Biocatalytic separation of (R, S)-1-phenylethanol enantiomers and fractionation of reaction products with supercritical carbon dioxide

Page 1

Bo

ASa

b

a

ARR2A

KSLE1P

1

fid

stfss

e[ctrp

tom

b

0d

J. of Supercritical Fluids 55 (2011) 963–970

Contents lists available at ScienceDirect

The Journal of Supercritical Fluids

journa l homepage: www.e lsev ier .com/ locate /supf lu

iocatalytic separation of (R, S)-1-phenylethanol enantiomers and fractionationf reaction products with supercritical carbon dioxide

lexandre Paivaa, Pedro Vidinhab, Maria Angelovaa, Sílvia Rebochob,usana Barreirosb,∗, Gerd Brunnera,∗

Technische Universität Hamburg-Harburg, Thermische Verfahrenstechnik, Eissendorfer Strasse 38, D-21073 Hamburg, GermanyREQUIMTE/CQFB, Departamento de Química, Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa, 2829-516 Caparica, Portugal

r t i c l e i n f o

rticle history:eceived 16 March 2010eceived in revised form7 September 2010

a b s t r a c t

The continuous kinetic resolution of (R, S)-1-phenylethanol via enzymatic transesterification was chosenas a benchmark reaction for the study of a complete reaction/separation process using supercriticalcarbon dioxide. Phase equilibrium data for binary and ternary systems was acquired as a starting point todetermine the best conditions of operation. Total conversion of the (R)-isomer of the alcohol was achieved

ccepted 28 September 2010

eywords:upercritical CO2

ipasenantioselectivity

using a 10% molar excess of vinyl laurate. Three separators operating at given temperatures and pressuresallowed the recovery of (S)-1-phenylethanol with a purity of 86%. The phase equilibrium data obtainedindicates that an additional separation step should allow the recovery of both (S)-1-phenylethanol and(R)-1-phenylethylaurate with over 95% purity. Due to the low solubilities of the target compounds, CO2

from the outlet gas stream of the last separator must be recycled to ensure the technical viability of theation

-Phenylethanol

hase equilibriumintegrated reaction/separ

. Introduction

Supercritical fluid technology is now well established in manyelds, including extraction [1], chemical processes [2], particleesign [3] and biocatalysis [4,5].

The most common and most successful industrial applications ofupercritical carbon dioxide (sc-CO2) are extraction processes. Dueo its low viscosity, sc-CO2 can generally penetrate a solid matrixaster than a liquid solvent, and rapidly extract and transport dis-olved solutes from it. A great advantage is the absence of toxicolvent residues in the processed compounds [6].

One of the best-known examples of application of sc-CO2 inxtraction processes is the decaffeination of green coffee beans7,8] and tea leaves [9], which are now established industrial pro-esses. Since then, many other examples have appeared, includinghe extraction of high value pharmaceutical precursors [10], envi-onmental pollutants [11], flavors, spices, and essential oils fromlant materials [12].

Other important commercial technologies are also emerginghat involve sc-CO2, such as polymerization [13] and nan-technology [14,15]. Supercritical technology is also useful toicroencapsulate materials that are difficult to treat with exist-

∗ Corresponding authors.E-mail addresses: [email protected] (S. Barreiros), [email protected],

[email protected] (G. Brunner).

896-8446/$ – see front matter © 2010 Elsevier B.V. All rights reserved.oi:10.1016/j.supflu.2010.09.037

of (R, S)-1-phenylethanol.© 2010 Elsevier B.V. All rights reserved.

ing techniques, while controlling particle size and morphology byadjusting nucleation and growth during the production of the par-ticles [16].

The use of sc-CO2 in reactions can facilitate downstream sepa-ration. Through the manipulation of temperature and pressure, thesame solvent can be used both as reaction medium and vehicle forthe continuous extraction and separation of reaction products. Thisapplies to biocatalysis as well. Enzymes are natural catalysts thatcan be easily decomposed in natural environments after use. Theuse of sc-CO2 in biocatalysis can be an additional factor contributingto a greener and more sustainable process.

Life depends on molecular chirality. The structural differencebetween enantiomers can be crucial with respect to the actionsof synthetic drugs. Chiral receptor sites in the human body interactonly with drug molecules having the proper absolute configuration,resulting in marked differences in the pharmacological activities ofenantiomers [17].

Enzymes are highly selective catalysts for the generation of pureenantiomers via the kinetic resolution of racemates. Interestingwork in the field of biocatalysis and enantiomeric separation insupercritical media has been carried out by Reetz et al. [18–20],Lozano et al. [21–23] and Knez et al. [5,24]. Reetz et al. looked

not only at the reaction itself, but also at the subsequent fraction-ation of the reaction mixture. In the case of (R, S)-1-phenylethanol(PhEtOH), the authors used a biphasic system composed of an ionicliquid, where the enzymatic reaction was carried out, and sc-CO2,which acted as a substrate carrier and selective extraction media

Page 2

9 ritical

faba

tNtttuvpaeoaie

2

2

Flr(la4gp

2

ivuwm

64 A. Paiva et al. / J. of Superc

or the reaction products [18]. These authors selected an acylatinggent leading to an ester product formed by the (R)-enantiomerearing a long alkyl chain and thus easier to separate from unre-cted ((S)-enantiomer) alcohol.

Reetz et al. [18] went farther in the study of the reac-ion/separation process than is usually found in the literature.onetheless, they did not carry out a complete study for

he integrated process, and this provided the motivation forhe present work. A substantiated characterization of a reac-ion/separation process using sc-CO2 as the solvent requires thetilization/acquisition of vapor–liquid equilibrium data for rele-ant combinations of the intervening species. The goal here is torovide such a detailed characterization, namely mass balances forll the operation units and fluid streams, allowing for the technical,conomical and environmental evaluation of the complete processf separation of the two enantiomers of PhEtOH. Unlike Reetz etl., we look at a fully continuous process, using only sc-CO2 (noonic liquid) for carrying the substrates to and through a fixed-bednzymatic reactor and for performing all the separations that ensue.

. Experimental

.1. Materials

(R, S)-1-Phenylethanol (PhEtOH; 98% purity) was supplied byluka, vinyl acetate (99% purity) by Merck, vinyl decanoate, vinylaurate (VL), vinyl stearate, methyl decanoate and methyl lau-ate (purity ranging from 96% and 99.9%) by Sigma-Aldrich. CO299.95% purity) was supplied by KWD Kohlensäurewerk Deutsch-and GmbH (Bad Hönningen) for the experiments done in Hamburg,nd by Air Liquide for the experiments done in Caparica. Novozym35 (immobilized Candida antarctica lipase B, CALB) was a kindift from Novozymes. (R, S)-1-phenyl-ethyl laurate (PhEtLau; >98%urity) was synthesized by Solchemar on demand.

.2. Enzymatic reaction experiments

The reaction mechanism is shown in Fig. 1.Enzymatic reactions were carried out in Caparica. Batch exper-

ments were done using an apparatus comprising a high pressureariable volume visual cell [25]. Reactions were also done contin-ously, using the apparatus configuration shown in grey in Fig. 2,ith the difference that the homogenization of the liquid substrateixture and CO2 was not done with a static mixer, as in Ham-

OH

OH

R

O

OCH2

O

R

R

O

OCH3

+

O

R

+

(R,S)-1-phenylethanol

Methyl ester

Vinyl ester

(R)-1-phe

Fig. 1. Reaction scheme using as acylating agent a vinyl ester (upper path) o

Fluids 55 (2011) 963–970

burg; also CO2 was admitted into the system with an HPLC pump.In both cases, a 1/4′ OD packed-bed enzymatic reactor was used.All reactions were carried out at conditions under which the fluidmixture was above its critical point and thus monophasic. Unlessstated otherwise, an alcohol/ester molar ratio = 1:1 was used, i.e. anester/(R)-isomer molar ratio = 2.

2.3. Phase equilibrium experiments

Phase equilibrium measurements were carried out in Hamburg,using a variable volume visual cell [26].

The first step is to heat the viewing cell to the desired temper-ature, by heating the water bath where it is immersed. Once tem-perature is stable, vacuum is made inside the cell and a previouslyprepared liquid mixture is loaded into it. CO2 is then compressedand inserted into the cell at the desired pressure. Before each sam-ple is taken, the system inside the cell is stirred for 1 h, after whichit is left to rest for 30 min. Vacuum is then made to the samplingline, and a sample from either the gas phase or the liquid phase istaken and collected in a cold trap. The solutes previously dissolvedin the CO2 remain in the trap. The amount of CO2 in the sampleis then measured in a flow meter. During sampling, the pressureinside the cell is kept constant by moving a piston and decreasingits internal volume, using water as the back-pressure fluid. In everyexperiment, the liquid and gas phases are sampled eight times each.Analysis of the results obtained show that stability is achieved afterthe second sampling round, which means that each experimentalcomposition reported is the average of six measurements.

2.4. Continuous enzymatic reaction and fractionation of reactionproducts

By combining phase equilibrium and reaction data, a continu-ous process for the kinetic resolution of PhEtOH and downstreamseparation can be designed. An apparatus was built in Hamburgfor that purpose, as shown in Fig. 2. This apparatus comprises twomain sections: a reaction section with a similar design to the onein Caparica, and a separation section. In the first separator (S01)kept at specific temperature and pressure conditions, the mixture

exiting the reactor is decompressed into a liquid phase collected atthe bottom, and a gas phase. The latter enters the second separa-tor (cyclone; C01), operated at a lower pressure than the first one,again to yield a liquid phase and a gaseous one. This process is againrepeated with a third separator (C02).

O

O

CH2

HO

H3C OH

+

H3C

O

+

nyl-ethyl ester

Vinyl alcohol

Methanol

Acetaldehyde

r a methyl ester (lower path). R = CH3(CH2)8, CH3(CH2)10, CH3(CH2)16.

Page 3

A. Paiva et al. / J. of Supercritical Fluids 55 (2011) 963–970 965

Fig. 2. Schematic diagram of the continuous reaction/fractionation apparatus. The section in grey was based on an existing reaction apparatus. V, valve; BPR, back pressurer romata

2

ll(n12

rye(6(O11i(smwvath(w

3.1.1. Batch reaction experimentsThe results obtained are shown in Fig. 3. The use of vinyl esters

is very common and many authors have reported on the impact of

egulator. P, liquid pump. S, gravimmetric separator. C, cyclone separator. 6-Way cht several points.

.5. Analysis

Samples obtained in phase equilibrium studies were ana-yzed with an HP5890 GC equipped with a J&W Scientific 30 mength × 0.25 mm id × 0.1 �m film column. The retention time forR, S)-1-phenylethanol was 5.4 min and for hexadecane (exter-al standard) was 9.2 min. Oven temperature program: 90 ◦C for0 min; 90–130 ◦C ramp at 1 ◦C min−1, injection temperature:50 ◦C. Flame ionization detection (FID) temperature: 250 ◦C.

Both the reaction conversion and the enantiomeric excess of theemaining ester product substrate (eep) were measured by GC anal-sis performed with a Trace 2000 Series Unicam gas chromatographquipped with a 30 m × 0.32 mm id fused silica capillary columnBGB-178) coated with a 0.25 mm thickness film of 20% 2,3-diethyl--tert-butyldimethylsilyl-beta-cyclodextrin dissolved in BGB-1515% phenyl-, 85% methylpolysiloxane), from BGB Analytik AG.ven temperature program: 90 ◦C for 10 min; 90–105 ◦C ramp at◦C min−1, 105–175 ◦C ramp at 10 ◦C min−1, 175–200 ◦C ramp at5 ◦C min−1, 200 ◦C for 2 min, injection temperature: 250 ◦C. Flame

onization detection (FID) temperature: 250 ◦C. Carrier gas: helium7.1 ml min−1). Split ratio: 1:20. Tridecane was used as internaltandard. The results reported are the average of least two replicateeasurements. No products were detected in assays carried outithout enzyme. In some reactions, the presence of acid formed

ia hydrolysis of the ester substrate was detected. This can be

voided by controlling the water introduced into the system byhe substrates, the enzyme and CO2. CALB works well at very lowydration. By ensuring that the major component CO2 is almost drywater activity below 0.1), good enzyme performance is achievedithout formation of by-products.

ographic valves with calibrated loops allowing sampling from liquid and gas phases

3. Results and discussion

3.1. Enzymatic reaction experiments

Fig. 3. Time course of (R)-PhEtOH consumption using different esters as acy-lating agents: (�) vinyl decanoate; (�) vinyl laurate; (�) vinyl stearate;(♦) methyl decanoate; (©) methyl laurate. T = 323 K, P = 15 MPa [immobilizedenzyme] = 0.75 mg mL−1, [alcohol] = [ester] = 50 mM.

Page 4

9 ritical Fluids 55 (2011) 963–970

vtawcppthamtsttrspp

3

s2flfi

bfi2uto

3

3

tpa

TVp

66 A. Paiva et al. / J. of Superc

inyl ester alkyl chain length on the catalytic efficiency and selec-ivity of CALB, a bell-shaped curve being obtained [27,28]. Highestctivity is usually reported for a number of carbon atoms around 4,hereas highest selectivity is usually obtained for shorter or longer

arbon chains. Reactions using methyl esters as acylating agentsroceeded at lower rates and reaction conversion was not com-lete, leveling out at around 40%. In the case of vinyl esters, one ofhe products is vinyl alcohol, which tautomerizes to form acetalde-yde. This does not happen with methanol that eventually becomesn obstacle to further conversion. The removal of methanol wouldake the process more complex and thus more expensive, and

his line of work was not followed. If the choice between vinyltearate and VL were based on Fig. 3 alone, the latter would behe best option. Not only are reaction rates slightly higher, but alsohe longer chain length of the ester product should make its sepa-ation from unreacted alcohol easier. However, vinyl stearate is aolid at room temperature. The fact that VL is a liquid facilitates itsrocessing with CO2 and avoids problems related with substraterecipitation.

.1.2. Continuous reaction experimentsThe continuous transesterification of PhEtOH with VL was

tudied in a 17-cm long reactor with an internal volume of.7 mL, packed with 300 mg of Novoym. Several flow rates of theuid (CO2 + substrates) mixture were tested, keeping the volume

raction of the substrates constant, to determine the optimum res-dence time, as shown in Fig. 4.

The conversion of the (R)-isomer is complete for flow rateselow 1 mL min−1, but starts to decline for higher flow rates,rst slightly and then more drastically for flow rates above.5 mL min−1, reaching 70% at 5.0 mL min−1. As shown in the fig-re, this is directly related with the decrease in residence time inhe reactor. The reaction reaches 95% conversion at a residence timef 1.1 min, and is complete for a residence time of less than 3 min.

.2. VLE of the system PhEtOH/PhEtLau/CO2

.2.1. Phase equilibrium experimentsThe phase equilibrium experiments were focused on the

ernary systems PhEtOH/PhEtLau/CO2. Isothermal vapor–liquidhase equilibrium was measured at temperatures between 313nd 333 K, in the pressure range 10–14 MPa, for (solvent-free) feed

able 1apor–liquid equilibrium data for the ternary systems PhEtOH (1)/PhEtLau (2)/CO2 (3). Foint is the average of four values. The average absolute deviation is 3.7%.

T (K) X ′feed

PhEtOH P (MPa) x1 x2

323 0.00 10 0.0000 0.2511 0.0000 0.2213 0.0000 0.19

0.25 10 0.1359 0.1611 0.1098 0.1413 0.0674 0.11

0.50 10 0.2529 0.1111 0.2385 0.1013 0.1933 0.0914 0.1380 0.06

0.75 10 0.3660 0.0611 0.3455 0.0513 0.3070 0.05

1.00 10 0.3658 0.0011 0.5277 0.0013 0.4824 0.00

313 0.50 13 0.1420 0.06323 0.50 13 0.1933 0.09328 0.50 13 0.2127 0.11333 0.50 13 0.2357 0.11

* Values below 0.0001.

Fig. 4. (R)-PhEtOH conversion as a function of the flow rate (lower X-axis, brokenline) and residence time (upper X-axis, full line). T = 323 K, P = 15 MPa, immobi-lized enzyme = 300 mg, [alcohol] = [ester] = 50 mM, corresponding to a fraction ofsubstrate mixture = 6.2% (v/v).

compositions of 0.25, 0.50, and 0.75 PhEtOH mass fraction. Data forthe binaries PhEtOH/CO2 and PhEtLau/CO2 was also collected. Theexperimentally measured liquid and vapor phase mole fractions (xand y, respectively) are given in Table 1.

Table 1 shows that the solubility of the alcohol–ester mixturein CO2 at 323 K increases with pressure, and so does the solubilityof CO2 in the alcohol–ester mixture. An increase in PhEtOH massfraction in the feed causes an expansion of the two-phase region,which is considerably more pronounced in the liquid side. The abil-ity of CO2 to separate the two compounds can be quantified usingthe separation factor, ˛, which can be calculated with data fromTable 1:

˛ = yPhEtOH/xPhEtOH

y /x(1)

PhEtLau PhEtLau

In Fig. 5, ˛ is plotted at 323 K, as a function of pressure, forhigher PhEtOH feed mass fractions. On the insert, ˛ is plotted at13 MPa, as a function of temperature, for a 0.50 feed mass fractioncomposition.

eed mass fractions of PhEtOH are on a solvent-free basis (X ′feed

PhEtOH). Each data

x3 y1 y2 y3

78 0.7422 0.0000 0.0001 0.999929 0.7771 0.0000 0.0020 0.998091 0.8009 0.0000 0.0036 0.996489 0.6951 0.0000* 0.0002 0.999829 0.7473 0.0022 0.0013 0.996509 0.8216 0.0044 0.0030 0.992635 0.6336 0.0007 0.0001 0.999271 0.6544 0.0028 0.0005 0.996717 0.7150 0.0087 0.0035 0.987847 0.7973 0.0152 0.0071 0.977741 0.5699 0.0003 0.0000* 0.999757 0.5988 0.0013 0.0000* 0.998654 0.6375 0.0136 0.0013 0.985100 0.6342 0.0006 0.0000 0.999400 0.4723 0.0055 0.0000 0.994500 0.5176 0.0159 0.0000 0.984115 0.7966 0.0174 0.0063 0.976317 0.7150 0.0087 0.0035 0.987813 0.6760 0.0063 0.0015 0.992218 0.6525 0.0016 0.0001 0.9982

Page 5

A. Paiva et al. / J. of Supercritical Fluids 55 (2011) 963–970 967

Ff

Pvsafi[lvcc(

attinefa

aa

Fa

ig. 5. Separation factor as a function of pressure at 323 K (main figure), and as aunction of temperature at 13 MPa (insert).

Fig. 5 shows that lower pressure, higher temperature and higherhEtOH mass fraction in the alcohol–ester mixture result in high ˛alues and more efficient separation of the two compounds. At con-tant temperature, the CO2 density and solvation ability are lowert lower pressure, and the discrimination ability of CO2 increases,avoring the more soluble compound, which is PhEtOH. The sames true at constant pressure, for higher temperatures. Reetz et al.18] worked with the PhEtOH/PhEtLau system, and reached simi-ar conclusions. Higher temperatures also bring about increases inapor pressure. The starting point of Reetz and co-workers was tohoose an ester with a carbon chain sufficiently long to make theorrespondingly (unfavorable) low volatility predominate over thefavorable) low polarity.

A separation process cannot be designed on the basis of ˛ valueslone. Although low pressures and high temperatures are requiredo achieve an efficient separation of the alcohol and the ester, athose conditions the solubility of the alcohol–ester mixture in CO2s very low (Table 1). This means that large amounts of CO2 areeeded to recover a pure component, which can make the processconomically unviable. A compromise between high separationactors and high solubility needs to be found. This can be better

nalyzed with Figs. 6 and 7.

Fig. 6 shows that high separation factors, which can be achievedt lower pressures, come at the cost of too low a solubility of thelcohol–ester mixture. For instance, at 10 MPa the separation fac-

ig. 6. Separation factor (left Y-axis) and solubility of the 0.5 PhEtOH mass fractionlcohol–ester liquid mixture in CO2 (right Y-axis), as a function of pressure, at 323 K.

Fig. 7. Separation factor (left Y-axis) and solubility of the 0.5 PhEtOH mass fractionalcohol–ester liquid mixture in CO2 (right Y-axis), as a function of temperature, at13 MPa.

tor is close to 4, but the mole fraction of the alcohol–ester mixturein CO2 is ca. 0.001. On the other hand, at 14 MPa the mole fractionof the alcohol–ester mixture in CO2 is about 20 times higher, butthe separation factor at these conditions is close to 1, which meansthat CO2 is not able to separate the two compounds. Intermedi-ate pressures provide an acceptable compromise between the twoparameters.

Fig. 7 shows that temperature can help improve the conditionsfor separation. This figure and the previous one allow the selectionof the boundary conditions for pressure and temperature to be usedin the separation. For example, for the separation of a 0.5 PhEtOHmass fraction alcohol–ester mixture, the separation process shouldbe operated at a temperature above 323 K and at pressures above11 MPa. Some caution needs to be taken with the upper limits ofoperation. At pressures higher than 14 MPa, a much higher tem-perature is necessary to achieve a successful separation and to stillhave enough liquid dissolved in the gas phase to make the processviable. This high temperature brings additional energy costs to theprocess.

3.2.2. ModelingThe pTxy experimental phase equilibrium data was fitted by the

Peng-Robinson equation of state (PR-EOS) [29] with the Mathias-Klotz-Prausnitz mixing rule (MKP-MR) [30], using the programpackage PE2000 [31]. Several other equations of state and mixingrules were tested, but the PR-EOS with the MKP-MR offered thebest results.

Pure component critical properties and acentric factors, ω(Table 2) were used to determine the parameters ai and bi in thecorrelation of the experimental data. Data for CO2 was readily avail-able [32], but in the case of PhEtLau and PhEtOH, critical propertieshad to be estimated using the simulation software ASPEN PLUS®.

The fitting of the PR-EOS/MKP-MR to the ternary system wasmade by finding the best set of interaction parameters that min-imized the deviations between the calculated and experimentallydetermined liquid and vapor phase compositions, as well as thedistribution factors of all the components of the mixture. The

Table 2Pure component critical parameters.

Compound MW (g/mol) Tb (K) Tc (K) Pc (MPa) ω

PhEtOH 122 477.2 668.0 3.99 0.706PhEtLau 304 741.5 931.9 12.81 0.858CO2 44 195.0 304.1 7.38 0.225

Page 6

968 A. Paiva et al. / J. of Supercritical Fluids 55 (2011) 963–970

Table 3Optimized interaction parameters for the ternary system PhEtOH (1)/PhEtLau(2)/CO2(3) at 323 K, using the PR-EOS/MKP-MR model.

1–2 1–3 2–3

kij 0.1348 −0.1591 −0.3483

otEPa

A

tilcMa

fmpaaaefiatla

FaMe

�ij 0.0033 −0.0440 −0.1765lij −0.1445 −0.0020 0.6518AAD < 2%

bjective function used to calculate the absolute average devia-ion (AAD) between the experimental and the correlated data wasq. (2). The optimum set of values obtained for the ternary systemhEtOH/PhEtLau/CO2 at 323 K and the corresponding AAD valuesre given in Table 3.

AD =

√√√√1n

n∑i=1

(zexpi

− zEOSi

)2

(2)

Experimental and correlated phase boundary lines for theernary system at 323 K are given in Fig. 8, which shows a compar-son between the calculated and the experimental tie-lines for theiquid and gas phases. Although there are slight deviations on thealculated tie-lines, especially for the gas phase, the PR-EOS/MKP-R model is able to correlate accurately the VLE data with a total

bsolute average deviation of less than 2%.Fig. 9 shows a comparison between the experimental separation

actors at 323 K and the values predicted with the PR-EOS/MKP-MRodel. Very good predictions were achieved at 13 MPa but as the

ressure decreases, AAD values increase. Overall the model yieldedrelative AAD of 29%. This is due to the comparatively higher devi-tions obtained for the vapor phase. The mass fractions of PhEtOHnd PhEtLau in the vapor phase are very low. Therefore, the slight-st deviation in the calculation of the PhEtOH and PhEtLau massractions in this phase has a high impact in the overall AAD, whichn turn impacts on the AAD obtained in the calculation of the sep-

ration factor. The highest deviations are obtained at 10 MPa. Athis pressure, the solubilities of the liquid mixtures in CO2 are veryow. To make the separation process possible, a pressure rangebove 10 MPa has to be chosen, at which the model can predict the

ig. 8. pTxy data for the ternary system PhEtOH/PhEtLau/CO2 at 323 K. The pointsre experimental data, and the lines were obtained by fitting with the PR-EOS/MKP-R model, using the program package PE2000. Data for the CO2-rich mixture is

xpanded in the right-hand figure.

Fig. 9. Comparison between calculated separation factors derived from experimen-tal data at 323 K and the values predicted with the PR-EOS/MKP-MR model, as afunction of X ′

feedPhEtOH: (�) 10 MPa; (�) 11 MPa; (�) 13 MPa.

separation factors between PhEtOH and PhEtLau with satisfactoryaccuracy.

3.3. Continuous kinetic resolution and separation of PhEtOH

The continuous process for the separation of PhEtOH enan-tiomers was carried out via an enzymatic transesterificationreaction in scCO2, followed by product separation and purifica-tion. As seen earlier, the enzyme is extremely selective for the(R)-isomer. Vinyl alcohol is highly volatile and remains solubilizedin CO2, from which it can be easily separated at a later step. Exper-iments were carried out at an alcohol/ester molar ratio = 2:1, i.e. anester/(R)-isomer molar ratio = 1. It was observed that over 92% ofthe (R)-isomer was consumed. Therefore, 100% conversion of the(R)-isomer requires an excess of VL, which complicates the post-reaction separation of the (S)-isomer of the alcohol from the esterproduct. To simulate this type of situation, experiments were per-formed with a 10% molar excess of VL in the feed, which was found

to be sufficient to convert all of the (R)-isomer of the alcohol. Theseparators were operated at pressures between 10 and 13 MPa, andtemperatures in the range 313–333 K. Fig. 10 shows some of thebest results obtained, at the conditions given in Table 4.

Fig. 10. Solvent free basis % molar compositions obtained in the outlet stream fromthe packed-bed reactor, as well as in each phase of the separators. PhEtOH at bottom,VL in the middle, PhEtLau at top. In this case, PhEtOH and PhEtLau are the (S)- andthe (R)-isomers, respectively.

Page 7

A. Paiva et al. / J. of Supercritical Fluids 55 (2011) 963–970 969

Table 4Operation conditions for Fig. 10.

Reaction Separator 1 Separator 2 Separator 3

Pdsfcf8

3heo0otbs

hofitv

Table 5Solubility of pure components in sc-CO2 as a function of pressure, at 323 K.

Pressure (MPa) VL PhEtOH PhEtLau

10.0 – 0.0006* 0.0001*

10.1 0.0029 – –10.3 0.0039 – –

2

Fa

Temperature (K) 318.15 325.65 325.65 325.65Pressure (MPa) 16.0 13.3 11.0 10.1

At the first separator it was possible to remove 71 mol% of (R)-hEtLau from the liquid mixture. The separation becomes moreifficult as the mixture is enriched in (S)-PhEtOH, and in the secondeparator it was only possible to remove 47 mol% of (R)-PhEtLaurom the gas stream. At the end of the third separation step itan be observed that practically all (R)-PhEtLau has been removedrom the mixture and (S)-PhEtOH can be recovered with a purity of6.3 mol%.

Due to the excess VL still present in the gas stream of separator(ca. 11% molar), a fourth separation step will be required to obtainigher purity alcohol. To this end, additional phase equilibriumxperiments for the ternary system PhEtOH/VL/CO2 were carriedut at 323 K and 7.5 MPa, with solvent free feed mole fractions of.89 for PhEtOH and 0.11 for VL. Solvent free molar compositionsf 96% and 4% were obtained for PhEtOH and VL, respectively. Athese experimental conditions, the solubility of PhEtLau in CO2 wille extremely low and its presence in the gas stream of a fourtheparator will be only residual.

Our separation strategy is aimed at recovering PhEtLau withigh purity (>95 mol%) as well. Fig. 10 shows that the compositionf the liquid phase of each separator does not change drastically

rom the feed stream composition. This is due to the low solubil-ty of the liquids in CO2. The higher solubility of PhEtOH comparedo PhEtLau is not enough to create a significant liquid compositionariation in just one step. There are several approaches to overcome

ig. 11. Mass balance (% molar compositions) at each outlet stream of the process at temnd PhEtLau = PEL. Data for the fourth separator are based on phase equilibrium data for P

10.5 0.0047 – –11.0 – 0.0055* 0.0020*

* Data taken from Table 1.

this. One consists in the recycling of the liquid phase of the last sep-arator, which will function as a counter-current extraction column.Another is the use of a mixer-settler. Also possible is the recircula-tion of pure CO2 to the feed of the third separator so that completeseparation of PhEtLau can be achieved. To assess the feasibility ofthe latter approach, the solubility of VL in CO2 was measured andcompared to the solubilities of PhEtLau and PhEtOH (Table 5). Thehigher solubility of both VL and PhEtOH indicates that it is possibleto obtain a liquid phase with high purity PhEtLau.

Fig. 11 shows a mass balance for the process referred in Fig. 10and Table 4, with the addition of a fourth separator (dashed lines),as required to obtain the two enantiomers of the alcohol with over95% purity, which was a main goal of the present work. In this pro-cess no reflux of the liquid phase was used, and neither was CO2recycled, making it possible to recover about 1% (w/w) of the totalfeed, as indicated. This means that about 99% of the initial PhEtOHremains in the liquid phases of the separators. This type of situationis a common problem with sc-CO , due to the typically low solubil-

ities of the target compounds. However, by recycling the CO2 fromthe outlet gas stream of the fourth separator, the integrated reac-tion/separation of (R, S)-1-phenylethanol can be made technicallyviable.

perature and pressure conditions of Fig. 10 and Table 4. In this figure, PhEtOH = PEhEtOH/VL/CO2 systems, as explained in the text.

Page 8

9 ritical

4

ttaapccswuoasst1tplgft

A

siB5tX

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

70 A. Paiva et al. / J. of Superc

. Conclusions

The great potential of using sc-CO2 in integrated reac-ion/separation processes is widely acknowledged. However, onlyhe combination of thermodynamic and reaction data can trulyllow a substantiated evaluation of the viability of a process. Suchcombined approach was the object of the present study. The

hysical separation of the two isomers of (R, S)-1-phenylethanol, aompound that has been dealt with in many scientific reports, washosen as a model to demonstrate the approach. We followed thetrategy described by Reetz et al. [18] of choosing an acylating agentith a long carbon chain, to increase the dissimilarity betweennreacted alcohol and ester product. But the complete conversionf the (R)-isomer of the alcohol required an excess of acylatinggent, which made the separation step more complex. By mea-uring phase equilibrium data for the relevant binary and ternaryystems, we were able to design a separation process that will leado the separation/recovery of both (S)-1-phenylethanol and (R)--phenylethylaurate with over 95% purity. The mass balances forhe complete process reflect very clearly the low solubilities of therocessed compounds in sc-CO2, which makes the recycling of the

atter a necessary condition for the technical viability of the inte-rated reaction/separation scheme. The characterization providedor each step allows the calculation of energy requirements, andhus the assessment of the economic viability of the process.

cknowledgments

This work has been supported by the European Commis-ion in the framework of the Marie Curie Research Train-ng Network ‘Green Chemistry in Supercritical Fluids: Phaseehaviour, Kinetics and Scale-up’ (EC contract no. MRTN-CT-2004-04005), Fundacão para a Ciência e a Tecnologia (FCT, Portugal)hrough the contract POCTI/BIO/57193/04 and the grant PRAXISXI/SFRH/BPD/41546/2007 (P. Vidinha), and by FEDER.

eferences

[1] G. Brunner, Counter-current separations, J. Supercritical Fluids 47 (2009)574–582.

[2] K. Arai, R.L. Smith, T.M. Aida, Decentralized chemical processes with supercrit-ical fluid technology for sustainable society, J. Supercritical Fluids 47 (2009)628–636.

[3] H. Okamoto, K. Danjo, Application of supercritical fluid to preparation of pow-ders of high-molecular weight drugs for inhalation, Advanced Drug DeliveryReviews 60 (2008) 433–446.

[4] T. Matsuda, T. Harada, K. Nakamura, Biocatalysis in supercritical CO2, CurrentOrganic Chemistry 9 (2005) 299–315.

[5] E. Knez, Enzymatic reactions in dense gases, J. Supercritical Fluids 47 (2009)357–372.

[6] G. Brunner, Gas Extraction, Steinkopff Darmstadt Springer, New York, 1994, pp.59–144.

[7] H. Peker, M.P. Srinivasan, J.M. Smith, B.J. McCoy, Caffeine extraction rates from

coffee beans with supercritical carbon dioxide, AIChE J. 39 (1992) 761–770.

[8] U. Kopcak, R. Mohamed, Caffeine solubility in supercritical carbon dioxide/co-solvent mixtures, J. Supercritical Fluids 34 (2005) 209–214.

[9] S. Lee, M.K. Park, K.H. Kim, Y.S. Kim, Effect of supercritical carbon dioxidedecaffeination on volatile components of green teas, J. Food Science 72 (2007)S497–S502.

[

[

Fluids 55 (2011) 963–970

10] L.S. Daintree, A. Kordikowski, P.T. York, Separation processes for organicmolecules using SCF technologies, Advanced Drug Delivery Reviews 60 (2008)351–372.

11] J.L. Leazer Jr., S. Gant, A. Houck, W. Leonard, C.J. Welch, Removal of commonorganic solvents from aqueous waste streams via supercritical CO2 extrac-tion: a potential green approach to sustainable waste management in thepharmaceutical industry, Environment Science & Technology 43 (2009) 2018–2021.

12] F. Temelli, Perspectives on supercritical fluid processing of fats and oils, J. Super-critical Fluids 47 (2009) 583–590.

13] N. Tsarevsky, K. Matyjaszewski, Green atom transfer radical polymerization:from process design to preparation of well-defined environmentally friendlypolymeric materials, Chemical Reviews 107 (2007) 2270–2299.

14] J. Li, M. Rodrigues, A. Paiva, H.A. Matos, E.G. Azavedo, Vapor–liquid equilibriaand volume expansion of the tetrahydrofuran/CO2 system: application to aSAS-atomization process, J. Supercritical Fluids 41 (2007) 343–351.

15] E. Reverchon, G. Della Porta, Particle design using supercritical fluids, ChemicalEngineering & Technology 26 (2003) 840–845.

16] F. Cansell, C. Aymonier, A. Loppinet-Serani, Review on materials science andsupercritical fluids, Current Opinion in Solid State & Materials Science 7 (2003)331–340.

17] G. Blaschke, H.P. Kraft, K. Fickentscher, F. Kohler, Chromatographic-separationof racemic thalidomie and teratogenic activity of its enantiomers, Arzneimittel-Forschung/Drug Research 29 (1979) 1640–1642.

18] M.T. Reetz, W. Wiesenhofer, G. Franciò, W. Leitner, Continuous flow enzymatickinetic resolution and enantiomer separation using ionic liquid/supercriticalcarbon dioxide media, Advanced Synthesis & Catalysis 345 (2003) 1221–1228.

19] M.T. Reetz, W. Wiesenhofer, Liquid poly(ethylene glycol) and supercritical car-bon dioxide as a biphasic solvent system for lipase-catalyzed esterification,Chemical Communications 23 (2004) 2750–2751.

20] M.T. Reetz, W. Wiesenhofer, G. Franciò, W. Leitner, Biocatalysis in ionic liquids:batchwise and continuous flow processes using supercritical carbon dioxide asthe mobile phase, Chemical Communications 9 (2002) 992–993.

21] P. Lozano, T. de Diego, D. Carrie, M. Vaultier, J.L. Iborra, Lipase catalysis inionic liquids and supercritical carbon dioxide at 150 degrees C, BiotechnologyProgress 19 (2003) 380–382.

22] P. Lozano, T. de Diego, D. Carrie, M. Vaultier, J.L. Iborra, Continuous green biocat-alytic processes using ionic liquids and supercritical carbon dioxide, ChemicalCommunications 7 (2002) 692–693.

23] P. Lozano, T. de Diego, C. Mira, K. Montagne, M. Vaultier, J.L. Iborra,Long term continuous chemoenzymatic dynamic kinetic resolution of rac-1-phenylethanol using ionic liquids and supercritical carbon dioxide, GreenChemistry 11 (2009) 538–542.

24] M. Paljevac, Z. Knez, M. Habulin, Lipase-catalyzed transesterification of (R, S)-1-phenylethanol in SC CO2 and in SC CO2/ionic liquid systems, Acta ChimicaSlovenia 56 (2009) 815–825.

25] N. Fontes, M.C. Almeida, S. Barreiros, Biotransformations in supercritical fluids,in: E.N. Vulfson, P.J. Halling, H.L. Holland (Eds.), Enzymes in Nonaqueous Sol-vents, Methods in Biotechnology series, Humana Press, New Jersey, 2001, pp.565–573.

26] A. Paiva, K. Gerasimov, G. Brunner, Phase equilibria of the ternary systemvinyl acetate/(R, S)-1-phenylethanol/carbon dioxide at high pressure condi-tions, Fluid Phase Equilibria 267 (2008) 104–112.

27] A.P. de los Rios, F.J. Hernandez-Fernandez, F. Tomas-Alonso, D. Gomez, G. Vil-lora, Synthesis of esters in ionic liquids – the effect of vinyl esters and alcohols,Process Biochemistry 43 (2008) 892–895.

28] A.P. de los Rios, F.J. Hernandez-Fernandez, F. Tomas-Alonso, D. Gomez, G. Vil-lora, Synthesis of flavour esters using free Candida antarctica lipase B in ionicliquids, Flavour and Fragrance J. 23 (2008) 319–322.

29] D.Y. Peng, D.B. Robinson, 2-Phase and 3-phase equilibrium calculationsfor systems containing water, Canadian J. Chemical Engineering 15 (1976)595.

30] P.M. Mathias, H.C. Klotz, J.M. Prausnitz, Equation-of-state mixing rules for mul-

ticomponent mixtures – the problem of invariance, Fluid Phase Equilibria 67(1991) 31–44.

31] O. Pfohl, S. Petkov, G. Brunner, Usage of PE – A Program to Calculate PhaseEquilibria, Herbert Utz Verlag, München, 1998.

32] S. Angus, B. Armstrong, K.M. Reuck, IUPAC – International ThermodynamicsTables of the Fluid State Carbon Dioxide, Pergamon Press, Oxford, UK, 1976.