A Highly Efficient Adsorbent Cu-Perusian Blue@Nanodiamond for Cesium in Diluted Artificial Seawater and Soil-Treated Wastewater

A new adsorbent Cu-Perussian blue@Nanodiamond (Cu-PB@DND) for Cs+ removal was prepared and characterized with IR, SEM, X-ray diffraction, particle size analysis, and zeta-potential. The adsorbent consists of a core of aggregated detonation nanodiamond (DND) particles with the surface treated with Cu-PB. Cesium adsorption was studied in two modes; a co-precipitation mode and a batch mode. In the co-precipitation mode, DND, CuCl2, and K4[Fe(CN)6] were added sequentially to a Cs+ solution in diluted artificial seawater. In the batch mode, adsorbent Cu-PB@DND was dispersed into a Cs+ solution with stirring. The distribution coefficient (Kd) of the co-precipitation mode was 8.8 × 107 (mL/g) at Cs+ 6.6 ppm in 0.07% seawater. The Kd value of the batch mode was 1.3 × 106 (mL/g). Precipitation of Cs+-incorporated particles was complete, and post filtration was not necessary. Excess copper and iron ions were completely removed and were not detected in the supernatant. The adsorption data for Cu-PB@DND were analyzed by assuming Langmuir isotherm and a good fit was obtained with a maximum adsorption capacity Qmax of 759 mg/g. The co-precipitation method was also applied to soil-treated wastewater.


Results and Discussion
Preparation and property of Cu-PB@DND. Schematic drawings of the structure of a DND particle and preparation of Cu-PB@DND are shown in Fig. 1.
For preparation of Cu-PB@DND, CuCl 2 solution was added to a dispersed solution of DND(II) (see the next IR section for the definition of DND(II)). At this stage, the added Cu 2+ was adsorbed on DND surface. In the next step, K 4 [Fe(CN) 6 ] solution was added to the solution. After a while, red-brown Cu-PB@DND appeared and precipitated in the solution. This adsorbent of Cu-PB@DND at this stage was fine powder, but as shown later, had larger average diameter than the original unmodified DND(II). The supernatant was removed by pipetting and the Cu-PB@DND residue was used in wet state for Cs + adsorption experiment. For spectroscopic characterization, the adsorbent was measured after drying. IR spectrum, zeta potential, SEM, particle size, and elemental analysis of DND and Cu-PB@ DND. The IR spectra of DND(I) (original DND), DND(II) (air-heated DND(I)) and DND(III) (NaOH-treated DND(I)) are shown in Fig. 2. The IR spectra of DND(I), DND(II) air-treated in 2011, DND(II) air-treated in 2016 and Cu-PB@DND are shown in Fig. 3. All of the three spectra in Fig. 2

have common absorption bands
Scientific RepoRts | (2018) 8:5807 | DOI: 10.1038/s41598-018-24129-0 at 3200-3600 cm −1 , 2900 cm −1 , 1700-1800 cm −1 , and 1600 cm −1 , but the relative intensities of the bands vary depending on the surface treatment. Such treatment-dependent IR property is already known for DNDs with different surface treatments 6,[39][40][41] . Referenced to the previous assignment in the literatures, the bands were assigned as follows: 3400 cm −1 is O-H stretching of water and phenols, 2900 cm −1 is C-H stretching, 1700-1800 cm −1 is C=O of ketones and carboxylic groups, and 1600 cm −1 is several C-O groups such as esters and lactones or water molecules in outer graphite-like layers. Surface water amount usually varies significantly depending on the drying condition and chemical surface treatment. Water molecules in DND are remarkably strongly associated   with DND, and are probably bound between the outer graphite-like layers or are associated with the interface of nanodiamond core and outer graphite-layers. Extensive depletion of the water needs vacuum and elevated temperature 42 . Therefore, IR spectra were measured at 120 °C in N 2 in the present experiment. It is obvious in Fig. 2 that sintering of DND(I) in air increases water amount probably bound to surface graphite layer, whereas NaOH treatment decreases water amount and most surface functional groups except C-H are lost. The surface of DND(III) has less functional groups and is thus more hydrophobic than DND(I) and DND(II), which suggests that DND(III) would be less favorable for metal ion adsorption than others.
Among the two air-treated DNDs in Fig. 3, significant difference is observed at the band 1900 cm −1 . The band still remains in DND(II) treated in 2016, but is almost lost by air-heating in 2011. The reason for such a significant difference is not known. The IR spectrum of Cu-PB@DND shows new bands, in addition to those of original DND(II), at 2100 cm −1 , 1600 cm −1 , and 670 cm −1 . These three bands are also reported for Cu II 3 30 and PB on graphene 37 , and are assigned to C≡N stretching (2100 cm −1 ) and Fe-CN-Fe bending (670 cm −1 ). The band at 1600 cm −1 has not been clearly assigned in literatures, but is always observed weakly or moderately in PB and Cu-PB.
Zeta potentials of DNDs in water were + 46 mV for DND(I), −14 mV for DND(II) and + 4 mV for DND(III). The X-ray diffraction pattern of Cu-PB@DND is shown in Fig. 4. All of the peaks correspond to K x Cu y [Fe(CN) 6 ] · xH 2 O or PB reported in literatures 30,38,[43][44][45][46] . Clear peak assignment is not reported, since relative peak intensities and positions vary depending on the amount of crystal water and hetero metal ions 30,37,38 . The two compounds give almost an identical X-ray diffraction pattern and are not distinguished in the literatures. Peaks of DND(II) are not observed in Fig. 4. Non-PB-modified DNDs usually give diamond peaks at 43 and 72 degree and a small graphite peak at 27 degree 4,6 . In Fig. 4, these DND peaks are not observed, maybe because only the thin surface layer of Cu-PB in Cu-PB@DND contributes to the X-ray diffracrion pattern. Similarly, only PB X-ray diffraction peaks are observed in other PB modified CNT, graphene and graphene oxide, for Cs + adsorption 28,38,44 . Existence of DND in Cu-PB@DND has been proved in the IR spectrum in Fig. 3, since longer wavelength of IR can penetrate into deeper sites of particles, and thus give spectrum peaks of both surface Cu-PB and core DND.
The SEM images of DND(II) and Cu-PB@DND are shown in Fig. 5. In Fig. 5(A and B), many aggregated spheres are observed, onto which smaller particles and primary particles of DND less than 10 nm are attached. In Fig. 5(B), the approximate aggregate size ranges from less than 100 nm to more than several μm. Since primary particles of DND in literatures are about 4-6 nm in diameter, and the particle size distribution in the initial dispersion of DND(II) is about 100 nm to 1 μm (Fig. S1), the aggregation in Fig. 5(A and B) would have been caused by drying. The EDS (energy dispersive semiconductor) analysis of DND(II) and Cu-PB@DND shows the elemental composition (%) of DND(II) as C 90.7, O 7.9, Zn 0.4, Fe 0.4, Si 0.3, Al 0.2, S 0.2 and Cu 0.1, and that of Cu-PB@DMD as C 54.9, Cu 21.7, Fe 13.2, N 8.3, O 1.9, and K 0.2. The elemental values of the same sample measured with ICP-MS using a nebuliser for dispersion solution, after HNO 3 digestion for 2 hrs, were less than 1/5 of the EDS values. This is due to the fact that for EDS, part of the water in DND is lost by vacuum, and for ICP-MS, most of the nanodiamond particles remain undecomposed in solution even after strong acid digestion, and is only partially introduced into the plasma with a conventional nebulizer, because of the high specific density of diamond particle (ca. 3.5) with heavy metal impurity. Many literatures report impurity values obtained with ICP-MS, but these must be considered with care.
The SEM images in Fig. 5(C and D) show that the average particle size is significantly increased by Cu-PB modification. The average particle size of the adsorbent dispersed in water was measured with LSD (laser scattering dynamics) (Fig. S1). Before PB modification, the average particle diameter of DND(II) was 0.2 ± 0.4 μm, whereas it is 5 ± 4μm after Cu-PB modification.
Comparison of DND(I), DND(II) and DND(III) in Cs + Removal. Removal efficiency, R eff (%) (R eff = 100 × (C 0 -C e )/C 0 , where C 0 and C e are original and equilibrium Cs + concentration in solution, respectively), was compared for DND(I), DND(II) and DND(III), by using a 10 ppm Cs + solution in diluted artificial seawater (0.07%). The Cs + adsorption and precipitation experiment was carried out by addition of DND to a Cs + solution to 0.25 μg/L. After the solution was stirred for 2 hours and left standing overnight, Cs + concentration in the supernatant was measured with ICP-AES. The removal efficiencies were as follows: unmodified DND(I) 22%, air-heated DND(II) 38%, and NaOH-treated DND(III) 13%. Therefore, subsequent experiments were carried out with DND(II). As expected, DND(III) is least effective, since it has the fewest number of polar functional groups, as observed in the IR spectrum (Fig. 2). Air-heated DND(II) having more water molecules around the diamond core, as observed in Fig. 2, has negative zeta-potential, and thus has the highest removal efficiency. Removal of Cs + by Cu-PB@DND. Cesium removal experiment was carried out in two modes: (i) co-precipitation by sequential addition of DND, CuCl 2 and potassium ferrocyanide. The latter two reagents were added to prepare PB in situ in the Cs + solution. (ii) batch treatment of Cu-PB@DND adsorbent in a Cs + solution. The results were evaluated in terms of removal efficiency, distribution coefficient of Cs + , capacity of adsorbent, and selectivity to Cs + . Detonation nanodiamond (DND) has advantage over other common non-carbon adsorbents that DND is resistant to chemicals, high temperatures up to 450 °C under atmospheric pressure in air and radioactivity.
Removal of Cs + in diluted artificial seawater with co-precipitation mode. Literatures show that not only PB but also several other PB-analogue complexes [M x Fe 1−x (CN) 6 ] y− , where M is Ni 2+ , Cu 2+ , Zn 2+ , Co 3+ , and Mn 3+ 17-32 , are effective for Cs + removal in freshwater and in some cases in seawater. These complexes having hetero metal M n+ ion are synthesised by addition of MCl 2 or 3 to a solution of potassium ferrocyanide. The removal efficiency for Cs + depends on M n+ , and also preparation method. Different preparation method gives different particle size, which affects the removal efficiency. In the present experiment, MCl 2 or 3 was first varied in co-precipitation mode, i.e., DND(II), MCl 2 or 3 , and potassium ferrocyanide were added sequentially in this order to seawater (0.07%) containing 20 ppm Cs + . After precipitate of Cs + -incorporated M-PB settled, Cs + removal efficiencies were calculated from the Cs + concentration in the supernatant. The results are as follows: Cu 2+ ( > 99%), Fe 3+ (92%), and Co 3+ (87%). Therefore, subsequent experiments were carried out with CuCl 2 . The addition order of the three reagents, DND(II), CuCl 2 and potassium ferrocyanide, critically affected the removal efficiency, and the addition order as described above gave the best removal. Addition of DND(II) after either one or both of the reagents remarkably decreased Cs + removal efficiency. The final Cs + concentrations in the supernatant of various addition combination of the reagents are listed in Table 1, which shows that all of the three reagents are necessary to achieve the highest efficiently (run 3). The effect of pH was examined at pH 4, 6.8 and 11 by addition of either NaOH or HCl after all the procedure was finished. The removal efficiency did not change significantly, and therefore the following experiments were carried out at pH 6.8 (without any pH adjustment). It is notable that when Cu-PB alone is used for Cs + removal, Cu-PB is effective only in neutral pH region 25,31 . In the present study, no pH dependence was observed both in co-precipitation and batch modes. Obviously the property of DND to adsorb Cs + as well as Cu-PB operates very much favorably to Cs + removal.
In Table 1, Cs + was removed perfectly when all of DND(II), CuCl 2 and K 4 [Fe(CN) 6 ] were added (run 3). In addition, Cu 2+ and Fe 3+ were also efficiently removed from the solution in run 3. This is in contrast to run 4, in which DND(II) was not added, and in this case Cu 2+ was only partially removed. It is also noteworthy that Fe 3+ was most efficiently removed when all of the three reagents were added (run 3), but was only moderately removed when DND(II) was not added (runs 2 and 3). It is notable that PB and Cu-PB precipitated completely and Cu 2+ and Fe 3+ were not detected in the colorless transparent supernatants, as shown in Table 1. Owing to such superb adsorption and precipitation effect of DND(II), it was not necessary to filter the supernatant to remove residual fine powder. This is a great advantage of DND(II) in real application to seawater and wastewater. The concentrations of Na + , K + , Mg 2+ and Ca 2+ ions in the final supernatants are listed in Table S1, which shows that none of these metal ions of seawater were adsorbed on Cu-PB@DND, i.e. the adsorbent is highly selective to Cs + . Such high removal efficiency and high selectivity to Cs + as well as the complete removal of Cu 2+ and Fe 3+ ions have not been reported for conventional PB and PB-analogue adsorbents, zeolites and other organic and inorganic adsorbents. The new adsorbent Cu-PB@DND has been found practically very useful, since the adsorbent retains its capability even in diluted artificial seawater and in a wide range of pH. This is obviously in contrast to previous adsorbents, whose Cs + removal efficiency decreases in diluted seawater or the Cs + removal efficiency is reported only in pure water 31,38,45,46 . Since high concentration of Na + (>175 ppm) in solution severely disturbed Cs + measurement on ICP-AES, all of the present experiments were performed in 0.07% artificial seawater, however, it is highly likely that the present method similarly works in more concentrated seawater.
All of the previous Cs + removal methods using PB or PB analogues used filtration for the final supernatant to remove Cs + -containing fine particles in the solution. In the present system, such filtration is not necessary, since all of the Cs + -adsorbed adsorbent precipitates completely, owing to the effect of DND(II). The remarkable difference of Cs + removal efficiency between run 3 and other runs in Table 1 suggests that DND(II) enhances precipitation of the adsorbent by aggregation. In literatures, several bead materials as PB supports have been reported for Cs + removal, such as silicate 26 , polymer with silicate 27 and polyvinyl alcohol beads 21 , with always filtration after adsorption. It should also be noted that the concentration of potassium ferrocyanide in Table 1 is approximately 1/100 to 1/10 of those of literature methods. Since PB contains toxic CN − , the concentration should be preferably as low as possible.
In Table 1, DND(II) itself adsorbs Cs + to some extent (run 2), but in addition to cation, DND(II) seemingly adsorbs also negatively charged PB and Cu-PB. Zeta potentials of DNDs are different depending on the different surface treatments, and the reported values range from negative to positive [3][4][5] . Zeta potential is only an average of the whole surface of DND. On the surface, a variety of functional groups having negative, neutral and possibly positive charges is actually distributed, since the functional groups have different pK a s and other cations might have neutralised the charges of the functional groups. It is no wonder that both cations and anions are seemingly adsorbed on DND, but it is also possible that Cs + adsorption on DND neutralises the surface charge of DND and facilitates subsequent adsorption of anions. Adsorption of metal ions other than Cs + onto DND in general is an interesting research target.
Mechanism of Cs + adsorption in co-precipitation mode. The mechanism of Cs + adsorption and precipitation by simple Cu-PB is explained in literatures as follows. In Cu-PB, Cu 2+ partially substitutes Fe 2+ sites in the 3-dimensional cubic lattice structure consisting of [Fe(CN) 6 ] 4− , and Cs + ion is adsorbed by being trapped into the void of the cubic lattice [Cu x Fe 1−x (CN) 6 ] 4− to form the complex with average composition of Cs 2 [Cu x Fe 1−x (CN) 6 ], which finally precipitates 18,22,26 . The partial substitution of copper ion with Fe 2+ ion in the lattice tunes the void volume of the cubic lattice to fit to the hydrated Cs + ion, thus enhancing adsorption efficiency and selectivity to Cs + .
Based on the above reported mechanism for simple Cu-PB, effect of DND in the present co-precipitation method is considered as follows. From run 2, DND itself has adsorption function for Cs + , which is probably due to metal cation coordination to surface functional groups of DND, and/or electrostatic attraction between metal cations and DND having negative zeta-potential. In contrast, single digit DND having positive zeta potential of +55 mV is reported to adsorb only negative ions, and is a very selective adsorbent to tungstate anion 10 . For our DND, both positive (Cs + and Cu 2+ ) and negative (ferrocyanide) ions seem to be adsorbed and are finally precipitated as Cs + -incorporated Cu-PB@DND. The adsorption behavior of DND can be explained that Cu 2+ adsorbed on DND reacts with ferrocyanide anion added in solution to form Cu-PB on the surface of DND. In such a reaction, electrostatic repulsion between the negative DND and ferrocyanide anion would be mitigated by Cu 2+ on the DND surface and by the formation of energetically favored Cu-PB. Cesium cation adsorbed on DND would be incorporated into Cu-PB during the formation reaction. In contrast, when independently pre-prepared Cu-PB was added to a Cs + solution already added with DND, the Cs + removal efficiency was only 55%. This fact suggests that formation of Cu-PB in the presence of Cs + -adsorbed DND is essential to high removal efficiency. From the optimized addition order of DND, CuCl 2 and K 4 [Fe(CN) 6 ] to a Cs + solution, it seems that the reaction first starts with adsorption of Cs + and then Cu 2+ to DND, which then reacts with potassium ferrocyanide added to the solution, to form Cs + -incorporated Cu-PB@DND precipitate. Such a reaction mechanism suggests that adsorbed Cs + and Cu 2+ ions react with potassium ferrocyanide in solution, as if the cations were free ions in solution. The  adsorbed metal ions and metal complexes obviously enhance aggregation and precipitation of DND, possibly by electrostatic attraction and bridging of several DND particles. The progress of the reaction is shown in Fig. 6.
The present experiment showed that DND did not adsorb appreciably either preformed [Cu x Fe 1−x (CN) 6 ] 4− or Cs 4 [Cu x Fe 1−x (CN) 6 ] in the solution, since Cs + removal efficiency was obviously decreased when DND was added after CuCl 2 and potassium ferrocyanide. Copper ion must be added after DND and prior to potassium ferrocyanide. Formation of red-brown [Cu x Fe 1−x (CN) 6 ] 4− was hindered, when potassium ferrocyanide was added after DND and prior to copper chloride. It seems that ferrocyanide ion is adsorbed on Cs + -adsorbed DND but does not react with copper ion, in the adsorbed state.
The final precipitation efficiency is obviously promoted by the presence of DND. The final supernatant after overnight standing was colorless and transparent, and the precipitate was brown.
Removal of Cs + in diluted artificial seawater by using Cu-PB@DND in batch mode. Removal efficiency of Cu-PB@DND adsorbent in batch mode was examined by using 0.015 g of the adsorbent added to 80 mL of Cs + solution in artificial seawater (0.07%). The solution was stirred for 6 hours. Clear supernatant was obtained after the solution was left standing overnight. The initial Cs + concentration was varied in the range of 1 to 150 ppm, and Cs + concentration of the supernatant was measured with ICP-MS. The results are shown in Table 2 for several different initial Cs + concentrations.

Calculation of Distribution Coefficient (K d ), Removal Efficiency (R eff ), Adsorption Capacity (Q e )
, and Adsorption Isotherm. The data of Cs + removal experiment were analyzed following the equations below. The distribution coefficient K d (mL/g) was calculated according to equation (1).
where C 0 and C e are initial and equilibrium concentrations of Cs + , and V and M are volume of solution (mL) and weight of adsorbent (g), respectively. Removal efficiency R eff (%) was calculated according to equation (2).
e 0 e Although most K d values are reported for a batch mode in literatures, K d value of the present co-precipitation method was calculated by using the data of run 3 in Table 1 and equation (1), assuming 2 mg of adsorbent (calculated from the concentrations of DND, CuCl 2 and potassium ferrocyanide and the solution volume of 80 mL used in the experiment). The K d value obtained was 8.8 × 10 7 mL/g ( Table 2). Considering that reported K d values for PB and PB analogue adsorbents in batch mode are in the range of 10 3 to 10 6 mL/g, depending on the hetero metal M in [M x Fe 1−x (CN) 6 ] y− and on the support bead material (silica, alumina, polymers etc.) 13,19,20,[25][26][27][28]31,38,[43][44][45] , the K d value of the present co-precipitation method is remarkably high, which is due to precipitation and aggregation effect of DND in co-precipitation mode. Co-precipitation effect gives often better removal efficiency than batch mode. When only DND is added to a Cs + solution in Table 1, K d is 2.4 × 10 4 mL/g, which is still a significantly high value. The K d and R eff values of both co-precipitation and batch modes are summarized in Table 2.
For evaluation of adsorption capacity and for understanding of the adsorption mechanism of Cu-ferrocyanide@ DND in batch mode, adsorption data were analyzed based on Langmuir adsorption model, and a good fit was found. From the fitting calculation, maximum adsorption capacity (Q max ) of the adsorbent was obtained. This value provides estimate of the adsorbent amount required to remove a unit mass of pollutant (Cs + ) under the system conditions. The data and the calculated curves of C e vs. Q e and C e vs. C e /Q e are shown in Fig. 7(A and B), respectively, in which Q e is equilibrium adsorption capacity (mg/g), i.e., Cs + concentration in adsorbent at equilibrium (Cs + in mg/adsorbent in g). In Fig. 7(A), Q e increases rapidly with increasing C e up to around 5 ppm, even after this point Q e increases but much gradually.
Langmuir adsorption equation is expressed in equation (4), which is derived from a model that adsorption takes place on surface monolayer.
In equation (4), K L (L/mg) is Langmuir constant related to energy of adsorption and desorption, and Q max is maximum adsorption capacity (mg/g). Equation (4) can be transformed to equation (5), which indicates that K L and Q max can be calculated from the intercept and slope of Fig. 7(B).
The calculated K L and Q max of Cu-PB@DND are listed together with the fitting residual R 2 in Table 3. Maximum adsorption capacity Q max and K d of other adsorbents in literature are also shown in Table 3 for comparison. Obviously Q max of the present adsorbent is remarkably high compared to those of simple PB, PB analogues and other common adsorbents. Recently nanocarbon materials such as CNT, graphene, graphene-oxide and reduced graphene-oxide are reported as solid support for PB and PB analogues in Cs + removal 38,47,49 . The performance of these new adsorbents are also included in Table 3. The new adsorbents have notably high Cs + adsorption capability. Another very new adsorbent CNF/PB/PVA, prepared from cellulose nanofiber (CNF) as a PB support with polyvinylalcohol (PVA) as a binder, gives high performance for Cs + adsorption both in batch and column mode 44 .    27 . In other literatures, maximum Cu 2+ adsorption on DND is reported to be 7.5 wt% 48 and 5μmol/g 49 . The latter value was obtained in the existence of SO 4 2− for DND from the same source, ALIT, Ukraine, for air-sintered sample similarly to DND(II).
With the specific surface area of 191 m 2 /g and number of acidic groups per area, 0.56 group/nm 2 reported in the literature for DND of ALIT 49 , the maximum Cu 2+ adsorption on DND can be calculated to 1.13 wt%, by assuming that all the acidic groups are occupied with Cu 2+ . This value is appreciably lower than the analytical Cu 2+ value of 21.7 wt% for Cu-PB@DND. From this fact we have an image of Cu-PB@DND that substantially thick Cu-PB layer covers multiple DND particles, and the adsorbent particle is actually something like a mixture of Cu-PB and DND rather than aggregated DND particles with thin surface layer of Cu-PB.
In the mechanistic Cs + adsorption study on Cu 2 [Fe(CN) 6 ], the maximum atomic ratio of Cs/Fe is reported to 1.5 30 . Assuming that all the Cu-PB in Cu-PB@DND participate in Cs + adsorption with the ratio of 1.5, the maximum Cs + adsorption on Cu-PB@DND was calculated to 314 mg/g.
On the other hand, maximum ion-exchange capacity of 2.0 mmol Cs + /g is reported for DND of ALIT 49 , which corresponds to 127 mg/g Cs + adsorption per Cu-PB@DND, assuming 47.8 wt% DND carbon in the adsorbent (DND C amount was estimated by subtracting cyanide C amount corresponding to N amount, from the total C amount). Addition of the two values of 314 and 127 mg/g amounts to 441 mg/g. This kind of calculation might be somewhat speculative, but explains the high Q max (mg/g) of 759 mg/g in Table 3. The remaining difference between 441 and 759 might be caused by errors in the cited literature values, experimental error of the elemental analysis, or existence of unexplained new mechanism specific to DND, such as formation of Cu-PB together with trapping Cs + in the DND hydrogel within the adsorbent. Hydrogel of DND would facilitate Cs + ion transfer deeper into the inner site of the adsorbent, thus increase the capacity. In silica gel, Cs + adsorption on PB is reported to proceed in a very different manner from usual PB adsorption, and yet Langmuir fitting is observed 27 .
Removal of Cs + in soil-treated wastewater. Wastewater (A) containing 5 w/v % fine clay powder was obtained after washing contaminated soil sample with high-temperature and high-pressure water. Then the wastewater (A) was treated with the present co-precipitation method, i.e., DND(II), CuCl 2 , and potassium ferrocyanide were added in this order to wastewater (A) with stirring. The wastewater was left standing for settling of the precipitate. In some cases, the supernatant was filtered to give a dark-brown transparent filtrate, which was subjected to 137 Cs γ-ray measurement. The results are shown in Table 4.
As shown in Table 4, most Cs + was removed by addition of only DND(II), CuCl 2 , and potassium ferrocyanide. When coagulant and a filter were applied in addition, Cs + level was completely decreased to N.D. (runs 4 and 5).
Another wastewater sample (B) contained fine soil at 3.2 wt/v %. Wastewater (B) was a turbid dispersion, and the soil powder never precipitated even after 12 hours standing. The sample was subjected directly to Cs + co-precipitation procedure. After the procedure, the supernatant was still turbid in most cases. The Cs + concentrations of the supernatants were measured with a 137 Cs γ-ray spectrometer. The results are shown in Table 5.
The effect of precipitation time of soil powder by using Cu-PB@DND was measured. In Fig. 8 are shown the photos of the soil solutions, which show how precipitation became rapid by addition of DND, CuCl 2 and potassium ferrocyanide.
In Fig. 8, sample # 9 almost completely precipitated only in 5 minutes, and the transparent supernatant was obtained, whose Cs + concentration was below detection limit (Table 5). Other solutions without DND addition in Fig. 8 are very turbid even after 720 minutes standing.
The concentrations of Sr + , Ca 2+ , Cu 2+ , Fe 3+ , K + , Mg 2+ , Na + , S and Si, in the supernatants were measured with ICP-AES, and are listed in Table S2. The table shows that Cu 2+ and Fe 2+ were not detected under any condition, whereas Ca 2+ , Mg 2+ , Na + and K + increased significantly by the treatment, which would be due to leaching from the soil powder.

Conclusions
Detonation nanodiamond (DND) is highly effective for Cs + removal when used together with Cu-PB. The co-precipitation method and the adsrobent Cu-PB@DND do not lose adsortion capability even in diluted seawater and fine soil-containing wastewater. The adsorbent Cu-PB@DND has a distinctly high K d of 1.3 × 10 6 mL/g and capacity Q max of 7.59 × 10 2 mg/g. Several other remarkable points of the present method are: (i) excess Cu 2+ and Fe 2/3+ added to the solution are not detected in the supernatant. (ii) the efficiency of Cu-PB@DND has no pH dependence, (iii) Cu-PB@DND retains its high removal efficiency even in diluted artificial seawater and soil-containing wastewater. The high capacity is explained with high specific surface area and high adsorption capacity of DND, together with well-known capacity of Cu-PB.  The powder DNDs thus prepared were dispersed to water by vigorous sonication to make 1 wt% dispersions. For comparison of Cs + removal efficiency, 1 wt% dispersions of DND(I), DND (II) and DND (III) were used in co-precipitation mode.
Reagents and samples. Water dispersion of DND(I) (1 wt%) from Vision Development Co., Ltd. Japan, was used as received. The dispersion was sonicated vigorously for a few minutes before use. Diluted artificial seawater was prepared by dissolving commercial natural seawater salt powder ("Nuchimasu" from Nuchimasu Co., Japan) into pure deionized water to 0.07 wt/v%. This concentration was the maximum concentration, at which Cs + could be measured on ICP-AES without Na + interference. To the diluted seawater, commercial standard 1000    Table 5.
ppm non-radioactive Cs + solution (Wako Pure Chemical Industries Ltd., Japan) was added to desired concentrations (pH 7.6). For measurement of Cs + , both ICP-AES and ICP-MS were used. Two soil-treated wastewaters (A) and (B) were obtained from two different spots near Fukushima prefecture. Sample (A) was a suspension of 5 wt/v% soil. Wastewater (B) was a supernatant of wet agricultural soil and had 3.2 wt/v% fine soil powder.
Instruments. The ICP-AES was SPS5510 from SII Co., Japan with a nebulizer for dispersion solution. The ICP-MS was 7700 × from Agilent. The 137 Cs γ-ray activity was measured on a Ge semiconductor detector from Ortec Co. The FT-IR spectra of DND were measured as KBr pellet on JASCO FT/IR-6100 at 120 °C under N 2 to avoid humidity for non-treated DND(I) and at ambient temperature in air for surface treated DND(II) and DND(III). Zeta-potential of DND was measured in 0.1 wt% aqueous solution with DLS (dynamic light scattering) on Zetasizer NanoZS from Malvern Instruments Ltd. The particle size distribution was measured with DLS on a UPA-EX from NIKKISO, Japan. Scanning electron microscopy (SEM) with elemental analysis (EDS) was JSM-7001F from JEOL and INCA x-act from Oxford Instruments.
Preparation of Cu-PB@DND adsorbent. To 40 mL of water was added 0.4 mL of ultrasonicated 1% DND(II) dispersion and the solution was stirred for a minute. To the solution was added 0.4 mL of 0.1 mol/L CuCl 2 solution, and the solution was stirred for 5 minutes. After stirring, 0.4 mL of 0.05 M of K 4 [Fe(CN) 6 ] was added and the solution was stirred for 1 hour. The solution was left standing overnight for precipitation. The supernatant was removed by pipetting, and the wet precipitate was used for Cs + adsorption experiment. For SEM and IR measurements, the adsorbent was dried at room temperature.
Procedure for cesium removal in diluted artificial seawater. Cesium removal in seawater was carried out in two modes; a co-precipitation mode using either one of DND(I), DND(II) or DND(III), and the other is a batch mode using Cu-PB@DND adsorbent.
The procedure of co-precipitation was as follows. To 80 mL of 0.07 wt/v % sea water containing 6.6 ppm Cs + , was added 2 mL of 1% DND(II) dispersion and the solution was stirred for a minute. The solution became gray and turbid. To the solution was added 0.2 mL of 0.02 mol/L CuCl 2 solution, and the solution was stirred for 5 minutes. After stirring, 0.2 mL of 0.01 mol/L of potassium ferrocyanide was added and the solution was stirred for 30 minutes. The solution was left standing for 30 minutes, to settle the red-brown precipitate. Colorless supernatant was taken and measured for Cs + , Cu 2+ , Fe 3+ , Na + and other metal ions with ICP-AES or ICP-MS.
For comparison, either CoCl 3 · 6H 2 O, FeCl 3 , or ZnCl 2 was added instead of CuCl 2 at the same molar concentration, and the above Cs + removal treatment was performed. Since CuCl 2 gave the best removal efficiency, all of the following experiments were carried out with CuCl 2 . The total weight of the air-dried final precipitate using CuCl 2 was 16 mg.
In the batch mode, 15 mg (dry base) of Cu-PB@DND adsorbent was added to 80 mL of 0.07% seawater containing 1 to 150 ppm of Cs + , and the solution was stirred for 6 hours. The solution was left standing overnight and the supernatant was measured for Cs + on ICP-MS.

Procedure for cesium removal in soil-treated wastewaters (A) and (B).
The co-precipitation procedure was performed to wastewaters (A) and (B) under several different conditions. In wastewater (A), either a coagulant (HimolocSS100 from HYMO CORPORATION) at 500 ppm, or a filter of φ 0.22 μm, or both were used to remove fine soil particles in the supernatants. The 137 Cs γ-ray activity of the supernatants was measured for both samples, in addition, 134 Cs was also measured for sample (B). For wastewater (B), a filter of φ1 μm was used.