The utility of several nonhybrid density functional approximations (DFAs) is considered for the prediction of gas phase enthalpies of formation for a large set of 3d transition metal-containing molecules. Nonhybrid DFAs can model thermochemical values for 3d transition metal-containing molecules with accuracy comparable to that of hybrid functionals. The GAM-generalized gradient approximation (GGA); the TPSS, M06-L, and MN15-L meta-GGAs; and the Rung 3.5 PBE+ΠLDA(s) DFAs all give root-mean-square deviations below that of the widely used B3LYP hybrid. Modern nonhybrid DFAs continue to show utility for transition metal thermochemistry.

The exact exchange energy and its energy density are useful but computationally expensive ingredients in density functional approximations for Kohn–Sham density functional theory. We present detailed tests of some exact nonempirical upper bounds to exact exchange. These “Rung 3.5” upper bounds contract the Kohn–Sham one-particle density matrix with model density matrices used to construct semilocal model exchange holes and invoke the Cauchy–Schwarz inequality. The contraction automatically eliminates the computationally expensive long-range component of the exact exchange hole. Numerical tests show that the exchange upper bounds underestimate total exchange energies while predicting other properties with accuracy approaching standard hybrid approximations.

AbstractAtomic partial charges are widely used to predict reactivity. Partial charge alone is often insufficient: the carbons of benzene and cyclobutadiene, or those of diamond, graphene, and C60, possess nearly identical partial charges and very different reactivities. Our atomic overlap distance complements computed partial charges by measuring the size of orbital lobes that best overlap with the wavefunction around an atom. Compact, chemically stable atoms tend to have overlap distances smaller than chemically soft, unstable atoms. We show here how combining atomic charges and overlap distances captures trends in aromaticity, nucleophilicity, allotrope stability, and substituent effects. Applications to recent experiments in organic chemistry (counterintuitive Lewis base stabilization of alkenyl anions in anionic cyclization) and nanomaterials chemistry (facile doping of the central atom in Au7 hexagons) illustrate this combination's predictive power.

AbstractAtomic partial charges are widely used to predict reactivity. Partial charge alone is often insufficient: the carbons of benzene and cyclobutadiene, or those of diamond, graphene, and C60, possess nearly identical partial charges and very different reactivities. Our atomic overlap distance complements computed partial charges by measuring the size of orbital lobes that best overlap with the wavefunction around an atom. Compact, chemically stable atoms tend to have overlap distances smaller than chemically soft, unstable atoms. We show here how combining atomic charges and overlap distances captures trends in aromaticity, nucleophilicity, allotrope stability, and substituent effects. Applications to recent experiments in organic chemistry (counterintuitive Lewis base stabilization of alkenyl anions in anionic cyclization) and nanomaterials chemistry (facile doping of the central atom in Au7 hexagons) illustrate this combination's predictive power.

•QC-FPT uses quantum chemistry to aid experimentalists in de novo annotation of tandem mass spectra.•Automatically predicts whether a proposed parent ion structure is thermodynamically likely to produce observed fragment.•Focused combinatorial search on only the peaks seen in experiment.•Applied to annotate previously un-annotated peak in tandem mass spectra of arsenic-containing metabolite.•Applied to hypothetical experiment in pesticide identification.Tandem mass spectrometry is widely used to assign and distinguish chemical structures in proteomics, metabolomics, lipidomics, and many other areas. Spectral annotation remains a major bottleneck. Our ”Quantum Chemical Fragment Precursor Tests” (QC-FPT) approach brings the accuracy and generality of modern quantum chemistry to combinatorial-search-based computer-aided spectral annotation. QC-FPT takes as input the dominant fragment peaks from a particular experiment, and one or more chemically reasonable hypotheses for the precursor ion's three-dimensional structure. The algorithm automatically generates possible precursor ion fragmentations matching the target experimental peaks, uses quantum chemistry calculations (geometry optimization with gas-phase semiempirical or density functional theory calculations) to predict each neutral or charged fragment's structure and energy, and reports the thermodynamically feasible predicted fragment assignments. Applications demonstrate that QC-FPT recovers known spectral annotations, can handle multiple ionization and fragmentation methods and adducts, and can capture some fragment rearrangements. We apply QC-FPT to assign previously unassigned peaks in an experimental LC-ESI-MS2 spectrum of dimethylarsinothioyl glutathione (Yehiayan et al., Chem. Res. Toxicol. 2014, 27, 754–764), and to a hypothetical experiment distinguishing two isomeric candidates for an ”unknown” pesticide's experimental LC-ESI-MS2 spectrum. Our results suggest QC-FPT is a practical tool to sharpen and refine the chemical intuition of experimentalists engaged in the laborious process of annotating tandem mass spectra.Download high-res image (104KB)Download full-size image

The electron delocalization range function EDR(r⃗; d) (J. Chem. Phys. 2014, 141, 144104) quantifies the extent to which an electron at point r⃗ in a calculated wave function delocalizes over distance d. This work illustrates how atomic averages of the EDR, and plots of the EDR on molecule surfaces, provide a chemically intuitive picture of the sizes of occupied orbital lobes in different regions. We show how the surface and atomic delocalization distinguish aminophosphine ligand’s hard N and soft P lone pairs, distinguish the site preference for Ag+ cation binding to conjugated oligomers, and provide information that is different from and complementary to conjugation lengths. Applications to strong correlation and the prototropic tautomerization of phosphinylidenes R1R2HPO illustrates how the surface and atomic delocalization can work with other tools to provide a nuanced picture of reactivity.

Electrides are ionic solids whose anions are electrons confined to crystal voids. We show that our electron delocalization range function EDR(r⃗;d), which quantifies the extent to which an electron at point r⃗ in a calculated wave function delocalizes over distance d, provides useful insights into electrides. The EDR quantifies the characteristic delocalization length of electride electrons and provides a chemically intuitive real-space picture of the electrons’ distribution. It also gives a potential diagnostic for whether a given formula unit will form a solid electride at ambient pressure, quantifies the effects of electron–electron correlation on confined electrons’ interactions, and highlights analogies between covalent bonding and the interaction of interstitial quasi-atoms in high-pressure electrides. These results motivate adding the EDR to the toolbox of theoretical methods applied to electrides.

Anionic cyclization of o-alkynylbenzamides is proposed as a crucial step in many heterocycle syntheses. The cyclization can produce three products: Z-3-methylenisoindolin-1-one (Z-5-exo), E-3-methylenisoindolin-1-one (E-5-exo), and isoquinolinone (6-endo). Under base catalysis, the selectivity is generally poor. However, a copper-involved domino reaction of coupling and cyclization gives surprising selectivity for the thermodynamically disfavored Z-5-exo product (Org. Lett. 2009, 11, 1309–1312). We study the selectivity of anionic cyclization in the presence of K2CO3 and copper-l-proline, using surveys of the experimental literature and density functional theory (DFT) calculations. The o-alkynylbenzamide is predicted to be readily deprotonated by many bases, with subsequent cyclization via nucleophilic attack of the amide N– to alkynyl. In the absence of copper, endo-exo selectivity is predicted to arise from substituent effects, while Z/E selectivity is a sensitive function of the tautomerization rate of an alkenyl anion intermediate. Most importantly, we predict that the remarkable selectivity of the copper-involved reaction occurs because copper-l-proline “locks” the alkene anion intermediates into the initially formed Z-5-exo configuration. Calculations on other metals suggest that soft Lewis acid additives provide a potential route to improved regiocontrol of other anionic cyclizations.

Delocalized, solvated electrons are a topic of much recent interest. We apply the electron delocalization range EDR(;u) (J. Chem. Phys., 2014, 141, 144104) to quantify the extent to which a solvated electron at point in a calculated wavefunction delocalizes over distance u. Calculations on electrons in one-dimensional model cavities illustrate fundamental properties of the EDR. Mean-field calculations on hydrated electrons (H2O)n− show that the density-matrix-based EDR reproduces existing molecular-orbital-based measures of delocalization. Correlated calculations on hydrated electrons and electrons in lithium–ammonia clusters illustrates how electron correlation tends to move surface- and cavity-bound electrons onto the cluster or cavity surface. Applications to multiple solvated electrons in lithium–ammonia clusters provide a novel perspective on the interplay of delocalization and strong correlation central to lithium–ammonia solutions' concentration-dependent insulator-to-metal transition. The results motivate continued application of the EDR to simulations of delocalized electrons.

Condensation of a hydrazine-substituted s-triazine with an aldehyde or ketone yields an equivalent to the widely used, acid-labile acyl hydrazone. Hydrolysis of these hydrazones using a formaldehyde trap as monitored using HPLC reveals that triazine-substituted hydrazones are more labile than acetyl hydrazones at pH >5. The reactivity trends mirror that of the corresponding acetyl hydrazones, with hydrolysis rates increasing along the series (aromatic aldehyde < aromatic ketone < aliphatic ketone). Computational and experimental studies indicate a reversal in stability around the triazine pKa (pH ∼5). Protonation of the triazine moiety retards acid-catalyzed hydrolysis of triazinyl hydrazones in comparison to acetyl hydrazone analogues. This behavior supports mechanistic interpretations suggesting that resistance to protonation of the hydrazone N1 is the critical factor in affecting the reaction rate.

Phosphinylidene compounds R1R2P(O)H are important functionalities in organophosphorus chemistry and display prototropic tautomerism. Quantifying the tautomerization rate is paramount to understanding these compounds’ tautomerization behavior, which may impact their reactivities in various reactions. We report a combined theoretical and experimental study of the initial tautomerization rate of a range of phosphinylidene compounds. Initial tautomerization rates are found to decrease in the order H3PO2 > Ph2P(O)H > (PhO)2P(O)H > PhP(O) (OAlk)H > AlkP(O)(OAlk)H ≈ (AlkO)2P(O)H, where “Alk” denotes an alkyl substituent. The combination of computational investigations with experimental validation establishes a quantitative measure for the reactivity of various phosphorus compounds, as well as an accurate predictive tool.

This work presents a computational mechanistic study of the acid-catalyzed hydrolysis of lignin β-O-4 linkages in ionic liquid solvents. Model compound 2-hydroyxyethyl phenyl ether undergoes dehydration to vinyl phenyl ether followed by hydrolysis to phenol and “Hibbert's ketones”. Larger model compound α-hydroxy-phenethyl phenyl ether illustrates an E1 dehydration mechanism involving resonance-stabilized carbocations. Continuum models for ionic liquid solvents indicate that solvation can significantly affect the reaction rates. The tested continuum ionic liquid solvents give similar results, and differ significantly from continuum organic solvents with comparable dielectric constants. The acidic ionic liquid cation 1-H-3-methylimidazolium has lower predicted catalytic activity than hydronium or HCl, consistent with the former's relatively small acid dissociation constant. Calculations with dispersion-corrected density functionals give similar behavior. Calculations on Lewis acidic metal chlorides used experimentally for lignin hydrolysis suggest that the metal chloride may participate in the initial dehydration. These results provide a baseline for future studies of improved hydrolysis catalysts.

Gold’s structure-dependent catalytic activity motivates the study of reactions on a range of gold nanostructures. Electronic structure methods used to model gold catalysis should be capable of treating atoms, clusters, nanostructures, and surfaces on an equal theoretical footing. We extend our previous density functional theory (DFT) studies of a model reaction, H2 adsorption and dissociation on unsupported Au3 clusters [J. Phys. Chem. C 2013, 117, 7487], to larger clusters, quasi-one-dimensional nanowires and nanoribbons, and surfaces. We focus on trends in DFT predictions made using various approximate exchange-correlation functionals. Most functionals predict qualitatively reasonable trends, i.e., decreasing adsorption energies and increasing dissociation barriers with increasing Au coordination number. However, significant quantitative differences motivate continued exploration of methods beyond the generalized gradient approximation.

Substituent effects on the π–π interactions of aromatic rings are a topic of much recent debate. Real substituents give a complicated combination of inductive, resonant, dispersion, and other effects. To help partition these effects, we present calculations on fictitious “pure” σ donor/acceptor substituents, hydrogen atoms with nuclear charges other than 1. “Pure” σ donors with nuclear charge <1 weaken π–π stacking in the sandwich benzene dimer. This result is consistent with the electrostatic model of Hunter and Sanders, and different from real substituents. Calculated inductive effects are largely additive and transferable, consistent with a local direct interaction model. A second series of fictitious substituents, neutral hydrogen atoms with an artificially broadened nuclear charge distribution, give similar trends though with reduced additivity. These results provide an alternative perspective on substituent effects in noncovalent interactions.

Regioregular poly(3-alkylthiophene)s are widely used in organic electronics applications such as solar cells and field effect transistors. Nickel, palladium, and platinum diphenylphosphinoethane complexes were tested as catalysts for the Grignard metathesis (GRIM) polymerization of 2,5-dibromo-3-hexylthiophene and 2-bromo-5-iodo-3-hexylthiophene. Nickel-mediated polymerization generated regioregular, low-polydispersity poly(3-hexylthiophene) with well-defined molecular weight consistent with a “living” chain-growth mechanism. By contrast, palladium-mediated polymerization proceeded by a step-growth mechanism and generated polymers with less than 80% head-to-tail couplings. Platinum-mediated polymerization gave very low molecular weight products. Kinetic and computational results suggested that the nickel catalyst acts as an initiator and remains associated with the growing polymer chain, while palladium dissociates from the growing chain. Computational and experimental evidence was provided for various side reactions of dissociated Pd(0) catalyst, which could yield a step growth mechanism and lower regioirregularity.

This work benchmarks density functional theory, with several different exchange-correlation functionals, for prediction of isotropic one-bond phosphorus–hydrogen NMR spin–spin coupling constants (SSCCs). Our test set consists of experimental SSCCs from 30 diverse molecules representing multiple phosphorus bonding environments. The results suggest the importance of a balance between the choice of correlation functional and the admixture of nonlocal exchange. Overall, standard DFT methods appear to suffice for usefully accurate predictions of 31P–1H SSCCs.

Simulations of surface chemistry often use density functional theory with generalized gradient approximations (GGAs) for the exchange-correlation functional. GGAs have well-known limitations for gas-phase chemistry, including underestimated reaction barriers, and are largely superseded by meta-GGAs and hybrids. Our simulations of O and Li adatoms on graphene add to a growing body of evidence that GGAs have similar limitations on surfaces and that meta-GGAs and screened hybrids are computationally feasible for such systems. Meta-GGAs and screened hybrids systematically improve accuracy, just as they do for gas-phase chemistry, motivating their continued exploration in surface chemistry.

This work proposes that zigzag graphene and graphane nanoribbons with alternating Lewis-acid and Lewis-base edge substituents can act as “frustrated Lewis pairs” (FLPs), with potential applications in heterogeneous catalysis and sensing. The nanoribbon scaffold prevents (frustrates) intramolecular Lewis pairing. Edge substitutents prevent inter-ribbon pairing and tune the Lewis acidity/basicity. Computational evidence suggests that graphene and graphane FLPs are synthetically feasible semiconductors, with heterolytic H2 splitting activity potentially comparable to molecular FLPs. These new materials show potential for extending FLPs to heterogeneous catalysis, coupling catalytic, electronic, and sensing activity on the nanoscale.

Screened hybrid exchange-correlation (XC) density functionals incorporating short-range exact exchange aid application of hybrid density functional theory to solids and surfaces. We explore screened hybrid XC functionals for a prototypical surface reaction, namely, adsorption and dissociation of ammonia on silicon. Screened hybrids are found to improve upon standard semilocal functionals for the dissociation barrier on Si9H12, reproducing accurate complete basis set extrapolated CCSD(T) results. Similar trends are found for realistic periodic Si(100)-2 × 2 surfaces. Screened hybrids also better reproduce experimental results for the relative barriers to different dissociation pathways. While the tested hybrid functionals tend to overestimate molecular adsorption energies, their good performance for kinetics motivates further exploration of screened exchange in surface chemistry.

Palladium-catalyzed cross-couplings of secondary alkyls are promising tools for the stereoselective formation of carbon–carbon bonds. We report a computational mechanistic study of the stereoselective Suzuki coupling between (S)-2-chloropropanenitrile and phenylboronic acid, following a recent experimental report on related α-cyanohydrin triflates ( J. Am. Chem. Soc. 2010, 132, 2524). Added Lewis base helps accelerate SN2 oxidative addition, leading to the experimentally observed inversion of configuration. Undesired β-hydride elimination side reactions are reduced by the activating cyano group’s inductive effects, by cyano-PdII coordination, and by excess boronic acid. The catalyst ligand’s trans influence and steric bulk also affect the rate of β-hydride elimination, suggesting design rules for alkyl cross-coupling ligands.

Dissolution of lignocellulose in ionic liquids is a promising route to synthesizing fuels and chemical feedstocks from woody plant materials. While cellulose dissolution is well-understood, less is known about the differences between ionic liquids' interactions with cellulose vs.lignin. This work uses dispersion-corrected density functional theory (DFT-D) to model the interactions of imidazolium chloride ionic liquid anions and cations with (1,4)-dimethoxy-β-D-glucopyranose and 1-(4-methoxyphenyl)-2-methoxyethanol as models for cellulose and the lignin polyphenol, respectively. The cellulose model preferentially interacts with Cl−, confirming previous experimental and theoretical studies. However, the lignin model has significant π-stacking and hydrogen bonding interactions with imidazolium cation. These results are robust to changes in the computational details, and suggest that the ionic liquid cations play important roles in tuning the relative solubilities of lignin and cellulose. Calculations predict that the extended π-systems of benzimidazolium ionic liquids yield stronger interactions with lignin, showing potential for improved lignocellulose solvents.

The Suzuki–Miyaura cross-coupling reaction is an important route to forming carbon–carbon bonds. Suzuki coupling of secondary alkyls containing β-hydrogens is challenging, due in part to a competing and undesired β-hydride elimination. We perform density functional electronic structure calculations on model compounds to study the selectivity of Suzuki coupling of secondary alkyl boranes. Results indicate that the rate and selectivity of the desired reductive elimination are strongly influenced by how the reactants and ligand interact with a coordinatively unsaturated PdII intermediate. Agostic interactions between PdII and reactant β-hydrogens provide facile routes to β-hydride elimination, while coordination of electron-donating reactant groups to PdII slows the reductive elimination. The bulky ligands used in typical Suzuki couplings, such as the SPhos dialkylbiarylphosphine, appear to block both types of undesired interaction while stabilizing the reductive elimination transition state.

Dative bonds to substituted boranes represent a challenge for the approximate exchange-correlation functionals typically used in density functional theory (DFT). Accurately modeling these bonds with DFT has usually required highly parametrized functionals, large admixtures of exact exchange, or computationally expensive double hybrids. This work shows that the nonempirical semilocal PBEsol functional, and the nonempirical semilocal PBE and TPSS functionals augmented with empirical interatomic dispersion corrections, accurately treat several representative problems in dative bonding. These methods typically surpass the MPW1K “kinetics” global hybrid previously recommended for dative bonds. This work also provides additional insights into the accuracy of the parametrized M06 functionals and indicates some deficiencies of the B97-D functional relative to PBE-D and TPSS-D. Applications to frustrated Lewis pairs illustrate the potential of these methods.

Highlights•The electron delocalization range function EDR quantifies the characteristic delocalization of surface and bulk F-centers, and other anionic surface defects.•Real-space plots of the EDR highlight the location of the trapped electron.•The density-matrix-based EDR quantifies the effects of electron correlation and surface morphology on the trapped electron.•EDR studies of nitrogen activation by a magnesium oxide reverse corner defect highlights how the trapped electron localizes into the adsorbed molecule's antibonding orbital.Electrons trapped in ionic crystal defects form color centers (F-centers) important in surface science, catalysis, and optoelectronic devices. We apply the electron delocalization range function (EDR) to quantify the delocalization of surface and bulk F-centers. The EDR uses computed one-particle density matrices to quantify “delocalization lengths” capturing the characteristic size of orbital lobes. Ab initio cluster model calculations confirm that the delocalization lengths of bulk alkali halide F-centers scale with the size of the anion vacancy. Calculations on magnesium oxide surface Fs and Fs+ centers, as well as other anionic surface defects, show how the trapped electrons' delocalization depends on the defect morphology, defect occupancy, and the approximate treatment of electron correlation. Application to N2 activation by anionic surface defects illustrate how the trapped electron localizes into the adsorbed molecule's unoccupied orbitals. The results confirm that the EDR provides a useful tool for understanding the chemistry of surface- and bulk-trapped electrons.Graphical abstract

Dissolution of lignocellulose in ionic liquids is a promising route to synthesizing fuels and chemical feedstocks from woody plant materials. While cellulose dissolution is well-understood, less is known about the differences between ionic liquids' interactions with cellulose vs.lignin. This work uses dispersion-corrected density functional theory (DFT-D) to model the interactions of imidazolium chloride ionic liquid anions and cations with (1,4)-dimethoxy-β-D-glucopyranose and 1-(4-methoxyphenyl)-2-methoxyethanol as models for cellulose and the lignin polyphenol, respectively. The cellulose model preferentially interacts with Cl−, confirming previous experimental and theoretical studies. However, the lignin model has significant π-stacking and hydrogen bonding interactions with imidazolium cation. These results are robust to changes in the computational details, and suggest that the ionic liquid cations play important roles in tuning the relative solubilities of lignin and cellulose. Calculations predict that the extended π-systems of benzimidazolium ionic liquids yield stronger interactions with lignin, showing potential for improved lignocellulose solvents.

Delocalized, solvated electrons are a topic of much recent interest. We apply the electron delocalization range EDR(;u) (J. Chem. Phys., 2014, 141, 144104) to quantify the extent to which a solvated electron at point in a calculated wavefunction delocalizes over distance u. Calculations on electrons in one-dimensional model cavities illustrate fundamental properties of the EDR. Mean-field calculations on hydrated electrons (H2O)n− show that the density-matrix-based EDR reproduces existing molecular-orbital-based measures of delocalization. Correlated calculations on hydrated electrons and electrons in lithium–ammonia clusters illustrates how electron correlation tends to move surface- and cavity-bound electrons onto the cluster or cavity surface. Applications to multiple solvated electrons in lithium–ammonia clusters provide a novel perspective on the interplay of delocalization and strong correlation central to lithium–ammonia solutions' concentration-dependent insulator-to-metal transition. The results motivate continued application of the EDR to simulations of delocalized electrons.

This work presents a computational mechanistic study of the acid-catalyzed hydrolysis of lignin β-O-4 linkages in ionic liquid solvents. Model compound 2-hydroyxyethyl phenyl ether undergoes dehydration to vinyl phenyl ether followed by hydrolysis to phenol and “Hibbert's ketones”. Larger model compound α-hydroxy-phenethyl phenyl ether illustrates an E1 dehydration mechanism involving resonance-stabilized carbocations. Continuum models for ionic liquid solvents indicate that solvation can significantly affect the reaction rates. The tested continuum ionic liquid solvents give similar results, and differ significantly from continuum organic solvents with comparable dielectric constants. The acidic ionic liquid cation 1-H-3-methylimidazolium has lower predicted catalytic activity than hydronium or HCl, consistent with the former's relatively small acid dissociation constant. Calculations with dispersion-corrected density functionals give similar behavior. Calculations on Lewis acidic metal chlorides used experimentally for lignin hydrolysis suggest that the metal chloride may participate in the initial dehydration. These results provide a baseline for future studies of improved hydrolysis catalysts.

Regioregular poly(3-alkylthiophene)s are widely used in organic electronics applications such as solar cells and field effect transistors. Nickel, palladium, and platinum diphenylphosphinoethane complexes were tested as catalysts for the Grignard metathesis (GRIM) polymerization of 2,5-dibromo-3-hexylthiophene and 2-bromo-5-iodo-3-hexylthiophene. Nickel-mediated polymerization generated regioregular, low-polydispersity poly(3-hexylthiophene) with well-defined molecular weight consistent with a “living” chain-growth mechanism. By contrast, palladium-mediated polymerization proceeded by a step-growth mechanism and generated polymers with less than 80% head-to-tail couplings. Platinum-mediated polymerization gave very low molecular weight products. Kinetic and computational results suggested that the nickel catalyst acts as an initiator and remains associated with the growing polymer chain, while palladium dissociates from the growing chain. Computational and experimental evidence was provided for various side reactions of dissociated Pd(0) catalyst, which could yield a step growth mechanism and lower regioirregularity.