Dataset Information

Prediction of organic homolytic bond dissociation enthalpies at near chemical accuracy with sub-second computational cost.

ABSTRACT: Bond dissociation enthalpies (BDEs) of organic molecules play a fundamental role in determining chemical reactivity and selectivity. However, BDE computations at sufficiently high levels of quantum mechanical theory require substantial computing resources. In this paper, we develop a machine learning model capable of accurately predicting BDEs for organic molecules in a fraction of a second. We perform automated density functional theory (DFT) calculations at the M06-2X/def2-TZVP level of theory for 42,577 small organic molecules, resulting in 290,664 BDEs. A graph neural network trained on a subset of these results achieves a mean absolute error of 0.58?kcal?mol^-1 (vs DFT) for BDEs of unseen molecules. We further demonstrate the model on two applications: first, we rapidly and accurately predict major sites of hydrogen abstraction in the metabolism of drug-like molecules, and second, we determine the dominant molecular fragmentation pathways during soot formation.

SUBMITTER: St John PC

PROVIDER: S-EPMC7214445 | biostudies-literature | 2020 May

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Prediction of organic homolytic bond dissociation enthalpies at near chemical accuracy with sub-second computational cost.

St John Peter C PC Guan Yanfei Y Kim Yeonjoon Y Kim Seonah S Paton Robert S RS

Nature communications 20200511 1

Bond dissociation enthalpies (BDEs) of organic molecules play a fundamental role in determining chemical reactivity and selectivity. However, BDE computations at sufficiently high levels of quantum mechanical theory require substantial computing resources. In this paper, we develop a machine learning model capable of accurately predicting BDEs for organic molecules in a fraction of a second. We perform automated density functional theory (DFT) calculations at the M06-2X/def2-TZVP level of theory ...[more]

PMID: 32393773

Similar Datasets

Project description:The concept of a bond energy between a pair of atoms is well-established in chemistry and physics. In materials science, it is a fundamental parameter in the development of thermodynamic models such as the regular, quasi-chemical and sub-regular solution models, as well as the central atoms model. Accurate bond dissociation enthalpies are available for gaseous molecular compounds, but these values are likely to differ significantly from single bond strengths between atoms in liquids and solids. While interatomic potential functions have been developed for atomic pairs by fitting to observed quantities, these functions often contain invariant transformations that yield bond energies that differ by up to a factor of four from values provided by other potentials for the same system, even though both may produce the same physical properties. Moreover, there is presently no widely used approach to determine bond enthalpies in condensed phases. An approach has been developed earlier to calculate bond enthalpies in solid and liquid phases using classical thermodynamic concepts, measured enthalpies for compound formation and elemental sublimation, and by counting bonds in the thermodynamically stable product and reactant phases (Miracle et al., 2011). That work reported bond enthalpies for essentially all stable solid and liquid elements, as well as bond strengths between unlike atoms for 71 different intermetallic compounds in 15 binary systems. This earlier work was validated by estimating elemental fusion enthalpies and surface energies, as well as formation enthalpies for ternary intermetallic compounds. Given the utility of this approach, the present dataset applies this earlier methodology to produce bond enthalpies between unlike atom pairs in an additional 904 binary intermetallic compounds from 443 systems, giving a total dataset of 975 bond enthalpies from 458 binary systems. Typical errors in the values reported here (from enthalpy measurements) are ±4%, and larger errors of about ±10-30% occur for a small subset of values where the number of bonds in the structure are difficult to establish. Used appropriately, these bond enthalpies enable classical approximations that-to first order-can capture critical material properties. A wide range of such estimates are possible, including the energies of vacancies and other atomic defects, solution enthalpies for complex, concentrated solid solution alloys (CCAs), formation enthalpies of higher-order compounds, and metallic glass stability. These bond enthalpies may also be useful for establishing trends in systematic studies that cover many systems, thus narrowing the scope of subsequent experimental measurements or computations which are more accurate, but are also more difficult and time consuming.

Dataset Information

Prediction of organic homolytic bond dissociation enthalpies at near chemical accuracy with sub-second computational cost.

Publications

Prediction of organic homolytic bond dissociation enthalpies at near chemical accuracy with sub-second computational cost.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets