We experimentally demonstrate a novel scheme to simultaneously confine two atomic species of 87Rb and 133Cs with adjustable spatial separation by a controllable double-well magneto-optic trap. Using a single-loop wire and a magnetic bias field, the two clouds, each containing more than 1×106 atoms, are spatially separated above and below the wire center of the double-well MOT. The cloud interdistance can be controlled by independently varying the wire current and external bias field. This allows to load the double-well magnetic trap, and to study the dynamics of cold collisions between two-species atoms.
©2008 Optical Society of America
A magneto-optical trap (MOT) is now a powerful and popular tabletop machine to produce cold atoms since the pioneer work by Chu et al. . So far, many MOT configurations have been proposed and realized to produce spin-polarized sample, high flux cold atomic beam, 2D surface MOT, and atom trap using a single input beam, etc . One important achievement is to use MOT atoms as the first stage to produce Bose-Einstein condensates of dilute gas .
Ultracold gases containing two different atomic species are expected to show interesting properties in quantum statistics and provide the perfect systems for the study of cold collisions. There have had intensive studies along this research direction for the past decade . This leads to the production of ultracold heteronuclear molecules  and reach simultaneous quantum degeneracy in a mixed gas of fermion and boson by sympathetic cooling . All the relevant experiments are carried out by starting with the two-species MOTs in which the two atomic ensembles are both confined and interact in the same trapping space.
Following the rapid development of this field there are more demands to study the time-dependent behaviors of the heteronuclear collisions [6, 7], sympathetic cooling  and phase separation in the two-species degenerate gases  for a better understanding of their dynamics. One way for achieving these goals is to prepare two independent ultracold atomic samples, each containing one single species, at separate locations initially and merge them together afterwards, to learn how the spatial mixing evolves with time.
Merging two spin-polarized atomic ensembles of the same species was demonstrated previously by gradually combining two magnetically trapped clouds [10, 11]. In those experiments, two clouds were nonsimultaneously produced from two independent MOTs. The setup can be much simplified if the two MOTs are simultaneously produced with only a few mm apart and are subsequently loaded into a double-well magnetic trap for further overlapping . However, there is an intrinsic difficulty to produce two magnetic quadrupole centers of only a few mm separation using the traditional MOT configuration or its combination.
Making two spatially separated MOTs for double-well magnetic trap loading is interesting in its owing right. Two spatially separated cold atomic samples are good candidates for the studies of quantum information , atomic tunneling , and for gravity field measurements . In addition, two magnetically trapped ultracold clouds of different species with a few mm apart permit to study the heteronulear collisions beyond S-wave scattering , which is still hard to explore under the current experimental condition.
More recently, Yun et al. showed that it is possible to use a single wire and apply an external bias field to produce two quadrupole centers, with opposite magnetic sign, at equal distance above and below the wire center along the symmetric axis , as shown in Fig. 1. This allows to create the double-well magneto-optic trap (DWMOT) and simultaneously confine two atomic clouds of the same or different species, with adjustable separation, by simply varying the bias field and wire current. In this letter, we show our experimental realization of this novel scheme by trapping Rb and Cs atoms and discuss its future applications.
The schematic of our DW MOT setup is shown in Fig. 1. It consists of a single-loop inner coil and a pair of Helmholtz coils. The inner coil sits inside a high vacuum glass cell. It is made from an oxygen-free copper plate of 2 mm thick and has an average radius R=5.25 mm. A current I flows into the inner coil to produce the required magnetic field gradient. The Helmholtz coil pair are located outside the cell and produce a magnetic bias field B b by feeding a current I b. The Rb and Cs dispensers are used as the atomic sources. The vacuum system is connected to an ion pump (8 ℓ/s pumping speed) and reaches a background pressure of 10-9 Torr when the dispensers are switched off.
To trap two species atoms, four stabilized diode lasers are used to generate the required trapping and repumping beams. Each horizontal incident trapping beam (in xy plane) has the elliptical profile with the long axis along the vertical (z) direction, and 1/e 2 diameters of 4 mm×8.5 mm for Rb, and 3.5 mm×8.5 mm for Cs, respectively. This provides the uniform horizontal trapping forces and allow smooth moving of the clouds along the z direction. The average horizontal MOT beam intensity is 20 mW/cm2 for Rb MOT, and 15 mW/cm2 for Cs MOT. The two vertical trapping beams have circular profiles and with the 1/e 2 diameter of 4 mm. The average vertical beam intensity is 20 mW/cm2 for Rb MOT, and 15 mW/cm2 for Cs MOT. The two sets of trapping beams have the opposite helicity for the two quadrupole centers.
In our setup, the Rb MOT is above the Cs MOT. The two sets of horizontal MOT beams are individually centered at z=+/-3 mm to +/-5 mm, depending on the practical requirements of the measurements. Since the glass cell surface is not with anti-reflection coating, this causes a typical 6.5% reflection of incident power and produces interference stripes on the beams while passing each cell surface. To achieve good intensity balance each horizontal trapping beam is made with a little convergence to reduce the retroreflected beam size to be ~90% of its corresponding incident one at the trapping region. Also, to allow for more flexible adjustment of the two MOTs, each horizontal trapping beam uses its own mirrors and λ/4-plates.
However, in the z direction, the two pairs of vertical trapping beams are retroreflected and share the same waveplates. This is accomplished by first overlapping the two MOT beams, with mutually perpendicular polarizations, at a polarization beam splitter cube, and sending them to the λ/4-plates and mirrors afterwards. We find the λ/4-plates, though made for 780 nm wavelength, can be used for Cs MOT too. The two incident vertical beams are also made with a slight convergence, similar to the horizontal ones, for intensity balance purpose.
The frequencies of the MOT beams are set to -13 MHz detuned from the 5S 1/2,F=2 to 5P 3/2,F=3 transition for Rb, and 6S 1/2,F=4 to 6P 3/2,F=5 for Cs. Each of the MOTs is shined with its individual repumping beam. The two repumping beams are linearly polarized and have the 1/e 2 diameter of 10 mm and power of 10 mW. The Rb repumper is locked to the 5S 1/2,F=1 to 5P 3/2,F=2 transition, and the Cs one is locked to the 6S 1/2,F=3 to 6P 3/2,F=4 transition.
In DW MOT operation, the numbers of trapped atoms are measured by imaging their fluorescence onto a calibrated photodiode and digital CCD camera. Using the fluorescence measurement, the accuracy in determining the absolute atom number is about ±15%. Additionally, when the atomic clouds are close to the inner coil, the scattered light from the coil surface also degrades the fluorescence measurement and causes more uncertainty on the atom number estimation. The temperature is measured by the time-of-flight method using a CCD. The Rb and Cs dispensers are operated at the current of 4.4 A to produce the highest loading rates for both MOTs, under the achievable vacuum condition.
To obtain the largest atom numbers in the MOTs, I and I b are set to generate the optimal axial magnetic field gradient. In general, using the trapping beams with the diameter of d, the steady-state atom number N in a cell MOT is proportional to ν 4 c d 2 , where νc is the capture velocity and is a function of the trapping laser detuning and intensity, and the applied magnetic field gradient . Due to the small beam size used in this experiment, a relative large magnetic field gradient is expected to provide larger vc and hence to capture more atoms.
3. Results and discussions
When the MOTs are formed the two clouds sit at a distance Z 0 above and below the wire center along the z axis. Z 0 is characterized by the following equation 
where µ 0 is the permeability of free space. To optimize the trapping condition we first find the suitable magnetic field gradient. This is done by making a series of atom number measurements while the Rb and Cs MOTs are individually located at Z 0=+/-3 mm, and only the magnetic field gradient is changed. The largest attainable atom number N max, for both species, is obtained while the axial magnetic field gradient B′z is set at 25 G/cm. However, the atom number quickly drops to 0.6 N max when B′z is increased to 35 G/cm or decreased to 15 G/cm.
Under our typical operation condition, the cloud separation is 2Z 0=9 mm and each cloud is under the local magnetic field gradient B′z ~21.5 G/cm, as I=13.87 A and B b=7.0 G. We normally trap 1.8×106 87Rb atoms and 1.2×106 133Cs atoms at spatial density of 4×1010 atoms/cm3 while the two sets of horizontal trapping beams are centered at z ~5 mm and z ~-5 mm, respectively. The MOT loading rates are 1.0×107 atoms/s and 1.3×107 for Rb and Cs atoms. The trap lifetimes are both 150 ms. The cloud temperatures are measured to be 350 µK and 300 µK for Rb and Cs, respectively. The measured atom numbers, densities, and temperatures of the two species MOTs agree reasonably well with the predictions in Ref. .
To verify the cloud separation can be adjusted by independently varying I and B b we first make the cloud center position measurement by keeping I invariant and changing B b alone. Figure 2 shows the images of the two clouds at each Z 0, from +/-4.5 mm to +/-1.5 mm, while B b varies from 7.0 G to 14.0 G as I=13.87 A. The center position and atom number of each cloud versus B b are plotted in Fig. 3. We also make the similar measurement by keeping B b invariant while changing I. The measured Z 0 and atom number for each I are plotted in Fig. 4. Both measurements show good agreement with the theoretical calculations by Eq. 1, and demonstrate the MOT centers are well controlled by adjusting I and B b.
Since the trapped atom number at each Z 0 strongly depends on the local magnetic field gradient, trapping beam parameters, such as the intensity and detuning, this leads to a large variation of the atom number at different MOT locations as seen in Figs. 3 and 4 . To further understand the origin of the observed fluctuation in atom number we first check the data shown in Fig. 3. This measurement is implemented under the typical running condition, except the two sets of horizontal trapping beams are centered at z ~+/-4 mm, respectively. As we described the setup for the horizontal MOT beams before, due to a slight convergence from its incidence, each retroreflected beam has the diameter (1/e 2) reduced to 3.6 mm×7.7 mm for Rb, and 3.2 mm×7.7 mm for Cs. The horizontal trapping beam intensity thus drops by about 55% from z=+/-4 mm to +/-1.5 mm. Furthermore, the interference fringes coming from each cell surface and the edge diffraction pattern produced while each horizontal MOT beam hits the inner wire somewhat contaminate the trapping beam profile and also cause fluctuation of atom number. The beam distortion is more significant as z approaches to 1.5 mm. In addition, our calculations show the local axial field gradient B′z, starting from 21.5 G/cm as B b=7 G, is gradually increased to a maximum 27.1 G/cm while B b=11 G, and then falls to 22.8 G/cm when B b is 14 G. In Fig. 3, the atom numbers, for both Rb and Cs, are gradually lowered as Z 0 is decreased. The data reasonably reflect the combination effect of the horizontal beam intensity distribution and the field gradient variation with Z 0.
The same discussion can be applied to the measurement shown in Fig. 4 while the two sets of horizontal trapping beams are brought even closer to the inner coil and centered at z ~+/-3 mm. Under the fixed bias field B b=10.35 G, the axial field gradient B′z, starting from 17.8 G/cm as I=10.4 A, is increased to 24.3 G/cm while I=12.4 A, and is further increased to 31.6 G/cm when I=20.3 A. Combining these two factors together it thus leads to the atom number reaches a maximum around I=14-15 A, and gradually falls due to power decrease and unfavorable field gradient at larger I. Although the horizontal trapping beam center is set closer to the inner coil and should allow trapping more atoms, it also produces larger stray light to contaminate the fluorescence signal. We suspect the unusual large atom number observed in Rb MOT as Z 0 ~1.5 mm might be due to this reason and is worth further investigation.
We also try to reduce the cloud separation and achieve a minimum value 2Z 0 ~2 mm, while each MOT contains no more than 105 atoms. Since our calculations show the trapping volume and depth of each MOT can be kept almost constant as Z 0 is reduced from few mm down to ~500 µm if I and B b are properly chosen, the above observed minimum Z 0 is apparently limited by the serious distortion of the horizontal MOT beams when blocked by the inner wire. Further decrease of the cloud separation is possible if a thinner wire or a chip-type DW MOT is used.
Another specialty about the DW MOT is that the two clouds can be easily brought to as close as a few mm, and each MOT is still allowed for independent manipulation by varying its individual trapping condition alone. For example, a 2D Rb MOT  can be formed without affecting the 3D Cs MOT by simply turning off the vertical Rb trapping beams, as shown in Fig. 5.
Currently, the two clouds, though ~3 mm apart, can be brought together for cold collision measurements. One can shine a red-detuned laser beam for Rb atoms to guide them towards the Cs MOT vertically [22, 23]. The center-of-mass kinetic energy of the Rb cloud obtained from the guiding beam through a distance of 3 mm is converted into the collision energy of the two MOTs . This MOT collider might allow to observe high-order partial waves interference in cold collisions between two species atoms . One can also suddenly switch on a new set of Rb MOT beams intersecting at the Cs MOT center , right after the presently used ones are turned off. The falling cold Rb atoms will be efficiently recaptured and collide with the Cs MOT. This permits to study the mixing dynamics of two MOTs of different species. Furthermore, the two MOT clouds separated in a few mm away are readily suitable for double-well magnetic trap loading.
In the future, the DW MOT configuration can be extended to the microtrap environment to make two spatially separated MOTs, with a few hundreds of micrometers apart. It is feasible to load ~106 atoms from the two MOTs into the species dependent magnetic microtrap  for further compression and cooling. This might provide a route to independently produce two nearby clouds of very high phase space density under a well controllable way.
We have demonstrated simultaneous trapping of 87Rb and 133Cs in a double-well MOT. Typically, 1.8×106 87Rb atoms and 1.2×106 133Cs atoms, at spatial density of 4×1010 atoms/cm3, with temperatures of 300 to 350 µK are trapped. We show that the two clouds can be smoothly moved from the initial cloud separation 9 mm to as close as 3 mm by independently varying the inner coil current and the external bias field. The DWMOT is suitable for studying the cold heteronuclear collisions and for double-well magnetic trap loading of two-species atoms. The DW MOT configuration can be further modified or miniaturized for future potential applications using magnetic microtraps.
We acknowledge S.T.Wu for assistance on the experiment. D.J.H. would like to acknowledge the support from the National Science Council of R.O.C. under NSC grant NO. 93-2112-M-194-009. Yin would like to thank the support from the National Nature Science Foundation of China (Grant Nos.10434060 and 10674047) and the National Key Basic Research and Development Program of China (Grant No 2006CB921604).
References and links
1. E. Raab, M. Prentiss, A. Cable, S. Chu, and D. Pritchard, “Trapping of Neutral Sodium Atoms with Radiation Pressure,” Phy. Rev. Lett. 59, 2631–2634 (1987). [CrossRef]
2. T. Walker, P. Feng, D. Hoffmann, and R. S. Williamson, III, “Spin-polarized spontaneous-force atom trap,” Phy. Rev. Lett.69, 2168–2172 (1992); C. J. Myatt, N. R. Newbury, R. W. Ghrist, S. Loutzenhiser, and C. E. Wieman, “Multiply loaded magneto-optical trap,” Opt. Lett. 21, 290–292 (1996); J. Reichel, W. Hansel, and T. W. Hansch, “Atomic Micromanipulation with Magnetic Surface Traps,” Phy. Rev. Lett. 83, 3398–3401 (1999); T. Pfau and J. Mlynek, “A 2D quantum gas of laser cooled atoms,” OSA Trends in Optics and Photonics 7, 33–38 (1997); Y. B. Ovchinnikov, I. Manek, and R. Grimm, “Surface Trap for Cs atoms based on Evanescent-Wave Cooling,” Phys. Rev. Lett. 79, 2225–2228 (1997); K. I. Lee, J. A. Kim, H. R. Noh, and W. Jhe, “Single-beam atom trap in a pyramidal and conical hollow mirror,” Opt. Lett. 21, 1177–1179 (1996). [CrossRef]
3. M. H. Anderson, J. R. Ensher, M. R. Matthews, C. E. Wieman, and E. A. Cornell, “Observation of Bose-Einstein Condensation in a Dilute Atomic Vapor,” Science 269, 198–201 (1995). [CrossRef] [PubMed]
4. M. S. Santos, P. Nussenzveig, L. G. Marcassa, K. Helmerson, J. Flemming, S. C. Zilio, and V. S. Bagnato, “Simultaneous trapping of two different atomic species in a vapor-cell magneto-optical trap,” Phys. Rev. A52, R4340–R4343 (1995); M. Taglieber, A.-C. Voigt, F. Henkel, S. Fray, T. W. Hansch, and K. Dieckmann, “Simultaneous magneto-optical trapping of three atomic species,” Phys. Rev. A 73, 011402(R).1–011402(R).6 (2006). [CrossRef] [PubMed]
5. M. W. Mancini, G. D. Telles, A. R. L. Caires, V. S. Bagnato, and L. G. Marcassa, “Observation of Ultracold Ground-State Heteronuclear Molecules,” Phy. Rev. Lett. 92, 133203.1–133203.4 (2004). [CrossRef]
6. G. Roati, F. Riboli, G. Modugno, and M. Inguscio, “Fermi-Bose Quantum Degenerate 40K-87Rb Mixture with Attractive Interaction,” Phy. Rev. Lett.89, 150403.1–150403.4 (2002); G. Modugno, G. Ferrari, G. Roati, R. J. Brecha, A. Simoni, M. Inguscio, “Bose-Einstein Condensation of Potassium Atoms by Sympathetic Cooling,” Science 294 1320–1324 (2001). [CrossRef]
7. M. Anderlini, E. Courtade, M. Cristiani, D. Cossart, D. Ciampini, C. Sias, O. Morsch, and E. Arimondo, “Sympathetic cooling and collisional properties of a Rb-Cs mixture,” Phys. Rev. A 71, 061401(R).1–061401(R).4 (2005). [CrossRef]
8. M. Anderlini, D. Ciampini, D. Cossart, E. Courtade, M. Cristiani, C. Sias, O. Morsch, and E. Arimondo, “Model for collisions in ultracold-atom mixtures,” Phys. Rev. A 72, 033408.1–33408.9 (2005). [CrossRef]
9. H. Pu and N. P. Bigelow, “Properties of Two-Species Bose Condensates,” Phy. Rev. Lett. 80, 1130–1133 (1997). [CrossRef]
10. W. Hansel, J. Reichel, P. Hommelhoff, and T. W. Hansch, “Magnetic Conveyor Belt for Transporting and Merging Trapped Atom Clouds,” Phy. Rev. Lett. 86, 608–611 (2001). [CrossRef]
11. J. F. Bertelsen, H. K. Andersen, S. Mai, and M. Budde, “Mixing of ultracold atomic clouds by merging of two magnetic traps,” Phy. Rev. A 75, 013404.1–013404.11 (2007). [CrossRef]
12. N. R. Thomas, A. C. Wilson, and C. J. Foot, “Double-well magnetic trap for Bose-Einstein condensates,” Phys. Rev. A 65, 063406.1–063406.8 (2002). [CrossRef]
13. G.-P. Guo and G.-C. Guo, “Entanglement of individual photon and atomic ensembles,” Quantum Information and Computation 3, 627–634 (2003).
14. S. Ashhab and C. Lobo, “External Josephson effect in Bose-Einstein condensates with a spin degree of freedom,” Phys. Rev. A 66, 013609.1–013609.10 (2002). [CrossRef]
15. A. B. Matsko, N. Yu, and L. Maleki, “Gravity field measurements using cold atoms with direct optical readout,” Phys. Rev. A 67, 043819.1–043819.12 (2003). [CrossRef]
16. N. Kjærgaard, A. S. Mellish, and A. C. Wilson, “Differential scattering measurements from a collider for ultracold atoms,” New J. Phys.6, 146.1–146.15 (2004); N. R. Thomas, N. Kjargaard, P. S. Julienne, and A. C. Wilson, “Imaging of s and d Partial-Wave Interference in Quantum Scattering of Identical Bosonic Atoms,” Phy. Rev. Lett. 93, 173201.1–173201.4 (2004). [CrossRef]
19. C. G. Townsend, “Laser Cooling and Trapping of Atoms,” Ph.D. thesis (Oxford University, 1995).
20. The model used to calculate the atom numbers in Ref.  is based on the low trapping beam intensity assumption, and uses a small laser detuning, which are different from what we use in the experiment.
21. K. Dieckmann, R. J. C. Spreeuw, M. Weidemuller, and J. T. M. Walraven, “Two-dimensional magneto-optical trap as a source of slow atoms,” Phys. Rev. A 58, 3891–3895 (1998). [CrossRef]
22. B. T. Wolschrijn, R. A. Cornelussen, R. J. C. Spreeuw, and H. B. van Linden van den Heuvell, “Guiding of cold atoms by a red-detuned laser beam of moderate power,” New J. Phys. 4, 69.1–69.10 (2002). [CrossRef]
23. A molasses cooling of a few ms on Rb is needed before the guiding beam is switched on.
24. Ch. Buggle, J. Leonard, W. von Klitzing, and J. T. M. Walraven, “Bose-Einstein Condensates Studied with a Linear Accelerator,” in Laser Spectroscopy, E. A. Hinds, A. Ferguson, and E. Riis, eds., (World Scientific, Singapore, 2005), pp. 199–206.
25. Since the two quadrupole centers have the opposite magnetic sign, the new Rb MOT beams should have the same helicity as the Cs ones.
26. Ph. W. Courteille, B. Deh, J. Fortagh, A Gunther, S. Kraft, C Marzok, S. Slama, and C. Zimmermann, “Highly versatile atomic micro traps generated by multifrequency magnetic field modulation,” J. Phys. B: At. Mol. Opt. Phys. 39, 1055–1064 (2006). [CrossRef]