AltSpanisch Lied

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/46648662
Laser Cooling and Trapping
Article in Journal of the Optical Society of America B · May 2003

DOI: 10.1364/JOSAB.20.000887 · Source: OAI
CITATIONS READS
632 17,074
2 authors, including:
Peter van der Straten

Utrecht University
129 PUBLICATIONS 4,879 CITATIONS
SEE PROFILE
All content following this page was uploaded by Peter van der Straten on 20 May 2014.
The user has requested enhancement of the downloaded file.

1
Laser Cooling and Trapping of Neutral Atoms
Q1 H. J. Metcalf•
Q2
Department of Physics, State University of New York, Stony Brook, N.Y. 11794-3800, USA•.
Q3
e-mail:• [email protected]
Q4
Peter van der Straten

Debye Institute, Department of Atomic and Interface Physics, Utrecht University, 3508 TA
Utrecht, The Netherlands. e-mail: [email protected]
Abstract
This article presents a review of some of the principal techniques of laser cooling and
trapping that have been developed during the past 20 years. Its approach is primarily
experimental, but its quantitative descriptions are consistent in notation with most of
the theoretical literature.
Keywords
laser cooling; atom trapping; optical lattice; Bose–Einstein condensation.
1 Introduction 3
1.1 Temperature and Entropy 3
1.2 Phase Space Density 4
2 Optical Forces on Neutral Atoms 4
2.1 Radiative Optical Forces 5
2.2 Dipole Optical Forces 5
2.3 Density Matrix Description of Optical Forces 6
2.3.1 Introduction 6
2.3.2 Open Systems and the Dissipative Force 6
2.3.3 Solution of the OBEs in Steady State 7
2.3.4 Radiative and Dipole Forces 8
2.3.5 Force on Moving Atoms 8
2 Laser Cooling and Trapping of Neutral Atoms
3 Laser Cooling 8
3.1 Slowing Atomic Beams 8
3.2 Optical Molasses 10
3.2.1 Doppler Cooling 10
3.2.2 Doppler Cooling Limit 11
3.2.3 Atomic Beam
Collimation – One-dimensional Optical Molasses – Beam Brightening 12
3.2.4 Experiments in Three-dimensional Optical Molasses 13
3.3 Cooling Below the Doppler Limit 14
3.3.2 Linear ⊥ Linear Polarization Gradient Cooling 14
3.3.3 Origin of the Damping Force 16
3.3.4 The Limits of Sisyphus Laser Cooling 17
4 Traps for Neutral Atoms 17
4.1 Dipole Force Optical Traps 18
4.1.1 Single-beam Optical Traps for Two-level Atoms 18
4.1.2 Blue-detuned Optical Traps 19
4.2 Magnetic Traps 20
4.2.2 Magnetic Confinement 20
4.2.3 Classical Motion of Atoms in a Quadrupole Trap 22
4.2.4 Quantum Motion in a Trap 23
4.3 Magneto-optical Traps 23
4.3.2 Cooling and Compressing Atoms in an MOT 25
4.3.3 Capturing Atoms in an MOT 25
4.3.4 Variations on the MOT Technique 26
5 Optical Lattices 26
5.1 Quantum States of Motion 26
5.2 Properties of 3-D Lattices 28
5.3 Spectroscopy in 3-D Lattices 29
5.4 Quantum Transport in Optical Lattices 30
6 Bose–Einstein Condensation 31
6.1 Introduction 31
6.2 Evaporative Cooling 31
6.2.1 Simple Model 32
6.2.2 Application of the Simple Model 33
6.2.3 Speed of Evaporation 34
6.2.4 Limiting Temperature 34
6.3 Forced Evaporative Cooling 35
7 Conclusion 36
References 36
OE005
Laser Cooling and Trapping of Neutral Atoms 3
1 assignment of a thermodynamic ‘‘temper-

Introduction ature’’ is completely inappropriate.
Nevertheless, it is convenient to use the
The combination of laser cooling and label of temperature to describe an atomic
atom trapping has produced astounding sample whose average kinetic energy Ek
new tools for atomic physicists [1]. These has been reduced by the laser light, and
experiments require the exchange of mo- this is written simply as kB T/2 = Ek ,
mentum between atoms and an optical where kB is the Boltzmann’s constant (for
field, usually at a nearly resonant fre- the case of one dimension, 1-D). It must
quency. The energy of light h̄ω changes be remembered that this temperature
the internal energy of the atom, and the assignment is absolutely inadequate for
angular momentum h̄ changes the orbital atomic samples that do not have a
angular momentum of the atom, as de- Maxwell–Boltzmann velocity distribution,
scribed by the well-known selection rule whether or not they are in thermal contact
= ±1. By contrast, the linear momen- with the environment; there are infinitely
tum of light p = h̄ω/c = h̄k cannot change many velocity distributions that have the
the internal atomic degrees of freedom, same value of Ek but are so different
and therefore must change the momen- from one another that characterizing them
tum of the atoms in the laboratory frame. by the same ‘‘temperature’’ is a severe
The force resulting from this momentum error. (In the special case where there
exchange between the light field and the is a true damping force, F ∝ −v, and
atoms can be used in many ways to control where the diffusion in momentum space
atomic motion, and is the subject of this is a constant independent of momentum,
article. solutions of the Fokker–Planck equation
can be found analytically and can lead
1.1 to a Maxwell–Boltzmann distribution that
Temperature and Entropy does indeed have a temperature.)
Since laser cooling decreases the tem-
The idea of ‘‘temperature’’ in laser cool- perature of a sample of atoms, there is
ing requires some careful discussion and less disorder and therefore less entropy.
disclaimers. In thermodynamics, temper- This seems to conflict with the second law
ature is carefully defined as a parameter of thermodynamics, which requires the
of the state of a closed system in thermal entropy of a closed system to always in-
equilibrium with its surroundings. This, crease with time. The explanation lies in
of course, requires that there be thermal the consideration of the fact that in laser
contact, that is, heat exchange, with the en- cooling, the atoms do not form a closed
vironment. In laser cooling, this is clearly system. Instead, there is always a flow of
not the case because a sample of atoms laser light with low entropy into the sys-
is always absorbing and scattering light. tem and fluorescence with high entropy
Furthermore, there is essentially no heat out of it. The decrease of entropy of the
exchange (the light cannot be considered atoms is accompanied by a much larger
as heat even though it is indeed a form increase in entropy of the light field. En-
of energy). Thus, the system may very tropy considerations for a laser beam are
well be in a steady state situation, but cer- far from trivial, but recently it has been
tainly not in thermal equilibrium, so the shown that the entropy lost by the atoms
OE005
is many orders of magnitude smaller than to use a force that is not conservative,
the entropy gained by the light field. such as a velocity-dependent force. In laser
cooling, the force on the atoms can be
1.2 a damping force, that is, always directed
Phase Space Density opposite to the atomic velocity, so that
the momentum part of ρφ increases. This
The phase space density ρ(r , p, t) is process arises from the irreversible nature
defined as the probability that a single of spontaneous emission.
particle is in a region dr around r and
has momentum dp around p at time t.
In classical mechanics, ρ(r , p, t) is just the 2
sum of the ρ(r , p, t) values of each of the N Optical Forces on Neutral Atoms
particles in the system divided by N. Since
the phase space density is a probability, it The usual form of electromagnetic forces is
is always positive and can be normalized given by F = q(E + v × B),
but for neutral
over the six-dimensional volume spanned atoms, q = 0. The next order of force is the
by position r and momentum p. For a gas dipole term, but this also vanishes because
of cold atoms, it is convenient to choose the neutral atoms have no inherent dipole
elementary volume for ρ(r , p, t) to be h̄3 , moment. However, a dipole moment can
so it becomes the dimensionless quantity be induced by a field, and this is most
efficient if the field is alternating near the
ρφ = nλ3deB , (1)
atomic resonance frequency. Since these
where λdeB is the deBroglie wavelength of frequencies are typically in the optical
the atoms in the sample and n is their range, dipole moments are efficiently
spatial density. induced by shining nearly resonant light
The Liouville theorem requires that ρφ on the atoms.
cannot be increased by using conservative If the light is absorbed, the atom
forces. For instance, in light optics one makes a transition to the excited state,
can focus a parallel beam of light with a and the subsequent return to the ground
lens to a small spot. However, that simply state can be either by spontaneous or by
produces a high density of light rays in the stimulated emission. The nature of the
focus in exchange for the momentum part optical force that arises from these two
of ρφ because the beam entering the lens different processes is quite different and
is parallel but the light rays are divergent will be described separately.
at the focus. The spontaneous emission case is
For classical particles, the same principle different from the familiar quantum-
applies. By increasing the strength of the mechanical calculations using state vec-
trapping potential of particles in a trap, one tors to describe the system. Spontaneous
can increase the density of the atoms in the emission causes the state of the sys-
trap but at the same time, the compression tem to evolve from a pure state into
of the sample results in a temperature a mixed state and so the density ma-
increase, leaving the phase space density trix is needed to describe it. Sponta-
unchanged. neous emission is an essential ingredient
In order to increase the phase space for the dissipative nature of the optical
density of an atomic sample, it is necessary forces.
OE005
2.1 the atoms). The force is thus velocity-

Radiative Optical Forces dependent and the experimenters’ task is
to exploit this dependence to the desired
In the simplest case–the absorption of goal, for example, optical friction for laser
well-directed light from a laser beam–the cooling.
momentum exchange between the light The maximum attainable deceleration
field and the atoms results in a force is obtained for high intensities of light.
High-intensity light can produce faster ab-
dp
F = = h̄kγp , (2) sorption, but it also causes equally fast
dt stimulated emission; the combination pro-
where γp is the excitation rate of the atoms. duces neither deceleration nor cooling.
The absorption leaves the atoms in their The momentum transfer to the atoms by
excited state, and if the light intensity stimulated emission is in the opposite di-
is low enough that they are much more rection to what it was in absorption, result-
likely to return to the ground state by ing in a net transfer of zero momentum•.
Q5
spontaneous emission than by stimulated At high intensity, Eq. (3) shows saturation
emission, the resulting fluorescent light of γp at γ /2, and since the force is given
carries off momentum h̄k in a random by Eq. (2), the deceleration saturates at a
/2M.
value amax = h̄kγ
direction. The momentum exchange from
the fluorescence averages zero, so the net
total force is given by Eq. (2). 2.2
The excitation rate γp depends on the Dipole Optical Forces
laser detuning from atomic resonance δ ≡
ωl − ωa , where ωl is the laser frequency While detuning •|δ| γ , spontaneous
Q6
and ωa is the atomic resonance frequency. emission may be much less frequent than
This detuning is measured in the atomic stimulated emission, unlike the case of the
rest frame, and it is necessary that the dissipative radiative force that is necessary
Doppler-shifted laser frequency in the rest for laser cooling, given by Eqs. (2) and (3).
frame of the moving atoms be used to In this case, absorption is most often fol-
calculate γp . In Sect. 2.3.3, we find that lowed by stimulated emission, and seems
to produce zero momentum transfer be-
γp for a two-level atom is given by the
cause the stimulated light has the same
Lorentzian
momentum as the exciting light. How-
s0 γ /2 ever, if the optical field has beams with at
γp = , (3)
1 + s0 + [2(δ + ωD )/γ ]2 least two different k-vectors present, such
as in counterpropagating beams, absorp-
where γ ≡ 1/τ is an angular frequency tion from one beam followed by stimulated
corresponding to the natural decay rate of emission into the other indeed produces a
the excited state. Here, s0 = I/Is is the ratio nonzero momentum exchange. The result
of the light intensity I to the saturation is called the dipole force, and is reversible
intensity Is ≡ πhc/3λ3 τ , which is a few and hence conservative, so it cannot be
mW cm−2 for typical atomic transitions. used for laser cooling.
The Doppler shift seen by the moving The dipole force is more easily calculated
atoms is ωD = −k · v (note that k opposite from an energy picture than from a
to v produces a positive Doppler shift for momentum picture. The force then derives
OE005
from the gradient of the potential of an intensity Imax at the antinodes is four times
atom in an inhomogeneous light field, that of the single traveling wave.
which is appropriate because the force is
conservative. The potential arises from the 2.3
shift of the atomic energy levels in the Density Matrix Description of Optical
light field, appropriately called the ‘‘light Forces
shift”, and is found by direct solution of the
Schrödinger equation for a two-level atom 2.3.1 Introduction
in a monochromatic plane wave. After Use of the density matrix ρ for pure states
making both the dipole and rotating wave provides an alternative description to the
approximations, the Hamiltonian can be more familiar one that uses wave functions
written as and operators but adds nothing new. Its
equation of motion is i h̄(dρ/dt) = [H, ρ],
h̄ −2δ
H= ∗ (4) and can be derived directly from the
2 0 Schrödinger equation. Moreover, it is a
√ straightforward exercise to show that the
where the Rabi frequency is | | = γ s0 /2
expectation value of any operator A that
for a single traveling laser beam. Solution
represents an observable is A = tr (ρ A).
of Eq. (4) for its eigenvalues provides
Application of the Ehrenfest theorem
the dressed state energies that are light-
gives the expectation value of the force as
shifted by
F = −tr (ρ∇ H). Beginning with the two-
level atom Hamiltonian of Eq. (4), we find
| |2 + δ 2 − δ
the force in 1-D to be
ωls = . (5)
2 ∂ ∂ ∗
∗
F = h̄ ρ + ρeg . (7)
For sufficiently large detuning |δ| | |, ∂z eg ∂z
approximation of Eq. (5) leads to ωls ≈
Thus, F depends only on the off-diagonal
| |2 /4δ = γ 2 s0 /8δ. ∗ , terms that are called
elements ρeg = ρge
In a standing wave in 1-D with |δ| | |,
the optical coherences.
the light shift ωls varies sinusoidally from
node to antinode. When δ is sufficiently
large, the spontaneous emission rate 2.3.2 Open Systems and the Dissipative
may be negligible compared with that of Force
stimulated emission, so that h̄ωls may be The real value of the density matrix
treated as a potential U. The resulting formalism for atom-light interactions is
dipole force is its ability to deal with open systems. By
not including the fluorescent light that is
2 lost from an atom-laser system undergoing
= − h̄γ ∇I,
F = −∇U (6)
8δIs cooling, a serious omission is being made
in the discussion above. That is, the closed
where I is the total intensity distribution system of atom plus laser light that can be
of the standing-wave light field of period described by Schrödinger wave functions
λ/2. For such a standing wave, the optical and is thus in a pure state, undergoes
electric field (and the Rabi frequency) at the evolution to a ‘‘mixed’’ state by virtue of
antinodes is double that of each traveling the spontaneous emission. This omission
wave that composes it, and so the total can be rectified by simple ad hoc additions
OE005
to the equation of motion, and the result is γρee . These equations have to be solved in
called the optical Bloch equations (OBE). order to evaluate the optical force on the
These are written explicitly as atoms.
dρgg i
= +γρee + ( ∗ ρ̃eg − ρ̃ge ) 2.3.3 Solution of the OBEs in Steady State
dt 2
In most cases, the laser light is applied
dρee i
= −γρee + ( ρ̃ge − ∗ ρ̃eg ) for a period long compared to the typical
dt 2 evolution times of atom-light interaction,
dρ̃ge γ i ∗ that is, the lifetime of the excited state τ =
=− + iδ ρ̃ge + (ρee − ρgg )
dt 2 2 1/γ . Thus, only the steady state solution of
dρ̃eg γ i the OBEs have to be considered, and these
=− − iδ ρ̃eg + (ρgg − ρee ), are found by setting the time derivatives in
dt 2 2
(8) Eq. (8) to zero. Then the probability ρee to
where ρ̃eg ≡ ρeg e−iδt for the coherences. be in the excited state is found to be
In these equations, the terms propor- γp s0 /2 s/2
tional to the spontaneous decay rate γ have ρee = = = ,
γ 1 + s0 + (2δ/γ )2 1+s
been put in ‘‘by hand”, that is, they have (9)
been introduced into the OBEs to account where s ≡ s0 /[1 + (2δ/γ )2 ] is the off-
for the effects of spontaneous emission. resonance saturation parameter. The
The spontaneous emission is irreversible excited-state population ρee increases lin-
and accounts for the dissipation of the early with the saturation parameter s for
cooling process. For the ground state, the small values of s, but for s of the order of
decay of the excited state leads to an in- unity, the probability starts to saturate to a
crease of its population ρgg proportional to value of 1/2. The detuning dependence of
γρee , whereas for the excited state, it leads γp (see Eq. 3) showing this saturation for
to a decrease of ρee , also proportional to various values of s0 is depicted in Fig. 1.
0.50
s0 = 100.0
s0 = 10.0
0.40 s0 = 1.0
Scattering rate gp (g)
s0 = 0.1
0.30
0.20
0.10
0.00
–10 –5 0 5 10
Detuning d (g)
Fig. 1 Excitation rate γp as a function of the detuning δ for several
values of the saturation parameter s0 . Note that for s0 > 1, the line
profiles start to broaden substantially from power broadening
OE005
2.3.4 Radiative and Dipole Forces atomic velocities that are small compared
Some insight into these forces emerges with γ /k, the motion can be treated as a
by expressing the gradient of the Rabi small perturbation in the atomic evolution
frequency of Eq. (7) in terms of a real that occurs on the time scale 1/γ . Then the
and imaginary part so that (∂ /∂z) = first-order result is given by
(qr + iqi ) . Then Eq. (7) becomes
d ∂ ∂ ∂
∗ ∗ = +v = + v(qr + iqi ) .
F = h̄qr ( ρeg + ρeg ) dt ∂t ∂z ∂t
(12)
∗ ∗
+ i h̄qi ( ρeg − ρeg ) (10) For the case of atoms moving in a standing
wave, this results in the same damping
Thus, the first term of the force is related force as Eq. (13) below.
to the dispersive part of the atom-light
interaction, whereas the second term is
related to the absorptive part of the atom- 3
light interaction. Laser Cooling
To appreciate the utility of the separation
of ∇ into real and imaginary parts, 3.1
consider the interaction of atoms with a Slowing Atomic Beams
traveling plane wave E(z) = E0 (ei(kz−ωt) +
c.c.)/2. In this case, qr = 0 and qi = k, and Among the earliest laser cooling exper-
so the force is caused only by absorption. iments was the deceleration of atoms
The force is given by Fsp = h̄kγρee and is in a beam [2]. The authors exploited the
the radiative force of Eqs. (2) and (3). Doppler shift to make the momentum
For the case of counterpropagating plane exchange (hence the force) velocity de-
waves, there is a standing wave whose pendent. It worked by directing a laser
electric field is E(z) = E0 cos(kz)(e−iωt + beam opposite an atomic beam so that the
c.c.). Thus, qr = −k tan(kz) and qi = 0, so atoms could absorb light, and hence mo-
there is only the dispersive part of the mentum h̄k, very many times along their
force, given by paths through the apparatus as shown in
Fig. 2 [2, 3]. Of course, excited-state atoms
2 h̄kδs0 sin 2kz cannot absorb light efficiently from the
Fdip = . (11)
1 + 4s0 cos2 kz + (2δ/γ )2 laser that excited them, so between ab-
sorptions they must return to the ground
This replaces Eq. (6) for the dipole force state by spontaneous decay, accompanied
and removes the restriction |δ| | |, by the emission of fluorescent light. The
thereby including saturation effects. Even spatial symmetry of the emitted fluores-
though the average of this force over a cence results in an average of zero net
wavelength vanishes, it can be used to momentum transfer from many such fluo-
trap atoms in a region smaller than the rescence events. Thus, the net force on the
wavelength of the light. atoms is in the direction of the laser beam,
and the maximum deceleration is limited
2.3.5 Force on Moving Atoms by the spontaneous emission rate γ .
In order to show how these forces can be Since the maximum deceleration a max =
used to cool atoms, one has to consider h̄kγ /2M is fixed by atomic parameters, it is
the force on moving atoms. For the case of straightforward to calculate the minimum
OE005
Extraction Laser 1
Profile magnets
magnet
Bias
PMT
Sodium magnet
80-MHz
source AOM l/4-plate
Laser 2
C.M. 80-MHz
AOM
4 cm
Pump Probe
beam beam
zs zc zp
40 cm 125 cm 25 cm 40 cm
Fig. 2Schematic diagram of apparatus for beam slowing. The tapered magnetic field is
produced by layers of varying length on the solenoid
stopping length Lmin and timetmin for Tab. 1 Parameters of interest for slowing
the rms velocity of atoms v̄ = 2 kB T/M various atoms. The stopping length Lmin and
time tmin are minimum values. The oven
at the source temperature. The result is temperature Toven that determines the peak
Lmin = v2 /2amax and tmin = v/amax . In velocity is chosen to give a vapor pressure of
Table 1 are some of the parameters for 1 Torr. Special cases are H at 1000 K for
slowing a few atomic species of interest dissociation of H2 into atoms, and He in the
from the peak of the thermal velocity metastable triplet state, for which two rows are
shown: one for a 4-K source and another for the
distribution. typical discharge temperature
Maximizing the scattering rate γp re-
quires δ = −ωD in Eq. (3). If δ is chosen Atom Toven v Lmin tmin
for a particular atomic velocity in the (K) (m s− 1) (m) (ms)
beam, then as the atoms slow down, their
changing Doppler shift will take them H 1000 5000 0.012 0.005
out of resonance. They will eventually He* 4 158 0.03 0.34
cease deceleration after their Doppler shift He* 650 2013 4.4 4.4
Li 1017 2051 1.15 1.12
has been decreased by a few times √ the Na 712 876 0.42 0.96
power-broadened width γ = γ 1 + s0 , K 617 626 0.77 2.45
corresponding to v of a few times Rb 568 402 0.75 3.72
vc = γ /k. Although this v of a few Cs 544 319 0.93 5.82
m s−1 is considerably larger than the typ-
ical atomic recoil velocity vr = h̄k/M of a
few cm s−1 , it is still only a small frac- In order to achieve deceleration that
tion of the average thermal velocity v of changes the atomic speeds by hundreds
the atoms, such that significant further of m s−1 , it is necessary to maintain
cooling or deceleration cannot be accom- (δ + ωD ) γ by compensating such large
plished. changes of the Doppler shift. This can be
OE005
done by changing ωD through the angular t = 10 − 50 µs with an acoustic optical

dependence of k · v, or changing δ either modulator (AOM). A pulse of atoms in
via ωl or ωa . The two most common the selected hfs passes the pump region
methods for maintaining this resonance and travels to the probe beam. The time
are sweeping the laser frequency ωl dependence of the fluorescence induced
along with the changing ωD of the by the probe laser, tuned to excite the
decelerating atoms [4–6], or by spatially selected hfs, gives the time of arrival, and
varying the atomic resonance frequency this signal is readily converted to a velocity
with an inhomogeneous d.c magnetic distribution. Figure 3 shows the measured
field to keep the decelerating atoms velocity distribution of the atoms slowed
in resonance with the fixed frequency by Laser 2.
laser [2, 3, 7].
The use of a spatially varying magnetic 3.2
field to tune the atomic levels along the Optical Molasses
beam path was the first method to succeed
in slowing atoms [2, 3]. It works as long 3.2.1 Doppler Cooling
as the Zeeman shifts of the ground and A different kind of radiative force arises
excited states are different so that the in low intensity, counterpropagating light
resonant frequency is shifted. The field beams that form a weak standing wave. It
can be tailored to provide the appropriate is straightforward to calculate the radiative
Doppler shift along the moving atom’s force on atoms moving in such a standing
path. A solenoid that can produce such wave using Eq. (3). In the low intensity
a spatially varying field has layers of case where stimulated emission is not
decreasing lengths. The technical problem important, the forces from the two light
of extracting the beam of slow atoms from beams are simply added to give F OM =
the end of the solenoid can be simplified by
reversing the field gradient and choosing a
transition whose frequency decreases with 6
increasing field [9]. Instrumental resolution
Atom flux (arb. units)
For alkali atoms such as sodium, a

4
time-of-flight (TOF) method can be used
to measure the velocity distribution of 2.97 m s–1
atoms in the beam [8]. It employs two 2 (FWHM)
additional beams labeled pump and probe

from Laser 1 as shown in Fig. 2. Because
these beams cross the atomic beam at 90◦ , 0
ωD = −k · v = 0, and they excite atoms at 125 130 135 140 145 150 155
all velocities. The pump beam is tuned Velocity (m s–1)
to excite and empty a selected ground Fig. 3 The velocity distribution measured with
hyperfine state (hfs), and it transfers the TOF method. The experimental width of
more than 98% of the population as the approximately 16 (γ /k) is shown by the dashed
atoms pass through its 0.5 mm width. vertical lines between the arrows. The Gaussian
fit through the data yields an FWHM• of
To measure the velocity distribution of 2.97 m s−1 (figure taken from Molenaar, P. A.,
Q7
atoms in the selected hfs, this pump vander Straten, P., Heideman, H. G. M.,
laser beam is interrupted for a period of Metcalf, H. (1997), Phys. Rev. A 55, 605–614)
OE005
F + + F − , where F ± are found from Eqs. (2) laser cooling and related aspects of optical
and (3). Then the sum of the two forces is control of atomic motion, the forces arise
because of the exchange of momentum
8 h̄k2 δs0 v between the atoms and the laser field.
F OM ∼
= ≡ −βv,
γ (1 + s0 + (2δ/γ )2 )2 These necessarily discrete steps of size h̄k
(13) constitute a heating mechanism that must
where terms of order (kv/γ )4 and higher be considered.
have been neglected. The force is pro- Since the atomic momentum changes
portional to velocity for small enough by h̄k, their kinetic energy changes on
velocities, resulting in viscous damping an average by at least the recoil energy
for δ < 0 [10, 11] that gives this technique Er = h̄2 k2 /2M = h̄ωr . This means that the
the name ‘‘optical molasses’’ (OM). average frequency of each absorption is
These forces are plotted in Fig. 4. For at least ωabs = ωa + ωr . Similarly, the en-
δ < 0, this force opposes the velocity ergy h̄ωa available from each spontaneous
and therefore viscously damps the atomic decay must be shared between the out-
motion.√The force F OM has maxima near going light and the kinetic energy of
v ≈ ±γ s0 + 1/2k and decreases rapidly the atom recoiling with momentum h̄k.
for larger velocities. Thus, the average frequency of each emis-
sion is ωemit = ωa − ωr . Therefore, the
3.2.2 Doppler Cooling Limit light field loses an average energy of
If there were no other influence on the h̄(ωabs − ωemit ) = 2 h̄ωr for each scatter-
atomic motion, all atoms would quickly ing event. This loss occurs at a rate of 2γp
decelerate to v = 0 and the sample would (two beams), and the energy is converted
reach T = 0, a clearly unphysical result. In to atomic kinetic energy because the atoms
0.4
0.2
Force (hkg)
0.0
– 0.2
– 0.4
–4 –2 0 2 4
Velocity (g /k)
Fig. 4 Velocity dependence of the optical damping forces for 1-D
optical molasses. The two dotted traces show the force from each
beam, and the solid curve is their sum. The straight line shows
how this force mimics a pure damping force over a restricted
velocity range. These are calculated for s0 = 2 and δ = −γ , so
there is some power broadening evident
OE005
recoil from each event. The atomic sample and step frequency 2γp , where the factor
is thereby heated because these recoils are of 2 arises because of the two beams. The
in random directions. random walk results in an evolution of
The competition between this heating the momentum distribution as described
and the damping force of Eq. (13) results by the Fokker–Planck equation, and can
in a nonzero kinetic energy in steady state, be used for a more formal treatment of
where the rates of heating and cooling are laser cooling. It results in diffusion in
equal. Equating the cooling rate, F OM · v, to momentum space with diffusion coeffi-
the heating rate, 4 h̄ωr γp , we find the steady cient D0 ≡ 2(p)2 /t = 4γp ( h̄k)2 . Then
state kinetic energy to be ( h̄γ /8)(2|δ|/γ + the steady state temperature is given by
γ /2|δ|). This result is dependent on kB T = D0 /β. This turns out to be h̄γ /2 as
|δ|, and has a minimum at 2|δ|/γ = 1, above for the case s0 1 when δ = −γ /2.
whence δ = −γ /2. The temperature found This remarkable result predicts that the
from the kinetic energy is then TD = final temperature of atoms in OM is inde-
h̄γ /2kB , where TD is called the Doppler pendent of the optical wavelength, atomic
temperature or the Doppler cooling limit. mass, and laser intensity (as long as it is
For ordinary atomic transitions, TD is not too large).
typically below 1 mK.
Another instructive way to determine 3.2.3 Atomic Beam
TD is to note that the average momentum Collimation – One-dimensional Optical
transfer of many spontaneous emissions Molasses – Beam Brightening
is zero, but the rms scatter of these When an atomic beam crosses a 1-D OM
about zero is finite. One can imagine as shown in Fig. 5, the transverse motion
these decays as causing a random walk of the atoms is quickly damped while
in momentum space, similar to Brownian the longitudinal component is essentially
motion in real space, with step size h̄k unchanged. This transverse cooling of
Optical Optical
molasses molasses
Oven
Very bright
atomic beam
Laser or
magnetic lens
Fig. 5 Scheme for optical brightening of an atomic beam. First,

the transverse velocity components of the atoms are damped out
by an optical molasses, then the atoms are focused to a spot,
and finally the atoms are recollimated in a second optical
molasses (figure taken from Sheehy, B., Shang, S. Q., vander
Straten, P., Metcalf, H. (1990), Chem. Phys. 145, 317–325)
OE005
an atomic beam is an example of a wander away from the center experience

method that can actually increase its no force directing them back. They are
brightness (atoms/s-sr-cm2 ) because such allowed to diffuse freely and even escape,
active collimation uses dissipative forces as long as there is enough time for
to compress the phase space volume their very slow diffusive movement to
occupied by the atoms. By contrast, the allow them to reach the edge of the
usual focusing or collimation techniques region of intersection of the laser beams.
for light beams and most particle beams Since the atomic velocities are randomized
is restricted to selection by apertures or during the damping time M/β = 2/ωr ,
conservative forces that preserve the phase atoms execute a random walk in position
space density of atoms in the beam. space√ with a step size of 2vD /ωr =
This velocity compression at low in- λ/(π 2ε) ∼ = few µm. To diffuse a distance
tensity in one dimension can be easily of 1 cm requires about 107 steps or about
estimated for two-level
√ atoms to be about 30 s [13, 14].
√
vc /vD = γ /ωr ≡ 1/ε. Here vD is the In 1985, the group at Bell Labs was the
velocity associated with the Doppler limit first to observe 3-D OM [11]. Preliminary
laser cooling discussed above: vD =
for measurements of the average kinetic en-
h̄γ /2M. For Rb, vD = 12 cm s−1 , vc = ergy of the atoms were done by blinking
γ /k 4.6 m s−1 , ωr 2π × 3.8 kHz, and off the laser beams for a fixed interval.
1/ε 1600. (The parameter ε charac- Comparison of the brightness of the fluo-
terizes optical forces on atoms.) In- rescence before and after the turnoff was
cluding two transverse directions along used to calculate the fraction of atoms that
with the longitudinal slowing and cool- left the region while it was in the dark.
ing discussed above, the decrease in The dependence of this fraction on the
three-dimensional 3-D phase space vol- duration of the dark interval was used to es-
ume for laser cooling of an Rb atomic timate the velocity distribution and hence
beam from the momentum contribution the temperature. This method, which is
alone •can exceed 106 . Clearly optical usually referred to as release and recapture,
Q8
techniques can create atomic beams enor- is specifically designed to measure the tem-
mously more times intense than ordinary perature of the atoms, since the usual way
thermal beams and also many orders of of measuring temperatures cannot be ap-
magnitude brighter. plied to an atomic cloud of a few million
atoms. The result was consistent with TD
3.2.4 Experiments in Three-dimensional as calculated from the Doppler theory, as
Optical Molasses described in Sect. 3.2.2.
By using three intersecting orthogonal Later a more sensitive ballistic technique
pairs of oppositely directed beams, the was devised at NIST that showed the
movement of atoms in the intersection astounding result that the temperature of
region can be severely restricted in all 3-D, the atoms in OM was very much lower than
and many atoms can thereby be collected TD [15]. These experiments also found that
and cooled in a small volume. OM was less sensitive to perturbations
Even though atoms can be collected and more tolerant of alignment errors
and cooled in the intersection region, it than was predicted by Doppler theory.
is important to stress that this is not a For example, if the intensities of the two
trap (see Sect. 4 below), that is, atoms that counterpropagating laser beams forming
OE005
an OM were unequal, then the force steady state have ground-state orientations
on the atoms at rest would not vanish, caused by optical pumping processes that
but the force on the atoms with some distribute the populations over the differ-
nonzero drift velocity would vanish. This ent ground-state sublevels. In the presence
drift velocity can be easily calculated by of polarization gradients, these orienta-
using unequal intensities s0+ and s0− , tions reflect the local light field. In the
to derive an analog of Eq. (13). Thus, low-light-intensity regime, the orientation
atoms would drift out of an OM, and the of stationary atoms is completely deter-
calculated rate would be much faster than mined by the ground-state distribution;
observed by deliberately unbalancing the the optical coherences and the excited-
beams in the experiments [16]. state population follow the ground-state
distribution adiabatically.
3.3 For atoms moving in a light field
Cooling Below the Doppler Limit that varies in space, optical pumping
acts to adjust the atomic orientation
3.3.1 Introduction to the changing conditions of the light
It was an enormous surprise to observe field. In a weak pumping process, the
that the ballistically measured temperature orientation of moving atoms always lags
of the Na atoms was as much as 10 behind the orientation that would exist for
times lower than TD = 240 µK [15], the stationary atoms. It is this phenomenon of
temperature minimum calculated from nonadiabatic following that is the essential
theory. This breaching of the Doppler limit feature of the new cooling process.
forced the development of an entirely new Production of spatially dependent optical
picture of OM that accounts for the fact pumping processes can be achieved in
that in 3-D, a two-level picture of atomic several different ways. As an example,
structure is inadequate. The multilevel consider two counterpropagating laser
structure of atomic states, and optical beams that have orthogonal polarizations,
pumping among these sublevels, must be as discussed below. The superposition of
considered in the description of 3-D OM. the two beams results in a light field
In response to these surprising mea- having a polarization that varies on the
surements of temperatures below TD , two wavelength scale along the direction of the
groups developed a model of laser cooling laser beams. Laser cooling by such a light
that could explain the lower tempera- field is called polarization gradient cooling.
tures [17, 18]. The key feature of this model In a 3-D OM, the transverse wave character
that distinguishes it from the earlier pic- of light requires that the light field always
ture is the inclusion of the multiplicity of has polarization gradients.
sublevels that make up an atomic state
(e.g., Zeeman and hfs). The dynamics 3.3.2 Linear ⊥ Linear Polarization
of optically pumping the moving atoms Gradient Cooling
among these sublevels provides the new One of the most instructive models for
mechanism for producing ultralow tem- discussion of sub-Doppler laser cooling
peratures [19]. was introduced in Ref. [17] and very well
The dominant feature of these models described in Ref. [19]. If the polarizations
is the nonadiabatic response of moving of two counterpropagating laser beams
atoms to the light field. Atoms at rest in a are identical, the two beams interfere
OE005
and produce a standing wave. When px

the two beams have orthogonal linear
s−
polarizations (same frequency ωl ) with →
l/2 k2
their Ê vectors perpendicular (e.g., x̂ and
→
ŷ), the configuration is called lin ⊥ lin k1 0
s+
or lin-perp-lin. Then the total field is the
py
sum of the two counterpropagating beams
given by 0 l/4 l/2
= E0 x̂ cos(ωl t − kz) + E0 ŷ cos(ωl t + kz) Fig. 6 Polarization gradient field for the lin ⊥ lin
E
configuration
= E0 [(x̂ + ŷ) cos ωl t cos kz
+ (x̂ − ŷ) sin ωl t sin kz]. (14) Since the coupling of the different states
of multilevel atoms to the light field de-
At the origin, where z = 0, this becomes pends on its polarization, atoms moving
in a polarization gradient will be coupled
= E0 (x̂ + ŷ) cos ωl t,
E (15) differently at different positions, and this
will have important consequences for laser
which corresponds to linearly polarized cooling. For the Jg = 1/2 → Je = 3/2 tran-
light at an angle +π/4 to√the x-axis. The sition (one of the simplest transitions that
amplitude of this field is 2E0 . Similarly, show sub-Doppler cooling [20]), the opti-
for z = λ/4, where kz = π/2, the field is cal pumping process in purely σ + light
also linearly polarized but at an angle −π/4 drives the ground-state population to the
to the x-axis. Mg = +1/2 sublevel. This optical pump-
Between these two points, at z = λ/8, ing occurs because absorption always
where kz = π/4, the total field is produces M = +1 transitions, whereas
the subsequent spontaneous emission pro-
= E0 x̂ sin ωl t + π
E
4 duces M = ±1, 0. Thus, the average
π M ≥ 0 for each scattering event. For σ − -
− ŷ cos ωl t + . (16) light, the population is pumped toward the
4
Mg = −1/2 sublevel. Thus, atoms trav-
Since the x̂ and ŷ components have eling through only a half wavelength in
sine and cosine temporal dependence, the light field, need to readjust their pop-
they are π/2 out of phase, and so ulation completely from Mg = +1/2 to
Eq. (16) represents circularly polarized Mg = −1/2 and back again.
light rotating about the z-axis in the The interaction between nearly resonant
negative sense. Similarly, at z = 3λ/8 light and atoms not only drives transitions
where kz = 3π/4, the polarization is between atomic energy levels but also
circular but in the positive sense. Thus, shifts their energies. This light shift of
in this lin ⊥ lin scheme the polarization the atomic energy levels, discussed in
cycles from linear to circular to orthogonal Sect. 2.2, plays a crucial role in this scheme
linear to opposite circular in the space of of sub-Doppler cooling, and the changing
only half a wavelength of light, as shown in polarization has a strong influence on the
Fig. 6. It truly has a very strong polarization light shifts. In the low-intensity limit of
gradient. two laser beams, each of intensity s0 Is , the
OE005
light shifts Eg of the ground magnetic

substates are given by a slight variation of
the approximation to Eq. (5) that accounts
for the multilevel structure of the atoms. 0
We write
Energy
M = +1/2
2 γ2
h̄s0 Cge
Eg = , (17)
8δ
M = −1/2
where Cge is the Clebsch–Gordan coeffi-
cient that describes the coupling between 0 λ /4 λ /2 3λ /4
Position (z)
the particular levels of the atom and the
light field. Fig. 7 The spatial dependence of the light shifts
of the ground-state sublevels of the
In the present case of orthogonal linear
J = 1/2 ⇔ 3/2 transition for the case of lin ⊥ lin
polarizations and J = 1/2 → 3/2, the light polarization configuration. The arrows show the
shift for the magnetic substate Mg = 1/2 path followed by atoms being cooled in this
is three times larger than that of the arrangement. Atoms starting at z = 0 in the
Mg = −1/2 substate when the light field is Mg = +1/2 sublevel must climb the potential
hill as they approach the z = λ/4 point where the
completely σ + . On the other hand, when
light becomes σ − polarized, and they are
an atom moves to a place where the light optically pumped to the Mg = −1/2 sublevel.
field is σ − , the shift of Mg = −1/2 is Then they must begin climbing another hill
three times larger. So, in this case, the toward the z = λ/2 point where the light is σ +
optical pumping discussed above causes polarized and they are optically pumped back to
the Mg = +1/2 sublevel. The process repeats
a larger population to be there •in the until the atomic kinetic energy is too small to
Q9
state with the larger light shift. This is climb the next hill. Each optical pumping event
generally true for any transition Jg → results in the absorption of light at a frequency
Je = Jg + 1. A schematic diagram showing lower than emission, thus dissipating energy to
the populations and light shifts for this the radiation field
particular case of negative detuning is
illustrated in Fig. 7. a position where the light field is σ − -
polarized, and are optically pumped to
3.3.3 Origin of the Damping Force Mg = −1/2, which is now lower than the
To discuss the origin of the cooling process Mg = 1/2 state. Again, the moving atoms
in this polarization gradient scheme, are at the bottom of a hill and start to
consider atoms with a velocity v at a climb. In climbing the hills, the kinetic
position where the light is σ + -polarized, as energy is converted to potential energy,
shown at the lower left of Fig. 7. The light and in the optical pumping process, the
optically pumps such atoms to the strongly potential energy is radiated away because
negative light-shifted Mg = +1/2 state. In the spontaneous emission is at a higher
moving through the light field, atoms must frequency than the absorption (see Fig. 7).
increase their potential energy (climb a Thus, atoms seem to be always climbing
hill) because the polarization of the light is hills and losing energy in the process.
changing and the state Mg = 1/2 becomes This process brings to mind a Greek
less strongly coupled to the light field. After myth, and is thus called ‘‘Sisyphus laser
traveling a distance λ/4, atoms arrive at cooling’’.
OE005
The cooling process described above is a few times Er ≡ kB Tr /2 = Mvr2 /2 is intu-

effective over a limited range of atomic itively comforting for two reasons. First,
velocities. The force is maximum for one might expect that the last spontaneous
atoms that undergo one optical pumping emission in a cooling process would leave
process while traveling over a distance λ/4. atoms with a residual momentum of the
Slower atoms will not reach the hilltop order of h̄k, since there is no control over
before the pumping process occurs and its direction. Thus, the randomness asso-
faster atoms will travel a longer distance ciated with this would put a lower limit
before being pumped toward the other of vmin ∼ vr on such cooling. Second, the
sublevel, so E/z is smaller. In both polarization gradient cooling mechanism
cases, the energy loss is smaller and described above requires that atoms be
therefore the cooling process less efficient. localizable within the scale of ∼ λ/2π in
Nevertheless, the damping constant β order to be subject to only a single polariza-
for this process is much larger than for tion in the spatially inhomogeneous light
Doppler cooling, and therefore the final field. The uncertainty principle then re-
steady state temperature is lower [17, 19]. quires that these atoms have a momentum
In the experiments of Ref. [21], the spread of at least h̄k.
temperature was measured in a 3-D The recoil limit discussed here has
molasses under various configurations been surpassed by evaporative cooling
of the polarization. Temperatures were of trapped atoms [22] and two different
measured by a ballistic technique, in which optical cooling methods, neither of which
the flight time of the released atoms was can be described in simple notions. One of
measured as they fell through a probe these uses optical pumping into a velocity-
a few centimeters below the molasses selective dark state [23]. The other one
region. The lowest temperature obtained uses carefully chosen, counterpropagating,
was 3 µK, which is a factor 40 below the laser pulses to induce velocity-selective
Doppler temperature and a factor 15 above Raman transitions, and is called Raman
the recoil temperature of Cs. cooling [24].
3.3.4 The Limits of Sisyphus Laser Cooling

The extension of the kind of thinking 4
about cooling limits in the case of Doppler Traps for Neutral Atoms
cooling to the case of the sub-Doppler
processes must be done with some care, In order to confine any object, it is
because a naive application of similar necessary to exchange kinetic for poten-
ideas would lead to an arbitrarily low tial energy in the trapping field, and in
final temperature. In the derivation in neutral atom traps, the potential energy
Sect. 3.2.2, it is explicitly assumed that must be stored as internal atomic energy.
each scattering event changes the atomic Thus, practical traps for ground-state neu-
momentum p by an amount that is a small tral atoms are necessarily very shallow
fraction of p and this clearly fails when compared with thermal energy because
the velocity is reduced to the region of the energy-level shifts that result from
vr ≡ h̄k/M. convenient size fields are typically con-
This limitation of the minimum steady siderably smaller than kB T for T = 1 K.
state value of the average kinetic energy to Neutral atom trapping therefore depends
OE005
on substantial cooling of a thermal atomic where w0 is the beam waist size. Such
sample, and is often connected with the a trap has a well-studied and important
cooling process. In most practical cases, macroscopic classical analog in a phe-
atoms are loaded from magneto-optical nomenon called optical tweezers [27–29].
traps (MOTs) in which they have been With the laser light tuned below reso-
efficiently accumulated and cooled to mK nance (δ < 0), the ground-state light shift
temperatures (see Sect. 4.3), or from op- is negative everywhere, but largest at
tical molasses, in which they have been the center of the Gaussian beam waist.
optically cooled to µK temperatures (see Ground-state atoms, therefore, experience
Sect. 3.2). a force attracting them toward this cen-
The small depth of typical neutral ter, given by the gradient of the light
atom traps dictates stringent vacuum shift, which is found from Eq. (5), and for
requirements because an atom cannot δ/γ s0 is given by Eq. (6). For the Gaus-
remain trapped after a collision with a sian beam, this transverse force at the waist
thermal energy background gas molecule. is harmonic for small r and is given by
Since these atoms are vulnerable targets
h̄γ 2 I0 r −2r 2 /w2
for thermal energy background gas, the F e 0. (19)
mean free time between collisions must 4δ Is w02
exceed the desired trapping time. The In the longitudinal direction, there is
cross section for destructive collisions is also an attractive force but it is more
quite large because even a gentle collision complicated and depends on the details
(i.e., large impact parameter) can impart of the focusing. Thus, this trap produces
enough energy to eject an atom from a an attractive force on the atoms in three
trap. At pressure P sufficiently low to be dimensions.
of practical interest, the trapping time is ∼ Although it may appear that the trap does
(10−8 /P) s, where P is in Torr. not confine atoms longitudinally because
of the radiation pressure along the laser
4.1 beam direction, careful choice of the laser
Dipole Force Optical Traps parameters can indeed produce trapping
in 3-D. This can be accomplished because
4.1.1 Single-beam Optical Traps for the radiation pressure force decreases as
Two-level Atoms
1/δ 2 (see Eqs. 2 and 3), but by contrast,
The simplest imaginable optical trap con- the light shift and hence the dipole force
sists of a single, strongly focused Gaussian decreases only as 1/δ for δ (see
laser beam (see Fig. 8) [25, 26] whose in- Eq. 5). If |δ| is chosen to be sufficiently
tensity at the focus varies transversely with large, atoms spend very little time in
r as the untrapped (actually repelled), excited
I(r) = I0 e−2r /w0 ,
2 2
(18) state because its population is proportional
to 1/δ 2 . Thus, a sufficiently large value
of |δ| produces longitudinal confinement
Laser and also maintains the atomic population
primarily in the trapped ground state.
2w0 The first optical trap was demonstrated
Fig. 8 A single focused laser beam produces in Na with light detuned below the D-
the simplest type of optical trap lines [26]. With 220 mW of dye laser light
OE005
tuned about 650 GHz below the atomic the optical frequency such as the Nd : YAG
transition and focused to an ∼10 µm laser trap [31] or the CO2 laser trap [32].
waist, the trap depth was about 15 h̄γ More important in some cases is that the
corresponding to 7 mK. trap relies on Stark shifting of the atomic
Single-beam dipole force traps can energy levels by an amount equal to the
be made with the light detuned by a trap depth, and this severely compromises
significant fraction of its frequency from the capabilities for precision spectroscopy
the atomic transition. Such a far-off- in a trap [33].
resonance trap (FORT) has been developed Attracting atoms to the region of low-
for Rb atoms using light detuned by nearly est intensity would ameliorate both these
10% to the red of the D1 transition at concerns, but such a trap requires positive
λ = 795 nm [30]. Between 0.5 and 1 W of detuning (blue), and an optical configu-
power was focused to a spot about 10 µm in ration having a dark central region. One
size, resulting in a trap 6 mK deep where of the first experimental efforts at a blue
the light-scattering rate was only a few detuned trap used the repulsive dipole
hundreds per second. The trap lifetime force to support Na atoms that were oth-
was more than half a second. erwise confined by gravity in an optical
There is a qualitative difference when the ‘‘cup’’ [34]. Two rather flat, parallel beams
trapping light is detuned by a large frac- detuned by 25% of the atomic resonance
tion of the optical frequency. In one such frequency were directed horizontally and
case, Nd : YAG light at λ = 1064 nm was oriented to form a V-shaped trough. Their
used to trap Na whose nearest transition Gaussian beam waists formed a region
is at λ = 596 nm [31]. In a more extreme 1 mm long where the potential was deep-
case, a trap using λ = 10.6 µm light from est, and hence provided confinement along
a CO2 laser has been used to trap Cs their propagation direction as shown in
whose optical transition is at a frequency Fig. 9. The beams were the λ = 514 nm
∼12 times higher (λ = 852 nm) [32]. For and λ = 488 nm from an argon laser, and
such large values of |δ|, calculations of
the trapping force cannot exploit the rotat-
0.6
ing wave approximations as was done for
Intensity
Eqs. (4) and (5), and the atomic behavior is 0.4
similar to that in a DC field. It is impor- 0.2

tant to remember that for an electrostatic 0
trap Earnshaw’s theorem precludes a field 1000
maximum, but that in this case there is in-
0
deed a local 3-D intensity maximum of the
y
focused light because it is not a static field. – 1000

50
0
– 50
x
4.1.2 Blue-detuned Optical Traps Fig. 9 The light intensity experienced by an
One of the principal disadvantages of the atom located in a plane 30 µm above the beam
optical traps discussed above is that the waists of two quasi-focused sheets of light
negative detuning attracts atoms to the traveling parallel and arranged to form a
V-shaped trough. The x and y dimensions are in
region of highest light intensity. This may µm (figure taken from Davidson, N., Lee, H. J.,
result in significant spontaneous emission Adams, C. S., Kasevich, M., Chu, S. (1995), Phys.
unless the detuning is a large fraction of Rev. Lett. 74, 1311–1314)
OE005
the choice of the two frequencies was not cooled Cs atoms to 3 µK at a density
simply to exploit the full power of the mul- of 3 × 1010 /cm3 in a sample whose 1/e
tiline Ar laser, but also to avoid the spatial height in the gravitational field was only
interference that would result from the use 19 µm. Simple ballistics gives a frequency
of a single frequency. of 450 bounces per second, and the 6-s
Obviously, a hollow laser beam would lifetime (limited only by background gas
also satisfy the requirement for a blue- collisions) corresponds to several thousand
detuned trap, but conventional textbook bounces. However, at such low energies,
wisdom shows that such a beam is not the deBroglie wavelength of the atoms is
an eigenmode of a laser resonator [35]. 1/4 µm, and the atomic motion is no
Some lasers can make hollow beams, longer accurately described classically, but
but these are illusions because they requires deBroglie wave methods.
consist of rapid oscillations between the
TEM01 and TEM10 modes of the cavity. 4.2
Nevertheless, Maxwell’s equations permit Magnetic Traps
the propagation of such beams, and
in the recent past there have been 4.2.1 Introduction
studies of the LaGuerre–Gaussian modes Magnetic trapping of neutral atoms is
that constitute them [36–38]. The several well suited for use in very many areas,
ways of generating such hollow beams including high-resolution spectroscopy,
have been tried by many experimental collision studies, Bose–Einstein conden-
groups and include phase and amplitude sation (BEC), and atom optics. Although
holograms, hollow waveguides, axicons or ion trapping, laser cooling of trapped ions,
related cylindrical prisms, stressing fibers, and trapped ion spectroscopy were known
and simply mixing the TEM01 and TEM10 for many years [42], it was only in 1985
modes with appropriate cylindrical lenses. that neutral atoms were first trapped [43].
An interesting experiment has been per- Such experiments offer the capability of
formed using the ideas of Sisyphus cooling the spectroscopic ideal of an isolated atom
(see Sect. 3.3.3•) with evanescent waves at rest, in the dark, available for interaction
Q10
combined with a hollow beam formed with electromagnetic field probes.
with an axicon [39]. In the previously re- Because trapping requires the exchange
ported experiments with atoms bouncing of kinetic energy for potential energy, the
under gravity from an evanescent wave atomic energy levels will necessarily shift
field [40, 41], they were usually lost to as the atoms move in the trap. These
horizontal motion for several reasons, in- shifts can severely affect the precision of
cluding slight tilting of the surface, surface spectroscopic measurements. Since one
roughness, horizontal motion associated of the potential applications of trapped
with their residual motion, and horizon- atoms is in high-resolution spectroscopy,
tal ejection by the Gaussian profile of the such inevitable shifts must be carefully
evanescent wave laser beam. The authors considered.
of Ref. [39] simply confined their atoms
in the horizontal direction by surround- 4.2.2 Magnetic Confinement
ing them with a wall of blue-detuned The Stern–Gerlach experiment in 1924
light in the form of a vertical hollow first demonstrated the mechanical action
beam. Their gravito-optical surface trap of inhomogeneous magnetic fields on
OE005
neutral atoms having magnetic moments, directions [44]. Its experimental simplicity
and the basic phenomenon was subse- makes it most attractive, both because of
quently developed and refined. An atom ease of construction and of optical access
with a magnetic moment µ can be con- to the interior. Such a trap was used in the
fined by an inhomogeneous magnetic field first neutral atom trapping experiments at
because of an interaction between the mo- NIST on laser-cooled Na atoms for times
ment and the field. This produces a force exceeding 1 s, and that time was limited
given by only by background gas pressure [43].
F = ∇(
µ
· B) (20) The magnitude of the field is zero at
the center of this trap, and increases in all
since E = −µ Several different mag-
· B. directions as
netic traps with varying geometries that
exploit the force of Eq. (20) have been B = ∇B ρ 2 + 4z2 , (21)
studied in some detail in the literature.
where ρ 2 ≡ x2 + y2 and the field gradient
The general features of the magnetic fields
is constant (see Ref. [44]). The field gradi-
of a large class of possible traps has been
ent is constant along any line through the
presented [44].
origin, but has different values in differ-
W. Paul originally suggested a quadru-
ent polar directions because of the ‘4’ in
pole trap composed of two identical coils
Eq. (21). Therefore, the force of Eq. (20)
carrying opposite currents (see Fig. 10).
that confines the atoms in the trap is
This trap clearly has a single center in
neither harmonic nor central, and orbital
which the field is zero, and is the simplest
angular momentum is not conserved.
of all possible magnetic traps. When the
The requisite field for the quadrupole
coils are separated by 1.25 times their
trap can also be provided in two dimen-
radius, such a trap has equal depth in the
sions by four straight currents as indicated
radial (x-y plane) and longitudinal (z-axis)
in Fig. 11. The field is translationally in-
variant along the direction parallel to the
currents, so a trap cannot be made this
way without additional fields. These are
provided by end coils that close the trap, as
I
shown.
Although there are very many different
kinds of magnetic traps for neutral parti-
cles, this particular one has played a special
− +
I I I
+ −
Fig. 11 The Ioffe trap has four straight current

Fig. 10 Schematic diagram of the coil elements that form a linear quadrupole field. The
configuration used in the quadrupole trap and axial confinement is accomplished with end coils
the resultant magnetic field lines. Because the as shown. These fields can be achieved with
currents in the two coils are in opposite many different current configurations as long as
= 0 point at the center
directions, there is a |B| the geometry is preserved
OE005
role. There are certain conditions required hundred Gauss depth, a ∼ 250 m s−2 , and
for trapped atoms not to be ejected in a the fastest trappable atoms in circular
region of zero field such as occurs at the orbits have vmax ∼ 1 m s−1 so ωT /2π ∼
center of a quadrupole trap (see Sects. 4.2.3 20 Hz. Because of the anharmonicity
and 4.2.4). This problem is not easily cured; of the potential, the orbital frequencies
the Ioffe trap has been used in many of the depend on the orbit size, but in general,
BEC experiments because it has |B| = 0 atoms in lower-energy orbits have higher
everywhere. frequencies.
For the quadrupole trap to work, the
4.2.3 Classical Motion of Atoms in a atomic magnetic moments must be ori-
Quadrupole Trap ented with µ < 0 so that they are
·B
Because of the dependence of the trapping repelled from regions of strong fields.
force on the angle between the field This orientation must be preserved while
and the atomic moment (see Eq. 20), the the atoms move around in the trap even
orientation of the magnetic moment with though the trap fields change directions
respect to the field must be preserved in a very complicated way. The condi-
as the atoms move about in the trap. tion for adiabatic motion can be written
Otherwise, the atoms may be ejected as ωZ |dB/dt|/B, where ωZ = µB/ h̄ is
instead of being confined by the fields the Larmor precession rate in the field.
of the trap. This requires velocities low Since v/ρ1 = v∇B/B = |dB/dt|/B for a
enough to ensure that the interaction uniform field gradient, the adiabaticity
between the atomic moment µ and the condition is
field B is adiabatic especially when the
atom’s path passes through a region ωZ ωT . (22)
where the field magnitude is small and
therefore the energy separation between More general orbits must satisfy a simi-
the trapping and nontrapping states is lar condition. For the two-coil quadrupole
small. This is especially critical at the trap, the adiabaticity condition can be eas-
√
low temperatures of the BEC experiments. ily calculated. Using v = ρ1 a for circular
Therefore energy considerations that focus orbits in the z = 0 plane, the adiabatic con-
only on the trap depth are not sufficient dition for a practical trap (∇B ∼ 1 T/m) re-
to determine the stability of a neutral quires ρ1 ( h̄2 /M2 a)1/3 ∼ 1 µm as well
atom trap; orbit and/or quantum state as v ( h̄a/M)1/3 ∼ 1 cm s−1 . Note that
calculations and their consequences must violation of these conditions (i.e., v ∼
also be considered. 1 cm s−1 in a trap with ∇B ∼ 1 T/m) re-
For the two-coil quadrupole magnetic sults in the onset of quantum dynamics for
trap of Fig. 10, stable circular orbits of the motion (deBroglie wavelength ≈ orbit
radius ρ1 in the z = 0 plane can be found size).
classically by setting µ∇B = Mv2 /ρ1 , so Since the nonadiabatic region of the trap
√
that v = ρ1 a, where a ≡ µ∇B/M is is so small (less than 10−18 m3 compared
the centripetal acceleration supplied by with typical sizes of ∼2 cm, corresponding
the field gradient (cylindrical coordinates to 10−5 m3 ), nearly all the orbits of most
are appropriate). Such orbits atoms are restricted to regions where
√ have an
angular frequency of ωT = a/ρ1 . For they are adiabatic. Therefore, most of
traps of a few centimeter size and a few such laser-cooled atoms stay trapped for
OE005
many thousands of orbits corresponding separation of the rapid precession from

to several minutes. At laboratory vacuum the slower orbital motion is reminiscent
chamber pressures of typically 10−10 torr, of the Born–Oppenheimer approximation
the mean free time between collisions that for molecules.
can eject trapped atoms is ∼2 min, so the On the other hand, the small-n states,
transitions caused by nonadiabatic motion whose orbits are confined to a region near
are not likely to be observable in atoms the origin where the field is small, have
that are optically cooled. much larger coupling to the continuum
states. Thus, they are rapidly ejected from
4.2.4 Quantum Motion in a Trap the trap. The transitions to unbound states
Since laser and evaporative cooling have resulting from the coupling of the motion
the capability to cool atoms to energies with the trapping fields are called Majorana
where their deBroglie wavelengths are on spin flips, and effectively constitute a ‘‘hole’’
the scale of the orbit size, the motional at the bottom of the trap. The evaporative
dynamics must be described in terms cooling process used to produce very cold,
of quantum mechanical variables and dense samples reduces the average total
suitable wave functions. Quantization of energy of the trapped atoms sufficiently
the motion leads to discrete bound states that the orbits are confined to regions
within the trap having µ < 0, and also a
·B near the origin and so, such losses
continuum of unbound states having µ ·B dominate [44, 45].
with opposite sign. There have been different solutions
Studying the behavior of extremely slow to this problem of Majorana losses for
(cold) atoms in the two-coil quadrupole confinement of ultracold atoms for the
trap begins with a heuristic quantization BEC experiments. In the JILA-experiment,
of the orbital angular momentum using the hole was rotated by rotating the
Mr 2 ωT = n h̄ for circular orbits. The en- magnetic field, and thus, the atoms do
ergy levels are then given by not spend sufficient time in the hole to
make a spin flip. In the MIT experiment,
3 the hole was plugged by using a focused
En = E1 n2/3 , where
2 laser beam tuned to the blue side of atomic
E1 = (Ma2 h̄2 )1/3 ∼ h × 5 kHz, (23) resonance, which expelled the atoms from
the center of the magnetic trap. In the
For velocities of optically cooled atoms of Rice experiment, the atoms were trapped
a few cm s−1 , n ∼ 10 − 100. It is readily in an Ioffe trap, which has a nonzero
found that ωZ = nωT , so that the adiabatic field minimum. Most BEC experiments
condition of Eq. (22) is satisfied only for are now using the Ioffe trap solution.
n 1.
These large-n bound states have small 4.3
matrix elements coupling them to the Magneto-optical Traps
unbound continuum states [45]. This can
be understood classically since they spend 4.3.1 Introduction
most of their time in a stronger field, and The most widely used trap for neutral
thus satisfy the condition of adiabaticity atoms is a hybrid employing both optical
of the orbital motion relative to the and magnetic fields to make a magneto-
Larmor precession. In this case, the optical trap (MOT), first demonstrated in
OE005
1987 [46]. The operation of an MOT de- Energy

pends on both inhomogeneous magnetic Me = +1
fields and radiative selection rules to ex-
ploit both optical pumping and the strong
radiative force [46, 47]. The radiative in- Me = 0
d+
teraction provides cooling that helps in d
d−
loading the trap and enables very easy op-
Me = −1
eration. MOT is a very robust trap that
does not depend on precise balancing of wl
the counterpropagating laser beams or on s+ beam s− beam
a very high degree of polarization. Mg = 0

The magnetic field gradients are modest z′ Position
and have the convenient feature that Fig. 12 Arrangement for an MOT in 1-D. The
the configuration is the same as the horizontal dashed line represents the laser
quadrupole magnetic traps discussed in frequency seen by an atom at rest in the center of
Sect. 4.2.2. Appropriate fields can readily the trap. Because of the Zeeman shifts of the
atomic transition frequencies in the
be achieved with simple, air-cooled coils.
inhomogeneous magnetic field, atoms at z = z
The trap is easy to construct because it can are closer to resonance with the σ − laser beam
be operated with a room-temperature cell than with the σ + beam, and are therefore driven
in which alkali atoms are captured from toward the center of the trap
the vapor. Furthermore, low-cost diode
lasers can be used to produce the light the magnetic field, therefore, tunes the
appropriate for many atoms, so the MOT M = −1 transition closer to resonance
has become one of the least expensive ways and the M = +1 transition further out of
to make atomic samples with temperatures resonance. If the polarization of the laser
below 1 mK. beam incident from the right is chosen
Trapping in an MOT works by opti- to be σ − and correspondingly σ + for the
cal pumping of slowly moving atoms in other beam, then more light is scattered
a linearly inhomogeneous magnetic field from the σ − beam than from the σ + beam.
B = B(z) (see Eq. 21), such as that formed Thus, the atoms are driven toward the
by a magnetic quadrupole field. Atomic center of the trap where the magnetic field
transitions with the simple scheme of is zero. On the other side of the center of
Jg = 0 → Je = 1 have three Zeeman com- the trap, the roles of the Me = ±1 states are
ponents in a magnetic field, excited by each reversed and now more light is scattered
of three polarizations, whose frequencies from the σ + beam, again driving the atoms
tune with the field (and therefore with po- toward the center.
sition) as shown in Fig. 12 for 1-D. Two So far, the discussion has been limited
counterpropagating laser beams of oppo- to the motion of atoms in 1-D. However,
site circular polarization, each detuned the MOT scheme can easily be extended
below the zero-field atomic resonance by to 3-D by using six instead of two laser
δ, are incident as shown. beams. Furthermore, even though very
Because of the Zeeman shift, the excited few atomic species have transitions as
state Me = +1 is shifted up for B > simple as Jg = 0 → Je = 1, the scheme
0, whereas the state with Me = −1 is works for any Jg → Je = Jg + 1 transition.
shifted down. At position z in Fig. 12, Atoms that scatter mainly from the σ +
OE005

√
laser beam will be optically pumped toward κ/M. For magnetic field gradients ∇B ≈
the Mg = +Jg substate, which forms a 0.1 T m−1 , the oscillation frequency is typ-
closed system with the Me = +Je substate. ically a few kHz, and this is much smaller
than the damping rate that is typically a
4.3.2 Cooling and Compressing Atoms in few hundred kHz. Thus, the motion is
an MOT overdamped, with a characteristic restor-
For a description of the motion of atoms ing time to the center of the trap of
in an MOT, consider the radiative force in 2MOT /ωMOT2 ≈ several milliseconds for
the low intensity limit (see Eqs. 2 and 3). typical values of detuning and intensity of
The total force on the atoms is given by the lasers.
F = F + + F − , where F ± can be found from
Eqs. (2) and (3), and the detuning δ± for 4.3.3 Capturing Atoms in an MOT
each laser beam is given by δ± = δ ∓ k · Although the approximations that lead to
v ± µ B/ h̄. Here, µ ≡ (ge Me − gg Mg )µB Eq. (24) for force hold for slow atoms near
is the effective magnetic moment for the the origin, they do not apply for the capture
transition used. Note that the Doppler of fast atoms far from the origin. In the
shift ωD ≡ −k · v and the Zeeman shift capture process, the Doppler and Zeeman
ωZ = µ B/ h̄ both have opposite signs for shifts are no longer small compared to
opposite beams. the detuning, so the effects of the position
The situation is analogous to the velocity and velocity can no longer be disentangled.
damping in an OM from the Doppler effect However, the full expression for the force
as discussed in Sec. 3.2, but here the effect still applies and the trajectories of the
also operates in position space, whereas atoms can be calculated by numerical
for molasses it operates only in velocity integration of the equation of motion [48].
space. Since the laser light is detuned The capture velocity of an MOT is
below the atomic resonance in both cases, serendipitously enhanced because atoms
compression and cooling of the atoms is traveling across it experience a decreasing
obtained simultaneously in an MOT. magnetic field just as in beam deceleration
When both the Doppler and Zeeman described in Sect. 3.1. This enables reso-
shifts are small compared to the detuning nance over an extended distance and ve-
δ, the denominator of the force can be locity range because the changing Doppler
expanded as for Eq. (13) and the result shift of decelerating atoms can be com-
becomes pensated by the changing Zeeman shift as
atoms move in the inhomogeneous mag-
F = −βv − κr , (24) netic field. Of course, it will work this
way only if the field gradient ∇B does not
where the damping coefficient β is defined demand an acceleration larger than the
in Eq. (13). The spring constant κ arises maximum acceleration amax . Thus, atoms
from the similar dependence of F on the are subject to the optical force over a dis-
Doppler and Zeeman shifts, and is given tance that can be as long as the trap size,
by κ = µ β∇B/ h̄k and can therefore be slowed considerably.
The force of Eq. (24) leads to damped The very large velocity capture range
harmonic motion of the atoms, where the vcap of an MOT can be estimated by using
damping rate is given by MOT = β/M Fmax = h̄kγ /2 and choosing a maximum
and the oscillation frequency ωMOT = size of a few centimeters for the beam
OE005
diameters. Thus, the energy change can low speed through the center of the MOT
be as large as a few K, corresponding until they emerge on the other side, into
to vcap ∼ 100 m s−1 [47]. The number of the region where light of both frequen-
atoms in a vapor with velocities below vcap cies is present and begin absorbing again.
in the Boltzmann distribution scales as Then they feel the trapping force and are
4 , and there are enough slow atoms to
vcap driven back into the ‘‘dark’’ center of the
fall within the large MOT capture range trap. Such an MOT has been operated at
even at room temperature, because a few MIT [57] with densities close to 1012 /cm3 ,
K includes 10−4 of the atoms. and the limitations are now from colli-
sions in the ground state rather than from
4.3.4 Variations on the MOT Technique multiple light scattering and excited-state
Because of the wide range of applications collisions.
of this most versatile kind of atom trap, a
number of careful studies of its properties
have been made [47, 49–56], and several 5
variations have been developed. One of Optical Lattices
these is designed to overcome the density
5.1
limits achievable in an MOT. In the
Quantum States of Motion
simplest picture, loading additional atoms
into an MOT produces a higher atomic
As the techniques of laser cooling ad-
density because the size of the trapped
vanced from a laboratory curiosity to a tool
sample is fixed. However, the density
for new problems, the emphasis shifted
cannot increase without limit as more
from attaining the lowest possible steady
atoms are added. The atomic density
state temperatures to the study of ele-
is limited to ∼1011 cm−3 because the
mentary processes, especially the quantum
fluorescent light emitted by some trapped
atoms is absorbed by others. mechanical description of the atomic mo-
One way to overcome this limit is to tion. In the completely classical description
have much less light in the center of the of laser cooling, atoms were assumed to
MOT than at the sides. Simply lowering have a well-defined position and momen-
the laser power is not effective in reducing tum that could be known simultaneously
the fluorescence because it will also reduce with arbitrary precision. However, when
the capture rate and trap depth. But those atoms are moving sufficiently slowly that
advantageous properties can be preserved their deBroglie wavelength precludes their
while reducing fluorescence from atoms at localization to less than λ/2π, these de-
the center if the light intensity is low only scriptions fail and a quantum mechanical
in the center. description is required. Such exotic be-
The repumping process for the alkali havior for the motion of whole atoms, as
atoms provides an ideal way of imple- opposed to electrons in the atoms, had not
menting this idea [57]. If the repumping been considered before the advent of laser
light is tailored to have zero intensity at cooling simply because it was too far out of
the center, then atoms trapped near the the range of ordinary experiments. A series
center of the MOT are optically pumped of experiments in the early 1990s provided
into the ‘‘wrong’’ hfs state and stop fluo- dramatic evidence for these new quantum
rescing. They drift freely in the ‘‘dark’’ at states of motion of neutral atoms, and
OE005
led to the debut of deBroglie wave atom means that their deBroglie wavelengths are
optics. equal to the optical wavelengths divided by
The quantum description of atomic a ‘‘few’’. If the depth of the optical potential
motion requires that the energy of such wells is high enough to contain such very
motion be included in the Hamiltonian. slow atoms, then their motion in potential
The total Hamiltonian for atoms moving wells of size λ/2 must be described
in a light field would then be given by quantum mechanically, since they are
confined to a space of size comparable
H = Hatom + Hrad + Hint + Hkin , (25) to their deBroglie wavelengths. Thus, they
where Hatom describes the motion of the do not oscillate in the sinusoidal wells as
atomic electrons and gives the internal classical localizable particles, but instead
atomic energy levels, Hrad is the energy of occupy discrete, quantum mechanical
the radiation field and is of no concern here bound states [60], as shown in the lower
because the field is not quantized, Hint part of Fig. 13.
describes the excitation of atoms by the The basic ideas of the quantum me-
light field and the concomitant light shifts, chanical motion of particles in a periodic
and Hkin is the kinetic energy operator of potential were laid out in the 1930s with the
the motion of the center of mass of the Kronig–Penney model and Bloch’s theo-
atoms. This Hamiltonian has eigenstates rem, and optical lattices offer important
of not only the internal energy levels and opportunities for their study. For example,
the atom-laser interaction that connects
them, but also that of the kinetic energy
operator Hkin ≡ P2 /2M. These eigenstates
will therefore be labeled by quantum
numbers of the atomic states as well as
the center of mass momentum p. An atom
in the ground state, |g; p, has an energy
Eg + p2 /2M, that can take on a range of
values.
In 1968, V.S. Letokhov [58] •suggested
Energy
Q11
that it is possible to confine atoms in the
wavelength-size regions of a standing wave
by means of the dipole force that arises
from the light shift. This was first accom-
plished in 1987 in 1-D with an atomic beam
traversing an intense standing wave [59]. 0 l/4 l/2 3l/4 l
Since then, the study of atoms confined Position
to wavelength-size potential wells has be- Fig. 13 Energy levels of atoms moving in the
come an important topic in optical control periodic potential of the light shift in a standing
of atomic motion because it opens up con- wave. There are discrete bound states deep in
figurations previously accessible only in the wells that broaden at higher energy, and
condensed matter physics using crystals. become bands separated by forbidden energies
above the tops of the wells. Under conditions
The limits of laser cooling discussed in appropriate to laser cooling, optical pumping
Sect. 3.3.4 suggest that atomic momenta among these states favors populating the lowest
can be reduced to a ‘‘few’’ times h̄k. This ones as indicated schematically by the arrows
OE005
these lattices can be made essentially free

of defects with only moderate care in
spatially filtering the laser beams to as-
sure a single transverse mode structure.
Furthermore, the shape of the potential
is exactly known and does not depend
on the effect of the crystal field or the
ionic energy level scheme. Finally, the
laser parameters can be varied to mod-
ify the depth of the potential wells without
changing the lattice vectors, and the lat-
tice vectors can be changed independently Fig. 14 The ‘‘egg-crate’’ potential of an optical
by redirecting the laser beams. The sim- lattice shown in two dimensions. The potential
wells are separated by λ/2
plest optical lattice to consider is a 1-D
pair of counterpropagating beams of the
same polarization, as was used in the first motion can also be observed by stimu-
experiment [59]. lated emission [62, 63] and by direct RF
Of course, such tiny traps are usually spectroscopy [64, 65].
very shallow, so loading them requires
cooling to the µK regime. Even atoms 5.2
whose energy exceeds the trap depth Properties of 3-D Lattices
must be described as quantum mechanical
particles moving in a periodic potential The name ‘‘optical lattice’’ is used rather
than optical crystal because the filling
that display energy band structure [60].
fraction of the lattice sites is typically
Such effects have been observed in very
only a few percent (as of 1999). The limit
careful experiments.
arises because the loading of atoms into
Because of the transverse nature of light,
the lattice is typically done from a sample
any mixture of beams with different k-
of trapped and cooled atoms, such as an
vectors necessarily produces a spatially MOT for atom collection, followed by an
periodic, inhomogeneous light field. The OM for laser cooling. The atomic density in
importance of the ‘‘egg-crate’’ array of such experiments is limited by collisions
potential wells arises because the asso- and multiple light scattering to a few times
ciated atomic light shifts can easily be 1011 cm−3 . Since the density of lattice sites
comparable to the very low average atomic of size λ/2 is a few times 1013 cm−3 , the
kinetic energy of laser-cooled atoms. A filling fraction is necessarily small. With
typical example projected against two di- the advent of experiments that load atoms
mensions is shown in Fig. 14. directly into a lattice from a BEC, the filling
Atoms trapped in wavelength-sized factor can be increased to 100%, and in
spaces occupy vibrational levels similar some cases it may be possible to load more
to those of molecules. The optical spec- than one atom per lattice site [66, 67].
trum can show Raman-like sidebands that In 1993 a very clever scheme was
result from transitions among the quan- described [68]. It was realized that an n-
tized vibrational levels [61, 62] as shown in dimensional lattice could be created by
Fig. 15. These quantum states of atomic only n + 1 traveling waves rather than
OE005
0.6 0.5
Fluorescence (arb. units)
Fluorescence (arb. units)

0.4
0.4
0.3
0.2 × 10
× 20 0.2
0.1
0
0
−200 −100 0 100 200 −200 −100 0 100 200
(a) Frequency (kHz) (b) Frequency (kHz)
Fig. 15 (a) Fluorescence spectrum in a 1-D lin ⊥ lin optical molasses. Atoms are first
captured and cooled in an MOT, and then the MOT light beams are switched off leaving a pair
of lin ⊥ lin because of spontaneous emission of the atoms to the same vibrational state from
which they are excited, whereas the sideband on the left (right) is due to spontaneous
emission to a vibrational state with one vibrational quantum number lower (higher) (see
Fig. 13). The presence of these sidebands is a direct proof of the existence of the band
structure. (b) Same as (a) except that the 1-D molasses is σ + − σ − , which has no spatially
dependent light shift and hence no vibrational states (figure taken from Jessen, P. S., Gerz, C.,
Lett, P. D., Phillips, W. D., Rolston, S. L., Spreeuw, R. J. C., Westbrook, C. I. (1992), Phys. Rev.
Lett. 69, 49–52)
2n. The real benefit of this scheme is 5.3

that in case of phase instabilities in the Spectroscopy in 3-D Lattices
laser beams, the interference pattern is
only shifted in space, but the interference The group at NIST developed a new
pattern itself is not changed. Instead method that superposed a weak probe
of producing optical wells in 2-D with beam of light directly from the laser upon
four beams (two standing waves), these some of the fluorescent light from the
authors used only three. The k-vectors atoms in a 3-D OM, and directed the
of the coplanar beams were separated by light from these combined sources onto
2π/3, and they were all linearly polarized on a fast photodetector [70]. The resulting
in their common plane (not parallel beat signal carried information about the
to one another). The same immunity Doppler shifts of the atoms in the optical
to vibrations was established for a 3- lattices [61]. These Doppler shifts were
D optical lattice by using only four expected to be in the sub-MHz range for
beams arranged in a quasi-tetrahedral atoms with the previously measured 50-
configuration. The three linearly polarized µK temperatures. The observed features
beams of the 2-D arrangement described confirmed the quantum nature of the
above were directed out of the plane motion of atoms in the wavelength-size
toward a common vertex, and a fourth potential wells (see Fig. 15) [15].
circularly polarized beam was added. The NIST group also studied atoms
All four beams were polarized in the loaded into an optical lattice using Bragg
same plane [68]. The authors showed that diffraction of laser light from the spatially
this configuration produced the desired ordered array [71]. They cut off the laser
potential wells in 3-D. beams that formed the lattice, and before
OE005
the atoms had time to move away from to a speed faster than that corresponding
their positions, they pulsed on a probe to the edge of a Brillouin zone, and
laser beam at the Bragg angle appropriate that at longer times the particles would
for one of the sets of lattice planes. The execute oscillatory motion. Ever since then,
Bragg diffraction not only enhanced the experimentalists have tried to observe
reflection of the probe beam by a factor these Bloch oscillations in increasingly
of 105 , but by varying the time between pure and/or defect-free crystals.
the shut-off of the lattice and turn-on of Atoms moving in optical lattices are ide-
the probe, they could also measure the ally suited for such an experiment, as was
‘‘temperature’’ of the atoms in the lattice. beautifully demonstrated in 1996 [69]. The
The reduction of the amplitude of the authors loaded a 1-D lattice with atoms
Bragg-scattered beam with time provided from a 3-D molasses, further narrowed the
some measure of the diffusion of the atoms velocity distribution, and then instead of
away from the lattice sites, much like the applying a constant force, simply changed
Debye–Waller factor in X-ray diffraction. the frequency of one of the beams of the
1-D lattice with respect to the other in
5.4 a controlled way, thereby creating an ac-
Quantum Transport in Optical Lattices celerating lattice. Seen from the atomic
reference frame, this was the equivalent of
In the 1930s, Bloch realized that applying a constant force trying to accelerate them.
a uniform force to a particle in a periodic After a variable time ta , the 1-D lattice
potential would not accelerate it beyond a beams were shut off and the measured
certain speed, but instead would result atomic velocity distribution showed beau-
in Bragg reflection when its deBroglie tiful Bloch oscillations as a function of ta .
wavelength became equal to the lattice The centroid of the very narrow velocity
period. Thus, an electric field applied to distribution was seen to shift in velocity
a conductor could not accelerate electrons space at a constant rate until it reached
Atom number (arb. units)
ta = t
ta = 3t/4
Mean velocity
0.5
ta = t/2
0
ta = t/4
ta = 0 −0.5
−1 0 1 −2 −1 0 1 2
(a) Atomic momentum (h k) (b) Quasi-momentum (k)
Fig. 16 Plot of the measured velocity distribution verses. time in the accelerated 1-D lattice.
(a) Atoms in a 1-D lattice are accelerated for a fixed potential depth for a certain time ta and
the momentum of the atoms after the acceleration is measured. The atoms accelerate only
to the edge of the Brillouin zone where the velocity is +vr , and then the velocity distribution
appears at −vr . (b) Mean velocity of the atoms as a function of the quasi-momentum, that
is, the force times the acceleration time (figure taken from Ben Dahan, M., Peik, E.,
Reichel, J., Castin, Y., Salomon, C. (1996), Phys. Rev. Lett. 76, 4508–4511)
OE005
vr = h̄k/M, and then it vanished and reap- Temperatures lower than the recoil limit
peared at −vr as shown in Fig. 16. The are readily achieved by evaporative cooling,
shape of the ‘‘dispersion curve’’ allowed and so all BEC experiments employ it in
measurement of the ‘‘effective mass’’ of their final phase. Evaporative cooling is
the atoms bound in the lattice. inherently different from the other cooling
processes discussed in Sect. 3, and hence
discussed here separately.
6 Since the first observations in 1995, BEC
Bose–Einstein Condensation has been the subject of intense investiga-
tion, both theoretical and experimental. No
6.1
attempt is made in this article to even ad-
Introduction
dress, much less cover, the very rich range
of physical phenomena that have been un-
In the 1920s, Bose and Einstein predicted
veiled by these studies. Instead, we focus
that for sufficiently high phase space
on the methods to achieve ρφ ∼ 1 and
density, ρφ ∼ 1 (see Sect. 1.2), a gas of
BEC.
atoms undergoes a phase transition that is
now called Bose–Einstein condensation. 6.2
It took 70 years before BEC could be Evaporative Cooling
unambiguously observed in a dilute gas.
From the advent of laser cooling and Evaporative cooling is based on the
trapping, it became clear that this method preferential removal of those atoms from
could be instrumental in achieving BEC. a confined sample with an energy higher
BEC is another manifestation of quanti- than the average energy followed by a
zation of atomic motion. It occurs in the rethermalization of the remaining gas by
absence of resonant light, and its onset is elastic collisions. Although evaporation is
characterized by cooling to the point where a process that occurs in nature, it was
the atomic deBroglie wavelengths are com- applied to atom cooling for the first time
parable to the interatomic spacing. This is in 1988 [72].
in contrast with the topics discussed in One way to think about evaporative
Sect. 5 where the atoms were in an optical cooling is to consider cooling of a con-
field and their deBroglie wavelengths were tainer of hot liquid. Since the most
comparable to the optical wavelength λ. energetic molecules evaporate from the
Laser cooling alone is inherently inca- liquid and leave the container, the remain-
pable of achieving ρφ ∼ 1. This is easily ing molecules obtain a lower temperature
seen from the recoil limit of Sect. 3.3.4 and are cooled. Furthermore, it requires
that limits λdeB to λ/‘‘few’’. Since the cross the evaporation of only a small fraction
section for optical absorption near reso- of the liquid to cool it by a considerable
nance is σ ∼ λ2 near ρφ ∼ 1, this limit of amount.
λdeB ∼ λ/‘‘few’’ results in the penetration Evaporative cooling works by remov-
depth of the cooling light into the sample ing the higher-energy atoms as sug-
being smaller than λ. Thus, the sample gested schematically in Fig. 17. Those
would have to be extremely small and con- that remain have much lower average en-
tain only a few atoms, hardly a system ergy (temperature) and so they occupy a
suitable for investigation. smaller volume near the bottom of the
OE005
Several models have been developed

to describe this process, but we present
here the simplest one [77] because of its
pedagogical value [22]. In this model, the
trap depth is lowered in one single step
and the effect on the thermodynamic
quantities, such as temperature, density,
and volume, is calculated. The process can
be repeated and the effects of multiple
steps added up cumulatively.
In such models of evaporative cooling,
Fig. 17 Principle of the evaporation technique. the following assumptions are made:
Once the trap depth is lowered, atoms with
energy above the trap depth can escape and the
remaining atoms reach a lower temperature 1. The gas behaves sufficiently ergodically,
that is, the distribution of atoms
trap, thereby increasing their density. For in phase space (both position and
trapped atoms, it can be achieved by lower- momentum) depends only on the
ing the depth of the trap, thereby allowing energy of the atoms and the nature
the atoms with energies higher than the of the trap.
trap depth to escape, as discussed first by 2. The gas is assumed to begin the
Hess [73]. Elastic collisions in the trap then process with ρφ 1 (far from the BEC
lead to a rethermalization of the gas. This transition point), and so it is described
technique was first employed for evapora- by classical statistics.
tive cooling of hydrogen [72, 74–76]. Since 3. Even though ρφ 1, the gas is cold
both the temperature and the volume de- enough that the atomic scattering is
crease, ρφ increases. pure (s-wave) quantum mechanical,
Recently, more refined techniques have that is, the temperature is sufficiently
been developed. For example, to sustain low that all higher partial waves do
the cooling process the trap depth can not contribute to the cross section.
be lowered continuously, achieving a Furthermore, the cross section for
continuous decrease in temperature. Such elastic scattering is energy-independent
a process is called forced evaporation and is and is given by σ = 8πa2 , where a is the
discussed in Sect. 6.3 below. scattering length. It is also assumed that
the ratio of elastic to inelastic collision
6.2.1 Simple Model rates is sufficiently large that the elastic
This section describes a simple model of collisions dominate.
evaporative cooling. Since such cooling is 4. Evaporation preserves the thermal na-
not achieved for single atoms but for the ture of the distribution, that is, the
whole ensemble, an atomic description of thermalization is much faster than the
the cooling process must be replaced by rate of cooling.
thermodynamic methods. These methods 5. Atoms that escape from the trap neither
are completely different from the rest of the collide with the remaining atoms nor
material in this article, and will therefore exchange energy with them. This is
remain rather elementary. called full evaporation.
OE005
6.2.2 Application of the Simple Model quantities for this process (the values af-
The first step in applying this simple model ter the process are denoted by a prime).
is to characterize the trap by calculating One of these is ν ≡ N /N, the fraction of
how the volume of a trapped sample atoms remaining in the trap after the cool-
of atoms changes with temperature T. ing. The other is γ (This γ is not to be
Consider a trapping potential that can be confused with the natural width of the ex-
expressed as a power law given by cited state.), a measure of the decrease in
temperature caused by the release of hot
s1 s2
x y atoms and subsequent cooling, modified
U(x, y, z) = ε1 + ε2
a1 a2 by ν, and defined as
s3
z
+ ε3 , (26) log(T /T) log(T /T)
a3 γ ≡ = . (28)
log(N /N) log ν
where aj is a characteristic length and sj the
This yields a power-law dependence for
power, for a certain direction j. Then the
the decrease in temperature caused by
volume occupied by trapped atoms scales
the loss of the evaporated particles, that
as V ∝ T ξ [78], where
is, T = Tν γ . The dependence of the
1 1 1 other thermodynamic quantities on the
ξ≡ + + . (27) parameters ν and γ can then be calculated.
s1 s2 s3
The scaling of N = Nν, T = Tν γ , and
Thus, the effect of the potential on the V = Vν γ ξ can provide the scaling of
volume of the trapped sample for a all the other thermodynamic quantities
given temperature can be reduced to a of interest by using the definitions for
single parameter ξ . This parameter is the density n = N/V, the phase space
independent of how the occupied volume density ρφ = nλ3deB ∝ nT −3/2 , and the
is defined, since many different definitions elastic collision rate kel ≡ nσ v ∝ nT 1/2 .
lead to the same scaling. When a gas is held The results are given in Table 2. For a
in a 3-D box with infinitely high walls, then given value of η, the scaling of all quantities
s1 = s2 = s3 = ∞ and ξ = 0, which means
that V is independent of T, as expected.
For a harmonic potential in 3-D, ξ = 3/2; Tab. 2 Exponent q for the scaling of the
for a linear potential in 2-D, ξ = 2; and for thermodynamic quantities X = Xν q with the
a linear potential in 3-D, ξ = 3. reduction ν of the number of atoms in the trap
The evaporative cooling model itself [77]
starts with a sample of N atoms in volume Thermodynamic Symbol Exponent q
variable
V having a temperature T held in an in-
finitely deep trap. The strategy for using
Number of N 1
the model is to choose a finite quantity η, atoms
and then (1) lower the trap depth to a value Temperature T γ
ηkB T, (2) allow for a thermalization of the Volume V γξ
sample by collisions, and (3) determine the Density n 1 − γξ
Phase space ρ 1 − γ (ξ + 3/2)
change in ρφ .
density
Only two parameters are needed to com- Collision rate k 1 − γ (ξ − 1/2)
pletely determine all the thermodynamic
OE005
depends only on γ . Note that for successive are considered. This leads to the Boltz-
steps j, ν has to be replaced with ν j . mann equation [81].
The fraction of atoms remaining is fully
determined by the final trap depth η for 6.2.4 Limiting Temperature
a given potential characterized by the trap In the models discussed so far, only
parameter ξ . In order to determine the elastic collisions have been considered,
change of the temperature in the cooling that is, collisions where kinetic energy
process, it is necessary to consider in is redistributed between the partners.
detail the distribution of the atoms in the However, if part of the internal energy
trap, and this is discussed more fully in of the colliding partners is exchanged with
Refs. [1, 22, 77]. their kinetic energy during •collision, then
Q12
it is inelastic. Inelastic collisions can cause
6.2.3 Speed of Evaporation problems for two reasons: (1) the internal
So far, the speed of the evaporative cooling energy released can cause the atoms to
process has not been considered. If the heat up and (2) the atoms can change their
trap depth is ramped down too quickly, the internal states, and the new states may
thermalization process does not have time no longer be trapped. In each case, such
to run its course and the process becomes collisions can lead to trap loss and are
less efficient. On the other hand, if the therefore not desirable.
trap depth is ramped down too slowly, Apart from collisions with the back-
the loss of particles by inelastic collisions ground gas and three-body recombination,
becomes important, thereby making the there are two inelastic processes that are
evaporation inefficient. important for evaporative cooling of alkali
The speed of evaporation can be found atoms: dipolar relaxation and spin relax-
from the principle of detailed balance[22]. ation. The collision rate nkdip for them
Its application shows that the ratio of the at low energies is independent of veloc-
evaporation time and the elastic collision ity [82]. Since the elastic collision rate is
time is √ η given by kel = nσ vrel , the ratio of good
τev 2e (= elastic) to bad (= relaxation) collisions
= . (29)
τel η goes down when the temperature does.
This limits the temperature to a value Te
Note that this ratio increases exponentially near which •the ratio between good and
with η. bad collisions becomes unity, and Te is
Q13
Experimental results show that ∼2.7 given by

elastic collisions are necessary to rether-
malize the gas [79]. In order to model the πMk2dip
rethermalization process, Luiten et al [80] kB Te = . (30)
16σ 2
have discussed a model based on the Boltz-
mann equation where the evolution of the The limiting temperature for the alkalis
phase space density ρ(r , p, t) is calculated. is of the order of 1 nK, depending on the
This evolution is not only caused by the values of σ and kdip .
trapping potential, but also by collisions In practice, however, this ratio has to
between the particles. Only elastic col- be considerably larger than unity, and
lisions, whose cross section is given by so the practical limit for evaporative
σ = 8πa2 with a as the scattering length, cooling occurs when the ratio is ∼103 [22].
OE005
In the model of Ref. [80], the authors atoms and waste them than it is to cool
discuss different strategies for evaporative them.
cooling. Even for the strategy of the For the evaporation of the atoms, it
lowest temperature, the final temperature is important that atoms with an energy
is higher than Te . above the cutoff are expelled from the trap.
The collision rate between atoms with By using RF-evaporation, one can expel
one in the excited state (S + P collisions) is the atoms in all three dimensions equally
also much larger at low temperatures than and thus obtain a true 3-D evaporation. In
the rate for such collisions with both atoms the case of the TOP•-trap, the atoms are
Q14
in the ground state (S + S collisions). Since evaporated along the outer side of the cloud
S + P collisions are generally inelastic and that is exposed to the highest magnetic
the inelastic energy exchange generally field on the average. This is a cylinder
leads to a heating of the atoms, increasing along the direction of rotation axis of the
the density increases the loss of cold magnetic field and thus the evaporation
atoms. To achieve BEC, resonant light takes place in 2-D.
should therefore be avoided, and thus laser Once the energy of the atoms becomes
cooling is not suitable for achieving BEC. very small, the atoms sag because of gravity
and the outer shell of the cloud is no longer
6.3
at a constant magnetic field. Atoms at the
Forced Evaporative Cooling bottom of the trap have the highest energy
and thus the evaporation becomes 1-D. In
In all the earliest experiments that achieved case of harmonic confinement, Utrap =
BEC, the evaporative cooling was ‘‘forced’’ U z2/2, the equipotential surface is at
by inducing rf transitions to magnetic z ≈ 2ηkB T/U . Now, the gravitational
sublevels that are not bound in the energy is given by Ugrav = mgz and
magnetic trap. Atoms with the highest thus the limiting temperature for 1-D
evaporation to take place is given by [22]
energies can access regions of the trap
where the magnetic field is stronger, and
2η(mg)2
thus their Zeeman shifts would be larger. kB T < (31)
µB
A correspondingly high-frequency rf field
would cause only these most energetic For a curvature of B = 500 T m−2 , the
atoms to undergo transitions to states limiting temperature becomes 1 µK for 7 Li,
that are not trapped, and in doing so, the 10 µK for 23 Na, and 150 µK for 87 Rb. Below
departing atoms carry away more than the this temperature, evaporation becomes
average energy. Thus, a slow sweep of less efficient.
the rf frequency from high to low would In the three experiments that obtained
continuously shave off the high-energy tail BEC for the first time in 1995, the problem
of the energy distribution, and thereby of this ‘‘gravitational sag’’ was not known,
continuously drive the temperature lower but it did not prevent the experimentalist
and the phase space density higher. The from observing BEC. The solution used
results of evaporative cooling from the first in those experiments was because of the
three groups that have obtained BEC have light mass (7 Li), tight confinement (23 Na),
shown that using this rf shaving technique, and TOP trap (87 Rb). In the last case, the
it is much easier to select high-energy axis of rotation is in the z-direction and
OE005
Tab. 3 Typical numbers for the phase space density as obtained in the
experiments aimed at achieving BEC. The different stages of cooling and
trapping the atoms will be discussed in the Appendix
Stages T λdeB n ρφ
Oven 300 ◦ C 0.02 nm 1010 cm−3 10−16

Slowing 30 mK 2 nm 108 cm−3 10−12
Pre-cooling 1 mK 10 nm 109 cm−3 10−9
Trapping 1 mK 10 nm 1012 cm−3 10−6
Cooling 1 µK 0.3 µm 1011 cm−3 3 × 10−3
Evaporation 70 nK 1 µm 1012 cm−3 2.612
thus the evaporation always remains 2- [5] Ertmer, W., Blatt, R., Hall, J. L., Zhu, M.
D. Table 3 shows typical values of ρφ for (1985), Phys. Rev. Lett. 54, 996–999.
[6] Watts, R., Wieman, C. (1986), Opt. Lett. 11,
various situations.
291–293.
[7] Bagnato, V., Lafyatis, G., Martin, A.,
Raab, E., Ahmad-Bitar, R., Pritchard, D.
7 (1987), Phys. Rev. Lett. 58, 2194–2197.
Conclusion [8] Molenaar, P. A., vander Straten, P., Heide-
man, H. G. M., Metcalf, H. (1997), Phys.
In this article, we have reviewed some Rev. A 55, 605–614.
[9] Barrett, T. E., Dapore-Schwartz, S. W., Ray,
of the fundamentals of optical control of
M. D., Lafyatis, G. P. (1991), Phys. Rev. Lett.
atomic motion. The reader is cautioned 67, 3483–3487.
that this is by no means an exhaustive [10] Dalibard, J., Phillips, W. (1985), Bull. Am.
review of the field, and that many impor- Phys. Soc. 30, 748.
tant and current topics have been omitted. [11] Chu, S., Hollberg, L., Bjorkholm, J., Ca-
Much of the material here was taken from ble, A., Ashkin, A. (1985), Phys. Rev. Lett.
55, 48–51.
our recent textbook [1], and the reader is
[12] Sheehy, B., Shang, S. Q., vander Straten, P.
encouraged to consult that source for the Metcalf, H. (1990), Chem. Phys. 145,
origin of many of the formulas presented 317–325.
in the present text, as well as for further [13] Gould, P., Lett, P., Phillips, W. D. (1987),
reading and more detailed references to New measurement with optical molasses,
the literature. in W. Persson, S. Svanberg, (Eds.), Laser
Spectroscopy VIII, Berlin: Springer, p. 64•.
Q15
[14] Hodapp, T., Gerz, C., Westbrook, C., Fut-
References rlehner, C., Phillips, W. (1992), Bull. Am.
Phys. Soc. 37, 1139.
[15] Lett, P., Watts, R., Westbrook, C.,
[1] Metcalf, H. J., vander Straten, P. (1999), Phillips, W., Gould, P., Metcalf, H. (1988),
Laser Cooling and Trapping. New York: Phys. Rev. Lett. 61, 169–172.
Springer-Verlag. [16] Lett, P. D., Watts, R. N., Tanner, C. E., Rol-
[2] Phillips, W., Metcalf, H. (1982), Phys. Rev. ston, S. L., Phillips, W. D., Westbrook, C. I.
Lett. 48, 596–599. (1989), J. Opt. Soc. Am. B 6, 2084–2107.
[3] Prodan, J., Phillips, W., Metcalf, H. (1982), [17] Dalibard, J., Cohen-Tannoudji, C. (1989), J.
Phys. Rev. Lett. 49, 1149–1153. Opt. Soc. Am. B 6, 2023–2045.
[4] Prodan, J., Phillips, W. (1984), Prog. Quant. [18] Ungar, P. J., Weiss, D. S., Chu, S., Riis, E.
Electron. 8, 231–235. (1989), J. Opt. Soc. Am. B 6, 2058–2071.
OE005
[19] Cohen-Tannoudji, C., Phillips, W. D. (1990), [42] Wineland, D., Itano, W., Bergquist, J.,
Phys. Today 43, 33–40. Bollinger, J. (1985), Trapped Ions and Laser
[20] Gupta, R., Padua, S., Xie, C., Batelaan, H., Cooling. Technical Report 1086, N.I.S.T.•.
Q17
Metcalf, H. (1994), J. Opt. Soc. Am. B 11, [43] Migdall, A., Prodan, J., Phillips, W., Berge-
537–541. man, T., Metcalf, H. (1985), Phys. Rev. Lett.
[21] Salomon, C., Dalibard, J., Phillips, W. D., 54, 2596.
Clairon, A., Guellati, S. (1990), Europhys. [44] Bergeman, T., Erez, G., Metcalf, H. (1987),
Lett. 12, 683–688. Phys. Rev. A 35, 1535.
[22] Ketterle, W., Vandruten, N. J. (1996), Adv. [45] Bergeman, T. H., Balazs, N. L., Metcalf, H.,
At. Mol. Opt. Phys. 37, 181–236. Mcnicholl, P., Kycia, J. (1989), J. Opt. Soc.
[23] Aspect, A., Arimondo, E., Kaiser, R., Vans- Am. B 6, 2249–2256.
teenkiste, N., Cohen-Tannoudji, C. (1988), [46] Raab, E., Prentiss, M., Cable, A., Chu, S.,
Phys. Rev. Lett. 61, 826–829. Pritchard, D. (1987), Phys. Rev. Lett. 59,
[24] Kasevich, M., Chu, S. (1992), Phys. Rev. Lett. 2631–2634.
69, 1741–1744. [47] Metcalf, H. (1989), J. Opt. Soc. Am. B 6,
[25] Ashkin, A. (1970), Phys. Rev. Lett. 24, 2206–2210.
156–159. [48] Molenaar, P. (1995), Photoassociative Reac-
[26] Chu, S., Bjorkholm, J., Ashkin, A., Ca- tions of Laser-Cooled Sodium. Ph.D. Thesis,
ble, A. (1986), Phys. Rev. Lett. 57, 314–317. Utrecht University•.
Q18
[27] Ashkin, A. (1980), Science 210, 1081–1088. [49] Cornell, E. A., Monroe, C., Wieman, C. E.
[28] Ashkin, A., Dziedzic, J. M. (1985), Phys. (1991), Phys. Rev. Lett. 67, 2439–2442.
Rev. Lett. 54, 1245–1248. [50] Steane, A. M., Foot, C. J. (1991), Europhys.
[29] Ashkin, A., Dziedzic, J. M. (1987), Science Lett. 14, 231–236.
235, 1517–1520. [51] Steane, A. M., Chowdhury, M., Foot, C. J.
[30] Miller, J. D., Cline, R. A., Heinzen, D. J. (1992), J. Opt. Soc. Am. B 9, 2142–2158.
(1993), Phys. Rev. A 47, R4567–R4570. [52] Gould, P. private communication•.
Q19
[31] Adams, C. S., Lee, H. J., Davidson, N., Ka- [53] Walker, T., Sesko, D., Wieman, C. (1990),
sevich, M., Chu, S. (1995), Phys. Rev. Lett. Phys. Rev. Lett. 64, 408–411.
74, 3577–3580. [54] Sesko, D. W., Walker, T. G., Wieman, C. E.
[32] Takekoshi, T., Knize, R. J. (1996), Opt. Lett. (1991), J. Opt. Soc. Am. B 8, 946–958.
21, 77–79. [55] Gibble, K. E., Kasapi, S., Chu, S. (1992),
[33] Metcalf, H., Phillips, W. (1986), Metrologia Opt. Lett. 17, 526–528.
22, 271–278. [56] Lindquist, K., Stephens, M., Wieman, C.
[34] Davidson, N., Lee, H. J., Adams, C. S., Ka- (1992), Phys. Rev. A 46, 4082–4090.
sevich, M., Chu, S. (1995), Phys. Rev. Lett. [57] Ketterle, W., Davis, K. B., Joffe, M. A., Mar-
74, 1311–1314. tin, A., Pritchard, D. E. (1993), Phys. Rev.
[35] Siegman, A. (1986), Lasers. Mill Valley: Lett. 70, 2253–2256.
University Sciences. [58] Lethokov, V. S. (1968), JETP Lett. 7, 272.
[36] Simpson, N., Dholakia, K., Allen, L., Pad- [59] Salomon, C., Dalibard, J., Aspect, A., Met-
gett, M. (1997), Opt. Lett. 22, 52–54. calf, H., Cohen-Tannoudji, C. (1987), Phys.
[37] McGloin, D., Simpson, N., Padgett, M. Rev. Lett. 59, 1659–1662.
(1998), App. Opt. 37, 469–472. [60] Castin, Y., Dalibard, J. (1991), Europhys.
[38] Beijersbergen, M. (1996), Phase Singulari- Lett. 14, 761–766.
ties in Optical Beams. Ph.D. Thesis, Uni- [61] Jessen, P. S., Gerz, C., Lett, P. D., Phillips,
versity Leiden•. W. D., Rolston, S. L., Spreeuw, R. J. C.,
Q16
[39] Ovchinnikov, Yu. B., Manek, I. Grimm, R. Westbrook, C. I. (1992), Phys. Rev. Lett. 69,
(1997), Phys. Rev. Lett. 79, 2225–2228. 49–52.
[40] Aminoff, C. G., Steane, A. M., Bouyer, P., [62] Verkerk, P., Lounis, B., Salomon, C., Cohen-
Desbiolles, P., Dalibard, J., Cohen-Tann- Tannoudji, C., Courtois, J. Y., Grynberg, G.
oudji, C. (1993), Phys. Rev. Lett. 71, (1992), Phys. Rev. Lett. 68, 3861–3864.
3083–3086. [63] Lounis, B., Verkerk, P., Courtois, J. Y.,
[41] Kasevich, M. A., Weiss, D. S., Chu, S. Salomon, C., Grynberg, G. (1993), 1-D Euro-
(1990), Opt. Lett. 15, 607–609. phys. Lett. 21, 13–17.
OE005
[64] Gupta, R., Padua, S., Xie, C., Batelaan, H., Kochanski, G. P. (1988), Phys. Rev. Lett. 61,
Bergeman, T., Metcalf, H. (1992), Bull. Am. 935.
Phys. Soc. 37, 1139. [73] Hess, H. F. (1986), Phys. Rev. B 34, 3476.
[65] Gupta, R., Padua, S., Bergeman, T., Met- [74] Doyle, J. M., Sandberg, J. C., Yu, I. A., Ce-
calf, H. (1993), Search for motional quan- sar, C. L., Kleppner, D., Greytak, T. J. (1991),
tization of laser-cooled atoms, in E. Ari- Phys. Rev. Lett. 67, 603.
mondo, W. Phillips, F. Strumia (Eds.), Laser [75] Luiten, O. J., Werij, H. G. C., Setija, I. D.,
Manipulation of Atoms and Ions, Proceedings Reynolds, M. W., Hijmans, T. W., Wal-
of Fermi School CXVIII, Varenna, Amster- raven, J. T. M. (1993), Phys. Rev. Lett. 70,
dam: North Holland. 544–547.
[66] Anderson, B. P., Kasevich, M. A. (1998), [76] Setija, I. D., Werij, H. G. C., Luiten, O. J.,
Nature 282, 1686–1689. Reynolds, M. W., Hijmans, T. W., Wal-
[67] Greiner, M., Mandel, O., Esslinger, T., raven, J. T. M. (1993), Phys. Rev. Lett. 70,
Hänsch, T. W., Bloch, I. (2002), Nature 415, 2257–2260.
39–44. [77] Davis, K. B., Mewes, M. O., Ketterle, W.
[68] Grynberg, G., Lounis, B., Verkerk, P., Cour- (1995), App. Phys. B 60, 155–159.
tois, J. Y., Salomon, C. (1993), Phys. Rev. [78] Bagnato, V., Pritchard, D. E., Kleppner, D.
Lett. 70, 2249–2252. (1987), Phys. Rev. A 35, 4354.
[69] Ben Dahan, M., Peik, E., Reichel, J., Cas- [79] Monroe, C., Cornell, E., Sackett, C., My-
tin, Y., Salomon C. (1996), Phys. Rev. Lett. att, C., Wieman, C. (1993), Phys. Rev. Lett.
76, 4508–4511. 70, 414.
[70] Westbrook, C. I., Watts, R. N., Tanner, [80] Luiten, O. J., Reynolds, M. W., Walraven,
C. E., Rolston, S. L., Phillips, W. D., Lett, J. T. M. (1996), Phys. Rev. A 53, 381.
P. D., Gould, P. L. (1990), Phys. Rev. Lett. [81] Huang, K. (1963), Statistical Mechanics. New
65, 33–36. York: Wiley.
[71] Birkl, G., Gatzke, M., Deutsch, I. H., Rol- [82] Landau, L. D., Lifshitz, E. M. (1958), Quan-
ston, S. L., Phillips, W. D. (1995), Phys. Rev. tum Mechanics (Non-Relativistic Theory).
Lett. 75, 2823–2826. Oxford: Pergamon Press.
[72] Masuhara, N., Doyle, J. M., Sandberg, J. C.,
Kleppner, D., Greytak, T. J., Hess, H. F.,
View publication stats

AltSpanisch Lied

Uploaded by

Copyright:

Available Formats

AltSpanisch Lied

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

AltSpanisch Lied

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Laser Cooling and Trapping

Article in Journal of the Optical Society of America B · May 2003

Peter van der Straten

The user has requested enhancement of the downloaded file.

Laser Cooling and Trapping of Neutral Atoms

Peter van der Straten

Laser Cooling and Trapping of Neutral Atoms 3

1 assignment of a thermodynamic ‘‘temper-

4 Laser Cooling and Trapping of Neutral Atoms

Laser Cooling and Trapping of Neutral Atoms 5

2.1 the atoms). The force is thus velocity-

6 Laser Cooling and Trapping of Neutral Atoms

Laser Cooling and Trapping of Neutral Atoms 7

8 Laser Cooling and Trapping of Neutral Atoms

Laser Cooling and Trapping of Neutral Atoms 9

10 Laser Cooling and Trapping of Neutral Atoms

done by changing ωD through the angular t = 10 − 50 µs with an acoustic optical

For alkali atoms such as sodium, a

additional beams labeled pump and probe

Laser Cooling and Trapping of Neutral Atoms 11

12 Laser Cooling and Trapping of Neutral Atoms

Fig. 5 Scheme for optical brightening of an atomic beam. First,

Laser Cooling and Trapping of Neutral Atoms 13

an atomic beam is an example of a wander away from the center experience

14 Laser Cooling and Trapping of Neutral Atoms

Laser Cooling and Trapping of Neutral Atoms 15

and produce a standing wave. When px

16 Laser Cooling and Trapping of Neutral Atoms

light shifts Eg of the ground magnetic

Laser Cooling and Trapping of Neutral Atoms 17

The cooling process described above is a few times Er ≡ kB Tr /2 = Mvr2 /2 is intu-

3.3.4 The Limits of Sisyphus Laser Cooling

18 Laser Cooling and Trapping of Neutral Atoms

Laser Cooling and Trapping of Neutral Atoms 19

Eqs. (4) and (5), and the atomic behavior is 0.4

similar to that in a DC ﬁeld. It is impor- 0.2

focused light because it is not a static ﬁeld. – 1000

20 Laser Cooling and Trapping of Neutral Atoms

Laser Cooling and Trapping of Neutral Atoms 21

Fig. 11 The Ioffe trap has four straight current

22 Laser Cooling and Trapping of Neutral Atoms

Laser Cooling and Trapping of Neutral Atoms 23

many thousands of orbits corresponding separation of the rapid precession from

24 Laser Cooling and Trapping of Neutral Atoms

1987 [46]. The operation of an MOT de- Energy

the counterpropagating laser beams or on s+ beam s− beam

a very high degree of polarization. Mg = 0

Laser Cooling and Trapping of Neutral Atoms 25

26 Laser Cooling and Trapping of Neutral Atoms

Laser Cooling and Trapping of Neutral Atoms 27

28 Laser Cooling and Trapping of Neutral Atoms

these lattices can be made essentially free

Laser Cooling and Trapping of Neutral Atoms 29

Fluorescence (arb. units)

2n. The real beneﬁt of this scheme is 5.3

30 Laser Cooling and Trapping of Neutral Atoms

Laser Cooling and Trapping of Neutral Atoms 31

32 Laser Cooling and Trapping of Neutral Atoms