Gallium Phosphide photoanode coated with TiO2 and CoOx for stable photoelectrochemical water oxidation

Gallium Phosphide (GaP) has a band gap of 2.26 eV and a valance band edge that is more negative than the water oxidation level. Hence, it may be a promising material for photoelectrochemical water splitting. However, one thing GaP has in common with other IIIV semiconductors is that it corrodes in photoelectrochemical reactions. Cobalt oxide (CoOx) is a chemically stable and highly active oxygen evolution reaction co-catalyst. In this study, we protected a GaP photoanode by using a 20 nm TiO2 as a protection layer and a 2 nm cobalt oxide co-catalyst layer, which were both deposited via atomic layer deposition (ALD). A GaP photoanode that was modified by CoOx exhibited much higher photocurrent, potential, and photon-to-current efficiency than a bare GaP photoanode under AM1.5G illumination. A photoanode that was coated with both TiO2 and CoOx layers was stable for over 24 h during constant reaction in 1 M NaOH (pH 13.7) solution under one sun illumination. Published by The Optical Society under the terms of the Creative Commons Attribution 4.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article’s title, journal citation, and DOI.


Introduction
The use of photoelectrochemcial (PEC) water splitting to harvest intermittent solar sources in the form of hydrogen is an attractive potential method to address the energy and environmental issues. Since Honda and Fujishima in 1972 demonstrated titanium dioxide (TiO 2 ) for PEC water splitting [1], extensive efforts have been devoted to the development of photoelectrodes with good stability and high solar-to-hydrogen efficiency. Metal oxides (e.g., TiO 2 , Fe 2 O 3 , BiVO 4 , and SrTiO 2 ) have been extensively studied but their wide band gap and sluggish charge transfer kineties typically limited their solar-to-hydrogen conversion efficiency [1][2][3][4]. III-V semiconductor materials have attracted for PEC water splitting due to their high-efficiency, optimal band gap, and excellent optical properties but are readily susceptible to corrosion in strongly acidic or basic aqueous solutions during PEC process [5][6][7][8][9][10][11]. Of III-V semiconductor materials, gallium phosphide (GaP) has an appropriate band gap of 2.26 eV for harvesting a large portion of solar energy and has a valence-band edge more negative than the water oxidation potential [5,12]. Although GaP-based photocathodes has been extensively studied [9,10,13,14], only relatively few studies demonstrated water oxidation using GaP photoanodes due to the challenges presented, which also include a requirement for high overpotential due to slow hole-transfer kinetics [5]. Recently, amorphous TiO 2 has been widely used to protect III-V photoelectrodes in combination with OER catalysts during water oxidation. For example, a leaky TiO 2 (~143 nm) coupled with Ni catalyst to stabilize the GaP photoanode and was achieved up to 6 h stability in 1M KOH [5]. In contrast, the earth-abundant CoO x is effective for both functions, enhancing the oxygen evolution and stabilizing photoanodes, because it possesses efficient hole-conductivity and high resistance to photocorrosion under highly alkaline media [15]. Many studies have demonstrated highly active CoO x as a co-catalyst for photoanodes. For instance, a silicon photoanode coated with ultra-thin CoO x via atomic layer deposition (ALD) exhibited high stability for 24 h in 1 M NaOH [16], and a BiVO 4 photoanode coated with 2 nm CoO x has shown stable photocurrent density in KOH solution [17]. By using amorphous TiO 2 as a protection layer and a layer made of highly active oxygen evolution reaction (OER) catalysts could further improve the stability of photoelectrodes under extreme pH conditions. Photocarriers is transferred to the solution via mid-gap states of a leaky amorphous TiO 2 layer and facilitated by the Cobalt Oxide (CoO x ) co-catalyst for water oxidation [18]. In this work, a 20 nm amorphous TiO 2 thin film was deposited as protection layer followed by ~2 nm cobalt oxide (CoO x ) as a co-catalyst both deposited via ALD on GaP surface. The GaP/TiO 2 /CoO x photoanode exhibited good stability for 24 hours in 1 M NaOH (pH 13.7) electrolyte under simulated one sun illumination. The GaP/TiO 2 /CoO x photoanode provides increased photocurrent density and improved onset potential compared to the reference GaP photoanode, ascribed to the high catalytic activity of cobalt oxide which is in agreement with previous reports [16,17,19].

Experiments
A GaP photoanode modified CoO x was fabricated as follows: firstly, a single-crystalline n-GaP substrate was immersed in HCl solution (50% v/v) for 45 s, and then washed with deionized water followed by drying with N 2 gas. The dried sample was then transferred to an ALD chamber. An amorphous layer of TiO 2 with ~20 nm thickness was deposited directly onto the GaP electrode at 150 °C using a home-built ALD reactor by using titanium isopropoxide (TTIP) and water as metal and oxygen precursors respectively [20]. TTIP was kept at room temperature (25 °C) while water was kept at 5 °C. The deposition temperature was maintained at 150 °C. The growth rate of ALD process was approximately 0.4 Å/cycle. Following this, ALD of CoO x layer (∼2 nm) was deposited onto the TiO 2 layer at 250 °C [21,22]. X-ray Photoelectron Spectroscopy (XPS) was performed on a Thermo monochromated aluminium K-alfa photoelectron spectrometer, using monochromic Al-Kα radiation (1486.7 eV). Low resolution survey scans were measured between 0 and 1300 eV and high-resolution scans were used to measure O 1s, Ti 2p and Co 2p spectra to determine oxidation and elemental concentration. All spectra were calibrated to adventitious carbon at 284.5 eV and data was analyzed using CasaXPS software. To produce simulated sunlight, a 200W Xe arc lamp was used as a light source with AM 1.5 G filter. A calibrated silicon solar cell was used to calibrate the illumination intensity. The PEC performances were conducted in a three-electrode system using the photoanode as a working electrode, Ag/AgCl reference electrode and platinum coil as a counter electrode in 1 M NaOH solution (pH 13.7). The linear sweep voltammetry measurements were conducted by using a potentiostat (Ivium CompactStat). The incident-photon-to-current conversion efficiency (IPCE) and the applied bias photon-to-current efficiency (ABPE) was measured using a published procedure (19).
The AFM images were acquired with a Veeco Dimension V Scanning Probe Microscope with tapping mode at atmospheric pressure with a Si cantilever with 10 nm of radius. Gas evolution measurements were conducted in a leak-tight glass reactor with a quartz window, in a three-electrode PEC cell configuration where the GaP/TiO 2 /CoO x electrode acted as the photoanode with a Pt coil counter electrode and a Ag / AgCl reference electrode in 1 M NaOH (pH 13.7) electrolyte. The reactor was connected to a high purity nitrogen gas circulation system to purge the reactor before illumination. The oxygen gas produced during the water splitting experiment was collected and quantified using SRI 310c gas chromatograph (GC) with a molecular sieve (5A) column and helium as the carrier gas.

Results and discussion
A schematic illustration of the GaP electrode modified by amorphous TiO 2 and cobalt oxide CoO x are shown in Fig. 1(a). XPS measurements on the GaP photoanode was used to confirm the deposition of TiO 2 and CoO x films, as shown in Figs. 1(b)-1(c). Figure 1(b) shows the Ti 2p ionisation with the Ti 2p 3/2 peak centered at 458.1 eV, corresponding to TiO 2 [23]. The Co 2p 3/2 spectrum was deconvoluted to five fitting peaks using the parameters given by Biesinger et al [24]. The principle peak at 780.0 eV matches well with that expected for a Co 3 O 4 /CoO layer [24]. It should also be noted that no Ga and no P were observable, suggesting the material is pinhole free. Atomic Force Microscopy (AFM) has been used to assess the surface morphology for the GaP/TiO 2 /CoO x photoanode. Figures 2(a)-2(c) shows AFM images of a planar view of the GaP/TiO 2 /CoO x photoanode surface and a 3D representation of the surface morphology. This measurement shows that the top CoO x layer is a smooth thin film. The root-mean-square (rms) surface roughness of the GaP/TiO 2 /CoO x photoanode has been measured at 0.2 nm confirming high quality of the top layer. The current-voltage ( -J V ) characteristics for both GaP/TiO 2 /CoO x and bare GaP photoanodes are shown in Fig. 3(a). The bare GaP photoanode here was used a reference and exhibited an onset potential of −1 V and a photocurrent density of 0.19 mA/cm 2 at 0.5 V versus reference electrode ( Ag / AgCl ). By comparison, GaP/TiO 2 /CoO x photoelectrode shows a great improvement in onset potential and photocurrent density, with the photocurrent was increased to around 0.788 mA/cm 2 at 0.5 V versus reference electrode Ag / AgCl in 1 M NaOH solution (pH 13.7) and the onset potential observed to be approximately −1.12 V. The improvement in photocurrent and onset potential could be attributed to enhance the charge transfer to semiconductor/electrolyte interface and the cobalt oxide activities. The PEC performance of the GaP photoanodes with respect to the applied potential was examined; the applied bias photon-to-current efficiency (ABPE) are shown in Fig. 3(b), with the peak ABPE values, measured at 0.5 V, for bare GaP and GaP/TiO 2 /CoO x photoanodes being 0.14% and 0.57% versus reference electrode ( Ag / AgCl ), respectively. The higher ABPE value for the GaP/TiO 2 /CoO x photoanode indicates the enhancement of hole collection and water oxidation likely due to the CoO x -based catalysts. Similar to NiO x , CoO x is a highly active OER catalyst in a strong alkaline solution due to its relatively high CB edge and hence excellent holetransfer property [15]. Moreover, the Ni and Co OER catalysts can result in oxyhydroxides on their surface, e.g. NiOOH and CoOOH, which can reduce the overpotential for OER activity [25,26]. To further investigate the wavelength-dependent photoelectrochemical contribution of GaP photoanode to the photocurrent, the incident-photon-to-current conversion efficiency (IPCE) was measured. As shown in Fig. 4(a), the maximum IPCE achieved at wavelength 400 nm for the GaP/TiO 2 /CoO x photoanode was 44.6%, which is higher than the value of the bare GaP photoanode 34.2%. Moreover, the IPCE up to 36% between (350-450 nm) for the GaP/TiO 2 /CoO x photoanode whereas the IPCE value of the bare GaP photoanode was 28.3% which indicates the improved efficiencies of carrier separation and collection while using TiO 2 protection layer and CoO x as a co-catalyst. The IPCE value decreases below 26.7% for GaP/TiO 2 /CoO x and 19.8% for bare GaP at the wavelength of 450 nm and further decreases towards longer wavelengths (> 500 nm). Marginal photoresponse was measured for the wavelength range from 500 to 550 nm while the absorption spectrum of the GaP/TiO 2 /CoO x extended to 550 nm. This is attributed to the indirect nature of GaP bandgap [9,13,27].
Oxygen production was evaluated for the GaP/TiO 2 /CoO x photoanode at zero V vs a reference electrode ( Ag / AgCl ) under AM 1.5G-simulated sunlight as shown in Fig. 4(b). The amount of gas produced during approximately 2 hours was collected and analysed using a gas chromatography system. The GaP/TiO 2 /CoO x photoanode showed increased oxygen evolution with time. An amount of approximately 0.15 mmol/cm 2 of oxygen was produced after 2 hours of continuous irradiation, indicating an efficient water oxidation process. The advantage of the CoO x -based catalyst can be clearly seen from the improved PEC water splitting stability and the improved -J V curve of the the GaP/TiO 2 /CoO x photoanode. The stability of the photoanode was studied by performing current density versus time ( -J t ) measurements. The photocurrent density of the GaP/TiO 2 /CoO x photoanode was evaluated for 24 h using chronoamperometry at a fixed potential of zero bias V vs ( Ag / AgCl ) in 1 M NaOH solution. As shown in Fig. 5, the photocurrent is stable for 24 hours under continuous illumination for GaP/TiO 2 /CoO x photoanode with marginal degradation. Clearly, the cobalt oxide catalyst along with the TiO 2 layer shows good stability because the TiO 2 layer has excellent hole-conducting and is chemical resistant under strongly basic electrolyte while cobalt oxides effectively transfer holes for water oxidation [15,28].
XPS analysis of the GaP/TiO 2 /CoO x photoanodes was used to probe the surface chemistry after the long stability test as shown in Figs. 5(b)-5(c). Before PEC analysis, only peaks for Co, and Ti were observed on the surface as shown in Figs. 1(b)-1(c). After PEC, the intensity of the Co 2p peaks are reduced (as evidenced by the low signal to noise ratio) but still visible therefore indicating the presence of the CoO x layer even after exposure to the harsh electrolyte. Again, similar to CoO x , the TiO 2 layer was still intact after 24 hours of exposure to the strongly basic solution as a Ti 2p signal was still seen with the Ti2p 3/2 peak at 457.9 eV, matching Ti in the 4 + oxidation state. Despite the CoO x and TiO 2 layers being present, a spectrum for P 2p was measured, this was most likely due to possible diffusion of P through the TiO 2 and CoO x layers or formation of pinholes after PEC experiment.

Conclusion
In conclusion, we have deposited a TiO 2 -protection layer along with an ultra-thin layer of cobalt oxides as an OER catalyst on a GaP photoanode by ALD. The TiO 2 /CoO x improved the activity and increased the stability of the GaP photoanode for 24 h in 1 M NaOH (pH 13.7) solution under the AM 1.5G simulated solar light irradiation. Moreover, the GaP/TiO 2 /CoO x photoanode provides increased photocurrent density, improved onset potential, and incident-photon-to-current conversion efficiency (IPCE) compared to a bare GaP photoanode due to the high catalytic activity and corrosion resistant of cobalt oxide.

Funding
This work is supported by King Abdulaziz City for Science and Technology (KACST), Riyadh, Saudi Arabia. The Future Compound Semiconductor Manufacturing Hub (CS Hub) funded by EPSRC grant EP/P006973/1.