A comparative study of the reaction dynamics of several potential energy surfaces of O(3P)+H2→OH+H. I

Abstract

Two new potential energy surfaces for th O+H2→OH+H reaction are presented, and a detailed comparision of the saddle point properties and thermal rate constants of these and of six other O+H2 surfaces is made. The two new surfaces are (1) an extended BEBO surface and (2) a rotated-Morse-oscillator-spine (RMOS) fit to the accurate ab initio POLCI surface of Walch and Dunning. In the BEBO surface, an improved end atom repulsive potential is used which leads to a much more accurate barrier estimate (11.52 kcal/mol) than with the usual anti-Morse expression. The POLCI–RMOS surface is an essentially quantitative fit to the ab initio points, and has a barrier of 12.58 kcal/mole. The other O+H2 surfaces examined include the LEPS surface of Westenberg and de Haas (LEPS-WDH) and of Johnson and Winter (LEPS-JW), the diatomics-in-molecules (DIM) surface of Whitlock, Muckerman, and Fisher, the ab initio surface of Howard, McLean, and Lester (HML), and a fit to HML’s surface by Schinke and Lester (SL). For the LEPS-JW and SL surfaces, the quasiclassical trajectory (QCT) and Wigner corrected transition state theory (TST) rate constants are in excellent agreement with each other over the 300–2000 K range. There is, however, poor agreement between the QCT and TST rate constants for the DIM surface. This is apparently due to the more product-like location of the saddle point on that surface. Using both the QCT and TST results to make best estimates of the true O+H2 rate constants, we find that the LEPS-JW, DIM, and POLCI rate constants all agree with experiment within the experimental uncertainties. Although the treatment of multiple surface and spin-orbit effects does introduce some uncertainty into this analysis, the experimental uncertainties are large enough to enable agreement with experiment almost irrespective of which electronic weighting factor is used. While the LEPS-JW and POLCI surfaces have many similar features (both have ∼12.5 kcal/mol barriers), the DIM surface is quite different ( its barrier is 13.35 kcal/mol) and yet its differences compensate one another in the rate constant calculation so as to give agreement with experiment. Other comparisons with experiment indicate that the BEBO and LEPS-WDH surfaces have too low a barrier while the SL surface has too high a barrier. The normal mode stretch frequencies published by HML are surprisingly different from SL (and from those of the other surfaces), and this causes HML’s rate constants to be significantly higher then experiment even when SL’s are significantly lower.