, Proc. Mater. Res. Soc. Symp. Proc., 1666, 716 (2014). [14] M. Hermle, F. Granek, O. Schultz, S. W. Glunz, J. Appl. Phys. 103, 054507 (2008). [15] P. Engelhart, D. Manger, B. Klöter, S. Hermann, A. Stekolnikov, S. Peters, H.-C. Ploigt, A. Eifler, C. Klenke, A. Mohr, G. Zimmermann, B. Barkenfelt, K. Suva, J. Wendt, T. Kaden, S. Rupp, M. Fischer, J. W. Müller, P. Wawer, “Q.ANTUM – Q-Cells next generation high-power silicon cell & module concept”, in Proc. 26th EUPVSEC, Hamburg, Germany 2011, pp. 821–826. [16] A. Khannah, T. Müller, R. A. Stangl, B. Hoex, P. K. Basu, A. G. Aberle, IEEE J. Photovolt. 3, 1170 (2013). [17] A. G. Aberle, W. Zhang, B. Hoex, Energy Procedia 8 (2011) pp. 244–249. [18] L. Tous, R. Russell, M. Debucquoy, N. Posthuma, F. Duerinckx, R. Mertens, J. Poortmans, Energy Procedia 38 (2013) pp. 467–473.
683
ScienceDirect Energy Procedia 92 (2016) 678 – 683
6th International Conference on Silicon Photovoltaics, SiliconPV 2016
Status of industrial back junction n-type Si solar cell development Stefan Bordihna, Verena Mertensa, Janko Cieslaka, Gregor Zimmermanna, Ronny Lantzscha, Jessica Scharfa, Steffen Geißlera, Stefan Hörnleina, Markus Neubera, Steffen Laubea, Alexandra Dietricha, Barbara Szramowskia, Sascha Esefeldera, Tabitha Ballmanna, Björn Eisenhawera, Andreas Mohra, Martin Schapera, Kai Pettera, Jörg W. Müllera, Daniel J. W. Jeonga, Ruiying Haob, T. S. Ravib a
Hanwha Q Cells GmbH, Sonnenallee 17-21, 06766 Bitterfeld-Wolfen, Germany b Crystal Solar Inc., 3050 Coronado Drive, Santa Clara, CA 95054, USA
Abstract Here we present the latest results of our double side contacted, all screen printed n-type mono silicon solar cell development. Ntype Czochralski (n-Cz) silicon solar cell results are compared to those of standard p-type Cz (p-Cz) silicon and n-type epitaxial (np+ epi) wafers, produced by Crystal Solar, having an integrated epitaxial boron doped p+-silicon layer. The np+ epi and p-Cz wafers are processed applying the Hanwha Q CELLS Q.ANTUM technology process flow to make PERC cells in our production line including process adaptations to mono wafers while the n-Cz wafers are processed with an extended Q.ANTUM sequence including additional processing steps like cleaning steps and a BBr3 tube furnace diffusion to create the rear side boron p+-silicon layer. We achieve conversion efficiencies up to 21.8 % for the n-type Cz silicon back junction solar cell with open circuit voltage values of 671 mV. The p-type Cz silicon solar cell shows non-stabilized efficiencies up to 21.2 %. The n-type epitaxial solar cells have efficiencies up to 21.7 % with fill factor values of up to 82.1% due to the high rear side conductivity of the integrated epitaxial boron doped p+-silicon layer. The latter solar cell results demonstrate a new path to industrial solar cells with efficiencies >22 % by combination of simple and robust solar cell processing and epitaxial wafer growth with built-in doping layers. © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 2016 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peerreview reviewbyby scientific conference committee of SiliconPV 2016 responsibility under responsibility of PSE AG. Peer thethe scientific conference committee of SiliconPV 2016 under of PSE AG. Keywords: n-type Si; back junction solar cell
1876-6102 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer review by the scientific conference committee of SiliconPV 2016 under responsibility of PSE AG. doi:10.1016/j.egypro.2016.07.042
Stefan Bordihn et al. / Energy Procedia 92 (2016) 678 – 683
1. Introduction Over the past years the so-called PERC technology, i.e. solar cells featuring a metal contact and a dielectric passivation on front and rear side, are introduced into mass production of solar cells. Mainly p-type (boron doped) multicrystalline (mc-Si) or Czochralski (p-Cz) are used but either for mc-Si [1, 2] or p-Cz [3] degradation losses of the wafer material is reported. Even if the p-Cz specific BO-related recombination is deactivated the material quality (in terms of minority carrier lifetime) is below the values reported for n-Cz, see [4] and [5, 6]. Hence, an improvement of the efficiency of PERC solar cells is possible by using phosphorous doped Cz Si material (n-Cz). It was shown that the higher segregation coefficient of phosphorous is not detrimental to the performance of the solar cell when keeping the base resistivity above 6 Ohm cm [7]. The highest processing synergy with established p-Cz PERC production sequences is achieved with a back junction concept which is also called n-PERT. This concept became more attractive during the last years and several research groups reported about efficiency results above 21 % [8–10] with record efficiencies of 22.5 % [11]. However, due to the low market share of n-Cz material its price level is still higher than those of p-Cz [12] although the productions costs are expected to be comparable. As alternative epitaxially (epi) grown n-type Si material has been introduced which has low cost potential [13]. By using epi Si material we presented in previous work independently confirmed efficiency values of 21.4 % [8]. In this article we present the latest results of our double side contacted, all screen printed n-Cz Si solar cell development. We compare n-Cz to standard p-Cz and n-type epitaxial (np+ epi) based solar cells and focus at first on the fill factor (FF)-loss analysis to evaluate the impact of the rear p+ layer. As second a short circuit current density (Jsc)-loss analysis is presented because the Jsc-value of back junction solar cells is known to be more sensitive to the front side recombination [14] than p-Cz front junction solar cells. We achieve conversion efficiencies of 21.8 % for the n-Cz solar cell concept with open circuit voltage values of 671 mV. The p-Cz reference shows nonstabilized efficiencies up to 21.2 % and the np+ epi solar cells have efficiencies up to 21.7 %. 2. Experimental In this work we compare n-Cz silicon solar cell results to those of standard p-Cz silicon and n-type epitaxially grown kerfless wafers having an integrated epitaxial boron doped p+-silicon layer. The cross sectional schematic view of the investigated three solar cell structures is shown in Fig. 1. The solar cell processing is based on an industrial p-type PERC process with all screen printed contacts. The ntype epitaxial wafers are produced by Crystal Solar and have a wafer thickness of 170 μm, a resistivity of 6 Ohm cm and an area of 243.4 cm². The np+ epi wafers and p-Cz Si wafers (resistivity 1 Ohm cm, thickness 180 μm, area 243.4 cm²) are processed applying the Hanwha Q CELLS Q.ANTUM technology process flow to make PERC solar cells [15] including adaptions to mono wafers (texturing process). The n-Cz Si wafers (resistivity ~10 Ohm cm, thickness 180 μm, area 243.4 cm²) were processed with an extended Q.ANTUM sequence including additional processing steps like cleaning processes and a BBr3 tube furnace diffusion to create the rear side boron p+-silicon layer. The rear dielectric passivation layer consists of AlOx and SiNx. Front and rear contacts of all three types of solar cells are screen printed using commercially available printing pastes and four bus bar layouts.
Fig 1. Cross sectional schematic view of the three different solar cell types under investigation: (left) n-Cz solar cell with additional rear side boron p+-silicon layer, (center) p-Cz solar cell with dielectric rear side passivation and local contacts and (right) np+ epi solar cell, made from epitaxial n-type Si wafers with built-in boron doped rear side epitaxial p+-Si layer, also with dielectric rear side passivation and local contacts.
679
680
Stefan Bordihn et al. / Energy Procedia 92 (2016) 678 – 683
3. Results and discussion Table 1. J–V data for the champion n-Cz, p-Cz and np+ epitaxial Si solar cells (in-house measurement, class AAA DC solar simulator calibrated to PTB standards). Cell Type
Jsc (mA/cm²)
Voc (mV)
FF (%)
Eta (%)
n-type Cz
40.4
671.0
80.2
21.8
p-type Cz
40.4
662.5
79.2
21.2
np+ epi
40.0
661.0
82.1
21.7
The champion solar cells of the n-Cz, p-Cz and np+ epi solar concept reflect best their technological differences. The J–V results of the champion cells are summarized in Table 1. The Jsc-values of n-Cz and p-Cz are similar whereas the np+ epi solar cell have a lower Jsc-value by 0.4 mA/cm². We achieve conversion efficiencies up to 21.8 % for the n-Cz solar cell with an open circuit voltage of 671 mV and therefore, 10 mV higher compared to pCz. The p-Cz solar cell shows non-stabilized efficiencies up to 21.2 % and the np+ epi solar cells efficiencies up to 21.7 % with fill factor values of up to 82.1 % which is ~1.9 %abs higher than for n-Cz and 2.9 %abs higher than for p-Cz. In Fig. 2 the illuminated J–V data (shifted by Jsc) and the SunsVoc data are shown. The high fill factor value of the np+ epi solar cell can be likely related to a low series resistance loss as the illuminated J–V data (shifted by Jsc) and the SunsVoc data overlaps in the voltage range of 550 mV – 650 mV.
Fig 2. J–V and SunsVoc data for the champion solar cells using (a) n-Cz, (b) p-Cz and (c) np+ epi Si wafer material.
681
Stefan Bordihn et al. / Energy Procedia 92 (2016) 678 – 683
3.1. Open circuit voltage differences As mentioned above the n-Cz solar cell shows an open circuit voltage (Voc) value of 671 mV which is ~10 mV higher than for p-Cz (662 mV) and np+ epi (661 mV). This difference corresponds to an efficiency gain of 0.33 %abs. To evaluate whether local recombination center causes the differences of the Voc, photoluminescence images are taken in Voc mode (excitation wavelength 808 nm). In Fig. 3 it is shown that no local spots with higher recombination can be observed. Hence, the higher Voc-values of the n-Cz solar cells can be related to a higher quality over the entire wafer area.
Fig 3. Photoluminescence images taken in Voc mode (scale 0–60k counts).
3.2. Fill factor analysis The differences in fill factor of the n-Cz, p-Cz and the np+ epi solar cells are summarized in Table 2. The fill factor analysis is based on the method proposed in Ref. [16]. The np+ epi solar cell gives the highest fill factor value of 82.1 % which is 1.9 %abs above the level of the n-Cz solar cell (FF = 80.2 %) and 2.9 %abs above the level of the p-Cz solar cell (FF = 79.2 %). The high fill factor level of the np+ epi solar cell can be attributed to a low series resistance of 0.1 Ohm cm² which corresponds to a fill factor loss of 0.55 %abs. The reason for the low series resistance is the high conductance of the integrated epitaxial boron doped p+-silicon layer. The series resistance of the n-Cz solar cell with a boron diffused rear p+-silicon layer has been 0.58 Ohm cm² which is slightly lower than the value obtained for the p-Cz solar cell (0.66 Ohm cm²). The pseudo fill factor (pFF) level achieved for the n-Cz solar cell is 83.3 % which is about 0.5 %abs above the level of the p-Cz and 0.7 %abs above the level of the np+ epi solar cells. These differences are also reflected by the fill factor loss caused by non-ideal recombination mechanisms. In the case of the np+ epi solar cells the high nonideal recombination indicates that its limits mainly the overall fill factor level of 82.1 %. Hence, FF-values of 82.6 % would be possible when the pFF of the np+ epi solar cells would be similar to those of the n-Cz solar cell which would increase the conversion efficiency by 0.13 %abs. Table 2. Fill factor loss caused by series resistance losses (FF-loss Rs), calculated series resistance (Rs), pseudo fill factor (pFF) and fill factor loss caused by non-ideal recombination (FF-loss J2) using I–V and SunsVoc data according to the method proposed in Ref. [16]. FF-loss Rs (%)
Rs (Ohm cm²)
pFF (%)
FF-loss J2 (%)
n-Cz
Cell Type
3.15
0.58
83.31
0.85
p-Cz
3.57
0.66
82.76
1.25
np+ epi
0.55
0.10
82.60
1.30
682
Stefan Bordihn et al. / Energy Procedia 92 (2016) 678 – 683
3.3. Short circuit current density analysis The short circuit current (Jsc) losses are analyzed using the method proposed in Ref. 17 and 18. The model uses the internal quantum efficiency (IQE) data, which are shown for the investigated solar cells in Fig 4-a. The model separates the short circuit current loss into the contributions: (a) front surface reflection, (b) front surface escape, (c) grid shading and (d) non-perfect IQE, see Fig. 4-b. The calculated loss contributions are listed in Table 3.
Fig 4. (a) Internal quantum efficiency data of the best n-Cz, p-Cz and np+ epi solar cells. (b) Schematic illustration of short circuit current (Jsc) loss contributions according to the method proposed in Ref. 17 and Ref. 18.
Table 3. Contributions of short circuit current losses (mA/cm²) for the best n-Cz, p-Cz and np+ epi solar cells. Cell Type
Front surface escape
Front surface reflectance
Grid shading
Non-perfect IQE
n-Cz
1.5
0.2
1.6
2.7
p-Cz
1.2
0.3
1.6
2.9
np+ epi
1.1
0.2
1.6
3.4
The main Jsc-loss for the n-Cz solar cell can be attributed to non-absorbed light that escapes at the front side after having reflected at the rear (called front surface escape). The Jsc-loss caused by the front surface escape is for n-Cz 0.3 mA/cm² higher than for p-Cz. This is likely related to the different processing of the n-Cz wafers which feature a smoother rear surface created by the cleaning steps prior to the BBr3 diffusion process. The Jsc of the np+ epi solar cell (Jsc = 40.0 mA/cm²) is lower by 0.6 mA/cm² due to a higher front side recombination compared to n-Cz solar cell (Jsc = 40.4 mA/cm²). The reason for this loss is presumably the lower base resistivity of the np+ epi wafer, ~6 Ohm cm, in comparison to the value of ~10 Ohm cm in the case of the n-Cz wafer. However, the loss of the Jsc due to a lower base resistivity is to some extend balanced by a slightly higher fill factor of the solar cells. Hence, the overall efficiency loss can be not directly extracted from the determined Jsc-loss. The impact of the base resistivity on the J–V data has been shown in previous work [7].
Stefan Bordihn et al. / Energy Procedia 92 (2016) 678 – 683
4. Summary In this article, we report on our work on solar cells made from n-type Czochralski silicon in comparison to results of standard p-type Cz silicon and epitaxially grown n-type silicon wafers with built-in boron rear side emitter. Using an extended version of the Q.ANTUM process flow we achieved efficiencies up to 21.8 % (in-house measurement) with standard screen printing metallization on front and rear and four bus bar design. A gain of approximately 0.6 % compared to p-type Cz silicon solar cells processed in parallel with the Q.ANTUM technology was shown. The solar cell made from n-type epitaxially grown silicon gives an efficiency of 21.7 % and a high fill factor value of 82.1 %. The high fill factor is attributed to the high conductivity of the integrated epitaxial boron doped p+-silicon layer. Additionally, the overall fill factor level can be increased further by decreasing the non-ideal recombination losses which have caused a reduction in the pseudo fill factor. The high efficiency obtained with np+ epi wafers using robust cell processing gives the chance to eliminate the need of a complex boron diffusion used in the conventional n-type Cz silicon process flow.
Acknowledgements Special thanks to the staff of the Reiner Lemoine Research Center and pilot line at Hanwha Q CELLS.
References [1] F. Kersten, P. Engelhart, H.-C. Ploigt, A. Stekolnikov, T. Lindner, F. Stenzel, M. Bartzsch, A. Szpeth, K. Petter, J. Heitmann and J. W. Müller, Sol. Energy Mater. Sol. Cells 142 (2015) pp. 83–86. [2] D. N. R. Payne, C. E. Chan, B. J. Hallam, B. Hoex, M. D. Abbott, S. R. Wenham, D. M. Bagnall, RRL Phys. Stat. Solidi 2016 doi: 10.1002/pssr.201510437. [3] K. Bothe, J. Schmidt, J. Appl. Phys. 99, 013701 (2006). [4] D. C. Walter, B. Lim, and J. Schmidt, Prog. Photovolt: Res. Appl., (2016) doi: 10.1002/pip.2731. [5] A. Cuevas, C. Samundsett, M. J. Kerr, D. H. Macdonald, H. Mäckel and P. P. Altermatt, in Proc 3rd WCPEC Osaka, Japan 2003 pp. 963–966. [6] A. Richter, S. W. Glunz, F. Werner, J. Schmidt, A. Cuevas, Phys. Rev. B 86, 165202 (2012). [7] V. Mertens, T. Ballmann, J. Y. Lee, M. Junghänel, F. Stenzel, L. Brandt, A. Schulze, P. Engelhart, J. W. Müller, P. Wawer, and K. H. Küsters, Energy Procedia 27 (2012) pp. 53–58. [8] V. Mertens, S. Bordihn, A. Mohr, K. Petter, J. W. Müller, D. J. W. Jeong, R. Hao, T. S. Ravi, “21.4% Efficient Fully Screen Printed n-Type Solar Cell on Epitaxially Grown Silicon Wafers With Built-In Boron Rear Side Emitter“, in Proc. 31st EUPVSEC, Hamburg, Germany 2015, pp. 1000–1002. [9] J. Cho, H.-N.-R. Shin, J. Lee, Y. Choi, J. Lee, H. Oh, T. Kim, M. Hwang, E.-C. Cho, Energy Procedia 77 (2015) pp. 279–285. [10] B. Lim, T. Brendemühl, T. Dullweber, R. Brendel, IEEE J. Photovolt. 6, 2 (2016) pp. 447–453. [11] http://www2.imec.be/be_en/press/imec-news/imec-22-5-percent-npert-solar-cel.html (last access 04.03.2016) [12] R. Lantzsch, V. Mertens, A. Mohr, S. Bordihn, S. Hörnlein, J. Cieslak, L. Niebergall, P. Kowalzik, T. Ballmann, M. Schaper, K. Petter, J. W. Müller, to be presented at nPV Workshop Chambéry, France 2016. [13] R. Hao, T. S. Ravi, V. Siva, J. Vatus, D. Miller, J. Custodio, K. Moyers, Proc. Mater. Res. Soc. Symp. Proc., 1666, 716 (2014). [14] M. Hermle, F. Granek, O. Schultz, S. W. Glunz, J. Appl. Phys. 103, 054507 (2008). [15] P. Engelhart, D. Manger, B. Klöter, S. Hermann, A. Stekolnikov, S. Peters, H.-C. Ploigt, A. Eifler, C. Klenke, A. Mohr, G. Zimmermann, B. Barkenfelt, K. Suva, J. Wendt, T. Kaden, S. Rupp, M. Fischer, J. W. Müller, P. Wawer, “Q.ANTUM – Q-Cells next generation high-power silicon cell & module concept”, in Proc. 26th EUPVSEC, Hamburg, Germany 2011, pp. 821–826. [16] A. Khannah, T. Müller, R. A. Stangl, B. Hoex, P. K. Basu, A. G. Aberle, IEEE J. Photovolt. 3, 1170 (2013). [17] A. G. Aberle, W. Zhang, B. Hoex, Energy Procedia 8 (2011) pp. 244–249. [18] L. Tous, R. Russell, M. Debucquoy, N. Posthuma, F. Duerinckx, R. Mertens, J. Poortmans, Energy Procedia 38 (2013) pp. 467–473.
683