Experimental Synthesis of Chlorite from Smectite at 300ºC in the Presence of Metallic Fe D

Home , Amesite, Chlorite group

Experimental synthesis of chlorite from smectite at 300ºC in the presence of metallic Fe D. Guillaume, A. Neaman, M. Cathelineau, Régine Mosser-Ruck, C. Peiffert, Mustapha Abdelmoula, J. Dubessy, F. Villieras, A. Baronnet, N. Michau

To cite this version:

D. Guillaume, A. Neaman, M. Cathelineau, Régine Mosser-Ruck, C. Peiffert, et al.. Experimental synthesis of chlorite from smectite at 300ºC in the presence of metallic Fe. Clay Minerals, Mineralogical Society, 2003, 38 (3), pp.281-302. 10.1180/0009855033830096. hal-01876608

HAL Id: hal-01876608 https://hal.univ-lorraine.fr/hal-01876608 Submitted on 18 Sep 2018

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. ClayMinerals (2003)38, 281–302

Experimentalsynthesisofchloritefrom smectiteat300ºCinthepresenceof metallic Fe

D.GUILLAUME 1,* , { , A . N E A MA N 2,M.CATHELINEAU 1, R.MOSSER-RUCK 1,C.PEIFFERT 1,M.ABDELMOULA 3, J. D U BE SSY 1, 2 4 5 F. V I L LI E´ RA S ,A.BARONNET AND N . M IC H A U

1 Geólogieet Gestion des Ressources Mine ´raleset Energe ´tiques(G2R ),UMR 7566CNRS-CREGU-INPL-UHP, Universite´ HenriPoincare ´,BP239,54506 Vandœ uvre-le `s-Nancy, 2 LaboratoireEnvironnement et Mine ´ralurgie (LEM),UMR 7569CNRS-INPL, EcoleNationale Supe ´rieurede Ge ólogie,BP 40,54501 Vandœ uvre-le `s-Nancy, 3 Laboratoirede Chimie Physique et Microbiologie pour l’ Environnement (LCPME), UMR 7564CNRS-UHP, Universite´HenriPoincare ´,405 ruede Vandœ uvre, 54600 Villers-le`s-Nancy, 4 Centrede Recherchesur les Me ćanismes deCroissance Cristalline CRMC2-CNRS, Campus Luminy,case 913, 13288 Marseille,and 5 Agencenationale pour la gestiondes de ćhetsradioactifs (ANDRA), DirectionScientifique/ ServiceMate ´riaux,Parc de la Croix Blanche, 1/ 7rue JeanMonnet, 92298 Cha ˆtenay-Malabry,France

(Received 28 January 2002; revised 7November 2002 )

ABSTRACT:The alteration and transformation behaviour of montmorillonite (bentonite from Wyoming, MX-80) in low-salinity solutions (NaCl, CaCl 2)in the presence of metallic Fe (powder and 86461mmplate) and magnetite powder was studied in batch experiments at 300ºC to simulate the mineralogical and chemical reactions of clays in contact with steel in anuclear waste repository. The evolutions of pH and solution concentrations were measured over aperiod of 9months. The mineralogical and chemical evolution of the clays was studied by XRD,SEM, Transmission Mo¨ssbauer Spectroscopy and TEM(EDS,HR imaging and EELS).Dissolution of the di-octahedral smectite of the starting bentonite was observed, in favour of newly formed clays (chlorite and saponite), quartz, feldspars and zeolite. The formation of Fe-chlorite was triggered by contact with the metallic Feplate and Fe-Mg-chlorite at distance from the Fe plate (>2 mm).

KEYWORDS:bentonite,smectite, Fe, magnetite,chlorite, saponite, experimental synthesis, EELS-TEM, EDS-TEM, HRTEM imaging,Mo ¨ssbauerspectroscopy.

Bentonitesare frequently recommended as backfill compactedto high density and their high sorption materialsfor the construction of high-level nuclear capacity.If the natural occurrence of old bentonites wastereposit ories.Differe ntconcept sforesee isan indication of their long-term stability, the very emplacingsteel containers in tunnels backfilled specificconditions of high-level nuclear wastes withcompactedbentonite(Madsen,1998). repositorysites has to be taken into account, Bentonitesare choosen because of their swelling particularlythe tempera tureincreas eandthe capacity,their low hydraulic conductivity when eventualpresence of available Fe. To this end, interactionsbetween the bentonite and the container corrosionproducts are of particular interest. The *E-mail: [email protected] keyproblem is the permanence or not of swelling { present address: Westfa¨lische Wilhelms-Universita ¨t Mu¨nster, Institut fu¨rMineralogie, Corrensstr. 24, 48149 minerals.Previous studies of the evolution of clay Mu¨nster, Germany mineralsin different natural contexts may provide DOI: 10.1180/0009855033830096 partof the answer.

# 2003The Mineralogical Society 282 D.Guillaume et al.

Duringdiagenesis and low-temperature meta- numberof studies, the formation of minor amounts morphismof intermediate to mafic volcanic rocks offramework silicates after clay minerals was andvolcanogenic sediments, chlorite may crystal- reported,but not studied in detail. The stability of lizefrom mineral precursors through complex smectitein the presence of Fe underlow f is still O2 processesincluding the crystallization of mixed- poorlystudied. Mu ¨ller-Vonmoos et al. (1991) and layerminerals (Bettison & Schiffman,1988). Madsen(1998) studied the hydrothermal stability of Previousworks suggest that precursors may be a bentonitein the presence of Fe andmagnetite smectiteor other minerals such as berthierine powdersat 80º C overa periodof 27 to 29 weeks, (Aagaard et al.,2000)which transform into chlorite butinasystemfreeof li quidwa ter.No inresponset oanincreaseo ftemperature mineralogicalchange was detected at this tempera- (Hoffmann&Hower,19 79;Horton,1 985; ture.However,thermodynamicmodelling Bettison&Schiffman,1 988;Schiffman& (Cathelineau et al.,2001)predic tsa strong Fridleifsson,19 91;Hu mpreys et al., 1994; mineralogicalevolution of the bentonite, especially Robinson& Santanade Zamora, 1999). The theformation of a chlorite-saponite-zeoliteassem- transitionmay occur in a steppedprogression of blage,when the smectite-Fe system interacts with a smectiteto chlorite via corrensite or as a gradual solutionat a temperature>150º C. changeinvolving randomly or regularly interlayered Inconclusion, the formation of chlorite is a structures(chlorite-smectit e:C-S), as dictated by possibility,establishe dbothfrom natural case theenvironment of formation (Bettison-Varga & studiesand thermodynamic simulations. However, Mackinnon,1997). The role of fluid composition, theavailability of native Fe incontact with clay porosityand water/ rockratio (W/ R)indetermining mineralsin a high-levelnuclear waste repository thestability of those minerals has been recognized. andits consequences for the evolution of clay Therole of low W/ Rratioin the stabilization of mineralscannot be establis hedfrom previous C-Sisemphasized (Alt et al.,1986;Schiffman & studies.Therefore,experimentshavebeen Staudigel,1995; Bettison-Var ga& Mackinnon, performedto investigate the hydrothermal stability 1997).Murakami et al.(1999)provided HRTEM ofa bentoniteat 300º C inthe presence of metallic evidencefor the mechanism of the saponite-to- Fe andmagnetite, and particularly the transforma- chloriteconversion in thermally metamorphosed tionof montmorillonite into Fe-rich clay phases, volcanoclasticrocks and Robinson et al. (2002) suchas sapo niteand /orchlo rite.Analy tical gavea reviewof reactio npathwaysfor the investigationsconcerned both solid and liquid smectite-to-chloritetransformation in geothermal phases.Clay minerals were characterized by their systems.However the complex structura land structural(X-ray Diffraction – XRD, Scanning compositionalchanges that accompany the trans- ElectronMicroscopy – SEM andHigh-Resolution formationare not entirely understood (Meunier et TransmissionElectron Microsco py– HRTEM al.,1991;Inoue & Utada,1991), and the conditions imaging)andchemicalproperties(Cation favourablefor the formation of chlorite have not ExchangeCapacity–CEC,Transmission beenfully determin edin most natural cases. Mo¨ssbauerSpectroscopy – TMS, ElectronEnergy- Elementarythermodynamic considerations indicate LossSpectroscopy – EELS andEnergy Dispersive thatFe-chlorite can only form when the Fe 2+/H+ Spectroscopy(EDS)-TEM). Theexperimental solu- ratiois high enough (Bowers et al., 1984). tionscoexistin gwiththe solid products were Thesmectite-to-chlo ritetransformation has not characterizedby inductively coupled-atomic emis- beenreproduced experimentally. Small et al. (1992) sionspectro scopy(ICP-AES) andinducti vely indicatedthat no chlorit ecouldbe obtaine d coupledplasma-mass spectroscopy (ICP-MS). experimentallyunder laboratory conditions below 400ºC, confirmingprevious statements from Gillery MATERIALSANDMETHODS (1959)and Velde (1973) .Thehyd rothermal stabilityof smectite has been studied extensively, Starting material especiallythe collapse of expandable smectite to non-expandablelayer silicates, in particular illite, TheMX80bentonite(Na/Ca-bentonite, generallyin mineral-solution systems devoid of Wyoming),referred to as ‘ startingbentonite’ , was addedFe (e.g.Eberl et al.,1993;Cuadros & usedfor the experiments. Its CEC is79 mEq/ 100g Linares,1996; Mosser-Ruck et al.,2001).In a (Table1) and its co mpositionis as fo llows Experimental synthesis of chlorite 283

(Guillaume et al.,2001a):montmorillonite (79%), preparationwas equilibrated at room temperature quartz(3%), K-feldspar (2%), plagioclase (9%), for6 days.The solution pH was then measured carbonates(2%), mica (3%) and other minerals usinga Mettler 1 combinationpH electrode at (mostlypyrite, phosphates and hematite, 2%). The 25ºC, andthe Na +/Ca2+ ratioof the reacting structuralformula of the montmorillonite (<2 mm solutionwas ~80 (± 2) as a resultof exchange fractionof the bentonite) determined from micro- processes.Two typesof assemblages, referred to as probeanalyses, consistent with EDS-TEM andICP- ‘startingsamples’ , wereused: (1) bentonite with 2+ MSanalyses,and taking into account the Fe /Fetot solutionand addition of metallic Fe powder+ ratiod eterminedbyTMSandEELS-TEM magnetitepowder; (2) the same with addition of an (Guillaume et al.,2001b),is: Fe platewith dimensions of 8 6461 mm. III II (Si3.98Al0.02)(Al1.55Mg0.28Fe0.09Fe0.08)O10 (OH)2Na0.18Ca0.10 (1) Experimental procedure Theexperimental conditions are listed in Table1. Thestarting samples were mounted under argon Preparation of the starting solid-solution atmospherein gold-lined warm-sealed great capa- system cityautoclaves (~18 ml). The autoclaves were then Thesolution was prepared from analytical-grad e heatedto 300º C ina furnace.The internal pressure, chemicals.A Milli-QReagent Water System from 86bars, is the liquid-vapour equilibrium pressure at MilliporeCorp. provided deionized water with a 300ºC. Thedurations of the experiments were 1 resistivity>18 M O cm – 1. Na+ and Ca2+ chlorides week,1 month,3 monthsand 9 months.The wereadded to deionized water to simulate the Na +/ temperaturecontrol was stable to ~2º C. Thevessels Ca2+ ratioof naturalwater in sediments which is, in wereremoved at specific time intervals, quenched general,between 10 and 100. The starting solution at25º C andopened under argon atmosphere. Run contained0.0207 mol/ kgNaCl and 0.0038 mol/ kg sampleswere collected under argon atmosphere and

CaCl2 andthe total chloride concentration was centrifugedto extract the solution. Solution aliquots 0.0282mol/ kg. weretaken, measured for pH and Eh and filtered Bentonite(1.5 g) was first added to the solution through0.45 then 0.02 mmfiltersinto cleaned witha liquid/solidmass ratio of 10, then metallic polypropylenebottles and analysed. When an Fe Fe andmagnetite powders with an Fe/ bentonite platewas added, the first 1 mm atthe contact with massratio of 0.1 (metallic Fe/ magnetite:1/ 1). theFe platewas collected before centrifugation and Argonwas bubbled through the preparation for analysedseparately. 45min to obtain an oxygen-free system and the Thesolid was dried under argon flux at room temperatureand gently ground in a mortar.Run sampleswere then stored in hermetically closed

TABLE 1. Experimental conditions of the experi- boxesunder argon atmosphere until they were ments and CEC(mEq/100 g) of the starting analysed.Solids were suspended again in pure bentonite, starting sample, WIPand IPrun samples. waterfor the preparation of oriented samples for XRD, orin methanol for the preparation of TEM grids. Sample CEC

Starting bentonite 79 R E S U L T S WIP Threecases will be distinguished for the run 25ºC Starting sample 73 300ºC 3 months 75 samplesformed:(1 )atth enearestc ontact 9 months 66 (<1mm) withthe Fe plate,denoted ‘ CIP run samples’;(2)at >1 mm fromthe Fe plate, IP 300ºC 1 week n.a. denoted‘ IPrunsamples’ ; and(3) during experi- 1 month n.a. mentsperformed without Fe plate,denoted ‘ WIP 3 months 73 runsamples’ . Inthe CIP, IPandWIP cases,the 9 months 30 sameamounts of Fe andmagnetite powders are presentat the beginning of the experiment. 284 D.Guillaume et al.

X-ray diffraction solidphases. Mo ¨ssbauerworks on ‘ environmental’ materialshave been mainly concerned with Fe in TheXRD datawere collected with a D8Bruker twomineral groups: clay-sized phyllosilicates (often diffractometerwithCo - Ka1 radiation (l = lookedupon as clay minerals sensustricto ) and 1.7902 AÊ ).TheXRD patternsof unorie nted oxides.Ideally, a distinctionof Fe inthese two powderswere taken in duplicate to identify non- groupsby Mo ¨ssbauerspectroscopy is relatively clayminerals. X-ray diffraction patterns were also straightforward:the phyllosilicates are paramagnetic takenfrom air-dried oriented and ethylene glycol atroom temperature and may contain both divalent (EG)-saturatedspecimens of the <2 mmfractionsof andtrivalent Fe, whereas the most common Fe thestarting- and run samples. oxidesshould be magnetically ordered and contain TheGreene-Kelly (Hoffman-Klemen) test was onlytrivalent Fe, except for magnetite which is the performed(Gree ne-Kelly,195 3;Ho ffman& onlypure oxide of mixed valence (Murad, 1998). Klemen,1950). The <2 mmfractionof each run TransmissionMo ¨ssbauerspectra were collected samplewas Li-saturated, heated overnight at 400º C usinga constant-accelerationspectrometer with a thenglycerol saturated. Pure silica slides were used 50mCi sourceof 57Coin Rh.The spectrometer was fororiented specimens of Li-saturated samples to calibratedwith a 25 mm foil of a-Fe at room avoidNa migration from the glass (Bystro ¨m- temperature.Spectra were obtained at 12 K or Brusewitz,1975). 150K. Thecryostat consisted of a closedcycle Additionaltreatments included (1) K-saturation, heliumMo ¨ssbauercryogen icworkstati onwith heatingovernigh tat110º C, heatingthen EG vibrationsisolation stand manufactured by Cryo saturation,(2) Mg-saturation, then glycerol satura- Industriesof America. Helium exchange gas was tion,and (3) Li-saturation, then EG saturation. usedto thermal lycouple the sample to the refrigerator,allowing variable temperature opera- tionfrom 12 to 300 K. Thesamples were set in the Cation exchange capacity sampleholder in a glovebox filled with an Argon TheCEC valuesof the starting and run samples atmosphereand quickly transferred in the cryostat weredetermined as follows. Samples were Ba 2+ forMo ¨ssbauermeasurements. Computer fittings saturatedby treatingthem three times each with 1 N wereperformed by using Lorentzian-shape lines.

BaCl2 solution,washed with distilled water and Theparameters that result from any computer centrifugeduntil the electrical conductivity of the fittingmust be both mathematically ( w2 minimiza- equilibriumsolution was <10 mohm/cm –1. The tion)and physically significant; in particular the Ba2+ wasthen exchanged by treating the samples widthof all lines must be small enough (Benali et 2+ threetimes with 1 N NaNO3 solution.The Ba al., 2001). concentrationsin the solutions were determined by atomicabsorption spectrometry (AAS). Electron energy-loss spectroscopy (EELS)- transmission electron microscopy

Scanning electronmicroscopy (SEM) 3+ The Fe /Fetot ratioin isolatedclay particles from TheSEM imageswere obtained using an Hitachi the <2 mmfractionof the samples was determined S-2500Fevex scanning electron microscope. The usinga Gatan 1 666parallel electron energy loss denserfraction of each sample was separated by spectrometerattached to a CM20-Philips 1 instru- successiveultrasonica tionand sedimentati onin mentoperating at 200 kV withan unsaturatedLaB6 alcohol,then disposed on carbon adhesives sticks. cathode.The following measuring conditions were Semi-quantitativechemical analyses were also chosen:a spectrometerentrance aperture of 2 mm, performed.The separated <2 mmfractionof each acollectionhalf angle of 10 mrad, an energy run-samplewas also checked for newly formed non- dispersionof 0.2 eV per channel. The energy clayminerals intimately associated with the clay. resolution,measured at the full width at half maximum(FWHM) ofthe zero-lo sspeak is 1.2eV. Spectra were acquired at the thinnest part Transmission Mo¨ssbauer spectroscopy (TMS) ofthe particles (<100 nm) and analysed with a 57Fe Mo¨ssbauerspectroscopyis the most Gatan1 EL/P2.1PEELS software.The low proficientmethod to characterize the Fe speciesin energy-lossspectrum was used to remove the Experimental synthesis of chlorite 285 multiple-ineslastic-scatteringeffect in the core-loss High-resolution transmission electron regionusing a Fourier-ratiodeconvolution method microscopy (HRTEM)

(Egerton,1986). Fitting of the Fe- L3 edge was performedin the range 705 to 720 eV using Opus Selectedsamples were prepared for HRTEM 3+ software1 fromBruker and Fe /Fetot ratio is usinga JEOL JEM2000FX instrumentoperating at determinedusingt hemethodd escribedby 200kV. Preparation of the <2 mmfractionof Guillaume et al. (2001b). samplesfor HR observationsconsisted of an

FIG.1. XRDpowder patterns of the starting sample, 3month and 9month IPrun samples. Mt: magnetite; Sm: smectite; Qz: quartz; Fs: feldspar; Chl: chlorite. Reflection values in A Ê . 286 D.Guillaume et al. exchangewith heptylamonium chloride (Lagaly et poly-atomicspecies, nitric acid was chosen as the al.,1976),followed by an embedding in an Epon acidifyingsolution for both ICP-AES andICP-MS resin(Amouric & Olives,1991). Ultrathin sections analysesat 1.5%w/ vand3% w/ v,respectively.The ofthe embedding blocks (100 nm) were prepared uncertaintyin measured solution concentrations was withan ultra-microtome equipped with a diamond ±10%. knife,and transferred onto 300-mesh carbon-coated Ni grids. Mineralogyof the Fephases XRDresults. TheFe powder,which was added to Energy dispersivespectroscopy (EDS) thestarting bentonite, was not detectable by XRD Microchemicalanalyses of isolated clay particles after1 monthof treatment and is not clearly of the <2 mmfractionwere obtained with an EDAX observedwith the SEM. Incontrast to the Fe energydispersive X-ray analyser attached to a powder,no significant difference in the relative CM20-Philipsinstrume ntoperati ngat 200 kV intensityof magnetite reflection lines was detected equippedwith Si-Li detector and Li superultrathin byXRD betweenthe starting and run samples windowsSUTW. Spectrawere collected under (Fig.1), indicating that the amount of magnetite in nanoprobemode, over 40 s froman area ~10 nm thesample did not change significantly. However, indiameter. Elemental composition was calculated thesize of magnetite-like crystals observed with assumingthe thin film criteria (SMTF program: SEM isstrongly modified. The starting magnetite semi-quantitativemetallurgicalthin film program) powderwas made of <5 mmgrains(Fig. 2h). After andusing k-factors calibrated with independently 9months,40 to 200 mmeuhedralcrystals, of analysedmacroscopic micas, with a maximumerror magnetitecomposition, are observed in the run of5% for each element. 30 to 50 analyses were samples(Fig. 2i). performedon isolated particles for each sample. TMSresults. TheTMS study(Fig. 3) also Thestructural formulae were calculated for each demonstratesthe consumption of the metallic Fe 3+ analysistaking into account the Fe /Fetot ratio powder.On the sides of the spectrum, one notices a determinedusing EELS andTMS. magneticcomponent S (~7%of the total surface of thespectrum)as cribabletothemaghemite ( -FeIII O ,theoxidized form of magnetite). This Analyses of run solutions g 2 3 indicatesthat magnetite is transformed to maghe- Chemicalanalyses of the experimental solutions miteduring the experiments. wereobtained by ICP-AES (Na +, K+, Ca2+, Mg2+, Fe2+ and Si4+ concentrations)and ICP-MS (Al 3+ Mineralogyof the non-clay phases concentration).Solutions were acidified to avoid precipitationof Al orFe hydroxidesand diluted to Observationsby SEM aresu mmarizedin bringconcentr atedanalytes into quantifi cation Table2.Newlyformedgrainsofquartz, rangesof the ICP-AES andICP-MS techniques. K-feldspar,plagioclase and zeolites were observed Becausehydrochloric acid and sulphuric acid form inall run samples and overgrowths are observed on

TABLE 2. Assemblage of non-clay minerals in the starting bentonite (Guillaume et al., 2001a) and as observed by SEMin the starting sample, CIPand IPrun samples.

Sample Qz Kfs Plagio Carb Phos Mica Pyr Sulph Iron Mt

Starting bentonite 3% 2% 9% 2% +3% ++ + 25ºC Starting + + + + + + + + + + 300ºC 1 month + + + + + ? + 3months +Gr +Gr +Gr + + 9months +Gr +Gr +Gr + +

+: presence of the mineral; Gr: evidence for growing and neoformation; space: absence of the mineral. Kfs: K-feldspars, Plagio: plagioclase, Carb: carbonate, Phos: phosphate, Pyr: pyrite, Sulph: sulphate, Mt: magnetite Experimental synthesis of chlorite 287

FIG.2. Evolution of the morphology of primary minerals and newly formed minerals as seen by SEM:(a) growth of quartz on aquartz particle, (b) growth of K-feldspar on aplagioclase particle, (c) zeolite crystal associated with the clay, (d) newly formed plagioclase, (e) newly formed K-feldspar, (f) clay in the starting bentonite, (g) clay in the 3month run sample, (h) magnetite powder added to the starting bentonite, (i) large euhedral crystal with magnetite composition. Micrographs a, g: 3month WIPrun sample; b, c, d, e, i: 9month IPrun sample. 288 D.Guillaume et al.

FIG.3. TMSspectra of the 9month IPrun sample, measured at 11 K(see text for explanation of peak labels). theanhedral detrital grains. Figures 2a and 2b show (characteristicof the presence of a monovalent overgrowthsof quartz grains and K-felds par cation)and the increase of the 14.9 A Ê peak (microcline),respectively, on a plagioclase(albite) (characteristicofthe presen ceof a divalent particle.Newly formed automorph particles of cation).This indicates that the sodi-calcic clay quartz,plagioclase (albite, Fig. 2d), and K-feldspar phaseof the starting bentonite became progressively (microcline,Fig. 2e) were also observed. The XRD calcic,and is mostly calcic after 9 months.The patternsof powder specimens of the 3 monthand 3monthrun sample presents an intermediate state. 9monthrun samples (Fig. 1) show a strong Thestarting sample and the 3 monthrun sample increasein the 1.54 A Ê and 3.34 AÊ reflection bothe xpandto ~1 7A Ê afterEGsaturation, (characteristicof quartz). This confirms that the indicatingthat the clay phase is of smectite type. amountof quartz is increasing. The9 monthsample expands to 16.8 A Ê ,indicating Nosignif icantcha ngewas notice dinthe thatthe clay phase is mostly of smectite type, and a morphologyof the clay mass, as shown by Fig. 2f smallamount of mixed-layer clay might be present. (clayfraction in the starting bentonite) and Fig. 2g TheGreene-Kelly test was performed on the (clayfracti oninthe 3 monthrun sample ). samethree samples. The XRD patternsobtained are Meanwhile,SEMobservationsshowed fibrous presentedin Fig. 5. The reflection around 9.5 A Ê for particlesof zeolites intimately associated with thestarting sample indicates the presence of a clayparticles of the <2 mmfractionsof 9 month montmorillonite.The strong reflection at 16.8 A Ê in IPandWI Prunsa mples(Fig. 2c) .Precise the9 monthrun sample indicates the presence of a determinationof thenature of these zeolite particles tetrahedralcharge deficiency. The 3 monthrun wasnot possible from semi-quantitativ echemical samplepresents an intermediate state. This means analyseson SEM (mostlysodic with Si/ Al &3.7, thatthe montmorillonite progressively transforms to closeto erionite composition). Based on XRD and anothermineral of the smectite group. SEM observations,zeolite was not present in the Astarting-samplewhich is K-saturated, heated startingbentonite. overnightat 110º C, thenEG saturated,expands to 16.9 AÊ (Fig.6b) because of its low charge (0.38). Afterthe same treatment, the 9 monthrun sample Mineralogy of the clay phase expandsslightly to ~14 A Ê ,indicatinga high-charge XRDresults- transformationof the mont- smectite(Fig. 6c). morillonitein the absence of Fe plate (WIP run A9monthrun sample which is Mg saturated samples). Orientedslides were prepared for the thenglycerol saturated expands to 17 A Ê , however startingsample and the 3 monthand 9 monthWIP (Fig.6d). This indicates that the clay phase of this runsamples. The XRD patternsobtained before and sampleis not of vermiculite type. afterEG saturationare presented in Fig. 4. In air- XRDresults- evidencefor a newlyformed clay driedspecimens, with the duration of experiment inexperime ntswith added Fe plate (IP run weobserve a strongdecrease of the 12 A Ê peak samples). Additionalpeaks at 13.9 A Ê (weak), Experimental synthesis of chlorite 289

FIG.4. XRDpatterns of the starting sample, 3month and 9month WIPrun samples. a, c, e: oriented, air-dried; b, d, f: EG-saturated. Reflection values in A Ê .

7.03 AÊ (strong)and 3.52 A Ê (medium)are present in followingthe decomposition of chlorite at 550º C theXRD patternof the oriented Li-satura ted canbe observed in the range 30 –33º2y. 9monthIP runsample (Fig. 7a). The 13.9 A Ê Chloritedoes not expand by EG orglycerol reflectionlineis more clearly observed after saturation.On both EG- andglycerol-satu rated heatingthe specimen at 330º C, sinceLi-smectite XRD patternsof the9 monthrun sample exchanged formeda ‘collapsed’structure (Fig. 7b). The above- withLi andMg, respectively, (Fig. 6e,f), reflections mentionedadditional peaks clearly indicate the around 17 AÊ indicatethe presence of smectite and presenceof a mineralof the chlorite group. shouldersaround 14.4 A Ê indicatethe presence of Heatingthe preparation at 550º C modifiedthe chloritelayers. Well-developed shoulders above intensitiesof chlorite lines, diminishing the reflec- 17 AÊ thatare not present in the diffractogram of tionsof the second, third and fourth orders thestarting sample are also observed (Fig. 6a). (Fig.7c). The formation of an amorphous phase Theseshoulders indicate that a randomlyinter- 290 D.Guillaume et al.

FIG.5. XRDpatterns of the starting sample, 3month and 9month WIPrun samples, after the Greene-Kelly test. Reflection values in A Ê . stratifiedchlorite-sme ctitestructure is probably presentin the 9 monthrun sample. TheXRD patternsof non-orientedpowders of the 3monthand 9 monthrun samples in the range 70 –75º2y (Fig.1), present two additional peaks at FIG.6. XRDpatterns of Li- K-and Mg-saturated 1.515 AÊ and 1.552 AÊ whichcan be attributed to samples. (a, b) starting sample, (a) Li- then EG-satu- trioctahedralchlorite. The relative intensity of the rated, (b) K-saturated, 110ºC heated then EG-saturated; (c, d) 9months WIPrun sample, (c) K-saturated, 060peak of dioctahedral smectite ( d = 1.496 AÊ ) 110ºC heated then EG-saturated, (d) Mg- then decreasedsignificantly and that of the 211 peak of glycerol-saturated; (e, f) 9month IPrun sample, quartzand/ orthe 060 peak of trioctahedral smectite (e) Li- then EG-saturated, (f) Mg- then glycerol- Ê (d = 1.541 A)increasedsignificantly. saturated. Reflection values in A Ê . TheXRD datademonstrated that in the absence ofadded Fe plate(WIP runsamples), mont- morilloniteprogressivelytransforms to another attributedto strong mineralogical changes, such as clayof the beidellite– saponite– nontronite series. theformatio nofnon-exp andableclay phases Whenan Fe plateis added to the experimental suggestedby the XRD study. system(IP runsamples), a chlorite-likemineral is HRTEM. High-resolutionTEM observationsof alsoformed and the presence of a randomly heptylammonium-saturatedclay fractions (<2 mm) interstratifiedclay is suspected. ofthe starting and 3 and9 monthIP runsamples CEC. Theresults of the CEC measurementsfor wereperformed to distinguish between different starting,IP andWIP runsamples are summarized in clayphases that might be present (low-charge Table1. The CEC decreaseswith the duration of smectite,saponite, vermiculite, chlorite, berthierine, thetreatment, and the decrease is very important etc.).According to Lagaly et al.(1976)and Vali & whenan Fe plateis added. This evolution can be Hesse(1990), the difference between low-charge Experimental synthesis of chlorite 291

FIG.7. XRDpatterns of 9month IPrun sample. (a) Li-saturated, un-heated, (b) 330ºC, (c) 550ºC heated. Sm: smectite, Chl: chlorite, Qz: quartz, Fs: feldspars. smectite(13 A Ê layerspacing), vermiculite and Crystal-chemistryof the clay phase saponite(17 –18 AÊ ),chlorite(14A Ê ), and berthierine(7 A Ê )shouldbe clearly observed . Thestructural formulae of clay minerals are Figure8a sh owst helay erstru ctureof th e calculatedfrom EDS analyses,taking into account montmorillonitein the starting -bentoniteafter theresults obtained by EELS andTMS forthe 3+ treatmentwith heptylammonium ions, with layer Fe /Fetot ratio(see below). The clay particles were spacingof ~13 A Ê .Inthe 3 monthIP runsample, selectedat random for EDS analysis.Consequently, ~13 AÊ layersare still dominant, and some particles theanalytical points presented on the graphs do not appearto be pa rtlytra nsformedtochlori te. necessarilyrepres entth evariouspopulations Figure8b shows the case of a particlepresenting (unreactedsmectite, interstratified or newly formed typicalsmectitelayers (13 A Ê layerspa cing) particles).All structural formulae have been calcu- surroundedby three well identified chlorite layers latedon the basis of 11 and 14 oxygens as well as (14 AÊ layerspacing). In the 9 monthIP runsample, referenceminerals because run samples are mixtures 14 AÊ layerspacing is dominant (Fig. 8c), but some ofdi-octahedral smectite and tri-octahedral chlorite particlesare still not totally transformed and present (mixed-layeredminerals, or mechanical association someremaining layers with 13 A Ê spacing(Fig. 8d). ofsmall-size particles). Following procedures already Neither 17 –18 AÊ nor 7 AÊ layerspacing was discussedin Schiffman & Fridleifsson(1991) for observed,indicating that no saponite, vermiculite or naturalsmectite– chlorite series, all figures show the berthierinewere formed during the experiments. No totalityof analytical data in diagrams calculated regularinterstra tifiedsmectite- chloriteparticles arbitrarilyon a singlebasis (14 oxygens), since no wereobserved, indicating that the smectite-to- quantitativeparameters provide the relative amount chloritetransition occurs via randomly interstrati- ofdi-/ trioctahedralphases in the studied run samples. fiedparticles. Alldiagrams using structural formulae based on 11 292 D.Guillaume et al.

FIG.8. HRTEMimages of clay fractions from the starting sample (a), 6month (b) and 9month (c, d) IPrun samples. Heptylamonium chloride-saturated specimens. oxygensshow, as expected, a similardistribution and (inwt.% of elements) of the 1 week,1 month, relationshipsamong analytical points and theoretical 3monthand 9 monthCIP, IPandWIP runsamples fieldsof reference minerals. Selected compositions arereported in Table 3. Experimental synthesis of chlorite 293

3+ Determinationof theFe /Fetot ratioby TMS and EELS -TMS results. Thespectrum obtained for the 9monthIP runsample (Fig. 3) shows a verystrong paramagneticco mponent(doubletD1) correspondingto Fe II characteristicof chlorite; ~90% ofthe total surface of the spectrum, indicating that FeII isfully incorporated into the clay. Another paramagneticdoublet (D2) in low abundan ce correspondingto Fe III isalso observed, indicating that FeIII remainsin the clay. 3+ The Fe /Fetot ratioin the clay fraction was calculatedfrom Mo ¨ssbauerspectra for the starting bentoniteand the 3 and9 monthIP runsamples. Resultsare reported as afunctionof time in Fig. 9a. Inthe montmorillonite of the starting-bentoni te,the 3+ Fe /Fetot ratiowas of 0.55. We observe a strong 3+ evolutionto a Fe /Fetot ratioof 0.05 after 9 months,the 3 monthrun sample presenting an intermediatesituation. 3+ Determinationof theFe /Fetot ratioby TMS and 3+ EELS -EELS results. The Fe /Fetot ratiowas also measuredwithEELS followingtheme thod publishedelsewhere (Guillaume et al., 2001b) in isolatedclay particles of the starting-be ntonite (Guillaume et al.,2001a)and clay fracti on (<2 mm)of the 1 month,3 monthand 9 month CIP andIP runsamples. Results are reported as a functionof time in Fig. 9b. The overall evolution is consistentwith TMS data,although EELS data showa largerdispersion which could be explained bythe heterogeneity of the samples (see below). 3+ Deviationto higher Fe /Fetot ratiocould also be dueto technical reasons such as sample preparation (thatcould not exclude selective sampling of particles)or sample oxidation under the beam. Forthe 1 month,3 and9 monthCIP andIP run samples,EELS andEDS analyseshave been

3+ FIG.9. Evolution of the Fe /Fetot ratio, (a) in the clay phase of the IPrun samples measured by TMS vs. experimental duration; (b) in isolated clay particles from CIPand IPrun samples measured by EELS vs. experimental duration; (c and d) in isolated clay particles measured by EELS vs.iron content measured by EDSon the same particle (from structural formula calculated on the basis of 14 oxygens), (c) IPrun samples, (d) CIPrun samples. Filled triangles: 1month; open diamonds: 3months; filled squares: 3+ 9month run samples. Comparison with the Fe /Fetot ratio in the starting montmorillonite (open squares, Guillaume et al., 2001b). 294 D.Guillaume et al.

TABLE 3. Selected compositions (EDSanalyses, in wt.% of elements) of CIP,IP and WIPrun samples.

1 week CIP 1 week IP SiAl Fe Mg KNa Ca SiAl Fe Mg KNa Ca

60.623.1 8.0 3.9 0.4 2.7 1.1 62.8 25.4 2.9 5.1 0.4 2.4 1.0 59.722.7 8.4 4.2 0.4 3.2 1.1 61.0 24.2 4.3 4.9 0.4 4.0 1.3 57.320.4 13.6 4.0 0.5 2.0 2.0 60.7 23.9 5.2 4.8 0.4 3.5 1.4 56.920.7 13.3 4.3 0.5 2.8 1.3 60.2 23.7 6.3 5.1 0.5 3.5 0.8 56.621.5 13.6 4.0 0.4 2.3 1.4 59.6 24.4 6.2 5.3 0.4 3.0 1.1 54.619.6 17.0 4.0 0.4 2.5 1.6 58.0 22.9 8.5 4.9 0.6 4.0 1.2 53.720.6 16.0 4.3 0.4 3.1 1.5 57.5 22.2 10.9 5.0 0.7 2.7 1.0 48.817.9 25.3 3.3 0.7 2.5 1.2 55.9 22.3 11.7 4.7 0.6 3.5 1.3 46.017.4 26.8 3.3 0.4 2.4 3.4 54.3 21.9 11.8 5.7 0.8 4.6 1.0 40.714.8 35.6 3.5 0.7 3.0 1.5 52.8 22.5 13.4 5.3 0.5 4.3 1.2

1 month CIP 1 month IP SiAl Fe Mg KNa Ca SiAl Fe Mg KNa Ca

55.218.4 19.5 3.9 0.4 1.3 0.9 63.3 24.1 4.2 4.8 0.2 2.7 0.8 43.217.5 31.6 3.7 0.2 1.4 2.1 63.1 23.5 6.4 4.4 0.4 1.1 1.1 34.37.7 51.4 2.7 0.4 1.7 1.5 61.8 23.3 6.1 4.8 0.6 2.5 0.8 35.114.9 43.2 3.5 0.7 1.8 0.6 61.1 23.4 6.2 5.4 0.4 2.4 1.1 31.911.7 49.8 3.0 0.6 1.9 0.9 60.7 24.6 6.7 4.8 0.5 1.8 0.9 31.311.3 50.9 2.9 0.6 1.8 0.9 57.1 21.0 13.0 5.2 0.7 1.9 1.1 29.79.6 55.5 2.4 0.4 1.3 0.9 58.1 24.0 11.4 4.0 0.7 0.6 1.3 30.012.2 53.4 2.5 0.2 0.8 0.7 56.8 24.4 11.4 4.2 0.5 1.5 1.3 28.910.7 54.7 2.3 0.9 1.2 1.0 52.5 20.7 16.6 5.4 0.7 2.9 1.2 24.96.4 65.4 1.4 0.3 0.5 0.9 52.0 23.1 18.2 3.8 0.5 0.7 1.6

3 months CIP 3 months IP 3 months WIP SiAl Fe Mg KNa Ca SiAl Fe Mg KNa Ca SiAl Fe Mg KNa Ca

45.417.6 26.2 3.5 0.3 5.6 1.3 62.9 24.5 4.5 3.2 1.7 1.7 1.5 62 23 6.2 3.9 0.6 3.2 1.2 45.417.5 29.9 3.5 0.4 2.3 1.1 58.2 23.5 10.4 3.4 0.9 1.7 1.8 61 24 7.1 3.2 0.9 2.0 1.2 44.919.1 28.4 3.2 0.5 2.6 1.3 56.7 22.3 12.4 4.2 0.6 2.4 1.4 58 23 10 3.9 0.7 3.4 1.0 44.319.2 28.5 3.7 1.4 1.5 1.4 55.2 21.7 12.9 3.3 4.1 0.2 2.5 58 22 12 3.5 0.5 3.1 1.3 44.118.9 28.7 4.0 0.4 2.5 1.3 51.7 20.2 19.8 4.1 0.6 1.7 2.0 57 22 12 3.2 0.5 2.4 1.8 43.219.9 28.8 4.3 0.4 2.3 1.2 52.0 21.5 16.4 4.8 0.8 2.6 1.8 55 22 15 3.3 0.5 2.3 1.3 42.019.2 30.7 4.1 0.3 2.7 1.0 51.7 22.6 17.2 3.5 0.8 1.9 2.3 51 18 24 2.7 0.6 1.9 1.5 41.921.6 28.2 4.4 0.5 2.1 1.3 50.5 21.6 19.1 4.2 0.6 2.6 1.3 51 20 21 3.3 0.5 2.1 1.3 38.320.3 35.9 2.8 0.5 1.3 1.0 47.6 19.1 23.9 4.8 0.8 2.1 1.7 50 19 21 3.8 0.7 3.3 2.1 34.012.8 48.7 3.0 0.3 0.8 0.4 45.6 18.4 26.3 4.8 1.0 2.3 1.6 49 19 22 4.3 0.3 3.4 1.6

9 months CIP 9 months IP 9 months WIP SiAl Fe Mg KNa Ca SiAl Fe Mg KNa Ca SiAl Fe Mg KNa Ca

29.813.7 53.3 1.3 0.0 0.6 1.2 40.3 26.8 22.1 8.1 0.4 1.5 0.9 54 25 13 4 0.62.4 1.3 29.015.0 50.7 1.8 0.0 0.7 1.4 36.4 25.4 26.2 9.4 0.0 1.5 1.1 49 24 20 3.4 0.4 1.8 1.6 28.215.4 51.9 2.2 0.0 1.3 1.0 32.8 20.3 38.9 5.8 0.3 0.7 1.2 46 23 24 3.7 0.6 1.7 1.6 26.318.0 51.2 1.9 0.3 1.1 0.8 31.9 20.6 41.2 4.6 0.4 0.4 0.8 45 21 25 5 0.32.5 1.6 27.318.5 50.5 1.8 0.0 0.8 1.1 34.8 25.5 27.9 9.6 0.2 1.2 0.7 41 19 31 5.2 0.4 2.5 1.1 27.719.3 48.9 2.0 0.0 0.9 1.1 33.8 25.0 33.6 4.8 0.9 0.9 1.0 39 17 35 4.9 0.3 1.5 1.5 25.115.9 54.4 2.0 0.4 1.0 0.9 33.5 26.5 27.6 10.7 0.2 0.9 0.5 40 21 29 5.3 0.2 2.1 3.0 27.419.2 49.8 1.6 0.0 1.3 0.7 27.8 19.8 45.5 3.6 0.3 1.8 0.9 40 19 31 5.9 0.5 1.6 2.5 26.618.6 51.9 1.8 0.0 0.3 0.8 27.0 20.8 45.3 4.9 0.4 1.0 0.7 37 18 38 4.7 0.0 0.8 2.1 20.914.3 61.8 1.6 0.3 0.3 0.9 25.3 20.8 46.6 5.4 0.4 0.8 0.7 Experimental synthesis of chlorite 295

FIG.10. Chemistry of the clay phase of the starting-sample (open square), 1week (open circles), 1month (filled triangles), 3month (open diamonds) and 9month (filled squares) CIP,IP and WIPrun samples. Si vs.interlayer cations (I.C. =Na+2Ca+K), from structural formulae calculated on the basis of 14 oxygens. Reference minerals are reported: low-charge and high-charge smectite (LC-Sm,HC-Sm), low-charge and high-charge beidellite (LC-Bei, HC-Bei), muscovite (Mu), chlorite (Chl). performedon thesame particle in order to study the transformedparticle s.This evoluti onis more 3+ relationshipbetween the Fe /Fetot ratioand the Fe efficientand rapid at the contact of the Fe plate. contentof the clay phase. Results are reported in Crystal-chemicalchanges of the run samples Fig.9c and d. The Fe enrichmentof the clay phase (EDS). Thegraphs presented in Figs 10 and 11 3+ iswell correlated with a decreasein the Fe /Fetot showa significantevolution in some cation site ratiotowards 0. This means that the newly formed occupanciesof the run products. The scatter of the claymineral is mostly an Fe 2+-richphase and analyticalpoints for each run product indicates a intermediatecompos itionsare due to pa rtly chemicalheterogeneity, which can be due to partly

FIG.11. Chemistry of the clay phase of the starting-sample (open square), 1week (open circles), 1month (filled triangles), 3month (open diamonds) and 9month (filled squares) CIP,IP and WIPrun samples. Fetotal vs. Mg, from structural formulae calculated on the basis of 14 oxygens. Reference minerals are reported: chamosite (Cham), clinochlore (Clino) and amesite (Am). 296 D.Guillaume et al. transformedparticles or mechanical mixing of experimentand the decrease of Si content(faster at particleseven at the smallest scale. The main thenearest contact with Fe plate).In the absence of changesin clay particl ecompositionare the addedFe plate(WIP run samples) the evolution is following. comparableto that of IP runsample. Tetrahedralsite: in both CIP andIP runsamples, Theoverall evolution of the clay (shown by the theSi contentdecreases drastically to a valueof 2.9 Si vs.interlayercations (I.C.), Fig. 10) can be forthe longest experimental duration (Fig. 10). explainedby two main changes in agreement with Thereis a continuousdecrease in Si contentfor CIP thepreliminary interpretation of the TMS, EELS, runsamples, at sub-constant interlayer charge first, CEC, HRTEM andXRD data;(1) a changein the andafter 3 months,at decreasing interlayer charge natureof the expandable layers, with increasing downto a valueof 0.2. The decrease in Si content tetrahedralcharge, from montmorillonite to high- islower in IP thanin CIP runsamples for durations chargesmectite; (2) the formation of chlorite. of1 weekto 3 months.However, the Si and Mixing(or interstratificat ion)between the starting interlayercharge are similar after 9 monthsin montmorilloniteand the high-ch argesmectit e bothCIP andIP runsamples. In the case of WIP explainsthe first trend, then mixing (or inter- runsamples, the transformation is different and the stratification)betweensmectiteand c hlorite decreasein Si contentonly reaches a valueof 4.1. explainsthe second trend. In the presence of a Octahedralsite: in CIP runsamples, a spectacular relativelylarge amount of Fe (presenceof added Fe increasein the total Fe contentis determined plate)these transformations are enhanced. (Fig.11). After 1 month,the CIP runsample Thehigh-charge smectite is a saponite,according contains2.5 Fe and3 to3.5 after 9 months,with tothe compositional field in diagrams and the 3+ Fe contentsup to 4.6 in some cases. In all CIP run Fe /Fetot ratio.Concerning the chlorite, crystal- samples,the Mg content remains sub-constant. For chemistryshows a solidsolution between an Fe-rich IPrunsamples, a similarincrease in Fe contentis end-memberand an Fe-Mg-rich chlorite depending noticedwith time; however, the highest Fe content onthe distance from the added Fe plate.The reachesa maximumvalue of 2.5 after 9 months,for recommendedname for a ferrous-richchlorite with aMgcontent of 1, e.g. a valuecompatible with an suchan Fe/ Al contentclose to 2/ 1ischamosite Fe-for-Mgsubstitution at constant R2+ (±3.5). For (Newman,1987). The Fe-Mg chlorite belongs to the WIPrunsamples, the same evolution as for IP is ‘amesite’series, amesite, a 7A Ê layersilicate, being observedwith Fe andMg content only reaching the usedas a compositionend-member only. valuesof 2 and0.6, respectively. Theoverall evolution of theCIP, IPandWIP run Solution chemistry samplesis well discrimin atedin the diagram Fe/(Fe+Mg) vs.Si (Fig.12). The discrimination of Thesolutions co-existing with the run samples thetwo series of CIP andIP runsamples is clear as wereanalysed when available. The results are afunctionof the Fe/ Mgratio, the duration of the summarizedin Table 4. The pH measured after

FIG.12. Chemistry of the clay phase of the starting-sample (open square), 1week (open circles), 1month (filled triangles), 3month (open diamonds) and 9month (filled squares) CIP,IP and WIPrun samples. Fe/(Fe+Mg) vs. Si, from structural formulae calculated on the basis of 14 oxygens. Experimental synthesis of chlorite 297

TABLE 4. Chemistry of the initial solution, starting solution (after 6days’equilibrium), WIPand IPrun solutions.

Sample pH Eh (mV) SiAl Fe Mg Ca Na K Cl Na/Ca

Initial solution 6.4 121 0000200 632 01032 5.5 WIP 25ºC Starting 7.2 –550 701427 1272 18 1035 83 300ºC 3months 5.5 –293 227 10011 802 34 n.a. 131 9 months 6.3 –157 76 10115 876 55 1362 101 IP 300ºC 1 day n.a. n.a. 486 36 17 925 908 82 1284 62 1week n.a. n.a. 410 10 5622 762 96 1232 62 1 month 6.5 –233 442 24223 913 72 n.a. 69 3 months 5.8 –260 265 1 2 1 9762 35 n.a. 150

All concentrations in mg/l. n.a.: not analysed cooling(at room temperature) and filtration of the presenceof metallic Fe (Fig.13b). This might be experimentalsolution is presented as a functionof dueto a significantdissolution of metallic Fe and theexperiment duration in Fig. 13a. There is an itspassivation by magnetite or maghemite around initialdecrease of run solution pH from 7.2 to 5.5, theFe plate. followedby an increase up to 6.3 for a durationof Theevolution with time of the composition of 9months.Eh measured after cooling (at room therun solutions is reported in Fig. 14a –d. After temperature)and filtration of the experimenta l 1weekof reaction with the bentonite at 300º C, the solutionis~ –260mVafter3monthsand Na+/Ca2+ ratioof the reacting solution is ~60 (± 2), –213mV after9 months,i.e. significantly higher butthen increases up to values ranging from 100 to thanin the starting solution ( –550mV) inthe 150depending on the experiments (Table 4). These changesin Na +/Ca2+ areprobably the result both of exchangeprocesses, and uptake by framework silicateslike feldspars and zeolites .TheCl – contentcan be considered as sub-constant, and is notaffected, as expected,by mineralogical changes. Theconcentrations of Si, Al, Fe andMg in the experimentalsolutions (Fig. 14a,b) display similar evolutionswith time: a rapidincrease during the firstwe eka ndt henastrongd ecreasef or experimentsof longer duration. The presence of Mg,Al andFe inthe solution during the first week suggestsa partialdissolution of smectite and Fe- bearingphases. Siliconis the most abundant element (except for Naand Cl initiallypresent in the experimental solution).The Si concentrationis high after 1 week (~500mg/ l,Fig. 14a). This result also suggests a rapiddissolution of several smectite layers. Then, theSi concentrationdecrea sesto ~250 mg/ l (3months), probably tending to values controlled bythe equilibrium with quartz. Indeed, at 300º C in moderatelysaline solution, Si saturationconcentra- tion(equilibrium with quartz) may be estimated at ~300 –350mg/ l(Wolery,1992; Wolery & Daveler, FIG.13. Chemistry of run solutions. Evolution of 1992).The val ueava ilablefor t helo ngest pH (a) and Eh (b) vs.duration of experiments. experimentis much smaller. The Si contentin the 298 D.Guillaume et al.

experimentalsolutions for the shortest duration explainwhy quartz crystall izedin signific ant amounts.The analyses of theexperimental solutions forexperiments with the longest durations revealed verysmall Fe andAl concentrations.These elementsare probably incorporated in the newly formedclays, zeolites and feldspars.

D I S C U S S I O N Evolution of Fe-bearing phases

TheXRD andTMS datademonstrated that: (1)magnetite is transformed to maghemite and the totalamount of magnetite-maghe miteis stable; (2)the added Fe powderis rapidly consumed and, asthe amount of magnetite-maghe miteis stable, rapidlytransformed to the soluble form (Fe 2+) duringthe experiment and/ orincorporated into otherminerals.

Newlyformed non-clay minerals:quartz, feldspars and zeolite TheSEM andXRD datademonstra tedthe formationof newly formed quartz, felspars and zeolites.Other non-clay minerals that were present inthe starting bentonite are rapidly consumed. Silicais not fully incorporated within the newly formedclay minerals which are much poorer in silicathan the montmorilloni teof the starting

bentonite.SiO 2 crystallizesas subeuhedral quartz. Equationswritten with constant Al andusing the structuralformulae of the starting- and run samples indicatethat one mole Na/ Ca-smectitemay trans- formapproximately into 0.91 moles of Fe-chlorite and1.78 moles of quartz. The distribution of the silicareleased during the transformation of Si-rich minerals(smectite) into clay minerals with lower Si/Al ratiosand quartz is difficult because at least twoother types of silicat esmay crystal lize, feldsparsand zeolites. The occurence of such mineralsas reaction products is consistent both withthermodynamic modelling predictions and data fromother experiments on bentonites. Feldspars werecommonly observed among reaction products inhydrothermal tests on Na- and K-montmorillonite between260 and 400º C (Eberl& Hower1977; FIG.14. Chemistry of run solutions. Evolution with Eberl1978). Inoue (1983) reported the formation of time of the composition (ppm scale) for different Cazeolites (wairakite and heulandite) as well as cations; Si(a), Al, Fe and Mg (b), Kand Ca (c), Na illiteand smectit e,from the reactio nofCa and Cl (d). montmorillonitewith 0.2 mol l – 1 KOH at300º C. Experimental synthesis of chlorite 299

Inaddition, zeolites and feldspars associated with 1991;Schiffman & Staudigel,1995; Beaufort et al., smectiteand chlorite are commonly observed in 1997).The main reason for such differences could alterationproducts of basic glasses of magmatic bethe experimental temperature (300º C) suffi- rockssubmitted to hydrothermal alteration (among cientlyhigh to prevent the formation of berthierine others,Cathelineau & Izquierdo,1985; Cathelineau ormixed-layered chlorite-smectite minerals. These et al.,1988;Schiffman & Fridleifsson,1991). mineralsare considered to be stable only at lower temperatures:stability field of berthierine from 65 to130º C (Iijima& Matsumoto,1982); from surface Newlyformed clay phases temperatureto 100º C (Velde et al.,1975);C-S Combinedmineralogic aland crystal-ch emical mineralsformedbetween200and240ºC studiesof the run samples allowed us to understand (Schiffman& Fridleifsson,1991). andcharacterize the evolution of the clay minerals Inother cases, more complex mineral assem- duringthe experiments. The XRD, CEC, TMS, blageshave been described, such as saponite- EELS, EDS andHRTEM datademonstrated that chloriteassociation as the ultimate product of hydrothermaltreatment of the bentonite at 300º C in alterationof beidellite (Schiffman & Fridleifsson, thepresence of Fe resultedin successive transfor- 1991).Such association is closeto our experimental mationof montmorillonite to saponite and triocta- runproducts association. hedralchlorite (± randomly-interstra tifiedchlorite- smectite). CONCLUSIONS Thesmectite-to-chlo ritetransformation occurred bothvia a dissolution-precipitationprocess and via Experimentsat 300º C inthe presence of Fe powder aprogressive,solid-state control, transformation of andmagnetite resulted in significant mineralogical theparticle s.Partly transfor medparticle sare changes,such as the formati onof saponi te, detectedby th eirint ermediatec ompositions trioctahedralchlorit e(±randoml y-interstratified betweenthe two end-members. chlorite-smectite),quartz, feldspars and zeolite. Thepresence of the Fe platein the system triggers Comparisonwith natural systems reactions.Besides, Si, Al, Fe andMg in thesolution showa concomitantstrong decreas eintheir Thefinal Fe mineralassemblage dominated by concentrations,probably due to their incorporation magnetiteand the expected f iscompatible with a O2 innew silicates. naturalreducinge nvironmentsuc hasdeep Concerningnuclear waste repositori es,these geothermalsystems, or deep oceanic sediments. experimentalsimulations demonstrated an evolution Therefore,the similarity between the newly formed towardclaymineralshavinglowerCEC. mineralassemblages in natural systems and the Meanwhile,this transformation is observed at the experimentalrun products obtained in this study nearestcontact with the added metallic Fe plateand (chlorite,zeolite s,quartz, saponite )mightbe underspecific experimental conditions that are not emphasized. representativeof a nuclearwastes repository site Amaindifference in the run product described in wherehigh temperature, and large amounts of water thepresent experiments with many natural mineral andFe prevail. assemblagesis the scarcity of mixed-layered clays. Thisstudy also demonstrated the necessity of Thegradua lorcontin uoustransfo rmationof usingcomple mentarytechni quesoperat ingat smectiteto chlorite, as accommodated through differentscales for a completeunderstanding of randomlyinterlayered C-S andregularly inter- thetransformation that occurred during experi- layeredC-S (corrensite),is generally reported ments. (Helmold& vande Kamp 1984; Chang et al. 1986;Bettison & Schiffman,1988; Bevins et al. ACKNOWLEDGMENTS 1991).However, as already stated by Betisson- Varga& Mackinnon(1997), the decrease in Experiments were conducted atthe Ge´ologie et Gestion smectiteproportion is discontinuous as shown by des Ressources Mine´rales et Energe´tiques laboratory thetransition from smectite (with <20% chlorite) to (G2R, CNRS-CREGU-INPL-UHP,Vandœuvre-le `s- corrensiteand chlorite (with <10 % smectite) Nancy, France). The XRDdata were collected at the (Inoue et al.,1984;Inoue, 1987; Inoue & Utada, Laboratoire Environnement et Mine´ralurgie (LEM, 300 D.Guillaume et al.

CNRS-INPL,Vandœ uvre-le `s-Nancy, France), the SEM structural variations of phyllosilicates from the Point work was carried out at the Universite´Henri Poincare´ Sal ophiolite, California. American Mineralogist , 73, (Vandœuvre-le `s-Nancy, France), the TMSwork at the 62 –76. Laboratoire de Chimie Physique et Microbiologie pour Bettison-Varga L.&Mackinnon I.D.R. (1997) The role l’Environnement (LCPME,CNRS-UHP,Villers-le`s- of randomly mixed-layered chlorite/smectite in the Nancy, France), the HRTEMimaging at the Centre de transformation of smectite to chlorite. Clays and Recherche sur les Mećanismes de Croissance Clay Minerals , 45, 506 –516. Cristalline (CRMC2, CNRS,National Facility for Bevins R.E., Robinson D.&Rowbotham G.(1991) TEM/AEM,Marseilles, France), the EDS-and EELS- Compositional variations in mafic phyllosilicates TEMworkat the Universite´HenriPoincare´ from regional low-grade metabasites and application (Vandœuvre-le `s-Nancy, France), and the ICP-AES of the chloritegeothermometer. Journal of and ICP-MSanalyses of solutions at the Centre de Metamorphic Geology , 9, 711 –721. Recherches Pe´trographiques etGeóchimiques (CRPG, Bowers T.S.,Jackson K.J. &Helgeson H.C.(1984) CNRS,Vandœuvre-le `s-Nancy, France). Equilibrium Activity Diagrams for Coexisting The authors wish to thank J.Ghanbaja (UHP, Minerals and Aqueous Solutions at Pressures and Vandœuvre-le `s-Nancy, France) for EELSand EDS Temperatures to 5kb and 600ºC .Springer-Verlag, analyses, I.Bihannic and F.Lhote (LEM,Vandœuvre- Berlin. le`s-Nancy, France) for their help and technical Bystro¨m-Brusewitz A.M. (1975) Studies of the Litest to assistance with the XRD,and S.Nitsche (CRMC2, distinguish beidellite and montmorillonite. Marseilles, France) for HRTEMimaging. This research Proceedings of the International Clay Conference, was supportedfinancially by Andra –Agence Mexico City, pp. Pp. 419 –428, Applied Publishing Nationale pour la gestion des Dećhets RadioActifs Ltd., Wilmette, Illinois, USA. (French national agency for the management of Cathelineau M.&Izquierdo G.(1988) Temperature- radioactive wastes). composition relationships of authigenic micaceous Wethank L.Bettison-Varga, an anonymous minerals in the Los Azufres geothermal system. reviewer and A.M.Karpoff, the Associate Editor, for Contributions to Mineralogy and Petrology , 100, their constructive comments. 418 –428. Cathelineau M., Oliver R., Nieva D.&Garfias A.(1985) Mineralogy and distribution of hydrothermal mineral REFERENCES zones in Los Azufres (Mexico) geothermal field. Aagaard P., Jahren J.S., Harstad A.O., Nilsen O.& Geothermics , 14, 49 –57. RammM. (2000) Formation of grain-coating chlorite Cathelineau M., Mosser-Ruck R.&Charpentier D. in sandstones. Laboratory synthesized vs. natural (2001) Interactions fluides/argilites en conditions de occurrences. Clay Minerals , 35, 261 –269. stockage profond des dećhets nucleáires. Inte´reˆtdu Alt J.C., Honnorez J., Laverne C.&Emmermann R. couplage expe´rimentation/mode´lisation dans la com- (1986) Hydrothermal alteration of 1km section pre´hension des mećanismes de transformation des through the upper oceanic crust, DSDPHole 504B: argiles et la pre´diction a`long terme du comporte- Mineralogy, chemistry and evolution of seawater- ment de la barrie`re argileuse. Pp. 305 –341 in: Actes des Journeés Scientifiques ANDRA ,EDPSciences basalt interaction. Journal of Geophysical Research , Nancy, France. 91, 10309 –10335. Chang H.K., Mackenzie F.T.& Schoonmaker J. (1986) Amouric M.&Olives J. (1991) Illitization of smectite as Comparison between the diagenesis of dioctahedral seen by high-resolution transmission electron micro- and trioctahedral smectite, Brasilian offshore basins. scopy. European Journal of Mineralogy , 3, Clays and Clay Minerals , 34, 407 –423. 831 –835. Cuadros J.&Linares J.(1996) Experimental kinetic Beaufort D., Baronnet A.,Lanson B.&Meunier A. study of the smectite-to-illite transformation. (1997) Corrensite: asingle phase or amixed layered Geochimica et Cosmochimica Acta , 60, 439 –453. phyllosilicate of the saponite-clorite conversion Eberl D.(1978) Series for dioctahedral smectites. Clays series? The case study of the Sancerre-Couy deep and Clay Minerals , 26, 327 –340. drill-hole (France). American Mineralogist , 82, Eberl D.&Hower J.(1977) The hydrothermal 109 –124. transformation of sodium and potassium smectite Benali O., Abdelmoula M., Refait P.&Geńin J.M. into mixed-layer clay. Clays and Clay Minerals , 25, (2001) Effect of orthophosphate on the oxidation 215 –227. products of Fe(II)-Fe(III) hydroxycarbonate: the Eberl D., Velde B.&McCormick T.(1993) Synthesis of transformation of green rust to ferrihydrite. illite-smectite from smectite at earth surface tem- Geochimica et Cosmochimica Acta , 65, 1715 –1726. peratures and high pH. Clay Minerals , 28, 49 –60. Bettison L.&Schiffman P.(1988) Compositional and Egerton R.F.(1986) Electron Energy-loss Spectroscopy Experimental synthesis of chlorite 301

in the Electron Microscope .Plenum, NewYork. Bloomington, Indiana, USA. Gillery F.H.(1959) The X-ray study of synthetic Mg-Al Inoue A.&Utada M.(1991) Smectite-to-chlorite serpentines and chlorites. American Mineralogist , transformation in thermally metamorphosed volca- 44, 143 –152. noclastic rocks in the Kamikita area, North Honshu, Greene-Kelly R.(1953) The identification of montmor- Japan. American Mineralogist , 76, 628 –640. illonoids in clays. Journal of Soil Science , 4, Inoue A., Utada M., Nagata H.&Watanabe T.(1984) 233 –237. Conversion of trioctahedral smectite to interstratified Guillaume D., Neaman A., Mosser-Ruck R., Dubessy J., chlorite/smectite in Pliocene acidic pyroclastic sedi- Cathelineau M.&Villie´ras F.(2001a) Experimental ments of the Ohyu district, Akita Prefecture, Japan. study of hydrothermal reactivity of bentonite at 80 Clay Science , 6, 103 –106. and 300ºC in the presence of iron and/or iron oxides. Lagaly G.,Fernandez Gonzalez M.&Weiss A.(1976) BerichtederDeutschenMineralogischen Problems in layer-charge determination of montmor- Gesellshaft, Beihefte zum European Journal of illonites. Clay Minerals , 11, 173 –187. Mineralogy, 13, 69 pp. Madsen F.T.(1998) Clay mineralogical investigations Guillaume D.,Pironon J.&Ghanbaja J.(2001b) related to nuclear waste disposal. Clay Minerals , 33, Valence determination of iron in clays by electron 109 –129. energy loss spectroscopy. Berichte der Deutschen MeunierA.,Inoue A.&Beaufort D.(1991) Mineralogischen Gesellshaft, Beihefte zum European Chemiographic analysis of trioctahedral smectite- Journal of Mineralogy , 13, 70 pp. to-chlorite conversion series from the Ohyu Caldera, Helmold K.P.& van de Kamp P.C.(1984) Diagenetic Japan. Clays and Clay Minerals , 39, 409 –415. mineralogy and controls on albitization and laumon- Mosser-Ruck R., Pironon J., Cathelineau M.&Trouiller tite formation in Paleogene arkoses, Santa Ynez A.(2001) Experimental illitization of smectite in a Mountains, California. Pp. 239 –276 in: Clastic K-rich solution. European Journal of Mineralogy , Diagenesis (D.D.McDonald &R.C.Surdam, 13, 829 –840. editors). American Association of Petroleum Mu¨ller-Vonmoos M., Kahr G., Bucher F., Madsen F.T. Geologists Memoir 37. &Mayor P.A.(1991) Untersuchungen zum Hoffman U.&Klemen E.(1950) Loss of exchange- Verhalten von Bentonit in kontakt mit Magnetit und ability of lithium ions in bentonites on heating. Eisen unter Endlagerbedingungen . NTB 91-14. Zeitschrift fu¨rAnorganische und Allgemeine Nagra, Hardstrasse 73, CH-5430 Wettingen, Chemie, 262, 95 –99. Switzerland. Hoffmann J. &Hower J.(1979) Clay mineral assem- Murad E.(1998) Clays and clay minerals: What can blages as low grade metamorphic geothermometers: Mo¨ssbauer spectroscopy do to help understand Application to the thrust-faulted disturbed belt of them? Hyperfine Interactions , 117, 39 –70. Montana, USA. Society of Economic Paleontologists Murakami T.,Sato T.&Inoue A.(1999) HRTEM and Mineralogists Special Publication , 26, 55 –79. evidence for the process and mechanism of saponite- Horton D.G.(1985) Mixed-layer illite/smectite as a to-chlorite conversion through corrensite. American paleotemperature indicator in the Amethyst vein Mineralogist , 84, 1080 –1087. system,Creeddistrict,Colorado,USA. Newman A.C.D., editor (1987) Chemistry of Clays and Contributions to Mineralogy and Petrology , 91, Clay Minerals . Monograph 6.Mineralogical Society, 171 –179. London. Humphreys B.,Kemp S.J.,Lott G.K., Bermanto Robinson D.&Santana de Zamora A.(1999) The Dharmayanti D.A.& Samsori I.(1994) Origin of smectite to chlorite transition in the Chipilapa grain-coating chlorite by smectite transformation: an geothermal system,ElSalvador. American example from Miocene sandstones, North Sumatra Mineralogist , 78, 607 –619. Back-Arc Basin, Indonesia. Clay Minerals , 29, Robinson D., Schmidt S.Th. &Santana de Zamora A. 681 –692. (2002) Reaction pathways and reaction progress for Iijima A.&Matsumoto R.(1982) Berthierine and the smectite-to-chlorite transformation: evidence chamosite in coal measures of Japan. Clays and Clay from hydrothermally altered metabasites. Journal Minerals, 30, 264 –274. of Metamorphic Geology , 20, 167 –174. Inoue A.(1983) Potassium fixation by clay minerals Schiffman P.&Fridleifsson G.O.(1991) The smectite- during hydrothermal treatment. Clays and Clay chlorite transition in drillhole Nj-15, Nesjavellir Minerals, 31, 81 –91. geothermal field, Iceland: XRD,BSE, and Electron Inoue A.(1987) Conversion of smectite to chlorite by Microprobe Investigations. Journal of Metamorphic hydrothermal diagenetic alterations, Hokuroku Geology, 9, 679 –696. Kuroko mineralization area, Northeast Japan. Schiffman P.&Staudigel H.(1995) The smectite to Proceedings of the International Clay Conference, chlorite transition in afossil seamount hydrothermal Denver, pp. 158 –164. The Clay Minerals Society, system: the basement complex of LaPalma, Canary 302 D.Guillaume et al.

Islands. Journal of Metamorphic Geology , 13, Cretaceous rocks from northeastern Algeria. 487 –498. Journal of Sedimentary Petrology , 39, 1275 –1280. Small J.S., Hamilton D.L.&Habesch S.(1992) Wolery T.J. (1992) EQ3NR,a computer program for Experimental simulation of clay precipitation within geochemical aqueous speciation solubility calcula- reservoir sandstones 1: Techniques and examples. tions: theoretical manual, user’s guide and related Journal of Sedimentary Petrology , 62, 508 –519. documentation (Version 7.0). UCRL-MA-110662- Vali H.&Hesse R.(1990) Alkylammonium ion PT-III, Laurence Livermore National Laboratory, treatment of clay minerals in ultrathin section: A Livermore, California. new method for HRTEMexamination of expandable layers. American Mineralogist , 75, 1443 –1446. Wolery T.J. &Daveler S.A.(1992) EQ6, acomputer Velde B.(1973) Phase equilibria studies in the system program for reaction path modeling of aqueous MgO-AlO-SiO-UO: chlorite and associated miner- geochemical systems: theoretical manual, user’s als. Mineralogical Magazine , 39, 297 –312. guide and related documentation (Version 7.0). Velde B.,Raoult J.-F.&Leikine M.(1975) UCRL-MA-110662-PT-IV, Laurence Livermore Metamorphosed berthierine pelletsin mid- National Laboratory, Livermore, California.