Arsenic (As) contamination has become a worldwide concerned issue in the water environment because of its potential toxic and carcinogenic effects through consumption of drinking water and food containing arsenic [1-5]. Due to their superior arsenic adsorption capacity, engineered nanomaterials are increasingly important for As removal from water [6-9]. Among these materials, iron nanomaterials [10] and their compositions (e.g., γ-Fe₂O₃, Fe₃O₄ [11], γ-Fe₂O₃ embedded silicon [12] or TiO₂ [13]) are the most used adsorbents because of their high sorption affinity toward inorganic arsenic (As(Ⅲ)/As(Ⅴ)) and easy separation. Nonetheless, the preparation of these nano-adsorbents is usually tedious and often needs relative expensive and pure salts. Sometimes, the practical application of nano-adsorbents for arsenic removal is unaffordable in developing countries. Unfortunately, such countries including Bangladesh, Vietnam and China have reported to have higher arsenic levels in the water environment that exceeds the World Health Organization (WHO) guideline (10 μg/L) [5, 14-15].

It is noteworthy that most previous works only focused on As (Ⅲ)/As(Ⅴ) removal [8]. This is probably because As(Ⅲ)/As(Ⅴ) was much more toxic over organic arsenics, and conventional adsorbents cannot efficiently remove organic arsenics [16-18]. However, large amounts of organic arsenic compounds (e.g., arsanilic acid and Roxarsone) are used as feed additives for breeding poultry, and the release of these compounds to the water environment is thus inevitable [19-22]. These organic arsenics can convert into As(Ⅲ)/As(Ⅴ) via chemical or biological transformation with dissolved oxygen, UV light and microorganisms [23-25]. Therefore, only removing inorganic arsenics from water is insufficient to prevent arsenic exposure to human. Significant challenges remain in developing a cost-effective adsorbent for the simultaneous removal of organic and inorganic arsenics.

Porous coordination polymers (PCPs) have been widely used in many fields because of their unique properties including porous, large surface area and a large amount of metal sites [26-30]. PCPs, particularly some metal-organic frameworks (MOFs) containing iron can easily form Fe-O-As bond on the surface and interior, significantly improving the absorption capacity of arsenic. To date, PCPs (ZIF-8 [31], MIL-53 [32] and UiO-66 [16]) have been applied for the removal of inorganic arsenic. Although few works focused on the removal of organic arsenic, PCPs actually contain large amounts of aromatic ligands, allowing them to also retain a great potential to remove organic arsenics through the π-π conjugation effect [20, 33].

Generally, PCPs are prepared via hydrothermal, solvothermal synthesis routines [34, 35]. Several disadvantages remain in these methods, including critical preparation conditions (high temperature and extreme pH), time-and solvent-consuming and producing secondary pollution [36, 37]. Recently, BASF succeeds in synthesis of HKUST-1 by electrochemical method [28]. Compared to conventional methods, electrochemical method is inexpensive and undertaken under mild conditions, requires less energy, and thus can easily realize mass production of industrial [28, 38-39].

In this work, we present a new, fast and facile electrochemical method to synthesize an Fe-PCPs by using iron and 1, 3, 5-benzenetricarboxylic acid (BTC) as sources of metal sites and organic linkers, respectively. The Fe-PCPs exhibits excellent adsorption capacity to both inorganic and organic arsenics via formations of Fe—O—As bond and π-π conjugation between the organic groups of organic arsenics and BTC, respectively. Scrap iron is one of the cheapest and most accessible metal materials and can be obtained from every waste recovery and recycling system. Therefore, scrap iron was used as an iron source in the current work, which eliminated the use of highly pure and relatively expensive iron salt, thereby making this method more acceptable and environmentally friendly. Additionally, the synthetic voltage is only 12 V, which is low enough to use a small portable battery to prepare this adsorbent. These advantages allow that the Fe-PCPs can be prepared whenever and wherever we want, particular for the emergent pollution of arsenic, see Fig. 1.

The as-prepared Fe-PCPs was characterized by SEM, TEM and powder X-ray diffraction (PXRD), as shown in Fig. S1 in Supporting information. The low-magnification SEM image shows that the FePCPs retains irregular shapes with diameter of 100–200 nm. The TEM image also indicates that the material has no fixed morphology. The PXRD plots of the Fe-PCPs and MIL-100(Fe) synthesized using iron ion and BTC are described in Fig. S1c, which demonstrate that the PXRD plot of Fe-PCPs has no strong and sharp peak compared to that of MIL-100 (Fe), revealing poor crystallinity associated with this material. According to the previous works [40], the Fe-PCPs is amorphous like F300. Furthermore, The PXRD characterization (Fig. S2 in Supporting information) shows that the structure of Fe-PCPs is not changed before and after adsorption of arsenic species.

The specific surface area and pore volume of the Fe-PCPs were determined by N₂ adsorption-desorption isotherms at 77 K. As can be seen from Figs. S1d–1f, the Brunauer-Emmett-Teller (BET) surface area and total pore volume of the prepared Fe-PCPs were calculated to be 1123.8 m²/g and 1.1 cm³/g, respectively. The mesopore size distribution curve (Fig. S1e) indicates that the average pore size of this material is about 3.9 nm. Table S1 in Supporting information summarized the comparison of the composition and textural parameters of Fe(BTC), MIL-100(Fe) and F300. Although the specific surface area and pore size of the Fe-PCPs are lower and bigger than those of MIL-100(Fe), respectively, they are comparable or better than those of F300. The high surface area and small pore size of Fe-PCPs result in its high As adsorption capacity.

Finally, the Fe-PCPs was characterized by XPS. The results described in Fig. S3a in Supporting information indicate that the as-prepared polymer contains Fe, O, and C. Carbon is ubiquitous and is present on all surfaces for XPS analysis, in which the carbon C 1s peak at 284.6 eV is used as a reference for charge correction. The XPS spectrum of C 1s for polymer is shown in Fig. S3b in Supporting information, which can be deconvoluted into three peaks centered at 284.6, 284.8 and 288.5 eV. The peaks at 284.6 eV and 288.5 eV corresponded to phenyl signals and carboxyl signals [41]. The peaks at 284.8 eV is the overlapping of the carbon signals on the surface. Fig. S3c in Supporting information shows the XPS peaks of Fe 2p3/2 and Fe 2p1/2. The binding energies of Fe 2p3/2 and Fe 2p1/2 obtained from this study are 711.5 eV and 725.5 eV. The Fe 2p3/2 peak has associated satellite peaks which obtained at 717.0 eV [42]. The O 1s peak at 531.7 eV corresponds to the Fe—O—C species [43]. According to these XPS information and the previously reported work [44], we speculate that the structure of the Fe-PCPs is similar to that described in Fig. S4 in Supporting information.

In order to attain the maximum adsorption capacity of arsenic on Fe-PCPs, the relevant operation conditions (pH, concentration of dissolved organic materials and ionic strength) were optimized as described in sections 4–9 of in Supporting information.

Adsorption kinetics are one of the most important factors to describe the adsorbent's performance and the mechanism of adsorption. Fig. 2a shows the time-dependent adsorption of As(Ⅴ), MMA, DMA, and ASA on Fe-PCPs. The initial concentrations of all arsenic species were 5 mg/L. Temperature and pH were fixed at 28 ℃ and 7, respectively. The results indicate that the adsorption equilibriums of these arsenic species were reached within 2 h, confirming rapid adsorption of inorganic and organic arsenic species on Fe-PCPs. These fast adsorption kinetics of all arsenic species were attributed to the well-developed mesopores, micropores, and high surface area of the prepared Fe-PCPs.

Two kinetic models including pseudo-first-order and pseudo-second-order models were further applied to gain an insight into the adsorption kinetics. For the pseudo-second-order kinetic model, the equation of the adsorption kinetics was evaluated as followed:

where q_e and q_t (mg/g) are the adsorption capacity of As species at equilibrium and a certain time t (h), respectively, and k₂ stands for the pseudo-second-order rate constant of adsorption (mg μg^-1 h^-1). Linear plots of t/q_t against t was achieved (Fig. 2b) and the values of q_e and k₂ were calculated from the slopes and summarized in Table S1 in Supporting information. The correlation coefficients of the pseudo-second-order rate model for the linear plots were better than 0.995, which suggested that kinetic adsorption of the four species were much better fitted with the pseudo-second-order kinetic model compared to the pseudo-firstorder kinetic model.

The adsorption isotherms of As(Ⅴ), MMA, DMA, and ASA were studied at room temperature (298 K) in the concentration range of 0.05–15 mg/L, as shown in Fig. 2c. The adsorption capacities of four species As were increased with increasing the initial concentration, indicating favorable adsorption of all these four As species at high concentrations. To evaluate their maximum adsorption capacities on Fe-PCPs, the adsorption isotherms were fitted with the Langmuir equation.

Where c_e (mg/L) and q_e (mg/g) are the equilibrium concentration and the equilibrium adsorption capacity of As(Ⅴ), MMA, DMA, or ASA, respectively. Q₀ (mg/g) is the maximum adsorption capacity, and b (L/mol) is the Langmuir constant. The plots of c_e/q_e against C_e with excellent linear coefficients were obtained for all of the tested initial concentrations (Fig. 2d), implying the adsorption of As(Ⅴ), MMA, DMA, and ASA on Fe(BTC) followed a Langmuir model. The obtained Langmuir parameters were summarized in Table S2 in Supporting information. According to the Langmuir isotherms, the adsorption capacities of Fe-PCPs adsorbent are as high as 18.65, 12.89, 18.39 and 40.44 mg/g to As(Ⅴ), MMA, DMA and ASA, respectively. Since studies on the adsorption of organic arsenics are relatively rare, the comparison of adsorption capacities between the Fe-PCPs and other adsorbents was only undertaken for As(Ⅴ) and summarized in Table S4 in Supporting information. The adsorption capacity obtained for As(Ⅴ) is comparable or even better than the reported adsorbents, which indicates the high potential of Fe-PCPs for the As removal in aqueous system. What the most important character is that Fe-PCPs retains good adsorption capacity not only for inorganic arsenics but also for organic arsenics.

To further explain the adsorption mechanism, the adsorption equilibrium constant (K₀), free energy change (ΔG, kJ/mol) for the adsorption of As species on Fe-PCPs were calculated based on the following equations:

where q_e (mg/g) is equilibrium adsorption capacity of As species adsorbed on Fe-PCPs, C_e (mg/L) is the equilibrium concentrations of As species, T is the temperature, R is the gas constant. The K₀ was obtained by ln (q_e/C_e) vs q_e and extrapolating q_e to zero (Fig. 2e). Table S3 shows the obtained values of ΔG and K₀. The values of ΔG are negative, suggesting that the adsorptions of As, MMA, DMA, and ASA on Fe-PCPs are spontaneous.

The mechanism of arsenic adsorption on iron and its compositions have been carefully discussed in previous works [7], which suggested this adsorption was attributed the formation of Fe-O-As. However, this is the first time to utilize the Fe-PCPs synthesized by electrochemical method for the adsorption and removal of both inorganic and organic arsenic species. Therefore, the Fe-PCPs before and after adsorption As(Ⅴ) and ASA were characterized by XPS and IR to further understand the mechanism of As species adsorption on Fe-PCPs. As can be seen from Fig. S10a in Supporting information, the arsenic signal can be clearly observed in the XPS spectrum of Fe-PCPs regardless of its adsorption of inorganic arsenics (As(Ⅴ)) or organic arsenics (ASA). However, the signal cannot be found in that of the FePCPs before As adsorption, indicating that both inorganic and organic arsenics can be adsorbed onto the as-prepared Fe-PCPs. The IR spectra of the Fe-PCPs before and after adsorption of As(Ⅴ) and ASA are described in Fig. S10b in Supporting information. The results show a new peak at 824 cm^-1 can be found in the spectrums of the Fe-PCPs after the adsorption of As(Ⅴ) or ASA compared to the IR spectrum of the original Fe-PCPs. According to the previous work [43], the 824 cm^-1 band belonged to the Fe-O-As groups. These results indicate that the Fe-O-As not only forms between the Fe sites and inorganic arsenic but also forms between the Fe sites and organic arsenic, resulting in the efficient removal of both inorganic and organic arsenics. Moreover, the maximum adsorption capacity of As(Ⅴ) on the Fe-PCPs (18.65 mg/g) is much higher than what reported (6.36 mg/g) [43] using Fe₂O₃ nanoparticles as adsorbent, which implies that the adsorption of arsenic species occurred not only on the outer surface of the Fe-PCPs but also in its interior. There is an interesting phenomenon that the maximum adsorption capacity of ASA on the Fe-PCPs is 2 times higher than that of As(Ⅴ). A reasonable explanation of this phenomenon is that the π-π conjugation not only forms between the organic groups of organic arsenics and Fe-PCPs but also plays an important role in the adsorption of ASA. TEM phase mapping characterization of the FePCPs after adsorption of As(Ⅴ) was investigated with corresponding phase mappings of Fe Kα1, O Kα1 and As Kα1, as shown in Fig. S11 in Supporting information. The images indicate that the As element (yellow) is evenly adsorbed in the Fe-PCPs structure, demonstrating that the arsenic ions really adsorbed both onto the interior and outer surface of the Fe-PCPs.

According to the previous works [9, 10, 43], the adsorption capacity of As(Ⅲ) on iron nanoparticles and its compositions are too low to be used for the removal of As(Ⅲ). As expected, the adsorption capacity of As(Ⅲ) on Fe-PCPs was 2.56 mg/g and was much lower than that of As(Ⅴ). Therefore, a pre-oxidation of As(Ⅲ) to As(Ⅴ) is required to improve the removal efficiency of As(Ⅲ). In order to accurately evaluate the oxidation efficiency of As(Ⅲ), a reported method based on HG-AFS was developed to selectively determine As(Ⅲ) from the mixture of As(Ⅲ) and As(Ⅴ) [45, 46]. The conversion efficiency of As(Ⅲ) was estimated from a comparison of the relative concentrations of As(Ⅲ) in the original and reacted solution. Fig. S12a shows that the efficiency obtained at different times after the 4% (v/v) of H₂O₂ were added into the solution of As(Ⅲ) with or without Fe-PCPs. It is very interesting that the conversion efficiency obtained in the presence of Fe-PCPs is much higher than that obtained only using H₂O₂, which implies the FePCPs did not only act as an adsorbent but also a catalyst to improve oxidation efficiency of As(Ⅲ). It is well known that iron ions, iron nanoparticles and iron based MOFs can be used as efficient Fenton or Fenton-like catalyst to catalytically produce powerful oxidative radicals of ^·OH from H₂O₂ [47]. Thus, these materials were widely used to determine H₂O₂, glucose, and other biological molecules and rapidly decompose organic pollutants. Therefore, it is speculated that the Fe-PCPs may also act as the role of Fenton catalyst and catalytically produce ^·OH from H₂O₂, improving the conversion efficiency of As(Ⅲ). To support this hypothesis, the oxidation radicals generated from H₂O₂ in the presence of the same amount of classical Fenton reagent (Fe²⁺), 30 nm of Fe₃O₄ nanoparticles and the prepared Fe-PCPs were measured by electron spin resonance spectroscopy (ESR). 5, 5-Dimethyl-1-pyrolin-N-oxide (DMPO, ACS reagent grade) was used as the spinning-trapping regent for hydroxyl radical and superoxide radical anion in this experiment. The ESR spectrums (Fig. S12b) exhibit that four characteristic peaks of the typical ^·OH/DMPO complex with an intensity ratio of 1:2:2:1 were found in all the tested catalysts. To some extent, the intensities of the identical lines of the DMPO adduct can reflect the catalytic activity of the catalysts on the generation of ^·OH. The productivity of ^·OH achieved using the Fe-PCPs is comparable and much better than those obtained using Fe²⁺ and Fe₃O₄ nanoparticles, respectively, clearly indicating that high specific surface area and 3D-ordered porous structure of Fe-PCPs are greatly helpful to improve the generation of ^·OH. Therefore, the As(Ⅲ) can also be efficiently removed from water via its conversion to As(Ⅴ) with H₂O₂.

The performance of Fe-PCPs on arsenic removal was validated by using three contaminated water samples. One of these samples is groundwater collected from a well and other two are surface water collected from a river of the Datong Basin, Shanxi province, China. The concentrations of arsenic species in these water samples before and after treatment with Fe-PCPs were analyzed by HPLC-ICP-MS, as shown in Figs. 3a–c. The results show that only As(Ⅴ) is found in the surface water samples. The original concentrations of As(Ⅴ) in these two samples are higher than 300.0 μg/L, which are much higher than that permitted by the WHO guideline (10 μg/L). Although As(Ⅲ) was observed in the groundwater sample, the concentrations of both As(Ⅲ) and As(Ⅴ) in this sample are much lower than those of surface water samples.

In order to demonstrate that Fe-PCPs can simultaneously remove inorganic and organic arsenics from water, a certain amount of As(Ⅴ), MMA, DMA and ASA were added to one of the surface water samples. The concentrations of the inorganic and organic arsenics before and after treatment were also analyzed by HPLC-ICP-MS, as shown in Fig. 3d. The results show that all the inorganic and organic arsenic were efficiently removed and their concentration in the treated water were much lower than their LODs obtained by HPLC-ICP-MS (ranging from 0.5 μg/L to 2 μg/L as sampling 20 μL of sample).

In conclusion, we have developed a simple, eco-economic and environmentally friendly method based on electrochemistry for the large-scale synthesis of iron porous coordination polymer from scrap iron. Besides the universal features of conventional PCPs including 3D-ordered porous structure and high surface areas, this prepared Fe-PCPs can not only form Fe-O-As bond with arsenic species but also produce hydrophobic and π-π conjugation through the aromatic rings of organic arsenics and the Fe-PCPs, allowing it to have high adsorption capacity for both inorganic and organic arsenic species. It is worth noting that this is the first report of the use of metal porous coordination polymer for the simultaneous removal of inorganic and organic arsenic species from contaminated water. It remains to explore the immobilization of Fe-PCPs on a substrate to improve its reusability for the arsenic removal.

Acknowledgment

We gratefully acknowledge the National Natural Science Foundation of China (Nos. 21575092 and 21622508) for financial support.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.cclet.2017.09.062.