Model Studies on the Photosensitized Isomerization of Bixin

Article abstract of DOI:10.1021/jf0349026

The photosensitized isomerization reaction of the natural cis carotenoid bixin (methyl hydrogen 9′-cis-6, 6′-diapocarotene-6, 6′-dioate) with rose bengal or methylene blue as the sensitizer in acetonitrile/methanol (1:1) solution was studied using UV-vis spectroscopy, high-performance liquid chromatography (HPLC), and time-resolved spectroscopic techniques, such as laser-flash photolysis and singlet oxygen phosphorescence detection. In both N₂- and air-saturated solutions, the main product formed was all-trans-bixin. The observed isomerization rate constants, k _obs, decreased in the presence of air or with increase in the bixin concentration, suggesting the participation of the excited triplet state of bixin, ³Bix*, as precursor of the cis→ trans process. On the other hand, bixin solutions in the absence of sensitizer and/or light did not degrade, indicating that the ground state of bixin is stable to thermal isomerization at room temperature. Time-resolved spectroscopic experiments confirmed the formation of the excited triplet state of bixin and its deactivation by ground state bixin and molecular oxygen quenching processes. The primary isomerization products only degraded in the presence of air and under prolonged illumination conditions, probably due to the formation of oxidation products by reaction with singlet molecular oxygen. An energy-transfer mechanism was used to explain the observed results for the bixin transformations, and the consequences for food color are discussed.

Full text of DOI:10.1021/jf0349026

J. Agric. Food Chem. 2004, 52, 367−373 367
Model Studies on the Photosensitized Isomerization of Bixin
MARIANA A. MONTENEGRO,† ALESSANDRO DE O. RIOS,‡
ADRIANA Z. MERCADANTE,‡ MOÄ NICA A. NAZARENO,† AND
CLAUDIO D. BORSARELLI*,†
Instituto de Ciencias Qu´ımicas, Universidad Nacional de Santiago del Estero, Av. Belgrano (S) 1912,
4200 Santiago del Estero, Argentina, and Departamento de Cieˆncia de Alimentos, Facultade de
Engenharia de Alimentos, UNICAMP, Cx. Postal 6121, 13083-970 Campinas, SP, Brazil
The photosensitized isomerization reaction of the natural cis carotenoid bixin (methyl hydrogen 9′-
cis-6, 6′-diapocarotene-6, 6′-dioate) with rose bengal or methylene blue as the sensitizer in acetonitrile/
methanol (1:1) solution was studied using UV-vis spectroscopy, high-performance liquid chroma-
tography (HPLC), and time-resolved spectroscopic techniques, such as laser-flash photolysis and
singlet oxygen phosphorescence detection. In both N2- and air-saturated solutions, the main product
formed was all-trans-bixin. The observed isomerization rate constants, kobs, decreased in the presence
of air or with increase in the bixin concentration, suggesting the participation of the excited triplet
state of bixin, 3Bix*, as precursor of the cisf trans process. On the other hand, bixin solutions in the
absence of sensitizer and/or light did not degrade, indicating that the ground state of bixin is stable
to thermal isomerization at room temperature. Time-resolved spectroscopic experiments confirmed
the formation of the excited triplet state of bixin and its deactivation by ground state bixin and molecular
oxygen quenching processes. The primary isomerization products only degraded in the presence of
air and under prolonged illumination conditions, probably due to the formation of oxidation products
by reaction with singlet molecular oxygen. An energy-transfer mechanism was used to explain the
observed results for the bixin transformations, and the consequences for food color are discussed.
KEYWORDS: Bixin; photoisomerization; photosensitization; singlet oxygen and triplet states
INTRODUCTION
Scheme 1. Chemical Structure of the Natural cis-Carotenoid Bixin
Bixin (methyl hydrogen 9′-cis-6, 6′-diapocarotene-6, 6′-dioate,
Scheme 1) is the main carotenoid existing in the seed coat of
the annatto tree (Bixa orellana L.), which is found in tropical
regions of South and Central America, Africa, and Asia (1, 2).
Annatto extracts are widely used in the food industry as natural
coloring agents in cheese, ice cream, yoghurt, sausages,
margarine, snacks, and dressings, and compared to natural
colors, are relatively inexpensive, thus increasing their com-
mercial relevance (1, 2).
The characterization of the coloring components of the
annatto extracts (3-5), annatto seeds (6- 9), and their degrada-
tion products (10, 11) has received much attention. In particular,
the stability of bixin was evaluated under the effect of several
factors, such as light, oxygen concentration, temperature, and
solvent (12, 13). Bixin showed great stability in the dark, in
the presence or absence of oxygen, but under direct illumination
was degraded (12). The thermo- and photostability of the 9′-
cis double bond in methyl bixin was reported as being so great
that the molecule underwent a second transfcis isomerization
giving considerable amounts of the 9, 9′-di-cis isomer, but in
the presence of iodine and light, 9′-cisftrans isomerization
occurred (14). Therefore, it seems that the fade of bixin could
be strongly dependent on the experimental conditions (e.g.,
direct illumination, heating, photosensitization, etc).
Most studies on bixin stability focused on thermal degradation
(10, 11), direct illumination (12), and the water activity solvent
effect (13). In the thermal degradation of bixin, the main
products formed were all-trans-bixin, a C17 carotenoid (trans-
monomethyl ester of 4,8-dimethyl tetradecahexaenedioic acid)
and volatile compounds, especially xylene (5, 10).
However, to our knowledge, there is little or no information
about the photosensitized degradation of bixin either in model
systems or in foods. Moreover, light can dramatically reduce
food stability due to the promotion of autoxidation and
photosensitized oxidation, the latter being more common,
because many natural sensitizers are commonly found in foods.
* To whom correspondence should be addressed. Fax: 54-385-4509585.
E-mail: cborsa@unse.edu.ar.
† Universidad Nacional de Santiago del Estero.
‡ UNICAMP.
10.1021/jf0349026 CCC: $27.50 © 2004 American Chemical Society
Published on Web 12/20/2003

368 J. Agric. Food Chem., Vol. 52, No. 2, 2004
Montenegro et al.
In photosensitized reactions, photons are absorbed by one
type of molecule (known as a sensitizer, S), resulting in a long-
lived energy-rich state(s) (typically triplet states) of the sensi-
tizer, eq 1, which can undergo reactions producing chemical
alteration of another molecule (substrate, P) in the system, eq
2 (15).
kisc
S
9
hν8 1S* 8 3S*
(1)
(2)
kSq,P
3S* + P
8 S + P; and/or products
In its primary reactions, the excited sensitizer molecule can
also react directly with some other molecule (frequently
molecular oxygen O2(3Σg-)) in the reaction mixture, giving
products (usually singlet molecular oxygen, O2(1∆g)) that in turn
react with the substrate, eqs 3 and 4 (15, 16).
Figure 1. HPLC chromatograms obtained with the C −Vydac 218TP54
18
column and monitored at 450 nm (left) and UV−vis spectra (right) of 8
µM bixin in air-saturated ACN/MeOH (1:1) solutions at 20 °C in the
presence of 10 µM Rose bengal after different times of illumination of
the sensitizer: (a) 0 min, (b) 2 min, (c) 10 min, (d) 150 min.
kSq,Σ
3S* + O2(3Σ g
)
8 S + O2(1∆g)
(3)
-
k∆q,P
O2(1∆g) + P
8 O2(3Σ g-) + P and/or oxidation products
Methods. Stationary UV-vis spectra were obtained either with a
Hewlett-Packard 8453 diode array spectrophotometer or a Beckman
DU-640 spectrophotometer. HPLC analysis was carried out using a
Waters HPLC system equipped with a photodiode array detector
(Waters, model 996). The equipment also included an on-line degasser,
a Rheodyne injection valve with a 20-µL loop and an external oven.
Data acquisition and processing were performed using the Millenium
Waters software. The carotenoids were separated on a C18 Spherisorb
ODS-2 column, 150 × 4.6 mm i.d. (3-µm particle size), with
acetonitrile/2% v/v acetic acid/dichloromethane (63:35:2) as mobile
phase at a flow rate of 1 mL/min, with the column temperature being
set at 25 °C. The chromatograms were processed at max plot (λmax),
and the spectra were recorded between 250 and 600 nm. The second
HPLC system used was a Konik 500 A chromatograph coupled to a
UV-vis detector Konik UV-vis 200, monitored at 450 nm. In this
case, the column was a C18 Vydac 218TP54 (5-µm, 4.6 × 250-mm),
and the mobile phase and flow conditions were identical to those
described above. All samples and solvents were filtered before use,
using 0.22 µm and 0.45 µm membranes, respectively.
(4)
On the other hand, it is well-known that carotenoids are
efficient quenchers of both excited triplet states and O2(1∆g),
principally by an energy-transfer mechanism producing the
ground state of the precursor species and the carotenoid triplet
state, as occurs in the photosynthetic organisms, eqs 5 and 6
(17).
kSq,Car
3S* + Car
8 S + 3Car*
(5)
(6)
k∆q,Car
O2(1∆g) + Car
8 O2(3Σ-g ) + 3Car*
Once formed, the carotenoid triplet state returns harmlessly
to the ground state with the liberation of heat. In general, the
reactive quenching reaction between O2(1∆g) and the carotenoids
yielding oxidation products is several orders of magnitude
smaller than the physical quenching process, eq 6. The bleaching
of all-trans-carotenoids by O2(1∆g)-mediated oxidation is com-
monly observed after prolonged photosensitization periods (18).
However, for cis carotenoids, the photosensitized process
produces the trans isomer (19, 20). Therefore, for bixin
contained in foods and beverages, a similar photosensitized
behavior can be expected, because in foods, several natural or
artificial pigments (e.g., riboflavin, chlorophyll, and organic
dyes) act as sensitizers (21).
The aim of this work was to present a kinetic study of the
photosensitized transformation of bixin in an organic solvent
mixture, using UV-vis spectroscopy, high-performance liquid
chromatography (HPLC) and time-resolved spectroscopic tech-
niques, such as laser-flash photolysis and O2(1∆g)-phosphores-
cence detection.
The photosensitization experiments were performed in acetonitrile/
methanol (1:1) solutions of bixin containing ca.10 µM RB or MB,
illuminated with a 150 W filament lamp coupled to an orange cutoff
filter to exclusively excite the band of the sensitizer above 520 nm.
The reaction was monitored by HPLC at 5, 15, and 20 °C under N2
(99.99% purity) and in air-saturated solutions.
Laser-flash photolysis (LFP) experiments and the transient spectra
were obtained with a homemade system previously described by
Borsarelli et al. (22), using a Q-switched Nd/YAG laser (Spectron
SL400) operating at the frequency-doubled output (532 nm, 20 ns
halfwidth) as excitation source. The transient spectra were performed
in both N2- and air-saturated solutions using fresh solution after a few
laser shots.
Singlet molecular oxygen O2(1∆g) sensitization in air-saturated
solutions was measured using time-resolved detection of the phospho-
rescence of O2(1∆g) at 1270 nm, using the same laser as excitation
source, as described previously (18).
In all cases, the experiments were performed in duplicate.
RESULTS AND DISCUSSION
EXPERIMENTAL PROCEDURES
Figure 1 shows the HPLC chromatograms and the UV-vis
spectra of 8 µM bixin in air-saturated solutions at 20 °C,
observed after different times of illumination of the sensitizer
RB. It can be observed that whereas the bixin peak was almost
completely consumed after 10 min of photosensitization, slight
changes in the UV-vis spectra were produced (Figure 1). In
addition, the consumption of bixin was accompanied by the
simultaneous formation of new peaks (labeled as products 1, 2,
Materials. Bixin (99% purity) was kindly donated by Dr. Werner
Simon from Roche Vitamins (Basel, Switzerland). Rose bengal (RB)
as its sodium salt and methylene blue (MB) were from Sigma Chemical
Co. (St. Louis, MO) and used without further purification. 9, 10-
Dimethylanthracene (DMA) from Aldrich (Milwaukee, WI) was
recrystallized from ethanol before use. All the organic solvents used
were HPLC grade from Mallinckrodt, and the distilled water was
purified by the Milli-Q Plus system (Millipore).

Photosensitized Isomerization of Bixin
J. Agric. Food Chem., Vol. 52, No. 2, 2004 369
Table 1. Retention Times tR, Absorption Maximum Wavelengths λMax
,
and Band Ratios %III/II and %AB/A Obtained from HPLC −
II
Photodiode Array Analysis (PDA) during the Photosensitized
Isomerization of Bixin in Aerated and Non-aerated ACN/MeOH (1:1)
Solutions
b
c
d
tR
(min)
tR
(min)
λmax
(nm)
peak labela
1
% III/II % AB/A
II
8.0 ± 0.5 11.8 ± 0.2 290, 352, 420 (sh),
10
28
21
39
40
12
10
18
5
455, 482
2
9.3 ± 0.3 12.3 ± 0.2 285, 420 (sh), 457,
483
3
11.0 ± 0.9 13.1 ± 0.1 285, 354, 422 (sh),
459, 483
Bix
4
12.7 ± 1.0 13.3 ± 0.1 355, 425 (sh), 461,
489
19.2 ± 1.4 13.7 ± 0.2 285, 430 (sh), 466,
495
0
Figure 3. Kinetic curves obtained from the HPLC peak analysis of 8 µM
bixin in air-saturated ACN/MeOH (1:1) solutions in the presence of 10
µM Rose bengal: (b) Bixin, (O) all-trans-bixin, and (9) isomer 3. The
solid and dotted lines represent the exponential fitting of the experimental
data (see text for details).
a Labeled according to the chromatogram from Figure 1. b C −Vydac 218TP54
18
column (average value from 10 chromatographic runs). c C −Spherisorb ODS-2
18
column (average value from 60 chromatographic runs). d (sh) ) shoulder.
Table 2. Observed Rate Constant kobs and Activation Parameters
∆H q and ∆S q for the Photosensitized Isomerization of Bixin in
ACN/MeOH (1:1) Solutions
temp
(deg C)
[Bix]
(µM)
k
obs × 104
(s-1
∆H q
(kcal/mol)
∆S q
(cal/mol ‚ K)
system
)
air saturated
5
15
20
20
76
76
76
8
2.3
3.0
3.7
5 ± 1
6 ± 2
−59 ± 11
−50 ± 30
42
N saturated
2
5
15
20
76
76
76
21
26
40
As compared with data from the literature, the observed
spectral features strongly suggested that products 1-4 were
geometrical isomers of bixin. For instance, the slight blue shift
of the λmax, together with the increase in %AB/AII and decrease
in %III/II for the minor products 1-3 indicated the formation
of different cis-isomers of bixin (23). Although the UV-vis
absorption spectra parameters reported in Table 1 are very useful
for structure characterization, they are not conclusive evidence
for the assignment of the position of the cis double bonds in
products 1-3. However, this is not the case for the major
product 4, because the bathochromic shift, together with the
total disappearance of the cis-band absorption and increased
spectral fine structure, indicated the quantitative cisftrans
conversion at the 9′-cis double bond of bixin, producing the
all-trans form (14, 23). In addition, the λmax values of product
4 were similar to those reported for all-trans-norbixin in a similar
solvent mixture (3) supporting, in this case, the assignment of
product 4 to all-trans-bixin, with the present UV-vis data. By
comparison of the HPLC peak areas, it can be concluded that
bixin is almost completely transformed to all-trans-bixin
(> 90%) and to a much less extension to the cis-isomers 1-3.
Figure 3 shows the kinetic curves, obtained from the HPLC
analysis, of bixin and products 3 and 4 (all-trans-bixin) as a
function of the photosensitization time. In all cases, the
disappearance of bixin followed an irreversible 1st-order kinetic
law, with a similar observed rate constant kobs to those for the
formation of the product isomers 3 and 4 (Table 2). However,
the formation of both bixin isomers was followed by slower
1st-order consumption only in air-saturated solutions. The solid
lines in Figure 3 represent the 1st-order fits for the decay of
bixin, and the 1st-order growth and decay of the isomers 3 and
Figure 2. Normalized UV−vis spectra obtained from HPLC−PDA analysis
of the photosensitized degradation products of bixin in ACN/MeOH (1:1)
solutions. The dotted spectra correspond to the bixin spectrum.
3, and 4). The same HPLC pattern was observed in photosen-
sitized experiments in N2-saturated solutions and with higher
bixin concentrations. In all cases, peak 4 was the main product
formed, with a greater retention time (tR) than bixin, whereas
the minor products 1-3 presented shorter tR values than bixin,
Table 1.
In all experiments, the sensitizer RB was not degraded as
confirmed by HPLC (Figure 1) and/or UV-vis spectra (data
not shown). Moreover, no consumption of bixin was observed
under dark conditions in the absence or presence of the
sensitizer, indicating that bixin did not form products from the
thermal degradation of the carotenoid and/or from direct reaction
with the sensitizer.
Figure
2 shows the normalized UV-vis absorption
spectra of products 1-4 obtained from the photodiode array
detection (PDA) of the HPLC peaks, and compared to that of
bixin (dotted line spectrum). Table 1 presents the %III/II ratio,
calculated as the values of the height ratio between the longest
and middle wavelength absorption bands (designated as III and
II, respectively) by taking as the baseline the valley between
both bands and of the relative intensity of the cis band (%AB/
AII), calculated as the ratio between the height of the cis band
(ca. 360 nm for bixin) and the band II. These are known to be
useful parameters for the spectral characterization of caro-
tenoids (23).

370 J. Agric. Food Chem., Vol. 52, No. 2, 2004
Montenegro et al.
Table 3. Rate Constant Values for the Deactivation Processes Indicated in Scheme 2
process
rate constant
kSd,o ) (1.7 × 104 s-1)a; (2.9 × 104 s-1 b
kSq,Bix ) (8 × 109 M-1s-1)a; (7× 109 M-1s-1 b
3S* f S
)
3S* + Bix f S + 3Bix*
3S* + Σ f S + ∆
∆ f Σ
∆ + Bix f Σ + 3Bix*
3Bix* f Bix
3Bix* + Bix f 2Bix
3Bix* + Σ f Bix + Σ
∆ + Isomer products f Oxidation products
)
kqS,Σ ) (1 × 109 M-1s-1)a; (2× 109 M-1s-1 b
)
k∆d,o ) 6.7 × 104 s-1
k∆q,Bix ) 1.3 × 1010 M-1s-1
kBd,iox ) 4.2 × 104 s-1
Bix
q,Bix
k
) 1.5 × 109 M-1s-1
kBq,iΣx ) 7.0 × 108 M-1s-1
k∆r ) 1.0 × 106 M-1s-1
a S ) RB. b S ) MB;. Σ ) O (3Σg-); ∆ ) O (1∆g).
2
2
(MB) triplet state 3S* was kinetically associated with the
formation of the 3Bix*, via a bimolecular energy-transfer
quenching pathway (17, 19), eq 7
kSq,Bix
3S* + Bix
8 S + 3Bix*
(7)
where kSq,Bix is the energy-transfer quenching rate constant of
3S* by bixin. In addition, the decay portion of the transient curve
3
at 520 nm represents the actual decay of Bix*, which is also
dependent on the bixin concentration. Because no laser intensity
effect was observed, the triplet-triplet annihilation process was
ruled out. On the basis of this fact, the shortening of the decay
3
time of Bix* with increases in the carotenoid concentration
was due to a self-quenching process (28) with a rate constant
Bix
Figure 4. Transient absorption spectra of 10 µM methylene blue (MB) in
k
, eq 8.
q,Bix
the presence of 9 µM bixin in N saturated ACN/MeOH (1:1) solutions
2
Bix
q,Bix
k
observed at different times after the 532 nm laser pulse: (b) 1 µs, (O)
4 µs, (9) 8 µs, and (0) 40 µs. Insets: (a) Decay traces at 420 and 520
nm, with the exponential and biexponential fittings (solid lines); (b)
Dependence of the observed decay constant kXd for 3MB* (b) and 3Bix*
(O), with the bixin concentration, eq 9.
3Bix* + Bix
8 2 Bix
(8)
The dependence of the unimolecular decay rate constants kdX
3
3
of the species X (with X ) S* or Bix*) showed a linear
relationship with the concentration of the carotenoid, eq 9, as
shown in inset b of Figure 4.
4, fitted with a biexponential function derived from the sum of
two consecutive 1st-order processes (dotted lines). Table 2 also
shows that the kobs decreased with increase in bixin concentration
and in the presence of air.
The temperature effect on kobs was also evaluated according
to the activated complex theory (24), allowing for the calculation
of the activation parameters ∆Hq and ∆Sq, Table 2. The
activation parameters were very similar in both N2- and air-
saturated solutions, suggesting that the same degradation mech-
anism operated under both conditions.
kXd ) kXd,o + kXq,Bix[Bix]
(9)
where kXd,o represents the unimolecular decay rate constant of
the species X in the absence of bixin. The quenching rate
S
Bix
3
3
constants of S* and Bix* by bixin, kq,Bix and kq,Bix, respec-
tively, are reported in Table 3. These values are equivalent to
those of quenching processes almost controlled by diffusion.
Bix
3
-1
The natural lifetime of Bix*, (kd,o
)
) 24 µs was calculated
from the intercept value of the plot for the self-quenching
process, with the value being in good agreement with recently
reported data for several carotenoids with different numbers of
conjugated double bonds (28).
To explore the participation of transient excited species in
the photosensitized isomerization of bixin, laser-flash photolysis
experiments were performed, using laser pulsed excitation at
532 nm. Figure 4 shows the transient absorption spectra
observed after laser excitation of the sensitizer MB in N2-
saturated solutions in the presence of 9 µM bixin. The transient
3
On the other hand, in air-saturated solutions, the S* was
efficiently quenched by ground-state molecular oxygen
O2(3Σ-g ), eq 3, with kqS,Σ ≈ 2 × 109 M-1s-1 (Table 3). At 20
°C, the concentration of dissolved molecular oxygen in air-
saturated methanol or acetonitrile was ca. 2 mM (25). Therefore,
under these conditions, 3S* was efficiently quenched by
O2(3Σ-g ) producing O2(1∆g) and the ground state of the sensi-
tizer, eq 3.
3
absorption due to the excited triplet state of the sensitizer, -
MB*, with a maximum at 420 nm (25) decayed with the
simultaneous build-up of an absorption band at ca. 520 nm, as
confirmed by the kinetic curves in the inset a of Figure 4. After
3
40 µs of the laser pulse, almost 95% of the MB* state was
depleted, and the residual transient spectrum was similar to those
observed for the triplet state of carotenoids such as â-carotene
or zeaxanthin (26, 27). The 1st-order kinetic analysis (solid lines
in the inset a) of the decay at 420 nm and the growth at 520
nm, afforded similar rate constant values as a function of the
bixin concentration, confirming that the decay of the sensitizer
The formation of O2(1∆g) was followed by monitoring the
phosphorescence decay of O2(1∆g) at 1270 nm (Figure 5). The
tail part of the O2(1∆g) decays was fitted with an exponential
function extrapolating to the same initial intensity value, due
to the signal distortion by the residual luminescence of the

Photosensitized Isomerization of Bixin
J. Agric. Food Chem., Vol. 52, No. 2, 2004 371
Scheme 2. Reaction Mechanism Proposed for the Photosensitized Isomerization of Bixin in Both Nonaerated and Aerated Solutions
3
O2(1∆g); and Bix and Bix* are the ground and triplet excited
states of bixin. According to this mechanism, the excited triplet
3
state of Bix* is produced under two different energy-transfer
pathways: (i) bixin quenching of 3S* in N2-saturated solutions,
and (ii) by sequential quenching of O2(1∆g) by bixin, after
-
3
deactivation of S* by O2(3Σg ) in air-saturated conditions.
The driving force for these processes is the energy cascade
involving S*, O2(1∆g) and Bix*. The triplet energies ET for
the sensitizers RB and MB were 39 and 33 kcal/mol, respec-
tively (25). In sequence, the O2(1∆g) lays at 22.5 kcal/mol above
the ground-state O2(3Σg-) (15), but the precise energy level of
3Bix* is still unknown. However, the ET of 3Bix* can be
expected to be no higher than that for â-carotene, reported as
20 kcal/mol (32, 33), by considering the similar k∆q values
observed for both carotenoids in the same solvent system (18).
Thus, the 3Bix* was the final long-lived excited state produced
in the degradation energy cascade.
Therefore, the unimolecular cisftrans isomerization reaction
of the 9′-double bond was occurring starting with 3Bix*,
explaining the 1st-order kinetic behavior of the consumption
of bixin and the formation of the isomerization products.
According to this mechanism, the same activation parameters
in the absence and presence of air can be expected, as was
experimentally found (Table 2).
3
3
Figure 5. Effect of the bixin concentration on the phosphorescence decay
of singlet molecular oxygen O (1∆g). Inset: Dependence of the observed
2
decay constant k∆d with the bixin concentration, eq 10.
sensitizer and the rise time of the signal response of the
germanium detector (<5 µs). In the absence of bixin, O2(1∆g)
decayed with a lifetime of 15 µs, in good agreement with the
lifetime expected for binary mixtures of similar polar organic
solvents (16, 29). The addition of bixin reduced the observed
decay rate constant of O2(1∆g), kd∆, following eq 10.
On the other hand, the smaller values for kobs observed by
increasing the bixin or molecular oxygen concentrations, are
k∆d ) k∆d,o + k∆q,Bix[Bix]
(10)
3
explained by the competitive deactivation of the Bix* by the
bimolecular self-quenching of bixin, eq 8, or by an enhanced
A quenching rate constant of k∆q,Bix ) 1.3 × 1010 M-1s-1
was obtained, Table 3, in good agreement with data obtained
for bixin in other solvent systems (18, 30). It is well-known
that the quenching process is more physical than chemical for
carotenoids with g9 conjugated double bonds, because the
chemical quenching rate constant k∆r is several orders of
magnitude smaller than the physical one, k∆q (16, 18, 28, 29).
Moreover, the HPLC analysis between 250 and 600 nm did
not show the presence of extra peaks other than the reported
products 1-4 during the consumption of bixin, confirming that
the O2(1∆g) quenching by bixin was totally physical, eq 11.
intersystem crossing quenching by O2(3Σg-), eq 12.
Bix
q,Σ
k
3Bix* + O2(3Σ -g )
8 Bix + O2(3Σ g
)
(12)
-
Bix
q,Σ
The rate constant quenching k was expected to be much
smaller than the solvent diffusion limit (ca. 2 × 1010 M-1s-1),
because energy-transfer quenching to yield O2(1∆g) is ender-
gonic (see above) (34). According to the quenching rate
constants reported in Table 3, the quenching efficiencies of 3S*
or O2(1∆g) by 76 µM bixin were almost in unity. Because both
energy-transfer pathways yield 3Bix* as common isomerization
precursor, the amount of carotenoid triplet state formed per
absorbed photon by the sensitizer was almost the same either
∆
O2(1∆g) + Bix kq,Bix8 O2(3Σ g-) + 3Bix*
(11)
3
The formation of Bix* during the reaction shown in eq 11
was also confirmed by the detection of the transient absorption
in N2- or air-saturated solutions. Thus, an estimation of the
Bix
q,Σ
k
can be obtained by using the kobs values for the isomer-
3
spectra of Bix* in air-saturated solutions (data not shown).
ization reaction in the absence and presence of air according to
eq 13, because the quenching processes by O2(3Σg-) competes
with the unimolecular isomerization pathway.
Therefore, all the above results support the participation of
3Bix* in the isomerization reaction of bixin, according to the
energy-transfer-based mechanism (19, 26, 31) proposed in
1
3
Scheme 2. In this mechanism S, S*, and S* symbolize the
ground, singlet, and triplet excited states of the sensitizer
molecules (RB or MB), respectively; Σ and ∆ are O2(3Σg-) and
(kNobs/kAobirs) - 1
2
Bix
q,Σ
k
)
(13)
τd,Bix[O2]

372 J. Agric. Food Chem., Vol. 52, No. 2, 2004
Montenegro et al.
In eq 13, τd,Bix () 1/kdBix) ) 6.5 µs was obtained as the actual
(4) Mercadante, A. Z. Composition of Carotenoids from Annatto.
In Chemistry and Physiology of Selected Food Colorants; Ames,
J. M., Hofmann, T. F., Eds.; ACS Symposium Series 775:
Washington, DC, 2001. pp 92-101.
(5) McKeown, G. G. Composition of oil-soluble annatto food colors.
III. Structure of the yellow pigment formed by the thermal
degradation of bixin. J. AOAC. 1965, 48, 835-837.
(6) Mercadante, A. Z.; Steck, A.; Pfander, H. Isolation and structure
elucidation of minor carotenoids from annatto (Bixa orellana
L.) seeds. Phytochemistry 1997, 46, 1379-1383.
3
lifetime of Bix*, considering the self-quenching of bixin at
[Bix] ) 76 µM, eq 9. Using the data reported in Table 3, the
Bix
calculation in eq 13 yielded k ) (7 ( 2) × 108 M-1s-1, this
q,Σ
magnitude being in agreement with the previous assumption
(34).
Finally, as shown in Figure 3, both all-trans-bixin and isomer
3 only degraded in aerated solutions, indicating a possibly slower
reaction with O2(1∆g), eq 14.
(7) Mercadante, A. Z.; Steck, A.; Pfander, H. Three minor caro-
tenoids from annatto (Bixa orellana) seeds. Phytochemistry 1999,
52, 135-139.
(8) Jondiko, I. J. O.; Pattenden, G. Terpenoids and an apocarotenoid
from seeds of Bixa orellana. Phytochemistry 1989, 28, 3159-
3162.
∆
kr
O2(1∆g) + Bixin isomers
98 Oxidation products (14)
This reaction is associated with the UV-vis spectral changes
observed after prolonged photosensitization periods (e.g., spec-
trum d in Figure 1). A chemical quenching rate constant kq∆,r
)
1 × 106 M-1s-1 was calculated by comparing, under identical
experimental conditions, the observed 1st-order rate constants
(2.8 × 10-4 s-1) for the degradation of the bixin isomers, with
that observed for the reaction of the reference compound 9,-
10-dimethylanthracene (DMA) with O2(1∆g), in which the
bimolecular reaction rate constant k∆r,DMA ) 5 × 107 M-1s-1
(16). Current research activities are being focused on the
isolation and identification of the O2(1∆g)-mediated final oxida-
tion products of the bixin isomers.
In summary, this study showed that in photosensitized
processes, bixin isomerizes mainly to all-trans-bixin, indepen-
dent of the presence of oxygen, via an energy-transfer based
mechanism, giving the triplet state of bixin, 3Bix*, as the
isomerization precursor. Therefore, the isomerization rate is
dependent on several factors that compete for the deactivation
of 3Bix*, such as self-quenching and O2(3Σg-) quenching
processes. In the presence of O2(1∆g) the isomerization products
are slowly degraded, resulting in small color changes in the
solution.
These results are important, considering that despite the
solvent used in the model-system, this study showed that
carotenoids with cis conformation are able to retain their color
during photosentitized oxidation reactions, because the main
product formed is the correspondent trans-carotenoid, which
shows a bathochromic shift of 2-5 nm. This is not the case for
trans-carotenoids, such as all-trans-â-carotene, which formed
oxidation products with shorter conjugated bond chain (35),
therefore presenting a softer color. Therefore, under the same
photosensitized conditions, the color of the bixin system can
be maintained, whereas that of â-carotene would fade.
(9) Mercadante, A. Z.; Steck, A.; Pfander, H. Isolation and identi-
fication of new apocarotenoids from annatto (Bixa orellana L.)
seeds. J. Agric. Food Chem. 1997, 45, 1050-1054.
(10) Scotter, M. J. Characterisation of the coloured thermal degrada-
tion products of bixin from annatto and a revised mechanism
for their formation. Food. Chem. 1995, 53, 177-185.
(11) Scotter, M. J.; Wilson, L. A.; Appleton, G. P.; Castle, L. Analysis
of Annatto (Bixa orellana) food coloring formulations. 2.
Determination of aromatic hydrocarbon thermal degradation
products by gas chromatography. J. Agric. Food Chem. 2000,
48, 484-488.
(12) Najar, S. V.; Bobbio, F. O.; Bobbio, P. A. Effect of light, air,
anti-oxidants and pro-oxidants on annatto extracts (Bixa orel-
lana). Food Chem. 1988, 29, 283-289.
(13) Gloria, M. B. A.; Vale, S. R.; Bobbio, P. A. Effect of water
activity on the stability of bixin in an annatto extract-microc-
rystalline cellulose model system. Food Chem. 1995, 52, 389-
391.
(14) Zechmeister, L.; Escue, R. B. A stereochemical study of
methylbixin. J. Am. Chem. Soc. 1944, 66, 322-330.
(15) Spikes, J. D. Photosensitization. In The Science of Photobiology,
2nd ed.; Smith, K. C., Ed.; Plenum Press: New York, 1989. pp
79-110.
(16) Wilkinson, F.; Brummer, J. G. Rate constants for the decay and
reactions of the lowest electronically excited singlet state of
molecular oxygen in solution. J. Phys. Chem. Ref. Data. 1981,
10, 809-999.
(17) Gust, D.; Moore, T. A.; Moore, A. L.; Jori, G.; Reddi, E. The
photochemistry of carotenoids. Some photosynthetic and pho-
tomedical aspects. Ann. NY. Acad. Sci. 1993, 691, 32-46.
(18) Montenegro, M. A.; Nazareno, M. A.; Durantini, E. N.; Bor-
sarelli, C. D. Singlet molecular oxygen quenching ability of
carotenoids in a reverse-micelle membrane mimetic system.
Photochem. Photobiol. 2002, 75, 353-361.
(19) Foote, C. S.; Chang, Y. C.; Denny, R. W. Chemistry of Singlet
Oxygen. XI. Cis-trans isomerization of carotenoids by singlet
oxygen and a probable quenching mechanism. J. Am. Chem. Soc.
1970, 92, 5218-5219.
(20) Manitto, P.; Speranza, G.; Monti, D.; Gramatica, P. Singlet
oxygen reactions in aqueous solution. Physical and chemical
quenching rate constants of crocin and related carotenoids.
Tetrahedron Lett. 1987, 28, 4221-4224.
ABBREVIATIONS
RB, Rose Bengal; MB, Methylene Blue; DMA, 9,10-
dimethylanthracene; O2(3Σg-), ground-state molecular oxygen;
O2(1∆g), singlet molecular oxygen; Bix*, bixin triplet excited
3
state; HLPC, high performance liquid chromatography; PDA,
photodiode array detection; LFP, laser-flash photolysis; TRPD,
time-resolved phosphorescence detection.
(21) Redmond, R. W.; Gamlin, J. N. A compilation of singlet oxygen
yields from biologically relevant molecules. Photochem. Pho-
tobiol. 1999, 70, 391-475.
LITERATURE CITED
(22) Borsarelli, C. D.; Bertolotti, S. G.; Previtali, C. M. Thermody-
namic changes in the photoinduced proton-transfer reaction of
the triplet state of safranine-T. Photochem. Photobiol. Sci. 2002,
1, 574-580.
(23) Britton, G. UV/visible spectroscopy. In Carotenoids Vol. 1B:
Spectroscopy; Britton, G., Liaaen-Jensens, S., Pfander, H., Eds.;
Birkha¨user Verlag: Basel, Switzerland, 1995, 13-62.
(24) Wilkinson, F. Theory of elementary reactions. In Chemical
Kinetics and Reaction Mechanisms; Van Nostrand Reinhold
Company: New York, 1980; pp 95-136.
(1) Mercadante, A. Z.; Pfander, H. Carotenoids from annatto: a
review. Recent Res. DeV. Agric. Food Chem. 1998, 2, 79-91.
(2) Preston, H. D.; Rickard, M. D. Extraction and chemistry of
annatto. Food Chem. 1980, 5, 47-56.
(3) Scotter, M. J.; Wilson, L. A.; Appleton, G. P.; Castle, L. Analysis
of Annatto (Bixa orellana) food coloring formulations. 1.
Determination of coloring components and colored thermal
degradation products by high-performance liquid chromatography
with photodiode array detection. J. Agric. Food Chem. 1998,
46, 1031-1038.

Photosensitized Isomerization of Bixin
J. Agric. Food Chem., Vol. 52, No. 2, 2004 373
(25) Murov, S. L.; Carmichael, I.; Hug. G. L. In Handbook of
Photochemistry, 2nd ed.; Marcel Dekker: New York, 1993; pp
289-291.
(26) Rodgers, M. A. J.; Bates, A. L. Kinetic and spectroscopic features
of some carotenoid triplet states: sensitization by singlet oxygen.
Photochem. Photobiol. 1980, 31, 533-537.
(27) Nielsen, B. R.; Mortensen, A.; Jørgensen, K.; Skibsted, L. H.
Singlet versus triplet reactivity in photodegradation of C40
carotenoids. J. Agric. Food Chem. 1996, 44, 2106-2113.
(28) Burke, M. B.; Land, E. J.; McGarvey, D. J.; Truscott, T. G.
Carotenoid triplet state lifetimes. J. Photochem. Photobiol., B
2000, 59, 132-138.
(32) Haley, J. L.; Fitch, A. N.; Goyal, R.; Lambert, C.; Truscott, T.
G. The S1 and T1 energy levels of all-trans-â-carotene. J. Chem.
Soc. Chem. Commun. 1992, 17, 1175-1176.
(33) Lambert, C.; Redmond, R. W. Triplet energy level of â-carotene.
Chem. Phys. Lett. 1994, 228, 495-498.
(34) Wilkinson, F. Quenching of electronically excited states by
molecular oxygen in fluid solution. Pure Appl. Chem. 1997, 69,
851-856.
(35) Stratton, S. P.; Schaefer, W. H.; Liebler, D. C. Isolation and
identification of singlet oxygen oxidation products of â-carotene.
Chem. Res. Toxicol. 1993, 6, 542-547.
(29) Bilski, P.; Holt, R. N.; Chignell, C. F. Properties of singlet
molecular oxygen O2(1∆g) in binary solvent mixtures of different
polarity and proticity. J. Photochem. Photobiol., A 1997, 109,
243-249.
Received for review August 7, 2003. Revised manuscript received
November 12, 2003. Accepted November 16, 2003. Financial support
provided by the Consejo Nacional de Investigaciones Cient´ıficas y
Te´cnicas de la Argentina (CONICET), Secretar´ıa de Ciencia y Te´cnica
de la Universidad Nacional de Santiago del Estero, Fundacio´n Antor-
chas (Argentina), and the Brazilian Funding Agencies (FAPESP,
CAPES, and CNPq).
(30) Speranza, G.; Manitto, P.; Monti, D. Interaction between singlet
oxygen and biologically active compounds in aqueous solution
1
III. Physical and chemical O2-quenching rate constants of 6,
6′-diapocarotenoids. J. Photochem. Photobiol., B 1990, 8, 51-
61.
(31) Chen, B. H.; Liu, M. H. Relationship between chlorophyll a and
â-carotene in a lipid-containing model system during illumina-
tion. Food Chem. 1998, 63, 207-213.
JF0349026