Sorption of Eu(III) on a Natural Hematite: Application of a Surface Complexation Model

JOURNAL OF COLLOID AND INTERFACE SCIENCE ARTICLE NO.

208, 153–161 (1998)

CS985788

Sorption of Eu(III) on a Natural Hematite: Application of a Surface Complexation Model Thomas Rabung,* Horst Geckeis,†,1 Jae-Il Kim,† and Horst Philipp Beck* *Universita¨t des Saarlandes, Anorganische und Analytische Chemie und Radiochemie, P.O. Box 15 11 50, D-66041 Saarbru¨cken, Germany; and †Forschungszentrum Karlsruhe, Institut fu¨r Nukleare Entsorgungstechnik, P.O. Box 3640, D-76021 Karlsruhe, Germany Received February 23, 1998; accepted August 6, 1998

The solid/water interface reaction is investigated for the sorption of the Eu31 ion onto a well-characterized natural hematite at pH <6 in 0.1 mol/L NaClO4. Sorption isotherms are determined at different pH. The sorption of Eu31 onto hematite at coverages <1% of the surface hydroxyl groups follows the ideal Nernst behavior and can be interpreted in terms of the surface complexation by the following equilibrium reaction: 'XOH 1 Eu31 N 'XOEu21 1 H1 (log Kint 5 2.50 6 0.19). Complexation conx stants are calculated by either graphically interpretating the sorption isotherms or fitting the data using the code FITEQL. At higher surface coverages a deviation from the Nernst behavior is observed which can be described satisfactorily by a two-site surface complexation model, taking into account the lateral electrostatic interaction of surface-bound Eu(III): 'YOH 1 Eu31 N 'YOEu21 1 H1 (log Kint 5 20.82 6 0.11). The experimental y results are compared with data published in the literature, and the interpretation of sorption isotherms with regard to the sorption mechanisms is critically discussed. © 1998 Academic Press Key Words: adsorption isotherms; surface complexation modeling; europium; hematite.

INTRODUCTION

The solid/water interface reactions of radionuclides have attracted much interest in conjunction with the long-term performance assessment of nuclear waste repositories (1, 2). In general, experimental K d values providing a conditional description of the sorption behavior of radionuclide ions are still applied in performance assessment calculations in order to describe retardation processes at surfaces (3). But, for obvious reasons, the general applicability of a K d value has been brought into question (4). A reliable predictive transport modeling of actinides and long-lived fission products in aquifer systems entails thermodynamic data and mechanistic understanding on sorption reactions at mineral surfaces. Surface complexation models (SCM) (9, 11) have been developed in order to allow a thermodynamic description of sorption reactions and have been successfully applied to the prediction of the migration of mostly divalent metal cations. Studies on tri1

and tetravalent ions which can be related to the sorption chemistry of actinide ions are rarely found in the literature. In general, SCM approaches still suffer from a lack of understanding about the surface species formed. Important parameters implemented into the individual models are mostly obtained by fitting titration data. As a consequence, various surface complexation models assuming different surface complexation species including different assumptions for the description of the surface potential, may be fitted into the same experimental data (5). In order to study the sorption mechanisms on a molecular scale, various spectroscopic methods have been applied. Especially X-ray absorption spectroscopy (XAS) has been found to be a promising tool for the identification of surface sorbed species and, thus, to prove the assumptions underlying the individual thermodynamic surface complexation models (6). The aim of this work is to check the applicability of the SCM as a thermodynamic concept for the interpretation of radionuclide sorption onto natural minerals. A model system is investigated consisting of Eu(III) used as a chemical homologue for trivalent actinides and a natural hematite. Ferric oxides and hydroxides are present in natural aquifers and known for their remarkable properties as sorbents for cations. For the modeling of the sorption reaction, we attempt to keep the system as simple as possible and to restrict the number of adjustable model parameters to a minimum. In order to come close to a “mechanistic” understanding of the sorption process various argumentations are given dealing mainly with the stoichiometry of the assumed surface reaction. For the unambiguous identification of surface-sorbed species and of mechanistic processes at the solid/liquid interface spectroscopic methods like XAS and laser fluorescence spectroscopy will be used in future work. In another paper (7) the SCM constants presented in this work are applied to the prediction of the Eu(III) behavior in the presence of anionic ligands.

To whom correspondence should be addressed. 153

MATERIALS AND METHODS

Natural hematite is ground to a grain size ,100 mm. The grain diameter shows a broad distribution with a maximum at 0021-9797/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.

154

RABUNG ET AL.

2.2 mm. The elemental composition is analysed by X-ray fluorescence (SRS-200, Fa. Siemens, Germany) and the mineral composition is characterized by X-ray diffraction analysis (XRD 3000, Fa. Seifert, Germany). No carbonate has been detected by analyzing the total inorganic carbon (TIC , 100 mg/g) content of the mineral. The surface area determined by the BET method using nitrogen is found to be 4 m2/g. All chemicals used are of analytical grade. Acidity constants of the surface hydroxyl groups and their total concentration are determined by potentiometric titration under Ar atmosphere in a glove box. For each titration 2.2 g of hematite is suspended in 50 mL of 0.001, 0.01, or 0.1 mol/L NaClO4. pH is measured by a Ross type electrode and adjusted by addition of either CO2free 0.1 mol/L NaOH or 0.1 mol/L HClO4. The equilibrium value is taken when showing a drift less than 0.01 units per minute. Sorption experiments are carried out batchwise under Ar atmosphere using a turnover mixer. Experiments under aerobic or anaerobic conditions show no difference in the Eu(III) distribution at pH #6. This fact meets with approval of our assumption that the carbonate complexation can be neglected in the pH range investigated (see discussion below). For each batch experiment 100 mg hematite is suspended in 15 mL 0.1 mol/L NaClO4 solution adjusted to pH 4.0, 5.0, 5.5, or 6.0 by buffering with 2-(N-morpholino)ethanesulfonic acid (MES) and b-(4-pyridyl)ethanesulfonic acid (PES). A preliminary study shows that an equilibrium of the Eu(III) sorption is established after 48 h. The Eu(III) concentration is determined after separation of the solid phase by centrifugation at 5000 rpm by ICP-AES, ICP-MS or by g-spectrometry for Eu-154 used as a tracer. Reversibility is confirmed by separating the solid after sorption and putting into contact with fresh NaClO4 solution at given pH. The Eu(III) concentration then is determined after 14 days. Surface acidity and complexation constants are calculated from the titration and sorption data using the code FITEQL (8). THEORY

A simple formalism for the description of sorption reactions is given by the Langmuir equation. The sorption of Eu(III) onto a theoretical surface covered homogeneously with the surface sites S then is represented by the equation: @SEu# 5

@ST#K ads@Eu# , 1 1 K ads@Eu#

@SEu# , @ST#

Q 5 K ads@Eu#. 12Q

[2]

[3]

The slope gives the adsorption constant K ads, which describes the sorption reaction: S 1 Eu N SEu,

K ads 5

@SEu# . @SF#@Eu#

[4]

Although this type of adsorption constant gives no insight into the underlying reactions concerned and is only valid for the specific experimental conditions, it leads to a descriptive modeling of experimental results. At very low surface coverages (1 2 Q ' 1), Eq. [3] is reduced to the Nernst equation, i.e., under these conditions the K d concept can be applied. Unfortunately, experimental sorption isotherms rarely follow the simple Langmuir equation, and usually it is necessary to combine several Langmuir isotherms. Opposite to the Langmuir formalism the SCM approach aims to the thermodynamic modeling of a complexation reaction with the surface hydroxyl groups ('SOH) of the hydrated metal oxide. Their amphoteric behavior and apparent acid– app base constants K app a1 and K a2 can be determined by potentiometric titration for the reactions (9): app 'SOH 1 H1 N 'SOH1 2 , K a1 ,

[5]

'SO2 1 H1 N 'SOH, K app a2 .

[6]

The resulting proton-induced surface charge Q can be calculated from titration data as a function of pH: Q5

2 @'SOH1 Ca 2 Cb 1 @OH2# 2 @H1# 2 # 2 @'SO # 5 , [7] m m

2 @H1#tot 5 @H1# 2 @OH2# 1 @'SOH1 2 # 2 @'SO # 5 Ca 2 Cb,

[8]

[1]

where [ST] is the total site concentration at the mineral surface (mol/kg), [Eu] is the Eu(III) concentration in solution (mol/L), and [SEu] is the sorbed Eu(III) concentration (mol/kg). The surface coverage Q is equal to: Q5

where [ST] 5 [SF] 1 [SEu]. [SF] corresponds to the free surface site concentration, which is accessible to the Eu(III) sorption. Q can be related to the free Eu(III) concentration according to:

where C a is the concentration of added acid (in mol/L), C b is the concentration of added base (in mol/L), and m is the mass of hematite in (g/L). A surface potential f is calculated from Q using the constant capacitance model (9) by introducing the capacitance of the surface double layer C: Q 5 Cf.

[9]

155

EU(III) SORPTION ON HEMATITE

FIG. 1. Titration curves for hematite suspensions at I 5 0.001, 0.01, and 0.1 mol/L NaClO4 (hematite concentration 5 44 g/L); dotted, dashed, and drawn lines correspond to the calculated values of Q plotted as a function of pH.

The constant capacitance model represents a simple model for the surface potential which requires only a low number of adjustable parameters and which can be applied to the relatively high ionic strength conditions of our investigations. Intrinsic values of K, i.e., acidity constants independent of pH, can be determined by fitting titration data to the model: log K app 5 log K int 1 F f /~2.3RT!.

[10]

Following the SCM approach suggested by Schindler and Stumm (9), which postulates two different surface sites, “strong” ('XOH) and “weak” ('YOH), the complexation reaction of Eu31 ions with the surface hydroxyl groups is represented by the following reactions: 1

1H ,K ,

[11]

'YOH 1 Eu31 N 'YOEu21 1 H1, Kapp y .

[12]

'XOH 1 Eu

31

N 'XOEu

21

app x

In order to take Coulombic interactions at the mineral surface into account, intrinsic surface complexation constants K int x,y can be calculated analogously to Eq. [10]. RESULTS AND DISCUSSION

for the generally lower pHpzc values for natural hematite is probably due to impurities involved as compared to synthetic materials of relatively higher purity. The amounts of inorganic elements in the natural hematite as analysed by XRF can be taken from Table 1. It should be noted that ground material is investigated in this study, so that the bulk composition corresponds roughly to that of the grain surface. Despite their relatively low concentration, the impurities might alter the general surface properties of the pure mineral, even though the XRD analysis is not sensitive enough to detect other minerals than hematite. The presence of impurities, like Al, Si, and As, might be responsible for the specific acid– base properties of the hematite surface and the lower pHpzc value. The code FITEQL is used to fit experimental data and C, int K int a1 , K a2 , and the total concentration of surface hydroxyl groups ['SOH]tot are obtained as fit parameters (Table 2) for three ionic strengths. Values for the WSOS/DF (weighted sum of squares divided by the degree of freedom (8)) provided by the FITEQL code as a measure for the goodness of fit are given in Table 2. Westall (8) notes that values between 0.1 and 20 indicate a reasonably good fit. Only for the titration data at I 5 0.001 mol/L a slightly higher value for WSOS/DF is obtained, which might be due to the fact that the validity of the constant capacitance model decreases with decreasing ionic strength. Only the data calculated from the titration data at I 5 0.1 mol/L NaClO4 are used for the subsequent SCM calculations. As is apparent from the data in Fig. 1 and Table 2, the influence of ionic strength on Q is quite small, inferring that the influence of surface charge is moderate, and hence it might also play a minor role for the sorption of Eu(III) onto hematite, especially in the pH range of the present experiments. Sorption Studies The Eu(III) uptake is measured as a function of pH for two concentration ratios, R 5 [Eu(III)]tot/['SOH]tot (R 5 0.04 and 0.4) (Fig. 2). A pronounced rise of the Eu(III) sorption from almost zero to about 100% within a range of about 2 pH units is observed, which is typical for many examples of metal ion sorption onto metal oxide surfaces. The shift of the adsorption “edge” to higher pH at the higher Eu(III) concentration indicates sorption behavior that is different from an ideal Nernst type adsorption, where the uptake is independent of the sorbate concentration. From Fig. 3 it can be seen that such an ideal Nernst adsorption can only be observed at rather low

Characterization of the Solid Phase Titration results for the investigated hematite at different ionic strength are shown in Fig. 1, where Q is plotted as a function of pH. The lines in Fig. 1 correspond to the calculated curves obtained at different ionic strengths and show a common intersection point at the point of zero charge, pHpzc ' 6.0 6 0.2. For synthetic hematite, a range of pHpzc 5 6.5–9.04 is found in the literature (10, 11), while for natural hematite, lower values (5.4 – 6.9) are determined (10). The main reason

TABLE 1 Elemental Composition of the Natural Hematite Analyzed by XRF (Limit of Detection for Other Elements Approximately 0.01 Mass %)

Mass %

SiO2

Al2O3

Fe2O3

TiO2

CaO

P2O5

As2O5

Sb2O5

1.78

0.27

97.6

0.02

0.04

0.06

0.09

0.07

156

RABUNG ET AL.

TABLE 2 Surface Acidity Constants, Number of Surface Hydroxyl Groups, and Capacitance as Calculated by FITEQL Using the Constant Capacitance Model NaClO4

log K int a1 log K int a2 C (F/m2) ['SOH]tot/(mol/g)a ['XOH]tot/(mol/g) WSOS/DF a

0.1 mol/L

0.01 mol/L

0.001 mol/L

4.80 6 0.02 7.49 6 0.02 2.24 (6.95 6 0.14) 3 1026 (2.64 6 0.06) 3 1027 7.15

4.72 6 0.02 7.75 6 0.02 2.24 (6.77 6 0.16) 3 1026 – 16.2

4.70 6 0.02 7.99 6 0.02 2.24 (6.39 6 0.17) 3 1026 – 23.5

2 ['SOH]tot 5 ['SOH1 2 ] 1 ['SO ] 1 ['SOH].

Eu(III) concentration, where a slope of 1 is found in the log–log plot of the isotherm data. The sorption isotherms in Fig. 3 are determined over several orders of magnitude of free Eu(III) concentration from 10210 to 1024 mol/L at four different pH values. Different to the “pH edge” experiments buffer substances have been used in these experiments. As the results of both experiments are consistent (see Discussion below), their influence on the sorption reaction can be considered negligible.

FIG. 2. (a) pH dependence of Eu31 sorption onto hematite at different ratios R 5 [Eu(III)]tot/['SOH]tot (R 5 0.04 and 0.4); solid lines are the result of model calculation. (b) Eu(III) speciation in solution under atmospheric conditions ( p(CO2) 5 0.031%); data taken from Ref. (14).

Usually, one would expect surface sites being gradually saturated with increasing the metal ion concentration and having an inclination toward zero slope of the isotherm (Langmuir isotherm). However, the experimental data deviate from the Langmuir shape. Such behavior is consistent with other investigations and also with a recent study on the sorption of trivalent elements onto silica (12). One explanation can be the presence of at least two surface sites of different complexation strength, so-called “strong” and “weak” sites according to their complexation strength (9, 11). At a low surface coverage the strong site complexation prevails, while at a higher coverage the surface complexation occurs at weak sites as well. Following this model assumption, the shape of the sorption isotherm can be described by combining at least two Langmuir isotherms. The sorption behavior may be also explained by the occurrence of surface precipitation at higher surface coverage (13). In that case the formation of a solid solution at the sorbent

FIG. 3. Sorption isotherms for Eu31 onto hematite at pH 4.0, 5.0, 5.5, 6.0 (I 5 0.1 mol/L NaClO4); drawn lines are the result of model calculations using an average of complexation constants determined by different methods (see text); data points obtained in reversibility studies are included (pH 6.0 rev., 5.0 rev.).

EU(III) SORPTION ON HEMATITE

157

surface is assumed with the fraction of incorporated metal ion increasing continuously with the sorbate concentration from 0 to 100%. According to that model the dissolved metal ion concentration gradually increases up to the thermodynamic solubility of the corresponding metal hydroxide. In Fig. 3 the results of reversibility studies for different Eu(III) concentrations at pH 5.0 and 6.0 are included. After 14 days the distribution of Eu(III) between solid and liquid by desorption has almost reached the values obtained by sorption experiments equilibrated for 48 h. Similar results are found for low and high surface coverages at different pH. This implies that at low surface coverage and in the nonideal sorption range at higher Eu(III) concentrations, a reversible sorption process takes place. Isotherm Description by the Langmuir Formalism Relevant actinide concentrations in the far field of a nuclear repository will certainly lie in the trace concentration range ,1028 mol/L, usually the range where natural actinides and rare earth elements are found in natural groundwaters. From Fig. 3 it follows that the Eu(III) sorption at such low concentrations follows ideal behavior in the investigated model system, which can be described by a Langmuir isotherm. The total accessible surface site concentration under the given experimental conditions is estimated by a simple linear extrapolation of the isotherms slopes in Fig. 3 at low and high Eu(III) concentrations (Fig. 4a). The intersection point of both lines can be taken as a number of total sites [ST] responsible for the Nernst type part of the sorption isotherm. [ST] is found to increase with increasing pH. Using these values for [ST], the sorption constants K ads depending on pH can be calculated according to Eq. [3] for the ideal sorption range (Fig. 4b). Results are given in Table 3 and are found to increase by several orders of magnitude with increasing pH. The adsorption constants can be transformed to a chemical surface complexation constant for the “strong” site complexation according to the definition in Eq. [11]. Values for [ST] are assumed to be an estimate for the available concentration of ['XOH] groups at given pH. The transformation of K ads into K app is then given x by Eq. [13]: log K *ads 5 log K app x 5 log K ads 2 pH.

[13]

The log K *ads values obtained from sorption isotherms (Eq. [3]) by applying such a simple graphical extrapolation method are quite consistent with the log K x values calculated according to the SCM approach discussed below (Table 3). The latter procedure requires pK s values of the surface sites obtained by fitting potentiometric titration data. The graphical method delivers [ST] in a very simple way without using fitted parameters such as pK a1 and pK a2. The independence of the estimated log K *ads values on pH can be taken as a confirmation of the

FIG. 4. (a) Graphical determination of [ST] and (b) calculation of K ads according to the Langmuir equation [3] for sorption experiments at pH 6.0.

stoichiometry of the assumed surface complexation reaction in Eq. [11]. Surface Complexation Modeling (SCM) A prerequisite for a proper chemical modeling of the surface complexation reaction is the knowledge on the aqueous speciation of Eu(III) under the experimental conditions. The pH range for the sorption experiments has been limited to #6.0 and to a relatively low Eu(III) concentration. A speciation calculation taking the data of Runde (14) (log b11(AmOH21) 1 5 6.3; log b12(Am(OH)1 2 ) 5 12.2; log b101(Eu(CO3) ) 5 2 6.75; log b102(Eu(CO3)2 ) 5 12.26) shows that at pH #6.0 the Eu(III) aquo complex prevails as dissolved species and neither carbonate complexation nor hydrolysis play an important role. Less than 5% of the total Eu(III) is present as EuOH21 at pH 6.0. A species plot for Eu(III) under atmospheric conditions is shown in Fig. 2. Therefore, it is reasonable to assume that only the surface complexation of the Eu31 aquo ion is observed under these conditions.

158

RABUNG ET AL.

TABLE 3 Surface Complexation Constants for Eu31 with Natural Hematite (I 5 0.1 mol/L NaClO4) Calculated from Sorption Experiments (Index x, Strong Sites; Index y, Weak Sites) pH

4.0

5.0

5.5

6.0

Mean values

Surface complexation constants calculated using FITEQL from isotherm data log K int x log K int y

2.52 –

2.18 20.94

2.60 20.81

2.43 20.72

2.43 6 0.18 20.82 6 0.11

Sorption constants calculated using the graphical method log K ads (Eq. [3]) log K *ads (Eq. [13])

6.53 2.53

7.35 2.35

8.33 2.83

8.55 2.55

2.57 6 0.20

Grand average values used for calculation log K int x 5 2.50 6 0.19 log K int y 5 20.82 6 0.11

A monodentate complexation is assumed in Eqs. [11] and [12]. As shown in Figs. 2 and 3, fitting experimental data to the given complexation reactions reveals that deviations appear only marginal, suggesting the prevailing monodentate complexation. The drawn lines are calculated using the averaged complexation constants of all experiments and the fitted concentration of strong and weak sites (see Table 2 and 3). The calculated curve in Fig. 2 for pH values .6.0 is plotted as a dotted line because hydrolyzed Eu(III) species and carbonate complexes are not negligible under these conditions but are not included into the calculation. Congruent results for the strong site complexation constant at different pH are obtained without invoking an electrostatic correction term. This can be explained by the potentiometric pH titration data of hematite. A small effect induced by the surface charge is observed in the pH range of our experiments and also small influence of ionic strength on the protonation reaction (Fig. 1). In the pH range of 4.0 – 6.0, the surface charge contribution to the Eu(III) sorption at low Eu(III) concentrations can be also considered as negligible. A possible lateral Coulombic interaction of surface sorbed metal ions disappears at low surface coverage (15). The similar effects are observed recently for the sorption of Np(V) onto argyllite, where a monodentate surface complexation without electrostatic SCM is found adequate to describe experimental data (16). Only small contribution of the surface charge effect on the sorption of Pb(II) onto goethite is reported by Lu¨tzenkirchen and Behra (17). It is well known that strongly adsorbing ions undergo inner-sphere complexation and thereby the chemical complexation contribution to the free energy of sorption exceeds the electrostatic contribution (5, 18). As the Eu(III) sorption edge occurs at pH , pHpzc, i.e., the sorption takes place against the electrostatic repulsion of the positively charged hematite surface, it becomes obvious that the formation of inner-sphere complexes like the Eu(III) sur-

face complexation is least influenced by the surface potential is equal to log K int effect. As a consequence, log K app x x . However, at higher surface coverages (“weak” site complexation), fitted curves deviate from experimental data, when no electrostatic term is considered. The inclusion of the surface charge effect for lateral interactions of 'YOEu21 species at higher coverages yields a better congruency between fitted and measured data (Fig. 3). The values for c(Eu sorbed) are calculated from solution concentrations only and, therefore, are afflicted by uncertainties especially at high surface coverages, where two large figures are subtracted from each other. This might explain to some extent why log K y values for the “weak” site complexation derived from such a fit vary somewhat with pH. For calculation as shown in Figs. 2 and 3 we used an average value for log K y from all experiments (Table 3). The necessity of postulating additional repulsive forces between trivalent element species sorbed at the solid–water interface, which contribute to the activity coefficient of these species has also been stated by Kosmulski in his study on the Gd(III) sorption onto silica (12). The impossibility of deriving the mechanism of a sorption process from isotherm data alone has been stated very clearly by Sposito (19). The assumed sorption reaction, however, can be supported by looking at the net proton exchange per sorbed Eu(III) ion, which in the present case should be equal to 1. Rearrangement of Eq. [11] yields the relation:

log

@'XOEu21# @'XOH# 5 log K int . x 1 log @Eu31# @H1#

[14]

The concentration of undissociated strong binding surface hydroxyl groups ['XOH] is calculated from intrinsic acid– base int constants K int a1 and K a2 (Eqs. [5] and [6]) and the total concentration of the strong binding surface hydroxyl groups ['XOH]tot: @'XOH# 5

@'XOH# tot 2 @'XOEu21# , int 1 1 1 ~@H1#/K int a1 ! 1 ~K a2 /@H #!

[15]

2 @'XOH# tot 5 @'XOH1 2 # 1 @'XO #

1 @'XOH# 1 @'XOEu21#.

[16]

The electrostatic term is disregarded in Eq. [14] because it is found to be negligible in the pH range 4 – 6. Log(['XOEu21]/ [Eu31]) is plotted versus log(['XOH]/[H1]) according to Eq. [14] for the isotherm data at low surface coverage, where only the “strong site” complexation is considered (Fig. 5). A linear relation with a slope of 0.98 6 0.14 indicates that the stoichiometry of Eq. [11] is valid and thus verifies the assumption for the monodentate complex formation. The intersection of the extrapolated line in Fig. 5 with the y axis gives log K int x according to Eq. [14], which is found to be 2.44 6 0.04 and agrees quite well with the values given in Table 3. The same

EU(III) SORPTION ON HEMATITE

159

however, proved the occurrence of exclusively monomeric surface-sorbed U(VI) species under these conditions (26, 27). These somewhat contradictory results make clear that the surface precipitation reaction might not always be a relevant sorption mechanism and that more work is necessary to explain the nonideal shape of sorption isotherms. Several considerations support the idea that in the present study the surface heterogeneity should be the more relevant reason for the nonideal sorption behavior at higher surface coverage, even though a final decision concerning the sorption mechanism presently cannot be made:

FIG. 5. coverage.

Function plot of Eq. [14] for isotherm data at low surface

relationship is found for the weak site complexation range at high surface coverage. Due to the nonideal sorption behavior, only those isotherm data at the same surface coverage can be correlated. Figure 6 contains exemplary plots of Eq. [14] for two values of c(Eu sorbed). The respective values from the isotherm at pH 4.0 are left out due to the large scatter of experimental data at higher Eu(III) concentrations. Slopes are close to 1, supporting the reaction stoichiometry underlying the SCM. The same findings on the stoichiometry of the surface complexation reaction have been reported in the literature for other trivalent elements, e.g., Y(III), Gd(III) (12), Al(III) (20), and Cr(III) (23). Surface Precipitation? As mentioned above, the shape of the isotherm curves at higher surface coverage resembles that observed in the case of surface precipitation (5, 13, 21). Surface precipitation models (SPM) combining surface complexation reactions at low sorbate concentrations with the formation of surface precipitates or surface polynucleation at higher surface coverage have been published by different authors and were successfully applied to describe the experimental isotherm data (22). According to Sposito (19) several parameters might favor surface precipitation processes over the surface sorption of distinct mononuclear species. However, he also notes that a final distinction between both processes can only be made by spectroscopic techniques. Surface precipitation reactions have been appraised experimentally for trivalent elements, e.g., for the sorption of Cr(III) at relatively high concentrations onto silica using EXAFS (23) and transmission electron microscopy (TEM) (24). In these experiments a considerable fraction of dissolved Cr(III) was present in hydrolyzed form. In the case of U(VI) sorption onto ferric oxyhydroxides it was concluded from isotherm data that a considerable fraction of sorbed U(VI) species should be oligomeric (25). Different EXAFS studies,

(1) Even at the highest pH adjusted during our experiments (pH 6.0) the fraction of hydrolyzed Eu(III) species in solution is negligible, and if we take the lowest solubility product found in the literature (log K sp 5 226.54; I 5 0.1 mol/L (28)) the calculated solubility concentration for Eu(III) at pH 5 6.0 is 2.9 3 1023 mol/L, which is higher by a factor of approximately 30 than the highest Eu(III) concentration used in our experiments (1 3 1024 mol/L). This factor increases to 6 3 103 at pH 5.0 and 6.3 3 106 at pH 4.0. As a consequence, the probability for the occurance of surface precipitation should decrease with decreasing pH and one would expect a flattening of the isotherm slope at higher surface coverage which in fact is not observed. (2) According to the rule of thumb given by Dzombak and Morel (11) the surface precipitation reaction has to be considered if the sorbate concentration exceeds one-tenth of its solubility or one-half of the total surface site concentration. Both the concentration of dissolved Eu(III) and the surface coverage (below one-third of the surface site concentration) remain below this criterion. (3) Desorption kinetics can be taken as an indicator for the kind of sorption process. A fast desorption reaction would imply that the sorbed ion is surface-bound by either surface

FIG. 6. Function plot of Eq. [14] for isotherm data at two different values in the range of high surface coverage.

160

RABUNG ET AL.

FIG. 7. Comparison of Eu31 sorption data onto hematite obtained in this work with literature data; dotted lines represent sorption data calculated from int pK int a1 and pK a2 values published in the literature but using our surface comint plexation model. *pK int a values from Ref. (30); **pK a values from Ref. (33).

complexation or ion exchange. Even though the kinetics of the Eu(III) desorption has not been investigated in detail, a reversible sorption reaction is observed after 14 days at the low and high surface coverage parts of the isotherm. (4) Finally, the pH dependency of the sorption reaction at the high Eu(III) concentration supports the assumed surface complexation reaction in Eq. [12] rather than a precipitation. In the latter case a different pH dependency would be expected. Comparison with Literature Data Some authors have investigated the sorption behavior of trivalent REE onto natural and synthetic hematite (29 –33). Because the interpretation of the results in each paper differs from another or is even missing, direct intercomparison of all published results is not possible. For the meaningful comparison the experimental data are calculated back to distribution ratios related to the surface area of hematite under investigation (K d [L/m2]). The concentration ratio of Eu(III) to hematite in experimental batches is very low in general, so that distribution ratios of Eu(III) should be comparable. All available data are plotted in Fig. 7 together with those from this work. Literature data show considerable differences both in the K d values itself and in its pH dependence. As no information is given in most cases about the number of surface sites or the acid– base characteristics of the given hematite (29, 31), it is difficult to compare the data on the basis of a surface complexation model. Such data are available only from Marmier et al. (30). Their results deviate most from the present data and appear to be the lowest K d values found in the literature with a steep pH dependence. They interpreted their results postulating the formation of the surface species ['SOHEu31] and ['SOEu21]. The formation of the first postulated species is doubtful, since

Eu31 should repulse protons upon complexation. As has been discussed before, an inner-sphere complexation, as represented by Eqs. [11] and [12], appears most plausible for the hard acid Eu31. Applying the present surface complexation model and the acidity constants of the hematite investigated by Marmier et int al. (pK int a1 5 26.38; pK a2 5 29.81) results in a shift of our K d values to higher pH values (dotted line in Fig. 7). Experimental data of Refs. (29) and (31) reasonably fit to that curve. A shift of K d values to higher pH, therefore, can be explained to a certain extent by different acid– base properties of hematite investigated by individual authors. However, this can not be the only reason for the divergence of data. In the case of data published by Cremers et al. (33), application of the present int model using their values of pK int a1 5 27.4 and pK a2 5 210.28 leads to a shift of K d to much higher pH (dashed line in Fig. 7). However, the experimental data of Cremers et al. (33) show the opposite trend and represent the strongest sorption of Eu(III) to hematite in this comparison. SUMMARY

The sorption reaction at low Eu(III) concentrations can be described by a simple Langmuir formalism. The Eu(III) sorption at surface coverages below the given concentration of “strong” binding sites [ST] and ['XOH], respectively, equally can be quantified by a pH dependent K d value. The prerequisites for the K d concept (distribution coefficient being independent of the sorbate concentration and reversibility of the sorption reaction) are fulfilled under the experimental conditions of the present study. Assuming a monodentate surface complex, the dependency of the sorption reaction on the Eu(III) concentration and on pH can be described by a SCM. For the low surface coverage range congruent surface complexation constants are obtained. The available surface site concentration can be calculated by fitting titration data using the constant capacitance model and by using a simple graphical extrapolation of sorption isotherms. The latter procedure appears to be useful for the simple validation of the number of surface sites involved in the reaction, which is usually obtained from protonation constants and the total number of “strong” surface hydroxyl groups. In general, the analysis of complete sorption isotherms providing the dependence of sorption on metal ion concentration is found to be more useful for the determination of surface complexation constants than fitting sorption edge data alone. Modeling of the sorption reaction at higher surface coverages is found more critical, because of difficulties involved in identifying individual surface reactions. The SCM needs to take into account the repulsion of surface-sorbed Eu(III) species in order to be able to describe the nonideal part of the sorption isotherms. Even though formation of polynuclear species or surface precipitation may also generate the observed nonideal sorption behavior (slope of the isotherm ,1) several indications suggest the presence of surface sites with different

EU(III) SORPTION ON HEMATITE

complexing strength. In this concentration range a simple K d model is not any more applicable. As the literature data on the surface sorption reaction of trivalent actinides or rare earth elements are scarce and experimental conditions in most cases are not clearly defined, a meaningful comparison and thus validation of our experimental results appears difficult. Surface hydroxyl groups of the hematite used in other studies may have acid– base properties different from those of the natural hematite used in this study. This fact may explain differences observed in the sorption data. Natural minerals mostly contain different amounts of amorphous or crystalline oxides and trace impurities, which might have a significant influence on their sorption properties. Information on such impurities is missing in the published works and thus a critical assessment is not possible for the sorption data obtained in different laboratories. REFERENCES 1. McKinley, I. G., and Alexander, W. R., J. Contam. Hydrol. 13, 249 –259 (1993). 2. Degueldre, C., Ulrich, H. J., and Silby, H., Radiochim. Acta 65, 173–179 (1994). 3. Warnecke, E., Hollmann, A., and Brennecke, P., Radiochim. Acta 66/67, 821– 827 (1994). 4. Stenhouse, M. J., and Po¨ttinger, J., Radiochim. Acta 66/67, 267–275 (1994). 5. Katz, L. E., and Hayes, K. F., J. Colloid Interface Sci. 170, 477–490 (1995). 6. Brown, G. E., in “Reviews in Mineralogy, Vol. 23, Mineral–Water Interface Geochemistry” (M. F. Hochella, and A. F. White, Eds.), pp. 309 –353. Mineralogical Society of America, Washington, DC, 1990. 7. Rabung, Th., Geckeis, H., Kim, J. I., and Beck, H. P., Radiochim. Acta (1998), in print. 8. Westall, J. C., “FITEQL: A Program for the Determination of Chemical Equilibrium Constants from Experimental Data, Version 2.0,” Report 82-02. Department of Chemistry, Oregon State University, Corvallis, OR, 1982. 9. Schindler, P. W., and Stumm, W., in “Aquatic Surface Chemistry” (Stumm, W., Ed.), pp. 83–107. Wiley-Interscience, New York, 1987. 10. Parks, G. A., Chem. Rev. 65, 177–198 (1965). 11. Dzombak, D. A., and Morel, F. M., “Surface Complexation Modelling: Hydrous Ferric Oxide.” Wiley-Interscience, New York, 1990. 12. Kosmulski, M., J. Colloid Interface Sci. 195, 395– 403 (1997). 13. Farley, K. J., Dzombak, D. A., and Morell, F. M. M., J. Colloid Interface Sci. 106, 226 –242 (1985).

161

14. Runde, W., “Carbonatkomplexierung von dreiwertigem Europium.” Thesis, Institut fu¨r Radiochemie, TU Mu¨nchen, 1990. 15. Schindler, P. W., in “Reviews in Mineralogy, Vol. 23, Mineral–Water Interface Geochemistry” (M. F. Hochella and A. F. White, Eds.), p. 291. Mineralogical Society of America, Washington, DC, 1990. 16. Del Nero, M., Made, B., Bontems, G., and Cle´ment, A., Radiochim. Acta 76, 219 –228 (1997). 17. Lu¨tzenkirchen, J., and Behra, P., in “Transport and Reactive Processes in Aquifers” (Th. Dracos and F. Stauffer, Eds.), pp. 391–396. Balkema, Rotterdam, 1994. 18. Davis, J. A., and Kent, D. B., in “Reviews in Mineralogy, Vol. 23, Mineral–Water Interface Geochemistry” (M. F. Hochella and A. F. White, Eds.), p. 225. Mineralogical Society of America, Washington, DC, 1990. 19. Sposito, G., in “Geochemical Processes at Mineral Surfaces” (J. A. Davis, and K. F. Hayes, Eds.), ACS Symposium Series 323, pp. 217–230. American Chemical Society, Washington, DC, 1986. 20. Lo¨vgren, L., Sjo¨berg, S., and Schindler, P. W., Geochim. Cosmochim. Acta 54, 1301–1306 (1990). 21. Katz, L. E., and Hayes, K. F., J. Colloid Interface Sci. 170, 491–501 (1995). 22. Tiffreau, C., Lu¨tzenkirchen, J., and Behra, P., J. Colloid Interface Sci. 172, 82–93 (1995). 23. Fendorf, S. E., Lamble, G. M., Stapleton, M. G., Kelley, M. J., and Sparks, D. L., Environ. Sci. Technol. 28, 284 –289 (1994). 24. Fendorf, S. E., and Sparks, D. L., Environ. Sci. Technol. 28, 290 –297 (1994). 25. Hsi, C. K. D., and Langmuir, D., Geochim. Cosmochim. Acta 49, 1931– 1941 (1985). 26. Waite, T. D., Davis, J. A., Payne, T. E., Waychunas, G. A., and Xu, N., Geochim. Cosmochim. Acta 58, 5465–5478 (1994). 27. Manceau, A., Charlet, L., Boisset, M. C., Didier, B., and Spadini, L., Appl. Clay Sci. 7, 201–223 (1992). 28. Aksel’rud, V. I., and Ermolenko, V. I., Zh. Neorg. Khim. 6, 4481 (1961). 29. Ledin, A., Karlsson, S., Du¨ker, A., and Allard, B., Radiochim. Acta 66/67, 221–228 (1994). 30. Marmier, N., Dumonceau, J., Chupeau, J., and Fromage, F., C. R. Acad. Sci. Paris 317, 311–317 (1993). 31. Music, S., Gessner, M., and Wolf, R. H. H., J. Radioanal. Nucl. Chem. 50, 91–100 (1979). 32. Koeppenkastrop, D., Decarlo, E. H., and Roth, M., J. Radioanal. Nucl. Chem. 152, 337–346 (1991). 33. Cremers, A., and Maes, A., in “Application of Distribution Coefficients to Radiological Assessment Models” (T. H. Sibley and C. Myttenaere, Eds.), pp. 4 –14. Commission of the European Communities, Elsevier, London, 1986.