Introduction
The increase in the surface-to-volume ratio that occurs when devices are scaled down in size makes friction increasingly problematic in miniature instruments, such as micro- and nanoelectromechanical systems. Indeed, the devices that work reliably usually have designs that avoid sliding contacts. Systems with moving components that come into contact with each other, on the other hand, suffer enormous problems due to stiction, friction and wear. Lubrication is not an option because the lubricant would be too viscous on the nanoscale and, moreover, the adhesion forces introduced by liquids are strong enough to damage tiny devices. However, the development of two new methods that allow the amount of friction to be varied could lead to greater control over this most troublesome force1, 2.
The typical approach to reducing friction is to optimize the properties of the surfaces that come into contact with each other. The idea is to make friction low by the appropriate choice of chemical composition, crystal structure, surface roughness, electrostatic interactions and other properties. Considerable progress has been made over the past decade with special coatings, such as the family of diamond-like carbon films3.
More delicate but perhaps less practical ways to reach ultralow friction involve either an extreme reduction of the contact pressure4 or a cancellation of lateral forces. This latter goal can be achieved by making use of the non-periodicity of quasicrystals5 or by introducing a deliberate lattice mismatch between the two sliding crystal surfaces (a mechanism referred to as superlubricity6). What these techniques all have in common is that they change the energy 'landscape' of the interaction between the surfaces at the atomic scale.
Recently, two new methods have been demonstrated that enable researchers to vary friction continuously by use of an easily adjusted, external control parameter — increasing the friction in one case and decreasing it in the other. Both experiments were performed under ultrahigh vacuum with atomic force microscopes (AFMs) that were operated as lateral or friction force microscopes. In a traditional AFM, an ultrasharp tip on a cantilever is scanned over a surface, and the forces between the surface and the tip cause the cantilever to bend in the vertical direction (that is, at right angles to the surface). By measuring the deflection of the cantilever, it is possible to produce an image of the surface with atomic resolution. And by monitoring how the cantilever twists, the amount of friction can be determined.
The approach described by Jeong Young Park and co-workers1 at the Lawrence Berkeley National Laboratory in California and the Ames Laboratory in Iowa involves a controversial contribution to the friction force, namely electronic friction. The extent to which electronic effects, such as the generation of pairs of electrons and 'holes'7, determine the energy dissipation rate of a sliding contact has been a long-standing matter of debate, and only a few experiments have managed to probe these effects8.
Park and co-workers elegantly and convincingly demonstrate that electronic effects are significant by measuring the friction force between the tip of a friction force microscope (FFM) and a piece of n-type silicon that contains stripes of p-type silicon (Fig. 1a). The p-type stripes were written by implanting boron into the surface. To make sure that structural and chemical details of the surface were not responsible for spatial variations of the lateral force, the entire surface was coated with a thin oxide layer. Two parameters could be adjusted independently: the normal force with which the tip was pressed against the surface; and an electrostatic voltage that could be applied between the silicon and the tip.
Figure 1: Schematic setups for the two friction force microscopy experiments with controllable friction.
a, Park et al.1 scanned a TiN-coated tip on a cantilever (grey) over an n-doped silicon sample that contained p-doped stripes (green). When there is a positive bias voltage on the sample, more friction is recorded over the p-doped regions. b, Socoliuc et al.2 applied an oscillating voltage to set the cantilever into resonant motion, which allows the friction force to be lowered to zero.
Full size image (7 KB)When the surface was negatively biased, the friction was the same for the n- and p-doped regions, even though the tunnelling current that was measured at the same time clearly indicated electronic contrast. At positive voltages above 2 V, however, friction was significantly higher for the p-doped regions. The applied electric field leads to extra attraction between the tip and the surface, but this electrostatic attraction was equally strong for n- and p-regions, ruling out a purely mechanical origin for the difference in the friction. Moreover, the fact that measurements were made on n- and p-type silicon at the same time with the same experimental setup meant that other factors (such as changes in the shape of the tip or temperature) could also be ruled out.
Park et al. suggest several electromechanical scenarios that could be responsible for the surprising differences in the friction they observe. In the semiconductor, the applied electric fields lead to substantial bending of the conduction and valence bands. When a positive voltage is applied to a p-doped region it causes the holes that carry the charge to accumulate near the interface between the silicon and the oxide, whereas a negative voltage leads to a depletion of these charge carriers.
These effects were indeed observed when the tunnelling current was measured as a function of voltage. Park et al. speculate that the high mechanical stress, which is localized under the FFM tip, might lead to the creation of new electronic surface states that need to be populated and depopulated as the tip is dragged over the surface. The discharging of this local quantum dot would increase the rate at which electrons and holes recombine, which would result in the dissipation of energy (that is, friction). It remains to be seen if the same effect could also be used to lower the friction force.
Anisoara Socoliuc and co-workers2 at the University of Basel in Switzerland and McGill University in Montreal took a completely different approach that involved dragging a silicon tip over cleaved surfaces of NaCl and KBr. As in Park's experiment, electric fields were applied between the tip (plus cantilever) and the sample. But this time an oscillating field was applied, so the cantilever started to oscillate as well (Fig. 1b). Similar mechanical vibrations, albeit on much larger scales, have been used to reduce friction for many years9.
In the Basel–McGill experiment the 'shake, rattle and slide' approach is applied at the atomic scale, and the tip is always in contact with the surface. Because of the oscillation of the cantilever, the 'normal' force that pressed the tip against the surface cycled rapidly between a high and a low value, while a (slow) control system kept the average normal force fixed, independent of the oscillation. The experiment was carried out at a range of frequencies, but a significant effect on the friction force was only observed at the resonance frequencies of the cantilever. The measurements made at resonance show that friction is reduced by an amount that is roughly proportional to the amplitude of the drive voltage. For sufficiently large amplitudes the friction force can be reduced to zero.
Because the apex of the tip behaves as a microscopic mass that is connected to the cantilever via a relatively soft spring (the rest of the tip), it has the freedom to move on timescales that are much shorter than the period of the drive signal. This means that we can regard the tip as spending a small but noticeable fraction of its time under conditions of ultralow normal force. As had been established before, the interaction between the tip and the surface becomes weaker when the normal force is reduced. This, in turn, makes the variations in this interaction smaller as the tip is moved over the surface, which ultimately leads to extreme slipperiness4. Socoliuc and co-workers have now performed model calculations to show that the friction lowering is a robust phenomenon that should take place over a range of conditions2.
Both effects could be useful in efforts to develop micro- and nanoelectromechanical systems (MEMS/NEMS)10 that do not tolerate lubrication. It should not be difficult to apply the extra d.c. or a.c. voltages needed to manipulate the friction in these devices. Moreover, if you need to stop the motion for any reason, simply switch off the voltage.