Multi-Junction Solar Cells and Photovoltaic Power Converters: High-Efficiency Designs and Effects of Luminescent Coupling

Abstract

Multi-junction photovoltaic devices based on III-V semiconductors have applications in space power systems and terrestrial concentrating photovoltaics, as well as in power-over-fibre and optical power conversion systems. These devices have between two and twenty junctions arranged in tandem, connected in series with optically transparent tunnel diodes. In some cases, they may include as many as eight different materials, including ternary and quaternary alloys, and >100 epitaxial layers in total. A general method for simulating performance of these devices using drift-diffusion based device simulation tools is reviewed. This includes discussion of the geometry, discretization, and physical equations to be solved. A set of material parameters for some important materials is listed, and solutions are shown for an example of a lattice-matched four-junction GaInP / (In)AlGaAs / InGaAsN(Sb) / Ge solar cell including a dilute nitride based p-i-n junction with ∼ 0.9 eV band gap. A sample of this dilute nitride junction with a 650 nm absorber layer was grown by molecular beam epitaxy and was shown to have short-circuit current density of 15.1 mA/cm2, sufficient for use in the 4-junction structure, while transmitting sufficient light through to the bottom (germanium) junction. Open-circuit voltage was up to 0.186 V at 1-sun, increasing to 0.436 V under 1500 suns concentration. The device simulation methodology was extended to include effects of luminescent coupling and photon recycling. These effects are included by adding a term to the electron and hole continuity equations, and the resulting coupled system of equations is solved. No external iterative loop is required, as has been the case in other efforts to model these effects. A five-junction photonic power converter (PPC) is simulated and it is shown that the quantum efficiency of the device is significantly broadened through luminescent coupling. There is a 350 mV reduction in simulated open-circuit voltage (70 mV per junction) if luminescent coupling is neglected. This work was later extended to a 12-junction PPC device, where the simulation predicts a wavelength sensitivity of -1.1%/nm in the absence of luminescent coupling; this is reduced to -0.4%/nm when luminescent coupling is included in the calculation. The latter result, and the overall shape of the simulated quantum efficiency curve agree closely with experimental measurements. Finally, two specific applications of PPCs are demonstrated. The first is in a step-up DC-to-DC converter, where a linear regulator combined with a laser/PPC pair can convert a 3.3 V input (commonly available from a single lithium polymer battery cell) into 12 V. Unlike conventional switching boost converters, this ‘photonic boost converter’ is not a source of ripple. In testing, a >80 dB reduction in ripple was measured compared with an equivalent switching boost converter, limited only by input noise of the instrument.The second application is in a 60 kW, 650 V switching circuit such as might be found in a hybrid or electric vehicle drivetrain. These circuits need several isolated power supplies to power gate drivers for the IGBT or SiC MOSFET switching components. This isolation is commonly provided by a small transformer, which inherently has a parasitic capacitance between primary and secondary windings and creates a path for EMI currents to flow from the high-power components to the power supply and control circuitry. By using a laser/PPC pair to provide the needed isolation, this parasitic capacitance can be largely eliminated; a 20 dB reduction in EMI current reaching the control FPGA is demonstrated.