Exploring the Iron Oxide Functionalized Biobased Carbon-silica-polyethyleneimine Composites for Hexavalent Chromium Removal from Dilute Aqueous Solutions

: The contamination of water resources by toxic hexavalent chromium remains a challenge. In this study, amino-functionalized iron oxide biobased carbon-silica composites were prepared through co-precipitation of Fe(II) and Fe(III) over Macadamia activated carbon and explored as feasible adsorbents for the removal of Cr(VI) from dilute aqueous solutions. The energy dispersive spectroscopy (EDS) elemental analysis conﬁrmed the existence of Fe, Si, O, and C atoms, which form the backbone of the composite. The FTIR also showed the presence of Fe-O and Si-O-Si and Si-OH spectral bands, afﬁrming the backbone of the adsorbents. Cr(VI) adsorption efﬁciency (5.76 mg/g) was achieved at pH 1 when an initial concentration of 2.5 mg/L, contact time of 90 min, and dosage concentration of 1.7 g/L were used. The data were best described by the Langmuir adsorption model and pseudo-second-order rate model. ∆ G ◦ ( − 3 to − 12 kJ/mol) and ∆ H ◦ (46, 12 and 5 kJ/mol) values afﬁrmed that the adsorption of Cr(VI) was spontaneous and endothermic and dominated by chemical interactions.


Introduction
Chromium, similar to other heavy metals, is a menace to the environment as it persists and is non-biodegradable [1]. Chromium can be present in both anionic and cationic forms in aqueous media depending on the pH/Eh conditions. In cationic form, Cr(III) is the most stable oxidation state. At trace concentrations, Cr(III) is a micronutrient required by animals and humans [2]. The anionic chromium, Cr(VI), because of its carcinogenic and genotoxic properties, is classified as the 17th most toxic hazardous substance in the environment by the Agency for Toxic Substances and Disease Registry. The major sources of Cr(VI) in the environment are through anthropogenic activities because the kinetic reactions governing the conversion of natural Cr(III) to Cr(VI) are very slow [3,4]. In particular, Cr(VI) is a major element for the electroplating and stainless-steel production industries [5]. These industries use lots of water during their production, hence, the probability of Cr(VI) compounds leaching to the environment is too high [6]. Industrial wastewater containing chromium concentrations in the range of 0.5 to 270 mg/L have been reported [7]. The World Health Organization put stringent regulations to limit the amount of Cr(VI) in drinking water and surface water to 0.05 mg/L and 0.1 mg/L, respectively [6].
On one hand, current treatment techniques for the elimination of toxic Cr(VI) from water, which include ion exchange, membrane separation, and chemical precipitation, have limitations [8][9][10]. Membrane technology suffers from clogging and fouling [11], ion exchange is limited by high costs and fouling of resins [10], and chemical precipitation is Water 2021, 13, 3081 2 of 20 made difficult by the generation of toxic sludge which is costly to dispose of [12]. Adsorption on the other hand is highly favored due to simplicity in its design, ease of operation, low cost, low energy inputs, and use of cost-effective adsorbents [5,13]. However, its costeffectiveness is dependent on the adsorbent being used, and its regeneration capabilities.
Biobased carbons, such as activated carbons (ACs), are widely used in the industry on a large scale as adsorbents for water purification, catalytic processes, and separation processes [14]. Activated carbons are applied in adsorption processes because of their large surface area, porous structure, and low cost of preparation [14][15][16]. Although efficient adsorption capacities have been reported for the removal of Cr(VI) using powdered ACs [17], challenges of separating the adsorbent from the solution have risen concerns of spent carbons being potential secondary pollutants.
The use of iron-based adsorbents in the removal of hexavalent chromium from aqueous solutions has attracted considerable research interest because of their power to transform Cr(VI) into Cr(III) during removal [18,19]. However, the iron-based materials have a shortfall of being applicable in a limited pH range due to precipitation of iron, and to address that, SiO 2 is often coated on the materials for applicability in a wider pH range [18]. Inorganic silica (SiO 2 ) is widely employed as a supporting material due to its favorable characteristics of being stable under acidic conditions, its inertness to redox reaction, and its high content of hydroxyl groups on its surface which allows further functionalization, ease of binding, and attachment of ligands [20].
Amino functionalized materials such as m-phenylenediamine, polyethyleneimine (PEI), polypyrrole (PPy), CM-dextran, and chitosan have also been explored as supporting materials for nanoparticles owing to their low cost of preparations and operations, porous structures, environmental stability, and regeneration properties [21]. The high content of deprotonated amine groups is critical for the adsorption of trace metal ions through electrostatic interactions and hydrogen bonding [22]. However, metal ions such as Cr(VI) undergo reductive transformation upon contact with electron donors to yield less toxic Cr(III) ions [23]. Immobilization of the formed-Cr(III) is also possible because the amino groups act as Lewis bases and form complexes with the Cr(III) [21]. The latter prepares the adsorbents via the functionalization of PEI to be of particular interest to researchers because both the adsorption and reduction mechanisms can be explored for the removal of Cr(VI). As such, adsorbents prepared by functionalization of PEI on a substrate have been widely used in the literature for the removal of Cr(VI) [24][25][26][27][28][29]. However, none of these were prepared from silica supported on biobased carbons originating from Macadamia nutshells. This study aimed to synthesize the iron oxide functionalized biobased carbon-silica-polyethyleneimine composites for the abatement of Cr(VI) in dilute aqueous solutions. The one-variable-at-a-time approach was used to study the effect of time, temperature, mass, concentration, and pH in batch mode.

Preparation of Adsorbents
(i) Hydroxylation of AC Macadamia activated carbon (AC) (150-300 µm) was firstly pre-treated in 5 mol/L KOH to achieve more porosity and possibly introduce OH − groups on its surface for ease of binding and functionalization. The resultant product was then filtered and washed several times with ultrapure water. The product was further neutralized by washing with 1% acetic acid, then dried in an oven at 60 • C for 24 h and finally labeled AC-KOH.

(ii) Preparation of iron oxide modified AC (AC-Fe 3 O 4 )
Iron oxide was immobilized on the carbon surface through a co-precipitation method by dissolving a known amount of FeCl 3 ·6H 2 O and FeCl 2 ·4H 2 O salts in 300 mL ultrapure water and refluxed at 70 • C under a nitrogen flow for 2 h. About 5 g of AC-KOH was added to the Fe 2+ /Fe 3+ solution to obtain a 1:1 ratio and stirred for half an hour while maintaining the pH between 10 and 11 using 5 mol/L NaOH solution. The solid material was separated by filtration, then washed several times with ultrapure water and ethanol and dried at 60 • C overnight.

(iii) Preparation of AC-Fe 3 O 4 -SiO 2
The method reported by [30] was used to prepare the silica-coated biobased carbon. About 5 g of the recovered AC-Fe 3 O 4 was refluxed in a 500 mL water and ethanol solution of 4:1 ratio at 60 • C under nitrogen flow. Tetraethyl orthosilicate (TEOS) was added dropwise to the solution to form AC-Fe 3 O 4 -SiO 2 maintaining the pH in basic conditions (pH 9 to 11). The formed product AC-Fe 3 O 4 -SiO 2 was separated by filtration, washed several times with ultrapure water and ethanol, and dried at 60 • C overnight.
(iv) Preparation of amino-modified adsorbent (AC-Fe 3 O 4 -SiO 2 -PEI) To bind the PEI ligand on AC-Fe 3 O 4 -SiO 2 , the material was first hydroxylated using a method reported by [20] to increase the density of hydroxyl groups on the adsorbent surface. A 500 mL piranha solution (3:1 v/v H 2 SO 4 : H 2 O 2 ) was prepared and allowed to cool for 24 h. About 5 g of AC-Fe 3 O 4 -SiO 2 was then added to the cooled piranha solution and sonicated for 10 min under nitrogen flow. The solid sample was then separated by centrifuge and washed several times with ultrapure water and ethanol followed by drying at 60 • C for 12 h. To the solid sample, 10 mL of 3-glycidyloxypropyltrimethoxysilane (GPS) and 20 mL toluene were added and stirred at 70 • C for 8 h under nitrogen flow. The adsorbent was separated from the solution by a centrifuge, washed several times with ultrapure water and ethanol, and dried at 60 • C overnight. About 5 g of GPS-treated AC-Fe 3 O 4 -SiO 2 was refluxed in a 5% solution of branched polyethyleneimine (PEI) 100 mL at 70 • C for 12 h. The material was filtered and washed several times with ultrapure water and ethanol, then dried at 60 • C in an oven overnight. The resultant product was designated AC-Fe 3 O 4 -SiO 2 -PEI.
where C o is the analyte initial concentration (mg/L), C e is the final concentration (mg/L), M is the mass of the adsorbent (g) and V is the volume of the eluent (L).

Adsorption Isotherms
The Langmuir (Equation (3)), Freundlich (Equation (4)), and Dubinin-Radushrkevich (Equation (5)) adsorption isotherms were used to deduce the removal mechanisms, the adsorption affinity and intensity of adsorption for Cr(VI) onto AC-Fe 3 O 4 , AC-Fe 3 O 4 -SiO 2, and AC-Fe 3 O 4 -SiO 2 -PEI adsorbents. The equilibrium data were fitted into nonlinear models, as illustrated in Equations (3)-(5). The two well-known two-parameter models describe the adsorption of adsorbates onto adsorbents in two distinct phenomena. The Langmuir isotherm predicts the adsorption to proceed by a monolayer attachment of adsorbate on a single binding site that is on a homogeneous adsorbent surface. The Langmuir further assumes that no interactive bonding between adsorbates from adjacent sites occurs [31]. Conversely, the Freundlich isotherm allows for the interaction of adsorbates from adjacent active sites to take place, thus resulting in a heterogeneous surface with a multi-layered coverage. The Dubinin-Radushkevich isotherm can be used to confirm if the adsorption process is through physical or chemical interactions.
where b is the Langmuir isotherm constant (L/mg), q mL is the adsorption capacity at equilibrium (mg/g), K F is Freundlich constant (mg/g)(mg/L) −1/n , n is the Freundlich adsorption intensity exponent (dimensionless), q mDR is the Dubinin-Radushkevich adsorption capacity (mg/g), K DR is the Dubinin-Radushkevich isotherm constant (mol 2 /kJ 2 ), R is the gas constant (8.314 J/(mol K)) and T is temperature (K).  [21] and/or C-O-C [28], while those at 948.73 and 791.46 cm −1 were assigned to the asymmetric and symmetric stretches of Si-O-Si of the SiO 2 . The AC-Fe 3 O 4 -SiO 2 -PEI exhibited a peak at 3289 cm −1 assigned to the -OH stretch which overlaps with the NH-stretch of the amine groups of PEI and the peaks at 2932.34 and 2816.03 cm −1 were assigned to the asymmetric and symmetric -CH 2 -stretching vibrations of the ethyl groups on PEI branches [32]. The new peaks at 1658, 1570, and 1454 cm −1 in AC-Fe 3 O 4 -SiO 2 -PEI were assigned to the vibration of C=O (-CONH amide band I), C-N (amide band II), and N-H (amide band III) C-N stretch of amide, suggesting a covalent bonding of PEI to the epoxide ring [21,28,33]. The C-O-C peak of AC-Fe 3 O 4 -SiO 2 at 1054.45 cm −1 was replaced by two peaks of the primary amine groups of the PEI at 1092-1030 cm −1 , representing the stretching N-H bends [28]. This suggested that the C-O-C functionality participated in the bonding of PEI to AC-Fe 3 O 4 -SiO 2 [28]. Moreover, these findings revealed that the PEI was bonded on the SiO 2 [32]. Although the peak of the SiO 2 was replaced by the amine groups, the symmetric shifted Si-O-Si peak at 759.24 cm −1 was still visible together with the vibrating stretches of the Fe-O at 559.94 cm −1 [21].

Adsorbent Characterization
Water 2021, 13, x FOR PEER REVIEW 5 of 20 asymmetric and symmetric -CH2-stretching vibrations of the ethyl groups on PEI branches [32]. The new peaks at 1658, 1570, and 1454 cm −1 in AC-Fe3O4-SiO2-PEI were assigned to the vibration of C=O (-CONH amide band I), C-N (amide band II), and N-H (amide band III) C-N stretch of amide, suggesting a covalent bonding of PEI to the epoxide ring [21,28,33]. The C-O-C peak of AC-Fe3O4-SiO2 at 1054.45 cm −1 was replaced by two peaks of the primary amine groups of the PEI at 1092-1030 cm −1 , representing the stretching N-H bends [28]. This suggested that the C-O-C functionality participated in the bonding of PEI to AC-Fe3O4-SiO2 [28]. Moreover, these findings revealed that the PEI was bonded on the SiO2 [32]. Although the peak of the SiO2 was replaced by the amine groups, the symmetric shifted Si-O-Si peak at 759.24 cm −1 was still visible together with the vibrating stretches of the Fe-O at 559.94 cm −1 [21]. The elemental composition and specific surface area (SBET) analyses of the prepared adsorbents together with that of AC reference material are shown in Table 1. The inclusion of Fe3O4 nanoparticles on AC lowered the %C from 83.27% to 44.03% and the SBET surface area decreased from 546 to 386.89 m 2 /g. The decrease in %C was attributed to the presence of Fe, which is a much heavier atom than C, while the surface area reduction was probably due to the filling of AC pores by Fe3O4 nanoparticles or surface coverage through the formation of a semi-permeable layer of Fe3O4 on the AC surface. The atomic percent of C and H increased upon functionalization of AC-Fe3O4 by silica due to the attachment of CH3CH2-branches of the TEOS, but the surface area decreased to 234.51 m 2 /g. The %residual increased as functionalization occurred implying that other components except those of C, H, N, and S were present. The inclusion of an impervious layer of PEI on AC-Fe3O4-SiO2 drastically decreased the BET value from 234.51 to 0.15 m 2 /g. Elsewhere, it was The elemental composition and specific surface area (S BET ) analyses of the prepared adsorbents together with that of AC reference material are shown in Table 1. The inclusion of Fe 3 O 4 nanoparticles on AC lowered the %C from 83.27% to 44.03% and the S BET surface area decreased from 546 to 386.89 m 2 /g. The decrease in %C was attributed to the presence of Fe, which is a much heavier atom than C, while the surface area reduction was probably due to the filling of AC pores by Fe 3 O 4 nanoparticles or surface coverage through the formation of a semi-permeable layer of Fe 3 O 4 on the AC surface. The atomic percent of C and H increased upon functionalization of AC-Fe 3 O 4 by silica due to the attachment of CH 3 CH 2 -branches of the TEOS, but the surface area decreased to 234.51 m 2 /g. The %residual increased as functionalization occurred implying that other components except those of C, H, N, and S were present. The inclusion of an impervious layer of PEI on AC-Fe 3 O 4 -SiO 2 drastically decreased the BET value from 234.51 to 0.15 m 2 /g. Elsewhere, it was also observed that functionalization of AC with PEI led to a decrease in specific surface area from 942 to 32 m 2 /g due to surface passivation [34]. All the adsorbents had pore sizes ranging between 2 and 50 nm, signifying that the adsorbents had a mesoporous character as per the IUPAC categorization of pore sizes. The nitrogen adsorption-desorption plots are shown in Figure 2. These plots were used to determine the adsorption type and porosity structure of the adsorbent. AC-Fe 3 O 4 and AC-Fe 3 O 4 -SiO 2 exhibited similar-shaped graphs. The graphs were of Type II adsorption isotherms hinting at a monolayer or multi-layered adsorption process. The graphs seemed to run parallel to P/P o at relative pressures below 0.9, in particular the AC-Fe 3 O 4 -SiO 2 and AC-Fe 3 O 4 -SiO 2 -PEI, and then manifested a hysterical loop which suggested the formation of a microporous structure. AC-Fe 3 O 4 -SiO 2 -PEI demonstrated a Type IV adsorption which is associated with capillary condensation of the mesoporous structure. A sharp hysterical loop at relatively low pressures represents adsorption-desorption hysteresis and is most common in oxide gels and mesoporous carbon materials [15]. also observed that functionalization of AC with PEI led to a decrease in specific surface area from 942 to 32 m 2 /g due to surface passivation [34]. All the adsorbents had pore sizes ranging between 2 and 50 nm, signifying that the adsorbents had a mesoporous character as per the IUPAC categorization of pore sizes. The nitrogen adsorption-desorption plots are shown in Figure 2. These plots were used to determine the adsorption type and porosity structure of the adsorbent. AC-Fe3O4 and AC-Fe3O4-SiO2 exhibited similar-shaped graphs. The graphs were of Type II adsorption isotherms hinting at a monolayer or multi-layered adsorption process. The graphs seemed to run parallel to P/Po at relative pressures below 0.9, in particular the AC-Fe3O4-SiO2 and AC-Fe3O4-SiO2-PEI, and then manifested a hysterical loop which suggested the formation of a microporous structure. AC-Fe3O4-SiO2-PEI demonstrated a Type IV adsorption which is associated with capillary condensation of the mesoporous structure. A sharp hysterical loop at relatively low pressures represents adsorption-desorption hysteresis and is most common in oxide gels and mesoporous carbon materials [15]. The X-Ray-diffraction patterns for AC-Fe3O4, AC-Fe3O4-SiO2, and AC-Fe3O4-SiO2-PEI are presented in Figure 3. AC-Fe3O4 and AC-Fe3O4-SiO2 exhibited six diffraction peaks at  [16]. These diffraction peaks belong to the magnetite reference with a PDF card 19-0629 showing that Fe 3 O 4 was successfully attached to the activated carbon [35]. The peak at 2θ of 35 • in AC-Fe 3 O 4 -SiO 2 -PEI was related to Bragg's reflection of 311 for the magnetite. This shows that even though the surface of the AC was fully covered by the PEI, some Fe content was still present in the adsorbent. A large amorphous peak observed at 2θ = 20 • in AC-Fe 3 O 4 -SiO 2 -PEI was attributed to the amorphous structure of the PEI. 2θ = 30°, 35°, 43°, 54°, 57°, and 63° which were related to 220, 311, 400, 511 and 440 crystal planes according to Bragg's reflection [16]. These diffraction peaks belong to the magnetite reference with a PDF card 19-0629 showing that Fe3O4 was successfully attached to the activated carbon [35]. The peak at 2θ of 35° in AC-Fe3O4-SiO2-PEI was related to Bragg's reflection of 311 for the magnetite. This shows that even though the surface of the AC was fully covered by the PEI, some Fe content was still present in the adsorbent. A large amorphous peak observed at 2θ = 20° in AC-Fe3O4-SiO2-PEI was attributed to the amorphous structure of the PEI.  content from 17.1 wt.% to 1.7 wt.% was attributed to leaching and oxidization of Fe3O4 as functionalization occurred. No N content was recorded on the EDS probably due to its lower atomic mass. The TEM images of AC-Fe3O4, AC-Fe3O4-SiO2, and AC-Fe3O4-SiO2-PEI are illustrated in Figure 5a-c. The darker shades in TEM were attributed to the presence of Fe3O4 while the lighter ones were due to AC backbone and silica.   Thermogravimetric analysis (TGA) and derivative thermogravimetric analysis (DTA) were used to study the thermal stability and decomposition stages of AC-Fe3O4, AC-Fe3O4-SiO2, and AC-Fe3O4-SiO2-PEI. The results are demonstrated in Figure 6. The decomposition curves exhibit different patterns as distinctly shown by the derivatives curves. However, some commonalities were observed, such as the decomposition stage taking place between 40 and 100 °C credited to moisture loss. This was quantified to 5, 5, and 10% weight loss for AC-Fe3O4, AC-Fe3O4-SiO2, and AC-Fe3O4-SiO2-PEI, respectively. The second weight loss was observed between 100 and 800 °C with 5% weight loss for AC-Fe3O4, between 100 and 300 °C for AC-Fe3O4-SiO2 with 15% weight loss, and lastly between 300 and 600 °C for AC-Fe3O4-SiO2-PEI with 40% weight loss. This stage was attributed to the loss of volatile compounds such as CO2, CH4, CO, and decomposition of adsorbents [36]. More of this weight loss was expressed with AC-Fe3O4-SiO2 and AC-Fe3O4-SiO2-PEI as compared to AC-Fe3O4. This was attributed to longer chains of CH3CH2of the TEOS and those of the branched PEI. The last decomposition stage was attributed to the loss of lignin of the adsorbent which was more pronounced with AC-Fe3O4-SiO2 and AC-Fe3O4-SiO2-PEI than with AC-Fe3O4. Less weight loss observed by AC-Fe3O4 was attributed to the thermal stability of the Fe3O4 nanoparticles.  Figure 6. The decomposition curves exhibit different patterns as distinctly shown by the derivatives curves. However, some commonalities were observed, such as the decomposition stage taking place between 40 and 100 • C credited to moisture loss. This was quantified to 5, 5, and 10% weight loss for AC-Fe 3 O 4 , AC-Fe 3 O 4 -SiO 2, and AC-Fe 3 O 4 -SiO 2 -PEI, respectively. The second weight loss was observed between 100 and 800 • C with 5% weight loss for AC-Fe 3 O 4 , between 100 and 300 • C for AC-Fe 3 O 4 -SiO 2 with 15% weight loss, and lastly between 300 and 600 • C for AC-Fe 3 O 4 -SiO 2 -PEI with 40% weight loss. This stage was attributed to the loss of volatile compounds such as CO 2, CH 4 , CO, and decomposition of adsorbents [36]. More of this weight loss was expressed with AC-

Influence of Solution pH on Adsorption of Cr(VI)
The zeta potential data of AC-Fe 3 Figure 7. The isoelectric point (IEP) can be determined from the graph as the pH point at which the surface charge of the adsorbent is zero or rather surface had electrical neutrality. It can be observed in Figure 7 that the zeta potential (mV) magnitude decreased with increasing solution pH for all adsorbents. At pH values below the IEP, the surface of the adsorbent is mostly positively charged and above IEP the surface is negatively charged [5,7,37]. The pH IEP of AC-Fe 3 O 4 , AC-Fe 3 O 4 -SiO 2, and AC-Fe 3 O 4 -SiO 2 -PEI was found to be 4.95, 3.53, and 11.69, respectively. The high IEP (pH 11.69) observed for AC-Fe 3 O 4 -SiO 2 -PEI was attributed to the cationic characteristic of PEI [38,39].

Influence of Solution pH on Adsorption of Cr(VI)
The zeta potential data of AC-Fe3O4, AC-Fe3O4-SiO2, and AC-Fe3O4-SiO2-PEI measured at different solution pH values are illustrated in Figure 7. The isoelectric point (IEP) can be determined from the graph as the pH point at which the surface charge of the adsorbent is zero or rather surface had electrical neutrality. It can be observed in Figure 7 that the zeta potential (mV) magnitude decreased with increasing solution pH for all adsorbents. At pH values below the IEP, the surface of the adsorbent is mostly positively charged and above IEP the surface is negatively charged [5,7,37]. The pHIEP of AC-Fe3O4, AC-Fe3O4-SiO2, and AC-Fe3O4-SiO2-PEI was found to be 4.95, 3.53, and 11.69, respectively. The high IEP (pH 11.69) observed for AC-Fe3O4-SiO2-PEI was attributed to the cationic characteristic of PEI [38,39]. The pH of a solution affects the surface chemistry of adsorbents, ionization, and bioavailability of adsorbates [25,40]. Hence, the investigation of its influence on adsorption is crucial. The influence of solution pH was investigated over the range of pH 1-12 and the result depicting the adsorption capacities against the pH is referred to in Figure 8a-c. All adsorbents exhibited a similar trend where the adsorption capacity of Cr(VI) decreased as the pH of the solution increased. The highest removal efficiencies were obtained between pH 1 and 3 for all three adsorbents. Explicitly, the optimum pH for AC-Fe3O4 (4.75 mg/g) was pH 3, and for both AC-Fe3O4-SiO2 (4.47 mg/g) and AC-Fe3O4-SiO2-PEI (3.63 mg/g), it was pH 1. The low removal efficiencies at high pH levels were attributed to forces of repulsion between negative sites on the adsorbent's surface and the anionic Cr(VI) ions in solutions. The negative sites on adsorbent surfaces are caused by the presence of a high concentration of OH − ions resulting from the NaOH solution used for the pH adjustments. In the case of AC-Fe3O4-SiO2-PEI with high IEP at pH 11.69, the repulsion of Cr(VI) at high pHs could be due to the preferential binding of OH − groups to the cationic sites of adsorbents.
The stability of the oxyanion character of Cr(VI) in solution is dependent on the solution pH, thus, the predominating species in acidic media (pH 1-5) is mostly HCrO4 -, which is yielded by the hydrolysis of the dichromate ion (Cr2O7 2− ) [25]. The oxyanions (HCrO4 − , CrO4 2− and Cr2O7 2− ) at acidic conditions turn to become electrostatically attracted The pH of a solution affects the surface chemistry of adsorbents, ionization, and bioavailability of adsorbates [25,40]. Hence, the investigation of its influence on adsorption is crucial. The influence of solution pH was investigated over the range of pH 1-12 and the result depicting the adsorption capacities against the pH is referred to in Figure 8a-c. All adsorbents exhibited a similar trend where the adsorption capacity of Cr(VI) decreased as the pH of the solution increased. The highest removal efficiencies were obtained between pH 1 and 3 for all three adsorbents. Explicitly, the optimum pH for AC-Fe 3 O 4 (4.75 mg/g) was pH 3, and for both AC-Fe 3 O 4 -SiO 2 (4.47 mg/g) and AC-Fe 3 O 4 -SiO 2 -PEI (3.63 mg/g), it was pH 1. The low removal efficiencies at high pH levels were attributed to forces of repulsion between negative sites on the adsorbent's surface and the anionic Cr(VI) ions in solutions. The negative sites on adsorbent surfaces are caused by the presence of a high concentration of OH − ions resulting from the NaOH solution used for the pH adjustments. In the case of AC-Fe 3 O 4 -SiO 2 -PEI with high IEP at pH 11.69, the repulsion of Cr(VI) at high pHs could be due to the preferential binding of OH − groups to the cationic sites of adsorbents.
tion. At basic pH conditions, the amino groups of the PEI may attract the OH − through hydrogen bonding (R-NH2-----OH − ), thus rendering the surface of adsorbent more electronegative, resulting in the repulsion of chromate ions and less adsorption effectiveness [42]. Moreover, the competition for adsorption sites between OH − and CrO4 2− at higher pH (pH>6) has been reported by numerous researchers [7,40,45]. From pH 4 onwards, AC-Fe3O4 appears to have inferior performance compared to AC-Fe3O4-SiO2, probably due to preferential binding of OH − by the exposed Fe 2+ /Fe 3+ ions on AC-Fe3O4.

Adsorption Isotherms
The equilibrium sorption data of Cr(VI) on AC-Fe3O4 at 45 °C, AC-Fe3O4-SiO2 at 45 °C, and AC-Fe3O4-SiO2-PEI at 45 °C were fitted into Langmuir and Freundlich isotherm models to infer the type of interactions between the adsorbate and the adsorbents. The results are illustrated in Figure 9 and further summarized in Table 2. In Figure 9a the approximation line of Langmuir was closer to the experimental data points as compared to the Freundlich estimation. The coefficient of determination (R 2 ), residual standard error (RSE), and the adsorption capacity (qm) were used to infer which isotherm model best fitted the results. The R 2 value for AC-Fe3O4 was 0.828 in Langmuir and 0.737 in Freundlich. Similarly, the RSE values were 0.172 and 0.263 in Langmuir and Freundlich models, respectively. In this regard, the higher R 2 value and lower RSE in Langmuir concurred with The stability of the oxyanion character of Cr(VI) in solution is dependent on the solution pH, thus, the predominating species in acidic media (pH 1-5) is mostly HCrO 4 − , which is yielded by the hydrolysis of the dichromate ion (Cr 2 O 7 2− ) [25]. The oxyanions (HCrO 4 − , CrO 4 2− and Cr 2 O 7 2− ) at acidic conditions turn to become electrostatically attracted to the positively charged surface functional groups of the adsorbent [41]. The positive surface functional groups may be the results of protonation of NH 2 , COOH, OH-functional groups on the surface of the adsorbents to form NH 3 + , COOH 2+ , OH + [36,42]. Because these functional groups are rich in electrons, it is possible that during Cr(VI) adsorption, some portion of the Cr(VI) is transformed to Cr(III) through receiving electrons from, for example, NH 2 . In addition to the electrostatic attraction of Cr(VI) to the positively charged functional groups of the adsorbents at low acidic solutions, the reduction of Cr(VI) to Cr(III) due to the presence of electron-donating groups such as OH, NH 2 , SH, COOH, and the oxidation of Fe 2+ to Fe 3+ are the other possible mechanisms for Cr(VI) removal by the prepared adsorbents [43]. The higher removal exhibited by AC-Fe 3 O 4 compared to AC-Fe 3 O 4 -SiO 2 and AC-Fe 3 O 4 -SiO 2 -PEI at acidic conditions could be attributed to the effects of adsorption and reduction of Cr(VI) to Cr(III) by Fe 2+ ions on the surface of AC-Fe 3 O 4 [37,44]. The same can be said of silica-containing adsorbent (AC-Fe 3 O 4 -SiO 2 ). The presence of PEI could enhance the repulsion of Cr(VI) due to hydrophobicity of PEI, hence low adsorption. At basic pH conditions, the amino groups of the PEI may attract the OH − through hydrogen bonding (R-NH 2 --OH − ), thus rendering the surface of adsorbent more electronegative, resulting in the repulsion of chromate ions and less adsorption effectiveness [42]. Moreover, the competition for adsorption sites between OH − and CrO 4 2− at higher pH (pH>6) has been reported by numerous researchers [7,40,45]. From pH 4 onwards, AC-Fe 3 O 4 appears to have inferior performance compared to AC-Fe 3 O 4 -SiO 2 , probably due to preferential binding of OH − by the exposed Fe 2+ /Fe 3+ ions on AC-Fe 3 O 4 .

Adsorption Isotherms
The equilibrium sorption data of Cr(VI) on AC-Fe 3 O 4 at 45 • C, AC-Fe 3 O 4 -SiO 2 at 45 • C, and AC-Fe 3 O 4 -SiO 2 -PEI at 45 • C were fitted into Langmuir and Freundlich isotherm models to infer the type of interactions between the adsorbate and the adsorbents. The results are illustrated in Figure 9 and further summarized in Table 2. In Figure 9a the approximation line of Langmuir was closer to the experimental data points as compared to the Freundlich estimation. The coefficient of determination (R 2 ), residual standard error (RSE), and the adsorption capacity (q m ) were used to infer which isotherm model best fitted the results. The R 2 value for AC-Fe 3 O 4 was 0.828 in Langmuir and 0.737 in Freundlich.
Similarly, the RSE values were 0.172 and 0.263 in Langmuir and Freundlich models, respectively. In this regard, the higher R 2 value and lower RSE in Langmuir concurred with the visual observation that the data were best described by a monolayer, adsorption phenomenon of Langmuir. Additionally, AC-Fe 3 O 4 -SiO 2 data fitted to the Langmuir model better using the same reasoning applied above. In addition, the experimental (q e ) and estimated (q m ) adsorption capacity values for AC-Fe 3 O 4 -SiO 2 were closer to each other for each respective adsorbent. However, the data for AC-Fe 3 O 4 -SiO 2 -PEI seemed to follow a different trend, as depicted by its better fitting to Freundlich (R 2 = 0.957; low RSE of 0.043), which suggests the complex adsorption of Cr(VI) through a multilayer process.
Water 2021, 13, x FOR PEER REVIEW 13 of 20 the visual observation that the data were best described by a monolayer, adsorption phenomenon of Langmuir. Additionally, AC-Fe3O4-SiO2 data fitted to the Langmuir model better using the same reasoning applied above. In addition, the experimental (qe) and estimated (qm) adsorption capacity values for AC-Fe3O4-SiO2 were closer to each other for each respective adsorbent. However, the data for AC-Fe3O4-SiO2-PEI seemed to follow a different trend, as depicted by its better fitting to Freundlich (R 2 = 0.957; low RSE of 0.043), which suggests the complex adsorption of Cr(VI) through a multilayer process.

Adsorption Kinetics
Pseudo-first-order (PFO) (Equation (6)) and pseudo-second-order (PSO) (Equation (7)) kinetic rate models were used to study the reaction kinetics for chromium (VI) adsorption by the three adsorbents (AC-Fe 3 O 4, AC-Fe 3 O 4 -SiO 2, and AC-Fe 3 O 4 -SiO 2 -PEI). The Elovich model was also fitted to the kinetic data to confirm the chemisorption nature of the adsorption process expressed using Equation (8).
where q t (mg/g) is the adsorption capacity of the adsorbent at time t, k 1 (1/min) and k 2 (g/mg/min) are the PFO and PSO rates constants, respectively. In Equation (8), β (mg/g) denotes the desorption rate, α (mg/g.min) is the initial adsorption rate and t is the time (min). The data from fitting kinetics curves are referred to in Figure 10 and the numerical data is summarized in Table 3. The R 2 and RSE were used to decide on which model was best fitted to the experimental data. The high R 2 (0.949, 0.948, and 0.925) and low RSE (0.051, 0.052, and 0.075) values observed with the PSO model indicated its better fit compared to the PFO for AC-Fe 3 O 4 , AC-Fe 3 O 4 -SiO 2, and AC-Fe 3 O 4 -SiO 2 -PEI, respectively. The Elovich error terms (R 2 and RSE) values were in agreement with the PSO data and suggested that chemisorption was the dominant mode of Cr(VI) removal. The high initial adsorption rates (α) show that the adsorption process was slow, probably due to the reaction being of chemical nature and poor porosity observed in Table 1.

Performance Comparison
The Cr(VI) adsorption capacities of AC-Fe3O4, AC-Fe3O4-SiO2, AC-Fe3O4-SiO2-PEI were compared to similar materials found in the literature ( Table 4). The adsorbents were compared based on initial solution pH, initial Cr(VI) concentration, and Langmuir adsorption capacity qmL. Table 4 reveals that the adsorbents worked better at acidic conditions for the uptake of Cr(VI) from an aqueous solution. Numerous studies in the literature have reported the existence of electrostatic attraction forces between anionic Cr(VI) and positively charged adsorbent surface groups as the mode of the removal [33,46,47]. The initial concentration of Cr(VI), Co, being studied varies as depicted in Table 3, depending on the type of water targeted for the clean-up. Those targeting wastewater normally use a high Co concentration of Cr(VI), while those aiming for analysis of drinking water use low concentration. The selected iron oxide functionalized materials presented in Table  4 also reveal that producing these types of materials that possess high adsorption capacity values presents a challenge [48]. Based on this observation, it can be concluded that the    (Table 4). The adsorbents were compared based on initial solution pH, initial Cr(VI) concentration, and Langmuir adsorption capacity q mL . Table 4 reveals that the adsorbents worked better at acidic conditions for the uptake of Cr(VI) from an aqueous solution. Numerous studies in the literature have reported the existence of electrostatic attraction forces between anionic Cr(VI) and positively charged adsorbent surface groups as the mode of the removal [33,46,47]. The initial concentration of Cr(VI), C o , being studied varies as depicted in Table 3, depending on the type of water targeted for the clean-up. Those targeting wastewater normally use a high C o concentration of Cr(VI), while those aiming for analysis of drinking water use low concentration. The selected iron oxide functionalized materials presented in Table 4 also reveal that producing these types of materials that possess high adsorption capacity values presents a challenge [48]. Based on this observation, it can be concluded that the materials produced in this study performed fairly well in comparison to the literature. Hence, the current material extends the research on the type of adsorbents that can be evaluated for Cr(VI) adsorption. It has been reported that despite the high content of amino groups on PEI, high loading of PEI may lead to reduced surface areas, which may lead to lower adsorption capacities [28]. Lower adsorption capacities were observed with PEI-loaded adsorbents as compared to magnetite-loaded AC due to reduced surface areas.

Thermodynamic Studies
The thermodynamic studies for the adsorption of Cr(VI) by AC-Fe 3 O 4 , AC-Fe 3 O 4 -SiO 2, and AC-Fe 3 O 4 -SiO 2 -PEI adsorbents were evaluated at different temperatures (25, 35, and 45 • C) to predict the adsorption nature and spontaneity of the process. Parameters such as the entropy change (∆S o , J/(mol K)), enthalpy change (∆H o , J/mol), and Gibb's free energy (∆G o , J/mol) were calculated by relating Gibb's free energy to the equilibrium constant Kc using Equations (9)- (12).
where ∆G • is Gibb's free energy (J/mol), R is the universal gas constant (8.3144 J/mol K), K o c is the dimensionless equilibrium constant, Mw (HCrO4-) is the molecular weight of model pollutant HCrO 4 − , γ is the coefficient of activity (dimensionless) [52], b is the Langmuir isotherm constant (L/mg), [HCrO 4 − ] • is the standard concentration of the adsorbate (1 mol/L) and T is the temperature (K).
The ∆G o , ∆H o and ∆S o results are stipulated in Table 5. The ∆G o values were found to be negative for AC-Fe 3 O 4 , AC-Fe 3 O 4 -SiO 2 and AC-Fe 3 O 4 -SiO 2 -PEI at all temperatures investigated (298, 308, and 318 K). The values of ∆G o ranged between −17 and −36 kJ/mol, indicating a spontaneous adsorption process. The positive ∆H o values (871.45 kJ/mol for AC-Fe 3 O 4 -SiO 2 -PEI) indicated that the Cr(VI) adsorption process was endothermic and fell outside the 2-30 kJ/mol range which is characteristic of chemisorption [49]. In addition, the positive ∆H o values are associated with the proven endothermic reaction of Cr(VI) to Cr(III) reduction [33,53]. The adsorption of Cr(VI) by AC-Fe 3 O 4 -SiO 2 and AC-Fe 3 O 4 -SiO 2 -PEI was regarded as highly randomized, as the ∆S o values were above zero [54]. The adsorption of Cr(VI) using AC-Fe 3 O 4 -SiO 2 and AC-Fe 3 O 4 -SiO 2 -PEI adsorbents was dominated by entropy change as opposed to enthalpy change due to |T∆S • | > |∆H • | [55].

Conclusions
Amino functionalized activated carbon silica composites were prepared and explored as adsorbents for the removal of Cr(VI) from aqueous solutions. The The experimental data were best fitted to Langmuir adsorption isotherm for all three prepared adsorbents, predicting that the adsorption process occurred over a homogenous, mono-layered coverage. The kinetics data were best fitted to the pseudo-second-order rate model. According to the thermodynamics parameters, the adsorption of Cr(VI) by the composites was spontaneous and endothermic. The high enthalpy change energy suggested chemical transformation of adsorbed Cr(VI) into Cr(III), thus overruling the physisorption as the dominant mechanism but chemisorption. Additionally, the poor porosity of the main sorbent favored the chemisorption uptake of Cr(VI). In comparison to similar adsorbents, the AC-Fe 3 O 4 , AC-Fe 3 O 4 -SiO 2, and AC-Fe 3 O 4 -SiO 2 -PEI showed promising adsorption capacities and can be used as adsorbents for the removal of Cr(VI) from dilute aqueous solutions but adsorbents need improvement or testing using other pollutants like organic dyes.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.