Research Article

A Supercritical Fluid-Assisted, IntegratedProcess for By-Products from Fat and LipidProduction

An integrated process for the enzymatic alcoholysis of lipids and the separationof the products by supercritical fluid extraction has been designed and tested witha commercial vegetable oil. The optimization of the reaction conditions and thefractionated separation in a cascade of three separators, which were operable atdifferent temperatures and pressures, have been tested. The fractions were ana-lyzed by HPLC for their content of fatty acid esters and different acylglycerides.

Keywords: Enzymatic processes, Process optimization, Supercritical Fluid Extraction

Received: January 2, 2007; accepted: February 8, 2007

DOI: 10.1002/ceat.200700002

1 Introduction

The manufacture of lipid products from natural resources nor-mally yields significant amounts of low-value fats and oilsalong with the focused high-value processes. Normally theseproducts are processed for practically zero profit, which al-though favorable for their disposal, can cost a considerableamount of money. With better refining technologies, thesewaste products can be turned into valuable products or at leastconverted into useful energy sources. It is possible to convertfats into glycerine and fatty acid esters by alcoholysis. Fattyacid esters can be used instead of petroleum in ovens or dieselengines to provide thermal or electrical energy, while glycerineis a raw material for the chemical industry.

As an alternative to chemical hydrolysis, which is difficult tocontrol, the biochemical conversion by lipase is uncomplicatedto carry out [1, 2]. Lipases are highly versatile and efficient bio-catalysts for these esterification reactions [3, 4]. They can actunder nonaqueous conditions, which are useful for the sup-pression of hydrolysis side reactions that are inevitable in thechemical hydrolysis of lipids. The mildness of the reaction con-ditions offered by lipases is especially favorable for processesinvolving the highly labile long-chain polyunsaturated fattyacids (PUFA). Microbial lipases are used to enrich PUFAs fromanimal and plant lipids, e.g., menhaden, tuna and borage oil[5]. A large number of additional hydrolytic applications havebeen described for microbial lipases, including flavor develop-ment for dairy products, e.g., cheese, butter, margarine, alco-holic beverages, milk chocolate and sweets, achieved by selec-tive hydrolysis of fat triglycerides to release free fatty acids.

These free fatty acids can either act as flavors or flavor precur-sors. When the reaction is applied to triglycerides, which are themain components of fats and lipids, fatty acid monoglyceridescan be generated along with fatty acid esters. Monoglyceridesare widely used as emulsifiers in the food industry and yieldhigher market prices than oils. The energetic use of fatty acid es-ters can improve the energy balance of the enzymatic process.

Lipases are activated only when adsorbed to an oil-water in-terface [6] and do not hydrolyze dissolved substrates in thebulk fluid. This configuration makes the reaction sensitive tothe influence of any amphiphilic substances. It has been foundin biphasic systems that specific kinds of lecithin can slowdown or even hinder the enzymatic reaction [7]. In water-freesystems, alcohols can also limit the efficiency of lipase [8]. Thisis independent of the presence of lipase in water micelles orimmobilized on a hydrophilic carrier.

Enzymatic reactions which are carried out in a supercriticalfluid solvent have higher mass transfer rates than reactionsperformed under classical conditions [9]. Supercritical fluidprocesses also offer unique possibilities for the continuous andselective extraction of products from the reaction mixture[10]. Due to the significant differences in their thermodynamicproperties, the pressure and temperature of the supercriticalfluid can be chosen in a way that only the fatty acid esters arewithdrawn. It is also possible to extract monoglycerides alongwith the fatty acid esters and separate them by a two-stage de-pressurization in two subsequently arranged separation vessels.

2 Experimental

2.1 Materials

Household grade corn oil (Ottogicorn oil, Korea) was used intransesterification reactions with ethanol (technical grade,

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim http://www.cet-journal.com

Andreas Weber1

Eun-Hee Lee1

Sang-Kyu Shin2

Byung-Soo Chun2

1 GoshenBitech, Namyangju-si,Kyonggi-do, Republic of Korea.

2 Pukyong National University,Nam-gu, Busan, Republic ofKorea.

–Correspondence: A. Weber (weber@igoshen.com), GoshenBiotech,83-2, Wolmun-ri, Waboo-eup, Namyangju-si, Kyonggi-do 472-90,Republic of Korea.

732 Chem. Eng. Technol. 2007, 30, No. 6, 732–736

95 %, Korean Ethanol Supplies Company, and 99.9 % HPLCgrade, Samchun Chemicals). Different kinds of immobilizedlipase were used including Novozyme 435, Lipozyme TL-IM,Lipozyme-RM, and Lipopan (Novozymes A/S, Bagsvaerd,Denmark).

2.2 Analytical Methods

HPLC analysis was carried out on a Waters 600 E Chromato-graphic system equipped with a Waters 486 UV detector andan Alltech Mk III ELSD. The different species of lipids wereanalyzed following the method of Holcapek et al. [11] and aWaters Symmetry C18 column (WAT054275) was used. Theeluent was fed at 1 mL min–1 in a gradient from 100 % metha-nol to 50 % methanol, 22.5 % hexane and 27.5 % isopropanolover a period of 15 min. The conversion of the enzymatic reac-tion was determined by the decrease of the triglyceride fractionin the samples. The activity of the enzymes was tested by asimple assay using the same corn oil as in the transesterifica-tion experiments. The tests were carried out in 250 mL Erlen-meyer flasks containing a mixture of 20 g of corn oil and1.2 mL of pure water using 2 wt % of each of the immobilizedlipases. The reaction mixture was incubated at 40 °C and sha-ken at 150 min–1 for 2 h. After the reaction, the amount ofhydrolyzed triglyceride was determined by titration of the freefatty acids with KOH in ethanol (0.02 mol/L) using phe-nolphthalein as an indicator.

2.3 Transesterification Experiments

2.3.1 Conventional Solvent-Free Ethanolysis

Experiments were carried out with asubstrate mixture of corn oil and etha-nol according to the solvent free meth-od proposed by Shimada et al., who in-vestigated the reaction of vegetable oiland methanol under conventional con-ditions [8]. The esterification of longchain lipids with short chain alcoholsis limited by the mutual solubility ofboth liquids. Also, the inactivation ofthe enzymes by methanol or isopro-panol under certain conditions hasbeen reported, which can be overcomeby the stepwise addition of a stoichio-metric amount of the alcohol [8, 12].Reactions were carried out in 250 mLErlenmeyer flasks using a mixture of20 g of corn oil, 3.14 g of ethanol, and0.6 g of immobilized lipase. Ethanol(95 %) was added to the mixture inthree successive steps of the same mo-lar equivalent (1.05 g) of corn oil for24 h. The reaction mixture was incu-bated at 40 °C and shaken at 120 min–1

on a shaking water bath.

2.3.2 Ethanolysis under Supercritical Carbon Dioxide

A laboratory scale supercritical fluid plant depicted in Fig. 1was build for the purpose of enzymatic reactions and subse-quent extraction processes. The plant consisted of a reactorwith a magnetically driven blade stirrer (550 mL, 40 MPa,100 °C), and three separators (150 mL, 15 MPa, 100 °C). CO2

was fed by a Lewa Ecoflow LDB1 pump with a working pres-sure of 44 MPa, at a flow rate of maximum 2 L/h. It was possi-ble to adjust the pressure drop between the separators E-4through to E-6 by using the back pressure regulators (TES-COM 26-1700 Series) to split off different products and by-products.

Based upon the experimental results under conventionalconditions, transesterification experiments were carried out inthe continuously stirred reactor. In the current experiments,the substrate containing 25–300 g of corn oil, 5.2–59 g of etha-nol and 4–18 g of immobilized enzyme, was added batch wiseinto the reactor. The reactor was flushed with CO2 and ad-justed to different pressures during the enzymatic reaction.The mixture was stirred at constant temperature and pressurefor a designated time period. Afterwards, the reaction mixturewas extracted by setting a CO2 flow of ca. 32 g/min. A pressureof 8.0 MPa was adjusted in separator E-4, while separators E-5and E-6 were operated at 5.0 MPa. The extraction rate wasdetermined by withdrawing samples from the separators andmeasuring their weight change over time. Finally, the productsin the reactor and separators were analyzed by HPLC.

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim http://www.cet-journal.com

Figure 1. Laboratory scale pilot plant for the reactive extraction of monoglycerides and fattyacid esters.

Chem. Eng. Technol. 2007, 30, No. 6, 732–736 Supercritical fluid extraction 733

3 Experimental Results

The activity of the different enzymes in the hydrolysis of cornoil was significantly different. Lipozyme-TL and Lipozyme-RM showed a higher activity than Novozyme 435 and Lipopan(see Fig. 2). The difference has been found to be even greaterin the transesterification of corn oil with ethanol. In the pres-ence of Lipozyme-TL and Lipozyme RM, 36 % and 50 % ofthe corn oil triglycerides were converted at 50 °C in 4 h, re-spectively. No significant reaction was determined when usingNovozyme 435 under the same conditions.

Lipozyme TL showed almost the same level of activity intransesterification experiments under supercritical conditions.The conversion of triglycerides was found to be ca. 50 % after18 h of reaction, when only one third of the stoichiometricamount of ethanol was contained in the substrate (see Tab. 1).Excess amounts of ethanol diminished the yield. The amountextracted in the separators increases by adding a second por-tion of ethanol after 18 h and proceeding with the reaction.The filling level of the reactor showed a significant influenceon the conversion rate (see Fig. 3). The reaction was more effi-cient when the reactor was filled to more than half of its vol-ume.

It was possible to efficiently extract the products at a rela-tively low pressure of 10 MPa. Fig. 4 shows the HPLC-analysisof the different products. The products in separator E-4 andE-5 consisted mainly of fatty acid ethyl esters while the reactorcontained mainly triglycerides and diglycerides. No significantamounts of extract were found in separator E-6. The extrac-

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim http://www.cet-journal.com

Figure 2. Activity of different enzymes determined by the in-crease of acid number during enzymatic hydrolysis.

Table 1. Conversion rates and amounts extracted at different ex-perimental conditions.

No. Substrate T(°C)

P(MPa)

Extractamount(g)

ConvertedTriglyceride(%)

LE051130 Corn oil 100 gEthanol 16 gLipozyme TL 4 g

40 10 9.0 45

LE060103 Corn oil 100 gEthanol 16 gLipozyme TL 4 g

40 10 21 49

LE060105 Corn oil 100 gEthanol 5.2 g + 5.2 gLipozyme TL 4 g

40 10 42 54

Figure 3. Conversion rate at different filling levels of the reactor.

Figure 4. HPLC-analysis of the products in the reactor and se-parator E-4 and E-5, after reaction and extraction. Experimentalconditions were: Reactor E-3: 10 MPa and 40 °C, Separator E-4:8.0 MPa and 40 °C, Separator E-5 and E-6: 5.0 MPa and 35 °C,CO2-flow: 32 g/min. The retention times of the different compo-nents are discussed in the text.

734 A. Weber et al. Chem. Eng. Technol. 2007, 30, No. 6, 732–736

tion conditions in this experiment were 10 MPa and 40 °C inthe reactor E-3, 8.0 MPa and 40 °C in separator E-4, and5.0 MPa and 35 °C, in separators E-5 and E-6. The CO2 flowwas 32 g/min. Peaks with a retention time of 4–5 min refer tofatty acids and monoglycerides, while fatty acid esters appearbetween 7 and 9 min, diglycerides between 12 and 16 min, andtriglycerides after 20 min. The ELSD detector is more sensitivefor diglycerides and triglycerides than for monoglycerides orfatty acid esters. The concentration of triglycerides was 2.1 %in separator E-4 and 2.7 % in separator E-5. Separator E-5contained visible amounts of water and ethanol.

Fig. 5 shows the separation rate in the two separators. Theextraction rate was found to be slightly higher than in a com-parable study by Gironi et al. [13]. This may be due to the factthat along with lipids, significant amounts of excess ethanoland water were also present in the extracts.

4 Modeling of the Reactions

In the transesterification of triglycerides a number of reactionscan be considered to happen simultaneously or subsequently.All of these reactions can be seen as reversible. For the purposeof modeling the measured concentrations of the components,a basic set of reactions was sought that provides basic data inthe form of velocity constants of the kinetic equations, andwhich can be used to compare results of different experimentalconditions, different lipids, or different enzymes. It was possi-ble to describe the measured data by the limited set of reac-tions detailed in Tab. 2, in the same way as proposed by Mo-quin et al. for the hydrolysis of triglycerides [14]. Furtherpossible reactions, e.g., the rearrangement of 2-monoglyceridesto 1-monoglycerides and the reaction of 1-monoglycerides toglycerol and ester have been excluded so far. Calculations werecarried out with the software CKS (Version 1.0.1) from IBM[15]. Fitting of data to the model requires estimation of theparameters as well as assembly of a suitable model. All reac-tions were assumed to be binary first-order reactions. Morecomplex reaction steps, e.g., the binding mechanism of thesubstrate to the enzyme, were neglected.

The model shows a good fit to the triglyceride data (seeFig. 6). As a result, other information such as the water con-

tent and the concentration of free fatty acids is obtained fromthe model calculations. Both parameters play an important rolefor the quality of the product in a production process. A com-parison with the kinetics obtained under supercritical carbondioxide shows only a slightly higher reaction rate (see Fig. 7).

5 Conclusions and Outlook

Lipozyme-TL and Lipozyme RM have been found to be effi-cient enzymes for the transesterification of corn oil. Lipozyme-TL is also applicable in supercritical CO2 and results in accept-

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim http://www.cet-journal.com

Figure 5. Total extraction amounts in Separators E-4 and E-5.

Table 2. Partial reactions during the ethanolysis of a triglyceridein the presence of water.

1. Triglyceride + Ethanol � Ethyl Ester + Diglyceride

2. Diglyceride + Ethanol � Ethyl Ester + Monoglyceride

3. Triglyceride + Water � Diglyceride + Fatty Acid

Figure 6. Model applied to experimental data from a conven-tional experiment. Experimental conditions: 40 °C, 6 % LipozymeTL-IM, 1/3 stoichiometric amount of ethanol on a shaking waterbath at 120 min–1.

Figure 7. Modeled data from an experiment under supercriticalcarbon dioxide conditions. Experimental conditions: 40 °C,10 MPa, 10 % Lipozyme TL-IM, 1/3 stoichiometric amount ofethanol, stirrer speed 200 min–1.

Chem. Eng. Technol. 2007, 30, No. 6, 732–736 Supercritical fluid extraction 735

able reaction yields. The concentration of ethanol in the sub-strate is critical for both supercritical conditions and conven-tional experiments, since excess amounts of ethanol inhibit theenzymatic reaction. Even though the reaction rate was notmuch higher under supercritical CO2 conditions, the possibili-ty of extracting ethyl ester and monoglycerides from the reac-tion mixture can be seen as a real advantage. The extraction offatty acid esters and monoglycerides from the reaction mixturewas relatively easy to achieve and worked at comparably lowpressures of 10–15 MPa.

The reaction was modeled with a simple approach usingonly three partial reaction steps. Despite its simplicity, the in-formation derived about free fatty acid concentration follow-ing reaction, and water and ethanol concentrations was suffi-ciently precise to predict the quality of the product. Futurework will detail the use of a pilot plant for a continuous modeoperation of the substrate feed and extraction of the products.The model will be applied to plan experiments and to calculatethe stationary concentrations of the components.

Acknowledgements

The authors gratefully acknowledge support for the currentproject by the Technological Innovation Strategy Fund of theSmall and Medium Business Administration of the KoreanGovernment, Project No. 2004-149.

References

[1] R. Sharma, Y. Chisti, U. C. Banerjee, Biotechnol. Adv. 2001,19, 627.

[2] A. L. Paiva, F. X. Malcata, J. Mol. Catal. B: Enzym. 1997, 3(1–4), 99.

[3] Lipases – Their Structure, Biochemistry and Applications (Eds:P. Woolley, S. B. Peterson), Cambridge University Press,Cambridge 1994.

[4] Enzymes in Lipid Modification (Ed: U. T. Bornscheuer), Wiley-VCH, Weinheim 2000.

[5] H. Gunnlaugsdottir, M. Jaremo, B. Sivik, J. Supercrit. Fluids1998, 12, 85.

[6] M. Martinelle, M. Holmquist, K. Hult, Biochim. Biophys.Acta 1995, 1258 (3), 272.

[7] J. P. Kupferberg, S. Yokoyama, F. J. Kezdy, J. Biol. Chem.1981, 256 (12), 6274.

[8] Y. Shimada, Y. Watanabe, A. Sugihara, Y. Tominaga, J. Mol.Catal. B: Enzym. 2002, 17, 133.

[9] Z. Knez, M. Habulin, Biochem. Eng. J. 2005, 27, 120.[10] J. W. King, E. Sable-Demessie, F. Temelli, J. A. Tee, J. Super-

crit. Fluids 1997, 10, 127.[11] M. Holcapek, P. Jandera, J. Fischer, B. Prokes, J. Chromatogr.,

A 1999, 858 (1), 13.[12] M. M. Soumanou, U. T. Bornscheuer, Enzyme Microb. Tech-

nol. 2003, 33, 97.[13] F. Gironi, M. Maschietti, 7th Italian Conf. on Supercritical

Fluids, I.S.A.S.F., Trieste, Italy, June 2004.[14] P. H. L. Moquin, F. Temelli, H. Sovova, M. D. A. Saldana,

J. Supercrit. Fluids 2006, 37 (4), 417.[15] Chemical Kinetics Simulator – version 1.0.1., IBM Corpora-

tion, 1996. http://www.almaden.ibm.com/st/computational_science/ck/msim/

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim http://www.cet-journal.com

736 A. Weber et al. Chem. Eng. Technol. 2007, 30, No. 6, 732–736