Aroma model

Quantification of the odorants and calculation of their OAVs are the next steps to develop an aroma model. 

---

The influence of the sensitivity of the assessors on AEDA has been studied [11], with the result that the differences in the FD factors determined by a group of six panellists amount to not more than two dilution steps (e.g. 64 and 256), implying that the key odorants in a given extract will undoubtedly be detected. However, to avoid falsification of the result by anosmia, AEDA of a sample should be independently performed by at least two assessors. As detailed in [6], odour threshold values of odorants can be determined by AEDA using a sensory internal standard, e.g. ( )-2-decenal. However, as shown in Table 16.6 these odour threshold values may vary by several orders of magnitude [8] owing to different properties of the stationary phases. Consequently, such effects will also influence the results of dilution experiments. Indeed, different FD factors were determined for 2-methyl-3-furanthiol on the stationary phases SE-54 and FFAP 2 and 2 , respectively. In contrast, 5-ethyl-3-hydroxy-4-methyl-2(5H)-furanone showed higher FD factors on FFAP than on SE-54 2 and 2, respectively. Consequently, FD factors should be determined on suitable GC capillaries [8]. However, the best method to overcome the limitations of GC-O and the dilution experiment is a sensory study of aroma models (Sect. 16.6.3). 

The highest OAVs were found for 4-hydroxy-2,5-dimethyl-3(2H)-fura-none, followed by ethyl 2-methylpropanoate, ethyl 2-methylbutanoate, methyl 2-methylbutanoate and ( ,Z)-l,3,5-undecatriene. It is assumed that these odorants contribute strongly to the aroma of pineapples [50]. However, FD factors and OAVs are functions of the odorants concentrations in the extract, and are not psychophysical measures for perceived odour intensity [71,72]. To take this criticism into account, aroma models are prepared on the basis of the results of the quantitative analysis (reviewed in [9]) and in addition omission experiments are performed [9]. 

In the case of pineapples, the 12 odorants listed in Table 16.7 were dissolved in water in concentrations equal to those determined in the fruit [50]. Then the odour profile of this aroma model was evaluated by a sensory panel in comparison to fresh pineapple juice. The result was a high agreement in the two odour profiles. Fresh, fruity and pineapple-like odour notes scored almost the same intensities in the model as in the juice. Only the sweet aroma note was more intense in the model than in the original sample [50]. In further experiments, the contributions of the six odorants showing the highest OAV (Table 16.7) were evaluated by means of omission tests [9]. The results presented in Table 16.8 show that the omission of 4-hydroxy-2,5-dimethyl-3(2H)-furanone, ethyl 2-methylbutanoate or ethyl 2-methylpropanoate changed the odour so clearly that more than half of the assessors were able to perceive an odour difference between the reduced and the complete aroma model. Therefore, it was concluded that these compounds are the character-impact odorants of fresh pineapple juice. 

The aroma model contains the odorants listed in Table 16.7. 

Number of panellists (out of 15) detecting an odour difference between the reduced and the complete aroma model in a triangle test... 

Most research on aroma recovery by organophilic pervaporation has been conducted using aqueous aroma model solutions [25-28], although in recent years significant interest has been devoted to the recovery of aroma compounds from natural complex streams, such as fruit juices [29-31], food industry effluents [32] and other natural matrixes [33]. The increasing demand for natural aroma compounds for food use, and their market value, opens a world of possibilities for a technique that allows for a benign recovery of these compounds without addition of any chemicals or temperature increase. However, in most situations, dedicated requests by industrialists are formulated in cooperation with marketing departments, which translate into the need for a correct public perception. 

Grosch, W. (2001). Evaluation of the key odorants of foods by dilution experiments, aroma models and omission. Chem. Senses, 26, 533-545. 

Preparation of a synthetic blend (aroma model) of the key odorants on the basis of the quantitative data obtained in step 5. Critical comparison of the aroma profile of the synthetic blend with that of the original... 

Comparison of the overall odour of the aroma model with that of models in which one or more components are omitted (omission experiments)... 

Table 6.28 Headspace odour profiles of a baguette and the corresponding aroma model [34]...

In the dilution experiments, on which the aroma models are based, the odour impact of the volatiles is evaluated after separation by GC (cf. 6.2.4.2.2). Perceptual interactions of odorants, which in most cases were characterised by inhibition and suppression [8, 36], are excluded by this procedure. Therefore, the question as to which compound among the volatiles evaluated by the dilution experiments actually contributes to the aroma has to be answered by omission experiments (step 7 in Table 6.23). Examples of such experiments are given in 6.2.4.3. 

To verify whether the volatiles listed in Table 6.32 are actually the key odorants, an aroma model was prepared by using an unripened cheese (UC) as base ]53J. The odorants and in addition the compoimds showing high taste activity values ]54] were quantified in UC and in Swiss cheese ]53]. The differences in the concentration of these compounds in both samples were calculated, and, accordingly, the compounds were dissolved in water and/or sunflower oil and then added to freeze-dried UC. The flavour model obtained agreed in colour, pH, water, protein and fat content with grated Swiss cheese, only the texture was more grainy ]53]. 

Aroma models were prepared for gruyere by applying the methods reported for emmental [57, 58], The aroma of the models was similar to that of the original cheeses. Consequently, it was concluded that 2-/3-methylbutanal, methional, dimeth-yltrisulphide, phenylacetaldehyde, 2-ethyl-3,5-dimethylpyrazine, methanethiol, as well as acetic, propionic, butyric, 3-methylbutyric and phenylacetic acids are the key odorants of gruyere. 

Table 6.36 Odour profiles of virgin olive oils and their aroma models [64]...

Aroma models prepared on the basis of the quantitative data shown in Table 6.37 agreed very well with the original oil samples (Table 6.36). The similarity scores amounted to 2.6 (oil I) and 2.7 (oil S), respectively. In these experiments [64] an odourless plant oil was used as the solvent for the odorants. Reduction of the aroma model for oil I to only seven odorants (nos. 3, 6, 7, 12, 16, 17, 19) lowered the similarity score to 2.2 but the characteristic overall odour remained was still preserved. In the case of oil S, a mixture containing only odorants nos. 1, 2, 8, 9 and 10 did not differ in the aroma from that of the complete aroma model. This result indicates that the other compounds quantified in oil S (Table 6.37) are not important for the aroma. 

An aroma model was very similar to the aroma of the juice when it contained 0.1% fat in addition to the volatiles listed in Table 6.38 [67], The terpene-like odour quality was decreased and the fruity note was enhanced. Omission experiments (Table 6.39) revealed that only the absence of acetaldehyde (no. 1) and (R)-limonene (no. 19) was detectable with high significance (a = 0.1) by either nasal or retronasal evaluation. Retronasally an aroma difference between the complete model and the model, in... 

Table 6.39 Odour differences between the complete aroma model and models containing various compounds less [67]...

All odorants listed were present in the complete aroma model. 

The odour intensities of volatiles showing similar odour qualities are partially additive [68]. To substantiate such additive effects, three groups of odorants (terpene hydrocarbons, esters or aldehydes) were omitted from the aroma model for orange juice. For all groups, a significant difference from the complete model was observed (Table 6.39). Omission of esters nos. 12,14 and 15 with ethyl butanoate (no. 13) still present was clearly detectable. This indicates that the fruity quality in the odour profile is enhanced by additive effects. In contrast, no difference was perceivable when (R)-a-pinene (no. 17) and myrcene (no. 18) were omitted. The concentration of the odorants in juice differs depending on the variety. Thus, the weaker citrus note of Navel oranges compared with the above discussed variety Valencia late is due to a 70% lower content of (R)-limonene [67]. 

In triangle tests the aroma model containing the 12 odorants was compared by six trained panellists with models lacking in one. N number of panellists detecting an odour difference. 

The volatiles detailed in Table 6.43 were identified in French fries prepared in palm oil [78, 79[. The flavour profile of a solution of these compounds in odourless sunflower oil was close to that of the original (Table 6.44). The greatest intensity differences were found for the caramel and malty notes when the model was retrona-sally examined. Variation of the composition of the aroma model indicated the odorants that are essential for the characteristic aroma [79[. An extract of the results is shown in Table 6.45. 

Table 6.45 Odour of the aroma model for French fries as affected by the absence of one or more odorants [79]...

Two aroma reconstitution experiments have been performed for tomato paste. In the first experiment [81 ] a synthetic mixture containing the odorants nos. 2, 8, 9, 15, 17 (3-isomer), 21 and l-nitro-2-phenylethane was used. The concentration of the first six odorants was in the range shown in Table 6.46. In the second experiment [82] the aroma model contained the 15 odorants in the concentration equal to those reported in Table 6.46. A comparison indicated that the aroma of this mixture was very close to the aroma of the original. Obviously, l-nitro-2-phenylethane does not contribute to the aroma of tomato paste. 

Compound present in the aroma model for black pepper [83]. 

The most important odorants of parsley leaves are listed in Table 6.48. An aroma model formulated on the basis of the quantitative data (Table 6.48) was described as clearly parsley-like [88]. Differences between the odour profile of the model and that of parsley leaves were observed for the spicy and the green-grassy notes which were stronger and weaker, respectively, in the model. The parsley-like character of the aroma model was completely lost when p-mentha-l,3,8-triene (no. 1 in Table 6.48) and myrcene (no. 2) were omitted [88],... 

Again choice and study will be done in close cooperation between the aroma and supports suppliers, the process engineer, and the flnal producer/user. Very often preliminary trials will be done using aroma model molecules (i.e., one or two, or oils), at a pilot scale, with different supports to study. Then the proper scale and conditions will be studied and chosen. 

The selection of odorants by dilution analyses (cf. 5.2.2) does not take into account additive (cf. 20.1.7.8) or antagonistic effects (example in Fig. 5.2) because the aroma substances, after separation by gas chromatography, are sniffed individually. Therefore, in view of the last mentioned effect, the question arises whether all the compounds occurring in the aroma model really contribute to the aroma in question. To answer... 

---