International audience Photocatalytic alcohol dehydrogenation is of great interest to produce green and sustainable H2. Our aim is to identify which parameters play a key role in the photocatalytic mechanism. To do so, we supported four transition metal sulfides (HgS, MoSx, NiS2 and RuS2) on commercial anatase (MSx/TiO2) through a method previously reported. [1] Photocatalytic tests were performed with the MSx/TiO2 in a mix of isopropanol and water (50%vol). The semi-batch reactor was illuminated either with a 370 nm LED or with a 300W Xe lamp equipped with filters. Photon Yield (PY) is calculated as the ratio between the rate of H2 production and the incident flux of photons between 290 and 390 nm. Values are reported on Figure 1. Each of the MSx/TiO2 is active for the conversion of isopropanol in H2 and more active than bare TiO2. Water circulation regulates the temperature of the photoreactor. The PY was therefore measured at various temperature. All the MSx/TiO2 followed an Arrhenius law at least between 5°C and 45°C. Two additional parameters are extracted: a pre-exponential factor (Aapp) and an apparent activation energy (Ea,app). To measured PY at four temperatures, the test lasts at least 20 hours. We confirmed, partly through XPS, that after photocatalytic tests MSx phases were kept. Additionally, we measured the PY at different wavelength, the so-called action spectrum. It evidence that photons only absorbed by the MSx do not participate significantly to the H2 production. As the TiO2, the mass of photocatalyst and the reactor geometry are unchanged, we assumed that the amount of photons absorbed is also unchanged. Variation of PY observed on Figure 1 must be due to intrinsic properties of the MSx.Indeed, the lowest Ea,app were obtained for MoSx and RuS2 (Ea,app = 19 kJ/mol). We made a screening of the bulk sulfides for the electrocatalytic H2 production and a screening of the supported sulfides for thiophene hydrodesulfurization (HDS). Such model reaction shows the ability of a sulfide to catalyze H-H recombination. It showed that MoSx and RuS2 are the two best electrocatalysts with similar activity. On the contrary, the rate of HDS for the MoSx is six times higher than for the RuS2 whereas Ea,app are equal. As a result, the reduction of proton in hydrogen radical is the rate-determining step and recombination of two hydrogen radicals in H2 is faster. Note also that depending on the MSx the driving force of the PY is either the Aapp or Ea,app (See Figure 1). It will be discussed regarding the intrinsic properties of the MSx/TiO2. The slow kinetic for proton reduction is either due to a limited amount of electrons available (electronic structure is the key parameter) or due to low electrocatalytic activity (electrochemical behavior is the key parameter). Those results agree with a microkinetic model established for the photocatalytic dehydrogenation of isopropanol. To distinguish which parameter is the key one, we determined the electronic structure of the MSx/TiO2 with a methodology reported in Maheu et al. [2] UPS evidences that the contribution of the MSx in the valence band states is crucial.