Optimization of an analytical method based on SPME-Arrow and chemometrics for the characterization of the aroma profile of commercial bread

A SPME-Arrow GC-MS approach, coupled with chemometrics, was used to thoroughly investigate the impact of different types of yeast (sourdough, bear's yeast and a mixture of both) and their respective leaving time (one, three and five hours) on VOCs of commercial bread samples. This aspect is of paramount importance for the baking industry to adjust recipe modifications and production parameters, as well as to meet consumer needs in formulating new products. A deep learning approach, PARADISe (PARAFAC2-based deconvolution and identification system), was used to analyse the obtained chromatograms in an untargeted manner. In particular, PARADISe, was able to perform a fast deconvolution of the chromatographic peaks directly from raw chromatographic data to allow a putatively identification of 66 volatile organic compounds, including alcohols, esters, carboxylic acids, ketones, aldehydes. Finally, Principal Component Analysis, applied on the areas of the resolved compounds, showed that bread samples differentiate according to their recipe and highlighted the most relevant volatile compounds responsible for the observed differences. •The performance of Arrow-SPME device was investigated in terms of blanks, reproducibility, and sensitivity.•Experimental design technique was used for the optimization of Arrow-SPME sampling of bread aroma.•The influence of different yeasts and leavening times were evaluated on final products.•PARADISe approach allowed to rapidly identify 66 VOCs including alcohols, esters, carboxylic acids, ketones and aldehydes.•SPME Arrow fiber combined with GC-MS and chemometrics is a powerful tool for developing new bread formulations.