Methanol on Anatase TiO2 (101): Mechanistic Insights into Photocatalysis

The photoactivity of methanol adsorbed on the anatase TiO2 (101) surface was studied by a combination of scanning tunneling microscopy (STM), temperature-programmed desorption (TPD), X-ray photoemission spectroscopy (XPS), and density functional theory (DFT) calculations. Isolated methanol molecules adsorbed at the anatase (101) surface show a negligible photoactivity. Two ways of methanol activation were found. First, methoxy groups formed by reaction of methanol with coadsorbed O2 molecules or terminal OH groups are photoactive, and they turn into formaldehyde upon UV illumination. The methoxy species show an unusual C 1s core-level shift of 1.4 eV compared to methanol; their chemical assignment was verified by DFT calculations with inclusion of final-state effects. The second way of methanol activation opens at methanol coverages above 0.5 monolayer (ML), and methyl formate is produced in this reaction pathway. The adsorption of methanol in the coverage regime from 0 to 2 ML is described in detail; it is key for understanding the photocatalytic behavior at high coverages. There, a hydrogen-bonding network is established in the adsorbed methanol layer, and consequently, methanol dissociation becomes energetically more favorable. DFT calculations show that dissociation of the methanol molecule is always the key requirement for hole transfer from the substrate to the adsorbed methanol. We show that the hydrogen-bonding network established in the methanol layer dramatically changes the kinetics of proton transfer during the photoreaction.

positions and heights are almost identical in both cases. The reaction products detected by TPD are also very similar.
The reaction with O2 was also observed by STM, see Figure S2. The surface was exposed to 1 L O2 at T = 10 K, followed by exposure to 0.1 L methanol at 110 K and annealing to 350 K for 10 minutes. The resulting STM image ( Figure S2a) shows a surface containing only methoxy groups as a single reaction product. Figure S2b shows the same area after 28 minutes of illumination by UV light.
The photoreaction proceeds in an identical as in the main text ( Figure 4). The methoxy species are again partially converted to formaldehyde.
The reaction pathway is more complicated, though. The thermally activated step -reaction of methanol with O2 -has multiple possible pathways. O2 rarely dissociates on the anatase (101), even at room temperature. The O=O bond breaking becomes feasible, however, when the O2 accepts a proton and forms OOH. 2 The first step of the reaction is The OOH can undergo a series of reactions described in ref. 2 , where it dissociates and further reacts with water, provided that the substrate can provide enough excess electrons: Here the (O2)O is so called the bridging oxygen dimer, which is essentially an O adatom incorporated in the surface layer. 3 The final product of this reaction cascade are terminal OH groups, which can react with methanol via the reaction in Equation (1) of the main text. Alternatively, both metastable reaction products, OOH and (O2)O, can react with methanol directly. XPS data in Figure S1 show that the major reaction product is again the methoxy species; hydrogen atoms are not removed from the methyl group.
According to the reaction scheme in Equations (2-4), each chemisorbed O2 molecule initially adsorbed at the surface results in the formation of 4 methoxy groups, regardless of the exact reaction cascade. The O2 either first reacts with water, providing terminal OH groups, or directly enters reactions with the methanol. The result is the same, as the terminal OH group has a higher affinity to protons then methanol, allowing the reaction in Equation (1) of the main text. In our case, the amount of chemisorbed O2 is determined by the availability of excess electrons in the sample (i.e. by its reduction state and extrinsic doping), though excess electrons can also be obtained by photoexcitation. The availability of excess electrons is the determining factor for the resulting concentration of methoxy groups.
In computations, we also considered the case where the OOH group from reaction (2) does not dissociate, but abstracts another H atom from the methoxy group, forming H2O2. This reaction mechanism is possible, see the pathway in Figure S3. It is likely occurs in samples that have a low concentration of excess electrons (non-reduced material).   formaldehyde. These hydrogen atom are transferred to the surface during the reaction, forming so called bridging hydroxyl groups. In order to identify these hydroxyls in STM images (Figure 4d), we exposed a clean anatase (101) surface to atomic hydrogen (created by cracking H2 at a hot W filament). The result in Figure S4 shows that the bridging hydroxyl groups appear as bright protrusions. The adsorbed species become unstable when the STM sample bias is increased above +2 V. We used this knowledge for identification of the adsorbed species shown in Figure 4d in the main text. Figure S5 shows the typical appearance of the discussed species when imaged by STM. The methanol, methoxy, and formaldehyde all appear as bright dimers centred above the Ti5c surface atoms ( Figure S5a,b,c, respectively). This appearance is in agreement with the DFT calculations of the corresponding species ( Figure S5d,e,f). These species differ in the apparent height; typical line profiles are shown in Figure S5g. The formaldehyde appear highest, followed by methanol and methoxy (lowest). The apparent height is, though, a rather weak tool for chemical assignment. It only allows to discern different species when they are present at the surface together.

Figure S6: Geometry used for calculation of the C1s core-level shift in XPS. Methanol is adsorbed on one side of the slab, methoxy is on the other side. The distribution of the excess electron in the presence of the core hole is shown (calculated by PBE+U)
We have performed calculations of the C1s core-level shift between the adsorbed methanol and methoxy groups according to the final-state scheme of ref 4 . First we examined the rutile (110) surface to verify that our setup provides reasonable values, next we repeated the calculations for anatase. The resulting energy shifts are reported in table ST1 while the slab geometry is shown Figure S6. The calculations were performed using both standard PBE and PBE+U, and assuming the core electron to be either completely ejected, so that the system is positively charged, or promoted to the conduction band, so as to maintain charge neutrality. The core level shifts obtained with these different setups show slight variations, but the trend between rutile and anatase is apparent in all of them.  0.20 Table S2 shows TPD cracking patterns measured for our experimental settings. Formaldehyde and methanol data were measured by dosing 0.2 ML of the respective molecules on the anatase (101) surface and integrating TPD peaks of the relevant m/z signals. The cracking pattern of methyl formate was measured by dosing 0.67 ML methanol on the rutile (110) surface, illuminating it by UV light for 30 minutes and measuring TPD. Here we used the rutile (110) surface as a substrate, because the methyl formate peak does not overlap with methanol, as is the case on the anatase (101) surface. Figure S7 shows TPD signals of 0.67 ML methanol dosed on the rutile (110) and anatase (101) surfaces after UV illumination for 30 minutes. The methyl formate (m/z=60) peak is higher on rutile, likely because this coverage provides the maximum photoactivity, 5 while for anatase it is at the lower limit of coverages where the material is photoactive. For rutile Figure S7a), the methyl formate peak is well separate from the methanol signal and allows calibration of the cracking pattern (Table S2). For anatase ( Figure. S7b), the methanol and formaldehyde peaks partially overlap. The data do not indicate production of formaldehyde, which was reported in previous studies. [6][7] This is possibly due to higher UV doses used in this study. Even though we did not detect any other reaction product than methyl formate, we note that we have detected water after the UV illumination, see Figure. S8. The amount of water is relatively high, and correlates with the amount of produced methyl formate, therefore it seems unlikely that it would originate solely from contamination of the sample from the background pressure. In principle, this water might be produced by dehydration of the methyl formate (producing dimethyl ether). Such a reaction has been reported on the anatase (001) surface. 8 We did not measure the m/z signals that correspond to dimethyl ether. Alternatively, the water could originate from hydrogen atoms, which are produced in the photocatalytic reaction, and are transferred to the surface. These hydrogen atoms may react with surface O atoms of TiO2 surfaces and desorb as water, 9 effectively reducing the surface. However, this scenario seems less likely because this reaction typically occurs at a temperature of ~500 K, 5 while our TPD peak appears below room temperature. Figure S8: TPD spectra of m/z = 18 (water) measured on the anatase (101) surface after exposing to different doses of methanol and 30 minutes of UV illumination. The inset shows the corresponding amount of water in monolayers. The values were obtained by integration of the m/z=18 peak up to 280 K (the shoulder above room temperature is attributed to desorption from the sample plate).