Parameterizing Complex Reactive Force Fields Using Multiple Objective Evolutionary Strategies (MOES). Part 1: ReaxFF Models for Cyclotrimethylene Trinitramine (RDX) and 1,1-Diamino-2,2-dinitroethene (FOX-7)

Abstract

ReaxFF (van Duin, A.C.T.; Dasgupta, S.; Lorant, F.; Goddard, W.A. J. Phys. Chem. A, 2001, 105, 9396-9409) reactive potentials are parametrized for cyclotrimethylene trinitramine (RDX) and 1,1-diamino-2,2-dinitroethene (FOX-7) in a novel application combining data envelopment analysis and a modern self-adaptive evolutionary algorithm to optimize multiple objectives simultaneously and map the entire family of solutions. In order to correct the poor crystallographic parameters predicted by ReaxFF using its base parametrization (Strachan, A.; van Duin, A. C. T.; Chakraborty, D.; Dasgupta S.; Goddard, W. A. Phys. Rev. Lett., 2003, 91, 098301), we augmented the existing training set data used for parametrization with additional (SAPT)DFT calculations of RDX and FOX-7 dimer interactions. By adjusting a small subset of the ReaxFF parameters that govern long-range interactions, the evolutionary algorithm approach converges on a family of solutions that best describe crystallographic parameters through simultaneous optimization of the objective functions. Molecular dynamics calculations of RDX and FOX-7 are conducted to assess the quality of the force fields, resulting in parametrizations that improve the overall prediction of the crystal structures.

Parameterizing Complex Reactive Force Fields Using Multiple Objective Evolutionary Strategies (MOES): Part 2: Transferability of ReaxFF Models to C–H–N–O Energetic Materials

“…All information used in the training set is that found in the Supporting Information in the accompanying paper. 9 The difference between the five-objectives, Type 2, and sixobjective training sets (Table 1) is in the partitioning of the SAPT(DFT) 11,12 information. The six objectives training set was explored to determine if partitioning the SAPT(DFT) FOX-7 data from the RDX data would result in a ReaxFF-lg type force field that better describes hydrogen bonded systems, such as FOX-7 or TATB.…”

“…Descriptions of Parameters Varied during the MOES Search. In the accompanying study, 9 parameters within a small subset of the ReaxFF parameters used in ref 6 [the 46 parameters corresponding to the vdW energy and the EEM in eq 1] were optimized using the MOES search process and five objectives, Type 2. The parameter set that best predicted the crystal structure of RDX produced results that were in significantly better agreement with experimental crystal parameters than those produced using the ReaxFF parameters in ref 6, but at the expense of the description of the molecular structures, as will be shown hereafter.…”

“…8 Application of ReaxFF-lg to TATB and NM also demonstrated a modest improvement in density predictions for TATB (4.09% error versus −6.30% using ReaxFF) and NM (−6.69% error versus −8.04% using ReaxFF). 8 In an accompanying paper, 9 we described the use of the Multiple Objective Evolutionary Strategies (MOES) method to obtain a set of parameters for the original ReaxFF function [eq 1] that would better describe the ambient state crystal structure of RDX. In that study, the 46 parameters corresponding to the van der Waals and EEM were allowed to vary in the MOES search.…”

“…In fact, it was noted by the developers that "since no data were included to train these vdW terms to fit the long-range London dispersion, they rather play a role in modulating the various valence interactions by including the Pauli repulsion or steric interactions so important for valence structure." 8 In the accompanying study, 9 the QM information within the original ReaxFF training set and the additional SAPT(DFT) information were used to define five objectives for use in the evolutionary fitting scheme, with the MOES search producing a Pareto-optimal set of force fields. From these, we selected the one that best described the crystal parameters of RDX.…”

“…We also explore the sensitivity of the results when separating the SAPT(DFT) information into two groups for separate objective evaluations, where the SAPT(DFT) information for the RDX and FOX-7 dimers reside in separate objectives rather than just one, and compare the force fields to the results obtained in the accompanying study. 9 The paper is structured as follows: Section 2 provides details of the computational approach, including description of the training set, the parameters that are allowed to vary, the screening procedure used to select the "best" force field from among the Pareto-optimal solutions, and a description of the methods to analyze the crystallographic and molecular structural parameters obtained in molecular dynamics simulations using the force fields. Section 3 describes the application of the force fields to the aforementioned set of energetic crystals to determine transferability.…”