2.1. Fabrication of the Composite Cryogels Containing Transition-Metal Ferrocyanides
Taking into account our earlier negative experience with the fabrication of composite PEI cryogel for cesium ion sorption via the sequential loading of PEI monolith with Cu(II) and [Fe(CN)
6]
4− ions [
20], and the assumption that weaker binding of precursor metal ions to PEI will be beneficial for the fabrication of the composite cryogel with good selectivity to cesium ions, we have tested the same approach (Method I,
Figure 1) with Zn(II), Ni(II), and Co(II) ions. The affinity of PEI cryogel to these ions is notably lower than to Cu(II) ions [
13,
21]; besides, in the case of a Co(II)-loaded monolith, it is important to treat cryogel with [Fe(CN)
6]
4− ions as soon as possible to avoid the oxidation of Co(II) to Co(III), which binds to PEI irreversibly [
13]. The level of the composite loading with inorganic sorbent using the Method I will be determined by the sorption capacity of the original PEI cryogel, which changes in the order Cu(II) > Zn(II) > Ni(II) > Co(II) (
Figure S1, Supplementary Material), meaning that the composite cryogel with Zn(II) ferrocyanide would be the preferable option in terms of maximal loading with the sorption active component.
Figure 2 shows that in situ modification of PEI cryogel with ferrocyanides did not affect porous structure. The surface of the cryogel containing Zn(II) ferrocyanide (ZnFC) was covered with aggregates of nanocrystals, while the surface of the composite containing CuFC remained smooth. Smooth surfaces without a visible crystalline phase were also observed for composites containing Co(II) and Ni(II) ferrocyanides (CoFC and NiFC, respectively). However, EDX analysis in all cases confirmed the presence of homogeneously distributed Cu, Zn, Ni, or Co, and Fe at a M/Fe ratio close to the theoretical value of 2. Only for PEI/ZnFC cryogel was a significant amount of potassium detected by SEM-EDX, confirming the formation of the mixed potassium–zinc ferrocyanide (
Table 1). Screening the sorption properties of the composites showed that only PEI/ZnFC cryogel was efficient for Cs
+ ion uptake (
Table 1).
Although the presence of alkali ions (mainly, K
+) in ferrocyanides is often correlated with the enhancement of uptake kinetics and the increase of sorption capacities [
8], good performance of ZnFC-based cryogel cannot be related directly to the formation of the mixed ferrocyanide in this case, since ion exchange of Cs
+ ions with K
+ is only one of the possible sorption mechanisms [
12]. We showed earlier, for example, that Cu
2+ ions were released upon efficient Cs
+ sorption on composite polyallylamine/CuFC cryogel fabricated via the precipitation of the ex situ-formed CuFC colloids in the porous matrix [
20]. SEM images of different areas of the monolith from top down and from the center to the periphery demonstrated homogeneous distribution of the ferrocyanide phase, which is an important advantage of the in situ fabrication method (Method I,
Figure 1).
Comparison of the swelling behavior and stiffness of the original PEI cryogel and PEI/ZnFC composite (
Figure 3) shows that the presence of the inorganic phase increased the stiffness of the cryogel, but had a minor effect on the overall swelling degree and contributions of free-flowing water (macropores) and bound water (polymer phase). As earlier reported for the composite cryogels [
17,
22], the introduction of inorganic particles into the porous polymer matrix could have a controversial effect on mechanical properties. Although the embedding of small TiO
2 nanoparticles (6 nm) in the polymer walls notably increased compressive moduli [
22], larger particles (25–27 nm) either improved [
17] or worsened [
22] mechanical properties of the composite.
The results obtained for the PEI/ZnFC composite here comply with the cryogel morphology. Since inorganic nanoparticles are located on the surface rather than embedded in the polymer, they do not have a drastic negative effect on cryogel stiffness, despite the large size of ZnFC nanoparticles (516 ± 146 nm) and the presence of aggregates of a size above 3 µm (
Figure 4a). These aggregates also do not affect free water flow in the monolith due to the large pore size of the swollen PEI cryogel (128 ± 30 µm [
23]). However, even a slightly decreased swelling degree of the composite cryogel can contribute to an increase in the Young modulus, since mechanical properties of cryogel significantly depend on water content. It should be noted that in our preliminary experiments aimed to control sorbent loading degree with PEI cross-linking density (i.e., the content of free amino groups capable of binding precursor metal ions), we found that highly cross-linked and less swellable cryogel was not applicable in columns, since the increase of stiffness after in situ formation of nanoparticles resulted either in a loss of permeability or in cryogel compression and formation of channels near the sorption column walls, meaning that the feeding solution went through these channels rather than through the cryogel bulk.
Method II (
Figure 1), which assumes cross-linking of the polymer in dispersion containing ex situ-formed sorbent nanoparticles, is attractive due to its simplicity and possibility to control loading degree. At the same time, sedimentation of the sorbent nanoparticles during freezing and cross-linking stages can lead to the inhomogeneous distribution of the inorganic phase in a porous polymer matrix. To shorten the freezing time, the solutions were precooled, but, due to the low viscosity of PEI solution, sedimentation was difficult to avoid. For the sake of comparison between different fabrication methods, we have focused only on the composite containing Zn(II) ferrocyanide, which can be obtained by both methods.
Figure 4b shows typical images for the composite fabricated via the Method II PEI porous structure with single ZnFC nanoparticles of average size 906 ± 165 µm, partially or completely embedded in the polymer matrix. At the same time, SEM-EDX analysis of different parts of the composite cryogel revealed inhomogeneous distribution of the inorganic phase from the center to the periphery and from top down, and the presence of the regions with large aggregates of ZnFC particles, which affected the integrity of the porous structure (
Figure S2, Supplementary Materials).
Method III (
Figure 1) is often used for the ligand-assisted fabrication of different nanoparticles, and, in comparison with Method II, this approach benefits from the stabilizing effect of the polymer at the stage of nanoparticle formation [
24]. Due to the instability of transition-metal ferrocyanides in alkaline media, at the initial pH of PEI solution (pH~11), ZnFC nanoparticles that were formed immediately after addition of K
4[Fe(CN)
6] solution to the solution of Zn(II)-PEI complex were dissolved within a few dozen seconds. Adjusting the pH value to 7.4 allowed the formation of stable nanoparticle dispersion, than the cross-linking agent was added within 1 min and the dispersion was frozen. After thawing, the composite cryogel was permeable to the flow, but the SEM-EDX analysis showed drastic changes in morphology (
Figure S2, Supplementary Materials). Although a typical macroporous structure of the PEI cryogel was preserved, large pores of a size from 1 to 15 µm appeared in the pore walls. The atomic ratio of Zn/Fe in the composite varied in a very broad range. In most spots, Zn was in non-stoichiometric excess, reaching a Zn/Fe atomic ratio of up to 15. The overall content of Zn and Fe was significantly lower than in composites obtained by the Methods I and II, although the same amounts of Zn(II) and [Fe(CN)
6]
4− were introduced to all types of composites. It is likely that, under cryoconcentration conditions, PEI negatively affects the stability of initially formed ferrocyanide nanocrystals; therefore, being originally embedded to the polymer phase during the cross-linking step (7 days), they were dissolved and played the role of template rather than that of sorption active component.
2.2. Cs+ Sorption on PEI/ZnFC Composite Cryogels in Batch and Fixed Bed
Figure 5a shows the breakthrough curves for Cs
+ ions sorption on PEI/ZnFC monoliths fabricated via Methods I and II. In all cases, residual Cs concentrations in the outlet solutions until breakthrough point were below the detection limit of AAS (<0.1 mg/L), so that Cs
+ uptake efficiency was above 99.5%. The observed at high flow rate (100 BV/h = 5.53 m/h) steep breakthrough profiles are typical for sorption without significant mass transfer limitation, which makes PEI/ZnFC composite cryogel different from the earlier reported composite cryogels, which showed sloppy breakthrough curves without a horizontal region of high removal efficiency even at significantly lower superficial velocities [
6,
11].
The location of the ferrocyanide nanoparticles inside a polymer matrix (Method II) or at the interface (Method I) had no effect on the shape of the breakthrough curve under the sorption conditions shown in
Figure 5a, although the sorption inactivity of nanoparticles embedded in a polymer matrix was reported for the composite cryogel for protein chromatography [
18]. This proved that the diffusion of cesium ions in the polymer matrix was not the limiting stage of the sorption, complying with our previous study [
23], which revealed a significant role of diffusion limitations only in the case of high-molecular-weight pollutant sorption on PEI cryogel. Thus, the sorption kinetics on ferrocyanide particles is expected to control the sorption performance of the PEI/ZnFC composite.
The release of K
+ ions under dynamic conditions (
Figure 5b) suggests that ion exchange between Cs
+ ions in solution and K
+ ions in the Zn
1.85K
0.3[Fe(CN)
6] lattice significantly contributes to the sorption mechanism on PEI/ZnFC (Method I) composite. However, the maximal sorption capacity (~0.4 mmol/g) determined from the Cs
+ sorption isotherm (
Figure 5d) is higher than the dynamic sorption capacity (0.26 mmol/g) determined from breakthrough curves (
Figure 5a) and the value 0.31 mmol/g, which can be expected from ZnFC chemical composition Zn
1.85K
0.3[Fe(CN)
6] and Zn(II) content in monolith (
Table 1). It is likely that the K
+/Cs
+ ion-exchange works for the fast sorption sites, which are active under dynamic conditions at high flow rates, but, under long equilibration times, other mechanisms or other sorption centers can contribute to the increased sorption capacity.
The presence of sorption sites with different sorption energies in the composite containing potassium–nickel ferrocyanide was hypothesized in [
11], since the bi-site Langmuir equation provided a better fit to the experimental Cs
+ sorption isotherm. Recently, we developed the extended rate-constant distribution (RCD) model for sorption in heterogeneous systems [
13,
14,
21], which provides complete information about sorption properties for material with a continuous distribution of the sorption sites via the calculation of the RCD function from experimental data—sorption breakthrough (dynamics) or sorption kinetic (batch) curves experimentally obtained at different solid:liquid ratios or flow rates, and/or initial concentrations of the adsorbate. All experimental data are processed simultaneously to find the RCD function, which equally well describes a full set of experimental data. Using the RCD function, one can calculate several other distribution functions, including the 2D distribution of sorption sites over constants of sorption rate (ρ(K
s) vs K
s); 2D distribution of sorption sites over constants of desorption rate (ρ(K
d) vs K
d); and 2D affinity—distribution of sorption sites over affinity constants (ρ(K
AF) vs K
AF, K
AF = K
s/K
d × Q
max).
The application of this model to the analysis of Cs
+ sorption breakthrough curves on the PEI/ZnFC composite (Method I,
Figure 5a) enabled the calculation of several distribution functions, as shown in
Figure 5c, and the identification of two types of Cs
+ sorption centers with differences in sorption rate constants of more than 1 log unit. It is likely that the “slow” sites did not contribute significantly to the sorption under dynamic conditions (fixed bed), but could still be accounted for during calculation of the overall sorption capacity, since theoretical sorption capacity was in good agreement with experimental data (
Figure 5d). It is worth mentioning that the investigation of sorption on composite cryogel (Method I) in batch under long equilibration times is complicated by the gradual destruction of the material with a release of nanoparticles into the solution, meaning that very thin discs of the composite were used instead of cryogel fines, which we used in metal ion sorption on the PEI cryogel to eliminate diffusion limitations [
14]. The high number of experimental errors in the initial region of the sorption isotherm and the deviation between the RCD model and experimental data can be explained by this phenomenon.
Modeling the breakthrough curves for Cs
+ ion sorption using the RCD function calculated for the PEI/ZnFC cryogel (
Figure S3, Supplementary Materials) showed that the sorption rate constant (K
s) is high enough to provide an efficient Cs
+ uptake at flow rates of up to 500 BV/h. However, aside from the sorption rate, the important limiting factor is the mechanical stability of the composite, i.e., emergence of operational defects due to the disruption of the porous structure, as earlier observed for Zn(II) and Cu(II) sorption on PEI cryogel at flow rate 242 BV/h (superficial velocity of 13.3 m/h) [
21]. Due to the higher stiffness of the composite (
Figure 3b), the risks of such defects are increasing, e.g.,
Figure 5b shows that the breakthrough point at flow rate 145 BV/h was observed earlier than can be expected from the model, and the column efficiency was 63% of the theoretical value. The most pronounced effects of mechanical properties on sorption performance were observed for composites fabricated via Method II, most likely due to the inhomogeneous distribution of the inorganic phase. This composite showed significant differences in permeability depending on column geometry, was impermeable at flow rates above 100 BV/h and in sea water, while the composite fabricated via Method I was able to uptake Cs
+ ions from sea water (
Figure S4, Supplementary Materials), but with lower dynamic sorption capacity and release of ferrocyanide particles into the solution.
2.3. Cs-137 Uptake with PEI/ZnFC Composite
To avoid the risks of operational defects, the evaluation of the PEI/ZnFC cryogel (Method I) efficiency for Cs-137 radionuclide uptake was investigated at volumetric flow rates of up to 880 mL/h that in the column used (monolith cryogel volume of 8.48 cm3, diameter: height aspect ratio of 2.5; dry composite weight 0.7 g) corresponding to 105 BV/h and superficial velocity of 1.24 m/h.
Figure 6a shows that increasing flow rate from 40 to 880 mL/h had no systematic effect on the decontamination factor, which varied between 1600 and 1900. More precise DF calculations were limited by spectrometer sensitivity; however, in γ-spectra, very weak signals at 662 keV, corresponding to the emission of Cs-137, were detected only for the solutions decontaminated at flow rates of 345 and 880 mL/h, and no difference was observed between background and solution decontaminated at flow rate 40 mL/h (
Figure 6b). This shows the significant advantages of PEI/ZnFC cryogel in comparison to porous chitin discs with immobilized K
2Ni[Fe(CN)
6] nanoparticles [
11], which were efficient under static conditions, but failed to provide high DF in continuous mode without solution recirculation even at low superficial velocity (0.3 m/h).
We have tested the Cs-137 uptake using the same column geometry as in fixed-bed experiments (
Figure 5a) at flow rates of 40, 80, and 120 mL/h corresponding to the superficial velocity 2.21, 4.42, and 6.63 m/h, respectively. In all cases, the residual γ-activity of the purified solutions corresponded to the background level. Thus, one can conclude that the efficient removal of Cs-137 with PEI/ZnFC fabricated with the Method I can be achieved at flow rates of at least up to 6.6 m/h.
This makes this composite cryogel attractive for POU application in flow-through devices in radioactively contaminated areas, primarily in household-type cartridges, e.g., jug water filters. High efficiency and high productivity in combination with small filter size, e.g., a relatively thin disc of the composite cryogel, as demonstrated above, are beneficial for in-field application to treat radioactively contaminated surface water, when other sources of drinking water are not available.
Cesium ion sorption on ferrocyanides is usually irreversible [
12], although efficient cesium elution was reported in exceptional cases for mixed ferrocyanides [
25].
Figure S5 (Supplementary Materials) that only about 25% of cesium was eluted with NaOH solution from the PEI/ZnFC cryogel fabricated via Method I at volumetric ratio eluent/column bed of 18. Virtually higher cesium elution efficiency can be reached with a larger volume of the eluent, but elution profile is not as steep as one expects for the reusable sorbents.
In POU application targeted to Cs-137 radionuclide removal, reusability of any kind of sorbents is not a crucial point, since regeneration would assume the production of the radioactive solution, which must be either processed or stored, while both options are not compatible with application in an emergency. It is more important that cryogel-based composites in dry form are light and, in comparison with matrices based on porous inorganic materials, can be efficiently compacted for safe storage until disposal. This feature can be also beneficial for analytical application, e.g., in Cs-137 ecological monitoring. High sorption efficiency at high flow rates allows cesium preconcentration from the surface water in field studies and collecting light and easy-to-transport samples, which can be further analyzed after oxidative destruction of the organic matrix or using non-invasive spectroscopic methods.