A semiconductor photocatalyst splits water when an absorbed photon promotes an electron across the bandgap, and both resulting carriers survive long enough to reach the surface and drive redox chemistry there (the Fujishima–Honda mechanism):
Absorption: hν ≥ E_g (photon energy must exceed the bandgap)
Reduction: 2H⁺ + 2e⁻ → H₂ needs E(CB) more negative than 0 V vs NHE
Oxidation: 2H₂O + 4h⁺ → O₂ + 4H⁺ needs E(VB) more positive than +1.23 V vs NHE
The two redox lines are fixed thermodynamic targets (at pH 0): the conduction band must sit above the H⁺/H₂ level and the valence band must sit below the O₂/H₂O level — this "straddling" condition is why TiO₂ works for overall water splitting while WO₃, whose conduction band is too low in energy, cannot reduce H⁺ without a sacrificial electron acceptor.
- Wavelength — sets photon energy E = 1240/λ(nm) eV. Only photons with E ≥ Eg are absorbed; the rest pass through unreacted (grey flashes).
- Photocatalyst / custom band edges — sets Eg and the absolute position of the bands on the NHE scale, which decides whether the straddling condition holds.
- Defect recombination — the rate at which a wandering electron and hole annihilate at a trap site before reaching a reaction site, releasing the absorbed energy as heat instead of chemistry (the dominant loss channel in most real photocatalysts).
- Pt co-catalyst loading — metal islands on the surface trap electrons and lower the kinetic barrier for H⁺ reduction; more loading means carriers are captured faster, out-competing recombination.
Apparent quantum yield (AQY) here is the fraction of absorbed photons whose electron-hole pair survives to complete both half-reactions — the same figure of merit reported in the photocatalysis literature, typically only a few percent even for good catalysts because recombination is fast compared to interfacial charge transfer.