Food Science and Technology Research
Online ISSN : 1881-3984
Print ISSN : 1344-6606
ISSN-L : 1344-6606
Original papers
Characterization of Potent Odorants in Three Different Cultivars (Madagascar, Comoro and Tahiti) of Vanilla Bean by Aroma Extract Dilution Analysis (AEDA)
Atsushi ShigetoShoko HachisukaKenji Kumazawa
Author information
Keywords: vanilla, flavor, aroma, cultivar, AEDA

2016 Volume 22 Issue 6 Pages 811-816


The aroma concentrates of three different cultivars of vanilla bean were prepared by combining the solvent extraction and solvent assisted flavor evaporation (SAFE) techniques. Aroma extract dilution analysis (AEDA) of the volatile fraction revealed 36 odorants, including nine newly identified vanilla odorants, which were identified or tentatively identified from the 49 odor-active peaks with FD factors between 41 and 49. Based on these results, it was proposed that the potent odorants, except for vanillin, were able to influence the characteristic flavor of vanilla beans from different growing regions.


Vanilla beans are used as an important raw material for the flavoring of various kinds of foods and fragrances. The pods are harvested from the vines of the Vanilla genus in the Orchidaceae family. While about 110 species of vanilla are currently known, only two species (V. planifolia and V. tahitensis) are cultivated commercially. V. planifolia is cultivated in many places of the world, such as Madagascar and Indonesia. In particular, Bourbon vanilla, which comprises the Madagascar, Comoro, Réunion cultivars, is well known as a high quality product. Meanwhile, V. tahitensis, which was recently reported to be a hybrid of V. planifolia and V. odorata (Lubinsky et al., 2008), is mainly cultivated in Tahiti and Papua New Guinea.

The most important characteristic of vanilla bean is its sweet attractive aroma, which has prompted numerous reports regarding its volatile compounds. More than 500 volatile compounds have been reported for vanilla (Klimes and Lampasky, 1976, Adedeji et al., 1993, Toth et al., 2011), with vanillin as the key aroma compound. Although there are some reports identifying the potent odorants of vanilla bean using the gas chromatography-olfactometry (GC-O) technique (Pérez-Silva et al., 2006; Zhang and Mueller, 2012; Brunschwig et al., 2012; Takahashi et al., 2013a,b), there are few studies (Zhang and Mueller, 2012; Brunschwig et al., 2012) comparing the odorant profiles of different vanilla cultivars using GC-O techniques such as aroma extract dilution analysis (AEDA). Therefore, the characteristic odorants of individual vanilla cultivars remain unclear.

In the present paper, we characterized the potent odorants of three different vanilla cultivars (Madagascar, Comoro and Tahiti) using AEDA. Furthermore, the relationship between species and the characteristic aroma profile was also investigated, focusing on 3- or 4-methoxylated aromatic compounds, in V. planifolia and V. tahitensis.

Materials and Methods

Materials and Chemicals    The Madagascar (16 – 20 cm, harvested in 2010), Comoro (14 – 17 cm, harvested in 2011) and Tahitian (14 – 20 cm, harvested in 2011) cultivars of vanilla bean were obtained from Kaneda Corporation (Tokyo, Japan).

The organic solvents (diethyl ether and pentane) used were of >99% purity (Kanto Chemical Co., Inc., Tokyo, Japan). 1, 2, 4, 5, 8, 9, 12, 13, 15, 19, 21, 22, 23, 25, 27, 28, 30, 31, 33, 35, 36, 38, 42, 44, 46, 48, 49, 50, 51, 52, 54 were purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). 3, 14, 20, 37, 45, 53 were purchased from Sigma Aldrich Japan Co., Ltd. (Tokyo, Japan). 6 (Czerny et al., 1996), 11 (Czerny et al., 1996), 16 (Guth and Grosch, 1990), 26 (Kumazawa et al., 2006), 47 (Kaneko et al., 2013) were synthesized according to the literature.

Isolation of Vanilla Volatiles    The vanilla aroma concentrates were obtained by organic solvent extraction, SAFE distillation and a concentration method. After removing the vanilla pod, 1 g of vanilla bean was suspended in 1 mL of distilled water, to which 20 mL of a mixture of pentane/diethyl ether (1:1 v/v) was added, and then the aroma compounds were extracted in an ultrasonic bath for 60 min.

This aroma extract was spiked with an internal standard (5 ng 2-octanol), dried using an excess amount of anhydrous sodium sulfate, and then distilled by solvent assisted flavor evaporation (SAFE) (40°C, <5.0 × 10−3 Pa). The vanilla aroma concentrate was obtained by rotary evaporation concentration, followed by nitrogen steam evaporation to about 100 µL. The resulting aroma concentrate was used as the sample for AEDA and GC-MS analyses.

Enrichment of Odorants for Identification    For the identification experiments, the vanilla volatiles were isolated from a larger amount of vanilla (50 g) by combining the solvent extraction and the SAFE technique as described above. The aroma concentrate (approximately 100 µL) was applied to a glass column (15°C, 20 cm × 0.7 cm i.d.) filled with 5 g of silica gel (Wakogel® C-200, Wako Pure Chemical Industries, Ltd., Tokyo, Japan) in n-pentane. Elution was performed using the following solvents: n-pentane (30 mL, fraction A), n-pentane/diethyl ether (30 mL, 10 + 1, v/v, fraction B), n-pentane/diethyl ether (30 mL, 10 + 2, v/v, fraction C), n-pentane/diethyl ether (30 mL, 1+1, fraction D) and diethyl ether (30 mL, fraction E). The solution was concentrated to approximately 100 µL as described above.

Gas Chromatography-Olfactometry (GC-O)    An Agilent 6850 gas chromatograph equipped with a thermal conductivity detector (TCD) and DB-WAX (J&W Scientific, California, USA) fused silica capillary column (30 m, 0.25 mm i.d., 0.25 µm film thickness) was used for GC-O analysis. One microliter of the sample was injected (splitless mode). The column temperature was programmed from 40 to 210°C at the rate of 5°C/min. The injector and detector temperatures were both 250°C. Helium was used as the carrier gas at the flow rate of 1 mL/min. A glass sniffing port was connected to the outlet of the TCD and heated to >210°C by a ribbon heater. Moist air was pumped into the sniffing port at 100 mL/min to quickly remove the odorant eluted from the TCD out of the sniffing port. The vanilla aroma concentrates were used for GC-O analysis. Determination of the odor qualities detected by sniffing was achieved by triplicate determinations for each sample.

Aroma Extract Dilution Analysis (AEDA)    The original odor concentrate of the vanilla aroma was stepwise diluted with dichloromethane to 4n (n = 1 – 9), and 1 µL of each fraction was analyzed by capillary GC on a DB-WAX column. The odorants were then detected by GC eluate sniffing (GC-O). The flavor dilution (FD) factors of the odorants were determined by AEDA. Before the FD factor measurement, three panelists repeatedly checked the retention time and odor quality of the odorants using each diluted sample (41 and 42), then the FD factor of the odorants was determined by detection of not less than two panelists at each dilution step.

Gas Chromatography–Mass Spectrometry (GC-MS)    An Agilent 7890A gas chromatograph equipped Agilent 5975C inert XL series MSD and DB-WAX fused silica capillary columns (60 m, 0.25 mm i.d., 0.25 µm film thickness) was used for identification and semi-quantification purposes. The column temperature was programmed from 80 or 40°C to 230°C at the rate of 3°C/min. The injector temperature was 250°C. Helium was used as the carrier gas at the flow rate of 1 mL/min. One microliter of the sample was injected, and the split ratio was 1:30 or splitless. Mass spectrometry was conducted at an ionization voltage of 70 eV (EI) and an ion source temperature of 150°C.

Identification and Semi-quantification of Aroma Compounds    The identification of each key aroma compound was performed by comparing its Kovats GC retention index (RI) and mass spectrum to those of the authentic compounds by GC-MS, in addition to comparison of its RI and odor quality with those of the authentic compounds by GC-O. Semi-quantitative data of the respective key aroma compounds were calculated on the basis of the internal standard material using the estimated response factor of 1 by GC-MS.

Results and Discussion

Potent odorants of the three different vanilla cultivars (Madagascar, Comoro and Tahiti)    Three different cultivars, belonging to V. planifolia (Madagascar, Comoro) and V. tahitensis (Tahiti), of commercial vanilla beans were used. The aroma of each vanilla bean had a different character, such as fatty and whisky-like for Madagascar, smoky for Comoro, and spicy and anise-like for Tahiti.

The volatile concentrates of the vanilla beans were prepared by combining the solvent extraction and SAFE techniques. The extraction solvent was optimized by comparing and assessing the aroma between the vanilla beans themselves and their extracts. Although the concentration process could result in the loss of some of the more volatile organic compounds, the volatile concentrates obtained using this sampling operation (without excessive heating) reproduced well the characteristic aromas of the vanilla beans. Thus, this was regarded as the preferred method to obtain analytical samples for comparing individual potent odorants using the GC-O technique.

The AEDA technique was applied to the volatile concentrates prepared from the three different vanilla bean cultivars (Madagascar, Comoro and Tahiti), and 49 odor-active peaks with FD factors between 41 and 49 were determined, thus confirming that the aroma of the vanilla beans consisted of many potent odorants (Table 1.). However, since many of the peaks of the unknown odorants on the gas chromatogram overlapped with other compounds (co-eluted), it was difficult to quantitate and qualitate these odorants at low levels. Therefore, to identify these low amounts of unknown odorants, highly concentrated volatile fractions were prepared from 50 g of Madagascar vanilla bean. The obtained volatile concentrate was fractionated into five fractions (from A to E) by silica gel column chromatography, so that these additional small amounts of odorants could be identified or tentatively identified by GC-MS and GC-O analyses. Finally, in these volatile concentrates, 36 odorants could be identified or tentatively identified from 49 odor-active peaks by GC-MS and GC-O. Besides these potent odorants, which included nine newly identified odorants (3, 4, 6, 8, 11, 15, 26, 38, 47), there were some important and unidentified remaining odorants in the vanilla aroma.

Table 1. Odorants (FD ≥ 4) of the vanilla beans in three cultivars (Madagascar, Comoro and Tahiti)
No. RIa Compoundb Odor quality FD factorc
Madagascar Comoro Tahiti
1 930 2- and 3-methylbutanald stimulus, malty 4 4 4
2 1296 1-octen-3-one mashroom-like 16 4 nde
3 1388 2-ethyl-6-methylpyrazinedf nutty, fruity 4 4 nde
4 1428 2-isopropyl-3-methoxypyrazinedf peacy 16 4 4
5 1439 acetic acid sour 4 16 4
6 1462 2-ethyl-3,5-dimethylpyrazinedf nutty 16 4 4
7 1502 NI putrid 4 16 4
8 1521 2-isobutyl-3-methoxypyrazinedf earthy, musty 64 64 16
9 1530 (E)-2-nonenald sweet, fatty 64 4 4
10 1541 NI lemon-like nde nde 4
11 1555 2-ethenyl-3,5-dimethylpyrazinedf nutty 16 4 4
12 1622 butanoic acid sweaty, rancid 4 4 nde
13 1660 3-methylbutanoic acid sweaty, rancid 64 64 16
14 1665 NI meaty, vitamine-like 16 16 nde
15 1696 (E,E)-2,4-nonadienalf sweet, fatty 256 16 nde
16 1716 3-methyl-2,4-nonanedioned green 4 16 4
17 1723 NI sweet, lactone-like 16 nde nde
18 1761 NI sweet 4 nde nde
19 1805 (E,E)-2,4-decadienal sweet, fatty 64 16 nde
20 1817 β-damascenone sweet 64 64 nde
21 1837 hexanoic acid sour, green nde 4 4
22 1851 guaiacol burnt 1024 65536 256
23 1913 4-octanolide sweet, juicy, lactone-like 4 4 nde
24 1938 NI sweet, juicy nde 4 nde
25 1948 2-methoxy-4-methylphenol burnt, spicy 16 64 nde
26 1996 trans-4,5-epoxy-(E)-2-decenaldf sweet, juicy 16 4 4
27 2020 anisaldehyde spicy, sweet, cherry 64 64 4096
28 2023 4-nonanolide sweet 64 64 nde
29 2052 NI metalic nde 4 nde
30 2073 p-cresol phenolic 256 256 16
31 2070 methyl cinnamate fruity, strawberry-like 4 16 nde
32 2102 NI burnt, spicy 4 nde nde
33 2124 ethyl cinnamate fruity, strawberry-like 256 1024 64
34 2141 NI lactone nde 4 nde
35 2146 anisyl acetate sweet, anise-like 4 4 64
36 2157 eugenol spicy 1024 1024 16
37 2185 2-methoxy-4-vinylphenol spicy 4 nde nde
38 2190 3-hydroxy-4,5-dimethyl-2(5H)-furanonef caramel-like 256 64 64
39 2236 NI juicy, sweet, butter-like 4 4 nde
40 2247 NI powderly nde 4 nde
41 2259 NI spicy nde 4 nde
42 2274 anisyl alcohol spicy, anise-like, cherry 16 16 1024
43 2304 NI meaty 4 nde nde
44 2446 coumarin sweet 4 4 16
45 2544 phenylacetic acidd sweet 4 4 nde
46 2551 vanillin vanilla-like 262144 262144 16384
47 2561 4-vinyl-2,6-dimethoxyphenoldf spicy, smoky 64 16 16
48 2613 3-phenylpropionic acidd animal-like 4 16 16
49 2686 isovanillin spicy, smoky nde nde 256

*NI, not identified.

a  RI, retention index calculated on DB-WAX columnusing C5-C32 n-alkanes.

b  The compound was identified by comparing its RI and mass spectrum with the authentic compound by GC-MS in comparison of its RI and odor quality with the authentic compound by GC-0.

c  FD factor, flavor dilution factor.

d  The compound was tentativoly identified by GC-0 analysis by comparison to the authentic compound, but no unequivocal mass spectrumwas available by GC-MS.

e  nd, not detected

f  The compound was newly identified in the vanilla beans.

Because the odor thresholds of the newly identified compounds were extremely low, these odorants in the vanilla can be assumed to be too low in concentration and thus difficult to identify. Pyrazines (3, 6, 11) and 3-hydroxy-4,5-dimethyl-2(5H)-furanone (38) are known to be produced from sugars and amino acids through multistep reactions (Koehler et al., 1969; Huang et al., 1994; Blank et al., 1996). Unsaturated aldehydes (15, 26) are reported to be generated from unsaturated fatty acids, such as linolenic acid, by autoxidation and enzymatic oxidation (Belitz et al., 2004). Additionally, it is well known that methoxypyrazines (4, 8) are produced by enzymatic reaction (Dunlevy et al., 2010). Therefore, the contents of these compounds are assumed to be dependent on their precursor contents and the manufacturing conditions of the vanilla bean.

As a result of the AEDA, vanillin (46) was obviously the most important odorant for each cultivar, which is consistent with previous reports. The odorant composition of cultivars of the same species (Madagascar and Comoro) was similar; various phenyl compounds (guaiacol (22), p-cresol (30), ethyl cinnamate (33) and eugenol (36)), which are derivatives of phenylpropanoids, showed high FD values in both vanillas. In contrast, certain compounds were characteristic for individual cultivars. For example, 2-alkenals and 2,4-dienals (9, 15, 19) in Madagascar vanilla, reported as lipid oxidation products formed during the curing process (Dunphy and Bala, 2014), were found to be higher than in Comoro vanilla. Meanwhile, guaiacol (22) showed the highest FD value, especially in Comoro vanilla, and was assumed to be responsible for the characteristic smoky note.

In food, guaiacol is reported to be produced by microorganisms or thermal degradation (Huang et al., 1993; Witthuhn et al., 2012; Belitz et al., 2004). The curing of vanilla beans generally includes fermentation and sun drying processes. Therefore, the higher guaiacol content of Comoro vanilla compared to Madagascar vanilla is assumed to be due to differences in curing conditions, such as the bacterial flora or drying conditions, between Madagascar and Comoro. In addition, plant maturity affects the lipoxygenase activity, which is involved in the production of 2-alkenals and 2,4-dienals (Belitz et al., 2004). Thus, it was assumed that differences in cultivation conditions between Madagascar and Comoro affects the aroma profile of vanillas. However, we compared only one sample of each cultivar in this study. In order to confirm the accuracy of these determinations, a larger study that compares various kinds of vanilla products is needed.

On the other hand, the Tahitian vanilla differed in odorant composition from V. planifolia. It had fewer odorants than V. planifolia, and four odorants (27, 35, 42, 49) showed characteristically high FDs. Surprisingly, these four odorants had the 4-methoxy aromatic structure, which is assumed to contribute the spicy and anise note.

The results of the AEDA indicated that a variety of trace and low-odor threshold odorants play an important role in the characteristic aroma of individual vanilla cultivars. It is assumed that differences in the aroma profile were derived from differences in the species, mainly in the curing process and growth conditions. Further clarification of the characteristics of each vanilla cultivar could be achieved by a more comprehensive investigation and comparison of the cultivars.

The 3- or 4-methoxylated aromatic compounds responsible for species differences in vanilla aroma    Based on the AEDA results, some compounds with the methoxybenzene (anisole) skeleton (25, 27, 35, 36, 37, 42, 46, 47, 49) were assumed to be responsible for the vanilla aroma. Table 2 shows the concentrations of anisole compounds (Figure 1.) in the three different cultivars. V. planifolia had a high content of 3-methoxylated compounds, such as vanillin and eugenol, while V. tahitensis contained abundant 4-methoxylated compounds, such as anisyl alcohol, anisaldehyde and isovanillin. This result suggests that V. planifolia and V. tahitensis have different biosynthesis routes for these aromatic compounds.

Table 2. The concentrations of the 3 or 4-position methoxylated compounds in the vanilla beans
No. Compounds Concentration (µg/g)a
V. planifolia V. tahitensis
Madagascar Comoro Tahiti
3-positionb methoxylated compounds 25 2-methoxy-4-methylphenol 2 6 0.5
36 eugenol 0.6 0.7 nd
37 2-methoxy-4-vinylphenol nd 1 0.1
46 vanillin 6201 8053 1501
50 acetovanillone 4 5 1
51 vanillyl alcohol 4 nd 1
4-positionb methoxylated compounds 27 anisaldehyde 0.3 0.3 19
35 anysyl acetate nd 0.3 14
42 anisyl alcohol 8 6 1175
49 isovanillin nd nd 34
52 methyl anisate nd 0.5 3
53 anisyl formate nd nd 0.9
54 anisic acid nd nd 238

nd, not detected

a  The concentration was determined by the ratio of the percentage peak area of each compound to the internai standard (2-octanol)

b  The number of positions was counted regarding any carbonio group (CH3, CH=CHCH2, CHO, COCH3, CH2OH, COOCH3, COOH, CH2CHO,CH2OAc) as the 1-position.

Fig. 1.

The structure of the 3 or 4-position methoxylated compounds in the vanilla beans

Vanillin is generally thought to be biosynthesized via a multi-step conversion from phenylalanine (Podstolski, 2011; Negishi et al., 2009). The 3-methoxylation of vanillin was achieved by 3-orthomethyltransferase (3-OMT). 3-OMT is thought to be activated in V. planifolia to a greater extent than in V. tahitensis. In addition, enzymes that methylate the phenol group in the 4-position (4-OMT) are thought to be present or highly active in V. tahitensis.

Although many kinds of 4-OMTs have been reported in plants (Koeduka, 2010), there has been little discussion of 4-OMT in vanilla. Besides, the biosynthesis route of aroma compounds in V. tahitensis has not yet been reported. In the future, investigation of vanilla enzymes related to the methoxylation of the phenol group will be important for further elucidation of the aroma profile of various cultivars of vanilla.

On the basis of these results, to better clarify the characteristic vanilla aroma of each cultivar, it is necessary to take into account not only vanillin, but also trace odorants such as those newly identified in this study. However, these results were obtained from only one product sample of each cultivar. To improve the accuracy of the flavor profile of each vanilla cultivar, it is important to understand the potent odorants of various vanilla products. Therefore, a more detailed study of the various kinds of vanilla products, such as from different raw materials and manufacturers, and their manufacturing conditions is needed.

© 2016 by Japanese Society for Food Science and Technology