Journal of Hazardous Materials 299 (2015) 316–324

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Fluoride adsorption by doped and un-doped magnetic ferrites CuCex Fe2-x O4 : Preparation, characterization, optimization and modeling for effectual remediation technologies Muhammad Abdur Rehman a,b,∗ , Ismail Yusoff a , Yatimah Alias b a b

Department of Geology, Faculty of Science, University of Malaya, Kuala Lumpur 50603, Malaysia Department of Chemistry, Faculty of Science, University of Malaya, 50603 Kuala Lumpur, Malaysia

h i g h l i g h t s • • • • •

A series of doped and un-doped magnetic ferrites adsorbents has been prepared by W/O micro-emulsion. The adsorption, electrochemical and magnetic properties of the adsorbents was compared. A fluoride adsorption model was developed based on central composite design of experiments. Effect of concomitants HCO3 −1 , SO4 2− , NO3 1− , Cl1− and Arsenic on fluoride adsorption. Response surface method for adsorption of fluoride.

a r t i c l e

i n f o

Article history: Received 24 April 2015 Received in revised form 5 June 2015 Accepted 19 June 2015 Available online 23 June 2015 Keywords: Fluorides Micro-emulsion CuFe2 O4 Adsorption Ion chromatography Ferrites

a b s t r a c t A series of doped and un-doped magnetic adsorbents CuCex Fe2-x O4 (x = 0.0–0.5) for fluoride were prepared with the micro-emulsion method. Fluoride adsorption was optimized for solution pH, temperature, contact time, and initial concentration and was monitored via normal phase ion chromatography (IC). The effect of concomitant anions was also explored to perform and simulate competitive fluoride adsorption in real water samples. Optimal adsorption was discovered by a simple quadratic model based on central composite design (CCD) and the response surface method (RSM). The adsorption, electrochemical and magnetic properties were compared between doped and un-doped ferrites. Doped ferrites (x = 0.1–0.5) were found to be superior to un-doped ferrites (x = 0) regarding the active sites, functional groups and fluoride adsorption. The characterization, optimization and application results of the doped ferrites indicated enhanced fluoride adsorption and easy separation with a simple magnet. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Fluoride is a necessary nutrient for dental health because it increases dental resistance to the attacks of acids formed by bacterial plaque and prevents the formation of cavities. However, when fluoride is present in public water supplies beyond certain limits, it is injurious to health [1–3]. The maximum allowable fluoride concentration level (MCL) of 1.5 mg/L has been set for drinking water by World Health Organization (WHO). The consumption of excess fluoride can cause impaired physical growth and intelligence, skeletal

∗ Corresponding author at: Department of Geology, Faculty of Science, University of Malaya, Kuala Lumpur 50603, Malaysia. E-mail address: [email protected] (M.A. Rehman). http://dx.doi.org/10.1016/j.jhazmat.2015.06.030 0304-3894/© 2015 Elsevier B.V. All rights reserved.

fluorosis, osteoporosis, arthritis, and even cancer in extreme cases [4–8]. Adsorption is considered the most efficient method due to its reduced cost, environmental friendliness and ease of operation. A number of adsorbents have been developed for fluoride adsorption [9–11]. The criteria for selecting a suitable adsorbent is based on the desirable properties, large specific surface area, larger number of active functional sites, pore size and volume, stability, and magnetic separation capability [12,13]. However, some obstacles remain in the development of adsorbents for fluoride adsorption, such as instability, matrix effects, mechanical strength and poor adsorption capacity. Therefore, more efforts should be devoted toward new classes of adsorbents with high de-fluoridation capacity. Recently, magnetic adsorbents with high adsorption capacity and environmental friendliness have been developed and

M.A. Rehman et al. / Journal of Hazardous Materials 299 (2015) 316–324

applied for the treatment of anionic contaminants, such as fluoride, nitrate, sulfate, arsenate and phosphate. These magnetic adsorbents include magnetic materials, such as iron, copper, cobalt and nickel oxides [14]. The flexible geometric structures of transition metal oxides enables the production of adsorbents with unique properties [15]. The cubic close packed geometry of CuFe2 O4 offer two types of sites, i.e., tetrahedral sites (A) and octahedral sites (B) [16]. The electrical and magnetic properties depend on the position of each metal on each site. These sites play a very critical role in the electromagnetic properties [17]. Ferrites exhibit the exchange of electrons during Fe–Fe interaction, causing polarizations and electrical conductivity [17]. However, doping rare earth cations changes the electrical, magnetic and physical properties [18,19]. Cerium (Ce) is a unique single electron system; it facilitates particle dispersion, which is a desirable and expected properties for an adsorbent. Cerium also promotes a larger number of active functional sites, surface area, and a larger pore size and pore volume. These improvements facilitate the adsorbate–adsorbent interactions because the adsorbate can easily diffuse and reach the active sites for adsorption. The design of experiments (DOE) strategy is an advantageous method compared with the use of the “OVAT strategy”, which incorporates one variable at a time [20,21]. DOE provides better insights into interactions among selected variables with a minimum number of experiments, thereby reducing the cost of research of new adsorbents [10,16,21]. Normally, multiple interactions of variables and non-linear behavior of adsorption processes result in a complex overall mechanism [11,14,17,20,22]. To understand such complex mechanisms, the response surface methodology (RSM) is a competitive modeling tool that can be used to analyze and present complex nonlinear relationships among variables either alone or in combination and their synergistic or opposed effects on adsorption [10,23–25]. The principle of RSM is to discover optimum system response with the help of polynomial equations [16,22]. From feasibility studies, significant experimental variables are co-varied over a pre-selected range to determine their interactive effects, finally combining all of the interactions to build a mathematical model [26]. The model is implemented in an adsorption system to achieve optimum performance through the use of a minimum number of experiments and by eradicating insignificant variables [27]. The ion chromatographic (IC) separation and detection method for quantifying anions provides multiple advantages. First, only a small amount of sample (20 ␮L) is required, which allows for the separation of concomitant ions, depending on their retention times (tR ). Second, inline ion suppressor controls the conductivity by suppressing the concomitant ions, thereby causing high conductivity. Third, a number of anions can be simultaneously detected and quantified in a single run due to the specific interaction and partition between the stationary phase and the mobile phase. Moreover, IC separation is a highly specific, selective, reproducible, robust and cost-effective method. In the study, magnetic CuCex Fe2-x O4 ferrites were prepared either with or without Ce doping to compare their capacity for fluoride adsorption. The characteristics were determined using the following techniques: X-ray diffraction (XRD), Fourier transform infrared (FTIR) spectroscopy, field emission scanning electron microscopy (FESEM), Brunauer–Emmett–Teller (BET) surface area, electrochemical impedance spectroscopy (EIS), Linear scan cyclic voltammetry (LSV) and vibrating sample magnetometer (VSM). A series of interactive experiments were conducted to study the effect of pH, temperature, adsorbent dose and fluoride initial concentration. The concomitant effects of HCO3 −1 , SO4 2− , NO3 1− , Cl1− and As were also studied to determined fluoride adsorption at different concentrations. Through this study, the model and mechanism of fluoride removal by magnetic ferrites were explained in detail.

317

2. Materials and method 2.1. Materials The analytical grade chemicals ferric nitrate heptahydrate Fe(NO3 )3. 9H2 O, copper nitrate Cu(NO3 )2 , cerium ammonium nitrate (NH4 )2 Ce (NO3 )6 , ammonium hydroxide NH4 OH, cyclohexane, 1-propanol, sodium carbonate Na2 CO3 , and sodium bicarbonate NaHCO3 were purchased and used in the form in which they were received from Sigma–Aldrich. A stock 1000 mg/L fluoride solution was prepared in deionized water with resistivity of 18 M cm−1 (Millipore, USA). Fluoride working solutions were prepared via the dilution method. Standard solutions of anions (chloride, nitrate, sulfate and bicarbonate) were prepared by dissolving sodium salts purchased from Sigma–Aldrich, Malaysia. Sulfuric acid (Merck, Malaysia) was used as a scavenger for the ion suppressor. 2.2. Instruments The IC system (Metrohm-850 Professional IC) equipped with a pump, column oven, sample injector, self-regenerating suppressor, Metrohm suppressor module (MSM), degasser and conductivity detector was used for the separation and detection of the anions. All separations using the IC system were performed on a column (Metrosep A supp 4 250/4.0 mm). The mobile phase containing 3.2 mM sodium carbonate, 1.7 mM sodium bicarbonate and 2% acetone was passed through the column at a flow rate of 0.70 mL/min. A sample volume of 20 ␮L was injected through a hexa-port valve sample injector, and the column temperature was maintained at 26 ◦ C. The ion suppressor was continuously regenerated with a constant flow of 0.5 mL/min of 50-mM sulfuric acid. Each analysis was performed within a period of 30 min, and the data were processed and analyzed using MagICNetTM 1.1. 2.3. Preparation of ferrite adsorbents A transparent micro-emulsion (1.0 dm3 ) was prepared through the sonication of a surfactant hexadecyl-trimethyl-ammonium bromide (CTAB) in cyclohexane, 1-propanol and water. Preweighed quantities of the respective aforementioned salts were separately dissolved and magnetically stirred at 60 ◦ C. Ammonium hydroxide was poured drop wise with continuous stirring to pH 8.0. Aging of the precipitates was completed by keeping the solution for 24 h at room temperature. Cleaning and removal of impurities was accomplished by successive washing with 80% methanol solution. Each ferrite was kept in an air electric furnace at 650 ◦ C for 3.0 h after heating at a rate of 5 ◦ C min−1 . Finally, the dried particles were ground into fine powder for the sequence experiments. 2.4. Characterization of the adsorbents The electric, magnetic and dielectric behavior of ferrites is a function of structural properties [28–30]. As a result, XRD, FTIR and FESEM were used for adsorbent characterization. The high penetrating power of X-rays provides important information regarding the structural properties of matter. The angle of diffraction and the intensity of the diffracted beam are both characteristics of a particular crystal structure. Every material is unique because of the sizes of its atoms, the arrangement of its atoms and the ability of each atom to scatter X-rays. Because no two atoms have exactly the same size and X-ray scattering ability, the intensities of the diffracted beam are unique for every material. This uniqueness helps identify the structure and determine the structural parameters of the material under study [16]. Multiphase structure, crystallite sizes and crystal strains of CuCex Fe2-x O4 were observed using an XRD instrument

318

M.A. Rehman et al. / Journal of Hazardous Materials 299 (2015) 316–324

Table 1 Retention times of the selected anions during ion chromatography. Anions

F−

Cl−

NO3 −

SO4 2−

tR

4.047

5.535

8.87

16.601

(PANanalytical’s Emprean) that was equipped with a monochromatic Cu-K␣ radiation source. The surface morphology and particle size distribution of the ferrites were studied using FESEM. FTIR analysis was performed on a PerkinElmer System 2000 series spectrophotometer (USA) over the spectral range of 4000–400 cm−1 . Magnetic hysteresis loops were recorded on a VSM. BET specific surface area analysis of the samples was performed on the basis of nitrogen adsorption desorption isotherms measured at 77 K using a BELSORP-max nitrogen adsorption apparatus (Japan Inc.). CV and EIS experiments were performed on an Autolab PGSTAT 302N instrument, and NOVA 1.10 software was used to analyze the experimental data. 2.5. Design of experiments Central composite design (CCD) was selected for the design of the fluoride adsorption experiments. Adsorption experiments were performed as suggested by the results of the software. The equilibrium fluoride adsorption capacity (qe mg g−1 ) was calculated using the following equation (Eq. (1)) [31]. qe =

(Co − Ce ) V m

(1)

where Co and Ce denote the initial and equilibrium adsorbate concentration (mg L−1 ), respectively, V is the total volume of solution in liters, and m is the mass of the adsorbent (g). Adsorption experiments were repeated to check for systematic variations to confirm the reproducibility of the results. The statistical models give a better insight into the complex, mutually influencing system of variables, the structure of the interactions, and the level of influence on the system overall response [20,21]. Moreover, statistical models also identify the most and least influential variables [27]. In this work, Eq. (2) presents the selected quadratic model [21,32]. k

k

i=1

i=1

k

k

qe = ˇo +  ˇi xi +  + ˇi xi2 +   ˇij xi xj + i = / j

(2)

i=1j=2

where ˇo is a constant coefficient,  represents the error, xi and xj are independent variables, and ˇi , ˇii , and ˇij represent coefficients of linear, quadratic, and interaction effects, respectively. Adsorption results against each experiment were analyzed and fitted into the quadratic models. The model was verified by statistical tests, e.g., analysis of variance (ANOVA), residuals analysis (RA), scaling residuals (SR) and prediction error sum of squares (PRESS) [21,23]. The verified and validated model was applied to optimize the four input variables and to maximize the output, i.e., fluoride adsorption capacity. 2.6. Determination of the anions The anions were identified and quantified based on their retention time (tR ) when performing ion chromatography (IC), as presented in Table 1. Anion calibration standards of 1, 5 and 10 mgL−1 were run to develop a calibration method. The tR values of standards and samples were compared and quantified with MagICNetTM 1.0. In suppressed ion chromatographic separation, an ion suppressor was inserted between the separating column and the detector to reduce the inherent conductivity of the eluent for improved anion

Fig. 1. XRD spectra [A] un-doped CuFe2 O4 , [B] doped CuCe0.1 Fe1.9 O4 , [C] CuCe0.2 Fe1.8 O4 , [D] CuCe0.3 Fe1.7 O4 , [E] CuCe0.4 Fe1.6 O4 , and [F] CuCe0.5 Fe1.5 O4 adsorbents for fluoride, annealed at 650 ◦ C for 3 h.

detection. The reactions involved with the chemical suppressor are represented by the following equation [33]. − − + + + R − SO− 3 H + Na HCO3 ↔ R − SO3 Na + H2 O + CO2 + R − SO− 3H

+

−

+ Na + F ↔

+ R − SO− 3 Na

+

+H +F

(3)

−

(4)

Na+

H+

ions coming in mobile phase are replaced with ions, quenching the conductivity of the background electrolyte, as represented by Eq. (3). The anion in sample, e.g., F− remains unaffected by the suppressor, detected and quantified by IC system, as represented by Eq. (4). 3. Results and discussion 3.1. XRD phase analysis Stacked XRD patterns of CuCex Fe2-x O4 (x = 0.0–0.5) are shown in Fig. 1. In the absence of cerium (x = 0.0), all of the identified peaks matched with copper ferrite (CuFe2 O4 , Reference code # 96901-2842), which indicated the formation of spinel-type CuFe2 O4 [34]. After the addition of Ce (x = 0.1), some additional peaks of low intensity appeared in the XRD spectra corresponding to CeO2 (CeO2 , Reference code # 96-900-9009) [35]. A systematic increase in dopant concentration (x) resulted in a gradual increase in the intensity of the CeO2 peaks. Moreover, we observed a decrease in the (3 1 1) peak. The addition of cerium into the pristine copper ferrite exhibited a distinct change in the percentages of the respective phases. The average crystallite size (nm) was calculated using Scherer’s formula (Table 2). A comparison among the crystallite sizes reveals a systematic decrease of crystallite size with a gradual increase in dopant (Ce). Table 2 Crystal sizes and the micro-strain and BET properties of a series of copper ferrites CuCex Fe2-x O4 (x = 0–0.5). Property 2

BET Surface area (m /g) Crystallite sizes D (nm) Average pore size (nm) Pore volume (cm3 /g) Lattice strain  (%)

x = 0.0

x = 0.1

x = 0.2

x = 0.3

x = 0.4

x = 0.5

70 72 38 0.44 0.17

75 50 34 0.4 0.25

79 34 30 0.38 0.38

83 30 25 0.34 0.42

88 20 20 0.29 0.64

92 18 17 0.26 0.69

M.A. Rehman et al. / Journal of Hazardous Materials 299 (2015) 316–324

319

Fig. 2. (A) Effect of the applied potential on the current response and (B) Nyquist diagrams of (x = 0) un-doped CuFe2 O4 (x = 0.1), doped CuCe0.1 Fe1.9 O4 (x = 0.2), CuCe0.2 Fe1.8 O4 (x = 0.3), CuCe0.3 Fe1.7 O4 (x = 0.4), and CuCe0.4 Fe1.6 O4 (x = 0.4) CuCe0.5 Fe1.5 O4 in 0.1 M KCl solution containing 1.0 mM Fe(CN)6 3−/4− (1:1).

Doping copper ferrites with Ce reduced the crystallite sizes of pristine ferrite. This decrease was attributed to the formation of secondary phases [36]. This decrease was also supported by a relative increase in the lattice parameters. Annealing at elevated temperature forced Ce3+ ions to migrate into the crystal structure, the interstitial regions, or the grain boundary sites. A large ionic radius of Ce3+ (1.020 Å) compared with that of Fe3+ (0.645 Å) offered a relative space hindrance to the competitive vacant crystal or interstitial sites in copper ferrites [37]. Consequently, the Ce3+ ions were oxidized to Ce4+ at this temperature and appeared as a distinct phase in the XRD spectrum (Fig. 1). The surface properties, surface area, and pore size and volume of the adsorbents were analyzed via BET. Table 2 indicates that a gradual increase in dopant concentration (x) corresponds to an increase in the BET surface area. Cerium oxide likely inhibits the growth of larger particles because of its high dispersion [18]. The BET results indicate that during adsorption, more fluoride can interact with the available active sites that are created by doping in these magnetic ferrites.

3.2. Electrochemical properties The electrochemical properties of ferrite modified electrodes were studied using electrochemical impedance spectroscopy (EIS). EIS measures the charge transfer resistance (Rct ) and separation efficiency between the holes and electrons. Rct is a critical factor in the charge separation efficiency. Fig. 2 shows a typical Nyquist plot of the ferrite. The arc radius of the Nyquist plot reflects the interface layer resistance at the surface of the electrode. A smaller arc radius indicates a higher charge transfer efficiency. The bare electrode response of a semicircle of larger radius compared with that of the ferrites modified electrodes due to the large value of Rct at the electrode–electrolyte interface. The arc radius decreases from x = 0 to x = 0.2, which indicates higher charge resistance with higher Ce content. According to the hopping mechanism, changes in the mobility of the charge carriers lead to the conduction of current by hopping from one iron atom to the next [38]. When cerium is doped in the place of iron, the resistivity of the sample is found to increases. This behavior can be explained by the greater resistivity of cerium (74.4 × 10−8  m) compared with iron (8.57 × 10−8  m) [39]. Further increase in the Ce dopant in the range of x = 0.3–0.5 again reveals that larger semicircles are associated with the higher amounts of Rct .

Fig. 3. FTIR overlay spectra of (A) un-doped CuFe2 O4 , (B) doped CuCe0.1 Fe1.9 O4 , (C) CuCe0.2 Fe1.8 O4 , (D) CuCe0.3 Fe1.7 O4 , (E) CuCe0.4 Fe1.6 O4 , and (F) CuCe0.5 Fe1.5 O4 .

LSV analysis was used to study the electrochemical properties of the adsorbents. Ferrite with x = 0.2 exhibited the maximum photocurrent response, followed by x = 0.1 (Fig. 2). The photoelectrochemical activity is well known to be determined by the light harvesting capacity and the separation of electron-hole pairs [40]. Generally, a high value of the photocurrent indicates that the sample has a strong ability to generate and transfer the photoexcited charger carrier under irradiation. The EIS and LSV results confirm that the ferrite adsorbents can be used as photo-catalysts and photo-electrode materials in addition to their excellent adsorbent properties. 3.3. FTIR spectroscopy of spinel structures The inter-metallic ionic mode of vibration in FTIR spectroscopy analysis provides information regarding the ions involved in the formation of the crystal lattice [40,41]. The characteristic peaks of the spinel structure observed in the range of 560–580 cm−1 are attributed to Fe O stretching vibration at the tetrahedral site

320

M.A. Rehman et al. / Journal of Hazardous Materials 299 (2015) 316–324

Fig. 4. FESEM images of (A) un-doped CuFe2 O4 , (B) doped CuCe0.1 Fe1.9 O4 , (C) CuCe0.2 Fe1.8 O4 , (D) CuCe0.3 Fe1.7 O4 , (E) CuCe0.4 Fe1.6 O4 , and (F) CuCe0.5 Fe1.5 O4 prepared in micro-emulsion (w = 15).

(Fig. 3). FTIR functional groups were compared in pristine and Ce-doped ferrites. All Ce-doped ferrites have distinct peaks near 850 cm−1 and 1440 cm−1 . The additional functionality is thought to be responsible for the electrostatic interaction with the adsorbate to enhance the overall fluoride adsorption. The hydroxyl and sulfate groups are clearly involved in the fluoride adsorption. The abundance of these groups in the Ce-doped ferrite may be the main reason for the high fluoride uptake. 3.4. FESEM results Fig. 4 shows FESEM images of the respective ferrites (x = 0.0–0.5). When x = 0.0, we observe normally sized CuFe2 O4 nano-crystals (∼74 nm) with uniform morphology along with some agglomerations of larger size (∼180 nm). This behavior was due to the large specific surface of the fine particles for van der Waals interactions. From a nanochemistry perspective, adjacent nanoparticles that share a common crystallographic orientation may collide with each other, which leads to coalescence [42] because adjacent primary particles attach to each other and form aggregates. In addition, due to the nanometer-scale sizes, these particles continue to collide and coalesce to form even larger particles. This phenomenon may eventually cause agglomeration in the fabricated ferrites. The agglomeration is affected by various processing parameters, such as speed, intensity and mixing time. Relatively smaller sizes were observed with further addition of ceria (x = 0.1–0.5). Moreover, the addition of ceria particles was found to increase the surface area by allowing for maximum dispersion and retarding large grain growth. These results complement the XRD results and provide evidence of an increased surface area, which can facilitate better fluoride adsorption. These results also agree with those of other studies, indicating that doping lanthanides hinders particle grain growth [36,43]. The observed incremental decrease in grain morphology via the systematic doping of ceria promotes fluoride adsorption. 3.5. Magnetic properties The magnetic hysteresis loops of CuCex Fe2-x O4 (x = 0.0–0.5) are shown in Fig. 5. The value of the squareness ratio (SQ) was calculated by dividing the value of the magnetic remanence (Mr) by

Fig. 5. Magnetic properties of (A) un-doped CuFe2 O4 , (B) doped CuCe0.1 Fe1.9 O4 , (C) CuCe0.2 Fe1.8 O4 , (D) CuCe0.3 Fe1.7 O4 , (E) CuCe0.4 Fe1.6 O4 , and (F) CuCe0.5 Fe1.5 O4 .

the magnetic saturation (Ms). The SQ value reflects the magnetocrystalline anisotropy and the super-exchange interactions of the ferrite series [44]. CuFe2 O4 exhibits a SQ value of 0.13 (x = 0.0), but after Ce doping, this value increases in the cases of x = 0.1 and x = 0.2. Note that the incorporation of Ce reduced the magneto-static interactions [45]. However, the further addition of Ce exhibits a regular declining trend from x = 0.3 to x = 0.5. This phenomenon indicates that the local canting effect is taking control over the surface canting effects associated with crystallite size and coercivity [46]. The relative change in Mr and Ms with Ce content has exactly the same trend, as expected from the SQ values for each member of the series Table 3. From Table 3, the Hc value is clearly increasing with the amount of dopant (x from x = 0.0 to x = 0.2), and further doping results in a regular decline (from x = 0.3; the maximum decrease occurs when x = 0.5), which is attributed to extrinsic and intrinsic factors. The extrinsic factors depend on the density, morphology and grain size [47]. The intrinsic factors, such as relative occupation of sub-lattice sites, contributed to the magnetism in the nano-ferrites [48]. The shape and width of the hysteresis loops depend on the method of preparation, composition, and distribution of the metal cations at

M.A. Rehman et al. / Journal of Hazardous Materials 299 (2015) 316–324

321

Table 3 Magnetic properties of the CuCex Fe2-x O4 adsorbents by VSM. Magnetic parameters

x = 0.0

x = 0.1

x = 0.2

x = 0.3

x = 0.4

x = 0.5

Coercivity (Hc)/T Saturation magnetization (Ms) (emu/g) Retentivity (Mr) (emu/g) Squareness (SQ)

0.0069 30.137 4.336 0.144

0.0091 25.307 3.889 0.154

0.0104 15.531 3.093 0.199

0.01 13 2.5 0.192

0.0087 11.646 2.106 0.181

0.0037 8.885 0.823 0.093

the tetrahedral and octahedral sites, the spin canting and the crystallite size [41]. The values of Mr and Ms exhibit a regular decline. This behavior could be explained by the reduction of D from 73 nm to 18 nm (Table 1) and the respective increase in the Ce content at the B-sites. Because the magnetic moment of Ce is less than that of Fe, theoretically, increasing the Ce content will reduce the Ms monotonically [46,49]. Generally, rare-earth elements have a strong anisotropic subsystem with a sufficiently high anisotropic exchange constant for iron [50]. Thermal excitation breaks the Kondo state, which causes a gradual recovery of magnetic moments [51]. Ce3+ magnetic moments are governed by direct or indirect exchange interaction, namely RKKY interaction mediated by conduction electrons [52]. Sufficiently strong exchange interaction produces a magnetically ordered ground state. In the present state, ferrite with x = 0.4 and x = 0.5 have such an interaction, as shown in Fig. 5. However, in the case of x = 0.3, a dominance of the Kondo interaction occurs, which results in a nonmagnetic ground state. A variation in the coercive field (Hc ) directly depends on the surface spin disorder [53]. The decrease in Hc of the present series affirmed the reduced surface spin disorder. The magnetic properties of the ferrites are controlled by Fe–Fe interaction via the spin

coupling of 3d valance electrons [54]. The present decrease in coercivity is attributed to morphology-increased magnetic moments and domain wall migration [55]. The lower values of Mr and Ms are due to the oxidation of cerium into diamagnetic states [56]. However, the present decrease in magnetic character has little effect on the fluoride adsorption capacity, but it affects the magnetic separation capability due to the change in its morphology, crystal structure and related electromagnetic properties. 3.6. Fluoride adsorption model The fluoride adsorption model is given by Eq. (5). The ANOVA test was applied to obtain the coefficients for each variable. A positive value of a coefficient symbolizes a synergistic effect, whereas a negative value indicates an opposite effect [57]. qe = +1039.71 + 74.29 × X1 + 25.31 × X2 + 36.99 × X3 + 8.31 × X4 − 0.32 × X1 X2 − 0.261 × X1 X3 − 0.21X1 X4 −0.23X2 X3 − 0.10 × X2 X4 − 0.13X3 X4 − 4.49 × X12 −0.93 × X22 − 0.44 × X32 − 0.03 × X42

(5)

Fig. 6. Response surfaces showing the effects of two variables on fluoride adsorption (A) pH and F−1 (mg L−1 ), (B) pH and Ads. Dose (mg), (C) F−1 (mg L−1 ) and Ads. Dose (mg), and (D) Temp. (◦ C) and Ads. Dose (mg).

322

M.A. Rehman et al. / Journal of Hazardous Materials 299 (2015) 316–324

Fig. 7. Effect of some concomitant anions on fluoride adsorption, m = 0.1 gL−1 , [F]0 = 10 mgL−1 , pH 7.0: (A) un-doped CuFe2 O4 , (B) doped CuCe0.2 Fe1.8 O4 , (C) CuCe0.3 Fe1.7 O4 , and (D) CuCe0.5 Fe1.5 O4 .

where X1 –X4 are the coded variables for pH, F−1 (mgL−1 ), Temp. (◦ C) and ferrite dose (mg), respectively. The adequacy of the model was ensured and tested by Fischer variation (F-value), probability values (p-value) and the correlation coefficient (R2 ). The lack of fit value of the model of 0.74 was insignificant relative to the pure error. The predicted R2 value of 0.9988 was in reasonable agreement with an adjusted R2 of 0.9994. The adequate precision of the model is a measure of the signal-tonoise ratio. A ratio greater than 4 is desirable. In the present case, a ratio of 194.521 indicates an adequate signal. The effect of variables was analyzed by comparing the values of the linear, quadratic and interaction terms. The pH linear term (X1 ) with an F-value of 2112.63 indicates the significant role of pH in fluoride adsorption. However, the pH quadratic term (X1 2 ) has an F-value of 2239.4, which is greater than that of the linear term and indicates a positive role of pH increase in the selected range on fluoride adsorption. An optimum fluoride adsorption was achieved at neutral pH, as shown in Fig. 6 (RSM A–C). According to the speciation of fluoride, it exists as a HF species instead of ionic (F−1 ), which may be the reason for the reduced adsorption under acidic conditions [58]. According to the electrostatic interaction mechanism, a relatively stronger attraction is obtained at a pH below 6.12. However, pH is not the only factor that controls the fluoride adsorption. Similarly, the linear and quadratic terms of the fluoride initial concentration are significant during fluoride adsorption. An increase in the fluoride initial concentration improved the rate of adsorption. However, increasing the fluoride concentration beyond the optimum reduces the fluo-

ride adsorption, as shown in Fig. 6 (RSM A and D). The linear and quadratic terms of temperature (X3 and X3 2 ) indicate a decrease in fluoride adsorption at elevated temperatures. The most prominent variable was adsorbent dose (X4 ), which has a significant influence on fluoride adsorption. This influence can be explained by the adsorbent dose being directly related with the number of active sites for fluoride adsorption. 3.7. Effect of competing anions Concomitant anions, such as chlorides, bicarbonate, nitrate, sulfate and arsenate, commonly exist in surface and ground water. Because these anions compete with fluorides for adsorbent active sites, their influence on fluoride adsorption was experimentally studied under different concentrations (0.1, 1.0 and 10 mM), as shown in Fig. 7. In addition, arsenate (As), which is a concomitant anion of widespread concern, was studied at four different concentrations (2, 5, 10 and 30 mgL−1 ). According to Fig. 7 (A–D), fluoride adsorption was affected by the presence of competitive anions. The red, orange, yellow, green and blue areas highlight the degree of adverse effect, with red indicating the greatest effect and blue indicating the least effect on fluoride adsorption. Fig. 7(A) shows the effect of concomitant anions on the fluoride adsorption capacity for pristine CuFe2 O4 (x=0.0). The presence of bicarbonate in solution exhibits the most negative effect on fluoride removal. Note that fluoride adsorption decreased by 12% in the presence of 0.1 mM HCO3 −1 at pH 7.0. Bicarbonate appears to have the most negative effect, followed by nitrates and arsenate.

M.A. Rehman et al. / Journal of Hazardous Materials 299 (2015) 316–324

These anions may strongly compete for ferrites active sites during adsorption. Fig. 7(B) shows the effect of HCO3 −1 , SO4 2− , NO3 1− , Cl1− and As on Ce-doped ferrite (x = 0.2). The adsorption of fluoride was unaffected by the lower concentration of the anions. However, the higher concentration, as usually encountered in industrial waste water, adversely affected the fluoride removal process. Fig. 7(C), demonstrates the effect of HCO3 −1 , SO4 2− , NO3 1− , Cl1− and As on ferrite (x = 0.3) at pH 7.0. For this particular adsorbent, the 10 mM NO3 1− exhibited a 45% decrease in fluoride adsorption, as indicated by the orange color. The presence of chlorides has a mild effect, even at a concentration of 10 mM. An arsenic concentration of 30 mgL−1 in the solution also exhibited a 39% decrease in fluoride adsorption. Fig. 7(D) shows the effect of HCO3 −1 , SO4 2− , NO3 1− , Cl1− and As on the last member of the ferrite series (x = 0.5). The elevated levels of HCO3 1− , NO3 1− and As were mapped in the danger zone of the graphs. In some regions with higher concentration of competing anions, careful pretreatment is mandatory before adsorption. However, lower concentration levels may not cause any concern during the fluoride removal process. Hence, this adsorbent is suitable for the de-fluoridation of the surface and ground water because the chloride, bicarbonate, nitrate, sulfate and arsenic levels are lower than those given above. 4. Conclusion In summary, by using a straightforward doping of cerium into magnetic ferrite, we successfully fabricated a series of novel magnetic adsorbents for fluoride adsorption and water remediation technology. Compared with un-doped parent material, superior results were obtained, and the present enhancement in adsorption behavior contributed to the active functional groups created after doping, the increased surface area with suitable porosity and the electro-magnetic properties. However, the presence of some concomitant species that are usually found in high concentration in industrial effluents reduced the efficiency of the present series of adsorbents due to competition for adsorbent active sites, which led to a decline in fluoride removal. These ferrite-based adsorbents for fluoride open new vistas for a broad range of applications and may lead to better insights into energy efficient magnetic adsorbents, electrochemical sensors and photo catalysts for environment remediation. Acknowledgments We gratefully thank Bright Spark (BSP), University of Malaya, and High Impact Research MoE Grant UM .C/625/1/HIR/MoE/SC/04/01 from the Ministry of Education Malaysia, University Malaya Center for Ionic liquids (UMCiL) and IPPP grant # PG215-2014B for funding the cost of chemicals and analysis. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jhazmat.2015.06. 030 References [1] R.E. Bain, S.W. Gundry, J.A. Wright, H. Yang, S. Pedley, J.K. Bartram, Accounting for water quality in monitoring access to safe drinking-water as part of the millennium development goals: lessons from five countries, Bull. World Health Org. 90 (2012) 228–235A, http://dx.doi.org/10.2471/BLT.11.094284 [2] C. Gonzalo, J.A. Camargo, Fluoride bioaccumulation in the signal crayfish Pacifastacus leniusculus (Dana) as suitable bioindicator of fluoride pollution in freshwater ecosystems, Ecol. Indic. 20 (2012) 244–251, http://dx.doi.org/10. 1016/j.ecolind.2011.12.019

323

[3] F. Shen, X. Chen, P. Gao, G. Chen, Electrochemical removal of fluoride ions from industrial wastewater, Chem. Eng. Sci. 58 (2003) 987–993, http://dx.doi. org/10.1016/S0009-2509(02) 639-5 [4] M.S. Onyango, Y. Kojima, O. Aoyi, E.C. Bernardo, H. Matsuda, Adsorption equilibrium modeling and solution chemistry dependence of fluoride removal from water by trivalent-cation-exchanged zeolite F-9, J. Colloid Interface Sci. 279 (2004) 341–350, http://dx.doi.org/10.1016/j.jcis.2004.06.038 [5] L.H. Velazquez-Jimenez, R.H. Hurt, J. Matos, J.R. Rangel-Mendez, Zirconium–carbon hybrid sorbent for removal of fluoride from water: oxalic acid mediated Zr(IV) assembly and adsorption mechanism, Environ. Sci. Technol. 48 (2013) 1166–1174, http://dx.doi.org/10.1021/es403929b [6] J. Kang, B. Li, J. Song, D. Li, J. Yang, W. Zhan, D. Liu, Defluoridation of water using calcined magnesia/pullulan composite, Chem. Eng. J. 166 (2011) 765–771, http://dx.doi.org/10.1016/j.cej.2010.11.031 [7] A.K. Yadav, C.P. Kaushik, A.K. Haritash, B. Singh, S.P. Raghuvanshi, A. Kansal, Determination of exposure and probable ingestion of fluoride through tea, toothpaste, tobacco and pan masala, J. Hazard. Mater. 142 (2007) 77–80, http://dx.doi.org/10.1016/j.jhazmat.2006.07.051 [8] W. Ma, T. Lv, X. Song, Z. Cheng, S. Duan, G. Xin, F. Liu, D. Pan, Characteristics of selective fluoride adsorption by biocarbon-Mg/Al layered double hydroxides composites from protein solutions: kinetics and equilibrium isotherms study, J. Hazard. Mater. 268 (2014) 166–176, http://dx.doi.org/10.1016/j.jhazmat. 2014.01.013 [9] A.M. Raichur, M. Jyoti Basu, Adsorption of fluoride onto mixed rare earth oxides, Sep. Purif. Technol. 24 (2001) 121–127, http://dx.doi.org/10.1016/ S1383-5866(00) 219-7 [10] M. Mohapatra, S. Anand, B.K. Mishra, D.E. Giles, P. Singh, Review of fluoride removal from drinking water, J. Environ. Manage. 91 (2009) 67–77, http://dx. doi.org/10.1016/j.jenvman.2009.08.015 [11] A. Bhatnagar, E. Kumar, M. Sillanpää, Fluoride removal from water by adsorption – a review, Chem. Eng. J. 171 (2011) 811–840, http://dx.doi.org/10. 1016/j.cej.2011.05.028 [12] B. Zhao, Y. Zhang, X. Dou, X. Wu, M. Yang, Granulation of Fe–Al–Ce trimetal hydroxide as a fluoride adsorbent using the extrusion method, Chem. Eng. J. 185–186 (2012) 211–218, http://dx.doi.org/10.1016/j.cej.2012.01.085 [13] P. Loganathan, S. Vigneswaran, J. Kandasamy, R. Naidu, Defluoridation of drinking water using adsorption processes, J. Hazard. Mater. 248–249 (2013) 1–19, http://dx.doi.org/10.1016/j.jhazmat.2012.12.043 [14] M. Hua, S. Zhang, B. Pan, W. Zhang, L. Lv, Q. Zhang, Heavy metal removal from water/wastewater by nanosized metal oxides: a review, J. Hazard. Mater. 211–212 (2012) 317–331, http://dx.doi.org/10.1016/j.jhazmat.2011.10.016 [15] C.R.N. Rao, J. Karthikeyan, Removal of fluoride from water by adsorption onto lanthanum oxide, Water Air Soil Pollut. 223 (2012) 1101–1114, http://dx.doi. org/10.1007/s11270-011-0928-0 [16] A.B. Gadkari, T.J. Shinde, P.N. Vasambekar, Structural analysis of Y3+-doped Mg–Cd ferrites prepared by oxalate co-precipitation method, Mater. Chem. Phys. 114 (2009) 505–510, http://dx.doi.org/10.1016/j.matchemphys.2008. 11.011 [17] M.A. Ahmed, E. Ateia, F.M. Salem, Spectroscopic and electrical properties of Mg–Ti ferrite doped with different rare-earth elements, Phys. B Condens. Matter 381 (2006) 144–155, http://dx.doi.org/10.1016/j.physb.2005.12.265 [18] H. Zhao, D. Zhang, F. Wang, T. Wu, J. Gao, Modification of ferrite–manganese oxide sorbent by doping with cerium oxide, Process Saf. Environ. Prot. 86 (2008) 448–454, http://dx.doi.org/10.1016/j.psep.2008.06.002 [19] M. Kobayashi, M. Flytzani-Stephanopoulos, Reduction and sulfidation kinetics of cerium oxide and Cu-modified cerium oxide, Ind. Eng. Chem. Res. 41 (2002) 3115–3123, http://dx.doi.org/10.1021/ie010815w [20] A. Witek-Krowiak, K. Chojnacka, D. Podstawczyk, A. Dawiec, K. Pokomeda, Application of response surface methodology and artificial neural network methods in modelling and optimization of biosorption process, Bioresour. Technol. 160 (2014) 150–160, http://dx.doi.org/10.1016/j.biortech.2014.01. 021 [21] M.A. Rehman, I. Yusoff, R. Ahmmad, Y. Alias, Arsenic adsorption using palm oil waste clinker sand biotechnology: an experimental and optimization approach, Water Air Soil Pollut. 226 (2015) 1–13, http://dx.doi.org/10.1007/ s11270-015-2411-9 [22] S. Bibi, A. Farooqi, K. Hussain, N. Haider, Evaluation of industrial based adsorbents for simultaneous removal of arsenic and fluoride from drinking water, J. Clean. Prod. 87 (2015) 882–896, http://dx.doi.org/10.1016/j.jclepro. 2014.09.030 [23] E. Naseri, A. Reyhanitabar, S. Oustan, A.A. Heydari, L. Alidokht, Optimization arsenic immobilization in a sandy loam soil using iron-based amendments by response surface methodology, Geoderma 232–234 (2014) 547–555, http:// dx.doi.org/10.1016/j.geoderma.2014.06.009 [24] J.A. Arcibar-Orozco, D.-B. Josue, J.C. Rios-Hurtado, J.R. Rangel-Mendez, Influence of iron content, surface area and charge distribution in the arsenic removal by activated carbons, Chem. Eng. J. 249 (2014) 201–209, http://dx. doi.org/10.1016/j.cej.2014.03.096 [25] W.-X. Gong, J.-H. Qu, R.-P. Liu, H.-C. Lan, Effect of aluminum fluoride complexation on fluoride removal by coagulation, Colloids Surf. A: Physicochem. Eng. Aspects 395 (2012) 88–93, http://dx.doi.org/10.1016/j. colsurfa.2011.12.010 [26] S.-Y. Oh, M.-K. Yoon, Chemical extraction of arsenic from contaminated soils in the vicinity of abandoned mines and a smelting plant under subcritical conditions, Soil Sediment Contam. 23 (2013) 166–179, http://dx.doi.org/10. 1080/15320383.2014.808169

324

M.A. Rehman et al. / Journal of Hazardous Materials 299 (2015) 316–324

[27] P. Roy, N.K. Mondal, K. Das, Modeling of the adsorptive removal of arsenic: a statistical approach, J. Environ. Chem. Eng. 2 (2014) 585–597, http://dx.doi. org/10.1016/j.jece.2013.10.014 [28] T. Larbi, M. Haj Lakhdar, A. Amara, B. Ouni, A. Boukhachem, A. Mater, M. Amlouk, Nickel content effect on the microstructural, optical and electrical properties of p-type Mn3 O4 sprayed thin films, J. Alloys Compd. 626 (2015) 93–101, http://dx.doi.org/10.1016/j.jallcom.2014.11.088 [29] T. Larbi, B. Ouni, A. Boukhachem, K. Boubaker, M. Amlouk, Investigation of structural, optical, electrical and dielectric properties of catalytic sprayed hausmannite thin film, Mater. Res. Bull. 60 (2014) 457–466, http://dx.doi.org/ 10.1016/j.materresbull.2014.09.007 [30] M.A. Ahmed, S.F. Mansour, H. Ismael, A comparative study on the magnetic and electrical properties of MFe12O19 (MBa and Sr)/BiFeO3 nanocomposites, J. Magn. Magn. Mater. 378 (2015) 376–388, http://dx.doi.org/10.1016/j. jmmm.2014.10.173 [31] M. Shanmugaprakash, V. Sivakumar, Development of experimental design approach and ANN-based models for determination of Cr(VI) ions uptake rate from aqueous solution onto the solid biodiesel waste residue, Bioresour. Technol. 148 (2013) 550–559, http://dx.doi.org/10.1016/j.biortech.2013.08. 149 [32] J. Wang, D. Kang, X. Yu, M. Ge, Y. Chen, Synthesis and characterization of Mg–Fe–La trimetal composite as an adsorbent for fluoride removal, Chem. Eng. J. 264 (2015) 506–513, http://dx.doi.org/10.1016/j.cej.2014.11.130 [33] C. Eith, M. Kolb, A. Seubert, K.H. Viehweger, Practical Ion Chromatography, Metrohm Ltd., Herisau, Switzerland, 2001. [34] Mexmain, Contribution a l’etude du ferrite cuivreux et de ses solutions solides avec le ferrite cuivrique, J. Ann. Chim. (1971) 297–308. [35] R.W.G. Wyckoff, Crystal Structures, Interscience Publishers, New York, 1963. [36] M.S. Khandekar, N.L. Tarwal, J.Y. Patil, F.I. Shaikh, I.S. Mulla, S.S. Suryavanshi, Liquefied petroleum gas sensing performance of cerium doped copper ferrite, Ceram. Int. 39 (2013) 5901–5907, http://dx.doi.org/10.1016/j.ceramint.2013. 01.010 [37] X. Wu, H. Yu, H. Dong, L. Geng, Enhanced infrared radiation properties of CoFe2 O4 by single Ce3+-doping with energy-efficient preparation, Ceram. Int. 40 (2014) 5905–5911, http://dx.doi.org/10.1016/j.ceramint.2013.11.035 [38] E. Rezlescu, L. Sachelarie, P.D. Popa, N. Rezlescu, Effect of substitution of divalent ions on the electrical and magnetic properties of Ni–Zn–Me ferrites, Magnetics, IEEE Trans. 36 (2000) 3962–3967, http://dx.doi.org/10.1109/20. 914348 [39] CRC Handbook of Chemistry and Physics, Internet Version, in: D.R. Lide (Ed.), CRC Press, Boca Raton, FL, 2005 http://www.hbcpnetbase.com [40] H. Zhang, J. Wang, X. Gao, Z. Wang, S. Wang, The electrochemical activity of polyaniline: An important issue on its use in electrochemical energy storage devices, Synth. Met. 187 (2014) 46–51, http://dx.doi.org/10.1016/j.synthmet. 2013.10.022 [41] A.I. Borhan, A.R. Iordan, M.N. Palamaru, Correlation between structural, magnetic and electrical properties of nanocrystalline Al3+ substituted zinc ferrite, Mater. Res. Bull. 48 (2013) 2549–2556, http://dx.doi.org/10.1016/j. materresbull.2013.03.012 [42] P. Balaz, Mechanochemistry in Nanoscience and Minerals Engineering, Springer, Berlin, Heidelberg, 2008. [43] M. Ishaque, M.A. Khan, I. Ali, H.M. Khan, M.A. Iqbal, M.U. Islam, M.F. Warsi, Impacts of yttrium substitution on FMR line-width and magnetic properties

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51] [52]

[53]

[54]

[55]

[56]

[57]

[58]

of nickel spinel ferrites, J. Magn. Magn. Mater. 382 (2015) 98–103, http://dx. doi.org/10.1016/j.jmmm.2015.01.057 C.N. Chinnasamy, M. Senoue, B. Jeyadevan, O. Perales-Perez, K. Shinoda, K. Tohji, Synthesis of size-controlled cobalt ferrite particles with high coercivity and squareness ratio, J. Colloid Interface Sci. 263 (2003) 80–83, http://dx.doi. org/10.1016/s0021-9797(03) 258-3 E. Domingo, Advances on structuring, integration and magnetic characterization of molecular nanomagnets on surfaces and devices, Chem. Soc. Rev. 41 (2012) 258–302, http://dx.doi.org/10.1039/c1cs15096k A.M. Wahba, M. Bakr Mohamed, Structural and magnetic characterization and cation distribution of nanocrystalline Cox Fe3-x O4 ferrites, J. Magn. Magn. Mater. 378 (2015) 246–252, http://dx.doi.org/10.1016/j.jmmm.2014.10.164 J.R. Potts, D.R. Dreyer, C.W. Bielawski, R.S. Ruoff, Graphene-based polymer nanocomposites, Polymer 52 (2011) 5–25, http://dx.doi.org/10.1016/j. polymer.2010.11.042 K. Raju, G. Venkataiah, D.H. Yoon, Effect of Zn substitution on the structural and magnetic properties of Ni–Co ferrites, Ceramics International 40 (2014) 9337–9344, http://dx.doi.org/10.1016/j.ceramint.2014.01.157 C. Petrovic, P. Pagliuso, M. Hundley, R. Movshovich, J. Sarrao, J. Thompson, Z. Fisk, Heavy-fermion superconductivity in CeCoIn5 at 2.3K, arXiv preprint cond-mat/0103168, (2001). O.V. Gornostaeva, K.V. Lamonova, S.M. Orel, Y.G. Pashkevich, Magnetic properties of Ce3+ ion in iron-containing oxypnictide CeFeAsO, Low Temp. Phys. 39 (2013) 343, http://dx.doi.org/10.1063/1.4801990 K.J. Buschow, Concise Encyclopedia of Magnetic and Superconducting Materials, Elsevier, 2005. ´ Nature of the J. Fikáˇcek, J. Prokleˇska, J. Prchal, J. Custers, V. Sechovsky, magnetic ground state in the mixed valence compound CeRuSn: a single-crystal study, J. Phys.: Condens. Matter 25 (2013) 416006. A. Kostopoulou, K. Brintakis, M. Vasilakaki, K.N. Trohidou, A.P. Douvalis, A. Lascialfari, L. Manna, A. Lappas, Assembly-mediated interplay of dipolar interactions and surface spin disorder in colloidal maghemite nanoclusters, Nanoscale 6 (2014) 3764–3776, http://dx.doi.org/10.1039/C3NR06103E G. Dixit, J. Pal Singh, R.C. Srivastava, H.M. Agrawal, Magnetic resonance study of Ce and Gd doped NiFe2 O4 nanoparticles, J. Magn. Magn. Mater. 324 (2012) 479–483, http://dx.doi.org/10.1016/j.jmmm.2011.08.027 M. Niaz Akhtar, M. Azhar Khan, M. Ahmad, G. Murtaza, R. Raza, S.F. Shaukat, M.H. Asif, N. Nasir, G. Abbas, M.S. Nazir, M.R. Raza, Y3Fe5 O12 nanoparticulate garnet ferrites: comprehensive study on the synthesis and characterization fabricated by various routes, J. Magn. Magn. Mater. 368 (2014) 393–400, http://dx.doi.org/10.1016/j.jmmm.2014.06.004 P. Dutta, S. Pal, M.S. Seehra, Y. Shi, E.M. Eyring, R.D. Ernst, Concentration of Ce3+ and Oxygen vacancies in cerium oxide nanoparticles, Chem. Mater. 18 (2006) 5144–5146, http://dx.doi.org/10.1021/cm061580n D. Badkar, K. Pandey, G. Buvanashekaran, Development of RSM- and ANN-based models to predict and analyze the effects of process parameters of laser-hardened commercially pure titanium on heat input and tensile strength, Int. J. Adv. Manuf. Technol. 65 (2013) 1319–1338, http://dx.doi.org/ 10.1007/s00170-012-4259-0 Y. Yu, L. Yu, J. Paul Chen, Adsorption of fluoride by Fe–Mg–La triple-metal composite: adsorbent preparation, illustration of performance and study of mechanisms, Chem. Eng. J. 262 (2015) 839–846, http://dx.doi.org/10.1016/j. cej.2014.09.006.