and may involve a disruption of the solvent (COa) shell around the dissolved .... bility, 1.5 g of cholesterol (J.T. Baker, ..... of trace chemicals in lipid-rich samples.
journal of Chromatographic Science, Vol. 35, October 1997
The Effectof DissolvedHelium on the Densityand SalvationPowerof SupercriticalCarbon Dioxide Zhouyao Zhang* and Jerry W. King Food Quality and Safety Research, National Center for Agricultural Utilization Research, Agricu!Jural Research Service, United States
Department of Agriculture+, 1815 North University Street, Peoria, IL 61604
Carbon dioxide (CO*) cylinders pressurized with helium in their headspace are widely used in supercritical fluid chromatography and extraction (SFCand SFE).A few percent of helium are dissolved in the liquid CO2 phase of the cylinder, which can result in such problems as retention time shift and poor reproducibility in SFCas well as reduced solubility and extraction rate in SFE.In this study, a high precision density meter and a gas chromatograph equipped with a thermal conductivity detector are used to monitor the density and composition of the supercritical fluid generated from a helium headspace CO2 cylinder over the duration of its use. These measurements are related to the solubility of soybean oil and cholesterol in the fluid. The density measurements (accurate to 1V g/ml) show that the density of the fluid is linearly proportional to the helium content of the CO*. However, the significant drop of solute solubility in helium-CO, mixtures cannot be explained by the reduction in fluid density alone and may involve a disruption of the solvent (COa) shell around the dissolved solute.
Introduction
Carbondioxide(C02)is probablythemostoftenusedfluidin supercriticalfluid chromatography (SFC)and supercritical fluidextraction(SFE).Thecompressibility of CO1is quitehigh when it is deliveredfrom a cylinder to a pump at room temperature,resultingin poorpumpingefficiency. Toreduce compressibility andimprovepumpingefficiency,COemustbe deliveredto thepumpeitherat highpressure(about2000psi) or at a temperature muchlowerthan25°C.Thelatterrequires thecoolingof thepumpheador pumpcylinder.Theformeris oftenrealizedby usinga heliumheadspace (HHS)insidethe CO2cylinder.Theuseof heliumheadspace COsin SFC,however,leadsto poorreproducibilityin peakretentiontimesand peakareas(l-3). In recentyears,researchers havealsonoticed that HH$ CO2producesadverseeffectsin SFEaswell.For example,thesolubilityof soybean oil in HHSCOsis substantially *Author to whom correspondence should be addressed. + Names are necessary to report factually on available data; however, the USDA neither guarantees nor warrants the standard of the product, and the use of the name by USDA implies no approval of the products to the exclusion of others that may also be suitable.
lowerthanthat in pureCOz(4).HHSCO2alsoreducestheextractionrateof cholesterol(5) andaffectspolymerizationreactionsin supercriticalCO2(6).Theproblemsassociated with HHSCO2areattributedto a smallpercentage of heliumdissolvedin liquid COs,which has’beenconfirmedby recent studies(3,7).Howdissolved heliumaffectsthefluiddensityand solutesolubilityin a supercriticalfluid is still not clear.In fact,to our knowledge, the densityof the HHSCOzhasonly beentheoreticallycalculated(2) but not experimentally measured. Our interestin HHSCO2stemsfrom the observation that thesolubilityof soybean oil is 50-70%lowerin HHSCO2than that in pureCO2at thesametemperatureandpressure(4).To betterunderstandthe reasonbehindthis observation, weutilizedan Anton Parr high-pressure densitymetercapableof measuring density accurate to 1O-4g/cm3, and a gas chromatograph(CC)equippedwith a thermal conductivity detector(TCD)to monitorthe compositionanddensityof the supercriticalfluid fromanHHScylinderduringSFEuntil depletionof the cylindercontents.Thesedatawerethenrelated to the solubilitiesof soybean oil andcholesterolin the binary fluid aswell asto thoseobserved in pureCOZ.
Experimental
A Spe-edSFEunit (AppliedSeparation,Allentown,PA) equipped with a liquidboosterpumpanda pumpheadchiller wereusedfor this study.Thedensitiesof supercriticalfluids weremeasured bya modelDMA48AntonParrdensity’meter in combinationwith a DMA512Phigh-pressuremeasuring cell (AntonParr,Graz,Austria).Thedensitymeterwascalibratedby a Haskelgasboosterpump(Haskel,Burbank,CA) andanISCOsyringepump(ISCO,Lincoln,NE)usingnitrogen and water,respectively.After the calibrationutilizing the knowndensities of nitrogenandwaterat selected temperatures and pressures,the densityof pure COzwas measuredand comparedwith the datareportedin the literature (8). This differencewasfoundto belessthan0.2%,indicatingthat the accuracyof the densitymeterwasquitesatisfactory. Theprecisionof the densitymeterwasbetterthan 0.02%from seven replicateruns.
Reproduction (photocopying) of editorial content of this journal is prohibited without publisher’s permission.
483
Journal of Chromatographic Science, Vol. 35, October 1997
anddensitycellwereidentical,andthedensity and compositionof the fluid usedfor each extraction was monitored and recorded. Themethodfor determiningsoybeanoil ; Extractor! solubilityin supercritical fluidshasbeenreF ported previously (9,lO). In thisstudy,about Liquid pump P 0 10g of soybean flakeswerepackedin a 25mL extractioncell (Keystone,Bellefonte, PA).Thefluid passingthroughthe cellwas p:i Syringe pump controlledat the selectedtemperatureand Collector pressure. Afterastaticholdfor 30min,a dynamicextractionwascommenced at a flow rateof 2 L/min (expanded gas).Theoil solubility wasthencalculatedfrom the linear portion of the extractioncurve,in which the weight of extractedoil was plotted Density meter versusthe volume of supercritical COs passed throughthe extractioncell. Figure 1. Experimental apparatusfor density, composition, and solubility measurementsusing pure For measurements of cholesterolsoluCO2 and HHS COz. bility, 1.5 g of cholesterol (J.T. Baker, Philipsburg,NJ)and4 g of Hvdromatrix,a diatomaceous earth-based soTbent(Variai,HarborCity,CA), werethoroughlymixedandthenpackedinto a 15-mLextracL*COIl tion cell.The fluid in the cell wasequilibratedat a selected 98n m158 temperature andpressurefor 30 min, andthe extractionwas I l .- 4.5 97.5.l .m= then carried out underthe sameselectedtemperatureand m .- 4 I pressure at a flow rateof 1 Urnin (expanded gas).In thecaseof ';: 97.r? .- 3.5 g the cholesterol extractions, the tubing between the exitof the h .- 3 extraction cell and the collection vial was washed with 10mL g 98.5.. , n l .- 2.5 .E E acetone to assure complete removal of the solute after the N l =I extraction was completed. This extract was then dried under g 98.. --2 .I 1.5 9 l l nitrogen and weighed. For solubility determinations, the .- 1 95.5.l extraction was not terminated until at least 80 mg of choles-* .- 0.5 terol hadbeenextracted. 95'1200 1300 1400 1500 1800 1700 :+ 180: TheSFC-SFEgradeCO2andHHSCO2(> 99.9999% purity) Cylinder head pressure (psi) usedin theseexperiments werepurchased fromAir Products (Allentown,PA)andusedwithoutfurtherpurification. Gas pump
A4w
-.-*-.m
I
A
I
Figure 2. Fluid composition of an HHS cylinder versusthe head pressure of the cylinder at room temperature.
Theexperimental apparatus usedfor makingthe solubility measurements is illustratedin Figure1.Here,the gasbooster pumpandsyringepumpwereonly usedfor calibrationpurposes. Thepressurized fluidcomingout of theliquidpumpw-as splitinto twostreams. Onestreamflowedtowardtheextraction cell housedin an ovenandthen exitedthrougha micrometeringvalveto a collectionvialbeforepassingthrougha flow meterand a dry test meter (AmericanMeter,Philadelphia, PA).Theotherstreamwasconnected to thethermostated DMA 512Phigh-pressuredensitymeasuringcell beforeexiting througha micrometering valve.It thenflowedthrougha2-mL gassamplingloop into a GCutilizing a 20 wt% DC-200on Chromasorb P,anAW-DMCS column,anda TCD(COW-MAC, LehighValley,PA).High-puritynitrogen(99.9995%) wasused asthe carriergasfor GC.TheTCDsignalof the GCwascalibratedwith pureCO2andpurehelium.Duringextractionexperiments,thetemperature andpressureof theextractioncell 484
Results and Discussion
TheHHSCO?cylinderusedin thisstudyhadaninitial head pressureof 2000psi.Because gaswasusedfor extraction,the cylinderpressuredropped.At about1200psi,the CO1content insidethe cylinderbecametoo lowfor conductingadditional extractions. Figure2 showsthe molarpercentages of CO2and heliumin theliquidphaseof theHHScylinderasthe cylinder headpressuredecreased. Thefluid composition wasmeasured bya GCanda TCDwith a precisionof about3%relativestandarddeviation(RSD)for heliumand1%RSDfor CO2(based on threereplicates). Thisdifferencein precisionindiceswasdue to thefactthatheliumhada smaller.peak areain theGCanalysisthancarbondioxide.Thecomposition of thefluid changed relativelyslowlyasthe cylinderheadpressuredecreased. The fluid wasconsidered ashavinga constantcompositionif the variationof the helium molar percentwasnot outsidethe .
Journal of Chromatographic Science, Vol. 35, October 1997
aboveRSDs.Thecomposition of the fluidchanged moreslowlywhenthe cylinderwasfull andmorerapidly Pressure(psi) Composition when the cylinder was almost 8,000 (mol %) 9,950 7,000 5,000 empty. As Figure 2 shows, the He Density (g/ml) Temperature Fluid co2 helium molar percentin the fluid 0.9795 1.0164 100 0 0.8968 0.9564 Pure CO1 decreased from4.9%with a cylinder 0.9420 4.9 0.9159 0.9827 95.1 0.8457 50°C CO*-He headpressureof 1800psi (almost full) to 2.3% at 1230 psi (near 4.2 3.8 3.3 5.7 Difference in density (%) depletion).The measuredhelium 0.9526 0 0.9276 0.9926 Pure COz 100 0.8609 contentwasquite consistentwith 0.9153 95.2 4.8 0.8882 0.9593 60°C C02-He 0.8135 thosereportedbyLeichteret al. (3). 4.2 3.9 5.5 3.4 Difference in density (%) Clearly,the fluid cameout of an 0.9254 0.9686 0 0.8242 0.8981 Pure CO2 100 HHSCO1cylinderasa mixedCOa0.8898 0.9372 95.4 4.6 0.7752 0.8598 70°C CO*-He heliumfluidthat changed composition over time instead of pure COZ. 4.3 3.8 3.2 Difference in density (%) 5.9 Theratioof COZ-helium in thefluid did not changeafterbeingpressurizedbythepump.Theheliumcon@ntmeasured directlyfrom the cylinder was exactly the same as the fluid that hadbeen 4 5,000 psi n 8,000psi As,950 psi pressurized to various pressures ranging from 2,000 to 10,000 11 1 psibythe boosterpump.Theobserved adverse effectson SFC and SFEusingHHS CO2wereobviouslycausedby the dis0.95 -1 solvedheliumin the extractionfluid. Oneof themostobviouseffectsof dissolved heliumin liquid z 0.9CO2 was the reduction of fluid density. Table I showsthe den3 sity differences between pure CO2 and helium-COa at selected ‘z; 0.85.temperaturesand pressures.The densityfor the latter was reducedby about3-6% comparedwith pure COZ.It appears g 0.8that temperaturedid not affectthe magnitudeof densitydrop I3 as significantly aspressure. Forthefluidwith thesamehelium 0.75content,the densitydrop at high pressure(about3%) was muchlessthanthat at lowerpressure(about6%). 0.7’ Figure3 showsthedensityof thefluid at 70°Candthreese4 2 0 8 6 lected extractionpressures asa functionof thefluid composiFluid helium content (mol%) tion. TheCO2contentcaneasilybeobtainedfromFigure3 by subtractingheliumcontentfrom 100%.Thedataat 100%CO2 Figure3. Density of helium-CO2 with different helium contents at 70°C (0% helium)werefrom the experimentsusinga pure CO2 and three selected pressures. cylinder.As Figure3 demonstrates, the densityof the fluid decreased linearlyasheliumcontentincreased underconstant pressure.Because the solvationpowerof supercriticalfluid is stronglyrelatedto its density,the dropin fluid densityaffects l 9OOOpsi n 8,000psi A9,950 psi the solubilityof a solute. Figure4 showsthe soybeanoil solubility in fluids with varyingheliumcontentat 70°Candthreeselectedpressures. Thedecrease of oil solubilitywith the increased heliumpresencein the extractionfluid is verypronouncedfor all three extractionpressures. However, thisrelationship is not linear.It appears thatat lowheliumcontent,thesolubilityof soybean oil decreasedmore rapidly for the samemolar percentage increasein helium.Because the densitychangedlinearlywith heliumcontent,thisresultsuggests thatthedensityofthefluid 4 2 0 6 alonedoesnot explainthe solubility drop in helium-CO2 mixtures. Fluid helium content (mol%) TableII summarizesthe solubilityof cholesterolin both pureCO;!andhelium-CO2mixtures.Theexperimental uncerFigure4. Soybean oil solubility in helium-CO2 having different helium tainty is based on three replicate determinations. At two content at 70°C and selected pressures. differentextractiontemperatures andpressures, thesolubility
Table I. Density of Pure CO2 and Helium-Entrained Temperatures and Selected Pressures
CO2 at Three Different
7
485
Journal of Chromatographic Science, Vol. 35, October 1997
with theobservation byGahrs(11)that whenCOsis mixedwith nitrogen,the solubilityof caffeineandnicotinein the 40°C and 3,500 psi 50°C and 4,000 psi mixedfluid decreases substantially (addition of 5 mol % of nitrogen in COs CO2 (96%kHe (4%) Fluid compositiqn Pure CO2 CO2 (96%)-He (4%) Pure CO* reducesthe solubility of caffeineby 0.055 f 0.001 0.034 f 0.001 0.074 2 0.001 0.053 ?: 0.002 Solubility (w/w%) 50%).ThisverifiesBnmner’sthesis(12) - that a modifyingagent(suchashelium or nitrogen)with a lowercriticaltemperaturethan that of the extractionfluid (COz)causesa decreasein the solubilityof a low-volatilitysolute;also,a modil 5,OOOpsi n 8,OOOpsi As,950 psi fying agent with a higher critical temperature (such as methanolor acetone)causesanincreasein solubility(12). Figure5 showsthedensityof thehelium-CO2mixturesasa percentageof pure COzdensityat 70°Cand three selected pressures. It is clearthat.thepercentage of thedensitydropin the helium-CO2mixturesis almostthe sameasthe molar percentageof helium in the fluid. For example,6 mol % heliumcausedthe densityto dropby 8% at 5,000psiandby about5%at 8,000psi. Figure6 compares the solubilityof soybeanoil in heliumCOzmixtureswith that in pureCOzat 70°Candthreeselected pressures. Thesolubilityin pureCOz(0%helium)wastakenas 100%.Theseresultsarequitedifferentfrom thoseshownin Figure5.With 6 mol % heliumin supercriticalCOz,thesoluI I 90 bility of soybean oil decreased by70%at 5,000psiand60%at 5 0 10 8,000and9,950psi.Althoughthepressurehadaverycleareffect on the relativedensityof the fluid, it did not havethe Fluid helium content (mol%) samestronginfluenceonthe relativesolubilityof soybean oil. For the fluid with the same helium content, the change in Figure 5. Relative density change expressed as the percentage of the relativesolubilityat 8,000psiwasoftenthelargest,whereas the density of pure CO2 at 70°C and three selected pressures. largestchangein relativedensityoccurredat 9,950psi.Consideringthat thesolubilityof soybean oil droppedby70%and its densitydroppedby only 8% at 5,000psi for 6 mol 96of heliumin thefluid,theseresultstogetherclearlysuggestthat +5,OOOpsi n 8,OOOpsi As,950 psi the solubility decreasein helium-CO;,mixturescannotbe 100 A explained bya decrease in densityalone.Asfurtherevidence of this observation, the solubility of soybean oil at 8,000 psi and q 90 -70°Cwas4.51wt% in pureCOzat a densityof 0.9254g/mL, e 80 -whereasthesolubilityof thesameoil was2.91wt%for COz-6 .@ 70 -mol % heliumat a densityof 0.9253g/mL at 9,950psi and G z 60 -70°C.In fact,asTableIII shows,the soybeanoil solubilities =:+ in COz-heliumfluidswith higherdensityweresmallerthan i 50 -I 4 that in pureCOzandhadlowerdensityat the sameextraction Q) 40 -temperature. + g 30 -The solvationpowerof a solventcanbe characterized by m 20 -solubilityparameter, a conceptfirst developed by Hildebrand B 10 -(13)andlaterextended to the supercriticalfluid stateby GidI1 dings et al. (14). The solubility parametertakesaccountof the 0 cohesive energyof thesolventandisa muchbetterdescription 5 0 IO of the solvationpowerof a solventthan density.For superFluid helium content (mol%) criticalfluids,the solubilityparameterasdefinedby Giddings Figure 6. Relative soybean oil solubility expressedas the percentage of is (14): solubility in pure CO2 at 70°C and three selected pressures. 6 = 1.25PcnXL Eq1 pr (14 where6 isthe fluidsolubilityparameter, PCis thecriticalpresdropin helium-COzmixturesis verysignificant.Thereis no doubtthat theheliumpresence in supercriticalCOzdepresses sureof the fluid, pr is the reduceddensityof the fluid, andpr the solubilityof cholesterol. Theseresultsarealsoconsistent (liq)isthereduceddensityfor thefluidat infinitecompression Table II. Solubility of Cholesterol in Pure CO2 and Helium-Entrained CO2 Based on Three Replicates
486
journal of Chromatographic Science, Vol. 35, October 1997
Table III. Soybean Oil Solubility and Fluid Density and Composition at 70°C Based on Three Replicates
I Extraction pressure (psi) 8,000 Helium (mol %) in CO2 0 Density (g/ml) 0.9254 Oil solubility (w/w%) 4.3kO.2
‘;:
b -
s$ 0 3
9,950
9,950
9,950
5.9 f 0.2 0.9253 2.920.1
4.4 t 0.1 0.9355 3.61t0.2
4.0 r 0.1 0.9391 3.950.2
8-
l
7--
8.. 5--
g 4-.Z O 3--
s nal 2-0” v)
I
.**
*+
l +* t
l
pressures werethus calculatedseparately usingEquation1. Thenthe solubilityparameterof the binaryfluid mixturewas calculatedusingEquation2. Figure7 showsthe supercriticalfluid solubilityparameter versusthesoybean oil solubility.All of thesolubilitydatafrom Figure4 wereusedhere,includingdatafor bothpureCO2and COZ-helium compositions at differentpressures (from5,000to 9,950psi)at 70°C.Asshownin Figure7, thesolubilityof soybeanoil increased exponentially asthefluid solubilityparameterincreased from 6.38to 8.29(cal/cm3P2. Thepressureand composition of thesupercritical fluidsthemselves didnot have averyobviouseffectonsolubility,asshownin TableIII, but the calculatedsolubility parameters,which are dependenton pressureandcomposition, demonstrated anunambiguous influence‘on the solubilityof the soybeanoil at variousfluid compositions. TheHildebrand-Scatchard equationcanbeusedto relatethe solutesolubilityto solubilityparametersof the soluteand solventby (13):
*
I-l
Eq3
@*
O+
8
8.5
7
7.5
8
8.5
Fluid solubility parameter (cal/cm3)1~2 Figure 7. Soybean oil solubility versus the solubility parameters of the fluids.
fb* (8-u* 0 -5
1 I 4
2 I
3 1
4 I
5 I
6 I
7
f?* = 0.9643
-62