Potential low-cost carbon-based adsorbent from gold mine tailings for anionic dye removal

In this study, a mesoporous carbon (CMK-3) was successfully synthesized using mesoporous silica (SBA-15) prepared from gold mine tailings (sample CGMT) for removal of anionic dye from aqueous solution. For comparison, CMK-3 was prepared by the same method from pure silica (sample CT), and the other CMK-3 sample was prepared by a one-pot route mixing with Pluronic P123 (sample CP). The effect of the carbonization time on the synthesis of all CMK-3 samples was investigated, and the samples were characterized by X-ray diffraction and N2 adsorption–desorption. The sample with the highest surface area was chosen as an adsorbent, for each CMK-3 obtained from different methods. Batch adsorption experiments were studied to determine the influence of pH, contact time, and initial dye concentration. The adsorption kinetics obeyed the pseudo-second-order model. All carbon-based adsorbents were observed to be quite effective for the removal of dye with adsorption percentage in the order of CP> CT>CGMT. The maximum adsorption capacities were 188.99 and 204.57 mg·g 1 for CT and CGMT, respectively. The comparative results of all carbon-based adsorbents show that CGMT can be applied as a low-cost alternative to CT for dye removal.


HIGHLIGHTS
• CMK-3 was synthesized from gold mine waste (C GMT ) for the first time.
• The usage of C GMT as low-cost adsorbent for the removal of dye was investigated. • For comparison, other adsorbents synthesized from different methods were investigated as an adsorbent in the dye adsorption.
• The specific surface area, pore volume and pore size dominate the adsorption capacity.
• C GMT can be applied as a novel, low-cost and alternative adsorbent for dye removal.

GRAPHICAL ABSTRACT INTRODUCTION
The development of mining has become very important with the increase in industrialization. However, with the industrial revolution, mining and operating activities have caused the environment to be more polluted (Abdul-Wahab & Marikar ). Recently, environmental pollution emerging with the development of mining has reached important dimensions in the world. The residue wastes, named as tailings, as a result of mining activities are a threat to the health of nearby communities and the environment (Goltz & Barnwal ). Therefore, it is very important to dispose of this type of mining tailings safely.
Gold mine tailings (GMT) are the ore waste of gold mines and are typically a mud-like material. Although GMT are stored in waste dams, the amount of the tailings is increasing day by day with the development of gold mining. Thus, the treatment of GMT is urgent.
Porous carbon materials have found great interest in many applications due to their surface areas and physicochemical properties. Porous carbon materials can be classified into three main groups based on their pore diameters: microporous, mesoporous, and macroporous (Liang et al. ). Among them, activated porous carbons are one of the most important types of industrial carbons (Manocha ). Activated carbon is commonly used in adsorption processes, due to its significant adsorption capacity and thermal stability. However, because of the dominant presence of micropores in its structure, it normally demonstrates low uptake capacity (Tsang et al. ; Bello & Ahmad ; Spagnoli et al. ). In this sense, recent studies have focused on the synthesis of mesoporous carbon-based adsorbents to improve the adsorption ability of the activated carbon.
Mesoporous carbon materials have a high surface area and ordered mesostructures, and they are used as catalysts, adsorbent, and advanced electronic materials in many scientific disciplines (Liang et al. ). CMK-n is the most important representative of ordered mesoporous carbon materials. SBA-n type mesoporous silica materials are generally used as templates in the synthesis of CMK-n due to their regular structure. CMK-3 is a diverse replica of the SBA-15, and has the same hexagonal structure (Wang et al. ). It has been used as adsorbents (Ezzeddine et al. ), electrocatalysts (Yu et al. ), and catalysts (Peng et al. ).
Synthetic dye wastewater is another important problem for the environment and human health. Depending on the diversity of these organic and inorganic synthetic compounds used in dyeing and other processes, the characteristics of the wastewater are different. Synthetic dye wastewaters discharged to surrounding waters reduce the light permeability in the water environment and affect the photosynthetic activity negatively (Correia et al. ). Moreover, these dyes are difficult to biodegrade due to their synthetic source and complex structure. For this reason, it is very significant to develop efficient treatment processes for the adsorption of synthetic dyes from wastewater.
Today, due to the inadequacy of water resources and more controlled discharge standards, industries require highly efficient and economical treatment technologies. In addition, recycling and reuse of treated water is an option (Manu & Chaudhari ). Various physical, chemical, and biological techniques such as chemical oxidation, solvent extraction, membrane separation, ion exchange, biodegradation, and adsorption are used for the removal of dyes. The adsorption process supplies an easy and common way to effectively remove dye from wastewaters (Wong et al. ; Yagub et al. ; Bilinska et al. ; Mokhtar et al. ; Liu et al. ). Much research has been performed to determine proper and affordable adsorbents for dye removal from wastewater.
Recently, the usage of CMK-3 as an adsorbent in the removal of dye adsorbent has received growing attention due to its perfect textural characteristics (Yu et al. ). Ezzeddine and colleagues produced CMK-3 from SBA-15 synthesized from pure silica for removal of methylene blue. They revealed that the synthesized CMK-3 sample exhibited very high adsorption capacity (Ezzeddine et al. ). Liu and co-workers revealed that CMK-3 showed stronger adsorption for Reactive Black 5 adsorption than did other adsorbents due to its open and interconnected pore features (Liu et al. ). Tripathi et al. () noticed that the dye removal capacity of ordered mesoporous carbon is directly proportional to the specific surface area and pore volume. Liu and colleagues revealed that CMK-3 exhibited larger adsorption rates for direct yellow 12 adsorption than the adsorption rates of commercial activated carbon (Liu et al. ).
In this study, the synthesis of CMK-3 using SBA-15 from GMT and the adsorption properties of the synthesized CMK-3 were investigated for the first time. In addition, two CMK-3 samples were synthesized using SBA-15 from pure silica and Pluronic P123 to compare their structural characterizations and adsorption properties. The synthesized samples were characterized by X-ray diffraction (XRD), N 2 adsorption-desorption, and high-resolution transmission electron microscopy (HRTEM). To determine the adsorption behavior of the prepared samples, the effect of various parameters were studied in detail. Adsorption isotherms and kinetics parameters were also analyzed.

EXPERIMENTAL Materials and methods
The chemicals used to synthesize SBA-15 and CMK-3 were Pluronic P123 from Aldrich, and tetraethyl orthosilicate (TEOS), hydrochloric acid (HCl), sulfuric acid (H 2 SO 4 ), ethanol, sucrose, and hydrofluoric acid (HF) from Merck. Bromcresol green (BCG) (molar mass 698.02 g·mol À1 ) was purchased from Merck ( Figure 1). GMT were obtained from Bergama Ovacık Gold Mine Treatment Plant, Turkey. It was treated by the alkali fusion process described in a previous study (Yilmaz Sari & Piskin a).
XRD data were measured by a diffractometer (PANalytical X'Pert Pro) equipped with CuKα radiation (45 kV/ 40 mA). Nitrogen adsorption-desorption measurement was carried out at 77.4 K using a volumetric adsorption system (Micromeritics ASAP 2020). Before the analyses, samples were degassed at 573 K. HRTEM measurements were obtained on a transmission electron microscope (JEOL-2100, 200 kV) with LaB 6 electron source. The elemental concentration of extracted solution obtained from the GMT was analyzed by inductively coupled plasma optical emission spectroscopy (Perkin Elmer Optima 2100 DV). The amounts of Si and Al in the extracted solution were found as 28,330 and 1,150 mg·L À1 , respectively, indicating that GMT are rich in SiO 2 .

Synthesis of silica template
Silica template SBA-15 was synthesized, for CMK-3 synthesis, from two raw materials, pure silica TEOS and GMT, according to the different procedures described in the literature (Zhao et al. ; Yilmaz Sari & Piskin a, b). The obtained samples from TEOS and GMT were labeled as S T and S GMT , respectively.

Synthesis of CMK-3
Mesoporous carbon CMK-3 was prepared from two different routes. In the first method, S T or S GMT samples were used as the template, and sucrose was used as the carbon source in the synthesis. Typically, an amount of 1 g of SBA-15 template was dissolved in some distilled water and then mixed with 0.6 g of sucrose and 0.1 g of H 2 SO 4 . After dissolution, the obtained solution was held in an oven at 100 C for 6 h, and then the temperature was increased to 160 C for 6 h. In order to have completely polymerized and carbonized sucrose inside the pores of the template, 2 g of distilled water, 0.05 g of H 2 SO 4 , and 0.4 g of sucrose were again introduced to the mixture and then it was treated thermally in the same way given above. The samples were then carbonized in argon flow at 900 C for 4, 5, and 6 h. Finally, the obtained products were washed with 5% HF followed by two more times with ethanol and then dried at 120 C. Samples obtained from S T and S GMT were labeled as C T -x and C GMT -x, respectively, where 'x' denotes the carbonization time.
In the second route, CMK-3 was prepared via the selfassembly method by using P123 copolymer, and the synthesis procedure was conducted based on the study done by Ting et al. () with some modifications. Briefly, the P123 copolymer, initially taken as 1.47 g, was first dissolved in distilled water and 2 M HCl with stirring at room temperature. Then, 0.63 g of sucrose was used as a carbon source, and H 2 SO 4 was added to the mixture. Then the obtained solution was stirred at room temperature for 30 min. TEOS (2.55 g) was introduced into the mixture under vigorous stirring. Then, the resulting solution was introduced into the autoclave and held at 100 C for 24 hours. The obtained sample was dried at 100 C for 6 hours, and then the temperature was set to 160 C for another 6 hours. After cooling the sample, it was washed with 10% HF and ethanol and dried overnight at 70 C for removal of the undesired substances. The resulting product was labeled as C P -x where 'x' denotes the carbonization time.

Adsorption experiments
Batch adsorption experiments were conducted to observe the effect of pH, initial concentration of dye, and contact time. All experiments were done on a Zhicheng incubator shaker with a shaker speed of 150 rpm at room temperature. For the adsorption studies, a 20 mL solution of known dye concentration and 0.006 g adsorbent were placed in an Erlenmeyer. The dye concentrations of the solutions were obtained by a UV-visible spectrophotometer (Perkin Elmer Lambda 35 model) with detecting wavelength at 445 nm. The dye removal and adsorption capacity of the adsorbent were found by the following equations: and where C o (mg·L À1 ), C t (mg·L À1 ), and C e (mg·L À1 ) are liquid phase equilibrium concentrations of BCG at the start, time t, and at equilibrium, respectively. q e is the amount of adsorbate adsorbed at equilibrium in mg.g À1 , V is the volume in L of the BCG dye solution, and W is the mass in g of the carbon-based adsorbent.

Characterization of the materials
The XRD patterns of S T , S GMT , and CMK-3 samples synthesized from different procedures are shown in Figure 2. As shown in Figure 2(a), S T demonstrates a very intense diffraction peak (100) and two obvious peaks (110) and (200), indicating that mesoporous structures form SBA-15 with a 2D hexagonal array and p6mm symmetry with high order (Santhanaraj et al. ). S GMT only has the (100) reflection, related to the concentration of high electronic density of silanol groups (Si-OH) on the sample surface ( Figure 2 . C T -4 and C T -5 samples have the weak (110) and (200) reflections. An ordered structure occurred in C T -6, and it has a uniform 2D hexagonally ordered structure, similar to S T . When the carbon sample was synthesized using the S GMT silica template, the XRD pattern revealed only a broad (100) peak, indicating a less ordered pore structure due to distortion of the hexagonal pore arrangement (Figure 2(b)). This result is similar to those of the S GMT , showing that CMK-3 carbons are a perfect replica of the SBA-15 template. For the C P -6 and C P -5 samples, the XRD analysis exhibited three (100, 110, and 200) reflections, which were the well-ordered CMK-3 mesostructure (Figure 2(c)). XRD analysis results showed that the C T and C P samples had a more ordered structure with an increase of the carbonization time. Figure 3 shows the N 2 adsorption-desorption isotherms and the related pore size distribution (PSD) curves of the synthesized carbon materials. These isotherms are attributed to a type IV isotherm with typical hysteresis loops in the mesopore range for all the carbon samples. As shown in Figure 3, while C P -5 has the highest N 2 adsorption, C GMT -5 has the lowest N 2 adsorption. The N 2 adsorptiondesorption isotherms of C T samples showed the largest hysteresis loop due to the strong chemical interaction of the pore wall with N 2 . N 2 adsorption-desorption isotherms of C p -5 demonstrated a smooth continuous increase, implying that it has larger pores. Table 1 presents the structural parameters of the carbon materials calculated from the N 2 adsorption isotherm data. The C GMT -x samples have a smaller surface area and pore volume values than those of the C T -x. The obtained result is compatible with the XRD, indicating that C T -x had a more ordered mesoporous structure as compared with C GMT -x. C T -4 has a higher surface area compared to the other C T -x samples, and also, C GMT -6 has a higher surface area compared to the other C GMT -x samples. From the table, it can be seen that C P -5 showed the highest surface area, pore volume, and pore size values compared to the other carbon samples. When the carbonization time increases from 4 to 5 h, specific surface area, pore  volume, and pore size values of C P increase noticeably. However, the C T -5 and C GMT -5 have the lowest surface area and pore volume compared to the other C T -x and C GMT -x samples, respectively. Nevertheless, different trends are observed for all the other samples, and there is no proper relation between samples obtained at different carbonization time in terms of the textural properties. The PSD of C P -6 exhibited a narrow distribution, while the PSD of C P -4 and C P -5 exhibited broad distributions. The large surface area can provide interactive sites for dye molecules within mesoporous samples (Zhuang et al. ; Chen et al. ). For this reason, C T -4, C GMT -6, and C P -5 with the highest specific surface area were selected as adsorbents for BCG removal in the adsorption studies regardless of whether the adsorbent has an ordered structure.
The HRTEM and TEM images of C T -4, C GMT -6, and C P -5 are shown in Figure 4. As can be seen, C T -4 and C P -5 samples possess a highly ordered hexagonal array of uniform pore structure (Figure 4(a) and 4(e)). Also, the TEM image of C T -4 and C P -5 samples exhibits wellorganized mesopores parallel to each other (Figure 4(b) and 4(f)). However, the HRTEM and TEM images of C GMT -6 became obscure, and clearly reveal that this material has a disordered wormlike mesoporous structure (Figure 4(c) and 4(d)). Further, the HRTEM and TEM analysis results also support the conclusion from XRD analysis.
Zeta potential analysis of the C T -4, C GMT -6, and C P -5 as a function of solution pH are given in Figure 5. The zeta potentials of the samples continuously decreased with the increase of solution pH. According to measurements, C T -4, C GMT -6, and C P -5 have zero charges of around 3, 2.7, and 2.5, respectively, and a positive surface of adsorbents, at pH values below the zero point.
Hence, an electrostatic attraction occurred between the anionic dye and positive charges of adsorbent surfaces.

Adsorption studies
The solution pH is one of the important parameters controlling the removal of dyes from wastewater. The pH effect on the removal of BCG was studied by changing the pH value from 2 to 12 while the contact time was kept as 30 min, initial dye concentration was fixed at 25 mg·L À1 and the amount of the adsorbent was 0.01 g. As demonstrated in Figure 6, the adsorption capacity of all adsorbents was high at low pH value but then decreased with the increasing pH of the dye solution. In pH ¼ 2, a high electrostatic attraction occurred between dye anions and adsorbent because of the tendency of H þ ions to interact with the adsorbent surface. The number of negatively charged surface sites on the adsorbent increases with the increase in solution pH so that the adsorbent does not favor the adsorption of anionic dye due to the repulsion (Malik ). From Figure 6, the dye removal capacities for all three adsorbents were maximum at the initial pH 2. A similar type of behavior is also reported in the literature for the BCG adsorption with different adsorbents (Shokrollahi et al. ). According to the results, subsequent adsorption experiments on all carbon-based materials were performed at pH 2.
The effect of contact time and initial dye concentration on the removal of BCG dye on the carbon-based adsorbents was investigated at pH 2, and the amount of adsorbents was 0.006 g for various time intervals. From Figure 7, the adsorption process of all adsorbents was fast in the first 60 min, and after 60 min, increase in contact time exhibited no important effect on the adsorption capacity value and remained nearly unchanged. The reason for the rapid adsorption of BCG at the beginning is the presence of a large number of sorption sites on the surface of the carbon-based adsorbent. When the active sites on the surface are filled, the adsorption rate decreases and eventually reaches equilibrium (Wu et al. ). It was observed that the adsorption of BCG onto C T -4 and C GMT -6 exhibited a similar trend. The highest adsorption capacities of C T -4, C GMT -6, and C P -5 adsorbents were found as 188.99, 204.57, and 328.74 mg·g À1 , respectively. It was also seen that the adsorption capacities increased by increasing the dye concentrations for all adsorbents, because more dye molecules can be found at higher initial concentrations. This leads to a greater propellant to accomplish the resistance of mass transfer of the dye among aqueous and solid phases (Mohammadi et al. ). It was noted that BCG was adsorbed the fastest on the C P -5 surface according to the C T -4 and C GMT -6 adsorbents because of the structural properties of C P -5.
The comparison of the maximum adsorption capacity of dye onto adsorbents synthesized from various wastes is presented in Table 2. The C GMT -6 has very high adsorption capacity as compared to the other samples synthesized from different wastes reported in the literature.

Adsorption kinetics
The kinetics of sorption is a significant factor for the design of a suitable adsorption process (McManamon et al. ). The various kinetic models were applied for a better understanding of the mechanism of dye adsorption on the adsorbent. The pseudo-first-order model is described below (Mathew & Tandon ): where q e and q t (mg.g À1 ) are adsorption capacities at equilibrium and time t, respectively, and k 1 (min À1 ) is a constant of pseudo-first-order (Harinath et al. ). The slope and the intercept of the plot of log(q e À q t ) versus t were obtained to find k 1 and q e . The pseudo-second-order equation can be explained in a linear form as follows (Yang et al. ; Yan et al. ): where k 2 (g·mg À1 ·min À1 ) is a constant. The slope and the intercept of the plot of t/q t versus t were obtained to find k 2 and q e . The adsorption kinetic plots are presented in Figure 8, and the calculated model parameters for all carbon-based adsorbents are summarized in Table 3. Comparison of the estimated adsorption capacity (q e,exp ) and the correlation coefficient (R 2 ) values of the kinetic models revealed that the best fit of experimental data was to the pseudo-second-order. The R 2 values of the pseudo-secondorder model were higher than the other models. The q e values and the experimental equilibrium sorption capacities were close to each other for all adsorbents. However, both the pseudo-first and pseudo-second orders cannot determine the mechanism of diffusion. Therefore, the intra-particle diffusion model recommended by Weber and Morris (Weber & Morris ) was used to analyze the kinetic parameters to determine if the intra-particle diffusion is a rate-limiting step in the adsorption of BCG on the adsorbent. The intra-particle diffusion model is expressed by the following formula: where k id (mg.g À1 ·min À1/2 ) is a constant (Harinath et al. ). C (mg·g À1 ) value provides knowledge about the thickness of the boundary layer. When a straight plot of q t versus t 0.5 passes through the origin, the adsorption process is found to be controlled by intra-particle diffusion only (Zolgharnein & Rastgordani ). The intra-particle diffusion plots for all adsorbents are given in Figure 8(c), and the calculated parameters are summarized in Table 3. All intra-particle diffusion plots appear to be divided into two regions (boxes 'a' and 'b' shown in the figure) during the adsorption process, and these plots do not cross the origin, showing that the intra-particle diffusion is not ratelimiting. The first linear region (a) indicates the external mass transfer and diffusion of the dye molecules to active sites on the surface of the adsorbents, while the subsequent

Adsorption isotherms
The adsorption isotherm models represent the relations between the concentration of adsorbate on the solid phase and the liquid phase equilibrium concentration of the adsorbate (Lata et al. ; Yan et al. ). In this study, Langmuir, Freundlich, and Temkin isotherms were used to better describe the adsorption process. The Langmuir model supposes that the adsorption process is a single-layered adsorption mechanism (Langmuir ). The Langmuir model is presented as: where q m (mg·g À1 ) is the maximum uptake capacity, K L is a constant, and C e is the equilibrium adsorbate concentration (Zhang et al. ). The Freundlich model explains multilayer and reversible adsorption and presumes a heterogeneous adsorbent surface as below: where K F is a constant and n is the heterogeneity factor (Freundlich ).
The Temkin isotherm represents adsorbent-adsorbate molecule interactions on the adsorption isotherm. An assumption is made that as surface coverage increases, the heat of adsorption of all molecules in the layer decreases linearly (Mall et al. ). The model equation is given by: where K T is the isotherm constant, and B is a constant associated with the adsorption heat. The obtained parameters derived from all isotherm models are summarized in Table 4. Figure 9 presents the plots of Langmuir, Freundlich, and Temkin models of all carbon-based adsorbents. According to the results of the dye adsorption isotherm experiments, the obtained parameters of C T -4 and C GMT -6 demonstrate a fitting with the Temkin model, having R 2 values of 0.985 and 0.990 with the heat of sorption values of 47.96 and 55.47, respectively. They also fit the Langmuir isotherm with correlation coefficients 0.979 and 0.980, and the maximum uptake capacity of C T -4 and C GMT -6, attributed to the complete monolayer coverages, is calculated to be 232.56 and 277.78 mg·g À1 . For C P -5 adsorbent, the Temkin isotherm with R 2 ¼ 0.980 with the heat of sorption value 250.54 presents a better fitting of experimental data than Freundlich and Langmuir models.

Reusability of the adsorbents
The regeneration for repeated reuse of the adsorbent is very important for commercial applications. The abilities of the regeneration of the adsorbent were determined in four consecutive adsorption-desorption cycles. After the adsorption process was complete under the optimum conditions, the loaded adsorbents were treated with 0.1 M NaOH. After the regeneration was completed, the adsorbents were mixed again with a fresh NaOH solution, and the second adsorption cycle was carried out. The effect of recycle times on the adsorption capacities of C T -4, C GMT -6, and C P -5 is shown in Figure 10. From the figure, the regenerated adsorbents can be used repeatedly at least four times with no evident performance loss. The percentage of decrease in the adsorption capacities of C T -4, C GMT -6, and C P -5 after the four cycles were found as 9.4%, 7.8%, and 5.8%, respectively. Results of the reusability study indicated that all adsorbents could be recycled for BCG adsorption, and the adsorbents can be reused.

CONCLUSION
In this study, the removal of BCG dye by CMK-3 synthesized from GMT (C GMT ) was studied. Also, two mesoporous carbons, CMK-3 synthesized using SBA-15 from pure silica (C T ) and CMK-3 synthesized by a one-pot route mixing with Pluronic P123 (C P ), were prepared for comparison purposes. Different carbonization times were applied to all synthesized samples, and the samples with the highest surface area were selected as an adsorbent (C T -4, C GMT -6, and C P -5) regardless of whether the adsorbent has an ordered structure. The following key points can be drawn from this study: -The adsorption capacities of the adsorbents are generally decreased by the increased pH of the dye solution. The optimum pH value was selected as 2. -The adsorption of BCG onto C T -4 and C GMT -6 adsorbents followed a similar trend. -Equilibrium data were fitted to the Temkin and Langmuir isotherm model for C T -4 and C GMT -6. -The adsorption kinetics of dye onto all adsorbents were best described by a second-order kinetic model. -The adsorption capacity of C T -4, C GMT -6, and C P -5 was found to be 188.99, 204.57, and 328.74 mg·g À1 , respectively, at an initial concentration of 100 ppm.