Suppression of Phase Separation in AlGaInAs Compositionally Graded Buffers for 1550 nm Photovoltaic Converters on GaAs

TL;DR

This study develops methods to suppress phase separation in AlGaInAs buffers, improving the quality and efficiency of 1550 nm photovoltaic devices on GaAs, with implications for scalable laser and power conversion technologies.

Contribution

It introduces optimized epitaxial growth strategies, including substrate misorientation and doping choices, to reduce defects and enhance device performance in AlGaInAs-based photovoltaic converters.

Findings

Suppression of phase separation via substrate misorientation and Zn-doping.

Achievement of low TDD (3.5 x 10^6 cm^-2) in AlGaInAs buffers.

High efficiency (31.9%) laser power conversion demonstrated.

Abstract

We investigate strategies to suppress phase separation and reduce threading dislocation densities (TDD) in AlGaInAs compositionally graded buffers (CGBs) that span the lattice constant range from GaAs to InP. Combining results from high resolution x-ray diffraction, cathodoluminescence, transmission electron microscopy, and photovoltaic device measurements, we correlate choices of epitaxial growth conditions with the defect structure of the CGBs and subsequent device performance. Both the use of substrates with high misorientation off (100) towards the (111)A plane and Zn-doping instead of Si-doping are shown to suppress phase separation and reduce TDD. We demonstrate a 0.74 eV GaInAs device grown on a (411)A GaAs substrate using a Zn-doped AlGaInAs CGB with TDD = 3.5 +/- 0.2 x 106 cm^-2 that has a bandgap-open circuit voltage offset of only 0.434 V measured under the AM1.5G solar spectrum. We characterized this device under high-intensity irradiance from a 1570 nm laser and measured a 31.9% peak efficiency laser power conversion efficiency at 3.6 W/cm2. These results provide a roadmap to the manufacture of laser- and thermal-power conversion devices with the performance and cost-effectiveness needed to drive adoption of these technologies at scale.