- Email: [email protected]
Enhancing oil removal from water by immobilizing multi-wall carbon nanotubes on the surface of polyurethane foam
Journal of Environmental Management 157 (2015) 279e286
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
Enhancing oil removal from water by immobilizing multi-wall carbon nanotubes on the surface of polyurethane foam Alireza Keshavarz a, Hamid Zilouei a, *, Amir Abdolmaleki b, Ahmad Asadinezhad a a b
Department of Chemical Engineering, Isfahan University of Technology, Isfahan 84156-83111, Iran Department of Chemistry, Isfahan University of Technology, Isfahan 84156-83111, Iran
a r t i c l e i n f o
a b s t r a c t
Article history: Received 23 February 2015 Received in revised form 16 April 2015 Accepted 17 April 2015 Available online
A surface modification method was carried out to enhance the light crude oil sorption capacity of polyurethane foam (PUF) through immobilization of multi-walled carbon nanotube (MWCNT) on the foam surface at various concentrations. The developed sorbent was characterized using scanning electron microscopy, Fourier transform infrared spectroscopy, thermogravimetric analysis, and tensile elongation test. The results obtained from thermogravimetric and tensile elongation tests showed the improvement of thermal and mechanical resistance of surface-modified foam. The experimental data also revealed that the immobilization of MWCNT on PUF surface enhanced the sorption capacity of light crude oil and reduced water sorption. The highest oil removal capacity was obtained for 1 wt% MWCNT on PUF surface which was 21.44% enhancement in light crude oil sorption compared to the blank PUF. The reusability of surface modified PUF was determined through four cycles of chemical regeneration using petroleum ether. The adsorption of light crude oil with 30 g initial mass showed that 85.45% of the initial oil sorption capacity of this modified sorbent was remained after four regeneration cycles. Equilibrium isotherms for adsorption of oil were analyzed by the Freundlich, Langmuir, Temkin, and Redlich ePeterson models through linear and non-linear regression methods. Results of equilibrium revealed that Langmuir isotherm is the best fitting model and non-linear method is a more accurate way to predict the parameters involved in the isotherms. The overall findings suggested the promising potentials of the developed sorbent in order to be efficiently used in large-scale oil spill cleanup. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Oil removal Polyurethane foam Multi-walled carbon nanotube Surface modification
1. Introduction In recent decades, oil demand has been escalated because of the rapid industrial development and urban area expansion (Wang et al., 2012). The global oil requirement is mainly transported through seas and oceans which pose a risk of oil spillage and leakage to the marine ecosystem. The spilled oil adversely affects the flora and fauna species such as birds and fish (Aguilera et al., 2010). The common methods employed to solve the oil spill problems include in-situ burning, mechanical methods (using booms, vacuum units, and skimmers), chemical methods (using chemical dispersants), and sorbents (Al-Majed et al., 2012). Each method with its own advantages and disadvantages, is normally selected depending on the type of oil, spill scale, the location, and the climate (Li et al., 2013a,b).
* Corresponding author. E-mail address: [email protected] (H. Zilouei). http://dx.doi.org/10.1016/j.jenvman.2015.04.030 0301-4797/© 2015 Elsevier Ltd. All rights reserved.
The advanced removal and recovery of oil by different sorbents can be regarded as one of the most effective techniques for oil spill cleanup (Banerjee et al., 2006). The criteria to select an ideal sorbent include high sorption capacity, low cost, high selectivity, and reusability (Deng, 2006). Oil sorbents may be divided into three basic types including inorganic mineral sorbents, such as aerogel (Wang et al., 2010), exfoliated graphite (Zheng et al., 2004), and expanded perlite (Bastani et al., 2006), organic natural sorbents, such askapok (Lim and Huang, 2007), saw dust (Husseien and Amer, 2009), and straw (Banerjee et al., 2006), and organic synthetic sorbents, such as polyurethane and polypropylene foams and carbon nanotubes (Gui et al., 2011). Typical advantages of inorganic mineral sorbents are inexpensiveness and easy availability. However, they tend to sink in water. On the other hand, organic natural sorbents are fascinating thanks to their good adsorption capacity, biodegradability and low cost, although they are dusty and have to be dried before using (Duong and Burford, 2006). Polyurethane foams (PUF) are cheap and already commercialized, with low density and good thermal properties. However, poor
280
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
hydrophobic property of PUF reduces its oil sorption efficiency since water is also adsorbed in large extent along with oil. In recent years, the surface modification with an aim to improve the hydrophobicity of PUF has attracted vast interest. Li et al. (2012) improved the oleophilic/hydrophobic properties of PUF for oil sorption by grafting the oleophilic monomer of laurylmethacrylate (LMA) on PUF surface. Their results showed a decrease by 24e50% in water sorption and an increase by 44% in the diesel sorption. Wu et al. (2014) also treated PUF with silica sol and gasoline, and their results showed an improvement of the sorption capacity and oil/ water selectivity after adsorption of 100 g of motor oil versus 0.1 g of water. Carbon nanotubes (CNTs) with two different versions, multiwalled carbon nanotube (MWCNT) and single-walled carbon nanotube (SWCNT), have received great attention in environmental pollution remediation because of their low mass density, high porosity and hollow structure, high specific surface area, strong interaction between CNTs and pollutant molecules, and their hydrophobic propensity (Ren et al., 2011). The relevant studies have indicated that CNTs are effective adsorbents for organic materials such as dyes (Li et al., 2013b; Zhao et al., 2013), benzene (Chin et al., 2010), phenol (Pacholczyk et al., 2011), dimethyl phthalate (Wang et al., 2013), and bisphenol A (Joseph et al., 2011). Fan et al. (2010) used CNTs as oil adsorbent and evaluated its recycling performance. Their results showed 69 g/g of oil sorption and good recyclability due to the unique mechanical strength of CNTs. Gui et al. (2011) also explored the recyclability of CNTs sponges in terms of the oil adsorption capacity. The results showed 112 g/g of sorption for diesel oil and CNTs could attain around 30% of the initial sorption capacity after 10 cycles of adsorption. In this work, carboxylic MWCNT is intended to be immobilized on the surface of PUF in order to enhance the oil sorption capacity and the sorption efficiency of the PUF. Carboxyl-functionalized MWCNT is immobilized on the PUF surface through the reaction between the OH groups existing in PUF structure and carboxyl functionalities on the surface of MWCNT and methylene diphenyl diisocyanate (MDI). Then, the synthesized sorbent is studied in terms of oil sorption capacity, oil sorption efficiency, and reusability. 2. Materials and methods 2.1. Materials Carboxylic multi-walled carbon nanotube was purchased from Neutrino Company (Iran). 1,1-dichoro-1-fluoroethane (HCFC 141b) was purchased from Lin'an E-COOL Refrigeration Equipment CO (China). NIXOL AM-313 polyether polyol was purchased from KPX chemical Co (South Korea) and methylene diphenyl diisocyanate (MDI) was obtained from Daeyang International Co (South Korea). Light crude oil was obtained from Isfahan refinery feed stream (Iran), industrial grade Xylene was purchased from Isfahan petrochemical complex (Iran) and petroleum ether with the boiling range of 30e60 C was supplied by Pars Chemie Co (Iran). General data obtained from the National Iranian Oil Company for characteristics of light crude oil was as below: specific gravity (at 15.56 C) 0.8597, API 33.09, Sulphur content 1.33 wt%, Mercaptane content 67 ppm, Nitrogen content 0.12 wt%, Asphaltenes 1.0 wt%, Wax content 5.1 wt%, Carbon residue conradson 4.00 wt%, Base sediment and water <0.05 vol%, Acidity 0.09 mg KOH/g.
as the chemical blowing agent and 1.5 g HCFC 141b as the physical blowing agent. Then, the resulting mixture was mixed with 4 g MDI by means of a mixer at 1000 rpm. At the end, the mixture was left for 1 h in order for the reaction to complete. 2.3. Surface modification of polyurethane foam The surface modification of PUF was carried out by allowing the reaction among MDI functionalities, carboxylic OH groups existing in MWCNT structure, and the unreacted OH units of polyol present in the foam structure. In this study, the samples with 0.5 wt%, 1 wt%, 2 wt%, and 3 wt% of MWCNT on PUF surface were prepared. To this end, first, 300 ml petroleum ether was poured inside a glass beaker. Then an amount equal to 10 times the final amount of MWCNT on the surface was added to the beaker and stirred for 5 min at 1000 rpm to obtain a homogeneous MWCNT suspension. Amount of 1 g PUF which was already cut into 1 1 1 cm3 cubes was added to the reaction media and the whole content was stirred for 15 min to allow the MWCNT moieties adsorb onto foam surface. Then, while mixing, MDI as much as 2 times the amount of the MWCNT was added to the reaction beaker gradually and stirred for another 40 min period. The foam cubes were taken out and thoroughly dried in an oven at 100 C for 10 min and weighed with high precision. Then, the cubes were washed by petroleum ether to remove the physically adsorbed loose particles trapped in the pores followed by complete drying in the oven at 100 C for 30 min. Finally, they were precisely weighed in order to estimate the amount of the adsorbed MWCNT on foam surface. 2.4. Sorbent capacity, removal and efficiency To study the sorption capacity of the prepared sorbents, light crude oil was used as water pollutant. The sorption experiments were performed based on the Standard Test Method for Sorbent Performance of Adsorbents (ASTMF726-99). Sorption experiments were performed in water-oil system. To do this, 250 ml fresh water was poured in a 600 ml beaker and then crude oil was added to water so that a layer with the thickness 2e4 mm of oil was formed on the water surface. Then, 1.0 g the adsorbent was weighed and added to the beaker. The container was then subjected to shaking on a shaker at the frequency 150 rpm for 15 min. The content of the beaker was allowed to settle for 2 min immediately after shaking. The sorbent cubes were withdrawn by a forceps and allowed to drain for 30 ± 3 s. They were carefully weighed. Then, the cubes were fully washed by the industrial grade Xylene to extract the adsorbed oil and water. The extracted solution was finally used to determine the water content of the oil based on the Standard Test Method for Water in Crude Oil by Distillation (ASTMD4006-81). To this end, the extracted solution was poured into a distillation apparatus. The distillation was performed until the volume of the water collected by the apparatus trap remained constant and no water was visible any longer in the apparatus. When the distillation operation was complete, the trap content was allowed to cool down to 20 C. At the end, the water collected by the trap was precisely weighed and the result was reported as the mass of the adsorbed water. If the value of any run (g/g) deviated by more than 15% from the mean of the three runs, then the samples would be rejected. The oil sorption capacity of the sorbents was calculated by Eq. (1):
ms mw m0 m0
2.2. Synthesis of polyurethane foam
M¼
To synthesize the polyurethane foam at ambient temperature (22 ± 3 C), 10 g polyol was fully mixed with 0.1 g deionized water
where M is the oil sorption capacity (g/g), ms is mass of saturated sorbent (mass of water plus that of oil and sorbent), mw is the mass
(1)
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
of adsorbed water (g), and m0 is initial dry mass of sorbent (g). Oil removal percent was determined by Eq. (2):
Pr ¼
mr 100 mi
(2)
where Pr is oil removal percent, mr is mass of removed oil (g), and mi is initial mass of oil (g). Oil sorption efficiency of the sorbent was defined as the ratio of adsorbed oil mass to total mass of adsorbed oil and water. Thus, Eq. (3) is:
E¼
mr 100 mo þ mw
(3)
where E is oil sorption efficiency, mr is mass of adsorbed oil (g) and mw is mass of adsorbed water. The data collected in the experiments were subjected to the twofactor analysis of variance (ANOVA) to determine the relationship between the adsorption capacity and amount of MWCNT on the surface of PUF and initial weight of light crude oil. Significance of the differences was defined as p < 0.05 in all the cases. The statistical processing was performed using Microsoft Excel software. 2.5. Sorbent regeneration Reusability property of the sorbent with 1 wt% multi-walled carbon nanotube was investigated by the chemical recovery of the sorbent cubes through four cycles of recovery and reuse. In each cycle, the sorbent cubes were washed three times using petroleum ether until no color change of the consumed petroleum ether took place. Finally, the regenerated sorbent cubes were dried inside an oven at 100 C for 30 min. 2.6. Sorbent characterization The morphology of the control sample (blank PUF) as well as the surface-modified PUF samples was studied by scanning electron microscopy (SEM) (TESCAN e VEGA3, SBU e Easy Probe, Czech Republic). FT-IR spectra of the samples were recorded on an FT-IR spectrometer (JASCO, Model 6300, Japan). This technique was in principle utilized to examine the surface chemical groups and to prove the occurrence of desirable surface modification reactions. To study the effect of surface modification on the thermal resistance of PUF, thermogravimetric analysis (TGA) was performed via a thermogravimetric analyzer (STA6000, PerkinElmer, USA). Tensile elongation test was also performed to exploreany changes of the mechanical strength of the developed sorbents by Zwick Universal Testing Machinee1446-60, USA. 2.7. Adsorption isotherms Both linear and non-linear equations of the Langmuir, Freundlich, and RedlichePeterson isotherm models were analyzed to identify which type of equation has better agreement with the experimental data of oil removal from water using blank PUF and MWCNT surface modified PUF. Linear regression analysis was done through least square method to determine the isotherm models parameters. However, linear method leads to uncontrollable parameter estimation error, thus it may be better to directly estimate the model parameters from the non-linear equation (Subramanyam and Ashutosh, 2011). In all cases, non-linear regression was done by use of GraphPad Prism 6 software. 2.7.1. Langmuir isotherm Langmuir model is based on a kinetic view point that propose a coherent adsorption theory onto a flat surface (Do, 1998). The
281
assumptions of the Langmuir model are: (i) surface is homogeneous with constant adsorption energy over the all sites, (ii) adsorption on surface is localized, (iii) monolayer adsorption; that is each site can accommodate only one molecule or atom. Langmuir isotherm is expressed as follow:
qe ¼
qm KL Ce 1 þ KL Ce
(4)
A linear expression of the Langmuir isotherm may be represented by:
Ce Ce 1 ¼ þ qe qm qm KL
(5)
where qe is the equilibrium solid-phase concentration (g g1), Ce is the equilibrium liquid-phase concentration (g L1), qm (g g1) is the maximal adsorption capacity, and KL (L g1) is the Langmuir model constant related to the adsorption free energy. The slope and intercept of a plot of Ce/qe vs. Ce enables the determination of qe and KL, respectively. 2.7.2. Freundlich isotherm Freundlich isotherm model is not restricted to monolayer adsorption and describes the non-ideal and reversible adsorption (Vasanth Kumar and Sivanesan, 2006). This model considers heterogeneous surfaces in the sense that the adsorption energy is distributed (Radhika and Palanivelu, 2006). The Freundlich model can be expressed as: 1=n
qe ¼ KF Ce
(6)
where KF is the constant indicative of the relative sorption capacity of the sorbent (g g1) (L/g)n, and 1/n is the heterogeneity factor. The linear form of Freundlich isotherm is expressed as:
log qe ¼
1 log Ce þ log KF n
(7)
which will have a straight line with a slope of 1/n and an intercept of log(KF) when log(qe) is plotted against log(Ce). 2.7.3. RedlichePeterson isotherm RedlichePeterson isotherm is a three-parameter isotherm model which includes the features of the Langmuir and Freundlich isotherm models (Hamdaoui and Naffrechoux, 2007). The RedlichePeterson isotherm can be described as follows:
qe ¼
KR Ce g 1 þ a R Ce
(8)
where KR (L g1) and aRðL g11=K RÞ are RedlichePeterson equation constants and g is the exponent. The exponent, g, is usually between zero and one (Wu et al., 2010). When g ¼ 1, this isotherm is the same as Langmuir isotherm and when g ¼ 0, the model is the same as Henry's law equation. The linear form of RedlichePeterson isotherm may be written as:
Ce 1 a ¼ þ R Cg qe KR KR e
(9)
To fit the linear equation of RedlichePeterson model, a trial and error procedure was performed.The value of g was assumed, then by substituting the assumed value in the linear equation the regular least square regression was performed and the value of coefficient of determination (R2) was determined. The procedure continued until the difference between the values of the R2obtained from two respective iteration was less than 104.
282
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
2.7.4. Temkin isotherm Temkin isotherm assumes the linear reduction of adsorption heat rather than logarithmic with coverage. The isotherm contains a factor that considers the adsorbenteadsorbate interactions. The Temkin isotherm is given as:
qe ¼
RT lnðKT Ce Þ b
(10)
which can be linearized as:
qe ¼ B1 ln KT þ B1 ln Ce B1 ¼
RT b
(11) (12)
where B1 is related to the heat of adsorption and KT (L g1) is the equilibrium binding constant. The isotherm constants B1 and KT can be determined from the slope and the intercept of a plot of qe versus lnCe, respectively. 2.7.5. Comparison of the models The coefficient of determination shows the fit between the experimental data and isotherm model (Subramanyam and Ashutosh, 2011) and represents the variance about the mean (Foo and Hameed, 2010). However, the value of R2 has no consideration for the degrees of freedom. Other obvious issues with R2 are sensitivity to extreme data points and the potential of being large in the case of models with more parameters (El-Khaiary and Malash, 2011). Therefore, it seems necessary to compare the models with different degrees of freedom based on an index that considers the number of parameters. Akaike's Information Criterion (AIC) (Burnham and Anderson, 2002) is a statistical method that provides an index to compare the models. It determines the more likely correct model based on information theory and maximum likelihood.
2Np Np þ 1 SSE þ 2Np þ AIC ¼ N ln N N Np 1
(13)
where SSE is the sum of squared deviations of the points from the regression curve, N is the number of data points, and Np is the number of parameters in the model. Having smaller AIC shows the more likely better model. 3. Results and discussion 3.1. FT-IR The results of FT-IR spectrometry are shown in Fig. 1. The presence of carboxylic acid groups on MWCNT surface is confirmed through the following identified bands; the strong peak at 1705 cm1 attributed to C]O, a very broad band at 3400-2400 cm-1 characteristic to OeH stretching vibration, and the peak at 1537 cm-1 corresponding to-COO vibration. As for the blank PUF, the peaks at 1070 cm-1 and 3375 cm-1, the strong peak at 1730 cm-1, and the peak at 1535 cm-1are assigned to etheric CeO, NeH, C]O and aromatic rings stretching vibrations, respectively. In the case of surface modified PUF, except for C]O and NeH groups, the observed peaks are almost similar to those of the blank PUF in terms of position and intensity. According to the pertinent spectra, C]O peak at 1126 cm-1 and NeH peak at 3357 cm-1 become weak and broad after surface modification. These changes imply the occurrence of various types of secondary interactions such as hydrogen bonding and also validate the presence of
MWCNT on the foam surface. The strong peak at 1132 cm-1 is associated with the etheric CeO stretching vibration. The peak at 1546 cm-1 attributed to C]C vibration band of aromatic rings. 3.2. SEM The SEM images of the blank PUF along with surface-modified PUFare shown in Fig. 2. Presence of MWCNT on the modified PUF surface is apparent from comparing the respective images (Fig. 2,b and d). Distinctively brighter spots in Fig. 2,d, which are due to the higher electron density, are possibly due to the MWCNTs moieties surrounded by MDI and then reacted with the foam surface. A slight difference between the morphology of the blank PUF and that of the surface-modified PUF is discernible from the respective images (Fig. 2,a and c). It is observed that the cells become slightly wider after surface modification. Therefore, the overall volume of the cells and the ratio of open to close cells are increased. Similar observation has also reported by Harikrishnan et al. (2006), however, in the case of polyurethane foam-clay nanocomposite, they used nanoclays as the cell openers in the structure of the polyurethane foam. The most probable reason behind the increase in cell size can be related with the modification method where the blank cubes are immersed inside petroleum ether. This makes the cubes swell and provide enough space for MWCNT moieties to penetrate into the pores, react with the surface molecules, and take up the pores. Due to this process, even after complete drying, the foam shrinkage is not allowed and the initial pores volume is not restored. In addition to the above observations, one can notice in the SEM images (Fig. 2b and d) the topographical difference between the surface of the blank and that of the modified PUF. The former has a smooth feature while the latter is rough. This is originated from the presence of the chemically attached MWCNTs. 3.3. Tensile elongation Tensile elongation test performed on sorbent samples with 0, 0.5, 1, and 2 wt% MWCNT on their surfaces and the results in terms of Fmax (N) and Elongation (mm) are reported as follow: (4.99 N, 16.08 mm), (8.23 N, 14.55 mm), (9.91 N, 13.27 mm), and (10.58 N, 12.31 mm), respectively. As can be inferred from data, for all samples, the maximal breaking force is increased and the breaking length is decreased, suggesting an increase in mechanical strength. Such an increase in strength can be explained based on the occurrence of strong interactions between MWCNTs and PU chains (like H-bonding) which hinder the mobility of polymer chains and enhance their stiffness under mechanical tensions. Moreover, the addition of MDI to PUF structure increases the extent of the hard segments resulting in an increase in stiffness and ultimately strength of the material. 3.4. TGA The results from TGA performed on the blank PUF and the modified sorbent with 1 wt% MWCNT are illustrated in Fig. 3. The presented curves show a significant difference between the final mass of the blank sample and that of the surface-modified sample. The lower mass loss of the surface-modified PUF, in comparison with that of the blank PUF, implies an improvement of the thermal resistance of the surface-modified sample. This can be attributed to the excellent inherent thermal resistance of MWCNT additive which increases the thermal performance of the PUF after surface modification (Sahoo et al., 2007). Similar degradation trend of the blank and surface-modified PUFs indicates that the bulk structure of the foam is not adversely affected after surface treatment (Alves et al., 2009).
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
283
Fig. 1. FTIR spectra of carboxylic MWCNT, blank PUF, and MWCNT surface-modified PUF.
3.5. Sorbent capacity and efficiency The oil removal percentage and the removal efficiency of the prepared sorbents together with calculated results of two-factor ANOVA are presented in Table 1 and Table 2, respectively. The presence of the chemically immobilized MWCNT units on PUF surface contributes positively to the oil sorbent capacity (Table 1).
Moreover, the results of ANOVA on the oil removal percentage, presented in Table 2, shows that MWCNT significantly improves the sorbent performance of polyurethane foam verified by the low value of P (P < 0.05) and the F > Fcrit (91.37 > 2.53). It is clear that the sorbent with 1 wt% MWCNT on its surface gives the best performance where the maximum sorption occurs at the oil initial mass of 30 g by adsorbing 24.75 g/g (corresponding to 21.44% increase in
Fig. 2. SEM images of blank PUF (a and b) and 1 wt% MWCNT surface-modified PUF (c and d).
284
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
Fig. 3. Thermogravimetric curves of the blank PUF along with 1 wt% MWCNT surfacemodified PUF.
Table 1 Oil removal, removal efficiency, and their corresponding variances (s2) for different prepared sorbents.
the oil release off the pores. Wu et al. (2014) mentioned in their report another reason behind similar finding on silica-treated PUF. They stated that an increase in surface roughness due to the attachment of MWCNTs could benefit the capacity of oil adsorption (Wu et al., 2014). Elsewhere, the researchers explained the reason based on the cell size. The increase in the foam cell size providing more space for the adsorbed oil could improve the sorption capacity of the modified foam (A Tanobe et al., 2009). The oil sorption efficiency is increased after MWCNT immobilization on PUF surface at each initial mass of oil (Table 1). In addition, the results of ANOVA show the significance of the sorbent efficiency enhancement by different amount of MWCNT immobilization. The sorbent with 3 wt% MWCNT on its surface possesses the maximum sorption efficiency. The explanation is that the hydrophobic nature of MWCNT causes the surface hydrophobicity to rise leading to an enhanced affinity towards the organic material through adsorbing less amount of water, thereby raising the sorption efficiency (Ren et al., 2011). The increase of the sorption efficiency could also be attributed to petroleum ether used during modification step which can raise the foam affinity towards organic material because of increased hydrophobicity. This explanation is supported by a study already conducted by Wu et al. (2014). The researchers treated PU sponges with gasoline and their results from contact angle analysis showed an increase in the level of hydrophobicity.
Initial amount Sorbent removal Blank PUF MWCNT surface modified PUF of oil (g) and efficiency (%) 0.5 wt% 1 wt% 2 wt% 3 wt%
3.6. Reusability of the sorbents
5
The reusability feature of the sorbent with 1 wt% MWCNT on the surface has been investigated through four regeneration cycles. The sorption removal and efficiency percentage of the regenerated sorbent are given in Table 3. As can be seen, the oil removal percentage is decreased after each regeneration cycle. As for 30 g initial mass of oil, 85.45% of the initial oil sorption capacity remains after the last cycle. These results also show the effectiveness of the surface modification method regarding this fact that the sorbents retain most of their initial capacity. The reduction in the sorption potential after each cycle mainly stems from the incomplete regeneration of the sorption sites which are permanently filled by the chemically adsorbed molecules. Another reason can be the loss of part of the MWCNTs away from the surface after each cycle. This may happen because of the adverse impact of various organic and inorganic materials present in the crude oil formulation on the bond strength among MWCNT and PUF units. Based on Table 3, one can find out that the oil sorption efficiency is increased after each cycle and the number of regeneration cycle positively improves the efficiency (using two factor ANOVA with P < 0.05 and F > Fcrit). This can be associated with the chemical regeneration by petroleum ether during which after drying the sorbent, some petroleum ether is retained on PUF surface enhancing the affinity towards oil rather than water.
removal
s2rem efficiency
s2eff 10
removal
s2rem efficiency
s2eff 20
removal
s2rem efficiency
s2eff 30
removal
s2rem efficiency
s2eff 50
removal
s2rem efficiency
s2eff 80
removal
s2rem efficiency
s2eff
83.40 2.49 27.80 1.20 91.10 1.51 39.76 0.65 89.00 1.00 71.34 5.50 67.93 3.18 75.82 0.80 40.54 1.11 90.61 1.71 25.14 1.43 99.02 0.06
91.20 2.08 39.96 1.25 93.10 1.48 52.42 0.45 93.40 1.17 76.62 3.67 73.00 3.68 92.09 0.16 42.00 1.19 96.24 0.28 26.10 1.29 97.98 0.23
95.60 2.74 45.27 0.25 97.30 0.98 52.77 0.16 99.00 1.33 79.42 2.57 82.50 4.70 93.93 0.24 48.64 1.59 97.79 0.06 27.96 2.18 99.38 0.02
93.20 90.20 2.44 2.60 46.28 50.11 1.35 0.85 95.00 94.40 1.55 1.57 57.58 60.40 0.24 0.05 98.45 94.80 1.19 1.31 84.00 85.33 1.42 1.21 78.90 77.97 4.30 4.20 94.91 95.86 0.17 0.04 44.42 43.80 1.29 1.33 98.40 99.10 0.04 0.02 27.61 27.19 0.72 2.73 100.00 100.00 0.00 0.00
oil removal percentage when compared with the blank PUF). Such an improvement of the oil removal capacity of the surface-modified PUF could be ascribed to the increase in the PUF strength already confirmed. An increase in strength decreases the extent of the mechanical deformation of the modified foam under gravitational force during foam removal from the polluted water. This reduces
3.7. Adsorption isotherm Linear regression of the crude oil sorption data was done by use of Langmuir, Freundlich, Temkin, and RedlichePeterson isotherm models. All the parameters for the analysis of these isotherm
Table 2 Analysis of variance on oil removal and oil removal efficiency of the prepared sorbents. Factor
Source of Variation
SS
df
MS
F
P-value
Fcrit
Oil removal (%)
Initial weight of oil Amount of MWCNT Initial weight of oil Amount of MWCNT
62,625.22 736.99 43,367.05 2205.80
5 4 5 4
12,525.04 184.25 8673.41 551.45
6211.35 91.37 10,560.94 671.46
4.7E-80 8.05E-25 5.85E-87 4.64E-49
2.37 2.53 2.37 2.53
Oil removal efficiency (%)
SS: sums of squares; df: degrees of freedom; MS: mean square.
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286 Table 3 The sorbent removal, removal efficiency, and their corresponding variances (s2) for the chemically regenerated 1 wt% MWCNT surface modified PUF. Initial amount of oil (g)
Sorbent removal and efficiency (%)
Regeneration cycle 1
2
3
4
5
removal
94.80 1.25 43.40 1.12 97.00 1.65 55.30 0.43 94.50 4.75 76.58 0.10 78.20 0.68 90.29 0.09 44.10 0.59 100.00 0.00 26.75 0.23 100.00 0.00
94.60 1.24 45.64 1.15 96.40 3.02 58.74 0.41 92.70 4.57 78.76 0.09 75.10 0.63 91.50 0.07 43.64 0.58 100.00 0.00 26.13 0.22 100.00 0.00
93.20 1.21 47.10 1.16 95.50 2.96 59.24 0.41 92.15 4.52 81.62 0.07 73.50 0.60 93.70 0.04 41.22 0.52 100.00 0.00 25.39 0.21 100.00 0.00
91.40 1.16 48.44 1.16 94.80 2.92 61.08 0.40 90.25 4.33 83.80 0.06 70.50 0.55 94.33 0.03 39.66 0.48 100.00 0.00 24.39 0.19 100.00 0.00
s2rem Efficiency
s2eff 10
removal
s2rem efficiency
s2eff 20
removal
s
2
rem
efficiency
s2eff 30
removal
s2rem efficiency
s2eff 50
removal
s2rem efficiency
s2eff 80
removal
s2rem efficiency
s2eff
models are reported in Tables 4 and 5 for the prepared sorbents. A comparison between the results of linear and non-linear methods by considering the values of AIC reveals that the non-linear method is more accurate and the results from this method have better agreement with the experimental data as in the all cases the value of AIC is smaller. From the both R2 and AIC quantities, it is obvious that the Langmuir is the best fitting model and has acceptable consistency with the results. Fig. 4 depicts the plot of the mentioned isotherms in the case of 1 wt% MWCNT surface modified PUF. Moreover, the values of AICc in all cases except for the blank PUF propose the ranking of the models as follow: Langmuir > Temkin > Freundlich > RedlichePeterson. Table 4 Analysis of adsorption of crude oil on the blank and surface modified PUF based on the Langmuir, Temkin, Freundlich, and RedlichePeterson equations through linear method. Isotherm
Blank PUF
MWCNT surface-modified PUF 0.5 wt %
Langmuir qm (g g1) KL (L g1) R2 AICc Freundlich KF (g g1) (L/g)n n R2 AICc RedlichePeterson g KR (L g1) aRðL g11=K RÞ R2 AICc Temkin KT (L g1) B1(g g1) R2 AICc
1 wt %
2 wt %
285
Table 5 Analysis of adsorption of crude oil on the blank and surface modified PUF based on Langmuir, Temkin, Freundlich, and RedlichePeterson equations through non-linear method. Isotherm
Blank PUF
MWCNT surface-modified PUF 0.5 wt %
Langmuir qm (g g1) KL (L g1) R2 AICc Freundlich KF (g g1) (L/g)n n R2 AICc RedlichePeterson g KR (L g1) aRðL g11=K RÞ R2 AICc Tempkin KT (L g1) B1(g g1) R2 AICc
1 wt %
2 wt %
3 wt %
21.44 1.009 0.963 9.979
22.660 0.938 0.742 22.839
23.390 4.280 0.601 26.895
22.860 2.650 0.541 26.992
23.670 1.282 0.837 20.723
11.03 5.635 0.7203 22.193
11.790 5.943 0.541 26.311
15.40 7.979 0.604 28.131
14.15 7.522 0.541 27.649
12.94 6.317 0.837 26.573
1.087 16.85 0.586 0.986 14.365
1.157 15.130 0.390 0.796 31.434
1.152 24.900 0.603 0.291 40.334
1.141 23.390 0.619 0.393 38.668
1.158 21.080 0.526 0.911 27.121
30.123 3.059 0.817 19.659
37.062 3.074 0.603 25.435
720.854 2.403 0.543 27.707
399.831 2.402 0.511 27.366
56.692 3.088 0.648 24.169
Langmuir isotherm, good fitting with the data, may suggest homogeneous and monolayer adsorption. In addition, as is seen in Fig. 4, the Langmuir model is the best model describing the saturation condition of the sorbent as in high initial concentration of oil the Langmuir plot becomes almost flat. RedlichePeterson isotherm with the power (g) near unity can satisfy the Langmuir model condition but as it is obvious from Table 5, the power (g) in all cases is more than one. The power larger than one can be explained by the other effective factors rather than the physical and chemical sorption conditions like the capillary action. The reported data in Table 5 reveal that in the case of blank PUF the RedlichePeterson model fits the equilibrium data better than Temkin. This may indicate that the addition of MWCNT to the surface affects the sorption mechanism and weakens the capillary action during the sorption process. In the case of Freundlich model, according to the reported values of n in the Table 5, the power (1/n) with the value less than one implies physical sorption phenomena and as its value gets closer to zero may show more surface heterogeneity which verifies the presented explanation about the Langmuir isotherm (Foo and Hameed, 2010).
3 wt %
20.445 1.421 0.999 9.969
21.413 0.905 0.997 25.441
23.202 2.535 0.998 28.983
22.321 2.435 0.999 46.374
22.075 1.690 0.998 37.549
13.657 2.195 0.800 29.376
9.210 3.913 0.512 26.411
13.317 5.665 0.462 45.657
12.060 5.367 0.443 53.144
10.329 4.076 0.572 74.878
1.060 14.240 0.639 0.999 20.353
1.007 19.011 0.878 0.997 33.431
1.192 24.570 0.498 0.999 40.556
1.091 23.250 0.720 0.999 38.794
1.038 25.570 0.992 0.999 30.057
30.115 3.0593 0.817 19.659
37.059 3.0743 0.603 25.435
720.917 2.403 0.543 27.706
399.850 2.4019 0.511 27.366
56.688 3.088 0.647 24.168
Fig. 4. Plot of the Langmuir, Temkin, Freundlich, and RedlichePeterson isotherm models for 1 wt% MWCNT surface-modified PUF obtained through non-linear regression method.
286
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
4. Conclusion It has been concluded from the findings of this research that surface modification of PUF by MWCNT is very favorable for increasing oil removal ability. The tensile elongation test has shown a correlation of the mechanical strength and the amount of MWCNT immobilized onto the PUF surface. The TGA results have revealed an increase in the level of thermal resistance. The results from sorption experiments have confirmed that the sorbent with 1 wt% MWCNT on its surface gives the maximum sorption performance (by adsorbing 24.75 g/g, about 21.44% sorption enhancement in comparison with the blank PUF). It has also been established that the sorption efficiency is increased after grafting MWCNT onto the PUF surface, since in all cases; less water is adsorbed in comparison with the blank PUF. The results on the reusability of the sorbent with 1 wt% MWCNTs on its surface after four cycles of regeneration has shown that 85.45% of the oil sorption capacity remains, however, the oil removal percentage is decreased and the sorption efficiency is increased after each regeneration cycle. Based on the values of R2 and Akaike's Information Criterion (AIC), Langmuir model showed better agreement with the results in comparison with the other fitted isotherms in both case of linear and non-linear methods. References ndez, J., P Aguilera, F., Me asaro, E., Laffon, B., 2010. Review on the effects of exposure to spilled oils on human health. J. Appl. Toxicol. 30, 291e301. Al-Majed, A.A., Adebayo, A.R., Hossain, M.E., 2012. A sustainable approach to controlling oil spills. J. Environ. Manage 113, 213e217. Alves, P., Coelho, J.F.J., Haack, J., Rota, A., Bruinink, A., Gil, M.H., 2009. Surface modification and characterization of thermoplastic polyurethane. Eur. Polym. J. 45, 1412e1419. Banerjee, S.S., Joshi, M.V., Jayaram, R.V., 2006. Treatment of oil spill by sorption technique using fatty acid grafted sawdust. Chemosphere 64, 1026e1031. Bastani, D., Safekordi, A.A., Alihosseini, A., Taghikhani, V., 2006. Study of oil sorption by expanded perlite at 298.15 K. Sep. Purif. Technol. 52, 295e300. Burnham, K.P., Anderson, D.R., 2002. Model selection and multimodel inference: a practical information-theoretic approach. Springer Sci. Bus. Media 149e205. Chin, C.-J.M., Shih, M.-W., Tsai, H.-J., 2010. Adsorption of nonpolar benzene derivatives on single-walled carbon nanotubes. Appl. Surf. Sci. 256, 6035e6039. Deng, S., 2006. Sorbent technology. Encycl. Chem. Process. 2825e2845. Do, D.D., 1998. Adsorption Analysis. World Scientific. Duong, H.T., Burford, R.P., 2006. Effect of foam density, oil viscosity, and temperature on oil sorption behavior of polyurethane. J. Appl. Polym. Sci. 99, 360e367. El-Khaiary, M.I., Malash, G.F., 2011. Common data analysis errors in batch adsorption studies. Hydrometallurgy 105, 314e320. Fan, Z., Yan, J., Ning, G., Wei, T., Qian, W., Zhang, S., Zheng, C., Zhang, Q., Wei, F., 2010. Oil sorption and recovery by using vertically aligned carbon nanotubes. Carbon 48, 4197e4200. Foo, K., Hameed, B., 2010. Insights into the modeling of adsorption isotherm systems. Chem. Eng. J. 156, 2e10. Gui, X., Li, H., Wang, K., Wei, J., Jia, Y., Li, Z., Fan, L., Cao, A., Zhu, H., Wu, D., 2011. Recyclable carbon nanotube sponges for oil absorption. Acta Mater. 59,
4798e4804. Hamdaoui, O., Naffrechoux, E., 2007. Modeling of adsorption isotherms of phenol and chlorophenols onto granular activated carbon: part II. Models with more than two parameters. J. Hazard. Mater. 147, 401e411. Harikrishnan, G., Patro, T.U., Khakhar, D., 2006. Polyurethane foam-clay nanocomposites: nanoclays as cell openers. Ind. Eng. Chem. Res. 45, 7126e7134. Husseien, M., Amer, A., 2009. Availability of barley straw application on oil spill clean up. Int. J. Environ. Sci. Technol. 6, 123e130. Joseph, L., Heo, J., Park, Y.-G., Flora, J.R.V., Yoon, Y., 2011. Adsorption of bisphenol A and 17a-ethinyl estradiol on single walled carbon nanotubes from seawater and brackish water. Desalination 281, 68e74. Li, H., Liu, L., Yang, F., 2012. Hydrophobic modification of polyurethane foam for oil spill cleanup. Mar. Pollut. Bull. 64, 1648e1653. Li, H., Liu, L., Yang, F., 2013a. Oleophilic polyurethane foams for oil spill cleanup. Procedia Environ. Sci. 18, 528e533. Li, Y., Du, Q., Liu, T., Peng, X., Wang, J., Sun, J., Wang, Y., Wu, S., Wang, Z., Xia, Y., Xia, L., 2013b. Comparative study of methylene blue dye adsorption onto activated carbon, graphene oxide, and carbon nanotubes. Chem. Eng. Res. Des. 91, 361e368. Lim, T.-T., Huang, X., 2007. Evaluation of hydrophobicity/oleophilicity of kapok and its performance in oily water filtration: comparison of raw and solvent-treated fibers. Ind. Crop. Prod. 26, 125e134. Pacholczyk, A., Terzyk, A.P., Wisniewski, M., Gauden, P.A., Wesołowski, R.P., Furmaniak, S., Szczes, A., Chibowski, E., Kruszka, B., 2011. Phenol adsorption on closed carbon nanotubes. J. Colloid Interf. Sci. 361, 288e292. Radhika, M., Palanivelu, K., 2006. Adsorptive removal of chlorophenols from aqueous solution by low cost adsorbentdKinetics and isotherm analysis. J. Hazard. Mater. 138, 116e124. Ren, X., Chen, C., Nagatsu, M., Wang, X., 2011. Carbon nanotubes as adsorbents in environmental pollution management: a review. Chem. Eng. J. 170, 395e410. Sahoo, N.G., Jung, Y.C., So, H.H., Cho, J.W., 2007. Synthesis of polyurethane nanocomposites of functionalized carbon nanotubes by in-situ polymerization methods. Korean Phys. Soc. 51, S1eS6. Subramanyam, B., Ashutosh, D., 2011. Adsorption isotherm modeling of phenol onto natural soilseApplicability of various isotherm models. Int. J. Environ. Res. 6, 265e276. A Tanobe, V., Sydenstricker, T., Amico, S., Vargas, J., Zawadzki, S., 2009. Evaluation of flexible post consumed polyurethane foams modified by polystyrene grafting as sorbent material for oil spills. J. Appl. Polym. Sci. 111, 1842e1849. Vasanth Kumar, K., Sivanesan, S., 2006. Equilibrium data, isotherm parameters and process design for partial and complete isotherm of methylene blue onto activated carbon. J. Hazard. Mater. 134, 237e244. Wang, D., Silbaugh, T., Pfeffer, R., Lin, Y.S., 2010. Removal of emulsified oil from water by inverse fluidization of hydrophobic aerogels. Powder Technol. 203, 298e309. Wang, J., Zheng, Y., Wang, A., 2012. Superhydrophobic kapok fiber oil-absorbent: preparation and high oil absorbency. Chem. Eng. J. 213, 1e7. Wang, J., Wang, F., Yao, J., Wang, R., Yuan, H., Masakorala, K., Choi, M.M.F., 2013. Adsorption and desorption of dimethyl phthalate on carbon nanotubes in aqueous copper(II) solution. Colloids Surf. A 417, 47e56. Wu, F.-C., Liu, B.-L., Wu, K.-T., Tseng, R.-L., 2010. A new linear form analysis of RedlichePeterson isotherm equation for the adsorptions of dyes. Chem. Eng. J. 162, 21e27. Wu, D., Fang, L., Qin, Y., Wu, W., Mao, C., Zhu, H., 2014. Oil sorbents with high sorption capacity, oil/water selectivity and reusability for oil spill cleanup. Mar. Pollut. Bull. 84, 263e267. Zhao, D., Zhang, W., Chen, C., Wang, X., 2013. Adsorption of methyl Orange dye onto multiwalled carbon nanotubes. Procedia Environ. Sci. 18, 890e895. Zheng, Y.-P., Wang, H.-N., Kang, F.-Y., Wang, L.-N., Inagaki, M., 2004. Sorption capacity of exfoliated graphite for oils-sorption in and among worm-like particles. Carbon 42, 2603e2607.
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
Enhancing oil removal from water by immobilizing multi-wall carbon nanotubes on the surface of polyurethane foam Alireza Keshavarz a, Hamid Zilouei a, *, Amir Abdolmaleki b, Ahmad Asadinezhad a a b
Department of Chemical Engineering, Isfahan University of Technology, Isfahan 84156-83111, Iran Department of Chemistry, Isfahan University of Technology, Isfahan 84156-83111, Iran
a r t i c l e i n f o
a b s t r a c t
Article history: Received 23 February 2015 Received in revised form 16 April 2015 Accepted 17 April 2015 Available online
A surface modification method was carried out to enhance the light crude oil sorption capacity of polyurethane foam (PUF) through immobilization of multi-walled carbon nanotube (MWCNT) on the foam surface at various concentrations. The developed sorbent was characterized using scanning electron microscopy, Fourier transform infrared spectroscopy, thermogravimetric analysis, and tensile elongation test. The results obtained from thermogravimetric and tensile elongation tests showed the improvement of thermal and mechanical resistance of surface-modified foam. The experimental data also revealed that the immobilization of MWCNT on PUF surface enhanced the sorption capacity of light crude oil and reduced water sorption. The highest oil removal capacity was obtained for 1 wt% MWCNT on PUF surface which was 21.44% enhancement in light crude oil sorption compared to the blank PUF. The reusability of surface modified PUF was determined through four cycles of chemical regeneration using petroleum ether. The adsorption of light crude oil with 30 g initial mass showed that 85.45% of the initial oil sorption capacity of this modified sorbent was remained after four regeneration cycles. Equilibrium isotherms for adsorption of oil were analyzed by the Freundlich, Langmuir, Temkin, and Redlich ePeterson models through linear and non-linear regression methods. Results of equilibrium revealed that Langmuir isotherm is the best fitting model and non-linear method is a more accurate way to predict the parameters involved in the isotherms. The overall findings suggested the promising potentials of the developed sorbent in order to be efficiently used in large-scale oil spill cleanup. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Oil removal Polyurethane foam Multi-walled carbon nanotube Surface modification
1. Introduction In recent decades, oil demand has been escalated because of the rapid industrial development and urban area expansion (Wang et al., 2012). The global oil requirement is mainly transported through seas and oceans which pose a risk of oil spillage and leakage to the marine ecosystem. The spilled oil adversely affects the flora and fauna species such as birds and fish (Aguilera et al., 2010). The common methods employed to solve the oil spill problems include in-situ burning, mechanical methods (using booms, vacuum units, and skimmers), chemical methods (using chemical dispersants), and sorbents (Al-Majed et al., 2012). Each method with its own advantages and disadvantages, is normally selected depending on the type of oil, spill scale, the location, and the climate (Li et al., 2013a,b).
* Corresponding author. E-mail address: [email protected] (H. Zilouei). http://dx.doi.org/10.1016/j.jenvman.2015.04.030 0301-4797/© 2015 Elsevier Ltd. All rights reserved.
The advanced removal and recovery of oil by different sorbents can be regarded as one of the most effective techniques for oil spill cleanup (Banerjee et al., 2006). The criteria to select an ideal sorbent include high sorption capacity, low cost, high selectivity, and reusability (Deng, 2006). Oil sorbents may be divided into three basic types including inorganic mineral sorbents, such as aerogel (Wang et al., 2010), exfoliated graphite (Zheng et al., 2004), and expanded perlite (Bastani et al., 2006), organic natural sorbents, such askapok (Lim and Huang, 2007), saw dust (Husseien and Amer, 2009), and straw (Banerjee et al., 2006), and organic synthetic sorbents, such as polyurethane and polypropylene foams and carbon nanotubes (Gui et al., 2011). Typical advantages of inorganic mineral sorbents are inexpensiveness and easy availability. However, they tend to sink in water. On the other hand, organic natural sorbents are fascinating thanks to their good adsorption capacity, biodegradability and low cost, although they are dusty and have to be dried before using (Duong and Burford, 2006). Polyurethane foams (PUF) are cheap and already commercialized, with low density and good thermal properties. However, poor
280
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
hydrophobic property of PUF reduces its oil sorption efficiency since water is also adsorbed in large extent along with oil. In recent years, the surface modification with an aim to improve the hydrophobicity of PUF has attracted vast interest. Li et al. (2012) improved the oleophilic/hydrophobic properties of PUF for oil sorption by grafting the oleophilic monomer of laurylmethacrylate (LMA) on PUF surface. Their results showed a decrease by 24e50% in water sorption and an increase by 44% in the diesel sorption. Wu et al. (2014) also treated PUF with silica sol and gasoline, and their results showed an improvement of the sorption capacity and oil/ water selectivity after adsorption of 100 g of motor oil versus 0.1 g of water. Carbon nanotubes (CNTs) with two different versions, multiwalled carbon nanotube (MWCNT) and single-walled carbon nanotube (SWCNT), have received great attention in environmental pollution remediation because of their low mass density, high porosity and hollow structure, high specific surface area, strong interaction between CNTs and pollutant molecules, and their hydrophobic propensity (Ren et al., 2011). The relevant studies have indicated that CNTs are effective adsorbents for organic materials such as dyes (Li et al., 2013b; Zhao et al., 2013), benzene (Chin et al., 2010), phenol (Pacholczyk et al., 2011), dimethyl phthalate (Wang et al., 2013), and bisphenol A (Joseph et al., 2011). Fan et al. (2010) used CNTs as oil adsorbent and evaluated its recycling performance. Their results showed 69 g/g of oil sorption and good recyclability due to the unique mechanical strength of CNTs. Gui et al. (2011) also explored the recyclability of CNTs sponges in terms of the oil adsorption capacity. The results showed 112 g/g of sorption for diesel oil and CNTs could attain around 30% of the initial sorption capacity after 10 cycles of adsorption. In this work, carboxylic MWCNT is intended to be immobilized on the surface of PUF in order to enhance the oil sorption capacity and the sorption efficiency of the PUF. Carboxyl-functionalized MWCNT is immobilized on the PUF surface through the reaction between the OH groups existing in PUF structure and carboxyl functionalities on the surface of MWCNT and methylene diphenyl diisocyanate (MDI). Then, the synthesized sorbent is studied in terms of oil sorption capacity, oil sorption efficiency, and reusability. 2. Materials and methods 2.1. Materials Carboxylic multi-walled carbon nanotube was purchased from Neutrino Company (Iran). 1,1-dichoro-1-fluoroethane (HCFC 141b) was purchased from Lin'an E-COOL Refrigeration Equipment CO (China). NIXOL AM-313 polyether polyol was purchased from KPX chemical Co (South Korea) and methylene diphenyl diisocyanate (MDI) was obtained from Daeyang International Co (South Korea). Light crude oil was obtained from Isfahan refinery feed stream (Iran), industrial grade Xylene was purchased from Isfahan petrochemical complex (Iran) and petroleum ether with the boiling range of 30e60 C was supplied by Pars Chemie Co (Iran). General data obtained from the National Iranian Oil Company for characteristics of light crude oil was as below: specific gravity (at 15.56 C) 0.8597, API 33.09, Sulphur content 1.33 wt%, Mercaptane content 67 ppm, Nitrogen content 0.12 wt%, Asphaltenes 1.0 wt%, Wax content 5.1 wt%, Carbon residue conradson 4.00 wt%, Base sediment and water <0.05 vol%, Acidity 0.09 mg KOH/g.
as the chemical blowing agent and 1.5 g HCFC 141b as the physical blowing agent. Then, the resulting mixture was mixed with 4 g MDI by means of a mixer at 1000 rpm. At the end, the mixture was left for 1 h in order for the reaction to complete. 2.3. Surface modification of polyurethane foam The surface modification of PUF was carried out by allowing the reaction among MDI functionalities, carboxylic OH groups existing in MWCNT structure, and the unreacted OH units of polyol present in the foam structure. In this study, the samples with 0.5 wt%, 1 wt%, 2 wt%, and 3 wt% of MWCNT on PUF surface were prepared. To this end, first, 300 ml petroleum ether was poured inside a glass beaker. Then an amount equal to 10 times the final amount of MWCNT on the surface was added to the beaker and stirred for 5 min at 1000 rpm to obtain a homogeneous MWCNT suspension. Amount of 1 g PUF which was already cut into 1 1 1 cm3 cubes was added to the reaction media and the whole content was stirred for 15 min to allow the MWCNT moieties adsorb onto foam surface. Then, while mixing, MDI as much as 2 times the amount of the MWCNT was added to the reaction beaker gradually and stirred for another 40 min period. The foam cubes were taken out and thoroughly dried in an oven at 100 C for 10 min and weighed with high precision. Then, the cubes were washed by petroleum ether to remove the physically adsorbed loose particles trapped in the pores followed by complete drying in the oven at 100 C for 30 min. Finally, they were precisely weighed in order to estimate the amount of the adsorbed MWCNT on foam surface. 2.4. Sorbent capacity, removal and efficiency To study the sorption capacity of the prepared sorbents, light crude oil was used as water pollutant. The sorption experiments were performed based on the Standard Test Method for Sorbent Performance of Adsorbents (ASTMF726-99). Sorption experiments were performed in water-oil system. To do this, 250 ml fresh water was poured in a 600 ml beaker and then crude oil was added to water so that a layer with the thickness 2e4 mm of oil was formed on the water surface. Then, 1.0 g the adsorbent was weighed and added to the beaker. The container was then subjected to shaking on a shaker at the frequency 150 rpm for 15 min. The content of the beaker was allowed to settle for 2 min immediately after shaking. The sorbent cubes were withdrawn by a forceps and allowed to drain for 30 ± 3 s. They were carefully weighed. Then, the cubes were fully washed by the industrial grade Xylene to extract the adsorbed oil and water. The extracted solution was finally used to determine the water content of the oil based on the Standard Test Method for Water in Crude Oil by Distillation (ASTMD4006-81). To this end, the extracted solution was poured into a distillation apparatus. The distillation was performed until the volume of the water collected by the apparatus trap remained constant and no water was visible any longer in the apparatus. When the distillation operation was complete, the trap content was allowed to cool down to 20 C. At the end, the water collected by the trap was precisely weighed and the result was reported as the mass of the adsorbed water. If the value of any run (g/g) deviated by more than 15% from the mean of the three runs, then the samples would be rejected. The oil sorption capacity of the sorbents was calculated by Eq. (1):
ms mw m0 m0
2.2. Synthesis of polyurethane foam
M¼
To synthesize the polyurethane foam at ambient temperature (22 ± 3 C), 10 g polyol was fully mixed with 0.1 g deionized water
where M is the oil sorption capacity (g/g), ms is mass of saturated sorbent (mass of water plus that of oil and sorbent), mw is the mass
(1)
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
of adsorbed water (g), and m0 is initial dry mass of sorbent (g). Oil removal percent was determined by Eq. (2):
Pr ¼
mr 100 mi
(2)
where Pr is oil removal percent, mr is mass of removed oil (g), and mi is initial mass of oil (g). Oil sorption efficiency of the sorbent was defined as the ratio of adsorbed oil mass to total mass of adsorbed oil and water. Thus, Eq. (3) is:
E¼
mr 100 mo þ mw
(3)
where E is oil sorption efficiency, mr is mass of adsorbed oil (g) and mw is mass of adsorbed water. The data collected in the experiments were subjected to the twofactor analysis of variance (ANOVA) to determine the relationship between the adsorption capacity and amount of MWCNT on the surface of PUF and initial weight of light crude oil. Significance of the differences was defined as p < 0.05 in all the cases. The statistical processing was performed using Microsoft Excel software. 2.5. Sorbent regeneration Reusability property of the sorbent with 1 wt% multi-walled carbon nanotube was investigated by the chemical recovery of the sorbent cubes through four cycles of recovery and reuse. In each cycle, the sorbent cubes were washed three times using petroleum ether until no color change of the consumed petroleum ether took place. Finally, the regenerated sorbent cubes were dried inside an oven at 100 C for 30 min. 2.6. Sorbent characterization The morphology of the control sample (blank PUF) as well as the surface-modified PUF samples was studied by scanning electron microscopy (SEM) (TESCAN e VEGA3, SBU e Easy Probe, Czech Republic). FT-IR spectra of the samples were recorded on an FT-IR spectrometer (JASCO, Model 6300, Japan). This technique was in principle utilized to examine the surface chemical groups and to prove the occurrence of desirable surface modification reactions. To study the effect of surface modification on the thermal resistance of PUF, thermogravimetric analysis (TGA) was performed via a thermogravimetric analyzer (STA6000, PerkinElmer, USA). Tensile elongation test was also performed to exploreany changes of the mechanical strength of the developed sorbents by Zwick Universal Testing Machinee1446-60, USA. 2.7. Adsorption isotherms Both linear and non-linear equations of the Langmuir, Freundlich, and RedlichePeterson isotherm models were analyzed to identify which type of equation has better agreement with the experimental data of oil removal from water using blank PUF and MWCNT surface modified PUF. Linear regression analysis was done through least square method to determine the isotherm models parameters. However, linear method leads to uncontrollable parameter estimation error, thus it may be better to directly estimate the model parameters from the non-linear equation (Subramanyam and Ashutosh, 2011). In all cases, non-linear regression was done by use of GraphPad Prism 6 software. 2.7.1. Langmuir isotherm Langmuir model is based on a kinetic view point that propose a coherent adsorption theory onto a flat surface (Do, 1998). The
281
assumptions of the Langmuir model are: (i) surface is homogeneous with constant adsorption energy over the all sites, (ii) adsorption on surface is localized, (iii) monolayer adsorption; that is each site can accommodate only one molecule or atom. Langmuir isotherm is expressed as follow:
qe ¼
qm KL Ce 1 þ KL Ce
(4)
A linear expression of the Langmuir isotherm may be represented by:
Ce Ce 1 ¼ þ qe qm qm KL
(5)
where qe is the equilibrium solid-phase concentration (g g1), Ce is the equilibrium liquid-phase concentration (g L1), qm (g g1) is the maximal adsorption capacity, and KL (L g1) is the Langmuir model constant related to the adsorption free energy. The slope and intercept of a plot of Ce/qe vs. Ce enables the determination of qe and KL, respectively. 2.7.2. Freundlich isotherm Freundlich isotherm model is not restricted to monolayer adsorption and describes the non-ideal and reversible adsorption (Vasanth Kumar and Sivanesan, 2006). This model considers heterogeneous surfaces in the sense that the adsorption energy is distributed (Radhika and Palanivelu, 2006). The Freundlich model can be expressed as: 1=n
qe ¼ KF Ce
(6)
where KF is the constant indicative of the relative sorption capacity of the sorbent (g g1) (L/g)n, and 1/n is the heterogeneity factor. The linear form of Freundlich isotherm is expressed as:
log qe ¼
1 log Ce þ log KF n
(7)
which will have a straight line with a slope of 1/n and an intercept of log(KF) when log(qe) is plotted against log(Ce). 2.7.3. RedlichePeterson isotherm RedlichePeterson isotherm is a three-parameter isotherm model which includes the features of the Langmuir and Freundlich isotherm models (Hamdaoui and Naffrechoux, 2007). The RedlichePeterson isotherm can be described as follows:
qe ¼
KR Ce g 1 þ a R Ce
(8)
where KR (L g1) and aRðL g11=K RÞ are RedlichePeterson equation constants and g is the exponent. The exponent, g, is usually between zero and one (Wu et al., 2010). When g ¼ 1, this isotherm is the same as Langmuir isotherm and when g ¼ 0, the model is the same as Henry's law equation. The linear form of RedlichePeterson isotherm may be written as:
Ce 1 a ¼ þ R Cg qe KR KR e
(9)
To fit the linear equation of RedlichePeterson model, a trial and error procedure was performed.The value of g was assumed, then by substituting the assumed value in the linear equation the regular least square regression was performed and the value of coefficient of determination (R2) was determined. The procedure continued until the difference between the values of the R2obtained from two respective iteration was less than 104.
282
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
2.7.4. Temkin isotherm Temkin isotherm assumes the linear reduction of adsorption heat rather than logarithmic with coverage. The isotherm contains a factor that considers the adsorbenteadsorbate interactions. The Temkin isotherm is given as:
qe ¼
RT lnðKT Ce Þ b
(10)
which can be linearized as:
qe ¼ B1 ln KT þ B1 ln Ce B1 ¼
RT b
(11) (12)
where B1 is related to the heat of adsorption and KT (L g1) is the equilibrium binding constant. The isotherm constants B1 and KT can be determined from the slope and the intercept of a plot of qe versus lnCe, respectively. 2.7.5. Comparison of the models The coefficient of determination shows the fit between the experimental data and isotherm model (Subramanyam and Ashutosh, 2011) and represents the variance about the mean (Foo and Hameed, 2010). However, the value of R2 has no consideration for the degrees of freedom. Other obvious issues with R2 are sensitivity to extreme data points and the potential of being large in the case of models with more parameters (El-Khaiary and Malash, 2011). Therefore, it seems necessary to compare the models with different degrees of freedom based on an index that considers the number of parameters. Akaike's Information Criterion (AIC) (Burnham and Anderson, 2002) is a statistical method that provides an index to compare the models. It determines the more likely correct model based on information theory and maximum likelihood.
2Np Np þ 1 SSE þ 2Np þ AIC ¼ N ln N N Np 1
(13)
where SSE is the sum of squared deviations of the points from the regression curve, N is the number of data points, and Np is the number of parameters in the model. Having smaller AIC shows the more likely better model. 3. Results and discussion 3.1. FT-IR The results of FT-IR spectrometry are shown in Fig. 1. The presence of carboxylic acid groups on MWCNT surface is confirmed through the following identified bands; the strong peak at 1705 cm1 attributed to C]O, a very broad band at 3400-2400 cm-1 characteristic to OeH stretching vibration, and the peak at 1537 cm-1 corresponding to-COO vibration. As for the blank PUF, the peaks at 1070 cm-1 and 3375 cm-1, the strong peak at 1730 cm-1, and the peak at 1535 cm-1are assigned to etheric CeO, NeH, C]O and aromatic rings stretching vibrations, respectively. In the case of surface modified PUF, except for C]O and NeH groups, the observed peaks are almost similar to those of the blank PUF in terms of position and intensity. According to the pertinent spectra, C]O peak at 1126 cm-1 and NeH peak at 3357 cm-1 become weak and broad after surface modification. These changes imply the occurrence of various types of secondary interactions such as hydrogen bonding and also validate the presence of
MWCNT on the foam surface. The strong peak at 1132 cm-1 is associated with the etheric CeO stretching vibration. The peak at 1546 cm-1 attributed to C]C vibration band of aromatic rings. 3.2. SEM The SEM images of the blank PUF along with surface-modified PUFare shown in Fig. 2. Presence of MWCNT on the modified PUF surface is apparent from comparing the respective images (Fig. 2,b and d). Distinctively brighter spots in Fig. 2,d, which are due to the higher electron density, are possibly due to the MWCNTs moieties surrounded by MDI and then reacted with the foam surface. A slight difference between the morphology of the blank PUF and that of the surface-modified PUF is discernible from the respective images (Fig. 2,a and c). It is observed that the cells become slightly wider after surface modification. Therefore, the overall volume of the cells and the ratio of open to close cells are increased. Similar observation has also reported by Harikrishnan et al. (2006), however, in the case of polyurethane foam-clay nanocomposite, they used nanoclays as the cell openers in the structure of the polyurethane foam. The most probable reason behind the increase in cell size can be related with the modification method where the blank cubes are immersed inside petroleum ether. This makes the cubes swell and provide enough space for MWCNT moieties to penetrate into the pores, react with the surface molecules, and take up the pores. Due to this process, even after complete drying, the foam shrinkage is not allowed and the initial pores volume is not restored. In addition to the above observations, one can notice in the SEM images (Fig. 2b and d) the topographical difference between the surface of the blank and that of the modified PUF. The former has a smooth feature while the latter is rough. This is originated from the presence of the chemically attached MWCNTs. 3.3. Tensile elongation Tensile elongation test performed on sorbent samples with 0, 0.5, 1, and 2 wt% MWCNT on their surfaces and the results in terms of Fmax (N) and Elongation (mm) are reported as follow: (4.99 N, 16.08 mm), (8.23 N, 14.55 mm), (9.91 N, 13.27 mm), and (10.58 N, 12.31 mm), respectively. As can be inferred from data, for all samples, the maximal breaking force is increased and the breaking length is decreased, suggesting an increase in mechanical strength. Such an increase in strength can be explained based on the occurrence of strong interactions between MWCNTs and PU chains (like H-bonding) which hinder the mobility of polymer chains and enhance their stiffness under mechanical tensions. Moreover, the addition of MDI to PUF structure increases the extent of the hard segments resulting in an increase in stiffness and ultimately strength of the material. 3.4. TGA The results from TGA performed on the blank PUF and the modified sorbent with 1 wt% MWCNT are illustrated in Fig. 3. The presented curves show a significant difference between the final mass of the blank sample and that of the surface-modified sample. The lower mass loss of the surface-modified PUF, in comparison with that of the blank PUF, implies an improvement of the thermal resistance of the surface-modified sample. This can be attributed to the excellent inherent thermal resistance of MWCNT additive which increases the thermal performance of the PUF after surface modification (Sahoo et al., 2007). Similar degradation trend of the blank and surface-modified PUFs indicates that the bulk structure of the foam is not adversely affected after surface treatment (Alves et al., 2009).
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
283
Fig. 1. FTIR spectra of carboxylic MWCNT, blank PUF, and MWCNT surface-modified PUF.
3.5. Sorbent capacity and efficiency The oil removal percentage and the removal efficiency of the prepared sorbents together with calculated results of two-factor ANOVA are presented in Table 1 and Table 2, respectively. The presence of the chemically immobilized MWCNT units on PUF surface contributes positively to the oil sorbent capacity (Table 1).
Moreover, the results of ANOVA on the oil removal percentage, presented in Table 2, shows that MWCNT significantly improves the sorbent performance of polyurethane foam verified by the low value of P (P < 0.05) and the F > Fcrit (91.37 > 2.53). It is clear that the sorbent with 1 wt% MWCNT on its surface gives the best performance where the maximum sorption occurs at the oil initial mass of 30 g by adsorbing 24.75 g/g (corresponding to 21.44% increase in
Fig. 2. SEM images of blank PUF (a and b) and 1 wt% MWCNT surface-modified PUF (c and d).
284
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
Fig. 3. Thermogravimetric curves of the blank PUF along with 1 wt% MWCNT surfacemodified PUF.
Table 1 Oil removal, removal efficiency, and their corresponding variances (s2) for different prepared sorbents.
the oil release off the pores. Wu et al. (2014) mentioned in their report another reason behind similar finding on silica-treated PUF. They stated that an increase in surface roughness due to the attachment of MWCNTs could benefit the capacity of oil adsorption (Wu et al., 2014). Elsewhere, the researchers explained the reason based on the cell size. The increase in the foam cell size providing more space for the adsorbed oil could improve the sorption capacity of the modified foam (A Tanobe et al., 2009). The oil sorption efficiency is increased after MWCNT immobilization on PUF surface at each initial mass of oil (Table 1). In addition, the results of ANOVA show the significance of the sorbent efficiency enhancement by different amount of MWCNT immobilization. The sorbent with 3 wt% MWCNT on its surface possesses the maximum sorption efficiency. The explanation is that the hydrophobic nature of MWCNT causes the surface hydrophobicity to rise leading to an enhanced affinity towards the organic material through adsorbing less amount of water, thereby raising the sorption efficiency (Ren et al., 2011). The increase of the sorption efficiency could also be attributed to petroleum ether used during modification step which can raise the foam affinity towards organic material because of increased hydrophobicity. This explanation is supported by a study already conducted by Wu et al. (2014). The researchers treated PU sponges with gasoline and their results from contact angle analysis showed an increase in the level of hydrophobicity.
Initial amount Sorbent removal Blank PUF MWCNT surface modified PUF of oil (g) and efficiency (%) 0.5 wt% 1 wt% 2 wt% 3 wt%
3.6. Reusability of the sorbents
5
The reusability feature of the sorbent with 1 wt% MWCNT on the surface has been investigated through four regeneration cycles. The sorption removal and efficiency percentage of the regenerated sorbent are given in Table 3. As can be seen, the oil removal percentage is decreased after each regeneration cycle. As for 30 g initial mass of oil, 85.45% of the initial oil sorption capacity remains after the last cycle. These results also show the effectiveness of the surface modification method regarding this fact that the sorbents retain most of their initial capacity. The reduction in the sorption potential after each cycle mainly stems from the incomplete regeneration of the sorption sites which are permanently filled by the chemically adsorbed molecules. Another reason can be the loss of part of the MWCNTs away from the surface after each cycle. This may happen because of the adverse impact of various organic and inorganic materials present in the crude oil formulation on the bond strength among MWCNT and PUF units. Based on Table 3, one can find out that the oil sorption efficiency is increased after each cycle and the number of regeneration cycle positively improves the efficiency (using two factor ANOVA with P < 0.05 and F > Fcrit). This can be associated with the chemical regeneration by petroleum ether during which after drying the sorbent, some petroleum ether is retained on PUF surface enhancing the affinity towards oil rather than water.
removal
s2rem efficiency
s2eff 10
removal
s2rem efficiency
s2eff 20
removal
s2rem efficiency
s2eff 30
removal
s2rem efficiency
s2eff 50
removal
s2rem efficiency
s2eff 80
removal
s2rem efficiency
s2eff
83.40 2.49 27.80 1.20 91.10 1.51 39.76 0.65 89.00 1.00 71.34 5.50 67.93 3.18 75.82 0.80 40.54 1.11 90.61 1.71 25.14 1.43 99.02 0.06
91.20 2.08 39.96 1.25 93.10 1.48 52.42 0.45 93.40 1.17 76.62 3.67 73.00 3.68 92.09 0.16 42.00 1.19 96.24 0.28 26.10 1.29 97.98 0.23
95.60 2.74 45.27 0.25 97.30 0.98 52.77 0.16 99.00 1.33 79.42 2.57 82.50 4.70 93.93 0.24 48.64 1.59 97.79 0.06 27.96 2.18 99.38 0.02
93.20 90.20 2.44 2.60 46.28 50.11 1.35 0.85 95.00 94.40 1.55 1.57 57.58 60.40 0.24 0.05 98.45 94.80 1.19 1.31 84.00 85.33 1.42 1.21 78.90 77.97 4.30 4.20 94.91 95.86 0.17 0.04 44.42 43.80 1.29 1.33 98.40 99.10 0.04 0.02 27.61 27.19 0.72 2.73 100.00 100.00 0.00 0.00
oil removal percentage when compared with the blank PUF). Such an improvement of the oil removal capacity of the surface-modified PUF could be ascribed to the increase in the PUF strength already confirmed. An increase in strength decreases the extent of the mechanical deformation of the modified foam under gravitational force during foam removal from the polluted water. This reduces
3.7. Adsorption isotherm Linear regression of the crude oil sorption data was done by use of Langmuir, Freundlich, Temkin, and RedlichePeterson isotherm models. All the parameters for the analysis of these isotherm
Table 2 Analysis of variance on oil removal and oil removal efficiency of the prepared sorbents. Factor
Source of Variation
SS
df
MS
F
P-value
Fcrit
Oil removal (%)
Initial weight of oil Amount of MWCNT Initial weight of oil Amount of MWCNT
62,625.22 736.99 43,367.05 2205.80
5 4 5 4
12,525.04 184.25 8673.41 551.45
6211.35 91.37 10,560.94 671.46
4.7E-80 8.05E-25 5.85E-87 4.64E-49
2.37 2.53 2.37 2.53
Oil removal efficiency (%)
SS: sums of squares; df: degrees of freedom; MS: mean square.
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286 Table 3 The sorbent removal, removal efficiency, and their corresponding variances (s2) for the chemically regenerated 1 wt% MWCNT surface modified PUF. Initial amount of oil (g)
Sorbent removal and efficiency (%)
Regeneration cycle 1
2
3
4
5
removal
94.80 1.25 43.40 1.12 97.00 1.65 55.30 0.43 94.50 4.75 76.58 0.10 78.20 0.68 90.29 0.09 44.10 0.59 100.00 0.00 26.75 0.23 100.00 0.00
94.60 1.24 45.64 1.15 96.40 3.02 58.74 0.41 92.70 4.57 78.76 0.09 75.10 0.63 91.50 0.07 43.64 0.58 100.00 0.00 26.13 0.22 100.00 0.00
93.20 1.21 47.10 1.16 95.50 2.96 59.24 0.41 92.15 4.52 81.62 0.07 73.50 0.60 93.70 0.04 41.22 0.52 100.00 0.00 25.39 0.21 100.00 0.00
91.40 1.16 48.44 1.16 94.80 2.92 61.08 0.40 90.25 4.33 83.80 0.06 70.50 0.55 94.33 0.03 39.66 0.48 100.00 0.00 24.39 0.19 100.00 0.00
s2rem Efficiency
s2eff 10
removal
s2rem efficiency
s2eff 20
removal
s
2
rem
efficiency
s2eff 30
removal
s2rem efficiency
s2eff 50
removal
s2rem efficiency
s2eff 80
removal
s2rem efficiency
s2eff
models are reported in Tables 4 and 5 for the prepared sorbents. A comparison between the results of linear and non-linear methods by considering the values of AIC reveals that the non-linear method is more accurate and the results from this method have better agreement with the experimental data as in the all cases the value of AIC is smaller. From the both R2 and AIC quantities, it is obvious that the Langmuir is the best fitting model and has acceptable consistency with the results. Fig. 4 depicts the plot of the mentioned isotherms in the case of 1 wt% MWCNT surface modified PUF. Moreover, the values of AICc in all cases except for the blank PUF propose the ranking of the models as follow: Langmuir > Temkin > Freundlich > RedlichePeterson. Table 4 Analysis of adsorption of crude oil on the blank and surface modified PUF based on the Langmuir, Temkin, Freundlich, and RedlichePeterson equations through linear method. Isotherm
Blank PUF
MWCNT surface-modified PUF 0.5 wt %
Langmuir qm (g g1) KL (L g1) R2 AICc Freundlich KF (g g1) (L/g)n n R2 AICc RedlichePeterson g KR (L g1) aRðL g11=K RÞ R2 AICc Temkin KT (L g1) B1(g g1) R2 AICc
1 wt %
2 wt %
285
Table 5 Analysis of adsorption of crude oil on the blank and surface modified PUF based on Langmuir, Temkin, Freundlich, and RedlichePeterson equations through non-linear method. Isotherm
Blank PUF
MWCNT surface-modified PUF 0.5 wt %
Langmuir qm (g g1) KL (L g1) R2 AICc Freundlich KF (g g1) (L/g)n n R2 AICc RedlichePeterson g KR (L g1) aRðL g11=K RÞ R2 AICc Tempkin KT (L g1) B1(g g1) R2 AICc
1 wt %
2 wt %
3 wt %
21.44 1.009 0.963 9.979
22.660 0.938 0.742 22.839
23.390 4.280 0.601 26.895
22.860 2.650 0.541 26.992
23.670 1.282 0.837 20.723
11.03 5.635 0.7203 22.193
11.790 5.943 0.541 26.311
15.40 7.979 0.604 28.131
14.15 7.522 0.541 27.649
12.94 6.317 0.837 26.573
1.087 16.85 0.586 0.986 14.365
1.157 15.130 0.390 0.796 31.434
1.152 24.900 0.603 0.291 40.334
1.141 23.390 0.619 0.393 38.668
1.158 21.080 0.526 0.911 27.121
30.123 3.059 0.817 19.659
37.062 3.074 0.603 25.435
720.854 2.403 0.543 27.707
399.831 2.402 0.511 27.366
56.692 3.088 0.648 24.169
Langmuir isotherm, good fitting with the data, may suggest homogeneous and monolayer adsorption. In addition, as is seen in Fig. 4, the Langmuir model is the best model describing the saturation condition of the sorbent as in high initial concentration of oil the Langmuir plot becomes almost flat. RedlichePeterson isotherm with the power (g) near unity can satisfy the Langmuir model condition but as it is obvious from Table 5, the power (g) in all cases is more than one. The power larger than one can be explained by the other effective factors rather than the physical and chemical sorption conditions like the capillary action. The reported data in Table 5 reveal that in the case of blank PUF the RedlichePeterson model fits the equilibrium data better than Temkin. This may indicate that the addition of MWCNT to the surface affects the sorption mechanism and weakens the capillary action during the sorption process. In the case of Freundlich model, according to the reported values of n in the Table 5, the power (1/n) with the value less than one implies physical sorption phenomena and as its value gets closer to zero may show more surface heterogeneity which verifies the presented explanation about the Langmuir isotherm (Foo and Hameed, 2010).
3 wt %
20.445 1.421 0.999 9.969
21.413 0.905 0.997 25.441
23.202 2.535 0.998 28.983
22.321 2.435 0.999 46.374
22.075 1.690 0.998 37.549
13.657 2.195 0.800 29.376
9.210 3.913 0.512 26.411
13.317 5.665 0.462 45.657
12.060 5.367 0.443 53.144
10.329 4.076 0.572 74.878
1.060 14.240 0.639 0.999 20.353
1.007 19.011 0.878 0.997 33.431
1.192 24.570 0.498 0.999 40.556
1.091 23.250 0.720 0.999 38.794
1.038 25.570 0.992 0.999 30.057
30.115 3.0593 0.817 19.659
37.059 3.0743 0.603 25.435
720.917 2.403 0.543 27.706
399.850 2.4019 0.511 27.366
56.688 3.088 0.647 24.168
Fig. 4. Plot of the Langmuir, Temkin, Freundlich, and RedlichePeterson isotherm models for 1 wt% MWCNT surface-modified PUF obtained through non-linear regression method.
286
A. Keshavarz et al. / Journal of Environmental Management 157 (2015) 279e286
4. Conclusion It has been concluded from the findings of this research that surface modification of PUF by MWCNT is very favorable for increasing oil removal ability. The tensile elongation test has shown a correlation of the mechanical strength and the amount of MWCNT immobilized onto the PUF surface. The TGA results have revealed an increase in the level of thermal resistance. The results from sorption experiments have confirmed that the sorbent with 1 wt% MWCNT on its surface gives the maximum sorption performance (by adsorbing 24.75 g/g, about 21.44% sorption enhancement in comparison with the blank PUF). It has also been established that the sorption efficiency is increased after grafting MWCNT onto the PUF surface, since in all cases; less water is adsorbed in comparison with the blank PUF. The results on the reusability of the sorbent with 1 wt% MWCNTs on its surface after four cycles of regeneration has shown that 85.45% of the oil sorption capacity remains, however, the oil removal percentage is decreased and the sorption efficiency is increased after each regeneration cycle. Based on the values of R2 and Akaike's Information Criterion (AIC), Langmuir model showed better agreement with the results in comparison with the other fitted isotherms in both case of linear and non-linear methods. References ndez, J., P Aguilera, F., Me asaro, E., Laffon, B., 2010. Review on the effects of exposure to spilled oils on human health. J. Appl. Toxicol. 30, 291e301. Al-Majed, A.A., Adebayo, A.R., Hossain, M.E., 2012. A sustainable approach to controlling oil spills. J. Environ. Manage 113, 213e217. Alves, P., Coelho, J.F.J., Haack, J., Rota, A., Bruinink, A., Gil, M.H., 2009. Surface modification and characterization of thermoplastic polyurethane. Eur. Polym. J. 45, 1412e1419. Banerjee, S.S., Joshi, M.V., Jayaram, R.V., 2006. Treatment of oil spill by sorption technique using fatty acid grafted sawdust. Chemosphere 64, 1026e1031. Bastani, D., Safekordi, A.A., Alihosseini, A., Taghikhani, V., 2006. Study of oil sorption by expanded perlite at 298.15 K. Sep. Purif. Technol. 52, 295e300. Burnham, K.P., Anderson, D.R., 2002. Model selection and multimodel inference: a practical information-theoretic approach. Springer Sci. Bus. Media 149e205. Chin, C.-J.M., Shih, M.-W., Tsai, H.-J., 2010. Adsorption of nonpolar benzene derivatives on single-walled carbon nanotubes. Appl. Surf. Sci. 256, 6035e6039. Deng, S., 2006. Sorbent technology. Encycl. Chem. Process. 2825e2845. Do, D.D., 1998. Adsorption Analysis. World Scientific. Duong, H.T., Burford, R.P., 2006. Effect of foam density, oil viscosity, and temperature on oil sorption behavior of polyurethane. J. Appl. Polym. Sci. 99, 360e367. El-Khaiary, M.I., Malash, G.F., 2011. Common data analysis errors in batch adsorption studies. Hydrometallurgy 105, 314e320. Fan, Z., Yan, J., Ning, G., Wei, T., Qian, W., Zhang, S., Zheng, C., Zhang, Q., Wei, F., 2010. Oil sorption and recovery by using vertically aligned carbon nanotubes. Carbon 48, 4197e4200. Foo, K., Hameed, B., 2010. Insights into the modeling of adsorption isotherm systems. Chem. Eng. J. 156, 2e10. Gui, X., Li, H., Wang, K., Wei, J., Jia, Y., Li, Z., Fan, L., Cao, A., Zhu, H., Wu, D., 2011. Recyclable carbon nanotube sponges for oil absorption. Acta Mater. 59,
4798e4804. Hamdaoui, O., Naffrechoux, E., 2007. Modeling of adsorption isotherms of phenol and chlorophenols onto granular activated carbon: part II. Models with more than two parameters. J. Hazard. Mater. 147, 401e411. Harikrishnan, G., Patro, T.U., Khakhar, D., 2006. Polyurethane foam-clay nanocomposites: nanoclays as cell openers. Ind. Eng. Chem. Res. 45, 7126e7134. Husseien, M., Amer, A., 2009. Availability of barley straw application on oil spill clean up. Int. J. Environ. Sci. Technol. 6, 123e130. Joseph, L., Heo, J., Park, Y.-G., Flora, J.R.V., Yoon, Y., 2011. Adsorption of bisphenol A and 17a-ethinyl estradiol on single walled carbon nanotubes from seawater and brackish water. Desalination 281, 68e74. Li, H., Liu, L., Yang, F., 2012. Hydrophobic modification of polyurethane foam for oil spill cleanup. Mar. Pollut. Bull. 64, 1648e1653. Li, H., Liu, L., Yang, F., 2013a. Oleophilic polyurethane foams for oil spill cleanup. Procedia Environ. Sci. 18, 528e533. Li, Y., Du, Q., Liu, T., Peng, X., Wang, J., Sun, J., Wang, Y., Wu, S., Wang, Z., Xia, Y., Xia, L., 2013b. Comparative study of methylene blue dye adsorption onto activated carbon, graphene oxide, and carbon nanotubes. Chem. Eng. Res. Des. 91, 361e368. Lim, T.-T., Huang, X., 2007. Evaluation of hydrophobicity/oleophilicity of kapok and its performance in oily water filtration: comparison of raw and solvent-treated fibers. Ind. Crop. Prod. 26, 125e134. Pacholczyk, A., Terzyk, A.P., Wisniewski, M., Gauden, P.A., Wesołowski, R.P., Furmaniak, S., Szczes, A., Chibowski, E., Kruszka, B., 2011. Phenol adsorption on closed carbon nanotubes. J. Colloid Interf. Sci. 361, 288e292. Radhika, M., Palanivelu, K., 2006. Adsorptive removal of chlorophenols from aqueous solution by low cost adsorbentdKinetics and isotherm analysis. J. Hazard. Mater. 138, 116e124. Ren, X., Chen, C., Nagatsu, M., Wang, X., 2011. Carbon nanotubes as adsorbents in environmental pollution management: a review. Chem. Eng. J. 170, 395e410. Sahoo, N.G., Jung, Y.C., So, H.H., Cho, J.W., 2007. Synthesis of polyurethane nanocomposites of functionalized carbon nanotubes by in-situ polymerization methods. Korean Phys. Soc. 51, S1eS6. Subramanyam, B., Ashutosh, D., 2011. Adsorption isotherm modeling of phenol onto natural soilseApplicability of various isotherm models. Int. J. Environ. Res. 6, 265e276. A Tanobe, V., Sydenstricker, T., Amico, S., Vargas, J., Zawadzki, S., 2009. Evaluation of flexible post consumed polyurethane foams modified by polystyrene grafting as sorbent material for oil spills. J. Appl. Polym. Sci. 111, 1842e1849. Vasanth Kumar, K., Sivanesan, S., 2006. Equilibrium data, isotherm parameters and process design for partial and complete isotherm of methylene blue onto activated carbon. J. Hazard. Mater. 134, 237e244. Wang, D., Silbaugh, T., Pfeffer, R., Lin, Y.S., 2010. Removal of emulsified oil from water by inverse fluidization of hydrophobic aerogels. Powder Technol. 203, 298e309. Wang, J., Zheng, Y., Wang, A., 2012. Superhydrophobic kapok fiber oil-absorbent: preparation and high oil absorbency. Chem. Eng. J. 213, 1e7. Wang, J., Wang, F., Yao, J., Wang, R., Yuan, H., Masakorala, K., Choi, M.M.F., 2013. Adsorption and desorption of dimethyl phthalate on carbon nanotubes in aqueous copper(II) solution. Colloids Surf. A 417, 47e56. Wu, F.-C., Liu, B.-L., Wu, K.-T., Tseng, R.-L., 2010. A new linear form analysis of RedlichePeterson isotherm equation for the adsorptions of dyes. Chem. Eng. J. 162, 21e27. Wu, D., Fang, L., Qin, Y., Wu, W., Mao, C., Zhu, H., 2014. Oil sorbents with high sorption capacity, oil/water selectivity and reusability for oil spill cleanup. Mar. Pollut. Bull. 84, 263e267. Zhao, D., Zhang, W., Chen, C., Wang, X., 2013. Adsorption of methyl Orange dye onto multiwalled carbon nanotubes. Procedia Environ. Sci. 18, 890e895. Zheng, Y.-P., Wang, H.-N., Kang, F.-Y., Wang, L.-N., Inagaki, M., 2004. Sorption capacity of exfoliated graphite for oils-sorption in and among worm-like particles. Carbon 42, 2603e2607.