Interpolating moving least-squares methods for fitting potential energy surfaces: a strategy for efficient automatic data point placement in high dimensions

An accurate and efficient method for automated molecular global potential energy surface (PES) construction and fitting is demonstrated. An interpolating moving least-squares (IMLS) method is developed with the flexibility to fit various ab initio data: (1) energies, (2) energies and gradients, or (3) energies, gradients, and Hessian data. The method is automated and flexible so that a PES can be optimally generated for trajectories, spectroscopy, or other applications. High efficiency is achieved by employing local IMLS in which fitting coefficients are stored at a limited number of expansion points, thus eliminating the need to perform weighted least-squares fits each time the potential is evaluated. An automatic point selection scheme based on the difference in two successive orders of IMLS fits is used to determine where new ab initio data need to be calculated for the most efficient fitting of the PES. A simple scan of the coordinate is shown to work well to identify these maxima in one dimension, but this search strategy scales poorly with dimension. We demonstrate the efficacy of using conjugate gradient minimizations on the difference surface to locate optimal data point placement in high dimensions. Results that are indicative of the accuracy, efficiency, and scalability are presented for a one-dimensional model potential (Morse) as well as for three-dimensional (HCN), six-dimensional (HOOH), and nine-dimensional (CH4) molecular PESs.     

Interpolating moving least-squares methods for fitting potential energy surfaces: computing high-density potential energy surface data from low-density ab initio data points

A highly accurate and efficient method for molecular global potential energy surface (PES) construction and fitting is demonstrated. An interpolating-moving-least-squares (IMLS)-based method is developed using low-density ab initio Hessian values to compute high-density PES parameters suitable for accurate and efficient PES representation. The method is automated and flexible so that a PES can be optimally generated for classical trajectories, spectroscopy, or other applications. Two important bottlenecks for fitting PESs are addressed. First, high accuracy is obtained using a minimal density of ab initio points, thus overcoming the bottleneck of ab initio point generation faced in applications of modified-Shepard-based methods. Second, high efficiency is also possible (suitable when a huge number of potential energy and gradient evaluations are required during a trajectory calculation). This overcomes the bottleneck in high-order IMLS-based methods, i.e., the high cost/accuracy ratio for potential energy evaluations. The result is a set of hybrid IMLS methods in which high-order IMLS is used with low-density ab initio Hessian data to compute a dense grid of points at which the energy, Hessian, or even high-order IMLS fitting parameters are stored. A series of hybrid methods is then possible as these data can be used for neural network fitting, modified-Shepard interpolation, or approximate IMLS. Results that are indicative of the accuracy, efficiency, and scalability are presented for one-dimensional model potentials as well as for three-dimensional (HCN) and six-dimensional (HOOH) molecular PESs.     

Interpolating moving least-squares methods for fitting potential energy surfaces: an application to the H2CN unimolecular reaction

Classical trajectories have been used to compute rates for the unimolecular reaction H2CN-->H+HCN on a fitted ab initio potential energy surface (PES). The ab initio energies were obtained from CCSD(T)/aug-cc-pvtz electronic structure calculations. The ab initio energies were fitted by the interpolating moving least-squares (IMLS) method. This work continues the development of the IMLS method for producing ab initio PESs for use in molecular dynamics simulations of many-atom systems. A dual-level scheme was used in which the preliminary selection of data points was done using a low-level theory and the points used for fitting the final PES were obtained at the desired higher level of theory. Classical trajectories were used on various low-level IMLS fits to tune the fit to the unimolecular reaction under study. Procedures for efficiently picking data points, selecting basis functions, and defining cutoff limits to exclude distant points were investigated. The accuracy of the fitted PES was assessed by comparing interpolated values of quantities to the corresponding ab initio values. With as little as 330 ab initio points classical trajectory rate constants were converged to 5%-10% and the rms error over the six-dimensional region sampled by the trajectories was a few tenths of a kcal/mol.     

Interpolating moving least-squares methods for fitting potential-energy surfaces: further improvement of efficiency via cutoff strategies

In standard applications of interpolating moving least squares (IMLS) for fitting a potential-energy surface (PES), all available ab initio points are used. Because remote ab initio points negligibly influence IMLS accuracy and increase IMLS time-to-solution, we present two methods to locally restrict the number of points included in a particular fit. The fixed radius cutoff (FRC) method includes ab initio points within a hypersphere of fixed radius. The density adaptive cutoff (DAC) method includes points within a hypersphere of variable radius depending on the point density. We test these methods by fitting a six-dimensional analytical PES for hydrogen peroxide. Both methods reduce the IMLS time-to-solution by about an order of magnitude relative to that when no cutoff method is used. The DAC method is more robust and efficient than the FRC method.     

Interpolating moving least-squares methods for fitting potential energy surfaces: Improving efficiency via local approximants

The local interpolating moving least-squares (IMLS) method for constructing potential energy surfaces is investigated. The method retains the advantageous features of the IMLS approach in that the ab initio derivatives are not required and high degree polynomials can be used to provide accurate fits, while at the same time it is much more efficient than the standard IMLS approach because the least-squares solutions need to be calculated only once at the data points. Issues related to the implementation of the local IMLS method are investigated and the accuracy is assessed using HOOH as a test case. It is shown that the local IMLS method is at the same level of accuracy as the standard IMLS method. In addition, the scaling of the method is found to be a power law as a function of number of data points N, N(-q). The results suggest that when fitting only to the energy values for a d-dimensional system by using a Qth degree polynomial the power law exponent q approximately Qd when the energy range fitted is large (e.g., E<100 kcalmol for HOOH), and q>Qd when the energy range fitted is smaller (E<30 kcalmol) and the density of data points is higher. This study demonstrates that the local IMLS method provides an efficient and accurate means for constructing potential energy surfaces.     

Interpolating moving least-squares methods for fitting potential energy surfaces: applications to classical dynamics calculations

As a continuation of our efforts to develop efficient and accurate interpolating moving least-squares (IMLS) methods for generating potential energy surfaces, we carry out classical trajectories and compute kinetics properties on higher degree IMLS surfaces. In this study, we have investigated the choice of coordinate system, the range of points (i.e., the cutoff radius) used in fitting, and strategies for selections of data points and basis elements. We illustrate and test the method by applying it to hydrogen peroxide (HOOH). In particular, reaction rates for the O-O bond breaking in HOOH are calculated on fitted surfaces using the classical trajectory approach to test the accuracy of the IMLS method for providing potentials for dynamics calculations.     

Gradient incorporation in one-dimensional applications of interpolating moving least-squares methods for fitting potential energy surfaces

We present several approaches to use gradients in higher degree interpolating moving least squares (IMLS) methods for representing a potential energy surface (PES). General procedures are developed to obtain smooth approximations of the PES and its derivatives from quasi-uniform sets of energy and gradient data points. These methods are illustrated and analyzed for the Morse oscillator and a 1-D slice of the ground-state PES for the HCO radical computed using density functional theory. Variations in the IMLS fits with the number and distribution of points and the degree of the polynomial fitting basis set are examined. We determine the effects of gradient inclusion on the accuracy of the IMLS values of the energy, first and second derivatives for two 1-D test cases. Gradient inclusion reduces the number of data points required by up to 40%.     

New insight into the gas-phase bimolecular self-reaction of the HOO radical

The singlet and triplet potential energy surfaces (PESs) for the gas-phase bimolecular self-reaction of HOO*, a key reaction in atmospheric environments, have been investigated by means of quantum-mechanical electronic structure methods (CASSCF and CASPT2). All the reaction pathways on both PESs consist of a first step involving the barrierless formation of a prereactive doubly hydrogen-bonded complex, which is a diradical species lying about 8 kcal/mol below the energy of the reactants at 0 K. The lowest energy reaction pathway on both PESs is the degenerate double hydrogen exchange between the HOO* moieties of the prereactive complex via a double proton transfer mechanism involving an energy barrier of only 1.1 kcal/mol for the singlet and 3.3 kcal/mol for the triplet at 0 K. The single H-atom transfer between the two HOO* moieties of the prereactive complex (yielding HOOH + O2) through a pathway keeping a planar arrangement of the six atoms involves a conical intersection between either two singlet or two triplet states of A' and A" symmetries. Thus, the lowest energy reaction pathway occurs via a nonplanar cisoid transition structure with an energy barrier of 5.8 kcal/mol for the triplet and 17.5 kcal/mol for the singlet at 0 K. The simple addition between the terminal oxygen atoms of the two HOO* moieties of the prereactive complex, leading to the straight chain H2O4 intermediate on the singlet PES, involves an energy barrier of 7.3 kcal/mol at 0 K. Because the decomposition of such an intermediate into HOOH + O2 entails an energy barrier of 45.2 kcal/mol at 0 K, it is concluded that the single H-atom transfer on the triplet PES is the dominant pathway leading to HOOH + O2. Finally, the strong negative temperature dependence of the rate constant observed for this reaction is attributed to the reversible formation of the prereactive complex in the entrance channel rather than to a short-lived tetraoxide intermediate.     

Ab initio benchmark study for the oxidative addition of CH4 to Pd: importance of basis-set flexibility and polarization

To obtain a state-of-the-art benchmark potential energy surface (PES) for the archetypal oxidative addition of the methane C-H bond to the palladium atom, we have explored this PES using a hierarchical series of ab initio methods (Hartree-Fock, second-order M?ller-Plesset perturbation theory, fourth-order M?ller-Plesset perturbation theory with single, double and quadruple excitations, coupled cluster theory with single and double excitations (CCSD), and with triple excitations treated perturbatively [CCSD(T)]) and hybrid density functional theory using the B3LYP functional, in combination with a hierarchical series of ten Gaussian-type basis sets, up to g polarization. Relativistic effects are taken into account either through a relativistic effective core potential for palladium or through a full four-component all-electron approach. Counterpoise corrected relative energies of stationary points are converged to within 0.1-0.2 kcal/mol as a function of the basis-set size. Our best estimate of kinetic and thermodynamic parameters is -8.1 (-8.3) kcal/mol for the formation of the reactant complex, 5.8 (3.1) kcal/mol for the activation energy relative to the separate reactants, and 0.8 (-1.2) kcal/mol for the reaction energy (zero-point vibrational energy-corrected values in parentheses). This agrees well with available experimental data. Our work highlights the importance of sufficient higher angular momentum polarization functions, f and g, for correctly describing metal-d-electron correlation and, thus, for obtaining reliable relative energies. We show that standard basis sets, such as LANL2DZ+1f for palladium, are not sufficiently polarized for this purpose and lead to erroneous CCSD(T) results. B3LYP is associated with smaller basis set superposition errors and shows faster convergence with basis-set size but yields relative energies (in particular, a reaction barrier) that are ca. 3.5 kcal/mol higher than the corresponding CCSD(T) values.     

Conformational preferences and internal rotation in alkyl- and phenyl-substituted thiourea derivatives

Potential energy surfaces (PES) for rotation about the N-C(sp(3)) or N-C(aryl) bond and energies of stationary points on PES for rotation about the C(sp(2))-N bond are reported for methylthiourea, ethylthiourea, isopropylthiourea, tert-butylthiourea, and phenylurea, using the MP2/aug-cc-pVDZ method. Analysis of alkylthioureas shows that conformations, with alkyl groups cis to the sulfur atom, are more stable (by 0.4-1.5 kcal/mol) than the trans forms. All minima adopt anti configurations with respect to nitrogen pyramidalization, whereas syn configurations are not stationary points on the MP2 potential surface. In contrast, analysis of phenylthiourea reveals that a trans isomer in a syn geometry is the global minimum, whereas a cis isomer in an anti geometry is a local minimum with a relative energy of 2.7 kcal/mol. Rotation about the C(sp(2))-N bond in alkyl and phenyl thioureas is slightly more hindered (9.1-10.2 kcal/mol) than the analogous motion in the unsubstituted molecule (8.6 kcal/mol). The maximum barriers to rotation for the methyl, ethyl, isopropyl, tert-butyl, and phenyl substituents are predicted to be 1.2, 8.9, 8.6, 5.3, and 0.9 kcal/mol, respectively. Corresponding PESs are consistent with the experimental dihedral angle distribution observed in crystal structures. The results of the electronic structure calculations are used to benchmark the performance of the MMFF94 force field. Systematic discrepancies between MMFF94 and MP2 results were improved by modification of selected torsion parameters and one of the van der Waals parameters for sulfur.     

Combustion chemistry: important features of the C3H5 potential energy surface, including allyl radical, propargyl + H2, allene + H, and eight transition states

The C(3)H(5) potential energy surface (PES) encompasses molecules of great significance to hydrocarbon combustion, including the resonantly stabilized free radicals propargyl (plus H(2)) and allyl. In this work, we investigate the interconversions that take place on this PES using high level coupled cluster methodology. Accurate geometries are obtained using coupled cluster theory with single, double, and perturbative triple excitations [CCSD(T)] combined with Dunning's correlation consistent quadruple-ζ basis set cc-pVQZ. The energies for these stationary points are then refined by a systematic series of computations, within the focal point scheme, using the cc-pVXZ (X = D, T, Q, 5, 6) basis sets and correlation treatments as extensive as coupled cluster with full single, double, and triple excitation and perturbative quadruple excitations [CCSDT(Q)]. Our benchmarks provide a zero-point vibrational energy (ZPVE) corrected barrier of 10.0 kcal mol(-1) for conversion of allene + H to propargyl + H(2). We also find that the barrier for H addition to a terminal carbon atom in allene leading to propenyl is 1.8 kcal mol(-1) lower than that for the addition to a central atom to form the allyl radical.     

Six-dimensional potential energy surface of the dissociative chemisorption of HCl on Au(111) using neural networks

We constructed a six-dimensional potential energy surface(PES)for the dissociative chemisorption of HCl on Au(111)using the neural networks method based on roughly 70000 energies obtained from extensive density functional theory(DFT)calculations.The resulting PES is accurate and smooth,based on the small fitting errors and good agreement between the fitted PES and the direct DFT calculations.Time-dependent wave packet calculations show that the potential energy surface is very well converged with respect to the number of DFT data points,as well as to the fitting process.The dissociation probabilities of HCl initially in the ground rovibrational state from six-dimensional quantum dynamical calculations are quite diferent from the four-dimensional fixed-site calculations,indicating it is essential to perform full-dimensional quantum dynamical studies for the title molecule-surface interaction system.     

Oxidative addition of the ethane C-C bond to Pd. An ab initio benchmark and DFT validation study

We have computed a state-of-the-art benchmark potential energy surface (PES) for the archetypal oxidative addition of the ethane C-C bond to the palladium atom and have used this to evaluate the performance of 24 popular density functionals, covering LDA, GGA, meta-GGA, and hybrid density functionals, for describing this reaction. The ab initio benchmark is obtained by exploring the PES using a hierarchical series of ab initio methods [HF, MP2, CCSD, CCSD(T)] in combination with a hierarchical series of five Gaussian-type basis sets, up to g polarization. Relativistic effects are taken into account either through a relativistic effective core potential for palladium or through a full four-component all-electron approach. Our best estimate of kinetic and thermodynamic parameters is -10.8 (-11.3) kcal/mol for the formation of the reactant complex, 19.4 (17.1) kcal/mol for the activation energy relative to the separate reactants, and -4.5 (-6.8) kcal/mol for the reaction energy (zero-point vibrational energy-corrected values in parentheses). Our work highlights the importance of sufficient higher angular momentum polarization functions for correctly describing metal-d-electron correlation. Best overall agreement with our ab initio benchmark is obtained by functionals from all three categories, GGA, meta-GGA, and hybrid DFT, with mean absolute errors of 1.5 to 2.5 kcal/mol and errors in activation energies ranging from -0.2 to -3.2 kcal/mol. Interestingly, the well-known BLYP functional compares very reasonably with a slight underestimation of the overall barrier by -0.9 kcal/mol. For comparison, with B3LYP we arrive at an overestimation of the overall barrier by 5.8 kcal/mol. On the other hand, B3LYP performs excellently for the central barrier (i.e., relative to the reactant complex) which it underestimates by only -0.1 kcal/mol.     

Modified feed-forward neural network structures and combined-function-derivative approximations incorporating exchange symmetry for potential energy surface fitting

The classical interchange (permutation) of atoms of similar identity does not have an effect on the overall potential energy. In this study, we present feed-forward neural network structures that provide permutation symmetry to the potential energy surfaces of molecules. The new feed-forward neural network structures are employed to fit the potential energy surfaces for two illustrative molecules, which are H(2)O and ClOOCl. Modifications are made to describe the symmetric interchange (permutation) of atoms of similar identity (or mathematically, the permutation of symmetric input parameters). The combined-function-derivative approximation algorithm (J. Chem. Phys. 2009, 130, 134101) is also implemented to fit the neural-network potential energy surfaces accurately. The combination of our symmetric neural networks and the function-derivative fitting effectively produces PES fits using fewer numbers of training data points. For H(2)O, only 282 configurations are employed as the training set; the testing root-mean-squared and mean-absolute energy errors are respectively reported as 0.0103 eV (0.236 kcal/mol) and 0.0078 eV (0.179 kcal/mol). In the ClOOCl case, 1693 configurations are required to construct the training set; the root-mean-squared and mean-absolute energy errors for the ClOOCl testing set are 0.0409 eV (0.943 kcal/mol) and 0.0269 eV (0.620 kcal/mol), respectively. Overall, we find good agreements between ab initio and NN prediction in term of energy and gradient errors, and conclude that the new feed-forward neural-network models advantageously describe the molecules with excellent accuracy.     

Free energy profiles for monomer capture in Grubbs- and SHOP-type olefin polymerization catalysts: a constraint ab initio molecular dynamics study

Density functional theory together with Car-Parrinello ab initio molecular dynamics simulation has been used to investigate the free energy profiles (FEP) of monomer capture in Grubbs- and SHOP-type olefin polymerization catalysts. The FEPs along the reaction coordinates at 300 K were determined directly by a point wise thermodynamic integration technique. Comparison between potential energy profile (PEP) and the FEP has been made. The results show that, for both catalysts, the PEP for the monomer ethylene uptake by the metal center is a typical Morse curve without energy barrier. However, a small barrier (1.8 kcal/mol for Grubbs catalyst and 2.4 kcal/mol for SHOP catalyst) exists on the FEP. The pi complexation energy on the FES at 300 K is higher by 10-12 kcal/mol over that on the PES. The differences between FES and PES are due to entropy contribution. Slow growth simulations on the ethylene capture process show that the ethylene attacks the metal center by an asynchronous mode. This indicates that the forming of the pi-bonding between the metal and ethylene is initiated by electrophilic attack of the metal to one of the ethylene carbons.     

Oxidative addition of the fluoromethane C-F bond to Pd. An ab initio benchmark and DFT validation study

We have computed a state-of-the-art benchmark potential energy surface (PES) for two reaction pathways (oxidative insertion, OxIn, and S(N)2) for oxidative addition of the fluoromethane C-F bond to the palladium atom and have used this to evaluate the performance of 26 popular density functionals, covering LDA, GGA, meta-GGA, and hybrid density functionals, for describing these reactions. The ab initio benchmark is obtained by exploring the PES using a hierarchical series of ab initio methods (HF, MP2, CCSD, CCSD(T)) in combination with a hierarchical series of seven Gaussian-type basis sets, up to g polarization. Relativistic effects are taken into account through a full four-component all-electron approach. Our best estimate of kinetic and thermodynamic parameters is -5.3 (-6.1) kcal/mol for the formation of the reactant complex, 27.8 (25.4) kcal/mol for the activation energy for oxidative insertion (OxIn) relative to the separate reactants, 37.5 (31.8) kcal/mol for the activation energy for the alternative S(N)2 pathway, and -6.4 (-7.8) kcal/mol for the reaction energy (zero-point vibrational energy-corrected values in parentheses). Our work highlights the importance of sufficient higher angular momentum polarization functions for correctly describing metal-d-electron correlation. Best overall agreement with our ab initio benchmark is obtained by functionals from all three categories, GGA, meta-GGA, and hybrid DFT, with mean absolute errors of 1.4-2.7 kcal/mol and errors in activation energies ranging from 0.3 to 2.8 kcal/mol. The B3LYP functional compares very well with a slight underestimation of the overall barrier for OxIn by -0.9 kcal/mol. For comparison, the well-known BLYP functional underestimates the overall barrier by -10.1 kcal/mol. The relative performance of these two functionals is inverted with respect to previous findings for the insertion of Pd into the C-H and C-C bonds. However, all major functionals yield correct trends and qualitative features of the PES, in particular, a clear preference for the OxIn over the alternative S(N)2 pathway.     

Amide–π interactions between formamide and benzene

High‐level ab initio calculations have been carried out using a formamide–benzene model system to evaluate amide–π interactions. The interaction energies were estimated as a sum of the CCSD(T) correlation contribution and the HF energy at the complete basis set limit, for the geometries of the model structures at the energy minimum obtained by potential energy surface (PES) scans. NH/π geometry in a face‐on configuration was found to be the most attractive among the various geometries considered, with interaction energy of ?3.75 kcal/mol. An interaction energy of ?2.08 kcal/mol was calculated for the stacked N/Center type geometry, where the nitrogen atom of formamide points directly toward the center of the aromatic ring. The weakest C?O/π geometry, where a carbonyl oxygen atom points toward the plane of the aromatic ring, was found to have energy minimum at an intermolecular distance of 3.67 Å from the PES, with a repulsive interaction energy less than 1 kcal/mol. However, if there are simultaneous attractive interactions with other parts of the molecule besides the amide group, the weak repulsion could be easily overcome, to give a C?O/π geometry interaction. © 2009 Wiley Periodicals, Inc. J Comput Chem, 2009     

Gas phase structures, energetics, and potential energy surfaces of disilacyclohexanes

The molecular structures of 1,4-, 1,3-, and 1,2-disilacyclohexanes (denoted as 14, 13, and 12, respectively) were investigated by means of gas electron diffraction (GED). Each molecule was found to possess a chair as the most stable conformation in the gas phase, the point group being C(2h), C(s), and C(2), respectively. Experimental GED structures are in good agreement with theoretical calculations (MP2/cc-pVTZ and B3LYP/cc-pVTZ). A qualitative ring strain analysis suggests 14 to be the most stable and 12 the least stable of the parent disilacyclohexanes. Relative energy calculations with the G4 model chemistry protocol, on the other hand, predict 13 to be the most stable isomer, 5.9 and 14.2 kcal/mol more stable than 14 and 12, respectively. The enhanced stability of 13 compared to 14 is in agreement with an analysis on endocyclic bond lengths and bond polarities. The heats of formation (G4 calculations) are predicted to be -12.3, -18.1, and -3.9 kcal/mol for 14, 13, and 12, respectively. The potential energy surface (PES) and the lowest energy path for the chair-to-chair inversion have been calculated for each isomer. In addition to the two chair forms in each case and some half-chair or sofa-like transition states (four in the case of 14, and two in the case of 13), there are two twist forms found as stationary points on the PES of 14, six twist and six boat forms on the PES of 13, and four twist and six boat forms on the PES of 12.     

Conformational analysis and rotational barriers of alkyl- and phenyl-substituted urea derivatives

Potential energy surfaces (PES) for rotation about the N-C(sp(3)) or N-C(aryl) bond and energies of stationary points on PES for rotation about the C(sp(2))-N bond are reported for methylurea, ethylurea, isopropylurea, tert-butylurea, and phenylurea, using the B3LYP/DZVP2 and MP2/aug-cc-pVDZ methods. The analysis of alkylureas reveals cis and (less stable) trans isomers that adopt anti geometries, whereas syn geometries do not correspond to stationary points. In contrast, the analysis of phenylurea reveals that the lowest energy form at the MP2 level is a trans isomer in a syn geometry. The fully optimized geometries are in good agreement with crystal structure data, and PESs are consistent with the experimental dihedral angle distribution. Rotation about the C(sp(2))-N bond in alkylureas and phenylurea is slightly more hindered (8.6-9.4 kcal/mol) than the analogous motion in the unsubstituted molecule (8.2 kcal/mol). At the MP2 level of theory, the maximum barriers to rotation for the methyl, ethyl, isopropyl, tert-butyl, and phenyl groups are predicted to be 0.9, 6.2, 6.0, 4.6, and 2.4 kcal/mol, respectively. The results are used to benchmark the performance of the MMFF94 force field. Systematic discrepancies between MMFF94 and MP2 results were improved by modification of several torsional parameters.     

Theoretical study of the dynamics of Cl + O3 reaction I. Ab initio potential energy surface and quasiclassical trajectory results

We present a global full dimensional potential energy surface (PES) for the Cl + O(3)→ ClO + O(2) reaction, which is an elementary step in a catalytic cycle that leads to the destruction of ozone in the stratosphere. The PES is constructed by interpolation of quantum chemistry data using the method developed by Collins and co-workers. Ab initio data points (energy, gradients and Hessian matrix elements) have been calculated at the UQCISD/aug-cc-pVDZ (unrestricted quadratic configuration interaction with single and double excitations) level of theory. The ab initio calculations predict a markedly non-coplanar (dihedral angle of 80°) transition state for the reaction, located very early in the reactant valley and slightly below the energy of the reactants as long as the spin-orbit splitting is neglected. Quasiclassical trajectory (QCT) calculations have been carried out at several collision energies to investigate the reaction dynamics. The QCT excitation function shows no threshold, displays a minimum at a collision energy of 2.5 kcal mol(-1), and then increases monotonically at larger collision energies. This behaviour is consistent with a barrierless reaction dominated by an oxygen-abstraction mechanism. The calculated product vibrational distributions (strongly inverted for ClO) and rate constants are compared with experimental determinations. Differential cross sections (DCS) summed over all final states are found to be in fairly good agreement with those derived from crossed molecular beam experiments.