Removal of mutagen X “MX” from drinking water using reduced graphene oxide coated sand particles

Abstract

Purpose

Mutagen X is a hazardous by-product of disinfection by chlorine, which is responsible for most of the mutagenicity in chlorinated drinking water. It has the cancer potency value of 100-fold higher than bromodichloromethane and 6000-fold higher than chloroform, In this study, Mutagen X was removed from aqueous media by a thermally reduced graphene oxide bonded on the surface of amino-functionalized sand particles.

Method

A Box-Behnken design was applied to optimize the adsorption process. Characterization of the adsorbent and graphene oxide was accomplished using scanning electron microscopy (SEM), Energy-dispersive X-ray spectroscopy (EDS), X-ray powder diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), and Raman analysis. The effects of three independent parameters, including initial concentration (20–200 μg L⁻¹), temperature (5–30 °C), and adsorbent dose (2–80 g L⁻¹) were examined using batch experiments.

Results

Characterization results confirmed that the graphene oxide was successfully coated on the surface of sand particles. Regression analysis of experimental results showed a great fit with a quadratic polynomial model with the R² = 0.999. Optimum conditions (initial concentration: 20 μg L⁻¹, temperature: 30 °C, and adsorbent dose: 80 g L⁻¹) with the desirability of 1.0 resulted in the minimum residual concentration of Mutagen X (2 μg L⁻¹). Equilibrium study results depicted that the experimental data were fitted well to the Freundlich and UT isotherm models.

Introduction

Water disinfection is a vital process of drinking water purification [1]. Millions of people use drinking water directly through their main public treatment centers. However, chlorine may react with Natural Organic Matter (NOM) and create disinfection by-products (DBPs), which have been proved to be hazardous for human health through exposure and consumption [2]. These DBPs usually include regulated compounds, such as trihalomethanes (THMs), and haloacetic acids (HAAs), and also less prevalent and unregulated compounds, such as 3-chloro-4(dichloromethyl)-2(5H)-furanone (MX) [3]. Most unregulated DBPs that cause tumors in some animals are intensively effective at the dosage higher than 75 mg kg⁻¹ day⁻¹ [1].

The U.S. EPA has set some regulations for permissible level of THMs (80 μg L⁻¹) in drinking water and five most common HAAs (60 μg L⁻¹) [4], while Health Canada has set minimum concentration of 100 μg L⁻¹ and 80 μg L⁻¹ for THMs and HAAs, respectively [5]. However, it has been proved that even dosages as low as 0.4 mg kg⁻¹ day⁻¹ of MX can result in rodent tumors [1].

Halogenated furanones have attracted researcher’s attention because of the possible exposure of the population to the extremely mutagenic compound of 3-chloro-4-(dichloromethyl)- 5-hydroxy-2(5H)- furanone, generally known as Mutagen X or MX [6]. It has been shown that there are 12 different forms of MX in drinking water disinfected by chlorine, which all of them are referred as halogenated furanes. Typically, they are formed as a result of the reaction of chlorine with humic acid. Up to now, MX is the most mutagenic DBP compound that has been identified, and it is responsible for 20–60% of the total mutagenicity of drinking water [7].

The cancer potency value for MX has been determined about 100-fold higher than that of bromodichloromethane and 6000-fold higher than that of chloroform by the U.S. Environmental Protection Agency [8]. In other words, a significant portion of the total mutagenicity of chlorinated drinking water has been attributed to MX. In addition, MX’s mutagenic potential is twice that of aflatoxin B1 [9]. Some clinical trials showed that MX is much stronger than THMs in causing DNA damage and changing paths involved in cell growth. Also, some epidemiological studies showed the considerable role of MX in increase of cancer rates associated with disinfected water. It has been reported that MX is likely to be stronger mutagen than any other chlorination by-products in causing cancer in animals [10]. MX has been classified as a group 2B, possibly carcinogenic to humans, based on the limited animal evidence and other relevant data by the International Agency for Research on Cancer (IARC) [11].

Various methods have been utilized for the removal of DBPs and NOM from drinking water including a numerous treatment processes such as coagulation by organic, inorganic, and polymeric coagulants [12], thermal and photochemical treatments, membrane filtration [6], ion exchange [13], advanced oxidation [14], and adsorption by activated carbon [15].

The absorption process, compared with other mentioned processes, is one of the most widely studied processes due to its practicality and simple operation characteristics [16]. In the adsorption process, the selection of an adsorbent among various candidates is a vital stage [17]. Carbon nanotubes and magnetic nanoparticles have been used for the removal of hydrophobic NOMs and humic acids [12, 13]. Also, granular activated carbon has been applied as an adsorbent for the removal of synthetic organic matters and NOMs [18].

Following the development of fullerene and carbon nanotubes (CNTs) in the prior decades, the advent of graphene with its exceptional properties such as a particular surface area (2620 m² g⁻¹), which has the potential to adsorb organic pollutants, has opened up a new field in the application of two dimensional (2D) nanomaterials for the management of environmental contaminants [15, 16]. Researches have exposed that graphene has better adsorption capacity for heavy metals and fluoride than carbon nanotubes [18, 19]. Moreover, unlike carbon nanotubes, preparation of graphene oxide (GO) nanosheets from graphite using Hummers strategy presents numerous oxygen-containing groups; for instance, carboxyl and hydroxyl groups, on the surface of GO nanosheets [20].

Gao et al. study [21] proved that GO@sand, or in other words, “Core-shell,” has a significant favorable adsorption capacity comparing to sand particles. The GO@sand can remove heavy metals and organic dyes at the concentrations at least 5-fold higher compared to pure sand [20–22]. In addition to the chemical bond formation between the GO and the amino-functionalized sand particles, suitable interaction between rGO and MX, T-shape π-π stacking, makes GO@sand a practical adsorbent for water treatment, which is recoverable and non-dispersible.

In the present study, we report the immobilization of rGO on the surface of sand particles using heat-treated process. For the first time, rGO@sand was proposed for the removal of the carcinogenic halogenated furan, MX, from drinking water. The characteristics of the produced adsorbent were investigated by FTIR, SEM, EDAX, and XRD analysis. To achieve the optimum conditions, response surface methodology (RSM) was applied. The equilibrium and kinetic studies has been assessed by the related models.

Material and methods

Materials

Hydrochloric acid (HCl, 7647-01-0, 37%), hydrogen peroxide (H₂O₂, 7722-84-1, 30%), sulfuric acid (H₂SO₄, 7664-93-9, 98%), and toluene (C₇H₈, 108–88-3, 99.8%) were purchased from Merck (Darmstadt, Germany). (3-Aminopropyl) triethoxysilane (APTS) (H₂N(CH₂)₃Si(OC₂H₅)₃, 919–30-2, 95%), sodium nitrate (NaNO₃, 7631-99-4, 99%), potassium permanganate (KMnO₄, 7722-64-7, 99%), and 3-Chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone, Mutagen X, (C₅H₃Cl₃O₃, 77,439–76-0, 96%) were purchased from Sigma–Aldrich (St. Louis, MO, USA). Sand particles were obtained from Firuzkooh glass company, Tehran, Iran.

Preparation of amino-functionalized sand particles

To provide more active sites for reacting with graphene oxide, the surface of sand particles was functionalized with amino functional groups. Firstly, 250 g of sand particles between the mesh of 30 and 80, were added in 200 mL HCl solution (1:1 v/v) and heated on a heater for two hours. Then the prepared sand particles were washed with distilled water several times and dried at a temperature of 70 °C in an oven. As a consequence of this treatment, the surface impurities of the sand particles were quite removed, which resulted in a clean surface for functionalization.

To add amin functional groups on the sand particles based on Mortazavi et.al study [23], a mixture of 250 g of sand particles, 200 mL of toluene, and 2.5 mL of APTS was refluxed for 24 h at a temperature of 180 °C. Finally, the resultant product was washed with methanol and distilled water several times and dried at ambient temperature to be prepared for the next step.

Preparation of rGO@sand

Synthesis of graphene oxide carried out using the modified Hummers method [24, 25]. Briefly, first, 500 mL of sulfuric acid (H₂SO₄) was poured into a 100 mL flat-bottom flask containing graphite powder (10 g), and placed in an ice bath. Sodium nitrate (NaNO₃, 10 g) was added to the solution, and then potassium permanganate (KMnO₄, 60 g) was gradually added while stirring. The temperature of the solution was kept below 10 °C. Afterward, the temperature increased to 40 °C and kept constant for 12 h, then distilled water (1000 mL) was added to the mixture and stirred continuously for 30 min at the temperature of 90 °C. Next, hydrogen peroxide (H₂O₂, 30%, 70 mL) was added, and the solution was cooled to the ambient temperature. Subsequently, the settled particles were washed for several times with HCl solution (5%) and deionized water. The separation of particles was carried out using a centrifuge. The washing by deionized water repeated to reach neutral pH. Eventually, the concentrated suspension was sonicated using an ultrasonic bath.

According to the Mortazavi et.al [23], preparation of rGO@sand was conducted using the following steps: 0.5 g of the GO was dispersed in 200 mL distilled water using an ultrasonic bath. Then the suspension was added gradually to the amino-functionalized sand particles while heating and mixing on a heater. After that, the particles incorporated GO on their surface was heated at 300 °C under N₂ atmosphere for 20 min. As a consequence, the surface amine groups reacted with the carboxylic acid groups of GO, resulting in strong chemical bonds. The color of the composite was changed from brown to black as a consequence of the partial thermal reduction of GO.

Batch experiment

The range and significant parameters in the absorption process of MX was determined using pretests and the impact of parameters was investigated using a Box-Behnken design (BBD) and response surface methodology (RSM). In this study, adsorbent mass, temperature, and concentration of MX were selected as significant parameters, while the pH of the solution and contact time were found to be ineffective parameters. Eq. (1) was applied to code the chosen variables:

$$x_i=\frac{x_i-x_0}{\mathrm{\Delta }x}$$ 1

where X is the dimensionless coded value of i_ththe independent variable; X₀ is the value of the variable at the center point, and ΔX is the step change value. The real values and coded levels of the independent variables are given in Table 1. The batch experiments were conducted according to the conditions mentioned in Table 2. In brief, the sample solutions (50 mL) prepared from the stoke solution of MX were treated with the adsorbent. The samples were stirred for 15 min, and the remaining MX in the solutions was determined using GC-ECD.

Table 1.

The real values and coded levels of the independent variables

Variables	Unit	Levels
		-1	0	1
A: Temperature	oC	5.0	17.5	30.0
B: MX	ppb	20	110	200
C: Dose	g L−1	2.0	41.0	80.0

Table 2.

EDS analysis results of acid-washed sand, amino-functionalized sand, and rGO@sand

Element	Acid washed sand	Amino-functionalized sand	rGO@sand
C%	0.51	0.03	12.04
N%	2.7	5.10	0
O%	58.62	57.71	51.22
Al%	0.82	1.51	1.67
Si%	37.35	35.65	35.07

Equilibrium studies

To investigate the adsorption behavior of the adsorption process, MX solutions with various concentrations (25–115 μg L⁻¹) were loaded with a constant dose of the adsorbent (80 g L⁻¹) at the temperature of 30 °C, and diverse adsorption models, such as Langmuir, Freundlich, Temkin, and UT isotherms were used and fitted the experimental adsorption data for MX. Langmuir model hypotheses monolayer adsorption [26] and provides a dimensionless separation factor (R_L) to evaluate the efficiency of adsorption by the following equation:

$$R_L=\frac{1}{1+K_L{C}_0}$$ 2

where K_L is the Langmuir constant (L μg⁻¹), and C₀ is the initial concentration of MX (μg L⁻¹). Furthermore, Freundlich isotherm is an empirical equation with an assumption of multilayer adsorption. The exponent of the Freundlich equation, n_F, is a dimensionless parameter which describes the deviation from the linearity of adsorption. When n_F < 1, the chemical adsorption (n_F < 1), and when n_F > 1, the physical interaction is involved in the adsorption process [27, 28]. The Temkin isotherm assumes that the adsorption heat of molecules would gradually decrease by the coverage of adsorbent surface [26, 27], and UT Isotherm is a dose-independent adsorption model [29].

Thermodynamic studies

To investigate the effect of temperature, thermodynamic parameters, such as Gibbs free energy (∆G^o), entropy (∆S^o), and enthalpy (∆H^o) were calculated at the initial concentration of 110 μg L⁻¹ and adsorbent dosage of 80 g L⁻¹. By investigating the variation of the equilibrium constant (K_L) by temperature, these parameters were obtained. Changes in enthalpy (∆H^o) and entropy (∆S^o) were calculated using the Van’t Hoff plot. The equations used to calculate free Gibbs energy are shown below:

$$ln{k}_L=\frac{\mathrm{\Delta }{S}^o}{R}-\frac{\mathrm{\Delta }{H}^o}{\mathit{RT}}$$ 3

$$\mathrm{\Delta }{G}^o=\mathrm{\Delta }{H}^o-T\mathrm{\Delta }{S}^o$$ 4

where R refers to the universal gas constant (kJ mol⁻¹ K⁻¹), T is the temperature (K), and K_L is the equilibrium constant, which can be obtained from Langmuir isotherms at various temperature.

Instruments

Scanning electron microscopic (SEM) images with 500-fold magnification were obtained using a scanning electron microscope (VEGA-II, TESCAN, Czech Republic) to characterize the morphology of GO and rGO@sand. In addition, the variations of the surface elemental composition of sand particles at each step were identified using EDAX analysis (VEGA-II, TESCAN, Czech Republic). In order to recognize the presence of functional groups in the GO structure, FTIR spectra were recorded on an infrared spectrometer (Tensor 27, Bruker, Germany) in the range of 400–4000 cm⁻¹. XRD analysis of powders was carried out on an X-ray diffractometer (PANalytical XPERT-PRO, Malvern Panalytical company, United Kingdom) equipped with a monochromatic CuK_α radiation source at λ = 1.5406 A^°. Data were collected from 5° to 90° at a scan rate of 0.1° min⁻¹. Raman spectra were collected by a Raman spectroscope equipped with a 532 nm laser source (Tekram, Teksan, Iran). In order to determine the amount of remaining MX in the solution, a BRUKER 450 gas chromatograph equipped with a Ni63 electron capture detector was employed. A DB-5 capillary column (30 m, 0.25 mm internal diameter, and 0.25 mm film thickness) purchased from Agilent Technologies Inc. (Santa Clara, CA, USA) was employed to separate the compounds. The injector temperature was 250 °C. The initial temperature of the oven was kept constant at 100 °C for 2 min, and then increased up to 230 °C at 10 °C min⁻¹, and finally kept at 230 °C for 7 min. The inlet was operated at the split ratio of 2:1, and the detector temperature was 300 °C. Also, nitrogen gas was used as the carrier gas at a constant flow of 1.4 mL min⁻¹.

Results and discussion

Characterization of GO and rGO@sand

Figure 1 (a) and (c) shows the SEM images of GO on the scales of 2 and 5 μm, displaying the layer structure of GO. Figure 1 (b) and (d) shows the SEM images of rGO@sand on the same scale. A layer of graphene oxide bonded to the surface of sand can be identified. Table 2 shows the EDAX analysis of acid-washed sand, amino-functionalized sand, and rGO@sand. The higher surface nitrogen content of amino-functionalized sand particles compared to that of acid-washed sand particles shows the modification of the surface of sand particles by amine functional groups. Moreover, the higher carbon content in rGO@sand shows the successful coating of rGO on the surface of sand particles. Figure 2 (b) shows the XRD pattern of GO. The sharp diffraction peak observed at 2θ =10.52° with a d-spacing of 8.39 Å corresponds to GO [30].

Fig. 1.

SEM images of GO at (a) 10 KX, (c) 6 KX, and rGO@sand at (b) 10 KX, (d) 6 KX

Fig. 2.

a The FTIR spectra of synthesized GO, b the XRD results for GO, and c Raman spectra of GO and rGO@sand

The FTIR spectra were recorded to represent the functional groups of GO (Fig. 2 (a)). The main adsorption band at 3362 cm ⁻¹ is allocated to the stretching vibration mode of the O-H group. C=C and C=O stretching vibrations were observed at 1616 cm⁻¹ and 1727 cm⁻¹, respectively. The two adsorption peak at 1168 cm⁻¹ and 1048 cm⁻¹ are assigned to the C–O stretching vibrations [20, 21, 25, 31]. The Raman spectroscopy is a powerful tool for the characterization of carbonaceous material, and it can depict the relative disorder in the crystal structure of GO. Raman spectra data of rGO@sand showed consistency with the Raman spectra data of GO with the same peaks at 1356 cm⁻¹ (D band) and 1608 cm⁻¹ (G band) (Fig. 2 (c)) [21]. A measure of the extent of disorder in the crystal structure is the relative intensity (I_D/I_G), which in this case for rGO@sand (0.95) was higher than that of GO (0.88) and it demonstrates a reduction in oxygen-containing functional groups of GO [23].

Statistical analysis

A Box-Behnken design (BBD) together with the response surface methodology (RSM) was used to design the experiments, model the adsorption process, and assess the significance of variables. There are three significant variables affecting the response, including the concentration of MX, the temperature of the solution, and the dosage of the adsorbent. Residual concentration of MX is shown in Table 3. The cubic model was aliased even though it had higher amounts of R² and adjust R² (adj-R²) in comparison to the quadratic model. Therefore, the quadratic model was chosen to be executed on the experimental data. The quadratic model for the effluent concentration of MX was as below:

$$\begin{array}{l}\mathit{\text{Sqrt}}\left(A\right)=0.018692+0.20884\times B+0.095976\times C+0.058897\times D-0.000305372\times B\times D\\ -0.000132041\times C\times D-0.00681485\times B^2-0.000225768\times C^2-0.000684587\times D^2\end{array}$$ 5

Table 3.

Experimental Design matrix and response

Run	Coded Values			Actual values			Ce (MX) μg L−1
	X1	X2	X3	X1	X2	X3	Observed	Predicted
1	0	0	0	17.5	110	41	91.2	97.2
2	+1	0	-1	30.0	110	2	65.6	64.7
3	0	-1	-1	17.5	20	2	12.3	12.4
4	+1	0	+1	30.0	110	80	40.1	41.1
5	0	0	0	17.5	110	41	95.4	97.2
6	0	+1	-1	17.5	200	2	138.2	139.3
7	+1	-1	0	30.0	20	41	8.1	7.6
8	0	+1	+1	17.5	200	80	92.6	91.0
9	0	0	0	17.5	110	41	98.2	97.2
10	0	0	0	17.5	110	41	98.9	97.2
11	+1	+1	0	30.0	200	41	105.9	102.4
12	0	-1	+1	17.5	20	80	10.3	9.7
13	0	0	0	17.5	110	41	98.2	97.2
14	-1	0	-1	5.0	110	2	79.5	77.5
15	-1	0	+1	5.0	110	80	60.4	60.3
16	-1	-1	0	5.0	20	41	14.2	14.6
17	-1	+1	0	5.0	200	41	125.3	124.9

Where A is the final concentration of MX; C is the initial concentration of MX; B is the temperature of the solution, and D is the dosage of the adsorbent.

Analysis of variance (ANOVA) was used to show the validation of the model and the significance of variables and optimize the variables and the corresponding responses (Table 4). Furthermore, the significance of the model was determined by the use of p value, F statistic, and the regression coefficient (R²). The F statistic of the model was 1813.61 relating to the p value of <0.0001, which proved the models’ adequacy. p values less than 0.05 indicates that the terms are significant, and terms with p values higher than 0.05 are not significant. As it is presented in Table 4, the term of the initial concentration had a significant effect on the removal of MX with the F-value of 9524.53 and p value of 0.001. The coefficient of determination (R²) was 0.9996, which means that 99.96% of the variation could be explained by the model, and there is only a 0.04% deviation. This result implies that just 0.04% of the total variance could not be explained by the quadratic model.

Table 4.

Analysis of variance (ANOVA) of MX for the selected quadratic model

Term	Sum of squares	df	Mean square	Regression and ANOVA factors R%		Significance
				F value	p value
Model	132.06	8	16.50	1813.61	<0.0001	*
A-Tem	1.78	1	1.78	195.67	<0.0001	*
B-Con	86.69	1	86.69	9524.53	<0.0001	*
C-Dose	3.56	1	3.56	391.23	<0.0001	*
AC	0.08	1	0.08	9.73	0.0206	*
BC	0.85	1	0.85	94.39	<0.0001	*
A2	4.03	1	4.03	442.91	<0.0001	*
B2	11.89	1	11.89	1306.36	<0.0001	*
C2	3.85	1	3.85	423.53	<0.0001	*
Residual	0.05	6	0.009
Lack of Fit	0.03	3	0.01	2.12	0.2761	#
Pure Error	0.01	3	0.005
Cor Total	132.11	14

R² = 0.9996, Adj-R² = 0.9990, * = significant, # = not-significant

Lack of fit is a measurement of model fit on the data. High values of lack of fit (p < 0.05) is an undesirable property because it indicates that the model does not fit the data well. Therefore, having an insignificant lack of fit is recommendable. Here, the p value of lack of fit was 0.2761, which indicated that our model fitted well and was significant. The adjusted R-squared and predicted R-squared should be within approximately 0.20 of each other to be in “reasonable agreement.” If they are not, there may be a problem with either the data or the model [32–35]. Here, the predicted R-square was 0.9951, and adjusted R-square was 0.9990, so they were in a good agreement to each other. Figure 3 shows the actual and predicted amounts of effluent concentration of MX. As shown clearly, the experimental points are almost completely conforming to the model-predicted line, indicating the perfect matching of the model with the experiments.

Fig. 3.

Actual and predicted amounts of effluence concentration of MX

Effect of the variable on the MX removal

Figure 4 (a) shows the interaction between the initial concentration of MX and the adsorbent dosage. The MX residual concentration in the solution increased by increasing the initial concentration of MX. Also, by increasing the adsorbent dose, the remaining concentration of MX decreased. At lower initial concentration (20 μg L⁻¹), the effect of the adsorbent dose was not considerably apparent due to the excessive presence of active sites on the adsorbent. At higher initial concentration (200 μg L⁻¹), because of the lack of active sites on the adsorbent surface, the adsorbent was rapidly saturated, which made the effect of the absorbent dose on the removal of MX significantly evident [22, 23]. Figure 4 (b) represents the effect of the temperature on the removal of MX. As can be seen, at the initial concentration of 200 μg L⁻¹, by increasing the temperature from 5 to 30, the residual concentration of MX decreased significantly. This can be described by the fact that the adsorption process of MX by rGO@sand was an endothermic one, so increasing the temperature resulted in an increase in the adsorption rate [36].

Fig. 4.

Response surface plots showing the interaction of (a) initial concentration and adsorbent dosage at the temperature of 17.5 °C, (b) temperature and initial concentration with the adsorbent dosage of 41 g L⁻¹ and (c) perturbation plot for the removal of MX by rGO@sand particles at the temperature of 17.5 °C, initial concentration of 110 μg L⁻¹, and adsorbent dosage of 41 g L⁻¹

Optimization

Optimization was conducted in a numerical model to achieve a minimum residual concentration of MX and minimum adsorbent dose, while other factors remained at the range. As depicted in Fig. 5, in the aim of minimizing the residual concentration of MX, while other factors were at the range, the residual concentration was calculated 2.0 μg L⁻¹, temperature optimized at 29.9 °C, initial concentration was set at 20.35 μg L^-1, and adsorbent dosage was 79.7 g L⁻¹. The desirability of this arrangement was at the highest level and equal to 1.00.

Fig. 5.

Optimization study of the effluent concentration of MX

Thermodynamic study

As it is shown in Table 5, the Gibbs free energy found to be negative, which approved the spontaneity of the adsorption process, and positive amounts of enthalpy approved that the adsorption was endothermic.

Table 5.

Parameters of thermodynamic study

T(°C)	ln K	∆Go (kJ mol−1)	∆Ho (kJ mol−1)	∆So (kJmol−1 K−1)
10	−2.09	−5.409	0.005
20	−2.04	−5.601		0.019
30	−1.67	−5.792

Adsorption isotherm

A comparison of the experimental results and adsorption isotherms at the temperature of 30 °C has provided in Fig. 6. (a), and residual plots of investigated models were also provided in Fig. 6. (b). The linear equations, parameters, and errors of investigated isotherm models are presented in Table 6. The maximum correlation coefficient was related to the UT isotherm and Freundlich isotherm, respectively [40].

Fig. 6.

a The equilibrium adsorption isotherms for the adsorption of MX by rGO@Sand particles. b Residual plots of investigated models. Conditions: Adsorbent dose 80 g L⁻¹; Temperature: 30 °C, (c, d, and e) K_L, n_f, and K_f values at different temperatures

Table 6.

Parameters and errors of the isotherm study

Isotherm	Equation	Parameters	Errors
Langmuir	$\frac{{\mathrm{C}}_{\mathrm{e}}}{{\mathrm{q}}_{\mathrm{e}}}=\frac{1}{{\mathrm{K}}_{\mathrm{L}}{\mathrm{q}}_{\mathrm{m}}}+\frac{{\mathrm{C}}_{\mathrm{e}}}{{\mathrm{q}}_{\mathrm{m}}}$	qm: 219.1 μg g−1	R2: 0.955
		KL: 0.073 L μg−1	RMS: 11.73	[37]
			X2: 9.015
Freundlich	$ln{\mathrm{q}}_{\mathrm{e}}=ln{\mathrm{K}}_{\mathrm{F}}+\frac{1}{\mathrm{n}}ln{\mathrm{C}}_{\mathrm{e}}$	nF: 2.30	R2: 0.995
		KF: 32.97 (μg g−1) (L μg−1)1/n	RMS: 3.849	[38]
			X2: 0.697
Temkin	qe = BllnA + BllnCe	A: 0.866 L μg−1	R2: 0.942
		Bl: 44.999	RMS: 9.738	[39]
			X2: 6.925
UT	$ln\left(\frac{{\mathrm{C}}_{\mathrm{e}}}{{\mathrm{C}}_{\mathrm{c}}^{\mathrm{ad}}}‐\frac{{\mathrm{K}}_{\mathrm{d}}}{{\mathrm{C}}_{\mathrm{m}}}\right)=‐\mathrm{bln}{\mathrm{C}}_{\mathrm{m}}+\mathrm{bln}{\mathrm{C}}_{\mathrm{e}}$	qm = 242.113 μg g−1	R2: 0.997
		Kd: 0.0 mol L−1	RMS: 3.849	[29]
		b: 0.564	X2: 0.697
		FCF: 0.0

Various amount of R_L has been obtained at different temperatures and depicted on Fig. 6 (c). 0 < R_L < 1 describes the favorability of the adsorption. Moreover, as it can be seen in Fig. 6 (d), the values of n_f were higher than 1.0 for all experiment temperatures, which indicates the physical adsorption process of MX. K_F ((μg g⁻¹) (L μg⁻¹)^1/n) is the Freundlich constant, and as it is depicted in Fig. 6 (e), by increasing the temperature, K_F increased. As can be seen, K_d is equal to zero, and the RMSE and R² of Freundlich isotherm were lower and higher than those of Langmuir isotherm, respectively. Also, the free capacity fraction (FCF) of zero indicates that the Freundlich isotherm is a better description of the adsorption [29]. B_l (mol μg⁻¹) is the Temkin isotherm constant and is defined as the variation of adsorption energy, which indicates the exothermic adsorption process (B_l > 1). A (L μg⁻¹) and B_l are the Temkin isotherm constants related to the binding and the heat of the adsorption.

Comparing to previous studies on the removal of MX from drinking water, GO@sand had a higher removal efficiency (90%) than activated carbon with an optimum removal of 40% [22], conventional treatment (61%), biofiltration with 0.2 mg L⁻¹ alum (27%) [3], photochemical treatment by UV (76%), thermal treatment(68%), reductive treatment by Na₂SO₃ (68%), and oxidative treatment using H₂O₂ (20%) [41].

Conclusion

In the current study, rGO@sand were synthesized, characterized, and applied for the removal of the halogenated furan MX from aqueous solution. Designing experiments and optimizing the variables were performed using RSM based on BBD. Three parameters such as MX initial concentration, temperature, and adsorbent dose were applied to investigate the adsorption process. The model p value of <0.0001 and high coefficient of determination of 0.999 showed the significance of the model and the excellent compatibility of the model with experimental results. The optimum residual MX concentration of 2 μg L⁻¹ was obtained by setting the parameters of initial concentration (20 μg L⁻¹(, temperature (30 °C), and adsorbent dose (80 g L⁻¹). Investigating the effect of temperature showed the endothermic nature of the equilibrium. The equilibrium adsorption results showed excellent compatibility of the Freundlich isotherm model with the experimental results with the R² of 0.995. These results confirmed that the rGO@sand can be considered as an alternative adsorbent to remove MX from aqueous solutions. However, further investigation is needed to evaluate the column study adsorption.

Compliance with ethical standards

Conflict of interest

There is no conflict of interest.