Analysis of triacylglycerols on porous graphitic carbon by high temperature liquid chromatography

Journal of Chromatography A, 1157 (2007) 462–466

Short communication

Analysis of triacylglycerols on porous graphitic carbon by high temperature liquid chromatography B´erang`ere Merelli, Marine De Person, Patrick Favetta, Michel Lafosse ∗ ICOA, CNRS UMR 6005, Universit´e d’Orl´eans, BP 6759, Rue de Chartres, 45067 Orleans Cedex 02, France Received 12 January 2007; received in revised form 29 April 2007; accepted 2 May 2007 Available online 7 May 2007

Abstract The retention behaviour of several triacylglycerols (TAGs) and fats on Hypercarb® , a porous graphitic carbon column (PGC), was investigated in liquid chromatography (LC) under isocratic elution mode with an evaporative light scattering detector (ELSD). Mixtures of chloroform/isopropanol were selected as mobile phase for a suitable retention time to study the influence of temperature. The retention was different between PGC and nonaqueous reversed phase liquid chromatography (NARP-LC) on octadecyl phase. The retention of TAGs was investigated in the interval 30–70 ◦ C. Retention was greatly affected by temperature: it decreases as the column temperature increases. Selectivity of TAGs was also slightly influenced by the temperature. Moreover, this chromatographic method is compatible with a mass spectrometer (MS) detector by using atmospheric pressure chemical ionisation (APCI): same fingerprints of cocoa butter and shea butter were obtained with LC–ELSD and LC-APCI–MS. These preliminary results showed that the PGC column could be suitable to separate quickly triacylglycerols in high temperature conditions coupled with ELSD or MS detector. © 2007 Elsevier B.V. All rights reserved. Keywords: Porous graphitic carbon; Triacylglycerols; Cocoa and shea butter; Column temperature; Liquid chromatography

1. Introduction The European Commission has recently published a new directive [1] about chocolate specifying the maximum amount of cocoa alternatives so-called cocoa butter equivalents (CBE) [2]. These are mostly mixtures of various vegetable fats such as shea butter. This European directive requires the implementation of analytical methods to control its application and to discover cases of fraud. In order to comply with this regulation, a method has been proposed using reversed-phase liquid chromatography on octadecyl bonded stationary phase with evaporative light scattering detection (ELSD) [3]. As the composition of the mobile phase influences the droplet size and the detector response, isocratic mode gives a response which is practically independent of the solute in homologous series [4]; that’s why this mode has been chosen in this study although gradient elution permits higher efficiency. Abbreviations: ELSD, evaporative light scattering detector; PGC, porous graphitic carbon; TAGs, triacylglycerols ∗ Corresponding author. Tel.: +33 238 49 45 75; fax: +33 238 41 72 81. E-mail address: [email protected] (M. Lafosse). 0021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2007.05.005

According to the conditions proposed by Dionisi et al. [3], the analysis of TAGs was long (about 54 min) although recent works have shown fast analysis of TAGs on ODS columns [5–7]. A simple method to decrease the analysis time consists in increasing the temperature of elution, but that is not suited for all bonded silica. Only few bonded silica can be used with temperature higher than 70 ◦ C and used for lipid analysis [8,9]. We therefore have investigated the potentiality of porous graphitic carbon (PGC) which can be used at high temperature [10] This support possesses rigid and planar surfaces and affords two types of solute–adsorbent interactions: dispersive interactions and polar retention effect of graphite (PREG) [11,12]. It represents an underutilised stationary phase for RPLC of lipophilic compounds and has shown a potential for the discrimination of lipid species containing carbon double bonds [13,14] and glycolipids [15,16]. Interactions between the carbon double bond and the surface of PGC can be expected due to its polarisable nature, the number of double bonds and the conformation of the molecule. In the few studies with non-aqueous mobile phase with PGC [13,17,18], the increase in the hydrocarbon chain length of the molecule always induced an increase in retention whereas retention and selectivity between saturated and

B. Merelli et al. / J. Chromatogr. A 1157 (2007) 462–466

unsaturated fatty acid methyl esters (FAMEs) were compared in different mobile phases [19]. The objective of the present work was to design an efficient and rapid chromatographic method to separate TAGs on PGC with cocoa and shea butters as model samples. The effect of experimental parameters such as organic solvent ratio and column temperature, on the chromatographic behaviour of the TAGs, is discussed here. 2. Materials and methods 2.1. Chemicals Chloroform, and isopropanol (HPLC gradient grade) were purchased from J.T. Baker (Noisy le sec, France). Tripalmitolein PoPoPo (16:1, cis-9), tripalmitelaidin PaPaPa (16:1, trans-9), tripalmitin PPP (16:0), trilinolein LLL (18:2, cis-9,12), trilinolenin LnLnLn (18:3, cis-9,12,15), tripetroselinin PePePe (18:1, cis-6), triolein OOO (18:1, cis-9), tristearin SSS (18:0) and trierucin EEE (22:1, cis-13) were obtained from Sigma (Saint Quentin Fallavier, France). Cocoa butter and shea butter were purchased at factories. All the samples were dissolved in chloroform at a concentration of 0.1 mg/mL. 2.2. Apparatus The liquid chromatographic system consisted of a Jasco (Nantes, France) model PU-2085 pump, a Rheodyne (Berkeley, CA, USA) model 7725 injector with a 5 ␮L sample loop and an evaporative light scattering detection (ELSD) system (Sedere, Alfortville, France) model Sedex 85. The usual ELSD detector settings were as follows: drift tube temperature 45 ◦ C, nebuliser gas pressure 3.5 bar, photomultiplier 8. The PGC separation was carried out on a Hypercarb® column (150 mm × 2.1 mm ID, particle size 5 ␮m) from Thermoquest, Hypersil (Runcorn, UK). The column was placed in a Jet Stream 2 oven (WO Industrial Electronics, Langenzersdorf, Austria). The flow rate was set at 0.2 mL/min. Data were collected and analysed using EZ-Chrom version 6.7 software. LC–MS analyses were carried out using a Perkin-Elmer (Toronto, Canada) model LC-200 binary pump and a Sciex (Forster City, CA, USA) API 300 triple quadrupole mass spectrometer with an APCI source in the positive ion mode. Operating conditions were as follows: nebuliser gas flow rate 1.22 L/min, curtain gas flow rate 1.31 L/min, temperature = 400 ◦ C, needle current = 3 ␮A, declustering potential = 34 V, focusing potential = 340 V, entrance potential = 10 V. Mass spectra were acquired by scanning the range m/z 800–1000 in order to obtain ions without fragmentation and to compare the chromatograms with the universal response of ELSD. Unit resolution was used as usual during spectra, the dwell-time was set at 2 ms and the pause-time was 5 ms. Injections were done by a Perkin-Elmer series 200 autosampler fitted with a 20 ␮L loop. All the results were acquired with the Analyst version 1.3.1 software (Sciex Applied Biosystems). PGC separation in LC–MS was carried out on a Hypercarb® column (150 mm × 4.6 mm ID, particle size 5 ␮m) from Thermoquest, Hypersil (Runcorn, UK) at a flow rate of 1 mL/min. Column temperature was reg-

463

ulated by a Jet Stream oven at 60 ◦ C. Although APCI enables TAG ionisation, it was improved by adding ammonium acetate (10 mM) to isopropanol in mobile phase without modification of the retention and the peak shape. 3. Results and discussion To assess the retention behaviour of TAGs on PGC, the high hydrophobicity of the packing required non-aqueous mixtures as mobile phase. Several TAGs were first tested as probes and mixtures of chloroform–isopropanol (CHCl3 –iPrOH) were chosen as the more suitable for elution of cocoa and shea butters and ELSD response. 3.1. Influence of the composition of mobile phase An analysis of five standards (PPP, LLL, LnLnLn, PoPoPo and PaPaPa) was performed on PGC at 60 ◦ C using various CHCl3 –iPrOH mobile phases (from 0.6 to 0.9 CHCl3 volume fraction). As expected in NARP-LC, the retention of TAGs on a PGC column decreases as the CHCl3 amount is increased (Fig. 1), and a good linearity is observed for the plot of log k versus the volume fraction of CHCl3 with a correlation coefficient r2 higher than 0.99. The slopes of variation are similar and the average value of the slope was about 3.97, greater than for FAMEs) (2.5–2.8) [13] showing a stronger interaction of TAGs on this packing. Moreover, it appeared that selectivity between these TAGs was weakly affected by the content of CHCl3 in the mobile phase. Nevertheless, selectivity slowly increased as the CHCl3 ratio decreases more particularly for the LnLnLn/PaPaPa and LLL/PaPaPa couples. For all the following analyses, the CHCl3 /iPrOH (80:20) mixture was selected as mobile phase for a suitable retention time to study the influence of temperature. 3.2. Influence of the nature of analytes In reversed phase liquid chromatography on octadecyl silica (ODS), TAGs are separated according to the combined effect of the chain-lengths of fatty acid moieties plus their degree of unsaturation: the logarithmic retention factor k of TAGs increases

Fig. 1. Influence of CHCl3 ratio on the retention of five TAGs. Column Hypercarb (150 mm × 2.1 mm I.D.). Mobile phase: CHCl3 –iPrOH mixtures. Flow-rate: 0.2 mL/min. Column temperature: 60 ◦ C. ELSD.

464

B. Merelli et al. / J. Chromatogr. A 1157 (2007) 462–466

was greater for the saturated TAGs than for the unsaturated solutes having the same alkyl chain length. Moreover, the α selectivity was influenced by the temperature for some couples of solutes. In Fig. 1, SSS was more retained than EEE at 30 ◦ C (αSSS/EEE = 1.65) while the opposite effect was observed at 70 ◦ C (αEEE/SSS = 1.359). Then the selectivity was also reversed for the couple PaPaPa/LLL. Finally, the increase of the temperature decreases the selectivity of the geometrical isomers (PaPaPa/PoPoPo). 3.4. Application Fig. 2. Plot of ln k against double bond number for various saturated (SSS, PPP) and unsaturated (EEE, OOO, PePePe, PaPaPa, LLL, LnLnLn and PoPoPo) TAGs on PGC, with CHCl3 –iPrOH (80:20) as mobile phase. Column temperatures: 30, 50 and 70 ◦ C. Flow-rate: 0.2 mL/min.

roughly in the order of increasing equivalent carbon number (ECN) defined as the number of carbons of fatty acids minus twice the number of double bonds [15]. Then the ECN values are 54, 48, 42 and 36 for SSS, OOO, LLL and LnLnLn, respectively involving a decrease of retention as the unsaturation increases. It results from the ECN values that the effect of the first double bond (SSS/OOO) on the ODS retention is the same than for the second (OOO/LLL) and the third one (LLL/LnLnLn). Thus, on ODS, the selectivity given by log α (i.e. log k2 − log k1 ) between these different couples will be similar. On PGC, Fig. 2 shows that the decrease in retention is much more significant for the first double bond than for the second and the third one. Thus, the α selectivities at 30 ◦ C were as follows: αSSS/OOO was 27.9, then αOOO/LLL was 2.58, then αLLL/LnLnLn = 1.097. In the same manner, the increase of the alkyl chain increases the retention on ODS. This retention increase is about the same for the couples SSS/PPP and OOO/PoPoPo which have the same ECN increase (ECNPoPoPo = 42). On PGC, Fig. 2 shows that the retention increases also as the alkyl chain increases but the selectivity α is higher for the saturated TAGs (αSSS/PPP = 6.0) than for unsaturated TAGs (αOOO/PoPoPo = 3.29) illustrating the influence of the analyte structure on this support. A previous study [11,21] has shown a similar behaviour of FAMEs on PGC. As PGC possesses an important steric selectivity [11,12], this may play a more important role than the polarity change induced by double bonds. Moreover, when the position of the double bond does not influence the retention (PePePe 18:1 cis-6 and OOO 18:1 cis9, αOOO/PePePe = 1.023), trans-structure increases the retention: PaPaPa (16:1 trans-9) was more retained than PoPoPo (16:1 cis9) (αPaPaPa/PoPoPo = 1.362). This behaviour was similar to that observed for fatty acids such as methyl oleate (cis) and methyl elaidate (trans) on pyrocarbon-modified silica gel [20] and on PGC [21].

The analysis of cocoa butter and shea butter was performed by LC on PGC with CHCl3 /iPrOH (80:20) as mobile phase with an ELSD. The chromatograms obtained at 50 ◦ C are shown in Fig. 3. The main advantage is the short analysis time (∼20 min) compared to C18 columns with a simple chromatographic method in isocratic mode where the ELSD response is universal [4] and enabling a simpler quantification than in gradient elution mode. Without standards such as POP, POS and SOS in cocoa butter and SOO and SOS in shea butter, we have used mass spectrometry which has become a widespread method [22–26] and particularly atmospheric-pressure chemical-ionisation (APCI) mostly used because of the relatively simple mass spectra obtained and the possibility of identifying the positional isomers [27–29]. The previous chromatographic conditions with LCELSD were transposed to LC-APCI–MS. In order to improve the ionisation, ionic species such as ammonium acetate has been added to the mobile phase [30]. As the aim of the MS use was only to verify the position of major components of butters and compare the MS and ELSD chromatograms, only [M + NH4 ]+ adduct was present for each peak of the cocoa butter and shea butter spectra in the range m/z 800–1000. Fig. 4 shows the fingerprints of butters obtained with LC-APCI–MS at 60 ◦ C. They were similar

3.3. Influence of the column temperature Analyses of standards were carried out at various temperatures in the interval 30–70 ◦ C. Fig. 2 shows a classical decrease of the retention as the temperature increases. This variation

Fig. 3. Separation of cocoa butter and shea butter. Mobile phase: CHCl3 –iPrOH (80/20, v/v). Flow-rate: 0.2 mL/min. Column: Hypercarb (150 mm × 2.1 mm I.D.), temperature: 50 ◦ C; ELSD.

B. Merelli et al. / J. Chromatogr. A 1157 (2007) 462–466

465

such as [M + NH4 RCOOH] may be obtained by using suitable mass range. However, the fingerprint of shea butter was more complicated due to the presence of UV-absorbing insaponifiable compounds such as sterols and tocopherols [31] in small quantity and less retained than TAGs. Using data obtained at 60 ◦ C for standards (OOO, SSS) and solutes identified in butters (POP, POS, SOS, SOA, SOO) on columns with 4.6 and 2.1 mm I.D., respectively, we can underscore a good correlation (correlation coefficient R2 > 0.99) between the retention and the ECN values, enabling thus to confirm the identity of SOA (Fig. 5). In this plot, the curve relative to the unsaturation effect concerns only one double bond on each alkyl chain. 4. Conclusion

Fig. 4. LC-APCI–MS of cocoa butter and shea-butter. Mobile phase: CHCl3 –iPrOH (80:20) + ammonium acetate. Flow-rate: 1 mL/min. Column: Hypercarb (150 mm × 4.6 mm I.D.), temperature: 60 ◦ C.

to the fingerprints obtained at 50 ◦ C with ELSD (selectivity variation between two consecutive peaks was minus than 15% on 10 ◦ C range) and the retention time was shorter due to the temperature effect. MS of the compounds confirms the identity of major peaks of butters: in cocoa butter the major peaks were POP ([M + NH4 ]+ = 851), POS ([M + NH4 ]+ = 879) and SOS ([M + NH4 ]+ = 907), as described in the literature [2,28] but in a shorter time, whereas the shea butter was composed primarily of SOO ([M + NH4 ]+ = 905) and SOS [2]. Moreover, on PGC, the resolution between SOO ([M + NH4 ]+ = 905) and POS peaks in cocoa butter was higher than in NARP-LC on ODS [30]. As shown in the two figures, ELSD gives a more exact fingerprint due to its universal response, whereas SM enhances minor peaks. The compound eluted at 18.8 min in each butter in Fig. 4 corresponds to SOA ([M + NH4 ]+ = 935). In order to unambiguously identify species, characteristic fragment ions

Fig. 5. Plot of ln k against ECN values for saturated and unsaturated TAGs (fatty acid with one double bond). Variation according to alkyl chain (♦); variation according to unsaturation (). Column temperature: 60 ◦ C. Column: Hypercarb (150 mm × 4.6 mm I.D. for POP, POS, SOS, SOA, SOO) and Hypercarb (150 mm × 2.1 mm I.D.) for OOO and SSS.

The PGC column presents a great potential for TAG separation and the behaviour of these analytes on the PGC is new in comparison to NARP-LC. Moreover, the retention of TAGs on PGC is greatly affected by temperature and decreases when the column temperature is increased, enabling a rapid analysis. According to these preliminary results, the chromatographic system consisting of a PGC column and a mobile phase composed of CHCl3 –iPrOH could be a suitable system to quickly separate TAGs in high temperature conditions coupled with ELSD or MS detector. References [1] EU Directive 2000/36/CE of 23 June 2000, Off. J. Eur. Communities: Legis, (2000) 197. [2] M. Lipp, E. Anklam, Food Chem. 62 (1998) 73. [3] F. Dionisi, P.A. Golay, B. Hug, M. Baumgartner, P. Callier, F. Destaillats, J. Agric. Food Chem. 52 (2004) 1835. [4] M. Lafosse, B. Herbreteau, in: Z. El Rassi (Ed.), Carbohydrate Analysis by Modern Chromatography and Electrophoresis. J. Chromatogr. Library, second ed., Elsevier Science, Amsterdam, 2002, p. 1101. [5] K. Nagy, D. Bongiorno, G. Avellone, P. Agozzino, L. Ceraulo, K. Vekey, J. Chromatogr. A 1078 (2005) 90. [6] K. Nagy, A. Jakab, J. Fekete, K. Vekey, Anal. Chem. 76 (2004) 1935. [7] A. Jakab, E. Forgacs, Chromatographia 56 (2002) S69. [8] A. Larsen, P. Molander, J. Sep. Sci. 27 (2004) 297. [9] P. Molander, A. Holm, E. Lundanes, T. Greibrokk, E. Onimundsen, J. Sep. Sci. 23 (2000) 653. [10] M.C. Pietrogrande, A. Benvenuti, F. Dondi, Chromatographia 51 (2000) 193. [11] H. Knox, P. Ross, Adv. Chromatogr. 37 (1997) 73. [12] P. Ross, LC–GC Europe 13 (2000) 310. [13] C. Viron, P. Andr´e, M. Dreux, M. Lafosse, Chromatographia 49 (1999) 137. [14] K. Gaudin, P. Chaminade, A. Baillet, J. Chromatogr. A 973 (2002) 69. [15] F. Deschamps, K. Gaudin, A. Baillet, P. Chaminade, J. Sep. Sci. 27 (2004) 1313. [16] S. Roy, A. Delobel, K. Gaudin, D. Touboul, D. Germain, A. Baillet, P. Prognon, O. Lapr´evote, P. Chaminade, J. Chromatogr. A 1117 (2006) 154. [17] K. Gaudin, P. Chaminade, D. Ferrier, A. Baillet, Chromatographia 50 (1999) 470. [18] S. Cassel, P. Chaimbault, C. Debaig, T. Benvegnu, S. Claude, D. Plusquellec, P. Rollin, M. Lafosse, J. Chromatogr. A 919 (2001) 95. [19] M. Buchgraber, F. Ulberth, H. Emons, E. Anklam, Eur. J. Lipid Sci. Technol. 106 (2004) 621. [20] H. Colin, N. Ward, G. Guiochon, J. Chromatogr. 149 (1978) 169.

466

B. Merelli et al. / J. Chromatogr. A 1157 (2007) 462–466

[21] K. Gaudin, P. Chaminade, A. Baillet, J. Sep. Sci. 27 (2004) 41. [22] A. Kukis, L. Marai, J.J. Myher, J. Chromatogr. 588 (1991) 73. [23] S. Heron, J. Bleton, A. Tchapla, in: J.-L. Sebedio, E.G. Perkins (Eds.), News Trends in Lipid, Lipoprotein Analyses, AOCS Press, Champagne, Illinois, 1995, p. 205. [24] H.R. Mottram, S.E. Woodbury, R.P. Evershed, Rapid Commun. Mass Spectrom. 11 (1997) 1240. [25] A. Jakab, K. Heberger, E. Forgacs, J. Chromatogr. A 976 (2002) 255. [26] M. Holcapek, P. Jandera, P. Zderadicka, L. Hruba, J. Chromatogr. A 1010 (2003) 195.

[27] [28] [29] [30]

W.E. Neff, W.C. Byrdwell, J. Am. Oil Chem. Soc. 72 (1995) 1185. W.E. Neff, W.C. Byrdwell, J. Liq. Chromatogr. 18 (1995) 4165. H.R. Mottram, R.P. Evershed, Tetrahedron Lett. 37 (1996) 8593. J. Cvaˇcka, O. Hovorka, P. Jiroˇs, J. Kindl, K. Str´ansky, I. Valterov´a, J. Chromatogr. A 1101 (2006) 226. [31] P. Favetta, A. Sevestre, P. Chaimbault, M. Lafosse, Rapid characterisation of cocoa and shea butters by LC, MS and LC–MS, poster presentation, 1st International Symposium on Recent Advances in Food Analysis, 5–7 November 2003, Prague, Czech Republic.