Rapid generation of metastable helium Bose-Einstein condensates
A. H. Abbas, X. Meng, R. S. Patil, J. A. Ross, A. G. Truscott, S. S. Hodgman
RRapid generation of metastable helium Bose-Einstein condensates
A. H. Abbas , , X. Meng , R. S. Patil , , ∗ J. A. Ross , A. G. Truscott , and S. S. Hodgman † Research School of Physics, Australian National University, Canberra 0200, Australia Department of Physics, Faculty of Science, Cairo University, Giza, Egypt and Indian Institute of Science Education and Research, Pune 411008, India (Dated: February 23, 2021)We report the realisation of Bose-Einstein condensation (BEC) of metastable helium atoms usingan in-vacuum coil magnetic trap and a crossed beam optical dipole trap. A novel quadrupole-Ioffe configuration (QUIC) magnetic trap made from in-vacuum hollow copper tubes provides fastswitching times while generating traps with a 10G bias, without compromising optical access. Thebias enables in-trap 1D doppler cooling to be used, which is the only cooling stage between themagneto-optic trap (MOT) and the optical dipole trap. This allows direct transfer to the dipoletrap without the need for any additional evaporative cooling in the magnetic trap. The entireexperimental sequence takes 3.3 seconds, with essentially pure BECs observed with ∼ atomsafter evaporative cooling in the dipole trap. The experimental creation of Bose-Einstein conden-sates (BECs) of dilute weakly interacting gases of atoms[1–3] has opened the possibility of exploring interest-ing phenomena of the quantum world on a macroscopicscale. Over the subsequent years the field has exploded,with BECs now used in diverse fields of quantum sci-ence, including quantum many-body systems [4], topo-logical physics [5] and precision inertial measurements[6]. While BEC experiments were initially limited to ob-serving collective properties of the ensemble, a number ofmore recent detection techniques allow for the detectionof individual atoms [7], opening up a broad new rangeof experimental possibilities. In experiments involvingalkali or rare earth atoms, such single atom detection isusually performed via high resolution fluorescence imag-ing, either in-situ in optical lattices [8, 9] or after ex-pansion from a trap [10, 11]. However, such techniquesusually have limitations on their spatial extent and theatom number able to be imaged.An alternative method of single atom detection ex-ploits the high internal energy of helium atoms trappedin the first atomic excited 2 S metastable state (He ∗ ),which has 19.8eV of internal energy. The high inter-nal energy allows direct detection of individual atomswith full 3D resolution using electronic detectors suchas multi-channel plate and delay-line detectors (MCP-DLD) [12]. This has opened up a wide range of excit-ing experimental possibilities, allowing quantum-opticsequivalent demonstrations with atoms of iconic experi-ments such as the Hanbury Brown-Twiss (HBT) effect[13], Wheeler’s delayed choice [14], the Hong-Ou-Mandeleffect [15], ghost imaging [16] and the measurement ofBell correlations [17]. More than just replicating quan-tum optics though, the atomic interactions allow effectsto be seen that are not possible with photons, such as ∗ Current address: Department of Physics, The PennsylvaniaState University, University Park, Pennsylvania 16802, USA † [email protected] Fermionic anti-bunching [18], quantum depletion [19] andstrongly interacting lattice physics [20].Crucial to many of these experiments are the measure-ment of HBT-style correlation functions - the key observ-able enabled by single atom detection. However, suchcorrelation functions require large amounts of data, es-pecially if higher order correlations are being measured[21, 22], often needing 10,000-100,000 individual exper-imental runs for a single experiment. This has led toconstant improvements across generations of experimen-tal He ∗ apparatuses aiming for ever shorter experimentalsequences. Most He ∗ experiments [23–27] use magnetictraps, where the sequence length is limited by the slowthermalisation rates in magnetic traps that have rela-tively weak trap frequencies. While using liquid heliumin the source stage can substantially reduce the sequencelength [26], it is less practical for everyday operation. Thetight traps provided by dipole traps can overcome this, ei-ther in a hybrid combination with a magnetic trap [28] orin a stand-alone cross beam configuration [29]. However,due to the limited depth of the dipole trap, to enable anefficient transfer the atoms need to first be cooled, eitheroptically via complex additional cooling schemes such asgrey molasses and/or evaporatively in a magnetic trap[29].In this work, we report on the construction of a He ∗ BEC machine capable of creating condensates in a sim-plified, rapid sequence that takes only 3.3s in total. Aninitial stage of 1D Doppler cooling [25] is performed in abiased magnetic trap, constructed using a simplified in-vacuum water-cooled coil geometry to produce a biasedquadrupole trap. This lowers the temperature enoughto permit direct loading of a crossed dipole trap, whichallows fast and efficient evaporative cooling. The finalBECs are produced with 10 atoms. As well as the rela-tively small size and minimal number of coil turns, whichallow fast switch off times, a major advantage of thismagnetic trap design is that the small size allows goodoptical access to the atomic cloud. This will make theapparatus ideal for future optical lattice experiments [20].A number of magnetic trap designs are used in BEC a r X i v : . [ c ond - m a t . qu a n t - g a s ] F e b Optical dipole beamsImaging beamCCDCameraZeeman slower beam1 st MOT beams Push beamQuantization coil2 nd Zeeman slower 1 st Zeeman slower Source chamberCollimatorHe inlet (b)
Cross opticaldipole beams2 nd MOT beamsMCP Bias coil2cmQuad coils1 st MOT chamber 2 nd MOT chamber (a) c m nd MOT beams x yzy z xz xy DLD
FIG. 1. Schematic of the experimental apparatus (see main text for details). (a) Top view of the experimental setup. Abeam of He ∗ atoms is produced in a cryogenically liquid nitrogen cooled source, before being optically collimated, slowed andcooled in a 1st MOT stage. This forms a source for the 2 nd MOT, where the atoms are then transferred to a magnetic trapand subsequently to an optical dipole trap, where the BEC is generated. (b) Side view of the 2 nd MOT chamber, showingthe quadruple-Ioffe configuration in-vacuum trap and MCP-DLD detector. The inset shows the trapping coil geometry from adifferent angle. experiments, with some commonly used ones includingthe cloverleaf design [30], the Quadrupole-Ioffe configu-ration (QUIC) trap [31] and atomic chip traps [32]. Anydesign has strengths and weaknesses, and inevitably in-volves compromises and trade-offs between factors suchas confinement, depth, stability, switch off time, physicalsize, current required and heat dissipation. In our mag-netic trap setup, no evaporation is implemented; we onlyperform 1D Doppler cooling in the trap. Hence we re-quire a large enough bias to split the atomic energy levels(as well as prevent Majorana and Penning losses), but arenot so concerned about the usual consideration of havinga tight confinement, as all evaporation is conducted inthe dipole trap. The trap is in a simple QUIC configura-tion, consisting of two 5 turn anti-Helmhotz (AH) coils of40 mm diameter centred on the y axis and an 8 turn biascoil of 20 mm diameter centred on the z axis, as shown inFig. 1. These coils are made from 2 mm hollow coppertubing (internal diameter 0.4 mm) mounted in-vacuumto ensure they are close to the atoms. Cooling is pro-vided by pumping chilled water through the tubes, witheach tube attached to hollowed out 1 / ∼ . z axis from the centre ofthe AH coils. The radial and axial trap frequencies are ω ⊥ = 2 π ×
89 Hz in the radial axes and ω z = 2 π ×
57 Hzrespectively.A cold source of He ∗ atoms suitable for loading intoa MOT is generated from our He ∗ beamline, shownschematically in Fig. 1. Helium atoms are excited intothe metastable state via a hollow cathode DC dischargesource [34], which is cryogenically cooled using liquid ni-trogen. The expanding atomic beam is then collimatedusing an optical collimation stage featuring 4 beamspropagating perpendicular to the He ∗ atoms with an in-tensity ∼ I sat ( I sat = 0 .
167 mW/cm ) and detuned by ∼ −
5Γ (Γ / π = 1 . S → P transition, with a negativesign indicating red detuning. All laser cooling and imag-ing beams are sourced from a home built external gain-chip laser [35] with linewidth < < ∼ I sat intensity and ∼ − − − − z (m ) B z ( G ) FIG. 2. The magnetic trap potential along the vertical axisˆ z for our in-vacuum magnetic trap. The trap minimum is 10Gauss located 6 mm above the centre of the quadrupole coils,while the total trap depth is ∼
25 Gauss. beams having ∼ I sat intensity (horizontal beam), 140and 110 I sat intensity (vertical beams), all with ∼ −
22Γ detuning. However, the background pressure from thesource is too high for this chamber to be used to forma BEC. The horizontal MOT retro mirror located in-side the vacuum chamber has a ∼ . nd MOT chamber, where thepressure is < × − Torr. To assist this process, anadditional ‘push’ beam is added ( ∼ I sat intensityand ∼ .
3Γ blue detuned), forming a low velocity intensesource scheme (LVIS) [26]. The flux from the LVIS, asmeasured ∼
10 cm beyond the location of the 2 nd MOT,is ∼ × atoms/s, measured using a Faraday cup ona rotation stage and a picoammeter.The second MOT consists of a quadrupole magneticfield generated by our in-vacuum AH coils with a gra-dient along the tight axis of B (cid:48) ∼ . ∼ −
33Γ and intensities of ∼ I sat , ∼ I sat and ∼ I sat . In 1s we load a cloud with ∼ . × atoms at a temperature of ∼ . ∼ n ( t ) = Av π (cid:18) ( gt / l ) t (cid:19) exp (cid:18) − ( gt / − l ) v t (cid:19) (1)where A = ( m/ πkT ) / , v = (cid:112) (2 kT /m ), m is themass of a He atom, k is Boltzmann’s constant and l is the falling distance. From this distribution we extractthe temperature T , with the only other free fit parameterbeing the amplitude A .Following loading, the MOT is then compressed byramping the magnetic field down to 1 . ∼ − .
4Γ and the intensity of the MOT beams are re-duced by two orders of magnitude. The MOT beams arethen switched off and the atoms in the m J = +1 state re-main trapped in the quadrupole magnetic field of the AHcoils. To improve the trap depth and atom density wetighten the quadrupole field to 16 . µ s. Thebias coil current is then ramped up in 100ms to form aQUIC trap with a 10G field offset, removing the magneticfield zero in the quadrupole trap. This configuration pro-vides weak trapping frequencies of ω r ∼ π × ω z ∼ π × ∼ . × atoms at ∼ . ∼ . I sat ,aligned vertically along the bias field axis, circularly po-larised to drive the σ + transition and red detuned by ∼ − Γ /
2. This Doppler cooling stage cools the cloud in500ms, after which we have ∼ . × atoms at ∼ µ K.This is sufficiently cold to directly transfer the atomsfrom the magnetic trap to the crossed optical dipole trap.The final trapping stage is a crossed optical dipole trap(CODT), formed from two intersecting laser beams fardetuned from the 1083nm helium resonance at 1550nmwavelength, from a 100kHz linewidth seed laser that isamplified to 30W of power. The power is distributedbetween the two beams such that after feedback loops toregulate the intensity and losses due to AOM and fibrecoupling the first beam (aligned along the y axis) has upto 6W and the second beam (aligned ∼ ◦ from the x axis in the x − y plane) has up to 1W of power. Thesebeams are focused down to Gaussian 1 /e waists of 73 µ mand 55 µ m, respectively, at the location of the atoms. Thetwo beams intersect at the centre of the 10G biased QUICtrap, forming a CODT with trapping frequencies ω x,y,z ∼ π × (1 . , . , . ∼ µ K atfull power. t (s) H e * f l u x ( a r b . un i t s ) datafit 0.27 0.28 0.29 0.30 0.31 0.32 0.33 t (s) H e * F l u x ( a r b . un i t s ) datacombined fitBECThermal 0.27 0.28 0.29 0.30 0.31 0.32 0.33 t (s) H e * F l u x ( a r b . un i t s ) datacombined fitBECThermal (a) (b) (c) FIG. 3. Time of flight signal traces taken from the MCP after the dipole trap is switched off and atoms are allowed to fallonto the detector. The three plots show different points in the evaporation sequence. Data is shown as blue circles, withfits shown in red. Above T c the fit is to a Maxwell-Boltzmann distribution (see text for details), while below T c a parabolicThomas-Fermi fit to the BEC component (green dot-dash line) is added to the thermal fit (black dashed line). The conditionsfor the plots shown are: (a) reduced temperature T /T c = 1.02(5), trap frequencies ω x,y,z ∼ π × (1100 , , Hz , totalatom number N = 3 . × . (b) T /T c = 0.88(7), ω x,y,z ∼ π × (720 , , Hz , N = 1 . × , condensate atomnumber N = 0 . × . (c) T /T c = 0.54(6), ω x,y,z ∼ π × (320 , , Hz , N = 1 . × , N = 1 . × . Eachimages represents data from 20 shots Since the 1D Doppler cooling process is more efficientat higher densities [25], we observe that the Doppler cool-ing is more effective with the dipole beams on, in addi-tion to the magnetic trap. Here the dipole beams mostlyserve to increase the density of the atomic cloud, makingthe cooling process more efficient and reducing the finaltemperature after 1D Doppler cooling. In our experi-ment, this improves the final atom number by ∼ ∼ ∼
50 cmdiameter square) coil mounted above the vacuum cham-ber was used to generate a uniform magnetic field of ∼ z direction. This field is switched onbefore the current through the QUIC trap starts ramp-ing down. However, once in the dipole trap, it was foundthat the background magnetic field (dominated by the z component of the Earth’s magnetic field) was sufficientto preserve the quantisation, so the extra quantisationcoil was switched off 700 ms after transfer to the CODT.Immediately after transfer we measure ∼ × atomsat 13 . µ K with a phase space density of ∼ . ∗ experiment[29], we did not observe any improvement in the evapo-ration process by adding a gradient to cancel the weakcomponents of the dipole potentials along the beam prop-agation directions.To determine the important parameters of the gas inthe CODT, we switch off the dipole trap and allow theatoms to fall under gravity onto the MCP to measurethe 1D TOF profile along the z axis, integrated in thex-y dimensions. Examples of the resulting integrated1D TOF profiles are shown in Fig. 3. Above the con-densation critical temperature T ∗ c (note that T c refersto the non-interacting critical temperature, while T ∗ c isthe critical temperature corrected for interactions [37]),the profile is fitted to the same Maxwell-Boltzmann dis-tribution as in Eqn. 1 to yield the temperature T .Below T ∗ c we fit with a bi-modal distribution consist-ing of the same Maxwell-Boltzmann distribution plusan inverted parabola to represent the far-field Thomas-Fermi density profile of the BEC [37]. The half width ofthis parabola is the far field Thomas-Fermi radius R T F ,from which we can extract the chemical potential µ from µ ≈ mR T F / t T OF [38]. From the chemical potentialwe can extract the number of atoms in the condensate N = (2 µ ) / / √ m (cid:126) ¯ ω a , where a is the He ∗ s − wavescattering length. The number of atoms in the thermalcloud N T is given by N T ≈ .
202 ( k B T / (cid:126) ¯ ω ) , for T < T ∗ c .To extract the atom number above T ∗ c , we sum the in-tegrated counts, correct for the fraction of the cloud attemperature T that will hit the detector and then scaleby an effective quantum efficiency of the detector γ QE .We vary γ QE until the thermal atom number N T matchesfor the clouds just above and below T ∗ c . Note that non-linearities in γ QE at high fluxes mean this is not accuratefor N .The above information, along with the total atom num-ber N = N + N T and k B T c = 0 . (cid:126) ¯ ωN / , is then com- T / T c N / N Theory - non-intTheory - intData
FIG. 4. Condensate fraction N /N verses reduced temper-ature T /T c for different points in the evaporation sequence.Data is shown as blue points, with error bars dominated by fituncertainties due to the condensate and thermal widths beingrelatively similar for many data points, making accurate fitsdifficult. The brown dot-dash line shows the non-interactingtheory, while the red dashed line shows a correction account-ing for two-body interactions [37]. For N /N (cid:38) . bined to produce Fig. 4. This shows the condensatefraction N /N vs the reduced temperature T /T c for oursequence as we evaporate through the transition temper- ature. We cross T ∗ c at ∼ µK with ∼ × atoms,and by evaporating further we are able to produce BECswith no discernible thermal fraction (not shown on thegraph) with ∼ × atoms.We demonstrate a comparatively simple and shortmethod for producing Bose Einstein condensates of He ∗ atoms. By using a magnetic trap with a 10G bias gen-erated from an in vacuum set of coils, we are able tocool the cloud below 100 µK in the magnetic trap via1D Doppler cooling. The degenerate state of metastablehelium atoms is accomplished via direct evaporative cool-ing in a crossed optical dipole trap. We cross the con-densation critical temperature with ∼ × atoms at ∼ µ K and by evaporating further are able to generateessentially pure BECs with ∼ × atoms. The entiresequence to produce a BEC only takes 3.3 sec. This willprovide an excellent starting point for a range of exper-iments with He ∗ BECs, such as many-body correlationexperiments [22], probes of quantum non-locality [17] orstrongly interacting optical lattice physics [20].
ACKNOWLEDGMENTS
The authors would like to thank David Clement andRobert Dall for helpful discussions, and Colin Dedman,Ross Tranter and Liangwei Wang for technical assistance.This work was supported through Australian ResearchCouncil (ARC) Discovery Project grants DP160102337and DP190103021. SSH was supported by ARC Discov-ery Early Career Researcher Award DE150100315. [1] M. H. Anderson, J. R. Ensher, M. R. Matthews, C. E.Wieman, and E. A. Cornell, Science , 198 (1995).[2] C. C. Bradley, C. A. Sackett, J. J. Tollett, and R. G.Hulet, Phys. Rev. Lett. , 1687 (1995).[3] K. B. Davis, M. O. Mewes, M. R. Andrews, N. J. vanDruten, D. S. Durfee, D. M. Kurn, and W. Ketterle,Phys. Rev. Lett. , 3969 (1995).[4] I. Bloch, J. Dalibard, and W. Zwerger, Rev. Mod. Phys. , 885 (2008).[5] N. R. Cooper, J. Dalibard, and I. B. Spielman, Rev.Mod. Phys. , 015005 (2019).[6] R. Geiger, A. Landragin, S. Merlet, and F. PereiraDos Santos, AVS Quantum Science , 024702 (2020).[7] H. Ott, Rep. Prog. Phys. , 054401 (2016).[8] W. S. Bakr, J. I. Gillen, A. Peng, S. F¨olling, andM. Greiner, Nature , 74 (2009).[9] J. F. Sherson, C. Weitenberg, M. Endres, M. Cheneau,I. Bloch, and S. Kuhr, Nature , 68 (2010).[10] R. B¨ucker, A. Perrin, S. Manz, T. Betz, C. Koller, T. Plis-son, J. Rottmann, T. Schumm, and J. Schmiedmayer,New Journal of Physics , 103039 (2009).[11] A. Bergschneider, V. M. Klinkhamer, J. H. Becher,R. Klemt, G. Z¨urn, P. M. Preiss, and S. Jochim, Phys.Rev. A , 063613 (2018).[12] W. Vassen, C. Cohen-Tannoudji, M. Leduc, D. Boiron, C. I. Westbrook, A. Truscott, K. Baldwin, G. Birkl,P. Cancio, and M. Trippenbach, Rev. Mod. Phys. ,175 (2012).[13] M. Schellekens, R. Hoppeler, A. Perrin, J. V. Gomes,D. Boiron, A. Aspect, and C. I. Westbrook, Science ,648 (2005).[14] A. G. Manning, R. I. Khakimov, R. G. Dall, and A. G.Truscott, Nature Phys. , 539 (2015).[15] R. Lopes, A. Imanaliev, A. Aspect, M. Cheneau, D. Bo-iron, and C. I. Westbrook, Nature , 66 (2015).[16] Khakimov R. I., Henson B. M., Shin D. K., Hodgman S.S., Dall R. G., Baldwin K. G. H., and Truscott A. G.,Nature , 100 (2016).[17] D. K. Shin, B. M. Henson, S. S. Hodgman, T. Wasak,J. Chwedenczuk, and A. G. Truscott, Nature Commu-nications , 4447 (2019).[18] T. Jeltes, J. M. McNamara, W. Hogervorst, W. Vassen,V. Krachmalnicoff, M. Schellekens, A. Perrin, H. Chang,D. Boiron, A. Aspect, and C. I. Westbrook, Nature ,402 (2007).[19] R. Chang, Q. Bouton, H. Cayla, C. Qu, A. Aspect,C. I. Westbrook, and D. Cl´ement, Phys. Rev. Lett. ,235303 (2016).[20] H. Cayla, S. Butera, C. Carcy, A. Tenart, G. Herc´e,M. Mancini, A. Aspect, I. Carusotto, and D. Cl´ement, Phys. Rev. Lett. , 165301 (2020).[21] Dall R. G., Manning A. G., Hodgman S. S., RuGway Wu,Kheruntsyan K. V., and Truscott A. G., Nature Phys. , 341 (2013).[22] S. S. Hodgman, R. I. Khakimov, R. J. Lewis-Swan, A. G.Truscott, and K. V. Kheruntsyan, Phys Rev Lett ,240402 (2017).[23] A. Robert, O. Sirjean, A. Browaeys, J. Poupard,S. Nowak, D. Boiron, C. I. Westbrook, and A. Aspect,Science , 461 (2001).[24] F. Pereira Dos Santos, J. L´eonard, J. Wang, C. J. Bar-relet, F. Perales, E. Rasel, C. S. Unnikrishnan, M. Leduc,and C. Cohen-Tannoudji, Phys. Rev. Lett. , 3459(2001).[25] A. S. Tychkov, T. Jeltes, J. M. McNamara, P. J. J. Tol,N. Herschbach, W. Hogervorst, and W. Vassen, Phys.Rev. A , 031603 (2006).[26] R. Dall and A. Truscott, Opt. Commun. , 255 (2007).[27] M. Keller, M. Kotyrba, F. Leupold, M. Singh, M. Ebner,and A. Zeilinger, Phys. Rev. A , 063607 (2014).[28] A. S. Flores, H. P. Mishra, W. Vassen, and S. Knoop,Appl. Phys. B , 391 (2015).[29] Q. Bouton, R. Chang, A. L. Hoendervanger, F. Nogrette,A. Aspect, C. I. Westbrook, and D. Cl´ement, Phys. Rev.A , 061402 (2015).[30] M.-O. Mewes, M. R. Andrews, N. J. van Druten, D. M.Kurn, D. S. Durfee, and W. Ketterle, Phys. Rev. Lett. , 416 (1996). [31] T. Esslinger, I. Bloch, and T. W. H¨ansch, Phys. Rev. A , R2664 (1998).[32] J. Fort´agh and C. Zimmermann, Rev. Mod. Phys. ,235 (2007).[33] Note that standard solder and flux was also found towork, although required an initial period under vacuumto remove the residual flux before being installed in thefinal UHV chamber. This prevented out-gassing from thenon-vacuum compatible flux, was sufficient to still be ableto reach UHV pressures.[34] J. Swansson, K. Baldwin, M. Hoogerland, A. Truscott,and S. Buckman, Appl. Phys. B , 485 (2004).[35] D. K. Shin, B. M. Henson, R. I. Khakimov, J. A. Ross,C. J. Dedman, S. S. Hodgman, K. G. H. Baldwin, andA. G. Truscott, Opt. Express , 27403 (2016).[36] I. Yavin, M. Weel, A. Andreyuk, and A. Kumarakrish-nan, American Journal of Physics , 149 (2002).[37] F. Dalfovo, S. Giorgini, L. P. Pitaevskii, and S. Stringari,Rev. Mod. Phys. , 463 (1999).[38] While this approximation for the chemical potential fromthe far field R TF is only valid for condensates in cigarshaped traps, numerical solutions of the equations of mo-tion following [39] show that it gives at most an ≈ µ for the trap geometries in our system, which iswithin our experimental uncertainties.[39] Y. Castin and R. Dum, Phys. Rev. Lett.77