(220n) A Generalized QSPR Model for Henry's Law Constant of Hydrogen in Organic Compounds | AIChE

(220n) A Generalized QSPR Model for Henry's Law Constant of Hydrogen in Organic Compounds


Odafin, C. E. - Presenter, Oklahoma State University
Gebreyohannes, S. B., Oklahoma State University
Yerramsetty, K. M., Oklahoma State University
Neely, B. J., Oklahoma State University
Gasem, K. A. M., Oklahoma State University

Henry's Law constant is a key physical property, which determines the distribution of gases in liquids. Hydrogen gas has a wide range of applications in chemical processes. For instance, in biomass conversion to biofuels, hydrogenation reactions are applied in the reduction of the oxygen content of the biofuel. Hydrogen solubilities in various solvents have been documented widely, and empirical models for Henry's law constants have been determined; however, these data are expensive to obtain and the empirical models are mostly suited for the available data. Hence, the need exists for generalized models for predicting Henry's constants. These generalized models will be applicable for a priori predictions of compounds lacking experimental data or for newly and yet to be synthesized compounds.

In this work, the objective was to develop QSPR generalized models for the a priori prediction of Henry's constant in various organic solvents. Specifically, the model results have been applied to hydrogenation reactions in a biphasic catalytic reactor for the upgrading of bio-oil to biofuels.

Hydrogen solubility and Henry's constant data were collected for a diverse range of compounds at various temperatures and pressures. The generalized equation of state (GEOS) software was used in the regression and evaluation of the available data. Quantitative structure-property relationship (QSPR) methodology was then applied to generalize the model parameters of the regressions. A representative database was employed in the model development and validation. The goal is to obtain hydrogen solubility predictions within 2-3 times the average experimental errors.