The key aim of this study is to evaluate the adsorption of heavy oil from aqueous solutions with different oil contents over mesoporous silicate materials having different surfactant template contents. The mesoporous silicate materials have been synthesized from tetraethylorthosilicate as a silica precursor and cetyltrimethylammonium bromide as a template using the sol-gel technique. Four samples were prepared by (1) totally removing the template using the calcination process, (2) partially removing the template via ethanol extraction, (3) partially removing the template via water extraction, and (4) keeping the template as synthesized, respectively. These four samples have been characterized using X-ray diffraction, nitrogen adsorption, thermal gravimetric analysis and Fourier transformed infrared. The effect of the degree of template removal of these mesoporous materials for the oil removal has been investigated. The oil removal is inversely proportional to the surfactant content in the mesoporous material, being highest for the calcined sample but lowest for the as-synthesized sample. The kinetic of oil adsorption over the calcined material has been also studied and the data obtained fit well a second-order model.
Adsorption is one of the common techniques used to recover products and remove pollutants from streams before they are reused or released into the environment. Adsorption processes usually have low operational and energy costs, making them popular in chemical industries and environmental applications [ 1]. Activated carbon has been widely used as an adsorbent for hydrocarbons; however, due to some limitations such as its regeneration capability and flammability, development of alternative adsorbents became a necessity [ 2]. Oil removal via adsorption process from produced water and contaminants in ground-water is reported utilizing several types of adsorbents such as activated carbon, bentonite, peat, sand, coal, organoclays [ 3] and mesoporous MCM-41 [ 4]. The organoclay was reported as asorbent in the removal of oil from water solutions with 5–7 times more effective compared to activated carbon. Functionalized siliceous MCM-41 with an amino-group was reported as a selective adsorbent of organic pollutants with high-capacity recyclable adsorbents [ 5]. Therefore, mesoporous MCM-41 were considered as alternative adsorbents because they have high adsorption capacity due to their physicochemical properties such as high surface area (1000–1500 m2/g) and average pore size of (2–10 nm). Also, it has an exceptional properties such as selectivity due to the presence of hydroxyl groups and moderate Bronsted surface acidity which support functioned groups [ 6, 7]. The thermal stability and low water affinity allow high recycle ability and high process retention of the mesoporous materials [ 8]. Hence, this material has potential applications in many areas such as adsorption, catalysis and support material.
Mesoporous siliceous materials are prepared with high controllability of their chemical and physical properties [ 9, 10] usually by using different silica sources and organic surfactant template. The surfactant usually is a long chain hydrocarbon with cationic, anionic or neutral ammonium functional group [ 11, 12]. The surfactant molecules aggregate into micellar tubes in aqueous solutions while silica molecules condense on the surface of the tubes through the hydrolysis and condensation reactions [ 13]. The mesoporous material is then obtained by subsequent removal of the template usually using air calcination process [ 14]. Besides the calcination process, several other methods were reported for template removal or partial removal from the mesoporous pores. Extraction technique is used for partial removal of the surfactant template using different solvents such as ethanol, methanol and supercritical fluids [ 15, 16] with the association of sonication [ 17, 18] for enhanced and faster template removals. The surfactant template can also be partially removed by oxidizing agents such as H2O2 [ 19] or ammonium perchlorate [ 20]. The oxidizing agents are more effective than air in reducing pore collapse during thermal treatments [ 21].
The hydrophobic behavior of siliceous mesoporous materials can be enhanced using organophilic surface [ 22]. Mesoporous organosilicates with the hydrophilic and positively charged end was reported as adsorbents for oil removal from produced water [ 23]. MCM-41 mesoporous materials are significantly altered by organic groups, which might be covalently or non-covalently (physically) bound to their silica surface [ 10]. Covalent attachment generally occurs through silane chemistry. The silane moiety binds to silica by direct co-condensation process or indirect post-synthesis such as grafting method [ 24]. The grafting typically involves reactions between hydroxyl groups on the surfaces of the mesoporous structure and a silane compound [ 25].
Partial template removal was reported to create hydrophobic groups on the mesoporous silica surface [ 26– 28]. Taking advantage of the remaining template through the synthesis process and utilizing it as an organic functional group at the surface has been investigated in the adsorption process to remove aniline, phenol and O-chlorophenol from aqueous solution [ 19, 21, 29, 30].
The main aim of this paper is to establish a link between the extent of template removal from the synthesized silicate mesoporous materials and their effectiveness in removing heavy oil from aqueous solutions.
The pure siliceous mesoporous material has been prepared by the sol-gel technique at room temperature in a 250 mL Pyrex round flask. Silica solution has been prepared by adding 15 mL tetraethylorthosilicate, a silica precursor, to a mixture of 45 mL deionized water and 1 mL 36 wt-% HCl. The template solution has been prepared by adding 7.5 g cetyltrimethylammonium bromide surfactant to a 40 mL absolute ethanol and then mixing vigorously under stirring for 30 min. The template solution was then added dropwise to the silica solution while stirring the solution vigorously using a magnetic stirrer for 24 h. The solid product was obtained after filtration, washed with deionized water and kept in an oven to dry overnight at 100 °C. The dried solid product was then divided into four equal portions: (1) the first portion was used as it synthesized without any further treatment, (2) the template of the second portion was extracted using water vapor in a soxhlet extraction unit for 4 h, (3) the template of the third portion was extracted using ethanol in a Soxhlet extraction unit for 4 h, and (4) the fourth portion was calcined in a muffle furnace at 550 °C for 6 h.
Powder X-ray diffraction (XRD) patterns of the as-synthesized samples were obtained on a Bruker AXS, D8 Advanced diffractometer using Cu-K radiation source at 2 θ in the range of 2–10° and step size of a 0.02°/10 s. The nitrogen adsorption-desorption isotherms of samples were obtained at p/ p0 in the range of 0–1 using the Quantachrome Autosorb-6B. The Brunauer, Emmett and Teller (BET) surface area was calculated in the p/ p0 range of 0–0.3. And the average pore size (APS) of the samples was calculated using Barrett-Joyner-Halenda (BJH) method. Thermogravimetric analysis (TGA) was carried out using ThermoStar TGA analyzer. The analysis was performed in a flowing air atmosphere (120 mL/min) from ambient temperature to 100 °C. After holding for 10 min, the heating was resumed and continued until 700 °C at a heating rate of 5 °C/min. Fourier transformed infrared (FTIR) spectra were recorded on powder samples 0.2 g using Bruker, VERTEX FTIR spectrometer in the range of 4000–450 cm–1. The samples were oven dried overnight at 100 °C and placed in a desiccator for 4 h before FTIR analysis is performed.
The oil-water solutions have been prepared by dissolving a specific amount of heavy oil (Horiba heavy oil standard) into distilled water. In order to form an oil-water suspension, the oil-water mixtures were mixed vigorously for 15 min using a magnetic stirrer, before use. The unsuspended oil was skimmed out from the solution. The initial oil concentration, which was determined using a Horiba OCMA 310 oil analyzer, was varied from 0 to 2000 mg/L. The Horiba OCMA 310 oil analyzer performs qualitative analysis via the infrared analysis technique. Prior to perform the oil content measurements, the analyzer should be standardized and standardizing curve was set up. The analyzer uses S-316 as standard solvent, and an oil content measurement is performed in a 10 mL standard cell.
Batch adsorption was tested by adding a specific amount of adsorbent into the oil-water suspension in a 100 mL conical flask. To eliminate mass transfer resistance, the solution and the adsorbent were slowly mixed using a controlled incubating shaker. At the end of each adsorption experiment, the adsorbent is removed by settling and the solution was sent for further oil content determination.
A sample of the adsorbent-free solution is withdrawn in order to determine the percent of the remaining oil content [ 31]. A solution sample (10 mL) is extracted using S-316 solvent, and then water (oil free water) was separated. Extraction and separation were carried out automatically in the analyzer. Oil concentration was obtained by measuring the infrared absorbance of the oil containing S-316 solvent. The analysis was repeated twice and the average result was recorded. If the difference of two Horiba analyses was higher than 10%, the result would be rejected and the adsorption experiment should be repeated.
In order to confirm the formation of mesoporous materials at the end of the synthesis reaction, characterization was carried out using XRD and nitrogen adsorption. Figure 1 shows the XRD pattern of the as-synthesized sample. The pattern reveals that a mesoporous material of the MCM-41 type is formed. The structure of the formed material has a typical hexagonally arranged channels, represented by (100), (110) and (200) reflections at 2 θ angles of 2.2°, 3.4° and 3.8°, respectively.
The nitrogen adsorption-desorption isotherms of various prepared MCM-41 materials are shown in Fig. 2. The surface area of calcined siliceous material is 1276 m2/g. The surface area of ethanol extracted sample is 585 m2/g, which is less than half the surface area of the calcined sample. The surface area of water extracted sample (82 m2/g) is even much lower. Furthermore, the as-synthesized sample shows a negligible surface area. These results indicate that the calcined sample is highly ordered materials with high regular porosity. Additionally, the hysteresis of the adsorption-desorption isotherm of the calcined sample indicate a narrow pore size distribution; the sample has an average pore size (APS) of 2.54 nm and a mesopore volume of 0.81 mL/g. The mesopore volumes of ethanol extracted, water extracted and as-synthesized samples are 0.45, 0.045 and ~0 mL/g, respectively, suggesting that the pores of the latter two samples are almost completely filled with the template, leading to limited or negligible nitrogen adsorption as clearly shown in Fig. 2. The nitrogen adsorption/desorption isotherm of the calcined MCM-41 sample follows type IV isotherm, with a capillary condensation step, as it is the case for the ethanol extracted MCM-41 sample [ 32].
Thermal gravimetric analysis of the four samples is shown in Fig. 3. The mass of the calcined sample remains almost constant upon heating up to 700 °C, which indicates that the previous calcination process has totally removed the organic template from the sample. Contrarily, the as-synthesized sample retains 100% of the template prior to the TGA analysis. Upon heating the sample, its mass remains constant up to about 250 °C, above which the total mass of the sample drops quickly before slowly approaching a steady value at 700 °C. The sample mass at 700 °C decreases by about 72%, which is the fraction of the template material in the as-synthesized sample. The reduction in the mass of the water extracted sample upon heating to 700 °C is about 58%, suggesting that the extraction process using water only removes about 20% of the original template. Thus, the water extracted sample retains 80% of the total template originally presented in the sample. Unlike water extraction, ethanol extraction process removes about 65% of the originally presented template in the sample. This value is estimated from the total reduction in the mass of the ethanol extracted sample, which is 75%, upon heating to 700 °C. The percentage of the template retained by the four samples prior to the TGA analysis is inversely nonlinearly related to the surface area of the sample.
Figure 4 shows the derivative thermal gravimetric (DTG) analysis of the four samples. The calcined sample does not show any peaks, confirming that it is completely template free sample. For the as-synthesized and water extracted samples there are two main mass derivative peaks at temperatures of 240 and 340 °C. The peak at 240 °C is associated with the removal of the template from the external surface of the mesoporous particles. The template materials in such a case are weakly bound to the external surface of silica. The peak depth is proportional to the amount of template on the surface. The as-synthesized sample has a deeper peek than the water extracted sample, whereas some of the templates on the surface of the latter sample has been already removed during the water extraction process. Unlike the as-synthesized and water extracted samples, ethanol extracted sample does not show any noticeable peak at 240 °C, indicating that the template on the external surface of the ethanol extracted sample has been almost totally removed during the ethanol extraction process. The peak at 340 °C shown by the as-synthesized and water extracted samples is related to the removal of the covalently bound template from the silica framework. The ethanol extracted sample does not show any noticeable peak at 340 °C because the extraction process might have already removed the insignificant amount of covalently bound template from the silica surface of the sample; however, a small peak appears at 280 °C, possibly due to the removal of the physically bound template within the pores of the sample after covalent bonds are partially dissociated from the silica surface.
In order to investigate the incorporation of the surfactant into the silica framework of MCM-41, FTIR analysis has been performed for the four samples as shown in Fig. 5. Bands in the wavelength range of 719–961 cm–1 are attributed to the hydrocarbon bonds, which increase with the increase of the surfactant content. The band at 1067 cm–1 is associated with the Si–O bond [ 33], which decreases with the increase of surfactant percentage on the silica surface. The bands at 1462 and 1482 cm–1 represent the alkylammonium vibration of the surfactant [ 12]. The surfactant-silica interaction is reflected in two bands at 2848 and 2916 cm–1, respectively. These bands are assigned to C–H vibrations of the surfactant alkylammonium, and only present in the spectra of the as-synthesized and water extracted samples. These two bands become weak as the surfactant removal increases, and disappear in the calcined sample because of the complete removal of the template [ 12].
This result indicates that both calcination and ethanol extraction processes have dissociated the surfactant template from the silica framework. Therefore, the presence of the surfactant in the ethanol extracted sample is only related to the surfactant molecules that are trapped inside the mesopores of MCM-41.
The removal, via adsorption, of oil from aqueous solution has been studied using the four samples as adsorbents. The performance of the calcined sample was used as a bench-marker for other samples containing different fractions of surfactant. The percentage of oil removal upon reaching the equilibrium state was calculated using Eq. (1) and the amount of oil intakes as a ratio of mass of oil adsorbed to the mass of adsorbent q t (mg oil/ g adsorbent) is also calculated using Eq. (2)
where C0 is the initial oil concentration in the solution (mg/L) and C t is the oil concentration (mg/L) at time t. V is the volume of the solution (L) and W is the mass of the adsorbent (g).
Figure 6 shows the oil intake achieved using four different amounts of the calcined sample. The oil concentrations in a 100 mL of oil solution are in the range of 0 to 2000 mg/L. For all experiments, the equilibrium is achieved within 60 min. Longer adsorption time does not significantly increase the extent of oil removal. Further increase in time has not significantly increased the percentage of oil removal. The maximum percent oil removal is achieved at oil initial concentrations of 300 mg/L. It is seen clearly that the increase in the amount of adsorbent increases the oil intake from the solution to the adsorbent.
The oil intake from aqueous solution using the other three samples is shown in Fig. 7. The amount of each adsorbent used to achieve the results shown in Fig. 7 is 0.4 g. The three samples show lower performance than the calcined sample, indicating that the presence of surfactant within the pores of these samples reduces their efficiency as adsorbents. Among all, the as-synthesized sample provides the lowest adsorption capacity. Water extraction and ethanol extraction provide better adsorption capacities compared to the as-synthesized sample, but they are still much weaker as adsorbents than the calcined sample. The differences in the performances are clearly related to the overall surface area of these samples, which differ significantly. As surface area increases, adsorption capacity also increases. The presence of the surfactant template within the pores of the sample renders its accessibility by the oil molecules (i.e., the pores are filled with the template), and thus its adsorption capacity is reduced. The adsorption mechanism of heavy oil into the different pore structures is proposed in Fig. 8. The extent of this reduction would depend on the fraction of the surfactant in the sample. For this reason, the as-synthesized sample shows the lowest performance because its pores are filled with the surfactant template. On the other hand, the calcined sample shows the best performance as its pores are almost template-free. Another factor that contributes to the above performance trend is the hydrophobicity of the surface of the sample. The hydrophobic interaction plays an important role in the overall uptake of oil by the mesoporous materials. Because the surfactant used carries cationic groups, the presence of the surfactant molecules on the sample surface makes it less hydrophobic, resulting in a reduction in the hydrophobic interaction between the sample surface and the organic components of the oil. Besides the hydrophobic interaction, hydrogen bonding between the oil molecules and the silanol groups on the surface of these mesoporous materials might also play a significant role in the overall oil removal.
The time-dependent removal of oil from aqueous solutions using 0.4 g of the calcined sample was investigated. Figure 9 shows the maximum oil intakes are 23.8, 52.9 and 82 (mg oil/g adsorbent) at three different initial oil concentrations of 42.8, 73.2 and 92.0 mg/L, respectively.
The kinetics of oil adsorption on the mesoporous silica material was analyzed using the following pseudo first-order model [ 34]
where q e and q t are the adsorption capacity at equilibrium and at time t, respectively (mg oil/g adsorbent), and k1 is the rate constant of pseudo first-order adsorption (1/min).
After integrating Eq. (3) and applying the boundary conditions of q t = 0 at t = 0 and q t = q t at t= t, the following equation is obtained
The left hand side of the above equation is plotted vs. time t (see Fig. 10) and the slope of the best linear fit was calculated for each data set.
In order to evaluate the quality of the experimental data fitting, R2 was calculated for each data set. High R 2 values closer to 1 indicate that the model successfully describes the kinetics of oil adsorption. On the other hand, as the difference between experimental values and model values increases, R2 deviation from unity also increases. Because the R2 values of the fitting of the three data sets are much lower than unity (in the range of 0.30–0.75), the first order model is inappropriate. Therefore, pseudo-second-order model was attempted to find out its capability of replicating the experimental data [ 34]. The model was derived from the following second-order ordinary differential equation
where k2is the rate constant of pseudo second-order adsorption (g adsorbent)/(mg oil·min).
By integrating the above equation and applying the aforementioned boundary conditions, the following pseudo second-order adsorption rate equation was obtained:
where
The plot of t/q t vs. t should be linear with a slope of 1/ q e and an intercept of 1/
Siliceous mesoporous materials have been synthesized using the sol-gel technique. The produced materials were characterized using XRD, nitrogen adsorption, thermal gravimetric analysis, and Fourier transformed infrared. The calcined mesoporous material shows the highest adsorption capacity of heavy oil compared to the as-synthesized, water extracted and ethanol extracted materials. The oil adsorption from aqueous solutions containing different initial oil concentrations on the calcined siliceous mesoporous materials was measured at different adsorption times. The obtained kinetic data were fitted using first- and second-order models, and the second-order model provides good fits with R2 in the range of 0.95–0.99 while the first-order fits poorly.