Physical spread and technical upshift in the band gaps of visible-light photocatalytic bismuth oxyhalide solid solutions

Semiconductor band gap determines the wavelength range of the utilized light, and is a key factor for many optical applications, e.g., photocatalysis and optoelectronic devices. As a superior group of bismuth oxyhalide, binary BiOIxBr1-x solid solutions have the visible-light band gap (epsilon(g) = 2.8 similar to 1.8 eV) that is readily tunable by changing the halogen composition. However, there remains a long-standing problem that the epsilon(g)s of BiOIxBr1-x have never been exactly confirmed by experimental measurements (with an error less than or similar to 0.5 eV). Using density-functional-theory calculations, we find that all of the sampled random structures for BiOIxBr1-x solid solutions have readily accessible stabilities, and the random arrangement of halogen atoms can result in a large physical spread (similar to 0.8 eV) in electronic epsilon(g). Such large band-gap variation with halogen-atom arrangement is attributed to the sensitive dependence of valence-band-maximum level on the local halogen composition. In the optical absorption spectra, we find the remarkable upshift of the optical epsilon(g) by 0.5 eV when only a small absorption-edge threshold (e.g., 2 x 10(4) cm(-1)) is used, which is ascribed to the small amount of conduction-band-minimum states. To further reveal the finite optical absorption within band gap that is responsible for the absorption-edge threshold existing in experimental characterizations, we finally calculate the optical absorption spectra of surfaces and find it is the metallic surfaces induced by halogen deficiency that causes the finite in-gap absorption. Such mechanism derived from the two-dimensional defects (i.e., surfaces) guides us to find similar phenomena on a zero-dimensional defect (i.e., halogen vacancy). The band-gap upshift, together with the physical spread, can explain the band-gap uncertainty existing in experimental results