ARTICLE pubs.acs.org/Langmuir
Influence of Surface Oxidation on the Aggregation and Deposition Kinetics of Multiwalled Carbon Nanotubes in Monovalent and Divalent Electrolytes Peng Yi and Kai Loon Chen* Department of Geography and Environmental Engineering, Johns Hopkins University, Baltimore, Maryland 21218-2686, United States
bS Supporting Information ABSTRACT: The aggregation and deposition kinetics of two multiwalled carbon nanotubes (MWNTs) with different degrees of surface oxidation are investigated using time-resolved dynamic light scattering (DLS) and quartz crystal microbalance with dissipation monitoring (QCM-D), respectively. Carboxyl groups are determined to be the predominant oxygen-containing surface functional groups for both MWNTs through X-ray photoelectron spectroscopy (XPS). The aggregation and deposition behavior of both MWNTs is in qualitative agreement with the Derjaguin-Landau-Verwey-Overbeek (DLVO) theory. The critical coagulation concentration (CCC) of the highly oxidized MWNTs (HO-MWNTs) is significantly higher than the lowly oxidized MWNTs (LO-MWNTs) in the presence of NaCl (210 and 53 mM, respectively) since HO-MWNTs have a higher surface charge density. In contrast, the aggregation inverse stability profiles of HO-MWNTs and LO-MWNTs are identical and yield comparable CCCs (0.9 and 1.0 mM, respectively) in the presence of CaCl2. Similar to the results obtained from the aggregation study, HO-MWNTs are considerably more stable to deposition on silica surfaces compared to LO-MWNTs in the presence of NaCl. However, both MWNTs have the same propensity to undergo deposition in the presence of CaCl2. The remarkable similarity in the aggregation and deposition kinetics of HO-MWNTs and LO-MWNTs in CaCl2 may be due to Ca2þ cations having a higher affinity to form complexes with adjacent carboxyl groups on HO-MWNTs than with isolated carboxyl groups on LO-MWNTs.
1. INTRODUCTION Carbon nanotubes (CNTs) are currently one of the most studied nanomaterials in numerous fields due to their unique mechanical, chemical, and electronic properties. In recent years, the number of applications that require the dispersion of CNTs in aqueous solutions, particularly in the area of biomedical and environmental engineering, has been escalating.1-5 Since unfunctionalized CNTs are extremely hydrophobic and tend to aggregate quickly in aqueous solutions, one of the key strategies to enhance the colloidal stability of CNTs is to covalently attach either charged or hydrophilic functional groups on the surfaces of CNTs through chemical treatment.6-8 A popular approach is to expose the CNTs to strong oxidants or concentrated acids in order to create oxygen-containing functional groups, such as carboxyl and hydroxyl groups, on the sidewalls and tube ends of CNTs.9,10 The use of these functionalized CNTs for applications in aqueous environments necessitates a thorough understanding of the aggregation and deposition (or adsorption) behavior of CNTs. In solution-based applications in which CNTs are suspended in water, it is critical that CNTs remain dispersed as individual strands so that their specific surface area can be maximized.11 When CNTs are utilized in optical devices, the colloidal stability of CNTs will greatly influence the optical properties of the CNT suspensions.9 Numerous studies have r 2011 American Chemical Society
also been conducted on the use of CNTs as transporters for drug delivery.4,11,12 In these applications, it is crucial that CNTs are resistant to aggregation so that they can remain mobile and be effectively delivered within the human body to the targeted area. In addition to applications that require the dispersion of CNTs in solution, other applications may involve the immobilization of CNTs on substrates. Some examples of these applications include thin film composites and coatings, electrodes, and biosensors.13-15 Several studies have shown that CNTs can be immobilized on flat solid surfaces and colloidal particles (e.g., glassy carbon, silica, and silicon) through physisorption,16,17 electrophoretic deposition,14,18 and layer-by-layer deposition.19-21 A comprehensive knowledge of the influence of the surface chemistry of CNTs, as well as the chemical composition of background solutions, on the deposition of CNTs on substrates will allow for effective implementation of the immobilization techniques. With consumer products that contain CNTs already available in the market, it is inevitable that some of the carbon-based nanomaterial will be released into natural aquatic systems.22-24 Recent studies have shown that CNTs exhibit toxic effects on Received: November 24, 2010 Revised: January 21, 2011 Published: February 28, 2011 3588
dx.doi.org/10.1021/la104682b | Langmuir 2011, 27, 3588–3599
Langmuir bacteria,25-28 thus raising concerns on the potential impacts of the nanomaterials on the environment and ecosystem. The fate and transport of CNTs in surface water and groundwater systems are dependent on the aggregation and deposition behavior of these nanoparticles.29,30 The aggregation of CNTs will result in the formation of CNT clusters, which will settle much faster compared to CNTs in their dispersed form. The aggregation state of CNTs is also likely to play an important role in controlling the reactivity of CNTs and their impacts on the environment.31 In addition, CNTs can deposit on other solid bodies in the environment, such as rocks, sediments, and mineral surfaces.32 The deposition of CNTs on these immobile surfaces will lead to the removal of CNTs from the aqueous phase and thus will have an effect on the bioavailability of CNTs in these systems. Therefore, a better understanding of the aggregation and deposition behavior of CNTs will allow for a more accurate prediction of the environmental fate and impact of these nanomaterials. Several groups have investigated the influence of solution chemistry on the aggregation behavior of CNTs. Saleh et al.33,34 employed time-resolved dynamic light scattering (DLS) to examine the influence of solution composition on the aggregation kinetics of single-walled and multiwalled CNTs (SWNTs and MWNTs, respectively) which were dispersed in water through ultrasonication using an ultrasonication probe. Their results showed that the aggregation behavior of both CNTs was in qualitative agreement with the classic Derjaguin-LandauVerwey-Overbeek (DLVO) theory.35,36 In another recent study, Smith et al.37 examined the effects of surface oxidation on the colloidal stability of MWNTs in a monovalent electrolyte solution (NaCl). By performing time-resolved DLS measurements on a suite of MWNTs with varying surface densities of oxygen-containing functional groups, the authors showed that the colloidal stability of MWNTs is enhanced as the degree of surface oxidation is increased.37 However, the dependence of MWNT stability on surface oxidation was not investigated in divalent electrolyte solutions in which charge neutralization38 may play the dominant role in controlling the colloidal stability of CNTs and thus confound the relationship between colloidal stability and surface oxidation. In recent years, the deposition (or adsorption) behavior of nanoparticles has been investigated using the quartz crystal microbalance with dissipation monitoring (QCM-D). This technique has gained popularity for nanoparticle deposition studies due to the high sensitivity and small sample volume requirements of the QCM-D. Chen and Elimelech39,40 employed this technique to demonstrate that the deposition kinetics of fullerene C60 nanoparticles on silica surfaces is controlled by electric double layer and van der Waals interactions. Similar studies have also been performed to derive the deposition kinetics of viruses,41 quantum dots,42 titanium dioxide,43 zerovalent iron,44,45 and zinc oxide nanoparticles.46 Until now, no study has been conducted to investigate the influence of surface oxidation on the deposition kinetics of CNTs using the QCM-D. The objective of this research is to quantify and compare the aggregation and deposition kinetics of two MWNTs with different degrees of surface oxidation in the presence of monovalent (NaCl) and divalent (CaCl2) electrolytes that are ubiquitous in biological and environmental systems. The two MWNTs were oxidized with acid mixtures of different concentrations, and the distributions of oxygen-containing functional groups were determined using X-ray photoelectron spectroscopy (XPS) and
ARTICLE
chemical derivatization. The aggregation kinetics of both MWNTs were obtained from time-resolved DLS measurements over a range of NaCl and CaCl2 concentrations. The deposition kinetics of the MWNTs were derived by performing both unfavorable and favorable (transport-limited) deposition experiments using the QCM-D. Our results show that the aggregation and deposition behavior of both MWNTs is consistent with classic DLVO theory. Furthermore, the degree of surface oxidation was found to have a significant effect on the aggregation and deposition kinetics of MWNTs in NaCl, while the aggregation and deposition kinetics of both MWNTs are comparable in CaCl2.
2. MATERIALS AND METHODS 2.1. Preparation and Characterization of MWNTs. A single batch of pristine MWNTs was purchased from NanoLab, Inc. (PD15L520, Newton, MA) and used as the starting material in this study. According to the manufacturer, the MWNTs have lengths and outer diameters of 5-20 μm and 15 ( 5 nm, respectively, and contain 99.75% carbon, 0.20% iron, and 0.05% sulfur in atomic percentage. The oxidized MWNTs were prepared and purified using the method described in the work of Smith et al.37 Highly oxidized MWNTs (HO-MWNTs) were prepared by refluxing the pristine nanotubes in a concentrated acid mixture of sulfuric acid (98% H2SO4 by mass) and nitric acid (69% HNO3 by mass) at 70 °C for 8 h. The volume ratio of concentrated sulfuric acid and nitric acid in the acid mixture is 3:1. Lowly oxidized MWNTs (LO-MWNTs) were prepared in a similar manner by refluxing the pristine nanotubes in a diluted acid mixture at 70 °C for 8 h. The diluted acid mixture was prepared by diluting the concentrated acid mixture four times by volume with deionized (DI) water. For both oxidation processes, a concentration of ca. 12.5 mg of MWNTs per mL of acid mixture was used. After oxidation, the MWNTs were extracted from the mixtures, which also contained residual acids, metallic byproducts, and amorphous carbon, by repeated cycles of dilution with DI water, centrifugation at 1850g for 5-10 min (Powerspin LX, Unico, Dayton, NJ), and decantation of supernatant until the resistivity of supernatant was greater than 0.5 MΩ 3 cm and the supernatant pH reached 5. The cleaned MWNTs were then dried overnight in an oven at 100 °C. The HO-MWNT and LO-MWNT powders were pulverized in a ball-mill (MM200, Retsch, Germany) for 15 min and stored in capped glass vials in the dark at room temperature. The total oxygen content of HO-MWNTs and LO-MWNTs was determined using XPS as described in Cho et al.47 The distributions of carbonyl (CdO), carboxyl (COOH), and hydroxyl (C-OH) groups for both HO-MWNTs and LO-MWNTs were quantified by performing XPS analysis in conjunction with vapor phase chemical derivatization as described in Smith et al.37 and references therein. Briefly, fluorinecontaining reagents (trifluoroethyl hydrazine, trifluoroethanol, and trifluoroacetic anhydride) were used to selectively label the oxygencontaining functional groups (carbonyl, carboxylic acid, and hydroxyl groups, respectively). After the chemical derivatization reaction, the F(1s) signal obtained from XPS was used to quantify the concentrations of the targeted functional groups. To prepare HO-MWNT and LO-MWNT stock suspensions, ca. 0.5 mg of the respective MWNT powder was dispersed in 200 mL DI water (Millipore, MA), which was contained in a 250-mL conical flask, through low-power ultrasonication in an ultrasonic bath (Branson 1510R-MT, output power 70 W, frequency 40 kHz) for 20 h. After ultrasonication, the MWNT suspension was centrifuged at 1400g (Avanti centrifuge J-20 XPI, Beckman Coulter Inc., Brea, CA) for 5-10 min in order to remove MWNT clusters that still remained in the suspensions. The supernatant, 3589
dx.doi.org/10.1021/la104682b |Langmuir 2011, 27, 3588–3599
Langmuir which contained mostly dispersed MWNTs, was carefully withdrawn using disposable sterile glass pipettes and stored in clean Pyrex bottles in the dark at 4 °C. Five and six batches of HO-MWNT and LO-MWNT stock suspensions were prepared, respectively. The HO-MWNT stock suspensions were used within 3 months after preparation, while the LOMWNT stock suspensions were used within 1 month. The average hydrodynamic diameters of the HO-MWNTs and LO-MWNTs in freshly produced stock suspensions were determined to be 104 and 150 nm, respectively, using DLS. No change in hydrodynamic diameter was observed for the HO-MWNTs even after three months, while the average increase in the hydrodynamic diameter of the LO-MWNTs within the time period of use was 30 nm. This observation is a first indication that the HO-MWNTs are more stable to aggregation than the LO-MWNTs in DI water. 2.2. Solution Chemistry. ACS-grade NaCl, CaCl2, and NaHCO3 electrolyte stock solutions were prepared and filtered using 0.1-μm syringe filters (Anotop 25, Whatman, Middlesex, UK). All aggregation and deposition experiments, as well as most electrophoretic mobility measurements, were performed at pH 7.1 ( 0.2 (buffered with 0.15 mM NaHCO3) and at the temperature of 25 °C. Some electrophoretic mobility measurements were conducted at pH 2.0 ( 0.1 (adjusted by adding HCl). 2.3. Electrophoretic Mobility Measurements. The electrophoretic mobilities (EPMs) of HO-MWNTs and LO-MWNTs were measured (ZetaPALS, Brookhaven Instruments Corp., Holtsville, NY) over a range of NaCl and CaCl2 concentrations. The HO-MWNT and LO-MWNT concentrations for EPM measurements were 0.40 and 0.20 times, respectively, of the concentrations of the corresponding MWNT stock suspensions. For most electrolyte concentrations, 10 measurements were conducted for each of at least three samples at each electrolyte concentration. However, at high NaCl and CaCl2 concentrations, fewer measurements (2-7) were conducted for each sample in order to minimize the effects of MWNT aggregation during measurements. 2.4. Time-Resolved DLS. In order to investigate the early stage aggregation kinetics of both MWNTs, time-resolved DLS measurements were conducted with the employment of a light scattering unit that comprises an argon laser unit (Lexel 95, Cambridge Laser Laboratories, Fremont, CA) that emits a laser with a wavelength of 488 nm, a photomultiplier tube mounted on a goniometer (BI-200SM, Brookhaven, NY), a digital correlator (BI-9000AT, Brookhaven, NY), and a thermostatted vat filled with an index-matching cis and trans mixture of decahydronaphthalene. For each aggregation experiment, a precalculated amount of electrolyte stock solution was introduced into the vial containing a predetermined volume of diluted MWNT suspension such that the total volume of the final suspension is 1 mL. The MWNT concentration for the DLS measurements was 0.04 times of the concentration of the MWNT stock suspension. The vial was capped, shaken for ca. 1 s with a vortex mixer, and quickly inserted into the vat of the light scattering unit. The DLS measurements were started immediately and the intensity-weighted hydrodynamic diameters of the aggregating MWNTs were monitored over time periods of 10-90 min. The time between the introduction of electrolyte stock solution and the start of the first DLS measurement was ca. 10 s. All glass vials were soaked in cleaning solution (Extran MA01, Merck KGaA, Darmstadt, Germany), thoroughly rinsed with DI water, and dried in an oven at 100 °C under dust-free conditions before use. The vials were only used once and disposed after use. All DLS measurements were performed at a scattering angle of 90° and each autocorrelation function was accumulated for 15 s. The hydrodynamic diameters were subsequently obtained through second-order cumulant analysis (Brookhaven software). 2.5. Determining Aggregation Kinetics of MWNTs. The early stage aggregation kinetics of colloidal particles undergoing Brownian diffusion can be obtained from time-resolved DLS measurements
ARTICLE
using the following equation:48,49 1 dDh ðtÞ kµ N0 dt tf0
ð1Þ
where k is the aggregation rate constant, N0 is the initial primary particle concentration, and Dh(t) is the intensity-weighted hydrodynamic diameter at time t. This relationship has also been used in previous investigations of the aggregation kinetics of SWNTs34 and MWNTs.33,37,50 For most of the electrolyte concentrations used in this study, the initial rates of increase in Dh(t) were obtained though linear least-squares regression analysis conducted to 100 nm in excess of the initial hydrodynamic diameters, Dh(0), for both MWNTs. However, at the lower range of electrolyte concentrations, the regression analyses were conducted to 20-60 nm in excess of Dh(0) over a time period of 20-90 min since aggregation took place very slowly. For all electrolyte concentrations, the y-intercepts of fitted lines are within 15% in excess of Dh(0). To quantify the aggregation kinetics of MWNTs, the aggregation attachment efficiency, RA (or inverse stability ratio, 1/W), was calculated by dividing the initial rate of increase in Dh(t) at the electrolyte concentration of interest by the initial rate of increase in Dh(t) under diffusion-limited conditions48,49 1 dDh ðtÞ N0 dt 1 k t f 0 ¼ RA ¼ ¼ ð2Þ 1 dDh ðtÞ W kfast ðN0 Þfast dt t f 0, fast The terms with subscript “fast” refer to diffusion-limited conditions. The aggregation attachment efficiency can be understood to be the probability of a permanent attachment resulting from the collision of two MWNTs.51 2.6. QCM-D. The deposition (or adsorption) kinetics of MWNTs on silica surfaces were measured using a QCM-D (E4, Q-Sense, V€astra Fr€olunda, Sweden). The E4 system has four identical flow modules and a 5 MHz AT-cut quartz crystal sensor with a silica-coated surface (QSX 303, Q-Sense) is mounted in each module. The frequency and dissipation of the crystals are continuously monitored by the E4 system during the entire experiment in order to derive the deposition kinetics of MWNTs. During the deposition experiments, the electrolyte solution, which was prepared at twice the concentration of interest, and MWNT suspension were withdrawn from separate reservoirs, combined in a T-junction, and introduced into each flow module using a peristaltic pump (ISM935C, Ismatec SA, Z€urich, Switzerland). The MWNT concentration in the combined solution entering the chamber in the modules was 0.20-0.48 times of the concentration of the MWNT stock suspension. The flow rate of the combined solution entering each module was maintained at 0.60 mL/min ((0.03 mL/min). According to the manufacturer, such a flow rate results in laminar flow in the flow module. The flow rates of the branches delivering MWNT suspension and electrolyte solution were also verified to be equal (within 5% deviation from the average of both flow rates) before every deposition experiment. This approach limits the time of MWNT exposure to the electrolyte solution to ca. 20 s before the MWNTs enter the chamber and thus significantly reduces the degree of concurrent MWNT aggregation, especially at high electrolyte concentrations. Before all deposition experiments, the quartz crystal sensors were soaked in 2% Hellmanex II cleaning solution (Hellma GmbH & Co. KG, M€ullheim, Germany) for 30 min, thoroughly rinsed with DI water, dried with ultrapure nitrogen, and oxidized in a UV-ozone chamber (Procleaner 110, BioForce Nanosciences, Inc., Ames, IA) for 20 min. All electrolyte solutions were degassed through ultrasonication (Branson 5510R-DTH, output power 135 W, frequency 40 kHz) for 3590
dx.doi.org/10.1021/la104682b |Langmuir 2011, 27, 3588–3599
Langmuir
ARTICLE
Figure 1. Total oxygen content and distributions of carbonyl (CdO), carboxyl (COOH), and hydroxyl (C-OH) groups on the surfaces of HO-MWNTs and LO-MWNTs.
Figure 2. EPMs of HO-MWNTs and LO-MWNTs as functions of NaCl and CaCl2 concentrations at pH 7.1. Error bars represent standard deviations.
10 min and the temperature of the electrolyte solutions and MWNT suspensions were maintained at 27 °C water bulk, 2 °C above the experimental temperature, in a water bath before the deposition experiments. In order to obtain the deposition kinetics of MWNTs at an electrolyte concentration of interest, both unfavorable and favorable deposition experiments were performed. All unfavorable and favorable experiments were duplicated for each experimental condition. For the unfavorable deposition experiments, the silica-coated crystal surfaces were first rinsed with DI water for 40 min and then rinsed with the electrolyte solution of interest for 30-40 min in order to achieve a drift in the normalized frequency of less than 0.2 Hz in a time period of 10 min. Following that, the MWNT suspension and electrolyte solution were both introduced into the chamber for 20-40 min for deposition to take place on the silica surface. For the favorable deposition experiments, a buffer comprising 10 mM N-(2-hydroxyethyl)-peperazine-N0 -(2-ethanesulfonic acid) (HEPES) (H4034-100G, Sigma-Aldrich, St. Louis, MO) and 100 mM NaCl was first prepared and filtered through 0.22-μm cellulose acetate filters (Corning, Lowell, MA). A 0.1 g/L cationic poly-L-lysine (PLL) polyelectrolyte (P-1274, Sigma-Aldrich, St. Louis, MO) solution was prepared using the HEPES buffer. Before each favorable deposition experiment, the crystal surface is modified by sequentially rinsing the crystal with 4.8 mL of HEPES buffer, 4.8 mL of PLL solution, and another 4.8 mL of HEPES buffer.40,52 The frequency and dissipation decreased and increased sharply, respectively, before reaching a plateau during the PLL adsorption process, indicating that the crystal surface is completely coated with PLL. The favorable deposition experiments were then conducted on the positively charged crystal surfaces using the same approach as for the unfavorable deposition experiments. As deposition or adsorption takes place on the crystal surface, the increase in the mass of the crystal will result in a corresponding decrease in the resonance and overtone frequencies of the crystal, as described by the Sauerbrey relationship53
also result in an enhancement in the crystal’s ability to dissipate energy, as reflected by the increase in the energy dissipation (D) monitored by the E4 system54
Δm ¼ -
CΔfn n
ð3Þ
where Δm is the deposited mass, Δfn is the frequency shift of the nth harmonic (n = 1, 3, 5, 7, 9, 11, and 13; the first harmonic is the fundamental oscillation), and C is the crystal constant (17.7 ng/ (Hz cm2)). This relationship only applies if a thin and rigid film is formed on the crystal surface.53 Deposition on the crystal surface may
D¼
Edissipation 2πEstored
ð4Þ
where Edissipation is the energy dissipated in one oscillation cycle and Estored is the total energy stored in the oscillator.
3. RESULTS AND DISCUSSION 3.1. Surface and Electrokinetic Characterization of MWNTs. The total oxygen composition and the distribution of
oxygen-containing surface functional groups for HO-MWNTs and LO-MWNTs obtained from XPS analysis are presented in Figure 1. Carbon and oxygen were the only elements detected through XPS analysis of both MWNTs, indicating that the MWNTs were free from contamination. The total oxygen composition for HO-MWNTs and LO-MWNTs are 10.6% and 6.3%, respectively. The considerably higher oxygen composition in HO-MWNTs compared to LO-MWNTs indicates that the degree of surface oxidation can be controlled by the concentration of acid mixture used to oxidize the MWNTs. As shown in Figure 1, carbonyl, carboxyl, and hydroxyl groups can be found on both MWNTs. The XPS results show that the concentrations of carbonyl, carboxyl, and hydroxyl groups for HO-MWNTs are all higher than those for LO-MWNTs. Also, carboxyl groups form the largest proportion of oxygen-containing groups for both MWNTs. The oxygen content contributed by other types of oxygen-containing functional groups (O(Others)) is calculated by subtracting the contributions by the carbonyl, carboxyl, and hydroxyl groups from the total oxygen composition. The small calculated values of O(Others) for both MWNTs (absolute values
