PHILOSOPHICAL MAGAZINE LETTERS,NOVEMBER 2003 VOL. 83, NO. 11, 699–708 Role of electron correlations in deoxyribonucleic acid duplexes: is an extended Hubbard Hamiltonian a good model in this case? E.B. STARIKOV Karolinska Institute, Department of Biosciences at NOVUM, Center for Structural Biochemistry, S-141 57 Huddinge, Sweden [Received in final form 20 May 2003 and accepted 20 May 2003 ] Abstract The parameters of an extended Hubbard Hamiltonian for poly(dA)–poly(dT) and poly(dG)–poly(dC) homopolynucleotide duplexes in A- and B- deoxyribonucleic acid (DNA) geometries have been estimated within the framework of the Fortunelli–Painelli dimer model using Hartree–Fock semiempirical quantum chemistry reinforced by a configuration interaction scheme to include one- and two-electron excitations. It is tentatively concluded that inclusion of electron correlations is essential in treating DNA duplexes. } 1. INTRODUCTION The problem of charge transfer and transport in deoxyribonucleic acid (DNA) has attracted appreciable attention during the last decade (for the most recent com- prehensive treatises see, for example, Treadway et al. (2002), Giese (2002) and Starikov (2002a)). Although much has been done, physical–chemical mechanisms of the phenomenon in question remain largely unclear. In clarifying these, a number of theoretical approaches are used (for example Rakhmanova and Conwell (2001), Bixon and Jortner (2002), Grozema et al. (2002), and Voityuk and Ro¨ sch (2002), and references therein). However, the conceptual schemes explicitly or implicitly used to describe the electronic structure of DNA do not exceed simple one-band tight- binding Hamiltonians and tacitly neglect electron correlations. Such an approach is justified when applied to explaining specific experiments such as those described by Treadway et al. (2002) and Giese (2002), where strictly one hole (or one electron) is injected into and travels through DNA duplexes. However, intrinsic conduction properties of DNA do not seem to comply with such a simple one-electron picture. Indeed, the important role of electron correlation in correct description of the DNA electronic structure has been known for a long time (for example Ladik (1988)). Moreover, a number of experimental results have recently been published, namely data on proximity-induced superconductivity in DNA (Kasumov et al. 2001) and on the decrease in the electron paramagnetic resonance signal intensity versus guanine- radical-generating intercalators in DNA (Schiemann et al. 2000), which could be indicative of antiferromagnetic excitations or even Bardeen–Cooper–Schrieffer-like Philosophical Magazine Letters ISSN 0950–0839 print/ISSN 1362–3036 online # 2003 Taylor & Francis Ltd http://www.tandf.co.uk/journals DOI: 10.1080/0950083031000151374