Characterisation of vanilla extracts based on sensory properties and chemical composition : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Food Technology at Massey University, New Zealand

Abstract

Although vanilla is one of the most commonly used flavourings in the world, there is only limited information available about its flavour and chemical composition. The aims of this research were to use sensory analysis and chemical composition analysis to characterise vanilla extracts produced from beans from different regions and to investigate correlations between sensory data and chemical composition of the vanilla extracts. Other aims were to investigate the effect of solvent extraction, concentration of extracts and the combination of vanilla and fat or sugar on the sensory profile of vanilla extracts and formulated matrices. The vanilla extracts (ethanol or glycerol based), either commercial or laboratory extracted samples using vanilla beans sourced from India, Madagascar, Papua New Guinea, Tonga and Uganda, were characterised for aroma and flavour by a sensory panel trained. The panel found that the aroma and flavour of vanilla extracts varied depending on both the growing region and the solvent or solvent concentration used for flavour extraction.Principal component analysis (PCA) showed that extracts from Madagascar and Tonga grown beans were similar, being high in sweet type aromas and flavours such as butterscotch flavour and raisin aroma. The xxtracts from India and Papua New Guinea beans were higher in the woody and bourbon notes. Glycerol extracts had a reduced aroma and flavour sensory profile intensity compared to the ethanol extracts. A range of concentrated vanilla extracts concentrated using vacuum concentration, maltodextrin flavour encapsulation and supercritical carbon dioxide extraction. The vacuum concentrate extract was the most similar to the standard single fold ethanol as found from sensory analysis and gas chromatography mass spectrometry (GCMS). Concentrations up to 35mg/ml of vanillin were reached with vacuum concentration, 20 times the concentration of 1.5mg/ml typically found in a single fold extract. The trained sensory panel were also asked to evaluate solutions containing vanilla extract, with either milk fat or sugar at different concentrations. Milk fat was found to reduce the aroma and flavour intensity of the vanilla and sucrose was found to increase the perception of vanilla extract aroma due to the ‘salting out’ effect and reduce bitter, woody and bourbon flavours while increasing butterscotch, raisin and vanilla flavours. Using GCMS it was found that more lower boiling point volatile compounds were extracted with polar solvents, such as methanol, than non-polar solvents, such as hexane. Fifteen of these volatile compounds were identified and quantified in 16 vanilla extracts and correlated withthe sensory attributes previously used to describe the vanilla extracts. Most of the volatile compounds in the vanilla extracts had a phenolic structure. Bourbon aroma and flavour correlated with syringaldehyde. Vanillyl alcohol was correlated with raisin aroma, raisin flavour and butterscotch flavour. Vanilla aroma and flavour were associated with creosol and vanillin. Sweet flavour was correlated with p-anisic acid, maltol and 4-hydroxybenzoic acidIn model systems with milk fat, vanilla and sugar, milk fat masked vanilla flavour and aroma in solution, whereas sugar enhanced the aromas and sweet flavours and only masked bitter, woody and bourbon flavour. Using GCMS, multiple correlations between volatile chemical compounds and sensory attributes were identified. Vanilla extracts characterised by sensory and analytical methods were found to vary based on vanilla bean region, solvent extraction conditions and concentration method used.