Abstract

A partial least-squares (PLS)–FTIR procedure is proposed for the direct determination of methanol and ethanol in liquid samples, based on vapour generation from small injected volumes of untreated samples, volatilized inside a Pyrex glass reactor, transported into a multi-pathlength cell with nitrogen carrier gas and then continuously measured by FTIR spectrometry. The spectra were processed using PLS regression. Several sets of standards and different wavenumber ranges were tested for calibration. A calibration set integrated by aqueous solutions of ethanol from 2.5 to 22.5% v/v and aqueous solutions of methanol from 0.5 to 3.5% v/v provided the lowest prediction errors. Average relative prediction errors for the PLS model of about 3% for ethanol and 2% for methanol were obtained for aqueous mixtures of the two compounds studied, with RSDs of 2.7 and 2.3%, respectively, calculated from four injections of 1 µl of sample, providing the method with a sampling frequency of 69 h–1. Results predicted by PLS–FTIR were compared with those found using a proportional equations approach. The method was applied to the determination of methanol and ethanol in high-level alcoholic beverages and cosmetic products, obtaining average recoveries of 107 ± 6 and 98 ± 2%, respectively.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call