Hydroxyl-Sulfate Exchange Stoichiometry on y-AbOa and KaoliniteL. M. He, L. W. Zelazny,* V. C. Baligar, K. D. Ritchey, and D. C. Martens

ABSTRACTThere is a release of OH~ from soil colloids with the addition of

SOj ~ . The ratio of moles of OH ~ released per mole of SOl' adsorbedis referred to as the OH/SO., exchange Stoichiometry. The OH/SO4Stoichiometry not only provides an important constraint on surfacecomplexation models but also is a critical area of research to explainthe effects on soil pH from the application of SOj" amendments suchas gypsum. The OH/SO4 Stoichiometry on y-AljOs and kaolinite wasdetermined by a back titration method with a pH-stat autotitrator aswell as calculated from SOl~ adsorption edges at various SO*~ concen-trations by the thermodynamic approach. The Stoichiometry increasedwith increasing solution pH and SOa~ adsorption density. At a lowpH, the Stoichiometry was low and was similar for y-AhOj and kaolin-ite. At higher pH levels, it was higher for y-AI2O., than for kaolinite. Theaverage values of the Stoichiometry calculated by the thermodynamicapproach were in agreement with those measured by back titrationfor both y-AhOj and kaolinite. The thermodynamic approach providesan easier experimental method to evaluate the OH/anion Stoichiometryas a function of pH and adsorption density. It also indicated that atlow pH levels (<5), the Stoichiometry might be overestimated due todissolution of adsorbents, whereas at higher pH levels (5-8), it mightbe underestimated because of precipitation of Al hydroxy sulfates andgibbsite.

SOIL ACIDITY is a major limiting factor in crop produc-tivity in many parts of the world (Baligar et al.,

1992; Sumner, 1993). The effect of soil acidity in topsoilscan be overcome by liming. This approach is much moredifficult in subsoils because lime does not readily movedown the profile and mechanical mixing is not usuallyfeasible (Shainberg et al., 1989; Sumner, 1990; Sumneret al., 1986). One of the possible approaches that canbe used to decrease subsoil acidity is incorporation ofgypsum into surface soil (Sumner, 1993); due to itsrelatively high solubility, gypsum readily leaches belowthe surface soil layer.

A commonly reported effect of gypsum application onsoils is the change in soil pH. Many indicate that soilpH increased with time after gypsum addition underfield and laboratory conditions (Curtin and Syers, 1990;Farina and Channon, 1988; Hue et al., 1985; Marcano-Martinez and McBride, 1989; Pavanetal., 1982; Ritchey

L.M. He, L.W. Zelazny, and D.C. Martens, Department of Crop andSoil Environmental Sciences, Virginia Polytechnic Inst. and State Univ.,Blacksburg, VA 24061; and V.C. Baligar and K.D. Ritchey, AppalachianSoil and Water Conservation Research Lab., USDA-ARS, Beckley, WV25802. Received 27 June 1994. *Corresponding author (zelazny@vt.edu).

Published in Soil Sci. Soc. Am. i. 60:442-452 (1996).

et al., 1980; Shainberg et al., 1989; Sumner, 1993). Incolumn leaching studies, however, gypsum applicationhad little effect on soil solution pH and even slightlydecreased soil pH (Gates and Caldwell, 1985; Pavan etal., 1984). Increases in soil pH due to gypsum additiondepend upon soil properties and range from 0.1 to 0.9pH units (Curtin and Syers, 1990; Shainberg et al.,1989). The range of pH change is apparently controlledby soil clay mineralogy, the initial pH of soils treated,the amount of SO?" addition, and the source of addedSOl". Since the amount of SO?" adsorption is propor-tional to the amount of Fe or Al oxides in soils (Chaoet al., 1964), a greater increase in soil pH is expectedfrom soils that have a larger quantity of sesquioxides.The largest increase in pH usually coincided with thezone of maximum SO?" adsorption within a soil profile(Shainberg et al., 1989). Curtin and Syers (1990) mea-sured an increase of 0.9 pH units for a soil with aninitial pH of 5.7 and only 0.4 pH units for another soilwith an initial pH of 3.8. They also found that soil pHincreased with an increase in SO?" addition. Marcano-Martinez and McBride (1989) observed a greater increasein soil pH following K2SO4 than CaSO4 additions.

An increase in suspension pH due to SO?" adsorptionis essentially accompanied by release of OH~ from solidsurfaces or consumption of H+ in solution. It is possibleto determine the ratio of moles of OH" released permole of SO?" adsorbed, which is referred to as the OH/SO4 exchange Stoichiometry and designated by %. Twomethods are commonly used to determine the OH/SO4Stoichiometry. One is a direct experimental method inwhich the suspension is back titrated to its original pHwith simultaneous determination of the amount ofSO?" adsorption. The other is an indirect method thatincludes the thermodynamic approach (Perona andLeckie, 1985; Honeyman and Leckie, 1986) and theKurbatov method (Kurbatov et al., 1951; Harvey et al.,1983). Since % depends on pH and adsorption density(F), use of the Kurbatov method is restricted because itconsiders neither pH nor F (Honeyman and Leckie,1986).

The OH/SO4 exchange Stoichiometry provides an im-portant constraint on surface complexation models andalso is a critical area of research aimed at explainingSO?"-induced effects on soil pH following the applicationof soil amendments containing SOi". With an increasein SO?" adsorption density, the % values were observedto increase from 0.3 to 0.7 for hydrous alumina (Rajan,

HE ET AL.: HYDROXYL-SULFATE EXCHANGE STOICHIOMETRY 443

1978) and from 0.2 to 0.7 for kaolinite (Rao and Srid-haran, 1984). In contrast, Inskeep (1989) obtained anopposite finding for kaolinite, that is, the x values de-creased from 0.7 to 0.1 with increasing adsorption den-sity of SOl~. Zhang et al. (1987) found that variablecharge soils in China had x values of 0.12 to 0.18, whichare much lower than those for pure minerals. However,Curtin and Syers (1990) and Guadalix and Pardo (1991)reported x values between 0.5 and 0.6 for variable chargesoils. More recently, Ajwa and Tabatabai (1995) calcu-lated x values with the Kurbatov method. They showedthat the x values varied with soil types and suspensionpH levels, ranging from 0.5 to 0.8 for K2SO4 additionand 0.1 to 0.5 for CaSO4 addition.

The conflicting reports on the OH/SO4 exchange stoi-chiometry in the literature may be attributed to differencesin experimental conditions and sample characteristics,such as initial pH of suspensions, amount of 864" added,and soil mineralogy. The objective of this study was todetermine the OH/SO4 exchange stoichiometry by theback titration method, using two model minerals, y-Al2Osand kaolinite, three levels of suspension pH, and variousSC>4~ adsorption densities, to obtain a better understand-ing of the OH/SO4 exchange stoichiometry. The experi-mental results were compared with those calculated fromthe adsorption edges at various SQl~ concentrations bythe thermodynamic approach. Kaolinite was selected forstudy because it is a major soil constituent. Althoughy-Al2O3 does not occur in soils, it was chosen for thisstudy because it has been well characterized and widelyused in cation and anion adsorption studies (Girvin etal., 1993).

THEORY OFTHERMODYNAMIC APPROACH

The mathematical derivation given here is based onthe report of Perona and Leckie (1985). They developedthis approach to determine the H+ stoichiometry forcation adsorption. This method is applied here to calcu-late the anion-exchange stoichiometry.

Suppose that there is a system consisting of a solidphase in equilibrium with a liquid phase. The two phasesare separated by an interfacial region, whose compositionat constant temperature is governed by the Gibbs equa-tion, which for a system containing c components is

[1]

where j is the surface tension at the solid-liquid interfaceand F, and (i, are the surface excess concentration andchemical potential of component i. The surface excessconcentration is given by

r, = «//SA [2]where n, is the surface excess moles of component i,and SA is the surface area of the solid.

Consider a typical experiment designed to measurethe adsorption of an anion A2~ by a solid adsorbent inequilibrium with an aqueous solution containing thestrong electrolytes M2A and MX, the strong base MOH,

and H2O. The salt MX is a supporting electrolyte, presentin large excess to maintain a constant ionic strength. Thecomponents are M2A, MX, MOH, H2O, and adsorbent. Ifthe solid adsorbent is insoluble in the solution, its chemi-cal potential is constant, and a term involving the adsor-bent does not appear in the Gibbs equation. From Eq.[1], the Gibbs equation for this system is

-dy = rM2AdM.M2A + FMxdHMx + rMoHdHMOH [3]where the surface excess concentration of water has beenset equal to zero.

The chemical potentials can be written in terms ofionic activities:

2/mnaw [4a]u&x + RTlnaM + RT\nax [4b]

M.MOH = HMOH + RTlnaM + RTlnaOn [4c]Where R is gas constant, T is absolute temperature, anda is ionic activity. In these and subsequent equations,the subscripts M, A, X, and OH will refer to the ionsM+, A2", X", and OH'.

The ions M+ and X~ are present in large excess relativeto the concentration of surface sites and changes in theirconcentrations due to adsorption are negligible. Thus,assuming ionic activity coefficients of unity, the activitiesof M+ and X~ are constants and we have

= 0

[5a][5b][5c]

In addition, we have the stoichiometric relationsrM2A = TA [6a]

TMOH = TOH [6b]Substitution of Eq. [5] and [6] into [3] yields

-dy = #r(rAdlnaA + rOHdlnaoH) [7]The surface tension is a state function since, at constant

temperature, it depends only on the ionic activities whichare themselves state functions. Thus, dy is an exactdifferential and we have

arA \ _ /arOH\

Furthermore, from Eq. [7], we have1 /ay(aA,aoH)\

A RT\ 31naA )a

=

°" RT\

[8]

[9a]

[9b]

and evidently FA and FOH each depend only on a\ anda0H at constant temperature. We can write for the totaldifferential of FA

\ainaAdlnaA + [10]

444 SOIL SCI. SOC. AM. J., VOL. 60, MARCH-APRIL 1996

Keeping TA = constant, Eq. [10] yields

^ dlnaA=-oH

dlnaoH [11]

[12]\dlna0H/rA (drA/dlnaA)aOH

which when combined with Eq. [8], yields/dlnaA\ = (dr0H/dlnaA)aoH =

rA (3rA/dlnaA)aOH 1 dl\ /aoH[13]

Thus, the surface excess concentrations of OH and Aare not independent of one another.

Using the relations

pOH = -logaoHa\ = YA[A]

we have

\dlna,OH/rA

= /aiog[A]\ /31ogYA\\ 3pOH /FA \3pOH /r

[14]

If the pH dependence of the activity coefficient can beneglected, Eq. [14] becomes

= /aiog[A]\ [15]

where the left side is the change in the amount of adsorbedhydroxyls divided by the change in the number of ad-sorbed anions. That is, the left side is the hydroxylstoichiometry, which will be symbolized by %. The rightside is experimentally measurable. It is the change inthe equilibrium concentration of anions with pOH, mea-sured at constant surface excess concentration of theanion.

Rewriting Eq. [15] yields/Alog[A]\

Since ApOH = -ApH, Eq. [16] becomes

/Alog[A]\\ ApH /rA

L J

[17]

The right side of Eq. [17] was evaluated by plottinglogF, in units of moles A per m2 of adsorbent, versuslog[A], in units of moles A per liter at constant pH, andthen by calculating Alog[A]/ApH at constant FA. Ingeneral, such a plot constitutes an adsorption isotherm.

MATERIALS AND METHODSAll reagents used in this study were analytical grade or

better. All experimental vessels and storage containers werepolyethylene or polycarbonate. All experiments were con-ducted at 25 ± 0.5°C. Back titration was duplicated forY-AlaOs, and every fifth sample was duplicated for kaolinite.

Solutions were made with deionized water from a nanopurewater system (Sybron Barnstead, Boston, MA).

Pretreatment on y-A^OaThe Al oxide was a Y-A12O3 made by the Degussa Corp.

(Akron, OH) under the name of Aluminum Oxide C. It hasa surface area of 100 m2 g~' with impurities <0.4% (Ettlingeret al., 1991). The Y-A12O3 was pretreated with 0.01 M KOHto minimize dissolution of a small but highly soluble portionof the solid during adsorption experiments (Girvin et al., 1993).Following a 1-h treatment with KOH, the solid was centrifugedfor 1 h at 15 000 rpm (27 500 X g) (Sorvall RC-5B RefrigeratedSuperspeed Centrifuge, Du Pont Instruments, Newtown, CT),rinsed four times with deionized water, and saturated with0.01 M KC1 solution. The pH of the KCl-saturated Y-A12O3suspensions was adjusted to 4.5, 5.5, or 6.5 with 0.1 M HC1or KOH. The solid/liquid ratio of Y-A12O3 suspensions was10.0 g A12O3 L-1.

Preparation of <2 um Kaolinite Clay SuspensionPoorly crystallized kaolinite (KGa-2) was purchased from

the Source Clay Minerals Repository, Clay Minerals Society,Dep. of Geology, University of Missouri, Columbia, MO.This kaolinite sample, from Warren County, Georgia, has aN2 gas surface area of 23.6 m2 g"1 and a cation-exchangecapacity of 3.3 cmolc kg"1 (van Olphen and Fripiat, 1979).Kaolinite was pretreated with 0.1 M KOH to remove oxideand hydroxide coatings on surfaces (Phelan and Mattigod,1984). Then, the <2-jim clay fraction was separated by centrif-ugation (L.W. Zelazny, 1993. Preparation of samples formineralogical analysis. Lecture notes). The separated clay waswashed four times with 1 M KC1 adjusted to pH 3.0, oncewith deionized water, saturated with 0.01 M KC1, and storedas a suspension. The weight of clay per unit volume of suspen-sion was determined by drying known volumes of suspensionat 110°Cfor 12 hand correcting for the occluded salt. Triplicatemeasurements indicated that the suspension contained 17.0 gkaolinite clay L~'. The suspension pH was adjusted to 4.3,5.5, or 7.0 with 0.1 M HC1 or KOH. The adjusted pH levelsof the stock suspensions of kaolinite and Y-A12O3 changed withtime, but it was <0.05 pH units.

Experimental Procedure and AnalysisBack Titration Method of Determining /

The OH~ released as a result of SO?r adsorption was mea-sured with a recording pH-stat autotitrator (Radiometer, Co-penhagen, Denmark). A 20-mL aliquot of Y-A12O3 or kaolinitesuspension was placed in a 50-mL titration vessel; the suspen-sion was stirred rapidly with a three-bladed polyethylene pro-peller and was purged with a continuous stream of moist N2.The pH end point was set to the required value. Standard stocksolution of SOT (made of K2SO4 in 0.01 M KC1) was theninjected into the suspension with a 1-mL Eppendorf pipette.Concentrations of SOI" in the suspension mixture ranged from0.05 to 2.0 mM for kaolinite and 0.1 to 5.0 mM for Y-A12O3.Volumes of the added SO2." stock solution were between 0.20and 1.00 mL. The standardized HC1 solution, 0.05 M forkaolinite and 0.1 Af for Y-A12O3, was automatically added tomaintain the pre-set pH, and the volume of HC1 consumedwas recorded as a function of time. After 30 min of reactionof SO?T with adsorbent, the suspension was filtered througha 0.2-um polycarbonate membrane (Nuclepore Co., Pleasan-ton, CA) with a 25-mm holder. Preliminary experiments indi-cated that >95 % of the adsorption was completed within 5 min,

HE ET AL.: HYDROXYL-SULFATE EXCHANGE STOICHIOMETRY 445

in agreement with previous reports that the rate of SC>4~ adsorp-tion was rapid (Rajan, 1978; Inskeep, 1989). Nuclepore niterswere chosen because they minimally affect concentrations ofions in solutions (Jardine et al., 1986). The filtrates wereanalyzed for S and Al, and SOi~ adsorption was calculatedby difference. It was necessary to account for changes involume resulting from (i) additions of standardized HC1 tokeep constant pH, and (ii) additions of SOJ" stock solution.

Adsorption Edges for SOi~Sulfate adsorption experiments were carried out in batch

systems to determine adsorption edges (percentage of SOj"adsorbed as a function of solution pH per fixed total SOl~concentration). Eight to fourteen 20-mL aliquots of y-Al2O3or kaolinite suspension were placed in 50-mL centrifuge tubes,and stock solutions of K2SO4 in 0.01 M KC1 were added.There were four batches of y-A!2O3 with initial concentrationsof 0.1, 1.0, 2.5, and 5.0 mM and five batches of kaolinitewith SC-4~ concentrations of 0.01, 0.02, 0.05, 0.10, and 1.00mM. After addition of SOl~, the suspensions were adjustedto the desired pH values with 1 M HC1 or 1 M KOH additionsthat changed the total volume by <2%. The treated suspensionswere shaken for 12 h in a water bath shaker at 100 oscillationsmin~'. Then, the pH of the suspension was determined witha pH meter and a KG2401 combined electrode (Radiometer).The samples were centrifuged for 30 min at 27 500 X g fory-Al2O3 and 30 min at 1950 x g for kaolinite. The supernatantswere then filtered through 0.2-|im Nuclepore membranes. Thefiltrates were analyzed for S and Al. The exact volume ofsolution in each tube was calculated from the volume of suspen-sion, volume of SO*' stock solution added, volume of acidor base added, and particle density of 2.9 g cm"3 for y-Al2O3(Ettlinger et al., 1991) or 2.63 g cm"3 for kaolinite (Dixon,1989).

Analysis of the filtrates for S and Al was performed withan inductively coupled Ar plasma (ICAP)-atomic emissionspectroscopy with simultaneous analysis on an ICAP 61 system(Thermo Jarrell Ash Corporation, Franklin, MA).

RESULTS AND DISCUSSIONAluminum Dissolving from y-AUOa and Kaolinite

Interpretation of OH/SO4 exchange stoichiometry re-quires the measurement of dissolved Al since (i) dissolu-tion reactions of y-AUOs and kaolinite may influence thestoichiometry (Carroll and Walther, 1990; Carroll-Webband Walther, 1988; Furrer and Stumm, 1983; Wielandand Stumm, 1992; Xie and Walther, 1992) and (ii)dissolved Al might complex with SOl" and precipitateas Al hydroxyl sulfates (Adams and Rawajfih, 1977;Adams and Hajek, 1978; Evans, 1991; Evans and Zel-azny, 1990; Nordstrom, 1982). The dissolved Al concen-trations from y-A!2O3 and kaolinite (Fig. 1) decreaseddramatically with increasing pH from 4 to 5, were negli-gible in the pH range of 5 to 9, and slightly increasedat high pH (9-11). The L- or U-shaped pattern of dis-solved Al is in agreement with the results of a-A!2O3and kaolinite (Carroll-Webb and Walther, 1988) andY-A12O3 (Girvin et al., 1993; Baumgarten et al., 1995).The high Al concentrations at low pH (<5) were attrib-uted to proton-promoted dissolution of y-Al2O3 (Furrerand Stumm, 1983, 1986) and kaolinite (Carroll-Webband Walther, 1988; Wieland and Stumm, 1992), while

20

15

-o 10v

v o y-AI203: 10 g L

O o.o• O.IV 1.0T 2.5D 5.0

20

15

10TDIB

Kaolinite: 17 g L

oa

2-0 (mM)

O 0.00• 0.01V 0.02T 0.05Q 0.10• 1.06

I.S.: 0.01 M KCI

•P *Or^

"•"

I

10 12

pH

Fig. 1. Dissolution of y-AhOj and kaolinite as a function of pH atvarious concentrations of SOI".

the relatively high Al concentrations at high pH (>9) wereregarded as caused by hydroxyl-promoted dissolution(Carroll-Webb and Walther, 1988). For y-Al2O3 (Fig.1), the concentration of dissolved Al was much greaterat 2.5 or 5.0 mM SOl~ compared with that at the lowerSC>4~ concentrations. In the case of kaolinite, however,Al dissolution was independent of SOl~ concentration.This might be explained by the mechanism of ligand-promoted dissolution for y-Al2O3 (Furrer and Stumm,1983, 1986). They found that the dissolution rate ofy-A!2O3 depended on the concentration of surface com-plexes formed in the presence of organic acids. Wethus postulate that y-A!2O3 dissolution increased withincreasing SO%~ concentration but only when SOi" con-centration exceeded 1 mM was dissolution significant.

In the process of dissolution, the number of protonsrequired to attach to the reaction site for each detachmentof an Al species into solution, i.e., the H/A1 stoichiometryof dissolution, is 3 for y-A!2O3 (Furrer and Stumm,1986) and 2 for kaolinite, which is the mean of 3 atthe tetrahedral basal surface and 1 at the edge surface(Wieland and Stumm, 1992). The dissolution processreleases OH~ and may result in an overestimation of the

446 SOIL SCI. SOC. AM. J. , VOL. 60, MARCH-APRIL 1996

OH/SO4 stoichiometry. According to the H/A1 ratio, wemay estimate the error in % resulting from dissolution.Considering pH 4.5 and 4.3 at which y-A!2O3 and kaolin-ite, respectively, are most soluble in these back titrationexperiments, the mean Al concentration (data not pre-sented) of 16 samples for y-A!2O3 was 11 mM, whichis equal to 0.22 mmol Al, and for kaolinite was 9.0mM, which equals 0.18 mmol Al in the 20-mL solution.Then, the H+ consumed by dissolution for y-A!2O3 is0.66 mmol, accounting for 6.6% of the total amount ofH+ consumed in the back titration, and for kaolinite is0.36 mmol, accounting for only 4.8%.

The activities of A13+ species in solutions for y-Al2O3and kaolinite were calculated from the total Al concen-trations with the MINTEQA2 computer program (Allisonet al., 1991). The A13+ activities were plotted on sol-ubility diagrams (Fig. 2), with the solubility constants(values of logK) reported by Nordstrom (1982), -17.8for jurbanite [A1(SO4)(OH)-5H2O], -85.4 for alunite[KA13(SO4)2(OH)6], -117.7 for basaluminite [AL,(SO4)(OH),o-5H2O], and -33.9 for gibbsite [A1(OH)3].Potassium activity was set at 10"2 M, and SO4" activitieswere selected at 10~3 M for y-Al2O3 and 10"4 M for

-30

-10

D)O

-15

-20 -

-25

-30

KaolinitejK| = 10 " M

jsoj = io~* u.

12pH

Fig. 2. Solubility diagrams for jurbanite, alunite, basaluminite, andgibbsite with sulfate activities of 10 "3 M for Y-A12O3 and 10 "4 Mfor kaolinite. Symbols represent the data points in this study.

kaolinite. It is evident that the suspensions of both y-A!2O3and kaolinite were undersaturated in the whole range ofpH examined with respect to jurbanite (Fig. 2). Withrespect to basaluminite, the Y-A12O3 suspension was atequilibrium from pH 5 to 7 and the kaolinite suspensionwas undersaturated. With respect to alunite, the y-A!2O3suspension was oversaturated in the pH range of 4 to7, whereas kaolinite was likely oversaturated in the pHrange of 5 to 6.5. These results suggest that alunite couldhave precipitated out of solution. But alunite may notbe important in controlling solution SO4" in these short-term experiments because of its slow nucleation andprecipitation (Nordstrom, 1982). Furthermore, if precip-itation had taken place in solutions, Al concentrationsin zero SO4" treatments should have been higher thanthose in any other SO4" treatments. But this is inconsis-tent with the data presented in Fig. 1. The Al concentra-tions with zero SO4~ addition were similar to or evenlower than those with SO4~ additions. Figure 2 alsoindicates that gibbsite may have precipitated from pH 5to 8 for both y-A!2O3 and kaolinite. The solubility linesin Fig. 2 indicate that all three Al hydroxy sulfates wereunstable compared with gibbsite at pH values >5.5, andthey would transform to gibbsite.

The resultant effect of possible precipitation as eitherAl sulfates, gibbsite, or both is a net depletion of OH",causing an underestimation of the OH/SO4 stoichiometry.Therefore, it may be postulated that the overestimationof x due to dissolution could, to some extent, cancel theunderestimation resulting from precipitation. This seemsmore likely for y-A!2O3 because the solutions were moreoversaturated (Fig. 2). On the other hand, this may alsoexplain why for kaolinite the % values from the backtitration differ from those calculated from the thermody-namic approach (see below for details).

Back Titration and Measured yThere was an instantaneous increase in suspension pH

after the addition of SO|" . This pH increase invoked animmediate addition of HC1 to regain the initial pH. Theperiod of continuous HC1 addition lasted for « 1 min withthe exact time depending on the rate of titrant deliveryspeed. Subsequently, there was relatively little releaseof OH". This relationship is similar to divalent cationadsorption in which metal adsorption on Fe oxides wasalso divided into two steps, fast and slow (Benjaminand Leckie, 1981; Bruemmer et al., 1988; Kinniburgh,1983).

The variation of HC1 addition and the exchange stoichi-ometry (x) with the total SO4~ concentration is shownin Fig. 3 for y-A!2O3 and kaolinite at three pH levels.The volume (milliliters) of HC1 consumed to maintainthe pre-set pH is assumed to indicate the amount of OH"that has been released after SO4" addition. That is, thegreater the volume of HC1 required to regain the pH tothe pre-set level, the higher the OH" release. Fory-A!2O3, the effect of a pre-set pH on OH" release waspH 5.5 « 6.5 > 4.5 (Fig. 3a), indicating that y-A!2O3released more OH" at a higher pH. This is consistentwith the reports of Rajan et al. (1974) and Shang et

HE ET AL.: HYDROXYL-SULFATE EXCHANGE STOICHIOMETRY 447

0.3

0.2 -

o>x 0.1

0.0

O pH4.5• pH5.5V pH6.5

7-AI20, O pH4.3• pHS.5V pH7.0

Kaolinite

1.5 -

1.0 h

0.5

V pH6.5• pH5.5O pH4.5

rr-— V —— ̂

. 7-AI20,

7_V- —— V ——— V

0.0 0.5

2-Total SO (mM)

1.0 1.5

~

2.0

Total SO, ~ (mM)*r **•

Fig. 3. Consumption of HC1 (VHC1) [(a) and (b)] and the OH-/SO|- exchange stoichiometry (jfl [(c) and (d)] as a function of the total SOi~concentration for y-AI2O3 and kaolinite at three pH levels.

al. (1992), in which phosphate was adsorbed by Alhydroxide. But for kaolinite, the effect of a pre-set pHon OH- release was pH 4.3 > 5.5 > 7.0 (Fig. 3b),suggesting that kaolinite released more OH at a lowerpH. Zhang et al. (1987) observed similar results inSOI" adsorption by variable charge soils. This may implythat, for soils consisting mainly of kaolinite, the moreacid the soil, the greater the release of OH" with additionof SOI". On the other hand, for both y-A!2O3 and kaolin-ite, there are higher values of the OH/SO4 stoichiometryat higher pH levels (Fig. 3c and 3d). Similar resultswere obtained by Rajan et al. (1974), Shang et al. (1992),and Zhang et al. (1987). It is unclear why the higherOH" release results in a greater x value for y-AhOa,whereas the higher OH" release has a lower % value forkaolinite. This trend for kaolinite might suggest thatthe change in the total amount of OH~ released is notnecessarily proportional to the change in the amount ofSOI" adsorbed. This might indicate that some fractionof SOI" that disappeared from the equilibrium solutionmay not be attributable to adsorption, if SO^" adsorptionis considered strictly as a ligand exchange reaction.Alternatively, it might mean that another mechanismother than ligand exchange is involved in SOl" adsorp-tion. It is evident, however, that pH has a greater effecton % than does SOl~ concentration. For instance, the

values of % for y-A!2O3 at pH 5.5 increased from 0.8to 1.0 as the total SOl~ concentration increased from0.5 to 5.0 mM, while at the concentration of 2 mMSO2", X increased from 0.4 at pH 4.5 to 1.0 at pH 5.5and to 1.3 at pH 6.5.

Adsorption Edges and Calculated xThe percentage SO4- adsorption is greater at lower

pH and decreases as pH increases for both y-A!2O3 andkaolinite (Fig. 4). The SO^ adsorption falls to zeroaround pH 10 for y-Al2O3 and pH 11 for kaolinite. Thecurves shift to lower pH and to lower adsorption plateauswith increasing SOi~ concentrations. Thus, as the totalSO|- concentration increases, the percentage adsorptionat a given pH decreases (Sigg and Stumm, 1981). At1.0 mM SOi", the adsorption edge becomes relativelyflat below pH 4.5, implying site saturation of kaolinite.However, in this experiment, site saturation of y-A!2O3 isnot evident at the highest SOT concentration (5.0 mM).

The amount of adsorbed BOl~ (F) and SO^ equilib-rium concentration (Ceq) at selected pH values can beobtained from Fig. 4. The pH values were chosen tobracket those in the back titration experiments. For exam-ple, since one of the pH levels in the back titration fory-A!2O3 was 4.5, two pH values were selected, i.e., 4.4

SOIL SCI. SOC. AM. }., VOL. 60, MARCH-APRIL 1996

100

-oD

XIl_O

Fig. 4. Adsorption edges for SOi" on y-AbOj and kaolinite.

and 4.6, which average 4.5. With a smaller pH interval,the calculated x is more representative of the target pH.A similar narrow pair of adsorption isotherms at twopH levels, which bracket the pH value examined in theback titration experiment, were obtained from Fig. 4and plotted in Fig. 5 for y-AhOs and kaolinite. The datapoints shown by closed symbols in Fig. 5 represent theamounts of SOl" adsorption at lower pH values, whileopen symbols are for higher pH values.

The calculated OH/SO4 stoichiometries are comparedwith experimentally measured values in Fig. 6. To calcu-late % from a narrow pair of adsorption isotherms inFig. 5, a value of logF must first be chosen. At thatselected logF value, logCi and logC2, which correspondto pHi and pH2, respectively, were obtained from theadsorption curves in Fig. 5. Since logC2 — logCi =AlogC and pH2 - pH, = ApH, the % value at this logFcan be calculated with AlogC/ApH, i.e., Eq. [17]. Thecalculated % occurs at the average pH between the narrowpair of adsorption isotherms (Fig. 6). The mean % valuesacross all measured F at each pH level examined arealso presented in Table 1. The negative sign of 7 resultsfrom the fact that adsorption of an anion is accompaniedby the desorption of OH~, i.e., dFA is positive, whereasdFoH is negative. At a low pH (e.g., 4.5), the OH/SO4stoichiometry values were similar between y-A!2O3 and

-6

-7 t-pH:

4.4• 4.6

o 5.4T 5.6

v 6.4• 6.6

KaolinitepH:

4.2 o 5.4• 4.4 - 5.6

» 6.8• 7.0

O>O

-7

-8 -6 -4

Log Ce (moles S04 L }-2

Fig. 5. Adsorption isotherms of SOi~ on y-AUOa and kaolinite atvarious pH levels (T = adsorption density; C«, = equilibriumconcentration).

kaolinite (Table 1). At higher pH levels, % values fory-AJzOs were greater than those for kaolinite, indicatingthat the eifect of pH on x is more significant for y-AhOsthan for kaolinite. Honeyman and Leckie (1986) observedthat titania shows little change in H/Cd stoichiometrywith either pH or surface coverage, while amorphousFe oxyhydroxide and Al oxide exhibit large ranges inthe stoichiometry. The calculated values of x are in goodagreement with experimental results for y-AbOa (Table1; Fig. 6). Although for kaolinite, the average % valuesdetermined from the back titration were rather close tothose calculated from the thermodynamic approach (Ta-ble 1), and there was a discrepancy between the twomethods, particularly at pH 4.3 and 5.5 (Fig. 6). Severalfactors may be involved in this discrepancy. Greaterdissolution of kaolinite might have taken place in theadsorption edge experiment than in the back titrationexperiment since the former had a much longer equilib-rium time. Dissolution of kaolinite by detaching Si aswell as Al depletes H+ in solution as indicated earlier.Furthermore, the detached Si may compete with SO|~in adsorption. Therefore, the calculated % values fromthe adsorption edge experiments may be overestimatedand greater than those from the back titration experi-ments. This is more evident at low pH levels (Fig. 6)where dissolution of kaolinite is more prominent.

The noninteger x values are reasonable consideringthe complexity of these heterogeneous systems. Thiscomplexity includes the reactions of surface protolysis,

HE ET AL.: HYDROXYL-SULFATE EXCHANGE STOICHIOMETRY 449

t.o

0.5

pH 4.5

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.

0.00.0 0.3 0.4 0.5

T (yLtmol S04 m )

Fig. 6. Variation of the sulfate exchange stoichiometry (y) with pHand adsorption density (T) for Y-A12O3 and kaolinite.

adsorbent dissolution, adsorbate precipitation, and elec-trolyte ion adsorption as well as the SOl~ adsorptionreaction. Thus, % may be better referred to as the macro-scopic OH/SO4 stoichiometry (Honeyman and Leckie,1986) because it represents the net release of OH~ orconsumption of H+ by all reactions, which results in therelease of OH~ and the removal of SOl~ from solution.

Mechanism of SO3~ AdsorptionThis study demonstrated that (i) at constant pH, the

OH/SO4 stoichiometry (x) increases with increasing sur-face coverage (T) (Fig. 6); (ii) at constant adsorptiondensity (F), x increases with increasing pH; and (iii) ata low pH (i.e., 4.5), x values for y-A!2O3 are very closeto those for kaolinite, while at a higher pH, x valuesfor y-A!2O3 are greater than those for kaolinite. Theseexperimental results may be explained by two completelyopposite mechanisms for SOl~ adsorption onto y-A!2O3and kaolinite.

Anion adsorption reactions take place by at least twomechanisms: (i) ligand exchange (i.e., inner-sphere com-plexation) on the surface (i.e., o-plane) and (ii) specificadsorption (i.e., outer-sphere complexation) on thep-plane that is away from the surface plane by somedistance. The type of adsorption reaction can vary with

Table 1. Mean values of x obtained from measurement and calcu-lation at selected pH levels for Y-A12O3 and kaolinite.

PH

4.55.56.5

Y-A1203

Measured-0.42-0.94-1.25

KaoliniteCalculated

-0.43-0.91-1.21

PH4.35.57.0

Measured-0.43-0.59-0.80

Calculated-0.48-0.65-0.79

differences in surface properties between y-A!2O3 andkaolinite. The reactive surface sites on y-A!2O3 are allAl-OH groups. The distance between two adjoining Alatoms in Al oxide is 0.29 nm (Rajan, 1978), which isclose to the distance of 0.24 nm between two oxygenatoms in SO?~ ions (Parfitt and Smart, 1977). It ispossible that one SO?" ion may complex with two adjoin-ing Al-OH groups to form a binuclear ring structure(Harrison and Berkheiser, 1982; Parfitt and Smart, 1977;Rajan, 1978). These surface complexes are formed bysharing the edge of a SO?" tetrahedron with two cornersfrom two adjacent Al octahedra in y-A!2O3 and thereforeis referred to as a corner-sharing structure (O'Day etal., 1994). Kaolinite surface sites are divided into threetypes, octahedral Al-OH (aluminol), tetrahedral Si-OH(silanol), and transitional tetrahedral-octahedral Al-OH-Si (White and Zelazny, 1988). Of the three types, Al-OH and Si-OH groups are considered as reactive surfacesites (Sposito, 1984), although they differ in protolysis ofsurface groups. Since the A1-A1 distance in the octahedralsheet of kaolinite is very similar to that in y-A!2O3, itis probable that each SO?" ion forms a binuclear bondingstructure with two corners of two adjacent Al octahedra.However, the Si-Si distance is =0.32 nm, from whichthe O-O distance between adjacent tetrahedra was calcu-lated at =0.5 nm based on the findings of Bish and vonDreele (1989). Therefore, the formation of binuclearbonding between SO%~ tetrahedra and Si tetrahedra isunlikely because of the large difference in O-O distance,and thus monodentate bonding must be assumed, i.e.,one corner of a SO?" tetrahedron is bound to the cornerof a Si tetrahedron in kaolinite.

Considering the inner-sphere complexation (ligand ex-change) mechanism for SO?" adsorption, possible reac-tions at varying pH levels may be expressed as follows.For y-A!2O3,

«£ISS *•«-<:=£< *»>at low pH [18]

O'- -Al—OH2+

"Al—OH°+ SOi •o: -Al—O-^.

-Al—Cr"at intermediate pH

at high pH

[19]

2OH-

[20]

(i) At low pH where surface reactive groups are moreprotonated, each SOi~ displaces two OH2 groups iny-A!2O3 (Eq. [18]), resulting in a low OH/SO4 stoichiom-etry. (ii) At intermediate pH where the surface is lessprotonated, each SOi~ may replace one OH and one

450 SOIL SCI. SOC. AM. J., VOL. 60, MARCH-APRIL 1996

OH2, causing a stoichiometry of 1 (Eq. [19]). (iii) Athigh pH, each SO4~ displaces two OH groups (Eq. [20]),causing a OH/SO4 value of 2. This implies that the OH/SO4 stoichiometry increases with increasing suspensionpH and would theoretically range from 0 to 2 for Y-A12O3.(iv) At low concentrations, SO4~ would possibly onlydisplace OH2 groups. As SO4~ concentration increases,more SO4~ ions would displace OH groups. Conse-quently, the stoichiometry increases with increasingSO4~ concentration (i.e., adsorption density).

For kaolinite, we need to consider the tetrahedral Si-OH groups as well as the octahedral Al-OH groups.The Al-OH groups on kaolinite would function similarto those on y-AlaOs, while the Si-OH groups could onlyadsorb SO4~ by forming a monodentate structure asfollows:

Si-OH° Si-SO4 [21]Thus, the OH/SO4 stoichiometry for kaolinite would bethe average of tetrahedra and octahedra adsorption andwould theoretically range from 0 to 1.5, which is lowerthan that in y-A!2O3, if the assumption is made that thereare two aluminol and one silanol group available forSOl" per unit area (Sposito, 1984).

The reactions of SO4~ adsorption at varying pH levelsfor the second case, i.e., outer-sphere complexation,may be described below. For y-Al2O3,

O: -Al—OH2+

^Al—OH2+

-Al—OH2+

"Al—OH°

+ SO2*- - O -Ai-OH2v~A1—OH2

+-'

H2o

o:/Al—OH° 2H2O

Al—OH}'at intermediate pH

4 ^A1-OH2+''"

at high pH

at low pH

[22]

D|-+OH-

[23]

+ 2OH~

[24]

(i) At low pH, the surface reactive groups are all proton-ated, thus each SO4~ is attracted by two positivelycharged OH2 sites (Eq. [22]), forming an electrostaticallybound ion pair, (ii) At high pH, some surface groupsare deprotonated into the form of OH, which must beprotonated before SO|~ is adsorbed by Y-A12O3 (Eq. [23]and [24]). Protonation of surface sites consumes H+ insolution in an amount equal to the release of OH" insolution. Thus, the stoichiometry of SO4~ adsorptionwould range from 0 to 2 for y-Al2O3 and increase withincreasing solution pH. (iii) At low SO4~ concentrations,SO|~ ions are paired with OH2 groups, resulting in alow OH/SO4 stoichiometry. As SOi" concentration in-creases, protonation of more surface OH groups followedby the interaction with SO4~ ions causes an increase inthe stoichiometry.

In the case of kaolinite, the Si-OH groups again needto be considered as well as Al-OH groups. The Al-OH groups would again behave like those on y-A!2O3,whereas the Si-OH groups would behave as follows:

Si-OH° + H+ Si-OH2+ •so?- [25]

On the analogy of inner-sphere complexation in kaolinite,

the stoichiometry for outer-sphere complexation wouldchange from 0 to 1.5. These OH/SO4 stoichiometryvalues are similar comparing ligand exchange to specificadsorption mechanisms.

Both of the mechanisms described above have beenproposed to account for the adsorption of SO4~ by soilsand soil constituents. Many studies have concluded that,like phosphate, SO4~ is adsorbed by a ligand exchangemechanism (Bolan et al., 1993; Guadalix and Pardo,1991; Marcano-Martinez and McBride, 1989; Parfitt,1978; Rajan, 1978, 1979; Zhang et al., 1987). An in-crease in suspension pH during SOl~ adsorption hasbeen regarded as evidence for this conclusion. But fromthe viewpoint of the surface complexation reactions de-scribed above, OH~ release is not necessarily responsiblefor the exchange reaction of surface OH~ with anionsin solution.

Sposito (1989) considered that SO4~ may possibly beadsorbed on soil colloids by an outer-sphere complex-ation mechanism because of its readily exchangeablecharacter. Ryden et al. (1987) have shown that SO4~adsorption by a Fe (III) oxide gel was completely elimi-nated when this anion was added together with equimolaramounts of phosphate, suggesting that SO4~ does notcompete effectively with phosphate for adsorption sites.Bolan and Barrow (1984) and Barrow (1985) consideredthe importance of the location of the plane of adsorptionand suggested that the mean plane of SO4~ adsorptionwas more distant from the adsorbent surface than thatof phosphate. Marsh et al. (1987) have explainedSO4~ adsorption by soils in electrostatic terms, with theprocess taking place in a plane distinct from the surfacebut closer than the plane of adsorption of monovalentanions such as Cl~ and NOs~. Curtin and Syers (1990)observed that adsorbed SO4~ was completely recoveredby washing with 1 MKC1 and concluded that SO4~ maynot be chemisorbed as commonly supposed. The findingsof Zhang and Sparks (1990) confirmed the outer-spherecomplexation mechanism for SO4~ adsorption on goe-thite, that is, SO4~ is adsorbed on a positive site byelectrostatic attraction.

Therefore, OH~ release is not unambiguous evidencethat a ligand exchange reaction occurs in anion adsorp-tion, even though it has long been used as such. Theionic strength effect on ion adsorption may be a bettertechnique to ascertain the mechanism of cation or anionadsorption (Hayes et al., 1988; Goldberg et al., 1993).For an inner-sphere complex, there is little effect of ionicstrength on the behavior of ion adsorption. In contrast,if an ion forms an outer-sphere complex with a surface,an increase in ionic strength markedly inhibits the adsorp-tion of the ion. However, obtaining direct molecular-levelinformation may only be possible with noninvasive tech-niques such as x-ray adsorption fine structure spectros-copy (Fendorf et al., 1994).

SUMMARY AND CONCLUSIONSThe precise determination of the OH/SO4 stoichiome-

try is not easily made by either the direct method (backtitration) or the indirect (thermodynamic) approach; how-

HE ET AL.: HYDROXYL-SULFATE EXCHANGE STOICHIOMETRY 451

ever, the macroscopic stoichiometries have been mea-sured by both methods and are in agreement with eachother for y-A!2O3 and kaolinite. Although a short reactiontime (30 min) was used in the back titration method, itis still time consuming since each value represents asample titration. The experiments for the indirect methodare more easily conducted, and the method is applicableat relatively low adsorbate concentrations. This methodpermits the evaluation of OH/anion stoichiometry as afunction of pH and adsorption density. But at a lowpH, the stoichiometry might be overestimated due todissolution and may be underestimated at a high pHbecause of possible precipitation of basic Al sulfates andgibbsite.

The results from the direct determination and the indi-rect calculation have substantiated that there was a sig-nificant difference in the OH/SO4 stoichiometry for thetwo minerals, y-Al2Os and kaolinite. The adsorption ofSC>4~ resulted in the net release of =0.9 mol OH~ permol SC>4~ adsorbed for y-A!2O3 vs. 0.6 mol OH~ permol SO4~ adsorbed for kaolinite. Variation in the stoichi-ometry with pH, adsorption density, and surface typemay be explained by two contrasting mechanisms forSO4~ adsorption on y-Al2Os and kaolinite, by eitherligand exchange (inner-sphere complexation) or specificadsorption (outer-sphere complexation). It suggests thatOH~ release during anion adsorption is not necessarilyresponsible for the exchange reaction of surface OH withanions in solution. It is plausible that similar OH~ releasecould occur through surface site protonation and outer-sphere complexation reactions.

From the results in this work, it may be concludedthat the amount of OH~ released from soils after theaddition of SOi" depends on the amount of SO%~ added,initial pH, and mineralogy of soils. The soils high inFe and Al oxides would have a high release of OH".These soils also would produce more OH" at a pH of= 6 than at a pH of 4.5. Considering that soil is abuffering system, it is uncertain if SOl~ addition wouldincrease soil pH. The released OH~ may be depleted byweak acids present in soils and the pH increase thus notdetectable. Nevertheless, the application of SO^" asK2SO4 to soils would not decrease soil pH provided thatthe precipitation of SO^ as a basic Al sulfate is negligibleand K+ ions cannot effectively exchange with the ad-sorbed A13+ and H+. On the other hand, for the additionof gypsum (CaSO4-2H2O) to soils, while soil pH wouldincrease in most cases, a decrease in pH may occur.For instance, in soils high in exchangeable Al, Ca2+

from CaSO4 could replace A13+ that hydrolyzes to giveH+ (Berry et al., 1990; Seaman et al., 1995). If thereleased H+ exceeds the released OH~ due to SOl~adsorption, soil pH would decrease.

ACKNOWLEDGMENTSGratitude is expressed to D.L. Sparks and A. Scheidegger,

University of Delaware, for their constructive comments onan early draft of this paper. Thanks are due to two anonymousreviewers for their elaborate reviews and most useful criti-cisms. This research was carried out under the collaborationof Virginia Tech with USDA-ARS at Beckley, WV, and sup-

ported by the contract of specific cooperative agreement no.58-1932-2-036. The senior author gratefully acknowledges thefinancial support from USDA, the C.I. Rich Fellowship, andthe graduate scholarships.

452 SOIL SCI. SOC. AM. J., VOL. 60, MARCH-APRIL 1996

Furrer, G., and W. Stumm. 1983. The role of surface coordinationin the dissolution of d-A!2O3 in dilute acids. Chimia 37:338-341.

Furrer, G., and W. Stumm. 1986. The coordination chemistry ofweathering: I. Dissolution kinetics of d-A!2O3 and BeO. Geochim.Cosmochim. Acta 50:1847-1860.

Girvin, D.C., P.L. Gassman, and H. Bolton, Jr. 1993. Adsorptionof aqueous cobalt ethylenediaminetetraacetate by d-Al2<D3. Soil Sci.Soc. Am. J. 57:47-57.

Goldberg, S., H.S. Forster, and E.L. Heick. 1993. Boron adsorptionmechanisms on oxides, clay minerals, and soils inferred from ionicstrength effects. Soil Sci. Soc. Am. J. 57:704-708.

Guadalix, M.E., and M.T. Pardo. 1991. Sulfate sorption by variablecharge soils. J. Soil Sci. 42:615-628.

Harrison, J.B., and V.E. Berkheiser. 1982. Anion interactions withfreshly prepared hydrous iron oxides. Clays Clay Miner. 30:97-102.

Harvey, D.T., J.E. Fulghum, and R.W. Linton. 1983. Determinationof adsorption stoichiometry for zinc adsorption on amorphoushydrous ferric oxide. J. Colloid Interface Sci. 94:276-278.

Hayes, K.F., C. Papelis, and J.O. Leckie. 1988. Modeling ionicstrength effects on anion adsorption at hydrous oxide/solution inter-face. J. Colloid Interface Sci. 125:717-726.

Honeyman, B.D., and J.O. Leckie. 1986. Macroscopic partitioningcoefficients for metal ion adsorption, p. 163-190. In J.A. Davisand K.F. Hayes (ed.) Geochemical processes at mineral surfaces.Am. Chem. Society, Washington, DC.

Hue, N.V., F. Adams, and C.E. Evans. 1985. Sulfate retention byan acid Be horizon of an Ultisol. Soil Sci. Soc. Am. J. 49:1196-1200.

Inskeep, W.P. 1989. Adsorption of sulfate by kaolinite and amorphousiron oxides in the presence of organic ligands. J. Environ. Qual.18:379-385.

Jardine, P.M., L.W. Zelazny, and A. Evans, Jr. 1986. Solutionaluminum anomalies resulting from various filtering materials. SoilSci. Soc. Am. J. 50:891-894.

Kinniburgh, D.G. 1983. The H+/M2+ exchange stoichiometry ofcalcium and zinc adsorption by ferrihydrite. J. Soil Sci. 34:759-768.

Kurbatov, M.H., G.B. Wood, and J.D. Kurbatov. 1951. Isothermaladsorption of cobalt from dilute solutions. J. Phys. Chem. 55:1170-1182.

Marcano-Martinez, E., andM.D. McBride. 1989. Calcium and sulfateretention by two Oxisols of the Brazilian Cerrado. Soil Sci. Soc.Am. J. 53:63-69.

Marsh, K.B., R.W. Tillman, and J.K. Syers. 1987. Charge relation-ships of sulfate adsorption by soils. Soil Sci. Soc. Am. J. 51:318-323.

Nordstrom, D.K. 1982. The effect of sulfate on aluminum concentra-tions in natural waters: Some stability relations in the system AI2O3-SO3-H2O at 298 K. Geochim. Cosmochim. Acta 46:681-692.

Gates, K.M., and A.G. Caldwell. 1985. Use of by-product gypsumto alleviate soil acidity. Soil Sci. Soc. Am. J. 49:915-918.

O'Day, P.A., G.A. Parks, and G.E. Brown, Jr. 1994. Molecularstructure and binding sites of cobalt (II) surface complexes onkaolinite from x-ray absorption spectroscopy. Clays Clay Miner.42:337-355.

Parfitt, R.L. 1978. Anion adsorption by soils and soil materials. Adv.Agron. 30:1-50.

Parfitt, R.L., andR.St.C. Smart. 1977. Infrared spectra frombinuclearbridging complexes of sulfate adsorbed on goethite (a-FeOOH).J. Chem. Soc. Faraday I. 73:796-802.

Pavan, M.A., F.T. Bingham, and P.P. Pratt. 1982. Toxicity ofaluminum to coffee in Ultisols and Oxisols amended with CaCO3,MgCO3, and CaSO4-2H2O. Soil Sci. Soc. Am. J. 46:1201-1207.

Pavan, M.A., F.T. Bingham, and P.P. Pratt. 1984. Redistribution

of exchangeable Ca, Mg, Al following lime or gypsum applicationsto a Brazilian Oxisol. Soil Sci. Soc. Am. J. 48:33-38.

Perona, M.J., and J.O. Leckie. 1985. Proton stoichiometry for theadsorption of cations on oxide surfaces. J. Colloid Interface Sci.106:64-69.

Phelan, P.J., and S.V. Mattigod. 1984. Adsorption of molybdateanion by sodium-saturated kaolinite. Clays Clay Miner. 32:45-48.

Rajan, S.S.S. 1978. Sulfate adsorbed on hydrous alumina, ligandsdisplaced, and changes in surface charge. Soil Sci. Soc. Am. J.42:39-44.

Rajan, S.S.S. 1979. Adsorption and desorption of sulfate and chargerelationships in allophanic clays. Soil Sci. Soc. Am. J. 43:65-69.

Rajan, S.S.S., K.W. Perrott, andW.M.H. Saunders. 1974. Identifica-tion of phosphate-reactive sites of hydrous alumina from protonconsumption during phosphate adsorption at constant pH levels.J. Soil Sci. 25:438-447.

Rao, S.M., and A. Sridharan. 1984. Mechanism of sulfate adsorptionby kaolinite. Clays Clay Miner. 32:414-418.

Ritchey, K.D., D.M.G. Souza, E. Lobato, and O. Corra. 1980.Calcium leaching to increase rooting depth in Brazilian SavannahOxisol. Agron. J. 72:40-44.

Ryden, J.C., J.K. Syers, and R.W. Tillman. 1987. Inorganic ionadsorption and interactions with phosphate adsorption by hydrousferric oxide gel. J. Soil Sci. 38:211-217.

Seaman, J.C., P.M. Bertsch, and W.P. Miller. 1995. Chemical con-trols on colloid generation and transport in a sandy aquifer. Environ.Sci. Technol. 29:1808-1815.

Shainberg, I., M.E. Sumner, W.P. Miller, M.P.W. Farina, M.A.Pavan, and M.V. Fey. 1989. Use of gypsum on soils: A review.Adv. Soil Sci. 9:1-101.

Shang, C., J.W.B. Stewart, and P.M. Huang. 1992. pH effect onkinetics of adsorption of organic and inorganic phosphates byshort-range ordered aluminum and iron precipitates. Geoderma 53:1-14.

Sigg, L., and W. Stumm. 1981. The interaction of anions and weakacids with the hydrous goethite (a-FeOOH) surface. Colloids Sur-faces 2:101-117.

Sposito, G. 1984. The surface chemistry of soils. Oxford UniversityPress, New York.

Sposito, G. 1989. The chemistry of soils. Oxford University Press,New York.

Sumner, M.E. 1990. Gypsum as an ameliorant for the subsoil aciditysyndrome. Florida Inst. of Phosphate Res. Rep. 82-01-024R. FIPR,Bartow, FL.

Sumner, M.E. 1993. Gypsum and acid soils: The world scene. Adv.Agron. 51:1-32.

Sumner, M.E., H. Shahandeh, J. Bouton, and J. Hammel. 1986.Amelioration of an acid soil profile through deep liming and surfaceapplication of gypsum. Soil Sci. Soc. Am. J. 50:1254-1258.

van Olphen, H., and J.J. Fripiat. 1979. Data handbook for claymaterials and other non-metallic minerals. Pergamon Press, NewYork.

White, G.N., and L.W. Zelazny. 1988. Analysis and implication ofthe edge structure of dioctahedral phyllosilicates. Clays Clay Miner.36:141-146.

Wieland, E., and W. Stumm. 1992. Dissolution kinetics of kaolinitein acidic aqueous solution at 25°C. Geochim. Cosmochim. Acta56:3339-3355.

Xie, Z., and J.V. Walther. 1992. Incongruent dissolution and surfacearea of kaolinite. Geochim. Cosmochim. Acta 56:3357-3363.

Zhang, P.C., and D.L. Sparks. 1990. Kinetics and mechanisms ofsulfate adsorption/desorption on goethite using pressure-jump relax-ation. Soil Sci. Soc. Am. J. 54:1266-1273.

Zhang, G.Y., X.N. Zhang, and T.R. Yu. 1987. Adsorption of sulfateand fluoride by variable charge soils. J. Soil Sci. 38:29-38.