Environ. Sci. Technol. 2008, 42, 2899–2905
Influence of Surface Oxides on the Adsorption of Naphthalene onto Multiwalled Carbon Nanotubes HYUN-HEE CHO,† BILLY A. SMITH,‡ JOSHUA D. WNUK,‡ D. HOWARD FAIRBROTHER,‡ AND W I L L I A M P . B A L L * ,† Department of Geography and Environmental Engineering and Department of Chemistry, Johns Hopkins University, 3400 North Charles Street, Baltimore, Maryland 21218
Received September 19, 2007. Revised manuscript received January 13, 2008. Accepted January 22, 2008.
As greater quantities of carbon nanotubes (CNTs) enter the environment, they will have an increasingly important effect on the availability and transport of aqueous contaminants. As a consequence of purification, deliberate surface functionalization, and/or exposure to oxidizing agents after release to the environment, CNTs often contain surface oxides (i.e., oxygen containing functional groups). To probe the influence that surface oxides exert on CNT sorption properties, multiwalled CNTs (MWCNTs) with varying oxygen concentrations were studied with respect to their sorption properties toward naphthalene. For pristine (as-received) MWCNTs, the sorption capacity was intermediate between that of a natural char and a granular activated carbon. Sorption data also reveal that a linear relationship exists between the oxygen content of MWCNTs and their maximum adsorption capacity for naphthalene, with 10% surface oxygen concentration resulting in a roughly 70% decrease in maximum adsorption capacity. The relative distribution of sorption energies, as characterized by Freundlich isotherm exponents was, however, unaffected by oxidation. Thus, the data are consistent with the idea that incorporated surface oxides create polar regions that reduce the surface area available for naphthalene sorption. These results highlight the important roleofsurfacechemistryincontrollingtheenvironmentalproperties of CNTs.
Introduction A carbon nanotube (CNT) consists of one or more graphene sheets rolled into a long, thin, hollow cylinder. Typical lengths and internal diameters range from 1 to 100 µm and from 1 to 25 nm, respectively (1). The unique combination of physical and chemical properties attributed to CNTs has made them one of the most widely utilized classes of engineered nanomaterials. Some current and proposed applications using CNTs include structural composites (2), microelectronic devices (3), and flat-panel displays (4). CNTs are also being developed as sensors (5), imaging agents in gene therapy (6), vehicles for targeted drug delivery (7), hydrogen storage devices (8), and catalyst supports (9). * Corresponding author phone: (410) 516-5434; fax: (410) 5168996; e-mail:
[email protected]. † Department of Geography and Environmental Engineering. ‡ Department of Chemistry. 10.1021/es702363e CCC: $40.75
Published on Web 03/18/2008
2008 American Chemical Society
Multiwalledandsingle-walledcarbonnanotubes(MWCNTs and SWCNTs) are also effective sorbents for low-molecularweight organic chemicals due to their high surface areas and hydrophobic graphene surfaces. Several recent studies have shown CNTs to be effective sorbents for vapor-phase toxins (10–12). CNT sorption properties in water have also begun to attract research interest with respect to the uptake of hydrophobic organic chemicals (HOCs). Indeed, Li et al. (13) reported that MWCNTs are better sorbents than carbon black for the sorption of volatile HOCs from water. Similarly, Peng et al. (14) and Lu et al. (15) observed that CNTs can effectively remove 1,2-dichlorobenzene and trihalomethanes, respectively, from water. Sorption studies on SWCNTs and MWCNTs have also been conducted on aqueous solutions of polycyclic aromatic hydrocarbons (16, 17). In single-solute systems, highly nonlinear adsorption isotherms were observed (16), but isotherms were more linear in multicomponent mixtures of naphthalene, pyrene, and phenanthrene (17). Such trends are expected for HOC adsorption with heterogeneous carbonaceous sorbents (18–20). Although many sorption studies reported to date have focused on the properties of pristine (as-received or freshly synthesized) CNTs, CNT surfaces often include oxygencontaining functional groups. These can be formed during purification procedures that use strong oxidizing acids for removal of amorphous carbon and metal contaminants (21–25), by deliberate oxidization to functionalize the surface (26–29), or by incidental exposure to oxidizing agents (e.g., O3 or OH · ) after release into the environment (30, 31). Although the exact location and structure of surface oxides on carbon nanotubes is extremely difficult to determine, indirect evidence for their location has nonetheless been obtained in previous studies by spatially imaging the position of metal and metal oxide (e.g., TiO2) nanoclusters deposited on the sidewalls of carbon nanotubes (32, 33). Results from these studies indicate that surface oxides, formed by chemical oxidation processes analogous to the ones used in the present study form randomly along the sidewalls, presumably at defect sites in the graphene sheet and at any open ends of CNTs. Theory predicts and prior studies have shown that the presence of surface oxides will influence the sorption properties of black carbon (BC) materials toward both organic and inorganic contaminants. For example, surface oxidation decreases the sorption capacity of soot particles toward HOCs (34). For metal contaminants (e.g., Cd2+, Pb2+, Cu2+), oxidized CNTs (35–37) and activated carbons (38) display higher adsorption capacities than the pristine materials. Although the qualitative effect of oxidation on the sorption properties of BC materials is known, quantitative understanding of the relationships between surface oxidation and sorption properties is still lacking. The sorption properties of CNTs are of environmental interest and concern not only because CNTs are being considered as commercial sorbents, but also because sorption properties will affect the impact of CNTs in the environment. In this regard, commercial production of CNTs is rapidly growing (39) and correspondingly greater numbers of CNT particles will be entering the natural environment, where they can also affect the transport of other chemicals. For example, immobilized CNTs may lead to enhanced retention of some dissolved chemicals in porous media, whereas mobile CNTs could facilitate the transport of adsorbed substances that would otherwise be strongly retained by immobile soils or sediments. VOL. 42, NO. 8, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2899
In the present study, we have used naphthalene to probe sorption properties of MWCNTs that contain various surface oxygen concentrations, and we have compared results with those obtained on a natural char and a granular activated carbon (GAC). We focus on MWCNTs rather than SWCNTs as nanoscale sorbents because MWCNTs are less expensive and more widely usedsin recent years, the annual production rate of MWCNTs has been 3-4 times greater than that of SWCNTs (40). Naphthalene was chosen because it is a common, low-molecular-weight HOC that is a common environmental contaminant and its sorption properties have been previously studied on CNTs and other BCs (16, 17). To better understand the influence that surface oxides exert on the properties of CNTs, we have combined our sorption studies with detailed surface characterization.
Experimental Section Chemicals and Reagents. 14C-labeled and unlabeled naphthalene were used as received (Sigma-Aldrich). For sorption experiments, naphthalene spiking solutions were prepared by adding 14C-labeled (31.3 mCi/mmol) and unlabeled naphthalene to HPLC-grade methanol (Fisher Scientific). Sorbents. Granular activated carbon (GAC) was used as received (F400, Calgon Carbon Corporation). The natural char sample (NC1) was prepared as described by Nguyen et al. (41). NC1 and F400 samples were pulverized, homogenized, and sieved to pass 400 mesh (38 µm) or 200 mesh (75 µm), respectively. Pristine MWCNTs (diameter 15 ( 5 nm, length 1–5 µm, purity 95%) were purchased from Nanolab Inc. To prepare MWCNTs with different levels of oxidation, we refluxed MWCNTs in solutions containing between 10% and 70% w/w HNO3. Prior to refluxing, pristine MWCNTs were sonicated (Sonicator: Branson 1510, operating at 70 W) for 1 h in HNO3. In each case, the ratio of MWCNT to HNO3 was maintained at 0.4 mg/mL. After sonication, the MWCNT and HNO3 mixtures were refluxed for 1.5 h at 140 °C and stirred vigorously. For comparison, some MWCNTs were oxidized using H2O2 (refluxed at 80 °C for 3 h) or KMnO4 (22). After each treatment, the samples were subjected to repeated centrifugation, decantation, and dilution with milli-Q water until the supernatant pH was steady at ∼5. Resultant MWCNTs were dried overnight in an oven at 100 °C. Characterization of Pristine and Oxidized MWCNTs. The MWCNTs’ chemical composition was determined using both X-ray photoelectron spectroscopy (XPS) and elemental analysis. For XPS, MWCNTs were adhered to double-sided copper tape and mounted onto a sample stub. Care was taken to ensure that the MWCNTs completely covered the copper tape. In all XPS experiments, Mg KR (1253.6 eV) X-ray radiation was generated from a Φ 04–500 X-ray source. Ejected photoelectrons were analyzed using a hemispherical electron energy analyzer operating at a pass-energy of 44.75 eV and 0.125 eV/step. Elemental analysis was conducted by Huffman Laboratories, Inc. (Golden, CO). The length distributions of the MWCNTs were evaluated using an atomic force microscope (AFM) (Pico SPM LE; Agilent Technologies) operating in tapping mode using magnetically coated probes oscillating at 75 kHz (NSC 18/ Co-Cr; MikroMasch). To exfoliate the MWCNTs prior to analysis, materials were sonicated for 2 min in ethanol. Immediately after sonication, two drops of dispersed MWCNT-suspension were deposited onto a silicon substrate using flash evaporation to minimize reaggregation. A sufficient number of 5.0-µm2 AFM images were acquired to fully resolve ∼300 MWCNTs and the construction of a statistically valid histogram of length distribution. The structural integrity of oxidized MWCNTs was examined using a transmission electron microscope (TEM). A drop of the dispersed MWCNT water suspension was placed on 2900
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 8, 2008
a holey-carbon TEM grid. The samples were imaged using a Philips CM 300 field emission gun TEM operating at 297 kV. Images were collected using a CCD camera mounted on a GIF 200 electron energy loss spectrometer. For BET measurements of specific surface area (SSA), N2 adsorption data at 77 K were obtained using a high-resolution gas adsorption analyzer with high vacuum capacity (5.0 × 10-7 Pa) (ASAP 2010, Micromeritics). Following previously established protocols for black carbon (41), all samples were outgassed at 300 °C for 5 h prior to analysis. Sorption Experiments. Experiments were conducted in 5-mL(nominalvolume)flame-sealedglassampules(Wheaton). Into each ampule, 0.5-2.0 mg of sorbent (gravimetrically determined) was added and the amplues were filled with 7.0 mL of naphthalene-containing solution. Except for experiments designed to test the effects of aggregation on sorption isotherms (Figures S7 and S8 and details therein; Supporting Information) this naphthalene-containing solution was prepared in advance by measuring a precise volume of naphthalene/MeOH stock solution into 100 mL of water containing 3.0 mM NaN3 (to prevent bacterial growth) and 5.0 mM CaCl2 (to facilitate coagulation). Coagulation is necessary to achieve good solid/water separation during sorption experiments. Separate control tests were conducted to verify the lack of impact of ionic strength on naphthalene sorption isotherms with MWCNTs, as described in Figure S8 (Supporting Information). To minimize cosolvent effects, the MeOH/H2O ratio was kept below 0.1% v/v. After naphthalene addition, each ampule was immediately flame sealed (Ampulmatic, Bioscience Inc.) and the system was allowed to equilibrate during end-over-end rotation. Equilibration was accomplished in a dark environmental chamber at 23 ( 0.5 °C. For MWCNTs, 14-day tests were conducted at Ce/Sw ) 0.001 (Figure S1) to confirm that 6 days was sufficient for sorption equilibrium. (Ce [mg/L] is equilibrium aqueous concentration and Sw is water solubility; 31.7 mg/L for naphthalene at 25 °C) Kd values were (2.6 ( 0.3) × 105 mL/g and (2.4 ( 0.2) × 105 mL/g at 6 and 14 days, respectively, confirming statistical equivalence at a significance level, R, of 5% (two-sided t test). (Kd ) qe/Ce, where qe [mg/g] is the sorbed phase concentration.) Kinetic experiments also revealed that less than 6 days was sufficient for equilibrium to be achieved with GAC (F400), but 28 days was required for NC1. After equilibration, the ampules were placed into centrifuge tubes containing a small quantity of water and a cushion of glass wool and centrifuged at 3000g for 30 min. The ampules were then broken and 1.0 mL of supernatant was withdrawn. The concentration of 14C-labeled naphthalene in the withdrawn solution was measured by liquid scintillation (model LS3801, Beckman Instruments, Fullerton, CA), using counting times that achieved
