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

net/publication/315764510

Design of Magneto-Optical Traps for Additive Manufacture by 3D Printing

Article · April 2017

CITATIONS READS
5 679

9 authors, including:

Yijia Zhou Nathan Welch

University of Nottingham University of Nottingham
6 PUBLICATIONS   15 CITATIONS    5 PUBLICATIONS   15 CITATIONS   

SEE PROFILE SEE PROFILE

Ricky Wildman Christopher John Tuck

University of Nottingham University of Nottingham
163 PUBLICATIONS   2,908 CITATIONS    164 PUBLICATIONS   4,664 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

OPM MEG simulations View project

Management of Rapid Manufacturing View project

All content following this page was uploaded by T. M. Fromhold on 14 April 2017.

The user has requested enhancement of the downloaded file.

 Design of Magneto-Optical Traps for Additive Manufacture by 3D Printing

Yijia Zhou,1, 2 Nathan Welch,1 Rosemary Crawford,1 Fedja Oručević,1, 3 Feiran Wang,1
Peter Krüger,1, 3 Ricky Wildman,4 Christopher Tuck,4 and T. Mark Fromhold1
1
School of Physics and Astronomy, University of Nottingham, Nottingham, NG7 2RD, UK
2
Department of Physics, Fudan University, Shanghai, 200433, P.R.China
3
Present address: Department of Physics and Astronomy,
University of Sussex, Falmer Campus, Brighton, BN1 9QH, UK
4
Faculty of Engineering, University of Nottingham, Nottingham, NG7 2RD, UK
A key element in the study of cold atoms, and their use in emerging quantum technologies, is
trapping the atoms in an ultra-high vacuum (UHV) chamber. Many methods have been used to trap
atoms including atom chips and magneto-optical traps (MOTs). However, the bulky apparatus, and
current-carrying coils, used so far in most MOTs restrict the reduction of power and physical size,
as required for quantum technology applications. The advent of 3D printing technology now offers
arXiv:1704.00430v1 [quant-ph] 3 Apr 2017

a new route to making MOTs with current paths that can be freely shaped and shrunk to several
centimetres, thereby helping to reduce the power consumption and simplify the production of the
MOT itself. In this paper, we present designs for 3D printed MOTs and analyse their performance
by using COMSOL simulations. We predict that the 3D-printed conductors can create magnetic
fields with gradients around 15 G/cm and passing through zero, as required for atom trapping, with
Joule heating as low as 0.2 W.

I. INTRODUCTION nents. Due to the scalable nature of such Additive Man-

ufacturing techniques, from metre down to sub-micron
Many devices have been developed to trap neutral scales, they may also be transferable to atom chip struc-
atoms in an ultra-high vacuum (UHV) environment by tures [9–11]. We present three designs for MOTs suitable
using light, magnetic and electric fields. One of these, the for production via 3D printing. We start from curved
magneto-optical trap (MOT), comprises laser light and a and twisting metal bars, which illustrate the principle
magnetic field [1–5] and operates by exploiting the atom- of shaping the current paths in order to produce the re-
photon interaction, which depends on both the position quired magnetic field. Guided by this, we then consider
and momentum of the atoms. Atoms can be excited by designs with successively more compact structures and
absorbing a photon and then spontaneously emitting an- volume-filling conductors designed to achieve lower power
other. The emission is usually isotropic, and thus does dissipation and more accurate field landscapes. For most
not change the mean momentum of the atom cloud. How- MOTs, the magnetic field gradients along the x, y and z
ever, the absorption process is anisotropic and can thus directions are chosen to be in the ratio of approximately
slow the atoms if the frequency and polarization of the 1:1:-2 in order to produce an atom cloud with similar di-
light is chosen correctly. The external magnetic field con- mensions in all three directions. The most efficient gra-
trols the internal states of the atoms and the absorption dients for Rb-87 cooling are around 15 G/cm [12, 13]
rate. For the MOT to work, we need the magnitude of and in our design we obtain 9.0, 9.2 and -17.6 G/cm re-
the magnetic field to increase with increasing distance spectively. A major advantage of the 3D-printed traps
from the trap centre. is that we can reduce their size to be the same order as
The traditional experimental MOT set-up includes the laser beam diameter, thereby enabling the current to
anti-Helmholtz coils outside the UHV chamber. The flow nearer to the trapped atom cloud. Thus, the heat
magnetic field is highly inhomogeneous and, in each di- dissipation in the conductors can be reduced to 0.2 W
rection, its strength varies approximately linearly around in our simulations. As noted above, such miniaturiza-
the trap centre. Another well-known trap configuration is tion and low power are essential for the development of
the Ioffe-Pritchard trap comprising four horizontal wires commercial applications in quantum technologies such as
[6, 7], which provide a quadrupole magnetic field in the sensors and clocks.
vertical plane, and two vertical rings carrying parallel
currents, which trap the atoms horizontally.
Traditional large (10s of cm) laboratory-based MOTs II. PRINCIPLES OF MAGNETO-OPTICAL
often have hand-made coils positioned outside the UHV TRAP OPERATION
chamber. It is laborious to make such coils, which inhibit
the miniaturisation of the apparatus and consume high MOTs enable many atoms, between 107 ∼ 1010 , to be
power of ∼ 20 W. cooled to µK temperatures. A unique feature of their
In this paper, we highlight the potential of 3D print- operation is that they can selectively cool atoms accord-
ing technology [8], to make current-carrying components ing to their position. When the atoms are not at the
for the MOTs, whose geometries can be shaped freely in trap centre, the light beams provide a force that push
order to reduce power and fit with surrounding compo- the atoms towards the centre. In the simplest model for
 2

quadrupole magnetic field at the trap centre.

Traditionally, a pair of anti-Helmholtz coils generates
the required field profile. Anti-Helmholtz coils comprise
a pair of parallel coils with current flowing in opposite
directions as shown in Fig.2(a)(left). The magnetic field
generated by each coil cancels at the trap centre, whilst
the field gradients add. Maxwell’s equations require that
∂Bx /∂x + ∂By /∂y + ∂Bz /∂z = 0 at the trap centre.
For the particular symmetry of anti-Helmholtz coils, this
relation yields ∂Bx /∂x = ∂By /∂y = −(1/2)∂Bz /∂z.
The conventional way of manufacturing anti-
Helmholtz coils is to use multiple turns of copper
wires held outside the UHV chamber. This makes the
FIG. 1. Schematic diagram showing the operation of a MOT coils far from the atom cloud and requires huge currents
for atomic motion along the x-axis (a similar picture holds (∼ 100 A) in the wires. To utilize the space between the
along the y and z directions): The strength of the magnetic coils, and thus achieve more power-efficient trapping,
field is proportional to x, and its direction is shown by the
Ioffe and Pritchard independently designed a magnetic
blue arrows. The Zeeman effect splits the energy of atoms by
an amount proportional to mF (values shown right). When trap to generate the field whose x, y components are
an s-state atom moves from x = 0 to the left, and absorbs a quadrupole as shown in Fig.2(a)(right) [6, 7]. The four
photon (represented by the red arrows) with energy slightly parallel wires generate the quadrupole magnetic field
below ∆E, it will be excited to the mF = 1 state. To maintain along the x and y directions, and an additional pair
momentum conservation, the atom is slowed by photon ab- of coils provides the magnetic field for z-directional
sorption. The de-excitation process emits photons randomly trapping.
in all directions so that, on average, the atom slows. A sim- To further reduce the heat dissipation resulting from
ilar deceleration occur for right-moving atoms that absorb the currents, we have investigated compact conductor ge-
left-moving photons (black arrow labelled σ − ). ometries, which wrap closely around the atomic cloud,
and are suitable for manufacturing by using 3D printing
technology. Our designs comprise either 2 or 4 separate
conducting parts and 4 current terminals. We consider
this restoring force, the atoms are taken to have only topologies inspired by both the Ioffe-Pritchard trap and
two internal states: a ground s-state and an excited p-
anti-Helmholtz coils.
state. The p-state splits into three substates in a mag-
netic field via the Zeeman effect. In MOTs, the mag-
netic field strength is proportional to the distance along
the x, y and z-axes from the trap centre. The excited III. TWISTED METAL BARS
state energy consequently splits as the atoms move away
from the trap centre, as shown in Fig.1. The frequency To illustrate how MOTs can also be made by produc-
of the laser beams applied along the axes are red de- ing curvilinear current paths that are neither straight
tuned to be slightly lower than the ground to excited nor circular, we now consider the use of twisted metal
state transition frequency. Due to the small energy differ- bars, each with a circular cross-section, to guide the
ence between the Zeeman-split energy levels, absorption current and generate the magnetic field for the MOT.
of photons preferentially occurs as the atoms move away Inspired by the Ioffe-Pritchard trap, we first analyse a
from the trap centre, which slows the atoms. The ex- twisted cage consisting of four twisted metal bars shown
cited atoms emit photons spontaneously. This emission in Fig.2(b). The shape of the bars is guided by para-
is isotropic and therefore does not change the mean mo- metric equations [14]. The current in adjacent bars flows
mentum of the atom cloud. For convenience we consider in opposite directions (white arrows). The vertical cur-
6 laser beams for cooling, making use of the conservation rent component contributes to the horizontal magnetic
of angular momentum in photon transitions. For exam- field, whilst the twisted current paths generate the ver-
ple, suppose one atom moves leftwards, as shown in Fig.1, tical field. Together, the four bars make the field van-
where x < 0 and the magnetic field points to the right. If ish at the trap centre and increase linearly around it.
a red-detuned right-going photon has σ + polarization, it This scheme can provide the magnetic field required for
will excite the atom to the mF = 1 state. Left-going σ − a MOT. The height of the trap in Fig.2(b) is 110 mm,
photons are also red-detuned, but the frequency needed and the outer width 55 mm. The copper bars are 10
to excite the atom to the mF = −1 state is higher, so mm in diameter. The holes in the bottom and top in-
such photons will be absorbed with much smaller prob- sulating mounts (white), and the space between wires
ability. Thus one pair of laser beams can slow atoms in allow access of 15 mm diameter orthogonal laser beams.
one direction. Three pairs of laser beams will slow and The trap is designed using 3D CAD software Solidworks
reduce the temperature of the atom cloud as a whole. and the physical fields are simulated by COMSOL Multi-
For this mechanism to work, the current must produce a physics. We calculated the current density and magnetic
 3

those, 1:1:-2, typically used in MOTs and the current re-

quired to generate these gradients is high. Consequently,
in the next section we consider a structure that also em-
ploys curvilinear current flows to generate the trapping
fields, but has improved field and power performance.

IV. COMPACT CONDUCTOR DESIGN

FIG. 2. (a) The traditional anti-Helmholtz coils (left) and

Ioffe-Pritchard trap. The white arrows show the current direc-
tions. (b) A MOT design using twisted metal bars to achieve
3D trapping. The white discs are dielectric mounts and the
purple parts are the bars. The white arrows show the current
direction in each bar. (c) Calculation of the magnetic field
magnitude, B, versus position along the x, y and z directions
(black, blue and red curves respectively). The gradients near
the trap centre along the x, y and z directions are 15.1, 6.5 FIG. 3. (a) Design for a 3D-printed version of the current
and -22.1 G/cm (black, blue and red dashed lines). topology shown in Fig.2(b). The four conducting parts are
electrically isolated, and the gaps between them can be filled
with dielectric (not shown). (b) An exploded view of the trap.
field profile with fixed voltages at the bar ends. The so- Red and blue arrows indicate the current input and outflow.
lution domain is a cube three times larger than the trap.
In Fig.2, we plot the field along three axes through the
trap centre where the field gradients have approximately We now present an improved conductor design, shown
constant values of 15.1, 6.5 and -22.1 G/cm along the x, y in Fig.3, which maintains the topology of the current
and z-axes respectively. paths shown in Fig.2 but offers superior performance. In
The current needed to make the trap is 100 A and the this case, the total height of the trap is 45 mm, and its
total electric power dissipation via Joule heating is 14.9 outer width is 24 mm. The holes for the laser beam access
W for copper. Although illustrative, this is currently are also 15 mm in diameter. The minimum gap between
not the best design we can produce. There is still a lot the four conductors is 0.5 mm, which can be adjusted
of unfilled space between the conductors and the trap according to the material stability in vacuum. The design
centre, making the trap power consumption unecessarily requires four pairs of voltage contacts to run the trap and
high. We found that a high-performance MOT should control the current in each conducting part. Compared
follow several key design criteria: to the twisted bar design in Fig.2(b), this structure better
satisfies the three design criteria specified in Section III.
• Conductors should wrap the optical molasses region It is smaller, makes better use of space around the atom
tightly and fill as much space as possible around it; cloud, and all the current flows contribute positively to
the desired field profile.
• The trap conductors should be as compact as pos- The current in the conductors can be divided approx-
sible; imately into vertical and horizontal components, with
• Each bar/conducting volume should always wind in the former creating the magnetic field in the xz-plane
the same direction either clockwise or anticlockwise and the latter generating the field along the z-axis. The
to maximize its contribution to the magnetic field. currents in adjacent conductors are anti-parallel at the
contacts and, at the trap centre, provide a linear mag-
The twisted bar design in Fig.2(b) does not satisfy all of netic gradient along the x, y and z-axes. To change the
the above criteria. It is large and the small-area twisted gradient ratio in the three directions, one may easily ad-
conductors are always winding in the same direction, just the length of the conducting prongs. To produce the
which decreases the performance of the device. In ad- field gradients close to the 1:1:-2 ratio typically used for
dition, the field gradient ratios are slightly different from MOTs, we choose the height of the device to be larger
 4

techniques. In the next section, we consider further sim-

plification of the design to remove this issue by reducing
the number of separate conductors.

V. TWO-PIECE CONDUCTOR DESIGN

FIG. 4. (a) Current density distribution calculated for the

MOT shown in Fig.3, with each conductor carrying 40 A. (b)
Calculation of the magnetic field magnitude versus position
along the x, y and z-axes (black, blue and red solid lines). FIG. 5. (a)A 3D-printed MOT design with conductors that
The gradients along the x, y and z-axes at the trap centre have a cylindrical outer curved surface. The shape is inspired
are 11.5, 11.9 and -22.5 G/cm respectively (black, blue and by the Ioffe-Pritchard trap. (b) An exploded view of this
red dashed lines). (c-e) The magnetic field magnitude in the model showing the two conductors. The pink lines and ar-
xz, xy and zy-planes through the centre of the MOT. In each rows show the directions of the current. Gaps between the
plane, there is a minimum at the centre. The gaps between the conductors appear white.
four conductors are just large enough for a 1.5 cm diameter
laser beam to pass through.
Suppose that we produce three orthogonal laser-access
holes, with circular cross sections, in a cylindrical con-
than its width. To further reduce the resistance and ductor and then separate that conductor into two nest-
power consumption, one may further increase the thick- ing parts as shown in Fig.5. The current will flow in and
ness of the volumetric conductors. out of the straight arms through a roughly circular path
We designed the trap in Solidworks and simulated it in in between. The parallel arms will provide a quadrupole
COMSOL. In Fig.4(a), we show a colour map of the cur- field in the trap centre and the approximately circular
rent density on the surface of the conductors. In Fig.4(b), current will create the field pointing towards the trap
we show the corresponding magnetic field components centre along the axial directions. This geometry takes
Bx , By , Bz along the x, y and z-axes and linear fits at advantage of filling almost the whole volume surround-
the trap centre. The field variations across the xz, xy and ing the laser beams with conductor, thereby reducing the
zy-planes through the trap centre are shown in Fig.4(c- resistance of the current paths. The ratio of the thickness
e), where the colour scale shows the field magnitude. of the arms of the conductor to its outer ring diameter is
In general, the current density is highest at sharp bends important to ensure that the gradients are constant and
in the conductors. To avoid current hot spots, we there- close to the ratio 1:1:-2; typically we achieved 1.0:1.0:-1.9.
fore fill the bend regions with curving surfaces, which is Our design in Fig.5 permits the required magnetic field
possible using 3D printing. The current on the inner con- to be generated with typically 25 A in each conductor.
ductor surfaces is higher than at the outer ones, which The trap in Fig.5 is 38 mm tall, and the outer diameter
helps to increase the magnetic field in the trap, thus re- is 26 mm. The arms are 3.1 mm wide at their narrow-
ducing the current and power consumption required. In est point, i.e., the upper and lower prongs in Fig.5(b).
Fig.4(b), the linear fitted slopes near the centre along The three holes for laser beams are 15 mm in diameter.
the x, y and z-axes are 11.5, 11.9 and -22.5 G/cm re- The dimensions are optimized with respect to power con-
spectively. The total Joule heat in the trap is 0.4 W for sumption. The white stripes in Fig.5 are dielectric mate-
copper conductors. rial to insulate the two conductors, or simply a vacuum
A shortcoming of this design is that we need to print gap if the positioning and clamping of the conductors is
four parts (Fig.3) and either fix them separately, or print accurate enough. Currently, we choose a gap of 0.5 mm.
dielectric material between each part to maintain the in- Despite possible capacitance effects, the gap should be
tegrity of the structure. Such an arrangement is possible, as thin as possible to minimize resistance and power dis-
but still a challenge for current Additive Manufacturing sipation. The approximate current paths are shown by
 5

tion in the conductors.

To keep the ratio of the field gradients to be 1:1:-2
along the three directions, a trap of the form in Fig.5
is close to ideal. Elongating or thickening the trap will
allow this ratio to be tuned as required for the other
MOT geometries.
MOTs usually work under UHV conditions, and the
trapping process is almost adiabatic. Consequently, heat
can only be dissipated through the ends of the arms.
The heat transfer is proportional to the transfer area and
temperature difference. Considering the total power dis-
sipation to be ∼ 0.2 W for copper, and the transfer area
to be ∼ 4 × 10−5 m2 , in order to maintain the tempera-
ture increase below an experimentally-tolerable limit of
200 K, the heat transfer coefficient should be around 25
W/m2 K. This is practical for experimental set-ups and,
if additional external cooling can be applied to the trap,
the operating temperature can be made lower than 470
K.
FIG. 6. (a) Current density distribution calculated for the
MOT shown in Fig.5, with each conductor carrying 25 A in
total. (b) Calculation of the magnetic field magnitude versus
VI. CONDUCTOR PARAMETERS AND
position along the x, y and z-axes (black, blue and red solid
OPTIMIZATION
lines). The gradients along the x, y and z-axes at the trap
centre are 8.98, 9.20 and -17.6 G/cm respectively (black, blue
and red dashed lines). (c-e) The magnetic field magnitude in The size and density of the trapped atom clouds de-
the xz,xy and zy-planes through the centre of the MOT. In pend strongly on the magnetic field and trapping effi-
each plane, there is a minimum at the centre. The circular ciency. Suppose that the trap volume is VT = l3 , where
holes in the trap are just large enough for a 1.5 cm diameter l is an effective linear size, the cross sectional area of the
laser beam to pass through. laser beams is A, the resistance of the conductor is R,
the operating temperature of the conductors is T , and
the heat dissipation rate is ρ. Then we can consider the
the pink arrowed curves in Fig.5(b). Note that these have general scaling relations: VT ∝ A×l ∝ l3 , R ∝ l/A ∝ l−1 ,
the same overall topology as the current paths through ρ ∝ A × T ∝ l2 . To avoid the traps becoming too hot,
the four conductors shown in Fig.3. the radiation rate should equal the electric power, i.e.,
The current density in each arm is ∼ 5 A/mm2 . In I 2 R ∝ A ∝ l2 . So the current passing through the trap
the ring-like regions, by contrast, the current density is must follow I ∝ l3/2 . The Biot-Savart law tells us that
significantly smaller, below 1 A/mm2 . By design, there the magnetic field B ∝ I/l2 ∝ l−1/2 , and the gradient
is no hot spot in the current density so the trap will be ∇B ∝ l−3/2 . Therefore, to reduce the operating power,
heated uniformly. Our simulations show that the con- it is advantageous to make the trap smaller providing
tributions to the magnetic field from the currents in the that all other experimental requirements can still be sat-
arms and rings are actually equivalent, which leads to the isfied.
ratio of the gradient along the y and z directions being Reduce the power consumption can also be achieved by
the same. thickening the conductors and reducing their resistance.
The trap is designed using Solidworks and the field is However, if the conductors are too thick, the current will
calculated with COMSOL. We set the current to be 25 be far from the trap centre and generate a lower mag-
A in each conductor, and calculate the static magnetic netic field. In practice, we need the current to be as
field within the interior space. Fig.6(a) shows a colour small as possible to minimize the power. As there is, in
map of the current density on the surface of the conduc- effect, only one current turn within the bulk conductors
tors. Fig.6(b) shows the magnetic field components Bx , that we consider, and the current needs to be as large as
By , Bz along the x, y and z-axes and linear fits (dashed 25 to 100 A, a special power supply is needed. Another
lines) near the trap centre. The field gradients along the key consideration is the length of the trap, which should
x, y and z-axes are 9.0, 9.2 and -17.6 G/cm respectively be neither too long nor too short. As mentioned above,
with an uncertainty of ∼ 0.03 G/cm in each case. The to- the ring-like regions of the conductors are responsible for
tal Joule heat in the trap is ∼ 0.2 W for copper. However, generating the field variation along the axial direction,
the most common materials for 3D printing are titanium which should be comparable to that in the x and y direc-
and nickel, which have ∼ 10 times larger resistance than tions, as is the case for our model designs. We may be
copper. Thus to provide the same magnetic field using able to further improve the geometry of the conductors
these materials will require about 2 W in power dissipa- and reduce the difference between the current density in
 6

the arms and the ring-like regions in Fig.6. We can also also needs to be verified in experiment. The next step is
remove some redundant parts of the conductors in order to design and build a multi-turn 3D printed conductive
to fit them more tightly around real experimental set-ups geometry, which can help to control the spatial current
and components of, for example, quantum sensors. density distribution in a more subtle way as well as reduc-
ing the current and power dissipation and improving the
fidelity of the magnetic field landscape. It may also be
VII. CONCLUSION possible to use Additive Manufacturing to produce atom
traps with more complex geometries, including multiple
Compared to traditional MOT traps, the designs show potential minima and lattices with sub-micron scales.
good performance, especially the device in Fig.5. Accord-
ing to our simulations, we need a 25 A current source
to drive the trap to generate the magnetic field profile
needed for a MOT. The resulting power consumption in ACKNOWLEDGEMENTS
the conductors is merely 0.2 W for copper. Whether
this idea works in practice depends on the capability of This work is funded by the Engineering and Physi-
3D printing technology, which, for metals, presently fo- cal Sciences Research Council (EPSRC) through the UK
cuses on titanium, aluminum, etc., whose resistance is Quantum Technology Hub for Sensors and Metrology.
larger than copper. Whether the traps are stable, espe- Y. Zhou is sponsored by the China Scholarship Council
cially the dielectric layers between the conductive parts, (CSC).

[1] E. L. Raab, M. Prentiss, A. Cable, S. Chu, and D. E. formation of substrate release mechanisms for gas metal
Pritchard. Trapping of neutral sodium atoms with radi- arc weld metal 3-D printing. J. Mater. Process. Technol.,
ation pressure. Phys. Rev. Lett., 59(23):2631–2634, 1987. 226:50–59, 2015.
[2] S. Chu, L. Hollberg, J. E. Bjorkholm, A. Cable, and [9] J. Schmiedmayer. Guiding and trapping a neutral atom
A. Ashkin. Three-dimensional viscous confinement and on a wire. Phys. Rev. A, 52(1):R13–R16, 1995.
cooling of atoms by resonance radiation pressure. Phys. [10] R. Folman, P. Kruger, D. Cassettari, B. Hessmo,
Rev. Lett., 55(1):48–51, 1985. T. Maier, and J. Schmiedmayer. Controlling cold atoms
[3] P. J. Ungar, D. S. Weiss, E. Riis, and Steven Chu. Optical using nanofabricated surfaces: atom chips. Phys. Rev.
molasses and multilevel atoms: theory. J. Opt. Soc. Am. Lett., 84(20):4749–52, 2000.
B, 6(11):2058, 1989. [11] József Fortágh and Claus Zimmermann. Magnetic micro-
[4] C. Monroe, W. Swann, H. Robinson, and C. Wieman. traps for ultracold atoms. Rev. Mod. Phys., 79(1):235–
Very cold trapped atoms in a vapor cell. Phys. Rev. Lett., 289, 2007.
65(13):1571–1574, 1990. [12] W. Suptitz, G. Wokurka, F. Strauch, P. Kohns, and
[5] Mark Keil, Omer Amit, Shuyu Zhou, David Groswasser, W. Ertmer. Simultaneous cooling and trapping of
85
Yonathan Japha, and Ron Folman. Fifteen years of Rb and 87 Rb in a magneto-optical trap. Opt. Lett.,
cold matter on the atom chip: promise, realizations, and 19(19):1571–3, 1994.
prospects. J. Mod. Opt., 63(18):1840–1885, 2016. [13] J. Goldwin, S. B. Papp, B. DeMarco, and D. S. Jin. Two-
[6] David E. Pritchard. Cooling neutral atoms in a mag- species magneto-optical trap with 40 K and 87 Rb. Phys.
netic trap for precision spectroscopy. Phys. Rev. Lett., Rev. A, 65(2), 2002.
51(15):1336–1339, 1983. [14] In spherical
 coordinates
p {r, θ, φ}, for 0 < θ < π/4,
[7] Kristian Helmerson, Alex Martin, and David E. r(θ) = l 2 − 1 − (4θ/π − 1)2 /4, φ(θ) = π/ 3 +
π 
Pritchard. Laser cooling of magnetically trapped neu- 3e−10(θ− 3 ) ; for π/4 < θ < 3π/4, r(θ) = l/4, φ(θ) =
tral atoms. J. Opt. Soc. Am. B, 9(11):1988, 1992. 2π 
π/ 3 + 3e−10(θ− 3 ) ; for 3π/4 < θ < π, r(θ) = l 2 −
 
[8] Amberlee S. Haselhuhn, Bas Wijnen, Gerald C. Anza- π
1 − (4θ/π − 3)2 /4, φ(θ) = π/ 3 + 3e−10(θ− 3 ) . Here
p   
lone, Paul G. Sanders, and Joshua M. Pearce. In situ
l = 110 mm is the height of the trap.

View publication stats