You are on page 1of 10
Contemporary Physies, 1992, volume 33, aumber $, pages 313-322 Casimir forces Prrer W. MiLONNI and Met-L1 SHiH Casimir effects are relevant 10 areas as diverse as quantum field theory, cosmology, high-precision spectroscopy of Rydberg atoms, forces between dielectrics and conductors, atom-surface interactions, and the wetting of surfaces, amongst many others. In this article the basic ideas and results are reviewed. The emphasis is on the basic physics of Casimir forces, but some applications are discussed. 1. Introduction In the 1940s, J.T. G. Overbeek atthe Philips Laboratory in The Netherlands carried out experiments on suspensions of quartz powder used in. manufacturing The results indicated that the theory of the stability of colloids which he had developed with E. J. W. Verwey might not be entirely correct, and that the interparticle interaction might fll off more rapidly at large distances than originally thought, Overbeck suggested that this hhad to do with the finite propagation velocity of light, fand this prompted H. B. G, Casimir and D. Polder at Philips to reconsider the van der Waals interaction. They found that Overbeck’s suggestion was correct; as. a consequence of retardation, the interaction energy varies as m7 rather than r°® for large. intermolecular separations r. Ths could have been nearly the end ofthe story, except that Casimir was intrigued by the simplicity ofthe result which he obtained with Polder and sought a physical explanation for it. This led him to propose a remarkable new effet that has been of interest ever since. Casimir recounts as follows how he arrived ata new way of thinking about the Casimir-Polder result (Casimir 1992 “Summer oF autumn 1947 (but 1 am not absolutely certain that it (was] not somewhat earlier or later) 1 mentioned my results to Niels Bobr, during a walk “That is nice,” he said, “That is something new.” I told +him that I was puzzled by the extremely simple form of the expressions for the interaction at very large distance and he mumbled something about zero-point energy. That was al, but it put me on a new track [ound that calculating changes of zero-point energy Aushors” adress: Theoretical Division, Laboratory. Los Alamos, NM 87545, USA. Los Alamos National co107st6 s1000 really leads to the same results as the calculations of Polder and myself ‘On 29 May, 1948 T presented my paper ‘On the attraction between two perfectly conducting plates’ at a meeting of the Royal Netherlands Academy of Arts and Sciences. It was published in the course of the year, It is the latter fect, a force between two perfectly conducting, uncharged plates, that physicists usually have in mind when they speak of the “Casimir effec’. Iti the association ofthe effect with zer0-point electromagnetic energy, following Casimit’s (1948) original derivation, that makes it so fascinating. Fascinating too is the fact that Casimir in The Netherlands was thinking in terms ofthe zero-point field energy at just about the same time that Welton and Weisskopfin the USA deduced that the Lamb shift could in large measure be attributed to this zero-point eneray. However, it would be false to the facts to regard this as ‘more than coincidence; again quoting Casimir (1992): “No, I was not at all familiar with the work of Welton and others. I went my own, somewhat clumsy way... T do not think there were outside influences... I did not myself contribute to further developments, nor to experimental confirmations.” Casimir calculated the force per unit area, he 240d” between two parallel, perfectly conducting plates separated by a distance (This result i derived in section 4) Thus F/d) = —0013/¢+ dyn em?, where d is in microns, and the force amounts to 04013 dyne for two 1 x Lem plates separated by 1 um. This isa tiny force, comparable with the Coulomb foree holding the electron to the hydrogen atom, or the gravitational attraction Fd) =~ w (192 Taylor and Ferns Lid aia PW. Milonni and Mei-Li Shi between two 1 Ib weights separated by 05 in, It does not involve the masses of the plates or their charge. Its often regarded as ‘one of the least intuitive consequences of ‘quantum electrodynamics” (Schwinger et af. 1978). 2. Zero-point energy AAs already mentioned, the Casimir force between two plates is related to the zero-point energy of the electromagnetic field. The zero-point enerey appeared for the first time in Planck’s ‘second theory’ of black-body radiation, more than a decade before it was actually derived from Heisenberg’s matrix mechanics in 1925. Planck obtained the expression tuo + tho Q) exp (huikT) 17 yn for the average energy of an oscillator of frequency @ in equilibrium with radiation at temperature 7. When T +0, U-+ 4hi. In other words, Planck's work implied the existence of a zero-point energy persisting even at the absolute zero of temperature where, according 10 conventional classical theory, all motion ceases. Planck soon became convinced that this constant additive zero-point energy is of no physical consequence. Einstein, however, was intrigued by the idea. In a paper with Stern in 1913, he noted that a zero-point energy seems necessary in order for equation (2) to give the classical formula U = kT in the ‘classical limit” KT" > he In this limit, retaining terms up to first order in hao/kT, Ga Ake exp (uo/kT)— 1 fo ss + tho TF hwiAT + WoikT—1* kr = ho Te thayer * = AT — tho + tho =ér. co) ‘The zero-point energy She» is thus necessary in order that U does not have first-order quantum corrections to KT in the classical limit kT > tuo. For this and other reasons, Einstein and Stern concluded, “The existence of a zero-point energy [4w] is probable.” Zero-point energy was a ‘hot topic’ during the decade preceding Heisenberg's matrix mechanics and Schré- inger's wave mechanics, mainly because of the great interest at the time in low-temperature phenomena such tas superconduetivity. Debye, for instance, showed that, if Planck's zero-point energy were real, there should be ‘reduction in the X-ray scattering from crystals even as T+ 0 because of zero-point fluctuations about the lattice sites. Mulliken's spectroscopic studies indicated that a zero-point energy of molecular vibrations was necessary to fit the data accurately, and Stern found a zero-point energy to be necessary in calculating the vapour pressure of solids. Early ideas about the zero-point energy have been reviewed elsewhere (Milonni and Shih 1991) It is typical in more recent times for textbooks t0 regard the zero-point energy as a curious prediction of quantum theory that can be discarded by simply redefining the zero of energy. This can be misleading if it implies that the zero-point energy is of no physical consequence whatsoever, for it turns out that changes in the zero-point energy are observable. The most precise observations have been made in connection with changes im the electromagnetic zero-point energy. 3. Electromagnetic zero-point energy ‘The basic idea is very simple: the electric and magnetic field vectors of a monochromatic wave of frequency 1 undergo harmonic oscillations and, when the field is described quantum mechanically it has the same allowed energy levels E, =(n + 49h, m= 0,1,2,3,..., a8 any other harmonic’ oscillator. In the case of the fleld the integer 1 corresponds to the number of photons. The term $a) is a zero-point energy and implies fluctuations of the electric and magnetic fields even when there are no measurable photons in the field In free space, or in a cavity whose dimensions are much larger than the wavelength A= 2xc/o, the number of modes per unit volume in the frequency interval [w, o + do] is given by the standard textbook formula AN, =(c7*/x?c*) de. Since each mode has the zero-point energy $f, the field has @ zero-point energy per unit volume, pate) = 2, thay to bs “ inthe frequency interval [o,0+do]. The total zero-point energy density ofthe field, Ep = J polo) do, is therefore infinite. We shall consider now an example of how one can calculate a physical consequence of a ‘change in this zero-point eneray. Consider a gas of refractive index n(o} contained in a box with conducting walls. The change in the zero-point field energy in the box due to the presence of the mediwn Casimir forces ais inside the box is (0) = $8 tho Able) = ay ko) =tho no) = ~ Shoo) — 1) 0 for |r) — 1] < 1, The change in zero-point energy for all the modes is then ap=v fasten, ae ~1555 ono) — 1] do, (6) where is the volume of the box. ‘Now suppose that the medium inside the box consists of identical one-electron atoms. We then have so) = 1 + 2nNa(a) for |r) ~ 1) <1, where NV is the number density of atoms of polarizability a(o) In the limiting case of a single atom, NV I and, from equation (6), ae=- lig o%e(a) do 0 We interpret AZ as a shift in the energy level of an atom ina state characterized by the polarizability 2). An immediate problem in relating AE to anything physical, however, is that it is infinite. Let us argue that a more physically relevant quantity than AE itself is the difference between the AF values for a bound electron and a free electron. For a bound electron, 2(«) is given by the well known Kramers-Heisenberg formula whereas, for a free electron, a(«») = —e*/mas*t. When we subtract the free-electron energy from equation (7), we are still left with an infinite energy, but let us suppose that whatever physical content might be associated with the change in the zero-point field energy is primarily non-relativistic, so that we can ignore frequencies © > me*/h after the subtraction. Then we can cut off the integration over e at frequency « = me’ /h. When this is done, we obtain exactly Bethe's non-relativistic approx- imation to the Lamb shift. Welton (1948) showed that Bethe’s result could be attributed to the zero-point fluctuations of the electromagnetic field. The derivation just outlined is based on a suggestion by Feynman (1961) and owes much to a paper by Power (19663. It shows 4 See, for instance, Milonni and Ebery (1988), equation (.A25) 1 The free-electron polarizability follows by taking the limit which ‘> is much larger than any transiion frequency appearing in the Kramers-Heisenberg formula, and using the Thomas-Reiche-Kuka sum rule. See, for instance, Milonns and Eberly (1988) p, 239. {See also Milonni and Shi (1992), and references therein, that the change in the zero-point energy of the field, due simply to the presence of an atom, can be related to a ‘measurable shift in the energy levels of the atom. 4. Casimir force between perfeetly-conducting plates {in the preceding discussion we considered the change in the zero-point electromagnetic energy in a box when a ‘medium of refractive index n(w) is placed in the box. Now suppose that there is no material medium in the box, which is assumed to be a parallelepiped with perfectly conducting walls of length L, = L, = Land L. = d. The allowed frequencies of any radiation ‘trapped’ in the box are restricted to a discrete set of values, whereas with no box all frequencies are possible. This means that the zero-point field energy is affected by the presence of the ‘box and is different from what it would be were there no box. The question arises: does this change in the zero-point energy result in any force on the box? The answer is yes, there is @ (Casimir) force on the box. The allowed frequencies are given by the well known formula, + where /, m and 1 take on all positive integer values and zero. The zero-point energy in the box is therefore El) =F’ (2)8hc ne ieee Ey ° That ate The factor of two arises from two independent polarizations of modes with J, m,n #0, and the prime ‘on the summation symbol implies that a factor of a half should be inserted if one of these integers is zero, for then ‘we have just one independent polarization, We are interested in the case Z >» d, where we can replace the sums over / and m by integrals, so that = an te f (ear +Zz) J If d were made arbitrarily large, the sum over m could also be replaced by an integral, and the zero-point energy (10) would be . ay The potential energy of the system when the plates are separated by a distance d is U/d) = Eq(d) ~ y(-2), the energy required to bring the plates from a large Eqlo0) 316 PW, Milonni and Mei-Li Shit separation to the separation d: onl (al w(esest) wt fafa eae] (12) Going to polar coordinates and making a simple ‘transformation of integration variables, we have wan PTS Candee tey -() fe [eee] he - ait E(B [ate -f ae f dx w+) (13) Thins eee ren oe Gop bord Were ue =" ufsro+ E r if sna) 0 ens fm-[en —4FO)— hF'(0) + rho FO) for the difference between a sum and an integral, where the primes denote derivatives. Noting that F'(2) we see that #(0) = 0, #0) = ~4, and that all higher derivatives F"(0) vanish. Thus 37", F(a) — f8' dz Fle) = —4F(0) — vio and the whe ua== 3 = Fee an This implies that the force per unit area is Fd) = — U(d) = —nhe)240d*, the Casimir force. + See, for instance, Abramowitz and Stegun (1968). We ure applying the formula under the assumption that Rc) -O- This can be realized by introducing & cu-of function analogous to the cae of the non: relativistic Lamb shit. The rest In this ease is independent of the ‘Choice for this out-of funtion, We can also derive the Casimir force between conduct ing plates in terms of radiation pressure (Milonni et al. 1988). The idea here is that virtual ‘zero-point photons carry linear momentum }fk, where k is the wave-vector of a plane wave of frequency = |kje. Then the reflec- tions off the plates of the zero-point field outside the plates act to push the plates together, while reflections of the field confined between the plates act to push them apart. The net effect, it turns out, is the Casimir force acting to push the plates together. The existence of the Casimir force between two con- ducting uncharged plates has fascinated physicists for many years. The typical reaction that they have when learning about the effect is illustrated by DeWitt’ re- collection (1989) of a seminar by Casimir around 1960: *What startled me, in addition to the crazy idea that fa pair of electrically neutral conductors should attract ‘one another, was the way in which Casimir said the force could be computed, namely, by examining the effect on the zero-point energy of the electromagnetic vacuum ‘caused by the mere presence of the plates. I had always been taught that the zero-point energy of a quantized field was unphysical To arrive at a less ‘crazy’ explanation for the Casimir force, we now turn to the quotidian van der Waals interaction, 5, yam der Waals forces tis household knowledge that van der Waals proposed an attractive force between molecules in order to explain deviations from the ideal gas law. It was nearly 60 years before such a “universal” interaction, independent of temperature and of whether the two molecules have permanent multipole moments, was derived from first principles. The derivation was given by London (1930), who obtained, using fourth-order quantum-mechanical perturbation theory, an attractive interaction varying as. the inverse sixth power of the intermolecular distance. London supported his calculation with a simplified derivation based on the zero-point energy of the two molecules, treating them as harmonic oscillators. It is also possible to derive the van der Waals interaction based on the zero-point field energy. Such a derivation gives both London’s result and its modification for large interatomic separations due to Casimir and Polder (1948)f, This derivation, or a more conventional one ‘based on perturbation theory, leads to the expression she = ne My (18) w= -pe +See also Milo and Stik (1992), and releronoss thers, Casimir forces 317 for interatomic separations r large compared with transi- tion wavelengths of the atoms, where 4. = 24,9(0) are the static polarizabilities. This is the Casimir-Polder result mentioned in section I. For r small compared with the atomic transition wavelengths, one obtains instead the usual ‘non-retarded” London-van der Waals interac tion of the form U(r) cc r-°. ‘The transition from an r~* to an 7-7 interaction at large interatomic separations is a consequence of re- tardation, that is the finite velocity of light. Another ‘example of this sort of behaviour, involving an atom near ‘a conducting wall, was also worked out by Casimir and, Polder. For small distances d of the atom from the wall the attractive interaction may be obtained from the dipole-dipole interaction of the atom with its image in the wall and varies as d~>. For large distances, however, Ud) falls off as d~*; as in the case of the van der Waals interaction between two atoms, the effect of retardation, is to weaken the interaction by a factor proportional to the reciprocal of the distance. This effect has just recently been corroborated experimentally (section 6). If two atoms have a van der Waals attraction, there must be a macroscopic manifestation of this force be- tween dielectric bodies. Consider first the example of a single atom at a distance d from a dielectric wall consist- ing of N identical atoms per unit volume, each of polariz~ ability a. For large d the interaction between the atom, atd and cach atom of the dielectric is given by equation (18). A schoolboy integration over all the atoms of the dielectric results in the following expression for the interaction energy between the atom at d and the wall sub a Pe 8. fe ua) = —n Pie fiefl of x dole? by + _ Bhea 1 (any40) a* 4nNa. a9) ‘The Clausius-Mosotti relation 1 = (3/4xNMe = 1)/le + 2), then gives 69hex 1 e~1 1600 de +2 Consider now the limit of a conducting wall. We can (usually) obtain results for a conductor by taking the limit ¢— oof). Taking this limit in equation (20), we have ua (20) 4 This typically requires to be large computed with 137 Bohr radi ‘{Forexample, the foros between a charge q and Wis image tra dest wall is (= 1)(e-F N)2)20%, and the limit «> ce gives the force when the wall i a perfect conductor. Fara recent discussion ofthis Tit, see Price (1992, z=0 ze=d Figure 1. Geometry assumed in the derivation ofthe Lifshitz force (22) between two semi-infinite dielectric slabs separated by 1 layer with dielectric constant, U(@) = ~69hex/160nd* for an atom of polarizability « near a conducting wall, in very good agreement with the Casimir-Polder result U(d) = —3healSnd In exactly the same fashion we can obtain the force per unit area between two conducting walls by an inte- gration over pairwise interatomic interactions (18) fol- lowed by the limit ¢—+ 20. The result is 9x 2h 640m? a which is about 80% of the Casimir force (1), ‘Why does not this simple approach give exactly the Casimir-Polder and Casimir forces? The reason can be traced to our assumption of an additive pairwise interaction between atoms. In fact the van der Waals forces are non-additive; the interaction between two ‘atoms is affected by the presence of a third atom (Axilrod and Teller 1943, Power and Thirunamachandran 1985, Milonni and Lerner 1992). The problem of including non-additive effects in the forces between macroscopic bodies was solved approximately by Lifshitz (1956). The most basic feature of the Lifshitz theory is that the field mediating the interaction between any two atoms propa- gates through the medium formed by all the other ‘molecules, and the effect of these other molecules is included through their effect on the refractive index, The result of the Lifshitz theory is that the force per unit area for the diketric configuration shown in figure 1 i atu at ter. (2indtd) 7 -l Slee al 681-47 Gap Ca} "See ako Milonni and Shi (1992), and references thers, Fid)= 1) Sor 2 é ea? Ao — 585 [ ave [area + Pate, Ss —PS2—P 318 PW. Milonni and Mei-Li Shik where «= fig) is the dielectric constant of region j evaluated at the imaginary frequency if and 42 = (9? — 14 6.9/6)". For rarified media this reduces to ‘our results Obtained by simple integrations over pairwise forces. For «oo it yields exactly the Casimir force (1. In other words, when non-additive contributions to the van der Waals force are accounted for, the Casimir force appears as just the force between dielectric slabs in the limite» 2c For finite dieletric constants, equation (22) yields a force per unit area varying as d~* for small separations and as d~* in the “retarded” regime of large separations. More generally, electromagnetic Casimir effects can be regarded as macroscopic manifestations of van der Waals forces, which ate themselves explainable in terms of changes of the zero-point electromagnetic energy because of the presenee of matter. As in the ease of the Lamb shift, the concept of the zero-point field energy is not necessary for the calculation of interatomic van der ‘Waals interactions, although it does in our opinion provide more physical insight than can be gleaned from standard perturbation theory. The calculation of Casimir eflects, however, is greatly simplified by employing ideas of zero-point energy, allowing us to replace a many-body problem by an essentially classical electromagnetic boun- dary-value problem (Milonni and Lerner 1992). 6. Experiments ‘The experimental literature on Casimir forces is vast if ‘one includes dielectries as well as conductors. The first successful experiments were reported by Derjaguin et al (1951). The measured forces were between a glass plate (4mm x 7mm) and spherical lenses, a configuration ‘more easily adjustable than that of two flat plates, and one that allows the closest distance of separation to be ‘measured optically from the diameter of Newton rings. From the measurement of the force between the lens and the plate, one can infer the interaction between two flat plates and compare the results with the Lifshitz theoryS, A negativeseedback technique was used in which a + Various other cases, such as that of imperfect conductors, may be ‘obiained rom the Lihi theory when some algebrae erors inthe ‘original paper are corrected. See for instance, Schwinge et al (1978) fr Milo (1993) Por more detailed discussion and references on this aspect of the Subject see Derjagun ef wl (1978) of Sparnaay (1589) This comparison rests om the soalled Degjaguin approximation, Which pies the force between two spheres with radii A and Ry epurated by a distance de Ry. Ry as Fld) = 2eRyRyaiRy +R) where id) the interaction energy per wait area between (wo flat Surftces separated by d. Thit result may be derived under the ‘sumption of pairwise adulitive ineratomie forss but has & more ality 30 Tong a8 d «Ry, Rs displacement of a knife-edge balance beam produces a current acting to restore the beam to its equilibrium position, and the force was determined through a ‘measurement of the current (Derjaguin and Abrikosova 1957). The data conflicted with estimates based on pairwise interatomic forces and provided the first “good. quantitative agreement’ with the Lifshitz theory in the retarded regime (Derjaguin er al, 1978). Precise com- parison with the Lifshitz theory was complicated by the appearance in the theory of the dielectric constant ‘over all frequencies, and in particular by a lack of data on high-frequency absorption spectra, Barly experiments between flat glass plates separated by 06-15 jm were carried out by Overbeck and ‘Sparnaay (1954). Tabor and Winterton (1969) reported, the first measurements of forces between dielectrics with such small separations that retardation effects are negligible, They used extremely smooth thin sheets of mica on glass cylinders to probe the transition between retarded and nonretarded forces, which they determined to occur between d= 120 and 500A. Hunklinger et al (1972) developed a ‘dynamic method” for measuring ‘macroscopic van der Waals forces, or actually the derivative of the force with respect to the separation, and results “in good agreement with theory” for borosilicate lass were reported for separations between 008 and 1-2 ym, van Blokland and Overbeck (1978) measured forces between a chromium plate and a chromium sphere at distances between 013 and 0-67 ym. Using in the Lifshitz formula an analytical expression for the dependence of the dielectric constant on frequency and conductivity, they state, “The measured force and the calculated force are in excellent agreement when the absorption band of chromium is taken into account in the calculation of the force. Surface roughness exists and it prevents measurements at small separations but has hhardly any influence on the measured forces.” The Lifshitz theory has also been corroborated in experi- ments of Sabisky and Anderson (1973) involving ‘acoustical interferometric measurements of the thickness of liquid-helium films. ‘The agreement with the Lifshitz theory was excellent, although, as in the work of van Blokland and Overbeek, a somewhat phenomenological expression for the dielectric constant as a function of frequency had to be employed. Sparnaay (1957, 1958) reported the first experiments to test the Casimir theory of the force between conducting plates. The force between chromium, chromium steel and aluminium plates at separations between about 03 and 2am was calculated from the deflection of a spring, which was determined from the change in the capacitance of a capacitor connected to a beam attached to one of the plates. Sparnaay’s results are often cited as experimental confirmation of the Casimir forces 319 Casimir force. Sparnaay himself was more cautious; he wrote ‘The observed attractions do not contradict Casimir’s theoretical prediction.” The relative errors in, the measured forces were large (about 100%) owing to errors Ad~0'3 wm introduced by hysteresis in the capacitive system used to measure the plate separation, Although it might be argued that the data do not provide an unambiguous test of equation (1), and that one might ‘even fit the experimental points reasonably well assuming a d~> dependence of the force on the plate separation, Sparnaay’s measurements remain impressive, It should be mentioned that the force derives mainly from field modes of frequency wg ~c/d. Finite- temperature corrections to the force, which are calculated by retaining the first term on the right-hand side of equation (2) for the energy of each field mode, are negligible when kT « cq. For d of the order of a few microns, then we can in effect assume the T=0 limit described by equations (1) and (23). More recent experiments have focused on the Casimir-Polder interaction between an atom and a conducting wall (section 5). At small distances the interaction energy varies asd * and is easily understood, from the interaction of the atomic dipole with its image* Shih and Parsegian (1975) observed the deflection of atomic beams by gold surfaces and obtained results, consistent with the d* interaction. Anderson ef al.(1988) achieved much larger atom-surface van der Waals forces by using highly excited caesium and sodium atoms; the electric dipole moment of such a Rydberg-atom scale as, the square of the effective principal quantum number n, and therefore the van der Waals force scales as n*. Furthermore the Rydberg atoms have large transition wavelengths, and so the range of their van der Waals interaction with the surface is substantially greater than that for ground-state atoms. The experiments involved passage of a beam of Rydberg atoms through an 8 mm channel formed by two gold mirrors separated by variable distance between 2:1 and 8:5 ym. Atoms with, larger n values are attracted to the mirrors more strongly than atoms with smaller m values, and so fewer of them escape the channel without sticking to one of the surfaces. One can define a maximal n for which an atom, can pass through, and a classical trajectory analysis indicates that this maximal n is proportional to the } power of the channel width if the atom-surface interaction varies as d~?. The experimental data were consistent with such a scaling. Deflections of ground- state atoms in similar experiments have also been observed by Hinds et al. (1991), As in the earlier experiments of Shih eal (1974) and Shih and Parsegian (1975), the data were not sufficiently accurate to distinguish between the d~? interaction and the Casimir-Polder retarded d~* interaction, Recent experiments, however, have measured the atom-surface interaction spectroscopicaly asa position- dependent energy level shift (Sandoghdar er a. 1992). In the most recent experiments the d"* interaction has been ‘observed, evidently for the frst time, by Hinds (1992) (ee Sukenik er af, 1992), These results appear to provide the most accurate evidence to date for the redaction in a van der Waals interaction by a factor proportional to the reciprocal distance due to retardation, Experiments of a different type have been in progress to observe the Casimir-Polder interaction with a glass surface of slowly moving (T~ 1 1K) atoms from an ‘atomic fountain® (Kasevich er al. 1991), Surprisingly enough, Casimir effects can occur even within a single atom, as first discussed by Spruch and Kelsey (1978), Consider first a stationary electron at a distance r from an atom, We can expect the electron to induce a quadratic Stark shift in the atom’s energy level, Proportional to the square of the Coulomb field and therefore to r*. It tums out, however, that for large r there is a contribution to the interaction given approximately by Ihe*a/4nmer’, where a is the static polarizability of the atom$, This result was obtained independently by Bernabeu and Tarrach (1976) for neutral atoms and by Kelsey and Spruch (1978) for a charged polarizable system. Spruch and Kelsey (1978) then showed that this result could be deduced by considering a change in the zero-point electromagnetic energy, in the same spirit as the derivations of the Lamb shift and the van der Waals interaction mentioned earlier. They suggested that this Casimir effect might be observed in the energy levels of Rydberg atomic states; in this case the polarizable system is the ionic core of the ‘atom and energy shift is due to an electron in a highly is given by an expectation value This suggestion stimulated high-precision spectro- scopic studies of Rydberg helium atoms by Lundeen e¢ al, (see Hessels et al. 1992, or the review by Lundeen 1992), Their experiments measure fine-structure inter- vals of high-angular-momentum states of Rydberg helium atoms with principal quantum number n= 10, for which the separations of states of different orbital angular momentum have been calculated very accurately * Atoms, of course, have no permanent dipole moments. What the theory lads to is the squat ofthe atomic dipole moment, which does have a non-vanishing expectation value. 414m contrast with the effet of retardation on the van der Waals interaction, ths is an addon to the non-rtarded interaction rather than a replacement, 320 PW. Milonni and Mei-Li Shih without the retarded Kelsey-Spruch interaction. When the differences between the measured and calculated energies are compared with the energy shifis predicted from the Kelsey-Spruch interaction, they are found to bbe an order of magnitude too small. This observation stimulated theorists to determine a more general form of the retarded interaction which reduces to the Kelsey- Spruch result at very large distances (r> 137d) but ‘which remains valid down to distances of the order of (10-50\a, which are important for the binding energy of n= 10 helium Rydberg states. These calculations are confirmed at the few percent level by comparison with experimental fine structure experiments. 7. Other applications and implications Interest in Casimic effects extends far beyond forces between atoms and surfaces, forces between dielectrics for conductors, or high-precision spectroscopy of Rydberg atoms, ‘A possibly very profound implication was suguested by Casimir (1953), Consider a spherical conducting shell of radius a, The zero-point field energy Bla) = 4 Y4 he, can be expected to come predominantly from frequencies to ~ c/a, and we can expect furthermore from dimen- sional considerations that 23) where Cis some dimensionless factor. Casimir proposed that an electron might be envisaged as such a spherical shell with a charge e, and that the attractive force associated with equation (23) might exactly balance the outward repulsive force associated with the electrostatic sell-energy U= e%/2a. The force per unit area corre- sponding to equation (23) is F(a) = —Che/8na*, and the repulsive electrostatic force per unit area is e*/8na*. The condition that these two forces balance each others then C=e?/he. In other words, Casimir suggested that an attractive force duc to the electromagnetic zero-point nergy might provide the “Poincaré stress’ necessary, from a classical perspective, to hold the electron together In particular, the value of the fine structure constant according to this ‘admittedly very erazy model’ (Casimir 1953) is just the number C appearing in the energy (23) and is independent of the presumed electron radius. Unfortunately this fascinating model fails because the factor C in equation (23) turns out to be negative, that is the force associated with the zero-point energy in this ease is repulsive rather than attractive. This result was first obtained by Boyer (1968), who computed C= ~009, implying that € not only has the “wrong” sian, but is also about 12 times larger than the fine Structure constant (see also Davies 1972), Similarly, Balian and Duplantier (1978) and Milton et al. (1978) have obtained C= ~0092 35 (soe also Candela 1982). It might be noted that Casimir (1978), a quarter of @ century after his original suggestion, wrote that he remained ‘reluctant entirely to give up the idea that the valve of [the fine structure constant] has something to do with a compensation of zero-point energy and electrostatic eneray’ and speculated on another possi- bility Thterest in the Casimir effects in quantum field theory and particle physics is not surprising when one recognizes that all quantum fields have zero-point energies. These zero-point energies are not necessarily positive; the zero-point energy of a fermionic oscillator is —}ho rather than the +Sho of a bosonic oscillator such as an electromagnetic field mode. The Casimir eflets arise, as in the electromagnetic case, when boundary conditions are imposed on these fields. ‘An example where these ideas have been applied isthe phenomenological “MIT bag model’, where a hadron is regarded as a finite region of space confining quark and gluon fields, such that the quark and gluon currents through the walls of the bag are zero, In the case of a sphere, one might reasonably guess, based on the electromagnetic case of a spherical shell and the dilference in sign between fermionic and bosonic zero-point eneries, that the quark contribution should be negative. However, it tums out that, when contributions from exterior modes are included, both the quark and the gluon contributions to the bag enerey are positive (Milton 1983), The Casimir contribution appears to be roughly about 9% of the total bag energy} (Mostepanenko and Trunov 1988) “The Casimir effect has also appeared in the context of black holes, particularly in connection with ‘wormholes” ‘A wormhole is a space-time tunnel connecting a black-hole-white-hole pair formed when a black hole turns inside out. The tunnel closes before anything can propagate across it, but it seems possible under certain circumstances to keep it open if conducting end caps are placed at the ends. This possibility results from the ‘modification ofthe vacuum by the Casimir effect and in particular from the negative energy density between the caps, and might in principle allow for “time travel” (Morris etal. 1988) ‘The most direct implication of the Casimir effects in cosmology arises in connection with the cosmological constant problem. The reality of zero-point energies demonstrated by the existence of the Casimir forces evidently means that zero-point energies should be taken seriously in general relativity. When thisis done, the total ‘In practice the bag parameters are taken tobe adj used to tthe hadron spectra and magetc moments Casimir forces 321 zero-point eneray density of the vacuum acts in effect as a cosmological constant of the type introduced by Einstein in order to have static solutions of his field equations. However, astronomical data indicate that any such cosmological constant must be many orders of magnitude smaller than predicted by quantum field theory (Weinberg 1989). This difficulty remains un- resolved. In ‘self-consistent’ models of the Universe, gravitational ficlds arc assumed to be generated by vacuum energy, which in turn is ‘polarized’ by the sravitational field (Mamaev and Mostepanenko 1980), In touching upon these more speculative aspects of the Casimir-type effects, we certainly do not wish to leave the impression that the original electromagnetic Casimir effects have themselves been fully explored, Barton (1991), hhas noted that the stress tensor appropriate to the vacuum between conducting plates does not commute with the Hamiltonian of the quantized electromagnetic field, and therefore that it is subject to fluctuations. He calculates mean-square fluctuations, including averages over finite times and areas appropriate to some ‘measurement apparatus, and finds them to be too small to be measured directly. Scharnhorst (1990) and Barton (1990) have shown that the refractive index of the vacuum, between parallel conducting plates can be less than unity for all wavelengths greater than the electron Compton radius. This results from virtual electron-positron pair production between the plates, leading to a decreased Polarizability of the vacuum and to phase and group velocities exceeding the speed ¢ of light in the free-space vacuum. The effect is far too small to imagine experimental tests, but it raises the important point of whether it is possible in principle for quantum. electrodynamics to allow signal velocities exceeding ¢, Milonni and Svozil (1990) argued on physical grounds that this effect cannot give measured signal velocities exceeding c, and Ben-Menahem (1990) arrived at a similar conclusion based on an analysis of a causal Perturbation expansion rather than Feynman graphs. More recently Barton and Scharnhorst (1992) have argued that, aside from measurement considerations, either the signal velocity c/n(zc) > e or the vacuum, between the plates must be unstable in the sense that it can amplify light signals in some frequency ranges. When dielectric as well as conducting surfaces are considered, the range of applicability of the electro- magnetic Casimir effects is enormous. For example, fone of the first successes of the Lifshitz theory was in explaining the wetting properties of liquid helium (Dzyaloshinskii er af. 1961). It turns out that, because the dielectric constant of liquid helium is so small (¢ = 1-057), the macroscopic van der Waals (or Casimir-Lifshitz) force across an adsorbed liquid helium film is generally repulsive and therefore tends to thicken the film. This explains the remarkable property of liquid helium of climbing the walls of a beaker; the experiments of Sabisky and Anderson (1973) cited earlier compared the thickness of a liquid-helium film as a function of height with the predictions of the Lifshitz theory. The repulsive nature of the force, in this case as a consequence of retardation, is also believed to account for the spreading of pentane on water, whereas many other hydrocarbons, experience attractive forces and consequently form Tens-like globules on a water surface In his autobiography, Casimir (1983) remarks that the effects named after him *have turned out to be of some theoretical significance.’ We hope that this survey has conveyed some sense of this enduring significance, Acknowledgements, We thank Dr Casimir for his kind reply to our query, and for granting us permission to quote from it. The first author acknowledges brief but informative discussions during the past year with G. Barton and E. A. Power. E. A. Hinds and S. R. Lundeen kindly provided information about their experimental work, and R. E. Peierls and L. Spruch identified some possibly misleading statements in an carlicr version of the manuscript. Finally we thank the editor for the invitation to write this article, which helped us to widen our perspectives on the Casimir effects. References Abramowitz, M. and Stegun, LA, 1968, Handbook of Mathematical urtions, Washington, DC-US Government Printing Oke. 16. ‘Anderson. A., Harochs,S. Hinds, E. A, She, W. and Meschede, D., 1988, Phy. Ret. A, 37, 3894, Axilcod, BM. and Teller, E_ 1943, J. Chem, Py, 1, 299 Balin, Rand Duplantier, By 1978, Ann. Pys. (N.Y), 11, 165 Barton, G. 1990, Phys.Lett, B, 237, $89 (1990) 191, J. Phys. A, 24 981, 5533. Barton, Gand Scharahorst, K., 1992, Preprint. ‘Ben-Mezahens, 8, 1980, Phy Let, B, 280,135 Bernabéa, J, and Tarrach, R, 1976, An Phys, (N-¥.), 102, 323. Boyer, TH. 1968, Phys. Ret 174, 1764, CCandelas, P1982, nm Phys. (N.Y), 148 281 Casimir, H.'B. G., 1948, Proc. K. Ned. Akad. Wer, St, 79%: 1953, Physica, 19, $46; 1978, Re. Rownaine de Physigu, 23, 723: 1983, Maphazard Realty (New York: Harper and Kos); (2 March 1992, private communication to P. W. Mion (Casimir, H. BG, and Polder, D, 1948, Pps. Ren. 78,360 Davies B, 1972, J. Math Py, 13 1324 Desagui, BV, and Abrikosova 1, 1957, Soir Phys. JETP, 3819. Deraguin. B. V,Abrikosova, LL, and Leib, E, 1951, Veuiike dkad ‘Nauk SSSR, 6 128 Derjaguln, 8. V, Rabinovich, ¥.1, and Churses, N. 1978, Nan, Mi. DeWitt, B, 1989, Physics inthe Making, edited by A. Sarlemiin and 1M. J. Sparaaay (Amsterdam: Elsevier Dayaloshinshi, FE Lititz, EM. and Pitaevski, L. P, 1961, Ads Phys. 10, 165. 32 PW. Milonni and Mei-Li Sih Feynman, R. P, 1961, The Quantum Theory af Felis, edited by R. ‘Stoops (New Vork: Wiley-lnterssenes} Hessel EA, Arcuni, P. W. Deck, F-4, apd Lundeen, SR, 192, Phys Rex A, 46,2622 Hinds, EA. 1992, Lectwe, Rank Prize Funds Minisymposium on Talorng Spontaneous Emission for Optical Devices, Grasmere, 22 September 1992, Hinds, EA, Sokenik, C. 1, Boshee, M.G, and Cho, D 1991, Atomic Physics, Vol. 12. edited by R. Lewis and J, Zorn (New York ‘American laitue of Physics) Hunklinger, S, Gosselmann, Hand Arnold, W., 1977, Rev sien Trsiram. 48, S84 Kasevich, M, Mole, K. Ris, E, Sunderman, E, Weis, D, and Chu, 'S, 198i, Atomic Physics, Vol 12, eaited by R. Lewis and J. Zorn (New York: American Insitute of Physics), Kelsey, and Spruch, L, 1978, Phys. Ren. A, 18,15 Lishit,E. ML 1986 Soviet Phys JETP, 2, 73 London, F 1930, . Pay, 68,28, Landen, SR,” 1992. Long Range Forces: Theory and Recent Experiments Atomic Systems, edited by FS, Levin ard D. Micha (New York: Plenum) Mamaey, 5. G. and Mostepanenko, V. M, 1980, Soviet Phys. JETP, 1,9, Milonni, P. W. 1993, The Quantum Vacuum (Boston, Massachusetts: ‘Academie Pres (tobe published) Milonni, P. W, Cook, Rand Goggin, M. E, 1988, Phys: Rew A, 38,16 Milonai P. W. and Eberly, 1H 1988, Lawes (New York: Wiley Milonn, P. W. and Lerner, P.B, 1992. Phy. Ret. A, 46, L18S Milonni, P. W, and Shih, M-L., 1991, Ame J. Phys. 58, 684; 1992 Phys.Rev. A, a8, 424, Milonni, P. W. and Svouil, K. 1990, Phys. Lett. B, 248,437 Milton, K. Ay 1983, Ann Phys (N.Y), 180,432 Milton, KA, DeRaad. L L, Jr, and Schwinger, J, 1978, Ann Phys (NY) MS, 388 Mortis, M.S, and Thoen Moris. M.S. Thorne. K.S. 61, 1446 Mostepanenko, ¥. M, and Trunoy, N.N, 1988, Suet Phys Usp. 3, 965 Overbeck, J.T. G. and Spasnaay, MJ, 1954, Discuss. Faraday Soe., 1812. Power, EA, 1966, dm J. Phys, 34,316 Power, EA, and Thiamachandran, T, 1985, Proc R Soc. A, OL, 267 Prive, RH, 1992, Bur J. Phys. 13,95. Sabinsky, ES, and Anderson, C. H, 1978, Phys: Ret. A, 7, 790 Sandoghdar, V, Sukeni, C.1, Hinds, E. A. and Haroche, S, 1992, Phy, Ret. Let, 68, 432, Scharnhorst, K. 1990, Phys Lett, B, 236, 354 Schwinger J, DeRaad, LL Jr, and Milton, KA, 1978, Ann. Ps: (NY) MS. Shih, A, and Parsegian, VAL 1975, Ps. Ret, Ay 12.835, Shih, A, Raskin, D, and Kusc, P, 1974, Phys. Rew. A, 8, 682. Sparnasy, M, J, 1957, Nature, 18, 334; 1958, Physica, 24, 75; 1989, Physics inthe Moking, edited by A. Sarlemin and M. J. Sparnaay (Amsterdam: Ekevier Spruch, L, and Keliey, EJ. 1978, Phys. Rer. A, Sukenik, C. L, Bother, M.G., Cho, D., Sandoghar, Vand Hinds, A 1993, Phys. Ree. Lett, 70, S60 “Tabor, D, and Winterton, RH. S, 1969, Proc. R. Soe, A312, 435 van Bloklond, P. H.G.M, and Overbeck, 1.T-G, 1978.1 chem. Soe. Faraday Trans, 74, 2637 Weinberg, S. 1989, Rev. mod Phys, 61, Welt, TA, 1948, Phys: Ren. 74, 11ST S, 1988, An J Pi. $6,395. ind Vurtsever, U., 1988, Ph. Ree: Lett, Peter W. Milonni has been a staff member in the Theoretical Division of the Los Alamos National Laboratory since 1986. Previously he was a Professor of Physics at the University of Arkansas, a Senior Staff Engineer with the Perkin-Elmer Corporation, and a Research Physicist at the Air Force Weapons Laboratory. His interests include quantum optics and electrodynamics, photonics, non-linear dynamics and the history of quantum theory. Dr Milonni is a co-author of Lasers 1988 (New York Wiley) and Chaos in Laser-Matter Interactions 1987 (Singapore: World Scientific), a co-editor of Physies and Probability (to be published by Cambridge University Press) and is currently trying to complete a book on The Quantum Vacuum to be published by Academic Press. He is a member of ‘The American Association of Physics Teachers, and Amnesty International, ‘Mei-Li Shih received a PhD degree in Physics from the University of Arkansas in 1985 and was awarded the Aubrey E. Harvey Award for Outstanding Graduate Research in the College of Arts and Sciences, She is a co-author of Chaos in Laser~Matter Inieractions and is currently a Collaborator in the Theoretical Atomic and Optical Physies Group at Los Alamos. Her interests include non-linear optics, chaos and fundamental physics.

You might also like