Geochemical effects on colloid-facilitated metal transport throughzeolitized tuffs from the NevadaTest SiteW. Um Æ C. Papelis

Abstract Natural colloids were generated fromzeolitized tuffs from the Nevada Test Site (NTS) andthe effects of colloids on the transport of a stronglysorbing metal, lead [Pb(II)], were investigated incolumn experiments under different geochemicalconditions. Because of the high sorption affinity ofPb(II) for zeolitized tuffs, the migration of Pb(II)without colloids was strongly retarded. The presenceof mobile colloids, however, enhanced the mobilityof Pb(II). Approximately 75–90% of the elutedPb(II) was transported as a colloid-associated phase.The migration of colloids was closely related to thegeochemical conditions in the background solution.Immobilization of colloids increased as the ionicstrength of the background electrolyte solutionincreased, because of double layer compaction.Remobilization of initially deposited colloidsoccurred when the ionic strength of the backgroundelectrolyte solution was reduced. The mobility ofcolloids increased with increasing pH and increasingflow rates. These results have significant implica-tions for the migration of strongly sorbing radio-nuclides and other metals at nuclear testing facilitiesand metal-contaminated sites and are consistentwith the hypothesis that inorganic contaminantmigration in the subsurface is a function ofgeochemical conditions.

Keywords Colloids Æ Pb(II) Æ Sorption Æ Colloid-facilitated transport Æ Zeolitized tuffs Æ Nevada TestSite (NTS), Nevada, USA

Introduction

Sorption onto particle surfaces and diffusion into a porousaquifer material matrix are considered the principalretardation mechanisms of solute transport in the sub-surface environment. The presence of colloids, however,may significantly enhance the migration of stronglysorbing ions in the subsurface environment. Becausemobile colloids can carry contaminants bound to theirsurfaces, heavy metals or radionuclides associated withmobile colloids may not be subject to the usual retardationmechanisms (McCarthy and Zachara 1989; Smith andDegueldre 1993).In natural groundwater, colloids and suspended particlesare ubiquitous. Colloids are typically thought to be smallparticles, smaller than 10 lm, and could include inorganicmineral fines, organic molecules, bacteria, and viruses(McCarthy and Zachara 1989; Wan and Wilson 1994;Stumm and Morgan 1996). The genesis of colloids ingroundwater results from a release, dissolution, or alter-ation of the primary aquifer matrix as a result of physical,chemical, and biological activity (McCarthy and Zachara1989).Although colloids differ in origin, their concentrationranges from 108 to more than 1012 particles per liter ofwater (Gschwend and Reynolds 1987; Kim et al. 1987;Degueldre et al. 1989). Because of their small size, colloidshave a high specific surface area (surface area per mass),and therefore are highly reactive particles from a phys-icochemical perspective. This increased surface activity isfrequently manifested in strong interactions with radio-nuclides and other toxic heavy metals. Natural colloidsgenerated from aquifer materials are thought to have thesame composition and surface charge characteristics as theminerals comprising the immobile aquifer matrix.Assuming that colloids have the same surface charge as theparent mineral surface, electrostatic repulsive forces wouldtend to increase the mobility of colloids, thereby alsoincreasing the mobility of contaminants associated withthese colloids.Mobile colloids behave like carriers for contaminants, andenhanced migrations of radionuclides associated withcolloids were found at a nuclear waste disposal site (Bateset al. 1992). In the presence of colloids, mobility of con-taminants was reported to be greater than predicted byconsidering dissolved solute transport alone, without col-loids (Buddemeier and Hunt 1988; Saiers and Hornberger

Received: 8 April 2002 / Accepted: 3 June 2002Published online: 17 July 2002ª Springer-Verlag 2002

W. Um Æ C. PapelisDivision of Hydrologic Sciences, Desert Research Institute,755 E. Flamingo Road, Las Vegas, 89119 Nevada, USA

Present address: W. UmPacific Northwest National Laboratory,P.O. Box 999, K6-81, Richland, 99352 Washington, USAE-mail: wooyong.um@pnl.govTel.: +1-509-3764627Fax: +1-509-3761638

Original article

DOI 10.1007/s00254-002-0646-4 Environmental Geology (2002) 43:209–218 209

1996; Jordan et al. 1997). Because of the potentially criticalrole of colloids in solute transport and remediation,colloid-facilitated transport has been widely studied.Mathematical models for predicting contaminant andcolloid-transport through porous and fractured mediahave been developed and applied in laboratory columnexperiments (Corapcioglu and Jiang 1993; Saiers andHornberger 1996; Grindrod and Lee 1997; van de Weerdet al. 1998).Since immobilization and remobilization of colloids areimportant processes for predicting the mobility of con-taminants bound to colloids, studies of colloid depositionand release in porous media have also been widely con-ducted (Rajagopalan and Chu 1981; McDowell-Boyer et al.1986; Elimelech and O’Melia 1990; Elimelech 1991; Faure etal. 1997). Champlin and Eichholz (1976) reported thatinfluent of a detergent solution through a sand column wasable to remobilize the adsorbed clay. McDowell-Boyer(1992) and Roy and Dzombak (1997) reported the remo-bilization of colloids initially deposited on the surface ofporous media by introducing low ionic strength solutions.Although the migration of contaminants bound to mobilecolloids is believed to be dependent on geochemical con-ditions in solution, few studies have been performed toinvestigate the geochemical factors influencing colloid-facilitated transport of contaminants at the Nevada TestSite (NTS). Because of the several hundreds of nucleartests conducted at the NTS during the Cold War, portionsof the soil and groundwater have been exposed to radio-nuclides, as well as other organic and inorganic trace metalcontaminants. These radionuclides include fission prod-ucts, such as 3H, 90Sr, and 137Cs, and transuranic elements,such as 239,240Pu. In addition to radionuclides, toxic metalcontaminants, such as lead, copper, cadmium, and chro-mate, were also released into the environment duringnuclear testing (Bryant and Fabryka-Martin 1991).Previous studies at the NTS indicated that radionuclidesfrom detonation cavities were associated with mobilecolloids having concentrations from 0.28 to 25 mg/L(Buddemeier and Hunt 1988; Kingston and Whitbeck1991). The concentration and mobilization of colloids atthe NTS, especially at the E-Tunnel site, which is located inarea 12 (Rainier Mesa), has been enhanced by anthropo-genic mechanical disturbances including drilling and nu-clear detonations (George et al. 2000). Additional studies,however, of physicochemical factors affecting the migra-tion of natural colloids at the NTS are needed.This study was to investigate the physical and chemicalfactors influencing colloid-facilitated transport of con-taminants at the NTS by performing systematic laboratorycolumn experiments. Batch equilibrium sorption experi-ments were also conducted to determine the sorption be-havior of Pb(II) on zeolitized tuffs. The effects of colloidson the transport of Pb(II) were studied in column exper-iments as a function of colloid concentration, pH, ionicstrength, and flow rate. Lead was selected as a target ele-ment because of its abundance at the NTS and its toxicityto humans. Lead poisoning has been known to cause braindamage and to affect the nervous system (Chang 1988). In

addition, Pb(II) is known to bind strongly on severaldifferent mineral surfaces and aquifer materials, includingoxides, hydroxides, clays, zeolites, and other aluminosili-cate minerals (Schindler et al. 1987; Bargar et al. 1997;Papelis 2001; Um 2001). Finally, because of strong sorptionon aquifer materials at the NTS, Pb(II) could be used as ananalog for other strongly sorbing radionuclides, such astransuranic elements.

Materials and methods

MaterialsNatural zeolitized tuffs were collected from Rainier Mesaat the NTS and were used as a source of colloids and as anadsorbent in batch and column experiments. Field sampleswere ground up using a Brinkman ZM 1000 centrifugalgrinder equipped with stainless-steel sieves of differentsizes. The ground zeolitized tuff samples had a medianmass diameter of 11.84 lm and a modal mass diameter of16.78 lm.The mineralogy of the zeolitized tuffs was determined byX-ray diffraction (XRD) using Cu Ka1 radiation(k=1.5405 A). Based on the XRD results, the major mineralconstituents of the zeolitized tuff are the zeolite clinoptil-olite and feldspars (Um 2001). In terms of elementalcomposition, determined by X-ray fluorescence (XRF) andexpressed as oxide equivalents, the major elements, asexpected, were silicon (Si) and aluminum (Al), specifically,73.7% Si and 13.8% Al. Other major elements includedK2O (5.43%), Na2O (3.44%), Fe2O3 (1.69%), and CaO(1.44%).The particle morphology was determined by scanningelectron microscopy (SEM) using a JEOL JSM-840A SEMwith an energy dispersive X-ray (EDX) attachment. AnSEM micrograph of a thin section of the zeolitized tuff isshown in Fig. 1. Zeolites and feldspars are clearly identi-fied in the figure and bright spots indicate the presence ofheavier elements, primarily iron.

Fig. 1Scanning electron micrograph of a zeolitized tuff thin section showingfeldspar, zeolites, and iron oxide minerals

Original article

210 Environmental Geology (2002) 43:209–218

The density and porosity of zeolitized tuffs were deter-mined using a Micromeritics AccuPyc 1330 mercury po-rosimeter. Based on measurements of mercury penetrationas a function of applied pressure, the total pore volume,pore area, average pore diameter, and porosity of theparticles can be determined. It should be kept in mind,however, that the results obtained by mercury porosimetryreflect characteristics associated primarily with macrop-ores (pores with diameter 500 A or larger). Characteristicsof finer pores were determined by nitrogen adsorption (seediscussion below). The calculated average true density andporosity of zeolitized tuff particles used for the batchsorption experiments were 2.32 g cm–3 and 20.12%, re-spectively. These results agree well with literature valuesfor the corresponding mineral phases present in thezeolitized tuffs.The specific surface area and the mesopore structure of theadsorbents were determined with a Micromeritics ASAP2010 surface area analyzer by nitrogen adsorption and theBET model (Brunauer et al. 1938). The specific surface areais the amount of total physical surface area per unit weightof the adsorbent. The complete sorption isotherm, in-cluding both adsorption and desorption branches, isshown in Fig. 2. Based on the IUPAC classification, thisisotherm belongs to a type IV nitrogen adsorption iso-therm, which is often found in mesoporous materialshaving pore diameters from 20 to 500 A (Gregg and Sing1982). The specific surface areas and pore diameters fordifferent size fractions are given in Table 1.All experiments were conducted using reagent-gradechemicals. NANOpure water (high purity water of at least18 MW-cm resistivity) was used for all solutions. Lead wasadded as lead nitrate [Pb(NO3)2]. Sodium nitrate (NaNO3)was used to adjust the ionic strength of the background

electrolyte. Nitric acid (HNO3) and sodium hydroxide(NaOH) were used to adjust the pH of the solution. Themetal concentration was determined by a Perkin Elmer 4110ZL atomic absorption spectrometer with graphite furnace(GFAA) and Zeeman background correction. Duplicateswere analyzed for each sample and the values were averaged.

Batch sorption experimentsBefore the onset of batch sorption experiments, the ad-sorbents were rinsed several times with NANOpure waterto remove the easily dissolvable salts. Particles smallerthan 0.25 mm were used for all batch sorption experi-ments. Batch sorption experiments were conducted inindividual 12-ml polypropylene centrifuge tubes by mixingground zeolitized tuffs with the metal ion solution. Sorp-tion of Pb to the test tubes was negligible. The initial metalconcentration used in the sorption experiments was 10–5 Mand the solid concentration was 3 g L–1.In general, eight to ten test tubes were prepared for eachbatch of sorption experiments. In any given experimentalset, Pb uptake was determined as a function of pH, all otherparameters being kept constant. A blank solution sample,containing no solid, was included to check the initial metalconcentration. The samples were equilibrated by end-over-end rotation for 48 h. Previous kinetic studies revealedthat a 48-h equilibration period was adequate for theparticles used in the batch equilibrium experiments (Bernot1999). After equilibrium, the samples were centrifuged at9,000 rpm for 30 min to separate the solution from the solid.The final pH was measured, and a 2-mL aliquot was removedfrom the supernatant. The metal concentration was mea-sured by atomic adsorption (AA) and the fractional sorptionuptake (%) was determined by the difference between theinitial and the final metal concentration.

ColloidsNatural colloids were obtained directly by passing NANO-pure water through columns packed with zeolitized tuffs.The effluent containing colloids or suspended particles wasfiltered using 10-lm-membrane nylon filters. Colloidssmaller than 10 lm were concentrated by centrifugation ofthe colloid containing solution at 9,000 rpm for 1 h. Afterdecanting the supernatant, NANOpure water was added tomake a new solution containing colloids and the solutionwas filtered through 0.45-lm-membrane nylon filters. Col-loids that remained on the filter (presumably larger than0.45 lm) were then dried in an oven overnight at 110 �C.The dried colloids were weighed and used to prepare theaqueous suspensions of known mass concentration.Based on the size of membrane filters used for colloidpreparation, the nominal size of colloids used in thecolumn experiments was between 0.45 and 10 lm.Although 10 lm is a relatively large size for colloids,

Fig. 2Isotherms of nitrogen adsorption and desorption for zeolitized tuffs

Table 1Specific surface area andaverage pore diameter forzeolitized tuff particles ofdifferent sizes

Nominal size (mm) <0.25 0.25–0.50 0.5–1.18 1.18–2.80 2.80–4.00Specific surface

area (m2 g–1)12.27 9.381 9.921 8.172 7.799

Average pore diameter(A) by BET

136.3 142.9 134.9 138.4 140.3

Original article

Environmental Geology (2002) 43:209–218 211

because colloids and suspended particles of this size arestill mobile in the column setup used, the objectives of thestudy could be accomplished using this particular sizefraction. The size and composition of colloids and sus-pended particles were confirmed by SEM/EDX. The majorcomponents were Si and Al, followed by lower concen-trations of Mg, K, and Fe, consistent with the compositionof the parent zeolitized tuff material.

Column experimentsGlass chromatographic columns with an internal diameterof 1.5 cm and 15 cm length were used in column experi-ments. These columns were oriented vertically and flow wasset up from bottom to top to prevent depositional effects.Both ends of the columns were covered with cheese-clothsized slightly larger than 1.0 mm to prevent the loss ofparticles. Steady flow was maintained using a peristalticpump. Particles ranging in size from 1.18 to 2.80 mm wereused for packing the columns. The wet packing techniquewas used to minimize air entrapping and to produce uni-form packing in the columns (Oliviera et al. 1996). Beforepacking, the particles were fully saturated with NANOpurewater and introduced into the column through a funnelwhile vibrating the column, which was already filled withNANOpure water. The bulk density of the adsorbents in thecolumn was determined by the volume of the column andthe dry weight of the particles used for packing. Theexperimental conditions in the column experiments areshown in Table 2.Before the start of the column experiments, NANOpurewater was fed through the column to remove easilybreakable particles and to stabilize the flow. This proce-dure was repeated until the turbidity measured in theeluted solution was less than 0.1 NTU (nephelometricturbidity units). Turbidity was measured using a DRT-15CE turbidimeter. Compared to the turbidity of NANOpurewater (�0.02 NTU), 0.1 NTU in the outflow solution wasexpected to be essentially colloid free. Bromide was usedas a tracer to determine the average pore-water velocity.After obtaining bromide breakthrough, the bromide so-lution was immediately replaced by the metal- and colloid-containing solution. The Pb(II) concentration was 10–5 M.Experiments with Pb(II) only, in the absence of colloids,were also conducted to compare the results with runscontaining colloids. A pulse-type input was used for allcolumn experiments.Solutions of different composition were prepared toinvestigate the effects of geochemical conditions on the

colloid-facilitated transport of metals. The range ofexperimental conditions included different pH values (3.2–8.9), colloid concentration (50 and 100 mg L–1) and ionicstrength (0.5, 0.05, and 0.005 M NaNO3) (Table 2). Dif-ferent flow rates (0.86 to 7.45 cm min–1) were also used toinvestigate the effects of flow rate on the migration ofcolloids. Dissolved and particulate forms of Pb(II) in theeluted solution were distinguished using membrane filters.Because the nominal size of the colloids used in this studywas between 0.45 and 10 lm, a 0.2-lm filter was used todistinguish between dissolved and particulate phases. ThePb(II) concentration in the filtrate was considered dis-solved. Lead associated with the particular matter wasdetermined by subtracting the dissolved Pb(II) concen-tration from the total Pb(II) concentration.The eluted solution was collected at the outlet of the columnand the metal concentration was determined by AA. Colloidconcentrations were also determined by turbiditymeasurements using standards prepared from natural col-loid suspensions of known mass concentrations (Roy andDzombak 1997). Before the measurement, the turbidimeterwas calibrated using turbidity standards ranging from 0.02to 10.0 NTU. The turbidity measurements of colloids wereduplicated and average values were used for the analysis.

Results and discussion

Lead sorptionThe fractional uptake of Pb(II) by zeolitized tuffs as afunction of ionic strength and pH is shown in Fig. 3. Ex-periments with three different ionic strengths (0.01, 0.1,and 1.0 M NaNO3) were conducted to evaluate Pb(II)sorption as a function of ionic strength. The effect of ionicstrength on metal ion sorption has been used to distin-guish between strongly and weakly binding cations onmineral surfaces. The sorption of strongly binding cationsis little affected by changes in ionic strength, whereas thesorption of weakly binding cations could be reduced sig-nificantly by a one or two order-of-magnitude increase inionic strength. Strong binding has often been interpretedas a sign of formation of inner-sphere complexes, wheremetal cations bind directly to the mineral surface throughformation of coordination bonds. Weak binding, on theother hand, has been attributed to formation of outer-sphere, ion-pair complexes, where metal ions retain theirprimary hydration sheath during sorption.

Table 2Experimental column andsolution conditions. heff Effectiveporosity was calculated byDarcy’s flow rate divided by Vx

Column Colloids pH Ionic strength Vx qb heff

(mg L–1) (M NaNO3) (cm min–1) (g cm–3) (cm3 cm–3)

Col-1 50 6.9 0.005 6.78 0.71 0.59Col-2 100 6.5 0.005 6.83 0.71 0.58Col-3 100 6.9 0.005 0.86 0.71 0.58Col-4 100 8.1 0.5 6.48 0.75 0.59Col-5 100 8.4 0.05 7.12 0.74 0.58Col-6 100 8.1 0.005 6.91 0.72 0.57Col-7 100 3.2 0.005 6.68 0.71 0.59Col-8 8.9 0.005 7.45 0.74 0.57

Original article

212 Environmental Geology (2002) 43:209–218

At the two lower ionic strength conditions, 0.01 and 0.1 MNaNO3, Pb(II) removal from solution appeared to be pHindependent and quantitative (uptake was at least 95%)(Fig. 3). The pH-independent sorption was consistent withsorption at cation exchange, permanent charge sites ofclinoptilolite. The essentially complete removal of Pb(II)from solution could be explained by the strong affinity ofPb(II) for cation exchange sites and the high cationexchange capacity of clinoptilolite. At the highest ionicstrength, however, Pb(II) was clearly pH dependent,increasing with increasing pH. Although sorption wasmoderate at low to neutral pH (20–30%), it increaseddramatically above pH 7 and Pb(II) removal was completeat around pH 8–9.The pH-dependent sorption is consistent with sorption atamphoteric surface hydroxyl sites or formation of surfaceprecipitates and is typical for metal ion sorption on oxidesurfaces. Amphoteric surface hydroxyl sites are likely toexist at the edges of zeolite channels and feldspars andoxide minerals present in the zeolitized tuff rock. On theother hand, sorption on cation exchange sites is likely tobe suppressed by the five-order-of-magnitude-higher Naconcentration compared to Pb(II) (1.0 vs. 10–5 M). Thelimited Pb(II) sorption at low pH, where hydrolysis andbinding on amphoteric surface hydroxyl sites is expectedto be very low, could be explained by limited cationexchange.Lead sorption on surface hydroxyl sites, leading to theformation of strong inner-sphere complexes at mineral–water interfaces, has been reported (Chisholm-Brause et al.1990; Bargar et al. 1997). In addition, based on X-rayabsorption spectroscopic (XAS) data analysis, Um (2001)concluded that Pb(II) was forming inner-sphere com-plexes above neutral pH. This type of bonding is consid-ered largely irreversible (Um 2001). Due to the irreversiblestrong sorption, the migration of Pb(II) is expected to behighly retarded by zeolitized tuffs under low ionic strengthand high pH conditions.

Lead transport as a function of colloidconcentration

Relative concentrations of Pb(II) and colloids (comparedto the inflow concentrations, respectively) versus porevolumes are shown in Fig. 4 for two different initialcolloid concentrations. The Pb(II) concentration in theeluted solutions, in the presence of colloids, was signif-icantly elevated compared to Pb(II) concentration in theabsence of colloids, consistent with enhanced Pb(II)migration because of association with mobile colloids.Based on the previous macroscopic experiments, Pb(II)is a strongly sorbing cation and reacts rapidly with thezeolitized tuff adsorbents. Although the flow rate inCol-8 (Vx=7.45 cm min–1) was orders of magnitudefaster compared to groundwater flow, because under thelow ionic strength conditions of this run (I=0.005 M,Table 2) Pb(II) removal was complete (Fig. 3), nosignificant Pb(II) concentration was observed in theeluted solution.Given the short contact time in this experiment, absence ofany significant breakthrough indicates rapid sorption.Previous kinetic experiments (Bernot 1999) have shownthat Pb sorption on particles of similar size is fast. Given

Fig. 3Fractional sorption uptake of Pb(II) on 3.0 g/L zeolitized tuffs as afunction of pH and ionic strength. Initial metal concentration was10–5 M. Triangles I=0.01 M NaNO3; circles I=0.1 M NaNO3; squaresI=1.0 M NaNO3

Fig. 4Breakthrough curves of Pb(II) and colloids. a Total Pb(II) concen-trations (dissolved and colloid-bound) for two different concentra-tions of colloids. b Relative concentrations of colloids for twodifferent initial concentrations. Arrows in a indicate the changingpoints to the solutions without colloids

Original article

Environmental Geology (2002) 43:209–218 213

the size and porous nature of these particles, however, it isunlikely that equilibrium could have been reached withinthe contact time of the experiment. It appears, however,that because of the high solid concentration in the col-umns compared to the batch sorption experiments, readilyavailable sites were in sufficient concentration to bind allavailable Pb(II), thereby resulting in nonequilibriumsorption. Papelis and Um (2001) found no Pb(II) break-through at low ionic strength (I=0.01 M) and flow rate(Vx=0.51 cm min–1), even after more than 1,800 porevolumes.The relative concentration of colloids in the elutedsolutions was also measured and plotted as a function ofpore volumes (Fig. 4b). The breakthrough of colloidsshowed a similar pattern to the measured Pb(II) effluentconcentration. These results suggest that the enhancedmobility of Pb(II) in zeolitized tuffs is a direct result ofmobile colloid migration. This assumption is reinforced bythe observed relationship between Pb(II) and colloid rel-ative effluent concentrations: doubling of the colloidconcentration from 50 to 100 mg L–1 resulted in approx-imate doubling of the relative effluent concentration from0.1 to 0.2 (Fig. 4a).To further test the hypothesis that colloids were primarilyresponsible for the observed Pb(II) breakthrough, bothtotal and dissolved Pb(II) concentrations were measured.First, an acidified aliquot of the effluent was analyzed. Thissample represented the total Pb(II) concentration. Anadditional aliquot was filtered using a 0.2-lm nylon filter.The filtrate was assumed to represent the dissolved Pb(II)concentration. The difference between the total Pb(II)concentration and the concentration in the filtrate wasassumed to correspond to the colloid-bound fraction ofthe metal. Concentrations of Pb(II) associated with thetotal and dissolved fraction, for two different colloidconcentrations, 50 and 100 mg L–1, are shown in Fig. 5.The dissolved concentrations are dramatically reducedcompared to the total Pb(II) concentrations, indicatingthat most of Pb(II) was transported by mobile colloids. In

addition, doubling of the colloid concentration from 50 to100 mg L–1 resulted in essentially doubling the Pb(II)relative concentration from approximately 0.1 to 0.2. Ap-proximately 75–90% of the total Pb(II) was transported asa colloidal phase, depending on initial colloid concentra-tion (Fig. 5).

Lead breakthrough as a function of ionic strengthThe eluted relative concentration of Pb(II), as a function ofionic strength, is shown in Fig. 6. A solution containing10–5 M Pb(II) and 100 mg L–1 colloids was used for theseexperiments. Before introducing the Pb(II)–colloid solu-tion, a solution with the same background electrolyteconcentration but without colloids or Pb(II) was used tosaturate and to condition the column. After two porevolumes of the conditioning solution were eluted, the flowwas switched to the Pb(II)–colloid solution. EnhancedPb(II) concentrations, relative to solutions not containingcolloids, were observed in the effluent. These results sug-gest that Pb(II) migration was greatly facilitated by thepresence of mobile colloids. Although the initial concen-tration of colloids was the same during all experimentalruns, Pb(II) concentrations in the effluent were definitely afunction of ionic strength, as can be clearly seen in Fig. 6.Lead concentration was lowest in the highest ionicstrength solution (Col-4, I=0.5 M), suggesting that Pb(II)-carrying colloids were intercepted more efficiently as ionicstrength increased. Deposition of colloids onto particles isaffected by the surface charge of the mobile colloids andthe immobile matrix and, therefore, could be influenced bycharge neutralization of colloid surfaces by sorbed metalions. This charge neutralization of the colloid surface bysorbing metal ions is significant at high metal concentra-tions (Roy and Dzombak 1997). Owing to the relatively lowPb(II) concentration used in these experiments, however,colloid interception and deposition was most likely causedby the shrinkage of the particle double layer. As the ionicstrength increases and the electrical charge in solutionincreases, the thickness of the electrical double layer isreduced (Elimelech 1991). Because colloids were generated

Fig. 5Total Pb(II) concentrations (dissolved and colloid-bound) anddissolved Pb(II) concentrations under different initial concentrationsof colloids

Fig. 6Total Pb(II) concentrations (dissolved and colloid-bound) underdifferent ionic strengths. Initial colloid concentration was 100 mg L–1

and pH=8.0

Original article

214 Environmental Geology (2002) 43:209–218

from the same material used to pack the columns, colloidsand zeolitized tuffs are likely to have similar surfacecharge. It should therefore be expected that electrostaticrepulsive forces would exist between mobile colloids andthe immobile matrix (McDowell-Boyer 1992). Due to therepulsive forces between colloids and particle surfaces,colloids were expected to be repelled from the particlesurface and would show increased mobility, especially atthe lower ionic strength conditions. As the ionic strengthincreased, however, the increasing particle charge neu-tralization and double layer compaction would allowincreased colloid deposition on the immobile particlematrix.Initially stabilized colloids could be released under ap-propriate conditions and increase the mobility of con-taminants; therefore, understanding the conditionspromoting remobilization of colloids is critical. Remobi-lization of attached colloids was negligible when thePb(II)–colloid solution was replaced by a solution of thesame ionic strength but without colloids, regardless ofionic strength. When NaNO3 solutions, however, werereplaced by NANOpure water, a significant increase in theeluted Pb(II) concentration was observed in the columnwith the highest ionic strength solution (I=0.5 M). Giventhe strong affinity of Pb(II) for the zeolitized tuff surfaces,the sharp Pb(II) concentration increase at approximately21 pore volumes in the highest ionic strength run (Col-4)suggested that initially deposited colloids may have beenreleased by introducing a very low ionic strength solution(Fig. 6). There was a marked contrast, however, betweenthe highest ionic strength (0.5 M) and the other ionicstrengths (0.005 and 0.05 M). At these lower ionicstrengths, no significant increase in the relative Pb(II)concentration was observed when the NaNO3 solution wasreplaced by NANOpure.To investigate the relationship between Pb(II) concen-tration and colloid mobilization, the breakthrough ofcolloids as a function of background electrolyte solutionconcentration was obtained and is shown in Fig. 7.

Comparison of Figs. 6 [Pb(II) breakthrough] and 7 (col-loid breakthrough) reveals similarities and differences.Replacing the Pb(II)–colloid solution by an NaNO3 solu-tion did not result in either Pb(II) or colloid break-through, regardless of the ionic strength of the solution.Under these conditions, both Pb(II) sorption and colloidattachment appear to be irreversible. When the NaNO3

solution, however, was replaced by NANOpure water, theresults obtained appeared to be a function of the ionicstrength of the Pb(II)–colloid solution. For example, atthe two lower ionic strengths (0.005 and 0.05 M), therelative Pb(II) effluent concentration was practicallynegligible (Fig. 6), although the colloid effluent concen-tration increased (Fig. 7). At the highest ionic strength,however, increases in both Pb(II) and colloid concentra-tions were observed.The remobilization of colloids appeared to be a functionof the initial background electrolyte concentration. Thelowest ionic strength (0.005 M) resulted in the highestinitial colloid breakthrough (approximately 0.3) and thelowest concentration during remobilization. Conversely,the highest ionic strength (0.5 M) resulted in the lowestinitial breakthrough and the highest breakthrough duringremobilization. The observed behavior is consistent withthe authors’ understanding of colloid attachment toparticles. The lower the solution ionic strength, the morestable the colloids are expected to be and therefore thebreakthrough concentration is expected to be the highest.On the other hand, at the highest ionic strength, colloidsare expected to be increasingly destabilized and thereforewould tend to coagulate and sorb on available matrixsurfaces. Colloid deposition and accumulation in thecolumn is therefore expected to increase with increasingionic strength. When the ionic strength decreases andcolloids are remobilized, the conditions that led tomaximum colloid attachment (highest ionic strength)should result in the highest colloid concentration and,conversely, the conditions that led to the minimum col-loid attachment (lowest ionic strength) should result inthe lowest colloid concentration. This behavior can beobserved in Fig. 7.The release of Pb(II) as a function of eluent solution(Fig. 6) can be contrasted with colloid release shown inFig. 7. At the highest ionic strength, 0.5 M, Pb(II) release isa direct image of colloid release, suggesting that Pb(II) isassociated with colloids under these conditions. At thelower ionic strengths, however, no significant Pb(II)breakthrough is observed, although remobilization ofcolloids is evident in Fig. 7, as was discussed above. Thesedifferences could be explained by differences in the Pb(II)sorption mechanism as a function of ionic strength. Aswas discussed earlier, sorption at the two lower ionicstrengths occurs, most likely, by ion exchange at readilyavailable cation exchange sites. At 0.5 M, however, the pH-dependent sorption suggests formation of strong inner-sphere complexes at amphoteric sites or formation ofsurface precipitates. The latter processes are thought to bemuch less reversible compared to cation exchange. Leadthat is bound to colloids irreversibly is therefore expected

Fig. 7Eluted and released concentrations of colloids under different ionicstrength conditions. Arrows indicate points of solution change

Original article

Environmental Geology (2002) 43:209–218 215

to be remobilized when the colloids are remobilized. Onthe other hand, Pb(II) bound by cation exchange is likelyto sorb on the stationary matrix phase when colloids areremobilized, given the much lower Pb(II) sorption densityon the stationary phase. In summary, the deposition ofcolloids did not appear to be readily reversible, unless theionic strength of the solution decreased significantly. Un-der such conditions, remobilization of colloids appearedto be instantaneous. These results have significant impli-cations for the colloid-facilitated transport of contami-nants.

Lead breakthrough as a function of pHand flow rate

Column experiments with different pH and flow rates werealso performed to investigate the effects of these importanthydrogeochemical parameters on colloid mobility. Becausemost of the naturally occurring inorganic colloids, such asfines of clay, silica, and other metal oxides, are assumed tobe relatively hydrophilic and charged either positively ornegatively depending on pH (Wan and Wilson 1994),electrostatic repulsion between colloids and particle sur-faces would exist if colloids and stationary particle sur-faces had the same charge. For colloids and particles of thesame charge, the magnitude of electrostatic repulsiveforces would increase with increasing charge.Runs at two different pH values, 3.2 and 8.1, resulted in adifferent breakthrough behavior for colloids and, there-fore, different Pb(II) transport characteristics (Fig. 8). ThePb(II) breakthrough concentration at the lower pH run(Col-7; pH=3.2) was lower compared to the breakthroughconcentration at the higher pH run (Col-6; pH=8.1). Thesolution pH was the only difference between the two runs,all other parameters being equal, including ionic strength(0.005 M) and flow rate. The differences between the tworuns can be attributed to the effects of surface charge oncolloid coagulation and deposition efficiency. The surfacecharge, in turn, is a function of solution pH. The surfacesof oxides and other mineral surfaces in solution arehydrated. Because of the hydration process, surfaces are

covered with either surface hydroxyl or other pH-depen-dent functional groups. Hydroxyl groups can be proto-nated or deprotonated, depending on solution pH. At lowpH, because of the abundance of hydrogen ions in solu-tion, surfaces are protonated and therefore positivelycharged, while at high pH the opposite is true, so theybecome negatively charged. The pH at which the net sur-face charge of the mineral is zero is broadly defined as thepoint of zero charge, pHPZC or simply PZC. The PZC of aparticular mineral is a function of the acidity of the surfacehydroxyl sites and can vary from about 2 for the silanolgroups of quartz to around 9 for hydroxyl groups ofaluminum and iron oxides.The observed colloid breakthrough dependence on pH canbe explained based on the expected surface charge ofcolloids and stationary mineral phases present. Becausethe colloids used in this study were derived from the samerock material used in the column studies, it is expectedthat the surface properties of the colloids and the sta-tionary matrix would be very similar. When the surfacecharge of colloids and matrix is increased, the electrostaticrepulsive forces between colloids and between colloids andsurface matrix would also be increased, leading to maxi-mum colloid stabilization and maximum colloid break-through. Contributing to the stabilization of colloids wasthe low ionic strength of the solution. Conversely, whenthe surface charge of colloids and matrix is minimized, theelectrostatic repulsive forces are also minimized, therebyenhancing colloid capture efficiency and minimizingcolloid breakthrough.The observed data are consistent with this hypothesis.The major components of the zeolitized tuff are thezeolite clinoptilolite and feldspars. Although very fewstudies on the surface charge properties of feldspars havebeen conducted, assumptions on these minerals’ PZC canbe made based on the limited studies and the knowncomposition of the minerals. Although aluminol sitesexist in these minerals, silanol sites predominate in allfeldspars. These sites are known to be acidic, resulting ina PZC of approximately 2 for quartz. Because of thepresence of aluminol sites, the overall PZC is expected tobe higher, and a value of 3–4, supported by limited data,appears to be reasonable. Similarly, the most commonzeolite in these tuffs had a high silica content, suggestinga high ratio of silanol to aluminol sites and, therefore, aPZC higher than that for quartz, probably also in the 3–4range. Under the lower pH experimental conditions,therefore, the surface charge of the minerals would beminimized and deposition efficiency would be maxi-mized, resulting in lower colloid breakthrough concen-tration. Conversely, at the higher pH value, the surfacecharge would be substantially higher, thereby stabilizingthe colloids and leading to increased colloid breakthroughconcentration.The effects of different flow rates on the mobility ofcolloids were also investigated. Approximately a one-order-of-magnitude lower flow rate reduced the elutedconcentration of Pb(II) substantially, although Pb(II) waspresumably still bound to the colloids (Fig. 8). These

Fig. 8Relative Pb(II) concentrations under different pH and flow rates

Original article

216 Environmental Geology (2002) 43:209–218

results are in good agreement with previous studies ofcolloid transport as a function of flow velocity (McDowell-Boyer et al. 1986; McDowell-Boyer 1992). Colloidsdeposited on the surface of porous media were immobi-lized when exposed to a lower flowrate without change insolution chemistry.

Conclusions

Facilitated transport of Pb(II) was observed when mobilecolloids were present in solution. Based on the resultsobtained from batch sorption experiments, Pb(II) was avery strongly reactive element and there was no significantbreakthrough of Pb(II) under low ionic strength or highpH conditions. However, owing to the presence of mobilecolloids, the concentration of eluted Pb(II) was enhanced,depending on geochemical conditions. The eluted Pb(II)concentration increased as the concentration of colloidsincreased.From this study, the mobility of colloids was found to beclosely dependent on geochemical conditions, primarilypH and ionic strength, of the background electrolyte so-lution. The ionic strength of the solution was one of themost important factors controlling the mobility of col-loids. Deposition efficiency of colloids increased with ionicstrength due to shrinkage of the double layer region.Therefore, the migration of colloids in low-ionic-strengthsolutions was greater than that in high-ionic-strength so-lutions. Remobilization of initially deposited colloids didnot occur, unless the initial background electrolyte solu-tion was replaced by a solution of lower ionic strength.Remobilization of Pb(II), however, was also a factor ofionic strength. Even when colloids were remobilized,Pb(II) breakthrough was only observed in the highest ionicstrength tested. These differences are most likely a factorof the sorption processes involved.The pH of the solution was also an important factor af-fecting colloid mobility. Because most of the naturallygenerated colloids had the same surface charge as theimmobile aquifer matrix, mobile colloids were excludedfrom the aquifer matrix by electrostatic repulsive forces.Thus, mobile colloids would travel more easily undernatural conditions (low ionic strength and above-neutralpH). Slow flow rates increased the residence time ofcolloids in pore waters, which increased the depositionefficiency and decreased the mobility of colloids.Because groundwater flow at the NTS occurs throughfractures and because anthropogenic disturbances (drill-ing wells and detonation tests) generate high concentra-tions of colloids, the migration of colloids in the NTS isexpected to facilitate the transport of radionuclides ortoxic metals under natural conditions (I�0.003 M andpH�8.3). Quantitative understanding of colloid transportmechanisms related to changes in water chemistry andmineral surface charge is required to predict the transportof contaminants bound to mobile colloids at the NTS.

Acknowledgements Financial support for this project was pro-vided by the US Department of Energy, National Nuclear Security

Administration Nevada Operations Office, under Contract DE-AC08–00NV13609, and by the Desert Research Institute. Thecontinued support of Robert Bangerter is greatly appreciated.Reference herein to any specific commercial product, process, orservice by trade name, trademark, manufacturer, or otherwisedoes not necessarily constitute or imply its endorsement, rec-ommendation, or favoring by the United States Government orany agency thereof or its contractors or subcontractors. Theauthors thank Dr. Spencer Steinberg for his careful review of themanuscript.

References

Bargar JR, Brown GE Jr, Parks GA (1997) Surface complexation ofPb(II) at oxide–water interfaces: I. XAFS and bond-valencedetermination of mononuclear and polynuclear Pb(II) sorptionproducts on aluminum oxides. Geochim Cosmochim Acta61:2617–2637

Bates JK, Bradley JP, Teetsov A, Bradley CR, Buchholts ten BrinkM (1992) Colloid formation during waste form reaction:implications for nuclear waste disposal. Science 256:649–651

Bernot PA (1999) Modeling the rates of lead and strontiumuptake by zeolitized tuff from Rainier Mesa, Nevada Test Site,Nevada. Master’s Thesis, Department of Geoscience, Universityof Nevada, Las Vegas, 101 pp

Brunauer S, Emmet PH, Teller E (1938) Adsorption of gases inmultimolecular layers. J Am Chem Soc 60:309–319

Bryant EA, Fabryka-Martin J (1991) Survey of hazardous mate-rials used in nuclear testing. Rep no LA-12014-MS, Los AlamosNational Laboratory, Los Alamos, New Mexico

Buddemeier RW, Hunt JR (1988) Transport of colloidal con-taminants in groundwater: radionuclide migration at theNevada Test Site. Appl Geochem 3:535–548

Champlin JBF, Eichholz GG (1976) Fixation and remobilization oftrace contaminants in simulated subsurface aquifers. HealthPhys 30:215–219

Chang R (1988) Chemistry, 3rd edn. McGraw-Hill, New YorkChisholm-Brause CJ, Hayes KF, Roe AL, Brown GE Jr, Parks GA,

Leckie JO (1990) Spectroscopic investigation of Pb(II)complexes at the c-Al2O3/water interface. Geochem CosmochimActa 54:1897–1909

Corapcioglu MY, Jiang S (1993) Colloid-facilitated groundwatercontaminant transport. Water Resour Res 29:2215–2226

Degueldre CA, Baeyens B, Goerlich W, Riga J, Verbist J, Stadel-mann P (1989) Colloids in water from a subsurface fracture ingranite rock Grimsel Test Site, Switzerland. Geochim Cosmo-chim Acta 53:603–610

Elimelech M (1991) Kinetics of capture of colloidal particles inpacked beds under attractive double layer interactions. J ColloidInterface Sci 146:337–352

Elimelech M, O’Melia CR (1990) Kinetics of deposition of col-loidal particles in porous media. Environ Sci Technol 24:1528–1536

Faure M-H, Sardin M, Vitorge P (1997) Release of clay particlesfrom an unconsolidated clay–sand core: experiments andmodeling. J Contam Hydrol 26:169–178

George S, Steinberg SM, Hodge V (2000) The concentration,apparent molecular weight and chemical reactivity ofsilica from groundwater in southern Nevada. Chemosphere40:57–63

Gregg SJ, Sing KSW (1982) Adsorption, surface area, and po-rosity, 2nd edn. Academic Press, Orlando, Florida, 303 pp

Grindrod P, Lee AJ (1997) Colloid migration in symmetricalnon-uniform fractures: particle tracking in three dimensions.J Contam Hydrol 27:157–175

Original article

Environmental Geology (2002) 43:209–218 217

Gschwend PM, Reynolds MD (1987) Monodisperse ferrousphosphate colloids in an anoxic ground plume. J ContamHydrol 1:309–327

Jordan RN, Yonge DR, Hathhorn WE (1997) Enhanced mobilityof Pb in the presence of dissolved natural organic matter.J Contam Hydrol 29:59–80

Kim JI, Buckau G, Klenze R (1987) Natural colloids and genera-tion of actinide pseudocolloids in groundwater. In: Come B,Chapman NA (eds) Natural analogues in radioactive wastedisposal. Graham and Trotman, London, pp 289–299

Kingston WL, Whitbeck M (1991) Characterization of colloidsfound in various groundwater environments in central andsouthern Nevada. Water Resources Center Publ 45083, DesertResearch Institute, Las Vegas, Nevada

McCarthy JF, Zachara JM (1989) Subsurface transport ofcontaminants. Environ Sci Technol 23:496–502

McDowell-Boyer LM (1992) Chemical mobilization of micron-sized particles in saturated porous media under steady flowconditions. Environ Sci Technol 26:586–593

McDowell-Boyer LM, Hunt JR, Sitar N (1986) Particle transportthrough porous media. Water Resour Res 22:1901–1921

Oliviera IB, Demond AH, Salehzadeh A (1996) Packing of sandsfor the production of homogeneous porous media. Soil Sci SocAm J 60:49–53

Papelis C (2001) Cation and anion sorption on granite from theProject Shoal Test Area, near Fallon, Nevada, USA. Adv EnvironRes 5:151–166

Papelis C, Um W (2001) Anion and cation transport in zeolitizedtuffs from the Nevada Test Site: effects of ion type, pH, and ionic

strength. In: Selim HM, Sparks DL (eds) Heavy metals release insoils. CRC Press, Boca Raton, pp 31–54

Rajagopalan R, Chu R (1981) Dynamics of adsorption of colloidalparticles in packed beds. J Colloid Interface Sci 86:299–317

Roy S, Dzombak DA (1997) Chemical factors influencing colloid-facilitated transport of contaminants in porous media. EnvironSci Technol 31:656–664

Saiers JE, Hornberger GM (1996) The role of colloidal kaolinite inthe transport of cesium laboratory sand columns. Water ResourRes 32:33–41

Schindler PW, Liechti P, Westall JC (1987) Adsorption of copper,cadmium and lead from aqueous solution to the kaolinite/waterinterface. J Neth Agric Sci 35:219–230

Smith PA, Degueldre C (1993) Colloid-facilitated transport ofradionuclides through fractured media. J Contam Hydrol13:143–166

Stumm W, Morgan JJ (1996) Aquatic chemistry, 3rd edn. Wiley,New York

Um W (2001) Sorption mechanisms and transport behavior ofSr(II) and Pb(II) on zeolitized tuffs from the Nevada Test Site.PhD Diss, University of Nevada, Reno, 173 pp

van de Weerd H, Leijnse A, van Riemsdijk WH (1998)Transport of reactive colloids and contaminants in ground-water: effect of nonlinear kinetic interactions. J Contam Hydrol32:313–331

Wan J, Wilson J (1994) Visualization of the role of the gas–waterinterface on the fate and transport of colloids in porous media.Water Resour Res 30:11–23

Original article

218 Environmental Geology (2002) 43:209–218