- Previous Article
- Next Article
- Table of Contents
Industrial & Engineering Chemistry Research, Vol.56, No.4, 815-831, 2017
Development of Predictive Models of the Kinetics of a Hydrogen Abstraction Reaction Combining Quantum-Mechanical Calculations and Experimental Data
The importance of developing accurate modeling tools for the prediction of reaction kinetics is well recognized. In this work, a thorough investigation of the suitability of quantum mechanical (QM) calculations to predict the effect of temperature on the rate constant of the reaction between ethane and the hydroxyl radical is presented. Further, hybrid models that combine a limited number of QM calculations and experimental data are developed in order to increase their reliability. The activation energy barrier of the reaction is computed using various computational methods, such as B3LYP, M05-2X, M06-2X, MP2 and PMP2, CBS-QB3, and WIBD, with a selection of basis sets. A broad range of values is obtained, including negative barriers for all of the calculations with B3LYP. The rate constants are also obtained for each method, using conventional transition state theory, and are compared with available experimental values at 298 K. The best agreement is achieved with the M05-2X functional with cc-pV5Z basis set. Rate constants calculated at this level of theory are also found to be in good agreement with experimental values at different temperatures, resulting in a mean absolute error of the logarithm (M.AE(in)) of the calculated values of 0.213 over a temperature range of 200-1250 K and 0.108 over a temperature range of 300-499 K. Tunnelling and vibrational anharmonicities are identified as important sources of discrepancies at low and high temperatures, respectively. Hybrid models are proposed and found to provide good correlated rate-constant values and to be competitive with conventional kinetic models, i.e., the Arrhenius and the three-parameter Arrhenius models. The combination of QM-calculated and experimental data sources proves particularly beneficial when fitting to scarce experimental data. The parameters of the model built on the hybrid strategy have a significantly reduced uncertainty (reflected in the much narrower 95% confidence intervals) compared with the conventional kinetic models while also capturing well the experimental reaction rates with a MAE(in) of the rate constant of 0.118. This provides a useful strategy for kinetic model development.