Study of the aroma of artificially flavoured custards by chemical sensor array fingerprinting

It is generally known that the chemical analysis of the volatile compounds of foods is not adequate to describe the complexity of flavours; as a consequence, sensory analysis is the fundamental methodology to determine quality and acceptability of food products. Nonetheless, instrumental estimations of human reactions to food are still highly demanded. To this regard the synthetic and qualitative representation of data originated by multivariate instruments has gained importance in the last years. In particular, fingerprints from chemical sensor arrays have been a subject of investigation aimed at developing instrumental measurements of human olfactory perceptions. In this paper, a study case on the application of a sensor array aimed at appraising the flavour properties of custard desserts seasoned with strawberry aroma is presented. These studies were intended to determine the effects of some key ingredients (milk fat content and sugar type) to the aroma and to the aroma temporal decay under controlled storage. Sensor data have been compared and complemented with those provided by gas-chromatography and sensory analysis. The results indicate that custard headspace fingerprinting can obviously recognize the addition of artificial flavours, and, more noteworthy, discriminate among fat milk content and sugar types used to prepare the samples. Furthermore, composition was also found to affect the temporal evolution of aroma evidencing distinct paths in a principal component analysis (PCA) score plot for samples prepared with different ingredients.