Report

Lubrication at Physiological Pressures by Polyzwitterionic Brushes

See allHide authors and affiliations

Science  27 Mar 2009:
Vol. 323, Issue 5922, pp. 1698-1701
DOI: 10.1126/science.1169399

Abstract

The very low sliding friction at natural synovial joints, which have friction coefficients of μ < 0.002 at pressures up to 5 megapascals or more, has to date not been attained in any human-made joints or between model surfaces in aqueous environments. We found that surfaces in water bearing polyzwitterionic brushes that were polymerized directly from the surface can have μ values as low as 0.0004 at pressures as high as 7.5 megapascals. This extreme lubrication is attributed primarily to the strong hydration of the phosphorylcholine-like monomers that make up the robustly attached brushes, and may have relevance to a wide range of human-made aqueous lubrication situations.

The rubbing of opposing bones during the articulation of mammalian joints is mediated by layers of articular cartilage coating their surfaces that provide particularly efficient lubrication as they slide past each other (1, 2). The associated very low friction at the high pressures of human joints such as hips or knees, with friction coefficients μ < 0.002, has to date not been emulated in human-made systems. Model studies (37) between smooth sliding surfaces bearing neutral or charged polymer brushes demonstrated sliding friction coefficients as low as μ < 0.001, which are values lower than with any other boundary lubricant system. As noted (8, 9), in earlier studies with polymer brushes (36), the friction increases sharply at mean pressures P > ∼ 0.3 MPa, which is far below the 5 MPa or more at which low friction persists in nature. In the present work, we sought to overcome the limitations of these earlier studies (36). We used polymer brushes, which make good boundary lubricants because they do not bridge the intersurface gap, that are strongly attached to each surface so as to resist being sheared off and that are highly hydrated so as to use the very efficient lubrication afforded by hydration sheaths that has been previously observed (10, 11).

Brushes of the polymer poly[2-(methacryloyloxy) ethyl phosphorylcholine] (pMPC) were grown from the surfaces of macroinitiator-coated mica sheets that were premounted onto cylindrical lenses for surface force balance (SFB) measurements (12, 13). After calibration in dry air, control measurements of the normal forces (Fn) and shear forces (Fs) between the macroinitiator-coated mica surfaces were performed (12, 14) as a function of their absolute separation D and their sliding velocity vs [Fn(D) and Fs(D, vs), respectively]. Measurements were performed both in polymer-free water and after incubation in a 0.3% w/v pMPC homopolymer solution in order to compare them with the subsequent pMPC brush profiles. These controls, shown in Fig. 1, reveal the high friction associated with the macroinitiator alone (Fig. 1, B and C) before the growing of the polymer layers. They also show that the pMPC homopolymer does not adsorb onto the macroinitiator-bearing mica (Fig. 1A), which confirms that surface growth of the polymer results in true brush formation. After the polymer brushes were grown via direct polymerization from the surface, Fn(D) and Fs(D, vs) profiles between the brush layers were measured in water and in aqueous NaNO3 solutions up to physiological salt concentrations.

Fig. 1.

Control measurements of (A) Fn versus D profiles and (B and C) of Fs between bare mica and between macroinitiator-coated mica surfaces across water and across a pMPC homopolymer solution. (A) Solid triangles indicate forces between bare mica in pure water on approach [normalized as Fn(D)/R, where R ≈ 1 cm is the mean surface curvature radius]. The dashed line is a fit to the Derjaguin-Landau-Vervey-Over-beek theory based on a numerical solution of the nonlinear Poisson-Boltzmann equation (32) for a 1:1 electrolyte concentration c = 5 × 10–5 M, with constant surface potential ψ0 = –80 mV and Hamaker constant A = 2 × 10–20 J. Solid and open circles indicate Fn(D) profiles between macroinitiator-coated (20 min of incubation) mica in pure water on approach and separation, respectively. Solid and open diamonds indicate profiles between macroinitiator-coated mica after 13 hours of incubation in 0.3% w/v pMPC homopolymer [Mw = 22,000; Mw/Mn = 1.3, where Mw and Mn are, respectively, the weight and number average molecular weights] in aqueous solution on approach and separation, respectively. The solid line is a guide to the eye. The illustrations show schematically the configurations corresponding to the three control measurements; the short chains (black) represent the macroinitiator, and the longer random-coil chains represent the pMPC homopolymer in solution. (B) A summary of friction force Fs versus load, extracted from traces as in (C), between macroinitiator-coated surfaces in water (open triangles) and in 0.3% w/v MPC homopolymer solution (solid triangles); the solid line corresponds to friction coefficient μ = 0.33. (C) Typical Fs versus time traces between two macroinitiator-coated mica surfaces in water at different D values in response to laterally applied motion of the top mica surface (top trace).

Figure 2 shows the typical Fn versus D profiles between pMPC brush–bearing mica surfaces in the standard crossed-cylinder SFB configuration in pure water and at salt concentrations of ∼0.01 M and ∼0.1 M NaNO3 (the pMPC brush and its monomer structure are shown in Fig. 2A, inset). Some contraction of the pMPC brushes is seen in the salt solutions relative to those in the pure water. The highest normal loads Fn applied are some two or more orders of magnitude higher than those in earlier brush studies (4, 5) that used the SFB. This leads to substantial flattening at the contact region, as indicated in the photo of the interference fringes (Fig. 2B and schematically in Fig. 2C), from which the contact area A between the surfaces is measured directly (the mean pressures P across the flattened contact area are given by P = Fn/A). Comparison with the control profiles from Fig. 1A in the absence of polymer brushes (the dotted and dashed curves in Fig. 2A) reveals the extension of the unperturbed brushes from the macroinitiator layer, whereas fits to the force profiles provide more detailed information on the brush characteristics (12, 13). The profiles both on approach and separation of the surfaces (characteristic of brush interactions) are similar as is in particular their insensitivity to repeated shear up to the highest compressions, which shows the robustness of the layers toward resisting friction and shear-off (15).

Fig. 2.

Normal interactions between pMPC brush–coated mica surfaces across water and across aqueous salt solutions (normalized as above). (A) Normal force profiles: circles, water (no added salt); triangles, 8.7 mM NaNO3; diamonds, 88.6 mM NaNO3. Solid and open symbols indicate compression and decompression, respectively. Different incubation times in the macroinitiator solution (20 min and 2 hours) were used, with a polymerization time of 1.5 hours in each case (the layers are designated as 20-1.5 and 2-1.5, respectively). The pMPC layers are 2-1.5 brushes (12). Before solvation, the brushes were characterized (12, 13) with x-ray photoelectron spectroscopy, multiple-beam interference, x-ray reflectometry, and atomic force micoscopy, which revealed a uniform smooth layer (root mean square roughness, 0.4 nm), whereas from a fit to the data (solid curves) (12) we estimate mean inter-anchor spacings s = 3.5 ± 1 nm and unperturbed brush-heights L = 37 ± 5, 28 ± 5, and 25 ± 5 nm for the solvated brushes at respective increasing salt concentrations. The brush-chain polydispersity is estimated at ∼1.2 (12, 13). The pMPC brush-like configuration and its monomer structure are also shown. The dashed and dotted lines summarize the brush-free controls between bare and between macroinitiator-coated mica, respectively, from Fig. 1A. (B) Equal chromatic order interference fringes revealing the flattening of the mica surfaces at a high load, indicated by the arrow in (A). (C) The schematic configuration corresponding to (B) (with contact area A ≅ πr2).

Characteristic traces of Fs versus time as two pMPC brushes slide past each other under different loads or compression ratios (corresponding to different surface separations) and salt concentrations, taken directly from the SFB, are shown in Fig. 3. Traces both at low compressions (higher D values) (Fig. 3, left) and high compressions (lower D values) (Fig. 3, right) show that the shear forces remain very weak—within the noise level of the SFB—up to mean contact pressures of ∼2 to 2.5 MPa and then increase measurably with the load, up to the largest loads applied. Such traces are characteristic of all pMPC brushes and salt concentrations studied; that the traces are un-changed in time indicates there is no detectable wear of the layers over the range of parameters studied. Shear force results are summarized in Fig. 4 and Table 1. These show that the friction coefficient μ (taken as the slope of the Fs versus Fn plot) is in the range of values μ ≈ (1.5 ± 1) × 10–3 up to the highest mean contact pressures applied, P ≈ 7.5 MPa (∼75 atmospheres). The values of μ, though remaining very low, are seen to increase slightly within this range as the salt concentration increases. The variation of the shear or friction force with vs is shown in Fig. 4B, revealing a weak Fs(vs) dependence over three orders of magnitude in vs.

Fig. 3.

Typical Fs versus time traces between pMPC-coated mica surfaces taken directly from the SFB. (Left) 20-1.5 brushes in pure water. (Right) 2-1.5 brushes in ∼0.1 M salt solution (12). The top trace A in each column is the lateral motion applied to the top surface, whereas traces B to E are the corresponding shear forces transmitted between the surfaces at the compression ratios β (2L/D) shown. Trace F is a frequency (ω) analysis of trace E, where the magnitude of Fs (ω) at the drive frequency (arrow) is used to reveal the onset of frictional force above any systematic signal (5, 10) at large D values.

Fig. 4.

Variation of friction Fs with load Fn and with sliding velocity vs between pMPC brush–coated surfaces. (A) Variation of Fs with Fn for both 20-1.5 (circles) and 2-1.5 (triangles) brushes across water; within the scatter there is little systematic difference between these two brushes. Solid circles, approach; open circles, separation. There is a double logarithmic scale; the solid curve corresponds to μ = 0.00043. (Inset) The same data on a linear scale; the solid line for Fn > 1.5 mN corresponds to the same friction coefficient as in the main plot. The sharply rising dotted line is the friction between macroinitiator-coated surfaces with no pMPC brushes (based on Fig. 1, inset). The arrows indicate the maximal loads applied in the polyelectrolyte brush friction study (5). (B) Typical variation of Fs with vs at different compressions and salt concentrations. In ∼10 mM NaNO3, triangles indicate D = 13 ± 0.3 nm on approach (solid) and separation (open) of surfaces, respectively. In ∼100 mM NaNO3, solid circles indicate D = 13.3 ± 0.3 nm in approach run; diamonds indicate D = 10.2 ± 0.3 nm, on approach (solid) and separation (open) of surfaces, respectively; and crossed circles indicate D = 8.5 ± 0.3 nm, in separation run. There is little variation between data sets whether on approach or separation of surfaces and whether in ∼10 mM or 100 mM salt.

Table 1.

Friction coefficients between pMPC brush–coated mica surfaces sliding past each other at mean pressures P (in the range of 2 to 7.5 MPa) at different salt concentrations. Details about the water purification can be found in (12).

View this table:

Our findings show that brushes of pMPC, a polyzwitterion whose monomers carry the phosphorylcholine (PC) group (as occurs in naturally occurring phospholipid headgroups), can reduce friction to levels previously seen only in natural joints (μ on the order of 0.001 at pressures of 5 MPa or more). Earlier studies (4, 5) between polymer brush–coated surfaces showed similarly low friction coefficients but only at very much lower pressures. At higher pressures, the coefficient of friction became large, either because of entanglement or vitrification of the brushes (4) or as a result of the increasing frictional shear at these pressures removing from the surface the relatively weakly attached charged brushes (5). What is the mechanism underlying the very low friction in the present study?

Earlier studies with charged brushes attributed comparably low friction coefficients (at much lower pressures) to a combination of factors. At low to moderate pressures, interpenetration (and therefore entanglement) of the opposing brushes was suppressed by configurational entropy effects (4, 6, 16). At the higher compressions of the present study, such interpenetration during the sliding is attributed instead to a self-regulation mechanism within the interfacial region (6, 17). According to this, the mutual interpenetration zone of the sliding brushes, where the viscous dissipation leading to frictional drag occurs, has a thickness δ. This thickness varies so that the relaxation rate of the polymer moieties within it equals the shear rate within the zone; that is, [1/τ(δ)] ≈ (vs/δ), where τ is the relaxation time of the polymer moieties. Such self-regulation can readily be seen to lead to a weak dependence of friction on vs, which is in line with the observed data at high loads (Fig. 4B), and a similar variation has been seen earlier for friction between sliding highly compressed brushes (17). Qualitatively, this occurs because at higher sliding velocities the extent of brush interpenetration decreases so that the overall frictional drag varies more weakly with increasing vs than if the interpenetration zone remained at a constant thickness (6, 17). At elevated pressures, lubrication by charged brushes was previously attributed to counterion osmotic pressure (5), together with the lubricating action of hydration layers surrounding the charged monomers. In the present study, the overall-neutral zwitterionic chains have no counterions associated with them so that osmotic pressure due to trapped counterions cannot be playing a role. The effect of any counterions associated with the residual charge of the underlying macroinitiator-coated mica is negligible. The primary mechanism underlying the low friction observed must therefore be attributed almost exclusively to the high level of hydration of the zwitterionic PC groups on the monomers. These are known to bind around 15 to 25 or more water molecules per monomer [depending on the method of measurement (1823)]. Similar strongly bound hydration layers about such phosphorylcholine-coated biomaterials are also responsible for repelling proteins from their surfaces (22, 24).

Water molecules in hydration layers have been shown to act as very efficient lubricants in the case of hydrated ions trapped between charged surfaces (10). This results from the water molecules being tenaciously attached, on an average over time, to the enclosed ionic charges so that the resistance to dehydration leads to their ability to support a large pressure. At the same time, the hydration water molecules are able to rapidly exchange with other hydration or free water molecules (10, 25) and so are able to behave fluidly when sheared at rates lower than these exchange (relaxation) rates. This, we believe, is also the origin of the low friction in the present system and, together with the stronger attachment of the chains grown directly from the surface, ensures the robust nature of the brushes when compared with the polyelectrolyte chains that were sheared off (15). We attribute the ability of the hydration shells about the (methacryloyloxy)ethyl phosphorylcholine (MPC) monomers to lubricate at much higher mean pressures (relative to those at which previously studied polyelectrolyte brushes were removed by friction) to the higher and more tenacious hydration (1821, 23) of their PC groups. The slight increase in the friction coefficient with increasing salt concentration (Table 1) may be attributed to some salting-out of the pMPC monomers at the higher salt concentration [as has been observed in other PC systems (26)], leading to reduced hydration of the monomers and thus to less efficient lubrication. Such an effect would also explain the small contraction of the brush height at the higher salt concentrations (Fig. 2).

A large reduction in wear of polymer/metal hip implants has recently been achieved (27) by growing a pMPC layer from a polyethylene acetabular surface. The corresponding μ ∼0.1 was not especially low. It is likely that this value of μ is dominated by a small number of high-friction asperity contacts between the sliding surfaces, which are far rougher than the mica substrate in our experiments, or by the bridging of the pMPC chains between the acetabular surface and the metal countersurface to which they may adsorb. This would result in higher friction relative to the values we measured between two pMPC brushes. A comparably high friction coefficient, ∼0.1, was also measured (28) in a recent pin-on-disk tribological study between pMPC-coated surfaces under water.

Finally, in view of their similarly low friction at comparably high pressures, it is of interest to consider the relevance of our study to mechanisms of lubrication between articular cartilage surfaces. Biological lubrication at such synovial joints, despite many decades of study, is not yet well understood at the molecular level (2931). Recent models have focused on the role of pressurized interstitial water (31) and of macromolecules at the outer cartilage surface (30). Our system of synthetic polyzwitterionic brushes polymerized from molecularly smooth mica surfaces does not have a clear analog at the cartilage surface, at which macromolecular components of the cartilage and the synovial fluid surrounding it are likely to be present [including proteins, hyaluronic acid, proteoglycans, glycoproteins, and lipidic molecules (30)]. The detailed role of these in the lubrication process, as well as of the cartilage substrate itself, has yet to be clarified. Moreover, cartilage is softer and much rougher than mica, though at these high pressures it may deform affinely (30); experiments on similarly soft and roughened brush-coated surfaces could potentially provide insight into the mechanisms that are operative in cartilage-cartilage friction. The highly hydrated PC-like monomers on our pMPC chains provide very efficient lubrication. The structure of these is similar to that of lipid headgroups, which provides some context for the recent controversial suggestion that lipid multilayers may have a role in mediating synovial lubrication (29), particularly in view of recent findings on the mechanism of boundary lubrication under water (11). Finally, although our polyzwitterionic brushes have no direct analog at cartilage surfaces, our results underline the possible importance at such surfaces of highly hydrated macromolecules in both chondroprotective and lubrication roles.

We have shown that brushes of a biocompatible polyzwitterion, pMPC, are capable of providing extremely efficient lubrication in aqueous media with coefficients of friction μ ≈ 0.001 at mean pressures up to 7.5 MPa, which are comparable to values in human synovial joints. We attribute the low friction at the high pressure primarily to the lubricating action of tenaciously attached but labile water molecules about the strongly hydrated MPC monomers. Our results may have relevance for boundary lubrication in human-made systems in aqueous or physiological media, as in biomedical devices, in which friction and wear are often an issue.

Supporting Online Material

www.sciencemag.org/cgi/content/full/323/5922/1698/DC1

Materials and Methods

References

References and Notes

View Abstract

Navigate This Article