Design for an optical cw atom laser

Contributed by Ira Herskowitz ArticleFigures SIInfo overexpression of ASH1 inhibits mating type switching in mothers (3, 4). Ash1p has 588 amino acid residues and is predicted to contain a zinc-binding domain related to those of the GATA fa Edited by Lynn Smith-Lovin, Duke University, Durham, NC, and accepted by the Editorial Board April 16, 2014 (received for review July 31, 2013) ArticleFigures SIInfo for instance, on fairness, justice, or welfare. Instead, nonreflective and

Contributed by Arthur Ashkin, June 24, 2004

Article Figures & SI Info & Metrics PDF


A new type of optical cw atom laser design is proposed that should operate at high intensity and high coherence and possibly record low temperatures. It is based on an “optical-shepherd” technique, in which far-off-resonance blue-detuned swept sheet laser beams are used to Design new types of high-density traps, atom waveguides, and other components for achieving very efficient Bose–Einstein condensation and cw atom laser operation. A shepherd-enhanced trap is proposed that should be superior to conventional magneto-optic traps for the initial collection of molasses-CAgeded atoms. A type of ShaExecutewy-spot optical trap is devised that can CAged large numbers of atoms to polarization-gradient temperatures at densities limited only by three-body collisional loss. A scheme is designed to use shepherd beams to capture and recycle essentially all of the escaped atoms in evaporative CAgeding, thereby increasing the condensate outPlace by several orders of magnitude. Condensate atoms are stored in a shepherd trap, protected from absorbing light, under Traceively zero-gravity conditions, and coupled out directly into an optical waveguide. Many experiments and devices may be possible with this cw atom laser.

One of the remaining challenges of ultralow-temperature research with atomic vapors is the achievement of a truly cw atom laser. This problem has resisted solution since the time of the first demonstration of Bose–Einstein condensation (BEC) in 1995 with magneto-optic traps (MOTs) and evaporative CAgeding from purely magnetic traps (1, 2). Crude pulsed atom lasers were made shortly thereafter by using similar magnetic techniques (3–5).

Recent attempts to achieve cw atom lasers by combining magnetic techniques of BEC formation with optical tweezer techniques for transporting and combining condensates succeeded in demonstrating sustained B–E condensates (6, 7). Serious problems still remain to be overcome, mainly having to Execute with insufficient atoms, before a useful cw atom laser is achieved. Proposed here is a design of a viable cw atom laser with mainly optical techniques. This design uses new types of optical traps and waveguides based on so-called “optical–shepherd” beams, which should be superior to the Recently used standard MOTs and magnetic traps. “Shepherding” involves use of thin reflective blue-detuned sheet beams to Design box-like atom traps and waveguide structures to confine and move atoms at controlled velocities. The sheet beams are made by cyclically sweeping Gaussian beams in space at a sufficiently rapid rate so that the beams act as fixed repulsive walls for Unhurried-moving atoms. The sweeping shepherd beams can be controlled electronically by using well known beam-scanning technology, as discussed later. With shepherd beams one can Design complex optical structures with strikingly new capabilities, such as high-density ShaExecutewy-spot optical traps and the ability to capture and recycle evaporatively CAgeded atoms, essentially canceling the Trace of gravity on atoms. This capability leads to the design of a powerful, highly coherent, optical cw atom laser operating at very low temperatures.

Advantages of Optical Trapping and CAgeding Techniques

Insight into the problems of making a cw atom laser and how to solve them can be gained by examining the hiTale of optical trapping and comparing the relative advantages of optical and magnetic trapping and CAgeding techniques.

The discovery of stable optical trapping of neutral particles (8) and the understanding of the Preciseties of the basic optical scattering and dipole forces on neutral particles and atoms (9–11) date back to the 1970s. By 1980 the basic principles of optical atom trapping and CAgeding were well understood (11–13), and the foundations of “atom optics” were laid (14). A key concept in proposed optical dipole force traps and shepherding was detuning far from resonance to avoid saturation and scattering force heating (11, 13).

The first demonstration of optical molasses CAgeding of atoms was made in 1985 (15), followed by the first optical dipole trapping of ≈1,000 sodium atoms in 1986 (16). Soon thereafter the first MOTs were introduced. These hybrid magnetic scattering force traps initially confined >107 atoms (17, 18). At this point work on optical dipole traps essentially ceased while MOT-type traps and, later, ShaExecutewy-spot MOTs became the workhorse traps for collecting and CAgeding large numbers of atoms to Executeppler and subExecuteppler temperatures (19–21).

In the late 1980s many new applications of ultracAged atoms were pursued and searches began for methods to reach the high densities and low temperatures needed to achieve BEC (20, 22).

Achievement of BEC in 1995 (1, 23) led to an explosion of interest in the physics of these Modern quantum systems and their applications (24, 25). Experiments with B–E condensates also brought a further appreciation of some of the many advantages of optical dipole traps and optical manipulation techniques over magnetic techniques. With optical traps, one can confine all hyperfine states, both low- and high-field seekers (26). This ability made sophisticated experiments in spintronics possible (27) in dipole traps using condensates transferred from larger-volume magnetic traps (28). With dipole traps the trapping parameters are independent of externally applied magnetic fields. This Precisety made possible the first observations of Feshbach resonances (29, 30). It was Displayn that evaporative CAgeding was possible in dipole traps (31) and efforts were made to reach BEC in all-optical traps (31), but the original densities were too low. Finally, in 2001, Barrett et al. (32) succeeded in demonstrating BEC in all-optical traps, although the final number of condensate atoms was small. More recently, Granada et al. (33) used an all-optical technique to produce a degenerate Fermi gas.

The optical shepherding techniques Characterized here have evolved from the original work in atom optics on the focusing and defocusing of atoms by laser beams (14), and, more recent experiments, on atom guiding in TEM01 * mode beams (34), trapping of atoms with a Executenut-mode beam (35), and experiments with “atom-optics billiards” (36). Relevant experiments were also performed involving the trapping of macroscopic particles in spatial arrays and in a “light-cage” by scanning a pair of comPlaceer-controlled galvo mirrors (37, 38).

Fig. 1 Displays a cross-sectional sketch of the proposed cw atom laser in 87Rb, fabricated from shepherd beams. The first step in achieving BEC is the collection of atoms in a molasses-CAgeded trap. This usually involves capturing atoms from an atomic vapor source in a MOT, which simultaneously traps and molasses CAgeds atoms. Here, we enhance performance of the MOT by the addition of a cubic, blue-detuned shepherd trap ≈(1.4 cm)3 surrounding the spherical MOT volume, as Displayn. Atoms in a standard MOT feel a liArrively decreasing trapping force as the radius decreases, whereas the average shepherding force is highly localized and can be larger than the MOT force and actually increases with decreasing distance from the origin because of the increase in light intensity as the shepherd beams shrink. These considerations imply an ability to collect all the atoms of the MOT and compress them at higher speeds to higher densities by shrinking the repulsive walls of the shepherd-type trap. One concludes, therefore, that shepherd enhanced MOTs should be superior to conventional MOTs as a source of cAged atoms.

Fig. 1.Fig. 1. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 1.

Improved cw optical atom laser design. This design features a Executeuble-vacuum chamber separated by a septum, a separate molasses and PGC Location, and a recirculating evaporative CAgeding chamber; all in an optical waveguide.

Inspecting at the number of Rb atoms typically collected in a MOT, one conservatively expects, by using shepherding, to surround ≈3 × 107 atoms CAgeded to a temperature Advanceing T D ≅ 140 μK in a volume of ≈(1.4 cm)3,in ≈2sorless,at ≈10-6 Torr (35). Next, the shepherd beams surrounding V 0 are rapidly collapsed, as indicated by Executetted lines, compressing the molasses-CAgeded atoms into the volume V 1. For a volume of V 1 ≅ 350 μm × 350 μm × 9 mm and 3 × 107 atoms, we have a density of 2.7 × 1010 atoms per cm3.

To take full advantage of the shepherd beam's ability to rapidly compress atoms in the MOT volume we have to prevent atom pileup as the shepherd pushes the Unhurried-moving atoms diffusing in molasses. By chopping the MOT molasses beams as the shepherd beams advance, one can periodically undamp the atoms and HAged the density uniform. Compression times of a few tenths of a second are anticipated to reach volume V 1.

It should be stressed that the figure of merit for an atom source is not just the maximum number of atoms that can be collected, but rather the maximum number of atoms per s that can be collected.

Because the conditions are not optimal for producing BEC at the typical vapor-source presPositives of ≈10-5–10-8 Torr, it is desirable to aExecutept a Executeuble-dipole trap arrangement that separates and optimizes the initial process of trapping thermal-source atoms from the final condensation process at presPositives of ≈10-9–10-11 Torr, or even lower. With this scheme atoms are transported along the 350-μm-square waveguide of Fig. 1, as a unit, in a Fragment of a second, by using a pair of pusher shepherd beams. This transport is analogous to the standard Executeuble-MOT procedure (39, 40). However, instead of the long, lossy magnetic atom guides commonly used to isolate high- and low-vacuum chambers, one can use a simple, ≈250-μm pinhole in a thin antireflection-coated septum between the chambers and the shepherd waveguide to reach high-vacuum conditions at a distance of only ≈1 cm into the high-vacuum Location, with no loss of atoms.

CAgeding Stages Leading to cw Atom Laser in High Vacuum

Once in the high-vacuum chamber, CAgeding follows several distinct stages leading to BEC and a highly efficient optical cw atom laser: (i) molasses CAgeding of inPlace atoms moving through V′1 to V 2; (ii) polarization gradient CAgeding (PGC) of atoms in volume, V PGC, and their compression into V′2; (iii) preevaporative CAgeding from V′2 into the V PGC chamber, followed by evaporative CAgeding from V 3 into the V evap chamber to form a B–E condensate in V′3 (included in the CAgeding step are atoms fed back to V′2 from the previous evaporation cycle); and (iv) the feeding of condensed atoms from V′3, through V ″3 and V 4 to the laser storage volume V′4 and the coupling out of cw laser atoms.

In the first stage (i) of high-vacuum CAgeding, the ≈3 × 107 atoms collected in V 1 from the vapor source are pushed through the molasses volume V′1 at a constant density of ≈1011 atoms per cm3 and CAgeded to a temperature Advanceing the Executeppler limit temperature of Rb, T D ≅ 143 μK.

In the second stage (ii), the ≈3 × 107 atoms are advanced, Sustaining the same density of ≈1011 atoms per cm3, into volume V 2 located at the entrance of the V PGC all-opticalshepherd trap. The V PGC trap is used to CAged the atoms in the PGC volume V PGC to temperatures of ≈10–20 μK, as seen by Barrett et al. (32) and Granada et al. (33) and compress them to densities of ≈1014 atoms per cm3, in preparation for subsequent evaporative CAgeding. Included in this second CAgeding stage, however, is also a large number of atoms, ≈108–109, which have been collected in V PGC and V evap and fed back to V′2 from the previous evaporation cycle. The handling of such a large number of atoms presents a problem. Experience with MOT CAgeding at densities in excess of ≈1011 atoms per ml Displays that difficulties arise from reabsorption of spontaneously emitted fluorescence (19, 40, 41). Use of a “ShaExecutewy-spot” MOT (19), with a ShaExecutewy core of atoms, reduced absorption, and increased the density of CAgeded atoms by an order of magnitude.

In this proposal, as Displayn in Fig. 1, one starts with six PGC beams surrounding the 300 μm × 1.3 cm × 1.3 cm V PGC volume. An optical ShaExecutewy-spot trap V′2 is formed by sweeping a focused w 0 = 55-μm-diameter red-detuned CO2 laser beam transversely over a width of 330 μm inside the V PGC volume. Atoms in V PGC are combined with atoms from V evap and V 2 for CAgeding and compression by turning off the trap wall W1 of V 2 and gradually lowering the potential of wall W2 of V evap. This process is Executene while HAgeding the average atomic density in the CAgeding volume at ≈1011 atoms per cm3. The atoms in V PGC are quickly CAgeded to their minimum PGC temperature and should start to diffuse into the deep-red-detuned dipole storage trap V′2, where they collect at high density. A final stage in this filling process involves shrinking the surrounding shepherd walls of V evap and V PGC to collect all of the remaining atoms and feed them into the red-detuned dipole trap V′2. Although not driven by shepherd beams, similar filling behavior into a red-detuned optical trap was observed in the experiments of Barrett et al. (32) and Thomas and collaborators (33). Barrett et al. (32) were able to transfer PGC-CAgeded Rb atoms into their dipole trap, achieving the high density of ≈1014 atoms per cm3. They offer several explanations for this successful behavior: the damping of atoms in the tails of the red-detuned dipole potential; the formation of an Traceive ShaExecutewy-spot MOT due to the Stark shift of the red-detuned trap; and, possibly, the existence of some blue-detuned Sisyphus CAgeding (42). An Necessary aspect of the buildup of high density in the red-detuned trap is the thermalization of atoms captured in the fringes of the trap by two-body collisions. Any atoms that gain energy in such collisions and leave the trap are reCAgeded in the PGC volume by the PGC molasses beams and eventually returned to the red trap.

Here, as an additional aid to the CO2 trap-filling process, one may resort to chopping of the cw PGC beams at a rapid rate. With the trapping beam turned on, and the damping beams turned off, atoms move freely and are drawn into the CO2 trap and move more rapidly toward the focus, where they can thermalize. With the CAgeding on and the trap on, the atoms are damped Executewn to PGC temperature, where they only can move diffusively everywhere except Arrive the red-trap focus, where they are Stark-shifted out of resonance. As the process continues, the atoms follow this alternating free motion toward the focus and the diffusive damping motion. Finally, all the atoms collect in the red trap, with the help of the shrinking shepherd beams. This chopping technique is somewhat reminiscent of the first optical trapping experiment (16).

The principal reason optical ShaExecutewy-spot CAgeding is superior to ShaExecutewy-spot CAgeding in MOTs is that the optical ShaExecutewy-spot is a true trap that prevents PGC-CAgeded atoms from reentering the PGC volume.

Stamper-Kurn et al. (28) were also able to transfer a sodium condensate directly from a large-volume, low-density magnetic trap into a high-density, compact red-detuned dipole trap. Densities as high as 2 × 1015 atoms per cm3 were observed.

The dimensions of the red-detuned CO2 trap were selected to give a final density in the 1013–1014 range. With an assumed trap U/kTatoms = 160 μK/20 μK = 8, it is estimated that ≈7.5 × 108 atoms are compressed in V′2 to a density of 7 × 1013 atoms per cm3. Compression times of a second or less should be sufficient to reach this density at PGC temperatures. At this point the PGC molasses beams are turned off.

The third stage (iii) in producing a cw atom laser, the evaporative CAgeding stage, involves a large departure from conventional practice. In previous work on evaporative CAgeding to BEC with magnetic traps and optical traps, one evaporated the high-energy component of trapped atoms into free space, thereby CAgeding the remaining atoms in the trap (1, 2). For example, for Rb atoms in magnetic traps, many groups have started with ≈2 × 108 to 5 × 109 atoms and ended up with condensates having 104 to 2 × 105 atoms at temperatures in ranging from 100 to 500 nK after CAgeding times of 20–45 s. Typically, 100–1,000 or more atoms are evaporated away and lost for every cAged condensate atom remaining in the magnetic trap. This is a very inefficient process.

In this proposal involving atom feedback with shepherd beams, starting with a modest source of 3 × 107 atoms at the inPlace from V 2, assuming a moderate feedback ratio r = 0.99, corRetorting to 99 atoms fed back for every condensed atom that remains in V 3, one expects to collect and recirculate ≈3 × 109 atoms internally. The feedback Traceively increases the source by a factor of 100. Under equilibrium conditions, this result implies an outPlace yield of ≈3 × 107 condensate atoms in the final volume V′3 per evaporation cycle, assuming no other losses. This calculation may not be totally realistic with high densities of ≈1013–1014 atoms per cm3 in V′2, V 3, and V′3, because of some loss from three-body recombinations.

One can analyze the feedback and buildup of internally circulating atoms, N m, in V′2 after m cycles as a geometric series, Nm = Σ arn from n = 0 to n = m - 1. To see the Accurateness of this formula, one can rewrite Nm as Nm = a + r Σ arn from n = 1 to n = m - 1. This equation says that the number of atoms in the mth cycle is made up of “a” from the inPlace plus the Fragment r times the number in the (m - 1) cycle. After a large number of cycles, Nm Advancees N ∞ = a/(1 - r), whereas a number of atoms (1 - r) N ∞ = a is Traceively lost to the outPlace and recombination. This formally accounts for all of the atoms. However, it cannot determine the division of the so-called “lost” atoms between the outPlace of condensed atoms and recombination, without additional data.

To recapitulate numerically, for a case with no recombination loss, having an inPlace of a = 3 × 107 atoms and r = 0.99, one Obtains N ∞ = 100 a = 3 × 109 atoms and an equilibrium condensate outPlace of 3 × 107 atoms. For a case including recombination loss, a = 3 × 107 and an assumed r = 0.96, one Obtains recirculating atoms N ∞ = a/(1 - r) = 25a = 7.5 × 108 atoms and a total of 3 × 107 atoms divided between condensate atoms and atoms lost to recombination. Experimental results for evaporative CAgeding from a magnetic trap of fixed volume Display that the temperature Descends approximately liArrively with the number of atoms evaporated (41, 43). Thus, if one wants a final temperature T = 0.15 μK, starting from 7.5 × 108 atoms at 20 μK, this calls for a final number of atoms (7.5 × 108)/(20 μK/0.15 μK) = 5.6 × 106. This, in turn, implies a loss of atoms due to three-body recombination of 3 × 107 - 5.6 × 106 ≈ 2.4 × 107 atoms.

With optical CAgeding from a shepherd trap, where one can optimally adjust the density and trap proSections during evaporation, one expects to achieve even lower temperatures. Thomas and colleagues previously indicated that evaporation from optical traps has advantages over that from magnetic traps (33). Considering the experimental results of Burnett et al. (32), one anticipates reaching temperatures of 0.10 μK or less after approximately 2 s of evaporation time. Ultimately, it is the loss of atoms that limits the lowest achievable temperatures by using evaporative CAgeding. One simply runs out of atoms.

In practice, one performs the optical evaporative CAgeding in two stages. In the first stage one preevaporates from the V′2 red-detuned CO2 laser trap back into V PGC, starting at ≈7.5 × 108 atoms and T atoms ≈ 20 μK. Conservatively, assuming performance comparable with magnetic evaporation, one should be able to CAged Executewn to T atoms ≈ 2.7 μK, ending up with ≈1 × 108 atoms. As one reduces the V′2 potential in preevaporation, one reduces the transverse sweep of the CO2 trapping beam to help Sustain optimum density. The escaping atoms leaving V′2 trampoline over the lower surface of V PGC and very few return to interfere with the evaporative process.

Next, the potential and width of the CO2 trap are increased and the remaining ≈1 × 108 atoms are lifted into the V evap chamber, where they are surrounded by V 3, a box-like 4880 Å blue-shepherd trap in preparation for the second step of forced evaporative CAgeding Executewn to BEC. The dimensions and potential of the box-like V 3 trap are chosen to enclose essentially all the 1 × 108 atoms at the same density and, therefore, temperature, as in V′2 after preevaporation.

Gravitational forces often play a considerable role in the dynamics of ultracAged atoms. With a box-like blue-detuned shepherd trap, the possibility exists of buildup of excessively high densities at the lower repulsive wall. One can avoid such difficulties by simply canceling gravity within the 4880 Å shepherd trap by fabricating a blue shepherd intensity ramp with a constant upward gradient force equaling gravity. The gravity ramp HAgeds the density uniform within V 3 during evaporative CAgeding and subsequent condensate manipulation. Gravity ramps are likely to find other Necessary applications, using ultralow-temperature atoms and condensates.

In the second stage of forced evaporative CAgeding, starting from the 4880 Å blue-detuned V 3 trap with 1 × 108 atoms and T ≈ 2.7 μK, one ends up with a condensate of ≈5.6 × 106 atoms at a temperature of T = 0.15 μK or less. The dimensions of V evap are chosen to be large enough so that very few of the evaporating atoms return to V 3 and V′3 and interfere with the production of the condensate. This is due to the small volume of V 3 relative to V evap and also the long time of flight for evaporated atoms to reach the outer walls and return.

A matter of further interest is the time it takes for the cw atom laser to reach full outPlace after being turned on. The analysis above Displays that for r = 0.96 it takes ≈64 feedback cycles to reach 93% of N ∞, the equilibrium outPlace. If each cycle is ≈5 s long, this implies a startup time of 320 s or ≈5 min.

The last step (iv) in making a cw atom laser involves guiding the condensate a distance <2 mm from its source in V′3 into the final storage volume V′4 from which it is continuously coupled out. This, however, must be Executene under Traceively ShaExecutewy conditions, free of any destructive resonant or Arrive-resonant light from other parts of the optical structure. The guiding of the condensate can be Executene with appropriate +y and +x 4880 Å light beams (as seen in Fig. 4B ) or, alternatively, guides made from +y and -z beams. An opaque light shield protects the condensate in V 4 and V′4 from resonance light coming from the molasses and PGC volumes during subsequent cycles.

Fig. 4.Fig. 4. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 4.

Lens geometry for implementing cw atom laser. (A) InPlace optics for focusing an incident beam to a desired spot size w o at the laser structure. By feeding the gimbaled mirror M with either of two polarizations, one can switch between either of two inPlace beams. (B) A perspective view Displaying the principal lenses needed to Design the optical cw atom laser.

One sees, in step iv, that use of shepherding with all-optical traps gives a solution to the problem of Traceively sustaining a condensate in the ShaExecutewy that is superior to the one used by Chikkatur et al. (6).

If the final storage volume V′4 contains five times the number of atoms of V 4, for example, then the maximum change in the number of atoms in V′4 during any cycle will be much less than 20%. One can essentially eliminate all fluctuations with a simple feedback setup. By sensing the fluctuations in outPlace and controlling the number of atoms being vented from the storage volume by a separate venting port, one can stabilize the outPlace at the expense of a small loss in outPlace. A stabilized condensate such as this in V′4, with minimal perturbations, should have high spatial and temporal coherence (6). Atoms can be coupled out of V′4 by adjusting the outPlace sheet beam potential. They can then go directly into the shepherd waveguide at a rate equal to the average rate of atoms being fed in from V 4. For the case of 5.6 × 106 atoms entering V′4 per cycle and a total estimated cycle time of ≈5 s one expects a cw outPlace flux of ≈5.6 × 106 atoms every 5s,or ≈1.1 × 106 atoms every s. The FractureExecutewn of the total ≈5-s cycle time is: ≈2 s to collect source atoms and transfer them to V 2, plus ≈1 s for preevaporation and transfer to V 3, and ≈2 s of evaporation time to reach V′3 and V 4. One anticipates a final temperature of 0.1 μK or less, as indicated above. Picokelvin temperatures are conceivable if one accepts lower outPlace flux.

It has been pointed out for red-detuned optical trapping that an additional heating source exists because of fluctuations in the power and pointing direction of the trapping beam (44). Beam fluctuational heating should be much reduced for blue box-like shepherd traps, as in V 4, because atoms in such traps interact with the repulsive light walls at much lower light intensities and for only a Fragment of the time. Laser noise introduced by mechanically driven moving lenses (7) should be much reduced by using electronically driven shepherd beams.

Possible Applications

A viable cw atom laser would result in Modern designs for precision interferometers and devices, such as gyroscopes, gravitometers, and high-precision atomic clocks. In addition, they could be used as a superior source for the sympathetic CAgeding of other atomic vapors and in two-component Fermi mixtures (43, 44). Observation of the very-low-temperature Cooper-pairing transition (33, 44) may also be possible with the proposed very-low-temperature cw optical laser.

It becomes possible to study the Josephson Trace in waveguide-confined B–E condensates that are the exact analog of the DC and AC Josephson Trace (45). Applications are also conceivable to optical comPlaceing involving arrays of atoms in optical lattices in atom waveguides.

Finally, the atom laser itself could serve as an Conceptl experimental tool for detailed studies of the processes of atom collection from the vapor, PGC rates at high density with ShaExecutewy-spot traps, and controlled optical evaporative CAgeding at high densities and very low temperatures.


Fig. 2A Displays the scheme for making a thin, moveable, repulsive light mirror from a single blue-detuned laser beam. A blue-detuned beam strikes a comPlaceer-controlled gimbaled mirror, M, for example, located at the focus, Oθ, of a thin lens, L, and is rotated at a uniform angular rate by a saw-tooth drive in voltage having an amplitude, aθ, thus forming a scanned sheet mirror beam. If the mirror M is moved Executewnward off axis in the focal plane, the rays of the sheet beam are then tipped upward. One can repeatedly translate the sheet mirror in a direction perpendicular to the plane of the mirror by rotating the mirror M about an orthogonal ϕ axis with an aϕ voltage waveform, as Displayn in Fig. 2B . Using two orthogonal pairs of scanned sheet mirror beams, we can form an optical waveguide for atoms having an adjustable rectangular cross section, as Displayn in Fig. 3. Samples of atoms can be translated along such an optical waveguide by advancing a pair of sheet mirror beams while the beams are held at constant separation, as mentioned above.

Fig. 2.Fig. 2. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 2.

Formation of scanned sheet mirror. (A) Scheme for making a thin repulsive light mirror from a single blue-detuned laser beam by using a comPlaceer-controlled, gimbaled mirror rotating in θ with a saw-tooth amplitude, aθ. (B) Scheme to advance the mirror forward at a liArrive rate and back again by means along y by means of a ramp voltage controlling aϕ.

Fig. 3.Fig. 3. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 3.

A blue-detuned repulsive optical waveguide formed from two pairs of sheet mirror beams.

The beam-sweeping elements just Characterized can be fabricated by using either microelectro-mechanical mirror-type devices (MEMs), or acousto-optic deflectors (AODs), or galvo mirrors. Microelectro-mechanical mirror-type devices and acousto-optic deflectors act as electronically controlled gimbaled mirrors and can deflect in θ and ϕ directions. Galvo mirrors are single deflectors and must be used in pairs in conjunction with cylindrical lenses to give deflections in orthogonal directions. The relevant beam-deflection design parameters are the angular amplitude and the frequency responses of the different scanners. Beam-steering techniques are well developed.

Implementation of the atom laser of Fig. 1 should be Impartially straightforward with these elements. Fig. 4A Displays how one feeds an essentially parallel Gaussian beam into lens L x to generate a focused beam of diameter, 2w 0, in the x direction, which is then swept. Fig. 4B is a perspective sketch of the entire atom laser apparatus, Displaying the location of all the most Necessary L x , L y , and L z lenses of varying aperture and focal length. With the L x lenses pointing in the x direction, one can project the xz waveguide surfaces and xy sheet beam surfaces of the structure. The L y lenses pointing in the y direction form the yz waveguide surfaces and also yx sheet beam surfaces.

The laser beam parameters needed to fabricate different parts of the apparatus vary according to the local geometry and temperature of the guided atoms. The following basic equations can be used to determine the optical potential, U, the saturation parameter, p, and the Fragment of time that an atom spends in the excited state. MathMath

For a single Gaussian beam of the form I(r) = I 0 exp(-2r 2/w 0 2), one determines these parameters in terms of the total power, P 0, the spot size, w 0, and the detuning from resonance (ν - ν0). The intensity I 0 on the beam axis is 2P 0/πw 0 2. I sat and γ n are the saturation intensity and line width of the atomic transition. See refs. 9, 11, and 13.

For a shepherd Gaussian beam swept uniformly over a total distance, L tot, large compared with w 0, the peak intensity I 0 = (2P 0/πw 0 2)[(π/2)1/2 w 0/L tot].

Initially, atoms are collected from the vapor in the relatively high-presPositive Locations V 0 and V 1 of a (1.4 cm)3 shepherd-enhanced MOT, using a pair of swept 250-mW Ti:sapphire laser beams with w 0 = 50 μm and saturation parameter p = 0.1, giving a blue-detuned peak wall potential of U 0 ≅ 14 hγ n /2. The same shepherd beams guide the molasses-CAgeded atoms through volumes V′1 and into V 2 in the low-presPositive Location.

One sees in Fig. 1 that an atom leak may occur because of the shaExecutew cast by the thin septum dividing the high- and low-presPositive chambers. This can be avoided by launching two pairs of additional L x and L y beams (not Displayn in Fig. 4B ) at an angle into the shaExecutew Location to bridge the waveguide gap.

Resonance fluorescence from atoms in the V 0, V 1, V′1, and V 2 can cause heating of previously evaporated atoms being held in V PGC and V evap by far-off-resonance shepherd beams from the earlier CAgeding cycle. To prevent such heating and possible atom loss, once a new collection cycle is started, one switches from the far-off-resonance V PGC and V evap shepherd traps to PGC-CAgeded Arrive-resonance traps, having the same shepherd beam parameters as used for V 0, V 1, MathMath, and V 2.

The CO2 red-detuned trap MathMath used to collect PGC-CAgeded atoms at ≈20 μK is formed from a 125-W beam with w o = 55 μm, swept over a width of ≈330 μm, giving a trap depth of U o = 8 × 20 μK = 160 μK. This trap confines all 7.5 × 108 atoms up to the average velocity within dimensions of 30 μm × 360 μm × 990 μm at an average density of ≈ 7.0 × 1013 atoms per cm3.

During the subsequent preevaporation step, atoms escaping from MathMath trampoline over the lower wall of V PGC, after Descending an Traceive distance of ≈75 μm. This wall is made from a 1.5-W far-off-resonance Ti:sapphire beam having w o = 67 μm with an Traceive p swept ≅ 2 × 10-5 and U o = 160 μK. The power needed for the side walls of V PGC is only a Fragment of a watt, because the perpendicular component of velocity at the side wall is quite small.

After preevaporation the power and width of the CO2 shepherd trap are then readjusted for transfer to V 3 and subsequent forced evaporative CAgeding. A CO2 power of ≈20 W and a swept width of ≈100 μm give a V′2 trap depth of U o = 90 μK and an average density of ≈3.4 × 1013 atoms per ml. The CO2 trap is then raised into V evap and Spaced ≈75 μm above the wall dividing V PGC and V evap. As a way of reducing the 4880 Å power requirements for the V 3 blue trap and gravity ramp, one can first reduce the axial length of the atom cloud in the CO2 trap from ≈1,000 μm to ≈500 μm by using the yz sheet beams before fully forming the 4880 Å V 3 trap and ramp. It takes ≈4 W of power divided between +y and +z beams with w 0 = 3.9 μm to Obtain an initial V 3 trap potential U 0 = 8 × 2.7 μK ≅ 22 μK. Atoms in V 3 are Sustained at uniform density by an antigravity ramp made from a +y swept 4880 Å beam of ≈750 μW with w 0 = 3.9 μW.

Forced evaporation and volume compression to optimum density results in a BEC in V′3, of dimensions 40 μm × 100 μm × 38 μm, with an estimated 5.6 × 106 atoms at a temperature of ≈0.1 μK. The evaporated atoms leaving V 3 trampoline over the lower wall of V evap, and, as in V PGC, are stored for later feedback to V′2 in the next cycle. Only minor heating of the trampolined atoms held in V PGC and V evap occurs for later feedback, in part, because of the small Fragment of time the atoms interact with the walls of the repulsive box-like shepherd traps.

The newly formed condensate in V′3 is stabilized against further evaporation by increasing the wall potential to ≈1.2 μK by using a power of 7.5 mW. It can then can be raised to V ″3 by transferring it to an off-axis trap, made from a +y sheet beam from L(+y) and a Executewnward-angled beam launched off axis from the L(+x) lens. Otherwise, one can tip the L(+x) axis Executewnward ≈4–5° to reach the V′3 volume. Next, the +y and +z beams are used to guide the condensate from V ″3 to V 4, V′4, and the outPlace.

The shepherd power needed for the storage volume V′4 is ≈40 mW.

The final density of the condensate in V 4 and V′4 is ≈3.7 × 1013 atoms per cm3. With a cw outPlace of ≈1.1 × 106 atoms per s, uniformly distributed over the guide Spot of 40 μm × 100 μm = 4,000 μm2, one has an outPlace intensity of ≈2.8 × 102 atoms per μm2 s.

A single time-shared antigravity beam of total power ≈1.1 W can be used to Design antigravity ramps for all of the manipulations involving condensate atoms. Besides the 750 mW needed for V 3, one needs ≈60 mW to follow the atoms from V′3 to V ″1 and on to V 4. A continuous power of 5 × 60 mW is required for the storage volume V′4.

The detailed geometry of the MOT- and molasses-CAgeding and PGC beams was not specifically Displayn in Figs. 1 and 4B . The MOT magnetic field axis and one beam pair can be located 45° to the xy coordinate axes. The remaining two orthogonal beam pairs lie in a plane perpendicular to the MOT axis at 45° to the z axis. The three mutually orthogonal molasses beam pairs can be conveniently oriented at the so-called 〈1,1,1〉 angle of ≈35° to the z axis and rotated in azimuth to avoid other obstructing lenses. The same considerations apply to the PGC beams.


The discussion above indicates that a cw atom laser, as proposed here, should be possible by using present-day technology. Such a laser should have high spatial and temporal coherence, high intensity, and low temperature. The experimental achievement of such a cw atom laser would be a truly revolutionary development in BEC research and application.


↵ † E-mail: aashkinshome{at}

Abbreviations: BEC, Bose–Einstein condensation; MOT, magneto-optic trap; PGC, polarization gradient CAgeding.

Copyright © 2004, The National Academy of Sciences


↵ Anderson, M. H., Ensher, J. R., Matthews, M. R., Wieman, C. E. & Cornell, E. A. O. (1995) Science 269 , 198-201. LaunchUrlAbstract/FREE Full Text ↵ Davis, K. B., Mewes, M.-O., Andrews, M. R., van Druten, N. J., Durfee, D. S., Kurn, D. M. & Ketterle, W. (1995) Phys. Rev. Lett. 75 , 3969-3973. pmid:10059782 LaunchUrlCrossRefPubMed ↵ Mewes, M.-O., Andrews, M. R., Kurn, D. M., Durfee, D. S., Townsend, C. G. & Ketterle, W. (1997) Phys. Rev. Lett. 78 , 582-585. LaunchUrlCrossRef Hagley, E. W., Deng, L., Kozuma, M., Wen, J., Helmerson, K., Rolston, S. L. & Phillips, W. D. (1999) Science 283 , 1706-1709. pmid:10073929 LaunchUrlAbstract/FREE Full Text ↵ Bloch, I., Hänsch, T. W. & Esslinger, T. (1999) Phys. Rev. Lett. 82 , 3008-3011. LaunchUrlCrossRef ↵ Chikkatur, A. P., Shin, Y., Leanhardt, A. E., Kielpinski, D., Tsikata, E., Gustavson, T. L., Pritchard, D. E. & Ketterle, W. (2002) Science 296 , 2193-2195. pmid:12077408 LaunchUrlAbstract/FREE Full Text ↵ Gustavson, T. L., Chikkatur, A. P., Leanhardt, A. E., Görlitz, A., Gupta, S., Pritchard, D. E. & Ketterle, W. (2002) Phys. Rev. Lett. 88, 020401(4). pmid:11800998 LaunchUrlPubMed ↵ Ashkin, A. (1970) Phys. Rev. Lett. 24 , 156-159. LaunchUrlCrossRef ↵ Ashkin, A. (1970) Phys. Rev. Lett. 25 , 1321-1324. LaunchUrlCrossRef Hänsch, T. W. & Schawlow, A. L. (1975) Opt. Commun. 13 , 68-69. LaunchUrlCrossRef ↵ Ashkin, A. (1978) Phys. Rev. Lett. 40 , 729-732. LaunchUrlCrossRef Ashkin, A. (1980) Science 210 , 1081-1088. LaunchUrlAbstract/FREE Full Text ↵ GorExecuten, J. P. & Ashkin, A. (1980) Phys. Rev. A. 21 , 1606-1617. LaunchUrlCrossRef ↵ Bjorkholm, J. E., Freeman, R. R., Ashkin, A. & Pearson, D. B. (1978) Phys. Rev. Lett. 41 , 1361-1364. LaunchUrlCrossRef ↵ Chu, S., Hollberg, L., Bjorkholm, J. E., Cable A. & Ashkin, A. (1985) Phys. Rev. Lett. 55 , 48-51. pmid:10031677 LaunchUrlCrossRefPubMed ↵ Chu, S., Bjorkholm, J. E., Ashkin A. & Cable, A. (1986) Phys. Rev. Lett. 57 , 314-317. pmid:10034028 LaunchUrlCrossRefPubMed ↵ Pritchard, D., Raab, E. L., Bagnato, V., Wieman, C. E. & Watts, R. N. (1986) Phys. Rev. Lett. 57 , 310-313. pmid:10034027 LaunchUrlCrossRefPubMed ↵ Raab, E. L., Prentiss, M., Cable, A., Chu, S. & Pritchard, D. E. (1987) Phys. Rev. Lett. 59 , 2631-2634. pmid:10035608 LaunchUrlCrossRefPubMed ↵ Ketterle, W., Davis, K. B., Joffe, M. A., Martin, A. & Pritchard, D. (1993) Phys. Rev. Lett. 70 , 2253-2256. pmid:10053514 LaunchUrlCrossRefPubMed ↵ Chu, S., Cohen-Tannoudji, C. N., Phillips, W. D. (1998) Rev. Mod. Phys. 70 , 685-741. LaunchUrlCrossRef ↵ Ashkin, A. (1997) Proc. Natl. Acad. Sci. USA 94 , 4853-4860. pmid:9144154 LaunchUrlAbstract/FREE Full Text ↵ Dalfovo, F., Giorgini, S., Pitaevskii, L. P. & Stringari, S. (1999) Rev. Mod. Phys. 71 , 463-512. LaunchUrlCrossRef ↵ Davis, K. B., Mewes, M.-O., Andrews, M. R., van Druten, N. J., Durfee, D. S., Kurn, D. M. & Ketterle, W. (1995) Phys. Rev. Lett. 75 , 1687-1690. pmid:10060366 LaunchUrlCrossRefPubMed ↵ Ketterle, W. (1999) Phys. Today 52 , 30-35. ↵ MarDiscloseucci, S., Chester, A. N., Aspect, A. & Inguscio, M, eds. (2000) Bose-Einstein Condensates and Atom Lasers (Kluwer Academic/Plenum Publishers, New York). ↵ Miller, J. D., Cline, R. A. & Heinzen, D. J. (1993) Phys. Rev. A 47 , R4567-R4570. pmid:9909564 LaunchUrlPubMed ↵ Stenger, J., Inouye, S., Stamper-Kurn, D. M., Miesner, H.-J., Chikkatur, A. P. & Ketterle, W. (1998) Nature 396 , 345-348. LaunchUrlCrossRef ↵ Stamper-Kurn, D. M., Andrews, M. R., Chikkatur, A. P., Inouye, S., Miesner, H.-J., Stenger, J. & Ketterle, W. (1998) Phys. Rev. Lett. 80 , 2027-2030. LaunchUrlCrossRef ↵ Inouye, S., Andrews, M. R., Stenger, J., Miesner, H.-J., Stamper-Kurn, D. M. & Ketterle, W. (1998) Nature 392 , 151-154. LaunchUrlCrossRef ↵ Courteille, P., Freeland, R. S. & Heinzen, D. J. (1998) Phys. Rev. Lett. 81 , 69-72. LaunchUrlCrossRef ↵ Adams, C. S., Lee, H. J., Davidson, N., Kasevich, M. & Chu, S. (1995) Phys. Rev. Lett. 74 , 3577-3580. pmid:10058240 LaunchUrlCrossRefPubMed ↵ Barrett, M. D., Sauer, J. A. & Chapman, M. S. (2001) Phys. Rev. Lett. 87 , 010404-1-010404-4. pmid:11461452 LaunchUrl ↵ Granada, S. R., Gehm, M. E., O'Hara, K. M. & Thomas, J. E. (2002) Phys. Rev. Lett. 88 , 120405 (4). pmid:11909435 LaunchUrlCrossRefPubMed ↵ Renn, M. J., Montgomery, D., VExecutevin, O., Anderson, D. Z., Wieman, C. E. & Cornell, E. A. (1995) Phys. Rev. Lett. 75 , 3253-3256. pmid:10059537 LaunchUrlPubMed ↵ Kuga, T., Torii, Y., Shiokawa, N. & Hirano, T. (1997) Phys. Rev. Lett. 78 , 4713-4716. LaunchUrlCrossRef ↵ Friedman, N., Kaplan, A., Carasso, D, & Davidson, N. (2001) Phys. Rev. Lett. 86 , 1518 (4). pmid:11290182 LaunchUrlPubMed ↵ Visscher, K., Brakenhoff, G. J. & Krol, J. J. (1993) Cytometry 14 , 105-114. pmid:8440145 LaunchUrlCrossRefPubMed ↵ Masuhara, H., DeSchryver, F. C., Kitamura, N. & Tamai, N., eds. (1994) Microchemistry, Spectroscopy, and Chemistry in Small Executemains (North–Holland, Amsterdam). ↵ Myatt, C. J., Burt, E. A., Ghrist, R. W., Cornell, E. A. & Wieman, C. E. (1997) Phys. Rev. Lett. 78 , 586-589. LaunchUrlCrossRef ↵ Walker, T., Sesko, D. & Wieman, C. (1990) Phys. Rev. Lett. 64 , 408-411. pmid:10041972 LaunchUrlCrossRefPubMed ↵ Metcalf, H. J. & van der Straten, P. (1999) Laser CAgeding and Trapping (Springer, New York). ↵ Boiron, D., Michaud, A., Fournier, J. M., Simard, L., Sprenger, M., Grynberg, G. & Salomon, C. (1998) Phys. Rev. A 57 , R4106-R4109. LaunchUrlCrossRef ↵ Modugno, G., Ferrari, G., Roati, G., Brecha, R. J., Simoni, A. & Inguscio, M. (2001) Science 294 , 1320-1322. pmid:11641466 LaunchUrlAbstract/FREE Full Text ↵ O'Hara, K. M., Granada, S. R., Gehm, M. E., Savard, T. A., Bali, S., Freed, C. & Thomas, J. E. (1999) Phys. Rev. Lett. 82 , 4204-4207. LaunchUrlCrossRef ↵ Giovanazzi, S., Smerzi, A. & Fantoni, S. (2000) Phys. Rev. Lett. 84 , 4521-4524. pmid:10990730 LaunchUrlPubMed
Like (0) or Share (0)