Surfactant modification of banana trunk as low-cost adsorbents and their high benzene adsorptive removal performance from aqueous solution

Abstract

The banana trunk was modified using cetyltrimethylammonium bromide (CTAB), sodium dodecyl sulphate (SDS), poly(ethylene glycol)-block-poly(propylene glycol)-block-poly(ethylene glycol) (Pluronic 123), and 4-(1,1,3,3-tetramethylbutyl)phenyl-polyethylene glycol (Triton X-100) to develop novel low-cost adsorbents for benzene removal from aqueous solution. The surface morphology and functional groups of the synthesized adsorbents were determined by a field emission scanning electron microscope (FESEM) and a Fourier transform infrared (FTIR) spectrophotometer. The Brunauer, Emmett and Teller (BET) analysis and X-ray photoelectron spectroscopy (XPS) were also conducted to study adsorbent characteristics. The benzene adsorptive performance of the synthesized adsorbents was evaluated in a batch adsorption experiment at various experimental conditions. It was found that the highest benzene adsorption capacity (280.890 × 10⁻³ mmol g⁻¹) was obtained for M-TX100-BT. The fundamental adsorption studies revealed that the benzene adsorption process was found to be thermodynamically non-spontaneous and all were fitted well by the Langmuir isotherm model. The adsorption kinetic data obeyed the pseudo-second kinetic model with the film diffusion as the rate-limiting step. The application prospects of the Triton X-100 modified banana trunk adsorbent were demonstrated through the regeneration study which revealed that it can also be repeatedly used for at least up to five-adsorption/desorption cycles and its adsorption capacity was comparable to the literature data of similar adsorbents. Thus, banana trunk agrowaste could be an alternative low-cost adsorbent precursor for the adsorptive benzene removal from an aqueous solution.

---

1. Introduction

In recent decades, water pollution caused by the release of hydrocarbons from petroleum production, transportation, processing and application activities into the environment has raised concerns all over the world. In fact, petroleum monoaromatic compounds such as benzene are frequently detected as the most common hydrocarbons in the industrial effluent and water bodies.¹ This may be attributed by cracked pipelines, underground fuel tank leakages as well as improper effluent discharge.² Benzene is commonly used as a solvent in various chemical and petrochemical industries.³ It is generally categorized as a volatile organic compound which is colourless and flammable. Many studies have reported that benzene is highly toxic and carcinogenic to humans.^4,5 Due to an acute toxicity of benzene attribute to the aquatic environment and human health, stringent regulations have been enforced on the benzene concentration in water bodies.⁶ According to the World Health Organization (WHO) drinking water guidelines, the maximum permissible benzene concentration is 0.01 ppm.⁷

A wide variety of methods for organic pollutant removal from aqueous phase, in particular chemical or thermal oxidation, bioremediation, volatilization, condensation, membrane separation, and adsorption have been reported.^6,8,9 The adsorption process is suggested to be the most cost-efficient and feasible method because it does not require a large amount of energy and additional chemicals,¹⁰ especially adsorption by activated carbon.¹¹ Activated carbon is well known as an effective adsorbent for various organic contaminants due to its high porosity and internal surface area.¹² However, activated carbon is expensive and hard to be regenerated due to the strong interaction between activated carbon and the adsorbed molecules.¹³

In recent years, there is a growing interest in the adsorption process using synthetic adsorbents from natural organic precursors such as agricultural wastes (i.e. agrowastes) for the removal of various types of inorganic (e.g. heavy metals) and organic (e.g. dyes, petroleum hydrocarbons, pharmaceuticals, phenolic compounds and pesticides) pollutants from water and wastewaters.^9,14–20 This is due to the fact that agrowastes are abundant, cheap, renewable and biodegradable. The agrowastes like banana trunk could easily be obtained since banana is one of the important crops cultivated in most tropical countries.²¹ Generally, the banana trunk (pseudo-stem) is disposed at a landfill or left decomposing in the plantation estates after harvesting.²² The inexhaustible supply of banana trunk agrowaste has made it a suitable candidate as an adsorbent precursor for the adsorptive removal of benzene from the aqueous solution. It was reported that the banana trunk contains up to 33% of holocellulose (hemicellulose and cellulose) on a dry basis. The banana trunk has also a high content of pectin and polyphenols in its extract.²³ The banana residues were used previously as adsorbents in adsorptive removal of heavy metal ions such as cobalt (Co(II)), cadmium (Cd(II)), copper (Cu(II)), iron (Fe(II)) and zinc (Zn(II)) as well as methylene blue.^21,22,24 The application of banana trunk based adsorbents for the removal of volatile organic compounds has so far not been reported. However, other agrowaste adsorbents such as angico saw-dust, peat, rice bran, coconut shell were used for the removal of volatile organic compounds.^25–30

It is well-known that raw/unmodified agrowastes have a low adsorption capacity and selectivity which can be enhanced by either through physical and/or chemical modifications.³¹ The chemical modifications of agrowastes by various functional groups can be easily conducted due to high content of hydroxyl groups on their surfaces.^32,33 These modifications may include the use of different types of surfactants to modify the adsorbent surfaces used for removal of various pollutants. For instance, the surfactant surface modification of agrowastes, yeast, activated carbon was studied for the removal of metal ions, oil and dyes,^31,34–40 while the surfactant modified natural zeolites and clays for the removal of hydrophobic organic compounds from aqueous solutions.^41,42 The cationic surfactants like hexadecyltrimethylammonium (HDTMA), n-cetylpyridinium bromide (CPB), benzyltrimethylammonium (BTMA) and benzyldimethyltetradecylammonium (BDTDA) are commonly studied.^11,42–44 The non-ionic surfactant of polyethylene glycol (PEG) was also used to modify montmorillonite which has shown promising results for the removal of volatile organic compounds.⁴⁵ It was suggested that the adsorption capacity of the adsorbents modified with non-ionic surfactants may be higher than that achieved with cationic surfactants. Anionic surfactants such as sodium stearate and sodium dodecyl sulphate (SDS) were also used to modify sawdust, montmorillonite, alumina and bentonite for the removal of total organic carbon (TOC) and dyes.^46–49

Therefore, this study was conducted to evaluate the surfactant modified banana trunk adsorbents prepared using different types of surfactant namely cetyltrimethylammonium bromide (CTAB, cationic), sodium dodecyl sulphate (SDS, anionic) and poly(ethylene glycol)-block-poly(propylene glycol)-block-poly(ethylene glycol) (Pluronic 123, non-ionic) and 4-(1,1,3,3-tetramethylbutyl)phenyl-polyethylene glycol (Triton X-100, non-ionic) for benzene removal from an aqueous solution. The benzene adsorption fundamentals (e.g. adsorption isotherms, kinetics and regeneration) were carried out for the most promising surfactant-modified banana trunk adsorbents so that the benzene adsorption mechanism could be ultimately elucidated. Finally, the adsorption results were compared with the published literature data to demonstrate the potential application of agrowastes in general and banana trunk in particular as adsorbent precursors for the benzene removal process.

2. Materials and methods

2.1. Materials

Benzene (99% purity) was purchased from Fisher Scientific (UK). Isooctane, ethanol, methanol, sodium hydroxide (NaOH pellets), hydrochloric acid (HCl, 37%), cetyltrimethylammonium bromide (CTAB, Fig. 1a), and 4-(1,1,3,3-tetramethylbutyl)phenyl-polyethylene glycol (Triton X-100, Fig. 1b) were obtained from Merck (Germany). Sodium dodecyl sulphate (SDS, Fig. 1c) was purchased from Fluka (Switzerland). Poly(ethylene glycol)-block-poly(propylene glycol)-block-poly(ethylene glycol) (Pluronic 123, Fig. 1d) was bought from BASF (Germany). Freshly prepared double-distilled water was used to prepare solutions used throughout the experiment. Agrowaste of banana trunk used in this study was collected at a nearby residential area (Johor, Malaysia).

Fig. 1 Chemical formula and molecular structure of (a) CTAB, (b) Triton X-100, (c) SDS, and (d) Pluronic 123.

2.2. Adsorbents preparation

The banana trunk was cut into small pieces and dried under the sunlight for 3 days. The naturally dried banana trunk was then further dried in the oven at 50 °C for another two days. The banana trunk was ground and then sieved in order to obtain particles of the desired size of 75–150 μm. The particles were washed by distilled water and ethanol to remove dirt and impurity. The sample was dried in the oven at 50 ± 1 °C overnight and finally kept in a desiccator for modification. This sample was denoted as Raw-BT.

Surface mercerization was conducted onto Raw-BT to improve its surface binding sites. A specified amount of the Raw-BT was immersed into 30% (wt) NaOH solution. The mixture was stirred for one hour at room temperature (30 ± 1 °C). The reacted Raw-BT was then filtered and washed by double-distilled water using vacuum filtration. The mercerized Raw-BT was dried in the oven at 50 ± 1 °C overnight and finally kept in a desiccator for characterization and further modification. This sample was denoted as M-BT.

After surface mercerization by concentrated NaOH solution, four types of surfactants namely CTAB, SDS, Pluronic 123, and Triton X-100 were used to modify the M-BT surface. The surfactant solution was prepared by dissolving a specified amount of the respective surfactant with double-distilled water. The desired concentration of surfactant solution is calculated based on twice of its critical micelle concentration (CMC). The CMC of CTAB, SDS, Pluronic 123, and Triton X-100 are 1.0, 8.2, 0.07, and 0.2 mmol L⁻¹, respectively.^50,51 A certain amount of M-BT (e.g. 500 mg) was immersed into the surfactant solution and stirred for 24 hours. The mixture was then filtered and repeatedly washed with double-distilled water to remove un-reacted surfactant. The sample was then dried in the oven at 50 ± 1 °C overnight and finally kept in a desiccator. The modified M-BT samples by CTAB, SDS, Pluronic 123 and Triton X-100 were denoted as M-CTAB-BT, M-SDS-BT, M-P123-BT, and M-TX100-BT, respectively. The modified M-BT was also prepared by reacting with the Triton X-100 solution having a concentration similar to its CMC and it was denoted as M-1TX100-BT.

2.3. Adsorbents characterization

The surface morphology of the synthesized adsorbents was obtained using a field emission scanning electron microscope (FESEM) (Hitachi S4800, Japan). The Brunauer, Emmett and Teller (BET) specific surface area of the adsorbents was determined by a nitrogen adsorption–desorption method at −195.9 °C (77.3 K) using a BET analyzer (Micromeritics ASAP 2000, USA). The adsorbent surface functional groups were determined using a Fourier transform infrared (FTIR) spectrophotometer (PerkinElmer Spectrum One, USA) equipped with the KBr sampling technique. The X-ray photoelectron spectroscopy (XPS) measurement was conducted using Shimadzu KRATOS AMICUS (Japan) equipped with aluminium Al Kα source (1486.8 eV) with an operating pressure of 10⁻⁹ mbar and current of 10 mA. The XPS spectra were obtained at constant analyzer energy (CAE) mode with 1 eV of energy step size, while the high-resolution XPS spectra of O1s, N1s and C1s were obtained at 25 eV of CAE and 0.05 eV of energy step size.

2.4. Adsorption experiment

A stock of benzene solution (adsorbate) in methanol was prepared using analytical grade benzene (99% purity). A desired concentration of benzene solution was then prepared by diluting the stock solution with double-distilled water. Batch benzene adsorption was carried out using a 50 mL glass Erlenmeyer flask with a glass stopper. Benzene solutions of different initial pH values were prepared by an addition of either 0.1 mmol L⁻¹ HCl or NaOH into the solution. An adsorbent dosage or solid to liquid (S/L) ratio of 1.0 mg mL⁻¹ was used to conduct batch benzene adsorption. The mixture was agitated in a temperature-controlled shaker at 200 rpm for 24 hours of contact time. A single-run experiment was carried out, except for the data which significantly deviated from the overall trend, the experiments were repeated and an average value was instead presented.

The residual benzene in the supernatant was extracted by isooctane and its concentration was determined by an UV-vis spectrophotometer (PerkinElmer, model Lambda 35, USA) at λ_max = 255 nm. The benzene concentration measurement was carried out in triplicate and an average value was presented. The relative standard deviation of benzene concentration measurement was less than 2%.

The adsorption capacity of benzene, q (mmol g⁻¹) was calculated as using eqn (1), meanwhile the adsorption removal efficiency, η (%) was evaluated using eqn (2).

where C₀ is the initial benzene concentration (mmol L⁻¹), C_e is the equilibrium benzene concentration (mmol L⁻¹), V is the volume of benzene solution used for adsorption (L) and W is the mass of adsorbent employed (g).

2.5. Desorption experiment

In order to evaluate the reusability of adsorbents, 100 mg of M-TX100-BT was added into a 100 mL of benzene solution having an initial benzene concentration (C₀) of 2.5 mmol L⁻¹. The mixture was agitated at 200 rpm for 24 hours of contact time. Once the adsorption equilibrium was achieved, the benzene solution was filtered and the equilibrium concentration of the benzene solution was determined by a US-vis spectrophotometer at λ_max = 255 nm. The exhausted M-TX100-BT was added into a 100 mL of 50% ethanol–water (desorbing agent) and was agitated at 200 rpm for 24 hours to ensure sufficient desorption period. The adsorption/desorption process was repeated for five regeneration cycles.

2.6. Error analysis

The validity of each model was analyzed by the linear coefficient of determination, R² and the Pearson's chi-squared test, χ² analysis (eqn (3)) as well as the normalized standard deviation, Δq_e (%) (eqn (4)).where z_exp is the experimental value of adsorption capacity (q_exp, mmol g⁻¹), z_cal is the calculated adsorption capacity (q_cal, mmol g⁻¹), and N is the number of measurements. The best suitable model can be determined when the R² value is equal to or near one whereas the Pearson's chi-squared test, χ² and the normalized standard deviation (Δq_e) are as small as possible.

3. Results and discussions

3.1. Adsorbent characterization

The field emission scanning electron microscope (FESEM) is the most commonly used device to determine the surface morphological properties of the solid samples. Fig. 2 illustrates FESEM image (5000× magnification) of (a) Raw-BT, (b) M-BT, (c) M-CTAB-BT, (d) M-SDS-BT, (e) M-P123-BT, and (f) M-TX100-BT. The Raw-BT has a hard and smooth surface structure. Meanwhile, the M-BT shows a rougher and irregular surface which is due to the alkali treatment (mercerization) of the Raw-BT producing the smooth Raw-BT surfaces being de-structured. On the contrary, the FESEM image of M-CTAB-BT, M-SDS-BT, M-P123-BT, and M-TX100-BT shows more homogenous and smoother surface morphologies as compared to M-BT. This may be attributed by the surfactants coverage on the adsorbent surfaces. Table 1 and ESI 1† show the FTIR wavenumber, υ (cm⁻¹) of the respective functional groups and spectra of BT adsorbents, respectively. The FTIR spectra show that the BT adsorbents show obvious broad O–H band (3600–3200 cm⁻¹), C–H stretching (3000–2850 cm⁻¹), C=C stretching (1680–1620 cm⁻¹), and alcohol C–O stretching (1120–1050 cm⁻¹). In addition, all BT adsorbents underwent CH₂ and CH₃ deformation at 1470–1350 cm⁻¹. This may be attributed by the alkali pre-treatment (mercerization) conducted on the Raw-BT before reacting it with various surfactants. However, there is an obvious and sharp peak at 3000–2850 cm⁻¹ for M-CTAB-BT (ESI 1a†). This is because CTAB possesses the higher number of hydrocarbon chain (C₁₉) as compared to SDS (C₁₂). Moreover, the FTIR spectrum of M-CTAB-BT shows additional adsorption bands at 3700 cm⁻¹, 1550 cm⁻¹ and 1050 cm⁻¹ which represent N–H stretching, N–O (aliphatic) and C–N stretching, respectively. The M-SDS-BT has additional peaks at 1030 cm⁻¹ and 700 cm⁻¹ which indicate S=O stretching and S–OR stretching, respectively (ESI 1b†). The M-TX100-BT has one pronounced adsorption band at 1450 cm⁻¹ which represents the aromatic C=C stretching.

Fig. 2 FESEM image of (a) Raw-BT, (b) M-BT, (c) M-CTAB-BT, (d) M-SDS-BT, (e) M-P123-BT, and (f) M-TX100-BT.

Table 1 FTIR wavenumber (cm⁻¹) of various functional groups

Functional group	Molecular motion	Characteristic wavenumber, υ (cm^−1)
Alkane	C–H stretching	3000–2850
	CH2 & CH3 deformation	1470–1350
Alkene	C=C stretching	1680–1620
Aromatic	C=C stretching	1600–1400
Alcohol	O–H stretching	3200–3600
	C–O stretching	1150–1050
Amine	N–H stretching	3700–3500
	C–N stretching (alkyl)	1200–1025
Nitro	N–O (aliphatic)	1600–1530
Ester	S–OR stretching	900–700
Sulfoxide	S=O stretching	1060–1030

---

3.2. Effect of surfactant types

Benzene adsorption capacity, q_e of the Raw-BT, M-BT and four surfactant modified BT adsorbents (i.e. M-CTAB-BT, M-SDS-BT, M-P123-BT, and M-TX100-BT) was evaluated at an initial benzene concentration of 2.5 mmol L⁻¹. Fig. 3 shows that the M-TX100-BT has the highest q_e (68.18 × 10⁻³ mmol g⁻¹) followed by M-CTAB-BT (43.22 × 10⁻³ mmol g⁻¹), M-P123-BT (37.13 × 10⁻³ mmol g⁻¹), M-SDS-BT (30.97 × 10⁻³ mmol g⁻¹), M-BT (23.49 × 10⁻³ mmol g⁻¹), and Raw-BT (13.72 × 10⁻³ mmol g⁻¹).

Fig. 3 Benzene adsorption performance of raw and modified BT adsorbents. Experimental conditions: initial benzene concentration (C₀), 2.5 mmol L⁻¹; contact time, 24 hours; pH, 7; temperature, 303 K; and adsorbent dosage (S/L), 1.0 mg mL⁻¹.

The specific surface area (S_BET, 2.247 m² g⁻¹) and pore diameter (10.970 nm) of the Raw-BT were found to be the largest followed by the M-TX100-BT (S_BET, 1.754 m² g⁻¹; pore diameter, 8.010 nm) and the M-CTAB-BT (S_BET, 0.834 m² g⁻¹; pore diameter, 4.081 nm). This greater surface area and pore diameter of Raw-BT may help enhancing the surfactant adsorption onto the respective adsorbent surface and thus improving the benzene adsorption capacity. The relative low S_BET and pore diameter of surfactant modified adsorbents suggested that the nitrogen gas could only access to the external surface of the adsorbents due to the occupancy by long and bulky hydrocarbon chains on the adsorbent structures.⁵² It was reported that the removal ability of non-polar organic pollutants from aqueous solution with non-ionic surfactant may be higher because adsorbents modified with non-ionic surfactant (Triton X-100) are more stable than those modified by cationic surfactant (CTAB).⁴⁵ This may be attributed by the complex hydrophobic chain possessed by the non-ionic surfactant.

The XPS spectrum of (a) Raw-BT and (b) M-TX100-BT shows the existing of oxygen (O1s), nitrogen (N1s) and carbon (C1s) peaks which represent the main lignocellulosic components of banana trunk (ESI 2†). Fig. 4a presents the high resolution spectra of the total C1s for both adsorbents. Obviously, there are two component peaks shown at 286.15 eV and 284.75 eV representing the C1 (C–H or C–C) and C2 (C–O–H or C–O–C), respectively. It can be observed that C1 peak is weaker after the Triton X-100 modification, however C2 peak increases slightly after the modification. This proves that the C–O–C bond of the Triton X-100 (which is known as copolymer) has been added to the adsorbent surface. Fig. 4b displays the high resolution spectra of the O1s, by which the component peak is observed at 532.6 eV. As shown in Fig. 4c, the intensity of N1s peak for the M-TX100-BT was found weaker than the N1s peak for the Raw-BT which proved the successful Triton X-100 modification of the M-BT.

Fig. 4 A high resolution XPS spectrum of Raw-BT and M-TX100-BT: (a) C1s; (b) O1s and (c) N1s.

3.3. Effect of initial pH values

The effects of initial pH on benzene adsorption by M-TX100-BT were conducted at five different pH values ranging from 3 to 11 and the results are displayed in ESI 3.† The increment of pH did not significantly increase the benzene adsorption by M-TX100-BT since the benzene adsorption was basically governed by the hydrophobic interactions between benzene molecules and the non-ionic Triton X-100 chains on the adsorbent surfaces. This indicates that there was no ion-exchange taking part in benzene adsorption,⁵³ which means that the M-TX100-BT is stable over the initial adsorbate pH range.⁵⁴

3.4. Effect of initial benzene concentrations

The effect of benzene initial concentrations on benzene adsorption performance was evaluated in terms of benzene removal efficiency, η (%) and benzene adsorption capacity, q_e (mmol L⁻¹) carried out within the initial benzene concentration, C₀ range of 0.1–10.0 mmol L⁻¹, adsorbent dosage (i.e. solid to liquid ratio), S/L range of 0.5–5 mg L⁻¹, temperature, T of 30 ± 1 °C and initial pH, pH_i of 7 using M-1TX100-BT and M-TX100-BT adsorbents. The benzene removal efficiency, η (%) was found very dependent on the initial benzene concentration and adsorbent dosages. For instances, at higher C₀ (i.e. 2.5 mmol L⁻¹), the η was found to be 3.0% (S/L = 2.5 mmol mL⁻¹) and 13.4% (S/L = 5 mmol mL⁻¹), while at lower C₀ (i.e. 0.1 mmol L⁻¹), the η was 14.3% (S/L = 1 mg mL⁻¹) and 61.6% (S/L = 5 mg mL⁻¹). These results indicate that the higher η can be obtained when the benzene concentration is relatively low and higher adsorbent dosage. This is due to the increase of the adsorptive sites at higher adsorbent dosage and limited amount of benzene molecules at low initial benzene concentration.

Fig. 5 shows the effect of the initial benzene concentration on the benzene adsorption capacity for both M-1TX100-BT and M-TX100-BT adsorbents observed for the adsorbent dosages of 0.5 and 1.0 mg mL⁻¹. The results clearly show that the M-TX100-BT has the higher benzene adsorption capacity, q_e as compared to M-1TX100-BT. This may be attributed by a higher amount of Triton X-100 loading on the adsorbent surfaces which enhanced the benzene adsorption performance. Contrarily, the q_e decreased by 19.45% as a result of increasing adsorbent dosage from 0.5 to 1.0 mg mL⁻¹. This might be due to the fact that the increment of benzene adsorption is not proportional to the increase of adsorbent active sites as a result of the thermodynamic limitations. Thus, the system reached equilibrium with unsaturated adsorbent active sites due to limited availability of benzene molecules in solution. It can be obviously observed that the q_e for benzene increased rapidly from 0.5 to 5.0 mmol L⁻¹, which might be due to only a limited amount of benzene molecules present in the solution at lower initial benzene concentrations. Hence, the amount of benzene molecules available for adsorption was found to be lower too. However, the q_e of all cases achieved a plateau at the initial benzene concentration of above 5.0 mmol L⁻¹, at which the adsorbent surface active sites are generally fully occupied by benzene molecules.

Fig. 5 Effect of initial benzene concentrations on benzene adsorption capacity of M-1TX100-BT and M-TX100-BT. Experimental conditions: contact time, 24 hours; pH, 7; temperature, 303 K; and adsorbent dosage (S/L), 0.5–1.0 mg mL⁻¹.

The benzene adsorption isotherm data were further analysed using the Langmuir (eqn (5)), Freundlich (eqn (6)), Temkin (eqn (7)) and Dubinin–Radushkevich (D–R, eqn (8)) isotherm models.⁵⁵

q_e = K_FC_e^1/n (6)

q_e = q_max exp(−βε²) (8)

where C_e is the equilibrium concentration (mmol L⁻¹); q_e is the adsorption capacity (mmol g⁻¹); q_max is the maximum adsorption capacity (mmol g⁻¹); K_L is the Langmuir constant; K_F is the Freundlich constant; n is the intensity of adsorption; a is the Temkin isotherm constant (L g⁻¹); b_T is the Temkin constant in relation to heat of adsorption (kJ mol⁻¹); R is the universal gas constant (8.314 J (mol⁻¹ K⁻¹)); T is the adsorption temperature (K); β is the constant proportional to the liquid molar volume; and ε is the Polanyi potential. The linearized isotherm model equations are tabulated in Table 2.

Table 2 Linearized isotherm model equations

Isotherm models	Linearization	Linear plotting
Langmuir		versus Ce
Freundlich		ln qe versus ln Ce
Temkin		qe versus ln Ce
Dubinin–Raduchkevich (D–R)	ln qe = ln qmax − βε^2	qe versus ε^2

---

Table 3 presents the isotherm model parameters obtained from the linear fitting of the respective isotherm models. It was found that the R² values for the Langmuir model are relatively higher than the Freundlich model. This suggests that the Langmuir model is more applicable for describing the benzene adsorption isotherm data. This is also supported by the lower Δq_e and χ² values. In addition, the q_L,max obtained from the Langmuir model equation (eqn (4)) was reasonably close to q_exp,max. Thus, it confirmed that the benzene adsorption is well-fitted by the Langmuir isotherm model. It is assumed that there was no further interaction between the adsorbent surface and benzene molecules when the M-TX100-BT surfaces were saturated with a single-layer of the benzene molecules (monolayer adsorption). Generally, it can be summarized that the best fitted adsorption isotherm models fall in the following order: Langmuir > Temkin > D–R > Freundlich. The non-linear fittings of the Langmuir and Temkin models by utilizing the model parameters tabulated in Table 3 can be found in ESI 4.†

Table 3 Isotherm model parameters obtained from benzene adsorption isotherm data analysis

Parameters/samples	M-1TX100-BT (S/L = 1/1 mg mL^−1)	M-TX100-BT (S/L = 1/1 mg mL^−1)	M-TX100-BT (S/L = 1/2 mg mL^−1)
qexp,max (×10^3 mmol g^−1)	154.533	226.200	280.890
Langmuir
KL (L mmol^−1)	0.315	0.463	0.774
qL,max (×10^3 mmol g^−1)	215.295	285.551	322.997
R^2	0.911	0.994	0.980
Δqe (%)	5.398	4.951	4.320
χ^2	0.035	0.041	0.025
Freundlich
KF (L^n mmol^n mmol^−1 g^−1)	0.049	0.090	0.141
n	1.622	2.060	2.797
R^2	0.838	0.890	0.870
Δqe (%)	5.752	5.115	4.342
χ^2	0.045	0.061	0.049
Temkin
a (L g^−1)	2.618	2.783	4.921
bT (kJ mol^−1)	47.465	32.946	31.556
R^2	0.961	0.961	0.952
Δqe (%)	5.703	5.589	4.791
χ^2	0.023	0.015	0.015
Dubinin–Raduchkevich (D–R)
qD–R,max (×10^3 mmol g^−1)	389.947	665.415	995.764
E (kJ mol^−1)	12.910	12.195	11.636
R^2	0.914	0.926	0.901
Δqe (%)	21.088	26.865	35.381
χ^2	1.468	2.533	3.906

---

The nature of interactions between benzene and adsorbent active sites can be elucidated by analysing the magnitude of the mean free energy (E) of the D–R isotherm model which is either physisorption (<8 kJ mol⁻¹), ion-exchange (8–16 kJ mol⁻¹) or chemisorption (20–40 kJ mol⁻¹).^56,57 Table 3 shows the calculated E values for all cases which lie between 8–16 kJ mol⁻¹ suggesting that the benzene adsorption by M-TX100-BT adsorbents was a chemical ion-exchange process. This should not be the case since the non-ionic nature of the benzene molecules and the hydrophobic surface of surfactant modified banana trunk adsorbents. It should be noted that the benzene adsorption by M-TX100-BT was found less fitted in the D–R isotherm model as compared to Langmuir and Temkin isotherm models. The Temkin model expresses the effects of adsorbate–adsorbate interactions indirectly on the benzene adsorption isotherm. The increase of adsorbate molecules coverage on the adsorbents would cause the reduction of the Temkin constant (b_T).⁴² This was observed in this study in which the b_T value was reduced when the adsorption capacity (q_exp,max) increased as shown in Table 3.

3.5. Effect of temperatures

Fig. 6 shows the effect of three different temperatures, ranging from 30 ± 1 to 50 ± 1 °C (303.15 to 323.15 K) on the benzene adsorption by M-TX100-BT at an initial benzene concentration of 2.5 mmol L⁻¹. The benzene adsorption capacities, q_e of each temperature are tabulated in Table 4. It was found that the benzene adsorption capacity, q_e increased with the temperature from 303.15 to 323.15 K. This indicates that the benzene adsorption by M-TX100-BT is controlled by an endothermic reaction.⁵⁸

Fig. 6 Effect of contact time on benzene adsorption capacity of M-TX100-BT at various reaction temperatures. Experimental conditions: initial benzene concentration (C₀), 2.5 mmol L⁻¹; pH, 7; and adsorbent dosage (S/L), 1.0 mg mL⁻¹.

Table 4 Thermodynamic parameters for benzene adsorptive by M-TX100-BT

Parameters	Temperature, T (K)
	303.15	313.15	323.15
Adsorption capacity, qe (×10^3 mmol g^−1)	133.55	143.86	155.68
Gibbs energy change, ΔG (kJ mol^−1)	7.24	7.28	7.28
Enthalpy change, ΔH (kJ mol^−1)		7.45
Entropy change, ΔS (J mol^−1 K^−1)		1.98

---

The change of thermodynamic parameters over a temperature range between 303.15 and 313.15 K was analysed. The thermodynamic relations of the Gibbs free energy, enthalpy and entropy are given by eqn (9)–(11), respectively and their calculated values are tabulated in Table 4.⁵⁹

ΔG = −RT ln k (9)

ΔG = ΔH − TΔS (10)

where ΔG is defined as the Gibbs free energy change (kJ mol⁻¹); k is the thermodynamic equilibrium constant; R is the ideal gas constant (8.314 J mol⁻¹ K⁻¹); T is the temperature in Kelvin (K); ΔH is the enthalpy change; and ΔS is the entropy change. The ΔH and ΔS of the adsorption process are obtained from the slope and the intercept of the linear plot of ln k versus 1/T, respectively. The k is determined by eqn (12), while the q_e is the adsorption capacity (mmol g⁻¹) and C_e is the equilibrium benzene concentration (mmol L⁻¹).

The positive ΔG shows that benzene adsorption by M-TX100-BT is an endergonic reaction. It is also recognized as a thermodynamically reversible non-spontaneous or an unfavourable reaction. The magnitude of ΔG increases with the temperature indicating that the benzene adsorption is more indicating that the benzene adsorption is more rapid at higher temperature.⁶⁰ As a result, the q_e increases with temperature.⁵⁹ It was reported that the ΔG value for physical adsorption (physisorption) falls between −20 to 0 kJ mol⁻¹, however, ΔG value for chemical adsorption (chemisorption) lies in the range −400 to −80 kJ mol⁻¹.⁵⁸ Generally, the magnitude of ΔG tabulated in Table 4 suggests that the benzene adsorption by M-TX100-BT might be classified as physisorption. Furthermore, the positive ΔH signposts that the benzene adsorption by M-TX100-BT is an endothermic reaction.⁶¹ Likewise, the ΔH in the order of 25 kJ mol⁻¹ suggests physisorption, while the ΔH in the order of 200 kJ mol⁻¹ represents chemisorption.⁶² Thus, the benzene adsorption might be justified as physisorption. It was reported that the low ΔH (7.45 kJ mol⁻¹) is obtained due to the physical interactions like non-polar interactions, water bridging, hydrogen bonding, ion exchange and Triton TX-100 aggregates.⁵⁶ The positive ΔS value may indicate that the benzene molecules became more disordered after adsorption onto the adsorbent surfaces. Generally, in the bulk water, benzene molecules are more ordered in which they are surrounded tightly bound with water molecules. In contrast, when the M-TX100-BT comes into close contact with benzene molecules in the solution, the ordered water molecules in the hydration layers are disturbed and thus the system becomes less ordered.⁶³

3.6. Effect of contact times

Fig. 6 also illustrates the effect of contact time on the benzene adsorption by M-TX100-BT at an initial benzene concentration of 2.5 mmol L⁻¹. It can be observed clearly that the benzene adsorption capacity, q_e increased very rapidly in the first 300 minutes. Thereafter, the q_e values increased gradually for another 200 minutes before reaching equilibrium after 720 minutes (12 hours) and become a plateau. The high adsorption rate at the first 300 minutes occurred due to the existence of a large amount of active site vacancies on the M-TX100-BT surfaces which causes the strong hydrophobic interaction forces between the M-TX100-BT active sites and the benzene molecules. Fig. 6 also shows that the q_e increased with the temperature as previously discussed in Section 3.5.

The adsorption kinetic data were further analysed which allow the determination of the process mechanism. This was done by assuming that the benzene adsorption mechanism can be described by four consecutive rate controlling steps: (a) external mass transfer from bulk solution to the adsorbent surface; (b) film diffusion across the liquid film from the adsorbent surfaces; (c) intraparticle diffusion (i.e. surface diffusion, pore diffusion or a combination of both processes); and (d) surface interactions between adsorbent active sites and benzene molecules.^64,65

(a) External mass transfer. In order to analyse the adsorption kinetic data, the external mass transport parameter should firstly be determined. The external mass transfer coefficient is evaluated by eqn (13), which was proposed by Furusawa and Smith in 1973.⁶⁶

When t → 0 then C_A → C_A0, eqn (14) can be derived.⁶⁷

where C_A is the adsorbate concentration (mmol L⁻¹); C_A0 is the initial adsorbate concentration (mmol L⁻¹); m is the adsorbent mass (g); S is the adsorbent external surface area per unit mass (m² g⁻¹); V is the adsorbate volume (m³); and k_L is the external mass transfer coefficient in the aqueous phase (m min⁻¹).

Additionally, k_L equals to the D_AB/δ, by which D_AB is the molecule diffusivity and δ is the Nernst film thickness. The k_L is inversely proportional to the solution viscosity according to the Wilke–Chang correlation.⁶⁸ The Wilke–Chang correlation (eqn (15)) is employed to evaluate the molecular diffusion of adsorbate molecules, D_AB in an aqueous solution.⁶⁹

where ϕ is the associate parameter of water (2.60); M_B is the water molecular weight (18.0 g mol⁻¹); T is the temperature (K); η_b is the water viscosity (0.89 cP); and V_A is the liquid molar volume at normal boiling temperature (m³ mol⁻¹).

Table 5 shows that the k_L increases with temperature from 303.15 to 323.15 K. This is due to the fact that the adsorbate molecular diffusion increases with temperature as given by eqn (15). In addition, at higher temperature the benzene molecules possess a higher internal energy (i.e. vibrational energy, translational energy and rotational energy) to transfer from one active site to another.⁶⁷

Table 5 Kinetic parameters obtained from benzene adsorption kinetic data analysis

Parameters		T = 303.15 K	T = 313.15 K	T = 323.15 K
Experimental adsorption capacity, qexp (×10^3 mmol g^−1)		140.000	150.000	160.000
External mass transfer coefficient, kL,exp (×10^7 m min^−1)		3.97	4.53	6.80
Diffusion based kinetic models	Fick's law
	Dfilm (×10^12 m^2 min^−1)	1.033	1.036	1.119
	R^2	0.938	0.923	0.781
	Boyd plot
	Deff (×10^12 m^2 min^−1)	2.564	3.205	7.404
	R^2	0.938	0.923	0.889
	Weber–Morris
	k1 (mmol g^−1 min^−0.5)	0.012	0.013	0.018
	R1^2	0.984	0.994	0.992
	k2 (mmol g^−1 min^−0.5)	0.005	0.005	0.002
	R2^2	0.966	0.965	0.693
	Ea (kJ mol^−1)	16.400
Chemical reaction based kinetic models	Pseudo-first order
	qcal (×10^3 mmol g^−1)	46.088	46.668	30.400
	k1 (g mmol^−1 min^−1)	0.002	0.002	0.003
	R^2	0.584	0.526	0.803
	Δqe (%)	77.866	80.650	84.764
	χ^2	5.355	8.319	14.996
	Pseudo-second order
	qcal (×10^3 mmol g^−1)	135.943	144.592	159.297
	k2 (g mmol^−1 min^−1)	0.353	0.428	0.574
	R^2	0.999	0.999	0.999
	Δqe (%)	12.966	12.010	12.902
	χ^2	0.009	0.004	0.001
	Ea (kJ mol^−1)	19.767
	Elovich
	β (g mmol^−1)	52.356	63.291	87.719
	α (mmol g^−1 min^−1)	0.031	0.181	16.123
	R^2	0.809	0.793	0.810
	Δqe (%)	16.064	13.799	13.462
	χ^2	0.010	0.006	0.003

---

(b) Film diffusion. The film diffusion coefficient (D_film) of adsorbate (e.g. benzene) in the stagnant film can be derived from the Fick's law model as given by eqn (16).⁷⁰where F is the function of q_t/q_e, q_t and q_e is the adsorption capacity at any time t and t equilibrium (mmol g⁻¹), respectively; while a is the adsorbent particle radius (5.63 × 10⁻⁵ m). Table 5 shows the D_film values for M-TX100-BT at various temperatures indicating it increased with temperature.

(c) Intraparticle diffusion. The intraparticle diffusion involves the diffusion of solute into the adsorbent pore (pore diffusion) and solute movement at the adsorbent surface active sites (surface diffusion). The effective diffusion coefficient (D_eff) is the combination of pore diffusion, D_p (m² min⁻¹) and surface diffusion, D_s (m² min⁻¹) which can be mathematically expressed by eqn (17).where q is the adsorption capacity (mmol g⁻¹); ρ_s is the solid density (g mL⁻¹); and C is the adsorbate concentration (mmol g⁻¹). The D_eff can be determined from the slope (S_eff) of the Boyd plot (B_t versus t) given by eqn (18).

The B_t is the function of F (=q_t/q_e) which can be calculated using eqn (19) and (20).⁷¹

F > 0.85; B_t = −0.4997 − ln(1 − F) (19)

It was reported that the D_eff is dependent on the solid to liquid ratio (S/L; mg mL⁻¹); initial adsorbate concentration and adsorbent particle size.^68,72 The D_eff values were found to have a similar increasing order in magnitude as the D_film shown with increasing temperatures (Table 5). This indicates that the temperature has a significant effect on adsorbate pore and surface diffusions.

The intraparticle diffusion can be analysed to determine the rate limiting step of the physical steps of the adsorption process by employing the Weber–Morris equation (eqn (21)).^71,73

q_e = k_idt^0.5 (21)

where k_id is the intraparticle diffusion constant (mmol g⁻¹ min^−0.5) and it was obtained from the slope of multi-linear straight lines which represent the first stage of the instantaneous adsorption process (k₁), the second stage of gradual adsorption step where the adsorption rate slows down (k₂), and the third stage of the final equilibrium step (k₃).⁷⁰ The analysis results are tabulated in Table 5. The intraparticle diffusion is not considered as a rate limiting step for the adsorption process, if the linear plot of eqn (20) yields a straight line which does not pass through the origin (ESI 5†).⁷⁴ Otherwise, the adsorption process is classified as boundary layer diffusion (external mass transfer or film diffusion). This result is similar to the lower D_film values obtained in comparison with D_eff values.

(d) Surface chemical interactions. The interactions of benzene molecules with M-TX100-BT external and internal surfaces can be viewed as a chemical process. The adsorption kinetic data can thus be analysed by using the reaction-based kinetic models namely the pseudo-first order kinetic (PFO) model (eqn (22)), the pseudo-second order kinetic (PSO) model (eqn (23)), and the Elovich equation (eqn (24)):

q_t = q_e(1 − e^−k₁t) (22)

where q_t is the adsorption capacity at time t (mmol g⁻¹); q_e is the adsorption capacity at equilibrium (mmol g⁻¹); k₁ is the pseudo-first order equilibrium rate constant; k₂ is the pseudo-second order equilibrium rate constant; α is the initial adsorption rate (mmol g⁻¹ min); and β is the desorption model (g mmol⁻¹). The linearized kinetic model equations are tabulated in Table 6.

Table 6 Linearized kinetic model equations

Kinetic models	Linearization	Linear plotting
Pseudo-first order	ln(qe − qt) = ln qe − k1t	ln(qe − qt) versus t
Pseudo-second order		versus t
Elovich		qt versus ln t
Weber–Morris	qe = kidt^0.5	qe versus t^0.5

---

The kinetic model parameters tabulated in Table 5, were determined from the linearized equations presented in Table 6. The results suggest that the benzene adsorption by M-TX100-BT can be well expressed by PSO kinetic model. This conclusion is drawn based on the fact that the R² values obtained from the fitting data of PSO model are relatively close to unity. Additionally, the Δq_e and χ² values for PSO model fitting are relatively smaller than another two kinetic models. The q_cal (mmol g⁻¹) value from the PSO kinetic model is close to the experimental adsorption capacity, q_exp (mmol g⁻¹).

All these results proved that the benzene adsorption by M-TX100-BT obeyed the PSO kinetic model. The non-linear fitting based on parameters tabulated in Table 5 shows that the PSO model provides the best fitting for the experimental kinetic data (ESI 6†). The constant rate value, k₂ (min⁻¹) of PSO was observed to rise with temperature indicating that the benzene molecules adsorbed by M-TX100-BT at the higher rate at higher temperature.

The pseudo-second order rate constant, k₂ was further employed to determine the activation energy by applying the Arrhenius equation which is shown in eqn (25).⁷⁵

where k₂ is the rate constant at different temperatures; A is the Arrhenius constant; E_a is the activation energy (kJ mol⁻¹), representing the minimum kinetic energy that the reactant (benzene molecules) requires to react with the adsorbent (M-TX100-BT) active sites; R is the universal gas constant (8.314 J (mol⁻¹ K⁻¹)); and T is the temperature (K). The experimental values of E_a and A were determined from the slope and the Y-intercept of the linear plot ln k₂ versus 1/T. It is reported that the physisorption takes part at low E_a (5–40 kJ mol⁻¹), while chemisorption is suggested at higher E_a (40–800 kJ mol⁻¹).⁷⁵ The calculated E_a value is 19.767 kJ mol⁻¹ which suggests that the benzene adsorption by M-TX100-BT is physisorption (Table 5). On the other hand, it is possible to calculate the sorption energy, E_a by applying the k_id obtained from the Weber–Morris equation (eqn (20)) using the Arrhenius equation (eqn (24)) for the determination of either physical or chemical phenomenon controlling the overall adsorption process. The process is controlled by the physical phenomenon if the E_a value is less than 25–30 kJ mol⁻¹. The calculated E_a value is 16.40 kJ mol⁻¹ suggesting that the benzene adsorption by M-TX100-BT is a physical phenomenon and thus controlled by the diffusion process.⁷⁴

3.7. Mechanism of benzene adsorption

Adsorption is a process in which the adsorbate diffuses from the bulk solution to the adsorbent surface and the interaction of the adsorbate to the adsorbent active sites. As previously discussed, the overall rate of the adsorption process was found to be controlled by the physical step namely the film diffusion which was the slowest step compared to other steps. It was thus considered to be the rate-limiting step of the benzene adsorption process. At the adsorbent surface the benzene molecules will interact with the adsorbent active sites through either chemical, physical, ion exchange, and/or chelating process.^64,76 As previously discussed, the benzene adsorption data obey the Langmuir isotherm model which suggests that the M-TX100-BT surfaces are saturated with a single-layer (monolayer) of adsorbed benzene molecules. The positive ΔH also indicates that the benzene adsorption process is an endothermic (adsorb energy) process. The calculated thermodynamic parameters and activation energy suggested that the nature of the benzene–adsorbent interactions was physisorption. The physisorption is also known as the van der Waal's adsorption where the interaction forces are weak. It is reported that low molecular weight aromatic compounds such as benzene are adsorbed by non-specific attraction such as the van der Waal's forces. The interaction between benzene molecules and M-TX100-BT surface active sites was expectedly weak since both benzene and Triton X-100 molecules do not have strong functional groups that might contribute to a stronger bonding force. As a result, this type of physisorption is thus non-spontaneous as suggested by the positive ΔG values. As stated earlier in Section 3.4, the D–R isotherm model analysis however found that the interaction could also be a chemical ion-exchange process. This deviation might be due to the fact that the model did not fit well to the adsorption isotherm data.

3.8. Application prospects

The potential application of any adsorbents depends among others on their regenerability since it determines the cost of the process.^59,77 The previous published studies reported that the regeneration of exhausted zeolite and activated carbon is somehow energy and time consuming.⁷⁸ In the present study, 50% ethanol–water mixture was used to desorb the benzene loaded M-TX100-BT.⁷⁹ Fig. 7 shows the benzene adsorption performance of the regenerate M-TX100-BT for five-adsorption/desorption cycles. It was found that the benzene adsorption capacity, q_e of M-TX100-BT presented a slight drop from 117.50 × 10⁻³ mmol g⁻¹ (first cycle) to 107.49 × 10⁻³ mmol g⁻¹ (fifth cycle). It marked only approximately 8.5% of q_e reduction. This shows that the benzene can be desorbed easily from the benzene loaded M-TX100-BT. In addition, there was no significant surfactant (i.e. Triton X-100) leaching observed in each adsorption–desorption cycle. This discovery is among the utmost advantages of mercerized and Triton X-100 modified banana trunk compared to the conventional adsorbents such as activated carbons and zeolites. Table 7 shows that the q_e (mmol g⁻¹) is highly dependent on the adsorbent dosage (S/L, mg mL⁻¹) and initial concentration (C₀, mmol L⁻¹). The M-TX100-BT gives a relatively better benzene adsorption capacity over some similar adsorbents such as angico sawdust, peat and commercial crystalline cellulose, but lower than cellulose microfibrils and coconut shell activated carbons. The modified BT adsorbent has favorably improved the benzene adsorption capacity, q_e up to 397% as compared to the raw BT adsorbent.

Fig. 7 Adsorption performance of M-TX100-BT as function of adsorption cycles. Experimental conditions: initial benzene concentration (C₀), 2.5 mmol L⁻¹; contact time, 24 hours; pH, 7; temperature, 303 K; and adsorbent dosage (S/L), 1.0 mg mL⁻¹.

Table 7 Benzene adsorption capacity for various adsorbents^a

Precursor	Modification		Adsorption capacity, qe (×10^3 mmol g^−1)	Experimental conditions	Reference
a ^EExperimental data; ^LLangmuir data; C0, initial concentration (mmol L^−1); T, temperature (°C); and S/L, solid to liquid ratio (mg mL^−1).
Banana trunk	Unmodified		13.718^E	pH: 6.5–7, T: 30, C0: 2.5, C^*0: 10.0, S/L: 1.0	Present study
	Mercerized		23.488^E
	Mercerized and loaded with surfactant	CTAB	43.219^E
		SDS	30.974^E
		Pluronic 123	37.126^E
		Triton X-100	68.180^E
			280.890^E*
			322.997^L*
Angico sawdust	—		0.028^E	T: 25, C0: 0.0013, S/L: 7.69	25
Peat	—		0.089^E
Commercial microcrystalline cellulose	Long hydrocarbon chain grafted and esterified with octanoic anhydride by solvent exchange		210.000^E	pH: 6.5–7, T: 25, C0: 2.5, S/L: 10	79
			220.000^L
Cellulose microfibrils (jute fibres)	Organosolv treated (delignification) and esterified with palmitic anhydride		340.502^L	T: 25, C0: 20, S/L: 0.5	80
	Esterified with palmitic anhydride		501.792^L
Coconut shell	Carbonized and potassium hydroxide activated		∼704.045^E	T: 30, S/L: 3.33, C0: 3.2	30
			2723.630^L
	Carbonized, potassium hydroxide activated and treated with ammonia		∼704.045^E
			4414.106^L

---

4. Conclusions

Agrowaste of banana trunk has the potential to be a precursor for benzene adsorbent synthesis because it is abundant and cheap. The surfactant modified banana trunk changed its surface chemistry; thus the benzene adsorption performance was enhanced. The banana trunk modified by Triton X-100 was found to have the highest benzene adsorption capacity which was independent of pH, but increasing with the initial benzene concentration, decreasing dosage (S/L) and increasing temperature. The benzene adsorption data of the M-TX100-BT obeyed the Langmuir isotherm model and the PSO kinetic model having the film boundary diffusion (physical phenomenon) as the rate limiting-step. The chemical surface interactions between the benzene molecules and adsorbent active sites were physisorption in nature. The synthesized adsorbent namely M-TX100-BT could be regenerated and reused up to five adsorption/desorption cycles with only a slight reduction in benzene adsorption capacity. These results demonstrate that the pre-treated and functionalized banana trunk could potentially be employed as an adsorbent for the adsorptive removal of not only benzene but might also be for other organic compounds through appropriate modifications.

Acknowledgements

The financial supports for the Ph.D scholarship and the Research University Grants (GUP 00H63 and GUP 06H85) from the Universiti Teknologi Malaysia (UTM), Malaysia are gratefully acknowledged.

References

1. M. Aivalioti, I. Vamvasakis and E. Gidarakos, J. Hazard. Mater., 2010, 178, 136–143.
2. L. Seifi, A. Torabian and H. Kazemian, Water, Air, Soil Pollut., 2011, 217, 611–625.
3. H. Hindarso, S. Ismadji, F. Wicaksana and N. Indraswati, J. Chem. Eng. Data, 2001, 46, 788–791.
4. M. Aivalioti, P. Papoulias, A. Kousaiti and E. Gidarakos, J. Hazard. Mater., 2012, 207–208, 117–127.
5. N. Wibowo, L. Setyadhi, D. Wibowo, J. Setiawan and S. Ismadji, J. Hazard. Mater., 2007, 146, 237–242.
6. M. N. Carvalho, M. da Motta, M. Benachour, D. C. S. Sales and C. A. M. Abreu, J. Hazard. Mater., 2012, 239–240, 95–101.
7. H. G. Gorchev and G. Ozolins, WHO Chron., 1984, 38, 104–108.
8. M. Aivalioti, D. Pothoulaki, P. Papoulias and E. Gidarakos, J. Hazard. Mater., 2012, 207–208, 136–146.
9. I. Ali, M. Asim and T. A. Khan, J. Environ. Manage., 2012, 113, 170–183.
10. P. Z. Ray and H. J. Shipley, RSC Adv., 2015, 5, 29885–29907.
11. A. Torabian, H. Kazemian, L. Seifi, G. N. Bidhendi, A. A. Azimi and S. K. Ghadiri, Clean: Soil, Air, Water, 2010, 38, 77–83.
12. M. Ghiaci, A. Abbaspur, R. Kia and F. Seyedeyn-Azad, Sep. Purif. Technol., 2004, 40, 217–229.
13. A. A. M. Daifullah and B. S. Girgis, Colloids Surf., A, 2003, 214, 181–193.
14. M. A. Hubbe, J. Park and S. Park, BioResources, 2014, 9, 7782–7925.
15. M. A. Hubbe, O. J. Rojas, M. Fingas and B. S. Gupta, BioResources, 2013, 8, 3038–3097.
16. M. A. Hubbe, K. R. Beck, W. Gilbert O'Neal and Y. C. Sharma, BioResources, 2012, 7, 2592–2687.
17. M. A. Hubbe, S. H. Hasan and J. J. Ducoste, BioResources, 2011, 6, 2161–2287.
18. Y. Zhou, L. Zhang and Z. Cheng, J. Mol. Liq., 2015, 212, 739–762.
19. V. S. Tran, H. H. Ngo, W. Guo, J. Zhang, S. Liang, C. Ton-That and X. Zhang, Bioresour. Technol., 2015, 182, 353–363.
20. Z. Aksu, Process Biochem., 2005, 40, 997–1026.
21. B. H. Hameed, D. K. Mahmoud and A. L. Ahmad, J. Hazard. Mater., 2008, 158, 499–506.
22. K. Sathasivam and M. R. H. Mas Haris, J. Chil. Chem. Soc., 2010, 55, 278–282.
23. R. G. Medeiros, M. L. Soffner, J. A. Thomé, A. O. Cacais, R. S. Estelles, B. C. Salles, H. M. Ferreira, S. A. Lucena Neto, F. G. Silva and E. X. Filho, Biotechnol. Prog., 2000, 16, 522–524.
24. I. G. Shibi and T. S. Anirudhan, Chemosphere, 2005, 58, 1117–1126.
25. A. S. Costa, L. P. C. Romão, B. R. Araújo, S. C. O. Lucas, S. T. A. Maciel, A. Wisniewski Jr and M. R. Alexandre, Bioresour. Technol., 2012, 105, 31–39.
26. A. Adachi, C. Ikeda, S. Takagi, N. Fukao, E. Yoshie and T. Okano, J. Agric. Food Chem., 2001, 49, 1309–1314.
27. M. Akhtar, S. Moosa Hasany, M. I. Bhanger and S. Iqbal, J. Hazard. Mater., 2007, 141, 546–556.
28. M. Akhtar, M. I. Bhanger, S. Iqbal and S. M. Hasany, J. Agric. Food Chem., 2005, 53, 8655–8662.
29. A. A. Daifullah and B. S. Girgis, Colloids Surf., A, 2003, 214, 181–193.
30. J. Mohammed, N. S. Nasri, M. A. Ahmad Zaini, U. D. Hamza and F. N. Ani, Int. Biodeterior. Biodegrad., 2015, 102, 245–255.
31. S. Ibrahim, I. Fatimah, H.-M. Ang and S. Wang, Water Sci. Technol., 2010, 62, 1177–1182.
32. F. Fu and Q. Wang, J. Environ. Manage., 2011, 92, 407–418.
33. F. Delval, G. Crini, S. Bertini, C. Filiatre and G. Torri, Carbohydr. Polym., 2005, 60, 67–75.
34. C. Namasivayam and M. V. Sureshkumar, Bioresour. Technol., 2008, 99, 2218–2225.
35. S. Ibrahim, H. M. Ang and S. Wang, Bioresour. Technol., 2009, 100, 4292–4295.
36. S. Ibrahim, S. Wang and H. M. Ang, Biochem. Eng. J., 2010, 49, 78–83.
37. B. Zhao, Y. Shang, W. Xiao, C. Dou and R. Han, J. Environ. Chem. Eng., 2014, 2, 40–45.
38. A. Bingol, H. Ucun, Y. K. Bayhan, A. Karagunduz, A. Cakici and B. Keskinler, Bioresour. Technol., 2004, 94, 245–249.
39. G. M. Nabil, N. M. El-Mallah and M. E. Mahmoud, J. Ind. Eng. Chem., 2014, 20, 994–1002.
40. S. Ibrahim, I. Fatimah, H.-M. Ang and S. Wang, Water Sci. Technol., 2010, 62, 1177–1182.
41. T. Munusamy, J. Ya-Ting and L. Jiunn-Fwu, RSC Adv., 2015, 5, 20583–20594.
42. C. B. Vidal, G. S. C. Raulino, A. L. Barros, A. C. A. Lima, J. P. Ribeiro, M. J. R. Pires and R. F. Nascimento, J. Environ. Manage., 2012, 112, 178–185.
43. S. M. Koh and J. B. Dixon, Appl. Clay Sci., 2001, 18, 111–122.
44. M. M. G. R. Vianna, F. R. Valenzuela-Díaz, V. F. J. Kozievitch, J. Dweck and P. M. Büchler, Mater. Sci. Forum, 2005, 498–499, 691–696.
45. H. Nourmoradi, M. Nikaeen, H. H. Khiadani and M. K. Hajian, Chem. Eng. J., 2012, 191, 341–348.
46. F. M. S. E. El-dars, M. Y. M. Hussien and A. H. T. Kandil, Int. J. Appl. Sci. Eng. Res., 2015, 6, 584–594.
47. R. Ansari, B. Seyghali, a. Mohammad-khah and M. a. Zanjanchi, Sep. Sci. Technol., 2012, 47, 1802–1812.
48. A. Adak, M. Bandyopadhyay and A. Pal, J. Environ. Sci. Health, Part A: Toxic/Hazard. Subst. Environ. Eng., 2005, 40, 167–182.
49. D. M. Chen, J. Chen, X. M. Wang, X. L. Luan, H. P. Ji and F. Xu, Adv. Mater. Res., 2010, 178, 29–34.
50. J. Aguiar, P. Carpena, J. A. Molina-Bolívar and C. Carnero Ruiz, J. Colloid Interface Sci., 2003, 258, 116–122.
51. C. Kim, H.-H. Kwon, I. K. Song, Y.-E. Sung, W. S. Chung and H.-I. Lee, J. Power Sources, 2007, 171, 404–411.
52. H. Koyuncu, N. Yıldız, U. Salgın, F. Köroğlu and A. Calımlı, J. Hazard. Mater., 2011, 185, 1332–1339.
53. C. J. M. Chin, M. W. Shih and H. J. Tsai, Appl. Surf. Sci., 2010, 256, 6035–6039.
54. F. Su, C. Lu and S. Hu, Colloids Surf., A, 2010, 353, 83–91.
55. J. Febrianto, A. N. Kosasih, J. Sunarso, Y. H. Ju, N. Indraswati and S. Ismadji, J. Hazard. Mater., 2009, 162, 616–645.
56. A. R. Cestari, E. F. S. Vieira, G. S. Vieira and L. E. Almeida, J. Colloid Interface Sci., 2007, 309, 402–411.
57. K. Johari, N. Saman, S. T. Song, H. Mat and D. C. Stuckey, Ind. Eng. Chem. Res., 2013, 52, 15648–15657.
58. U. F. Alkaram, A. a. Mukhlis and A. H. Al-Dujaili, J. Hazard. Mater., 2009, 169, 324–332.
59. F. Su, C. Lu, K. R. Johnston and S. Hu, in Environanotechnology, Elsevier, 1st edn, 2010, pp. 71–97.
60. B. S. Chu, B. S. Baharin, Y. B. Che Man and S. Y. Quek, J. Food Eng., 2004, 62, 97–103.
61. S. Nethaji and A. Sivasamy, Chemosphere, 2011, 82, 1367–1372.
62. G. Crini and P. M. Badot, Prog. Polym. Sci., 2008, 33, 399–447.
63. G. D. Sheng, D. D. Shao, X. M. Ren, X. Q. Wang, J. X. Li, Y. X. Chen and X. K. Wang, J. Hazard. Mater., 2010, 178, 505–516.
64. Y. S. Ho, J. C. Y. Ng and G. McKay, Sep. Purif. Rev., 2000, 29, 189–232.
65. R. B. Garcia-Reyes and J. R. Rangel-Mendez, Bioresour. Technol., 2010, 101, 8099–8108.
66. T. Furusawa and J. M. Smith, Ind. Eng. Chem. Fundam., 1973, 12, 197–203.
67. R. Ocampo-Perez, R. Leyva-Ramos, P. Alonso-Davila, J. Rivera-Utrilla and M. Sanchez-Polo, Chem. Eng. J., 2010, 165, 133–141.
68. C.-F. C.-F. Chang, C.-Y. C.-Y. Chang, W. Höll, M. Ulmer, Y.-H. Y.-H. Chen and H.-J. H.-J. Gross, Water Res., 2004, 38, 2559–2570.
69. R. Ocampo-Pérez, M. M. Abdel daiem, J. Rivera-Utrilla, J. D. Méndez-Díaz and M. Sánchez-Polo, J. Colloid Interface Sci., 2012, 385, 174–182.
70. G. L. Dotto and L. A. A. Pinto, J. Hazard. Mater., 2011, 187, 164–170.
71. S. T. Song, N. Saman, K. Johari and H. Mat, Ind. Eng. Chem. Res., 2013, 52, 13092–13101.
72. G. McKay and B. Al-Duri, J. Chem. Technol. Biotechnol., 2007, 48, 269–285.
73. N. Saman, K. Johari, S. S. Tien and H. Mat, Clean: Soil, Air, Water, 2014, 42, 1541–1548.
74. Y. S. Ho, J. C. Y. Ng and G. McKay, Sep. Purif. Rev., 2000, 29, 189–232.
75. H. Nollet, M. Roels, P. Lutgen, P. Van Der Meeren and W. Verstraete, Chemosphere, 2003, 53, 655–665.
76. L. Chen, H. Wang, H. Wei, Z. Guo, M. a. Khan, D. P. Young and J. Zhu, RSC Adv., 2015, 5, 42540–42547.
77. Z. Li and R. S. Bowman, Water Res., 2001, 35, 322–326.
78. S. Alila, F. Aloulou, W. Thielemans and S. Boufi, Ind. Crops Prod., 2011, 33, 350–357.
79. F. Aloulou, S. Boufi and J. Labidi, Sep. Purif. Technol., 2006, 52, 332–342.
80. L. Serrano, I. Urruzola, D. Nemeth, K. Bela and J. Labidi, Desalination, 2011, 270, 143–150.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra00911e

---