Core-Shell Si-Ta3N5 Heterostructures As Photoanodes for Photoelectrochemical Water Splitting

Abstract

Tantalum nitride (Ta3N5­) is a promising photoanode material for photoelectrochemical water splitting due to its n-type doping, bandgap of 2.1 eV, and band positions that straddle the redox potentials of both the hydrogen and oxygen evolution reactions.1 However, there are other challenges associated with Ta3N5 including poor charge transport properties, late photocurrent onset and photodegradation. To address all of these challenges, we fabricated efficient and stable core-shell heterostructured Si-Ta3N5 nanowire photoanodes coated with CoTiOx and NiO oxygen evolution co-catalysts. For metal oxide semiconductors with charge transport limitations such as short charge carrier diffusion lengths and/or lifetimes, previous device architecture strategies have included nanostructuring the oxide or coating it onto a conductive, high-aspect ratio scaffold.2, 3 In this work, a similar approach was applied to Ta3N5 to make a nanostructured core-shell Si-Ta3N5 device where the Si is a conductive, nanowire scaffold and the Ta3N5 is coated onto Si as a thin shell through which charge extraction should be efficient. Furthermore, if n-type Si is used, then a heterojunction between Si and Ta3N5 is formed enabling a tandem, dual-absorber configuration. This results in a 200 mV cathodic shift for photocurrent onset due to the photovoltage gained from Si and therefore improves the overall performance of the Si-Ta3N5 photoanode. The Si nanostructures were fabricated by a pseudo-Bosch deep reactive ion etching process,4 then tantalum oxide was coated onto Si by atomic layer deposition (ALD), and finally the tantalum oxide was nitrided in pure ammonia gas at 900 °C to yield crystalline Ta3N5. To gain a deeper understanding of charge transport within the thin Ta3N5 shells and to optimize the electrode performance, we varied the thickness of the shell between 10 and 70 nm and tested for photoelectrochemical ferrocyanide oxidation in a reversible ferri/ferrocyanide redox couple. We found that in a core-shell Si-Ta3N5 nanowire device, where light absorption should not vary as a function Ta3N5 thickness, photocurrent decreased with increasing Ta3N5 thickness, which indicates that charge extraction becomes problematic with thicker Ta3N5 films. As a clearer demonstration of this charge transport issue, we performed absorbed photon to current efficiency (A.P.C.E.) measurements using 10-70 nm films of Ta3N5 on planar Si and quartz substrates. Again, we found that A.P.C.E. decreased once the films were thicker than 30 nm, demonstrating that the minority carrier diffusion length is likely tens of nanometers. Finally, to address the problem of Ta3N5 photodegradation and to demonstrate photoelectrochemical performance for water oxidation with the core-shell system, CoTiOx and NiO catalysts were coated onto Si-Ta3N5. We found that CoTiOx, deposited by dipcoating from a sol-gel,5 pairs well with Ta3N5 as a cocatalyst and yields a photocurrent density of 2.6 mA/cm2 at 1.23 V vs. RHE. However, the device is not stable. With NiO deposited by ALD,6 we were able to achieve a photocurrent density of 2.4 mA/cm2 after Fe incorporation into NiO. A bare Si-Ta3N5 yields 0.7 mA/cm2. Furthermore, the Si-Ta3N5 coated with only ~1 nm of Ni(Fe)O loses 40-50% of its peak photocurrent over a 24 hour chronoamperometry test. Compared to the current Ta3N5 reports in the literature, the Ni(Fe)O catalyst affords noteworthy stability.7, 8 Therefore, in this work, we demonstrate a heterostructured Ta3N5 system with a core-shell Si-Ta3N5 dual absorber configuration that enabled earlier onset of photocurrent, efficient charge extraction from Ta3N5 and relatively stable long term water oxidation with a Ni(Fe)O co-catalyst.