The tandem photoelectochemical (PEC) cell based on oxide semiconductors for water splitting offers a potentially inexpensive route for solar hydrogen generation. At the heart of the device, a nanostructured photoanode for water oxidation is connected in series with one or two dye-sensitized solar cells (DSCs) that provide an extra bias to photogenerated electrons in order to perform water reduction at the platinum cathode. In Part A of this thesis, after reviewing the different technologies available for solar hydrogen production, I focus on different possible architectures for photoanode / DSC tandem cells enabled by the most recent advances in the field. First, the development of all organic squaraine dyes having a narrow absorption bandwidth extending into the near infrared region and a broad transparent window in the visible region of the spectrum have allowed the design of a new photoanode / DSC / DSC tandem architecture. The "Back DSC" tandem cell, where two DSCs placed side-byside exploit the photons transmitted by a single photoanode is the conventional architecture. In this thesis, I demonstrate the inverted "Front DSC" architecture, which allows the use of non-transparent lower cost metallic substrate for the photoanode, as well as the "tri-level" tandem cell where a panchromatic "black" dyebased DSC exploits the energy transmitted by the photoanode and the squaraine dyebased DSC. Opto-electronic studies on each of these three architecture are performed and the respective solar-to-hydrogen conversion (STH) efficiencies of 1.16 %STH, 0.76 %STH and 1.36 %STH are assessed. Finally, I leverage recent findings in the field of high voltage DSCs to demonstrate for the first time a tandem cell using only one photovoltaic cell to perform complete water splitting at efficiencies as high as 3.10 % STH. Such a result represents a ten-fold improvement over previous demonstrations with this class of device. This work describes a breakthrough in the inexpensive solar-to-chemical conversion using improved photon management in a dual-absorber tandem cell and undoubtedly constitutes a benchmark for solar fuel production by solution processable oxide-based devices. In Part B, I focus on the photoanode for the oxygen evolution reaction. Hematite (α-Fe2O3) is a great candidate material for this application due to its availability, low cost, non-toxicity and appropriate band gap allowing extensive absorption in the visible part of the spectrum. However, it suffers severe drawbacks, among which are poor majority carrier conductivity and a short diffusion length of minority charge carrier with regard to photon penetration depth. This circumstance causes most photogenerated charges to have a low probability of reaching the semiconductor / liquid junction and thus to participate in the water oxidation reaction. This feature implies the use of hematite morphologies having feature size in the range of 10-20 nm. A solution-based colloidal approach offers a simple and easily scalable method to obtain nanostructured hematite. However, this type of nanostructure needs to be exposed to an annealing step at 800 °C to become photoactive. This has been attributed to the diffusion of dopants from the substrate. In addition, this annealing step sinters the 10 nm particles colloid into a feature size approaching 100 nm. Here, I first show the effect of intentionally doping this material on the sintering and photoactivity. I then demonstrate a method that allows for the first time to apply high temperature annealing steps while controlling the feature size of the nanoparticles. Such an approach can be applied to any kind of nanostructure. This latter strategy coupled to further passivation of surface states allowed an improvement of the net photoactivity of this type of photoanode by a factor of two. A reproducible net photocurrent exceeding 4 mA.cm-2 was obtained. This result represents the highest performance reported for hematite under one sun illumination. In Part C of this thesis, I present several synthesis methods for titania nanostructures acting as photoanodes in the DSC. In this photovoltaic cell, the electronic loss that is typically discussed is the slow transport induced recombination. Charges recombining before reaching the electrode have a detrimental influence on the photocurrent and photovoltage. Several approaches have been proposed in order to reduce interfacial recombination and improve charge collection in liquid electrolyte and solid state-DSCs (ss-DSCs) including the use of radial collection nanostructures, one-dimensional ZnO and TiO2 nanorods, and nanowires as photoanodes. Even though these approaches show great promise, they have yet to achieve power conversion efficiencies above 5% in liquid electrolyte DSCs and 1.7% in ss-DSCs. In this part of my thesis, I expose results on the strategies I have pursued during the course of my PhD to improve the dynamics in the liquid and ss-DSC. From one dimensional titania nanotube arrays, I have moved to tri-dimensional fibrous network of crystalline TiO2 and more complex host-passivation-guest photoelectrodes.
EPFL2012
