Transverse laser cooling of a thermal atomic beam of dysprosium
N. Leefer, A. Cingöz, B. Gerber-Siff, Arijit Sharma, J. R. Torgerson, D. Budker
aa r X i v : . [ phy s i c s . a t o m - ph ] M a r Transverse laser cooling of a thermal atomic beam of dysprosium
N. Leefer, ∗ A. Cing¨oz † , B. Gerber-Siff, Arijit Sharma, J. R. Torgerson, and D. Budker
1, 5, ‡ Department of Physics, University of California at Berkeley, Berkeley, CA 94720-7300, USA Swarthmore College, Swarthmore, PA 19081, USA Raman Research Institute, Sadashivanagar, Bangalore 560080, India Physics Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA Nuclear Science Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA (Dated: October 22, 2018)A thermal atomic beam of dysprosium (Dy) atoms is cooled using the 4 f s ( J = 8) → f s p ( J = 9) transition at 421 nm. The cooling is done via a standing light wave orthogo-nal to the atomic beam. Efficient transverse cooling to the Doppler limit is demonstrated for allobservable isotopes of dysprosium. Branching ratios to metastable states are demonstrated to be < × − . A scheme for enhancement of the nonzero-nuclear-spin-isotope cooling, as well as amethod for direct identification of possible trap states, is proposed. PACS numbers: 06.20.Jr, 32.30.Jc, 37.10.De, 37.10.Vz
I. INTRODUCTION
Laser cooling and trapping of atoms has been a pro-lific area of research, with applications to the study ofnovel quantum phases [1, 2], matter-wave interferom-etry [3], quantum information [4], and precision mea-surements and tests of fundamental physics [5]. Atomsamenable to direct laser cooling have recently expandedoutside of the alkalis and noble gases to include elementssuch as chromium, indium, ytterbium and mercury [6–9].The lanthanides are of particular interest. The generallylarge ground-state magnetic moments of these elementspresent opportunities for new studies in areas such as de-terministic single atom sources [10], degenerate dipolarFermi gases [11], and quantum computing [12]. Whilethe complex electronic structures and numerous possi-ble decay channels for excited states do not make mostlanthanides obvious candidates for laser cooling, someprogress has been made. In erbium, the large magneticmoment allows for magnetic confinement in the trap re-gion while excited states recycle to the ground state, per-mitting laser cooling to low temperatures [13]. A success-ful effort to form a Dy magneto-optical trap (MOT) hasdemonstrated that the same principle works for dyspro-sium [14].In this work, we explore transverse laser cooling of dys-prosium (Dy, Z=66) in an atomic beam. In addition tosharing the largest ground-state magnetic moment in theperiodic table with terbium ( µ ∼ µ B ), Dy contains apair of nearly degenerate, opposite-parity states the havebeen investigated in detail [15]. These levels have re-cently been used in a search for a variation of the fine- † currently at: JILA, National Institute of Standards and Technol-ogy and University of Colorado at Boulder, Boulder, CA 80309-0440, USA ∗ Electronic address: [email protected] ‡ Electronic address: [email protected] structure constant that does not rely on optical-frequencycombs [16, 17]. The implementation of laser cooling inthis experiment as a control over atomic-beam parame-ters is a crucial aspect of systematics management [18].Previously our group reported on the manipulation of athermal beam of Dy atoms using a bright, visible transi-tion in Dy at 421 nm [19] (which was the first report oflaser cooling in Dy) and the spectroscopy of this transi-tion [20]. In this work, we demonstrate successful trans-verse laser cooling of a thermal beam for all isotopes ofdysprosium without repumping, propose a method forthe simultaneous cooling and optical pumping within thehyperfine manifold of the odd-neutron-number isotopeswith non-zero nuclear spin, and show that the maxi-mum possible branching ratio of the cooling transitionto metastable states is less than 5 × − . II. EXPERIMENTA. Dysprosium
Dysprosium has seven stable isotopes, ranging from A = 156 → , Dy, withnuclear spin I = 5 /
2. Cooling is accomplished using the4 f s ( J = 8) → f s p ( J = 9) transition at 421nm [20]. Cooling of the odd-neutron-number isotopeswas accomplished using the only closed transition in thehyperfine manifold from F = 10 . → F ′ = 11 .
5. The
TABLE I: Parameters of the cooling transitionParameter ValueWavelength λ c τ v r = ~ k/M v d = p ~ / Mτ I sat = hcπ/ τ λ c
60 mW/cm relevant cooling parameters of this system are given inTable I. B. Apparatus
This work was carried out using a beam source pre-viously used in a search for a variation of the fine-structure constant, a search for parity nonconservation(PNC), and rf-spectroscopy of nearly degenerate levelsin Dy [15, 17, 22, 23]. A detailed description of theatomic-beam source is given in Ref. [23]. The beam isproduced by an effusive oven with a multislit nozzle ar-ray operating at ≃ ≃ × cm/s.The 1D transverse velocity profile was measured via flu-orescence spectroscopy to be Gaussian (Eqn. 1), with σ v = 1 . × cm/s.To generate the 421-nm light, 12 W of Ar-ion laser light(Coherent Innova 400) was used to pump a Ti:Sapphirering laser (Coherent 899), producing up to 650 mW of842-nm light. After diagnostics and passing through anoptical fiber, approximately 300 mW of 842-nm light en-tered a resonant bow-tie cavity containing a 1 × × periodically poled potassium titanyl phosphate(PPKTP) crystal (Raicol Crystals Ltd.) located at the22- µ m beam waist of the cavity. The output power of thefrequency doubler was consistently 50-60 mW. Outputspowers of >
100 mW were observed, but at powers above60 mW, gray tracking in the crystal caused rapid degra-dation of the conversion efficiency [24]. It was determinedexperimentally that laser-cooling did not noticeably im-prove above ∼
30 mW of light power, so the cavity wasoperated at sufficiently lower power to eliminate notice-able degradation. The 421-nm light passed through aquarter- and a half-wave plate to adjust the polariza-tion to vertical and a 4x cylindrical telescope to create a0 . × . µ s. Assuminga cycle time of 4 τ ≃
20 ns for a saturation parameter of ∼ × cycles.To reduce long term drifts of the laser frequency,the 421-nm light was scanned over the desired reso-nance while monitoring fluorescence with a photomul-tiplier tube (PMT) oriented at 90 degrees with respectto the laser and atomic beams. Once at the desired fre-quency, the Ti:Sapph laser was stabilized using a scan-ning transfer-cavity lock [26]. In this control system, lightat 842 nm from the Ti:Sapph laser is injected into a con-tinuously scanning Fabry-Perot (FP) cavity along withlight from a metrological HeNe (Spectra-Physics 117A) Ar-ion Ti:Sapph842 nm HeNeAr-ionDye Laser658 nmFPPDmicrocontroller PDλ/2 λ/2λ/4lin.pol. irisFrequency Doubler fiber launcherfiber launcher DyOven PMTPMT
FIG. 1: (Color online) Simplified layout of the experimen-tal system. Light from two Ar-ion lasers pumped a Coherent899 Ti:Sapphire and Coherent 699 dye laser respectively. Theoutput of the Ti:Sapph laser was split, with 10% of powergoing to diagnostics and the frequency stabilization system.The rest was fiber coupled to the frequency doubling system,the output of which was expanded and polarized before en-tering the vacuum chamber. Light from the dye laser wasfiber coupled to the beam apparatus, where it passed throughpolarization optics and an iris diaphragm before entering thechamber. with a frequency drift rate of < . < ∼
15 cm to the back of the apparatus for de-tection. The Doppler profile of the atomic beam wascharacterized by scanning a probe-laser frequency overthe 4 f s ( J = 8) → f d s ( J = 7) transition at658 nm. For a Gaussian velocity distribution describedby P ( v ) = 1 √ πσ v e − v σ v (1)with characteristic half-width 1/e velocity 2 σ v , the pureDoppler spectral profile is also Gaussian with the rela-tionship σ f = kσ v / π , where k = 2 π/λ . The Dopplercooling limit, v d = 19 . σ v D = 300 kHz at 658nm. Thus, the 658-nm transition is an ideal probe ofthe atoms’ Doppler profile, as the natural width of thetransition, 120 kHz [21], is considerably smaller than theDoppler limited width. The sensitivity in our experimentwas ultimately limited by the probe-laser linewidth; ef-fects such as magnetic-field broadening and power broad- æææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ ííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííí ææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ æ æææ ííííííííííííííííííííííííííííííííííí æææææææææ íííííí ææææææææææææææææææææ ííííííííí ææææææææææææææ ííí æææ íííí æææææææææææææææ íí æææææææææææææææææææææææææææææææææææææ íí ææææææ ííí ææææææææææææææææææææææææææææææææææ íííí ææææææææææ ííííííí ææ íí ææ íííí æææææ ííí ææææææææææææææææææææææææææææ ííí ææææææææææææææææææææææææææææææææææææææ íííííííí æ íí ææ íííííííí æææææææææææ íí ææææ ííííííííí ææ íííííííííííí æææææææææææææ íííííí ææææææææ íííííííííííííííííííííííííííííííííííííííííííííííííííííííííí æææææ ííííí ææ íííííííííííííííííííííííííííííííííííííííííí æ íííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííí ææææææ íí æææææ í ææææææ ííííííí æææææææææ íííí ææææææææ ííííííííííííííí ææææææææææææææ íííííí ææ íííí ææææææææææææ íííí æææææ íííííí ææææææææææææ ííííí æææææææææææ í ææ íí ææææ í ææææææææææ íí ææææææææææææ íííí ææææææææææææææææææææææææææææææ í ææææææ ííí ææææææææææ íííííí ææææææææææææææææææææææææææææ íííííí ææææææææææææææææ ííííí æææææææææææææææææææ ííí ææææææææææææææ í æææææææææææææææææ íííí æææææææ íííííííí ææææææææææ íííí æææææææææææææ íííí æææææ ííííííí ææææææææææææææææææ í æææ í æææææ ííííííí æææææææææææææææ ííííííííí ææææææææææ ííííííííííííí æææ í æææ í ææ íííííííííííííííí æææ ííííííííí æ íí ææææææææ íííííí æææææææ ííííííí ææææ íííííííííííííííí æææææææ íííííííííííííííííííííííííííííííííííííííííí æ íííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííí ææ íííííííí ææ íííííííííííííííííííí ææææææ ííí æææææææææ íííí ææææææææææ íííííí ææææ íííí ææ ííííííííí æææææææææææ íííí ææææææææææææææææææææææææææææææææææææææææææææææææ íííííí æææ ííí æææææ íí ææææææææææææææææææææææææææææææææ íííí æææ ííííííí ææææ í ææ í ææææ ííí ææ íí æææ íí æææææææææææ íííí ææææææææ íííííííí ææ íí ææ í æææ íí ææææææ íííí ææææ ííí æ í ææ íííííí ææææ íííííí æææææææ ííííííííííí ææ ííííííí æææææææææææææææææææ í æ í æææææ ííí æ ííí ææææææææææææææ ííííí ææææ íííí ææææ íí ææ ííííí ææ íí ææææææææææ ííííííí æææææ ííííííííííííí ææææ íííííííííííííííííííííííííííííííííííííííííí æææ íííííííííííííííííí æææ íííííííííííí æææææææææææææææææææææææææææææ í æææ ííí ææææææ íííííí æææææææ ííí æææææææææ ííííííí æ ííííí ææææ ííí æ í æ íííííííííííííííííí æ í ææææ íííííííí ææ ííí æææææ ííííííííííííííííííí ææææ ííííííííííííí ææ í ææ ííííííí æææææææ ííííí ææææææææ íí ææææææææ ííí æææææææææææ íííííí æææææææ íííííííí æææææææ íí æææææææææ ííííí æææææ ííííííííí æææ íí ææææ íííí ææææææææ ííííííííííííííííííííííííííííííííííííííííííííííííí ææææææææææææ íí æææææææ ííííí ææææææ ííííííí ææ íííííííííííííííííííííííííííííí ææææ ííí æææææ íí æææ íííííííííííííííííííííí ææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ íí æææ íííí æææææææ ííí æææææææææææææææ í ææææææææææææææææææææææ ííííííííííííí ææææææææææææææææææææææææææææææææææææææææææææææææææææ ííííí ææææ íí ææ ííííííí æææææ ííííííí æææææ íí æææææææææææ ææææææ íí ææ íí -
50 0 50 100 1500.280.300.320.340.360.380.400.42 Frequency H MHz L P M T S i gn a l H V L æææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ - -
10 0 10 20Frequency H MHz L FIG. 2: (Color online) Example probe spectrum of the
Dy658-nm transition with 40 mW of cooling light. The un-cooled Doppler profile is shown by the unfilled diamonds.Optical pumping in the cooling region to magnetic sublevelswith larger Clebsch-Gordan coefficients accounts for the non-conservation of area under the curve. The solid red line is anonlinear least-squares fit to the data. The inset shows thenarrow feature with the broad background subtracted and asingle Voigt profile fit to the remaining peak. ening were controlled and maintained at a smaller level.The 658-nm light was produced using a Coherent 699dye laser pumped by 5.5 W of Ar-ion laser light. Af-ter passing through an optical fiber and diagnostics, ≃ µ W of vertically polarized light in a 4-mm diameterbeam entered the apparatus and intersected the atomicbeam at a ninety-degree angle. Fluorescence from theprobe transition was detected using a PMT oriented per-pendicular to the laser and atomic beams, with a 660-nm central wavelength interference filter (10-nm band-width) on the detection port. The dye-laser frequencywas scanned continuously over the transition, while thefluorescence signal from the PMT and the signal from areference photodiode (PD) were simultaneously acquiredusing an Agilent DSO5014A oscilloscope. Normalizingthe fluorescence signal with the reference PD signal en-sured that fluctuations in the probe laser power did notcontribute to features in the Doppler profile.
C. Results
A scan over the 658-nm
Dy resonance is shown inFig. 2. The broad uncooled Doppler profile has a Gaus-sian profile with characteristic width, σ f i = 20 MHz,corresponding to σ v i = 1 . × cm/s. At low tempera-tures the measured lineshape is expected to be describedby a Voigt profile, a convolution of the Doppler spectrumof the atoms and the Lorentzian lineshape of the probelaser, given by æææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ ííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííí a L - -
50 0 50 1000.340.360.380.400.420.440.46 Frequency H MHz L P M T S i gn a l H V L æææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ ííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííííí b L - -
50 0 50 1000.320.330.340.350.360.37 Frequency H MHz L P M T S i gn a l H V L P M T S i gn a l H V L FIG. 3: (Color online) Probe spectra of the a)
Dy and b)
Dy 658-nm, F = 10 . → . ∼
40 mWof cooling light. The unfilled diamonds show the uncooledDoppler profile, and the solid red line is a nonlinear least-squares fit with a Voigt profile used for the narrow feature.The slope on the right side of a) is from the edge of the
Dytransition. V ( f ) = (2) Z ∞−∞ √ πσ f c exp − ( f − f ′ ) σ f c ! π γ L / f ′ + γ L / df ′ , where σ f c and γ L are the Gaussian width of the cooledthermal distribution and the full-width half-maximum(FWHM) of the Lorentzian describing the laser spec-trum, respectively. For the best cooling, a good fit wasachieved using a combination of a Gaussian and a Voigtprofile: a Gaussian profile for the uncooled backgroundand a Voigt profile for the narrow feature (see inset inFig. 2). The uncooled background is due to atoms outsidethe capture range of the cooling laser. The values for σ f c and γ L obtained from a least-squares fit are 0 . . σ f c gives σ v c = 53(33) cm/s, which is consistent with the expectedDoppler cooling limit of 19.8 cm/s (Table I). The valuefor γ L is also consistent with a separate measurement ofthe laser linewidth (Sec. II D).Laser cooling of Dy and
Dy, both of which havehyperfine structure, was accomplished by tuning the cool-ing laser onto the closed F = 10 . → . F is the total angular momentum (Fig. 3). TheGaussian widths of the Voigt profiles fit to the narrowfeatures of Dy and
Dy spectra are 0 . . æææ æææææææ æ æ æ æææææ æ ææææææææææææ æ æ æ æææææ æ æ æææææ æææææææ æ æ æ ææææ æ æ ææææ S Ν H f L = S - f + S ´ ´ ´ ´ ´ ´ Frequency H Hz L S Ν H H z (cid:144) H z L FIG. 4: (Color online) Power spectral density of frequencyfluctuations of the dye laser used to observe the 1D tempera-ture of the atomic beam. The solid red line is a least-squaresfit to a function of the form given by Eqn. 3
The number of cycles can be estimated from the initialvelocity, ( σ v i = 1300 cm/s), the final velocity, and therecoil velocity given in Table I as N ≈ ( σ v i − σ v f ) /v r ≈ σ v i /v r ∼ /N < × − . This result is in agreement witha more sensitive measurement of branching ratios thathas been performed in a MOT, yielding 7 × − [14]. D. Broadening Mechanisms
Our measurement of the laser cooled transverse tem-perature of the atomic beam was ultimately limited bythe linewidth of the dye laser. The linewidth was mea-sured by coupling the 658-nm light into a FP cavity sta-bilized to a metrological HeNe (Spectra-Physics 117A).The dye laser was tuned to the side of a transmissionpeak, where the slope is linear and the fluctuations inthe transmitted power were recorded with a spectrumanalyzer. The known free-spectral range (FSR) and mea-sured finesse of the cavity allowed us to to convert thevoltage power spectral density (PSD) of the fluctuationsto the frequency PSD [ S ν ( f )]. The result of this mea-surement is shown in Fig. 4. A function of the form, S ν ( f ) = S − f + S (3)was fit to the data, where S − and S are the 1 /f - andwhite-noise contributions, respectively [27]. The least-squares fit values for S − and S are 2 . × Hz and9 . × Hz respectively. Interpretation of the shapeof the laser linewidth from this data is non-trivial, butthe FWHM can be estimated by solving the equation [28]1 rad ≈ Z ∞ f dB S ν ( f ) f df, (4) where 2 f dB is the FWHM of the laser lineshape. Solv-ing Eqn. 4 for f dB we find a FWHM of 3.2(5) MHz, inagreement with the value obtained from a Voigt profilefit (Sec. II C).The most important remaining broadening mecha-nisms of the probe transition linewidth are magnetic-fieldbroadening and power broadening. The magnetic-fieldbroadening is given by∆ ν B = | ( m J u g u − m J l g l ) | µ B B, (5)where g u = 1 .
26 and g l = 1 .
24 are the g -factors of theupper and lower energy levels and m J u and m J l are themagnetic sublevels involved. The worst-case scenario ofdriving m → m ± ∼ ∼ I S , of the probetransition is approximately 50 µ W/cm , and the corre-sponding power broadened linewidth can be calculatedusing γ P = γ √ κ, (6)where γ is the natural linewidth and κ = I/I S is thesaturation parameter. With an approximate beam ra-dius of 2 mm, an input power of 50 µ W gives κ ≃ γ P ≃
360 kHz, well below the linewidth of our laser.In addition, data were taken as a function of probe laserpower to confirm the absence of measurable power broad-ening.
E. Hyperfine Structure
The dense structure of the 421-nm cooling transitiongives rise to interesting optical pumping dynamics (spec-troscopic information on the 421-nm spectrum can befound in Ref. [20]). In
Dy, all of the ∆ F = 0 transi-tions are unresolved due to the broad natural linewidthand Doppler width. Because the cooling transitionis a ∆ J = +1 transition in the uncoupled angular-momentum basis ( J is the total electronic angular mo-mentum), driving all of the ∆ F = 0 transitions makesthe ground state with the smallest value of F a darkstate. In Fig. 5, the 421-nm laser has been tuned ontothe cluster of ∆ F = 0 transitions while the probe laseris swept over the entire 658-nm spectrum, demonstratingoptical pumping of the atoms into the F = 5 . Dy. The 421-nm car-rier frequency can be tuned to drive the F = 5 . → . ∼ F = 0 : F = . : F = . : F = . : F = . : F = . , . : F = . , . , . : F = . : F = . : F = . a L - - - - -
500 0Frequency H MHz L P M T S i gn a l H V L b L - - - - -
500 0Frequency H MHz L S i gn a l H V L S i gn a l H a r b . s ca l e L FIG. 5: (Color online) Example of the optical pumping dy-namics observed. a) A typical spectrum of the 658-nm tran-sition. For the odd-neutron number isotopes, the strongest∆ F = − F = 0 transitions. SeeSection II E for details. transitions. The sideband will optically pump atoms to-wards the F = 5 . F = 5 . → . III. FUTURE WORK
The extension of laser cooling to include all observableisotopes of Dy has been demonstrated, with cooling of thenon-zero-nuclear-spin isotopes,
Dy and
Dy, accom-plished using the closed F = 10 . → . F = 10 . → . F = 10 . Dy is shown in Fig. 6.This can be accomplished by passing the 421-nmlight through a running-wave (non-resonant) electroopticmodulator [30]. The desired modulation frequencies canall be applied to the same crystal to generate sidebandsat the appropriate frequencies. Atoms being driven onthe F → F + 1 transitions will eventually be pumped tothe F = 10 . M H z M H z M H z M H z M H z M H z E n e r gy F = = = = = = = f H J = L f s H J = L FIG. 6: (Color online) Diagram of the
Dy hyperfine struc-ture. The relative transition strengths || d F || / || d J || areshown alongside transition arrows, where || d || is the reduceddipole matrix element normalized so that the F = 10 . → . F → F + 1transitions relative to the F = 10 . → . F = − ∆ J transitions are negligibly weak and are notshown. methods such as those described in Sec. II E may allowus to circumvent these limitations.Given the complex electronic structure of the dyspro-sium system, the existence of a closed optical transitionis unlikely. A theoretical calculation was performed thatestimated the total branching ratio to the most likely trapstates at less than the 5 × − level [31] and a measure-ment in a MOT obtained a value of 7 × − [14]. Detailedknowledge of the dominant trap states will be importantfor optimization of MOTs and future work with lasercooled and trapped Dy.Our current set-up is insensitive to trap states forwhich the time of return to the ground state is less thanthe relatively long time-of-flight, τ tof ∼ µ s, betweenthe cooling and probe regions. We have developed aprocedure to directly identify trap states. The 421-nmlight will be tuned on resonance with the Dy transi-tion while light from a pulsed dye laser operating between600 and 700 nm will be superimposed over the fluores-cence region. A PMT with a UV short-pass filter on theviewport will be oriented perpendicular to the atomicbeam and the cooling laser while the dye laser is scannedfrom 600 to 700 nm. An increase in the fluorescence sig-nal from the PMT will be taken to indicate a transitionfrom the trap state to a highly excited odd-parity state,the fluorescence from the ground state decay channel ofthis state being less than 300 nm. A diagram of likelyrepump wavelengths is presented in Fig. 7. H J = L H J = L H J = L H J = L H J = L H J = L H J = L H J = L H J = L H J = L H J = L H J = L . m . m . m . m . m . m ® m E n e r gy Even Odd
FIG. 7: Partial energy level diagram showing possible repumpwavelengths from the likely trap states. The circled state isthe upper level of the cooling transition. The trap statesand corresponding repump wavelengths were identified in theNIST Atomic Spectroscopy database.
IV. DISCUSSION
We have presented results on transverse cooling of athermal dysprosium beam. A final laser linewidth lim-ited spectrum of a narrow probe transition in the cooledbeam consistent with cooling to the Doppler limit wasobtained. Cooling has been demonstrated for all de-tectable isotopes of Dy. A separate measurement of thelaser linewidth was consistent with the hypothesis thatthe laser linewidth is the limiting factor in measuring theDoppler spectrum. An attempt was made at sub-Doppler cooling, but no narrowing of the line was observed dueto the large probe laser linewidth. The remaining dom-inant broadening mechanisms of the probing method,background magnetic-field and power broadening, wereestimated and controlled for. A plan for increasing theefficiency of cooling the odd-neutron number isotopes ofDy has been presented. The branching ratios of the cool-ing transition have been confirmed to be < × − anda procedure for direct identification of the dominant trapstates has been outlined.Future work includes using a femtosecond frequencycomb as a probe of the Doppler profile of the atomicbeam. The narrow width of the comb teeth should besufficient to directly observe Doppler and sub-Dopplercooling. The ultimate goal of this work is the extensionto 2D transverse cooling and longitudinal slowing in anexperiment dedicated to a search for variation of the fine-structure constant. Acknowledgments
The authors would like to acknowledge the contribu-tions of V. Flambaum and V. Dzuba for the theoret-ical calculations of branching ratios and H. M¨uller forcomments on the manuscript. We would like to thankA. Skliar of Raicol Crystals Ltd. for extensive techni-cal advice regarding the PPKTP crystal. This work hasbeen supported by the Foundational Questions Institute(fqxi.org), the UC Berkeley Committee on Research andNSF/DST Grant No. PHY-0425916 for U.S.-India coop-erative research. [1] L. Santos, G. V. Shlyapnikov, P. Zoller, and M. Lewen-stein, Phys. Rev. Lett. , 1791 (2000).[2] A. Schirotzek, Y. Shin, C. H. Schunck, and W. Ketterle,Phys. Rev. Lett. , 140403 (2008).[3] H. M¨uller, S.-W. Chiow, Q. Long, S. Herrmann, andS. Chu, Phys. Rev. Lett. , 180405 (2008).[4] C. Trefzger, C. Menotti, and M. Lewenstein, Phys. Rev.A , 043604 (2008).[5] S. N. Atutov, V. Biancalana, A. Burchianti, R. Calabrese,L. Corradi, A. Dainelli, V. Guidi, B. Mai, C. Marinelli,E. Mariotti, et al., Hyperfine Interactions , 83(2003).[6] Z. W. Tao, Z. B. Hua, X. X. Ming, Y. Z. Heng, Z. B. Wu,M. Yan, and L. T. Bao, Nuc. Instr. and Meth. in Phys.Res. B , 5171 (2008).[7] J.-I. Kim, D. Haubrich, and D. Meschede, Opt. Expr. ,21216 (2009).[8] C. Y. Park and T. H. Yoon, Phys. Rev. A , 055401(2003).[9] H. Hachisu, K. M. andS. G. Porsev, A. Derevianko, V. D.Ovsianniko, V. G. Pal’chikov, M. Takamoto, and H. Ka-tori, Phys. Rev. Lett. , 053001 (2008).[10] S. B. Hill and J. J. McClelland, Appl. Phys. Lett. ,3128 (2003). [11] B. M. Fregoso, K. Sun, E. Fradkin, and B. Lev, New J.Phys. , 103003 (2009).[12] H. Y. Ban, M. Jacka, J. L. Hanssen, J. Reader, and J. J.McClelland, Opt. Expr. , 3185 (2005).[13] J. J. McClelland and J. L. Hanssen, Phys. Rev. Lett. ,143005 (2006).[14] M. Lu, S. H. Youn, and B. L. Lev, Phys. Rev. Lett. ,063001 (2010).[15] D. Budker, D. DeMille, E. D. Commins, and M. S. Zolo-torev, Phys. Rev. A , 132 (1994).[16] A.-T. Nguyen, D. Budker, S. K. Lamoreaux, and J. R.Torgerson, Phys. Rev. A , 022105 (2004).[17] A. Cing¨oz, A. Lapierre, A.-T. Nguyen, N. Leefer, D. Bud-ker, S. K. Lamoreaux, and J. R. Torgerson, Phys. Rev.Lett. , 040801 (2007).[18] A. Cing¨oz, Ph.D. thesis, University of California, Berke-ley (2009).[19] N. Leefer, A. Cing¨oz, D. Budker, S. J. Ferrell, V. V.Yaschuk, A. Lapierre, A.-T. Nguyen, S. K. Lamoreaux,and J. R. Torgerson, in Proc. 7th Symposium FrequencyStandards and Metrology (2009), p. 34.[20] N. Leefer, A. Cing¨oz, and D. Budker, Opt. Lett. , 2548(2009).[21] J. J. Curry, E. A. D. Hartog, and J. E. Lawler, J. Opt. Soc. Am. B , 2788 (1997).[22] S. J. Ferrell, A. Cing¨oz, A. Lapierre, A.-T. Nguyen,N. Leefer, D. Budker, V. V. Flambaum, S. K. Lamoreaux,and J. R. Torgerson, Phys. Rev. A , 062104 (2007).[23] A.-T. Nguyen, D. Budker, D. DeMille, and M. Zolotorev,Phys. Rev. A , 3453 (1997).[24] J. P. F´eve, B. Boulanger, and G. Marnier, Appl. Phys.Lett. , 277 (1996).[25] H. Metcalf and P. v. d. Straten, Laser Cooling and Trap-ping (Springer, 2002).[26] J. H. T. Burke, O. Garcia, K. J. Hughes, B. Livedalen,and C. A. Sackett, Rev. Sci. Inst. , 116105 (2005). [27] F. Riehle, Frequency Standards: Basics and Applications (Wiley-VCH, 2004).[28] J. Hall, in
International School of Physics ‘Enrico Fermi’,Course CXVIII, Laser Manipulation of Atoms and Ions (1992), pp. 671–702.[29] W. Childs, Phys. Rev. A , 1692 (1970).[30] R. G. DeVoe and R. G. Brewer, Phys. Rev. A20