migration and entrapment of mercury in porous media

Journal of Contaminant Hydrology 117 (2010) 60–70

Contents lists available at ScienceDirect

Journal of Contaminant Hydrology

j ourna l homepage: www.e lsev ie r.com/ locate / jconhyd

Migration and entrapment of mercury in porous media

M. Devasena 1, Indumathi M. Nambi⁎Environmental and Water Resources Engineering Division, Department of Civil Engineering, Indian Institute of Technology Madras, Chennai, 600036 India

a r t i c l e i n f o

⁎ Corresponding author. Tel.: +91 44 225742892574252.

E-mail addresses: [email protected] ([email protected] (I.M. Nambi).1 Tel.: +91 09600004594(mobile).

0169-7722/$ – see front matter © 2010 Elsevier B.V.doi:10.1016/j.jconhyd.2010.06.005

a b s t r a c t

Article history:Received 4 August 2009Received in revised form 13 April 2010Accepted 9 June 2010Available online 18 June 2010

Elemental mercury is an immiscible liquid with high density and high interfacial tension withwater. Its movement in the saturated subsurface region is therefore considered as a case of twophase flow involving mercury and water and is expected to be governed by gravity, viscous,hydrodynamic and capillary forces. This paper investigates the migration and capillaryentrapment of mercury in the subsurface based on controlled laboratory capillary pressure–saturation experiments. In the first place, entrapment of mercury was observed inhomogeneous porous media. Residual mercury saturation and van Genuchten's parametersfor mercury entrapment were generated. These data will provide vital inputs for mercurymigration and entrapment models. Secondly, the dependency of residual saturation on fluidproperties was brought out in this work by comparing the experimental results of mercury–water system and DNAPL–water systems. Capillary forces were large enough inmercury–watersystems to counteract the high gravity forces and caused the entrapment of mercury. Largedensity differences between mercury and water lead to a high Bond number and thus a lowresidual mercury saturation was obtained which corroborates with existing DNAPL theories.However, the inverse relationship between residual saturation and capillary numberestablished for NAPL–water systems cannot be compared with mercury–water systems.Moreover, the critical capillary numbers and Bond numbers to mobilize DNAPLs may not beapplicable to mercury since mercury has a low capillary number and high Bond number. Thiswork has enabled the understanding of the process of migration and entrapment of mercuryand provided useful inputs for two phase flow models specific to mercury–water systems. Ithas also highlighted the influence of fluid properties on entrapment and mobilizationparticularly for highly dense, viscous fluid which also possesses high interfacial tension withwater.

© 2010 Elsevier B.V. All rights reserved.

Keywords:MercuryResidual saturationCapillary pressureBond numberCapillary number

1. Introduction

Elemental mercury, although a metal, is unique since it is aliquid at normal temperature. This property, in addition to itshigh specific gravity and electrical conductivity, has broughtabout its widespread use in industries and various types oflaboratory equipments and instruments (U.S.EPA, 2007).Mercury is often found in soils located close to industrial

(office); fax: +91 44

Devasena),
2
in

All rights reserved.

facilities that either usemercury in their processes (chlor-alkaliplants) or in products such as thermometers, barometers andcompact fluorescent lights. Although mercury is present invarious toxic forms, elementalmercury exists as a contaminantof concern in nearly 290 National priorities list (NPL) sitesaccording to U.S. EPA (2007). Mercury amalgamation for theextraction of gold from ore is widely practiced inmany tropicalcountries including Brazil, Peru, Indonesia, Columbia andVietnam (Lechler et al., 2000). One of the largest thermometermanufacturing plants in Kodaikanal, India, once producing100,000 to 150,000 thermometer pieces a month has recentlybeen the source of the 7 ton mercury spill as a result ofimproper disposal practices (IPT, 2003). Such wide usage andimproper land disposals of elemental mercury are menacingproblems threatening the subsurface.

http://dx.doi.org/10.1016/j.jconhyd.2010.06.005

mailto:[email protected]

mailto:[email protected]

http://dx.doi.org/10.1016/j.jconhyd.2010.06.005

http://www.sciencedirect.com/science/journal/01697722

61M. Devasena, I.M. Nambi / Journal of Contaminant Hydrology 117 (2010) 60–70

Owing to its low solubility, mercury tends to stay as aseparate phase immiscible with water (Kudo and Miyahara,1991; Melamed et al., 1997; Wang et al., 2004; Martı´nez etal., 2005). An immiscible fluid migrates under the influence ofcapillary, gravity, viscous and buoyancy forces, spreads bothlaterally and vertically displacing air from pores of theunsaturated zone and water from pores of the saturatedzone. Eventually, it is trapped as discontinuous blobs ofvarying sizes and shapes due to the effect of capillary forces.Oil, petroleum compounds, and chlorinated organic solventsare a few examples of immiscible fluids. In petroleumreservoir engineering, the discontinuous entrapped oil iscalled residual oil saturation (Morrow, 1979; Chatzis andMorrow, 1984; Anderson, 1988). In the case of chlorinatedorganic solvents, the entrapped volume is referred to asresidual non aqueous phase liquid (NAPL) saturation orresidual organic saturation. A residual liquid may bring abouta wide spread of groundwater contamination by its slow anddirect dissolution and by its volatilization into the soil gas(Mercer and Cohen, 1990; Kueper and Frind, 1991; Chevalierand Fonte, 2000).

Mercury has a low water solubility of 0.06 mg/l. It is also13.5 times denser than water (U.S.DOE, 2001). Mercury–water has an interfacial tension of 375 dyn/cm. Thereforemercury entrapment in porous media is expected to begoverned by the same principles as for dense non aqueousphase liquid or DNAPLs. Chemical propertiesmay vary but thebasic principles governing the distribution of elementalmercury in the subsurface are expected to be the same asany DNAPL. In this work, entrapped mercury will be termedas residual mercury saturation and it is the measure of thevolume of mercury trapped in the pores relative to its voidspace.

Entrapped mercury is expected to cause serious conse-quences in the subsurface. The metal can be subjected tooxidation to far more soluble inorganic mercury compoundsthus contaminating the groundwater. Mercury's high vaporpressure would serve as a continuous source of vapor phasecontamination (Lindqvist and Rodlhe, 1985; Eichholz et al.,1988). It can also be subjected to biological activity, whichwill lead to the methylation of mercury, a more seriousproblem considering the high toxicity and solubility ofmethyl mercury (Pandit et al., 1997; Wang et al., 2004).Thus the mass transfer of elemental mercury to vapor phasemercury and other chemical and biological transformationswill all likely be related to its residual saturation. Knowledgeabout the location and distribution of residual mercury in thesubsurface is therefore essential to effectively remediatemercury contaminated sites.

Capillary trapping and residual immiscible fluid saturationwere investigated by a number of researchers for a widerange of immiscible fluids. Controlled laboratory scalecapillary pressure–saturation experiments provide a bettermeans of understanding the physics of transport anddynamics of NAPL entrapment. The data generated from theexperiments are fitted using standard procedures and thefitting parameters are useful inputs to the models developedfor two phase flow migration and entrapment. In addition,entry pressure, irreducible wetting phase saturation, andresidual non wetting phase saturation are crucial dataobtained from these experiments which can be directly

used for making gross field scale estimations. Several modelshave been developed to predict NAPL entrapment phenom-enon and all of them need fundamental input parametersfrom capillary pressure saturation experiments (Kueper andFrind, 1991; Oostrom and Lenhard, 1998; Gerhard andKueper, 2003). The lack of experimental data and constitutiverelationships has been frequently reported as a major lacunain NAPL entrapment literature (Oostrom and Lenhard, 1998).The goal of this study was to investigate the migration ofmercury in homogeneous porous media by conductingcapillary pressure–saturation experiments using mercuryand water as the non wetting fluid and wetting fluidrespectively. The main objective was to independentlydetermine capillary entrapment parameters for mercuryentrapmentmodels. It is also important to test our hypothesiswhether the established theories developed for organicDNAPLs can be extended to mercury. Comparison of exper-imental data and theoretical investigations will be performedfor various DNAPLs including mercury and based on thesestudies the role of fluid properties and the three predominantforces (capillary, gravity and viscous forces) on entrapmentand residual mercury saturation will be elucidated.

For bringing out the influence of fluid properties onmercury entrapment, tetrachloroethylene (PCE)–water andtrichloroethylene (TCE)–water capillary pressure saturationcurves were also generated in this work. PCE is a DNAPL witha solubility of 200 mg/L. Its density, though not analogous tomercury, is still 1.63 times heavier than water (Gillham andRao, 1990). Similarly, TCE is a DNAPL with a density of 1.46 g/cm3 and a solubility of 1100 mg/l (Powers et al., 1998).Therefore, comparison to PCE–water and TCE–water exper-imental results facilitates our understanding of the behaviourof elemental mercury. It also serves to scale up the effects ofdensity, viscosity and interfacial tension on estimatingmercury residual saturation.

2. Contemporary studies

Elemental mercury has been extensively used in mercuryporosimetry to basically analyze the structure of porous mediaand the pore size distribution. Ioannidis et al. (1991) studiedmercury saturation and capillary pressure relation formercury-air using glass etched micromodels. Rigby et al. (2003)conducted porosimetry experiments using sol–gel silica pow-der fragments and pellets. They showed that during porosi-metry experiments with silica gel powders (size∼30–40 μm),neither hysteresis nor entrapment was observed, whereaswhen whole pellets (size ∼3 mm) were used as porous media,the results showed apparent hysteresis and entrapment.However, in their study using pellets, the larger pore sizeswere enclosed by other smaller pore sizes thus creating anartificial heterogeneity. Similar mercury porosimetry experi-ments on glass etched micromodels by Wardlaw and McKellar(1981) also attribute tonon randomstructural heterogeneity asthe cause of mercury entrapment. Their study revealed thatneither grids of uniform pore size nor grids with different andrandomly distributed pore size showed any mercury entrap-ment. But a non random model, which had clusters of largerpores within a continuous network of smaller pores, showedmercury entrapment. Even though the studies by Rigby et al.(2003) and Wardlaw and Mckellar (1981) confirmed the

Table 1Physical properties of porous media.

Porousmedia

Median grainSize (cm)

Permeability(cm2)

Uniformitycoefficient

Coarse 0.1 5.0E−07 1.65Fine 0.04 1.5E−07 2.2

Table 2Physical properties of fluids.

Fluid Densitykg/m3

Viscositykg/m.s

Interfacialtension dyn/cm

Mercury (Hg) 13,500 a 0.001554 a 375 a

Tetrachloroethylene (PCE) 1630 b 0.00089 c 47.8 d

Trichloroethylene (TCE) 1470 e 0.00059 e 35 e

Water 1000 0.001 NA

a U.S.DOE (2001).b Gillham and Rao (1990).c Li et al. (2007).d Pennell et al. (1994).e Powers (1992).

62 M. Devasena, I.M. Nambi / Journal of Contaminant Hydrology 117 (2010) 60–70

entrapment of mercury purely in heterogeneous domains,residual saturation quantification and capillary entrapmentparameters have not been reported.

Precedent studies on DNAPL migration and entrapmentwill aid in understanding the phenomena involved inmercury migration and entrapment. The general conclusionfrom earlier research works is that NAPL saturation stronglydepends on the soil texture and heterogeneity and isinvariant with respect to fluid properties such as density,viscosity and interfacial tension. According to Wilson et al.(1990), soil texture and heterogeneity have far more controlon capillary trapping and residual non wetting fluid satura-tion as long as the Capillary and Bond numbers are low.Capillary numbers (NC) and Bond numbers (NB) are dimen-sionless numbers that bring out the influence of capillary,gravity and viscous forces on NAPL entrapment. Residualsaturation was identical for the six organic liquids (kerosene,xylene, gasoline, p-xylene, n-decane, soltrol and PCE) testedby Wilson et al., 1990 and they found no correlation betweenresidual saturation and viscosity, density or interfacialtension under low NC and NB conditions. However, PCE wasthe only non wetting fluid tested having a density greaterthan water in their study. The average residual organic liquidsaturation over all the fluids was 26.4%. Chatzis et al. (1983)also have shown that only pore geometry controls entrappedNAPL distribution under low NC and low NB conditions.Similar results were obtained by Demond and Roberts (1991)for non halogenated and halogenated organic liquids.

The extent of influence of NC and NB on capillary trapping ofnon wetting fluids have been studied by Conrad et al., 1992,Meakin et al., 2000, Morrow et al., 1988 and Ovdat andBerkowitz, 2006. An increase in either of these numbersdecreases the residual nonwettingfluid saturation.Dombrowskiand Brownell (1954) revealed that the increase in NC resultedin a significant decrease in residual saturation only forNCN10−2. Ng et al. (1978) show that at low fluid velocitieswith NC≤2×10−5, capillary forces dominate viscous forcesand residual saturation either increases or becomes invariant.When NC exceeds 2×10−5, viscous forces become significantand residual non wetting saturation decreases. Morrow andSongkran (1981) found that residual non wetting saturationvaried from 14% at low Nc and to almost zero at higher NC forexperiments conducted with beadpacks. Moreover, intention-ally varying the NC either by reducing the interfacial tension orby increasing the aqueous flow rate is one of the NAPLremediation techniques. Chatzis and Morrow have found acritical NC (under low NC and low NB) above which completeNAPL mobilization takes place. Pennell et al. (1996) found theconcept of trapping number (vectorial sumof NC andNB) underlow NC and high NB conditions. They have found a criticaltrapping number (1×10−3) at which complete NAPL mobili-zation takes place.

Many researchers have reported an inverse relationshipbetween NAPL saturation and NB (Morrow et al. 1988; Guptaand Mohanty, 2001). Relationship between DNAPL saturationand NB depends on the sign of NB. DNAPL–water systemsgenerally have a positive NB since they are always operated ina downward displacement mode or in other words upwardentry of DNAPL. This is to minimize fingering and achieve100% NAPL saturation during its initial invasion. For systemsoperated in a downward displacement mode with a positive

NB, residual saturation is inversely proportional to NB.Negative Bond numbers are generally associated with LNAPLsor with DNAPLs when operated in an upward displacementmode. Results of Dawson and Roberts (1997) show a directrelationship between NAPL saturation and NB since theymeasured DNAPL saturations through an upward displace-ment mode. These results are questionable since an upwarddisplacement mode may result in fingering and randomvariation in residual NAPL saturations. However, it simulatesnatural migration of DNAPL during a spill. Therefore, therelation between residual NAPL saturation and NB dependsupon the sign of NB and mode of displacement.

With an alarming increase in the number of mercury spillsites and the lack of constitutive entrapment models formercury the need for experiments for understanding migra-tion and capillary entrapment of mercury and comparison toorganic NAPLs is justified.

3. Experimental methods

3.1. Materials

Bench top, short column experiments were conducted tomeasure the residual saturation of mercury under watersaturated conditions using two sizes of porous media. Theresults were compared with DNAPL–water experimentalstudies. Analytical grade elemental mercury was used formercury–water experiment. Laboratory grade (99%) PCE andTCE were selected as representative DNAPLs. A red organicdye (Oil-red-O, Fischer Scientific) was added to DNAPLs at aconcentration of 0.5 g/L to aid DNAPL visualization. Theporous media consisted of sand procured from M/s TaminIndustries. The wetting fluid used in the experiments wasdegassed, deionized water. Tables 1 and 2 summarize thesand and fluid properties respectively. The sand sampleswerewashed repeatedly with distilled water and air dried.Porosities and bulk densities of the different sand sizesweremeasured at the time of column packing by knowing themass of soil packed into the column. Hydraulic conductivity


and intrinsic permeability were determined using theconstant head permeability method.

3.2. Column

Fig. 1 shows the schematic of the experimental column thatwas used for the investigation of the relationship betweendegree of water saturation and capillary pressure applied.Mercury–water experiments were conducted in a one dimen-sional acrylic column of 50 mm inner diameter (i.d.) and a50 mm internal length supported by flanges on both sides. Theflanges on both sides were specially fabricated in a parabolicshape as to provide some headspace for the fluids used.Capillarybarriers are placedusually at the ends of the columntoensure that they have an entry pressure large enough toprevent the breakthrough of the fluids both in the upwarddisplacement of wetting fluid and the downward displacementof dense non wetting fluid. Here, a nylon filter was placed onthe top as water wet membrane which allowed the passage ofwater and not mercury. No filter was placed at the bottom. Itwas expected that mercury's high density will not lead tocapillary end effects and moreover the parabolic flanges heldmercury evenly asnot to cause anypreferential pathwayduringits upward movement. In this work, it was visually observedthat even without a capillary barrier, there was no break-through of mercury during the drainage of water. The outflowof water was steady with neither pockets of air or mercury

Fig. 1. Schematic representation of experimental setup. Head of mercury (hm) and hcolumn.

maintaining the filter integrity. Similarly the displacement ofmercury by water did not encounter any breakthrough ofwater.

The column was filled with water approximately halfway.The known weight of sand was gradually poured in thecolumn. After every 1.5 cm of sand layer, the top was tappedto ensure tight packing. After completion of packing, thecolumn was closed tightly. Tubing to and from the columnwas 3 mm i.d silicon tubing attached to two burettes on eitherside. Mercury being a heavy liquid was passed from below tomaintain a stable displacement of water. The column wassubjected to increase in capillary pressure by graduallyelevating the mercury filled burette. Capillary pressure (PC)is given by PC=Pm−Pw=ρghm−ρghw. Pm and Pw are thepressures of mercury and water respectively. hm and hw arethe heights of mercury and water head measured from thebottom of the sand column respectively. Capillary pressurefor mercury–water was measured in terms of water equiv-alent capillary pressure head hc. ρw is the density of water.

hc = Pc = ρwg ð1Þ

For every step, an equilibration time was given and anincrease in water level in the burette was noted down. Theequilibrium time was taken as the time when there was nowater flow into the water filled burette. Water saturation isgiven by the volume of water inside (after increase in

ead of water (hw) are measured from a reference point at the bottom of the


capillary pressure) over the initial volume of water. This wascontinued until the system reached stabilization, that is nomore water level increase in the burette connected to the top.Following the water drainage, the column was slowlysubjected to decrease in capillary pressure by lowering themercury filled burette. The displacement of mercury tookplace gradually and was stopped until no mercury came outof the column. Capillary pressure saturation relationshipswere constructed by plotting water saturation versus capil-lary pressure. Replicate experiments were conducted and theresults are reproducible.

PCE–water and TCE–water experiments were carried outin a similar pressure column made of aluminum. The sameprocedure was adopted except that a teflon filter placed on aporous teflon disc at the PCE/TCE entrance port and a nylonfilter at the water side acted as capillary barriers.

The van Genuchten's model describing the relationbetween capillary pressure and water saturation was fittedto both drainage and imbibition data. Water saturation isexpressed as Se (effective saturation) for the convenience ofnormalizing the data between the aqueous phase saturationof 0 and 1 (Powers, 1992). Effective saturation correspondingto drainage and imbibition were calculated by Eqs. (2) and(3) respectively.

Se = Sw � Sirw = 1� Sirw ð2Þ

Se = Sw � Sirw = 1� Sirw � Snr ð3Þ

where Sw is the water saturation, Sirw is the irreducible watersaturation and Snr is the residual non wetting saturation. vanGenuchten's capillary pressure saturation model is repre-sented as

Se = f1 + αhcð Þng�m ð4Þm = 1−1 = n ð5Þ

where α, n and m are constants. α is a linear scaling factorand has the dimension of 1/L. The constants n and m aredimensionless (van Genuchten, 1980). hc is the capillarypressure head corresponding to the degree of saturation. van

Fig. 2. Observed and fitted capillary pressure — effective wat

Genuchten's parameters were generated from the vanGenuchten et al. (1991) retention curve model RETC. With acomprehensive knowledge on capillary entrapment para-meters, mercury spill can be numerically modeled toillustrate the effects of mercury entrapment and residualmercury saturation in predicting the distribution of mercuryin the subsurface zones.

4. Experimental results

Two phase capillary pressure saturation relationshipswere determined for mercury–water and DNAPL–watersystems. PCE and TCE were the DNAPLs tested. Fig. 2 presentsthe experimental results of the mercury–water systemconducted for coarse sand and illustrates how the residualsaturation of mercury was measured. In the main drainagecurve, it was observed that water started draining only afterreaching an average capillary pressure head of 18.5 cm ofwater. Eventually at the end of the drainage cycle, watersaturation reached a minimum value and no further drainagetook place giving an average irreducible water saturation of0.07. Furthermore in the imbibition pathway, the degree ofwater saturation reached only 0.96 although the mercuryhead was lowered well below zero capillary pressure. Thisshowed that 4% of mercury was entrapped inside the column.Drainage and imbibition curves of mercury–water systemswere fitted with van Genuchten's model and model para-meters were found. RETC was used for fitting the experimen-tal data and generating the fitting parameters. Capillaryentrapment parameters ‘α’ and ‘n’ corresponding to drainagewere found to be 0.03 cm−1 and 8.2 respectively. The valuesmentioned here are the averages of duplicate experiments.

Fig. 2 also illustrates the quality of fit of capillary pressuresaturation functions of van Genuchten's model to theobserved data for coarse sand. The results of trial 2experiment are also shown in Fig. 2. Similarly, the experi-ments were carried out for fine sand and results are tabulatedin Table 3 and the corresponding observed and fitted resultsare shown in Fig. 3 for trials 1 and 2. The residual saturation ofmercury for fine sand was 0.08 and the entry pressure was an

er saturation for mercury–water system (coarse sand).

image of Fig.�2

Table 3Results of mercury–water experiments.

Parameters Units Porous media

Coarse Hg–waterTrial 1

Coarse Hg–waterTrial 2

Average Fine Hg–waterTrial 1

Fine Hg–waterTrial 2

Average

Porosity (θ) 0.30 0.30 0.30 0.30 0.30 0.30Irreducible Water Saturation (Sirw) 0.056 0.08 0.07 0.113 0.108 0.221Residual mercury saturation (Snr) 0.04 0.04 0.04 0.08 0.083 0.08

van Genuchten parametersDrainageAlpha cm−1 0.01 0.05 0.03 0.009 0.008 0.0085n 8.05 8.3 8.2 10 8.8 9.4Capillary pressure (initial) cm 19 18 18.5 25 33 29Capillary pressure (final) cm 130 113 121.5 129 157 143

ImbibitionAlpha cm−1 0.012 0.02 0.016 0.024 0.05 0.074n 8 5.16 6.6 3.7 3.8 3.75Capillary pressure (initial) cm 126 113 119.5 96.5 86.4 91.5Capillary pressure (final) cm −30 −68 −49 −42 −27.2 −34.6

Force analysisViscous force kg/s2 2.16E−06 1.47E−06 1.8E−06 0.47E−06 0.4E−06 0.45E−06Gravity force kg/s2 3.07E−02 3.07E−02 3.07E−02 0.49E−02 0.49E−02 0.49E−02NC 5.7E−09 3.92E−09 4.8E−09 1.26E−09 1.12E−09 1.2E−09NB 8.2E−05 8.2E−05 8.2E−05 1.3E−05 1.3E−05 1.3E−05


average of 29 cm of water. ‘α’ and ‘n’ corresponding todrainage were found to be 0.0085 cm−1 and 9.4 respectively.

The experimental results of the PCE–water system aresummarized in Table 4. Figs. 4 and 5 show PCE–water (coarseand fine sand) and TCE–water experimental results respec-tively. Residual PCE saturation was found to be 0.17 for coarsesand and 0.275 for fine sand. PCE entry pressures weremeasured as 14 cm of water for coarse sand and 23 cm ofwater for fine sand. TCE–water experiments were conductedfor coarse sand only. Residual TCE entry pressure and TCEsaturation was found to be 11.5 cm of water and 0.14respectively.

NC and NB were calculated for all experiments usingEqs. (6)–(8).

NB = ðρnw−ρwÞ gr2 = σ ð6Þ

Fig. 3. Observed and fitted capillary pressure — effective wa

NC = μwUw = σ ð7Þ

Uw = kρwgi= μw ð8Þ

where g is gravitational constant, ρnw and ρw are the densitiesof the non wetting and wetting fluids respectively, r is therepresentative grain size radius, σ is the interfacial tensionbetween the two fluids, μw is the viscosity of the wetting fluid,Uw is the velocity of the wetting fluid, k is the intrinsicpermeability of the porous media and i is the hydraulicgradient (Mayer and Miller, 1993). Maximum hydraulicgradient (i) measured in the experiments was used to findthe viscous forces. Gravity, viscous, capillary forces, NC and NB

are given in Table 3 formercury–water systems and in Table 4for PCE–water systems. Capillary numbers were in the orderof 10−9 to 10−8 for mercury and PCE respectively. Bond

ter saturation for mercury–water system (fine sand).

image of Fig.�3

Table 4Results of PCE–water experiments.

Parameters Units Porous media

Coarse (withcorrection factor)

Coarse (withoutcorrection factor)

Fine (withcorrection factor)

Fine (withoutcorrection factor)

Porosity (θ) 0.38 0.38 0.393 0.393Irreducible water saturation (Sirw) 0.163 0.163 0.288 0.288Residual PCE saturation (Snr) 0.168 0.168 0.275 0.275

van Genuchten parametersDrainageAlpha cm−1 0.05 0.16 0.044 0.2n 15.5 5.8 15.9 4.3Capillary pressure (Initial) cm 14 2 23 3.37Capillary pressure (Final) cm 21.6 9.5 30 10.9

Imbibitionalpha cm−1 0.17 – 0.06 0.33n 2.74 – 14 3.5Capillary pressure (initial) cm 16.8 4 21.8 9Capillary pressure (final) cm 0.9 −11 11.7 −1

Force analysisViscous force kg/s2 8.2E−07 1.7E−07Gravity force kg/s2 1.55E−03 0.25E−03NC 1.71E−08 0.36E−08NB 3.23E−05 0.5E−05


numbers were in the order of 10−5.Variation of residualmercury and PCE saturation with Nc and NB are shown inFigs. 6 and 7 respectively.

5. Discussions

This study investigates the entrapment of mercury inhomogeneousporousmedia and thevarious factors influencingthe residual mercury saturation. The dependency of residualsaturation on fluid properties was brought out in this work bycomparing the experimental results of the mercury–watersystemandDNAPL–water systems. The role of thepredominantforces such as capillary, gravity, and viscous during themigration of mercury in porous media and their influence onmercury entrapment are captured by the dimensionlessnumbers, Nc and NB. The study also investigates whether the

Fig. 4. Observed and fitted capillary pressure — effective water

well established theories developed for organic DNAPLs areapplicable to mercury. Comparison with the existing DNAPLtheories will provide an understanding in choosing properremedial technology to treat mercury contaminated sites. Theresults suggest that there is a significant difference between thebehaviour of mercury and other organic DNAPLs such as PCEand TCE tested in this study. Mercury–water systems exhibiteda lower residual saturation compared toDNAPL–water systems.The van Genuchten's parameters were also different formercury–water systems. The anomalies between mercuryand DNAPL systems elucidate that capillary trapping is equallydependent on the fluid characteristics such as density,interfacial tension and viscosity especially for high densityimmiscible fluids like mercury.

The following paragraphs discuss the influence of fluidand porous media properties on mercury saturations and

saturation for PCE–water system (coarse and fine sand).

image of Fig.�4

Fig. 5. Observed and fitted capillary pressure — effective water saturation for TCE–water system (coarse sand).


compare the theoretical estimates and experimental observa-tions. The effect of Nc and NB are also discussed.

5.1. Influence of fluid properties

Theoretical estimates for mercury–water and PCE–watersystems for similar sand sizes were arrived by using theLeverett scaling and then were compared to experimentallymeasured values. The function by Leverett (1941) is tradi-tionally used to scale up capillary pressures between differentsystems with varying interfacial tension as given in Eq. (9).

PC2 = PC1ðσ2 = σ1Þ ð9Þ

where PC1 and PC2 are the capillary pressures of fluid 1 and 2,σ1 and σ2 are the interfacial tensions of fluids 1 and 2. Theabove equation neglects the influence of the contact angleand assumes similar porosities in the two systems. Based onLeverett scaling, the entry pressure of mercury should be 7.85and 10.7 times higher than PCE and TCE entry pressuresrespectively. The reported values in Tables 3 and 4 show that

Fig. 6. Variation of residual satura

the capillary entry pressure for mercury is only 1.3 timeshigher than PCE. It is only 1.6 times higher than TCE. It is to benoted that mercury–water experiments were conductedwithout any capillary barrier unlike PCE–water and TCE–water experiments. Apart from the teflon filter which acted asa capillary barrier, a porous teflon disc was also placed to holdthe filter paper. A correction factor was therefore used inDNAPL–water experiments to account for the headloss acrossthe teflon disc and teflon paper. The results (Tables 3 and 4)show that the mercury–water experiment data nearlymatched the theoretical estimates when the correction factorwas not included. The Capillary pressure of mercury was 9.25times that of PCE for coarse sand and 8.6 times for fine sand.Capillary pressure of mercury was 12.3 times that of TCE'scapillary pressure. With the correction factor, there is muchmore deviation from theory. Other works in literature havenot reported any correction factor while scaling acrossdifferent systems with and without barrier.

Residual saturation of organic liquids do not vary much ifthe capillary pressure force exerted at the fluid–waterinterface remains essentially the same. The residual

tion with capillary number.

image of Fig.�5

image of Fig.�6

Fig. 7. Variation of residual saturation with Bond number.


saturation of PCE and TCE were 0.17 and 0.14 whereasmercury saturation was 0.04 (coarse sand). With all theparameters being the same for the two systems, highinterfacial tension of mercury–water, high density andviscosity of mercury may be the prime reasons for thissignificant variation in residual saturation between the twofluid pairs. Different capillary pressures exerted by mercuryand PCE lead to varied mercury and PCE saturation, thus,stressing the importance of fluid properties in quantifyingresidual saturations of mercury.

It has been well established that any fluid with a highdensity and viscosity forms an unconditionally stable inter-face (Powers, 1992) and end up in minimum residualsaturation. Mercury being a dense and more viscous fluidthan water forms an unconditionally stable interface bothduring drainage and imbibition. Since the interface is stable,the pressure of water does not alter the interface and waterflows from the pore throat to the pore body and results inminimum residual mercury saturation. In the case of PCE andTCE, they form an unstable interface with water since theirviscosity is less than water. The interface becomes highlycurved as water enters the pore throat during imbibition andthus severs the interface formed in the pore throat resultingin high PCE and TCE residual saturation.

5.2. Influence of sand size

Theoretical estimation of capillary pressures required formercury–water system for coarse and fine sand was calcu-lated by Eq. (10).

PCcoarse = PCfine r1 = r2ð Þ ð10Þ

Capillary pressure was found to be inversely proportionalto the sand size in both mercury–water and PCE–watersystems. Sands with decreasing size require more entrypressure for the non wetting fluid to enter the watersaturated column. The same trend was observed in thiswork. In the case of the mercury–water system, coarse sandrequired an average capillary pressure head of 18.5 cm ofwater while fine sand required an average of 29 cm of water.This shows that the fine sand required approximately 1.5times more capillary pressure than the coarse sand. PCE–

water also required an increase of 1.6 times in capillarypressure to enter the column packed with fine grain sand.Eq. (10) estimates that the entry pressure should be 2.5 timeshigher for fine sand compared to coarse sand irrespective ofthe fluids. The small deviation could be due to difference innon uniformity of sand grain sizes.

5.3. Influence of Capillary number

The independent role of Nc is essential to know whethermercury can be hydraulically removed since hydraulicmobilization is one of the potential remediation techniqueswith respect to organic DNAPLs. Nc is usually increased to therange of 10−4 by reducing the interfacial tension withsurfactants or by increasing the aqueous flow rate in orderto completely mobilize the entrapped NAPL blobs.

Chatzis and Morrow conducted a large number ofexperiments under low NB and Nc conditions to explore therelationship between Nc and residual saturation. They cameup with a critical Nc of 2×10−5 to initiate mobilization oftrapped NAPL blob and a critical Nc of 1.3×10−3 tocompletely mobilize all the trapped blobs. Pennell et al.,1996 varied Nc from 9.21×10−7 to 6.10×10−4 by usingvarious surfactants and by increasing the flow rate. They alsoincreased the NB from 2.62×10−7 to 1.31×10−4 by reducingthe interfacial tension of the displacing fluid. Their studiesshow that mere Nc cannot be used to predict NAPLmobilization especially when the NB is higher. Nc in thesestudies are in the order of 10−9 for the mercury–watersystem which is an order of magnitude less than the PCE–water system (10−8). Such low Nc reveals that migration andentrapment of mercury is independent of viscous forces andis dependent on the combined effect of gravity and capillaryforces. Very low Nc reported for the mercury–water systemalso indicates that it may not be possible to raise the Nc byseveral orders of magnitude to mobilize residual mercuryblobs.

Fig. 6 shows the residual saturation-Nc data from Pennellet al., 1996 and Gupta andMohanty, 2001 and also the resultsof this study.. Gupta and Mohanty (2001) and Pennell et al.,1996 varied Nc by reducing the PCE–water interfacial tensionusing surfactants. Morrow et al. (1988) varied the Nc of the

image of Fig.�7


crude oil–water system by varying flow rates. Irrespective ofhow Nc was altered, an inverse relationship was observed inall systems when comparing Nc versus residual saturation foreach fluid in isolation. Even when comparing across differentorganic NAPL–water systems, Nc showed an inverse correla-tion with residual saturation. But, when the mercury–watersystem and NAPL–water systems were compared, Nc variesdirectly with residual saturation which is contradictory. Thisimplies that mercury, indeed behaves differently whencompared to organic NAPLs and the need for separateconstitutive models for mercury is justified.

5.4. Influence of Bond number

Theoretically, residual non wetting saturation decreaseswith increasing NB during downward displacement of NAPL.Mercury–water systems, show a large density difference anda high interfacial tension compared to PCE–water system.Gravity force was twenty times higher for mercury–watersystems than PCE–water systems. Gravity force, neverthelessa predominant control factor in the migration of highly densemercury, was counteracted by not less trivial capillary force.The capillary forces surmounted the gravity forces and lead tomercury entrapment. The capillary force was 1.22×104 timeshigher than the gravity force in mercury–water system forcoarse sand. It was 3.09×104 times higher in PCE–watersystem again for coarse sand. Similarly, the capillary forcewasmuch higher in fine sand for bothmercury–water (7.65×104)and PCE–water (19.3×104) (Tables 3 and 4).

Inverse relationships between Bond number and residualsaturation are well established in literature as shown in Fig. 7irrespective of the type of NAPL. NB for mercury–water wasfound to be 8.2×10−5 (coarse sand) and 1.3×10−5 (finesand), approximately 2.5 times higher than PCE–watersystems. NB for PCE–water was 3.23×10−5 and 5.17×10−6

for coarse and fine sand respectively. In this work, a 2.5 foldincrease in NB for mercury–water system compared to thePCE–water system resulted in low residual mercury satura-tion of 0.04 (coarse sand). At low NC, gravity and capillaryforces are the significant causes of mercury entrapment.

6. Summary and conclusions

This paper investigates mercury entrapment in thesaturated subsurface by generating capillary pressure satu-ration curves. These curves were used to estimate entrypressures, irreducible water saturations and residual mercurysaturations which provide valid inputs for mercury entrap-ment models. The experimental results obtained by consid-ering flow of mercury as a two phase flow in the saturatedsubsurface were different in a few aspects from the findingsin the precedent studies. In the first place, the entrapment ofmercury was observed in homogeneous media which wasdenied so far. Secondly, mercury–water systems exhibited alow residual saturation of 0.04 as compared to 0.17 for PCE–water systems and 0.14 for TCE–water system. Resultspresent herein show that unlike the previous findings, fluidproperties had an equal effect on the process of non wettingfluid entrapment and residual non wetting fluid saturationespecially for fluids like mercury which have high interfacialtension and density. Migration of mercury is solely under the

influence of gravity and capillary forces and is independent ofthe viscous forces. High NB and low NC observed in mercury–water systems indicate that the type of fluid is very crucialand the critical capillary number or trapping numberestablished for DNAPL systems may not be directly applicableto mercury. The cut off values for Nc for entrapment andmobilization cannot be applied for the mercury–watersystem since its Nc is much lower. A new critical trappingnumber has to be established for mercury–water systems ifsimilar treatment is considered. Deviations from establishedtheories were observed for mercury–water systems whencompared with DNAPL–water system for residual saturation-capillary number relationship. Future research will investi-gate the pore level capillary trapping of mercury and theapplication of these findings to large scale aquifers.

References

Anderson, W.G., 1988. Wettability literature survey — part 6: the effects ofwettability on water flooding. Journal of Petroleum Technology 39 (12),1605–1622.

Chatzis, I., Morrow, N.R., 1984. Correlation of capillary number relationshipsfor sandstone. Society of Petroleum Engineers Journal 24 (5), 555–562.

Chatzis, I., Morrow, N.R., Lim, H.T., 1983. Magnitude and detailed structure ofresidual oil saturation. SPE Journal 23 (2), 311–325.

Chevalier, L.R., Fonte, J.M., 2000. Correlation model to predict residualimmiscible organic contaminants in sandy soils. Journal of HazardousMaterials B72, 39–52.

Conrad, S.H., Wilson, J.L., Mason, W.R., Peplinski, W.J., 1992. Visualization ofresidual organic liquid trapped in aquifers. Water Resources Research 28,467–478.

Dawson, H.E., Roberts, P.V., 1997. Influence of viscous, gravitational andcapillary forces on DNAPL saturation. Ground Water 35 (2), 261–269.

Demond, A.H., Roberts, P.V., 1991. Effect of interfacial forces on two phasecapillary pressure saturation relationships. Water Resources Research27 (3), 423–437.

Dombrowski, H.S., Brownell, L.E., 1954. Residual equilibrium saturation ofporous media. Journal of Industrial and Engineering Chemistry 46,1207–1219.

Eichholz, G.G., Petelka, M.F., Kury, R.L., 1988. Migration of elemental mercurythrough soil from simulated burial sites. Water Resources 22, 15–20.

Gerhard, J.I., Kueper, B.H., 2003. Capillary pressure characteristics necessaryfor simulating DNAPL infiltration, redistribution, and immobilization insaturated porous media. Water Resources Research vol. 39 (No.8), 1212.doi:10.1029/2002WR001270.

Gillham, R.W., Rao, P.S.C., 1990. In significance and treatment of volatileorganic compounds in water supplies. In: Ram, N.M., Russel, F.C., Contor,K.P. (Eds.), Lewis Publishers, Chelsea, MI, pp. 141–181.

Gupta, D.K., Mohanty, K.K., 2001. A laboratory study of surfactant flushing ofDNAPL in the presence of macroemulsion. Environmental Science andTechnology 35, 2836–2843.

Ioannidis, M.A., Chatzis, I., Payatakes, A.C., 1991. A mercury porosimeter forinvestigating capillary phenomena and microdisplacement mechanismsin capillary networks. Journal of Colloid and Interface Science 143 (1),22–36.

IPT, 2003. The Indian People's Tribunal on Environment and Human Rights.Combat Law Publications Pvt Ltd, Indian people tribunal report.

Kudo, A., Miyahara, S., 1991. A case history: Minamata mercury pollution inJapan— from loss of human lives to decontamination. Water Science andTechnology 23, 283–290.

Kueper, B.H., Frind, E.O., 1991. Two phase flow in heterogeneous PorousMedia 1. Model Development. Water Resources Research 27, 1049–1057.

Lechler, P.J., Miller, J.R., Lacerda, L.D., Vinson, D., Bonzong, J.C., Lyons, W.B.,Warwick, J.J., 2000. Elevated mercury concentrations in soils, sediments,water, and fish of the Madeira River basin, Brazilian Amazon: a functionof natural enrichments? The Science of the Total Environment 260,87–96.

Leverett, M.C., 1941. Capillary behaviour in porous solids. American Instituteof Mining and Metallurgical Engineers Transactions, Petroleum Devel-opment and Technology 142, 152–169.

Li, Y., Abriola, L.M., Phelan, T.J., Ramsburg, C.A., Pennell, K.D., 2007.Experimental and numerical validation of the total trapping numberfor prediction of DNAPL mobilization. Environmental Science andTechnology 41, 8135–8141.

http://dx.doi.org/10.1029/2002WR001270


Lindqvist, O., Rodlhe, H., 1985. Atmospheric mercury: a review. Tellus 37B,136–159.

Martı´nez, R.F., Loredo, J., Ordo´n˜ez, A., l Rucandio, M.I., 2005. Physicochemicalcharacterization and mercury speciation of particle-size soil fractions froman abandoned mining area in Mieres, Asturias (Spain). EnvironmentalPollution 142, 217–226.

Mayer, A.S., Miller, C.S., 1993. An experimental investigation of pore-scaledistributions of non aqueous phase liquids at residual saturation.Transport in Porous Media 10, 57–80.

Meakin, P., Wagner, G., vedvik, A., Amundsen, H., Feder, J., Jossang, T., 2000.Invasion percolation and secondary migration: experiments and simula-tions. Marine and Petroleum Geology 17, 777–795.

Melamed, R., Villas Boas, R.C., Gonalves, G.O., Paiva, E.C., 1997. Mechanisms ofphysico-chemical interaction of mercury with fiver sediments from agold mining region in Brazil: Relative mobility of mercury species.Journal of Geochemical Exploration 58, 119–124.

Mercer, J.W., Cohen, R.M., 1990. A review of immiscible fluids in thesubsurface: properties, models, characterization and remediation.Journal of Contaminant Hydrology 6, 107–163.

Morrow, N.R., 1979. Interplay of capillary, viscous and buoyancy forces in themobilization of residual oil. Journal of Canadian Petroleum Geology 18,35–46.

Morrow, N.R., Songkran, B., 1981. Effect of viscous and buoyancy forces onnon wetting phase trapping in porous media. Surface Phenomena inEnhanced Oil Recovery. Plenum Press, New York, pp. 387–411.

Morrow, N.R., Chatzis, I., Taber, J.J., 1988. Entrapment and mobilization ofresidual oil in bead packs. SPE Reservoir Engineering 3 (3), 927–934.

Ng, K.M., Davis, H.T., Scriven, L.E., 1978. Visualization of blob mechanisms inflow through porousmedia. Chemical Engineering Science 33, 1009–1017.

Oostrom, M., Lenhard, R.J., 1998. Comparison of relative permeability-saturation-pressure parametric models for infiltration and redistributionof a light non aqueous phase liquid in sandy porous media. Advances inWater Resources 21 (2), 145–157.

Ovdat, H., Berkowitz, B., 2006. Pore-scale study of drainage displacement undercombined capillary and gravity effects in index-matched porous media.Water Resources Research 42, W06411. doi:10.1029/2005WR00455.

Pandit, G.G., Jha, S.K., Tripati, R.M., Krishnamoorthy, T.M., 1997. Intake ofmethyl mercury by the population of Mumbai, India. The Science of theTotal Environment 205, 267–270.

Pennell, K.D., Jin, M., Pope, G.A., Abriola, L.M., 1994. Surfactant enhancedremediation of soil columns contaminated by residual tetrachloroethylene.Journal of Contaminant Hydrology 16, 35–53.

Pennell, K.D., Pope, G.A., Abriola, L.M., 1996. Influence of viscous andbuoyancy forces on the mobilization of residual tetrachloroethyleneduring surfactant flushing. Environmental Science and Technology 30,1328–1335.

Powers, S.E., (1992). Dissolution of non aqueous phase liquids in saturatedsubsurface systems. Ph.D Dissertation, University of Michigan.

Powers, S.E., Nambi, I.M., Curry Jr, G.W., 1998. Non-aqueous phase liquiddissolution in heterogeneous systems: mechanisms and a local equilib-rium modeling approach. Water Resources Research 34, 3293–3302.

Rigby, S.P., Fletcher, R.S., Riley, S.N., 2003. Determination of the cause ofmercury entrapment during porosimetry experiments on sol–gel silicacatalyst supports. Applied Catalysis A General 247 (1), 27–39.

U.S.DOE, 2001. Mercury Contaminated Material Decontamination Methods:Investigation and Assessment. DE-FG21-95EW55094. .

U.S.EPA, 2007. Treatment Technologies for Mercury in Soil, Waste, andWater. EPA Report 542-R-07-003. .

van Genuchten, M.T., 1980. A closed form equation for predicting thehydraulic conductivity of unsaturated soils. Soil Science Society ofAmerican Journal 44, 892–898.

van Genuchten, M.Th., Leij, F.J., Yates, S.R., 1991. The RETC Code forQuantifying the Hydraulic Functions of Unsaturated Soils. Version 1.0.EPA Report 600/2-91/065. U.S. Salinity Laboratory, USDA, ARS, Riverside,California.

Wang, Q., Kim, D., Dionysiou, D.D., Sorial, G.A., Timberlake, D., 2004. Sourcesand remediation for mercury contamination in aquatic systems — aliterature review. Environmental Pollution 131, 323–336.

Wardlaw, N.C., McKellar, M., 1981. Mercury porosimetry and the interpre-tation of pore geometry in sedimentary rocks and artificial models.Powder Technology 29, 127–143.

Wilson, J.L., Conrad, S.H., Mason, W.R., Peplinski, W., Hagan, E., 1990.Laboratory Investigation of Residual Liquid Organics from Spills, Leaksand the Disposal of HazardousWastes in Groundwater. Final report, EPA/600/6-90/004. US Environmental Protection Agency, Washington, DC.

http://dx.doi.org/10.1029/2005WR00455

migration and entrapment of mercury in porous media

Documents