Optimizing Fe0/Ni0/alginate beads as a stable and recoverable catalyst for removing highly toxic water contaminants: full-factorial design

Pollution by highly toxic contaminants poses a great threat to the aquatic environment and human life. Bimetallic materials have been proven to be efficient for the removal of such contaminants. In this study, the bimetallic Fe0/Ni0/alginate beads have been prepared using solvothermal technique followed by polymerization of alginate. Full-factorial design has been utilized to optimize the preparation conditions including the weight ratios of Fe and Ni, and time of the solvothermal process. The bimetal made from equal weights of Ni and Fe metals along with alginate has shown the best efficiency for the removal of 2,4,6-trichlorophenol (2,4,6-TCP, ~ 70%) as a model organic pollutant under acidic conditions (pH = 2) with minor contribution of alginate component in the removal process. Adsorption followed by reduction has been defined as the removal mechanism using GC/MS/MS chromatography in the case of 2,4,6-TCP. A mathematical model was established to describe the 2,4,6-TCP removal by the optimally prepared material. Differently, Pb2+ ions were removed predominantly by the alginate with minor effect of the bimetal (max. removal = 96.91% at pH 5 and after 2 h contact time). Accordingly, it was difficult to describe this process by a mathematical model. However, in both cases, the optimally prepared material has exhibited superior stability and reusability rather than the neat Fe0/Ni0 bimetal under the same experimental conditions. To sum up, the designed bimetallic Fe0/Ni0/alginate beads can be considered as an efficient catalyst for the removal of highly toxic pollutants from aqueous media.


Introduction
The population growth and consequent industrial, agriculture, and urbanization progress brought out a serious water scarcity and water pollution problems globally. Potentially toxic metals (PTMs) and chlorophenols (CPs) are two classes of pollutants that have been frequently detected in natural waters (Chang et al. 2015;Xie et al. 2021a, b) and have become of high concern due to the threat they pose to the public health and ecosystems (Ai et al. 2021;Kuganathan et al. 2021;Srikanth et al. 2021;Xie et al. 2021a, b;Xie et al. 2021a, b). The high toxicity, bio-recalcitrant, and accumulation in tissues are common features of these two classes of pollutants (Chang et al. 2015;Kuganathan et al. 2021;Srikanth et al. 2021;Xie et al. 2021a, b;Xie et al. 2021a, b) which make them priority environmental pollutants.
Lead (Pb) and 2,4,6-trichlorophenol (2,4,6-TCP) were selected as model pollutants for PTMs and CPs, respectively, because of their environmental impacts. Lead is a lifethreatening pollutant due to its high toxicity, non-biodegradability, accumulation in living beings, including human, high consumption (11.5 million tons in 2020 worldwide), and ubiquitous detection in environment (Nzediegwu et al. 2021;Xu et al. 2021). Chronic exposure to low levels of lead affects the nervous system and kidneys leading to their damage, while exposure to high levels can induce permanent learning difficulties, hearing loss, depression and antisocial behavior (Kuganathan et al. 2021). On the other hand, 2,4,6-TCP has been regarded as a significant problem in the recent decades owing to its toxicity, non-biodegradability, carcinogenicity, ease of adsorption on and penetration into the human skin and its negative impacts on the human nervous and respiratory systems (Chang et al. 2015;Radwan et al. 2017;Hwa et al. 2021;Karimipour et al. 2021). 2,4,6-TCP has been detected in surface water, groundwater and wastewater (Karimipour et al. 2021). Therefore, mitigating the contamination effect of lead and 2,4,6-TCP is urgently needed. Various methods were investigated for the removal of Pb 2+ including adsorption (Shi et al. 2018(Shi et al. , 2019Du et al. 2020;Geisse et al. 2020;Taghavi et al. 2021;Adebowale and Egbedina 2022) or degradation of 2,4,6-TCP by catalytic oxidation (Kruanak and Jarusutthirak 2019;Li et al. 2021).
In the last few decades, zero-valent iron nanoparticles (Fe 0 NPs) have been emerged as a versatile material that can mitigate the negative impacts of several classes of pollutants including halogenated organic compounds, azo and nitro compounds, potentially toxic metals, and oxyanions (Bezbaruah et al. 2011;Blundell and Owens 2021;Wei et al. 2021;Xie et al. 2021a, b). In spite of the numerous advantages of Fe 0 NPs such as large surface area, high and robust reducing ability, feasible in-situ operation, wide pollutant applicability, cost-effectiveness and eco-friendly nature (Bezbaruah et al. 2011;Angaru et al. 2021;Blundell and Owens 2021;Hemmat et al. 2021;Wei et al. 2021), there still some drawbacks. One of the main drawbacks is the rapid formation of iron hydroxide and oxide layer on its surface which suppresses the Fe 0 NPs catalytic efficiency (Blundell and Owens 2021;Hemmat et al. 2021;Pereira et al. 2021;Wei et al. 2021;Xie et al. 2021a, b). A route to overcome this drawback is to combine the Fe 0 NPs with a second metal to form a bimetallic nanocomposite. This route has been demonstrated to decrease passivation and enhance the reduction reactivity of Fe 0 NPs (Angaru et al. 2021;Hemmat et al. 2021;Xie et al. 2021a, b), however, the zero-valent bimetallic nanocomposite usually suffers from limited dispersibility due to its aggregation under normal water chemistry conditions as a result of the magnetic attraction and van der Waals forces. Such aggregation decreases the effective surface area, and consequently decreases remediation efficiency (Bezbaruah et al. 2011;Chang et al. 2015;Blundell and Owens 2021;Hemmat et al. 2021). Another two critical drawbacks of NPs are (i) reaching undesired receptors such as microorganisms and humans (Bezbaruah et al. 2011), and (ii) requirement of an additional separation process after the treatment process (Chang et al. 2015). Providentially, immobilizing the bimetal on a separable support decreases the bimetal oxidation, limits its mobility and aggregation, and negates the need for additional separation process (Chang et al. 2015;Hemmat et al. 2021). Among the various supports, the natural biopolymer alginate gained researcher's attention due to its non-toxicity, biodegradability, ease of handling, and low-cost (Bezbaruah et al. 2011;Pereira et al. 2021). Alginate beads enjoy significant porosity that enables the access of pollutants to the particles immobilized inside them. Furthermore, alginate can enhance the stability of immobilized particles without significant change in reactivity when compared to free particles (Pereira et al. 2021).
Therefore, in this study, bimetallic Fe 0 /Ni 0 nanocomposite encapsulated in calcium alginate beads was prepared via a solvothermal method followed by ionotropic gelation method and applied to remove 2,4,6-TCP and Pb 2+ cations. In this composite, nickel metal was selected owing to its low-cost and good corrosion stability (Angaru et al. 2021). Commonly, the removal efficiency depends mainly on the characteristics of the prepared materials, which in turn are affected by the preparation conditions. To avoid errors caused by the traditional optimization process in which the effect of one factor variation is examined while upholding all other factors fixed, a full-factorial design (FFD) providing the relative contribution of the examined preparation conditions in addition to their possible interactions with a minimum number of experiments was applied in this study. Thus, for the first time, the influences of time of the solvothermal process as well as the weight ratios of Fe 0 and Ni 0 on the characteristics and efficiency of Fe 0 /Ni 0 /alginate beads were analyzed using FFD of the experiments. The composition was optimized based on the obtained results. In addition, the characteristics of both bimetallic Fe 0 /Ni 0 and Fe 0 /Ni 0 /alginate beads were investigated thoroughly. Finally, the efficiency, stability and reusability of the bimetallic Fe 0 / Ni 0 and Fe 0 /Ni 0 /alginate beads were analyzed and compared.

Synthesis of zero-valent metals and bimetal
An accurately weighed amounts of FeCl 3 .6H 2 O, NiCl 2 .6H 2 O or their mixture was dissolved in 20 mL ethanol. Then, 5.0 g NaOH or 10 mL of NH 4 OH, and 10.0 mL N 2 H 4 .H 2 O were added in order to the former solution. After vigorous stirring to attain good homogeneity, the mixture was transferred to 150 mL stainless steel autoclave lined with Teflon. The autoclave was heated up to 80 °C and maintained at this temperature for a predetermined time. The formed black precipitate was magnetically separated, washed twice with copious amount of deoxygenated hot deionized water (DDW) followed by ethanol for several times, and finally dried under dry N 2 atmosphere.

Preparation of alginate beads and their composite with zero-valent metals and bimetals
The as-prepared zero-valent metals or bimetal were homogeneously suspended in 50 mL of 2% (w/v) sodium alginate solution and sonicated for 20 min. Then, the mixture was dropped into 100 mL of 2% (w/v) deoxygenated aqueous solution of CaCl 2. Alginate beads were formed once the alginate drops contacted the CaCl 2 solution. The beads were kept in the CaCl 2 solution for ∼9 h then washed with DDW and dried with absorbent paper for immediate use.

Characterization of the prepared materials
Crystal structures of the prepared materials were identified from the X-ray diffraction (XRD) diffractograms collected by PANalytical X'Pert Pro diffractometer with CuKα source (λ = 1.5406 Å). The morphology and elemental composition were investigated using field emission scanning electron microscopes (FE-SEM) equipped with energydispersive X-ray spectrometer (EDX) (JEOL 6400 F) and high-resolution transmission electron microscopy (HR-TEM, JEOL TEM-2100). The functional groups were assigned using attenuated total reflection Fourier transform infrared (ATR-FTIR) spectra obtained via Vertex 70 / Bruker Optics infrared spectrophotometer. The magnetic properties of the prepared materials were investigated using Riken Denshi BH-55 vibrating sample magnetometer (VSM) at room temperature. In addition, the textural analysis was determined by N 2 adsorption-desorption at 77 K using BELSORP-max surface analyzer.

Evaluating the efficiency of 2,4,6-TCP and Pb 2+ removal
Batch experiments were performed to evaluate the efficiency of the prepared materials for the removal of 2,4,6-TCP and Pb 2+ from aqueous medium. 3 g/L of the prepared material was agitated with 2,4,6-TCP or Pb 2+ using water bath shaker (DAIHAN MaXturdy TM, Korea) preset at 26 °C and 125 rpm. Samples were frequently collected and preserved to monitor the change of model contaminant concentration during the experiment. Effect of the medium initial pH on the removal efficiency was investigated by performing the experiment after adjusting the initial pH of the model contaminant solution at the desired value using a pH meter (OHAUS STARTER 3100).
where, C i and C t are the concentrations of the model substrate at time zero and t, respectively.

Experimental design: 2 3 FFD with center points
An FFD was applied to determine precisely the preparation variables that affect the removal of Pb 2+ and 2,4,6-TCP, and their interactions by the least possible number of experiments. In this design, the response was assumed to depend linearly on the change of a certain level from its minimum to its maximum value.
Generally, the 2 k factorial design can be simply expressed in terms of a regression model response enlightened by the following first-degree polynomial equation (Eq. 2): where, Y = response, β = regression model coefficients, x = coded variables, and ∈ = random error (Hribernik et al. 2009).
A pre-understanding of the process and its variables are required for attaining a realistic model. Accordingly, preliminary experiments were conducted to determine the main variables in preparing Fe 0 /Ni 0 /alginate beads. In this study, three level (k = 3) FFD with three center points and onereplicate have been used. The three investigated parameters were Fe 0 and Ni 0 loadings along with time of the solvothermal process. Selected variables were studied at the different levels of the other factors such as low (− 1), medium (0) and high (+ 1) levels. Besides, the calculated P-value was considered statistically significant when P < 0.05. The low levels and the high levels of the factors and the matrix containing the randomized runs are presented in Tables 1 and 2. The matrix was created using Design-Expert Version 13 (Stat-Ease Co.). Main effects, interaction effects, Pareto plot, normality plot and response surface plots were demonstrated and interpreted. Finally, a suitable model was generated in the response and the significance of the model was determined according to the statistical parameters.
The statistical analysis of the model was performed in the form of analysis of variance (ANOVA). This analysis included the Fisher's F-test (overall model significance), its associated probability P(F), correlation coefficient (CR), and regression coefficient (R 2 ) that measures the goodness of fit of the regression model. The analysis also includes the Student's t-value for the estimated coefficients and associated probabilities P(t) (Fang et al. 2012).

Characterization of the zero-valent metals and bimetal
The prepared zero-valent metals and bimetal were fully characterized before incorporation into alginate beads in order to ensure the successful preparation of the targeted materials and to evaluate the suitability of the preparation method.
Two alkaline reagents, namely; NH 4 OH and NaOH, have been investigated for the formation of zero-valent metals. According to the XRD patterns shown in Fig. S1 and its discussion in the supplementary materials, zero-valent metals cannot be produced using NH 4 OH. On the other hand, XRD patterns of the zero-valent metals and bimetal prepared using NaOH are depicted in Fig. 1a. It is obvious that pure well-crystallized Fe 0 , Ni 0 , and Fe 0 /Ni 0 were obtained as a single phase. Specifically, the single intense sharp peak appearing at 2θ = 44.83° in the XRD pattern of Fe 0 matches well with the ICDD #01-087-0721 of α-Fe 0 , while the two diffraction peaks at 2θ = 44.59° and 51.87° in the pattern  of Ni 0 are distinctive for pure face-centered cubic (fcc) Ni 0 (ICDD #00-003-1051). The XRD pattern of Fe 0 /Ni 0 bimetal exhibits two diffraction peaks at about 44.00° and 51.00° indicating the successful formation of Fe 0 /Ni 0 bimetal. Morphological examination of the Fe 0 sample under HR-TEM revealed the formation of cuboid crystals of Fe 0 of an average particle size of about 14 nm as shown in Fig. 1b. In Fig. 1c, the Ni 0 sample constitutes of aggregated sheets forming particles (average size ~ 68 nm) larger than those of Fe 0 . These sheets were not observed at all in the Fe 0 /Ni 0 sample (Fig. 1d), alternatively, large aggregates of about 127 nm composing of tiny particles were observed. Therefore, both Fe 0 and Ni 0 are nanoparticles, but their co-existence affected significantly the morphology and increased the particle size of the final product.
It is important to test the magnetic properties to demonstrate the possibility of using magnetic separation for the recovery of the prepared zero-valent metals and bimetal. Accordingly, the VSM hysteresis loops were collected. As shown in Fig. 2a, Fe 0 , Ni 0 , and Fe 0 /Ni 0 exhibit Langevin-like magnetic response for the applied magnetic field that is characteristic for superparamagnetic materials. This behavior is efficient for the water/wastewater application because the materials respond quickly and attracted to the applied magnetic field ( Figure S2 a and b); while, they are well dispersed and have high surface area in the absence of magnetic field (Fig. S2 c). Hence, the recovery and reuse of these materials are easy.
Deeper analysis of the collected hysteresis loops allowed extraction of the magnetic parameters depicted in Table 3. Fe 0 NPs exhibited the best magnetic properties of the highest saturation magnetization (Ms) and very low coercivity (Hc) and remanent magnetization (Mr). It is well established that decreasing the particle size is usually accompanied by surface spin disorder and the coercivity may approach zero if the crystal size is small enough (Tian et al. 2011;Kaliyamoorthy et al. 2016). Therefore, the observed low Hc (inset of Fig. 2a) can be ascribed to the small particle size of the materials. Interestingly, the Ms value for Fe 0 sample is seven times higher than that reported by Nicolás et al. (Nicolás et al. 2014) for the Fe 0 NPs prepared using NaBH 4 as reducing agent. Similarly, Ni 0 NPs exhibited Ms value that is double that of Ni NPs prepared previously (Chen et al. 2007;Sundar et al. 2014). Although the Ms value for Ni 0 sample is much lower than that for Fe 0 sample, it still has very good superparamagnetic properties. As expected, the The BET surface area of Fe 0 NPs, Ni 0 NPs, and Fe 0 /Ni 0 was found to be 23, 6, and 7 m 2 /g, respectively. These values agree well with the particle size observed in the HR-TEM images (Fig. 1b, c and d); since the particle size of the Fe 0 NPs were considerably smaller than Ni 0 NPs which, in turn, was comparable to that of Fe 0 /Ni 0 bimetal. The BET surface areas of Fe 0 NPs and Ni 0 NPs prepared in this study are considerably higher than those reported in the literature. The BET surface area of Fe 0 was reported to be 0.12 (Kim and Carraway 2000), 0.53 (Choi et al. 2008), 0.80 (Triszcz et al. 2009), and 2.02 m 2 /g (Triszcz et al. 2009); while that of Ni 0 was reported to be 1.07 (Choi et al. 2008), 0.17 (Xu et al. 2012), and 0.49 m 2 /g (Xu et al. 2012). Figure 2 b&c shows the N 2 adsorption-desorption curves and the NLDFT pore size distribution, respectively, of the zero-valent metals and bimetal. According to the IUPAC classification (Thommes et al. 2015), the N 2 adsorption-desorption curves of the prepared materials (Fig. 2b) belong to the reversible type II isotherm, which is distinctive for nonporous or microporous materials and indicates unlimited monolayer-multilayer  adsorption up to high relative pressure. The progressive curvature of the curves indicates considerable overlapping of monolayer coverage and the outset of multilayer adsorption. Figure 2b indicates that Fe 0 NPs are mainly microporous (pore width = 0.7 nm), while Ni 0 NPs, and Fe 0 /Ni 0 bimetal are mainly mesoporous (pore width = 3.3, and 3.5 nm, respectively). Correlating the results of the adsorption-desorption and NLDFT curves suggests that the prepared materials are nonporous with interparticle void spaces of different widths resulting from materials' aggregation.
To sum up, the NaOH/hydrazine hydrate method was efficient for the preparation of well crystalline magnetically separable superparamagnetic zero-valent iron NPs, zero-valent nickel NPs and zero-valent iron/nickel bimetal of high surface area.

Characterization of the bimetal/alginate beads
After the successful preparation of Fe 0 /Ni 0 bimetal, it was incorporated within the framework of alginate polymer to improve its adsorption properties, stability, recovery and reusability. Several Fe 0 /Ni 0 /alginate beads were prepared according to Table 2, some of them are shown in Figure S3.
The FTIR spectra shown in Fig. 3, indicate that pure alginate beads (std #1 in Table 2) and all prepared Fe 0 /Ni 0 /alginate beads have the same functional groups. Specifically, the absorption bands appearing at 3397 cm −1 and 2927 cm −1 can be attributed to the stretching vibration of hydroxyl groups in free water, whereas the bands at 1629 cm −1 , 1418 cm −1 and 1124 cm −1 are the bending and vibration absorption peaks of -COO-, C-O-C and -OH in alginate, respectively. Bending and vibration bands of -OH attached to Fe appeared at 1029 cm −1 , while that corresponding to the vibration of Fe-O appeared at 478 cm −1 . Figure 4 depicts the morphologies and elemental compositions of Fe 0 /Ni 0 bimetal, pure alginate beads, and the Fe 0 / Ni 0 /alginate beads (std order #7 in Table 2). Fe 0 /Ni 0 bimetal (Fig. 4a) appears as large aggregates composed of small homogenous spherical particles of about 187 nm diameter, which is consistent with HR-TEM images (Fig. 1a). The EDX analysis of Fe 0 /Ni 0 bimetal (Fig. 4b) proved that it is composed mainly of Fe and Ni in addition to trace amount of oxygen. As expected, pure alginate beads exhibit a smooth surface (Fig. 4c) and are rich of oxygen, carbon and calcium (Fig. 4d). Incorporating the Fe 0 /Ni 0 bimetal into the alginate beads increased the roughness of beads surface and induced the formation of rupture-like voids on its surface as presented in Fig. 4. Interestingly, it is obvious that Fe 0 /Ni 0 bimetal particles are distributed homogeneously inside and on the surface of the alginate beads and no aggregations could be observed (see inset of Fig. 4). The EDX analysis of Fe 0 /Ni 0 /alginate beads (Fig. 4f) indicates the presence of Fe, Ni, O, C, and Ca which proves the successful incorporation of the Fe 0 /Ni 0 bimetal into the alginate beads. Figure 5 displays the influences of changing the time of solvothermal process and loadings of Fe 0 and Ni 0 NPs on the removal of Pb 2+ and 2,4,6-TCP by the Fe 0 /Ni 0 /alginate beads prepared according to FFD. According to Fig. 5a, all the prepared materials, including the pure alginate beads, can efficiently remove Pb 2+ . The removal percentage (shown in Table 2) ranged between 82.47% for the pure alginate beads (Std order #2) and 97.94% for the Fe 0 /Ni 0 /alginate beads (Std order #8). The pure alginate beads have several surface functional groups as discussed in FTIR results. These functional groups endow the pure alginate beads with high affinity toward Pb 2+ adsorption as reported before (Khezri et al. 2016). However, the incorporation of the Fe 0 /Ni 0 bimetal improved the removal efficiency of Pb 2+ by 15.47% over the pure alginate beads which is significantly increased.

Removal of 2,4,6-TCP and Pb 2+ by the prepared Fe 0 / Ni 0 /alginate beads
On the other hand, Fig. 5b shows that the prepared Fe 0 / Ni 0 /alginate beads can efficiently remove 2,4,6-TCP as well, however, the removal efficiency was considerably lower and requires longer contact time relative to that of Pb 2+ (Fig. 5a) (3 h for Pb 2+ and 6 h for 2,4,6-TCP). The removal efficiency of 2,4,6-TCP shown in Table 2 varied significantly from 39.93% for the pure alginate beads (Std order #2) to 67.78% for Fe 0 /Ni 0 /alginate beads (Std order #7). This great improvement in 2,4,6-TCP removal by the Fe 0 /Ni 0 /alginate beads implies that Fe 0 /Ni 0 plays a key role in the removal process. After collecting the responses data, the significance of the FFD model was investigated by checking the probability (appropriate probability plots) of the main effects of the factors and their interaction terms as presented in Fig. 6. Generally, the effects distributed normally (i.e., fall along a straight line in the plot) are considered insignificant, while the significant effects show up as outliers on the normal probability plot (Hu et al. 2016;Campos et al. 2020). Accordingly, the variables under study and their interactions were found to have insignificant effects on Pb 2+ removal (Fig. 6a). On contrary, the studied variables were affecting the removal of 2,4,6-TCP significantly (Fig. 6b). Loadings of Ni 0 and Fe 0 were the most affecting factors, while, the interaction among the three variables was found to be insignificant. Based on these results further analysis of the FFD model was focused on 2,4,6-TCP.
The above results of 2,4,6-TCP were further confirmed from the Pareto chart (Fig. 7). The Pareto chart is a frequency histogram, which shows the amount of effect that each factor has on the response in decreasing order (Fang et al. 2012). In this chart, the factors possessing effects higher than the Bonferroni limit are considered very significant, those having effect lower than the t-limit are generally insignificant; while those laying between the two limits possess moderate importance (Liang et al. 2015). Consequently, order of the factorial effects of the studied factors and their interactions were found to be: C > B > A > AB ≈ AC > BC. Therefore, the factors themselves have strong influences on the 2,4,6-TCP removal, but their interactions have moderate effects on 2,4,6-TCP removal.

Interaction effects of the studied parameters
Variation in the efficiency of 2,4,6-TCP removal while changing each parameter from the low to the high levels is represented by the main plot effects. On the other hand, Fig. 7 Pareto chart illustrating the influence of each factor on the removal of TCP Fig. 8 Plots of the factors a main and b interaction effects for the 2,4,6-TCP removal efficiency dependence of the different levels of one parameter on the levels of other parameters yield the interaction plot effects. The interaction between any two parameters is considered to be effective only in case of having nonparallel plot lines of the two parameters (Moghazy et al. 2020). The main and interaction plots of the studied parameters are illustrated in Fig. 8. The main effect plots in Fig. 8a indicates obviously that the three studied parameters affect significantly 2,4,6-TCP removal. However, increasing the time of solvothermal process has a negative impact on 2,4,6-TCP removal, which is opposite to the other two parameters (Fe 0 and Ni 0 loadings). This behavior might be attributed to the fact that increasing the time of solvothermal process result in the formation of large particles of relatively lower surface area. The distribution of these particles within the framework of alginate is relatively difficult than that of fine particles. Accordingly, shorter time of solvothermal process is preferred in order to have homogeneous Fe 0 /Ni 0 /alginate beads of available active surface. Undeniably, this composite should include high loadings of Fe 0 /Ni 0 bimetal since the effect of Fe 0 and Ni 0 loadings on the 2,4,6-TCP removal is more pronounced than the effect of solvothermal time.
The interaction effect of a specific factor at the low level and the high level of another factor is illustrated in Fig. 8b. Generally, an interaction exists when the relationship exhibits nonparallel lines; while, parallel lines denote no relationship between the studied parameters (Morshed et al. 2020). According to the presented data, all parameters affect each other as indicated by the nonparallel-like lines in the interaction plots. In detail, the increase of either Fe 0 or Ni 0 to the maximum level (1 wt.%) with decreasing the time of solvothermal process enhances the 2,4,6-TCP removal. However, the interaction between time of solvothermal process and Ni 0 loading is slightly more significant than that of Fe 0 loading. Also, increasing both Fe 0 and Ni 0 loadings in the Fe 0 /Ni 0 /alginate beads is expected to improve the 2,4,6-TCP removal according to the interaction plot of Fe and Ni parameters. Therefore, the interaction plots are appropriate for investigating the preparation parameters of Fe 0 /Ni 0 /alginate beads.
The significance of the linear model suggested by Design Expert software was tested by depurating and evaluating the analysis of variance (ANOVA) method without performing transformation according to the Box-Cox plot presented in Fig. S4. Table 4 summarizes the ANOVA analysis of the effects of the selected factors and their interaction on the 2,4,6-TCP removal efficiency along with the model fit statistics. According to the presented data, the suggested model is significant as indicated by its high F-value (116.60) with only a 0.02% chance that this large F-value could occur due to noise. In addition, the calculated P-values were < 0.05 confirming the significance of the model terms (i.e., studied preparation parameters). In this case, A, B, C, AB, AC, and BC are the significant model terms. Looking at the fit statistics of the model reveals that the adjusted R 2 (0.9838) indicates the validity of this model. Furthermore, the adequate precision (a term measures the signal-to-noise ratio) is 38.89 that is much higher than the recommended value (adeq precision = 4), which supports that the suggested model can be used to navigate the design space. In addition, the calculated variance inflation factor (VIF) was found to be one at the initial data evaluation demonstrating no multicollinearity (Akinwande et al. 2015). Conclusively, the suggested model for predicting the 2,4,6-TCP removal by Fe 0 /Ni 0 /alginate beads can be represented in terms of codded and actual factors by Eqs. 3 and 4, respectively.
The validity of the suggested model has been evaluated using the normal probability plot of internally studentized residuals and the plot of the residuals versus predicted response values as presented in Fig. 9. Most of the internally   Fig. 9a indicating that normal pattern is valid for the regression residuals. The competence of the obtained model was also investigated by the plot of studentized residuals versus predicted values of the responses (Fig. 9b). The residuals were homogenously scattered above and below the x-axis in the range of -and + 3, demonstrating the adequacy of the model and there is no reason to suspect any violation (Hu et al. 2016).
The predictive ability of the model was assessed by comparing the actual values with the predicted values. The plot of this comparison ought to display a random scatter around a 45° line (Azizi Namaghi et al. 2015). As shown in Fig. 9c, all predicted values are very close to the experimental findings, revealing that 2,4,6-TCP removal by Fe 0 /Ni 0 /alginate beads could be predicted by the suggested FFD model.
The ideal levels for the significant factors were assessed in order to achieve a balance between the highest 2,4,6-TCP removal and lowest cost of Fe 0 /Ni 0 /alginate beads preparation. The three factors were adjusted to lay within the range given in Table 1 and the response surface plots are presented in Fig. 10. The highest 2,4,6-TCP removal was selected as the optimizing target. According to the obtained results, the highest TCP removal was 67.78% by combining the parameters: solvothermal time = 2 h, Fe 0 wt% = 1, and Ni 0 wt% = 1.

Effect of solution pH, reusability and stability of Fe 0 / Ni 0 /alginate beads studies
The medium pH can influence different aspects such as surface charge and solution chemistry (Zazouli et al. 2013). Accordingly, Pb 2+ and 2,4,6-TCP removal at different pH were tested. The Fe 0 /Ni 0 /alginate beads (Std order #7) and (Std order #8) were used in these experiments as they achieved the highest removal of Pb 2+ and 2,4,6-TCP, respectively. The collected data points are graphed in Fig. 11.
The removal of Pb 2+ was minimal at pH 2 and increased monotonically with increasing the solution pH up to 5 as indicated in Fig. 11a. Generally, adsorption of cationic metal ions occurs via one or more of three mechanisms, namely; electrostatic interaction, ion exchange, and chelation. At pH 2 the surface functional groups of the Fe 0 /Ni 0 /alginate beads become protonated and positively charged, consequently, electrostatic repulsion of the Pb 2+ cations and the protonated functional groups dominates which diminishes the removal percentage. What's more, the hydronium cations formed at highly acidic pH competes the Pb 2+ cations on the adsorption sites of the Fe 0 /Ni 0 /alginate beads. As the pH increased up to pH 5 the degree of protonation of surface functional groups and formation of competing hydronium cations decreases, hence, the removal percentage increased. Further increase of pH to 6 caused a decrease in Pb 2+ removal. Similar behavior has been observed for the adsorption of Pb 2+ cations on Fe 3 O 4 @TATS@ATA nanocomposite (Alqadami et al. 2020), activated carbons (Giraldo and Moreno-Piraján 2008), and calcium sulfate hemihydrate whiskers (Wang et al. 2017). It is known that Pb 2+ exists in different forms according to the solution pH. At pH 6 the concentration of Pb 2+ species decreases and the concentration of hydroxylated species (Pb(OH) + , Pb(OH) 0 2 , and Pb 3 (OH) 2− 4 ) increases resulting in decreasing the removal percentage.
In contrast to Pb 2+ , 2,4,6-TCP removal (Fig. 11b) was trivial under basic conditions and increased by decreasing the pH of the medium. Similar behavior was observed in case of TCP adsorption on Azolla filiculoides biomass (Ren et al. 2011;Zazouli et al. 2013) and insolubilized humic acid (Radwan et al. 2017). This behavior can be interpreted based on the FTIR results; the Fe 0 /Ni 0 /alginate beads enfold various functional groups. Consequently, the interactions between the beads surface and 2,4,6-TCP can be regarded as electron donor-acceptor interactions between the aromatic phenolic ring and the basic surface oxygens depending on the medium pH (Ren et al. 2011). The pKa of calcium alginate is 3 (Jodra and Mijangos 2001), meanwhile, 2,4,6-TCP is a weakly acidic compound of pKa = 5.99 (Ren et al. 2011). Therefore, at pH < pKa (i.e., acidic pH), both the functional groups on Fe 0 /Ni 0 /alginate beads and 2,4,6-TCP are in the non-ionized form leading to the predominance of hydrogen bonding (Zhu et al. 2018). Hence, only attractive forces occur between 2,4,6-TCP and the Fe 0 /Ni 0 /alginate beads. On contrary, under basic conditions 2,4,6-TCP dissociates forming phenolate anions, and the surface functional groups of Fe 0 /Ni 0 /alginate beads become either neutral or negatively charged. In this case, an electrostatic repulsion happens resulting in low 2,4,6-TCP removal. Besides, the phenolate anion possesses higher solubility in aqueous solution and forms stronger bonds with water molecules which decreases its removal.
To further understand the mechanism of 2,4,6-TCP removal, the probability of reduction of 2,4,6-TCP was investigated by analyzing samples collected during the reaction course using GC/MS/MS to monitor the byproduct formation. Phenol, 2-chlorophenol, and 2,4-dichlorophenol were identified as the reduction byproducts as illustrated in Fig. S5. The concentration of these compounds was increased with the reaction time revealing the reduction capability of Fe 0 /Ni 0 /alginate beads. Therefore, the removal mechanism of 2,4,6-TCP can be considered as adsorption followed by reduction.
The main objective of incorporating Fe 0 /Ni 0 bimetal inside the alginate matrix is to enable the easy recovery, and reuse as well as to enhance the stability of the bimetallic catalyst. To demonstrate this hypothesis, the reusability of the Fe 0 /Ni 0 /alginate beads and dissolution of iron and nickel from them were evaluated and compared to that of the Fe 0 /Ni 0 bimetal under identical experimental conditions. Figure 12a and b shows that in spite of the tremendous large particle size of the Fe 0 /Ni 0 /alginate beads, it can remove Pb 2+ and 2,4,6-TCP more efficiently than the Fe 0 / Ni 0 bimetal. This might be owed to the well dispersion of Fe 0 /Ni 0 bimetal within the alginate matrix. Hence, aggregation of Fe 0 /Ni 0 bimetal seems to be limited and more active sites are available for Pb 2+ and 2,4,6-TCP removal. Additionally, Table 2 shows the high removal efficiency of pure alginate beads toward Pb 2+ which could explain the significantly higher removal efficiency of the Fe 0 /Ni 0 /alginate beads relative to the Fe 0 /Ni 0 bimetal. Also, Fig. 12a and b indicates that the removal efficiency of Fe 0 /Ni 0 /alginate beads slightly decreased after the first cycle then remained nearly constant. In contrast, the removal efficiency of Fe 0 / Ni 0 bimetal decreased continuously. The amounts of Fe and Ni leached from Fe 0 /Ni 0 bimetal and Fe 0 /Ni 0 /alginate beads were measured at pH 5 and pH 2 (Fig. 12 c and d, respectively). Generally, the amount of Fe and Ni leached from the Fe 0 /Ni 0 bimetal was substantially higher than those leached from the Fe 0 /Ni 0 /alginate beads. All above-mentioned observations suggest that incorporating the Fe 0 /Ni 0 bimetal in alginate beads not only enhanced its removal efficiency, but also improved its reusability and stability. Noteworthy that Fe leaches more easily than Ni. Moreover, decreasing the solution pH and prolonging the contact time resulted in leaching of higher amounts Fe and Ni.

Conclusion
A superparamagnetic Fe 0 /Ni 0 of 127 nm diameter was prepared successfully using a solvothermal process in presence of hydrazine hydrate as the reducing agent. XRD analysis confirmed the formation of pure zero-valent bimetal without metal oxide impurities. Later on, an FFD enclosing weight ratios of Fe 0 and Ni 0 and time of solvothermal treatment variables was utilized to optimize the preparation conditions of Fe 0 /Ni 0 /alginate beads. The prepared beads were applied for the removal of the highly toxic water contaminants, 2,4,6-TCP and Pb 2+ . The selected variables were significant only in case of 2,4,6-TCP removal, however, the interaction effects of the variables were insignificant. Accordingly, a linear model was suggested to describe the 2,4,6-TCP. About 68% of 2,4,6-TCP could be removed by combining the optimum parameters: solvothermal time = 2 h, Fe 0 wt% = 1, and Ni 0 wt% = 1. Interestingly, the Fe 0 /Ni 0 /alginate beads prepared using the determined optimum conditions were more stable than neat Fe 0 /Ni 0 bimetal, and can be reused without losing much of their activity. On the other hand, Pb 2+ could be removed efficiently (95%) by the prepared Fe 0 /Ni 0 /alginate beads. Finally, the optimized Fe 0 /Ni 0 /alginate beads can be considered as a promising material for the decontamination of water and protection of environment.