Sorption of long-lived 94Nb on magnetite: spectroscopic and electrochemical investigation of the associated mechanism

The sorption study of long lived 94Nb isotope on magnetite was carried out in the pH range of 1–10, both in aerobic and anaerobic conditions. The present study is focused to understand the mechanism behind the sorption and to predict the role of magnetite in retarding the migration of the radionuclide. The sorption mechanism and the role of Fe(II) site of magnetite were investigated using solvent extraction, cyclic voltammetry, X-ray photoelectron and absorption spectroscopy. Insignificant difference in the sorption pattern and percentage sorption under aerobic and anaerobic conditions suggests similar sorption mechanism in both conditions. The oxidation states of Nb and Fe of magnetite remained unchanged after sorption process. In acidic medium, the sorption mainly occurs via ion exchange whereas in neutral/basic medium via covalent bond formation of Nb with magnetite.


Introduction
The sorption and migration of radionuclides in the geosphere became a matter of concern, particularly for those radionuclides having (1) high fission or activation yield in nuclear reactor and (2) long half-life [1]. 94 Nb is one of those isotopes with significant activation yield (0.67%, considering residence time in reactor: 10 y, neutron flux: 10 14 cm −2 s −1 ) and long half-life (20,300 y) but not studied to the same extent as other fission and activation products 137 Cs, 90 Sr, 239 Pu, 237 Np [2][3][4][5]. The generation of 94 Nb occurs from a stable 93 Nb isotope of niobium (Nb) upon neutron irradiation of Zr-Nb pressure tube material, used in pressurized heavy water reactors (PHWR). The presence of 94 Nb, along with other fission and activation products in discharged pressure tubes, results in a large radiation field. These discharged pressure tubes are generally stored in deep tile holes with concrete cap without further processing. The short and medium-lived radioisotopes (half-life: ≤ 1.5 y) mostly decay within 10 years of cooling but 94 Nb remains as the major contributor of radioactivity for a prolonged period owing to its long half-life [6].
Niobium predominantly exists in + V oxidation state in aerobic condition [7]. The high tendency to hydrolyse followed by forming intrinsic colloid in aqueous medium and the strong affinity towards suspended colloid in aquatic systems make Nb a prominent candidate for colloid assisted migration. The sorption studies of Nb on various materials like silica/glass surface, iron oxides, manganese dioxides and cement have been reported but detailed investigation regarding the sorption mechanism is not available [8][9][10][11][12][13]. The knowledge regarding the sorption mechanism helps in accessing the migration probability in natural aquatic conditions. There are several mechanisms like electrostatic forces, ion exchange, surface complexation, and surface precipitation, via which radionuclides get attached to colloid surface. The extent of sorption may depend upon the concentrations of radionuclide; the colloids suspended in aqueous medium; pH, ionic strength of the solution; and natural complexing species present in the solution. Some classical and spectroscopic methods were utilized to elucidate the sorption mechanism and speciation. Ionic strength [14], thermodynamic study [15], solvent extraction based classical methods can be used for this purpose [16]. Concerning spectroscopic techniques, Infrared/Raman spectroscopy, Time Resolved Laser Fluorescence spectroscopy (TRLFS), X-ray Photoelectron spectroscopy (XPS), X-ray Reflectivity and X-ray Absorption Fine structure (XAFS) were used extensively [17]. Generally, the spectroscopic techniques are supposed to provide detailed information regarding the sorption mechanism at the molecular level.
Iron oxides are expected to play a crucial role in controlling the migration of radionuclides because of widespread abundance in the earth crust (about 5%) and in ground water (1-40 mg L −1 depending upon the region) and high sorption capacity [18,19]. The aim of the present study is to investigate the sorption behaviour of 94 Nb on synthetic magnetite with the variation of pH of the aqueous medium. So, the sorption of 94 Nb on magnetite, was studied both in aerobic and atmospheric controlled (anaerobic) conditions. We have also tried to explore the role of Fe(II) site during the sorption of Nb(V) on magnetite, especially any change in Fe(II)/ Fe(III) ratio in magnetite before/after the sorption and at the same time any change in oxidation state of the Nb(V). Solvent extraction and X-ray photoelectron spectroscopy (XPS) methodologies were utilized to probe the sorption mechanism in anaerobic condition, whereas X-ray absorption spectroscopy (XAS) and cyclic voltammetry (CV) were utilized to understand the sorption mechanism in aerobic condition. The sorption and solvent extraction studies were carried out using 94 Nb radiotracer whereas the XPS, XAS and CV experiments were carried out using inactive Nb solution.

Materials and methods
Ferric nitrate (AR, Loba Chemie), ammonium ferrous sulphate (AR, Sarabhai chemicals), ammonium hydroxide (S D Fine Chem), sodium hydroxide (Thomas baker), acetone (Extra pure Ar, Cisco Research Laboratory), sodium perchlorate (Sigma Aldrich), nitric acid (AR, Pallav), 2-theonyl trifluoro acetone (Sigma Aldrich), xylene (Chemco fine Chemicals) were used during the present study. Millipore water (Resistivity 18.2 MΩ cm) was used throughout the experiment and nitrogen atmosphere (oxygen content ≤ 0.5 ppm) was used to maintain the inert atmosphere. The sorption study of 94 Nb on magnetite was carried out in batch method. The 94 Nb activity was measured using high purity germanium (HPGe) detector (Canberra) coupled with 8 k channel analyzer (Relative efficiency: 30%; Resolution: 1.9 keV at 1332 keV of 60 Co). Cyclic voltammetry (CD) was performed using an Autolab Potentiostat Galvanostat 302N (Metrohm, Switzerland). Data acquisition and analysis were carried out using GPES software.
X-ray photoelectron spectroscopy (XPS) was carried out to determine the chemical state of Nb and Fe before and after sorption at different pH conditions. Mg-K α (1253.6 eV) excitation source and DESA-150 electron analyzer operated at 40 eV pass energy (Staib Instruments, Germany) were used during the XPS measurement. The pressure in the chamber was 9 × 10 −7 Pa. The applied potential, filament current and emission current of the X-ray source were 15.5 kV, 2.25 A, and 15.5 mA respectively. The energy calibration was carried out using a standard Au sample with Au-4f 7/2 photo peak (binding energy of 84.0 eV) before data acquisition. XPS spectra were analyzed using XPS PEAK 4.1 software. The background was subtracted by means of the Shirley method and the deconvolution of the XPS spectra was carried out using the Gaussian and Lorentzian line shapes for each component in the spectra.
Energy-Scanning XAS beam line (BL-9) of the Indus-2 Synchrotron source (2.5 GeV, 100 mA) in Raja Ramanna Centre for Advanced Technology (RRCAT), Indore, India was utilized for the XAS measurements [20]. The operating energy range was 4 keV to 25 keV. The beam line optics consist of a Rh/Pt coated collimating meridional cylindrical mirror and the collimated beam reflected by the mirror is monochromatized by a Si (111) (2d = 6.2709 Å) based double crystal monochromator (DCM). The second crystal of DCM is a sagittal cylinder used for horizontal focusing while a Rh/Pt coated bendable post mirror, facing down, is used for vertical focusing of the beam at the sample position. Rejection of the higher harmonics in the X-ray beam was performed by de-tuning the second crystal of DCM.

Preparation of magnetite and characterization
About 50 g each of the ferric nitrate [Fe(NO 3 ) 3 ] and Mohr salt [(NH 4 ) 2 SO 4 FeSO 4 ·6H 2 O] were dissolved in 100 mL deionized water separately in two beakers and then mixed. After that, dilute ammonia solution was added gradually to the mixed solution and stirred continuously at 80 °C for 2 h. The brownish black colored precipitate was then filtered and washed using dilute ammonia solution, followed by deionized water till pH of the washings was within 6.5 to 7.0. Thereafter, the precipitate was further washed with high purity acetone and stored in nitrogen atmosphere (oxygen content ≤ 0.5 ppm).
X-ray diffraction (XRD) pattern of the synthesized iron oxide was recorded using a X-ray diffractometer (Rigaku Smart Lab) equipped with Ni filtered Cu K α X-rays. An indigenous surface area analyzer (BARC make) was used to measure the specific surface area. The zeta potential of magnetite was measured using the Litesizer TM500 instrument (Anton-Paar). Dynamic light scattering (DLS) measurements were used for particle size determination. 94 Nb radiotracer was separated from irradiated Zr-Nb pressure tube in HCl-HF mixed acid medium using anion exchange methodology [21]. Both Zr and Nb form anionic complexes in the medium and when they pass through an anion exchange column, the anionic complexes were sorbed. Then, using gradient elution, pure Zr and Nb fractions were separated from each other. The detailed separation procedure was discussed in our previous work [8]. Niobium concentration in tracer solution was 3.12 × 10 -5 M and the concentration corresponding to 94 Nb radioisotope was 2.2 × 10 -7 M.

Sorption study of 94 Nb on magnetite
About 100 μL of 94 Nb (2.2 × 10 -7 M) radiotracer was introduced into 25 mL solution containing 0.05 M NaClO 4 in air tight polypropylene centrifuge tubes (Tarson, 60 mL). The pH of the solutions was adjusted between 1 and 10 using dilute HNO 3 or NaOH. Then about 50 mg of magnetite added to each of the solution. The experiments in aerobic condition were carried out at ambient atmospheric conditions. The experiments in anaerobic condition were performed in controlled atmospheric conditions (high purity N 2 ) and then the centrifuge tubes were capped tightly and, for further precaution, tubes were wrapped using para film. Solutions were stored for 48 h with occasional shaking and afterwards, centrifugation (temperature controlled) was carried out for 30 min at 10,000 rpm. As centrifugation facility in inert atmosphere was not available, the phase separation by centrifugation was carried out after hermitically sealing the centrifuge tubes using polyethylene. Then the supernatant was collected for 94 Nb activity measurement (time: 5000 s) using a HPGe. The percentage sorption was calculated by comparing the activities of initial solutions and the final supernatants at 702.7 keV for 94 Nb using the following equation where A i and A f are the 94 Nb activity (counts) in initial solutions and supernatant respectively. Similar experiments were performed under aerobic conditions to check the difference in sorption of Nb on magnetite in the above said experimental conditions.

Solvent extraction of 94 Nb sorbed magnetite
Solvent extraction methodology is one of the simplest methods to check any change of oxidation state during sorption, as it does not require any complicated instrumental facility or analysis procedure. First, 100 µL of 94 Nb radiotracer was equilibrated with magnetite (controlled atmospheric condition maintained by high pure N 2 gas) at different pH conditions. After centrifugation, the 94 Nb activity in the supernatant was measured by HPGe detector. From the difference between initial activity and the activity in supernatant, the 94 Nb activity on magnetite was determined. Then a mixture of 25 mL of 0.5 M theonyltrifluoro acetone (TTA) in xylene and 25 mL of 1.0 M HNO 3 solution was added together to Nb sorbed magnetite. The mixture was shaken for 5 min and thereafter, the aqueous and organic phases were separated. The 94 Nb activity in the aqueous and organic phase were measured and these corresponds to + V and + IV oxidation states of Nb respectively.

Electrochemical investigation of Nb sorption on magnetite
Magnetite is an unique variant of iron oxides as it contains both Fe(II) and Fe(III) sites. To check any change in oxidation state of both adsorbate Nb(V) and adsorbent (magnetite) at different pH conditions, an electrochemical investigation was carried out. A three electrodes system was adopted for the experiment; where a modified glassy carbon electrode (GCE) was used as working electrode, Ag/AgCl/KCl (saturated) as reference electrode and a graphite rod as counter electrode. All the electrochemical experiments were carried out at room temperature in ambient atmospheric condition. The glassy carbon electrode was polished with 5 µm and 1 µm alumina powder and cleaned under ultra-sonication at 20 kHz for 15 min to get a clean surface. Carbon black and magnetite composite (1:1) was prepared by mixing both (2 g L −1 ) at 1:1 (weight) ratio in acetone medium containing 0.1% nafion. The mixture was sonicated for 5 min at 20 kHz to disperse the solid particles in acetone medium to attain homogeneity prior to casting on the glassy carbon electrode surface. A 5µL aliquot of the homogeneous mixture was drop cast on the glassy carbon electrode surface and allowed to dry.

Sample preparation for XPS and XAS
The batch sorption experiments were carried out using 94 Nb radiotracer in very low concentration level. But it is very difficult to get good spectroscopic signal in XPS/XAS measurement in such low concentration level. Therefore, to overcome the problem, non-active Nb solution of elevated concentration (mM) was used for the sample preparations.

XAS measurement and methodology
XAS measurements of the samples were performed in fluorescence mode at Nb K-edge energy (18.9856 keV).
In transmission measurements, three ionization chambers (300 mm length each) were used, one ionization chamber for measuring incident flux (I 0 ), second one for measuring transmitted flux (I t ) and the third one for measuring X-ray absorption spectrum of a reference metal foil for energy calibration. Appropriate gas mixtures and pressure were chosen to achieve 10-20% absorption in the first ionization chamber and 70-90% absorption in the second ionization chamber to improve the signal to noise ratio. The absorption coefficient (μ) at specific energy (E) for specific absorber thickness (x), was obtained using the relation, In fluorescence mode, the fluorescence intensity (I f ) emitted from the sample was measured using a silicon drift detector (SDD) and an ionization chamber to measure the incident flux (I 0 ). The relation μ = I f /I 0, was used to obtain the absorption coefficient.
In order to take care of the oscillations in the absorption spectra μ(E) were converted to absorption function χ(E) defined as given in following Eq. (3).
where, E 0 is absorption edge energy, μ 0 (E 0 ) is the bare atom background and Δ μ 0 (E 0 ) is the step in μ(E) value at the absorption edge. The energy dependent absorption coefficient χ(E) was converted to the wave number dependent absorption coefficient χ(k) using the relation given in Eq. (4). (2) where, m is the mass of electron. χ(k) is weighted by k 2 to amplify the oscillation at high k and the χ(k)k 2 functions are Fourier transformed in R space to generate the χ(R) versus R spectra in terms of the real distances from the center of the absorbing atom [22]. The XAS data analysis program available within the Demeter software package was used for XAS data analysis. This includes background reduction and Fourier transform to derive the χ(R) vs R plot from the absorption spectra (using ATHENA software), generation of the theoretical XAS spectra starting from an assumed crystallographic structure and finally fitting of experimental data with the theoretical spectra using ARTEMIS software [23].

Characterization of synthesized magnetite
The XRD pattern of the synthesized iron oxide (Fig.S1) indicates the crystalline nature which matches with magnetite (JCPDS:19-0629). The specific surface area of magnetite was found to be 35 m 2 g −1 . The point of zero charge (PZC) of magnetite was found to be at pH 6.3 (Fig. S2). DLS measurement revealed the hydrodynamic particle size was in the range 180-320 nm.

Sorption of Nb on magnetite in aerobic and anaerobic condition
The variation of the sorption of Nb on magnetite with pH both in aerobic and anaerobic conditions is shown in Fig. 1. The sorption was found to be increasing with the increase in pH of the medium and became almost quantitative in neutral / basic conditions. The sorption patterns as well as percentage sorption in aerobic and anaerobic conditions are almost identical. This suggests that the sorption mechanisms in aerobic and anaerobic conditions are similar. Present observation is different from that of the other pentavalent element Np(V) sorption on magnetite where the authors reported significant difference in sorption under aerobic and anaerobic conditions [24][25][26][27][28]. The point of zero charge of magnetite was found at pH 6.4. So, below pH 6.4, magnetite surface remains positively charged and above that, surface gradually becomes negatively charged. Niobium exists as NbO 2 + in acidic solution up to pH 3.0, while between pH 3.0 to 7.0 neutral species like HNbO 3 or NbO(OH) 3 become dominant and beyond that pH, the anionic species like NbO 2 (OH) 2 or NbO(OH) 4 − or Nb(OH) 6 − become main species [29][30][31]. Therefore, the low sorption up to pH 3.0 may be because of the electrostatic repulsion forces and whatever the sorption took place, was due to physi-sorption or ion exchange phenomenon between the surface hydroxyl group of magnetite and the positively charged Nb species. Again, the electrostatic repulsion was expected to cause the reduction in sorption above pH 7.0, but a noticeable decrease in sorption was not observed. This indicates that the sorption of Nb on magnetite above pH 7.0 may not follow the same mechanism as followed below pH 3.0. So, the sorption above pH 7.0 cannot be explained by physisorption or ion exchange phenomenon. Surface complexation between the hydroxyl group of neutral or anionic species Nb species and the surface hydroxyl group (-OH) of magnetite may be the reason behind high sorption. The similar sorption features in aerobic and anaerobic conditions suggests that, probably a similar sorption mechanism is followed in both the conditions. The solubility of magnetite with variation of pH of medium also checked as it can affect the magnitude of sorption. Negligible dissolution of magnetite was observed above pH 3.0 but increases to 2 wt.% at pH 1 (Fig. S3). So, dissolution of magnetite might have mild effect on Nb sorption only in acidic pH (≤ pH 3.0). We found ~ 5 (± 2) % loss of Nb on container walls in low pH region (≤ pH 3.0) and the loss decreases to 2 (± 1) % in neutral and basic pH region. The correction due to the sorption on container wall was incorporated while calculating the final sorption values.

Solvent extraction outcome
2-theonyltrifluoroacetone (TTA) is used for selective extraction of the + IV and + VI species from the aqueous phase, leaving + V species in the aqueous phase in the pH range of 3.0-4.0. About pH 1.0, although the extraction of IV species remains quantitative, but become negligible for VI species [32,33]. Again, if the acidity of the aqueous medium maintained above 2 M, then extraction of the V species become significant [34]. As we intended to separate Nb(IV) and Nb(V), the acidity of the aqueous medium was kept 1 M during TTA extraction. The methodology adopted in the present case, was also utilized by several authors to understand the change in oxidation states (reduction) of Pu(V), Np(V) during their sorption on various iron oxides [24,28,35]. The addition of 1 M HNO 3 helps in desorption of Nb sorbed on magnetite. If Nb(V) is reduced to Nb(IV) during the sorption, it will form a neutral complex with TTA and will be immediately extracted into organic phase (xylene) during solvent extraction. In the present study, insignificant amount of 94 Nb activity was detected in the organic ase after solvent extraction of Nb sorbed magnetite, which corresponds to + IV oxidation state of Nb. The Nb(IV) fraction in tracer solution (0.84%) is comparable to the Nb(IV) fraction found after equilibrating with magnetite (0.9-1.20%) in the atmospheric controlled system (Table S1). In other words, although significant sorption of Nb on magnetite was observed especially in neutral and basic pH regions, we did not observe any change in oxidation state of Nb(V) in all the studied pH conditions even in inert atmosphere.

XPS data interpretation
The XPS spectrum of untreated as synthesized Fe 3 O 4 is shown in Fig. 2a. Two distinct peaks of Fe 2p 3/2 and Fe 2p 1/2 were observed and the area of Fe 2p 3/2 peak is greater than that of Fe 2p 1/2 because of spin-orbit coupling (2p 3/2 : degeneracy = 4, 2p 1/2 : degeneracy = 2). Magnetite contains both Fe(II) and Fe(III) sites, having an inverse spinel structure where the Fe(II) is octahedrally coordinated and the Fe(III) is both octahedrally and tetrahedrally coordinated. Both peaks can be de-convoluted into three peaks which correspond to Fe(II) oct , Fe(III) oct and Fe(III) tet with binding energy (BE) 709.4 eV, 711.0 and 712.8 eV respectively in 2p 3/2 peak. These BE values are in close agreement with the literature reported values [36][37][38][39][40]. The 2p 3/2 to 2p 1/2 peak ratio was found to be 1.8, which is close to its predicted value of 2 (degeneracy of 2p 3/2 divided by the degeneracy of 2p 1/2 ). The ratio of Fe(II) to Fe(III) from the deconvolution of both the 2p 3/2 and 2p 1/2 peaks were found to be 0.56 and 0.59 respectively and the values are slightly higher than the theoretical value of 0.5 (calculated from the stoichiometry of Fe 3 O 4 ) but in accordance with the literature [36,37]. Figure 2b shows the XPS spectrum of Fe 3 O 4 , equilibrated in the aqueous medium in ambient atmosphere for 48 h.
The XPS spectra of Fe 3 O 4 equilibrated with Nb(V) in inert atmosphere pH 1.2 medium is shown in Fig. 3. The XPS spectrum of the similar systems at pH 6.6 and pH 10.4 conditions are shown in Fig. S4 & Fig. S5   The XPS spectra of Nb (3d levels) in Nb sorbed magnetite systems at different pH conditions are shown in Fig. 3, Fig.  S4 & Fig. S5. The binding energy of the doublet 3d 3/2 and 3d 5/2 were found to be at 209.5 and 206.9 eV respectively, which are very close to the literature reported peak positions of Nb(V) in Nb 2 O 5 [41,42]. The 3d 5/2 to 3d 3/2 peak ratio was  found to be 1.37, which is close to its predicted value of 1.5 (degeneracy of 3d 5/2 /degeneracy of 3d 3/2 ). This suggests that oxidation state of Nb remain same even after sorption on magnetite in anaerobic condition.

Electrochemical investigation
The GCE surface was initially coated with 1:1 mixture of carbon black and synthesized magnetite and cyclic voltammetry (CV) (staring potential: 0.1 V, switching potential: 1 V) was performed in different pH conditions (pH 1.2, 6.6 and 10.4). During anodic scanning, the oxidation peak was found at 0.65 V, while during cathodic scanning, the reduction peak was observed at 0.55 V. With repeated scanning at a regular time interval (5 min.) the peak height of both the peaks were found to decrease gradually (Fig. 4a).
The electrochemical reactions of the bulk Fe 3 O 4 were postulated in the following way During the oxidation process, Fe 3 O 4 is converted into Fe 2 O 3 . The cathodic reaction results in the reductive dissolution of magnetite. With repeated scan, the Fe(II) content on the electrode surface decreases and this may be the reason for the gradual decrease in the peak current for both the oxidation and reduction peaks [43][44][45].
Freshly prepared carbon black and Fe 3 O 4 coated GCE was dipped into 100 mg L −1 Nb(V) solution for 5 min, and then CV was performed as described above. Like in the case of magnetite, a gradual decrease in oxidation/reduction peak current due to Fe(II)/Fe(III) sites in magnetite was observed Fig. 4b). The rate of decrease in peak current is comparable with the rate of decrease in peak current for the magnetite system. This suggests that the Fe(II) sites in magnetite do not act as reducing agent during the sorption of Nb(V) on magnetite surfaces. The structural/surface Fe(II) reported to be stronger reducing agent (Potential: − 0.34 to − 0.65 V) compared to aqueous Fe(II) in homogeneous solution (− 0.77 V) [43].  [43][44][45][46]. The complete process has two individual part, i.e. oxidation of structural Fe(II) of magnetite and the reduction of the metal ion in solution. The potential of the whole process can be written as The potential of the whole process must be positive for the feasibility of the process and the highest oxidation potential of Fe(II) was observed during the reduction of Cu(II) to Cu. This makes structural Fe(II) a stronger reducing agent and the electrochemical reductions, which are not possible in homogeneous solution, become possible by Fe(II) containing minerals. No literature report found where Fe(II) of magnetite reduces metal ion of negative reduction potential. The standard reduction potential of Nb(V)/Nb(IV) system is in the range − 0.25 to − 0.38 V [47,48] and if the reduction process is supposed to occur, then Fe(II)/Fe(III) oxidation potential must be greater than 0.25 V. But in the present case, the Fe(II)/Fe(III) oxidation potential in the presence of 100 mg L −1 Nb(V) was found at − 0.65 V. Therefore, the potential of the whole oxidation-reduction process will be − 0.90 V, and this explains the non-feasibility of the reduction of Nb(V) during sorption on magnetite surface.

XAS data interpretation
The normalized X-ray absorption near edge structure (XANES) spectra at Nb K-edge for metallic Nb foil, Nb(V) standard solution and Nb(V) sorbed Fe 3 O 4 system for all three pH conditions is shown in Fig. 5. It is evident from  Table 2. The sample for Nb(V) sorbed Fe 3 O 4 system at pH 1.2 shows that 1.80 numbers of oxygen atoms at a distance of 1.72 Å, which may be because of the two axial Nb=O bonds of NbO 2 + group. The second and third peaks correspond to the longer and less rigid Nb-O bond with the oxygen atoms of hydroxyl group or water molecules. These O atoms may be present in equatorial positions surround the Nb atom and satisfying its coordination number. Multiple k-weighted test is a method in EXAFS to identify the atom present surrounding the central atom [49]. The presence of O atoms and the absence of Fe atomsin the second coordination peak was identified by mmultiple k-weighted test. The dissimilar amplitudes of different for Fe k-weighted spectra for the first and second peaks confirmed the absence of Fe atoms in both first and second coordination shells (Fig S6). Increase in pH of the medium make the NbO 2 + species less stable, which is also evident from the decreasing number of oxygen atoms in the first co-ordination shell of Nb.
In case of Nb(V) sorbed Fe 3 O 4 system at pH 6.6, there are 1.38 and 4.22 oxygen atoms at distance of 1.74 Å and 1.92 Å, respectively from the Nb atom and it is confirmed by the multiple k-weighted test (Fig S7, S8). The contribution of Fe atoms in second coordination sphere (at 2.93 Å and 3.24 Å) is confirmed by k-weighted test. This confirmed the formation of chemical bond between the magnetite surface and the Nb during the sorption at pH 6.6. The Nb(V) sorbed Fe 3 O 4 system at pH 10.4 shows decreasing number of O atoms (1.08) at 1.73 Å, which signifies that the rigidity of the NbO 2 + group gradually reduces with increase in pH. Along with that, the Fe atom density around Nb atom is higher compared to the system at pH 6.6. At the same time the Nb-Fe distance also reduced to 2.72 Å compare to the system at pH 6.6.
The longer Nb-Fe bond of 3.24 Å and shorter Nb-Fe bonds 2.93/2.72 Å were observed in XAS study of Nb(V)magnetite systems in neutral and basic medium. The inner sphere complexation of Nb with the magnetite surface (≡Fe-O-Nb) may lead to the formation of longer Nb-Fe bond. Similar types of observation was reported during the sorption of Np, U on hematite, goethite, iron (oxyhydr) oxides and ferrihydrite where the authors reported typical Np-Fe or U-Fe bond length of 3.4-3.5 Å [50][51][52][53].
The short Nb-Fe bond length suggests the probability of structural inclusion of Nb(V). This is in accordance with other literature reports where the authors found that Np(V) was incorporated into hematite structure and U(VI) incorporated into ferrihydrite during the sorption through direct  replacement of octahedrally coordinated Fe(III) [52,54]. In ambient aquatic condition, magnetite slowly transforms to maghemite (intermediate) and finally to hematite [55]. This is also evident from the current XPS study of the magnetite system where, Fe(II) to Fe(III) ratio was found to be gradually decrease under aqueous aerobic condition. So, the short Nb-Fe bond length suggests the possible replacement of the octahedrally occupied Fe(III) by Nb(V) during the crystallization of maghemite to hematite. Almost identical ionic radii (r) of Fe(III) (r: 0.645 Å, CN: 6) and Nb(V) (r: 0.640 Å, CN: 6) (http:// abula fia. mt. ic. ac. uk/ shann on/ ptable. php) may have amplified the replacement probability. The shortest Nb-Fe distance (2.72 Å) may be because of the replacement of Fe(III) at face centered octahedral site and relative larger Nb-Fe distance (2.93 Å) may be due to the replacement of Fe(III) at edge centered octahedral site. Further investigation is required to ascertain whether Nb directly incorporated in the magnetite structure or during its gradual transformation to hematite in aerobic condition. The number of Fe atoms coordinated to Nb at longer distance is higher compared to the numbers at shorter distance. This suggests that, the major fraction Nb undergoes surface complexation rather than inclusion in the structure.
Therefore, XAS study reveals that, in acidic medium the sorption of on magnetite takes place through outer sphere complexation, whereas in neutral and mild basic medium via inner sphere complexation.

Conclusion
The sorption study of Nb on magnetite was carried out in both aerobic and anaerobic conditions. The sorption was highly dependent on the pH of the medium, and it was observed that at low pH the sorption was quite low but increased gradually with increase in pH. Under both aerobic and anaerobic conditions, the sorption was almost identical. In anaerobic condition, solvent extraction study of Nb sorbed magnetite ruled out the probability of Nb(V) reduction during its sorption on magnetite. XPS study of the Nb sorbed magnetite under anaerobic condition unfolded the puzzle more clearly. The Fe(II) to Fe(III) ratio in synthesized magnetite remain unchanged even after equilibration of magnetite with Nb(V) solution at different pH conditions and at the same time, only Nb(V) was detected in all the samples. This indicated that, there was no redox reaction between Fe(II) site of magnetite and Nb(V) in solution during the sorption process. Cyclic voltammetry study using carbon black-magnetite coated GCE confirmed that Fe(II) site in magnetite does not play any special role during the sorption of Nb(V) on magnetite surfaces in aerobic condition also. XAS study at Nb k-edge, reinforces the previous observation and, along with that, the XAS study inferred that the sorption in acidic pH conditions mainly occurs via physi-sorption or ion exchange phenomenon as Fe atom was not detected in both the primary and secondary coordination sphere of Nb atom. In neutral and basic medium, Fe atom was identified in the second coordination shell of Nb. Two different types of Nb-Fe bond distances suggest probable inclusion of Nb in the structure and inner sphere surface complexation.

Declarations
Conflict of interest It is declared that, there is no known conflict of interest regarding the work reported in this manuscript.