Rational Design of Reliable Computational Protocols for Predicting Dielectric Constants of Gaseous Molecules

Abstract

A rapid prediction of the dielectric constants from a wide range of organic compounds is of paramount importance given the pressing need to find alternatives to SF6, one of the seven greenhouse gases. However, the availability of a universally applicable equation for predicting dielectric constants remains limited. This study endeavors to systematically develop a universal equation for predicting the dielectric constants of gaseous organic molecules in a systematic manner. The reliability of these newly developed equational protocols is evaluated through both quantitative (i.e., root-mean-squared deviation) and qualitative (i.e., Spearman’s rank correlation coefficient) analyses. Equational optimization of the traditionally unreliable Clausius–Mossotti equation highlights the critical role of selecting a suitable variable to be incorporated into an adapted Clausius–Mossotti equation, ultimately enhancing the predictive accuracy. Furthermore, it is revealed that the nature of the chosen variable has a more significant impact on prediction accuracy than the quantity of variables introduced. These findings shed light on the ongoing efforts of developing a dependable protocol for predicting not only dielectric constants but also other vital insulating properties, such as dielectric strength.

Highly Efficient Prediction of Dielectric and Electrochemical Properties of Organic Compounds Assisted by Artificial Intelligence

“…Also, note that distinct sets of physical/chemical parameters were introduced as features for characterizing the target variables (i.e., dielectric strength, dielectric constant, and electron affinity), respectively (Figure and Table ). Notably, the chosen features for the dielectric (i.e., dielectric strength and dielectric constant) and electrochemical (i.e., electron affinity) properties are derived not only from previously reported relationships in the literature but also from an intuitive grasp of their physical and chemical relevance to the target variables. ,, …”

“…As illustrated in Scheme , all of these organic compounds are collected from various resources, such as commercial databases and the literature. The equational protocols in conjunction with the DFT modeling approach developed in our prior studies are utilized to reliably calculate the dielectric properties (i.e., dielectric strength and dielectric constant) of the organic compounds for the constructed databases. ,, The electron affinity of each organic compound for the constructed database is computed through the investigation of the change in the DFT-calculated energy during the reduction of the compound, the reliability of which has been validated in prior studies . In addition to the target variables, various input variables, such as HOMO, LUMO, and polarizability, were calculated by DFT calculations.…”

“…More recently, computational methodologies have been leveraged to enhance the prediction accuracy of dielectric and electrochemical properties. By incorporating computationally predictable key parameters into existing equational protocols, researchers have achieved notable improvements in the prediction accuracy of dielectric strengths and dielectric constants. , However, the computational overhead associated with identifying these key parameters poses a challenge to efficiently predicting the dielectric and electrochemical properties across a wide range of organic compounds.…”