Ternary Oxides of $s$- and $p$-Block Metals for Photocatalytic Solar-to-Hydrogen Conversion

Oxides containing metals or metalloids from the $p$ block of the periodic table (e.g., $\mathrm{In}$, $\mathrm{Sn}$, $\mathrm{Sb}$, $\mathrm{Pb}$, and $\mathrm{Bi}$) are of technological interest as transparent conductors and light absorbers for solar-energy conversion due to the tunability of their electronic conductivity and optical absorption. Comparatively, these oxides have found limited applications in hydrogen photoelectrolysis, primarily due to their high electronegativity, which impedes electron transfer for reducing protons into hydrogen. We have shown recently that inserting $s$-block cations into $p$-block metal oxides is effective at lowering electronegativities while affording further control of band gaps. Here, we explain the origins of this dual tunability by demonstrating the mediator role of $s$-block cations in modulating orbital hybridization while not contributing to frontier electronic states. From this result, we carry out a comprehensive computational study of 109 ternary oxides of $s$- and $p$-block metal elements as candidate photocatalysts for solar hydrogen generation. We down-select the most desirable materials using band gaps and band edges obtained from Hubbard-corrected density-functional theory, with Hubbard parameters computed entirely from first principles, evaluate the stability of these oxides in aqueous conditions, and characterize experimentally four of the remaining materials, synthesized with high phase uniformity, to validate and further develop the computational models. We thus propose nine oxide semiconductors, including $\mathrm{Cs}\mathrm{In}$${}_3$$\mathrm{O}$${}_5$, $\mathrm{Sr}$${}_2$$\mathrm{In}$${}_2$$\mathrm{O}$${}_5$, and $\mathrm{KSbO}$${}_2$, which, to the extent of our literature review, have not been previously considered as water-splitting photocatalysts.

Popular Summary

Oxides that contain s- and p-block elements offer tunable electronic conductivity and optical absorption, which is promising for photocatalytic hydrogen generation. Here, the authors explore this dual tunability by combining materials modeling, computational design, experimental synthesis, and electrochemical characterization. As a result, they discover a versatile class of photocatalytic semiconductors that incorporate electropositive elements into photoactive oxides. Varying the incorporated electropositive elements enables simultaneous tuning of redox activity, solar absorption, and aqueous stability, thereby opening a route toward corrosion-resistant semiconductors that can produce carbon-neutral hydrogen under solar illumination.

Figure 1

Inserting $s$-block cations into $p$-block metal oxides enables band realignment. (a) An experimental band alignment with respect to the redox potentials of water for representative $p$-block metal oxides ($\mathrm{In}$${}_2$$\mathrm{O}$${}_3$ [7], $\mathrm{Sn}\mathrm{O}$${}_2$ [7], $\mathrm{Sb}$${}_2$$\mathrm{O}$${}_3$ [8], and $\mathrm{Bi}$${}_2$$\mathrm{O}$${}_3$ [9]) with the mid-band-gap potential (approximating the flat-band potential) shown as a horizontal line. As indicated by the open arrow, the band edges of an oxide tend to shift toward less reducing electrode potentials as the electronegativity of its cations increases. (b) An experimental band alignment for $\mathrm{In}$${}_2$$\mathrm{O}$${}_3$ [7] compared to that of $\mathrm{In}$${}_2$$\mathrm{O}$${}_3$-based compounds with the addition of $s$-block metal cations ($\mathrm{Na}\mathrm{In}\mathrm{O}$${}_2$, $\mathrm{Ba}\mathrm{In}$${}_2$$\mathrm{O}$${}_4$, and $\mathrm{Sr}\mathrm{In}$${}_2$$\mathrm{O}$${}_4$ [6]) that cause the band edges to shift toward more reducing potentials.

Figure 2

Inserting $s$-block cations into $p$-block metal oxides enables band-gap tuning. (a) Experimental and numerical band gaps (open and filled symbols, respectively) of stannate-based ternary oxides with perovskite [13], pyrochlore [14], and spinel [6, 15] structures. The band gaps are calculated using generalized-gradient density functionals, with scissors corrections of 2 eV ($A$$\mathrm{Sn}\mathrm{O}$${}_3$), 1.7 eV ($A$$\mathrm{In}$${}_2$$\mathrm{O}$${}_4$), and 2.3 eV ($A_2$$\mathrm{Sb}$${}_2$$\mathrm{O}$${}_7$). (b) Octahedra of the perovskite $A$$\mathrm{Sn}\mathrm{O}$${}_3$ materials rotate away from the vertical axis as the radius of the $s$-block metal cation $A$ decreases, creating the distortions visible in the atomic representations. (c) Electronic structures of the highest occupied and lowest unoccupied bands in $\mathrm{Ca}\mathrm{Sn}\mathrm{O}$${}_3$ (top) and $\mathrm{Ba}\mathrm{Sn}\mathrm{O}$${}_3$ (bottom). The black and orange lines are energy bands calculated from the Kohn-Sham Hamiltonian and by Wannier interpolation [16, 17], respectively. The decomposition of Bloch electrons into maximally localized Wannier orbitals at high-symmetry wave vectors—$\mathrm{\Gamma }=(0,0,0)$ and $S=(\frac{1}{2},\frac{1}{2},0)$—shows that the top valence bands are composed of oxygen $2p$ orbitals, whereas the lowest conduction band is a mixture of oxygen $2p$ orbitals and $A_{\text{1g}}$ orbitals centered around stannate cations. The centers of the Wannier orbitals are displaced away from the central $s$-block cation for ease of visualization.

Figure 3

The down-selection of the 109 ternary oxides containing $s$- and $p$-block metals using band gaps (${\epsilon }_{\mathrm{g}}$) and band edges ($E_{\mathrm{VB}}$ and $E_{\mathrm{CB}}$) predicted at the DFT (step 1) and DFT+$U$ (step 2) levels, and Pourbaix free energies of decomposition under electrochemical conditions in aqueous environments.

Figure 4

(a) A comparison of the computationally determined (within DFT+$U$) and experimentally measured conduction- and valence-band edges for the $s$- and $p$-block metal oxides that were synthesized and the phases of which were confirmed to be uniform. (b) Photoelectrochemical characterization of the synthesized oxides via voltage-dependent photocurrent measurements under chopped illumination.

Figure 5

The dependence of the free energy of electrochemical decomposition $\mathrm{\Delta }{G}_{\mathrm{redox}}$, calculated at $p$$\mathrm{H}$ 7 and at the flat-band potential of the oxide, as a function of its band gap ${\epsilon }_{\mathrm{g}}$, showing superior aqueous stability for wider energy gaps.