High Efficiency on Boron Emitter n-Type Cz Silicon Solar Cells With Industrial Process

118

IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 1, NO. 2, DECEMBER 2011

High Efficiency on Boron Emitter n-Type Cz Silicon Solar Cells With Industrial Process Yannick Veschetti, Raphael Cabal, Pierre Brand, Vincent Sanzone, Gaetan Raymond, and Armand Bettinelli

Abstract—In this study, we describe the fabrication of n-type solar cells using an industrial process. The open-circuit voltage limitation is discussed by investigating the influence of the screenprinted metallization at the front and rear sides. Efficiencies above 19.0% were obtained on 125PSQ Cz–Si wafers with a reference process. Narrow front metalized fingers were deposited by means of a stencil screen associated with a silver-plating step. Recombination below the contacts due to a metal area reduction resulted in a Vo c improvement up to 5 mV. The process flow was then modified to develop an improved back-surface field (BSF). Measurements of implied Vo c values on the cell precursors confirmed the interest of reducing the BSF doping level. Nevertheless, no gain in efficiency was achieved on full-metalized solar cells. A Vo c limitation due to the metallization impact was also observed on several batches of solar cells. Specific light beam induced current measurements confirmed the need for a deep BSF profile to minimize recombination activity under the contact area. Finally, stability of the fabricated cell was investigated under light soaking. After over 100-h exposures, the cell efficiency was improved by 0.2% absolute, leading to a maximum efficiency of 19.3%. Index Terms—Photovoltaic cells, silicon.

I. INTRODUCTION ONVENTIONAL silicon solar cells are fabricated on p-type substrates. Nevertheless, n-type silicon is known to provide excellent electrical properties resulting in higher efficiency potential. The highest efficiencies obtained so far on industrial solar devices involve technologies implemented on ntype substrates [1], [2]. N-type silicon is also considered for its stabilized electrical performances. For instance, B-doped Cz–Si is known to endure light-induced degradation (LID), contrary to Cz Si(n) [3]. The front boron emitter bifacial cell architecture could contribute to cost reduction by allowing higher efficiencies with a cost-effective process on Cz Si(n) substrates [4]–[6]. This cell structure developed in this study is presented in Fig. 1. It includes a boron emitter on front side and a fully P-diffused backsurface field (BSF) on the back. BCl3 diffusion—which has already been introduced in previous studies [7]—was used here for the emitter formation. The emitter passivated by SiO2 \SiN stack was used since saturation current density values below 100 fA/cm2 were already obtained [7]. The first aim of this pa-

C

Manuscript received June 28, 2011; revised August 23, 2011; accepted August 24, 2011. Date of publication October 6, 2011; date of current version December 27, 2011. The authors are with the Atomic Energy Center, LITEN, National Institute for Solar Energy, 73370 Le Bourget du Lac, France (e-mail: [email protected]; [email protected]; [email protected]; vincent. [email protected]; [email protected]; [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JPHOTOV.2011.2167958

Fig. 1. Capture cross section of n-type silicon solar cells fabricated according to the reference process

per is to describe the latest improvements in the cell process, including advanced developments in the front metallization. The second aim is to evaluate the Vo c limitations of this device by investigating the interaction between the metal area and the surface passivation. Finally, the stability of the fabricated cells under light-soaking measurements is presented.

II. EXPERIMENTAL DETAILS Solar cells were fabricated on n-type 148.6-cm2 substrates according to the reference process flow given in Fig. 2. Substrates are Cz type since we foresee studying the LID impact on the fabricated devices. Both sides of the wafer were alkaline textured and RCA cleaned. A silicon oxide plasma-enhanced chemical vapor deposition diffusion barrier was used to allow the B- and P-diffusion on one side of the cell. The boron diffusion was performed at a temperature of 940 ◦ C, resulting in a sheet resistance of 60 Ω/sq. The thermal oxidation step induced an increase of the emitter sheet resistance to 90 Ω/sq [7]. Front metalized fingers were deposited by screen printing with a width of 100 μm. In order to optimize the BSF profile, an alternative process flow was developed (see Fig. 2). To avoid the redistribution of the BSF during the thermal oxidation, the POCl3 diffusion was performed after the front-side process. Different diffusion temperatures were selected in order to reduce the BSF doping level. Secondary Ion Mass Spectroscopy (SIMS) profiles are presented further below in the study. For such process, the P-BSF was only passivated by an SiN layer. The implied Vo c was measured before and after the firing step on cell precursor for both processes (prior to the metallization step). This method helps to compare the passivation potential independently of the metallization step.

2156-3381/$26.00 © 2011 IEEE

VESCHETTI et al.: HIGH EFFICIENCY ON BORON EMITTER N-TYPE CZ SILICON SOLAR CELLS WITH INDUSTRIAL PROCESS

119

TABLE II PERFORMANCE OF SOLAR CELLS COMPARING STANDARD SP TECHNIQUE WITH STENCIL PRINTING + SILVER PLATING

Fig. 2. Fabrication of n-type silicon solar cells. Reference process (left) and advanced process for BSF optimization (right). TABLE I AVERAGE AND BEST N-TYPE SOLAR CELL (148.6 CM2 ) UNDER STC (1000 W/M2 ; 25 ◦ C) Fig. 3. Evolution of FF and Jsc of solar cell for different Ag plating time (one cell per time). The reference values using standard screen printing are given on the left side.

III. REFERENCE PROCESS The reference process with the deep BSF results in an average efficiency of 18.9% when measured on a full-metalized chuck (see Table I). Cell certification at Instituts f¨ur Solare Energiesysteme (ISE) CalLab using an adapted bifacial chuck gives an efficiency of 18.7% due to a difference in short-circuit current (influence of rear chuck reflection). This loss should be minimized at the module level due to the back sheet reflectivity. The best efficiency obtained with referenced process was 19.1%. Recent published works on similar cell architecture show slightly higher efficiency due to superior open-circuit voltage values. Benick et al. achieved Vo c values closed to 650 mV using Al2 O3 /SiN passivation stack and plated front contact on large solar cells [8]. B¨oscke et al. presented efficiency of 19.7% on 156PSQ Cz–Si with a Vo c of 643 mV with a similar process to the reference one presented in Fig. 2 [9].

step at different times. The open-circuit voltage was not affected by the electroplating step. Fig. 3 describes the evolution of the FF and short-circuits current values with the plating time. As expected, a compromise was found to increase the finger’s conductivity without inducing too much shadowing. An optimized plating time of 11 min was found, resulting into a limited improvement in efficiency (+0.1% absolute). The corresponding finger width was 70 μm. For longer deposition time of 20 min, the FF values saturate to 78% as the reference metallization. This shows that no additional contact resistance occurs on the advanced metallization devices, despite the reduced contact area. As a conclusion, the combination stencil/electroplating solution leads to a limited gain in efficiency. In terms of mass of deposited silver, this technique does not result in a considerable cost reduction. Nevertheless, further optimizations in the first print line definition could allow the reduction of finger resistance without inducing an important shadowing effect.

IV. FRONT METALLIZATION IMPROVEMENT A study was carried out in order to understand the effect of contact area on the surface recombination. The cells performances for different metallization schemes are given in Table II. Very thin metalized fingers were deposited via the use of a stencil screen, resulting into a finger width of 50 μm. At this stage, the fill factor (FF) remains very low due to the relatively high finger resistance (1 Ω/cm). The short-circuit current was improved by 0.6 mA/cm2 due to a reduced shadowing. An increase in the open circuit which can be explained by a reduction in the recombination activity below the contacts is also worth noticing. In a second stage, the solar cells were subjected to a silver-plating

V. REAR-SIDE DEVELOPMENT Open-circuit voltage limiting factor was determined by computing symmetrical (n+ /n/n+ lifetime samples implied Vo c as a function of surface recombination velocity (SRV) on PC1D). The simulated structures involve different BSF doping profiles that can be achieved by POCl3 diffusion. The different diffusion profiles are presented in Fig. 4. Specific low energy conditions were chosen in order to obtain a better resolution at the surface at the expense of the detection limit. The resulting profile of the BSF obtained by the reference process flow (see Fig. 4) has a very deep (n4+ due to the

120

Fig. 4.

IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 1, NO. 2, DECEMBER 2011

SIMS profile of the phosphorus BSF on n-type solar cells.

Fig. 6. profile.

Comparison between implied Vo c and cell Vo c for different BSF

Fig. 5. Simulated implied Vo c of a symmetrical (n+ /n/n+ ) lifetime sample versus SRV for several BSF profiles.

influence on the thermal oxide step. The increase in the phosphorus concentration at the surface can be related to the oxidation step. During the oxide growth, the phosphorus atoms accumulate in the crystal. According to the simulation results presented in Fig. 5, no drastic Vo c improvement is expected by keeping the reference n4+ BSF phosphorus profile. The Vo c remains below 647 mV, even for extremely low SRV. Nevertheless, the simulation study indicates that higher open-circuit voltage values could be achieved with a narrower doping profile, providing excellent surface passivation is associated with it. An experimental procedure was thus developed in order to fabricate solar cells with a specific BSF profile (process flow described in Fig. 2). For the very narrow BSF (n+ ), a high doping of 35 Ω/sq was locally made by means of laser doping from the phosphorus silicon glass (PSG) source to ensure a low contact resistance by screen-printing. This technology is generally applied for the selective emitter formation on p-type solar cells [10]. For each BSF profile, the implied Vo c of the solar cell was measured by quasi-steady-state photoconductance after the firing step in order to improve the emitter passivation of the SiN/SiO2 stack (see Fig. 6). In agreement with simulation results, the reduction of the profile depth results in an improvement of the implied Vo c . The narrow profile (n+ ) gives the best potential.

Fig. 7. Effective diffusion length measured in a solar cell with n4 + , n3 + , and n2 + BSF profiles

Nevertheless, the final Vo c measured on metalized cells was lower than the reference cells with the deep n4+ BSF. This is clearly the result of a very high recombination activity below the contacts. Light-beam-induced current (LBIC) measurements presented in Fig. 7 were made between two metalized fingers and along the central metalized plot required for the I–V measurement. A decrease in the LBIC signal can be observed on the central plot for the non-redistributed BSF profiles (n+ , n2+ , and n3+ ). The effect is more pronounced when the BSF is lighter, which indicates a more intense recombination activity below the contact. The use of a deep BSF (n4+ ) induces an efficient screening of the near surface from the metal. In terms of cell

VESCHETTI et al.: HIGH EFFICIENCY ON BORON EMITTER N-TYPE CZ SILICON SOLAR CELLS WITH INDUSTRIAL PROCESS

Fig. 8. Comparison between Implied Vo c on cell precursor and final Vo c on complete metalized cells for several batches of solar cells fabricated with an identical process flow.

121

Fig. 9. Delta in open-circuit voltage for a cell under light-soaking at AM1.5 at 50 ◦ C. TABLE III PERFORMANCE OF n-TYPE SOLAR CELLS BEFORE AND AFTER LIGHT SOAKING

performances, the use the selective BSF scheme shows a considerable drop in Vo c after metallization and, finally, leads to the same range of efficiency as the reference cell with the deep BSF profile. As a consequence, an alternative selective BSF process flow must be investigated in order to develop a deep BSF profile below the contacts and a shallow BSF between the metallization. VI. Vo c COMPARISON BETWEEN CELLS PRECURSOR AND METALIZED CELLS The limitation of the open-circuit voltage on complete solar cells was also investigated through the fabrication of different batches using the reference process flow. For each batch of solar cells, the implied Vo c was measured on cell precursors (p+ /n/n+ ) after the firing step to simulate the contact firing step (see Fig. 8). The implied Vo c was found to vary between 640 and 660 mV. This gap of 20 mV was due to the variation of several parameters (oxidation temperature, surface preparation, etc.) and to the substrate’s quality. After the metallization step, the final Vo c was found to be limited between 630 and 632.5 mV. This result confirms that the open-circuit voltage is strongly limited by the screen-printed metallization, which is consistent with the results presented previously. At this stage of development, further improvements in terms of emitter and BSF passivation would not result in a gain in efficiency. VII. CELL STABILITY The evolution of cells Vo c was measured at 1 sun at 50 ◦ C for more than 100-h duration (see Fig. 9). The Vo c was shown to be improved before getting stabilized at +1.5 mV toward initial value. This specific behavior highlights the advantage of n-type toward B-doped Si cells that are known to be degraded after a prolonged exposure to light (LID). The electrical parameters of the reference cell are presented in Table III, before and after a prolonged illumination. An absolute gain (+0.2%) was observed after light exposure, leading to a maximum efficiency of 19.3%. The cause of such an improvement could not be explained at this

stage of the study. Longer exposure time should be considered to verify the stability of the SiO2 /SiN passivation stack. VIII. CONCLUSION The latest improvements made on the n-type reference process using BCl3 diffusion were presented in this study. Maximum efficiency of 19.1% was achieved on large-area Cz substrates (148.6 cm2 ). Front-side metallization combining narrow printed fingers with electroplating resulted in a limited gain of 0.1% absolute due improvements in Vo c and Jsc . The limitation of the open-circuit voltage of the fabricated devices was investigated. In agreement with simulation data, the process flow was adapted in order to form shallow BSF to improve rear passivation. The passivation potential through implied Vo c measurements on cell precursors was found to be superior to the reference process with a very deep BSF. Nevertheless, the integration of metallization onto the fabricated devices resulted on an important loss in the final Vo c value. This showed the detrimental impact of the screen-printed contacts on the cell performance. Differences between implied Vo c and final cell Vo c were also presented for several batches of solar cells made with an identical process. The open-circuit voltage was limited to a maximum value of 632 mV due to metallization. As a consequence, this study underlines the importance of developing an advanced cell architecture combining selective emitter and selective BSF to reach higher efficiency. Finally, the cell stability was investigated under light soaking at 50 ◦ C. There was no degradation of the cell performances after more than 100 h of light soaking. An improvement in efficiency was even measured, which resulted in the best efficiency of 19.3%. The origin of this gain still needs to be investigated.

122

IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 1, NO. 2, DECEMBER 2011

REFERENCES [1] T. Mishima, M. Taguchi, H. Sakata, and E. Maruyama, “Development status of high efficiency HIT solar cells,” Sol. Energy Mater. Sol. Cells, vol. 95, pp. 18–21, 2011. [2] P. J. Cousins, D. D. Smith, H.-C. Luan, J. Manning, T. D. Dennis, A. Waldhauer, K. E. Wilson, G. Harley, and W. P. Mulligan, “Gen III: Improved performance at lower cost,” in Proc. 35th IEEE Photovoltaic Spec. Conf., Honolulu, HI, 2010, pp. 275–277. [3] T. Schutz-Kuchly, J. Veirman, S. Dubois, and D. R. Heslinga, “LightInduced-Degradation effects in boron–phosphorus compensated n-type Czochralski silicon,” Appl. Phys. Lett., vol. 96, pp. 093505-1–093505-3, 2010. [4] A. R. Burgers, R. C. G. Naber, A. J. Carr, P. C. Barton, L. J. Geerligs, X. Jingfeng, L. Gaofei, S. Weipeng, A. Haijiao, H. Zhiyan, P. R. Venema, and A. H. G. Vlooswijk, “19% efficient n-type Si solar cells made in pilot production,” in Proc. 25th Eur. Photovoltaic Solar Energy Conf. Exhib., Valencia, Spain, Sep.2010, pp. 1106–1109. [5] V. D. Mihailetchi, J. Jourdan, A. Edler, R. Kopecek, R. Harney, D. Stichtenoth, J. Lossen, T. S. B¨oscke, and H.-J. Krokoszinski, “Screenprinted n-type silicon solar cells for industrial applications,” in Proc. 25th Eur. Photovoltaic Solar Energy Conf. Exhib., Valencia, Spain, Sep. 2010, pp. 1446–1448.

[6] A. Richter, S. Henneck, J. Benick, M. H¨orteis, M. Hermle, and S. Glunz, “Firing stable Al2O3/SiN layer stack passivation, for the front side boron emitter of n-type silicon solar cells,” in Proc. 25th Eur. Photovoltaic Solar Energy Conf. Exhib., Valencia, Spain, Sep. 2010, pp. 1453–1459. [7] Y. Veschetti, V. Sanzone, F. Diagne, N. Auriac, and D. Heslinga, “High efficiency n-type silicon solar cells with novel diffusion technique for emitter formation,” in Proc. 25th Eur. Photovoltaic Solar Energy Conf. Exhib., Valencia, Spain, Sep. 2010, pp. 2241–2244. [8] J. Benick, A. Richter, D. Suwito, U. J¨ager, M. H¨orteis, A. Kalio, M. Hermle, and S. Glunz, “High-efficiency n-type solar cells with a front side boron emitter,” presented at the NPV Workshop, ISC Konstanz, Konstanz, Germany, May 2011. [9] T. B¨oscke, R. Hellriegel, D. Kania, A. Helbig, M. Weiß, C. Wiendl, D. Stichtenoth, T. Aichele, J. Lossen, and H.-J. Krokoszinski, “Development of industrial n-type cells with boron emitters,” in Proc. NPV Workshop, ISC Konstanz, Konstanz, Germany, May, 2011. [10] B. Paviet-Salomon, S. Gall, S. Manuel, R. Monna, and A. Slaoui, “Laser doped selective emitter silicon solar cells,” presented at the Eur. Photovoltaic Solar Energy Conf. Exhib., Valencia, Spain, Sep. 2010.

Authors’ photographs and biographies not available at the time of publication.