ww.sciencedirect.com

b i om a s s a n d b i o e n e r g y 6 1 ( 2 0 1 4 ) 1 7 9e1 8 6

Available online at w

ScienceDirect

http: / /www.elsevier .com/locate/biombioe

Optimization of lipase-catalyzed synthesis ofdiglycerol monooleate by response surfacemethodology

Noureddin El-Boulifi, Jose Aracil, Mercedes Martınez*

Chemical Engineering Department, Chemistry Faculty, Complutense University, 28040 Madrid, Spain

a r t i c l e i n f o

Article history:

Received 23 May 2013

Received in revised form

10 December 2013

Accepted 15 December 2013

Available online 17 January 2014

Keywords:

Diglycerol

Diglycerol monooleate

Novozym 435

Response surface methodology

Surfactant

* Corresponding author. Tel./fax: þ34 91 394E-mail address: mmr1@quim.ucm.es (M.

0961-9534/$ e see front matter ª 2013 Elsevhttp://dx.doi.org/10.1016/j.biombioe.2013.12.

a b s t r a c t

The optimization of the lipase-catalyzed esterification of pure diglycerol (3-(2,3-

dihydroxypropoxy)propane-1,2-diol) with pure oleic acid to produce pure diglycerol mon-

ooleate (E475) which is a non-ionic surfactants, was performed. Six immobilized lipases

were tested and the best oleic acid conversion was attained with Novozym 435 from

Candida antarctica which was selected to optimize the reaction conditions by response

surface methodology (RSM). Well-fitting quadratic polynomial regression model for acid

conversion was established with regard to temperature (65 �Ce75 �C) and catalyst con-

centration (mass fraction of 1e5%). The two factors investigated positively affected acid

conversion, with catalyst concentration having the greatest effect. The regression equation

obtained by central composite design of RSM predicted optimal reaction conditions of 77 �C

and 5.8%. Under these optimal conditions the model obtained in this work has been tested

in scale-up experiment, and the resulting acid conversion was 93.9% with an accuracy of

97.4%. Within the experimental range studied the results model give good agreement with

the experimental data.

ª 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Polyglycerol esters of fatty acids (PGE, E475) are non-ionic

surfactants that have been used during the last years as

emulsifiers in food and personal care products [1,2]. More

recently, new industrial applications based on PGE have been

developed, includes their utilization as antifogging and anti-

static additives, lubricants and plasticizers [1,3]. The PGE

are an amphipathic molecules with hydrophilic moiety (hy-

droxyl groups) attached to a hydrophobic backbone (alkyl

chain). These two parts provide a compound with interfacial

activity and give rise to a wide range of surface chemistry

functions.

4167.Martınez).ier Ltd. All rights reserved009

Glycerol is the main by-product of biodiesel production. As

the biodiesel production is increasing exponentially, the crude

glycerol generated from the transesterification of oils has also

been generated in a large quantity. The valorization of glycerol

as by-product increases the economic sustainability of the

biodiesel industry. Within this context, among the wide

application of glycerol in food, the use of glycerol for the

production of non-ionic surfactants like glycerol esters of fatty

acids, can be a good way forward to development of surfac-

tants products entirely derived from renewable resource [4,5].

However, glycerol itself is not a good primary constituent of

the hydrophilic part of the surfactant and polyglycerols are

needed to increase the hydrophilicity and to adjust the

hydrophilic-hydrophobic balance of the products. Since the

.

Scheme 1 e .TIF. Diglycerol monooleate ester synthesis.

b i om a s s a n d b i o e n e r g y 6 1 ( 2 0 1 4 ) 1 7 9e1 8 6180

phase behavior of diglycerol and glycerol esters of fatty acid,

shown that the presence of strong intramolecular H-bonds

within the diglycerol headgroup increases the polar part vol-

ume and thus causes a swelling of the lamellar phase [6]. PGE

have an unexpectedly low solubility in water; they are of in-

terest as stable carriers for drugs, cosmetics, or specific food

ingredients. Among the commercial mixtures of PGE, the

main component is diglycerol ester [7] and pure diglycerol

esters have been shown to be efficient surfactants [8], thereby

creating a demand for more specific synthesis procedures of

PGE.

Current processes for production of fatty acid esters of

polyglycerols are based on direct esterification of the poly-

glycerol with fatty acids [9], transesterification of the poly-

glycerol with a triglyceride or a fatty acid methyl ester [10] or

addition polymerization of glycidol to a fatty acid or to a fatty

acid monoglyceride [11]. Esterification method is the best

suited for the production of designer polyglycerol ester

because the desired free fatty acid and polyglycerol can easily

be selected prior to polyglycerol ester formation.

Conventional production of polyglycerol esters by esterifi-

cation commonly involves the use either alkali [9,12] or acid

catalysts [10]. But, besides the environmental problem, these

catalysts favor side-reactions from the degradation of the

fatty acid (oxidation and dimerization) or from the poly-

glycerol (dehydration into acrolein and oxidation) [13].

A few decades ago, the employment of lipases as bio-

catalysts for esterification reactions has emerged as a poten-

tial route to replace the conventional chemical process [14,15].

The main reason is that they are employed as active and se-

lective catalysts in a variety of reactions, with fewer envi-

ronmental problems, and in the case of immobilized lipases,

they can be reused.

So far, to our knowledge, the use of pure diglycerol to

produce diglycerol fatty acid mono-esters has received little

Table 1 e Properties of immobilized lipases used in diglycerol

Microorganisms Lipase Support

Rhizomucor miehei Lipozyme IM-20 Anion exchange res

Lipozyme IM Anion exchange res

Lipozyme IM-50 Anion exchange res

Candida antarctica Novozym 435 Acrylic resin

Novozym SP 435-A Acrylic resin

Novozym SP 435-L Acrylic resin

attention. The objective of our study was to model the

lipase-catalyzed esterification of pure diglycerol with oleic

acid to produce pure diglycerol monooleate ester (Scheme 1)

in a solvent-free system. Six different commercially avail-

able lipases were evaluated for their catalytic activity in the

reaction. For selected lipase (Novozym 435), the modeling

and optimization of the reaction by RSM was performed

with respect to temperature and catalyst concentration as

independent variables and acid conversion as a response

variable, and then the optimal reaction conditions were

proposed.

2. Materials and methods

2.1. Materials

Diglycerol (purity > 98%) was supplied by Solvay Quımica S.L.

(Spain) and oleic acid (purity > 98%) was supplied by Henkel

Iberica. The six immobilized thermostable lipases were pro-

vided by Novozymes A/S (Bagsvœrd, Denmark). Table 1 sum-

marizes some of the properties of reported lipases. All

solvents used for the chromatographic analysis were of high-

performance liquid chromatography (HPLC) grade, supplied

by SigmaeAldrich.

2.2. Equipment

Reactions were performed in a solvent-free system using a

batch stirred reactor of 500 cm3 volume, under fixed condi-

tions of pressure and temperature. Pressure, stirring speed

and temperature controllers were provided. The propeller

used was marine-type and the speed was set at 62.8 rad s�1.

The desired working pressure (7.998 kPa) was maintained by a

vacuum pump. This permitted ready elimination of water

from the system in range of temperature studied, without

significant variations of viscosity of the liquid phase or reac-

tion volume. The reaction temperature was achieved

immersing the reactor into a thermostatic bath with an elec-

trical device connected to a PID controller which allows a

temperature control of �0.1 �C.

2.3. Lipase-catalyzed esterification

Oleic acid (45 g, 160 mmol) and diglycerol (72 g, 433 mmol) were

added to the reactor and the stirring was started. When the

desired temperature was reached, the catalyst was added and

the vacuum pump was turned on in order to drive the equi-

librium towards the diglycerol monooleate ester synthesis.

monooleate production.

Surface area (dam2 kg�1) Pore diameter (nm)

in 3.6 14.1

in 6.0 17.6

in 6.2 24.3

2.9 29.2

6.7 21.0

9.5 17.9

40

50

6070

80

90

vers

ion(

%)

b i om a s s a n d b i o e n e r g y 6 1 ( 2 0 1 4 ) 1 7 9e1 8 6 181

The reaction was stirred during 240 min. Samples were taken

at regular intervals (at 30, 60, 120 and 240 min) and analyzed

by gas chromatography-mass spectrometry (GC-MS). During

the experiments, the acid:alcohol molar ratio (1:1), pressure

(7.998 kPa) and agitation speed (62.8 rad s�1) were maintained

constant.

0

10

2030

(A) (B) (C) (D) (E) (F)Lipases

wac

id c

on

Fig. 1 eMS. Acid conversionmediated by various enzymes:

Novozym 435 (A), Lipozyme IM (B), Novozym SP 435-L (C),

Lipozyme IM-50 (D), Novozym SP 435-A (E) and Lipozyme

IM-20 (F) (all performed at a molar ratio of 1:1 (diglycerol to

2.4. Purification and characterization of diglycerolmonooleate

At the optimal condition of the reaction, the diglycerol mon-

ooleate ester was isolated from the reaction mixture by silica

gel 60 (with a mesh size: 40 mme63 mm) column chromatog-

raphy using chloroform/methanol (90:10, v/v) as eluents. The

purified diglycerol monooleate was set aside for analysis by

GC-MS, infra-red (IR) spectrometry and nuclear magnetic

resonance (1H and 13C NMR) spectrometry.

oleic acid), temperature of 75 �C and mass fraction of 3% of

catalyst).

2.5. Analytical method

2.5.1. GC-MS analysisThe analysis was performed using a HewlettePackard 5890 II

GC, equipped with a fused-silica capillary column (OV-1,

12 m � 310 mm i.d., 17 nm film) and a mass spectrometer

5971A. Column temperature was initially 60 �C for 5 min, then

gradually increased to 170 �C at 5 K min�1, and finally

increased to 270 �C at 10 K min�1. For GC-MS detection an

electron ionization system was used with an ionization en-

ergy of 70 eV. A HewlettePackard 3396A integrator was con-

nected to the chromatograph to process and collect the output

signal. The detector system was an FID type at 270 �C. Theinjection system is a split/splitless. The carrier gaswas helium

and the flow rate 10 mm3 s�1. Quantification was based on

external calibration using standard solutions of octyl octa-

noate, over the range of 20 mg mL�1 to 60 mg mL�1 and reaction

Table 2 e Experimental range and levels of independent variaresults.

Factors

þ1.414 þ1

X1: Temperature (�C) 77 75

X2: wcatalyst concentration (%) 5.8 5

Experimental matrix and experimental results

Run number Uncoded factor values

X1 (�C) X2 (%) At 30

1 65 1 25.

2 75 1 39.

3 65 5 42.

4 75 5 52.

5 70 3 36.

6 70 3 59.

7 70 3 40.

8 70 3 48.

9 77 3 52.

10 63 3 22.

11 70 5.8 49.

12 70 0.2 33.

aliquots were always prepared in CS2 to a final concentration

of 300 mg mL�1.

2.5.2. NMR analysis1H and 13C NMR spectra were measured on a Brucker Avance-

DPX-300 NMR spectrometer, in CDCl3 solution [6]. Chemical

shifts (d) were expressed in ppm relative to tetramethylsilane

(TMS), and coupling constants J were given in hertz (Hz).

2.5.3. IR analysisIR spectra were recorded as a thin film using NaCl plate on a

PERKIN-ELMER 781 spectrophotometer (nmax in cm�1). IR

(neat): 3429, 2935, 2849, 1733 cm�1.

bles together with experimental matrix and experimental

Level of variables

0 �1 �1.414

70 65 63

3 1 0.2

wacid conversion (%)

min At 60 min At 120 min At 240 min

0 33.0 40.5 37.9

5 51.4 56.3 59.3

3 58.4 74.3 71.6

3 63.2 80.5 91.5

5 49.6 52.5 65.9

7 62.4 68.9 71.5

1 58.6 65.4 63.6

9 58.7 71.2 66.2

1 61.2 70.4 74.9

7 40.1 48.9 51.2

9 75.9 85.1 90.1

9 42.0 45.6 46.3

Table 3 e Statistical analysisemain effects andinteraction observed as a result of the 22 experimentaldesign (experiments 1e8 in Table 1).

Response: w acid conversion (%)

Main effects and interactions:

At 30 min At 60 min At 120 min At 240 min

IC ¼ 15.1 IC ¼ 18.6 IC ¼ 29.0 IC ¼ 33.0

IT ¼ 12.3 IT ¼ 11.6 IT ¼ 13.6 IT ¼ 20.7

ITC ¼ �2.3 ITC ¼ �6.8 ITC ¼ �4.8 ITC ¼ �0.8

Significance test (Confidence level: 95%)

Mean responses

Y ¼ 26:7 Y ¼ 51:5 Y ¼ 62:9 Y ¼ 65:1

t ¼ 3.182

s ¼ 10.30

t ¼ 3.182 s ¼ 5.45 t ¼ 3.182

s ¼ 8.33

t ¼ 3.182

s ¼ 3.34

Confidence interval:

�16.34 �8.61 �13.3 �5.30

Significant variables

e T, C T, C T, C

Curvature

C ¼ 6.54 C ¼ 5.84 C ¼ 1.62 C ¼ 1.71

T, temperature; C, catalyst concentration; I, main effect or inter-

action; s, standard deviation; t, Student’s t-test.

b i om a s s a n d b i o e n e r g y 6 1 ( 2 0 1 4 ) 1 7 9e1 8 6182

2.6. Statistical experimental design

The synthesis of diglycerol monooleate ester using immobi-

lized lipase as catalyst was carried out using a factorial design

of experiments. The experimental design applied to this study

was a full 22 factorial design (two factors each at two levels)

with four central points were added, to evaluate the experi-

ments errors. The factorial design was performed to study the

effect of the variables on the process and the interaction be-

tween them. The RSM was applied to optimize the process.

The response selected was the previously defined, acid con-

version. It is relative to the initial amount of oleic acid. The

two factors considered to model and optimize the lipase-

catalyzed esterification were temperature (65 �Ce75 �C) andcatalyst concentration (mass fraction of 1e5%). The ranges of

these two parameters were chosen based on results obtained

in our laboratory in preliminary kinetic study of diglycerol

monooleate synthesis [16] and are given in Table 2. The

impeller speed was set to 62.8 rad s�1, working pressure was

hold at 7.998 kPa and the molar ratio of substrates (diglycerol

to oleic acid) was fixed at 1:1 to avoid any excess of the re-

actants and improve the purification process.

3. Results and discussion

3.1. Enzyme screening for optimization

The effects of six immobilized lipases on esterification of pure

diglycerol with oleic acid were investigated, and the results

Table 4 e The predictive response equations relating the a

At 30 min Y ¼ 46.30 þ 8.26X1 þ 6.59X2 � 4.4

At 60 min Y ¼ 57.32 þ 6.63X1 þ 10.64X2 � 4

At 120 min Y ¼ 64.5 þ 6.55X1 þ 14.23X2 � 2.3

At 240 min Y ¼ 66.80 þ 9.35X1 þ 15.98X2 � 2

are represented in Fig. 1. Novozym 435 from Candida antarctica

gave the highest acid conversion (73%). The acid conversion

for Lipozyme IM from Rhizomucor meihei and Novozym SP 435-

L from C. antarctica were 69% and 65%, respectively. However,

the rest of the other ones show little activity (less than 60% of

acid conversion) for the esterification of pure diglycerol with

oleic acid. Comparing the acid conversions (56%, 65% and 73%)

obtained with immobilized lipases Novozym SP 435-A

(2.9 dam2 kg�1), Novozym SP 435-L (6.7 dam2 kg�1) and Novo-

zym 435 (9.5 dam2 kg�1) respectively, it can be seen that with

increasing surface area of support, the acid conversion is

higher. The surface area of a support is of vital importance for

its performance. A large surface area has the positive effect of

spreading out the enzyme and thereby making the active site

accessible [17]. However, although Lipozyme IM and Lipozyme

IM-50 have similar surface area of support (6 dam2 kg�1 and

6.1 dam2 kg�1), the acid conversion using Lipozyme IM is

higher than when Lipozyme IM-50 is used. This would indi-

cate that the acid conversionmaybe is affected by other factor

like, pore diameter. Since using Lipozyme IM-50 with a pore

diameter of 24.3 nm, the acid conversion is 57.6%, lower than

when Lipozyme IM (pore diameter 17.6 nm) is used. This result

shows that the acid conversion increases as the pore diameter

decreases. Similar conclusion is drawn from the analysis of

results for lipases from C. antarctica. The highest acid con-

version (73%) is obtained with Novozym 435 with the smallest

pore diameter. This means to an increase in acid conversion

as the pore diameter decreases. Consequently, Novozym 435

was selected as the more suitable lipase to optimize the re-

action conditions for the production of pure diglycerol mon-

ooleate ester by esterification of diglycerol with oleic acid.

3.2. Optimization by 2-level factorial design

The first step in the process of seeking optimum conditions

was to identify the input variables that have the greatest in-

fluence on the acid conversion. The standard experimental

matrix for the factorial design is shown in Table 2 (Experi-

ments 1e8). Columns 2 and 3 represent the factor levels on

nature scale and columns 4, 5, 6 and 7 represent the acid

conversion, at 30, 60, 120 and 240 min, respectively. The

analysis of main effects and their interactions on acid con-

version, together with the test of statistical significance, a

two-sided t-test with a confidence level of 95% are given in

Table 3. Statistical analysis shows that the two factors, tem-

perature and catalyst concentration has a significant effect,

since the main effect values were higher than the corre-

sponding confidence interval Table 3. The catalyst concen-

tration (mass fraction of 1e5%) is the most important factor

throughout the esterification process for the experimental

range studied. This factor has a positive influence on acid

conversion. It can be observed the low influence of catalyst

cid conversion to process parameters.

7X21 � 1.12X1X2 � 2.17X2

2 (r ¼ 0.860) (2)

.16X21 � 3.40X1X2 � 0.012X2

2 (r ¼ 0.957) (3)

2X21 � 2.40X1X2 þ 0.52X2

2 (r ¼ 0.950) (4)

.01X21 � 0.37X1X2 þ 0.56X2

2 (r ¼ 0.987) (5)

Fig. 2 e TIF. Response surface plot and contour plots of acid conversion as a function of temperature and catalyst

concentration: at 30 min (a), at 60 min (b), at 120 min (c) and at 240 min (d).

b i om a s s a n d b i o e n e r g y 6 1 ( 2 0 1 4 ) 1 7 9e1 8 6 183

Table 5 e The analysis of variance (ANOVA) for responsesurface quadratic polynomial model at 240 min.

Source SS df MS F-value P-value

Regression 2742.66 2 1371.33 161.7 <0.0001

Catalyst

concentration (X2)

2042.96 1 2042.96 182.95 0.0009

Temperature (X1) 699.69 1 699.69 62.66 0.0042

X22 2.025 1 2.025 0.18 0.6989

X21 25.921 1 25.921 2.32 0.225

X1X2 0.5625 1 0.5625 0.05 0.8368

Residual 76.3262 9 8.48

Lack-of-fit 10.1345 3 3.378 0.3 0.8238

SS, sum of squares; df, degrEES OF FREEDom; MS, mean square.

b i om a s s a n d b i o e n e r g y 6 1 ( 2 0 1 4 ) 1 7 9e1 8 6184

concentration at low reaction times. However, this influence

gets more significance at large reaction time. At 240 min of

reaction time, the variation of acid conversion with catalyst

concentration was found not to be uniform in the whole

catalyst concentration range. The high acid conversion in-

crease observed for higher catalyst concentration can prob-

ably be due to formation of diester [18]. Temperature also have

a significant and positive effect on the acid conversion in the

experimental range studied (65 �Ce75 �C) (Table 3). The tem-

perature effect diminishes progressively with the reaction

time from 30 min to 180 min, perhaps due to the decrease in

reaction rate produced by decrease in reactant concentrations

[5]. However, there was no significant interactive effect be-

tween catalyst concentration and temperature in all reaction

times. At 30min of reaction time, neither of the two individual

variable has significant effect on acid conversion.

The results obtained in this first set of experiments were

adjusted by linear regression because the number of different

combinations in the 22 factorial design (experiments 1e8 in

Table 3) was insufficient to carry out a more complex model.

However, there was no evidence that a linear model could be

use to describe the strong relationship that existed between

the process variables. To evaluate if this factorial design was

sufficient to accurately describe the diglycerol monooleate

production process, the statistical significance of the curva-

ture was studied (Table 3). The curvature effect was found

statistically significant and it was necessary to consider a

different design. Consequently, it is necessary to consider a

more complex design to fit the data to a second-ordermodel in

two variables.

r = 0.985

37

47

57

67

77

87

97

37 47 57 67 77 87 97Predicted values

Obv

erve

d va

lues

Fig. 3 e MS. Parity plot showing relationships between

observed and predicted values by the regression model for

the acid conversion at 240 min.

3.3. Optimization by response surface methodology

To better predict the effect of reaction variables, a second-

order model is required, because of the significant curvature

effect found in linear stage. Additional experiments called star

points and coded �a (where a, the distance from the origin to

the star point, is given by a ¼ 2n/4. In this design a ¼ 1.414)

must be incorporated into the 22 factorial design (experiments

9e12 in Table 2) [19]. The corresponding model is a central

composite design. The adjustment of the response variable to

the quadratic model resulted in the following equation (1).

Y ¼ a0 þX2

k¼1

akXk þX2

k¼1

akkX2k þ

X2

ks1

akjXkXj (1)

where Y is the response; Xk and Xj are the uncoded indepen-

dent variables and a0, ak, akk and akj are intercept, linear,

quadratic and interaction constant coefficients, respectively.

The fit of themodel was evaluated by coefficient of correlation

(r) and analysis of variance (ANOVA). This fit showed that the

esterification of pure diglycerol with oleic acid was most

suitably described with a quadratic polynomial model. The

equation of the model based on the coded values Eq. (1) be-

comes Eqs. (2)e(5) (Table 4).

The relationships between reaction parameters and re-

sponses can be better understood by studying response sur-

face plots and the planned series of contour plots generated

from the predicted model. Fig. 2 represents the effect of

varying temperature and catalyst concentration on the acid

conversion as response surface and contour plots, at 30, 60,

120 and 240 min.

Analysis of variance (ANOVA) of this newmodel at 240min

of reaction time (Table 5) showed that its statistical signifi-

cance and predictability (p < 0.0001) was much greater than

that of the linear model developed above. The goodness of fit

of the regressionmodel was evaluated by a test for lack of fit at

95% level of significance. Parity plot between observed values

and values predicted by the model for the acid conversion,

and the coefficients of correlation r, are show in Fig. 3. The plot

showed a satisfactory linear correlation between observed

and predicted values (r ¼ 0.985). This implies that 98.5% of the

variation could be explained by fittedmodel. Joglekar andMay

[20] suggested that for a good fit of amodel, r should be at least

0.80. According to the value of adjusted r (0.983) which is a

modification of r that adjust for the number of variables in a

model, 98.3% of the variance can be explained by the model.

Therefore, these results suggest that well-fitting of the re-

lationships between the reaction parameters investigated and

acid conversion was successfully established.

3.3.1. Influence of operating variables on acid conversion at240 minThe analysis of variance, estimated effect and interactions on

acid conversion at 240 min were listed with statistical signif-

icance in Table 5. The two reaction parameters investigated

Table 6 e Observed values and values predicted byregression model for acid conversion in six additionalexperiments.

Reaction parameters Responses

X1 (�C) X2 (%) wacid conversion (%) Accuracy(%)Observed Predicted

65a 3a 76.1a 75.0 98.50

70a 1a 55.0a 55.4 99.20

75a 3a 51.9a 51.4 99.00

70a 5a 81.4a 83.6 97.30

77a 5.8a 94.1a 96.4 97.80

77b 5.8b 93.9b 96.4 97.40

a Small scale.b Using a 5 dm3 batch reactor.

b i om a s s a n d b i o e n e r g y 6 1 ( 2 0 1 4 ) 1 7 9e1 8 6 185

positively affected the levels of acid conversion, with catalyst

concentration having the greatest effect. In this case, the two

effects have p-values less than 0.05, indicating that they are

significantly different from zero at the 95% confidence level.

Whereas quadratic terms of catalyst concentration and tem-

perature and their interaction had not significant effect

(p > 0.05) on the acid conversion. These results are in agree-

ment with the results shown in the prediction plots for the

acid conversion level by the effect of individual reaction pa-

rameters obtained by our research group in a previous work

[16].

3.4. Optimum conditions and model validation

According to the model, the optimal conditions are likely to

provide the best process response, leading to the highest acid

conversion and process efficiency. Using RSM the two reaction

parameters, temperature and catalyst concentration should

be kept at the highest level (77 �C) and (5.8%), respectively,

within their given ranges to obtain the highest acid

conversion.

To examine the adequacy of the model established, an

experiment at the predicted optimal conditions was carried

out, together with five additional experimental sets selected

from the given ranges of reaction parameters (Table 6). The

regression model was also verified with an experiment at the

predicted optimal conditions, using a 5� 10-3 m3 batch reactor

for the scale up of the process (Table 6). Using this reactor, the

acid conversion reached 93.9%, with an accuracy of 97.4%; the

result was close to that obtained on the smaller scale. There

was considerable agreement between the experimental and

the predicted acid conversion values and this confirmed that

the quadratic model was suitable to predict the course of this

reaction.

4. Conclusion

Novozym 435 was the most effective biocatalyst for the

esterification of pure diglycerol with oleic acid to produce pure

diglycerol monooleate ester, among six commercially avail-

able lipases tested in the present study. The modeling and

optimization of Novozym 435-catalyzed synthesis of

diglycerol monooleate ester was successfully performed by

central composite design and response surface methodology.

The influence of twomain reaction parameters were elucidate

within the experimental range considered and optimal reac-

tion condition were obtained using the established model.

Under these optimal conditions the acid conversion was

93.9%. The coefficient of correlation r (0.985) and analysis of

variance implied that themodel satisfactorily represented the

real relationship of the main reaction parameters and the

response, and have been found to be valid for the scale up the

process.

Acknowledgments

The authors gratefully acknowledge financial support of this

work by the PRI-PIBAR-2011-1375 and “Ministerio de Economıa

y Competitividad” (CTQ-2009-09-088).

r e f e r e n c e s

[1] Benvegnu T, Plusquellec D, Lemiegre L. Surfactants fromrenewable sources: synthesis and applications. In: Naceur M,Gandini A, editors. Monomers, polymers and compositesfrom renewable sources. Amsterdam: Elsevier Limited; 2008.pp. 153e78.

[2] Zhu Q, Li T, Wang Y, Yang B, Ma Y. A two-stage enzymaticprocess for synthesis of extremely pure high oleic glycerolmonooleate. Enzym Microb Tech 2010;48(2):143e7.

[3] Sakamoto M, Ohba A, Kuriyama J, Maruo K, Ueno S, Sato K.Influences of fatty acid moiety and esterification ofpolyglycerol fatty acid esters on the crystallization of palmmid fraction in oil-in-water emulsion. Colloids Surf B2004;37(1e2):27e33.

[4] Kjellin M, Johansson I. Surfactants from renewableresources. 1st ed. Chichester: John Wiley & Sons Inc; 2010.

[5] Sanchez N, Martinez M, Aracil J. Selective esterification ofglycerine to 1-glycerol monooleate. 2. Optimization studies.Ind Eng Chem Res 1997;36(5):1529e32.

[6] Pitzalis P, Monduzzi M, Krog N, Larsson H, Ljusberg-Wahren H, Nylander T. Characterization of the liquid-crystalline phases in the glycerol monooleate/diglycerolmonooleate/water system. Langmuir 2000;16(15):6358e65.

[7] Kato T, Nakamura T, Yamashita M, Kawaguchi M, Kato T,Itoh T. Surfactant properties of purified polyglycerolmonolaurates. J Surfactants Deterg 2003;6(4):331e7.

[8] Alvarez JM, Rodrıguez JM. Interfacial properties of diglycerolesters and caseinate mixed films at the air-water interface.J Phys Chem C 2007;111(12):4790e9.

[9] Nakamura T, Yamashita M, inventors; Taiyo Kagaku Co.,Ltd., assignee. Polyether-polyol compound. United Statespatent US 20020035238; 2002 Mar 21.

[10] Marquez-Alvarez C, Sastre E, Perez-Pariente J. Solid catalystsfor the synthesis of fatty esters of glycerol, polyglycerols andsorbitol from renewable resources. Top Catal2004;27(1e4):105e17.

[11] Endo T, Maruo K, inventors; Taiyo Kagaku Co., Ltd., assignee.Production of polyglycerol mono-fatty acid ester. Japanpatent JP 2000143794; 2000 May 26.

[12] Lemke DW, inventor; Lonza Inc., assignee. Processes forpreparing linear polyglycerols and polyglycerol esters. WorldIntellectual Property Organization patent WO 2002036534;2002 May 10.

b i om a s s a n d b i o e n e r g y 6 1 ( 2 0 1 4 ) 1 7 9e1 8 6186

[13] Corma A, Iborra S, Velty A. Chemical routes for thetransformation of biomass into chemicals. Chem Rev2007;107(6):2411e502.

[14] Treichel H, Oliveira D, Mazutti MA, Di Luccio M, Oliveira JV. Areview on microbial lipases production. Food BioprocessTech 2010;3(2):182e96.

[15] Aehle W. Enzymes in industry: production and applications.3th ed. Weinheim: Wiley-VCH; 2007.

[16] Martinez M, Oliveros R, Aracil J. Synthesis of biosurfactants:enzymatic esterification of diglycerol and oleic acid. 1.Kinetic modelling. Ind Eng Chem Res 2011;50(11):6609e14.

[17] Pollard D, Kosjek B. Industrial-scale applications of enzymesin non-aqueous solvents. In: Carrea G, Riva S, editors.

Organic synthesis with enzymes in non-aqueous media.Weinheim: Wiley-VCH Verlag GmbH & Co. KGaA; 2008.pp. 169e88.

[18] El-Boulifi N, Aracil J, Martınez M. Lipase-catalyzed synthesisof isosorbide monoricinoleate: process optimization byresponse surface methodology. Bioresour Technol2010;101(22):8520e5.

[19] Barros Neto B, Scarminio IS, Bruns RE. Como FazerExperimentos: Pesquisa e desenvolvimento na ciencia e naindustria. 2nd ed. Campinas: Unicamp; 2003.

[20] Joglekar AM, May AT. Product excellence through design ofexperiments. Cereal Foods World 1987;32(12):857e68.