ClayMinerals, (2013)48,309–329 Influence of alkaline (pH 8.3(cid:2)12.0) and saline solutions on chemical, mineralogical and physical properties of two different bentonites T. HEIKOLA1,*, S. KUMPULAINEN2, U. VUORINEN1, L. KIVIRANTA2 AND P. KORKEAKOSKI3 1VTTTechnicalResearchCentreofFinland,Otakaari3KEspoo,02044VTT,Finland,2B+TechOy,Laulukuja4, 00420Helsinki,Finland,and3PosivaOy,Olkiluoto,27160Eurajoki,Finland (Received4December2012;revised25March2013;Editor:JohnAdams) ABSTRACT:Theinteractionoftwodifferentbulkbentonites(Na-andCa-types)withthreetypes ofsimulatedcementwaters(pH9.7,11.3and12.0)andonesalinegroundwatersimulate(pH8.3)as areference,wasstudiedinbatchreactorsat25ºC.ThesolutionpHwasmonitoredinordertokeep the pHas steady aspossible byreplacing theleaching solution with freshonewhen needed. After 554days, onesetofparallel sampleswasremoved fromtheexperimentinordertoinvestigatethe possiblechanges inthebentonitematerials. Thebufferingcapacityofbentonitewasclearlyobserved,especiallyatthebeginningofthe high-pHexperiments, as thepH oftheleachingsolutions decreased quitedramatically due to interactionwith bentonite.The solutionchemistryresultsshowedadecreaseofCa contentin all leachate samples, but especially in pH 12.0 experiments. Small amounts of silica were released throughout the experiment. Both bentonites in pH 12.0 experiments also released detectableamountsofAl,whileinthelowerpHexperimentsthelevelswerebelowdetection limit. These observations were also supported by chemical analyses of the bentonite materials. Only minor changes were detected in the mineralogy, and they were mainly concentrated on experiments at pH 11.3 and pH 12.0. The measured swelling pressure showed an increase in pH 12.0 experiments. The results obtained in this research may facilitate modelling of bentonite interaction with high-pH solutions. KEYWORDS: bentonite buffer, alteration, batch, bentonite, alkaline, saline spent nuclear fuel, underground repository. In Finland, disposal of spent nuclear fuel is buffer and backfill materials. Construction of the planned in a deep bedrock repository based on a repository will also need various supporting multi-barrier concept. Bentonites and smectite-rich cementitious structures. In saturated groundwater clays,e.g.Friedlandclay,areintendedtobeusedas environments,degradationofcementitiousmaterials produces chemically aggressive and highly alkaline solutions which may affect the established safety functions of the bentonite by inducing mineralo- *E-mail:[email protected] gical and chemical changes over considerable DOI:10.1180/claymin.2013.048.2.12 periods of time. #2013TheMineralogicalSociety Downloaded from https://pubs.geoscienceworld.org/claymin/article-pdf/48/2/309/3312912/gsclaymin.48.2.12-hei.pdf by guest on 10 April 2019 310 T.Heikolaetal. In order to reduce the effects of a high-pH bentonite, diffusion controls the mass transport and plume, investigations for development of low-pH hence the amount of bentonite that can dissolve. cement formulations started in 2002, among others Furthermore, possible mineral alteration that takes in a joint project between Posiva, SKB and NUMO place may cause a change in the porosity and as a (Bode´n & Sieva¨nen, 2005). The aim of the project result change the hydraulic conductivity and was to achieve at least one well-quantified, tested thereby affect solute diffusivities. Mineral dissolu- andapprovedlow-pHinjectiongrouttobeusedina tion rates are known to increase with pH and/or repository. The target was to obtain material with a temperature, but considerable uncertainties exist poresolutionpH411.ThepHoftheporesolution concerning the montmorillonite dissolution kinetics depends on the phases that are present. In OPC at a high pH range. The rate of dissolution may be (Ordinary Portland Cement) materials the pore inhibited by the dissolved Si or otherwise affected solution pH stays high (>12.5) as long as alkali by the presence of the secondary minerals. hydroxides and portlandite are present. The pH The objective of this study was to investigate the value of the pore solution starts to decrease when buffering capacity of bentonite against high-pH dissolution of C-S-H phases are involved. In the solutions and the possible alteration of bentonite in low-pH materials pozzolans such as silica fume, alkaline cement leachates. Batch experiments with which is very reactive, are used to bind most of the two different bulk bentonites at 25ºC were started portlandite toformC-S-Hphasesand therebylower inordertogaininformationonthesequestions.The the pH of cement leachates that are less harmful to pH values chosen bound the pH domain of interest bentonite clay (Kronlo¨f, 2005). whenlow-pHcementitiousmaterialsareconsidered. Bentonites are predominantly composed of the clay mineral smectite which is the key mineral in METHODS respect to the set safety functions of bentonite, e.g. high expandability. It is known that the most Solution pH-values were determined with a important parameter affecting the dissolution of commercial glass electrode (ROSS1 Sure-flow1) smectites is pH, and that smectite dissolution and standard calibration solutions (pH 7, 10 and becomes significant at pH > 13. On the other 13). The solution chemistry was determined using hand, Kaufhold & Dohrmann (2011) have shown inductively coupled plasma atomic emission spec- that bentonites are surprisingly resistant against troscopy (ICP-AES: Ca, Mg, Na, Si), flame atomic solutions at pH around 12 and up to 90ºC. When absorption spectroscopy (FAAS: K), ion chromato- considering long periods of time in the scale of a graphy (IC: SO , Br), titration (Cl) and high- 4 repository performance and the modelling needs of resolutioninductivelycoupledplasmamassspectro- assessing the dissolutionandprecipitationprocesses metry (HR-ICP-MS: Al, Fe). Mineralogical compo- in systems of bentonite and alkaline solutions, it is sitions of the bentonites were studied with Philips also important to know the processes even at X’Pert MPD diffractometer (XRD), using Rietveld pH < 12.5. The interaction between bentonite clay refinement, Fourier transform infrared spectroscopy andcementporefluidsisacomplicatedprocessand (FTIR) and quantitative Greene-Kelly testing. anumberofuncertaintiesandlimitationsneedtobe Oriented XRD mounts were prepared from sepa- considered. Savage & Benbow (2007) have rated clay fractions (<2 mm) using filter-membrane provided a comprehensive review of the factors peel-off technique. The mineralogical sample that contribute to the degradation of bentonite due preparationandmeasurementmethodsaredescribed to interaction with cement. According to them, in more detail in Kiviranta & Kumpulainen (2011). there are a number of potential constraints, The chemical composition of the bentonites was including mass balance, thermodynamics, mass determined by analysing LOI (loss on ignition), transport and kinetics. Lack of knowledge of the concentrations of elements using ICP-AES and type of secondary minerals formed complicates carbon content with combustion (Leco). The mass balance issues, as the minerals may consume cation exchange capacity of bentonite was or generate hydroxide ions. From the thermody- measured spectroscopically at 620 nm from the namic point of view the variation of silicon supernatantafterCu(II)-triethylenetetramineabsorp- speciation and quartz solubility with pH signifi- tion (Meier & Kahr, 1999; Ammann et al., 2005). cantly affects the hydrolysis reactions of mont- The amount of exchangeable cations was deter- morillonite and, thus its stability. In compacted, mined with ICP-AES and FAAS after NH Cl 4 Downloaded from https://pubs.geoscienceworld.org/claymin/article-pdf/48/2/309/3312912/gsclaymin.48.2.12-hei.pdf by guest on 10 April 2019 Influence of alkaline and saline solutions on bentonites 311 extraction (Belyayeva, 1967; Jackson, 1975). The Solutions swelling pressure of bulk bentonite was measured by compacting air-dry material to a fixed-volume Three types of simulated cement waters of swelling pressure cell (24(cid:2)25 mm in diameter) different pH values (9.7, 11.3 and 12.0) and one with a sample density of approximately 1.6 g/cm3. saline groundwater simulate (pH 8.3) were used as In order to compare clay materials of different leaching solutions as a reference. The pH values densities the effective montmorillonite dry density were chosen to demarcate the pH values of (EMDD) was calculated after Dixon et al. (2002). designed low-pH grout pore solution values, both The measurement method is described in detail in at earlier and later stage in the course of cement Kiviranta & Kumpulainen (2011). degradation. The compositions were based on earlier cement leaching tests by Vuorinen et al. (2005) and Heikola (2009). The EQ3 code for MATERIALS aqueous geochemistry (Wolery, 1983) was used to Bentonite calculate the equilibrium composition of the solutions in the absence of CO . In order to avoid 2 Two bentonites, Wyoming-type Na-bentonite and CO interference in the experiments with high-pH 2 Ca-bentonite from Milos, Greece, being considered solutions the experiments were conducted inside an for use in the underground nuclear waste facility in anaerobic glove-box (O <1 ppm and low CO ). 2 2 Olkiluoto (Finland), were studied. Both materials The composition of the leaching solutions and each were provided by Posiva Oy. The initial bentonite pH are given in Table 2. In order to simplify materials were characterized in Finland by B+Tech modelling of cement-bentonite interaction, some of (Kiviranta & Kumpulainen, 2011) and are referred the minor components were omitted from the to later in the text as the initial materials/samples. leaching solution recipes, as well as components The mineralogical composition of the samples as which could indicate dissolution of bentonite (Al, well as the water ratios, cation exchange capacities Mg). Nevertheless, silicon was included in the andswellingindexesofthebulkmaterialsaregiven recipes even if it is expected to indicate dissolution in Table 1. of bentonite, because pore-water leached from low- TABLE 1. Mineralogical composition (in wt.%), water ratios, CEC and swelling indexes of bulk materials (Kiviranta & Kumpulainen, 2011). Wyoming bentonite Milos bentonite Smectite 88.2 79.4 Quartz 3.5 1.0 Illite 0.1 3.2 Cristobalite 0.1 Plagioclase 2.9 1.8 Calcite 0.2 4.3 Dolomite 5.7 K-feldspar 2.4 0.6 Biotite 0.3 tr Chlorite 0.4 0.9 Hematite 0.1 0.7 Pyrite 0.8 1.5 Opal-A 0.3 0.4 Rutile 0.5 0.4 Gypsum 0.4 Water ratio (%) 12.3 17.0 CEC (eq/kg) 0.863 0.909 Swelling index (ml/2 g) 22.0 10.9 tr = present as traces Downloaded from https://pubs.geoscienceworld.org/claymin/article-pdf/48/2/309/3312912/gsclaymin.48.2.12-hei.pdf by guest on 10 April 2019 312 T.Heikolaetal. TABLE 2. Composition of the solutions (meq/l). Experiment => ol-gw8.3 cem9.7 cem11.0 cem12.0 pH 8.3 9.7 11.3 12.0 Na+ 208 205 210 235 K+L 0.6 0.8 0.9 0.9 Ca2+ 200 200 206 218 SiO (aq) (mmol/L) – 0.04 0.02 0.004 2 Cl(cid:2) 408 408 421 432 Br(cid:2) 1.2 1.2 1.2 1.2 Ionic strength 0.50 0.50 0.52 0.53 pH cement materials contains silicon (Heikola, appropriate solution. The samples were placed on 2009). a platform shaker in order to maintain good contact All leaching solutions were prepared by first of bentonite materials with the solutions. The adding the main chemicals NaCl and CaCl to evolution of pH in the samples was followed by 2 MilliQ water (water purified with Millipore Milli-Q measuring the pH value in the solution phase labwatersystem,ISO3696)andflushedwithN or approximately twice a week.The leaching solutions 2 Ar (grade 6.0) in a vessel fitted with two quick- were renewed once a month. For each renewal, couplings allowing flushing with gas. (Note: there centrifugation was used to separate the phases, then was a change of the inert gas used, from N to Ar). the solution was withdrawn and its chemical 2 Thereafter the solutions were brought into the composition analysed. All the solutions were ultra- glove-box and the other minor components were filtered(Macrosep1AdvanceCentrifugalDevices, addedtothesolutionsfromstocksolutionsprepared 10K) before analysis in order to remove possible and kept in the glove-box. Also the stock solutions colloids. The amount of solutions exchanged in the had been flushed with Ar or N prior to placing experiments was approximately 145 ml/g of bento- 2 them in the glove-box. nite at the end of the experiment. This is much more than expected in repository conditions; e.g. Suzuki et al. (2008) have estimated that the L/S in EXPERIMENTAL SET-UP real repository environment is 0.2 ml/ g. However, Preparation of the bentonites for the experiments the target of our experiments was not to simulate began by drying batches of each bulk material in a actual repository conditions, but to gain knowledge temperature cabinet at 105ºC overnight. After that of bentonite alteration and buffering capacity in the bentonites were allowed to cool in a desiccator alkaline conditions. and then each sample was transferred into a glass bottle and closed with a cap fitted with quick RESULTS couplings to enable flushing with nitrogen gas (grade N 6.0). Both bentonites were flushed for pH evolution 2 two days in order to remove oxygen and carbon dioxide, and then transferred into the glove-box. In The results of the pH measurements are depicted order to remove any moisture and remaining in Fig. 1. In the Olkiluoto groundwater type oxygen each bottle was emptied over a wide solution (pH 8.3 experiment) the pH-values metal vessel and heated overnight on a hot plate. showed rather constant values around the initial The oxygen release was monitored with an oxygen value. In the pH 9.7 experiment the pH drop from probe (Orbisphere) to ensure that no more oxygen the initial solution value was distinct, about one pH was released. unit and remained rather constant throughout the In each experiment the solid to solution ratio experiment. A rather distinct wave-like pattern was used was 1/10. The individual samples were detectedinthehigherpHexperiments(pH11.3and prepared by weighing 20 g of dried bentonite into pH 12.0),asa quick dropof pH values followed by a centrifuge bottle and adding 200 g of the gradual levelling off was observed after each Downloaded from https://pubs.geoscienceworld.org/claymin/article-pdf/48/2/309/3312912/gsclaymin.48.2.12-hei.pdf by guest on 10 April 2019 Influence of alkaline and saline solutions on bentonites 313 FIG. 1. Measured pH values in the leaching solutions (vertical dashed lines indicate the time of the solution exchange). Leachate chemistry renewaloftheleachates,especiallyatthebeginning of the experiments. This decrease in the pH values Some results of the chemical analyses of the diminished gradually during the experiment but a leaching solutions are presented in Fig. 2. As slight overall increase in the pH values was expected, ion-exchange processes in bentonites observed in the pH 11.3 experiments, whereas caused depletion of Ca and increase of Na at the towards the end of the pH 12.0 experiment the pH beginning of the test in all the experiments, but values showed an overall levelling off. more distinctly in Na-bentonite experiments. In As a general comment, of all the pH values general, depletionof Ca in the leachates diminished measured it can be stated that pH values in the Ca- quite quickly except in the pH 12.0 experiments bentonite experiments were somewhat lower which showed slight continuous Ca depletion. In compared to those of Na-bentonite, although the addition toNa release,K and Mg were also quickly difference decreased with time (ecpecially in released in the first couple of days in all experiment at pH 12.0). This behaviour may be experiments; one exception was the pH 12.0 due to the higher carbonate mineral content of Ca- experiment of Ca-bentonite, which showed constant bentonite, which plays a major role in the pH release of small amounts of K throughout the buffering capacity of bentonite. experiment. In the case of Mg, approximately four Downloaded from https://pubs.geoscienceworld.org/claymin/article-pdf/48/2/309/3312912/gsclaymin.48.2.12-hei.pdf by guest on 10 April 2019 314 T.Heikolaetal. Downloaded from https://pubs.geoscienceworld.org/claymin/article-pdf/48/2/309/3312912/gsclaymin.48.2.12-hei.pdf by guest on 10 April 2019 Influence of alkaline and saline solutions on bentonites 315 FIG. 3. XRD patterns of bulk Na-bentonite samples. times more was present in the leachates of experiment, while in the other experiments the Ca-bentonite experiments compared to those of levels were below detection limit. Na-bentonite, which can be explained by the higher initial content of Mg in Ca-bentonite. Another Bentonite samples observation of the Mg contents was that more was present in the leachates of the pH 8.3 and 9.7 So far only one set of three sets of parallel experiments compared to those of the higher pH samples in the experiment was analysed after 554 experiments. InthepH8.3andpH 9.7 experiments, days from starting the experiments. The results of small amounts of Si were released relatively the analyses are given here. steadily from both bentonites, whereas in the pH XRD: Bulk samples. Randomly oriented XRD 12.0 experiments an abrupt release occurred only in patterns of bulk materials are presented in Figs 3 the first couple of days, while the pH 11.3 and 4. The Na-bentonite samples showed a small experiments showed a different pattern, an increase decrease in the position of the smectite d line 101 of Si content to a maximum around 40(cid:2)50 days along the pH series. The most noticeable change from the start was observed and thereafter a was observed in the pH 12.0 experiment in which continuous decrease up to 150 and 380 days (Na- the intensity of calcite (CaCO ) lines increased as 3 andCa-bentonite,respectively)until levelling offto well as the appearance of vaterite (CaCO ) lines in 3 a rather steady release towards the end. This trend boththepH11.3and12.0experiments.Anincrease was more notable in Ca-bentonite as more Si was in intensity of the calcite line and a decrease in released. An increase in SO concentration in the intensityof thesmectite lines werealsoobserved in 4 leachates was observed only within the first couple the Ca-bentonite samples in the pH 12.0 experi- of days and almost three times more SO was ment. In the pH 9.7 and 12.0 experiments, there 4 released in Na-bentonite experiments compared to weresmalladditionallinesthatcouldbeassignedto those of the Ca-bentonite experiments. Both initial laumontite (CaAl Si O 4H O), a member of the 2 4 12 2 bentonites contained only small amounts of soluble zeolite group. sulfate according to Kiviranta & Kumpulainen The XRD patterns of samples in the d region 060 (2011), but the gypsum content of Na-bentonite are presented in Fig. 5. The position of the d 060 was higher even if small. Detectable amounts of Al lines decreased slightly in Na-bentonite in the were released from both bentonites in the pH 12.0 pH 11.3 and pH 12.0 experiments, and in FIG. 2 (facing page). Chemical composition (Ca, Mg, SiO2, SO42–) of the leaching solutions of both bentonites (Na-bentonite on the left hand side, Ca-bentonite on the right hand side). Downloaded from https://pubs.geoscienceworld.org/claymin/article-pdf/48/2/309/3312912/gsclaymin.48.2.12-hei.pdf by guest on 10 April 2019 316 T.Heikolaetal. FIG. 4. XRD patterns of bulk Ca-bentonite samples. Ca-bentonite in the pH 12.0 experiment. A totally XRD: Oriented clay fractions. During the new line (or rather humps) appeared at 1.52 A˚ in preparation of the oriented XRD mounts, both the pH 12.0 experiments for both materials (Fig. 5, bentonites in the pH 12.0 experiments were Table 3). This might indicate the presence of observed to have changed to lighter colours. trioctahedral clay minerals such as saponite These samples (both bentonites at pH 12.0) also (Mg-rich smectite). However, carbonates have resisted remaining in dispersion, and they addition- overlapping lines in that same region, and due to ally fractured or peeled off during drying, thus increases in carbonate content in the bentonite preventing the preparation of a proper mount. As samples,itwashypothesizedthatthemoreprobable proved by chemical analysis and oriented XRD reason for the development of the 1.52 A˚ lines was patterns(Fig. 6,Tables3and6),accessoryminerals the presence of carbonates. were still present despite the separation of the clay FIG. 5. XRD patterns of Na-bentonite (left) and Ca-bentonite (right) sample series at d060 region. Downloaded from https://pubs.geoscienceworld.org/claymin/article-pdf/48/2/309/3312912/gsclaymin.48.2.12-hei.pdf by guest on 10 April 2019 Influence of alkaline and saline solutions on bentonites 317 ngthemethod Interpretedminerals S,Q,CrS,Q,CrS,Q,CrS,Q,Cr,Plag.S,Q SS,Plag.S,Plag.S,K-Fs,Plag.S,Q,Plag. usi I/I/I/I/I/ I/I/I/I/I/ d e at C alcul 550ºd001 9.569.709.729.759.56 9.809.799.599.629.72 c S I/ n i S nterlayers ———I%inI/ 0.91.42.22.42.3 3.85.65.35.34.1 i — (I) — illite ——d003 5.575.595.605.645.60 5.545.625.575.565.60 of — nt G u E mo — ar, andthea(1989). ————d002 8.358.408.458.538.44 8.348.548.418.408.47 K-feldsp mectite(I/S),&Reynolds ———d001 16.4116.2516.7117.5716.60 16.4017.6716.5716.2416.94 orite,K-Fs= ationofillite/sofMoore Orientedd001 13.9913.7513.9914.9614.52 14.3814.9314.2414.1814.59 uartz,Cl=chl c q entifi Q= id e, ˚portantlines(inA)usedin BulksamplesRandomlyorientedd060 1.4971.4951.4971.4921.492;1.515 1.4981.4961.4961.4971.494;1.517 Kumpulainen(2011).e/smectite,Cr=cristobalit im &illit of a= T3.PositionsABLE TreatmentLine/interpretation Na-bentoniteInitialsample*pH8.3pH9.7pH11.3pH12.0 Ca-bentoniteInitialsample*pH8.3pH9.7pH11.3pH12.0 DatafromKivirantAbbreviations:I/SPlag.=plagioclase Downloaded from https://pubs.geoscienceworld.org/claymin/article-pdf/48/2/309/3312912/gsclaymin.48.2.12-hei.pdf by guest on 10 April 2019 318 T.Heikolaetal. FIG. 6. XRD patterns of clay fractions after different treatments (Na-bentonite left, Ca-bentonite right). fraction. Carbon analyses showed that the carbon indicate the presence of carbonate and/or CSH content had increased at pH 12.0, and thus phases; secondary minerals contained carbonates. (3) an increase in water absorption bands in the FTIR. The results from FTIR analyses are given pH 12.0 experiments in both bentonites at in Fig. 7 and in Table 4. The results of both 3430 cm(cid:2)1 and 1630 cm(cid:2)1, and; bentonite materials (Wyoming and Milos) show (4) a decrease in transmissivity at four changes in the pH series towards the alkaline 3700(cid:2)2500 cm(cid:2)1. samples: Most changes that occurred in the pH 12.0 (1) a decrease in smectite OH-bendings over the experiments can be explained by the presence of wavelength region 1000(cid:2)500 cm(cid:2)1, especially in carbonates, which cause bands at 1482 cm(cid:2)1, samples in the pH 12.0 experiments; 1425 cm(cid:2)1 and 865 cm(cid:2)1 for amorphous calcite (2)thedevelopmentofbroadbandsat1497 cm(cid:2)1 (Wilson, 1987) and at 1490 cm(cid:2)1, 1420 cm(cid:2)1 and and 1420 cm(cid:2)1, and additionally in the Na- 870 cm(cid:2)1 for vaterite (Sato & Matsuda, 1969), or bentonite sample in the pH 12.0 experiment a by CSH or a calcium hydroxide (portlandite) weak band at 958 cm(cid:2)1 was noted, which could phases, which according to Delgato et al. (1996) Downloaded from https://pubs.geoscienceworld.org/claymin/article-pdf/48/2/309/3312912/gsclaymin.48.2.12-hei.pdf by guest on 10 April 2019