Journal of Colloid and Interface Science 318 (2008) 5–14www.elsevier.com/locate/jcis

Immobilization of trivalent actinides by sorption onto quartz andincorporation into siliceous bulk: Investigations by TRLFS

S. Stumpf a,∗, Th. Stumpf b, J. Lützenkirchen b, C. Walther b, Th. Fanghänel a,c

a European Commission, Joint Research Centre, Institute for Transuranium Elements, Karlsruhe, Germanyb Forschungszentrum Karlsruhe, Institut für Nukleare Entsorgung, P.O. Box 3640, 76021 Karlsruhe, Germany

c Ruprecht-Karls-Universität Heidelberg, Physikalisch-Chemisches Institut, Im Neuenheimer Feld 253, 69120 Heidelberg, Germany

Received 29 June 2007; accepted 28 September 2007

Available online 3 October 2007

Abstract

The adsorption of Cm(III) on quartz is studied by time resolved laser fluorescence spectroscopy (TRLFS) in the pH range from 3.75 to 9.45.The raw spectra are deconvoluted into three single components. The first one has a peak maximum at 593.8 nm and can be attributed to the Cm(III)aquo ion with an emission lifetime of 68 ± 3 µs. The second one corresponds to an adsorbed species and has a peak maximum at 601.4 nm andan emission lifetime of 123 ± 10 µs. The peak maximum of the third component is shifted to higher wavelength (603.6 nm) while the lifetimeremains constant. Additionally, the adsorption of Am(III) on quartz is investigated in batch experiments. Based on the spectroscopic data asorption mechanism is suggested. In addition, the obtained Am uptake data and the Cm-TRLFS data are modeled simultaneously using a singlesite Basic Stern model in combination with the charge distribution concept of Pauling. The finally suggested model consists of two bidentatesurface complexes where the second one is the product of hydrolysis of the first sorption species. In a separate set of experiments the influenceof silicic acid at different concentrations on the Cm(III) speciation in a quartz system is investigated by TRLFS. In suspension silicic acid at lowconcentration (3.5×10−4 mol/L) has no influence on the Cm(III) speciation. At high concentration (3.5× 10−2 mol/L) the Cm(III) speciation isdefinitely influenced. The results at higher concentration indicate the formation of Cm(III)/silicic acid complexes and the incorporation of Cm(III)into siliceous bulk. This is confirmed by measurements at a quartz single crystal surface. Moreover, these measurements indicate the formation ofquartz/Cm(III)/silicic acid ternary complexes at the mineral surface.© 2007 Elsevier Inc. All rights reserved.

Keywords: Curium; Quartz; Silicic acid; Surface complexation; Sorption; Ternary complexes; TRLFS; Single crystals; Adsorption model

1. Introduction

Radionuclide migration in natural aqueous systems is an on-going concern in environmental research in particular in thecontext of the long term performance of nuclear waste repos-itories. The transport of actinides is strongly influenced by ad-sorption onto mineral surfaces and interaction with organic andinorganic ligands. Fundamental insight into sorption and com-plexation mechanisms such as identification of dissolved andadsorbed species is of cardinal importance for reliable predic-tion of actinide reactions in natural systems. Therefore it is

* Corresponding author.E-mail address: silvia.stumpf@ec.europa.eu (S. Stumpf).

0021-9797/$ – see front matter © 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.jcis.2007.09.080

necessary to characterize the different actinide species and toelucidate the reaction mechanisms involved.

Time resolved laser fluorescence spectroscopy (TRLFS) en-ables the speciation of lanthanides like Eu(III) and actinides likeU(VI), Am(III) and Cm(III) [1] in aqueous solution and on thewater/mineral interface. Due to the high fluorescence yield ofCm(III) TRLFS allows speciation studies in the nanomolar con-centration range corresponding to a surface loading far below amonolayer. Up to now the interaction of Cm(III) with mineralsurfaces like γ -alumina [2], clay minerals [3], feldspars [4],CSH phases [5], cement [6], calcite [7] and α-alumina singlecrystal surfaces [8] has been investigated. The characterizationof the adsorbed species is deduced from excitation and emissionspectra and from the fluorescence emission lifetime of Cm(III).

6 S. Stumpf et al. / Journal of Colloid and Interface Science 318 (2008) 5–14

TRLFS investigations of the interaction of Eu(III) withamorphous silica led to the conclusion that the trivalent lan-thanide is not adsorbed but incorporated into the bulk struc-ture [9]. Comparable results have been found for the interactionof Cm(III) with amorphous silica colloids [10]. These resultsindicate different mechanisms for the interaction of trivalentlanthanides/actinides with amorphous silica and with the aforementioned minerals. One possible explanation is the presenceof dissolved silica from weathering of silicate minerals as de-scribed in literature [11–13]. The solubility of amorphous silicais known to be 2 × 10−3 M at pH < 9, dominated by monosili-cic acid and increases rapidly with pH via deprotonation andthrough formation of polysilicic acid [14–16]. The different Sispecies may be involved in various reactions like complexation[17], formation of colloids [18,19] and precipitation [13]. Thesereactions have to be taken into account when the speciation ofCm(III) is investigated in the presence of silicate minerals.

The present TRLFS investigation is part of a study, which in-tends to give a complete Cm(III) speciation in a natural systemmainly composed of silicates like quartz. Hereby, Cm(III) wasselected as a representative of a trivalent actinide ion. Becauseof their omnipresence in nature silicates can have a dominatinginfluence on the aqueous chemistry and hence on the migra-tion behavior of actinides in the natural environment. In thisstudy the adsorption of Cm(III) on quartz and the influence ofsilicic acid, which can be generated by the dissolution of the sil-icate mineral, at different concentrations on the Cm(III) speci-ation was investigated in dependence on pH. The studies of theCm(III) speciation were performed on quartz particles in sus-pension as well as on quartz single crystal surfaces. In additionto TRLFS measurements the adsorption of Cm(III) and Am(III)was investigated by batch experiments and α-spectrometry. Thebatch sorption and TRLFS data for the quartz particles in ab-sence of silicic acid are described by a surface complexationmodel.

1.1. Surface complexation model

Adsorption of cations on oxides has been the subject ofmany studies. A number of different models have been usedto describe the adsorption behavior [20]. Most models interpretthe adsorption as an interaction of metal ions with functionalgroups at the surface. Those models usually consider competi-tive adsorption of protons and metal ions in proton–metal ion–adsorbent systems. Before interacting with the surface func-tional groups the metal ions have to overcome a potential dif-ference between the bulk solution and the interface. A surfacecomplexation model is therefore composed of two parts: a dis-cription of the electric double layer (e.g., that by Gouy andChapman [21]) and conventional mass law equations and bal-ances for the reactions at the surface. With this, the overallreaction for the absorption of a metal ion to a surface func-tional group can be separated into a chemical (intrinsic) and avariable electrostatic part:

K = e−�Gads/RT = e−(�G◦r +�Gel)/RT

(1)= Kine−�Gel/RT = KinKel,

Fig. 1. Electrostatic double layer model for the interpretation of the sorptiondata.

where K : overall reaction constant, Kin: intrinsic reaction con-stant, Kel: electrostatic reaction constant, �Gads: Gibbs freeenergy for the adsorption, �G◦

r : standard Gibbs free energy,�Gel: electrostatic energy.

The electrostatic energy change, �Gel, is variable due to thechange in charge upon adsorption of ions and is determined bythe electrostatic potential Ψ which is calculated by the use ofan electrostatic model, in which the charge distribution (CD) ofions is involved [22].

Models may differ in the formulation of the surface chemi-cal reactions (stoichiometries) and the assumed structure of theelectrostatic double layer. In this study the basic Stern modelwas used as the electrostatic double layer model for the in-terpretation of the adsorption data (Fig. 1). Two electrostaticplanes are considered that are separated from each other by acharge free layer in between called the Stern layer. The chargesof protons are allocated to the surface (σ0). Adsorption of back-ground electrolyte ions is considered by treating them as pointcharges and placing that charge at the head end of the diffusepart of the double layer (σ1). The sum of σ0 and σ1 is com-pensated by the charge σd of the diffuse part of the electrosta-tic double layer, which is calculated from the Gouy–Chapmanequation. The Stern layer is characterized by a capacitance.This capacitance is usually determined from model fits to acidbase titration data [23].

As demonstrated in Eq. (2) the interface of an oxide may becomposed of several functional groups, which are coordinatedto one or more metal ions of the solid.

(2)O−2 + 2H+ ⇔ OH−1 + H+ ⇔ OH02.

This picture leads to a discrete surface heterogeneity becauseeach group has its own proton affinity and charge characteris-tics, which may even vary with the crystallographic plane. Forthe determination of the overall charge σ0 at the surface onemust take into account that the surface oxygen is not only par-tially neutralized by protons but also by the metal ions in themineral structure. Vice versa, the charge of the cation is com-pensated by the charge of the surrounding oxygens. As it wasintroduced by Pauling [24] for neutralization, the charge is dis-tributed over the surrounding ligands, which can be expressedper bond. With this, the bond valence v is defined as the charge

S. Stumpf et al. / Journal of Colloid and Interface Science 318 (2008) 5–14 7

z of the cation divided by its coordination number CN:

(3)v = z/CN.

A charge distribution (CD) concept like this is also applied forcations adsorbed to the mineral surface. Whereas electrolyteions are still considered as point charges as stated above (trueouter-sphere surface complexes), the charge of sorbed cationslike Cm and Am is distributed between the surface plane andthe head end of the diffuse layer thus rationalizing their sizecompared to that of a proton. The simple Pauling bond va-lence concept can be used as a first order estimate in relatingthe charge distribution needed in the model to the structure ofthe adsorbing ion. In case of a structural change of the cationadsorbed to the surface a change of the charge distribution andwith this of the surface potential Ψ results.

The structure of the surface, the structure of the adsorbedspecies, and the electrostatic potential profile are all essentialfeatures of an adsorption model using physically realistic sur-face species.

2. Experimental

2.1. Methods

2.1.1. TRLFS (time resolved laser fluorescence spectroscopy)TRLFS measurements were performed using a dye laser

(Lambda Physics, scanmate) pumped by a XeCl-excimer laser(Lambda Physics, EMG, 308 nm, 24 ns) for excitation. Emis-sion spectra were recorded from 580 to 620 nm. The fluo-rescence emission is detected by an optical multichannel an-alyzer consisting of a polychromator (Jobin Yvon, HR 320)with a 300/600/1200 lines/mm grating and a photodiode array(Spectroscopy instruments, ST 180, IRY 700G). The Cm(III)was excited at 396.6 nm. The emission spectra of Cm(III)were recorded in the 580–620 nm range (1200 lines/mm grat-ing). For measuring the time dependent emission decay (300lines/mm grating), the delay time between laser pulse and cam-era gating was scanned with time intervals between 5 and 10 µs.For measuring the decay time of the fluorescence emission thedelay time was shifted in steps between 5 and 10 µs using aconstant time window of 1 ms.

2.1.2. Data modelingAll sorption data were modeled as surface charge densities

simultaneously using the FITEQL 2.0 program [25] in combi-nation with the optimization code UCODE [26]. The Daviesequation was used to correct for ionic strength effects. Thisallows fitting of all parameters, including capacitance values,which can only be adjusted manually in FITEQL. The linearcorrelation coefficient (between the capacitance value and theelectrolyte binding constant) which can be obtained with thiscode combination has an absolute value of 0.91. This indicatesa relatively strong correlation between the two adjustable para-meters. This information would not be obtained using FITEQLin optimization mode. Another advantage of using UCODE forparameter estimation is that a high degree of flexibility is ob-tained in the combination of experimental data pertaining to

the system. Thus in the present study, for the first time ex-perimentally obtained solution and surface speciation (for Cmfrom TRLFS) could be directly used in the modeling exercisein simultaneous combination with classical Am uptake data (as-suming that both exhibit equal behavior).

3. Materials

All used chemicals were of analytical grade and carbonatefree.

3.1. Quartz particles

The used quartz was obtained from Redon/Bretagne, France.The quartz was crushed in a mortar and sieved to get frac-tions <20 µm. A <5 µm fraction (BET surface >1 m2/g) thatwas used for the sorption experiments was selectively obtainedvia centrifugation. Purity and crystallinity of the quartz powderwere verified by ICP-MS and XRD measurements.

3.2. Quartz single crystal

The quartz single crystal was obtained from Belo Hori-zonte, Brazil and was cut (X-cut) in 10 × 10 × 1 mm singlecrystal slices. XPS measurements indicated a carbon impurityon the surface that was removed by cleaning the surface withmethanol.

3.3. Silicic acid

The silicic acid solution was prepared by diluting a sil-icon standard for ICP (1 mg/ml Si in 2% NaOH, ACROSFisher Scientific GmbH) with HClO4 to a desired concentrationof 3.5 × 10−4 mol/L (undersaturation) and 3.5 × 10−2 mol/L(oversaturation) [14,15,27,28].

3.4. Experimental set-up

A stock solution of the long-lived curium isotope Cm-248(t 1

2= 3.4 × 105 years) with the isotopic composition 97.3%

Cm-248, 2.6% Cm-246, 0.04% Cm-245, 0.02% Cm-247 and0.009% Cm-244 in 1.0 M HClO4 was used for the TRLFSexperiments. The initial curium concentration determined byICP-mass spectroscopy was adjusted to 2.0 × 10−7 mol/L. Toavoid complexation by carbonate all experiments were made ina glove box under argon atmosphere at 25 ± 1 ◦C. The Cm(III)adsorption was investigated in batch experiments with a quartzsuspension of 1 g/L quartz particles <5 µm in size. The pHof the batch samples was varied by adding NaOH and HClO4.During the curium/quartz contact time the samples were shakenperiodically. All samples had an electrolytic background of0.1 mol/L NaClO4.

Additionally batch sorption experiments with Am(III) wereperformed. The samples were prepared analogous to the Cm(III)samples. The pH edges were obtained by analyzing the Am(III)concentration of the solution by α-spectrometry after equilibra-tion (1 day) and solid liquid separation.

8 S. Stumpf et al. / Journal of Colloid and Interface Science 318 (2008) 5–14

The influence of silicic acid was investigated by adding thesilicon standard to the quartz suspension at acidic pH (≈3.5)before adding the Cm(III). The concentration of silicic acidwas 3.5×10−4 mol/L (undersaturation) and 3.5×10−2 mol/L(oversaturation), respectively.

The sorption of Cm(III) onto single crystals was obtainedcontacting the single crystal with a Cm(III) solution (2 ×10−7 mol/L) at different pH values. An additional set ofexperiments was done contacting the single crystal with aCm(III)/quartz suspension (1 g/L quartz; <5 µm; 2 × 10−7

mol/L Cm(III)) in presence of 3.5 × 10−2 mol/L silicic acidat different pH values.

4. Results and discussion

4.1. Adsorption of Cm(III) on quartz-TRLFS measurements

Fluorescence emission spectra of 2 × 10−7 mol/L Cm(III)in a quartz suspension, recorded at different pH values rangingfrom 3.75 to 9.45, are shown in Fig. 2. At low pH, the emis-sion band with a peak maximum at 593.8 nm can be attributedto the Cm3+ aquo ion [29]. With increasing pH, the intensity ofthis peak decreases and a red-shift of the fluorescence emissionup to 603.6 nm at pH 9.45 appears. Under the present exper-imental conditions hydrolysis of Cm(III) can be ruled out andthe change of fluorescence spectra can be solely explained bya stepwise complexation of the curium aquo ion at the min-eral surface and the formation of inner-sphere complexes. Themixed spectra were deconvoluted using a factoranalysis pro-gram [30]. The species identified by the deconvolution proce-dure are plotted in Fig. 3. Moreover, from the peak deconvolu-tion data and the respective FI values (fluorescence intensity)the mole fractions of the three Cm-species are determined andplotted in Fig. 3 as a function of pH (speciation plot). The aquoion dominates the system up to a pH value of 4.5. Then theformation of the first surface complex (601.4 nm) starts witha maximum fraction of 30% at pH 5.5. With increasing pH

(a)

(b)

Fig. 3. (a) Calculated single components and (b) speciation plot for the sorptionof Cm(III) onto quartz.

Fig. 2. Fluorescence emission spectra of 2 × 10−7 mol/L Cm(III) in a suspension of 1 g/L quartz at different pH values.

S. Stumpf et al. / Journal of Colloid and Interface Science 318 (2008) 5–14 9

Fig. 4. Calculated slope for the formation reaction of sorption Species 2.

the second surface complex (603.6 nm) is formed. This speciesdominates the system up to pH 9. The fluorescence emission ofthe Cm3+ aquo ion decays with a fluorescence emission life-time of 68 µs, while the first as well as the second Cm complexemitting at 601.4 nm and 603.6 nm both have an emission life-time of 123 ± 10 µs. The increase in lifetime by complexationreflects the exclusion of water molecules from the first coordi-nation sphere of Cm(III). In the first coordination sphere H2Oacts as a fluorescence quencher by energy transfer and exci-tation of H2O vibrations and, therefore, causes a decrease ofthe emission lifetime. The short lifetime of the Cm3+ aquo ion(68 ± 3 µs) compared to a lifetime of 1300 µs in D2O [29] is aresult of these quenching processes. A mathematical expressionfor the correlation of lifetime and number of H2O quenchers isgiven in literature by Kimura et al. [29]. Complexes 1 and 2are characterized by the same fluorescence lifetime, implyingthat the number of quenching ligands in the first Cm(III) co-ordination sphere remains unchanged. Applying the correlationof Kimura et al. the number of water molecules in the first co-ordination shell for both surface complexes can be estimatedto be 5 ± 0.2. However, the change in the emission spectraclearly indicates that there are two different species with dif-ferent composition or structure. With increasing pH a stepwisedeprotonation according to Eq. (4) is very likely.

(4)Species 1 → Species 2 + nH+.

From this equation it follows the correlation for the law of massaction:

(5)K = [Species 2]/[Species 1] × [H+]n

.

A plot of the ratio log(Species 2/Species 1) with pH allowsthe determination of the proton stoichiometry for the givenreaction. Here, the plot gives a slope of 1 (Fig. 4), indicat-ing the exchange of one proton during the formation reac-tion. The increase of the error bar value together with an in-crease of pH is a result of the change of concentration ratio[Species 2]/[Species 1] during the deprotonation reaction.

This result is in good accordance with spectroscopic in-vestigations of the adsorption of Cm(III) on alumina [2], clayminerals [3] and feldspar surfaces [4]. Relying on the cited in-vestigations and on the basis of the actual study a mechanismfor the Cm adsorption on a quartz surface can be deduced. Ina first reaction step a surface complex is formed by exchangeof approximately half of the hydration shell from the first Cmcoordination sphere. According to the determined proton stoi-chiometry in Eq. (5) in a second reaction step the release of aproton is proposed. Whether this proton is released by the hy-drolysis of the surface complex or by a deprotonation of thesurface itself is not clear at this point. Moreover, the spectro-scopic data give no evidence how many oxygen of the surfaceare involved into the bonding to the sorption species.

To get more structural insight the adsorption of Am(III) onquartz at different concentrations of the actinide was investi-gated in an analogue experiment. The Cm and Am sorption datawere then modeled and a sorption mechanism is proposed.

4.2. Adsorption of Cm(III) and Am(III) on quartz-surfacecomplexation modeling

As mentioned above, a surface complexation model is com-posed of two parts—one part describes the electric double layerwhereas the other one gives conventional mass law equationsfor the reactions at the surface. Here, the Basic Stern Modelwas used as the electrostatic double layer model for the inter-pretation of the adsorption data. With regard to the mass lawequations two types of surface groups can be considered for thequartz surface. The logK value for the protonation of the dou-bly coordinated Si2–O group is extremely low and can thereforebe regarded as inert [31]. Only the singly coordinated groupsare reactive and the protonation reactions can be formulated as:

≡SiOH ↔ ≡SiO− + H+, (6)

≡SiOH+2 ↔ ≡SiOH + H+. (7)

Due to the very low logK value for Eq. (7) the protonation ofthis surface group is not very likely. Therefore, in this study apotentially complex situation is simplified by the use of onlyone charging reaction (Eq. (6); 1-pK approach) [32,33]. Atthe point of zero charge of quartz (around pH 2–3) all surfacegroups are protonated and are uncharged (SiOH0). Therefore,in the pH range that is relevant for the sorption experiments thesurface charge of quartz is negative and the quartz surface issupposed to be homogeneous [34]. With this, the reactions atthe quartz surface within the 1-pK , 1-site basic Stern model arethe following:

SiOH ↔ SiO− + H+, (6)

SiOH + Na+ ↔ SiO− · · ·Na+ + H+. (9)

The logK value of reaction (6) is −7.9 for I = 0 according toprevious work [35] and is calculated to be −7.79 for I = 0.1(i.e., for the background electrolyte used in the is study: 0.1 MNaClO4). The logK for reaction (9) was optimized numeri-cally (logK = −6.93) [31]. Based on previous work [31], the

10 S. Stumpf et al. / Journal of Colloid and Interface Science 318 (2008) 5–14

Fig. 5. pH edges of the sorption of 248Cm and 243Am onto quartz.

Fig. 6. Sorption of trivalent actinides onto a quartz surface: modeled mechanismand sorption species.

number of sites at the quartz surface was set to 4.6 nm−2. More-over the capacitance of the Stern layer was set to a value of1.44 F/m2 according to acid base titration data [23]. All sorp-tion data were used in one UCODE set-up file and fitted usingthe Basic Stern Model outlined above and considering idealcharge distribution. The experimental data were all used as thefraction of species with respect to the total concentration of Cm(Cm3+, Species 1, Species 2) or Am (Amads), respectively.

Fig. 5 shows pH sorption edges of Am(III) at three dif-ferent Am(III) concentrations (2 × 10−7/10−6/10−5 mol/L).Moreover, the Cm sorption data ([Cm(III)] = 2 × 10−7 M), ob-tained by the calculation of the speciation from the TRLFS data(Fig. 3), are added to the plot in Fig. 5. The pH edges of bothactinides fit very well for nearly identical conditions. This re-sult is in good agreement with the assumption of equal chemicalbehavior for americium and curium. As expected, with increas-ing actinide concentration the pH edges are shifted to higherpH values. The fit of the sorption data is also given in Fig. 5(black line). The proposed structure model for the two differentsorption species is shown in Fig. 6. The structure of the surfacecomplexes determines the charge distribution at the mineral sur-face and with this also determines the electrostatic part of theequation constant which finally describes the given sorptiondata. Here, the trivalent actinide ion is attached to the surfacein a bidentate conformation. Once the location of the complexhas been defined one has to consider the position of charge forboth surface species. Part of the charge of the surface complexis shared with the surface itself. The remaining part is attributedto the other ligands of the complex which are oriented towardsthe solution. By application of the Pauling bond valence con-cept (Eq. (3)) and assuming a coordination number of 8 for theactinide a value of v = 3/8 can be attributed to each metal bond.It follows for the charge of the actinide in the first complex

Fig. 7. Slope analysis of the sorption data in consideration of different assump-tions for logK .

+3 − 2 × 3/8 = 2.25. The surface oxygen has a bond valenceof v = −2/2 = −1 whereas a value of v = −2/2 + 1 (whichcomes from H+) can be attributed to the hydroxid group. Tak-ing the contribution of the actinide valence bond into accountthe charge at the surface becomes −1 + 0 + 2 × 3/8 = −0.25.As a result of a proposed twofold hydrolysis (v(OH–) = −1)for the second complex, the charge of the actinide changes to avalue of +3 − 2 × 3/8 − 1 = 1.25 and for the surface to a valueof −2 + 2 × 3/8 = −1.25. It is obvious that in the proposedmodel a change of surface complex structure is associated witha change of charge at the solid solution interface. Finally, theelectrostatic part in the reaction equation of the sorption processis changed. This means, that the classical plot for the estima-tion of proton stoichiometries (i.e., log[Species 2]/[Species 1]vs pH) is not applicable, since this involves the assumption ofone invariant overall stability constant. As it was already shown,by application of a surface complexation model the stabilityconstant K is separated into an intrinsic and electrostatic part(K = Kin × Kel; Eq. (1)). A plot taking Kin as overall stabilityconstant results in a value of 2 for the slope (Fig. 7). It fol-lows that two protons are released during the surface reaction(Fig. 6) and not one as inferred from the classical slope deter-mination that was applied for the interpretation of the TRLFSdata. Additionally, taking the electrostatic part into account asgiven in Eq. (1) and applying K as overall stability constanta plot of log[Species 2]/[Species 1] vs pH results a line witha slope of 1 (Fig. 7). With the separation of K into an intrin-sic and electrostatic part the stoichiometry of the formationreaction of Species 2 from Species 1 as well as the slope de-termination are very well reflected. With the proposed structuremodel it is possible to fit the sorption data of the Cm(III) sorp-tion as well as of the Am(III) sorption onto quartz (Fig. 5). Atthe highest Am concentration (2 × 10−5 mol/L) Am is presentin excess of the surface sites. Therefore, the observed uptakedata beyond site saturation may be attributed to the formationof a solid Am-phase which is either amorphous or crystallineor a mixture of both. Therefore, two fitting curves from two

S. Stumpf et al. / Journal of Colloid and Interface Science 318 (2008) 5–14 11

solubility products, namely logK0L = −25.1 (amorphous Am-

hydroxide) and logK0L = −26.4 (crystalline Am-hydroxide),

have been calculated for the Am sorption at higher concentra-tions. As the experimental data are located in between the twofitting lines (Fig. 5) obviously a mixture of an amorphous andcrystalline Am-phase is formed. The logK values for the sorp-tion of Cm(III) and Am(III) at the quartz surface are:

logK = 5.05, 2SiOH ↔ Species 1 + H+,

logK = −6.15, Species 1 ↔ Species 2 + 2H+.

The interpretation of TRLFS data by application of the slopeanalysis in the simplest form gives an idea for the adsorption oftrivalent actinides on a quartz surface. The modeling of sorptiondata by application of a surface complexation model confirmsthe initial assumption. Moreover, it results in the formulation ofan unique sorption mechanism, that is, the formation of a biden-tate at the mineral surface in a first step (sorption Species 1) andthe twofold hydrolysis of the bidentate in a second step (sorp-tion Species 2).

4.3. Adsorption of Cm(III) on quartz-influence of silicic acid

The dissolution of quartz results in the formation of silicicacid that tends to act as inorganic ligand for cations. The ques-tion arises weather silicic acid has an influence on the Cm(III)speciation in a quartz system. The interaction of Cm(III) withsilicic acid at low concentrations (undersaturation) were inves-tigated by Panak et al. [36]. Two different Cm(III)/silicic acidcomplexes together with their stability constants were deter-mined. Moreover, the influence of silicic acid at higher concen-trations (oversaturation) was investigated [36]. At these concen-trations silicic acid polymerizes and forms colloids. As a resultof colloid formation the trivalent actinide is incorporated intothe amorphous silica phase resulting in the total loss of its hy-dration sphere.

According to the investigations performed by Panak et al.,in this study, the influence of silicic acid at different concen-trations (3.5 × 10−4 and 3.5 × 10−2 mol/L) on the Cm(III)speciation in the presence of quartz was investigated by TRLFS.The emission spectra were recorded in a pH range from 3.48to 9.25 and deconvoluted as described above. In Fig. 8 the sin-gle components together with the speciation are shown for theinteraction of Cm(III) with a quartz surface in absence and pres-ence of 3.5×10−4 mol/L (undersaturated) silicic acid. For bothsystems the same species with the same fluorescence emissionlifetimes are observed. Moreover, silicic acid at this concen-tration has no influence on the Cm speciation. However, thespeciation changes significantly when increasing the silicic acidconcentration to a value of 3.5 × 10−2 mol/L (oversaturated).A comparative presentation of the single components togetherwith the speciation for Cm(III) in the presence of quartz withand without addition of silicic acid at high concentrations isgiven in Fig. 9. The influence of silicic acid results in a shift ofemission maximum for the first surface complex from 601.4 nmwithout silicic acid to 602.6 nm in the presence of the complex-ing ligand. The emission band for the second surface complex

(604.0 nm) is broadened indicating that this peak must be at-tributed to more than one surface species. Up to a pH of 6 thespeciation plots of the two systems coincide with each other.At higher pH the Cm speciation is obviously influenced bythe silicic acid. Panak et al. report on the formation of sili-cic acid colloids by formation of polymers at pH values >6[36]. Furthermore, these colloids complex Cm(III) in an unde-fined stoichiometry. Taking these investigations into account,the influence of silicic acid at high concentration on the Cm(III)speciation in a quartz system can be explained by the formationof such polysilicic acid colloids. The complexation of Cm(III)by these colloids results in the formation of Cm species withan unknown stoichiometry. As reported by Panak et al. the in-corporation of Cm(III) into silica colloids is associated with atotal loss of the Cm hydration sphere [36] which is indicatedby lifetimes of 310 ± 19 µs. Here, the measured lifetimes atpH values <6 are in accordance with the lifetimes of the sili-cic acid free system. But, with increasing pH, when silicic acidcolloids are formed, also the lifetimes increase up to a value of750 ± 30 µs. This increase of lifetimes indicates the loss of theCm(III) hydration sphere (i.e., incorporation) which is in goodaccordance with the investigations performed by Panak et al.

The question arises if adsorbed Cm(III) is desorbed from thequartz surface and preferentially incorporated into the formedsilica colloids in solution instead which would result in a mo-bilization of Cm(III). Another possibility is the formation ofpolysilicic gel-like layers and incorporation of Cm(III) in suchlayers at the quartz surface (ternary quartz/Cm(III)/silicic acidcomplexes) which could result in an immobilization of Cm(III).To solve this issue, Cm sorption experiments in the presence ofsilicic acid at high concentration were performed at a singlecrystal quartz surface.

4.4. Adsorption of Cm(III) on quartz-single crystalmeasurements

The adsorption of Cm(III) on a quartz single crystal in thepH range from 2.5 to 6.2 was observed by autoradiographicmeasurements (Fig. 10). With increasing pH adsorption at thesurface increases. Simultaneously, the intensity of the Cm(III)fluorescence emission signal increases, indicating the forma-tion of a surface complex (Fig. 11). Because of the small singlecrystal surface area compared to the one provided in suspen-sion the adsorption process is shifted to higher pH values andonly the first sorption species is observed in the selected pHrange. The fluorescence emission maximum of this species is at600.3 nm and shows a shoulder at lower wavelength. The peakposition differs slightly from the fluorescence emission max-ima of the first surface complex given in a quartz suspension(601.4 nm). In suspension Cm(III) adsorbs on different crystalplanes. Here, only the x-plane is available for adsorption. Re-cently it was found that with changing the crystal plane of a sin-gle crystal the maximum of Cm fluorescence emission changes[8]. With this, the determined fluorescence emission maximain suspension can be attributed to an averaged wavelength overall planes whereas the maximum at 600.3 nm corresponds toone species adsorbed at one designated plane. The lifetime of

12 S. Stumpf et al. / Journal of Colloid and Interface Science 318 (2008) 5–14

(a)

(b)

Fig. 8. (a) Calculated single components and (b) speciation plot for the sorption of Cm(III) onto quartz in presence of silicic acid at low concentration.

the adsorbed species was measured at pH 5.7 and 6.2. For bothpH values a biexponential decay behavior is observed whereasthe rate of the species with the shorter lifetime becomes less athigher pH. The quartz surface is covered by a thin liquid film ofcurium solution as a result of preparation. Therefore, the shorterlifetime with a value of 68 ± 3 µs can be attributed to the non-surface complexed Cm3+ aquo ion. Moreover, the shoulder atlower wavelength in the emission spectrum can be explainedby the fact, that with increasing pH the ratio of the aquo ion de-creases because of increasing adsorption. The longer lifetimehas a value of 125 ± 10 µs. This value corresponds to a surfacespecies with approximately five water molecules in the first co-ordination sphere of curium which is in good accordance withthe lifetimes that can be attributed to the afore determined Cmsorption species in a quartz suspension. We conclude that thesame surface complex is formed at the single crystal surfaceand in suspension. Fig. 12 presents the Cm(III) emission spec-tra after adding silicic acid to the system. The peak maximum

is shifted to 602.7 nm indicating a change of the ligand field ofthe adsorbed Cm(III) by complexation with silicic acid. Besidethe spectral shift, the emission signal is broadened indicatingthe formation of more than one species as it was found for themeasurements in suspension in the presence of high silicic acidconcentrations. The measured lifetimes in presence of silicicacid increase up to a value of 234 µs at pH 6.7 and 323 µs atpH 9.1. The observed increase can be attributed to the reduc-tion of the first Cm(III) hydration sphere to one water moleculeas a result of the incorporation of Cm(III) into the bulk mater-ial. The Cm(III) emission spectrum at pH 9.1 was measured asecond time after 78 days. The spectrum with a maximum at602.7 nm was again broadened. The decay behavior that couldbe attributed to the spectrum was fitted with a function of higherorder. This again indicates that the emission spectrum is com-posed of several complexation species. The determined lifetimeincreased to a value >500 µs which can be attributed to thecomplete loss of the first Cm(III) hydration sphere. As it was

S. Stumpf et al. / Journal of Colloid and Interface Science 318 (2008) 5–14 13

(a)

(b)

Fig. 9. (a) Calculated single components and (b) speciation plot for the sorptionof Cm(III) onto quartz in presence of silicic acid at high concentration.

Fig. 10. Autoradiographic images of the Cm(III) sorption onto a quartz surface(X-cut) in the pH range 2.5 to 6.2.

additionally confirmed by autoradiographic measurements theincorporation of Cm in the bulk structure does not result inthe desorption of Cm from the crystal surface. The measure-ments show a constant Cm(III) occupancy of the single crystalsurface instead of a reduction caused by desorption processes.The influence of silica colloids solely results in the formationof ternary quartz/Cm(III)/silicic acid complexes at the quartzsurface. Hence, we conclude that higher concentrations of sili-cic acid do not cause any mobilization of the already sorbedactinide. This result is of great importance in view of the as-

Fig. 11. Fluorescence emission spectra of 2 × 10−7 mol/L Cm(III) at a quartzsurface (X-cut) at pH 4.0 and 6.2.

Fig. 12. Fluorescence emission spectra of 2 × 10−7 mol/L Cm(III) at a quartzsurface (X-cut) in presence of 3.5 × 10−2 mol/L silicic acid at pH 6.7, 9.1 andpH 9.1 after 78 days contact time.

sessment of the actinide migration in natural systems and withthis the safety assessment of nuclear waste repositories.

5. Summary

Depending on pH Cm(III) adsorbs on quartz by formationof two sorption species. The modeling of the sorption data sug-gests the following sorption mechanism. Cm(III) adsorbs in abidentate fashion in a first step and is hydrolyzed in a secondstep. Silicic acid at low concentration (3.5 × 10−4 mol/L) hasno influence on the Cm(III) speciation. At higher silicic acidconcentration (3.5×10−2 mol/L) and pH values >6 Cm(III) isincorporated into the gel-like bulk. This complexation results inthe total loss of the first Cm(III) hydration sphere. The TRLFSmeasurements and α-spectrometric measurements show that

14 S. Stumpf et al. / Journal of Colloid and Interface Science 318 (2008) 5–14

Cm(III) is not desorbed from the surface and mobilized bythis “incorporation” process. But, the spectroscopic results canbe interpreted by the formation of quartz/Cm(III)/silicic acidternary complexes at the surface.

Acknowledgment

We thank Dr. André Rossberg for providing the factoranaly-sis program code designed for the analysis of EXAFS, UV–visand TRLFS spectra.

References

[1] H. Geckeis, R. Klenze, J.I. Kim, Radiochim. Acta 87 (1999) 13.[2] Th. Stumpf, Th. Rabung, R. Klenze, H. Geckeis, J.I. Kim, J. Colloid In-

terface Sci. 238 (2001).[3] Th. Stumpf, A. Bauer, F. Coppin, J.I. Kim, Environ. Sci. Technol. 35

(2001) 3691.[4] S. Stumpf, Th. Stumpf, C. Walther, D. Bosbach, Th. Fanghänel, Ra-

diochim. Acta 94 (2006) 243.[5] J. Tits, Th. Stumpf, Th. Rabung, E. Wieland, Th. Fanghänel, Environ. Sci.

Technol. 37 (2003) 3568.[6] Th. Stumpf, J. Tits, C. Walther, E. Wieland, Th. Fanghänel, J. Colloid

Interface Sci. 276 (2004) 118.[7] Th. Stumpf, Th. Fanghänel, J. Colloid Interface Sci. 249 (2002) 119.[8] Th. Rabung, D. Schild, H. Geckeis, R. Klenze, Th. Fanghänel, J. Phys.

Chem. B 108 (2004) 17160.[9] Y. Takahashi, T. Kimura, Y. Kato, Y. Minai, T. Tominaga, Radiochim.

Acta 82 (1998) 227.[10] K.H. Chung, R. Klenze, K.K. Park, P. Paviet-Hartmann, J.I. Kim, Ra-

diochim. Acta 82 (1998) 215.[11] R.K. Iler, in: The Chemistry of Silica, Wiley–Interscience, New York,

1997.[12] M. Dietzel, Geochim. Cosmochim. Acta 64/19 (2000) 3275.[13] W. Stumm, J.J. Morgan, in: Aquatic Chemistry, Wiley–Interscience, New

York, 1981, pp. 540–541.

[14] G. Lagerstrom, Acta Chem. Scand. 13 (1959) 722.[15] N. Ingri, Chem. Scand. 13 (1959) 758.[16] M. Dietzel, in: I. Stober, K. Bucher (Eds.), Water–Rock Interaction,

Kluwer Academic, Netherlands, 2002, pp. 207–235.[17] W.J. Weber, W. Stumm, J. Inorg. Nucl. Chem. 27 (1965) 237.[18] M.A. Kim, P.J. Panak, J.I. Yun, J.I. Kim, R. Klenze, K. Köhler, Colloids

Surf. A 216 (2003) 97.[19] P.J. Panak, M.A. Kim, J.I. Yun, J.I. Kim, Colloids Surf. A 227 (2003) 93.[20] P. Venema, T. Hiemstra, W.H. Van Riemsdijk, J. Colloid Interface Sci. 181

(1996) 45.[21] D.C. Grahame, Chem. Rev. 41 (1947) 441.[22] T. Hiemstra, W.H. Van Riemsdijk, J. Colloid Interface Sci. 179 (1996)

488.[23] G.H. Bolt, J. Phys. Chem. 61 (1957) 1166.[24] L. Pauling, J. Am. Chem. Soc. 51 (1929) 1010.[25] J. Westall, FITEQL: A Computer Program for Determination of Chemi-

cal Equilibrium Constants from Experimental Data, Version 2.0’, Report82-02, Dept. of Chemistry, Oregon State University, Corvallis, OR, USA,1982.

[26] E.P. Poeter, M.C. Hill, Documentation of UCODE, a computer code foruniversal inverse modeling, US Geological Survey Water-Resources In-vestigations Report 98-4080, 1998, 116 pp.

[27] R. Siever, Am. Mineral. 42 (1957) 821.[28] S.M. Elgawhary, L. Lindsay, Soil Sci. Soc. Am. Proc. 36 (1972) 439.[29] T. Kimura, G.R. Choppin, J. Alloys Compd. 213/214 (1994) 313.[30] A. Roßberg, T. Reich, G. Bernhard, Anal. Bioanal. Chem. 376 (2003) 631.[31] T. Hiemstra, J.C.M. De Wit, W.H. Van Riemsdijk, J. Colloid Interface

Sci. 133/1 (1989) 105.[32] W.H. Van Rimsdijk, J.C.M. de Wit, L.K. Koopal, G.H. Bolt, J. Colloid

Interface Sci. 116 (1987) 511.[33] T. Hiemstra, W.H. Van Rimsdijk, M.G.M. Bruggenwert, Neth. J. Agric.

Sci. 35 (1987) 281.[34] T. Hiemstra, W.H. Van Rimsdijk, G.H. Bolt, J. Colloid Interface Sci. 133

(1989) 91.[35] T. Hiemstra, P. Venema, W.H. Van Rimsdijk, J. Colloid Interface Sci. 184

(1996) 680.[36] P.J. Panak, M.A. Kim, R. Klenze, J.I. Kim, Th. Fanghänel, Radiochim.

Acta 93 (3) (2005) 133.