American Mineralogist, Volume 76, pages211-217, 1991

Dissolution rates of plagioclase at pH : 2 and 3

Wrr-r-rau H. C.nsnv,flnNnv R. Wnsrnrcn Geochemistry Research,Sandia National Laboratories,Albuquerque, New Mexico 87185, U.S.A. GBoncn R. Holonnx NSI Technology ServicesCorporation, E.P.A. Environmental ResearchLaboratory, 200 Southwest35th Street, Corvallis, Oregon 97333, U.S.A.

Ansrru,cr The dissolution rates of plagioclasein acids are very sensitive to the mineral composi- tion. At pH : 2.0 and25 oC, rates range from =l x l0 15feldspar-mol/cm2/s for albite to greater than I x l0 r, feldspar-mol/cm2lsfor anorthite. At pH : 3.0, the respective ratesrange from =3 x lQ-to feldspar-mol/cm2/sto=7 x l0-t4 feldspar-mol/cm2ls.The relationship betweenmineral composition and dissolution rate, however, is not simple. In order to better define such a relation at pH : 2, rateswere measuredon eight plagioclase minerals and compared to existing data. In general, the dissolution rates of a given plagioclasemineral estimated by different researchersgenerally agree to within a factor of two to five. The uncertainty is much larger for Ca-rich minerals such as bytownite and anorthite, and reasonsfor the disag"reement are unclear. Even with this uncertainty, however, it is apparent that dissolution rates vary nonlinearly with composition and that the rates for the Ca-rich minerals vary more with composition than Na-rich minerals.

INrnonucrroN last, 1984, 1985;Holdren and Speyer,1985; Casey et al., 1988,r989). It has long beenobserved that mineral weatheringrates The disparity in reactivity reflects differencesin the vary considerably with the mineral structure and com- bonding characterof Al and Si to O atoms at the mineral position. Goldich (1938), for example, noted that silicate surface. Furthermore, becausehydrolysis reactions are materials that dissolve slowly tend to contain a highly catalyzedby H and hydroxyl ions (Wendt, 1973), The polymerized (e.g., tectosilicate) structure (seealso Hurd disparity in reactivity reflectsdifferences in the acid-base et al., 1979; Lin and Clemency, 198l). These materials properties of the exposedaluminate and silicate sites.Al also tend to be selectively leachedby aqueoussolutions, is rapidly leachedfrom the surfaceoffeldspar in acid but as there are sufficient unreactive bonds near the mineral is not detectableat near-neutraland basic pH conditions surface to maintain integrity once the reactive constitu- (Caseyet al., 1988, 1989). The leaching reaction creates ents are removed (Murata, 1943). Under certain condi- an Al-, Na-, and Ca-depletedlayer on the surfaceof some tions, thesematerials form a thick leachedlayer adjacent plagioclaseminerals in the early stagesof a dissolution to the aqueoussolution. Conversely,silicates with a poor- experiment. ly connectedfabric, such as olivine, tend to dissolve rap- This selectivehydrolysis not only affectsthe chemistry idly and uniformly (Schott and Berner, 1985; Petit et al., of the mineral surfacebut should also be manifested in 1989) as long as the adjacent solution is undersaturated the bulk dissolution ratesofminerals, suchas plagioclase, with respectto secondaryphases. which have a wide range in Si and Al concentrations. Both the leaching behavior and bulk dissolution rate Plagioclase ranges in composition from albite (NaAl- of minerals are controlled by the relative concentration SirOr) to anorthite (CaAl$irO8). This variation corre- of reactive and unreactive sites at the mineral surface. sponds to a coupled, isostructural substitution of alumi- Aluminate (AlO;5) sites on an aluminosilicate mineral nate (AlO;') groups andCa2'for silicate (SiO;a) groups surface tend to be more rapidly hydrolyzed in a strong and Na". acid than the silicate (SiO;') sites (Murata, 1943). Acid The actual variation in plagioclasedissolution rate at treatment, for example,is usedto dealuminate zeolitesin a constant pH is poorly known, as much previous work order to make hydrocarbon-crackingcatalysts (e.g., Bar- was limited to a small range of mineral compositions. rer, 1978; Man et al., 1990). Similar dealumination of Even in more comprehensive studies, however, impor- clay has been proposed as a nonbauxitic source of AI tant variables such as the mineral surface area and purity (Aglietta et al., 1988).Acid dealumination reactionsalso were not determined, or the solution composition was affectthesurfaceofdissolvingfeldspars(ChouandWol- allowed to vary widely during the experiment. These

0003{04xl91/0102-021l$02.00 2rr 2t2 CASEY ET AL.: DISSOLUTION RATES OF PLAGIOCLASE

TABLE1. Plagioclasecompositions by Holdren and Speyer(1987). Once prepared,the pow- ders were treated identically. All of the feldsparsexcept the Grass Valley sio, 66.83 62.89 63.74 54.71 52.34 49.45 44.97 anorthite Tio, 0.00 0.00 0.01 0.05 0.07 0.03 0.01 were acquired as large crystals. These crystals were ground Alros 19.26 21.95 23.02 27.93 29.17 32.13 34.53 and wet-sievedto isolatethe 25-7 5 rrm sizefraction. This Fe2O3 0.04 0.05 0.09 0.28 0.31 0.40 0.31 powder FeO 0.04 0.06 0.09 0.25 0.29 0.36 o.28 was repeatedlywashed in distilled HrO and spec- Mgo 0.01 0.01 0.04 0.23 0.14 0.14 0.02 trophotometric-grade acetonein order to remove ultra- MnO 0.01 0.01 0.01 0.02 0.02 0.00 0.01 fine particles that cling to the larger grains. Washing re- CaO 0.16 4.15 4.47 9.45 12.31 15.49 18.52 Nio 0.12 0.05 0.08 0.10 0.00 0.24 0.10 moves most, but not all, ultrafine particles,as determined Naro 11.51 9.05 7.93 5.23 4.30 2.73 0.89 by scanningelectron microscopy. t&o 0.11 0.35 1.43 0.93 0.30 0.12 0.o2 The anorthite was separatedfrom an anorthosite from Note.' Separate grains within each powder were analyzed, averaged, Grass Valley, California. This material was crushed in a and normalized to 100%. Molar Fe concentrations are divided equally steelmortar and pestle,followed by sieving to isolate the between terric and ferrous oxidation states. The average stoichiometry and the total number ol analyses are includedas footnotes. In all cases, 25-75 prn size fraction. This powder was washedrepeat- the analytical precision analysis is less than the compositionalvariation edly in distilled HrO and acetoneto remove ultrafine par- within a singlemineral. 1 : Keystonealbite: (NaoeTGoloao o2Alossi3 @Os oo), 337 analyses.2 : Mitchellcounty oligoclase:(Nao,ulG*CaoroAl,,ESirsO6@), ticles. The resulting powder consistedof a mixture of sep- 317 analyses.3 : Bancroft albite: (Nao6slc osoao.,Al, roSi. roO"*),266 anal- arate grains of mafic minerals and feldspar. These minerals yses.4 : Saranac Lake andesine:(Nao 16Ko6Oao6Alr $Sir soos@),294anal- were sepaxatedwith a Franz Isodynamic Separator,which yses.5 : PuebloPark bytownite:(NaosKo@Cao60A11$Sireosoo),48 anal- yses. 6 : Crystal Bay bytownite: (Naor4GolcaoT6AllT4sirreoooo),314 produced a nearly pure feldspar powder. Final purifica- analyses.7 : Grass Valleyanorthite: (NaoosG@Cao@Alreosi2@06@), 46 tion was accomplishedusing high-density liquids. Anor- analyses. thite grains were then washedwith distilled HrO and ac- etone. variations can lead to precipitation of secondaryminerals Feldspar compositionsand sample purity onto the dissolving surface(see review by Helgesonet al., The feldsparcompositions were measuredwith an elec- 1984) or to important changesin mineral surfacechem- tron microprobe and averagecompositions are reported istry during the experiment. in Table l The total range of measuredcompositions is The goal paper of this is to define a relation between shown on ternary diagrams in Figure l. The diameter of the dissolution rates and composition of plagioclasefeld- the electron beam used to analyze these sampleswas 16 spars.To achieve this goal it is necessaryto supplement rlm. the existing data with additional experimentsso that rates Each feldspar exhibits some variation in composition are available plagioclase for a wide range of composi- as measuredwith the electron probe, and the traditional tions. The investigation is restricted to dissolution in name of some samples can be misleading. Specifically, strongly acidic solutions where the character of the re- the Pueblo Park bytownite and the Bancroft albite are action is relatively well understood (e.g.,Chou and Wol- labradorite and oligoclase, respectively. Much of this last, 1985;Holdren and Speyer,1985; Casey et al., 1988, variation in composition is attributable to microscopic 1989) precipitation and the of secondaryminerals can be mineralogic heterogeneities.Zoning, exsolution lamellae, easily suppressed. and accessoryphases are apparent from microscopic ex- amination of material (see Marnnllrs AND TEcHNTeuES the Appendix). Although all but one of the compositions are consistent with the per- Preparation of samples missible range of solid solution (Fig. l), most of these All samplesexcept the Pueblo Park bytownite were ac- materials contain more than a single feldspar phase(Ap- quired from Wards Natural Science Establishment, pendix). The Mitchell County andesine, for example, Rochester,New York. The samplesand their trace min- contains exsolved potassium feldspar (Appendix), which eralogy are described in the Appendix l.' Two samples introducesconsiderable scatter in the ternary plot of com- of the Mitchell County oligoclaseand SaranacLake an- position (Fig. l). desine were prepared independently. Granular samples were independentlyprepared from crystalsin 1983-1984 Specific surface areas (seeHoldren and Speyer, 1987) and as part ofthis study Specificsurface area of each sample was measuredpri- in 1988. The procedures used by each researcherwere or to dissolution using a Micromeritics Digisorb 2600 similar, except that a ball mill was used to crush the ma- device. After degassing overnight at 200 oC, varying terial in this study, rather than the disk grinder employed mixtures of N and He were adsorbed onto the powder surface.The areaswere calculatedfrom the sorption iso- therm and the BET theory of area measurement (Bru- ' A copy of Appendix I may be ordered as Document AM- nauer 1938). The precision of this 9l-449 from the BusinessOfrce, Mineralogical Society of Amer- et al., and accuracy ica, 1130 SeventeenthStreet NW, Suite 330, Washington, DC technique is generally better than 50/0,as determined by 20036, U.S.A. Pleaseremit $5.00 in advancefor the microfiche. repeated measurement of standards and one sample CASEYET AL.: DISSOLUTIONRATES OF PLAGIOCLASE 213

An Fig. 1. Variation in Gldspar composition for each single powder as a function of the end-member compositions: albite (Ab), anorthite (An), and orthoclase (Or). Analyses for the Keystone albite (unfilled triangle) and Saranac Lake andesine (filled circle) are shown in (A). Analyses for the Bancroft albite (filled circle) and the Crystal Bay bytownite (unfilled triangle) are shown in (B). Analyses for the Mitchell County oligoclase (fllled circle), the Pueblo Park bytownite (unfilled triangle) and the Great Valley anorthite (unfilled square)are shown in (C). The dashedline representsthe range offeldspar solutions (Smith, 1974).

(Pueblo Park bytownite). Measured areasand uncertain- ter than 5ol0,as determined by repeatedanalysis of samples ties are listed in Table 2. Becauseonly a small amount and standards. of Crystal Bay bytownite was separated,the area was de- Dissolution rates were calculated by monitoring the termined by using mixtures of Kr and He rather than concentrationsof Al, Si, Ca, and Na in the solutions (Fig. N-He mixtures. The precision of areameasurements with 2). Rates were calculated by linearly regressingthe con- Kr is comparable to those made with N. The accuracyis centrations against time. Dissolution rates were calculat- only within about 20010,as indicated by repeated mea- ed from the slope of these curves, and from knowledge surement of standards. of the specificsurface area and composition of each min- The exposedarea of feldspar probably did not change eral. The rates are compiled in Table 3. Data used to significantly during the experiment and only a small frac- calculate the rates are given in Appendix l. tion of the powder mass(typically =0.1 mass9o)was con- sumed. The feldspar areas could not be measured after Rnsur,rs the experimentsbecause of the artifacts of sample leach- The accuracyof any relation between mineral compo- ing and drying discussedby Caseyet al. (1990). Specifi- sition and dissolution rate is limited by the abundance cally, the leached surfacespartly separatefrom reacted and quality of data. The set of plagioclasedissolution plagioclaseduring drying, and this separation exposesnew rates is illustrated in Figure 3, where results from this surface area. study are comparedwith data from published sources.As one can see,the dissolution rates vary dramatically with Dissolution experiments mineral composition.Rates increasefrom =l x l0-t5 Batch dissolution experimentswere conducted by dis- mol/cm2ls for albite to well over I x 10-12moycm2/s for persing each feldspar in a polyethylenebottle containing anorthite. All of these data are from experiments con- 0.010 N HCI (see Appendix). Solution volumes and ducted in HCI solutions, except possibly for the datum massesof mineral powder are compiled in the Appendix. The solutions were continuously agitated at 25 .C. Sam- ples of the solution were periodically extracted from the TlsLe 2. Specificsurface areas of the feldsparpowders bottles, pressedthrough a0.2 pm filter, and analyzed.The Sample Area m'zlg samplevolumes varied between l0 and 25 ml. After sam- Bancroft albite 0.405+ 0.03 pling, an equal volume ofunreacted background electro- Mitchell County oligoclaseno. 1 0.18+ 0.01 lyte was immediately added to the bottles to maintain a Mitchell County oligoclaseno. 2 0.645+ 0.05 + constant total volume. All data are correctedfor changes Grass Valley anorthite 0.52 0.02 Keystone albite 0.49+ 0.05 in concentration as a result of sampling. There was neg- Pueblo Park bytownite 0.20+ 0.01 ligible drift in pH during the experiments. Crystal Bay bytownite 0.27+ O.O1 + 0.03 Si concentrationswere measuredusing the Heteropoly Saranac Lake andesineno. 1 0.405 Saranac Lake andesineno. 2 0.35+ 0.03 blue technique (American Public Health Association, (one 1970).Al concentrationwas measuredwith a DC-plasma Note.'The powders range in size from 25 to 75 pm. Uncertainties estimatedstandard deviation)are reported from the BET r€ression (Low- spectrophotometer.Na and Ca concentrations were de- ell, 1979) or the average of repeated measurementsof the Pueblo Park termined by atomic absorption spectrophotometry.Pre- labradorite,whichever is larger. The Mitchell County oligoclaseno. 2 and no. 2 were ground in 1983-1984. cision and accuracyofthese techniquesare generallybet- the Saranac Lake andesine 214 CASEY ET AL.: DISSOLUTION RATES OF PLAGIOCLASE

-8 TABLE3. Plagioclasedissolution rates at pH : 2.0 X10 1.5. Silicon Rate (teldspar-mol/cmr/s)x 1015 6 v Sample Si Ca Feldspar- moles 1 1.08+ 0.6 0.47 + 0.01 5 cm2 2 1.41! 0.2 2.57 + O.79 v. 3 3.62 + O.2 4.98 + 0.14 4 2.O5+ 1.1 2.51 + 0.87 2.53 + O.27 a 5 6.85 + 3.0 15.0a 7.5 8.46 + 0.51 8.32 + 1.20 .3 4 6 9.88 + 4.3 26.8 ! 5.7 15.8+ 0.62 12.9+ 1.63 7 14.0+ 3.3 49.2 + 4.1 42.9 + 2.08 39.5 + 3.35 o o o e 66 8 212.0+ 95.0 359.0 + 174.0 425.0 + 24.9 419.0 + 28.8 1.O 1.5 io" to6 9 643.0 + 40.0 600.0 + 48.0 565.0 + 48.0 573.0 + 65.1 Seconds Note.' Uncertaintiesfor Si and Al fluxes are estimated from duplicate -8 experiments.Uncertainties for Ca and Na fluxes are from the enor of the x1O linear regression.All uncertaintiescorrespond to one estimated standard deviation. 1 : Keystone albite, 2 : Mitchell County oligoclaseno. 1, 3 : Aluminum 6 MitcheffCounty oligoclase no.2,4 : Bancroft albite, 5 : Saranac Lake " 3.O andesineno.1,6: SaranacLake andesineno.2,7: PuebloPark bytownite, 8 : Crystal Bay bytonite, I : Great Valley anorthite. Feldspar - moleS 5, cm'2.o ve . -d

,v ' 1.O -vt --lat dissolution rates ofbytownite and anorthite. The rate of v" -rf _3 dissolution of the Grass Valley anorthite from this study, o+ for example, is approximately 75 times smaller than that o do"ro6 reported for the Pacayaanorthite by Fleer (1982) (Fig. 3). Seconds Brady and Walther's (1989) rate for anorthite dissolution is also considerablylower than Fleer's (1982) datum, but Fig. 2. (a) Concentrationof Si in solutionas a functionof is also higher than the rate measuredin this study for the time. The Si concentrationsare normalized to the initial powder Grass Valley anorthite. Brady and Walther (1989) dis- surfacearea, the total solutionvolume, and the compositionof solved the unnamed anorthite (Annr) at pH : 1.9. The the feldspar.The resultingconcentrations are plotted in unitsof pH feldspar-mol/cmr.Expressed in this way,the slopeof the curve slight differencesamong the experiments cannot ac- equalsthe feldspardissolution rate in mol/cm2ls.The numbers count for such large differencesin reported rates. correspondto thesamples identified in Table3. (b) Theconcen- The discrepancy, however, may be attributable to trationsof Al in solutionas a functionof time. unique aspectsof Fleer's experiments. Most of the data compiled in Figure 3 are from experimentsconducted on granular material. The datum of Fleer (1982) is from a from Brady and Walther (1989). A complete description batch experiment conducted on a single crystal of anor- of this particular experiment by the original authors is in thite. She estimated the exposedsurface area of this crys- preparation. tal geometrically,rather than through the sorption ofgas- In severalexperiments either separateminerals of sim- es onto exposed surfaces, as was done by the other ilar composition were dissolved (e.g., the Keystone and researchers.If this crystal contained many small frac- Amelia albites), or separateexperiments were conducted tures, which were accessibleto the aqueoussolution but on minerals from the samelocation (e.g.,the Crystal Bay too small to be incorporated into the areacalculation, her bytownite). In general,the dissolution rates ofthe plagio- estimate of the dissolution rate would be anomalously clase minerals in the approximate composition range of high. The total disagreementover anorthite dissolution Anno to Anuoagree to within a factor of between two and rates decreasesfrom over a factor of 100 to approxi- five. Estimates for albite, for example, range from 1.08 mately a factor of four if Fleer's (1982) datum is exclud- x l0 '5 to 2.5 x l0-r5 feldspar-moycm2/s.The disso- ed. lution rate of the Keystone albite is lower than that re- The dissolution rate of the Crystal Bay bytownite re- ported for Amelia albite at pH : 2.1 by Chou and Wol- ported by Brady and Walther (1989) is lower by almost last (1984, 1985)but compareswell with that published a factor of ten than is reported here for the samematerial by Knauss and Wolery (1986) for an unidentified albite. at pH : 2.0 (Fig. 3). Reasons for this discrepancy are To make this comparison, data of Knauss and Wolery unclear. Although Brady and Walther (1989) conducted 'C (1986)were adjustedfrom 70 "Cto 25 usingthe stan- their experiments at sli8htly more basic conditions (pH -- dard Arrhenius relation and the activation energy(8 kcal/ 2.15) than those employed here, the large discrepancy mol) measuredby Fleer (1982) for anorthite dissolution in rates cannot be attributed solely to differencesin so- at pH 2.0. Note also that the batch dissolution rate of lution pH. albite reported here (Table 3) compareswell with exper- iments conductedin chemostatcells (Knaussand Wolery, DrscussroN 1986;Chou and Wollast, 1985). There are two important points to summarize from the There is considerable disagreement,however, about the precedingsection. First, rates vary with plagioclasecom- CASEYET AL.: DISSOLUTIONRATES OF PLAGIOCLASE 215

-10 -11

-11 -12 log {Rate} tos {Rate} -12 -13

-13 -14

6 -14 4 + -15 F---a+ rL .5 o IC -15 -16 o 20 40 60 80 100 40 60 80

Anorthite{mole percent} Anorthite{ mole percent} Fig. 3. Specifrcdissolution rates of plagioclasesamples as a Fig-4. Dissolutionrates of plagioclasesamples as a function functionof compositionin pH : 2.0 solutions.The numbers of mole percentanorthite in pH : 3.0 solutions.Sample C is correspondto tho samplesidentified in Table 3. Ratesare re- from Chouand Wollast(1984), sample H is from Holdrenand portedas feldspar-moVcm2/s asdetermined by Si fluxes.Sample Speyer(1985, 1987), and sample A is from Amrheinand Suarez C correspondsto the Amelia albite(Chou and Wollast,1985). (I 988). SampleF correspondsto thePacaya anorthite (Fleer, unpublished data). SampleK is an unidentifiedalbite describedby Knauss and Wolery(1986). Samples identified with a B correspondto trend of the logarithms of rate vs. composition is not an unidentified anorthite of compositionAno, and the Crystal constant, but increasesin slope with increasedCa and Al Bay bytownite (Patrick Brady, personalcommunication). The content (Fig. 3). Although the data are scattered,the rates curve is an empirical correlationbetween dissolution rate and apparently increasemore rapidly for the calcic minerals feldsparcomposition. than they do for minerals in the composition range Anoo to Anro-ro.This relation is apparent even if Fleer's (1982) datum is excluded from the set of data. To verify this position by a factor of well over 103.The large difference relation, additional dissolution rates are compiled from in dissolution rate of very sodic and very calcic plagio- experimentsat pH : 3 and are shown in Figure 4. clase has been pointed out before (Lasaga, 1983; Brady As in the casefor experimentsat pH : 2, ralesincrease and Walther, 1989).The new data presentedhere for pla- with anorthite content. The total variation in rates with gioclaseminerals of intermediate composition, however, mineral composition is, however, not as pronounced as add much detail to the relation (Fie. 3). in the more acidic solution (Fig. 3). The relation between Second,this wide variation in rate is much larger than the logarithm of dissolution rate and mineral composi- the uncertainty surrounding any singlecomposition. Dis- tion is nonlinear, however,just as was observedat pH : solution rates of a single plagioclasecomposition from 2 (Fig.3). The dissolution rates of plagioclasein the ap- different investigators generally agree to within a factor proximate composition range Anoo to An,oo vary more of approximately 2 to 5 for minerals in the approximate dramatically with composition than plagioclasein the ap- composition range An00to Anuo.The disagreementis much proximate composition range Anooto Anoo.The relation larger for more Ca-rich plagioclase,and additional data between plagioclasedissolution rate and mineral com- are critically neededon these compositions. position is, ofcourse, heavily influenced by the few data An empirical correlation of dissolution rate with min- for calcic compositions. The following empirical corre- eral composition in pH : 2 solutions is given by: lation describesthe distribution ofrates: log,o[R(X)]: -14.7484 - 0.0135083x X log,o[R(X)]: -15.023 - 0.001131x X + 0.000505238x lG. (l) + 0.0003176x l3 (2\ In Equation l, X representsthe mole percent anorthite where all variables are as previously defined. AII of the in the plagioclaseand R (X) is the dissolution rate in feld- exp€riments used to compile Figure 4 were conducted in spar-mol/cm'zls(Fig. 3). All data shown in Figure 3 are HCI solutions. included in this regression,including Fleer's (1982) esti- It is not surprising that the rates ofdissolution ofpla- mate of the anorthite dissolution rate. No statistical sig- gioclasevary so dramatically and nonlinearly with com- nificance is assignedto this nonlinear regression,and the position. The structure of albite contains an extensive correlation is heavily weighted by the datum of Fleer framework of linked silicate tetrahedra which are resis- (1982). The regressionis shown as a bold, nearly contin- tant to hydrolysis relative to the aluminate sites. Con- uous line in Figure 3. versely, silicate tetrahedra in ordered anorthite are iso- An important feature of the data is that dissolution lated from one another by aluminate groups. In contrast rates of calcic plagioclaseare apparently more sensitive to albite, selective hydrolysis of the Al from anorthite to variations in composition than sodic materials. The completely disrupts the structure. 216 CASEY ET AL.: DISSOLUTION RATES OF PLAGIOCLASE

The relationship between dissolution rate and mineral nessofthe leachedlayer is controlled by the solution and composition is likely to vary with the solution composi- mineral compositions. In any case, the overall rate of tion. As noted by Wendt (1973), the pH-dependenceof dissolution is controlled by the rate of destruction of the hydrolysis rates in simple solutions is controlled by the silicate framework, and these kinetics can be described acid-baseproperties of the important bonds (seealso Bales with an approximately zero-order rate law. Thus rates of and Morgan, 1985; Carroll-Webb and Walther, 1988; dissolution become nearly constant with time after the Blum and Lasaga, 1988). Other sorbed ligands can also early stagesof an experiment. promote hydrolysis rates(Amrhein and Suarez,1988) and Mineralogic heterogeneities,however, influence con- the relationships shown in Figures 3 and 4 are not ap- siderably the measured rate of dissolution. Ca-rich pla- propriate for solutions containing ligands, such as fluo- gioclasedissolves many times more rapidly than either ride ion or certain organic acids, which changethe char- Na- or K-rich feldspar (compare the rates compiled in acter of hydrolysis from proton-promoted dissolution. Figure 3 with that reported for potassium feldspar by Sorption ofthese ions onto a hydrolyzing bond enhances Holdren and Speyer,1985). Most plagioclasesamples used the rate and may also modify the relation between dis- in this study contain fine-grained intergrowths of two solution rate and mineral composition. Of the data com- feldspar minerals, or are zoned in composition. The Ban- piled in Figure 3, only Knauss and Wolery (1986) em- croft albite, for example, is antiperthitic (Appendix). ployed a potentially catalytic solute (biphthalate) in their Likewise, the Mitchell County oligoclase contains an experiment. The close agreement between Knauss and exsolved K-rich feldspar and exhibits a wide range in Wolery's rate estimate and those of other investigators composition (Fie. l). The net dissolution rate is different suggeststhat the biphthalate did not affect their result. than would be measuredon homogeneousmaterial with Many potential factors may accountfor the uncertainty the same nominal composition. of a factor of approximately two to five for minerals in Regular variations in the dissolution rate with time can the composition rangeAnoo to Anuo.Any singlebatch rate also lead to uncertainties in the rate estimate. Rates de- measurement,for example,is only reproducibleto within creasewith time in the early stagesof an experiment by this factor (Table 3). Furthermore, rates estimated from as much as a factor of ten as the particularly reactive sites batch experiments ofshort duration difer depending upon dissolve (e.g.,Holdren and Berner, 19791,Holdren and which element is examined. Rates calculatedfrom the Al Speyer,1985; Knauss and Wolery, 1986).Such a period flux, for example, are larger than rates calculated from of enhancedreactivity is apparent in Figure 2 as anom- the Si flux (Table 3). alously rapid increasesin Si concentration in the first 0.5 The uncertainty that arisesfrom this nonstoichiometric x 106s ofeach experiment. Subsequentchanges in con- dissolution is too small to influence dramatically the re- centration are approximately constant (Fig. 2), and it is lation between rate and composition shown in Figure 3. these data that are used to calculate the dissolution rate. In fact, both Knauss and Wolery (1986) and Chou and Holdren and Speyer(1985), however, also show that the Wollast (1985) establishedstoichiometric dissolution in rates decreaseslowly with time over many hundreds of their chemostat experimentsand their rates differ signif- hours of reaction. All of the data compiled in Figure 3 icantly. The discrepancymust be attributed either to dif- are from experiments that lasted severalhundred hours. ferencesin the experimental design,or to inherent differ- Subtletiesof sample preparation can also causerates to encesin the reactivities of the two albites.Note, however, vary by factors ofbetween two and five. Egglestonet al. that the results of Knauss and Wolery (1986) compare (1989), for example,found that diopside dissolution rates well with the rate estimate for the Keystone albite. Stoi- decreasedfor several months after the mineral powder chiometric dissolution was not achievedin this batch ex- was initially prepared. They attributed the decreasein penment. rate with time to slow recrystallization of the damaged The nonstoichiometric dissolution arises becauseAl, mineral surfacesto a relatively undamagedstate. A sim- Na, and Ca dissolve more rapidly from the plagioclase ilar mechanism may explain the duplicate experiments structure than Si in the early stagesof an experiment. The on separatelyprepared samples ofSaranac Lake andesine differencein rate disappearswith time as a leachedlayer and the Mitchell County oligoclase (Table 3). The dis- of constant thicknessdevelops on the mineral surface.At solution rates ofthe duplicatesvary by factors ofapprox- this stage, the difiirsive flux of Al through the leached imately two in each case.One clear distinction between layer is constant and proportional to the rate of destruc- thesematerials is the time that elapsedsince preparation. tion of the residual silicate framework (Chou and Wol- The slowly dissolving material was prepared 4-5 yr ear- last, 1984). lier than the more rapidly dissolving material. There have been no studies to date documenting the The essentialpoint is, however, that the variation in time required to reach a leachedlayer of constant thick- rate for any single plagioclasesample is small relative to ness for plagioclase minerals at pH : 2. The ratio of the entire variation in rate with composition. Many po- dissolved substancesreleased from the mineral cannot be tential factors can causerates to vary by a factor of be- trusted, since these fluxes may reflect the dissolution of tween two and five besidesthose listed above. Thus the trace impurities and exsolvedphases. The ultimate thick- relation between the plagioclase dissolution rates and CASEY ET AL.: DISSOLUTION RATES OF PLAGIOCLASE 217 mineral composition derived above is meaningful. This Fleer, V.N. (1982) The dissolution kinetics of anorthite and synthetic relation will be improved, of course, as additional data strontium feldspar in aqueous solutions at temperatures below 100 eC: are acquired. With applications to the goologicaldisposal ofradioactive nuclear wasles, 197 p. Unpublished Ph.D disserlation, The Pennsylvania State Uni- AcxNowr-nrcMENTs versity, University Park, Pennsylvania. Goldich, S.S.(1938) A study in rock-weathering.Joumal ofGeology, 46, This paper was considerably improved through reviews provided by r 7-58. Susan Brantley, Phillip Bennett, and an anonyrnous referee. The authors Helgeson,H.C., Murphy, W.M., and Aagaard,P. (1984)Thermodynamic thank Pat Brady for sharing his unpublished data on plagioclase dissolu- and kinetic constraints on reaction rates among minerals and aqueous tion, Fran Nirnick for providing samples for this study, and C-arol S solutions II: Rate const4nts, effective surface area and tlte hydrolysis Ashley for measuring the surface areas. This research was supported by of feldspar. Geochimica et Cosmochimica Acta, 48, 2405-2432. the Department of Energy under contract DE-AC04-76DP00789. Holdren, G.R., and Bemer, R.A. (1979) Mechanism of feldspar weath- ering l: Experimental studies. Geochimica et Cosmochimica Acta, 43, RrrnnnNcns crrED l 161-l171. Holdren, G.R., and Speyer,P.M. (1985) pH-dependent changesin the Aglietta, E.F., Porto-Lopez, J.M., and Pereira,E. (1988) Structural alter- rates and stoichiometry of dissolution of an alkali feldspar at room ations in kaolinite by acid treatment. Applied Clay Science, 3, 155- temperature. American Journal of Science, 285, 994-1026. 163. -(1987) Reaction rate-surface area reliationships during the early American Public Health Association (1970) Standard methods for ex- stagesofweathering-Il Data on eightadditionalfeldspan. Geochimica amination of wastewater(l3th edition), New York. et CosrnochimicaActa, 51. 23ll-2318. Amrhein, C., and Suarez,D.L. (1988) The use ofa surfacecornplexation Hurd, D.C., Fraley,C., and Fugate,J.K. (1979)Silica apparentsolubilities model to describe the kinetics of ligand-promoted dissolution of an- and rates ofdissolution and precipitation. In E.A. Jenne, Ed., Chemical orthite. Geochimica et CosmochimicaActa, 52,2785-2793. modeling in aqueous systems. American Chemical Society Symposium Bales,R.C., and Morgan, J.J. (1985) Dissolution kinetics of chrysotile at Series93, Washington, DC. : pH 7 to 10. Geochimica et C.osmochimicaAct,,49,2281-2288. Knauss, I(G., and Wolery, T.J. (1986) Dependenceof albite dissolution Barrer, R.M. (1978) Zeolitesand clay minerals as sorbentsand molecular kinetics on pH and time at 25 f and 70 'C. Geochimica et Cosmo- sieves.Academic Press,New York. chimica Acta, 50, 2481-2497. Blum, A., and Lasaga,A.C. (1988) Role of surfacespeciation in the low- Iasaga, A.C. (1983) Kinetics ofsilicate dissolution. Fourth international temperaturedissolution of minerals. Nature, 33 l, 4ll-433. symposium on water-rock interaction, p. 269-274. Misasa,Japan. Brady, P.V., and Walther, J.V. (1989) Controls on silicate dissolution Lin, F.-C., and Clemency,C.V. (1981) The dissolution kinetics of brucite, rates in neutral and basic pH solutions at 25 f. Geochimica et Cos- antigorite, talc, atrd phlogopite at room temperatue and pressure. mochimica Aaa - 53. 2823-2830. American Mineralogist, 66, 801-806. Brunauer, S., Emmet, P.H., and Teller, E. (1938) Adsorption of gasesin I-owell, S. (1979) Introduction to powder surface area, 199 p. Wiley- multimolecular layers. Joumal of the American Chemical Society, 60, Interscience, New York. 309-3 I 9. Man, P.P., Peltre, M.J., and Barthomeuf, D. (1990) Nuclear magnetic Carroll-Webb, S.A., and Walther, J.V. (1988) A surfacecomplex reaction resonance study ofthe de-alumination ofan amorphous silica-alumina model for the pH-dependence of corundum and kaolinite dissolution catalyst.Journal ofChemical SocietyFaraday Transactions86, 1599- rates. Geochimica et Cosmochimica Actz, 52, 2609-2623. 1602. Casey,W.H., Westrich, H.R., and Arnold, G.W. (1988) Surfaceof lab- Murata, K.J. (1943) Internal structure of silicate minerals that gelatinize radorite feldspar reacted with aqueous solutions at pH : 2,3, and 12. with acid. American Mineralogist, 28, 545-562. Geochimica et Cosmochimica Acta, 52, 2795-2807 . Petit, J.-C., Dran, J.-C., Paccagnella,A., and Della Mea, G. (1989) Struc- Casey, W.H., Westrich, H.R., Arnold, G.W., and Banfield, J.F. (1989) tural dependence of crystalline silicate hydration during aqueous dis- The surface chemistry of dissolving labradorite feldspar. Geochirnica solution. Earth and PlanetaryScience ktters, 9t,292-298. et Cosmochimica Acta. 53. 82I-832. Schott, J., and Berner, J.A. (1985) Dissolution mechanismsofpyroxenes Casey,W.H., Westrich, H.R., Massis,T., Banfield,J.F., and Arnold, G.W. and olivines during weathering. In J.I. Drever, Ed., The chemistry of (1990) The surface oflabradorite feldspar after acid hydrolysis. Chem- weathering,p. 35-53. Reidel, Boston. ical Geology, 78, 205-218. Smith, J.V. (1974) Feldspar minerals- l: Crystal structure and physical Chou, L., and Wollast, R. (1984) Study of the weathering of albite at properties, 627 p. Springer-Verlag, Heidelberg. room temperature and pressurewith a fluidized bed reactor. Geochimi- Wendt, W. (1973) Die kinetik typischer Hydrolysereaktionen von rnehr- ca et CosmochimicaActa, 48,2205-2217. wertigen Kationen. Chimica, 27, 57 5-588. - (1985) Steady-statekinetics and dissolution mechanismof albite. American Joumal of Science, 285, 963-993. Eggleston,C.M., Hochella, M.F., and Parks, G.A. (1989) Sample prepa- ration and aging effects on the dissolution rate and surface composition MeNuscnrrr RrcErvEDApnrl 12, 1990 of diopside. Geochimica et Cosmochimica Acta, 53, 497 -8O4. Mem;scnrvr AccEprEDNowvrpn 15, 1990