Ion exchange and structural properties of a new cyanoferrate mesoporous silica material for Cs removal from natural saline waters

Journal of Environmental Chemical Engineering 5 (2017) 810–817

Contents lists available at ScienceDirect

Journal of Environmental Chemical Engineering journal homepage: www.elsevier.com/locate/jece

Ion exchange and structural properties of a new cyanoferrate mesoporous silica material for Cs removal from natural saline waters Caroline Michela,b,* , Yves Barréa , Laurent De Windtb , Caroline de Dieuleveultb , Emmanuelle Brackxc , Agnès Grandjeana a b c

CEA, DEN, DTCD, SPDE, Laboratoire des Procédés Supercritiques et de Décontamination, F-30207 Bagnols-sur-Ceze, France MINES ParisTech, PSL Research University, Centre de Géosciences, F-77300 Fontainebleau, France CEA, DEN, DTEC, Laboratoire de Métallographie et d’Analyse Chimique, F-30207 Bagnols-sur-Ceze, France

A R T I C L E I N F O

Article history: Received 20 October 2016 Received in revised form 15 December 2016 Accepted 20 December 2016 Available online 21 December 2016 Keywords: Cesium decontamination Ferrocyanide Inorganic ion exchanger Modeling Sorption Vanselow

A B S T R A C T

The demand for effective and inexpensive treatment for decontamination of waters from radionuclides, such as 137Cs, is presently high. In this context a selective adsorbent material for Cs (SORBMATECH1 202) was designed, consisting of potassium/copper ferrocyanide nanoparticles deposited in the mesoporosity of silica grains. Several batch experiments were carried out in order to obtain kinetic and thermodynamic data concerning the ion exchange. Isotherms and measurements of distribution coefficients (Kd,Cs) in waters of increasing salinity (pure water, Ca-bicarbonate fresh water and seawater) were conducted in a radioactive environment. This study shows a fast ion-exchange kinetics (<5 min), due to the open silica porosity combined to cyanoferrate nanoparticles, as well as high selectivity with Kd,Cs in fresh and sea waters (106 and 2.0
105 mL/g, respectively). These experiments also demonstrated the competitive effects of the major cations present in natural waters (K+, Na+, Mg2+ and Ca2+) and led to the determination of the Vanselow’s selectivity coefficients. Integrating this dataset into the CHESS geochemical speciation code allowed the correct modeling of Kd,Cs values in these different water types over a wide range of Cs concentrations. The dataset can be extrapolated to the modeling of other Kferrocyanides and effluent compositions. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction The ever-increasing pressure to reduce the release of radioactive species into the environment requires constant improvement of technologies for waste treatment and for dose minimization. Recently, the Fukushima disaster required the development of appropriate processes for the removal of radionuclides resulting from accidental discharge. Various treatments may be applied depending on the effluent’s composition and radionuclide targeted for extraction (co-precipitation, ion exchange, sorption) [1,2]. Ion exchange in a fixed-bed column is one of the most common and efficient methods to treat radioactive effluents. Moreover, this process produces a minimal amount of final waste. Among these radionuclides, 137Cs is considered the most abundant and hazardous element due to its presence in many types of waste and its relatively long half-life (30 years). A number

* Corresponding author. Present address: CEA, DEN, DTCD, Laboratoire de Développement des Procédés de Vitrification, F-30207 Bagnols-sur-Ceze, France. E-mail address: [email protected] (C. Michel). http://dx.doi.org/10.1016/j.jece.2016.12.033 2213-3437/© 2017 Elsevier Ltd. All rights reserved.

of studies have been carried out on the extraction of cesium using inorganic ion exchangers such as zeolite [2,3], silicotitanate [4] or inorganic phosphate [5]. Nevertheless, cyano-bridged coordination polymers based on hexacyanometallates and transition metal ions KxMy[Fe(CN)6]z (where M = a bivalent transition metal ion), also called Prussian blue analogous (PBA), present a higher selectivity for the extraction of radioactive cesium ions from contaminated waters over a wide range of pH values [6–14]. These bulk materials are actually used in stirred reactors at industrial scale for the selective extraction of 137Cs+ from contaminated effluents. However, the main drawback of these materials is related to their small grain size, preventing their use in fixed-bed columns due to pressure loss and clogging. In powder form, their low adsorption kinetics is also a problem for a column process. One way to overcome these drawbacks is to insert a selective hexacyanoferrate compound in a solid porous matrix so that it becomes suitable for a column process: the porosity of the support and the nanosize of the hexacyanoferrate inside the porosity improve the adsorption rate. Since a few years, numerous composite solids (organic or inorganic) loaded with hexacyanoferrate particles have

C. Michel et al. / Journal of Environmental Chemical Engineering 5 (2017) 810–817

been proposed for cesium removal [15–22]. Most of these composites use silica supports and are obtained by successive impregnation of porous silica or polymer modified silica by bivalent transition metal ions and a hexacyanoferrate precursor. These materials are able to reach high capacity and selectivity, but the kinetics of sorption is relatively slow which is an obstacle to their use in a continuous process, and moreover in a column process. However, their final compositions are not well controlled and thus their sorption properties are relatively poorly reproducible. Some of us have recently reported an original approach for preparing porous silica-based nanocomposites containing Prussian blue-type nanoparticles (3–6 nm) covalently grafted inside the pores of silica [15,17]. These nanocomposites were obtained by successive coordination of cobalt ions Co2+ and hexacyanoferrate [Fe(CN)6]3 (Fe(III)) precursors on amino sites present on the surface of the support. However, these materials were synthesized with expensive reagents, organic solvents and many steps, hindering the industrialization of this process. Moreover, the hexacyanoferrate particles thus obtained do not contain alkali ions inside the crystal lattice. The sorption capacity is therefore low and the Cs adsorption process leads to the release of the divalent metal that could be an issue for the final decontaminated liquid effluent. Here we use a new simpler synthetic route based on silica supports containing pure potassium copper hexacyanoferrate (with Fe(II)) KCuFC nanoparticles inserted inside the pores. This new material called Sorbmatech 202 (S202) could also be easily synthesized on a large scale. In this paper we report the structural characterization and the ion exchange properties of S202, as well as very fast ion exchange ability and high selectivity. This makes S202 promising for an efficient decontamination of saline effluent by a column process. From the perspective of industrial development, ion exchange occurring in the case of KCuFC must be perfectly mastered and modeled. In this work, several water compositions were studied, from pure water (deionized water) to a Ca-bicarbonate fresh water and seawater. This study further demonstrates that the competitive effect of ions present in natural waters is only active when cesium is at trace concentrations. In this study, first the exchange reaction coefficients between the cation K+ from the solid exchanger S202 and the cations from the aqueous solutions (Cs+, Na+ and Mg2+) were experimentally determined through isotherms using 133Cs and a doped 137Cs solution in batch mode. Then, ion exchange data were used to determine the full set of selectivity coefficients. Subsequently, modeling of the isotherms was performed for each water composition to check dataset consistency and the ability to model the cesium uptake by this new material in complex aqueous solutions with competitive cations. These coefficients have been set into the CHESS geochemical speciation code allowed the correct modeling of Kd,Cs values in these different water types over a wide range of Cs concentrations and they can be extrapolated to other K-ferrocyanide and many effluent compositions. 2. Material and method 2.1. Material synthesis The porous support consisted of a commercial silica-gel (SIGMA ALDRICH 35–60 mesh) with a mean initial pore volume of 0.75 cm3/g, a mean pore size diameter of 6 nm, and a specific area of 500 m2/g. This support was loaded with nanoparticles of KCuFC distributed over the porosity. For this, the silica-gel was functionalized with a two-step method: first APTES (3-Aminopropyl)-triethoxysilane was grafted on the silica surface by a hydrolysis-condensation reaction with silanol groups on the

811

surface. This grafting was performed in absolute EtOH (99.9%) solvent at 60  C over 24 h. The amino-grafted silica was then analyzed by TGA to calculate the amount of amino group inserted into the silica support: about 0.150 mmol/g of amino group was grafted onto silica. Details of the synthesis procedure is given in [23]. The growth of KCuFC nanoparticles (NP) was first initiated by contacting the amino silica with a 102 mol/L Cu(NO3)2 solution at room temperature and with stirring over 24 h. Silica granules turned instantly blue. Then the solid was washed with pure water and filtered. The second step was to add this solid to a solution composed of K+ and Fe(CN)64 agitated for 24 h. Both these steps, Cu-solution and then K-solution, consisted of an impregnation cycle. Three impregnation cycles were performed to obtain the final material. 2.2. Characterization method The concentrations of K, Cu and Cs in an aqueous solution from non radioactive batch sorption experiments were analyzed by the following three methods: inductively-coupled plasma mass spectrometry (ICP-MS Thermo Scientific) used for very low Cs concentrations, Atomic Absorption Spectrometry (AAS with a PERKIN ELMER spectrometer) used for higher Cs concentrations and inductively-coupled plasma atomic emission spectroscopy (ICP-AES Thermo Scientific) for other cations. Radiochemical analyses were performed by gamma counting (Eurisys, measured with a germanium detector) to analyze the amount of residual 137 Cs in the radioactive solutions. To determine the chemical composition of the materials, a sample of approximately 50 mg was weighed for mineralization. This aliquot was dissolved in a hot (130  C) HNO3/HF solution leading to brown precipitates. This mixture was filtered and rinsed with pure water, and a first solution (50 mL) was obtained (Solution 1). The residue remaining in the filter was calcined for 5 h at 900  C. After cooling, the deposit obtained was digested in HNO3, resulting in a second solution (Solution 2). These solutions were analyzed by ICP-AES (atomic emission spectrometry coupled with a plasma torch) Thermo ICAP 7400DV for the elements K, Cu, Fe and Si. X-ray diffraction data were collected using an X’Pert PROPANalytical apparatus with a Cu anticathode. The angular range of 2u = 10–70 was scanned at 0.08 min1 using 0.02008 steps. The KCuFC raw phases were analyzed by Rietveld refinement, using TOPAS 4.2 software and Le Bail pattern matching [24]. The profile parameters (cell, dimensions, peak shape . . . ) were refined. The peak shape was described by a pseudo-Voigt function with the Caglioti formula [25]. Surface area was obtained using nitrogen adsorption isotherms on an ASAP2020 apparatus from Micrometrics. Samples were degassed under vacuum at 70  C over two days prior to analysis. Surface area was determined using the Brunauer-Emmet-Teller (BET) method. Pore distribution was determined using the BarretJoyner-Halenda (BJH) method. SEM image analysis were performed with a high-resolution MERLIN field-emission SEM (Carl Zeiss). TEM image were collect with a MET FEI Titan Themis apparatus. 2.3. Batch sorption experiments Sorption experiments were performed in batch mode using a rotating agitator at room temperature by contacting the solid with the solution. The solid and liquid phases were separated by filtration through a 2 mm syringe filter and the liquid was analyzed

812

C. Michel et al. / Journal of Environmental Chemical Engineering 5 (2017) 810–817

3. With a seawater solution doped with radioactive initial activity of 42.5 kBq/L.

to determine the ion exchange capacity as following: Q ¼ ðC 0  CÞ:

V m

137

Cs with an

ð1Þ

where Q is the Cs exchange capacity (meq/g or mmol/g), C0 is the initial Cs concentration in solution (meq/L or mmol/L), C is the remaining concentration of Cs in solution (meq/L or mmol/L), V is the volume (L) and m the mass of adsorbent involved (g). It is worth noting that concentrations expressed in meq/L and mmol/L are identical in the case of a monovalent cation such as Cs+. 2.3.1. Kinetic experiment To determine the time to reach equilibrium, the amount of Cs adsorbed (Cs exchange capacity) on the solid is plotted against time. These sorption kinetics were followed by contacting 100 mg of solid with 100 mL of a solution containing an initial cesium nitrate concentration of 7.520104 mol/L and 0.01 mol/L sodium nitrate as a basic salt for 24 h. Each point of the kinetic curve corresponds to a separate sorption test. After each time period, the shaking was stopped, and the solution was filtered with a 0.2 mm syringe filter and analyzed. 2.3.2. Isotherm experiments To establish an adsorption isotherm, the Cs concentration in solution at equilibrium Ceq (mol/L) is compared with the solid ion exchange capacity (see Eq. (1)). The relationship Qeq = f(Ceq) is known as the “sorption isotherm”. Isotherm experiments consisted of contacting 100 mg of adsorbent with 100 mL of a solution containing an initial cesium concentration varying from 0 to 2.25
103 mol/L. The contact time was set to 24 h, which was more than sufficient to reach equilibrium as discussed below. Isotherms were performed at different pH values and with different water compositions. The pH was adjusted before the sorption with 0.1 mol/L HNO3/NaOH. The influence of water composition was performed using fresh water and seawater. The compositions of these aqueous solutions are given in Table 1. After 24 h of stirring, the solutions were filtered with a 0.2 mm syringe filter and analyzed. Radioactive isotherms were performed with trace concentration of Cs under different operating conditions: 1. At pH = 7 with a solution containing radioactive 137Cs with an initial activity of 144 kBq/L (3.3
107 mol/L) and stable 133Cs ranging from 0 to 1.0
107 mol/L. 2. With an initial solution containing radioactive 137Cs with an initial activity of 50 kBq/L and containing a K+ concentration varying from 1.0
105 to 1.0
102 mol/L (the latter is the seawater concentration).

Table 1 Chemical composition of the fresh water and seawater used in this study; the total ion concentrations are given in mol/L (N.D. = non determined). Ion

Fresh water

Seawater

Na+ K+ Mg2+ Ca2+ Cl SO42 HCO3 B Br NO3 SiO2(aq) pH (natural buffer)

2.8
104 4.4
105 1.1
103 2.0
103 1.9
104 1.3
104 5.9
103 N.D. N.D. 6.0
105 2.5
104 7.9

0.485 0.0106 0.0550 0.0107 0.566 0.0293 5.9
103 4.3
104 8.7
104 N.D. N.D. 8.3

For all experiments, a volume of 50 mL of the studied solution was stirred with 50 mg of solid over 24 h. Then, the solutions were filtered with a 0.2 mm syringe filter and the remaining 137Cs concentration was measured. Assuming that both radioactive and stable Cs were equally adsorbed onto the material, the determination of the amount of the radioactive Cs allowed the determination of the residual traces of stable Cs and the calculation of the sorption capacity at equilibrium. 2.3.3. Experiments with competitive ions The effect of Na+ was studied by varying the amount of Na+ in the solution from 0.1 mol/L to 0.5 mol/L (Na+ concentration in seawater) with an initial Cs+ concentration of 3.750105 mol/L. These solutions were stirred for 24 h with 1 g/L of adsorbent, then filtered with a 0.2 mm syringe filter for analysis. The same approach was performed concerning Mg2+ effect on Cs sorption. The concentration range of Mg2+ was chosen to scan the concentration that can be found in fresh water and seawater (from 1
102 to 5
102 mol/L). 2.4. Cation exchange reactions and formalism A general ion exchange reaction may be written as: zb Azaþ þ za Bzbþ Ð za Bzbþ þ zb Azaþ

ð2Þ

where A and B are multivalent cations with za and zb positive charge, respectively. Azaþ stands for the exchangeable ion in the solid whereas Azaþ is the ion in solution. Since seawater and fresh water have a significant amount of competitive cations, complementary ion exchange reactions were characterized (i.e. K+$Na+; 2 K+$Mg2+;). The exchange between protons from the solution with exchangeable K+ from the solid was also studied (K+$H+). Considering Eq. ()2, a law of mass action at thermodynamic equilibrium is described as follows: 0 za
zb B B2þ Þ Aþ A=B B ð3Þ K ¼ @  za zb Aþ Þ B2þ There are different formalisms to estimate the ion activity fixed into the solid. This study uses the Vanselow’s formalism where the activity of an ion bound to the solid is calculated on the molar fraction scale. This formalism is based on a selectivity coefficient, A=B , and is commonly used in the literature on ion exchange noted K v;s (e.g. [31]). The activity of the B2+ ion in the solid phase is

considered to be equal to the molar fraction of B2+, X B , which in the case of binary exchange can be written as: 2 6 X B ¼ 4h

B2þ  h A  þ B2þ  þ

ð4Þ

where the square brackets represent concentrations in mole per liter of solution or mole per gram of solid. Non idealities in the solid exchanger were not taken into account in this study. 2.4.1. Cation exchange capacity (CEC) The cation exchange capacity (CEC) corresponds to the maximum amount of ions that the exchanger can exchange. This value can be estimated from the chemical formula of the material. The CEC should not be confused with the maximum experimental capacity, Qmax, which is determined by isotherm experiment. In

C. Michel et al. / Journal of Environmental Chemical Engineering 5 (2017) 810–817

this study, the CEC corresponds to the K+ ions present in the S202 material. When the adsorbent completely exchanges all available K+, the maximum experimental Cs+ capacity is equal to the CEC. 2.4.2. Distribution coefficient Kd The distribution coefficient Kd is defined as follows: Kd ¼

½Csþ 
1000 ½Csþ 

ð5Þ

where Kd is the distribution coefficient in mL/g, ½Csþ  is the concentration on the solid in mol/g and ½Csþ  is the concentration remaining in the solution in mol/L. This parameter is an indicator of the performance of the adsorbent. When the latter is used to decontaminate trace concentration of Cs, Kd is constant and can therefore be compared with other adsorbents used under the same experimental conditions. 2.4.3. Modeling code The CHESS geochemical code and the EQ3/6 thermodynamic database [26] were used to model the aqueous chemistry (acid/ base and complexation reactions, quartz solubility), as well as cation exchange, at equilibrium without kinetics. The activity coefficients of species in solution were calculated with the B-dot model applicable for low to highly mineralized solutions such as seawater (i.e. ionic strength up to unity). 3. Experimental results and discussion 3.1. Adsorbent characterization 3.1.1. Porous structure and microstructure A type IV adsorption-desorption of N2 was obtained for both materials with an H1 hysteresis loop characteristic of a mesoporous structure (Fig. 1-A). This demonstrates the preservation of the porous structure after growth of the KCuFC nanoparticles. The total pore volume for the silica support was equal to 0.79 cm3/g whereas it was equal to 0.57 cm3/g after incorporation of nanoparticles. BET specific area calculated for the silica support was 534 m2/g compared to 283 m2/g for the KCuFC-loaded material. The pore size distribution calculated using the BJH model is reported on Fig. 1-B. The growth of nanoparticles in the porous structure of silica led to clogging of the smaller pores. The significant decrease of both the mesoporous volume and the specific area clearly demonstrates that KCuFC nanoparticles are inserted inside the porosity leading to a modification of the porous structure.

813

The shape form and average size of S202 grain was collected through SEM analysis in Fig. 2-A). S202 granules were closed to regular octahedron with a volume shape factor of 0.471. The average size of S202 was 482 mm. The microstructure of the loaded materials was analyzed by TEM. A lamella was taken from inside of a grain of S202. The slide was cut and collected by FIB (Focused Ion Beam). A TEM image is shown on Fig. 2-B). In this figure, the white spots are the silica pores. Their size was about 60 Å, which is in good agreement with the results of the BJH method (Fig. 1-B). The black spots are KCuFC nanoparticles. The nanoparticles in this image were 8 nm across on average. This particle size is in good agreement with the Scherrer formula [27] applied to the XRD spectrum. Nanoparticles were well distributed in the silica pores, which was an advantage ensuring high sorption rates. Indeed, under these conditions, rate limitation due to intraparticle diffusion became less important. 3.1.2. Crystalline structure and chemical composition The loaded material before and after the sorption experiment was analyzed by XRD (the diffractograms are given in Fig. S1 as Supporting information). The XRD results showed that the KCuFC nanoparticles are crystalline with a face-centered cubic lattice. The Cu2+ and Fe(CN)64 ions form the lattice while K+ ions are located in the center of the tetrahedral sites in cubic lattice, balancing with the negative charge. These K+ ions are mobile and are able to exchange with other monocations such as Cs+ from an aqueous solution. The amorphous phases correspond to the support material consisting of amorphous silica. The Rietveld refinement performed on the diffraction was used to determine a lattice parameter of 7.08 Å. This value is slightly smaller than for the K2CuFe(CN)6 used for the refinement (7.064 Å). This could be explained by the fact that the chemical formula of our KCuFC was between KCu1.5Fe(CN)6 and K2Cu Fe(CN)6. After the sorption experiment, the shift of the peaks at 2u = 25.05 , 36 , 51.8 and 58.1 indicates the incorporation of Cs+ which is a larger ion than K+. The disappearance of the peaks at 2u = 18 , 40.3 and 55 and the appearance of the peak at 2u = 30.8 shows that all K+ was totally exchanged. The average chemical composition of the loaded material was analyzed by elementary analysis, using three material batches in order to assess of the reproducibility and robustness of the synthesis method. The chemical formula of KCuFC nanoparticles inserted into the silica support was written according to the K2xCu2-xFe(CN)6 formula [28]. The analytical results are reported in Table 2, for the three batches.

Fig. 1. A) Nitrogen adsorption isotherms for the initial silica-gel material (SiO2) and after KCuFC nanoparticle growth (S202). B) Pore diameter distribution from nitrogen BJH adsorption.

814

C. Michel et al. / Journal of Environmental Chemical Engineering 5 (2017) 810–817

Fig. 2. A) SEM picture of S202 grain B) TEM image of the S202 material; the white spots correspond to the silica pores, the black spots to KCuFC nanoparticles.

3.2. Cesium sorption experiments 3.2.1. Sorption kinetics Kinetic experiment showed that more than 90% of the total Cs capacity is reached within less than 5 min (see Fig. 3). The sorption rate is extremely high compared to literature data for loaded ferrocyanide materials [29,30] and for bulk ferrocyanides [6,31], which require a dozen hours to reach equilibrium. This property is of great importance in fixed-bed continuous mode, as in a column process. Indeed, the higher the sorption rate, the higher the flow rate in the column. This high rate results from the porosity of the materials allowing fast diffusion of Cs inside the solid support on the one hand, and the nanosize of the KCuFC particles inserted into this porosity leading to fast diffusion inside the adsorbent on the other hand. 3.2.2. Sorption isotherms Two main parameters can be derived from sorption isotherms. First, the initial slope at low concentration is linked to the affinity of Cs for the solid. At trace concentration, this slope is often used as the distribution coefficient Kd. The second parameter corresponds to the plateau reached by the isotherm curves. This is the experimental sorption capacity of Cs, which has to be compared to the theoretical maximum CEC of the solid. The sorption isotherms are reported as Supplementary information (see Figs. S2 and S3).

maximum Cs+ trapped in the solid was 0.22 meq/g, which is less than K+ released in solution. Since no other ion was present in solution, the excess of K+ in solution is attributed to an exchange with protons from the solution[6,28]. At a pH = 3, a partial oxidation of the KCuFC occurred, which led to a restructuring of the material and the release of Cu2+ into solution that was not due to an ion exchange [6,22,32,33]. No traces of Cu nor Fe were detected in solution for pH > 3. The difference between K+ release and Cs+ sorption decreases while pH increases up to pH = 9. At this pH value, the concentration of protons is negligible and one K+ exchanges with one Cs+. Moreover, the concentration of the released K+ was close to the CEC, for which all K+ available for an ion exchange were exchanged with Cs+ from solution. This trend is advantageous for industrial applications because, for example, in the case of KNiFC [29] in a supported material, not only K+ but also part of the Ni2+ was also involved in the exchange. At maximum capacity, the exchange was only about 20% of the CEC. Eventually, the experimental capacity of the S202 material tends to decrease at pH  10. This pH trend has already been observed in the literature for bulk ferrocyanides [13], probably because hexacyanoferrate are no more chemically stable. 3.2.2.2. Role of competitive ions. For both solutions, plateaus were reached for a minimal aqueous Cs concentration at equilibrium of 7.52
104 mol/L. This threshold is higher than the one measured

3.2.2.1. Role of pH. The maximum sorption capacities are rather similar over the full pH range 3–10. This is in agreement with the literature showing that pure potassium hexacyanoferrates, and especially KCuFC, are selective for Cs sorption over a large range of pH values.The concentrations of Cs+ depleted from the solution were compared with the K+ concentration released by the material into the solution. At pH = 3, the K+ concentration released in solution is equal to the CEC (CEC = 0.28 meq/g). However, the Table 2 Elementary analysis of the S202 exchanger and calculation of the theoretical CEC based on the K+ stoichiometry. Sample

Cu/Fe

K/Fe

Chemical formula

CEC meq/g

KCuFC@SiO2  1 KCuFC@SiO2  2 KCuFC@SiO2  3

1.46 1.41 1.42

1.08 1.18 1.16

K1.08Cu1.46Fe(CN)6, 57 Si(O)2 K1.18Cu1.41Fe(CN)6, 58 Si(O)2 K1.16Cu1.42Fe(CN)6, 59 Si(O)2

0.286 0.305 0.295

Fig. 3. Kinetics of Cs sorption on the S202 material ([NaNO3] = 0.01 mol/L, [Cs+]i = 7.4
104 mol/L, pH = 6.5).

C. Michel et al. / Journal of Environmental Chemical Engineering 5 (2017) 810–817

for isotherms performed on pure water (3.76
104 mol/L). This is a sign of a loss of selectivity in these more complex solutions that contain many competitive cations (K+, Na+, Mg2+, Ca2+). Nevertheless, the maximum sorption capacity in fresh water is still high (0.286 meq/g) and closed to the maximum capacity obtained for pure water at pH = 9 (around analytical error). The concentration of K+ released by ion exchange is 0.295 meq/g, which is close to the concentration of Cs+ in the solid (0.286 meq/g). As for pure water, one Cs+ is exchanged with one K+. In contrast, for seawater, the maximum sorption capacity is lower (0.202 meq/g). This can be attributed to the higher concentrations of competitive ions. Similar trends have been observed in the literature while using high concentrations of Mg(NO3)2 in solution with a bulk KCuFC material for Cs+ removal [33]. Distribution coefficients Kd,Cs were calculated for each of the isotherm experiments, as a function of Cs+ remaining in solution. Isotherms performed with radioactive 137Cs (deionized water at pH 7, fresh water and seawater) were also used to calculate Kd,Cs for very low Cs concentration. Each Kd,Cs point is reported in Fig. 4. A first zone corresponds to high concentrations of Cs (i.e. [Cs+]eq > 1.00106 mol/L), where log(Kd,Cs) linearly decreased with log ([Cs+]eq). The slopes of these linear regressions vary between 0.910 and 0.981 depending on the nature of the water. A slope value of 1 means that ion exchange involved only one ion from the material with one ion from the solution [28], in this case the exchange reaction between one Cs+ and one K+. The second zone in Fig. 4 corresponds to constants Kd,Cs values at low Cs+ concentrations. In this region the competitive ion effect is comparatively more effective. Kd,Cs is about 2.50106 mL/g at pH = 7 without competitive cation but decreases by one order of magnitude in seawater (Kd = 20105 mL/g). The same applied to the Kd value in fresh water (Kd = 106 mL/g), although it decreases at a lower rate since the concentration of competitive cations is lower than in seawater. Despite these competitive effects, this material is undoubtedly a good candidate to decontaminate seawater effluent contaminated with traces of 137Cs. A Kd,Cs in seawater of 20105 mL/ g means that 200 L of seawater can be decontaminated of 137Cs with only 1 g of the S202 material. 3.2.2.3. Selectivity coefficients  vanselow model. K/Cs exchange: The K/Cs exchange reaction was first studied using a pure water solution containing different concentrations of K+ (from 1.00105 to 1.00102 mol/L) doped with radioactive 137Cs with an initial activity of 50 kBq/L. This experiment was then compared to sorption isotherms performed in fresh water and seawater. The selectivity coefficient was derived from the experiment at trace

Fig. 4. Variation of Kd,Cs as a function of [Cs+]eq for deionized water at different pH, fresh water and saline water.

815

concentrations of Cs and a pH of 7, conditions under which ½Csþ    and proton sorption are negligible compared to K þ . Since the activity coefficients of K+ and Cs+ in solution are equal, the relationship is as follows: K d;Cs ¼

K=Cs K v;s  CEC  1000  þ K

ð6Þ

where Kd,Cs is in mL/g and CEC is in eq/g or mol/g. The plot of Kd,Cs as a function of 1/[K+] gave a straight line. The linearity indicates K=Cs is a constant value over the range of K+ concentrations, that K v;s K=Cs ) = 3.71. leading to log(K v;s For all the other isotherm experiments performed at higher concentrations of Cs, whatever the type of solution, the selectivity

is no longer a constant value but its logarithm coefficient K K=Cs v;s linearly decreases with the molar fraction of Cs, X Cs , as shown in Fig. 5. Similar trends have been observed with massive hexacyanoferrate in previous studies [28,31]. K/Na exchange: Three ion exchange reactions, were considered to calculate the selectivity coefficients, involving cation in the solid and cation in solution: K þ $Csþ , K þ $Naþ and Naþ $Csþ . The molar fraction of Cs on solid, X Cs , was not influenced by Na+ concentration in solution, X Cs varying from 0.138 to 0.139 only, while the total Na2+ concentration increased from 0.1 to 0.5 mol/L. The following mean values of selectivity coefficients were obtained, using result from the experiments with competitive ions described in the experimental

part:

Na=Cs log(K v;s ) = 5.3,

K=Cs log(K v;s ) = 3.3

and

log

K=Na ) = 2.0. Taking into account the linear dependency of log (K v;s K=Cs ) at X Cs = 0.1385, the values of the selectivity coefficients (K v;s K=Na show a good consistency since K v;s can also be deduced from       K=Na K=Cs Na=Cs log K v;s ¼ log K v;s  log K v;s ¼ 3:3  5:3 ¼ 2:0. K/Mg exchange: Lee and al. studied the competitive effect of Mg2+ ions on Cs+ sorption using bulk KCuFC [33]. They observed a slight decrease of the capacity while Mg2+ was in significant concentration.The competitive effect of Mg was more significant than in the case of

Na+, although it remained weak. The X Cs , varied from 0.215 to 0.185 while the total Mg2+ concentration increased from 0.1 to 0.5 mol/L. The mean values of the selectivity coefficients were log K=Mg ) = 11.2 and log(K v;s ) = 4.5. These results were obtained (K Mg=Cs v;s

Fig. 5. Evolution of the logarithm of K K=Cs with X Cs for a large range of Cs v;s concentrations and different solutions (pure water, fresh water and seawater).

816

C. Michel et al. / Journal of Environmental Chemical Engineering 5 (2017) 810–817

Table 3 Selectivity coefficients (Vanselow’s formalism) used for modeling. Ion exchange reaction

Log(Kv,s)

Csþ þ K þ $

Hþ þ K þ $

Naþ þ K þ $

Mg2þ þ 2 K þ $

Ca2þ þ 2K þ $

K þ þ Csþ

Hþ þ K þ

K þ þ Naþ

Mg2þ þ 2K þ

Ca2þ þ 2K þ

3:83X Cs þ 3:9

2.0

2.0

4.5

4.4

from the experiments with competitive ions described in the experimental part. K/Ca exchange: The K/Ca exchange was not determined experimentally. Since Mg2+ and Ca2+ are alkaline earth ions, we assumed that the

single ion to be exchanged with Cs+. The divalent cations Ca2+ and Mg2+ represent almost 50% of the exchanger population. In seawater the main exchangeable cations are K+ (68%) and Na+ (32%), the divalent cations being negligible.

equals K Ca=Mg . Other authors selectivity coefficient of K Ca=Mg v;s v;s working on clays mentioned the similar behavior of the two cations [34]. K/H exchange: K/H exchange was eventually investigated since the pH changed during the experiments due to exchange of K+ with H+ ions. This is reported in Table S1 (Supplementary information) for blank experiments, which consisted of placing into contact 1 g of material with 1 L of solution at a desired pH without Cs for 24 h. The exchange between protons from the solution with K+ from the solid exchanger induces an increase of pH due to proton depletion for pH 3 to 7. However, the final pH was lower than initial value for the pH 9 and pH 10 isotherms. This is due to the slight dissolution of the silica support (SiO2) under alkaline conditions that consumes hydroxyl ions OH. The selectivity coefficient was estimated from the evolution of pH with the isotherm performed

3.3.3. Modeling of the sorption isotherms While modeling isotherms, the initial pH values were set to be the same as those measured in the experiments. For instance, NaOH or HNO3 were added to the modeled solutions as in the experiments. Fig. 6 shows the Kd,Cs values and K+ concentrations released in solution as a function of the Cs concentration at equilibrium. The model is in good agreement with experiment for the medium to high Cs concentrations, whatever the water composition. The amount of K+ ions released by the material due to exchange with Cs+ is correct, as well as the decrease of Kd,Cs with Cs concentrations (i.e. Kd,Cs is no longer a constant value, but becomes Cs-dependent). Concerning the lowest Cs concentrations, both calculated and experimental Kd,Cs values are constant (characterized by a plateau in the curves of Fig. 6). However, the Kd plateau is dependent upon the type of water (Kd in pure water > Kd in fresh water > Kd in seawater). Modeling is in good agreement with experiment for pure water and fresh water, but underestimates the Kd,Cs by one order of magnitude in seawater. Seawater is a complex

K=H ) = 2.0. at pH = 7, leading to a log(K v;s

3.3. Modeling 3.3.1. Ion exchange database The database of selectivity coefficients used for modeling is reported in Table 3. A single value of selectivity coefficient was considered, excepted for the K/Cs exchange where a linear regression log(K K=Cs v;s ) = f(3:83X Cs þ 3:9) was set in order to scan the entire concentration range of Cs. The CEC was fixed to 0.286 meq/g, which corresponds to the total amount of K+ in the material. 3.3.2. Modeling of the initial cation exchange population A complementary test was conducted in order to measure the initial cation exchange population of the S202 materials in contact with fresh water without any Cs. The experiment consisted of placing into contact 1 g/L of material in freshwater over 24 h with stirring. Then, the material was placed in a solution with Cs in excess (2.26
103 mol/L) for 24 h with stirring to displace the exchangeable cation population of the exchanger. The final solution was filtered and analyzed. The molar fraction of cations are reported in Table 4. The rather good agreement between the experimental and calculated values supports the accuracy of the full database. The slight discrepancy may be due to the uncertainty K=Ca , which was not directly measured. It is worth on the value of K v;s noting that the initial state of the S202 exchanger is strongly modified while in contact with fresh water. K+ is no longer the

Table 4 Molar fractions of cation fixed on the solid without Cs in freshwater and seawater.

Freshwater Seawater

Molar fraction

XK

XNa

X Mg

XCa

Experiment Modeling Modeling

0.53 0.59 0.68

0.04 0.01 0.32

0.12 0.12 0.01

0.31 0.28 0.001

Fig. 6. Modeling of sorption isotherms of Cs+ in pure water (pH = 7), Mg-Cabicarbonate fresh water and seawater, showing the evolution of Kd,Cs and K+ released concentration in solution as a function of Cs+ concentration in solution at equilibrium.

C. Michel et al. / Journal of Environmental Chemical Engineering 5 (2017) 810–817

environment and it is likely that all effects acting at trace concentrations were not considered by the model. 4. Conclusions A new selective adsorbent material for Cs was designed, consisting of potassium/copper ferrocyanide nanoparticles deposited in the mesoporosity of silica grains. High selectivity was obtained in solutions representative of natural fresh and sea waters with Kd,Cs of 106 and 2.0
105 mL/g, respectively. The competitive effect of the major cations present in these natural waters (K+, Na+, Mg2+ and Ca2+) plays a role but not in a manner that would prevent such a material to be used for treating contaminated natural waters as found at Fukushima for instance. A large range of chemical conditions was tested in order to build a large dataset and model parameterization using the Vanselow’s formalism. Integrating this dataset into the CHESS speciation code allowed the correct modeling of Kd,Cs values in these different water types over a wide range of Cs concentrations. The reasonably correct modeling of the full set of isotherm experiments confirms that the present selectivity coefficients are relevant for a large range of natural water compositions. The dataset of selectivity coefficients can be extrapolated to the study of other K-ferrocyanides and effluents, as well as to scale up these batch results for industrial purposes.

[8] [9]

[10]

[11]

[12] [13] [14]

[15]

[16]

[17]

[18]

5. Associated content

[19]

Supporting information presents one table on pH blank experiments (Table S1), one figure on XRD patterns (Fig. S1) and two figures on batch isotherms (Figs. S2-3).

[20]

Acknowledgement

[21] [22]

The authors acknowledge the members of the LMAC laboratory in CEA Marcoule for ICP-MS analysis and their contribution to the analytical methods. The authors are grateful to the French “Programme d’Investissements d’Avenir” (ANR-11-RSNR-0005) and AREVA for the financial support of this work. This work was also partly supported by the French RENATECH network.

[23]

[24] [25] [26]

Appendix A. Supplementary data

[27]

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jece.2016.12.033.

[28]

References

[29]

[1] J. Flouret, Y. Barre, H. Muhr, E. Plasari, Design of an intensified coprecipitation reactor for the treatment of liquid radioactive wastes, Chem. Eng. Sci. 77 (2012) 176–183. [2] M.W. Munthali, E. Johan, H. Aono, N. Matsue, Cs+ and Sr2+ adsorption selectivity of zeolites in relation to radioactive decontamination, J. Asian Ceram. Soc. 3 (2015) 245–250. [3] P.K. Sinha, P.K. Paniker, Treatment of radioactive liquid waste containing caesium by indigenously availalble synthetic zeolite: a comparative study, Waste Manage. 15 (1995) 149–157. [4] A.J. Aaron, A. Clearfield, The origin of ion exchange selectivity in a porous framework titanium silicate, J. Mater. Chem. 17 (2007) 4839–4842. [5] S.P. Mishra, D. Tiwari, S.K. Prasad, Inorganic ion exchangers in radioactive waste management. Part XVI: Uptake of some metal phosphates (stannic and zirconium) for 134Cs, J. Radioanal. Nucl. Chem. 268 (2006) 191–199. [6] S. Ayrault, B. Jimenez, E. Garnier, M. Fedoroff, D.J. Jones, C. Loos-Neskovic, Sorption mechanisms of cesium on (Cu2FeII)-Fe-II(CN)(6) and Cu-3(II) Fe-III (CN)(6) (2): Hexacyanoferrates and their relation to the crystalline structure, J. Solid State Chem. 141 (1998) 475–485. [7] I.M. Ismail, M.R. El-Sourougy, N.A. Moneim, H.F. Aly, Preparation, characterization, and utilization of potassium nickel hexacyanoferrate for

[30]

[31]

[32]

[33]

[34]

817

the separation of cesium and cobalt from contaminated waste water, J. Radioanal. Nucl. Chem. 237 (1998) 97–102. H. Mimura, J. Lehto, R. Harjula, Ion exchange of cesium on potassium nickel hexacyanoferrate (II)s, J. Nucl. Sci. Technol. 34 (1997) 484–489. I.M. Ismail, M.R. EI-Sourougy, N. Abdel Moneim, H.F. Aly, Equilibrium and kinetic studies of the sorption of cesium by potassium nickel hexacyanoferrate complex, J. Radioanal. Nucl. Chem. 240 (1999) 59–67. M.T.G. Valentini, R. Stella, M. Cola, Characterization of copper hexacyanoferrate(II) and (III) with reference to their use as cesium adsorbers, J. Radioanal. Nucl. Chem. 102 (1986) 99–110. M.T.G. Valentini, R. Stella, L. Maggi, G. Ciceri, Copper hexacyanoferrate(II) and hexacyanoferrate(III) as trace cesium adsorbers from natural-waters, J. Radioanal. Nucl. Chem. 114 (1987) 105–112. C. Loosneskovic, M. Fedoroff, Fixation mechanisms of cesium on nickel and zinc ferrocyanides, Solvent Extr. Ion Exch. 7 (1989) 131–158. P.A. Haas, A review of information on ferrocyanide solids for removal of cesium from solutions, Sep. Sci. Technol. 28 (1993) 2479–2506. Y.H. Qing, J. Li, B. Kang, S.Q. Chang, Y.D. Dai, Q. Long, C. Yuan, Selective sorption mechanism of Cs+ on potassium nickel hexacyanoferrate(II) compounds, J. Radioanal. Nucl. Chem. 304 (2015) 527–533. C. Delchet, A. Tokarev, X. Dumail, G. Toquer, Y. Barré, Y. Guari, C. Guerin, J. Larionova, A. Grandjean, Extraction of radioactive cesium using innovative fonctionalized porous materials, RSC Adv. 2 (2012) 5707–5716. K. Kawatake, N. Shigemoto, Preparation of potassium iron(III) hexacyanoferrate(II) supported on activated carbon and Cs uptake performance of the sorbent, J. Nucl. Sci. Technol. 49 (2012) 1048–1056. R. Turgis, G. Arrachart, C. Delchet, C. Rey, Y. Barre, S. Pellet-Rostaing, Y. Guari, J. Larionova, A. Grandjean, An original click and bind approach for immobilizing copper hexacyanoferrate nanoparticles on mesoporous silica, Chem. Mater. 25 (2013) 4447–4453. C. Vincent, A. Hertz, T. Vincent, Y. Barre, E. Guibal, Immobilization of inorganic ion-exchanger into biopolymer foams  Application to cesium sorption, Chem. Eng. J. 236 (2014) 202–211. C.Y. Chang, L.K. Chau, W.P. Hu, C.Y. Wang, J.H. Liao, Nickel hexacyanoferrate multilayers on functionalized mesoporous silica supports for selective sorption and sensing of cesium, Microporous Mesoporous Mater. 109 (2008) 505–512. Z.H. Du, M.C. Jia, X.W. Wang, Cesium removal from solution using PAN-based potassium nickel hexacyanoferrate (II) composite spheres, J. Radioanal. Nucl. Chem. 298 (2013) 167–177. L. Sharygin, A. Muromskii, M. Kalyagina, S. Borovkov, A granular inorganic cation-exchanger selective to cesium, J. Nucl. Sci. Technol. 44 (2007) 767–773. Y. Lin, G.E. Fryxell, H. Wu, M. Engelhard, Selective sorption of cesium using selfAssembled monolayers on mesoporous supports, Environ. Sci. Technol. 35 (2001) 3962–3966. A. Grandjean, Y. Barré, Procédé de préparation d’un matériau solide nanocomposite à base d’hexa- et octacyanométallates de métaux alcalins, 2016, WO 2016/038206. Brucker-AXS, Software TOPAS Version 4.2. C. Caglioti, A. Paoletti, E.P. Ricci, Choice of collimators for acrystal spectrometer for a neutron diffraction, Nucl. Instrum. Methods 3 (1958) 223. J. van der Lee, L. De Windt, V. Lagneau, P. Goblet, Module-oriented modeling of reactive transport with HYTEC, Comp. Geosci. 29 (2003) 265–275. J.I. Langford, A.J.C. Wilson, Scherrer after sixty years: a survey and some new results in the determination of crystallite size, J. Appl. Crystallogr. 11 (1978) 102–113. J. Letho, S. Haukka, R. Harjula, Mechanism of caesium ion exchange on potassium cobalt Hexacyanoferrates(II), J. Chem. Soc. Dalton Trans. 3 (1990) 1007–1011. C. Michel, Y. Barré, C. de Dieuleveult, A. Grandjean, L. De Windt, Cs ion exchange by a potassium nickel hexacyanoferrate loaded on a granular support, Chem. Eng. Sci. 137 (2015) 904–913. C. Loos-Neskovic, C. Vidal-Madjar, B. Jimenez, A. Pantazaki, V. Federici, A. Tamburini, M. Fedoroff, E. Persidou, A copper hexacyanoferrate/polymer/silica composite as selective sorbent for the decontamination of radioactive caesium, Radiochim. Acta 85 (1999) 143–148. H. Mimura, N. Kageyama, K. Akiba, M. Yoneya, Y. Miyamoto, Ion-exchange properties of potassium nickel hexacyanoferrate(II) compounds, Solvent Extr. Ion Exch. 16 (1998) 1013–1031. C. Loos-Neskovic, S. Ayrault, V. Badillo, B. Jimenez, E. Garnier, M. Fedoroff, D.J. Jones, B. Merinov, Structure of copper-potassium hexacyanoferrate (II) and sorption mechanisms of cesium, J. Solid State Chem. 177 (2004) 1817–1828. E.F.T. Lee, M. Streat, Sorption of caesium by complex hexacyanoferrates III. a study of the sorption properties of potassium copper ferrocyanide, J. Chem. Technol. Biotechnol. 33 (1983) 80–86. L. Charlet, C. Tournassat, Fe(II)-Na(I)-Ca(II) cation exchange on montmorillonite in chloride medium: evidence for preferential clay adsorption of chloride  metal ion pairs in seawater, Aquat. Geochem. 11 (2005) 115–137.