Protected ultrathin cuprous oxide film for photocatalysis: Excitation and relaxation dynamics

In the search for cost-effective ways to produce green hydrogen, harnessing the most abundant renewable energy source—sunlight—directly through photoelectrochemical water splitting is highly desirable. Cuprous oxide (${\mathrm{Cu}}_2\mathrm{O}$) has proven itself as a material with great potential for the hydrogen evolution reaction (HER) and stands out by its Earth abundance and low processing cost. In this photoelectron spectroscopy study we investigated the electron dynamics in a heterostructure comprised of a ${\mathrm{Cu}}_2\mathrm{O}$ surface oxide grown on Cu(111) underneath a hexagonal boron nitride ($h$-BN) film, which replaces a conventional capping layer for corrosion protection. Our results show that it can be a viable cuprous oxide-based photoelectrode material. The $h$-BN film stays intact during the oxidation, and the oxide layer forms an ordered structure with defined valence and conduction bands. The conduction band is at a suitable energy to drive the HER for water splitting, but the photoexcited electrons display short lifetimes. Electrons are mainly excited in the copper substrate and are then captured in long-lived defect states in the ${\mathrm{Cu}}_2\mathrm{O}$ layer. The water splitting efficiency of this photocathode may still be improved by reducing the defect density.

Figure 1

LEED images at (a) $120\mathrm{eV}$ and (b) $48\mathrm{eV}$ after the first preparation. (c) $h$-BN/${\mathrm{Cu}}_2\mathrm{O}$/Cu(111) ARPES, combined from angular scans in the $\overline{\mathrm{K}}$ and $\overline{\mathrm{M}}$ directions.

Figure 2

(a) Detail from a He IIα spectrum of $h$-BN/Cu(111) including the Fermi edge, the Shockley surface state, the He IIβ satellite of the $h$-BN $\mathit{\pi }$-band, and the onset of the Cu $d$ bands. (b) Same region after oxidation with logarithmic color scale for both images. The surface state vanishes and an additional band above the $d$ bands appears. (c) Gaussian fit with exponential background of the spectrum after oxidation integrated between $-0.7$ and $-1.2{\AA }^{-1}$, with the spectrum before oxidation for reference. The Gaussian peak assigned to the ${\mathrm{Cu}}_2\mathrm{O}$ valence band is centered at $E_B=1.28\pm 0.01\mathrm{eV}$ with the FWHM $0.48\pm 0.01\mathrm{eV}$.

Figure 3

Composite pump-only angle-resolved 2PPE and 3PPE spectra for bare Cu(111), $h$-BN/Cu(111), and $h$-BN/${\mathrm{Cu}}_2\mathrm{O}$/Cu(111). The spectra are individually normalized and the relevant features are labeled. The dashed lines denote $E_F+2h\nu$, and the white circle highlights the resonant transition between SS and IPS.

Figure 4

(a) Time-resolved 2PPE scan after the first preparation, $3\mathrm{nJ}$ pump energy, linear color scale normalized to the maximum intensity. The spectra at negative delays were used for background subtraction. (b) Background-subtracted spectra at different delays. (c) Background-subtracted detector image at $50\mathrm{fs}$ delay, showing the dispersion of defect state and conduction band.

Figure 5

Least-squares fits of electron dynamics at the conduction band energy, defect state energy and in between with the model described in the text.

Figure 6

Fit results for the relaxation time of hot electrons in the $h$-BN/${\mathrm{Cu}}_2\mathrm{O}$/Cu(111) system excited with $3$ and $20\mathrm{nJ}$ pump pulses. Fit results obtained with the same model on $h$-BN/Cu(111) and literature values for Cu(111) taken from [37] are included for comparison. The literature data are extrapolated to low energies.

Figure 7

Energy scheme of the heterostructure showing the different excitation and relaxation pathways.