The role of phase behavior in the enzyme catalyzed synthesis of glycerol monolaurate

Abstract

Partial glycerides such as monoacylglycerides (MAGs) are important functional ingredients with various applications in the cosmetics and food industry. Direct synthesis of high purity MAGs is commercially not feasible due to low selectivity and substrate miscibility problems. In this article, a selectivity increase in synthesis towards MAG with the addition of 5 wt% of the amphoteric surfactant cocamidopropyl betaine (CAPB) compared to a solvent free reaction is demonstrated. In comparison to tert-butanol, the addition of CAPB leads to similar MAG contents in the equilibrium state and furthermore CAPB does not need to be removed from the reaction, because it is an approved additive in cosmetic products. Additionally it acts as a solubilizer in the product mixture. By adding 5 wt% of CAPB to the enzymatic glycerolysis of TL, the selectivity towards ML could be increased by 17% compared to the solvent free reaction. Also in the enzymatic esterification of glycerol and lauric acid the addition of 5 wt% CAPB shows an increase in selectivity towards ML up to 9% compared to the solvent free esterification. The major function is a direct influence on the phase behavior of the reaction mixture.

---

Introduction

Glycerol monolaurate (ML) is a non-ionic monoacylglyceride (MAG) with emulsifying properties. It is widely used in the food, pharmaceutical, and cosmetic industries and fulfils various functions as an emulsifier and thickening agent. Mono- and diacylglycerides (MAG, DAG) are, with a market share of almost 75%, the most commonly used synthetic, non-ionic emulsifiers in the food industry.¹ Worldwide they are permitted as food additives and in Europe they belong to the additive class E471.² Common industrial synthesis routes for MAG production are direct esterification of glycerol and fatty acids, glycerolysis of triacylglycerides (TAG) with glycerol, partial hydrolysis of TAG as well as interesterification between glycerol and fatty acid esters.³ During the past decade, extensive research has been carried out to identify high yielding reaction routes leading to MAG, using chemical or enzymatic catalysis. In consequence, several strategies were developed targeting the increase of MAG selectivity. Some highlights include the use of cosolvents such as the organic additive tert-butanol or solvation by ionic liquids such as coco pentaethoxy methyl ammonium methosulfate (CPMA·MS).^1,4,5 The focus of this work is the usage of cocamidopropyl betaine (CAPB) as additive in the enzymatic glycerolysis of glycerol trilaurate (TL) to increase the selectivity towards ML instead of glycerol dilaurate (DL). To ensure a high enzyme activity and stability an immobilized lipase (here Novozym 435®) was applied in the biocatalysis. Especially MAGs have a high surface activity due to their molecular structure with two hydroxyl functions. Therefore product mixtures with an increased proportion of MAGs are of particular interest to the cosmetic industry, for example as thickening agent in cosmetic formulations.^6,7 MAGs are industrially produced by chemical glycerolysis of TAGs or esterification of glycerol with fatty acids in the presence of strong acids or bases as catalyst at high temperatures up to 230 °C. Depending on the glycerol–fatty acid ratio, this results in a MAG content of 35–60 wt% in the lipid phase, which consists of MAG, DAG and TAG and glycerol. The maximum MAG content relative to total glycerides is obtained with a substantial excess of glycerol and is limited by the reaction equilibrium and the limited solubility of reactants in each other.^2,5,7 Increasing glycerol beyond a certain excess does not result in further MAG formation since the glycerol solubility in the lipid phase is limited and a biphasic system is obtained. If higher MAG contents are desired, the mixtures obtained by classical synthesis need to be further purified, e.g. by molecular distillation, which can substantially add to production cost. Among other approaches, lipases have been used to reach higher MAG contents during synthesis by employing different synthetic strategies (compare Table 1). Therefore the enzymatic reaction route to higher purity MAGs becomes increasingly important.

Table 1 Overview of strategies to increase reaction selectivity towards MAG.¹ All given results refer to the MAG content in the lipid phase

Strategies	Selective precipitation of MAG^8	Rational design of heterogeneous catalysts		Solvation^5,11
		Pore size^9	Catalyst surface^12
Principle	T↓ ⇒ MAG solubility in reaction mixture↓	ϕ pore < ϕDL	c glycerol,surface > cester,surface limitation in DAG/TAG formation	K eq
				A + B ⇌ C + D
Idea	Reaction equilibrium shifted towards MAG	Prevention of DAG and TAG formation	Hydrophilic surface interacts more with glycerol	Miscibility of reactants↓ solvent X interacts mostly with C (MAG) ⇒ activity coef. γML↓ ⇒ molar fraction of MAG↑
Results	Glycerolysis of olive oil: 90% MAG	Esterification of glycerol + lauric acid 75% MAG	81% MAG selectivity; 60% MAG molar yield (based on glycerol)	∼80% MAG in tert-alcohols up to 90% MAG in glycerolysis
	- 16 h at 42 °C
	- 4 d at 5 °C
Drawback	Reaction rates↓	Only esterification & no enzymes	No enzymes + dimethyl sulfide (DSMO)	Ionic liquids are toxic organic solvent ⇒ high effort in down streaming processing

---

Zhong et al. reported in 2014 in a review article several strategies for the increase of MAG selectivity. The main strategies are listed in Table 1, which are the selective precipitation of MAG,⁸ the rational design of a heterogeneous catalyst,^9,10 as well as using of solvation effects.^4,5,11 In the case of selective precipitation of MAG its decreasing solubility with decreasing the reaction temperature is made use of, whereby in consequence the reaction equilibrium is shifted towards MAG formation. The main drawback is the very slow reaction rate due to the low temperature.

Another strategy is the rational design of the heterogeneous catalyst by control of the pore size or the catalyst surface. However, both approaches are not applicable with enzymes yet. Solvation is a further technique to increase reaction selectivity towards MAG by using organic solvents like tert-butanol, which are not converted by most lipases, or ionic liquids. Enzymatic catalysis in non-aqueous media has the limitation of low activity, due to the insoluble form of the enzyme in organic reagents. An enzyme–surfactant complex could better disperse enzyme in organic solvents and improve the enzyme performance.¹³ Here the solvent increases the solubility of the reactants, especially the solubility of glycerol in the apolar phase, leading to an equilibrium shift towards MAG. Main drawbacks are the toxicity of ionic liquids and complicated downstream processing.

ML is a non-ionic emulsifier with amphiphilic character due to its two hydrophilic hydroxyl functions and its hydrophobic fatty acid residue. ML consists of three isomers, two regioisomers sn-1 ML and sn-2 ML, with two enantiomers of sn-1 ML. The thermodynamic equilibrium of sn-1 ML and sn-2 ML is temperature dependent and lays clearly on the side of sn-1 ML (K = sn-2 ML/sn-1 ML = 0.74).¹⁴ Also DL consists of three isomers, sn-1,3 DL and the two enantiomers of sn-1,2 DL. The acyl migration is also favoured to sn-1,3-DL.¹⁵ In this work ML as well as DL were considered as a sum of their isomers. The main research of this publication is the influence of the additives tert-butanol and CAPB on the formation rate of ML instead of DL to increase its selectivity. Knowledge about the phase behaviour in the multicomponent reaction medium is essential to investigate the effect of additives systematically. Therefore we used the thermodynamic model software COSMO-RS (conductor-like screening model for real solvents) to predict ternary liquid–liquid equilibria (LLE) containing glycerol and ML, DL and to study a priori effects of the two additives tert-butanol and CAPB on the individual phase composition. COSMO-RS is a quantum chemical model for thermodynamic properties based on molecular interacting surfaces of compounds in mixtures as well as statistical thermodynamics. For each of the systems glycerol/ML/DL, glycerol/ML/TL and glycerol/DL/TL with and without additives, 4 tie-lines of the liquid–liquid equilibrium were measured to show the favourable alteration of the phase compositions.

Experimental

Chemicals

Glycerol was purchased at Carl Roth GmbH & Co. KG (Karlsruhe, Germany), with a purity of 99.5%. Glycerol mono, di- and trilaurate as analytical standards were purchased at TCI Germany GmbH (Eschborn, Germany), all with purities higher than 98.5%. tert-Butanol was purchased at Carl Roth GmbH & Co. KG (Karlsruhe, Germany), with a purity of 99.5%. Glycerol trilaurate as starting material for the glycerolysis was produced enzymatically by our own in a 2 L bubble column at lab-scale^16,17 (lauric acid conversion of 96%; product composition: 91.3 wt% TL, 5.7 wt% DL, 0.1 wt% ML and 2.9 wt% LA). Powdered cocamidopropyl betaine (CAPB, TEGO® Betain CK D, Evonik), and Novozym 435® was provided from Evonik Industries AG.

GC analysis

For GC analysis, 2.8 mg sample were dissolved in 70 μL chloroform/pyridine mixture (50 v/v) and reacted with 10 μL silylating reagent MSTFA (n-methyl-n-trimethylsilyl-trifluoracetamide; Macherey-Nagel, Düren, Germany) for 30 min at 80 °C. As internal standard 20 μL tricaprin (Sigma Aldrich GmbH) was used resulting in a GC insert concentration of 2.8 μg μL⁻¹. GC analysis was performed with an HP 5980 on a VF5-ht Ultimetal (Agilent Technologies Deutschland GmbH & Co. KG, Waldbronn, Germany, length: 15 m), using a temperature gradient of 90–380 °C with a heating rate of 15 °C min⁻¹; FID detection was employed. Typical retention times were: glycerol 2.5 min, lauric acid 5.7 min, glycerol monolaurate 9.7, glycerol dilaurate 14.7 min, tricaprin 16 min and glycerol trilaurate 18.2 min.

Experimental setup

All liquid–liquid equilibrium measurements were set up in 1 mL scale in Eppendorf tubes (Eppendorf AG, Hamburg, Germany). For the determination of different tie-lines, all chemicals were weighed out (on a balance with an accuracy of 0.1 mg) to reach an equilibrium phase ratio of approximately one. After heating up to 65 °C the samples were mixed at this temperature and 500 rpm for one hour. Phase separation was reached after one hour of settling at 65 °C. Samples from each phase (polar phase at bottom and lipid phase at top) were withdrawn and analyzed with the GC, respectively.

All glycerolysis reactions were carried out at 65 °C with a stirrer rate of 800 rpm in a 50 mL (in case of solvent free glycerolysis and with tert-butanol as additive) or 100 mL (in case of glycerolysis with CAPB as additive) stirred tank reactor with a thermostating jacket. In each case a molar ratio (glycerol/glycerol trilaurate) of 5 and 5 wt% of Novozym 435® was used. Besides the additive-free glycerolysis, experiments with 5 wt% CAPB or tert-butanol as cosolvent were conducted. For comparison to a glycerolysis reaction an esterification of glycerol and lauric acid with a subtract ratio (glycerol/lauric acid) of 2 and 10 wt% Novozym 435® with addition of 5 wt% CAPB was carried out in a bubble column reactor at 65 °C and 1 L min⁻¹ aeration with air. In all reactions the equilibrium conversion was reached latest after 48 h.

Reaction equations/equilibrium constants

Glycerolysis

GLY + TL ⇌ ML + DL (1)

GLY + DL ⇌ 2ML (2)

2GLY + TL ⇌ 3ML (3)

Esterification

GLY + LA ⇌ ML + H₂O (7)

ML + LA ⇌ DL + H₂O (8)

DL + LA ⇌ TL + H₂O (9)

GLY + 3LA ⇌ TL + 3H₂O (10)

Calculation of selectivity towards glycerol monolaurate

LLE prediction with COSMO-RS

COSMO-RS (conduct or-like screening model for realistic solvation) is based on quantum chemical calculations and statistical thermodynamics and can be employed for the calculation of molecular interactions of molecules in solution.^18–22 For a priori calculations of liquid–liquid equilibria the program COSMOthermX (COSMOlogic GmbH & Co. KG, Leverkusen, Germany), where the COSMO-RS model is implemented, with the parameterization BP-TZVP-C30-1501 was applied. In this work, the “Liquid extraction” option for multi-component multi-phase liquid–liquid extraction equilibria calculation was used to derive the phase composition of the polar and lipid phase. For the calculations, COSMO-RS requires DFT/COSMO files for all involved components. The conformers and the corresponding DFT/COSMO files were created with COSMOconf [M. Diedenhofen, A. Klamt, COSMOconf, COSMOlogic GmbH & CO. KG, Leverkusen, Germany (2013)] and Turbomole 6.6 [TURBOMOLE Version 6.6, TURBOMOLE GmbH, Karlsruhe, Germany]. Further informations about COSMO-RS can be found in the corresponding publications by Klamt et al.^18–22

Results and discussion

Preliminary remarks

As model reaction in this work the enzymatic glycerolysis of TL to their partial glycerides glycerol monolaurate (ML) and glycerol dilaurate (DL) was investigated (see Fig. 2). The biocatalyzed glycerolysis of glycerol trilaurate (TL) was carried out with an excess of glycerol. However, the reaction mixture stayed multiphasic without the addition of any cosolvent. The overall reaction system consists of three phases: the polar bulk phase of glycerol, the disperse apolar lipid phase of all partial glycerides and solubilized glycerol, as well as the solid enzyme immobilizate, which is located in the apolar lipid phase due to its hydrophobic surface. The enzyme-catalyzed reaction between TL and glycerol takes place in the apolar lipid phase. However, glycerol forms the polar phase and has to diffuse into the lipid phase. Therefore, the phase equilibrium between glycerol and all partial glycerides (ML, DL and TL) is of major importance for the reaction (Fig. 2).

Fig. 1 Molecular structure of cocamidopropyl betaine (CAPB).

The reaction principle of glycerolysis is shown in Fig. 2. In general, free glycerol acts as a nucleophile in the transesterification of trilaurate and in consequence the partial glycerides ML and DL will be formed. The reaction scheme shows also detailed information on the formation of the regioisomers of ML and DL and the involved non enzyme-catalyzed acyl migration. However, the lipase used in this study catalyzes preferably the (trans)esterification of primary hydroxyl functions.^23,24 In a previous work the behaviour was proven by comparing the formation of sn-1 ML to sn-2 ML, as well as sn-1,3 DL to sn-1,2 DL.²⁵

In this study the focus of research is the influence of additives on the selectivity towards formation of ML, whereas ML and DL are considered as a sum of their isomers.

As additives, cocamidopropyl betaine (CAPB) was compared to tert-butanol in view of selectivity increase towards ML and its respective influence of the phase behaviour. CAPB is a zwitterion, whose molecular structure includes a quaternary ammonium cation and a carboxylate anion (Fig. 1). In this work a water free powdered form of CAPB was used. It is widely used as surfactant in personal care products, especially as mild surfactant in shampoos.

Fig. 2 Schematic principle of the glycerolysis reaction with glycerol (continuous phase in dark blue) in excess. Gly: glycerol, ML: glycerol monolaurate, DL: glycerol dilaurate, TL: glycerol trilaurate.

In view of the limitation reaction equilibrium and the limited solubility of reactants in each other by the phase behaviour, the idea of using CAPB as solubilizer instead of an organic solvent has arisen. Based on the molecular structure, the presumption was obvious that CAPB can dissolve glycerol in the apolar lipid phase due to the amphiphilic character. Compared to tert-butanol, CAPB has some advantages in the biocatalytic process for personal care products. Firstly, it can stay in the product mixture after the synthesis due to its application in cosmetic products. Hence, the costly purification step for tert-butanol removal is no longer necessary. Secondly, less safety and apparative limitations apply to the operation of the process with CAPB compared to the organic solvent tert-butanol. Finally, the non-volatile CAPB can be employed in an esterification process as an alternative to glycerolysis, which is not practically feasible with organic solvent present. The removal of reaction water by stripping, like in a bubble column or via vacuum is no problem, because CAPB is not evaporating under the conditions applied in contrast to tert-butanol.

Ternary phase diagrams

To prove the assumption and presumption of CAPB as ideal cosolvent in the system investigated and its application in the process, the phase behaviour was investigated and ternary plots were experimentally determined (Fig. 3–5). Therefore, defined mixtures of glycerol and its partial glycerides ML, DL and TL were prepared and both phases, the polar glycerol phase and the apolar lipid phase, were analyzed by GC.

Fig. 3 Experimental ternary phase diagrams at 65 °C of glycerol (Gly) and all partial glycerides of the reaction system (glycerol monolaurate: ML; glycerol dilaurate: DL and glycerol trilaurate: TL), showing the influence of 5 wt% CAPB on the phase behavior (red circles). Tie-lines are displayed as black dotted lines in each diagram. The green and red dotted lines are added as visual guide. [A] Ternary plot of Gly/ML/DL; [B] ternary plot of Gly/ML/TL; [C] ternary plot of Gly/DL/TL.

Fig. 4 Experimental ternary phase diagrams of glycerol [Gly], glycerol monolaurate [ML] and glycerol dilaurate [DL] (without additive: green squares), showing the influence of 5 wt% tert-butanol (red dots) and 5 wt% CAPB (blue triangles) on the phase behavior. The colored dotted lines are added as visual guide.

Fig. 5 Comparison of experimental and predicted (COSMO-RSRS) data for a ternary plot of glycerol, glycerol monolaurate and glycerol dilaurate with the influence of 5 wt% of tert-butanol. The colored dotted lines are added to guide the eye.

For comparison, all mixtures were prepared without (Fig. 3; green squares) and with addition of 5 wt% CAPB and analysed, respectively (Fig. 3; red circles).

First, it can be seen that two phases are formed, an almost pure polar glycerol phase and an apolar lipid phase containing all partial glycerides and dissolved glycerol. Also, mixtures containing ML (see Fig. 3A and B) are forming smaller miscibility gaps compared to mixtures without ML (see Fig. 3C). The reason for that is due to the glycerol monolaurate itself. It is more polar than the DL and TL and has a high surface activity due to its molecular structure with two hydroxyl functions. ML has a higher hydrogen bonding donor capacity compared to DL and TL. In the presence on ML, the hydrogen bonding donor capacity in the apolar lipid phase increases which lead to a rise in the glycerol solubility in the apolar phase. In Fig. 3A and B, the decrease of the miscibility gap by addition of 5 wt% CAPB is clearly visible. On the other hand CAPB shows no influence on the phase behaviour when ML is not present in the mixture to be investigated (see Fig. 3C).

Fig. 4 shows in comparison the influence of CAPB and tert-butanol on the phase behaviour of the ternary mixture of glycerol, glycerol monolaurate and glycerol dilaurate. The addition of 5 wt% tert-butanol also decreases the miscibility gap of this mixture. This means that more glycerol is soluble in the apolar lipid phase.

However, CAPB shows a slightly stronger effect on the phase behaviour than tert-butanol. Combined with the other benefits described above, this can make CAPB a preferable option compared to the organic additive tert-butanol. CAPB can shift the miscibility gap towards glycerol. In consequence more glycerol is soluble in the apolar lipid phase and by this means a higher concentration thereof is established for the reaction which takes place in the apolar lipid phase.

The influence of tert-butanol and CAPB on the phase behaviour of the ternary mixture of glycerol, glycerol monolaurate and glycerol dilaurate were also modeled with COSMO-RS. Fig. 5 shows the liquid–liquid equilibrium of Gly, ML and DL without and with 5 wt% tert-butanol (number of conformers: 10 for Gly, ML and DL, and 1 for tert-butanol, respectively). The results show qualitatively good agreement with the experimental data in case of the lipid-rich phase whereas the ML solubility in the polar phase is overestimated. However, it is known that the choice of conformers has a great influence on the predicted properties with COSMO-RS, especially for large molecules like ML, DL, and CABP.²⁶ A screening with COSMO-RS to identify suitable emulsifiers like CAPB did not lead to satisfactory results. Using diverse sets of conformers for CAPB, only small changes in the miscibility gap and both, an increased and decreased glycerol concentration was calculated. In order to achieve more reliable predictions, further studies for a reasonable choice of conformers are required which was not in the scope of this work. Nevertheless, the good results for the ternary LLE and the correct estimated trend in case of addition of tert-butanol demonstrates COSMO-RS to be able to predict liquid–liquid equilibria of complex mixtures containing large molecules like ML and DL with a suitable choice of conformers.

Application and comparison of additives in the reaction

Esterification. The application of CAPB and tert-butanol as cosolvents were shown in the previous results. Both additives have an influence on the phase behaviour of the model reaction system explicitly the decrease of the miscibility gap and as result the increase of the solubility of glycerol in the apolar lipid phase. In this subchapter the effect of these additives in the reaction itself were studied and compared to each other on the example of maximum ML synthesis. On the one hand, the additives were compared to solvent free reactions and on the other hand two different reaction types were tested. Here an esterification of glycerol and lauric acid, accomplished in a bubble column reactor, was compared to a glycerolysis of trilaurate, carried out in a stirred tank reactor (see Fig. 6 and 7). In the case of an esterification reaction, the content of ML increases compared to a solvent free reaction (see Fig. 6), as well as the selectivity towards ML increases by 9% from 0.580 to 0.633 (see Fig. 6). Table 2 shows the equilibrium constants for the esterification reaction (cf. reaction eqn (7)–(10) and equilibriums constants eqn (11)–(14)). By comparing the equilibrium constants, it is clearly obvious that the equilibrium of the esterification with addition of CAPB favoured the formation of ML.

Fig. 6 Comparison of reactions in the equilibrium state in the overall reaction mixture. The numbers on the green bands are the selectivity values towards ML S_ML. Two independent measurements with triplicates were carried out. Reaction conditions: glycerolysis reactions 65 °C, 800 rpm in a 50 mL (solvent free and tert-butanol) or 100 mL (CAPB) stirred tank reactor, molar ratio (glycerol/trilaurate) of 5 and 5 wt% of Novozym 435®, addition of 5 wt% CAPB or tert-butanol. Esterification reaction : substrate ratio (glycerol/lauric acid) of 2 and 10 wt% Novozym 435® with addition of 5 wt% CAPB, bubble column reactor, 65 °C, 1 L min⁻¹ aeration.

Fig. 7 Comparison of reactions in the equilibrium state in the lipid apolar phase. Reaction conditions: glycerolysis reactions 65 °C, 800 rpm in a 50 mL (solvent free and tert-butanol) or 100 mL (CAPB) stirred tank reactor, molar ratio (glycerol/trilaurate) of 5 and 5 wt% of Novozym 435®, addition of 5 wt% CAPB or tert-butanol. Esterification reaction : substrate ratio (glycerol/lauric acid) of 2 and 10 wt% Novozym 435® with addition of 5 wt% CAPB, bubble column reactor, 65 °C, 1 L min⁻¹ aeration.

Table 2 Equilibrium constants for the esterification reaction

	K 1,esterification	K 2,esterification	K 3,esterification	K 4,esterification
CAPB	0.309	0.483	0.201	0.030
Solvent free	0.282	0.637	0.135	0.024

---

From the results in Table 2, some conclusions can be made:

• K_{1,CAPB,esterification} > K_{1,solvent free,esterification}

In the first esterification step (cf. reaction eqn (7)) the formation of ML with addition of 5 wt% CAPB is slightly higher compared to a solvent free reaction.

• K_{2,CAPB,esterification} < K_{2,solvent free,esterification}

In the second esterification step (cf. reaction eqn (8)) the conversion of ML to DL with addition of 5 wt% CAPB is lower compared to a solvent free reaction.

• K_{3,CAPB,esterification} > K_{3,solvent free,esterification}

In the third esterification step (cf. reaction eqn (9)) the conversion of DL to TL with addition of 5 wt% CAPB is slightly higher compared to a solvent free reaction.

Glycerolysis. In the case of the glycerolysis, two additives, an organic solvent tert-butanol as well as an zwitterionic/amphoteric surfactant CAPB, were compared to the solvent free reaction in respect to the selectivity towards ML. The addition of both additives, respectively, shows almost the same results (see Fig. 6). The content of ML increases compared to a solvent free glycerolysis, as well as the selectivity towards ML increases by 17% from 0.548 to 0.642. Comparing the equilibrium constants (see Table 3), the effect of both additives shifting the equilibrium on the ML site, is clearly obvious. Furthermore the conversion of glycerol increases compared to the solvent free reaction due to the lower amount of glycerol at the equilibrium point.

Table 3 Equilibrium constants for the glycerolysis reaction

	K 1,glycerolysis	K 2,glycerolysis	K 3,glycerolysis
CAPB	1.806	0.623	1.125
tert-Butanol	2.108	0.590	1.243
Solvent free	1.040	0.299	0.311

---

From the results in Table 3, some conclusions can be made:

• K_{1,tert-butanol,glycerolysis} > K_{1,CAPB,glycerolysis} > K_{1,solvent free,esterification}

In the first glycerolysis reaction step (cf. reaction eqn (1)) the formation of ML and DL with addition of 5 wt% CAPB or tert-butanol is higher compared to a solvent free reaction.

• K_{2,CAPB,glycerolysis} > K_{2,tert-butanol,glycerolysis} > K_{2,solvent free,esterification}

In the second glycerolysis reaction step (cf. reaction eqn (2)) the conversion of DL to ML with addition of 5 wt% CAPB or tert-butanol is higher compared to a solvent free reaction. The conversion of DL to ML is slightly higher comparing CAPB to tert-butanol.

• K_{3,tert-butanol,glycerolysis} > K_{3,CAPB,glycerolysis} > K_{3,solvent free,esterification}

The overall conversion of TL and Gly to ML with addition of 5 wt% CAPB or tert-butanol is higher compared to a solvent free reaction.

In this case CAPB shows great potential to be used as additive in esterification or glycerolysis reactions instead of tert-butanol increasing the selectivity towards ML in equal measure. Also the increase of the glycerol solubility in the apolar lipid phase is shown in Fig. 7. In both reaction types, the addition of 5 wt% CAPB or tert-butanol increases the glycerol solubility in the apolar lipid phase (see Fig. 7). Therefore, more glycerol is available as substrate, reacting to partial glycerides.

Conclusions

The influence of the cosolvents/additives CAPB and tert-butanol on the phase behaviour of reaction mixtures of glycerol and their partial glycerides ML, DL and TL was experimentally and computationally investigated. The resulting knowledge was applied to increase the reaction selectivity in biotransformations. The addition of 5 wt% of each additive decreases the miscibility gap whereby the solubility of glycerol in the apolar lipid phase increases and more glycerol can interact as substrate due to the location of the catalyst in the apolar lipid phase. CAPB and tert-butanol were also applied as cosolvents in the reaction itself. In the esterification, as well as in the glycerolysis reaction, the addition of CAPB increases the selectivity towards ML resulting in a more polar product mixture compared to a solvent free reaction. Therefore, the application of CAPB has great potential for the industrial production of partial glycerides. It enables the usage in the esterification reaction i.e. in a bubble column due to no evaporation of CAPB instead of running a reaction sequence containing a solvent free esterification followed by a glycerolysis reaction with addition of tert-butanol. The predictive model COSMO-RS was able to predict qualitatively the phase behavior of the ternary system Gly/ML/DL after addition of tert-butanol. Still, for complex molecules like CAPB the predicted trend strongly depends on the chosen set of conformers. The system presented is a model system. Further approaches will focus on the transfer to similar reaction systems including fatty acids with shorter chain lengths.

Acknowledgements

The authors thank the BMBF (German Federal Ministry of Education and Research) cluster “Biocatalysis 2021” (project number: 031A089) for the financial support.

References

1. N. Zhong, L.-Z. Cheong and X. Xu, Org. Process Res. Dev., 2014, 116(2), 97.
2. D. S. Negi, Base Catalyzed Glycerolysis of Fatty Acid Methyl Esters: Investigation towards the Development of a Continuous Process, Dissertation, TU Berlin, Berlin, 2006.
3. M. S. Machado, J. Pérez-Pariente, E. Sastre, D. Cardoso and A. de Guereñu, Appl. Catal., A, 2000, 203(2), 321.
4. Z. Guo and X. Xu, Green Chem., 2006, 8(1), 54.
5. M. L. Damstrup, T. Jensen, F. V. Sparsø, S. Z. Kiil, A. D. Jensen and X. Xu, J. Am. Oil Chem. Soc., 2006, 83(1), 27.
6. A. Corma, S. Iborra, S. Miquel and J. Primo, J. Catal., 1998, 173(2), 315.
7. G. L. Hasenhuettl, Food Emulsifiers and Their Applications Second Edition, NY Springer Science+Business Media, LLC, New York, 2008.
8. G. P. McNeill, S. Shimizu and T. Yamane, J. Am. Oil Chem. Soc., 1991, 68(1), 1.
9. I. Díaz, F. Mohino, J. Pérez-Pariente and E. Sastre, Appl. Catal., A, 2001, 205(1), 19.
10. F. Jérôme, G. Kharchafi, I. Adam and J. Barrault, Green Chem., 2004, 6(2), 72.
11. Z. Guo and X. Xu, Org. Biomol. Chem., 2005, 3(14), 2615.
12. G. Kharchafi, F. Jérôme, I. Adam, Y. Pouilloux and J. Barrault, New J. Chem., 2005, 29(7), 928.
13. Y. Zhang, Y. Dai, M. Hou, T. Li, J. Ge and Z. Liu, RSC Adv., 2013, 3(45), 22963.
14. G. Boswinkel, J. T. Derksen, K. Van't Riet and F. P. Cuperus, J. Am. Oil Chem. Soc., 1996, 73(6), 707.
15. X. Wang, J. Xiao, W. Zou, Z. Han, Q. Jin and X. Wang, Process Biochem., 2015, 50(3), 388.
16. L. Hilterhaus, O. Thum and A. Liese, Org. Process Res. Dev., 2008, 12(4), 618.
17. J. J. Müller, Enzymatische Veresterung in Blasensäulenreaktoren: Kinetik, Stofftransport und Online Analytik, Dissertation, Hamburg University of Technology, Hamburg, 2012.
18. A. Klamt, J. Phys. Chem., 1995, 99(7), 2224.
19. A. Klamt, V. Jonas, T. Bürger and J. C. W. Lohrenz, J. Phys. Chem. A, 1998, 102(26), 5074.
20. A. Klamt and F. Eckert, Fluid Phase Equilib., 2000, 172(1), 43.
21. A. Klamt, F. Eckert and M. Diedenhofen, Fluid Phase Equilib., 2009, 285(1), 15.
22. A. Klamt, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2011, 1(5), 699.
23. R. Irimescu, T. Saito and K. Kato, J. Mol. Catal. B: Enzym., 2004, 27(2), 69.
24. Z. Cabrera, G. Fernandez-Lorente, R. Fernandez-Lafuente, J. M. Palomo and J. M. Guisan, J. Mol. Catal. B: Enzym., 2009, 57(1), 171.
25. J. J. Mueller, S. Baum, L. Hilterhaus, M. Eckstein, O. Thum and A. Liese, Anal. Chem., 2011, 83(24), 9321.
26. M. Buggert, C. Cadena, L. Mokrushina, I. Smirnova, E. J. Maginn and W. Arlt, Chem. Eng. Technol., 2009, 32(6), 977.

---