Novel Aluminum Oxide-Impregnated Carbon Nanotube Membrane for the Removal of Cadmium from Aqueous Solution

An aluminum oxide-impregnated carbon nanotube (CNT-Al2O3) membrane was developed via a novel approach and used in the removal of toxic metal cadmium ions, Cd(II). The membrane did not require any binder to hold the carbon nanotubes (CNTs) together. Instead, the Al2O3 particles impregnated on the surface of the CNTs were sintered together during heating at 1400 °C. Impregnated CNTs were characterized using XRD, while the CNT-Al2O3 membrane was characterized using scanning electron microscopy (SEM). Water flux, contact angle, and porosity measurements were performed on the membrane prior to the Cd(II) ion removal experiment, which was conducted in a specially devised continuous filtration system. The results demonstrated the extreme hydrophilic behavior of the developed membrane, which yielded a high water flux through the membrane. The filtration system removed 84% of the Cd(II) ions at pH 7 using CNT membrane with 10% Al2O3 loading. A maximum adsorption capacity of 54 mg/g was predicted by the Langmuir isotherm model for the CNT membrane with 10% Al2O3 loading. This high adsorption capacity indicated that adsorption was the main mechanism involved in the removal of Cd(II) ions.


Introduction
Cadmium is a well-known highly toxic metal found in drinking water, and is associated with major negative health impacts. The World Health Organization guidelines suggest an allowable limit of cadmium ions, Cd(II), in water of 0.003 mg/L [1]. Cadmium primarily accumulates in the kidneys, and has a relatively long biological half-life of 10 to 35 years in humans [2]. A potential source of cadmium contamination in drinking water is industrial wastewater, such as that produced by the manufacturing processes for smelting, pesticides, fertilizers, dyes, pigments, refining, and textile operations. Cadmium contamination of drinking water might also be caused by the presence of Cd(II) ions as an impurity in the zinc of galvanized pipes and certain metal fittings [3,4].
Various toxic metal decontamination techniques (such as adsorption, precipitation, reduction, ion exchange, precipitation, solvent extraction, electrolytic recovery, and chemical oxidation) have been applied for the removal of toxic metals from water [3,5]. However, the majority of these methods have limited applications due to economical or technical constraints. Water treatment by adsorption offers the most practical and economical treatment alternative. Numerous adsorbents have been used

X-ray Diffraction (XRD)
The XRD patterns for raw and impregnated CNTs were measured at a rate of 1.0°/min in the range of 10-80° (2α) using an X-ray diffractometer equipped with a Cu Kα radiation source (40 kV, 20 mA).

Porosity Measurement
The dry-wet method [16] was used to determine the porosity of the membranes using Equation (1): where W1 (g) and W2 (g) are the weights of the dry and wet membranes, respectively; V (cm 3 ) is the volume of the membrane; and ρ (g/cm 3 ) is the density of distilled water at ambient temperature. The membrane was immersed in distilled water for 24 h, and its wet weight was measured.
The membrane was subsequently dried in an oven at 90 °C for 24 h, and the dry weight of the membrane was measured. The experiment was performed in triplicate, and the data are presented as the mean value of all experiments.

Contact Angle Measurement
A contact angle analyzer (model DM-301, KYOWA, Niiza, Japan) was used to measure the contact angle of the membrane surface and hence the hydrophobicity/hydrophilicity of the membrane.

Zeta Potential Measurement
The zeta potential for a suspension of 0.5 g/L CNTs-Al2O3 in distilled water was determined using a Malvern ZEN2600 Zetasizer Nano Z (Malvern, Worcestershire, UK) at pH 2.0-10 (adjusted with 0.1 M NaOH or HNO3).

Continuous Filtration System
The main components of the continuous filtration system used in this study (presented in Figure 2) were a membrane cell with an effective surface area of 5.7 cm 2 , a 10 L feed tank, and the required Flowchart for the synthesis of aluminum oxide-impregnated carbon nanotube (CNTs-Al 2 O 3 ) membrane.

X-ray Diffraction (XRD)
The XRD patterns for raw and impregnated CNTs were measured at a rate of 1.0 • /min in the range of 10 • -80 • (2α) using an X-ray diffractometer equipped with a Cu Kα radiation source (40 kV, 20 mA).

Porosity Measurement
The dry-wet method [16] was used to determine the porosity of the membranes using Equation (1): where W 1 (g) and W 2 (g) are the weights of the dry and wet membranes, respectively; V (cm 3 ) is the volume of the membrane; and ρ (g/cm 3 ) is the density of distilled water at ambient temperature. The membrane was immersed in distilled water for 24 h, and its wet weight was measured.
The membrane was subsequently dried in an oven at 90 • C for 24 h, and the dry weight of the membrane was measured. The experiment was performed in triplicate, and the data are presented as the mean value of all experiments.

Contact Angle Measurement
A contact angle analyzer (model DM-301, KYOWA, Niiza, Japan) was used to measure the contact angle of the membrane surface and hence the hydrophobicity/hydrophilicity of the membrane.

Zeta Potential Measurement
The zeta potential for a suspension of 0.5 g/L CNTs-Al 2 O 3 in distilled water was determined using a Malvern ZEN2600 Zetasizer Nano Z (Malvern, Worcestershire, UK) at pH 2.0-10 (adjusted with 0.1 M NaOH or HNO 3 ).

Continuous Filtration System
The main components of the continuous filtration system used in this study (presented in Figure 2) were a membrane cell with an effective surface area of 5.7 cm 2 , a 10 L feed tank, and the required pressure pump and flow meter. The membrane was placed in a circular housing with a mesh underneath it as a support structure to maintain the stability of the membrane during the flow experiments. The pure water flux analysis was performed before the Cd(II) remediation studies were performed. The pure water flux was measured using Equation (2) [16,28,29]: where J (L·m −2 ·h −1 ) is the water flux, t (h) is the time required for permeate water to pass through the membrane, and V (L) is the volume of permeate water. experiments. The pure water flux analysis was performed before the Cd(II) remediation studies were performed. The pure water flux was measured using Equation (2) [16,28,29]: where J (L·m −2 ·h −1 ) is the water flux, t (h) is the time required for permeate water to pass through the membrane, and V (L) is the volume of permeate water. For cadmium removal, the experimental runs began with the circulation of a 1 ppm solution of Cd(II) from the feed tank through the system. An initial volume of approximately 10 L was added to the feed tank, and the pH was adjusted using 1 M NaOH or 1 M HNO3, as required. The pressure and flow rate were adjusted to the desired values. Permeate (purified water) passing through the membrane was collected from the sample collection point (shown in Figure 2) at different time intervals using sample bottles with volumes of approximately 20 mL. The effects of the initial concentration, time, pH, and membrane Al2O3 loading on cadmium removal were studied. Table 1 shows the experimental conditions for Cd(II) removal using different CNT-Al2O3 membranes. First, a CNT membrane with 10% Al2O3 loading was used to determine the optimum pH for the maximum removal of Cd(II). The transmembrane pressure difference and concentration of Cd(II) were held constant during these experiments. The optimum pH (pH = 7) was held constant during the remainder of the experiments, and the effects of Al2O3 loading and the initial Cd(II) concentration in the solution (water) on the removal efficiency of Cd(II) ions were determined.

Analytical Methods
Inductively coupled plasma mass spectrometry (X-Series 2 Q-ICP-MS, Thermo Fisher Scientific, Waltham, MA, USA) was used to measure the concentration of Cd(II) before and after the experiments. For cadmium removal, the experimental runs began with the circulation of a 1 ppm solution of Cd(II) from the feed tank through the system. An initial volume of approximately 10 L was added to the feed tank, and the pH was adjusted using 1 M NaOH or 1 M HNO 3 , as required. The pressure and flow rate were adjusted to the desired values. Permeate (purified water) passing through the membrane was collected from the sample collection point (shown in Figure 2) at different time intervals using sample bottles with volumes of approximately 20 mL. The effects of the initial concentration, time, pH, and membrane Al 2 O 3 loading on cadmium removal were studied. Table 1 shows the experimental conditions for Cd(II) removal using different CNT-Al 2 O 3 membranes. First, a CNT membrane with 10% Al 2 O 3 loading was used to determine the optimum pH for the maximum removal of Cd(II). The transmembrane pressure difference and concentration of Cd(II) were held constant during these experiments. The optimum pH (pH = 7) was held constant during the remainder of the experiments, and the effects of Al 2 O 3 loading and the initial Cd(II) concentration in the solution (water) on the removal efficiency of Cd(II) ions were determined.

Analytical Methods
Inductively coupled plasma mass spectrometry (X-Series 2 Q-ICP-MS, Thermo Fisher Scientific, Waltham, MA, USA) was used to measure the concentration of Cd(II) before and after the experiments. Figure 3 shows the SEM images of the prepared membranes with various Al 2 O 3 contents. The particles are well dispersed at low Al 2 O 3 loadings, whereas the particles tend to agglomerate at higher Al 2 O 3 loadings (i.e., 20% or above).  Figure 3 shows the SEM images of the prepared membranes with various Al2O3 contents. The particles are well dispersed at low Al2O3 loadings, whereas the particles tend to agglomerate at higher Al2O3 loadings (i.e., 20% or above).  Figure 4 displays the XRD patterns of the raw and impregnated CNTs powders. In the XRD pattern of the raw CNTs, the characteristic peaks at 2θ = 27° and 44° correspond to the CNTs. However, the XRD pattern of the impregnated CNTs presents new peaks at 2θ = 17°, 33°, and 40° in addition to the two apparent peaks associated with the CNTs. These peaks correspond to the Al2O3 nanoparticles and indicate the successful impregnation of Al2O3 particles onto the surface of the CNTs. The XRD spectrum shows a minor shift of CNTs peaks for the CNT-Al2O3, possibly due to the residual stresses imposed by the Al2O3 particles embedded onto the CNTs [30,31].   Figure 4 displays the XRD patterns of the raw and impregnated CNTs powders. In the XRD pattern of the raw CNTs, the characteristic peaks at 2θ = 27 • and 44 • correspond to the CNTs. However, the XRD pattern of the impregnated CNTs presents new peaks at 2θ = 17 • , 33 • , and 40 • in addition to the two apparent peaks associated with the CNTs. These peaks correspond to the Al 2 O 3 nanoparticles and indicate the successful impregnation of Al 2 O 3 particles onto the surface of the CNTs. The XRD spectrum shows a minor shift of CNTs peaks for the CNT-Al 2 O 3 , possibly due to the residual stresses imposed by the Al 2 O 3 particles embedded onto the CNTs [30,31].  Figure 3 shows the SEM images of the prepared membranes with various Al2O3 contents. The particles are well dispersed at low Al2O3 loadings, whereas the particles tend to agglomerate at higher Al2O3 loadings (i.e., 20% or above).  Figure 4 displays the XRD patterns of the raw and impregnated CNTs powders. In the XRD pattern of the raw CNTs, the characteristic peaks at 2θ = 27° and 44° correspond to the CNTs. However, the XRD pattern of the impregnated CNTs presents new peaks at 2θ = 17°, 33°, and 40° in addition to the two apparent peaks associated with the CNTs. These peaks correspond to the Al2O3 nanoparticles and indicate the successful impregnation of Al2O3 particles onto the surface of the CNTs. The XRD spectrum shows a minor shift of CNTs peaks for the CNT-Al2O3, possibly due to the residual stresses imposed by the Al2O3 particles embedded onto the CNTs [30,31].

Measurement of the Zeta Potential and Point of Zero Electric Charge (pH PZC )
The zeta potentials of CNT-10% Al 2 O 3 in distilled water were determined in a pH range of 2.0-10. As displayed in Figure 5, the surface charge of the membrane surface is positive at pH < 6.5 and negative at pH > 6.5. The zero electric charge (pH PZC ) value of the membrane was noted at 6.5. The membrane is expected to have a relatively higher removal of Cd(II) ions at pH 7, as discussed in Section 3.6, because of the electrostatic interactions between the negatively charged membrane surface and cationic Cd(II) ions.

Measurement of the Zeta Potential and Point of Zero Electric Charge (pHPZC)
The zeta potentials of CNT-10% Al2O3 in distilled water were determined in a pH range of 2.0-10. As displayed in Figure 5, the surface charge of the membrane surface is positive at pH < 6.5 and negative at pH > 6.5. The zero electric charge (pHPZC) value of the membrane was noted at 6.5. The membrane is expected to have a relatively higher removal of Cd(II) ions at pH 7, as discussed in Section 3.6, because of the electrostatic interactions between the negatively charged membrane surface and cationic Cd(II) ions.

Porosity Measurement
The dry-wet method was used to determine the porosity of the membranes. Figure 6 displays the porosity versus Al2O3 loading for loadings of 1 to 20%. Considering the standard deviation reported for the measured porosity values, the variation in porosity with Al2O3 content is rather minor, implying that the Al2O3 loading does not have a significant effect on the membrane porosity.

Contact Angle Measurement
Contact angle measurement is an index of the hydrophobicity/hydrophilicity of the membrane surface. As shown in Figure 7, the contact angle of the membrane decreased with increasing Al2O3 content. In other words, the hydrophilicity of the membrane increases with increasing Al2O3 loading.

Porosity Measurement
The dry-wet method was used to determine the porosity of the membranes. Figure 6 displays the porosity versus Al 2 O 3 loading for loadings of 1 to 20%. Considering the standard deviation reported for the measured porosity values, the variation in porosity with Al 2 O 3 content is rather minor, implying that the Al 2 O 3 loading does not have a significant effect on the membrane porosity.

Measurement of the Zeta Potential and Point of Zero Electric Charge (pHPZC)
The zeta potentials of CNT-10% Al2O3 in distilled water were determined in a pH range of 2.0-10. As displayed in Figure 5, the surface charge of the membrane surface is positive at pH < 6.5 and negative at pH > 6.5. The zero electric charge (pHPZC) value of the membrane was noted at 6.5. The membrane is expected to have a relatively higher removal of Cd(II) ions at pH 7, as discussed in Section 3.6, because of the electrostatic interactions between the negatively charged membrane surface and cationic Cd(II) ions.

Porosity Measurement
The dry-wet method was used to determine the porosity of the membranes. Figure 6 displays the porosity versus Al2O3 loading for loadings of 1 to 20%. Considering the standard deviation reported for the measured porosity values, the variation in porosity with Al2O3 content is rather minor, implying that the Al2O3 loading does not have a significant effect on the membrane porosity.

Contact Angle Measurement
Contact angle measurement is an index of the hydrophobicity/hydrophilicity of the membrane surface. As shown in Figure 7, the contact angle of the membrane decreased with increasing Al2O3 content. In other words, the hydrophilicity of the membrane increases with increasing Al2O3 loading.

Contact Angle Measurement
Contact angle measurement is an index of the hydrophobicity/hydrophilicity of the membrane surface. As shown in Figure 7, the contact angle of the membrane decreased with increasing Al 2 O 3 content. In other words, the hydrophilicity of the membrane increases with increasing Al 2 O 3 loading. The decrease in contact angle with increase in Al 2 O 3 loading might be due to change in membrane pore size due to agglomeration at higher Al 2 O 3 loading. This agglomeration leads to enhanced water flux (as shown in Figure 8), and hence, the contact angle value is decreased. This hydrophilic nature of the membrane is primarily responsible for the enhanced flux through the membrane [28,29]. The decrease in contact angle with increase in Al2O3 loading might be due to change in membrane pore size due to agglomeration at higher Al2O3 loading. This agglomeration leads to enhanced water flux (as shown in Figure 8), and hence, the contact angle value is decreased. This hydrophilic nature of the membrane is primarily responsible for the enhanced flux through the membrane [28,29].

Water Flux Measurements: Effect of Transmembrane Pressure Difference and Aluminum Oxide Loading
The effects of transmembrane pressure difference, aluminum oxide loading, and time, on the water flux through the membranes, were studied, as shown in Figure 8. The transmembrane pressure difference was varied from 1 to 40 psi. A nearly linear relationship exists between the pressure and flux for all membranes with different Al2O3 loadings. The water permeate flux was measured by holding the transmembrane pressure difference constant at 20 psi for 30 min. The permeate flux values for different pressures were obtained using the same procedure. For each reading, the pressure was maintained for 10 min before noting the readings. Figure 8 illustrates that the permeate water flux increased as the Al2O3 content increased from 1 to 20%. The increased permeate flux for membranes with high Al2O3 loading can be explained based on two mechanisms. First, the hydrophilic surface of the membrane at high Al2O3 loading facilitates the transport of water through the membrane (Figure 7). Second, the agglomeration of Al2O3 particles (Figure 3c) at high loading (20%) results in the formation of relatively large pores in the membrane, thus contributing to the higher permeate flux. The decrease in contact angle with increase in Al2O3 loading might be due to change in membrane pore size due to agglomeration at higher Al2O3 loading. This agglomeration leads to enhanced water flux (as shown in Figure 8), and hence, the contact angle value is decreased. This hydrophilic nature of the membrane is primarily responsible for the enhanced flux through the membrane [28,29].

Water Flux Measurements: Effect of Transmembrane Pressure Difference and Aluminum Oxide Loading
The effects of transmembrane pressure difference, aluminum oxide loading, and time, on the water flux through the membranes, were studied, as shown in Figure 8. The transmembrane pressure difference was varied from 1 to 40 psi. A nearly linear relationship exists between the pressure and flux for all membranes with different Al2O3 loadings. The water permeate flux was measured by holding the transmembrane pressure difference constant at 20 psi for 30 min. The permeate flux values for different pressures were obtained using the same procedure. For each reading, the pressure was maintained for 10 min before noting the readings. Figure 8 illustrates that the permeate water flux increased as the Al2O3 content increased from 1 to 20%. The increased permeate flux for membranes with high Al2O3 loading can be explained based on two mechanisms. First, the hydrophilic surface of the membrane at high Al2O3 loading facilitates the transport of water through the membrane (Figure 7). Second, the agglomeration of Al2O3 particles (Figure 3c) at high loading (20%) results in the formation of relatively large pores in the membrane, thus contributing to the higher permeate flux.

Water Flux Measurements: Effect of Transmembrane Pressure Difference and Aluminum Oxide Loading
The effects of transmembrane pressure difference, aluminum oxide loading, and time, on the water flux through the membranes, were studied, as shown in Figure 8. The transmembrane pressure difference was varied from 1 to 40 psi. A nearly linear relationship exists between the pressure and flux for all membranes with different Al 2 O 3 loadings. The water permeate flux was measured by holding the transmembrane pressure difference constant at 20 psi for 30 min. The permeate flux values for different pressures were obtained using the same procedure. For each reading, the pressure was maintained for 10 min before noting the readings. Figure 8 illustrates that the permeate water flux increased as the Al 2 O 3 content increased from 1 to 20%. The increased permeate flux for membranes with high Al 2 O 3 loading can be explained based on two mechanisms. First, the hydrophilic surface of the membrane at high Al 2 O 3 loading facilitates the transport of water through the membrane (Figure 7). Second, the agglomeration of Al 2 O 3 particles (Figure 3c) at high loading (20%) results in the formation of relatively large pores in the membrane, thus contributing to the higher permeate flux.

Cadmium Removal
The Cd(II) ion removal studies were performed in the flow loop system, as shown in Figure 2. The cadmium solution was passed through the CNT-Al 2 O 3 membrane. Cd(II) ions are retained in the feed side while purified water permeates the membrane. The cadmium removal (R) can be determined using the following equation: where C p and C f are the concentrations of the solute in the permeate and feed, respectively.

Effect of Feed pH
Solution pH is an important parameter that determines the removal of toxic metal by carbon-based materials. The Cd(II)removal experiments were performed for the membrane with 10% Al 2 O 3 loading in the pH range of 3-10 (experimental set 1). The results of the analysis are shown in Figure 9.

Cadmium Removal
The Cd(II) ion removal studies were performed in the flow loop system, as shown in Figure 2. The cadmium solution was passed through the CNT-Al2O3 membrane. Cd(II) ions are retained in the feed side while purified water permeates the membrane. The cadmium removal (R) can be determined using the following equation: where Cp and Cf are the concentrations of the solute in the permeate and feed, respectively.

Effect of Feed pH
Solution pH is an important parameter that determines the removal of toxic metal by carbonbased materials. The Cd(II)removal experiments were performed for the membrane with 10% Al2O3 loading in the pH range of 3-10 (experimental set 1). The results of the analysis are shown in Figure 9. Cadmium species are present in deionized (DI) water in the form of Cd 2+ , Cd(OH) + , and Cd(OH)2(s) [32,33]. At pH < 8, the dominant cadmium species is Cd 2+ in the form of complex [Cd(H2O)6] 2+ [34]. The pHPZC value for Al2O3-doped CNTs is pH 6.5, as shown in Figure 4. This value demonstrates that Al2O3 is basic in DI water [27]. At pH < pHPZC, the membrane surface is positively charged, and repulsion exists between the Cd(II) ions and surface, causing a low removal rate for Cd(II) ions. In addition, competition between H + and Cd 2+ ions for the active sites decreases the Cd(II) ions adsorption rate. At pH > pHPZC, the surface of the membrane becomes more negatively charged, and thus, additional Cd(II) ions are attracted to the surface, due to electrostatic interactions.
A maximum removal of 84% was observed at pH 8. However, cadmium might have precipitated as Cd(OH)2 at pH > 8, as reported elsewhere [35,36], and the removal might be due to both adsorption and precipitation. Therefore, pH 7 was used in all experiments as an optimum value to avoid the precipitation of cadmium ions. Moreover, at pH > 8, the concentration of Cd(II) ions is low, and the predominant ions are HCO . The repulsion between the negatively charged surface of the membrane at pH > pHPZC, and the HCO ions, causes a decrease in the removal of cadmium ions. In addition to electrostatic interactions, the van der Waals interactions occurring between the cadmium ions and carbon atoms could also induce the adsorption of cadmium ions [27,37]. Cadmium species are present in deionized (DI) water in the form of Cd 2+ , Cd(OH) + , and Cd(OH) 2(s) [32,33]. At pH < 8, the dominant cadmium species is Cd 2+ in the form of complex [Cd(H 2 O) 6 ] 2+ [34]. The pH PZC value for Al 2 O 3 -doped CNTs is pH 6.5, as shown in Figure 4. This value demonstrates that Al 2 O 3 is basic in DI water [27]. At pH < pH PZC , the membrane surface is positively charged, and repulsion exists between the Cd(II) ions and surface, causing a low removal rate for Cd(II) ions. In addition, competition between H + and Cd 2+ ions for the active sites decreases the Cd(II) ions adsorption rate. At pH > pH PZC , the surface of the membrane becomes more negatively charged, and thus, additional Cd(II) ions are attracted to the surface, due to electrostatic interactions.

Effect of Time
A maximum removal of 84% was observed at pH 8. However, cadmium might have precipitated as Cd(OH) 2 at pH > 8, as reported elsewhere [35,36], and the removal might be due to both adsorption and precipitation. Therefore, pH 7 was used in all experiments as an optimum value to avoid the precipitation of cadmium ions. Moreover, at pH > 8, the concentration of Cd(II) ions is low, and the predominant ions are HCO − 3 . The repulsion between the negatively charged surface of the membrane at pH > pH PZC , and the HCO − 3 ions, causes a decrease in the removal of cadmium ions. In addition to electrostatic interactions, the van der Waals interactions occurring between the cadmium ions and carbon atoms could also induce the adsorption of cadmium ions [27,37].

Effect of Time
To study the effect of time on the removal of Cd(II) ions, the experiments were performed at constant pH 7, and an initial concentration of 1 ppm. The samples were collected every 30 min and analyzed. The results of the analysis are presented in Figure 10. The percentage removal of Cd(II) increases with time until 2 h of operation, after which no significant increase in removal was observed. Equilibrium was reached within nearly 2 h for all membranes. The maximum removal rate of 84% was achieved with the CNT-10% Al 2 O 3 membrane. Membranes with 1% and 20% Al 2 O 3 loadings achieved Cd(II) removal rates of 80% and 74%, respectively, under similar conditions. increases with time until 2 h of operation, after which no significant increase in removal was observed. Equilibrium was reached within nearly 2 h for all membranes. The maximum removal rate of 84% was achieved with the CNT-10% Al2O3 membrane. Membranes with 1% and 20% Al2O3 loadings achieved Cd(II) removal rates of 80% and 74%, respectively, under similar conditions. The relatively higher removal of Cd(II) ions by the CNT-10% Al2O3 membrane might be due to the greater number of adsorption sites than the membrane with 1% Al2O3 loading. The relatively lower removal of Cd(II) ions by the membrane with 20% Al2O3 loading could be attributed to the agglomeration of Al2O3 particles at higher loading i.e., 20% Al2O3 (as discussed in Section 3.1 (Figure 3c)). This agglomeration might create large pores in the membrane that leads to poor separation of Cd(II) ions. Figure 11 presents the effect of the initial concentration of the solution on the percentage removal of Cd(II) ions. The initial concentration was varied from 0.5 to 10 ppm, and the other experimental parameters were pH 6, a contact time of 2 h, and a transmembrane pressure difference of 15 psi. The relatively higher removal of Cd(II) ions by the CNT-10% Al 2 O 3 membrane might be due to the greater number of adsorption sites than the membrane with 1% Al 2 O 3 loading. The relatively lower removal of Cd(II) ions by the membrane with 20% Al 2 O 3 loading could be attributed to the agglomeration of Al 2 O 3 particles at higher loading i.e., 20% Al 2 O 3 (as discussed in Section 3.1 (Figure 3c)). This agglomeration might create large pores in the membrane that leads to poor separation of Cd(II) ions. Figure 11 presents the effect of the initial concentration of the solution on the percentage removal of Cd(II) ions. The initial concentration was varied from 0.5 to 10 ppm, and the other experimental parameters were pH 6, a contact time of 2 h, and a transmembrane pressure difference of 15 psi.

Effect of the Initial Concentration
The removal increased slightly from 78 to 84%, as the concentration increased from 0.5 to 1 ppm, and remained nearly constant until approximately 5 ppm. At an initial concentration of 0.5 ppm, the membrane still had available adsorption sites and did not reach saturation; thus, more Cd(II) ions could be adsorbed. The percentage removal remained constant as the concentration was increased from 1 to 5 ppm. However, as the concentration increased further beyond 5 ppm, the removal decreased slightly, likely because all available adsorption sites were covered by the Cd(II) ions and the membrane had reached its adsorption equilibrium limit. The membrane was effective in removing a low concentration of Cd(II) ions. separation of Cd(II) ions.
3.6.3. Effect of the Initial Concentration Figure 11 presents the effect of the initial concentration of the solution on the percentage removal of Cd(II) ions. The initial concentration was varied from 0.5 to 10 ppm, and the other experimental parameters were pH 6, a contact time of 2 h, and a transmembrane pressure difference of 15 psi.

Adsorption Isotherms
The nonlinear forms of the Langmuir and Freundlich adsorption isotherms for the adsorption of cadmium ions on the CNT-Al 2 O 3 membrane surface are presented in Figure 12. Representative equations and the results of the analysis are summarized in Table 2. The removal increased slightly from 78 to 84%, as the concentration increased from 0.5 to 1 ppm, and remained nearly constant until approximately 5 ppm. At an initial concentration of 0.5 ppm, the membrane still had available adsorption sites and did not reach saturation; thus, more Cd(II) ions could be adsorbed. The percentage removal remained constant as the concentration was increased from 1 to 5 ppm. However, as the concentration increased further beyond 5 ppm, the removal decreased slightly, likely because all available adsorption sites were covered by the Cd(II) ions and the membrane had reached its adsorption equilibrium limit. The membrane was effective in removing a low concentration of Cd(II) ions.

Adsorption Isotherms
The nonlinear forms of the Langmuir and Freundlich adsorption isotherms for the adsorption of cadmium ions on the CNT-Al2O3 membrane surface are presented in Figure 12. Representative equations and the results of the analysis are summarized in Table 2.  In the Langmuir model, qe (mg/g) represents the concentration of adsorbate on the surface of adsorbent, Ce (mg/L) indicates the concentration of adsorbate in solution when equilibrium was reached, qm (mg/g) is the maximum adsorption capacity, and KL is the Langmuir adsorption equilibrium constant (L/mg). In the Freundlich isotherm model, KF (mg/g)·(L/mg) 1/n and n (dimensionless) are Freundlich constants. Mathematica version 10 (Wolfram 2015, Long Hanborough Oxfordshire, UK) was used to plot the isotherm data and determine the values of various parameters.  In the Langmuir model, q e (mg/g) represents the concentration of adsorbate on the surface of adsorbent, C e (mg/L) indicates the concentration of adsorbate in solution when equilibrium was reached, q m (mg/g) is the maximum adsorption capacity, and K L is the Langmuir adsorption equilibrium constant (L/mg). In the Freundlich isotherm model, K F (mg/g)·(L/mg) 1/n and n (dimensionless) are Freundlich constants. Mathematica version 10 (Wolfram 2015, Long Hanborough Oxfordshire, UK) was used to plot the isotherm data and determine the values of various parameters.
Both models can describe the experimental data satisfactorily, but the correlation coefficient (R 2 ) value for the Langmuir isotherm model is slightly higher than that for the Freundlich isotherm model.

Mechanism of Cadmium Ion Removal by the CNT-10% Al 2 O 3 Membrane
The possible mechanism underlying Cd(II) ion interactions with the CNT-Al 2 O 3 membrane is presented in Figure 13. As discussed in Section 3.6.1, the dominant cadmium species in deionized (DI) water is Cd(II), or Cd 2+ , in the form of complex [Cd(H 2 O) 6 ] 2+ at pH 7, used as an optimum value in all experiments. When pH < pH PZC , the membrane surface is positively charged, and the low removal of Cd(II) ions can be attributed to electrostatic repulsions between the Cd(II) ions and the surface of the CNT-Al 2 O 3 membrane. Similarly, the higher removal of Cd(II) ions at pH > pH PZC might be due to the strong electrostatic interactions between the negatively charged CNT-Al 2 O 3 membrane surface and the cationic Cd(II) ions [27].

Mechanism of Cadmium Ion Removal by the CNT-10% Al2O3 Membrane
The possible mechanism underlying Cd(II) ion interactions with the CNT-Al2O3 membrane is presented in Figure 13. As discussed in Section 3.6.1, the dominant cadmium species in deionized (DI) water is Cd(II), or Cd 2+ , in the form of complex [Cd(H2O)6] 2+ at pH 7, used as an optimum value in all experiments. When pH < pHPZC, the membrane surface is positively charged, and the low removal of Cd(II) ions can be attributed to electrostatic repulsions between the Cd(II) ions and the surface of the CNT-Al2O3 membrane. Similarly, the higher removal of Cd(II) ions at pH > pHPZC might be due to the strong electrostatic interactions between the negatively charged CNT-Al2O3 membrane surface and the cationic Cd(II) ions [27]. This observation suggests that electrostatic interaction is the main mechanism involved in the sorption of Cd(II) ions onto the CNT-Al2O3 membrane surface. In addition to electrostatic interaction, Cd(II) ions might also adsorb on the surface of CNT-Al2O3 membrane due to van der Waals interactions (physical adsorption) occurring between Cd(II) ions and carbon atoms in the CNT-Al2O3 composite. At the adsorption saturation of the membrane surface and internal structure, certain pores among the CNTs might be covered by Cd(II) ions, thus blocking further ions from passing through and acting as sieves (size exclusion).

Comparative Analysis
The adsorption capacity and removal efficiency of Cd(II) ions from water for the membrane developed in this study are compared to those of similar studies reported in the literature in Table 3. The estimated maximum Cd(II) adsorption capacity of the CNT-10% Al2O3 membrane is 54.42 mg/g, which is higher than those of related adsorbents (used in batch experiments), such as raw CNTs (1.661 mg/g) [38], acid-modified CNTs (4.35 mg/g) [3], ethylenediamine-functionalized multi-walled This observation suggests that electrostatic interaction is the main mechanism involved in the sorption of Cd(II) ions onto the CNT-Al 2 O 3 membrane surface. In addition to electrostatic interaction, Cd(II) ions might also adsorb on the surface of CNT-Al 2 O 3 membrane due to van der Waals interactions (physical adsorption) occurring between Cd(II) ions and carbon atoms in the CNT-Al 2 O 3 composite. At the adsorption saturation of the membrane surface and internal structure, certain pores among the CNTs might be covered by Cd(II) ions, thus blocking further ions from passing through and acting as sieves (size exclusion).

Comparative Analysis
The adsorption capacity and removal efficiency of Cd(II) ions from water for the membrane developed in this study are compared to those of similar studies reported in the literature in Table 3. The estimated maximum Cd(II) adsorption capacity of the CNT-10% Al 2 O 3 membrane is 54.42 mg/g, which is higher than those of related adsorbents (used in batch experiments), such as raw CNTs (1.661 mg/g) [38], acid-modified CNTs (4.35 mg/g) [3], ethylenediamine-functionalized multi-walled carbon nanotubes (MWCNTs) (25.70 mg/g) [33] and nano-alumina on single-walled carbon nanotube (SWCNTs) (2.18 mg/g) [39]. This result suggests that the CNT-Al 2 O 3 membrane is effective in removing low concentrations of Cd(II) ions from water.

Conclusions
A novel approach was developed to synthesize an aluminum oxide-impregnated CNT membrane. No binder was used in the membrane synthesis; instead, aluminum oxide particles served as a binder to hold the 3D CNT network together. The membrane surface demonstrated extreme hydrophilic behavior and yielded a high water flux. The membrane was able to remove low concentrations of Cd(II) ions from aqueous solution. The removal was affected by the aluminum oxide loading, initial cadmium solution concentration, pH, and time. The maximum Cd(II) removal of 84% was obtained at pH 7 and an initial concentration of 1 ppm using the CNT membrane with 10% Al 2 O 3 loading and 2 h of operation. Membranes with 1% and 20% Al 2 O 3 loadings were able to remove 80% and 74% of Cd(II) ions, respectively, under similar experimental conditions. These results suggest that the CNT-Al 2 O 3 membrane can be effectively used in a continuous filtration system for the removal of Cd(II) ions.