A big data framework to validate thermodynamic data for chemical species

Abstract

The advent of large sets of chemical and thermodynamic data has enabled the rapid investigation of increasingly complex systems. The challenge, however, is how to validate such large databases. We propose an automated framework to solve this problem by identifying which data are consistent and recommending what future experiments or calculations are required. The framework is applied to validate data for the standard enthalpy of formation for 920 gas-phase hydrocarbon species retrieved from the NIST Chemistry WebBook. The concept of error-cancelling balanced reactions is used to calculate a distribution of possible values for the standard enthalpy of formation of each species. The method automates the identification and exclusion of inconsistent data. We find that this enables the rapid convergence of the calculations towards chemical accuracy. The method can exploit knowledge of the structural similarities between species and the consistency of the data to identify which species introduce the most error and recommend what future experiments and calculations should be considered.

Designing compact training sets for data-driven molecular property prediction through optimal exploitation and exploration

“…Our method is benchmarked on three different sets of data: (1)∼7000 molecules consisting of C,O,N,S,H from the QM7 database [42,43], (2) 920 hydrocarbon species from NIST chemistry webbook [44,45] , and (3) 591 surface intermediates on a transition metal facet [46]. The QM7 dataset contains atomization energies (computed using density functional theory, or DFT) of molecules with up to seven (7) non-H atoms.…”

Extended first-principles thermochemistry for the oxidation of titanium tetrachloride

“…The isodesmic and isogyric reaction classes were used in this work [44,45,46]. This methodology has been extensively tested and validated for test data sets including carbon, hydrogen, oxygen, chlorine and titanium [47,48].…”

“…The effect of the reference data (Table 1) on the accuracy of the method was assessed using a cross-validation technique. The method was described in full elsewhere [47,48] and is only summarised here. The standard enthalpy of formation was iteratively estimated for each species in the reference set, assuming that the enthalpy of the species under investigation is unknown.…”

“…The calculations in this work did not rely on a single EBR, but used a set of EBRs for each species. This has been shown to significantly improve the accuracy of calculations based on EBRs [47,48], providing some compensation from the use of isogyric reactions.…”

A systematic method to estimate and validate enthalpies of formation using error-cancelling balanced reactions

“…Ground state geometries and vibrational frequencies for all species used in this work were calculated using density functional theory (DFT) at the B97-1/6-311+G(d,p) level of theory, as per previous works [54,56,5,67]. This functional has shown to be accurate [68,69] and well suited for transition metal complexes [70,71,72].…”