Article pubs.acs.org/Langmuir
Adsorption and Friction Behavior of Amphiphilic Polymers on Hydrophobic Surfaces Giacomo Fontani,†,‡ Roberto Gaspari,†,§ Nicholas D. Spencer,‡ Daniele Passerone,† and Rowena Crockett*,† †
Swiss Federal Laboratories for Materials Science and Technology, Empa, Ueberlandstrasse 129, 8600 Duebendorf, Switzerland Department of Materials, ETH Zurich, Wolfgang-Pauli-Strasse 10, 8093 Zurich, Switzerland
‡
S Supporting Information *
ABSTRACT: The ability of amphiphilic polymers to self-assemble and form a gel or gel-like layer has been investigated by means of both experimental and theoretical studies on alkylated derivatives of poly(acrylic acid). Experiments were performed to determine the relationship between amphiphilic polymer chemistry, structure, water retention, and friction in the presence of hydrophobic substrates. The results indicate that the amphiphilic polymer forms a water-enriched, friction-reducing adsorbed layer on hydrophobic surfaces. The shear moduli and viscosities of the adsorbed layers, as determined by fitting the Voigt model to QCM-D data, were consistent with the presence of a gel. Computational studies on HPAA-12 were performed and are consistent with the presence of adsorbed conformations, in which the lowest free energy in the model corresponded to a partially adsorbed molecule, with a small fraction of hydrophobic side chains being compelled, for configurational reasons, to point into the bulk water. This would support the possibility of the formation of either a gel-like layer or surface aggregation. However, because the adsorption experiments showed no evidence of aggregation, this strongly suggests the formation of a gel.
1. INTRODUCTION Among the various proposals that have been made to improve the lubricating ability of water, the most common involves the adsorption of brush-like polymers onto the contacting surfaces.1 It has been proposed that the ability of these brushes to maintain a layer of water at the surface, together with the entropic and osmotic properties of the hydrated polymers, results in the very low values of friction that have been observed.1,2 Another strategy that has been studied is the formation of gels from hydrophilic polymers.3,4 These gels also lead to low friction between sliding contacts in aqueous solutions. However, as the polymers are covalently cross-linked, they cannot easily be replenished during sliding, in contrast to the brush-like systems. We have previously proposed a mechanism for natural lubrication in articular joints that involved the formation of a gel-like layer of hyaluronan.5,6 This mechanism involves the intertwining of the hyaluronan chains, enriched at the articular surface, together with phospholipids, to form a highly hydrated layer. To investigate the feasibility of such a mechanism, an experimental model system was constructed, based on the simplest chemical components. The aim was to determine whether a gel-like layer could form and, in turn, lead to a reduction in friction. Hyaluronan is a highly negatively charged polymer that has a secondary structure resulting from the hydrophobic faces on each saccharide unit.7 These two functions are mimicked in the model system by an alkylated © 2013 American Chemical Society
poly(acrylic acid), which contains a negatively charged backbone in aqueous solution and hydrophobic alkyl side chains, allowing the possibility of secondary-structure formation. This polymer is not intended to mimic hyaluronan but only the chemical properties that may influence the formation of a gel layer on adsorption. The mechanical properties of hydrogels that maintain their structure through hydrophobic interactions have previously been investigated by means of acrylic acid and methyl methacrylate.8 The gels were formed by polymerizing varying ratios of the two monomers using 2,2′-azobis(amidinopropane) as initiator. In recent studies, methylcellulose and hydroxypropyl methylcellulose were shown to form gels as a result of hydrophobic interactions.9 As expected, the ability of the hydrogel to adsorb water decreases as the hydrophobicity increases; however, the elastic modulus also increases.8 The aim of the present work is to study a system that combines the ability to form a highly hydrated layer of hydrogels with loadbearing capacity and to use it for tribological applications. The advantage of a low-friction, gel-like layer rather than a crosslinked gel is that it can re-form from solution after being removed by sliding of the contacting surfaces. Charged polymers adsorb onto oppositely charged surfaces in different ways, depending on the ionic strength of the Received: January 20, 2013 Revised: March 18, 2013 Published: March 19, 2013 4760
dx.doi.org/10.1021/la400263r | Langmuir 2013, 29, 4760−4771
Langmuir
Article
The sensor was coated with trichloro(octadecyl)silane (ODTS) by immersion in a solution of anhydrous decahydronaphthalene (“decalin”, mixture of cis and trans) (20 mL) containing ODTS (3 drops) and carbon tetrachloride (2 drops) for 15 min, washed with chloroform, and then reimmersed in the same solution for a further 15 min before finally washing with chloroform and drying with a stream of nitrogen. Decahydronaphthalene and carbon tetrachloride were dried and distilled prior to use. TInAS is a mass-sensing device based on Fabry−Pérot white light interference, in which interference fringes caused by partial reflections at optical interfaces, such as transparent dielectric multilayer structures, are analyzed. This can be used to determine small changes of film thickness due to molecular adsorption. The resulting effective optical thickness D can be transformed into adsorbed mass, M, per unit area, by the equation:
solution, the charge density on the polymer, and the polymer concentration in solution.10 At low ionic strength, the polymers adsorb onto the surface in a straight conformation, as the repulsion between charges causes the polymer to stretch when in solution. At high ionic strength, screening of the charges results in a more folded polymer structure, both in solution and on the surface.10 Amphiphilic polymers have been shown to behave in a similar manner when adsorbed onto charged surfaces.11 At high salt and polymer concentrations, multilayers or aggregates form on the surface.11 In this work, the adsorption behavior of the amphiphilic, alkylated poly(acrylic acid), a hexyl derivative, from aqueous solutions onto hydrophobic surfaces was investigated both by experimental techniques, using a combination of the transmission interferometric adsorption sensor (TInAS) and the quartz crystal microbalance with dissipation monitoring (QCM-D), and by modeling the polymer at the surface by means of classical molecular dynamics (MD) methods. Over the last two decades, many computer simulation studies have been carried out on the self-assembly of amphiphilic molecules.12−22 Smit et al. carried out MD simulations of the spontaneous aggregation of surfactants for a simple oil/water/ surfactant system and analyzed the detailed structure of a water/oil interface in the presence of micelles.12,13 In coarsegrained MD studies with “hard” non-bonded interactions (Lennard−Jones potential and electrostatics), self-assembling of model amphiphilic polymers in water and the elastic properties and interfacial tension of the assembled layers have been studied in some detail.17,19,22 A very long-range attraction between hydrophobic surfaces immersed in water has been observed in a variety of experiments.23,24 Conventional MD simulation has been applied to systems including flat hydrophobic surfaces25 and atomically rough hydrophobic surfaces.26 According to these reports, the surfaces produce density oscillations and significant orientational preferences in the confined water. The hydrophobic effect as well as the influence of kosmotropes and chaotropes on the structure of water have been implemented in coarse-grained simulations by Miller et al.27
M=ρ×D where ρ is the density of the adsorbate. Because adsorbates may remain partially solvated after adsorption, the density can be expressed in terms of the refractive-index difference between the adsorbate, nA, and the solvent, nC, normalized to the concentration dependence of the refractive index in the mixture, dn/dc:29 n −n M = A dn C × D dc
For adsorbed polymer, a dn/dc value of 0.182 cm3 g−1 was used. This is the value that has been used for polyacrylamide.30,31 The TInAS measurements were conducted in a flow cell with a flow rate of 20 μL min−1, using a 100 mM potassium dihydrogen phosphate/disodium hydrogen phosphate buffer solution at pH 7 and a polymer concentration of 0.1 mg mL−1 and 1.0 mg mL−1. The ionic strength of the buffer was approximately 0.5 mol dm−3, and the maximum possible change caused by the polymer, for fully protonated HPAA-26 at 1.0 mg mL−1, was approximately 1.2%. Transmission spectra were recorded with a USB2000 spectrometer (Ocean Optics, Dunedin, FL), using halogen lamp illumination and a fiber optic readout. The QCM-D measurements were performed with a QCM-D (QE301: electronics unit, QAFC301: axial flow chamber, QSoft 301: software version) from Q-Sense AB (Göteborg, Sweden). Quartz crystals (diameter = 25 mm) coated with SiO2 (QSX 303, LOT Oriel Group, Germany) with a fundamental frequency of 5 MHz were used to study the mass of the polymer adsorbed from the buffer solution at pH 7. The quartz crystal microbalance is a mass-sensing device that allows adsorption kinetics to be studied by focusing on the change in the resonant frequencies of vibrating quartz crystals. Frequencies can be measured over different overtones (n = 1 to n = 13), which have different surface sensitivities associated with them. The mass of the adsorbed polymer, along with its associated water molecules, can be calculated by the Sauerbrey equation:32
2. MATERIALS AND METHODS 2.1. Experimental Details. All reagents were purchased from Sigma-Aldrich GmbH (Buchs, Switzerland) and used without further purification unless otherwise stated. To synthesize hexyl poly(acrylic acid) with a substitution of 1 hexyl group in 26 carboxylate groups (HPAA-26), poly(acrylic acid) (Mw = 15 000) (1.8 g, 0.025 mol of monomer) was dissolved in 20 mL of freshly distilled, anhydrous DMSO containing hexylamine, (1.26 g, 0.025 mol), after which diisopropylcarbodiimide (0.32 g, 0.0025 mol), dissolved in 20 mL of dichloromethane, was added dropwise to the DMSO solution with stirring. Substitutions of 1 in 12 (HPAA-12) and 1 in 5 (HPAA-5) were achieved using 0.63 g (0.005 mols) and 1.58 g (0.0125 mols) of diisopropylcarbodiimide in dichloromethane (20 mL), respectively. The resulting solutions were then stirred overnight at room temperature. Sodium hydroxide (1 M, 20 mL) was added, and stirring continued for 1 h. Dichloromethane was then removed under reduced pressure and the sodium salts of the hexylated poly(acrylic acid)s precipitated with ethanol. The amphiphilic polymers were washed with ethanol and dried under reduced pressure. The products were characterized by means of infrared spectroscopy (Bio-Rad FTS 6000, Bio-Rad Laboratories AG, Switzerland) and 1H NMR (Bruker ASX-400 MHz). The yields were 56 ± 5%. TInAS measurements were performed with a homemade device.28 A TInAS sensor chip with a 25 nm thick aluminum mirror layer and a 3 μm amorphous SiO2 spacer layer was used for the measurements.28
mSauberbrey = − C ×
Δf n
where mSauerbrey is the wet mass of the polymer adsorbed, Δf is the change in frequency of the quartz crystal upon adsorption, C is the characteristic constant of the instrument, and n is the shear wavenumber. The QCM-D is an extension of the quartz-crystal microbalance designed for liquid−solid-interface studies, which uses a pulsed excitation of the quartz crystal to determine both the mass of the adsorbed layer and a dissipation factor associated with the viscoelastic properties of the adsorbed polymer. The Sauerbrey equation is valid only for thin and rigid films; when in the presence of a viscoelastic layer, both the change in frequency of the crystal and the dissipation factor are strongly dependent on the shear viscosity and the shear elasticity of the adsorbed layer. To determine the mechanical properties of the adsorbed layer, the Voigt model was used, in which two viscoelastic layers covering the surface of a piezoelectric plate oscillate in a pure shear mode in a bulk liquid. 4761
dx.doi.org/10.1021/la400263r | Langmuir 2013, 29, 4760−4771
Langmuir
Article
The model was simplified to a single viscoelastic layer, assuming that the OTDS coating is sufficiently thin and rigid to be neglected. The shift of the quartz resonance frequency and the dissipation factor were recalculated according to the equations:33
Δf ≈ −
⎧ 1 ⎪ η3 ⎨ + 2πρ0 h0 ⎪ ⎩ δ3
poly(acrylic acid) chain was neutralized with 71 sodium counterions (Na+). This corresponded to a pH of 7. An additional model was constructed by adding 30 sodium counterions, Na+, and 30 chloride counterions, Cl− to increase the ionic strength by 0.2. Simulations were performed with both models. An octane monolayer was used as a model of the hydrophobic surface, to imitate the ODTS monolayer with the alkyl chains oriented perpendicular to the surface. The octane chains were arranged parallel to the x-axis to form a regular lattice and the central atoms of the chains (C4 and C5) were fixed to simulate a rigid wall. The stable structure of the surface was determined with a series of geometry optimization runs, and the equilibrium structure was found to exhibit a spacing of 4.5 Å between adjacent octane molecules. The dynamics of the octane atoms was explicitly taken into account in the simulation.25 As a first step, an equilibration in implicit solvent was performed, in which all the water molecules were replaced by a similar dielectric constant of 80 (εrH2O (20 °C) = 80.1). This allowed an acceleration of the MD runs by excluding the computational effort in calculating the force field for each water molecule and by preventing diffusive effects. The starting configuration was equilibrated for 1000 ps at room temperature and constant volume. This was followed by a metadynamics run in implicit solvent, using as CVs the gyration radius of the amphiphilic polymer, rg, and the distance between its center of mass and the closest y-plane containing the octane carbon atoms, d. The run lasted 100 ns, and a height of 0.4 kcal mol−1 and a width of 0.35 units were chosen for the added Gaussians. A classical dynamics run was then performed in explicit solvent, using standard TIP3P water molecules.38 The equilibrium implicit solvent configuration was chosen as the starting configuration of the MD run, and the simulation was terminated when no further conformational changes were observed.39 Subsequently, to study the influence of the solvent toward the configuration of the bulk polymer, HPAA-12 was immersed in two different solvents, namely a box of TIP3P water molecules and toluene molecules. For each solvent, a metadynamics run was performed using the HPAA-12 gyration radius as the collective variable and starting from a folded conformation.
2 ⎡ ⎤⎫ ηω ⎛ η ⎞2 ⎪ j ⎢h ρ ω − 2h ⎜ 3 ⎟ ⎥⎬ j j j 2 2 2⎥ ⎢⎣ ⎝ δ3 ⎠ μj + ω ηj ⎦⎪ j = 1,2 ⎭
∑
⎧ 1 ⎪ η3 ⎨ + ΔD ≈ − 2πfρ0 h0 ⎪ ⎩ δ3
2 ⎡ ⎛ η ⎞2 ⎤⎫ ηω ⎪ j 3 ⎢ ⎥⎬ ∑ ⎢2hj⎜ ⎟ 2 2 2⎥ ⎪ δ μ ω η + ⎝ 3⎠ j j ⎦⎭ j = 1,2 ⎣
where h and ρ are the thickness and the density of the layers, η and δ are the shear viscosity and the viscous penetration depth, and μ is the elastic shear modulus (0 = quartz plate, 1, 2 = adsorbed layer, 3 = bulk liquid). The Voigt model was fitted to the frequency and dissipation for the third, fifth, and seventh overtones using the Q-sense software (Q-tools, Version 2.0.1).29 The crystals were plasma-treated for 4 min and subsequently coated with ODTS using the same protocol described above. HPAA buffer solutions with concentrations of 0.1 mg mL−1 and 1 mg mL−1 were used with a flow rate of 20 μL min−1, and the chamber temperature was maintained at 25 °C during all of the measurements. Friction measurements were carried out at normal atmospheric conditions (24 °C and 50% relative humidity) on a home-made reciprocating sliding rig under a normal load of 0.2 N at a sliding velocity of 0.02 m/s, using poly(methyl methacrylate) (PMMA) cylinders of 10 mm in height (radius 5 mm, maximum contact pressure 2.1 N mm−2) sliding against glass slides that had either been cleaned using an oxygen plasma or coated with ODTS in the same way as the TInAS silicon sensors. The PMAA was supplied by Bö rling Kunststofftechnik AG, Switzerland, and had a Young’s Modulus of 3.2 GPa, a roughness (Ra) of
