The generation of chemical fuels (e.g., hydrogen, hydrocarbons) using sunlight offers a carbon-neutral route to meet the ever-increasing world’s energy needs. In recent years, significant progress has been reported, especially in solar water splitting. Compared to indirectly coupling photovoltaic cells with water electrolyzers, direct photoelectrochemical (PEC) water splitting offers potential advantages in terms of system and thermal integration. Nevertheless, the demonstrated solar-to-hydrogen (STH) efficiencies are still insufficient to obtain technological competitiveness. Efforts in increasing efficiency are, unfortunately, faced with a classic material science dilemma: high STH efficiencies (~20%) are achieved in devices employing high quality semiconductors that are cost-prohibitive,[1-2] while devices based on low-cost metal oxide semiconductors have only demonstrated modest STH efficiencies (<10%).[3-4] These metal oxide-based PEC water splitting devices typically utilize BiVO₄ as the absorber, and further increase of the efficiency is hindered by the optical absorption limit of BiVO₄ that has a ~2.4 eV bandgap. Novel metal oxide semiconductors with a bandgap of 1.7-1.9 eV, which are stable and efficient, are therefore desired. In this talk, our recent efforts in developing two complex metal oxides will be presented. The first metal oxide is alpha-SnWO₄ photoanode with a bandgap of ~1.9 eV. We deposited alpha-SnWO₄ thin films using pulsed laser deposition and investigated the pH stability window using a combination of inductively coupled plasma optical emission spectroscopy, x-ray photoelectron spectroscopy and in situ spectro(photo)electrochemistry measurements.[5] We found that photocorrosion occurs in alkaline pH electrolytes, while at acidic to neutral pH, a self-passivating oxide layer is formed on the surface of alpha-SnWO₄ that blocks the transfer of photogenerated holes to the electrolyte. The latter could be overcome by depositing NiO_x as a protection layer, but it is also accompanied by the reduction of the photovoltage. A thorough interface investigation using synchrotron-based hard x-ray photoelectron spectroscopy (HAXPES) and surface photovoltage spectroscopy (SPV) reveals the presence of an interfacial SnO₂ layer,[6] which must be avoided to overcome the photovoltage limitation. The second metal oxide is modified BaSnO₃. While pristine BaSnO₃ has a large bandgap (> 3 eV), the introduction of defects, such as oxygen vacancy and sulfur substitution, extends the optical absorption onset to ~1.7 eV and increases the charge carrier mobility and photocurrent by a factor of ~20.[7] By combining analyses from photoluminescence spectroscopy, modulated SPV, and wavelength-dependent time-resolved microwave conductivity, these defects can be attributed to the generation of delocalized mid-bandgap states through which charge transport via a long-lived carrier-hopping mechanism is enabled. Further outlook on these two promising complex metal oxides, alpha-SnWO₄ and modified BaSnO₃, will be discussed.