The many faces of rutile titania

Abstract

Titania has been extensively studied for over three decades since it was discovered in 1972 that it acts as a photocatalyst for the water-splitting reaction [1]. This led to a proliferation of surface science studies focused on single crystal rutile TiO2(110), as it is the most thermodynamically stable face [2]. This enormous body of work gives TiO2(110) its status as the model substrate [3] with which to explore the surface science of metal oxides. The superb real space insights made possible by scanning probe microscopy have enabled this surface to be understood to an extraordinary level on the atomicscale [3–9]. The article by Dulub et al. [10] in this issue of Surface Science employs the scanning tunneling microscope (STM) to image point defects on the TiO2(011) surface. As with TiO2(110) [3–7], the (011) face contains a rich array of defects. However, whereas the (110) surface is dominated by individual O-vacancies, to the extent that theoretical calculations suggest that O-vacancies repel each other [11], the results for the (011) surface indicate that the majority of O-vacancies lie in pairs with a smaller number arranged in triangles or linear arrays. Theoretical calculations show that an ideal rutile TiO2 crystal is predominantly terminated by (110), (100), and (011) faces [2]. Thus, in order to advance the understanding of titania in real applications, it is imperative to expand our knowledge of both the (100) and (011) planes. An illustration of how the literature is skewed towards the (110) face is given with a search for single crystal rutile TiO2 studies in 2005 which gives 38 hits for the (110) surface, 6 for the (100) face, two for the (001) face, and only one for the (011) face. Although lagging behind in detail and volume to the (110) studies, the (100) face is relatively well-understood [12–16]. Individual steps in the 1 · 1 to 1 · 3-microfacet transition mechanism were imaged with STM and non-contact atomic force microscopy (NC-AFM) [16] with one of the intermediate phases proposed receiving some support from theoretical calculations [15]. On the TiO2(110) surface, a series of high temperature STM and low energy electron microscopy movies give evidence for mass-flow during the transition from 1 · 1 to 1 · 2 surfaces [8,9,17,18]. The STM movies suggest that interstitial Ti species react at the surface of TiO2(110) so that the crystal grows outwards in cycles of 1 · 1 and 1 · 2 terminations. Similar high temperature experiments could yield valuable insights into the TiO2(100) 1 · 1 to 1 · 3 phase transition. The (111), (210), and (001) faces of rutile TiO2 are not exposed to any sizeable extent on rutile TiO2 crystals. However, these surfaces may still be important because they represent phases which could be engineered to be exposed in order to enhance or suppress some reaction. For instance, whereas formate anions bridge between two Ti sites on the (110) surface, because of the greater distances between Ti sites on the (111) surface, the formate appears to be coordinated in a mixture of monodentate and bidentate chelating configurations [19,20]. For the TiO2(210) surface which lies formally between the (110) and (100) planes, atomic-scale rows were observed in STM and, with support from atomistic modelling, attributed to a sawtooth nanofacet model [21]. This model is similar to that initially proposed for the (100)-1 · 3 reconstruction [12], whereby nanoscale (110) planes are exposed. As for the (001) surface, various terminations have been observed, two of which were attributed to (011) facets and (114) facets [22–25]. However, a more recent article suggests that the various terminations all result from combinations of (111) and (001) microfacets [26].