Development of hollow δ -FeOOH structures for mercury removal from water

δ -FeOOH, a magnetic iron oxyhydroxide, has a signi ﬁ cant number of -OH groups on its surface. These provide an attractive platform for heavy metal species in contaminated water, giving it potential as an adsorbent. Its performance can be improved by increasing the number of active surface sites. δ -FeOOH hollow structures were synthesized on a mesoporous silica surface then treated with NaOH solution. X-ray diffraction (XRD) and transmission electron microscopy (TEM) con ﬁ rmed that structure synthesis was successful. δ -FeOOH, 5,27 nm, hollow crystals were produced with 63 m 2 g (cid:1) 1 surface area and 20 nm average pore size. The point of zero charge was 4.72, which is bene ﬁ cial for Hg(II) adsorption near neutral pH. The maximum Hg(II) adsorption capacity at pH 7 was determined as 89.1 mg g (cid:1) 1 . The kinetics data were best ﬁ tted by a pseudo-second-order model with k 2 equal 0,1151 g mg (cid:1) 1 min (cid:1) 1 . Finally, a nanomaterial ﬁ lter was developed and used to remove mercury in water samples from a Brazilian river. δ -FeOOH hollow structures were prepared on a mesoporous silica simple chemical . The δ -FeOOH hollow structures are an effective Hg(II) adsorbent. (cid:129) The adsorbent had good anti-interference ability to co-existing anions and cations. (cid:129) A ﬁ lter made from the adsorbent treated water effectively from a Brazilian river.


INTRODUCTION
Mercury (Hg) is one of the most toxic metals that can affect human health, due to its volatility, persistence, and bioaccumulation (Yu et al. 2016). Severe damage to the nervous system, kidneys, and other organs has been reported after Hg exposure (Shen et al. 2014). Thus, there is great concern about the release of mercury into the environment, e.g., through mining, and/or urban and industrial waste disposal (Ahamad et al. 2017). Among the main chemical forms of Hg, inorganic Hg(II) species in water pose a serious threat because of their mobility and complexity, showing features of wide distribution, low concentration and difficulty in control (Santana et al. 2016).
The US Environmental Protection Agency (USEPA), and WHO have set/recommended maxima of 1 μg-Hg L À1 for water and 5 μg-Hg L À1 for wastes discharged to the environment (Homayoon et al. 2017). In recent years various techniques have been used to remove Hg(II) from water, including solvent extraction, precipitation, coagulation, ion-exchange, reverse osmosis, membrane separation, coagulation, and photoreduction (AlOmar et al. 2017). Adsorption is considered an efficient and low cost technique for Hg (II) removal from wastewater effluents (Burakov et al. 2018).
Iron oxyhydroxide (FeOOH) nanoparticles are cheap, harmless and eco-friendly (Patra & Kim 2017), and used widely for efficient removal of various metals from aqueous media (Kokkinos et al. 2015). Iron hydroxides have been used widely as decontaminants to remove oxyanions like arsenite, arsenate, chromate, vanadate, phosphate, etc (Davantès et al. 2016). Recently, δ-FeOOH, magnetic iron oxyhydroxide, has been reported as an efficient adsorbent for Hg(II) from contaminated water (Maia et al. 2019). The maximum Hg(II) adsorption capacity of δ-FeOOH at pH 7 was determined as 35 mg g À1 . δ-FeOOH has a significant number of -OH groups on its surface, providing an attractive platform for coordination with Hg(II) metal species. Thus, to pursue high adsorptive performance, it was interesting to increase the number of active surface hydroxyl groups on it.
In this work, δ-FeOOH hollow structures were synthesized on a mesoporous silica surface, which was then treated with NaOH to dissolve the silica. The δ-FeOOH hollow structures were characterized by X-ray diffraction (XRD), transmission electron microscopy (TEM), BET specific surface and zeta potential. The effect of variables on Hg(II) adsorption was investigated and the material was used to treat mercury-contaminated water from the Doce River, Governador Valadares, Minas Gerais, Brazil.

Synthesis of mesoporous silica (MCM)
The mesoporous silica was synthesized as by Coelho et al. (2013) reported by dissolving 1.63 g of CTAB in a mixture of 30 mL of ultrapure water and 10 mL of 1 mol L À1 NaOH, adding 3.7 ml of TEOS and stirring the mixture for 24 hours at pH 11. After the contact time, the white precipitate was collected by centrifugation and washed several times until the pH of the final wash was 7. The material obtained was dried in a vacuum desiccator at ambient temperature for 48 hours. To remove the CTAB, 1,0 g of the dried solid was calcined in a muffle furnace at 550°C for 5 hours.
Synthesis of δ-FeOOH hollow structures MCM (1.0 g) was dispersed into 50 mL of 0.3 mol L À1 of Fe 2þ solution and sonicated for 5 minutes. Fe 2þ was precipitated as a green rust in the presence of MCM with 50 mL of 2 mol L À1 NaOH, after which 10 mL of 30% H 2 O 2 was added immediately, with stirring. Stirring continued for 30 minutes. The reddish brown precipitate FeOOH/MCM formed was washed several times with deionized water and dried in a vacuum desiccator at room temperature.
δ-FeOOH hollow structures were prepared by dissolving 1.0 g of δ-FeOOH/MCM in 50 mL of 5 mol L À1 NaOH and stirring at room temperature for 2 hours. Finally, the yellowish-brown material obtained was washed three times with deionized water and dried in a vacuum desiccator.

Materials characterization
Powder X-ray diffraction (XRD) data were collected from 5 to 80°2θ at a step width of 0.02°, 0.6 s per step, using Cu-Kα (λ ¼ 1.540560 Å) radiation in a Rigaku Geigerflex diffractometer equipped with a graphite diffracted-beam monochromator. Silicon was used as the external standard. Morphology analyses were conducted with a JEOL JEM 2000EXII transmission electron microscope. The specific surface of the nanoparticles was determined by the BET method, using N 2 adsorption/desorption in an Autosorb 1 Quantachrome instrument and the Barrett-Joyner-Halenda (BJH) method to determine the pore diameter distribution (Barrett et al. 1951). The zeta potential was measured by dynamic light scattering (Zetasizer Nano ZS, Malvern, UK). OriginPro (OriginLab, MA, USA, 2021) software was used for data processing.

Adsorption experiments
Hg(II) adsorption was investigated by batch method studies to examine the effect of various parameters -pH, contact time, initial Hg(II) concentration, and interferentsand determine the conditions leading to maximum Hg(II) sorption capacity. All experiments were carried in duplicate at 25 + 0.1°C in 50 ml, metal free tubes metal stirred at 200 rpm, using 10 mg of adsorbent. The effect of pH on Hg(II) adsorption was investigated at pH ¼ 3, 5, 7, 9 and 11. The Hg(II) adsorption rate were determined by batch experiments with varying reaction times. Solutions were in contact with hollow δ-FeOOH nanoparticles for a period of time between 0 and 360 minutes Adsorption isotherms were obtained by varying the initial Hg(II) concentration in the range 0-40 mg/L. The effect of interferent anions on Hg(II) removal from solution was investigated by performing adsorption experiments in the presence of solutions containing of 0 and 50 mg L À1 of Cl À , NO 3 À , SO 4 2À , PO 4 3À . Binary mixtures of Hg 2þ ions with different Cu 2þ , Co 2þ , Cd 2þ , Zn 2þ , Ni 2þ , Na þ or Mg 2þ metals in 0.1 mg L À1 concentration were investigated to determine the adsorption capacity of hollow δ-FeOOH in the presence of competitive cations. The initial and residual concentrations of mercury were determined with an inductively coupled plasma mass spectrometer (ICP-MS) (Perkin Elmer NexIon 300D). The analytical conditions are summarized in Table 1.
All samples analyzed after adsorption were submitted to dilution in 2% (v/v) nitric acid (HNO 3 ), using the USEPA 200.8 protocol and the Hg detection limit was determined as 2 ng L À1 (Ewa & Bosnak 2012). The Hg(II) removal efficiency and adsorption capacities were calculated using Equations (1) and (2): where R represents the Hg(II) removal efficiency (%), qt the mercury adsorption capacity on the adsorbent (mg g À1 ) at a given time, V and m the solution volume (L) and mass of adsorbent (g) respectively, and C i and C f the initial and final concentrations of Hg(II) in solution (mg L À1 ) respectively.

Sample collection
The water contaminated with mercury was collected in Governador Valadares, Minas Gerais, Brazil, in November 2015, using the procedure adopted by Environmental Company of the State of São Paulo (CETESB or Environmental Company of the State of São Paulo) (CETESB 2011). The samples were collected at two sites on the Doce River: P1 is within the area serviced by the water and sewage treatment system at 18°52 0 55.31 (S) 41°57 0 .86″ (W) and P2, on the urban perimeter, is at 18°51 0 22.54 (S) 41°56 0 10.55″ (W).

Nanofilter development
A nano-adsorbent filter was developed using the procedure reported by Maia et al. (2019) to remove mercury from water. 5 mL of a standard solution containing 50 μg-Hg(II) L À1 was applied to each filter, which then drained by gravity. The filtrate was collected after 20 minutes of continuous flow. The XRD patterns of the synthesized samples are shown in Figure 1. A broadening peak, centered around 2θ ¼ 23.5°, corresponds to amorphous silica (Coelho et al. 2013;Xiong et al. 2017) - Figure 1(a). As shown in Figure 1(a), all diffraction peaks in the XRD patterns can be readily indexed to the phase of δ-FeOOH (JCPDS Card no. 13-87). Figure 1(b), shows a decrease in the peak intensity for amorphous silica, indicating that some of it was removed in treatment with NaOH. The average particle size was calculated using the Scherrer Equation (3)for the most intense peak (100).
where 0.94 is the value assigned to the constant 'K', λ is the X-ray wavelength, θ half the diffraction angle, and β the angular width (radians) at the half-maximum intensity. The average crystal sizes calculated before and after treatment were 5.23 and 5.27 nm, respectively. The TEM image for the δ-FeOOH/MCM hybrid in Figure 2(a) shows a nanoparticle agglomerate, with a clear contrast between the dark δ-FeOOH and pale amorphous SiO 2 can be seen. It is impossible to define a characteristic form. The δ-FeOOH hollow structures were obtained by treating δ-FeOOH/MCM with 5 mol L À1 NaOH to remove the silica. It is possible from the STEM image (Figure 2(b)) to infer the formation of cavities on the nanoparticle surfaces, confirming that much of the nanomaterial has a hollow interior.
The N 2 adsorption-desorption isotherms for the δ-FeOOH/MCM and hollow δ-FeOOH are shown in Figure 3. The isotherms exhibit high N 2 adsorption, reaching 170 and 156 cm 3 g À1 at p/p 0 ¼ 0.95 for δ-FeOOH/MCM and hollow δ-FeOOH, respectively. A type IV isotherm was exhibited, characteristic of mesoporous materials, and a type H3 hysteresis loop, which is related to the filling of the mesopores due to capillary condensation. The BET area decreased from 94 m 2 g À1 in δ-FeOOH/MCM to 63 in hollow δ-FeOOH. This occurred because part of the silica was removed in the treatment with NaOH. The pore size distribution (Figure 3, inset) calculated from the desorption branch of the N 2 -isotherms by the BJH method, showed an average pore size of 20 nm, thus confirming the presence of mesopores.

Effect of pH
The effect of pH on Hg(II) adsorption onto hollow δ-FeOOH is shown in Figure 4(a). As the pH increased up to 9, mercury adsorption also increased. However, at pH 11, there was a slight decrease. The Hg(II) adsorption capacity at pH 9 was 8.01 mg g À1 corresponding to a removal efficiency of 93.3%. Subsequent studies at pH 7 showed Hg(II) removal efficiency of 92.4%.

Uncorrected Proof
The change in solution pH results in the formation of different aqueous Hg(II) species, including Hg 2þ , HgOH þ , and Hg(OH) 2 (Lu et al. 2014), and different hollow δ-FeOOH surface charges. When the mercury solution pH is below the PZC of hollow δ-FeOOH (PZC ¼ 4.72, Figure 4(b)), the surface of hollow δ-FeOOH is positively charged and interacts very weakly with mercury cations, due to protonation of the surface functional groups, and competition between the mercuric and H þ ions present in the solution (Husnain et al. 2016). On the other hand, when pH exceeded the PZC, the hollow δ-FeOOH surface was negatively charged and more functional groups were available for metal cation binding due to deprotonation, resulting in electrostatic attraction between Hg(II) and hollow δ-FeOOH. At pH above 9, insoluble metal hydroxide species (Hg(OH) 2 ) are formed, lowering adsorption efficiency (Shen et al. 2014).

Adsorption kinetics
The kinetics of Hg(II) adsorption are shown in Figure 5(a). Preliminary studies on the rate of mercury removal by hollow δ-FeOOH at pH 7 indicated that some 89.7% of the total mercury was removed after 60 minutes' adsorption, suggesting that Hg(II) adsorption on hollow δ-FeOOH is rapid initially but decreases gradually with time. Subsequently, the amount adsorbed increased slightly, and a maximum removal efficiency of 91.7% (8.13 mg g À1 ) was determined. This could arise from the high specific surface of δ-FeOOH due to its small particle size.
The pseudo-second-order equation (Equation (4)) was used to fit the experimental data to investigate the adsorption mechanism.
where q t and q e are the amounts of Hg(II) adsorbed at time t and equilibrium, respectively (mg g À1 ), and k 2 the rate constant of the pseudo-second-order adsorption (g mg À1 min À1 ).
The linear plot of t/q t versus t is shown in Figure 5(a). The correlation coefficient for Hg(II) was 0.999. Moreover, the values of q e calculated show good agreement with the experimental data (Table 2). Hg(II) adsorption in  Uncorrected Proof hollow δ-FeOOH can thus be approximated favorably by the pseudo-second-order adsorption model. The value of k 2 was calculated as 0.1151 g mg À1 min À1 and q e (calculated) was 8.141 mg g À1 .

Adsorption isotherm
The effect of the initial Hg(II) concentration on adsorption capacity was also studied ( Figure 6). The value of q e when the Hg(II) concentration was 1.5 mg L À1 was 2.56 mg g À1 , showing that the adsorbent is effective in adsorbing Hg(II) even at lower concentrations.
The adsorption mechanism at a given temperature is predicted using an adsorption isotherm (Luo et al. 2017).
The experimental data of mercury adsorption in hollow δ-FeOOH nanoparticles were fitted to the Langmuir and Freundlich isotherm models. The Langmuir model is used for the adsorption process on homogeneous, monolayer surfaces, where adsorption occurs at specific sites on the adsorbent. The Langmuir isotherm is represented by Equation (5): where Q m (mg g À1 ) and C e (mg L À1 ) are the maximum adsorption capacity and concentration at equilibrium, respectively, and k L is the Langmuir constant, representing the energy of adsorption (L mg À1 ). The Freundlich model is another that is used frequently to describe heterogeneous systems. It is represented by Equation (6):.
where C e is the equilibrium concentration of the adsorbate (mg L À1 ), and k F (L g À1 ) and 1/n are Freundlich constants representing the adsorption capacity and its intensity, respectively. 1/n values ,1.0 indicate an ordinary Langmuir isotherm, whereas a value of 1/n exceeding 1.0 suggests cooperative adsorption. The parameters derived from each model are shown in Table 3. The hollow δ-FeOOH experimental data fit best with the Langmuir model (R 2 ¼ 0.988). The Freundlich model showed a slightly lower correlation coefficient (R 2 ¼ 0.984), suggesting that Hg(II) adsorption on the hollow structures is monolayer.
The maximum Hg(II) adsorption capacity calculated using the Langmuir model was 89.1 mg g À1 , which is higher than that found for δ-FeOOH (Maia et al. 2019), possibly due to the increased number of hydroxyl groups (main binding site) in the hollow interior of the nanoparticles. The value of R L (¼1/(1 þ k L C 0 )) was below 1.0, indicating that the adsorption process was favorable, which was further supported by the 1/n values below 1 in the Freundlich model. The maximum Hg(II) adsorption capacity of hollow δ-FeOOH was compared with that of other iron oxides and hydroxides reported previously in the literature. The 89.1 mg g À1 found for hollow δ-FeOOH exceeded those for materials like FeOOH and FeMnOOH (6.4 and 11.9 mg g À1 ) (Kokkinos et al. 2015); Fe 3 O 4 and Fe 3 O 4 @SiO 2 (24.80 and 29.52 mg g À1 ) (Afraz et al. 2017), and Fe-Mn/graphene oxide composite (32.9 mg g À1 ) (Tang et al. 2016), but was lower than the 200 mg g À1 found for mesoporous α-FeOOH adsorbent (Patra & Kim 2017), with its high specific surface (152.4 m 2 g À1 ).

Interference study
Coexisting ions in solution affect many aspects of the adsorption process (Zhou et al. 2017). For example, different species can compete for the limited number of active sites, influence electrostatic interactions and impede the ion transfer from solution to the adsorbent surface (Tan et al. 2015). Figure 7(a) shows that Hg(II) adsorption efficiency was affected little by the presence of NO 3 À or SO 4 2À . The decrease of about 5% adsorbed mercury in the presence of Cl À was due to the formation of aqueous Hg(II)-Cl complexes, i.e., HgCl 3 À and HgCl 4 2À (Shan et al. 2015), leading to electrostatic repulsion from the negative surface of hollow δ-FeOOH. With phosphate the reduction was slightly greater at about 7%, because species formed that have little or no affinity with the variable-charge surfaces (Sarkar et al. 2000).
Figure 7(b) shows the competitive adsorption of Hg(II) in the presence of various metal ions in water. There was no significant interference in adsorption capacity at the concentration studied.

Application to real samples
The filter containing 0.5 g of hollow δ-FeOOH was tested initially for the treatment of water contaminated with a mercury standard at 50 μg-Hg L À1 . After filtration, the concentration was 1.150 μg-Hg L À1 , a removal efficiency of 97.7%. Samples of water collected from the Doce River and filtered in the same way contained Hg concentrations below the limit of detectionsee Table 4.   Uncorrected Proof