Understanding the Decomposition of Dimethyl Methyl Phosphonate on Metal-Modified TiO2(110) Surfaces Using Ensembles of Product Configurations.

Abstract

The decomposition of dimethyl methyl phosphonate (DMMP), a simulant for the nerve agent sarin, was investigated on Cu4/TiO2(110) and K/Cu4/TiO2(110) surfaces using a combination of near-ambient-pressure X-ray photoelectron spectroscopy (NAP-XPS) and density functional theory calculations (DFT). Mass-selected Cu4 clusters and potassium (K) atoms were deposited onto TiO2(110) as a metal catalyst and alkali promoter to improve the reactivity and recyclability of the TiO2 surface after exposure to DMMP. Surface reaction products resulting from decomposition of DMMP were probed by NAP-XPS measurements of phosphorus (P) 2p and carbon 1s core-level spectra. The Cu4/TiO2(110) surface is found to be very active for DMMP decomposition with highly reduced P-species observed even at room temperature (RT). The codeposition of K atoms and Cu4 clusters further improves the reactivity with no intact DMMP detectable. Temperature-dependent measurements show that the presence of K atoms promotes the removal of residual P-species at temperatures > 600 K. Detailed DFT calculations were performed to determine the surface structures and energetically accessible pathways for DMMP decomposition on Cu4/TiO2(110) and K/Cu4/TiO2(110) surfaces. The calculations show that DMMP and P-containing reaction products preferentially bind to the TiO2 surface, while the molecular fragments, i.e., methoxy and methyl, bind to both the Cu4 clusters and TiO2. The Cu4 clusters make the P-O, O-C, and P-C bond cleavages of DMMP markedly more exothermic. The Cu4 clusters are highly fluxional with atomic structures that depend on the configuration of fragments bound to them. Finally, the manifold of P 2p chemical shifts calculated for a large number of energetically favorable configurations of decomposition products is in good agreement with the observed XPS spectra and provides an alternative way of interpreting incompletely resolved core-level spectra using an ensemble of observed structures.