Busca de descritores para modelagem quimiométrica de inibidores de corrosão

• Autor(a) principal: Ramos, Ana Luísa de Queiroz Baddini
• Data de Publicação: 2005
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da Universidade Federal Fluminense (RIUFF)
• Texto Completo: https://app.uff.br/riuff/handle/1/17485
• topic: Inibidores de corrosão; Modelagem quimiométrica; Corrosão e anti-corrosivos; Modelagem; Quimiometria; Inhibitors of corrosion; chemiometrics modeling; CNPQ::CIENCIAS EXATAS E DA TERRA::QUIMICA

This work develops a quantitative study correlated to the efficiency in the inhibition of corrosion with molecular properties from three different sources. The two first ones involve the building of mathematical models with experimental data obtained from the literature, while in the third one the experimental data were provided in S. P. Cardoso s doctorate course thesis (Corrosion Lab COPPE/UFRJ). We made use of molecular quantum properties calculated by the AM1 method and also some descriptors associated with the contribution of groups. Most of the models were obtained by the multivaried linear regression and the reply functions used were log icorr and ln Kads. The first study elaborated multivaried models correlating iron corrosion inhibition in an HC1 5% medium using pyridine, dibenzyl sulfoxide and a homologue series of pyridine N-oxide as inhibitors. The second study developed multivaried models for the inhibition of the N-80 steel corrosion in an H2S saturated solution in ammonium chloride (pH ~ 4) using 32 molecules (aliphatic amines, imidazolines e amidoamines) as potential inhibitors. In that study quantum descriptors, descriptors by group contributions and products of descriptors were used. In the third study we built models for the corrosion inhibition in three different steels simultaneously in HC1 15% p/v using 23 organic compounds (amines, thiourea and derivatives and acetylenics alcohols). Besides the quantum descriptors, for contributions of groups of descriptors, we made use of steels descriptors. It´s important to emphasize that this initiative is unprecedented in the literature, for the previous works never studied more than one steel at a time, besides making use of much inferior number of compounds to that used in our study. In this study besides the multivaried linear regression we used the Partial Least Square method (PLS).