Ulb

MGa

b

a

ARAA

KUcTNEBO

1

ddBeoBphAhbd

m

h1

Journal of Chromatography B, 983–984 (2015) 94–100

Contents lists available at ScienceDirect

Journal of Chromatography B

jou rn al hom epage: www.elsev ier .com/ locate /chromb

ltrahigh performance supercritical fluid chromatography ofipophilic compounds with application to synthetic and commercialiodiesel

. Ashraf-Khorassania, J. Yangb, P. Rainvilleb, M.D. Jonesb, K.J. Fountainb,. Isaacb, L.T. Taylora,∗

Department of Chemistry, Virginia Tech, Blacksburg, VA 24061, United StatesWaters Corporation, 34 Maple Street, Milford, MA 01757, United States

r t i c l e i n f o

rticle history:eceived 2 September 2014ccepted 14 December 2014vailable online 6 January 2015

eywords:ltrahigh performance supercritical fluidhromatographyobacco seed oilormal phase chromatography

a b s t r a c t

Ultrahigh performance supercritical fluid chromatography (UHPSFC) in combination with sub-2 �m par-ticles and either diode array ultraviolet (UV), evaporative light scattering, (ELSD), or mass spectrometric(MS) detection has been shown to be a valuable technique for the determination of acylglycerols in soy-bean, corn, sesame, and tobacco seed oils. Excellent resolution on an un-endcapped single C18 column(3.0 mm × 150 mm) with a mobile phase gradient of acetonitrile and carbon dioxide in as little as 10 minserved greatly as an improvement on first generation packed column SFC instrumentation. Unlike highresolution gas chromatography and high performance liquid chromatography with mass spectrometricdetection, UHPSFC/MS was determined to be a superior analytical tool for both separation and detec-tion of mono-, di-, and tri-acylglycerols as well as free glycerol itself in biodiesel without derivatization.

vaporative light scattering detectioniodiesel purity testcta-decyl bonded silica particles

Baseline separation of residual tri-, di-, and mono-acylglycerols alongside glycerol at 0.05% (w/w) waseasily obtained employing packed column SFC. The new analytical methodology was applied to bothcommercial B100 biodiesel (i.e. fatty acid methyl esters) derived from vegetable oil and to an “in-house”synthetic biodiesel (i.e. fatty acid ethyl esters) derived from tobacco seed oil and ethanol both before andafter purification via column chromatography on bare silica.

© 2015 Elsevier B.V. All rights reserved.

. Introduction

The development of new technologies that afford the pro-uction of fuels that are obtained from renewable resources isriven by both environmental concerns and fossil fuel deficiency.iofuels, such as the biodiesel produced via base catalyzed trans-sterification of vegetable oils and animal fats using either ethanolr methanol in the U.S. is a promising alternative fuel source [1].iodiesel has similar chemical structure and energy content toetro-diesel in terms of fuel quality. Compared to petroleum diesel,owever, biodiesel can reduce CO2 emissions approximately 78%.s an energy source, biodiesel (i.e. fatty acid alkyl esters) should

ave certain criteria [2]. The feed-stocks available for producingiodiesel are highly dependent upon the region, climate, soil con-itions, and geography. For example, rapeseed is the dominant

∗ Corresponding author. Tel.: +1 540 231 6680; fax: +1 540 232 3255;obile: +1 540 239 3944.

E-mail address: ltaylor@vt.edu (L.T. Taylor).

ttp://dx.doi.org/10.1016/j.jchromb.2014.12.012570-0232/© 2015 Elsevier B.V. All rights reserved.

feedstock for biodiesel production in Europe, soybean in the UnitedStates and palm oil in tropical countries such as Malaysia and thePhilippines. Coconut is another feedstock used for biodiesel pro-duction in the Pacific Rim region.

The advantages of biodiesel in addition to being consideredmore environmentally friendly than petro diesel are higher octanenumber and higher combustion efficiency. Biodiesel is usuallycommercially blended with petro-diesel to 5% biodiesel. A majorproblem related to the use of biodiesel is its inherent instability tooxidation. For example, biodiesel degradation can change the fuelproperties by formation of gums, acids, insolubles, and other oxi-dation products. The level of impurities in commercial biodieselB100 as cited in ASTM D 6751 for (a) tri-acylglycerol which is unre-acted starting material, (b) di-acylglycerol and mono-acylglycerolwhich are reaction intermediates, and (c) free glycerol which is aby-product should be 0.02 wt% for glycerol itself and 0.24 wt% total

glycerol which includes mono-, di-, and tri-acylglycerols as wellas glycerol itself. It is interesting to note that each year produc-ers of biodiesel create more than a million tons of “just” glycerolworldwide much of which goes to waste [3]. The presence and

hroma

cu

yi(ghtGfttlptn[dcpd

ottpaeaeftonntwtatswle

luiBictmmaeaaagHbdUf

M. Ashraf-Khorassani et al. / J. C

oncentration of these impurities obviously lead to biodiesel prod-cts that vary in complexity, polarity, solubility, and volatility.

Analysis of acylglycerols typically follows two routes: (1) anal-sis of the whole molecule which can be a mixture of glyceroltself, mono-acylglycerol, di-acylglycerol, and tri-acylglycerol and2) analysis of the range of fatty acids that originally esterify thelycerol molecule [4]. Analyses of whole acylglycerol moleculesave been reported in the literature and in ASTM D 6584. Suchechniques use GC, HPLC, TLC or a combination of these [5]. MostC techniques employ flame ionization detection; while HPLC

requently uses evaporative light scattering [6] and mass spec-rometric [7] detection. For glycerol, mono-, and di-acylglycerols,he analytical method is based on conversion of the molecu-ar hydroxyl function into silyl derivatives in the presence ofyridine and N-trimethylsilyltrifluoroacetamide followed by highemperature GC on a short capillary column with thin film thick-ess using on-column injection and flame ionization detection4]. Most techniques, on the other hand, dedicated to HPLC areesigned to deal with tri-acylglycerols (TAGs). In most of theseases, the methodology employs multi-dimensional chromatogra-hy wherein argentation chromatography is one of the componentsesigned to achieve separation of unsaturated acylglycerols

Packed column supercritical fluid chromatography (SFC), on thether hand, can be essentially considered to be a more powerfulool for lipid analyses [8] than either HRGC or HPLC. In comparisono GC, acylglycerols are separated via SFC at a much lower tem-erature and compared to HPLC, different selectivity and shorternalysis times are obtained with SFC. Contrary to numerous refer-nces that have appeared in the older literature, polar solutes suchs free glycerol can be analyzed via SFC without derivatization. Forxample, TAGs have been analyzed by packed column SFC usingor example 2-ethylpyridine [9] and silver-ion exchange [10] sta-ionary phases. Using a reversed phase stationary phase such asctadecylsilica [11], the SFC separation may be based upon carbonumber (i.e. total number of carbons in all fatty acids) and/or on theumber of double bonds. In a more recent reversed phase separa-ion [12], three Zorbax SB-C18 columns (4.6 mm × 250 mm, 5 �m)ere coupled in series. Temperature was 25 ◦C with an acetoni-

rile/methanol modifier gradient. With a flow rate of 2.5 mL/min, mixture of eight TAGs eluted between 45 and 60 min, althoughhey were not baseline resolved. On the other hand, separation ofelected unsaturated vegetable oils on the silver loaded columnas mainly based on the number of double bonds. Evaporative

ight scattering detection results were reproducible and providednhanced sensitivity compared to UV detection.

An improvement in chromatographic performance for similaripophilic analytes is reported here. Fast determination of resid-al glycerol, free fatty acids, and mono-, di-, and tri-acylglycerols

n both synthetic biodiesel prepared “in-house” and commercial100 biodiesel (e.g. typically referred to as fatty acid alkyl esters)

s described here using ultrahigh performance supercritical fluidhromatography (UHPSFC). UHPSFC is a relatively new separationechnique that uses compressed carbon dioxide as the primary

obile phase [13]. It takes advantage of sub-2 �m particle chro-atography columns and advanced chromatography systems that

re designed to achieve fast and reproducible separation with highfficiencies and unique selectivity. The established UHPSFC/MSpproach has potential application in lipidomics and food testings a complementary method alongside HPLC–MS and HRGC–MS,s the former can separate both polar and non-polar lipids in a sin-le run to improve both detection limits and peak shape. UnlikeRGC/MS, low volatile and very long chain fatty acids (>24 car-

on atoms) can be easily analyzed without concern for analyteegradation. In this regard, the possibilities for lipid analyses viaHPSFC were recently illustrated by Jones et al. [14] who per-

ormed the separation of neutral and amphipathetic lipids using

togr. B 983–984 (2015) 94–100 95

both BEH (unbonded organic/inorganic silica hybrid) and HSS C18(un-endcapped silica) columns packed with sub-2 �m particles.

Studies using UHPSFC coupled with photodiode array (DAD),mass spectrometry (MS), and evaporative light scattering (ELSD)detection for the separation of TAGs isolated from tobacco, corn,sesame, and soybean seed oils are presented here. A single un-endcapped C18 column with acetonitrile-modified carbon dioxidewas employed for separation of all seed oils. Excellent chro-matography of associated, hydroxyl-containing di-acylglycerols,mono-acylglycerols, and glycerol wherein methanol was employedas a secondary modifier in addition to acetonitrile was alsoachieved.

2. Experimental

2.1. Sample description

Pure fatty acid ethyl esters (C16, C18, C18:1, C18:2, andC18:3) were purchased from Sigma–Aldrich (St. Louis, MO) andmixed to form a model biodiesel. Pure mono-, di-, and tri-octadecylglycerol plus free glycerol and soybean oil were alsoobtained from Sigma–Aldrich. All standards were prepared ina 50/50 dichloromethane/methanol (DCM/MeOH) mixture. Themodel biodiesel was prepared in the same mixture as a 5% (v/v)solution. Tobacco seed oil was obtained from R.J. Reynolds TobaccoCo. (Winston-Salem, NC). Soybean oil, corn oil, and sesame seed oilwere obtained from Sigma–Aldrich (St. Louis, MO). Five percent ofthe different oils was dissolved in DCM/MeOH (1/1, v/v) for bothUV and ELS detection and then 0.1% for the Xevo G2 QTof MS.

2.2. Chromatographic conditions

Supercritical fluid chromatography experiments were per-formed using a Waters Acquity UPC2 system (Milford, MA, USA)equipped with (a) high pressure mixing binary solvent deliverymanager, (b) fixed loop design autosampler, (c) active back pres-sure regulator, (d) column compartment with active heating andcolumn switching control, (e) photodiode array (PDA) and evap-orative light scattering (ELS) detectors. Experiments were carriedout using an ACQUITY UPC2 HSS C18 SB column (150 mm × 3.0 mm,1.8 �m) at a temperature of 25 ◦C. The mobile phase consisted ofcompressed CO2 (component A) and either 100% acetonitrile oracetonitrile/methanol (90:10) (component B). The mobile phaseflow rate was maintained at 1–2 mL/min under all different gra-dient conditions. Backpressure was maintained isobarically andautomatically by a backpressure regulator (ACQUITY CCM) at apressure of 1500 psi. The injection volume was varied 2–8 �L forboth PDA and ELSD; while, 0.1 �L was the volume for MS detection.Photodiode array detection was monitored at a wavelength range190–400 nm with a reference of 400–500 nm. The Waters ACQUITYELSD detector was operated with nebulizer cooling, drift tube:50 ◦C, gas pressure: 40 psi and gain 10, make up flow (isopropylalcohol) was added at 0.2 mL/min before the ELSD. The solvent flowwas split prior to the back pressure regulator for ELSD (split ratio1:3) and MS electrospray probe.

2.3. Mass spectrometric conditions

Mass spectrometry was performed using Xevo G2 QTof (WatersCorp., Milford, MA, USA). The solvent flow was split post PDAdetection using a pre-BPR flow Upchurch cross 1/16 PEEK splitter.CO2-miscible make-up solvent (MeOH in 10 mM NH4OAc), deliv-

ered by a HPLC 515 make-up pump (Waters Corp., Milford, MA,USA), was added at a flow rate of 0.2 mL/min and mixed withthe chromatographic effluent to aid ionization. A fraction of thetotal flow was directed from the union to the ESI source through

9 hromatogr. B 983–984 (2015) 94–100

att3v1tp

3

3

asattboflp(tdfh

mctoAalcmcdc

TDTbdsbriMepmp28apisaia

Table 1List of detected TAGs in each seed oil with their retention time via UHPSFC/MS.

Ret. time (min) Soybean oil Corn oil Sesame oil Tobacco seed oil

10.86 LnLnL – – –11.11 LnLL LnLL LnLL LnLL11.40 LLL LLL LLL LLL11.60 OLLn OLLn OLLn –11.78 PLLn – –11.90 SLLn SLLn SLLn SLLn12.11 OLL OLL OLL OLL12.32 POLn POLn POLn –12.49 OOL OOL OOL OOL12.71 POL POL POL POL13.00 PPL PPL PPL PPL13.08 OOO OOO OOO –13.22 SLL SLL SLL SLL13.40 POO POO – POO13.93 SOL SOL SOL SOL14.26 PSL PSL PSL PSL14.73 SSL SSL SSL SSL15.14 PSO PSO PSO –

6 M. Ashraf-Khorassani et al. / J. C

transfer line, whereas the remaining mobile phase was directedo the BPR PEEK connection. The ESI source was operated in posi-ive ionization mode with capillary and cone voltages of +3 kV and0 V, respectively. The source temperature, cone gas flow, desol-ation temperature, and desolvation gas flow were set at 150 ◦C,0 L/h, 500 ◦C and 600 L/h, respectively. MS data were acquired inhe range of 400–1200. Data handling and instrument control wereerformed with Masslynx 4.1 (Waters Corp., Milford, MA, USA).

. Results and discussion

.1. UHPSFC of tri-acylglycerols

SFC is useful for the separation of lipids such as carotenoidsnd tri-acylglycerols (TAGs) which have numerous analogs withimilar structures. Currently, there is much interest in rapid char-cterization of TAGs which are natural compounds produced byhe esterification of glycerol with fatty acids. While first genera-ion supercritical fluid chromatographic (SFC) instrumentation haseen successfully applied to the characterization of TAGs in vari-us seed oils, the application of ultrahigh performance supercriticaluid chromatography (UHPSFC) to the range of byproducts that areroduced during the conversion of TAGs to fatty acid alkyl estersi.e. biodiesel) has not been considered. Separation and detection ofhe wide analyte polarity range afforded by underivatized mono-,i-, and tri-acylglycerols, glycerol, fatty acid alkyl esters, and freeatty acids by UHPSFC with three different detectors is discussedere.

Fig. 1 shows the UV (i.e. diode array), ELSD, and MS chro-atograms for the separation of TAGs in soybean oil. All of the

hromatographic effluent first passes through the DAD flow cellhen via a 3/1 split to the back pressure regulator and either ELSDr MS. Each gradient separation was performed at 25◦C with anCQUITY UPC2 HSS C18 SB column (HSS = high strength silica) and

mobile phase of acetonitrile-modified carbon dioxide. Near base-ine peak resolution was observed with a single 3.0 mm × 150 mmolumn packed with 1.8 �m bonded silica particles in approxi-ately 16 min. Baseline stability was excellent even under gradient

onditions. The resulting chromatographic system allowed repro-ucible separations and possible quantification of analytes at lowoncentrations in various samples.

These optimized conditions were used to separate and profileAGs in tobacco seed oil, soybean oil, corn oil, and sesame seed oil.ata were acquired in UHPSFC-MSE mode, which is an unbiasedof acquisition method in which the mass spectrometer switchesetween low and high energy on alternate scans for structural eluci-ation and identification. In all cases, distinct profiles and excellenteparations were obtained for all oil types when using UHPSFC withoth MS and ELSD detection. Fig. 2 shows the UHPSFC/MS sepa-ation and detection of various TAGs in different oils. TAGs weredentified using accurate mass spectra collected by QTof MS with

SE and the Waters Progenesis QI. In positive ion mode MSE lownergy, TAGs produced intact ammonium ion adducts [M+NH4]+

recursor exact mass when ammonium acetate was present in theake-up solvent. In MSE high energy, abundant fragment ions were

roduced corresponding to the neutral loss of one of the sn-1, sn- or sn-3 fatty acids plus ammonia. For example, the ion at m/z74.7823 corresponding to 52:3 TAG (calculated fatty acid carbontom: total number of double bonds), can be identified as POL (i.e.almitic–oleic–linoleic) due to the presence of abundant fragment

ons at m/z 575.5038, 577.5201 and 601.5186. These ions corre-pond to the neutral loss of fatty acyl groups P, O, and L plusmmonia, respectively [4]. Table 1 shows a list of all identified TAGsn the different oils based on the low energy precursor exact massnd corresponding high energy fragment ions.

15.70 PSA PSA PSA –

– not detected.

3.2. Feasibility of impurities assay

A single ACQUITY HSS C18 SB column (3.0 mm × 150 mm) packedwith 1.8 �m particles was used to develop the separation of amodel mixture of anticipated biodiesel components such as freeglycerol, acylglycerols, and fatty alkyl esters. When a faster gra-dient of modifier was employed, all the components were elutedin less than 9 minutes without a significant loss in resolution evenamong the mono-, di-, and tri-acylglycerols. Application of a steepermodifier gradient ramp in mobile phase under the same chromato-graphic conditions further decreased the analysis time to 4 minwith again minimal loss in peak resolution. These data indicatedthat the gradient profile, flow rate, and mobile phase modifier canbe readily changed in order to optimize the separation regard-less of the biodiesel analysis required. Furthermore, all the listedparameters are compatible with MS detection, such that posi-tive identification of all acylglycerol species plus glycerol itselfis possible. Finally a faster separation without significant loss inanalyte resolution could translate into higher sample throughputwhen monitoring product quality. Equally important is the obser-vation that baseline separation of all compounds was achieved. Itis also imperative that one must also use a method whereby theimpurities do not interfere with proper quantification of resid-ual acylglycerols in the sample. This situation can become criticalsince most biodiesels have minor components which elute earlybut after elution of the fatty acid ethyl esters (i.e. biodiesel com-ponents). For selected analyses, flow was lowered from 1.2 to1.0 mL/min to minimize these interferences. It should be notedthat reproducibility was excellent for all analytes. To demonstratethis feature, five replicate chromatographic injections of our modelbiodiesel mixture that contained the four spiked impurities usingELSD were performed. Percent RSD for all peak areas was between1 and 4%, and the RSD for retention time of all peaks was lessthan 0.1%.

Next a series of model biodiesel mixtures were preparedwherein the compounds varied in concentration. In each case,FAEE, soybean oil TAGs, and free glycerol were mixed at differ-ent concentrations and the resulting mixture was injected into theACQUITY HSS C18 SB column (3.0 mm × 150 mm). A series of chro-matograms were thus generated wherein the spiking percentage

varied from 0.2% to 5.0% (v/v). At the lowest concentration (0.2%),the glycerol peak was clearly seen and well resolved from the othercomponent peaks for quantification purposes. In order to estimatesystem detection limit, larger injection volumes (4 and 8 �L) of the

M. Ashraf-Khorassani et al. / J. Chromatogr. B 983–984 (2015) 94–100 97

Fig. 1. Sequential UHPSFC of soybean oil with UV, ELSD, and MS detection using CO2 with CH3CN as modifier. Chromatography conditions: gradient elution: 2–20% CH3CNin 18 min. Flow: 1.5 mL/min, 25 ◦C. Column ACQUITY UPC2 HSS C18 SB column (150 mm × 3.0 mm, 1.8 �m).

Fig. 2. UHPSFC/MS of different oils using CO2 with CH3CN as modifier. Gradient elution: 2–20% CH3CN in 18 min. Flow: 1.5 mL/min, 25 ◦C. Column ACQUITY UPC2 HSS C18 SBcolumn (150 mm × 3.0 mm, 1.8 �m).

98 M. Ashraf-Khorassani et al. / J. Chromatogr. B 983–984 (2015) 94–100

Fig. 3. UPSFC/ELSD detection of glycerol at 0.05% level at two different injection volume (4 and 8 �L). Chromatography conditions: T = 0 min, 98/2 CO2/modifier (90/10CH3CN/CH3OH), T = 10 min, 80/20, T = 12 min, 60/40, T = 15 min, 60/40, flow = 1.2 mL/min., 25 ◦C. Column ACQUITY UPC2 HSS C18 SB column (150 mm × 3.0 mm, 1.8 �m)volume.

F ondit9 B colu

mpwu

ig. 4. UHPSFC/ELSD of 3 different batches of synthetic biodiesel. Chromatography c0/10, T = 12 min, 90/10, flow = 1.0 mL/min., 25◦ C. Column ACQUITY UPC2 HSS C18 S

ixture spiked with impurities at 0.05% of glycerol yielded com-arable results (Fig. 3). The S/N ratio for glycerol using ELSDas about 10/1 at the 0.05% level with a 4 �L injection vol-me which was equivalent to 2 �g mass injected and 40/1 at the

ions: T = 0 min, 90/10 CO2/modifier, T = 10 min, 50/50, T = 11 min, 50/50, T = 11.1 min,mn (150 mm × 3.0 mm, 1.8 �m).

same level with an 8 �L injection which was equivalent to 4 �ginjected on-column. The detection limit demonstrated here viathis method is compatible with ASTM 6751for biodiesel impuritiesanalysis.

M. Ashraf-Khorassani et al. / J. Chromatogr. B 983–984 (2015) 94–100 99

Fig. 5. Injection of commercial biodiesel B100 and synthetic biodiesel spiked with different impurities. Chromatography conditions: T = 0 min, 90/10 CO2/modifier, T = 10 min,50/50, T = 11 min, 50/50, T = 11.1 min, 90/10, T = 12 min, 90/10, flow = 1.0 mL/min., 25 ◦C. Column ACQUITY UPC2 HSS C18 SB column (150 mm × 3.0 mm, 1.8 �m).

Fig. 6. UHPSFC/ELSD separation of synthetic biodiesel before and after purification. Chromatography conditions: T = 0 min, 90/10 CO2/modifier, T = 10 min, 50/50, T = 11 min,50/50, T = 11.1 min, 90/10, T = 12 min, 90/10, flow = 1.0 mL/min., 25◦ C. Column ACQUITY UPC2 HSS C18 SB column (150 mm × 3.0 mm, 1.8 �m).

1 hroma

tihltattnt(dwTenamirhtmtg

3

dhCQotmtcotlctBrmdo

aoigarMusw

[

[

[

[

00 M. Ashraf-Khorassani et al. / J. C

This study yielded a surprising result regarding the retentionimes of the various components in the model mixture. An increasen amount of soybean oil tri-acylglycerols injected from 0.2% to 5.0%ad little or no effect upon retention time of specific lipophilic ana-

ytes. The result insofar as glycerol was quite different. Retentionime of glycerol for 4 �g injected was 14.63 min. An increase inmount injected from 4 �g to 100 �g injected caused the retentionime to reproducibly go from14.63 min to 14.57 min to 14.44 mino 14.28 min. Since all of the separations in the total study wereormal phase as evidenced by the observation that analyte reten-ion increased with the number of non-derivatized hydroxyl groupsi.e. elution order is tri-acylglycerol with no hydroxyls, followed byi-acylglycerols with one hydroxyl, followed by mono-acylglycerolith two hydroxyls, followed by glycerol with three hydroxyls).

hus glycerol appears to become less polar as the mass of glyc-rol increases in the experiment. Since the mass injected haso effect on the soybean acylglycerol retention tine, it wouldppear that the effect with glycerol is unrelated to a change inobile phase physical properties or to a change from supercrit-

cal to subcritical conditions. On the other hand, it does seemeasonable that a highly polar molecule like glycerol with threeydroxyl groups would exhibit greater intermolecular interac-ion at a greater analyte concentration in a basically nonpolar

edium of 98% CO2. In other words, the decrease in retentionime could be explained by a reduction in effective polarity forlycerol.

.3. Biodiesel purity test

The above method was used to determine the purity of threeifferent synthetic biodiesels from tobacco seed oil prepared “in-ouse” (submitted for publication: M. Ashraf-Khorassani, W.M.oleman III, M.F. Dube, L.T. Taylor, “Synthesis, Purification, anduantification of Fatty Acid Ethyl Esters After trans-Esterificationf Large Batches of Tobacco Seed Oil”). Due to the complexity ofhe biodiesel, the method was further optimized by lowering the

obile phase flow rate from 1.2 to 1.0 mL/min in order to enhancehe separation of impurities that eluted after biodiesel (Fig. 4). Asan be observed, all three biodiesel samples showed the presencef multiple impurities as well as mono-acylglycerols at a reten-ion time of 6.2 min. These impurities were highest in batch 2 andeast in batch 3. For comparison, B100 biodiesel from a commer-ial source was obtained and also analyzed, Fig. 5A. It is interestingo note that mono-acylglycerols were observed in the commercial100 biodiesel based on comparison of retention time. Other impu-ities, however, were not detected. Fig. 5B shows the separation ofodel (i.e. pre-mixed fatty acid alkyl esters) biodiesel spiked with

ifferent, but known, impurities for comparison. No quantificationf impurities was attempted in this study

In order to remove the impurities from synthesized biodiesel, simple two-step column chromatography process was devel-ped. In this process, tobacco seed oil biodiesel with associatedmpurities was passed through a bare silica column and eluted viaravity first using hexane and followed by elution of the remainingnalytes using ethanol. Solvent from each fraction was evapo-ated and the proper amount of each sample was dissolved in

eOH/DCM (1/1, v/v). Analysis of the fractions was performed

sing the UHPSFC method developed previously. Fig. 6 shows theeparation of biodiesel (a) before purification, (b) after purificationhen hexane was the eluting solvent, and (c) when ethanol was

[

togr. B 983–984 (2015) 94–100

the initial eluting solvent. The results suggested that nearly puresynthetic biodiesel was obtained in the hexane fraction. Mono-acylgycerols were again easily separated and detected in thesebiodiesel samples using UHPSFC/ELSD. See Fig. 6 caption for silicachromatography conditions.

4. Conclusions

UHPSFC with ELSD was determined to be much faster and easierthan either high temperature GC or GC with preliminary derivatiza-tion as an analytical tool for determination of biodiesel impurities.A complete separation of tri-acylglycerols, di-acylglycerols, mono-acylglycerols, and free glycerol from model biodiesel was obtainedin less than 15 min. Faster separation was obtained by increasingthe gradient elution. The ELSD S/N ratio for glycerol for an injectedmass of 2 �g was about 10/1. This level of detection could easilybe lowered with higher injection volume (10 �L) to be compatiblewith ASTM D 6751 for detection and analysis of glycerol (0.02%).Low volatile and very long chain fatty acids (>24 carbon atoms)could be easily analyzed with UHPSFC although it was not demon-strated here. With GC, high MW components or impurities mayelute in the same chromatographic region as the mono-, di-, andtri-acylglycerols, which can cause poor accuracy in quantification.With UHPSFC, complete baseline separation of mixtures of FAME,FFA, mono-, di-, and tri-acylglycerols and free glycerol affords quan-titative results with good accuracy and precision.

References

[1] J.K. Rodriguez-Guerrero, R.M. Filho, P.T.V. Rosa, Production of biodiesel fromcastor oil using sub and supercritical ethanol: effect of sodium hydroxide onthe ethyl ester production, J. Supercrit. Fluids 83 (2013) 124–132.

[2] J.W. Diehl, F.P. DiSanzo, Determination of total biodiesel fatty acid methylethyl esters, and hydrocarbon types in diesel fuels by supercritical fluidchromatography-flame ionization detection, J. Chromatogr. Sci. 45 (2007)690–693.

[3] http://cenm.ag/two14[4] S.C. Moldoveanu, Y. Chang, Dual analysis of triglycerides from certain common

lipids and seed extracts, J. Agric. Food Chem. 59 (2011) 2137–2147.[5] J.L. Bernal, M.T. Martin, L. Toribio, Supercritical fluid chromatography in food

analysis, J. Chromatogr. A 1313 (2013) 24–36.[6] M. Buchgraber, F. Ulberth, E. Anklam, Comparison of HPLC and GLC techniques

for the determination of the triglyceride profile of cocoa butter, J. Agric. FoodChem. 48 (2000) 3359–3363.

[7] C.A. Dorschel, Characterization of the TAG of peanut oil by electrosprayLC–MS–MS, J. Am. Oil Chem. Soc. 79 (2002) 749–753.

[8] T. Bamba, J.W. Lee, A. Matsubara, E. Fukusaki, Metabolic profiling of lipids bysupercritical fluid chromatography/mass spectrometry, J. Chromatogr. A 1250(2012) 212–219.

[9] E. Lesellier, A. Tchapla, Retention behavior of triglycerides in octadecyl packedsubcritical fluid chromatography with CO2/modifier mobile phases, Anal.Chem. 71 (1999) 5372–5378.

10] Q. Zhou, B. Gao, X. Zhang, Y. Xu, H. Shi, L. Yu, Chemical profiling of tri-acylglycerols and di-acylglycerols in cow mile fat by ultra-performanceconvergence chromatography combined with a quadrupole time-of-flight massspectrometry, Food Chem. 143 (2014) 199–204.

11] A. Dermaux, M. Medvedovici, E. Ksir, M. Van Hove, P. Talbi, Sandra, Eluci-dation of the triglycerides in fish oil by packed column supercritical fluidchromatography fractionation followed by capillary electrochromatographyand electrospray mass spectrometry, J. Microcol. Sep. 11 (1999) 451–459.

12] E. Lesellier, A. Tchapla, Supercritical fluid chromatography with packedcolumns, techniques and applications, in: K. Anton, C. Berger (Eds.), Chro-matographic Science Series, vol. 75, Marcel Dekker Inc., New York, 1998, p.195.

13] A.G.-G. Perrenoud, J.-L. Veuthey, D. Guillarme, Comparison of ultra-high perfor-

mance supercritical fluid chromatography and ultra-high performance liquidchromatography for the analysis of pharmaceutical compounds, J. Chromatogr.A 1266 (2012) 158–167.

14] M.D. Jones, G. Isaac, G. Astarita, A. Aubin, J. Shockor, V Shulaev, Lipid ClassSeparation Using UPC2/MS, Waters Corp., Application Note 720004579.