Modulating Interfacial Energy Dissipation via Potential-Controlled Ion Trapping.

As a metal (gold) surface at a given, but variable potential slides past a dielectric (mica) surface at a fixed charge, across aqueous salt solutions, two distinct dissipation regimes may be identified. In regime I, when the gold potential is such that counterions are expelled from between the surfaces, which then come to adhesive contact, the frictional dissipation is high, with coefficient of friction μ ≈ 0.8-0.9. In regime II, when hydrated counterions are trapped between the compressed surfaces, hydration lubrication is active and friction is much lower, μ = 0.05 ± 0.03. Moreover, the dissipation regime as the surfaces contact is largely retained even when the metal potential changes to the other regime, attributed to the slow kinetics of counterion expulsion from or penetration into the subnanometer intersurface gap. Our results indicate how frictional dissipation between such a conducting/nonconducting couple may be modulated by the potential applied to the metal.

Introduction

Whenever two contacting surfaces are in relative motion, energy will be dissipated at their interface and manifested as a sliding friction force. Such friction, while desirable in some cases such as for improving traction or coupling between the surfaces, may also lead to large energy loss and wear,[1,2] which is generally undesirable, so that the ability to control friction in situ is clearly at a premium. Indeed, achieving facile external control of the friction coefficient between sliding surfaces[3] is one of the holy grails in the field of friction and lubrication. The shear forces between two contacting surfaces are largely dictated by surface interactions such as electrostatic, van der Waals, and hydrophobic forces.[4] This sets a preferable route for controlling friction forces through manipulating surface interactions. So far, alteration of surface interactions has mainly been attained using “smart surfaces” that respond to various external stimuli such as pH, solvent, temperature, electric potential, and light.[3,5]

Modulation of the interfacial friction via control of the potential of conducting surfaces is an especially attractive method, as it may provide an easily applied, rapid, and reversible switching of their physicochemical properties.[6−18] In aqueous systems, the potential of interacting surfaces controls the interfacial distribution of ions, and through that the nature of the interfacial forces and dissipation.[19] In particular, the trapping of hydrated counterions between like-potential surfaces can readily overcome the vdW attractions between them[20−22] and can massively reduce the interfacial energy dissipation as the surfaces slide past each other, via the hydration lubrication mechanism.[21,23] This remarkable ability of hydrated ions to act as a lubricating element offers a direct and readily attained approach for controlling and manipulating frictional dissipation between bare surfaces without the need for a prior chemical modification or coating, which is often prone to wear and tear. Here, we exploit controlled changes in surface potential to either expel or trap hydrated ions between compressed surfaces and thereby strongly modulate frictional dissipation between them as they slide past each other.

In our proof-of-concept model system, a smooth (rms roughness = 0.3 nm) metal surface (gold) at a variable, externally controlled potential interacts with an atomically smooth single-crystal dielectric, a mica surface at a constant (negative) surface charge. By tuning the surface potential of the gold, its interaction across aqueous electrolyte solution with the negatively charged mica is manipulated between two regimes:[24,25] An electrostatic attraction (regime I), where the approaching surfaces come into strong adhesive contact and exhibit high frictional dissipation on sliding; and a hydration repulsion regime (II), where hydrated counterions are trapped between the surfaces, preventing their adhesion. In this latter regime (II), hydration lubrication is active and the interfacial dissipation on sliding, monitored via the friction coefficient μ (= [force to slide]/[load compressing the surfaces]), is massively reduced compared with regime I.

Methods

Materials

Pure water with a total organic content of less than 1 ppb (TOC < 1 ppb), a resistivity of 18.2 MΩ cm (so-called conductivity water), and pH 5.8 was prepared by passing tap water twice through a reverse osmosis system, then passing through an ion-exchange column, and then through mechanical filters of mesh size 5 and 2 μm before processing in a Barnstead Nanopure Diamond UV/UF system. Gold pellets, 99.999% pure, were purchased from Kurt J. Lesker Co. and evaporated from a graphite crucible. Lithium perchlorate, LiClO4, 99.99% pure, was purchased from Merck Millipore and used as received. Nitric acid 65% (HNO3, pKa = −1.4) was purchased from Merck Millipore and used as received.

Atomic Force Microscopy (AFM)

AFM (MFP-3D, Asylum Research) is used for the morphological analysis of surfaces. Scans are performed immediately after exposure of the surface to water or saline to avoid contamination and time effect. All scans are conducted in tapping mode using silicon nitride tips (Asylum Research) with a spring constant of ∼80 N/m.

Surface Force Balance (SFB)

Normal and shear force measurements were performed using a surface force balance (SFB) modified with a custom-designed three-electrode electrochemical cell, as schematically shown in Figure S1 and as previously elaborated elsewhere.[25,26] Briefly, molecularly smooth gold (upper lens) and back-silvered mica (lower lens) surfaces are mounted on fused silica lenses in a cross-cylinder configuration, equivalent to the geometry of sphere over a flat surface, with the upper lens mounted on a sectored piezo tube (PZT) and lower lens on a leaf spring of spring constant Kn = 81.5 ± 2.7 N/m (Figure S1A). The absolute separation D between the surfaces is accurately determined with a ca. 3 Å resolution through analysis of the “fringes of equal chromatic order” (FECO; see inset in Figure S1A), formed between the two reflective surfaces, using the “multilayer matrix method”.[27] External potentials are applied to the gold surface across the three-electrode configuration shown in Figure S1B with the gold surface as the working electrode (W) and two platinum wires (Kurt, J Lesker, 99.95%) as counter and (quasi)-reference electrodes. To avoid any electric leakage, quartz boat and Teflon lower lens holders are used in all measurements. Before each measurement, platinum wires are washed in hot nitric acid (30%, ∼80 °C) for at least 30 min and then passed through a flame three times to oxidize and remove the adsorbed organic residues. The electrodes are connected to a potentiostat (CHI600C, CH Instruments, Inc.), which serves as a control unit.

Normal Force Measurements

In the experiments, quasi-static and dynamic force measurements are used to determine the normalized normal force Fn/R (R is the radius of curvature) as a function of separation D.[24,25] In the quasi-static approach, the upper lens is moved toward the lower lens in a stepwise manner and at a fixed step distance (ΔDPZT) using a piezo tube (PZT). During the approach, the actual change in distance (ΔD), measured using FECO fringes, is determined and the normal force between the surface could be then calculated by ΔFn = Kn(ΔDPZT – ΔD). In the dynamic approach, the lower lens is moved continuously towards the upper lens at a constant velocity using a step motor while recording the FECO fringes at a frame rate of either 30 or 60 frames/s (XR60 camera, Sony). Through video analysis of the interference fringes motion, the surface separation D(t) can be determined at any time and the normal force can be then calculated by solving the equation of motion (including hydrodynamic forces) relevant for our configurational setup.[28,29]

Shear Force Measurements

Shear force measurements are performed by laterally moving the upper lens (gold) using either a PZT or a differential micrometer (Figure S4), depending on the required magnitude of applied shear force. The upper lens, mounted on the piezo tube (lateral shear motion range, ca. 1.5 μm), is connected to vertical leaf springs (spring constant Ks = 400 N/m), and its lateral motion is monitored in real time using an air gap capacitance probe. Shear forces Fs between the surfaces can then be measured through monitoring of the bending of the vertical springs Δd[25] (Figure S4). In cases where a large applied shear motion (>1.5 μm) is required (e.g., diamond and star data points in Figure B), the upper lens mount is gently moved using a differential micrometer. In this way, a lateral force can be applied to the upper lens by bending the vertical leaf springs as schematically shown and described in detail in Figure S4.

Figure 2

Measurement of frictional forces between gold and mica surfaces across 1 and 2 mM LiClO4 solutions, at different loads Fn in regime I and regime II. (A) Typical shear force Fs vs time traces, taken directly from the SFB in regime II, at different loads (corresponding mean contact pressures shown) in 2 mM LiClO4. For regime I, the much higher friction was measured differently, as detailed in the Supporting Information (SI). (B) Summary of friction vs load in the two regimes (regime I: 1 mM LiClO4 (stars); regime II: 1 mM LiClO4 (squares) and 2 mM LiClO4 (triangles and circles)). Each symbol represents a different experiment or a different contact position (different colors). The shaded area illustrates the friction coefficient range measured in regime II, and its average is indicated by the red dashed line. The black and red solid lines correspond to friction coefficients in 1 mM LiClO4 and in pure water under regime I. Also shown (diamonds) is the friction–load variation in water with no added salt, where the surfaces always come to adhesive contact. The inset shows the low load regime II data on an expanded scale.

All of the glassware is cleaned with piranha solution (30:70 H2O2/H2SO4), then rinsed with water and ethanol.

Results and Discussion

Normal and Adhesion Forces

Interactions between the curved gold and mica surfaces, at closest separation D apart across aqueous salt solution, under externally set gold potentials, were measured using a surface force balance (SFB) custom-designed as a three-electrode electrochemical cell (Figure S1), as described in detail elsewhere[25] and are shown in Figure . Figure A shows typical normalized normal force Fn(D)/R profiles between gold and mica in 2 mM LiClO4, where different potentials Ψapp are applied to the gold relative to a platinum quasi-reference electrode. The actual gold surface potential Ψgold is not identical to Ψapp, but is extracted, as described earlier,[24] by fitting the Fn(D)/R curves to the solution of the nonlinear Poisson–Boltzmann (PB) equation with constant potential (gold) and constant charge (mica) boundary conditions, augmented by vdW attraction.[24] The values of Ψgold corresponding to different Ψapp are given in Figure A.

Figure 1

(A) Interaction profiles Fn(D)/R between curved gold and mica surfaces, mean radius of curvature R, under different applied potentials Ψapp (color-coded to match curves) and across 2 mM LiClO4. Empty gray symbols represent an out (retraction) profile taken immediately after approach (filled gray symbols) while maintaining an applied potential of −0.3 V. Gold surface potentials Ψgold were extracted from fits (black curves) to numerical solutions of the nonlinear Poisson–Boltzmann (PB) equation d2Ψ/dx2 = (8πeno/ε0ε) sinh (eΨ/kBT) with constant potential (gold, Ψgold) and constant charge (mica, σmica = 5.23 mC/m2 for all fitting curves) boundary conditions, augmented by vdW attraction Fn(D)/R = −AH/6D2, where AH = 9 × 10–20 J is the gold–water–mica Hamaker constant estimated from fits, n0 is the bulk ion concentration (number of ions per unit volume), ε0 is the permittivity of free space, ε is the dielectric constant of the solvent, kB is the Boltzmann constant, T is the absolute temperature, and e is the electronic charge. Lower and upper insets are schematic representations of the two different interaction regimes as described in the text. (B) Adhesion energy between gold and mica across 1 mM (diamonds) and 2 mM (circles) LiClO4 at different gold surface potentials Ψgold. Each symbol (solid) color represents a different experiment with a new set of surfaces, while empty symbols represent different contact positions in the same experiment. For each experiment, surface potential values were obtained through fits of the gold–mica interaction curves to the nonlinear PB equation, as similarly depicted in (A).

The measured force profiles show a gradual change of the interaction from attraction to repulsion following a change of the gold surface potential Ψgold from positive to negative values. At positive Ψgold values (red and green data points in Figure ), the interaction is purely attractive and the surfaces jump in to adhesive contact, with a well-defined adhesion energy WA, expelling all electric double layer (EDL) counterions to the bulk. When Ψgold is slightly more negative than the surface potential of mica Ψmica, the interaction onsets as an electrostatic repulsion (due to counterion osmotic pressure), but gradually turns to electrostatic attraction due to surface charge inversion at the gold, which is then followed by a jump-in to adhesive contact (blue curve), as discussed in detail earlier.[24] However, when Ψgold is much more negative than Ψmica (gray curve), the electrostatic interaction is purely repulsive at all measured compressions. In this case, the interaction energy curve Fn(D)/R is also fully reversible on separating the surfaces (empty gray circles). Similar results were also obtained in 1 mM LiClO4 (Figure S2). Such reversibility is a clear signature of hydration repulsion,[20,21] where short-range hydration forces between confined hydrated ions overcome the vdW attraction at all separations, hindering the surfaces from reaching a primary minimum in the Fn(D) curve (i.e., an adhesive contact). Overall, this indicates two interaction regimes at which either adhesion (regime I) or hydration repulsion (regime II) occurs as the surfaces come into contact. The insets in Figure A schematically illustrate the interfacial condition following approach to contact in the two regimes.

Within regime I, the adhesion, as quantified by an adhesion energy per unit area WA, may also be controlled by varying the gold surface potential.[30,31]WA was evaluated by measuring the force Fpulloff required to separate the two curved surfaces (mica and gold) from adhesive contact at different values of Ψgold, according to the Johnson–Kendall–Roberts expression[32]WA = −2Fpulloff/(3πR), and is shown in Figure B for two different salt concentrations. While there is some scatter, possibly arising from the different extents of gold roughness at different contact points (SI), at both salt concentrations, WA decreases as Ψgold becomes more negative up to a value where no adhesion is measured due to hydration repulsion, i.e., once regime II (|Ψgold|≫|Ψmica|) is reached.

Potential-Regulated Frictional Dissipation

Interfacial energy dissipation between the contacting mica and gold surfaces is manifested by the friction force between them as they slide past each other, and may be modulated by changing the gold potential, as shown in Figure . The friction coefficient μ = μ(I) ≈ 0.8 when the surfaces approach to vdW contact in adhesive regime I. When they are compressed in regime II, however, in which hydrated counterions are trapped between the surfaces, the frictional dissipation decreases dramatically, yielding μ = μ(II) ≈ 0.05 ± 0.03. Figure A shows typical friction traces in 2 mM LiClO4 solution, recorded directly from the SFB during sliding of the top (gold) surface past the lower (mica) one, from which the friction forces Fs are extracted. Figure B summarizes friction as a function of the load between the surfaces. The much lower frictional dissipation on sliding in regime II is attributed to hydration lubrication by the hydrated Li+ counterions trapped between the contacting surfaces, and as seen in Figure B, it is therefore the same for both 1 and 2 mM salt concentrations (unlike the adhesion energy in regime I; see also SI). We note however that the value of μ(II) (between gold and mica), while low, is nonetheless about an order of magnitude higher than that between two atomically smooth mica surfaces across trapped, hydrated Li+ ions under similar loads, for which μ ≈ (4 ± 2) × 10–3, as reported in ref (33). This suggests that, despite the very low RMS roughness (∼0.3 nm) of the template-stripped gold[26] (Figure S3), asperity contacts form between the gold and mica (such contacts are absent from mica–mica contact, where both surfaces are atomically smooth single-crystal contact). Such contacts, across which hydrated ions may not be trapped, would lead to availability of additional dissipative pathways on sliding,[4,34] in particular hysteretic breaking and reforming of vdW bonds, and also possibly plastic deformation of the metal asperities and some wear.[35]

Notably, since each gold–mica contact region has slightly different gold roughness, the variance in the asperity contact formation may be a source of the scatter observed in our Fs vs Fn data (inset to Figure B). This is because only asperities whose height is comparable to or higher than the diameter (ca. 0.72 nm[36]) of the trapped hydrated Li+ ions would be expected to make vdW contact with the opposing mica surface, while the “valleys” between such asperities—accounting for most of the nominal contact region—trap the hydrated Li+ ions (in regime II). The areal number density of such high asperities, as seen in the gold height profiles (Figure S3), is relatively small so that their variance across the different small contact regions (with diameters of order 10 μm) is likely to be correspondingly large.

Also shown are the friction vs load data between sliding mica and gold surfaces in adhesive contact, when immersed in water with no added salt, revealing a large friction coefficient μ ≈ 0.9, similar to regime I in the 1 mM LiClO4 salt solution (see Methods and Figure S4 for a detailed description of Fs measurement). In such a largely salt-free medium, there are no trapped hydrated ions between the contacting surfaces even at |Ψgold|≫ |Ψmica|, which in a salt solution would correspond to trapped counterions in regime II.[25] This is because the counterions are predominantly hydrated protons, which readily condense into and neutralize the mica surface, leading to adhesive vdW (as well as electrostatic) interaction between the surfaces.[30] The similarity of the frictional dissipation in the two cases (regime I in LiClO4 solution, and water with no added salt) is attributed in both to strong adhesion, where trapped counterions are absent (or largely so), and the associated dissipation on sliding (Figure S4).

Shear Force Tuning

Further insight into the nature of the interfacial dissipation is provided by examining the change in friction in situ, i.e., during sliding of the compressed surfaces past each other, as a function of the applied potential on the metal surface, as shown in Figure . Figure A shows the effect on the friction of mildly changing the potential applied to the gold surface. Initially, the surfaces approach to contact when the potential applied to the gold is Ψapp = −0.3 V—well in regime II where hydration repulsion and hydration lubrication apply, see Figure A—and are compressed by a force Fext. The friction force Fs required to slide the upper surface (gold) against the lower one (mica) yields an effective friction coefficient (Fs/Fext) ≈ 0.05, consistent with μ(II) (Figure B), where Fn is Fext. The potential Ψapp applied to the gold is then changed to either −0.2 V (Figure A(ii)) or −0.15 V (Figure A(iii)). As seen in the caption of Figure A, this results in gold surface potential Ψgold = −0.058 and +0.004 V, respectively, where the surfaces are in regime I. Under the applied load Fext, in equilibrium, counterions should be expelled from between the surfaces and the friction coefficient should be high. However, despite the increase in friction as the gold potential changes (Figure A(ii) and (iii)), the frictional dissipation remains characteristic of regime II.

Figure 3

Variation of in situ frictional force Fs between mica and gold surfaces across 2 mM LiClO4 (traces (ii) and (iii) in (A)) and 1 mM LiClO4 (traces (i) and (ii) in (B)) as the upper (gold) surface is moved back and forth laterally, and the applied potential is toggled periodically (as marked by arrows). Shear force values for each trace are indicated. (A) (i): applied lateral motion ΔX0 to the gold surface; (ii) and (iii): surfaces compressed by Fn in regime II (Ψapp = −0.3 V). Trace (ii): Fn = 314 μN, Ψapp toggled from −0.3 to −0.2 V (corresponding Ψgold = −0.18 to −0.058 V). Trace (iii): Fn = 255 μN, Ψapp toggles from −0.3 to −0.15 V (corresponding Ψgold = −0.18 to +0.004 V). (B) Trace (i): surfaces compressed by Fn = 250 μN in regime II (Ψapp = −0.3 V) then toggled from Ψapp = −0.3 to +0.2 V (corresponding Ψgold = −0.17 to +0.098 V). Trace (ii): surfaces compressed by Fn = 462 μN in regime I (Ψapp = +0.2 V) then toggled from Ψapp = +0.2 to −0.3 V (corresponding Ψgold = 0.092 to −0.165 V).

To see this, we estimate the additional force between the surfaces arising from the change in gold potential. This may be approximated as the pull-off force Fpulloff between the surfaces at the corresponding potentials, available from the WA vs Ψgold plot of Figure B. The total compression of the surfaces may then be estimated as Fn ≈ Fext + Fpulloff and the effective friction coefficient as μ = (Fs/Fn). This yields for traces Figure A(ii) and (iii) μ ≈ 0.09 and 0.1 respectively (SI). These values of the friction coefficient are slightly higher than the regime II values, Figure B, for which μ(II) = 0.05 ± 0.03, but very much lower than the regime I value μ(I) = 0.8. The implication of this, to be further considered below, is that hydrated Li+ ions are trapped between the gold and mica surfaces even at gold potentials corresponding to regime I, thereby greatly reducing the frictional dissipation, through the hydration lubrication mechanism.

In Figure B, we examine the effect on the friction of approaching the surfaces to contact with the gold potential corresponding to either regime II or regime I and then toggling the gold potential deeply into the other regime. Figure B(i) shows gold vs mica initially compressed by Fext at Ψapp = −0.3 V—well in regime II where hydration lubrication applies—as the surfaces slide past each other, yielding μ = (Fs/Fext) ≈ 0.04, in line with earlier values (Figure ). The potential is then toggled to Ψapp = 0.2 V, strongly in regime I (Figure A), and back again, repeatedly. The larger friction force in this case (Ψapp = 0.2 V) exceeds the maximal applied shear force Fs,max so that the surfaces do not slide, and one can only surmise that μ > (Fs,max/Fn), where Fn ≈ Fext + Fpulloff (where Fpulloff is evaluated from WA) as above, yielding μ > 0.095. This value, while consistent with the values seen in Figure A(ii) and (iii), for which μ ≈ 0.09–0.1 and which indicate the presence of trapped counterions, cannot in itself confirm such a presence. This is because it is a lower limit, and it might also be much closer to μ(I) ≈ 0.8, where no counterions are trapped. Importantly, however, we note in Figure B(i) the near-instantaneous reversion of the friction coefficient to μ = (Fs/Fext) ≈ 0.04 on toggling the potential back to Ψapp = −0.3 V. This immediate (<1 s) reversion to the lower friction does strongly point to the presence of trapped counterions even at Ψapp = 0.2 V, deep in the regime where, in equilibrium, such ions should be excluded. This is because it is known both theoretically[37,38] and from direct experiments[29] that penetration of ions into such a narrow slit (<1 nm) between two surfaces, even when favored by their potentials, is kinetically limited, and would be expected to require of the order of 102 s or longer (see also Figure S5). The fact that on toggling from deep in regime I (Ψapp = 0.2 V) to deep in regime II (Ψapp = −0.3 V), the transition to the lower friction—which implicates hydration lubrication by trapped hydrated Li+ ions—occurs so rapidly (<1 s), shows unambiguously that the hydrated counterions are largely trapped between the compressed surfaces throughout.

This scenario is corroborated, as shown in Figure B(ii), when the surfaces are brought to initial contact while in regime I (Ψapp = 0.2 V), compressed by Fext, and the potential toggled repeatedly between regime I and regime II while sliding. On toggling to regime II (Ψapp = −0.3 V), the friction force should be reduced, if hydrated Li+ ions were to enter the intersurface gap (which is the equilibrium state in regime II), to a value corresponding to μ = μ(II). We observe from the trace in Figure B(ii), however, that no sliding occurs. This implies a friction force larger than the maximal applied shear force Fs,max (=228 μN) and thus a friction coefficient μ > (Fs,max/Fn) > 0.25, consistent with μ(I) but in any case far larger than μ(II) ≈ 0.05 ± 0.03 (with Fn estimated as above). This strongly indicates that few if any hydrated counterions entered the gap after being expelled during the initial approach in regime I. This is attributed—as for the opposite case (Figure B(i)) where hydrated counterions were trapped in the gap—to the slow kinetics of hydrated ions entering or leaving the narrow subnanometer gap between the gold and mica surfaces.

Conclusions

To conclude, we have shown that frictional dissipation across aqueous salt solution between a smooth metal surface (gold) at a controlled potential, sliding past a dielectric (mica) at a fixed charge, depends strongly on the hydrated counterions trapped between the surfaces, whose presence in turn depends crucially on the gold potential when the surfaces approach to contact. In particular, hydrated ions in the interfacial gap can strongly reduce the sliding friction through the hydration lubrication mechanism, to friction coefficients as low as μ = 0.05 ± 0.03, compared with the high friction observed (μ ≈ 0.8–0.9) when ions are excluded from the interfacial gap. In this scenario, most of the dissipation is attributed to sliding of asperity contacts between the gold and mica surfaces, where hydrated ions prevent van der Waals adhesion in the valleys separating the asperities. Importantly, within the parameters of our experiments, the initial equilibrium presence or absence of interfacial ions on compressing the surfaces at a given potential is largely maintained during sliding also when the potential of the metal is changed to favor a different equilibrium state. This is attributed to the slow kinetics of hydrated ions leaving or entering the gap, and, for the case where hydrated ions are trapped between the surfaces on initial contact, enables facile, reversible, in situ control of the friction through changes in the metal potential.