Design and implementation of a next-generation software interface for on-the-fly quantum and force field calculations in automated reaction mechanism generation

https://doi.org/10.1016/j.compchemeng.2012.11.009Get rights and content

Abstract

A software interface for performing on-the-fly quantum and force field calculations has been developed and integrated into RMG, an open-source reaction mechanism generation software package, to provide needed estimates of thermodynamic parameters. These estimates based on three-dimensional molecular geometries bypasses the traditional group-additivity-based approach, which can suffer from lack of availability of necessary parameters; this issue is particularly evident for polycyclic species with fused rings, which would require ad hoc ring corrections in the group-additivity framework. In addition to making extensive use of open-source tools, the interface takes advantage of recent developments from several fields, including three-dimensional geometry embedding, force fields, and chemical structure representation, along with enhanced robustness of quantum chemistry codes. The effectiveness of the new approach is demonstrated for a computer-constructed model of combustion of the synthetic jet fuel JP-10. The interface also establishes a framework for future improvements in the chemical fidelity of computer-generated kinetic models.

Highlights

► RMG is an open-source automated reaction mechanism generation tool. ► An interface was developed to incorporate 3D molecular structures into RMG. ► The interface improves reliability of thermodynamic estimates for polycyclic species. ► The interface creates opportunities for improvement in other areas.

Section snippets

Background

Automated reaction mechanism generation software is an important tool in the generation of detailed chemical kinetic models for complex reacting systems (e.g. pyrolytic and combustion systems). RMG (Reaction Mechanism Generator) is one example of such software (Allen et al., 2009), which uses a rate-based mechanism construction algorithm (Susnow, Dean, Green, Peczak, & Broadbelt, 1997), and has been applied to a number of systems, including combustion of butanol (Harper, Van Geem, Pyl, Marin, &

Design overview

An overview of the QMTP interface is shown in Fig. 1. The workflow starts by estimating a three-dimensional molecular structure using RDKit, followed by calls to an outside program to perform quantum mechanics or force field calculations to refine that geometry and compute its enthalpy and vibrational frequencies; the results of the calculation are then read and used to calculate the desired thermodynamic properties using standard statistical mechanical relationships within the framework of the

Failure checking and recovery

Although the codes used to perform quantum mechanics or force field calculations are relatively robust, they are not error proof. When calculations of this nature are performed manually, troubleshooting is often required. We have incorporated automated troubleshooting into the design for the QMTP interface, since QMTP must successfully return a sensible estimate of the thermochemistry for every molecule considered in order to automatically generate large kinetic models – a failure rate of even

Selective use of more accurate methods

As shown in Table 3, MM4 is significantly more accurate than PM3 for certain types of molecules. Similarly, in some situations it would be better to run a DFT or high-level quantum chemistry calculation rather than relying on PM3 or MM4. It would be easy to modify QMTP to use these other types of calculations–the challenge is writing the logic which would decide which type of calculation to perform for each molecule.

Improvements to treating conformational flexibility

As discussed previously, the typical treatment is to assume harmonic behavior

Summary

An interface for performing on-the-fly quantum mechanics or force field calculations in the context of automated reaction mechanism generation has been described. Testing has demonstrated the impact, accuracy, and robustness of the interface, which make it suitable for routine use during mechanism generation. The interface is particularly useful for obtaining more-reliable thermochemical parameters for cyclic species (for which alternative automated group additivity-based approaches are prone

Acknowledgements

The authors would like to acknowledge helpful discussions with Professor Linda Broadbelt of Northwestern University, Laurent Catoire of CNRS, and Professor Richard H. West, now of Northeastern University. The authors also acknowledge work by Nick Vandewiele of Ghent University for development of the standalone thermodynamic property estimation tool mentioned here. Simon E. Albo, Oluwayemisi O. Oluwole, and Hsi-Wu Wong of Aerodyne Research, Inc. are acknowledged for work with the authors on

References (54)

  • S.W. Benson

    Thermochemical kinetics

    (1976)
  • S.W. Benson et al.

    Additivity rules for the estimation of molecular properties. Thermodynamic properties

    The Journal of Chemical Physics

    (1958)
  • N. Cohen et al.

    The thermochemistry of alkanes and cycloalkanes

  • N. Cohen et al.

    Estimation of heats of formation of organic compounds by additivity methods

    Chemical Reviews

    (1993)
  • G.M. Crippen

    Chemical distance geometry: Current realization and future projection

    Journal of Mathematical Chemistry

    (1991)
  • L.A. Curtiss et al.

    Gaussian-4 theory

    The Journal of Chemical Physics

    (2007)
  • J.-P. Ebejer et al.

    Freely available conformer generation methods: How good are they?

    Journal of Chemical Information and Modeling

    (2012)
  • Y. Feng et al.

    Homolytic Csingle bondH and Nsingle bondH bond dissociation energies of strained organic compounds

    The Journal of Organic Chemistry

    (2004)
  • M.J. Frisch et al.

    Gaussian 03, Revision D.01

    (2003)
  • Gilbert, K., & Guha, R. (2011) smi23d,...
  • O. Herbinet et al.

    Primary mechanism of the thermal decomposition of tricyclodecane

    Journal of Physical Chemistry A

    (2006)
  • InChI (2007)....
  • K.K. Irikura et al.

    Computational thermochemistry: Prediction and estimation of molecular thermodynamics

    American Chemical Society

    (1998)
  • Landrum, G. (2012). RDKit,...
  • T.H. Lay et al.

    Hydrogen atom bond increments for calculation of thermodynamic properties of hydrocarbon radical species

    The Journal of Physical Chemistry

    (1995)
  • T.B. Leite et al.

    Frog: A FRee Online druG 3D conformation generator

    Nucleic Acids Research

    (2007)
  • Lii, J.-H. (2010). Personal...
  • Cited by (56)

    • Method and automatic program for accurate thermodynamic data of reaction mechanisms for combustion modeling

      2022, Fuel
      Citation Excerpt :

      There are additional programs which were developed to evaluate thermodynamic data. An on-the-fly software interface termed quantum mechanics thermodynamic property (QMTP) [30] was developed for RMG by using semiempirical methods of PM3 or MM4 to obtain ΔfH° whose accuracy is still insufficient. An up-to-date program termed Automated Calculation of Reaction Energy Profiles (autodE) can obtain ΔfH° during the calculations of reaction energies [39], but its ΔfH° are not very accurate due to the low-level PBE0-D3BJ/def2-TZVP method for single point energies.

    • Fast estimation of standard enthalpy of formation with chemical accuracy by artificial neural network correction of low-level-of-theory ab initio calculations

      2021, Chemical Engineering Journal
      Citation Excerpt :

      This issue is partially remedied in the Reaction Mechanism Generator (RMG) [4] via a tree-based hierarchical group search, at the cost of decreased accuracy. Although on-the-fly ab-initio quantum chemistry computation of thermodynamics and kinetics [12–15] is more frequently used by automatic kinetic model generators, this is practical only at a lower level-of-theory (LLOT), e.g. B3LYP or similar hybrid-DFT methods. Unfortunately, species’ standard enthalpy of formation and reaction energetics by these methods are much less accurate than so-called chemical accuracy (approximately 1 kcal/mol or 4.2 kJ/mol [16]) [17,18].

    View all citing articles on Scopus
    1

    Present address: Aerodyne Research, 45 Manning Rd., Billerica, MA 01821, United States.

    View full text