22.2% Efficiency n-type PERT Solar Cell - Science Direct

Available online at www.sciencedirect.com

ScienceDirect Energy Procedia 92 (2016) 399 – 403

6th International Conference on Silicon Photovoltaics, SiliconPV 2016

22.2% efficiency n-type PERT solar cell Wenhao Cai*ˈShengzhao YuanˈYun ShengˈWeiyuan DuanˈZigang Wangˈ Yifeng ChenˈYang Yang, Pietro. P. Altermatt, Pierre J. Verlinden and Zhiqiang Feng State Key Laboratory of Photovoltaic Science and Technology, Trina Solar, No. 2 Trina Road, Trina PV Park, New District, Changzhou, Jiangsu Province, China 213031

Abstract N-type PERT (passivated emitter rear totally diffused) silicon solar cells promise high and stabilized conversion efficiencies. As relative big contact recombination and shading losses were introduced by traditional front screen-printed metallization, we applied Ni/Cu/Ag plating to form the front metallization grid, and we also applied PVD (physical vapor deposited) aluminum for the rear side contacts for better optical reflection. With help of laser ablation, the front and rear metallization fraction can be reduced to less than 1%, which benefits both Voc and Jsc. In order to increase Jsc further, we optimized the front AR coating and the back reflection. Based on these technologies, we reach efficiencies of up to 22.2% on 5 inch, commercial grade Cz n-type wafers. For further improvement, we reproduce the cell performance by simulation with Sentaurus to do a power loss analysis to quantify recombination and resistive losses. These simulations indicate that emitter recombination and the internal resistance are the top two power loss sources for our 22% PERT cells. © 2016 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/). Peer review reviewbybythe thescientific scientific conference committee of SiliconPV 2016 under responsibility of PSE AG. Peer conference committee of SiliconPV 2016 under responsibility of PSE AG. Keywords: n-type; PERT; plating;

1. Introduction N-type PERT solar cells promise high and stabilized conversion efficiency, the recent champion efficiency is reported as 22.5% in [1] by A. Urueña et al. Traditional screen-printed metallization has limited the efficiency increase due to a considerable contact area fraction and because the emitter needs to be highly doped at the contact. Using laser ablation, plating and PVD (physical vapor deposition) of Al, the total metallization and the front shading were significantly reduced, while the back reflection can be considerably increased with optimized back film stacks [2-4]. In this paper, the recent developments in our laboratory are described. To improve the efficiency further, a power loss analysis is carried out to identify the main limitations and to decide on further improvement strategies.

1876-6102 © 2016 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.2016.07.119

400

Wenhao Cai et al. / Energy Procedia 92 (2016) 399 – 403

2. Experiment The n-type PERT solar cells were fabricated on 5 inch Cz phosphorous-doped silicon wafers with resistivity of 1–5 cm and thickness of 180 ȝm. The process flow mainly includes texturing, boron diffusion, single-side polishing, phosphorous diffusion, passivation, laser ablation process, Al PVD deposition and front metal plating. The as-fabricated cell structure has random pyramids, a homogenous emitter, a stack of dielectrics and plated fingers on the front side, and a chemical polished surface, a homogenous BSF, a stack of dielectrics and a PVD metal film on the rear side, as illustrated in Fig. 1. Parts of the cells were characterized with optical microscopy, ellipsometry, four-probe electric measurement, and electrochemical capacitance voltage (ECV) measurement of the dopant profiles. From the finished cells, IQE data were collected for wavelength in the range of 300-1200 nm at 10 nm intervals with a set-up from PV measurements, Model QEX10. In lifetime samples, quasi-steady-state photo conductance (QSSPC) was used to measure the saturation current densities to evaluate the diffusions, and to determine the bulk lifetime. The characterized parameters served as input for the simulator Sentaurus to analyze the cell performance.

Fig. 1. Schematic of the fabricated n-type PERT solar cells.

3. Results and discussions Table 1 shows cell performance of one n-PERT cell of the past, measured in-house and subsequently confirmed by Fraunhofer ISE CalLab. It seems that the Jsc measured in-house is underestimated, and the FF is overestimated. However, the cell efficiency is comparable. Table 1. I-V parameters of one n-type PERT solar cell Condition

Jsc (mA/cm2)

Voc (mV)

FF (%)

Eff (%)

In-house

39.79

0.6841

80.50

21.91

Fraunhofer ISE CalLab

40.13

0.6838

80.11

21.98

We have used this calibrated cell for calibration of our more recent cells. Table 2 lists the I-V parameters of our newest batch with the improved laser ablation process. Per batch there are only seven cells processed due to restricted capacity of our PVD equipment. The batch has a median efficiency of 22.14%. The champion cell reaches an efficiency of 22.20%. To avoid any capacitance effects in the I-V sweep, an advanced algorithm to correct FF provided by H.A.L.M is adopted. The high Jsc can be attributed to the following factors: Table 2. I-V parameters of n-type PERT solar cells, measured in-house. Jsc (mA/cm2)

Voc (mV)

FF (%)

Eff (%)

Avg.

40.60

687.2

79.34

22.14

Max.

40.62

688.3

79.39

22.20


x Fig. 2 shows that both the EQE and the IQE are high at short wavelengths, due to the lightly-doped, homogeneous emitter. The EQE and IQE curves are also high at long wavelengths, indicating that the back surface field (BSF) passivates the bulk very well without causing much Auger recombination.; x The rather small differences between EQE and IQE are due to the optimized anti-reflection coating (ARC) stack on the front and rear surfaces; x The plated narrow fingers reduce both the front shading and front contact recombination, see Fig. 3. The contact width is almost 15 ȝm, determined by the laser opening width, which effectively bring to a metal ratio below 1%. And the final finger width is around 27 ȝm. x The optimized laser opening pattern combined with the PVD Al metallization on the rear decrease the rear metallization fraction to 0.4% and, accordingly, reduce the contact recombination while still keeping the contact resistivity manageably low.

Fig. 2. QE and reflectance curves of the champion cell.

Fig. 3. SEM image of plated Ni/Cu/Ag fingers.

4. Power loss simulation 4.1. Simulation of recombination losses The relevant device parameters, measured as outlined in the previous section, are now used as input to Sentaurus simulations. Sentaurus does not accept J0 as input, but the dopant profiles. The surface recombination velocity is

401

402


adjusted in a Sentaurus simulation of the J0 experiment such that it reproduces the measured J0. The resulting recombination velocity can then be compared to experiments performed in other institutions as in Refs. [5-10] for all kinds of passivation techniques. This gives not only the correct amount of recombination losses at all voltages, but also a distinction between Auger and SRH in the bulk and SRH at the surface, and an assessment of the passivation quality. Besides of this, Sentaurus yields also recombination in the space charge regions, the collection efficiency and – if relevant – impact ionization of UV light in the diffusion. We find these analytic tools more useful for our loss assessment and road maps than the ease of handling of the simulator Quokka. Figure 4 shows the recombination losses in the various device parts, expressed as current-density. The emitter dominates the overall recombination losses. However, recombination in the rear diffusions increases faster than in the emitter towards Voc, so future improvements of the emitter will be additionally offset by increased losses at the rear. A similar plot but for the SRH, Auger and radiative recombination losses in the base (not shown here) reveals that mainly Auger recombination determines the dynamics in dependence of applied voltage. All of this needs to be considered in road maps that suggest specific processing improvements. Also, note that SRH recombination at both front and rear contacts do not substantially reduce efficiency any more due to the small metallization fractions. The metallization fraction may even be slightly increased to lower the resistive losses.

Fig. 4. Amount of recombination, expressed in current-density, in the various device parts, simulated with Sentaurus using the measured input parameters as input.

4.2. Simulation of resistive losses The series resistance is simulated by a series of three IV curves under three different illumination intensities [11], but all near 1-sun (otherwise, injection-dependent lifetimes would influence the extracted Rs value as is the case e.g. in suns-Voc curves). The resulting series resistance is shown in Fig. 5. The internal Rs is slightly bias-dependent due to the varying amount of lateral current flow in the emitter compared to the base, and due to a slight decrease of the base resistivity with higher bias and, hence, higher injection levels. Internally, there is Rs.mpp = 0.648 :cm2, Rs.oc = 0.607 :cm2, and the external Rs.met is 0.282 :cm2, independent of bias. Particularly the internal Rs is rather large and bears a significant potential for reduction. Nevertheless, the achieved FF is 79.39.


Fig. 5 Lumped series resistance in dependence of bias, simulated with Sentaurus and a spice simulation.

5. Conclusion In this work, n-type PERT cells with high efficiencies beyond 22% are reported. Metal plating, Al PVD metallization, a lightly doped emitter and BSF contribute to the good performance. Sentaurus simulations suggest that the recombination losses in the emitter dominate the total recombination losses. Moreover, emitter doping and passivation, rear contact resistance are also limitations of high efficiency. Acknowledgements This work is financially supported by the National “863” Project (2015AA050302) and the Natural Science Foundation of Jiangsu Province for Young Scientist (No. BK20140273). References [1] Urueña1A, Aleman M, Cornagliotti E, Sharma A, Deckers J, Haslinger M, Tous L, Russell R, John J, Yao Y, Söderström T, Duerinckx F, Szlufcik J, Beyond 22% large area n-type silicon solar cells with front laser doping and a rear emitter, 30th EUPVSEC, 2015. [2] Steinhauser B , Mansoor MB , Jäger U, Benick J, Hermle M, Firing-stable PassDop passivation for screen printed n-type PERL solar cells based on a-SiNx:P, Solar Energy Materials and Solar Cells, vol. 126, 96-100 (2014) [3] Bartsch J, Kamp M, Hartleb D, Wittich C, Mondon A, Steinhauser B, Feldmann F, Richter A, Benick J, Glatthaar M, Hermle M, Glunz SW, 21.8% Efficient n-type Solar Cells with Industrially Feasible Plated Metallizatio Energy Procedia, vol.55, 400-409 (2014) [4] Benick J, Steinhauser B, Müller R, Bartsch J, Kamp M, Mondon A, Richter A, Hermle M, Glunz S, High efficiency n-type PERT and PERL solar cells, Photovoltaic Specialist Conference (PVSC), 2014 IEEE 40th. p. 3637 – 3640 [5] Black LE, Allen T, McIntosh KR, Cuevas A, Effect of boron concentration on recombination at the p-Si-Al2O3 interface, J. Appl. Phys. 115, 093707 (2014). [6] Altermatt PP, Schumacher JO, Cuevas A, Kerr MJ, Glunz SW, King RR, Heiser G, Schenk A, Numerical modeling of highly doped Si:P emitters based of Fermi-Dirac statistics and self-consistent material parameters" J. Appl. Phys. 92, 3187-3197 (2002). [7] Kimmerle A, Rahman MM, Werner S, Mack S, Wolf A, Richter A, Haug H, Precise parameterization of the recombination velocity at passivated phosphorus doped surfaces, J. Appl. Phys. 119, 025706 (2016). [8] Kerr MJ and Cuevas A, Recombination at the interface between silicon and stoichiometric plasma silicon nitride”, Sem. Sc. Tech. 17, 166– 172 (2002) . [9] Albohn J, Füssel W, Sinh ND, Kliefoth K, Fuhs W, Capture cross sections of defect states at the Si/SiO2 interface, J. Appl. Phys. 88, 842 (2000). [10] McIntosh KR, Baker-Finch SC, Grant NE, Thomson AF, Singh S, Baikieb ID, Charge density in atmospheric pressure chemical vapor deposition TiO2 on SiO2-passivated silicon, J. Elchem. Soc. 156, G190-G195 (2009). [11] Fong KC, McIntosh KR, Blakers AW, Accurate series resistance measurement of solar cells, Prog. PV 21, 490 – 499 (2013).

403