The superior performance of silica gel supported nano zero-valent iron for simultaneous removal of Cr (VI)

Pure nano zero-valent iron (NZVI) was fabricated under optimum conditions based on material production yield and its efficiency toward acid blue dye-25 decolorization. The optimum prepared bare NZVI was immobilized with two different supports of silica and starch to fabricate their composites nanomaterials. The three different prepared zero-valent iron-based nanomaterials were evaluated for removal of hexavalent chromium (Cr(VI)). The silica-modified NZVI recorded the most outstanding removal efficiency for Cr(VI) compared to pristine NZVI and starch-modified NZVI. The removal efficiency of Cr(VI) was improved under acidic conditions and decreased with raising the initial concentration of Cr(VI). The co-existence of cations, anions, and humic acid reduced Cr(VI) removal efficiency. The removal efficiency was ameliorated from 96.8% to 100% after adding 0.75 mM of H2O2. The reusability of silica-modified NZVI for six cycles of Cr(VI) removal was investigated and the removal mechanism was suggested as the physicochemical process. Based on Langmuir isotherm, the maximal Cr(VI) removal capacity attained 149.25 mg/g. Kinetic and equilibrium data were efficiently fitted using the pseudo-second-order and Langmuir models, respectively confirming the proposed mechanism. Diffusion models affirmed that the adsorption rate was governed by intraparticle diffusion. Adsorption thermodynamic study suggested the spontaneity and exothermic nature of the adsorption process. This study sheds light on the technology that has potential for magnetic separation and long-term use for effective removal of emerging water pollutants.


Results and discussion
Optimization of the preparation process of NZVI. The effects of reaction time, reducing agents, and iron precursors on the preparation process of NZVI were optimized based on NZVI yield and the performance of the prepared NZVI towards the decolorization of acid blue-25. The preparation process of NZVI requires a suitable reaction time to obtain metallic iron nanoparticles, so the preparation of NZVI was conducted at reaction times of 10, 30, 60 and 120 min using NaBH 4 and FeCl 2 .4H 2 O as a reducing agent and iron precursor, respectively. The yield of NZVI and removal efficiency of dye decreased with raising the reaction time as shown in Fig. 1a,b, respectively. The NZVI yield decreased from 37.38% to 8.4% by extending the time from 10 to 120 min and the removal efficiency of the dye decreased from 66.7 to 16.5%. The extension of reaction time may increase the tendency of NZVI surface oxidation and the formation of iron oxides. It was noticed that the increase in time resulted in the raising of rusting. So, the optimum reaction time for NZVI production was considered as 10 min.
The reducing agent has a significant role in the preparation process of NZVI. To specify the most proper reducing agent for NZVI synthesis, NaBH 4 , N 2 H 4 , NaOH and NH 4 OH with 4 M were used as reducing agents following the previously mentioned synthesis procedures of NZVI.
In the case of using NaBH 4 , a black magnetic precipitate was produced. However, NaOH and N 2 H 4 produced a brown precipitate showing lower magnetic properties. In the case of NH 4 OH, a non-magnetic green color precipitate was formed. The results showed that N 2 H 4 , NaOH and NH 4 OH could not effectively reduce the iron salt to its zero state. Therefore, NaBH 4 was the most suitable reducing agent for the preparation of NZVI.
Four iron salts were used (FeCl 2 .4H 2 O, FeCl 2 , FeSO 4 .7H 2 O and FeCl 3 .6H 2 O) to specify the best iron salt for the preparation of NZVI based on the yield (%) and dye removal efficiency (%) as demonstrated in Fig. 1c,d. The preparation process was conducted using the optimum reaction time and reducing agent. The results indicated www.nature.com/scientificreports/ that FeCl 2 .4H 2 O attained the highest yield (66.35%) and removal efficiency of dye (99.9%) compared to other salts. Moreover, the results confirmed that the preparation of NZVI by Fe(II) was better than that produced using Fe(III). Therefore, the optimum iron salt for NZVI production was FeCl 2 .4H 2 O. In the synthesis of NZVI, the economic cost into consideration for the need of the clean water, especially in developing countries. Hence, in the calculations for removal efficiency, it is important to calculate the costeffectiveness index of the nZVI-synthesis parameters, especially with the optimization of different reducing agents due to the cost of reducing agent is so high. In addition, the relevant cost and energy consumption should be taken into consideration for a practical optimization 38 . Physico-chemical characteristics of the synthesized nanomaterials. TEM image of bare NZVI (Fig. 2a) showed a chain-like agglomeration of nanoparticles, whereas the aggregation of nanoparticles in the case of NZVI-St and NZVI-Si as shown in (Fig. 2b,c) was lower confirming the role of different supports as stabilizers in providing the required repulsive forces between nanoparticles. The nanoparticles had a spherical shape with low tendency of aggregation for the composite materials with a particle size ranged from 20 to 50 nm in the case of pure NZVI and modified NZVI. The grey area in TEM images of NZVI-Si and NZVI-St is attributed to silica gel and starch supporting materials confirming the excellent support of NZVI on silica gel and starch. Moreover, high-resolution TEM images in Fig. 2d-f insured the excellent support of NZVI on silica gel or starch.
To investigate the chemical composition of the synthesized nanomaterials, the EDS pattern in Fig. S1a-c showed the growth of Fe and O in the case of NZVI confirming the partial oxidation of NZVI, whereas the presence of Fe, O and Si in the case of NZVI-Si and Fe, O and C for NZVI-St was affirmed. Moreover, EDS elemental mapping in Fig. 3a-h reconfirmed the chemical composition of the synthesized nanomaterials. Additionally, the EDS and elemental mapping analyses were performed after the removal of Cr(VI) and the results demonstrated the introduction of Cr to the chemical composition in the case of NZVI and NZVI-Si as shown in Fig. 3i-q which insures the adsorption of Cr(VI) on the surface of the adsorbent. SAED patterns in Fig. S2 showed the crystallinity of the synthesized materials.  www.nature.com/scientificreports/ Figure 4a show the XRD patterns of NZVI, NZVI-St and NZVI-Si. The significant peak at around 45 o is ascribed to the NZVI diffraction plane (110) (JCPDS, file No. 87-0722), whereas there is no peak nearly 35° is imputed to the (311) diffraction plane of Fe 3 O 4 , confirming there is no or rare formation of the iron oxide layer on the NZVI surface during the preparation process 39,40 . Chemical structures of the three zero-valent based nanomaterials were confirmed using FTIR analysis. The bands at around 1618 and 3423 cm −1 are ascribed to bending and vibration of the O-H bond owing to the adsorbed water molecules on the adsorbent's surface in the case of NZVI, whereas the bands at 3417 and 1635 cm −1 in the case of NZVI-Si are ascribed to the presence of O-H bond besides the existence of stretching vibrations of Si-O-Si group at 1095 cm −1 as depicted in Fig. 4b 41,42 . Regarding NZVI-St, the band at 2912 cm −1 is allocated to the CH 2 OH group in the starch which confirms the excellent support of NZVI on starch and the bands at 1624 and 3424 cm −1 are imputed to the adsorbed water molecules on the surface 41 . The bands around 1030 cm −1 and 1380 cm −1 are ascribed to the C-O and COObonds respectively which are presence at NZVI, NZVI-Si and NZVI-St 43 . The bands at nearly 484, 476 and 565 cm −1 are allocated to the stretching vibrations of Fe-O in the case of NZVI and its composites NZVI-Si and NZVI-St, respectively 44 . As shown in Fig. 4c, both NZVI and NZVI-St showed type-II adsorption isotherm, while NZVI-Si showed type-IV isotherm. The porous texture of NZVI-Si is different from NZVI and NZV-St, confirming the characteristic effect of the mesoporous structure of silica in NZVI-Si. In addition, NZVI-Si presents a developed few micropore structures that strongly favors high surface area values compared to those for NZVI and NZVI-St. The calculated surface areas of the three prepared materials NZVI, NZVI-St and NZVI-Si were 12, 14, and 60 m 2 /g, respectively, confirming the significant enhancement of surface area after the support on silica that resulted to the excellent adsorption performance of NZVI-Si. In Fig. 4d, NZVI-Si has a higher mesopore volume than NZVI-St, resulting in the higher removal of pollutants via pore. The magnetic saturation values (Fig. 4e) Fig. 5c,e are imputed to Fe 2p 1/2 , Fe 2p 3/2 and Fe 0 , respectively which reconfirms the formation of the iron oxide layer in the case of NZVI and NZVI-Si 46 . The peak at nearly 101.56 eV is ascribed to the Si 2p which affirms the excellent support of NZVI on Si (Fig. 5f) 26 .
The existence of the Cr 2p peak revealed that the Cr(VI) was successfully trapped on the NZVI-Si (Fig. 5g). The high resolution XPS spectra were recorded after the Cr(VI) removal process and insignificant shifts in the www.nature.com/scientificreports/ binding energies were observed due to the complexation between Cr(VI) and the adsorbents' surface as shown in Fig. 5g-m. The peak at 581.9 eV is ascribed to Cr(VI) which affirms the adsorption of Cr(VI) on the surface (Fig. 5n). Additionally, the peaks at 577.68 and 588.23 eV are attributed to Cr 2p 3/2 and Cr 2p 1/2 of Cr(III), respectively in the case of NZVI and confirming the reduction of Cr(VI) to Cr(III) after adsorption on the materials surfaces 40 . The existence of Fe 0 peak after Cr(VI) removal suggests the high reusability capability of both NZVI-Si and NZVI-St.
Effect of pH and initial Cr(VI) concentration on the removal performance. Figure    Furthermore, iron hydroxides can be easily formed on the adsorbents' surface in alkaline conditions, and they can block the active sites which decreases the reactivity of NZVI, NZVI-Si and NZVI-St 48 . Moreover, the iron oxides on the adsorbents' surface can de-accelerate the electron transfer from the adsorbent to sorbent 49 . The removal efficiency was enhanced in alkaline conditions using modified NZVI, as silanol groups in the support can be separated and provide protons that maintain the pH and prevents the passivation 4 . www.nature.com/scientificreports/ Figure 6b shows the effect of initial Cr(VI) concentration on the adsorption capacity of NZVI, NZVI-Si and NZVI-St and the removal efficiency of Cr(VI). NZVI-Si and NZVI-St showed higher removal performance compared to pristine NZVI at different initial Cr(VI) concentrations due to the reduction of aggregation of nanoparticles after the support on silica and starch as explained in the characterizations section. Moreover, the introduction of these supports can reduce the oxidation of the adsorbents' surface and inhibit the formation of passivation layer which facilitates the electron transfer between NZVI and Cr(VI). Also, the amelioration of the removal performance of Cr(VI) in the case of modified NZVI was due to the increase in the surface area.
The increase of adsorption capacity from 4.95 to 91.5 mg g −1 by raising the initial concentration of Cr(VI) from 5 to 100 mg L −1 at pH 1, contact time of 120 min, the temperature of 20 °C and NZVI-Si dose of 0.1 g/100 mL was owing to the ameliorated driving forces between Cr(VI) and reactive sites. On the other hand, the removal efficiency of Cr(VI) went down from 90.8 to 72.4%, 95.6 to 78.3% and 99 to 91.5%, respectively with the increase of initial Cr(VI) concentration from 5 to 100 mg L −1 using NZVI, NZVI-St and NZVI-Si, respectively. In the case of low concentrations of Cr(VI), the active sites on the NZVI, NZVI-Si and NZVI-St are adequate for the reduction and adsorption of Cr(VI). However, in the case of high concentrations of Cr(VI), the reactive sites are not enough to adsorb the high number of Cr(VI) ions which inhibits the appropriate contact between the adsorbents and Cr(VI) ions 50 . Moreover, the formation of passivation layer on the adsorbents' surface can be accelerated in the case of high Cr(VI) concentration which reduces the reactivity of the synthesized adsorbents with Cr(VI) and prevents the electron transfer from NZVI surface to Cr(VI) 51,52 . Zhou et al. (2022) reported the same trend during the removal of Cr(VI) by a modified NZVI 53 . Due to the superiority of NZVI-Si for Cr(VI) adsorption is over NZVI-St and NZVI, NZVI-Si. It was selected to investigate the remaining parameters affected on Cr(VI) removal as well as the experimental data of Cr(VI) removal using NZVI-Si were fitted using adsorption kinetic, isotherm and thermodynamic models. , respectively compared to 96.8% in the case of no anions at NZVI-Si dose of 0.1 g/100 mL, initial Cr(VI) concentration of 25 mg L −1 , contact time of 120 min and temperature of 20 °C. In spite of the improvement of the ionic strength after adding the aforementioned anions, they can compete with Cr(VI) and occupy the binding sites which de-accelerate the removal of Cr(VI) 53 . Furthermore, anions such as CO 3 2and PO 4 3can form innersphere complexes with iron (oxy) hydroxides and the formed complexes can block the active sites reducing the removal efficiency 54 .
The investigation of the influence of the existence of cations such as Mg 2+ , Zn 2+ and Cu 2+ on the removal efficiency of Cr(VI) was carried out as shown in Fig. 7b at NZVI-Si dose of 0.1 g/100 mL, initial Cr(VI) concentration of 25 mg L −1 , contact time of 120 min, cations concentrations of 40 mg L −1 and temperature of 20 °C. The removal efficacy of Cr(VI) was 96.8% in the case of no-cations, whereas the addition of Mg 2+ and Zn 2+ decreased the removal efficiencies to 68% and 84%, respectively. Mg 2+ and Zn 2+ ions can occupy the reactive sites instead of Cr(VI) which decreases the removal efficiency of Cr(VI). On the other hand, the removal efficiency of Cr(VI) was 96% after adding Cu 2+ . The electron transfer and corrosion of Fe 0 can be improved owing to the formed bimetallic surface after adding Cu 2+ . The enhancement of electron transfer and corrosion after the addition of Cu 2+ can overweigh the negative effect of the occupation of binding sites by Cu 2+ ions leading to high removal efficiency approximately the same as the blank sample. Chen et al. (2016) stated the improvement of the removal of hexachlorobenzene after Cu 2+ addition using NZVI composited with activated carbon 54 .
The removal efficiency of Cr(VI) was enhanced to 98.4% in the presence of 5 mg L −1 of HA compared to 96.8% in the absence of HA as shown in Fig. 7c at NZVI-Si dose of 0.1 g/100 mL, initial Cr(VI) concentration  Adsorption kinetics. The pseudo-first-order rate constant and coefficient of determination (R 2 ) were estimated via the linear plot of ln (q e -q t ) versus time (t) as shown in Fig. 9a. The R 2 was high (0.97); however, there was a significant difference between experimental q e (23.39 mg g −1 ) and obtained q e (18.04 mg g −1 ) from the pseudo-first-order model. Therefore, the pseudo-first-order model was not efficient to describe the adsorption of Cr(VI) on the NZVI-Si surface. Regarding the pseudo-second-order model, the model constants were estimated from the slope and the intercept of the linear relation between t/q t and t as depicted in Fig. 9b. The significance of the pseudo-second-order model was affirmed by the high R 2 (0.9972) and the slight difference between experimental q e (23.39 mg g −1 ) and calculated q e (26.8 mg g −1 ) from the model. The intraparticle diffusion model constants and R 2 were estimated by the linear plot of q t versus t 0.5 . The multi-linearity in Fig. 9c indicated that the adsorption process took place in three stages as well as the multistage sorption of Cr(VI) on the NZVI-Si surface. The first stage describes the transfer of Cr(VI) from the solution to NZVI-Si outer surface or boundary layer diffusion. The second stage represents the entrance of Cr(VI) ions into the pores by intraparticle diffusion. The third stage refers to the diffusion of Cr(VI) into the small pores till reaching the equilibrium. The lines in Fig. 9c did not pass through the origin affirming that the adsorption of Cr(VI) could be attained via intraparticle diffusion but it was not the only rate-governing step 56 . Moreover, the non-zero intercept in the case of the Boyd model in Fig. 9d reaffirmed that the rate of adsorption was controlled by intraparticle diffusion. The high R 2 of the parabolic diffusion model affirmed that the adsorption of Cr(VI) could occur by intraparticle diffusion as shown in Fig. 9e. Linear plot of q t versus ln(t) as depicted in Fig. 9f was used to estimate the Elovich model constants and R 2 . The high R 2 indicated that Elovich model was satisfactory to describe the adsorption process. Power kinetic model constants were determined via the linear plot of ln(q t ) versus ln(t) and the modest R 2 value expressed that this model could not effectively describe the adsorption process (Fig. 9g). Table 1 shows the kinetic models' constants and R 2 . www.nature.com/scientificreports/ Adsorption isotherms. The linear relation between C e and C e /q e was plotted to estimate the Langmuir isotherm constants and R 2 as portrayed in Fig. 10a. The high R 2 (0.9976) indicated that the adsorption isotherm of Cr(VI) on the adsorbent's surface could be expressed by the Langmuir model. The maximum monolayer adsorption capacity was 149.25 mg g −1 . R L value was lower than 1 confirming the favorable adsorption of Cr(VI) on NZVI-Si surface confirming the suggested mechanism for Cr(VI) removal. Moreover, Freundlich and Temkin isotherm models' constants were estimated from the slope and intercept of the linear plots shown in Fig. 10b,c. The parameters of the different studied equilibrium isotherm models are listed in Table 2. The R 2 was lower in the case of Freundlich and Temkin indicating that Langmuir was more suitable to fit the experimental data. The value of 1/n in Freundlich equation was lower than 1 suggesting the favorability of the adsorption of Cr(VI) on the NZVI-Si surface. The linear plot of ε 2 versus ln(q e ) could be employed to estimate Dubinin-Radushkevish constants (Fig. 10d). The mean adsorption energy calculated as 17.5 kJ mol -1 indicating that the adsorption of Cr(VI) on NZVI-Si surface is physico-chemical adsorption. The low R 2 in the Hankins-Jura model is as shown in Fig. 10e depicted that the adsorption process was monolayer which was in agreement with the results obtained from Langmuir model. The Generalized isotherm model constants were estimated via the linear plot of ln( q m q e − 1) versus ln(C e ). The generalized isotherm model could not describe the adsorption process of Cr(VI) on NZVI-Si because of its low R 2 compared to other isotherm models (Fig. 10f).   www.nature.com/scientificreports/ Reusability of NZVI/Si. NZVI-Si was used in repetitive cycles to investigate its reusability performance as shown in Fig. 12. The NZVI-Si particles were collected after each cycle by a magnet, washed with water and dried before the subsequent use. The removal efficiencies were 96.8%, 93.67%, 90.1%, 86.7, 82.9% and 74.8% in the six consecutive runs, respectively. The results indicated the efficient reusability of NZVI-Si. The removal efficiency was higher in the first cycle owing to the availability of active sites. However, the number of binding sites decreased in the following runs. Moreover, the removal efficiency decreased in successive cycles due to the oxidation of NZVI-Si and the formation of a passivation layer in successive cycles 8 . Figure 13 shows the three-step removal mechanism of Cr(VI) (adsorption, reduction and precipitation). NZVI nanoparticles consist of a core and shell. The shell is composed of iron oxides that can be formed via the environmental oxidation of NZVI 51 . Moreover, NZVI can Table 2. Constants and coefficients of determination of isotherm models.

Isotherm models Parameters
Langmuir model q m = 149.25 mg g -1 K L = 0.113 L mg -1 R 2 = 0.9976 R L = 0.0815 Freundlich   59 . Then, the Cr(OH) 3 can be formed on the NZVI surface and then precipitated as shown in Eq. (4). Moreover, Cr 3+ can be incorporated into the iron oxide/hydroxide layer forming Fe 3+ -Cr 3+ complexes. The formed Fe 3+ -Cr 3+ hydroxides can de-accelerate the electron transfer from the core to the surface which inhibits the reduction of Cr(VI) and reduces the removal performance especially at high initial Cr(VI) concentrations 59 . The adsorption of Cr(VI) on NZVI surface was confirmed by EDS, EDS elemental mapping and XPS. Moreover, XPS analysis affirmed the reduction of Cr(VI) to Cr(III). The support on silica or starch can facilitate the electron transfer and inhibit the rapid oxidation of NZVI surface which improves the removal of Cr(VI) 50 . Moreover, supporting on silica or starch can decrease the agglomeration of NZVI which improves the reactivity, reducibility and dispersibility of NZVI as shown in TEM images. Further, the Cr(VI) can diffuse into the pores of the supporting materials as well as supporting materials can provide NZVI with higher surface area and active sites which increased the adsorption and reduction  www.nature.com/scientificreports/ performance of bare NZVI 56 . Additionally, the starch or silica can employ as nanoreactors for accelerating the reaction between supported-NZVI and Cr(VI) and the NZVI in the pores can effectively reduce Cr(VI) 60 . The difference between Cr(VI) concentration in the outer and inner pores creates a driving force that participates in attaining frequent adsorption and diffusion of Cr(VI) into the pores till reaching the equilibrium 61 . Thus, the supported-NZVI can effectively remove Cr(VI).

Preparation of pure nano zero-valent iron (NZVI) and modified NZVI composite. NZVI was
synthesized via the liquid-phase reduction method in which ferrous ions (Fe 2+ ) in an aqueous solution can be rapidly reduced to zero-valent ion (Fe 0 ) using sodium borohydride (NaBH 4 ) as a reducing agent as shown in Eq. (5). The optimum reaction time, reducing agent and iron salt were specified based on the results of the optimization of NZVI preparation process that were discussed in the results and discussion section.
In detail, 1 M of FeCl 2 .4H 2 O was added to 50 mL of ethanol. Subsequently, 4 M of NaBH 4 was mixed with 20 mL of distilled water and the formed solution was added dropwise (50-60 drops/min) to the ferrous solution during mixing in an aerobic condition (Without purging N 2 or Ar). The mixture color turned to black after adding the NaBH 4 solution affirming the reduction of ferrous ions to zero-valent iron and the mixture was further stirred for 10 min to secure the time required for the complete reduction of ferrous ions to Fe 0 . Then, the particles were collected using an external magnet and washed five times with distilled water and absolute ethanol to ensure the removal of reducing agent residuals. After washing, the nanoparticles were collected by centrifugation at 6000 rpm followed by drying at 50 °C overnight in a vacuum condition.
To prepare the NZVI based composite nanomaterials through supporting on either organic starch (NZVI-St) or inorganic silica gel (NZVI-Si), the same procedures for the preparation of the bare NZVI were followed beside the addition of ferrous solution and ethanol to 0.3 g of silica gel or starch and sonication for 30 min to affirm the dispersion of supporting materials before the addition of NaBH 4 solution.
Experimental procedures. The preparation of NZVI was conducted at reaction times of 10, 30, 60 and 120 min using different iron salts and reducing agents under vigorous stirring. The yield of the prepared NZVI samples was estimated and the performance of the prepared NZVI for the removal of acid blue-25 was evaluated at initial dye concentration of 50 mg L -1 , NZVI dose of 0.1 g, solution volume of 100 mL, contact time of 60 min and agitation rate of 200 rpm. The synthesis of NZVI supported on the surface of silica or starch was performed by adding the iron solution (FeCl 2 .4H 2 O) and ethanol on the supporting material. Then, the reducing agent (sodium borohydride (NaBH 4 )) was added to reduce iron ions to zero-valent iron supported on silica or starch. The adsorption integrated with chemical reduction of chromium Cr(VI) or acid blue-25 by the synthesized nanoparticles was conducted in a screw cap glass bottle. The bottle was filled with 100 mL of Cr(VI) or acid blue-25 dye solution and 0.1 g of the adsorbent was added at 20 °C. Then, the contaminated solution with the added adsorbent was placed in an incubator shaker for 120 min to remove Cr(VI) and 60 min for the removal of acid blue-25 (200 rpm). A stock solution of Cr(VI)) with a concentration of 1000 mg L -1 was prepared by dissolving 0.2829 g of K 2 Cr 2 O 7 in distilled water (1000 mL) and another stock solution of acid blue-25 dye was prepared.
The pH values were adjusted using 0.1 M of NaOH or HCl and the effect of pH (1-11) was investigated as well as the effect of initial Cr(VI) concentration (5-100 mg L −1 ) on the removal efficiency was studied at an adsorbent dose of 0. www.nature.com/scientificreports/ PO 4 3and humic acid on the removal efficacy were investigated. The concentration of anions and cations was 40 mg L −1 and the investigation of the effect of humic acid on the adsorption system was conducted using two concentrations of humic acid (5 mg L −1 and 20 mg L −1 ) at NZVI-Si dose of 0.1 g/100 mL, pH 1, initial Cr(VI) concentration of 25 mg L −1 , contact time of 120 min and temperature of 20 °C. The reusability study was conducted for six cycles at NZVI-Si dose of 0.1 g/100 mL, initial Cr(VI) concentration of 25 mg/L, pH 1, contact time of 120 min, and temperature of 20 °C. The particles were collected after each cycle using a magnet, and then they were washed with water and dried for successive use. The effect of adding H 2 O 2 (0.25-1 mM) on the removal performance was investigated at NZVI-Si dose of 0.1 g/100 mL, initial Cr(VI) concentration of 25 mg L −1 , contact time of 120 min and temperature of 20 °C. The samples were withdrawn during the contact time and centrifuged to separate the nanoparticles. Thereafter, measurement of Cr(VI) and acid blue-25 concentrations was performed using a UV spectrophotometer instrument (JASCO V-630) at 540 and 602 nm, respectively. Adsorption kinetics were studied during the time interval (0-120 min) at an initial Cr(VI) concentration of 25 mg L −1 and pH 1 using first-order, second-order, intraparticle diffusion, Elovich, power function, parabolic diffusion and Boyd kinetic models and investigation of the adsorption equilibrium was conducted at initial Cr(VI) concentrations of 5, 10, 25, 50 and 100 mg L −1 , pH 1 and time of 120 min using Langmuir, Freundlich, Dubinin-Radushkevish, Temkin, Harkins-Jura and Generalized isotherm models. Thermodynamic study was performed at different temperatures (293, 313, 328, 343 and 358 K). The equations and discussion of kinetic, thermodynamic and isotherm models were provided in the supplementary file (Text S1).
The removal percentage of Cr(VI) or acid blue-25 dye was calculated as shown in Eq. (6): where C o is the initial Cr(VI) or dye concentration (mg L −1 ) and C e is the Cr(VI) or dye concentration at equilibrium (mg L −1 ). The adsorption capacities of the synthesized materials were calculated using Eq. (7): where q e is the adsorption capacity of the synthesized materials at equilibrium (mg g -1 ); V is the solution volume (L); and m is the mass of synthesized nanomaterials (g).
Analytical methods. The diffraction planes of the synthesized nanomaterials were specified using X-ray diffraction analysis (XRD, Siemens model D-5000 diffractometer). The morphology, crystallinity, and chemical composition of the synthesized materials were investigated by performing transmission electron microscopy (TEM) coupled with energy dispersive X-ray spectroscopy (EDS), elemental mapping and selected area electron diffraction (SAED) (JEOL JEM-2100). The chemical bonds existing in the synthesized nanomaterials were specified using Fourier transform infrared spectroscopy (Shimadzu, FTIR-8400S). The surface area and pore size distribution of the synthesized nanomaterials were estimated using Belsorp-max automated apparatus (BEL Japan). Moreover, the chemical composition and oxidation states of the prepared nanoparticles were studied by performing an X-ray photoelectron spectroscopy analysis (Thermo-Fisher, USA). The magnetic characteristics of the synthesized nanomaterials were evaluated using vibrating sample magnetometer (VSM, Lake Shore-7410, USA), magnetic field up to 20 kOe and the sensitivity up to 1 μ emu. The point of zero charge was estimated using solid addition method as reported in our previous work 56 .

Conclusions
Sodium borohydride and ferrous chloride tetrahydrate were the optimum reducing agent and iron precursor, respectively for the preparation of pure NZVI after a reaction time of 10 min. The NZVI surface was modified it by supporting on starch or silica gel. The excellent support on modifiers was confirmed by various analyses such as TEM, EDS and XPS. The optimized NZVI can attain full removal of acid blue-25 dye after 60 min. NZVI-Si showed higher performance than pure NZVI and NZVI-St. The adsorption capacity was improved at elevated concentrations of Cr(VI) under acidic conditions. Adsorption kinetics, isotherms and thermodynamics studies indicated that the adsorption process was physical, favorable, spontaneous and endothermic. The existence of cations such as Mg 2+ , Zn 2+ and Cu 2+ and anions like NO 3 -, CO 3 2-, SO 4 2-, PO 4 3decreased the removal efficiency of Cr(VI). The addition of alow concentration of HA (5 mg L −1 ) can improve the removal efficiency compared to reduced removal performance at high HA concentration (10 mg L −1 ). The addition of H 2 O 2 with a concentration over 0.75 mM reduced the removal efficiency. The removal efficiencies were 96.8%, 93.67%, 90.1%, 86.7, 82.9% and 74.8% after six repetitive cycles using NZVI-Si. Reduction, adsorption and precipitation were the major removal Cr(VI) mechanisms onto the prepared NZVI-Si.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.