Environmental Pollution 186 (2014) 43e49
Contents lists available at ScienceDirect
Environmental Pollution journal homepage: www.elsevier.com/locate/envpol
The kinetic and thermodynamic sorption and stabilization of multiwalled carbon nanotubes in natural organic matter surrogate solutions: The effect of surrogate molecular weight Tingting Li a, Daohui Lin a, *, Lu Li a, b, c, Zhengyu Wang b, Fengchang Wu c, * a b c
Department of Environmental Science, Zhejiang University, Hangzhou 310058, China College of Environmental Science and Engineering, Ocean University of China, Qingdao 266100, China State Key Laboratory of Environmental Criteria and Risk Assessment, Chinese Research Academy of Environmental Sciences, Beijing 100012, China
a r t i c l e i n f o
a b s t r a c t
Article history: Received 21 May 2013 Received in revised form 21 November 2013 Accepted 26 November 2013
Styrene sulfonate (SS) and polystyrene sulfonates (PSSs) were used as surrogates of natural organic matter to study the effect of molecular weight (from 206.2 to 70,000 Da) on their sorption by a multiwalled carbon nanotube (MWCNT) and an activated carbon (AC) and on their stabilization of MWCNT suspension. Results indicate that surface-diffusion through the liquid-sorbent boundary was the ratecontrolling step of the kinetic sorption of both MWCNTs and AC, and surface-occupying and porefilling mechanisms respectively dominated the thermodynamic sorption of MWCNTs and AC. Sorption rates and capacities of MWCNTs and AC in molecular concentration of SS and PSS decreased with increasing molecular weight. The PSSs but not SS facilitated the stabilization of MWCNT suspension because of the increased electrosteric repulsion. The PSSs with more monomers had greater capabilities to stabilize the MWCNT suspension, but the capabilities were comparable after being normalized by the total monomer number. Ó 2013 Elsevier Ltd. All rights reserved.
Keywords: Polystyrene sulfonate Carbon nanotube Activated carbon Adsorption Suspension
1. Introduction Carbon nanotubes (CNTs) are promising nanomaterials that are being increasingly used in a variety of industrial areas (Baughman et al., 2002; Mauter and Elimelech, 2008), which may result in their exposure to the aquatic environment (Petersen et al., 2011; Wiesner et al., 2006). An increasing number of studies have observed hazardous effects of CNTs on cells, tissues, and various organisms (Helland et al., 2007; Johnston et al., 2010; Long et al., 2012), which raises concerns over the environmental behavior and ecological effects of CNTs. Pristine CNTs are highly hydrophobic and can readily aggregate and precipitate in water, and thus their transport is restricted (Lin et al., 2010b). However, ubiquitous natural organic matter (NOM) can adsorb on the hydrophobic CNTs, thus promoting their aqueous stabilization and transport (Hyung et al., 2007; Hyung and Kim, 2008; Lin and Xing, 2008a; Lin et al., 2009, 2010a, 2012; Saleh et al., 2010; Schwyzer et al., 2012; Smith et al., 2012; Zhou et al., 2012) and influencing their ecotoxicity (Edgington et al., 2010). Hence, CNT-NOM interactions are receiving
* Corresponding authors. E-mail addresses: [email protected] (D. Lin), [email protected] (F. Wu). 0269-7491/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.envpol.2013.11.016
more and more attention from researchers (Lin et al., 2010b; Petersen et al., 2011). The underlying mechanism of the CNT-NOM interaction is becoming obvious. NOM can adsorb on CNTs mainly through hydrophobic interactions (Wang et al., 2011), pep interactions (Hyung and Kim, 2008; Lin and Xing, 2008a), hydrogen bonding, and/or electrostatic interactions (Lin et al., 2009). The adsorbed NOM can more or less disperse and form stable suspension of the CNTs via electrostatic repulsion, steric repulsion, and/or solvation (Hyung et al., 2007; Lin and Xing, 2008b; Lin et al., 2012; Saleh et al., 2010; Zhou et al., 2012). However, NOM is a highly heterogeneous mixture, and the interaction of the specific components of NOM with CNTs is still being studied (Liu et al., 2012, 2013; Louie et al., 2013) and merits more specific investigations. It was reported that the NOM adsorption on CNTs varied greatly with the type of NOM and was proportional to the aromatic carbon content of the NOM, and that the concentration of the CNT suspension depended on the amount of NOM adsorbed per unit mass of CNTs (Hyung and Kim, 2008). Our previous study investigated the interaction between CNTs and humic acids (a typical NOM analogue) and found that the humic acid adsorption on CNTs increased with decreasing polarity of the humic acids, whereas the capability of the CNT-adsorbed humic acids for the CNT stabilization increased with increasing polarity of the humic acids (Lin et al.,
44
T. Li et al. / Environmental Pollution 186 (2014) 43e49
2012). The molecular weight or size of the NOM components can play an important role in their interaction with nanoparticles (Louie et al., 2013). However, limited studies have specifically examined the effect of the molecular weight or size of the NOM on its sorption on CNTs. Hyung and Kim (2008) observed that the high-molecular weight fractions of NOM could be preferentially adsorbed by CNTs. The higher adsorption capacity of CNTs for highmolecular weight NOM was also reported in a study by employing tannic acid and gallic acid as high- and low-molecular weight NOM analogues, respectively (Liu et al., 2012). Furthermore, to the best of our knowledge, no study has specifically investigated the effect of the molecular weight or size of NOM on the stabilization of CNTs. Therefore, polystyrene sulfonates (PSSs) with different molecular weights were used in this study as NOM surrogates, in particular, as the aromatic fraction of NOM, to investigate the effect of the molecular weight or size of NOM on its sorption by a type of CNT and the stabilization of the CNTs. PSSs with aromatic rings and oxygen-containing functional groups have been widely used as NOM surrogates (Muller et al., 2000; Kawasaki et al., 2011; Ando et al., 2010; Karanfil et al., 1996; Li et al., 2003; Revchuk and Suffet, 2009) and to suspend CNTs (Huang et al., 2011; Li and Adronov, 2007; O’Connell et al., 2001). An activated carbon (AC) having a similar elemental composition but a different pore structure and morphology as the CNTs was chosen as an alternative sorbent to explore the sorption mechanism and compare the results. The results of this study are expected to promote an understanding of the CNT-NOM interaction and help the evaluation of the environmental behavior and ecological effects of CNTs.
Qt ¼ Qe 1 eK1 t
(1)
1 1 1 1 1 1 1 1 ¼ $ $ þ ¼ $ þ Qt K2 Qe2 t Qe V0 t Qe
(2)
Qt ¼ A þ Ka t 0:5
(3)
where Qt and Qe are the sorbed amounts of sorbate by sorbent at time point t and equilibrium (mg g1), respectively; K1 and K2 are the Lagergren sorption rate constants of the pseudo first-order (h1) and second-order (g mg1 h1) kinetics models, respectively; t is the contact time (h); V0 is the initial sorption rate (mg g1 h1); A is the intercept of the vertical axis (mg g1); Ka is the overall diffusion constant for sorption (mg (g h0.5)1). 2.3. Thermodynamic sorption experiment Sorption isotherms of SS and the PSSs by the MWCNTs and AC were also obtained using a batch experiment in screw cap vials. The MWCNTs and AC were added into 8 mL of the sorbate solutions with initial sorbate concentrations of 0, 5, 10, 15, 20, 40, 60, 80, and 100 mg L1. Each concentration point, including blanks (i.e., without MWCNTs and AC), was run in duplicate. The mixtures were equilibrated in a thermostat shaker (110 rpm) for 24 h. The kinetic adsorption experiments indicated that the apparent sorption equilibrium was reached within 12 h. The equilibrium dissolved and the sorbed concentrations of the sorbates were determined as in the kinetic sorption experiments. The Langmuir model (Eq. (4)) was employed to fit the equilibrium sorption data (Lin and Xing, 2008b; Wang et al., 2011): qe ¼
q0 bCe 1 þ bCe
(4)
where qe (mg g1) is the amount of sorbate, which is sorbed per unit mass of the sorbent, Ce (mg L1) is the sorbate concentration at equilibrium, q0 (mg g1) is the sorption capacity of the sorbent, and b (L mg1) is the constant related to the molar heat of adsorption.
2. Materials and methods 2.1. Materials
2.4. Suspending experiment
Multi-walled carbon nanotubes (MWCNTs) with outer diameters of 60e100 nm were purchased from Shenzhen Nanotech Port Co., China. AC powders were obtained from Sinopharm Group, Shanghai, China. The length and outer diameter of the MWCNTs was measured with the aid of a transmission electron microscope (TEM, JEM-1230, JEOL, Japan). The specific surface areas and total pore volumes of the MWCNTs and AC were calculated from the adsorptionedesorption isotherm of N2 at 77 K by the multipoint BrunauereEmmetteTeller (BET) method, and the mesopore and micropore volumes were calculated by the BarretteJoynereHalenda (BJH) and DubinineAshtakhov (DA) methods, respectively. The ash contents were measured by heating the MWCNTs and AC at 900 C for 10 h. The dry weight-based C, H, and N contents of the MWCNTs were determined using an elemental analyzer (Flash EA 1112, ThermoFinnigan, Italy). The oxygen content of the MWCNTs was calculated by the mass differences. Styrene sulfonate (SS, C8H7O3S.Na) and PSS (C8H7O3S.Na)x with molecular weights of 4300, 6800, 10,000, 17,000, and 70,000 Da were purchased from Sigmae Aldrich Co. The structures and sizes of the SS and SS monomer in the PSS molecules were modeled by the Molecular Operating Environment (MOE) Software (2009). A 0.01 mol L1 NaCl solution at pH 7, representing the typical ionic strength and pH of natural surface water, was used as the background solution in the following kinetic and thermodynamic sorption and stabilization experiments.
The suspending kinetics of the MWCNTs in the SS and PSS solutions were carried out as a function of the sonication time at room temperature. Four mg of the MWCNTs were added into 20 mL of a 20 mg L1 SS or PSS solution. The mixtures were sonicated (600 W, 50 kHz) for 0, 5, 10, 20, 40, 60, 90, 120, and 180 min and were then centrifuged at 3000 g for 15 min. The concentrations of the stabilized MWCNTs in the resultant supernatants were measured by UVevis spectrometry at 800 nm. Samples from each point in time were run in triplicate. The absorbance at 800 nm was calibrated with the MWCNT concentration measured by a total organic carbon analyzer (TOC analyzer, Shimadzu, TOC-VCPH) to quantify the stabilized MWCNTs (Lin et al., 2010a, 2012). The thermodynamic suspending experiment of the MWCNTs in the SS and PSS solutions was conducted by sonicating (600 W, 50 kHz) mixtures of 4 mg of the MWCNTs and 20 mL of the SS and PSS solutions with concentrations of 0, 0.1, 0.5, 1.0, 1.5, 2.0, 2.5, 5, 10, 20, and 40 mg L1 for 3 h. Each concentration point, including the blanks (i.e., without MWCNTs), was run in triplicate. The suspensions were then centrifuged at 3000 g for 15 min. The stabilized MWCNTs in the resultant supernatants were then quantified. The zeta potentials of the suspended MWCNTs were measured by a zetasizer (Nano ZS90, Malvern Instrument, UK).
3. Results and discussion 2.2. Kinetic sorption experiment The kinetic sorptions of SS and the PSSs by the MWCNTs and AC were studied using a batch experiment in screw cap vials at 25 1 C. Eight milliliters of the SS or PSS solutions (20 mg L1) were added into the screw cap vials containing certain amounts of the MWCNTs and AC (Table S1 in the Supporting Information). All vials were shaken in a thermostat shaker (110 rpm) for 1, 2, 3, 4, 5, 10, 20, 40, 60, 120, 180, 360, 720, 1080, and 1440 min. Each point in time was sampled, including blanks (i.e., without MWCNTs and AC), and was run in duplicate. After separating the MWCNTs and AC from the sorptive solution using a 0.2-mm filter membrane (Millipore, PTFE), the residual SS and PSSs were quantified with a UVevis spectrometer (Shimadzu, UV-2540) at 255 and 225 nm because they had marked absorption peaks at these wavelengths, respectively. A negligible mass loss of SS and the PSSs was observed in the sorption and filtration processes. The adsorbed amounts of SS and PSSs by the MWCNTs and AC were then calculated by the mass differences. The Lagergren pseudo first-order (Eq. (1)) and second-order (Eq. (2)) models and the Weber-Morris model (Eq. (3)) were employed to fit the kinetic sorption data (Ho and Mckay, 1999; Svilovic et al., 2010; Wang et al., 2011).
3.1. Properties of the sorbents and sorbates Selected properties of the MWCNTs and AC are summarized in Table 1. The pore size distributions of the MWCNTs and AC are shown in Fig. S1 in the Supporting Information. The specific surface area and the total pore, mesopore, and micropore volumes of the AC were much greater than those of the MWCNTs. The AC had an abundant pore structure with a large amount of micropores. The structure of the MWCNTs was mainly mesoporous. The three-dimensional molecular structure diagrams of SS and the repeating SS monomer of PSS are shown in Fig S2. The molecular size of SS is approximately 6.026e8.430 A. The molecular sizes of the PSSs were not measured by the MOE Software owing to the complex spatial molecular structures.
T. Li et al. / Environmental Pollution 186 (2014) 43e49
45
Table 1 Selected properties of the MWCNTs and AC. Materials
MWCNTs AC
Puritya
95 95
Lengthb (mm)
Outer diameterb (nm)
Asurfc (m2 g1)
Vtotalc (cm3 g1)
Vmicroc (cm3 g1)
Vmesoc (cm3 g1)
Ashd (wt%)
3.2 0.9 e
70 9 e
60.3 1280
0.410 0.952
0.044 0.769
0.394 0.402
2.88 1.98
Elem. Cont.e (%) C
H
O
98.01 e
0.11 e
e e
a
Provided by the supplier. Measured by TEM, n ¼ 100. c Specific surface area (Asurf) and total pore volume (Vtotal) were calculated from the adsorptionedesorption isotherm of N2 at 77 K by multipoint BET method. Mesopore volume (Vmeso) and micropore volume (Vmicro) were calculated from the adsorptionedesorption isotherm of N2 at 77 K by BJH method and DA method, respectively. d Ash content was measured by heating the CNTs at 900 C for 10 h. e Elemental contents were determined using an elemental analyzer. b
and AC could thus dominate the sorption rate. However, the Lagergren pseudo first- and second-order models cannot identify the diffusion mechanism; the kinetic sorption data were thus further analyzed by the WebereMorris model, an intraparticle diffusion model, to look into the potential diffusion mechanism. The sorption parameters fitted by the Weber-Morris model are listed in Table 2. Previous research has indicated that pore diffusion can be the rate-controlling step if Qt is linearly correlated with t0.5 and the regression line passes through the origin (Shen et al., 2009; Wang et al., 2011). The fitted results (Table 2) show that Qt of SS and the PSSs to the MWCNTs and AC are all linearly correlated with t0.5, but the regression lines have positive intercepts, suggesting that the diffusion in the pores is not the rate-controlling step for the kinetic sorptions (Shen et al., 2009). Therefore, the surfacediffusion step may primarily regulate the sorption kinetics of SS and the PSSs to the MWCNTs and AC in this study. It appears that the Lagergren pseudo second-order model can fit the kinetic sorption data better than the Lagergren pseudo first-order model, as indicated by the higher r2 values (Table 2). The sorption rate constant, K2, fitted by the Lagergren pseudo second-order model, was thus used to compare and analyze the sorption kinetics of SS and the PSSs by the sorbents. The AC has a much greater pore volume (1.3 times greater total pore volume and 16 times greater micropore volume) than the MWCNTs (Table 1). However, the AC has a much higher K2 (4.0 times higher for SS and 1.0e221 times higher for the PSSs) than the MWCNTs (Table 2), further suggesting that the pore-diffusion process cannot be the rate-controlling step for the kinetic sorptions of SS and PSS. The huge aspect ratio (tube length to outer diameter ratio) of the MWCNTs cause them to readily aggregate in the sorptive solution (Lin and Xing, 2008b), and together with their much smaller size compared with the AC, may more or less inhibit the diffusion of the sorbate molecules through the liquid-sorbent boundary and account for the relatively slower sorption of the
3.2. Kinetic sorption The kinetic sorption curves of SS and the PSSs by the MWCNTs and AC are presented in Fig. 1. The sorption sharply increased and then leveled off with increasing contact time, presenting a twostage sorption, i.e., an initial rapid sorption and a subsequent slow and stagnant sorption. The sorbents could be heterogeneous with various surface structures and sorption sites with different affinities to the sorbates. The highly active surface sites were available for adsorption in the initial stage, while the remaining adsorption sites were difficult to be occupied owing to their relatively low affinity and/or the steric hindrance from the previouslyadsorbed sorbate molecules. Similar two-stage sorption kinetics have been generally observed in other related studies (Al-Johani and Salam, 2011; Ghaedi et al., 2011; Lu and Su, 2007; Shen et al., 2009; Yu et al., 2012). To deepen our understanding of the kinetic sorption behavior and mechanism, widely used kinetics models, including the Lagergren pseudo first- and second-order models and the WeberMorris model, were employed to fit the kinetic sorption data. The assumption for the Lagergren pseudo first- and second-order models is that the key driving force for the sorption is the difference between the sorption at a given time and the sorption at equilibrium (Wang et al., 2011). According to the Lagergren kinetic sorption models, three steps could be involved in the sorption of SS and the PSSs by the MWCNTs and AC (Ho and Mckay, 1999; Wang et al., 2011): (1) the sorbate molecules diffused from the liquid phase to the liquid-sorbent boundary; (2) the sorbate molecules moved from the liquid-sorbent boundary to the sorbent surfaces; and (3) the sorbate molecules diffused into the MWCNTs and AC. The diffusion of the SS and PSS molecules from the aqueous phase to the liquid-sorbent boundary could be relatively fast under the shaking conditions employed, and the step moving from the boundary to the surfaces along with the diffusion into the MWCNTs
Sorbed amount (mg g-1)
30
30
MWCNTs
AC
25
25
20
20
15
15
10
10
5
5
0
0 0
4
8
12
16
20
24
0
4
8
12
16
20
24
Contact time (h) Fig. 1. Sorption kinetics of SS (Mw ¼ 206.2 Da) and PSSs (with Mw ranging from 4300 Da to 70,000 Da) to the MWCNTs (left) and AC (right). : Mw ¼ 206.2 Da; : Mw ¼ 4300 Da; : Mw ¼ 6800 Da; : Mw ¼ 10,000 Da; þ: Mw ¼ 17,000 Da; : Mw ¼ 70,000 Da.
T. Li et al. / Environmental Pollution 186 (2014) 43e49
0.973 0.966 0.926 0.822 0.985 0.871 0.853 0.904 0.726 0.959 0.934 0.891 0.01 0.08 0.09 0.30 0.03 0.07 0.14 0.09 0.23 0.06 0.11 0.02
0.2 0.6 1.2 1.1 0.4 1.3 3.4 0.3 0.4 0.2 0.2 0.5 4.8 22.1 18.5 16.8 16.7 16.7 41.4 7.0 4.4 5.2 3.5 5.2 0.698 0.934 0.878 0.925 0.843 0.927 0.089 0.722 0.881 0.848 0.423 0.870 0.23 2.89 1.28 2.29 2.02 2.36 1.16 0.83 0.50 0.47 0.22 0.29 0.839 0.790 0.654 0.789 0.813 0.849 0.998 0.769 0.760 0.707 0.558 0.407 17.0 0.2 2.3 0.5 0.3 0.1 84.3 1.4 4.7 7.6 18.8 13.1
1.36 3.70 3.32 3.15 3.03 1.98 19.43 2.46 1.66 1.66 1.45 1.48
3.3. Thermodynamic sorption
a
AC
Molecular weight (MW) of SS is 206.2; the PSSs have MW varied from 4300 to 70,000.
68.0 40.3 104.3 58.2 26.5 8.3 45,900 42.9 49.5 61.9 83.2 67.6 0.1 1.3 0.5 1.0 0.8 1.1 0.1 0.3 0.2 0.2 0.2 0.2 2.0 14.1 6.8 10.8 9.8 11.8 23.3 5.5 3.3 2.9 2.1 2.3 0.750 0.690 0.468 0.705 0.713 0.806 0.998 0.629 0.563 0.517 0.528 0.123 206.2 4300 6800 10,000 17,000 70,000 206.2 4300 6800 10,000 17,000 70,000 MWCNTs
1.9 13.2 6.4 10.4 9.4 10.7 23.3 5.1 3.1 2.7 2.0 2.1
0.1 1.3 0.6 1.0 0.8 1.0 0.1 0.4 0.2 0.2 0.1 0.2
21.7 2.0 11.0 3.8 1.4 0.5 121.4 6.3 11.5 16.0 25.7 54.5
5.1 0.8 3.9 1.4 0.5 0.1 25.1 2.1 3.4 4.5 7.7 35.4
R2 K1 (h1)
MWCNTs than the AC. However, there are also studies reporting faster sorption of MWCNTs than AC. For example, a much faster adsorption of Reactive Red M-2BE dye on a type of MWCNT compared to that on an AC adsorbent was observed and attributed to the expandable pores of the MWCNT aggregates in the sorptive solution (Machado et al., 2011). Therefore, the difference in the sorption rates between the MWCNTs and AC can be dependent on the physical characteristics of the sorptive system, including the properties of the MWCNTs, AC, and background sorptive solution. It may be misleading to make a generalized conclusion based on the comparison of the sorption rates of the MWCNTs and AC according to which one is faster. Detailed mechanisms underlying the difference in the sorption kinetics between MWCNTs and AC merit more specific studies. The sorption rate constant (K2, in sorbate mass concentration) of SS was much higher than that of the PSSs to both the MWCNTs and the AC (Table 2), and the difference became greater when the unit of sorption was changed to sorbate molecular concentration (Fig. S3). This could be because SS, with a smaller molecular size compared to PSS, had a lower steric hindrance. The bulkier nature of the PSS may prevent it from moving through the liquid-sorbent boundary and interacting with the sorbents. Similarly, the PSSs with a higher molecular weight had a slower sorption to the sorbents in sorbate molecular concentration (Fig. S3). The trend that K2 (in sorbate mass concentration) of the PSSs first increased and then decreased with increasing molecular weight, as shown in Table 2, may result from the balance between the decreasing sorption rate of the sorbate molecules and the increasing sorbate molecular weight. This trend also indicated that the steric hindrance for the PSSs adsorbing on the sorbent surfaces was not linearly correlated with their SS monomer number; the hindrance for the PSSs with relatively higher molecular weights (>6800 Da to the MWCNTs and >17,000 Da to the AC) could increase faster with an increasing number of monomers.
0.10 0.50 0.31 0.73 0.24 0.20 0.30 0.39 0.47 0.35 0.45 0.08
b (L mg1) q0 (mg g1) Ka (mg (g h0.5)1) A (mg g1) R2 K2 (g mg1 h1) Qe (mg g1) Qe (mg g1)
V0 (mg g1 h1)
Lagergren pseudo second-order model Lagergren pseudo first-order model SS/PSSa (MW, Da) S Orbents
Table 2 Kinetic and thermodynamic sorption models-fitted parameters for the sorption of SS and PSSs to the MWCNTs and AC.
Weber-Morris model
R2
Langmuir model
R2
46
The sorption isotherms of SS and the PSSs by the MWCNTs and AC are presented in Fig. 2, and the isotherm data were fitted by the Langmuir model, with r2 ranging from 0.726 to 0.985 (Table 2). The sorption capacities (16.7e22.1 mg g1) of the MWCNTs for the PSSs were 3.5e4.6 times that for SS (4.8 mg g1) and slightly decreased with the increasing molecular weight of the PSSs. However, when converted to the molecular concentration (mmol g1), the sorption capacity of the MWCNTs for SS was 4.6e100 times those for the PSSs (Fig. S4), and the decreasing trend of the sorption capacities with the increasing molecular weight of the PSSs became relatively marked. The inner spaces between the graphene layers of MWCNTs are generally unavailable for the sorption of organic chemicals owing to their extremely small size and their being blocked by amorphous carbons and/or residual metal catalysts (Yang and Xing, 2010). However, hydrophobic MWCNTs would bind and agglomerate together in the aqueous phase (Lin and Xing, 2008b), and the bound and agglomerated MWCNTs could form active sorption sites in the interstitial spaces and interconnected groove regions (Pan et al., 2008; Yang and Xing, 2010). The PSSs with molecular weights greater than 4300 could be more or less restricted from entering into the interstitial spaces and groove regions owing to the molecular sieve effect compared with the small molecule SS (Pan et al., 2008; Yang and Xing, 2010). This could contribute to the higher sorption capacity of SS compared to the PSSs in terms of the molecular concentration. The MWCNT surface-sorbed PSSs with larger molecules had greater steric hindrance among the sorbate molecules, which could largely account for the decreasing sorption capacity with the increasing molecular weight of the PSSs.
T. Li et al. / Environmental Pollution 186 (2014) 43e49
50
30
Sorbed amount (mg g-1)
47
AC
MWCNTs
40 20
30 20
10
10 0
0 0
20
40
60
80
100
0
20
40
60
80
100
Equilibrium SS/PSS conc. (mg L-1) Fig. 2. Sorption isotherms of SS (Mw ¼ 206.2 Da) and PSSs (with Mw ranging from 4300 Da to 70,000 Da) to the MWCNTs (left) and AC (right). : Mw ¼ 206.2 Da; : Mw ¼ 4300 Da; : Mw ¼ 6800 Da; : Mw ¼ 10,000 Da; þ: Mw ¼ 17,000 Da; : Mw ¼ 70,000 Da.
Stabilized MWNTs (mg L-1)
80
60
Mw=206.2 Mw=4300 Mw=6800 Mw=10,000 Mw=17,000 Mw=70,000 Water
40
20
0 0
50
100
150
200
Sonication time (min) Fig. 3. The kinetic stabilization of the MWCNTs (200 mg L1) in 20 mg L1 SS (Mw ¼ 206.2 Da) and PSS (with Mw ranging from 4300 Da to 70,000 Da) solutions and ultrapure water after the sonication (600 W, 50 kHz) for different times.
The pore volume of the MWCNTs was limited, especially that of the micropores, as shown in Table 1; hence, surface adsorption rather than pore filling dominated the sorption of the MWCNTs. The oxygen content of the MWCNTs was negligible (Table 1), and the zeta potential of the MWCNTs in the sorptive solution in the absence of the PSSs was close to zero (Fig. 4B); therefore, hydrogen bonding and electrostatic interactions may not contribute to the sorption of PSS to the MWCNTs. The forces regulating the PSS-CNT interaction could mainly be hydrophobic and/or pep interactions. The benzene ring in the SS molecule/monomer could bind to the carbon rings on the MWCNT surfaces with a planar configuration through pep and/or hydrophobic interactions (Lin and Xing,
2008a), but a number of the SS monomers in the PSS molecules must extended into the solution. The PSSs having about 21e339 SS monomers could thus have higher sorption capacities in mass concentration (mg g1) on the MWCNT surfaces than SS, though fewer PSS molecules were sorbed by the MWCNTs. The fact that the sorption capacity of the MWCNTs for the PSSs was higher in mass concentration but lower in molecular concentration than that for SS further demonstrated that pore filling could not be the main sorption mechanism of the MWCNTs, even for SS. In general, micropore filling is the key mechanism for the sorption of small organic molecules by AC (Zhang et al., 2010). The large amount of micropores in the AC (Table 1) could accommodate SS but not the larger PSSs because of the molecular sieve effect. The PSSs could only adsorb on the very limited outer surfaces of the AC. Therefore, the AC had a much higher sorption capacity both in molecular concentration (120e2700 times higher, Fig. S4) and in mass concentration (4.9e11 times higher, Table 2) for SS than for the PSSs. Similar to the trend of the sorption on the MWCNT surfaces, fewer PSS molecules with a larger size were sorbed on the AC (Fig. S4). It can be calculated from Table 2 that the sorption capacity of the MWCNTs for a given PSS was 2.2e3.8 times greater than that of the AC, while the sorption capacity of the AC for SS was 7.6 times greater than that of the MWCNTs. This difference further demonstrated the surface-occupying nature of the MWCNTs and the porefilling nature of the AC for the sorption behavior and was in accordance with previously recorded results (Liu et al., 2012, 2013). For example, it was reported that compared with four types of MWCNTs, a powdered AC had a higher sorption capacity for gallic acid (a small molecule) because of the pore-filling effect but had a lower sorption capacity for tannic acid (a larger molecule) owing to the molecular sieve effect (Liu et al., 2012). 10
A
70
B Zetapotential (mV)
Stabilized MWNTs (mg L-1)
80 60 50 40 30 20 10 0
-10
-30
-50
-70
0
1
2
3
4
5 10 20 30 40 50
0
10
20
30
40
50
Initial SS/PSS conc. (mg L-1) Fig. 4. Changes of the stabilized concentration (A) and zeta potential (B) of the MWCNTs with the initial concentrations of SS (Mw ¼ 206.2 Da) and PSSs (with Mw ranging from 4300 Da to 70,000 Da). : Mw ¼ 206.2 Da; : Mw ¼ 4300 Da; : Mw ¼ 6800 Da; : Mw ¼ 10,000 Da; þ: Mw ¼ 17,000 Da; : Mw ¼ 70,000 Da.
48
T. Li et al. / Environmental Pollution 186 (2014) 43e49
3.4. Stabilization of the MWCNT suspension As a result of the sorption, a fraction of the MWCNTs were stably suspended in the PSS solutions after the sonication. The MWCNTs remaining in the supernatants of the MWCNT-PSS mixtures after centrifugation at 3000 g for 15 min following the sonication were considered to be stably suspended or stabilized (Lin and Xing, 2008b; Lin et al., 2012). The suspending kinetics of the MWCNTs in the SS and PSS solutions are presented in Fig. 3. No stabilized MWCNTs were detected in the SS solution and ultra-pure water even after sonication for 3 h. The amount of stabilized MWCNTs increased with the sonication time in the PSS solutions and then leveled off after about 2 h, while no significant difference in the suspending kinetics among the PSSs was observed. The MWCNT stabilization in the PSS solution could be due to the increased electrostatic and steric repulsions from the PSS sorption on the MWCNT surfaces, as discussed later. However, the MWCNTs would not be stably suspended if there were no sonication to disperse the MWCNT aggregates. The dispersing or disaggregating of the MWCNT aggregates increases with increasing sonication energy (i.e., sonication time in this study) (Yang et al., 2013), which, together with the increasing PSS sorption, contributed to the increasing MWCNT stabilization in the PSS solutions with increasing sonication time. The equilibrium MWCNT stabilizations in the SS and PSS solutions are shown in Fig. 4A. The MWCNTs could not be stabilized in the SS solutions with concentrations up to 40 mg L1, while a threestage stabilization was observed in the PSS solutions with an increasing concentration of the PSSs. Below 1.5 mg L1, the PSSs could not stabilize the MWCNTs; the MWCNT stabilization sharply increased in the PSS solutions with increasing PSS concentration from 1.5 to 2.5 mg L1 and then leveled off above 2.5 mg L1. Similar to the suspending kinetics, there was no marked difference in the equilibrium MWCNT stabilization in the five PSS solutions. However, when the PSS concentration was expressed as the molecular concentration, a lower concentration of PSS with a higher molecular weight was needed to obtain the maximum MWCNT stabilization (Fig. S5). This implies that polymers with more monomers per molecule have a greater capability per molecule to stabilize CNT suspensions, but the capabilities will be comparable if they are normalized by the total monomer number, i.e., the capability to stabilize CNTs per monomer of the polymers with a different number of repeat units is similar. The MWCNT stabilization in the PSS solutions could be due to the combined effects of increased electrostatic repulsion and steric repulsion as a result of the PSS sorption. The zeta potential of the MWCNTs became more negative, decreasing to less than 30 mV and 50 mV in the SS and PSS solutions with increasing SS and PSS concentrations, respectively (Fig. 4B). The adsorption of SS could offer strong electrostatic repulsion between the MWCNTs by lowering the zeta potential below 30 mV; however, it could not effectively disperse and stabilize the MWCNTs owing to the lack of sufficient steric repulsion. It has been observed in our previous studies that MWCNTs could not be stabilized in water even when the zeta potential of the MWCNTs was lowered to around 50 mV by increasing the solution pH to 11 (Lin et al., 2009), while MWCNTs with a zeta potential around 30 mV could be stably suspended in the presence of a nonionic surfactant (Lin et al., 2010a). SS molecules could bind to the MWCNT surfaces with a planar configuration, while a fraction of the SS monomers in the MWCNT-adsorbed PSS molecules must extend toward the solution. Thus, the adsorbed PSSs could offer steric repulsion between the MWCNTs in addition to the increased electrostatic repulsion and effectively disperse and stabilize the MWCNTs. Low concentrations (
Contents lists available at ScienceDirect
Environmental Pollution journal homepage: www.elsevier.com/locate/envpol
The kinetic and thermodynamic sorption and stabilization of multiwalled carbon nanotubes in natural organic matter surrogate solutions: The effect of surrogate molecular weight Tingting Li a, Daohui Lin a, *, Lu Li a, b, c, Zhengyu Wang b, Fengchang Wu c, * a b c
Department of Environmental Science, Zhejiang University, Hangzhou 310058, China College of Environmental Science and Engineering, Ocean University of China, Qingdao 266100, China State Key Laboratory of Environmental Criteria and Risk Assessment, Chinese Research Academy of Environmental Sciences, Beijing 100012, China
a r t i c l e i n f o
a b s t r a c t
Article history: Received 21 May 2013 Received in revised form 21 November 2013 Accepted 26 November 2013
Styrene sulfonate (SS) and polystyrene sulfonates (PSSs) were used as surrogates of natural organic matter to study the effect of molecular weight (from 206.2 to 70,000 Da) on their sorption by a multiwalled carbon nanotube (MWCNT) and an activated carbon (AC) and on their stabilization of MWCNT suspension. Results indicate that surface-diffusion through the liquid-sorbent boundary was the ratecontrolling step of the kinetic sorption of both MWCNTs and AC, and surface-occupying and porefilling mechanisms respectively dominated the thermodynamic sorption of MWCNTs and AC. Sorption rates and capacities of MWCNTs and AC in molecular concentration of SS and PSS decreased with increasing molecular weight. The PSSs but not SS facilitated the stabilization of MWCNT suspension because of the increased electrosteric repulsion. The PSSs with more monomers had greater capabilities to stabilize the MWCNT suspension, but the capabilities were comparable after being normalized by the total monomer number. Ó 2013 Elsevier Ltd. All rights reserved.
Keywords: Polystyrene sulfonate Carbon nanotube Activated carbon Adsorption Suspension
1. Introduction Carbon nanotubes (CNTs) are promising nanomaterials that are being increasingly used in a variety of industrial areas (Baughman et al., 2002; Mauter and Elimelech, 2008), which may result in their exposure to the aquatic environment (Petersen et al., 2011; Wiesner et al., 2006). An increasing number of studies have observed hazardous effects of CNTs on cells, tissues, and various organisms (Helland et al., 2007; Johnston et al., 2010; Long et al., 2012), which raises concerns over the environmental behavior and ecological effects of CNTs. Pristine CNTs are highly hydrophobic and can readily aggregate and precipitate in water, and thus their transport is restricted (Lin et al., 2010b). However, ubiquitous natural organic matter (NOM) can adsorb on the hydrophobic CNTs, thus promoting their aqueous stabilization and transport (Hyung et al., 2007; Hyung and Kim, 2008; Lin and Xing, 2008a; Lin et al., 2009, 2010a, 2012; Saleh et al., 2010; Schwyzer et al., 2012; Smith et al., 2012; Zhou et al., 2012) and influencing their ecotoxicity (Edgington et al., 2010). Hence, CNT-NOM interactions are receiving
* Corresponding authors. E-mail addresses: [email protected] (D. Lin), [email protected] (F. Wu). 0269-7491/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.envpol.2013.11.016
more and more attention from researchers (Lin et al., 2010b; Petersen et al., 2011). The underlying mechanism of the CNT-NOM interaction is becoming obvious. NOM can adsorb on CNTs mainly through hydrophobic interactions (Wang et al., 2011), pep interactions (Hyung and Kim, 2008; Lin and Xing, 2008a), hydrogen bonding, and/or electrostatic interactions (Lin et al., 2009). The adsorbed NOM can more or less disperse and form stable suspension of the CNTs via electrostatic repulsion, steric repulsion, and/or solvation (Hyung et al., 2007; Lin and Xing, 2008b; Lin et al., 2012; Saleh et al., 2010; Zhou et al., 2012). However, NOM is a highly heterogeneous mixture, and the interaction of the specific components of NOM with CNTs is still being studied (Liu et al., 2012, 2013; Louie et al., 2013) and merits more specific investigations. It was reported that the NOM adsorption on CNTs varied greatly with the type of NOM and was proportional to the aromatic carbon content of the NOM, and that the concentration of the CNT suspension depended on the amount of NOM adsorbed per unit mass of CNTs (Hyung and Kim, 2008). Our previous study investigated the interaction between CNTs and humic acids (a typical NOM analogue) and found that the humic acid adsorption on CNTs increased with decreasing polarity of the humic acids, whereas the capability of the CNT-adsorbed humic acids for the CNT stabilization increased with increasing polarity of the humic acids (Lin et al.,
44
T. Li et al. / Environmental Pollution 186 (2014) 43e49
2012). The molecular weight or size of the NOM components can play an important role in their interaction with nanoparticles (Louie et al., 2013). However, limited studies have specifically examined the effect of the molecular weight or size of the NOM on its sorption on CNTs. Hyung and Kim (2008) observed that the high-molecular weight fractions of NOM could be preferentially adsorbed by CNTs. The higher adsorption capacity of CNTs for highmolecular weight NOM was also reported in a study by employing tannic acid and gallic acid as high- and low-molecular weight NOM analogues, respectively (Liu et al., 2012). Furthermore, to the best of our knowledge, no study has specifically investigated the effect of the molecular weight or size of NOM on the stabilization of CNTs. Therefore, polystyrene sulfonates (PSSs) with different molecular weights were used in this study as NOM surrogates, in particular, as the aromatic fraction of NOM, to investigate the effect of the molecular weight or size of NOM on its sorption by a type of CNT and the stabilization of the CNTs. PSSs with aromatic rings and oxygen-containing functional groups have been widely used as NOM surrogates (Muller et al., 2000; Kawasaki et al., 2011; Ando et al., 2010; Karanfil et al., 1996; Li et al., 2003; Revchuk and Suffet, 2009) and to suspend CNTs (Huang et al., 2011; Li and Adronov, 2007; O’Connell et al., 2001). An activated carbon (AC) having a similar elemental composition but a different pore structure and morphology as the CNTs was chosen as an alternative sorbent to explore the sorption mechanism and compare the results. The results of this study are expected to promote an understanding of the CNT-NOM interaction and help the evaluation of the environmental behavior and ecological effects of CNTs.
Qt ¼ Qe 1 eK1 t
(1)
1 1 1 1 1 1 1 1 ¼ $ $ þ ¼ $ þ Qt K2 Qe2 t Qe V0 t Qe
(2)
Qt ¼ A þ Ka t 0:5
(3)
where Qt and Qe are the sorbed amounts of sorbate by sorbent at time point t and equilibrium (mg g1), respectively; K1 and K2 are the Lagergren sorption rate constants of the pseudo first-order (h1) and second-order (g mg1 h1) kinetics models, respectively; t is the contact time (h); V0 is the initial sorption rate (mg g1 h1); A is the intercept of the vertical axis (mg g1); Ka is the overall diffusion constant for sorption (mg (g h0.5)1). 2.3. Thermodynamic sorption experiment Sorption isotherms of SS and the PSSs by the MWCNTs and AC were also obtained using a batch experiment in screw cap vials. The MWCNTs and AC were added into 8 mL of the sorbate solutions with initial sorbate concentrations of 0, 5, 10, 15, 20, 40, 60, 80, and 100 mg L1. Each concentration point, including blanks (i.e., without MWCNTs and AC), was run in duplicate. The mixtures were equilibrated in a thermostat shaker (110 rpm) for 24 h. The kinetic adsorption experiments indicated that the apparent sorption equilibrium was reached within 12 h. The equilibrium dissolved and the sorbed concentrations of the sorbates were determined as in the kinetic sorption experiments. The Langmuir model (Eq. (4)) was employed to fit the equilibrium sorption data (Lin and Xing, 2008b; Wang et al., 2011): qe ¼
q0 bCe 1 þ bCe
(4)
where qe (mg g1) is the amount of sorbate, which is sorbed per unit mass of the sorbent, Ce (mg L1) is the sorbate concentration at equilibrium, q0 (mg g1) is the sorption capacity of the sorbent, and b (L mg1) is the constant related to the molar heat of adsorption.
2. Materials and methods 2.1. Materials
2.4. Suspending experiment
Multi-walled carbon nanotubes (MWCNTs) with outer diameters of 60e100 nm were purchased from Shenzhen Nanotech Port Co., China. AC powders were obtained from Sinopharm Group, Shanghai, China. The length and outer diameter of the MWCNTs was measured with the aid of a transmission electron microscope (TEM, JEM-1230, JEOL, Japan). The specific surface areas and total pore volumes of the MWCNTs and AC were calculated from the adsorptionedesorption isotherm of N2 at 77 K by the multipoint BrunauereEmmetteTeller (BET) method, and the mesopore and micropore volumes were calculated by the BarretteJoynereHalenda (BJH) and DubinineAshtakhov (DA) methods, respectively. The ash contents were measured by heating the MWCNTs and AC at 900 C for 10 h. The dry weight-based C, H, and N contents of the MWCNTs were determined using an elemental analyzer (Flash EA 1112, ThermoFinnigan, Italy). The oxygen content of the MWCNTs was calculated by the mass differences. Styrene sulfonate (SS, C8H7O3S.Na) and PSS (C8H7O3S.Na)x with molecular weights of 4300, 6800, 10,000, 17,000, and 70,000 Da were purchased from Sigmae Aldrich Co. The structures and sizes of the SS and SS monomer in the PSS molecules were modeled by the Molecular Operating Environment (MOE) Software (2009). A 0.01 mol L1 NaCl solution at pH 7, representing the typical ionic strength and pH of natural surface water, was used as the background solution in the following kinetic and thermodynamic sorption and stabilization experiments.
The suspending kinetics of the MWCNTs in the SS and PSS solutions were carried out as a function of the sonication time at room temperature. Four mg of the MWCNTs were added into 20 mL of a 20 mg L1 SS or PSS solution. The mixtures were sonicated (600 W, 50 kHz) for 0, 5, 10, 20, 40, 60, 90, 120, and 180 min and were then centrifuged at 3000 g for 15 min. The concentrations of the stabilized MWCNTs in the resultant supernatants were measured by UVevis spectrometry at 800 nm. Samples from each point in time were run in triplicate. The absorbance at 800 nm was calibrated with the MWCNT concentration measured by a total organic carbon analyzer (TOC analyzer, Shimadzu, TOC-VCPH) to quantify the stabilized MWCNTs (Lin et al., 2010a, 2012). The thermodynamic suspending experiment of the MWCNTs in the SS and PSS solutions was conducted by sonicating (600 W, 50 kHz) mixtures of 4 mg of the MWCNTs and 20 mL of the SS and PSS solutions with concentrations of 0, 0.1, 0.5, 1.0, 1.5, 2.0, 2.5, 5, 10, 20, and 40 mg L1 for 3 h. Each concentration point, including the blanks (i.e., without MWCNTs), was run in triplicate. The suspensions were then centrifuged at 3000 g for 15 min. The stabilized MWCNTs in the resultant supernatants were then quantified. The zeta potentials of the suspended MWCNTs were measured by a zetasizer (Nano ZS90, Malvern Instrument, UK).
3. Results and discussion 2.2. Kinetic sorption experiment The kinetic sorptions of SS and the PSSs by the MWCNTs and AC were studied using a batch experiment in screw cap vials at 25 1 C. Eight milliliters of the SS or PSS solutions (20 mg L1) were added into the screw cap vials containing certain amounts of the MWCNTs and AC (Table S1 in the Supporting Information). All vials were shaken in a thermostat shaker (110 rpm) for 1, 2, 3, 4, 5, 10, 20, 40, 60, 120, 180, 360, 720, 1080, and 1440 min. Each point in time was sampled, including blanks (i.e., without MWCNTs and AC), and was run in duplicate. After separating the MWCNTs and AC from the sorptive solution using a 0.2-mm filter membrane (Millipore, PTFE), the residual SS and PSSs were quantified with a UVevis spectrometer (Shimadzu, UV-2540) at 255 and 225 nm because they had marked absorption peaks at these wavelengths, respectively. A negligible mass loss of SS and the PSSs was observed in the sorption and filtration processes. The adsorbed amounts of SS and PSSs by the MWCNTs and AC were then calculated by the mass differences. The Lagergren pseudo first-order (Eq. (1)) and second-order (Eq. (2)) models and the Weber-Morris model (Eq. (3)) were employed to fit the kinetic sorption data (Ho and Mckay, 1999; Svilovic et al., 2010; Wang et al., 2011).
3.1. Properties of the sorbents and sorbates Selected properties of the MWCNTs and AC are summarized in Table 1. The pore size distributions of the MWCNTs and AC are shown in Fig. S1 in the Supporting Information. The specific surface area and the total pore, mesopore, and micropore volumes of the AC were much greater than those of the MWCNTs. The AC had an abundant pore structure with a large amount of micropores. The structure of the MWCNTs was mainly mesoporous. The three-dimensional molecular structure diagrams of SS and the repeating SS monomer of PSS are shown in Fig S2. The molecular size of SS is approximately 6.026e8.430 A. The molecular sizes of the PSSs were not measured by the MOE Software owing to the complex spatial molecular structures.
T. Li et al. / Environmental Pollution 186 (2014) 43e49
45
Table 1 Selected properties of the MWCNTs and AC. Materials
MWCNTs AC
Puritya
95 95
Lengthb (mm)
Outer diameterb (nm)
Asurfc (m2 g1)
Vtotalc (cm3 g1)
Vmicroc (cm3 g1)
Vmesoc (cm3 g1)
Ashd (wt%)
3.2 0.9 e
70 9 e
60.3 1280
0.410 0.952
0.044 0.769
0.394 0.402
2.88 1.98
Elem. Cont.e (%) C
H
O
98.01 e
0.11 e
e e
a
Provided by the supplier. Measured by TEM, n ¼ 100. c Specific surface area (Asurf) and total pore volume (Vtotal) were calculated from the adsorptionedesorption isotherm of N2 at 77 K by multipoint BET method. Mesopore volume (Vmeso) and micropore volume (Vmicro) were calculated from the adsorptionedesorption isotherm of N2 at 77 K by BJH method and DA method, respectively. d Ash content was measured by heating the CNTs at 900 C for 10 h. e Elemental contents were determined using an elemental analyzer. b
and AC could thus dominate the sorption rate. However, the Lagergren pseudo first- and second-order models cannot identify the diffusion mechanism; the kinetic sorption data were thus further analyzed by the WebereMorris model, an intraparticle diffusion model, to look into the potential diffusion mechanism. The sorption parameters fitted by the Weber-Morris model are listed in Table 2. Previous research has indicated that pore diffusion can be the rate-controlling step if Qt is linearly correlated with t0.5 and the regression line passes through the origin (Shen et al., 2009; Wang et al., 2011). The fitted results (Table 2) show that Qt of SS and the PSSs to the MWCNTs and AC are all linearly correlated with t0.5, but the regression lines have positive intercepts, suggesting that the diffusion in the pores is not the rate-controlling step for the kinetic sorptions (Shen et al., 2009). Therefore, the surfacediffusion step may primarily regulate the sorption kinetics of SS and the PSSs to the MWCNTs and AC in this study. It appears that the Lagergren pseudo second-order model can fit the kinetic sorption data better than the Lagergren pseudo first-order model, as indicated by the higher r2 values (Table 2). The sorption rate constant, K2, fitted by the Lagergren pseudo second-order model, was thus used to compare and analyze the sorption kinetics of SS and the PSSs by the sorbents. The AC has a much greater pore volume (1.3 times greater total pore volume and 16 times greater micropore volume) than the MWCNTs (Table 1). However, the AC has a much higher K2 (4.0 times higher for SS and 1.0e221 times higher for the PSSs) than the MWCNTs (Table 2), further suggesting that the pore-diffusion process cannot be the rate-controlling step for the kinetic sorptions of SS and PSS. The huge aspect ratio (tube length to outer diameter ratio) of the MWCNTs cause them to readily aggregate in the sorptive solution (Lin and Xing, 2008b), and together with their much smaller size compared with the AC, may more or less inhibit the diffusion of the sorbate molecules through the liquid-sorbent boundary and account for the relatively slower sorption of the
3.2. Kinetic sorption The kinetic sorption curves of SS and the PSSs by the MWCNTs and AC are presented in Fig. 1. The sorption sharply increased and then leveled off with increasing contact time, presenting a twostage sorption, i.e., an initial rapid sorption and a subsequent slow and stagnant sorption. The sorbents could be heterogeneous with various surface structures and sorption sites with different affinities to the sorbates. The highly active surface sites were available for adsorption in the initial stage, while the remaining adsorption sites were difficult to be occupied owing to their relatively low affinity and/or the steric hindrance from the previouslyadsorbed sorbate molecules. Similar two-stage sorption kinetics have been generally observed in other related studies (Al-Johani and Salam, 2011; Ghaedi et al., 2011; Lu and Su, 2007; Shen et al., 2009; Yu et al., 2012). To deepen our understanding of the kinetic sorption behavior and mechanism, widely used kinetics models, including the Lagergren pseudo first- and second-order models and the WeberMorris model, were employed to fit the kinetic sorption data. The assumption for the Lagergren pseudo first- and second-order models is that the key driving force for the sorption is the difference between the sorption at a given time and the sorption at equilibrium (Wang et al., 2011). According to the Lagergren kinetic sorption models, three steps could be involved in the sorption of SS and the PSSs by the MWCNTs and AC (Ho and Mckay, 1999; Wang et al., 2011): (1) the sorbate molecules diffused from the liquid phase to the liquid-sorbent boundary; (2) the sorbate molecules moved from the liquid-sorbent boundary to the sorbent surfaces; and (3) the sorbate molecules diffused into the MWCNTs and AC. The diffusion of the SS and PSS molecules from the aqueous phase to the liquid-sorbent boundary could be relatively fast under the shaking conditions employed, and the step moving from the boundary to the surfaces along with the diffusion into the MWCNTs
Sorbed amount (mg g-1)
30
30
MWCNTs
AC
25
25
20
20
15
15
10
10
5
5
0
0 0
4
8
12
16
20
24
0
4
8
12
16
20
24
Contact time (h) Fig. 1. Sorption kinetics of SS (Mw ¼ 206.2 Da) and PSSs (with Mw ranging from 4300 Da to 70,000 Da) to the MWCNTs (left) and AC (right). : Mw ¼ 206.2 Da; : Mw ¼ 4300 Da; : Mw ¼ 6800 Da; : Mw ¼ 10,000 Da; þ: Mw ¼ 17,000 Da; : Mw ¼ 70,000 Da.
T. Li et al. / Environmental Pollution 186 (2014) 43e49
0.973 0.966 0.926 0.822 0.985 0.871 0.853 0.904 0.726 0.959 0.934 0.891 0.01 0.08 0.09 0.30 0.03 0.07 0.14 0.09 0.23 0.06 0.11 0.02
0.2 0.6 1.2 1.1 0.4 1.3 3.4 0.3 0.4 0.2 0.2 0.5 4.8 22.1 18.5 16.8 16.7 16.7 41.4 7.0 4.4 5.2 3.5 5.2 0.698 0.934 0.878 0.925 0.843 0.927 0.089 0.722 0.881 0.848 0.423 0.870 0.23 2.89 1.28 2.29 2.02 2.36 1.16 0.83 0.50 0.47 0.22 0.29 0.839 0.790 0.654 0.789 0.813 0.849 0.998 0.769 0.760 0.707 0.558 0.407 17.0 0.2 2.3 0.5 0.3 0.1 84.3 1.4 4.7 7.6 18.8 13.1
1.36 3.70 3.32 3.15 3.03 1.98 19.43 2.46 1.66 1.66 1.45 1.48
3.3. Thermodynamic sorption
a
AC
Molecular weight (MW) of SS is 206.2; the PSSs have MW varied from 4300 to 70,000.
68.0 40.3 104.3 58.2 26.5 8.3 45,900 42.9 49.5 61.9 83.2 67.6 0.1 1.3 0.5 1.0 0.8 1.1 0.1 0.3 0.2 0.2 0.2 0.2 2.0 14.1 6.8 10.8 9.8 11.8 23.3 5.5 3.3 2.9 2.1 2.3 0.750 0.690 0.468 0.705 0.713 0.806 0.998 0.629 0.563 0.517 0.528 0.123 206.2 4300 6800 10,000 17,000 70,000 206.2 4300 6800 10,000 17,000 70,000 MWCNTs
1.9 13.2 6.4 10.4 9.4 10.7 23.3 5.1 3.1 2.7 2.0 2.1
0.1 1.3 0.6 1.0 0.8 1.0 0.1 0.4 0.2 0.2 0.1 0.2
21.7 2.0 11.0 3.8 1.4 0.5 121.4 6.3 11.5 16.0 25.7 54.5
5.1 0.8 3.9 1.4 0.5 0.1 25.1 2.1 3.4 4.5 7.7 35.4
R2 K1 (h1)
MWCNTs than the AC. However, there are also studies reporting faster sorption of MWCNTs than AC. For example, a much faster adsorption of Reactive Red M-2BE dye on a type of MWCNT compared to that on an AC adsorbent was observed and attributed to the expandable pores of the MWCNT aggregates in the sorptive solution (Machado et al., 2011). Therefore, the difference in the sorption rates between the MWCNTs and AC can be dependent on the physical characteristics of the sorptive system, including the properties of the MWCNTs, AC, and background sorptive solution. It may be misleading to make a generalized conclusion based on the comparison of the sorption rates of the MWCNTs and AC according to which one is faster. Detailed mechanisms underlying the difference in the sorption kinetics between MWCNTs and AC merit more specific studies. The sorption rate constant (K2, in sorbate mass concentration) of SS was much higher than that of the PSSs to both the MWCNTs and the AC (Table 2), and the difference became greater when the unit of sorption was changed to sorbate molecular concentration (Fig. S3). This could be because SS, with a smaller molecular size compared to PSS, had a lower steric hindrance. The bulkier nature of the PSS may prevent it from moving through the liquid-sorbent boundary and interacting with the sorbents. Similarly, the PSSs with a higher molecular weight had a slower sorption to the sorbents in sorbate molecular concentration (Fig. S3). The trend that K2 (in sorbate mass concentration) of the PSSs first increased and then decreased with increasing molecular weight, as shown in Table 2, may result from the balance between the decreasing sorption rate of the sorbate molecules and the increasing sorbate molecular weight. This trend also indicated that the steric hindrance for the PSSs adsorbing on the sorbent surfaces was not linearly correlated with their SS monomer number; the hindrance for the PSSs with relatively higher molecular weights (>6800 Da to the MWCNTs and >17,000 Da to the AC) could increase faster with an increasing number of monomers.
0.10 0.50 0.31 0.73 0.24 0.20 0.30 0.39 0.47 0.35 0.45 0.08
b (L mg1) q0 (mg g1) Ka (mg (g h0.5)1) A (mg g1) R2 K2 (g mg1 h1) Qe (mg g1) Qe (mg g1)
V0 (mg g1 h1)
Lagergren pseudo second-order model Lagergren pseudo first-order model SS/PSSa (MW, Da) S Orbents
Table 2 Kinetic and thermodynamic sorption models-fitted parameters for the sorption of SS and PSSs to the MWCNTs and AC.
Weber-Morris model
R2
Langmuir model
R2
46
The sorption isotherms of SS and the PSSs by the MWCNTs and AC are presented in Fig. 2, and the isotherm data were fitted by the Langmuir model, with r2 ranging from 0.726 to 0.985 (Table 2). The sorption capacities (16.7e22.1 mg g1) of the MWCNTs for the PSSs were 3.5e4.6 times that for SS (4.8 mg g1) and slightly decreased with the increasing molecular weight of the PSSs. However, when converted to the molecular concentration (mmol g1), the sorption capacity of the MWCNTs for SS was 4.6e100 times those for the PSSs (Fig. S4), and the decreasing trend of the sorption capacities with the increasing molecular weight of the PSSs became relatively marked. The inner spaces between the graphene layers of MWCNTs are generally unavailable for the sorption of organic chemicals owing to their extremely small size and their being blocked by amorphous carbons and/or residual metal catalysts (Yang and Xing, 2010). However, hydrophobic MWCNTs would bind and agglomerate together in the aqueous phase (Lin and Xing, 2008b), and the bound and agglomerated MWCNTs could form active sorption sites in the interstitial spaces and interconnected groove regions (Pan et al., 2008; Yang and Xing, 2010). The PSSs with molecular weights greater than 4300 could be more or less restricted from entering into the interstitial spaces and groove regions owing to the molecular sieve effect compared with the small molecule SS (Pan et al., 2008; Yang and Xing, 2010). This could contribute to the higher sorption capacity of SS compared to the PSSs in terms of the molecular concentration. The MWCNT surface-sorbed PSSs with larger molecules had greater steric hindrance among the sorbate molecules, which could largely account for the decreasing sorption capacity with the increasing molecular weight of the PSSs.
T. Li et al. / Environmental Pollution 186 (2014) 43e49
50
30
Sorbed amount (mg g-1)
47
AC
MWCNTs
40 20
30 20
10
10 0
0 0
20
40
60
80
100
0
20
40
60
80
100
Equilibrium SS/PSS conc. (mg L-1) Fig. 2. Sorption isotherms of SS (Mw ¼ 206.2 Da) and PSSs (with Mw ranging from 4300 Da to 70,000 Da) to the MWCNTs (left) and AC (right). : Mw ¼ 206.2 Da; : Mw ¼ 4300 Da; : Mw ¼ 6800 Da; : Mw ¼ 10,000 Da; þ: Mw ¼ 17,000 Da; : Mw ¼ 70,000 Da.
Stabilized MWNTs (mg L-1)
80
60
Mw=206.2 Mw=4300 Mw=6800 Mw=10,000 Mw=17,000 Mw=70,000 Water
40
20
0 0
50
100
150
200
Sonication time (min) Fig. 3. The kinetic stabilization of the MWCNTs (200 mg L1) in 20 mg L1 SS (Mw ¼ 206.2 Da) and PSS (with Mw ranging from 4300 Da to 70,000 Da) solutions and ultrapure water after the sonication (600 W, 50 kHz) for different times.
The pore volume of the MWCNTs was limited, especially that of the micropores, as shown in Table 1; hence, surface adsorption rather than pore filling dominated the sorption of the MWCNTs. The oxygen content of the MWCNTs was negligible (Table 1), and the zeta potential of the MWCNTs in the sorptive solution in the absence of the PSSs was close to zero (Fig. 4B); therefore, hydrogen bonding and electrostatic interactions may not contribute to the sorption of PSS to the MWCNTs. The forces regulating the PSS-CNT interaction could mainly be hydrophobic and/or pep interactions. The benzene ring in the SS molecule/monomer could bind to the carbon rings on the MWCNT surfaces with a planar configuration through pep and/or hydrophobic interactions (Lin and Xing,
2008a), but a number of the SS monomers in the PSS molecules must extended into the solution. The PSSs having about 21e339 SS monomers could thus have higher sorption capacities in mass concentration (mg g1) on the MWCNT surfaces than SS, though fewer PSS molecules were sorbed by the MWCNTs. The fact that the sorption capacity of the MWCNTs for the PSSs was higher in mass concentration but lower in molecular concentration than that for SS further demonstrated that pore filling could not be the main sorption mechanism of the MWCNTs, even for SS. In general, micropore filling is the key mechanism for the sorption of small organic molecules by AC (Zhang et al., 2010). The large amount of micropores in the AC (Table 1) could accommodate SS but not the larger PSSs because of the molecular sieve effect. The PSSs could only adsorb on the very limited outer surfaces of the AC. Therefore, the AC had a much higher sorption capacity both in molecular concentration (120e2700 times higher, Fig. S4) and in mass concentration (4.9e11 times higher, Table 2) for SS than for the PSSs. Similar to the trend of the sorption on the MWCNT surfaces, fewer PSS molecules with a larger size were sorbed on the AC (Fig. S4). It can be calculated from Table 2 that the sorption capacity of the MWCNTs for a given PSS was 2.2e3.8 times greater than that of the AC, while the sorption capacity of the AC for SS was 7.6 times greater than that of the MWCNTs. This difference further demonstrated the surface-occupying nature of the MWCNTs and the porefilling nature of the AC for the sorption behavior and was in accordance with previously recorded results (Liu et al., 2012, 2013). For example, it was reported that compared with four types of MWCNTs, a powdered AC had a higher sorption capacity for gallic acid (a small molecule) because of the pore-filling effect but had a lower sorption capacity for tannic acid (a larger molecule) owing to the molecular sieve effect (Liu et al., 2012). 10
A
70
B Zetapotential (mV)
Stabilized MWNTs (mg L-1)
80 60 50 40 30 20 10 0
-10
-30
-50
-70
0
1
2
3
4
5 10 20 30 40 50
0
10
20
30
40
50
Initial SS/PSS conc. (mg L-1) Fig. 4. Changes of the stabilized concentration (A) and zeta potential (B) of the MWCNTs with the initial concentrations of SS (Mw ¼ 206.2 Da) and PSSs (with Mw ranging from 4300 Da to 70,000 Da). : Mw ¼ 206.2 Da; : Mw ¼ 4300 Da; : Mw ¼ 6800 Da; : Mw ¼ 10,000 Da; þ: Mw ¼ 17,000 Da; : Mw ¼ 70,000 Da.
48
T. Li et al. / Environmental Pollution 186 (2014) 43e49
3.4. Stabilization of the MWCNT suspension As a result of the sorption, a fraction of the MWCNTs were stably suspended in the PSS solutions after the sonication. The MWCNTs remaining in the supernatants of the MWCNT-PSS mixtures after centrifugation at 3000 g for 15 min following the sonication were considered to be stably suspended or stabilized (Lin and Xing, 2008b; Lin et al., 2012). The suspending kinetics of the MWCNTs in the SS and PSS solutions are presented in Fig. 3. No stabilized MWCNTs were detected in the SS solution and ultra-pure water even after sonication for 3 h. The amount of stabilized MWCNTs increased with the sonication time in the PSS solutions and then leveled off after about 2 h, while no significant difference in the suspending kinetics among the PSSs was observed. The MWCNT stabilization in the PSS solution could be due to the increased electrostatic and steric repulsions from the PSS sorption on the MWCNT surfaces, as discussed later. However, the MWCNTs would not be stably suspended if there were no sonication to disperse the MWCNT aggregates. The dispersing or disaggregating of the MWCNT aggregates increases with increasing sonication energy (i.e., sonication time in this study) (Yang et al., 2013), which, together with the increasing PSS sorption, contributed to the increasing MWCNT stabilization in the PSS solutions with increasing sonication time. The equilibrium MWCNT stabilizations in the SS and PSS solutions are shown in Fig. 4A. The MWCNTs could not be stabilized in the SS solutions with concentrations up to 40 mg L1, while a threestage stabilization was observed in the PSS solutions with an increasing concentration of the PSSs. Below 1.5 mg L1, the PSSs could not stabilize the MWCNTs; the MWCNT stabilization sharply increased in the PSS solutions with increasing PSS concentration from 1.5 to 2.5 mg L1 and then leveled off above 2.5 mg L1. Similar to the suspending kinetics, there was no marked difference in the equilibrium MWCNT stabilization in the five PSS solutions. However, when the PSS concentration was expressed as the molecular concentration, a lower concentration of PSS with a higher molecular weight was needed to obtain the maximum MWCNT stabilization (Fig. S5). This implies that polymers with more monomers per molecule have a greater capability per molecule to stabilize CNT suspensions, but the capabilities will be comparable if they are normalized by the total monomer number, i.e., the capability to stabilize CNTs per monomer of the polymers with a different number of repeat units is similar. The MWCNT stabilization in the PSS solutions could be due to the combined effects of increased electrostatic repulsion and steric repulsion as a result of the PSS sorption. The zeta potential of the MWCNTs became more negative, decreasing to less than 30 mV and 50 mV in the SS and PSS solutions with increasing SS and PSS concentrations, respectively (Fig. 4B). The adsorption of SS could offer strong electrostatic repulsion between the MWCNTs by lowering the zeta potential below 30 mV; however, it could not effectively disperse and stabilize the MWCNTs owing to the lack of sufficient steric repulsion. It has been observed in our previous studies that MWCNTs could not be stabilized in water even when the zeta potential of the MWCNTs was lowered to around 50 mV by increasing the solution pH to 11 (Lin et al., 2009), while MWCNTs with a zeta potential around 30 mV could be stably suspended in the presence of a nonionic surfactant (Lin et al., 2010a). SS molecules could bind to the MWCNT surfaces with a planar configuration, while a fraction of the SS monomers in the MWCNT-adsorbed PSS molecules must extend toward the solution. Thus, the adsorbed PSSs could offer steric repulsion between the MWCNTs in addition to the increased electrostatic repulsion and effectively disperse and stabilize the MWCNTs. Low concentrations (
