Assessment of semi-empirical molecular orbital calculations for describing magnetic interactions. (4th November 2017)

• Record Type: Journal Article
• Authors: Saito, Toru; Kitagawa, Yasutaka; Kawakami, Takashi; Yamanaka, Shusuke; Okumura, Mitsutaka; Takano, Yu
• Is Part Of: Polyhedron. Volume 136(2017)
• Page Start: 52
• Page End: 57
• Publication Date: 2017-11-04
• Subjects: Magnetic exchange interaction -- Semi-empirical method -- Open-shell -- DFTB -- PM6; Chemistry, Inorganic -- Periodicals; Chimie inorganique -- Périodiques; Organometaalverbindingen; Anorganische chemie; 546.05
• DOI: 10.1016/j.poly.2017.02.029
• Languages: English
• ISSNs: 0277-5387

Abstract

Graphical abstract: A thorough comparison is made between the spin-unrestricted semi-empirical methods (UPM6, UrPM6, and UDFTB) and UCAM-B3LYP using π-conjugated organic diradicals. Overall, the degree of diradical character decreases as follows: UPM6 > UrPM6 ≈ UCAM-B3LYP > UDFTB. Thus, UrPM6 is able to reproduce UCAM-B3LYP well at much lower computational cost. Abstract: We report a thorough comparison between the spin-unrestricted semi-empirical molecular orbital (SE-UMO) and density functional theory (UDFT) calculations for a variety of π-conjugated diradical molecules which exhibit through-bond magnetic exchange interactions. The spin-unrestricted neglect of diatomic differential overlap (NDDO)-based methods (UPM6 and UrPM6) and self-consistent charge density functional tight-biding (UDFTB) are examined. The UCAM-B3LYP method is used as a representative of UDFT. We have found that NDDO-based and UDFTB methods give different performances with respect to the results of UCAM-B3LYP. The degree of diradical character in the target systems decreases in the order of UPM6 > UrPM6 ≈ UCAM-B3LYP > UDFTB. The conventional UPM6 and UDFTB calculations tend to overestimate and underestimate the diradical character, respectively, whereas UrPM6 is able to provide much the same results as UCAM-B3LYP. This is convincing evidence that UrPM6 will be useful for in silico screening of materials science at much lower computational cost.