Predicting organoleptic scores of sub-ppm flavour notes. Part 1. Theoretical and experimental details

Most existing electronic nose systems have limited commercial application since they only provide a relative description of the flavour under investigation, rather than one against a universal standard. However, the development of a set of universally accepted standards for flavour description has been problematic due to the lack of any comprehensive model relating the molecular structure of an odorant with its flavour-impact during the act of perception. Instead, industries have tended to develop their own flavour models (flavour terminology systems) for specific consumer products that are based upon practical experience of a particular foodstuff, cosmetic, or beverage. We report here on the novel application of chemical multi-sensor arrays to the prediction of organoleptic flavour notes, as defined under a specific terminology system suitable for describing and communicating specific flavours. A novel odour mapping scheme is proposed that may be applied generally to multi-sensor arrays and provides more detailed characterisation of odour quality than is currently achievable. As part of our study, a flow injection analyser (FIA) system has been developed that combines chemical and electronic hardware driven by a microcomputer to achieve accurate and independent control over odour-stream temperature, flow-rate and flow profile, sensor head temperature, and sample times. An array of 24 conducting polymer sensors (11 different types) is used within the FIA system giving an overall experimental coefficient of variation below 7%. The application of this odour mapping technique is demonstrated by way of an experimental study, using the FIA system reported here. The details for this study are given in Part 1, and the computational analysis of the data is carried out in Part 2 (T. C. Pearce and J. W. Gardner, Analyst, 1998, 123, 2057.