|
Type |
Poster Presentation |
Area |
Industrial Chemistry |
Room No. |
Exhibition Hall 2+3 |
Time |
10월 20일 (금요일) 13:00~14:30 |
Code |
IND.P-64 |
Subject |
Prediction of the diffusion coefficient for infinite water based on quantitative structure-property relationship modeling |
Authors |
YE-EUN KIM, Byeong Hun Lee, Sung Kwang Lee* Department of Chemistry, Hannam University, Korea |
Abstract |
In physical chemistry, the diffusion coefficient is the amount of material that diffuses from one space to another under constant conditions of volume and concentration, and is the area of each cross-sectional unit per unit time. This is one of the important parameters used in diffusion, distillation, adsorption, and chemical reaction processes of various fields. The factors affecting the diffusion coefficient of chemicals are the bond types, the melting point, the crystal structure of the diffusing material, the presence of the medium and the temperature. In this study, quantitative structure-property relationship models were developed for predicting diffusion coefficient for infinite water from 2D chemical structure. The experimental diffusion coefficients from 5000 compounds at 298 K were collected for model development. In the 556 predefined descriptor pools, a suitable descriptors for learning model was pre-filtered. Learning method for linear and non-linear fitting was performed using multiple linear regression and support vector machines. Also Y-scrambling test was performed to verify that chance correlation of the models and the confidence range of the model were defined as kNN-based applicability domain. These models can predict the diffusion coefficient limit for infinite water and new compounds. |
E-mail |
tkfkdzldnal9@naver.com |
|