Food Research International 38 (2005) 847–853

www.elsevier.com/locate/foodres

Analysis of a complex mixture of carotenes from oil palm

(Elaeis guineensis) fruit extract
A. Mortensen *

Colour Research, Development & Application, Chr. Hansen A/S, Bøge Allé 10-12, DK-2970 Hørsholm, Denmark

Received 1 June 2004; accepted 21 January 2005

Abstract

A carotene extract from the fruits of the oil palm (Elaeis guineensis) was analysed by HPLC employing a C30 column for better
separation efficiency. A multitude of cis-isomers of a-, b- and c-carotene were separated. Detailed assignment was possible by sub-
jecting pure standards of a-, b- and c-carotene to isomerisation and comparing spectral data and order of elution to literature data.
a- and b-carotene were found to be the most abundant carotenoids comprising 12.3% and 17.9%, respectively, of a (roughly) 30% oil
suspension of oil palm carotenes in vegetable oil. A large proportion (about 40%) of a- and b-carotene was in the form of cis-
isomers. The c-carotene content was found to be 0.38% and other carotenes like phytoene, phytofluene, f-carotene, lycopene
and possibly b-zeacarotene were found as well but were not quantified.

2005 Elsevier Ltd. All rights reserved.

Keywords: Oil palm fruit; Elaeis guineensis; a-Carotene; b-Carotene; c-Carotene; HPLC

1. Introduction The crude palm oil is usually processed to yield either

Oil palms (Elaeis guineensis) are grown extensively in
Southeast Asia and Equatorial Africa because the fruits
are rich in vegetable oil and because the oil palm pro-
duces more oil per area than any other plant (Poku,
2002). The kernel as well as the mesocarp contain oil.
The oil obtained from the yellow, fleshy mesocarp is
called crude palm oil and is rich in carotenes (Poku,
2002). A number of carotenes have been found in palm
oil: phytoene, phytofluene, f-carotene, neurosporene, a-
zeacarotene, b-zeacarotene, lycopene, d-carotene, a-car-
otene, b-carotene and c-carotene (Tay & Choo, 2000;
Tan, Grady, & Gawienowski, 1986; Yap, Choo, Ooi,
Ong, & Goh, 1991) of which a- and b-carotene comprise
more than 90% (Tay & Choo, 2000; Yap et al., 1991).
The other carotenes are present in small amounts and
are precursors in the biosynthesis of a- and b-carotene.
*
Fax: +45 45 74 89 11.
E-mail address: alan.mortensen@dk.chr-hansen.com.
a red or bleached cooking oil or detergents. During this
processing, the carotenes are often destroyed, but they
may also be isolated and suspended in vegetable oil at
a concentration of up to 30%. This product can be used
as a food colour or in health products such as dietary
supplements.
Before the oil is pressed from the fruits, the fruits are
subjected to heat treatment employing pressurised steam
or boiling (Poku, 2002). Heating causes cis/trans-isom-
erisation of the carotenes (Doering, Sotiriou-Leventis,
& Roth, 1995). Thus, a complex mixture of highly apolar
carotenes and their cis-isomers is obtained making sepa-
ration difficult. A decade ago a new stationary phase
(called C30) was developed (Sander, Sharpless, Craft, &
Wise, 1994). This new column has since then gained wide
acceptance for analysis of carotenoids and has proven
useful for the simultaneous separation of 11 different
carotenoids used as food colourants (Breithaupt, 2004).
Furthermore, this column not only provides excellent
separation of closely related carotenes like a- and

0963-9969/$ - see front matter
2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.foodres.2005.01.009
848 A. Mortensen / Food Research International 38 (2005) 847–853
b-carotene, but also a better separation of their cis-iso-
mers than was previously possible using a conventional
C18 column (Sander et al., 1994).
2. Materials and methods
Oil palm carotenes were suspended in vegetable oil at
a concentration of 30%. a- and c-carotene were obtained
from CaroteNature and b-carotene (30% oil suspension)
from BASF. Methyl tert-butyl ether (MTBE) and meth-
anol were from Merck (HPLC grade). HPLC analysis
was performed on a Merck-Hitachi system employing
a photodiode-array detector for detection. The column
(thermostatted at 10 C) used was from YMC (C30,
3 lm, 250 · 4.6 mm) and a linear gradient going from
0% to 100% B in 75 min (A: MTBE:methanol:water
(15:81:4) and B: MTBE:methanol (10:1)) at a flow rate
of 1.1 ml/min was employed. Samples for HPLC analy-
sis were dissolved in MTBE and filtered prior to
injection. a-, b- and c-carotene were isomerised by dis-
solving the carotene in MTBE, adding a small crystal
of iodine and leaving the sample under fluorescent light
for 1 h. Response curves of a-, b- and c-carotene as a
function of concentration were constructed – the
concentration of the standard solutions was determined
by using published absorption coefficients ðE1% 1 cm Þ:
a-carotene (2710 in hexane), b-carotene (2500 in cyclo-
hexane) and c-carotene (2780 in hexane) (Britton,
1995; CaroteNature, private communication).
3. Results and discussion
3.1. Assignments
In Fig. 1 is shown the chromatogram of oil palm caro-
tenes in the region where a- and b-carotene elute. As
expected, based on the close structural similarity between
a- and b-carotene, a- and b-carotene elute close together.
A large number of isomers are formed in relatively high
concentration (Fig. 1) due to the heat treatment the caro-
tenes have undergone (vide supra). In assigning the
peaks, standards of trans-a- and trans-b-carotene were
used to assign the largest and second-largest peaks to
trans-b- and trans-a-carotene, respectively. Because
standards of cis-a- and cis-b-carotenes are not readily
available (except for 9- and 13-cis-b-carotene), assign-
ments of the cis-isomers were done using the following
procedure. Firstly, isomerised samples of trans-a- and
trans-b-carotene were used to clarify which peaks in
Fig. 1 belonged to a-carotene and which belonged to
b-carotene. Secondly, relative abundance and spectral
data of the individual isomers were used (see below) to
assign (most of) the mono-cis-isomers and, thirdly,
assignments in the literature were used to corroborate
t- α t-β 9-β
9,13-β
9,13,9'-β 9,9'-β
9-α
15-α α+15-β 9'-α
13'-α+β 9,9'-α
13-β
13-α
α+ β 2
α 1 4
β α 7-β 3
24 29 34 39
Time / min
Fig. 1. Chromatogram at 456 nm of oil palm carotenes. Peaks are
labelled as follows: number indicates position of cis-bond and greek
letter indicates carotene, e.g., 15-a means 15-cis-a-carotene. Assign-
ments in bold are tentative. See text for details.
the assignments and assign a few di-cis-isomers. In
assigning mono-cis-isomers, the following guidelines
(Britton, 1995) are useful: (1) the absorption maximum
is blue-shifted a few nm compared to the trans-isomer
(di- and poly-cis-isomers show a larger shift) and (2) a
new band around 340 nm (the cis-band) is observed, the
intensity of which increases the closer the cis-bond is to
the centre of the molecule (the cis-band may be com-
pletely absent in di- and poly-cis-isomers). Furthermore,
9-cis- and 13-cis-b-carotene are the most stable mono-
cis-isomers (based on calculations, not actual measure-
ments) followed by 7-cis, 15-cis and 11-cis (Doering
et al., 1995) which is reflected in the relative abundance
of these isomers: 9-cis and 13-cis are often observed in
carotenoid-containing samples, 15-cis is sometimes ob-
served, whereas 7-cis and 11-cis are rarely/never observed.
Based on this, it is possible to assign (most of) the
mono-cis-isomers of a- and b-carotene without using
standards. However, based on spectral data alone it is
not possible to distinguish between 13-cis- and 13 0 -cis-
or 9-cis- and 9 0 -cis-a-carotene. It has been shown that
13-cis-a-carotene elutes before 13 0 -cis-a-carotene and
9-cis elutes before 9 0 -cis (Emenhiser, Englert, Sander,
Ludwig, & Schwartz, 1996) under conditions similar to
those used here. Therefore, it is possible to unambigu-
ously assign 15-, 13-, 13 0 -, 9- and 9 0 -cis-a-carotene and
15-, 13- and 9-cis-b-carotene (Fig. 1). However, this only
accounts for around half the number of peaks. Compar-
ison with the literature assignments allows a tentative
assignment of some of the remaining peaks. Thus,
9,13-di-cis-b-carotene has been shown to elute between
13-cis- and trans-b-carotene (Breitenbach, Braun, Stei-
ger, & Sandmann, 2001; Lacker, Strohschein, & Albert,
1999). The peak at 31.8 min is a good candidate for 9,
13-di-cis-b-carotene due to its relative intensity and be-
cause the intensity of the cis-band of this compound
matches that of 9,13-di-cis-b-carotene (cis-band more in-
tense than in 9-cis-b-carotene but less intense than in
13-cis-b-carotene) (Lacker et al., 1999). Other di- and
 A. Mortensen / Food Research International 38 (2005) 847–853
tri-cis-b-carotenes have been identified using a C30 col-
umn (Breitenbach et al., 2001; Strohschein, Pursch, &
Albert, 1999), but different eluents from those used here
were used making comparison to the results obtained
here more difficult. However, based on order of elution
the peak at 33 min (Fig. 1) could be 9,13,9 0 -tri-cis-b-car-
otene. This is further ‘‘corroborated’’ by results ob-
tained using a normal-phase calcium hydroxide
column, where 9,13,9 0 -cis-b-carotene elutes before
trans-b-carotene but after 13-cis-b-carotene (the order
of elution using the reversed-phase C30 column seems
to be very similar to the order of elution using the nor-
mal-phase calcium hydroxide column) (Koyama et al.,
1988; Tsukida, Saiki, Takii, & Koyama, 1982).
A b-carotene isomer eluting between trans-b-carotene
and 9-cis-b-carotene has only been observed once before
using a C30 column (Breitenbach et al., 2001 – in this case
sub-ambient column temperature was used as it is here,
whereas elevated column temperature is more commonly
employed), probably because it is usually not separated
from trans-b-carotene (in the chromatogram of Stro-
hschein et al., 1999, a pronounced asymmetry is seen in
the peak of trans-b-carotene, but it was not noted). A
tentative assignment of this peak is 9,9 0 -di-cis-b-carotene
because it elutes between trans- and 9-cis-b-carotene on a
calcium hydroxide column (Koyama et al., 1988) and be-
cause it does not show a pronounced cis-band (when
symmetry is restored, as in 9,9 0 -di-cis-b-carotene, the
cis-band is expected to be ‘‘absent’’ (a small cis-band is
always seen, even in trans-b-carotene)). Breitenbach
et al. (2001) did not assign this peak but they did observe
that it featured a weak cis-band. The peak at 38.4 min
is not only observed in the oil palm carotene extract
but also in an isomerised sample of trans-b-carotene (it
is not found in the sample of trans-b-carotene before
isomerisation) indicating that it is an isomer of b-caro-
tene (as shown below this peak is not pure, but contains
other carotenes besides the cis-b-carotene). This peak
could be due to 7-cis-b-carotene (again, this compound
has not been observed using a C30 column, but 7-cis-b-
carotene has been shown to elute after 9-cis-b-carotene
on a calcium hydroxide column (Koyama et al., 1988;
Tsukida et al., 1982)).
Di-cis-isomers of a-carotene have not been assigned
previously, but the peak at 33.4 min could be 9,9 0 -di-
cis-a-carotene based on its absent cis-band and relative
retention time (compared to the order of elution of the
b-carotene isomers).
The peaks marked 1, 2, 3 and 4 have not been assigned
yet. Peak 1 exhibits absorption maximum at 432 nm and
could thus be a poly-cis-a- or b-carotene. The trans-b-car-
otene standard does show a very small peak in this region,
but it does not increase in intensity on isomerisation indi-
cating that it is not a b-carotene isomer. The absorption
maximum suggests that it could be b-zeacarotene
(Britton, 1995). Peak 2 could be a b-carotene isomer since
849
isomerised b-carotene shows a shoulder on the 13-cis-b-
carotene peak, but due to overlap of the peak of this
compound with the peaks of 13-cis-b-carotene and
trans-a-carotene, the spectrum of this peak could not be
obtained and compared to that of the corresponding peak
in isomerised b-carotene, making assignment uncertain.
Peak 3 is actually a double peak and can be assigned
to c-carotene (see below). The peak of 9-cis-b-carotene
is not pure but contains another compound showing
as a shoulder at 37.5 min (Fig. 1, peak 4). This com-
pound is not observed in isomerised b-carotene (nor in
isomerised a-carotene) and is thus most likely another
carotene. f-Carotene has been shown to co-elute with
9-cis-b-carotene (Yeum et al., 1996). However, f-caro-
tene has absorption maximum at 400 nm (Britton,
1995), whereas the compound eluting with 9-cis-b-caro-
tene shows a spectrum similar to that of 9-cis-b-caro-
tene, i.e., absorption maximum around 450 nm (since
the peaks of 9-cis-b-carotene and the unknown carotene
overlap, the spectrum of the unknown carotene alone
could not be obtained, but spectra taken at 37 and
37.5 min were virtually identical). Possible candidates,
based on published absorption maxima (Britton,
1995), could be either neurosporene or e-carotene but
a definitive assignment is impossible without an authen-
tic standard.
Besides a- and b-carotene, palm oil is known to
contain other carotenes (precursors in the biosynthesis
of a- and b-carotene) in small amounts (Tay & Choo,
2000; Yap et al., 1991), which may make them diffi-
cult to detect. Looking at the associated spectra in de-
tail reveals a number of these minor carotenes that
can be distinguished based on their absorption charac-
teristics. Phytoene (3 conjugated double bonds) shows
an absorption maximum at 286 nm (Britton, 1995)
and was found at 23.4 min. Phytofluene (5 conjugated
double bonds) absorbs at 348 nm (Britton, 1995) and
was found with retention times of 26.1, 27.6, 28.8
and 29.7 min. The phytofluene isomers elute with
some of the other carotenes (primarily a- and b-caro-
tene isomers) and can only be detected due to the
large differences in absorption spectra. f-Carotene
with 7 conjugated double bonds absorbs at 400 nm
(Britton, 1995) and may thus be difficult to detect if
they co-elute with some of the a- and b-carotene iso-
mers. Two of the f-carotene isomers co-elute with
other carotenes in low abundance making it possible
to detect them. One of the isomers elutes as part of
the peak tentatively assigned to 7-cis-b-carotene and
the other as part of peak 3 (Fig. 1).
Other carotenes with 9 or more conjugated double
bonds, e.g., neurosporene, e-carotene and d-carotene,
cannot be identified without the use of proper standards
because the absorption spectra of these compounds
resemble those of a- and b-carotene isomers and they
can be expected to elute in the same region.
850 A. Mortensen / Food Research International 38 (2005) 847–853
13/13'-γ+7 β+ζ
t-γ
5'-γ
13/13'-γ
9/9'-γ γ present. 15-cis-c-carotene was found in the isomerised
ζ
γγ γ γ
38 48 58
Time / min
Fig. 2. Chromatogram at 456 nm of oil palm carotenes. Labelling of
the peaks follows the same conventions as in Fig. 1. A higher
concentration of carotene (compared to Fig. 1) was used in order to be
able to detect the less abundant carotenes. Retention times have
increased by
1 min compared to Fig. 1. See text for details.
The most abundant of the minor carotenes is c-caro-
tene. In Fig. 2 is shown the chromatogram of oil palm
carotenes in the region where the c-carotene isomers
elute. trans-c-Carotene is identified unequivocally by
comparison with a standard. By comparing with an
isomerised sample of trans-c-carotene, it is possible to
assign some of the peaks to cis-c-carotenes (Fig. 2).
Assignment of the individual mono-cis-isomers is done
by looking at the relative intensity of the cis-band and
the relative abundance of that isomer, as was done for
a- and b-carotene (see above). Hence, the most abun-
dant cis-isomer at 53.5 min (Fig. 2) can safely be as-
signed to 5 0 -cis-c-carotene because: (1) its spectrum is
identical to that of trans-c-carotene (the chromophore
is the same), (2) the order of elution (compared to lyco-
pene, see below) and (3) its high abundance (compare
with lycopene, see below). In contrast to a- and b-caro-
tene, separation and assignment of c-carotene isomers
using a C30 column have never been published before.
However, isomerised c-carotene has been analysed using
a calcium hydroxide column (Schmitz, Emenhiser, &
Schwartz, 1995), but apart from trans-carotene, only
two isomers eluting after trans-c-carotene were tenta-
tively identified as 9- and 9 0 -cis-c-carotene (this assign-
ment was based on the fact that ‘‘peripheral cis-
isomers elute after the all-trans compound’’ and not by
using spectral data (eluent interfered with the UV spec-
trum)). However, the most peripheral cis-isomer would
be 5 0 -cis-c-carotene making their assignments question-
able, see above and below. A distinction between
9- and 9 0 -cis-c-carotene and between 13- and 13 0 -cis-c-
carotene can therefore not be made (Fig. 2) as it was for
a-carotene. With respect to order of elution c-carotene
resembles lycopene rather than the (likewise) asymmetric
a-carotene, i.e., the 9- and 9 0 -cis isomers elute before the
trans-isomer and not after (Sander, Pursch, Märker, &
Wise, 1999). The peak that was tentatively assigned to
7-cis-b-carotene and a f-carotene isomer should also
contain some c-carotene isomers (13- or 13 0 -cis and oth-
ers) by comparison with an isomerised c-carotene stan-
dard. From the spectral data associated with this peak
no more than two compounds are revealed, but it is
not unlikely that more than two compounds may be
standard, but in the oil palm carotene sample it would
elute together with 9-cis-b-carotene and would hence
Lycopene escape detection.
The two peaks around 69 min are due to lycopene,
68
the smaller one being due to trans-lycopene and the lar-
ger to 5-cis-lycopene (Sander et al., 1999). It may seem
surprising that trans-lycopene is not the most abundant
isomer, which it usually is in tomatoes and tomato prod-
ucts. Ab initio calculations have shown that 5-cis-lyco-
pene is more stable than trans-lycopene (Chasse et al.,
2001), so in a thermally isomerised sample like this
5-cis-lycopene may be the most abundant lycopene
isomer. Other cis-lycopenes, which would elute between
trans-b-carotene and trans-lycopene (Sander et al.,
1999), would be too rare to be detected. Some peaks
(Fig. 2) could neither be assigned to c-carotene nor lyco-
pene. The nature of these compounds is not known.
3.2. Quantitative analysis
Quantification of carotenoids using HPLC is not a
trivial task for a number of reasons. First of all, base-
line separation of closely related carotenoids like a-
and b-carotene (or lutein and zeaxanthin) is inherently
difficult due to their close structural similarity. This
means that although base-line separation of trans-a-
and trans-b-carotene is easily achieved using a C30 col-
umn (Fig. 1), base-line separation of such a complex
mixture with more than 20 closely related compounds
(i.e., a- and b-carotene isomers as well as other caro-
tenes) as the one analysed here is not possible. Secondly,
standards of trans-a- and trans-b-carotene are available,
but standards of their cis-isomers (except for 9- and
13-cis-b-carotene) are not readily available. Most often
cis-isomers are quantified using the response of the corre-
sponding trans-isomer standard, because the cis-isomers
are not available to determine their response curves.
However, this will tend to underestimate the concentra-
tion of the cis-isomers because they have lower extinc-
tion coefficients than the corresponding trans-isomer
(Britton, 1995). Furthermore, the absorption maximum
of cis-isomers is blue-shifted compared to the absorption
maximum of the trans-isomer. Therefore, if the absorp-
tion maximum of the trans-isomer is chosen for quanti-
fication, this will even further underestimate the
concentration of the cis-isomers. In one study, response
factors for a number of b-carotene isomers were deter-
mined by using known absorption coefficients (Schierle,
Härdi, Faccin, Bühler, & Schüep, 1995). It was then pos-
sible to determine the concentration of these cis-isomers
 A. Mortensen / Food Research International 38 (2005) 847–853 851

from the response curve of trans-b-carotene by applying
these relative response factors. In another study (Tsuk-
ida et al., 1982), correction factors (relative to trans-b-
carotene) were applied, but it was not clear how these
correction factors were determined.
Instead of using a single wavelength for determination
it was decided to use an integrated chromatogram from
365 to 545 nm (for a- and b-carotene) to alleviate the
problem of blue-shifted absorption spectra. Correction
for the lower absorption coefficients of the cis-isomers
was not attempted (the published relation factors
(Schierle et al., 1995) or correction factors (Tsukida
et al., 1982) could not be used, since they were determined
for entirely different systems). The correction factor for
9-cis-b-carotene is close to one (Schierle et al., 1995; Tsuk-
ida et al., 1982) and this is the most abundant cis-isomer
(Fig. 1). The correction factors are higher for other iso-
mers (Schierle et al., 1995; Tsukida et al., 1982), but they
are much less abundant than trans-b-carotene and 9-cis-
b-carotene in this sample (Fig. 1), so that not correcting
for the lower response of these isomers would not intro-
duce too much of an error (the response factors of a-car-
otene isomers are not known, but they can be expected to
follow the same trends as the b-carotene isomers).
Instead of using peak areas for quantifying a- and b-
carotene, peak heights were used because of overlapping
peaks (Fig. 1). In case of the peaks eluting between 28
and 29 min (Fig. 1) three peak heights were used and
half of the total height was arbitrarily assigned as a-
carotenes and the other half as b-carotenes. c-Carotene
was quantified using peak areas, otherwise the isomer
eluting as a shoulder on the peak of trans-c-carotene
(Fig. 2) could not be quantified. A single wavelength
(465 nm) was used rather than an integrated chromato-
gram because the signal-to-noise ratio would be poor for
the smaller peaks. This would only introduce a small er-
ror since the two most abundant isomers, trans- and 5 0 -
cis-c-carotene, both have maximum absorption at this
wavelength (the other isomers show the expected blue-
shift).
The responses of trans-a- and trans-b-carotene were
linear up to a concentration of 0.3 mg/ml, the highest
concentrations that could be used while still getting a
detector response. The response of c-carotene was linear
up to at least 0.12 mg/ml.
In Table 1 is shown the content of a-, b- and c-caro-
tenes in the oil palm extract. The ratio between a- and b-
carotene in this batch is 40.6:59.4, and trans-a-carotene
accounts for 59% of total a-carotene and trans-b-caro-
tene accounts for 61% of total b-carotene. The 9/9 0 -cis
isomers of a- and b-carotene account for half the total
amount of cis-isomers (Table 1). The concentration of
c-carotene in this batch was found to be 3.8 g/kg. A split
between trans-c-carotene and cis-c-carotenes could not
be obtained due to the overlap of trans-c-carotene with
a cis-isomer (Fig. 2). It was not possible to determine the
concentration of other carotenes due to overlap with
the more intense peaks of a- and b-carotene and due
to the lack of standards. However, their concentrations
are small compared to a- and b-carotene. It is estimated
that a- and b-carotene constitute more than 95% of total
carotenes in the oil palm extract.

Table 1
Retention times and concentrations of a- and b-carotene isomers in the oil palm extract
Compound Peak maximaa (nm) Retention time (min) Concentration (g/kg)
15-cis-a-Carotene 441 332 26.3 2.06
13-cis-a-Carotene 439 334 27.0 3.37
13 0 -cis-a-Carotene 438 333 28.1 15.4b
15-cis-b-Carotene 448 339 28.4 10.79c
13-cis-b-Carotene 445 339 30.3 8.19
trans-a-Carotene 448 269 31.1 72.4
9,13-Di-cis-b-carotene 440 338 31.8 8.81
9-cis-a-Carotene 442 334 32.6 13.66
9,13,9 0 -Tri-cis-b-carotene ND 33.0 ND
9,9 0 -Di-cis-a-carotene 438 334 33.3 3.83
trans-b-Carotene 456 275 34.4 109.2
9,9 0 -Di-cis-b-carotene ND 35.2 6.16
9 0 -cis-a-Carotene 444 334 35.5 12.08
9-cis-b-Carotene 449 342 36.9 36.2
7-cis-b-Carotene ND 38.4 ND
Total a-carotene 122.8 (59%)d
Total b-carotene 179.4 (61%)d
Total c-carotene 3.8
ND, not determined due to overlap with other compounds.
a
The spectral resolution of the detector was not good enough to allow resolution of vibrational fine structure.
b
Includes other a-carotenes eluting between 25 and 29 min (Fig. 1).
c
Includes other b-carotenes eluting between 28 and 29 min (Fig. 1).
d
Percentage trans-isomer.
852 A. Mortensen / Food Research International 38 (2005) 847–853
3.3. Comparison with previous studies
Most of the studies on oil palm carotenes are based on
crude palm oil or red palm oil (Darnoko, Cheryan,
Moros, Jerrel, & Perkins, 2000; Ng & Tan, 1988; Tay
& Choo, 2000; Tay, Choo, Ee, & Goh, 2000; Tan
et al., 1986), i.e., the carotenes have undergone extensive
heat treatment. Solvent-extracted palm fruits have also
been analysed (Tay et al., 2000; Yap et al., 1991) but in
one study (Yap et al., 1991) the fruits were heat-treated
prior to extraction. Thus, in the majority of these studies
a high degree of isomerisation of the carotenoids can be
expected. Identification of carotenoids is based on a
combination of a few standards (only phytoene, a- and
b-carotene and lycopene have been used, but not in all
studies) and comparison of spectral data with literature
values. Phytoene, phytofluene and f-carotene can be
safely identified without the use of standards due to the
large blue-shift of their absorption spectra. However,
other carotenes present in small amounts like neurospo-
rene, e-carotene, d-carotene, a-zeacarotene and b-zeacar-
otene may easily be taken for cis-isomers of the abundant
a- and b-carotene or vice versa. Therefore, it is important
to unambiguously identify the peaks as either a-carotene,
b-carotene or another carotene by employing isomerisa-
tion of pure compounds as was done in this study.
Most of the studies employed a C18 column (Ng &
Tan, 1988; Tay & Choo, 2000; Yap et al., 1991) or open
column chromatography (Tan et al., 1986) for separa-
tion. Neither of these methods are as good at separating
cis-isomers as the C30 column. The inadequate separa-
tion of these methods manifests itself in the levels of
cis-isomers found: of b-carotene only around 1% was
cis whereas cis-isomers constituted up to 14% of total
a-carotene (Tay & Choo, 2000; Yap et al., 1991). This
is in stark contrast to the results presented here and
what could be expected, i.e., similar levels of cis-a- and
cis-b-carotene. Tay and Choo (2000) also noted that
the peaks of trans-a- and trans-b-carotene were not pure
but contained other compounds – most likely cis-iso-
mers – which would explain the anomalous results. Like-
wise, open column chromatography was not able to
provide a complete separation of a-carotene from b-car-
otene (Tan et al., 1986), leading to an apparently very
high percentage of b-carotene (more than 80%). Only
in two studies was a C30 column employed (Darnoko
et al., 2000; Tay et al., 2000) and in one of them they
were only looking for a- and b-carotene (Tay et al.,
2000). However, the separations achieved in those two
studies were inferior to the one presented here. Further-
more, the study by Darnoko et al. (2000) is marred by
wrong assignments. Thus, two peaks, which are proba-
bly 13- and 13 0 -cis-a-carotene (based on their relative
retention times), have been assigned to e-carotene and
c-carotene, the fourth-largest peak is assigned to 15-
cis-b-carotene despite the fact that this isomer is not
very abundant (a more likely assignment would be 13-
cis-b-carotene), and assignment of two peaks to b-cryp-
toxanthin and echinenone is also obviously wrong (by
comparing their spectral data to literature values).
All in all, results reported previously should be viewed
with some caution both with respect to assignments (be-
cause they are largely based on spectral data and not on
comparison with isomerised standards) and with respect
to quantification (because of inadequate separation).
4. Conclusions
A C30 column provides the best separation of a com-
plex mixture of closely related carotenes. By employing
pure standards and subjecting them to isomerisation, it
is possible to unambiguously assign most peaks to either
a-, b- or c-carotene. Careful examination of the spectral
data revealed that minor carotenes like phytoene, phy-
tofluene and f-carotene co-eluted with some of the
a- and b-carotene isomers. Other carotenes, like neuro-
sporene and e-carotene, could possibly also co-elute with
the a- and b-carotene isomers, but would elude detection
due to near-coincidence of the absorption spectra.
a- and b-carotene, in the ratio 2:3, constituted by far
the major part of the carotenes. Sixty percent of the a-
and b-carotene were in the trans-form.
References
Breitenbach, J., Braun, G., Steiger, S., & Sandmann, G. (2001).
Chromatographic performance on a C30-bonded stationary phase
of monohydroxycarotenoids with variable chain length or degree of
desaturation and of lycopene isomers synthesized by various
carotene desaturases. Journal of Chromatography A, 936, 59–69.
Breithaupt, D. E. (2004). Simultaneous HPLC determination of
carotenoids used as food coloring additives: applicability of
accelerated solvent extraction. Food Chemistry, 86, 449–456.
Britton, G. (1995). UV/visible spectroscopy. In G. Britton, S. Liaaen-
Jensen, & H. Pfander (Eds.), Carotenoids. Spectroscopy (Vol. 1B,
pp. 13–62). Basel: Birkhäuser Verlag.
Chasse, G. A., Mak, M. L., Deretey, E., Farkas, I., Torday, L. L.,
Papp, J. G., et al. (2001). An ab initio computational study on
selected lycopene isomers. Journal of Molecular Structure (Theo-
chem), 571, 27–37.
Darnoko, D., Cheryan, M., Moros, E., Jerrel, J., & Perkins, E. G.
(2000). Simultaneous HPLC analysis of palm carotenoids and
tocopherols using a C-30 column and photodiode array detector.
Journal of Liquid Chromatography & Related Technologies, 23,
1873–1885.
Doering, W. von E., Sotiriou-Leventis, C., & Roth, W. R. (1995).
Thermal interconversions among 15-cis-,13-cis-, and all-trans-b-
carotene: kinetics, Arrhenius parameters, thermochemistry, and
potential relevance to anticarcinogenicity of all-trans-b-carotene.
Journal of the American Chemical Society, 117, 2747–2757.
Emenhiser, C., Englert, G., Sander, L. C., Ludwig, B., & Schwartz, S.
J. (1996). Isolation and structural elucidation of the predominant
geometrical isomers of a-carotene. Journal of Chromatography A,
719, 333–343.
 A. Mortensen / Food Research International 38 (2005) 847–853
Koyama, Y., Hosomi, M., Miyata, A., Hashimoto, H., Reames, S. A.,
Nagayama, K., et al. (1988). Supplementary and revised assign-
ment of the peaks of the 7,9-, 9,9 0 -, 13,13 0 -, 9,13 0 -di-cis and 9,9 0 ,13-
tri-cis isomers of b-carotene in high-performance liquid chroma-
tography using a column of calcium hydroxide. Journal of
Chromatography, 439, 417–422.
Lacker, T., Strohschein, S., & Albert, K. (1999). Separation and
identification of various carotenoids by C30 reversed-phase high-
performance liquid chromatography coupled to UV and atmo-
spheric pressure chemical ionization mass spectrometric detection.
Journal of Chromatography A, 854, 37–44.
Ng, J. H., & Tan, B. (1988). Analysis of palm oil carotenoids by HPLC
with diode-array detection. Journal of Chromatographic Science,
26, 463–469.
Poku, K. (2002). Small-scale palm oil processing in Africa. FAO
Agricultural Services Bulletin, 148.
Sander, L. C., Pursch, M., Märker, B., & Wise, S. A. (1999).
Separation of carotenoid isomers by capillary electrochromatog-
raphy with C30 stationary phases. Analytical Chemistry, 71,
3477–3483.
Sander, L. C., Sharpless, K. E., Craft, N. E., & Wise, S. A. (1994).
Development of engineered stationary phases for the separation of
carotenoid isomers. Analytical Chemistry, 66, 1667–1674.
Schierle, J., Härdi, W., Faccin, N., Bühler, I., & Schüep, W. (1995).
Example 8: Geometrical isomers of b,b-carotene. In G. Britton, S.
Liaaen-Jensen, & H. Pfander (Eds.), Carotenoids. Isolation and
analysis (Vol. 1A, pp. 265–272). Basel: Birkhäuser Verlag.
Schmitz, H. H., Emenhiser, C., & Schwartz, S. J. (1995). HPLC
separation of geometric carotene isomers using a calcium hydrox-
853
ide stationary phase. Journal of Agricultural and Food Chemistry,
43, 1212–1218.
Strohschein, S., Pursch, M., & Albert, K. (1999). Hyphenation of high
performance liquid chromatography with nuclear magnetic reso-
nance spectroscopy for the characterization of b-carotene isomers
employing a C30 stationary phase. Journal of Pharmaceutical and
Biomedical Analysis, 21, 669–677.
Tan, B., Grady, C. M., & Gawienowski, A. M. (1986). Hydrocarbon
carotenoid profiles of palm oil processed fractions. JAOCS, 63,
1175–1179.
Tay, B. Y. P., & Choo, Y. M. (2000). Practical guide to establishing
palm carotenoids profiles by HPLC with three dimensional diode
array detector. Palm Oil Developments, 33, 13–17.
Tay, B. Y. P., Choo, Y. M., Ee, G. C. L., & Goh, S. H. (2000).
Geometrical isomers of the major provitamin A palm carotenes, a-
and b-carotenes in the mesocarp oil of fresh and sterilized palm
fruits, crude palm oil and palm carotene-based products: red palm
olein and carotene concentrates. Journal of Oil Palm Research, 13,
23–32.
Tsukida, K., Saiki, K., Takii, T., & Koyama, Y. (1982). Separation
and determination of cis/trans-b-carotenes by high-performance
liquid chromatography. Journal of Chromatography, 245, 359–364.
Yap, S. C., Choo, Y. M., Ooi, C. K., Ong, A. S. H., & Goh, S. H.
(1991). Quantitative analysis of carotenes in the oil from different
palm species. Elaeis, 3, 369–378.
Yeum, K.-J., Booth, S. L., Sadowski, J. A., Liu, C., Tang, G., Krinsky,
N. I., et al. (1996). Human plasma carotenoid response to the
ingestion of controlled diets high in fruits and vegetables. American
Journal of Clinical Nutrition, 64, 594–602.