A comparison of gold and molybdenum nanoparticles on TiO2(110) 1x2 reconstructed single crystal surfaces
John R. Kitchin, Mark A. Barteau, and Jingguang G. Chen, “A comparison of gold and molybdenum nanoparticles on TiO2(110) 1x2 reconstructed single crystal surfaces”, Surface Science 526(3), 323-331 (2003). https://doi.org/10.1016/s0039-6028(02)02679-1
This summary was written by a language model from the paper and has not been checked line by line. Rely on the published paper for details.
Question
Gold nanoparticles of 1 to 5 nm on oxides are active for reactions such as low-temperature CO oxidation, but they sinter readily, and particle growth can destroy that activity. Early transition metals are expected to interact more strongly with oxide supports than noble metals do, yet the growth of Mo on TiO2(110) had not been examined with atomic-resolution scanning tunneling microscopy. The paper asks how Au and Mo nanoparticles grown on the same TiO2(110)-(1x2) surface differ in their resistance to sintering and in their shapes, and what interfacial chemistry explains the difference (Kitchin et al. 2003).
Approach
Experiments were done in an ultrahigh vacuum chamber (base pressure 2 x 10-10 Torr) with Auger electron spectroscopy and a variable-temperature STM. The rutile TiO2(110) crystal was cleaned by Ar+ sputtering and annealing at 850 K, which gave the 1x2 reconstruction. Au (from a W-wire-heated pellet) and Mo (from a resistively heated Mo ribbon) were deposited by physical vapor deposition at room temperature; the Mo flux could not be calibrated with the quartz microbalance because the source ran too hot. Samples were annealed and then imaged at 300 K in constant current mode (about +2 V, 0.1 nA). Particle diameters, heights, fitted-ellipse axes, aspect ratios and volumes were extracted with Matlab image analysis routines that subtract a background found by morphological opening and then threshold the particles; the authors note these values are not corrected for tip convolution and that heights are systematically underestimated. The discussion then uses literature adhesion energies on alumina, heats of oxygen adsorption, and NIST-JANAF heats of formation of Ti and Mo oxides to compare the energetics of oxygen transfer from titania to Mo (Kitchin et al. 2003).
Key findings
Room-temperature deposition produced particles 1 to 3 nm across for both metals. Au particles averaged 2.7 nm in diameter and 0.45 nm in height; after a flash anneal to 773 K they grew to 5.5 nm and 0.73 nm, with fewer particles, a fairly uniform size and an average aspect ratio of 1.3. The annealed Au particles favored step edges and the ends of reconstructed rows, although many sat on terraces. Mo particles were under 2 nm after deposition and only reached 2.7 nm after the same flash anneal, with heights of 0.31 nm in both cases, so Mo sintered much less than Au. (The abstract refers to annealing at 700 K, while the figure captions and text give 773 K for flash anneals and 783 K for long anneals.) (Kitchin et al. 2003)
Longer anneals did coarsen the Mo. After 1 h at 783 K, Mo formed anisotropic particles aligned along the [001] direction and frequently centered on the bright Ti rows, with an average minor axis of 1.5 nm (close to the 1.33 nm row spacing of the reconstructed surface) and aspect ratios of 1.2 to 3.2 (mean about 1.8). After 19.5 h, almost only large elongated particles remained, some over 1 nm high. Au never showed this row templating. The total Mo particle volume stayed roughly constant up to about 800 K; above that temperature Mo disappeared from both STM images and Auger spectra, which the authors attribute to evaporation of an Mo oxide or diffusion into the bulk. No surface reconstruction around the Mo particles was seen, unlike reports for Al on TiO2 (Kitchin et al. 2003).
To explain the difference, the authors show that adhesion energies of metals on alumina correlate with the initial heat of oxygen adsorption on the metal, about 86.5 kcal/mol O for Mo versus about 20 kcal/mol O for Au. Heats of reaction derived from NIST-JANAF data put the reduction of TiO2 to lower oxides at roughly 84 to 94 kcal/mol O, while forming MoO2 or MoO3 releases only about 70 or 59 kcal/mol O. Full oxidation of Mo by titania is therefore unfavorable, but partial oxidation of Mo at the interface (comparable to the 86 kcal/mol O adsorption heat) with partial reduction of the oxide is feasible. This bonding to surface oxygen accounts for both the slow growth and the anisotropy: Mo particles centered on Ti rows can bond to the oxygen rows on both sides (Kitchin et al. 2003). The authors note the consistency with reports that metals with high oxygen adsorption heats (Al, Ti, Nb, Cr and others) reduce TiO2(110) while Rh, Ni, Pd, Pt and Cu do not.
Significance
The paper ties nanoparticle thermal stability on an oxide to a simple descriptor, the heat of oxygen adsorption on the metal, and proposes it as an indicator of adhesion energy and sintering resistance. It also suggests using sinter-resistant Mo clusters as anchors for a second metal with lower surface energy, which might segregate to the surface and form stabilized core-shell particles (Kitchin et al. 2003).