Highly efficient removal of arsenate and arsenite with potassium ferrate: role of in situ formed ferric nanoparticle

It is well known the capacity of potassium ferrate (Fe(VI)) for the oxidation of pollutants or co-precipitation and adsorption of hazardous species. However, little information has been paid on the adsorption and co-precipitation contribution of the Fe(VI) resultant nanoparticles, the in situ hydrolytic ferric iron oxides. Here, the removal of arsenate (As(V)) and arsenite (As(III)) by Fe(VI) was investigated, which focused on the interaction mechanisms of Fe(VI) with arsenic, especially in the contribution of the co-precipitation and adsorption of its hydrolytic ferric iron oxides. pH and Fe(VI) played significant roles on arsenic removal; over 97.8% and 98.1% of As(V) and As(III) removal were observed when Fe(VI):As(V) and Fe(VI):As(III) were 24:1 and 16:1 at pH 4, respectively. The removal of As(V) and As(III) by in situ and ex situ formed hydrolytic ferric iron oxides was examined respectively. The results revealed that As(III) was oxidized by Fe(VI) to As(V), and then was removed though co-precipitation and adsorption by the hydrolytic ferric iron oxides with the contribution content was about 1:3. For As(V), it could be removed directly by the in situ formed particles from Fe(VI) through co-precipitation and adsorption with the contribution content was about 1:1.5. By comparison, As(III) and As(V) were mainly removed through adsorption by the 30-min hydrolytic ferric iron oxides during the ex situ process. The hydrolytic ferric iron oxides size was obviously different in the process of in situ and ex situ, possessing abundant and multiple morphological structures ferric oxides, which was conducive for the efficient removal of arsenic. This study would provide a new perspective for understanding the potential of Fe(VI) treatment on arsenic control.


Introduction
Arsenic contamination is abundant in drinking water ranks and groundwater which threat to the lives of millions of people throughout the globe, especially in the USA, India, Bangladesh, China, Canada, Hungary, Japan, Mexico, and Argentina, varying from ∼50 to >3000 μg/L, far higher than that of World Health Organization's recommended limit for drinking water (10 μg/L) (Kuo et al. 2017). Sources of arsenic can be either of natural origin such as soils, sediments, and natural waters which contain arsenic or as a result of anthropogenic activities such as processing of petroleum refineries, fossil fuel power plants, nonferrous, and smelting (Kolarik et al. 2018). Arsenic is always considered as a highly toxic element that increases risk of developing different types of cancer including skin, bladder, liver, and lung and causes damage to immune, nervous, and respiratory system (Mertens et al. 2016). Thus, optimizing treatment technologies for arsenic removal is currently of great urgency and high priority in many countries.
The dissolved forms of arsenic in water are predominantly the trivalent arsenite (As(III), such as H 3 AsO 3 , H 2 AsO 3 − , HAsO 3 2− ) and pentavalent arsenate (As(V), such as H 3 AsO 4 , H 2 AsO 4 − , HAsO 4 2− ) oxyanions (Guan et al. 2009). Various Responsible Editor: Tito Roberto Cadaval Jr approaches have been explored for arsenic removal, including coagulation/filtration, adsorption, ion exchange, photooxidation, and membrane separation, etc. (Kolarik et al. 2018;Matsui et al. 2017). Among them, coagulation and adsorption were viewed as affordable, cheap, and effective methods for large flow rates or high As(V) waters. At present, many materials have been investigated for arsenic removal by adsorption technique Yu et al. 2021;Yu et al. 2022). As(III) is more mobile and toxic than that of As(V) and has a low affinity to the surface of various adsorbents compared to As(V) (Jain et al. 2009). As(III) removal during coagulation with ferric coagulant has been shown to be less efficient than As(V) under comparable conditions, which should be due to As(V)-containing compounds show a negative charge contrary to As(III) species that are found to be mainly neutral, remarkably decreasing the removal efficiency for As(III)-containing compounds.
In addition, arsenite is easily oxidized to arsenate and once oxidized, several As(V)-removing technologies can be employed with more or great removal efficiency (Prucek et al. 2013;Qiao et al. 2012). Therefore, it is necessary and recommended a pre-treatment of As(III) oxidized to As(V) before coagulation-precipitation or adsorption processes for As(III) effective removal. Within these technologies, multifunctional water treatment agents for arsenic removal have attracted wide attention of many researchers. Potassium ferrate [K 2 FeO 4 , Fe(VI)] has been proved to be an environmental friendly agent for treating various organic and inorganic contaminants, which should be attributed to its ability of oxidation, flocculation, adsorption, co-precipitation, disinfection, etc. (Lee et al. 2014;Talaiekhozani et al. 2017). As a strong oxidant, especially in acidic conditions, Fe(VI) tends to attack electron-rich organic moieties, some kind of inorganic metal ions, and metal(II)-iminodiacetic acid complexed species (Acosta-Rangel et al. 2020;Yang et al. 2018b). Meanwhile, Fe(VI) would self-decay into ferric particle and act as adsorbent (Goodwill et al. 2015), and these in situ formed nanoparticles were identified to play a crucial role in predicting their efficiency in removing of pollutants. Specifically, these in situ formed ferric nanoparticles generated in the ferrate reduction process, such as Fe 2 O 3 , FeOOH, amorphous ferric, and these hydrolytic ferric iron oxides possessing the properties of highly dispersed, small in size (nanoparticle), and have abundant hydroxylation group, which could interact with oxidation products through the function of chemical bonds and hydrogen bond and adsorb them (Luo et al. 2021;Yang et al. 2018a). Recent research demonstrated that the in situ hydrolytic ferric iron oxides of Fe(VI) played an important role for removing high level of metal ions such as cadmium(II), cobalt(II), nickel(II), and copper(II), especially have a great potential to remove hazardous ions As(III), which might be due to its coagulation, adsorption, and co-precipitation (Liu et al. 2017;Wang et al. 2022b;Yunho et al. 2003). Studies have examined the removal of thallium by in situ formed and ex situ formed ferric particle from Fe(VI) and the results revealed that thallium removal should be attributed to the combination of adsorption and coprecipitation processes (Liu et al. 2017). Similarly, it was found that the phosphate removal efficiency by Fe(VI) dramatically changes during in situ and ex situ conditions, which indicated that the dominant Fe(VI) decomposition product γ-Fe 2 O 3 played a crucial role (Kralchevska et al. 2016). Lan et al. (2016) noted that As(III) could be oxidized by Fe(VI) and then adsorption by its in situ hydrolytic ferric iron oxides. Previous studies found that Fe(VI) could in situ formed ferric nanoparticles with γ-Fe 2 O 3 as the core and γ-FeOOH as the shell during the arsenic removal process, which reacted with As(III) and As(V) and then removed them from the solution in the form of core-shell nanoparticles and was not easy to leach and re-release into the environment. In particular, the ex situ formed ferric particle from the reduction of Fe(VI) could only remove 75% of As(III), while the in situ formed ferric particle from the reduction of Fe(VI) removed nearly 100% of As(III) (Prucek et al. 2013). Wang et al. (2020) showed that As(III) removed with Fe(VI) mainly through forming iron arsenate (FeAsO 4 ) precipitation and by the Fe(OH) 3 adsorption. In summary, previous studies have showed Fe(VI) could not only oxidize As(III) to As(V), but also remove As(V) by co-precipitation and adsorption of its in situ hydrolytic ferric iron oxides. However, little information has been paid on the adsorption and co-precipitation contribution of the Fe(VI) in situ hydrolytic ferric iron oxides.
Therefore, the interaction mechanisms of Fe(VI) and arsenic, especially contribution of the co-precipitation and adsorption of the in situ hydrolytic ferric iron oxides to arsenic removal, were first discussed in this study. The removal of aqueous As(III) and As(V) by Fe(VI) through batch experiments was investigated. The objectives of this paper are (1) to determine the efficiency of Fe(VI) and its different time hydrolytic ferric iron oxides on As(III) and As(V) removal under different Fe(VI) dosage, pH, and reaction time; (2) to compare the interaction mechanisms of Fe(VI) on As(III) and As(V) removal; and (3) to investigate the contribution content of co-precipitation and adsorption of the in situ hydrolytic ferric iron oxides on As(III) and As(V) removal. This study would provide a new insight into the interaction and mechanisms between arsenic and the hydrolytic ferric iron oxides of Fe(VI).

Preparation and characterization of chemicals
All chemicals were reagent-grade and used without any purification. Na 2 HAsO 3 and Na 3 AsO 4 ·7H 2 O were purchased from Sigma and were used to prepare the stock solutions of As(III) and As(V), respectively. Hydrochloric acid (HCl), tetramethylammonium hydroxide pentahydrate ((CH 3 ) 4 NOH·5H 2 O, TMA), nitric acid (HNO 3 ), and other reagents were purchased from Sinopharm Chemical Reagent Co., Ltd, China. Fe(VI) was prepared in the laboratory according to a wet method Liu et al. 2019). Briefly, calcium hypochlorite and potassium carbonate were used to produce potassium hypochlorite, and then Fe(VI) was produced by reacting potassium hypochlorite and iron nitrate under alkaline conditions with a purity higher than 95% by ABTS detection method (Acosta-Rangel et al. 2020). The tap water has been left in the open air for 24 h to prepare the water samples for the designed experiments, and the characteristics of the water samples were shown in Table S1.

Jar tests
Jar tests were performed open to the air with a jar testing device (ZR4-6, Shenzhen Zhongrun Co., Ltd.). The jar testing procedure was initiated with a rapid mixing at 300 rpm for 5 min, followed by 200 rpm for 2 min and then slow stirring at 40 rpm for 50 min and finally there was a 30-min settling. For the investigation the efficiency of Fe(VI) and its hydrolytic ferric iron oxides on arsenic removal, two sets of samples, termed as "in situ" and "ex situ," were prepared. The "in situ" samples originated from simultaneous additions of Fe(VI) and arsenic to the prepared tap water, after Fe(VI) and arsenic were added, the pH of the solutions was adjusted, and the rapid mixing was started immediately. The "ex situ" samples were formed in two steps. First, Fe(VI) was added to the prepared tap water and adjusted the pH of the solution to the setting pH and by subsequent shaking of the samples for 30 min. Then, arsenic stock solution was added to the mixture prepared in the first step. Moreover, in order to compare the arsenic removal efficiency by the hydrolytic ferric iron oxides of Fe(VI), "1-min ferric oxide" samples were prepared, which were similar to the "ex situ" sample procedure with shaking of the samples for 1 min before arsenic stock solution was added. To determine the influence of coexisting ions on As(V) or As(III) removal by Fe(VI), the cations Ca 2+ , Mg 2+ , and K + and anions Cl − , SO 4 2− , NO 3 − , PO 4 3− , SiO 3 2− , and CO 3 2− that ranged from 1 to 15 mg/L were dosed to the arsenic solution before Fe(VI) application. Note that 0.1 mol/L TMA and HCl were employed to adjust the desired pH during reaction. After each test, an appropriate amount of supernatant was taken, filtrated immediately through a 0.45-μm membrane (Shanghai ANPEL, China), and acidified with concentrated HNO 3 for determination of arsenic by ICP-AES (PerkinElmer, Optima 2000, UK). All experiments were carried out in triplicates.

Analytical and characterization techniques
The precipitated flocs were collected for particle size distribution using a particle size analyzer (MalvernZatasizer Nano ZS90, UK). The surface morphology and elemental content distribution of the flocs were observed using a scanning electron microscope (SEM-EDX, JSM-6490LV, JEOL, Japan). The flocs would be collected and processed, such as centrifugation, lyophilized, and grinding, and the crystal structures of the flocs were examined by X-ray diffraction (XRD, D8ADVANCE, BRUCKNER, Germany). To evaluate the coordination of complexes, a Fourier transform infrared spectroscope (FTIR, Nicolet6700, Nicolet, USA) was applied to determine the distribution of functional groups on the flocs. Chemical binding energies of arsenic, iron, and oxygen in the precipitates were analyzed by X-ray photoelectron spectroscopy (XPS).

Removal characteristics of As(V) and As(III) with Fe(VI)
To determine the influence of Fe(VI) dosage on As(V) and As(III) removal, 0.5-8 mg/L Fe(VI) was dosed to the 250 μg/L arsenic-containing solution at pH 4. The influence of pH on As(V) or As(III) removal by Fe(VI) was determined at different pH levels from 4 to 9 while keeping the optimum dosage of Fe(VI) obtained in the previous step. The effect of Fe(VI) dosage and pH on As(V) and As(III) removal was shown in Fig. 1. The results revealed that Fe(VI) dosage and pH have great influence on As(V) and As(III) removal by Fe(VI). As shown in Fig. 1a, at pH 4.0, the optimal As(V) and As(III) removal were observed of 97.5% and 98.3% under 6 mg/L and 4 mg/L Fe(VI), respectively. In addition, the concentrations of As(V) and As(III) in the filtrate were about 6.33 and 4.23 μg/L, lower than that of the limit (10 μg/L) specified by the Chinese drinking water standard ( GB5749-2006). Obviously, the removal of As(V) and As(III) increased as Fe(VI) dosage increased. As(V) removal experienced a considerable increased from 15.7 to 90.9% as Fe(VI) dosage increased from 0.5 to 4.0 mg/L and then slowly increased to the optimum removal rate with increasing Fe(VI) dosage to 6 mg/L (mFe(V)/mAs = 24). For the case of As(III) removal, it showed a similar trend to that of As (V) that increased significantly from 14.3 to 98.3% as Fe(VI) dosage increased from 0.5 to 4.0 mg/L (mFe(V)/mAs = 16) and then experienced basically unchanged with further increase Fe(VI) dosage. Note that As(III) removal was slight higher than that of As(V) when Fe(VI) was given a certain amount, which should be mainly associated with promoting the formation of Fe(OH) 3 during the oxidation reaction of As(III) to As(V) (Zheng et al. 2021).
As(V) and As(III) removed by Fe(VI) as functions of pH under the corresponding optimal Fe(VI) dosage of 6 mg/L and 4 mg/L respectively were shown in Fig. 1b. The results showed that As(V) and As(III) removal decreased rapidly from 97.5 to 56.5% and from 98.3 to 51.46% as pH increased from 4.0 to 9.0, respectively. Previous studies have shown that Fe(VI) could oxidize As(III) in 1 s under acidic conditions, and Fe(VI) would be more easily reduced to iron(III) oxides, the final products of Fe(VI), under this conditions (Yunho et al. 2003). However, Fe(VI) was relatively stable and difficult to be reduced to iron(III) oxides under alkaline conditions, and the corresponding coagulation and adsorption effect was poor (Wang et al. 2022a). At the same time, the electrostatic repulsion between the negatively charged iron nanoparticles formed by Fe(VI) and the negatively charged arsenic prevented the interaction between Fe(V) and arsenic under this condition .
As Fe(VI) was a promising alterative coagulant, the effect of flocculation time on arsenic removal by Fe(VI) was investigated. Fig.S1(a) shows that the residual concentrations of As(V) and As(III) decreased from 250 to 2.94 μg/L and to 0.58 μg/L respectively after 5-min flocculation time under the optimal condition. As the flocculating time increased, the residual content of arsenic decreased slightly, indicating that the flocculating time had little effect on arsenic removal. The residual iron contents during arsenic removal by Fe(VI) under different pH conditions were shown in Fig.S1(b). The residual iron contents were less than 0.3 mg/L in a wide pH range, which meet the requirements of water quality standards (GB5749-2006). It indicated that the residual iron content experienced a considerable decrease and then increased trend with pH increased, reached the lowest residual iron content at pH 5.0 with 0.030 mg/L and 0.065 mg/L during As(V) and As(III) removal, respectively. The high residual iron content under alkaline condition might be attributed to the strong stability of Fe (VI) and was difficult to be reduced to iron(III) oxides under this condition. Figure S2 shows the influence of coexisting ions on As(V) or As(III) removal by Fe(VI). In general, the influences of coexisting ions on As(III) removal were stronger than that of As(V) removal by Fe(VI), which might be due to the differences in the interaction of As(V) or As(III) with Fe(VI), especially with the in situ formed ferric particles. The cations of Ca 2+ , Mg 2+ , and K + showed a certain promoting effect on arsenic by Fe(VI) with the enhancement trend of Mg 2+ > Ca 2+ > K + , and its promoting effect decreased with the increase of ion concentration. The coexisting anions of Cl − , SO 4 2− , and NO 3 − showed little influence on arsenic removal by Fe(VI), while PO 4 3− , SiO 3 2− , and CO 3 2− had obvious inhibition effect on arsenic removal by Fe(VI) with the inhibition trend of PO 4 3− > SiO 3 2− > CO 3 2− . Studies have showed that ionic strength was an important parameter to determine the specific and non-specific adsorption removal rates of heavy metals by adsorbents (Egbosiuba et al. 2022). The specific adsorption process of heavy metals was basically not affected by the ionic strength of solution, while the non-specific adsorption of metal ions was generally affected by the change of ionic strength of solution . Therefore, As(V) or As(III) removal by Fe(VI) in situ formed ferric particles was more suitable a non-specific adsorption behavior.

Removal characteristics of As(V) and As(III) with Fe(VI) and ferric oxides
The contribution of Fe(VI) and the hydrolytic ferric iron oxides on As(V) and As(III) removal was examined and demonstrated in Fig. 2. Obviously, the results showed that the oxidation ability of the hydrolytic oxides of Fe(VI) in 1 min was greatly reduced, and the removal of arsenic by the 30-min hydrolytic oxides was mainly due to adsorption. As shown in Fig. 2a, Fe(VI) exhibited remarkable effect on As(V) removal that the residual concentration of As(V) reduced to 2.94 μg/L with the removal rate 98.8%, Fig. 1 Effect of Fe(VI) dosage and pH on As(V) and As(III) removal (initial concentration of arsenic 250 μg/L). a Influence of Fe(VI) dosage and b the influence of pH on As(V) removal at Fe(VI) dosage 6 mg/L and As(III) removal at Fe(VI) dosage 4 mg/L and the remaining concentration of As(V) was almost unchanged with further the reaction time. The removal efficiency of As(V) by the 1-min hydrolytic ferric iron oxides decreased. As the reaction time increased from 5 to 115 min, the residual As(V) concentration was reduced from 16.6 to 7.40 μg/L with the removal rate increased from 93.4 to 97.0%, respectively. As for As(V) removal by the 30-min hydrolytic ferric iron oxides, the removal efficiency was significantly reduced and the residual concentration of As(V) was 83.5 μg/L with the removal rate of 66.6% at 5-min reaction time, while it reduced to 37.3 μg/L with the removal rate of 85.1% at 115 min. Therefore, the oxidation of Fe(VI) has little effect on As(V) removal, and the adsorption and co-precipitation of the hydrolytic ferric iron oxides played an important role in the removal of As(V).
As illustrated in Fig. 2b, Fe(VI) showed remarkable effect on As(III) removal, and the residual As(III) concentration decreased to 0.58 μg/L after 5 min, while the removal effect of As(III) by the hydrolytic ferric iron oxides was poor. During As(III) removal by the 1-min hydrolytic ferric iron oxides, the residual As(III) concentration was 151.1 μg/L and the removal rate was only 39.6% after 5-min reaction time. When the reaction time increased to 115 min, the residual As(III) concentration decreased gradually to 100.2 μg/L with the removal rate 59.9%. For As(III) removal by the 30-min hydrolytic ferric iron oxides, the effect of As(III) experienced a considerable decrease. The residual concentration was reduced from 204.8 to 151.9 μg/L with the removal rate increased from 18.1 to 39.3% as reaction time increased from 5 to 115 min, respectively. The above phenomena further illustrated that the mechanisms of As(III) removal by Fe(VI) included the oxidation of Fe(VI), the adsorption and co-precipitation of the hydrolytic ferric iron oxides.

Surface characterization and structure of the characterization
The SEM images of the particles formed in Fe(VI) and the 30-min hydrolytic ferric iron oxides for arsenic removal were investigated to further understand their microscopic morphology, as illustrated in Fig. 3. As shown in Fig. 3a and b, the particles produced by Fe(VI) in situ for arsenic removal presented a large number of lamellar structures, of which the particles of As(V) removal were relatively loose and that of As(III) were more dense. Figure 3c and d shows the particles generated in the reaction of arsenic by the 30-min hydrolytic ferric iron oxides that consisted of a large number of spherical particles aggregated. Compared with the in situ removal of arsenic by Fe(VI), the particles formed by the ex situ removal of As(V) from the 30-min hydrolytic ferric iron oxides were more porous and that of As(III) were more dense. The difference of these particles structure between in situ removal of arsenic by Fe(VI) and that of ex situ removal of arsenic by the 30-min hydrolytic ferric iron oxides was related to the different arsenic removal mechanisms. In the process of arsenic removal during ex situ reaction, Fe(VI) self-decomposition formed a spherical hydrolysate, which played an important role in arsenic removal by adsorption and co-precipitation. The elements of the particles formed in Fe(VI) and hydrolytic ferric iron oxides for arsenic removal by EDS analysis were shown in Table S2, as it illustrated, arsenic in solution entered precipitation which indicated arsenic could be effectively removed by Fe(VI) and its hydrolytic ferric iron oxides. In addition, the elements of K, Fe, and O in the precipitation should be from Fe(VI). Since the solid Fe(VI) was synthesized through a wet chemical method using calcium hypochlorite, potassium carbonate, potassium hydroxide, and iron nitrate, with a purity higher than 95%, thus, the element of K might be due to impurities in Fe(VI).
The size distribution of flocs formed in Fe(VI) and hydrolytic ferric iron oxides for arsenic removal were monitored over the whole flocculation phase. As shown in Fig. S3, the average flocs size validated by three times experiments followed the order: ex situ-As(III) > ex situ-As(V) > in situ-As(III) > in situ-As(V), with the average size of 59.1 nm, 52.4 nm, 38.0 nm, and 35.0 nm, respectively. This result could be consistent with Kralchevska Fig. 2 Effect of As(V) and As(III) removal by Fe(VI) and ferric oxide at different reaction times (a initial As(V) concentration was 250 μg/L, Fe(VI) dosage was 6 mg/L, pH = 4.0, T = 25 °C, b initial As(III) concentration was 250 μg/L, Fe(VI) dosage was 4 mg/L, pH = 4.0, T = 25 °C) et al. (2016) research on phosphate removal by Fe(VI), which demonstrated that flocs generated in ex situ reaction by hydrolytic iron oxides were smaller and more dense than that of in situ process. In the ex situ process, the particle size would increase with the increase of hydrolysis time, and the harder co-precipitation worked; therefore, floc size was large while arsenic removal efficiency was poor. Moreover, As(V) and As(III) in the process of removing by Fe(VI) would be doped into the hydrolysis oxides earlier and inhibited the growth of the particle size of the hydrolysis oxides.

The composition of flocs
The FTIR spectra of the flocs formed in situ and ex situ of Fe(VI) for arsenic removal were illustrated in Fig. 4A. Specifically, the broadband near 3400 cm −1 was assigned to the stretching vibration of hydroxyl groups binding with iron or in H 2 O molecules (Ristić et al. 2007). The sharp peak near 1630 cm −1 corresponded to the bending vibration of hydroxyl groups in H 2 O molecules, indicating the presence of adsorbed water in the samples (Filip et al. 2011). As it shown in Fig. 4A(a), the band at approximately 1450 cm −1 and 1375 cm −1 attributed to the vibration of CO 3 2− and NO 3 − , respectively, and CO 3 2− might come from water, while NO 3 − should be ascribed to the residues in Fe(VI) (Jia et al. 2007). The peaks at 1400 cm −1 and 1053 cm −1 observed, indicating that carbonates existed in the flocs of arsenic removal, which should be due to both of the low crystalline and amorphous iron hydroxides were susceptible to CO 2 in the air, as illustrated in Fig. 4A(b-e) (Zhang et al. 2005). The new peaks emerged at 829 cm −1 and 833 cm −1 should be caused the stretching vibration of As-O-Fe, consistent with the relevant studies which noted that arsenic in the solution co-precipitates with the hydrolysates of Fe(VI) (Jia et al. 2007). In addition, previous studies had showed that the main mechanism of As(V) adsorption by amorphous iron oxides was through the formation of inner sphere arsenate complexes on the surface (Cheng et al. 2009;Senn et al. 2018). Moreover, the weak broadband at 829 cm −1 also indicated that As(III) removal by hydrolysis products of Fe(VI) was poor. The bank in the region 460-600 cm −1 was the typical characteristics of low crystallinity rust (Lan et al. 2016), and this conclusion was verified by XRD results that no sharp diffraction peaks observed, indicating the amorphous structure of the arsenic removal flocs (as shown in Fig. 4B). The diffraction peaks appeared around the 2 theta value of 26°, 34°, 58°, and 63°, respectively, which were the spectrum of ferric arsenate and ferric hydroxides according to the JCPDS database (Tang et al. 2011). Lan et al. (2016) also reported that the line 2 hydrous ferric oxides existed in the flocs of As(III) by Fe(VI), which possessed the low crystallinity, high specific surface area, and excellent arsenic removal performance.

Binding state of the material elements
The flocs of As(III) and As(V) removal by in situ Fe(VI) and ex situ 30-min hydrolytic ferric iron oxides were collected for XPS analysis to investigate the mechanisms of arsenic removal more accurately, as illustrated in Figs. 5, 6, 7, and 8, respectively. The peaks of Fe 2p, O 1s, C 1s, and As 3d were observed in XPS full spectrum (Figs. 5, 6, 7, and 8a), indicating that the obtained residue contained Fe, O, C, and As elements. Furthermore, the Fe 2p of the four flocs XPS spectrum results were similar, among them, the peaks near 713 eV, 720 eV, and 725 eV correspond to the Fe 2p 3/2 , satellite peak, and Fe 2p 1/2 , indicating that abundant kinds of iron oxides coexisted in the flocs (Kralchevska et al. 2016). For example, iron oxides/hydroxides, such as Fe 2 O 3 , Fe(OH) 3 , and FeOOH, were observed in the four reactive precipitates, as shown in the curve fitting of Fe 2p (Fig.S4) and the same phenomenon has been found in other studies (Liu et al. 2014). Previous studies have shown that Fe(VI) hydrolysis produces nanoparticles with Fe 2 O 3 core and FeOOH shell, as Fe 2 O 3 and FeOOH have similar XPS characteristic peaks (Prucek et al. 2015;Xu et al. 2022); the O1s spectrum of the precipitated product after the reaction was used to investigate the changes of components in detail.
The curve fitting of O1s of As(III) removal by in situ Fe(VI) was shown in Fig. 5c; the binding energy peaks at 533.58 eV, 531.38 eV, and 529.58 eV were assigned to H 2 O, OH − , and O 2− , respectively (Xu et al. 2022). The presence of H 2 O indicated the adsorbed water existed in the sample, which was in accordance with the result from the FTIR analysis. The presence of OH − groups indicated that hydrolytic iron oxides of Fe(VI) existed in the forms as Fe(OH) 3 and FeOOH, while that of O 2− could be corresponded to FeOOH or Fe 2 O 3 (Liu et al. 2014;Sun et al. 2013). The ratio of the surface OH − group to O 2− could reflect the species of iron oxides and hydroxides in samples. Previous studies have shown that the stoichiometric ratios of OH − and O 2− in FeOOH were about 0.9-1.1 (Liu et al. 2014), which was much less than that of the results in Fig. 5c, confirming that the iron oxide in this sample was mainly Fe(OH) 3 . Figure 5d reveals the forms of arsenic in the flocs that two obvious characteristic peaks near 45 eV and 48 eV were observed, which should belong to the As(V) characteristic peak (Liu et al. 2018), consistent with Prucek et al. (2015) had reported. It is noteworthy the contribution of adsorption and co-precipitation to As(III) removal. In detail, the characteristic peak near 45 eV indicated that As(III) was oxidized and then adsorbed on the surface of iron oxide nanoparticles, while that of peak near 48 eV might attributed to the oxidized As(V) co-precipitated with iron oxide nanoparticles. Therefore, comparing the curve fitting area of adsorbed and coprecipitated, it can be seen that the ratio of the removal amount of As(III) by co-precipitation and adsorption in solution was about 1:3.
As illustrated in Fig. 6c, the analysis of O1s for removing As(V) flocs by in situ Fe(VI) showed that the binding energy peaks of H 2 O, OH − , and O 2− were located at 534.58 eV, 531.38 eV, and 529.38 eV, respectively (Xu et al. 2022). Compared with the removal of As(III) by in situ Fe(VI), the content of adsorbed water was reduced, while the proportion of OH − and O 2− was almost unchanged, indicating that the iron oxide in the sample was mainly Fe(OH) 3 . As shown in Fig. 6d, the results of As 3d analysis showed that the amount of arsenic removed by co-precipitation at 48 eV embedded in nano-oxides and adsorption at 45 eV was about 1:1.5. It is noticed that the different contribution content of co-precipitation and adsorption in the As(III) and As(V) removal by in situ Fe(VI) might be due to the somewhat different interaction between arsenic and Fe(VI). That is, the morphology and structure of the two flocs were obviously different, and the intraparticle diffusion of arsenic is favored along the defective, high-energy, and nonequilibrium polycrystalline grain boundary of iron oxides (Tucek et al. 2017).
The O 1s spectra of removing As(III) by ex situ 30-min hydrolytic ferric iron oxides of Fe(VI) was investigated in Fig. 7c, it revealed the binding energy peaks of 535.08 eV, 531.98 eV, and 529.78 eV belonged to H 2 O, OH − , and O 2 respectively, among them, the content of OH − occupied the majority, and that of H 2 O and OH − were relatively low (Xu et al. 2022). Figure 7d demonstrates that the binding energy peaks at 44.58 eV and 45.18 eV corresponded to As Fig. 7 XPS results of the precipitates from the 30-min hydrolytic ferric iron oxides removing As(III) (a full spectrum, b Fe.2p, c O 1s, d As 3d) 3d 5/2 and As 3d 3/2 , where that of As(III) characteristic peak located, implying there was no oxidation existed during As(III) removal under this condition . It is important to note that the depth of XPS detection is only a few nanometers, reflecting most of the arsenic was present on the outer surface and there was less As(III) embedded in the nano-oxide structure (Yan et al. 2012). Therefore, the results confirmed that As(III) removal was mainly through adsorption by the hydrolytic ferric iron oxides of Fe(VI). Figure 8c shows the O 1s spectra of removing As(V) by ex situ 30-min hydrolytic ferric iron oxides of Fe(VI); the binding energy peaks corresponding to H 2 O, OH − , and O 2 were located at 532.48 eV, 531.28 eV, and 529.78 eV, respectively (Xu et al. 2022). The content of H 2 O was higher than that of As(III) removal by the hydrolytic ferric iron oxides, while that of OH − was lower than that of As(III) removal process. The peaks at 45.18 eV and 45.78 eV belonged to As(V) as shown in Fig. 8d, and the depth of probe detected by XPS was only a few nanometers which implied As(V) removal was enriched mainly by adsorption on the surface of the 30-min hydrolytic ferric iron oxides of Fe(VI) (Yan et al. 2012).

Analysis of the mechanisms on arsenic removal by Fe(VI)
The interpretation of arsenic removal mechanisms and reactivity characteristics were illustrated in Fig. 9 for Fe(VI) and its hydrolytic ferric iron oxides. Fe(VI) has a high reduction potential that can rapidly oxidize As(III) to As(V) in solution under acidic conditions. Importantly, the in situ iron oxide nanoparticles formed possessing low crystalline rust and amorphous iron oxide/ hydroxide, such as Fe 2 O 3 , Fe(OH) 3 , and FeOOH, and these iron oxides played an important role on arsenic removal. Arsenic removal efficiency in the process of in situ formed and ex situ formed ferric particles revealed that the mechanisms on arsenic removal by Fe(VI) and its ex situ formed hydrolytic ferric iron oxides were clearly different. That is, arsenic can be removed by adsorption and co-precipitation of the hydrolysis of Fe(VI) in the in situ process. In the ex situ process, arsenic removal was mainly through adsorption by the 30-min hydrolytic ferric iron oxides of Fe(VI). The co-precipitation process consumed the Fe-O-Fe of oxide hydrolyzed by Fe(VI) to generate Fe-O-As, namely, arsenate. During the adsorption process, arsenic in the solution was adsorbed by the hydrolytic ferric iron oxides and formed an internal spherical complex on the surface Zhu et al. 2021). In brief, during the process of As(III) removed by Fe(VI), As(III) could be oxidized instantaneously and then the formed As(V) removed through co-precipitation and adsorption by the in situ iron oxide nanoparticles. Significantly, the contribution of co-precipitation and adsorption was about 1:3. For As(V), there was only coprecipitation and adsorption through the in situ iron oxide nanoparticles with the contribution proportion about 1:1.5. In addition, adsorption was the main mechanism for arsenic removal by ex situ 30-min hydrolytic ferric iron oxides of Fe(VI), and As(V) removal effect was better than that of As(III) might be due to the low affinity of As(III) to the hydrolytic ferric iron oxides.

Conclusion
The interaction mechanisms of Fe(VI) and arsenic, especially contribution of the co-precipitation and adsorption of the in situ hydrolytic ferric iron oxides on arsenic removal, were investigated. The results revealed that pH and Fe(VI) played significant roles on arsenic removal, and the optimum removal efficiency was achieved with the removal rate of 97.8% and 98.1% when Fe(VI):As(V) and Fe(VI):As(III) were 24:1 and 16:1 at pH 4, respectively. The flocs size in the process of in situ and ex situ was obviously different and these hydrolytic ferric iron oxides possessed abundant and multiple morphological structures ferric oxides, which provided a large number of adsorption sites for arsenic. During the process of As(III) removed by Fe(VI), As(III) was oxidized to As(V) and then was removed though co-precipitation and adsorption by the hydrolytic ferric iron oxides with the contribution content was about 1:3. For As(V), it could be removed directly by the in situ formed particles from Fe(VI) through co-precipitation and adsorption with the contribution content of 1:1.5. In the ex situ process, As(III) and As(V) were mainly removed through adsorption by the 30-min hydrolytic ferric iron oxides.