You are on page 1of 7

Process Biochemistry 36 (2001) 933– 939

www.elsevier.com/locate/procbio

Cocoa butter equivalent through enzymic interesterification of


palm oil midfraction
Daniel Undurraga a, Andrés Markovits a,*, Sonia Erazo b
a
Escuela de Ingenierı́a Bioquı́mica, Uni6ersidad Católica de Valparaı́so, A6. Brasil 2147, Valparaı́so, Chile
b
Instituto de Quı́mica, Uni6ersidad Católica de Valparaı́so, A6. Brasil, Valparaı́so, Chile

Received 6 July 2000; received in revised form 31 August 2000; accepted 29 October 2000

Abstract

The production of a cocoa butter equivalent (CBE) through enzymic interesterification of palm oil midfraction (POMF) with
stearic acid in a solvent free system using Novo lipase Lipozyme™ as a catalyst was analyzed. A two level factorial design was
used to study the effect of the initial ratio of stearic acid– POMF, initial humidity of the enzyme preparation and the
enzyme–substrate ratio on the yield, mass productivity and specific productivity. Studies were carried out both in batch and in
a continuous packed bed reactor. The highest specific productivity obtained in shake flask was 0.0393 g/Batch Interesterification
Unit (BIU) h at a stearic acid–POMF ratio of 1.6 and enzyme– substrate ratio of 23 BIU/g. In the continuous packed bed reactor
the highest mass productivity observed was 1.54 g/g·h, using an enzymic load of 73 BIU/g. Unreacted fatty acids were separated
from the intereresterified products by short path distillation at 0.2 mbar and 140°C obtaining a product practically without free
fatty acids. Thermograms of the products obtained by scanning differential calorimetry were similar to cocoa butter (CB), but
exhibited several distinct peaks, due presumably to the presence of diglycerides and trisaturated triglycerides. © 2001 Elsevier
Science Ltd. All rights reserved.

Keywords: Interesterification; Lipases; Cocoa butter equivalent; Palm oil midfraction; Solvent free system; Diferential scanning calorimetry

1. Introduction composition, but its triglycerides derive from a source


other than cocoa beans [6].
The modification of fats and oils through interester- CBE contains approximately 40% 1-palmito, 2-olein,
ification reactions catalyzed by lipases has been studied 3-sterin glycerol (POS), 27% of 1,3 distearin-
since the 1980’s. This is a method with significant monooleate glycerol (SOS) and 21% of 1,3 dipalmitin-
technical advantages over the traditional methods of 2-monooleato glycerol (POP) and minor amounts of
chemical interesterification. Among the advantages are other triglycerides. A suitable raw material for the
lower energy consumption, absence of isomerization by production of CBE in terms of cost, availability and
products and better control of products. composition is palm oil midfraction (POMF), which is
The transformation of low-cost fats and oil triglyce-
obtained by double fractionation of palm oil. POMF
rides through enzymic interesterification for the elabo-
contains approximately 73% POP, 13% POS, 2% SOS
ration of cocoa butter (CB) substitutes or equivalents
and 12% of other triglycerides.
has been previously reported [1 – 5]. A substitute is a
type of fat, which does not have a similar chemical The purpose of interesterification is to incorporate
composition, but presents a fusion profile similar to selectively stearic residues into POMF triglycerides, in
CB. On the other hand, a cocoa butter equivalent positions 1 and 3 until a composition resembling CB
(CBE) is a type of fat that has a very similar chemical composition is obtained. The stearate donor could be
either free stearic acid or an ester of stearic acid,
however, for practical applications stearic acid is pre-
ferred due to its availability and low cost.
* Corresponding author. For that purpose, the enzyme used should be 1,3

0032-9592/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 3 2 - 9 5 9 2 ( 0 0 ) 0 0 2 6 0 - 0
934 D. Undurraga et al. / Process Biochemistry 36 (2001) 933–939

regiospecific, exchanging triglyceride residues only at 2.2. Analysis of glycerides


positions sn-1 and sn-3, and preferably immobilized
because in this form it is easily separated from the The composition of diglycerides and triglycerides
products and is also more thermostable than the free were determined by HPLC in a HPLC Shimadzu deliv-
enzyme. ery system (Shimadzu Scientific Instrument, Columbia,
The interesterification reaction can be carried out in MD) provided with refraction index detector. A CLC-
a system containing triglycerides and stearic acid ODS C-18 4.6 mm ID and 15 cm long Shimpack
only (‘solvent free system’) or the mixture could be column (Shimadzu) was used along with a mixture of
diluted with some suitable non-polar solvent. There are 3.5/l (v/v) acetone–acetonitrile as the mobile phase.
distinct advantages and disadvantages in using The samples were diluted in acetone up to a concentra-
either system, but in food application processes solvent tion of 2 g/l approximately. The injection volume was
free systems are clearly preferable. For technological of 20 ml. Detector and column were maintained at
purposes, it is important to determine the parameters 35°C. The chromatograms were processed in a Shi-
that affect the productivity of a CBE and find madzu CR4A integrator.
the optimal operation strategy that maximize produc-
tivity. 2.3. Interesterification in batch reactor
In this research, the production of a CBE through
enzymic interesterification of POMF with Lipozime™ A 23 factorial design (experimental design I) was used
in a solvent-free system using stearic acid as stearate to investigate the effect on productivity of the stearic
donor was studied. Two factorial experimental designs acid–POMF ratio (RS 1.2 and 1.4), the enzyme–sub-
were used to evaluate the effect of different parameters strate ratio (RE 7.7 and 15.4 BIU/g) and the initial
on productivity. humidity of the enzyme preparation (H 4.5 and 5.9%
The variables studied were the stearic acid– POMF d.w. basis).
ratio, which affects the equilibrium and, therefore, the Each experience consisted of a batch interesterifica-
time at which a composition similar to that of CB is tion reaction under conditions imposed by the experi-
obtained; the enzyme– substrate ratio which directly mental design. Reactions were carried out in 50 ml
affects the rate of reaction, and the humidity or water Erlenmeyer flasks with about 6 g of substrate each and
activity of the enzymic preparation, which have to be incubated with Lipozyme™ at 65°C in a rotary Shaker.
over a minimum value to activate the enzyme but not in The initial humidity was fixed as follows, samples of
such excess as to promote the hydrolysis of triglycerides Lipozyme™ were dried in vacuum at 40°C overnight
[7]. and then placed into a constant temperature, constant
relative humidity closed chamber containing a satu-
rated solution of an inorganic salt, and maintained in
2. Materials and methods the chamber until reaching constant weight.
This usually occurred in about 12 h. Using different
2.1. Materials salts the initial equilibrium humidity of the biocatalyst
could be fixed at the values of 4.5 and 5.9% d.w. basis.
The Novo-Nordisk commercial preparation of lipase Samples of about 10 mg were withdrawn at timed
(triacyl glycerol hydrolase EC 3.1.1.3) called intervals until equilibrium conditions were obtained.
Lipozyme™ (76.9 Batch Interesterification Unit (BIU)/ Samples were diluted in acetone and analyzed by
g) which presents 1, 3 regiospecificity, is immobilized on HPLC. Time of reaction, yield and productivity were
a macroporous anionic exchange resin with a particle recorded for each sample.
diameter of 0.2–0.6 mm. One BIU is defined as micro- Time of reaction (tr) is defined as the time required
moles of palmitic acid incorporated into triolein per for the incorporation of stearate into POMF triglyce-
minute [8]. rides at CB level, quantified by a stearate index (SI) as
Refined, bleached and deodorized POMF was kindly defined by Bloomer [1].
donated by The Palm Oil Research Institute of
2[SOS] + [POS]
Malaysia (PORIM). Stearic acid used had a purity over SI = (1)
99% (Riedel De Haen). Triolein (OOO) and trilinolein 2([POP] + [POS] +[SOS])
(LiLiLi) were obtained from Sigma Chemical Co. (St. where [triglyceride] is the triglyceride weight fraction. SI
Louis, MO). Other triglycerides such as POP, POO, for CB and POMF were 0.52 and 0.1, respectively.
SOS, SOO, PLiP, PLiLi, SLiS and SLiLi were pro- The yield (R or mass of triglycerides produced per
duced from the above standards by enzymic interester- unit mass of initial substrate), mass productivity (PM or
ification with stearic or palmitic acid. Acetone and mass of product/(time·initial mass of substrate)) and
acetonitrile (Baker) were of high-performance liquid specific productivity (PE or mass of product/(time
chromatography (HPLC) grade. BIU)) were calculated as follows:
D. Undurraga et al. / Process Biochemistry 36 (2001) 933–939 935

R
PM = (2) column. The packing was loaded up to 15 cm height
(1+ RS)tr
approximately. The substrate at a constant RS value of
R 1.4 was fed from the bottom using a peristaltic pump.
PE = (3) All the experimental set up including the reactor, feed
Re(1+ RS)tr
and product lines and pump head were enclosed in an
insulating box whose interior was controlled at 70°C by
2.4. Interesterification in a continuous reactor means of an air heating system. The feed rate was set to
obtain a product with a SI value of 0.529 0.01.
In order to appreciate the effect of higher RE values
on productivity than those attainable in batch reactors, 2.5. Separation of the product
a continuous interesterification system using a column
reactor packed with Lipozyme™ and an inert packing The interesterified product was distilled in a short
(0.4 mm glass balls) was employed. The reactor con- path distillation unit KDL 4 (UIC GmbH) at P=0.2
sisted of a 2.5-cm diameter and 25-cm long glass mbar and T= 140°C. At these conditions, free fatty
acids were totally eliminated, but not diglycerides pro-
duced in the reaction.

2.6. Functional analysis

The distilled product was subjected to scanning dif-


ferential calorimetry analysis using a DSC-7 Perkin–
Elmer calorimeter. The samples were tempered by
heating them up to 80°C, fast cooling to 0°C and
keeping for 7 days at 25°C. For the analysis, around 15
mg samples were taken and heated at the rate of
5°C/min from 5 to 50°C.

3. Results and discussion

3.1. Interesterification in batch reactor

Fig. 1. Time course of the interesterification in experiment 7 of The course of the interesterification reaction in expe-
experimental design I, showing changes in triglyceride composition.
, POS;
, POP; , SOS.
rience 7 (RS 1.2; RE 15.5 y H 5.9) is plotted in terms of
the composition of triglycerides in Fig. 1 and in terms
of level of incorporation of the stearic acid, in Fig. 2. In
these figures, it is observed that the desired composition
(with a SI value of 0.52) is achieved at 1.2 h approxi-
mately and equilibrium is reached at a SI value of 0.56.
Time of reaction, yield and productivity results were
adjusted to a linear model with interactions. Model
coefficients had a 95% level of confidence. The correla-
tions obtained were the following:
tr = 3.174−0.115 · RS − 1.906 · RE − 0.4035 · H
+ 0.2665 · RE · H (4)
R=0.9318− 0.0077 · RE − 0.006 · H (5)
PM = 0.2029+ 0.121 · RE + 0.0235 · H (6)
PE = 0.0157+ 0.0052 · RE + 0.0018 · H (7)
Symbols in italics stand for the coded variables (−1,
0, + 1).
Fig. 2. Time course of the interesterification in experiment 7 of Batch productivities slightly increased with H as a
experimental design I, showing its degree of stearate incorporation result of the combined effect of increasing both transes-
into triglycerides. terification and tryglyceride hydrolysis rates, water be-
936 D. Undurraga et al. / Process Biochemistry 36 (2001) 933–939

compounds has a pronounced negative effect on the


fusion characteristics of the product.
In order to asses the effect of higher RS y RE values
a 22 design (experimental design II; RS 1.2 and 1.6; RE
15.4 and 23.2) was carried out. Given the slight influ-
ence of initial humidity and to preclude possible hy-
drolytic effects the value of the initial humidity was not
varied but was fixed at the value 5.9% d.w. basis as
explained above. Results were as follows.

tr = 0.6176− 0.2258 · RE − 0.1763 · RS


+ 0.0922 · RE · RS (8)

R=0.8864− 0.0192 · RS (9)

PM = 0.6591+ 0.1972 · RE + 0.0807 · RS (10)

PE = 0.0336+ 0.0037 · RE + 0.0044 · RS (11)


Fig. 3. Time course of the interesterification in experiment 4 of
experimental design II, showing changes in triglyceride composition. The time course of the reaction of experiment 4 (RS
, POS;
, POP; , SOS. 1.6 and RE 23.2) expressed as the variation in the
percentage of triglycerides is shown in Fig. 3, and
expressed as the variation of the level of incorporation
of stearic (SI) in Fig. 4. In both figures, it can be seen
that the times of reaction are shorter than those of first
design. Furthermore, a higher SI equilibrium value was
observed due to the higher RS values used in the second
design.
PM increases with higher values of RE as expected
from theory. The PE also increased with increasing
values of RE, although the effect was less pronounced
than in the first design. Theoretically there is no rela-
tion between these variables, and the variations ob-
served might be due to the decrease of humidity, either
by evaporation or desorption into the oil phase. In the
batch system humidity was not controlled and dehydra-
tion should be expected to increase with time. Therefore
the reaction rate should be higher at higher values of
RE, which corresponded to shorter reaction times, when
the humidity was still higher.
Fig. 4. Time course of the interesterification in experiment 4 of The effect of RS on PM is rather complex. An in-
experimental design II, showing its degree of stearate incorporation crease of RS should cause a decrease in PM according to
into triglycerides. Eq. (1). Nevertheless, RS also affects — inversely —
the time of reaction, therefore, the overall effect is
ing a reactant of this later reaction (it is assumed that, diminished. This would explain the low value of the
hydrolysis is an intermediate step of the whole inter- coefficient of RS in Eq. (10).
esterification reaction [9]). RE is the single factor that most affects productivity.
For the purpose of producing high quality CBE, However, in batch reactors it is not practical to increase
hydrolysis reactions pose another difficulty, the possible its value over 25 BIU/g of substrate (32% in weight of
production of diglycerides, which have a strong effect Lipozyme™) due to operational problems (poor mix-
in the fusion profile of the product. Fig. 7 is the ing, sampling difficulties etc.), therefore, in order to
chromatogram of an interesterified product after short investigate the effect of this variable at higher values a
path distillation. Several smaller distinct peaks corre- continuous system was utilized.
sponding to compounds of lower retention time than Results of batch interesterification experiments car-
triglycerides can be observed. These compounds, which ried out by Blumer et al. in a model system consisting
make up approximately 9% of the mixture, are believed of triolein and palmitic acid using Lipozyme™ were
to correspond to diglycerides. The presence of these similar to those of this work [10].
D. Undurraga et al. / Process Biochemistry 36 (2001) 933–939 937

In summary the best results in batch systems were RE = 23.3 (BIU/g), a ratio between substrates, RE =1.6
obtained using the highest enzyme– substrate ratio, and an initial humidity of H=5.9% (d.w. basis). Under
these conditions values of PM, PE and R were 0.913
(g/h·g), 0.0393 [g/(BIU·h)] and 86.7%, respectively.
Table 1
Enzyme–substrate ratio (RE), residence time (~), yield (R), specific
productivity (PE) and mass productivity (PM) in continuous packed 3.2. Interesterification in continuous reactor
bed reactor
In order to prevent or to minimize dehydration of the
RE (UI/g) ~ (min) R PE (g/BIU·h) PM (g/g·h) biocatalyst, whose initial value was set at 5.9% as
73.13 14.0 0.86 0.0210 1.536
described above, the substrate fed to the reactor was
52.6 20.3 0.872 0.0204 1.074 saturated with water. In this manner, it was possible to
48.16 24.1 0.866 0.0187 0.901 maintain constant SI during at least six residence times.
Results obtained in the continuous system are shown in
Table 1. Extrapolating the results of the batch system
and based on previous experiences with the continuous
system it was possible to determine a suitable feed rate
(and residence time) to produce a SI value of 0.52 at the
reactor outlet. An increase in RE caused a decrease in
residence time (~), an increase in productivity but prac-
tically no effect on yield (Table 1).
A packed bed continuous reactor in a plug flow
regime behaves as a perfectly mixed batch reactor.
What occurs in a continuous reactor along its length is
equivalent to what occurs in a batch reactor along time.
Therefore, it should be possible to make a plot showing
the effect of RE on productivities along the whole range
(15.4− 73.1 BIU/g) as shown in Figs. 5 and 6. In these
figures, it can be observed that the behaviour at high
RE in the continuous reactor is not the extension of
what occurs at lower RE in the batch reactors, as might
have been expected.
These behaviours can be in part explained assuming
Fig. 5. Changes in mass productivity in the whole range of BIU/sub- that only a fraction of the enzymic activity loaded into
strate ratio studied.
, Batch reactor; , continuous packed bed the reactor is expressed, and in part, by considering
reactor.
that in the continuous reactor the loss of water was less
(better control) than in the batch system, thus reducing
or eliminating the dependence between PE and RE
found in the latter.
Non-expressed enzymic activity can be attributed to
external diffusion restrictions since the enzyme is ab-
sorbed on to the particle surface and not inside the
micropores [11].
It can be seen that the yield was nearly constant at
different RE values, which would indicate that humidity
was similar in all experiments in the continuous system,
because otherwise yield would have fluctuated widely
(humidity strongly affects yield).
The best results, as far as production was concerned,
were obtained with the highest enzyme–substrate ratio
RE = 73.1, RS = 1.4 and ~= 14 min, obtaining a mass
productivity of 1.54 [g/(g·h)] and a specific productivity
of 0.02 [g/BIU·h] with a yield of 86%. These results can
be compared with those obtained by Forsell et al. [7]
Fig. 6. Changes in specific productivity in the whole range of BIU/ who interesterified rape seed oil with lauric acid in a
substrate ratio studied.
, Batch reactor; , continuous packed bed solvent-free system using a packed bed continuous reac-
reactor. tor with Lipozyme™. The ratio between lauric
938 D. Undurraga et al. / Process Biochemistry 36 (2001) 933–939

Table 2
Triglyceride composition POMF, CB and CBE produced by inter-
esterification

Sample POP POS SOS Other TGs DGs

POMF 74.3 14.3 2.0 9.4 –


CB 23.4 42.8 27.5 3.6 2.7
CBE 23.4 38.5 20.2 8.2 9.7

3.3. Scanning differential calorimetry

As mentioned earlier, the chromatogram of the inter-


esterified product after distillation at vacuum to remove
impurities, still presented compounds other than
triglycerides, which are believed to correspond to
diglycerides. These latter strongly affect the fusion
characteristics of the products as can be observed in the
thermograms (Figs. 7–10). Table 2 shows the composi-

Fig. 7. Chromatogram of interesterified and distilled product.

Fig. 9. Thermogram of CB.

Fig. 8. Thermogram of POMF.

acid:rape seed oil acid used was 1:3. With a residence


time of 30 min, a conversion of 60% of the equilibrium
value was obtained, (continuous interesterification in
this work was carried out at 85% conversion) and mass
productivity and yield were of 1.34 [g/(g·h)] and 89%,
respectively. Svanholm [12] using soybean oil instead of
rape seed oil and a ratio between acid and oil of 2:5
obtained a conversion of 65% of the equilibrium value
and a specific productivity 1.8 [g/(g·h)]. These and the
present results indicate that conversion by an interester-
ification reaction with Lipozyme™ are similar irrespec-
tive of the substrates used. The yield of the reaction is
high (80–90%) and mass productivity is less than 2
[g/(g·h)]. Fig. 10. Thermogram of interesterified and distilled product.
D. Undurraga et al. / Process Biochemistry 36 (2001) 933–939 939

Table 3
Scanning differential calorimetry parameters of POMF, CB and CBE produced by interesterification

Sample Onset temperature (°C) Peak temperature (°C) Area (J/g)

POMF 21.47 29.76 107.86


CB 24.08 31.04 86.48
CBE (first peak) 16.24 18.67 2.47
Second peak 20.94 22.33 1.01
Third peak 25.07 29.73 56.84

tion of POMF, a sample of pure CB, and a sample of there is little to be done to improve in this aspect.
CBE of this research, whose chromatogram is shown in Further purification of the reaction products to obtain
Fig. 7. a high quality CBE, although technically feasible, could
From the thermograms three characteristics for each render the overall process uneconomical.
sample can be obtained: the onset temperature (TO),
peak temperature (TP) and the area of the curve under
each peak. These values for POMF, CB and a sample References
of CBE of this study are shown in Table 3. The
thermogram of CB exhibts a narrow range of fusion [1] Bloomer S, Adlercreutz P, Mattiasson B. Triglyceride interester-
ification by lipases. I. Cocoa butter equivalents from Fraction
between 24 and 35°C with a steep descending slope.
Palm Olein (FPO). J Am Oil Chem Soc 1990;67:519 – 24.
The area under this peak is 88.7 J/g and represents the [2] Sridhar R, Lakshminarayana G, Kaimal T. Modification of
heat of fusion of CB. In addition two smaller fusion selected Indian vegetable fats into cocoa butter substitutes by
peaks are noted between 5 and 15°C and 15 and 23°C, lipase-catalized ester interchange. J Am Oil Chem Soc
which are likely due to the crystal structures I and II of 1991;68:726 – 30.
[3] Chang M, Abraham G. Production of cocoa butter-like fat from
CB. The thermogram of CBE resembles that of CB, but
interesterification of vegetable oils. J Am Oil Chem Soc
a closer scrutiny indicates several important differences, 1990;67:832 – 4.
which could impair seriously its utilization in choco- [4] Chong C, Hoh Y, Wang C. Fractionation procedures for obtain-
lates. As can be observed, CBE shows a ‘shoulder’ ing cocoa butter like fat from enzymatically interesterified palm
between 35 and 39°C in the descending portion of its olein. J Am Oil Chem Soc 1992;69:137 – 40.
[5] Mojovic L, Siler-Marinkovic S, Kukic G, Vunjak-Novakovic G.
largest peak, due probably to trisaturated triglicerides
Rhizopus arrhizus lipase-catalyzed interesterification of the mid-
(PPS, PPP, PSS and others), which may have formed fraction of palm oil to a cocoa butter equivalent fat. Enzyme
during the interesterification reaction. These triglice- Microb Technol 1993;15:438 – 43.
rides have larger fusion temperature than the disatu- [6] Minifie B. Chocolate, Cocoa, and Confectionary, third ed. New
rated triglicerides. In addition, it exhibits a fairly large York: Chapman & Hall, 1989.
[7] Bulletin Novo B 347c-GB 200. December 1992. Novo Nordisk.
peak between 5 and 15°C, probably due to triglycerides
[8] Forsell P, Paravuori P, Linko P, Poutanen K. Enzymatic tranes-
in the crystalline from I, and two smaller peaks between terification of rapeseed oil and lauric acid in a continuous
15 and 25°C, which could be attributed to the presence reactor. J Am Oil Chem Soc 1993;70:1105 – 9.
of diglycerides. It have been noted that the presence of [9] Malcata FX, Reyes H, Garcı́a H, Hill C, Amundson C. Kinetic
diglycerides in fats affects their fusion behaviour and mechanisms of reactions catalyzed by immobilized lipases.
Enzyme Microb Technol 1992;14:426 – 46.
[10,13].
[10] Bloomer S, Adlercreutz P, Mattiasson B. Triglyceride interester-
ification by lipases. 2-Reaction parameters for the reduction
trisaturated impurities and diglycerides in batch reactions. Bio-
catalysis 1991;5:145 – 62.
4. Conclusions [11] Vázquez F, Pyle D, Asenjo J. Factors affecting the esterification
of lauric acid using an immobilized biocatalyst: enzyme charac-
POMF appears to be a suitable raw material for .
terization and studies in a well-mixed reactor. Biotechnol Bioeng
interesterification with stearic acid using Lipozyme™, 1995;46:69 – 79.
[12] Svanholm H. Enzymatic Interesterification and Esterification.
leading to products whose composition matches closely
Novo Report. A-06052 1989: p. 1 – 27.
that of CB. Nevertheless the fusion profile of the prod- [13] Cebula DJ, Smith KW. Differential scanning calorimetry of
ucts of the reaction does not warrant their use as a confectionary fats: part II — effects of blends and minor
good quality CBE for the reasons discussed above and components. J Am Oil Chem Soc 1992;69:992 – 8.

You might also like