WATER RESOURCES RESEARCH,
VOL. 34, NO. 10, PAGES 2529-2537, OCTOBER
Partitioning of hydrophobic contaminants in the vadose zone in the presence of a nonaqueous phase C. E. Schaefer,• D. R. Unger, and D. S. Kosson Department of Chemicaland BiochemicalEngineering,Rutgers,The State Universityof New Jersey,Piscataway
Abstract. Partitioningbehaviorof organiccontaminantsin the vadosezone may be influencedby the presenceof a nonaqueousphaseliquid (NAPL). The bulk NAPL phase not only servesas a sink for hydrophobiccontaminantsbut may also cover a significant fraction of the soil surface,reducingthe surfacearea availablefor vapor-solidadsorption. Experimentswere carried out to examinethe applicabilityof binary partitioning parameterswhen appliedto a multiphasesystem.Dodecanewas usedas the model NAPL phase,while toluene and benzenewere used as the model hydrophobiccontaminants. Severalsoil typesand water contentswere examined.Resultsindicatedthat two-phase equilibriumparameterswere sufficientto predict partitioningamongthe variousphasesif fractional soil surfacecoverageby each phasewas considered.A previouslydeveloped model was extendedto accountfor the presenceof NAPL and was used to describethe
multiphase contaminant partitioning presented in thiswork. Modeling of aqueousand vapor-solidadsorptionhas been further developedand quantified by examiningsoil surface The presenceof a nonaqueous phaseliquid (NAPL) in a soil area and pore size distribution via mercury intrusion and matrix has a significanteffect on environmentalrisk assess- Brunauer, Emmett, and Teller (BET) nitrogen adsorption ments, as well as the bioavailabilityof contaminantsin reme- analysis[Ungeret al., 1996].In this model the massadsorbedto diation efforts.NAPL often becomesentrappedin a soil sys- the soil surfacemay be expressedas tem in the form of ganglia [Cohenand Mercer, 1993]. These Msv = fvMs•(dry) (1) NAPL globulesmay serve as reservoirsfor hydrophobiccontaminants.The spread or dispersionof these contaminants where then may be controlledby their releasefrom the NAPL phase Msv massof contaminant sorbedto the solidfrom [Baehr,1987;Hatfieldand Stauffer,1993].In addition,contamthe vapor phasein a water unsaturatedsystem inant partitioningin the presenceof a NAPL often determines [m]; the amount of contaminantpresent in the vapor or aqueous fv fractionof the hydrated(not liquid water) soil phasesthroughmultiphaseequilibrium.Reductionin vapor or surfacearea exposedto the vapor phaseat a give aqueousphase contaminantconcentrationsmay significantly moisture content [dimensionless]; reducethe effectivenessof remediationtechniquessuchas in Msv(dry) massof contaminantsorbedto the solidfrom situ vapor extractionor bioremediation. the vapor phaseon oven-driedsoil (liquid water removed)[m]. 1.
In this work, isotherms obtained for oven-dried soils were
Much attentionhasbeen givento volatileorganiccontami- describedusingthe nonlinearBET adsorptionisotherm. nant partitioningbehaviorin a systemcontainingsolid,vapor, CrnonoK(P/Po) and aqueousphases.Sorptionfrom the aqueousphaseto the solid phasehas been shownto be related to the soil organic Cs= [1- (P/Po) ][1+ (K- 1)(P/Po)] (2) content [Karickhoffet al., 1979; Chiou et al., 1979; Gschwend where and Wu, 1985]. Solid-vaporadsorptionhas been shownto be related to the mineral surfaceof the soil [McCartyet al., 1981; Cs sorbedconcentration[mg/kgdry soil]; Gossand Eisenreich,1996].Upon coverageof the internal soil Cmono monolayeradsorptioncapacity[mg/kgdry soil]; surfacearea by liquid water the amount of adsorptionis no P partial pressureof the volatile contaminant[mg/L]; longer affectedby the moisturecontentof the system[Chiou Po saturationpressureof the volatile contaminant and Shoup, 1985]. This occursbecausenone of the mineral [mg/L]; surfacesare exposedto the vapor phase above this moisture K affinityconstant[dimensionless]. content.
Systemscontaininga residualNAPL phasehave also been examined.The residual saturation(NAPL volume/voidvol•Now at Departmentof Civil and EnvironmentalEngineering, Terman EngineeringCenter, Stanford University, Stanford, Califor- ume) of NAPL in a soil bed rangesfrom 5 to 40% [Bouchard et al., 1990]. For a systemcontaininga soil, aqueous,and nia. NAPL phase,the following partitioning relation has been reCopyright1998 by the American GeophysicalUnion. ported [Boydand Sun, 1990]: Paper number 98WR01905.
K = foMKoMq-foiLKoiL
0043-1397/98/98 WR-01905 $09.00 2529
overall aqueous-soil+ oil sorptioncoefficient[L/kg]; fraction of organicmaterial occurringnaturallyin soil [kg/kg]; KoM soilorganic-aqueous sorptioncoefficient[L/kg]; fore fraction of NAPL occurringin the soil [kg/kg]; Kom NAPL-aqueouspartitioningcoefficient[L/kg].
It alsohasbeenobserved[Nkedi-Kizzeet el., 1987;Reo et el., 1990] that organicsolventsthat are misciblein water can influence the aqueoussolubilityof hydrophobiccontaminants. Even liquids that are virtually immisciblemight be slightly solubleand affect contaminantpartitioning[Preusnitzet el., 1986]. Partitioningin systemscontainingsoil,aqueous,NAPL, and vaporphaseshasalsobeen studied.One approachhasbeen to ignorevapor-solidand NAPL-solid adsorption,assumingcoverage of the soil surfaceby water. Vapor-NAPL and vaporaqueouspartitioningmay then be approximatedby Raoult's law and Henry's law, respectively[Corapciogluand Baehr, 1987].Deviationsfrom Raoult'slaw mayoccur,in whichcase a Scatchard-Hildebrand expressionfor the activitycoefficient may be used [Preusnitzet el., 1986]. Xl = T1P1
not ignored.In addition,the useof Raoult'slaw wasevaluated asa meansof predictingcontaminantpartitioningbetweenthe NAPL (dodecane)and vaporphases.
Hydrophobic contaminantssuch as benzene and toluene may partition into any of four phases:vapor, water, solid, or NAPL. The total massof contaminantin the systemmay be calculatedby a massbalanceon the four phasesas follows:
Mr = CvVv + CwVw + CsVs + CNVN
M r total contaminantmass[mg]; Ci contaminantconcentrationin phasei (V = vapor, W = water, S = soil,N = NAPL)[mg/L]; Vi volumeof phasei (V = vapor, W - water, S - soil, N = NAPL)[L]. Two-phaseequilibriumdata were usedto simplifythe above expression. The followingtwo-phaseequilibriumrelationships were determined
In7,=-•-• [(15,-152)2 + 2e,2•i,•i2] (4b) mole fraction of component1 in the liquid; liquid phaseactivitycoefficientof component1; partial pressureof component1; solubilityparameter of componenti; interactionconstant; molar volume of component1; volume fraction of component2; gasconstant; temperature.
Equation (4a) reducesto Raoult'slaw when the activitycoefficient is equal to unity. Batch experimentsalsohavebeen employedto studypartitioning effects in a four-phase(solid, liquid, aqueous,and vapor) systemunderwater-saturatedconditions[Gen endDupont, 1989]. Their resultssuggestedthat partition coefficients obtained in binary phase systemsgenerallydo not apply to multiphasesystems.Becausepreviousmodelsoften ignorevapor-solidand NAPL-solid sorption[Bensonet el., 1993], it is unknownwhether binary vapor-solidand NAPL-solid partitioningparameterswill applyto four-phasesystems.Thus the applicabilityof binary phaserelationshipsin multiphasesystems is in question. The purpose of this researchwas to examine multiphase partitioning(soil, aqueous,NAPL, and vapor phases)in an unsaturatedsoil systemat low-NAPL relative saturations.In particular, extensionof the previouslydevelopedmodel by Ungeret el.  was studied to accountfor soil surface coverageby NAPL aswell as liquid water. Two-phaseequilibrium parametersto predict overall contaminantpartitioning were then comparedwith experimentaldata to examine the validity of such an approachin a multiphasesystem.Unlike previousmodels,sorptionto the soilphasefrom the vaporand NAPL phasesin the presenceof a residualNAPL phasewas
for both benzene
Cv = HCw
Scatchard-Hildebrand expression or Raoult's
x• 7• P• 15i l•2 v• (I)2 R T
Cv = [CN]
H Henry'slaw constant[mg/L per mg/L]; Ci concentrationof contaminantin phasei [mg/L];
Cs,i concentration of contaminant in phaseS (soil)in equilibriumwith phasei when only the soil phaseand phasei are present[mg/kg].
The bracketedconcentrationterms representthe nonlinear partitioningshownin the labelsto the left of (6d) and(6e). For convenience and clarity in presentingthe model derivation, this notationis usedrather than rewritingthe BET or activity coefficientequations. Henry'slaw coefficients(H) havebeendeterminedfor several organiccontaminantsin aqueoussolution[Turner,1995]. Contaminant partitioning between the NAPL and vapor phases([CN]) was determinedexperimentallyin this work usinga liquid-vaporpartitioningexperiment(describedin section 5). In general,a Scatchard-Hildebrand or other type of equilibriumrelationshipmaybe usedto obtain [CN]. KSN was determinedfor oven-driedsoil usinga batch isothermtechnique. Ksw, which is the soil-waterpartition coefficientobtained under water-saturated conditions, has also been deter-
mined by batch adsorptionexperiments[Ungeret el., 1996].
Cs,v was determinedfor oven-dried(105øC)soils.This allowed for the removal of liquid phasewater but still allowed the solid surfaceto remain hydrated.This relationshippreviously was obtained for each contaminant-soilcombination usedin thisstudyandfoundto be described(with a correlation
Table 1. Propertiesof SelectedSoils Quakertown Silt Loam
Sand, wt % Silt, wt %
Clay, wt % SoilœH Cation exchangecapacity, meq/100g dry soil Organicmatter, dry wt %
20 6.8 12.9
14 5.2 7.1
2.0 4.2 4.2
62 30. 8.0 7.0 68
BET surface area,m2/g
coefficient>0.96) by the BET adsorptionisotherm[Ungeret correlationbetweenpore volume and surfacearea needed to estimateF•,, F•v, and Fw. al., 1996]. The work of Ungeret al. wasthen extendedto include Dividing the massbalance(5) by C•, and substitutingthe both aqueousand NAPL soil surfacecoverage.Pore filling in two-phaseequilibriumdata (alongwith (7) to determineCs), the soilwasassumedto occurfrom smallestto largestdiameter the followingexpressionis obtained: pore in the order of water,NAPL, andthen air (on the basisof Mt Cs [CN] Vw surfacetensionand contactangles)[Cohenand Mercer,1993]. Vs + H (8) It is significantto note that liquid layerswere not assumedto coverall of the soil surfaceat liquid saturations99% of the sorbed contaminant mass is either at the
soil-vapor interface, at the soil-NAPL interface, or in the Soil NAPL itself. The Cohanseysoil experimentat 10% water RS Contaminant andthe Adelphiasoilexperimentat 55% RS (Figures7 and6, Benzene Quakertown respectively)may be describedsimplyby contaminantparti- Toluene Quakertown tioning betweenthe vapor and NAPL phasesbecauseessenAdelphia Cohansey tially no soil surfacearea is exposedto the vapor or NAPL
phases(thusno soil-vapor or soil-NAPLadsorption). The ability of the NAPL-vaporrelationshipobtainedin Figure 3 to BET, Brunauer,Emmett, Turner. BET parametersare obtained predictpartitioningin a multiphasesystemis verifiedby these from Ungeret al. .
SCHAEFER ET AL.: PARTITIONING
Figure10. Experimental datacompared to themodelpredictions for benzene on Quakertown soilat low concentrations. Twoliquidrelativesaturations wereexamined (circle,18%RS;square, 57%RS).RS refers to the relative saturationof water only.
wasmostpronounced at lowcontaminant concentrations and Hatfield,K., andT. B. Stauffer,Transportin porousmediacontaining at low water relativesaturations(below40%).
residualhydrocarbon, I, Model,J. Environ.Eng.N.Y., 119(3),540558, 1993.
KarickhOff,S. W., D. S. Brown, and T. A. Scott, Sorptionof hydro-
Acknowledgments. The authorswouldlike to thankJenniferGlick phobicpollutantson naturalsediments, WaterRes.,13, 241-248, 1979. for her help in obtainingthe experimental data. Fundingfor this researchwassponsored in part by the North East HazardousSub- Lide,D. R. (Ed.),Handbookof Chemistry andPhysics, ChemicalRubber, New York, 1990. stanceResearch Center,ProjectR-49.The authorsalsoappreciate the McCarty,P. L., M. Reinhard,andB. E. Rittmann,Traceorganics in input from the manuscriptreferees.
groundwater, Environ. $ci.Technol., 15,40-51,1981. Nkedi-Kizza,P., P.S. C. Rao, andA. G. Hornsby,Influenceof organic cosolvents on leachingof hydrophobic organicchemicalsthrough Baehr,A. L., Selectivetransportof hydrocarbons in the unsaturated soils,Environ.$ci. Technol.,21, 1107-1111, 1987. zonedue to aqueousandvaporphasepartitioning,WaterResour. Prausnitz,J. M., R. N. Lichtenthaler,and E.G. de Azevedo,Molecular Thermodynamics of Fluid-PhaseEquilibria,pp. 274-291, PrenticeRes., 23, 1926-1938, 1987. Hall, EnglewoodCliffs,N.J., 1986. Benson,D. A., D. Huntley,and P. C. Johnson,Modelingvapor extractionand generaltransportin the presenceof NAPL mixtures Rao, P.S. C., L. S. Lee, and R. Pinal, Cosolvencyand sorptionof hydrophobic organiccontaminants, Environ.Sci.Technol., 24, 647and nonideal conditions,Ground Water,31,437-446, 1993. 654, 1990. Bouchard,D.C., S.C. Mravik, and G. B. Smith,Benzeneand naphin unsatthalenesorptionon soilcontaminated with highmolecularweight Schaefer,C. E., Diffusivemobilityof aromatichydrocarbons urated and saturated soils, Ph.D. dissertation,195 pp., Rutgers residual hydrocarbons fromunleaded gasoline, Chernosphere, 21(8),
Univ.;NewBrunswick, N., J., 1997. 975-989, 1990. Schaefer,C. E., R. R. Arands, H. van der Sloot, and D. S. Kosson, Boyd,S.A., andS. Sun,Residual petroleum andpolychlorobiphenyl Modelingof the gaseous diffusioncoefficient throughunsaturated oilsassorptive phases for organiccontaminants in soils,Environ.Sci.
soil systems, J. Contam.Hydrol.,29, 1-21, 1997. Technol.,24, 142-144, 1990. equilibrium of highlynon-ideal, aqueousChiou,C. T., and T. D. Shoup,Soil sorptionof organicvaporsand Turner,H. L., Vapor-liquid organicsystems anda computermodelfor masstransportin gravity effectsof humidityon sorptivemechanismand capacity,Environ. flow networks,Ph.D. dissertation,279 pp., RutgersUniv., PiscatSci. Technol.,19, 1196-1200, 1985. away,N.J., 1995. Chiou, C. T., L. J. Peters,and V. H. Freed, A physicalconceptof andD. S. Kosson, Predicting soil-water equilibriafor nonionic organiccompounds, Science, 206, Unger,D. R., T. T. Lam,C. E. Schaefer, the effectof moistureon vapor-phase sorptionof volatileorganic 831-832, 1979. compounds to soil,Environ. Sci.Technol., 30,1081-1091, 1996. Cohen,R. M., and J. W. Mercer,DNAPL SiteEvaluation,p. 45, CRC Press,Boca Raton, Fla., 1993.
author)andD. R. Unger,Department Corapcioglu, M. Y., and A. L. Baehr,A compositional multiphase D. S. Kosson(corresponding Engineering,Rutgers,the StateUnivermodelfor groundwater contamination by petroleumproducts,1, of ChemicalandBiochemical sityof NewJersey, 98 BrettRoad,Piscataway, NJ 08854-8058. (e-mail: Theoreticalconsiderations,WaterResour.Res.,23, 191-200, 1987. Gan, D. R., and R. R. Dupont,Multiphaseandmulticompound mea- [email protected]
) C. E. Schaefer,Departmentof Civil and EnvironmentalEngineersurements of batchequilibrium distribution coefficients for sixvolatile Center,StanfordUniversity,Stanford,CA organic compounds, Hazard.Waste Hazard.Mater.,6, 363-383,1989. ing,TermanEngineering Goss,K., and S. J. Eisenreich,Adsorptionof VOC's from the gas 94305-4020.(e-mail:[email protected]
) phaseto differentmineralsand a mineralmixture,Environ.Sci. Technol.,30, 2135-2142, 1996.
Gschwend,P.M., and S. Wu, On the constancyof sediment-water
partitioncoefficients of hydrophobic organicpollutants, Environ. (ReceivedMarch 18, 1998;revisedJune3, 1998; Sci. Technol.,19, 90-96, 1985.