NUREG/CR-6603SAND98-0419

RW

Characterization of RetardationMechanisms in Soil

Manuscript Completed: January 1998Date Published: April 1998

Prepared byH. R. Westrich, P. V. Brady, R. T. Cygan, S. E. Gruenhagen, H. L. Anderson, SNLK. L. Nagy*

Sandia National LaboratoriesAlbuquerque, NM 87185-0750

*University of ColoradoBoulder, CO

E. O’Donnell, NRC Project Manager

Prepared forWaste Management BranchDivision of Regulatory ApplicationsOffice of Nuclear Regulatory ResearchU. S. Nuclear Regulatory CommissionWashington, DC 20555-0001NRC Job Code B6608

Abstract

Realistic performance assessment of low-level radioactive waste sites is dependent upon ourunderstanding of the transport and retardation of radionuclides in soils under a variety ofgeochemical conditions. Improved fate and transport codes should include a mechanistic modelof radionuclide retardation so as to minimize uncertainties. We have investigated metal sorption(Cs+, Sr2+, and Ba2+) on a simple clay mineral (kaolinite) to better understand the geochemicalinteractions of common soil minerals with contaminated groundwaters. Our studies includedetailed characterizations of kaolinite surfaces and sorption sites, experimental adsorptionmeasurements, surface complexation modeling, and theoretical simulations of cation sorption.Spectroscopic analyses of Cs-exchanged samples show that Cs is sorbed at mineral surfaces andat the interlayer site of smectite clays and on edge and basal sites of kaolinite. Atomic ForceMicroscopic examination of kaolinite crystallites suggests that edges comprisefrom 10 to 50% of its measured surface area. Proton adsorption isotherms forkaolinite and molecular modeling results show that aluminol edge binding sites are more reactiveat typical soil pH than silanol edge sites, and are the most likely sites for metal sorption onkaolinite in natural solutions. Relative metal binding strengths decrease from Ba2+ to Sr2+ to Cs+,with some portion sorbed on both kaolinite edges and basal surfaces. Some Cs+ also appears tobe irreversibly sorbed on all sites. Molecular dynamics simulations suggest that Cs+ is sorbed ataluminol (010) edge sites as an inner-sphere complex and weakly sorbed as an outer-spherecomplex on (001) basal surfaces. The first, all-atom structure energy minimization for kaolinitewas completed, and calculated cell dimensions, bond lengths and bond angles agree very wellwith observed values. These results provide the validation for accurate descriptions of morecomplex structures that will be used in simulations of metal sorption onto realistic clay minerals.Recent measurements and characterization of Sr2+ and Ba2+ sorption onto montmorillonite clayindicate that montmorillonite metal binding strengths are similar to those seen for kaolinite. Incontrast to kaolinite sorption, the basal plane residual charge in smectite clays greatly influencesmetal sorption. On the other hand, phase transformation kinetics (ferrihydrite to goethite) may bea more important control over metal sorption and desorption from Fe-oxides/hydroxides. Thesemeasurements are being extended and will be interpreted in light of new atomistic simulations.These results provide a basis for understanding and predicting metal sorption onto kaolinite, anda framework for characterizing sorption processes on more complex soil minerals.

Contents1.0 Introduction ............................................................................................................................ 8

1.1 Background ........................................................................................................................... 92.0 Procedure.............................................................................................................................. 10

2.1 Task 1: Completion of characterization of Cs sorption onto kaolinite .............................. 112.2 Task 2: Computer simulations of ionic interactions with clay surfaces............................. 112.3 Task 3: Isothermal measurements of adsorption onto clays ............................................... 122.4 Task 4: AFM microcharacterization of clay surfaces ........................................................ 132.5 Task 5: Sorption measurements during Fe-oxide/hydroxide phase transformation........... 13

3.0 Results................................................................................................................................... 143.1 Task 1: Completion of characterization of Cs sorption onto kaolinite .............................. 143.2 Task 2: Computer simulations of ionic interactions with clay surfaces............................. 163.3 Task 3. Isothermal adsorption measurements onto clays................................................... 183.4 Task 4: AFM microcharacterization of clay surfaces ........................................................ 203.5 Task 5: Sorption measurements during Fe-oxide/hydroxide phase transformation........... 21

4.0 Discussion.............................................................................................................................. 255.0 Conclusions........................................................................................................................... 286.0 References............................................................................................................................. 29

List of Figures

Figure 1. Schematic of kaolinite structure; note that the SiO4 tetrahedral layer is apically bondedto a gibbsite-like octahedral layer, where the basal spacing is 7.1 Å. .................................... 10

Figure 2. High-resolution XPS spectra for unsputtered kaolinite and montmorillonite afterreaction with a CsCl solution; the single peaks do not identify the presence of multiplesurface or interlayer sorption sites. ......................................................................................... 15

Figure 3. 133Cs MAS NMR spectrum for kaolinite after reaction with CsCl solutions at 25 and50°C; one broad and three minor peaks have been assigned to edge, surface, and ‘interlayer’Cs sorption sites...................................................................................................................... 15

Figure 4. Comparisons of calculated (dark) and experimental (light) kaolinite structures. .......... 16Figure 5. Neutron powder diffraction patterns for Keokuk kaolinite (observed) and that

calculated for an LDF-optimized structure............................................................................. 16Figure 6. Molecular electrostatic potential surface of hydrated kaolinite crystal viewed along

(010) edge; arrows point to areas of high negative charge. .................................................... 17Figure 7. MD Simulation of inner-sphere sorption of Cs+ ion onto aluminol site at (010) edge of

the kaolinite structure. ............................................................................................................ 17Figure 8. Sorption (filled circles) and desorption (open circles) of Cs+ onto kaolinite as a function

of ionic strength at 25°C......................................................................................................... 18Figure 9. Sorption isotherms for Sr and Ba onto kaolinite; the line is best fit for the constant

capacitance surface complexation model. .............................................................................. 19Figure 10. Ba sorption on Swy-1 (pretreated with NaCl to remove exchangeable Ca) at 25°C as a

function of solution pH and Ca/Na ratio. ............................................................................... 19Figure 11. Ba exchange onto unwashed SWy-1............................................................................ 20Figure 12. Sr (lower) exchange on unwashed SWy-1. .................................................................. 20

Figure 13. AFM image of kaolinite crystallites (001) orientation, with well-developed hexagonalhabit; upper right crystal has surface steps of ~800 Å in height............................................. 20

Figure 14. Tapping mode AFM image of Swy-1 montmorillonite; particle size ranges from 0.1-1µm........................................................................................................................................... 21

Figure 15. Percent transformation [(1-FeOx/FeHCl X 100] of ferrihydrite aged in 0.01 M KNO3 atpH12 and 50°C. ...................................................................................................................... 22

Figure 16. Ba sorption on ferrihydrite at pH 12 and 50°C; note the transformation of ferrihydriteto goethite with time and the associated drop in Ba sorption. ................................................ 22

Figure 17. XRD patterns of ferrihydrite samples aged in 0.01 M KNO3 at pH12 and 50°C(F=ferrihydrite; G= goethite). .................................................................................................22

Figure 18. XRD patterns of ferrihydrite samples aged in 0.01 M KNO3 and 0.0005 M BaNO3 atpH12 and 50°C (F = ferrihydrite; G = goethite). .................................................................... 23

Figure 19. BET surface areas of ferrifydrite aged at 50°C and pH12 in 0.01 M KNO3................ 23

TableTable 1. Ba/Na Exchange Constants…………………………………………………………….12

Disclaimer:

NUREG\CR-6603

Is not a substitute for NRC regulations, and compliance is not required. The approaches , end ormethods described in the NUREG\CR are provided for information only. Publication of thisreport does not necessarily constitute an NRC approval of or agreement with the informationcontained herein.

1.0 Introduction

The Nuclear Regulatory Commission (NRC)and several State agencies have been chargedwith decommissioning of civilian nuclearfacilities, as well as licensing and regulationof uranium mill tailings sites and commercialradioactive waste disposal facilities. TheNRC and state regulators must be able toevaluate the expected performance of adecommissioned site or a waste disposal sitein accordance with federal or stateregulations. Current performance assessment(PA) models of radionuclide transport in soiland groundwater use simplified conceptualmodels for radionuclide retardation that arebased on linear and reversible partitioncoefficients (KD s) measured for a specific setof experimental conditions. The KD approachshould be appropriate when the site is wellcharacterized, both chemically andhydrologically. Unfortunately, experienceshows that this empirical approach often failsto correlate with field measurements of actualradionuclide transport. The reason is thatexperimental KD s have localized applicationand are sometimes unrealistic whenconsidering the range of nonlineargeochemical parameters that can significantlyaffect radionuclide transport mechanisms andkinetics (temperature, pH, fluid composition,ionic strength, mineral structure, substratereactivity, organic complexation, etc.). Topredict radionuclide retardation over a widerange of environmental conditions that arerelevant to NRC licensing concerns, it iscritical to understand and model each of theseeffects, especially for common soil minerals.Since sorption and desorption phenomenaoften dominate radionuclide retardation insoils, they are probably the most importantgeochemical processes to model in PA codes.Once radionuclide sorption and desorptionmechanisms are better understood, more

appropriate retardation models can beincorporated into PA codes to allow forimproved treatment of uncertainty whenmaking estimates of dose to the public.

Transport and attenuation of radionuclides insubsurface environments are largelycontrolled by the geochemical interactionsof dissolved aqueous waste with commonoxide and silicate minerals. These minerals,which range in complexity from quartz tosmectite-like clays to poor and/or well-crystalline Fe- oxides/hydroxides, are foundin soils and rocks adjacent to the waste site.An understanding of the surface structureand sorptive properties of soil minerals incontact with dissolved radionuclides isfundamental to modeling metal retardation.Hydrogeochemical transport codes canaccurately predict the transport andretardation of dissolved radionuclides onlywhen there is a mechanistic understandingof the adsorption and fixation behavior ofmetals in near-field soils.

While theoretical models exist for modelingsurface complexation, sorption, anddesorption at mineral surfaces, experimentalverification of those models lags, as doesatomistic characterization of retardationmechanisms. This NRC project,Characterization of Retardation Mechanismsin Soils; JCN B6608, is designed to providemechanistic and kinetic data about thesorption behavior of metals with complexmineral surfaces. Unraveling themechanistic controls on sorption underdiverse geochemical conditions is necessaryto predict metal transport and retardation insoils. These data will add confidence toregulatory decisions on radionuclidetransport and retardation. The regulatoryobjectives of this project are to provide the

NRC, state regulators, and disposal siteoperators with data and models describingradionuclide retardation by soil-formingminerals. Specifically, it will provideevaluation and quantification of those criticalmechanisms, geochemical parameters, andkinetic data that control the retardation ofradionuclides on common sorbing phases innear-field soils, principally clay minerals(kaolinite and montmorillonite) and Fe-oxides/hydroxides (ferrihydrite, hematite, andgoethite).

This work is multidisciplinary in nature andwill involve topographic and spectroscopiccharacterizations of mineral surfaces byatomic force microscopy (AFM) and nuclearmagnetic resonance spectroscopy (NMR)techniques, experimental measurements ofadsorption isotherms (KD s) of Cs+, Sr2+, Ra2+

(using Ba2+ as an analog) on clay and Fe-oxyhydroxide phases, and theoreticalsimulations of the kaolinite mineral surfaceand Cs sorption onto kaolinite. Thisintegrated investigation will provide thetechnically defensible methodology forevaluating radionuclide migration underdiverse geochemical conditions in soilenvironments, especially for those sites withnew contaminants or with insufficientbudgets to measure local KD s.

1.1 Background

Much of the work documented in this study isbased upon our earlier metal sorption andcharacterization results for the first three-yeargrant of the NRC project (JCN B6608),1-4 aswell as recent Department of Energy/BasicEnergy Sciences/Geosciences research onsilicate mineral dissolution.5-9 In this earlierwork, Sandia National Laboratories initiateda comprehensive investigation (analytical,experimental, and theoretical) of alkali metalsorption onto the surface of kaolinite. This

research included Cs+ sorptionmeasurements on silica, corundum, andkaolinite, and the NMR and X-rayphotoelectron (XPS) spectroscopiccharacterizations of the sorption of Cs+ ontothe multiple surface and interlayer sites ofclay minerals. These results demonstratedthe critical role of the mineral latticeframework in controlling dissolution andleaching processes, and of the effect ofsecondary leach layers upon their sorptionproperties. Furthermore, this researchdemonstrated the importance of modernspectroscopic characterization of the surfaceof the reacting mineral phase, includingchemistry, atomic structure, and bondingconfigurations at the surface sorption sites.

Complementary to the spectroscopiccharacterizations and the experimentalsorption measurements are state-of-the-artempirically based molecular simulations ofthe interactions of a monovalent cation (Cs)sorbate with a simple clay substrate(kaolinite), using consortium softwareprovided by Molecular Simulation Inc.(MSI) for their Catalysis and SorptionProject. The computer simulations help toevaluate experimental or analytical researchresults and to develop a useful strategy forfuture work. Our work unravelingmechanistic controls on sorption is anecessary first step to understanding thegeneral controls on metal transport in soils.In particular, our intensive, multifocusedapproach to Cs sorption onto kaoliniteestablishes a portable framework in which toconsider the binding of other radionuclidesto other minerals. It also allows us to scale-up our results to predict radionuclidemovement in real-world situations by usingthe mechanistic work to quantify thesorptive sink in site-specific scenarios. Inthe renewal phase of this project, we beganto extend this research to other sorbatemetals (Sr and Ba) and substrates (e.g., the

structurally complex smectite clays such asmontmorillonite, and Fe-oxyhydroxideminerals). In this document, we report on thedata obtained during the course of the entireproject, recognizing that many aspects of thisresearch are unfinished and will be addressed

by a new NRC project entitled“Incorporation of Mechanistic SorptionModels in SEDSS” (JCN W6811).

2.0 Procedure

Recognizing the complex chemistryassociated with characterizing and evaluatingthe interactions of a metal sorbate withsimple and interlayer clay substrates or Fe-oxides/hydroxides, we have emphasizedcomplementary experimental, analytical, andtheoretical studies of metal sorption ontomodel clays and colloid minerals. Weinitiated a comprehensive investigation of themechanism of alkali and alkaline earthsorption onto the surfaces of several clay andFe-oxyhydroxide minerals. This researchcombines spectroscopic analyses, batchsorption, surface complexation, molecularmodeling, and atomic force microscopy tocharacterize the mechanisms by which metalsadhere to mineral surfaces. Most of the workdescribed here is for Cs sorption ontokaolinite clay, because of the clay’s relativelysimple chemistry and structure. Generalanalytical and experimental studies of cationsorption processes were performed using bothnatural and synthetic soil mineral samples.Natural samples include KGa-1 kaolinite(Al 2Si2O5(OH)4 from Warren County,Georgia), SAz-1 montmorillonite (Ca-interlayer clay from Apache County, Arizona)and SWy-1 montmorillonite (Na/Ca-interlayer clay from Crook County,Wyoming). ). Synthetic soil mineral samplesinclude goethite [FeO(OH)] and ferrihydrite[Fe2O3

.nH2O]. The clay and Fe-oxideminerals were chosen to provide increasinglycomplex structures and sorption substrates bywhich the mechanism of metal sorption onto

mineral surfaces might be determined. Forexample, kaolinite is a 1:1 layer type claymineral composed of a tetrahedral SiO4 layerlinked to an octahedral gibbsite layer(tetrahedral-octahedral [TO] structure, seeFigure 1). Montmorillonite is a complex 2:1layer type smectite clay (TOT structure),composed of an Al-deficient octahedralsheet sandwiched between two inward-pointing SiO4 tetrahedral sheets, whereinterlayer cations satisfy charge balance.Characterizations of powdered minerals andreacted samples were made using standardanalytical techniques (atomic absorption[AA], direct current plasma [DCP], energydispersive spectroscopy [EDS], inductivelycoupled plasma-mass spectrometry [ICP-MS], transmission electron microscopy

c

a

(100) V iew o f K ao lin ite

Figure 1. Schematic of kaolinite structure;note that the SiO4 tetrahedral layer is apicallybonded to a gibbsite-like octahedral layer,where the basal spacing is 7.1 Å.

NUREG/CR-660311

[TEM], and X-ray diffractometry [XRD]) inorder to verify the structure of startingmaterials and reacted phases, and to quantifysolution compositions. Specific researchprocedures are described separately as tasks1-5, which reflects the work structure of thisproposal.

2.1 Task 1: Completion ofcharacterization of Cs sorption ontokaolinite

After a series of batch Cs sorption treatments,several surface-sensitive spectroscopictechniques were used to characterize theextent and site-specific control of Cs sorptiononto these minerals. These techniquesincluded X-ray photoelectron spectroscopy,which is sensitive to the near-surface (<20 Å)composition and chemical state of elements,and nuclear magnetic resonance (NMR)spectroscopy, which yields structural andbonding information at the molecular level.In these tests, approximately 1 g of themineral was reacted with 100 ml of 0.1 or0.01 M CsCl solutions in polyethylene bottlesfor 5 days at 25 and 50°C. A shaking bathwas used to maintain constant temperatureand suspension of the sample. After the test,samples were separated by filtration. TheXPS analyses were performed under vacuum(~ 2 x 10-9 torr) using a Perkin Elmer PHI5400 instrument employing Mg K X-raysand a takeoff angle of 45°. Adventitiouscarbon (determined by the C1s bindingenergy) provided an internal standard forcorrecting energy shifts resulting from samplecharging. The 133Cs NMR spectra werecollected at room temperature and at 65.5mHz using a spectrometer with a 11.7-Tsuperconducting magnet. Spectra for mostsamples were collected at room humidity;however, a few kaolinite and montmorillonitesamples were collected at 100% relative

humidity (RH) in order to evaluate interlayersite occupancy and exchange. Somesamples were also cooled, or heated, in orderto evaluate thermal effects upon siteoccupancy. The 133Cs chemical shifts arereported in parts per million (ppm) relativeto an external 0.1 M CsCl solution at roomtemperature.

2.2 Task 2: Computer simulations ofionic interactions with clay surfaces

A variety of computational models wereutilized to simulate dissolved Cs sorptiononto kaolinite surfaces. These modelsincluded lattice energy minimization, MonteCarlo, and molecular dynamics (MD). Bycombining all of these techniques, we havedeveloped a general approach for evaluatingsorption sites, isosteric sorption energies,docking mechanisms, and packing of thesorbates. Computer modeling resultsprovide a fundamental basis forinterpretation of analytical and experimentalresults. However, molecular-scalesimulations of Cs sorption onto kaoliniterequire an accurate description of claystructures and surfaces, based either on X-ray crystal cell refinements10 or quantumchemical structure calculations.11 Accuratereproduction of structure and bonding in asheet aluminosilicate mineral is necessaryprior to realistic simulations of surfacerelaxation phenomena for the extendedmineral, water, and adsorbate complex. Thefirst all-atom geometry optimization ofkaolinite has been calculated using aperiodic density functional method. Sincethe kaolinite structure contains bothinterlayer OH groups that form hydrogenbonds with the adjacent silicon tetrahedralsheet and inner OH groups associated withvacant octahedral aluminum sites, it isimportant to include hydrogen atoms in thestructure optimization. Numerous

12NUREG/CR-6603

experimental12,13 and theoretical11,14 studieshave addressed heavy atom positions andproton orientations in kaolinite; however, nooptimization calculations involving all atomsof the unit cell in a periodic structure havebeen conducted. Interlayer and inner OHgroups also have been suggested to besignificant in the surface electrostatics thatinfluence the formation of host/guestcomplexes and thus may play an importantrole as a model in evaluating radionuclidetransport.

The first-principles periodic bulk calculationsused the local density approximation todensity functional theory in a Gaussian-basedlinear combination of atomic orbitals(LCAO) basis. The calculations wereperformed using the QUEST (QUantumElectronic STructure) code, designed formassively-parallel, distributed-memorycomputer architectures. Full geometryoptimization was accomplished starting withatomic coordinates from two recentrefinements of X-ray and neutrondiffraction.12,13 Geometry optimizations wereunconstrained for all atomic coordinates;however, experimental unit cell parameterswere not adjusted relative to new atomicpositions during the minimizations. AllQUEST calculations were run on an 1824-processor Intel Paragon computer.

2.3 Task 3: Isothermal measurements ofadsorption onto clays

Sorption isotherms for Cs+, Sr2+, and Ba2+ (asan analog for Ra2+) onto kaolinite andsmectite (montmorillonite) clays weremeasured as a function of temperature (25,50, and 70°C ) over a wide pH range (2-12).The titrant for the adsorption measurementswas 0.1 M NaOH or HCl and the pHelectrode was calibrated using NationalInstitute of Standards and Technology

samples at the temperature of interest. Fourgrams of kaolinite or montmorillonitepowders (Brunauer, Emmett, and Teller[BET]) surface area ~10 m2/g for KGa-1 and~32 m2/g for SWy-1) were titrated in 50 mLof 0.01-1.0 N NaClO4 or NaCl backgroundelectrolyte solutions using a Mettler DL12titrator. Slurries were continuously stirredand N2 gas was pumped directly into thesolution prior to and during each titration topurge CO2 and to minimize carbonate levelsin solution. Temperature was maintained byadjusting the hot plate that controlled thewater bath in which the reaction vessel wasimmersed. Certified Cs, Sr, and Ba atomicadsorption analytic standards were diluted toparts per million (ppm) levels and used asstock solutions. Solutions containingNaClO4 or SrCl2 or BaCl2 were used asbackground electrolytes in these adsorptionexperiments because Cl- and ClO-

4 do notcomplex with these metals strongly.Solution aliquots were taken after pHstabilization and were analyzed for metalsby directly coupled plasma (DCP) orinductively coupled plasma (ICP)spectroscopy using matrix-matchedstandards. Metal sorption was calculated bydifference from starting solutions. Thedissolved metal concentrations and pH datawere regressed using the computer codeFITEQL15 to calculate best-fit values of pKx

and site density for Cs, Sr, and Ba as afunction of temperature.

The cation exchange capacity of SWy-1(CEC = 76.4 mEq/100 g) is due to thepermanent negative charge on the TOTlayers, and it is pH-independent. This clayhas a BET surface area of 31.8 m2/g, and theprimary exchangeable cations are Na+ andCa2+.16 Exchange equilibria onto mineralssaturated with Na+ are written as:

Ba2+ + 2>Na ↔ >Ba + 2Na+ KBa =

[>Ba][Na+]2/[Ba2+][>Na]2

NUREG/CR-660313

Sr2+ + 2>Na ↔ >Sr + 2Na+ KSr =

[>Sr][Na+]2/[Sr2+][>Na]2

where “>“ denotes an exchange site. Thetotal number of exchange sites is equal to thecation exchange capacity. Bracketed termsdenote concentrations here, as no effort ismade at this point to correct for nonidealityof the aqueous or exchange species. Notethat by using Ba exchange as an example, theexchange constant KBa can be calculated at asingle pH given Ba levels measured insolution from known starting points.Specifically, for the case of Ba in a 0.001MNaCl solution:

KBa = [Bao - Ba2+,anal][0.001 + 2(Bao -Ba2+,anal)]2/[Ba2+,anal][2(CEC-(Bao -

Ba2+,anal))]2

Bao is the spike Ba concentration (1 ppm)and Ba2+,anal is the analyzed Ba concentration.The equation above assumes that the surfaceis saturated initially with Na+, and that Ba2+

exchanges 2:1 for Na+. An analogousequation is used to calculate the exchangeconstant for Sr.

2.4 Task 4: AFM microcharacterizationof clay surfaces.Metal sorption onto clay surfaces is afunction of the charge distribution on thesurfaces as well as the absolute proportions ofexposed planes of differing surface charge.AFM studies of the starting kaolinite surfacewere initiated for comparison with Cs-sorbedsurfaces and for characterization of exposedbasal and edge crystallographic features.AFM acquisition of surface topography to theangstrom scale is based on tip/surfaceinteraction forces, including short-rangerepulsive and attractive (van der Waals)forces. Relative proportions of basaland edge surface area for untreatedkaolinite crystals were determined

using AFM techniques. Kaolinitepowder was ultrasonicallysuspended in water for 1 minute,after which 50 µL of this solutionwas placed on freshly cleavedmuscovite and air dried. Imageswere obtained in contact mode usinga Park Scientific InstrumentsAutoProbe LS (5 µm scanner) and0.6 µm thick Ultralevers (siliconcantilever and tip). Scanningspeeds ranged from 0.5 to 2 Hz.Particle dimensions and stepheights were measured on rawimages of a few representativegrains. These dimensions were usedto calculate relative amounts of edgeand basal surface area. Estimatederrors for calculated surface areasare 10% based on an averaged 5%uncertainty in a single height orlength measurement. Likewise, AFMtopographic images of montmorillonite wereobtained with a Digital InstrumentsNanoscope III using tapping mode in airwith silicon tips. Montmorillonite particleswere suspended in water and pipetted ontofreshly cleaved muscovite for imaging.Image quality seemed to improvedramatically when the humidity was high(about 50%).

2.5 Task 5: Sorption measurementsduring Fe-oxide/hydroxide phasetransformation

Ba sorption and desorption onto the surfacesof ferrihydrite, goethite, and hematite weremeasured as a function of pH, solutioncomposition, and phase transformation. Allexperiments were conducted at pH 12 sinceearlier workers had found that high pHhastens the conversion of ferrihydrite tomore crystalline products. For example, half

14NUREG/CR-6603

conversion of ferrihydrite to goethite at 25°Cand pH 12 occurs in less than 4 days, whilehalf conversion to hematite and goethite atpH 7 takes 112 days.17 Since heating alsohastens ferrihydrite transformation, our testswere conducted at 50°C using two differentionic strengths (0.01-1 M KNO3) to modelsalt contents approaching those of fresh waterand brine.

Approximately 40 g of ferrihydrite wereprecipitated by batch neutralization of 1 MFe(NO3)3 with 1 M KOH to pH 8. Theresulting suspension was adjusted to pH 12and divided into two portions. Barium nitrate(0.0005 M) was added to one portion andthen both portions were aged at 50°C. Thesecond half of the original ferrihydritesuspension was washed free of KNO3 bydialysis, re-suspended in 0.01 M KNO3,adjusted to pH 12, and divided into 2portions. Barium nitrate (0.0005 M) wasadded to one portion of the secondsuspension and then both portions were agedat 50°C. To measure Ba sorption ordesorption on the precipitates, aliquots of thesuspensions were removed periodically andcentrifuged. The supernatant was passedthrough a 0.2 µm syringe filter and the filtrate

was analyzed for Ba by DCP techniques.The precipitates from subsamples werewashed and freeze dried. Iron was extractedfrom filtrates with 0.2 M ammonium oxalateand 6 M HCl to determine the amounts of Fepresent as ferrihydrite and the total Fecontent of the precipitates.18 These extractswere also analyzed for Ba to determine thepartitioning of Ba between ferrihydrite(BaOx) and the goethite formed as theprecipitates were aged (BaHCl - BaOx).Barium adsorbed on the surface of goethitecrystals may also be present in the oxalateextracts.19 Iron and Ba in the extracts weredetermined by DCP. The percenttransformation of ferrihydrite to goethite canbe estimated from the iron analyses:

(1 - FeOx / FeHCl) × 100

Pressed powder mounts of the solidsubsamples were analyzed by X-raydiffraction (XRD). Surface area of selectedsolid samples was determined by thenitrogen BET adsorption method.

3.0 Results

Results from analytical, experimental andtheoretical studies of metal sorption ontocommon soil minerals are given in the nextsections and are discussed by task.

3.1 Task 1: Completion ofcharacterization of Cs sorption ontokaolinite.133Cs MAS NMR and Cs XPS were used tostudy the adsorption of Cs on kaolinite,montmorillonite, corundum, and gibbsitefrom 0.1 N and 0.01 N CsCl solutions. XPS

spectra for all samples except gibbsitecontain strong signals associated with Cs3d3/2 and 3d5/2 photopeaks. Peak intensityanalyses of the XPS spectra indicate thatpowdered corundum sorbed ~0.3 mol% Cs;the kaolinite slab exhibited Cs sorption atthe 1 mol% level; and powderedmontmorillonite sorbed ~2 mol% Cs.Sorbed Cs concentrations were reduced by50-75% after Ar sputtering of the nearsurface. Based upon these results, therelative Cs sorption capacity for theseminerals is montmorillonite > kaolinite >corundum > gibbsite.

NUREG/CR-660315

In addition to near-surface compositionalprofiles, XPS can measure shifts in thebinding energy of core electrons resultingfrom a change in chemical environment(chemical shift). These shifts result from achange in the nearest neighbor, oxidationstate or crystal structure. We did not observea chemical shift for Cs sorbed ontomontmorillonite nor kaolinite. Multiplesurface or interlayer sites for Cs could not beresolved from the single peak Cs spectraobserved for clay mineral samples (Figure 2).

These same samples were also characterizedby 133Cs magic angle spinning (MAS)-NMRspectroscopy. There was no observed 133CsMAS-NMR spectral signal from thecorundum or gibbsite samples. The NMRspectra for the montmorillonite samplescontain one broad peak at -18 to -12 ppm,whereas the kaolinite samples containmultiple peaks (Figure 3), including a large,relatively narrow peak at a more shieldedchemical shift (-40 to -25 ppm) along withseveral other weak peaks. Higherconcentrations of Cs in the reacting solution,

as well as lower temperatures and higherhumidity, resulted in less shielded chemicalshifts. There was increased peak broadeningfor kaolinite and montmorillonite withdecreasing temperature, although separatepeaks were unresolved, even at -80°C.

The 133Cs NMR data, combined withTEM/XRD data for kaolinite, show that formontmorillonite most of the Cs is adsorbed inthe interlayers and is motionally-averagedamong two or more sites at room temperatureand humidity. The spectra for kaolinite aresimilar to those of montmorillonite, and thetightly bonded Cs in our sample appears to be inexpandable interlayers which exist as mixedlayers with kaolinite, and not as discretesmectite. The Cs environments in bothmontmorillonite and kaolinite are sensitive tohumidity. For kaolinite, the 133Cs chemical shiftchanges at low temperature and 100% humidity,indicating that the interlayer charge of itsexpandable layers is very small. No 133Cs NMRsignal could be detected for gibbsite orcorundum.

Figure 2. High-resolution XPS spectrafor unsputtered kaolinite andmontmorillonite after reaction with aCsCl solution; the single peaks do notidentify the presence of multiple surfaceor interlayer sorption sites.

Figure 3. 133Cs MAS NMR spectrum forkaolinite after reaction with CsCl solutionsat 25 and 50°C; one broad and three minorpeaks have been assigned to edge, surface,and ‘interlayer’ Cs sorption sites.

16NUREG/CR-6603

3.2 Task 2: Computer simulations of ionicinteractions with clay surfaces

Symmetries of the Lattice Density Functional(LDF)-calculated kaolinite structures areconsistent with the unit cell possessingoverall C1 symmetry. All heavy atompositions (aluminum, silicon, and oxygen)produce an excellent mapping and reductionof the P1-based LDF structure to a C1symmetry structure. The interlayer hydrogenpositions exhibit C1 symmetry to a toleranceof 0.018 Å, whereas the inner hydrogens ofkaolinite agree to within 0.064 Å. Bothhydrogen position tolerances are similar tothe offsets expected from the standarddeviations obtained for the atomic positionsfrom the neutron diffraction data.12 Overallstructural agreement between heavy atomsand protons of the energy-optimized structureand recent refinement is excellent. Bondlengths and bond angles (see Figure 4) aregenerally consistent with what has beenfound for other clays, kaolinite12 and dickite.20

Differences in O-H bond distances are alsofound for the inner and interlayer OH groups,and are on the order of 0.01 Å. Hydroxylgroup orientations in the geometry-optimized

structures are also similar to thosedetermined through model fitting of theneutron diffraction data. Only minordifferences in bond angles (less than 0.5°)were observed between the full- andhydrogen-only optimized structures.Moreover, the two inner OH positions of theP1 symmetry unit cell used for allcalculations are identical.

Comparisons of the calculated neutronpowder diffraction intensities, based uponcoordinates from the energy-optimized and

observed structures, are in excellentagreement (Figure 5). The calculateddiffraction pattern for kaolinite wasgenerated from the unit cell obtained for theall-atom LDF optimization. All major andalmost all minor peak positions and theirrelative intensities are comparable for bothstructures. In a similar fashion, weexamined the X-ray diffraction patternassociated with the LDF-based structure.No C1-violating peaks were observed at 9.9o

and within the 15.5o to 19.0o 20 range. Theseresults support earlier conclusions12 that thekaolinite structure can be characterized by acrystal lattice that is nonprimitive and

Figure 4. Comparisons of calculated(dark) and experimental (light) kaolinitestructures.

0 20 40 60 80 100 1202 Θ (°)

Inte

nsity

(co

unts

)

Observed

Calculated

Figure 5. Neutron powder diffraction patterns forKeokuk kaolinite (observed) and that calculatedfor an LDF-optimized structure.

NUREG/CR-660317

possesses C-centering. Most convincing inthese results is that our LDF calculationswere performed on a unit cell with P1symmetry (no unit cell symmetry) thatultimately energy optimizes (regularizes) tothe more symmetrical C1 unit cell.

Molecular electrostatic potential (MEP)calculations, which represent the netsummation of the atomic partial charges atthe mineral cluster surface, were performedusing the observed crystallographic structureof kaolinite5 and energy minimized positionsfor hydroxyl groups on selected surfaces.The MEP surface (Figure 6) depicts the mostfavorable sites for ionic sorption on kaolinite.MEP results show that there is a significant

difference in the electrostatic potential ofedge and basal surface sites.21 That is,siloxane and gibbsite basal planes (001) arenot as negatively charged as edge silanol oraluminol edge sites, and are more reactive forcation sorption (greater negative electrostaticcharge) than the edge silanol sites.

Both energy minimization and moleculardynamics techniques were used to simulatethe interactions of dissolved cesium with akaolinite surface. One Cs+ atom, along with

its waters of hydration, was positionedadjacent to either the (010) edge or the (001)basal aluminol surface of kaolinite prior toenergy minimization. Simulations indicatethat Cs+ sorbs directly to the aluminol ((010)edge) as an inner sphere complex (Figure 7).Note that the solvating water

molecules are not positioned between theCs+ and the (010) edge of kaolinite, butrather coordinate to silanol and otherunreacted edge sites. Energy minimizationresults for the (001) surface indicate that Cs+

is coordinated to the basal aluminol plane asan outer sphere complex, with waters ofhydration completely surrounding the Cs+.This contrasts sharply with the inner sphereconfiguration of Cs+ sorption at the (010)edge.

However, shortly after equilibration of thewater molecules on the (001) surface, Cs+

(outer sphere complex) migrates toward the(010) edge and repositions itself as a aninner sphere complex coordinated to an

-a

c

b

0-1 Charge/Angstrom

Hydrated(010) Surface

(001) Basal Surface

+-

Figure 6. Molecular electrostatic potentialsurface of hydrated kaolinite crystal viewedalong (010) edge; arrows point to areas of highnegative charge.

H2O

Si-tetrahedra

Al-octahedra

(010) View

T = 300Kt = 15.6 psec

Cs+

Figure 7. MD Simulation of inner-sphere sorption of Cs+ ion onto aluminolsite at (010) edge of the kaolinitestructure.

18NUREG/CR-6603

aluminol site. Clearly, these computersimulations suggest that the edge aluminolsite is the preferred Cs+ sorption site on astoichiometric kaolinite.

3.3 Task 3. Isothermal adsorptionmeasurements onto clays

Potentiometric surface charge titrations weremade with quartz, corundum, and kaoliniteslurries at 25 and 50°C to evaluate the pHdependence of multisite surface charge as afunction of temperature.21 Proton donor-acceptor reactions were found to occursimultaneously on the Si and Al sites exposedat basal planes and edges. We found that theSi site acidity at the kaolinite-solutioninterface differs minimally from that of pureSiO2, whereas Al sites became appreciablymore acidic with substitution into thekaolinite matrix. Increasing temperaturecauses both Al and Si sites to become moreacidic, the Si sites more so than the Al sites.Calculated site densities increase withincreasing temperature, suggestingappreciable surface roughening. Thecombination of increasing site acidity anddensity points to kaolinite having a greatersorptive potential at higher temperatures.Measured Cs+ sorption isotherms ontokaolinite surfaces at 25°C and in threebackground electrolyte solutions (0.01-1.0 MNaCl) are shown in Figure 8. These titrationswere conducted from low to high pH andthen reversed from high to low pH. At eachionic strength (IS), there is clear hysteresis inthe sorption behavior. Cs+ is irreversiblysorbed by kaolinite, particularly at pH <6. Inaddition, we see that as the IS of Na+

increases by 2 orders of magnitude, thesorbed Cs+ decreases by about an order ofmagnitude and that the absolute concentrationof Na+ is 5-6 orders of magnitude greater thanthat of Cs+.

We hypothesize that irreversibly-sorbed Cs+

is associated with the siloxane basal surface,which might have a small pH-independentresidual charge. However, the slight buttangible pH dependence of Cs+ sorption atall ionic strengths suggests that some sorbedCs+ is also associated with edge sites(probably aluminols) that possess a pH-dependent charge. Sorption isotherms forSr2+ and Ba2+ onto kaolinite surfaces at 25°Care shown in Figure 9, where >Me denotessorbed metal and the solid line is a best fitusing a constant capacitance surfacecomplexation model. Note that at a givenpH, the relative binding strength of Ba2+ tokaolinite is significantly stronger than thatmeasured for Sr2+ (pKSr = 6.44 vs. pKBa =6.02). Both metals demonstrate the samepH-dependent sorption isotherm at high pH,but significant pH-independent sorption

0.01 M NaCl

pH

2 3 4 5 6 7 8 9 10

Sor

bed

Cs

(mol

/L)

2.6e-5

2.8e-5

3.0e-5

3.2e-5

3.4e-5

0.1 M NaCl

pH

2 3 4 5 6 7 8 9 10Sor

bed

Cs

(mol

/L)

1.3e-5

1.4e-5

1.5e-5

1.6e-5

1.7e-5

1.8e-5

1.9e-5

1.0 M NaCl

pH2 3 4 5 6 7 8 9 10S

orbe

d C

s (m

ol/L

)

0.0e+0

1.0e-6

2.0e-6

3.0e-6

4.0e-6

5.0e-6

Figure 8. Sorption (filled circles) anddesorption (open circles) of Cs+ onto kaoliniteas a function of ionic strength at 25°C.

NUREG/CR-660319

occurs in acid solutions (pH<7). The mostlikely explanation is that metal binding inacid solutions occurs at the basal siloxanelayer because of a small permanent and pH-independent charge arising from minorsubstitution of Al3+ and/or Fe3+ fortetrahedral Si4+. Figure 10 shows the amountof Ba exchanged onto prewashed SWy-1 as afunction of pH and the Ca/Na ratio of thesolution. Table 1 lists the calculated constant

Table 1. Ba/Na Exchange Constants

pH KBa

4.3 2.36.2 6.06.5 4.37.8 5.29.1 5.3

for Ba/Na exchange as a function of pH in0.001M NaCl solutions. Note that the

exchange constant is relatively independentof pH, suggesting that sorption onto edgesites is minor. Note that adding Ca2+ to thesolution decreases the amount of Baexchanged onto the clay. Millimolar levelsof Ca2+ decrease Ba retention by about 40%.Again, there is no clear pH dependenceobserved in the retention of Ba.

Figures 11 and 12 show Ba and Sr exchangeonto Swy-1 that was not washed orpresaturated with Na+. Instead, the basalplanes probably possessed both Ca and Naas opposed to just Na+. The same trends arenoted as in the Na-saturated case; i.e., metalsorption (via exchange) is independent ofpH, and Ca suppresses exchange. The onlyexception to this behavior appears to be Sr

>Ba(mol/L)

100.0

1.0 10 -5

2.0 10 -5

3.0 10 -5

4.0 10 -5

5.0 10 -5

6.0 10 -5

7.0 10 -5

3 4 5 6 7 8 9

[Ba2+]o = 7.8 x 10 -5M

pKBa = 6.02

[Kaol] = 800 m 2 L-1

C = 1 F m -2

25oC 0.1M NaCl, KGa-1

pH

0.0

2.0 10-6

4.0 10-6

6.0 10-6

8.0 10-6

1.0 10-5

1.2 10-5

1.4 10-5

4 5 6 7 8 9 10 11

[Sr2+]o = 1.4 x 10-5M

pKSr = 6.44

[Kaol] = 800 m2 L-1

C = 1 F m-2

25oC 0.1M NaCl, KGa-1 >Sr(mol/L)

Figure 9. Sorption isotherms for Sr and Ba ontokaolinite; the line is best fit for the constantcapacitance surface complexation model.

3 10-6

4 10-6

5 10-6

6 10-6

7 10-6

8 10-6

4 5 6 7 8 9 10

>B

a (

mo

l/l)

pH

[Ba2+]o = 7.3 x 10 -6 mol/l

[SWy-1] = 39.14 m 2/l

NaCl prewashed

25 oC

Na+ = 0.001 mol/l

Ca++ = 0.001 mol/l

Na+ = 0.001 mol/l

Ca++ = 0.005 mol/l

Figure 10. Ba sorption on Swy-1 (pretreatedwith NaCl to remove exchangeable Ca) at 25°Cas a function of solution pH and Ca/Na ratio.

20NUREG/CR-6603

exchange at pH > 7, where appreciabledesorption occurs.

3.4 Task 4: AFM microcharacterizationof clay surfaces

Images of KGa1 kaolinite were obtainedusing contact-mode AFM in air, wherereported particle dimensions were based onimages of only 7 discrete particles (Figure13). More images were not acquired becauseof the difficulty of mounting the kaolinite onmica surfaces and imaging via contact modein air. Practical experience has shown thatthe exceptionally low humidity ofAlbuquerque, NM (~15%) makes imagingsmall particles in air even more difficult.Particles of untreated kaolinite rangefrom 0.1 to 0.8 µm in diameter and100 to 1200 Å in total thickness(average 7-8 unit cells in height; seeFigure 13). Assuming particles arehexagonal in shape and that similarsteps occur on both basal surfaces,we estimate that edge-to-basalsurface area ratios range from 15 to50%. Edge surface areas were

estimated from small steps on theclay surface (Figure 13b), and are

typically 1 to 10 unit cells high (7 to70 Å). They increase the edgesurface area obtained from the grossshape of the grain by a minimum of10%. The additional contribution ofstep edge area depends on the

0

2 10-6

4 10-6

6 10-6

8 10-6

1 10-5

4 5 6 7 8 9

>B

a (

mo

l/l)

pH

Ca2+ = 0.001 mol/l

Na+ = 0.005 mol/l

Ca2+ = 0.001 mol/lNa+ = 0.005 mol/l

no pre-wash

[Ba ]o = 7.3 x 10-6

[SWy-1] = 39.14 m 2/l

25oC

Figure 11. Ba exchange onto unwashed SWy-1.

2 10-6

4 10-6

6 10-6

8 10-6

1 10-5

1.2 10-5

4 5 6 7 8 9

>S

r (m

ol/l

)pH

Ca2+ = 0.001 mol/lNa+ = 0.005 mol/l

Na+ = 0.005 mol/l

Ca2+ = 0.001 mol/l

[Sr++ ]o = 1.14 x 10-5

[SWy-1] = 39.14 m 2/l

25oCno pre-wash

Figure 12. Sr (lower) exchange on unwashed SWy-1.

AB

C 0.2 µm

Edge Area

A: 13%

B: 33%

C: 47%

KGa-1

Figure 13. AFM image of kaolinite crystallites(001) orientation, with well-developedhexagonal habit; upper right crystal has surfacesteps of ~800 Å in height.

NUREG/CR-660321

number and average height of thesteps. The potentiometric titrationresults require a site density of 2.25sites/nm2 to explain the amount ofadsorption assuming that the entireBET surface area sorbs hydrogen orhydroxyl ions. Crystallographic sitedensities are estimated to be 3 to 6sites/nm2. Therefore, if adsorptionoccurs only at edge sites where thedensity of charged sites is highest, aminimum of 38% of the BET surfacearea must be edge surface. Thisamount of edge surface isapproximately the median of therange gleaned from limited AFMobservations. Recent measurements ofparticle dimensions and surface morphologiesof 150 grains of KGa-1 kaolinite led to thesame conclusions as ours. In terms ofparticle statistics, this confirms earlier workon fibrous illites22 where the mean particlethickness was the same for 13 or 60 particles,although particle size distribution was morestatistically significant for the larger samplesize. The particle morphology of SWy-1

montmorillonite also was measured by AFMtechniques (Figure 14). This sample ofmontmorillonite contains trace quartz andcalcite. Its BET surface area, measuredusing N2(g), is 31.82 m2/g. Montmorilloniteparticles range in diameter from a minimumof about 0.1 µm to a maximum of about 1µm, and particle thickness ranges from about2 to 45 nm. The thicker particles appear tobe composed of overlapping thinnerparticles 20 to 30 nm thick. The thinparticles are also extremely flat on theirbasal surfaces. Aspect ratio (thickness todiameter) ranges from about 0.02 to 0.45.Edge surface area ranges from 4 to 9%,much less than that observed for the KGa-1kaolinite. This percentage of edge surfacearea is more in line with values typicallyestimated for clays. Most particles appearirregular in outline with minimal basalsurface topography (i.e., rare unit-cell highsteps). Although samples were sonicated foronly 20-30 sec in an ultrasonic bath, it ispossible that some cleavage and breakageoccurred, particularly in the case of thesmaller, thinner particles.

3.5 Task 5: Sorption measurementsduring Fe-oxide/hydroxide phasetransformation

Barium sorption data from the ferrihydritetransformation experiment at pH 12 and50°C in 0.01 M KNO3 with Ba added arepresented in Figure 15. Analyses of thesuspension filtrate indicate that 100% of theBa added was retained by the iron oxide forat least 25.2 h of aging and that 4 and 7% ofthe Ba was desorbed after 45.9 and 68.4 h ofaging, respectively. Barium desorptionbegan when ferrihydrite transformation togoethite was 92 to 94% complete.Consequently, the amount of Ba sorbed byferrihydrite (BaOx) began to decrease after

25nm

50nm

0nm

1 µm

Height

SWy-1

Figure 14. Tapping mode AFM image ofSwy-1 montmorillonite; particle size rangesfrom 0.1-1 µm.

22NUREG/CR-6603

25.2 h of aging. The Ba sorbed by goethite(BaHCl - BaOx) increased at the beginning ofthe aging period (0.4 to 19.3 h) and thenremained constant as ferrihydrite

transformation neared completion.

Figure 16 compares the percenttransformation of ferrihydrite for samplesaged at 50 °C in 0.01 M KNO3 with andwithout Ba. Although this graph shows aslight decrease in transformation rate insamples with Ba added during the first 12.4h of aging, the decrease is not consistentlygreater than the standard error of the data.After 25.2 h of aging, there is no differencebetween the rate of ferrihydritetransformation with and without Ba. Otherworkers have found that cation sorption mayretard ferrihydrite transformation.19, 23-25

The very slight decrease in transformationrate indicates that addition of 0.0005 M Bawas not sufficient to prevent ferrihydritedissolution or goethite nucleation and crystalgrowth.

Figure 17 displays XRD patterns forsubsamples taken during 68.4 h of agingwithout Ba at 50°C in 0.01 M KNO3, whilethe XRD patterns of subsamples aged withBa during the same time period are shown in

Figure 18. These data also show a slight

Time (h)

0 10 20 30 40 50 60 70

% F

errih

ydrit

e Tr

ansf

orm

atio

n

30

40

50

60

70

80

90

100

0.0005 M BaNO3

Without Ba

Figure 15. Percent transformation [(1-FeOx/FeHCl X 100] of ferrihydrite aged in 0.01M KNO3 at pH12 and 50°C.

(1 - FeOx/FeHCl) X 100

BaHCl - BaOx

BaOx

Ba in filtrate

Time (h)

0 20 40 60

% F

errih

ydrit

e Tr

ansf

orm

atio

n

30

40

50

60

70

80

90

100

mg

Ba S

orbe

d or

in S

olut

ion

g-1

Iron

Oxi

de

0

1

2

3

4

5

6

7

Figure 16. Ba sorption on ferrihydrite at pH 12and 50°C; note the transformation of ferrihydriteto goethite with time and the associated drop in Basorption.

Ferrihydrite Aging Without Ba

15 25 35 45 55 65 75

2 θ (degrees, Cu Kα)

Inte

nsity

4.7 h

2.5 h

0.4 h

12.4 h

25.2 h

45.9 h

68.4 hG

G

G

GG

G

GG G G GGG

G G G G

FF

Figure 17. XRD patterns of ferrihydrite samples aged in0.01 M KNO3 at pH12 and 50°C (F=ferrihydrite; G=goethite).

NUREG/CR-660323

decrease in transformation rate in sampleswith Ba added during the first 12.4 h of aging.No difference can be detected in this studybetween XRD patterns for samples aged withor without Ba from 12.4 to 68.4 h .

Figures 17 and 18 show that some poorlycrystalline goethite formed during the first0.4 h of aging and that the goethite XRDpeaks increase in intensity and decrease inwidth during the first 12.4 h of aging. Thisindicates an increase in the amount ofgoethite present and its crystallinity.Between 12.4 h and 25.2 h of aging, there areno further changes in XRD peak width and aslight increase in peak height, indicating anincrease in the amount of crystalline goethitepresent. After 25.2 h of aging, when thetransformation of ferrihydrite to goethite is 94to 95% complete (Figure 16), there are nofurther changes in the XRD patterns of thesamples. Figure 15 also shows that the

amount of Ba partitioned to goethite (BaHCl -BaOx) appears to reach a constant level afterabout 25.2 h.

Figure 19 displays changes in surface area ofthese samples during aging. The surface

area of samples aged with and without Badecreased rapidly during the first 19.3 h ofaging and then approached a constant level.The surface area decreased rapidly until thetransformation of ferrihydrite to goethiteneared completion (Figure 16). Bariumdesorption may be caused by this decrease insurface area. The surface areas of samplesaged without Ba appear to be slightly lowerthan those of samples aged for the same timewith Ba during the first 19.3 h of theexperiment.

In the pH range of 6 to 10, Ba is known tosorb on ferrihydrite26 and sorbs morestrongly at pH values above its zero point ofcharge (pH > 8). Goethite is thought to formfrom ferrihydrite by dissolution andreprecipitation.27 Although Ba2+ is too largeto replace Fe(III) in the goethite structure,Ba released after ferrihydrite dissolution

Ferrihydrite Aging with Ba

15 25 35 45 55 65 75

2θ (degrees, Cu Kα)

Inte

nsity

2.4 h

4.7 h

12.4 h

25.2 h

45.9 h

68.4 h

0.4 h

G

G

G

GG

G

GGG G G

G

GGGG G GG

F F

Figure 18. XRD patterns of ferrihydrite samplesaged in 0.01 M KNO3 and 0.0005 M BaNO3 atpH12 and 50°C (F = ferrihydrite; G = goethite).

Time (h)

0 10 20 30 40 50 60 70

Surfa

ce A

rea

(m2 g

-1)

40

50

60

70

80

90

100

110

120

130

140

150

Without Ba With Ba

Figure 19. BET surface areas of ferrifydriteaged at 50°C and pH12 in 0.01 M KNO3.

24NUREG/CR-6603

may have been adsorbed and then occludedduring rapid nucleation of goethite crystals.As the goethite crystals grew larger, surfacearea decreased and the number of sites for Basorption and/or occlusion also decreased,causing some Ba to be desorbed. Futureexperiments with longer aging periods willdetermine whether Ba desorption reaches a

constant level or ceases with time. Inparticular, atomic force microscopy will beused to relate ferrihydrite aging with andwithout Ba to changes in crystal size andmorphology. Initial trends indicate that Bamay slow the transformation of ferrihydriteto goethite.

NUREG/CR-660325

4.0 Discussion

One of the primary goals of this study is toidentify specific sites and mechanisms formetal sorption, as well as to quantify thekinetics and magnitude of alkali or alkalineearth sorption and desorption on clay andFe-oxyhydroxide minerals. XPS analyses ofCs-reacted clay minerals suggest that Cs issorbed on montmorillonite and possibly onkaolinite, although the spectral resolution ofXPS analyses is insufficient to differentiateamong basal, edge, or interlayer sites. 133CsMAS- NMR results demonstrate that Cs isadsorbed in a motionally averaged interlayersite of montmorillonite. NMR spectra forthe kaolinite samples, however, show thepresence of as many as four distinct sites.The bulk of these sites is thought to consistof edge and basal plane sites, in addition tosome smectite-like “exchangeableinterlayer” sites. The latter probably resultfrom the presence of small quantities ofsmectite-like layers on the kaolinitecrystallites. Given the sheet-like structure ofkaolinite, exposed Si and Al sites at thecrystal’s edge would provide likely sorptionsites suggested by the 133Cs NMR spectra. Itis not surprising that the NMR results formontmorillonite indicate a continuum of Cssorption sites, given the basal planeinterlayer exchange capacity of smectite.

Metal sorption onto clay surfaces isa function of the charge distributionon the surfaces as well as theabsolute proportions of exposedplanes of differing surface charge.Although total surface areas can bemeasured using routine BETmethods, the best way to determineproportions of edge and basalsurfaces is to use an imagingtechnique in which all threedimensions can be determined on

individual particles (e.g., AFM).25

Quantification by AFM of edge andbasal surface areas for kaoliniteshowed that edges can comprise asmuch as 50% of the total BETsurface area. This is in contrast tountreated SWy-1 montmorillonite,where edge surface areas typically measure<10%, much less than that observed for theKGa-1 kaolinite. This percentage of edgesurface area for smectite clays is more inline with values typically estimated forclays. Obviously, better statisticalcharacterization of surface areaproportions for a variety of clays andother fine-grained soil phases couldsignificantly affect the interpretationof both spectroscopic andexperimental sorption data.However, it is apparent that thedecreased grain size and thinnerclay platelets of the montmorillonitesample result in increased surfacearea and available basal reactivesites. In addition, smectites havevery reactive basal surface sites dueto heterovalent octahedral andtetrahedral lattice substitutions,which induce a permanent charge tothe (001) surfaces.

Surface complexation modeling of protonpotentiometric measurements of kaolinite andits component oxides suggest that metalbinding on clay minerals occurs on multiplesites. Of the four sites indicated by the NMRwork, only two dominate the overall sorptivecapacity of kaolinite. The surface charge andmetal sorption properties of kaolinite werefound to be largely pH dependent, reflectingthe acid-base properties of constituent Si- andAl-oxide and hydroxides. Regressed surfacecharge equilibrium constants and site

26NUREG/CR-6603

densities for KGa1, using a multisitereactivity model (both Al and Si on edge andbasal plane sites), suggest that either thesiloxane basal plane plays a greater role inoverall surface charge than would beexpected for fully satisfied siloxane bonds orthat the number of pH dependent edge sites isunderestimated.21 However, AFM studies ofKGa1 particle surfaces have shown that up to50% of kaolinite edges sites are exposed,21

where most of the KGa1 surface charge isattributed to a combination of exposed Al-and Si-sites on the basal plane and edgeswithout calling upon proton sorption ordesorption from the siloxane basal plane.This is consistent with an interpretation ofH+/OH- adsorption occurring primarily atcharged edge sites, and by implication, withmetal sorption also occurring primarily atedge sites. However, multivalent octahedraland tetrahedral cationic substitutions thatcommonly give rise to the compositional andsurface charge variability of expandable TOTlayer clays (smectites) will probably posespecial problems when predicting metalsorption properties.

In order to simulate, at an atomistic level,specific site acidity at edge and basal planesites in kaolinite and metal sorptionreactions, clay structures for realisticcompositions need to be ascertained eitherfrom experimental diffraction, or ab initiosimulation techniques. For the first time, thestructural features of kaolinite have beenaccurately predicted by LDF molecularorbital methods, including the general C-centering and C1 symmetry in the purecrystal as well as the hydroxyl orientationsin the interlayer region.12 Further, thecalculated O-H bond lengths and hydrogenbonds correlate well with observed single-crystal infrared spectra of kaolinite.28

Empirical MD simulations suggest that Cs+

is sorbed on kaolinite at aluminol (010) edgesites as an inner-sphere complex and weakly

sorbed as an outer-sphere complex on (001)basal surfaces. These computer simulationsare consistent with results of sorption anddesorption isotherms as a function of ionicstrength. MEP calculations also show thatthere is a significant difference in theelectrostatic potential of kaolinite edge andbasal surface sites,21 where siloxane andgibbsite basal planes (001) are not asnegatively charged as edge silanol oraluminol sites (010), and aluminol edge siteshave greater negative electrostatic chargethan edge silanol sites. The combination ofcharge distributions calculated frompotentiometric titrations, proportions of edgeand basal plane areas measured from AFMimages, and MEP surface charge densitiescalculated from atomistic simulationsprovides a consistent picture of distinctcrystallographic sites that control kaolinitesurface charge and reactivity. That is,aluminol edge sites are preferred metalsorption sites over edge silanol or basal Si-or Al-sites. It should be noted that a smallpermanent negative charge may exist on thebasal planes due to ionic substitutions in theoctahedral or tetrahedral layers.

Relative metal binding strengths over allkaolinite sites were found to decrease fromBa2+ to Sr2+ to Cs+, where metals are sorbedon both kaolinite edges and basal surfaces.Some fraction of these metals also appearsto be irreversibly sorbed and cannot beeasily desorbed from the kaolinite surface.For unwashed montmorillonite, the relativebinding strength for Ba is slightly greaterthan that for Sr. Also for montmorillonite,metal sorption appears to occur via a simpleexchange reaction and is independent of pH;the presence of Ca in solution seems tosuppress cation exchange. The onlyexception to the pH-independent metalsorption behavior appears to be Sr exchangeat pH > 7, where appreciable desorptionoccurs. The Sr behavior is anomalous, and

NUREG/CR-660327

we do not at present have an explanation forit.

Barium is known to sorb on ferrihydrite inthe pH range of 6 to 10 and sorbs morestrongly at pH values above its zero point ofcharge (pH > 8).24 Goethite is thought toform from ferrihydrite by dissolution andreprecipitation.25 Although Ba2+ is too largeto replace Fe(III) in the goethite structure, Bareleased after ferrihydrite dissolution mayhave been adsorbed and then occluded duringrapid nucleation of goethite crystals. As thegoethite crystals grew larger, surface area

decreased and the number of sites for Basorption and/or occlusion also decreasedcausing some Ba to be desorbed. Futureexperiments with longer aging periods willdetermine whether Ba desorption reaches aconstant level or ceases with time. Inparticular, atomic force microscopy will beused to relate ferrihydrite aging with andwithout Ba to changes in crystal size andmorphology. Initial trends indicate that Bamay slow the transformation of ferrihydrite togoethite.

28NUREG/CR-6603

5.0 Conclusions

A combination of analytical, experimentaland theoretical techniques was used tocharacterize the sorption mechanism for Cs+,Ba2+, and Sr2+ onto several common soilminerals, including kaolinite andmontmorillonite clays as well as goethiteand ferrihydrite Fe-oxyhydroxide phases.Most of these efforts were devoted to amechanistic understanding of Cs sorptiononto kaolinite, and provide a basis forunderstanding and predicting metal sorptiononto simple clays and a general frameworkfor characterizing the sorption of othermetals onto more complex soil minerals.Sorption models describing surface

complexation reactions are based onmeasured adsorption isotherms, bulksolution-mineral surface complexationreactions, and molecular or atomistic modelsof the mineral-solution interface. Careclearly should be exercised when applyingthese preliminary models to contaminatedsites because they are only a hypotheticalinterpretation of metal sorption in verycomplex systems and are undoubtedly,affected by multiple physicochemicalprocesses.

NUREG/CR-660329

6.0 References

1. Kim, Y., Cygan, R. T., and Kirkpatrick,R. J., 133Cs NMR and XPS investigationof Cs adsorbed on clay minerals andrelated phases, Geochimica etCosmochimica Acta, 60, 1041-1052,1996.

2. Kim Y., Kirkpatrick, R. J. and Cygan, R.T., 133Cs NMR Study of Cs on theSurfaces of Kaolinite and Illite,Geochimica et Cosmochimica Acta, 60,4059-4074, 1996.

3. Kim, Y., Kirkpatrick, R. J., and Cygan, R.T., 133Cs NMR study of Cs reaction withclay minerals, In R. A. Palmer and V.Jain, Eds., Environmental Issues andWaste Management Technologies inCeramic and Nuclear Industry, Am.Ceramic Soc., Columbus, 629-636, 1995.

4. Westrich, H. R., Cygan, R. T., Brady, P.V., Nagy, K. L., Anderson, H. L., Kim,Y., and Kirkpatrick, R. J., The sorptionbehavior of Cs and Cd onto oxide andclay surfaces, In Proceedings of the WasteManagement Conference, WM’95, 24-4,1995.

5. Brady P. V., Silica surface chemistry atelevated temperatures, Geochimica etCosmochimica Acta, 56, 2941-2946,1992.

6. Brady P. V., Alumina surface chemistry at25, 35, and 60 C, Geochimica etCosmochimica Acta, 58, 1213-1217,1994.

7. Casey, W. H., Westrich, H. R., Banfield,J. F., Ferruzzi, G., and Arnold, G. W.,Leaching and reconstruction at thesurfaces of dissolving chain-silicateminerals, Nature, 366, 253-256, 1993.

8. Nagy, K. L., Dissolution andprecipitation kinetics of sheet silicates,In Chemical Weathering Rates ofSilicate Minerals, Reviews inMineralogy Vol. 31, Mineral. Soc. Am.,173-233, 1995.

9. Westrich, H. R., Cygan, R. T., Arnold,G. W., Zemitis, C., and Casey, W. H.,The dissolution kinetics of mixed-cationorthosilicate minerals, Am. Journal ofSci., 293, 1-25, 1993.

10. Bish, D. L. and Von Dreele, R. B.,Rietveld refinement of non-hydrogenatomic positions in kaolinite, Clays andClay Minerals, 37, 289-296, 1989.

11. Hess, A. C. and Saunders, V. R.,Periodic ab initio Hartree-Fockcalculations of the low-symmetrymineral kaolinite, Journal of PhysicalChemistry, 96, 4367-4374, 1992.

12. Bish, D. L., Rietveld refinement of thekaolinite structure at 1.5 K, Clays andClay Minerals, 41, 738-744, 1993.

13. Young, R. A. and Hewat, A. W.,Verification of the triclinic crystalstructure of kaolinite, Clays and ClayMinerals, 36, 225-232, 1988.

14. Giese, R. F. and Datta, P., Hydroxylorientations in kaolinite, dickite, andnacrite, American Mineralogist, 58,471-479, 1973.

15. Westall, J. C., FITEQL - A computerprogram for determination of chemicalequilibrium constants fromexperimental data, Report 82-02, Dept.of Chemistry, Oregon State University,Corvalis, OR, 1982.

30NUREG/CR-6603

16. Van Olphen, H., and J. J. Fripiat, DataHandbook for Clay Materials and otherNon-Metallic Minerals, Pergamon Press,New York, 1979.

17. Schwertmann, U., and Murad, E., Theeffect of pH on the formation of goethiteand hematite from ferrihydrite, Clays andClay Minerals, 31, 277-284, 1983.

18. Schwertmann, U., Differenzierung dereisenoxide des bodens durch extraktionmit ammoniumoxalat-lösung, Z.Pflanzenernähr. Bodenkd., 105, 194-202,1964.

19. Cornell, R.M., The influence of somedivalent cations on the transformation offerrihydrite to more crystalline products.,Clays and Clay Minerals, 23, 329-332,1988.

20. Bish, D. L. and Johnston, C. T., Rietveldrefinement and Fourier-transform infraredspectroscopic study of the dickitestructure at low temperature, Clays andClay Minerals, 41, 297-304, 1993.

21. Brady P. V., Cygan, R. T., and Nagy, K.L., Molecular controls on kaolinitesurface charge, Journal of Colloid andInterface Science, 183, 356-364, 1996.

22. Nagy, K. L., Application ofmorphological data obtained usingscanning force microscopy toquantification of fibrous illite growthrates, In Scanning Probe Microscopy ofClay Minerals, K. L. Nagy and A. E.Blum, Eds., The Clay Minerals SocietyWorkshop Lectures, Vol. 7, pp. 203-239,1994.

23. Axe, L., and Anderson, P. R., Sr diffusionand reaction within Fe oxides: Evaluationof the rate-limiting mechanism forsorption, Journal of Colloid and InterfaceScience,175, 157-165, 1995.

24. Baltpurvins, K.A., Burns, R.C.,Lawrance, G. A., and Stuart, A. D.,Effect of Ca2+, Mg2+, and anion type onthe aging of iron (III) hydroxideprecipitates, Environmental Science andTechnology, 31, 1024-1032, 1997.

25. Ford, R.G., Bertsch, P.M., and Farley,K.J., Changes in transition and heavymetal partitioning during hydrous ironoxide aging, Environmental Science andTechnology, 31, 2028-2033, 1997.

26. Kinniburgh, D.G., Jackson, M. L., andSyers, J.K., Adsorption of alkalineearth, transition, and heavy metalcations by hydrous oxide gels of ironand aluminum, Soil Science Society ofAmerica Journal, 40, 796-799, 1976.

27. Schwertmann. U., and Fischer, W., ZurBildung von α-FeOOH und α-Fe2O3 ausamorphem Eisen(III) hydroxide, Z.Anorg. Allg. Chem., 346, 137-142,1966.

28. Johnston, C. T., Agnew, S. F., and Bish,D. L., Polarized single-crystal Fourier-transform infrared microscopy of Ouraydickite and Keokuk kaolinite, Clays andClay Minerals, 38, 573-583, 1990.