Phytochemical Analysis

Phytochem. Anal. 17: 121–128 (2006)

ESSENTIAL OIL DETERMINATION BY NIR SPECTROSCOPY Phytochemical
121

Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/pca.895

Analysis

Application of Near-infrared Spectroscopy in Quality Control

and Determination of Adulteration of African Essential Oils
HECTOR R. JULIANI,1 JEREMY KAPTEYN,1 DAYTON JONES,2 ADOLFINA R. KOROCH,1 MINGFU WANG,1
DENYS CHARLES3 and JAMES E. SIMON1*
1
New Use Agriculture and Natural Plant Products Program, Rutgers, The State University of New Jersey, New Brunswick, New Jersey, USA
2
Analytical Spectral Devices Inc., Boulder, Colorado, USA
3
Frontier Natural Products, Co-Op, Norway, Iowa, USA

Received 28 April 2005; Revised 5 May 2005; Accepted 20 May 2005

Abstract: An evaluation has been made of the potential of near-infrared (NIR) technologies in the assessment of essential oil com-
ponents and in the identification of individual essential oils. The results showed that cross-validation models are able to predict
accurately almost all of the components of essential oils. In different cinnamon (Cinnamomum zeylanicum) and clove (Syzygium
aromaticum) essential oils, which showed a similar composition, 23 components (representing 97.8 – 99.9% of the oil) were accu-
rately predicted, as well as 20 components (93.0 – 99.1%) in Cinnamomum camphora (ravintsara), 32 components (92.3 – 98.1%) in
Ravensara aromatica (ravensara), and 26 components (96.6 – 98.4%) in Lippia multiflora. For almost all of the components, the
modelled and reference values obtained by GC-FID were highly correlated (r 2 ≥ 0.985) and exhibited a low variance (less than 5%).
The model was also able to discriminate between the ravintsara and ravensara essential oils. It was shown that two commercial
oils labelled as R. aromatica were actually ravintsara (C. camphora), revealing the misidentification of these essential oils in the
marketplace. The study demonstrates the application of NIR technology as a quality control tool for the rapid identification of
individual essential oils, for product authentication, and for the detection of adulteration. Copyright © 2006 John Wiley & Sons,
Ltd.

Keywords: near-infrared spectroscopy; GC-FID; quality control; volatile oils; Cinnamomum zeylanicum; Cinnamomum camphora;
Lippia multiflora; Ravensara aromatica; Syzygium aromaticum.

INTRODUCTION identify those samples that might require further

Near-infrared (NIR) technology has been applied to the
quantification of agricultural commodities of plant and
animal origin and their derived products, forages,
foods, feeds and their ingredients (Williams, 2001).
In the field of natural products, NIR has been used for
the rapid determination of active constituents of
medicinal plants (Gray et al., 2001; Laasonen et al.,
2002; Rager et al., 2002). Since most of the recent ad-
vances in NIR have been concerned with chemometric
software (Williams, 2001), its use in the quantification
of essential oil components is a relatively new applica-
tion. The technology was useful in the rapid and accu-
rate determination of essential oil content and
composition in intact fennel fruits (Steuer and Schulz,
2003) and dried mint leaves (Schulz et al., 1999),
and in the estimation of different components in
isolated essential oils, including mint, eucalyptus and
citrus (Schulz et al., 1999; Steuer et al., 2001; Wilson
et al., 2001). One of the potential applications of the
technology is in quality control, since large lots or
single samples can rapidly be pre-screened in order to
* Correspondence to: J. E. Simon. New Use Agriculture and Natural
Plant Products Program, Rutgers, The State University of New Jersey,
New Brunswick, New Jersey, USA.
E-mail: jesimon@aesop.rutgers.edu
testing by more time-consuming and expensive meth-
ods (Williams, 2001).
Many valuable essential oils are produced in Africa,
and the continent has a significant potential for pro-
ducing new and traditional essential oils for the inter-
national market. The genus Cinnamomum (Laureaceae)
includes trees that typically contain important eco-
nomic essential oils in almost all of the plant parts. The
leaf and bark of the tree Cinnamomum zeylanicum
Blume (syn. C. verum J.S. Presl), popularly known
as cinnamon, are used as spices throughout the
world. The essential oil from the leaves is dominated
by eugenol, while the stem bark contains
cinnamaldehyde. The essential oil of the evergreen tree
Syzygium aromaticum (L.) Marr and Perryl, belonging to
the family Myrtaceae, is also dominated by eugenol,
with minor and varying amounts of eugenyl acetate
and (E)-caryophyllene (Juliani et al., 2004).
The camphor tree, C. camphora (L.) J. Presl, is native
to China, Japan and Taiwan, and is known to have dif-
ferent chemotypes. On the island of Madagascar, the
1,8-cineole chemotype was introduced in the nineteenth
century from Taiwan (formerly Formosa). Currently,
this tree grows wild in mid-eastern Madagascar, and is
locally know as ravintsara. The essential oil of the
leaves of ravintsara contains 1,8-cineole in amounts

Copyright © 2006 John Wiley & Sons, Ltd. Phytochem. Anal. 17: 121–128 (2006)
DOI: 10.1002.pca
122 H. R. JULIANI ET AL.
ranging from 50 to 60% as the main component, with
minor amounts of sabinene (10–12%) and α-terpineol
(7%; Behra et al., 2001; Juliani et al., 2005a).
Ravensara aromatica Sonn. (Lauraceae, syn
Ravensara anisata; ravensara) is a well-known en-
demic aromatic tree from Madagascar and is locally
known as ‘Havozo’, which means ‘aromatic tree’ in the
Malagasy language. All the aerial parts contain essen-
tial oils. The leaf oils are dominated by limonene
(14–23%) and sabinene (10–16%) with minor amounts
of methyl chavicol (2–12%). The reddish bark is very
aromatic and contains volatile oils that are rich in
methyl chavicol (90–95%), whilst the fruits also contain
essential oils (Behra et al., 2001).
The essential oils of ravensara and ravintsara are
frequently misidentified in the marketplace (Juliani
et al., 2005a), through a confusion that appears
to have originated years ago (Behra et al., 2001). The
botanical name of the genus Ravensara came from
latinisation of the Malagasy word ‘ravintsara’ (meaning
‘good leaves’), which refers to the introduced species
C. camphora (Rasonaivo and de la Gorce, 1998).
Lippia multiflora Moldenke (Verbenaceae) is another
important aromatic plant from West Africa that is
widely used as a tea and also has many medicinal
properties. The essential oils extracted from the leaves
have long been used locally as spices, condiments,
flavours in foods and drinks and as cosmetics, and
these oils have a great potential on the international
marketplace (Juliani et al., 2005b).
The lack of quality assurance and control of African
natural materials using assays designed to ensure the
authenticity of botanical products is the major obstacle
to international trade. The present study evaluates the
potential application of NIR technology in the assess-
ment of essential oil quality, and in the identification
and adulteration of individual essential oils of a range
of selected African essential oils.
EXPERIMENTAL
Materials. Cinnamon (C. zeylanicum) leaf oil and clove
(S. aromaticum) bud oils were obtained from the
Malagasy companies EPAM (Fianarantsoa), Phael Flor
(PF; Antananarivo) and Parapharma (PP; Antananarivo).
For comparison purposes, essential oils of these
plants obtained from US companies were also include
in this study. Ravintsara (C. camphora) and ravensara
(R. aromatica) oils were obtained from the Malagasy
companies Biosave (Antananarivo), EPAM, Label Con-
servation Biodiversity and Development (Antananarivo),
and the National Center for Application of Phar-
maceutical Research (Antananarivo). Two ravensara
(R. aromatica) oils derived from Madagascar were
obtained in the US marketplace for comparison
purposes.
Copyright © 2006 John Wiley & Sons, Ltd.
Wild plants of Lippia multiflora were collected at full
flowering from Ghana (Nyankpala, Sari, Atebubu, Aman-
tin and Kobre). Voucher specimen (Nyankpala GC
47801, Sari GC 47807, Atebubu GC 47803, Amantin
GC 47812 and Kobre GC 47808) were deposited
at Ghana Herbarium (GC). The essential oils were
extracted in Ghana (by A.Y. Mensah and C. Quansah,
Knust Kumasi, Ghana) by steam distillation from
the dried leaves and inflorescences. Pure eugenol and
1,8-cineole as standards were obtained from Sigma
(St. Louis, MO, USA).
Reference analyses. Samples were analysed using an
Agilent (Palo Alto, CA, USA) 6890 GC system coupled to
Agilent 5973 series FID and mass selective detectors,
and fitted with an HP5 column (30 m × 0.25 mm i.d.;
0.25 µm). Samples were injected using an Agilent 7683
series autosampler with an inlet temperature of 200°C.
The column temperature programme was: 60°C for
1 min, increased by 4°C/min to 200°C, and held for
15 min. The carrier gas was helium at a flow rate of
1 mL/min, and the FID temperature was 220°C. In-
dividual identifications were made by matching spectra
with those from MS libraries (Wiley 275.L), and
identities were confirmed by retention indices (RI;
Adams, 1995). The relative percentages of each
component, listed in order of elution from the HP5
(DB5 equivalent) column, were calculated using the
FID detector by peak integration without calculating
the detector response factor. The relative percentages
of those components that were absent or present in
only trace amounts (less than 0.01%) were assumed to
be zero.
Near-infrared spectroscopy. Essential oils (1 mL) were
placed in quartz cuvettes (path length 5 cm) and
scanned 10 times in the range 300–2500 nm using a
portable NIR spectrometer (LabSpec Pro; Analytical
Spectral Devices, Boulder, CO, USA). Each individual
scan was the result of an average of 25 scans auto-
matically made by the equipment. Data processing
and development of appropriate chemometrics methods
were carried out using the program GRAMS 386
(Galactic Industries Corporation, Salem, NH, USA).
Calibration accuracy was defined in terms of the
multiple coefficient of determination (r 2), that measures
the linearity between the modelled and reference val-
ues (GC-FID), and the variance (V), that measures the
overall error between modelled and reference values,
expressed as the percentage of standard error of cross-
validation (SECV) in relation to the average of each
essential oil component (V = SECV*100/mean value).
Values ranging from 0 to 5% are considered excellent
and acceptable for any NIR application, whilst values
>5% are considered unsuitable for NIR analysis
(Williams, 2001). The numbers of factors were set to
minimize the variance of the calibration.
Phytochem. Anal. 17: 121–128 (2006)
DOI: 10.1002.pca
 ESSENTIAL OIL DETERMINATION BY NIR SPECTROSCOPY 123

RESULTS AND DISCUSSION
Essential oils of cinnamon leaf and clove buds
Six cinnamon (C. zeylanicum) leaf, five clove (S.
aromaticum) bud, and one clove leaf essential oil
samples, together with pure eugenol (99.9%), were
used to build a cross-validation model (Table 1). The
compositions of the cinnamon and clove oils were simi-
lar, being dominated by eugenol (59 – 82 and 69–81%,
respectively) with minor and varying amounts of
eugenyl acetate (2–11 and 1–15%, respectively) and (E)-
caryophyllene (2– 4 and 5–10%, respectively). The cin-
namon leaf oils exhibited a more complex mixture of
volatile constituents containing significant amounts of
monoterpenes (4– 7%) that were almost absent from
the clove oils (0– 0.22%). The cinnamon leaf oils also
contained higher levels of safrole and benzyl benzoate
(1–11 and 5%, respectively) than the clove oils (0– 0.1
and 0– 0.7%).
Correlation statistics showed that the model was
able to predict with high accuracy the relative percent-
age of all 23 essential oil components constituting 98–
99.9% of the total oil (Table 1). The coefficient of
determination of the cross-validation model for each of
the components was very high (r 2 ≥ 0.993), while the
variance between the model and reference values for
eugenol, eugenyl acetate and (E)-caryophyllene was low
(<1.75%). For the majority of the minor components,
the variance was less than 4.6%. However, for com-
ponents with relative percentages of less than 0.18%,
such as furfural, α-thujene and sabinene, the variation
was much higher (≥7%; Table 1).
Ravintsara and ravensara essential oils
Ten essential oils of ravintsara (C. camphora) and 20
components were used to build a cross-validation
model. The major components of the essential oil,

Table 1 Percentage compositions of cinnamon leaf (Cinnamomum zeylanicum), clove leaf and clove bud (Syzygium aromaticum)
essential oils by GC and the corresponding NIR correlation statistics

Reference analysis (GC/FID) NIR correlation statistics

Peak Retention Compound Compositiona Factors Coefficient of Standard error of Variance

number index determination cross-validation (V)b
Minimum Maximum Mean (r 2) (SECV)

1 843 Furfural 0.0c 0.2 0.0 17 0.998 0.003 7.0

2 932 α-Thujene 0.0 0.2 0.0 14 0.997 0.003 8.3
3 940 α-Pinene 0.0 1.3 0.4 16 0.999 0.008 1.9
4 955 Camphene 0.0 0.4 0.1 13 0.999 0.005 3.5
5 972 Sabinene 0.0 0.1 0.0 19 0.996 0.003 6.9
6 982 β -Pinene 0.0 0.4 0.1 10 0.999 0.005 3.6
7 994 Myrcene 0.0 0.3 0.0 16 0.999 0.0014 3.3
8 1009 α-Phellandrene 0.0 1.2 0.4 16 0.999 0.01 2.3
9 1028 p-Cymene 0.0 1.4 0.4 13 0.999 0.01 2.3
10 1031 Limonene 0.0 0.8 0.3 13 0.999 0.01 3.2
11 1035 1,8-Cineole 0.0 0.2 0.04 22 0.999 0.001 2.4
12 1098 Linalool 0.0 2.3 0.8 16 0.999 0.03 3.8
13 1268 (E)-Cinnamaldehyde 0.0 0.3 1.0 19 0.998 0.004 4.6
14 1283 Unknown 0.0 2.8 0.7 10 0.999 0.02 2.7
15 1295 Safrole 0.0 11.0 1.4 16 0.999 0.05 3.6
16d 1370 Eugenol 59.7 99.9 77.0 6 0.993 0.8 1.0
16e 59.7 85.4 75.1 6 0.995 0.48 0.63
17 1382 α-Copaene 0.0 1.0 0.4 13 0.998 0.013 3.0
18 1427 (E)-Caryophyllene 0.0 16.1 5.9 15 0.999 0.1 1.7
19 1457 α-Humulene 0.0 7.3 1.4 14 0.999 0.05 3.6
20 1462 Unknown 0.0 1.8 0.7 10 0.998 0.02 2.8
21 1540 Eugenyl acetate 0.0 14.7 6.6 15 0.999 0.1 1.5
22 1598 Caryophyllene oxide 0.0 1.0 0.5 14 0.997 0.015 3.3
23 1776 Benzyl benzoate 0.0 4.9 1.9 16 0.999 0.05 2.6

a
Relative percentage composition derived from GC-FID.
b
Overall error between modelled and reference values calculated from (SECV*100/mean value) for each component.
c
The value 0.0 implies that the component was absent or present in trace amounts of less than 0.01%.
d
The range was extended by including pure eugenol (99.9% by FID) in the model.
e
Pure eugenol was removed from the model.

Copyright © 2006 John Wiley & Sons, Ltd. Phytochem. Anal. 17: 121–128 (2006)
DOI: 10.1002.pca
124 H. R. JULIANI ET AL.

Table 2 Percentage composition of ravintsara (Cinnamomum camphora) essential oils by GC-FID and the corresponding NIR
correlation statistics

Reference analysis (GC/FID) NIR correlation statistics

Peak Retention Compound Compositiona Factors Coefficient of Standard error of Variance

number index determination cross-validation (V )b
Minimum Maximum Mean (r 2) (SECV)

1 932 α-Thujene 0.4 0.8 0.7 7 0.997 0.01 1.2

2 940 α-Pinene 3.0 7.1 4.6 7 0.994 0.08 1.7
3 955 Camphene 0.1 0.3 0.2 10 0.985 0.01 3.4
4 979 Sabinene 5.1 13.7 11.3 6 0.999 0.07 0.6
5 982 β -Pinene 2.3 3.5 3.1 5 0.987 0.04 1.3
6 993 Myrcene 0.4 1.6 1.2 6 0.997 0.02 1.6
7 1022 α-Terpinene 0.2 1.0 0.6 11 0.999 0.01 1.6
8 1028 p-Cymene 0.2 2.1 1.2 10 0.999 0.02 1.4
9 1031 Limonene 0.8 5.5 1.9 11 0.999 0.03 1.6
10 1035 1,8-Cineole 54.9 63.6 58.2 7 0.993 0.20 0.3
11 1053 (E)-β -Ocimene 0.2 0.8 0.4 14 0.997 0.01 2.3
12 1064 γ -Terpinene 0.1 1.6 0.9 10 0.998 0.02 2.3
13 1081 (Z)-Sabinene hydrate 0.1 0.5 0.2 13 0.993 0.01 4.2
14 1092 Terpinolene 0.2 0.4 0.3 13 0.999 0.05 15.7
15 1112 Unknown 0.1 0.3 0.2 12 0.997 0.01 3.1
16 1179 Terpin-4-ol 0.4 1.0 0.8 9 0.998 0.01 1.3
17 1188 Unknown 1.4 3.1 2.6 10 0.994 0.04 1.4
18 1193 α-Terpineol 7.0 10.6 8.3 9 0.993 0.08 1.0
19 1427 (E)-Caryophyllene 0.3 1.4 0.7 11 0.998 0.02 2.2
20 1462 α-Humulene 0.2 1.2 0.8 12 0.997 0.01 1.8

a
Relative percentage composition derived from GC-FID.
b
Overall error between modelled and reference values calculated from (SECV*100/mean value) for each component.

namely, 1,8-cineole (58%), sabinene (11%) and α-
terpineol (8%), were accurately predicted by the
cross-validation model with a high coefficient of deter-
mination (r2 ≥ 0.993) and the variance between the
modelled and the reference values was less than 1%
(Table 2).
The remaining components also showed a high co-
efficient of determination (r 2 ≥ 0.985; Table 2). The
components with relative percentages ranging from 1.9
to 4.6% (α-pinene, β-pinene, limonene and an
unknown) showed a higher variance (1.3–1.7%). The
minor components ranging from 0.4 to 1.2% showed a
variance of 2.3% or less. The components with relative
percentages less than 0.3% (i.e. camphene, sabinene
hydrate, unknown, and terpinolene) showed a very
high variance (3–16%; Table 2).
Five essential oils samples of ravensara (R.
aromatica) and 32 components were used to build a
cross-validation model (Table 3). The major compo-
nents of the essential oil, namely, sabinene (13%),
limonene (17%) and methyl eugenol (11%), were
predicted with a very high coefficient of determination
(r2 = 0.999), and the variance between modelled and
reference values was less than 1% (Table 3). The
remaining 29 minor components were also predicted
with a high coefficient of determination (r2 > 0.992) and
with a low variance (2% and less; Table 3).
Essential oil of L. multiflora
Five essential oils of L. multiflora plus the GC-FID data
of 26 components were used to build a cross-validation
model (Table 4). The major components p-cymene
(14%), γ-terpinene (10%), thymol (37%), carvacrol (5%)
and thymyl acetate (14%) were accurately predicted
with a high coefficient of determination (r 2 ≥ 0.985) and
a low variance (V < 2%). The remaining 21 minor com-
ponents were also predicted with a high coefficient of
correlation (r 2 ≥ 0.978) and with variance ranging from
1 to 4% (Table 4).
The results showed that the cross-validation
models were able to predict accurately nearly all of the
components in the essential oils. For almost all of the
20–32 compounds present in the different essential oil
samples, the modelled and reference values obtained
by GC-FID were highly correlated (r 2 ≥ 0.985) and
exhibited a low variance (V ≤ 5%). In previous
reports using NIR models for the analysis of isolated
essential oils (Schulz et al., 1999; Steuer et al., 2001;

Copyright © 2006 John Wiley & Sons, Ltd. Phytochem. Anal. 17: 121–128 (2006)
DOI: 10.1002.pca
 ESSENTIAL OIL DETERMINATION BY NIR SPECTROSCOPY 125

Table 3 Percentage composition of ravensara (Ravensara aromatica) essential oils by GC-FID and the corresponding NIR correla-
tion statistics

Reference analysis (GC/FID) NIR correlation statistics

Peak Retention Compound Compositiona Factors Coefficient of Standard error of Variance

number index determination cross-validation (V )b
Minimum Maximum Mean (r 2) (SECV)

1 932 α-Thujene 0.4 1.1 0.8 5 0.999 0.008 1.0

2 940 α-Pinene 2.9 5.2 4.2 4 0.998 0.03 0.7
3 955 Camphene 0.4 0.9 0.7 4 0.999 0.004 0.5
4 979 Sabinene 8.4 16.2 12.6 8 0.999 0.04 0.3
5 982 β -Pinene 2.0 2.9 2.5 4 0.998 0.01 0.6
6 993 Myrcene 1.9 3.4 2.9 4 0.999 0.02 0.5
7 1009 α-Phellandrene 0.2 1.9 1.2 5 0.999 0.02 1.4
8 1015 ∆3-Carene 2.5 4.9 3.4 4 0.992 0.07 2.0
9 1022 α-Terpinene 0.1 7.0 3.7 7 0.999 0.05 1.4
10 1028 p-Cymene 1.2 3.4 1.9 4 0.999 0.02 1.1
11 1031 Limonene 13.9 20.9 17.1 4 0.999 0.04 0.2
12 1042 (Z)-β -Ocimene 0.7 1.2 1.1 4 0.999 0.005 0.5
13 1045 Unknown 1.0 1.7 1.3 4 0.998 0.009 0.7
14 1053 (E)-β -Ocimene 0.2 0.3 0.3 4 0.998 0.003 1.0
15 1064 γ -Terpinene 0.2 2.6 1.5 7 0.999 0.02 1.1
16 1092 Terpinolene 0.1 0.5 0.4 4 0.999 0.004 1.1
17 1098 Linalool 3.9 8.9 5.9 4 0.999 0.03 0.6
18 1179 Terpin-4-ol 0.1 0.3 0.1 6 0.999 0.002 1.1
19 1193 α-Terpineol 2.4 4.4 3.3 6 0.999 0.02 0.5
20 1195 Unknown 0.3 0.4 0.3 4 0.999 0.0007 0.2
21 1198 Methyl chavicol 2.8 16.8 7.0 6 0.999 0.06 0.9
22 1297 Unknown 0.1 0.4 0.3 4 0.999 0.003 0.9
23 1344 δ -Elemene 0.1 0.9 0.4 6 0.999 0.008 1.9
24 1356 Eugenol 0.2 0.7 0.4 4 0.999 0.006 1.4
25 1382 α-Copanene 0.5 1.9 1.1 6 0.999 0.001 0.9
26 1395 β -Elemene 0.4 0.9 0.6 6 0.999 0.005 0.9
27 1414 Methyl eugenol 7.8 16.9 10.9 4 0.999 0.09 0.8
28 1427 (E)-Caryophyllene 2.1 7.4 4.4 4 0.999 0.06 1.3
29 1462 α-Humulene 0.5 1.7 1.0 6 0.999 0.008 0.8
30 1469 Allo-aromadendrene 0.3 1.1 0.6 7 0.999 0.005 0.9
31 1488 Germacrene D 3.0 6.9 4.0 6 0.999 0.04 0.9
32 1530 δ -Cadinene 0.4 1.1 0.7 4 0.999 0.009 1.4

a
Relative percentage composition derived from GC-FID.
b
Overall error between modelled and reference values calculated from (SECV*100/mean value) for each component.

Wilson et al., 2001) only 10 components or fewer were
predicted, whereas in the present study two to three
times the number of components were included since
the essential oils were far more complex.
The reference values obtained by GC-FID for six
components of seven Citrus oils were predicted by NIR
(Steuer et al., 2001). The main components, limonene
and γ-terpinene, and the minor ones, myrcene and β-
pinene, were predicted with a high coefficient of deter-
mination (r 2 = 0.99) and with a low variance, whilst the
remaining two components, α-pinene and sabinene,
exhibited a higher variance (10% or higher). Through
NIR and GC-FID, it was possible to predict 10 com-
ponents in different mint essential oils. The main
components of the oils, menthone, isomenthone and
menthol, showed a high coefficient of determination
and low variance (Schulz et al., 1999); however, the
variance of the minor components (limonene, 1,8-
cineole, menthofuran, menthyl acetate, pulegone and
piperitone) was much higher, ranging from 8 to 25%.
In a series of Eucalyptus oils, the content of the main
constituent 1,8-cineole (71.5 – 85%), measured by the
freezing point method (Wilson et al., 2001) was ac-
curately predicted using NIR. By extending the range of
1,8-cineole content in the NIR assay by up to 99%
(with pure 1,8-cineole), a more accurate calibration
was obtained (Wilson et al., 2001). Results from the
present study showed that the extension of the range

Copyright © 2006 John Wiley & Sons, Ltd. Phytochem. Anal. 17: 121–128 (2006)
DOI: 10.1002.pca
126 H. R. JULIANI ET AL.

Table 4 Percentage composition of Lippia multiflora essential oils by GC-FID and the corresponding NIR correlation statistics

Reference analysis (GC/FID) NIR correlation statistics

Peak Retention Compound Compositiona Factors Coefficient of Standard error of Variance

number index determination cross-validation (V )b
Minimum Maximum Mean (r 2) (SECV)

1 932 α-Thujene 1.7 1.0 2.0 10 0.983 0.05 3.0

2 940 α-Pinene 0.2 0.1 0.3 10 0.978 0.008 3.6
3 979 Sabinene 0.4 0.2 0.5 5 0.979 0.02 4.3
4 993 Myrcene 2.3 1.9 2.6 7 0.987 0.03 1.3
5 1009 α-Phellandrene 0.3 0.2 0.4 5 0.992 0.004 1.4
6 1015 ∆3-Carene 0.2 0.1 0.2 5 0.982 0.002 1.4
7 1022 α-Terpinene 1.9 1.2 2.5 5 0.994 0.04 1.9
8 1028 p-Cymene 14.3 11.8 16.3 5 0.985 0.2 1.3
9 1031 Limonene 0.6 0.5 0.6 5 0.976 0.006 1.1
10 1053 (Z)-β -Ocimene 0.5 0.1 1.2 9 0.999 0.007 1.3
11 1064 γ -Terpinene 10.1 5.7 14.5 5 0.995 0.2 2.0
12 1081 cis-Sabinene hydrate 0.3 0.2 0.4 5 0.989 0.007 2.2
13 1092 Terpinolene 0.1 0.1 0.1 6 0.993 0.002 2.2
14 1098 Linalool 0.2 0.1 0.3 5 0.997 0.004 1.8
15 1109 Unknown 0.2 0.1 0.3 10 0.996 0.003 1.8
16 1113 Unknown 0.2 0.1 0.2 5 0.995 0.003 1.7
17 1179 Terpin-4-ol 0.6 0.5 0.7 5 0.984 0.01 1.8
18 1290 Thymol 37.4 31.0 41.1 7 0.999 0.04 0.1
19 1299 Carvacrol 5.4 4.8 5.7 7 0.999 0.005 0.1
20 1365 Thymyl acetate 14.4 11.8 16.2 7 0.997 0.09 0.6
21 1382 Carvacryl acetate 0.8 0.6 1.1 5 0.982 0.03 3.7
22 1427 (E)-Caryophyllene 1.6 1.3 2.1 7 0.997 0.02 1.2
23 1462 (E)-β -Farnesene 2.2 1.3 2.9 10 0.998 0.03 1.3
24 1483 Germacrene D 0.2 0.1 0.3 10 0.982 0.01 7.5
25 1488 β -Cubebene 1.5 0.2 2.7 10 0.998 0.04 2.7
26 1514 β -Bisabolene 0.6 0.4 0.7 7 0.995 0.008 1.4
27 1597 Caryophyllene oxide 0.3 0.1 0.6 0 0 0 0

a
Relative percentage composition derived from GC-FID.
b
Overall error between modelled and reference values calculated from (SECV*100/mean value) for each component.

of eugenol concentration by adding pure standard did
not improve the correlation but slightly increased the
variation from 0.63 to 1% (Table 1).
A case study
In order to verify the identity of the essential oil of
ravensara (R. aromatica), two discrimination models
were built with essential oils of ravintsara (C.
camphora) and ravensara. Authentic ravensara and
ravintsara oils were analysed using both discrimination
models and the samples passed the appropriate tests.
Two commercial samples (obtained from companies A
and B) both labelled as being oil of R. aromatica were
obtained in the US market place. These oils were
scanned using the NIR equipment. Neither sample
passed the ravensara discrimination model, whilst
both passed the ravintsara (C. camphora) model, show-
ing that the R. aromatica samples from the US com-
panies were actually from C. camphora. We were able
to confirm this misidentification on the basis of the
traditional organoleptic, physicochemical and chemical
properties of these oils (Juliani et al., 2005a), thus
validating the qualitative conclusion reached by NIR
technology. Other essential oils, such as citrus and
mint, were also discriminated on the basis of their indi-
vidual spectral properties (Steuer et al., 2001; Schulz
et al., 1999).
We then prepared an adulterated essential oil by
mixing pure 1,8-cineole with authentic ravensara (R.
aromatica) oil in order to mimic the relative percentage
of 1,8-cineole found in ravintsara (C. camphora) oil.
When this adulterated product was analysed according
to the NIR models, it failed to pass either the ravintsara
or the ravensara discrimination model even though the
oil contained 1,8-cineole in an amount similar to that
present in true ravintsara oil.

Copyright © 2006 John Wiley & Sons, Ltd. Phytochem. Anal. 17: 121–128 (2006)
DOI: 10.1002.pca
 ESSENTIAL OIL DETERMINATION BY NIR SPECTROSCOPY 127

Table 5 Actual and predicted values of the components of ravintsara (Cinnamomum camphora) essential oils obtained from US
companies A and B and incorrectly labelled Ravensara aromatica essential oils

Composition of oil from company A Composition of oil from company B

Peak Retention Actual Predicted Differencea Actual Predicted Differencea

number index Compound (GC-FID) (NIR) (GC-FID) (NIR)

1 932 α-Thujene 0.7 0.6 13.2 0.8 0.9 3.9

2 940 α-Pinene 4.3 4.1 5.7 4.7 4.8 1.7
3 955 Camphene 0.3 0.2 43.7 0.2 0.3 19.4
4 979 Sabinene 10.4 10.3 1.2 13.6 13.6 0.6
5 982 β -Pinene 3.1 2.8 7.9 3.5 3.6 3.1
6 993 Myrcene 0.7 0.8 12.9 1.5 1.5 0.6
7 1022 α-Terpinene 0.0 0.1 – 0.7 0.6 7.2
8 1028 p-Cymene 2.5 1.8 28.6 0.6 0.9 35.9
9 1031 Limonene 1.1 1.7 50.9 1.8 1.6 12.9
10 1035 1,8-Cineole 59.6 59.5 0.0 57.7 58.7 1.8
11 1053 (E)-β -Ocimene 0.0 0.0 – 0.6 0.6 11.9
12 1064 γ -Terpinene 0.1 0.3 221.7 1.1 1.2 4.6
13 1081 (Z)-Sabinene hydrate 0.3 0.5 75.3 0.4 0.2 52.6
14 1092 Terpinolene 0.0 0.1 – 0.3 0.3 5.8
15 1112 Unknown 0.2 0.2 19.3 0.2 0.0 90.1
16 1179 Terpin-4-ol 1.0 0.7 31.9 0.7 0.7 4.3
17 1188 Unknown 3.2 2.8 12.1 2.2 2.7 24.8
18 1193 α-Terpineol 9.1 9.3 2.45 7.1 6.7 5.24
19 1427 (E)-Caryophyllene 0.5 0.6 26.09 0.8 0.7 10.58
20 1462 α-Humulene 0.7 0.5 26.62 0.8 1.0 16.13

a
Difference between actual and predicted values expressed as the percentage of the actual value.

In order to predict the composition of essential oils
in the two commercial samples incorrectly labelled as
R. aromatica, a calibration model was built using
ravintsara oil and the result of the prediction was
compared with the actual value from the GC-FID data
(Table 5). The model was able to predict with high
accuracy the actual value for the major component
1,8-cineole, viz. actual values of 59.6 and 57.7 and
predicted values of 59.5 and 58.7, respectively, for the
products from companies A and B. The differences
between the predicted and actual values (expressed as
percentages of the actual values) for α- and β-pinene,
sabinene and α-terpineol were also low (<8%) (Table 5).
However, for those minor components present in the
range from 0 to 3%, the differences were much higher
(8 –220%).
Whilst the calibration model was able to predict
almost all of the components in the two mislabelled
essential oils, only the major compounds (1,8-cineole,
sabinene, α-terpineol, and α- and β-pinene) were accu-
rately predicted in relation to the reference values. By
international standards, only the major components
are analysed and required to be between specific
ranges in order to meet a defined grade and standard.
Most international standards for essential oils, for ex-
ample, have a maximum of three to seven constituents
defined by their minimum–maximum ranges. The
cross-validation models could be further improved by
analysing more samples so that the entire variability of
the product could be used to predict more accurately
the minor components.
Although a small number of essential oils (five to 13
samples of each type) were used to build the models,
our results showed that accurate models could be built
and that the data show the great potential of NIR
technology in the accurate analysis of almost all the
components in essential oils. This technology has
the ability to identify adulterated essential oils and to
distinguish essential oils of different chemotypes and
species. Nowadays, GC analysis is very rapid (through
the application of Fast and Ultra-Fast GC) and may
provide competitive and comparable results in 1–3 min.
However, the described NIR procedure is even more
rapid since, once the models have been created, it re-
quires only a few seconds to analyse the individual
oil. For this reason, NIR is ideal for routine and quality
control analysis.
Coupled with GC, as the reference technique, NIR
technology is a very fast and reliable method. With the
added value of NIR to identify individual essential oils
and the possible detection of adulterations, NIR
spectroscopy becomes a very promising and powerful
technique in the field of research and quality control of
essential oils.

Copyright © 2006 John Wiley & Sons, Ltd. Phytochem. Anal. 17: 121–128 (2006)
DOI: 10.1002.pca
128 H. R. JULIANI ET AL.
Acknowledgements
The authors wish to acknowledge the United States
Agency for International Development (US AID) for pro-
viding the funds to conduct quality control studies in
natural plant products as part of the Agribusiness in
Sustainable Natural African Plant Products (ASNAPP;
www.asnapp.org) project. The authors wish to thank
John Enterline from Analytical Spectral Devices for
graciously providing the NIR equipment, Phael-Flor,
EPAM and Parapharma for kindly providing samples of
essential oils that were used to assess oil quality and
composition, and Olivier Behra, Lalasoa Ranarivelo,
Abraham Y. Mensah, Charles Quansah, and Dan
Acquaye for fruitful discussions. The authors are
indebted to the New Jersey Agricultural Experiment
Station, Chemonics International, LDI, and PRONABIO
(Association of Malagasy Exporters of Natural Prod-
ucts) for supporting the research program.
REFERENCES
Adams PA. 1995. Identification of Essential Oil Components by Gas
Chromatography/Mass Spectroscopy. Allured: Carol Stream, IL.
Behra O, Rakotoanison C, Harris R. 2001. Ravintsara vs ravensara, a
taxonomic clarification. Int J Aromather 15: 4–7.
Gray DE, Roberts CA, Rottinghaus GE, Garrett HE, Pallardy SG. 2001.
Quantification of root chicoric acid in purple coneflower by near
infrared reflectance spectroscopy. Crop Sci 41: 1159–1161.
Juliani HR, Simon JE, Ramboatiana MMR, Behra O, Garvey A, Raskin
I. 2004. Malagasy aromatic plants: essentials, antioxidant and
antimicrobial activities. Acta Hortic 629: 77– 81.
Copyright © 2006 John Wiley & Sons, Ltd.
Juliani HR, Behra O, Moharram H, Ranarivelo L, Ralijerson B,
Andriantsiferana M, Ranjatoson N, Rasoarahona J, Ramanoelina
P, Wang M, Simon JE. 2005a. Searching for the real ravensara
(Ravensara aromatica Sonn.) essential oil: a case study for ‘Natiora’
the Malagasy natural product label. Perf Flavor 30: 60– 65.
Juliani HR, Wang M, Moharram H, Asante-Dartey J, Acquaye D,
Koroch AR, Simon JE. 2005. Intraspecific Variation in Quality
Control Parameters, Polyphenol Profile and Antioxidant Activity
in Wild Populations of Lippia multiflora from Ghana, pp. 126–142.
In: Wang M, Sang S, Hwang LS, Ho CT (eds). Herbs: Challenges in
Chemistry and Biology. American Chemical Society Symposium
Series 925. American Chemical Society, Washington, D.C. USA.
Laasonen M, Harmia-Pulkkinen T, Simard CL, Michiels E, Rasanen M,
Vuorela H. 2002. Fast identification of Echinacea purpurea dried
roots using near-infrared spectroscopy. Anal Chem 74: 2493–
2499.
Rager I, Roos G, Schmidt PC, Kovar KA. 2002. Rapid quantification
of constituents in St. John’s Wort extracts by NIR spectroscopy.
J Pharm Biomed Anal 28: 439– 446.
Rasoanaivo P, de la Gorce P. 1998. Essential oils of economic value in
Madagascar: present state of knowledge. Herbalgram 43: 31– 59.
Schulz H, Drews HH, Kruger H. 1999. Rapid NIRS determination of
quality parameters in leaves and isolated essential oils of Mentha
species. J Essent Oil Res 11: 185–190.
Steuer B, Schulz H. 2003. Near-infrared analysis of fennel (Foeniculum
vulgare Miller) on different spectrometers—basic considerations for
a reliable network. Phytochem Anal 14: 285–289.
Steuer B, Schulz H, Lager E. 2001. Classification and analysis of citrus
oils by NIR spectroscopy. Food Chem 72: 113–117.
Williams PC. 2001. Implementation of near-infrared technology.
In Near-infrared Technology in the Agricultural and Food Industries,
Williams P, Norris K (eds). American Association of Cereal
Chemists. Minnesota, USA; 145–169.
Wilson ND, Watt RA, Moffat AC. 2001. A near-infrared method for the
assay of cineole in Eucalyptus oil as an alternative to the official BP
method. J Pharm Pharmac 53: 95–102.
Phytochem. Anal. 17: 121–128 (2006)
DOI: 10.1002.pca