In 1948, Hendrik Casimir predicted1 that two uncharged, perfectly conducting plates in a vacuum would be attracted to each other because of quantum fluctuations in the vacuum's electromagnetic field between the plates. Generalized for real materials by Evgeny Lifshitz2 in 1956, Casimir's prediction has been verified many times and is now known as the Casimir–Lifshitz (C–L) force. But for all systems studied experimentally so far, the C–L force is attractive. Writing in this issue (page 170), Munday et al.3 report the first experimental measurement of a repulsive C–L force.
The attractive C–L force has been measured with great precision, and has been taken into account in the design of nanoscale mechanical devices. But in many instances, the attractive nature of the force has led to more problems than solutions. One such problem is that the components in a nanodevice can stick together irreversibly. The desirability of a repulsive C–L force stems from its potential to fix this problem and also to enable objects to be levitated in fluids, which could find applications in nanotechnology. Proposals for the design of 'metamaterials' capable of producing such a repulsive force have been put forward, but attempts to achieve this have been unsuccessful4.
Munday and colleagues' experiment3 is based on a further generalization5 of Lifshitz's formulation of the force, which allows the vacuum to be replaced by a material — here, a liquid. One of the most precise tests of Lifshitz's theory was performed6 by Edward Sabisky and Charles Anderson in 1973, when they measured the binding energy of a superfluid helium film to a crystal surface.
But Lifshitz's theory also asserts that if the properties of the liquid and plates are appropriately chosen, the C–L force can be repulsive. A repulsive C–L force can be generated by judicious choice of the dielectric permittivities of the plates and the liquid, which describe their ability to store electric-field energy. If the dielectric permittivities of the plates are 
1 and 2, respectively, and that of the liquid in the gap is 3, the force will be repulsive when 1 > 3 > 2. And because these permittivities depend on the frequency of the electromagnetic field, this relationship must hold over the broad range of frequencies that contribute to the C–L force.
If this relationship is met, the C–L force will cause the liquid to wet the material's surface. For example, if one plate is replaced by air or a vacuum (2 = 1), and if the liquid's permittivity is less than that of the other plate, the liquid will spread out in a thin film, rather than forming droplets as is the case with water on an oily glass surface. For instance, liquid helium, which has a very small permittivity, readily forms a thin film on almost all surfaces (except caesium films) because it is 'repelled' by the vacuum (1 > 3 > 2 = 1), or highly attracted to the surface, and so wets the surface. In contrast, liquid mercury, which has a high permittivity, does not wet glass (1 < 3 > 2 = 1).
Although many liquids can wet surfaces such as glass or silica, only a few sets of materials (plate–liquid–plate) will satisfy the requirement for a repulsive force between the plates. The set used by Munday et al. consisted of silica and gold, with bromobenzene as the liquid separating them. The authors' experimental set-up (see Fig. 2 on page 171) used an atomic force microscope that was modified to detect average surface forces rather than atomic-scale point forces. A typical atomic force microscope consists of a microcantilever with a sharp tip at its end that is moved above the specimen's surface. As the tip scans the surface, the cantilever bends in response to the surface force felt by the tip. This bending is monitored by measuring the angular displacement of laser light reflected from the top surface of the cantilever, and allows the force's topography to be mapped out.
To measure the C–L force, Munday and colleagues replaced the sharp tip by a microsphere (of diameter about 40 micrometres) coated with gold. This served as the gold plate. Using a spherical surface for one plate simplifies the geometry of the system, which is completely defined by the radius of the sphere and the distance of closest approach to the flat silica plate. Although this leads to a significant — but easily calculated — modification of the force, it eliminates the need for angular alignment of the plates.
A problem associated with all C–L force measurements is the calibration of the system. Munday and colleagues have come up with a clever technique to overcome this problem. When the separation between the gold sphere and the silica plate is changed, the fluid produces a hydrodynamic force that changes linearly with the velocity at which the separation is altered. By measuring the total force at two different velocities, the hydrodynamic force can be isolated with high accuracy from the C–L force. Scaled to the appropriate velocity, the hydrodynamic force can then be subtracted from the total force at a given sphere–plate separation, yielding a clean measurement of the C–L force. The measurements spanned a range in separation from 20 nm to several hundred nanometres, with the minimum distance being limited by the roughness of the gold and silica surfaces, and the maximum distance limited by the system's sensitivity.
Munday and colleagues' demonstration of a repulsive C–L force is pivotal for both fundamental physics and nanodevice engineering. For example, it might be possible to 'tune' the liquid (possibly by mixing two or more liquids) so that the force becomes attractive at large separations, but remains repulsive at short range. This would provide the means for quantum levitation of an object in a fluid at a fixed distance above another object, and so could lead to the design of ultra-low-friction devices. The applications of the C–L force to nanodevices remain to be investigated, but the prospects look exciting.
