High-accuracy QSPR models for azeotropic property prediction of binary aromatic hydrocarbon mixtures: a genetic...

Area of Science:

• Chemical Engineering
• Physical Chemistry
• Computational Chemistry

Background:

• Aromatic hydrocarbons like benzene are vital industrial solvents.
• Accurate azeotropic data is crucial for optimizing separation processes like distillation.
• Existing methods for determining azeotropic properties can be resource-intensive.

Purpose of the Study:

• To develop predictive quantitative structure-property relationship (QSPR) models for azeotropic temperature and composition.
• To utilize only molecular structural information for model development.
• To establish robust and accurate models for binary mixtures of aromatic hydrocarbons.

Main Methods:

• Molecular geometries optimized using HyperChem (MM+ and PM3 methods).
• Molecular descriptors calculated via Online Chemical Modeling Environment (OCHEM).
• Genetic Function Approximation (GFA) and Multiple Linear Regression (MLR) used for model construction and descriptor selection.

Main Results:

• Developed two QSPR models with high agreement to experimental data (R² = 0.9454 and 0.9448).
• Achieved strong internal validation (R²cv = 0.9308 and 0.9364) and external validation (Q²ext = 0.8939 and 0.9364).
• Demonstrated superior predictive performance compared to existing models.

Conclusions:

• The developed QSPR models provide accurate predictions of azeotropic temperature and composition for aromatic hydrocarbon mixtures.
• These models offer a computationally efficient alternative for process design and optimization in chemical manufacturing.
• The study highlights the potential of QSPR in predicting physical properties of chemical mixtures.