All Publications

Correcting pidelocalisation errors in conformational energies using densitycorrected DFT, with application to crystal polymorphs
MOLECULAR PHYSICS
2022
View details for DOI 10.1080/00268976.2022.2138789
View details for Web of Science ID 000873334300001

Detection and Correction of Delocalization Errors for Electron and Hole Polarons Using DensityCorrected DFT.
The journal of physical chemistry letters
2022: 52755284
Abstract
Modeling polaron defects is an important aspect of computational materials science, but the description of unpaired spins in density functional theory (DFT) often suffers from delocalization error. To diagnose and correct the overdelocalization of spin defects, we report an implementation of densitycorrected (DC)DFT and its analytic energy gradient. In DCDFT, an exchangecorrelation functional is evaluated using a HartreeFock density, thus incorporating electron correlation while avoiding selfinteraction error. Results for an electron polaron in models of titania and a hole polaron in Aldoped silica demonstrate that geometry optimization with semilocal functionals drives significant structural distortion, including the elongation of several bonds, such that subsequent singlepoint calculations with hybrid functionals fail to afford a localized defect even in cases where geometry optimization with the hybrid functional does localize the polaron. This has significant implications for traditional workflows in computational materials science, where semilocal functionals are often used for structure relaxation. DCDFT calculations provide a mechanism to detect situations where delocalization error is likely to affect the results.
View details for DOI 10.1021/acs.jpclett.2c01187
View details for PubMedID 35674719

Software for the frontiers of quantum chemistry: An overview of developments in the QChem 5 package
JOURNAL OF CHEMICAL PHYSICS
2021; 155 (8)
View details for DOI 10.1063/5.0055522
View details for Web of Science ID 000687352200007

Hidden Hemibonding in the Aqueous Hydroxyl Radical
JOURNAL OF PHYSICAL CHEMISTRY LETTERS
2021; 12 (33): 80538060
Abstract
The existence of a twocenter, threeelectron hemibond in the first solvation shell of •OH(aq) has long been a matter of debate. The hemibond manifests in ab initio molecular dynamics simulations as a smallr feature in the oxygen radial distribution function (RDF) for H2O···•OH, but that feature disappears when semilocal density functionals are replaced with hybrids, suggesting a selfinteraction artifact. Using periodic simulations at the PBE0+D3 level, we demonstrate that the hemibond is actually still present (as evidenced by delocalization of the spin density) but is obscured by the hydrogenbonded feature in the RDF due to a slight elongation of the hemibond. Computed electronic spectra for •OH(aq) are in excellent agreement with experiment and confirm that hemibondlike configurations play an outsized role in the spectroscopy due to an intense chargetransfer transition that is strongly attenuated in hydrogenbonded configurations. Apparently, 25% exact exchange (as in PBE0) is insufficient to eliminate delocalization of unpaired spins.
View details for DOI 10.1021/acs.jpclett.1c02283
View details for Web of Science ID 000692014200020
View details for PubMedID 34406021

Role of hemibonding in the structure and ultraviolet spectroscopy of the aqueous hydroxyl radical
PHYSICAL CHEMISTRY CHEMICAL PHYSICS
2020; 22 (47): 2782927844
Abstract
The presence of a hemibond in the local solvation structure of the aqueous hydroxyl radical has long been debated, as its appearance in ab initio simulations based on density functional theory is sensitive to selfinteraction error (favoring a twocenter, threeelectron hemibond) but also to finitesize effects. Simulations reported here use a mixed quantum mechanics/molecular mechanics (QM/MM) framework in a very large periodic simulation cell, in order to avoid finitesize artifacts and to facilitate testing of various density functionals, in order to probe the effects of delocalization error. The preponderance of hemibonded structures predicted by generalized gradient approximations persists in simulations using the hybrid functionals B3LYP and PBE0, but is reduced to a minor population if the fraction of exact exchange is increased to 50%. The hemibonded population is also small in simulations employing the longrange corrected functional LRCωPBE. Electronic spectra are computed using timedependent density functional theory, and from these calculations emerges a consensus picture in which hemibonded configurations play an outsized role in the absorption spectrum, even when present as a minority species. An intense 1b2(H2O) → 2pπ(˙OH) chargetransfer transition in hemibonded configurations of the radical proves to be responsible for an absorption feature at 230 nm that is strongly shifted with respect to the gasphase absorption at 307 nm, but this intense feature is substantially diminished in aqueous geometries where the hemibond is absent. Although not yet sufficient to quantitatively establish the population of hemibonded ˙OH(aq), these simulations do suggest that its presence is revealed by the strongly shifted ultraviolet absorption spectrum of the aqueous radical.
View details for DOI 10.1039/d0cp05216g
View details for Web of Science ID 000599460800034
View details for PubMedID 33245735

Ab Initio Investigation of the Resonance Raman Spectrum of the Hydrated Electron
JOURNAL OF PHYSICAL CHEMISTRY B
2019; 123 (38): 80748085
Abstract
According to the conventional picture, the aqueous or "hydrated" electron, e(aq), occupies an excluded volume (cavity) in the structure of liquid water. However, simulations with certain oneelectron models predict a more delocalized spin density for the unpaired electron, with no distinct cavity structure. It has been suggested that only the latter (noncavity) structure can explain the hydrated electron's resonance Raman spectrum, although this suggestion is based on calculations using empirical frequency maps developed for neat liquid water, not for e(aq). Allelectron ab initio calculations presented here demonstrate that both cavity and noncavity models of e(aq) afford significant redshifts in the OH stretching region. This effect is nonspecific and arises due to electron penetration into frontier orbitals of the water molecules. Only the conventional cavity model, however, reproduces the splitting of the HOD bend (in isotopically mixed water) that is observed experimentally and arises due to the asymmetric environments of the hydroxyl moieties in the electron's first solvation shell. We conclude that the cavity model of e(aq) is more consistent with the measured resonance Raman spectrum than is the delocalized, noncavity model, despite previous suggestions to the contrary. Furthermore, calculations with hybrid density functionals and with HartreeFock theory predict that noncavity liquid geometries afford only unbound (continuum) states for an extra electron, whereas in reality this energy level should lie more than 3 eV below vacuum level. As such, the noncavity model of e(aq) appears to be inconsistent with available vibrational spectroscopy, photoelectron spectroscopy, and quantum chemistry.
View details for DOI 10.1021/acs.jpcb.9b04895
View details for Web of Science ID 000488335100014
View details for PubMedID 31442044

Variational Formulation of the Generalized ManyBody Expansion with SelfConsistent Charge Embedding: Simple and Correct Analytic Energy Gradient for FragmentBased ab Initio Molecular Dynamics
JOURNAL OF PHYSICAL CHEMISTRY LETTERS
2019; 10 (14): 38773886
Abstract
The manybody expansion (MBE) and its extension to overlapping fragments, the generalized (G)MBE, constitute the theoretical basis for most fragmentbased approaches for largescale quantum chemistry. We reformulate the GMBE for use with embedding charges determined selfconsistently from the fragment wave functions, in a manner that preserves the variational nature of the underlying selfconsistent field method. As a result, the analytic gradient retains the simple "sum of fragment gradients" form that is often assumed in practice, sometimes incorrectly. This obviates (without approximation) the need to solve coupledperturbed equations, and we demonstrate stable, fragmentbased ab initio molecular dynamics simulations using this technique. Energy conservation fails when chargeresponse contributions to the Fock matrix are neglected, even while geometry optimizations and vibrational frequency calculations may yet be accurate. Stable simulations can be recovered by means of straightforward modifications introduced here, providing a general paradigm for fragmentbased ab initio molecular dynamics.
View details for DOI 10.1021/acs.jpclett.9b01214
View details for Web of Science ID 000476694300009
View details for PubMedID 31251619

Analytic gradient for the QM/MMEwald method using charges derived from the electrostatic potential: Theory, implementation, and application to ab initio molecular dynamics simulation of the aqueous electron
JOURNAL OF CHEMICAL PHYSICS
2019; 150 (14): 144115
Abstract
We report an implementation of periodic boundary conditions for mixed quantum mechanics/molecular mechanics (QM/MM) simulations, in which atomic partial charges are used to represent periodic images of the QM region. These charges are incorporated into the Fock matrix in a manner that preserves the variational nature of the selfconsistent field procedure, and their interactions with the MM charges are summed using the conventional Ewald technique. To ensure that the procedure is stable in arbitrary basis sets, the atomic charges are derived by leastsquares fit to the electrostatic potential generated by the QM region. We formulate and implement analytic energy gradients for the QM/MMEwald method and demonstrate that stable molecular dynamics simulations are thereby obtained. As a proofofconcept application, we perform QM/MM simulations of a hydrated electron in bulk liquid water at the level of HartreeFock theory plus empirical dispersion. These simulations demonstrate that the "cavity model" of the aqueous electron, in which the spin density of the anionic defect is localized within an excluded volume in the liquid, is stable at room temperature on a time scale of at least several picoseconds. These results validate cavityforming pseudopotential models of e(aq) that have previously been derived from staticexchange HartreeFock calculations, and cast doubt upon whether noncavityforming pseudopotentials are faithful to the underlying HartreeFock calculation from which they were obtained.
View details for DOI 10.1063/1.5089673
View details for Web of Science ID 000464451300018
View details for PubMedID 30981237