New Concepts In Electron Correlation With Application To Extended Systems

The objective of our work for ONR has been to develop and apply new, correlated electronic structure methods for infinite periodic systems. The methods of interest include many body perturbation theory (MBPT) and its infinite order coupled cluster generalizations and complementary density functional methods. Today we routinely do highly correlated studies of molecules (including making reliable predictions of structure, photoelectron, vibrational, and electronic spectra). The next frontier in electronic structure theory is the treatment (with the same level of confidence) of polymers, surfaces, and crystals, which are semiconductors or insulators and are collectively termed "extended systems." The current paradigm for molecules is SCF, MBPT(2), CCSD, CCSD(T), CCSDT, and full CI. Though convergence is not monotonic, such a sequence, augmented by experience and approximate error bars, typically offers purely ab inito, extrapolated solutions that are predictive. In contrast, no such paradigm exists for extended systems. Furthermore, there is no full CI so only high-level CC methods can offer reference results. The only methods currently available for extended systems are periodic Hartree Fock (i.e., SCF) and density functional theory (DFT), usually in the local density approximation (LDA). The former lacks the critical electron correlation effects responsible for so many of the interesting phenomena in solids, from band gaps to high Tc superconductivity, while the latter might introduce some correlation effects, but an unclear amount. In fact, the greatest weakness of DFT is that, unlike the above CC/MBPT paradigm, there is no way to systematically converge to the correct result.