Jorge Vega Martín, Daniel G. Trabada, Diego Soler-Polo, C. Roldán-Piñero, Jesús I. Mendieta-Moreno, José Ortega
{"title":"Partial Atomic Charges, Dipole Moments and Electrostatic Potentials","authors":"Jorge Vega Martín, Daniel G. Trabada, Diego Soler-Polo, C. Roldán-Piñero, Jesús I. Mendieta-Moreno, José Ortega","doi":"10.1002/jcc.70496","DOIUrl":null,"url":null,"abstract":"<p>Partial atomic charges are indispensable tools in chemistry, materials science, and atomic-scale modeling of materials, yet they are not uniquely defined. In this work, we introduce a new charge-assignment model—the Dipole-Projected Overlap Density (DPOD) charges—and assess it systematically against several other schemes: Mulliken (M), Löwdin (L), electrostatic-potential-fitted (ESP), Weighted Löwdin (WL), also introduced here, and Mulliken-Dipole (MD) on which DPOD builds. Using Kohn–Sham DFT electron densities for a diverse test set of 37 molecules, we evaluate the ability of each scheme to: (i) reproduce the molecular dipole moment, (ii) accurately describe the electrostatic potential in regions relevant for intermolecular interactions, and (iii) yield chemically meaningful and transferable charge distributions. The DPOD model, which incorporates both monopole and bond-directed dipolar contributions from overlap densities and distributes intra-atomic dipoles locally, is shown to reproduce the molecular dipole moment by construction while providing a robust and chemically consistent description of atomic charges. Across the entire test set, DPOD and ESP charges are the only methods that consistently yield accurate dipole moments; however, ESP charges often suffer from weakly determined fits leading to chemically unreasonable values, particularly in hydrocarbons. DPOD charges, in contrast, remain stable across chemical environments and achieve good electrostatic-potential description. Surprisingly, WL charges, which are related to Natural Population Analysis, also display reasonably good electrostatic-potential performance despite poor dipole-moment accuracy. Overall, our results demonstrate that the DPOD scheme combines strong physical grounding, excellent dipole-moment reproduction, good electrostatic-potential accuracy, and chemical interpretability, making it a promising and reliable option for atomistic simulations, force-field parameterization, and studies requiring chemically meaningful partial atomic charges.</p>","PeriodicalId":188,"journal":{"name":"Journal of Computational Chemistry","volume":"47 25","pages":""},"PeriodicalIF":2.9000,"publicationDate":"2026-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13594691/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Computational Chemistry","FirstCategoryId":"92","ListUrlMain":"https://onlinelibrary.wiley.com/doi/10.1002/jcc.70496","RegionNum":3,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"CHEMISTRY, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0
Abstract
Partial atomic charges are indispensable tools in chemistry, materials science, and atomic-scale modeling of materials, yet they are not uniquely defined. In this work, we introduce a new charge-assignment model—the Dipole-Projected Overlap Density (DPOD) charges—and assess it systematically against several other schemes: Mulliken (M), Löwdin (L), electrostatic-potential-fitted (ESP), Weighted Löwdin (WL), also introduced here, and Mulliken-Dipole (MD) on which DPOD builds. Using Kohn–Sham DFT electron densities for a diverse test set of 37 molecules, we evaluate the ability of each scheme to: (i) reproduce the molecular dipole moment, (ii) accurately describe the electrostatic potential in regions relevant for intermolecular interactions, and (iii) yield chemically meaningful and transferable charge distributions. The DPOD model, which incorporates both monopole and bond-directed dipolar contributions from overlap densities and distributes intra-atomic dipoles locally, is shown to reproduce the molecular dipole moment by construction while providing a robust and chemically consistent description of atomic charges. Across the entire test set, DPOD and ESP charges are the only methods that consistently yield accurate dipole moments; however, ESP charges often suffer from weakly determined fits leading to chemically unreasonable values, particularly in hydrocarbons. DPOD charges, in contrast, remain stable across chemical environments and achieve good electrostatic-potential description. Surprisingly, WL charges, which are related to Natural Population Analysis, also display reasonably good electrostatic-potential performance despite poor dipole-moment accuracy. Overall, our results demonstrate that the DPOD scheme combines strong physical grounding, excellent dipole-moment reproduction, good electrostatic-potential accuracy, and chemical interpretability, making it a promising and reliable option for atomistic simulations, force-field parameterization, and studies requiring chemically meaningful partial atomic charges.
期刊介绍:
This distinguished journal publishes articles concerned with all aspects of computational chemistry: analytical, biological, inorganic, organic, physical, and materials. The Journal of Computational Chemistry presents original research, contemporary developments in theory and methodology, and state-of-the-art applications. Computational areas that are featured in the journal include ab initio and semiempirical quantum mechanics, density functional theory, molecular mechanics, molecular dynamics, statistical mechanics, cheminformatics, biomolecular structure prediction, molecular design, and bioinformatics.