Parametrização do método RM1 para ferro e níquel

• Ano de defesa: 2021
• Autor(a) principal: Silva, Gustavo de Santana
• Orientador(a): Freire, Ricardo Oliveira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Programa de Pós-Graduação: Pós-Graduação em Química
• Palavras-chave em Português: Química; Metodologia científica; Métodos semiempíricos; Parametrização; RM1; Metais de transição
• Palavras-chave em Inglês: Semiempirical methods; Parameterization; Transition metals
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::QUIMICA
• Link de acesso: https://ri.ufs.br/jspui/handle/riufs/14908

The need to study increasingly large systems led to the development of semiempirical methods in the second half of the 20th century. Such systems were impossible to study using ab initio methodologies, which, although more accurate, demand a lot of computational processing. Currently, even with advances related to hardware and software, the calculation of systems containing hundreds or thousands of atoms is not viable using such a methodology. Since 2006 made available to the scientific community, the RM1 currently has parameters for 25 elements, in a list that does not include transition metals. This work presents the parameterization of the RM1 method for geometry calculations of systems containing the metals Iron (Fe) and Nickel (Ni). The reference data were obtained from the Cambridge Structural Database (CSD) crystallographic structures database. A comparison between RM1 and the PM6 and PM7 methods, the only ones that present parameters for these metals, was performed, where RM1 was more accurate. The relative mean errors (RMEs) for the Fe and Ni systems were less than 4% and 5% in connection distances and 3.54 and 8.10 degrees in angles, respectively, when compared with crystallographic data, in addition to being smaller error variability among the evaluated methods. We tried to parameterize the RM1 to also calculate the heats of formation of the systems, however, the unsigned mean errors (UMEs) presented for Fe and Ni systems showed errors, respectively. Finally, we carried out a comparative study between the semiempirical and DFT methods, in which the DFT was more accurate for predicting polyhedra in the evaluated parameters, however, the RM1 showed close accuracy, with a computational cost several thousand times lower