Solar-to-Hydrogen Production on Multi-Band Photoelectrodes: Surpassing the Current Matching Requirements of Conventional Tandem Devices
The Si/GaN/p-InGaN PV-PEC device heterostructure consists of a planar n+-p Si solar cell wafer, ~ 150 nm n-GaN and ~ 600 nm p-InGaN nanowire segments along the axial direction. The top p-InGaN nanowire arrays are designed to absorb the ultraviolet and a large portion of the visible solar spectrum. The remaining photons with wavelengths up to 1.1 µm are absorbed by the underlying planar Si p-n junction. In contrary to conventional semiconductor photocatalysts, InGaN can uniquely straddle water oxidation and hydrogen reduction potentials under deep visible light irradiation. The n-GaN and p-InGaN are connected via an n++-GaN/InGaN/p++-GaN polarization-enhanced tunnel junction, which enables the transport of photo-excited holes from the p-InGaN to the n-GaN within each single nanowire. The presented device differs from conventional PV-PEC electrodes in that both the top p-InGaN and the bottom GaN/Si light absorbers can simultaneously drive proton reduction due to the lateral carrier extraction scheme of nanowires. That is, due to the relatively small offset between the n+-Si and n-GaN conduction band edges, photo-excited electrons of the bottom Si solar cell can readily inject into the n-GaN nanowire segment. A large fraction of the injected electrons can drive proton reduction on GaN surfaces, with the rest recombining with holes from the p-InGaN in the tunnel junction. It is seen that such a novel design, with the use of Si/GaN-nanowire as the bottom light absorber, can surpass the restriction of current matching in conventional dual absorber devices and simultaneously provide energetic photo-excited electrons to the hydrogen-evolution-reaction (HER) catalyst. Compared to the conventional PEC-PV, such adaptive junction can reduce chemical loss by allowing charge carriers with different overpotentials to be utilized for HER. Such a monolithically integrated photocathode shows an applied bias photon-to-current efficiency of 8.7% at a potential of 0.33 V vs. normal hydrogen electrode. The Faradaic efficiency for hydrogen generation is also measured to be nearly unity.