Assessment of the performance of MP2 and MP2 variants for the treatment of noncovalent interactions

For many years, MP2 served as the principal method for the treatment of noncovalent interactions. Until recently, this was the only technique that could be used to produce reasonably accurate binding energies, with binding energy errors generally below ~35%, at a reasonable computational cost. The past decade has seen the development of many new methods with improved performance for noncovalent interactions, several of which are based on MP2. Here, we assess the performance of MP2, LMP2, MP2-F12, and LMP2-F12, as well as spin component scaled variants (SCS) of these methods, in terms of their abilities to produce accurate interaction energies for binding motifs commonly found in organic and biomolecular systems. Reference data from the newly developed S66 database of interaction energies are used for this assessment, and a further set of 38 complexes is used as a test set for SCS methods developed herein. The strongly basis set-dependent nature of MP2 is confirmed in this study, with the SCS technique greatly reducing this behavior. It is found in this work that the spin component scaling technique can effectively be used to dramatically improve the performance of MP2 and MP2 variants, with overall errors being reduced by factors of about 1.5-2. SCS versions of all MP2 variants tested here are shown to give similarly accurate overall results.     

Analytic energy gradients for orbital‐optimized MP3 and MP2.5 with the density‐fitting approximation: An efficient implementation

Efficient implementations of analytic gradients for the orbital‐optimized MP3 and MP2.5 and their standard versions with the density‐fitting approximation, which are denoted as DF‐MP3, DF‐MP2.5, DF‐OMP3, and DF‐OMP2.5, are presented. The DF‐MP3, DF‐MP2.5, DF‐OMP3, and DF‐OMP2.5 methods are applied to a set of alkanes and noncovalent interaction complexes to compare the computational cost with the conventional MP3, MP2.5, OMP3, and OMP2.5. Our results demonstrate that density‐fitted perturbation theory (DF‐MP) methods considered substantially reduce the computational cost compared to conventional MP methods. The efficiency of our DF‐MP methods arise from the reduced input/output (I/O) time and the acceleration of gradient related terms, such as computations of particle density and generalized Fock matrices (PDMs and GFM), solution of the Z‐vector equation, back‐transformations of PDMs and GFM, and evaluation of analytic gradients in the atomic orbital basis. Further, application results show that errors introduced by the DF approach are negligible. Mean absolute errors for bond lengths of a molecular set, with the cc‐pCVQZ basis set, is 0.0001–0.0002 Å. © 2017 Wiley Periodicals, Inc.     

On the accuracy of DFT-SAPT, MP2, SCS-MP2, MP2C, and DFT+Disp methods for the interaction energies of endohedral complexes of the C(60) fullerene with a rare gas atom

Selected points on the potential energy surface for the complexes Rg@C(60) (Rg = He, Ne, Ar, Kr) are calculated with various theoretical methods, like symmetry-adapted perturbation theory with monomers described by density functional theory (DFT-SAPT), supermolecular M?ller-Plesset theory truncated on the second order (MP2), spin-component-scaled MP2 (SCS-MP2), supermolecular density functional theory with empirical dispersion correction (DFT+Disp), and the recently developed MP2C method that improves the MP2 method for long-range electron correlation effects. A stabilization of the endohedral complex is predicted by all methods, but the depth of the potential energy well is overestimated by the DFT+Disp and MP2 approaches. On the other hand, the MP2C model agrees well with DFT-SAPT, which serves as the reference. The performance of SCS-MP2 is mixed: it produces too low interaction energies for the two heavier guests, while its accuracy for He@C(60) and Ne@C(60) is similar to that of MP2C. Fitting formulas for the main interaction energy components, i.e. the dispersion and first-order repulsion energies are proposed, which are applicable for both endo- and exohedral cases. For all examined methods density fitting is used to evaluate two-electron repulsion integrals, which is indispensable to allow studies of noncovalent complexes of this size. It has been found that density-fitting auxiliary basis sets cannot be used in a black-box fashion for the calculation of the first-order SAPT electrostatic energy, and that the quality of these basis sets should be always carefully examined in order to avoid an unphysical long-range behavior.     

Scaled MP3 Non‐Covalent Interaction Energies Agree Closely with Accurate CCSD(T) Benchmark Data

Scaled MP3 interaction energies calculated as a sum of MP2/CBS (complete basis set limit) interaction energies and scaled third‐order energy contributions obtained in small or medium size basis sets agree very closely with the estimated CCSD(T)/CBS interaction energies for the 22 H‐bonded, dispersion‐controlled and mixed non‐covalent complexes from the S22 data set. Performance of this so‐called MP2.5 (third‐order scaling factor of 0.5) method has also been tested for 33 nucleic acid base pairs and two stacked conformers of porphine dimer. In all the test cases, performance of the MP2.5 method was shown to be superior to the scaled spin‐component MP2 based methods, e.g. SCS–MP2, SCSN–MP2 and SCS(MI)–MP2. In particular, a very balanced treatment of hydrogen‐bonded compared to stacked complexes is achieved with MP2.5. The main advantage of the approach is that it employs only a single empirical parameter and is thus biased by two rigorously defined, asymptotically correct ab‐initio methods, MP2 and MP3. The method is proposed as an accurate but computationally feasible alternative to CCSD(T) for the computation of the properties of various kinds of non‐covalently bound systems.     

Noncovalent complexes between dimethyl ether and formic acid--an ab initio and matrix isolation study.

The complexes formed by noncovalent interactions between formic acid and dimethyl ether are investigated by ab initio methods and characterized by matrix isolation spectroscopy. Six complexes with binding energies between -2.26 and -7.97 kcal mol(-1) (MP2/cc-pVTZ+zero point vibrational energy+basis set superposition erros) are identified. The two strongest bound complexes are, within a range of 0.3 kcal mol(-1), isoenergetic. The binding in these six dimers can be described in terms of OH...O, C=O...H, C-O...H and CH...O interactions. Matrix isolation spectroscopy allowed to characterize the two strongest bound complexes by their infrared spectra.     

Anchoring the potential energy surface of the cyclic water trimer

Six cyclic stationary points on the water trimer potential energy surface have been fully optimized at the MP2 level with the aug-cc-pVQZ basis set. In agreement with previous work, harmonic vibrational frequencies indicate that two structures are minima, three are transition states connecting minima on the surface while the remaining stationary point is a higher-order saddle point. The 1- and n-particle limits of the electronic energies of each of these six structures were estimated by systematically varying both the basis sets and theoretical methods. The former limit was approached with the cc-pVXZ and aug-cc-pVXZ families of basis sets (X=2-7) while MP2, CCSD(T), and BD(TQ) calculations helped examine the latter. Core correlation effects have also been assessed at the MP2 level with the cc-pCVXZ series of basis sets (X=2-5). These data have been combined to provide highly accurate relative energies and dissociation energies for these stationary points.     

On geometries of stacked and H-bonded nucleic acid base pairs determined at various DFT, MP2, and CCSD(T) levels up to the CCSD(T)/complete basis set limit level

The geometries and interaction energies of stacked and hydrogen-bonded uracil dimers and a stacked adeninecdots, three dots, centeredthymine pair were studied by means of high-level quantum chemical calculations. Specifically, standard as well as counterpoise-corrected optimizations were performed at second-order Moller-Plesset (MP2) and coupled cluster level of theory with single, double, and perturbative triple excitations [CCSD(T)] levels with various basis sets up to the complete basis set limit. The results can be summarized as follows: (i) standard geometry optimization with small basis set (e.g., 6-31G(*)) provides fairly reasonable intermolecular separation; (ii) geometry optimization with extended basis sets at the MP2 level underestimates the intermolecular distances compared to the reference CCSD(T) results, whereas the MP2/cc-pVTZ counterpoise-corrected optimization agrees well with the reference geometries and, therefore, is recommended as a next step for improving MP2/cc-pVTZ geometries; (iii) the stabilization energy of stacked nucleic acids base pairs depends considerably on the method used for geometry optimization, so the use of reliable geometries, such as counterpoise-corrected MP2/cc-pVTZ ones, is recommended; (iv) the density functional theory methods fail completely in locating the energy minima for stacked structures and when the geometries from MP2 calculations are used, the resulting stabilization energies are strongly underestimated; (v) the self-consistent charges-density functional tight binding method, with inclusion of the empirical dispersion energy, accurately reproduces interaction energies and geometries of dispersion-bonded (stacked) complexes; this method can thus be recommended for prescanning the potential energy surfaces of van der Waals complexes.     

Comparative study of BSSE correction methods at DFT and MP2 levels of theory

A comparative study of intermolecular potential energy curves is performed on the complexes H₂O(SINGLE BOND)HF, H₂O(SINGLE BOND)H₂O, H₂O(SINGLE BOND)H₂S, and H₂S(SINGLE BOND)H₂S using nine different basis sets at the MP2 and DFT (BLYP and B3LYP) levels of theory. The basis set superposition error is corrected by means of the counterpoise scheme and based on the “chemical Hamiltonian approach.” The counterpoise and CHA-corrected DFT curves are generally close to each other. Using small basis sets, the B3LYP functional cannot be favored against the BLYP one because the BLYP results sometimes get closer to the MP2 values than those of B3LYP. From the results—including the available literature data—we suggest that one has to use at least polarized-valence triple-zeta-quality basis sets (TZV, 6-311G) for the investigation of hydrogen-bonded complexes. Special attention must be paid to the physical nature of the binding. If the dispersion forces become significant DFT methods are not able to describe the interaction. Proper correction for the basis set superposition error is found to be mandatory in all cases. © 1998 John Wiley & Sons, Inc. J Comput Chem 19: 575–584, 1998     

Weak interactions between hypohalous acids and dimethylchalcogens

The complexes formed between dimethylchalcogens X(CH(3))(2) (X = S, Se, and Te) and hypohalous acids YOH (Y = F, Cl, Br, and I) have been studied at the MP2/aug'-cc-pVTZ computational level, five minima structures being located. Two of them correspond to hydrogen bonds (HB), another two to halogen bonds (XB) with the chalcogen acting as an electron donor, the last one showing a C-H···O contact. The most stable complexes of IOH and BrOH acids present halogen···chalcogen interactions with interaction energies, E(i), up to -49 kJ mol(-1). In the case of the ClOH and FOH molecules, the hydrogen bonded complexes are more stable with interaction energies between -27 and -34 kJ mol(-1). Linear correlations between the molecular electrostatic potential (MEP) stationary points at the van der Waals surface and the interaction energy have been found. The contribution of the different energy terms to the total interaction energy was analyzed by means of the DFT-SAPT theory finding that the electrostatic attractive term is dominant in the complexes with HB and XB, excepting a few cases in which the dispersion and induction terms become more important than the electrostatic one.     

Evaluation of the performance of post-Hartree-Fock methods in terms of intermolecular distance in noncovalent complexes

Dissociation curves calculated using multiple correlated QM methods for 66 noncovalent complexes (?ezá? et al., J Chem Theory Comput 2011, 7, 2427) have allowed us to interpolate equilibrium intermolecular distances for each studied method. Comparison of these data with CCSD(T)/complete basis set reference geometries provides information on how these methods perform in geometry optimizations. The large set of systems considered here is necessary for reliable statistical evaluation of the results and assessment of the robustness of the studied methods. Our results show that advanced methods such as MP3 and CCSD provide significant improvement over MP2 only when empirical scaling is used. The best results can be achieved with spin component scaled CCSD optimized for noncovalent interactions, with a root mean square error of 0.4% of the equilibrium distance. Scaled MP3, the MP2.5 method, yields comparably good results (error 0.5%) while being substantially cheaper. © 2011 Wiley Periodicals, Inc. J Comput Chem, 2012     

Interaction of metal porphyrins with fullerene C60: a new insight

The electronic structure and bonding in the noncovalent, supramolecular complexes of fullerene C60 with a series of first-row transition metal porphines MP (M=Fe, Co, Ni, Cu, Zn) have been re-examined with DFT methods. A dispersion correction was made for the C60-MP binding energy through an empirical method (J. Comput. Chem. 2004, 25, 1463). Several density functionals and two types of basis sets were employed in the calculations. Our calculated results are rather different from those obtained in a recent paper (J. Phys. Chem. A 2005, 109, 3704). The ground state of C60.FeP is predicted to be high spin (S=2); the low-spin (S=0), closed-shell state is even higher in energy than the intermediate-spin (S=1) state. With only one electron in the Co-dz2 orbital, the calculated Co-C60 distance is in fact rather short, about 0.1 A longer than the Fe-C60 distance in high-spin C60.FeP. Double occupation of an M-dz2 orbital in MP prevents close association of any axial ligand, and so the Ni-C60, Cu-C60, and Zn-C60 distances are much longer than the Co-C60 one. The evaluated MP-C60 binding energies (Ebind) are 0.8 eV (18.5 kcal/mol) for M=Fe/Co and 0.5 eV (11.5 kcal/mol) for M=Ni/Cu/Zn (Ebind is about 0.2 eV larger in the case of C60-MTPP). They are believed to be reliable and accurate based on our dispersion-corrected DFT calculations that included the counterpoise (CP) correction. The effects of the C60 contact on the redox properties of MP were also examined.     

Anomeric effect and theoretical vibrational spectra compared to the experiment in 2-chloro-1,3,2-dioxaphosphorinane-2-oxide, -sulfide,and -selenide. I. Anomeric effect and structures

2-Chloro-1,3,2-dioxaphosphorinane-2-oxide, -sulfide, and -selenide are studied at the DFT/B3LYP level and several ab initio methods using a 6-311G** basis set. Our energy optimizations by all these methods show that for oxide DFT and ab initio methods are not much different, while for the sulfide and the selenide the DFT relative energies are higher by about a kcal/mol as compared to those of MP2, MP3, MP4(SDTQ)//MP2, and CCSD(T)//MP2 (//MP2 indicates that a single-point calculation based on the MP2 optimized geometry is performed). However, regardless of rather large relative energies, that does not change the fact that in all three cases the conformational equilibrium mixture contains more than 95% of the lowest, chair-equatorial conformer (this indicates that the P=X bond is in equatorial position). This one and the next higher conformer (chair-axial) are confirmed to be real conformers (energy minima) in all cases. The energetically much higher twist and boat forms are probably just stationary states and local maxima because in many cases, geometry optimizations do not converge to them. Only for MP2 and the selenide do all optimizations converge to the desired stationary state. The relative energies could all be explained in terms of anomeric effects and ring strains. The decreasing covalent character of the P=X bond, with X changing from O to S and to Se, shows itself in the increasing bond lengths and the decreasing strength of anomeric effects.     

Mechanisms of [1,3]-Sigmatropic Migrations of the Nitroso Group in the ON-X-CH=X Systems (X = O,S, Se,NH, CH<Subscript>2</Subscript>)

[1,3]-Sigmatropic migrations of the nitroso group in the systems ON-X-CH=X (X = O, S, Se, NH, CH₂) were studied by MP2(fc)/6-311+G** and B3LYP/6-311+G** quantum-chemical calculations. The energy barrier in the process was estimated at 2.4 (2.5), 20.0 (25.0), and 22.3 (23.4) kcal/mol for X = O, NH, and CH₂, respectively. The energy minima for X = S and X = Se correspond to cyclic structures with two-coordinate NO group, which are more stable than acyclic structures by 9.3 (4.3) (X = S) or 13.1 (5.7) kcal/mol (X = Se).     

Comparison of aromatic NH···π, OH···π, and CH···π interactions of alanine using MP2, CCSD, and DFT methods

A comparison of the performance of various density functional methods including long‐range corrected and dispersion corrected methods [MPW1PW91, B3LYP, B3PW91, B97‐D, B1B95, MPWB1K, M06‐2X, SVWN5, ωB97XD, long‐range correction (LC)‐ωPBE, and CAM‐B3LYP using 6‐31+G(d,p) basis set] in the study of CH···π, OH···π, and NH···π interactions were done using weak complexes of neutral (A) and cationic (A⁺) forms of alanine with benzene by taking the Møller–Plesset (MP2)/6‐31+G(d,p) results as the reference. Further, the binding energies of the neutral alanine–benzene complexes were assessed at coupled cluster (CCSD)/6‐31G(d,p) method. Analysis of the molecular geometries and interaction energies at density functional theory (DFT), MP2, CCSD methods and CCSD(T) single point level reveal that MP2 is the best overall performer for noncovalent interactions giving accuracy close to CCSD method. MPWB1K fared better in interaction energy calculations than other DFT methods. In the case of M06‐2X, SVWN5, and the dispersion corrected B97‐D, the interaction energies are significantly overrated for neutral systems compared to other methods. However, for cationic systems, B97‐D yields structures and interaction energies similar to MP2 and MPWB1K methods. Among the long‐range corrected methods, LC‐ωPBE and CAM‐B3LYP methods show close agreement with MP2 values while ωB97XD energies are notably higher than MP2 values. © 2010 Wiley Periodicals, Inc. J Comput Chem 2010     

True stabilization energies for the optimal planar hydrogen-bonded and stacked structures of guanine...cytosine, adenine...thymine, and their 9- and 1-methyl derivatives: complete basis set calculations at the MP2 and CCSD(T) levels and comparison with experiment

Planar H-bonded and stacked structures of guanine...cytosine (G.C), adenine...thymine (A...T), 9-methylguanine...1-methylcytosine (mG...mC), and 9-methyladenine...1-methylthymine (mA...mT) were optimized at the RI-MP2 level using the TZVPP ([5s3p2d1f/3s2p1d]) basis set. Planar H-bonded structures of G...C, mG...mC, and A...T correspond to the Watson-Crick (WC) arrangement, in contrast to mA...mT for which the Hoogsteen (H) structure is found. Stabilization energies for all structures were determined as the sum of the complete basis set limit of MP2 energies and a (DeltaE(CCSD(T)) - DeltaE(MP2)) correction term evaluated with the cc-pVDZ(0.25,0.15) basis set. The complete basis set limit of MP2 energies was determined by two-point extrapolation using the aug-cc-pVXZ basis sets for X = D and T and X = T and Q. This procedure is required since the convergency of the MP2 interaction energy for the present complexes is rather slow, and it is thus important to include the extrapolation to the complete basis set limit. For the MP2/aug-cc-pVQZ level of theory, stabilization energies for all complexes studied are already very close to the complete basis set limit. The much cheaper D-->T extrapolation provided a complete basis set limit close (by less than 0.7 kcal/mol) to the more accurate T-->Q term, and the D-->T extrapolation can be recommended for evaluation of complete basis set limits of more extended complexes (e.g. larger motifs of DNA). The convergency of the (DeltaE(CCSD(T)) - DeltaE(MP2)) term is known to be faster than that of the MP2 or CCSD(T) correlation energy itself, and the cc-pVDZ(0.25,0.15) basis set provides reasonable values for planar H-bonded as well as stacked structures. Inclusion of the CCSD(T) correction is essential for obtaining reliable relative values for planar H-bonding and stacking interactions; neglecting the CCSD(T) correction results in very considerable errors between 2.5 and 3.4 kcal/mol. Final stabilization energies (kcal/mol) for the base pairs studied are very substantial (A...T WC, 15.4; mA...mT H, 16.3; A...T stacked, 11.6; mA...mT stacked, 13.1; G...C WC, 28.8; mG...mC WC, 28.5; G...C stacked, 16.9; mG...mC stacked, 18.0), much larger than published previously. On the basis of comparison with experimental data, we conclude that our values represent the lower boundary of the true stabilization energies. On the basis of error analysis, we expect the present H-bonding energies to be fairly close to the true values, while stacked energies are still expected to be about 10% too low. The stacking energy for the mG...mC pair is considerably lower than the respective H-bonding energy, but it is larger than the mA...mT H-bonding energy. This conclusion could significantly change the present view on the importance of specific H-bonding interactions and nonspecific stacking interactions in nature, for instance, in DNA. Present stabilization energies for H-bonding and stacking energies represent the most accurate and reliable values and can be considered as new reference data.     

S66: A Well-balanced Database of Benchmark Interaction Energies Relevant to Biomolecular Structures

With numerous new quantum chemistry methods being developed in recent years and the promise of even more new methods to be developed in the near future, it is clearly critical that highly accurate, well-balanced, reference data for many different atomic and molecular properties be available for the parametrization and validation of these methods. One area of research that is of particular importance in many areas of chemistry, biology, and material science is the study of noncovalent interactions. Because these interactions are often strongly influenced by correlation effects, it is necessary to use computationally expensive high-order wave function methods to describe them accurately. Here, we present a large new database of interaction energies calculated using an accurate CCSD(T)/CBS scheme. Data are presented for 66 molecular complexes, at their reference equilibrium geometries and at 8 points systematically exploring their dissociation curves; in total, the database contains 594 points: 66 at equilibrium geometries, and 528 in dissociation curves. The data set is designed to cover the most common types of noncovalent interactions in biomolecules, while keeping a balanced representation of dispersion and electrostatic contributions. The data set is therefore well suited for testing and development of methods applicable to bioorganic systems. In addition to the benchmark CCSD(T) results, we also provide decompositions of the interaction energies by means of DFT-SAPT calculations. The data set was used to test several correlated QM methods, including those parametrized specifically for noncovalent interactions. Among these, the SCS-MI-CCSD method outperforms all other tested methods, with a root-mean-square error of 0.08 kcal/mol for the S66 data set.     

Description of noncovalent interactions involving π-system with high precision: An assessment of RPA,MP2, and DFT-D methods

Efficient approaches with high precision are essential for understanding the formation and stability of noncovalent interaction complexes. Here, 21 noncovalent interaction complexes involving π-system are selected and grouped in three subsets according to ETS–NOCV method: dispersion-dominated, electrostatic-dominated, and mixed. We mainly focus on examining the performance of random-phase approximation (RPA) on these π systems. The tested RPA-based method includes standard RPA and its variants including the related single excitations (SEs), renormalized single excitations (rSEs), second-order screened exchange (SOSEX), and the renormalized second-order perturbation theory (rPT2). The routine second-order Møller–Plesset perturbation theory (MP2) and three popular DFT-D functionals (M06-2X-D3, ωB97XD, and PBE-D3(BJ)) are also assessed for comparison. In this work, besides the calculation of interaction energies at Dunning-type aug-cc-pVDZ and aug-cc-pVTZ basis set, we also present a larger database of interaction energies calculated using MP2 and RPA methods with Dunning-type aug-cc-pVQZ basis set. An accurate CCSD(T)/CBS scheme is used to provide benchmark database. In addition to the high-level results, we also provide potential energy surfaces (PES) of different interaction type. Among all the tested methods, MP2 has a satisfactory performance on electrostatic-dominated and mixed-type systems, except for dispersion-dominated systems. DFT-D functionals, especially ωB97XD functional, has a balanced performance across all the tested systems. Importantly, for RPA-based methods, the calculation accuracy can be dramatically improved by taking into account SE or exchange effects, especially in the mixed complexes. We conclude that rPT2 among all the test RPA-based methods gives an overall satisfactory performance across different interaction types. © 2019 Wiley Periodicals, Inc.