Glyphosate Removal from Water Using Biochar Based Coffee Husk Loaded Fe 3 O 4

: Glyphosate is an herbicide that is usually used by farmers and is considered harmful to the environment in excess amounts. To address these issues, coffee-husk-biochar-loaded Fe 3 O 4 (CHB-Fe 3 O 4 ) was used as an adsorbent to remove glyphosate from water. CHB-Fe 3 O 4 characteristics such as pH pzc , FTIR, and SEM were measured to understand the properties of this adsorbent. The best conditions for glyphosate removal by CHB-Fe 3 O 4 were obtained at pH 2.0, where the adsorption capacity and percentage removal are 22.44 mg/g and 99.64%, respectively, after 4 h of adsorption. The Freundlich model provided the best ﬁt for the adsorption isotherm, demonstrating multilayer sorption. The most effective model for characterizing the adsorption kinetics was the pseudo-second-order model with a chemical adsorption mechanism. The desorption studies found that the use of 0.1 M NaOH was the best concentration to effectively desorb glyphosate with a desorption percentage of 69.4%. This indicates that CHB-Fe 3 O 4 is a feasible adsorbent for glyphosate removal from water.


Introduction
Glyphosate (C 3 H 8 NO 5 P), also known as N-phosphonomethyl glycine, is one of the active ingredients in herbicides that is commonly used by farmers to eradicate postharvest weeds. In the last few years, glyphosate usage has increased annually, by about 8.6 million kilograms from 1974 to 2014 [1]. The uncontrolled application of herbicides containing this active ingredient (glyphosate) without adhering to the recommended levels can cause serious damage to environmental ecosystems, especially soil and water. Glyphosate enters the water system through several sources such as agricultural runoff from rainfall or irrigation and groundwater leakage from contaminated crop residues [2]. The current detection of glyphosate, and its transforming products amino methyl phosphonic acid, in food, human bodily fluids, waterways, and soil demonstrates an urgent need for further research on how this herbicide affects the environment [3]. The usage of glyphosate exceeding the threshold level can have adverse effects on the environment, human health, and aquatic organisms, and can also impart ecological threats to sustainable agriculture [4]. This herbicide can penetrate into water systems and threaten and cause damage to ecosystems as it generally remains in the environment for a long time and is carcinogenic. The presence of herbicides in drinking water can cause major health problems to humans, therefore it is necessary to monitor the concentration of this herbicide in tap water [5]. A previous study has reported that tap water and river water have the slowest degradation rate of glyphosate. This is

Preparation of Coffee-Husk-Biochar-Loaded Fe3O4 (CHB-Fe3O4)
CHB (3 g) was mixed with FeSO4.7H2O (0.5 M) and FeCl3 (1 M). Then, the mixed solution was shaken for 1 h while heated at 60 °C. Then, the mixture was stirred for 24 h at 25-30 °C by using a magnetic stirrer (Mag-mixer MG600) after adjusting the pH to 10 by using 3 M NaOH. Following that, the filtration process by filtering the mixture using filter paper (qualitative filter paper no. 5C) was performed, and the filtrate was dried for 72 h at 60 °C. This adsorbent is called CHB-Fe3O4.

Characterization of Adsorbent
The pH zero-point charge (pHzpc) was also defined in this study for better understanding of the characteristics of this adsorbent. Samples were subjected to different initial pHs (2.0, 4.0, 6.0, 8.0, and 10.0) of NaCl (0.01 M). Mixture of NaCl (50 mL) with 0.1 g adsorbent was shaken for 24 h using a shaker before being centrifuged for 15 min at 1500 rpm. Then, the pH was measured for calculation of pHzpc. Analysis of SEM images and functional groups CHB-Fe3O4 before and after adsorption were carried out by using SEM (JIED-2300, Shimadzu, Kyoto, Japan) and FTIR (Thermo Scientific Nicolet iS10, Waltham, MA, USA), respectively.

Detection of Glyphosate and Calibration Curve
The technique for detecting glyphosate was conducted by using the method discovered by [29]. The standard solution of 1000 mg/L glyphosate was used to prepare different concentrations of glyphosate (10,20,30,40, and 50 mg/L). For detection of glyphosate, 0.5 mL of each standard glyphosate/sample were added into test tubes containing 0.5 mL of 5% (m/v) C9H6O4 and 0.5 mL of 5% (m/v) Na2MoO4. The solution mixture was then heated in a water bath at a temperature ranging from 80 to 90 °C for 12 min until the sample solution turned a purple color. The sample was cooled down to room temperature, placed in a volumetric flask, and added 1 mL of distilled water. A spectrophotometer from Kyoritsu Chemical-Check (Lab., Corp, Yokohama, Japan) with a wavelength of 570 nm was used to measure the absorbance. From the measurement of the standard solution data, linear absorbance of the standard curve was obtained as the concentration of glyphosate increased, as presented in Figure 2.

Characterization of Adsorbent
The pH zero-point charge (pH zpc ) was also defined in this study for better understanding of the characteristics of this adsorbent. Samples were subjected to different initial pHs (2.0, 4.0, 6.0, 8.0, and 10.0) of NaCl (0.01 M). Mixture of NaCl (50 mL) with 0.1 g adsorbent was shaken for 24 h using a shaker before being centrifuged for 15 min at 1500 rpm. Then, the pH was measured for calculation of pH zpc . Analysis of SEM images and functional groups CHB-Fe 3 O 4 before and after adsorption were carried out by using SEM (JIED-2300, Shimadzu, Kyoto, Japan) and FTIR (Thermo Scientific Nicolet iS10, Waltham, MA, USA), respectively.

Detection of Glyphosate and Calibration Curve
The technique for detecting glyphosate was conducted by using the method discovered by [29]. The standard solution of 1000 mg/L glyphosate was used to prepare different concentrations of glyphosate (10,20,30,40, and 50 mg/L). For detection of glyphosate, 0.5 mL of each standard glyphosate/sample were added into test tubes containing 0.5 mL of 5% (m/v) C 9 H 6 O 4 and 0.5 mL of 5% (m/v) Na 2 MoO 4 . The solution mixture was then heated in a water bath at a temperature ranging from 80 to 90 • C for 12 min until the sample solution turned a purple color. The sample was cooled down to room temperature, placed in a volumetric flask, and added 1 mL of distilled water. A spectrophotometer from Kyoritsu Chemical-Check (Lab., Corp, Yokohama, Japan) with a wavelength of 570 nm was used to measure the absorbance. From the measurement of the standard solution data, linear absorbance of the standard curve was obtained as the concentration of glyphosate increased, as presented in Figure 2.
where C a and C e are the initial and after concentration of glyphosate (mg/L), respectively. q e is the adsorption quantity (mg/g), W is the mass of adsorbent (g), and V is the glyphosate volume (L).

Desorption Studies
Desorption studies were performed using the equilibrium data. NaOH was used as the desorption solvent. Different concentrations of NaOH (0.1, 0.5, and 1 M), contact times (30-1440 min), and temperatures (30,40, and 50 • C) were conducted to determine the effect of these parameters on desorption efficiency. The calculation for desorption percentage is presented in Equation (3) below [16].
where % desorption is the percent desorption (%), De and Ae are the desorption equilibrium and adsorption equilibrium (mg/g), respectively.  Figure 3 shows SEM images of CHB-Fe 3 O 4 before and after adsorption of glyphosate. It can be observed that the surface of CHB-Fe 3 O 4 before adsorption is rougher and contains many available pores compared to the surface of the adsorbent after glyphosate adsorption. This could be due to the interactions between the adsorbate and adsorbent, where the adsorbate (glyphosate) diffuses into the available pores on the surface of the adsorbent (CHB-Fe 3 O 4 ), and the binding of glyphosate to the active sites of the adsorbent containing Fe 3 O 4 . Based on the results of a study by [32], the porosity structure of magnetite coffee husk, obtained from SEM results, is in accordance with the BET results of coffee husk biochar and magnetic coffee husk biochar samples, which are 48.84 m 2 /g and 142.32 m 2 /g, respectively. This information signifies that the adsorption ability of magnetite coffee husk biochar is better than coffee husk biochar.

FTIR Spectra of Adsorbent before and after Glyphosate Adsorption
The FTIR data before and after the adsorption are shown in Figure 4. The band at 3369 cm −1 changed to 3341 cm −1 after adsorption, indicating a strong interaction between O-H surface groups and glyphosate [33,34]. The band at 2119 cm −1 changed to 2117 cm −1 after adsorption of glyphosate which was specific to the functional group of C≡C [35]. The peak band at 1630 cm −1 's move to 1629 cm −1 after adsorption is attributed to the C=O group [36].The peak at the 1118 cm −1 and 1124 cm −1 bands demonstrates the (C-O) polysaccharide carbohydrate region [37]. In contrast, bands at 883 and 888 cm −1 were assigned to C-H vibrations in the aromatic structures [36]. The bands at 632 cm −1 and 598 cm −1 are the result of the stretching vibrational interaction which connected to the metal-oxygen bonds in the Fe-O bond of the Fe 3 O 4 crystal lattice [38]. Based on previous research conducted by [32], comparison of FITR results showed that the OH, C-H, and C=C functional groups in raw coffee husk become more obvious after turned into coffee husk biochar, because of hydrothermal carbonization process making the intensity of the FTIR spectrum of the coffee husk biochar become higher and more obvious. The thermal process causes the raw coffee husk to have more OH and C=C functional groups. Furthermore, the chemical treatment by adding magnetite (Fe 3 O 4 ) into coffee husk biochar has resulted in more oxygen-containing functional groups in the adsorbent. respectively. This information signifies that the adsorption ability of magnetite coffee husk biochar is better than coffee husk biochar.

FTIR Spectra of Adsorbent before and after Glyphosate Adsorption
The FTIR data before and after the adsorption are shown in Figure 4. The band at 3369 cm −1 changed to 3341 cm −1 after adsorption, indicating a strong interaction between O-H surface groups and glyphosate [33,34]. The band at 2119 cm −1 changed to 2117 cm −1 after adsorption of glyphosate which was specific to the functional group of C≡C [35]. The peak band at 1630 cm −1 's move to 1629 cm −1 after adsorption is attributed to the C=O group [36].The peak at the 1118 cm −1 and 1124 cm −1 bands demonstrates the (C-O) polysaccharide carbohydrate region [37]. In contrast, bands at 883 and 888 cm −1 were assigned to C-H vibrations in the aromatic structures [36]. The bands at 632 cm −1 and 598 cm −1 are the result of the stretching vibrational interaction which connected to the metal-oxygen bonds in the Fe-O bond of the Fe3O4 crystal lattice [38]. Based on previous research conducted by [32], comparison of FITR results showed that the OH, C-H, and C=C functional groups in raw coffee husk become more obvious after turned into coffee husk biochar, because of hydrothermal carbonization process making the intensity of the FTIR spectrum of the coffee husk biochar become higher and more obvious. The thermal process causes the raw coffee husk to have more OH and C=C functional groups. Furthermore, the chemical treatment by adding magnetite (Fe3O4) into coffee husk biochar has resulted in more oxygencontaining functional groups in the adsorbent.

Interaction Glyphosate with CHB-Fe3O4
After adsorption of glyphosate using CHB-Fe3O4, it is assumed that there will be several interactions occurring between the adsorbent and the adsorbate as shown in Figure  5. Previous studies explained that there are several interactions defined in glyphosate adsorption such as pore diffusion, electrostatic interactions, and H-bonding [39,40]. Pore diffusion is the interaction that is most likely to occur, because the glyphosate will move into the adsorbent through the available pore sites.

Interaction Glyphosate with CHB-Fe 3 O 4
After adsorption of glyphosate using CHB-Fe 3 O 4 , it is assumed that there will be several interactions occurring between the adsorbent and the adsorbate as shown in Figure 5. Previous studies explained that there are several interactions defined in glyphosate adsorption such as pore diffusion, electrostatic interactions, and H-bonding [39,40]. Pore diffusion is the interaction that is most likely to occur, because the glyphosate will move into the adsorbent through the available pore sites.

Initial pH Effects
The pH is one of the most important parameters that affect the surface charge of the adsorbate (glyphosate) and adsorbent (CHB-Fe3O4). As shown in Figure 6a the highest adsorption was obtained at pH 2, with an adsorption capacity of 3.07 mg/g after 30 min of adsorption. Figure 6b shows that the pHzpc was at pH 2. If the pH < pHzpc, the adsorbent surface will be positively charged, while if the pH > pHzpc, the surface of the adsorbent will be negatively charged. At pH 2, the surface of the adsorbent is positively charged, while glyphosate is negatively charged because the pKa of glyphosate is 2.23 (<2.23 is negative charge). Thus, electrostatic interactions between the net charge of the adsorbent and the negative charge of glyphosate occurred. The lower pH caused the deprotonation of amine (NH − ), phosphate (PO3 − ), and carboxyl (COOH − ) groups in glyphosate molecules [10,11,15,41].

Initial Concentration Effects
The effect of the initial concentration is shown in Figure 7. The result displays an increase in adsorption capacity from 2.77 to 17.13 mg/g when the glyphosate concentration increases. This is because the quantity of glyphosate molecules increases with a b Figure 5. Proposed mechanism of glyphosate adsorption.

Initial pH Effects
The pH is one of the most important parameters that affect the surface charge of the adsorbate (glyphosate) and adsorbent (CHB-Fe 3 O 4 ). As shown in Figure 6a the highest adsorption was obtained at pH 2, with an adsorption capacity of 3.07 mg/g after 30 min of adsorption. Figure 6b shows that the pH zpc was at pH 2. If the pH < pH zpc , the adsorbent surface will be positively charged, while if the pH > pH zpc , the surface of the adsorbent will be negatively charged. At pH 2, the surface of the adsorbent is positively charged, while glyphosate is negatively charged because the pKa of glyphosate is 2.23 (<2.23 is negative charge). Thus, electrostatic interactions between the net charge of the adsorbent and the negative charge of glyphosate occurred. The lower pH caused the deprotonation of amine (NH − ), phosphate (PO 3 − ), and carboxyl (COOH − ) groups in glyphosate molecules [10,11,15,41].

Initial pH Effects
The pH is one of the most important parameters that affect the surface charge of the adsorbate (glyphosate) and adsorbent (CHB-Fe3O4). As shown in Figure 6a the highest adsorption was obtained at pH 2, with an adsorption capacity of 3.07 mg/g after 30 min of adsorption. Figure 6b shows that the pHzpc was at pH 2. If the pH < pHzpc, the adsorbent surface will be positively charged, while if the pH > pHzpc, the surface of the adsorbent will be negatively charged. At pH 2, the surface of the adsorbent is positively charged, while glyphosate is negatively charged because the pKa of glyphosate is 2.23 (<2.23 is negative charge). Thus, electrostatic interactions between the net charge of the adsorbent and the negative charge of glyphosate occurred. The lower pH caused the deprotonation of amine (NH − ), phosphate (PO3 − ), and carboxyl (COOH − ) groups in glyphosate molecules [10,11,15,41].

Initial Concentration Effects
The effect of the initial concentration is shown in Figure 7. The result displays an increase in adsorption capacity from 2.77 to 17.13 mg/g when the glyphosate concentration increases. This is because the quantity of glyphosate molecules increases with a b

Initial Concentration Effects
The effect of the initial concentration is shown in Figure 7. The result displays an increase in adsorption capacity from 2.77 to 17.13 mg/g when the glyphosate concentration increases. This is because the quantity of glyphosate molecules increases with increasing concentration, resulting in a greater tendency for attachment of the adsorbate onto the adsorbent surface [42]. increasing concentration, resulting in a greater tendency for attachment of the adsorbate onto the adsorbent surface [42].

Adsorption Isotherm Studies
Isotherm models are a crucial concept for observing or predicting the processes that occur during adsorption equilibrium [43]. The Langmuir and Freundlich models are the most commonly used models for glyphosate adsorption and are used for isotherm analysis in this study. The Langmuir isotherm shows that active sites of the adsorbent have equal energy and that the adsorption process primarily occurs with regards to monolayers [39,40,44]. In contrast, the Freundlich isotherm shows that energy declines logarithmically as the number of molecules adsorbed on the adsorbent sites increases [7,9,40,45]. The experiment was conducted by adding 0.05 g of adsorbent into 25 mL of glyphosate at different initial concentrations ranging from 10 to 50 mg/L. The linear equations for the Langmuir and Freundlich models are listed in Table 1.

Langmuir
C /q e = C q max +1/(K 1 q max ) R 1 1 K C

Freundlich
Ln q e = L n k F + 1 n ×L n C b Note: qe = adsorption capacity (mg/g), Kl = equilibrium constant of adsorption (L/mg), Kf = equilibrium constant of adsorption, qmax = maximal adsorption capacity (mg/g), Ce and Ca (mg/L) = equilibrium concentration and initial concentration, respectively. qmax in the Langmuir equation indicates that the maximum adsorption capacity obtained in this study was 18.315. RL is the separation factor, where 0 < (favorable), > 1 (unfavorable) and = 1 (linear). Table 2 shows that the RL value of glyphosate adsorption was 0.0146, indicating that the adsorption was favorable [16]. The graph in Figure 8a shows that the regression coefficient R 2 in the Freundlich equation was close to one, making it more suitable than the Langmuir model (Figure 8b).

Adsorption Isotherm Studies
Isotherm models are a crucial concept for observing or predicting the processes that occur during adsorption equilibrium [43]. The Langmuir and Freundlich models are the most commonly used models for glyphosate adsorption and are used for isotherm analysis in this study. The Langmuir isotherm shows that active sites of the adsorbent have equal energy and that the adsorption process primarily occurs with regards to monolayers [39,40,44]. In contrast, the Freundlich isotherm shows that energy declines logarithmically as the number of molecules adsorbed on the adsorbent sites increases [7,9,40,45]. The experiment was conducted by adding 0.05 g of adsorbent into 25 mL of glyphosate at different initial concentrations ranging from 10 to 50 mg/L. The linear equations for the Langmuir and Freundlich models are listed in Table 1. Table 1. Linear equations of isotherm models.

Model Equation Supporting Equation
Langmuir C e /q e = C e q max +1/(K 1 q max R L = 1 (1+KL×Ca)

Freundlich
Ln q e = L n k F + 1 n × L n C b Note: q e = adsorption capacity (mg/g), K l = equilibrium constant of adsorption (L/mg), K f = equilibrium constant of adsorption, q max = maximal adsorption capacity (mg/g), C e and C a (mg/L) = equilibrium concentration and initial concentration, respectively. q max in the Langmuir equation indicates that the maximum adsorption capacity obtained in this study was 18.315. R L is the separation factor, where 0 < (favorable), > 1 (unfavorable) and = 1 (linear). Table 2 shows that the R L value of glyphosate adsorption was 0.0146, indicating that the adsorption was favorable [16]. The graph in Figure 8a shows that the regression coefficient R 2 in the Freundlich equation was close to one, making it more suitable than the Langmuir model (Figure 8b).   Figure 9 shows the effect of contact time on glyphosate adsorption. The adsorption capacity increased rapidly from 30 to 60 min, and then steadily increased to 240 min with an adsorption capacity of 22.29 mg/g, then decreased and increased again until 1440 min. This is because the adsorbent's surface site was saturated with adsorbate, and this condition could lead to adsorption or desorption of the glyphosate. Finally, adsorption capacity equilibrium was attained at 1440 min with an adsorption capacity of 22.44 mg/g. Kinetic analysis can be helpful in determining the mechanism and pace of the adsorption process [18]. In this study, pseudo-first-order (Equation (4)) and pseudo-secondorder (Equation (5)) models were used to predict the adsorption kinetic of this study.

Adsorption Kinetic Studies
Log (qe − qt) = Log qe − k1t (4) t/qt = (1/k2 q 2 qe) + t/qe (5)  Figure 9 shows the effect of contact time on glyphosate adsorption. The adsorption capacity increased rapidly from 30 to 60 min, and then steadily increased to 240 min with an adsorption capacity of 22.29 mg/g, then decreased and increased again until 1440 min. This is because the adsorbent's surface site was saturated with adsorbate, and this condition could lead to adsorption or desorption of the glyphosate. Finally, adsorption capacity equilibrium was attained at 1440 min with an adsorption capacity of 22.44 mg/g.   Figure 9 shows the effect of contact time on glyphosate adsorption. The adsorption capacity increased rapidly from 30 to 60 min, and then steadily increased to 240 min with an adsorption capacity of 22.29 mg/g, then decreased and increased again until 1440 min. This is because the adsorbent's surface site was saturated with adsorbate, and this condition could lead to adsorption or desorption of the glyphosate. Finally, adsorption capacity equilibrium was attained at 1440 min with an adsorption capacity of 22.44 mg/g. Kinetic analysis can be helpful in determining the mechanism and pace of the adsorption process [18]. In this study, pseudo-first-order (Equation (4)) and pseudo-secondorder (Equation (5)) models were used to predict the adsorption kinetic of this study.
Log (q e − q t ) = Log q e − k 1 t (4) t/q t = (1/k 2 q 2 qe) + t/qe (5) where k 1 (min −1 ) is the constant of the pseudo-first order, and k 2 (g/mg min −1 ) is the constant of the pseudo-second order. t (time) (min), and q t (mg/g) is the adsorption capacity at time.
The results show that the pseudo-second-order model (R = 0.999) ( Figure 10) is more appropriate for describing the kinetic study data in this research, with an adsorption capacity of 22.57 mg/g (Table 3). These results are in line with those reported by a previous study [46], which showed that glyphosate adsorption using nanosized copper hydroxide-modified resin fits the pseudo-second-order model, and chemisorption could be the ratedetermining process.
where k1 (min −1 ) is the constant of the pseudo-first order, and k2 (g/mg min −1 ) is the constant of the pseudo-second order. t (time) (min), and qt (mg/g) is the adsorption capacity at time.
The results show that the pseudo-second-order model (R = 0.999) ( Figure 10) is more appropriate for describing the kinetic study data in this research, with an adsorption capacity of 22.57 mg/g (Table 3). These results are in line with those reported by a previous study [46], which showed that glyphosate adsorption using nanosized copper hydroxidemodified resin fits the pseudo-second-order model, and chemisorption could be the ratedetermining process.

Desorption Study
The desorption percentages of glyphosate are shown in Figure 11. The results showed that the NaOH concentration significantly affects the release of glyphosate from the adsorbent. Increases in the concentrations of NaOH have the contrary effect on the desorption percentage, as the concentration of glyphosate released decreases from 69.40 to 5.99% (Figure 11a). Subsequently, different temperatures were also tested for desorption effectivity, and the results showed that 30 °C is the best temperature for desorption of glyphosate ( Figure 11b). Interestingly, the desorption of glyphosate varied as the time increased. The desorption percentage increased and then decreased for up to 1440 min, indicating the occurrence of adsorp/desorp during these processes (Figure 11c). The best results for desorption of glyphosate were obtained at a temperature of 30 °C by using NaOH (0.1 M) for 30 min, with a desorption percentage of 69.4%.

Desorption Study
The desorption percentages of glyphosate are shown in Figure 11. The results showed that the NaOH concentration significantly affects the release of glyphosate from the adsorbent. Increases in the concentrations of NaOH have the contrary effect on the desorption percentage, as the concentration of glyphosate released decreases from 69.40 to 5.99% (Figure 11a). Subsequently, different temperatures were also tested for desorption effectivity, and the results showed that 30 • C is the best temperature for desorption of glyphosate ( Figure 11b). Interestingly, the desorption of glyphosate varied as the time increased. The desorption percentage increased and then decreased for up to 1440 min, indicating the occurrence of adsorp/desorp during these processes (Figure 11c). The best results for desorption of glyphosate were obtained at a temperature of 30 • C by using NaOH (0.1 M) for 30 min, with a desorption percentage of 69.4%. Various materials have been reported in the literature for the adsorption-based removal of glyphosate, as shown in Table 4. The results indicate that CHB-Fe3O4 is a feasible adsorbent for the removal of glyphosate from water.  Various materials have been reported in the literature for the adsorption-based removal of glyphosate, as shown in Table 4. The results indicate that CHB-Fe 3 O 4 is a feasible adsorbent for the removal of glyphosate from water.

Conclusions
In this study, biochar-based coffee-husk-loaded Fe 3 O 4 (CHB-Fe 3 O 4 ) was used as an adsorbent to remove glyphosate from water. The results show that the best glyphosate adsorption was achieved at pH 2 solution after a contact time of 1440 min, with maximum adsorption capacity of 22.44 mg/g. The Freundlich model provided the best fit to the isotherm analysis, demonstrating multilayer sorption. The most effective model for characterizing the adsorption kinetics was the pseudo-second-order model, with chemical adsorption as the rate-limiting phase, possibly including valence forces by sharing or exchanging electrons between the adsorbent and the adsorbate. The desorption studies showed that the use of 0.1 M NaOH is the best concentration to achieve a desorption percentage of 69.4% in 30 min.