Competitive Sorption of Pyrene, Phenanthrene, and Naphthalene on

In this study, competitive sorption of pyrene, phenanthrene, and naphthalene on a multiwalled CNT material was investigated. All isotherms in single-,...
1 downloads 6 Views 288KB Size
Download PDF
Environ. Sci. Technol. 2006, 40, 5804-5810

Competitive Sorption of Pyrene, Phenanthrene, and Naphthalene on Multiwalled Carbon Nanotubes K U N Y A N G , †,‡ X I L O N G W A N G , † L I Z H O N G Z H U , ‡ A N D B A O S H A N X I N G * ,† Department of Plant, Soil and Insect Sciences, University of Massachusetts, Amherst, Massachusetts 01003, and Department of Environmental Science, Zhejiang University, Hangzhou, Zhejiang 310028, China

Knowledge of toxic chemical sorption by carbon nanotubes (CNTs) is critical for environmental application of CNTs as superior sorbents and for environmental risk assessment of both CNTs and toxic chemicals. Single-solute sorption results were reported in the literature, however, they cannot be used for predicting pollutant sorption by CNTs in wastewater and natural water systems where multiple organic contaminants are present. In this study, competitive sorption of pyrene, phenanthrene, and naphthalene on a multiwalled CNT material was investigated. All isotherms in single-, bi-, and tri-solute systems were fitted well by the Dubinin-Ashtakhov (DA) model. The isotherm of a given primary solute changed from being significantly nonlinear to nearly linear when competitors were added. The observed competitive sorption depended on the relative equilibrium concentrations of both primary and cosolutes. Significant competition was observed at relatively low concentrations of primary solute and high concentrations of competitors, while competition was much weaker in the case of relatively high concentrations of primary solute and low competitor concentrations. When the relative concentration of primary solute (Ce/Cs) approached 1, competition by other solutes seemed to disappear. Sorption and competition of three polycyclic aromatic hydrocarbons (PAHs) on CNTs could not be explained with either pore-filling or partition-adsorption mechanisms. A Polanyi-based surface adsorption mechanism was proposed to interpret the observed sorption and competition.

Introduction Carbon nanotubes (CNTs) are novel manufactured nanomaterials, having widespread potential applications such as drug delivery, optical devices, quantum computing, and energy conversion (1). Due to their large surface area and high reactivity, potential environmental applications of CNTs as superior sorbents have been suggested for removal of toxic chemicals from water and gases (2-8). However, nanoparticles can enter cells (9, 10) and cross the blood-brain barrier (11, 12). Therefore, recent studies (13-18) on CNTs highlight their potential risks to human beings and ecosystems once they are released to the environment, comparable to ultrafine particles from fuel burning and vehicle sources (19). CNTs * Corresponding author phone: (413)545-5212; fax: (413)545-3958; e-mail: [email protected]. † University of Massachusetts. ‡ Zhejiang University. 5804

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 18, 2006

are cytotoxic (13-15), and they can initiate a cell irritation response (16) and induce granulomas in rat lungs (17, 18). Adsorption of toxic chemicals on CNTs may enhance the toxicity of CNTs (20, 21) and affect the fate and transport of toxic pollutants in the environment (22), thus making the risk assessment of nanoparticles complicated (23). Therefore, knowledge of the sorption behavior of toxic chemicals by CNTs is critical for application of CNTs and environmental risk assessment of both CNTs and toxic chemicals. To date, very few studies have dealt with sorption of toxic organics by CNTs in the aqueous environment, and only single-solute sorption was examined (7, 8, 24). However, single-solute sorption data may not be useful for predicting pollutant sorption by CNTs when multiple organic contaminants are present. Competitive sorption and displacement of multiple organic contaminants on CNTs would likely be observed in multi-solute systems, as observed on activated carbon and natural geosorbents (25-30). Competitive sorption would decrease the removal efficiency of pollutants by CNTs from wastewater, and may result in serious alteration in the health and environmental risks of both CNTs and organic contaminants, relative to that predicted from noncompetitive models. Therefore, it is important to understand the alteration caused by competitive sorption and the underlying mechanisms. Two possible types of adsorption sites were proposed for carbonaceous (geo-) sorbents (22): (i) adsorption on external surfaces, and (ii) adsorption in nanopores. Though adsorption of polycyclic aromatic hydrocarbons (PAHs) by CNTs in our recent work (24) was described well by the Polanyi theory, the exact adsorption sites and mechanisms are unclear because (i) Polanyi theory is applicable for both pore-filling and flat surfaces (31); and (ii) isotherm parameters of PAHs for various CNTs were correlated well with both surface areas and pore volumes (24). Results of competitive sorption would help understand the adsorption sites and underlying mechanisms. Therefore, competitive sorption of pyrene, phenanthrene, and naphthalene on a multiwalled carbon nanotube material (MWCNT15) was investigated in this study. MWCNT15 was used in this study mainly due to the large production capacity of multiwalled CNTs. PAHs were selected due to their notable concentrations in the environment, high toxicity, and persistence (22).

Materials and Methods Chemicals. 14C-radiolabeled pyrene (with a specific activity of 55 µCi/µmol), phenanthrene (8.2 µCi/µmol), and naphthalene (31.3 µCi/µmol) were purchased from Sigma-Aldrich Chemical Co. Unlabeled pyrene (99+%), phenanthrene (98+%), and naphthalene (99+%) were also obtained from the same company and used without further purification. Selected properties of these chemicals are listed in Table S1 (Supporting Information). Carbon Nanomaterials. MWCNT15 was from the same batch as the one used in our previous study (24), which has a purity of 95+%, elemental carbon content of 99.8%, length of 10-50 µm, outer diameter of 8-15 nm, and inner diameter of 3-5 nm. Its surface area, mesopore and micropore volumes are approximately 174 m2/g, 0.597 cm3/g, and 0.068 cm3/g, respectively (24). Sorption Experiments for PAHs by MWCNT15. All singlesolute sorption isotherms by MWCNT15 were obtained using a batch equilibration technique at 25 ( 1, 40 ( 1, and 55 ( 1 °C. Detailed procedures were described elsewhere (24). Briefly, 1 mg of MWCNT15 and 40 mL of background solution 10.1021/es061081n CCC: $33.50

 2006 American Chemical Society Published on Web 08/18/2006

FIGURE 1. Single-point distribution coefficients (Kd) of phenanthrene at its initial phenanthrene concentrations (C0) of 0.25 mg/L (A) and 0.89 mg/L (B) as a function of relative equilibrium concentrations (Ce/Cs) of pyrene and naphthalene. The inserts are the same data in linear scale. in 40 mL screw cap vials were used for phenanthrene and pyrene, while 10 mg of MWCNT15 and 40 mL solution were for naphthalene. Mixtures of 14C-labeled and unlabeled compounds in methanol were injected into the solutions. After 5-day equilibration, centrifugation (1000g for 20 min at 25 ( 1 °C) or 24 h vertical settlement (at 40 ( 1 or 55 ( 1 °C) was used to separate MWCNT15 from water. Then, 1 mL of supernatant was added to 8 mL of Scintiverse cocktail (Fisher Scientific, PA) for liquid scintillation counting (Bechman [Fullerton, CA] LS6500). The mass loss of solutes in the experiments was less than 4% of the initial concentrations. Therefore, sorbed solute by MWCNT15 was calculated by mass difference. All competitive sorption experiments were conducted at 25 ( 1 °C. Procedures of competitive experiments were the same as single-solute sorption experiments except for the following: (i) mixtures of 14C-labeled and unlabeled primary solute as well as unlabeled competing solutes were injected into the suspensions; (ii) single-point phenanthrene sorption experiments were performed at two initial phenanthrene concentrations (0.25 and 0.89 mg/L) with various concentrations of competitor (pyrene or naphthalene); for other competitive experiments, a fixed dose of competing solutes was added, calculated from the saturated sorption capacity of their single-solute isotherms and water solubility to control the equilibrium concentrations equal to water solubility; and (iii) centrifugation (1000g for 20 min) was used to separate MWCNT15 from water. Independent tests with crystalline PAHs were conducted to demonstrate that pyrene, phenanthrene, and naphthalene did not affect each other’s solubility. Sorption Experiments for Phenanthrene by Solid Pyrene Particles. Two forms of solid pyrene were investigated in 8 mL vials at 25 ( 1 °C: big and small pyrene particles. A 40 mg portion of crystalline pyrene (big particles with diameter ∼1 mm) was weighed and mixed with 8 mL of background solution containing various phenanthrene concentrations. For small pyrene particles (with diameter ∼0.8 µm), 100 µL of unlabeled pyrene in methanol solution with concentration of 6000 mg/L was injected into 8 mL of background solution. The final solid pyrene mass was calculated from the difference between the added pyrene dose (40 mg for big particles, and 100 µL × 6000 mg/L ) 0.6 mg for small particles) and its water solubility (0.132 mg/L × 8 mL ) 0.00106 mg). Other procedures were the same as the single-solute sorption experiments. Sorption Models and Regression Analysis. The Polanyi theory is useful for describing the sorption data of PAHs by CNTs. However, molar volume (Vs) alone as an abscissa

scaling factor failed to obtain a single correlation curve (24). Hence, the Dubinin-Ashtakhov (DA) model (32) was employed to fit the sorption data.

log qe ) logQ0 - ( /E)b where qe [mg/g] is the equilibrium sorbed concentration; Q0 [mg/g] is the saturated sorbed capacity;  [kJ/mol],  ) RTln(Cs/Ce), is the effective adsorption potential; Ce [mg/L] is the equilibrium aqueous concentration; Cs [mg/L] is the water solubility; R [8.314 × 10-3 kJ/(mol K)] is the universal gas constant; and T [K] is the absolute temperature; E [kJ/mol] is the ‘‘correlating divisor”; and b is the fitting parameter. The Freundlich model was also used to fit the sorption data.

Log qe ) logKf + n logCe where Kf [(mg/g)/(mg/L)n] is the Freundlich affinity coefficient; and n is the Freundlich exponential coefficient. For dilute solute concentrations, the Henry’s law, showing a linear isotherm, is approached. In contrast, for high Ce/Cs concentrations, the isotherm could be described by the Freundlich model. All model parameters with their standard errors were determined by a commercial software program (SPSS10.0). Mean-weighted-square-errors (MWSE), equal to 1/v[(qmeasured - qmodeled)2/q2measured], and correlation coefficients (r2) were used to evaluate the goodness of fit; where v is the degree of freedom (v ) N - 2 for Freundlich model, v ) N - 3 for DA model), N is the number of experimental data points, qmeasured is the measured equilibrium sorbed concentration, and qmodeled is the estimated equilibrium sorbed concentration by respective models.

Results and Discussion Single-Point Competitive Sorption of Phenanthrene by Pyrene or Naphthalene. Figure 1 shows the competition of pyrene and naphthalene with phenanthrene at two initial phenanthrene concentrations: 0.25 and 0.89 mg/L. Competition depended largely on equilibrium competitor concentrations (Figure 1). Single-point distribution coefficients (Kd ) qe/Ce) of phenanthrene at 0.25 mg/L decreased from ∼2650 L/g to 48.3 L/g and to 212 L/g as pyrene and naphthalene relative equilibrium concentrations (Ce/Cs) increased from ∼0.002 to 0.884 and to 0.804 (Figure 1A), respectively. Phenanthrene Kd at 0.89 mg/L decreased also from ∼373 L/g to 50.8 L/g and to 102 L/g as pyrene and VOL. 40, NO. 18, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5805

FIGURE 2. Sorption isotherms of PAHs on MWCNT15 without or with competitors at Ce/Cs ) 1: (A) pyrene alone and with naphthalene or phenanthrene as competitor; (B) phenanthrene alone and with naphthalene, pyrene, or their mixture as competitors; and (C) naphthalene alone and with pyrene or phenanthrene as competitor. All isotherms were normalized by the primary solute solubility. Solid lines (;;), the isotherms fitted by DA model; long dashed lines (- - -), the isotherms fitted by Freundlich model. Short dashed lines (-----) represent the limits of competition, computed using DA model with E ) 5.71 kJ/mol, b )1, and Q0 of single-solute isotherms. Phen and Naph represent phenanthrene and naphthalene, respectively. naphthalene Ce/Cs increased from ∼0.002 to 0.930 and to 0.856 (Figure 1B), respectively. According to the trend shown in Figure 1, maximum competition of phenanthrene by pyrene and naphthalene would be at the competitor Ce/Cs ) 1. The competition also depended on the primary solute (phenanthrene) concentration (Figure 1), being greater at low phenanthrene concentration (0.25 mg/L) than at high phenanthrene concentration (0.89 mg/L). Sorption Isotherms in Single-, Bi-, and Tri-Solute Systems. Single-solute isotherms and multi-solute isotherms for pyrene, phenanthrene, and naphthalene by MWCNT15 are presented in Figure 2. Isotherm data were fitted by both DA and Freundlich models (Figure 2). Respective fitting parameters are given in Table 1. The DA model had good fits for both single-solute and multi-solute isotherms (Figure 2), supported by the low MWSE and high r2 values (Table 1). The Freundlich model had good fits for naphthalene and multi-solute isotherms, but it failed to fit the single-solute isotherms of pyrene and phenanthrene (Figure 2 and Table 1). Since the Freundlich model is a special form of the DA model (b ) 1), it is not surprising that the Freundlich model 5806

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 18, 2006

is applicable for a few cases. Linear model is also a special form of the DA model, where b ) 1 and E ) RTln10 ≈ 5.71 kJ/mol. For the discussion below, we refer the DA fitted isotherms with b > 1 as a “typical Polanyi isotherm”, showing nonlinear isotherms even in a log-log scale plot. Consequently, single-solute isotherms of pyrene and phenanthrene in Figure 2 are “typical Polanyi isotherms”. All single-solute isotherms are nonlinear (Figure 2 and Table 1), indicating the heterogeneous energy distribution of sorption sites on MWCNT15. Single-solute isotherms of phenanthrene and pyrene had the same sorption capacity (DA fitted logQ0 values are 1.62 ( 0.01 for phenanthrene and 1.63 ( 0.02 for pyrene), which was lower than that of naphthalene (logQ0 ) 1.87 ( 0.02) (Table 1). However, phenanthrene and pyrene had higher DA fitted E (E ) 17.3 ( 0.1 kJ/mol for phenanthrene and 14.3 ( 0.2 kJ/mol for pyrene) and b values (b ) 2.10 ( 0.06 for phenanthrene and 2.48 ( 0.13 for pyrene) than naphthalene (E ) 11.4 ( 0.3 kJ/mol and b ) 0.968 ( 0.03) (Table 1). The difference of DA fitted E and b values for pyrene, phenanthrene, and naphthalene suggests that available sorption sites for MWCNT15

TABLE 1. Results of Freundlich and DA Model Fits to Sorption Data of Pyrene, Phenanthrene, and Naphthalene on MWCNT15a primary sorbate pyrene phenanthrene

naphthalene

primary sorbate pyrene phenanthrene

naphthalene

with competitors none naphthalene phenanthrene none naphthalene pyrene phenanthrene/naphthalene none pyrene phenanthrene

with competitors none naphthalene phenanthrene none naphthalene pyrene phenanthrene/naphthalene none pyrene phenanthrene

Freundlich Model logKf 2.37 ( 0.11d 2.43 ( 0.45 2.51 ( 0.03 1.94 ( 0.05 1.93 ( 0.03 1.66 ( 0.02 1.58 ( 0.02 1.11 ( 0.01 0.945 ( 0.013 0.271 ( 0.017 Dubinin-Ashtakhov Model (DA) logQ0 E 1.63 ( 0.02 1.71 ( 0.02 1.76 ( 0.03 1.62 ( 0.01 1.62 ( 0.04 1.65 ( 0.04 1.59 ( 0.05 1.87 ( 0.02 1.71 ( 0.04 1.61 ( 0.07

14.3 ( 0.2 9.72 ( 0.19 6.18 ( 0.25 17.3 ( 0.1 9.58 ( 0.36 6.73 ( 0.34 6.21 ( 0.40 11.4 ( 0.3 11.1 ( 0.4 6.49 ( 0.51

n

MWSEb

r2

Nc

0.498 ( 0.044 0.672 ( 0.023 0.890 ( 0.017 0.426 ( 0.024 0.747 ( 0.020 0.857 ( 0.015 0.911 ( 0.019 0.485 ( 0.004 0.591 ( 0.015 0.905 ( 0.021

0.228 0.0325 0.0100 0.0864 0.0255 0.0102 0.0163 0.00311 0.0207 0.0306

0.866 0.975 0.992 0.936 0.984 0.994 0.991 0.999 0.987 0.988

22 24 24 24 24 24 24 24 22 24

b

MWSE

r2

N

2.48 ( 0.13 1.41 ( 0.07 0.938 ( 0.052 2.10 ( 0.06 1.35 ( 0.08 1.01 ( 0.06 0.989 ( 0.069 0.968 ( 0.025 1.23 ( 0.08 1.03 ( 0.08

0.0168 0.00984 0.0101 0.00343 0.0122 0.0107 0.0171 0.00297 0.0159 0.0313

0.988 0.972 0.992 0.997 0.992 0.994 0.992 0.999 0.991 0.988

22 24 24 24 24 24 24 24 22 24

a All estimated parameter values and their standard errors were determined by a commercial software program (SPSS10.0); K unit is (mg/ f g)/(mg/L)n; Q0 unit is mg/g; and E unit is kJ/mol. b MWSE is mean-weighted-square-errors, equal to 1/v[(qmeasured - qmodeled)2/q2measured], where v is c d the amount of freedom; v ) N - 2 for Freundlich model and v ) N - 3 for DA model. N ) number of observations. Mean ( standard deviation.

vary with adsorbates. The adsorbed volume capacities of naphthalene, phenanthrene, and pyrene, calculated from their mass capacities (Q0) and respective solid-phase density (Table S1), were about 0.074 , 0.040, and 0.033 cm3/g, respectively. This order is negatively related to molecular size: 126.9 for naphthalene, 169.5 for phenanthrene, and 186.0 Å3 for pyrene. Figure 2 shows strong competition between pyrene, phenanthrene, and naphthalene. Since the equilibrium concentrations of competitors were controlled at their water solubility, the isotherms in Figure 2 represent the maximum competition of the competing solute. In bi-solute systems, the more similar the molecular structure and physicochemical properties between the competing and primary solutes, the greater the competition: phenanthrene, whose molar volume and other physicochemical properties such as water solubility are between pyrene and naphthalene (Table S1), had greater competition with pyrene than naphthalene (Figure 2A) or than naphthalene with pyrene (Figure 2C). Compared to single-solute isotherms, primary-solute isotherms with competing solutes changed toward linearity, supported either by Freundlich fitted n f 1 or by DA fitted E f 5.71 kJ/mol and b f 1 (Table 1). For example, phenanthrene isotherm changed from a “typical Polanyi isotherm” (DA fitted b > 1) for single solute to a Freundlich curve (DA fitted b f 1) with pyrene as a competitor and then to nearly linear isotherm (DA fitted E f 5.71 kJ/mol and b f 1) with the mixtures of naphthalene and pyrene as competitors (Figure 2B and Table 1). Thus, the phenanthrene isotherm with naphthalene and pyrene in Figure 2B was highly linear (qe ) 41.7Ce, r2 ) 0.985, where the unit of the sorption coefficients is L/g). Linear isotherms (E ) 5.71 kJ/ mol and b )1) seem to be the limit of competition (represented by short dashed lines in Figure 2): no more sorbed solute molecules could be competed out by others when these linear isotherms were reached. Figure 2 also shows concentration-dependent competition: significant competition at low primary-solute con-

centrations, while insignificant competition at high concentrations. Figure S1 displays this concentration-dependent competition more clearly for phenanthrene as the primary solute. Percent reduction (R, %) of phenanthrene was calculated by the equation, Ri ) 100 × (qisingle - qimulti)/qisingle, where Ri, qisingle, and qimulti are the percent reduction, sorbed amount in single-solute, and multi-solute systems at phenanthrene concentration of i, respectively. Percent reductions of sorbed phenanthrene by naphthalene, pyrene, and the mixture of naphthalene and pyrene decreased from ∼100% to ∼0 when relative phenanthrene concentrations approached 1 (Figure S1), where the competitive effect disappeared. DA model fitted log Q0 (Table 1) for phenanthrene remained practically constant, slightly increased for pyrene, and slightly decreased for naphthalene (Table 1). The small changes of log Q0 for pyrene and naphthalene could be negligible if experimental errors were considered. Possible Mechanisms for Sorption and Competition. The good fit of the DA model for all sorption isotherms shows that the Polanyi theory is useful to describe the single-solute and multi-solute sorption of PAHs by CNTs, as indicated by our previous study (24). The Polanyi theory has been related with sorption as micropore filling (33). For a given undeformable sorbent, pore filling is to give a limiting volume of the total available pores for sorption at zero sorption potential ( ) 0) (33). At Ce/Cs ) 1 ( ) 0) of all solutes, lack of competitive sorption among pyrene, phenanthrene, and naphthalene (Figure 2 and Table 1) indicates that the total volume of sorbed solutes in tri-solute systems (0.147 cm3/g) is much higher than that in bi-solute or single-solute systems, and also higher than the micropore volume of MWCNT15 (0.068 cm3/g). This is contradictory with the pore-filling mechanism for un-deformable sorbents. Though pore-filling mechanism accompanied with possible pore swelling (25, 34) may explain varied adsorption volume, it is difficult to interpret the observed concentration-dependent competition: significant competition at low primary-solute concentrations (Figures 1 and 2) implies that a lot of pores exist VOL. 40, NO. 18, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5807

FIGURE 3. Schematic model of PAHs on surface of carbon nanotubes in single-, bi-, and tri-solute systems: O solute type 1; b solute type 2; X solute type 3. in MWCNT15, which was available either for phenanthrene or pyrene as well as naphthalene; in contrast, the absence of competition at Ce/Cs ) 1 of any solute implies that the available sorption pores would be completely different for pyrene, phenanthrene, and naphthalene. Therefore, pore filling seems not to be the dominant mechanism for sorption of PAHs by MWCNT15. Polanyi theory, however, is applicable for either porefilling or flat surface adsorption (31). Furthermore, sorption parameters of single-solute isotherms for PAHs by various carbon nanomaterials were correlated with the nanomaterial’s surface area (24). It is also observed in our previous study that available adsorption spaces of carbon nanotubes were the cylindrical external surfaces and not the inner cavities or inter-wall spaces (35). In this case, therefore, we think that surface adsorption is a dominant mechanism for sorption of PAHs by MWCNT15 rather than pore filling. A schematic model for this surface sorption mechanism is shown in Figure 3, assuming that (i) potential energy of sorption sites on solute-coated sorbent surface is lower and more homogeneous than that on un-coated MWCNT15 surface; (ii) sorbed solute molecules on the sorbent surface have attractive forces for solute molecules of another type; solute-coated sorbent surface can be available for sorption of other solutes; and (iii) for a given solute, the maximum adsorption capacity depends on the sorbent surface area in single- and multi-solute systems. Experiments of phenanthrene sorption by solid pyrene particles were conducted in this study to examine the first two assumptions, showing linear phenanthrene isotherms (Figure 4). The linear sorption coefficient by small pyrene particles (25.5 L/g) was 330 times higher than that by big pyrene particles (0.0771 L/g). This large difference of sorption between small and big pyrene particles indicates surface adsorption rather than partitioning. Based on pyrene particle size, the theoretically calculated surface area of small particles (with diameter ∼0.8 µm) is 1250 times higher than that of big particles (with diameter ∼1 mm), showing more available surfaces for phenanthrene adsorption. Significant sorption of phenanthrene by pyrene particles (Figure 4) means molecule-molecule attraction between different solutes, consistent with assumption ii. The linear isotherms (Figure 4) suggest more homogeneous sorption sites on the surface of pyrene particles (a pure, uniform material) than that of MWCNT15 with a nonlinear phenanthrene isotherm (Figure 2B and Table 1). This is consistent with assumption i. At a given concentration, the higher sorption coefficients of phenanthrene by MWCNT15 (Figure 2B) than pyrene par5808

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 18, 2006

FIGURE 4. Sorption isotherms of phenanthrene by two forms of solid pyrene particles: large and small. ticles (Figure 4) implies the higher energy of sorption sites on MWCNT15 surface, in agreement with assumption i. The last assumption given here is based on the fact that singlesolute isotherm parameters for PAHs by various carbon nanomaterials are correlated with nanomaterial’s surface area (24), and the nearly constant logQ0 in multi-solute systems (Table 1). However, this assumption was only invoked to explain experimental data in this work; it needs further investigations and tests. Due to phenanthrene sorption by solid pyrene (Figure 4), it is necessary to recheck whether the lack of competition between phenanthrene and pyrene at their Ce/Cs ) 1 was real or induced by the possible sorption of phenanthrene on pyrene solid particles. Since the added dose of pyrene was the sum of its saturated sorption capacity (Q0) of singlesolute isotherm and water solubility, the maximum amount of pyrene, possibly as small solid particles (assuming no sorption of pyrene by MWCNT15), is Q0 (101.63 ) 42.7 mg/g, in Table 1). Therefore, one can calculate the enhanced phenanthrene sorption coefficient by small pyrene particles is 1.09 L/g ()25.5 L/g × 42.7 mg/g). At Ce/Cs ) 1 of phenanthrene, the maximum sorbed amount of phenanthrene on MWCNT15 enhanced by small pyrene particles is 1.09 mg/g ()1.09 L/g × 1.00 mg/L). Comparing with the sorbed amount of phenanthrene by MWCNT15 at Ce/Cs )

1 (101.65 ) 45 mg/g, in Table 1), the induced phenanthrene sorption by small pyrene particles is negligible (∼2%), thus, lack of competition between phenanthrene and pyrene at their Ce/Cs ) 1 was not an artifact, caused by sorption on the solid of competitors. The surface-dominated sorption mechanism (Figure 3) can be used to interpret the sorption and competition data observed in this study. Based on the last two assumptions, one can conclude that (i) sorption is monolayer for singlesolute system but could be multilayers in multi-solute systems and the maximum number of sorption layers on the sorbent surface depends on the number of solute species in the system (e.g., three layers are corresponding to tri-solute system, shown in Figure 3C); and (ii) sorption for a single solute as well as any solute in bi- and tri-solute systems is limited by available surface area (Figures 3B and 3C), thus, the saturated sorbed amounts of a given solute with or without other solutes are relatively invariant. The greater competition of primarysolute at relatively low concentrations by competing solute at high concentrations was attributed to the heterogeneous energy distribution of surface sorption sites for MWCNT15 (assumption i), where the high-energy sites are largely held by the competing solute because of its high concentrations (Figures 3B and 3C). However, a solute-coated sorbent surface is more homogeneous than an uncoated sorbent surface (assumption i), thus, it is not surprising that the sorption isotherm of the primary solute became more linear in the presence of competing solutes (Figure 2 and Table 1). Enhanced isotherm linearity of primary solute by competing solutes has also been observed for organic chemicals including PAHs by soils and sediments, and widely explained by a partition-adsorption model (25, 27-29, 36). The partition-adsorption model is based on assumptions that natural organic matter (NOM) in soils and sediments consists of two types of domains: a glassy (or “hard”) and a rubbery (or “soft”) NOM domain (27-29). The glassy state NOM has low O/C atomic ratio and high specific surfaces, while the rubbery state NOM has high O/C ratio and low specific surfaces (25, 27, 36-38). If there are partition and adsorption for MWCNT15, the linear phenanthrene isotherm with the mixture of pyrene and naphthalene as competitors would represent the partitioning contribution (25, 27-29, 36). Consequently, the difference between total sorption (singlesolute isotherm) and partitioning (linear isotherm in tri-solute system) would be the adsorption contribution. According to the partition-adsorption model, adsorption contribution will reach a constant maximum value (25, 27-29, 36) with increasing primary solute concentration. Figure S2 shows the concentration dependence of phenanthrene adsorption contribution (qad), calculated by the equation qiad ) qisingle qitri, where qisingle and qitri are the sorbed amount in singlesolute isotherm and in tri-solute isotherm, respectively, at a given phenanthrene concentration i. The adsorption contribution decreased from 29 mg/g to ∼0 as relative phenanthrene concentrations (Ce/Cs) increased from 0.15 to 1 (Figure S2), which is inconsistent with the partitionadsorption model with a constant adsorption maximum. The calculated maximum phenanthrene adsorption (29 mg/g) accounts for 70% of the total phenanthrene sorption at Ce/Cs )1 (Q0 ) 101.62 ) 42 mg/g, Table 1), indicating that adsorption contribution could not be negligible and also the decrease in adsorption contribution could not be attributed to the experimental error. Therefore, the partition-adsorption model could not be used to explain the sorption and competition data of PAHs by MWCNT15. Absence of a rubbery state domain (partition medium) may be responsible for this inapplicability of the partition-adsorption model because carbon nanomaterials are predominated by crystalline, graphitic-type carbon. The graphite and elemental carbon contents of MWCNT15 used in this study are higher

than 95% and 99.8%, respectively, showing O/C atomic ratio f 0. Assessment of Competitive Sorption Models. Ideal adsorbed solution theory (IAST), a thermodynamic model derived from the Gibbs equation, has been applied successfully to simulate the competitive sorption of solutes from their single-solute isotherms on activated carbon (26) and soils (27, 28). The two main assumptions of IAST (33) are (i) the effective sorption potential of sorbent to each solute is determined by the total adsorption volume, independent of whether adsorption is from single-solute systems or multisolute systems, i.e., each solute has access to all sorption sites and complete exclusion of one solute by another saturated solute with greater adsorption is expected; and (ii) the adsorbed mixed components of sorbates are completely miscible and will behave approximately as an ideal mixture, so that the partial pressure of the ith component at the interface is (xCe)i, where xi is the mole fraction of the ith component in solution and Σxi ) 1. The IAST for predicting the multi-solute sorption under Polanyi theory was described in detail by Manes (33) and Xia and Ball (28). The basic IAST equations based on the DA model are given in the Supporting Information. IAST-predicted competitive sorption of phenanthrene by pyrene is given in Figure S3 as an example, showing that IAST greatly overestimated the competition and, thus, it is incapable of simulating competition of PAHs on MWCNT15. Two plausible reasons for the failure of IAST prediction are (i) no complete exclusion of one solute by another saturated solute with greater adsorption and no common limiting adsorbate volume were observed in this study (Figure 2); and (ii) sorbed adsorbates may form a solid phase in the adsorption space rather than ideal mixed solutions because solid adsorbates were investigated in this work (33). Figure S4 shows the temperature-dependent isotherms of pyrene, phenanthrene, and naphthalene by MWCNT15. The coincidence of solute solubility-normalized isotherms at 25, 40, and 55 °C (inserted plots in Figure S4) provides the evidence that heats of adsorption are essentially the same as that of dissolution (in Table S1) (36), implying that sorbed PAHs may form solid phases. Assuming adsorbed adsorbates will form solid phase in the adsorption space, the Polanyi-based immiscible model was developed to simulate the competitive sorption of solid adsorbates from water to activated carbon (33) and soil (28). The immiscible model, however, still assumes the complete exclusion and a common limiting adsorbate volume, the same as with assumption i of IAST. This assumption is inconsistent with the observed results in Figure 2, i.e., no complete exclusion and no common limiting adsorbate volume. Therefore, the immiscible model was also incapable of describing the competitive data observed in this study. In conclusion, the application of existing sorption models based on the Polanyi theory for describing competitive sorption of PAHs on carbon nanotubes was not very successful in this study. This is because these models often assume micropore filling (33) along with a common limiting adsorbate volume. However, the data in this work indicate that the Polanyi theory can be also applicable for flat surface adsorption. Therefore, new empirical or theoretical models based on a surface-dominated sorption mechanism need to be developed and tested with additional experimental data.

Acknowledgments This work was supported in part by the Massachusetts Agricultural Experiment Station (MAS 00090) and the National Natural Science Foundation of China (40503015). VOL. 40, NO. 18, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5809

Supporting Information Available Four figures, one table, and the basic IAST equations. This material is available free of charge via the Internet at http:// pubs.acs.org.

Literature Cited (1) Warheit, D. B. Nanoparticles: Health impacts? Mater. Today 2004, 7, 32-35. (2) Liu, Z.; Gao, Y.; Bando, Y. Highly effective metal vapour absorbents based on carbon nanotubes. Appl. Phys. Lett. 2002, 81, 4844-4846. (3) Long, R. Q.; Yang, R. T. Carbon nanotubes as superior sorbent for dioxin removal. J. Am. Chem. Soc. 2001, 123, 2058-2059. (4) Kim, C.; Chio, Y. S.; Lee, S. M.; Park, J. T.; Kim, B.; Lee, Y. H. The effect of gas adsorption on the field emission mechanism of carbon nanotubes. J. Am. Chem. Soc. 2002, 124, 9906-9911. (5) Li, Y.; Wang, S.; Luan, Z.; Ding, J.; Xu, C.; Wu, D. Adsorption of cadmium(II) from aqueous solution by surface oxidized carbon nanotubes. Carbon 2003, 41, 1057-1062. (6) Wang, X.; Chen, G.; Hu, W.; Ding A.; Xu, D.; Zhou, X. Sorption of 243Am(III) to Multiwall Carbon Nanotubes. Environ. Sci. Technol. 2005, 39, 2856-2860. (7) Peng, X.; Li, Y.; Luan, Z.; Di, Z.; Wang, H.; Tian, B.; Jia, Z. Adsorption of 1,2-dichlorobenzene from water to carbon nanotubes. Chem. Phys. Lett. 2003, 376, 154-158. (8) Lu, C.; Chung, Y.-L.; Chang, K.-F. Adsorption of trihalomethanes from water with carbon nanotubes. Water Res. 2005, 39, 11831189. (9) Lu, Q.; Moore, J. M.; Huang, G.; Mount, A. S.; Rao, A. M.; Larcom, L. L.; Ke, P. C. RNA polymer translocation with single-walled carbon nanotubes. Nano Lett. 2004, 4, 2473-2477. (10) Kam, N. W. S.; Jessop, T. C.; Wender, P. A.; Dai, H. J.; Nanotube molecular transporters: Internalization of carbon nanotubeprotein conjugates into mammalian cells. J. Am. Chem. Soc. 2004, 126, 6850-6851. (11) Lockman, P. R.; Oyewumi, M. O.; Koziara, J. M.; Roder, K. E.; Mumper, R. J.; Allen, D. D. Brain uptake of thiamine-coated nanoparticles. J. Controlled Release 2003, 93, 271-282. (12) Brigger, I.; Morizet, J.; Aubert, G.; Chacun, H.; Terrier-Lacombe, M. J.; Couvreur, P.; Vassal, G. Poly(ethyleneglycol)-coated hexadecylcyanoacrylate nanospheres display a combined effect for brain tumor targeting. J. Pharmacol. Exp. Ther. 2002, 303, 928-936. (13) Jia, G.; Wang, H.; Yan, L.; Wang, X.; Pei, R. Yan, T.; Zhao, Y.; Guo, X. Cytotoxicity of carbon nanomaterials: Single-wall nanotube, multi-wall nanotube, and fullerene. Environ. Sci. Technol. 2005, 39, 1378-1383. (14) Shvedova, A. A.; Castranova, V.; Kisin, E. R.; Schwegler-Berry, D.; Murray, A. R.; Gandelsman, V. Z.; Maynard, A.; Baron, P. Exposure to carbon nanotube material: Assessment of nanotube cytotoxicity using human keratinocyte cells. J. Toxicol. Environ. Health, Part A 2003, 66, 1909-1926. (15) Cui, D. X.; Tian, F. R.; Ozkan, C. S.; Wang, M.; Gao, H. J. Effect of single wall carbon nanotubes on human HEK293 cells. Toxicol. Lett. 2005, 155, 73-85. (16) Monteiro-Riviere, N. A.; Nemanich, R. J.; Inman, A. O.; Wang, Y. Y. Y.; Riviere, J. E. Multi-walled carbon nanotube interactions with human epidermal keratinocytes. Toxicol. Lett. 2005, 155, 377-384. (17) Warheit, D. B.; Laurence, B. R.; Reed, K. L.; Roach, D. H.; Reynolds, G. A.; Webb, T. R. Comparative pulmonary toxicity assessment of single-wall carbon nanotubes in rats. Toxicol. Sci. 2004, 77, 117-125. (18) Lam, C. W.; James, J. T.; McCluskey, R.; Hunter, R. L. Pulmonary toxicity of single-wall carbon nanotubes in mice 7 and 90 days after intratracheal instillation. Toxicol. Sci. 2004, 77, 126-134. (19) Borm, P. J. A. Particle toxicology: From coal mining to nanotechnology. Inhalation Toxicol. 2002, 14, 311-324. (20) Oberdorster, G.; Oberdorster, E.; Oberdorster, J. Nanotoxicology: An emerging discipline evolving from studies of ultrafine particles. Environ. Health Perspect. 2005, 113, 823-839.

5810

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 18, 2006

(21) Li, N.; Sioutas, C.; Cho, A.; Schmitz, D.; Misra, C.; Sempf, J.; Wang, M. Y.; Oberley, T.; Froines, J.; Nel, A. Ultrafine particulate pollutants induce oxidative stress and mitochondrial damage. Environ. Health Perspect. 2003, 111, 455-460. (22) Cornelissen, G.; Gustafsson, O.; Bucheli, T. D.; Jonker, M. O.; Koelmans, A. A.; Van noort, P. M. Extensive sorption of organic compounds to black carbon, coal, and kerogen in sediments and soils: mechanisms and consequences for distribution, bioaccumulation, and biodegradation. Environ. Sci. Technol. 2005, 39, 6881-6895. (23) Biswas, P.; Wu, C. Y. Nanoparticles and the environment. J. Air Waste Manage. Assoc. 2005, 55, 708-746. (24) Yang, K.; Zhu, L. Z.; Xing, B. S. Adsorption of polycyclic aromatic hydrocarbons by carbon nanomaterials. Environ. Sci. Technol. 2006, 40, 1855-1861. (25) Yu, Z.; Huang, W. Competitive sorption between 17 R-ethinyl estradiol and naphthalene/ phenanthrene by sediments. Environ. Sci. Technol. 2005, 39, 4878-4885. (26) Crittenden, J. C.; Luft, P.; Hand, D. W.; Oravitz, J. L.; Loper, S. W.; Ari, M. Prediction of multicomponent adsorption equilibria using ideal adsorbed solution theory. Environ. Sci. Technol. 1985, 19, 1037-1043. (27) McGinley, P. M.; Katz, L. E.; Weber, W. J., Jr. A distribution reactivity model for sorption by soils and sediments. 2. Multicomponent systems and competitive effects. Environ. Sci. Technol. 1993, 27, 1524-1531. (28) Xia, G.; Ball, W. P. Polanyi-based models for the competitive sorption of low-polarity organic contaminants on a natural sorbent. Environ. Sci. Technol. 2000, 34, 1246-1253. (29) Xing, B. S.; Pignatello, J. J.; Gigliotti, B. Competitive sorption between atrazine and other organic compounds in soils and model sorbents. Environ. Sci. Technol. 1996, 30, 2432-2440. (30) Sander, M.; Pignatello, J. J. Characterization of charcoal adsorption sites for aromatic compounds: insight drawn from singlesolute and bi-solute competitive experiments. Environ. Sci. Technol. 2005, 39, 1606-1615. (31) Condon, J. B. Equivalency of the Dubinin-Polany equations and the QM based sorption isotherm equation. B. Simulations of heterogeneous surfaces. Microporous Mesoporous Mater. 2000, 38, 377-83. (32) Dubinin, M. M.; Astakhov, V. A. Development of concepts of volume filling of micropores in adsorption of gases and vapors by microporous adsorbents. Izv. Akad. Nauk SSSR, Ser. Khim. 1971, 1, 5-11. (33) Manes, M. Activated carbon adsorption fundaments. In Encyclopedia of Environmental Analysis and Remediation; Meyers, R.A., Ed.; John Wiley & Sons, Inc.: New York, 1998; pp 26-68. (34) Lu, Y.; Pignatello, J. J. Demonstration of the “conditioning effect” in soil organic matter in support of a pore deformation mechanism for sorption hysteresis. Environ. Sci. Technol. 2002, 36, 4553-4561. (35) Yang, K.; Xing, B. S. Desorption of polycyclic aromatic hydrocarbons from carbon nanomaterials in water. Environ. Pollut. 2006 (in press). (36) Chiou, C. T. Partition and Adsorption of Organic Contaminants in Environmental Systems; John Wiley & Sons, Inc.: New York, 2002. (37) Xing, B. S.; Pignatello, J. J. Dual-mode sorption of low-polarity compounds in glassy poly(vinyl chloride) and soil organic matter. Environ. Sci. Technol. 1997, 31, 792-799. (38) Gunasekara, A. S.; Simpson, M. J.; Xing, B. S. Identification and characterization of sorption domains in soil organic matter using structurally modified humic acids. Environ. Sci. Technol. 2003, 37, 852-858.

Received for review May 5, 2006. Revised manuscript received July 18, 2006. Accepted July 19, 2006. ES061081N