Highly Accurate CCSDT(Q)/CBS Reaction Barrier Heights for a Diverse Set of Transition Structures: Basis Set Convergence and Cost-Effective Approaches for Estimating Post-CCSD(T) Contributions

Abstract

The ability to accurately calculate reaction barrier heights is of central importance to many areas of chemistry. We report an extensive study examining the basis set convergence of post-CCSD(T) contributions (up to CCSDT(Q)) for a diverse set of 28 reaction barrier heights. In contrast to previous studies, we focus here on larger transition structures (TSs) involving 4-7 non-hydrogen atoms. The set of reaction barrier heights includes pericyclic, bipolar cycloaddition, cycloreversion, and multiple proton transfer reactions. We find that in most cases post-CCSD(T) contributions converge rapidly towards the basis set limit, such that even double-z and truncated double-z basis sets provide useful estimates of the T-(T) and (Q) contributions, respectively. In addition, we find that due to the tendency of these small basis sets to systematically underestimate the T-(T) and (Q) components, scaling is an effective approach for improving performance. For example, scaling the T-(T)/cc-pVDZ contribution by 1.25 results in an RMSD of merely 0.4 kJ mol -1 relative to basis set limit reference values from W3lite-F12 theory. Similarly, calculating the (Q) contribution with a cc-pVDZ basis set without d functions and scaling by 1.6 results in an RMSD of 0.5 kJ mol -1 . We also examine the magnitude of post-CCSD(T) contributions for a wide range of TSs. We find that for pericyclic, bipolar cycloaddition, and multiple proton transfer reactions there is an effective cancellation between the T-(T) and (Q) components (i.e., they have opposite signs and are of similar magnitude), such that overall post-CCSD(T) contributions to the reaction barrier heights are below ~1 kJ mol -1 (in absolute value). However, for the barrier heights of cycloreversion reactions, the T-(T) and (Q) components are both negative and large and consequentially post-CCSD(T) contributions reduce the reaction barrier heights by significant amounts ranging between 4.1-6.7 kJ mol -1 .

Kinetics and Thermodynamics of Reactions Involving Criegee Intermediates: An Assessment of Density Functional Theory and Ab Initio Methods Through Comparison with CCSDT(Q)/CBS Data

“…Thus, overall post-CCSD(T) contributions to reaction barrier heights normally do not exceed the 1 kJ mol -1 mark. A striking exception to this are the challenging cycloreversion fragmentation reactions for which the T-(T) and (Q) contributions reinforce each other, 131 and thus post-CCSD(T) contributions can reach up to 6.7 kJ mol -1 . Inspection of Table S2 of the Supporting Information reveals that post-CCSD(T) contributions to the reaction barrier heights in the Criegee22 database do not exceed the 1.5 kJ mol -1 mark and are therefore consistent with the general finding of ref.…”

“…A recent study examined a set of CCSDT(Q)/CBS barrier heights for a diverse range of reactions, including pericyclic, bipolar cycloaddition, cycloreversion, and multiple-proton transfer reaction. 131 They found that CCSD(T)/CBS reaction barrier heights normally benefit from an effective error cancellation between the higher-order triples, T-(T), which tend to increase the barrier heights, and the quasiperturbative quadruples, (Q), which tend to reduce the barrier heights.…”

The HOX⋯SO2 (X=F, Cl, Br, I) Binary Complexes: Implications for Atmospheric Chemistry

“…The DBOC corrections ranged from −0.08 to −0.09 kcal mol −1 , showing no evidence of surface crossings. Based on the convergence of our methods we assume a conservative error from the ab initio limit of 0.25 kcal mol −1 [152–155] . The Gibbs Free energy correction was computed for each complex at 298.15 K using the harmonic frequency results.…”

SVECV-f12: benchmark of a composite scheme for accurate and cost effective evaluation of reaction barriers.

“…Some small differences were noticed between the two sets of calculations. Results at the ccCA level of calculation were taken from the paper by Grimes et al [29], while the most recent results by Karton at the W3lite-F12 level were also considered [5,37].…”

“…Since the barriers depend on the precise evaluation of both stable species (reactants and products) and transition states, normally very resource intensive methods are required to achieve high precision. CCSD(T) calculations [2][3][4] or beyond (including non-perturbative triple and quadruple excitations) are being presently employed in combination with very extensive basis sets (from quadruple up to sextuple zeta) extrapolated to the CBS limit, to obtain sub-kJ/mol precision [5]. But, of course, at an enormous computer cost.…”