Description:
<jats:p> In order to economically generate renewable hydrogen fuel from solar energy using semiconductor-based devices, the U.S. Department of Energy Fuel Cells Technology Office has established technical targets of over 20% solar-to-hydrogen (STH) efficiency with several thousand hours of stability under operating conditions [1]. We have modeled attainable efficiencies of tandem absorbers that, for the first time, considered the absorption of sunlight by water [2]. We used this modeling to identify top and bottom semiconductor bandgap combinations that should be targeted to achieve maximal STH efficiency. </jats:p>
<jats:p>We had to employ several key solid-state technological advances to achieve STH efficiencies exceeding 16%. The first improvement was to increase the device current via a non-lattice-matched 1.2 eV InGaAs grown using the inverted metamorphic multijunction technique developed by NREL’s III-V photovoltaics group. The second modification was to add a thin n-GaInP<jats:sub>2</jats:sub> layer to p-GaInP<jats:sub>2</jats:sub> to generate a "buried junction", which increased the photocurrent onset or Voc of the device by several hundred mV and enabled 14% STH efficiency. Finally, we increased the top junction photon conversion efficiency by adding an AlInP "window layer", which is commonly used in solid-state PV devices to reduce surface recombination. Through the use of a collimating tube, we measured our devices outdoors under direct solar illumination and verified over 16% STH conversion efficiency. I will also briefly introduce pitfalls of common experimental procedures that can influence the accuracy of measured STH efficiencies, which can be exaggerated for mulitjunction absorbers. </jats:p>
<jats:p>The largest loss in our current system is reflection at the semiconductor/electrolyte interface, so I will address the photon management strategies we use to achieve greater parity between measured efficiency and the theoretical limit. Capturing a significant portion of the ~25% of photons lost to reflection at this interface should allow the realization of devices that exceed 20% STH efficiency. </jats:p>
<jats:p>[1] http://energy.gov/sites/prod/files/2015/06/f23/fcto_myrdd_production.pdf </jats:p>
<jats:p>[2] H. Döscher et al., Energy Environ. Sci. 7, 2951 (2016). </jats:p>