|57||Last Update: 12/23/99|
By Eleanor Wiesenfeld, Noville, South Hackensack, NJ
The procedure for determining aroma profiles of various species of live Lavandula plants is discussed. Techniques employed include: dynamic headspace purge, short-path thermal desorption, GC/MSD/FID.
The Lavandula genus consists of about 20 species of small evergreen shrubs having aromatic foliage and flowers (1). The genus is a member of the Labiatae family, along with the thymes, mints, rosemary, sage, and other herbs, sharing with them the characteristic squared stems, two-lipped flowers and paired leaves (2). The plants are useful as decorative hedges in the garden, while the dried flowers are used in potpourris, as cooking herbs (3), and as insect repellents (7). Several species grow wild in the rocky soil of Southern Europe: Spain, Portugal, Southern France, etc. In the past, these wild plants were harvested by the local people and distilled for their essential oil (9). At present, fields are cultivated for this purpose in the calcareous soil of Mediterranean countries as well as other areas, especially Bulgaria and the countries of the former Yugoslavia (5).
Commercially, the Lavandula genus provides several important essential oils to the fragrance industry. Because of this, the composition of these oils has been extensively investigated (4). However, little has been written concerning the actual aroma of the various lavender species, that is, the aroma profile as determined by headspace analysis.
The aroma profile is determined by analyzing the headspace of a growing plant, a non-destructive technique. In comparison, the essential oil is produced by steam distillation of the above-ground parts (flowers, leaves and stems) of harvested lavender plants (5). Due to the harsh conditions prevalent during the distillation process, one would expect the odor profile of the living plant to be different than that of the processed oil. The aroma profile may be thought of as correlating to the perceived aroma of the living plant. The oil, however, reflects the composition of volatiles and semi-volatiles present in the plant, with molecular transformations occurring during the distillation process and during storage. Adulteration with synthetic chemicals (principally, linalool and linalyl acetate) and other essential oils of lesser value, is also a common practice. In this paper, aroma profiles of various Lavandula species will be investigated. In addition, commercially available lavender oil will be compared to its respective aroma profile.
One-year-old Lavandula plants were obtained from a supplier and identified as to species. Upon arrival, the plants were repotted and acclimated in the laboratory for several weeks. The specimens were sampled after each put forth active growth. All sampling took place within a three month period, prior to flowering. Each specimen was sampled at least twice, and the results averaged. The lavender species involved in this study were: L. officinalis or angustifolia (English Lavender), L. dentata (French or fringed Lavender), L. stoechas (Spanish Lavender), L. spica, L. viridis (green lavender), L. lanata (wooly lavender) (11), L. pinnata, L. multifida, and L. x heterophylla 'Goodwin Creek'. Of these, L. angustifolia is of commercial interest as the source of an essential oil used extensively in the fragrance industry.
Nomenclature of the individual species is sometimes confusing and contradictory and the common names are sometimes inconsistent. For instance, Lavandula angustifolia, usually referred to as English lavender, grows wild in Southern France. Lavandula stoechas, which grows wild in Spain, is sometimes referred to as French Lavender. Additionally, Lavandula spica is thought to be a non-specific name, possibly a form of angustifolia (8). To eliminate confusion, common names will not be referred to in the text.
Sampling and Analysis Technique
In preparation for sampling, several stems of each plant are enclosed in a two-piece spherical, custom-designed glass vessel, taking care to not damage or stress the plant. The two sections of the vessel are clamped around the stems and, over the course of several hours, the headspace is sampled by means of a low-flow vacuum pump. The volatile components are trapped on an adsorbent resin (Tenax® or equivalent), which is packed within a desorption tube attached to the pump. Subsequently, these trapped components are desorbed by means of a short path thermal desorption unit(10) into a gas chromatograph equipped with a cryo trap (10). A blank is purged and analyzed to determine artifacts from the system.
Instruments and conditions: The GC/MS (Hewlett Packard 5890/5971) is configured with two matched (50 meter, methyl silicone) capillary columns, one inlet, and two detectors (MSD and FID). The inlet is modified to accept the thermal desorption unit. The cryo trap is installed at the head of the columns and cooled to -50oC by carbon dioxide. Desorption takes place at 220oC for 4 minutes. After desorption is completed, the cryo trap is heated to match the inlet temperature, while the GC temperature is programmed from 350C to 2400C at 3oC per minute.
A proprietary, mass spectrum library provides peak identification. Quantitation
is determined by FID area normalization. Because of the complexity of the
chromatograms, no correction is made for response differences.
Legend For Chromatographs
* may include: beta-phellandrene, cis-ocimene, limonene, and 1,8-cineoleResults
All results are reported as averages; refer to Tables 1 and 2 for the
data. Although no ranges are given in the results, in some instances the
variation is considerable. In spite of attempts to standardize the sampling
parameters, differences still occur due to the inherent variable nature
of living things. However, despite the quantitative differences, in general,
the peak pattern of each specimen's profile is maintained from one sampling
|Table 1 - AROMA PROFILES
OF SELECTED LAVANDULA SPECIES
All data semi-quantitative--results expressed as percent
Species with 'typical' scent
|* sum of isomers
** may include beta-phellandrene, cis-ocimene, limonene, cineole--see text
|Table 2 - AROMA PROFILES
OF SELECTED LAVANDULA SPECIES
All data semi-quantitative--results expressed as percent
Species with 'a typical' scent
|* sum of isomers
** may include beta-phellandrene, cis-ocimene, limonene, cineol
Because of the large number of components present in each of the samples and the nature of the non-polar column, coelution is practically inevitable. This occurs in the case of beta-phellandrene, limonene, cis-ocimene, and 1,8-cineole (a terpene oxide). An exact quantitation of their distribution within the peak cannot be determined, but an approximation can be made by mass spectrometry. (See tables 1 and 2 for totals.) An analysis of this peak shows: limonene dominates in multifida, pinnata, and heterophylla; cineole dominates in viridis, dentata, lanata, and spica. Lavandula angustifolia contains a greater percentage of beta-phellandrene and L. stoechas seems to contain almost equal amounts of limonene and cineole.
Cymene most commonly occurs as the para isomer. However, Lavandula angustifolia shows two other isomers of cymene in addition to the para form. Presumably, these are meta and ortho, but this has not been proven. Quantitatively, all the cymenes are added together in tables 1 and 2.
The sesquiterpenes are reported as a sum of the various isomers (beta, gamma, delta, etc.) for that particular sesquiterpene. Those sesquiterpenes listed under 'other sesquiterpenes' are mostly unknown, but are usually not the same unknown.
The tentative identification of unknown diterpenes found in L. viridis, L. stoechas, and L. lanata is made by molecular weight, as determined by mass spectrometry molecular ion.
In the following discussion the scent of each Lavandula species will be correlated with the composition of its respective aroma profile derived by dynamic headspace analysis.
These Lavandulas can be grouped by chemotype, that is chemically, according to the major components in their aroma profiles. Some are dominated by terpenes and terpene oxides (viridis, angustifolia, dentata, and spica), others fall into the high camphor classification (stoechas, lanata), while still others are in the ocimene/carvacrol group (pinnata and multifida). The scent of most, but not all Lavandulas, is somewhat similar namely: refreshing, herbaceous and sweet, imparting a sense of 'clean'. In addition to having a similar scent, most lavenders have a similar appearance as well: small shrubs with narrow, flat, gray-green leaves (7). But, there are always exceptions. Lavandula dentata has the characteristic scent, but a rather distinctive leaf, being bright green and finely toothed (7). Lavandula species pinnata and multifida have a distinctly different scent and appearance than the others in this study. They are warm and balsamic, while their appearance is rather like a fern, having finely cut grayish leaves. Unlike the other lavenders, L. x heterophylla "Goodwin Creek" has a flat, greenish-gray, tomentose leaf. It exhibits a rather atypical aroma, being more floral and less herbaceous than the others.
Coelution of the four terpenes and terpene oxide in the 'multi-component peak' (see tables) is significant due to the abundance of this peak in the chromatograms and the differing olfactive properties of the various components. The aroma of the plant can depend upon which component dominates this complex peak. Beta-phellandrene is 'peppery-minty and slightly citrusy', contributing to the herbaceous scent of the angustifolia species. Cineole is 'camphoraceous, cool', adding a substantial sharp, penetrating quality to the viridis, dentata, lanata and spica species. Limonene is either dextro or laevo: d-limonene being 'sweet citrusy', whereas l-limonene is 'very clean, not reminiscent of Citrus fruits'. Limonene dominates the terpene fraction of the heterophylla species and contributes to the scent of stoechas. Trans-ocimene, the dominant terpene in the pinnata and multifida species, has a "warm-herbaceous odor", enhancing the balsamic qualities of the carvacrol component (all descriptions see ref. 6).
The total sesquiterpene content of the different species varies notably, from 1.7% in L. stoechas to 35.9% in L. angustifolia. Referring to tables 1 and 2, the variety of sesquiterpenes is distinctive for each species, and this unique distribution undoubtedly contributes to the variation in scent. The high amount of caryophyllene, along with a considerable amount of cadinenes, in L. angustifolia contributes to the distinctive woody note found in this species. It is not surprising to see a large amount of bisabolenes (sweet-spicy-balsamic (6)) in the pinnata and multifida species, enhancing their warm, sweet, spicy scent. Positive identification of individual sesquiterpenes can be difficult, if not impossible. All have the same molecular formula (C15H24) and although the molecular configuration of each is different, the fragmentation pattern can sometimes be remarkably similar, thus making mass spec identification imprecise. Retention time can sometimes be used in determining differences, but not to confirm identification. In some cases, retention time data was used to declare a sesquiterpene 'unknown'.
Lavandulol and borneol cannot always be separated by the non-polar column used for this study. However, these two alcohols can be differentiated by mass spectrometry. The two species containing lavandulol in the effluent are L. x heterophylla and L. lanata. Of these, heterophylla contains a significant amount of this terpene alcohol and its scent reflects this. Lavandulol, an isomer of geraniol, has a 'warm-rosy odor', 'with a slightly spicy note' (6). This, along with the floralcy of the germacrenes, likely contributes to the uncharacteristically floral note in the heterophylla specimen.
Borneol, more widely distributed throughout the genus than lavandulol, has a 'dry-camphoraceous, woody-peppery odor' (6), more characteristic of the lavenders. This terpene alcohol, along with its acetate and ketone form (camphor), adds to the distinctive warm, minty, herbaceous aroma of the typical lavender.
Comparison of Aroma Profile To Distilled Oil
Since these aroma profiles are produced by headspace analysis of the
leaves and stems of living plants, while the essential oils are prepared
by distillation of the leaves, stems, and flowers of harvested plants,
comparisons between the oil and headspace compositions are of limited value.
We can assume that inclusion of the flowers in the oil greatly influences
its composition. However, since comparisons will inevitably be made, one
is included here.
|Table 3 - Composition
Lavandula angustifolia aroma profile
compared to Bulgarian Lavender Oil
(results expressed as percent)
|* may include beta-phellandrene, cis-ocimene,
** contains approximately 30% of a tricyclo sesquiterpene
Aroma profile is from dynamic headspace purge
Bulgarian Oil is steam distilled
Table 3 compares the composition of the aroma profile of Lavandula angustifolia with that of the steam distilled oil of Bulgarian lavender, presumably derived mostly from this same species. (Hortus Third gives the source of Lavender Oil as Lavandula angustifolia and Lavandula stoechas (8). This may refer to the plants growing wild in Southern France and Spain as harvested in years past. Most other sources give L. angustifolia, the cultivated species, as the usual source for the oil. It can be assumed that this commercially available oil is derived from cultivated plants.)
Lavender oil of Bulgarian origin was chosen for this comparison because:
1) a large part of the lavender oil commercially available today is of Bulgarian origin;
2) the sample used for analysis was considered reliably free of adulterants; and
3) the composition of the sample is typical for a lavender oil.
Referring to the table, the most obvious differences between the headspace and the distilled oil are in the amounts of linalool and linalyl acetate. The level of linalyl acetate is of great interest in a lavender oil, because the quality of the oil is evaluated by its ester content; the higher the ester content, the finer the oil. Therefore, the low percentage of these two components in the headspace of the leaves is of particular interest. Again, exclusion of the flowers from the aroma profile, is an important point and may account for most of the disparity. However, the inherent differences between a headspace sampling and a steam distillation method cannot be discounted as a factor. Due to the high temperatures and harsh conditions prevailing during the distillation process, this technique often leads to degradation products and molecular rearrangements. Storage time of the oil must also be considered as a factor.
Another interesting difference between the composition of the headspace and the oil is the distribution of the sesquiterpenes. Both contain caryophyllene, a sesquiterpene found widely in nature. However, the headspace contains approximately four percent of a tricyclo sesquiterpene eluting on the back of the caryophyllene peak. This peak does not appear in the oil. In addition, the oil contains farnesene and germacrene isomers, while the headspace contains considerable amounts of cadinene and bergamotene isomers (listed as 'other sesquiterpenes' in the legend).
In general, the headspace is dominated by terpenes and sesquiterpenes, while the oil contains mostly alcohols and esters.
Dynamic headspace analysis coupled with thermal desorption is a powerful technique for studying the correlation of effluent composition with scent in living plants. A reliable mass spectrometry library is a necessity for identification of the components.
This work was not intended to be a exhaustive study of the subject, but the data can be used for comparative purposes and as a guide to direct future projects.
Future studies will involve analyzing the flowers of various Lavandula species, especially those of L. angustifolia to determine their contribution to the oil composition.
The author wishes to thank Sandra Seals and Laura Ferrara for their help in preparing this presentation.
Presented at Pitt Con 97, Atlanta, GA, March 1997.
(1) Allaby, M., (editor), The Concise Oxford Dictionary of Botany, Oxford University Press, 1992.
(2) Van Pelt Wilson, H., and Bell, L., The Fragrant Year, M. Barrows & Company, Inc., 1967.
(3) Janulewics, M., Traditional Home Book of Herbs, Longmeadow Press, 1996.
(4) Boelens, Perfumer and Flavorist, May/June 1995.
(5) The H&R Book of Perfume.
(6) Arctander, S., Perfume and Flavor Chemicals, vols. 1&2, Published by the author, original publication 1961.
(7) Bremner, Lesly, Complete Book of Herbs, Penguin Books, 1994.
(8) Hortus Third, Macmillan, 1976.
(9) Arctander, Steffen, Perfume and Flavor Materials of Natural Origin, Published by the author, original publication 1961.
(10) Available from Scientific Instruments Services, Ringoes, NJ
(11) Brown, Deni, Encylopedia of Herbs, and Their Uses, Dorling Kindersley, 1995.
Yttria coated filament at start
Yttria coated filament after 16,000 cycles