Divergence of Many-Body Perturbation Theory for Noncovalent Interactions of Large Molecules

• DOI: 10.1021/acs.jctc.9b01176
• Publication Date: 2020-02-27T20:32:15Z
• Authors: Brian D. Nguyen; Guo P. Chen; Matthew M. Agee; Asbjörn M. Burow; Matthew P. Tang; Filipp Furche
• Keywords: Coupled cluster; Random phase approximation

Prompted by recent reports of large errors in noncovalent interaction (NI) energies obtained from many-body perturbation theory (MBPT), we compare the performance second-order Mo̷ller-Plesset MBPT (MP2), spin-scaled MP2, dispersion-corrected semilocal density functional approximations (DFAs), and post-Kohn-Sham random phase approximation (RPA) for predicting binding supramolecular complexes contained S66, L7, S30L benchmarks. All are extrapolated to basis set limit, corrected superposition errors, compared reference results domain-based local pair-natural orbital coupled-cluster (DLPNO-CCSD(T)) or better quality. Our confirm that MP2 severely overestimates complexes, producing relative over 100% several benchmark compounds. RPA consistently range between 5 10%, significantly less than reported previously using smaller sets, whereas methods show limitations similar albeit pronounced, empirically DFAs perform almost as well RPA. Regression analysis reveals a systematic increase energy with system size at rate approximately 0.1% per valence electron, DFA virtually independent size. These observations corroborated comparison computed rotational constants organic molecules gas-phase spectroscopy data ROT34 benchmark. To analyze these results, an asymptotic adiabatic connection symmetry-adapted (AC-SAPT) is developed, which uses monomers full coupling, whose ground-state constrained complex. Using fluctuation-dissipation theorem, obtain nonperturbative "screened second-order" expression dispersion terms monomer quantities, exact non-overlapping subsystems free induction terms; first-order RPA-like Hartree, exchange, correlation kernel recovers macroscopic Lifshitz limit. The AC-SAPT expansion Taylor coupling strength integrand. Explicit expressions convergence radius series derived within numerically evaluated. While always convergent nondegenerate when used, it found spuriously diverge MBPT, except smallest least polarizable monomers. divergence confirmed RPA; prior numerical on SAPT revisited support this conclusion once sufficiently high orders included. cause failure NIs systems missing incomplete "electrodynamic" screening Coulomb due induced particle-hole pairs electrons different monomers, leaving effective too strong converge. Hence, cannot be considered reliable quantitative predictions NIs, even moderately few tens atoms. accurately account electrodynamic polarization makes qualitatively unsuitable applications such nanostructures, macromolecules, soft materials; more robust approaches coupled cluster should used instead whenever possible.