Aggregation and interactions of chemical mechanical planarization nanoparticles with model biological membranes: role of phosphate adsorption

Abstract

In the semiconductor industry, chemical mechanical planarization (CMP) slurries are used in large quantities for the planarization of wafers and the release of CMP nanoparticles (NPs) into the natural environment may pose a threat to human health. In this study, the aggregation behavior of four model CMP NPs, namely, colloidal SiO₂, fumed SiO₂, CeO₂, and Al₂O₃ NPs, was investigated through dynamic light scattering in the presence and absence of phosphate at pH 7.4. The colloidal and fumed SiO₂ NPs were observed to have remarkable stability under neutral pH conditions both in the presence and absence of phosphate. Phosphate adsorption enhanced the negative surface charge of CeO₂ NPs and reversed the charge of Al₂O₃ NPs, resulting in an increase in the stability of both NPs. In order to evaluate the propensity for the CMP NPs to attach to and disrupt cell membranes, the interactions of these NPs with supported lipid bilayers (SLBs) and lipid vesicles composed of zwitterionic 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) were examined using a quartz crystal microbalance. The attachment efficiencies of both SiO₂ NPs on SLBs increased to 1 when the NaCl concentration was raised to 100 mM in the presence of phosphate. In contrast, favorable attachment of CeO₂ and Al₂O₃ NPs to SLBs was not achieved in the presence of phosphate even at elevated NaCl concentrations, likely due to the enhancement of nanoparticle surface charge resulting from phosphate adsorption. When the supported vesicles were exposed to the SiO₂ NPs, no disruption of vesicles was detected despite favorable conditions for nanoparticle–membrane association.

---

Nano impact

With the extensive use of CMP slurries in the semiconductor industry, there is a growing concern on their effect on human health. In this study, the surface charge and colloidal stability of CMP NPs, as well as their interactions with model membranes, were investigated, with an emphasis on the role of phosphate adsorption. Our results show that the presence of phosphate altered the surface charge of CeO₂ and Al₂O₃ NPs, which in turn enhanced their stability and reduced their attachment to lipid bilayers. Since phosphate is a key constituent in biological fluids, these findings will allow for a better understanding of the aggregation behavior of CMP NPs and their association with cell membranes when they enter the human body.

Introduction

Chemical mechanical planarization (CMP) is a key process in the semiconductor industry for the production of integrated circuit devices.¹ In CMP processes, slurries of abrasive particles, including SiO₂, CeO₂, and Al₂O₃, are employed to planarize wafers through a combination of mechanical abrasion and chemical etching.² The planarized wafers are subsequently rinsed with ultra-pure water, resulting in the generation of CMP wastewater.^3,4 Although CMP wastewaters are generally subjected to on-site wastewater treatment (e.g., electrocoagulation⁵ and microfiltration⁶) and/or municipal biological wastewater treatment, the wastewater effluent from municipal wastewater treatment plants may still contain some unremoved CMP NPs which can be discharged, together with the treated effluent, into surface waters.⁷ In their study on the removal efficiency of CeO₂ NPs by a model wastewater treatment plant, Limbach et al.⁸ reported that about 6 wt% of the added CeO₂ NPs was found in the outflow of the model plant. The release of CMP NPs into the natural environment potentially poses a threat to the ecosystem and human health. In addition to the risks associated with the environmental release of CMP NPs, the exposure of workers to CMP NPs at the workplace may give rise to health risk concerns.^9,10

There have been extensive studies concerning the toxicity of SiO₂, CeO₂, and Al₂O₃ NPs. The SiO₂ NPs used in CMP processes mainly include colloidal silica (c-SiO₂) and fumed silica (f-SiO₂).⁷ Colloidal SiO₂ is synthesized through the hydrolysis and condensation of a Si precursor (e.g., silicon alkoxide),¹¹ while fumed SiO₂ is produced by pyrolysis of silicon tetrachloride (SiCl₄) at high temperatures (1800 °C).^12,13 It has been reported that f-SiO₂ has a greater ability than c-SiO₂ to generate hydroxyl radicals and cause red blood cells (RBCs) to undergo lysis due to their strained three-membered rings and chainlike aggregate structure.¹⁴ The CeO₂ NPs used in CMP processes are made from high temperature calcination (>300 °C).⁷ CeO₂ NPs have been shown to induce oxidative stress in human cancer cells and subsequently result in the damage of cell membranes.¹⁵ Finally, Al₂O₃ NPs can be produced through dehydration of γ-AlOOH (boehmite) at high temperatures (500–700 °C).¹⁶ Previous studies have reported that Al₂O₃ NPs can show inhibitory effect on human bronchial epithelial cells and cause hemolysis of RBCs.^17,18

While the mechanisms for the cytotoxicity of different classes of nanomaterials are still under investigation, studies have shown that the direct association of NPs with cell membranes is often a critical step that precedes the subsequence biological responses.^19,20 The attachment of NPs on cell membranes can also result in damage of the membranes.²⁰ Therefore, a detailed assessment of the interactions of CMP NPs with cell membranes is expected to improve our understanding on the biological effects of these nanomaterials. Due to the inherent heterogeneity and diversity of cell membranes, the use of model (or artificial) cell membranes that are composed of lipid bilayers is an attractive approach to gain fundamental insights into the interactions between NPs and cell membranes.^20–22 A common approach to investigate the attachment of NPs to cell membranes involves the employment of supported lipid bilayers (SLBs) to evaluate the NP's propensity to attach on lipid bilayers.^23,24 Since lipid bilayer is a key component of cell membranes, the use of SLBs will enable the elucidation of the nonspecific interactions between NPs and cell membranes at the nano–bio interface.^{19,20,24–28} A recent study utilizing model membranes showed that the presence of proteins can decrease the attachment of NPs to membranes and hence reduce the uptake of NPs by human epithelial cells.²⁹ Additionally, lipid bilayer vesicles have been used as model cell membranes to probe the propensity of NPs to disrupt bilayers.^20,30–32

The objective of this work is to investigate the interactions of model CMP NPs with model biological membranes which include SLBs and supported vesicular layers (SVLs). Since phosphate is a key constituent in biological fluids, such as human blood serum (0.81–1.45 mM) and saliva (5–14 mM),³³ the role of phosphate in these interactions will be studied. In addition, we will examine the aggregation behavior of CMP NPs in model biological media since the cellular uptake of NPs, as well as the subsequent biological responses, has been shown to be dependent on the aggregation state of the NPs.^34,35 The results of this study will provide new insights into the nonspecific interactions of CMP NPs with cell membranes when they enter biological systems.

Experimental methods

CMP NP suspensions

Four model CMP NPs, namely, c-SiO₂, f-SiO₂, CeO₂, and Al₂O₃, were provided by Cabot Microelectronics Corporation (Aurora, Illinois, USA) as slurries. These NP slurries are shared within a research consortium and have been extensively characterized.⁷ Their physicochemical properties (including shape, size, crystalline structure, and surface charge) are presented in a recently published paper by Speed et al.⁷ These slurries were maintained colloidally stable by pH adjustment and did not contain organic stabilizers.⁷ Here, we highlight some of the properties of the CMP NPs reported by Speed et al.⁷ Through transmission electron microscopy (TEM) imaging, the c-SiO₂ NPs were observed to be mostly spherical, while the f-SiO₂ NPs appeared irregular-shaped and fused together. X-ray diffraction (XRD) analysis showed that both c-SiO₂ and f-SiO₂ NPs had amorphous SiO₂ structures. The CeO₂ NPs were rectangular and the XRD patterns of the CeO₂ slurry revealed the cubic fluorite structure of the NPs.³⁶ Finally, the Al₂O₃ NPs contained a broad range of particles sizes and XRD results indicated that the Al₂O₃ NPs were γ-Al₂O₃.³⁷ The mean diameters of the c-SiO₂, CeO₂, and Al₂O₃ NPs determined through TEM imaging were 36, 39, and 38 nm, respectively. The size of the f-SiO₂ NPs could not be determined through TEM due to their coalesced state. By employing dynamic light scattering (DLS), the hydrodynamic diameter of the f-SiO₂ NPs was determined to be ca. 148 nm. All slurries were used directly after being diluted in deionized (DI) water (18 MΩ cm, Millipore, MA) without further purification.

Solution chemistry

ACS-grade NaCl, NaH₂PO₄, Na₂HPO₄, and NaHCO₃ were used for the preparation of stock solutions. All stock solutions were filtered using 0.1 μm polyvinylidene fluoride syringe filters (Millipore, MA). The pH of the diluted CMP dispersions was maintained at 7.4 with either a phosphate (1.13 mM NaH₂PO₄ and 3.87 mM Na₂HPO₄) or bicarbonate (0.4 mM NaHCO₃) buffer. Three buffers containing N-(2-hydroxyethyl)-piperazine-N′-(2-ethanesulfonic acid) (HEPES, Sigma-Aldrich) were also used in the experiments using a quartz crystal microbalance with dissipation monitoring (QCM-D). They are denoted as Buffer A (10 mM HEPES, 150 mM NaCl, pH adjusted to 7.4), Buffer B (10 mM HEPES, 100 mM NaCl, with unadjusted pH ca. 5.5), and Buffer C (20 mM HEPES, 150 mM NaCl, pH adjusted to 7.4).

Electrophoretic mobility measurements

The electrophoretic mobilities (EPMs) of the CMP NPs were measured (ZetaPALS, Brookhaven Instruments Corp., Holtsville, NY) over a range of NaCl concentrations in either phosphate or bicarbonate buffer at pH 7.4. The concentrations of the CMP NPs used in the EPM measurements are presented in Table S1 of the ESI.† Three samples were measured at each electrolyte concentration. For most of the experimental conditions, ten measurements were conducted for each sample. At high NaCl concentrations, only five measurements were conducted in order to minimize the effect of NP aggregation.

Phosphate adsorption by CMP NPs

The adsorption of phosphate on the CMP NPs was determined after an exposure time period of 10 min. The concentration of phosphate was measured using the colorimetric technique of Murphy and Riley.³⁸ The details for the adsorption experiments are provided in the ESI.†

Dynamic light scattering

Time-resolved DLS measurements were conducted with a light scattering unit to investigate the aggregation kinetics of four CMP NPs in phosphate and bicarbonate buffers. This unit employs an argon laser with a wavelength of 488 nm (Lexel 95, Cambridge Lasers Laboratories Ltd., CA).³⁹ The details for the time-resolved DLS measurements are provided in our previous publications^39,40 and also in the ESI.† The early-stage aggregation kinetics of CMP NPs can be derived from time-resolved DLS measurements using eqn (1):^41,42where k is the aggregation rate constant, N₀ is the initial primary particle concentration, and D_h(t) is the intensity-weighted hydrodynamic diameter at time t. Note that D_h(t) is calculated from the diffusion coefficient of the CMP NPs/aggregates and is equivalent to the diameter of a sphere that has the same diffusion coefficient as the CMP NPs/aggregates. The DLS technique has also been used to investigate the aggregation kinetics of other non-spherical colloids, such as nickel hydroxycarbonate,⁴³ carbon nanotubes,⁴⁴ and graphene oxide.⁴⁵ The NP concentrations used in the DLS experiments are presented in Table S2 of the ESI.† For most NaCl concentrations employed in the DLS measurements, the initial aggregation rates were derived by performing linear least-square regression in the early aggregation stage before the intensity-weighted hydrodynamic diameter reached 1.4D_h(0).⁴⁶ At low NaCl concentrations, however, the hydrodynamic diameter failed to reach 1.4D_h(0) and the linear regressions were performed over the first 50 min of aggregation. The aggregation attachment efficiency, α_A, was calculated to quantify the aggregation kinetics of the CMP NPs at different NaCl concentrations:^42,46The terms with subscript “fast” refer to diffusion-limited conditions.

Preparation of vesicle stock suspensions

Two types of vesicles were used in the QCM-D experiments: 1) zwitterionic 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) vesicles for the formation of SLBs to investigate the propensity for CMP NPs to attach to lipid bilayers, and 2) DOPC vesicles encapsulated with carboxyfluorescein (CF) dye (to be referred to as CF-DOPC vesicles) to assess the disruption of lipid membranes by the NPs. For the preparation of DOPC vesicles, 0.2 mL of 25 g L⁻¹ DOPC chloroform solution (Avanti Polar Lipids, Inc., Alabaster, AL) was dried in an Erlenmeyer flask using a gentle stream of ultrapure nitrogen gas. A DOPC thin film was formed at the bottom of the flask and the film was subjected to vacuum desiccation for at least 4 h. The DOPC lipids were then re-suspended in 5 mL Buffer A and the solution was magnetically stirred for ca. 30 min. To obtain unilamellar vesicles, the DOPC solution was extruded using a mini-extruder (Avanti Polar Lipids) through a 50 nm polycarbonate membrane (Whatman) for at least 15 times.⁴⁷ For the preparation of CF-DOPC vesicles, the aforementioned DOPC thin film was hydrated with a 100 mM CF solution prepared in Buffer C and was subsequently subjected to the same extrusion treatment.³² All vesicle suspensions were stored in glass vials at 4 °C and used within four days after preparation.

Quartz crystal microbalance with dissipation monitoring

The interactions of CMP NPs with DOPC SLBs and vesicles were studied using a QCM-D E4 setup (Biolin Scientific, Västra Frölunda, Sweden) that is equipped with four measurement chambers. Each chamber housed a 5 MHz quartz crystal sensor with either a silica-coated (QSX 303) or gold-coated (QSX 301) surface. The sensors were cleaned prior to the QCM-D experiments according to a protocol that was described in our previous paper.⁴⁸ The temperature inside the QCM-D chambers was maintained at 25 °C. The rate of frequency shift at the 3rd overtone, dΔf₍₃₎/dt, was used to quantify the deposition rates of GO on SLBs.^48,49 Duplicate experiments were carried out for each condition.

For the deposition of CMP NPs on SLBs, a SLB was first formed on the surface of a silica-coated sensor using the method of Keller and Kasemo.⁵⁰ The sensor was rinsed with Buffer A until a stable baseline was achieved (i.e., frequency signal drifts of less than 0.2 Hz per 10 min). Following that, a 0.07 g L⁻¹ DOPC vesicle suspension was flowed through the QCM-D chamber for ca. 5 min. The vesicles deposited on the silica surface and subsequently ruptured on the surface to form a SLB.⁵⁰ The formation of a DOPC SLB resulted in a characteristic frequency shift of −25 ± 0.5 Hz and a simultaneous dissipation shift of 0.1 × 10⁻⁶.⁵⁰ After the SLB was formed, the chamber was rinsed with Buffer A for ca. 10 min to remove the unadsorbed vesicles. A NaCl solution with phosphate or bicarbonate buffer was then flowed through the measurement chamber to obtain a stable baseline, followed by the introduction of CMP NPs in the same solution chemistry to initiate the deposition process. The NP concentrations used in the QCM-D experiments are presented in Table S3 of the ESI.† The flow rate was maintained at 0.10 ± 0.002 mL min⁻¹ using a peristaltic pump (ISM935C, Ismatec SA, Zürich, Switzerland). This flow rate resulted in a laminar flow in the QCM-D chamber.⁴⁹

The deposition kinetics of CMP NPs on SLBs were quantified using the deposition attachment efficiency, α_D, which can be calculated by normalizing the rate of frequency shift on SLBs to the rate under favorable condition obtained at the same solution chemistry:^40,51

The subscript “fav” refers to favorable deposition condition. For c-SiO₂, f-SiO₂, and CeO₂ NPs in both phosphate and bicarbonate buffers, as well as Al₂O₃ NPs in phosphate buffer, the favorable deposition experiments were performed on silica sensors coated with positively charged poly-L-lysine (PLL).^51,52 To prepare the PLL-coated sensors, PLL was diluted in Buffer B to a concentration of 0.01 g L⁻¹ before the mixture was used to modify the silica crystals.⁵¹ For Al₂O₃ NPs in bicarbonate buffer, favorable deposition conditions were achieved by using bare silica surfaces since the Al₂O₃ NPs are positively charged in bicarbonate buffer at pH 7.4 while silica is negatively charged.⁵³

Detection of dye leakage from supported vesicles

A supported CF-DOPC vesicular layer was formed on a gold-coated sensor in order to investigate the propensity for the CMP NPs to disrupt lipid membranes.^48,50,54 After a stable baseline was achieved by rinsing the sensor with Buffer C, a 0.05 g L⁻¹ CF-DOPC vesicle suspension was flowed through the chamber for ca. 30 min. The CF-DOPC vesicles deposited on the gold surface to form a SVL.^50,54 The CF dye outside the vesicles was removed by rinsing the SVL with Buffer C for at least 2 h. Following the rinsing stage, a NaCl solution (150 mM NaCl, phosphate buffer, pH 7.4) was introduced into the chamber to achieve a stable baseline. Next, a CMP NP suspension prepared in the same solution chemistry was introduced into the chamber for 40 min. At the end of each experiment, the vesicles were exposed to a 32 mM Triton X-100 solution, a membrane solubilizer,⁵⁵ to induce the complete rupture of the deposited vesicles. The fluorescence of the outflow was measured using a spectrofluorometer (Horiba Jobin Yvon, France) at an excitation wavelength of 490 nm and an emission wavelength of 517 nm.

Results and discussion

Electrokinetic properties of CMP NPs

The EPMs of the CMP NPs are presented in Fig. 1a as a function of NaCl concentration in phosphate buffer at pH 7.4. As shown in the figure, all four NPs carried negative charge in the presence of phosphate. The magnitude of the EPMs for all four NPs decreased with an increase in NaCl concentration due to the screening of surface charge.⁵⁶ In our recent work,⁷ we have measured the pH of zero point of charge (pH_ZPC) of c-SiO₂, f-SiO₂, CeO₂, and Al₂O₃ NPs in the absence of phosphate to be below 2, 2, 8, and 10, respectively. These values are consistent with the pH_ZPC values of silica (ca. 2),^53,57 CeO₂ (6.0–8.5), and γ-Al₂O₃ (7.6–9.3)^58,59 reported in the literature. The negative charge of SiO₂ NPs at pH 7.4 in the presence of phosphate is consistent with the pH_ZPC of silica. Interestingly, in the presence of phosphate, the CeO₂ and Al₂O₃ NPs carried considerable negative charge at pH 7.4.

Fig. 1 (a) EPMs of CMP NPs over a wide range of NaCl concentrations at pH 7.4 in phosphate buffer. (b) EPMs of CMP NPs at 1 and 100 mM NaCl and pH 7.4 in phosphate and bicarbonate buffer. The data in phosphate buffer was reproduced from Fig. 1a for comparison. (c) Phosphate adsorption by CMP NPs within a time duration of 10 min. The adsorption experiments were conducted at an initial phosphate concentration of 100 μM, 1 g L⁻¹ NPs, 1 mM NaCl, and pH 7.4. Error bars represent standard deviations.

The EPMs of the four CMP NPs were further measured in 1 and 100 mM NaCl in bicarbonate buffer (i.e., in the absence of phosphate) at pH 7.4 and the EPMs in the absence and presence of phosphate are compared in Fig. 1b. In bicarbonate buffer, the c-SiO₂ and f-SiO₂ NPs were negatively charged, while the CeO₂ and Al₂O₃ NPs carried slight negative charge and considerable positive charge, respectively. These results were consistent with the pH_ZPC values for SiO₂, CeO₂, and Al₂O₃ that had been reported in the literature.^53,57–59 For both c-SiO₂ and f-SiO₂ NPs, the EPMs in the two buffers at the same NaCl concentration was not statistically different (two tailed t-test, significant level of 0.05). In contrast, for CeO₂ and Al₂O₃ NPs, their EPMs in the two buffers were significantly different. This discrepancy is likely due to the adsorption of the phosphate ions on the surface of CeO₂ and Al₂O₃ NPs. In order to test this hypothesis, phosphate adsorption experiments were further conducted for the four NPs (Fig. 1c). The amount of phosphate adsorbed on CeO₂ and Al₂O₃ NPs over a time period of 10 min (expressed as μmol phosphate per g NPs) was one to two orders of magnitude higher than that on f-SiO₂ NPs (no adsorption detected on c-SiO₂ NPs). Since the densities of CeO₂ (ca. 7.3 g cm⁻³)⁶⁰ and Al₂O₃ (ca. 3.9 g cm⁻³)⁶¹ are both higher than that of amorphous SiO₂ (ca. 2.2 g cm⁻³),⁶² a more pronounced difference in phosphate adsorption on CeO₂ and Al₂O₃ NPs versus SiO₂ NPs would be expected if the amount of adsorbed phosphate is expressed on a particle number basis. Therefore, the significant change in the EPMs of CeO₂ and Al₂O₃ NPs was very likely due to the adsorption of phosphate on the NPs.

At pH 7.4, phosphate (pK_a1 = 2.1, pK_a2 = 7.2, pK_a3 = 11.9) mainly exists in the form of monovalent (H₂PO₄⁻) and divalent (HPO₄²⁻) phosphate ions (calculated to be 39 and 61%, respectively).⁵³ As shown in Fig. 1b, the Al₂O₃ NPs underwent charge reversal in the presence of phosphate. It has been shown through ³¹P NMR measurements that phosphate can adsorb to γ-Al₂O₃via both outer- and inner-sphere complexation.⁶³ Outer-sphere complexation occurs through the electrostatic attraction between the positively charged –AlOH₂⁺ groups and negatively charged phosphate ions,⁶³ whereas inner-sphere complexation takes place through direct bond formation between the Al atom and the oxygen donor atoms in the phosphate ion.^63,64 Hunter⁶⁵ reported that inner-spherically adsorbed ions have a higher propensity to reverse the sign of surface charge, whereas indifferent ions can only screen the surface charge. Therefore, inner-sphere complexation of phosphate is expected to be the primary mechanism for the reversal of surface charge of Al₂O₃ NPs. For CeO₂ NPs which already carried a negative charge at pH 7.4 (Fig. 1b), the enhancement of negative surface charge in the presence of phosphate, as observed elsewhere,^66,67 was likely a result of the inner-sphere complexation between phosphate ions and the CeO₂ NPs.⁶⁵ For both SiO₂ NPs, the surface charges in the presence and absence of phosphate were not significantly different (Fig. 1b), consistent with the negligible adsorption of phosphate on the SiO₂ NPs compared to that on CeO₂ and Al₂O₃ NPs (Fig. 1c).

Stability of c-SiO₂ and f-SiO₂ NPs at pH 7.4

The aggregation profiles of the four CMP NPs at 150 mM NaCl and pH 7.4 in the presence and absence of phosphate are presented in Fig. S1 of the ESI.† This NaCl concentration was chosen because the ionic strength in physiological fluids is ca. 150 mM.¹⁹ The hydrodynamic diameter of c-SiO₂ and f-SiO₂ NPs did not change significantly within a duration of 20 min, whereas CeO₂ and Al₂O₃ NPs underwent appreciable aggregation. This observation shows that both SiO₂ NPs were considerably more stable than CeO₂ and Al₂O₃ NPs at pH 7.4. Since phosphate ions do not adsorb favorably on both NPs, as discussed in the previous section, they are not expected to be involved in the mechanisms for the elevated stability of the SiO₂ NPs.

In order to elucidate the mechanisms for the high stability of SiO₂ NPs at pH 7.4, aggregation experiments were further carried out at pH 2.0 (adjusted with 1 M HCl), which is the reported pH_ZPC of SiO₂.^53,57 Thus, at pH 2.0, the electric double layer (EDL) repulsion between the SiO₂ NPs was expected to be largely eliminated. As shown in Fig. 2a, the f-SiO₂ NPs underwent considerable aggregation at pH 2.0 compared to at pH 7.4 (both in the presence and absence of phosphate). It can therefore be concluded that the elevated stability of f-SiO₂ NPs at pH 7.4 was due to the high surface charge that was not completely screened even at 150 mM NaCl.

Fig. 2 Aggregation profiles of (a) f-SiO₂ NPs at 150 mM NaCl at pH 7.4 (in phosphate and bicarbonate buffer) and pH 2.0 (pH adjusted with 1 M HCl), (b) c-SiO₂ NPs at 150 mM NaCl at pH 7.4 (in phosphate and bicarbonate buffer) and pH 2.0 (pH adjusted with 1 M HCl), and (c) c-SiO₂ NPs at pH 12 (pH adjusted with 1 M NaOH) at varying NaCl concentrations.

In contrast to f-SO₂ NPs, c-SiO₂ NPs were stable at pH 2.0 (Fig. 2b). Additional aggregation experiments of c-SiO₂ NPs were carried out at pH 1.0 and, similar to at pH 2.0 and 7.4, no aggregation was observed (data not shown). The high stability of c-SiO₂ colloids at ca. pH 2 has been reported by different researchers.^68–70 The presence of a repulsive force between silica surfaces at short separation distances, in addition to EDL repulsion, has been detected through force measurements using atomic force microscopy and surface force apparatus.^71,72 Israelachvili and co-workers^72,73 proposed that a hairy layer of protruding silanol and silicic acid groups on silica surface was the origin of the repulsive force between SiO₂ colloids and also their stability in water. Borkovec and co-workers⁷⁰ reported that SiO₂ particles underwent fast aggregation at pH 12 which they attributed to the dissolution of the hairy layers at high pH conditions. The aggregation behavior of c-SiO₂ at pH 12 is presented in Fig. 2c. In the absence of NaCl at pH 12, c-SiO₂ was stable, likely due to the EDL repulsion between the NPs. With the increase in NaCl concentration, the aggregation rates of the NPs increased and fast aggregation was achieved at 150 mM NaCl. This result supports the hypothesis that the presence of hairy layer on the surface of c-SiO₂, which was dissolved at pH 12, contributed to the remarkable stability of the c-SiO₂ NPs at pH 7.4. Zhang et al.¹⁴ investigated the surface silanol densities of both c-SiO₂ and f-SiO₂ through Fourier transform infrared spectroscopy and found that f-SiO₂ had a lower total silanol content (ca. 2.8 OH per nm²) than c-SiO₂ (ca. 4.5 OH per nm²). It is thus expected that the protruding hairy silanol and silicic acid groups were less abundant on f-SiO₂ NPs than on c-SiO₂ NPs. Therefore, the repulsive force caused by the hairy layers is likely to play a more important role in the stability of c-SiO₂ than that of f-SiO₂. We do not, however, rule out the possibility of other mechanisms for the stability of c-SiO₂ NPs, such as hydration force resulting from the hydrogen bonding network between silanol groups.⁷⁴

Phosphate adsorption enhanced the stability of CeO₂ and Al₂O₃ NPs

The aggregation kinetics of CeO₂ and Al₂O₃ NPs were investigated over a wide range of NaCl concentrations in the presence of phosphate and the attachment efficiencies are presented in Fig. 3. At NaCl concentrations of ca. 10–90 mM, the attachment efficiencies of both CeO₂ and Al₂O₃ NPs increased with increasing NaCl concentrations (reaction-limited regime) as a result of the screening of their surface charge. At NaCl concentrations higher than 90 mM, a further increase in the salt concentration did not result in any substantial increase in the attachment efficiencies (diffusion-controlled regime) since the energy barrier to aggregation was eliminated due to charge screening under these conditions. The critical coagulation concentrations (CCCs) that demarcated the reaction- and diffusion-limited regimes were ca. 90 mM for both CeO₂ and Al₂O₃ NPs in the presence of phosphate.

Fig. 3 Aggregation attachment efficiencies of (a) CeO₂ and (b) Al₂O₃ NPs as a function of NaCl concentration at pH 7.4 in phosphate and bicarbonate buffer. The dashed lines are extrapolated from the favorable and unfavorable aggregation regimes, and their intersections yield the CCCs.

In addition, aggregation experiments were carried out at pH 7.4 in the absence of phosphate in order to elucidate the role of phosphate adsorption on the stability of CeO₂ and Al₂O₃ NPs (Fig. 3). Reaction- and diffusion-limited regimes were also observed for both CeO₂ and Al₂O₃ NPs in bicarbonate buffer, indicating that the aggregation kinetics were controlled by electrostatic interactions in the absence of phosphate. The CCC values were determined to be ca. 2 mM for CeO₂ (Fig. 3a) and 25 mM for Al₂O₃ NPs (Fig. 3b), both considerably lower than their respective CCCs in the presence of phosphate. This comparison implies that the stability of CeO₂ and Al₂O₃ NPs was enhanced as a result of the adsorption of phosphate on the NP surface. As shown in Fig. 1b, the adsorption of phosphate resulted in a significant increase in the magnitude of the surface charge of CeO₂ NPs, thereby increasing the energy barrier to NP aggregation. For Al₂O₃ NPs, the surface charge was reversed in the presence of phosphate and, additionally, the magnitude of the surface charge was increased (as shown in the increase in the magnitude of EPM in Fig. 1b), thus raising the EDL repulsion between the NPs. Because the increase in the magnitude of surface charge due to the adsorption of phosphate was smaller for Al₂O₃ than for CeO₂ (Fig. 1b), the rise in CCC in the presence of phosphate was less pronounced for Al₂O₃ than for CeO₂ NPs (Fig. 3).

Deposition of CMP NPs on DOPC SLBs in the presence of phosphate

The deposition of the CMP NPs on DOPC SLBs was investigated in order to probe the propensity for CMP NPs to attach to lipid membranes. In our previous work, the DOPC bilayers were determined to be negatively charged at pH 7.²³ Fig. S2 of the ESI† presents the rates of frequency shift during the deposition of CMP NPs on the SLBs in the presence of phosphate. When the NaCl concentration was raised from 1 to 30 mM, the deposition rates of c-SiO₂ and f-SiO₂ NPs on SLBs increased due to the reduction of the EDL repulsion between the NPs and SLBs. In contrast, no deposition of CeO₂ and Al₂O₃ NPs on SLBs was detected over the NaCl concentration range of 1–100 mM.

By normalizing the deposition rates of CMP NPs on SLBs to the transport-limited deposition rates on PLL (presented in Fig. S3 of the ESI†), we obtained the deposition attachment efficiencies of the four NPs on SLBs, as presented in Fig. 4. For the two SiO₂ NPs, the attachment efficiencies increased when the NaCl concentration was increased from 1 to 100 mM. At NaCl concentrations higher than 100 mM, the attachment efficiency did not increase with a further increase in NaCl concentration, most likely due to the complete screening of surface charges under these high ionic strength conditions. Interestingly, for both CeO₂ and Al₂O₃ NPs, the attachment efficiency did not reach unity even at 300 mM NaCl.

Fig. 4 Deposition attachment efficiencies of (a) c-SiO₂, (b) f-SiO₂, (c) CeO₂, and (d) Al₂O₃ NPs on DOPC SLBs at pH 7.4 in phosphate buffer. Error bars represent standard deviations. The lines are meant to guide the eye. Asterisks (*) indicate that no frequency shift was detected on SLBs.

Comparing deposition kinetics of CMP NPs on DOPC SLBs in the presence and absence of phosphate

In order to obtain further insights into the low propensity for CeO₂ and Al₂O₃ NPs to deposit on SLBs in the presence of phosphate, additional deposition experiments on SLBs were performed in bicarbonate buffer. As shown in Fig. 5a and b, both c-SiO₂ and f-SiO₂ NPs had similar propensities to attach to SLBs in phosphate and bicarbonate buffers, owing to their similar surface charge in the two buffers (Fig. 1b).

Fig. 5 Deposition attachment efficiencies of (a) c-SiO₂, (b) f-SiO₂, (c) CeO₂, and (d) Al₂O₃ NPs on DOPC SLBs at pH 7.4 in phosphate and bicarbonate buffer. Error bars represent standard deviations. The data in phosphate buffer was reproduced from Fig. 4 for comparison. Asterisks (*) indicate that the attachment efficiencies were 0 (i.e., no frequency shift was detected on SLBs).

For the CeO₂ NPs, the deposition attachment efficiency in bicarbonate buffer increased from zero to close to unity when the NaCl concentration was raised from 1 to 100 mM (Fig. 5c) due to the reduction in the EDL repulsion between the NPs and SLBs. In the presence of phosphate, however, the deposition attachment efficiencies of CeO₂ NPs on SLBs were zero at both 1 and 100 mM NaCl. According to the EPM measurements (Fig. 1b), additional negative charge was imparted to the CeO₂ NPs through phosphate adsorption. Thus, the EDL repulsion between the CeO₂ NPs and SLBs may still be strong enough even at 100 mM NaCl to inhibit the deposition of the NPs.

For the deposition of Al₂O₃ NPs on SLBs, a more pronounced contrast in attachment efficiency was observed in phosphate and bicarbonate buffers (Fig. 5d). In bicarbonate buffer, Al₂O₃ was positively charged at both 1 and 100 mM NaCl (Fig. 1b). The attractive EDL and vdW interactions between Al₂O₃ NPs and SLBs resulted in close-to-unity attachment efficiencies at both 1 and 100 mM NaCl in bicarbonate buffer. In the presence of phosphate, however, the attachment of Al₂O₃ NPs on SLBs was not detected at both 1 and 100 mM NaCl due to the negative charge of Al₂O₃ NPs imparted by phosphate adsorption (Fig. 1b). The distinctly different deposition behaviors of both CeO₂ and Al₂O₃ NPs on SLBs in the presence and absence of phosphate shows that it is imperative to account for the influence of phosphate adsorption in the assessment of NP association with cell membranes.

No disruption of vesicles by CMP NPs detected

The attachment of NPs on cell membranes has been reported in some studies^75,76 to result in the damage of the membranes and the leakage of intracellular contents. Dye leakage experiments have been employed in several studies on the disruption of cell membranes by silica,³¹ gold,³⁰ TiO₂,³² and graphene oxide⁵⁴ NPs. Herein, the disruption of the supported vesicles upon the attachment of CMP NPs was investigated by QCM-D coupled with fluorescence detection measurements. This technique enables us to simultaneously monitor the deposition of CMP NPs on the supported vesicles using QCM-D and detect the disruption of the vesicles by measuring the dye leakage from the vesicles.⁵⁴ These experiments were only performed in phosphate buffer in the context of cell membrane disruption in biological systems. Since no attachment of CeO₂ and Al₂O₃ NPs on DOPC bilayers was expected to take place at 150 mM NaCl (i.e., ionic strength of physiological fluids), as shown in Fig. 4c and d, these experiments were only conducted for both SiO₂ NPs.

The frequency and dissipation shifts during the formation of a SVL and the deposition of f-SiO₂ NPs on the SVL are presented in Fig. 6a. Fig. 6b and c show the frequency and dissipation shifts and the dye concentration in the outflow of QCM-D, respectively, during the deposition of f-SiO₂ NPs on the SVL. The experimental protocol involves three stages: 1) deposition of dye-encapsulated vesicles on a gold-coated sensor to form a SVL (26–57 min, only in Fig. 6a); 2) removal of unadsorbed vesicles with buffer rinse (57–190 min); and 3) deposition of NPs on the SVL at 150 mM NaCl (213–257 min). The fluorescence intensity of the outflow was monitored from close to the end of stage 2 onwards (from 176 min). In a typical experiment, four QCM-D chambers were employed. In two of the chambers, deposition experiments were performed with a suspension of CMP NPs prepared in 150 mM NaCl. In the other two chambers, control experiments were conducted using the same procedure as the deposition experiments, except that a 150 mM NaCl solution was used in lieu of the NP suspension.

Fig. 6 (a) Frequency and dissipation shifts during the formation of a SVL (26–57 min), the deposition of f-SiO₂ NPs on the SVL at 150 mM NaCl and pH 7.4 (213–257 min), and the rupture of the vesicles when they were exposed to 32 mM Triton X-100 (277–301 min). (b) Frequency and dissipation shifts when supported vesicles were exposed to f-SiO₂ NPs (in 150 mM NaCl), 150 mM NaCl, and 32 mM Triton X-100. (c) Release of CF dye from supported vesicles upon exposure to f-SiO₂ NPs (in 150 mM NaCl), 150 mM NaCl, and 32 mM Triton X-100. Insert: Close-up of the release of dye in the f-SiO₂ NP deposition stage.

As shown in Fig. 6a and b, f-SiO₂ NPs deposited on the SVL continuously without eliciting significant rupture of the vesicles as evidenced from the monotonic decrease in frequency and increase in dissipation shifts (213–257 min). At the same time, no significant difference in dye leakage concentration was observed between the deposition and control experiments (inset of Fig. 6c, two-tailed t test, significance level of 0.05). As a positive control, the exposure of the vesicles to Triton (276 min) resulted in precipitous frequency and dissipation shifts (Fig. 6b), as well as a dramatic increase in fluorescence intensity (Fig. 6c), indicating that the vesicles had been ruptured by the surfactant. Hence, it can be inferred that the deposition of f-SiO₂ NPs on the SVL did not induce appreciable disruption of the lipid membrane. Similar experiments were performed for c-SiO₂ NPs and no significant dye leakage was observed during the deposition of c-SiO₂ NPs on the vesicles (Fig. S4†). These results demonstrate that both of SiO₂ NPs have a low propensity to disrupt DOPC bilayers at 150 mM NaCl despite their high affinity to associate with the model membranes under these conditions.

Conclusions

In this study, the aggregation behavior of four CMP NPs and their interactions with model cell membranes at neutral pH conditions were investigated. One important aspect of this study is to examine the influence of phosphate, a key component of biological fluids, on the surface charge and colloidal stability of the NPs, as well as on the nanoparticle–membrane interactions. The surface charge of both c-SiO₂ and f-SiO₂ NPs was not appreciably altered by phosphate as phosphate did not adsorb favorably on the NP surface. As a result, both SiO₂ NPs had similar aggregation and deposition kinetics on SLBs in the presence and absence of phosphate. Conversely, the CeO₂ NPs became more negatively charged while the Al₂O₃ NPs underwent charge reversal (positive to negative) in the presence of phosphate, both likely due to the adsorption of phosphate on the NPs through inner-sphere complexation. The enhancement of surface charge in the presence of phosphate led to an increase in the stability of CeO₂ and Al₂O₃ NPs and decreased the NPs' propensity to attach to SLBs. These result imply that the nonspecific interactions between NPs and cell membranes and potentially the biological effects of NPs may be controlled by phosphate. Thus, the role of phosphate needs to be accounted for in the assessment of the stability and cellular interactions of NPs. Additionally, the deposition of both c-SiO₂ and f-SiO₂ NPs on supported vesicles did not result in detectable disruption of the vesicles. Future studies that include a systematic variation of membrane properties (e.g., lipid composition and incorporation of proteins in membranes) and CMP NP properties (e.g., sizes and shapes) will enable additional mechanistic insights into the interactions between the nanoparticles and cell membranes. The effects of other major constituents in biological fluids (e.g., proteins) should also be investigated.

Acknowledgements

This work is supported by the Semiconductor Research Corporation (award 425-MC-2001, project 425.041) and National Science Foundation (CBET-1133559). We acknowledge Cabot Inc. for providing the CMP slurries used in this study, which are currently available through Dr. Farhang Shadman (University of Arizona). X. L. acknowledges funding support from the Gary Anderson Fellowship. We thank Dr. Kalina Hristova from the Department of Materials Science and Engineering at Johns Hopkins University for the use of the fluorescence spectrophotometer.

References

1. J. M. Steigerwald, S. P. Murarka and R. J. Gutmann, Chemical mechanical planarization of microelectronic materials, John Wiley & Sons, 2008.
2. M. Krishnan, J. W. Nalaskowski and L. M. Cook, Chem. Rev., 2010, 110, 178–204.
3. C. L. Lai and S. H. Lin, Chemosphere, 2004, 54, 235–242.
4. S. H. Lin and C. R. Yang, J. Hazard. Mater., 2004, 108, 103–109.
5. W. Den and C. P. Huang, Colloids Surf., A, 2005, 254, 81–89.
6. Y. Li, Microelectronic applications of chemical mechanical planarization, John Wiley & Sons, 2007.
7. D. Speed, P. Westerhoff, R. Sierra-Alvarez, R. Draper, P. Pantano, S. Aravamudhan, K. L. Chen, K. Hristovski, P. Herckes, X. Y. Bi, Y. Yang, C. Zeng, L. Otero-Gonzalez, C. Mikoryak, B. A. Wilson, K. Kosaraju, M. Tarannum, S. Crawford, P. Yi, X. Liu, S. V. Babu, M. Moinpour, J. Ranville, M. Montano, C. Corredor, J. Posner and F. Shadman, Environ. Sci.: Nano, 2015, 2, 227–244.
8. L. K. Limbach, R. Bereiter, E. Mueller, R. Krebs, R. Gaelli and W. J. Stark, Environ. Sci. Technol., 2008, 42, 5828–5833.
9. M. N. Shepard and S. Brenner, Ann. Occup. Hyg., 2014, 58, 251–265.
10. S. A. Brenner and N. M. Neu-Baker, J. Chem. Health Saf., 2014, 22, 10–19.
11. H. E. Bergna and W. O. Roberts, Colloidal silica: fundamentals and applications, CRC Press, 2005.
12. S. E. Pratsinis, Prog. Energy Combust. Sci., 1998, 24, 197–219.
13. P. B. Zantye, A. Kumar and A. K. Sikder, Mater. Sci. Eng., R, 2004, 45, 89–220.
14. H. Y. Zhang, D. R. Dunphy, X. M. Jiang, H. Meng, B. B. Sun, D. Tarn, M. Xue, X. Wang, S. J. Lin, Z. X. Ji, R. B. Li, F. L. Garcia, J. Yang, M. L. Kirk, T. Xia, J. I. Zink, A. Nel and C. J. Brinker, J. Am. Chem. Soc., 2012, 134, 15790–15804.
15. W. S. Lin, Y. W. Huang, X. D. Zhou and Y. F. Ma, Int. J. Toxicol., 2006, 25, 451–457.
16. M. G. Song, J. H. Lee, Y. G. Lee and J. H. Koo, J. Colloid Interface Sci., 2006, 300, 603–611.
17. L. Otero-Gonzalez, R. Sierra-Alvarez, S. Boitano and J. A. Field, Environ. Sci. Technol., 2012, 46, 10271–10278.
18. M. P. Vinardell, A. Sorde, J. Diaz, T. Baccarin and M. Mitjans, J. Nanopart. Res., 2015, 17.
19. A. E. Nel, L. Madler, D. Velegol, T. Xia, E. M. V. Hoek, P. Somasundaran, F. Klaessig, V. Castranova and M. Thompson, Nat. Mater., 2009, 8, 543–557.
20. K. L. Chen and G. D. Bothun, Environ. Sci. Technol., 2014, 48, 873–880.
21. K. Yang and Y. Q. Ma, Nat. Nanotechnol., 2010, 5, 579–583.
22. J. M. Troiano, L. L. Olenick, T. R. Kuech, E. S. Melby, D. H. Hu, S. E. Lohse, A. C. Mensch, M. Dogangun, A. M. Vartanian, M. D. Torelli, E. Ehimiaghe, S. R. Walter, L. Fu, C. R. Anderton, Z. H. Zhu, H. F. Wang, G. Orr, C. J. Murphy, R. J. Hamers, J. A. Pedersen and F. M. Geiger, J. Phys. Chem. C, 2015, 119, 534–546.
23. P. Yi and K. L. Chen, Environ. Sci. Technol., 2013, 47, 5711–5719.
24. X. F. Zhang and S. H. Yang, Langmuir, 2011, 27, 2528–2535.
25. B. Jing and Y. Zhu, J. Am. Chem. Soc., 2011, 133, 10983–10989.
26. R. Frost, G. E. Jonsson, D. Chakarov, S. Svedhem and B. Kasemo, Nano Lett., 2012, 12, 3356–3362.
27. K. H. Jacobson, I. L. Gunsolus, T. R. Kuech, J. M. Troiano, E. S. Melby, S. E. Lohse, D. Hu, W. B. Chrisler, C. J. Murphy, G. Orr, F. M. Geiger, C. L. Haynes and J. A. Pedersen, Environ. Sci. Technol., 2015, 49, 10642–10650.
28. P. R. Leroueil, S. A. Berry, K. Duthie, G. Han, V. M. Rotello, D. Q. McNerny, J. R. Baker, B. G. Orr and M. M. B. Holl, Nano Lett., 2008, 8, 420–424.
29. A. Lesniak, A. Salvati, M. J. Santos-Martinez, M. W. Radomski, K. A. Dawson and C. Åberg, J. Am. Chem. Soc., 2013, 135, 1438–1444.
30. C. M. Goodman, C. D. McCusker, T. Yilmaz and V. M. Rotello, Bioconjugate Chem., 2004, 15, 897–900.
31. H. Pera, J. M. Kleijn and F. A. M. Leermakers, Chem. Lett., 2012, 41, 1322–1324.
32. B. Y. Moghadam, W. C. Hou, C. Corredor, P. Westerhoff and J. D. Posner, Langmuir, 2012, 28, 16318–16326.
33. S. L. Tobey and E. V. Anslyn, Org. Lett., 2003, 5, 2029–2031.
34. A. Albanese and W. C. W. Chan, ACS Nano, 2011, 5, 5478–5489.
35. X. S. Liu, Y. J. Chen, H. Li, N. Huang, Q. Jin, K. F. Ren and J. Ji, ACS Nano, 2013, 7, 6244–6257.
36. T. Masui, K. Fujiwara, K. Machida, G. Adachi, T. Sakata and H. Mori, Chem. Mater., 1997, 9, 2197–2204.
37. K. M. Parida, A. C. Pradhan, J. Das and N. Sahu, Mater. Chem. Phys., 2009, 113, 244–248.
38. J. Murphy and J. P. Riley, Anal. Chim. Acta, 1962, 27, 31–36.
39. K. A. Huynh and K. L. Chen, Environ. Sci. Technol., 2011, 45, 5564–5571.
40. P. Yi and K. L. Chen, Langmuir, 2011, 27, 3588–3599.
41. K. L. Chen, S. E. Mylon and M. Elimelech, Environ. Sci. Technol., 2006, 40, 1516–1523.
42. K. L. Chen and M. Elimelech, Langmuir, 2006, 22, 10994–11001.
43. L. L. Hoekstra, R. Vreeker and W. G. M. Agterof, J. Colloid Interface Sci., 1992, 151, 17–25.
44. N. B. Saleh, L. D. Pfefferle and M. Elimelech, Environ. Sci. Technol., 2008, 42, 7963–7969.
45. I. Chowdhury, M. C. Duch, N. D. Mansukhani, M. C. Hersam and D. Bouchard, Environ. Sci. Technol., 2013, 47, 6288–6296.
46. H. Holthoff, S. U. Egelhaaf, M. Borkovec, P. Schurtenberger and H. Sticher, Langmuir, 1996, 12, 5541–5549.
47. N. J. Cho, C. W. Frank, B. Kasemo and F. Höök, Nat. Protoc., 2010, 5, 1096–1106.
48. P. Yi and K. L. Chen, Environ. Sci. Technol., 2013, 47, 5711–5719.
49. X. L. Qu, P. J. J. Alvarez and Q. L. Li, Environ. Sci. Technol., 2012, 46, 13455–13462.
50. C. A. Keller and B. Kasemo, Biophys. J., 1998, 75, 1397–1402.
51. K. L. Chen and M. Elimelech, Environ. Sci. Technol., 2008, 42, 7607–7614.
52. A. J. de Kerchove and M. Elimelech, Macromolecules, 2006, 39, 6558–6564.
53. W. Stumm and J. J. Morgan, Aquatic chemistry: an introduction emphasizing chemical equilibria in natural waters, John Wiley, 1981.
54. X. Liu and K. L. Chen, Langmuir, 2015 DOI:10.1021/acs.langmuir.5b02414.
55. C. A. Schnaitman, J. Bacteriol., 1971, 108, 545–552.
56. M. Elimelech, J. Gregory, X. Jia and R. A. Williams, Particle Deposition and Aggregation: Measurement, Modelling and Simulation, Butterworth-Heinemann, Oxford, England, 1995.
57. T. Hiemstra, P. Venema and W. H. VanRiemsdijk, J. Colloid Interface Sci., 1996, 184, 680–692.
58. M. Kosmulski, J. Colloid Interface Sci., 2002, 253, 77–87.
59. M. Kosmulski, J. Colloid Interface Sci., 2011, 353, 1–15.
60. S. Tsunekawa, T. Fukuda and A. Kasuya, J. Appl. Phys., 2000, 87, 1318–1321.
61. G. J. Zhang, J. F. Yang and T. Ohji, J. Am. Ceram. Soc., 2001, 84, 1395–1397.
62. S. W. Wang, M. Di Ventra, S. G. Kim and S. T. Pantelides, Phys. Rev. Lett., 2001, 86, 5946–5949.
63. B. B. Johnson, A. V. Ivanov, O. N. Antzutkin and W. Forsling, Langmuir, 2002, 18, 1104–1111.
64. E. Laiti, P. Persson and L. O. Ohman, Langmuir, 1996, 12, 2969–2975.
65. R. J. Hunter, Zeta potential in colloid science: principles and applications, Academic press, 2013.
66. K. Van Hoecke, J. T. K. Quik, J. Mankiewicz-Boczek, K. A. C. De Schamphelaere, A. Elsaesser, P. Van der Meeren, C. Barnes, G. McKerr, C. V. Howard, D. Van De Meent, K. Rydzynski, K. A. Dawson, A. Salvati, A. Lesniak, I. Lynch, G. Silversmit, B. De Samber, L. Vincze and C. R. Janssen, Environ. Sci. Technol., 2009, 43, 4537–4546.
67. G. Cornelis, B. Ryan, M. J. McLaughlin, J. K. Kirby, D. Beak and D. Chittleborough, Environ. Sci. Technol., 2011, 45, 2777–2782.
68. K. Higashitani, M. Kondo and S. Hatade, J. Colloid Interface Sci., 1991, 142, 204–213.
69. J. Depasse, J. Colloid Interface Sci., 1999, 220, 174–176.
70. M. Kobayashi, F. Juillerat, P. Galletto, P. Bowen and M. Borkovec, Langmuir, 2005, 21, 5761–5769.
71. W. A. Ducker, T. J. Senden and R. M. Pashley, Nature, 1991, 353, 239–241.
72. G. Vigil, Z. H. Xu, S. Steinberg and J. Israelachvili, J. Colloid Interface Sci., 1994, 165, 367–385.
73. J. Israelachvili and H. Wennerstrom, Nature, 1996, 379, 219–225.
74. J. P. Chapel, Langmuir, 1994, 10, 4237–4243.
75. Y. S. Lin and C. L. Haynes, J. Am. Chem. Soc., 2010, 132, 4834–4842.
76. T. Yu, A. Malugin and H. Ghandehari, ACS Nano, 2011, 5, 5717–5728.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5en00176e

---