A two-dimensional phase of TiO₂ with a reduced bandgap

Junguang Tao,¹

Tim Luttrell¹

& Matthias Batzill¹

Affiliations

Contributions

Corresponding author

Journal name:: Nature Chemistry
Volume:: 3,
Pages:: 296–300
Year published:: (2011)
DOI:: doi:10.1038/nchem.1006

Received: 25 October 2010
Accepted: 04 February 2011
Published online: 13 March 2011

Abstract

Titanium dioxide is the prototypical transition metal oxide photocatalyst. However, the larger than 3 eV bandgap of common bulk phases of TiO₂ limits its light absorption to UV light, making it inefficient for solar energy conversion. Attempts at increasing visible light activity by narrowing the bandgap of TiO₂ through doping have proven difficult, because of defect-induced charge trapping and recombination sites of photo-excited charge carriers. Here, we report the existence of a dopant-free, pure TiO₂ phase with a narrow bandgap. This new pure TiO₂ phase forms on the surface of rutile TiO₂(011) by oxidation of bulk titanium interstitials. We measure a bandgap of only ~2.1 eV for this new phase, matching it closely with the energy of visible light.

Figures at a glance

left

Figure 1: Atomic-resolution STM images of the rutile TiO₂(011) surface.

a,b, The (2 × 1) reconstructed surface. c,d, The new TiO₂ phase formed after annealing in a 1 × 10⁻⁶ torr O₂ atmosphere. Image size (a,c), 50 × 50 nm². The surface unit cell for the rectangular (2 × 1) reconstruction is indicated in b and the quasi-hexagonal symmetry of the new TiO₂ phase is shown in d. A line defect in the new TiO₂ phase can also be seen in d. This defect is an anti-phase boundary due to the registry of the new phase with the TiO₂(011) substrate. e, Line profiles indicating the atomic corrugation along the indicated lines in b and d.

Figure 2: Scanning tunnelling spectroscopy (STS) measurements.

a, Sample prepared so that both ‘new’ and ‘original’ phases are present simultaneously at the surface. The original (2 × 1) reconstructed surface is shown in blue, and the new TiO₂ phase in red. b, I–V spectra taken in the two surface areas (blue, original surface; red, new structure). c, Numerically differentiated dI/dV curves. The new surface phase exhibits additional electronic states at negative voltages (filled states). The empty states (positive bias voltage) are shifted slightly towards the Fermi level for the new TiO₂ phase. Thus, although STS cannot provide absolute values for the bandgap due to the strong electric field between the measuring probe and the sample, comparison of the STS spectra on the two surface structures unambiguously identifies a bandgap narrowing for the ‘new’ structure relative to the ‘original’ structure.

Figure 3: Comparison of TiO₂ phases by photoemission spectroscopy.

a,b, UPS spectra taken with a photon energy of 80 eV: full valence band, with 0 eV binding energy corresponding to the Fermi level (a); zoomed in view of the low binding energy region (b). The black curve corresponds to the photoemission spectrum for the original (2 × 1) reconstructed surface. The red curve was acquired after low-pressure oxidation and (partial) formation of a new TiO₂ phase. The blue line indicates a photoemission spectrum that has been taken after annealing the sample to ~600 °C in vacuum and re-formation of the original (2 × 1) surface structure. For the new TiO₂ phase a new state is formed with a peak at a binding energy of 2.1 eV. Also, a ~0.3 eV upward shift of the bulk bands is observed due to band bending at the surface. c, XPS spectra of the Ti 2p core level before (black line) and after (red line) oxygen treatment. No indication for formation of a Ti³⁺ sub-oxide can be detected, and the line shape barely changes with the oxidation procedure, apart from a slight upward shift of the spectrum after oxidation induced by band bending. d, Band diagram, deduced from UPS and STS, together with an indication of a proposed electron–hole separation of photo-excited electron–hole pairs at this surface. The hole is indicated by an open circle and is trapped at the surface, and the electron, indicated by a filled circle, can diffuse into the bulk.

Figure 4: Resonant photoemission measurement for the valence band maximum of the new TiO₂ phase.

a, Valence band spectra acquired at different photon energies. b, Variation of photoemission intensity at 2.1 eV as a function of photon energy; error bars represent the standard deviation of the data. The full red line in b is a fit of the experimental data to a Fano line shape. The resonance at ~55 eV corresponds to Ti 4sp states, thus showing that the valence band maximum is due to hybridization of O 2p with Ti 4sp, indicative of a TiO₂ phase.