Solvent-Dependent Friction Force Response of Poly(ethylenimine

May 1, 2009 - Lateral and normal forces between a surface-bound, brushlike copolymer, poly(ethylenimine)-graft-poly(ethylene glycol) (PEI-g-PEG), and ...
0 downloads 0 Views 1MB Size
Download PDF
pubs.acs.org/Langmuir © 2009 American Chemical Society

Solvent-Dependent Friction Force Response of Poly(ethylenimine)-graftpoly(ethylene glycol) Brushes Investigated by Atomic Force Microscopy Michael A. Brady,† F. T. Limpoco,‡ and Scott S. Perry†,* †

Department of Materials Science & Engineering and ‡Department of Chemistry, University of Florida, Gainesville, Florida 32611 Received January 29, 2009. Revised Manuscript Received March 30, 2009

Lateral and normal forces between a surface-bound, brushlike copolymer, poly(ethylenimine)-graft-poly(ethylene glycol) (PEI-g-PEG), and a silica colloidal probe were investigated with atomic force microscopy (AFM) and related to the relative mass of the solvent absorbed within the polymer as measured with the quartz crystal microbalance. PEI-gPEG was adsorbed onto an oxide-passivated silicon wafer through its exposure to physiologically buffered solutions of the polymer. Frictional forces were measured between the colloidal probe and the substrate by AFM as the polarity of the solvent was systematically varied. Reduced friction forces and greater film thicknesses were encountered under solvents of higher polarity, which are attributed to the extended conformation of the brushlike copolymer under these conditions. Lateral and normal forces detected between the colloidal probe and this surface-bound PEI-g-PEG were found to be similar under certain solvent conditions to those measured for poly(L-lysine)-graft-poly(ethylene glycol), a brushlike copolymer with a different molecular architecture. To this end, friction force studies of both symmetric and asymmetric PEI-g-PEG-coated interfaces served to identify the contributions of conformational and bridging effects in the observed tribological behavior.

1. Introduction The lubrication of most solid surfaces is achieved primarily through the use of hydrocarbon oils. While the use of hydrocarbon oils is acceptable in many instances, alternative forms of lubrication must be developed for more technologically specific applications such as moving assemblies designed for operation in space, the food packaging industry, and throughout the biomedical field. Vapor and solid phase lubrication are being actively researched for high temperature and space-based applications; however, alternative approaches to liquid phase lubrication are needed for operation in biomedical or other “clean” environments. Here, lubrication in an aqueous environment is of particular interest, and biology provides a number of successful examples for potential approaches;namely, the presence/activity of brushlike biopolymers at or near interfaces subjected to interfacial shear.1,2 In the past, very low friction forces and coefficients of kinetic friction have been measured at surfaces in sliding motion in aqueous environments following adsorption of synthetic polymer brushes, as demonstrated through studies employing atomic force *To whom correspondence should be addressed. (1) Kumar, P.; Oka, M.; Toguchida, J.; Kobayashi, M.; Uchida, E.; Nakamura, T.; Tanaka, K. J. Anat. 2001, 199, 241–250. (2) Mow, V. C.; Gu, W. Y.; Chen, F. H. In Basic Orthopaedic Biomechanics and Mechano-Biology, 3rd ed.; Mow, V. C., Huiskes, R., Eds.; Lippincott, Williams, and Wilkins: Philadelphia, 2005; pp 182-198. (3) Limpoco, F. T.; Advincula, R. C.; Perry, S. S. Langmuir 2007, 23, 12196– 12201. (4) Gourdon, D.; Lin, Q.; Oroudjev, E.; Hansma, H.; Golan, Y.; Arad, S.; Israelachvili, J. Langmuir 2008, 24, 1534–1540. (5) Vyas, M. K.; Schneider, K.; Nandan, B.; Stamm, M. Soft Matter 2008, 4, 1024–1032. :: (6) Berger, R.; Cheng, Y.; Forch, R.; Gotsmann, B.; Gutmann, J. S.; Pakula, T.; :: Rietzler, U.; Schartl, W.; Schmidt, M.; Strack, A.; Windeln, J.; Butt, H.-J. Langmuir 2007, 23, 3150–3156. (7) Kobayashi, M.; Yamaguchi, H.; Terayama, Y.; Wang, Z.; Kaido, M.; Suzuki, A.; Takahara, A. J. Soc. Rheo., Jpn. 2008, 36, 107–112. :: :: :: (8) Lee, S.; Muller, M.; Ratoi-Salagean, M.; Voros, J.; Pasche, S.; De Paul, S. M.; Spikes, H. A.; Textor, M.; Spencer, N. D. Tribol. Lett. 2003, 15, 231–239.

Langmuir 2009, 25(13), 7443–7449

microscopy (AFM),3-7 pin-on-disk tribometry,8,9 and surface forces apparatus (SFA),4,10-12 as well as through computational and theoretical work.13-16 These same polymer brushes have also been studied to increase the protein resistance of the surface.17-24 Both the structure and the conformation of adsorbed polymer brushes were found to influence observed interfacial properties. In the present study, the modification of normal and frictional forces by the adsorption of a polymer brush with a branched backbone at oxide interfaces was investigated on the molecular scale through systematic considerations of solvent-induced conformational changes. Poly(ethylenimine)-graft-poly(ethylene glycol) (PEI-g-PEG) is a polycationic copolymer that physisorbs onto negatively charged surfaces through Coulombic and non-Coulombic interactions.22,23 At physiological pH (7.4), the PEI backbone becomes :: (9) Muller, M.; Lee, S.; Spikes, H. A.; Spencer, N. D. Tribol. Lett. 2003, 15, 395– 405. (10) Raviv, U.; Giasson, S.; Kampf, N.; Gohy, J.-F.; Jer^ome, R.; Klein, J. Nature (London) 2003, 425, 163–165. (11) Drobek, T.; Spencer, N. D. Langmuir 2008,, 24, 1484–1488. (12) Raviv, U.; Giasson, S.; Kampf, N.; Gohy, J.-F.; Jer^ome, R.; Klein, J. Langmuir 2008, 24, 8678–8687. :: (13) Kreer, T.; Muser, M. H.; Binder, K.; Klein, J. Langmuir 2001, 17, 7804– 7813. :: (14) Kreer, T.; Muser, M. H. Wear 2003, 254, 827–831. (15) Sokoloff, J. B. Macromolecules 2007, 40, 4053–4058. (16) Sokoloff, J. B. J. Chem. Phys. 2008, 129, 014901/1–014901/9. :: :: (17) Lee, S.; Voros, J. Langmuir 2005, 21, 11957–11962. (18) Michel, R.; Pasche, S.; Textor, M.; Castner, D. G. Langmuir 2005, 21, 12327–12332. (19) Fukai, R.; Dakwa, P. H. R.; Chen, W. J. Polym. Sci., Part A: Polym. Chem. 2004, 42, 5389–5400. (20) Drobek, T.; Spencer, N. D.; Heuberger, M. Macromolecules 2005, 38, 5254– 5259. (21) Pasche, S.; Textor, M.; Meagher, L.; Spencer, N. D.; Griesser, H. J. Langmuir 2005, 21, 6508–6520. (22) Nnebe, I. M.; Tilton, R. D.; Schneider, J. W. J. Colloid Interface Sci. 2004, 276, 306–316. (23) Claesson, P. M.; Blomberg, E.; Paulson, O.; Malmsten, M. Colloids Surf., A 1996, 112, 131–139. (24) Ai, H.; Pink, J. J.; Shuai, X.; Boothman, D. A.; Gao, J. Wiley InterScience (Online) 2005, 303–312.

Published on Web 05/01/2009

DOI: 10.1021/la900371k

7443

Article

positively charged through protonation of amine groups; SiO2 surfaces are negatively charged under these conditions as the pH exceeds its isoelectric point. As such, spontaneous adsorption of PEI-g-PEG onto SiO2 occurs, thus offering the opportunity to reduce the interfacial kinetic friction in aqueous environments. Prior studies documenting the low friction detected at interfaces modified with poly(L-lysine)-graft-poly(ethylene glycol) (PLL-gPEG) as a function of solvent and molecular architecture8,9,25-27 provide further support for this approach. While PLL-g-PEG is known to take on a rodlike structure when adsorbed onto oxide surfaces due to its linear PLL backbone, PEI-g-PEG is expected to have a more globular structure (Figure 1), thus presenting an alternative in adsorption architecture with the opportunity to study its effects on tribological performance. Hartung et al. conducted an analogous study in which the tribological properties of PLL-g-PEG and PAAm-g-PEG, with differing backbone architectures, were compared.28 The more rigid PAAm backbone resulted in less adsorption of PEG-grafted copolymers, producing less lubricious surfaces as compared to PLL-g-PEG. The structure of PEI-g-PEG consists of a PEI backbone with PEG side chains grafted at the amine groups of PEI. The specific copolymer used in this study has a PEI molecular mass of 25 kDa, a grafting ratio of 3.5 ethylenimine units per PEG side chain, and a PEG molecular mass of 4 kDa. In this study, interfacial kinetic friction and normal forcedisplacement measurements were conducted using AFM as a function of solvent for PEI-g-PEG-modified silica microsphere vs oxide-passivated silicon interfaces. Microsphere probes were employed to avoid the extreme contact pressures (GPa) encountered with typical sharp probes. The amount of solvent absorbed within the PEI-g-PEG brush was detected as a function of solvent polarity using the quartz crystal microbalance (QCM) and related to the friction and normal force behavior. This work has demonstrated that, like PLL-g-PEG, PEI-g-PEG exhibits extremely lubricious behavior under good solvent conditions due to the stretched conformation of the brush in such an environment and that friction is greatly reduced for a symmetric PEI-g-PEGmodified interface as compared to that for an asymmetric one.

2. Experimental Section 2.1. Preparation of PEI-g-PEG. PEI(25)-g[3.5]-PEG(4)

:: was synthesized by SurfaceSolutionS GmbH (Zurich, Switzerland) following a modified procedure for the synthesis of PLL-gPEG,31 producing brushlike copolymers with a PEI molecular mass of 25 kDa, a grafting ratio of 3.5 ethylenimine units/PEG side chain, and a PEG molecular mass of 4 kDa. The grafting ratio was determined by two methods: from the ratio of the integrated peaks of the 1H NMR chemical shifts of the terminal methyl protons of PEG (-OCH3) and the methylene protons of the PEI residues (-NCH2CH2-) and from the C:N ratio from total elemental analysis. PEI-g-PEG-coated silicon wafer substrates were likewise analyzed by SurfaceSolutionS using XPS and variable angle spectroscopic ellipsometry (VASE). The C:N atom(25) Yan, X.; Perry, S. S.; Spencer, N. D.; Pasche, S.; De Paul, S. M.; Textor, M.; Lim, M. S. Langmuir 2004, 20, 423–428. :: (26) Muller, M. T.; Yan, X.; Lee, S.; Perry, S. S.; Spencer, N. D. Macromolecules 2005, 38, 5706–5713. :: (27) Muller, M. T.; Yan, X.; Lee, S.; Perry, S. S.; Spencer, N. D. Macromolecules 2005, 38, 3861–3866. :: (28) Hartung, W.; Drobek, T.; Lee, S.; Zurcher, S.; Spencer, N. D. Tribol. Lett. 2008, 31, 119–128. (29) De Smedt, S. C.; Demeester, J.; Hennink, W. E. Pharm. Res. 2000, 17(2), 113–126. (30) Petersen, H.; Fechner, P. M.; Fischer, D.; Kissel, T. Macromolecules 2002, 35(18), 6867–6874. :: :: (31) Pasche, S.; DePaul, S. M.; Voros, J.; Spencer, N. D.; Textor, M. Langmuir 2003, 19, 9216–9225.

7444 DOI: 10.1021/la900371k

Brady et al.

Figure 1. (A) Schematic structure of PEI-g-PEG, based on the branched PEI structure reported previously.29 Note that the PEI molecular weight, PEG molecular weight, and grafting ratio are represented qualitatively. (B) Schematic depiction of the adsorption of PEI-g-PEG on SiO230 surfaces at pH values greater than 2 through the electrostatic attraction between the negatively charged surface and the positively charged amine groups of the PEI backbone. ic concentration ratio measured from XPS was 12-18% higher than what was found by total elemental analysis, across several grafting ratios examined (2.5-10). This is consistent with a layered structure in which the PEI backbone lies close to the surface, while the PEG side chains extend away from the surface, resulting in the attenuation of the N signal. Dry ellipsometric thicknesses of adsorbed PEI-g-PEG films ranged from 1.1 to 1.7 nm, decreasing with the ethylene imine/PEG grafting ratio; this is consistent with a higher electrostatic affinity of the PEI backbone to the substrate as more amino groups become available with increased grafting ratio. These values changed very little (0.03 nm, maximum) after exposure of the PEI-g-PEG film to human serum, indicating excellent resistance to serum adsorption. 2.2. Preparation of PEI-g-PEG-Coated Substrates. Silicon (100) wafers were cleaved to the appropriate size (approximately 0.3 cm  0.3 cm) and treated by the following cleaning procedure: sonication in acetone for 5 min and 2-propanol for 5 min, copious rinsing with ultrapure water (18.2 MΩ cm) (Barnstead International, Dubuque, IA), exposure to fresh piranha solution (30% H2O2/70% H2SO4) at 80 °C for 10 min, copious rinsing with ultrapure water, drying under a nitrogen flow, and exposure to an O2/H2O2 plasma generated with a PDCLangmuir 2009, 25(13), 7443–7449

Brady et al. 32G chamber (Harrick Scientific Corp., Ossining, NY) for 2 min. (Caution! Piranha solution is extremely corrosive and oxidizing!) The oxidized substrate was transferred to a 10 mM HEPES buffer solution (4-[2-hydroxyethyl]piperazine-1-[2-ethanesulfonic acid], pH 7.4) containing 0.25 mg/mL PEI-g-PEG and incubated there for 60 min. The polymer-coated substrates were removed from solution, rinsed with HEPES buffer to remove unbound and multilayered polymer, and dried under a nitrogen flow. AFM experiments immediately followed deposition of the polymer on the substrate. 2.3. Preparation of Colloidal AFM Tips. Silica microspheres 5 μm in diameter attached to the end of AFM cantilevers (0.58 N/m nominal spring constant, Novascan Technologies, Inc., Ames, IA) were employed as counterfaces. For normal and friction force measurements involving the bare oxide probe tips, the cantilever assemblies were rinsed sequentially in 0.1 M HCl and ultrapure water, exposed to O2/H2O2 plasma for 15 s, and submerged in HEPES buffer solution immediately prior to the experiments. For measurements entailing polymer-coated probes, the tips were cleaned by the same method as the bare oxide tips and additionally immersed in 0.25 mg/mL solutions of PEI-gPEG for 60 min following the plasma treatment. Upon withdrawal from polymer solution, these tips were rinsed with HEPES buffer solution to remove unbound and multilayered polymer and dried under a nitrogen flow prior to use in AFM measurements.

2.4. Normal and Friction Force Measurements Using AFM. Normal and frictional forces were monitored by AFM at the interface of the polymer-modified SiO2 substrate and the bare or polymer-modified silica microsphere under liquid environments, made possible with the use of a liquid cell tip holder (Digital Instruments, Santa Barbara, CA). The AFM was controlled by AFM100/STM100 electronics and SPM32 software (RHK Technology, Inc., Troy, MI). Movement of the substrate relative to the fixed AFM tip in the x, y, and z directions was accomplished through the use of a single-tube piezoelectric scanner. Normal and lateral tip deflections were detected by the motion of a laser beam reflected off the back of the cantilever onto a four-quadrant photodiode. This AFM assembly has been discussed in greater detail previously.32,33 Normal forces were detected as a function of separation of the two interfaces by monitoring the deflection of the cantilever over a distance of ∼200 nm in the direction normal to the surface. In both friction and force-distance measurements, the normal load was limited to maximum values of 25 nN to avoid the possibility of tip and substrate damage. Kinetic friction was measured as a function of loading and unloading between the AFM probe and the modified substrates through the torsional deflection of the cantilever. Scan rates of approximately 2000 nm/s over a lateral distance of 200 nm were employed throughout the study, keeping the tribological conditions well within the boundary regime. Using the nominal spring constant of the cantilever (k = 0.58 N/m), normal loads were found based on the known substrate displacement. Friction forces were determined by sliding tips of nominal radius 20 nm, attached to similar cantilever assemblies, on known slopes of SrTiO3 planes (WITec GmbH, Germany) at selected normal force set points. This method, which yields an instrumental lateral force sensitivity, based on its geometric relationship with the normal load, has been discussed in greater detail elsewhere34 and enables an approach to quantitatively compare frictional data within an experimental set. Solvents were exchanged in the liquid cell in the order of HEPES, methanol, ethanol, 2-propanol, and HEPES by transferring sufficient volumes using two 5 mL syringes. Normal and friction force measurements were conducted under each solvent (32) Perry, S. S. MRS Bull. 2004, 29, 478–483. (33) Perry, S. S.; Somorjai, G. A.; Mate, C. M.; White, R. L. Tribol. Lett. 1995, 1, 233–246. (34) Ogletree, D. F.; Carpick, R. W.; Salmeron, M. Rev. Sci. Instrum. 1996, 67, 3298–3306.

Langmuir 2009, 25(13), 7443–7449

Article less than 30 min following their injection. The same cantilever/tip assembly was used throughout a set of measurements in which the friction and normal force response was compared under the different solvents. 2.5. Solvent Uptake Measurements Using QCM. The QCM consisted of a SA250B-1 Network Analyzer and test fixture (Saunders & Associates, Inc., Phoenix, AZ), a liquid flow cell (Maxtek, Inc., Beaverton, OR), and QTZ control software (Resonant Probes GmBH, Germany). The resonators used in QCM experiments were AT-cut quartz crystals (Maxtek) with silicasputtered gold electrodes and a 5 MHz fundamental resonance frequency. Prior to measurement, QCM crystals were cleaned according to the following procedure: sonication in acetone, 2-propanol, and ultrapure water for 5 min each with an ultrapure water rinse in between each sonication; drying under a nitrogen flow; and plasma cleaning in an O2/H2O2 environment for 1 min. Immediately after plasma cleaning, the crystal was installed in the liquid flow cell that had just been cleaned with 2-propanol and dried under a nitrogen flow and made to sit overnight to allow time for the stress relaxation of the Viton O-ring seal. Data were recorded in air for the first 60 min to ensure stabilization of the QCM assembly. Solvents were exchanged in 10 min intervals into the liquid flow cell in the following order: methanol, ethanol, 2propanol, and HEPES (Figure 4). This series of solvents represents the “blank” data to which solvent mass uptake measurements on the polymer-coated quartz crystal were compared. The deposition of PEI-g-PEG was performed in situ through the injection of the 0.25 mg/mL PEI-g-PEG HEPES solution. The PEI-g-PEG was allowed to adsorb onto the silica surface over a 10 min period and then was followed by a HEPES rinse to remove unbound and multilayered polymer. The above series of solvent injection was then repeated, this time on the polymer-coated quartz crystal, to determine changes as a result of adsorption and solvent association. Frequency shifts and bandwidth changes were calculated by averaging the data over 5 min intervals collected after the system had reached equilibrium. The change in resonance frequency was related to the adsorbed mass by the Sauerbrey equation.35 Solvent exchange on a blank quartz crystal would cause a negative frequency shift proportional to the square roots of the viscosity and density of the liquid, according to the Kanazawa relation;36 this negative frequency shift (Δf ) represents the dissipative interaction of the resonator with a viscous fluid as can be seen in a positive shift in the half-bandwidth (ΔΓ) of the same magnitude. On a polymer-coated quartz crystal, the frequency shift includes, additionally, mass loading due to the mechanical coupling of the solvent that is intimately associated with the polymer segments. In the thin film limit, this mass uptake can be extracted from the observed frequency shift by the subtraction of the viscous load contribution as determined from the blank run. This “wet mass” (from ΔΔf ) can be used to compare the relative solvent uptake of the polymer under different solvent environments. Details of this approach have been reported elsewhere.3,37

3. Results and Discussion 3.1. Effect of Solvent on Friction Forces of PEI-g-PEG Brushes. Figure 2A displays the plot of interfacial kinetic friction vs increasing normal load for the symmetric contact of a PEI(25)g[3.5]-PEG(4)-modified tip and substrate as a function of solvent. The same probe tip and substrate were employed throughout this series of measurements. It is evident that the interfacial friction of this system is a strong function of the solvent quality of the surrounding environment. While coefficients of friction for the (35) Rodahl, M.; Kasemo, B. Sens. Actuators, A 1996, 54, 448–456. (36) Kanazawa, K. K.; Gordon, J. G. Anal. Chem. 1985, 57(8), 1770–1771. (37) Goubaidoulline, I.; Reuber, J.; Merz, F.; Johannsmann, D. J. Appl. Phys. 2005, 98, 014305/1–014305/4.

DOI: 10.1021/la900371k

7445

Article

Figure 2. (A) Interfacial friction measured as a function of increasing load between a silica colloidal probe and SiO2 surface, both (symmetrically) modified by the adsorption of PEI(25)-g[3.5]PEG(4) at varying solvent environments, measured 30 min after solvent injection. The measurements were all performed using the same AFM cantilever/tip assembly. (B) Interfacial friction measured as a function of increasing load between the probe and the surface, symmetrically modified by PEI(25)-g[3.5]-PEG(4) in one experiment and PLL(20)-g[3.2]-PEG(5) in another. Both measurements were performed in HEPES buffer solution (pH 7.4) with the same AFM cantilever/tip assembly.

polymer interface in HEPES are exceptionally low, significant increases in friction are also observed with the decreasing polarity of the solvent. The similarly low friction force values encountered at normal loads less than ∼5 nN for all solvents is attributed to the boundary layer of solvated PEI-g-PEG, as solvent molecules are not “squeezed out” until loads are sufficiently high. The trend seen in the coefficient of kinetic friction as a function of solvent is due to differences in solvent polarity. With the HEPES buffer solution (pH 7.4) being the aqueous solvent system, the polarity varies for the solvent series as follows: HEPES buffer solution > methanol > ethanol > 2-propanol. In HEPES buffer solution, the PEG side chains form hydrogen bonds with nearby water molecules.38 Moreover, when the solvent is water, the distance between adjacent oxygen atoms in the gauche form would be commensurate with the O-H 3 3 3 O distance in liquid water, allowing PEG to participate in its hydrogen-bonding network.20,39 Control experiments show a very similar contact behavior between a colloidal probe and a (38) Tasaki, K. J. Am. Chem. Soc. 1996, 118, 8459–8469. (39) Kjellander, R.; Florin, E. J. Chem. Soc., Faraday Trans. 1981, 9, 2053–2077.

7446 DOI: 10.1021/la900371k

Brady et al.

substrate symmetrically coated with PEI-g-PEG under HEPES and deionized water, suggesting that the effect on friction seen in the solvent exchange was dominated by solubility effects as opposed to removal of the constituents of the HEPES solution. With these considerations, it can be expected that the interaction between PEG side chains and nearby solvent molecules would decrease with decreasing solvent polarity. In previous work on the solvent-dependent tribology of PLL-g-PEG, it was found that the ability for solvent absorption by the PEG brush increases as the hydrogen-bonding and polar Hansen interaction parameters increase.26 It was also determined that there were an average of 14 water molecules associated with each ethylene glycol unit, as compared to 0.8 for 2-propanol, the worst solvent in the series. In the same manner, the amount of trapped solvent in the brushlike structure of PEI-g-PEG is expected to be greater than for less polar solvents. The lubricity increases (friction decreases) for a good solvent such as HEPES, in which the copolymer takes on a swollen brushlike structure. For a poor solvent such as 2-propanol, the lubricity decreases as a result of a more collapsed brush conformation and decreased solvent uptake in the copolymer. Previous studies have shown similar results of low friction for polymer brush-modified interfaces in good solvents, specifically for the case of polystyrene brushes in toluene, as measured with SFA40 and AFM.3 Previous work in our laboratory and others’ on the tribological behavior of polymer brush-modified interfaces has focused primarily on PLL-g-PEG-coated SiO2 surfaces.25-27 Figure 2B exhibits a comparative plot of friction as a function of increasing normal load for symmetric (both substrate and microsphere tip have adlayers) PEI(25)-g[3.5]-PEG(4)- and PLL(20)-g[3.2]-PEG (5)-coated interfaces in HEPES buffer solution, measured with the same AFM cantilever/tip assembly. Notably, low friction was observed for the entire range of measured loads for both interfaces, as observed previously for PLL-g-PEG interfaces in buffer solution.25,26 Although the molecular architectures of these two copolymer systems are distinguished through their backbones, they have approximately the same grafting ratios. The results presented here demonstrate that the frictional properties in HEPES are determined largely by solvated components extending away from the surface. 3.2. Effect of Solvent on Normal Forces of PEI-g-PEG Brushes. Normal forces monitored as a function of substrate displacement provided information on the conformation of the adsorbed polymer brushes as a function of solvent environment. Symmetrically coated interfaces were employed in measurements of force-distance curves obtained using the same AFM cantilever/tip assembly as a function of solvent. Although the exact separation distance of the probe and silica surfaces is convoluted with the elastic deformation of the polymer coatings, the qualitative behavior clearly highlights changes in conformation as a function of the solvent quality. Figure 3 presents (A) loading and (B) unloading curves for the symmetric PEI-g-PEG-modified interfaces; the hysteresis between them is mainly due to the lag in the piezo response typical of open-loop systems such as used in these experiments. In both data sets, it is evident that a solvent dependence exists for the load-displacement behavior. The loading trace under HEPES shows a strong rise in normal force at 90 nm interpreted as the start of the compressive regime; by contrast, compression starts at around 140 nm under the alcohols. When solvated by HEPES, the onset of compression occurs at a distance approximately 50 nm further from the substrate as compared to contact under less polar (40) Klein, J.; Perahia, D.; Warburg, J. Nature (London) 1991, 352, 143–145.

Langmuir 2009, 25(13), 7443–7449

Brady et al.

Article Table 1. Wet Mass of PEI(25)-g[3.5]-PEG(4) Adsorbate on SiO2Coated QCM Crystals As a Function of Solvent Environmenta solvent

ΔΔmwet (μg/cm2)

HEPES (first rinse) 1.39 methanol 0.68 ethanol 0.39 2-propanol 0.28 HEPES (second rinse) 1.58 a The mass loading was extracted from the shift in frequency by subraction of the viscous loading as determined from the blank run.

Figure 3. Normal load detected as a function of the z displacement of the piezoelectric tube scanner, upon which the sample rests, for the contact between a PEI-g-PEG-coated silica colloidal probe and a SiO2 surface under various solvents. For zero z displacement, the tip-sample separation is maximized. The approach curve (A) represents increasing load, while the retract curve (B) represents decreasing load.

solvents. This difference suggests that the PEI-g-PEG brush is in a more extended conformation under HEPES than under the alcohols. The theoretical length of a fully extended 4 kDa PEG chain is estimated to be 31.5 nm;41 in considering film thickness, additional solvation (and therefore expansion) of the globular PEI backbone must also be included. Furthermore, it also takes almost twice the displacement to compress the PEI-gPEG brush under HEPES as compared to being under the alcohols to attain the same maximal normal force. In a qualitative sense, it is thus evident that there is a softer contact between the surfaces under HEPES than under the alcohols. Contact has been taken to be the rapid departure from an equilibrium cantilever position in the force-distance plot, which is interpreted as the onset of compression. Also depicted in Figure 3A, there is a long-range repulsive interaction42 prior to contact when the brush is solvated with HEPES. While the normal load remains virtually zero until contact under the alcohols, it gradually increases prior to contact under HEPES. Because of the long-range nature of the repulsive force, this behavior can likely be attributed to electrostatic repulsion that results from the overcompensation of the substrate’s negative surface charge due to the adsorbed polycations. (41) Rixman, M. A.; Dean, D.; Ortiz, C. Langmuir 2003, 19(22), 9357–9372. (42) Butt, H. J. Biophys. J. 1991, 60, 1438–1444.

Langmuir 2009, 25(13), 7443–7449

The unloading curves displayed in Figure 3B show similar solvent dependence to the increasing load curves in Figure 3A. The retract curves for the alcohols show very little hysteresis when compared to the approach curves. Comparison of the loading and unloading traces generally shows the compression to be elastic, with negligible adhesion hysteresis under the alcohols and only a minor one under HEPES. The small pull-off force (at ∼50 nm) shown in Figure 3B for HEPES can also be attributed to physical entanglement of polymer chains as the substrate is pulled away from the tip. Previous work on PLL-g-PEG has shown that the pull-off forces for a symmetric interface are negligibly small.21,26 While it has been shown that PLL-g-PEG exhibits a rodlike conformation with extended PEG brushes when adsorbed onto oxide interfaces,21,25-27 the difference in pull-off forces suggests that these two lubricious systems have distinct conformational structures when adsorbed onto such interfaces. The solvent dependence of surface forces monitored at PEI-g-PEG-modified interfaces demonstrates the ability of this copolymer to adopt collapsed and extended conformations under solvents of varying polarity. The more stretched conformation observed under HEPES can be attributed to entrapment of water molecules within the PEG brush structure. 3.3. Relative Solvent Uptake of PEI-g-PEG Brushes. Changes in polymer mass as a function of solvent absorption was measured by QCM. For these measurements, adsorption of PEI-g-PEG from a HEPES solution on to SiO2-coated QCM crystals occurred in situ over a period of 10 min. The adsorbed mass of the solvated polymer was then compared to solvated masses measured for this same polymer film under different solvents. The changes in mass are taken to reflect the degree of solvent uptake as a function of solvent quality. Systematic variation in the dissipative component of the frequency shift in the QCM data additionally reflects changes in the state of the adsorbed polymer as a function of solvent quality. Table 1 reports the solvated mass of the PEI-g-PEG brush as a function of solvent. This solvated mass is a measure of the mass of adsorbed copolymer as well as that of solvent molecules absorbed within the brush. Before polymer deposition (blank), the frequency shift is due to viscous loading of the solvent on the crystal, as reflected from how the magnitude of the frequency shift approximates that of the bandwidth shift but opposite in sign (Figure 4). The subsequent mass loading of the solvent on the adsorbed polymer brush can be obtained from the frequency shift by subtracting the viscous contribution estimated from the blank. The sequential exposure of the adsorbed PEI-g-PEG to different solvents in these QCM measurements followed the solvent switching conditions used in AFM friction measurements. The data provided clearly show a direct correlation between the solvent quality and the mass of solvent absorbed within the brush. The adsorbed mass increases significantly as the polarity of the solvent increases. This relationship follows from the increased interactions between PEG chains and solvent molecules as the polarity of the solvent increases. The rise in mass uptake in DOI: 10.1021/la900371k

7447

Article

Brady et al.

Figure 4. Fractional shifts in frequency (Δf/f ) and half-bandwidth (ΔΓ/f ) under methanol, ethanol, 2-propanol, and aqueous HEPES solution of a quartz crystal resonator before and after the in situ adsorption of PEI-g-PEG.

HEPES from the first to the second rinse can be rationalized in terms of solvent trapping within the brush, caused by incomplete solvent exchange upon injection of the next solvent. Injection of the alcohols likely does not remove all of the water molecules trapped within the brush after the HEPES injection, due to the very high affinity of water to PEG. The increase in energy dissipation in the QCM resonator, related to the increase or decrease in bandwidth, was greatest when PEI-g-PEG was solvated with HEPES and methanol and decreased as the solvent polarity decreased. In HEPES, where the adsorbed polymer mass is greatest due to maximized solvation of the brush, the increased viscoelastic damping of the piezoelectric resonator originates from the adsorbed film becoming more plasticized with increased solvation. These QCM data thereby corroborate the copolymer’s lubricious behavior and brush extension, as observed through AFM measurements in polar solvents, by displaying the direct relationship between absorbed solvent within the brush and solvent quality. 3.4. Symmetric vs Asymmetric PEI-g-PEG-Coated Interfaces. Frictional forces were measured using AFM as a function of increasing normal load for the contact between the PEI-g-PEG-modified SiO2 substrate and both bare and PEIg-PEG-modified counterfaces. The tribology of these asymmetric and symmetric interfaces, respectively, was studied under various solvents as follows: HEPES, methanol, ethanol, 2-propanol, and HEPES. Figure 5 displays friction-load behavior in each solvent for the (A) asymmetric and (B) symmetric PEI-g-PEG-modified interface. Both the coefficient of kinetic friction and the magnitude of friction force for a particular normal load are significantly greater for the asymmetric interface. The friction response under 2-propanol, for example, is more than twice at the maximum load of ∼20 nN as compared to the symmetric case. The symmetric interface exhibits more lubricious behavior due to greater interfacial repulsions than for the asymmetric interface as the substrate approaches the microsphere. These repulsions are steric in nature and characteristic of polymer brush structures. For the symmetric interface, the PEG chains tethered to each surface repel one another due to unfavorable increases in entropy and decreases in free volume as they are compressed. In the asymmetric case, there is only one “layer” of tethered PEG chains because the opposing surface is an uncoated silica microsphere. Increased polymerpolymer repulsions in the symmetric interface result in greater normal loads required to achieve equal compression as compared to the asymmetric interface. Qualitative evidence of distinctions in compression of the interface can be seen in comparing the change 7448 DOI: 10.1021/la900371k

Figure 5. Interfacial friction recorded as a function of increasing normal load. (A) Asymmetric interface in which only the SiO2 substrate is coated with PEI-g-PEG (the silica tip is uncoated). (B) Symmetric interface in which both the silica tip and the SiO2 substrate are coated with PEI-g-PEG. The polymer films were exposed to different solvents during measurement as follows: HEPES, methanol, ethanol, 2-propanol, and HEPES.

in friction force over 0-5 nN loads. The friction begins to increase immediately for the asymmetric interface, while it remains constant for loads up to approximately 5 nN for the symmetric interface. As described earlier, the friction is independent of load in this low-load regime for the symmetric interface due to the inability to force solvent molecules out of the brush. Thus, the brush at the asymmetric interface is more easily compressed at low normal loads than that of the symmetric interface. Symmetrically coated surfaces, representing “softer” brushlike interfaces, experience lower shear forces as they slide relative to one another due to these increased polymer-polymer repulsions. Although the observed trends in friction-load behavior for symmetrically coated interfaces were independent of the order of solvent exchanges, there exists hysteresis in the behavior between the two HEPES solvations for the asymmetric interface. Material transfer from the coated substrate to the negatively charged bare tip would cause friction to be lower after the transfer event due to effective sliding between a polymer-polymer interface, but control experiments suggest this to not be the case. Friction-load behavior was measured under HEPES for an asymmetric PEIg-PEG-modified interface after polymer deposition, after ethanol rinse, and, finally, after cleaning the AFM tip. This showed a reduction in friction response after the ethanol rinse that remained the same even after the probe/cantilever assembly was Langmuir 2009, 25(13), 7443–7449

Brady et al.

Article

removed and cleaned with 0.1 M HCl followed by O2/H2O2 plasma to remove any polymer material that fouled the probe, thus excluding the effect of possible material transfer. Bridging through the interaction of amine groups in the PEI backbone with both the bare silica microsphere and the substrate is responsible for the observed hysteresis. Amine groups not utilized in PEG grafting are positively charged at the buffer pH, while the microsphere is negatively charged. Upon adsorption of PEI-g-PEG on the SiO2 substrate, not all of the available charged amine groups may be attached to the substrate due to the globular adsorption structure of this copolymer.22 As the bare, negatively charged microsphere approaches the substrate during friction measurements, these freely moving portions of the PEI backbone would adhere to the probe through electrostatic attractions, thus causing increased friction due to adhesion between the two surfaces. Upon solvation in the alcohols, the dehydration (collapse) of the brush due to decreased polymer-solvent interactions may effectively cause these free amine groups to finally bind to the substrate. When the brush is again solvated with HEPES, the entrapped water molecules would give rise to an extended PEG brush conformation, with the PEI backbone now more securely tethered on the substrate. The burying of positively charged amine groups of the PEI backbone reasonably accounts for the transition from high to vanishingly low friction in HEPES for asymmetrically coated interfaces.

experiments, fluid drag was negligible, and thus, these measurements were well within the boundary regime of lubrication. The solvent-dependent friction observed in AFM experiments was corroborated by solvent mass uptake studies using QCM. Greater mass uptake in the PEI-g-PEG brush was seen under HEPES buffer solution as compared to the alcohols. Normal force-displacement relationships monitored using AFM revealed an increased film thickness for the PEI-g-PEG brush under HEPES buffer solution as compared to that measured under the alcohols. The increased solvent mass uptake and film thickness under solvents of greater polarity allowed the observed lubricious behavior to be attributed to the expanded or stretched PEI-g-PEG brush conformation. Friction force measurements of symmetric (PEI-g-PEG-coated substrate and probe) and asymmetric (PEI-g-PEG-coated substrate and unmodified probe) interfaces demonstrated differences in the solvent-dependent friction due to bridging effects present between the brush and the unmodified probe in the asymmetric interface. Comparison of the friction of PEI-g-PEG and PLL-g-PEG brushes under HEPES buffer solution revealed vanishingly low friction forces in good solvent for both systems, even though there are significant differences in molecular architecture. The similar tribological behavior of these two brushes allows for the use of PEIg-PEG as an alternative to PLL-g-PEG in aqueous biomimetic lubrication.

4. Conclusion

Acknowledgment. We acknowledge the Air Force Office of Scientific Research (AFOSR) for support of this research. We also thank Benjamin Raterman for his assistance in our laboratory during this study. We also acknowledge the thoughtful input :: of Prof. Nic Spencer and Dr. Seunghwan Lee of ETH-Zurich.

The tribological properties of a surface-bound, brushlike copolymer (PEI-g-PEG) have been studied as a means to decrease the normal and shear forces experienced between a silica colloidal probe and an oxide surface. PEI-g-PEG was adsorbed from solution onto oxide-passivated silicon surfaces at physiological pH to form molecularly thin films. Lateral force measurements detected very low friction forces in HEPES buffer solution, a good solvent, while friction was found to increase as the solvent polarity decreased. Under the very low sliding velocities typical in AFM

Langmuir 2009, 25(13), 7443–7449

Supporting Information Available: AFM height contrast image taken in AC mode of PEI-g-PEG film on a Si wafer under HEPES buffer. This material is available free of charge via the Internet at http://pubs.acs.org.

DOI: 10.1021/la900371k

7449