Development of Porous Template Carbons from Montmorillonite Clays and Evaluation of Their Toluene Adsorption Behaviors

Carbon replicas were developed with two montmorillonites as the template and sucrose as the carbon source. The fabrication procedures included sucrose doping, polymerization with H2SO4, carbonization at 500–900°C, and liberation of carbon replicas from the template skeleton. N2 adsorption isotherms indicated the all of the resulting carbon replicas contained both micropores and mesopores. A higher carbonization temperature typically led to greater initial N2 adsorption, suggesting the presence of a greater number of micropores. Since no measurable micropores were found in the parent montmorillonites, those observed in the carbon replicas may have developed via activation of inherent water due to dehydration of montmorillonite clay or sucrose during the high temperature carbonization. XRD examination confirmed the formation of both graphite-like structure and amorphous carbon. The C content, C/H molar ratio, and the extent of πelectron resonance of the carbon replicas increased as the carbonization temperature rose, indicating that a higher temperature caused more thorough carbonization in the montmorillonite templates. Toluene adsorption experiments at 1000–8000 ppmv and 30–90°C demonstrated the effectiveness of carbon replicas with regard to capturing toluene. Model simulations further suggest that pore filling of toluene inside the mesopores of template carbon may occur, because the adsorption data showed better agreement with the DR model.


INTRODUCTION
Due to the versatile porous structure and surface functionality, carbon materials have been extensively applied in gas storage, pollution control, catalysis, and solvent recovery.Development of porous carbon replicas with non-carbonaceous templates having ordered structures has been considered an innovative technology for novel carbon production.Great attention has been paid to fabricate carbonaceous materials from silica templates (Bandosz et al., 1996;Johnson et al., 1997).In contrast to conventional activated carbons with a disordered and non-uniform porous structure, carbons developed from templating carbonization in general possess well-structured pores, pronounced surface area and pore volume, and minimum amount of impurity, namely, ash or minerals.
While many studies have concentrated on preparation and characterization of carbons replicas derived from various silica templates and on the application on H 2 storage or as supercapacitors, research pertaining to understanding the adsorption of volatile organic compounds (VOCs) using template carbons at a relatively high concentration level from a solvent recovery/reuse perspective is limited.The aim of the present study was at preparing and characterizing the carbon replicates from natural montmorillonite clay using a cheap and environmental friendly carbon source, namely, sucrose.Toluene was chosen as the representative VOC for adsorption tests since it is a hazardous air pollutant and widely used as benzene production, toluene diisocyanate production, gasoline blending, and solvent in large quantity (USEPA, 2013).The toluene adsorption dependence of carbon replicas on tested temperature and the inlet concentration were subsequently discussed.The success of preparing carbonaceous adsorbent from montmorillonite provides a treating alternative for waste bentonite clay, a leftover in considerable amount in Taiwan as a vegetable oil bleaching agent.

Preparation of Carbons from Montmorillonite Template
Two montmorillonites were used as templates.K10 montmorillonite (H 2 Al 2 (SiO 3 ) 4 -nH 2 O, designated as M) was obtained from Aldrich Chemical Co.The other one, an industry-grade acidified active clay (designated as AC), was obtained from an edible cooking oil manufacturer in Taiwan.Both samples were dried at 110°C for 24 h before use.Reagent-grade sucrose was dissolved in deionized (DI) water (1:2 by mass) to prepare the carbon source.Reagent-grade Sulfuric acid (H 2 SO 4 ) was used as the acid catalyst to prepare porous carbon.The sucrose solution was added into 6.0 g of montmorillonite according to a mass ratio of montmorillonite:sucrose = 1:1.H 2 SO 4 of 1 or 4 mL was subsequently introduced into the mixture to polymerize the sucrose in the montmorillonite samples.The mixture was then washed with DI water to remove the additional sucrose, heated at 100°C for 6 h, and subsequently heated at 160°C for another 6 h.This polymerization process was repeated twice.Carbonization was performed at 500, 700, and 900°C with a heating rate of 5 °C/min for 3 h under a N 2 environment.After carbonization, the montmorillonitecarbon composites were demineralization with 48 wt% HF solution at 85°C for 5 h.The demineralization process was performed twice to completely remove the montmorillonite framework.Subsequently, the insoluble carbons were treated with 36 wt% HCl solution for 2 h to purify the carbons.The resulting carbons were washed with DI water, collected by filtering, and drying at 110°C for 24 h.Samples were designated X-SY-Z with X representing the parent template, Y representing the amount of H 2 SO 4 used, and z representing the carbonization temperature.For example, sample M-S1-500 refers to the carbon from K10 montmorillonite treated with 1 mL of H 2 SO 4 and carbonized at 500°C.

Characterization of Carbons from Montmorillonite Template
The total surface area (S BET ), total pore volume (V t ), micropore (pore width < 2 nm) surface area (S micro ), micropore volume (V micro ), and pore size distribution of montmorillonite template and carbon replicas were determined by N 2 adsorption at 77 K (Beckman Coulter SA3100).Samples were degassed at 10-20 Torr vacuum and 150°C for 24 h before the N 2 adsorption measurements occurred between 10 -3 and 1 atm.S BET was calculated by the Brunauer-Emmett-Teller equation based on ASTM D4820-96a.S micro and V micro were calculated from t-plot analyses using the Jura-Harkins equation: t = [13.99/(0.034log(p/p 0 )] 0.5 (Lippens and de Boer, 1965).The range of relative pressures used to determine S micro and V micro was based on thickness t values between 0.45 and 0.80 nm.Micropore size distribution was determined by the 3-D model (Sun et al., 1998).Mesopore size distribution was determined by the Barrett-Joyner-Halenda method (Gregg and Sing, 1982).
The chemical composition, including the mass concentrations of C, H, N, and S of carbon replicas, was determined with an elemental analyzer (Elementar Vario EL III).The ash contents of samples were carried out by burning away the samples at 800°C for 3 h, or until total elimination of fixed carbon and volatile matter.Crystalline phases of the montmorillonite template and resulting carbonaceous materials were analyzed with X-ray diffraction (XRD; Rigaku Rint 2000 with CuKα radiation).The surface morphology and element composition were characterized using an environmental scanning electron microscope (ESEM; Quanta 400F) with an energy dispersive spectroscope (EDS) system.The samples' morphology was also illustrated with a transmission electron microscope (TEM; JEOL TEM-3010).X-ray photoelectron spectroscopy (XPS) was carried out to examine the surface functional groups of the resulting samples.Carbon functional groups were deconvoluted within 280-294 eV on the basis of their characteristic binding energies.

Toluene Adsorption Experiments
The adsorption testing system comprised an electronic, gravimetric balance (Thermo Cahn TG-2121) with a capacity up to 1.5 g and a weighing range of 150 mg.The prepared gas stream with a toluene concentration in the range of 1000-8000 ppm v passed a mixture coil, and then passed the carbon sample suspended in a thermogravimetric balance at a 90 NmL/min flow rate, with adsorption temperatures controlled at 30-90°C.Mass flow controllers (Brooks 5850E and 5850E-Kr) were used to dilute and regulate the flow rate of gas mixture entering the testing system.As the sample adsorbed the toluene, its mass increased until adsorption equilibrium was reached.The difference between the initial and final masses of the adsorbent represented the total mass of toluene adsorbed at given concentration and temperature.The preliminary test also showed that toluene adsorption by the thermogravimetric balance itself is negligible.The mass of toluene adsorbed divided by the initial dry mass of the adsorbent provided the equilibrium adsorption capacity of the adsorbent (milligram toluene per gram carbon; mg/g).
The Freundlich, Langmuir, and Dubinin-Radushkevich (DR) equations were applied to the toluene adsorption isotherms obtained at various inlet toluene concentrations and adsorption temperatures.Freundlich equation is an empirical model, assuming that the equilibrium adsorption capacity is linear with change in adsorbate concentration on a log-log scale at a given temperature (Ruthven, 1984): where x is the mass of the adsorbate adsorbed at equilibrium, m is the mass of the adsorbent, C is the equilibrium gas phase concentration of the adsorbate, and k and n are Freundlich constants.
The theoretical Langmuir equation is the best known of all isotherms, assuming monolayer adsorption onto a surface with a finite number of identical sits and is represented as follows (Ruthven, 1984): where q is the equilibrium adsorption capacity (mg/g) at a given concentration C, q m is the maximum adsorption capacity required to form a monomolecular layer, and K L is the Langmuir constant.
The DR model was developed to determine physisorption in microporous adsorbents.Detailed description about the development and application of DR model can be found elsewhere (Dubinin, 1989).A brief statement is presented here for clarity.The DR equation is given by: where W is the volume of adsorbate per mass of adsorbent (cm 3 /g), W 0 is the limiting pore volume of adsorption per unit mass of adsorbent (cm 3 /g), which is typically referred to as the volume of micropore.β is the similarity coefficient, a unitless constant equal to the ratio of the adsorption potentials for the adsorbate of interest and a reference adsorbate.E 0 is the characteristic adsorption energy of the reference adsorbate (kJ/mol).A is the differential molar work need to transport one mole of the adsorbate to the surface of an infinitely large adsorbent (kJ/mol), which is given by: where -ΔG is the Gibbs free energy change, R is the ideal gas law constant, T is the absolute temperature, p s is the saturation partial pressure of adsorbate, and p is the partial pressure of adsorbate.

Nitrogen Adsorption Isotherm, Surface Area, and Pore Volume of Carbon Replicas
Fig. 1 illustrates the N 2 adsorption isotherms of montmorillonite templates and corresponding carbon replicas.The isotherms for both M (Fig. 1(a)) and AC (Fig. 1(b)) templates were in type IV shape.The raw M and AC had a small N 2 adsorption of approximately 30 and 40 cm 3 -N 2 /g, respectively, near zero pressure followed by a gradual increase in adsorption at higher pressure, suggesting the absence of micropores.The presence of adsorption/desorption hysteresis at relative pressure = 0.4 further indicated the existence of mesopores in templates.For carbon replicas, the shape of isotherms assembled to those of their parent templates.The initial part of the isotherm (relative pressure < 0.1) maintained a Langmuir shape with various N 2 adsorption amounts, depending on the carbonization temperature.A higher carbonization temperature resulted in larger initial N 2 adsorption for template carbons, suggesting an increasing microporosity.It is worth to address that the creation of micropores should not be a result of replication of the montmorillonite templates, in which microporosity does not exist.These micropores may be developed via activation by inherent water from dehydration of montmorillonite clay and sucrose during the high temperature carbonization.
The surface area and pore volume of montmorillonite templates and carbon replicas were deconvoluted from the N 2 adsorption isotherms and listed in Table 1.The M and AC templates had S BET = 276.7 and 257.0 m 2 /g respectively and undetected microporosity, indicating its dominating mesoporosity.The resulting carbon replicas possessed appreciable microporosity, of which the amount depended on carbonization temperature and the properties of templates.Carbons developed from AC had significant amounts of micropores at carbonization temperature of 500°C.On the contrary, for carbons from M, micropores cannot be developed unless the temperature reached 700°C.This trend was similar for experiments using different amounts of H 2 SO 4 for polymerization.These experimental results suggest that montmorillonites from different sources/batches can lead to difference consequences in resulting physical properties of carbon replicas.Even though it is not yet thoroughly comprehended, the M and AC templates may contain a different amount of crystal water that can affect the magnitude of activation.In addition, AC appeared to possess greater mineral crystallinity than M (this speculation was confirmed by the XRD examination and will be  presented later); consequently, both the micropore and mesopore structures of carbon replicas, namely AC-S1-500 and AC-S4-500, can be retained at a lower carbonization temperature.For carbon replicas derived from M at 500°C, in contrast, the pore structures appeared to be unstable and destructed during the HF demineralization; very small micro-and mesoporosity was thus shown (Table 1).When carbonization temperature was higher than 700°C, the trend in development of S BET and S micro for carbons from M and AC was difficult to conclude.Overall, a higher carbonization temperature caused a greater S BET (for both M and AC series), but did not guarantee a greater S micro (i.e., AC series).In additional, the amount of H 2 SO 4 used for polymerization also led to different effects on S BET of carbons from different templates.A greater amount of H 2 SO 4 resulted in a larger S BET for M-templated carbons but a smaller S BET for AC-templated carbons (Table 2).These results again suggest that templates from different sources/batches cause various consequences in resulting properties of carbon replicas.
Fig. 2 shows the micropore size distributions of the carbon replicas.Because both M and AC are highly mesoporous, these two templates had less significant pore peaks in micropore range.Nevertheless, M template demonstrated a trimodal size distribution with peak volume < 0.05 cm 3 /g-nm.
For the carbon replicas from M, micropores with pronounced trimodal peaks at approximately 0.9, 1.6, 2.0 nm were presented.For the carbon replicas from AC, the micropore size distribution was more uniform with a sharp peak at 9 nm.These modeling results suggest that initial pore size of template may determine the final pore size distribution of carbon replicas.
For both templates and carbon replicas, the dominant size of mesopores ranged between 2 and 10 nm, except for those samples developed under 500°C carbonization (Fig. 3).Results from Fig. 3 further indicate that the initial pore size of template may determine the final size distribution of carbon replicas, especially for AC series because the curve shapes of mesopore size distribution were similar.These results also support the success of replication of porous structure from parent montmorillonite templates.

ESEM and TEM Observations
The ESEM micrographs of templates M and AC and carbon replicas (M-S1-900 and AC-S1-900) are shown in Fig. 4. In general, both M and AC were irregular shaped particles with smooth surface and a diameter of approximately 2-15 μm.For their corresponding carbon replicas, the size and morphology appeared to be preserved, again indicating the success of replication from parent templates.Note that for M-S1-900, the surface was rougher (indicated by the arrow 1 in Fig. 4(b)) than its parent template M, suggesting that extra carbon structure may also built up on the external surface of the M particles, which may cause not only the observed surface roughness but also the greater surface area and pore volume shown in Table 1.
Fig. 5 illustrates the TEM images of template M (Fig. 5(a)) and carbon replica M-S1-900 (Figs. 5(b) and 5(c)).Replication of initial structure of template can be clearly seen on the carbon replica by comparing Figs.5(a) and 5(b), again supporting the success of development of template carbons.The nano-channels in mesopore size range (approximately 1-2 nm) within the crystal grains were shown to develop.The curved graphene fragments disorderly packing and constituting the carbon structures in the interior of carbon replica are also clearly illustrated in Fig. 5(c).The observational results are consistent with those in Armandi et al. (2007).These results also in part support the observations from N 2 adsorption isotherms (Table 1 and Figs. 2 and 3).

XRD Examination
The XRD patterns of templates, template-carbon composites after polymerization and carbonization, and resulting carbon replicas developed at 900°C are presented in Fig. 6.Both M and AC had a distinct diffraction pattern consisting of narrow characteristic peaks.However, we also noticed that AC possessed better crystallinity than M because the peaks of AC appeared to be sharper.In addition, AC had an extra low-angle peak at 2θ = 6.08°, indicating the diffraction from the (001) plane with a d-spacing of approximately 1.5 nm.The lack of the peak at d-spacing = 1.5 nm for template M may be because template M contains a great content of water that causes the disappearance of regular stacking of montmorillonite layers (Fukushima, 1984).After polymerization and carbonization, the XRD patterns of template-carbon composites were similar to the parent template.However, when the template was removed with HF, the peak at 2θ = 26.8°being the index as the diffraction from the (111) plane of montmorillonite disappeared.On the other hand, peaks with 2θ at 15.2, 29.8, and 31.1°were presented, indicating the graphite-like structures were formed.Furthermore, two broad diffraction effects at approximately 2θ = 24 and 44° in template carbons were another indication of the presence of amorphous carbon.

Elemental Analysis
The chemical composition of the template carbons obtained under different carbonization temperature is presented in Table 2.The C content in carbon replicas was between 43 and 58 wt%.Increasing the carbonization temperature led to a greater C content and C/H ratio, suggesting carbonization is more complete at an elevating temperature.These trends are in agreement with most of the previous studies pertaining to preparation of carbons from silica templates.However, compared to the works using acrylonitrile, furfuryl alcohol or vinyl acetate (Meyers et al., 2001), propylene (Chen et al., 2007), and pyrrole or styrene (Garsuch et al., 2006)   a smaller C content and a greater O content.Since O is mainly contributed by sucrose, the analytical results shown here suggest that the polymerization process in this study should be further optimized to achieve a better dehydration of template-carbon composites.It was also noteworthy that a greater H 2 SO 4 addition can result into a higher S content.
In this study, template-carbon composites polymerization at 4-mL H 2 SO 4 resulted in S content between 6 and 9 wt%.The ash content of all the carbon replicas was measured to be < 1.4 wt%, suggesting that removal of the montmorillonite skeleton with HF was thorough.EDS results also confirmed that the template carbons contained a significant amount of C and O, as well as a small amount of F (from HF), S (from H 2 SO 4 ), Mg, Na, and Al (from montmorillonite template).Si was not found with EDS, indicating that the liberation of carbon from template skeletons was successful.These data are in consistence with those from the elemental analysis.

Characterization of Surface Functional Groups
Carbon functional groups including C-C and C-H bonding (284.4-284.6 eV), hydroxyl (286.1-286.5 eV), carbonyl (287.7-288.2 eV), carboxyl groups (289.3-289.9eV), and π-electron resonance (292.7-294.2eV) were deconvoluted and the corresponding percentages of total C 1s peak area were determined.A decrease in the peak area of hydroxyl group from 19.3 to 6.8% was observed as increasing carbonization temperature.Furthermore, the elevating carbonization temperature also enhanced the peak area of π-electron resonance from 2.3 to 5.7%, supporting our previous conclusion that a more thorough carbonization was achieved at a higher temperature.

Toluene Adsorption and Model Simulation
To examine the equilibrium toluene adsorption behaviors of carbon replicas, M-S4-900 and AC-S1-900 were chosen as the representative samples because these two carbons had the largest S BET and V t among their homologues.Tables 3 and 4 list the equilibrium adsorption capacities, the modeled Freundlich, Langmuir, DR parameters, and the corresponding R 2 values, respectively.Results showed that the toluene adsorption capacity enhanced with increasing toluene concentration, and decreased with elevating adsorption temperature.Furthermore, correlations for toluene adsorption capacities of the template carbons had good agreement from DR (R 2 > 0.908) simulations (Table 4).Freundlich equation can also fairly simulate the isotherms of AC-S1-900 (R 2 > 0.930) but had a greater deviation for the isotherm of M-S4-900 obtained at 90°C.Interestingly, Langmuir equation showed a poorer simulation, especially for M-S4-900.These observations infer that pore filling instead of monolayer adsorption may limit the toluene adsorption since the adsorption isotherms were better fitted with DR model than Langmuir model.
It is also imperative to note that the adsorption characteristic energy (E 0 ) for template carbon was small, within 11-17 kJ/mol (Table 4).The experimental results suggest that toluene is easy to adsorb onto and desorb from the template carbon.Additionally, physisorption is the key mechanism resulting the partitioning of the toluene vapor between the gas and adsorbed phases.These data are highly encouraging that the template carbon could be utilized to recover high-cost nonpolar volatile organic solvents under conditions similar to the tests.Smaller adsorption energy leads to rapid adsorption/desorption, by which an effective recovery/regeneration sorption system can be designed and processed from a material reuse/recycle viewpoint.

CONCLUSIONS
In this work, two montmorillonite clays were used as mineral templates to prepare carbon replicas via sucrose impregnation, polymerization with H 2 SO 4 , carbonization at 500-900°C, and demineralization of template framework with HF.All the resulting carbon replicas possessed both microporous and mesoporous structures.A higher carbonization temperature resulted in larger initial N 2 adsorption, suggesting that a greater amount of micropores was created.Because no microporosity can be observed in

Table 1 .
Physical properties of montmorillonite clay templates and resulting carbon replicas (n = 2).

Table 3 .
Equilibrium adsorption capacity (in mg/g) of carbon replicas at given temperatures and inlet toluene concentration.

Table 4 .
Freundlich, Langmuir, and Dubinin-Radushkevich (DR) parameters for toluene adsorption with carbon replicas., the micropores in carbon replicas appeared to be developed via activation by inherent water from dehydration of montmorillonite clay or sucrose during the high temperature carbonization.The success of replication of carbon template and the presence of mesoporosity was supported by ESEM and TEM observations.XRD examination confirmed the formation of graphite-like structure and amorphous carbon.The C content, C/H molar ratio, and the extent of π-electron resonance in carbon replicas were enhanced with increasing carbonization temperature, suggesting that a higher temperature resulted in a more complete carbonization inside the montmorillonite template.Toluene adsorption equilibrium showed the dependence of capacity on inlet concentrations and adsorption temperature.Model simulations additionally suggested that the pore filling of toluene inside the pores of template carbon took place due to better simulation results on adsorption isotherms with DR model.