DE GRUYTER OPEN HUNG. J. IND. CHEM. 2017 45(2), 1–49 Table of Contents Methyl-tert-Butyl-Ether Synthesis Reactor Modelling and Optimization Using Aspen Custom Modeler ZEESHAN NAWAZ .............................................................................................................................................. 1–7 Modelling the Partial Demineralization Process of Cow Milk by SuperPro Designer ATTILA CSIGHY, ANDRÁS KORIS, AND GYULA VATAI ......................................................................................... 9–12 Investigations of the TlInP2Se6–In4(P2Se6)3 System and Its Optical Properties VALERIA TOVT, IGOR BARCHIY, MICHAL PIASECKI, IWAN KITYK, AND ANATOLII FEDORCHUK ............................. 13–18 Examination of Innovative High-Throughput Fermentations ÁRON NÉMETH ............................................................................................................................................. 19–21 The Effect of Advanced Oxidation Pre-Treatment on the Membrane Filtration Parameters of Dairy Wastewater MIHÁLY ZAKAR, ILDIKÓ KOVÁCS, PÉTER MUHI, ERIKA HANCZNÉ LAKATOS, GÁBOR KESZTHELYI-SZABÓ, AND ZSUZSANNA LÁSZLÓ .............................................................................................................................. 23–27 Microencapsulation of Vegetable Oil: Alternative Approaches Using Membrane Technology and Spray Drying KRISZTINA ALBERT, GYULA VATAI, AND ANDRÁS KORIS .................................................................................. 29–33 Effect of Chain Length and Order of the Alcohol on Enzyme Activity during Enzymatic Esterification in Organic Media ZSÓFIA MÁRKUS, KATALIN BÉLAFI-BAKÓ, GÁBOR TÓTH, NÁNDOR NEMESTÓTHY, AND LÁSZLÓ GUBICZA ........... 35–39 State-of-the-Art Recovery of Fermentative Organic Acids by Ionic Liquids: An Overview KONSTANTZA TONOVA .................................................................................................................................. 41–44 Application of a Hydrophobic Polymeric Membrane for Carbon Dioxide Desorption from an MEA-Water Solution ZENON ZIOBROWSKI AND ADAM ROTKEGEL .................................................................................................... 45–49 HUNGARIAN JOURNAL OF INDUSTRY AND CHEMISTRY Vol. 45(2) pp. 1–7 (2017) hjic.mk.uni-pannon.hu DOI: 10.1515/hjic-2017-0012 METHYL-TERT-BUTYL-ETHER SYNTHESIS REACTOR MODELLING AND OPTIMIZATION USING AN ASPEN CUSTOM MODELER ZEESHAN NAWAZ SABIC Technology & Innovation, PO Box 42503, Saudi Basic Industries Corporation (SABIC), Riyadh, 11551, SAUDI ARABIA A pseudo-homogeneous model of methyl-tert-butyl-ether (MTBE) synthesis in a multi-tubular packed-bed reactor has been developed using an Aspen Custom Modeler (ACM) for selecting optimum operating strategies, for the maximization and enhancement of MTBE production, and isobutylene consumption, respectively. The model accounts for mass, energy and momentum balances; and the effectiveness factor is evaluated in a one- dimensional pseudo-homogeneous model. The kinetic investigation contains kinetic rate expressions as given by the effectiveness factor for accounting the resistance of pellets in terms of mass and heat transfer. An activity coefficient can be used in order to systematically obtain a new steady-state solution. The model used literature-based correlations for the estimation of heat transfer coefficients. The value of the coefficient for gas- coolant heat transfer can be adjusted by using a tuning coefficient in order to enrich the process data. Reasonable agreement was found between model predictions and data under similar conditions. The studies concerning model sensitivity compute the optimum temperature, pressure, feed flow rate, methanol/isobutylene ratio, heat removal rate, etc. of the reactor and suggest optimum operating conditions of the reactor. Keywords: methyl-tert-butyl-ether (MTBE) synthesis, lead reactor, multi-tubular packed-bed, Aspen Custom Modeler (ACM), isobutylene 1. Introduction Methyl-tert-butyl-ether (MTBE) is of importance among petrochemicals due to its physicochemical properties and as an additive in gasoline. It contributes to the gradual elimination of lead-based additives, increasing the octane number (as an octane booster) and serves as a volume extender [1]. It is formed by the etherification reaction of an alcohol with a tertiary olefin. Ethers are preferable over alcohols due to their lower sensitivity values, i.e. the difference between the RON (Research Octane Number) and MON (Motor Octane Number) [2]. Today, standard (Huls AG) and Ethermax are the known processes commercialized by Snamprogetti, CDTech, Universal Oil Products (UOP), Axens, Fortum, Arco, Phillips, BP-Intevep, Sinopec, Sumitomo, etc. These are similar processes to each other. In this study a Snamprogetti-lead reactor is modelled to maximize its productivity. 2. Experimental Conditions Considered 2.1. Reaction Chemistry The following five reactions (Eqs.(1-5)) are involved in MTBE synthesis when isobutylene reacts with methanol *Correspondence: zeeshan@sabic.com (MeOH). MTBE synthesis is an exothermic liquid- phase reversible reaction catalysed by a cationic ion- exchange resin (sulphonated macroporous polystyrene). For each mole of isobutylene converted, ~37.7 kJ of heat is released and the thermodynamic equilibrium determines the extent of conversion [1-2]. MTBE: CH3OH + (CH3)2C=CH2  (CH3)3COCH3 (1) However, undesirable side reactions may take place such as the dimerization of isobutylene to diisobutenes (DIB), 2,4,4-trimethyl-1-pentene (TMP-1) and 2,4,4- trimethyl-2-pentene (TMP-2); and the formation of methyl sec-butyl ether (MSBE), dimethyl ether (DME) and tert-butyl alcohol (TBA). The water produced during the formation of DME may react with isobutylene to form tert-butyl alcohol (TBA) and its blending octane number is lower than that of MTBE. On the other hand, the presence of water reduces the acidity of the catalyst (reducing activity) and therefore a higher reaction temperature is required. DIB: 2 (CH3)2C=CH2  (CH3)3CH2(CH3)C=CH2 + (CH3)3C(H)C=C(CH3)2 (2) MSBE: CH3OH + (C2H5)CH=CH2  CH3CH(OCH3)(CH2CH3) (3) DME: 2 CH3OH  CH3OCH3 + H2O (4) TBA: (CH3)2C=CH2 + H2O  (CH3)3COH (5) NAWAZ Hungarian Journal of Industry and Chemistry 2 2.2. Thermodynamics Etherification is an exothermic equilibrium reaction between a primary alcohol and an iso-olefin containing a double bond on a tertiary carbon atom (such as isobutene (iBu) under the operating temperature and pressure. Owing to the non-ideality of liquid mixtures due to their disparate polarities and equilibria, the kinetic rate expressions for the synthesis of MTBE are generally given in terms of component activities. Equilibrium and rate equations based on activities were first proposed by Rehfinger et al. [3] and kinetic (including the effectiveness factor) and equilibrium data from an isothermal packed-bed reactor using the catalyst Amberlyst-15 were discussed by Zhang et al. [4]. The activity-based equilibrium constant is expressed as: �a = � �MTBE �iBu∙�MeOH � eq. = � �MTBE �iBu∙�MeOH ∙ �MTBE �iBu∙�MeOH � eq. (6) where aMeOH, aiBu, and aMTBE are the activities of methanol, isobutene, and MTBE, respectively, while γ and x are the activity coefficients and mole fractions of these components, respectively. The values of the equilibrium constant (Ka) published in the literature vary significantly (Fig.1) as a function of temperature [3-10]. Deviations may stem from the different bases used for the evaluation of the activity coefficient as well as from the operating temperature of the reactor, for example, employing primary reactors or reactive distillation columns. 2.3. Catalysis and Kinetics The acidic cation exchange macroporous resin catalysts used commercially for the synthesis of MTBE are provided by Bayer, Dow Chemical (Rohm & Haas), Purolite, Kairui, etc. The resins are prepared by the suspension of styrene in the presence of an appropriate cross-linking agent (divinylbenzene (DVB)) for polymerization functionalized by means of sulphonation with sulphuric or chlorosulphonic acid [11]. A reduction in volume of ~20-30% is observed because of the shift of the resin from the hydrated form (as it is generally charged in water) to a more contracted form due to a much less polar-reacting medium (a hydrocarbon/methanol mixture). The usability of the catalyst largely depends upon the process conditions, namely the type of reactor, e.g. a reactor receiving thermal support is better than adiabatic and thermal degradation with breakage of the carbon-sulphur bond (therefore, temperatures below 130 °C are recommended) and feed impurities (cations, e.g. Na, Ca, Fe, Al, Cr, Si, etc., strong N-bases like ammonia and amines, weak N-bases like acetonitrile and propionitrile, and dienes). The rate expressions, essential kinetic parameters and thermodynamic data such as the activation energy, effectiveness of the catalyst, heat of reaction, equilibrium constant and so forth have been selected from the literature [3-14]. The rate of reaction depends substantially on two parameters: acidity (type and number of acidic sites) and accessibility (porosity, content, particle diameter and treating medium). In the homogeneous model (in the absence of the diffusion phenomena), the reaction rate can be described in terms of confined catalyst pores, where the reactant concentration is in a partition equilibrium with the corresponding concentrations in the external solution. The rate-determining step was estimated to be the surface reaction between protonated adsorbed methanol and isobutylene [3-13]. This simplifies the rate of consumption for isobutylene after assuming that polar methanol molecules are preferentially adsorbed onto the ion-exchange resin catalyst (Table 1) and the fraction of unoccupied sites over the catalyst surface was small as follows [4]: a- iBu MTBE iBu 0 2 MeOH a MeOH 1            E RT a a r a e a K a (7) where iBur is the rate constant of the reaction, R is the universal gas constant, aE is the activation energy of 85.4 kJ mol-1 (for Amberlyst-15 this value varies between 71 and 93 kJ mol-1 and is influenced by not Figure 1. Values of the equilibrium constant Ka as a function of temperature. Table 1. Properties of the catalyst Amberlyst-15. Properties Specifications Porosity of Catalyst (ɛp) 0.31-0.39 Tortuosity of Catalyst (t) 0.7-1.7 Density of catalyst (kg m-3) ~2000 Physical Shape spherical beads Ionic Form hydrogen Moisture Content (g/g %) ~50 Particle Size (mm) 0.3-1.5 Functional Group RSO3H Surface Area (m2 g-1) ~45 Pore Diameter (Å) ~250 Acid Sites (eq(H+) kg-1) 4.6-5.2 MTBE SYNTHESIS REACTOR MODELLING 45(1) pp. 1–7 (2017) 3 only the type of catalyst but also by the composition of the reaction mixture); and a0 = 6.3 x 1012 mol h-1 g-1. The effective diffusion coefficient and activation energy of methanol over Amberlyst-15 at 60 oC was estimated to be 2.3·10-9 m2 s-1 and ~35.4 kJ mol-1, respectively [4]. The given value of the effective diffusion coefficient is slightly lower than the one Rehfinger and Hoffmann reported (3.5·10-9 m2 s-1) [3]. The catalysts Kastel CS 381 and Amberlyst CSP used in the synthesis of MTBE possess similar rate constants to Amberlyst-15, whereas Amberlyst XE 307 and Duolite ES 276 have significantly higher; and Duolite C16P and Duolite C26 have substantially lower values for their rate constants [3-15]. Every resin undergoes deactivation, if sodium ions are exchanged for their protons. Due to the acidic nature of resins, the activity is strongly reduced by the presence of basic substances and/or salts in the reaction mixture. The heat of reaction, ∆Hº(MTBE), in the liquid phase at 25 ºC was reviewed by Iborra et al. and found to be within the range of -34 to -40 kJ mol-1 [5]. The rate for dimerization of isobutene (DIB formation) [4, 10] can be calculated as follows: -66.7kJ/mol 1 1 2- 333 iBu DIB 0 ad MeOH iBu           R T a r k e K a a (8) where the unit of rDIB is mmol s-1 eq-1, Kad is the ratio of adsorption equilibrium constants, and k0 is the frequency factor of the kinetic constant. 2.4. Reactor and Process Generally, the multi-tubular packed-bed reactor is used exclusively as a front-end for the reactor performing the synthesis of MTBE, where the catalyst is lodged in the tubes (~10,000 tubes of diameter ~30 mm, length ~6 m, ɛ = 0.4). The cooling water flows in the shell side and can either flow co- or counter-current. The reactor column is used solely for finishing and exploits the principle of catalytic distillation possessing the normal fractionation trays with reactive trays (where the catalyst is packed) [2, 16]. Therefore, the maximum degree of isobutene conversion was achieved and reactants were separated from the product simultaneously, as MTBE has a much higher boiling temperature. Reactor parameters used for the modelling are given in Table 2. Almost all the commercial etherification technologies use similar sections with regard to operation and separation, but are different in terms of type of reactors, numbers and process schemes [16-17]. The MTBE plant consists of a multi-tubular packed-bed reactor, and reactive distillation and methanol recovery sections. There are two separation towers after each reactor that recover the C4/methanol azeotrope from the top and MTBE from the bottom, followed by the washing tower where water removes the methanol from the C4, and then a distillation tower to separate water and methanol. 3. Reactor Modelling Classical models of multi-tubular packed-bed reactors (pseudo-homogeneous one-dimensional) have been extensively discussed in the literature [1, 3-5]. In this study, a custom model of an MTBE water-cooled multi- tubular packed-bed reactor was built using an Aspen- based platform (ACM) with the ultimate objective of developing a complete process flow-sheet using the same method as in Refs. [18-19]. Model equations have been developed based on the following assumptions:  steady-state operation;  plug flow with no axial mixing;  one dimensional as no temperature and composition gradients exist in a radial direction (perfect radial mixing);  pseudo-homogeneous conditions for fluid - solid phase interactions: transport limitations in terms of catalyst particles are taken into account by using the concept of an effectiveness factor;  heat transfer coefficient of the liquid phase to coolant (constant value along the reactor axis) is implemented as a sub-model. The heat transfer coefficient was evaluated under various changeable process parameters as defined the coolant inlet temperature, inlet flow rate, inlet pressure and inlet composition;  pressure drops according to the Ergun equation in the model. 3.1. Mass and Energy Balance The steady-state material balance in the liquid phase expressed in terms of vectors of the molar flow rates of the components in z direction is d d  A z f r and T b kef   r S r (9) where r is the vector of the rate of production or consumption per unit volume of a component, S is the stoichiometric matrix, rkef is the vector of effective reaction rates, A is the cross-sectional area of the reactor, f is a molar flow rate vector of the component, and b the bulk density of the catalyst. The liquid side of the heat balance equation under steady-state conditions is as follows: Table 2. Specifications of the reactor system. Specification Units Value Reactor Length m ~8 Reactor Diameter m ~3.5 LHSV h-1 ~2-9 Porosity - ~0.35 Feed Inlet Temperature oC ~40 i-C4/MeOH Ratio molar ~1.1 Pressure bar ~10 Recycling Ratio % ~15 NAWAZ Hungarian Journal of Industry and Chemistry 4    T R kef f t c T p        cA h D T TdT dz h f c r (10) where ΔhR is the vector concerning the heat of reaction, cp is the vector concerning the molar specific heat of a component, T and Tc are the liquid phase and coolant absolute temperatures, respectively, hfc is the liquid-to- coolant heat transfer coefficient, and Dt is the diameter of the reactor tube. The shell side of the heat balance equation under steady-state conditions is as follows:  fc t cc Vcool cool s,cool         h D T TdT dz F c (11) where FVcool is the volumetric flow rate of the coolant, cool is the density of the coolant, and cs,cool is the specific heat capacity of the coolant. In terms of modelling, the negative sign stands for the counter- current and the positive sign for the co-current. The boundary conditions at the inlet of the reactor in terms of the material and energy balance differential equations are z = 0, Fi(0) = F0,i , T(0) = T0, Tc(0) = Tcool,in and Tc(RL) = Tcool,in. 3.2. Effectiveness Factor The effectiveness factor for the synthesis of MTBE has been evaluated on the basis of data concerning the generalized Thiele modulus (φg) [3]:  = tanh(g) / g . (12) In terms of the pseudo-homogeneous model, the effectiveness factor is expressed as a function of temperature and conversion: 2 3x 1 4 1 conversion % 1.1 100                x RTx e x . (13) The parameters x1 – x4 have been evaluated by minimizing the absolute error between evaluations by Zhang et al. [4] and the values predicted by Eq. 13. The accuracy of the prediction is shown in Fig.2. 3.3. Overall Heat Transfer Coefficient The overall heat transfer coefficient, taking into account the small value of the wall thickness, is evaluated as follows: fc te ti fouling fw pipe shell 1 1 1 2           h D D r h h (14) where Dti and Dte are the internal and external diameters of the reactor tube, respectively, λpipe is the thermal conductivity of the reactor tube, hfw is the fluid-to-wall heat transfer coefficient, hshell is the shell side (wall-to- cooling water) heat transfer coefficient, and rfouling is the fouling factor. The evaluation of the fluid-to-wall heat transfer coefficient was conducted on the basis of a newly proposed correlation for the estimation of the total (wall and effective thermal conductivity effects) heat transfer coefficient [16]: 0.643 0.333 f f f ti p 1.37 Nu 3.87 3.77 exp Re Pr                      D d .(15) In this case, both Nuf and Ref are expressed based on particle diameters. The heat transfer coefficient of the shell side is expressed as a function of Nu and Re and are both based on the external diameter of the tube, where ɛφ stands for a correction factor accounting for the angle between the coolant stream and reactor tubes: Nu s  0.4 e  Re s 0.6Pr s 0.333 . (16) 3.4. ACM Property Estimations The physical properties of components and mixtures like viscosity, density and specific heat, were estimated by Aspen Properties (using UNIFAC - Dortmund property package). The “.appdf” file prepared by Aspen Properties is the one called in ACM for seamless transfer of properties. The kinetic sub-model helps in terms of the calculation of reaction heats using the Aspen Properties database. Figure 2. Effectiveness factor for the MTBE synthesis reaction based on data from Ref. [4]. MTBE SYNTHESIS REACTOR MODELLING 45(1) pp. 1–7 (2017) 5 3.5. Performance Parameters Reactor performance is evaluated by the following equations: key_reactant key_reactant 0 z key_reactant 0 Conversion( )   F F z F (17) key_product key_product z 0 key_reactant key_reactant 0 z Selectivity( )    F F z F F (18) Yield( ) Conversion( ) Selectivity( ) z z z (19) where z stands for the axial coordinate. 4. Model Predictions and Analysis ACM is the preferred choice for robust reactor modelling. Its equation-oriented modelling platform can be easily exported to Aspen Plus process flow-sheets. The coding of a multi-tubular packed-bed reactor in terms of the synthesis of MTBE was conducted in a modular form. The main reactor model defines the constitutive equations in the bulk phase and sub-models in terms of the reaction kinetics, the estimation of the heat transfer coefficient, and the development of the pellet design. The normal operating conditions for the primary reactor were selected as the base case in terms of modelling (Table 3), where MTBE productivity is ~50 T h-1. The molar concentration profile of the reactants and product along the axis of the reactor is shown in Fig.3. The full length of the reactor is used for the synthesis of MTBE. Fig.4 shows the temperature profiles of the reactor and coolant. The temperature sensitivity is controlled by keeping the temperature difference between the coolant and feed temperatures constant (Fig.5). It was observed that due to the exothermic nature of the reaction, the maximum temperature increased significantly with the Table 3. Typical base-case operating conditions and results for the primary reactor. Specification Units Value tube length m 6 tube diameter (ID) mm 21 number of tubes - 10,000 porosity - 0.33 feed inlet temperature ºC 50 isobutene/MeOH ratio molar 1.1 pressure bar 8 coolant inlet temperature ºC 45 feed flow rate kmol h-1 2,000 isobutene mole fraction - 0.35 ∆P bar 0.723 conversion % 84.7 MTBE selectivity % 97.86 Figure 3. Molar concentration profile along the length of the reactor. Figure 4. Temperature profiles of the reactor and coolant. Figure 5. Temperature sensitivity by varying the feed temperature and keeping constant the temperature difference between the coolant and feed temperatures. NAWAZ Hungarian Journal of Industry and Chemistry 6 increase in reaction temperature. At a constant pressure and feed flow rate, the increase in the reaction temperature led to an outer mass transfer limitation that strongly affects the rate of conversion. At the same time the selectivity also decreased due to a significant increase in the maximum temperature of the reactor (product degradation) and the maximum rate of conversion was achieved in almost ~30% of the reactor. The isobutylene/methanol molar ratio is one of the key operating parameters and is always kept higher than its stoichiometric value. Table 4 shows that by deviating from the stoichiometric ratio the effect on the rate of conversion and selectivity also decreases with the increase in the isobutylene/methanol molar ratio. The predictions according to reactor modelling were found to be in good agreement with data from the literature. The reaction must reach an equilibrium at the rear end of the lead reactor. The estimated effectiveness factor is 0.75 which was also used in the model. The possible explanation for the occurrence of minor discrepancies could be the use of a different catalyst or errors in terms of experimental measurements. It was observed that the maximum degree of productivity corresponded to an isobutene/MeOH molar ratio of close to 1 and a feed inlet temperature of between 50 and 55 ºC. However, it is necessary to synchronize this with the secondary reactor and other unit operations. 5. Conclusion An ACM-based pseudo-homogeneous model with regard to the synthesis of MTBE using an industrial multi-tubular packed-bed (methyl-tert-butyl-ether) reactor has been developed for enhancing operation strategies. The UNIFAC - Dortmund property package was used for reactor modelling. The effectiveness factor was implemented as a function of the reaction temperature. The model was validated using data from the literature. Various sensitivity evaluations were conducted to determine the operational optimization and maximization of MTBE production under different operating parameters. The model was able to predict reaction behaviour and produce temperature and concentration profiles along the length of the reactor. Sensitivity studies were able to calculate the optimum temperature, feed flow rate, methanol/isobutylene ratio as well as heat removal rate of the reactor and thus provide insights into a reasonable operational strategy. SYMBOLS A cross-sectional area of the reactor a activity aMeOH, aiBu, aMTBE, ainert activities of methanol, isobutene, MTBE and inert isobutane, respectively a0 6.3 x 1012 mol h-1 g-1 cp molar specific heat vector of a component cs,cool specific heat capacity of the coolant Dt diameter of the reactor tube Dti, Dte internal and external diameters of the reactor tube, respectively Ea activation energy in kJ mol-1 FVcool volumetric flow rate of the coolant f molar flow rate vector of a component hfc liquid-to-coolant heat transfer coefficient hfw fluid-to-wall heat transfer coefficient hshell shell side (wall-to-cooling water) heat transfer coefficient ΔhR heat of reaction vector γ activity coefficient Ka equilibrium constant (based on activities) Kad ratio of adsorption equilibrium constants KMeOH, KiBu, KMTBE, Kinert adsorption equilibrium constants for methanol, isobutene, MTBE and inert isobutane, respectively k0 frequency factor of the kinetic constant kr reaction rate constant λpipe thermal conductivity of the reactor tube R universal gas constant r vector of the rate of production or consumption per unit volume of a component rkef vector of effective reaction rates rfouling fouling factor b bulk density of the catalyst cool density of the coolant S stoichiometric matrix T , Tc absolute temperatures of the liquid phase and coolant, respectively x mole fraction Acknowledgement The author gratefully acknowledges the technical support of Prof. Dr. Ing. Teodor Todinca at the University Politechenica Timisoara in Romania. REFERENCES [1] Nawaz, Z.: Light alkane dehydrogenation to light olefin technologies: A comprehensive review, Rev. Chem. Engng., 2015 31(5), 413–436 DOI 10.1515/revce-2015-0012 Table 4. Dependence of conversion (%), selectivity (%) and the maximum temperature of the reactor (Tmax) on the isobutene/methanol molar ratio (iBu/MeOH). iBu/MeOH 0.8 0.9 1.0 1.1 1.2 1.3 conversion 84.7 85.4 85.5 84.6 80.7 76.7 selectivity 98.1 98.1 98.1 97.9 97.1 95.4 Tmax, ºC 56.1 56.5 56.9 57.7 59.5 62.0 MTBE SYNTHESIS REACTOR MODELLING 45(1) pp. 1–7 (2017) 7 [2] DiGirolamo, M.; Sanfilippo, D.: Etherification: Process to improve the quality of distillates, in Beccari M.; Romano, U.; eds. Encyclopaedia of hydrocarbons (ENI S.p.A and Istituto della Enciclopedia Italiana fondata da Giovanni Treccani S.p.A, Roma, Italy) 2006 Vol. II. pp. 193–209 [3] Rehfinger, A.; Hoffmann, U.: Kinetics of MTBE liquid phase synthesis catalyzed by ion-exchange resin. I. Intrinsic rate expression in liquid phase activities, Chem. Eng. Sci., 1990 45(6), 1605–1617 DOI 10.1016/0009-2509(90)80014-6 [4] Zhang, T.; Datta, R.: Integral analysis of MTBE synthesis kinetics, Ind. Eng. Chem. Res., 1995 34, 730–740 DOI 10.1021/ie00042a004 [5] Iborra, M.; Tejero, J.; Ben El-Fassi, M.; Cunill, F.; Izquierdo, J.F.; Fite, C.: Experimental study of the liquid phase simultaneous synthesis of MTBE and TBA, Ind. Eng. Chem. Res., 2002 41, 5359–5365 DOI 10.1021/ie010640q [6] Chen, F.; Huss, R.S.; Doherty, M.F.; Malone, M.F.: Multiple steady-states in reactive distillation: kinetic effects, Comp. Chem. Engng., 2002 26, 81– 93 DOI 10.1016/S0098-1354(01)00750-5 [7] Lisal, M.; Smith, W.R.; Nezbeda, I.: Molecular simulation of multicomponent reaction and phase equilibria in MTBE ternary system, AIChE J., 2000 46(4), 867–875 DOI 10.1002/aic.690460419 [8] Isla, M.; Irazoqui, H.: Modeling, analysis and simulation of MTBE reactive distillation column, Ind. Eng. Chem. Res., 1996 35, 2696–2708 DOI 10.1021/ie9505930 [9] Venimadhavan, G.; Malone, M.F.; Doherty, M.F.: Bifurcation study of kinetic effects in reactive distillation, AIChE J., 1999 45(3), 546–556 DOI 10.1002/aic.690450311 [10] Oktar, N.; Murtezaoglu, K.; Dogu, T.; Dogu, G.: Dynamic analysis of adsorption equilibrium and rate parameters of reactants and products in MTBE, ETBE and TAME production, Can. J. Chem. Eng., 1999 77, 406–412 DOI 10.1002/cjce.5450770229 [11] Girolamo, M.; Tagliabue, L.: MTBE and alkylate co-production: fundamentals and operating experience, Catal. Today, 1999 52, 307–319 DOI 10.1016/S0920-5861(99)00084-X [12] Ancilotti, F.; Fattore, V.: Oxygenate fuels: Market expansion and catalytic aspect of synthesis, Fuel Proc. Technol., 1998 57, 163-194 DOI 10.1016/S0378- 3820(98)00081-2 [13] Girolamo, M.; Lami, M.; Marchionna, M.; Pescarollo, E.; Tagliabue, L.; Ancilotti, F.: Liquid phase etherification/dimerization of isobutene over sulfonic acid resins, Ind. Eng. Chem. Res., 1997 36, 4452–4458 DOI 10.1021/ie9700932 [14] Miracca, I.; Tagliabue, L.; Trotta, R.: Multitubular reactors for etherifications, Chem. Engng. Science, 1996 51, 2349–2358 DOI 10.1016/0009-2509(96)00091-7 [15] Panneman, H.J.; Beenackers, A.: Influence of the isobutene/methanol ratio and of the MTBE content on the reaction rate of the synthesis of MTBE, Ind. Eng. Chem. Res., 1995 34, 3817–3825 DOI 10.1021/ie00038a019 [16] Mariani, N.J.; Martinez, O.M.; Barreto, G.F.: Evaluation of heat transfer parameters in packed beds with cocurrent downflow of liquid and gas, Chem. Engng. Science, 2001 56, 5995–6001 DOI 10.1016/S0009-2509(01)00225-1 [17] Elkanzi, E.M.: Optimization of MTBE synthesis in a packed-bed reactor system, Chem. Engng. Proc., 1996 35, 131–139 DOI 10.1016/0255-2701(95)04122-2 [18] Nawaz, Z.: Dynamic modeling of CATOFIN® packed-bed isobutane dehydrogenation reactor for operational optimization, Int. J. Chem. Reactor Engng., 2016 14(1), 491–515 DOI 10.1515/ijcre-2015- 0087 [19] Nawaz, Z.: Heterogeneous reactor modeling of an industrial multitubular packed-bed ethylene oxide reactor, Chem. Engng. Technol., 2016 39(10), 1845–1857 DOI 10.1002/ceat.201500603 HUNGARIAN JOURNAL OF INDUSTRY AND CHEMISTRY Vol. 45(2) pp. 9–12 (2017) hjic.mk.uni-pannon.hu DOI: 10.1515/hjic-2017-0013 MODELLING THE PARTIAL DEMINERALIZATION PROCESS OF COW MILK BY SUPERPRO DESIGNER ATTILA CSIGHY,* ANDRÁS KORIS, AND GYULA VATAI Department of Food Engineering, Szent István University, Ménesi út 44, Budapest, 1118, HUNGARY Milk and dairy products contain a number of biologically active compounds (proteins, lipids, vitamins and minerals) that are essential for human nutrition. The most common procedures for demineralization are based on ion exchange-, nanofiltration- and electrodialysis-based technologies. In this study, the application of membrane filtration-based partial demineralization of cow milk was investigated and the process modelled. Using design equations, the partial demineralization process was designed and the economy of the process calculated. The modelling and simulation of the partial demineralization process was carried out by the SuperPro Designer programme. As the first step the unit operations of the demineralization technology were defined using the tools of the programme. The SuperPro Designer possesses industrial tools with reactor models, chemical components, a database of mixtures, and price estimations. By analysing the influence of the operation parameters, the feasibility of the proposed process was investigated. From the results of the modelling it can be concluded that the partial demineralization process can be successfully implemented, achieving the expected demineralization rates with a relatively good payback time of two years. Keywords: partial demineralization, SuperPro Designer, modelling, economic analysis 1. Introduction Dairy products play an important role in the health of humans as milk contains a number of biologically active compounds (proteins, lipids, vitamins and minerals). Therefore, the consumption of milk and dairy products is highly recommended [1-2]. Milk is the main raw material in the cheese and casein industry. The disposal of whey creates a major environmental problem for the cheese industry due to the high amount of organic compounds it contains. In the cheese industry to manufacture 1 kg of cheese, 9 kg of whey is produced as a by-product. Whey possesses a large ‘biochemical oxygen demand’ (BOD) value, therefore, treatment is required before it is released into the environment or recycled. Suárez [3] observed that the nanofiltration membrane is an effective medium for the demineralization of whey. The demineralization efficiency depends on the transmembrane pressure and the volume concentration ratio (VCR). Experimentally, monovalent ions exhibit the highest degree of permeation. [3] Whey is a watery, dilute liquid, which generally contains 0.8-1.0% total protein, 4.5-5.0% lactose, 0.5- 0.7% minerals and 93-94% water. Whey can be divided into two different compositions, namely sweet and acid whey [4]. Another process as part of the treatment of whey is demineralization. Upon demineralization, whey *Correspondence: csighy.attila@gmail.com can be used for manufacturing drinks, desserts or ice- cream products. A novel alternative process for demineralization is nanofiltration. Nanofiltration is a membrane separation method that lies between ultrafiltration and reverse osmosis. Nanofiltration membranes can maintain monovalent ions (NaCl) and also organic compounds between 300 and 100 in the Dalton range. Nanofiltration is an effective technique to remove salts, while preserving valuable components [5]. In order to examine the feasibility of the process, technological and economic experiments were carried out by the SuperPro Designer software [6]. This programme is widely used in the pharmaceutical, biotechnology and food industries. The SuperPro Designer is capable of preparing technological and economic documentations and reports about the modelled process. The programme possesses an industrial tool, where the unit operations (reactions, solid/liquid separation, tanks) can be prepared [7]. The purpose of this study is to model the demineralization process using the SuperPro Designer programme and by analyzing the influence of the operation parameters, the feasibility of the proposed process was investigated. 2. Experimental 2.1. Materials and Methods In this study, one ultrafiltration (UF) and two nanofiltration (NF) membranes were used. The first step was the pre-concentration of milk using the UF CSIGHY, KORIS, AND VATAI Hungarian Journal of Industry and Chemistry 10 membrane. A Schumasiv type 100 nm pore size membrane was used for the pre-concentration phase, a Schumasiv type 5 nm pore size membrane for nanofiltration of the ultrafiltration retentate, and a Membralox type 5 kDa membrane for nanofiltration of the ultrafiltration permeate. The experiments were carried out using ultra- and nanofiltration units in a laboratory unit designed by the Department of Food Engineering. The effective area of the membranes was 0.005 m2. The milk sample was circulated at a constant temperature (21±1 ºC) maintained by a thermostat. The optimal working parameters were measured during an experimental design in both the UF and NF processes. During the UF process the transmembrane pressure was 1.5 bars and the recirculation flow rate was 150 dm3 h-1. In both NF processes, the transmembrane pressure was 2 bars and the recirculation flow rate was 200 dm3 h-1. The transmembrane pressure and the recirculation flow rate were controlled by regulating valves. 2.2. Process simulation The partial demineralization process was modelled using SuperPro Designer software version 8.5 by Intelligen, Inc. The main target task was partial demineralization. The SuperPro Designer software possesses an industrial tools section, where models of reactors, chemical components, a database of mixtures, price estimations and economic evaluations can be used. The parameters used were calculated from laboratory experiments as shown in Table 1. Firstly, The SuperPro Designer tool was used to install the unit operations. The first unit is the ultrafiltration system. The raw milk enters the system and is separated into its permeate and retentate. In the next step, the two fractions are transferred to each diafiltration unit. Another major step is the definition of essential material streams for milk, the addition of distilled water to the diafiltration unit and the provision of electricity. The diafiltrated, partially demineralized milk and whey components are then sent to the storage tank. The last part of the process is the packaging system, where the products are completed. The demineralized whey is finally loaded into a truck. The mass of the filled entity is 40 metric tons (MT). The basic process flow diagram of the combined partial demineralization of cow milk is shown in Fig. 1. The composition of the milk was defined as illustrated in Table 2. In the next step, the transmembrane pressure, flow rate and flux as operating parameters were set and the cost of purchase assigned to each operating unit. The annual operating time is 300 days for this process. The cost of purchase of the ultrafiltration system was set at $100k, the two diafiltration units at $75k, the storage Table 1. The characteristics of the membranes used. Membrane type Process parameters Pressure, bar Flow rate, dm3 h-1 UF (100 nm) 1.5 150 NF (5 kDa) 2.0 200 NF (5 nm) 2.0 200 Figure 1. Process flow diagram of the demineralization of cow milk. Table 2. The composition of the cow milk considered Ingredient’s name Mass (%) Casein 3.4 Fats 3.4 Lactose 5.2 Sodium Chloride 1.0 Water 87 MODELLING THE PARTIAL DEMINERALIZATION PROCESS OF COW MILK 45(2) pp. 9–12 (2017) 11 tanks at $75k, and the packaging unit at $50k. In addition, the cost of raw materials was set at $1/kg, of packaging material at $0.1/piece (100g), and of water at $0.1/kg. The product must be sold at a price that is in line with the market conditions (est. $2.2/kg). Each unit operation needs a minimum of one operator, thus the cost of labour was set at a basic rate of $10 per hour. The electricity, cooling and heating energy demands were estimated by the programme (Table 3). 3. Results and Analysis Upon the establishment of costs, technological and economic simulations were executed. The summary of the run simulations for the partial demineralization process yielded the relevant costs, revenue and payback time as shown in Table 4. According to the simulation results, the annual total revenue is higher than the operating costs. The production costs are given with and without amortization. Table 5 summarizes the indices of the project. The ‘payback time’ or ‘return on investment’ (ROI) for the partial demineralization process in this case was calculated for four years. The ROI is defined by Eq. (1): ROI (%) = annual net profit capital cost ·100 (1) Other important parameters were the ‘internal rate of return’ (IRR) before and after tax. The IRR is a discount value when the ‘net present value’ (NPV) is zero. The NPV is determined by calculating the costs and benefits of the technological investment. The NPV is defined by Eq. (2): NPV (USD) = ∑ �t (���)t − �� (2) where Ct is the cash flow, t is the lifetime in years, d is the discount rate, and C0 is the initial investment. If the NPV remains positive, the process will be economically viable [8]. The costs of materials and supplies are shown in Table 6 with the most expensive item being the raw material (cow milk) while the cost of packaging materials is far less. Table 7 presents the total cost of the plant for the demineralization process. In addition to purchasing the equipment, other costs were the piping, instrumentation, and electrical, building and construction fees. According to the simulated revenues of the partial demineralization process, the production of milk and lactose solution cost $2.2 and $0.3 per kg, respectively. Assuming that the demineralized whey is transferred into a truck, it can be sold to food industries for $40 per metric ton. The composition of the demineralised whey in the form of lactose solution is 4.3% lactose, 0.8% NaCl and 95% water, which is suitable for the production of candy, yoghurt and ice cream. It is also important to consider the cost of waste treatment. The SuperPro Designer generated two different costs, namely those of utilities ($25k) and transportation ($874k). 4. Conclusion This study examined the feasibility of the partial demineralization process which was successfully Table 5. Summary of project indices (IRR = internal rate of return, NPV = net present value). gross margin, % 6.78 return on investment, % 24.90 payback time, years 4.02 IRR before tax, % 28.36 IRR after tax, % 18.20 NPV (at 7.00%) 5,381,025 Table 6. Summary of the costs of materials (in USD per kg annually). raw materials unit amount cost milk $1.00 21,766,316 21,766,316 water $0.10 10,879,641 1,087,964 packaging material $0.20 10,800,661 2,160,132 total - - 25,014,412 Table 7. Total cost of the plant (in thousands of USD). equipment purchase $563k installation $266k process piping $197k instrumentation $225k insulation $17k electrical $56k buildings $253k yard improvements $84k auxiliary facilities $225k engineering $472k construction $66k Table 3. Cost estimates for energy supplies Energy supplies Price Steam (High P) [242 ℃] $10/MT Steam [152 ℃] $6/MT Chilled Water [5 ℃ - 10 ℃] $0.4/MT Cooling Water [25 ℃ - 30 ℃] $0.05/MT Std. Power $0.1/kWh Table 4. Summary of economy indices (MP stands for “flow of discrete entity”). total investment $6,279,438 total annual revenue $30,315,042 annual operating cost $28,258,168 annual unit production reference rate (MP entity) $10,800,660 unit product cost (per MP entity) including depreciation $2.62 excluding depreciation $2.59 CSIGHY, KORIS, AND VATAI Hungarian Journal of Industry and Chemistry 12 modelled by the SuperPro Designer software. According to an economic evaluation, the net present value, return on investment, and internal rate of return were $5.3Million for four years, 28% and 18%, respectively. The revenues of the products were $2.2 per unit and $0.3 per kg (or $40 per metric ton) for the milk and lactose solution, respectively. Future research will focus on the optimization of costs and exploration of alternative ways of recycling whey. Acknowledgement The research was supported by the Doctoral School of Food Sciences at Szent István University. REFERENCES [1] Csapo, J.; Csapone, K.Zs.: Milk and Dairy Products in Food Consumption (Mezőgazdasági Kiadó, Budapest) 2002 (in Hungarian) ISBN 963 9358 68 1 [2] Mass, S.; Lucot, E.; Gimbert, F.; Crini, N.; Badot, P.M.: Trace metals in raw cow’s milk and assessment of transfer to Comté cheese, Food Chem., 2011 129(1), 7–12 DOI: 10. 1016/j.foodchem.2010.09.034 [3] Suárez, E.; Lobo, A.; Álvarez, S.; Riera, F.A.; Álvarez, R.: Partial demineralization of whey and milk ultrafiltration permeate by nanofiltration at pilot-plant scale, Desalination, 2006 198(1-3), 274–281 DOI 10.1016/j.desal.2005.12.028 [4] Tsakali, E.; Petrotos, K.; Alessandro, A.D.; Goulas, P.: A review on whey composition and methods used for its utilization for food and pharmaceutical products, Proc. 6th Int. Conf. Simul. Modelling Food Bioind., 2010 (CIMO Research Centre, Bragança, Portugal) pp. 195-201 [5] Pan, K.; Song, Q.; Wang, L.; Cao, B.: A study of demineralization of whey by nanofiltration membrane, Desalination, 2011 267(2-3), 217–221 DOI 10.1016/j.desal.2010.09.029 [6] Mel, M.; Yong, A.S.H.; Avicenna; Ihsan, S.I.; Setyobudi, R.H.: Simulation Study for Economic Analysis of Biogas Production from Agricultural Biomass, Energy Procedia, 2015 65, 204-214 DOI 10.1016/j.egypro.2015.01.026 [7] Flora, J.R.V.; McAnally, S.A.; Petrides, D.: Treatment plant instructional modules based on SuperPro Designer® v.2.7, Environ. Model. Software, 1998 14(1), 69–80 DOI 10.1016/S1364- 8152(98)00059-0 [8] Kwan, T.H.; Pleissner, D.; Lau, K.Y.; Venus, J.; Pommeret, A.; Lin, C.S.: Techno-economic analysis of a food waste valorization process via microalgae cultivation and co-production of plasticizer, lactic acid and animal feed from algal biomass and food waste, Biores. Technol., 2015 198, 292–299 DOI 10.1016/j.biortech.2015.09.003 HUNGARIAN JOURNAL OF INDUSTRY AND CHEMISTRY Vol. 45(2) pp. 13–18 (2017) hjic.mk.uni-pannon.hu DOI: 10.1515/hjic-2017-0014 INVESTIGATIONS OF THE TlInP2Se6–In4(P2Se6)3 SYSTEM AND ITS OPTICAL PROPERTIES VALERIA TOVT, 1 IGOR BARCHIY, 1 * MICHAL PIASECKI, 2 IWAN KITYK, 3 AND ANATOLII FEDORCHUK 4 1 Department of Chemistry, Uzhgorod National University, Pidgirna St. 46, 88000 Uzhgorod, UKRAINE 2 Institute of Physics, Jan Dlugosz University, Armii Krajowej 13/15, 42-200 Częstochowa, POLAND 3 Faculty of Electrical Engineering, Częstochowa University of Technology, Dabrowskiego 69, 42201 Częstochowa, POLAND 4 Department of Inorganic and Organic Chemistry, Lviv National University of Veterinary Medicine and Biotechnologies, Pekarska St. 50, 79010 Lviv, UKRAINE The equilibrium phases were investigated and the corresponding phase diagram constructed for the TlInP2Se6– In4(P2Se6)3 system from physical and chemical analyses, namely differential thermal analysis (DTA), X-ray diffraction (XRD), and microstructural analysis (MSA). It was established that this system belongs to the eutectic type and is characterized by the formation of boundary solid phases containing complex compounds. Single crystals of the compounds TlInP2Se6 and In4(P2Se6)3 were grown using the Bridgman method. Both crystals were found to exhibit diffuse reflection spectra and photoinduced dependence of birefringence at various IR wavelengths generated by CO2 laser irradiation. Birefringence properties were investigated using the Senarmont method. Keywords: phase diagram, solid solution, crystal structure, optical properties, direct-gap semicon- ductor, indirect-gap semiconductor, photoinduced birefringence 1. Introduction Compounds with the formula M2P2Se6 possess promis- ing magneto-electric, piezoelectric, electro-optical, and thermoelectric properties that indicate their suitability as functional materials in optoelectronics [1-2]. Due to their crystal structure, they exhibit anisotropy in terms of their physical properties. In a multilevel structure of M2P2Se6 compounds, metal cations and pairs of phos- phorous atoms occupy the octahedral positions between planes of selenium atoms. This structure is characterized by its layered arrangement of atoms, which contributes to the formation of a dipole moment between the layers of cationic and anionic groups. The replacement of the metal cation М2+ by other metal cations (М+, М3+ or М4+) leads to the deformation of the structure [3-4], changes the magnitude of the dipole moment and, con- sequently, its physical properties. The Tl2Se–In2Se3–“P2Se4” ternary system is com- posed of binary Tl2Se–In2Se3, Tl2Se–“P2Se4” and In2Se3–“P2Se4” systems. The Tl2Se–In2Se3 system is characterized by the formation of two intermediate ter- nary compounds: TlInSe2 melts congruently at 1023 K and TlIn5Se8 is formed according to the peritectic reac- tion L + In2Se3  TlIn5Se8 at 1029 K [5-6]. In the sys- *Correspondence: i_barchiy@ukr.net tem Tl2Se–“P2Se4” with a ratio of 2 to 1, interoperable components form the compound Tl4P2Se6 which pos- sesses a congruent nature of melting at 758 K [7]. The In2Se3–“P2Se4” system is characterized by the formation of the compound In4(P2Se6)3 in a syntectic reaction of L1 + L2  In4(P2Se6)3 at 880 K [8]. In the Tl2Se– In2Se3–“P2Se4” system at the intersection of incisions, the phases Tl4P2Se6–In4(P2Se6)3 and TlInSe2–“P2Se4” form the complex compound TlInP2Se6 [9]. 2. Experimental Ternary Tl4P2Se6 and In4(P2Se6)3 compounds were pre- pared by melting stoichiometric quantities of binary Tl2Se with elementary indium, phosphorous and seleni- um under a vacuum of 0.13 Pa in quartz ampoules using a single temperature method. In all syntheses, compo- nents were used that possess a purity greater than 99.999 %. The maximum temperatures of synthesis were 993 and 893 K for In4(P2Se6)3 and TlInP2Se6, re- spectively. The rate of heating up to the maximum tem- perature was 50 K h-1. The melts were maintained at the maximum temperature for 72 hours. Cooling was per- formed at a rate of 50 K h-1 down to an annealing tem- perature of 573 K. The linearity of the heating and cool- ing processes was achieved by a RIF-101 temperature controller. The homogenization process occurred over 120 hours. Identification of the complex compounds and alloys was conducted by differential thermal analy- TOVT, BARCHIY, PIASECKI, KITYK, AND FEDORCHUK Hungarian Journal of Industry and Chemistry 14 sis (DTA) (PRA-01, chrome-alumina thermocouple 5 K), X-ray diffraction (XRD) (DRON-3 diffractometer, CuKα radiation, Ni filter) and microstructural analysis (MSA) (metallurgical microscope Lomo Metam R-1). Crystal structural calculations were conducted using the software package WinCSD [10]. Optical properties were investigated using an SF-18 spectrophotometer within the wavelength range of 400 – 750 nm. A CO2 laser was used for photoinduced electrons in samples employing 200 ns pulses with a pulse repetition fre- quency of about 10 Hz, a fundamental frequency of 10.6 μm and a frequency doubling of 5.3 μm beams. The birefringence was measured using a Er:glass cw laser at 1540 nm by application of the Senarmont method. 3. Results and Analysis 3.1. Phase diagram of the TlInP2Se6– In4(P2Se6)3 system The TlInP2Se6–In4(P2Se6)3 system is a quasi-binary sec- tion of the Tl2Se–In2Se3–“P2Se4” ternary system (Figs.1 and 2). It belongs to the eutectic type (V-type diagram by Rozeboom). The complex compounds TlInP2Se6 and In4(P2Se6)3 melt congruently at 875 K and 963 K, re- spectively. TlInP2Se6 is characterized by two polymor- phic transformations ltTlInP2Se6  mtTlInP2Se6 at 680 K and mtTlInP2Se6  htTlInP2Se6 at 711 K. The prefixes lt–, mt– and ht– represent low–, medium–, and high–temperature modifications, respectively. In4(P2Se6)3 is also characterized by two polymorphic transformations ltIn4(P2Se6)3  mtIn4(P2Se6)3 at 665 K and mtIn4(P2Se6)3  htIn4(P2Se6)3 at 903 K. When the temperature rises above 791 K, an invariant eutectic process is observed L  htTlInP2Se6 + mtIn4(P2Se6)3 (in the presence of 15 mol% In4(P2Se6)3). The system is described by the sequence of the ef- ficient peritectic processes htTlInP2Se6 + mtIn4(P2Se6)3  mtTlInP2Se6 (714 K) and mtTlInP2Se6 + mtIn4(P2Se6)3  ltTlInP2Se6 (689 K) based on the pol- ymorphic transformation of TlInP2Se6. The polymor- phism of In4(P2Se6)3 produces metatectic htIn4(P2Se6)3  L + mtIn4(P2Se6)3 (884 K) and eutectic mtIn4(P2Se6)3  ltTlInP2Se6 + ltIn4(P2Se6)3 (652 K) processes. Re- gions of homogeneity in solid solutions, based on the batched complex selenides during annealing at a tem- perature of 573 K, do not exceed 10 mol%. 3.2. Crystal structure of the compounds In4(P2Se6)3 and TlInP2Se6 The crystal structures of the compounds TlInP2Se6 and In4(P2Se6)3 were solved using the Rietveld method. As an initial model for TlInP2Se6 [2], the parameters of In4(P2Se6)3 were used [8]. Analysis of the crystalline structures of the investigated compounds (Table 1) showed that it is possible to define the structural group of the anionic group [P2Se6] 4–, which is formed by two single tetrahedra (Fig.3). Cationic atoms occupy posi- tions between the anionic groups and none are located between the layers. Figure 1. Results of the XRD analysis of the TlInP2Se6–In4(P2Se6)3 system. (I rel – Intensity, 2 theta - Angle of reflection) Figure 2. Phase diagram of the TlInP2Se6–In4(P2Se6)3 system. (1–L, 2–L+htIn4(P2Se6)3, 3–htIn4(P2Se6)3, 4–htTlInP2Se6, 5–L+mtIn4(P2Se6)3, 6–htIn4(P2Se6)3+mtIn4(P2Se6)3, 7–htTlInP2Se6, 8–htTlInP2Se6+mtIn4(P2Se6)3, 9–mtIn4(P2Se6)3, 10–htTlInP2Se6+mtTlInP2Se6, 11–mtTlInP2Se6, 12–mtTlInP2Se6+mtIn4(P2Se6)3, 13–mtTlInP2Se6+ltTlInP2Se6, 14–ltTlInP2Se6+mtIn4(P2Se6)3, 15–mtIn4(P2Se6)3+ltIn4(P2Se6)3, 16–ltTlInP2Se6, 17–ltTlInP2Se6+ltIn4(P2Se6)3, 18–ltIn4(P2Se6)3). Table 1. Crystal data of TlInP2Se6 and In4(P2Se6)3 compounds. Compound Crystal system Space group Lattice constant In4(P2Se6)3 [8] trigonal R3 h (146) a = 6.362(3), c = 19.929(6) Å In4(P2Se6)3 trigonal R3 h (146) a = 6.3808(8), c = 20.014(4) Å TlInP2Se6 [2] triclinic P-1 (2) a = 6.4310, b = 7.5002, c = 12.124 Å, TlInP2Se6 triclinic P-1 (2) α = 100.553, β = 93.735, γ = 113.451 INVESTIGATIONS OF THE TLINP2SE6–IN4(P2SE6)3 SYSTEM 45(2) pp. 13–18 (2017) 15 The structure of In4(P2Se6)3 can be derived from the structure of Sn2P2Se6 [11]. It is composed of multi- ple substitutions of the isovalent cations according to 2M2+  M4+. The crystal structure of the compound In4(P2Se6)3 can be presented based on the composition of the anionic group [P2Se6] 4– (Fig.4), in which the indi- um atoms occupy the space between the anionic groups. The second coordination environment (SCE) [12] is of cuboctahedron form. Indium cations are surrounded by a triangular environment of anionic atoms of the group [P2Se6] 4–and within the frames of its environment bonding exists with six atoms of selenium while the coordination form is octahedral (Fig.5). The structural and chemical properties of the МеІМеІІІР2Se6 compositions are related to the important role concerning the dimension of the cation on its loca- tion between the layers of the anionic [P2Se6] 4–groups. Crystallographic analysis showed that smaller cations occupy a position in the plane perpendicular to the main axis. Atoms located in a second coordination environ- ment of anionic groups in the structure of TlInP2Se6 compounds can be presented as a strongly distorted hexagonal-equivalent cuboctahedron (Fig.6). The atoms of metallic cations, located in the cavi- ties between the atoms of the anionic groups, are within an asymmetric environment (Fig.7). In3+ cations move toward tetrahedral cavities on the boundary between tetrahedral and octahedral cavities, and Tl+ cations move in the direction of the octahedral cavities. Moreover the In3+ cations are located in the same plane together with the centres of the anionic [P2Se6] 4– groups (Fig.8) and some Tl+ cations are shifted relative to the plane. Therefore, this arrangement is a source of the interesting electro-physical and optical properties of materials based on compounds of this type. 3.3. Optical response of single crystals of TlInP2Se6 and In4(P2Se6)3 The most important parameter of the energy spectra of semiconductors is the width of the band gap, Eg, which is defined by the difference in energy between the bot- tom of the conduction band, EC, and the top of the va- lence band, EV. All semiconductors can be divided into two groups. In the first group, the minimum of the con- duction band and the maximum of the valence band occupy the same point in the Brillouin zone, i.e. at an identical location in the space of quasi-moments. In this case, the optical transitions of electrons from the va- lence band to the conduction band (with the absorption of a quantum of light) and from the conduction band to the valence band (with the emission of a quantum of light) occur so that the electrons practically do not change their quasi-moments. Such transitions are char- acteristic of direct-gap semiconductors. For the second group, the absolute minimum of the conduction band and the absolute maximum of the valence band are at different points in the Brillouin zone, and optical inter- Figure 3. Structure of the anionic group [P2Se6] 4–. Figure 4. Arrangement of the polyhedra anionic group [P2Se6] 4– in In4(P2Se6)3. Figure 5. Coordination environment of the indium atoms in the structure of In4(P2Se6)3. Figure 6. Second (SCE) and nearest (NCE) coordina- tion environments of atoms in the [P2Se6] 4– anionic groups in the structure of TlInP2Se6. TOVT, BARCHIY, PIASECKI, KITYK, AND FEDORCHUK Hungarian Journal of Industry and Chemistry 16 band transitions must be accompanied by a large change in the electron quasi-moment. These are characteristic of indirect-gap semiconductors. Since the photon mo- ment is negligibly small compared with the electron quasi-moment, the latter case is possible only when the electron interacts with the phonon. According to the phase diagram, the single crystals of TlInP2Se6 and In4(P2Se6)3 were grown using the Bridgman method in two vertical zone furnaces. Exper- imental studies of optical spectra in the absorption re- gion yielded information on the energy spectrum of electrons near the edges of the conduction band and band gap. Studies concerning the dependence of diffuse reflection on wavelength (R = f(λ)) have shown that the compound TlInP2Se6 refers to indirect-gap semiconduc- tors. On the graph there are two rectilinear sections, one of which (for small wavelengths, , and large values of E) characterizes the interband transitions of electrons with phonon emission, and the other (for large  and small E) describes the processes of phonon absorption (Fig.9). The intersection of the first section with the wave- length axis, , yields the value of Eg + Ephonon ( = 560 nm, E = 2.21 eV), and the intersection of the second characterizes Eg – Ephonon ( = 605 nm and E = 2.05 eV). The length of the segment between the points of inter- section of both straight lines with the wavelength axis, , is equal to the doubled energy of the phonons, 2Ephonon (0.16 eV), interacting with the electron. The middle of this segment corresponds to the photon ener- gy equal to the width of the band gap of the indirect-gap semiconductor, Eg. Experimental calculations in terms of the compound TlInP2Se6 have shown that Eg = 2.13 eV and Ephonon = 0.08 eV. The compound In4(P2Se6)3 refers to direct-gap semiconductors, which characterizes the interband tran- sitions of electrons in terms of photon absorption (Fig.10). The intersection of the line with the wave- length axis,  ( = 651 nm), yields the value of Eg = 1.91 eV. The crystals of In4(P2Se6)3 and TlInP2Se6 were il- luminated by 10.6 μm and (its second harmonic) fre- quency doubling of 5.3 μm beams. Each channel of the beam was split by 200-ns CO2 laser pulses with a pulse repetition frequency of about 10 Hz. The angle between these two laser beams was changed from 18º to 22º. Figs.11 and 12 present these dependences. Treatment with a 10.6 μm beam achieved a smaller maximum bire- fringence (about 1.5510-2) in comparison to the 5.2 μm beam. This indicates a different photoinduced anisotro- py for the In4(P2Se6)3 and TlInP2Se6 crystals. Because a) b) . Figure 7. Coordination environments of the thallium (a) and indium (b) atoms in the structure of TlInP2Se6. Figure 8. Arrangement of the polyhedra anionic group [P2Se6] 4– in the structure of TlInP2Se6. Figure 9. Dependence of the diffuse reflection R on the wavelength  for the compound TlInP2Se6. Figure 10. Dependence of the diffuse reflection R on the wavelength  for the compound In4(P2Se6)3. INVESTIGATIONS OF THE TLINP2SE6–IN4(P2SE6)3 SYSTEM 45(2) pp. 13–18 (2017) 17 these crystals contain chalcogenide anions that contrib- ute to the anharmonicity of the phonon, they play a cru- cial role in terms of the second harmonic generation [13-14]. The maximum changes in the birefringence achieved were less than 210-2 and 6.310-2 for CO2 laser wavelengths of 10.6 μm and 5.3 μm, respectively. 4. Conclusion Differential thermal analysis, X-ray diffraction and mi- crostructural analysis were used to construct a phase diagram for the TlInP2Se6–In4(P2Se6)3 system, which can be characterized by a eutectic-type interaction. The invariant eutectic process L  htTlInP2Se6 + mtIn4(P2Se6)3 (15 mol% In4(P2Se6)3) occurs at 791 K. Two polymorphic transformations were identified for TlInP2Se6 at 680 K and 711 K and for In4(P2Se6)3 at 665 K and 903 K. New compounds were not detected in the binary system. The regions of solid phases of the complex compounds TlInP2Se6 and In4(P2Se6)3 do not exceed 10 mol%. Single crystals of both test compounds were achieved by the Bridgman method. Investigations concerning the dependence of the diffuse reflection spectrum showed that the compound TlInP2Se6 is char- acteristic of indirect-gap semiconductors (Eg = 2.13 eV, Ephonon = 0.08 eV), while the compound In4(P2Se6)3 is characteristic of direct-gap semiconductors (Eg = 1.91 eV, Ephonon = 0.08 eV). The dependence of the birefrin- gence was photoinduced by wavelengths of 5.3 μm and 10.6 μm, which is indicative of different photoinduced anisotropy. Acknowledgement We are grateful for the financial support of this work by the Ministry of Education and Science of Ukraine under the project DB874P_0117U000380. SYMBOLS ht high–temperature modification mt middle–temperature modification lt low–temperature modification SCE second coordination environment NCE nearest coordination environment Eg band gap, eV Ephonon phonon energy, eV R diffuse reflection  wavelength, nm REFERENCES Galdamez, A., Manriquez, V., Kasaneva, J., Avila, [1] R.E.: Synthesis, characterization and electrical properties of quaternary selenodiphosphates: AMP2Se6 with A – Cu, Ag and M – Bi, Sb, Mat. Res. Bull., 2003 38, 1063-1072 DOI: 10.1016/S0025- 5408(03)00068-0 McGuire, M.A.; Reynolds, T.K.; Di Salvo, F.J.: [2] Exploring thallium compounds as thermoelectric materials: seventeen new thallium chalcogenides, Chem. Mater., 2005 17, 2875-2884 DOI: 10.1021/cm050412c Gave, M.A.; Bilc, D.; Mahanti, S.D.; Breshears, [3] J.D.; Kanatzidis, M.G.: On the lamellar compounds CuBiP2Se6, AgBiP2Se6 and AgBiP2S6. antiferroe- lectric phase transitions due to cooperative Cu+ and Bi3+ ion motion, Inorg. Chem., 2005 44, 5293-5303 DOI: 10.1021/ic050357 Pfeiff, R.; Kniep, R.: Quaternary selenodiphos-[4] phates(IV): M(I)M(III)[P2Se6], (M(I) = Cu, Ag; M(III) = Cr, Al, Ga, In), J. Alloys Compd., 1992 186, 111-133 DOI: 10.1016/0925-8388(92)90626-K Figure 11. Photoinduced birefringence dependence at wavelengths of 5.3 μm and 10.6 μm - the two coherent beams were illuminated under different angles. During the two-beam coherent treatment at optimal power densities (about 400 MW/cm2) and incident angles (18-22 degrees) for crystals of In4(P2Se6)3. The bire- fringence scale should be multiplied by 10-2. Figure 12. Photoinduced birefringence dependence at wavelengths of 5.3 μm and 10.6 μm - the two coherent beams were illuminated under different angles. During the two-beam coherent treatment at optimal power densities (about 400 MW/cm2) and incident angles (18-22 degrees) for crystals of TlInP2Se6. The bire- fringence scale should be multiplied by 10-2. TOVT, BARCHIY, PIASECKI, KITYK, AND FEDORCHUK Hungarian Journal of Industry and Chemistry 18 Mucha, I.: Phase diagram for the quasi-binary thal-[5] lium(I) selenide–indium(III) selenide system, Thermochimica Acta, 2012 550, 1-4 DOI: 10.1016/j.tca.2012.09.028 Guseinov, G.D.; Abdullaev, G.B.; Godzhaev, [6] E.M.; Rzaeva, L.A.; Agaev, G.A.: Constitutional diagram and physical properties of thallium sele- nide–indium selenide pseudobinary system, Mater. Res. Bull., 1972 7(12), 1497-1503 DOI: 10.1016/0025- 5408(72)90187-0 Brockner, W.; Ohse, L.; Pätzmann, U.; Eisenmann, [7] B.; Schäfer, H.: Kristallstruktur und Schwingungsspektrum des Tetra-Thallium- Hexaselenidohypodiphosphates Tl4P2Se6, Z. Naturforsch. A, 1985 40, 1248-1252 Voroshilov, Y.V.; Gebesh, V.Y.; Potorii, M.V.: [8] Phase equilibria in the system In–P–Se and crystal structure of β-In4(P2Se6)3, Inorg. Mater., 1991 27, 2141-2144 Tovt, V.A.; Barchiy, I.E.; Piasecki, M.; Kityk, I.V.; [9] Fedorchuk, A.O.; Solomon, A.M.; Pogodin, A.I.: Triangulation of the Tl2Se–In2Se3–“P2Se4” qua- siternary system, Nauch. Vestn. Uzhgorod. Univ. (Ser. Khim.), 2016 35(2), 14–19 (in Russian) Akselrud, L.; Grin, Y.: WinCSD: Software package [10] for crystallographic calculations (Ver.4), J. Appl. Crystallogr., 2014 47, 803-805 DOI: 10.1107/S1600576714001058 Israel, R.; De Gelder, R.; Smits, J.M.M.; [11] Beurskens, P.T.; Eijt, S.W.H.; Rasing, T.H.; van Kempen, H.; Maior, M.M.; Motrya, S.F.: Crystal structures of di-tin-hexa(seleno)hypodiphosphate, Sn2P2Se6, in the ferroelectric and paraelectric phase, Z. Kristallogr., 1998 213, 34-41 DOI: 10.1524/zkri.1998.213.1.34 Fedorchuk, A.O.; Parasyuk, O.V.; Kityk, I.V.: Sec-[12] ond anion coordination for wurtzite and sphalerite chalcogenide derivatives as a tool for the descrip- tion of anion sub-lattice, Mat. Chem. Phys., 2013 139, 92-99 DOI: 10.1016/j.matchemphys.2012.12.058 Kityk, I.V.: IR-stimulated second harmonic genera-[13] tion in Sb2Te2Se–BaF2–PbCl2 glasses, J. Modern Optics, 2004 51, 1179-1189 DOI: 10.1080/09500340408230415 Kityk, I.V.: IR-induced second harmonic genera-[14] tion in Sb2Te3–BaF2–PbCl2 glasses, J. Phys. Chem. B, 2003 107, 10083-10087 DOI: 10.1021/jp030058a HUNGARIAN JOURNAL OF INDUSTRY AND CHEMISTRY Vol. 45(2) pp. 19–21 (2017) hjic.mk.uni-pannon.hu DOI: 10.1515/hjic-2017-0015 EXAMINATION OF INNOVATIVE HIGH-THROUGHPUT FERMENTATIONS ÁRON NÉMETH* Department of Applied Biotechnology and Food Science, Budapest University of Technology and Economics, Műegyetem rkp. 3, Budapest, 1111, HUNGARY During the investigation of fermentations, issues such as the need for numerous parallel experiments with re- gard to strain improvement or screening were often met, or in the case of media optimization the need for online measurements to avoid a lack of night-samples was also required. Therefore, several new instruments were in- troduced to solve one or more of these problems: impedimetric- and reverse-spin-technologies (RST) were compared via fermentation of a well-known species of yeast, Saccharomyces cerevisiae, under both aerobic and anaerobic conditions, resulting in a diauxic growth curve. To identify the most accurate method, a well- known mathematical description was fitted to the measured data. Since the initial parameters were considered reliable as they originated from real experiments, during model fitting, the parameters were further fine-tuned, and the less modifications reported the better the system since it produces a growth curve that is more similar to standard bioreactors. According to our study, the impedimetric equipment was more efficient, and could run 40 parallel experiments, but the RST was more flexible. Keywords: fermentation, high-throughput, scale-down, online measurement, mathematical model- ling 1. Introduction Developments in fermentations face numerous chal- lenges which may require expensive analytics, media components or special tools to facilitate aseptic work and sampling. Furthermore, these biological processes vary significantly. To overcome these difficulties, the process should be scaled-down in combination with high-throughput methods, resulting in many parallel, small-scale experiments. Such experiments are used in terms of strain and technological improvements as well as media optimization. A good solution may be the consideration of mi- cro-bioreactors. However, because of their high invest- ment and operational costs, they have not become wide- spread in Hungary. While each can provide almost eve- ry service required for bioreactors, for example, aera- tion, agitation and sampling in addition to pH and tem- perature control, they possess considerable limitations, namely non-standard conformations resulting in scale- up difficulties, or special measurement techniques that are incompatible with standard methods. A readily available alternative, to be more precise, Microtiter-Plates (MTP), for microscale high- throughput fermentations has already been presented and reported [1]. The basic principle is to use sterile ’96-Well’ microtiter plates with a special “sandwich cover” that facilitate sufficient aeration but reduce the likelihood of cross-infection. This system requires an *Correspondence: naron@f-labor.mkt.bme.hu adapter to be able to mount microtiter plates into a commercial rotary incubator shaker. The next issue is to analyse and follow the processes in the wells since their volumes are so small (ca. 100 L) that sampling is im- possible. Therefore, either a microplate reader is re- quired or a simple office scanner to produce a grey- scale photo taken from the bottom of the plate. The col- our of high cell-densities is close to white, but empty broths have a black background. In the case of species that produce high levels of acid, like Lactobacillus, even a pH indicator can be applied and besides a grey- scale photo a coloured one has to be taken as well; al- ternatively, CaCO3 should be added at the start but this can disturb the scanner-based “photometry”. Our partner (enzyscreen.com) even offers micro- titer plates for fed-batch fermentations. To achieve this, the feed components are adsorbed onto the material of the MTP, and are programmed to slowly release the fresh substrate during cultivation. However, another innovative solution has been developed for small-scale fermentations using online monitoring: Biosan Ltd. (Lithuania) applies Reverse-Spin Technology in the equipment of their Personal Bioreactor (RTS-1). This cost-effective equipment rotates a standard Falcon tube, filled with ca. 10 ml of fermentation media, at different rotation speeds in several directions at various con- trolled temperatures using a variety of aeration holes on the cap. This instrument also involves a photometer to facilitate the programming of measuring frequencies at a given wavelength (= 850 nm). For calibrated and reproducible measurements, a constant film layer is necessary, therefore, the instruments increase the rate of NÉMETH Hungarian Journal of Industry and Chemistry 20 rotation until 2000 rpm. The changes in parameters ef- fect shear forces as well as levels of aeration. Finally, this comparative study used an impedimet- ric system by SY-LAB (Austria) which is called Bac- Trac 4100 [2]. This equipment possesses a block ther- mostat composed of 40 measurement cells, each con- taining 4 electrodes. One pair of them follows the changes in the impedance of the media, M%, caused by the secreted acids and metabolites. In the case of micro- organisms that exhibit high levels of ionic strength in the media, it is hard to detect M%, therefore, with the application of a different frequency the changes in im- pedance on the other electrode surface (E%) can be fol- lowed. In direct measurements, these electrodes are immersed directly into the culture, but in the case of indirect measurements, they are rinsed with KOH which can adsorb the formed CO2 released by the culture. While this system does not possess mixer/aerator solu- tions, this result can be transferred carefully to the known systems, namely benchtop fermenters or shaking flasks. However, it is able to follow forty different cul- tures. In this study, a well-known model organism (S. cerevisiae) was chosen that exhibits special biochemical behaviour. It was used to test the compare the ability of the three systems introduced above. What is special about S. cerevisiae is that it can change from aerobic to anaerobic cultivation according to Pasteur and Crabtree effects; i.e. under lack of oxygen or excess to sugar, respectively. After changing to anaerobic metabolism, it produces mostly alcohol but later this can be consumed by yeast as well resulting in a stepwise growth curve, also referred to as a diauxic growth profile. Thus, the question was whether such a system could show and follow this diauxic growth. 2. Experimental Commercial S. cerevisiae, i.e. baking yeast produced by Lesaffre, was cultured on a media of molasses that were diluted by a factor of 10 resulting in a saccharose con- centration of ca. 75 g dm-3 and a 20:1 volume of molas- ses to NH4OH ratio at 34°C. The 100 L of inoculum possessed a cell-dry-weight (CDW) content of 10 g/dm3. RTS-1 collected the data in a Microsoft Excel database. BacTrac only provided the data collected on screen plots, but with the help of Digitizelt v.2.3 soft- ware the measurement data was transported into Mi- crosoft Excel. To compare the data in Microsoft Excel, the structured model of Blanch et al. [3] was adopted and programmed in Berkeley Madonna for Windows 8.1. This model can describe both anaerobic cell growth on excess sugar with the formation of alcohol and aero- bic cell growth on alcohol as a substrate. It divides cells into two main compartments, i.e. substructures: one is responsible for metabolism (both aerobic and anaero- bic), and the other is responsible for cell division. The parameters, for example reciprocal yields and stoichio- metric coefficients of the model, were partly determined experimentally, but others were determined by non- linear model fitting, i.e. model calibration on real sam- ples. 3. Results and Analysis 3.1. Reverse-Spin Technology vs. Personal Bioreactor (RTS-1) Fig.1 presents the results of RTS-1. While optical densi- ty (OD), i.e. turbidity at = 850nm, changed slowly, the specific growth rate calculated online only reflected the uncertainty of the OD measurements, but the tempera- ture remained constant as expected. Additionally diaux- ic growth was also detected but over a very long period of time. The model fitting was quite difficult because a satisfactory fit was only achieved after remarkable changes to basic constants, for example maximum spe- cific growth rates on both substrates, etc., had been ap- plied. 3.2. Impedimetric System: BacTrac In the case of the impedimetric experiments, three dif- ferent arrangements were tested: an anaerobic cell with (A) (B) Figure 1. The measured parameters (temperature, , OD850) and calculated data (CDW from OD850) along with the data of the predicted (i.e. fitted) model. (A) green line: temperature; blue line: measured turbidity at 850nm; brown line: measured specific growth rate. (B) green line: temperature; red crosses: calculated cell dry weight from the measured turbidity at 850nm; purple line: fitted model-based prediction for CDW. 0 5 10 15 20 25 30 35 40 -2 -1 0 1 2 3 4 0 50 100 150 200 250 300 350 T(°C)OD (850nm) Fermentation Time (h) OD(850nm) µ (h ̄ ¹) T °C Growth on sugar Growth on ethanol D I A U X I C G R O W T H 0 5 10 15 20 25 30 35 40 0 2 4 6 8 10 12 14 16 18 0 50 100 150 200 250 300 350 T(°C)CDW Fermentation Time (h) CDW Predicted CDW T °C EXAMINATION OF INNOVATIVE HIGH-THROUGHPUT FERMENTATIONS 45(2) pp. 19–21 (2017) 21 an incorporated valve for gas release, an aerobic one, and an indirect one (Fig.2). Only M% values yielded explainable curves. Indirect measurements yielded an inverse growth curve (decreasing) as expected, but did not exhibit a two-step decrease, i.e. diauxic growth, therefore, M% values of direct measurements were evaluated. The two curves of aerobic and anaerobic M% values were very similar to each other, but perhaps the anaerobic example is more relevant as in the case of high sugar content, the metabolism of yeast shifted in the anaerobic direction. Fig.3 shows the fits of the mod- el in which less constants had to be changed and diauxic growth was detected. 4. Conclusion Both tested systems – Personal Bioreactor (RTS-1, Bio- san) and BacTrac (SY-LAB) – detected diauxic cell growth of baking yeast. RTS-1 seemed to be a little bit more flexible, but BacTrac gave faster results, was able to make 40 measurements at the same time and offered three options in terms of evaluation. Maybe in the near future a solution to regular automatic sampling from larger-scale fermenters will be found and then the re- sults can be compared with the ones presented here. Acknowledgement We are sincerely grateful to SY-LAB for the support provided with regard to BacTrac, and to Biocenter Kft. for importing RTS-1. REFERENCES [1] Németh, Á.; Kiss, Á.; Sevella, B.: Experiments for D-lactic acid production with fermentation, Hung. J. Ind. Chem., 2011 39(3), 359–362 [2] Bankovsky, V.; Bankovsky, I.; Bankovsky, P.; Isakova, J.; Djackova, I.; Sharipo, A.; Eskin, J.; Dišlers, A.; Rozenstein, R.; Saricev, V.; Djacenko, S.; Makarenko, V.; Balodis, U.: Reverse–Spin® Technology - Innovative Principle of Microbial Cultivation, Manufacturer's online leaflet: https://biosan.lv/images/uploads/content/files/reverse_spinner.pdf [3] Blanch, H.W.; Clark, D.S.: Biochemical Engineering (Marcel Dekker, NY, USA), 1996, pp. 231–236 ISBN 9780824700997 (A) (B) (C) Figure 2. The results of three BacTrac measurements: relative changes in impedancy in the M% of media vs. time (h) - (A) indirect-; (B) aerobic-direct-; (C) anaerobic-direct measurements. Figure 3. Model fitting to the anaerobic BacTrac curve (M%): red crosses: measured data; purple line: model- predicted values; green line: temperature (°C). 0 5 10 15 20 25 30 35 40 0 2 4 6 8 10 12 0 20 40 60 80 100 120 T(°C) Measured (M%) and predicted impedancy Fermentation Time (h) M(%) Predicted Impedancy T °C D I A U X I C G R O W T H Growth on ethanol Growth on sugar HUNGARIAN JOURNAL OF INDUSTRY AND CHEMISTRY Vol. 45(2) pp. 23–27 (2017) hjic.mk.uni-pannon.hu DOI: 10.1515/hjic-2017-0016 THE EFFECT OF ADVANCED OXIDATION PRE-TREATMENT ON THE MEMBRANE FILTRATION PARAMETERS OF DAIRY WASTEWATER MIHÁLY ZAKAR, 1,2 ILDIKÓ KOVÁCS, 1 PÉTER MUHI, 1 ERIKA HANCZNÉ LAKATOS, 2 GÁBOR KESZTHELYI-SZABÓ, 1 ZSUZSANNA LÁSZLÓ 1, * 1 Institute of Mechanical and Process Engineering, Faculty of Engineering, University of Szeged, Moszkvai krt. 9, Szeged, 6724, HUNGARY 2 Institute of Food Sciences, Széchenyi István University, Lucsony u. 15-17, Mosonmagyaróvár, 9200, HUNGARY The dairy industry generates wastewater characterised by high levels of biological and chemical oxygen de- mands representative of their high degree of organic content; mainly carbohydrates, proteins and fats that origi- nate from milk. Several investigations have been conducted into the reuse of dairy wastewater, e.g. membrane processes are a promising method to treat such wastewater. Earlier works have proven that with membrane fil- tration an appropriate degree of retention can be achieved and the permeate can be reused. However, mem- brane fouling is a limiting factor in these processes. Advanced oxidation processes (AOPs) are widely used in the fields of water and wastewater treatments and are known for their capability to mineralise a wide range of organic compounds. AOPs also exhibit some other effects on the filtration process, e.g. the microflocculation ef- fect of ozone may play a significant role in increasing the elimination efficiency and causing a decreased level of irreversible fouling. By comparing ozone and Fenton pre-treatment (FPT) processes it can be shown that the fouling propensity of pre-treated pollutants does not depend on the pre-treatment method, while FPT was prov- en to be more efficient in improving the level of flux. Keywords: ultrafiltration, ozone pre-treatment, Fenton-reaction, fouling resistances, dairy wastewater 1. Introduction The dairy industry is considered to be the largest source of food-processing wastewater in many countries. Dairy wastewater exhibits high degrees of biological oxygen demand (BOD) and chemical oxygen demand (COD); contains high levels of dissolved or suspended solids including fats, oils and grease; as well as nutrients such as ammonium ions or phosphates. Therefore, proper attention must be paid to them before disposal [1]. There are several research projects that aim to identify possibilities of reusing or recycling dairy wastewater [2- 9]. Membrane treatment of dairy wastewater with the aim of water reuse could simultaneously lower the total water consumption and effluent production of dairy plants, as the purified water produced could be reused in a dairy plant to heat or cool water. Besides additional advantages, e.g. a high degree of separation efficiency in the absence of chemical changes and low levels of energy intensity, membrane filtration also has draw- backs, namely compounds in dairy wastewater that con- tain protein were found to be significant foulants in terms of existing membrane materials [10-12]. *Correspondence: zsizsu@mk.u-szeged.hu The combination of membrane separation and pre- treatment with advanced oxidation processes (AOPs: using ozone, hydrogen peroxide, UV light, or a combi- nation of these) opens up new opportunities, since the ozone and the resulting oxidizing radicals (mainly hy- droxyl radicals) efficiently change the characteristics of the colloidal particles or oxidizing compounds, which cause membrane fouling [13]. Earlier studies have shown that the microflocculation effect of ozone may play a significant role in increasing the elimination effi- ciency and may decrease the extent of membrane foul- ing and increase the degree of gel formation. In addi- tion, AOPs can be used as a pre-treatment stage before a biological step in order to increase the biodegradability of the recalcitrant compounds and thus lower the toxici- ty of the wastewaters [13-14]. According to economic evaluation studies, the Fenton process is more economical than ozone pre- treatment [15]. However, there is little data concerning its effect on membrane filtration parameters. The aim of the present work was to investigate and compare the effect of ozone pre-treatment and the Fenton’s reaction on ultrafiltration parameters, fouling mechanisms and the pollutant removal efficiency on a model dairy wastewater. ZAKAR, KOVÁCS, MUHI, LAKATOS, KESZTHELYI-SZABÓ, AND LÁSZLÓ Hungarian Journal of Industry and Chemistry 24 2. Experimental 2.1. Samples and Measurements Model solutions were prepared from milk powder (Milk Quick, Instantpack Kft., Hungary) composed of 0.3% (g/g) concentrations, 32% (g/g) proteins, 5% (g/g) fat and 50% (g/g) lactose. Ozone was produced from oxy- gen (Linde, 3.0) with a flow-type ozone generator (Ozomatic Modular 4, Wedeco Ltd., Germany). The ozone-containing gas was bubbled continuously through a batch reactor during the treatment. The volume of the treated water was 0.45 dm3. The durations of the treat- ment were 5, 10 and 20 mins; and the flow rate was 1 dm3 min–1. The ozone concentrations of the bubbling gas before and after it was passed through the batch reactor were measured with a ultraviolet–visible (UV- VIS) spectrophotometer (Nanocolor NUV 0113) at a wavelength of 254 nm (Fig.1). The absorbed ozone concentrations were 6.8·10–4 M, 1.43·10–3 M and 2.67·10–3 M, respectively. Fenton’s reaction was conducted in a batch stirred ultrafiltration cell with 1.5 mmol dm-3 FeSO4×7H2O (purity 99%, Spektrum-3D, EU) adjusted to pH 3 with H2SO4 (purity 96%, Farmitalia Carlo Erba SPA, Italy), 0.3 wt.% milk powder solution and H2O2 solution (30%, purity 99%, Spektrum-3D Kft.), the [H2O2]:[Fe] ratio was 5:1 (Fenton (5:1)) or 50:1 (Fenton (50:1)). The ozone or Fenton pre-treated samples were used as a feed in ultrafiltration (UF) experiments. The UF experiments were carried out in a batch stirred ultrafiltration cell (Millipore, Serial N°94, USA) with a capacity of 50 cm3, and the filtrations were per- formed at transmembrane pressures of 0.1 (only in the case of Fenton (50:1)) or 0.3 MPa and the feed solutions were stirred at 350 rpm. For filtration experiments, flat sheet polyethersulfone (PES) membranes (PES-10 se- ries, New Logic Research Inc., USA) and a molecular weight cut-off (MWCO) of 10 kDa were used with an effective membrane surface area of 1.73 dm2. The initial feed volume was 50 cm3, the ultrafiltration experiments were conducted until 40 cm3 of the total sample had been filtered, when the volume reduction ratio (VRR) was equal to 5. Determination of the COD was based on the standard method involving the oxidation of potassium dichromate; for the analysis, standard test tubes (Lov- ibond Tintometer Ltd.) were used. The digestions were conducted in a COD digester (Lovibond ET108 ther- moreactor); and the COD values were measured with a COD photometer (Lovibond PCCheckit). For the de- termination of the residual amount of hydrogen perox- ide, COD measurements were performed before and after the addition of the enzyme catalase. 2.2. Theoretical Methodologies In order to investigate mechanisms of membrane foul- ing, filtration resistances were calculated according to the Resistance-In-Series Model, Eqs.(1-4). The membrane resistance ( MR , m-1) was calculat- ed as 1 M w w [m ] p R J    (1) where p is the difference in pressure either side of the membrane (in MPa), wJ is the water flux of the clean membrane, and w is the viscosity of water (in Pa·s). The total resistance ( TR , in m–1), can be evaluated from the steady-state flux by using the Resistance-In- Series Model: revT M irrevR R R R   (2) where irrevR is the irreversible resistance (mainly caused by the fouled pores) and revR is the reversible re- sistance. The irreversible resistance was determined by measuring the water flux through the membrane after filtration, rinsing it with deionized water to remove any particles of the residue layer from the surface, and sub- tracting the resistance of the clean membrane: Mirrev WW A p R R J     (3) where WAJ is the water flux after concentration tests. The reversible resistance of the layer deposited on the membrane surface was calculated as: r e v Mir r e v W WC p R R R J      (4) where CJ is the constant flux at the end of the concen- tration test and WW is the viscosity of the wastewater viscosity [16]. Mathematical modelling of the fouling mechanism was studied based on the Hermia’s model [17]. The Hermia’s model describes the mechanism of membrane fouling based on blocking filtration laws, consisting of complete pore blocking, standard pore blocking and intermediate pore blocking, in addition to cake filtration (Table 1) to illustrate the different fouling mechanisms. Figure 1. Experimental set-up of ozonation. EFFECT OF ADVANCED OXIDATION PRE-TREATMENT 45(2) pp. 23–27 (2017) 25 The Hermia’s model was then linearized for each model using a fitting equation in terms of the permeate flux versus time as presented in Table 1. In terms of the evaluation of the results these models were fitted to ex- perimental data. In Table 1, J is the flux, J0 is the initial flux, the various K’s are the fouling coefficients, and A is a constant. To compare the performance of different AOPs, the oxygen-equivalent chemical-oxidation capacity (OCC, kg O2 m-3) was used to quantify the oxidants used in the ozone treatment and Fenton’s reaction, and was determined based on stoichiometric calculations [14]: OCC =1.000[O3] = 0.471[H2O2] (5) where [O3] is the required ozone concentration (kg O3 m-3), and [H2O2] is the required hydrogen peroxide con- centration (kg H2O2 m -3). 3. Results and Analysis 3.1. Experiments The effect of pre-oxidation on filtration parameters was investigated by fitting equations in Table 1 to measured data. Based on the value of the coefficient of determina- tion, the cake layer filtration yielded the best correla- tion. In order to compare the different pre-oxidation methods, normalised values of the initial flux (J0, L m-2 h-1 bar-1) and fouling coefficients (k) were calculated and compared (Figs.2 and 3). It was found that the ef- fect of ozone treatment and Fenton-treatment is differ- ent in the case of initial normalised flux. Not only the Fenton pre-treatment but the addition of reagents in the absence of hydrogen peroxide exhibited coagulation- flocculation effects that resulted in an enhanced initial flux. In the case of the Fenton’s reaction this effect is independent of the [H2O2]:[Fe] ratio. The fouling coefficient also changes by the addi- tion of oxidants, (Fig.3) but in this case, the tendency is more likely to depend on the OCC than on the applied AOP method. At lower oxidation capacities the fouling coefficient decreases resulting in lower degrees of foul- ing than in non-treated solutions, however, at higher oxidation grades, the fouling coefficient increases. To obtain more information concerning the fouling mechanisms, the filtration resistances of ozone-treated and Fenton (5:1) pre-treated solutions were calculated and compared (Fig.4). It was found that - in accordance with the values of the fouling coefficient - filtration re- sistances decrease as the duration of oxidation pre- treatment increases. In particular, mainly pre-treatments of short durations decreased the irreversible fouling resistance and increased the reversible fouling re- sistance. 4. Discussion As an effect of the pre-oxidation of model dairy wastewater two typical pathways were observed that influence membrane filtration parameters: the i) micro- Figure 2. Normalised initial flux values as a function of OCC. 0 5 10 15 20 25 0,0000 0,0050 0,0100 0,0150 J 0 (L / m 2 h b a r) OCC (kg O2/m3) Fenton (50:1) Fenton (5:1) ozone untreated Figure 3. Fouling coefficient as a function of OCC. 0 0,001 0,002 0,003 0,004 0,005 0,006 0,00 0,10 0,20 0,30 0,40 0,50 k OCC (kg O2/m3) Fenton (50:1) Fenton (5:1) ozone untreated Figure 4. Filtration resistances of untreated, Fenton (5:1) and ozone pre-treated solutions. 0,00E+00 1,00E+13 2,00E+13 3,00E+13 4,00E+13 5,00E+13 6,00E+13 7,00E+13 F il tr a ti o n r e si st a n ce s (1 / m ) RM R(irrev) R(rev) RT Table 1. Hermia’s filtration laws. Fouling mecha- nism Filtration law Constant-pressure filtration J0 A = cont. Complete pore blocking J = J0 e -kt ln J = ln J0 - kt Gradual pore blocking (stand- ard pore block- ing) J = J0·(1 + ½ Ks (A·J0 ) ½·t)-2 1/J0.5 =1/J0 0.5+ks·t ks = 0.5 Ks A 0.5 Intermediate filtration J = J0 · (1 + Ki·A·J0 ·t) -1 1/J = 1/J0 + ki·t ki = Ki A Cake filtration J = J0 · (1 + 2Kc(A·J0) 2·t)-0.5 1/J2 = 1/J0 2 + kc·t kc = 2Kc A 2 ZAKAR, KOVÁCS, MUHI, LAKATOS, KESZTHELYI-SZABÓ, AND LÁSZLÓ Hungarian Journal of Industry and Chemistry 26 flocculation effect produces associated colloidal parti- cles, and ii) degradation of organic matter (Fig.5). The former resulted in decreased fouling of membrane pores as shown by the decreased fouling coefficient and irre- versible resistance. This can be observed only during short-term ozone or Fenton treatments (OCC < 0.05 kg O2 m -3). The latter point may increase the degree of pore fouling [13, 18] due to the formation of small degrada- tion by-products, which can enter membrane pores as the increased values of irreversible resistance also prove. By comparing the ozone and Fenton pre- treatment processes with similar OCCs, it can be con- cluded that the Fenton pre-treatment may be more effec- tive in terms of enhancing the flux, probably due to the coagulation-flocculation effect of the ferrous salts them- selves. 5. Conclusion The comparison of ozone and Fenton processes as pre- treatments before ultrafiltration of a model sample of dairy wastewater showed that such pre-treatments may improve the filtration parameters in terms of flux or fouling mitigation. By examining the effect of the oxi- dation capacities of ozone and Fenton pre-treatment processes, it was found that the fouling propensity of pollutants does not depend on the pre-treatment method. However, it depends on the OCC of the pre-treatment method. Although the method of pre-treatment affects the flux, the Fenton pre-treatment proved to be more efficient in terms of enhancing the value of the flux. SYMBOLS RM membrane resistance (m–1) Rrev reversible resistance (m–1) Rirrev irreversible resistance (m–1) p pressure difference between the two sides of the membrane (MPa) J flux (1/s) JW water flux (1/s) JC constant flux at the end of the concentration (1/s) J0 initial flux (1/s) ηW water viscosity (Pa·s) k fouling coefficient OCC oxygen-equivalent chemical-oxidation capacity (kg O2·m –3) Acknowledgement This research was supported by the János Bolyai Re- search Fellowship of the Hungarian Academy of Sci- ences. The authors are also grateful for the financial support of the National Research, Development and Innovation Office (NKFIH K112096). REFERENCES [1] Farizoglu, B.; Uzuner, S.: The investigation of dairy industry wastewater treatment in a biological high performance membrane system, Biochem. Eng. J., 2011 57, 46–54 DOI: 10.1016/j.bej.2011.08.007 [2] Vourch, M.; Balance, B.; Chaufer, B.; Dorange, G.: Treatment of dairy industry wastewater by reverse osmosis for water reuse, Desalination, 2008 219(1-3), 190–202 DOI 10.1016/j.desal.2007.05.013 [3] Perle, M.; Kimchie, S.; Shelef, G.: Some biochemical aspects of the anaerobic degradation of dairy wastewater, Water Res., 1995 29(6), 1549–1554 DOI 10.1016/0043-1354(94)00248-6 [4] Bick, A.; Plazas, T.J.G.; Yang, F.; Raveh, A.; Hagin, J.; Oron, G.: Immersed Membrane Bio Reactor (IMBR) for treatment of combined domestic and dairy wastewater in an isolated farm: An exploratory case study implementing the Facet Analysis (FA), Desalination, 2009 249(3), 1217– 1222 DOI 10.1016/j.desal.2009.06.035 [5] Sarkar, B.; Chakrabarti, P.P.; Vijaykumar, A.; Kale, V.: Wastewater treatment in dairy industries: possibility of reuse, Desalination, 2006 195(1-3), 141–152 DOI 10.1016/j.desal.2005.11.015 [6] Andrade, L.H.; Motta, G.E.; Amaral, M.C.S.: Treatment of dairy wastewater with a membrane bioreactor, Braz. J. Chem. Engng., 2013 30(4), 759–770 DOI 10.1590/S0104-66322013000400008 [7] Yip, V.; Arntfield, S.D.; Hydamaka, A.W.: Potential for stainless steel microfiltration processing to reduce effluent from a fluid milk and ice-cream processing plant, J. Dairy Sci., 1996 79(4), 710–716 DOI 10.3168/jds.S0022-0302(96)76417-2 [8] Koyuncu, I.; Turan, M.; Topacik, D.; Ates, A.: Application of low pressure nanofiltration membranes, for the recovery and reuse of dairy industry effluents, Water Sci. Tech., 2000 41(1) 213–221 web: http://wst.iwaponline.com/content/41/1/213 [9] Luo, J.; Ding, L.; Qi, B.; Jaffrin, M.Y.; Wan, Y.: A two-stage ultrafiltration and nanofiltration process for recycling dairy wastewater, Bioresour. Technol., 2011 102(16), 7437–7442 DOI 10.1016/j.biortech.2011.05.012 Figure 5. Possible effect of ozone pre-treatment on the membrane filtration of dairy wastewater. EFFECT OF ADVANCED OXIDATION PRE-TREATMENT 45(2) pp. 23–27 (2017) 27 [10] Bégoin, L.; Rabiller-Baudry, M.; Chaufer, B.; Faille, C.; Blanpain-Avet, P.; Bénézech, T.; Doneva, T.: Methodology of analysis of a spiral- wound module. Application to PES membrane for ultrafiltration of skimmed milk, Desalination, 2006 192(1-3), 40–53 DOI 10.1016/j.desal.2005.10.010 [11] Koutake, M.; Uchida, Y.; Sato, T.; Shimoda, K.; Watanabe, A.; Nakao, S.: Filtration membrane fouling in ultrafiltration of skim milk, 1: Causes and cleaning, J. Agric. Chem. Soc. Jpn., 1987 198, 677–681 web: http://agris.fao.org/agris-search/search.do? recordID=JP880163288 [12] Rabiller-Baudry, M.; Le Maux, M.; Chaufer, B.; Begoin, L.: Characterisation of cleaned and fouled membrane by ATR-FTIR and EDX analysis coupled with SEM: Application to UF of skimmed milk with a PES membrane, Desalination, 2002 146(1-3), 123–128 DOI 10.1016/S0011-9164(02)00503-9 [13] László, Zs.; Kertész, Sz.; Beszédes, S.; Hovorka- Horváth, Zs.; Szabó, G.; Hodúr, C.: Effect of preozonation on the filterability of model dairy waste water in nanofiltration, Desalination, 2009 240(1-3), 170–177 DOI 10.1016/j.desal.2007.12.040 [14] Cańizares, P.; Paz, R.; Sáez, C.; Rodrigo, M.A.: Costs of the electrochemical oxidation of wastewaters: a comparison with ozonation and Fenton oxidation processes, J. Environ. Mgmt., 2009 90(1), 410–420 DOI 10.1016/j.jenvman.2007.10.010 [15] Özge, S.; Taner, Y.: Determination of the acute toxicities of physicochemical pretreatment and advanced oxidation processes applied to dairy effluents on activated sludge, American Dairy Science Association, J. Dairy Sci., 2015 98(4), 2337–2344 DOI 10.3168/jds.2014-8278 [16] Kertész, Sz.; László, Zs.; Forgács, E.; Szabó, G.; Hodúr, C.: Dairy waste-water purification by vibratory shear enhanced processing, Desalination Water Treat., 2012 37(1-3), 1–7 DOI 10.5004/dwt.2011.2485 [17] Hermia, J.: Constant pressure blocking filtration law: Application to power law non-newtonian fluids, Trans. Ind. Chem. Eng., 1982 60(3), 183– 187 web: http://hdl.handle.net/2078.1/57489 [18] Zhu, H.T.; Wen, X.H.; Huang, X.: Pre-ozonation for dead-end microfiltration of the secondary effluent: suspended particles and membrane fouling, Desalination, 2008 231(1-3), 166–174 DOI 10.1016/j.desal.2007.11.044 HUNGARIAN JOURNAL OF INDUSTRY AND CHEMISTRY Vol. 45(2) pp. 29–33 (2017) hjic.mk.uni-pannon.hu DOI: 10.1515/hjic-2017-0017 MICROENCAPSULATION OF VEGETABLE OIL: ALTERNATIVE APPROACHES USING MEMBRANE TECHNOLOGY AND SPRAY DRYING KRISZTINA ALBERT,* GYULA VATAI, AND ANDRÁS KORIS Department of Food Engineering, Szent István University, Ménesi út 44, Budapest, 1118, HUNGARY Microencapsulation technology is a method that is widely used in the food industry. By comparing the latest en- capsulation techniques, a significant number of publications concern membrane technology. The term “mem- brane-based encapsulation” entails that the first step of the technique is the preparation of emulsion with the help of microporous membranes. Generally, in microencapsulation technologies, the wall material is dissolved in a continuous phase and oil is dispersed within it. In the present investigation, a new method of preparing micro- capsules composed of vegetable oil and maltodextrin was developed. In the first step, the wall material (malto- dextrin) was dissolved in oil and considered as a dispersed phase, subsequently, it was introduced into a con- tinuous phase (water) through a microporous membrane. A comparative study was conducted between conven- tional microencapsulation techniques and one developed in our laboratory. The average particle size of micro- capsules prepared by our method is smaller than the size allowed by other methods. After encapsulation prepa- ration, fine-tuned microcapsules were produced by spray drying. However, the main disadvantage of our pro- posed technology is rapid membrane fouling, because of high concentrations of solute in the dispersed phase. This problem can be eliminated by judicious and systematic investigations. Keywords: vegetable oil, microencapsulation, membrane technology, spray drying 1. Introduction The controlled release of food ingredients at the right place and the right time is a key functionality that can be provided by microencapsulation. In food products, fats and oils, aroma compounds, vitamins, minerals, colourants and enzymes are encapsulated. The process of the encapsulation of sensitive compounds consists of two steps: the first is often emulsion production: emulsification of core materials with dense solutions of wall material; the second is drying or cooling of emulsion by some chemical or mechanical process; and at the end of these processes microparticles can be obtained [1]. The two major industrial encapsulation processes are spray drying and extrusion [2]. Table 1 summarizes literature reviews that focus on these technologies, where the two steps of microencapsulation are performed by the combination of so-called membrane emulsification and spray drying. Membrane emulsification (ME) is a relatively new, simple emulsion-production method which can be conducted with the use of a microporous membrane. Other emulsification processes within this category, besides ME and spray drying, include simple blending as well as homogenization; and secondary reactions that recover the capsules from the emulsion are interfacial *Correspondence: albert.krisztina@etk.szie.hu polymerization, vacuum heat treatment, solvent diffusion and freeze drying. Different materials are used as the dispersed phase and wall materials of the microcapsules. The continuous phase is typically water. 2. Experimental The main objective of this study was to initially prepare emulsions and form microcapsules using three different methods. The first approach was conventional membrane emulsification. The second one was conducted by a modified ME process. The third involved the use of a laboratory blender. In all three cases, the second step was the spray-drying technique to recover the microcapsules from the emulsions. 2.1. Samples and Measurements All chemicals (maltodextrin, Tween-80 emulsifier) were procured from Shop.Builder, Hungary and Sigma- Aldrich, Germany. Commercial-grade sunflower-seed oil was purchased from a local market in the vicinity of Budapest. 2.2. Membrane Emulsification Apparatus A cross-flow membrane emulsification system was used as the emulsification process (Fig.1). The apparatus included two manometers positioned at the opposite ALBERT, VATAI, AND KORIS Hungarian Journal of Industry and Chemistry 30 ends of the membrane to measure the drop in pressure along the membrane. The pressure of the dispersed phase was guaranteed by compressed air from an air pump which was injected from the outer surface of the membrane. The continuous phase was recirculated on the lumen side of the membrane by a pump. A rotameter, placed at the exit of the membrane, allowed the flow rate of the continuous phase to be measured. 2.3. Microencapsulation Process Microcapsules were prepared by a tubular ceramic membrane with a pore size of l.4 μm (PALL Austria Filter GmbH). The membrane was composed of α- alumina and the surface area of the active membrane was 50 cm2. Cross-flow operation was adopted for emulsion production. The first emulsion was prepared using a conventional membrane emulsification technique. The wall material (maltodextrin) was dissolved in the continuous phase (water) and sunflower-seed oil was dispersed into it. The direction of flow of the dispersed phase was tangential with the surface of the membrane. Otherwise, in the second case the wall material was mixed with sunflower-seed oil. This mixture was considered as the dispersed phase and it was pressed through the membrane pores under pressure. In both cases the pressure of the dispersed phase was 2.5 bars and the recirculation flow rate of the continuous phase was 150 dm3 h-1. In the third case the emulsion was prepared at room temperature using a laboratory blender at 2,000 rpm for 40 minutes. 2.4. Spray Drying The emulsions prepared were spray dried with a laboratory-scale spray dryer (LabPlant SD-05) equipped with a nozzle of 0.5 mm in diameter. The pressure of the compressed air in terms of the flow of the spray was adjusted to 3.6 bars. The air temperature at the inlet was maintained at 180±5 °C, and the feed rate was adjusted to 475 cm3 h-1, respectively. Emulsions were prepared during the spray drying process and were continuously stirred by a magnetic stirrer throughout. The microcapsules were collected from the collecting chamber and stored in darkness until analysed. 2.5. Analysis of Microcapsules Following the emulsion preparation, the average droplet size and span value were measured by a FRITSCH Laser Particle Sizer ANALYSETTE 22 NanoTec. Each sample was analyzed three times and the average data reported. The span value was considered as an indication of the dispersity of the droplet size. The lower the span value, the more monodisperse the emulsion. A VHX-6000 digital microscope manufactured by KEYENCE was used to check the formation of microcapsules and evaluate their morphology. 2.6. Surface-Oil Content and Microencapsulation Efficiency The procedure modified by Calvo et al. [9] to determine the surface-oil content and microencapsulation efficiency was employed to measure the amount of unencapsulated oil present on the surface of the powders. Briefly, 5 g of microcapsules were precisely Table 1. Summary of examples from the literature of microencapsulation technologies. Emulsification Secondary reaction Dispersed phase / Wall material membrane emulsification spray-drying oil / polyvinylpyrrolidone (PVP) [3] fish oil / whey protein isolate (WPI), whey protein hydrolysate (WPH), sodium caseinate, maltodextrin [4] interfacial polymerization benzene, xylene, liquid paraffin / terephthaloyl dichloride (TDC) [5] vacuum heat treatment oil / polyethersulphone (PES) [6] solvent diffusion oil / polycaprolactone (PCL), dichloromethane (DCM) [7] freeze-drying chloroform + curcumin / poly(lactic-co-glycolic acid) (PLGA) [8] blending and homogenization spray-drying extra virgin olive oil / gelatin, gum arabic, starch, lactose, maltodextrin [9] olive oil + α-Tocopherol / maltodextrin, agave inulin, acacia gum [10] walnut and chia oil / maltodextrin, (hydroxypropyl)methyl cellulose [11] ginger oil / cashew gum, inulin [12] walnut oil / skimmed milk powder (SMP), SMP+Tween 80, SMP+maltodextrin [13] Nigella sativa oil / sodium caseinate, maltodextrin [14] chili seed oil / sodium octenylsuccinate starch, maltodextrin [15] freeze-drying Ziziphora clinopodiodes essential oil / whey protein isolate (WPI), pectin [16] Figure 1. Experimental set-up for the cross-flow membrane emulsification process: (1) continuous phase tank, (2) graduated disperse phase tank, (3) membrane module, (4) pump, (5) compressor, (6) pressure meter, (7) rotameter, (8) pressure controller, (9) valve on drain, (10) valve, (11) heating / cooling (thermostat). MICROENCAPSULATION OF VEGETABLE OIL 45(2) pp. 29–33 (2017) 31 weighed in a beaker and 50 cm3 of hexane was added and shaken by hand for 15 s at ambient temperature to extract the superficial oil. The solvent mixture was then filtered through filter paper, and subsequently the unencapsulated oil was collected after evaporation of the hexane under a vacuum. To measure the total amount of encapsulated oil, the same procedure was conducted, but the powder of microcapsules in hexane as an extraction solvent was stirred for 4 h using a magnetic stirrer. For the production of microcapsules in oil, the encapsulation efficiency (EE) is an important parameter which is strongly related to the amount of oil on the surface, and was calculated using the following equation: �� = ����� ��� – ������� ��� ����� ��� × 100 (1) 3. Results and Analysis The first part of this section presents the results in terms of the production of sunflower-seed oil/water emulsions by conventional and modified-membrane emulsification techniques using a laboratory blender, then the encapsulation efficiency and the physical characterization of the obtained microcapsules are presented. 3.1. Results of Emulsification The results of particle-size measurements are summarized in Table 2. The average droplet sizes were 9.18 µm, 7.58 µm and 7.12 µm in the three different emulsification cases. The average droplet size of microcapsules prepared in the modified way, when the wall material was mixed with the dispersed phase, was lower than those prepared using the conventional technique. In 2000, Joscelyne, along with his co-worker, reported that for membrane emulsification, the size of synthesized emulsion particles might be 2-10 times greater than the pore size of the membrane. The present investigators used a microporous membrane with a pore size of 1.4 μm and it was found that in all three cases the average size of the synthesized microcapsules is in line with the conclusion reached by a previously mentioned research group [17]. As an example, the shape and morphology of synthesized microcapsules prepared by the modified ME methods are shown in Fig.2. 3.2. Spray-Drying Results Sunflower-seed oil microcapsules were obtained by spray drying the oil-in-water (O/W) emulsions. The novelty of the approach adopted in the present study is that it utilizes membrane emulsification, which is a low- energy technology, to produce sunflower-seed oil/water emulsions stabilized by maltodextrin which has been dried by spray drying. The results, represented in Table 3, show that the encapsulation efficiency is enhanced by this modern method, the highest values correspond to microcapsules produced by this membrane technology. By comparing the encapsulation efficiency of microcapsules produced by conventional microencapsulation techniques with our laboratory- developed method, it should be mentioned that there is Figure 2. Microscopic image of emulsion sample at 1000 magnification. Table 3. Oil encapsulation efficiency of microcap- sules. Conventional method with ME (1) 36.4 % New method with ME (2) 34.5 % Laboratory blender (3) 28.2 % Table 2. Results of emulsification. Conventional meth- od using ME (1) Droplet size = 9.18 µm Span = 1.5 Average flux = 368.36 dm3 m-2 h-1 New method using ME (2) Droplet size = 7.58 µm Span = 1.72 Average flux = 159.29 dm3 m-2 h-1 Laboratory blender (3) Droplet size = 7.12 µm Span = 1.88 Figure 3. Image of microencapsulated sunflower-seed oil powder produced by conventional ME at 500 magnification. ALBERT, VATAI, AND KORIS Hungarian Journal of Industry and Chemistry 32 no significant difference in terms of encapsulation efficiency. The surface and external morphologies of the microcapsules of sunflower-seed oil were studied using optical microscopy (Figs.3-5). In terms of particle morphology, the microstructures of the powders containing maltodextrin as a wall material were generally spherical in shape, homogeneous and exhibited a smooth surface as seen in the microscopic images. In 2015, Koç et al. [18] reported that the smooth surface of microencapsulated extra virgin olive oil using maltodextrin is related to the low-molecular- weight sugar content of maltodextrin. These low molecular weight sugars may act as plasticizers on the surface of the particles during spray drying. In their study, when WPI (whey protein isolate) was selected as the wall material, the particles exhibited a rough surface compared to those containing maltodextrin. 4. Conclusion Our work focused on basic research in terms of microcapsule production under a laboratory set-up. The main aim of the present investigation was to gain experience with regard to the preparation of microcapsules by membrane emulsification in combination with spray drying using a novel technique and to compare it with the conventional one. The operation of the process and its basic correlations were the primary foci of this paper. Based on the results obtained it is possible to draw up experimental plans that refine and optimize the process. The results of this work demonstrate that a modified membrane emulsification technique, in which the wall material is mixed with the dispersed phase and not with the continuous phase, combined with spray drying can be used to produce microcapsules of sunflower-seed oil that possess an appropriate oil-encapsulation efficiency. Acknowledgement The authors acknowledge the financial support of the ÚNKP-16-3-III New National Excellence Program of the Ministry of Human Capacities. They would like to thank KEYENCE Magyarország for providing the microscopic images. REFERENCES [1] Madene, A.; Jacquet, M.; Scher, J.; Desobry, S.: Flavour encapsulation and controlled release – a review, IJFST, 2006 41(1), 1–21 DOI 10.1111/j.1365- 2621.2005.00980.x [2] Gouin, S.: Microencapsulation: industrial appraisal of existing technologies and trends, TIFS, 2004 15(7-8), 330–347 DOI 10.1016/j.tifs.2003.10.005 [3] Choi, H.G.; Yong, C.S.; Yang, K.Y.: Oral solid preparation composition containing silymarin using membrane emulsification technique, and its pro- duction method, 2010, KR20120034264 [4] Ramakrishnan, S.; Ferrando, M.; Güell, C.: Food grade microcapsules produced by membrane emul- sification, Procedia Engng., 2012 44, 1530–1533 DOI 10.1016/j.proeng.2012.08.855 [5] Chu, L.-Y.; Xie, R.; Zhu, J.-H.; Chen, W.-M.; Ja- maguchi, T.; Nakao, S.-I.: Study of SPG membrane emulsification processes for the preparation of monodisperse core-shell microcapsules, J. Colloid Interface Sci., 2003 265(1), 187–196 DOI 10.1016/S0021-9797(03)00350-3 [6] Figoli, A.; Shanthana Lakshmi, D.; Piacentini, E.; Giorno, L.; Drioli, E.: Preparation of novel ionic liquid loaded polymeric microspheres by mem- brane emulsification process, Procedia Engng., 2012 44, 1287–1290 DOI 10.1016/j.proeng.2012.08.757 Figure 4. Image of microencapsulated sunflower-seed oil powder produced by modified ME at 500 magnification. Figure 5. Image of microencapsulated sunflower-seed oil powder produced by a laboratory blender at 500 magnification. MICROENCAPSULATION OF VEGETABLE OIL 45(2) pp. 29–33 (2017) 33 [7] Imbrogno, A.; Piacentini, E.; Drioli, E.; Giorno, L.: Preparation of uniform polycaprolactone micropar- ticles by membrane emulsification/solvent diffu- sion process, JMSC, 2014 467, 262–268 DOI 10.1016/j.memsci.2014.05.037 [8] Ho, T.H.; Dao, T.P.T.; Nguyen, T.A.; Le, D.D.; Dang, M.C.: Cross-flow membrane emulsification technique for fabrication of drug-loaded particles, Adv. Nat. Sci. Nanosci. Nanotechnol., 2013 4, 045008 1–6 DOI 10.1088/2043-6262/4/4/045008 [9] Calvo, P.; Hernández, T.; Lozano, M.; González- Gómez, D.: Microencapsulation of extra-virgin ol- ive oil by spray-drying: Influence of wall material and olive quality, Eur. J. Lipid Sci. Technol., 2010 112(8), 852–858 DOI 10.1002/ejlt.201000059 [10] Turchiuli, C.; Jimenez Munguia, M.T.; Hernandez Sanchez, M.; Cortesferre, H.; Dumoulin, E.: Use of different supports for oil encapsulation in powder by spray drying, Powder Technol., 2014 255, 103– 108 DOI 10.1016/j.powtec.2013.08.026 [11] Martínez, M.E.; Curti, M.I.; Roccia, P.; Llabot, J.M.; Penci, M.C.; Bodoira, R.M.; Ribotta, P.D.: Oxidative stability of walnut (Juglans regia L.) and chia (Salvia hispanica L.) oils microencapsulated by spray drying, Powder Technol., 2015 270, 271– 277 DOI 10.1016/j.powtec.2014.10.031 [12] Fernandes, R.V.; Botrel, D.A.; Silva, E.K.; Borges, S.V.; Oliveira, C.R.; Yoshida, M.I.; Feitosa, J.P.; de Paula, R.C.: Cashew gum and inulin: new alter- native for ginger essential oil microencapsulation, Carbohydrate Polym., 2016 153, 133–142 DOI 10.1016/j.carbpol.2016.07.096 [13] Shamaei, S.; Seiiedlou, S.S.; Aghbashlo, M.; Tso- tsas, E.; Kharaghani, A.: Microencapsulation of walnut oil by spray drying: Effects of wall material and drying conditions on physicochemical proper- ties of microcapsules, IFSET, 2017 39, 101–112 DOI 10.1016/j.ifset.2016.11.011 [14] Mohammed, N.K.; Tana, C.P.; Manap, Y.A.; Alhelli, A.M.; Hussin, A.S.M.: Process conditions of spray drying microencapsulation of Nigella sati- va oil, Powder Technol., 2017 315, 1–14 DOI 10.1016/j.powtec.2017.03.04 [15] Wang, Y.; Liu, B.; Wen, X.; Li, M.; Wang, K.; Ni, Y.: Quality analysis and microencapsulation of chili seed oil by spray drying with starch sodium octenylsuccinate and maltodextrin, Powder Tech- nol., 2017 312, 294–298 DOI 10.1016/j.powtec.2017.02.060 [16] Hosseinnia, M.; Khaledabad, M.A.; Almasi, H.: Optimization of Ziziphora clinopodiodes essential oil microencapsulation by whey protein isolate and pectin: a comparative study, Int. J. Biol. Macro- mol., 2017 101, 958–966 DOI 10.1016/j.ijbiomac.2017.03.190 [17] Joscelyne, S.M.; Tragardh, G.: Membrane emulsi- fication – a literature review, JMSC, 2000 169(1), 107–117 DOI 10.1016/S0376-7388(99)00334-8 [18] Koç, M.; Güngör, Ö.; Zungur, A.; Yalçin, B.; Sel- ek, İ.S.; Ertekin, F.K.; Ötles, S.: Microencapsula- tion of extra virgin olive oil by spray drying: Effect of wall materials composition, process conditions, and emulsification method, Food Bioprocess Tech- nol., 2015 8, 301–318 DOI 10.1007/s11947-014-1404-9 HUNGARIAN JOURNAL OF INDUSTRY AND CHEMISTRY Vol. 45(2) pp. 35–39 (2017) hjic.mk.uni-pannon.hu DOI: 10.1515/hjic-2017-0018 EFFECT OF CHAIN LENGTH AND ORDER OF THE ALCOHOL ON ENZYME ACTIVITY DURING ENZYMATIC ESTERIFICATION IN ORGANIC MEDIA ZSÓFIA MÁRKUS, KATALIN BÉLAFI-BAKÓ, GÁBOR TÓTH, NÁNDOR NEMESTÓTHY AND LÁSZLÓ GUBICZA* Research Institute on Bioengineering, Membrane Technology and Energetics, University of Pannonia, Egyetem u. 10, Veszprém, 8200, Hungary Esters of short chain acids and alcohols are found in nature as compounds of flavors. Lately the method for their manufacture has been the enzymatic esterification in non-conventional media. Although several reactions have been studied in various media (organic solvents, ionic liquids, supercritical fluids, solvent-free systems), there has been no systematic investigation to clarify the effects of chain length and order of alcohols on the ac- tivity of the enzyme. In this work acetic acid was used as an acyl donor and the roles of the linear and branched chains of C2-C8 primary, secondary and tertiary alcohols on the activity of Novozym 435, the widely used lipase preparation were studied. Both the length of the carbon chain and the order of the alcohol were found to strong- ly influence the activity of the enzyme using the same operational parameters for the reactions. As a result of this project general conclusions were made with regard to the characters of alcohols affecting the reaction rates, which can be applied to other similar reactions. Keywords: enzymatic esterification, non-conventional media, effect of alcohol chain lengths, lipase activity 1. Introduction Enzyme technology provides a promising solution for the biosynthesis of natural flavor esters, since several enzymes are able to catalyze the synthesis of aroma compounds from precursor molecules [1]. Nowadays most flavor compounds are manufactured by conven- tional methods: chemical synthesis or recovery from natural sources. Esters produced chemically are quite common, but their method of production is not consid- ered environmentally safe and cannot be classed as “natural”. Recently interest has been growing in the production of these components by biotransformation, that is the manufacture of natural flavor esters by using natural raw materials. Lipases belong to the most di- verse class of enzymes, they catalyze various reactions due to their wide spectrum of industrial applications. Lipase enzymes have been applied in many industrial sectors, e.g. the food and pharmaceutical industries, in the production of biological detergents (esters of carbo- hydrates), moreover in the manufacture of certain cos- metics and fragrances. Recently interest has grown in the production of natural flavor esters by the biosynthe- sis of short chain acids and alcohols [2,3]. Several similar reactions have been studied, most of which focused on the synthesis of acetates, like ethyl *Correspondence: gubiczal@almos.uni-pannon.hu [4], butyl [5], hexyl [3], cinnamyl [6], and benzyl [7] acetates. During the investigation of the reactions‘ parame- ters the roles of temperature, the molar ratio of acid to alcohol and the amount of enzyme were described in almost every paper. From these data the optimal values of these parameters could be estimated. Numerous reac- tions were carried out in organic solvents [8, 9], ionic liquids [10, 11], supercritical fluids (mainly in super- critical carbon dioxide) [12], solvent-free systems [13], in addition to in the gas phase [14]. Since these reac- tions can be conducted in non-aqueous media, the water content, to be more precise the water activity of the re- action mixture plays an extremely important role in terms of the reaction rate. For the operation of the en- zyme lipase it is necessary to provide a minimal amount of water. On the other hand it is an equilibrium reaction, thus the greater excess of water shifts the reaction to- wards hydrolysis, decreasing the conversion rate signif- icantly. The investigation of the effect of water content, or at least an intention to adjust the initial water content by a constant value is missing in several papers. The water content during the reaction continuously changes due to the production of water in the esterification. This effect can be neglected during the investigation of the initial reaction rates, but in terms of the development of continuous production it should be taken into account. A number of methods are known to maintain water con- tent/activity. From a practical point of view membrane separation processes, like pervaporation provide attrac- tive procedures [15]. MÁRKUS, BÉLAFI-BAKÓ, TÓTH, NEMESTÓTHY AND GUBICZA Hungarian Journal of Industry and Chemistry 36 Beyond the parameters mentioned and the water content, only a limited attention was paid to the role of substrates present in the reaction. As pointed out earlier, acetic acid was the acidic component used in most cas- es. The alcohols applied, however, were much more diverse: linear and branched chains, aromatic and dif- ferent orders of alcohols were investigated. Neverthe- less our group has not found a single study in the litera- ture where systematic research was conducted to assess the role of alcohols on enzyme activity. The effect of the chain length of various alcohols was investigated in reactions catalyzed by enzymes in non-conventional media. Romero and colleagues con- ducted experiments in supercritical carbon dioxide us- ing four different alcohols: propanol, butanol, pentanol and octanol. It was found that higher degrees of conver- sion could be obtained by applying longer chain alco- hols. Enzymes have a higher affinity towards longer chain alcohols, the difference, however, was small [16]. In another paper Romero used butyric acid as an acyl donor and the enzyme Novozym 435. An insignifi- cant difference was observed in terms of the reaction rate or conversion during esterification when the four primary alcohols (propanol, butanol, hexanol and oc- tanol) were applied. Applying secondary alcohols, e.g. 2-hexanol, however, yielded higher reaction rate and conversion than was the case with 2-butanol [17]. Pan et al. studied how the chain length of alcohol compounds can influence the resolution reaction of mandelic acid. Using methanol, ethanol, butanol, hep- tanol and octanol it was found that the highest degree of enantioselectivity could be obtained by ethanol. Moreo- ver the reaction was described as following Michaelis- Menten kinetics in all cases and the inhibition constant increased as the carbon chain go longer [18]. Varma and Madras investigated the esterification of propionic acid and three different alcohols by the enzyme Novozym 435 in supercritical carbon dioxide. Primary (isobutanol, isoamyl alcohol) and secondary alcohols (isopropyl alcohol) were used, as well. Based on the measurements, it was concluded that enzymatic esterification was faster with primary alcohols than with secondary alcohols, moreover a greater degree of con- version was achieved with isobutanol than with isoamyl alcohol [19]. Therefore the aim of this paper was to study the role of alcohols by the preparation of a given enzyme whilst maintaining the operation parameters as con- stants. The esterification primary, secondary and tertiary alcohols of C2-C8 carbon chain lengths were investigat- ed to be able to draw general conclusions concerning the role of alcohol structure on reaction rate. For the measurements a popular immobilized lipase enzyme preparation, Novozym 435®, was used. 2. Experimental 2.1. Chemicals and Enzymes All chemicals: acetic acid, ethanol, 1-propanol. 1- butanol and n-hexane (Merck), 1-pentanol, 2-pentanol tert-butanol (2-methylpropan-2-ol) and 1-hexanol (Sig- ma-Aldrich); 1-heptanol (BDH Chemicals); 1-octanol and isobutanol (2-methylpropan-1-ol) (Spektrum-3D); isoamyl alcohol (3-methylbutan-1-ol) (Molar Chemi- cals); and 2-propanol, 2-butanol, tert-amyl alcohol (2- methylbutan-2-ol) and toluene (Reanal) were of analyti- cal grade. The water content of the chemicals varied greatly, that is why they were dewatered over a 3Å mo- lecular sieve in the form of beads (Sigma-Aldrich). The enzyme used was Novozym 435® from Can- dida antarctica lipase B, immobilized on a macroporous acrylic resin with a water content of 1-2% w/w, which was kindly provided as a gift by Novo Nordisk A/S, Denmark. According to their commercial product man- ual, its catalytic activity was 7000 PLU/g (propyl laurate units/gram). 2.2. Reaction and Analysis Reactions were carried out in 50 mL Erlenmeyer flasks on a laboratory incubator shaker (IKA incubator shaker, KS 4000i) at 150 rpm and 50 oC. The typical reaction mixture contained acetic acid (0.5 mmol), alcohol (3.0 mmol), Novozym 435® lipase (60 mg) and n-hexane (20 mL). The reaction was commenced by adding the en- zyme. The gas chromatography (GC) analyses for the de- termination of ester concentrations were conducted by a HP 5890 A gas chromatograph, with an HP-FFAP col- umn (Macherey-Nagel), split: 70 kPa, N2: 19 cm3/min, using a flame ionization detector (FID). Toluene was used as an internal standard, the changes in ester yield were followed during the reaction. Samples were taken after reaction times of 0.5, 1.0, 2.0 4.0 and 6.0 min. The water contents of the reaction mixtures were determined by a Mettler DL35 Karl Fisher titrator. 3. Results and discussion Although several publications have presented results on the production of flavor esters, the optimal initial condi- tions suggested were quite different and a high degree of deviation was found among data in the literature. Firstly the average of the literature data was used for our preliminary experiments. Based on these figures the following initial parameters were applied: in the reac- tion mixture the molar ratio of acetic acid to alcohol was 1:6, and 20 ml of n-hexane, 20 mmol of toluene and 60 mg of the enzyme Novozym 435 were added to it. ENZYMATIC ESTERIFICATION 45(2) pp. 35–39 (2017) 37 3.1. The effect of water content Experiments had to be conducted to determine the cor- rect water content since it could not be found in the lit- erature. The esterification of acetic acid and isoamyl alcohol – a reaction that has been quite frequently stud- ied – was investigated under the conditions mentioned earlier. During the measurements completely dried reac- tion mixtures (0 % w/w water content) were used and others adjusted the initial water content to the levels of 0.3, 0.5 and 0.7 % w/w by adding water. As can be seen from the data of Table 1, the reaction was extremely slow in the case of completely dried solvents and rea- gents – as was expected. The reaction rate began to rise when the water content grew slowly (and approached the optimal value) due to the water forming in the reac- tion. Based on the experimental results an initial water content of 0.3 % w/w was applied to further measure- ments since this water concentration provided the high- est yield. The yields of esterification after a reaction time of 4 h were presented in Fig.1, where the meas- urements were taken under the conditions given earlier, with an adjusted and the same initial water contents. The yields of esterification were sufficiently high to observe the differences caused by the different struc- tures of alcohols, but saturation levels were not reached and the distinct amounts of water formed during the reaction did not affect such a tendency either. 3.2. Primary alcohols Our study involved linear and branched alcohols with a chain length of C2-C8. As can be seen from Fig.1 for primary alcohols, the yield increased as the length of the carbon chain grew. The effect of chain length was investigated by Romero using propanol, butanol, hexa- nol as well as octanol and a similar conclusion was drawn: acetic acid conversion was greater with alcohols of longer chain lengths, thus the yields of esterification were higher, as well [16]. As far as branched primary alcohols were con- cerned, the opposite trend was observed: the yield of esterification decreased as the chain length increased. At the beginning of the reactions the differences were only minor: only a difference of 3 % was observed in the yields in the cases of alcohols consisting of a carbon chain of 4 or 5. The behavior of isooctanol was espe- cially interesting, because a significant drop in yield was observed compared to the other linear, 8-carbon- chain alcohols – a far smaller amount of ester was formed in the reaction mixture. 3.3. Secondary alcohols In this work three secondary alcohols: 2-propanol, 2-butanol and 2-pentanol were used. As presented in Table 2, an increase in ester yields was observed as the carbon chain length of secondary alcohols grew, as well. The values, however were not as high as for pri- mary alcohols. Neji et al. observed a similar behavior when butanol and 2-butanol were used in the esterifica- tion reaction [20]. Although both alcohols could per- form esterification, yields of esterification were 50 % lower for secondary alcohols. 3.4. Tertiary alcohols Among tertiary alcohols tert-amyl alcohol and tert- butanol were used, however, the enzyme was not able to convert them into esters using acetic acid. From the literature, Stavarache et al. applied tertiary alcohols for transesterification in the production of biodiesel [21]. Similarly no activity was observed during their experi- ments, not even when using ultrasonic radiation. 3.5. Discussion of the experimental results Although our experimental results did not reveal entire- ly general conclusions, which are valid in all cases, ob- vious relationships could be formulated for certain Figure 1. Ester yields during the reactions of acetic acid and primary alcohols after a reaction time of 4 h 31,4 37,3 46,2 76,1 49,6 73,6 53,9 63,2 76,4 7,2 0 10 20 30 40 50 60 70 80 90 Yi e ld [ % ] Table 1. The effect of water content on ester yield in the esterification of acetic acid and isoamyl alcohol Time (h) 0.0% 0.3% 0.5% 0.7% Ester yield (%) 0 0.0 0.0 0.0 0.0 0.5 3.2 16.7 18.5 20.5 1 8.3 28.8 33.4 30.7 2 15.8 50.2 48.3 46.7 4 24.2 73.5 68.2 60.3 6 33.1 85.4 79.7 68.5 Table 2. Ester yields using secondary alcohols Secondary alcohol Ester yield after 4 h (%) 2-propanol 2-butanol 35.8 44.8 2-pentanol 50.5 MÁRKUS, BÉLAFI-BAKÓ, TÓTH, NEMESTÓTHY AND GUBICZA Hungarian Journal of Industry and Chemistry 38 groups of alcohols. It is certain that for alcohols with a linear carbon chain of C2-C8 in length, the conversion rate increases proportionally to the lengthening of the carbon chains. As for branched alcohols, the opposite tendency can be observed: yields were found to de- crease as the carbon chain grew. The relationship is more obvious when the order of alcohols is taken into consideration. Yields of esterifica- tion decreased in the following order: primary alcohol > secondary alcohol > tertiary alcohol (noting that tertiary alcohols did not react at all under the conditions used by Novozym 435). By applying other enzymes, e.g. car- boxylesterase from Bacillus licheniformis, small de- grees of conversion were measured, but remained close to the limit of detection [22]. In an attempt to justify such behavior, it can be assumed that access of the hy- droxyl group of the alcohol to the active centre of the enzyme is severely sterically hindered in the case of secondary and especially tertiary alcohols, which cause enzyme activity to decline or even cease. 4. Conclusion The expected reaction rate produced by a given enzyme can be predicted according to characteristics of the al- cohol used, namely carbon chain length, linear or branched, and the order in production of flavor esters by enzymatic esterification of natural acids and alcohols. The expected ester yield of the esterification reaction using acetic acid as an acyl donor depends on certain characteristics of the alcohol according to a well- defined tendency. It can be assumed that a similar ten- dency (though distinct in terms of rate) could be ob- served for other enzymes regarding the effect of the alcohol. This should be studied separately to decide whether a similar tendency could be detected if various acids are used with the same alcohol. REFERENCES [1] Bommarius, A.S.; Riebel, B.R. Biocatalysis (Wiley-VCH Verlag GmbH & Co. KGaA, Wein- heim, Germany) 2004 pp. 339-372 DOI: 10.1002/aoc.651 [2] Yan, H-D.; Zhang, Q.; Wang, Z.: Biocatalytic syn- thesis of short-chain flavor esters with high sub- strate loading by a whole-cell lipase from Aspergil- lus oryzae, Catal. Commun., 2014 45, 59–62 DOI: 10.1016/j.catcom.2013.10.018 [3] Almeida, A.G.; de Meneses, A.C.; de Araújo, P.H.; de Oliveira, D.: A review on enzymatic synthesis of aromatic esters used as flavor ingredients for food, cosmetics and pharmaceuticals industries, Trends Food Sci. Tech., 2017 69, 95–105 DOI: 10.1016/j.tifs.2017.09.004 [4] Xu, Y.; Wang, D.; Mu, X.Y.; Zhao, G.A.; Zhang, K.C.: Biosynthesis of ethyl esters of short-chain fatty acids using whole-cell lipase from Rhizopus chinesis CCTCC M201021 in non-aqueous phase, J. Mol. Cat. B-Enzym., 2002 18, 29–37 DOI: 10.1016/S1381-1177(02)00056-5 [5] Martins, A.B.; Schein, M.F.; Friedrich, J.L.; Fer- nandez-Lafuente, R.; Ayub, M.A.; Rodrigues, R.C.: Ultrasound-assisted butyl acetate synthesis catalyzed by Novozym 435: Enhanced activity and operational stability, Ultrason. Sonochem., 2013 20(5), 1155–1160 DOI: 10.1016/j.ultsonch. 2013.01.018 [6] Tomke, P.D.; Rathod, V.K.: Ultrasound assisted lipase catalyzed synthesis of cinnamyl acetate via transesterification reaction in a solvent free medi- um, Ultrason. Sonochem., 2015 27, 241–246 DOI: 10.1016/j.ultsonch.2015.04.022 [7] McGinty, D.; Vitale, D.; Letizia, C.S.; Api, A.M.: Fragrance material review on benzyl acetate, Food Chem. Toxycol., 2012 50, S363–S384 DOI: 10.1016/j.fct.2012.02.057 [8] Gubicza, L.; Kabiri-Badr, A.; Keoves, E.; Belafi- Bako, K.: Large-scale enzymatic production of natural flavour esters in organic solvent with con- tinuous water removal, J. Biotechnol., 2000 84, 193–196 DOI: 10.1016/S0168-1656(00)00352-7 [9] Doukyu, N.; Ogino, H.: Organic solvent-tolerant enzymes, Biochem. Eng. J., 2010 48, 270–282 DOI: 10.1016/j.bej.2009.09.009 [10] de los Ríos, A.P.; Hernández-Fernández, F.J.; Tomás-Alonso, F.; Gómez, D.; Víllora G.: Synthe- sis of flavour esters using free Candida antarctica lipase B in ionic liquids, Flavour Frag. J., 2008 23, 319–322 DOI: 10.1002/ffj.1884 [11] Major, B.; Nemestóthy, N.; Bélafi-Bakó, K.; Gu- bicza, L.: Enzymatic esterification of lactic acid under microwave conditions in ionic liquids, Hung. J. Ind. Chem., 2008 36, 77–81 [12] dos Santos, P.; Meireles, M.A.; Martínez, J.: Pro- duction of isoamyl acetate by enzymatic reactions in batch and packed bed reactors with supercritical CO2, J. Supercrit. Fluids, 2017 127, 71–80 DOI: 10.1016/j.supflu. 2017.03.019 [13] Sun, J.; Yu, B.; Curran, P.; Liu, S.Q.: Lipase- catalysed ester synthesis in solvent-free oil system: Is it esterification or transesterification?, Food Chem., 2013 141, 2828–2832 DOI: 10.1016/j.foodchem.2013.05.109 [14] Csanádi, Z.; Kurdi, R.; Bélafi-Bakó, K.: Ethyl- acetate synthesis in gas phase by immobilized li- pase, Hung. J. Ind. Chem., 2012 40, 39–44 [15] Findrik, Z.; Nemeth, G.; Vasic-Racki, D.; Belafi- Bako, K.; Csanadi, Z.; Gubicza, L.: Pervaporation- aided enzymatic esterifications in non-conventional media, Process Biochem., 2012 47, 1715–1722 DOI: 10.1016/j.procbio.2012.08.003 [16] Romero, M.; Calvo, L.,; Alba, C.; Habulin, M.; Primozic, M.; Knez, Z.: Enzymatic synthesis of isoamyl acetate with immobilized Candida antarctica lipase in supercritical carbon dioxide, J. Supercrit. Fluids, 2005 33, 77-84 DOI: 10.1016/j.supflu.2004.05.004 ENZYMATIC ESTERIFICATION 45(2) pp. 35–39 (2017) 39 [17] Romero, M.D.; Gómez, J.M.; Díaz-Suelto, B.; García-Sanz, A.: Study of the influence of alcohols in the synthesis of short chain esters, Chem. Eng. Trans., 2011 24, 37-42 DOI: 10.3303/CET1124007 [18] Pan, Y.; Tang, K.V.; He, C.; Yi, W.; Zhu, W.; Liu, Y.: Effect of alcohol chain length on the enzymatic resolution of racemic mandelic acid and kinetic study, Biotechnol. Appl. Biochem., 2014 61(3), 274-279 DOI: 10.1002/bab.1170 [19] Kumar, R.; Modak, J.; Madras, G.: Effect of chain length of alcohol on the lipase-catalyzed esterification of propionic acid in supercritical carbon dioxide, Appl. Biochem. Biotechnol. 2010 160, 2342-2354 DOI: 10.1016/j.bej.2005.01.007 [20] Neji, S.B.; Trabelsi, M.; Frikha, M.H.: Esterification of fatty acids with short-chain alcohols over commercial acid clays in a semi- continuous reactor, Energies, 2009 2, 1107-1117 DOI: 10.3390/en20401107 [21] Stavarache, C.; Vinatoru, M.; Nishimura, R.,; Maeda, Y.: Fatty acids methyl esters from vegetable oil by means of ultrasonic energy, Ultrason. Sonochem., 2005 12, 367-372 DOI: 10.1016/j.ultsonch.2004.04.001 [22] Berger, R.G.: Biotechnology as a source of natural volatile flavors, Curr. Opin. Food Sci., 2011 1(2), 38–43 DOI: 10.1016/j.cofs.2014.09.003 HUNGARIAN JOURNAL OF INDUSTRY AND CHEMISTRY Vol. 45(2) pp. 41–44 (2017) hjic.mk.uni-pannon.hu DOI: 10.1515/hjic-2017-0019 STATE-OF-THE-ART RECOVERY OF FERMENTATIVE ORGANIC ACIDS BY IONIC LIQUIDS: AN OVERVIEW KONSTANTZA TONOVA* Institute of Chemical Engineering, Bulgarian Academy of Sciences, Acad. G. Bonchev Str., Bl. 103, 1113 Sofia, BULGARIA The main achievements of liquid–liquid extraction (LLE) of fermentative organic acids from their aqueous sources using a diverse range of ionic liquids are summarized since the first study appeared in 2004. The litera- ture survey is organized in consideration of the distinct chemical structures of the organic acids. The acids dis- cussed include mono– or dicarboxylic ones (butyric, L-malic and succinic acids), acids bearing both carboxyl and hydroxyl groups (L-lactic, citric and mevalonic acids), and volatile organic acids (mainly acetic acid). Infor- mation is given about ionic liquids applied in recovery, and the resultant extraction efficiencies and partition co- efficients. As the topic is novel and experimental studies scarce, the selection of the ionic liquids that were test- ed still seems random. This may well change in the future, especially after improving the ecological and toxico- logical characteristics of the ionic liquids in order to bring about an “in situ” method of extraction without harming the microbial producers of the organic acids. Keywords: extraction, ionic liquid, organic acid, recovery, re–extraction 1. Introduction Room temperature ionic liquids (ILs) exist as molten salts at ambient temperature and consist entirely of ions, usually a charge–stabilized organic cation and an inorganic or organic anion. ILs can be tailored to a wide variety of applications by combining different ions [1] and for this reason they are often called “designer solvents”. ILs exhibit a broad range of unique properties, including negligible vapor pressure, high thermal stability and low chemical reactivity [2]. The union of these particular properties, together with finely tunable density, viscosity, polarity and miscibility with other common solvents favor the application of ILs in different kinds of separation and reaction processes [3– 8]. Considering the benefits that arise from the properties of ILs, Matsumoto et al. [9] first proposed an environmentally friendly system for the extraction of fermentative L-lactic acid. They used hydrophobic [CnC1im][PF6] instead of volatile organic solvents as diluents of reactive organic bases. These ILs proved to be nontoxic towards the lactic acid producing bacterium Lactobacillus rhamnosus, but provided low degrees of solubility of the reactive amines which resulted in insufficient levels of extraction efficiency. Nevertheless, these results suggest possible applications of ILs in extractive fermentations. *Correspondence: konstantzatonova@yahoo.com 2. Discussion on the organic acids extracted and the ionic liquids applied 2.1. Butyric acid and phosphinate–based ILs The most remarkable results regarding the partition coefficient of an organic acid in an IL have been documented with regards to the extraction of butyric acid, the four–carbon fatty acid, with phosphinate-based ([Phos]) ILs. [P6,6,6,14][Phos] and a novel ammonium phosphinate, [CnCnCnC1N][Phos], were studied [10-11]. Distribution coefficients of about 80 were obtained using the low concentrations of butyric acid, and the extraction efficiency was just as high in the pure (water saturated) IL as in the IL/water/dodecane reversed micellar solution. The ammonium phosphinate absorbed a relatively high amount of water until saturation was achieved, ca 21 wt%. (about 12 water molecules per ion pair of the IL), which implies that an aqueous biphasic system was formed. 2.2. Dicarboxylic acids and phosphonium- or imidazolium-based ILs Among phosphonium-based ILs, [P6,6,6,14]Cl seems the most suitable extractant for the recovery of low and moderate concentrations of dicarboxylic L-malic acid in aqueous solutions [12]. The other phosphonium-based ILs and higher acid concentrations entrain third-phase formation, especially in the case of [P6,6,6,14][Phos] when a large amount of the acid content (ca 40%) remains TONOVA Hungarian Journal of Industry and Chemistry 42 uncovered in both phases. The [P6,6,6,14]Cl–rich phase is also the best extractant for another dicarboxylic acid, succinic acid [12]. Extractions with [Dec]- and [Phos]– based ILs resulted in a substantial amount of undetectable acid in both phases, which was attributed to the formation of complexes between the organic acid and the extractants that were not quantified. More recently succinic acid attracted special attention in a comprehensive study where the extraction was carried out by aqueous biphasic systems (ABS) of alcohols/salts or imidazolium-based ILs/salts [13]. Successful recovery was achieved by both systems. Succinic acid preferentially migrates to the IL–rich phase in all systems formed of [C6C1im]Br and a kosmotropic salt (phosphate, sulfate, carbonate or citrate). The IL salted out by (NH4)2SO4 or K2CO3 exhibited the highest levels of extractability. The pH values of these systems were quite different. The pH of the system with (NH4)2SO4 was 3.43 which is below the pKa values of succinic acid (pKa1 = 4.21, pKa2 = 5.72), while pH = 10.50 for K2CO3 greatly exceeded the pKas. This suggests that unlike the aqueous biphasic systems with alcohols, the extraction capacity of the IL/salts systems towards succinic acid is not pH–dependent and is most likely related to the proper nature of the solvent (IL/salt) and the solute (acid). For the same IL, [C6C1im]Br, an excellent solvating capacity to the lactic acid was reported [14] so that the acid could be extracted from a concentrate of white wine. This way the extraction efficiencies of the ABS of [C6C1im]Br/(NH4)2SO4 or K2CO3 are comparable to those obtained with the hydrophobic IL [P6,6,6,14]Cl [12]. Moreover, the re–extraction efficiency achieved was superior at ~71%. Succinic acid was obtained in a crystalline form by direct precipitation with sodium hydroxide. 2.3. Acids with both hydroxyl and carboxyl groups and phosphonium- or imidazolium-based ILs Different types of phosphonium-based IL biphasic systems were applied for L–lactic acid recovery. An extraction efficiency of above 80% was achieved by using either pure [P6,6,6,14][Phos] [12] or a mixed biphasic system of [P6,6,6,14]Cl and an inorganic kosmotrope, MgSO4 [15]. The kosmotropic salt engages more water molecules when hydrated thus rendering the microenvironment of the acid more hydrophobic which favors the undissociated form of acid suitable for extraction. All extraction systems of phosphonium- based ILs with long side chains suffer from the common disadvantage of forming stable emulsions or a third phase between the IL–rich phase and aqueous solution. This drawback is avoided by applying ILs of an imidazolium cationic moiety, however, in the majority of the cases these ILs exhibit low levels of extraction efficiency towards lactic acid [9,16] and other acids bearing both hydroxyl and carboxyl groups (citric and mevalonic acids) [16]. An advantageous ABS of imidazolium saccharinate, that possesses a long side chain, [C8/10C1im][Sac], has been exploited lately and it was shown that when it is combined with an inorganic kosmotropic salt (that retains water from solubilization into the IL–rich phase) an extraction efficiency of 81% and partition coefficient of 5.5 could be achieved [17]. The extraction yield of lactic acid was as high as 90% in a two–step recovery by [C8C1im][Sac] with or without the addition of a kosmotropic salt (MgSO4). Moreover, successful acid re–extraction of 95% from the IL–rich phase was attained by means of a solution containing an alkaline kosmotrope, K2HPO4. 2.4. Volatile fatty acids and phosphonium- based ILs Apart from culture broths, fermented wastewater streams still represent an unexploited source of platform chemicals, including volatile organic acids. Volatile fatty acids are versatile carboxylic acids involved in the synthesis of bioplastics and other value–added chemicals [18]. The composition of fermented wastewater typically contains ~1 wt% of volatile fatty acids, but also a significant amount of various dissolved salts. The low concentrations of the volatile fatty acids and the large quantity of inorganic salt–originating ions result in pH–values of between 4 and 6, which are in favor of the deprotonated acid form and thus do not support complexation with the IL. The distribution of acetic acid between model solutions with or without salts and different solvents, including phosphonium- based ILs, was recently studied [19]. Similarly to the butyric and lactic acids [10,20], the low concentration of acetic acid and the use of [P6,6,6,14][Phos] were the best conditions to obtain the highest partition coefficient in the IL–rich phase starting from an idealized aqueous solution containing only the acetic acid. In the presence of salts (KCl, Na2SO4 or Na2HPO4), however, the partition coefficients reported for [P6,6,6,14]Cl were the highest in the series of ILs tested and exceeded even those obtained in the classical extraction by trioctylamine (TOA)/n-octanol. [P6,6,6,14]Cl as a solvent has an inevitable drawback related to its measurable level of leaching into the aqueous phase due to the hydrophilicity of the [Cl]–. Contrary to [P6,6,6,14]Cl, [P6,6,6,14][Phos] and [P6,6,6,14][N(CN)2] were found to be highly stable as significant leaching was not detected in the aqueous phases [19]. Extraction by [P6,6,6,14][Phos], however, was strongly affected by the ions of the salts present in the feed, while [P6,6,6,14]Cl and [P6,6,6,14][N(CN)2] kept extraction capacities constant for acetic acid. When the source contained different acids, mimicking actual fermented wastewater, it was found that the growing hydrophobic domain in the acid leads to higher degrees of extraction. Butyric acid was the most extracted acid from the fermented wastewater, while lactic acid was the most challenging acid to extract. By modifying the solvent properties of [P6,6,6,14][Phos] by sparging pressurized CO2, a further increase in the extractability of acetic acid was observed RECOVERY OF FERMENTATIVE ORGANIC ACIDS BY IONIC LIQUIDS: OVERVIEW 45(2) pp. 41–44 (2017) 43 [21]. The effect was attributed to the altered structure of the fluid which becomes more accessible for the acetic acid. This finding constitutes a general concept for the improvement of extraction processes other than those involving volatile fatty acids. ILs can act as solvents and simultaneously mediate reactive extraction to valorize low–titer volatile fatty acids. This has been recently shown through an IL– mediated esterification of acetic acid recovered from dilute aqueous streams [22]. The acids produced in anaerobic digestion or fermentation were transferred to a nonvolatile hydrophobic phase where they reacted with an alcohol (ethanol) in order to generate volatile, value–added esters of low solubility. [P6,6,6,14]–ILs were selected for their potentially high extracting capacity and hydrophobicity. Their hydrophobic character provides a water-excluding site for esterification and a nonvolatile carrier for the evaporation of the ester produced. Significant accumulation of acetic acid in the IL was achieved by using [P6,6,6,14][N(CN)2], but this was mainly due to the exchange of [N(CN)2] – for the acetate anion as the dicyanamide anion was found to hydrolyze under the extraction conditions used, including at an elevated temperature (75 °C). Contrary to the extraction, [P6,6,6,14][N(CN)2] and [P6,6,6,14]Cl appeared to be the worst media for performing esterification, while the best was [P6,6,6,14][Tf2N], which, however, is poor and costly extractant. Thus an IL of combined anions, Cl– + [Tf2N]–, was tested which could be used in a multistage way. Starting from an aqueous stream of 0.33 mol dm-3 acetic acid, 0.44 mol dm-3 accumulated in the mixed [P6,6,6,14]Cl+[Tf2N] which allowed an esterification conversion of 56% to be achieved over 30 min. 3. Conclusion ILs are commonly considered more sustainable than classical organic solvents. It is well known that the toxicity level of conventional solvents to microbes limits their compatibility with fermentation broths. However, the label of “green solvent”, assigned to the ILs, has led to the delusion that they are nontoxic and biodegradable, which is not true about some of the most employed ILs. For example, the commonly used [P6,6,6,14]Cl may be regarded as toxic in aquatic environments exhibiting much higher levels of ecotoxicity compared to ordinary organic solvents [23]. The biocompetitiveness and biodegradability of ILs are not still convincingly argued for [24-25]. The need for novel extractants with improved characteristics from ecological and toxicological standpoints can be put forward. By taking into account that aqueous streams and bioorganics are treated, the environmental impact of ILs should be resolved as a result of future studies. SYMBOLS IL’s cationic moiety: [CnC1im] 1-alkyl-3-methylimidazolium [CnCnCnC1N] trialkylmethylammonium [P6,6,6,14] tetradecyl(trihexyl)phosphonium IL’s anionic moiety: [Dec] decanoate [N(CN)2] dicyanamide [Phos] bis(2,4,4-trimethylpentyl)phosphinate [Sac] saccharinate (which is a benzoic sulfimide) [Tf2N] bis(trifluoromethylsulfonyl)imide Other: TOA trioctylamine Acknowledgement This research was supported by the Bulgarian Science Fund (Contract grant DFNI–B01/23). REFERENCES [1] Blundell, R.K.; Licence, P.: Quaternary ammonium and phosphonium based ionic liquids: A compari- son of common anions, Phys. Chem. Chem. Phys., 2014 16(29), 15278–15288 DOI: 10.1039/C4CP01901F [2] Freemantle, M.: An Introduction to Ionic Liquids (RSC Publishing, Cambridge, UK) 2009 [3] Mutelet, F.; Jaubert, J.-N.: Interactions between organic compounds and ionic liquids. Selectivity and capacity characteristics of ionic liquids, Chap- ter 10 in Ionic Liquids: Theory, Properties, New Approaches, Ed.: Kokorin, A. (InTech) 2011 DOI: 10.5772/14291 [4] Tonova, K.: Separation of poly- and disaccharides by biphasic systems based on ionic liquids, Sep. Purif. Technol., 2012 89, 57–65 DOI: 10.1016/j.seppur.2012.01.007 [5] Keremedchieva, R.; Svinyarov, I.; Bogdanov, M.G.: Ionic liquid–based aqueous biphasic systems – A facile approach for ionic liquid regeneration from crude plant extracts, Processes, 2015 3(4), 769–778 DOI: 10.3390/pr3040769 [6] Tonova, K.; Bogdanov, M.G.: Partitioning of α- amylase in aqueous biphasic system based on hy- drophobic and polar ionic liquid: Enzyme extrac- tion, stripping and purification, Sep. Sci. Technol., 2017 52(5), 812–823 DOI: 10.1080/01496395.2016.1267211 [7] Fehér, E.; Illeová, V.; Kelemen-Horváth, I.; Bélafi- Bakó, K.; Polakovič, M.; Gubicza, L.: Enzymatic production of isoamyl acetate in an ionic liquid– alcohol biphasic system, J. Mol. Catal. B: Enz., 2008 50(1), 28–32 DOI: 10.1016/j.molcatb.2007.09.019 [8] Major, B.; Nemestóthy, N.; Bélafi-Bakó, K.; Gu- bicza, L.: Enzymatic esterification of lactic acid under microwave conditions in ionic liquids, Hung. J. Ind. Chem., 2008 36(1-2), 77–81 TONOVA Hungarian Journal of Industry and Chemistry 44 [9] Matsumoto, M.; Mochiduki, K.; Fukunishi, K.; Kondo, K.: Extraction of organic acids using imid- imidazolium–based ionic liquids and their toxicity to Lactobacillus rhamnosus, Sep. Purif. Technol., 2004 40(1), 97–101 DOI: 10.1016/j.seppur.2004.01.009 [10] Marták, J.; Schlosser, Š.: Liquid–liquid equilibria of butyric acid for solvents containing a phosphonium ionic liquid, Chem. Pap., 2008 62(1), 42–50 DOI: 10.2478/s11696-007-0077-5 [11] Blahušiak, M.; Schlosser, Š.; Marták, J.: Extraction of butyric acid with a solvent containing ammonium ionic liquid, Sep. Purif. Technol., 2013 119, 102–111 DOI: 10.1016/j.seppur.2013.09.005 [12] Oliveira, F.S.; Araújo, J.M.M.; Ferreira, R.; Rebelo, L.P.N.; Marrucho, I.M.: Extraction of L- lactic, L-malic, and succinic acids using phosphonium–based ionic liquids, Sep. Purif. Technol., 2012 85, 137–146 DOI: 10.1016/j.seppur.2011.10.002 [13] Pratiwi, A.I.; Yokouchi, T.; Matsumoto, M.; Kondo, K.: Extraction of succinic acid by aqueous two–phase system using alcohols/salts and ionic liquids/salts, Sep. Purif. Technol., 2015 155, 127– 132 DOI: 10.1016/j.seppur.2015.07.039 [14] Lateef, H.; Gooding, A.; Grimes, S.: Use of 1- hexyl-3-methylimidazolium bromide ionic liquid in the recovery of lactic acid from wine, J. Chem. Technol. Biotechnol., 2012 87(8), 1066–1073 DOI: 10.1002/jctb.3843 [15] Tonova, K.; Svinyarov, I.; Bogdanov, M.G.: Biocompatible ionic liquids in liquid–liquid extraction of lactic acid: A comparative study, Int. J. Chem. Nuclear Mater. Metallurgical Eng., 2015 9(4), 526–530 https://www.waset.org/publications/10001024 [16] Li, Q.Z.; Jiang, X.L.; Zou, H.B.; Cao, Z.F.; Zhang, H.B.; Xian, M.: Extraction of short–chain organic acids using imidazolium–based ionic liquids from aqueous media, J. Chem. Pharm. Res., 2014 6(5), 374–381 http://www.jocpr.com/articles/extraction-of- shortchain-organic-acids-using-imidazoliumbased-ionic-liquids- from-aqueous-media.pdf [17] Tonova, K.; Svinyarov, I.; Bogdanov, M.G.: Hydrophobic 3-alkyl-1-methylimidazolium saccharinates as extractants for L-lactic acid recovery, Sep. Purif. Technol., 2014 125, 239–246 DOI: 10.1016/j.seppur.2014.02.001 [18] Straathof, A.J.J.: Transformation of biomass into commodity chemicals using enzymes or cells, Chem. Rev., 2014 114(3), 1871–1908 DOI: 10.1021/cr400309c [19] Reyhanitash, E.; Zaalberg, B.; Kersten, S.R.A.; Schuur, B.: Extraction of volatile fatty acids from fermented wastewater, Sep. Purif. Technol., 2016 161, 61–68 DOI: 10.1016/j.seppur.2016.01.037 [20] Marták, J.; Schlosser, Š.: Extraction of lactic acid by phosphonium ionic liquids, Sep. Purif. Technol., 2007 57, 483–494 DOI: 10.1016/j.seppur.2006.09.013 [21] Reyhanitash, E.; Zaalberg, B.; Ijmker, H.M.; Kersten, S.R.A.; Schuur, B.: CO2–enhanced extraction of acetic acid from fermented wastewater, Green Chem., 2015 17(8), 4393–4400 DOI: 10.1039/C5GC01061F [22] Andersen, S.J.; Berton, J.K.E.T.; Naert, P.; Gildemyn, S.; Rabaey, K.; Stevens, C.V.: Extraction and esterification of low–titer short– chain volatile fatty acids from anaerobic fermentation with ionic liquids, Chem. Sus. Chem., 2016 9(16), 2059–2063 DOI: 10.1002/cssc.201600473 [23] Wells, A.S.; Coombe, V.T.: On the freshwater ecotoxicity and biodegradation properties of some common ionic liquids, Org. Process Res. Dev., 2006 10(4), 794–798 DOI: 10.1021/op060048i [24] Siedlecka, E.M.; Czerwicka, M.; Neumann, J.; Stepnowski, P.; Fernández, J.F.; Thöming, J.: Ionic liquids: Methods of degradation and recovery, Chapter 28 in Ionic Liquids: Theory, Properties, New Approaches, Ed.: Kokorin, A. (InTech) 2011 DOI: 10.5772/15463 [25] Egorova, K.S.; Ananikov, V.P.: Toxicity of ionic liquids: Eco(cyto)activity as complicated, but unavoidable parameter for task–specific optimization, Chem. Sus. Chem., 2014 7(2), 336– 360 DOI: 10.1002/cssc.201300459 HUNGARIAN JOURNAL OF INDUSTRY AND CHEMISTRY Vol. 45(2) pp. 45–49 (2017) hjic.mk.uni-pannon.hu DOI: 10.1515/hjic-2017-0020 APPLICATION OF A HYDROPHOBIC POLYMERIC MEMBRANE FOR CARBON DIOXIDE DESORPTION FROM AN MEA-WATER SOLUTION ZENON ZIOBROWSKI * , ADAM ROTKEGEL Institute of Chemical Engineering of the Polish Academy of Sciences, ul. Balycka 5, 44-100 Gliwice, POLAND Carbon dioxide desorption from a monoethanolamine (MEA) solution using a hydrophobic polydimethylsiloxane (PDMS) tubular membrane on a ceramic support is presented. The effects of operating parameters such as feed temperature, liquid flow rate and MEA concentration on mass transfer were examined. The mass transfer of CO2 from the liquid to gaseous phase was predicted by a multilayer film model with an accuracy of ±25%. Research into new selective materials is needed to develop more efficient and environmentally friendly CO2 capture tech- nology Keywords: MEA, desorption, carbon dioxide, hydrophobic membrane, PDMS 1. Introduction Fossil fuel combustion from power plants is one of the most significant sources of CO2 emissions [1]. The sep- aration of carbon dioxide from gases can be realized by processes such as adsorption, absorption, low tempera- ture distillation and membrane separation. The absorp- tion of carbon dioxide in amine based solutions is cur- rently the most widespread method in industry for the post-combustion capture of CO2 [2]. The advantage of chemical absorption in amine so- lutions is the fact that at higher temperatures the chemi- cal reaction can be reversed and the amine recycled. On the other hand, obstacles include a relatively low CO2 capture capacity, solvent losses caused by evaporation, thermal stability, highly corrosive characteristics, eco- toxicity and biodegradability in the natural environment [2-4]. It was shown that MEA and diethanolamine (DEA) might promote potential long-term toxicity ef- fects towards living organisms [5,6]. In addition the regeneration step may increase the total operating costs of the capture plant by up to 70%, especially for prima- ry and tertiary amines where the heat of reaction is quite high [7]. The amine scrubbing processes carried out in packed columns are currently most widely used in in- dustry for the post-combustion capture of CO2. Limiting factors for the application of this technology are its size and large capital costs. The mass transfer performance of this solution can be reduced by flooding, foaming and entrainment conditions. *Correspondence: zenz@iich.gliwice.pl In comparison to the studies on CO2 absorption in MEA solutions there are only a few concerning CO2 desorption, despite the fact that the stripping unit is re- sponsible for most of the separation cost of the process [8]. It is important that materials used in the processes concerning post-combustion capture of CO2 exhibit low or no environmental effects. Various tubular membranes were operated as catalyst supports [9]. Recently a new type of ceramic hollow fiber membrane contactor has been studied [10]. This kind of membrane can be modi- fied to be hydrophobic which enables it to be applied for CO2 absorption-desorption in amine solutions. In this study the process of CO2 removal from an MEA solution using a hydrophobic polydimethylsiloxane (PDMS) tubular membrane on a ceramic support was investigated. 2. Experimental 2.1. Experimental setup The experimental setup shown in Fig.1 consisted of a membrane module, reactor vessel, cooling system, as well as circulation and vacuum pumps. The hydropho- bic PDMS membranes on ceramic support (ceramic tubes with an outer diameter of 0.01 m and length of 0.25 m using a PVM 250 membrane module made by Pervatech BV) was studied. The feed was circulated by a pump and the flow rate was controlled by a flowmeter. In all experiments the feed temperature was stabilized by a thermostat (1C). The permeate was condensed and collected in cold traps immersed in liquid nitrogen. The vacuum pump was used to maintain the pressure between 7 and 10 mmHg on the permeate side. The concentration of ZIOBROWSKI AND ROTKEGEL Hungarian Journal of Industry and Chemistry 46 carbon dioxide in the permeate was calculated by meas- uring the mass of carbon dioxide and water in the ana- lyzed permeate sample. The pressures on the feed and permeate sides were measured by pressure gauges. The temperatures of the feed in the reactor vessel, before and after the membrane module were measured by thermo- couples. Pure monoethanolamine (MEA) and deionised wa- ter were used to prepare the liquid-feed solution. After- wards the obtained solution was loaded with CO2 by bubbling pure CO2 in a magnetically stirred vessel until the required carbonation ratio, , was achieved. In our experiments the carbonation ratio was determined by measuring the mass of absorbed CO2 in the amine solu- tion at a given temperature. Additionally, independent pervaporation experi- ments with the same PDMS membrane under similar thermal and hydrodynamic conditions for a 2-propanol – water mixture were performed to estimate the mem- brane resistance (1/kM). 2.2. Experimental results The performance of the PDMS membrane was exam- ined experimentally. The operating temperature was between 323 and 348K (50 and 75°C), liquid flow rate between 20 and 600 l/h and the MEA concentrations were 5, 10 and 15 wt%. The effect of liquid flow rate on the CO2 mass flux and selectivity is presented in Figs.2 and 3 for the tem- perature of 323K (50°C) and 10% MEA concentration. The selectivity of the process is defined as follows: 2 2 2 2 CO CO CO CO ( (1 )) ( (1 )) p f w w S w w    (1) The measured fluxes increase with the Reynolds num- ber. The highest values were obtained for Re>10,000 (turbulent flow). This can be explained by the CO2 mass transfer increase in the liquid phase for turbulent re- gime. The measured selectivities rise with the Reynolds number and for turbulent flows reach the value of 10. The operating temperature is an important parame- ter as far as the efficiency of the membrane is concerned as shown in Fig.4. For a given turbulent liquid flow rate the measured CO2 mass fluxes rise with the feed tem- peratures due to the increased driving force in favour of CO2 mass transfer. The selectivity does not change sig- nificantly with the operating temperature, Fig.5. The effect of the MEA concentration on mass flux and selectivity is presented in Figs.6-7 at an operating temperature of 323K (50°C) and turbulent flow (Re of about 40,000). The measured mass fluxes do not change signifi- cantly with MEA concentration (Fig.6), because of the Figure 1. The experimental setup: 1 – membrane contactor, 2 – feed tank, 3 – cold traps, 4 – circulation pump, 5 – vacuum pump, 6 – heater Figure 3. The effect of Re number on selectivity (T = 50°C and wMEA = 10 wt%) Figure 2. The effect of Re number on CO2 mass flux (T = 50°C, wMEA = 10 wt%) Figure 4. The effect of feed temperature on CO2 mass flux (wMEA = 10 wt%) APPLICATION OF HYDROPHOBIC POLYMERIC MEMBRANE FOR CO2 DESORPTION ... 45(2) pp. 45–49 (2017) 47 relationship between equilibrium constants of the CO2 - MEA reaction and the CO2 solubility in water at a given temperature. The selectivity decreases with MEA con- centration as a result of the rising amount of CO2 ab- sorbed in the MEA solution and the constant CO2 flux in the permeate, see Fig.7. 3. Mathematical model and calculation results When CO2 is absorbed in aqueous monoethanolamine (MEA) solution, the following reactions can be written as [11]: slow 2 2 2CO RNH RN H COO  (2) fast 2 2 3RN H COO RNH RNH RNHCOO     (3) The formation of carbamate is well understood and the rate of the forward reaction has been determined as first order with respect to both CO2 and RNH2: CF 2 2[CO ][RNH ]r k (4) During the desorption process the differences in the concentration of the component and the temperature between the inlet and outlet in the liquid phase are very small. Therefore, the desorption rate may be simply calculated using the arithmetic mean value of CO2 in the liquid phase. With this assumption we can calculate the mass fluxes of CO2 can be calculated as follows: 2 2 2 * LCO CO CO( )N K x x  (5) where NCO2 [kmol/s] is the flux of CO2 and KL [kmol/m2s] is the overall mass-transfer coefficient of the liquid phase. The overall mass-transfer coefficient for CO2 can be evaluated by a resistance-in-series model [12]. The numerical calculations based on model equa- tions were performed and estimated values of mem- brane resistance (1/kM) used. In the calculations the ex- perimental values of the Henry’s constant for CO2 in water and MEA under standard conditions are 1.2456 and 1.5732, respectively [13]. The enhancement factor of the chemical reaction of CO2 in the liquid phase, as defined by DeCoursey [14], was between 20 and 60. The viscosity of the water–MEA mixture was calculated according to a Grunberg and Nissan equation [15]. Cal- culated and experimental values of CO2 mass fluxes are Figure 5. The effect of feed temperature on selectivity (wMEA = 10 wt%) Figure 6. The effect of MEA concentration on CO2 mass flux Figure 7. The effect of MEA concentration on selectivity Figure 8. Comparison of calculated values of CO2 fluxes with experimental ones ZIOBROWSKI AND ROTKEGEL Hungarian Journal of Industry and Chemistry 48 shown in Fig.8. The scattering of calculated and exper- imental values of CO2 mass fluxes was within the range of ±25% . The experimental values of CO2 mass fluxes were compared with those obtained from the literature for CO2 stripping in a ceramic hollow fiber membrane con- tactor [10]. In spite of the different types of membrane type and hydrodynamic conditions the measured values of CO2 mass fluxes were comparable in both cases. Conclusions The application of a membrane in the process of CO2 stripping from MEA solutions avoids some technical problems that are encountered in industrial practices. The PDMS hydrophobic tubular membrane on a ceramic support can be applied for the removal of CO2 from MEA solutions. In developed turbulent flows the measured CO2 mass fluxes and selectivities do not change significantly with Re number (Figs.2-3). The measured CO2 mass fluxes increase as the feed tempera- ture rises (Fig.4) and slightly depend on the MEA con- centration (Fig.6). The measured and calculated CO2 mass fluxes are in good agreement with each other (Fig.8). The ±25% variation in scattering can be ex- plained by the accuracy of the correlations, experi- mental precision and simplification of the model. 4. SYMBOLS C – concentration, kmol m-3 D – diffusion coefficient, m2 s-1 KL – overall mass transfer coefficient, kmol m-2 s-1 kM – mass transfer coefficient of the membrane, kmol m-2 s-1 N – mass flux kmol m-2s-1 r – reaction rate, kmol s-1 Re – Reynolds number S – selectivity T – temperature, K w – mass fraction x – mole fraction of CO2 in the liquid phase superscripts * - refers to equilibrium subscripts calc – calculation CO2 – carbon dioxide exp - experimental f – feed G – gaseous phase L – liquid phase p - permeate REFERENCES [1] Budzianowski W.M.: Single solvents, solvent blends, and advanced solvent systems in CO2 cap- ture by absorption: a review, Int. J. Global Warm- ing, 2015 7 (2), 184-225, DOI: 10.1504/IJGW.2015.067749 [2] Rochelle G.T.: Amine scrubbing for CO2 capture, Science, 2009 325, 1652-1654, DOI: 10.1126/science.1176731 [3] Zhao B.; Sun Y.; Yuan Y.; Gao J.; Wang S.; Zhuo Y.; Chen C.: Study on corrosion in CO2 chemical absorption process using amine solution, Energy Procedia, 2011 4, 93–100, DOI: 10.1016/j.egypro.2011.01.028 [4] Eide-Haugmo I. et al.: Environmental impact of amines, Energy Procedia 2009 1, 1297–1304, DOI: 10.1016/j.egypro.2009.01.170 [5] Libralato G.; Volpi Ghirardini A.; Avezzù F.: Seawater ecotoxicity of monoethanolamine, dieth- anolamine and triethanolamine, J. Hazard. Mat., 2010 176, 535–539, DOI: 10.1016/j.jhazmat.2009.11.062 [6] Ethanolamine Compounds (MEA, DEA, TEA And Others), online: http://www.safecosmetics.org/get- the-facts/chemicals-of-concern/ethanolamine- compounds/, accessed: 2017-10-05 [7] Schäfer B.; Mather A.E.; Marsh K.N.: Enthalpies of solution of carbon dioxide in mixed solvents, Fluid Phase Equilib., 2002 194, 929-935, DOI: 10.1016/S0378-3812(01)00722-1 [8] Dugas R.; Rochelle G.: Absorption and desorption rates of carbon dioxide with monoethanolamine and piperazine, Energy Procedia, 2009 1, 1163- 1169, DOI: 10.1016/j.egypro.2009.01.153 [9] Keil F. J.; Flügge U.: High Performance Catalytic Tubular Membrane Reactors Owing to Forced Convective Flow Operation, Hung. J. Ind. Chem.,2005, 32(1-2), 31-42 [10] Koonaphapdeelert S.; Wu Z.; Li K.: Carbon diox- ide stripping in ceramic hollow fibre membrane contactors, Chem. Eng. Sci., 2009 64, 1-8, DOI: 10.1016/j.ces.2008.09.010 [11] Astarita G.; Savage D.W.; Bisio A.: Gas treating with chemical reaction, John Wiley & Sons, New York, 1983 [12] Kreulen H.; Smolders C.A.; Versteeg G.F.; Van Swaaij W.P.M.: Microporous hollow fiber mem- brane module as gas-liquid contactors. Part 2: Mass transfer with chemical reaction, J. Membrane Sci., 1993 78, 217-238, DOI: 10.1016/0376- 7388(93)80002-F [13] Browning G.J.; Weiland R.H.: Physical solubility of carbon dioxide in aqueous alkanolamine via ni- trous oxide analogy, J. Chem. Eng. Data, 1994 39, 817-822, DOI: 10.1021/je00016a040 [14] DeCoursey W.J.: Enhancement factors for gas ab- sorption with reversible reaction, Chem. Eng. Sci., 1982 37, 1483-1489, DOI: 10.1016/0009-2509(82)80005- 5 APPLICATION OF HYDROPHOBIC POLYMERIC MEMBRANE FOR CO2 DESORPTION ... 45(2) pp. 45–49 (2017) 49 [15] Meng-Hui L., Yei-Chung L.: Densities and Viscos- ities of Solutions of Monoethanolamine + N- Methyldiethanolamine + Water and Monoethano- lamine + 2-Amino-2-methyl-1-propanol + Water, J. Chem. Eng. Data, 1994 39, 444-447, DOI: 10.1021/je00015a009