Enzymatic production of cocoa butter equivalents high in 1-palmitoyl-2-oleoyl-3-stearin in continuous packed bed reactors

Article abstract of DOI:10.1007/s11746-014-2412-7

This study aimed to optimize the lipase-catalyzed transesterification of high oleic sunflower oil (A) with a mixture of ethyl palmitate and ethyl stearate (B) to produce cocoa butter equivalents with a weight ratio of 1-palmitoyl-2-oleoyl-3-stearoyl-rac-glycerol (POS) to total symmetric monounsaturated triacylglycerols (SMUT) that is similar to that of cocoa butter by response surface methodology. The reaction was performed in a continuous packed bed reactor, using 0.45 g of Lipozyme RM IM as the biocatalyst. The effects of temperature (Te), residence time (RT), substrate molar ratio (SR, B/A), and water content (WC) of the substrates on the composition of reaction products were elucidated using the models established. Optimal reaction conditions for maximizing total SMUT and POS contents while minimizing the levels of diacylglycerol formation and acyl migration were: Te, 60°C; RT, 28.5 min; SR, 8.5; WC, 300 mg/kg. The contents of total SMUT, POS, and diacylglycerol in the reaction products and the content of palmitoyl and stearoyl residues at the sn-2 position of triacylglycerols in the products were 52.0, 25.1, 9.4, and 4.8 %, respectively, under these conditions. Successful scale-up of the reaction was achieved under the optimal conditions, using 5 g of the lipase. A silver-ion high performance liquid chromatography analysis showed that the products obtained by the larger scale reaction contained 49.1 % total SMUT and 6.1 % of their positional isomers.

Full text of DOI:10.1007/s11746-014-2412-7

J Am Oil Chem Soc
DOI 10.1007/s11746-014-2412-7
ORIGINAL PAPER
Enzymatic Production of Cocoa Butter Equivalents High
in 1-Palmitoyl-2-oleoyl-3-stearin in Continuous Packed Bed
Reactors
Sohee Kim
Byung Hee Kim
•
In-Hwan Kim
•
Casimir C. Akoh
•
Received: 10 July 2013 / Revised: 24 November 2013 / Accepted: 30 December 2013
Ó AOCS 2014
Abstract This study aimed to optimize the lipase-cata-
lyzed transesterification of high oleic sunflower oil (A) with
a mixture of ethyl palmitate and ethyl stearate (B) to pro-
duce cocoa butter equivalents with a weight ratio of
performance liquid chromatography analysis showed that
the products obtained by the larger scale reaction contained
49.1 % total SMUT and 6.1 % of their positional isomers.
1
-palmitoyl-2-oleoyl-3-stearoyl-rac-glycerol (POS) to total
Keywords Cocoa butter equivalents ꢀ Lipase ꢀ
symmetric monounsaturated triacylglycerols (SMUT) that
is similar to that of cocoa butter by response surface
methodology. The reaction was performed in a continuous
packed bed reactor, using 0.45 g of Lipozyme RM IM as the
biocatalyst. The effects of temperature (Te), residence time
Continuous packed bed reactor ꢀ 1-Palmitoyl-2-oleoyl-3-
stearoyl-rac-glycerol ꢀ Response surface methodology ꢀ
Silver-ion high performance liquid chromatography
(
(
RT), substrate molar ratio (SR, B/A), and water content
WC) of the substrates on the composition of reaction
Introduction
products were elucidated using the models established.
Optimal reaction conditions for maximizing total SMUT
and POS contents while minimizing the levels of diacyl-
glycerol formation and acyl migration were: Te, 60 °C; RT,
Cocoa butter, the fat component of chocolate, is the most
expensive constituent in chocolate formulations and is
composed mainly of symmetric monounsaturated triacyl-
glycerols (SMUT) containing an oleoyl (O) residue at the
sn-2 position and palmitoyl (P) or stearoyl (S) residues at
sn-1,3 positions of the glycerol backbone, represented as
POP (15–19 %), POS (36–41 %), and SOS (25–31 %) [1,
2]. The large amounts of total SMUT (79–87 %) in cocoa
butter are responsible for its unique physical attributes,
such as the dominating existence of stable b crystals and
rapid melting at body temperature, thereby providing
chocolate with snap, gloss, and a smooth mouthfeel [3].
Cocoa butter equivalents (CBE), which are produced
from cheaper fat/oil resources than cocoa butter and have
similar total SMUT content to cocoa butter, have been used
to replace cocoa butter in the manufacturing of chocolate
products, because they reduce production costs. Typical
commercial CBE contain more SOS and POP and less POS
than cocoa butter since they are produced by blending fats
rich in POP, such as palm mid-fraction with SOS-rich fats,
which are obtained by fractional crystallization (e.g., shea
stearin, sal stearin) or enzymatic interesterification (e.g.,
high oleic sunflower oil esterified with stearic acid) [2].
2
8.5 min; SR, 8.5; WC, 300 mg/kg. The contents of total
SMUT, POS, and diacylglycerol in the reaction products
and the content of palmitoyl and stearoyl residues at the
sn-2 position of triacylglycerols in the products were 52.0,
2
5.1, 9.4, and 4.8 %, respectively, under these conditions.
Successful scale-up of the reaction was achieved under the
optimal conditions, using 5 g of the lipase. A silver-ion high
S. Kim ꢀ B. H. Kim (&)
Department of Food Science and Technology,
Chung-Ang University, Anseong 456-756, Korea
e-mail: bhkim@cau.ac.kr
I.-H. Kim
Department of Food and Nutrition, Korea University,
Seoul, Korea
C. C. Akoh
Department of Food Science and Technology,
The University of Georgia, Athens, GA, USA
1
23

J Am Oil Chem Soc
In recent years, CBE manufacturers and researchers
have attempted to prepare POS-rich fats for formulating
CBE with weight ratios between the three SMUT species
that are similar to those of cocoa butter. The POS-rich fats
are produced by esterifying POP-rich fats with stearic acid
Diacylglycerols (DAG) are intermediates in lipase-cat-
alyzed transesterification where ester hydrolysis is
involved. The formation of more DAG and the esterifica-
tion of less DAG during the reaction results in the presence
of more DAG in the final product. The higher DAG content
in cocoa butter causes a slower crystallization rate of the
cocoa butter, thereby resulting in faster bloom development
in chocolate [13]. Because DAG formation is undesirable,
but unavoidable in the lipase-catalyzed production of CBE,
it is necessary to minimize it to improve chocolate quality
as well as CBE yield.
(
or its alkyl esters) or tristearin (SSS) using sn-1,3 specific
lipases as biocatalysts [4–7]. A possible alternative for
producing POS-rich fats is to esterify triolein (OOO)-rich
vegetable oils, such as high oleic sunflower oil [8], olive–
pomace oil [9], and tea seed oil [10] with mixtures of
palmitic and stearic acids (or their methyl esters) with the
same enzymes. These published studies have employed
batch reaction systems for the production of POS. How-
ever, despite using sn-1,3-specific lipases, use of a stirred
batch reactor for performing lipase-catalyzed esterification
often leads to significant migration of acyl groups from one
site on the glycerol backbone to another site, thereby
forming positional isomers of POS, such as PSO and SPO.
Furthermore, these studies did not report the exact contents
of SMUT species and SMUT positional isomers in the
reaction products. They simply analyzed the fatty acid
compositions of triacylglycerols using gas chromatography
The aims of this study were to model a small-scale
lipase-catalyzed transesterification process and to optimize
the reaction conditions through response surface method-
ology (RSM). The effects of four reaction parameters (i.e.,
temperature, residence time, substrate molar ratio, and
water content) on the total SMUT, POS, and DAG contents
in the reaction products and the content of palmitoyl and
stearoyl residues at the sn-2 position of triacylglycerols in
the products, which is indicative of the level of acyl
migration during the reaction, were evaluated using the
regression models established. Larger scale transesterifi-
cation was conducted under the optimal conditions estab-
lished for the small-scale reaction, and the reaction
products were evaluated for the contents of total SMUT
and their positional isomers.
(
GC) [6, 8, 10] or utilized high performance liquid chro-
matography (HPLC) with a C18 column, where isomeric
triacylglycerols cannot be separated [4, 5, 7, 9].
Continuous packed bed reactors are the most commonly
used reactors for immobilized lipase in industrial scale
applications because of their easy construction and opera-
tion and the effective recovery and reuse of the lipase. The
rupture of the supporting materials for immobilized lipase,
which facilitates acyl migration, can be avoided in these
reactors [11]. Moreover, the ratio between substrate and
lipase is much lower in a continuous packed bed reactor
than in a stirred batch reactor. The lower substrate/enzyme
ratio results in a shorter reaction time and a higher reaction
rate, thereby reducing acyl migration [12]. In these
respects, continuous packed bed reactors are most desirable
and feasible for the industrial scale production of CBE or
individual SMUT species.
Materials and Methods
Materials
High oleic sunflower oil (oleic acid content, *82 % of
total fatty acids; triolein content, *73 % of total acylgly-
cerols) was obtained from Nongshim Corp. (Seoul, Korea).
Palmitic and stearic acids were purchased from Sigma
Chemical Co. (St. Louis, MO, USA). Lipozyme RM IM, a
sn-1,3 specific immobilized lipase from Rhizomucor mie-
hei, was provided by Novozymes North America, Inc.
(Franklinton, NC, USA). Pancreatic lipase was purchased
from Sigma. POP and its positional isomer PPO were
purchased from Varian Inc. (Lake Forest, CA, USA), POS
and PSO were obtained from Tsukishima Food Industry
Co. (Tokyo, Japan), and SOS and SSO from Sigma. All
other reagents were analytical or HPLC grade.
In the present study, we successfully prepared CBE rich
in POS whose weight ratios between the three SMUT
species were similar to those of cocoa butter. This was
accomplished via lipase-catalyzed transesterification of
high oleic sunflower oil with a mixture of ethyl palmitate
and ethyl stearate that were more lipid-soluble forms of
palmitic and stearic acids, respectively, in a continuous
packed bed reactor. Using fatty acid ethyl esters as acyl
donors allows the reaction to proceed at lower temperature
than using the corresponding free fatty acid forms, thereby
reducing acyl migration. Furthermore, we investigated the
relative contents of total SMUT and total SMUT positional
isomers in our CBE samples using a silver-ion HPLC
system.
Preparation of Ethyl Palmitate and Ethyl Stearate
Palmitic and stearic acids were converted into their ethyl
ester forms according to previously described methods [14]
with slight modification. Palmitic or stearic acids (100 g)
were solubilized in chloroform (200 mL) and ethylated
with 2 % sulfuric acid in ethanol (400 mL) in a boiling
1
23

J Am Oil Chem Soc
flask equipped with a reflux condenser for 30 min. After
being cooled to room temperature, the product was washed
with water in a separatory funnel. The chloroform layer
containing ethyl palmitate or ethyl stearate was passed
through an anhydrous sodium sulfate column, solvent
removed in a rotary vacuum evaporator, and then com-
pletely dried under a nitrogen flush. The purities of the
resulting ethyl palmitate and ethyl stearate were[99 % by
GC analysis.
in the same manner as those used for the small-scale
reaction.
Separation of Triacylglycerols
Triacylglycerols present in the reaction products were
separated by thin-layer chromatography (TLC). Aliquots of
the reaction medium (150 mg) were dissolved in chloro-
form (350 lL), placed on silica gel 60 F254 TLC plate
(Merck, Darmstadt, Germany), and developed with petro-
Lipase-Catalyzed Transesterification
leum ether/diethyl ether/acetic acid (90:10:1, v/v/v). The
plate was dried in air and the bands were visualized under
ultraviolet light. The band corresponding to the triacyl-
glycerols was scraped from the plate and then the triacyl-
glycerols were extracted three times with anhydrous
diethyl ether. After removing the diethyl ether under a
nitrogen flush, the triacylglycerols were used for analyzing
fatty acids at the sn-2 position and positional isomers of
SMUT.
The transesterification of high oleic sunflower oil with the
2
:3 (mol/mol) mixture of ethyl palmitate and ethyl stearate
for producing SMUT was performed in a continuous
packed bed reactor, which consisted of a stainless steel
tube (4.9 cm 9 2.6 mm i.d.) packed manually with 0.45 g
of Lipozyme RM IM. A syringe pump (model Legato 100;
KD Scientific, Holliston, MA, USA) was used to continu-
ously supply the substrates in a 50-mL glass syringe to the
reactor. The syringe was wrapped in heating tape to
maintain the substrates in the liquid phase. The reactor was
immersed in a water bath equipped with a water circulator
and a digital temperature controller to maintain it at a
desired temperature. The outlet port of the reactor was
irradiated with an infrared lamp to avoid clogging by
solidification of the reaction products. To adjust water
content, substrates were dried under a nitrogen flush and
then distilled water was added to the dried substrates and
uniformly dispersed using ultrasonication. The water con-
tent of the substrates was monitored using a Karl-Fisher
moisture analyzer (model 803 Ti Stand; Metrohm Ltd.,
Herisau, Switzerland). Prior to starting the reaction, the
reactor was flushed with 10 void volumes of the substrates
to be used at a flow rate of one void volume per minute.
Then, reaction products were collected after the enzyme
bed had been conditioned with two void volumes of the
substrates at a flow rate of one void volume per residence
time. A new enzyme bed was used for each trial.
Analysis of Fatty Acid Residues at the sn-2 Position
Pancreatic hydrolysis was used to determine the fatty acid
residues at the sn-2 position of triacylglycerols [15]. The
extracted triacylglycerols were mixed with 2 mL of 1 M
Tris–HCl buffer (pH 7.6), 0.05 % sodium cholate solution
(0.5 mL), and 2.2 % calcium chloride solution (0.2 mL),
and the mixture was vortexed thoroughly. After adding
pancreatic lipase (20 mg), the mixture was incubated at
40 °C for 2 min, vortexed vigorously for 0.5 min, and then
6 M hydrochloric acid (5 mL) was added. Anhydrous die-
thyl ether (15 mL 9 2) was added and the mixture was
vortexed. The upper diethyl ether layer containing pancre-
atic lipase hydrolysates was collected and passed through an
anhydrous sodium sulfate column. The hydrolysates were
then placed on silica gel 60 F254 TLC plate and developed
with hexane/diethyl ether/acetic acid (50:50:1, v/v/v). The
plate was dried in air, sprayed with 0.2 % 2,7-dichloroflu-
orescein in methanol, and the bands were visualized under
ultraviolet light. The band corresponding to the sn-2
monoacylglycerols was scraped from the plate, methylated
with 14 % boron trifluoride in methanol, and analyzed by
GC as described previously [16].
A larger scale transesterification was carried out under
the optimal conditions established for the small-scale
reaction described above in a continuous packed bed
reactor, which consisted of a glass column (27.7
cm 9 7.0 mm i.d.) with a water jacket. The column was
manually packed with 5 g of Lipozyme RM IM, which was
then fixed in place using a stainless steel sieve (50 mesh).
The substrates were kept in a reservoir, which consisted of
a glass vessel with a water jacket. The reactor and reservoir
were maintained at a desired temperature with a water
circulator (Model RW-0525G, Jeio Tech, Seoul, Korea).
The substrates were pumped upward into the column by a
fluid metering pump (model QG6; Fluid Metering Inc.,
Syosset, NY, USA). The enzyme bed had been conditioned
Analysis of Triacylglycerol Composition
The triacylglycerol composition of the reaction products
was analyzed by GC according to the method described by
Kang et al. [16]. An Agilent Technologies 7890N gas
chromatograph (Agilent Technologies Inc., Palo Alto, CA,
USA), equipped with a flame ionization detector and a
fused silica capillary column (DB-17HT, 15 m 9 0.25 mm
i.d. 9 0.15 lm film thickness, J & W Scientific, Folsom,
1
23

J Am Oil Chem Soc
CA, USA) was used for the analysis. Aliquots of the
reaction medium (25 mg) were dissolved in chloroform
Analysis of SMUT Positional Isomers
(
1 mL) and the solution (1 lL) was injected in split mode
split ratio of 50:1). The carrier gas was helium at a flow
The relative contents of SMUT (POP, POS, and SOS) and
their positional isomers (PPO, PSO, SPO, and SSO) were
analyzed by silver-ion HPLC following the method
described by Kang et al. [16]. An HPLC system (Gilson
Inc., Middleton, WI, USA) composed of a model 305 pump
and a Rheodyne 7161 injector (Rheodyne, Inc., Cotati, CA,
USA) with a 20-lL injection loop was used for the ana-
lysis. The detector was an Alltech 3300 evaporative light-
scattering detector (Alltech Associates Inc., Deerfield, IL,
USA) in which the drift tube was maintained at 40 °C and
the flow rate of nitrogen was 1.5 L/min. The separation of
(
rate of 2.0 mL/min. The injector and detector temperatures
were maintained at 360 °C, respectively. The oven tem-
perature was initially held at 280 °C for 1 min and was
programmed to increase to 360 °C at a rate of 3 °C/min
and held at 360 °C for 5 min. Triacylglycerols were
identified by comparing their retention times with those of
the triacylglycerol standards or those of triacylglycerol
species present in cocoa butter, and their relative contents
were calculated as w/w%.
Table 1 Central composite circumscribed design arrangement and responses for the Lipozyme RM IM-catalyzed transesterification of high oleic
sunflower oil with a mixture of ethyl palmitate and ethyl stearate
Exp. no.
Factor
Response
Te
(
RT
(min)
SR
WC
(mg/kg)
Total SMUT
content (w/w%)
POS content
(w/w%)
DAG content
(w/w%)
sn-2 P ? S
content (w/w%)
°C)
a
1
2
3
4
5
6
7
8
9
40
60
40
60
40
60
40
60
40
60
40
60
40
60
40
60
35.9
64.1
50
50
50
50
50
50
50
50
50
15
3
3
3
3
9
9
9
9
3
3
3
3
9
9
9
9
6
6
6
6
300
300
27.9 ± 0.8
14.0 ± 0.4
14.7 ± 0.5
16.9 ± 0.5
17.8 ± 0.4
19.5 ± 0.2
22.1 ± 2.0
27.4 ± 0.7
26.8 ± 1.4
14.2 ± 0.3
15.4 ± 1.9
16.0 ± 0.1
16.8 ± 0.3
21.4 ± 0.3
23.5 ± 1.5
23.2 ± 2.6
21.7 ± 1.0
19.3 ± 2.5
22.1 ± 1.1
5.9 ± 0.6
8.0 ± 2.1
5.2 ± 2.7
11.3 ± 2.1
6.2 ± 1.8
8.1 ± 0.2
5.7 ± 3.2
9.7 ± 0.5
8.6 ± 2.5
13.3 ± 0.6
9.0 ± 2.0
15.1 ± 0.5
10.3 ± 1.2
15.4 ± 3.7
15.8 ± 3.5
18.5 ± 2.2
18.2 ± 0.8
10.6 ± 3.7
10.5 ± 2.7
4.3 ± 1.4
16.1 ± 0.3
13.2 ± 0.4
14.4 ± 4.5
9.5 ± 0.8
12.3 ± 1.6
10.3 ± 2.7
10.5 ± 4.0
5.7 ± 0.7
1.2 ± 0.3
2.7 ± 2.0
7.5 ± 3.2
9.0 ± 1.7
2.6 ± 2.2
2.7 ± 0.2
5.3 ± 1.6
9.3 ± 1.7
1.5 ± 0.3
2.6 ± 0.6
5.7 ± 0.5
10.1 ± 0.5
2.6 ± 0.6
3.5 ± 1.4
7.6 ± 0.2
15.6 ± 4.0
5.7 ± 1.4
7.6 ± 0.8
1.2 ± 0.2
12.8 ± 2.3
4.1 ± 0.9
5.6 ± 1.1
4.9 ± 0.2
4.8 ± 1.0
5.1 ± 0.3
7.1 ± 2.4
5.7 ± 1.4
15
29.1 ± 0.2
33.8 ± 1.0
35.7 ± 0.8
39.1 ± 0.6
44.2 ± 4.2
55.2 ± 1.4
53.9 ± 2.8
28.5 ± 0.9
30.6 ± 3.3
32.3 ± 0.2
33.6 ± 0.6
42.9 ± 0.7
47.3 ± 3.1
46.6 ± 4.9
43.9 ± 1.7
40.1 ± 3.6
46.2 ± 0.5
12.3 ± 1.8
42.8 ± 1.6
22.0 ± 0.3
53.8 ± 3.1
44.6 ± 3.6
45.2 ± 3.4
45.6 ± 0.9
46.2 ± 0.1
47.6 ± 0.9
80
300
80
300
15
300
15
300
80
300
80
300
15
1,000
1,000
1,000
1,000
1,000
1,000
1,000
1,000
650
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
15
80
80
15
15
80
80
47.5
47.5
1.5
93.5
47.5
47.5
47.5
47.5
47.5
47.5
47.5
650
650
650
20.5 ± 1.8
10.6 ± 0.0
26.5 ± 1.7
21.4 ± 2.8
21.7 ± 2.7
21.9 ± 1.5
22.2 ± 1.0
22.8 ± 1.7
1.8
650
10.2
650
6
6
6
6
6
155
1,145
650
650
650
Te temperature, RT residence time, SR substrate molar ratio (high oleic sunflower oil to 2:3 (mol/mol) mixture of ethyl palmitate and ethyl
stearate), WC water content, SMUT symmetric monounsaturated triacylglycerols, POS 1-palmitoyl-2-oleoyl-3-stearoyl-rac-glycerol, DAG dia-
cylglycerols, sn-2 P ? S palmitoyl and stearoyl residues found at the sn-2 position of triacylglycerols
a
Mean ± standard deviation (n = 2)
1
23

J Am Oil Chem Soc
SMUT and their positional isomers was performed on a
ChromSpher Lipids column (250 mm 9 4.6 mm, 5 lm
particle size, silver-ion impregnated, Varian Inc., Lake
Forest, CA, USA). The column was conditioned with an
isocratic mobile phase for 48 h prior to analysis. The
mobile phase was a mixture of toluene, ethyl acetate, and
hexane (10:10:80, v/v/v) at a flow rate of 1 mL/min.
the reaction products and sn-2 P ? S content of triacyl-
glycerols present in the products in each of the 27 exper-
imental settings generated by the principles of RSM used in
this study. The total SMUT content ranged from as low as
12.3 % to as high as 55.2 % and the POS content was in
the range of 5.9–27.4 %. The weight ratio of POS to total
SMUT in the reaction products ranged from 0.48 to 0.50.
Natural cocoa butter contains POS and total SMUT in
weight ratios of 0.46–0.48 [2, 17]. The weight ratio of POP
to total SMUT (0.14–0.19), and the weight ratio of SOS to
total SMUT (0.31–0.38) in the reaction products, were also
similar to those of natural cocoa butter (0.17–0.23 for POP/
total SMUT and 0.31–0.35 for SOS/total SMUT, respec-
tively; data not shown). The DAG content in the reaction
products ranged from 4.3 to 18.5 % and sn-2 P ? S content
of triacylglycerols in the products was in the range of
1.2–15.6 %.
Experimental Design for RSM
The experimental work was designed according to the
principles of RSM with the assistance of the commercial
software, Design expert 8.0 from Stat-Ease (Minneapolis,
MN, USA). Twenty-seven experimental settings consisting
of eight star points and three center points (star distance is
1
.414) were generated with four factors and five levels for
each factor by central composite circumscribed (CCC)
design. The four factors used to model and optimize the
lipase-catalyzed transesterification were temperature
The four best-fitting quadratic models for the data were
determined by backward elimination. Analysis of variance
(ANOVA) showed that all regression models were
(
Te = 40–60 °C), residence time (RT = 15–80 min),
substrate molar ratio [SR = 3–9, high oleic sunflower oil
to the 2:3 (mol/mol) mixture of ethyl palmitate and ethyl
stearate], and water content (WC = 300–1,000 mg/kg).
Response variables were total SMUT (i.e., POP ?
POS ? SOS), POS, and DAG contents in the reaction
products and the content of palmitoyl and stearoyl residues
at the sn-2 position (sn-2 P ? S content) of triacylglycerols
in the products. Data from the CCC design were analyzed
by the least square method to fit the quadratic polynomial
regression model, given by the following Eq. (1):
Table 2 ANOVA table
Variable
df
SS
MS
F value
p value
Total SMUT content
Corrected total
Model
25
1,929.22
1,814.48
114.74
112.64
2.11
8
17
15
2
226.81
6.75
33.60
7.13
\0.0001
Residual
Lack of fit
Pure error
POS content
Corrected total
Model
7.51
0.1298
1.05
4
4
3
4
X
X
X X
2
i
Y ¼ b þ
b X þ
b X þ
b X X ;
ii i j
ð1Þ
25
8
449.47
419.37
30.11
29.69
0.42
0
i
i
ii
i¼1
i¼1
i¼1 j¼iþ1
52.42
1.77
1.98
0.21
29.60
9.42
\0.0001
Residual
17
15
2
where, Y represents the response variables, i.e. total SMUT
content, POS content, DAG content, sn-2 P ? S content,
b , b , b , and b are constant coefficients of intercept,
Lack of fit
Pure error
DAG content
Corrected total
Model
0.1000
0
i
ii
ij
linear, quadratic, and interaction terms, respectively, and Xi
25
7
361.23
289.71
71.52
56.77
14.75
and X are independent variables (Te, RT, SR, and WC).
j
41.39
3.97
3.55
7.37
10.42
0.48
\0.0001
Residual
18
16
2
Lack of fit
Pure error
sn-2 P ? S content
Corrected total
Model
0.8423
Results and Discussion
Model Fitting
25
11
14
12
2
303.22
283.54
19.68
17.57
2.11
25.78
1.41
1.46
1.05
18.34
1.39
\0.0001
Response surface methodology is one application of mul-
tiple linear regression models that is useful for investigat-
ing and optimizing complex processes, such as enzymatic
reactions. In this study, it was applied to model the Lipo-
zyme RM IM-catalyzed transesterification of high oleic
sunflower oil with mixed ethyl palmitate and ethyl stearate,
with four reaction factors, i.e. Te, RT, SR, and WC.
Table 1 lists the contents of total SMUT, POS, and DAG in
Residual
Lack of fit
Pure error
0.4931
df degree of freedom, SS sum of squares, MS mean square, SMUT
symmetric monounsaturated triacylglycerols, POS 1-palmitoyl-2-
oleoyl-3-stearoyl-rac-glycerol, DAG diacylglycerols, sn-2 P ? S pal-
mitoyl and stearoyl residues found at the sn-2 position of
triacylglycerols
1
23

J Am Oil Chem Soc
statistically significant (p \ 0.0001) (Table 2). The good-
ness-of-fit of the regression models was evaluated by a test
for lack of fit from ANOVA and by the coefficients of
determination of the models. The models had no lack of fit
at the 95 % level of significance. Parity plot between
observed values and values predicted by the models for
total SMUT, POS, DAG, and sn-2 P ? S contents and
indicate that the models can predict 84.0, 81.5, 66.4, and
75.9 % of the variances in total SMUT, POS, DAG, and
sn-2 P ? S contents, respectively. For each model, the
2
2
adjusted R value was close to the Q value, thereby
indicating that there was no block effect with the model.
Therefore, these results suggest that we successfully
established well-fitting models of the relationships between
the four reaction factors investigated and each response
variable.
2
2
2
coefficients of determination, i.e., R , adjusted R , and Q
2
are shown in Fig. 1. R values of the regression models
indicate that they can explain 94.1, 93.3, 80.2 and 93.5 %
of the variances in total SMUT, POS, DAG, and sn-2
P ? S contents, respectively. According to the value of
Model Validation
2
2
adjusted R , which is a modification of R that adjusts for
the number of variables in a model, 91.3 % (for total
SMUT content), 90.2 % (for POS content), 72.5 % (for
DAG content), and 88.4 % (for sn-2 P ? S content) of the
In order to examine the adequacy of the four models
established, observed responses were compared to the
predicted values given by the models, with ten additional
experimental sets selected from the given ranges of reac-
tion parameters (Table 3). The percentage prediction errors
2
variances can be explained by each model. The Q values
Fig. 1 Parity plots showing the relationships between observed
values and values predicted by the regression models. Numbers inside
the graph represent experimental numbers shown in Table 1. SMUT
symmetric monounsaturated triacylglycerols, POS 1-palmitoyl-2-
oleoyl-3-stearoyl-rac-glycerol, DAG diacylglycerols, sn-2 P ? S pal-
mitoyl and stearoyl residues found at the sn-2 position of
triacylglycerols
1
23

J Am Oil Chem Soc
on observed and predicted total SMUT and POS contents
were in the range of 0.5–9.8 and 0.6–11.2 %, respectively,
indicating that the models showed great predictability.
Whereas, the predictability of the models for DAG and
sn-2 P ? S contents was relatively low since considerable
discrepancies between observed and predicted values were
found in some experimental sets. Overall correlation
between observed and predicted values for the models for
2
2
total SMUT (R = 0.960), POS (R = 0.962), and sn-2
2
P ? S contents (R = 0.891) showed good linearity,
Table 3 Observed values and values predicted by the regression models for the responses in ten additional experimental sets
Exp.
no.
Factor
Response
Te RT
(°C) (min)
SR WC
(mg/
kg)
Total SMUT
content (w/w%)
POS content
(w/w%)
DAG content
(w/w%)
sn-2 P ? S content
(w/w%)
Observed
Predicted %PE Observed
Predicted %PE Observed
Predicted %PE Observed
Predicted %PE
a
1
2
3
4
5
6
7
8
9
40
60
40
57
40
55
43
58
53
42
15
50
52
17
78
30
31
30
70
45
3.0 500
3.0 500
4.8 850
4.8 850
5.8 580
5.8 580
8.5 900
7.8 320
3.1 960
3.1 960
24.4 ± 9.6
33.2 ± 0.2
41.4 ± 0.6
38.0 ± 1.5
43.5 ± 1.8
45.3 ± 0.9
51.9 ± 1.9
52.1 ± 1.2
32.4 ± 0.3
32.5 ± 0.5
25.4
34.8
40.1
38.2
45.0
42.9
46.8
47.3
32.8
32.3
4.1
4.8
3.1
0.5
3.4
5.3
9.8
9.2
1.2
0.6
12.1 ± 0.2 12.7
16.5 ± 0.2 17.0
20.5 ± 0.4 19.2
18.9 ± 0.8 18.7
21.6 ± 0.9 21.8
22.5 ± 0.5 20.7
25.8 ± 1.0 22.9
25.9 ± 0.7 23.1
16.1 ± 0.2 16.0
16.1 ± 0.2 15.7
5.0
3.0 11.1 ± 1.3
6.3 15.5 ± 0.3 12.8
1.1 10.4 ± 0.0 8.9
0.9 17.1 ± 2.0 12.0
8.0 11.6 ± 4.4 8.2
11.2 11.6 ± 3.0 14.2
10.8 7.5 ± 1.0 6.8
7.7 ± 3.0 10.0
29.9
29.7
17.4
2.9 ± 0.6 2.7
8.3 ± 0.9 6.2
5.7 ± 1.1 4.7
6.9
25.3
17.5
15.6
23.5
41.2
27.6
43.5
30.7
34.1
7.8
14.4 4.5 ± 0.0
3.8
29.8 10.2 ± 2.9 7.8
29.3
22.4
9.3
8.5 ± 4.1 5.0
5.8 ± 0.1 4.2
6.9 ± 2.3 3.9
0.6 15.5 ± 0.6 12.2
2.5 14.3 ± 0.9 13.1
21.3 11.4 ± 0.1 7.9
8.4 4.1 ± 0.5 2.7
1
0
Te temperature, RT residence time, SR substrate molar ratio (high oleic sunflower oil to 2:3 (mol/mol) mixture of ethyl palmitate and ethyl stearate), WC water content, SMUT
symmetric monounsaturated triacylglycerols, POS 1-palmitoyl-2-oleoyl-3-stearoyl-rac-glycerol, DAG diacylglycerols, sn-2 P ? S palmitoyl and stearoyl residues found at sn-2
position of triacylglycerols, %PE percentage prediction error
a
Mean ± standard deviation (n = 2)
Table 4 Regression coefficients of the quadratic polynomial models for the Lipozyme RM IM-catalyzed transesterification of high oleic
sunflower oil with a mixture of ethyl palmitate and ethyl stearate
Effect
Total SMUT content (w/w%)
POS content (w/w%)
DAG content (w/w%)
sn-2 P ? S content (w/w%)
Coefficient
45.07
p value
Coefficient
21.63
p value
Coefficient
10.16
p value
Coefficient
6.28
p value
Intercept
Linear
Te
\0.0001
\0.0001
\0.0001
\0.0001
a
a
1.03
2.77
0.0938
0.0004
0.51
1.28
0.1060
0.0010
-1.03
1.43
0.0333
0.0072
1.22
3.21
0.57
0.45
0.0004
\0.0001
0.0509
RT
SR
8.33
\0.0001
0.3026
4.11
\0.0001
0.2846
1.16
0.0176
a
a
WC
-0.62
-0.33
2.84
\0.0001
0.1096
Quadratic
Te*Te
RT*RT
SR*SR
WC*WC
Interaction
Te*RT
Te*SR
Te*WC
RT*SR
RT*WC
SR*WC
-2.73
-3.40
0.0144
0.0019
-0.99
-1.30
0.0709
0.0141
0.89
-0.77
-0.77
0.0821
0.0916
0.0916
1.27
0.85
1.18
0.0792
0.1053
0.0299
a
a
-0.85
-1.95
0.2081
0.0080
-0.44
-0.96
0.2060
0.0101
0.91
0.47
0.0082
0.1313
a
a
a
0.45
0.64
0.1512
0.0494
Te temperature, RT residence time, SR substrate molar ratio (high oleic sunflower oil to 2:3 (mol/mol) mixture of ethyl palmitate and ethyl
stearate), WC water content
a
Not significant at p \ 0.1
1
23

J Am Oil Chem Soc
2
whereas the R value of the model for DAG content was
.438, which is relatively low compared to those of the
significantly (p \ 0.05) and negatively affected DAG
content, whereas it significantly (p \ 0.01) and positively
affected sn-2 P ? S content. This result was in accordance
with Xu [11] who indicated that elevated temperature
promoted acyl migration during a lipase-catalyzed reac-
tion. Residence time and substrate molar ratio were sig-
nificantly and positively correlated with all four responses,
i.e., total SMUT, POS, DAG, and sn-2 P ? S contents,
respectively. Control of the level of water in the lipase-
catalyzed reaction system is crucial for maintaining the
catalytic activity of lipase and determining the orientation
of the reaction process (e.g., hydrolysis vs. esterification)
and the composition of the final reaction products [12, 18–
20]. Water content had no significant effects on total
SMUT, POS, and sn-2 P ? S contents, whereas it
0
others.
Effects of Reaction Parameters
The significant (p \ 0.1) coefficients of the regression
model equations in terms of coded factors are listed with
their statistical significances in Table 4. Some linear and
interaction terms were not removed by backward elimina-
tion in the process of model fitting described before to
enhance the fitness of models, although they were not
significant at the 10 % level. There were no effects of
temperature on total SMUT and POS contents, respec-
tively, at the 5 % significance level. Temperature
Fig. 2 Prediction plots for response variables by the effects of
reaction parameters. The coded values of all reaction parameters were
zero (i.e., 50 °C for temperature, 47.5 min for residence time, 6 for
substrate molar ratio, 650 mg/kg for water content) except those of a
reaction parameter represented on the X axis. SMUT symmetric
monounsaturated triacylglycerols, POS 1-palmitoyl-2-oleoyl-3-stear-
oyl-rac-glycerol, DAG diacylglycerols, sn-2 P ? S palmitoyl and
stearoyl residues found at the sn-2 position of triacylglycerols
1
23

J Am Oil Chem Soc
significantly (p \ 0.01) and positively affected DAG
content. These results indicate that the water content of the
substrates investigated in the present study is below a
critical level where acyl migration is promoted, but is
sufficient to induce hydrolysis during the lipase-catalyzed
transesterification. The quadratic term of residence time
had significant (p \ 0.1) negative effects on total SMUT
and POS contents, whereas it was significantly (p \ 0.1)
and positively correlated with sn-2 P ? S content. Simi-
larly, the quadratic term of substrate molar ratio signifi-
cantly (p \ 0.05) and negatively affected total SMUT and
POS contents, but showed significant (p \ 0.1) positive
impacts on DAG content. These results are in agreement
with the results shown in the prediction plots for each
response level by the effect of individual reaction param-
eter (Fig. 2). The rate of increase in total SMUT and POS
contents was expected to decline gradually as residence
time or substrate molar ratio increased within the given
ranges. In contrast, the rate of increase in DAG and sn-2
P ? S contents was predicted to increase gradually with
the increase in residence time and substrate molar ratio,
respectively. The 3D response surface plots obtained when
residence time and substrate molar ratio were modified but
temperature and water content were maintained at the
center points also showed that the responses of total SMUT
and POS contents were represented by convex surfaces,
respectively, whereas the response surface of DAG content
had a concave form (data not shown).
Optimization of Reaction Parameters
Optimal reaction conditions for maximizing total SMUT
and POS contents while minimizing DAG and sn-2 P ? S
contents were generated using the optimization function of
the Design expert 8.0 software. The established optimal
conditions were: temperature, 59.9–60.0 °C; residence
time, 25.9–31.2 min; substrate molar ratio, 8.4–8.7; and
water content, 300 mg/kg. The total SMUT, POS, DAG,
and sn-2 P ? S contents were predicted to be 47.7–48.8,
2
3.5–23.9, 6.9–7.3, and 3.1–3.6 %, respectively, under
these conditions. These results suggest that temperature
should be kept at the highest level and water content
should be minimized within the given ranges to increase
the contents of SMUT species, while suppressing the for-
mation of DAG, although elevated temperature promotes
acyl migration. Reducing the temperature, residence time,
and substrate molar ratio is necessary to develop an
industrially applicable and cost-effective process for pro-
ducing CBE. The possible minimum levels of the reaction
factors while maximizing total SMUT and POS contents
and minimizing DAG and sn-2 P ? S contents were
40.0°°C for temperature, 18.2–20.5 min for residence time,
and 5.3–5.5 for substrate molar ratio. Under these
1
23

J Am Oil Chem Soc
Table 6 Contents of symmetric monounsaturated triacylglycerols
and their positional isomers (w/w%)
respectively. These results were in reasonable agreement
with the values (7.1 % for experiment 1 and 9.4 % for
experiment 2) that were estimated using the following
equation: sn-2 P ? S content/total SMUT content 9 100,
from Table 5.
Triacylglycerols
Exp. no. 1
Exp. no. 2
a
POP ? POS ? SOS
49.1 ± 1.4
6.1 ± 1.8
31.3 ± 0.9
4.2 ± 1.3
PPO ? PSO ? SPO ? SSO
Experiment 1 was performed with 5 g of Lipozyme RM IM, at
optimal levels of reaction factors (Te, 60 °C; RT, 28.5 min; SR, 8.5;
WC, 300 mg/kg)
Conclusion
The modeling of Lipozyme RM IM-catalyzed transesteri-
fication of high oleic sunflower oil with a mixture of ethyl
palmitate and ethyl stearate in a small-scale continuous
packed bed reactor for producing CBE with a weight ratio
of POS to total SMUT that is similar to that of cocoa butter
was successfully performed by RSM. The effects of tem-
perature, residence time, substrate molar ratio, and water
content on total SMUT and POS contents in the reaction
products as well as the levels of DAG formation and acyl
migration that deteriorate the quality of CBE were eluci-
dated within their given ranges. Optimal reaction condi-
tions for maximizing total SMUT and POS contents, while
suppressing the phenomena of DAG formation and acyl
migration, were then obtained using the established
regression models. Successful scale-up of the reaction was
also achieved under the optimal conditions. A considerable
amount of ethyl palmitate and ethyl stearate remained
unreacted in the reaction products and ethyl oleate was also
formed as the main co-products of the reaction. Therefore,
further investigations on the purification (e.g., distillation
process) and characterization (e.g., physical and textural
properties) of the CBE would be required for their possible
usage in chocolate products.
Experiment 2 was performed with 5 g of Lipozyme RM IM, at
minimum levels of reaction factors (Te, 40 °C; RT, 19.5 min; SR,
5
.4; WC, 300 mg/kg)
Te temperature, RT residence time, SR substrate molar ratio (high
oleic sunflower oil to 2:3 (mol/mol) mixture of ethyl palmitate and
ethyl stearate), WC water content, P palmitic, S stearic, O oleic acids
a
Mean ± standard deviation (n = 2)
conditions, the maximum total SMUT and POS contents
were predicted to be 35.3–36.1 and 17.3–17.6 %, respec-
tively, and the minimum DAG and sn-2 P ? S contents
were predicted to be 7.3–7.4 and 3.4 %, respectively.
Additional reactions with two different amounts of Lipo-
zyme RM IM, 0.45 g (small-scale reaction) and 5 g (larger
scale reaction) were performed at two conditions selected
from the optimal levels (experiment 1) and the minimum
levels (experiment 2) of the reaction factors described
above, respectively (Table 5). The percentage prediction
error values indicate that observed responses were in rea-
sonable agreement with the predicted values given by the
models. Good reproducibility for total SMUT, POS, and
sn-2 P ? S contents was observed, with the relative stan-
dard deviation being less than 10 %. These results suggest
that the process successfully increased 11-fold. The weight
ratios of POS to total SMUT in the reaction products were
Acknowledgments This research was supported by the Basic Sci-
ence Research Program through the National Research Foundation of
Korea (NRF) funded by the Ministry of Education, Science and
Technology (2011-0005361).
0
.48 (small-scale reaction) and 0.50 (larger scale reaction),
which were similar to those of cocoa butter, under each set
of conditions.
SMUT and Positional Isomer Profile
References
In the modeling and optimization process, triacylglycerol
composition of the reaction products was analyzed using a
GC system in which isomeric triacylgycerols cannot be
separated. Therefore, total SMUT content presented in
Tables 1, 2, 3, 4, 5 or Figs. 1, 2 includes the content of
their positional isomers, i.e., PPO, PSO, SPO, and SSO.
Likewise, the POS content given in the Tables and Figures
is the sum of POS, PSO and SPO contents. Thus, the
contents of total SMUT and their positional isomers in the
products obtained by larger scale reaction were analyzed
using a silver-ion HPLC system, and the results are given
in Table 6. Positional isomers of SMUT species occupy
1
2
. Shukla VKS (2005) Cocoa butter, cocoa butter equivalents, and
cocoa butter substitutes. In: Akoh CC (ed) Handbook of func-
tional lipids. CRC Press, New York, pp 279–307
. Yamada K, Ibuki M, McBrayer T (2005) Cocoa butter, cocoa
butter equivalents, and cocoa butter replacers. In: Lai OM, Akoh
CC (eds) Healthful lipids. AOCS Press, Champaign, pp 642–664
3. Lipp M, Simoneau C, Ulberth F, Anklam E, Crews C, Brereton P,
de Greyt W, Schwack W, Wiedmaier C (2001) Composition of
genuine cocoa butter and cocoa butter equivalents. J Food
Compost Anal 14:399–408
4
. Undurraga D, Markovits A, Erazo S (2001) Cocoa butter equiv-
alent through enzymic interesterification of palm oil midfraction.
Process Biochem 36:933–939
. Abigor RD, Marmer WN, Foglia TA, Jones KC, DiCiccio RJ,
Ashby R, Uadia PO (2003) Production of cocoa butter-like fats
by the lipase-catalyzed interesterification of palm oil and
hydrogenated soybean oil. J Am Oil Chem Soc 80:1193–1196
5
1
1.1 % (for experiment 1) and 11.8 % (for experiment 2)
of total SMUT contents that were analyzed by GC,
1
23

J Am Oil Chem Soc
6
. Pinyaphong P, Phutrakul S (2009) Synthesis of cocoa butter
equivalent from palm oil by Carica papaya lipase-catalyzed in-
teresterification. Chiang Mai J Sci 36:359–368
. Tchobo FP, Piombo G, Pina M, Soumanou MM, Villeneuve P,
Sohounhloue DCK (2009) Enzymatic synthesis of cocoa butter
equivalent through transesterification of Pentadesma butyracea
butter. J Food Lipids 16:605–617
15. Luddy FE, Barford RA, Herb SF, Magidman P, Riemenschneider
RW (1964) Pancreatic lipase hydrolysis of triglycerides by a
semimicro technique. J Am Oil Chem Soc 41:693–696
16. Kang KK, Kim S, Kim IH, Lee C, Kim BH (2013) Selective
enrichment of symmetric monounsaturated triacylglycerols from
palm stearin by double solvent fractionation. LWT Food Sci
Technol 51:242–252
7
8
9
. Carrin ME, Crapiste GH (2008) Enzymatic acidolysis of sunflower
oil with a palmitic-stearic acid mixture. J Food Eng 84:243–249
. Ciftci ON, Fadiloglu S, Gogus F (2009) Utilization of olive-
pomace oil for enzymatic production of cocoa butter-like fat.
J Am Oil Chem Soc 86:119–125
17. Gunstone FD, Harwood JL (2007) Occurrence and characterisa-
tion of oils and fats. In: Gunstone FD, Harwood JL, Dijkstra AJ
(eds) The lipid handbook with CD-ROM, 3rd edn. CRC Press,
Boca Raton, pp 37–142
18. Lee SB, Kim K (1995) Effect of water activity on enzyme
hydration and enzyme reaction rate in organic solvents. J Ferment
Bioeng 79:473–478
19. Xu X, Balchen S, Høy CE, Adler-Nissen J (1998) Pilot batch
production of specific-structured lipids by lipase-catalyzed in-
teresterification: preliminary study on incorporation and acyl
migration. J Am Oil Chem Soc 75:301–308
20. Goderis HL, Ampe G, Feyten MP, Fouw e´ BL, Guffens WM, Van
Cauwenbergh SM, Tobback PP (2013) Lipase-catalyzed ester
exchange reactions in organic media with controlled humidity.
Biotechnol Bioeng 30:258–266
1
0. Wang HX, Wu H, Ho CT, Weng CX (2006) Cocoa butter
equivalent from enzymatic interesterification of tea seed oil and
fatty acid methyl esters. Food Chem 97:661–665
1
1
1. Xu X (2000) Enzymatic production of structured lipids: process
reaction and acyl migration. INFORM 11(October):1121–1131
2. Mu H, Xu X, Høy CE (1998) Production of specific-structured
triacylglycerols by lipase-catalyzed interesterification in a labora-
tory-scale continuous reactor. J Am Oil Chem Soc 75:1187–1193
3. Tietz RA, Hartel RW (2000) Effects of minor lipids on crystal-
lization of milk fat-cocoa butter blends and bloom formation in
chocolate. J Am Oil Chem Soc 77:763–771
1
1
4. Suppalakpanya K, Ratanawilai SB, Tongurai C (2010) Produc-
tion of ethyl ester from crude palm oil by two-step reaction with a
microwave system. Fuel 89:2140–2144
1
23