Nano Gold Rush: On the Origin of the Photocurrent Enhancement in Hematite Photoanodes Decorated with Gold Nanoparticles

Hematite (α-Fe₂O₃) photoanodes are widely studied as candidates for solar water splitting. Recent reports suggest that the photocurrent of thin film hematite photoanodes could be enhanced by using gold nanoparticles (Au NPs) that give rise to plasmonic light trapping close to the hematite/electrolyte interface. This work examines the effect of Au NPs on the optical, electrochemical (in the dark), and photoelectrochemical (under illumination) properties of thin film (20–115 nm thick) Ti-doped hematite photoanodes. Au NPs were obtained by annealing 2.5–15 nm thick Au layers which led to dewetting and formation of 15–150 nm Au NPs, respectively. Au NPs on glass substrates displayed broad and shallow plasmonic peaks in the visible range, commensurate with the size distribution of the Au NPs. Two photoanode configurations with Au NPs decorating the surface or embedded under the hematite films were examined. Photoanodes of the first configuration displayed smaller photocurrents compared to counterpart photoanodes without Au NPs, most likely due to wasted absorption by the Au NPs and scattering into inactive parts of the device. The Au NPs underwent a redox reaction (Au + 3OH^– ⇌ Au­(OH)₃ + 3e^–) that gave rise to spurious contribution to the current. In addition, they also reduced the onset potential of water oxidation by ∼200 mV due to an electrocatalytic effect. Photoanodes of the second configuration displayed considerable enhancement (up to 92%) in absorption with respect to counterpart photoanodes without Au NPs. The enhancement was broadband, as expected for Mie scattering. The plasmonic resonances in their absorption spectra were red-shifted to wavelengths above the hematite absorption edge (600 nm); therefore, no plasmonic peaks were observed in the photocurrent action (IPCE) spectra. Small (<18%) enhancement in the plateau photocurrent (at 1.53 V_RHE) was observed in some cases, whereas in other cases the photocurrent was smaller than that of counterpart photoanodes without Au NPs. The photocurrent enhancement was considerably higher close to the onset potential, reaching up to 124% at 1.23 V_RHE, but the photocurrents were quite low (<0.5 mA/cm²) at these low potentials. No obvious correlation was observed between the absorption and photocurrent gains in photoanodes with different hematite film thicknesses and Au NP sizes, except for 4 photoanodes (out of 14 specimens that were examined) with a hematite film thickness of 75 nm. Therefore, we conclude that the photocurrent enhancement observed in our hematite photoanodes with embedded Au NPs was largely due to electrochemical effects rather than optical ones. This study highlights the intricate nature of several effects, both optical and electrochemical, of the Au NPs on hematite photoanodes, which come together to produce different contributions to the photocurrent. Depending on the device structure, some effects may enhance the water photo-oxidation current but other ones may not, and some effects may even suppress it. Therefore, careful design and optimization must be carried out in order to take advantage of the beneficial effects and mitigate the deleterious ones.