Towards the Extraction of Radioactive Cesium-137 from Water via Graphene/CNT and Nanostructured Prussian Blue Hybrid Nanocomposites: A Review

Abstract

Cesium is a radioactive fission product generated in nuclear power plants and is disposed of as liquid waste. The recent catastrophe at the Fukushima Daiichi nuclear plant in Japan has increased the ¹³⁷Cs and ¹³⁴Cs concentrations in air, soil and water to lethal levels. ¹³⁷Cs has a half-life of 30.4 years, while the half-life of ¹³⁴Cs is around two years, therefore the formers’ detrimental effects linger for a longer period. In addition, cesium is easily transported through water bodies making water contamination an urgent issue to address. Presently, efficient water remediation methods towards the extraction of ¹³⁷Cs are being studied. Prussian blue (PB) and its analogs have shown very high efficiencies in the capture of ¹³⁷Cs⁺ ions. In addition, combining them with magnetic nanoparticles such as Fe₃O₄ allows their recovery via magnetic extraction once exhausted. Graphene and carbon nanotubes (CNT) are the new generation carbon allotropes that possess high specific surface areas. Moreover, the possibility to functionalize them with organic or inorganic materials opens new avenues in water treatment. The combination of PB-CNT/Graphene has shown enhanced ¹³⁷Cs⁺ extraction and their possible applications as membranes can be envisaged. This review will survey these nanocomposites, their efficiency in ¹³⁷Cs⁺ extraction, their possible toxicity, and prospects in large-scale water remediation and succinctly survey other new developments in ¹³⁷Cs⁺ extraction.

1. Introduction

Cesium-137 is a radioactive element with a half-life of 30.4 years emitting both beta and gamma radiations. Under normal circumstances, its release is an outcome of radioactive testing of fission reactions, whereby making it the most abundant radioactive atmospheric pollutant capable of entailing health hazards. The ¹³⁷Cs isotope is produced when uranium and plutonium undergo fission after having absorbed neutrons in a nuclear reactor and is detectable and measurable by gamma counting. It decays into ¹³⁷Ba with a half-life of 2.6 min. Small amounts of ¹³⁷Cs⁺ are released on a regular basis in spent fuel ponds through cracks in the fuel rod that reach the coolant and fuel reprocessing waters, which are all subsequently discharged into the sea as effluents [1]. Today, in view of the Fukushima-Daiichi catastrophe, the extraction of radioactive cesium from soil and water is a necessity and calls for the development of new technologies. The projected radiation effects of ¹³⁷Cs will remain at their maximum for the next 100 years, seconded by isotopes of strontium and plutonium. Other sources of ¹³⁷Cs contamination can be attributed to the Chernobyl accident in 1986; [2] it took 10 years for the levels to drop but their effects are still being perceived even today [3]. In general, ¹³⁷Cs is not very mobile and tends to accumulate on the soil surface and is subsequently absorbed by plants and also captured by fungi [4]. A soil cleanup of 40,000 km² reduced the radioactive contamination to 1/10th of the original value in Chernobyl.

¹³⁷Cs is the primary cause of water contamination since the recent Fukushima Daiichi catastrophe posing more far-reaching threats than via soil contamination [5], which tends to be more localized [6]. ¹³⁷Cs decontamination has now become a priority recommended by the International Atomic Energy Agency [7]. Mobility of ¹³⁷Cs⁺ through moving water bodies displaces the contamination to other areas, thus widening the ‘contamination reduction’ zone. Therefore, aquatic life and fish-farming products are also contaminated, which are thereafter consumed by humans. Additionally, ¹³⁷Cs⁺ behaves very similarly to K⁺ and Na⁺, thus facilitating its digestion and assimilation in living organisms [8]. Consumption of contaminated game and fish would introduce ¹³⁷Cs in the human body, which would then emit harmful radiations directly targeting the cell nucleus. Moreover, in aqueous media ¹³⁷Cs⁺ ions tend to be rather robust and unaffected by changes in pH and redox conditions, whereby making them a menace [9].

Several techniques were developed for the extraction of ¹³⁷Cs⁺ from water: reverse osmosis [10], coagulation-sedimentation [11], ion-exchange [12], nanofiltration [13], electro-dyalysis [14] and more recently, fibrous zeolite-polymer composites [15]. On the other hand, adsorption is a highly efficient and cost-effective process with very high ion selectivity provided that the right sorbents are used [16]. Various adsorbents: inorganic adsorbents, polymer-inorganic adsorbents [7] and bioadsorbents [17], have shown efficiency in ¹³⁷Cs⁺ extraction. Bioadsorbents however, have several disadvantages such as low sorption efficiency, especially in the presence of other salts in the aqueous medium viz., Na⁺ and K⁺. They also suffer from degradation in extreme conditions (high temperature and low pH). On the other hand, inorganic sorbents have shown large-scale applicability with high cation sorption capacities (clays viz., zeolites, bentonite and coal). However, activation of sorption sites is necessary through a surface functionalization treatment, which conversely, increases their production cost. For nanozeolites, the Q_max or maximum adsorption capacity values could reach up to 69 mg/g, but on the other hand, it is difficult to recover the exhausted material. Moreover, the K_d value, which is the ratio of the equilibrium adsorption of the sorbent to the equilibrium concentration of the solute, is low, implying an ineffectiveness in high solute concentrations [18]. Finally yet importantly, they do not withstand harsh aqueous environments. Synthetic polymers combined with inorganic materials seem to be more robust with a better sorption capacity than inorganic sorbents; they also show stability in harsh environments and have been already applied on a large scale in Fukushima [7].

Among the various methods available for the extraction of ¹³⁷Cs⁺ from aqueous solutions, hybrid materials manifest enormous potential in selectively targeting and extracting ¹³⁷Cs⁺ ions [19]. Inorganic ligands such as macrocyclic o-benzo-p-xylyl-22-crown-6-ether (OBPX22C6) ligand bonded to the hydroxyl groups of the mesoporous silica, exhibited a yield of 60% attributed to the Cs-π interaction of the OBPX22C6 benzene. Zeolite-Poly(ethersulfone) composite fiber having 30 wt % loading showed excellent properties for the decontamination of radioactive ¹³⁷Cs⁺ [7]. The decontamination with such composites has also been demonstrated for a contamination of 823Bq/L with pH = 12. Other silicates have also been employed for their known selectivity to ¹³⁷Cs⁺; these include crystalline silicotitanate [20,21], sodium mica [22] and sodium zirconium silicate [23]. Since they are usually in the form of very fine powders, they are therefore unsuitable for column loading. Moreover, they are difficult to separate from aqueous solutions by filtration or centrifugation therefore, reclaiming them once expended becomes problematic.

This review will focus on the recent developments in the extraction of ¹³⁷Cs⁺ from water as depicted in the schematic outline of Figure 1. It will more specifically survey PB-CNT-Graphene based nanocomposite efficiencies in ¹³⁷Cs⁺ extraction. Large-scale applicability in real case scenarios of such nanocomposites will be probed and their nanotoxicity issues will be discussed. A short summary of other new materials is also provided at the end, which opens-up new possibilities in combining these new materials with PB-CNT-graphene based nanocomposites.

2. Prussian Blue

2.1. Structure

Prussian blue (PB) is a dark blue pigment synthesized by ferrous ferrocyanide salts with chemical formula Fe₇(CN)₁₈. It has a porous structure with the capacity to adsorb the ¹³⁷Cs⁺ ions into its pores and store them there. It is a metal organic framework (MOF) where the inorganic vertices, which donate electrons in the structure, are linked to each other via organic compounds. The complete chemical formula is Fe^III₄[Fe^II(CN)₆]₃·xH₂O. The compound has a face-centered cubic structure (FCC) structure (Figure 2) belonging to the Fm3̅m space group with a lattice parameter of 10.166 Å. Fe exists in two oxidation states within the structure: Fe³⁺ and Fe²⁺. These ions form two different FCC lattices displaced by half a lattice parameter with respect to each other. However, the bi and tri-valent Fe are coordinated differently. Furthermore, they are linked to each other via cyanide groups (C ≡ N) i.e., C groups are linked to Fe²⁺ and N groups to Fe³⁺ with high and low spins respectively, in octahedral configurations. The Fe²⁺ and Fe³⁺ ratio of 3:4 implies that in order to obtain a charge neutrality within the structure a 25% vacancy of [Fe⁺²(CN)₆]⁴⁻ molecules is necessary [24]. Coordinated water molecules occupy the resulting octahedral cavities created by such vacancies; six water molecules are linked to Fe²⁺. The other interstitial water molecules occupy the eight corners of the unit cell (¼, ¼, ¼) and are essential for the insertion of the ¹³⁷Cs⁺ ions in the structure.

Fe⁺² can be replaced by other transition metals with the same +2 oxidation states such as Ni, Mn, Cu and Co, coordinated exactly like Fe⁺² in the structure and are called PB analogs. However, there are reports of Cd and Zn with slightly larger atomic radii also being incorporated into the structure owing to their +2 oxidation state [25]. The aim in including different species into the structure is to provoke a distortion of the PB lattice by producing vacancies of the high spin state molecule along with distortions in the vacant cages, in order to facilitate the capture and sequestration of the ¹³⁷Cs⁺ [26].

2.2. ¹³⁷Cs⁺ Ion Capture Mechanism in Prussian Blue

The compound is insoluble in water and the basic mechanism consists of ion exchange of ¹³⁷Cs⁺ and H⁺ with the former occupying hydrophilic vacancies [28]. PB analogs have very different mechanisms of ion exchange or capture depending upon the anionic and alkali metal cation concentrations. Since PB and its analogs contain large amounts of interstitial and coordinated water, ¹³⁷Cs⁺ is captured by a defect created by a [Fe⁺²(CN)₆] vacancy, which creates a spherical cavity whose size is equivalent to the hydration radius of ¹³⁷Cs⁺. Nevertheless, recent calculations have demonstrated that a completely dehydrated ¹³⁷Cs⁺ ion can be incorporated into the structure with the release of a water molecule from the interstitial sites [29]. This is similar to certain clays, where on dehydrating the interlayers the ¹³⁷Cs⁺ selectivity increases [30]. On the other hand, water soluble analogs such as metal hexacyanoferrates (HCF) consisting of a alkali metal cation with a [Fe⁺²(CN)₆] anion, used for the extraction of ¹³⁷Cs⁺ have shown less efficiency. In such compounds Na⁺ or K⁺ are incorporated during the synthesis of the MOF in order to render them water-soluble [31]. In addition to ¹³⁷Cs⁺ capture mechanisms for non-soluble analogs; the water-soluble analogs mainly depend on the Na⁺ or K⁺ ion exchanges with Cs⁺. Takahashi et al., have studied the ¹³⁷Cs⁺ uptake in KCuHCF PB analog in order to understand their lower adsorption capacity [31]. Three main mechanisms governed the ¹³⁷Cs⁺ ion exchange according to them, with the ¹³⁷Cs⁺-K⁺ ion exchanges being predominant, as also stipulated by other research groups. In case of low anionic vacancies, the percolation of ¹³⁷Cs⁺ through the vacancies was prevalent. Finally, for low K⁺ incorporation in the structure, proton exchange between ¹³⁷Cs⁺ and K⁺ ions was evidenced. Ayrault et al., report a degradation in the crystal structure of the KCuHF soluble compound after ¹³⁷Cs⁺ adsorption which was not observed in the non-soluble counterpart [32].

2.3. Nanostructured Prussian Blue

¹³⁷Cs⁺ adsorption in PB crystals is a very slow process. Fujita et al., have demonstrated that after two weeks of adsorption experiments, the depth of ¹³⁷Cs⁺ adsorption was at most between 1–2 nm, irrespective of the crystal size [33]. This implies that most of the adsorption occurs on the surface of the crystallites. This low diffusion depth is mainly attributed to the blocking of the vacancies by captured ¹³⁷Cs⁺ ions, which in turn hinders further ¹³⁷Cs⁺ diffusion. Since the diffusion depth appears to be a constant, increasing the specific surface would therefore be a solution to increasing the ¹³⁷Cs⁺ uptake. One way of augmenting the specific surface is by synthesizing nanoparticles of PB. The surface to volume ratio of crystallites increases as their size decreases; therefore nanoparticles have an extremely large surface to volume ratio. For example, a 3 nm nanoparticle will have 50% of its atoms on its surface. This would also imply that in the case of PB nanocrystals most of the vacancies and sites responsible for ¹³⁷Cs⁺ adsorption would be available on the surface, thus enhancing its specific surface. To this end, different research groups have produced various PB analogs of type Metal(M)-Co, where the nature of M defines the efficiency of the uptake. Liu et al., have demonstrated that Zn assisted Fe-Co PB analogs present high ¹³⁷Cs⁺ uptake efficiency [34]. They also observed that the size of the PB analog particle reduced with the reduction of Fe in the structure; for pure Zn-Co analogs, a crystallite size of ~73 nm was calculated, displaying the highest ¹³⁷Cs⁺ adsorption as depicted in Figure 3.

Considering that the ¹³⁷Cs⁺ adsorption depth is only about 1–2 nm, hollow PB nanoparticles may offer many more advantages. They not only have a very active surface area due to their large surface to volume ratio but their hollow interior is also capable of capturing and storing ¹³⁷Cs⁺ [35]. A surfactant polyvinylpyrrolidone (PVP) was used to stabilize the nanoparticle and increase their dispersion in aqueous solutions. Figure 4 compares the efficiency of solid and hollow PB cubes of ~200 nm, in the capture of ¹³⁷Cs⁺. The elemental mapping of Figure 4B depicts a higher concentration of captured ¹³⁷Cs⁺ ions within the hollow structures than the filled ones in Figure 4A. Nevertheless, ¹³⁷Cs⁺ diffusion depth greater than 2 nm would require higher activation energy at room temperature. Other methods are required to determine the exact diffusion depth in such structures, as these results are mainly qualitative. Besides, it is well known that the use of a surfactant shields the active sites and prevents the capture of ¹³⁷Cs⁺. In order to avoid such shielding effects, PB could be coated onto support materials instead through hydroxyl bonds that anchor the PB to the support material. Carboxylic groups also tend to immobilize the PB particles in a sturdier manner. Wi et al., used a polyvinyl support surface functionalized with acrylic acid. This allowed converting the OH groups to COOH and providing a better adhesion of the PB [36]. The PB nanoparticles were immobilized on the PVP sponge and an increase in ¹³⁷Cs⁺ uptake efficiency by five times was reported, compared to the hydroxyl bond functionalization.

3. Magnetic Extraction Using Prussian Blue

There are reports of photo-induced magnetism where an electromagnetic radiation induces a residual magnetization even after the excitation is turned off, [37] due to low and high spin combinations of the transition metals in PB analogs. On their own PB and its analogs exhibit ferromagnetism at a Curie temperature of 11 K with a saturation magnetization of 3.4 emu/g as obtained by Tokoro et al., for Mn-Rb-Fe PB analogs [38]. The presence of Mn in the structure creates a Jahn-Teller distortion [39] by changing the M–CN–M bond angle and deviating it from 180°, whereupon inducing ferromagnetism. Among the various PB analogs, Mn based ones have shown the highest saturation magnetization [40]. One method of decreasing the Curie temperature of PB is by synthesizing nanoparticles of PB. Uemura et al., have demonstrated a decrease in Tc from 5.5 K in bulk PB to 4 K for PB nanoparticles protected by PVP [41]. Nevertheless, finding practical applications involving magnetic extraction would require having a Tc at around room temperature. Also, humidity increases the Curie temperature for Co-Cr PB analogs, [38] thus making it difficult for their direct application in aqueous media. This implies that most methods using PB for ¹³⁷Cs⁺ extraction, do not prescribe any efficient approach to recover the exhausted adsorbent.

Literature on nanostructured PB alone is very scarce as they are generally combined with magnetic nanomaterials like superparamagnetic iron oxide nanoparticles (SPIONs) i.e., Fe₃O₄ or γ-Fe₂O₃ nanoparticles. In the past, there have been reviews briefly describing ¹³⁷Cs⁺ adsorption employing magnetic PB-Fe₃O₄ nanoparticles [16]. However, this review goes further as it not only describes more recent developments in the latter but also discusses the development of the magnetic nanocomposite in detail from a nanoscale point of view. In nanostructures, physical properties such as magnetic moment as well as adsorption vary as a function of the nanoparticle size and further depend upon the surfactants used to stabilize them during synthesis. In the paragraphs that follow, the efficiency of magnetic PB nanoparticles and their combination with carbon allotropes are assessed. To the best of our knowledge, in the literature such nanostructures have not been evaluated in detail.

Core-shell structures with the magnetite constituting the core and the PB active layer constituting the shell have been employed. Jang et al., have reported that the poly(diallyldimethylammoniumchloride) (PDDA)@Iron oxide nanoparticles can act as nucleation sites for the precipitated PB, resulting in the coating of a negatively charged PB on the PDDA@Iron oxide nanoparticle surface [42]. Furthermore, they have also studied the magnetic properties of Fe₃O₄ and have observed a decrease in the saturation magnetization from 56 emu/g for pure Fe₃O₄ to 12 emu/g for the PB-Fe₃O₄ nanocomposite. The reduction was mostly due to the shielding of the superparamagnetism of Fe₃O₄ by the PB capping. Nevertheless, successful magnetic extraction was achieved with the nanocomposite [43]. PB analog compounds tend to show degradation in various applications after successive cycles of reuse. Chang et al., Figure 5A, have synthesized Fe₃O₄ (shell)-PB (core) nanocomposites with sizes between 20–40 nm; [44] two different concentrations of FeCl₃ were used during the synthesis. The higher Fe concentrations produced Fe₃O₄ cores with a higher magnetic saturation moment (Figure 5B), which conversely had an adverse effect on the ¹³⁷Cs⁺ adsorption due to the higher shielding of the active sites in the PB core as depicted in Figure 5C.

Some researchers have also used different ferrimagnetic nanoparticles like CoFe₂O₄ combined with PB for the extraction of cesium and have indicated very high efficiencies compared to pure CoFe₂O₄, which has some interesting ¹³⁷Cs⁺ adsorption capacity by itself [45]. CoFe₂O₄ possesses the spinel structure and tends to be physically more robust but is a hard magnetic compound compared to Fe₃O₄. The latter is a soft magnet with a very small residual magnetization.

4. Nanostructured Carbon Allotropes in ¹³⁷Cs⁺ Capture

Carbon based materials have a large surface area with good ion-exchange capabilities. These are further enhanced by the presence of functional groups such as hydroxyl or carboxyl groups on their surfaces, which can trap heavy metals from aqueous solutions. Carbon has various allotropes such as fullerenes, graphite, and diamond. The 2D carbon allotropes include carbon nanotubes and graphene [46]. The latter is the building block for graphite, which consists of stacked graphene sheets. On the other hand, a carbon nanotube (CNT) is a rolled-up graphene sheet and belongs to the fullerene or C60 family. The number of walls on a CNT is determined by the number of times the graphene sheet is rolled-up. The defects in graphene include folds and wrinkles operating as active sites towards their decoration. In a CNT, the edges, bends and breaks in the walls promote the attachment of functional groups that are subsequently linked to other organic or inorganic nanoparticles. Therefore, CNT and graphene are attractive nanoscale structures with large surface areas displaying high ion adsorption capacities and constituting versatile catalytic supports with tailored properties.

There are a few works based on activated carbon or carbonaceous materials in the extraction of low level ¹³⁷Cs⁺ contamination [47]. Mesoporous carbon based magnetic nanoparticles for ¹³⁷Cs⁺ sequestration are also being studied [48]. However, activated carbon is more efficient in extracting organic species than metal ions, [49] with an estimated adsorption efficiency between 0–10% for inorganic contaminants [50]. Activated carbon and PB analogs have shown moderate efficiency in ¹³⁷Cs⁺ extraction with a Q_max = 63 mg/g [51]. PB acts as the active material while as activated carbon is the support material for the PB analog. On the other hand, nanoscale carbon-based materials are gaining importance as they can extract heavy metal ions as well as radioactive cesium ions on their own. PB and their analogs also show a higher efficiency and a greater adsorption capacity towards ¹³⁷Cs⁺ when combined with carbon-based nanomaterials.

Table 1. Various Graphene/carbon nanotubes (CNT)/PB/Fe₃O₄ based nanocomposites arranged in decreasing order of their Q_max. The table provides the ¹³⁷Cs⁺removal efficiency, K_d, intitial concentration, temperature of extraction, isotherm and kinetic models used.

Nanocomposite	Removal Efficiency %	Kd L/g	pH	Initial Concentration mg/L	Temperature °C	Isotherm	Kinetic Model	Qmax mg/g
Pristine graphene [52]	41	0.115	1–12	10–10000	27	Langmuir	NA	465
Fe3O4-GO-pectin [56]	NA	50	1–11	9–822	30	Langmuir Freundlich Tempkin	PFO PSO	432
MWCNT-CuHCF [77]	95	568	7	1	28	Langmuir Fruendlich	NA	310
SWCNT-DMAD [70]	NA	NA	NA	NA	25	NA	NA	250
SWCNT-CuHCF [74]	NA	NA	NA	NA	25	NA	NA	230
Chitosan-PB-CNT [73]	NA	42.5	1–9	200	20	Langmuir Freundlich Redlich Peterson	PFO PSO Boyd Webbe Morris Elovich	219
GO-PANI [53]	100	94.5	2–10	1–100	25	Langmuir Freundlich	NA	185
Bentonite [69]	89	34.2	3–10	10	25	Langmuir Freundlich	NA	182.7
MWCNT-Cu ferrocyanide [75]	NA	NA	1–12	0.027–55	25	Langmuir	NA	151
MWCNT-Na-CoHCF-alginate [72]	53	23	2–10	200	25	Langmuir Freundlich	PFO PSO	133.33
MWCNT-Amino functionalized [71]	95	7–13	1–9	4.5 × 10−6–2.28 × 10−4	15–35	Langmuir Freundlich Tempkin	PFO PSO	117–136
PB-GO-Carbon fiber [78]	85.48	159.8	2–8	5–80	25	Langmuir Freundlich	PFO PSO	81.25
MWCNT [75]	NA	NA	1–12	0.027–55	25	Langmuir	NA	46
Chitosan-MWCNT [68]	25–70	26.7	3–10	1–42	25	Langmuir Freundlich	NA	45
Fe3O4-PB-GO- alginate beads [82]	80	48.7	7	25–150	0–30	Langmuir Freundlich	PSO	43.5
MWCNT-betonite [69]	80	23	3–10	10	25	Langmuir Freundlich	NA	159.4
Fe3O4-PB-hydrogels [81]	99.5	0.4	4–10	100–500	25	Langmuir	PSO	41.5
GO [54]	55	481	1–13	10	23	Langmuir Freundlich	PFO PSO	40
MWCNT [69]	35	27.8	3–10	10	25	Langmuir Freundlich	NA	45.6
Raw MWCNT [68]	12–55	55.6	3–10	1–42	25	Langmuir Freundlich	NA	29
RGO-PB [57]	99.5	6.455	NA	0.2	NA	Langmuir Freundlich	NA	18.67
MWCNT [67]	45	35	1–10	5–20	25	Langmuir Freundlich	NA	12.75
Pristine MWCNT [67]	NA	14	1–10	5–20	25	Langmuir Freundlich	NA	1.63

NA-not available.

lists the various nanocomposites of CNT/graphene used in ¹³⁷Cs⁺ extraction from aqueous solution. All of these nanocomposites are discussed in the paragraphs that follow.

4.1. Graphene Based Nanocomposites for ¹³⁷Cs⁺ Extraction

There is not much work on pristine graphene in the adsorption of ¹³⁷Cs⁺ from aqueous solutions. Only one paper reporting a very high efficiency of ¹³⁷Cs⁺ extraction using pristine graphene has been published [52]. Graphene oxide and graphene functionalized with polyaniline (PANI) have also been used in radionuclide extraction, including ¹³⁷Cs⁺ from aqueous solutions with the former showing a lower Q_max than the latter [53,54]. Combining graphene with nano PB further increases the efficiency as described below. Other combinations of graphene and PB with other bioabsorbants are also available in the literature viz. chitosan and biopolymers such as pectins [55,56]. For practical application, reclaiming the exhausted nanomaterials is an important aspect to consider. In addition, once the nano PB sorbent is exhausted it takes a very long time for a small quantity of it to precipitate, as PB has a tendency to disperse like a colloid in aqueous solutions. Combining them with graphene nanosheets cannot really solve the problem of recovering the exhausted materials either, as they too tend to form a colloidal suspension in aqueous solutions depending upon the pH. One way of overcoming the problem of reclaiming these PB nanoparticles is to pack them in a graphene oxide foam (Figure 6) that could be loaded directly into the filtration column [57]. This has already been applied to organic dyes such as methylene blue where wood was impregnated with graphene, intended for a continuous flow system. The sorbent showed potential of being regenerated and was reused multiple times [58].

As mentioned earlier, PB nanoparticles have a tendency to agglomerate because of the magnetic moments generated by the various oxidation states of the transition metals [40]. In general, nanosize materials tend to agglomerate in order to reduce their surface energy [59]. The superparamagnetic properties of Fe₃O₄ allow easy recovery of the PB from the aqueous solution when combined together to form Fe₃O₄-PB nanocomposites. Nevertheless, the ¹³⁷Cs⁺ uptake efficiency of such nanocomposites tends to decrease with a higher magnetization of Fe₃O₄. This is mainly because when the magnetization increases, magnetic nanoparticles tend to show higher agglomeration, which conversely reduces their active surface area. To this end, PVP surfactant was employed, which increases the mono-dispersion of the nanomaterials and provides larger surface areas for adsorption. However, surfactants tend to reduce the adsorption efficiency as seen previously. In this regard, graphene based nanocomposites are therefore an alternative to the above-mentioned shortcomings. Firstly, Fe₃O₄ nanoparticles are immobilized on the graphene surface in a dispersed manner, whereby reducing their agglomeration. Secondly, these nanoparticles serve as spacers between the graphene sheets discouraging the graphene sheets to stick together due to Van der Waals interactions [60]. Thirdly, PB is then directly decorated onto the graphene without being linked to the Fe₃O₄ nanoparticles. Here again, PB hosts the active adsorption sites whereas Fe₃O₄ aids in magnetically extracting the exhausted nanocomposites. Graphene mainly serves as a support even though its synergistic interaction with PB cannot be neglected. Yang et al., obtained the nanocomposite by wet impregnation of graphene with FeCl₃ followed by K₄[Fe(CN)₆] with constant magnetic stirring for 1 h at room temperature. The precipitates were subsequently recovered with a magnet by them [61].

4.2. Carbon Nanotubes for ¹³⁷Cs⁺ Extraction

Similarities in adsorption exist between graphene and CNT despite their opposite extreme aspect ratios. Like in the case of graphene, CNT also promises exciting ¹³⁷Cs⁺ extraction properties. Both have the capacity to be used either as colloids that can be magnetically extracted or as membranes which allow reclaiming them easily once exhausted [62]. The mechanism of adsorption is similar to graphene and relies on electrostatic, hydrogen and π-π interactions between the carbon allotrope and the contaminant [63]. Moreover, CNT have an additional advantage compared to graphene, which is the capacity to create mesopores when bundled together. Therefore, it presents several more sites for the adsorption of ions: within the walls of the CNT, between CNT or within the mesopores, on the walls of the CNT and in peripheral grooves [64].

In addition, nitric acid treated CNT have demonstrated efficacy in the extraction of heavy metals ions such as Cu²⁺, Mn²⁺, Co²⁺, Zn²⁺, Pb²⁺ owing to amine groups on the surface. This has motivated researchers to employ CNT for the extraction of other radioactive ions from aqueous solutions viz., U⁴⁺, U⁶⁺, Th⁴⁺ and radioactive Co²⁺ and Cu²⁺ ions [65]. Many groups have therefore carried out treatment or functionalization of CNT in order to make them more reactive to ¹³⁷Cs⁺ species. Several studies have shown that ¹³⁷Cs⁺ is attracted to OH groups [66]. CNT easily harbor different groups such as amine, hydroxyl, lactone, phenol or carboxyl groups on their surfaces post-functionalization [67]. They therefore are ideal candidates for the adsorption of ¹³⁷Cs⁺. Yang et al., have therefore prepared chitosan grafted CNT owing to the abundance of OH groups present in the former [68]. Their study demonstrated that the pH of the aqueous solution had a very important role to play on the ¹³⁷Cs⁺ adsorption by the OH groups of chitosan. They also manifested a higher adsorption than the bare CNT. Another study determined that increasing the amount of hydroxyl groups does not necessarily bring about an increase in ¹³⁷Cs⁺ adsorption. To this end, Yang et al., combined CNT with bentonite which are both known ¹³⁷Cs⁺ adsorbents linked to each other via chitosan [69]. They demonstrated that cation exchange is more effective than hydroxyl capture. More importantly, they established that hydroxyl capture depends upon the host matrix and increasing the number of OH groups will not increase the ¹³⁷Cs⁺ adsorption. This mainly suggests that the OH groups should be linked directly to the CNT matrix and are therefore limited by the number of active sites on the CNT walls. In this regard, single-walled carbon nanotube (SWCNT) could be more effective [70] than multi-walled carbon nanotube (MWCNT) [71], considering that the former possess a higher specific surface. In any case, the type of functional group also plays a role in ¹³⁷Cs⁺ selectivity.

Other groups have studied the effects of Prussian blue and their analogs on the adsorption efficiency of ¹³⁷Cs⁺ when combined with CNT. These tend to have a catalytic effect on the PB analogs by increasing the adsorption efficiency of the latter [72]. In one study by Li et al., Cu-Co-Ni PB analogs were combined with CNT [73]. Chitosan was wrapped around the CNT to increase their dispersion and also to link them to the PB analogs. Two studies have combined CNT with Cu cyanoferrates linked via different amine groups i.e., propargylamine (Figure 7) and dietheleneamine on SWCNT [74] and MWCNT [75], respectively. The Q_max increased from 150 mg/g (MWCNT) to 239 mg/g (SWCNT). These differences could be attributed to the use of different types of nanotubes and to variety of the amine ligands. In any case, the SWCNT study demonstrated that only 30% of ¹³⁷Cs⁺ ions were adsorbed by the PB nanoparticles. The rest were adsorbed onto other active sites of the functionalized SWCNT. Tsuruoka et al., have studied ferrocyanides embedded into diatomite nanoparticles encaged in both MWCNT and SWCNT which were impregnated in a polyurethane spongiform [76]. They have observed a synergistic effect owing to the porous structures created by both types of CNT. Zheng et al., have also demonstrated that it was possible to electrochemically clean the nanocomposite of ¹³⁷Cs⁺ ions and reuse them [77]. Water-soluble sodium CoHCF encapsulated in alginate beads reinforced with highly dispersed CNT was employed in the extraction of ¹³⁷Cs⁺. The beads were stable in a broad pH range of 4–10 and revealed potential for large scale applications [72]. A combination of graphene, carbon fiber and PB has also been used to extract ¹³⁷Cs⁺ directly from a lake in China [78]. Such a combination therefore allows an increased specific surface area with prospective scalable applications. A high efficacy of the graphene-CNT nanocomposites has been demonstrated for the extraction of aromatic compounds and Cu⁺² heavy metal ions [79]. These hybrid carbon allotropes exhibit 25% higher adsorption capacities in both cases than their individual counterparts. They have also shown high desalination capacities along with adsorption of heavy metal ions [80]. This suggests that alkali metal ions would tend to have a higher affinity towards the hybrid carbon allotrope compared to PB, making ¹³⁷Cs⁺ extraction by PB more effective in a high salinity aqueous medium such as seawater. Combining them with Fe₃O₄ would then be a suitable method of reclaiming the exhausted sorbents. Figure 8 provides the schematic of the ¹³⁷Cs⁺ extraction mechanism of graphene/CNT/magnetic PB nanocomposites.

5. Efficiency of Various Carbon Based Materials in ¹³⁷Cs⁺ Capture

Various batch tests performed on the different sorbents have shown increased efficiency of the graphene or CNT based sorbents (Table 1). Adsorption kinetics and equilibrium studies provide the efficacy of an adsorbent in a given adsorption system and are necessary to understand the underlying adsorption mechanisms. Most of the equilibrium adsorption studies that are provided in Table 1, employ the Langmuir and Freundlich adsorption models [67,77]. In the Langmuir model, the adsorption of ¹³⁷Cs⁺ from an aqueous solution onto a graphene-based nanocomposite is mainly used to obtain Q_max values. The model would hold for graphene-based nanocomposites as only a single monolayer of adsorption is considered, as defined by the active sites on the graphene/CNT surfaces. In the Langmuir model, all sites are considered equally favorable. This is contrary to the Freundlich model, which considers that the active sites possess varying adsorption energies and such is the case of graphene-PB based nanocomposites. In the Tempkin model it is assumed that the adsorption energy reduces as and when the sites become saturated [71,77]. The Brunauer-Emmet-Teller model assumes multilayer adsorption by considering the Langmuir model for each adsorbed layer [75]. Figure 9 provides examples of some of the isotherm models used in CNT/graphene-PB nanocomposites surveyed in this work. However, graphene and CNT based PB nanocomposites can be regarded as monolayer adsorption considering that the active sites lie on the surface of CNT and graphene, implying that all three models, Langmuir, Freundlich and Tempkin are appropriate. These equilibrium models are seconded by kinetic models such as, the pseudo first order (PFO) and second order kinetic models (PSO), which define the adsorption rate limiting factors [70,78,81,82]. The Elovich kinetic model has also been applied to chitosan-CNT-PB composites in consideration of their elemental heterogeneity, as the solid surfaces under study are energetically heterogeneous also [74].

Table 1 below therefore provides Q_max values obtained from batch tests of several different CNT, graphene, Fe₃O₄ and PB combinations. Q_max gives a direct indication of the available active sites or the maximum monolayer capacity for ¹³⁷Cs⁺ adsorption. Moreover, the Langmuir equation is valid over a wide concentration range. The table indicates that PB analogs along with CNT statistically provide the highest Q_max values reaching 310 mg/g. Pristine CNT or graphene in general do not show much efficiency compared to their hybrid counterparts. However, when they are decorated with amine groups their Q_max reaches 240 mg/g. This value is higher than CNT attached with hydroxyl groups of chitosan. However, chitosan when used as a linker between CNT and PB has a Q_max of 219 mg/g. Nevertheless, the value of the Q_max depends on various factors such as pH, temperature of the solution, presence of competing ions and the initial concentration of ¹³⁷Cs⁺. The particle size of the sorbent is also an important consideration but since we are only considering nanosize sorbents in this review, we can neglect this parameter in our discussions. Various studies on graphene or CNT based materials in this study show that the temperature of the solvent is instrumental in increasing the Q_max of the sorbent [54,71,78]. This is mainly because the process becomes endothermic with temperature increase and ¹³⁷Cs⁺ in the solution is therefore more mobile due to which they attain active sites more efficiently on the sorbent. This further implies that ¹³⁷Cs⁺ capture efficiency and can increase if the number of active sites obtained by various treatment or functionalization on the carbon-based materials are increased to their maximum capacity. The adsorption of ¹³⁷Cs⁺ is also highly dependent on the initial pH of the solution, which has mostly to do with the functional groups of the carbon-based materials becoming protonated at low pH [67,72]. Since an OH group is necessary for the capture of ¹³⁷Cs⁺ as discussed above a reduction in the former due to protonation would decrease the Q_max of the sorbent. At higher pH, a deprotonation of the functional group and an increase in OH^- ions in the solution makes the ¹³⁷Cs⁺ capture process more efficient.

The presence of other ions or salts such as Mg, Na, Li and K also affects the ¹³⁷Cs⁺ adsorption due to the inherent competition between them. They all manifest similar affinities to the active sites on the carbon-PB based nanocomposites. In the case of PB or its analog, a higher affinity to K⁺ and Na⁺ has been observed with a given order of competition: K⁺ > Ca²⁺ > Mg²⁺ > Na⁺. When combined with MWCNT, the Q_max value decreased only slightly but showed a higher selectivity towards ¹³⁷Cs⁺ [75]. Lin et al., have used CNT-PANI-NiHCF composites for electrically switching ion exchange studies and have demonstrated a higher affinity towards ¹³⁷Cs⁺ than Na⁺ ions [83]. Such nanocomposites are also employed for the extraction of Sr²⁺ at the same time. However, they appear to have a greater affinity towards ¹³⁷Cs⁺ than ⁸¹Sr²⁺, which may have to do with the difference of oxidation states of the two cations [73]. The contact time between the sorbent and the sorbate are also necessary parameters towards understanding the adsorption of ¹³⁷Cs⁺. Optimal adsorption times are required in order to evaluate the feasible scalability of the developed methods. This means that the contact time should be calculated in such a way that optimum adsorption should be obtained with minimum leaching of the adsorbed ¹³⁷Cs⁺ as well as the transition metals and alkali metals constituting the PB framework. Both adsorption and equilibrium times should be short for large-scale applications. The rate of adsorption depends on the temperature, number of adsorption sites, and diffusion of the ¹³⁷Cs⁺ within the porous matrix, which in turn depends on the density of the packed material. In the case of MWCNT, a contact time of 80 min was required to attain equilibrium [67]. A study with graphene oxide required a long contact time of 24 h [54] similarly to stirred pectin stabilized magnetic PB-GO nanocomposites [56]. For PB-graphene-carbon fiber composite the equilibrium was reached after 8 h [78].

6. Toxicity of Nanomaterials

Toxicity of nanomaterials is a key issue that is under debate [84]. If used in water remediation systems, their release and disposal need to be properly planned and evaluated. Nanomaterials are small enough to pass the skin barrier and enter the blood stream. In addition, they can be inhaled causing lung damage or ingested causing kidney damage. In the case of heavy metals, their ions can be fatal and can engender life-threatening maladies. Both CNT and graphene have been studied to evaluate their toxicity and environmental impact. Studies on zebra fish have shown that both their growth and cardiac rates are affected by these 2D nanocarbon allotropes [85]. The toxicity of CNT is an important enough issue to incite the development of related counter small-molecule-drugs [86]. CNT present risks during their entire lifecycle and mainly during occupational exposure. This implies that their manipulation needs to be regulated. CNT could also interact with biomolecules in the water and produce toxic effects [87]. The stability of PB needs to be thoroughly evaluated in various aqueous conditions. Studies have demonstrated the possibility of leaching of cyanide into ground water on decomposition of the ferrocyanides [88]. This also implies that ¹³⁷Cs⁺ could leach out and contaminate ground water. One method to stabilize PB analogs viz., titanium ferrocyanide and curb leaching of cyanide and ¹³⁷Cs⁺ includes transforming the spent material into lithium titanate, which consequently immobilizes the ¹³⁷Cs [89]. In some studies, the release of iron from PB in Fenton type reactions were studied by Doumic et al. [90]. They studied the catalytic effect in a pH = 3 environment and have observed that insoluble PB nanocomposites were more stable and showed negligible Fe release (10% after 13 cycles). Kim et al., also have observed negligible leaching of Fe during the ¹³⁷Cs⁺ adsorption process in PB-cellulose hydrogel composites [91]. Yang et al., have also carried out a systematic study of the possible leaching of Fe from PB-Fe₃O₄-graphene nanocomposites in a wide pH range from 4 to 10 [61]. Even in the highest ionic strength seawater, the amount of Fe leaching ranged from 0.95% to 0.61% while in natural water, Fe leaching as low as 0.0026% was observed. In another study with PB embedded magnetic hydrogels, leaching of Fe was studied over a period of two weeks and only a 0.3% increase in Fe was observed in the solution [81]. Graphene is also considered a chemically stable material. Nevertheless, changes in pH affect the protonation of the carbonyl or hydroxyl groups. At high pH graphene dissolves like a salt while at low pH, it tends to form aggregates and at neutral pH it stays suspended in the solution [92]. This also implies that the stability of graphene oxide depends upon the functional groups on its surface and their particle sizes [93]. In such a scenario, the dose related toxicity could vary. Overall, the formation of colloids should be absolutely avoided. However, the effect of graphene on the human body is still unknown; on the other hand, graphene oxide has been known to accumulate in the lungs of mice when inhaled, but no pathological outcome was further reported. Nano-graphite and Fe₃O₄ were also tested for Fe leaching. A very low pH < 3 showed an exponential increase of Fe release. Fe₃O₄ are linked to graphene via electrostatic forces or via functional groups on their surfaces similar to CNT therefore, the risk of nanoparticles being loosely bonded does exist. Nevertheless, in general, many studies have shown that linking Fe₃O₄ to graphene and CNT exhibit exceptionally high stability with very little leaching of Fe [94]. In any case, the exhausted nanoadsorbent has to be removed from the water and presently; magnetic extraction appears to be the most adapted solution.

7. Conclusions and Perspectives

7.1. The Road so Far for ¹³⁷Cs⁺ and the Prospects of Graphene/CNT-PB Based Nanocomposites Towards Its Extraction

Research on ¹³⁷Cs⁺ contamination has taken giant leaps since the Chernobyl catastrophe. Cesium is known to damage soft tissue, bones and provoke bone cancer. Today, eight years after the Fukushima Daiichi disaster, cesium still seems to be prevalent in soil and water; a large-scale cleanup at both fronts is still being pursued. Radioactive ¹³⁷Cs⁺ has also been detected in tap waters of Tokyo. Salty ground water containing ¹³⁷Cs⁺ is one artery for land contamination to reach the sea. Even though there is some on-site storage capacity at Fukushima Daiichi of the contaminated water, a large amount of the least contaminated water has already been rejected into the sea. There has been some progress in the cleansing of waters, however the problem is still at large. Tokyo Electric Power Company (TEPCO) forecasts the decontamination procedure to continue for the next 30–40 years. Therefore, new materials and methods towards the capture and sequestration of ¹³⁷Cs⁺ are being developed at an accelerated pace in order to quickly reply to the concern.

Adsorption and ion exchange have been found to be efficient methods in the capture of ¹³⁷Cs⁺. The contamination in Fukushima has been cleaned by the Kurion process, which consisted of effective capture of the ions (⁸¹Sr²⁺ and ^134,137Cs⁺) via an extremely porous aluminosilicate based zeolite [95]. Areva, the French nuclear company then further treated the water with a nickel ferrocyanide (NiHFC) and sand-polymer mixtures [96], which reduced the contamination by at least more than 1000 times. Advanced liquid processing system (ALPS) technology uses different combinations of resins, zeolites, titanates, activated carbons and hexacyanoferrates with a selectivity towards 62 radionuclides but is only applicable after ¹³⁷Cs, ¹³⁴Cs and ⁹⁰Sr quantities are reduced by the Kurion and Areva methods [97]. Therefore, the Prussian blue analog or hexacyanoferrates have already shown efficiency at a large scale and a total of 482,000 m³ of water has been decontaminated so far. The major drawback of the methods was the similarity between Na⁺ and ^134,137Cs⁺ affinities thus, making the whole process less efficient. The storage of sludge obtained via such processes was initially planned in large underground tanks on the Fukushima-Daiichi premises with no radioactive leaching detected. The waste was then immobilized by cementation treatment [98]. The major disadvantage of cementation was the increase in volume of the waste to be stored in a nuclear repository along with the risk of the cement cracking. To this end, vitrification has been proposed as a method that not only immobilizes ¹³⁷Cs more securely without the risk of breaking or cracking but also compresses the quantity of stored radioactive waste.

Graphene based nanomaterials seem to have a catalytic effect on PB as the Na⁺ and K⁺ selectivity decreases compared to that of ¹³⁷Cs⁺. This selectivity towards ¹³⁷Cs⁺ can be further enhanced on electrical switching as in the case of PB with CNT. In addition, ⁸¹Sr²⁺ and ¹³⁴Cs⁺ ions are also extracted simultaneously with ¹³⁷Cs⁺. Nevertheless, PB does show a higher affinity to ^134,137Cs⁺ than ⁸¹Sr²⁺, which can be further enhanced by using high specific surface materials such as graphene, CNT or hollow nano PB and their analogs, which exhibit higher Q_max values than their bulk counterparts. These nanopowders can be dispersed into a water body to be decontaminated and recovered via magnetic extraction on attaching Fe₃O₄ nanoparticles to the graphene surfaces. The drawback of Fe₃O₄ is that the saturation magnetization is rather low for large-scale applications as they usually contain surfactants. This would imply a risk of leaving behind some nanomaterial uncaught by the applied magnetic field. Another scenario could consist of making membranes of functionalized graphene and CNT with PB, which could then be easily recovered once exhausted. The main drawback of using PB is an increase of alkali metal content in the water due to ion exchange. This would not have detrimental effects on seawater; however, increased salinity in ground water could render it unsafe to drink.

Graphene and CNT are resistant in harsh environments and further offer recycling capabilities. Moreover, carbon is the fourth most abundant element in the Earth’s crust implying continuous availability for large-scale applications. However, no publications reporting large-scale use of graphene or CNT in water filtration are available. This suggests that new technologies and methods employing them are potential candidates. Nevertheless, large-scale applications require large cost-effective production and functionalization capacities. Moreover, a choice between fixed bed and batch adsorption has to be made when employing graphene. In certain cases, batch methods appear to be more efficient for contaminant adsorption than fixed bed methods. Such a choice would allow obtaining maximal adsorption with minimal amounts of absorbent, thereby reducing the quantities of radioactive wastes to be treated [99]. Further studies on the toxicity of these carbon-based materials are also required. The most important parameter is the contact time of the adsorbent with the adsorbate, which should be as low as possible for large-scale applications especially considering the thousands of cubic meter waiting to be remediated. Regeneration of graphene and CNT has been studied using acids and bases, making the recovery of the metallic contaminants feasible. However, from an industrial point of view, there is much work that needs to be conducted in this very promising field.

7.2. Development of Other Nanocomposites

Different materials need to be combined in order to increase the ¹³⁷Cs⁺ selectivity and hence its extraction efficiency, as already proven by the Kurion-Areva-ALPS combinations. With regards to other potential materials, PB has also been combined with zeolites for the adsorption of ¹³⁷Cs⁺ ions [100,101]. Zeolites were seen as promising materials [101] for ¹³⁷Cs⁺ adsorption [102,103] and were subsequently combined with fibrous polymer adsorbents to enhance their ¹³⁷Cs⁺ extraction capacity. These zeolite-polymer composite fibers were initially studied for heavy metal ion removal such as Pb²⁺, Cu²⁺, Ni²⁺, and Cd²⁺ [104,105]. One advantage is that the surface to volume ratio could be tuned through the zeolite fibrous polymer ratio with a specific surface value of 145 m²/g. The study suggested that with such zeolite polymer composite fibers, the Pb²⁺ extraction takes place by both ion exchange and inclusion mechanisms. Zeolite and poly(ethersulfone) were then combined for ¹³⁷Cs⁺ extraction with 30 wt% of zeolite in porous fibrous polymer [15]. These composites were applied to the decontamination of radioactive ¹³⁷Cs⁺ in the city of Fukushima for a period of 28 days. A total of 7700 Bq/kg of radioactive Cs was extracted during this period, while with the zeolite alone only 33 Bq/kg was extracted [7]. The main advantage of this composite is that water can flow unobstructed through the composite compared to the zeolite powders. Fiber based porous structures seem to be more adapted for large flow extraction than magnetic extraction that needs a specific set-up. The latter nevertheless has the advantage of limiting the quantity of nuclear waste produced during the radioactive ¹³⁷Cs⁺ extraction process. Functionalized graphene and CNT could very well be included in such porous structures without clogging the pores and increasing the ¹³⁷Cs⁺ selectivity during extraction.