- Email: [email protected]
Equilibrium adsorption of thorium by metal oxides in marine electrolytes
EquiIibrium adsorption of thorium by metal oxides in marine electrolytes I(EIm A. HUNTER,DAVID J. HAWEEand LEEKWEECHOO ChemistryDepartment,Univnsity of Otago, Dunedin, New Zealand (Received February23, 1987;accepted in revisedfm
November 26, 1987)
Abstract--The equilibriumadsorptionof Th by the hydrousoxides goetbite(a-FeOOH)and &MnQIin mprineek-ctrolytesis not alkted by the majorcationsCa*+and MS?+,relativeto NaClelectrolyte,while a- deaeeuw adsorption through competitive ion pairing with Th in solution. A triple layer model of %ufm speciation in N&.3 ektrolyte qgested that adsorptionof Th mainly involves the hydrolyzedforms Th(OHp, Th(OHg and Th(OHk. Intrinsic surfPee~adi~~~~fort~~~i~~fi~~~rn~~~~~ex~rn~~~n data. In ~~~n~~~ ekctrolytcs including seawater,the model conzetiy prcdiaed the skpc of the adsorption awvcs fix both oxids, with the pH of the adsorptionedge depending on value used for Th-sulphateion-pairing equilibrium oonstants. Competition experimentsusing synthetic organicligands to bulk the Th concentrationswith &Mn~ loweredtbe concentrationof freeTh availablefor adsorptionto withia the rangeof oceanic dissolvedTh concentrations.This gave rise to shitts in the adsorptionedge to higher pH close to that of seawater,which were consistently pmdktcd by the model. Application of the model to the deep ocean watercolumn suggc&d that Shin@ would not k a pomrful scsvenga for ‘I%becauseof the developmentof a strongnegativesurfacechat8eat pH 8, whik adsorptionon 8oethite shoul&be significant. effects of carbonate alkalinity on Th adsorption by goethite
INTRODUCMON
in 0.1 M NaNOr ekctrolyte. Laboratory studies of ion adsorption have the advantage over field studies that different factors influencing the adsorption process may be systematically varied to reveal their part in the mechanism. As with earlier work on other trace metals (e.g. BALERRER~andMu~~+~, 1982a,b, 1983.1984; DAVIS, 1984), our investigation has focuJod on the thermodynamics of the adsorption ps~%, and has sought to determine the factors that infhrence the degree of Th uptake by metals oxides in seawater media. in particular, the possible competitive effects of the major ions and the influence of pH and the concentration of available surface were studied. For most trace elements, however, laboratory studies have the disadvantage that it is usually difhcuh to reproduce the very low concentrations pmaent natumlly in the ocean he-
THORIUMISOTOPES are usefi11tools for studying the removal of reactive trace elements from the ocean through scavenging by particles sinking through the water column (TIJREKIAN,
1977). The deep-sea distribution of Th isotopes is consistent with a reversible exchange of Th between dissolved and particulate forms (NOZAIUet 1, 1981; BACONand ANDERSON, 1982). The available evidence suggests that Th and other u-a@ ekments of differing reactivity underga removal from the water column by maintaining adsorptive equilibrium, with the surfaces of sinking particles (BREWERand HAO,
1979). Surface-adsorbed oqanic matter and surface coatings on f-manganese oxides, known for their adsorptive afhnity for metal ions, have been implicated as likely scavenging p~~~~e~~~~(B~~~, 1981; HUNTER,1983). Accordingty, recent attention has been focused on the adsorptive equilibria of trace metals in seawater media with either w&defined model phases such as goethite (a-F&OH)
by experimental In the pm&t work we-have addressed this shortcoming through the use of solution organic ligands which compete with the surface for Th, thereby reducing the concentration of free Th iu solution. The hgands wem chosen to reduce the concentration of free thorium a&able for adsorption, through metal ion buff&r& to kvels comparable to those measumd in deep ocean waters.
or MnOr (BALLWRDXI and MURRAY,1981,1982a,b, 1983), or natural marine particulate material obtained from scdiment traps (NYEEEEER et al., 1984) and from interfacial sediment (BAUSTRIERI and MURRAY,1984). This information is essential in providing the framework for a surface &em&d explanation for the scavenging behavior of reactive elements. In the present work we report on studies of the surface adsorption of Th onto the surfaces of two synthetic oxides, go&rite (a-F&OH) and the manganese oxide &MnOz. Goethite is a hydrous Fe(III) oxide that has been extensively studied as a model for iron oxide components of marine particulate matter (BALISTRIERI and MURRAY, 1981), while 6-MtQ believed to be similar in surface properties to the
MATERULS
ANU METHODS
Gocthite (&%OOH) was pnpand accordingto the method of ATKIIWNet al. (1967), whik b-MnOzwas prqared accordingto the method of MU-Y (1974) and BALWWERIand MURRAY (1982b). Both productswereptu%al by dialysis and suned at pH 5 in hard borosihcategkss. The oxides wtre cbaractnized to allow comparisonwith other work by severalmeasurementmetho& The @Cthiteshowedchasactainic X-raydif%ction re&aions at 4.2,2.7,awl2.4A,MdhadapH~thcpointoflnocba%e,pH(PZCX ~~~~~~~~ ~MIJRRAY (1981) of 7.1, similar to previous work. The total titratablecharge was 65 + 3 mmol of ions (as H+) per k8 of oxide. The &MnOl was poorly orckr~Iwith the most distinct refkctio~
manganese oxide components of marine particulate matter (BALISTRIERI and MURRAY, 1982bf. The surface chemistry of Th in seawater media has not previously been reported. However LAFLAMMEand MURRAY(1987) have studied the 627
628
K. A. Hunter, D. J. Hawke and L. K. Choo
being at 3.90 A, had a pH(PZC) of 1.7 * 0.3 and a B.E.T. surface area of 130 m*/g. Stoichiometric analysis by oxalate titration and atomic absorption (MURRAY,1974) gave an O:Mn ratio of 1.89 + 0.05. These results are all consistent with the results of other workers (e.g. MORGAN and -MM, 1964; MURRAY, 1974). For this oxide, the concentration of surface sites was measured by fast tritium exchange (YATESand HWLY, 1976), giving a result of 5.0 i 0.7 mol/ kg This compares favorably with a value of 8.6 md/kg determined recently by C~rrs and LANGMUIR (1986), and to values derived from adsorption studies (GVU et d, 1975; MURRAY, 1974; DE.htPSEY and SINGER,1980),but is considerably lower than a value of 27 mol/ kg measured by BALEXRIERIand MURRAY( I982b). A range of electrolytes was used in the adsorption studies to assess the effects of the seawater major ions (Table I). Potassium was not included because we expected it to be similar in behaviour to sodium and therefore not signiticant at its relatively low concentration in seawater. Results of LAFLAMMEand MURRAY (1987) suggest that
the effectsof carbonateand hicarhonateions on Th adsorptionby goethitecan he negketed at seawateraikahnities. The electrolyte solutions were prepamd from AR grade reagents in distilled, deionised, and t&red water from a “Milli+” system. Standard&&n of the calcium electrolyteswas hy EDTA titration. Th(IV) solutions were I)reDILndfromA.R. Th(NO&. All storane and e&&ally glass vessels, to control potential organic co%minati&, from plasticixers, since preliminary work sugg&ed that this could be severe. Some experiments used natural seawater electrolytes, with organic material removed by prolonged irradiation with ultra violet light (HUNTER and Lms, 198 I ). The seawater was collected locally about 50 m ot%hore and had a salinity of 34.2%
Adsorptionexperimentswereconductedover a rangeof pH values through which the degree ofTb adsorption increasedfiom low values
to essentially 100%removal. A small volume of oxide suspension was added to IOmL of pH-adjustedelectmlyte in a 5@mL pyrex tlsskand allowedto equilibrateovernight,f-which Th soiution and distilled waterwere added to make a total volume of 12 mL. Tbe mixture was then allowed to equilibrate at room temperature (20 f 2’c) until emiiliium untakewasattained.Kineticexnetiments &blisbed that complete uptake of Th by the oxide was-complete aRer3hrintbecaseofgoetbite,and6hrinthecaseofCMnGa. Samples were not stirred or shaken continuously during the adsorption period, since separate experiments established that this had no e@xt on the measured uptake of Tb. After measurement of the suspe&on pH, the mixture was centrif@ at 45oog for 15 minutes. The supernatant liquid was then assayed for unads&ed Tb by a calorimetric method using the ThArsenaxoIII complex (SAWIN,1961). The detection limit for the Th assay was about 0.05 pmol/L. Possible interference from any residual Fe(III) in experiments using goethite was removed by reduction to Fe(B) with ascorbic acid.
Experimentsaimed at elucidating the effects of the major ions MI?. Ca*‘. and ti-
to&&p&
on the adsorntion of Tb bv aoetbite commued
in NaCl electrolytew&ecar&d out &ing fourseparate
synthdic electrolytes as detailed in Table 1. The ionic proportions werederivedfrom HANSSON ( 1973)for synthetic seawater. For each electrolyte, Th adsorption measurements were carried out using three different combinations of total Th and goethite so as to assess the
T&h
1.
Il.ctrolyt* solutioxu w.d for thorium ldoorption .%P.tinata
3.
4. 5.
0.422 mollk; N&l + 0.010 mol/kij Chr 0.422 mollkq NaCl + 0.028 wl/kg NarSOq U.V. irrdl~ted sem*ater
effect of changing both adsorbent and adsorbate concentrations. For goethite these combinations were: 1. 0.54 g/L goethite + 4.5 rmol/L fh 2. 0.54 g/L goethite + 9.0 pmol/L Th 3. 8.6 g/L goethite + 9.0 pmol/L Th. For bMn02, measurements were carried out mainly at two different total thorium concentrations (9 and 45 rmol/L) with a 6-MnOr concentration of 8.3 mg/L. Working concentrations of Th were chosen so as to minim& the effects of losses through adsorption onto the walls of the Pyrex ap paratus. The flask adsorption was measured both in separate experiments without oxide present, and by nitric acid leashing of the 8ask walls after adsorption by the oxide. Losses were less than 10% of total Th added. The total Th concentrations employed are eight or more orders of magnitude bigber than those found in the open ocean (BACON
and ANDERSON, 1982; LANGMUIR and I-&R~%N,[email protected] are also higherthan the theoretical solubility Emit determined by Th02 (s), allowing the possibility of Th removal from soiution by precig it&on rather than adsorption. We discount this possibility because, as described below, Th removal Born solution as a function of pH was d&rent in the presence of the two oxides, and &pen&d on the ratio of Th to oxide concentration in the case of go&&e. We suggest that uncertainties in the speciation of Tb, espe&lly Midged species, make exact solubility limits in concentrated e&eetrolytes difficult to determine. Experiments involving competing solution ligands were conducted with ethylenediamine tetra-acetic acid (EDTA), and tram-1,2-diamirwcyclohexane tetm-acetic acid (CDTA) using &MnOa as adsorbent. These were cat&d out in 0.422 mol/kg NaCl ekcvolyte at a ligand concentration of 8.3 pmol/L. Experimental procedure was similartothatusedintheotherexprimentgwiththeLigand_thorum solution being added to the suspension last of all. Sepnrate kinetic measurements showed Tb a&orption to be complete wit&n 24 hours for both EDTA and CDTA. Total thorium concentration was 9 rmol/L. RESULTS
AND DISCUSSION
The results obtained in this study were analysed by plotting the percentage of total thorium adsorbed as a titnction of pH. Each set of results (e.g. Fig. 1) showed a relatively sharp transition with increasing pH from zero to 100% adsorption, the so-called “adsorption edge.” This is characteristic of the adsorption of hydrolysable metal ions on oxides (JAMESand HEALY, 1972).
E&cf
of[Thl/[oxide]
ratio
For goethite, the position of the adsorption edge depended strongly on the oxide:metal ratio employed in all of the electrolytes. The results for the NaCl electrolyte are presented in Fig la, which shows that the adsorption edge position increased by 0.5 pH units each time the [gaethite$[Th] ratio was decreased by about an order of magnitude. Comparable pH shifts were obtained in the other three electrolytes containing magnesium, calcium and sulphate ions. BALI~TRIERI and MURRAY (1982a) found that the adsorption edge positions for Pb” and Cu*+ ions with goethite became indepen-
dent of oxide:metai ratio at values exceeding 40 moles of surface sites per mole of metal. This can be compared with a maximum-ratio of 70 achieved in our experiments. BENJAMIN and LECKIE (1980) indicate that when the adsorption edge is independent of oxide:metal ion ratio, the surface adsorntion sites are nresent in excess, with the result that ad-
Metal oxide adsorption of Th
3
5
4
6
PH
(b) kQ Effactof IlN/toxidel ratio I
1 0
OOO
629
the margins ofexperimental variability. In contrast, sulphate caused a shift of the adsorption edge of 0.5 pH units towards higher pH, i.e. a significant decrease in fractional adsorption throughout the adsorption edge region. This effect was found at all [Th]:[oxide] ratios studied. BALISTRIERI and MURRAY (1982a) found different major ion effects on the adsorption of copper, lead, zinc, and cadmium ions by goethite. Sulphate enhanced the adsorption of these metal cations, while magnesium suppressed their adsorption. The effects of the major ions with &MnOr adsorbent are illustrated in Fig. 2b which shows adsorption data at [Th] = 9 pmol/L for the four electrolytes. The results are similar to those observed with goethite. Magnesium and calcium ions had a very small effect on Th adsorption. However, sulphate ions caused a significant decrease in Th adsorption in the low pH region of the adsorption edge below about 80% adsorption. Unlike with goethite, however, the decrease caused by sulphate on this oxide became less important at higher pH where adsorption approached 100%. Adsorption in seawater
FIG. 1. Adsorption of Th as a function of pH in 0.422 md/k8 NaCl ekctrolyte at difbnt Fp[oxideJ ratios (a) Goethite,Triangle 9rmol/LTh8.6g/l~te;Cirdes:9rmd/LnZ0.54g/lgodhitc; Squarex 45 rmoVL Th, 0.54 g/L gocthitc. (b) b-Mn0~. Circks 9 rmol/L Th, 8.3 mg/L bMn@; Squares: 45 pmol/L Tk, 8.3 me/L &Mn@ .
to lower metal ion concentrations. Clearly the surface sites for Tb adsorption on goethite are not present in excess in our experiments, making constants derived from the results conditional in nature. For b-I&t@, the position of the adsorption M was nearly independent of the concentration of Th employed in all of the electrolytes. The results obtained for the NaCl electrolyte are presented in Fw 1b, which shows no significant shift in the adsorption curve when the thorium concentration was decmased by a factor of 5. This indicates that the surface adsorption sites are present in excess under the experimental conditions. The same effect was observed in the other three electrolytes. This finding means that the surface sites for thorium adsorption on this oxide are present in excess under the conditions of the experiments, so that adsorption constants derived from the experimental data may be extrapolated to lower thorium concentrations (BENJAMIN and LECKIE, 1980).
Figure 3a shows adsorption results obtained using U.V. irradiated seawater electrolyte with 9 pmol/L Th and 0.54 g/L goethite, while Fig. 3b shows comparable results for b Mn02 at 9 and 45 pmol/L Th and 8.3 mg/L oxide. The results for CMn02 at the three different total Th concentrations are essentially identical, especially in the region of the
(b)Hoi:Effectofsmmtumajoriom
The eficts of the major seawater ions The etlkcts of the electrolyte composition on Th adsorption by goethite are shown in Fig. 2a, where data for each of the four electrolytes at fixed concentrations of Th and goethite are presented. The effect of magnesium and calcium ions on the adsorption were very small, and are likely to be within
FIG. 2. Adsorption of Th as a function of pH in mixed ekctrolytes containing seawater major ions; (a) Guethite (b) 6-MnO,, Circles: NaCk Squares: NaCl + MgcI,; Triangksz Nail + C&l,; Diamonds: NaCl + Na$O,.
630
K. A. Hunter,
3
5
k
D. J. Hawke and L. K. CRoo
6
eluding any quantitative comparison with results obtamed in the presence of competing ligands. The data for Th adsorption in the presence of the competing ligands EDTA and CDTA are given in Figs. Sa and 5b. For both ligands there is a large shift of the adsorption edge to higher pH as a result of competition for Th between the surface and the solution l&and, the shift amounting to 5-6 pH units. If either the &and or Its metal ion complexes adsorb on the oxide surface, interaction of the results using competing equilibria is ambiguous and ditfrcult. We believe that neither EDTA nor CDTA adsorb significantly under the conditions of our experiments, either directly or in the form of complexes with Th. BOWERSand HUANG ( 1986) investigated adsorption of metal-EDTA complexes on a range of hydrous oxides, and found that EDTA adsorption decreased with increasing pH down to low values at a pH above 8, as expected for anion adsorption. This suggests EDTA adsorption in the region of the Th adsorption edge above pH 9 should be n~i~ble~ BOWERS and HUANG (1986) aiso showed that metal-EDTA complexes show a similar pH-adsorption dependence to EDTA itself. Moreover, our experimental results show no significant adsorption of Th in the presence of either ligand below pH 9, ruling out a significant adsorption of Th-ligand complexes.
PH FIG.3.Adsorptionof Th as a function of pH in u.\r.-irradiated seawater: (a) 0.54
g/L goethk, [Thj = 9 rmol/L (squares). (b) 8.3 n&L &MI&&.%nrare~ D’hl = 9 tunol/L: CircleszflItI= 45 rrmoll L.-k! ~~ne-~~ from a ~~~~~~ h&ding &odef. Thesolidtine shows the same model results atIer dcxmsing Th-W,- ion-pairing constant by a fmor of 10. adsorption edge. This independence of the adsorption edge on Th concentration confirms that found with the individual major ion electrolytes (Fig. 2). For both oxides, the adsorption results in U.V. irradiated seawater were not greatly ditlkent from those for the NaCl + Na$O, electrolyte, as shown by Figs. 4a and 4b in which the sulphate and seawater electrolyte results are plotted together. This is consistent with the effects of the major ions observed with the individual electrolytes. Adsorption in the presence of competing ligands The principle of using competing solution ligands in ion adsorption studies is that complexation of the metal ion in solution by the liind will reduce the free metal ion concentration available for adsorption, thereby shifting the adsorption curve to higher pH values. The magnitude of the pH shift will increase as the relative stability of the rn~-~~~nd complex increases compared to that of metal-surface complexes, and can be calculated using an equilibrium model for adsorption. This general approach has been used by VAN DEN BERG (1982) to determine Cu*+-organic matter stability constants using Mn02 as a competing adsorbent l&and. Th adsoqnion in the presence of competing organic hgands was only investigated for QMnOz , Since the goethite results depended on total Th concentration, adsorption constants derived from modelling of the results are conditional, pre-
3
4
5
6
OH IW. 4. Adsorption of Th as a function of pH in seawater and NaCl + Na$GO, clectrolytcs, [‘I%] = 9 rmol/L; (a) 0.54 S/L gorthite (b) 8.3 mg/L &Ma; Cimk seawateq IBismonk Nail + Na#O,. The dotted line is cakxlated fmm a triple-layer site bindi~ model for the NaCl + Na&I, ekctroiyte. The solid line shows the same
model results afkr &cfeasinS TM@tor of 10.
ion-pairing constant by a fsc-
Metal oxide adsorption of Th
4
6
B
lo
l2
w FIG. 5. Adsorption of 9 pmol/LTb by &-h&O3in Nat3ektrolyte in the pnana of competing @aads: (a) EDTA (FJ)CDTA Circles = no ligand added, Squaw 8.3 gtmol/L &and. The solid line is cakbtcd from a ~~~~ff site binding model for m] = 9 lrmolf L Tb and 8.3 mg/L oxide.
Because of the structural similarities between EDTA and CDTA, we expect the same conclusions to apply to the CDTA ligand as well. Application of a site-binding model
Recent work (BALIS~UERIand MURR+Y,1981, 1982a,b; DAVIS,1984; DAVISand LSXE, 1998) on the adsorption of m~ontometaloxi~afappliedtonaturalaquaticsystems has made use of the Triple Layer Model (YATESet al., 1974; DAVISd al., 1978) of the solid-solution interface, usually applied with the computer program MINEQL (WESTALLet al., 1976; DAVISet al., 1978). This model envisions the interface as consisting of three linear, parallel layers: an inner layer of surface hydmxyl groups which undergo proton exchange; a compucr layer of specifically adsorbed ions, and a di$use layer of non-speci&ally adsorbed ions. Each layer has an associated electricai potential and surface charge. Triple-layer model calculations require a number of variable parameters. These include the surface area of the solid, the concentration of surface adsorption sites (NJ and the inner and outer layer capacitances, C, and C,. In addition, equilibrium constants must be specified for all reactions taking place both in solution and at the surfhce. Some parameters can be fixed by experimental measurement. The others are then systematically varied to obtain a best-fit to the experimental adsorption data. The results must be interpreted with
631
care, since a good fit can usually be obtained by using enough variable parameters. Thus, a good fit does not necessarily mean that the model has any physical reality. For our calculations, inner and outer layer capacitance values of 140 pF/cm2 and 20 rF/cm’ were used (DAVISet al., 1978). Solution equilibrium constants used were those in the thermodynamic data file of the MLNEQL computer program, with the exception of those involving Th, which were obtained from TURNERet al. fl98lf and BAESand MESMER( 1976). Valuesfor the surface association constants for H+ and the major ions of seawater were obtained from BALISTRIERI and MURRAY(l981, 1982b). For goethite the concentration of surface sites (NJ was derived from the total titratable charge and concentration of oxide in the suspension. For &MnOr N, was initially derived from the fast tritium exchange measurements. However, model calculations with this value failed because the model always had insufficient sites to allow for complete Th adsorption, even at the lowest Th concentrations and the highest PH. Our ~cu~tions showed that site ~n~ntmtions of at least 20 mol/kg were required to obtain satisfactory agreement with experiment. There is no general agreement over what is the most reliable way of measuring the number of surface sites available for chemisorption. Measurements are difficult for acidic oxides like &Mn02 which have a high negative charge at most pH values. It is possible that our measurements may underestimate the site concentration. For the purposes of modelling, we selected the value repotied by BAJXIXERIand MURRAY ( 1982b) of 27.2 mol/kg, also measured by tritium exchange. This value for N, did give sufficient sites for 100%adsorption. There is some doubt as to the applicability of the surface association constants for H+ and seawater major ions measured by BALISTRIERJ and MURRAY(1981, 1982b) for modellihg our systems because the oxides, although synthesised by the same methods may have subtle differences in their surface properties. The concentration of surface sites N, particularly affects surface potential (JAMESet al.. 1978). However, we carried out calculations which showed that increasing or decreasing the values of the surface association constants for Na, Mg, and Ca species by upward of five orders of magnitude had no significant effect on the predicted Th adsorption. In contrast, the position of the predicted Th adsorption edge for goethite was quite sensitive to changes in the values of the association constants for sulphate and chloride with goethite, but the slope of the adsorption edge was not a&cted. Therefore, any errors in the values of these anion surface association constants were effectively incorporated into the conditionaI nature of the Th adsorption constants. This problem does not atise for &MnOr since sulphate and chloride do not adsorb on this oxide (BALISTRIERI and MURRAY, 1982b). The critical aspect of modelhng Th adsorption is to specify the different species of Th involved and their adsorption reactions. To achieve a realistic equilibrium mode1 for adsorption, it is essential that the basis set of adsorption reactions used in the calculations include all of the important reacting species and their adsorption adducts. The solution speciation for Th as a function of pH in major ion seawater, and in
632
K. A. Hunter, D. J. Hawke and L. K. Choo
0.422 mol/kg NaCI, is given in Fig. 6. In the pH range of the adsorption edge, hydrolysed fm of Th are clearly important in solution. This implies that the adsorption of Th, like that of other hydrolysable metal ions, also involves hydrolysed species (JAMESand HEALY, 1972). The hydrolysis of Th in solution can be represented by the following stepwise reactions, and equilibrium constants at zero ionic strength (BAES and MWMER, 1976): Th4+ + Hz0 + ThOH3+ + H+
log Kz = -3.20
ThOH3+ + HI0 + Th(OH):+ i- H* Th(OH):’
log *Kz = -3.73
+ Hz0 -, Th(OH)$ f H+
log *K3 = -4.77
Th(OH)f + Hz0 + Th(OH)., + H+ log
*K4
-4.20
=
(la-d)
and for which Kg, *K, , *K3, and *K4 are the respective stepwise hydrolysis constants. Each of these hydrolysed Th species, along with Th*‘, could be adsorbed through association with surface hydroxyl groups (SOH) according to the following reactions: SOH + Th4+ -) SO--Th4+ + H’
*K:
SOH + ThOH’+ -) SO--ThOH3+ C H+ SOH + Th(OH):+ --c SO--Th(OH)j+
SOH f Th(OH); * SO--Th(OH);
t H’
SOH + Th(OH)4 + SW-Th(OH),
+ H+
*k”a *K:
r2a-e)
for which *KY to *KJ are the respective intrinsic adsorption constants. In the case of 6-Mn02, the adsorption of Th was observed to be independent of Th concentration below 45 rmol/L. Therefore adsorption constants obtained by modelling may be applied to systems having lower Th concentmtions. However for goethite, the fractional adoption was not independent of total Th concentration in our experiments. Therefore only conditional constants may be obtained by modeling the experimental data, and these constants will not apply to other Th concentrations. The model calculations showed that the position of the calculated adsorption edge depended on the magnitude of the intrinsic adsorption constants, while the slope depended on the overall proton stoichiometry of the reactions involved. The model calculations presented here are confmed to a total Th inanition of 9 Fmol/L, with oxide ~n~ntmtions of 0.54 gjL for goethite or 8.3 mg/L for 6-MnOz. Modeiiing Th adsorption in NaCl electrolyte
*K;
+ H+
*K;
FIG. 6. Solution speciation of Th calculated at 9 pmol/L Th in (a) 0.422 mol/kg Nail (b) m&r ion seawater. Species of less than 1% are omitted, except that ThfOH)$+is a minor species(- 2%)between pH 5.2 and 5.6.
Initially, a series of model calculations was carried out by determining the best fit of the model to the experimental adsorption data using the assumption that, in turn, each of the individual Th hydrolysed ions was solely responsible for Th adsorption, i.e. reactions 2a-e above. This approach to dete~ining adsorption speciation is similar to that used by SANCHEZ et al. (1985). These results were then used to refine the model by incorporating mixtures of the hydroIysed species so as reproduce the position and slope of the adsorption edge. The model results for goethite are shown in Fii. 7a, which shows that the calculated adsorption of the species Th(OH)$+ has a proton stoichiometry close to that of the experimental data in the region of the adsorption edge up to about 60% adsorption. The fit at higher pH was improved by including the adsorption of higher hydrolysis products Th(OH)$ and ThfOHk , while the species Th4’ and ThOH3’ made little di@erence to the total lotion. The overall fit to the data is very good. The conditional adsorption constants for each of the hydrolysed Th species are i&ted in Table 3. Because the inclusion of other surface hydrolysed species introduces further variables, the uniqueness of the adsorption constants derived from the modelling may be questioned. Various combinations of (Th’+ + ThOH3+) and (Th(OH): f Th(OH)4) species, at low and high pH, respectively, can yield an acceptable fit to the experimental data. Further experimental measurements involving competing i&an& which shift the pH range of the lotion edge are necessary to refine the modef in the high pH region. A similar series of model calculations was carried out for &Mn02 (Fig. 7b), with different results. Adsorption of Th(OH), was required to account for the total adsorption observed at higher pH, with the lower hydrolysis products Th(OH)p and Th(OH)E being required to account for adsorption toward the lower pH end of the adsorption edge. As with goethite, the overall calcuiatecl adsorption was not improved by inciuding Th4” and ThOH’+ as adsorbing species.
633
Metal oxide adsorption of Th
3
5
4
6
PH (b) hhO$ NoClelectrolyte
adsorption reactions and ion-pairing with Th. Figures 4a and 4b show the calculated adsorption curves for both oxides in the NaCl + Na,SO, electrolyte made using the intrinsic adsorption constants derived from fitting the model to the NaCl results. The dotted curves show results obtained using Thsulphate solution ion-pairing constants taken from TURNER ef al. (198 I), values obtained from linear free energy correlations with other elements. Both curves are about 0.25 pH units to the right of the experimental data, but predict the correct slope. The discrepancy could be easily corrected by modifying slightly the adsorption constants for the Th species. However, this would remove any quantitative consistency with the model for the NaCl electrolyte. The position of the modelled adsorption edge was strongly dependent on the Thsulphate ion-pairing constant used. The solid lines in Fii. 4 show model curves calculated using values one order of magnitude lower than those estimated by TURNERet al. (198 l), and show a quite satisfactory fit to the experimental data. Obviously it is desirable to secure actual measurements of the Th-sulphate ion pairing constants in view of the importance of this interaction in seawater. iUodeliing Th adsorption in seawater
3
5
4
6
PH FIG. 7. Tb adsorption as a function of pH mlculatod using a triplelayer site binding model for [Th] = 9 pmol/L in NaCl electrolyte showing individual species(a) 0.54 g/L goethite (b) 8.3 mg/L CMnQ.
This overall fit is reasonable, but not as successful as with goethite. Modelling Th adsorption in mixed electrolyles Model calculations for thorium adsorption on both oxides with the magnesium and calcium-containing electrolytes were carried out using the conditional constants for hydrolyzed Th species already determined for the NaCl systems. These resulted in adsorption curves that were essentially identical to the calculated curves for the NaCl electrolyte. In view of the uncertainties associated with the modelling approach, and the minimal effects of both calcium and magnesium ions on Th adsorption found experimentally for both oxides, we believe that this implies a reasonable agreement between the model and experiment for calcium and magnesium effects. In the model, both of these ions did not adsorb strongly enough to compete with Th for surface sites. The experimental data for the NaCl + Na$O, electrolyte (Figs. 2a,b) shows that sulphate significantly decreases Th adsorption at low pH on both oxides. In general, this effect may result from both competition for surface adsorption sites and the formation of thoriumsulphate complexes in solution, with the exception that sulphate does not adsorb on &MnOz (BALISTRIERI and MURRAY, 1982b). The solution speciation calculations shown in Fig. 6 indicate that thorium-sulphate complexes are important in electrolytes containing sulphate. The adsorption model can allow for both of these effects through the inclusion of equilibrium constants for sulphate
The constants derived from the modelling for the adsorption of Th species are listed in Table 3 for both oxides. This set of constants was also used to model Th adsorption in seawater by including thermodynamic information for all of the seawater major ion species and their reactions in solution. The results are shown in Figs. 3a and 3b. For both oxides, the correct slope was predicted by the model, as in the case of the sulphate electrolyte. In seawater, however, the effect of changing the Th-sulphate ion pairing constants was not as great, presumably because of ion pairing of sulphate with calcium and magnesium ions. The agreement between the model and experiment is quite reasonable, both with the original ion pairing constants of TURNERet al. (1981), and with values an order of magnitude lower. Modelling Th adsorption in the presence of competing ligands As already noted, the presence of competing organic ligands caused a substantial shift of the adsorption edge for Th on &Mn@ to higher pH. This results from the decrease in fne Th in solution brought about by complexing with the ligand.
lDTA
BDTA-Th'* uDTA-TbOIP*
27.51: 21.3* 91.3'
27.6 a4.2
CDT1
CDT*-Th" CDTA-TbOlP'
31.49: 20.31: 26.0' 10.71
30.7 27.5
-.~~-....I-I.~.II..~~....~.~.~..~.....~.~~~~~.~~..~....~.. 1:
BOlTAl
end ANDSRCGa (19671
2: IlIWIIQL tbwmodynuie 3: HSN .t al 119651
data file (VCSTUL .t .I. 19761
634
K. A. Hunter. D. J. Hawke and L. K. Choo Table
3.
Conditioaul intrinsic equilibrium coneants for thorium adsorption onto gootbit~ (0.540/L) U%d (I-U801 IS.3 m&L) at 9 us&/L Th
____________________~~~~~~~~~~__~~~~~~~-~~~~-~~~~~-~~~~~___~~~~---log .B aoothitr s-U&h
1.00 -0.75 -2.38 -9.85 -16.75 9.25
-0.50 -6.00 -12.40
adsorption reaction
SOH + SOH + SOW + SOII + SO8 + SOR +
Th" Th" Th" Th" Th" Th.+
+ + + + +
Ha0 2HsO 3&O OHIO soda-
+ + . .
SO--Th" SO--ThOH" SO--ThlOH)rz' SO--ThlOH),' SO--Th(OHl. SO--ThSOd=*
+ n* + H' + 38' + 4H* + 5H' + ii+
_____________________________________-________________________-__
We made cakulations which showed that in the presence of 8.3 rmol/L of ligand and 9 pmol/L Th, as used in the present experiments, the total concentration of free, hydrolysed forms of Th in solution would be reduced to only 10-l’ mol/L by compkxing with CDTA at the pH of seawater. Similar large reductions occur with EDTA. Thus the free thorium concentration in these experiments is of the same magnitude as that found in the deep ocean (e.g. 5 X lo-” mol/kg in the deep Pacific, NOZAKI and HORIBE,1983). The ability of the adsorption model to cope with these very ditkent conditions was tested by including solution complexes formed between the liind and Th in the calculations. Because both liida shifted the adsorption edge to much higher pH, their inclusion in the model should be sensitive to the combination of Th hydrolysis products used as adsorbing species in the model, since the speciation of the hydrolysed forms is strongly pH-sensitive. The cakuiations were carried out by systematicaUy varying the stability constants of the Th-l&and solution complexes used for the calculations until the best fit of the modelled adsorption to the experimental data was obtained. No adsorption of ligand or Th-&and complexes was allowed for, and the same adsorption constants for hydrolysed thorium species were used as before (Table 3). The Th-ligand complex formation constants estimated in this manner are compared in Table 2 with available literature values measured by other techniques. The results are within the general range of the literature data. The experimental data and model calculations are compared in Fig. 4. The general success of the model in coping with the large changes in free Th concentration brought about by competitive complex formation with both CDTA and EDTA, and the consequent shift of the adsorption edge by 5-6 pH units, without having to postulate any further surface adsorption reactions, suggests that the most important adsorbing species have been included in the modeI. This underlines that competitive ligands are useful in evaluating the applicability of surface adsorption models. Extrapolation to natural marine systems The model cahzulations involving competing hgands indicate that the adsorption model adequately describes Th adsorption by &MnOz in seawater electrolytes at natural pH (7.9-8.2) and at concentrations of free Th which are comparable to those observed in the deep ocean. It is instructive,
therefore, to use the model to predict Th adsorption by ferromanganese oxide components of marine par&late matter under natural conditions. This calculation can be simplified by taking account of the major species present. At the pH of seawater (7.9-8.2), the predominant adsorbed thorium species is SO--Th(OH), , while the only important inorganic solution species is Th(OHk. Thus adsorption at seawater pH can be described adequately by the two reactions: Th’+ + 4H20 -+ Th(OH)’ + 4H+ SOH+ + Th(OH)’ -
SO--Th(OH),
+ H’.
(6)
The partition coefficient for Th bound by an adsorbing oxide phase in particles may be calculated by combining the equilibrium expressions for the two reactions in [3]: [SO-Th(OH)4] P’WW41
=
@-h-[=Hl+)
exd-#‘ko 84.
-
4$b]/kT)
(7j
W+l
where $0 and fib are the surface and Compact Layer potentials mspectively and e, k and T have heir usual meanings. The intrinsic adsorption constant for Th(OHk on &MnOr , 10%&I = - 12.40, is given in Table 3 while the solution hydrolysis constant log B4 has a value of - 15.90 (TURNER et al., 1981). As expected, the partition coeflicient of Th is proportional to the concentration of adsorbing particulate matter, in agreement with the field observations made by &CoN ami ANDERSON(1982). A value of this distribution coegkient appropriatetodeepoceanwatersmaybecaku&dbyestimating the remaining terms on the right hand side of the equation. MINEQUL calculations of the surtke speciation of bMnOr at pH 8 gave values for +. = -382 mV and tit, = 10.3 mV respectively for the electrical terms, and showed that 97% of the availabie surface sites exist as free SQH. Thus the site density for Th adsorption is 0.97 X 27.2 = 26 mol/kg of &MnOz . For an estimate of the oceanic concentration of MnOr in oceanic suspended matter, we calculated a geomek mean value of 7.8 ng Mn/L of 250 analyses of parkulate Mn in deep and intermediate waters of the AtIantic Ckean from the
635
Metal oxide adsorption of Th
compilation by BUAT-MENARD (1977). Comparison with levels of particulate Al, an indicator of detrital crustal minerals, shows that on average particulate Mn is enriched over its crustal abundance by a factor of 4.6 (BUAT-MENARD, 1977). Thus the concentration of particulate Mn in excess of that in detrital minerals, expressed as Mn% is 9.4 rig/l. Combining this with a site density of 26 mol/kg 6-Mn02 yields an adsorbing site concentration [SOH] of 2.5 X lo-” mol/L of seawater. Substitution of this value into Eon. (7), together with the appropriate values for the equilibrium constants and a deep water pH of 8.0, gave the numerical result: [So-Th(OH),]/[Th(OH),]
~5: lo-$.
(8)
This value is a tactor of up to IO’ smaller than those measured by BACON and ANDERSON(1982), e.g. for the Guatemala and Panama Basins, coefficients between 0.03 for Th-234 and 0.1 for Th-228 were observed. In coastal and estuarine waters the concentration of particulate Mn can be as high as I-10 rg/L. Under these conditions, sign&ant adsorption of Th by manganese oxides is expected on the basis of our model. These simple calculations, while clearly limited by the extrapolations and assumptions involved, suggest that manganese oxides may not be important surface scavenging phases in the deep ocean for thorium isotopes, More detailed modelling using the MINBQL program indicated that the principal reason for the low adsorption of Th at seawater pH was the development of a very large negative surface charge by the oxide. In a triple-layer model, protons interact with the surface itself while Th ions are adsorbed in the compact layer. Thus the electrosmtic attraction of the surface for Th is smaller than that for the equivalent number of protons that must be released to allow Th adsorption because the latter are closer to the surface and experience a greater potential. The iron oxide goethite does not suffer from this problem because it has a pH(PZC) close to that of seawater and does not themfore develop a large surface charge in this electrolyte. Evaluation of Bqn. (4) in a similar way for goethite using the adsorption constants for Th in Table 3 yielded a much higher estimated distribution coefficient of -0.1, within the range of values encountered in the deep ocean (BACON and ANDERSON, 1982). This estimate is uncertain because of the conditional natun of the adsorption constants for goethite determined by modelhng. However, the adsorption constants are expected to increase in magnitude as Th concentration is decreased until the adsorption sites are present in excess. Therefore the estimate of Th binding made here is, if anything, lower than expected. The result suggests, in any case, that iron oxide components may be important scavengers of thorium. Acknowledgements-We are grateful to J. Davis for supplying us with a copy of the computer program MINEQL, and for his helpful comments made while visiting our laboratory. We also thank J. Murray, R. Anderson, B. Honeyman, L. Balisuieri and one anonymous reviewerfor their constructive comments on this work Special thanks to Peter Carpenter for his help with the computing.
Editorial handling: E. R. Sholkovitz
REFERENCES ATKINSONR. J., FOSNERA. M. and
QUIRK1. P. ( 1967) Adsorption of potential detennining ions at the krric 0xidapsuecu.Sdectrdyte interface. J. Phys. Chem. 71,550-558. BACONM. P. and ANLERSONR. F. (1982) Distribution of thorium isotopes between dissolved and particulate forms in the deep sea. J. Geophys. Res. 81,2045-2056. BAESC. F. JR. and ME~MERR. E. (1976) The Hydrolysis of Cations, Wiley, New York, 489~. BALISTRIERIL. S. and MURRAYJ. W. ( 198 I ) The surfacechemistry of goethite in major ion sea water. Amer. J. Sci. Z&788-806. BALIS~UERIL. S. and MURMY J. W. (19820) The adsorption of Cu, Pb, Zn, and Cd on goethite from major ion sea water. Geochim.
Cosmochim. Acta 46, 1253-1265. BALISTRIERIL. S. and MURRAYJ. W. (1982b) The surface chemistry of delta manganese dioxide in major ion sea water. Geochim. Cos-
mochim. Acta 46, 1041-1052. Bw L. S. and MURRAYJ. W. (1983) MemI-s&d intemctions in the marine environment: estimating apparent equilibrium binding constants. Geochim. Cosmochim. Acta 47, 109I - 1098. BALISTRIERI L. S. and MURIUY J. W. (1984) Marine Scavenging: Trace metal adsorption by inter&al sediment from MANOP Site
H. Geoehim. Cosmachim. Acta 48,92 I-930. BAL~ST~IERI L S., BREWERP. G. and MURRAY J. W. (198 1) Scavenging msidence times of trace metals and surface chemistry of sinking particles in the deep ocean. Deep-Sea Res. 28, 10 I - 121. BENJAMIN M. M. and LECWEJ. 0. (1980) Adsorption of metals at oxide interfacesz effects of the concentrations of adsorbate and competing metals. In Contaminants and Sediments (ad. R. A. BAKER),pp. 305-322. Ann Arbor, Michigan. B0TTARlE. and ANDG. (1967) Die Untersuchung der 1:IKompkxe von einigen dreiund verwertigen Metall-Ionen mit polyaminocarboxykten mittels Redoxmessungen. Helv. Chim.
Acta SO,2349-2356. BOWERSA. R. and HUANGC. P. (1986)Adsorption characteristics of metal-EDTA complexesonto hydrous oxides.J. Cdloid Intetjbce Sci. I IO, 575-590. BREWERP. G. and HAO W. M. (1979) Ocean chemical scavenging. In Chemical Modellingin Aqueous Systems-@&bon. Sorption, Sohbility and Kinetics (cd. E. A. JENNE),pp. 26 l-274, A.C.S. Symposium Series 93, Amer. Chem. Sot. BUAT-MENARD P. B. (1977) Iniluence de Ia tetombce atmospber@ sur la chimie des m&aux en tmce dans la mati& en suspension de I’Atlantique Nord. Tb&sed’itat, Univers& de Paris VII, 434~. CAT'S J. G. and LANGMUIRD. (1986) Adsmtion of Cu. Pb and Zn by &MnOs : Applicability of the she bin&g-surface complexation model. Appl. Geochem. 1,255-264. DAvrsJ.A(1984).Complexationoftraametrlsby~natural organic matter. Geochim. Cosmochim. Acta 48,67969 1. DAVISJ. A. and LECKIEJ. 0. (1978) Surface ion&ion and complexation at the oxide-water interface. II. Surface properties of amorphous iron oxyhydroxide and adsorption of metal ions J.
Colloid Interfare Sci. 61,90- 107.
D~vrs J. A., JAMES R 0. and Venue J. 0. (1978) Surface ion&ion and complexation at the ox&water interface. I. Computation of electrical doubk layer properties in simple electrolytes. J. Colfoid
Interface Sci. 63,480-499. DEMPSEYB. A. and SINGERP. C. (1980) The effect of cakium on the adsorption of zinc by MnO, (s) and Fe(OHh (am). In Contaminants and Sediments (cd. R. A. BAKER),pp. 333-355. Gw R. D., CrrAkaAaAan C. L. and SCHRAMM L. L. ( 1975) The application of a simple chemical model of natural waters to metal fixation in particulate matter. Can. J. Chem. 53,66 I-669. HANSSON I. ( 1973) A new set of acidity constants for carbonic acid and boric acid in sea water. Deep& Res. 20,461-478. . HSEUT. M., WV S. F. and CHUANGT. J. (1965) The stabilities and thermodynamic proper& of thorium(Iv) and zimonium(IV) chelates of tranc I ,2diaminocyclo~N,N,N’,N-mtmaceuc acid (H,CDTA). 1. Itwrg. Nucl. Chem. 27, 1655-1662. HER K. A. ( 1983) The adsorptive properties of sinking particles in the deep ocean. DeepSea Res. 30,669-675.
636
K. A. Hunter, D. J. Hawke and L. K. Choo
HUNTERK. A. and LLSSP. S. (198 1) Polarographic measurement of surface active material in natural waters. Water Res. 15,203-2 15. JAMB R. O., DA~LSJ. A. and LECKIEJ. 0. (1978) Computer simulation of the conductimettic and potentiometric titrations of the surface groups on ion&able latexes. J. Colloid Int%ce Sci. 65, 331-344. JAMESR. 0. and HEALYT. W. (1972) Adsorption of hydrolysable metal ions at the oxide-water interface. J. Colloid Int%ce Sci. 4@,42-8 1. JAFLAMMEB. D. and MURIUV J. W. (1987) Solid/solution interaction: The efkct of carbonate alkahnity on adsorbed thorium.
Geochim. Cosmochim. Acta S&243-250. LANGMUIRD. and HERMANJ. S. (1980) The mobility of thorium in natural waters at low temperatures. Geochim. Cosmochim. Acta 44, 1753-1766. MoRG~J.J.andSTUhuwW.(1964)Thecohoid&emicalproper& of manganese dioxide. J. CoUoid Inter&e Sci. 19,347-359. MLJRRAYJ.W.(1974)ThesnrEaceche&tryofhy&ousmangam~ dioxide. J. Cdloid Inte&ce Sci. 46.357-37 1. NOZAKIY. and HORUIEY. (1983) Alpha emitting thorium isotopes in northwest P&tic deep waters. Earth Planer. Sci. I.&t. 65,39-
50. lUoz.u Y., HO?MBEY. and TSUWTA H. (1981) The water column distributions of thorium iso@pea in the we&em North P&lIc. Earth Planet. Sci. Lett. y 203-2i6. NYFFELERU. P., Lr Y. H. and SANZXX P. H. (1984) A kinetic approach to describe tmce-element distribution between particles
and solution in natural aquatic systems. Geochim. Cosmochlm.
Acta48, 1513-1522. SANCHEZA. L., MURRAYJ. W. and SIBLEYT. H. (1985) The ad-
sorption of plutonium IV and V on 8oethite. Geuchim.Cosmochim.
Acta 49,2297-2307. SAWIN S. B. (1961) Analytical use of Arsenaxo III: determination
of thorium, zirconium, uranium, and rare earth ekments. Tuknta
8,673-685. TUREKUN K. K. (1977) The fate of metals in the oceans.
Geochim
Cosmochim. Acta 41, 1139- 1144. TURNERD. R.. WHITF~ELD M. and DICKSONA. G. t I98 1) The eoui-
librium speciation of dissolved components in freshwater and &awater at 25C and 1 atm pressure. Geochim. Cosmochim. Acta 45.
855-881. VAN DEN BERGC. M. G. (1982) Determination
of copper complexation with natural organic ligands in seawater by equilibration with MnC&. II. Experimental pmcedures and application to surface seawater. Marine Chem. II, 323-342. WESXUL J., ZACHARYJ. and MORELF. (1976) MINEQL: a computer program for the calculation of chemical equilibrium cornposition of aqueous systems. In Technical Note no. 18. Ralph M. Parsons Laboratory, M.I.T. YAP D. E. and HEALYT. W. (1976) The structure of the silica/ electrolyte inter&e. J. Colloid Inte&ce Sci. 55,9- 19. YATE$D. E., LEVINES. and HEALY T. W. ( 1974) Site binding model of the dechical double layer at the oxide-water interface. .I. Chem. Sot 70, 1807-1818.
November 26, 1987)
Abstract--The equilibriumadsorptionof Th by the hydrousoxides goetbite(a-FeOOH)and &MnQIin mprineek-ctrolytesis not alkted by the majorcationsCa*+and MS?+,relativeto NaClelectrolyte,while a- deaeeuw adsorption through competitive ion pairing with Th in solution. A triple layer model of %ufm speciation in N&.3 ektrolyte qgested that adsorptionof Th mainly involves the hydrolyzedforms Th(OHp, Th(OHg and Th(OHk. Intrinsic surfPee~adi~~~~fort~~~i~~fi~~~rn~~~~~ex~rn~~~n data. In ~~~n~~~ ekctrolytcs including seawater,the model conzetiy prcdiaed the skpc of the adsorption awvcs fix both oxids, with the pH of the adsorptionedge depending on value used for Th-sulphateion-pairing equilibrium oonstants. Competition experimentsusing synthetic organicligands to bulk the Th concentrationswith &Mn~ loweredtbe concentrationof freeTh availablefor adsorptionto withia the rangeof oceanic dissolvedTh concentrations.This gave rise to shitts in the adsorptionedge to higher pH close to that of seawater,which were consistently pmdktcd by the model. Application of the model to the deep ocean watercolumn suggc&d that Shin@ would not k a pomrful scsvenga for ‘I%becauseof the developmentof a strongnegativesurfacechat8eat pH 8, whik adsorptionon 8oethite shoul&be significant. effects of carbonate alkalinity on Th adsorption by goethite
INTRODUCMON
in 0.1 M NaNOr ekctrolyte. Laboratory studies of ion adsorption have the advantage over field studies that different factors influencing the adsorption process may be systematically varied to reveal their part in the mechanism. As with earlier work on other trace metals (e.g. BALERRER~andMu~~+~, 1982a,b, 1983.1984; DAVIS, 1984), our investigation has focuJod on the thermodynamics of the adsorption ps~%, and has sought to determine the factors that infhrence the degree of Th uptake by metals oxides in seawater media. in particular, the possible competitive effects of the major ions and the influence of pH and the concentration of available surface were studied. For most trace elements, however, laboratory studies have the disadvantage that it is usually difhcuh to reproduce the very low concentrations pmaent natumlly in the ocean he-
THORIUMISOTOPES are usefi11tools for studying the removal of reactive trace elements from the ocean through scavenging by particles sinking through the water column (TIJREKIAN,
1977). The deep-sea distribution of Th isotopes is consistent with a reversible exchange of Th between dissolved and particulate forms (NOZAIUet 1, 1981; BACONand ANDERSON, 1982). The available evidence suggests that Th and other u-a@ ekments of differing reactivity underga removal from the water column by maintaining adsorptive equilibrium, with the surfaces of sinking particles (BREWERand HAO,
1979). Surface-adsorbed oqanic matter and surface coatings on f-manganese oxides, known for their adsorptive afhnity for metal ions, have been implicated as likely scavenging p~~~~e~~~~(B~~~, 1981; HUNTER,1983). Accordingty, recent attention has been focused on the adsorptive equilibria of trace metals in seawater media with either w&defined model phases such as goethite (a-F&OH)
by experimental In the pm&t work we-have addressed this shortcoming through the use of solution organic ligands which compete with the surface for Th, thereby reducing the concentration of free Th iu solution. The hgands wem chosen to reduce the concentration of free thorium a&able for adsorption, through metal ion buff&r& to kvels comparable to those measumd in deep ocean waters.
or MnOr (BALLWRDXI and MURRAY,1981,1982a,b, 1983), or natural marine particulate material obtained from scdiment traps (NYEEEEER et al., 1984) and from interfacial sediment (BAUSTRIERI and MURRAY,1984). This information is essential in providing the framework for a surface &em&d explanation for the scavenging behavior of reactive elements. In the present work we report on studies of the surface adsorption of Th onto the surfaces of two synthetic oxides, go&rite (a-F&OH) and the manganese oxide &MnOz. Goethite is a hydrous Fe(III) oxide that has been extensively studied as a model for iron oxide components of marine particulate matter (BALISTRIERI and MURRAY, 1981), while 6-MtQ believed to be similar in surface properties to the
MATERULS
ANU METHODS
Gocthite (&%OOH) was pnpand accordingto the method of ATKIIWNet al. (1967), whik b-MnOzwas prqared accordingto the method of MU-Y (1974) and BALWWERIand MURRAY (1982b). Both productswereptu%al by dialysis and suned at pH 5 in hard borosihcategkss. The oxides wtre cbaractnized to allow comparisonwith other work by severalmeasurementmetho& The @Cthiteshowedchasactainic X-raydif%ction re&aions at 4.2,2.7,awl2.4A,MdhadapH~thcpointoflnocba%e,pH(PZCX ~~~~~~~~ ~MIJRRAY (1981) of 7.1, similar to previous work. The total titratablecharge was 65 + 3 mmol of ions (as H+) per k8 of oxide. The &MnOl was poorly orckr~Iwith the most distinct refkctio~
manganese oxide components of marine particulate matter (BALISTRIERI and MURRAY, 1982bf. The surface chemistry of Th in seawater media has not previously been reported. However LAFLAMMEand MURRAY(1987) have studied the 627
628
K. A. Hunter, D. J. Hawke and L. K. Choo
being at 3.90 A, had a pH(PZC) of 1.7 * 0.3 and a B.E.T. surface area of 130 m*/g. Stoichiometric analysis by oxalate titration and atomic absorption (MURRAY,1974) gave an O:Mn ratio of 1.89 + 0.05. These results are all consistent with the results of other workers (e.g. MORGAN and -MM, 1964; MURRAY, 1974). For this oxide, the concentration of surface sites was measured by fast tritium exchange (YATESand HWLY, 1976), giving a result of 5.0 i 0.7 mol/ kg This compares favorably with a value of 8.6 md/kg determined recently by C~rrs and LANGMUIR (1986), and to values derived from adsorption studies (GVU et d, 1975; MURRAY, 1974; DE.htPSEY and SINGER,1980),but is considerably lower than a value of 27 mol/ kg measured by BALEXRIERIand MURRAY( I982b). A range of electrolytes was used in the adsorption studies to assess the effects of the seawater major ions (Table I). Potassium was not included because we expected it to be similar in behaviour to sodium and therefore not signiticant at its relatively low concentration in seawater. Results of LAFLAMMEand MURRAY (1987) suggest that
the effectsof carbonateand hicarhonateions on Th adsorptionby goethitecan he negketed at seawateraikahnities. The electrolyte solutions were prepamd from AR grade reagents in distilled, deionised, and t&red water from a “Milli+” system. Standard&&n of the calcium electrolyteswas hy EDTA titration. Th(IV) solutions were I)reDILndfromA.R. Th(NO&. All storane and e&&ally glass vessels, to control potential organic co%minati&, from plasticixers, since preliminary work sugg&ed that this could be severe. Some experiments used natural seawater electrolytes, with organic material removed by prolonged irradiation with ultra violet light (HUNTER and Lms, 198 I ). The seawater was collected locally about 50 m ot%hore and had a salinity of 34.2%
Adsorptionexperimentswereconductedover a rangeof pH values through which the degree ofTb adsorption increasedfiom low values
to essentially 100%removal. A small volume of oxide suspension was added to IOmL of pH-adjustedelectmlyte in a 5@mL pyrex tlsskand allowedto equilibrateovernight,f-which Th soiution and distilled waterwere added to make a total volume of 12 mL. Tbe mixture was then allowed to equilibrate at room temperature (20 f 2’c) until emiiliium untakewasattained.Kineticexnetiments &blisbed that complete uptake of Th by the oxide was-complete aRer3hrintbecaseofgoetbite,and6hrinthecaseofCMnGa. Samples were not stirred or shaken continuously during the adsorption period, since separate experiments established that this had no e@xt on the measured uptake of Tb. After measurement of the suspe&on pH, the mixture was centrif@ at 45oog for 15 minutes. The supernatant liquid was then assayed for unads&ed Tb by a calorimetric method using the ThArsenaxoIII complex (SAWIN,1961). The detection limit for the Th assay was about 0.05 pmol/L. Possible interference from any residual Fe(III) in experiments using goethite was removed by reduction to Fe(B) with ascorbic acid.
Experimentsaimed at elucidating the effects of the major ions MI?. Ca*‘. and ti-
to&&p&
on the adsorntion of Tb bv aoetbite commued
in NaCl electrolytew&ecar&d out &ing fourseparate
synthdic electrolytes as detailed in Table 1. The ionic proportions werederivedfrom HANSSON ( 1973)for synthetic seawater. For each electrolyte, Th adsorption measurements were carried out using three different combinations of total Th and goethite so as to assess the
T&h
1.
Il.ctrolyt* solutioxu w.d for thorium ldoorption .%P.tinata
3.
4. 5.
0.422 mollk; N&l + 0.010 mol/kij Chr 0.422 mollkq NaCl + 0.028 wl/kg NarSOq U.V. irrdl~ted sem*ater
effect of changing both adsorbent and adsorbate concentrations. For goethite these combinations were: 1. 0.54 g/L goethite + 4.5 rmol/L fh 2. 0.54 g/L goethite + 9.0 pmol/L Th 3. 8.6 g/L goethite + 9.0 pmol/L Th. For bMn02, measurements were carried out mainly at two different total thorium concentrations (9 and 45 rmol/L) with a 6-MnOr concentration of 8.3 mg/L. Working concentrations of Th were chosen so as to minim& the effects of losses through adsorption onto the walls of the Pyrex ap paratus. The flask adsorption was measured both in separate experiments without oxide present, and by nitric acid leashing of the 8ask walls after adsorption by the oxide. Losses were less than 10% of total Th added. The total Th concentrations employed are eight or more orders of magnitude bigber than those found in the open ocean (BACON
and ANDERSON, 1982; LANGMUIR and I-&R~%N,[email protected] are also higherthan the theoretical solubility Emit determined by Th02 (s), allowing the possibility of Th removal from soiution by precig it&on rather than adsorption. We discount this possibility because, as described below, Th removal Born solution as a function of pH was d&rent in the presence of the two oxides, and &pen&d on the ratio of Th to oxide concentration in the case of go&&e. We suggest that uncertainties in the speciation of Tb, espe&lly Midged species, make exact solubility limits in concentrated e&eetrolytes difficult to determine. Experiments involving competing solution ligands were conducted with ethylenediamine tetra-acetic acid (EDTA), and tram-1,2-diamirwcyclohexane tetm-acetic acid (CDTA) using &MnOa as adsorbent. These were cat&d out in 0.422 mol/kg NaCl ekcvolyte at a ligand concentration of 8.3 pmol/L. Experimental procedure was similartothatusedintheotherexprimentgwiththeLigand_thorum solution being added to the suspension last of all. Sepnrate kinetic measurements showed Tb a&orption to be complete wit&n 24 hours for both EDTA and CDTA. Total thorium concentration was 9 rmol/L. RESULTS
AND DISCUSSION
The results obtained in this study were analysed by plotting the percentage of total thorium adsorbed as a titnction of pH. Each set of results (e.g. Fig. 1) showed a relatively sharp transition with increasing pH from zero to 100% adsorption, the so-called “adsorption edge.” This is characteristic of the adsorption of hydrolysable metal ions on oxides (JAMESand HEALY, 1972).
E&cf
of[Thl/[oxide]
ratio
For goethite, the position of the adsorption edge depended strongly on the oxide:metal ratio employed in all of the electrolytes. The results for the NaCl electrolyte are presented in Fig la, which shows that the adsorption edge position increased by 0.5 pH units each time the [gaethite$[Th] ratio was decreased by about an order of magnitude. Comparable pH shifts were obtained in the other three electrolytes containing magnesium, calcium and sulphate ions. BALI~TRIERI and MURRAY (1982a) found that the adsorption edge positions for Pb” and Cu*+ ions with goethite became indepen-
dent of oxide:metai ratio at values exceeding 40 moles of surface sites per mole of metal. This can be compared with a maximum-ratio of 70 achieved in our experiments. BENJAMIN and LECKIE (1980) indicate that when the adsorption edge is independent of oxide:metal ion ratio, the surface adsorntion sites are nresent in excess, with the result that ad-
Metal oxide adsorption of Th
3
5
4
6
PH
(b) kQ Effactof IlN/toxidel ratio I
1 0
OOO
629
the margins ofexperimental variability. In contrast, sulphate caused a shift of the adsorption edge of 0.5 pH units towards higher pH, i.e. a significant decrease in fractional adsorption throughout the adsorption edge region. This effect was found at all [Th]:[oxide] ratios studied. BALISTRIERI and MURRAY (1982a) found different major ion effects on the adsorption of copper, lead, zinc, and cadmium ions by goethite. Sulphate enhanced the adsorption of these metal cations, while magnesium suppressed their adsorption. The effects of the major ions with &MnOr adsorbent are illustrated in Fig. 2b which shows adsorption data at [Th] = 9 pmol/L for the four electrolytes. The results are similar to those observed with goethite. Magnesium and calcium ions had a very small effect on Th adsorption. However, sulphate ions caused a significant decrease in Th adsorption in the low pH region of the adsorption edge below about 80% adsorption. Unlike with goethite, however, the decrease caused by sulphate on this oxide became less important at higher pH where adsorption approached 100%. Adsorption in seawater
FIG. 1. Adsorption of Th as a function of pH in 0.422 md/k8 NaCl ekctrolyte at difbnt Fp[oxideJ ratios (a) Goethite,Triangle 9rmol/LTh8.6g/l~te;Cirdes:9rmd/LnZ0.54g/lgodhitc; Squarex 45 rmoVL Th, 0.54 g/L gocthitc. (b) b-Mn0~. Circks 9 rmol/L Th, 8.3 mg/L bMn@; Squares: 45 pmol/L Tk, 8.3 me/L &Mn@ .
to lower metal ion concentrations. Clearly the surface sites for Tb adsorption on goethite are not present in excess in our experiments, making constants derived from the results conditional in nature. For b-I&t@, the position of the adsorption M was nearly independent of the concentration of Th employed in all of the electrolytes. The results obtained for the NaCl electrolyte are presented in Fw 1b, which shows no significant shift in the adsorption curve when the thorium concentration was decmased by a factor of 5. This indicates that the surface adsorption sites are present in excess under the experimental conditions. The same effect was observed in the other three electrolytes. This finding means that the surface sites for thorium adsorption on this oxide are present in excess under the conditions of the experiments, so that adsorption constants derived from the experimental data may be extrapolated to lower thorium concentrations (BENJAMIN and LECKIE, 1980).
Figure 3a shows adsorption results obtained using U.V. irradiated seawater electrolyte with 9 pmol/L Th and 0.54 g/L goethite, while Fig. 3b shows comparable results for b Mn02 at 9 and 45 pmol/L Th and 8.3 mg/L oxide. The results for CMn02 at the three different total Th concentrations are essentially identical, especially in the region of the
(b)Hoi:Effectofsmmtumajoriom
The eficts of the major seawater ions The etlkcts of the electrolyte composition on Th adsorption by goethite are shown in Fig. 2a, where data for each of the four electrolytes at fixed concentrations of Th and goethite are presented. The effect of magnesium and calcium ions on the adsorption were very small, and are likely to be within
FIG. 2. Adsorption of Th as a function of pH in mixed ekctrolytes containing seawater major ions; (a) Guethite (b) 6-MnO,, Circles: NaCk Squares: NaCl + MgcI,; Triangksz Nail + C&l,; Diamonds: NaCl + Na$O,.
630
K. A. Hunter,
3
5
k
D. J. Hawke and L. K. CRoo
6
eluding any quantitative comparison with results obtamed in the presence of competing ligands. The data for Th adsorption in the presence of the competing ligands EDTA and CDTA are given in Figs. Sa and 5b. For both ligands there is a large shift of the adsorption edge to higher pH as a result of competition for Th between the surface and the solution l&and, the shift amounting to 5-6 pH units. If either the &and or Its metal ion complexes adsorb on the oxide surface, interaction of the results using competing equilibria is ambiguous and ditfrcult. We believe that neither EDTA nor CDTA adsorb significantly under the conditions of our experiments, either directly or in the form of complexes with Th. BOWERSand HUANG ( 1986) investigated adsorption of metal-EDTA complexes on a range of hydrous oxides, and found that EDTA adsorption decreased with increasing pH down to low values at a pH above 8, as expected for anion adsorption. This suggests EDTA adsorption in the region of the Th adsorption edge above pH 9 should be n~i~ble~ BOWERS and HUANG (1986) aiso showed that metal-EDTA complexes show a similar pH-adsorption dependence to EDTA itself. Moreover, our experimental results show no significant adsorption of Th in the presence of either ligand below pH 9, ruling out a significant adsorption of Th-ligand complexes.
PH FIG.3.Adsorptionof Th as a function of pH in u.\r.-irradiated seawater: (a) 0.54
g/L goethk, [Thj = 9 rmol/L (squares). (b) 8.3 n&L &MI&&.%nrare~ D’hl = 9 tunol/L: CircleszflItI= 45 rrmoll L.-k! ~~ne-~~ from a ~~~~~~ h&ding &odef. Thesolidtine shows the same model results atIer dcxmsing Th-W,- ion-pairing constant by a fmor of 10. adsorption edge. This independence of the adsorption edge on Th concentration confirms that found with the individual major ion electrolytes (Fig. 2). For both oxides, the adsorption results in U.V. irradiated seawater were not greatly ditlkent from those for the NaCl + Na$O, electrolyte, as shown by Figs. 4a and 4b in which the sulphate and seawater electrolyte results are plotted together. This is consistent with the effects of the major ions observed with the individual electrolytes. Adsorption in the presence of competing ligands The principle of using competing solution ligands in ion adsorption studies is that complexation of the metal ion in solution by the liind will reduce the free metal ion concentration available for adsorption, thereby shifting the adsorption curve to higher pH values. The magnitude of the pH shift will increase as the relative stability of the rn~-~~~nd complex increases compared to that of metal-surface complexes, and can be calculated using an equilibrium model for adsorption. This general approach has been used by VAN DEN BERG (1982) to determine Cu*+-organic matter stability constants using Mn02 as a competing adsorbent l&and. Th adsoqnion in the presence of competing organic hgands was only investigated for QMnOz , Since the goethite results depended on total Th concentration, adsorption constants derived from modelling of the results are conditional, pre-
3
4
5
6
OH IW. 4. Adsorption of Th as a function of pH in seawater and NaCl + Na$GO, clectrolytcs, [‘I%] = 9 rmol/L; (a) 0.54 S/L gorthite (b) 8.3 mg/L &Ma; Cimk seawateq IBismonk Nail + Na#O,. The dotted line is cakxlated fmm a triple-layer site bindi~ model for the NaCl + Na&I, ekctroiyte. The solid line shows the same
model results afkr &cfeasinS TM@tor of 10.
ion-pairing constant by a fsc-
Metal oxide adsorption of Th
4
6
B
lo
l2
w FIG. 5. Adsorption of 9 pmol/LTb by &-h&O3in Nat3ektrolyte in the pnana of competing @aads: (a) EDTA (FJ)CDTA Circles = no ligand added, Squaw 8.3 gtmol/L &and. The solid line is cakbtcd from a ~~~~ff site binding model for m] = 9 lrmolf L Tb and 8.3 mg/L oxide.
Because of the structural similarities between EDTA and CDTA, we expect the same conclusions to apply to the CDTA ligand as well. Application of a site-binding model
Recent work (BALIS~UERIand MURR+Y,1981, 1982a,b; DAVIS,1984; DAVISand LSXE, 1998) on the adsorption of m~ontometaloxi~afappliedtonaturalaquaticsystems has made use of the Triple Layer Model (YATESet al., 1974; DAVISd al., 1978) of the solid-solution interface, usually applied with the computer program MINEQL (WESTALLet al., 1976; DAVISet al., 1978). This model envisions the interface as consisting of three linear, parallel layers: an inner layer of surface hydmxyl groups which undergo proton exchange; a compucr layer of specifically adsorbed ions, and a di$use layer of non-speci&ally adsorbed ions. Each layer has an associated electricai potential and surface charge. Triple-layer model calculations require a number of variable parameters. These include the surface area of the solid, the concentration of surface adsorption sites (NJ and the inner and outer layer capacitances, C, and C,. In addition, equilibrium constants must be specified for all reactions taking place both in solution and at the surfhce. Some parameters can be fixed by experimental measurement. The others are then systematically varied to obtain a best-fit to the experimental adsorption data. The results must be interpreted with
631
care, since a good fit can usually be obtained by using enough variable parameters. Thus, a good fit does not necessarily mean that the model has any physical reality. For our calculations, inner and outer layer capacitance values of 140 pF/cm2 and 20 rF/cm’ were used (DAVISet al., 1978). Solution equilibrium constants used were those in the thermodynamic data file of the MLNEQL computer program, with the exception of those involving Th, which were obtained from TURNERet al. fl98lf and BAESand MESMER( 1976). Valuesfor the surface association constants for H+ and the major ions of seawater were obtained from BALISTRIERI and MURRAY(l981, 1982b). For goethite the concentration of surface sites (NJ was derived from the total titratable charge and concentration of oxide in the suspension. For &MnOr N, was initially derived from the fast tritium exchange measurements. However, model calculations with this value failed because the model always had insufficient sites to allow for complete Th adsorption, even at the lowest Th concentrations and the highest PH. Our ~cu~tions showed that site ~n~ntmtions of at least 20 mol/kg were required to obtain satisfactory agreement with experiment. There is no general agreement over what is the most reliable way of measuring the number of surface sites available for chemisorption. Measurements are difficult for acidic oxides like &Mn02 which have a high negative charge at most pH values. It is possible that our measurements may underestimate the site concentration. For the purposes of modelling, we selected the value repotied by BAJXIXERIand MURRAY ( 1982b) of 27.2 mol/kg, also measured by tritium exchange. This value for N, did give sufficient sites for 100%adsorption. There is some doubt as to the applicability of the surface association constants for H+ and seawater major ions measured by BALISTRIERJ and MURRAY(1981, 1982b) for modellihg our systems because the oxides, although synthesised by the same methods may have subtle differences in their surface properties. The concentration of surface sites N, particularly affects surface potential (JAMESet al.. 1978). However, we carried out calculations which showed that increasing or decreasing the values of the surface association constants for Na, Mg, and Ca species by upward of five orders of magnitude had no significant effect on the predicted Th adsorption. In contrast, the position of the predicted Th adsorption edge for goethite was quite sensitive to changes in the values of the association constants for sulphate and chloride with goethite, but the slope of the adsorption edge was not a&cted. Therefore, any errors in the values of these anion surface association constants were effectively incorporated into the conditionaI nature of the Th adsorption constants. This problem does not atise for &MnOr since sulphate and chloride do not adsorb on this oxide (BALISTRIERI and MURRAY, 1982b). The critical aspect of modelhng Th adsorption is to specify the different species of Th involved and their adsorption reactions. To achieve a realistic equilibrium mode1 for adsorption, it is essential that the basis set of adsorption reactions used in the calculations include all of the important reacting species and their adsorption adducts. The solution speciation for Th as a function of pH in major ion seawater, and in
632
K. A. Hunter, D. J. Hawke and L. K. Choo
0.422 mol/kg NaCI, is given in Fig. 6. In the pH range of the adsorption edge, hydrolysed fm of Th are clearly important in solution. This implies that the adsorption of Th, like that of other hydrolysable metal ions, also involves hydrolysed species (JAMESand HEALY, 1972). The hydrolysis of Th in solution can be represented by the following stepwise reactions, and equilibrium constants at zero ionic strength (BAES and MWMER, 1976): Th4+ + Hz0 + ThOH3+ + H+
log Kz = -3.20
ThOH3+ + HI0 + Th(OH):+ i- H* Th(OH):’
log *Kz = -3.73
+ Hz0 -, Th(OH)$ f H+
log *K3 = -4.77
Th(OH)f + Hz0 + Th(OH)., + H+ log
*K4
-4.20
=
(la-d)
and for which Kg, *K, , *K3, and *K4 are the respective stepwise hydrolysis constants. Each of these hydrolysed Th species, along with Th*‘, could be adsorbed through association with surface hydroxyl groups (SOH) according to the following reactions: SOH + Th4+ -) SO--Th4+ + H’
*K:
SOH + ThOH’+ -) SO--ThOH3+ C H+ SOH + Th(OH):+ --c SO--Th(OH)j+
SOH f Th(OH); * SO--Th(OH);
t H’
SOH + Th(OH)4 + SW-Th(OH),
+ H+
*k”a *K:
r2a-e)
for which *KY to *KJ are the respective intrinsic adsorption constants. In the case of 6-Mn02, the adsorption of Th was observed to be independent of Th concentration below 45 rmol/L. Therefore adsorption constants obtained by modelling may be applied to systems having lower Th concentmtions. However for goethite, the fractional adoption was not independent of total Th concentration in our experiments. Therefore only conditional constants may be obtained by modeling the experimental data, and these constants will not apply to other Th concentrations. The model calculations showed that the position of the calculated adsorption edge depended on the magnitude of the intrinsic adsorption constants, while the slope depended on the overall proton stoichiometry of the reactions involved. The model calculations presented here are confmed to a total Th inanition of 9 Fmol/L, with oxide ~n~ntmtions of 0.54 gjL for goethite or 8.3 mg/L for 6-MnOz. Modeiiing Th adsorption in NaCl electrolyte
*K;
+ H+
*K;
FIG. 6. Solution speciation of Th calculated at 9 pmol/L Th in (a) 0.422 mol/kg Nail (b) m&r ion seawater. Species of less than 1% are omitted, except that ThfOH)$+is a minor species(- 2%)between pH 5.2 and 5.6.
Initially, a series of model calculations was carried out by determining the best fit of the model to the experimental adsorption data using the assumption that, in turn, each of the individual Th hydrolysed ions was solely responsible for Th adsorption, i.e. reactions 2a-e above. This approach to dete~ining adsorption speciation is similar to that used by SANCHEZ et al. (1985). These results were then used to refine the model by incorporating mixtures of the hydroIysed species so as reproduce the position and slope of the adsorption edge. The model results for goethite are shown in Fii. 7a, which shows that the calculated adsorption of the species Th(OH)$+ has a proton stoichiometry close to that of the experimental data in the region of the adsorption edge up to about 60% adsorption. The fit at higher pH was improved by including the adsorption of higher hydrolysis products Th(OH)$ and ThfOHk , while the species Th4’ and ThOH3’ made little di@erence to the total lotion. The overall fit to the data is very good. The conditional adsorption constants for each of the hydrolysed Th species are i&ted in Table 3. Because the inclusion of other surface hydrolysed species introduces further variables, the uniqueness of the adsorption constants derived from the modelling may be questioned. Various combinations of (Th’+ + ThOH3+) and (Th(OH): f Th(OH)4) species, at low and high pH, respectively, can yield an acceptable fit to the experimental data. Further experimental measurements involving competing i&an& which shift the pH range of the lotion edge are necessary to refine the modef in the high pH region. A similar series of model calculations was carried out for &Mn02 (Fig. 7b), with different results. Adsorption of Th(OH), was required to account for the total adsorption observed at higher pH, with the lower hydrolysis products Th(OH)p and Th(OH)E being required to account for adsorption toward the lower pH end of the adsorption edge. As with goethite, the overall calcuiatecl adsorption was not improved by inciuding Th4” and ThOH’+ as adsorbing species.
633
Metal oxide adsorption of Th
3
5
4
6
PH (b) hhO$ NoClelectrolyte
adsorption reactions and ion-pairing with Th. Figures 4a and 4b show the calculated adsorption curves for both oxides in the NaCl + Na,SO, electrolyte made using the intrinsic adsorption constants derived from fitting the model to the NaCl results. The dotted curves show results obtained using Thsulphate solution ion-pairing constants taken from TURNER ef al. (198 I), values obtained from linear free energy correlations with other elements. Both curves are about 0.25 pH units to the right of the experimental data, but predict the correct slope. The discrepancy could be easily corrected by modifying slightly the adsorption constants for the Th species. However, this would remove any quantitative consistency with the model for the NaCl electrolyte. The position of the modelled adsorption edge was strongly dependent on the Thsulphate ion-pairing constant used. The solid lines in Fii. 4 show model curves calculated using values one order of magnitude lower than those estimated by TURNERet al. (198 l), and show a quite satisfactory fit to the experimental data. Obviously it is desirable to secure actual measurements of the Th-sulphate ion pairing constants in view of the importance of this interaction in seawater. iUodeliing Th adsorption in seawater
3
5
4
6
PH FIG. 7. Tb adsorption as a function of pH mlculatod using a triplelayer site binding model for [Th] = 9 pmol/L in NaCl electrolyte showing individual species(a) 0.54 g/L goethite (b) 8.3 mg/L CMnQ.
This overall fit is reasonable, but not as successful as with goethite. Modelling Th adsorption in mixed electrolyles Model calculations for thorium adsorption on both oxides with the magnesium and calcium-containing electrolytes were carried out using the conditional constants for hydrolyzed Th species already determined for the NaCl systems. These resulted in adsorption curves that were essentially identical to the calculated curves for the NaCl electrolyte. In view of the uncertainties associated with the modelling approach, and the minimal effects of both calcium and magnesium ions on Th adsorption found experimentally for both oxides, we believe that this implies a reasonable agreement between the model and experiment for calcium and magnesium effects. In the model, both of these ions did not adsorb strongly enough to compete with Th for surface sites. The experimental data for the NaCl + Na$O, electrolyte (Figs. 2a,b) shows that sulphate significantly decreases Th adsorption at low pH on both oxides. In general, this effect may result from both competition for surface adsorption sites and the formation of thoriumsulphate complexes in solution, with the exception that sulphate does not adsorb on &MnOz (BALISTRIERI and MURRAY, 1982b). The solution speciation calculations shown in Fig. 6 indicate that thorium-sulphate complexes are important in electrolytes containing sulphate. The adsorption model can allow for both of these effects through the inclusion of equilibrium constants for sulphate
The constants derived from the modelling for the adsorption of Th species are listed in Table 3 for both oxides. This set of constants was also used to model Th adsorption in seawater by including thermodynamic information for all of the seawater major ion species and their reactions in solution. The results are shown in Figs. 3a and 3b. For both oxides, the correct slope was predicted by the model, as in the case of the sulphate electrolyte. In seawater, however, the effect of changing the Th-sulphate ion pairing constants was not as great, presumably because of ion pairing of sulphate with calcium and magnesium ions. The agreement between the model and experiment is quite reasonable, both with the original ion pairing constants of TURNERet al. (1981), and with values an order of magnitude lower. Modelling Th adsorption in the presence of competing ligands As already noted, the presence of competing organic ligands caused a substantial shift of the adsorption edge for Th on &Mn@ to higher pH. This results from the decrease in fne Th in solution brought about by complexing with the ligand.
lDTA
BDTA-Th'* uDTA-TbOIP*
27.51: 21.3* 91.3'
27.6 a4.2
CDT1
CDT*-Th" CDTA-TbOlP'
31.49: 20.31: 26.0' 10.71
30.7 27.5
-.~~-....I-I.~.II..~~....~.~.~..~.....~.~~~~~.~~..~....~.. 1:
BOlTAl
end ANDSRCGa (19671
2: IlIWIIQL tbwmodynuie 3: HSN .t al 119651
data file (VCSTUL .t .I. 19761
634
K. A. Hunter. D. J. Hawke and L. K. Choo Table
3.
Conditioaul intrinsic equilibrium coneants for thorium adsorption onto gootbit~ (0.540/L) U%d (I-U801 IS.3 m&L) at 9 us&/L Th
____________________~~~~~~~~~~__~~~~~~~-~~~~-~~~~~-~~~~~___~~~~---log .B aoothitr s-U&h
1.00 -0.75 -2.38 -9.85 -16.75 9.25
-0.50 -6.00 -12.40
adsorption reaction
SOH + SOH + SOW + SOII + SO8 + SOR +
Th" Th" Th" Th" Th" Th.+
+ + + + +
Ha0 2HsO 3&O OHIO soda-
+ + . .
SO--Th" SO--ThOH" SO--ThlOH)rz' SO--ThlOH),' SO--Th(OHl. SO--ThSOd=*
+ n* + H' + 38' + 4H* + 5H' + ii+
_____________________________________-________________________-__
We made cakulations which showed that in the presence of 8.3 rmol/L of ligand and 9 pmol/L Th, as used in the present experiments, the total concentration of free, hydrolysed forms of Th in solution would be reduced to only 10-l’ mol/L by compkxing with CDTA at the pH of seawater. Similar large reductions occur with EDTA. Thus the free thorium concentration in these experiments is of the same magnitude as that found in the deep ocean (e.g. 5 X lo-” mol/kg in the deep Pacific, NOZAKI and HORIBE,1983). The ability of the adsorption model to cope with these very ditkent conditions was tested by including solution complexes formed between the liind and Th in the calculations. Because both liida shifted the adsorption edge to much higher pH, their inclusion in the model should be sensitive to the combination of Th hydrolysis products used as adsorbing species in the model, since the speciation of the hydrolysed forms is strongly pH-sensitive. The cakuiations were carried out by systematicaUy varying the stability constants of the Th-l&and solution complexes used for the calculations until the best fit of the modelled adsorption to the experimental data was obtained. No adsorption of ligand or Th-&and complexes was allowed for, and the same adsorption constants for hydrolysed thorium species were used as before (Table 3). The Th-ligand complex formation constants estimated in this manner are compared in Table 2 with available literature values measured by other techniques. The results are within the general range of the literature data. The experimental data and model calculations are compared in Fig. 4. The general success of the model in coping with the large changes in free Th concentration brought about by competitive complex formation with both CDTA and EDTA, and the consequent shift of the adsorption edge by 5-6 pH units, without having to postulate any further surface adsorption reactions, suggests that the most important adsorbing species have been included in the modeI. This underlines that competitive ligands are useful in evaluating the applicability of surface adsorption models. Extrapolation to natural marine systems The model cahzulations involving competing hgands indicate that the adsorption model adequately describes Th adsorption by &MnOz in seawater electrolytes at natural pH (7.9-8.2) and at concentrations of free Th which are comparable to those observed in the deep ocean. It is instructive,
therefore, to use the model to predict Th adsorption by ferromanganese oxide components of marine par&late matter under natural conditions. This calculation can be simplified by taking account of the major species present. At the pH of seawater (7.9-8.2), the predominant adsorbed thorium species is SO--Th(OH), , while the only important inorganic solution species is Th(OHk. Thus adsorption at seawater pH can be described adequately by the two reactions: Th’+ + 4H20 -+ Th(OH)’ + 4H+ SOH+ + Th(OH)’ -
SO--Th(OH),
+ H’.
(6)
The partition coefficient for Th bound by an adsorbing oxide phase in particles may be calculated by combining the equilibrium expressions for the two reactions in [3]: [SO-Th(OH)4] P’WW41
=
@-h-[=Hl+)
exd-#‘ko 84.
-
4$b]/kT)
(7j
W+l
where $0 and fib are the surface and Compact Layer potentials mspectively and e, k and T have heir usual meanings. The intrinsic adsorption constant for Th(OHk on &MnOr , 10%&I = - 12.40, is given in Table 3 while the solution hydrolysis constant log B4 has a value of - 15.90 (TURNER et al., 1981). As expected, the partition coeflicient of Th is proportional to the concentration of adsorbing particulate matter, in agreement with the field observations made by &CoN ami ANDERSON(1982). A value of this distribution coegkient appropriatetodeepoceanwatersmaybecaku&dbyestimating the remaining terms on the right hand side of the equation. MINEQUL calculations of the surtke speciation of bMnOr at pH 8 gave values for +. = -382 mV and tit, = 10.3 mV respectively for the electrical terms, and showed that 97% of the availabie surface sites exist as free SQH. Thus the site density for Th adsorption is 0.97 X 27.2 = 26 mol/kg of &MnOz . For an estimate of the oceanic concentration of MnOr in oceanic suspended matter, we calculated a geomek mean value of 7.8 ng Mn/L of 250 analyses of parkulate Mn in deep and intermediate waters of the AtIantic Ckean from the
635
Metal oxide adsorption of Th
compilation by BUAT-MENARD (1977). Comparison with levels of particulate Al, an indicator of detrital crustal minerals, shows that on average particulate Mn is enriched over its crustal abundance by a factor of 4.6 (BUAT-MENARD, 1977). Thus the concentration of particulate Mn in excess of that in detrital minerals, expressed as Mn% is 9.4 rig/l. Combining this with a site density of 26 mol/kg 6-Mn02 yields an adsorbing site concentration [SOH] of 2.5 X lo-” mol/L of seawater. Substitution of this value into Eon. (7), together with the appropriate values for the equilibrium constants and a deep water pH of 8.0, gave the numerical result: [So-Th(OH),]/[Th(OH),]
~5: lo-$.
(8)
This value is a tactor of up to IO’ smaller than those measured by BACON and ANDERSON(1982), e.g. for the Guatemala and Panama Basins, coefficients between 0.03 for Th-234 and 0.1 for Th-228 were observed. In coastal and estuarine waters the concentration of particulate Mn can be as high as I-10 rg/L. Under these conditions, sign&ant adsorption of Th by manganese oxides is expected on the basis of our model. These simple calculations, while clearly limited by the extrapolations and assumptions involved, suggest that manganese oxides may not be important surface scavenging phases in the deep ocean for thorium isotopes, More detailed modelling using the MINBQL program indicated that the principal reason for the low adsorption of Th at seawater pH was the development of a very large negative surface charge by the oxide. In a triple-layer model, protons interact with the surface itself while Th ions are adsorbed in the compact layer. Thus the electrosmtic attraction of the surface for Th is smaller than that for the equivalent number of protons that must be released to allow Th adsorption because the latter are closer to the surface and experience a greater potential. The iron oxide goethite does not suffer from this problem because it has a pH(PZC) close to that of seawater and does not themfore develop a large surface charge in this electrolyte. Evaluation of Bqn. (4) in a similar way for goethite using the adsorption constants for Th in Table 3 yielded a much higher estimated distribution coefficient of -0.1, within the range of values encountered in the deep ocean (BACON and ANDERSON, 1982). This estimate is uncertain because of the conditional natun of the adsorption constants for goethite determined by modelhng. However, the adsorption constants are expected to increase in magnitude as Th concentration is decreased until the adsorption sites are present in excess. Therefore the estimate of Th binding made here is, if anything, lower than expected. The result suggests, in any case, that iron oxide components may be important scavengers of thorium. Acknowledgements-We are grateful to J. Davis for supplying us with a copy of the computer program MINEQL, and for his helpful comments made while visiting our laboratory. We also thank J. Murray, R. Anderson, B. Honeyman, L. Balisuieri and one anonymous reviewerfor their constructive comments on this work Special thanks to Peter Carpenter for his help with the computing.
Editorial handling: E. R. Sholkovitz
REFERENCES ATKINSONR. J., FOSNERA. M. and
QUIRK1. P. ( 1967) Adsorption of potential detennining ions at the krric 0xidapsuecu.Sdectrdyte interface. J. Phys. Chem. 71,550-558. BACONM. P. and ANLERSONR. F. (1982) Distribution of thorium isotopes between dissolved and particulate forms in the deep sea. J. Geophys. Res. 81,2045-2056. BAESC. F. JR. and ME~MERR. E. (1976) The Hydrolysis of Cations, Wiley, New York, 489~. BALISTRIERIL. S. and MURRAYJ. W. ( 198 I ) The surfacechemistry of goethite in major ion sea water. Amer. J. Sci. Z&788-806. BALIS~UERIL. S. and MURMY J. W. (19820) The adsorption of Cu, Pb, Zn, and Cd on goethite from major ion sea water. Geochim.
Cosmochim. Acta 46, 1253-1265. BALISTRIERIL. S. and MURRAYJ. W. (1982b) The surface chemistry of delta manganese dioxide in major ion sea water. Geochim. Cos-
mochim. Acta 46, 1041-1052. Bw L. S. and MURRAYJ. W. (1983) MemI-s&d intemctions in the marine environment: estimating apparent equilibrium binding constants. Geochim. Cosmochim. Acta 47, 109I - 1098. BALISTRIERI L. S. and MURIUY J. W. (1984) Marine Scavenging: Trace metal adsorption by inter&al sediment from MANOP Site
H. Geoehim. Cosmachim. Acta 48,92 I-930. BAL~ST~IERI L S., BREWERP. G. and MURRAY J. W. (198 1) Scavenging msidence times of trace metals and surface chemistry of sinking particles in the deep ocean. Deep-Sea Res. 28, 10 I - 121. BENJAMIN M. M. and LECWEJ. 0. (1980) Adsorption of metals at oxide interfacesz effects of the concentrations of adsorbate and competing metals. In Contaminants and Sediments (ad. R. A. BAKER),pp. 305-322. Ann Arbor, Michigan. B0TTARlE. and ANDG. (1967) Die Untersuchung der 1:IKompkxe von einigen dreiund verwertigen Metall-Ionen mit polyaminocarboxykten mittels Redoxmessungen. Helv. Chim.
Acta SO,2349-2356. BOWERSA. R. and HUANGC. P. (1986)Adsorption characteristics of metal-EDTA complexesonto hydrous oxides.J. Cdloid Intetjbce Sci. I IO, 575-590. BREWERP. G. and HAO W. M. (1979) Ocean chemical scavenging. In Chemical Modellingin Aqueous Systems-@&bon. Sorption, Sohbility and Kinetics (cd. E. A. JENNE),pp. 26 l-274, A.C.S. Symposium Series 93, Amer. Chem. Sot. BUAT-MENARD P. B. (1977) Iniluence de Ia tetombce atmospber@ sur la chimie des m&aux en tmce dans la mati& en suspension de I’Atlantique Nord. Tb&sed’itat, Univers& de Paris VII, 434~. CAT'S J. G. and LANGMUIRD. (1986) Adsmtion of Cu. Pb and Zn by &MnOs : Applicability of the she bin&g-surface complexation model. Appl. Geochem. 1,255-264. DAvrsJ.A(1984).Complexationoftraametrlsby~natural organic matter. Geochim. Cosmochim. Acta 48,67969 1. DAVISJ. A. and LECKIEJ. 0. (1978) Surface ion&ion and complexation at the oxide-water interface. II. Surface properties of amorphous iron oxyhydroxide and adsorption of metal ions J.
Colloid Interfare Sci. 61,90- 107.
D~vrs J. A., JAMES R 0. and Venue J. 0. (1978) Surface ion&ion and complexation at the ox&water interface. I. Computation of electrical doubk layer properties in simple electrolytes. J. Colfoid
Interface Sci. 63,480-499. DEMPSEYB. A. and SINGERP. C. (1980) The effect of cakium on the adsorption of zinc by MnO, (s) and Fe(OHh (am). In Contaminants and Sediments (cd. R. A. BAKER),pp. 333-355. Gw R. D., CrrAkaAaAan C. L. and SCHRAMM L. L. ( 1975) The application of a simple chemical model of natural waters to metal fixation in particulate matter. Can. J. Chem. 53,66 I-669. HANSSON I. ( 1973) A new set of acidity constants for carbonic acid and boric acid in sea water. Deep& Res. 20,461-478. . HSEUT. M., WV S. F. and CHUANGT. J. (1965) The stabilities and thermodynamic proper& of thorium(Iv) and zimonium(IV) chelates of tranc I ,2diaminocyclo~N,N,N’,N-mtmaceuc acid (H,CDTA). 1. Itwrg. Nucl. Chem. 27, 1655-1662. HER K. A. ( 1983) The adsorptive properties of sinking particles in the deep ocean. DeepSea Res. 30,669-675.
636
K. A. Hunter, D. J. Hawke and L. K. Choo
HUNTERK. A. and LLSSP. S. (198 1) Polarographic measurement of surface active material in natural waters. Water Res. 15,203-2 15. JAMB R. O., DA~LSJ. A. and LECKIEJ. 0. (1978) Computer simulation of the conductimettic and potentiometric titrations of the surface groups on ion&able latexes. J. Colloid Int%ce Sci. 65, 331-344. JAMESR. 0. and HEALYT. W. (1972) Adsorption of hydrolysable metal ions at the oxide-water interface. J. Colloid Int%ce Sci. 4@,42-8 1. JAFLAMMEB. D. and MURIUV J. W. (1987) Solid/solution interaction: The efkct of carbonate alkahnity on adsorbed thorium.
Geochim. Cosmochim. Acta S&243-250. LANGMUIRD. and HERMANJ. S. (1980) The mobility of thorium in natural waters at low temperatures. Geochim. Cosmochim. Acta 44, 1753-1766. MoRG~J.J.andSTUhuwW.(1964)Thecohoid&emicalproper& of manganese dioxide. J. CoUoid Inter&e Sci. 19,347-359. MLJRRAYJ.W.(1974)ThesnrEaceche&tryofhy&ousmangam~ dioxide. J. Cdloid Inte&ce Sci. 46.357-37 1. NOZAKIY. and HORUIEY. (1983) Alpha emitting thorium isotopes in northwest P&tic deep waters. Earth Planer. Sci. I.&t. 65,39-
50. lUoz.u Y., HO?MBEY. and TSUWTA H. (1981) The water column distributions of thorium iso@pea in the we&em North P&lIc. Earth Planet. Sci. Lett. y 203-2i6. NYFFELERU. P., Lr Y. H. and SANZXX P. H. (1984) A kinetic approach to describe tmce-element distribution between particles
and solution in natural aquatic systems. Geochim. Cosmochlm.
Acta48, 1513-1522. SANCHEZA. L., MURRAYJ. W. and SIBLEYT. H. (1985) The ad-
sorption of plutonium IV and V on 8oethite. Geuchim.Cosmochim.
Acta 49,2297-2307. SAWIN S. B. (1961) Analytical use of Arsenaxo III: determination
of thorium, zirconium, uranium, and rare earth ekments. Tuknta
8,673-685. TUREKUN K. K. (1977) The fate of metals in the oceans.
Geochim
Cosmochim. Acta 41, 1139- 1144. TURNERD. R.. WHITF~ELD M. and DICKSONA. G. t I98 1) The eoui-
librium speciation of dissolved components in freshwater and &awater at 25C and 1 atm pressure. Geochim. Cosmochim. Acta 45.
855-881. VAN DEN BERGC. M. G. (1982) Determination
of copper complexation with natural organic ligands in seawater by equilibration with MnC&. II. Experimental pmcedures and application to surface seawater. Marine Chem. II, 323-342. WESXUL J., ZACHARYJ. and MORELF. (1976) MINEQL: a computer program for the calculation of chemical equilibrium cornposition of aqueous systems. In Technical Note no. 18. Ralph M. Parsons Laboratory, M.I.T. YAP D. E. and HEALYT. W. (1976) The structure of the silica/ electrolyte inter&e. J. Colloid Inte&ce Sci. 55,9- 19. YATE$D. E., LEVINES. and HEALY T. W. ( 1974) Site binding model of the dechical double layer at the oxide-water interface. .I. Chem. Sot 70, 1807-1818.