Unraveling ingredients in complex mixtures by chromatographic spectrum recognition: Application to perfume deformulation

In the past 40 years, progress in GC‐MS and GC‐FID has lent the modern perfume industry powerful tools for determining raw ingredients in fragrances, which is still often manual and based on molecular markers thus subjective and unreliable due to variability and chromatographic challenges. This article presents an alternative approach by introducing the concept of chromatographic spectrum. This approach was evaluated on a commercial database containing molecular composition of 4106 perfumery ingredients and on real and simulated mixtures. Five hundred and seven out of 565 database ingredients classes were differentiated by their chromatographic spectrum against 164 with distinctive markers. The 5 ingredients of a real Eau de Cologne mixture were identified, and their proportions estimated with less than 12% relative error for 4 of them. The usefulness of chromatographic spectra in a deformulation support algorithm is discussed based on the deformulation of 210 simulated mixtures of 5, 10, and 15 ingredients.