1822 J. A@. Food Chem.

1002, 40, 1922-1924

Vanillin and Related Flavor Compounds in Vanilla Extracts Made from

Beans of Various Global Origins
Arvind S. Ranadive
Premier Vanilla, Inc., East Brunswick, New Jersey 08816

Vanilla beans from seven different vanilla growing regions of the world were analyzed for vanillin,
p-hydroxybenzoic acid, p-hydroxybenzaldehyde, and vanillic acid. Beans were extracted to obtain
singlefold extracts, and analysis was carried out using reversed-phase liquid chromatography. Analysis
of the extracts, obtained from cured and uncured beans treated with @-glucosidase,indicates that all
of the above-mentioned compounds are present in green beans as glycosides and are released upon
curing. In addition to glycosides of these four known monophenols, there are a t least three other major
glycosides in green vanilla beans which are hydrolyzed during the curing process.

INTRODUCTION Another aim of this work was to confirm the presence

of these aromatic volatile compounds as glycosides in green
The characteristic flavor and aroma of vanilla, developed Jamaican beans and that they are released in their free
in properly cured beans, is the result of a number of form upon curing. The data obtained are presented in
biochemical and chemical transformations. More than this paper.
170 volatile aromatic compounds have been identified in
Madagascar beans (Klimes and Lamparsky, 1976). Of EXPERIMENTAL PROCEDURES
these, vanillin is the most abundant. Among the other
major volatile constituents of vanilla aroma are p- Apparatus and Reagents. ( a ) The liquid chromatograph
consisted of an Applied Biosystems Spectroflow 400 solvent
hydroxybenzoic acid, p-hydroxybenzaldehyde, vanillic delivery system, a Spectroflow480 injector/valvemodule, and a
acid, p-hydroxybenzyl alcohol, and vanillyl alcohol. Spectroflow 757 absorbance detector with UV detection set at
Guarino and Brown (1985)have developed a liquid 254 nm.
chromatographic method that can be routinely and ( b ) Column. An Applied Biosystems Brownlee MPLC RP-8
efficiently used for the quantitative analysis of vanillin, Spheri-5 column (100 X 4.6 mm, catalog no. OS-MP)) packed
p-hydroxybenzoic acid, p-hydroxybenzaldehyde, and van- with Cs stationary phase, 5-pm particle size, was used in
illic acid. They analyzed 13commercially available vanilla conjunction with a MPLC NewGuard cartridge (15 X 3.2 mm,
extracts and reported the quantitative data. Hermann catalog no. G08-013) containing Ca stationary phase.
(c) Solvents used were LC grade methanol (J. T. Baker), LC
and Stkkli (1982)reported a high-performance liquid grade water (Waters),95% ethanol, and glacial acetic acid (Fisher
chromatography (HPLC) method for the control of vanilla Scientific).
products through quantitative analysis of p-hydroxybenzyl (d) The mobile phase used was methanol-acidified water (10
alcohol, aldehyde, and acid, vanillin and vanillyl alcohol, + 90) pumped at 1.5mL/min. Water was acidified by adding 10
ethyl vanillin, and coumarin. Wallace (1983)used an mL of glacial acetic acid/800 mL.
HPLC method to quantitatively analyze vanillin and seven (e) Sample filters used were 0.45-pm nylon 66 membrane
other related phenolic compounds produced during the alcohol-compatible filters (Rainin).
manufacture of vanillin from pulp mill effluent. Standards. Vanillin (Sigma), p-hydroxybenzoic acid,
A correlation was shown to exist between vanillin and p-hydroxybenzaldehyde, and vanillic acid were obtained from
Kodak. p-Hydroxybenzyl alcohol and vanillyl alcohol were
p-hydroxybenzaldehydecontent of Madagascar vanilla by obtained from Aldrich. Caution should be exercised while using
Jurgens (1981). He suggested the use of this ratio to these chemical irritants.
identify the geographic origin of beans in an extract. Archer Standard Preparation. Vanillin (1.2000 g), vanillic acid
(1989)has suggested a similar correlation between vanillin (0.0800 g), p-hydroxybenzoic acid (0.0200 g), and p-hydroxy-
and p-hydroxybenzoic acid and between vanillin and benzaldehyde (0.0600 g) were weighed into a 100-mLvolumetric
vanillic acid for genuine vanilla essences. flask and diluted to volume with 95% ethanol. A 10-mL aliquot
Vanillin, vanillyl alcohol, and one other aroma constit- was further diluted to 100 mL with 40% ethanol and filtered
uent of Bourbon vanilla were found to be present in their through 0.45-pm nylon 66 filter prior to injection into chro-
glycosidicforms in green beans by Goris (1947). Leong et matograph.
Extract Preparation. Approximately 100g of cured vanilla
al. (1989) found vanillin, p-hydroxybenzoic acid, p- pods was finely chopped without crushing. A 10-g sample of the
hydroxybenzaldehyde, and vanillic acid only in their chopped beans waa used for extraction with 75mL of 44 76 aqueous
glycosidic forms in ripe green Bourbon beans. Sagrero- ethanol for 48 h at 45 "C. The mixture was occasionally stirred
Nieves and Schwartz (19881,on the other hand, showed during this time. The mixture was then filtered and the filter
the presence of significant levels of free vanillin and cake pressed and washed with 36 % ethanol until the total volume
p-hydroxybenzaldehyde in maturing green Mexican beans. of filtrate and washings was 100 mL (this is singlefold vanilla
The objective of this work was to collect quantitative extract).
data on the major aromatic volatile5 in the cured vanilla Enzyme Treatments. Uncured beam. Jamaican green
beans from Madagascar, Indonesia, Mexico, Jamaica, vanilla beans were provided by Elan, Inc., Newark, NJ, and were
kept refrigerated after harvest until used for these experiments.
Costa Rica, Tonga, and Tahiti. Except for Tahiti beans One hundred grams of green beans was immersed in a boiling
which belong to Vanilla tahitensis species, beans from all water bath for 5 min to inactivate all native enzymes. After
other world origins were of Vanilla fragrans (Salisbury) cooling to ambient temperature, the beans were pureed in a
species, also known as Vanilla planifolia Andrews. Waring Blendor and separated into two equal portions. To one

0021-856 119211440-1922$03.0010 0 1992 Amerlcan Chemical Society

Vanilla Extracts J. AQY~C.~oodchem.,VOI. 40, NO. io, 1902 192s

Table I. important Flavoring Components of Cured Vanilla Beans of Different Geographic Origins (Milligrams per 100 mL
of Singlefold Extract)
bean source p-hydroxybenzoic acid p-hydroxybenzaldehyde vanillic acid vanillin HPLC method vanillin W method
Madagascar 5.6 13.7 15.0 164.0 184.0
Indonesia 3.4 9.3 7.7 117.0 131.0
Mexico 4.0 7.0 13.0 90.0 100.0
Costa Rica 5.2 14.0 12.0 135.0 161.0
Jamaica n.d. 8.4 4.2 216.0 265.0
Tonga 2.1 10.0 7.6 197.0 320.0
Tahiti 32.8 13.0 4.4 103.0 120.0
commercial extract 43.8
1 2.2 8.5 7.5 133.0
2 2.0 10.6 8.3 154.0
3 1.5 9.2 6.8 133.0

portion was added 20 mL of &glucosidase solution (0.25 mg of Table 11. Ratios of Vanillin to Other Important
Flavoring Components
enzyme/mL) to obtain an approximate concentration of 0.5 mg
of enzyme/g of dry bean weight, followed by thorough mixing ratio of vanillin to
and incubation at 40 "C for 1 h. The mixture was then dried to p-hydroxybenzoic p-hydroxy- vanillic
.
a moisture content of less than 5 % The second portion was bean source acid benzaldehyde acid
treated similarly except 20 mL of water containing no enzyme
was added to the puree. Both the enzyme-treated and control Madagascar 29.3 12.0 10.9
dry beans were then extracted as described under Extract Indonesia 34.4 12.6 15.2
Preparation to obtain singlefold extracts. Mexico 22.5 12.9 6.9
Cured Beans. Madagascar, Tonga, Jamaica, and Tahiti beans Costa Rica 26.0 9.6 11.3
were chopped to about '/*-in. pieces. To 20 g of each of the Jamaica n.d. 25.7 51.4
Tonga 93.8 19.7 25.9
chopped beans was added 25 mL of &glucosidase solution (0.25 Tahiti 2.4 7.9 23.4
mg of enzyme/mL) to obtain an approximate concentration of
0.5 mg of enzyme/g of dry bean weight, and the beans were
incubated at 40 OC for 1h. The enzyme-treated beans were then indicate that the maturity, geographic origin, or curing
extracted with aqueous ethanol to obtain singlefold extracts. proceases may have little effect on their formation. Overall
Singlefoldextracts were also prepared from the beans not treated lower values for all four parameters for the three com-
with enzyme. mercial extracts may be the reflection of inadequate
Sample Preparation and Calculations. A 10-fold dilution amounts of beans used for their preparation.
of the singlefold extract, containing less than 0.3 g of vanillin/ The most significant difference between V.fragram
100 mL of extract, was made using 40% aqueous ethanol. The
samples were filtered through 0.45-pm nylon 66 filters. Concen- and V.tahitemisbeans wasseen in theirp-hydroxybenzoic
trations of the major components in the samples were calculated acid content, which appears to be about 10 times higher
by peak area proportioning using external standards. in V. tahitemis beans. V.tahitensis beans also contain
anisic acid, anisic aldehyde, and heliotropin, all of which
RESULTS AND DISCUSSION are not detected in V. fragram beans.
Quantitative data on vanillin, p-hydroxybenzoic acid, The ratios of vanillin to p-hydroxybenzoicacid, vanillin
p-hydroxybenzaldehyde, and vanillic acid present in to p-hydroxybenzaldehyde, and vanillin to vanillic acid
singlefold extracts made from cured vanilla beans of are given in Table 11. The ratio of vanillin to p-hydroxy-
different geographicorigins are presented in Table I. Both benzaldehyde has been used to authenticate vanilla
p-hydroxybenzyl alcohol and vanillyl alcohol were also extracts, and its range is reported to be 10.9-18.4 (Jurgens,
found to be present in all extracts, but, due to their 1981). This ratio for Madagascar, Indonesia, Mexico, Costa
inadequate separation from other peaks in our method, it Rica, and Tonga beans appears to be consistent with the
was not possible to obtain quantitative data on these range reported by Jurgens. As can be seen from the data
compounds. Although only one batch of beans per in Table 11,other ratios did not show any particular trend.
geographic source was used, representative samples from Archer (1989), working on commercial and laboratory-
large batches (1-3 tons) of beans from each geographic prepared Bourbon and Java vanilla extracts, also observed
area were used for this analysis. Also, reproducibility of much larger variation in the vanillin to p-hydroxybenzoic
the extraction procedure was ascertained by replicate acid and vanillin to vanillic acid ratios than in the vanillin
extractions of Madagascar beans. It should be mentioned, to p-hydroxybenzaldehyde ratio. It is not surprising to
however, that the results obtained from only one sample see the inconstancy in the ratios for beans of different
per bean origin allow the most general conclusions to be regions since their vanillin content, which is dependent
drawn. on bean maturity a t harvest, environmental factors, and
Beans of V. fragrans species of different geographic curing practices, showed large variation.
origins showed little variation in p-hydroxybenzoic acid The data on Jamaica vanilla beans, green and after
and p-hydroxybenzaldehyde content. Their vanillin and curing, presented in Table I11show that all four important
vanillic acid contents, however, showed considerable flavoring componentsincreased after the beans were cured.
variation. It is known that the bean maturity at harvest The same components increased dramatically when green
and the manner in which the beans are cured have beans whose native enzymes were heat inactivated were
considerable effect on their vanillin content (Ranadive et treated with added @-glucosidase. The results indicate
al., 1983). Since the Tonga, Jamaica, and Indonesia beans that these four flavoring components of vanilla are present
showed the vanillin content of fully matured adequately in green beans in their glycosidic forms and are produced
cured beans, the presence of lower vanillic acid in these in their free form by the native glycosidases during curing.
beans may be related to their geographic origin. Smaller It is interesting to see that the beans treated with added
variations in p-hydroxybenzoic acid and p-hydroxy- enzyme produced higher amounts of these components
benzaldehyde contents in the beans which showed con- compared to beans cured normally (with native enzymes
siderable variation in their vanillin and vanillic acid content system intact and no added enzymes). This may be due
1824 J. A@c. F d C h e m . , Vol. 40, No. 10, 1992 Ranadlve

Table 111. Determination of Selected Important Flavoring ens our observations that the hydrolysis of glycosides is
Components of Green, Cured, and Enzyme-Treated Killed not completely accomplished in the normal curing process.
Green Jamaica Vanilla Beans (Milligrams per 100 mL of Tahiti beans did not show increase in the vanillin content
Singlefold Equivalent Extract)
upon enzyme treatment, probably due to complete hy-
p-hydroxy- p-hydroxy- vanillin drolysis of glucovanillin during commercial curing. This
benzoic benzal- vanillic HPLC could also have happened if the added enzyme was
bean source acid dehyde acid method
inhibited by Tahiti bean vanilla phenolics.
Jamaican 0.00 3.5 0.3 40.0
green bean ACKNOWLEDGMENT
extract
Jamaica lab-cured 1.2 11.1 4.8 240.0 I thank Elan, Inc., of Newark, NJ, for use of their
bean extract laboratory facility. I am grateful to Mr. George Kapp of
Jamaica killed 2.8 22.5 8.9 370.0 Elan, Inc., for providing Jamaican green vanilla beans used
green bean in this work.
extract after
external LITERATURE CITED
j3-glucosidase
treatment Archer, A. W. Analysis of vanilla essences by high-performance
liquid chromatography. J. Chromutogr. 1989,462,461.
Table IV. Effect of &Glucosidase Treatment of Cured Goris, M. A. Formation du parfum de la vanille. Ind. Parfum.
Vanilla Beans on the Vanillin Content 1947,2,4.
vanillin, mg/100 mL of extract Guarino, P. A.; Brown, S. M. Liquid Chromatographic Deter-
mination of Vanillin and Related Flavor Compoundsin Vanilla
treated with Extract: Cooperative Study. J.Assoc. Off. Anal. Chem. 1985,
bean source control &glucosidase 68,1198.
Madagascar 171 182 Herrmann, A.; Stdckli M. Rapid control of vanilla-containing
Tonga 147 165 products using high-performance liquid chromatography. J.
Jamaica 229 285 Chromatogr. 1982,246,313.
Tahiti 80 80 Jurgens, U. The vanillin/p-hydroxybenzaldehyde ratio in bourbon
vanilla. Lebensmittelchem. Gerichtl. Chem. 1981,s(5),97.
to (a) inadequate amounts of native enzymes present in Klimes, I.; Lamparsky, D. Vanilla Volatiles-A Comprehensive
green beans, (b)inadequate amounts of enzyme-substrate Analysis. Int. Flavours Food Addit. 1976,Nov/Dec, 272.
contact, or (c) poisoning of the native enzymesby oxidized Leong, G.; Archavlis, A,; Derbesy, M. Research on the Glucoside
phenols produced simultaneously during curing. This Fraction of the Vanilla Bean. J.Essent. Oil Res. 1989,1,33.
Ranadive,A. S.;Szkutnica,K.; Guerrera, J. G.; Frenkel, C. Vanillin
oxidation of phenols was (unintentionally) kept to a Biosynthesis in Vanilla Beans. Proceedings of the 9th
minimum in the enzyme-treatedbeans in our experiments, International Congress of Essential Oils, Singapore; Allured
which helped added enzyme to complete its process more Publishing: Wheaton, IL, 1983;Book 2, p 147.
effectively. Sagrero-Nieves,L.; Schwartz, S. J. Phenolic Contents of Vanilla
If the native vanilla bean enzymes do not complete the planifolia as Affected by Harvest Period. J. Food Compos.
hydrolysis of vanilla glycosides during normal curing,then Anal. 1988,1, 362.
the commerciallycured beans should produce more vanillin Wallace, E. M. HPLC of Vanillin and Related Phenols. J.
if treated with external enzyme prior to extraction than Chromatogr. Sci. 1983,21,139.
the nontreated beans. To test this hypothesis, commer-
cially cured Madagascar, Tonga, Jamaica, and Tahiti beans Received for review March 30,1992. Revised manuscript received
were treated with @-glucosidasedissolved in water 1h prior July 8, 1992. Accepted July 20,1992.
to extraction with aqueous ethanol. Control treatments Regietry No. Vanillin, 121-33-5; p-hydroxybenzoic acid, 99-
were carried out without the enzyme. The results in Table 96-7;p-hydroxybenzaldehyde, 123-08-0;vanillic acid, 121-34-6;
IV indicate that commercially cured Madagascar, Tonga, p-hydroxybenzyl alcohol, 623-05-2;vanillyl alcohol, 498-00-0;
and Jamaica beans did produce increased amounts of @-glucosidase,9001-22-3; anisic acid, 1335-08-6;
anisic aldehyde,
vanillin upon external enzyme treatment. This strength- 50984-52-6; heliotropin, 120-57-0.