Transcript
a r X i v : q u a n t - p h / 0 6 0 1 1 6 3 v 2 2 1 A p r 2 0 0 6 Trapping atoms on a transparent permanent-magnet atom chip A. Shevchenko ∗ , M. Heili¨o, T. Lindvall, A. Jaakkola, I. Tittonen, and M. Kaivola Optics and Molecular Materials, Helsinki University of Technology, P.O.Box 3500, FI-02015 TKK, Finland T. Pfau 5. Physikalisches Institut, Universit¨ at Stuttgart, 70550 Stuttgart, Germany (Dated: February 1, 2008)We describe experiments on trapping of atoms in microscopic magneto-optical traps on an opti-cally transparent permanent-magnet atom chip. The chip is made of magnetically hard ferrite-garnetmaterial deposited on a dielectric substrate. The confining magnetic fields are produced by minia-ture magnetized patterns recorded in the film by magneto-optical techniques. We trap Rb atoms onthese structures by applying three crossed pairs of counter-propagating laser beams in the conven-tional magneto-optical trapping (MOT) geometry. We demonstrate the flexibility of the concept increation and in-situ modification of the trapping geometries through several experiments. PACS numbers: 03.75.Be, 32.80.Pj, 39.25.+k, 85.70.Ge Microfabricated devices for trapping and manipulation of ultracold neutral atoms, known as atom chips [1, 2, 3],have been demonstrated to provide remarkable control of the internal and external atomic states [4, 5, 6, 7, 8], andin an essential way to simplify the production of Bose-Einstein condensates [8, 9, 10, 11, 12]. The magnetic-fieldpatterns needed for trapping atoms on these chips are typically created by driving current through metal wires thatare lithographically fabricated on the chip surface [13]. During the last few years, however, much attention has beenpaid to the development of atom chips based on selectively magnetized permanent magnets [14, 15, 16, 17, 18]. Suchdevices, in principle, allow one to get rid of the electric-power dissipation in the wires, avoid magnetic-field noisesrcinating from temporal and spatial fluctuations of the currents, and reduce the near-field noise srcinating fromthe thermal motion of free electrons in the chip. Permanent magnets also make it possible to create novel geometriesfor surface traps, such as storage rings that can serve as miniature rotation sensors [19, 20, 21, 22]. In view of futureapplications for atom chips, these developments have a great practical importance.We introduce a new kind of permanent-magnet atom chip and demonstrate magneto-optical trapping of 85 Rb atomson the surface of this device. The chip is of optically homogeneous material and transparent to light at near-infraredand infrared wavelengths. Owing to this property we can make use of the ordinary MOT geometry with threeorthogonal pairs of counter-propagating laser beams to collect and trap atoms on the surface instead of using thereflection MOT configuration of the traditional atom chips. We routinely capture more than 10 6 atoms in a micro-MOT on a magnetized pattern at a distance of ∼ 100 µ m above the chip surface. Being optically transparent, thedevice allows unimpeded control and probing of the on-chip atoms with laser light. Another important feature thatmakes our atom chip flexible and simple to operate is the possibility and ease of in-situ reconfiguration of its trappingpotentials. The atoms are trapped above miniature magnetized patterns which can readily be remotely recorded anderased by means of conventional magneto-optical recording techniques, even in the presence of trapped atoms.To fabricate the chip, a 1 . 8 µ m thick film of magnetically hard ferrite-garnet, (BiYTmGd) 3 (FeGa) 5 O 12 , was grownon a 500 µ m thick substrate of gadolinium-gallium-garnet (Gd 3 Ga 5 O 12 ) [23]. The film only absorbs about 10 % of the light at λ = 780 nm, which is the wavelength for trapping Rb atoms. The preferred direction of magnetization inthe film is normal to the surface and the nearly squared hysteresis loop of the film is characterized by a saturationmagnetization of ∼ 20 mT and coercivity of higher than 10 mT. We first magnetize the film uniformly and then createa desired magnetization pattern by locally heating the film with a scanned, focused cw laser beam at λ = 532 nm,at which wavelength 80 % of the light is absorbed in the film. During the patterning, an external magnetic fieldof ∼ 1 mT is applied in the direction opposite to the initial magnetization of the film. By reversing the directionof the applied magnetic field the patterns can be erased with the same laser beam. We have created patterns withdimensions down to the order of µ m with this technique [23]. Typically, 10 mW of power in the beam is needed towrite 10 µ m thick lines on the chip.The film is placed in a rectangular UHV cell made of fused silica and connected to a vacuum system that keeps thepressure of 10 − 11 mBar (see Fig. 1). The cell is located in the center of a set of 6 square-shaped magnetic coilsof 35 cm dimension that are used to compensate the background magnetic fields and to create a uniform externalmagnetic field on the chip surface. By controlling the strength and direction of the external field, a quadrupole field ∗ Email:
[email protected].fi; Fax. +358 9 451 3155 2 FIG. 1: Experimental setup. In (a), CCD - CCD camera, FGF - ferrite-garnet film on gadolinium-gallium-garnet substrate,laser - laser head that collimates and σ + -polarizes the light from a fiber. Two anti-Helmholtz coils (AH coils) are positionedclose to the right and left windows of the UHV cell. One pair of the cooling laser beams, the optical axis of the polarizationimaging system, and the laser beam for magneto-optical patterning pass through the openings of the coils. (b) Cooling-beamalignment. (c) Transparency of the atom chip under illumination at 780 nm. structure can be created at the location of the magnetized surface pattern. Based on such a magnetic-field structure, aminiature magneto-optical atom trap at a short distance from the surface is then created by applying three orthogonalpairs of retro-reflected laser beams that are circularly polarized and intersect on the chip surface at the location of amagnetized pattern. Two of the beam pairs propagate along the surface, and the third one is let directly through thechip.The cooling laser beams are produced by three separate single-frequency diode lasers that are injection-locked to asingle home-built transmission-grating external-cavity diode laser [24]. The external-cavity laser is locked close to the | 5 2 S 1 / 2 ,F = 3 →| 5 2 P 3 / 2 ,F = 4 transition of 85 Rb. The light from the cooling lasers is delivered to the setup in threepolarization-maintaining optical fibers. The maximum power in each of the beams is 20 mW, and the 1 /e 2 diameterof the beams in the cell is about 10 mm. The light frequency is tuned to the red from the atomic resonance by oneatomic linewidth Γ. To obtain the repumping radiation, another laser locked to the | 5 2 S 1 / 2 ,F = 2 →| 5 2 P 3 / 2 ,F = 2 transition is used. This radiation is guided to the cell in the same fiber as the light of one of the cooling lasers. Thepower in the repumping beam is several mW. The trapped atoms are observed from directions normal and parallelto the film by using two CCD cameras, SSC-M370CE (Sony) and Pixelfly (PCO).The atoms are collected to the micro-MOTs from rubidium vapor evaporated in the cell from a resistively heatedRb dispenser. The chip is positioned at a distance of 4 cm from the dispenser with the substrate side facing it. In allthe experiments the dispenser is operated in continuous mode.To record and erase magnetization patterns in the film, a simple mechanical beam-scanning system was built (seeFig. 2). A laser beam from a cw laser (Coherent Verdi-V10; λ = 532 nm) is focused onto the film by reflecting it froma mirror whose tilt angle is mechanically controlled via a computer. The beam spot size on the film is adjusted byshifting the lens L along the beam axis. A mechanical shutter is used to switch the recording/erasing beam on and off.The external field that defines the magnetization direction in the recorded pattern is produced by co-running currentsin the AH coils shown in Fig. 1. The setup includes a polarization-microscopy imaging system for in-situ observationof the recordings. Faraday rotation in the film is made visible by observing the transmission of thermal light throughthe film and two nearly crossed polarizers placed in front and after the cell [23]. An image of the transmitted-lightpattern is recorded with the same camera as is used to detect the trapped atoms. Figures 3a and e show two examplesof magnetization patterns visualized with this imaging system.The magnetic field produced by a given pattern can be calculated by using the Biot-Savart law [23]. For example,the magnetic field strength at the center of the magnetized circular spot shown in Fig. 3a is equal to 130 µ T. Byapplying a uniform magnetic field of 60 µ T in the opposite direction one can obtain a localized quadrupole field abovethe pattern. The absolute value of the field, | B q | , calculated as a function of distance z from the center of the spotalong the normal to the surface is plotted in Fig. 3h. At z = 200 µ m, B q is zero, and the gradient ∂B q /∂z is equalto 3 . 5 mT/cm. The gradients along the x and y directions are half of this.For trapping atoms, the intensities of the cooling laser beams are carefully balanced at the position where theminiature magnetic quadrupole is created above a selected magnetized pattern. We first trap atoms in a largemagneto-optical trap created with the aid of two external anti-Helmholtz coils placed close to the cell windows (see 3 FIG. 2: Combined magneto-optical patterning and polarization imaging system. To record/erase magnetization patterns inthe film (FGF), the electric current in coils C 1 and C 2 (AH coils in Fig. 1) is switched on and a cw laser beam ( λ = 532 nm) isreflected from a scanning mirror (SM) and focused with lens (L) onto the film. The imaging system consists of a CCD camera,a tungsten halogen lamp, a color filter (F), and two nearly crossed polarizers (P 1 and P 2 ).FIG. 3: Top row: two different magnetization patterns on the chip. Second row: front views of the resulting trapped atomicclouds. Third row: side views of the same traps (in (g) the clouds are located at different distances from the side-view camera).Bottom row: (d) front view of an atomic cloud above the pattern (a) when the capture volume of the trap is increased byapplying an additional large-extent quadrupole field; (h) the absolute value of the microscopic magnetic quadrupole field as afunction of distance z from the center of spot (a). Fig. 1). Then the trap center is shifted towards the pattern by adjusting the positions of the coils, and the currents inthe coils are gradually decreased to zero. Simultaneously, the number of atoms in the trap is optimized by readjustingthe retro-reflection angles of the cooling beams. When eventually the currents in the external coils are switched off,the atoms remain trapped in the surface trap. Figure 3b shows the fluorescence image of an atomic cloud above thepattern of Fig. 3a. In the side view, Fig. 3c, the cloud is seen together with its reflection from the surface. Thedistance of the trap center from the surface is 200 µ m. The number of atoms in the trap is 7 × 10 4 . In order toincrease the number of trapped atoms, we increased the capture volume of the trap by superimposing on the steepquadrupole field of the surface trap a weaker quadrupole field of larger spatial extent produced by the two externalanti-Helmholtz coils outside the cell. The spatial gradient of this field was a few hundreds of µ T/cm. As a result,the number of atoms in the trap exceeded 10 6 . This case is illustrated in Fig. 3d. We then trapped atoms withintwo identical magnetized spots positioned at a distance of 3 mm from each other. The spots and the trapped atomicclouds are shown in Figs. 3e, f, and g. The additional external quadrupole field was not used in this case. We notethat such micro-MOTs could be created even further apart from each other if the cooling laser beams were made tohave a larger diameter.In order to demonstrate the in-situ reconfigurability of the trapping potentials, we consecutively created and erased 4 FIG. 4: Consecutively recorded and erased magnetization patterns (top row) and atomic clouds trapped within them (bottomrow). The trap (d) was modified to obtain the trap (e) without significantly affecting the trapping of the atoms in the threesrcinal traps. A weak auxiliary quadrupole field centered in the middle of the structure was used to increase the trappingefficiency.FIG. 5: A toroidal trap: (a) magnetization pattern, (b) front view and (c) side view of trapped atomic cloud. magnetization patterns within the same area of the film. Some of the patterns and the atoms trapped above them areshown in Fig. 4. In Fig. 4b, a curved strip of reversed magnetization is added to the magnetized spot of Fig. 4a, and theresulting L-shaped continuation of the MOT is filled with atoms. This structure was then erased and a square-shapedpattern was recorded in its place (see Fig. 4c). The number of atoms in this trap is comparable to that in the trap of Fig. 4a. We note that, if the new pattern has similar dimensions and position as the previous one, it is not necessaryto readjust the applied magnetic and optical fields for the new trap. The trap can be reconfigured even while atomsare confined in it. The array of micro-MOTs shown in Fig. 4e was obtained by modifying the array of Fig. 4d withoutdestroying the functioning of the srcinal micro-traps. The fourth square was magnetized purely optically, i.e., noadditional writing magnetic field was applied. Even if the value of magnetization within this additional square isclose to zero, the saturation magnetization of the surrounding area of the film is high enough for the creation of amicro-MOT. MOT arrays, similar to those shown in Figs. 4d and e but with somewhat larger dimensions, have beendemonstrated previously by using ribbon UHV cables and applying the reflection-MOT principle [25].The magneto-optical trapping efficiency turns out to be rather insensitive to the details of the magnetization-patterngeometry. It is, for example, possible to collect atoms within some part of a pattern and then guide the trap along thepattern’s structure. This possibility could provide extra flexibility for the design and operation of atom-chip circuits.We recorded a toroidal pattern shown in Fig. 5a that, as a matter of fact, would be difficult if not impossible to realizeusing current-carrying wires. If a uniform magnetic field is applied in the direction opposite to the magnetization of the torus, the atoms settle in a ring-shaped MOT, as shown in Figs. 5b and 5c. The slightly uneven distribution of the atoms inside the torus is mainly explained by the interference and diffraction of the cooling laser beams at theposition of the trap. We believe that this destructive effect can be substantially reduced by polishing the film andAR-coating the surfaces of the device.By periodically modulating the x - and y -components of a weak external magnetic field of the compensation coils witha mutual phase difference of π/ 2, we could drive the trap center into circular motion along the torus. The modulationis accomplished by modulating the currents in the 35 cm compensation coils. Figures 6b-e show a sequence of imagesof the trapped atoms separated in time by a quarter of the modulation period. In this case, an auxiliary quadrupolefield, with a spatial gradient of 300 µ T/cm, was added to increase the number of atoms. The modulation amplitudeswere 27 µ T.In conclusion, we have demonstrated magneto-optical trapping of atoms on a transparent permanent-magnet atomchip. This novel approach to the creation of atom chips can provide several advantages over conventional techniquesbased on current-carrying wires and over the other up-to-date techniques employing permanent magnets. Our trapsare readily reconfigurable in situ . Essentially free-format trap patterns can be realized, as demonstrated by theexample of a ring-shaped trap. Ring-shaped traps are particularly interesting due to the possibility of applying themin a Sagnac-type atom interferometer [19, 20, 21, 22]. The device is transparent to light, which provides unimpeded 5 FIG. 6: A surface MOT translated along a toroidal magnetization pattern: (a) magnetization pattern, (b)-(e) a series of imagesof the trapped atomic cloud. control of atoms with laser radiation. In particular, an ordinary MOT instead of a reflection MOT geometry wasused to collect atoms close to the surface. Microscopic magneto-optical traps described in this work are formed at adistance of a few 100 µ m from the surface and they contain more than 10 6 atoms. We also created surface-mountedarrays of micro-MOTs. Such an array can be used to prepare multiple atomic samples on the chip. Each of thesesamples can then be processed individually, by using, e.g., a nearly resonant focused optical field.There are no electric currents applied to the device. Consequently, there is no electric-power dissipation nortemporal or spatio-temporal current fluctuation. Since the device is made of dielectric material, magnetic-field noisedue to thermal electrons is insignificant at short distances from the surface [26]. Purely magnetic micro-traps witha trap depth of up to 1 mK are realizable on an atom chip of this type [23]. Such traps can be loaded with atomsfrom surface-mounted micro-MOTs described in the present work and be used for Bose-Einstein condensation and forexperiments on atom interferometry.We acknowledge financial support from the Academy of Finland and the Jenny and Antti Wihuri Foundation andthank E.I. Il’yashenko and T.H. Johansen, University of Oslo, for fruitful collaboration. [1] J. Denschlag et al ., Appl. Phys. B 69 , 291 (1999).[2] R. Folman et al ., Phys. Rev. Lett. 84 , 4749 (2000).[3] D. Cassettari et al ., Appl. Phys. B 70 , 721 (2000).[4] P. Treutlein, P. Hommelhoff, T. Steinmetz, T. W. Hansch, J. Reichel, Phys. Rev. Lett. 92 , 203005 (2004).[5] P. Kr¨uger et al ., Phys. Rev. Lett. 91 , 233201 (2003).[6] P. Hommelhoff et al ., New J. Phys. 7 , 3 (2005).[7] K. Brugger et al ., Phys. Rev. A 72 , 023607 (2005).[8] T. Schumm et al ., Nature Phys. 1 , 57 (2005).[9] W. H¨ansel, P. Hommelhoff, T. W. H¨ansch, and J. Reichel, Nature 413 , 498 (2001).[10] H. Ott, J. Fortagh,G. Schlotterbeck, A. Grossmann, C. Zimmermann, Phys. Rev. Lett. 87 , 230401 (2001).[11] S. Schneider et al ., Phys. Rev. A 67 , 023612 (2003).[12] A. Kasper et al ., J. Opt. B 5 , S143 (2003).[13] R. Folman et al ., Adv. At. Mol. Phys. 48 , 263 (2002).[14] S. Eriksson et al ., Appl. Phys. B 79 , 811 (2004).[15] C. D. S. Sinclair et al ., Eur. Phys. J. D 35 , 105 (2005).[16] C. D. S. Sinclair et al ., Phys. Rev. A 72 , 031603(R) (2005).[17] B. V. Hall et al ., arXiv:cond-mat/0507435.[18] I. Barb et al ., Eur. Phys. J. D 35 , 75 (2005).[19] J. A. Sauer, M.D. Barrett, M. S. Chapman, Phys. Rev. Lett. 87 , 270401 (2001); gtresearch-news.gatech.edu/newsrelease/STORAGE.html.[20] S. Gupta, K. W. Murch, K. L. Moore, T. P. Purdy, D. M. Stamper-Kurn, Phys. Rev. Lett. 95 , 143201 (2005).[21] A. S. Arnold, C. S. Garvie, and E. Riis, arXiv:cond-mat/0506142.[22] Kunal K. Das, M. D. Girardeau, E. M. Wright, Phys. Rev. Lett. 89 , 170404 (2002).[23] A. Jaakkola et al ., Eur. Phys. J. D 35 , 81 (2005).[24] M. Merimaa et al ., Opt. Commun. 174 , 175 (2000).[25] A. Grabowski, T. Pfau, Eur. Phys. J. D 22 , 347 (2003).[26] C. Henkel, P. Kr¨uger, R. Folman, J. Schmiedmayer, Appl. Phys. B 76 , 173 (2003).
