Anion exchange on hydrous zirconium oxide materials: application for selective iodate removal

The radioactive 129I is a top-priority radionuclide due to its the long half-life (1.57 × 107 years) and high mobility. Because of the planned and accidental releases to the environment, specific separation technologies are required to limit the potential radiation dose to human beings. Zirconium oxides are known for their adsorption capability and selectivity to oxyanions and here the applicability to selective IO3− removal has been investigated regarding the uptake mechanism, regeneration and competition caused by other anions, like environmentally relevant SO42−. Granular aggregates of hydrous zirconium oxides with and without Sb doping showed high potential for the selective IO3− removal in the presence of competing anions, like the forementioned SO42− (apparent capacity between 0.1–0.4 meq g−1 depending on SO42− concentration). The main uptake mechanism was found to be outer-sphere complexation (ion-exchange) to the protonated hydroxyl groups of hydrous zirconium oxides, but also minor mechanisms were identified including inner-sphere complexation and reduction to I−. The materials were observed to be easily and successively regenerated using dilute acid. Hydrous zirconium oxides showed high potential for IO3− removal from waste solutions regarding technical (high selectivity and apparent capacity) and ecological/economic (feasible regeneration) aspects.


Introduction
Iodine is a vital element for human beings and other mammals for the proper functioning of the thyroid gland. Although the most abundant isotope of iodine is stable 127 I, also radioactive isotopes are formed during uranium and plutonium ssion. Like stable iodine, these radioactive isotopes also concentrate in the thyroid gland if inhaled or digested causing an elevated risk of thyroid cancer. The most relevant risks of the radioactive iodine isotopes can be divided into two categories: the longterm risk of 129 I with an extremely long half-life (15.7 million years) and acute risk mainly from 131 I with a short half-life (8 days). The former is important from an environmental perspective because of its long half-life and high mobility. Therefore, its risk must be assessed in the case of groundwater contamination and nuclear waste disposal. For example, 129 I is regarded as one of the top priority radionuclides in the biosphere safety assessment of the nal disposal of spent nuclear fuel in Finland. 1 131 I is only important in the case of fresh fallout during, for instance, nuclear accidents like at the Chernobyl or Fukushima Daiichi nuclear power plants. The effects of acute iodine uptake and internal dose can be minimized by saturating the thyroid in advance with non-radioactive iodine for which iodine pills are intended.
Iodine has complex chemistry in the environment with its three main oxidation states −1, 0, +5. Iodate (IO 3 − ) is the main species with its +5 oxidation state found in oxidizing environments, whereas iodide (I − ) prevails with its −1 oxidation state in reducing conditions. For example, in anoxic waters, such as the Baltic sea, iodine exists mainly as I − , 2 but in oxidizing waters such as oceans, 3 IO 3 − is the dominant species. Molecular iodine (I 2 ), i.e. 0 oxidation state, is the major species only at low pH. 4 In addition, iodine readily reacts with organic molecules forming a wide range of different organo-iodine compounds. 5,6 Iodine removal is strongly dependent on its speciation and the immobilization of different iodine species has been comprehensively reviewed in literature. 4 Silver-based materials have proved to be feasible for I − decontamination due to the formation of AgI with extremely low solubility (K sp = 8.5 × 10 −17 ). [7][8][9][10][11] The solubility of AgIO 3 is however much higher (K sp = 3.2 × 10 −8 ) which makes silver-based materials inefficient in the removal of IO 3 − . In fact, the different affinity of I − /IO 3 − to Ag + can be utilized in analytical separations of these iodine species. 12 Other removal techniques for I − include for example ion exchange resins, 13 activated carbons, 14 organoclays 14-16 and hydrotalcites. 17 For the removal of IO 3 − , a wide range of different adsorbent materials have been studied including zero valent iron, 18 hydroxyapatites 19,20 and hydrotalcites. 17,21,22 Despite the intensive research, no selective, highly performing state-of-art iodate removal technique has yet been established. ZrO 2 is an amphoteric metal oxide known for its cation and anion exchange capabilities depending on the solution pH. [23][24][25] ZrO 2 is a widely used material that can be doped with different cations like Y, 26,27 Ce 28 or Sb [29][30][31] in order to enhance its mechanical, electronic or chemical properties such as the zeta potential of the material. ZrO 2 have several crystal structures from which the monoclinic, tetragonal, and cubic phases can be formed in ambient pressures. The monoclinic structure is the most stable in low temperatures, but cubic and tetragonal structures can be stabilized either by the doping or by limiting the crystallite size, i.e. forming nanocrystalline materials. The properties of the crystal structures differ signicantly regarding their physical and chemical properties like toughness or the number of anion and oxygen vacancies. 32 Sb is known to stabilize the tetragonal form of ZrO 2 , 29 which has been shown to be the active phase for the adsorption of different anions. [33][34][35] Our previous work 31 demonstrated the effectiveness of hydrous ZrO 2 materials for selective iodate removal. However, the ndings could not holistically reveal the mechanism behind the adsorption of IO 3 − or other anions. Within this study, we have extensively studied the basic ion exchange properties of pure ZrO 2 and antimony doped Zr(Sb)O 2 with different anions and the basic adsorption experiments have been complemented with supportive XAS (X-ray Absorption Spectroscopy) measurements. The focus has been on the mechanism of IO 3 − adsorption but also the competition of SO 4 2− , due to its relevance to environmental decontamination and strong affinity to adsorbent materials, has been investigated extensively. Different conditions like concentration and type of competing anions and the reversibility of the adsorption have been investigated to understand the mechanisms of IO 3 − uptake on ZrO 2 materials.
The reversibility has not only signicance in studying the mechanism of uptake but also regarding the regeneration of the materials, which is an important ecological and economic practical aspect regarding actual application of adsorbents in decontamination processes.

Chemicals
The reagents used within the study were of analytical grade (Alfa Aesar, Sigma-Aldirch, Riedel de Häen) and were used as received. oxidation, the radioactivity of the original and the separated solution were equal as IO 3 − adsorption to the material is insig-nicant whereas I − adsorption is extremely efficient. If the oxidation was not complete, more NaOCl was added, and the analysis step was repeated before use.

Synthesis of materials
The materials were synthesized and characterized as described earlier in literature. 31 In brief, two different zirconium oxides were synthesized by the precipitation method. First, Sb doped Zr(Sb)O 2 was synthesized by dissolving 45 g of ZrCl 4 (Riedel de Häen) and 2 g of SbCl 3 (Sigma-Aldrich) to 2 L of 3 M HCl under vigorous stirring using a mechanical stirrer. 1.2 L of 6 M NH 3 was added slowly to the solution until pH reached 7.8 and white gel was precipitated. ZrO 2 was synthesized similarly but here 100 g of zirconium basic carbonate (Alfa Aesar) was dissolved in 1 L of 6 M HNO 3 . Precipitates were let to stand in their mother solution overnight and clear supernatants were discarded. The precipitates were washed with deionized water until the conductivity of the supernatant was less than 4.0 mS cm −1 . Aer the wash, the materials were dried in an oven at 70°C for three days. The dried materials were ground and sieved to particle size 74-149 mm. Processor and using a mixture of 3.2 mM Na 2 CO 3 and 1 mM NaHCO 3 as an eluent. The anions were identied based on their retention times and the concentrations were calculated from the chromatogram peak areas compared with measured references.

Batch ion exchange experiments
In a typical batch experiment, 20 ± 1 mg of the ground and sieved material was weighed into a 20 mL polyethylene scintillation vial to which 10 mL of appropriate solution was pipetted. In some experiments larger masses and volumes were used but the batch factor, i.e. the solution to solid ratio, was kept the same at approximately 500 mL g −1 . If needed, pH was adjusted with appropriate volumes of NaOH or HNO 3 . The samples were equilibrated in a rotary mixer for 24 h and the solids were separated from the liquid phase by a combination of centrifuging (2100 G, 10 min) and ltering with a 0.2 mm lter (PVDF LC. Arcodisc, Gellman Sciences). The equilibration time was chosen for practical reasons and to ensure that equilibrium was attained. Earlier studies 31 have shown that IO 3 − the adsorption on ZrO 2 and Zr(Sb)O 2 is in equilibrium before 1 hour and pH already aer 5 minutes of contact. Finally, 5 mL of ltered solution was measured with a gamma counter as described in previous section (2.3.1 Iodine analyses) and the equilibrium pH measurement was conducted from the remaining ltered solution with an Orion™ 9103BNWP combination pH electrode (Thermo Scientic™). The uncertainty of pH measurements was estimated to be 0.1 units. All the experiments were conducted in normal laboratory air. 2.4.1 Equilibrium pH experiment. The standard batch procedure described in the previous section was used to study equilibrium pH in deionized water, 5 mM and 50 mM solutions of NaNO 3 , NaCl, Na 2 SO 4 and KIO 3 with the initial pH of about 5.4-5.9 before the contact with Zr(Sb)O 2 material. The samples containing the material and appropriate solutions were equilibrated in a rotary mixer for 24 h and solid and solution were separated, and pH of the solutions were measured.

Column experiments
In a typical column experiment 0.2-0.5 g of the sieved (74-149 mm) material was weighed to a beaker and rinsed with deionized water. The water was decanted and discarded and remaining white solid was mixed with a small volume (∼5 mL) of fresh deionized water and transferred with a disposable pipette into a low-pressure borosilicate glass column with a diameter of 0.5 or 0.7 cm and a porous polymer bed support at the bottom (Econo-Column ® . Bio-Rad Laboratories, Inc.). The feed solution was pumped through the column from the inlet with a ow velocity of about 20 bed volumes (BV) per hour and sample fractions were collected from the outlet. In columns with a smaller mass, a higher ow velocity of 40 BV per hour was used. From the collected fractions, pH and probe (mainly IO 3 − but also NO 3 − , Cl − and SO 4 2− in some experiments) concentrations were measured with corresponding methods described earlier in Section 2.3. The feed solution pH and concentration of the elements of interest were monitored throughout the experiments. 2.5.1 pH equilibration column experiment. Two columns were prepared for ZrO 2 and one for Zr(Sb)O 2 using 0.5 g of the material for each column. One of the ZrO 2 columns was fed with deionized water with pH 5.6 and the rest were fed with 1 mM KIO 3 solution with the same pH. The effluent was collected with a fraction collector using 120 min collection time before pH measurement. In the rst experiment the elution solutions were 100 mM NaNO 3 , Na 2 SO 4 and NaOH and in the second experiment only 100 mM Na 2 SO 4 and NaOH were used. Every elution step was continued until the desorption of IO 3 − was negligible and aer that the eluent was changed. The fractions were collected with a fraction collector and nally pH and the radioactivity of 125 IO 3 − of the samples were measured.

Regeneration experiments in column.
Two sets of regeneration experiments were performed within the study. In the rst experiment, 0.2 g columns of Zr(Sb)O 2 and ZrO 2 were rst loaded with a solution containing 10 mM Na 2 SO 4 and 10 mM KIO 3 traced with 125 IO 3 − followed by desorption with 0.1 M NaOH. Aer desorption, the columns were washed with deionized water until pH was 5.5, to remove the remaining NaOH solution and the columns were treated with solutions containing 1 M NaCl or 0.1 M of either NaOH or HCl until pH of the feed and the effluent were the same. Aer washing the columns with a few BV of deionized water, they were loaded again with the same 10 mM Na 2 SO 4 and 10 mM KIO 3 solution and uptake proles were measured. In the second experiment, the regeneration experiment was performed with the same 10 mM Na 2 SO 4 solution but with higher KIO 3 concentration (1 mM). Only 0.1 M HCl was used as the regeneration solution and in total the material was regenerated three times and its IO 3 − uptake performance was investigated.
2.6. Solid sample characterization 2.6.1 X-ray absorption spectroscopy. The oxidation states and the local coordination environment of I, Zr and Sb were determined by X-ray Near-Edge Structure (XANES) and Extended X-ray Absorption Fine structure (EXAFS) at their K-edges respectively situated at 33.169 keV, 17.998 keV and 30.491 keV. The measurements were done before (Zr and Sb only) and aer (Zr, Sb and I) IO 3 − adsorption in order to see the possible changes in their speciation. The XAS measurements were performed at PETRA III P64 beamline, the Deutsches Elektronen-Synchrotron (DESY), Germany. The spectra were collected in transmission (all the reference materials and Zr measurements) or uorescence (I and Sb measurements of the samples) mode depending on the concentration level of the probe element. The sample temperature was maintained below 10 K using an He cryostat to minimize the Debye-Waller factor (thermal effects) and to reduce any potential beam damage. Energy calibration was achieved using a Zr foil, a Sb foil and an I 2 cellulose sample for Zr, Sb and I respectively. The spectra of Sb 2 O 3 , Sb 2 O 5, I 2 , KI, KIO 3 were also collected as references. The collected data was normalized, analyzed and repeated scans were merged with Athena soware 36 and nally compared with the measured data of the reference materials. EXAFS data analyses were done with Artemis soware package 36 using rst shell tting for I (R-space 1.1-1.75Å based on k-space 2.8-12.5Å −1 ), Sb (R-space 1.25-2Å based on k-space 3-8Å −1 ) and Zr (R-space 1.1-2.0Å based on kspace 3-12Å −1 ). The small ranges in R-space were selected as the rst shell tting does not require a large range and as extending the range only resulted in the addition of noise. Also, the larger ranges increased R-factors of the ttings without signicantly changing the structural parameters within the reasonable values. The k-space ranges were selected by taking the maximum available range without signicant amount of noise and considering the fact that the signal of the rst shell damped aer about 10Å −1 . 2.6.2 Specic surface area measurements. Samples were analyzed with nitrogen physisorption at 77 K. Specic surface areas were calculated from the adsorption branch using the Brunauer-Emmett-Teller (BET) method and the total pore volume and average pore size were calculated from the desorption branch using the Barrett-Joyner-Halenda (BJH) method. The measurements were performed in an external laboratory as a service according to the standard ISO 9277:2010 using nitrogen adsorption at the temperature of liquid.
2.6.3 Thermogravimetric analysis. Thermogravimetric analysis (TGA) of the materials was performed with STA 449F3 Jupiter, Netzsch instrument connected to JAS-Agilent gas chromatography-mass spectrometer (7890B GC/MSD5977A). In the TGA experiments, about 25 mg of dried sample was weighed on Al 2 O 3 70 mL crucibles. The samples were heated to 1200°C under constant He ow with the heating rate of 20°C min −1 .  (Table 1). In deionized water, the contact with the material lowered the pH from 5.4 (deionized water in equilibrium with the atmospheric CO 2 ) to 3.3. Similarly, in NaCl and NaNO 3 solutions the pH dropped from ∼5.5 to 3.4 and 3.7 in 5 mM and 10 mM solutions, respectively. A completely different behaviour was observed with Na 2 SO 4 and KIO 3 , where the drop was negligible in 5 mM concentration and the pH rose to 5.9 and 6.4 in 50 mM solution, respectively.

Results and discussion
Fundamentally, the pH drop can be due either an increase in the concentration of H 3 O + or the decrease of the concentration of OH − in the solution. Because pH behaviour was different between the anions (and not the cations) the most probable reason for the pH drop is anion exchange between the OH − of the solution and the exchangeable anions in the materials originated from the synthesis. Evidently, the solutions containing SO 4 2− and IO 3 − ions prevented this drop of pH, probably because they are exchanged instead of OH − , and in higher concentrations the equilibrium pH even rose which indicates they are exchanged to OH − of the material thus indicating higher selectivity. The higher 50 mM concentration of Cl − and NO 3 − resulted only in a slightly higher equilibrium pH (3.7) compared to the 5 mM concentration (pH 3.4) and this is likely because with higher concentration more Cl − /NO 3 − is adsorbed instead of OH − but the selectivity is much lower compared with SO 4 2− and IO 3 − . The ion exchange reaction is described by eqn (1) for Cl − exchange (similarly with NO 3 − in the case of ZrO 2 ): In the exchange reaction, a neutral water molecule is exchanged to hydrochloric (or nitric) acid, which causes the decrease of pH.
The ion exchange between NO 3 − and Cl − is then described by eqn (2):   During the loading of materials with of 1 mM Na 2 SO 4 , the release of synthesis derived anions from the materials were studied (Fig. 3). In total 0.49 ± 0.03 meq g −1 SO 4 2− was exchanged to Cl − (0.73 ± 0.02 mmol g −1 ) and OH − in the case of Zr(Sb)O 2 . During the rst 300 BV's, pH steadily increased and stabilized to pH 5.6 aer an instant drop to pH 3 at the beginning of the experiment. The ratio of adsorbed (SO 4 2− ) and desorbed (Cl − and OH − ) equaled about 1.1. In total, 0.33 ± 0.01 meq g −1 of SO 4 2− was exchanged to NO 3 − (0.53 ± 0.02 mmol g −1 ) and OH − in the case of ZrO 2 which equals adsorption/desorption ratio of 1.25. Compared with Zr(Sb)O 2 , pH stabilized to about 6.5 that is slightly higher than the feed solution pH. In the second set, the same was done with SO 4 2− with varying concentrations in the range of 0.1-10 mM. In the third set, the desorption of IO 3 − from the loaded materials were studied with sequential elution with different solutions (100 mM NaNO 3 , Na 2 SO 4 and NaOH). Finally, the regeneration of the materials was studied with different solutions (NaCl, NaOH and HCl) and then tested for four successive regeneration cycles. , Cl − and NO 3 − ) at ten times higher concentration (10 mM) was studied in columns. The BT-curves were rather similar between the two materials (le side graph in Fig. 4)       and NaOH (1.82 ± 0.03 meq g −1 ) only partial regeneration was achieved. With ZrO 2 , similar results were observed (ESI †). In addition, only HCl was able to regenerate the pH lowering capability of the material (right side graphs in Fig. 9 and 10). Aer the treatment with 1 M NaCl, the pH rose from 6 to about 8 which indicates a release of OH − from the material. However, it remains unexplained why the pH rose less (from 6 to 7) aer 0.1 M NaOH treatment. The regeneration behavior is explained by ion exchange: aer the treatment with NaOH the material is in OH − form:

Application on selective IO
Without any further treatment, IO 3 − is not able to exchange with OH − in the material to the same extent as with the fresh material. The treatment with 0.1 M HCl returns the material to Cl − form: The same applies most probably to 1 M NaCl solution as well, but the conversion is not complete at neutral pH.  The practical regeneration of Zr(Sb)O 2 material was tested with four IO 3 − uptake/eluent cycles using 0.1 M NaOH as an eluent and 0.1 M HCl for the regeneration of the material between the cycles (Fig. 11). The regeneration efficiency remained high for all the cycles and the IO 3 − uptake was approximately 0.10 ± 0.01 meq g −1 (the uptake and elution curves in ESI †). The successive cycles showed some variation in the eluted IO 3 − fraction and the largest deviation was associated to the rst cycle where non-treated material was used. This resulted in the lower IO 3 − uptake and the lower elution percentage compared with HCl regenerated material at cycles 2 to 4, due to unexchangeable (inner-sphere complexation) IO 3 − uptake.

Solid sample characterization
3.3.1 Specic surface area measurements. The specic surface areas of the materials were analysed with nitrogen adsorption-desorption (Table 2). Zr(Sb)O 2 exhibited slightly larger specic surface area compared with ZrO 2 . That was expected due to the disorder caused by a guest atom, Sb, in the ZrO 2 structure. In general, Zr(Sb)O 2 had higher maximum IO 3 − uptake in the uptake experiments compared with ZrO 2 , but much less than the difference in the surface areas. Most probably the surface area is not signicant attribute in the anion exchange behaviour of the material on this scale.
3.3.2 I and Sb K-edge XANES. I K-edge XANES spectra of the IO 3 − loaded materials were measured to determine iodine oxidation state aer adsorption (Fig. 12). The iodine K-edge XANES spectra of all the samples show similar strongly characteristic shape of IO 3 − , with some slight differences around the white line due to small changes in average geometry around iodine, except for the Zr(Sb)O 2 in the most concentrated 10 mM SO 4 2− solution. In the latter case, the IO 3 − spectral features were still visible but the overall spectrum was attened indicating the partial reduction of iodine by comparison to KI and I 2 reference spectra. The reduction of IO 3 − to I − was observed only for the Zr(Sb)O 2 sample from the concentrated SO 4 2− solution, which should not be redox active as such. However, in column experiments it was observed that SO 4 2− considerably lowers the  This would also indicate that SO 4 2− is not preventing this redox dependant adsorption mechanism effectively, which is logical as the reduction of S(VI) by Sb(III) is not energetically favoured.
The Sb K-edge XANES spectra were measured to see if Sb oxidation state changes in the material during the uptake process (Fig. 13). Sb remained as Sb(III) aer the Zr(Sb)O 2 synthesis, but aer the contact with IO 3 − solution it partly oxidizes to Sb(V) as seen on the derivative spectra (Fig. 13) which is showing a bimodal white-line. However, the possible oxidation by dissolved oxygen needs to be considered. Since reduction of I was observed in I K-edge XANES, it seems highly probable that Sb is the reason for this. In total context, this must be a secondary adsorption mechanism as 1 gram of 5  (Fig. 14) showed a strong peak centred approximately at 1.4Å which is related to three oxygen atoms covalently bound to iodine in the molecule. 39 Outside the oxygens, nothing signicant was observed. This indicates that there is no close Zr neighbour for iodine. However, the possibility of heterogeneous distribution of local environments cannot be excluded. The EXAFS spectra of the materials with the adsorbed iodine were tted (ESI †) with a simple rst O shell path using Artemis, assuming I-O distance of 1.8Å and coordination number of 3. The optimized DE 0 was slightly elevated (15.24 ± 0.04 eV) for an unknown reason but also previously published values, even for reference materials, like KIO 3 , have been relatively high (10 eV). 39 The ts including the closest oxygens reproduced the experimental data sufficiently well regarding the simplicity of the tting approach (Rfactor 0.0102 . In Zr K-edge EXAFS spectra the signal at about 1.5Å is assigned to the 1st shell oxygen and the second peak at 3.0Å to 2nd shell Zr atoms (ESI †). These values correspond well to the values found in literature. 40,41 Similar tting as with I EXAFS was done for Zr. However, in all the samples no signicant differences were found between the pristine and loaded ZrO 2 /Zr(Sb) O 2 regarding the local coordination environment of Zr. Sb Kedge EXAFS spectra exhibited peaks at about 1.5Å but the rst shell ts of Zr and Sb (ESI †) did not reveal any signicant differences, except that s 2 was higher for pure Zr(Sb)O 2 compared with the material aer IO 3 − loading.
3.3.4 Thermo gravimetric analysis. Before performing TGA, Zr(Sb)O 2 was treated with 0.1 M NaOH or HCl in a column followed by washing with deionized water and drying in oven at 70°C. In general, the TG data was similar for both NaOH and HCl treated material (Fig. 15, see ESI † for the GS-MS chromatograms). The phase transition from amorphous to tetragonal took place at about 500°C which can be seen as an exothermic peak in DSC. The most signicant mass loss was observed between 100 and 200°C which was caused by the evaporation of adsorbed water conrmed by the collected MSspectrum where a signicant signal was observed at m/z 18. At the same temperature also notable signal from m/z 44 corresponding to CO 2 was observed. It seems that the material is adsorbing signicant amounts of CO 2 from either or both air and solution like have been reported previously. 42 The signal of    corresponds to HCl with different Cl isotopes ( 35 Cl: 75.77% 37 C: 24.23%). The HCl treated material released a high amount of HCl at 700°C with a sharp rise in the signal. NaOH treated material did not release HCl before the very end of the measurement where heating was already stopped. This suggests two different sites for Cl − in the material: the rst released at 700°C originates from the ion exchange in the column whereas the later released tracks down to the synthesis. The previous has a much higher signal and the latter is only released aer the structure starts to transform to a monoclinic structure. For the HCl treated material also SbCl 5 was detected aer the HCl release. This could be caused by the reaction of released HCl with the structural Sb.

Consideration on IO 3
− adsorption mechanism and competition of other anions The adsorption of IO 3 − is strongly associated with the pH in the solution and indications on the sorption process can be drawn from the pH changes during the uptake. Firstly, the surface charge of the adsorbent materials changes according to the solution pH, which strongly affects their affinity to anions in the solution. The anion exchange on zirconium oxides have been extensively studied earlier 24,25,37,43 and in principle, the anion exchange site on zirconium oxide can be represented as: where the protonated surface hydroxyls act as positively charged anion exchange sites. In addition, the possible ligand exchange is described as: where anion X − , e.g., sulphate or iodate, is exchanged with OH − in the material structure. 23 Secondly, the uptake of IO 3 − (or SO 4 2− ) itself affects the solution pH. Zirconium oxides synthesized within this study tend to lower pH even in deionized solutions, e.g., in the column experiments pH was lowered from 5.6 to 2.5 at the beginning of the experiments. The mechanism of this pH change is related to the synthesis derived anions (Cl − and NO 3 − depending on the synthesis conditions), which remain in the structure due to the incomplete exchange with OH − during the synthesis, i.e. the materials are initially partly in the OH − and partly in the Cl −form. 23 In addition, the form of the material can be changed by regeneration with dilute acid like HCl that changes the material back to the Cl − -form. The exchange of the adsorbed anions with OH − leads to the pH drop in the solution, like shown in eqn (1) ), the competing reaction takes place preventing the drop of pH: This explains why pH (i.e., the concentration of OH − ) and SO 4 2− concentration are critical parameters regarding the IO 3 − adsorption on zirconium oxides. 31 This study demonstrates that both adsorption mechanisms (eqn (5) and (6)) contribute to the adsorption of IO 3 − , although the outer-sphere complexation (eqn (7)) is evidently the main mechanism of uptake as the IO 3 − sorption was observed to be highly reversible and efficient and fast desorption was achieved with relatively dilute NaOH (100 mM). Also, the sorption capability was efficiently regenerated using dilute acid like HCl which changes the material back to the Cl − -form. In addition, certain fractions of adsorbed IO 3 − were eluted efficiently by

Conclusion
Hydrous zirconium oxide materials ZrO 2 and its antimonydoped Zr(Sb)O 2 counterpart exhibited excellent IO 3 − adsorption properties regarding apparent capacity (>0.6 meq g −1 ) and especially selectivity in high excess of competing anions, such as environmentally relevant SO 4 2− . The selectivity differences of zirconium oxides to different anions were observed, as NO 3 − , SO 4 2− and OH − seem to compete with IO 3 − for different available adsorption sites (competition decreasing in order OH − > SO 4 2− > NO 3 − ). The materials exhibited the highest IO 3 − removal when changed to Cl − form with dilute HCl (about 5 times higher apparent capacity compared with the OH − -form).
The materials also showed constant uptake performance during three load-regeneration cycles when regenerated with dilute acid (0.1 M HCl) which demonstrates the potential feasibility of the material for practical applications regarding sustainability and nancial perspectives. Based on the easy regeneration in dilute conditions and fast uptake, the main mechanism of uptake was concluded to be the ion-exchange between IO 3 − and anions e.g., NO 3 − , Cl − and OH − forming the outer-sphere complex with the materials. In the XAS data no external neighboring atoms were observed in the Zr or I K-edge EXAFS. This supports the conclusion regarding the outer-sphere complexation, although certain precautions should be taken as the reason for this could also be the relatively low concentrations of exchangeable anions (e.g., in 4% of total mass Cl − -form) or the amorphous structure of the materials resulting in the homogeneous distribution of the local coordination environments of the adsorption sites. Ligand exchange (inner-sphere complexation) between IO 3 − and surface OH − was observed to take place as a minor secondary adsorption mechanism in conditions without competing anions. IO 3 − is known to form both inner-and outer-sphere complexes with oxides and the dominating mechanism depends on the ionic strength and pH of the solution. 44 It seems that at least in the case of zirconium oxides, the type of competing anion can affect the proportions of the available sites as well due to electivity differences. In the presence of Sb doping, also a redox reaction between Sb and I was discovered and conrmed by the XANES data, but the mechanism only contributes slightly (theoretical capacity 0.04 mmol g −1 ) to the overall IO 3 − (maximum apparent capacity about 1 mmol g −1 ) uptake. It can, however, become signicant in concentrated matrices. Further studies would be required for the identication of the different ion exchange sites and what is the fundamental chemical or physical explanation for the selectivity of zirconium oxides to IO 3 − and SO 4 2− . This knowledge could be utilized for the manipulation of the material structure during the synthesis to furthermore improve the IO 3 − selectivity.

Conflicts of interest
There are no conicts of interest to declare.