Novel electroblowing synthesis of tin dioxide and composite tin dioxide/silicon dioxide submicron fibers for cobalt(ii) uptake

Nanoscale SnO2 has many important properties ranging from sorption of metal ions to gas sensing. Using a novel electroblowing method followed by calcination, we synthesized SnO2 and composite SnO2/SiO2 submicron fibers with a Sn : Si molar ratio of 3 : 1. Different calcination temperatures and heating rates produced fibers with varying structures and morphologies. In all the fibers SnO2 was detected by XRD indicating the SnO2/SiO2 fibers to be composite instead of complete mixtures. We studied the Co2+ separation ability of the fibers, since 60Co is a problematic contaminant in nuclear power plant wastewaters. Both SnO2 and SnO2/SiO2 fibers had an excellent Co2+ uptake with their highest uptake/Kd values being 99.82%/281 000 mL g−1 and 99.79%/234 000 mL g−1, respectively. Compared to the bare SnO2 fibers, the SiO2 component improved the elasticity and mechanical strength of the composite fibers which is advantageous in dynamic column operation.


Introduction
Inorganic materials are interesting as industrial adsorbents due to their good resistance to decomposition at high temperatures and under ionizing radiation and with a wide operating pH range. Furthermore, they are typically much more selective than organic resins for metal ions. The high selectivities stem from their regular, porous and rigid structures that exhibit ion sieve functionality. 1 There has been research on the sorption characteristics of metal oxides, such as SiO 2 , TiO 2 , ZrO 2 and SnO 2 since the mid-20th century. 2 Among them, tin dioxide nanoand microparticles have shown promising sorption properties for both organic 3 and inorganic 4-13 compounds. Nanoscale SnO 2 has attracted attention as a multifunctional material and it has also been used in transistors, 14 protective coatings 15 and gas sensors. 16,17 Similarly, both bare SnO 2 and composite SnO 2 /SiO 2 nanobers have exhibited promising gas sensing, [18][19][20][21][22][23][24][25] electrochemical, 26 optical 27 and molecular ltration properties. 28 The large specic surface area and porous structure of bers provide plenty of contact area for target gases 24,25 and adsorbing species 28 and improve the selectivity for adsorbing molecules with different sizes. 28 The incorporation of an amorphous SiO 2 component into the SnO 2 bers has been shown to improve both the gas sensing properties compared to SnO 2 nanoparticles 25 and the structural stability 24 and mechanical strength 28 of the bers.
For removal of heavy metal ions from aqueous solutions, there are plenty of techniques such as chemical precipitation, ion exchange, adsorption, membrane ltration, electrochemical methods and phytoremediation. 29,30 Adsorption has certain advantages over the other methods including high removal efficiency even at low concentrations, cost-effectiveness, simple design and less production of toxic sludge. 29,31 Adsorbents comprise both organic and inorganic materials which may be synthetic or of natural origin, such as agricultural by-products. For example, titanium(IV) iodovanadate particles have been utilized for Pb 2+ and Hg 2+ separation 32 and coco-peat biomass for Pb 2+ , Cd 2+ , Cu 2+ and Ni 2+ removal. 33 Inorganic submicron and nanobers could be excellent adsorbents for a variety of industrial applications. Inorganic bers have a large specic surface area due to their high surface-to-volume ratio and oen form porous structures which can lead to a good adsorption capacity. Zirconium dioxide submicron bers have performed well in ow-through column operation and shown better mechanical stability, less pressure build-up in the column and faster adsorption kinetics than the corresponding granules. 34 Submicron and nanobers can also be an eco-friendly alternative for the purication of nuclear power plant (NPP) wastewaters, since they can markedly reduce the volume of solid radioactive waste that requires a permanent repository.
The most common method to prepare polymeric or inorganic submicron and nanobers is electrospinning. In this technique, a high voltage is applied to a precursor solution containing a polymer, solvent(s) and for example a metal salt, depending on the desired ber composition. During the electrospinning process, repulsive electrostatic forces cause the polymer chains to stretch and form bers that are deposited on a grounded collector. Via post-electrospinning calcination, the polymer may be removed and the nal brous product, typically a metal oxide, is formed. Different precursor solutions and operational parameters allow the control of the properties and morphology of the bers. 35 Electrospinning is therefore a simple and cost-efficient method and upscalable for industry. [36][37][38][39] Solution blowing (airbrushing) method in contrast uses pressurized gas to draw bers from the precursor solution. 40 Solution blowing offers even 15 times faster production rate than electrospinning 41 but it tends to result in bundles of aligned bers whereas electrospinning yields an entanglement of individual, mainly un-aligned bers. 42,43 In electroblowing, the bers are formed by both electrostatic force and air ow and hence the technique combines electrospinning and solution blowing and their advantages. Moreover, electroblowing enables 2.5 times faster solution feeding rate and thus ber production rate than solution blowing. 41,44 Compared to electrospinning, the additional air ow in electroblowing permits the use of more viscous precursor solutions 45 and produces bers with smaller diameters and fewer beads. 46 It should be feasible to scale up electroblowing for industry, similarly to electrospinning.
To tackle climate change, global energy production is increasingly based on clean low-carbon power including nuclear power. However, the production of nuclear energy generates radioactive ssion and activation products that pose a risk to human health and environment without proper treatment of nuclear waste effluents. Among the activation products, 60 Co caused by steel corrosion is one of the most hazardous ones due to its rather long half-life of 5.3 years and high gamma decay energies of 1.17 and 1.33 MeV. 47 Removal of 60 Co is therefore essential to diminish workers' radiation exposure, to reduce radioactive emissions to aquatic systems and to create safe waste forms for the nal disposal. However, the separation of this radionuclide is quite difficult owing to its low concentration in the liquid waste of NPPs (402 Bq L À1 60 Co in oor drain water, Loviisa NPP, Finland). 48 The process for its uptake has to be very selective, since generally the liquid waste contains large amounts of inactive metal ions (e.g. 37 mg L À1 Na + in oor drain water, Loviisa NPP, Finland). 48 In this research, we prepared SnO 2 and composite SnO 2 /SiO 2 submicron bers by the electroblowing technique and calcination. We also studied the ability of the bers to remove 57 Co 2+ from an aqueous solution. To investigate the effect of SiO 2 on both the structure and Co 2+ uptake performance of the SnO 2 bers, we added 25 mol% of silicon in the synthesis solution of the composite bers. The structure and morphology of all the bers were characterized and their Co 2+ removal capability was studied. To the best of our knowledge, this is the rst report on the electroblowing synthesis of SnO 2 and SnO 2 /SiO 2 submicron bers and on their Co 2+ uptake. This is also the rst report on the uptake of metal ions by brous SnO 2 and SnO 2 /SiO 2 .

Materials
The precursor solutions for the electroblowing experiments were prepared from SnCl 4 $5H 2 O ($98%, Sigma-Aldrich), tetraethoxysilane (TEOS, Si(C 2 H 5 O) 4 , 98%, Sigma-Aldrich), polyvinylpyrrolidone (PVP, (C 6 H 9 NO) n , M w ¼ 1 300 000, Alfa Aesar), N,N-dimethylformamide (DMF, C 3 H 7 NO, $99.9%, Sigma-Aldrich) ethanol (C 2 H 5 OH, 96 vol%, GPR RECTAPUR) and deionized water. In the uptake experiments 57  For the electroblowing, a self-assembled apparatus was used. 49 In a typical experiment, 12 mL of precursor solution was withdrawn into a syringe and a 27 G (inner diam. 0.21 mm) needle was attached to the syringe. The syringe was placed on a syringe infusion pump (KD Scientic Legato® 101) and the solution feed rate was set to 15 mL h À1 . This feed rate is 4 to 500 times as high as reported for the electrospinning of SnO 2 and SnO 2 /SiO 2 bers. [21][22][23][24]26,50,51 The needle was pushed through a 3 mm metal adapter of a box enclosing a cylindrical side collector and a planar back collector at 80 cm distance, both being made of a metal wire mesh. The potential difference between the needle and collectors was set to 15 kV with a high voltage power source, and compressed air was delivered through the adapter at a rate of 30 NL min À1 . The solution jet erupted from the needle tip was deposited as bers on the grounded collectors and additional drying air was delivered to the box at a rate of 40 NL min À1 to enhance solvent evaporation and to control the relative humidity within the box (#20%). The as-electroblown brous mats were detached from the collectors and calcined in an air furnace at 400, 450 and 500 C for 4 hours with a heating rate of 1 C min À1 in order to remove the polymer and to form the desired ceramic material. The as-electroblown SnO 2 /SiO 2 /PVP bers were also calcined at 400 C for 4 hours with heating rates of 5 and 10 C min À1 . The yields of SnO 2 and SnO 2 /SiO 2 bers were at best 0.54 and 0.53 g per hour of electroblowing, respectively.

Characterization of the SnO 2 and SnO 2 /SiO 2 bers
Morphology of the bers was analysed by imaging with secondary electrons (SE) and transmitted electrons (TE) with a Hitachi S-4800 eld emission SEM. Prior to the imaging with secondary electrons, the samples were placed on carbon tape and sputter coated with a 4 nm layer of Au/Pd alloy to improve conductivity. Elemental analysis of the bers including elemental mapping was conducted with an Oxford INCA 350 energy dispersive X-ray spectroscopy (EDX) system connected with the Hitachi S-4800. A Quanta 3D 200i focused ion beam SEM (FIB-SEM) equipped with an Oxford INCA 350 EDX system and an Omniprobe nanomanipulator was used for extracting single SnO 2 /SiO 2 bers onto a copper grid for cross-sectional elemental mapping. The average diameters of the bers and their grains were determined with a Fiji ImageJ soware. The crystallinity of the bers was analysed with a PANalytical X'Pert PRO MPD X-ray diffractometer using Cu Ka radiation and focusing optics. The ber samples were powdered prior to the analysis. The mean crystallite sizes and their weight ratios were determined from the XRD data by the Rietveld renement using a MAUD soware. 52 Thermogravimetric analysis (TGA) of the aselectroblown bers was conducted with a NETZSCH STA 449 F3 Jupiter® system using a heating rate of 10 C min À1 in a temperature range of 25 to 1000 C in a ow of air (50 mol%) and N 2 (50 mol%, the purge gas). The specic surface area and porosity of the bers were measured by N 2 physisorption at 77 K with a Micromeritics ASAP 2020 Gas sorption analyser.
2.4 Co 2+ uptake studies 2.4.1 Effect of calcination temperature and heating rate. Co 2+ uptake by the SnO 2 and SnO 2 /SiO 2 bers calcined at different temperatures and with different heating rates was studied at a pH of 6. 20 mg of ground bers was weighed into 20 mL scintillation vials and 10 mL of 0.01 M NaNO 3 solution containing 30 Bq mL À1 57 Co 2+ was added into the vials. The pH of the solution was adjusted to 6 with a small volume of NaOH. The samples were equilibrated in a constant rotary mixer (50 rpm) for 24 hours aer which the equilibrium pH was measured. The samples were phase separated by centrifugation at 4000 rpm (2100g) and syringe ltration (Acrodisc LC PVDF, 0.2 mm). The 57 Co 2+ uptake efficiency of each sample was determined by pipetting 5 mL of the ltrate into a scintillation vial and measuring the remaining activity with a PerkinElmer Wallac Wizard 3 00 1480 automatic gamma counter. The 57 Co 2+ uptake results are presented by means of distribution coefficient K d , that describes the distribution of the adsorbate between the adsorbent and solution: where c 0 (Bq L À1 ) is the initial concentration, c eq (Bq L À1 ) is the equilibrium concentration, V (mL) is the volume of the solution and m (g) is the mass of dry adsorbent. Background activity was subtracted before the calculations. Uncertainty of K d was calculated using the error propagation law. 2.4.2 Effect of pH. Co 2+ uptake by SnO 2 bers calcined at 500 C was investigated in the pH range of 4 to 12. The batch samples were prepared as described above and the pH of the solution was adjusted with a small volume of either HCl or NaOH. The equilibrium pH was measured aer the 24 hours of constant rotary mixing. The 57 Co 2+ uptake by the bers was calculated by means of the distribution coefficient K d as described above.
2.4.3 Effect of coexisting ions. Selectivity of SnO 2 bers towards Co 2+ in the presence of competing ions Na + and Ca 2+ was examined. The batch samples were prepared as reported Fig. 1 As-electroblown SnO 2 /PVP (a) and SnO 2 /SiO 2 /PVP (b) fibers, SnO 2 fibers calcined at 400 C (c) and 500 C (d) as well as SnO 2 /SiO 2 composite fibers calcined at 400 C (e) and 500 C (f) on 150 mm silicon wafers. In the calcination, a heating rate of 1 C min À1 was used. The calcined fiber mats are shrunken due to the removal of the polymer. above and the initial pH was adjusted to 6. Four concentrations of the competing ion were used: 0.001, 0.01, 0.1 and 1 M for NaCl, and 0.001, 0.01, 0.1 and 0.5 M for CaCl 2 . The 57 Co 2+ removal by the bers was calculated as the distribution coefficient K d .

EDX analysis.
Elemental analysis of the SnO 2 and SnO 2 /SiO 2 bers aer adsorption of Co 2+ including elemental mapping was conducted. For the analysis, 20 mg of bers was weighed into 20 mL scintillation vials and 10 mL of 0.01 M NaNO 3 solution containing 1 mM non-radioactive Co 2+ (Co(NO 3 ) 2 $6H 2 O) was added into the vials. The pH of the solution was adjusted to 6. The samples were equilibrated for 24 h, phase separated and dried in an oven at 70 C overnight.

Results and discussion
3.1 Electron microscopy and TG analysis of the SnO 2 and SnO 2 /SiO 2 bers Photographs of as-electroblown SnO 2 /PVP and SnO 2 /SiO 2 /PVP bers as well as SnO 2 and SnO 2 /SiO 2 bers calcined at 400 and 500 C are shown in Fig. 1. The calcined ber mats have shrunk because the polymer has been removed. Except for the SnO 2 /SiO 2 bers calcined at 400 C, the colour of the calcined bers is white, which implies no major amounts of carbon residues i.e. efficient combustion of the polymer. In regard to the SnO 2 /SiO 2 bers with a brownish hue, both the lower calcination temperature of 400 C and the presence of SiO 2 probably cause incomplete combustion of the PVP. Based on TG analysis, 53 combustion of the bare PVP bers is not complete until at 700 C and hence there may be carbon residues in the calcined bers although not necessarily to a visible extent. The amorphous SiO 2 phase in the composite bers may be more prone to retain the amorphous polymer compared to the crystalline SnO 2 bers and SiO 2 and PVP are known to form hydrogen bonds. 54,55 Fig. 2 TG curves of as-electroblown SnO 2 /PVP and SnO 2 /SiO 2 /PVP fibers in air (50 mol%) and N 2 (50 mol%, the purge gas).   and 500 C (f-j) with a heating rate of 1 C min À1 as well as SnO 2 /SiO 2 composite fibers calcined at 400 C with heating rates of 5 C min À1 (k and l) and 10 C min À1 (m and n).
According to the TG analysis of the as-electroblown SnO 2 / PVP and SnO 2 /SiO 2 /PVP bers (Fig. 2), the combustion of the PVP in the SnO 2 /SiO 2 composite bers is not complete until at 600 C whereas it is complete at 500 C in the case of the bare SnO 2 bers. Furthermore, it can be seen that the mass of pure SnO 2 is ca. 15% and the mass of pure SnO 2 /SiO 2 is ca. 20% of the mass of the as-electroblown material. The measured weights of the calcined SnO 2 and SnO 2 /SiO 2 bers are in line with this, also for the lowest calcination temperature of 400 C. This implies a more effective combustion of the polymer in the calcining furnace which is probably due to a higher oxygen content as the furnace atmosphere comprises air instead of air/N 2 mixture and a long 4 hour duration of the calcination. Owing to the presence of SiO 2 , the calcined SnO 2 /SiO 2 composite ber mats were elastic and could be bent without causing fractures to them. The SnO 2 ber mats were more brittle and fractured when bent. The differences in elasticity between the SnO 2 and SnO 2 /SiO 2 bers are demonstrated in the video "Bending experiments with the SnO 2 and SnO 2 /SiO 2 bers".
FESEM images of the uncalcined SnO 2 /PVP bers and SnO 2 bers calcined at 400, 450 and 500 C are presented in Fig. 3. The average diameter of the uncalcined bers containing the polymer was 1.5 mm whereas the average diameter of the calcined bers was 560 nm. The morphology of the calcined SnO 2 bers was similar regardless of the calcination temperature. The bers seemed to have a uniform structure that consisted of roundish grains approximately 20 nm in diameter, although there was some variation between individual grains. A similar granular structure has also been reported for both dense 21,23,26,50 and hollow 22,56 electrospun SnO 2 nanobers. The dense character of the bers in the current study is proved by the FESEM image of a ber cross-section (Fig. 3e) and was veried by imaging with transmitted electrons (Fig. 4a and b).
FESEM images of the uncalcined SnO 2 /SiO 2 /PVP bers and SnO 2 /SiO 2 bers calcined at 400, 450 and 500 C with a heating rate of 1 C min À1 as well as at 400 C with heating rates of 5 and 10 C min À1 are presented in Fig. 5. The average diameters of the uncalcined and calcined bers were 1.2 mm and 580 nm, respectively. Irrespective of the calcination temperature, the  appearance of the bers calcined with a heating rate of 1 C min À1 was the same (Fig. 5b-h). The bers seemed to have a smooth core that was rather sparsely covered with roundish grains approximately 30 nm in diameter. A rather similar structure comprising a smooth core and granular surface has been reported previously for SnO 2 /SiO 2 nanobers, although with a higher proportion of Si from 50 to 83 mol% compared to ours of 25 mol%. 24 As for the bers calcined with the faster heating rates, they also had a smooth core sparsely covered with grains that were quite angular in shape and rather large (Fig. 5k-n). The average diameter of the grains was ca. 200 nm in the bers calcined with heating rates of 5 and 10 C min À1 , although there was some variation between individual bers. Among the bers calcined at 500 C with the slow heating rate of 1 C min À1 , there were also some bers with a special morphology with the ber core covered with large, irregular grains of 500 nm to 1 mm in diameter ( Fig. 5i and j). FESEM (Fig. 5h) and TE imaging ( Fig. 4c and d) conrmed that the structure of the SnO 2 /SiO 2 composite bers was dense like that of the bare SnO 2 bers.   3.2 EDX analysis of the SnO 2 and SnO 2 /SiO 2 bers EDX spectra of the SnO 2 bers are shown in Fig. S1 and S3. † As for the SnO 2 /SiO 2 composite bers, the quantitative Sn : Si EDX results were 77 at% : 23 at% and 80 at% : 20 at% for the bers calcined at 400 and 500 C, respectively ( Fig. S2 and S4 †). Considering possible variation in sample homogeneity, the results are as expected and indicate a successful synthesis. Owing to the difficult quantication of light elements with EDX and because detected carbon may also originate in the environment, the amount of residual carbon in the bers could not be reliably determined.
EDX elemental maps of a single SnO 2 /SiO 2 ber calcined at 500 C are shown in Fig. 6. EDX spectra recorded from both inner and outer section of the smooth part of the ber are presented in Fig. 7. A TE image of a thin inner section of the ber is shown in Fig. 8. For the EDX measurement, a portion of the ber was FIB milled away to expose a at longitudinal crosssection surface from which the spectra were recorded. For the TE imaging, equal longitudinal portions from both sides of the ber were FIB milled away leaving a thin slab in the middle of which the imaging was done. The elemental maps prove the presence of Sn, Si and O in the ber but on the basis of them it is difficult to tell any difference between the distribution or the concentration of Sn and Si in the ber. The EDX spectra of the center and outer smooth parts of the ber are quite identical and they both show the presence of Sn. This implies that the core of the SnO 2 /SiO 2 composite bers in this study is different from the single-phase amorphous SiO 2 core reported for the SnO 2 /SiO 2 composite bers with a higher proportion of Si. 24 The TE image from inside the ber supports this as it suggests that the inner part of the ber is not homogeneous but consists of distinct nanoscale domains, possibly small SnO 2 grains embedded in a SiO 2 matrix. Previously it has been proved that more than ca. 1 mol% of SnO 2 cannot be dissolved in SiO 2 but SnO 2 forms crystalline nanoclusters dispersed in amorphous SiO 2 . 57,58 EDX elemental maps of the SnO 2 /SiO 2 composite bers calcined at 400 C with a heating rate of 10 C min À1 are presented in Fig. 9. As can be seen, Sn is more concentrated at the sites where the large grains are located while Si is more concentrated at the smooth part of the bers. This implies that the large grains on the surface of the bers consist primarily of SnO 2 . More evidence is provided by EDX spectra (Fig. 10) recorded at the smooth part (spectrum 1) and at a surface grain (spectrum 2) of the bers. In the spectrum recorded at the grain there are strong signals for Sn and a very weak signal for Si while in the spectrum recorded at the smooth part the case is the opposite, i.e. the signal for Si is very intense and the signals for Sn are quite low.

Crystal structure of the SnO 2 and SnO 2 /SiO 2 bers
X-ray diffraction patterns of both bare SnO 2 and composite SnO 2 /SiO 2 bers are presented in Fig. 11-13. All the bers Fig. 11 X-ray diffraction patterns of the SnO 2 fibers calcined at 400, 450 and 500 C. Cryst. refers to average crystallite size in this and Fig. 12 and 13.  irrespective of the calcination temperature or heating rate have the tetragonal rutile structure. The presence of SiO 2 doesn't seem to affect the crystalline phase nor the lattice parameters compared to bare SnO 2 bers but instead it affects the crystallite size. Interestingly, in both SnO 2 and SnO 2 /SiO 2 bers there were crystallites of two different sizes for which the weight ratios were determined (Table 1). Rening the XRD data with a single crystallite size led to unsatisfactory results while clear improvement was achieved by using two different crystallite sizes with varying weight ratios. This was pronounced with the composite bers. However, the bimodal size distribution may not be sufficient and a more complex distribution is probable, but from the perspective of this study it is adequate to show that there are at least two different crystallite sizes present. The fraction of one size was always much higher than that of the other size, ranging from 63 to 93 wt%. In all the bers the larger crystallites were predominant over the smaller ones, except for the composite bers calcined at 400 and 450 C with a heating rate of 1 C min À1 (Fig. 5b-e). The sizes of the proportionally larger crystallites varied between 10 and 243 nm; the bers calcined at 400 and 450 C with slow heating possessed the smallest of them and the bers calcined at 500 C or at 400 C with fast heating the largest of them. The sizes of the proportionally smaller crystallites varied from 2 to 15 nm and also here the size followed calcination temperature and heating rate: the bers calcined at 400 and 450 C with slow heating possessed the smallest of them and the bers calcined at 500 C or at 400 C with fast heating the largest of them. Rietveld rened Xray diffractograms of SnO 2 and SnO 2 /SiO 2 bers calcined at 500 C are presented in Fig. S13 and S14. † The crystallites in the SnO 2 bers calcined at 400 and 450 C are in average 11 and 10 nm in size, respectively, whereas in the corresponding SnO 2 /SiO 2 bers they are only 2 nm in size. The silica in the composite bers probably acts as a crystal growth inhibitor, as is also the case with SnO 2 nanoparticles embedded in a SiO 2 network. 59 It should be noted though that the renement results of the broad bumps in this data (Fig. 12) leading to below 10 nm sized crystallites represent rather the lack of long range order than an actual precise size, since severe overlapping of the peaks showing poor crystallinity undermines the accuracy of the results. In both the bare and composite bers, the average crystallite size increases substantially as the calcination temperature is increased and in the composite bers the increase is quite steep ( Table 1). It is known that the crystallite size of ceramic submicron bers depends on the calcination temperature. 44 In the bare SnO 2 bers, the crystallite growth is probably accompanied with a grain growth at the higher temperatures 21 although no prominent difference in grain size between the bers calcined at 400, 450 and 500 C was observed on the basis of the FESEM analysis (Fig. 3). As for the composite bers calcined at 500 C with a heating rate of 1 C min À1 and at 400 C with heating rates of 5 and 10 C min À1 , both the high calcination temperature and fast heating rates could explain the large crystallite sizes of 243, 230 and 111 nm, respectively (Table  1). It is likely that the large, irregular grains on the bers (Fig. 5i-n) consist mainly of SnO 2 that is known to form various shapes in nanoscale. 50,60 High calcination temperatures and fast heating rates may inuence both the size and shape of crystals, 61 which is reected in the morphology of grains comprising the crystals.
The presence of crystallites of two different sizes in the SnO 2 / SiO 2 composite bers is reasonable in the light of the electron microscopy and EDX analysis results. Based on EDX analysis, both Sn and Si are present in the core of the composite bers but the surface grains consist primarily of SnO 2 (Fig. 7, 9 and 10). It is seen in the TEM image (Fig. 8) that the core of the bers comprises different nanoscale domains probably of SnO 2 and SiO 2 . As discussed, SiO 2 in the core of the bers presumably prevents SnO 2 crystallite growth. This might cause the presence of small SnO 2 crystallites and grains in the core of the bers. The SnO 2 crystallites can grow more freely on the surface of the bers resulting in larger crystallite and grain sizes. For the SnO 2 /SiO 2 bers calcined at 400 and 450 C with a heating rate of 1 C min À1 the relatively high amount of small 2 nm crystallites (83 and 80 wt%, respectively, Table 1) and the moderate size of the larger crystallites (14 and 11 nm, respectively, Table  1) are probably due to the low calcination temperature and slow heating rate. As regards the bare SnO 2 bers, the presence of the smaller crystallites (3, 2 and 12 nm for the bers calcined at 400, 450 and 500 C, respectively, Table 1) might be at least partly due to some PVP residues hindering crystal growth. This may especially be the case for the bers calcined at the lowest temperature of 400 C for which the fraction of the small crystallites, 37 wt%, is the highest of the SnO 2 bers. The average crystallite sizes of 11, 10 and 12 nm (SnO 2 , Table 1) as well as 14, 11 and 11 nm (SnO 2 /SiO 2 , Table 1) for the bers calcined at 400, 450 and 500 C with slow heating, respectively, coincide with some SnO 2 grain sizes seen in the FESEM images (Fig. 3d, e, g, i and 5c, e, g). Previously it has been found that the crystal size of granular polycrystalline SnO 2 nanobers is approximately the same as the size of the SnO 2 grains meaning that each SnO 2 grain is possibly a single crystal. 26 Perhaps this is the case for some SnO 2 grains in the bers of this study as well.
3.4 Specic surface area, pore volume and pore size analysis of the SnO 2 bers Specic surface area of the SnO 2 bers calcined at 500 C was analysed by the Brunauer-Emmett-Teller (BET) method and total pore volume, pore size distribution and average pore diameter by the Barrett-Joyner-Halenda (BJH) method using nitrogen gas adsorption and desorption. The specic surface area, total pore volume and average pore size of the bers were 41 m 2 g À1 , 0.15 cm 3 g À1 and 10 nm, respectively. The surface area of the bers is somewhat higher than that found in literature for both dense 51 and hollow 56,62 electrospun SnO 2 bers calcined at 500 or 600 C, 6.9 to 36 m 2 g À1 . The same is true of the total pore volume, as a smaller pore volume of 0.079 cm 3 g À1 has been reported by Mudra et al. 62 while a larger average pore size of 16 nm has been reported by Xia et al. 56 The surface area of the SnO 2 bers is quite high considering the large 71 nm crystallites, as large crystallites commonly result in a small surface area. For ZrO 2 bers of almost the same diameter (570 nm) crystallites of 9 and 63 nm produced surface areas of 14 and 1.7 m 2 g À1 , respectively. 44 The large surface area of the SnO 2 bers of this study can probably be explained by the granular and porous structure. The average pore size of 10 nm seems to match quite well with the occasional interstices between the grains of the bers (Fig. 3d, e, g and i). N 2 adsorption and desorption isotherms and pore size distribution of the SnO 2 bers are presented in Fig. S15 and S16, † respectively.

Co 2+ uptake by the SnO 2 and SnO 2 /SiO 2 bers
The adsorption ability of electroblown submicron bers is greatly affected by their crystal structure and size, morphology and the amount of possible polymer residues. These characteristics, in turn, depend on the calcination temperature. 44 Therefore, we investigated the effects of calcination temperature and heating rate on the Co 2+ uptake by the SnO 2 and SnO 2 / SiO 2 bers. As revealed by Table 2, all the bers have a good Co 2+ uptake performance with an average uptake of 99.71% and 99.53% for SnO 2 and SnO 2 /SiO 2 bers, respectively. However, there were some differences between the bers. As seen from the K d values, the bare SnO 2 bers have somewhat better Co 2+ uptake than the SnO 2 /SiO 2 composite bers. The SiO 2 in the composite bers seems to impair their Co 2+ uptake, although chemisorption of Co 2+ on SiO 2 is known. 63 In this study, however, the crystalline SnO 2 appears to be the major adsorbent. The high surface area of the SnO 2 bers is likely to enhance the adsorption even more as it should provide plenty of adsorption sites. EDX elemental maps and spectra of the SnO 2 and SnO 2 /SiO 2 bers calcined at 400 and 500 C with a heating rate of 1 C min À1 aer adsorption of Co 2+ are shown in Fig. S5-S12. † In regard to the SnO 2 and SnO 2 /SiO 2 bers calcined with a heating rate of 1 C min À1 , the calcination temperature had a different effect on their Co 2+ uptake performance. Calcination temperatures of 500 and 400 C produced the best adsorption performance for SnO 2 and SnO 2 /SiO 2 bers, respectively ( Table  2 and Fig. 14). Lower calcination temperature tends to produce smaller crystallites and thus increase the surface area of the material, which, in turn, improves its adsorption properties. 44 Therefore it is surprising that the SnO 2 bers calcined at 500 C possessing 71 nm crystallites perform better than the SnO 2 bers calcined at 400 and 450 C possessing 11 and 10 nm crystallites, respectively. One possible reason is that the lower calcination temperatures leave some PVP residues in the material blocking some of the adsorption sites. However, despite their different crystallite size, the K d value of the SnO 2 bers calcined at 450 C is almost as high as that of the bers calcined at 500 C. The crystallite size doesn't seem to be a determining factor in the Co 2+ uptake performance of the SnO 2 bers.
In the SnO 2 /SiO 2 bers calcined with a heating rate of 1 C min À1 the Co 2+ uptake correlated better with the crystallite size, since the bers calcined at 400 C and with the smallest 2 nm crystallites performed the best (Table 2 and Fig. 14). It should be noted that although the bers calcined at 450 C also possessed 2 nm crystallites, their sorption and K d values were lower. One reason for the weaker Co 2+ uptake performance of the SnO 2 /SiO 2 bers calcined at 450 and 500 C with a heating rate of 1 C min À1 could be that due to the higher calcination temperature, they have fewer surface hydroxyl groups that are assumed to play a signicant role in the uptake process. As for the bers calcined at 500 C, the peculiar morphology of some  of the bers (Fig. 5i and j) and large crystallite size of 243 nm that decreases the surface area, may also impair their Co 2+ uptake performance. The SnO 2 /SiO 2 bers calcined at 400 C with the fast heating rates of 5 and 10 C min À1 exhibited excellent Co 2+ uptake despite their large crystallite sizes. This might be explained by the roundish shape and quite sparse distribution of their SnO 2 grains (Fig. 5k-n) compared to the angular and rather tightly packed SnO 2 grains of the bers calcined at 500 C ( Fig. 5i and j). The roundish and sparsely distributed surface SnO 2 grains might be better accessible to adsorbing ions in an aqueous solution. Moreover, the lower target temperature of 400 C may cause more hydroxyl groups remaining on the surface of the bers enhancing the uptake of Co 2+ . The pH of the solution may have a major impact on the uptake properties of a material because it affects both the speciation of the adsorbate and the surface charge of the adsorbent. As for Co 2+ , it exists as Co(H 2 O) 6 2+ in the pH range of 2 to 9. 64 It can be inferred from the K d values (Fig. 15), that the SnO 2 bers adsorb Co 2+ the best in the neutral to mildly basic pH region, from pH 6 to 9, and they reach the highest K d value at the pH of 7. This is in accordance with previous research 7 and promising for the use of the bers in purication of NPP waste waters, since the pH of the primary coolant water in NPPs is about 7. The point of zero charge (pH pzc )of SnO 2 lies at a pH of 4, 7 and above this pH the surface charge of pure SnO 2 is negative. The Co 2+ uptake by the rutile structured SnO 2 is evidently based on electrostatic forces 65 which explains the good uptake in the neutral to mildly basic pH region where the charges of the Co 2+ species and the surface of the SnO 2 bers are the opposite. The adsorption of Co 2+ on hydrous SnO 2 most probably occurs via substitution of H + ions of the surface water molecules or hydroxyl groups by Co 2+ ions. [4][5][6]66 An assumed ion exchange reaction between hydrous SnO 2 and Co 2+ is illustrated in Fig. 17. 5 A good selectivity of the adsorbent is crucial when a trace amount of a specic ion is separated from a solution containing much higher concentrations of other ions. Thus, we examined how the Co 2+ adsorption on SnO 2 bers is inuenced by Na + and Ca 2+ ions that are among the most common cations in natural and nuclear waste waters with concentrations of 0.47 and 0.01 M in sea water, respectively. 67 As Fig. 16 reveals, the Co 2+ uptake remains high in the presence of Na + ions irrespective of their concentration. By contrast, there is a marked weakening of the uptake in the CaCl 2 solution at concentrations of 0.01 M or higher. With both of the ions, the interfering effect increases with concentration up to 0.1 M, aer which there is no further decreasing effect on the uptake of Co 2+ . The good selectivity of the bers for Co 2+ over Na + is promising for their use in decontamination of radioactive waste water, since Na + is the most abundant coexisting ion in nuclear waste effluents. 67 Fig. 16 Effects of concentrations of competing ions Na + or Ca 2+ on the uptake of 30 Bq mL À1 57 Co 2+ by SnO 2 fibers calcined at 500 C. The initial pH was 6 and the pH at the end of the experiment is shown (2 g L À1 SnO 2 ).

Conclusions
We have synthesized SnO 2 and composite SnO 2 /SiO 2 submicron bers with a Sn : Si molar ratio of 3 : 1 and studied the ability of the bers to remove Co 2+ from an aqueous solution. For the synthesis, a novel and efficient electroblowing method was used. The as-electroblown bers were calcined in air at 400, 450 and 500 C with varying heating rates in order to produce the desired ceramic material and to investigate the effect of calcination temperature and heating rate on the structure and Co 2+ uptake by the bers. The bare SnO 2 bers had a granular structure in the tetragonal rutile phase with an average diameter of 560 nm. The SnO 2 /SiO 2 composite bers had a smooth core possibly comprising small SnO 2 grains in a SiO 2 matrix with large SnO 2 grains dispersed on the core and the average diameter of the bers was 580 nm. The morphology of the surface SnO 2 grains of the composite bers was dependent on the calcination temperature and heating rate. Irrespective of the calcination temperature, a heating rate of 1 C min À1 yielded surface SnO 2 grains that were roundish and ca. 30 nm in diameter. In the composite bers calcined at 500 C with a heating rate of 1 C min À1 or at 400 C with faster heating rates of 5 and 10 C min À1 there were also some surface SnO 2 grains that were irregularly shaped and 200 nm to 1 mm in diameter.
Both bare SnO 2 and composite SnO 2 /SiO 2 bers had a high Co 2+ uptake with SnO 2 bers exhibiting slightly more efficient Co 2+ separation. Calcination temperature and heating rate affected the Co 2+ uptake by the bers. Among the SnO 2 bers, the bers calcined at 500 C performed the best. Among the composite bers, the bers calcined at 400 C performed the best, and within them the bers calcined with a heating rate of 10 C min À1 were superior to the bers calcined with heating rates of 1 and 5 C min À1 . The SnO 2 /SiO 2 composite bers were more elastic and durable and easier to handle than the somewhat brittle SnO 2 bers which makes them ideal for use in owthrough separation columns. Overall, on the basis of our results, the mechanical strength of SnO 2 submicron bers can be enhanced by adding a moderate amount of SiO 2 without compromising their adsorption performance too much. This approach might also be extended to other ceramic submicron and nanobers that require improved mechanical properties in various applications.

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