Narrow-line magneto-optical trap for europium
Yuki Miyazawa, Ryotaro Inoue, Hiroki Matsui, Kenta Takanashi, Mikio Kozuma
aa r X i v : . [ c ond - m a t . qu a n t - g a s ] F e b Narrow-line magneto-optical trap for europium
Yuki Miyazawa, Ryotaro Inoue, Hiroki Matsui, Kenta Takanashi, and Mikio Kozuma
Department of Physics, Tokyo Institute of Technology,2-12-1 O-okayama, Meguro-ku, Tokyo 152-8550, Japan (Dated: February 24, 2021)We report on the realization of a magneto-optical trap (MOT) for europium atoms using a narrow-line cooling transition with a natural linewidth of 97 kHz. Our starting point is continuous capturingand cooling of optically pumped metastable europium atoms. We have employed simultaneous MOTfor the metastable and ground-state atoms. The trapped metastable atoms are successively pumpedback to the ground state and then continuously loaded to the narrow-line MOT, where up to 4 . × atoms are captured. A spin-polarized sample at a temperature of 6 µ K and with a peak numberdensity of 2 . × cm − is obtained through the compression process, resulting in a phase spacedensity of 3 × − . I. INTRODUCTION
Quantum degenerate gas of atoms with a large dipolemoment has opened up a new avenue for studying ultra-cold atoms with long-range and anisotropic interaction[1–3]. These quantum degenerate gases were first real-ized by using chromium [4] with a magnetic moment of6 µ B ; subsequently, the dipolar lanthanide atoms of dys-prosium [5] and erbium [6] with a magnetic moment of10 µ B and 7 µ B , respectively, too were used. Europium(Eu) also belongs to the lanthanide group and has a largedipole moment of 7 µ B . It has two stable bosonic iso-topes: Eu and
Eu with natural abundances of 48 %and 52 %, respectively. Since both isotopes have nuclearspins of I = 5 /
2, they have hyperfine structures in theirground state. This enables us to control their scatteringlength using radiofrequency fields [7, 8]: such control isuseful for examining the behavior of dipolar spinor Bosegases under an ultralow magnetic field [9–11]. Since theelectronic angular momentum of the ground state is zero,a less dense and nonchaotic Feshbach resonance spectrumis expected in the case of Eu, which is in contrast to thespectra of other dipolar lanthanides [12–14].The laser cooling of other lanthanide atoms was suc-cessfully conducted by combining Zeeman slowing usinga broad optical transition and a magneto-optical trap(MOT) with a narrow optical transition [15–17]. Coldatoms in the MOT were loaded to an optical dipole trap,and quantum degenerate gases were obtained throughsuccessive evaporative cooling [5, 6, 18]. However, thisapproach cannot be directly applied for Eu atoms be-cause the optical transition broad enough for Zeemanslowing has large optical leaks; excited atoms decay toat least six metastable states with a total probability of1 . × − [19]. Hence, we pumped atoms to an-other metastable state that exhibits a quasicyclic tran-sition at 583 nm and implemented the Zeeman slowingand MOT [20]. In this paper, we report on the laser cool-ing of Eu in the ground state. We optically pumped themetastable atoms back to the ground state and imple-mented a narrow-line MOT such that the MOT opera-tions for the metastable and ground states of atoms weresimultaneously performed. The rest of the paper is organized as follows. In sectionII, the narrow-line MOT procedure is described with fo-cus on optical transitions. Our experimental setup is ex-plained in section III. The experimental results of MOTcharacteristics are presented in section IV, and the paperis concluded in section V. II. NARROW-LINE MOT PROCEDURE
Figure 1 shows the schematic of our laser cooling pro-cedure. We start from a hot atomic beam of Eu inthe ground state. Atoms in the ground state a S ◦ / are transferred to the metastable state a D ◦ / as de-scribed below. By driving the 460 nm transition, atomsare pumped from the ground state into two intermediatestates a D ◦ / and a D ◦ / ; then, they are pumped to a D ◦ / by driving the 507 nm and 513 nm transitions,respectively. The total transfer efficiency is estimated asup to 19 % [19, 22]. The atoms in a D ◦ / are then Zee-man slowed and captured in a MOT (yellow-MOT) us-ing the a D ◦ / ↔ z F / cooling transition at 583 nmwith a natural linewidth of Γ / π = 8 . a S ◦ / ↔ z P / cooling transition at 687 nmwith a natural linewidth of Γ / π = 97 kHz [21]. Notethat this series of optical pumping and laser coolingprocesses were executed simultaneously. Atoms in theground state are thus loaded to the red-MOT continu-ously. The cooling transition at 687 nm has optical leaksfrom the excited state z P / to three metastable states a D ◦ / , a D ◦ / , and a D ◦ / . The decay rates wereexperimentally measured in this work (see Appendix B),and the sum of them was estimated as 1 . × s − ,which corresponds to a branching fraction of 2 . y P / Z ee m a n s l o w i ng & y e ll o w - M O T li gh t n m , . M H z n m n m n m n m op ti ca l l ea k s . % n m red-MOT light687 nm, 97 kHz op ti ca l pu m p i ng t o op ti ca l pu m p i ng t o FIG. 1. Energy levels and transitions of Eu relevant for our laser cooling procedure [21, 22]. Solid lines indicate laser-driventransitions and dashed lines indicate spontaneous decay channels. state (see Appendix A).
III. EXPERIMENTAL SETUP
Our Eu atomic beam was produced from an effusiveoven operating at 770 K. Optically pumped metastableatoms were Zeeman slowed and loaded to the yellow-MOT, which was formed in a quadrupole magneticfield provided by anti-Helmholtz coils and three pairsof counter-propagating, circularly polarized cooling-lightbeams at 583 nm. The beam diameter was about 25 mmand truncated by a circular aperture of 21 . µ m so that cooled atomswere selectively pumped back.The narrow-line red-MOT was formed in the samemagnetic field as the yellow-MOT. The cooling light wasproduced by an external cavity laser diode and amplifiedby a tapered amplifier. The laser linewidth was sup-pressed below 33 kHz by stabilizing the laser frequencyto the resonance of an ultralow-expansion cavity. Thecooling laser beams overlapped the 583-nm beams withthe same polarization and beam diameter. The intensityratio of the laser beams propagating in axial and radialdirections of the coil was set as 2:1. Here, the axial di-rection was parallel to gravity. The cooling light wasred-detuned by δ with respect to the F = 6 ↔ F ′ = 7cyclic transition. We have added some additional lasersto repump from the levels a D (see Appendix A for de-tails).The number of trapped atoms was determined by anabsorption imaging technique using the 460 nm transi- tion with spin polarization. We turned off all the lightsand quadrupole magnetic field after loading the atomsand applied a magnetic field of approximately 10 − Talong the probe axis. Then, atoms were optically pumpedto the | F = 6 , m F = 6 i Zeeman sublevel by a two-color σ + polarized light pulse tuned to F = 5 ↔ F ′ = 6and F = 6 ↔ F ′ = 6 transitions. Subsequently,the polarized atoms were illuminated by a σ + polarizedprobe laser beam near-resonant on | F = 6 , m F = 6 i ↔| F ′ = 7 , m F ′ = 7 i transition. The absorption by theatoms casts a shadow on an imaging device. We con-firmed that the degree of the spin polarization is sufficientfor determining the number of atoms within a few per-cent errors by comparing the optical depth of two imagesobtained by σ + and σ − polarized probe beams. IV. EXPERIMENTAL RESULTS
By using the aforementioned laser-cooling procedure,up to 4 . × Eu atoms were trapped. The num-ber of atoms in our red-MOT was maximized underthe following experimental conditions. The axial mag-netic field gradient was set to ∂B z /∂z = 3 × − T / m.The detuning and total intensity of the cooling laserfor the yellow-MOT were − .
75 Γ and 3 . I s, , re-spectively, whereas these values for the narrow-line red-MOT were δ = −
13 Γ and I = 40 I s, . Here, I s, = 5 . / cm , and I s, = 3 . µ W / cm were thesaturation intensities for the cooling transitions. Figure2 (a) summarizes the number of trapped atoms after 4 sof loading as a function of the detuning δ and the in-tensity I . The maximum number of atoms is obtainedaround δ = −
13 Γ and I = 40 I s, . The load-ing and decay curves of the red-MOT under the optimalconditions are shown in Fig. 2 (b) and (c), respectively.From curve fitting [23], we obtained a loading rate R =6 × atoms / s, a one-body loss rate of 1 /τ = 0 .
38 s − ,and a two-body loss rate of β = 7 × − cm / s. Themeasured one-body loss rate is well described by the in-completeness of our repumping scheme (see Appendix A)assuming that the population in the excited state is 0.02.This population is estimated from equilibrium of the ra-diative force, gravity, and magnetic force [24]. Note thatour vacuum chamber with < − Pa background pres-sure does not limit the decay rate. (b)(a) (c)
FIG. 2. (a) The number of trapped atoms as a function of thedetuning δ and the intensity I . The parameters for theyellow-MOT are fixed to the optimal conditions (see text).Loading (b) and decay (c) curves of the red-MOT under theoptimal conditions. The solid lines are fitted to the data, andthe dashed line in (c) is the exponential asymptote of the fit. One of the features of a narrow-line MOT is the spon-taneous spin polarization of the atoms in the MOT, asobserved in the narrow-line MOTs of other lanthanideatoms [6, 24]. As shown in Fig. 3, we evaluated the meanspin projection of the atoms in our red-MOT by usingthe Stern–Gerlach effect. After releasing the atoms fromthe red-MOT, we applied a 7-ms-long vertical magneticfield gradient of about 0 . / m. An absorption imagewas then obtained in the same manner as explained insection II. The typical absorption images are shown inFig. 3 (a) and (b) along with the reference images, whichwere obtained without applying a magnetic field gradient.The populations of individual spin levels were estimatedwith multiple Gaussian fitting of the absorption images,considering the corresponding reference images. We es- timated the population distributions as a function of thedetuning δ , and calculated the mean spin projections h F z i = P m F = − p m F m F from the obtained populations p m F as shown in Fig. 3(c). The figure reveals that atomswere well spin-polarized for larger detuning. With the de-tuning of δ = −
21 Γ , 95 % of the atoms populatedthe lowest Zeeman sublevel | F = 6 , m F = − i . (a) (b)(c) ∇ |B| ∇ |B| OD0123 OD00.20.40.6
FIG. 3. Spontaneous spin polarization in the narrow-line red-MOT. (a and b) Absorption images are taken at the samedetuning of δ = −
21 Γ (a) and δ = − (b). Theleft images are taken after free expansion with applying amagnetic field gradient for spin separation, whereas the rightimages are taken in the absence of the magnetic field gradient.(c) The estimated mean spin projection h F z i as a function ofthe detuning δ . The intensity is I = 3 I s, , and theaxial magnetic field gradient is ∂B z /∂z = 3 × − T / m. To increase the phase space density of the atomiccloud, a compression sequence was performed after theloading stage. Using the optimal parameters ∂B z /∂z =2 . × − T / m, I = 2 I s, , and δ = − ,we achieve a temperature of 6 µ K and a peak numberdensity of 2 . × atoms / cm with an atom numberof 3 . × . Under this condition, the population of | F = 6 , m F = − i was inferred as about 0 . × − . V. CONCLUSION
In conclusion, we demonstrated a MOT for Eu atomsusing the narrow-line transition with a natural linewidthof 97 kHz. Although Eu has no suitable transition forZeeman slowing of an atomic beam in the ground state,atoms are successfully loaded to the narrow-line MOT byusing Zeeman slowing and a MOT operated at the 583-nm cooling transition originating from the metastablestate in a continuous manner. The phase space densityand the number of atoms after the compression stage pro-vide us with good starting conditions for direct loadingto an optical dipole trap.
ACKNOWLEDGMENTS
This study was supported by JSPS KAKENHI GrantsNumbers JP16K13856 and JP17J06179. Y.M. acknowl-edges partial support from the Japan Society for the Pro-motion of Science.
Appendix A: Repumping scheme
The 687-nm cooling transition has optical leaksto some metastable states. The upper-state | z P / , F = 7 i has six dominant decay chan-nels to | a D ◦ / , F = 6 i , | a D ◦ / , F = 6 , i , and | a D ◦ / , F = 6 , , i , besides one to the lower-statehyperfine manifold. In this work, we plugged merely thefive most dominant relaxation pathways as schematicallyshown in Fig. 4. Here we omit the repumping from thestate | a D ◦ / , F = 6 i since the optical leak probabilityto this state is comparable with that to the other statesreferred to as “others” in Fig. 4. One can estimatethis overall repumping efficiency of the scheme as up to99 . ∼
10 mW, respectively.
TABLE I. Transition wavelengths λ and transition probabil-ities A ki in our repumping scheme.upperlevel lowerlevel λ (nm) A ki (s − ) y P / a S ◦ /
460 1 . × [22] a D ◦ / . × [19] a D ◦ / . × [19] a D ◦ / . × [19]others 1577-4880 ≧ . × [19] F=6F=5F=1
F=2F=6F=7
F=2F=6F=7 o t h e r s F=6F=5F=7F=2 F=6F=7F=3F=8F=66.9 GHz120 MHz 1.5 GHzF=5F=4F=1 n m n m l a s e r c oo li ng … … … … … …… … … … … … n m n m n m r e pu m p i ng r e pu m p i ng r e pu m p i ng op ti ca l l ea k s op ti ca l l ea k s ≧ . % FIG. 4. Energy levels and transitions of
Eu relevant forour repumping scheme including hyperfine structures. Solidlines indicate laser-driven transitions and dashed lines indi-cate spontaneous decay channels. Hyperfine splittings arecalculated based on hyperfine constants in references [25, 26].
Appendix B: Transition probabilities for optical-leaklines
We measured the transition probabilities for the leak-age transitions from z P / to a D ◦ / , a D ◦ / , and a D ◦ / . After loading the atoms, we turned off one ofthe three repumping beams, and kept an optical leakunplugged. Under this condition, the lifetime of thered-MOT is limited by the optical leak. By measuringthe lifetime of the red-MOT, we estimated the transitionprobability for the unplugged leakage transition. In thismanner, we measured the three transition probabilities,which are listed in Table II. TABLE II. The transition wavelengths λ and measured tran-sition probabilities A ki for the leakage transitions.upperlevel lowerlevel λ (nm) A ki (s − ) z P / a D ◦ / . × a D ◦ / . × a D ◦ / . × [1] T. Lahaye, J. Metz, B. Fr¨ohlich, T. Koch, M. Meister,A. Griesmaier, T. Pfau, H. Saito, Y. Kawaguchi, andM. Ueda, Phys. Rev. Lett. , 080401 (2008).[2] L. Chomaz, D. Petter, P. Ilzh¨ofer, G. Natale, A. Traut-mann, C. Politi, G. Durastante, R. M. W. van Bijnen,A. Patscheider, M. Sohmen, M. J. Mark, and F. Ferlaino,Phys. Rev. X , 021012 (2019).[3] S. Baier, M. J. Mark, D. Petter, K. Aikawa, L. Chomaz,Z. Cai, M. Baranov, P. Zoller, and F. Ferlaino,Science , 201 (2016).[4] A. Griesmaier, J. Werner, S. Hensler, J. Stuhler, andT. Pfau, Phys. Rev. Lett. , 160401 (2005).[5] M. Lu, N. Q. Burdick, S. H. Youn, and B. L. Lev,Phys. Rev. Lett. , 190401 (2011).[6] K. Aikawa, A. Frisch, M. Mark, S. Baier,A. Rietzler, R. Grimm, and F. Ferlaino,Phys. Rev. Lett. , 210401 (2012).[7] T. M. Hanna, E. Tiesinga, and P. S. Julienne,New J. Phys. , 083031 (2010).[8] D. J. Papoular, G. V. Shlyapnikov, and J. Dalibard,Phys. Rev. A , 041603(R) (2010).[9] Y. Kawaguchi, H. Saito, and M. Ueda,Phys. Rev. Lett. , 130404 (2006).[10] Y. Kawaguchi, H. Saito, and M. Ueda,Phys. Rev. Lett. , 080405 (2006).[11] M. Takahashi, S. Ghosh, T. Mizushima, and K. Machida,Phys. Rev. Lett. , 260403 (2007).[12] K. Zaremba-Kopczyk, P. S. ˙Zuchowski, and M. Tomza,Phys. Rev. A , 032704 (2018).[13] T. Maier, H. Kadau, M. Schmitt, M. Wenzel,I. Ferrier-Barbut, T. Pfau, A. Frisch, S. Baier,K. Aikawa, L. Chomaz, M. J. Mark, F. Ferlaino,C. Makrides, E. Tiesinga, A. Petrov, and S. Kotochigova,Phys. Rev. X , 041029 (2015).[14] V. A. Khlebnikov, D. A. Pershin, V. V. Tsyganok,E. T. Davletov, I. S. Cojocaru, E. S. Fe- dorova, A. A. Buchachenko, and A. V. Akimov,Phys. Rev. Lett. , 213402 (2019).[15] A. Frisch, K. Aikawa, M. Mark, A. Rietzler,J. Schindler, E. Zupaniˇc, R. Grimm, and F. Ferlaino,Phys. Rev. A , 051401(R) (2012).[16] T. Maier, H. Kadau, M. Schmitt, A. Griesmaier, andT. Pfau, Opt. Lett. , 3138 (2014).[17] D. D. Sukachev, E. S. Kalganova, A. V. Sokolov,S. A. Fedorov, G. A. Vishnyakova, A. V. Aki-mov, N. N. Kolachevsky, and V. N. Sorokin,Quantum Electron. , 515 (2014).[18] E. T. Davletov, V. V. Tsyganok, V. A. Khlebnikov,D. A. Pershin, D. V. Shaykin, and A. V. Akimov,Phys. Rev. A , 011302(R) (2020).[19] Y. Miyazawa, R. Inoue, K. Nishida, T. Hosoya, andM. Kozuma, Opt. Commun. , 171 (2017).[20] R. Inoue, Y. Miyazawa, and M. Kozuma,Phys. Rev. A , 061607(R) (2018).[21] B. Fechner, R. H. Rinkleff, and A. Steudel,J. Phys. B , 31 (1987).[22] E. A. D. Hartog, M. E. Wickliffe, and J. E. Lawler,Astrophys. J. Suppl. Ser. , 255 (2002).[23] A. Browaeys, J. Poupard, A. Robert, S. Nowak,W. Rooijakkers, E. Arimondo, L. Marcassa,D. Boiron, C. I. Westbrook, and A. Aspect,Eur. Phys. J. D , 199 (2000).[24] D. Dreon, L. A. Sidorenkov, C. Bouazza,W. Maineult, J. Dalibard, and S. Nascimbene,J. Phys. B , 065005 (2017).[25] G. J. Zaal, W. Hogervorst, E. R. Eliel, K. A. H. vanLeeuwen, and J. Blok, Z. Phys. A , 339 (1979).[26] W.-G. Jin, T. Endo, T. Wakui, H. Uematsu, T. Minowa,and H. Katsuragawa, J. Phys. Soc. Jpn71