Canadion MineralogistVo l .2 l , pp . 611-623 (1983)

LEACHING STUDIES OF NATURAL AND SYNTHETIC TITANITE,A POTENTIAL HOST FOR WASTES FROM THE REPROCESSINd

OF CANADIAN NUCLEAR FUELT

P.J. HAYWARD, F.E. DOERN, E.v. CECCHETTo eNn S.L. MITCHELLAtomic Energy of Canada Limited, Whiteshetl Nuclear Research Establishment, Pinawa, Manitoba R}E ILT

ABSTRACT

Glass ceramics Q'.e., glasses subjected to controlledcrystallization) with synthetic titanite as the major crystallinephase are being considered as potential hosts ior theradioactive wastes arising from possible future reprocess_ing of nuclear fuel in Canada. In order to assess the;tabilitvoJ titanite in the anticipated environment of a disposal vauitsited 500-1ffi0 m deep within a granitic pluton in the Cana_dian Shield, leaching experiments have been performed withnatural and synthetic titanite, using a slmthetic groundwaterwhose composition is based on findings from a recentborehole-survey. The results are in qualitative agreementv/ith calculations of solution equilibria for titaniie and itsmain alteration products, and indicate that titanite shouldbe stable and suffer no net leaching under anticipated con_ditions in the vault.

Keywords: nuclear waste, titanite, stability, solubility,leaching, groundwater.

SolaNaatns

On 6tudie actuellement la possibilitd d,uriliser, commehdte, la c6ramique de verre (c'est-ir-dire le verre soumis iune cristallisation contr6l€e), dont la titanite synthdtiqueconstitue la principale phase cristalline, pour les ddchetsradioactifs pouvant provenir du retraitement dventuel ducombustible nucl6aire au Canada. A I'aide d'une eau phr6a-tique synth6tique dont la composition est d6riv6e des don-n6es obtenues lors de r6centes 6tudes de forage, on a pro-c€d6 d des expdriences de lixiviation en utilisant de la tita-nite naturelle ou synthetique dans le but de connaltre lastabilite de la titanite dans le milieu pr6vu d'une enceinted'€vacuation situde d 500-1000 mbtres de profondeur, ausein d'un pluton granitique du bouclier canadien. Du pointde vue qualitatif, les rdsultats correspondent aux calculs des6quilibres de solutions propres i la titanite et e ses princ!paux produits d'alt6ration; ils indiquent que la titanitedevrait Otre stable et ne pas subir de lixiviation nette en pr6-sence des conditions pr6vues dans I'enceinte.

Mots-cl6s: d€chets nucl6aires, titanite, stabilit6, solubilitd,lixiviation, eau phrfutique.

INTRoDUcTIoN

The Canadian concept for disposing of nuclear

*CANDU (CanadaDeuteium L/ranium) rea$ors use heavywater as moderator and coolant and natural UO, as fuel.A thorium-based fuel cycle is a furure possibiliiy.flssued as AECL 7856.

fuel wastes is based upon multiple barriers to inhi-bit the migration of radionuclides from a deep under-ground vault @oulton 1978). In addition to the naru-ral geological barriers, various engineered barriersare also being investigated. Two classes of nuclearfuel wastes are being considered for disposal,namely, irradiated CANDU't fuel and also the high-level radioactive wastes that would arise from pos-sible future reprocessing of irradiated fuel fiomCANDU reactors. High-inteerity materials are beingdeveloped to immobilize the latter class of wastes,and the disposal concept under investigation invol-ves burial of the high-integrity material (,,the wasteform") in an engineered vault, at a depth of500-1000 m, within a gr€udte or granodiorite plutonin the Canadian Shield in Ontario. The waste formwould be contained within a metal or ceramic canis-ter surrounded by a suitable "buffer" material (typi-cally bentonite-based), which would swell if con-tacted by underground water and thus retard accessof the water to the waste form. Finally, the vaultwould be backfilled and sealed.

The purpose of this system of engineered barriersis to retard the release of radionuclides to the bio-sphere until the radioactivity has decayed to negli-gible levels. The only credible mechanism for returuof radionuclides would be by partial dissolution ofthe waste form in underground water, followed bysolution or colloidal transport to the surface. Con-sequently, a major criterion in selecting a waste-formmaterial will be its geochemical stability or resistanceto leaching in a flooded vault under epithermalconditions.

An additional stipulation arising from the disposalconcept under consideration (Boulton 1980, Dixon& Rosinger 1981, Cameron & Strathdee 1979) is thatthe mean temperature of the vault, which may bevaried by design, should not exceed - 100"C. At oneproposed spacing of waste canisters, this temperaturecorresponds to an upper limit of - I wt.go fissionproducts in the waste form for wastes from irradiatedfuel that has been cooled for five years beforereprocessing. Cesium-l27 and strontium-9O are themajor heat-producing radionuclides, each with ahalf-life of approximately 30 years, so that the vaulttemperature would approach that of the host rockafter several hundred years.

6 1 1

6t2 THE CANADIAN MINERALOGIST

CHorcE oF WASTE-FoRM MATERTALS

Materials under investigation include borosilicateand aluminosilicate glasses, and glass ceramics bas-ed on the mineral titanite (CaTiSiO). The lalter,partly crystalline materials, are produced by con-trolled crystallization during reheating of precursorglasses from the system Na2O-Al2O3-CaO-TiO2-SiO2 (Hayward & Cecchetto 1982). The prod-uct consists of 0.5-5 pm titanite crystallites in amatrix of residual aluminosilicate glass. Waste ionsincorporated in the parent glass will partition intoeither the crystalline or the vitreous phase, or may,perhaps, form a separate phase. However, a majorfactor in the choice of titanite as a candidate phasefor waste immobilization is its well-documented abili-ty to take a wide variety of foreign ions into its struc-ture (Ribbe 1980).

In order to understand the leaching behavior ofthese composite materials, leaching experiments havebeen performed on representative glass-ceramics andalso on specimens of natural and synthetic titanite.This report discusses the results of the leaching ex-periments on titanite with reference to anticipatedconditions in the vault. Results for the glass ceramicswill be published separately.

AxrrcrparEo CuErratstnv oF THE GnouNowaren

A recent survey of more than fifty locations acrossthe Canadian Shield has confirmed that highly salinegroundwaters of the Ca-Na-Cl type occur widely atdepths below a few hundred metres (Frape & Fritz1981, 1982), the salinity generally increasing withdepth. This trend has been observed in samples fromdeep boreholes at Sudbury, Ontario, at Thompson,Manitoba, and at Yellowknife, N.W.T., and is il-lustrated in Figure I by groundwater compositionsfrom the Con mine, Yellowknife (Boyle 1979). Atdepths of 700 metres and below, Ca2* is the domi-nant cation, and Cl- is by far the dominant anion.The K* and Mg2* concentrations increase slightlywith depth, contrary to their behavior at Sudburyand Thompson. At Yellowknife, HCO3- andSOa2- concentrations fall to negligible levels withincreasing depth. Concentrations of dissolved silica,which were not measured in the Yellowknife andThompson data, are likely to lie between the satura-tion values for quartz (log HaSiOo molality :-4.0) and amorphous silica (log H4SiO4 molality: -2.7), Thus, dissolved silica in mine waters inthe Sudbury area give a mean log HaSiOa molalityof -3.2 for various depths down to 800 m.

too

50

c l -

Ca2+

N a +

soo loooBorehole sample depth, metres

Ftc. l. Variation of groundwater composition

roooBorehole sample depth, metres

with depth at the Con mine, Yellowknife, N.W.T.

!-r

'a

E@qc

o.9o

!o)i.9(!

(DoCo

.9o

o ro 500o

LEACHING STUDIES OF NATURAL AND SYNTHETIC TITANITE 6 1 3

TABLE 'I.

CATCULATED ACIIVITIES OF I'{A.JOR IONIC SPECIES IN THESYNT}IETIC OROUNDWATER IN THE TEMPEMTURE MNGE 25.150O C

-Ioc lrl

z5oc 6ooc loooc 150oc

It may be concluded from the above that anvpenetration of the engineered barriers within thevault by groundwater is likely to bring the waste forminto contact with solutions rich in Ca2+. Na* andCl-,.and poor in K*, Mg2*, SOo2- and HCO3*. Astandard synthetic groundwater designed to corres-pond approximately to the Yellowknife groundwaterat - 700 m depth has, therefore, been prepared foruse,in leaching experiments. Its composition (in mgL* ' ) is : caz" 15500, Na+ 5047. Cl - 34310.Sio2(aq) 50, Sr2* 20,Mg2* 200, K+ 50, So42- 790;total carbonate as HCOr- 10, and NO.- 50, witha total dissolved solids content of 56 g L-r. Thissynthetic groundwater has also been used as a modelcomposition in the calculations of thermodynamicstability, discussed below.

TirANrrE Srasrt-rry rN THE Vaulr ENvrnoNNasNr

Of the species present in natural brines and thesynthetic groundwater, reactions might be anti-cipated between titanite and (a) H,O*, (b) sulfatespecies in solution (SOl- and HSO.-) and (c) car-bonate species in solution (CO:2-, HCO3- andH2CO3). In this section, equilibrium conditions forthese three classes of reactions are presented, andsample calculations are made for titanite stabilitv inthe synthetic groundwater.

Free-energy data for these calculations have beentaken from the compilations of Robie et al. /19791and Barner & Scheuerman (1978). High-temperaturedata for sulfate and carbonate species are taken fromMarshall & Jones (1966) and Helgeson (1967),respectively. Saturation solubilities for quartz andamorphous silica, and free-energy data at hightemperature for HoSiOo have been computed fromRimstidt & Barnes (1980). Calculations of speciationin solution and of the activities of the resultingspecies have been made using the compuler programSOLMNEQ (Kharaka & Barnes 1973). Acrivity coef-ficients are calculated by SOLMNEe using the B'version of the extended Debye-Hrickel equation(Lewis & Randall 1961, Helgeson 1969) and are likelyto be in error for high-ionic-strength solutions, suchas the synthetic groundwater, by factors of up to 3for the species considered here (H.W. Nesbitt, pers.conun. 1983). The effect on the equilibrium calcula-tions, however, is expected to be relatively small.Sample calculations for sulfate and carbonatespecies, described below, confirm that the trends ofthe results remain unaffected.

Calculated activities of the major ionic species inthe synthetic groundwater are listed in Table I.Square brackets are used throughout to denote ac-tivities. Species present at very low activities, suchas HrSiOn-, H2SiO42- and HSO.-, are ignored inthe following calculations.

3 . 1 8

4.61

o,29

4.22

1 . 1 6

2.80

0.87

3 . 1 3

4.50

o . 2 5

3.80

1.09

2.73

0.85

3 . 1 1

4,43

o.23

7.85

8.14

4.70

5.10

3.30

2.93

5.43

9.92

6.7r

2 . 6 9

0.83

3.09

4.38

ceil+

t+Mg-

+Na'

4R '

- 2 +br

L.26

2.90

o.92

cl

'oi-E"O;

col-nco]

E2@3

nor.

E4StO4

n srof,t -

E2srol

E- (pE)

0.21

3.41

8.46

7.66

4.33

4.96

3 . 2 9

2.90

5.54

x0.23

7 . @

7 . L 7 6 . 2 9

9.00 10 .55

5.23 6 .02

5 . 1 9 5 . L 2

3.33 3 .37

2.94 2 .95

5.60 6 .24

10.23 11.43

6.37 5 .81

Eh (@1t6) - o.zo4 - 0.239 - 0.272 - 0.294

IonLc strstrgth

(t l2zf l2\

r .457 X.453 1 .450 r ,446

Reoction with HrO*

Activity diagrams for the H*-HrO-CaO-TiO2-SiO2 system have been published for 25.C(Nesbitt et a/. l98la, l98lb) and for 150.C(Hayward & Cecchetto 1982). The correspondingdiagram calculated for 100"C is shown in Figure 2.These diagrams demonstrate that, within thetemperature range anticipated for the waste vAult,and at [SiOr(aq)] values found in most naturalgroundwaters, either titanite or rutile is the stablephase in this system, depending on the log([Ca'z *

]/[H' ]2) values.

In general, for the reaction

CaTiSiO5 + 2H* + HrO + Ca2* + TiO, + H4SiO4

the equilibrium-activity expression is:

log ([Ca2*]/[H-]2)+log [H4SiO4]=log Kr (l)

Values of K,, the equilibrium constant for thereaction, have been computed as a function oftemperature in the range 25-150.C, and used tocalculate equilibrium values of log ([Ca2+]/[H*]2)at log [HoSiOa] values corresponding to (a) quartzsaturation, and (b) amorphous sil.ica saturation. The

Perovskite Titanitei it l

l ir iL Il ir\Ruti le

614 THE CANADIAN MINERALOGIST

+I

+oo

t 5

t o

F'T

+oo

J

Log [Hc SiOe I

Frc. 2. The system H+-H2O-CaO-TiO2-SiO2 at 100'C.Dashed lines (a) and (b) represent quartz saturation andamorphous silica saturation, respectively. Square andcircle represent the synthetic groundwater at pH 6,37(unbuffered) and at pH 8.5 (buffered), respectively.

results are shown graphically in Figure 3. Also shownin the figure are the log (Ca2+l/[H*]2) values com-puted for the synthetic groundwater in thistemperature range. The groundwater calculationsrefer 1o (i) unbuffered conditions, and (ii) solutionsbuffered to pH 8.5 by addition of a crushed CaO-TiO2-SiO2 glass frit (as discussed in the section onexperimental procedure).

Providing that the silica content of the ground-water is maintained at a value between the satura-tion level for quartz and that for amorphous silica,Figure 3 shows that titanite will be stable relative torutile in the presence of either solution (i) or (ii), thestabitty increasing at higher temperatures. If a sourceof Ti ions in solution were made available. as in thecase of the groundwater buffered to pH 8.5 by fine-ly powdered glass frit, titanite should precipitatefrom solution.

Reaction with SOo?-

For the reactions

CaTiSiO, + 2H' + SOo'- * H2O+casoo(s) + Tio2 + H4sio4

CaTiSiOr+2H* +SOo'- + 1.5HrO+CaSoa.0.5H2O(s) + Tio2 + H4Sio4

CaTiSiO, + 2H* + SC)o'- * 3H2O +CaSOa.2H2O(s) + TiO2 + H4SiO4

computations of the free-energy changes involved ineach reaction indicate that anhydrite is the prefer-red product in the range 150-60oC, with gypsum thepreferred product below 60oC. Hemihydrate ismetastable with respect to anhydrite or gypsumthroughout this temperature range.

t6.0

t4.o

t2.o

Titanite

- a \

( o )

Rut i le

( b

J

to.o

8 .O25 60 loo 150

Temperature, 'C

Frc. 3. Equilibrium log (lc*+l/lll*12) values in thetemperature range 25-l50oC for the reaction:CaTiSiO5 + 2H+ + H2O + Ca2+ + TiO2 + H4SiO4, atlog [HaSiOa] values corresponding to quartz saturationand amorphous silica saturation. Dashed line (a)represents log ([Caz+],/[H+]2) values for unbufferedsynthetic groundwater. Dashed line O) represents log(Caz+l/[H+1) values for buffered synthetic ground-water, pH 8.5.

The equilibrium expressions for the three reactionsall take the form:

log [HaSiOa]-log [SOo2-] -2log [H*]:log K2 A)

Equation (2) has been used to calculate values ofK2, the equilibrium constant for the preferred reac-tion (i.e., to give anhydrite above 60'C and gypsumbelow 60'C) in the temperature range 25-150"C. Byinserting the synthetic groundwater values for log[SOa2-] into the equation, the equilibrium values ofpH for the above reaction at temperatures in thisrange have been computed for log [HaSiOa] valuescorresponding to (a) quartz saturation, and (b) amor-phous silica saturation, and are shown in Figure 4.Also shown is the pH variation of the syntheticgroundwater with temperature. The results indicatethat titanite should be stable with respect to reac-tion with sulfate in the goundwater at temperaturesabove -25oC, provided that the level of dissolvedsilica remains close to these saturation values.

Errors in these calculations that result from theuse of the B' version of the extended Debye-Hiickelequation will be greatest for reactions involvinghighly charged species in solution, such as SOo'-.

-o-4-8-12

u.o525

Their magnitude and influence may be judged byassuming, as a worst possible case, that the calcutatedactivity-coefficients for SO.2- are in error by a fac-tor of 3. It then follows from Equation (2) rhat theboundary between titanite and (rutile + anhydrite) orbetween titanite and (rutile+gypsum) in Figure 4 willbe shifted by x. 0.24 on the pH axis. The temperaturefor equilibrium between titanite and its reaction pro-ducts at a given value of [H4SiO4] will be altered ac-cordingly. However, the trend shown in Figure 4,namely, the destabilization of titanite with respectto its reaction products as temperature is reduced,remains unchanged.

Reaction with carbonaceous species

The carbonaceous species occurring in ground-waters are COr2-, HCO3- and HrCOr, plusdissolved COr. The reactions between titanite andthese species are given by:

CaTiSiO, + CO.'- + 2H* * H2O =CaCO3 + TiO2 + H4SiO4

CaTiSiO, + HCO3- + H* * H2O +CaCO3 + TiO2 + H4SiO4

CaTiSiO, + H2CO3 + HrO+CaCO, +TiO2+ H4SiO4

Equilibrium equations for these three reactionsare, respectively:

615

60 roo r50

tog [HaSiOa] -2log [H+] -log

log [HaSiOa] - log [H+] - log [HCOr-] =log Ka @)

log [HoSiO.]- log [H2COr]= loe K5 (5)

where K3, Ka and l(5 are the equilibrium constantsfor the reactions. Values for log K:, log /(a and log"l(, have been calculated from free-energy data fortemperatures in the range 25-l50oc, and are shownin Figure 5. Formation of calcite in the reaction prod-ucts has been assumed, since it is the stable CaCO,polymorph at atmospheric pressure in thistemperature range. Although metastable aragoniteis a possible product of reaction at temperaturesabove - 40'C and in the presence of other cationsin solution (Deer et al. 1962), its formation insteadof calcite would have a negligible effect on the abovecalculations, since the free energies of formation arewithin - 0. I 9o of each other in the range 25-150'C.Also shown in Figure 5 are the correspondinglogarithmic activity-products, log AP3, log APo artdlog AP5, obtained by inserting the activities of theappropriate species in the synthetic groundwater(from Table l) into Equations 3-5 at severaltemperatures in this range. For each possible reac-tion, titanite stability is indicated by the conditionlog AP > log K.

The influence of errors in the calculated activity-coefficients for the charged species in solution can

LEACHING STUDIES OF NATURAL AND SYNTHETIC TITANITE

T-o zoo)(53

Ecfo

3 e.o(D-c

a

TITANITE

AmorphousSi l icaSaturation

GYPSUM +R U T I L E

o,"f -Saturation

ANHYDRITE+ RUTI LE

Temperature, 'C

Ftc. 4. Variation of equilibrium pH with temperature for the reactionCaTiSiO5 + 2H+ + H2O + SOoz- e CaSOa + TiO2 + H4SiO4, in syntheticgroundwater at specified [H4SiO4] values. The dashed line represents thecalculated variation in pH with temperature for the unbuffered syntheticgroundwater.

l co32- l :log ](: (3)

6t6

?o

l 5

- l o

THE CANADIAN MINERALOCIST

oo

5 t o(L

qs(!

5 4 O

- - 5

Los APj- co3-

Log Kg

Log APIg t\r4 HCO'Log Ka

Log AP

Log KsH2 CO3

of waste ions not easily accommodated in the titanitestructure, and (iv) ion exchange at the titanite sur-face between, for example, fission products in theCa sites and Ca2* from the groundwater. Thesecomplications will need to be addressed in future ex-perimental studies.

The remainder of this report describes leaching ex-periments with natural and synthetic titanite indeionized water and under conditions designed tosimulate a flooded waste-vault. The experiments wereintended to measure rates of release of ions fromtitanite in these leaching solutions, and do notnecessarily represent equilibrium situations. Never-theless, many of the experimental observations arein agreement with thermodynamic predictions, andsuggest that the simplified thermodynamic analysesdescribed above may be useful in predicting long-term stability. In contrast, kinetically-based leachingmodels require extrapolation beyond their timeframes ro be applicable to predictions of leachingbehavior over extended periods of time, and this ex-trapolation is always open to question.

ExprntnasNrat. PRocEDURES

Sample preparation ond analYsis

Samples of synthetic (ceramic) titanite wereprepared by melting a stoichiometric mixture ofCaCOr, TiO, and SiO, in a platinum crucible at1450'C, pouring the melt into water, drying theresulting "frit", reheating to 1050'C for one hourto crystallize the "frit", and then grinding underbutanol in an alumina jar with alumina grindingmedia. The product after drying is a fine (< l0 pm)powder, to which was added lVo stearic acid and 390wax binder in an ether - carbon tetrachloride solu-tion to form a slurry. The slurry was dried andgranulated to less than 800 pm, and 10-20 g quan-tities were pressed into 25-mm diameter pellets in ahardened steel die at 25 MPa. The pellets weresintered by heating at 5oC per minute to 1310"C andheld at this temperature for six hours. The finalceramic consists of interlocking CaTiSiO,crystallites 5-20 p.m in diameter, with an estimatedporosity of - 5-1090. Scanning-electron-microscope(SEM) examination of numerous fracture-surfacesshowed that the pores are typically 5-25 p'm indiameter and not interconnected. ASTM test C20(ASTM 1980) was used to confirm lhat a representa-tive sample is impervious to water.

Analyses have been performed on these ceramicsand on similar ceramics prepared with partialreplacement of calcium and titanium by simulatedfission products, uranium and thorium. Methods in-cluded HF-HCIOa dissolution followed by atomicabsorption (AA) spectroscopy, or neutron-activationanalysis, or X-ray fluorescence (XRF) spectroscopic

?5 50 rooTemperature, 'C

1 5 0

Ftc. 5. Variation with temperature of the logarithmicequilibrium-constants, log K3, log Ka and K5, for reac-tion between titanite and soluble carbonaceous species,calculated from equations 3-5. Also shown is the varia-tion with temperature of the logarithmic activity-products log AP3,log .4Pa and log AP5 in unbufferedsynthetic groundwater. Dashed lines explained in text.

be estimated, as in the preceding section, by assum-ing that the values used for COrz- und HCO3-arein error by a (maximum) factor of 3. It follows, fromEquations 3 and 4, that the log A\ and log APocurves in Figure 5 may be shifted vertically by upto t 0.48. This is represented by the areas withinthe dashed lines.

For all three reactions, a similar trend is seen,namely, convergence of the groundwater activity-products and the equilibrium constants as thetemperature falls toward 25oC. The implication fromFigure 5 is that titanite should be stable in thepresence of the CO32-, HCO3- and HrCO, in thesynthetic groundwater at all temperatures in thisrange, its stability increasing at higher temperatures.

The conclusion from these simplified ther-modynamic analyses is that titanite should be stablein the synthetic groundwater at all temperatures inthe range 25-150'C. However, it should be notedthat this conclusion may not be directly applicableto a crystalline waste-form based on titanite, sincethe following factors have not been considered: (i)radiation-damage and transmutation effects in thetitanite structure resulting from the incorporation offission products, (ii) stability-field changes due to theincorporation of fission products and processingchemicals (e.9., Na+, used for solution-scrubbingand adjustment of plutonium valency duringreprocessing), (iii) any grain-boundary segregation

analysis on powdered and solid specimens. Meanresults for the undoped ceramic were CaO 24.4,TiO240.9, SiO2 30.7, N2O32.9, other oxides (prin-cipally NarO, K2O and MgO) l.l wr.9o by dif-ference, indicating that the method of preparationhad produced slight loss of calcium and enrichmentin aluminum. X-ray-diffraction results showedtitanite to be the only detectable crystalline phase,and SEM examination of fractured surfaces show-ed no inclusions or obvious secondary phases.

Samples of natural titanite from North CrosbyTownship, Ontario, have also been used in these ex-periments; electron microprobe and XRF analysesgave CaO 27.7, TiO2 36.5, SiO2 30.3, FeO 1.25,MnO 0.25, Al2O3 1.7, NarO 0.04, V 0.5, La 0.3, y0.04, Zr 0.04, Sr 0.02 wt. r/0,

Leaching experiments

Monolithic samples of both the ceramic and NorthCrosby Township (NCT) ritanite have been used instatic leaching experiments, with both deionizedwater and synthetic groundwater as leachants. TheNCT titanite specimens were cut from the massivemineral using a fine diamond saw and light oil ascutting fluid. Orthogonal specimens of 10-20 mmside dimension were ground to 600-mesh finish us-ing carborundum paper and a light oil lubricant,followed by rinsing in acetone. Cylindrical ceramicspecimens, approximately l8 mm in diameter andl-4 mm high, were prepared and ground in a similarmanner.

Leaching experiments were performed in 150-mlsealed teflon vessels, heated to l00oC in a circulatingoven. In all cases, an apparent ratio of surface areato leachant volume of 0.1 cm-l was maintained.The surface-area measurements were madegeometrically, and thus did not take surfaceroughness or the presence of exposed pores into ac-count. Apparent leach-rates calculated for thesesamples would, therefore, be higher than true leach-rates. The problem of surface-area measurements inthe preparation of ceramic samples for leaching ex-periments has been discussed by Coles & Bazan(1982), who concluded that alternative methods suchas those based on measurements of N2 or Kr sorp-tion give results at least one order of magnitude toohigh.

Leaching was followed by weight-change meas-urements on the samples at 30-day intervals for upto one year, followed by re-immersion of the samplesin the leaching solutions. Solution analyses were per-formed at the termination of each test, and surfaceexaminations of selected specimens were made us-ing SEM combined with energy-dispersion X-rayanalysis (EDX). To identify some of the finersurface-precipitates and reaction products, extrac-tion replicas were made using acetyl cellulose foils,

617

which were then used for scanning-transmissionelectron-microscopy (STEM) EDX analyses of in-dividual grains.

Leaching experiments with the synthetic ground-water were performed in two ranges of pH. Somewere done with no added pH buffer (except fordissolved COr, introduced during the weight-changemeasurements); here, the pH remained in the range6.0-6.5. In some experiments, however, 5 g of apowdered glass (l'.e., amorphous) frit of composi-tion 30.8 CaO,3l,4 TiO2, 37.8 wt.9o SiO, was add-ed to increase the pH to the range 8.0-9.0 withoutthe addition of foreign ions to the system. The fritalso served as a source of Ca, Ti and Si ions, whichcould dissolve and reprecipitate as crystalline titaniteif it were thermodynamically stable under the ex-perimental conditions.

Finally, 5 g of powdered biotite granite (ColdSpring quarry, Lac du Bonnet, Manitoba) were add-ed to each teflon vessel to simulate rock-groundwaterinteractions. It was anticipated that the major effectof these additions would be rapid saturation of thegroundwater with respect to Al3+ and SiO, (aq). Itwas also recognized, however, that the use of addi-tions of powdered granite and glass frit allowed thepossibility of their adherence to the specimen sur-faces and interference with weight-changemeasurements, and that surface analyses would berequired to distinguish this situation.

Experiments involving titanite plus deionized waterwere performed in the same manner, except that noadditions of glass frit or powdered granite weremade.

ExprnttragxreI- RESULTS

NCT tilonite + deionized water

Determinations of solution concentrations andsample-weight changes were performed in triplicatedexperiments, terminated after 30, 60 and 90 days,respectively. Elemental leach-rates (kg.m-z.s-t;

TASLE 2. INDIVIDUAT ION ATID BUL( LEACH.RAITS, I{AIURA]. ND SYIITTFTIC TITAIIITE

LEACHING STUDIES OF NATURAL AND SYNTHETIC TITANITE

kilq Bdl &t6 dc&!d fr@ bcdq &ldld Myse

&tsdd hatld fr@.6.1oss Ml !at6 & Fle 9! rar6 L rar6

10 &y6

ao 2,4(2.0'

m e 5.4<2.9)t16dt6 90 2.3(1.4)

2.4(1.1) 5.6(3.8) 2.0(0.J) <0.0132.1(0.9) 5.9(?.2) 1.4(1.1) 0.02(0.02)r.4(0.2) 3,7(0.1) 1.3(0,5) 0.003(0.002)

30&90

t20

- 4 . 7 ( 0 . 9 )3.4(0.1) 3.5(0.2)2.7(0.!) 2.7(0.r)2.4(0.2) 2.7(0.1)

4.3G.3) rr .0(2.2) <0.244.2(0.2) 7.6(0.8)<0.163.2(0.2) J.7(0.5) 0,m3(0.@6)3 . 4 ( 0 . 4 ) 5 . 6 ( 0 . 4 ) < 0 . @ 2

Leach rtes ( i 10-t0) expressed ln kg.f2.s_!, f , i th standari devlat lon in paren-th€ses. l{aiural titaniie frch Iodh Ctusby Tomshlp, 0nt8d6. Leaching rascarr ied out at 100'C tn deionized Babr.

6 1 8 THE CANADIAN MINERALOGIST

Frc. 6. Scanning-electron photomicrographs of (a) titanite ceramic, fracture surface, and O) titanite ceramic, leached

120 davs in deionized water at 100"C.

were calculated from the equation: leach rate =CY/Fbt, where C; is the concentration of theanalyzed element or ion (kg L-t), V is the leachantvolume (L), A is the geometric surface-area (m2), tis the duration of the leaching experiment (s), andF1 is the element or ion fraction in the mineral asdefined by F,'={.W-2100 Wo' where X, is theweight percent of the oxide of element i in themineral, W; is the atomic weight of the element orion multiplied by the number of ions in the oxideformula, and Wo, is the molecular weight of the ox-ide. F1 relates the ion concentration in solution tothe ion concentration in the mineral, and theresulting leach-rate for each element or ion is thusnormalized to unit concentration in the mineral.

Bulk leach-rates were calculated both from theweight-loss measurements and from the ion concen-

trations in solution. Discrepancies between the twomay be due to (i) analytical or weighing errors, (ii)precipitation of surface phases from solution, (iii)formation of a hydroxylated layer on the mineral sur-face and (iv) back precipitation of a leached phaseonto the mineral surface.

Leach-rate data for NCT titanite in deionizedwater are summarized in Table 2. The bulk leach-rates determined from composition of solution andweight-loss measurements are, within experimentalerror, in agreement. The leach rates of the individualions demonstrate a preferential removal of Ca2*compared to Sia+, with Tia* almost immobile, sug-gesting thai a surface layer rich in a titanium oxidespecies is formed at the mineral surface. Metson e/al, (1982) independently observed the formation oftitanium-rich surface layers and selective leaching of

FT3eJ (.,

Ftui=Eo u JF ( )I <e&HA

80

60

40

20

o-20

-40-60

-80

o 12c 24o 360DAYS

Ftc. 7. Leaching results for North Crosby Township tilanite using (a) syntheticgroundwater plus crushed granite, pH 6.G{.5, and (b) synthetic groundwater pluscrushed granite plus CaO-TiO2-SiO2 glass frit, pH 8.0-9.0. Static tests at 100'C.

C** and Sia* on samples of leached titanite usingX-ray-photoelectron spectroscopy and secondary ionmass-spe$rometry.

Synthetic titonite + deionized water

Compositions of solution and weight-changemeasurements from triplicated experiments were us-ed to calculate bulk and elemental leach-rates in tesrsterminated after 30, 60, 90 or 120 days. The leachrates are presented in Table2, and are in good agree-ment with those obtained for NCT titanite except forthe Sia" rates, which are somewhat larger. Thehigher rates found for Sia* may be due to loss ofCa2* (and gain of Al3+) during ceramic fabrication,so that the resulting ceramic probably containedstoichiometric CaTiSiO, plus traces of surplus TiO,and SiO, together with Al2Or. Although CaTiSiO,was the sole phase identified by XRD and SEM-EDX, considerations of phase equilibriasuggest thatquartz, rutile, corundum and, possibly, mullite couldhave been present as minor coexisting phases andmay have contributed to an enhanced leach-rate fors i4* .

Scanning-electron photomicrographs of an un-leached and a leached ceramic surface are shown inFigure 6. Two distinct phases are seen to have pre-cipitated on the surface of the leached sample. SEM-EDX resolution was not adequate to analyze thefibrous phase. The prismatic phase was identified byEDX as being Al- and Si-rich, although the crystalmorphology did not appear to resemble those ofhydrothermal aluminosilicates such as kaolinite orpyrophyllite.

NCT titanite + synt hetic groundwater + crus hedgronite, pH 6.0-6.5

The mean results of weight-change measurements

6t9

made on three specimens in continuous experimentslasting 360 days are shown in Figure 7, curve (a).After an initial gain in weight, a slight but pro-gressively increasing loss was recorded, so that themean bulk leach-rate between days 90 and 360 was(2.2x.0.3) x 10-10 kg.6-2.s- t .

SEM-EDX examination of the leached surfaces(Fig. 8) showed that little dissolution of the matrixhad occurred (original polishing scratches were still

TABLE 3. OIIPOSITION OF PARTICLES FNOil EX]MCTION REPLICAS OF TEAC}IED TITANI]E

Mysls& !18 [ St r 6 n la I S

LEACHING STUDIES OF NATURAL AND SYNTHETIC TITANITE

FIc. 8. Scanning-electron photomicrographs of North Crosby Township titanite leached in synthetic groundwater pluscrushed granite, pH 6.0-6.5 for 360 days at 100.C.

tus iddtilld

([)

(rv)(v)

2 7

- 4 2 1 5 0 5

o . 5 7 9 3 9 0

20 56 23

1 - 2 6 6 5 -3 g -

4 l -

1 -

2 -

cilorita?

alt€rd? tr feld6Fr

chy dmtal?

23 62

26 70

2 4 6 4

2 9 6L 7 g

day turall

cby dGiel?

clat tfural?

cblodte?

5 4

L 2 L

3 3- l

- 2 7

(c)

(d )

(dr)(d!)

(dv)

( r )

7 2 -

6 0 -3 7 2

2 9 -

3 7 r t32 0.8 -

24

a

39 m@ or adyddto

- clay ddoral?

I - ro@tllte?

. ?

- tldt€

(d)

(Fl)

(dr)(dr)

(*)(s)

2 5 6 7 -

2 5 5 -

5 1 8 6 9 -

2 0 . 9 3 6 -

r - dst &6rd?

1n!6t8rm r@ts

Utdldt6?

cby deral?

34 t73 5 5

6 1 1

3 1 2 7 1

C@posttlons detendned by STf,ll-EoX analysls' eipressed in catJon rt. t'dcludlng oxygen and hydrcxyl . (a) l iodh Crcsby Tomship t ibnl te,360 daysln synthet ic grcunddater plus cdshed granite at 100"C, pH 6.0-6.5. (b) SJn-thet lc t lbnl te, s@ condit ions as (s). (c) l {orth Crcsby Tmshlp t i tani te,360 days ln synthetic grcundrat€r plus crushed granlte plus glass frlt at100oc, pH 8.0-9.0. (d) Synthet lc t i tu l t€ ' s@ condit lo is as (c).

620

visible). Surface phases, mainly K-Al-Si, werepredominantly those of debris from the addition ofcrushed granite. The fine precipitate visible in Figure8b was too small to permit EDX analysis.

STEM-EDX measurements were made on fiveparticles adhering to an extraction replica. The EDXmeasurements, converted to cation weight percents,are given in Table 3a. Although it was not alwayspossible to identify these phases on the basis of theEDX analyses and morphologies, the compositionscorrespond to granite or alteredgranite debris. Thehigh content of Fe in composition (ii) may representreprecipitated Fe leached from either the underly-ing titanire or the crushed granite.

Synthetic titanite + synthetic groundwater + crushedgranite, pH 6.0-6.5

Weight-change measurements were made on threespecimens in continuous experiments up to 360 days;the results are plotted in Figure 9, curve (a). Asomewhat larger initial weight-gain was found thanfor NCT titanite, and this persisted up to day 180,after which the ceramic lost weight, with a bulkleach-rate of (0.65 t 0.36) x l0-r0 kg.s1-2.s-t.

SEM examination of a leached surface after 360days showed little visible evidence of bulk dissolu-tion of the ceramic: the original polishing scratcheswere still clearly visible (Fie. l0). The closed poresin the original unleached ceramic (Fig. 6a), whichhad been exposed during cutting and grinding of thesample, are seen to be largely infilled with debrisfrom the crushed granite; EDX analysis of this debris

dt!E

t!ctfEfoF s lz El c )c eu J t .o-ul(,zToF?

IUJB

THE CANADIAN MINERALOGIST

t20 240 360DAYS

Ftc. 9. Leaching results for synthetic titanite using (a) synthetic groundwater pluscrushed granite, pH 6.0-6.5, and (b) synthetic groundwater plus crushed graniteplus CaO-TiO2-SiO2 glass frit, pH 8.0-9.0. Static tests at 100'C.

Frc. 10. Scanning-electron photomicrograph of synthetictitanite leached for 360 days at 100"C in syntheticgroundwater plus crushed granite, pH 6.0-6.5'

demonstrate the presence of quartz and K-Al-Siphases.

STEM-EDX measurements were performed onfive particles from an extraction replica, and the pro-portions of cations (wt. 9o) are quoted in Table 3(b).The results confirm the presence of granitic or alteredgranitic phases, and are similar to those of Table3(a).

NCT titanite * synthetic groundwoter + crushedgranite + gloss frit, pH 8,0-9.0.

Weight-change measurements were made for threespecimens during 360 days of leaching, and the mean

LEACHING STUDIES OF NATURAL AND SYNTHETIC TITANITE 621

Frc. ll. Scanning-electron photomicrograph of NorthCrosby Township titanite leached for 360 days at l@oCin synthetic groundwater plus crushed granite plusCaO-TiO2-SiO2 glass frit, pH 8.0-9.0.

. * . .results are shown in Figure 7, curve (b). In contrastto the results of the experiments at pH 6.0-6.5, theinitial gains in weight observed after 30 days leachingwere maintained for the duration of the experiment.

SEM examination of specimen surfaces after 360days leaching showed the buildup of a multiphasesurface-layer (Fig. ll). The elongate crystals wereindentified by EDX as a Ca-Ti-Si phase, and therhomboidal morphology is characteristic of titanite.Other phases were found to be Al- and Si-rich withsome K and Ca and traces of Na and Mg, and mostprobably represent original or altered granite debris.STEM-EDX analytical data on 5 particles from anextraction replica are given in Table 3(c), and are ingeneral agreement with the SEM-EDX results.

Synthetic titonite + synthetic groundwater + crushedgranite + gloss frit, pH 8.0-9.0,

Weight-change measurements were made for threespecimens during the course of 360 days of leaching,and mean results are shown in Figure 9, curve (b).The observed trend, namely, a significant gain inweight during the first 30 days followed by littleweight-change thereafter, is similar to that observedfor the NCT titanite samples, but approximately onone order of magnitude larger.

SEM examination of the specimen surfaces after360 days leaching showed a buildup of a multi-phaselayer (Fig. 12). The rhomboidal crystals were againidentified by EDX and morphology as titanite. ThecoralJike growth on the titanite crystals in the figureis a calcium silicate with traces of Mg, Al, K and Cl,possibly xonotlite CauSiuO,r(OH)2 or robermoriteCa5Si6O,r.5HrO.

Results of STEM-EDX analyses ol five grainsfrom an extraction replica of a leached surface aregiven in Table 3(d), and confirm the presence oftitanite. Composition (xviii) may correspond to in-

Frc. 12. Scanning-electron photomicrograph of synthetictitanite leached for 360 days at 100'C in syntheticgroundwater plus crushed granite plus CaO-TiO2-SiO2glass frit, pH 8.0-9.0.

tergrown titanite plus a hydrated calcium silicatephase, such as the coral-like phase observed in Figure12. The other compositions probably represent clayminerals or feldspars (or mixtures of the two).

DISCUSSIoN AND CoNCLUSIONS

The results of leaching of the two varieties oftitanite with deionized water are similar, and sug-gest that Ca and Si are preferentially leached to leavea surface layer enriched in Ti. This is consistent withthe thermodynamic arguments presented earlier,since rutile or, possibly, anatase or brookite (rutilepolymorphs) would be the stable phase under thesecondiiions (Fig.2). Nesbitt et al. (l98la) identifiedbrookite on the surfaces of perovskite leached indeionized water, which may be regarded as a similarreaction involving Ca2* leaching. The buildup of aTi-rich surface layer would progressively reduce theexposed surface-area of unleached titanite, so thatthe leach rates of Ca2* and Sia* in deionized watershould decrease with time until they reach a valuecontrolled by the rate of dissolution of theTiO2-enriched surface layer. There is evidence inour experiments of a slight fall with time in theCa2* and Sia* leach rates, but Figure 6 confirmsthat the surface TiO, layer is not sufficientlycoherent to constitute a significant barrier to furtherleaching. Longer-term experiments are required toconfirm this prediction.

The weight losses observed for both NCT and syn-thetic titanite in the experiments with slntheticgroundwater at pH 6.0-6.5 are not easily understood.Figure 3 suggests that titanite should be stable underthese conditions, and no visible evidence for leaching,such as etch pits or obvious alteration-products, wasfound in the SEM examinations of either material.At this pH range, and assuming AlrO, saturation(from the addition of crushed granite), calculations

622 THE CANADIAN MINERALOGIST

of mineral-solution equilibria using the SOLMNEQprogram indicate that the leachant is supersaturatedwith respect to at least 29 minerals, the majority be-ing clays. The early gains in weight in the experimentsare presumably due to precipitation of one or moreof these supersaturated phases on the surfaces of thespecimens. It is possible that part of the observedweight-losses was the result of repeated drying of thesamples prior to weighing, followed by rewettingwhen the samples were replaced in the groundwater,causing the adhering layer to loosen and separate.It is also possible that the local activity of Ca2* andH* at the surfaces of the titanite specimensdeparted significantly from the mean values in thesolution, making titanite a nonequilibrium phase andconsequently causing leaching of Ca2+ and Sia*ions from the surfaces. The observed weight-lossescorrespond to removal of approximately 0.1 pm ofmaterial from exposed surfaces; it is doubtfulwhether such a small loss would be visible usingSEM.

In the groundwater experiments at pH 8.0-9.0, theaddition of the glass frit raised the pH by surfaceexchange of Ca2* for H*, and also provided asource of Ca, Si and Ti ions in solution. The obser-vation that fresh titanite crystals formed on the sur-faces of both the NCT and the specimens of synthetictitanite confirms that titanite is a stable phase at highCa2* concentrations in solution or high pH whenthe HoSiOa activity is at, or close to, saturationvalues. Thus the gypsum or anhydrite deposited inthe surface layers [Table 3(c), analysis (xi)] was pro-duced by precipitation from the groundwater ratherthan from alteration of titanite. This deduction issupported by SOLMNEQ calculations of solutionequilibria for the groundwater at 100'C and pH 8.5,which indicate that the latter was significantly super-saturated with respect to gypsum and to anhydrite,and also by the failure to observe rutile, anatase orbrookite in the altered layers. In the same way, theunaltered nature of the titanite surfaces, where visi-ble through the precipitated layers, and the absenceof either calcite or aragonite in these layers, suggesta lack of reaction with carbonate species in solution.

The significant difference in magnitude of theweight gains observed for synthetic and NCT titaniteat 100"C and pH 8.5 may be due to the differingabilities of the ceramic surfaces and the mineral sur-faces to cause precipitation and adherence of super-saturaled phases from the groundwater solution, orto permit adherence of debris and alteration productsfrom the crushed granite. It should be noted that thesurfaces of the ceramic specimens contain numerousexposed pores, none of which are interconnected, butwhich could have acted as sites for precipitation fromsolution or for infilling by granite debris.

The overall conclusion from these experiments is,therefore, that titanite has excellent durability under

anticipated conditions in a waste repository. For suc-cessful development of a titanite glass-ceramic waste-form, however, it will be necessary to demonstratethat a significant fraction of the waste ions in theparent glass partitions into the titanite structure dur-ing controlled crystallization. This aspect is beingexplored at the Whiteshell Nuclear ResearchEstablishment, McMaster University and the Univer-sity of Western Ontario using ion-beam or electron-beam microanalytical techniques, in many cases attheir limits of spatial and spectroscopic resolution.

AcrNowrsncEMENTS

Sincere thanks are due to P.G. Richardson for theSTEM/EDX analyses, and to the staff of theAnalytical Science Branch at WNRE and J.B. Met-son of the University of Western Ontario for thesolution and solid analyses. The advice and com-ments of many colleagues, including B.W. Goodwin'R.J. Lemire, H.W. Nesbitt, J.B. Metson, G.M. Ban-croft, J.C. Tait, K.B. Harvey and A.G. Wikjord,are also gratefully acknowledged. The referees' com-ments were most useful in preparing the final ver-sion of this contributiorr.

REEenBNcrs

AvEnrcaN Socrsrv eNo MatsnraLs (1980): AnnualBook of ASTM Standards 17, C2, 6-8.

BanNrn, H.E. & ScsrurnueN, R.V. (1978): Handbookof Thermochemical Data for Compounds andAqueous Species. John Wiley & Sons, New York.

BoulroN, J., ed. (1978): Management of radioactivefuel wastes: the Canadian disposal program, AtomicEnergy Can. Ltd., Rep. AECL-6314.

(1980): Second annual report of the Canadiannuclear fuel waste management program. AtomicEnergy Can. Ltd,, Rep. AECL-6E04.

Boyrs, R.W. (1979): The geochemistry of gold and itsdeposits. Geol. Surv. Can. Bull.2E0.

Cen,renoN, D.J. & Srnersoee, G.G. (1979): Materialaspects of nuclear waste disposal in Canada. InCeramics in Nuclear Waste Managemeni (T.D.Chikalla & J.E. Mendel, eds.). U.S. Dep. EnergyRep. CONF-790420.

Corrs, D.G. & BezaN, F. (1982): Continuous-flowleaching studies of crushed and cored SYNROC.Nucl. Tech. 56. 226-237.

Drrn, W.A., HowrE, R.A. & Zussuall, J. (1962): Rock-Forming Minerals. Y, Non-Silicales. Longmans,London.

Drxor, R.S. & RosrNosn E.L.J. (1981): Third annual

report of the Canadian nuclear fuel waste manage-ment program. Atomic Energy Con. Ltd., Rep.AECL-6E2I.

FnepE, S.K. & Fnrrz, P. (1981): A preliminary reporton the occurrence and geochemistry of salinegroundwaters in the Canadian Shield. AtomicEnergy Can. Ltd., Tech. Rec. TR-136.

& - (1982): Saline groundwaters in theCanadian Shield, a first overview. Chem. Geol.36.r79-t90.

Havwenn, P.J. & Crccsnrro, E.V. (1982): Develop-ment of sphene-based glass ceramic tailored forCanadian waste disposal conditions. ,Irl ScientificBasis for Nuclear Waste Management (S.V. Topp,ed.). Elsevier, New York.

HrtcrsoN, H.C. (1967): Thermodynamics of complexdissociation in aqueous solution at elevatedtemperature. J. Phys. Chem. 71,312l-3136.

(1969): Thermodynamics of

623

350o evaluated from solubilities of calcium sulphatein sulphuric acid solutions. J. Phys. Chem. 70,4028-4040.

MErsoN, J.B., BaNcnoFr, G.M., KeNrrran, S.M.,Nessrrr, H.W., Fvnr, W.S. & Havwano, P.J.(1982): Leaching of natural and synthetic sphene andperovskite, 1lz Scientific Basis for Nuclear WasteManagement 5 (W. Lutze, ed.). Elsevier, New York.

Nrssr r r , H .W. , BeNcnor r , G.M. , Fvrs , W.S. ,KanxsaNrs, S.N., Nrsurriua, A. & SutN, S. (1981a):Thermodynamic stability and kinetics of perovskitedissolution. Nature 289, 358-362.

KenrseNrs, S,N. & Fvrr, W.S.(1981b): The stability of perovskite and sphene inthe presence ofbackfill and repository materials: ageneral approach. .Ir Scientific Basis for NuclearWaste Management 3 (C. Northrup, ed.). Plenum,New York.

Rrana, P.H., ed. (1980): Orthosilicates. Mineral. Soc.Amer., Rev. Mineral. 5.

Rrvsrror, J.D. & BenNEs, H.L. (1980): The kinetics ofsilica-water reactions. Geochim. C osmochim, Ac-ta 44.1683-1699.

Ronrr, R.A., HpurNcwav, B.S. & FrsHsn, J.R. (1979):Thermodynamic properties of minerals and relatedsubstances at 298.15 K and I bar (l0s pascals)pressure and at higher temperatures. U,S. Geol,Surv. Bull. 1452.

Received December 9, 1982, revised manuscript oc-cepted April 16, 1983.

LEACHING STUDIES OF NATURAL AND SYNTHETIC TITANITE

systems at elevated temperaturesAmer. J. Sci. 267, 729-804.

hydrothermaland pressures.

Kuenara, Y.K. & Ba,nNrs, I. (1973): SOLMNEQ:solution-mineral equilibrium'computations. U.S.Geol. Surv. Comp. Contr. PB-215-E99 (The ver-sion used was modified and updated by B.W. Good-win, Atomic Energy Can. Ltd.).

Lrwrs, G.N. & RaNoalr-, M. (1961): Thermodynamics(2nd edition). McGraw-Hill, New York.

MansHarr, W.L. & Joxps, E.V. (1966): Seconddissociation constant of sulphuric acid from 25 to