Progress in Low-Cost Gallium Arsenide Solar Cells

Citation

Jahelka, Phillip Robert (2022) Progress in Low-Cost Gallium Arsenide Solar Cells. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/btbp-6h76. https://resolver.caltech.edu/CaltechTHESIS:05242022-235428954

Abstract

In order to prevent disastrous global warming the manufacturing capacity of renewable energy power sources must grow rapidly. Solar photovoltaics will likely be one of humanity's main sources of energy in the future due to the enormous available resource but increasing the manufacturing capacity of solar panels is hamstrung by both the limited profit margins of the highly-competitive renewable energy market and the enormous capital cost of building the factories that convert sand into semiconductor-grade silicon. Gallium arsenide is a material that can potentially help with the capital bottleneck because it absorbs light much more strongly than silicon and so the capital cost per unit weight of making the semiconductor can be spread over a larger number of devices and therefore effectively reduced. We present a number of results aimed at enabling low capital-cost GaAs solar cell manufacturing. First is a technique for open-tube, vapor phase zinc diffusion in GaAs. This method is dramatically simpler than its historical counterparts. Second, we use this technique to fabricate solar cells with Voc's greater than 960 mV and uncertified efficiencies over 23%, large improvements over the state of the art. We further demonstrate a base-metal, air-tolerant ohmic contact to n-type GaAs which is an improvement over traditional contacts that require noble metals and inert atmospheres. We also found the existence of melt-grown n-type GaAs with minority carrier diffusion length comparable to vapor grown material which helps with the economic viability of these devices. We also performed a technoeconomic analysis on our proposed devices and find that they satisfy the desired properties of both the capital and electricity being cheaper than silicon solar cells. We also demonstrate the first n-on-p diffused junction GaAs solar cells.

As a parallel path to low capital intensity GaAs solar cells we also investigated non-epitaxial heterojunction devices. In the course of this work we both developed and characterized passivation chemistries for GaAs. Results in include the first use of a carbene and dithiothreitol for GaAs passivation and achieving surface recombination velocities comparable to GaInP passivation. With passivated organic heterojunction solar cells we were able to achieve a Voc of 840 mV which is a record for this class of devices, but its unclear how to improve the result to make them competitive with diffused junctions.

We also explored nanowire solar cells as an alternative strategy to reducing material usage by exploiting their strong light-absorption. We developed a computational model for a non-epitaxial GaAs heterojunction nanowire solar cell and predict an optimized efficiency over 30%. Towards fabrication we used metal-assisted-chemical-etching to make nanowire arrays and found we were able to cleanly cleave the nanowires embedded in a polymer from a 110 oriented wafer.

We also share some preliminary work on using total internal reflection in a solar cell encapsulant to mitigate shading loss due to the contacts on the front of a solar cell. We developed a computational model arguing that these structures could increase energy yield by 8% and demonstrated proof-of-principle experiments.

Finally, we share work on designing solar cells for operation on Venus. We developed models for the optical properties and recombination that correctly model the temperature dependence of a reference solar cell and using that model predict that a GaInP single-junction solar cell is a good solar-cell design for general usage in the atmosphere.

Thesis Files