A wide variety of semiconductor systems have proven capable of harnessing near-ultraviolet and visible light and using that energy to drive the reduction of carbon dioxide through photocatalysis or photoelectrochemistry. The improved production of solar fuels is an essential component of the renewable energy portfolio of the future, enabling intermittent power sources to become more prevalent by storing excess energy with fewer losses and thus greater recoverability.

Whereas early research in this field employed suspensions of simple semiconductor materials in aqueous solution, recent developments have focused on various doping and nanostructuring techniques, combinations of multiple semiconductors, and the addition of heterogeneous or homogeneous catalysts and sensitizers. The use of nanomaterials in particular has led to improved charge separation and to enhanced catalysis. Metals, both directly in contact with the light absorber and in a separate but electrically connected electrolysis cell, serve as electron sinks and catalysts, and they sometimes even enable CO₂ reduction on semiconductors whose own surfaces have excessively high overpotentials.

Many of the semiconductors described in this review are oxides, with TiO₂ receiving the bulk of scrutiny in this regard. Often these materials have band gaps approaching 3 eV; this hinders the absorption of a large proportion of the solar spectrum available at the Earth’s surface and diminishes the insolation to current efficiency. The efficiency of smaller band gap (∼1.3 eV) semiconductors is still capped at the Shockley–Queiesser limit of ∼30%. Therefore, multijunction cells may be necessary to capture the majority of the solar spectrum, overcome the SQ limit, maintain long-term stability under illumination, and still straddle the CO₂ reduction and H₂O oxidation potentials. Although some efforts have already been made to find working combinations of semiconductors, much of the domain of pairs (or even multiples) has yet to be explored.

Some of the innovations recently taken up in photocatalysis, such as photosensitization, were first studied for use in photovoltaics. Another such innovation, the use of organic semiconductors (OSCs), has, to our knowledge, not yet been employed in CO₂ reduction catalysis. Although organic photovoltaics are still less efficient than silicon, the use of these semiconductors in catalysis may prove to be particularly interesting due to their mutability. The vast scope of synthetic methods available permits tuning of the electronics, but functional groups could also be appended to bind and reduce CO₂ more effectively.

As diverse as these studies have been, the vast majority of CO₂-reduced products are single-carbon species such as methane, methanol, carbon monoxide, and formic acid. Some multicarbon products, primarily light hydrocarbons, have been reported to form, although they are generated almost exclusively at metallic copper and typically in low yields. Whereas the established Fischer–Tropsch process is capable of taking solar-generated CO and H₂ to liquids, an important advance in catalysis would be the development of materials that can make carbon–carbon bonds and, thus, higher order hydrocarbons and alcohols at ambient conditions directly from CO₂, water, and light.