Buffer Layer Effects on Tandem InGaAs TPV Devices
Single junction indium gallium arsenide (InGaAs) based TPV devices have demonstrated efficiencies in excess of 20% at radiator temperatures of 1058 C. Modeling suggests that efficiency improvements in single bandgap devices should continue although they will eventually plateau. One approach for extending efficiencies beyond the single bandgap limit is to follow the technique taken in the solar cell field, namely tandem TPV cells. Tandem photovoltaic devices are traditionally composed of cells of decreasing bandgap, connected electrically and optically in series. The incident light impinges upon the highest bandgap first. This device acts as a sieve, absorbing the high-energy photons, while allowing the remainder to pass through to the underlying cell(s), and so on. Tandem devices reduce the energy lost to overexcitation as well as reducing the current density (Jsc). Reduced Jsc results in lower resistive losses and enables the use of thinner and lower doped lateral current conducting layers as well as a higher pitch grid design. Fabricating TPV tandem devices utilizing InGaAs for all of the component cells in a two cell tandem necessitates the inclusion of a buffer layer in-between the high bandgap device (In0.53 Ga0.47As - 0.74eV) and the low bandgap device (In0.66Ga0.34As - 0.63eV) to accommodate the approximately 1% lattice strain generated due to the change in InGaAs composition. To incorporate only a single buffer layer structure, we have investigated the use of the indium phosphide (InP) substrate as a superstrate. Thus the high-bandgap, lattice- matched device is deposited first, followed by the buffer structure and the low-bandgap cell. The near perfect transparency of the high bandgap (1.35eV) iron-doped InP permits the device to be oriented such that the light enters through the substrate. In this paper we examine the impact of the buffer layer on the underlying lattice-matched InGaAs device. 0.74eV InGaAs devices were produced in a variety of configurations both with and without buffer layers. All structures were characterized by reciprocal space x-ray diffraction to determine epilayer composition and residual strain. Electrical characterization of the devices was performed to examine the effect of the buffer on the device performance. The effect of the buffer structure depends upon where it is positioned. When near the emitter region, a 2.6x increase in dark current was measured, whereas no change in dark current was observed when it was near the base region.