Advanced Device Design for Photoelectrochemical Water Splitting Derived By a Detailed Balance Approach

Abstract

Solar fuel generation by direct photoelectrolysis of water is a pathway towards energy storage and transportation fuel demands in a sustainable energy economy. High band gap semiconductor materials may drive the water splitting reaction, but cannot use substantial parts of the solar spectrum, while common photovoltaic (PV) absorbers fail to provide sufficient threshold voltage for photoelectrochemical (PEC) conversion. Tandem absorber structures overcome this trade-off featuring promising efficiencies [1]. Tandem cell implementation by epitaxial III-V growth has enabled landmark solar-to-hydrogen (STH) conversion efficiencies [2]. In contrast to the maximum power metric in PV, efficient STH conversion requires maximizing the photocurrent versus a load voltage consisting of the thermodynamic potential of water splitting (1.23V) and various voltage loss mechanisms accumulating throughout the photoelectrochemical cell, often summarized in a generalized overvoltage term for modelling approaches. We discovered that sunlight absorption during illumination through the aqueous PEC electrolyte flawed both efficiency and band gap predictions in earlier theoretical studies and derived PEC tandem device guidelines defining an interdependence between allowable overvoltage loss and electrolyte transmission length [1]. Even though optimum band gap combinations for STH conversion have been discussed for decades, state-of-the-art inverted metamorphic multijunction (IMM) III-V growth and processing strategies developed in the context of high-efficiency PV [3] finally enabled us to tailor water splitting device structures according to maximum performance limits [Fig. 1] and validate our prediction. A first step was to relax the harsh current limitation imposed by the bottom absorber of the classical GaInP/GaAs PEC tandem [2] by replacing it with a metamorphic GaInAs with a band gap of approximately 1.2 eV [Fig. 1(a)]. Further reductions in overvoltage loss and electrolyte layer thickness enable further reductions of top and bottom junctions to unlock higher STH limit efficiency [Fig. 1(b)] that could be implemented by the integration of a 1.7 eV absorber on Si. Advanced PEC measurement protocols [4] drastically reduce performance estimation errors induced by the spectral mismatch of laboratory light sources and light transmission through epoxy used for device area confinement. Valid PEC results of our tailored devices as well as of analogue PV device characterization provide us with important feedback for our detailed balance model, where we start to unravel overvoltage loss into its fundamental components to obtain advanced guidance for future technical development.