19.2% efficient c-Si solar cells using ion implantationJian Wu, Yumei Li, Xusheng Wang, Linjun Zhang Canadian Solar Inc., 199 Lushan Road, SND, Suzhou, Jiangsu, ChinaABSTRACTIon-implantation offers numerous advantages for solar cell manufacturing. Canadian Solar Inc. (CSI) has developed an average efficiency 19.23%, champion efficiency 19.39% solar cell (156mm) process using a high throughput Varian (Applied Materials) Solion ion-implant tool based on a beam line design. The path to commercialization has included module process development, achieving 265W with 6×10 cells. With the installation of the high-throughput ion implanter for PV, optimized unit processes were transferred from the lab to the production floor at CSI.1. INTRODUCTIONIon implantation has been reported as a breakthrough technology in PV industry to drive down $/W cost. Ion implantation is a process of precisely introducing a known amount of energetic particles into any substrate to alter its material properties. Emitters formed by ion implantation show improved cell efficiency due to the benefit in blue wavelength region. The quality of the emitters formed by ion implantation followed by activation has clear advantages over emitters made by conventional POCl3diffusion, in terms of lattice defect recovery as well as dopant activation (no dead layer), as depicted by TEM micrographs and SIMS profiles. By tailoring the activation temperature and ambient, a layer of oxide can be formed during activation [1]. J0e data show surface passivation is better with SiO2/SiN x stack than single SiN x layer. We prove ion implantation technology is mass production ready for PV cell manufacturing, and it is an essential enabler of lower $/W.2. EXPERIMENTSFig. 1Process simplification for fabrication of ion implantation emitter cells: (a) standard POCl3 process;(b) implanted process.Ion implanted process is described in Fig.1 (a). Fig.1 (b) presents traditional POCl3diffusion solar cell process as reference. Ion implantation is a single side process, so edge isolation is not necessary. SIMS and TEM are conducted on both POCl3diffused and ion implanted samples, to understand and characterize the emitter quality. Lower surface concentration from ion implantation is observed. The precise dose control of ion implanter provides excellent uniformity and repeatability within wafer and wafer-to-wafer on 156mm pseudo sq. wafers. Metallization was optimized to account for the modified doping profile and SiO2/SiN x stack passivation layer. The internal quantum efficiency (IQE) and dopantprofile are compared between cells made by ion implantation and incumbent POCl 3 diffusion process. Modules are fabricated and characterized to ensure that the efficiency gain is fully transferred into increased module power output.3. RESULTS AND DISCUSSIONS3.1 Emitter uniformity on Si-wafersFig. 2 Sheet resistance comparisons between ion implanted (top) and POCl 3-diffused (bottom) wafers.Fig.2 shows the sheet resistance mapping with 49 points by ion implantation (top), and POCl 3-diffusion (bottom). For a fair comparison, the target sheet resistance (R sq ) for both cases was set ~60Ω/sq. The R sq standard deviation of diffused wafer is 5.47Ω/sq, larger than 2.38Ω/sq of implanted wafer. It indicates the emitter fabricated by ion implantation is relatively more uniform and acceptable for mass productionprocess.3.2 Superior Junction QualityOne concern about the implantation process is if the implanted regions can be perfectly re-grown without any residual damage within the layer. Defects in the re-grown lattice may cause increased carrier recombination in the emitter.(a)(b)(c)Fig. 3 Cross section images of: (a) as implanted; (b) furnace annealed; (c) POCl 3 diffused wafer samples, illustrating no residual defects in the crystal lattice inthe implanted emitter.Fig.3 shows a TEM comparison of implanted and diffused emitters with similar R sq values of 60Ω/sq. Image (a) shows as-implanted sample and reveals a high uniform and fully amorphized ~30nm region. After a furnace annealing, the amorphous region completely recrystallized without any defect, as shown in Fig.3 (b). The cross-section comparisons illustrate the perfectly re-grown Si crystal lattice with no dead regions in the near surface region of the implanted emitter. On contrary, POCl 3diffused image (c) presents ~10nm dark region, correlated with the junction defects, or called dead region. Image (b) also illustrates the presence of a thin layer of thermally grown silicon dioxide during the post-implantation annealing. The smoother interface between the oxide layer and the Si sample for the implanted emitters (reduced surface recombination area) combined with the high quality thermal oxide (with very few interface states) is also evident in the picture.3.3 Comparison of dopant junction profileIon implantation provides a powerful method to precisely control the amount of dopant in the emitter layer. The dopant profile may use an appropriate annealing step to maximize cell performance. For the same sheet resistance, the implanted emitter has a lower surface concentration than diffused one, thus enabling lowerrecombination in the emitter region.Fig. 4Comparison of dopant junction profiles between POCl3-diffused and ion implanted emitter. The implanted emitter with and without furnace annealed both had been measured.Figure 4 shows the SIMS profile of the emitters of POCl3-diffusion and ion-implantation before and after annealing. All test samples have similar R sq values of 60Ω/sq. POCl3emitter profile shows a thin layer with non-activated peak concentration up to 1022cm-3, relating to the dead layer. For the dopant activation profiles, the as-implanted sample presents surface concentration of 7×1020cm-3. After high temperature annealing, phosphorus atoms are activated and driven deeper into Si substrate, to form an even lower (2×1020cm-3) concentration dopant emitter.3.4 Improved Surface Passivation with thermal oxideOxide passivation has been used successfully on very high efficiency solar cells [3]. Since ion implantation does not require any parasitic dopant containing layers (such as the phosphor-silicate glass in POCl3 diffusion), it is possible to grow and retain a high quality thermal oxide for implanted emitters without the removal/ additional process. As seen in the Table 1, implied V oc@1 Sun value over 650mV of ion implanted solar cell is indicative of very low recombination in the emitters due to the excellent surface passivation after the anneal. After SiN x and firing, the value increases even more, due to H atoms passivation in Si bulk. Table 1. Implied V oc data after activation/diffusion, SiN x and firing of the ion implanted and POCl3 emitter.As further evidence of superior emitter quality, we refer to Fig. 5 that shows the internal quantum efficiencies(IQE) of POCl 3-diffused and ion implanted solar cells. The IQE of the ion implanted solar cell was higher than that of POCl 3-diffused in the short wavelength region, the so-called blue response, as shown in Fig 5. The excellent blue response of the cell is indicative of very low recombination in the emitter and excellent surface passivation, as well as the absence of dead layer of the emitter.Fig. 5 Comparison of the internal quantum efficiency (IQE) for POCl 3-diffused and ion implanted solar cell.3.6 Performance of the cell efficiency in productionIon implantation provides a powerful method to precisely control the amount of dopant in the emitter layer. The uniformity and precision doping via implant enables a repeatable process for fabrication of lightly doped emitter regions [4-6]. The efficiency distribution of over 2000pcs cells produced recently is shown in Fig. 6 with an average efficiency of 19.23% and champion efficiency of 19.39%. The absolute efficiency gain reaches 0.5%, accordingto Table 2.Fig. 6 Histogram of cell efficiency for selected cells in production.Table 2. Comparison of the average I-V parameters between ion-implanted and POCl 3-diffused solar cells of more than 2000 pcs.3.7 Module powerFull size modules (60cells) with conventional layout have been made with 19.2% ion implanted solar cells. They can achieve over 265W module power, with the cell to module (CTM) loss ~ 3%. It is proved the ion implanted process repeatability and stability for manufacturing process. The EL image of the modules reveals the good current response in Fig. 7.Fig. 7 The EL image of the module made with ion implanted cells.4. DISCUSSION & CONCLUSIONSIn this study, we have reported the approach to boost the cell performance on standard p-Cz cells with an ion implantation process flow. The key benefits for improving cell efficiency are a) Process Simplification through elimination of PSG strip and junction isolation steps b) improved quality of junction doping (lower surface dopant concentration and elimination of dead layer), c) improved surface passivation through the integration of thermal oxidation in the process flow. A large number of cells efficiency over 19.2% is achieved by using ion implantation technology. Improved cell efficiency is transferred to higher module power output over 265W. REFERENCES[1] M. B. Spitzer, C. J. Keavney, Proceedings 18th IEEE Photovoltaic Specialist Conference (1985) 43.[2] J. Ben ick, et al., “Very Low Emitter Saturation Current Densities on Ion Implanted Boron Emitters”, 25th EU- PVSEC, Valencia, Spain, 2010.[3] J. Zhao et al., ”24.5% Efficiency Silicon PERT Cells on MCZ Substrates and 24.7% Efficiency PERL Cells on Fz Substrates”, Prog. Photovolt: Res. Appl. 7, 1999, pp. 471-474.[4] R. Low, A. Gupta et al., “High Efficiency Selective Emitter Enabled through Patterned Ion Imp lantation”, 35th IEEE PVSC, Honolulu, 2010. [5] J. Benick, et al., “Very Low Emitter Saturation Current Densities on Ion Implanted Boron Emitters”, 25th EU- PVSEC, Valencia, Spain, 2010.[6] T. Janssens, et al., “Implantation for an Excellent Definition of Doping Profiles in Si Solar Cells”, 25th EUPVSEC, Valencia, Spain, 2010.。