ABSTRACT
In a previous report, we validated the use of a database that compiled the relative response factors of flavor and fragrance compounds under standard GC conditions for a flame ionization detector. Here we investigate the prediction of unknown response factors from the molecular structure by using combustion enthalpies. In a first step, this enthalpy was well-predicted with either ab initio calculation or multiple linear regression based on the molecular formula. In a second step, good correlation was observed between these combustion enthalpies and experimental relative response factors, and so the response factors were predictable from only the molecular formula. With a database of 351 compounds, about 60% of them exhibited a difference of less than 5% between the predicted and experimental relative response factors and about 80% exhibited a difference of less than 10%.
ABSTRACT
In a previous report, we validated the use of a database that compiled the relative response factors of flavor and fragrance compounds under standard GC conditions for a flame ionization detector. Here we investigate the prediction of unknown response factors from the molecular structure by using combustion enthalpies. In a first step, this enthalpy was well-predicted with either ab initio calculation or multiple linear regression based on the molecular formula. In a second step, good correlation was observed between these combustion enthalpies and experimental relative response factors, and so the response factors were predictable from only the molecular formula. With a database of 351 compounds, about 60% of them exhibited a difference of less than 5% between the predicted and experimental relative response factors and about 80% exhibited a difference of less than 10%.
