Probing UV photo-oxidation on oxide surfaces

Photochemistry at oxide surfaces is important for solar-energy conversion.^1,2 The most promising candidate for solar-powered water splitting (i.e., hydrogen production) and, thus, benign energy production is titania (TiO₂)-based photocatalysis. Since the initial work of Fujishima and Honda,² numerous improvements related to water splitting have been reported.³ However, even after nearly 40 years of research, major issues concerning the photoactivity of this important oxide are still debated. For example, a convincing explanation as to why the anatase polymorph of TiO₂ shows much more photocatalytic activity than its rutile form, by an order of magnitude,^4,5 is unavailable.³

Even simple reactions, such as carbon monoxide (CO) oxidation (see Figure 1), are not understood because of the almost exclusive focus on oxide-powder samples. Such samples are structurally undefined, while all possible surface orientations—often with a high defect density—are probed simultaneously.

The surface-science approach to understanding heterogeneous catalysis has been quite successful.⁶ Consequently, progress toward unraveling fundamental principles governing oxide-surface chemistry urgently necessitates detailed experimental and theoretical studies on model systems.

Figure 1.Photo-oxidation of carbon monoxide (CO) adsorbed on a rutile titania (TiO₂) (110) surface.⁷ C: Carbon. O: Oxygen. O₂: Molecular oxygen. O_2c: Bridging oxygen. CO₂: Carbon dioxide. Ti_5c: Fivefold coordinated titanium.

Especially for TiO₂, surface-oxide research within the past two decades has proceeded along a different direction than that of metal surfaces. Primarily driven by the tremendous success of microscopic interrogation, particularly by scanning-tunneling microscopy (STM), a wealth of structural information has become available about processes on the atomic scale, e.g., reactions occurring at defect sites on rutile^8–15 and, to a lesser extent, anatase.^16–18 By employing density functional theory (DFT), many STM observations have been explained.⁹

Figure 2.Reflection-absorption IR spectroscopy data recorded for CO adsorbed on a rutile TiO₂(110) substrate. (A) Prior to UV exposure. (B)–(F) Increasing exposure to UV photons.⁷

However, identification of chemical intermediates by STM/ DFT is subject to some pitfalls (e.g., reliable identification of hydroxyl species at TiO₂ surfaces remains a challenge). It has become clear that spectroscopic methods, and in particular IR vibrational spectroscopy, are indispensable for establishing a reliable foundation of chemistry on oxide surfaces, as has been demonstrated with respect to chemistry on metal surfaces through the surface-science approach.

Unfortunately, specific optical properties of oxides lead to severe problems for applications of reflection-absorption IR spectroscopy (IRRAS), the standard experimental method in this field. Sensitivity to molecular vibrations within adsorbates on oxides is reduced by two orders of magnitude with respect to metals. Consequently, despite the availability of a large set of IR data recorded in transmission mode for powders, data for well-defined oxide model systems is virtually unavailable. This is one of the primary reasons why our understanding of chemistry and photochemistry on oxide surfaces is still relatively limited.

We recently overcame these intensity problems by employing a novel, carefully optimized apparatus, where an IR spectrometer is directly attached to an ultrahigh-vacuum chamber.¹⁹ We detected (for the first time) the internal CO-stretch vibration of a TiO₂single crystal surface through IRRAS⁷ (see Figure 2) and consequently demonstrated that application of IRRAS on oxide surfaces offers huge potential for other molecular adsorbate species.²⁰ When adsorbed CO is exposed to UV photons in the presence of molecular oxygen (O₂), photo-oxidation proceeds without any intermediates. Activated O₂ reacts directly with CO, yielding carbon dioxide.⁷ Furthermore, we determined the photo cross section of the photo-oxidation reaction and found that it agrees with previous data.⁷ In very recent measurements on anatase single crystals, we found that the corresponding cross section is much larger. This parallels observations reported for powders and indicates that the surprisingly large photo cross sections seen for anatase powders do not result from either special types of surface-active sites or a particular form of defects characteristic for anatase, as proposed previously. In contrast, they originate from special features of the (bulk) electronic structure, specifically the presence of an indirect band gap.²¹

These results have important implications for our fundamental understanding of photochemical energy conversion in general, as well as for fabrication of materials with high photochemical cross sections. For nanoparticles with diameters in the sub-10nm region, the electronic structure of particles will be strongly disturbed with respect to the bulk. Consequently, the longer lifetimes of electronic excitations (electron-hole pairs) of anatase are predicted to be reduced to the values characteristic of rutile, since the indirect band gap will no longer reduce the electron-hole recombination rate. We next plan to look at other photochemical reactions on TiO₂and zinc oxide surfaces, with particular emphasis on doping effects.