The direct photoelectrochemical (PEC) splitting of water into hydrogen and oxygen remains a promising route for producing sustainable and renewable hydrogen fuel. PEC devices use both photocathodes for the hydrogen generation (p-type semiconductors) and photoanodes for the oxygen evolution (n-type semiconductors), respectively. From chemical stability concerns, n-type metal oxides and p-type non-metal oxides are the desirable photoelectrodes. Unfortunately, none of the current existing metal oxides have demonstrated high solar-to-hydrogen conversion efficiencies under visible light irradiation, due largely to the fact that most metal oxides have too large bandgaps to effectively absorb the main portion of the sunlight spectrum, i.e., visible light. It is essential to engineer the bandgaps of metal oxides for the applications of PEC water splitting. For p-type photocathodes, most considered materials such as CuInS₂, Cu(In,Ga)Se₂, CuZnSnS₄ (CZTS) and WSe₂ have bandgap not wide enough or have poor performance when their bandgaps are widened. In this talk, I will present our recent results on two new PEC materials, one metal oxide (barium bismuth niobate) and one non-metal oxide (copper barium tin sulfide). Through density-functional theory (DFT) and experimental synthesis, we demonstrate a new approach to effectively engineer the bandgap of barium bismuth niobate double perovskites (Ba₂Bi_1+xNb_1-xO₆), i.e, varying the composition. We find that the film with the composition of Ba₂Bi_1.4Nb_0.6O₆ forms single pure phase and exhibits a nearly direct bandgap of about 1.64 eV, which is small enough to absorb visible light at wavelengths below 756 nm. For copper barium tin sulfide (CBTS), our DFT calculations indicate that it exhibits much better defect tolerance than CZTS. Our preliminary results show a saturated photocurrent of ~11 mA cm^-2 were reached at -0.55 V vs RHE in a neutral electrolyte.