J.

of Supercritical Fluids 100 (2015) 175183

Contents lists available at ScienceDirect

The Journal of Supercritical Fluids

journal homepage: www.elsevier.com/locate/supflu

Extraction of bixin from annatto seeds using combined technologies

Thiago Taham a , Fernando A. Cabral b , Marcos A.S. Barrozo a,
a
Federal University of Uberlandia School of Chemical Engineering, A. Joo Naves de vila, 2121, Block 1K Campus Santa Mnica, Uberlandia,
MG 38400-902, Brazil
b
University of Campinas (UNICAMP), School of Food Engineering, R. Monteiro Lobato, 80, Campinas, SP 13083-862, Brazil

a r t i c l e

i n f o

Article history:
Received 7 November 2014
Received in revised form 2 February 2015
Accepted 3 February 2015
Available online 12 February 2015
Keywords:
Annatto
Bixin
Supercritical uid extraction
Tocols

a b s t r a c t
Bixin is the most used carothenoid in food industry. It is conventionally extracted from annatto seeds
(Bixa orellana L.) by alkaline solutions or organic solvents, producing bixin with low purity and generating
toxic waste. This study aimed to compare different clean technologies for extraction of nutraceuticals
(bixin and tocols) from annatto seeds, using a xed bed extractor. The solvents used were supercritical
CO2 , ethanol and a mixture of ethanol and water. Conventional extractions using water and chloroform
were also conducted, for comparison. Scanning electron microscopy was used to evaluate the surface of
seeds and conrm experimental results. The best extraction efciencies (for bixin) were using ethanol
(ambient pressure) and a sequential extraction (supercritical CO2 as pretreatment followed by ethanol).
This last result has also shown a good recovery of -tocotrienol, a powerful natural antioxidant. Higher
concentrations of bixin in extracts were obtained using pure ethanol at ambient and high pressures.
2015 Elsevier B.V. All rights reserved.

1. Introduction
Annatto is a natural colorant that imparts colors ranging from
yellow to red due to the concentration of color compounds in the
solution [1]. This pigment is obtained from the seed coat of the
tropical shrub Bixa orellana L. This tree is native to tropical South
America, where it has been a traditional ingredient of some foods
for centuries [2].
The main pigments of annatto seeds are bixin and norbixin, whose structures are shown in Fig. 1. These pigments are
carotenoids whose colors vary between yellow and red, colors of
huge importance in the food, pharmacological and cosmetic industries. In food industries these natural pigments are used in cheeses,
sausages, meats and candies [3]. Annatto seems to be an important
natural colorant for food and drug industries owing to its potential
uses as a substitute for Tartrazine which is a synthetic colorant that
is prohibited in many countries [4]. Later annatto is classied as a
color additive exempt of certication by FDA of United States of
America [5].
Studies have shown that other peculiarity of annatto is
that its lipid fraction contains a large amount of -tocotrienol
[7,8]. Tocotrienols have been associated with hypocholesterolemic effects and are believed to be useful in the treatment of

Corresponding author. Tel.: +55 34 3230 9400.

E-mail addresses: thiago.taham@iftm.edu.br (T. Taham), cabral@fea.unicamp.br
(F.A. Cabral), masbarrozo@feq.ufu.br (M.A.S. Barrozo).
http://dx.doi.org/10.1016/j.supu.2015.02.006
0896-8446/ 2015 Elsevier B.V. All rights reserved.

cardiovascular diseases and cancer [8,9]. The lipid-rich fraction of

annatto can also be effective in the prevention of lipid peroxidation
[10]; probably, the tocotrienols combined with bixin act synergistically to protect the unsaturated lipids from oxidation [11].
Annatto pigment can be separated from annatto seeds by many
ways, including immersion of seeds in hot vegetable oil, dilute
alkaline aqueous solutions and solvents. In the rst case, the pigment is obtained by abrasion of the exocarp submerged in warm
vegetable oil (70 C). This process produces more concentrated
suspensions pigment but it may contains degradation products,
considering that, at elevated temperatures, annatto pigment gets
degraded and form several products [12,13]. Industrially, the most
used process is the extraction with an alkaline solution, usually
potassium or sodium hydroxide [14]. This process is known to
transform bixin (present in the seeds as an ester) into a diacid salt
norbixate that is soluble in water [15,16]. The removal of alkaline solution from the bixin-rich extract, as well as from the seeds,
requires several subsequent unit operations, elevating the energy
costs of the process [17]. Many solvents such as acetone, chloroform, ethanol, ethyl acetate and hexane are used for extraction
of annatto pigment due to the strong solubility of this pigment.
Solvent extraction results in the purest form of bixin pigment
[18]. The commercial preparations of annatto colors with organic
solvents have the disadvantage of small concentrations of pigments and a residual toxic solvent in the product. Solvents can also
be removed from these high color-concentrated solutions to give
bixin crystals, elevating, however, the energy costs of the process
[2,19].

176

T. Taham et al. / J. of Supercritical Fluids 100 (2015) 175183

Fig. 1. Molecular structures of bixin (R COOCH3 ) and norbixin (R COOH): (a) -cis and (b) -trans.
Source: [6].

Various techniques have been studied to develop clean

extraction technologies with environmental benets [20]. These
techniques are used to extract bioactive substances, to shorten
the processing time, reduce or eliminate solvent consumption,
using mechanical extraction [21,22], increase the extraction yield
and improve the quality of the extracts. These studies are needed
because of the general trend of the market to identify products that
generate economic, social and environmental advantages [2,23].
Supercritical extraction with CO2 could be a good alternative in
these cases avoiding environmental problems.
Supercritical CO2 with different pressures and temperatures
proved to be advantageous for removing compounds from complex food systems than conventional methods [24] because of least
thermal effects on products, high quality of recovered products,
low energy requirement for solvent recovery and high selectivity
in the separation process. Recently, many attempts have been carried out to use supercritical carbon dioxide uid for the extraction
of pigment from annatto seed. However, the results of these studies
seemed to be unsatisfactory both in terms of economics and efciency [1,6,2427]. These studies showed that the efciency of the
extraction was improved with increasing temperature and pressure, the seed oil acted as a co-solvent in the extraction of bixin
from the seeds and the yield of bixin was signicantly increased
with the use of organic solvents as co-solvents [1,24,26,27]. Supercritical CO2 method has been studied, also, as a pretreatment for
defatting of annatto seeds [19,20]. These studies aimed to remove
a lipid layer strongly associated to the bixin in the seed; the removal
of the lipidic layer could increase bixin recovery.
Based on the improvement of the clean technologies, the objective of this work was to carry out supercritical uid extraction from
annatto seeds with CO2 combined with other extraction methods
which have low environmental impact, applying both high and low
pressure, in order to increase the results and make the process more
economically attractive.
2. Materials and methods
2.1. Materials
Annatto seeds were obtained from Rio Vermelho MG and
stored at 18 C, protected from light in dark bags of polyethylene
until further analysis and rst tests. An 18 kg cylinder of liquid CO2
was provided by White Martins (Praxair Inc., Campinas, Brazil). The
solvents used in analyses were: chloroform, hexane, acetone and
ethanol. All of them were of analytical-grade.
2.2. Seeds characterization
A batch of the seeds was characterized for moisture, protein, ash,
total lipids and bixin content. Moisture, protein and ash contents
were determined according to the ofcial methods published by the
Association of Ofcial Analytical Chemists (AOAC) [28]. The seed
oil content was quantied by gravimetric analysis, evaporating

until dry the total volume of an extraction in a Soxhlet apparatus

with hexane (P.A.). The carbohydrate content was determined as
the difference, experiments were carried out in duplicate.
The annatto oil extracted was also characterized in term of
its fatty acids prole using capillary gas chromatography (GC)
according to AOCS ofcial method Ce 1-62 [29]. The analysis
was performed on a CGC AGILENT 68650 SERIES GC SYSTEM,
column DB-23 AGILENT (50% cyanopropil) methylpolysiloxane,
60 m 0.25 mm 0.25 m, column ow rate of 1.0 mL/min, carrier
gas: Helium and volume injected 1.0 L.
The tocols content (tocopherol and tocotrienol) was determinated by high performance liquid chromatography (HPLC)
according AOCS ofcial method Ce 8-89 [29]. The isomers (alpha,
beta, gamma, and delta tocotrienols and tocopherols) were quantied using a standard curve prepared under the same conditions as
that of the analysis. The analysis was performed on a Hibar RT Column (250 mm 4 mm Li Chrosorb Si 60, 5 mm). The mobile phase,
hexane/isopropanol (99:1) was used under isocratic conditions at
a ow rate of 1.0 mL/min.
The pigment content of the seeds was determined by a method
using organic extraction [12], which consists in repeated extraction
of the seeds with chloroform, until the exhaustion of pigment contained in the seeds. This extraction was conducted at 50 C using a
balance cell coupled to a reux condenser, whose purpose was to
prevent the loss of desirable compounds (volatiles). The temperature of the bath connected to the Soxhlet was 5 C. These conditions
were chosen based in the work performed by Silva et al. [30] whose
results show that, besides the good performance of chloroform as
solvent, temperatures below 80 C cause minimum degradation in
bixin.
For results comparison, the same extraction was performed
with water, also at 50 C. The bixin content of the seeds was
determinate, then, using spectrophotometric method according
to Joint FAO/WHO Expert Committee on Food Additives Monographs [31]. Sample Absorbance was measured in a 1 cm quartz
cuvette at 487 nm, using a UVvis spectrophotometer. The percentage of bixin in the annatto seeds was calculated, then, according to
LambertBeer law, using E11%cm = 3090 [31], in Eq. (1):
Bixin (%) =

A V1 . . . Vn
E msample V1 . . . Vn

(1)

In which:
A = average absorbance of the samples;
Vi = dilution volume (i = 1, 2,. . ., n);
E = absorptivity coefcient E (equal to 3090 for bixin);
Msample = sample mass, in grams; and
Vi  = volume of aliquot for dilution (i = 1, 2,. . ., n).
2.3. Fixed bed extraction
The xed bed experiments were conducted in an experimental unit bench, in the Laboratory of Extraction, Applied

T. Taham et al. / J. of Supercritical Fluids 100 (2015) 175183

177

Fig. 2. Scheme of the equipment.

thermodynamics and Equilibrium (EXTRAE), Faculty of Food

Engineering, UNICAMP, Campinas, Brazil. Fig. 2 shows a schematic
diagram of the equipment.
The experimental apparatus consisted mainly of a CO2 cylinder
(1), refrigerated bath (2), high-pressure pump (3), supply tank (4),
extractor (7), collection ask (9), gas ow meter (10), volume totalizer (11), a thermostatic bath, two Bourdon pressure gauges, one
located in the supply tank (5) and the other on the inlet side of the
extractor (6), a peristaltic pump (8) used to inject the ethanol (to
wash the internal walls of the pipe), and a high pressure pump
for ethanol/water extractions (12). The extractor consisted of a
50 mL equilibrium cell (stainless steel) immersed in a constanttemperature water bath controlled by a heater (accurate to within
0.1 C). The CO2 from the supply tank was cooled to the liquid
state (by a refrigeration bath) and compressed into the supply tank
and the extractor by a high pressure pump. In liquid extractions,
a beaker containing the solvent was positioned under the water
bath and a high pressure pump was used to inject the uid into
the extractor, without refrigeration of the solvent. After reaching
the equilibrium conditions of temperature and pressure, a sample
of the extract mixture was transferred to a collector of known volume and then collected in the glass tube. To recover the material
remaining in the collector, a peristaltic pump circulated solvent
through the collector by entrance valve.
Approximately 15 g of whole annatto seeds were placed into
the extraction cell, completing the volume with glass beads (3 mm
in diameter). This arrangement makes the system completely
full, forcing the solvent to pass through the particles, aiming to
improve the solubilization of the interest compounds. After the
unit preparation, the temperature-controlled bath was lled with
water to cover the extractor and the valves. The temperature of all
extractions was 50 C. In experiments conducted at high pressure
(400 bar), a pressurization test of 20 min was performed to check
leaks and provide a static period of contact between the solvent and
the seeds. In the experiments at ambient pressure, a small amount
of solvent was pumped into the system, providing a pressure equal
to 5 bar, to ensure the complete lling of the bed and optimizing
the contact between the solvent and seeds. For extraction, the
solvent was continuously drained through the bed and the extract
collected in a previously weighed collection bottle. This one was

wrapped in aluminum foil and kept refrigerated (5 C) during and

after extraction to avoid oxidation of the extract components.
The rst step of the extractions was the ones carried out using
a single solvent (SSE). The solvents used were ScCO2 (400 bar),
ethanol (400 bar and ambient pressure) and a mixture between
ethanol and distillated water (70:30 volume/volume at 400 bar and
ambient pressure). Besides the single-solvent extractions (SSE) and
conventional ones mentioned in Section 2.2, sequential extractions
(SE) were performed in order to obtain the bixin extraction kinetics. The solvents used were, in sequence, supercritical CO2 , ethanol
and a mixture between ethanol and distillated water (70:30 v/v).
These solvents were pumped through the same seeds in the extraction cell, which is characteristic of the sequential extraction. This
extraction was carried out in triplicate. In the rst and second
experiments of the triplicate, samples were collected over time, in
order to construct an extraction curve. In the third experiment, one
single sample was taken at the end of each time step of extraction.
The last step of extraction was carried out according the
methodology described by Albuquerque and Meireles [19], using
partially defatted seeds by supercritical extraction (DSE). It consists in a sequential extraction, wherein the seeds are pre-treated
with ScCO2 (which extracts the annatto oil) followed by an extraction with solvent at ambient pressure. For comparison, this annatto
oil extracted was also characterized in term of its fatty acids prole
according to AOCS ofcial method Ce 1-62 [27] and tocols content
according AOCS ofcial method Ce 8-89 [27]. The solvents used in
the second step of the extraction were ethanol and the mixture
between ethanol and water mentioned above. Table 1 shows the
operating conditions of the experiments carried out in xed bed.
The gas ow rate (CO2 ) is monitored by a gas ow meter. After
the time of extraction, the pipe used to remove the extract (after
the extraction cell) was washed with ethanol, using of a peristaltic
pump, to recover the extract retained in the pipe. The solvent was
evaporated off in a rotary evaporator at 50 C with a pressure of
0.93 bar, and the extraction yield was determined using Eq. (2).
All vials were stored, then, at 18 C and protected from light, until
further analyses were conducted.
Global extraction yield(X0 %) =

mass of extract(g)
100
mass of seeds(g)

(2)

178

T. Taham et al. / J. of Supercritical Fluids 100 (2015) 175183

Table 1
Operating conditions.a
Experiment codication

Solvent

Solvent ow rate

Temperature ( C)

Pressure (bar)

Time (min)

SSE1
SSE2
SSE3
SSE4
SSE5

scCO2
Ethanol
Ethanol
EthanolWater
EthanolWater
scCO2
Ethanol
EthanolWater
scCO2
Ethanol
EthanolWater
scCO2
Ethanol
EthanolWater
scCO2
Ethanol
scCO2
EthanolWater

1.5 L/min
0.5 mL/min
0.5 mL/min
0.5 mL/min
0.5 mL/min
1.5 L/min
0.5 mL/min
0.5 mL/min
1.5 L/min
0.5 mL/min
0.5 mL/min
1.5 L/min
0.5 mL/min
0.5 mL/min
1.5 L/min
0.5 mL/min
1.5 L/min
0.5 mL/min

50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50

400
5
400
5
400
400
400
400
400
400
400
400
400
400
400
5
400
5

300
540
540
540
540
210
360
360
210
360
360
210
360
360
120
420
120
420

SE1

SE2

SE3
DSE1
DSE2
a

All ow rate measures were done at ambient temperature (25 C). The CO2 ow rate was measured in gaseous state at ambient pressure (0.93 bar).

The bixin content in the extract (%) was expressed according to

Eq. (3) that involves the mass of bixin obtained and the total mass
of extract.
Bixin content in extract(C0 %) =

mass of bixin(g)
100
mass of extract(g)

(3)

It also was calculated the ratio between the mass of bixin

extracted and the mass of raw material (seeds) used, according Eq.
(4).
Ratio between bixin recovery and seeds(Q )
=

mass of bixin obtained(g)

100
mass of seeds(g)

(4)

Finally, it was calculated the recovery of bixin (R), according to

Eq. (5). The process that had the highest Q values was considered
as 100% recovery.
Recovery(R) =

Qtested
100
Qhighest

(5)

3. Results and discussion

3.1. Annatto seeds characterization
The average seed moisture was 11.5% (wet basis). The content
of ash, protein, oil and carbohydrate were, respectively, 3.82%;
10.34%; 4.99% and 69.35%. The values obtained in this study are similar to those found in the literature for annatto seeds [1,6,19,30,31].
The bixin content was determined performing a conventional
extraction until the total depletion of the seeds colorant. The
conventional extractions, using separately water and chloroform,
resulted in the overall extraction prole shown in Table 2.
The extraction with chloroform was considered, in this work,
as it reached 100% efciency (recovery equal to 100%). The bixin
content was calculated based on the mass of extract obtained (C0 ),
and it was equal to 2.47%.

Table 2
Results of bixin content in the seeds performing conventional extraction.
Solvent

Mraw material (g)

Mextract (g)

Mbixin (g)

C0 (%)

X0 (%)

Q (%)

Water
Chloroform

1.0519
1.0563

0.4494
0.3638

0.0053
0.0090

1.18
2.47

42.72
34.44

0.50
0.85

3.2. Fixed bed extraction

Five single solvent extractions (SSE) were conducted; they
lasted different total times based on preliminary tests. Table 3
shows the results obtained in each extraction and Fig. 3 presents,
graphically, the kinetic curves found.
All values of bixin (%) obtained in the extract were higher than
the results found in Table 2. This fact occurs because, in conventional extraction, the experiment was carried out until the
exhaustion of the annatto seeds. It was dragged then, in addition to
the pigments, all other components of the pericarp. This fact contributed to increase the total mass of extract obtained, reducing the
percentage of this bixin in this.
The supercritical extraction (SSE1) had the lowest capacity
to extract compounds from the whole seed. In the rst hour, it
was extracted annatto oil, with an intense red color, followed by
a decreasing extraction of pigments. Comparing the extractions
(SSE2 and SSE3; SSE4 and SSE5), it is noticed that the pressure
increase causes a reverse effect on the extraction efciency. When
the experiment was conducted at high pressure, the total mass of
extract and bixin obtained were smaller, in both cases. The increase
in solvent polarity was also not effective to improve the extraction
efciency (comparing SSE2 and SSE4; SSE3 and SSE5): despite the
mixture between ethanol and water increased the overall yield, the
content of bixin were much smaller (in terms of mass or percentage). These results show that water is able to extract, apart from
bixin, other compounds of the seed. The results in Table 3 show
that the extraction of bixin, under the operating conditions tested,
was more efcient at lower pressures and using ethanol as solvent.
Three sequential extractions (SE) were also performed in the
same operating conditions (see Table 1, Section 2.3), in order to
obtain the bixin extraction kinetics. The solvents used in all extractions were, in sequence, supercritical CO2 , ethanol and a mixture
between ethanol and distillated water (70:30 v/v). The pressure of
the system was 400 bar during the entire experiment. Table 4 shows
the results obtained and Fig. 4 presents, graphically, the kinetic
curves found:
The total time of these extractions was the highest among all
(15 h and 30 min), which was done to ensure a sufcient contact
time between the seeds and the respective solvent in its pure form
since, during the exchange of the solvent, a mixture occurs in the
equilibrium cell. The ratio between the bixin and the mass of seeds
in the extraction shows a good repeatability in these extractions,
performed in the same operating conditions, as we can see in Fig. 4.
The SE3 experiment is shown without a curve in the graphic; as

T. Taham et al. / J. of Supercritical Fluids 100 (2015) 175183

179

Table 3
Single solvent extraction results (SSE).
Exp.

Solvent

P (bar)

Mraw material (g)

Mextract (g)

Mbixin (g)

X0 (%)

C0 (%)

Q (%)

SSE 1
SSE 2
SSE 3
SSE 4
SSE 5

ScCO2
Ethanol
Ethanol
EthanolH2 O
EthanolH2 O

400
5
400
5
400

15.1259
15.8366
15.0177
15.2854
15.0017

0.2591
0.9243
0.6958
1.3160
0.9535

0.0129
0.1328
0.0777
0.0842
0.0496

1.71
5.84
4.63
8.61
6.36

4.98
14.37
11.17
6.40
5.20

0.09
0.84
0.52
0.55
0.33

Fig. 3. Kinetics of bixin extraction using single-solvent.

Table 4
Results of sequential extractions (SE).
Exp.

Mraw material (g)

Mextract (g)

Mbixin (g)

X0

C0 (%)

Q (%)

SE1

14.9153

0.2525
0.5002
0.4506

0.0121
0.0607
0.0196

1.69
3.35
3.02

4.79
12.14
4.35

0.62

SE2

15.1358

0.2591
0.4979
0.4314

0.0085
0.0714
0.0234

1.71
3.30
2.85

3.28
14.34
5.42

0.68

SE3

15.066

0.2691
0.4947
0.4339

0.0113
0.0663
0.0242

1.79
3.28
2.90

4.20
13.40
5.58

0.67

Fig. 4. Kinetics of the sequential extractions of bixin (SSE).

180

T. Taham et al. / J. of Supercritical Fluids 100 (2015) 175183

Table 5
Results of defatted seeds sequential extractions (DSE).
Exp.

Mraw material (g)

Mextract (g)

Mbixin (g)

X0

C0 (%)

Q (%)

DSE1

15.8270

0.2736
1.1102

0.0023
0.1089

1.73
7.01

8.40
9.81

0.70

DSE2

15.5874

0.2498
1.7663

0.0020
0.0324

1.60
11.33

8.00
1.83

0.22

explained before (Section 2.3), the samples were collected only at

the end of each time step of extraction. Despite the higher extraction time, none of these three results was able to overcome the ratio
between the mass of bixin recovered and the mass of raw material
obtained using ethanol at low pressure in a single step.
Highest amounts of bixin recovered and its high levels (%) in
the extract were obtained using ethanol as solvent, as also shown
in SSE extractions. The percentages of bixin contained in ethanol
extracts after the supercritical extraction showed that a large proportion of the colorant is not removed in the rst extraction step.
Similar results were found in experimental works of Albuquerque
and Meireles [19] and Rodrigues et al. [20].
The last three extractions performed (DSE) were also considered
as sequential extractions. The difference between them (DSE) and
the sequential ones (SE) was the change in operating conditions
(see Table 1). Table 5 shows the results obtained and Fig. 5 presents,
graphically, the kinetic curves found.
The results shown in Fig. 5 ignore the time that the annatto seeds
was subjected to pretreatment with supercritical uid, since the
bixin recovery in this step is very small. The application of a subsequent extraction with solvent at low pressure increased the amount
of bixin obtained in both cases (DSE1 and DSE2), being signicantly
higher using only ethanol as solvent (DSE1). Similar results were
obtained in this work (see Fig. 3, Section 3.2); the presence of water
in extraction increases the amount of global extract (Table 5) and
reduces the amount and percentage of bixin in this, conrming the
ability of water to extract other compounds than bixin.
Similar operating conditions and results were found in experimental works of Albuquerque and Meireles [19] and Rodrigues et al.
[20], indicating that the pretreatment using ScCO2 could improve
the bixin extraction. In this work, however, the amount of bixin
recovered was not higher than those ones obtained with the lowpressure extraction using ethanol (SSE2), although the total time

of the two experiments was exactly the same. One can be concluded, then, that among the conditions tested, no method was
more efcient for bixin recovery than extraction with ethanol at
low pressure. Table 6 shows the values of bixin recovery compared
with each other and research works found in literature.
The experiment conducted using only carbon dioxide as solvent
(SSE1) showed a slightly higher value of bixin recovery comparing
with the work presented by Albuquerque et al. [19], since the time
of the process was higher too. Rodrigues et al. [20] obtained 31.62%
of bixin recovery using an extraction with ethanol at low pressure after pre-treatment of the seeds using ScCO2 . The low value
of recovery may be due to the time of process (20 min). The use of
ethanol as cosolvent in supercritical uid extraction improved the
bixin recovery (50%), according to Nobre et al. [1]. Even so, it did not
exceed the result that uses pure ethanol at low or high pressures
(SSE 2, SSE 3). The other research result presented in Table 6 [27]
had also lower values of bixin yield (Q = 0.27), despite adding an
organic solvent in the extraction (acetonitrile containing 0.05% triuoroacetic acid). These facts conrm that the use of ethanol at low
pressures can recover almost entirely the bixin of the seeds, with a
lower environmental impact than the use of other organic solvents
(such as chloroform or acetonitrile) and using a classical technology. Furthermore, as seen in Table 3, the highest concentration of
bixin (14.4%) in the extract has achieved with its use at low pressure
(SSE 2), among all conditions tested.
It is important to emphasize that annatto extracts rich in bixin
have been frequently used as basis in several research works.
Suo et al. [32] studied bixin micronization aiming to evaluate the
efciency of the supercritical uid precipitation in processing of
natural pigments. Popoola [33] studied the use of puried bixin
and curcumin as color additives in two brous constructions of oor
carpet and cotton fabric. Microencapsulation of the extract for food
industry application has also been studied [34]. The positive effects

Fig. 5. Kinetics of the bixin extractions from defatted seeds (DSE).

T. Taham et al. / J. of Supercritical Fluids 100 (2015) 175183

181

Table 6
Values of bixin recovery (R).
Exp.

Solvent

P (bar)

Time (min)

Q (%)

Recovery (%)

Conventional
Conventional
SSE 1
SSE 2
SSE 3
SSE 4
SSE 5
SE 1
SE 2
SE 3
DSE 1
DSE 2
Nobrea [1]
Albuquerqueb [19]
Rodriguesc [20]
Andersond [27]

Chloroform
Water
ScCO2
Ethanol
Ethanol
Ethanol/H2 O
Ethanol/H2 O
ScCO2 /Ethanol/Ethanol/H2 O
ScCO2 /Ethanol/Ethanol/H2 O
ScCO2 /Ethanol/Ethanol/H2 O
ScCO2 /Ethanol
ScCO2 /Ethanol/H2 O
ScCO2 /Ethanol
ScCO2
Ethanol
ScCO2

Ambient
Ambient
400
5
400
5
400
400
5
400
400/5
400/5
200
300
Ambient
606

270
270
300
540
540
540
540
930
930
930
420
420

210
20

0.85
0.50
0.09
0.84
0.52
0.55
0.33
0.62
0.68
0.67
0.70
0.22

0.07e

0.27e

100
58.82
1.05
98.82
61.18
64.71
38.82
72.94
80.00
78.82
82.35
25.88
50.00e
0.80e
31.62e

a
b
c
d
e

Extraction with ScCO2 using ethanol as co-solvent.

Extraction using only ScCO2 .
Low pressure solvent extraction assisted by ultrasound.
Extraction with ScCO2 containing acetonitrile.
The calculation of these parameters consider the information given in original article.

Fig. 6. Structure of whole seeds. (a) 50 and (b) 1000.

of the annatto extract as anticancer [35] or on glucose levels in diabetic rats [36] were also observed, demonstrating a wide range of
use of these extracts.
3.3. Scanning electron microscopy
The surfaces of annatto seeds were analyzed by scanning electron microscopy (SEM), aiming to visualize in photographs the
experimental results obtained. Increases of 50 and 100 times in
size of the seeds were observed. The photographs were taken from
whole seed and the seeds collected in the equilibrium cell after the

experiments SSE1 (only ScCO2 ), SSE2 (ethanol at low pressure) and

DSE1 (defatted seeds plus ethanol at low pressure).
Fig. 6(a and b) shows a whole annatto seed. It could
be noticed an intact structure, highly brous and a globular
format.
Fig. 7(a and b) shows an annatto seed after treatment with at
400 bar during 5 h. The surface of the seeds (the most of them) has
been disrupted (Fig. 7a), revealing the interior. The removal of oil
(using ScCO2 ) also induced a higher porosity of the surface and
wilting of the globules (Fig. 7b). Similar results of damages were
obtained in the sequential extractions (SE).

Fig. 7. Structure of seed treatment using ScCO2 . SSE1 (a) 50 and (b) 1000.

182

T. Taham et al. / J. of Supercritical Fluids 100 (2015) 175183

Fig. 8. Structure of seed ethanol at ambient pressure SSE2. (a) 50 and (b) 1000.

Fig. 9. Structure of seed ScCO2 /ethanol at low pressure DSE1. (a) 50 and (b) 1000.

Fig. 8(a and b) shows an annatto seed after extraction using

ethanol at ambient pressure. Although the intact globular structure of the seeds has still remained, the surface is more wilted and
less brous, comparing with the whole seed. This fact indicates that
treatment using ethanol at low pressures causes fewer injuries to
the seed structure among the tests done.
Finally, Fig. 9(a and b) shows an annatto seed after the
pre-treatment using ScCO2 and ethanol at low pressure (DSE1).
Although the seeds have maintained their globular structure, some
disruption was observed (Fig. 9a). The seeds have shown, also, a
higher porosity, comparing with whole ones (Fig. 9b).

The extraction of oil using ScCO2 was performed using 32.9323 g

of whole seeds; 0.5272 g of oil were obtained (1.60% of yield). Conventional extraction with n-hexane was performed using 19.9374 g
of whole seed, in 4 different batches; 0.7966 g of oil were obtained
(3.99% of yield). Similar result was found by Gaydou and Ramanoelina [37], whose conventional extraction obtained 3.9% of yield.
It was performed the free fatty acids characterization of the
annatto oil extracted by different ways described before: through
the pre-treatment with ScCO2 (which extracts the annatto oil) and
the conventional extraction with hexane. The results of the principal free fatty acids found on these two samples of oil are shown in
Table 7.

Similar results were found by Matos [38], in their studies of

Brazilian tropical oil seeds. Their values for palmitic, stearic, oleic
and linoleic acids were, respectively: 19.3; 13.0; 15.5 and 19.3%. The
composition of the oil obtained by ScCO2 and n-hexane showed no
signicant difference in terms of fatty acid prole.
The tocols (tocopherol and tocotrienol) content was also performed in the two samples of oil. It was not found tocopherol in any
of the samples (oil extracted with ScCO2 or Hexane). The results
found are in agreement with the literature, which characterizes
annatto seeds as tocopherol-free [39]. The results of tocotrienol
content are shown in Table 8.
The results, expressed in g/mL are close to the experimental
ones obtained by Albuquerque and Meireles [19], whose quantities of gamma-tocotrienol ranged from 1.57 to 1.97 g/mL and
delta-tocotrienol ranged from 11.7 to 14.6 g/mL. The results
obtained by extraction using hexane (conventional extraction)
were just slightly higher than those obtained with supercritical
uid extraction using CO2 as solvent. This fact indicates that the
pre-treatment of the seeds using supercritical uid extraction can
result on a recovery of a component of high biological value, since
delta-tocotrienol is the most potent isomer of vitamin E and the
compound most considerable referring to their pharmacological
properties [40]. After this procedure for vitamin E recover, one
can see that the bixin yield obtained was satisfactory and your
ratio between bixin recovery and the mass of the seeds (Q) were
closer between each other (0.84 using ethanol at low pressure and

Table 7
Free fatty acids characterization of annatto oil.

Table 8
Results of defatted seeds sequential extractions (DSE).

3.4. Oil characterization

Conc. (%)

Palmitic
C16:0

Stearic C18:0

Oleic C18:1

Linoleic
C18:2

Conc. (g/mL)

Sample mass
(g)

Alpha

Beta

Gamma

Delta

Sample 1 (ScCO2 )
Sample 2 (Hexane)

19.36
17.95

5.11
8.07

22.44
20.53

21.29
20.90

Sample 1 (ScCO2 )
Sample 2 (Hexane)

0.01034
0.01104

0.00
0.00

0.00
0.00

2.1749
2.2922

13.2481
13.4928

T. Taham et al. / J. of Supercritical Fluids 100 (2015) 175183

0.70 using defatted seeds). The combined application of these two

technologies can, then, improve the nal results obtaining a large
amount of delta-tocotrienol in the pre-treatment step and causing
no strong interference on the extraction of bixin.
4. Conclusion
In this study some extraction methods were tested and combined with the focus on bixin recovery from annatto seeds. None of
the combinations tested were more effective than ethanol extraction at ambient pressure. This condition was responsible for higher
values of bixin obtention among all conditions tested. Moreover,
this condition still resulted in the highest percentages of bixin in the
extract, characterized as the most pure bixin obtained among the
conditions tested. In the other hand, we could see that the pretreatment of the seeds, using supercritical CO2 , results in obtainment
of an oil rich in vitamin E without removing the most part of the
bixin present in the annatto seeds. This important pigment can be
extracted, then, in a subsequent unit operation using low pressures.
Acknowledgment
The authors would like to acknowledge National Council for Scientic and Technological Development (CNPq: 301563-2010-01)
for nancial support.
References
[1] B.P. Nobre, R.L. Mendes, E.M. Queiroz, F.L.P. Pessoa, J.P. Coelho, A.F. Palavra,
Supercritical carbon dioxide extraction of pigments from Bixaorellana seeds
(experiments and modeling), Brazilian J. Chemical Engineering 23 (2) (2006)
251258.
[2] H.V. Chuyen, N.T. Ngoc Hoi, J.-B. Eun, Improvement of bixin extraction yield
and extraction quality from annatto seed by modication and combination of
different extraction methods, International J. of Food Science & Technology 47
(2012) 13331338.
[3] M. Furtado, Dyes: food industry chooses natural dyes, Chemicals and Derivatives Magazine 421 (2003) 110 (in Portuguese).
[4] JECFA, Evaluation of Certain Food Additives and Contaminants; 26th Report
of the Joint FAP/WHO Expert Committee on Food Additives; Technical Report
Series, vol. 683, WHO, Geneva, 1982, pp. 51.
[5] J.B. Hallagan, D.C. Allen, J.F. Borzelleca, The safety and regulatory status of
food, drug and cosmetics colour additives exempt from certication, Food and
Chemical Toxicology 33 (6) (1995) 515528.
[6] G.F. Silva, F.M.C. Gamarra, A.L. Oliveira, F.A. Cabral, Extraction of bixin from
annatto seeds using supercritical carbon dioxide, Brazilian J. Chemical Engineering 25 (2008) 419426.
[7] N. Frega, M. Mozzon, F. Bocci, Identication and estimation of tocotrienols in the
annatto lipid fraction by gas chromatographymass spectrometry, J. American
Oil Chemists Society 75 (1998) 17231727.
[8] B. Tan, Appropriate spectrum vitamin E and new perspectives on desmethyl
tocopherols and tocotrienols, J. American Nutraceutical Association 8 (2005)
3542.
[9] B. Tan, J. Foley, Tocotrienols and geranylgeraniol from Bixaorellana byproducts,
Patent US6350453 (2002).
[10] R.C. Chist, A.Z. Mercadante, A. Gomes, E. Fernandes, J.L.F.C. Lima, N. Bragagnolo,
In vitro scavenging capacity of annatto seed extracts against reactive oxygen
and nitrogen species, Food Chemistry 127 (2011) 419426.
[11] W.F. Castro, L.R.B. Mariutti, N. Bragagnolo, The effects of colorico on lipid oxidation, colour and vitamin E in raw and grilled chicken patties during frozen
storage, Food Chemistry 124 (2011) 126131.
[12] G.G. McKeown, E. Mark, The composition of oil-soluble annatto food colors,
Journal of the AOAC 45 (3) (1962) 761766.
[13] J. Smith, Annatto extracts chemical and technical assessment, in: Chemical
and Technical Assessment Manual, 2006, pp. 121.
[14] P.R.N. Carvalho, M.G. Silva, E.G. Fabri, P.E.R. Tavares, A.L.M. Martins, L.R. Spatti,
Bixin and lipids content in annatto seeds from collection of the Instituto
Agronmico (IAC), Bragantia 69 (2010) 519524.

183

[15] P.R.N. Carvalho, The colorants from Annatto, in: Proceedings of the 2nd National
Meeting of the Annatto Production Chain, Campinas, Brazil, 2010 (Full paper
available in 9th ISSF CD-ROM (ISBN 978-85-7029-105-9)).
[16] M. Scotter, The chemistry and analysis of annatto food colouring: a review,
Food Additives Contaminants 26 (2009) 11231145.
[17] M.A.A. Meireles, P.F., Leal, P.T.V. Rosa, Extraction process and purication process in series of active compounds and colorants from solid matrices using
supercritical CO2 : bixin from annatto, Patent BRPI0605425, 2008 (in Portuguese).
[18] C. Prentice-Hernndez, O. Rusig, Bixin extract from annatto (Bixaorellana L.)
obtained using ethanol as solvent, Biology and Technology 35-1 (1992) 6374
(in Portuguese).
[19] C.L. Albuquerque, M.A.A. Meireles, Defatting of annatto seeds using supercritical carbon dioxide as a pretreatment for the production of bixin: experimental,
modeling and economic evaluation of the process, Journal of Supercritical Fluids 66 (2012) 8695.
[20] L.M. Rodrigues, S.C. Alczar-Alay, A.J. Petenate, M.A.A. Meireles, Bixin
extraction from defatted annatto seeds, ComptesRendusChimie 17 (2014)
268283.
[21] F.G. Cunha, K.G. Santos, C.H. Atade, N. Epstein, M.A.S. Barrozo, Annatto powder
production in a spouted bed: an experimental and CFD study, Industrial and
Engineering Chemistry Research 48 (2009) 976982.
[22] F.G. Cunha, K.G. Santos, M.A.S. Barrozo, Mechanical extraction of natural dye
from Bixa orellanna seeds in spouted bed, Industrial Crops and Products 45
(2013) 279282.
[23] L.J. Wang, C.L. Weller, Recent advances in extraction of nutraceuticals from
plants, Trends in Food Science & Technology 17 (2006) 300312.
[24] R.R. Chao, S.J. Mulvaney, D.R. Sanson, F. Hsieh, M.S. Tempesta, Supercritical CO2
extraction of annatto (Bixaorellana) pigments and some characteristics of the
color extracts, J. Food Science 56 (1991) 8083.
[25] A. Satyanarayana, P.G. PrabhakaraRao, D.G. Rao, Chemistry, processing and toxicology of annatto (Bixaorellana L.), J. Food Science and Technology 40 (2003)
131141.
[26] A.J. Degnan, J.H. Von Elbe, R.W. Hartel, Extraction of annatto seed pigment by
supercritical CO2 , J. Food Science 56 (1991) 16561659.
[27] S.G. Anderson, M.G. Nai, A. Chandra, E. Morrison, Supercritical uid carbon dioxide extraction of annatto seeds and quantication of trans-bixin
by high pressure liquid chromatography, Phytochemical Analysis 8 (1997)
247249.
[28] Association of Ofcial Analytical Chemists (AOAC), Ofcial Methods of Analysis,
16th ed., 1997, Washington.
[29] American Oil Chemists Society (AOCS), Ofcial Methods and Recommended
Practices of the American Oil Chemists Society, 5th ed., AOCS Press, Champaign,
USA, 2004.
[30] G.F. Silva, S.A. Cavalcanti, M.C. Sobral, Extraction of annatto colourants II,
Proceedings of Brazilian Association of Chemistry 43-1/2 (1994) 5864.
[31] Food and Agriculture Organization of the United Nations (FAO), Combined
Compendium of Food Additive Specications, in: 67th Joint FAO/WHO Expert
Committee on Food Additives, vol. 3, 2006, p. 11.
[32] Q.L. Suo, W.Z. He, Y.C. Huang, C.P. Li, H.L. Hong, Y.X. Li, M. Da Zhu, Micronization
of the natural pigment-bixin by the SEDS process through prelming atomization, Powder Technology 154 (23) (2005) 110115.
[33] A.V. Popoola, Comparative fastness assessment performance of cellulosic bers
dyed using natural colorants, J. Appl. Polym. Sci. 77 (4) (2000) 752755.
[34] R. De Marcoa, A.M.S. Vieira, M.C.A. Ugrib, A.R.G. Monteiroa, R.d.C. Bergamasco,
Microencapsulation of annatto seed extract: stability and application, Chemical
Engineering Transactions 32 (2013) 17771782.
[35] J.D. Tibodeau, C.R. Isham, K.C. Bible, Annatto constituent cis-bixin has selective
antimyeloma effects mediated by oxidative stress and associated with inhibition of thioredoxin and thioredoxin reductase, Antioxidants & Redox Signaling
13 (7) (2010) 987997.
[36] M. Roehrs, C.G. Figueiredo, M.M. Zanchi, G.V. Bochi, R.N. Moresco, A. Quatrin,
S. Somacal, L. Conte, T. Emanuelli, Bixin and norbixin have opposite effects on
glycemia, lipidemia, and oxidative stress in streptozotocin-induced diabetic
rats, International Journal of Endocrinology 2014 (2014) 110.
[37] E.M. Gaydou, A.R.P. Ramanoelina, Study of fatty acids composition of plant
oil extracts from Madagascar, French Review of Body Fat 30 (1983) 2125 (in
French).
[38] F.J.A. Matos, J.W. Alencar, A.A. Craveiro, N.I.L. Machado, Fatty acids from tropical
oilseeds in Brazilian northwest, New Chemistry 15 (3) (1992) 181185.
[39] B. Tan, Appropriate spectrum vitamin E and new perspectives on desmethyltocopherols and tocotrienols, J. American Nutraceutical Association 8 (2005)
3542.
[40] D. Hussein, H. Mo, -[delta]-Tocotrienol-mediated suppression of the proliferation of human PANC-1, MIA PaCa-2, and BxPC-3 pancreatic carcinoma cells,
Pancreas 38 (4) (2009) 124136.