Unravelling Site-Specific Photo-Reactions of Ethanol on Rutile TiO2(110)

Finding the active sites of catalysts and photo-catalysts is crucial for an improved fundamental understanding and the development of efficient catalytic systems. Here we have studied the photo-activated dehydrogenation of ethanol on reduced and oxidized rutile TiO2(110) in ultrahigh vacuum conditions. Utilizing scanning tunnelling microscopy, various spectroscopic techniques and theoretical calculations we found that the photo-reaction proceeds most efficiently when the reactants are adsorbed on regular Ti surface sites, whereas species that are strongly adsorbed at surface defects such as O vacancies and step edges show little reaction under reducing conditions. We propose that regular Ti surface sites are the most active sites in photo-reactions on TiO2.

): STM image (420 Å × 420 Å) acquired after illumination of the EtOH-covered r-TiO 2 (110) surface with UV-light for 11 min at 290 K in UHV. All the EtOH Ti / EtO Ti species disappeared but not the EtO br and EtO S ethoxides. Few of the EtO S ethoxides are indicated by white arrows. In the inset, the Ti troughs are indicated by thin white lines. The STM image was collected with a tunnelling current ≤ 0.1 nA and a tunnelling voltage of ~1.2 V.      Fig. S10a. Subsequently, the same area was scanned again, but this time with V t = +3V. We then zoomed out to a larger scanning area (166 Å × 166 Å) and switched back to our usual scanning conditions (see Fig. S10b). In this image, the area that was scanned earlier with a high bias (V t = +3V) is indicated by a white broken square. It can be seen that the EtO br species were not removed during scanning with high bias. However, as expected 3 , all the H ad species disappeared after scanning with V t = +3V. As a result, the O br rows and the Ti troughs became clearly visible in this area.
In the area scanned with high bias the white lines are superimposed on the Ti troughs.
By extrapolating the white lines to the area outside the white broken square it becomes evident that the many new species that appeared after UV-light illumination ascribed to H ad species are indeed centered at the O br rows (i.e. between the lines), as expected for the H ad species 3 . Moreover, it can be seen that the EtO br are centered about the O br rows. Also this result is expected, since the EtO br species are formed via EtOH dissociation at O br vacancy sites 4 . The described control experiment further supports the assignments given in the main text. Figure S10: (a) STM image (80 Å × 80 Å) acquired within a similar experiment as the image shown in Fig. S4 after exposing an r-TiO 2 (110) surface to EtOH at 300 K followed by UV light illumination at 290 K. The most apparent protrusions arise from EtO br ethoxides, three of which are indicated by white dotted circles. The other protrusions arise from H ad species. After this image was acquired, we scanned the same surface area again but this time with an increased bias of V t = +3 V, resulting in the removal of H ad species. (b) STM image (166 Å × 166 Å) acquired after scanning with V t = +3 V in the indicated area (same as in (a)). The white lines are superimposed on the Ti troughs in the squared region scanned with high bias. By extending these white lines to the area outside the square it becomes clear that the EtO br species and the H ad species are located at the O br rows between the Ti troughs (thin white lines), confirming their assignments. In (b), the same three EtO br species are indicated as in (a). In addition, a newly appeared species in a Ti trough (probably a water monomer from the background) is indicated by a blue circle.

Confirmation of H ad formation by TPD measurements.
The water-TPD spectra (m/z = 18) displayed in Fig. S11 further corroborate the formation of H ad species in the photo-reaction on EtOH / r-TiO 2 (110). The black TPD spectrum corresponds to EtOH / r-TiO 2 (110) and was acquired after UV-light illumination for 11 min. This spectrum shows a small and narrow desorption feature at ~350 K (denoted "") and a larger and broader feature with a maximum at ~520 K (denoted ""). For comparison, we show a water-TPD spectrum acquired from an h-TiO 2 (110) surface (green curve). This surface was prepared by exposing an r-TiO 2 (110) surface to water, using the same TiO 2 (110) sample. Here, the -feature is similar as in the previous case, but the -feature is less intense by a factor of ~2.6. The -feature arises from water molecules in the residual gas, which adsorb in the Ti troughs 1,5 while cooling the sample down to ~100 K. The -feature, however, arises from the recombination of H ad species with O br atoms 1,5 .
Consequently, the integrated area of the -feature is a measure of the H ad density.
An STM analysis of the h-TiO 2 (110) surface corresponding to the green spectrum in Fig. S11 yielded an H ad density of (11 ± 0.5) %ML, which is a factor of ~2.45 less than on the illuminated EtOH / r-TiO 2 (110) surface [(27 ± 1) %ML, see Fig. 1d,e]. Accordingly, the TPD and STM data are in excellent agreement. Figure S11: Water-TPD spectra of an illuminated EtOH / r-TiO 2 (110) surface (black curve) and a non-illuminated h-TiO 2 (110) surface (green curve). The same TiO 2 (110) crystal was used for these experiments.