- Email: [email protected]
Adsorption separation of 2-phenylethanol and l -phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria
FBP-567; No. of Pages 10
ARTICLE IN PRESS food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Contents lists available at ScienceDirect
Food and Bioproducts Processing journal homepage: www.elsevier.com/locate/fbp
Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria ˇ Ivan Simko, Emanuel Roriz, Michal Grambliˇcka, Viera Illeová, Milan Polakoviˇc ∗ Department of Chemical and Biochemical Engineering, Institute of Chemical and Environmental Engineering, Faculty of Chemical and Food Technology, Slovak University of Technology, Radlinského 9, 812 37 Bratislava, Slovakia
a r t i c l e
i n f o
a b s t r a c t
Article history:
Eleven polymeric resin adsorbents were investigated for their potential to bind selec-
Received 23 June 2014
tively 2-phenylethanol from solutions containing also l-phenylalanine. The capacity for
Received in revised form 24 October
single-component and the capacity/selectivity for binary adsorption of 2-phenylethanol
2014
and l-phenylalanine were determined to eliminate adsorbents with low performance.
Accepted 27 November 2014
Single-component equilibrium data of both compounds were well described by the
Available online xxx
Langmuir equation for five selected adsorbents. Two most suitable adsorbents, which were both non-functionalised poly(styrene-divinylbenzene) resins with different pore
Keywords:
structures, were then used in measurements of the binary adsorption equilibrium of 2-
2-phenylethanol
phenylethanol and l-phenylalanine. The modified competitive Langmuir equation and the
l-phenylalanine
LeVan–Vermeulen equation provided good description of co-adsorption of 2-phenylethanol
Adsorption capacity
and l-phenylalanine in a broad concentration range of both components when high capac-
Selectivity
ity and selectivity for 2-phenylethanol binding was observed. In order to obtain data for
Binary equilibrium
characterization of desorption performance, adsorption isotherms of 2-phenylethanol were
Separation
also determined for pure ethanol and ethanol/water solutions as solvents. © 2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
1.
Introduction
2-Phenylethanol (2-PE) is an aromatic alcohol with a characteristic rose-like odour and it is one of the basic compounds of rose oil. Its annual production is approximately 10,000 t (Hua and Xu, 2011). Most of this production covers the needs of cosmetic industry; a smaller part is used for chemical synthesis and in food industry as aromas (Etschmann et al., 2002). The major part of 2-PE is still produced by chemical synthesis where harmful and dangerous materials and extreme operational conditions are used (Chaudhari et al., 2000; Etschmann et al., 2002). The main advantage of such production is the low price of about 5 USD/kg of 2-PE (Hua and Xu, 2011). A traditional alternative approach is based on the extraction of 2-PE
from rose petals. Due to a very low concentration of 2-PE and high cost of the raw material, the price of the final product is rather high—about 1000 USD/kg (Hua and Xu, 2011). Since flavouring substances are, in general, applied in very low concentrations, the expensive flavours of natural origin find their place in the food market. A significant group of consumers prefer their use. As far as foodstuffs are concerned, the European legislation allows to use the term natural flavouring substances only for substances obtained by physical, enzymatic or microbiological processes from a material of vegetable or animal origin (Council directive 88/388/EEC). The existence of such legislation and the increasing demand for natural products in the food industry encouraged remarkable efforts towards the development of biotechnological processes
∗
Corresponding author. Tel.: +421 2 59325 254. ˇ E-mail address: [email protected] (M. Polakovic). http://dx.doi.org/10.1016/j.fbp.2014.11.005 0960-3085/© 2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
FBP-567; No. of Pages 10
2
ARTICLE IN PRESS food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
for the production of flavour compounds over the past few years (Longo and Sanroman, 2006). This has also opened an opportunity for the production of 2-PE by a biotechnological process. 2-PE can be produced by biotransformation using yeasts such as Saccharomyces cerevisiae, Pichia anomala, Pichia membranaefaciens, Kluyveromyces marxianus etc. (Etschmann et al., 2003). To increase the production of 2-PE, l-phenylalanine (l-Phe) is added as a precursor to the fermentation broth. l-Phe is transformed to 2-PE via the Ehrlich pathway (Ehrlich, 1907), where the biotransformation of l-Phe into 2-PE is connected with the growth of yeasts on ␣ketoglutarate acid. It means that 2-PE is produced only in the exponential phase of the microbial growth. One of the important problems during the fermentation, which has significant consequences for 2-PE recovery, is the product inhibition. 2-PE is rather toxic to all microorganisms used for its production (Etschmann et al., 2002). Total growth inhibition was observed already at the 2-PE concentration of 2 g l−1 for K. marxianus and 3.8 g l−1 for S. cerevisiae, respectively (Etschmann et al., 2002). The concentration of 3.8 g l−1 was also the highest one achieved for a fed-batch fermentation (Stark, 2001). A conventional 2-PE production process thus includes down-stream processing steps where this flavour compound is recovered from rather dilute biotransformation product mixtures. To overcome the problem of low productivity of the biotransformation step, in situ product recovery techniques have often been investigated in the recent period. The concentration of 2-PE in the fermentation medium was maintained under the inhibition level using hybrid bioreactor/separator systems based on microcapsule extraction (Stark et al., 2003a), membrane extraction (Mihal et al., 2013) or adsorption (Gao and Daugulis, 2009; Mei et al., 2009). The integration of reaction and separation units either in a single equipment unit or in a recycle loop makes the whole process more complex. The implementation of hybrid systems on the production scale thus requires higher research/development costs and stricter process control. These requirements reduce the benefits of a higher volumetric productivity provided by the in situ recovery, for which the conventional biotransformation is still a preferred option. The isolation and purification of both natural and synthetic 2-PE were considered primarily in patent literature. Distillation of 2-PE alcoholic or aqueous/alcoholic solutions is a typical purification step (Hopff et al., 1958; Nienhaus and Hopp, 1990; Savina et al., 1999). Liquid–liquid extraction or adsorption are suitable isolation processes for the recovery of 2-PE from raw materials. For example, the extraction by hexane is the conventional process of obtaining 2-PE from rose petals (Baser and Buchbauer, 2009). Adsorption on polymeric resins was applied for the recovery of 2-PE from alcohol distillation residues (Savina et al., 1999) or from a biotransformation product mixture (Subbiah, 1999). Desorption was done using ethanol or methanol, respectively. The latter patent includes also a comparison of the adsorption recovery process with the liquid–liquid extraction by hexane, ethyl acetate and butanol (Subbiah, 1999). The performance of the adsorption process was better in regard to the 2-PE yield and concentration when using methanol. For 2-PE aqueous solutions, the adsorbent capacity is higher than the extractant capacity (Gao and Daugulis, 2009). On the contrary, the selectivity of 2-PE to l-Phe is higher for the extraction process because l-Phe is insoluble in organic solvents (Mihal et al., 2012a,b). This drawback of lower selectivity in the adsorption phase can be eliminated by a choice of suitable desorbent. l-Phe
can then be completely separated from 2-PE in the desorption phase. Few equilibrium and kinetic characteristics of 2-PE and l-Phe adsorption, which are prerequisites for engineering design of separation equipment, are available in the cited works. Mei et al. determined a single value of adsorption capacity for each tested macroporous resin (Mei et al., 2009). Only one of these resins provided the 2-PE adsorption capacity higher than 100 mg per gram of adsorbent dry mass. This is not a high value for low-molecular compound adsorption even if it is considered that the corresponding liquid-phase equilibrium concentration was relatively low—about 1 g l−1 . Moreover, the adsorption capacity of l-Phe was quite high for this particular adsorbent, about 80 mg g−1 , which indicated not a good selectivity for the separation of 2-PE and l-Phe. Gao and Daugulis used a hydrophobic thermoplastic elastomer Hytrel 8206, which has an infinite value of selectivity because it does not bind l-Phe (Gao and Daugulis, 2009). They achieved a 2-PE partition coefficient of about 80 for the liquidphase equilibrium concentrations of about 1 g l−1 and lower. Drawbacks of this adsorbent appear to be the absence of regular pores at low water content (up to 30%) and large particle size (over 2 mm) which must result in a significant mass transfer resistance. This slow kinetics probably caused that, in a semi-continuous operation investigated by these authors, the total volume of three packed bed adsorbers used was very high; about 50% of the fermentation medium volume. The overall objective of our research activities in this area is to make a systematic design and optimisation of 2-PE separation from post-biotransformation solutions. This particular work was focused on the investigation of equilibrium properties of suitable adsorbent(s). Our search started with a group of commercial polymeric resins from two major world producers of adsorbents. In order to evaluate their capacity for 2-PE and the selectivity of 2-PE and l-Phe, batch adsorption experiments were carried out using single-component and binary aqueous solutions, respectively. A broad range of concentration values and different ratios of 2-PE and l-Phe were used in these experiments to determine single-component and binary adsorption isotherms for selected adsorbents.
2.
Materials and methods
2.1.
Chemicals and resins
2-Phenylethanol and l-phenylalanine were purchased from Merck Schuchardt OHG (Hohenbrunn, Germany). All other chemicals were of analytical grade and they were purchased from local vendors. Milli-Q-grade water was used for the preparation of all solutions. The solutions used for HPLC analysis were filtered through a 0.45 m cellulose nitrate membrane filter. The tested adsorbents were different types of Amberlite (Dow Chemical Company, Michigan, USA) and Macronet resins (Purolite, Philadelphia, USA). The list of the adsorbents together with their properties, such as the surface area and pore size are shown in Table 1. Before use, the adsorbents were conditioned in a 96% ethanol aqueous solution for 24 h. They were then gently stirred and rinsed with redistilled water.
2.2.
Measurement of adsorbent dry mass
Adsorbent slurry was filtered through a 0.45 m cellulose nitrate membrane filter. In this way, the water outside the
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
3
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 1 – Adsorbents used and their properties provided by the producers. Adsorbent Amberlite XAD 4 Amberlite XAD16 N Amberlite XAD16HPN Amberlite XAD 18 Amberlite XAD1180N Amberlite XAD 7HP Amberlite XAD 761 Macronet MN 100 Macronet MN 200 Macronet MN 270 Macronet MN 500 a b c d
Matrix a
SDVB SDVB SDVB SDVB SDVB Acrylic PFAb SDVB SDVB SDVB SDVB
Functional group
Surface area [m2 g−1 ]
Mean pore size [nm]
Porosity
– – – – – Carboxyl Fenol Tertiary amine – – Sulfonic acid
≥750 ≥800 ≥800 ≥800 ≥450 380 200 900 900 1200 900
10 20 20 15 40 45 60 80c /1.4d 80c /1.5d 80c /2.5d 80c /1.4d
≥0.50 ≥0.55 ≥0.60 ≥0.55 ≥0.60 ≥0.50 ≥0.50 0.5 0.5 0.2–0.4 0.5
SDVB – styrene-divinylbenzene. PFA – phenol-formaldehyde. Meso- and macropores. Micropores.
particles was removed by suction. The water remained only in the internal pores of the adsorbent. About 2–3 g of the wet adsorbent were taken and dried at the constant temperature of 60 ◦ C until constant weight was reached.
2.3.
Adsorption/desorption experiments
About 100 mg of the wet adsorbent were weighed into a vial or test-tube and 4 ml or 12 ml, respectively, of aqueous/ethanol solution of 2-PE, l-Phe or their mixture were added. The vials and test-tubes were closed with teflon-sealed cups and the suspensions were stirred at the ambient temperature for 21 h using a reciprocal shaker to achieve adsorption equilibrium. The concentrations of solution components were then measured either spectrophotometrically for one-component adsorption or using HPLC for binary adsorption. Adsorption capacity was determined from a material balance in the following form:
a=
(c0 − ceq )V mDA
(1)
where a is the component adsorbed mass per mass of dry adsorbent; c0 and ceq are its initial and equilibrium concentrations in the solution; V is the solution volume; and mDA is the mass of dry adsorbent. For the investigation of 2-PE desorption, an adsorption experiment was first made using the 2-PE initial concentration of 10 g l−1 . The particles of Amberlite XAD 18 with adsorbed 2-PE were then separated from the aqueous solution using a 0.45 m cellulose–acetate membrane filter (Sartorius, Göttingen, Germany) and added into 100 ml of a desorbent. The conditions of the experiment and analysis were the same as for adsorption. Afterwards, two further desorption steps were made in the same way to increase the recovered amount of 2-PE.
2.4.
Analytical methods
The absorbance of one-component solutions was measured by a through-flow spectrophotometer HP8452A (Hewlett Packard, Palo Alto, CA, USA) at the wavelength of 260 nm using an 80 l flow cuvette. The mean value was obtained from triplicate measurements. The composition of ethanol/water solutions was determined using an Abbé refractometer.
The concentration of binary solutions was determined by HPLC (Agilent 1100, Palo Alto, CA, USA) using a slightly modified version of the method presented in (Stark et al., 2003b). A gradient from 100% of mobile phase A (25 mM HCl) to 100% of mobile phase B (20 vol% of 25 mM HCl and 80 vol% of acetonitrile) was applied for 18 min. The flow rate was 1 ml/min. A reversed-phase column (LiChroCART 250-4 HPLC-Cartridge, Merck, Darmstadt, Germany) was used at 30 ◦ C. A DAD UV detector was used at the fixed wavelength of 254 nm. The injection volume was in the range of 5–100 l.
2.5.
Adsorption isotherms
For the mathematical description of single-component adsorption experiments, the Langmuir equation was used in the form: a = am
bceq 1 + bceq
(2)
where the isotherm parameters are: am , the maximum adsorption capacity, and b, the affinity constant. The parameter values and their errors were estimated using the non-linear regression procedure of process engineering software Athena Visual Studio (Athena Visual Software, Inc., Madison, WI). The parameter values of one-component Langmuir isotherm were also used to predict binary equilibria. Three binary adsorption isotherms (Eqs. (3)–(6)), which are reducible to the Langmuir equation for the zero concentration of one component, were applied to the description of experimental data. The binary equations contain the same symbols as those in the Langmuir equation but the subscript 1 denotes 2phenylethanol and subscript 2 l-phenylalanine. Eqs. (3a) and (3b) represent the simplest binary isotherm used, the competitive Langmuir equation assuming a single type of adsorption sites for which both components compete. a1 =
am,1 b1 ceq,1 1 + b1 ceq,1 + b2 ceq,2
(3a)
a2 =
am,2 b2 ceq,2 1 + b1 ceq,1 + b2 ceq,2
(3b)
The modified competitive Langmuir equation (Eqs. (4a) and (4b)) distinguishes between two types of adsorbent sites. Sites 1 are accessible only to one of the components (2-PE was
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
4
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
chosen in this case.) whereas both components compete in binding to sites 2. a1 =
(am,1 − am,2 )b1 ceq,1 am,2 b1 ceq,1 + 1 + b1 ceq,1 1 + b1 ceq,1 + b2 ceq,2
(4a)
a2 =
am,2 b2 ceq,2 1 + b1 ceq,1 + b2 ceq,2
(4b)
The LeVan–Vermeulen equation is based on the ideal adsorbed solution theory and its definition for the ith component has the following form: ai =
bi a¯ m ceq,i 1 + b1 ceq,1 +b2 ceq,2
+ ceq,i
∂a¯ m ln(1 + b1 ceq,1 + b2 ceq,2 ) ∂ceq,i
(5)
where a¯ m is the weighted average monolayer capacity. Its derivative with respect to the equilibrium solution concentration is typically expressed through the Taylor series expansion. In this work, the third order expansion was used. The following equations were then obtained for the description of adsorption equilibrium: a1 =
a¯ m b1 ceq,1 + 12 (1 + 13 ) 1 + b1 ceq,1 + b2 ceq,2
(6a)
a2 =
a¯ m b2 ceq,2 − 12 (1 + 23 ) 1 + b1 ceq,1 + b2 ceq,2
(6b)
a¯m =
2 am,1 b1 ceq,1 + am,2 b2 ceq,2 b1 ceq,1 b2 ceq,2 (am,1 − am,2 ) +2 2 × b1 ceq,1 + b2 ceq,2 am,1 + am,2 b c 1 eq,1 + b2 ceq,2
12 = (am,1 − am,2 )
13
b1 ceq,1 b2 ceq,2 b1 ceq,1 + b2 ceq,2
am,1 − am,2 1 = × am,1 + am,2 b1 ceq,1 + b2 ceq,2
+
b2 ceq,2
2
3 b2 ceq,2
2
× ln 1 + b1 ceq,1 + b2 ceq,2
2 (6e)
2 (b1 ceq,1 ) + (2b1 ceq,1 ) − 4b2 ceq,2 − b2 ceq,2
b1 ceq,1 + b2 ceq,2
+ 4b2 ceq,2 + b2 ceq,2 b1 ceq,1 − 2b1 ceq,1 − 2(b1 ceq,1 )2 1 + b1 ceq,1 + b2 ceq,2
Results and discussion
Two groups of hydrophobic polymeric adsorbents with separation potential were selected for this investigation (Table 1). All Amberlite resins were adsorbents with a broad size distribution of mesopores whereas the bimodal pore structure of Macronet resins was characteristic by two narrower regions of pore sizes; one formed by micropores and the second one by the pores which size is at the boundary of meso- and macropores. Except for Amberlites XAD 761 and XAD 7HP, all other adsorbents had a poly(styrene-divinylbenzene) matrix. Most adsorbents did not contain functional groups.
(6c)
(6d)
b1 ceq,1 + b2 ceq,2
ln 1 + b1 ceq,1 + b2 ceq,2 − 1
1 + b1 ceq,1 + b2 ceq,2
3.
1 1 + b1 ceq,1 + b2 ceq,2 2
+ 2b2 ceq,2 − (4b1 ceq,1 ) − (b1 ceq,1 )2
3(b1 ceq,1 )2 + 4b1 ceq,1 + b1 ceq,1 b2 ceq,2 − 2b2 ceq,2 − 2 b2 ceq,2
am,2 − am,1 1 = × am,1 + am,2 b1 ceq,1 + b2 ceq,2
+
2 ln 1 + b1 ceq,1 + b2 ceq,2
23
The first series of adsorption experiments were made at a single value of the initial concentration of 2-PE and l-Phe, respectively. Concentration values pertinent to the fermentative production of 2-PE were chosen (Stark et al., 2003b). The results of these experiments are presented in Table 2. Single-component adsorption of 2-PE was carried out for all tested resins. The calculated values of the distribution coefficient, a1 /ceq,1 , show that several tested adsorbents have very good adsorption capacity for 2-PE. Table 2 also shows that the functionalised resins, except for Macronet MN 100 with the tertiary amine group, have much lower adsorption capacity for 2-PE. For Amberlite XAD 761 functionalized with phenolic groups, the low capacity can be explained by its larger pore size and therewith lower specific surface area. The low capacity of Amberlite XAD 7HP and Macronet MN 500 is a consequence of the functionalisation of the base matrix with cationic groups. The mentioned three adsorbents were not used in further experiments. The distribution coefficient, a2 /ceq,2 , was also determined for l-Phe single-component adsorption. Table 2 shows that these values were by about 20–100 lower than the corresponding values for 2-PE. It must however be emphasised that the differences in the adsorbed amounts, a1 and a2 , were not higher than 20%. In some cases, the adsorbed amount of lPhe exceeded that of 2-PE. The differences in the capacity ratios were thus caused by approximately the same adsorbed
2
× ln 1 + b1 ceq,1 + b2 ceq,2
(6f)
amounts obtained at very different equilibrium solution concentrations of about 0.1–0.3 g l−1 for 2-PE but 8–9 g l−1 for l-Phe. The single-component adsorption results thus indicate a potential application of several adsorbents for 2-PE recovery; however, the binary adsorption experiments where 2-PE molecules had to compete for binding sites with much larger molar amounts of l-Phe present in the solution (as is the case in actual bioproduction processes) were the real test of the adsorbents’ suitability. Table 2 presents the results of the distribution coefficients a1 /ceq,1 and a2 /ceq,2 respectively. Due to the competitive effect and larger overall amount used, the distribution coefficients of 2-PE were somewhat lower but still
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
5
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 2 – Results of adsorbent selection by single-concentration experiments. The standard deviations of the partition coefficients were 0.038 l g−1 for 2-PE and 0.001 l g−1 for l-Phe. Adsorbent
Single-component adsorptiona a1 ceq,1
Amberlite XAD 4 Amberlite XAD16 N Amberlite XAD16 HPN Amberlite XAD 18 Amberlite XAD 1180N Amberlite XAD 7HP Amberlite XAD 761 Macronet MN 100 Macronet MN 200 Macronet MN 270 Macronet MN 500 a
l g−1
0.519 0.408 0.469 0.504 0.269 0.093 0.094 1.106 2.326 2.626 0.052
a2 ceq,2
l g−1
Binary adsorptiona a1 ceq,1
l g−1
a2 ceq,2
l g−1
a1 ceq,2 ceq,1 a2
dimensionless
0.016 0.017 0.016 0.016 0.013
0.329 0.375 0.373 0.286 0.194
0.009 0.011 0.010 0.009 0.008
37.1 35.5 36.7 31.2 24.2
0.020 0.021 0.021
0.896 0.846 1.504
0.007 0.009 0.019
121 100 77.2
Initial solution concentration: 2-PE 1 g l−1 , l-Phe 10 g l−1 .
relatively high. The favourable adsorption of 2-PE compared to l-Phe is demonstrated by the high selectivity ratio values presented in Table 2. All tested adsorbents have thus a good potential for 2-PE/l-Phe separation but only five of them (with a/ceq > 0.5), Amberlite XAD 4 and XAD 18, and Macronet MN 100, MN 200, and MN 270, were chosen for the next stage of this investigation. Single-component adsorption isotherms were determined for the selected five adsorbents. The equilibrium data for both 2-PE and l-Phe were fitted with the Langmuir equation (Eq. (2)) using non-linear regression. Fig. 1 illustrates a good match of the model and experimental data for Amberlite XAD18 and Macronet MN 270. The estimated parameters of the Langmuir equation for all five adsorbents are summarised in Table 3. This table also presents the mean errors of the calculated values of adsorption capacity, sa . They confirm a good
Fig. 1 – Comparison of experimental (points) and model (lines) single-component adsorption data; 2-PE on Amberlite XAD 18 (), l-Phe on Amberlite XAD 18 (), 2-PE on Macronet MN 270 () and l-Phe on Macronet MN 270 (䊉). Curves were calculated using the parameter values of the Langmuir equation presented in Table 3.
accuracy of the description of single-component adsorption on all adsorbents by the Langmuir equation. The values of parameters presented in Table 3 provide several interesting observations. The maximum adsorption capacity of l-Phe on the individual adsorbents was quite high, from 0.29 g g−1 to 0.35 g g−1 for four of the adsorbents and even higher, 0.54 g g−1 , for Macronet MN 270, this value is only by about 25% lower than am of 2-PE for this adsorbent. The am values of 2-PE were in the range of 0.45–0.68 g g−1 , which shows that the selectivity of these adsorbents is expected to decrease strongly in the saturation concentration region. The much better selectivity values for the low concentration region of 2-PE presented in Table 2 can be explained by the values of the second parameter of the Langmuir equation, the affinity constant b. Table 3 shows that they were by an order of magnitude higher for 2-PE than for l-Phe. The values of the affinity constants of 2-PE and l-Phe within Amberlite and Macronet series had differences which were not statistically significant. For that reason, the capacity values were decisive for a further reduction of the number of adsorbents tested. Two adsorbents with the highest am value were Amberlite XAD 18 and Macronet MN 270. In spite of their different pore structure, they seem to be equally suitable for the 2-PE and l-Phe separation. They were used in the next series of experiments where binary adsorption isotherms were measured for mixtures with different ratios of 2-PE and l-Phe. The results of the equilibrium experiments, presented in Fig. 2, confirm the preferable adsorption of 2-PE from the binary mixtures in a broad range of concentration ratios of the two studied compounds. This observation favours the potential use of the investigated adsorbents for 2-PE recovery from a fermentation broth. Fig. 2 moreover displays interesting differences in the behaviour of Amberlite XAD 18 and Macronet MN 270, respectively. In case of Amberlite XAD 18, the adsorbed mass of 2-PE was not affected by the content of l-Phe in the binary solutions (Fig. 2a) whereas the adsorbed mass of l-Phe strongly decreased with the 2-PE concentration (Fig. 2b). The adsorbed amounts were always less than 50% of the corresponding values achieved for single-component adsorption. On the other hand, Fig. 2c shows that the adsorbed mass of 2-PE on Macronet MN 270 was affected by the presence of l-Phe. Most values are by about 20% lower than the corresponding ones for single-component adsorption. The values of the adsorbed mass of l-Phe on this adsorbent were affected by the content
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
6
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 3 – Parameters of the Langmuir equation for single-component adsorption of 2-PE and l-Phe: n – number of equilibrium data, sa – mean error of adsorption capacity. Compound
am (g g−1 )
Adsorbent
n
2-PE
Amberlite XAD 18 Amberlite XAD 4 Macronet MN 100 Macronet MN 200 Macronet MN 270
15 14 15 15 15
0.68 0.60 0.45 0.48 0.67
± ± ± ± ±
0.05 0.04 0.02 0.02 0.03
l-Phe
Amberlite XAD 18 Amberlite XAD 4 Macronet MN 100 Macronet MN 200 Macronet MN 270
16 13 13 12 12
0.30 0.29 0.30 0.35 0.54
± ± ± ± ±
0.03 0.05 0.04 0.02 0.05
of 2-PE (Fig. 2d), but the decrease of the adsorption capacity was lower than in the case of the adsorption on Amberlite XAD 18. It is evident that the binary adsorption data were scattered more than single-component data presented in Fig. 1. The reason is that a rule of thumb for a good design of batch equilibrium experiments, to adsorb about 50% of the initial amount of component, could not be kept for both 2-PE and l-Phe, simultaneously. The design of experiments was complicated by the limited solubility of 2-PE and, in the case of Macronet MN 270, also by the high affinity of 2-PE to this adsorbent.
b (l g−1 )
sa × 102 (g g−1 )
0.44 0.78 1.66 4.44 2.26
± ± ± ± ±
0.12 0.28 0.50 1.39 0.47
2.46 3.03 1.97 2.29 2.26
0.11 0.10 0.12 0.14 0.11
± ± ± ± ±
0.03 0.03 0.03 0.02 0.02
0.72 0.83 0.69 0.47 0.61
This often resulted in significant differences between the initial and equilibrium concentrations of 2-PE. Fig. 2c shows that most 2-PE solution equilibrium concentrations were lower than 0.5 g l−1 . The next step in the investigation of the co-adsorption of 2-PE and l-Phe was to examine whether the equilibrium data can be described by suitable binary adsorption isotherms. For this purpose, predictive modelling was applied using three binary adsorption isotherms (Section 2.5). The binary adsorption data were not fitted but they were calculated using parameters adopted from the single-component isotherms
Fig. 2 – Binary adsorption equilibrium data for different concentration ratios: (a) 2-PE on Amberlite XAD 18, (b) l-Phe on Amberlite XAD 18, (c) 2-PE on Macronet MN 270 and (d) l-Phe on Macronet MN 270. The ratios of initial solution concentrations of l-Phe and 2-PE are labelled as follows: 1:1 (), 1:3 () and 3:1 (). The single-component adsorption isotherms from Fig. 1 are plotted (lines) for comparison. The inset graph in Fig. 2c displays the region of low concentrations. ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
FBP-567; No. of Pages 10
ARTICLE IN PRESS 7
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 4 – Comparison of the accuracy of predictive modelling of binary adsorption data of 2-PE and l-Phe on Amberlite XAD 18 and Macronet MN 270. Table values are the standard deviations of experimental and predicted data, sa × 102 in g g−1 . The degrees of freedom were 28 and 30, respectively. Isotherm
Competitive Langmuir Modified competitive Langmuir LeVan-Vermeulen
Amberlite XAD 18
Macronet MN 270
2-PE
l-Phe
2-PE
l-Phe
11.99 3.38 3.69
7.73 2.86 1.50
6.89 7.19 7.16
1.46 1.46 2.07
(Table 3). The match of the experimental and predicted data was quantified by the standard deviations presented in Table 4. The standard deviations show that all three binary isotherms were equally suitable for the description of adsorption data for Macronet MN 270 whereas the adsorption data for Amberlite XAD 18 were described much better by the modified competitive Langmuir equation and LeVan–Vermeulen equations. The ordinary competitive Langmuir equation was not suitable in this case. The cause lies in the significantly different maximum saturation capacity of this adsorbent for 2-PE and l-Phe—0.68 g g−1 vs. 0.30 g g−1 . It is an inherent feature of this equation that the overall adsorbed amount at high concentrations of both components will be a weighted average of their maximum saturation capacities. Other equations assume that a full potential of adsorbent sites can be used in such case and the overall adsorbed amount will be closer to the higher value of am . On the other hand, the competitive Langmuir equation was good for the Macronet MN 270 binary data because the am -values were much closer, 0.67 g g−1 vs. 0.54 g g−1 , and the competition effect of l-Phe was weaker. An interesting result was obtained when the values of the standard deviations of binary data (Table 4) and corresponding single-component data (Table 3) were compared using F-statistics at 95% confidence level. It was found that the variances of the 2-PE adsorption data on Amberlite XAD 18 for the modified competitive Langmuir equation and LeVan–Vermeulen equation were not statistically different from the variance of the single-component data. On the other hand, the variance values of the binary 2-PE adsorption data on Macronet MN 270 were twice higher than those for Amberlite XAD 18. However, no systematic deviations between the calculated and experimental data were observed. The higher values of the variance were thus caused by the higher random errors mentioned above. Tables 3 and 4 show that the standard deviations of l-Phe binary data for both adsorbents were at least twice higher than those of single-component data. A closer inspection of the data showed that the reason of the discrepancy were systematic errors for the experiments with the l-Phe/2-PE initial ratio of 3:1. The calculated values were overestimated by about 0.03–0.05 g g-1 . These systematic errors are not large and would not have a significant impact on the design of adsorption equipment. It can therefore be concluded that two types of binary isotherms, the modified competitive Langmuir equation and the LeVan-Vermeulen equation, describe the co-adsorption of 2-PE and l-Phe on both Amberlite XAD 18 and Macronet MN 270 with acceptable accuracy when their parameters can be effectively adopted from single-component experiments. The application of the binary isotherm equations to the co-adsorption of 2-PE and l-Phe from a fermentation broth
could be complicated by non-specific adsorption of other compounds present in the complex medium. For that reason, verification experiments were made using a supernatant from a fed-batch production of 2-PE by S. cerevisiae (Mihal et al., 2012a,b). 2-PE and l-Phe were added into the supernatant to adjust the initial concentrations to the same values as those in pure binary solutions. Fig. 3a and b show the equilibrium data for the adsorption of 2-PE and l-Phe, respectively, from the fermentation medium on Amberlite XAD 18. Fig. 3 also contains the values presented in Fig. 2 for the same adsorbent. It is evident that the adsorbed amounts of 2-PE and l-Phe were not affected by other components of the fermentation
Fig. 3 – Comparison of 2-PE and l-Phe co-adsorption on Amberlite XAD 18 from pure binary solutions (open symbols) and real fermentation medium (closed symbols). (a) 2-PE, (b) l-Phe. Different ratios of initial solution concentrations of l-Phe and 2-PE are denoted with the same geometrical shapes as in Fig. 2.
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
8
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 5 – Desorption of 2-PE from Amberlite XAD 18. The mean error of this experiment was about 2 percentage points. Desorbent
Methanol Ethanol Isopropanol Butanol
Fraction of 2-PE desorbed in the 1st step (%)
Fraction of 2-PE desorbed after three steps (%)
65.0 69.7 73.1 63.4
84.3 89.1 95.2 81.2
medium. The parameters of the binary competitive isotherms presented in Table 3 were therewith verified for a real process medium and can be used in the design and optimization of an adsorption process. The last part of this work was dedicated to the selection of a suitable desorbent. Four lower aliphatic alcohols were tested in batch desorption experiments. Particles of Amberlite XAD containing about 0.2 g g−1 of adsorbed 2-PE were subjected to three successive desorption steps. Table 5 shows that about 65−70% of adsorbed 2-PE was released in the first step and the total released fraction was from 81% for butanol to 95.2% for isopropanol. Desorption performance of ethanol was only slightly worse than that of isopropanol. Considering ethanol’s lower cost and better acceptance for food applications, it was decided that this solvent will be used as desorbent in future investigations. In order to characterize the equilibrium in desorption conditions, the adsorption isotherms of 2-PE from water/ethanol solutions of different composition were determined. They are displayed in Fig. 4. Since the presence of ethanol increases the solubility of 2-PE significantly, the adsorption isotherms for ethanol solutions were determined in a much wider range of the liquid-phase equilibrium concentrations of 2-PE. For 100% ethanol, the upper boundary of this concentration range was close to 300 g l−1 . It is evident that the presence of ethanol in the solutions significantly decreased the affinity of 2-PE to the adsorbents. On the other hand, the effect of ethanol on the maximum capacity is not so obvious from the plots. An accurate picture of the effect of ethanol on 2-PE capacity and affinity is provided by the values of the parameters of the Langmuir isotherm presented in Table 6. The affinity values for water as a solvent were two or three orders of magnitude higher than those for 100% ethanol. These values demonstrate the effectiveness of the use of ethanol as desorbent. Fig. 4 shows that the 2-PE adsorption isotherms for the solutions with a high content of ethanol deviated only a little from a linear trend. The highest adsorbed amounts were observed for the adsorption from 100% ethanol—approximately 0.6 g g−1 for Macronet MN 270 and 0.8 g g−1 for Amberlite XAD 18. A small deviation from the linear trend in the experimental range means that the corresponding estimated parameters am presented in Table 6 are extrapolated values. A simple calculation can quickly show that they cannot be even approached within the real range of 2-PE concentrations. (The maximum value of liquid-phase concentration is the density of 2-PE which is 1020 g l−1 .) Nonetheless, it is evident that the maximum capacity values for adsorption from ethanol are higher than those for adsorption from water. Fig. 5 presents the equilibrium isotherms for ethanol adsorption from aqueous solutions on Macronet MN 270 and Amberlite XAD 18. It is apparent from the ratio of the
Fig. 4 – Adsorption isotherms of 2-PE from water/ethanol solutions: a) Macronet MN 270, b) Amberlite XAD. The symbols represent the following mass content of ethanol: 100% (), 80% (䊉), 60% (), 40% (), 20% () and 0% (). The curves were calculated using the parameter values of the Langmuir equation presented in Table 6.
Fig. 5 – Adsorption isotherms of ethanol from aqueous solutions on Macronet MN 270 () and Amberlite XAD 18 (䊉). The curves were calculated using the parameter values of the Langmuir equation presented in Table 7.
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
9
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 6 – Parameters of the Langmuir equation for adsorption of 2-PE from water/ethanol solutions where n is the number of equilibrium data. Adsorbent
% Ethanol
n
am (g g−1 )
b (10−3 l g−1 )
Macronet MN 270
100 80 60 40 20 0
8 8 9 8 6 12
1.38 0.60 0.49 0.47 0.56 0.67
± ± ± ± ± ±
0.40 0.05 0.08 0.05 0.08 0.03
2.67 5.95 13.9 27.4 279 2260
± ± ± ± ± ±
1.24 0.86 5.3 4.8 92 470
Amberlite XAD 18
100 80 60 40 20 0
5 8 9 8 6 15
2.41 1.87 0.96 0.73 0.65 0.68
± ± ± ± ± ±
1.21 0.35 0.29 0.04 0.13 0.05
1.78 2.35 5.19 10.6 164 440
± ± ± ± ± ±
1.26 0.57 2.69 1.0 68 120
Table 7 – Parameters of the Langmuir equation for ethanol adsorption from aqueous solutions where n is the number of equilibrium data. Adsorbent
n
am (g g−1 )
b (10−3 l g−1 )
Macronet MN 270 Amberlite XAD 18
15 15
0.87 ± 0.06 1.69 ± 0.13
8.6 ± 1.9 2.3 ± 0.3
solid- and liquid-phase concentrations that ethanol not only fills the pores but also binds to the adsorbent pore surface. This is also confirmed by the maximum capacity parameter values of the Langmuir isotherm presented in Table 7. The affinity constants of ethanol are much lower than the affinity constants of 2-PE for adsorption from water but they are about 2–3 times higher than the affinity constant of 2-PE for adsorption from 100% ethanol. This means that ethanol during the course of desorption primarily weakens the bonds of 2-PE to the adsorbent. When intraparticle water is replaced by ethanol in a large extent, ethanol acts then also as a displacer.
4.
Conclusions
Two types of hydrophobic poly(styrene-divinylbenzene) resins were found to be good adsorbents for the 2-PE/l-Phe separation from biotransformation mixtures. Single-component adsorption isotherms parameters showed that that maximum adsorption capacities of both 2-PE and l-Phe were comparable but the affinity constants for 2-PE were one order of magnitude higher than those for l-Phe. Amberlite XAD 18 and Macronet MN 270 exhibited preferential binding of 2-PE in binary adsorption measurements. The binary adsorption equilibria were successfully predicted by competitive binary isotherms using single-component adsorption isotherm parameters. Ethanol was found to be a suitable desorbent of 2-PE.
Acknowledgements This study was supported by the Slovak Grant Agency for Science VEGA 1/0531/13.
References Baser, K.H.C., Buchbauer, G., 2009. Handbook of Essential Oils: Science, Technology, and Applications. CRC Press, Boca Raton.
Chaudhari, R.V., Telkar, M.M., Rode, C.V., 2000. Process for the preparation of 2-phenyl ethanol. US patent 6166269. Ehrlich, F., 1907. Über die Bedingungen der Fuselölbildung und über ihren Zusammenhang mit dem Eiweißaufbau der Hefe. Ber. Dtsch. Chemi. Ges. 40, 1027–1047, http://dx.doi.org/10.1002/cber.190704001156. Etschmann, M.M.W., Bluemke, W., Sell, D., Schrader, J., 2002. Biotechnological production of 2-phenylethanol. Appl. Microbiol. Biotechnol. 59, 1–8, http://dx.doi.org/10.1007/s00253-002-0992-x. Etschmann, M.M.W., Sell, D., Schrader, J., 2003. Screening of yeasts for the production of the aroma compound 2-phenylethanol in a molasses-based medium. Biotechnol. Lett. 25, 531–536, http://dx.doi.org/10.1023/A:1022890119847. Council Directive 88/388/EEC of 22 June 1988 on the approximation of the laws of the Member States relating to flavourings for use in foodstuffs and to source materials for their production. Gao, F., Daugulis, A.J., 2009. Bioproduction of the aroma compound 2-phenylethanol in a solid-liquid two-phase partitioning bioreactor system by Kluyveromyces marxianus. Biotechnol. Bioeng. 104, 332–339, http://dx.doi.org/10.1002/bit.22387. Hopff, H., Kuhn, H., Hoffmann, U., 1958. Process for the preparation of beta-phenyl ethyl alcohol. US patent 2822403. Hua, D., Xu, P., 2011. Recent advances in biotechnological production of 2-phenylethanol. Biotechnol. Adv. 29, 654–660, http://dx.doi.org/10.1016/j.biotechadv.2011.05. 001. Longo, M.A., Sanroman, M.A., 2006. Production of food aroma compounds: microbial and enzymatic methodologies. Food Technol. Biotechnol. 44, 335–353. Mei, J., Min, H., Lü, Z., 2009. Enhanced biotransformation of l-phenylalanine to 2-phenylethanol using an in situ product adsorption technique. Process Biochem. 44, 886–890, http://dx.doi.org/10.1016/j.procbio.2009.04.012. ˇ Mihal’, M., Vereˇs, R., Markoˇs, J., Stefuca, V., 2012a. Intensification of 2-phenylethanol production in fed-batch hybrid bioreactor: biotransformations and simulations. Chem. Eng. Process 57-58, 75–85, http://dx.doi.org/10.1016/j.cep.2012.03.006. Mihal’, M., Gavin, S.P., Markos, J., 2013. Airlift reactor – membrane extraction hybrid system for aroma production. Chem. Pap. 67, 1485–1494, http://dx.doi.org/10.2478/s11696-012-0261-0. Mihal’, M., Veres, R., Markos, J., 2012b. Investigation of 2-phenylethanol production in fed-batch hybrid bioreactor: membrane extraction and microfiltration. Sep. Purif. Technol. 95, 126–135, http://dx.doi.org/10.1016/j.seppur.2012.04.030. Nienhaus, J., Hopp, R., 1990. Process for the purification of crude beta-phenylethyl alcohol. European patent 0185249. Savina, J., Kohler, D., Brunerie, P., 1999. Method for extracting 2-phenylethanol. US patent 5965780. Stark, D., 2001. Extractive bioconversion of 2-phenylethanol from l-phenylalanine by Saccharomyces cerevisiae. École
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
FBP-567; No. of Pages 10
10
ARTICLE IN PRESS food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Polytechnique Fédérale de Lausanne, Lausanne, Switzerland (PhD thesis). Stark, D., Kornmann, H., Munch, T., Sonnleitner, B., Marison, I.W., von Stockar, U., 2003a. Novel type of in situ extraction: Use of solvent containing microcapsules for the bioconversion of 2-phenylethanol from l-phenylalanine by Saccharomyces cerevisiae. Biotechnol. Bioeng. 83, 376–385, http://dx.doi.org/10.1002/bit.10679.
Stark, D., Zala, D., Münch, T., Sonnleitner, B., Marison, I.W., von Stockar, U., 2003b. Inhibition aspects of the bioconversion of l-phenylalanine to 2-phenylethanol by Saccharomyces cerevisiae. Enzyme Microb. Technol. 32, 212–223, http://dx.doi.org/10.1016/S0141-0229(02)00237-5. Subbiah, V., 1999. Solid phase extraction of phenethyl alcohol. US patent 5919991.
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Contents lists available at ScienceDirect
Food and Bioproducts Processing journal homepage: www.elsevier.com/locate/fbp
Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria ˇ Ivan Simko, Emanuel Roriz, Michal Grambliˇcka, Viera Illeová, Milan Polakoviˇc ∗ Department of Chemical and Biochemical Engineering, Institute of Chemical and Environmental Engineering, Faculty of Chemical and Food Technology, Slovak University of Technology, Radlinského 9, 812 37 Bratislava, Slovakia
a r t i c l e
i n f o
a b s t r a c t
Article history:
Eleven polymeric resin adsorbents were investigated for their potential to bind selec-
Received 23 June 2014
tively 2-phenylethanol from solutions containing also l-phenylalanine. The capacity for
Received in revised form 24 October
single-component and the capacity/selectivity for binary adsorption of 2-phenylethanol
2014
and l-phenylalanine were determined to eliminate adsorbents with low performance.
Accepted 27 November 2014
Single-component equilibrium data of both compounds were well described by the
Available online xxx
Langmuir equation for five selected adsorbents. Two most suitable adsorbents, which were both non-functionalised poly(styrene-divinylbenzene) resins with different pore
Keywords:
structures, were then used in measurements of the binary adsorption equilibrium of 2-
2-phenylethanol
phenylethanol and l-phenylalanine. The modified competitive Langmuir equation and the
l-phenylalanine
LeVan–Vermeulen equation provided good description of co-adsorption of 2-phenylethanol
Adsorption capacity
and l-phenylalanine in a broad concentration range of both components when high capac-
Selectivity
ity and selectivity for 2-phenylethanol binding was observed. In order to obtain data for
Binary equilibrium
characterization of desorption performance, adsorption isotherms of 2-phenylethanol were
Separation
also determined for pure ethanol and ethanol/water solutions as solvents. © 2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
1.
Introduction
2-Phenylethanol (2-PE) is an aromatic alcohol with a characteristic rose-like odour and it is one of the basic compounds of rose oil. Its annual production is approximately 10,000 t (Hua and Xu, 2011). Most of this production covers the needs of cosmetic industry; a smaller part is used for chemical synthesis and in food industry as aromas (Etschmann et al., 2002). The major part of 2-PE is still produced by chemical synthesis where harmful and dangerous materials and extreme operational conditions are used (Chaudhari et al., 2000; Etschmann et al., 2002). The main advantage of such production is the low price of about 5 USD/kg of 2-PE (Hua and Xu, 2011). A traditional alternative approach is based on the extraction of 2-PE
from rose petals. Due to a very low concentration of 2-PE and high cost of the raw material, the price of the final product is rather high—about 1000 USD/kg (Hua and Xu, 2011). Since flavouring substances are, in general, applied in very low concentrations, the expensive flavours of natural origin find their place in the food market. A significant group of consumers prefer their use. As far as foodstuffs are concerned, the European legislation allows to use the term natural flavouring substances only for substances obtained by physical, enzymatic or microbiological processes from a material of vegetable or animal origin (Council directive 88/388/EEC). The existence of such legislation and the increasing demand for natural products in the food industry encouraged remarkable efforts towards the development of biotechnological processes
∗
Corresponding author. Tel.: +421 2 59325 254. ˇ E-mail address: [email protected] (M. Polakovic). http://dx.doi.org/10.1016/j.fbp.2014.11.005 0960-3085/© 2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
FBP-567; No. of Pages 10
2
ARTICLE IN PRESS food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
for the production of flavour compounds over the past few years (Longo and Sanroman, 2006). This has also opened an opportunity for the production of 2-PE by a biotechnological process. 2-PE can be produced by biotransformation using yeasts such as Saccharomyces cerevisiae, Pichia anomala, Pichia membranaefaciens, Kluyveromyces marxianus etc. (Etschmann et al., 2003). To increase the production of 2-PE, l-phenylalanine (l-Phe) is added as a precursor to the fermentation broth. l-Phe is transformed to 2-PE via the Ehrlich pathway (Ehrlich, 1907), where the biotransformation of l-Phe into 2-PE is connected with the growth of yeasts on ␣ketoglutarate acid. It means that 2-PE is produced only in the exponential phase of the microbial growth. One of the important problems during the fermentation, which has significant consequences for 2-PE recovery, is the product inhibition. 2-PE is rather toxic to all microorganisms used for its production (Etschmann et al., 2002). Total growth inhibition was observed already at the 2-PE concentration of 2 g l−1 for K. marxianus and 3.8 g l−1 for S. cerevisiae, respectively (Etschmann et al., 2002). The concentration of 3.8 g l−1 was also the highest one achieved for a fed-batch fermentation (Stark, 2001). A conventional 2-PE production process thus includes down-stream processing steps where this flavour compound is recovered from rather dilute biotransformation product mixtures. To overcome the problem of low productivity of the biotransformation step, in situ product recovery techniques have often been investigated in the recent period. The concentration of 2-PE in the fermentation medium was maintained under the inhibition level using hybrid bioreactor/separator systems based on microcapsule extraction (Stark et al., 2003a), membrane extraction (Mihal et al., 2013) or adsorption (Gao and Daugulis, 2009; Mei et al., 2009). The integration of reaction and separation units either in a single equipment unit or in a recycle loop makes the whole process more complex. The implementation of hybrid systems on the production scale thus requires higher research/development costs and stricter process control. These requirements reduce the benefits of a higher volumetric productivity provided by the in situ recovery, for which the conventional biotransformation is still a preferred option. The isolation and purification of both natural and synthetic 2-PE were considered primarily in patent literature. Distillation of 2-PE alcoholic or aqueous/alcoholic solutions is a typical purification step (Hopff et al., 1958; Nienhaus and Hopp, 1990; Savina et al., 1999). Liquid–liquid extraction or adsorption are suitable isolation processes for the recovery of 2-PE from raw materials. For example, the extraction by hexane is the conventional process of obtaining 2-PE from rose petals (Baser and Buchbauer, 2009). Adsorption on polymeric resins was applied for the recovery of 2-PE from alcohol distillation residues (Savina et al., 1999) or from a biotransformation product mixture (Subbiah, 1999). Desorption was done using ethanol or methanol, respectively. The latter patent includes also a comparison of the adsorption recovery process with the liquid–liquid extraction by hexane, ethyl acetate and butanol (Subbiah, 1999). The performance of the adsorption process was better in regard to the 2-PE yield and concentration when using methanol. For 2-PE aqueous solutions, the adsorbent capacity is higher than the extractant capacity (Gao and Daugulis, 2009). On the contrary, the selectivity of 2-PE to l-Phe is higher for the extraction process because l-Phe is insoluble in organic solvents (Mihal et al., 2012a,b). This drawback of lower selectivity in the adsorption phase can be eliminated by a choice of suitable desorbent. l-Phe
can then be completely separated from 2-PE in the desorption phase. Few equilibrium and kinetic characteristics of 2-PE and l-Phe adsorption, which are prerequisites for engineering design of separation equipment, are available in the cited works. Mei et al. determined a single value of adsorption capacity for each tested macroporous resin (Mei et al., 2009). Only one of these resins provided the 2-PE adsorption capacity higher than 100 mg per gram of adsorbent dry mass. This is not a high value for low-molecular compound adsorption even if it is considered that the corresponding liquid-phase equilibrium concentration was relatively low—about 1 g l−1 . Moreover, the adsorption capacity of l-Phe was quite high for this particular adsorbent, about 80 mg g−1 , which indicated not a good selectivity for the separation of 2-PE and l-Phe. Gao and Daugulis used a hydrophobic thermoplastic elastomer Hytrel 8206, which has an infinite value of selectivity because it does not bind l-Phe (Gao and Daugulis, 2009). They achieved a 2-PE partition coefficient of about 80 for the liquidphase equilibrium concentrations of about 1 g l−1 and lower. Drawbacks of this adsorbent appear to be the absence of regular pores at low water content (up to 30%) and large particle size (over 2 mm) which must result in a significant mass transfer resistance. This slow kinetics probably caused that, in a semi-continuous operation investigated by these authors, the total volume of three packed bed adsorbers used was very high; about 50% of the fermentation medium volume. The overall objective of our research activities in this area is to make a systematic design and optimisation of 2-PE separation from post-biotransformation solutions. This particular work was focused on the investigation of equilibrium properties of suitable adsorbent(s). Our search started with a group of commercial polymeric resins from two major world producers of adsorbents. In order to evaluate their capacity for 2-PE and the selectivity of 2-PE and l-Phe, batch adsorption experiments were carried out using single-component and binary aqueous solutions, respectively. A broad range of concentration values and different ratios of 2-PE and l-Phe were used in these experiments to determine single-component and binary adsorption isotherms for selected adsorbents.
2.
Materials and methods
2.1.
Chemicals and resins
2-Phenylethanol and l-phenylalanine were purchased from Merck Schuchardt OHG (Hohenbrunn, Germany). All other chemicals were of analytical grade and they were purchased from local vendors. Milli-Q-grade water was used for the preparation of all solutions. The solutions used for HPLC analysis were filtered through a 0.45 m cellulose nitrate membrane filter. The tested adsorbents were different types of Amberlite (Dow Chemical Company, Michigan, USA) and Macronet resins (Purolite, Philadelphia, USA). The list of the adsorbents together with their properties, such as the surface area and pore size are shown in Table 1. Before use, the adsorbents were conditioned in a 96% ethanol aqueous solution for 24 h. They were then gently stirred and rinsed with redistilled water.
2.2.
Measurement of adsorbent dry mass
Adsorbent slurry was filtered through a 0.45 m cellulose nitrate membrane filter. In this way, the water outside the
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
3
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 1 – Adsorbents used and their properties provided by the producers. Adsorbent Amberlite XAD 4 Amberlite XAD16 N Amberlite XAD16HPN Amberlite XAD 18 Amberlite XAD1180N Amberlite XAD 7HP Amberlite XAD 761 Macronet MN 100 Macronet MN 200 Macronet MN 270 Macronet MN 500 a b c d
Matrix a
SDVB SDVB SDVB SDVB SDVB Acrylic PFAb SDVB SDVB SDVB SDVB
Functional group
Surface area [m2 g−1 ]
Mean pore size [nm]
Porosity
– – – – – Carboxyl Fenol Tertiary amine – – Sulfonic acid
≥750 ≥800 ≥800 ≥800 ≥450 380 200 900 900 1200 900
10 20 20 15 40 45 60 80c /1.4d 80c /1.5d 80c /2.5d 80c /1.4d
≥0.50 ≥0.55 ≥0.60 ≥0.55 ≥0.60 ≥0.50 ≥0.50 0.5 0.5 0.2–0.4 0.5
SDVB – styrene-divinylbenzene. PFA – phenol-formaldehyde. Meso- and macropores. Micropores.
particles was removed by suction. The water remained only in the internal pores of the adsorbent. About 2–3 g of the wet adsorbent were taken and dried at the constant temperature of 60 ◦ C until constant weight was reached.
2.3.
Adsorption/desorption experiments
About 100 mg of the wet adsorbent were weighed into a vial or test-tube and 4 ml or 12 ml, respectively, of aqueous/ethanol solution of 2-PE, l-Phe or their mixture were added. The vials and test-tubes were closed with teflon-sealed cups and the suspensions were stirred at the ambient temperature for 21 h using a reciprocal shaker to achieve adsorption equilibrium. The concentrations of solution components were then measured either spectrophotometrically for one-component adsorption or using HPLC for binary adsorption. Adsorption capacity was determined from a material balance in the following form:
a=
(c0 − ceq )V mDA
(1)
where a is the component adsorbed mass per mass of dry adsorbent; c0 and ceq are its initial and equilibrium concentrations in the solution; V is the solution volume; and mDA is the mass of dry adsorbent. For the investigation of 2-PE desorption, an adsorption experiment was first made using the 2-PE initial concentration of 10 g l−1 . The particles of Amberlite XAD 18 with adsorbed 2-PE were then separated from the aqueous solution using a 0.45 m cellulose–acetate membrane filter (Sartorius, Göttingen, Germany) and added into 100 ml of a desorbent. The conditions of the experiment and analysis were the same as for adsorption. Afterwards, two further desorption steps were made in the same way to increase the recovered amount of 2-PE.
2.4.
Analytical methods
The absorbance of one-component solutions was measured by a through-flow spectrophotometer HP8452A (Hewlett Packard, Palo Alto, CA, USA) at the wavelength of 260 nm using an 80 l flow cuvette. The mean value was obtained from triplicate measurements. The composition of ethanol/water solutions was determined using an Abbé refractometer.
The concentration of binary solutions was determined by HPLC (Agilent 1100, Palo Alto, CA, USA) using a slightly modified version of the method presented in (Stark et al., 2003b). A gradient from 100% of mobile phase A (25 mM HCl) to 100% of mobile phase B (20 vol% of 25 mM HCl and 80 vol% of acetonitrile) was applied for 18 min. The flow rate was 1 ml/min. A reversed-phase column (LiChroCART 250-4 HPLC-Cartridge, Merck, Darmstadt, Germany) was used at 30 ◦ C. A DAD UV detector was used at the fixed wavelength of 254 nm. The injection volume was in the range of 5–100 l.
2.5.
Adsorption isotherms
For the mathematical description of single-component adsorption experiments, the Langmuir equation was used in the form: a = am
bceq 1 + bceq
(2)
where the isotherm parameters are: am , the maximum adsorption capacity, and b, the affinity constant. The parameter values and their errors were estimated using the non-linear regression procedure of process engineering software Athena Visual Studio (Athena Visual Software, Inc., Madison, WI). The parameter values of one-component Langmuir isotherm were also used to predict binary equilibria. Three binary adsorption isotherms (Eqs. (3)–(6)), which are reducible to the Langmuir equation for the zero concentration of one component, were applied to the description of experimental data. The binary equations contain the same symbols as those in the Langmuir equation but the subscript 1 denotes 2phenylethanol and subscript 2 l-phenylalanine. Eqs. (3a) and (3b) represent the simplest binary isotherm used, the competitive Langmuir equation assuming a single type of adsorption sites for which both components compete. a1 =
am,1 b1 ceq,1 1 + b1 ceq,1 + b2 ceq,2
(3a)
a2 =
am,2 b2 ceq,2 1 + b1 ceq,1 + b2 ceq,2
(3b)
The modified competitive Langmuir equation (Eqs. (4a) and (4b)) distinguishes between two types of adsorbent sites. Sites 1 are accessible only to one of the components (2-PE was
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
4
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
chosen in this case.) whereas both components compete in binding to sites 2. a1 =
(am,1 − am,2 )b1 ceq,1 am,2 b1 ceq,1 + 1 + b1 ceq,1 1 + b1 ceq,1 + b2 ceq,2
(4a)
a2 =
am,2 b2 ceq,2 1 + b1 ceq,1 + b2 ceq,2
(4b)
The LeVan–Vermeulen equation is based on the ideal adsorbed solution theory and its definition for the ith component has the following form: ai =
bi a¯ m ceq,i 1 + b1 ceq,1 +b2 ceq,2
+ ceq,i
∂a¯ m ln(1 + b1 ceq,1 + b2 ceq,2 ) ∂ceq,i
(5)
where a¯ m is the weighted average monolayer capacity. Its derivative with respect to the equilibrium solution concentration is typically expressed through the Taylor series expansion. In this work, the third order expansion was used. The following equations were then obtained for the description of adsorption equilibrium: a1 =
a¯ m b1 ceq,1 + 12 (1 + 13 ) 1 + b1 ceq,1 + b2 ceq,2
(6a)
a2 =
a¯ m b2 ceq,2 − 12 (1 + 23 ) 1 + b1 ceq,1 + b2 ceq,2
(6b)
a¯m =
2 am,1 b1 ceq,1 + am,2 b2 ceq,2 b1 ceq,1 b2 ceq,2 (am,1 − am,2 ) +2 2 × b1 ceq,1 + b2 ceq,2 am,1 + am,2 b c 1 eq,1 + b2 ceq,2
12 = (am,1 − am,2 )
13
b1 ceq,1 b2 ceq,2 b1 ceq,1 + b2 ceq,2
am,1 − am,2 1 = × am,1 + am,2 b1 ceq,1 + b2 ceq,2
+
b2 ceq,2
2
3 b2 ceq,2
2
× ln 1 + b1 ceq,1 + b2 ceq,2
2 (6e)
2 (b1 ceq,1 ) + (2b1 ceq,1 ) − 4b2 ceq,2 − b2 ceq,2
b1 ceq,1 + b2 ceq,2
+ 4b2 ceq,2 + b2 ceq,2 b1 ceq,1 − 2b1 ceq,1 − 2(b1 ceq,1 )2 1 + b1 ceq,1 + b2 ceq,2
Results and discussion
Two groups of hydrophobic polymeric adsorbents with separation potential were selected for this investigation (Table 1). All Amberlite resins were adsorbents with a broad size distribution of mesopores whereas the bimodal pore structure of Macronet resins was characteristic by two narrower regions of pore sizes; one formed by micropores and the second one by the pores which size is at the boundary of meso- and macropores. Except for Amberlites XAD 761 and XAD 7HP, all other adsorbents had a poly(styrene-divinylbenzene) matrix. Most adsorbents did not contain functional groups.
(6c)
(6d)
b1 ceq,1 + b2 ceq,2
ln 1 + b1 ceq,1 + b2 ceq,2 − 1
1 + b1 ceq,1 + b2 ceq,2
3.
1 1 + b1 ceq,1 + b2 ceq,2 2
+ 2b2 ceq,2 − (4b1 ceq,1 ) − (b1 ceq,1 )2
3(b1 ceq,1 )2 + 4b1 ceq,1 + b1 ceq,1 b2 ceq,2 − 2b2 ceq,2 − 2 b2 ceq,2
am,2 − am,1 1 = × am,1 + am,2 b1 ceq,1 + b2 ceq,2
+
2 ln 1 + b1 ceq,1 + b2 ceq,2
23
The first series of adsorption experiments were made at a single value of the initial concentration of 2-PE and l-Phe, respectively. Concentration values pertinent to the fermentative production of 2-PE were chosen (Stark et al., 2003b). The results of these experiments are presented in Table 2. Single-component adsorption of 2-PE was carried out for all tested resins. The calculated values of the distribution coefficient, a1 /ceq,1 , show that several tested adsorbents have very good adsorption capacity for 2-PE. Table 2 also shows that the functionalised resins, except for Macronet MN 100 with the tertiary amine group, have much lower adsorption capacity for 2-PE. For Amberlite XAD 761 functionalized with phenolic groups, the low capacity can be explained by its larger pore size and therewith lower specific surface area. The low capacity of Amberlite XAD 7HP and Macronet MN 500 is a consequence of the functionalisation of the base matrix with cationic groups. The mentioned three adsorbents were not used in further experiments. The distribution coefficient, a2 /ceq,2 , was also determined for l-Phe single-component adsorption. Table 2 shows that these values were by about 20–100 lower than the corresponding values for 2-PE. It must however be emphasised that the differences in the adsorbed amounts, a1 and a2 , were not higher than 20%. In some cases, the adsorbed amount of lPhe exceeded that of 2-PE. The differences in the capacity ratios were thus caused by approximately the same adsorbed
2
× ln 1 + b1 ceq,1 + b2 ceq,2
(6f)
amounts obtained at very different equilibrium solution concentrations of about 0.1–0.3 g l−1 for 2-PE but 8–9 g l−1 for l-Phe. The single-component adsorption results thus indicate a potential application of several adsorbents for 2-PE recovery; however, the binary adsorption experiments where 2-PE molecules had to compete for binding sites with much larger molar amounts of l-Phe present in the solution (as is the case in actual bioproduction processes) were the real test of the adsorbents’ suitability. Table 2 presents the results of the distribution coefficients a1 /ceq,1 and a2 /ceq,2 respectively. Due to the competitive effect and larger overall amount used, the distribution coefficients of 2-PE were somewhat lower but still
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
5
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 2 – Results of adsorbent selection by single-concentration experiments. The standard deviations of the partition coefficients were 0.038 l g−1 for 2-PE and 0.001 l g−1 for l-Phe. Adsorbent
Single-component adsorptiona a1 ceq,1
Amberlite XAD 4 Amberlite XAD16 N Amberlite XAD16 HPN Amberlite XAD 18 Amberlite XAD 1180N Amberlite XAD 7HP Amberlite XAD 761 Macronet MN 100 Macronet MN 200 Macronet MN 270 Macronet MN 500 a
l g−1
0.519 0.408 0.469 0.504 0.269 0.093 0.094 1.106 2.326 2.626 0.052
a2 ceq,2
l g−1
Binary adsorptiona a1 ceq,1
l g−1
a2 ceq,2
l g−1
a1 ceq,2 ceq,1 a2
dimensionless
0.016 0.017 0.016 0.016 0.013
0.329 0.375 0.373 0.286 0.194
0.009 0.011 0.010 0.009 0.008
37.1 35.5 36.7 31.2 24.2
0.020 0.021 0.021
0.896 0.846 1.504
0.007 0.009 0.019
121 100 77.2
Initial solution concentration: 2-PE 1 g l−1 , l-Phe 10 g l−1 .
relatively high. The favourable adsorption of 2-PE compared to l-Phe is demonstrated by the high selectivity ratio values presented in Table 2. All tested adsorbents have thus a good potential for 2-PE/l-Phe separation but only five of them (with a/ceq > 0.5), Amberlite XAD 4 and XAD 18, and Macronet MN 100, MN 200, and MN 270, were chosen for the next stage of this investigation. Single-component adsorption isotherms were determined for the selected five adsorbents. The equilibrium data for both 2-PE and l-Phe were fitted with the Langmuir equation (Eq. (2)) using non-linear regression. Fig. 1 illustrates a good match of the model and experimental data for Amberlite XAD18 and Macronet MN 270. The estimated parameters of the Langmuir equation for all five adsorbents are summarised in Table 3. This table also presents the mean errors of the calculated values of adsorption capacity, sa . They confirm a good
Fig. 1 – Comparison of experimental (points) and model (lines) single-component adsorption data; 2-PE on Amberlite XAD 18 (), l-Phe on Amberlite XAD 18 (), 2-PE on Macronet MN 270 () and l-Phe on Macronet MN 270 (䊉). Curves were calculated using the parameter values of the Langmuir equation presented in Table 3.
accuracy of the description of single-component adsorption on all adsorbents by the Langmuir equation. The values of parameters presented in Table 3 provide several interesting observations. The maximum adsorption capacity of l-Phe on the individual adsorbents was quite high, from 0.29 g g−1 to 0.35 g g−1 for four of the adsorbents and even higher, 0.54 g g−1 , for Macronet MN 270, this value is only by about 25% lower than am of 2-PE for this adsorbent. The am values of 2-PE were in the range of 0.45–0.68 g g−1 , which shows that the selectivity of these adsorbents is expected to decrease strongly in the saturation concentration region. The much better selectivity values for the low concentration region of 2-PE presented in Table 2 can be explained by the values of the second parameter of the Langmuir equation, the affinity constant b. Table 3 shows that they were by an order of magnitude higher for 2-PE than for l-Phe. The values of the affinity constants of 2-PE and l-Phe within Amberlite and Macronet series had differences which were not statistically significant. For that reason, the capacity values were decisive for a further reduction of the number of adsorbents tested. Two adsorbents with the highest am value were Amberlite XAD 18 and Macronet MN 270. In spite of their different pore structure, they seem to be equally suitable for the 2-PE and l-Phe separation. They were used in the next series of experiments where binary adsorption isotherms were measured for mixtures with different ratios of 2-PE and l-Phe. The results of the equilibrium experiments, presented in Fig. 2, confirm the preferable adsorption of 2-PE from the binary mixtures in a broad range of concentration ratios of the two studied compounds. This observation favours the potential use of the investigated adsorbents for 2-PE recovery from a fermentation broth. Fig. 2 moreover displays interesting differences in the behaviour of Amberlite XAD 18 and Macronet MN 270, respectively. In case of Amberlite XAD 18, the adsorbed mass of 2-PE was not affected by the content of l-Phe in the binary solutions (Fig. 2a) whereas the adsorbed mass of l-Phe strongly decreased with the 2-PE concentration (Fig. 2b). The adsorbed amounts were always less than 50% of the corresponding values achieved for single-component adsorption. On the other hand, Fig. 2c shows that the adsorbed mass of 2-PE on Macronet MN 270 was affected by the presence of l-Phe. Most values are by about 20% lower than the corresponding ones for single-component adsorption. The values of the adsorbed mass of l-Phe on this adsorbent were affected by the content
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
6
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 3 – Parameters of the Langmuir equation for single-component adsorption of 2-PE and l-Phe: n – number of equilibrium data, sa – mean error of adsorption capacity. Compound
am (g g−1 )
Adsorbent
n
2-PE
Amberlite XAD 18 Amberlite XAD 4 Macronet MN 100 Macronet MN 200 Macronet MN 270
15 14 15 15 15
0.68 0.60 0.45 0.48 0.67
± ± ± ± ±
0.05 0.04 0.02 0.02 0.03
l-Phe
Amberlite XAD 18 Amberlite XAD 4 Macronet MN 100 Macronet MN 200 Macronet MN 270
16 13 13 12 12
0.30 0.29 0.30 0.35 0.54
± ± ± ± ±
0.03 0.05 0.04 0.02 0.05
of 2-PE (Fig. 2d), but the decrease of the adsorption capacity was lower than in the case of the adsorption on Amberlite XAD 18. It is evident that the binary adsorption data were scattered more than single-component data presented in Fig. 1. The reason is that a rule of thumb for a good design of batch equilibrium experiments, to adsorb about 50% of the initial amount of component, could not be kept for both 2-PE and l-Phe, simultaneously. The design of experiments was complicated by the limited solubility of 2-PE and, in the case of Macronet MN 270, also by the high affinity of 2-PE to this adsorbent.
b (l g−1 )
sa × 102 (g g−1 )
0.44 0.78 1.66 4.44 2.26
± ± ± ± ±
0.12 0.28 0.50 1.39 0.47
2.46 3.03 1.97 2.29 2.26
0.11 0.10 0.12 0.14 0.11
± ± ± ± ±
0.03 0.03 0.03 0.02 0.02
0.72 0.83 0.69 0.47 0.61
This often resulted in significant differences between the initial and equilibrium concentrations of 2-PE. Fig. 2c shows that most 2-PE solution equilibrium concentrations were lower than 0.5 g l−1 . The next step in the investigation of the co-adsorption of 2-PE and l-Phe was to examine whether the equilibrium data can be described by suitable binary adsorption isotherms. For this purpose, predictive modelling was applied using three binary adsorption isotherms (Section 2.5). The binary adsorption data were not fitted but they were calculated using parameters adopted from the single-component isotherms
Fig. 2 – Binary adsorption equilibrium data for different concentration ratios: (a) 2-PE on Amberlite XAD 18, (b) l-Phe on Amberlite XAD 18, (c) 2-PE on Macronet MN 270 and (d) l-Phe on Macronet MN 270. The ratios of initial solution concentrations of l-Phe and 2-PE are labelled as follows: 1:1 (), 1:3 () and 3:1 (). The single-component adsorption isotherms from Fig. 1 are plotted (lines) for comparison. The inset graph in Fig. 2c displays the region of low concentrations. ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
FBP-567; No. of Pages 10
ARTICLE IN PRESS 7
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 4 – Comparison of the accuracy of predictive modelling of binary adsorption data of 2-PE and l-Phe on Amberlite XAD 18 and Macronet MN 270. Table values are the standard deviations of experimental and predicted data, sa × 102 in g g−1 . The degrees of freedom were 28 and 30, respectively. Isotherm
Competitive Langmuir Modified competitive Langmuir LeVan-Vermeulen
Amberlite XAD 18
Macronet MN 270
2-PE
l-Phe
2-PE
l-Phe
11.99 3.38 3.69
7.73 2.86 1.50
6.89 7.19 7.16
1.46 1.46 2.07
(Table 3). The match of the experimental and predicted data was quantified by the standard deviations presented in Table 4. The standard deviations show that all three binary isotherms were equally suitable for the description of adsorption data for Macronet MN 270 whereas the adsorption data for Amberlite XAD 18 were described much better by the modified competitive Langmuir equation and LeVan–Vermeulen equations. The ordinary competitive Langmuir equation was not suitable in this case. The cause lies in the significantly different maximum saturation capacity of this adsorbent for 2-PE and l-Phe—0.68 g g−1 vs. 0.30 g g−1 . It is an inherent feature of this equation that the overall adsorbed amount at high concentrations of both components will be a weighted average of their maximum saturation capacities. Other equations assume that a full potential of adsorbent sites can be used in such case and the overall adsorbed amount will be closer to the higher value of am . On the other hand, the competitive Langmuir equation was good for the Macronet MN 270 binary data because the am -values were much closer, 0.67 g g−1 vs. 0.54 g g−1 , and the competition effect of l-Phe was weaker. An interesting result was obtained when the values of the standard deviations of binary data (Table 4) and corresponding single-component data (Table 3) were compared using F-statistics at 95% confidence level. It was found that the variances of the 2-PE adsorption data on Amberlite XAD 18 for the modified competitive Langmuir equation and LeVan–Vermeulen equation were not statistically different from the variance of the single-component data. On the other hand, the variance values of the binary 2-PE adsorption data on Macronet MN 270 were twice higher than those for Amberlite XAD 18. However, no systematic deviations between the calculated and experimental data were observed. The higher values of the variance were thus caused by the higher random errors mentioned above. Tables 3 and 4 show that the standard deviations of l-Phe binary data for both adsorbents were at least twice higher than those of single-component data. A closer inspection of the data showed that the reason of the discrepancy were systematic errors for the experiments with the l-Phe/2-PE initial ratio of 3:1. The calculated values were overestimated by about 0.03–0.05 g g-1 . These systematic errors are not large and would not have a significant impact on the design of adsorption equipment. It can therefore be concluded that two types of binary isotherms, the modified competitive Langmuir equation and the LeVan-Vermeulen equation, describe the co-adsorption of 2-PE and l-Phe on both Amberlite XAD 18 and Macronet MN 270 with acceptable accuracy when their parameters can be effectively adopted from single-component experiments. The application of the binary isotherm equations to the co-adsorption of 2-PE and l-Phe from a fermentation broth
could be complicated by non-specific adsorption of other compounds present in the complex medium. For that reason, verification experiments were made using a supernatant from a fed-batch production of 2-PE by S. cerevisiae (Mihal et al., 2012a,b). 2-PE and l-Phe were added into the supernatant to adjust the initial concentrations to the same values as those in pure binary solutions. Fig. 3a and b show the equilibrium data for the adsorption of 2-PE and l-Phe, respectively, from the fermentation medium on Amberlite XAD 18. Fig. 3 also contains the values presented in Fig. 2 for the same adsorbent. It is evident that the adsorbed amounts of 2-PE and l-Phe were not affected by other components of the fermentation
Fig. 3 – Comparison of 2-PE and l-Phe co-adsorption on Amberlite XAD 18 from pure binary solutions (open symbols) and real fermentation medium (closed symbols). (a) 2-PE, (b) l-Phe. Different ratios of initial solution concentrations of l-Phe and 2-PE are denoted with the same geometrical shapes as in Fig. 2.
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
8
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 5 – Desorption of 2-PE from Amberlite XAD 18. The mean error of this experiment was about 2 percentage points. Desorbent
Methanol Ethanol Isopropanol Butanol
Fraction of 2-PE desorbed in the 1st step (%)
Fraction of 2-PE desorbed after three steps (%)
65.0 69.7 73.1 63.4
84.3 89.1 95.2 81.2
medium. The parameters of the binary competitive isotherms presented in Table 3 were therewith verified for a real process medium and can be used in the design and optimization of an adsorption process. The last part of this work was dedicated to the selection of a suitable desorbent. Four lower aliphatic alcohols were tested in batch desorption experiments. Particles of Amberlite XAD containing about 0.2 g g−1 of adsorbed 2-PE were subjected to three successive desorption steps. Table 5 shows that about 65−70% of adsorbed 2-PE was released in the first step and the total released fraction was from 81% for butanol to 95.2% for isopropanol. Desorption performance of ethanol was only slightly worse than that of isopropanol. Considering ethanol’s lower cost and better acceptance for food applications, it was decided that this solvent will be used as desorbent in future investigations. In order to characterize the equilibrium in desorption conditions, the adsorption isotherms of 2-PE from water/ethanol solutions of different composition were determined. They are displayed in Fig. 4. Since the presence of ethanol increases the solubility of 2-PE significantly, the adsorption isotherms for ethanol solutions were determined in a much wider range of the liquid-phase equilibrium concentrations of 2-PE. For 100% ethanol, the upper boundary of this concentration range was close to 300 g l−1 . It is evident that the presence of ethanol in the solutions significantly decreased the affinity of 2-PE to the adsorbents. On the other hand, the effect of ethanol on the maximum capacity is not so obvious from the plots. An accurate picture of the effect of ethanol on 2-PE capacity and affinity is provided by the values of the parameters of the Langmuir isotherm presented in Table 6. The affinity values for water as a solvent were two or three orders of magnitude higher than those for 100% ethanol. These values demonstrate the effectiveness of the use of ethanol as desorbent. Fig. 4 shows that the 2-PE adsorption isotherms for the solutions with a high content of ethanol deviated only a little from a linear trend. The highest adsorbed amounts were observed for the adsorption from 100% ethanol—approximately 0.6 g g−1 for Macronet MN 270 and 0.8 g g−1 for Amberlite XAD 18. A small deviation from the linear trend in the experimental range means that the corresponding estimated parameters am presented in Table 6 are extrapolated values. A simple calculation can quickly show that they cannot be even approached within the real range of 2-PE concentrations. (The maximum value of liquid-phase concentration is the density of 2-PE which is 1020 g l−1 .) Nonetheless, it is evident that the maximum capacity values for adsorption from ethanol are higher than those for adsorption from water. Fig. 5 presents the equilibrium isotherms for ethanol adsorption from aqueous solutions on Macronet MN 270 and Amberlite XAD 18. It is apparent from the ratio of the
Fig. 4 – Adsorption isotherms of 2-PE from water/ethanol solutions: a) Macronet MN 270, b) Amberlite XAD. The symbols represent the following mass content of ethanol: 100% (), 80% (䊉), 60% (), 40% (), 20% () and 0% (). The curves were calculated using the parameter values of the Langmuir equation presented in Table 6.
Fig. 5 – Adsorption isotherms of ethanol from aqueous solutions on Macronet MN 270 () and Amberlite XAD 18 (䊉). The curves were calculated using the parameter values of the Langmuir equation presented in Table 7.
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
ARTICLE IN PRESS
FBP-567; No. of Pages 10
9
food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Table 6 – Parameters of the Langmuir equation for adsorption of 2-PE from water/ethanol solutions where n is the number of equilibrium data. Adsorbent
% Ethanol
n
am (g g−1 )
b (10−3 l g−1 )
Macronet MN 270
100 80 60 40 20 0
8 8 9 8 6 12
1.38 0.60 0.49 0.47 0.56 0.67
± ± ± ± ± ±
0.40 0.05 0.08 0.05 0.08 0.03
2.67 5.95 13.9 27.4 279 2260
± ± ± ± ± ±
1.24 0.86 5.3 4.8 92 470
Amberlite XAD 18
100 80 60 40 20 0
5 8 9 8 6 15
2.41 1.87 0.96 0.73 0.65 0.68
± ± ± ± ± ±
1.21 0.35 0.29 0.04 0.13 0.05
1.78 2.35 5.19 10.6 164 440
± ± ± ± ± ±
1.26 0.57 2.69 1.0 68 120
Table 7 – Parameters of the Langmuir equation for ethanol adsorption from aqueous solutions where n is the number of equilibrium data. Adsorbent
n
am (g g−1 )
b (10−3 l g−1 )
Macronet MN 270 Amberlite XAD 18
15 15
0.87 ± 0.06 1.69 ± 0.13
8.6 ± 1.9 2.3 ± 0.3
solid- and liquid-phase concentrations that ethanol not only fills the pores but also binds to the adsorbent pore surface. This is also confirmed by the maximum capacity parameter values of the Langmuir isotherm presented in Table 7. The affinity constants of ethanol are much lower than the affinity constants of 2-PE for adsorption from water but they are about 2–3 times higher than the affinity constant of 2-PE for adsorption from 100% ethanol. This means that ethanol during the course of desorption primarily weakens the bonds of 2-PE to the adsorbent. When intraparticle water is replaced by ethanol in a large extent, ethanol acts then also as a displacer.
4.
Conclusions
Two types of hydrophobic poly(styrene-divinylbenzene) resins were found to be good adsorbents for the 2-PE/l-Phe separation from biotransformation mixtures. Single-component adsorption isotherms parameters showed that that maximum adsorption capacities of both 2-PE and l-Phe were comparable but the affinity constants for 2-PE were one order of magnitude higher than those for l-Phe. Amberlite XAD 18 and Macronet MN 270 exhibited preferential binding of 2-PE in binary adsorption measurements. The binary adsorption equilibria were successfully predicted by competitive binary isotherms using single-component adsorption isotherm parameters. Ethanol was found to be a suitable desorbent of 2-PE.
Acknowledgements This study was supported by the Slovak Grant Agency for Science VEGA 1/0531/13.
References Baser, K.H.C., Buchbauer, G., 2009. Handbook of Essential Oils: Science, Technology, and Applications. CRC Press, Boca Raton.
Chaudhari, R.V., Telkar, M.M., Rode, C.V., 2000. Process for the preparation of 2-phenyl ethanol. US patent 6166269. Ehrlich, F., 1907. Über die Bedingungen der Fuselölbildung und über ihren Zusammenhang mit dem Eiweißaufbau der Hefe. Ber. Dtsch. Chemi. Ges. 40, 1027–1047, http://dx.doi.org/10.1002/cber.190704001156. Etschmann, M.M.W., Bluemke, W., Sell, D., Schrader, J., 2002. Biotechnological production of 2-phenylethanol. Appl. Microbiol. Biotechnol. 59, 1–8, http://dx.doi.org/10.1007/s00253-002-0992-x. Etschmann, M.M.W., Sell, D., Schrader, J., 2003. Screening of yeasts for the production of the aroma compound 2-phenylethanol in a molasses-based medium. Biotechnol. Lett. 25, 531–536, http://dx.doi.org/10.1023/A:1022890119847. Council Directive 88/388/EEC of 22 June 1988 on the approximation of the laws of the Member States relating to flavourings for use in foodstuffs and to source materials for their production. Gao, F., Daugulis, A.J., 2009. Bioproduction of the aroma compound 2-phenylethanol in a solid-liquid two-phase partitioning bioreactor system by Kluyveromyces marxianus. Biotechnol. Bioeng. 104, 332–339, http://dx.doi.org/10.1002/bit.22387. Hopff, H., Kuhn, H., Hoffmann, U., 1958. Process for the preparation of beta-phenyl ethyl alcohol. US patent 2822403. Hua, D., Xu, P., 2011. Recent advances in biotechnological production of 2-phenylethanol. Biotechnol. Adv. 29, 654–660, http://dx.doi.org/10.1016/j.biotechadv.2011.05. 001. Longo, M.A., Sanroman, M.A., 2006. Production of food aroma compounds: microbial and enzymatic methodologies. Food Technol. Biotechnol. 44, 335–353. Mei, J., Min, H., Lü, Z., 2009. Enhanced biotransformation of l-phenylalanine to 2-phenylethanol using an in situ product adsorption technique. Process Biochem. 44, 886–890, http://dx.doi.org/10.1016/j.procbio.2009.04.012. ˇ Mihal’, M., Vereˇs, R., Markoˇs, J., Stefuca, V., 2012a. Intensification of 2-phenylethanol production in fed-batch hybrid bioreactor: biotransformations and simulations. Chem. Eng. Process 57-58, 75–85, http://dx.doi.org/10.1016/j.cep.2012.03.006. Mihal’, M., Gavin, S.P., Markos, J., 2013. Airlift reactor – membrane extraction hybrid system for aroma production. Chem. Pap. 67, 1485–1494, http://dx.doi.org/10.2478/s11696-012-0261-0. Mihal’, M., Veres, R., Markos, J., 2012b. Investigation of 2-phenylethanol production in fed-batch hybrid bioreactor: membrane extraction and microfiltration. Sep. Purif. Technol. 95, 126–135, http://dx.doi.org/10.1016/j.seppur.2012.04.030. Nienhaus, J., Hopp, R., 1990. Process for the purification of crude beta-phenylethyl alcohol. European patent 0185249. Savina, J., Kohler, D., Brunerie, P., 1999. Method for extracting 2-phenylethanol. US patent 5965780. Stark, D., 2001. Extractive bioconversion of 2-phenylethanol from l-phenylalanine by Saccharomyces cerevisiae. École
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005
FBP-567; No. of Pages 10
10
ARTICLE IN PRESS food and bioproducts processing x x x ( 2 0 1 4 ) xxx–xxx
Polytechnique Fédérale de Lausanne, Lausanne, Switzerland (PhD thesis). Stark, D., Kornmann, H., Munch, T., Sonnleitner, B., Marison, I.W., von Stockar, U., 2003a. Novel type of in situ extraction: Use of solvent containing microcapsules for the bioconversion of 2-phenylethanol from l-phenylalanine by Saccharomyces cerevisiae. Biotechnol. Bioeng. 83, 376–385, http://dx.doi.org/10.1002/bit.10679.
Stark, D., Zala, D., Münch, T., Sonnleitner, B., Marison, I.W., von Stockar, U., 2003b. Inhibition aspects of the bioconversion of l-phenylalanine to 2-phenylethanol by Saccharomyces cerevisiae. Enzyme Microb. Technol. 32, 212–223, http://dx.doi.org/10.1016/S0141-0229(02)00237-5. Subbiah, V., 1999. Solid phase extraction of phenethyl alcohol. US patent 5919991.
ˇ Please cite this article in press as: Simko, I., et al., Adsorption separation of 2-phenylethanol and l-phenylalanine on polymeric resins: Adsorbent screening, single-component and binary equilibria. Food Bioprod Process (2014), http://dx.doi.org/10.1016/j.fbp.2014.11.005