Accurate prediction of standard enthalpy of formation based on semiempirical quantum chemistry methods with artificial neural network and molecular descriptors

Abstract

This work investigates possible improvements in the accuracy of semiempirical quantum chemistry (SQC) methods for the prediction of standard enthalpy of formation (Δ f H o) through the use of an artificial neural network (ANN) with molecular descriptors. A total of 142 organic compounds with enough structural diversity has been considered in the training set. Standard enthalpy of formation for the selected compounds at the semiempirical PM3 and PM6 quantum chemistry methods is collected from literature and is calculated by using the semiempirical PM7 method in this work. The multiple stepwise regression is first used to screen effective molecular descriptors, which are highly correlated with the error terms of the standard enthalpy of formation compared with experimental values. The obtained seven effective molecular descriptors are then used as input set to establish three 7-11-1 neural network-based correction models to improve the accuracy of SQC methods. By using the developed correction models, the mean absolute errors for Δ f H o of PM3, PM6, and PM7 methods are reduced from 22.36, 18.60, and 17.27 to 9.86, 9.83, and 8.95, respectively, in kJ/mol. Meanwhile, the results of the test set show that the neural network does not have the problem of overfitting. Detailed analysis of the seven effective molecular descriptors indicates that the major source of the correction models is the electron-withdrawing effect. The developed ANN models for the three selected SQC methods provide an efficient method for the quick and accurate prediction of thermodynamic properties. K E Y W O R D S artificial neural network, enthalpy of formation, molecular descriptors, semiempirical quantum chemistry methods 1 | INTRODUCTION Thermodynamic properties are critical in almost all branches of chemistry and chemical engineering, such as chemical kinetics, combustion processing, and petroleum industry. [1,2] They are used to determine heat of reaction, equilibrium constant, reaction rate constants, and so on. Thus, accurate prediction of thermodynamic properties is one of the major focuses for both experimental and theoretical researchers. [3,4] Among the various thermodynamic properties, standard enthalpy of formation (Δ f H o) is a key parameter of species. Generally, experimental measurements of thermodynamic properties are reliable but become difficult and expensive as the research systems become larger. Hence, empirical methods were developed. Among the empirical methods, the group additivity (GA) method developed by Benson et al [5] is one of the quickest

Towards Computational Screening for New Energetic Molecules: Calculation of Heat of Formation and Determination of Bond Strengths by Local Mode Analysis

“…Given the limitations of these two methods, herein we have pursued the use of PM7 to determine ΔH f(g) (Stewart, 2013). This has been shown by Wan et al to out-perform previous semiempirical methods for a set of 142 organic molecules (Wan et al, 2020). Elioff et al evaluated its capabilities compared to both the isodesmic and the group equivalence methods for nitrogencontaining organic molecules (Elioff et al, 2016).…”

“…The PM7 method also carries the advantage that no further calculations beyond a geometry optimisation are required, which renders it attractive as part of a high throughput study. Moreover, semi-empirical calculation methods have a wide application and user base, and are being continuously improved (Wan et al, 2020).…”

On the limiting stage of oxidation of alkylarenes by potassium permanganate

“…Потрібні для оцінки змін ентальпії реакцій термохімічні дані багатьох хімічних частинок експериментально визначено і наведено в базах даних [13], для інших частинок термохімічні характеристики можуть бути обчислені як неемпіричними [14,15], так і напівемпіричними [16] методами сучасної квантової хімії. Термохімічні характеристики як окремих частинок, так і реакцій за їх участю обчислювали з використанням напівемпіричного методу РМ7 (пакет МОРАС2016).…”

Predicting Experimental Heats of Formation via Deep Learning with Limited Experimental Data

“…In 2018, a group of researchers predicted solar cell efficiencies with simple machine learning algorithms such as linear regression, the artificial neural network (ANN), and random forest with only 280 experimental data points . Similarly, the ANN was again proven to be capable of predicting semiempirical quantum chemical properties . In one of the most recent studies, a deep-learning ANI-1ccx method was proposed to compute enthalpies of formation of molecules to near chemical accuracy without training directly on experimental values, with time complexity O ( n ) .…”