Journal of the Korean Physical Society, Vol. 60, No. 1, January 2012, pp.

L1∼L5 Letters

Systematic Study of the Rf-induced Evaporation of 87 Rb Atoms in a Spherical

Magnetic Quadrupole Trap

Dahyun Yum, Jina Park, Wanhee Lee and Wonho Jhe∗

Department of Physics and Astronomy, Seoul National University, Seoul 151-747, Korea

(Received 28 September 2011, in final form 28 November 2011)

We trap 87 Rb atoms in a spherical magnetic quadrupole potential consisting of a pair of anti-

Helmholtz coils. The number of trapped atoms is 2.16(7) × 108 , and the lifetime is 26 seconds. The
number, the temperature and the size of the atoms in a magnetic trap are controlled by the forced rf-
evaporative cooling technique. We obtain a sub-Doppler temperature of 33.7(5) µK in these atoms
at an evaporation frequency of 4.2 MHz in the magnetic quadrupole trap. We discuss the properties
of the atoms in the spherical magnetic quadrupole trap and the results of the rf-evaporation process.

PACS numbers: 32.80.Pj, 42.50.Vk

Keywords: 87 Rb, rf-Evaporation, Spherical magnetic quadrupole trap, Cold atoms
DOI: 10.3938/jkps.60.1

I. INTRODUCTION quadrupole trap is used as the preparation stage before

loading in the optical dipole trap [11].
In this paper, we report the experimental results of
The development of trapping and cooling neutral trapping 87 Rb atoms in a spherical magnetic quadrupole
atoms in the past few decades provides a wide research trap and applying a rf-evaporation technique. After trap-
pool for atomic physics, precision measurement, con- ping in a MOT and optical pumping, 2.16(7) × 108 atoms
densed matter physics and quantum information. The are trapped by magnetic trap at a temperature of 58.2(3)
study of the dilute and cold neutral atomic system µK. We obtain the properties of the atoms in the mag-
was extended considerably by the achievement of the netic quadrupole trap by in-situ absorption imaging. Be-
magneto-optical trap (MOT) [1, 2]. Though the var- cause the absorption image profile of an atomic cloud in
ious studies have been reported for the MOT system, a magnetic trap is deformed by the Zeeman shift [12], the
it has a limitation to lower the temperature of trapped image has to be modified by considering the spatial ab-
atoms due to the scattering of the resonant photons. For sorption rate. The temperature of and the number of the
colder atomic cloud, another method of trapping atoms atoms in a magnetic quadrupole trap decreased by using
without resonant photons and a new cooling mechanism rf-evaporative cooling technique. For magnetic trapping,
are necessary. The pure magnetic trap [3,4] and the rf- the atoms have to be in a low-field-seeking state as there
evaporative cooling method [5–7] have come to the fore are no magnetic maxima. In our case, the trapped 87 Rb
for cold atomic systems. Various magnetic traps with rf- atoms are in the |F = 1, mF = −1> state. By applying
evaporation reduce the temperature of the atomic cloud a radio frequency field, the hot |F = 1, mF = −1> state
beyond the limitation of the MOT to the transition tem- atoms are transferred to the |F = 1, mF = +1> state
perature of Bose-Einstein condensation [8,9]. selectively and hot atoms are repelled from the magnetic
Currently, various studies are underway on cold atomic trap. We can control the number and the temperature
systems in an effort to investigate many types of trapping below Doppler limit of the atomic cloud in a magnetic
potentials, not only magnetic potentials but also optical trap by using the rf-evaporation technique. The evapo-
potentials, with the evaporative cooling mechanism. The ration result demonstrates the linearity of the spherical
spherical magnetic quadrupole trap is the simplest and magnetic quadrupole trap’s shape. This experimental
most basic method of creating a pure magnetic potential, work on the spherical magnetic quadrupole trap and rf-
and rf-evaporative cooling is the most widely used way to evaporative cooling will be a step forward to a more sta-
reduce the temperature of atoms in a magnetic potential ble trap without spin-flip loss for colder atoms like Bose
[6,10]. The spherical magnetic quadrupole trap provides Einstein condensation, for example, the time-averaged
the advantage of effective evaporation due to its linear orbiting potential (TOP) trap [4], the optical dipole trap
potential shape. Therefore, rf-evaporation in a magnetic [11], QUIC trap [13].

∗ E-mail: [email protected]
-L1-
-L2- Journal of the Korean Physical Society, Vol. 60, No. 1, January 2012

Fig. 1. (Color online) Schematic diagram of the experi-

mental setup. Laser beams perpendicular to the page are not
shown.

Fig. 2. (Color online) 87 Rb D2 line data and the frequen-

II. EXPERIMENTAL SETUP cies of the laser used in the experiment. ωTrap is the frequency
of the trap line, ωDep is the frequency of the depump line, and
ωRep is the frequency of the repump line.
We used a double-MOT system with differential pump-
ing to overcome a contradiction about the vacuum. If a
large number of atoms is to be trapped, sufficient atomic sisting of a piece of hollow 4 mm × 4 mm cooper wire,
gas in the background is needed. However, If the atomic and it is water cooled. The coils both have 30 turns (6
cloud in a trap is to be kept cold, it is necessary to scat- turns and 5 layers). The inner diameter of the coil is 8.5
ter as little as possible with other thermal gas in the cm, and the outer diameter is 15 cm. The separation
vacuum, meaning that a low vacuum pressure is impor- between the two coils is 6.5 cm. The AH coil pair at
tant. For this reason, we utilize two chambers connected the 2nd glass chamber is powered by a 15 kW DC power
by a pipe 1 cm in diameter and 40 cm in length. The supply (Lambda EMI) that is controlled by a RS232 con-
pressure of the 1st chamber is below 1 × 10−9 Torr, and trol and is switched by an IGBT (insulated gate bipolar
that of the 2nd chamber is below 1 × 10−11 Torr, both transistor). The maximum magnetic field gradient is 270
values are measurement limits of the ion pump gauge. G/cm at 400 A in the z direction (135 G/cm in the ra-
The 2nd chamber is made of Pyrex glass to prevent the dial direction). Three power resistors (1 Ω, 100 W) are
generation of undesired eddy currents. A schematic di- connected in parallel at the switching circuit to consume
agram of the chamber and the laser setup is shown at the remaining current in the coil when it is switched off.
Fig. 1. To protect the IGBT switch from high-power electronic
We use three different laser frequencies, named here as shocks, a varistor (TNR 20E391K) is used. A diagram
trap line, repump line and depump line. The laser fre- of the high-power DC supply switching circuit is shown
quency of the trap line is resonant on the F = 2 to the F in Fig. 3.
= 3 cycling transition of the 87 Rb D2 line. This is used The evaporation coil pair has Helmholtz geometry
for the trap, the push and the probe laser. The repump with two turns, and it has a diameter of 6.5 cm. The
line is resonant on the F = 1 to the F = 2 transition, rf current is created by a HP3658B rf signal generator
and the depump line is resonant on the F = 2 to the F and is amplified by an amplifier (MINI Circuit, ZHL-
= 2 transition. The 87 Rb D2 line and all laser frequen- 5W-1). We obtain all images by fluorescent imaging or
cies used are shown at Fig. 2. The trap lasers used for absorption imaging using a CCD camera (PCO, Pixel Fly
the MOT at the 1st and the 2nd chambers are 2Γ red XS). All machines and circuits are controlled by a FPGA
detuned from the trap line. (Γ is the natural linewidth (fieed programable gate array, NI PCI-7833R) and an
of the 87 Rb atom; its value is 5.9 MHz.) The intensity of AO/DO board (NI PCI-6713). The FPGA has an inner
the 2nd trap laser is 2 mW/cm2 . The push laser 1Γ red clock running at 40 MHz that is used as a time trigger
detuned from the trap line is focused 3 mm above and so that it is not interrupted by the multi-tasking process
1 cm in front of the 2nd MOT’s center to avoid scatter- of the computer.
ing with the trapped atoms in the 2nd glass cell. The
probe laser is resonant on the trap line at an intensity of
0.5 mW/cm2 . The intensity and the frequency of all of
the lasers are controlled by using an acousto optic mod- III. RESULTS AND DISCUSSION
ulator (AOM) and a mechanical shutter.
The spherical magnetic quadrupole potential used for The magneto-optically trapped atoms in the 1st cham-
the MOT and the magnetic trap in the 2nd chamber is ber are transferred to the 2nd glass chamber by the push
created by using an anti-Helmholtz (AH) coil pair con- laser. We recapture the transferred atoms by MOT. The
 87
Systematic Study of the Rf-induced Evaporation of Rb Atoms · · · – Dahyun Yum et al. -L3-

Fig. 3. Schematic diagram of the switching circuit for con-

trolling the AH coil pair. R is the three 1 Ω, 100 W resistances
in parallel.

magnetic field gradient is 15 G/cm, and the number of

trapped atoms is 1 × 109 after 5 s of pushing. We obtain Fig. 4. (Color online) (a) Absorption image of the atomic
the atom number by the fluorescence signal using a pho- cloud in a spherical magnetic quadrupole trap. The scale
todiode and the absorption image using the CCD cam- bar is 500 µm. (b),(c) The vertical(horizontal) profile of the
era. The absorption image is calibrated with the fluores- absorption image in (a). The line is a Gaussian fitting curve,
cence result. The magneto optically trapped atoms are and the y axis is the optical depth. (d) Life time of atoms in
compressed by increasing the magnetic field gradient to the trap. The line is an exponential fitting curve. From the
exponential fitting, the life time is 26 seconds.
70 G/cm within 10 ms. At the compressed MOT stage,
the detuning of the trap laser is increased to 3Γ, and the
intensity is lowered from 2 mW/cm2 to 1.4 mW/cm2 .
After compression, we turn off the repump laser and absorption image and its profile are shown in Figs. 4(a)
turn on the depump laser for optical pumping so that – (c).
the atoms are in the F = 1 ground state in a magnetic There is no fluorescent light when the atoms are in a
trap. We change the detuning of the trap laser to 4.5Γ pure magnetic trap, so the absorption imaging method
for efficient optical pumping. By 1 ms optical pumping is used to detect the atomic cloud. The atoms expo-
process, the ground state is changed from the F = 2 state nentially attenuate the intensity of the probe laser as it
to the F = 1 state. We verify whether the optical pump- passes through the cloud via the optical depth (OD) in
ing process operates well or not by observing a shadow a manner known as Beer’s rule;
image of the atoms when the probe laser is illuminated.
If the atoms are in the F = 2 state, the absorption signal Iabs = I0 e−OD , (1)
is detected by the CCD camera whereas atoms in the F Iatom − Ibg
OD = ln[ ]. (2)
= 1 state are not observed by the probe laser when the Iprobe − Ibg
optical pumping works well.
After optical pumping, the atoms are trapped by the We take three images to observe the trapped atoms. The
spherical magnetic quadrupole trap potential. There are Iatom is an absorption image when the atoms exist. The
no field maxima in the magnetic trap, only the weak- I0 is an image of the probe laser without trapped atoms,
field-seeking state atoms are trapped magnetically. In and the Ibg is a background image with neither atoms
the D2 transition of the 87 Rb atom, there are three nor a probe laser for noise correction. The trapped Rb
possible states in which to trap magnetically, the |F = atoms are in the F = 1 state, but the probe laser is
2, mF = +2>, |F = 2, mF = +1> and |F = 1, mF = resonant on the F = 2 to the F = 3 transition. We
−1>. The magnetic sub-level of the atoms in the F = 1 turn on the repump laser with the probe laser during
state is chosen naturally when the magnetic quadrupole the imaging process. We obtain the density distribution
trap is turned on. The other ground-state atoms, mF = of the atoms when the sample is in the magnetic trap.
+1 or 0, are repelled or fall from the trap. In spite of The in-situ imaging method has the advantage of being
losing 2/3 of the atoms in magnetic trap, the optical able to observe a relatively small atomic cloud compared
pumping process from F = 2 to F = 1 is a convenient to the TOF (time of flight) method detecting atoms in
way to select the ground state. ballistic expansion. By the Zeeman effect, the absorp-
The magnetic field created by the anti-Helmholtz coil tion rates of the probe laser at various positions of the
pair has the form B = B  (x, y, −2z), and the potential is magnetic trap are different due to the inhomogeneous
Uquad (r) = µB mF B  r, where µB is the Bohr magneton, magnetic field strength. The detuning change has to be
mF is the projection of the total angular momentum of considered when analyzing the image obtained by in-situ

the atom on the magnetic field axis, B is the magnetic imaging. The column density at each pixel in the image
field gradient and r = x + y + (2z)2 . A typical
2 2 is obtained by the simple equation of OD/A, where the
-L4- Journal of the Korean Physical Society, Vol. 60, No. 1, January 2012

Fig. 5. Graphs of the (a) size, (b) number, and (c) temperature of atoms in a spherical magnetic quadrupole trap vs. rf
evaporation frequency. The magnetic field gradient is 100 G/cm for the z direction. In (a), solid line is the calculated size versus
the transition frequency from mF = −1 to mF = +1 in the magnetic trap.

absorption cross-section A is expressed as follows; to deal with it and indicates that the magnetic trap is
  2 stable.
Λ 3λ 1 After the magnetic trap, we apply a radio frequency
A= , (3)
2 2π δ(r)2 field, which is the transition frequency from the magnetic
1+4 2 sub-level, |F = 1, mF = −1> to |F = 1, mF = +1>.
Γ
The atom is coupled selectively with the rf field, and
where λ is the wavelength of the transition, the detuning the atoms in the mF = +1 level are evaporated out of
δ(r) is a function of the distance r from the trap center, the trap. We observe that the atoms in the magnetic
and the branching ratio Λ for the F = 2 to F = 3 tran- spherical quadrupole trap after an evaporative cooling
sition at the linear polarization beam is 7/3 in our case. process lasting 3 s for several radio frequencies. The
Though the distribution of atoms in the spherical mag- results pertaining to the size, number and temperature
netic quadrupole trap is not Gaussian, the size of the of the atomic cloud at each frequency are shown in Fig. 5.
atoms is obtained by a Gaussian fitting for calculational The spherical magnetic quadrupole trap has a linear
convenience, and the profile is not far from a Gaussian potential shape, and the energy difference between mF =
profile [14]. The hot atoms tend to be on the edger of −1 and mF = +1 is also linear at each position. This
the potential whereas the cold atoms are near the center. is well explained by the magnetic Zeeman shift, and the
The temperature of the atoms in the spherical magnetic energy shift of the magnetic sublevel at low fields is [15,
quadrupole trap could be obtained from the size of the 16]
atomic cloud, and the equation is
∆E = gµB mF B  r. (5)
2 µB g 
T = B σF W HM , (4) The result of the atomic cloud size is shown in the
5 kb r
Fig. 5(a). The solid line in Fig. 5(a) is the size of the
where µB is the Bohr magneton, g is the landé g-factor, atoms in the magnetic trap calculated from the transition
kB is Boltzmann’s constant, Br is the radial magnetic energy between the mF = −1 state and the mF = +1
field gradient and σF W HM is the radial full width at state for a magnetic field gradient of 100 G/cm in the
half maximum size of the cloud. The number of atoms experimental condition. We surmise that the difference
in Fig. 4(a) is 2.16(7) × 108 , and the temperature is between the experimental result and the calculated value
58(3) µK. The image is taken after holding the atoms of the size is caused by the power broadening effect of
for 3 s in the spherical magnetic quadrupole trap with the rf-field [7]. The size of the atomic cloud for vari-
100 G/cm magnetic field gradient in the z direction. Fig- ous evaporation frequencies demonstrate the linearity of
ure 4(b) shows the profile of the Fig. 4(a) along the ver- the magnetic trap, showing that the magnetic spherical
tical direction (the z direction), and the solid line is the quadrupole trap is stable and that the evaporation pro-
Gaussian fitting. Figure 4(c) is the horizontal profile (ra- cess works well.
dial direction) of the atomic cloud in the trap. We can The number and the temperature of the atomic cloud
detect the absorption signal of atoms in the trap after at various evaporation frequencies are shown in Fig. 5(b)
60 s of magnetic trapping and the lifetime is obtained and (c). The number of atoms is shown on a log scale
26 s (Fig. 4(d)). The losses of atoms are caused by the for presentational convenience, and the line is a linear
scattering of the background gas, the Majorana spin-flip fitting as a guide line. It is difficult to analyze an absorp-
transitions at the magnetic field is zero near the center tion image at frequencies below 4.2 MHz even when the
of the trap and the fluctuation of the current of the AH atomic shadow can be distinguished by eye. We deter-
coil. The 26 s life time of the atoms in the trap is easy mined that the temperature of the atoms in the magnetic
 87
Systematic Study of the Rf-induced Evaporation of Rb Atoms · · · – Dahyun Yum et al. -L5-

trap was T = 33.7(5) µK at 4.2 MHz. The number and REFERENCES

the temperature of the atomic cloud could be controlled
by rf-evaporation technique below the Doppler limit in a [1] E. L. Raab, M. Prentiss, A. Cable, S. Chu and D. E.
spherical magnetic quadrupole trap. Pritchard, Phys. Rev. Lett. 59, 2631 (1987).
[2] W. D. Phillips, P. L. Gould and P. D. Lett, Science 239,
877 (1987).
[3] A. L. Migdall, J. V. Prodan, W. D. Phillips, T. H. Berge-
IV. CONCLUSIONS
man and H. J. Metcalf, Phys. Rev. Lett. 54, 2596 (1985).
[4] W. Petrich, M. H. Anderson, J. R. Ensher and E. A.
We trapped 2.16(7) × 108 87 Rb atoms in a spheri- Cornell, Phys. Rev. Lett. 74, 3352 (1995).
cal magnetic quadrupole trap, showing that the lifetime [5] N. Masuhara, J. M. Doyle, J. C. Sandberg, D. Kleppner
was 26 s, which is enough time to study the trapped and T. J. Greytak, Phys. Rev. Lett. 61, 935 (1988).
cold atoms and to conduct experiments on them. The [6] K. B. Davis, M. Mewes, M. A. Joffe, M. R. Andrews and
W. Ketterle, Phys. Rev. Lett. 74, 5202 (1995).
properties of the atomic cloud in a spherical magnetic
[7] W. Ketterle and N. J. Van Druten, Adv. At. Mol. Opt.
quadrupole trap were obtained by an in-situ absorption Phys. 37, 181 (1996).
imaging method. After the atoms had been trapped [8] M. H. Anderson, J. R. Ensher, M. R. Matthews, C. E.
magnetically, we applied a rf-field resonance on the Wieman and E. A. Cornell, Science 269, 198 (1995).
mF = −1 state to the mF = +1 state transition. We [9] K. B. Davis, M -O. Mewes, M. R. Andrews, M. J. Van
controlled the size, number and temperature of the atoms Druten, D. S. Durfee, D. M. Kum and W. Ketterle, Phys.
in the magnetic trap below the Doppler limit by using Rev. Lett. 75, 3969 (1995).
the rf-evaporation technique. The results show that the [10] K. B. Davis, M -O. Mewes and W. Ketterle, Appl. Phys.
spherical magnetic quadrupole trap is a stable linear trap B 60, 155 (1995).
shape. We expect that well-controlled cold atoms in a [11] Y -J. Lin, A. R. Perry, R. L. Compton, I. B. Spielman,
spherical magnetic quadrupole trap can be used to lower and J. V. Porto, Phys. Rev. A 79, 063631 (2009).
[12] O. J. Luiten, H. G. C. Werij, I. D. Setija, M. W.
the temperature even more by, for example, an optical
Reynolds, T. W. Hijmans and J. T. M. Walraven, Phys.
dipole trap to prevent loss of atoms due to the spin flip Rev. Lett. 70, 544 (1993).
transitions. [13] T. Esslinger, I. Bloch and T. W. Hänsch, Phys. Rev. A
58, R2664 (1998).
[14] H. J. Lewandowski, D. M. Harber, D. L. Whitaker and
E. A. Cornell, J. Low Temp. Phys. 132, 309 (2003).
ACKNOWLEDGMENTS
[15] H. J. Metcalf and P. van der Straten, Evaporative Cooling
of Trapped Atoms (Springer, New York, 1999).
This work was supported by the Acceleration Research [16] C. J. Pethick and H. Smith, Bose-Einstein Condensa-
Program of the Korean Ministry of Science and Technol- tion in Dilute Gases (Cambridge University Press, Cam-
ogy. bridge, 2002).