| Crystalline silicon cells dominate the photovoltaic (PV) market today with ~90% market share. However, a majority of Si cells use the screen-printed, full Al-BSF structure which cannot achieve ≥ 20% efficiency using conventional screen-printing technology. This provided the motivation to produce ≥ 20% efficient Si solar cells in this thesis. The approach that was adopted in this research involved first fabricating and analyzing a high-efficiency, full Al-BSF cell, and then developing modifications to that cell that would lead to efficiencies ≥ 20%. Issues of commercial viability and throughput were also kept in mind during the development of the modified structure and process.;In the first phase of this thesis, a ~19% efficient, 300 μm thick, screen-printed Al- BSF cell (4 cm2) was fabricated and analyzed to determine and quantify its major efficiency-limiting loss mechanisms.;The cell concept that is pursued in this thesis is the boron back surface field (B-BSF) structure. For meeting the BSRV goal of 200 cm/s, a B-BSF is a technically attractive alternative to a conventional Al-BSF as: 1) a B-BSF can provide stronger field-effect passivation due to boron having higher solubility in Si than aluminum, and 2) the surface of a B-BSF can be more easily passivated with a dielectric. Furthermore, a B-BSF can be capped with a highly reflective material (or a dielectric/reflector stack) allowing the RB target of ~95% to be achieved.;Another goal of this thesis is to develop a fabrication process that can deliver a ≥ 20% efficient, full-area B-BSF cell while: 1) ensuring that the cycle time of each process step is no longer than the longest step used for the baseline full Al-BSF cell (90 minutes), and 2) using low-cost, screen-printing technology for contact formation.;The first step towards realizing a 20% B-BSF cell was to pick a boron diffusion source. In this thesis, dilute spin-on solutions of boric acid in de-ionized (DI) water were investigated as a novel, low-cost and non-toxic alternative to more conventional boron diffusion sources like boron tribromide (BBr3) which are toxic and pyrophoric. It was found that boron emitters with a wide range sheet resistances (~20 – 200 Ω/sq.) could be achieved with very dilute boric acid sources (~0.5-2 wt.% boric acid in DI water) by controlling the diffusion time and temperature. In addition, uniform diffusion was obtained on both planar and pyramid textured surfaces using these spin-on sources. The stability of the spin-on boric acid sources were also tested.;The next phase of the research involved selecting a diffusion condition/doping profile that can achieve the target BSRV of 200 cm/s and thereby deliver a 20% efficient B-BSF cell. The simplest B-BSF structure is one with full-area boron diffusion and full-area rear metallization because this structure requires just one additional step (boron diffusion) over the baseline Al-BSF process. Finding a dielectric that: 1) provides high-quality passivation of a p+ surface in a process that lasts . 90 minutes, and 2) is stable through a high-temperature (700-800°C) screen-printed contact firing cycle, became the next focus of this thesis.;In the passivation studies, several different dielectrics . thermal SiO2, low-frequency plasma-enhanced chemical vapor deposited (LF-PECVD) SiNX, thermal SiO2/PECVD SiNX stacks, atomic layer deposited (ALD) Al2O3, a commercially available spin-on SiO2, and a commercially available Al-doped spin-on glass (SOG) . were examined.;The next phase of the thesis dealt with finding a back surface reflector (BSR) material that has RB = 95% and diffusivity (â) = 100%. Since the passivation studies established that the rear contacts needed to be point contacts, the BSR also needs to be electrically conductive so that it can electrically interconnect the rear point contacts. Several materials were tested in this phase of the thesis – evaporated Ag, evaporated Al, a screenprinted Ag paste and a Ag colloid that is deposited with a brush. While none of these four materials met both the RB and the â targets simultaneously, the Ag colloid came the closest.;The final phase of the thesis involved integrating the results of the modeling, passivation, lifetime and BSR studies into a cell fabrication sequence. This was successfully completed and a confirmed efficiency of 20.2% was achieved on a 4cm2, passivated B-BSF cell on FZ Si. However, device modeling showed that if this cell structure and process were transferred to commercial p-type Cz Si material, the lightinduced degradation (LID) effect that affects commercial p-type Cz material could reduce the efficiency to ~19%. Via both device modeling and device theory this large drop in cell efficiency due to LID was shown, to be linked to the well-passivated surfaces of the B-BSF cell. (Abstract shortened by UMI.). |
