Reliability of interacting quantum atoms (IQA) data computed from post-HF densities: impact of the approximation used.

The performance of BBC1, BBC2 and Müller approximations, in terms of reliability of IQA data, was investigated at the CCSD, CCSD(T) and MP2 levels using glycol, as a case study, in interpreting the relative stability of its conformers, one with H-bond type intramolecular interaction and the other with a steric clash between two O-atoms. The CCSD/BBC1 level appeared to be perfectly suited as a reference needed to evaluate all possible levels of theory/approximation combinations (LoT/LoA). We found the reliability trend LoT/BBC1 > LoT/BBC2 > LoT/Müller (as well as its origin) and concluded that the Müller approximation should not be used when the accuracy of IQA-defined energy terms is considered. Moreover, we have established that the requirement of reproducing, by IQA calculations, electronic energy is desirable but not a necessary requirement when a comparative approach is used, such as in FAMSEC-based analysis (FAMSEC = fragment attributed molecular system energy change). A new criterion is proposed to assess the quality of IQA data for comparative analyses, ΔE(IQA) ≈ ΔE, where ΔE(IQA) and ΔE are the IQA and electronic energy differences, respectively, between the fin-state and ref-state of a molecular system. The closer ΔE(IQA) approaches ΔE, the closer the FAMSEC data approach values obtained at the exceptionally well performing CCSD/BBC1 level, regardless of the LoT/LoA combination used. Importantly, the MP2/BBC1 level performed nearly as well as the CCSD/BBC1 level in comparative studies. The origin of the MP2/BBC1 approximation's exceptional and the MP2/Müller approximations's acceptable performance in explaining the relative stability of glycol conformers has been uncovered and discussed in detail.