J. of Supercritical Fluids 35 (2005) 128–132

Isolation of phenolic antioxidant compounds by SFC

Pilar Ramıreza, Tiziana Fornaria, F. Javier Senoransa,Elena Ibanezb,∗, Guillermo Regleroa

a Area de Tecnolog´ıa de Alimentos (Unidad Asociada al CSIC), Facultad de Ciencias, Universidad Aut´onoma de Madrid,28049 Cantoblanco, Madrid, Spain

b Departamento de Caracterizaci´on de Alimentos, Instituto de Fermentaciones Industriales (CSIC),Juan de la Cierva 3, 28006 Madrid, Spain

Received 18 May 2004; received in revised form 11 January 2005; accepted 13 January 2005

Abstract

Specially designed chromatographic columns were evaluated for antioxidant compounds separation by supercritical fluid chromatography(SFC) with pure CO2. Columns studied involved the use of commercial octadecylsilica particles (ODS) and silica particles coated witha stationary phase commonly used in gas chromatography, such as CW20M (polyethylene glycol) of high polarity. In order to select thea del moleculea utm No elutionw active sitesr xtraction)u©

K

1

iwotbflsacfl[t

nsibleri-reer to

egiesSFC

aseion,lica-laterweasyvena firsttion.ationOrder

0d

ppropriate chromatographic conditions to elute antioxidant compounds of phenolic structure, carnosic acid was selected as a mond a theoretical study of the solubility of carnosic acid was performed. Carnosic acid was eluted using CO2 as mobile phase (withoodifiers) with a pressure programming up to 37 MPa (and at different temperatures) but only when coated particles were used.as possible when ODS particles were tested, probably due to the strong interactions of the acid group of the molecule with the

emaining in the column. A separation of a complex mixture of a rosemary antioxidant extract (obtained by supercritical fluid esing the columns designed in the present work is shown.2005 Elsevier B.V. All rights reserved.

eywords:Carnosic acid; Antioxidant compounds; Supercritical fluid chromatography; Packed capillary columns; Coated particles

. Introduction

At present, one of the most important areas of researchn food technology is the isolation of natural compoundsith functional properties from natural sources. Isolationf pure compounds or even groups of compounds with

herapeutic properties, such as antioxidants, is usually doney HPLC at preparative scale (Prep-LC)[1,2]. Supercriticaluid chromatography (SFC) at preparative scale has beenuggested as an alternative to Prep-LC with some advantagesssociated to its use such as the easy recovery of the isolatedompounds by a simple decompression of the supercriticaluid where the mobile phase is spontaneously eliminated3] and the possibility of obtaining pure compounds withouthe use of solvents.

∗ Corresponding author. Tel.: +34 91 562 2900x388; fax: +34 91 564 4853.E-mail address:elena@ifi.csic.es (E. Ibanez).

Phenolic compounds have been described as respoof the antioxidant activity of different matrices of natural ogin such as aromatic plants[4]. Considering the polar natuof some of these compounds, a study is needed in ordbe able to elute such compounds in SFC. Different strathave been described to separate polar compounds bywith packed capillary columns: tuning the mobile phpolarity by adding polar cosolvents at low concentrator tuning the polarity of the stationary phase through sibased particles deactivation and/or polymer coating andelution with pure CO2 [5]. Packed capillary columns shohigh efficiency, a reasonable sample loadability and areto prepare with a wide variety of packing materials. Emore important, these types of columns can be used asstep towards a scaling-up to a preparative scale separa

The objective of the present work has been the separof phenolic antioxidant compounds by SFC with pure C2using specially designed chromatographic columns. In o

896-8446/$ – see front matter © 2005 Elsevier B.V. All rights reserved.oi:10.1016/j.supflu.2005.01.002

P. Ram´ırez et al. / J. of Supercritical Fluids 35 (2005) 128–132 129

to select the appropriate chromatographic conditions to eluteantioxidant compounds of phenolic structure, carnosic acidwas selected as a model molecule. A thermodynamic model-ing based on the use of the Group Contribution AssociatingEquation of State (GCA-EoS)[6], was applied to estimatethe solubility of carnosic acid in pure CO2. Elution wasattempted using packed capillary columns with commercialoctadecylsilica particles (ODS) and silica particles coatedwith a polymeric stationary phase of high polarity suchas CW20M (polyethylene glycol). Also, evaluation of thecolumns used in the present study in terms of efficiency andsurface activity was performed. Moreover, a separation ofa complex mixture of a supercritical rosemary antioxidantextract using the columns designed in the present work isshown to demonstrate the usefulness of this columns forthe isolation of certain phenolic compounds in complexsamples.

The present study is presented as a first step towards thedevelopment of a process for off-line extraction and frac-tionation (SFE/SFC) at preparative scale of antioxidant com-pounds from aromatic plants.

2. Experimental

2.1. Samples

( e ac-t thylb wasu s( turea ed top blin,I 77 Ki

dr ain).S ands

2

rloE torw ting-v ng1 tec-tC 300p wass ff ces-t ead-v tion

valve via a flow splitter consisting of a silica tubing of 13�mi.d.× 20 cm length. Conditions of analysis are detailed in thefigure caption.

2.3. Columns

Packed capillary columns were prepared according to areported procedure[7] by using 500�m i.d. stainless steeltubing (Symta Ltd., Madrid, Spain) of 25 cm length. Thestainless steel tubing was deactivated with 20% polyethyleneglycol (Supelco, Bellefonte, USA). The packing procedurewas performed at a starting pressure of 8 MPa followed by apressure rate of 0.3 MPa min−1 up to 34 MPa. The tubing wasintroduced into an ultrasonic bath maintained at room tem-perature. Once filled, the column was allowed to depressurizeovernight. The columns were conditioned prior to their usein SFC using a pressure and temperature program as follows:from 12 to 34 MPa at 0.4 MPa min−1 and from 313 to 453 Kat 3 K min−1.

2.4. Octadecylsilica particles

Octadecylsilica particles of 10�m (Spherisorb ODS2,Hichrom Ltd., Reading, UK) were conditioned prior to theiruse by washing them with ethanol, and dried by heating at4a t thew tepsi e in-t

2

,U ed asm

2 sed

b 0M( oat-i urgeg

2

PA,U 5 g,d l ex-t sc

tiont mice in ag

For efficiency measurements, a mixture of puren-alkanesC12–C28) (Sigma, St. Louis, MO) was used. For surfacivity evaluation, a mixture of menthol, benzoic acid, meenzoate and 2,6-dimethylaniline (Fluka, Switzerland)tilized. Carnosic acid standard fromRosmarinus officinali93% purity) was used as model sample of phenolic nantioxidants (Sigma). Hexane and dichloromethane usrepare the solutions were purchased from LabScan (Du

reland). Carnosic acid solutions were maintained at 2n dark glass flasks.

The rosemary sample (R. officinalisL.) consisted of drieosemary leaves (Murciana de Herboristeria, Murcia, Spamples were ground under cryogenic carbon dioxidetored in amber flasks at 253 K until use.

.2. Instrumentation

A supercritical fluid chromatograph SFC 3000 (Carba, Milan, Italy) equipped with a flame ionization detecas used. Sample was loaded by a time-controlled, rotaalve injection device (Vici, Houston, TX, USA) containi�L internal loop. Injector temperature was 313 K. De

or temperature was kept at 623 K. SFC grade CO2 (Liquidarbonic, Madrid, Spain) was pumped by using a SFCump (Carlo Erba). The flow rate of the mobile phaseet by using a linear restrictor of 13�m i.d.× 20 cm made oused silica tubing (Composite Metal Services Ltd., Worershire, UK) connected to the column through a zero dolume union. The columns were connected to the injec

33 K for 1 h in a fluidized bed reaction vessel[8] whichllowed continuous inert gas (helium) purge throughouashing, drying and coating. By performing all of these s

n the same container, handling was minimized and thegrity of the packing material maintained.

.5. Coating of silica particles

Porous silica particles (10�m, 60A, Hichrom, ReadingK) were used as base material. Particles were washentioned above.

.5.1. Silica particles coating with GC stationary phaseSilica particles (0.3 g) were placed in the fluidiz

ed reaction vessel and mixed with 3% (w/w) CW2polyethylene glycol) dissolved in dichloromethane. Cng was performed at room temperature with He as pas.

.6. Supercritical fluid extraction of rosemary

A Suprex PrepMaster (Suprex Corp., Pittsburgh,SA) supercritical fluid extractor was used. Sample (0.8ry weight basis) was placed into a 5 mL stainless stee

raction cell. The supercritical CO2 flow rate (3 mL/min) waontrolled using a needle valve as variable restrictor.

Sample was extracted at 10 MPa and 313 K. Extracime was 5 min static extraction followed by 60 min dynaxtraction. The supercritical fluid extract was collectedlass vial (2 cm× 0.5 cm).

130 P. Ram´ırez et al. / J. of Supercritical Fluids 35 (2005) 128–132

3. Results and discussion

3.1. Column efficiency and surface activity

Packed capillary columns of special dimensions (500�mi.d.× 25 cm length) were used to test the elution of phe-nolic compounds. Previous studies have demonstrated theperformance of packed capillary columns of large diameter(500�m i.d.) packed with 10�m particles in terms of effi-ciency, sample capacity and sample volume loadability com-pared to packed capillary columns of 180�m i.d.[9]. Samplecapacity (c) and sample loadability (l) represent the influenceof the quantity of solute injected on the separation power ofthe tested column, and are of great importance for scalingup the separation to preparative scale. Columns made using500�m tubing have five times more sample capacity and 4.4times more sample loadability than columns prepared with180�m tubing[9]; therefore, the expected separation powerof the columns prepared with 500�m tubing would be lessaffected for both the volume injected and the concentrationof the solutes in the sample.

Before studying the use of the columns for phenolic com-pounds separation, column efficiency and activity were evalu-ated. Efficiency was measured by separation of ann-alkanessolution at a constant pressure of 9 MPa and at a constanttemperature of 353 K. The expected efficiency obviously de-p typeo worki thc n1 th1 ase( 8000t nd2

ac-c d po-l ,t f thec tingc ac-t ningc rac-ts inedb ied,C facew yerc atedb se.

earlyn ses,e Thec ohols( zoica n. If

Table 1Surface activity (as asymmetry factor, As) of the different columns evaluatedin this study

Compound 500�m i.d.×25 cm, 10�mODS column

500�m i.d.× 25 cm,10�m particles coatedwith 3% CW20M

Methyl benzoate 1.2 1.12,6-Dimethylaniline 1.5 (T) 1.4 (T)Menthol 1.6 (T) 1.5 (T)Benzoic acid – (no elution) 1.4 (T)

T: peak tailing; separation conditions: 393 K, 9–30 MPa at 0.4 MPa min−1.Experiments performed by triplicate.

the deactivation is not complete, these polar solutes can in-teract with the residual silanol groups resulting in tailing andincreased retention. Comparing data of Columns 1 and 2 (us-ing pure CO2), it can clearly be seen that the coated particlesoffer the advantage of the separation of the polar compoundswith, in most cases, only slight peak tailing. Menthol, ani-line, and even benzoic acid can be eluted in all cases withthis type of columns although some peak tailing can be seenfor benzoic acid; nevertheless, elution of benzoic acid assuresa complete deactivation of the active sites of the column.

3.1.1. Analysis of carnosic acid solubility insupercritical CO2

After testing the columns for efficiency and surface activ-ity, carnosic acid was selected as model compound to studyits elution in SFC with pure carbon dioxide. Temperature andpressure ranges which guarantee high carnosic acid solubil-ity in CO2, would be appropriate chromatographic conditionsfor elution. Thus, carnosic acid solubilities in pure CO2 wereestimated using a theoretical analysis.

The isofugacity criteria provides the general equation ofequilibrium between the solid solute (s) and the supercriticalphase. Taking into account that the solubility of the supercrit-ical fluid in the solid phase can be considered negligible, thesolute mole fraction (ys) in the supercritical phase is given by

y

w ures dE so icalp asingi

ri-m ticalC d outa suresr them Ow molef peri-m ol-

ends on the quality of the packing procedure and on thef material packed. The columns studied in the present

ncluded a 500�m i.d.× 25 cm length column packed wiommercial 10�m octadecylsilica (ODS) particles (Colum) and a 500�m i.d.× 25 cm length column packed wi0�m particles coated with 3% CW20M stationary phColumn 2). The average values obtained ranged fromo 10500 plates/m atk′ = 4.5 and 7 (Column 1) and arou0 000 plates/m (k′ = 3) for Column 2.

The lack of polarity of neat carbon dioxide makes it uneptable as a mobile phase for high molecular weight anar compounds on packed columns. When neat CO2 is usedhe risk of surface activity increases and the behavior oolumns in terms of activity is very important for separaompounds of different polarity. In the present work, theivity was evaluated by separating a test mixture contaiompounds of different polarity that have different inteions with either the mobile or the stationary phase.Table 1hows data of surface activity (asymmetry factor) obtay injection of the test mixture in the two columns studolumn 1 (ODS), deactivated by reaction of the silica surith a monomeric silylation reagent to obtain a monolaoverage with a C18 functionality, and Column 2, deactivy coating the particles with 3% CW20M stationary pha

As can be seen, methyl benzoate, a compound with no interactions with neither mobile nor stationary phalute from both columns with symmetrical peak shapes.ompounds that have strong interactions, such as alcmenthol), amines (2,6-dimethylaniline) and acids (bencid) are much more affected by the type of deactivatio

s = Psats

PE

herePsats is the sublimation (vapor) pressure of the p

olid,P the pressure, the ratioPsats /P the ideal solubility an

the enhancement factor[10]. Usually, very small valuef the fugacity coefficient of the solute in the supercrithase, produces very large enhancement factors incre

deal solubility by various orders of magnitude.In a previous work[6] the authors reported expe

ental data on carnosic acid solubilities in supercriO2 + ethanol as a modifier. Measurements were carriet temperatures in the range of 313.15–333.15 K, presanging from 28 to 40 MPa, and at different content ofodifier ethanol (from 0.7 to 10%). Solubilities in pure C2ere not possible to measure, because carnosic acid

ractions were in the same order of magnitude of the exental error (10−7). The correlation of the experimental s

P. Ram´ırez et al. / J. of Supercritical Fluids 35 (2005) 128–132 131

Table 2Carnosic acid physical properties[6] used in the solubility calculations

Critical temperature,Tc 950.5 KCritical pressure,Pc 1.04 MPaSolid volume,vsolid

s 393.03× 10−6 m3/molSublimation pressure,Psat

s 8.22× 10−12 MPa at 313.15 K4.33× 10−9 MPa at 343.15 K7.34× 10−6 MPa at 373.15 K

ubility data using the GCA-EoS[11,12]was also presented inthis work. Pure carnosic acid physical properties (i.e. criticalproperties, solid volume and sublimation pressure) were alsoestimated and are reproduced inTable 2. A detailed descrip-tion of the methods employed to estimate these properties,together with the thermodynamic modeling details, are givenin [6].

In the present work carnosic acid solubilities in pure CO2are calculated, extrapolating the thermodynamic modelingprocedure given in[6] to 0% ethanol content. The values ob-tained in the temperature range of 313.15–333.15 K are ingood agreement with the experimental data reported in[6].Fig. 1shows the variation of carnosic acid solubility with CO2density, calculated at 313.15, 343.15 and 373.15 K, and pres-sures from 20 to 37 MPa. As expected, the model predictionsresult in an increase of carnosic acid solubility with CO2density. Taking into account that 37 MPa is the maximumpressure achievable in the chromatographic system tempera-tures around 313.15 K would be the best conditions to elutecarnosic acid using pure CO2.

The low solubilities predicted by the thermodynamicmodel (mole fractions in the order of 10−8 to 10−7), indi-cated that carnosic acid would not be easy to elute usingpure CO2; moreover, considering that the column packedwith chemically bonded ODS (C18) particles has an excess ofsilanol groups on the silica surface (which have not been com-

F as af

Fig. 2. SFC chromatogram of carnosic acid in the 25 cm× 500�m i.d. col-umn packed with 3% CW20M coated silica particles (10�m). Chromato-graphic conditions: column temperature 313.15 K, pressure program from15 to 37 MPa at 1 MPa min−1.

pletely deactivated by the reaction with silylation reagents)this can explain the impossibility of eluting carnosic acidin such columns. Nevertheless, elution of carnosic acid wasaccomplished when a packed column with silica particlescoated with 3% of CW20M was used.Fig. 2 shows thechromatogram obtained for the elution of carnosic acid at313.15 K and using a pressure program up to 37 MPa. Ex-perimental data showed that carnosic acid was eluted at veryhigh pressures and at moderate temperatures (from 313.15 to373.15 K) increasing its retention while increasing the tem-perature of the column, according to the decrease of solubilitywith temperature predicted by the model.

By comparing the results obtained with the two columnsevaluated, it is easy to realize that the separation of carnosicacid was possible only when homogeneous and controlledcoating was achieved, that is, only when exhaustive deacti-vation of the silica surface was performed. When coating thesilica particles with liquid stationary phases the deactivationtakes place through different mechanisms, one is associatedto the degree of physical coverage of the active sites of thesilica particles, while a different one is related to the presenceof certain polar functional groups, such as phenyl groups, inthe polymeric stationary phase that interact with the residualsilanol groups on the silica phase surface through hydrogen-bond interaction; such interactions can provide further deac-tivation of the residual silanol groups on the particles thusf

mnst plexs thea t thec ons,a ajorc cted)

ig. 1. Carnosic acid solubilities predicted by the GCA-EoS modelunction of CO2 density: (�) 313.15 K; (�) 343.15 K; (©) 373.15 K.

avoring the separation of more polar compounds.As an example of the usefulness of the designed colu

owards the separation of phenolic compounds in a comample,Fig. 3shows the chromatogram corresponding tonalysis of a supercritical rosemary extract obtained aonditions described under experimental. At this conditi

quite complex mixture is obtained where the mompound was identified as carnosic acid (as expe

132 P. Ram´ırez et al. / J. of Supercritical Fluids 35 (2005) 128–132

Fig. 3. SFC-FID of SFE rosemary on CW20M-coated silica columns. Ex-tract corresponding to 10 MPa and 313 K. Column temperature 373 K, pres-sure program from 15 to 37 MPa at 1 MPa min−1.

while the other compounds can correspond to other phenoliccompounds previously described in such antioxidant extracts[13]. As can be seen, elution of carnosic acid was achievedunder conditions of high selectivity allowing the isolation ofthis compound once working at preparative scale.

As a conclusion, the present study shows, for the first time,the possibility of separating phenolic compounds with antiox-idant properties, such as carnosic acid, by SFC using coatedpacked capillary columns and pure CO2. This work can beconsidered as a first step towards the optimization of separa-tion conditions to fractionate complex extracts using a prepar-ative scale SFC. At present, studies are conducted in ourlaboratory to achieve the purification of carnosic acid fromrosemary extracts by preparative-SFC using coated packedcolumns.

Acknowledgements

This work was supported by Project AGL2000-0448(Ministerio de Ciencia y Tecnologia, Spain). P. Ramirez

greatly acknowledges the Spanish Ministry of Science andTechnology for the FPI grant. T. Fornari would like to ac-knowledge the financial support of the Secretarıa de Estadode Educacion y Universidades for the fellowship at the Uni-versidad Autonoma de Madrid, Spain.

References

[1] H. Esaki, H. Onozaki, Y. Morimitsu, S. Kawakishi, T. Osawa,Potent antioxidative isoflavones isolated from soybeans fermentedwith Aspergillusi isaitoii, Biosci. Biotechnol. Biochem. 62 (1998)740.

[2] W.C. Sang, J.C. Eun, Y.H. Tae, H.C. Kyoung, Antioxidative activityof acylated anthocyanin isolated from fruit and vegetables, J. FoodSci. Nutr. 2 (1997) 191.

[3] K. Coleman, R. Boutant, F. Verillon, Practical Aspects of Prepara-tive Supercritical Fluid Chromatography, Gordon and Breach SciencePublishers, UK, 1999.

[4] K. Schwarz, W. Ternes, Antioxidative constituents ofRosmarinusofficinalis and Salvia officinalis. II. Isolation of carnosic acid andformation of other phenolic diterpenes, Z. Lebensm. Unters. Forsch.195 (1992) 99.

[5] E. Ibanez, F.J. Senorans, Tuning of mobile and stationary phasepolarity for the separation of polar compounds by SFC, J. Biochem.Biophys. Meth. 43 (2000) 25.

[6] A. Chafer, T. Fornari, A. Berna, E. Ibanez, G. Reglero, Solubility ofsolid carnosic acid in supercritical CO2 with ethanol as a cosolvent,J. Supercrit. Fluids, in press.

ry993)

Lee,s for62

para-illary998)

[ ilib-

[ ilib-and

[ ofting003)

[ uidl-fluid

[7] A. Malik, W. Li, M.L. Lee, Preparation of long packed capillacolumns using carbon dioxide slurries, J. Microcol. Sep. 5 (1361.

[8] K.M. Payne, B.J. Tarbet, J.S. Bradshaw, K.E. Markides, M.L.Simultaneous deactivation and coating of porous silica particlemicrocolumn supercritical fluid chromatography, Anal. Chem.(1990) 1379.

[9] E. Ibanez, S. Lopez-Sebastian, J. Tabera, G. Reglero, Setion of carotenoids by sub-SFC with coated, packed capcolumns and neat carbon dioxide, J. Chromatogr. A 823 (1313.

10] J.M. Prausnitz, Molecular Thermodynamics of Fluid-phase Equria, Prentice-Hall, 1969, p. 164.

11] H.P. Gros, S.B. Bottini, E.A. Brignole, High pressure phase equrium modeling of mixtures containing associating compoundsgases, Fluid Phase Equilibria 139 (1997) 75.

12] O. Ferreira, T. Fornari, E.A. Brignole, S.B. Bottini, Modelingassociation effects in mixtures of carboxylic acids with associaand non-associating components, Latin Am. Appl. Res. 33 (2217.

13] F.J. Senorans, E. Ibanez, S. Cavero, J. Tabera, G. Reglero, Liqchromatographic–mass spectrometric analysis of supercriticaextracts of rosemary plants, J. Chromatogr. A 870 (2000) 491.