Macroscopic and spectroscopic investigations on Eu(III) and Cm(III) sorption onto bayerite (β-Al(OH)3) and corundum (α-Al2O3)

Journal of Colloid and Interface Science 461 (2016) 215–224

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science journal homepage: www.elsevier.com/locate/jcis

Macroscopic and spectroscopic investigations on Eu(III) and Cm(III) sorption onto bayerite (b-Al(OH)3) and corundum (a-Al2O3) Tomas Kupcik a,⇑, Thomas Rabung a, Johannes Lützenkirchen a, Nicolas Finck a, Horst Geckeis a, Thomas Fanghänel b,c a b c

Karlsruhe Institute of Technology, Institute for Nuclear Waste Disposal, P.O. Box 3640, 76021 Karlsruhe, Germany University of Heidelberg, Institute of Physical Chemistry, Im Neunheimer Feld 253, 69120 Heidelberg, Germany Joint Research Centre – JRC, Rue du Champs de Mars 21, 1050 Brussels, Belgium

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


Sorption studied by macroscopic

(batch sorption) and microscopic (TRLFS) techniques.
Similar pH dependent sorption behaviour at trace metal ion concentrations.
Surface transformation of the oxide surface into an hydroxidic structure.
Metal ion sorption via singly coordinated aluminol groups.

a r t i c l e

i n f o

Article history: Received 20 May 2015 Revised 4 August 2015 Accepted 8 September 2015 Available online 8 September 2015 Keywords: Bayerite Corundum Curium Europium Aluminum oxides Aluminum hydroxides Sorption Time resolved laser fluorescence spectroscopy (TRLFS) Isoelectric point

a b s t r a c t The interaction of trivalent Cm and Eu with the aluminum hydroxide bayerite (b-Al(OH)3) and the aluminum oxide corundum (a-Al2O3) was investigated by batch sorption experiments and time resolved laser fluorescence spectroscopy (TRLFS). The experimental methods for both polymorphs show similar pH dependent sorption behavior at trace metal ion concentrations (107 M), i.e. similar Eu sorption edges and nearly identical Cm speciation between pH = 3 and 13. In this pH range the Cm aquo ion as well as the Cm(III) surface species surface  Cm(OH)x(H2O)5x (x = 0, 1, 2) can be distinguished by TRLFS. The similar sorption data point to a (surface) transformation of the thermodynamically unstable Al2O3 surface into bayerite, in agreement with the similar isoelectric points obtained for both minerals (pHIEP = 8.6–8.8). The pH dependent surface charge is most likely due to the protonation/deprotonation of singly coordinated Al–OH surface groups, prevailing on the edge planes of the rod-like bayerite crystals and the surface of the colloidal Al2O3 particles. These surface groups are also believed to act as ligands for lanthanide/actinide(III) surface complexation. In contrast to the similar sorption behavior at trace metal ion concentrations, discrepancies are observed at higher Eu levels. While similar sorption edges occur up to 7  107 M Eu for corundum, the pH edge on bayerite is gradually shifted to higher pH values in this Eu concentration range. The latter behavior may be related either to the existence of multiple sorption sites

⇑ Corresponding author at: Institute for Nuclear Waste Disposal (INE), Karlsruhe Institute of Technology (KIT), Postfach 36 40, D-76021 Karlsruhe, Germany. E-mail addresses: [email protected] (T. Kupcik), [email protected] (T. Rabung), [email protected] (J. Lützenkirchen), [email protected] (N. Finck), [email protected] (H. Geckeis), [email protected] (T. Fanghänel). http://dx.doi.org/10.1016/j.jcis.2015.09.020 0021-9797/Ó 2015 Elsevier Inc. All rights reserved.

216

T. Kupcik et al. / Journal of Colloid and Interface Science 461 (2016) 215–224

with different sorption affinities, or to the influence of an additional amorphous Al-phase, forming in the course of the bayerite synthesis. Ó 2015 Elsevier Inc. All rights reserved.

1. Introduction Fundamental knowledge of the parameters controlling the interaction of aquatic species with mineral surfaces is crucial for assessing the fate of pollutants in geosystems. In this respect, the chemical nature and reactivity of mineral/water interfaces are decisive for the retention of metal ions. Frequently a-Al2O3 (corundum) and c-Al2O3 have been taken as model minerals for the investigation of metal ion sorption at variable metal ion concentration and pH [1–5]. Even though pure aluminum oxides are not very abundant in nature, reactions with aluminol groups are thought to be relevant for natural systems. They exist at surfaces of aluminosilicates and their alteration products which are present in soil and subsoil environments [6,7]. Especially the secondary minerals of aluminosilicates belonging to the clay group are known to be highly efficient in metal ion retention [8,9]. In addition, aluminum oxides hold as isomorphous model phases for ferric oxides/ hydroxides, known for their strong sorptive properties in geosystems. Aluminum oxides are available as high purity solids with well-defined crystal cuts so that their crystal/solution interface can be studied in great detail. The structure of corundum and the aluminum hydroxides gibbsite (a-Al(OH)3) and bayerite (b-Al(OH)3) consists of hexagonal close packed layers of AlO6 octahedra, where 2/3 of the octahedral sites are occupied by Al atoms. In the case of corundum, the aluminum atoms are slightly displaced toward the free octahedral sites, thus lie slightly above and below the mid-plane between the oxygen layers. In contrast, the aluminum atoms are located directly in the mid-plane of the adjacent oxygen layers for the aluminum hydroxides, with an hydrogen attached to each oxygen forming a hydroxyl ion. A structural difference is apparent for c-Al2O3, where Al is also located in tetrahedrally coordinated interstitial sites (AlO4) [10,11]. Investigations with aluminum oxides are complicated by their instability with regard to dissolution and secondary phase formation when in contact with aqueous solution for more than 10 h [12]. The thermodynamically more stable gibbsite (a-Al(OH)3) appears to form predominantly in the acidic pH range, while bayerite (b-Al(OH)3) forms as a kinetically favoured phase at near-neutral pH and above. Lee and Condrate [13] observed the presence of a mixture of gibbsite and bayerite on a-Al2O3 in an aqueous solution using diffuse reflectance Fourier transformed infrared (DR-FTIR) spectroscopy. In the case of aged c-Al2O3 suspensions, the surface is transformed into bayerite as indicated by acid-base and dissolution rate measurements [14], Raman spectroscopy and X-ray diffraction [14,15], as well as Fourier transformed infrared (FTIR) spectroscopy [15–17]. Based on the results of Lefèvre et al. [14], surface transformation is characterized by an induction period of about four days with the formation of a transient amorphous hydrated phase, after which the bayerite content increases until it levels off after two months. Gibbsite was not found by these authors, indicating that under a wide pH range bayerite is a kinetically stabilized polymorph being formed in aqueous suspensions. Different types of surface groups are present in different ratios on the surfaces of the crystalline aluminum oxides and hydroxides. Bayerite crystals appear as rods usually well developed in z direction, which yields a strong contribution from 0 1 0 and 1 0 0 faces

[18]. On these crystal planes, singly and doubly coordinated aluminum groups can be found, i.e. the surface hydroxyl is bound to one (Al–OH) or two Al atoms (Al2–OH) of the solid structure. As the latter are ‘‘set deeper in the surface”, they are believed to be sterically hindered and therefore less reactive [11]. In contrast, only doubly coordinated aluminum groups exist on the ideal 0 0 1 basal plane. In the case of crystalline a- and c-Al2O3, triply coordinated Al3–OH entities may also occur on the hydroxylated surfaces [19]. However, spherical aluminum oxide powders consist of edges, steps, vacancies and defect sites and are composed of a high fraction of singly coordinated aluminum groups [11]. As indicated above, there is a structural difference between the latter minerals, in that effect that also tetrahedrally coordinated Al is present in c-Al2O3. Due to the lower coordination compared to the AlO6, these sites are assumed to have a different reactivity. However, as c-Al2O3 is, at least superficially, transformed into an hydroxidic structure [14–16], these sites are not relevant for metal ion sorption. Which specific types of surface aluminol moieties act as coordinating surface ligands to bind metal ions still appears to be under debate (see. e.g. [19,20], or [21] for the case of Pb(II) adsorption). Their affinities toward metal ion sorption is to some extent related to the reactivity of the respective aluminol sites with regard to protonation/deprotonation. As a result of experiments and modeling, the coordination of surface oxygen groups to Al-atoms in the crystal structure is believed to mainly determine the acidity of the aluminol groups. The singly and triply coordinated aluminol groups are charged and/or proton reactive over a broad pH range [18], the corundum (0 0 1) plane (as the isomorphous bayerite and gibbsite basal plane) exclusively exposes doubly coordinated hydroxyl groups and is expected to be uncharged and of low reactivity. However, an unexpected pH dependent charge has been observed and mono- and divalent ions appear to adsorb on this crystal plane [22,23]. The doubly coordinated hydroxyl groups on gibbsite basal planes were also postulated to be proton active with pKa values around 4–5.2 [24–26], in line with a pHIEP at about 4–5 observed by streaming potential measurements on the corundum (0 0 1) plane [23]. Moreover, surface roughness, or steric aspects concerning the ordering of surface functional groups at the crystal plane might as well play a significant role for hydrated metal cation sorption as the structure of surface bound water layers [27] and crystal imperfections [28]. The aim of the present work is to apply a comprehensive approach combining classical batch sorption experiments with spectroscopy in order to provide insight into surface speciation of metal ions on model mineral phases, acting as surrogates for naturally occurring minerals (e.g. clays or iron oxides). Hence, trivalent metal ion sorption onto corundum (present work) and c-alumina [29,30] is compared to the secondary polymorph bayerite (b-Al(OH)3) in order to identify differences and similarities. We focus on trivalent metal ions such as lanthanide and actinide ions, which strongly adsorb on metal oxide surfaces. Notably the actinide ions are of specific interest in the context of radionuclide behavior in nuclear waste disposal systems. The present study covers wet chemistry (batch sorption studies with the lanthanide Eu) and spectroscopic investigations (using time resolved laser fluorescence spectroscopy (TRLFS) with the actinide Cm as a fluorescent probe) on corundum and bayerite. The latter yields spectroscopic

217

T. Kupcik et al. / Journal of Colloid and Interface Science 461 (2016) 215–224

insight into metal ion surface speciation at trace metal ion concentration. Earlier studies investigating Cm(III) interaction by TRLFS with c-alumina [29,30], well defined sapphire single crystal planes [2] and gibbsite [31] revealed formation of inner-sphere surface complexes in all cases. Fluorescence lifetime analysis suggests predominant formation of Al–O–Cm(OH)x(H2O)5x type complexes at c-alumina, gibbsite and sapphire (0 0 1) surfaces. At sapphire crystal cuts other than the (0 0 1) face longer lifetimes indicate a higher denticity of the Cm(III) surface complex with a smaller number of water molecules (2.5–3.2) left in the first coordination sphere [2]. The investigations presented here supplement the existing sorption data for trivalent lanthanides and actinides on aluminum oxides and hydroxides and shall help to improve the understanding of Ln(III)/An(III) interaction with oxide surfaces. 2. Experimental 2.1. Chemicals and materials

a-Al2O3 powder (corundum) was obtained from Taimei Chemicals, Tokyo, Japan (TAIMICRON TM-DAR). According to the technical data the purity of the solid material is >99.99% and therefore the powder was used without a further purification. It was characterized by XRD, SEM and BET prior to the experiments. The synthesis of bayerite roughly followed a procedure described by Lefèvre et al. [32]. An aluminate solution (0.1 M) (OH/Al ratio of 5) was prepared by mixing AlCl3  6H2O and 1 M NaOH. A volume of 200 mL of this solution was heated in a polyethylene vessel to 50 °C in a water bath with argon gas over the solution. Acid (1 M HCl) was added at a flow rate of 7 mL h1 by a peristaltic pump until the pH decreased to 9. The obtained precipitate was left in the liquid overnight at room temperature and subsequently washed four times with deionized water. The solid powder was obtained by freeze drying. The mineralogical purity, morphology and surface area of the final product was examined by XRD (D8 Advance diffractometer (Bruker) equipped with a Cu Ka source and an energy dispersive (Sol-X) detector), SEM (CamScan CS44FE) and N2-BET (AUTOSORB-1, Quantachrome Corporation). To determine the isoelectric point of both corundum and bayerite, f-potential measurements were performed with a 1 g L1 mineral suspension in 0.1, 0.01 and 0.001 M NaCl, respectively. 2.2. Batch sorption study The macroscopic sorption behavior of europium, a common lanthanide analogue to trivalent actinides [33], onto corundum and bayerite was investigated by batch experiments performed in a glove box (O2 < 1 ppm) to exclude atmospheric CO2. An ICP stan-

dard Eu solution (MerckÒ; [Eu] = 6.6  103 M) was diluted to secondary standards (6.6  105 M and 6.6  107 M) prior to use. For adsorption experiments, a 20 g L1 mineral suspension was diluted in 0.1 M NaClO4 and aliquots of the Eu standard solutions were added to give a metal ion concentration between 6.6  109 and 6.6  105 mol L1 and a solid-to-liquid (S:L) ratio of 6 g L1 and 1 g L1 for corundum and bayerite, respectively. For bayerite the S:L ratio was lower due to the limited amount of solid material available. The experimental conditions are summarized in Table 1. Eu was added at low pH values (pH  4) and the target pH in the various batches was adjusted to the desired value by adding CO2-free NaOH solutions (0.001–0.1 M). To simplify the system, no buffer solution was used. The samples were shaken periodically for 7 days to reach equilibrium. Subsequently the solid was separated from the liquid phase by ultracentrifugation at 18,000 rpm, which had been checked before to be sufficient for complete removal of the solid phase. Several aliquots (up to three) of the supernatant solution were sampled and Eu concentrations analyzed by ICP-MS. The error bars represent the standard deviation of these sub-samples. The pH dependent Eu uptake was expressed as a sorption edge, either as percentage uptake (%), or as distribution ratios Kd (L kg1) vs. pH. The latter is defined in the usual manner as:

Kd ¼

ctot  ceq V  m ceq

ð1Þ

where ctot is the initial aqueous metal ion concentration (mol L1), ceq is the measured Eu concentration in the supernatant (mol L1), V is the volume of the liquid phase (L) and m is the mass of the solid phase (kg). The presentation in terms of Kd (L kg1) has the advantage, that the solid-to-liquid ratio does not affect the presentation of the data. Surface normalized Kd values (L m2) will be used to compare the sorption of trivalent metal ions onto the different aluminum oxide and hydroxide minerals, to further include the dependence of the metal ion uptake on solid materials on their surface area [34]. 2.3. TRLFS study Like in the batch sorption study, sample preparation for the TRLFS experiments was performed in a glove box under argon atmosphere. The S:L ratio was reduced 0.5 g L1 to minimize light scattering effects. For corundum and bayerite, aliquots of a 20 g L1 mineral suspension were diluted in 0.1 mol L1 NaClO4 and spiked with a 248-Cm stock solution (t1/2 = 3.5  105 years; 2  105 mol L1 stored in 0.1 M HClO4; isotopic composition: 89.7% Cm-248, 9.4% Cm-246, 0.4% Cm-243, 0.3% Cm-244, 0.1% Cm-245 and 0.1% Cm-247) to give a fixed metal ion concentration of 2  107 M (Table 1). Starting in the acidic pH range (pH  4), the suspensions were shaken periodically for 2–3 days to reach

Table 1 Experimental conditions for batch sorption and TRLFS experiments on the sorption of trivalent lanthanides/actinides onto corundum, c-Al2O3 and bayerite. Element

c (mol/L)

S:L (g L1)

Solvent

Ref.

c-Al2O3

Eu Eu Am Eu Eu

6.6  109–6.6  105 106 <108 6.6  109–6.6  105 9.0  108–3.1  105

6 1 1 1 3.6

0.1 M 0.1 M 0.1 M 0.1 M 0.1 M

NaClO4 NaClO4 NaCl NaClO4 NaClO4

This work [48] [47] This work [1]

TRLFS study Corundum Bayerite c-Al2O3

Cm Cm Cm

2  107 2  107 2.5  107

0.5 0.5 0.57

0.1 M NaClO4 0.1 M NaClO4 0.1 M NaClO4

This work This work [30]

Material Batch sorption study Corundum

Bayerite

218

T. Kupcik et al. / Journal of Colloid and Interface Science 461 (2016) 215–224

sorption equilibrium, transferred into cuvettes, sealed and used for TRLFS. After the measurements, the solution was pipetted back into the reaction vials and the pH was increased in small steps by adding CO2-free sodium hydroxide solutions (0.001– 0.1 mol L1). As for the batch experiments, no buffer was used to keep the system as simple as possible. The TRLFS measurements were performed with a pulsed Nd:YAG pumped dye laser system (Continuum, Powerlite, ND 6000, laser dye; Exalite 398). The Cm fluorescence emission was detected by an optical multichannel analyser consisting of a polychromator (Chromex 250) with a 1200 lines mm1 grating. To filter out light scattering and background fluorescence, the emission spectra were recorded in the range 580–620 nm, 1 ls after the exciting laser pulse in a time window of 1 ms at kex = 396.6 nm. For lifetime measurements, the time delay between the laser pulse and the camera gating was scanned with time intervals between 10 and 15 ls. The laser pulse energy, controlled by a photodiode, was between 2.0 and 3.0 mJ.

Table 2 BET surface area and the isoelectric points (pHIEP) for different several aluminum oxides and hydroxides.

a b

Material

N2-BET (m2 g1)

pHIEP

Ref.

Corundum

14.5 15.0 12.0

8.8 9.4a 9.1b

This work [48] [47]

c-Al2O3

119

8.6a

[1]

Bayerite

14.0

8.7

This work

Gibbsite

49.5

11.1

[31]

Point of zero charge (PZC). Point of zero net proton charge (PZNPC).

3. Results 3.1. Material characterization 3.1.1. Corundum SEM pictures of the commercially obtained corundum powder show a homogenous size distribution of the particles with diameters ranging from 150 to 200 nm (Fig. 1a). Due to their spherical shape, distinct crystal faces cannot be distinguished. XRD data (Fig. 1 supplementary information) indicate that the sample consists of corundum, with trace amounts of diaspore (peak at 39.0° 2h) and NaCl (peak at 31.6° 2h). Due to the nature and amount of the impurities, interference with the sorption studies is not expected. The specific surface area was determined to be 14.5 m2 g1 by N2-BET (Table 2). An isoelectric point (pHIEP) of pHIEP 8.8 was obtained by microelectrophoresis in 0.1–0.001 M NaCl (Fig. 2a), close to literature data on the same material (pHIEP = 9.1 [35]) and close to the majority of isoelectric points reported for colloidal a-alumina [36]. 3.1.2. Bayerite Mineralogically pure and crystalline bayerite was identified by XRD (Fig. 1 supplementary information). SEM pictures show 1–5 lm long rod shaped, but irregular crystals with a heterogeneous size distribution (Fig. 1b). The particle shape is dominated by the edge surfaces, while the basal plane has only a small contribution to the total surface area. Beside the well-crystalline phase, small particles with mixed morphologies are visible, pointing to the presence of amorphous or weakly crystalline Al(OH)3. The specific surface area was 14.0 m2 g1 (N2-BET method, Table 2), lower than the specific surface area of bayerite synthesized under similar experimental conditions (27 and 44 m2 g1; [32]). The isoelectric

Fig. 2. f-potential as a function of pH for 1 g L1 (a) corundum and (b) bayerite in 1–100 mM NaCl.

point was determined at pHIEP 8.7 by microelectrophoresis in 0.001–0.1 M NaCl background electrolyte solutions (Fig. 2b). The value of 8.7 for the pHIEP agrees with published isoelectric points measured for bayerite by the same method, which are around

Fig. 1. SEM images of (a) corundum powder and (b) bayerite. For the latter, beside the rod shaped crystals an amorphous phase is visible.

T. Kupcik et al. / Journal of Colloid and Interface Science 461 (2016) 215–224

pHIEP 9 [18,36,37]. Significantly lower pHIEP have been reported for different preparation conditions and with varying experimental methods (e.g. potentiometric and mass titrations; see [36,37]). Our bayerite sample exhibits a much larger variability in size and crystal shape as compared to corundum.

219

saturation effects). The latter were postulated for the interaction of Eu and Am with c-Al2O3 in analogy to investigations on natural hematite [39], or in describing metal ion sorption to hydrous ferric oxides [38]. However, the existence of sorption sites with different sorption properties on c-Al2O3 was not observed by Cm-TRLFS investigations [29].

3.2. Batch sorption experiments 3.2.1. Eu sorption onto corundum The pH-dependent uptake of Eu onto corundum for different metal ion concentrations (6.6  109–6.6  105 M) is shown in Fig. 3, both as percentage of sorbed Eu and log Kd. As expected, a strong increase in the europium uptake is noticed in a narrow pH range. As an initial experiment performed with 6.6  107 M Eu indicated an almost quantitative sorption at pH values above 8, the investigated pH range for the other experiments was reduced (3.5 < pH < 8.0). Note that in the log Kd vs. pH graph, a constant value of 5.2 L kg1 occurs for pH P 6 and 6.6  107 M (Fig. 3b). The remaining Eu concentrations in solution under these conditions (60.1% of initial concentration) are within the limit of quantification of ICP-MS and the value of 5.2 can be interpreted as a lower limit for log Kd. For the lowest Eu concentration, sorption starts at pH  4.5 and is almost complete at pH 6.0 on the percentage uptake scale. Sorption edges for europium concentrations up to 6.6  107 M are similar, indicating ideal sorption behavior within this Eu concentration range. As pure electrostatic interactions of the positively charged metal ion with the mineral surface can be ruled out due a pHIEP of 8.8 (Table 2), the assumption of inner-sphere coordination to the corundum surface is reasonable. At higher metal ion concentrations the sorption edge is shifted to higher pH values. The non-linear sorption behavior in this Eu concentration range can be related to an increasing electrostatic repulsion of adsorbed Eu ions [19,38], or to the presence of different surface sites (site

Fig. 3. pH-dependent Eu sorption onto corundum (6 g L1) at different Eu concentrations in 0.1 M NaClO4 plotted as (a) %-sorbed vs. pH and (b) distribution ratio, log Kd, against pH. When not shown, the experimental error is within the data points.

3.2.2. Eu sorption onto bayerite The pH-dependent sorption of Eu onto bayerite is presented in Fig. 4 (again as percentage of sorbed Eu and log Kd). Note that in contrast to corundum, the quantification limit of ICP-MS (60.1% of initial Eu concentration remaining in solution) refers to a log Kd value of 6.0. For the lowest metal ion concentration (6.6  109 M) the sorption edge is situated between pH 4 and 6 (Fig. 4a), where the surface is positively charged (pHIEP = 8.7; Table 2). Following the interpretation for corundum, this again points to innersphere coordination. The sorption edge for [Eu] = 6.6  108 M is already shifted to higher pH values, whereas almost congruent sorption edges are detected for the three highest Eu concentrations (Eu = 6.6  107–6.6  105 M). This uptake behavior significantly differs from that observed for corundum. Following the discussion in the previous section, the non-linear sorption behavior at low Eu concentrations (6.6  109–6.6  107 M) may be ascribed to the existence of several distinctive sorption sites with, however, low abundance (at least lower compared to corundum) but different sorption affinities. By increasing the Eu concentration these sites are successively saturated, which may lead to the observed gradual shift of the sorption edge to higher pH values. Subsequently, only one sorption site (with a high capacity, but having a low affinity toward Eu) is responsible for metal uptake at Eu concentrations above 6.6  107 M. Alternatively, the presence of an amorphous phase (as indicated by SEM results; Fig. 1b) with concomitant

Fig. 4. pH-dependent Eu sorption onto bayerite (1 g L1) at different Eu concentrations in 0.1 M NaClO4 plotted as (a) %-sorbed and (b) log Kd against pH. When not shown, the experimental error is within the data points. Note that the detection limit of the method is 6 L kg1.

220

T. Kupcik et al. / Journal of Colloid and Interface Science 461 (2016) 215–224

stronger binding sites can possibly influence Eu uptake at trace metal ion concentrations. As macroscopic batch sorption studies are not suitable to investigate the underlying sorption mechanisms in detail, TRLFS experiments with the actinide curium provide further information on the metal ion complexes formed at mineral–water interfaces [5]. 3.3. TRLFS experiments 3.3.1. Cm sorption onto corundum Fig. 5a presents the evolution of the fluorescence emission spectra for Cm (2  107 M) in a corundum suspension (0.5 g L1) between pH 4 and 13. For better visibility only eight emission spectra at selected pH values are depicted. As in previous investigations on Cm sorption onto c-Al2O3 or clay minerals [29,30,40] only the Cm3+ aquo ion is visible below pH 4. An increase in pH leads to a broader emission spectrum and a shoulder at higher wavelengths becomes visible (Fig. 5a). Its fraction increases with increasing pH, while the Cm3+ aquo species contribution decreases. By peak deconvolution, i.e. by subtracting the spectrum of the well-known Cm3+ aquo from the respective emission spectrum, a single component spectrum with an emission peak maximum at 600.7 nm can be derived. The shift of the emission peak maxima to higher wavelengths compared to the free Cm3+ aquo ion indicates a change in the Cm ligand field and the formation of a Cm surface complex [30]. Subsequently, this single component spectrum can be used as an input parameter to deconvolute the emission spectra at higher pH values. By this method, three single component spectra (in addition to free Cm3+ aquo) are derived in the pH range 4 < pH < 13, with kmax = 600.7, 603.0 and 605.4 nm (Fig. 5b and Table 3). The positions of the emission peak maxima are consistent with the emission bands observed for the interaction of Cm with c-Al2O3 [30]. Using these single component emission spectra and the spectrum for Cm3+ aquo, a pHdependent Cm species distribution can be calculated (Fig. 6). For

pH < 5 the Cm speciation is dominated by Cm3+ aquo, while for pH P 5 the first Cm surface complex (Cm species 1) is formed, its fraction increasing to 50% at pH 6.0. At this pH value, the second Cm surface species appears with its abundance increasing and dominating the species distribution up to pH  10. Beyond a third Cm surface species prevails. In order to calculate an accurate species distribution, relative fluorescence intensity factors (FIi) for the different Cm species have to be taken into account [41]. By fixing the FIi factor for the Cm aquo ion to FI (Cm3+ aquo) = 1, the relative fluorescence intensity factors for the three curium surface species are FI (Cm species 1) = 0.55, FI (Cm species 2) = 0.21 and FI (Cm species 3) = 0.22, which agrees well with FIi factors for Cm/c-Al2O3 (FI = 0.45, 0.29 and 0.26 [30]). The decrease in FI by inner-sphere complexation can be related to a shift of the absorption maximum for surface Cm compared to the free Cm ion in aqueous solution when the exciting wavelength is kept constant [42]. In general, the differences in FIi factors for the identified curium species arise from differences in the molar absorptivity at a given excitation wavelength, the fluorescence quantum yield and the overall efficiency of the instrumentation used. 3.3.2. Cm sorption onto Bayerite Selected emission spectra for the interaction of curium with bayerite in the pH range between 4 and 13 are shown in Fig. 5c. Three distinct Cm surface species were identified with kmax at 600.6, 603.6 and 606.7 nm, respectively (Fig. 5d and Table 3). As for corundum, the single component emission spectrum for the third Cm species resembles the one recorded at pH 12.9. The emission peak maxima are in close agreement with the results obtained for corundum and c-Al2O3 [30]. The extracted relative fluorescence intensity factors are FIi (Cm3+ aquo) = 1, FI (Cm species 1) = 0.63, FI (Cm species 2) = 0.19 and FI (Cm species 3) = 0.05. While the former resemble the ones determined for corundum and c-Al2O3 [30], FI for Cm species 3 is significantly lower. Similar (i.e. low) FIi factors

Fig. 5. Selected fluorescence emission spectra (kex = 396.6 nm) at various pH and single component emission spectra derived by peak deconvolution for Cm (2  107 M) in contact with 0.5 g L1 a-Al2O3 (a and b) and bayerite (c and d). For a better view, the spectra are scaled to the same peak height.

221

T. Kupcik et al. / Journal of Colloid and Interface Science 461 (2016) 215–224

Table 3 Emission peak maxima (kmax) for the Cm3+ aquo ion and the Cm species formed on the surfaces of several aluminum oxides and hydroxides together with the determined lifetimes (s) and the number of H2O/OH entities in the first Cm coordination sphere.

a b

kmax (nm)

s (ls)

n (H2O/OH)

Ref.

Cm3+ aquo

593.8 Cm species 1

64 ± 1

9

[45]

Cm species 2

Corundum

600.7

603.0

605.4

110

5±1

This work

c-Al2O3

600.6 601.2

602.5 603.3

605.7 –

110 110

5±1 5±1

[30] [29]

Bayerite

600.6

603.6

606.7

110

5±1

This work

Gibbsite

–

603.0

605.2a

140–150a

3.5–3.8/2.4–2.7

[31]

Sapphire 0 0 1

601.3

107 ± 1

5

[2]b

Sapphire 1 1 0

603.6

192 ± 3

2.5–3.2

[2]b

Cm species 3

Apart from surface sorbed Cm also an incorporated Cm species was detected (kmax = 609 nm; s = 180–200 ls) [31]. TRLFS only performed at pH 4.5 and 5.1.

Fig. 6. Cm species distribution derived by peak deconvolution of the TRLFS emission spectra for Cm in aqueous corundum (full symbols) and bayerite suspension (open symbols). The lines are included to guide the eye.

Fig. 7. Cm fluorescence decay and the calculated fluorescence lifetimes for the Cm3+ aquo ion at pH = 1, Cm in aqueous corundum and bayerite suspensions at pH  7. The fitted lines refer to a fluorescence lifetime of s = 110 ls.

were also reported for the interaction of curium with clay minerals [40,43] and related in this case to a transfer of energy from the excited Cm state to the clay mineral structure by a phonon assisted processes. As these processes are presently not understood in detail, a clear explanation for the low FI factor for the third Cm surface species in the case of bayerite cannot be given here. As for a-Al2O3, the Cm species distribution for pH < 5 is dominated by Cm3+ aquo, while the three Cm surface species are formed successively with increasing pH (Fig. 6). The Cm distribution closely resembles the one for corundum.

fluorescence lifetime of the Cm3+ aquo (s = 64 ± 1 ls; 9 H2O ligands in the first Cm coordination sphere). A further increase in pH is accompanied by an increase in fluorescence lifetime to s = 110 ± 10 ls at pH  7 (Fig. 7 and Table 3), which remains constant up to pH 13. Based on the latter lifetime 5 ± 1 H2O/OH entities should remain in the first coordination sphere. Assuming that the total coordination number of 9 for Cm remains constant, the loss of water ligands has to be compensated by coordinating oxygen atoms from the mineral surface, indicating the formation of inner-sphere-bound curium.

3.3.3. Fluorescence lifetimes Besides the number of distinct Cm species in mineral suspension, TRLFS can also provide insight into the hydration state of Cm(III). The fluorescence lifetime s can be correlated to the number of OH and/or H2O entities in the vicinity of the curium ion, which act as fluorescence quenchers due to energy transfer of the excited Cm f state to high-energy OH vibronic states [44]. Adsorption of the free Cm ion leads to a loss of H2O or OH ligands from the first Cm coordination sphere and consequently to a decreased number of fluorescence quenchers, thus increasing the value of s. Based on an empirical linear equation [45,46], the number of H2O or OH units can be determined from the fluorescence lifetime. For bayerite and corundum the fluorescence lifetimes in the weakly acidic pH range (4–5) are consistent with the

4. Discussion 4.1. Eu/Cm sorption onto b-Al(OH)3, a- and c-Al2O3 and the chemical nature of lanthanide/actinide(III) surface complexes A comparison of the present results together with literature data on the interaction of trivalent lanthanides and actinides with aluminum oxides and hydroxides at trace Eu concentrations is given in Fig. 8a and b [1,47,48]. The metal ion uptake is normalized to the reactive surface area (N2-BET). Based on the experimental conditions summarized in Table 1, similar sorption behavior is observed for a- and c-Al2O3 and bayerite at metal concentrations ranging from 6  108 to 2  107 M (Fig. 8a). Furthermore, the results by both batch sorption and laser spectroscopy are in good

222

T. Kupcik et al. / Journal of Colloid and Interface Science 461 (2016) 215–224

the Cm(III) ion, a very similar chemical environment of the innersphere bound metal ion on those surfaces can be deduced. According to the fluorescence lifetime analysis given before, the initial sorption reaction onto the mineral surfaces and the formation of the first surface Cm complex can therefore be schematically expressed by Eq. (2), while the increase in pH is expected to cause hydrolysis of the surface complexes (Eqs. (3) and (4)).

h i surface þ Cm3þ ðH2 OÞ9 $ surface    Cm3þ ðH2 OÞ5 þ 4H2 O

ð2Þ

h i h i surface    Cm3þ ðH2 OÞ5 $ surface    CmðOHÞ2þ ðH2 OÞ4 þ Hþ ð3Þ h i   surface   CmðOHÞ2þ ðH2 OÞ4 $ surface  CmðOHÞþ2 ðH2 OÞ3 þ Hþ ð4Þ

Fig. 8. (a) Eu and Cm uptake on different aluminum oxides and hydroxides at trace metal ion concentrations investigated by batch sorption experiments and laser fluorescence spectroscopy (TRLFS) (Rabung et al. (2000) [1]: S:L = 3.6 g L1, Eu = 9  108 M, 0.1 M NaClO4; Rabung et al. (2006) [30]: S:L = 0.57 g L1, Cm = 2.5  107 M, 0.1 M NaClO4; Huittinen et al. [31]: S:L = 2.2 g L1, Gd = 7  108 M, 0.1 M NaClO4). (b) Comparison of Eu and Am sorption data onto a-Al2O3 at trace metal ion concentrations. (Janot et al. [48]: S:L = 1 g L1, Eu = 106 M, 0.1 M NaClO4; Alliot et al. [47]: S:L = 1 g L1, Am < 108 M, 0.1 M NaClO4).

agreement. Note that Kd values from TRLFS data treatment are not considered when the contribution of the sorbed species based on peak deconvolution is <10% of the total intensity, i.e. the estimated error of the method. While the present sorption edge for a-Al2O3 is in good agreement with data reported by Janot et al. [48], significantly higher Kd values have been reported by Alliot et al. [47] for americium uptake (Fig. 8b). A straightforward explanation for this discrepancy is not available. The similarity of sorption data for a- and c-Al2O3 and bayerite supports previous reports that Al2O3 surfaces in aqueous solution are converted to bayerite. We also found that they exhibit very similar surface charge properties, denoted by the almost identical pHIEP = 8.6–8.8 (Fig. 2 and Table 2). About 25% of the initial material were found to be converted into Al(OH)3 phases after 5 years contact time for colloidal c-Al2O3 [30]. As the pH dependent surface charge for bayerite (but also for non-transformed Al2O3) is mostly due to the protonation/deprotonation of singly coordinated Al–OH groups, these surface groups may in turn be assigned as relevant coordinating entities for lanthanide/actinide(III) surface complexation. Finally, TRLFS investigations of Cm(III) sorption onto a-, c-Al2O3 and bayerite show very similar spectral features. Three Cm surface species with similar single component emission spectra and emission lifetimes successively appear with increasing pH (Table 3). As TRLFS spectra are very sensitive to the first coordination sphere of

Application of the linear correlation between the Cm fluorescence lifetime and the number of H2O/OH quenchers implies two assumptions: First, no quench processes except those originating from H2O or OH ligands occur in the vicinity of Cm. Experiments with Cm adsorbed to c-Al2O3 performed in D2O demonstrated that fluorescence in this case is not quenched by the solid [49]. However, MoO2 4 entities in powellite were assumed to quench the fluorescence for incorporated Cm [50], though a detailed explanation for this energy transfer was not given. Second: the quenching behavior of H2O and OH is similar. While this hypothesis was used by several authors (and adopted to different Cm/mineral systems, see e.g. [40,43]), the H2O entity can also be regarded as equivalent to two OH units. Following this interpretation, a H2O ligand provides two OH oscillations, making the quenching process for a water molecule twice as effective as for OH [51]. However, a closer spatial OH/Cm distance with respect to H2O may also enhance its quenching properties. Slight variations in Cm–H2O and Cm–OH distances in surface complexes were indeed deduced from EXAFS (extended X-ray absorption fine structure) studies [30,49]. The discussion above shows that the interpretation of luminescence lifetimes as to the structure of adsorbed Cm species has to be considered with some care. Despite the uncertainties related to the interpretation of fluorescence lifetimes the speciation proposed in Eqs. (2)–(4) is supported by quantum chemistry calculations and EXAFS studies. A tri- to fourfold coordination of trivalent lanthanide and actinide ions to surface oxygen atoms and 5–6 water molecules is concluded for various metal oxides [30,49,52,53]. As the Cm fluorescence lifetime of 110 ls does not vary up to pH = 13, the existence of ternary complexes with other ligands like aluminates can be excluded at high pH. In the case of aluminosilicates such ternary complexes including silicate ions have been suggested by Huittinen et al. [54]. The similar sorption properties determined by batch sorption and spectroscopic experiments point to a comparable chemical environment and a similar coordination geometry for the surface sorbed metal ion on the investigated aluminum oxide and hydroxide surfaces. Which aluminol group the metal ion binds to (e.g. singly, doubly and triply coordinated aluminol groups) cannot be inferred by the methods applied in the present study. However, Bargar et al. [19] describe a metal cation coordination to Al–OH and Al3–OH surface groups for colloidal a-Al2O3, while Huittinen et al. [55] state that the protons of both the singly and doubly coordinated entities on c-Al2O3 are being removed upon Y(III) sorption. The former study was performed with divalent Pb and the results cannot be transferred directly to trivalent lanthanides/actinides due to different ionic radii [56] and different ionic charge. In case of bayerite, Al–OH and Al2–OH groups can be considered as being

T. Kupcik et al. / Journal of Colloid and Interface Science 461 (2016) 215–224

responsible for metal ion uptake. Siretanu et al. [22] find Ca2+ adsorbing to the Al2–OH groups on the gibbsite (0 0 1) basal plane and Lützenkirchen et al. [23] conclude that Al3+ ions as well interact with doubly coordinated hydroxyl groups on the isostructural sapphire (0 0 1) surface. There are indications from spectroscopic and quantum chemical studies that surface complexes might have slightly variable structures and chemical environments. Fluorescence line narrowing experiments suggest that the single component fluorescence emission bands isolated by peak deconvolution (Fig. 5) cover several Cm inner-sphere surface complexes characterized by a certain variability [57]. Density Functional Theory (DFT) studies on the interaction of trivalent lanthanum with a cluster model for sapphire (0 0 1) and (1 1 0) surfaces (taken as an analogue for the crystal surfaces present on bayerite) report energetically distinct but close energy minima [52,53]. Both findings suggest that surface complexes are characterized by some disorder and that the surface bound metal ions seemingly coexist in several slightly different chemical environments, where binding to multiple surface groups is possible. 4.2. Eu/Cm sorption onto gibbsite (a-Al(OH)3) and sapphire (a-Al2O3) single crystals While the sorption edges for bayerite and a- and c-Al2O3 at trace metal concentrations are almost identical, the one for the second aluminum hydroxide polymorph gibbsite (a-Al(OH)3) is shifted to higher pH values (Fig. 8; [31]). Note that a Gd(III) instead of Eu(III) was used by the authors. In addition, a Cm species with an emission peak maximum similar to the ones found for bayerite as well as a- and c-Al2O3 was absent in their TRLFS emission spectra in the lower pH range [31]. Both results were attributed to the higher isoelectric point (pHIEP = 11.1; Table 2) of the material due to a higher contribution of Al2–OH sites compared to bayerite (and the hydroxylated a- and c-Al2O3 surfaces) [18]. However, this interpretation can be questioned and the gibbsite sorption behavior may also be explained by blocking of the available surface sites (in the case of gibbsite these are mainly the Al2–OH groups located on the 0 0 1 basal plane) by sorbed Al. Such reactions on the isostructural 0 0 1 sapphire surface were recently reported for Alions [23] and Al13 polycations [58], as well as for Ca2+ on the gibbsite basal plane [22]. For the sapphire single crystals, which can be taken as model surfaces for aluminum hydroxides, the emission peak positions and fluorescence lifetimes for Cm sorbed on the (0 1 2), (1 1 0), (0 1 8), and (1 0 4) planes [2] are similar to the ones observed by Huittinen et al. [31] on gibbsite (Table 3), pointing to comparable chemical environments. The spectroscopic results for the sapphire (0 0 1) surface differ from the other crystal planes, i.e. the emission peak maximum is shifted to lower wavelengths and the Cm fluorescence lifetime is significantly lower, indicating a lower denticity of the Cm(III) surface complex with a higher number water molecules left in the first coordination sphere (5) [2]. A respective spectroscopic fingerprint is absent in gibbsite. Consequently Cm seems to be solely associated with the gibbsite crystal edges and sorption onto the basal plane seems to be absent. As discussed above, this result can be interpreted in terms of competition effects occurring on the gibbsite basal surface, i.e. sorbed cations compete with Cm (added at trace concentrations) thus suppressing its sorption on the 0 0 1 surface. With respect to bayerite (and hence a- and c-Al2O3) the information gained by a comparison to the sapphire single crystals is ambiguous: while the fluorescence lifetime for Cm sorbed on b-Al(OH)3 equals to one determined for the sapphire (0 0 1) surface, i.e. s = 110 ls, the emission peak maxima show significant differences (Table 3). Thus a conclusive interpretation of these results cannot be given here. However, the results point to the fact that metal ion coordination to the sapphire

223

0 0 1/bayerite surfaces is, at least, comparable. As mentioned in the last paragraph, the metal ion may reside in slightly different chemical environments and the Cm inner-sphere surface complexes may exhibit a certain variability. In order to determine the distinct structure of these surface complexes, EXAFS investigations on Gd(III) interaction with gibbsite, bayerite and a-Al2O3 are on the way. 5. Conclusions Our present results suggest similar sorption behavior of trivalent lanthanides/actinides on corundum (a-Al2O3), c-Al2O3 and bayerite (b-Al(OH)3) at trace metal ion concentrations ([M]  107 M) by both macroscopic batch sorption experiments and time resolved laser fluorescence spectroscopy (TRLFS). The latter method indicates the formation of three inner-sphere surface Cm species, i.e. surface  Cm(OH)x(H2O)5x (x = 0, 1, 2), in the pH range between 4 and 13, in agreement with data for c-Al2O3 [30]. These results agree with a surfacial phase transformation of the aluminum oxides to bayerite resulting in a similar chemical environment for Eu/Cm surface complexes on the three nominally distinct sorbents. Based on the available datasets and by comparison with literature results, it is not possible to define the exact type of surface sites (singly-, doubly- or triply coordinated aluminol groups) participating in Eu/Cm coordination. However, the results suggest that not only one, but several similar Cm surface species coexist simultaneously, differing slightly in the nature of the coordinating ligands, the binding angles, etc. As the coordination environment of surface sorbed (metal) ions can be probed by extended X-ray absorption fine structure (EXAFS) spectroscopy, this method will be applied in a subsequent study to determine the distinct metal ion coordination to aluminum oxides and hydroxides. While similar sorption properties are observed at Eu/Cm concentrations of 107 M, the uptake behavior under variable metal ion concentrations is different. Ideal Eu sorption is apparent for corundum for [Eu] ranging from 7  109 M to 7  107 M, but non-ideal behavior is observed in the case bayerite under similar conditions. The latter cannot be explained based on our present batch and TRLFS investigations, but may be related to a stepwise saturation of multiple, different sorption sites within this Eu concentration range. On the other hand, Eu sorption can also be influenced by an additional (likely amorphous) phase, which is indeed visible by SEM. Additionally, our findings indicate similarities to published data for the interaction of Cm with illite, smectite [40,59] and kaolinite [43,54], i.e. formation of 3–4 Cm surface species with increasing pH and comparable Cm emission spectra and fluorescence lifetimes. Acknowledgements This work was supported by the Stiftung Energieforschung Baden-Württemberg funded by ENBW (EnBW Energie Baden-Württemberg AG). The authors would like to thank T. Kisely (N2-BET determination) and E. Soballa (SEM). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jcis.2015.09.020. References [1] T. Rabung, T. Stumpf, H. Geckeis, R. Klenze, J.I. Kim, Radiochim. Acta 88 (2000) 711. [2] T. Rabung, D. Schild, H. Geckeis, R. Klenze, T. Fanghanel, J. Phys. Chem. B 108 (2004) 17160.

224

T. Kupcik et al. / Journal of Colloid and Interface Science 461 (2016) 215–224

[3] Y. Tang, R.J. Reeder, Geochim. Cosmochim. Acta 73 (2009) 2727. [4] S.E. Mason, C.R. Iceman, T.P. Trainor, A.M. Chaka, Phys. Proc. 4 (2010) 67. [5] H. Geckeis, J. Lützenkirchen, R. Polly, T. Rabung, M. Schmidt, Chem. Rev. 113 (2013) 1016. [6] M.E. Essington, Soil and Water Chemistry: An Integrative Approach, CRC Press, Boca Raton, FL, 2004. [7] A.K. Karamalidis, A.D. Dzombak, Surface Complexation Modeling: Gibbsite, John Wiley & Sons, Hoboken, New Jersey, 2010. [8] L.J. Evans, Environ. Sci. Technol. 23 (1989) 1046. [9] M.F.J. Hochella, A.F. White (Eds.), Reviews in Mineralogy, Mineralogical Society of America, 1990. [10] I. Levin, D. Brandon, J. Am. Ceram. Soc. 81 (1998) 1995. [11] G.V. Franks, Y. Gan, J. Am. Ceram. Soc. 90 (2007) 3373. [12] X. Carrier, E. Marceau, J.-F. Lambert, M. Che, J. Colloid Interface Sci. 308 (2007) 429. [13] D.H. Lee, R.A. Condrate, Mater. Lett. 23 (1995) 241. [14] G. Lefevre, M. Duc, P. Lepeut, R. Caplain, M. Fedoroff, Langmuir 18 (2002) 7530. [15] C. Dyer, P.J. Hendra, W. Forsling, M. Ranheimer, Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 49 (1993) 691. [16] E. Laiti, P. Persson, L.O. Ohman, Langmuir 14 (1998) 825. [17] H. Wijnja, C.P. Schulthess, Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 55 (1999) 861. [18] T. Hiemstra, H. Yong, W.H. Van Riemsdijk, Langmuir 15 (1999) 5942. [19] J.R. Bargar, G.E. Brown, G.A. Parks, Geochim. Cosmochim. Acta 61 (1997) 2617. [20] J.R. Bargar, S.N. Towle, G.E. Brown Jr., G.A. Parks, J. Colloid Interface Sci. 185 (1997) 473. [21] S.E. Mason, C.R. Iceman, K.S. Tanwar, T.P. Trainor, A.M. Chaka, J. Phys. Chem. C 113 (2009) 2159. [22] I. Siretanu, D. Ebeling, M.P. Andersson, S.L.S. Stipp, A. Philipse, M.C. Stuart, D. van den Ende, F. Mugele, Sci. Rep. 4 (2014). [23] J. Lützenkirchen, A. Abdelmonem, R. Weerasooriya, F. Heberling, V. Metz, R. Marsac, Geochem. Trans. 15 (2014) 9. [24] Y. Gan, G.V. Franks, Langmuir 22 (2006) 6087. [25] J. Rosenqvist, P. Persson, S. Sjöberg, Langmuir 18 (2002) 4598. [26] M.-C. Jodin, F. Gaboriaud, B. Humbert, J. Colloid Interface Sci. 287 (2005) 581. [27] J. Lützenkirchen, R. Zimmermann, T. Preocˇanin, A. Filby, T. Kupcik, D. Küttner, A. Abdelmonem, D. Schild, T. Rabung, M. Plaschke, F. Brandenstein, C. Werner, H. Geckeis, Adv. Colloid Interface Sci. 157 (2010) 61. [28] S.J. Freij, G.M. Parkinson, Hydrometallurgy 78 (2005) 246. [29] T. Stumpf, T. Rabung, R. Klenze, H. Geckeis, J.I. Kim, J. Colloid Interface Sci. 238 (2001) 219. [30] T. Rabung, H. Geckeis, X.K. Wang, J. Rothe, M.A. Denecke, R. Klenze, T. Fanghanel, Radiochim. Acta 94 (2006) 609. [31] N. Huittinen, T. Rabung, J. Lutzenkirchen, S.C. Mitchell, B.R. Bickmore, J. Lehto, H. Geckeis, J. Colloid Interface Sci. 332 (2009) 158.

[32] G. Lefevre, M. Fedoroff, Mater. Lett. 56 (2002) 978. [33] G.R. Choppin, J. Alloys Compd. 223 (1995) 174. [34] T. Rabung, Einfluss von Huminstoffen auf die Europium(III)-Sorption an Hämatit Universität des Saarlandes, Saarbrücken (1998). [35] K. Rezwan, L.P. Meier, L.J. Gauckler, Langmuir 21 (2005) 3493. [36] M. Kosmulski, Surface Charging and Points of Zero Charge, CRC Press, 2009. [37] M. Kosmulski, Adv. Colloid Interface Sci. 152 (2009) 14. [38] D.A. Dzombak, F.M.M. Morel, Surface Complexation Modeling – Hydrous Ferric Oxide, John Wiley & Sons, New York, 1990. [39] T. Rabung, H. Geckeis, J.I. Kim, H.P. Beck, J. Colloid Interface Sci. 208 (1998) 153. [40] T. Rabung, M.C. Pierret, A. Bauer, H. Geckeis, M.H. Bradbury, B. Baeyens, Geochim. Cosmochim. Acta 69 (2005) 5393. [41] T. Fanghanel, H.T. Weger, T. Konnecke, V. Neck, P. Paviet-Hartmann, E. Steinle, J.I. Kim, Radiochim. Acta 82 (1998) 47. [42] X.K. Wang, T. Rabung, H. Geckeis, P.J. Panak, R. Klenze, T. Fanghanel, Radiochim. Acta 92 (2004) 691. [43] N. Huittinen, T. Rabung, P. Andrieux, J. Lehto, H. Geckeis, Radiochim. Acta 98 (2010) 613. [44] A. Heller, J. Am. Chem. Soc. 88 (1966) 2058. [45] T. Kimura, G.R. Choppin, J. Alloys Compd. 213 (1994) 313. [46] T. Kimura, G.R. Choppin, Y. Kato, Z. Yoshida, Radiochim. Acta 72 (1996) 61. [47] C. Alliot, L. Bion, F. Mercier, P. Vitorge, P. Toulhoat, Radiochim. Acta 93 (2005) 435. [48] N. Janot, M.F. Benedetti, P.E. Reiller, Environ. Sci. Technol. 45 (2011) 3224. [49] T. Stumpf, C. Hennig, A. Bauer, M.A. Denecke, T. Fanghanel, Radiochim. Acta 92 (2004) 133. [50] D. Bosbach, T. Rabung, F. Brandt, T. Fanghänel, Radiochim. Acta 92 (2004) 639. [51] R.M. Supkowski, W.D. Horrocks Jr., Inorg. Chim. Acta 340 (2002) 44. [52] R. Polly, B. Schimmelpfennig, T. Rabung, M. Flörsheimer, R. Klenze, H. Geckeis, Radiochim. Acta 98 (2010) 627. [53] R. Polly, B. Schimmelpfennig, M. Flörsheimer, T. Rabung, T. Kupcik, R. Klenze, H. Geckeis, Radiochim. Acta 101 (2013) 561. [54] N. Huittinen, T. Rabung, A. Schnurr, M. Hakanen, J. Lehto, H. Geckeis, Geochim. Cosmochim. Acta 99 (2012) 100. [55] N. Huittinen, P. Sarv, J. Lehto, J. Colloid Interface Sci. 361 (2011) 252. [56] R.D. Shannon, Acta Crystallogr., Sect. A 32 (1976) 751. [57] M. Marques Fernandes, T. Stumpf, B. Baeyens, C. Walther, M.H. Bradbury, Environ. Sci. Technol. 44 (2010) 921. [58] J. Lützenkirchen, T. Kupcik, M. Fuss, C. Walther, A. Sarpola, O. Sundman, Appl. Surf. Sci. 256 (2010) 5406. [59] A. Schnurr, R. Marsac, T. Rabung, J. Lützenkirchen, H. Geckeis, Geochim. Cosmochim. Acta 151 (2015) 192.