With growing concerns on future energy supplies, solar energy appears as an energy source whose potential remains to be tapped at a large scale. In the last two decades, dye-sensitized solar cells (DSCs) have been considered as a competitive means to convert solar energy into electricity thanks to their low manufacturing and environmental costs, simple fabrication and high efficiency. Indeed, record efficiencies as high as 13% have been obtained for liquid DSCs and recently, solid-state DSCs using solid hole transport materials as well as perovskite materials as sensitizers have showed efficiencies above 17%. From a physical point of view, charge carrier generation in a DSC takes place in a photoactive light harvester chemisorbed on a mesoscopic semiconductor oxide. Due to the excellent electronic overlap, the exciton separation occurs by the injection of the excited electrons and holes from the dye into the conduction band of the semiconductor and redox electrolyte, respectively. A high surface area mesoscopic photoanode is necessary to ensure a high-density bulk heterojunction and efficient light harvesting in the visible part of the solar spectrum due to effective dye loading. This implies that the DSC has a large, heterogeneous interface through which photogenerated electrons may be intercepted as a result of the slow electron transport. The latter is governed by an ambipolar diffusion mechanism controlled by trap-limited hopping through a long path to the transparent conductive electrode. It is however hindered by the low electron mobility in anatase TiO2 nanoparticles and the multiple grain boundaries in the mesoporous film. The work conducted during this thesis aimed at developing new architectures for the electrodes of DSCs using notably atomic layer deposition (ALD) and inverse opal host-guest structures with different semiconducting metal oxides. First, using a self-assembled polymeric template and ALD, we fabricated a porous ZnO inverse opal backbone. Advantages are numerous: a high electronic mobility of ZnO, a precise control of the metal oxide deposition using ALD and in situ growth of ZnO nanowires at low temperatures. Besides increasing the surface area available for photoactive dye molecules, single-crystal nanowires offer a defect-free path for photoinjected electrons. However, because of the relative instability of ZnO in acidic environments, a TiO2 overlayer was deposited by ALD. This proved to increase the recombination resistance of photogenerated electrons and increased their lifetime in the porous film. This host-guest approach was further transposed to three related fields in photo-electrochemistry using different ALD precursors. For water splitting, a niobium-doped tin oxide inverse opal backbone was constructed, which served as a host for hematite deposited by atmospheric pressure chemical vapour deposition or ALD. In the field of plasmonics, a TiO2 inverse opal host was made to act as a scaffold for silver nanoparticles. For solid-state DSCs, we strived to achieve an ultra-thin ZnO inverse opal, which acted as a host for perovskite crystals. Finally, we studied the effect of very thin TiO2 and ZnO overlayers deposited by ALD on top of mesoporous ZnO and TiO2, respectively. We were able to accurately monitor the thickness of overlayers and study the effects on photovoltaic parameters such as transport and recombination rates of photogenerated electrons, photocurrent, photovoltage, fill factor and efficiency.