Silicon photovoltaic (PV) is a promising solution for energy shortage and environmental pollution. We are experiencing an era when PV is exponentially increasing. Global cumulative installation had reached 380 GW in 2017. Among which, silicon-based PV productions share more than 90% market. Performance of the first two-generation commercial popular silicon solar cells - Al-BSF and PERC - are limited by metal/Si contacts, where interface defects significantly reduce the open-circuit voltage. In this context, full-area passivation concepts are proposed for c-Si solar cells, with expectation to enhance the efficiency via reducing carrier recombination loss at the contact regions. In this thesis, poly silicon on oxide (POLO) passivating contact is developed for high efficiency c-Si solar cells. We unveiled the working mechanisms of POLO cells and then optimized the device performance based on our conclusion. We use boiling nitric acid to oxidize c-Si surface, which is of significance to determine the POLO working mechanisms. Phosphorus and boron doped silicon films are deposited by plasma enhanced vapor deposition (PECVD) or low-pressure vapor deposition (LPCVD) followed by high temperature (>800°C) annealing. SiOx structural evolution process under different annealing temperature was observed and the corresponding effects on passivation have been elucidated. The carrier transport mechanisms in the POLO contact annealed at high temperature, e.g. 800°C  900°C, were explored. We unveil that carrier transport in POLO structure is a combination of tunneling and pinhole transport, but dominant at varied temperature regions. Phosphorus-doped n-type POLO contact is optimized by several parameters, such as doping concentration, film thickness, annealing temperature, film deposition temperature, film relaxation time during annealing process, etc. We successfully obtained minority carrier lifetime over 10ms and contact resistivity lower than 30 mΩ·cm2. Boron-doped p-type POLO contact is also optimized by changing the doping concentration and annealing temperature. Finally, further hydrogen passivation is applied to enhance p-type POLO contact passivation, achieving an iVoc>690 mV, J0 <5 fA/cm2 and contact resistivity 1.3 mΩ·cm2. With the optimized n-type and p-type POLO contacts, an efficiency over 18% is achieved on n-type c-Si solar cells with a flat front surface.

Renewable and clean energy is urgently needed to cope with the climate crisis. Photovoltaics (PV) has been the fastest growing technology in the clean energy market due to its low cost, and the abundance of solar energy. The capacity of silicon-based PV is rapidly expanding with evolving technologies. Passivating the solar cell’s electrical contacts is a widely accepted strategy for the PV industry to improve device power conversion efficiency (PCE). Polycrystalline silicon (Poly-Si) passivating contacts are one of the promising concepts in the emerging class of passivating contacts. In this dissertation, the passivation mechanism of Poly-Si passivating contacts is investigated. Moreover, the influence of dopant diffusion on the passivation quality is revealed. To address the side-effects of dopant diffusion, a thin buffer layer is inserted between the Poly-Si(p) layer and the $SiO_x$ layer. With such a buffer layer, the passivation of the Poly-Si passivating contact is improved, which in turn, enhances the device PCE. In addition to passivating contacts, this dissertation also explores carrier-selective contact of crystalline silicon (c-Si) and low work function metal – Li. Li is a very reactive metal which makes the fabrication process a challenge. To overcome such a challenge, the c-Si/ Li contact is fabricated by thermally decomposing stable $Li_3N$ powder instead of metal evaporation. The c-Si/Li contact shows an excellent electron-selective transport performance with a 0.39 eV energy barrier. Full-area Si/Li rear contact devices are fabricated, and >19% PCE and >80% fill factor are achieved. To accelerate the device optimization, a physical model embedded machine-learning approach is applied to transparent conductive oxide (TCO) materials optimization. In this work, empirical correlations between sputtering parameters and the deposited TCOs’ electrical properties are established. Then a Bayesian Parameter Estimation (BPE) algorithm is applied to learn the empirical model. With this BPE network, the TCOs’ electrical properties are successfully predicted with limited material characterizations. Thanks to the combination of BPE and a physical model network, the material optimization process is significantly accelerated. In summary, this dissertation explores different aspects to develop novel passivating and carrier-selective contacts for c-Si solar cells, and introduces an approach to accelerate the development processes.