A general fully automated implementation of the incremental scheme for molecules and embedded clusters in the framework of the coupled cluster singles and doubles theory is presented. The code can be applied to arbitrary order of the incremental expansion and is parallelized in a master/slave structure. The authors found that the error in the total correlation energy is lower than 1 kcal/mol with respect to the canonical CCSD calculation if the incremental series is truncated in a proper way.

In this work we present a new partitioning scheme for the incremental approach in combination with the efficient (F12*) approximation for explicitly correlated coupled cluster (J. Chem. Phys. 2010, 132, 231102). Furthermore we establish a black-box truncation scheme which provides chemical accuracy for the absolute energies of 81 molecules and 51 reaction energies. The errors in the absolute CCSD(T)/cc-pVTZ-F12 energies due to the local approximations are characterized by mean = -0.24 kJ/mol, σ = 0.49 kJ/mol, mae = 0.37 kJ/mol, rmsd = 0.54 kJ/mol, and range = 3.63 kJ/mol. For the reaction energies we find mean = 0.07 kJ/mol, σ = 0.49 kJ/mol, mae = 0.33 kJ/mol, rmsd = 0.49 kJ/mol, and range = 2.40 kJ/mol. On the basis of these findings it is evident that the incremental scheme provides highly accurate CCSD(T) energies of benchmark quality.

A fully automated implementation of the incremental scheme for CCSD energies has been extended to treat MP2 and CCSD(T) energies. It is shown in applications on water clusters that the error of the incremental expansion for the total energy is below 1 kcal/mol already at second or third order. It is demonstrated that the approach saves CPU time, RAM, and disk space. Finally it is shown that the calculations can be run in parallel on up to 50 CPUs, without significant loss of computer time.

We present a theoretical study of carbene formation from the 1-ethyl-3-methylimidazolium acetate ionic liquid in the absence and presence of CO2 in gas and liquid phase. Although CO2 physisorption constitutes a precursory step of chemisorption (the CO2's reaction with carbenes, which forms from cations via proton abstraction by anions), it also enables a very stable CO2-anion associate. However, this counteracts the chemical absorption by reducing the basicity of the anion and the electrophilicity of the CO2, which is reflected by charge transfer. Accordingly, the observable carbene formation in the gas phase is hindered in the presence of CO2. In the neat liquid, the carbene formation is also suppressed by the charge screening compared to the case of the gas phase; nevertheless, indications for carbene incidents appear. Interestingly, in the CO2-containing liquid we detect more carbene-like incidents than in the neat one, which is caused by the way CO2 is solvated. Despite the weakness of the CO2-cation interaction, the CO2-anion associate is distorted by cations, which can be seen in longer associate distances and reduced "binding" energies. While the single solvating anion is shifted away from CO2, many more solvating cations approach it compared to the case of the gas phase. This leads to the conclusion that while the ionic liquid effect stabilizes charged species, introducing neutral species such as CO2 provides an opposite trend, leading to an inverse ionic liquid effect with the facilitation of carbene formation and thus of chemical absorption.

A systematic screening procedure for small contributions in the incremental expansion of the correlation energy is presented. The performance of the proposed scheme is checked for the calculation of intermolecular interactions in realistic test systems as large as a guanine-cytosine base pair. It is found that the computational cost for the incremental expansion can be reduced considerably without significant loss of accuracy. Typically, the errors of the systems investigated here amount to <3.4, 0.22, and 0.06% for second-, third-, and fourth-order expansions, respectively. For almost all cases, the error in the total correlation energy can be kept below 1 kcal/mol with respect to the canonical CCSD result if the incremental series is truncated in a proper way.