About 400 semiconductor solids are known to have photocatalytic activity for water splitting. Yet there is no single material that could satisfy all the requirements for desired photocatalysts: i) suitable band gap energy (1.7 eV< Eg < 2.3 eV) for high efficiency, ii) proper band position for reduction and/or oxidation of water, iii) long-term stability in aqueous solutions, iv) low cost, v) high crystallinity, and vi) high conductivity. Hence, in the selection of photocatalytic materials, we better start from intrinsically stable materials made of earth-abundant elements. Upon selection of the candidate materials, we can also modify the materials for full utilization their potentials. The main path of efficiency loss in photoelectrochemical water splitting process is recombination of photoelectrons and holes. We discuss the material designs to minimize the e^- - h⁺ recombination including; i) heterojunction photoanodes for effective charge separation, ii) band engineering to extend the range of light absorption, iii) metal or anion doping to improve conductivity of the semiconductor and, iv) one-dimensional nanomaterials to secure a short hole diffusion distance and vectoral electron transfer, and v) loading co-catalysts for facile charge separation. Finally, we need to construct a stand-alone solar fuel production system by combining with a solar cell in tandem, which provides bias voltage needed for the photolytic reactions making possible the fuel production only with solar energy without any external energy supply.