Remediation of arsenic-contaminated water by green zero-valent iron nanoparticles

The optimal conditions for the green synthesis of nano zero-valent iron (G-NZVI) using mango peel extract were investigated using a Box-Behnken design approach. Three factors were considered, namely the ratio of iron solution to mango peel extract ratio (1:1–1:3), feeding rate of mango peel extract (1–5 mL min−1), and agitation speed (300–350 rpm). The results showed that the optimal conditions for the synthesis of G-NZVI for arsenate removal were a 1:1 ratio of iron solution to mango peel extract, a mango peel extract feeding rate of 5 mL min−1, and an agitation speed of 300 rpm. Under these conditions, nearly 100% arsenate removal was achieved. X-ray diffractometry (XRD), Fourier transform infrared spectroscopy (FT-IR), transmission electron microscopy (TEM), Brunauer–Emmett–Teller (BET), and scanning electron microscopy (SEM) with energy-dispersive X-ray analysis (EDX) methods were used to characterize the properties of the G-NZVI. Finally, the arsenate removal efficiency of the G-NZVI was compared against that of commercial nano zero-valent iron (C-NZVI). The results revealed that the G-NZVI was roughly five times more efficient at arsenate removal than the C-NZVI. The influence of background species such as chloride (Cl−), phosphate (PO43−), calcium (Ca2+), and sulfate (SO42−) was studied to evaluate their effects on arsenate removal. As a result, Cl− and Ca2+ were shown to play a role in promoting arsenate removal, whereas SO42− and PO43− were observed to play an inhibiting role.


Introduction
Arsenic is pervasive in the environment and can be poisonous as well as carcinogenic. Widespread arsenic contamination is currently regarded as a serious environmental issue (Rahaman et al. 2022). Chronic ingestion of inorganic arsenic, particularly via water consumption, is associated with skin and bladder cancers as arsenic accumulates and concentrates in the liver, lungs, kidneys, and skin tissue (Fatoki et al. 2019). In surface water, the prevalent form of arsenic is arsenate (As(V)). Therefore, developing a highly efficient process for As(V) removal is critical. To remove arsenic from contaminated water, various techniques have been employed including co-precipitation, membrane separation, coagulation, and adsorption (Suwannatrai et al. 2020;Phanthasri et al. 2018;Khamdahsag et al. 2021). Among these techniques, adsorption is one of the most appropriate for arsenic removal due to its convenient operation, costeffectiveness, efficacy, and low consumption requirements. There are many adsorbents for arsenic removal such as commercial and synthetic activated carbons, agricultural product and by-products, and industrial by-products. Therefore, an adsorbent for the efficient removal of As(V) would be valuable.
The adsorption process using zero-valent iron (ZVI) is a promising option for removing arsenic because using this Responsible Editor: George Z. Kyzas 1 3 sorption medium has enabled both As(V) and As(III) to be removed concurrently without the need for pre-oxidation or extra chemical reagents (Tanboonchuy et al. 2011a, b;Nakseedee et al. 2015). However, there are limitations to using micro-scale ZVI, because the needed reaction time is in days and generally requires a lot of iron to remove the target pollutant effectively (Sasaki et al. 2009;Zhang et al. 2004). Therefore, the development and synthesis of a more effective nanoscale ZVI (NZVI) for arsenic removal are needed.
NZVI can be synthesized using precision milling, ultrasound-assisted production, or carbothermal reduction, or via electrochemical methods (Wang and Zhang 1997). From the methods mentioned, the chemical synthesis approach has certain disadvantages such as toxic by-products and expensive chemicals (Tanboonchuy et al. 2012). Additionally, the synthesized NZVI particles tend to aggregate due to the loss of stabilization. Inadequate synthesis conditions may also result in a decrease in the reducing capacity of the substance because of the high reactivity of NZVI with oxygen and water, as shown in Eqs. (1-6) (Tanboonchuy et al. 2010(Tanboonchuy et al. , 2011a. Therefore, the development of environmentally friendly synthesis methods using plant extracts, such as green tea, oolong tea, pomegranate peel, mulberry, and cherry, has resulted in these products being considered "green chemicals for NZVI" (G-NZVI) (Xiao et al. 2020;Zhu et al. 2018;Rashtbari et al. 2020;Poguberović et al. 2016). Tree leaves are a potential reducing agent for the synthesis of NZVI due to their high polyphenol content and antioxidant capabilities. In addition, polyphenol is biodegradable and water-soluble at room temperature. They are also an alcohol functional group molecule that can reduce the size of NZVI which can prevent aggregation and aid in nanoparticle stabilization (Hoag et al. 2009;Mystrioti et al. 2015;Nadagouda et al. 2010).
Based on the aforementioned, the main objective of this study was to investigate the effect of the conditions of G-NZVI synthesis for application in As(V) removal. Mango peel was used as the raw material for G-NZVI synthesis because it has high polyphenol content, as noted by earlier studies (Berardini et al. 2005;Ignat et al. 2011). Moreover, mango peel contains phenolic compounds such as auroxanthin, carotenoid anthocyanin, and flavonoid protein antioxidants (Ajila et al. 2010). Phenolic compounds can produce complex compounds with metal ions and reducing agent characteristics that are biodegradable and have good solubility at room temperature. In a previous study, it was found that phenolic compounds are higher in mango peel than in mango flesh (Kim et al. 2010).
The Box-Behnken design (BBD) approach was applied in the experimental design to study the optimum conditions for G-NZVI synthesis. Subsequently, synthesized G-NZVI generated from each condition was tested for As(V) removal. Finally, X-ray diffraction (XRD), point of zero charges (pH pzc ), transmission electron microscopy (HR-TEM), scanning electron microscopy (SEM) with energy-dispersive X-ray analysis (EDX), and Brunauer-Emmett-Teller (BET) tests were used to characterize the synthesized G-NZVI.

G-NZVI synthesis, efficiency test, and characterization
Ripe mango peels were prepared by washing them with deionized water. The extraction of polyphenols from mango peels followed our previous study (Deewan and Tanboonchuy 2022). The G-NZVI was synthesized from extracted solution and pumped into a 0.1 M FeCl 3 aqueous solution. The mixture was stirred using a rotating propeller. Ferric iron (Fe 3+ ) was reduced by polyphenol, according to Eq. (7).
where n is the amount of hydroxyl group oxidized by Fe 3+ and Ar is a phenyl group. The experimental factors considered were the ratio between the iron solution and mango peel extract (X 1 ), the feeding rate of the extracted solution (mL min −1 ) (X 2 ), and the agitation speed rate (rpm) (X 3 ) ( Table 1). The combination of these factors resulted in 15 conditions according to BBD (Table 2).
Each synthesized G-NZVI (Table 2) was tested for As(V) removal by adding 0.3 g to a 100-mL aqueous solution of arsenic (initial concentration is 0.5 mg L −1 ). A sample was collected after 40 min and analyzed using ICP-OES. The results were calculated as the percentage of As(V) removed by G-NZVI, according to Eq. (8).
where Y is the percentage of As(V) removed, and C 0 and C t are As(V) the initial concentration (0.5 mg L −1 ) and concentration following reaction time, respectively. The G-NZVI that provided the highest percentage of As(V) removal was then characterized by XRD, BET, SEM with EDX, FT-IR, and TEM.

Effect of background species on arsenate removal
A fractional factorial design (FFD) was applied to create the experiments used to evaluate the performance of arsenate removal in the presence of ions (SO 4 2− , PO 4 3− , Ca 2+ , and Cl − ). The synthesized G-NZVI (2.5 g L −1 ) was added to a reactor with a volume of 50 mL for arsenate treatment (initial concentration of 5 mg L −1 ). Two levels of low and high concentrations for each selected ion were presented as follows: PO 4 3− : 1, 5 mg L −1 ; SO 4 2− : 10, 100 mg L −1 , Cl − : 50, 500 mg L −1 ; Ca 2+ : 50, 500 mg L −1 . A sample was collected after 40 min and measured.

Characterization of G-NZVI
The XRD results for crystallite samples of G-NZVI are presented in Fig. 1(a). The formation of NZVI was confirmed by the outstanding peaks at 44.8° (Eljamal et al. 2020). However, peaks of iron oxides were observed, which agreed with previous studies that reported the NZVI's morphology is composed of two separate layers. Iron oxides such as FeOOH and Fe 2 O 3 make up the exterior layer that covers the Fe 0 , whereas the inner core is represented by Fe 0 itself (Phanthasri et al. 2022;Fu et al. 2016).
It was also confirmed by the SEM-EDX profile, which demonstrated Fe, O, and Cl are the primary components with the following weight percentages: 90.66: 7.66: 1.68 ( Fig. 1(b)). The histogram in Fig. 1(c) shows that the average particle size was 4.30 nm, and more than 95% of the particles were smaller than 6 nm based on a count of 100 particles. The small G-NZVI can be produced because the alcohol functional group in the extracted solution can stabilize nanoparticles and decrease aggregation (Kim et al. 2010). The pH pzc of the lab-synthesized G-NZVI solution was around 2.5 ( Fig. 1(d)). Finally, the specific surface area was calculated as 26.67 m 2 g −1 according to the BET method.
FT-IR spectra show the surface structures and functional groups of the commercial NZVI (C-NZVI) and G-ZNVI that are responsible for their stability and reduction. Figure 1(e) depicts the FT-IR spectra of the commercial NZVI (C-NZVI) and G-ZNVI. The G-NZVI's peak at 3285 cm −1 is attributed to the O-H group, which includes chemicals like phenol and/or carboxylic acids found in lignin, pectin, and cellulose (Desalegn et al. 2019). It has been revealed that phenolic compounds can decrease Fe 3+ to Fe 0 and improve the stability of the synthesized G-NZVI. The peak at 1636 cm −1 identifies the C = O stretching vibration in aldehyde and ketones (Karavasilis and Tsakiroglou 2019). The peak at 1430 cm −1 was derived from the C-C stretch in ring aromatics. Mango peel contains lignin, pectin, cellulose, and flavonoids, all of which contain diverse functional groups including ketone, carboxyl, aldehydes, and hydroxyl groups (Fazlzadeh et al. 2017). The range of 1126-1190 cm −1 , which corresponds to the carbonyl group,  demonstrates heterocyclic compounds produced by plant extract proteins. These substances cause nanoparticles to create capping ligands. The aggregation of newly synthesized G-NZVI is prevented by the present proteins (Khashij et al. 2020). This result is consistent with the TEM analysis, as shown in Fig. 1(c). The intense band found at 562 cm −1 for the G-NZVI sample corresponds to the stretching vibration mode of Fe-O bonds. Generally, the bands observed at 562 cm −1 show the presence of iron (hydr)oxides (Lin et al. 2017). The Fe-O stretching vibration band was observed at 562 cm −1 in the G-NZVI, but not prominently observed in C-NZVI. This result agrees with the XRD and SEM-EDX analyses, as presented in Fig. 1(a) and (b).

Statistical analysis
The optimal operational and experimental conditions were determined using the response surface method. According to the results presented in Table 2, the percentage of As(V) removal ranged from 42.69 to 97.81%. Regression analysis techniques were used to fit a quadratic model to predict the As(V) removal rates. To determine Fig. 1 Characteristics of G-NZVI from (a) XRD pattern, (b) SEM-EDX image, (c) TEM image and particle size distribution (inserted), (d) point of zero charge plot, and (e) FT-IR spectra in comparison to C-NZVI coefficients, the response was plotted as a function of independent variables, as shown in Eq. (9): where Y represents the percentage of As(V) removal. The coefficients prior to X x X y are the interaction between two factors and the quadratic effects, respectively. The negative sign of each term denotes an antagonistic effect, whereas a positive sign denotes a synergistic effect. The relationship between the factors and their response values was evaluated at a significance level (α) of 0.05 using a regression model. Overall, the model achieved a coefficient of determination (R 2 ) greater than 95% (p = 0.005) with several factors exceeding the significance level (Table 3) (Rakhmania et al. 2021;Zhang et al. 2022).
The results revealed a strong correlation between the predicted values obtained from Eq. (9) and the experimental results (R 2 = 97.89%, Fig. 2). The ratio of iron solution to extracted solution (X 1 ), which had a p-value of < 0.001 (Table 3), was the most significant factor determining the removal of As(V). On the contrary, the feeding rate of mango peel extract (X 2 ) and agitation speed (X 3 ) were not significant factors (p = 0.848 and p = 0.232, respectively).
The main effects plot for the percentage of As(V) removal is shown in Fig. 3. The lower ratio of iron solution to mango peel extract produced the highest concentration of G-NZVI. Considering the feeding rate, it is preferable to complete the reaction as quickly as possible to avoid the potential for surface oxidation of the G-NZVI. Additionally, it is beneficial to reduce the reaction time for G-NZVI synthesis. However, rapid feeding of the extracted solution can cause aggregation of the (9) Y = 577 − 58.4X 1 + 31.4X 2 − 2.95X 3 + 17.08X 2 1 + 1.211X 2 2 + 0.00516X 2 3 − 1.51X 1 ⋅ X 2 − 0.0860X 1 ⋅ X 3 − 0.1091X 2 ⋅ X 3 G-NZVI particles, leading to a decrease in the active surface area of particles (Tanboonchuy et al. 2012). In other words, there is a trade-off between reducing the reaction time and the removal efficiency of G-NZVI particles. However, the results indicated that the feeding rate had a negligible impact on the removal of arsenate under these conditions (1-3 mL min −1 ). The results also showed a slight decrease in the reactivity of G-NZVI particles with increasing agitating speed. These results might be explained by the higher agitation speed introducing a significant amount of oxygen, resulting in the surface oxidation potential of G-NZVI, as presented in Eqs.
(1)-(6). Figure 4 shows the surface plot and contour plot for the effectiveness of arsenate removal. The maximum As(V) removal was greater than 90% with the optimum conditions being a 1:1 ratio of iron solution to mango peel extract, a feeding rate of 5 mL min −1 , and an agitation speed of 300 rpm.

Comparison between G-NZVI and commercial NZVI (C-NZVI)
The kinetics of arsenate removal efficacy were compared between commercial NZVI (C-NZVI) and G-NZVI. The potential rate reaction was determined by pseudo-firstorder and pseudo-second-order models. The linear form of pseudo-first-order and pseudo-second-order equations based on solid capacity is shown in Eqs. (10) and (11): Fe:extract ratio . * feeding rate X 1 * X 2 − 1.51 0.266 Fe:extract ratio . * agitation speed X 1 * X 3 − 0.086 0.414 Feeding rate * agitation speed X 2 * X 3 − 0.1091 0.074 where q e and q t are the amounts of arsenate removal at equilibrium and at time t (mg g −1 ) and k 1 and k 2 are the equilibrium rate constant of pseudo-first-order kinetics (min −1 ), and pseudo-second-order kinetics (g mg −1 min −1 ), respectively. The G-NZVI outperformed the C-NZVI in As(V) removal (Fig. 5). G-NZVI achieved 90% removal within t q e 60 min, while C-NZVI achieved around 60% in the same amount of time. The kinetic parameters (Table 4) indicated that the experimental results for both G-NZVI and C-NZVI were best fitted by the pseudo-second-order kinetic model, according to the correlation coefficient (R 2 ). The pseudo-second-order kinetic model concluded that the rate of kinetic reaction relied on the As(V) concentration. There are 3 steps for the reaction mechanism, namely mass transfer, pore diffusion, and adsorption. The pseudosecond-order kinetic replied that the adsorption step is a rate-limiting step. Moreover, the amount of As(V) removal Fig. 3 Plots of the main effects including iron solution to the extract solution ratio (X 1 ), feeding rate of mango peel extract (X 2 ), and agitation speed rate (X 3 ) on As(V) removal

Fig. 4
Contour plots and surface plots of the As(V) removal efficiency corresponding to the effects of the iron solution to extract solution ratio * feeding rate of mango peel extract (X 1 * X 2 ), iron solution to extract solution ratio * agitation speed rate (X 1 * X 3 ), and feeding rate of mango peel extract * agitation speed rate (X 2 * X 3 ) at equilibrium time for the G-NZVI (0.8344 mg g −1 ) was approximately five times higher than that of C-NZVI (0.1573 mg g −1 ). Adsorption capacities of G-NZVI and C-NZVI on arsenate were investigated by an equilibrium adsorption isotherm study. Adsorption data were fitted with Freundlich and Langmuir isotherms, as in Eqs. (12) and (13), respectively: where q e and q m are the amounts of arsenate uptake on the adsorbent at the equilibrium (mg g −1 ) and maximum adsorption capacity (mg g −1 ), respectively. C e is the concentration of arsenate at the equilibrium (mg L −1 ) and n is the adsorption intensity. K L and K F are the rate constants of Langmuir isotherm and Freundlich isotherm, respectively.
Compared to the Freundlich isotherm and Langmuir isotherm shown in Table 5, the Langmuir isotherm fitted well with the adsorption of arsenate on both G-NZVI and (12) q e = K F C 1∕n e (13) q e = q max K L C e 1 + K L C e C-NZVI compared to the Freundlich isotherm. The maximum adsorption capacity (q m ) calculated from the Langmuir model was 0.2773 mg g −1 and 0.0403 mg g −1 for G-NZVI and C-NZVI, respectively.

Effect of background species on arsenate removal
As shown in Fig. 6, the removal of arsenate was decreased in the presence of SO 4 2− and PO 4 3− . A previous study reported that the electrical repulsion between SO 4 2− and negatively charged arsenate species (H 2 AsO 4 − or HAsO 4 2− ) resulted in a decrease in the efficiency of removal (Su and Puls 2001), while the presence of PO 4 3− inhibited the removal performance of arsenate because the predominant dissociation species of phosphate and arsenate have comparable chemistry. By forming inner-sphere complexes with the hydroxyl groups, it may be said that phosphate and arsenate species compete for the same adsorption sites on the surface of iron (oxy)hydroxides (Jegadeesan et al. 2005;Hsu et al. 2008).
As presented in Fig. 6, the removal of arsenate was more effective when Ca 2+ and Cl − were present. Arsenate can remain on the adsorption sites because the presence of Ca 2+ reduces electrostatic repulsion between negatively charged arsenate and iron (oxy)hydroxides, which act to neutralize their negative surface charges (Parks et al. 2003). In an alternative interpretation, the Ca 2+ could form a complex with the surface sites (SS) of iron (oxy)hydroxides, as shown in Eqs. (14) and (15).

Material
Pseudo-first-order Pseudo-second-order k 1 (min −1 ) q e (mg g −1 ) R 2 k 2 (g mg −1 min −1 ) q e (mg g −1 )  By raising the positive surface charge of the adsorbent, both reactions create a bridge between its surface and the negatively charged arsenate, which encourages the adsorption of more negatively charged arsenate e species (Vaishya and Gupta 2004). In a previous report, it was found that the presence of Cl − in solutions caused the pitting corrosion of iron surfaces, potentially increasing the reactive area of iron for arsenate adsorption (Choe et al. 2004).

Conclusions
Experiments for green nano zero-valent iron (G-NZVI) synthesis were designed according to the Box-Behnken design to assess the ability of G-NZVI to remove arsenate in an aqueous solution. The results showed that the optimal conditions for the synthesis of G-NZVI were a 1:1 ratio of iron solution to mango peel extract, a mango peel extract feeding rate of 5 mL min −1 , and an agitation speed of 300 rpm. The most important parameter associated with arsenate removal was the ratio of iron solution to mango peel extract. A multiple linear regression model, including the three factors and interaction effects, thoroughly described the experimental data (R 2 = 0.9789). A kinetic study showed that the reaction rate of G-NZVI was best described using a pseudo-second-order model and that G-NZVI outperformed a commercial NZVI in terms of the amount of arsenate removed at equilibrium time (0.8344 mg g −1 compared to 0.1573 mg g −1 , respectively). In terms of the effect of ions, SO 4 2− and PO 4 3− play an inhibiting role in removing arsenate, whereas Ca 2+ and Cl − enhance the effect. , Ca 2+ , and Cl − (experimental conditions: volume ratio of iron solution to extract solution = 1, 2, 3, feeding rate of mango peel extract = 1, 3, 5 mL min −1 , agitation speed rate = 300, 325, 350 rpm, G-NZVI, and C-NZVI loading = 0.3 g, and [As(V)] = 0.5 mg L − . 1 )