Optics Communications

Transcription

1 Optics Communications 290 (2013) Contents lists available at SciVerse ScienceDirect Optics Communications journal homepage: A simple and highly reliable laser system with microwave generated repumping light for cold atom experiments Daniel Sahagun b,n, Vasiliki Bolpasi a, Wolf von Klitzing a a Institute of Electronic Structure and, Foundation for Research and Technology Hellas, P.O. Box 1527, GR Heraklion, Greece b Centre for Quantum Technologies, National University of Singapore, 3 Science Drive 2, Singapore , Singapore article info Article history: Received 7 August 2012 Received in revised form 20 September 2012 Accepted 3 October 2012 Available online 26 October 2012 Keywords: Diode laser Cold atoms Bose Einstein condensation Microwave sideband generation system abstract The increasing complexity of cold atom experiments puts ever higher demands on the stability and reliability of its components. We present a laser system for atom cooling experiments, which is extremely reliable yet simple to construct and low-cost, thus forming an ideal basis for ultracold atom experiments such as Bose Einstein condensation and degenerate Fermi gases. The extended cavity (master) diode and slave lasers remain locked over a period of months with a drift in absolute frequency well below 1 MHz with a line-width of less than 300 khz. We generate the repumper light by modulating the current of an injection locked slave laser at a frequency of 6.6 GHz. The construction of the laser is simple and largely based on off-the-shelf electronic and optomechanical components. Crown Copyright & 2012 Published by Elsevier B.V. All rights reserved. 1. Introduction s are the principal tool in most of the modern atomic physics experiments. cooling of dilute atomic vapours [1] precipitated the achievement of Bose Einstein condensation (BEC) in the mid- 1990s [2 4]. Experimental setups in this area of research have reached a complexity level, which poses new technological challenges for the stability and reliability of frequency stabilised diode lasers. With hundreds of components in an experiment it is crucial that they function reliably for long periods. Here, we report on the solution developed for our BEC experiments: a diode laser system that remains operational for periods of months at a time without any user intervention. This level of reliability enabled the use of a single master laser as the reference for three separate ultracold atom experiments. We describe the key points of the construction of this extremely reliable laser setup, which is inexpensive and sufficiently simple to construct even in a very basic workshop. In the supplementary material we provide electronic diagrams and mechanical drawings of the main components. 2. Overview over the laser system An overview of the laser system of this paper is shown in Fig. 1. It is based on functional modules connected to each other by single-mode, polarisation-maintaining optical fibres. The primary module contains the master laser with its frequency stabilisation spectroscopy. It is a home-made extended-cavity diode laser (ECDL) providing up to 60 mw of light, which is distributed to three separate cold atom experiments. The light of the master laser is delivered to the AOM board where we control the frequency and power of the light coupled into the optical pumping (5 mw), imaging (1 5 mw), and cooling (2 mw) fibres. The light of the cooling fibre is amplified in a double-pass tapered diode laser [5] with about 500 mw delivered to the distribution board (Fig. 2) where the light is split into the six fibres for the magneto-optic trap (3D-MOT) and the two fibres for the cold-atom beam source (2D-MOT). Each of the modules is based on a granite board ð mm 3 Þ onto which we glue the optical components via short aluminium +25 mm posts. 1 In order to ensure good long-term stability, we set the beam height to only 5 cm above from the granite surface and limit the degrees of freedom to the absolute minimum required. The distribution board (Fig. 2), for example, has exactly four adjustment screws to couple the light into each of the fibres. For ease of alignment, we mount FC-APC fibre collimators (SchäfterþKirchhoff 60FC-4-A8-07) directly into 1/2 in mirror mounts (New Focus 9882). In order to ensure that the double-passed AOMs achieve their maximum modulation bandwidth possible (one half of the carrier frequency), we have n Corresponding author. address: lasers@bec.gr (D. Sahagun). 1 We use cyanoacrylate (Loctite 408) to glue the posts onto the granite as well as some of the optics onto their mounts [6] /$ - see front matter Crown Copyright & 2012 Published by Elsevier B.V. All rights reserved.

2 D. Sahagun et al. / Optics Communications 290 (2013) Master Slave Microwave AOM Board Absorpt. Imaging Tapered Optical Pumping Distribution 2D/3D MOT Fig. 1. Layout of the laser system. Each box represents a separate module with the arrows depicting the optical fibres links between them. The darker boxes represent the function taken by the generated light on the experiment. Diode Holder and Collimator Lens Fastening Diode Holder Screw (and Collimator) 14 mm Peltier Aspheric Lens Diffraction Grating PZT Mirror Mount Fig. 3. The ECDL master laser. The cavity is formed by the laser diode, the lens (both mounted into the aluminium block on the left side of the image), and the diffraction grating, which is glued onto two piezo stack, which are in turn are glued onto a commercial mirror mount. The aluminium base-plate of the cavity is glued onto a large Peltier element covering most of its surface. The Peltier with its cavity assembly is then glued onto an aluminium block, which serves as heat sink and acoustic shield. Fig. 2. Distribution board, where light is delivered from the tapered amplifier and split amongst the six beams of the 3D-MOT and the two beams of the 2D-MOT. There are additional ports for imaging, optical pumping, and push beams. The dimensions of the board are cm. developed an optical alignment algorithm, which uses the optical fibres as ideal mode-filters at each stage of the alignment process The ECDL design The mechanical design of our master laser (Fig. 3) combines the high level of thermal stability of the Hänsch-type [7] ECDL with the simple mechanical construction of the Sussex design [9]. The mirror mount holding the grating, the diode, and collimator are all mounted directly onto a temperature-stabilised base-plate. The base-plate itself is glued onto a Peltier element, which in turn is glued onto a solid aluminium base. 3 The laser diode and the collimation lens are mounted into a small aluminium block. The diode (Sharp GH0781JA2C) is held in place simply by three small screws and the collimations lens (Thorlabs C230TM-B) sits in a threaded hole with clamping mechanism. 4 The grating (Thorlabs GH13-18U-UV) is glued onto two piezo stacks (Thorlabs AE0203D08F), which in turn are glued directly onto a mirror mount (New Focus 9882) such that the laser beam hits the grating close to the outer piezo stack. Using two stacks not only simplifies the construction, but also gives us independent control over the length of the cavity length and the angle of the grating. This allows one to obtain a tuning range of more than 6 GHz, sufficient to tune over the D2 lines of both isotopes of rubidium. In practice after the initial alignment we only use a single piezo, since the tuning range of 4 6 GHz obtained this way is sufficient to lock the laser without requiring any subsequent mechanical realignment over the lifetime of the diode. We align the laser following 2 The layout of the AOM and distribution boards including alignment instructions can be found in the supplementary material. See Appendix A. 3 Due to the small thickness of the layer of glue, the thermal conductance is very high despite cyanoacrylate having a mediocre thermal conductivity. The elements can readily be disassembled by heating the components to the glass transition of the glue at 120 1C. The absence of any screws improves the mechanical stability and simplifies the construction of the laser. 4 The mechanical drawings can be found in the supplementary material in Appendix A. standard procedures for Littrow configuration lasers [7,8]. The precision of the screws is sufficient to optimise the extended cavity lasing threshold and to tune easily to the atomic absorption line of interest. One of the main sources of frequency fluctuations in diode lasers are acoustic vibrations either through air or the optical table. We shield the laser from air-borne sound and thermal fluctuations by placing it inside an aluminium box of 10 mm wall thickness with the inside coated by an additional 10 mm layer of standard pluming foam. The laser beam exits through a microscope coverslip placed at Brewster angle. The diode laser, the spectroscopy, and the fibre coupling optics are glued onto the granite plate resting on a second stone plate (100 mm thick), which in turn sits on rubber spacers on a home-built stone table. All the cables for spectroscopy are clamped onto the optical table and the bread board. The thick aluminium box together with the insulating foam of the laser box provides good acoustic and thermal isolation. The temperature of base-plate is monitored by a thermistor placed close to the laser diode, which according to the temperature controller (Thorlabs TED200C) is stable to within 72 mk. The residual temperature fluctuations then result in a frequency drift of the ECDL of less than 100 khz/s, which greatly facilitates locking the laser to an atomic reference. The three individual experimental setups amplify the light supplied by the master laser thus requiring only a few milliwatts of power. We choose to maximise the lifetime of the master laser diode rather than its output power by operating it at a current of ma, which is well bellow the nominal maximum injection current (120 ma). This yields an output power after the optical isolator of around 70 mw, from which we use 1 mw for the laser-stabilization spectroscopy and distribute the remainder to the three experiments. We found that our commercial current sources degraded the short-term frequency stability of our laser. Using an improved and simplified version of the electronics described in Ref. [10] (see Fig. 4 and supplementary material) we were able to improve the short-term stability of the free-running laser by more than one order of magnitude at much reduced cost Active laser frequency stabilisation We lock the master laser to the saturation dip of the 95S 1=2,F ¼ 2S-95P 3=2,F ¼ 1,3S crossover at the D2 line of 87 Rb. Since the light provided by the master laser must not contain sidebands, we modulate the atomic transition frequencies using

3 112 D. Sahagun et al. / Optics Communications 290 (2013) ~ 2 7 mw each λ/2 λ/2 λ/2 4 λ/2 λ/4 Rb Attenuator Mirror Fig. 4. A simple ultra-low noise current supply. The design is a simplified, improved version of [10]. In order to reduce mains noise, the current supply is placed in a magnetically shielded box supplied by an external 18 V DC power supply, which is regulated to 715 V at the entrance to the box (not shown in figure). The O current-sensing precision resistors are mounted on a small heat sink. We use a double shielded BNC cable to carry the DC current to the bias- Tee and then on to the laser diode, which is grounded onto the electrically floating base-plate of the laser. the Zeeman shifts rather than modulating the diode laser frequency itself [11,12]. The optical arrangement can be seen in Fig. 5. For the spectroscopy, we use less than 1 mw of circularly polarised light in a 7 mm diameter beam. After passing through the absorption cell once, the light goes through a db optical attenuator, is retro-reflected, traverses the attenuator and the cell again, and is reflected by a polarising beamsplitter onto an amplified photodiode. The optical attenuator ensures that the probe beam is much weaker than the pump beam, which results in a suppression of the non-saturated signal compared to the saturation dip and thus in a five-fold reduction of the offset in error signal. This is particularly useful, since this offset is pressure and thus temperature sensitive. The transitions in Rb atoms is modulated by a small AC magnetic field (B 0:2G, f¼1.2 MHz). The resulting absorption signal is demodulated by a mixer, low pass filtered, and conditioned by a PI circuit, the output voltage of which is fed back to the PZT of the ECDL. We generate the oscillating magnetic field using a simple coil (80 mm long, 25 mm in diameter, 12 windings) wound directly onto the Rb-cell with an HV-capacitor (15 nf Type 940C, Cornell Dubilier) connected in parallel. The circuit is driven near its resonance frequency of about 1.2 MHz by a standard function generator. We demodulate the signal from the photodiode using a standard RF-mixer (MiniCircuits ZAD-6) and filter the resulting error signal with a single pole 1 khz low pass filter, resulting in a time constant of 150 ms. We use a PI-circuit to control the voltage on one pole of the piezo stack. The other pole is connected to a low-impedance circuit providing manual control over voltage for scan and offset. The voltages on the two poles are displayed separately. Using the manual offset voltage on one pole, we can tune the voltage of the locked PI-feedback on the other pole to zero. This way, when we break the lock, the laser will be scanning around the previous locking point, which provides very useful information for trouble shooting. In order to minimise noise, we use exclusively OP-27A operational amplifiers. 5 We found it very useful to place a large silicon photodiode (100 mm 2 ) between the coil and the absorption cell, which allows us to monitor the fluorescence of the atoms including the saturation dips even if the absorption signal is partially obscured by mode-jumps of the laser diode. This also allowed us to easily observe the desired saturation intensity ð0:5 I sat Þ. An example of the resulting fluorescence and the error signal used for locking can be seen in Fig The diagram of the PI-circuit and a block diagram of the detection electronics can be found in the supplementary material. See Appendix A. PI Mixer Amplified PD Fig. 5. Layout of the laser frequency stabilisation spectroscopy board. We couple most of the output power of our ECDL to the optical fibres leading to our four experimental stations (2 7 mw each). About 1 mw are used for the spectroscopy. We polarise the beam circularly and expand it to 5 mm diameter in order to increase the interaction volume and hence, absorption of light by the atoms. After passing through the absorption cell, it passes through a db attenuator traverses the cell and is reflected by a polarising beamsplitter onto an amplified photodiode. The transitions in Rb atoms is modulated by a small ac magnetic field (B 0:2 G, f¼1.2 MHz). The resulting absorption signal is demodulated by a mixer passed through a low pass filter, read by a PI circuit, and fed back to the laser via the voltage applied to the PZT of the ECDL. We found it extremely useful to use a 100 mm 2 photodiode placed between the coil and the spectroscopy cell to observe the generated fluorescence directly on an oscilloscope. Lock Signal Fluorescence Frequency Fig. 6. Fluorescence (top) and differential absorption (bottom) signal. The arrow indicates the ð5s 1=2,F ¼ 2Þ2ð5P 3=2,F ¼ 1c:o: 3Þ crossover transition of the D2 line of rubidium 87, to which we lock our master laser. The fluorescence signal (top) was taken by a simple 100 mm 2 photodiode placed between the coil and the spectroscopy cell and recorded via a variable resistor directly on a digital oscilloscope. The differential absorption signal (bottom) was low pass filtered with a time constant of 150 ms. Note the relatively small contribution of the nonsaturated signal, thus giving only a small pressure dependent offset. 3. Slave laser with microwave sidebands -cooling of rubidium and most other alkali atoms requires light at a repumping transition to keep the atoms in the cooling cycle. For a rubidium MOT this typically amounts to tens of milliwatts of light at the cooling transition and about a milliwatt at the repumping transition. Normally, this requires a full second ECDL laser together with its frequency stabilisation. Here, we generate the repumper light by modulating the drive current of the slave laser at a frequency of GHz the difference between the cooling and repumping transitions [13]. Figs. 7 and 8 show example spectra of this microwavemodulated slave laser for injection currents of ma and for microwave powers of mw. The spectra were taken using a scanning Fabry Pérot cavity with the line shapes having purely technical origin. In many cases, for example in a MOT (Fig. 7c), one requires only very little light at the repumping transition. For optical pumping (Fig. 7a and b), the power of the AC

4 D. Sahagun et al. / Optics Communications 290 (2013) a) Optical Pumping b) Optical Pumping c) MOT / Imaging d) e) f) repumper Detuning from Injection Light [GHz] Detuning from Injection Light [GHz] 0 Fig. 8. Frequency comb generated by modulating the slave laser with (a) 0.2 W and (b) 0.86 W of the microwave power at 7 GHz and injection-locking it to the master laser. The laser current was 128 ma. The line shapes are due to the Fabry Perot cavity used. Current leads Fig. 7. Spectrum, taken with a Fabry Perot cavity, of the slave laser for different laser currents and microwave powers. The frequency on the horizontal axes is relative to the one of the injected light (the D2-line in 87 Rb at 780 nm) with the cooling and repumping transitions marked with dotted lines. The wavelength of the free-running slave laser is 783 nm. The laser current was ma and microwave power 5 50 mw at a frequency of GHz. (a) and (b) are useful for optical pumping, where the repumper light has to be the same or stronger than the trapping light. (c) is a typical spectrum used for a MOT, which requires only a weak repumper. (d) (f) have been added in order to demonstrate the versatility of the sideband modulation technique. The only difference in the experimental settings between the curves (a) (f) is the laser current and microwave power. repumping light needs to be equal or stronger than the one of the optical pumping/cooling light. Fig. 7d f shows how one can control the distribution of the sidebands relative to the trapping light by small changes of the laser current (DIo10 ma) and microwave power (DP 15 mw). Once adjusted, the amplitude of the sidebands is observed to remain stable for many months. The microwave source for modulation is a low-cost PLL synthesiser (AME-engineering LO-45B-680) with a resolution of 100 khz and absolute precision of 10 khz. We control the amplitude of the microwave signal with a commercial mixer (Mini- Circuits ZMX-8GLH) followed by either a 10 mw (Kuhne LNA BB0180A) or a 5 W (Stealth Microwave SM6451) amplifier. We then combine the DC-current from our current source with the microwave signal using a bias-tee (MiniCircuits Z85-12G or Picosecond Pulse Labs 5585). 6 In the design of the slave laser (Fig. 9), care was taken to optimise the coupling of the microwave to the slave diode (Sharp GH0781JA2C) by soldering the microwave cable directly to the diode. The Peltier element is glued onto the mount and its heat sink. The alignment of the slave laser is simplified by mounting the whole slave laser assembly directly into a 1 in kinetic mirror mount. The microwave modulated slave laser costs only about h2500 with all electronics and optics included. 7 Fig. 8 shows for the first time the generation of a phasecoherent frequency comb using microwave current modulation of a slave laser: With about 1 W of microwave power at 7 GHz we can create a phase-coherent frequency comb spanning more than 100 GHz. This corresponds to a modulation index m ¼ Df =f mod ¼ 14, with f mod being the carrier frequency and Df being the maximum frequency deviation from the carrier frequency. The highest modulation observed previously was 6 A block diagram of the microwave electronics and a technical drawing of the slave laser can be found in the supplementary material. See Appendix A. 7 This includes all of the low-power microwave electronics. The 5 W version of the microwave electronics costs an additional h1000. Collimator Lens Locking Ring 1.9 [14]. The microwave sideband frequency comb presented here is potentially very useful for example for metrology applications. 4. Stability and performance The free-running master laser has a frequency drift smaller than 100 khz/s or 30 MHz/min. We measured the stability of actively stabilised laser by beating two lasers locked to adjacent absorption features in the D2 line of rubidium. The resulting beat signal has a line-width (FWHM) of less than 500 khz, corresponds to a line-width of the individual lasers of 300 khz. The long-term drift is well below 1 MHz over a duration of months. Given the natural line-width of the transition of 6 MHz, this is more than sufficient for laser-cooling experiments of rubidium. The most important consideration for complex cold atom experiments is the reliability of the laser: The master and slave lasers including the tapered amplifier typically stay locked for many months at a time. For example between the end of November 2010 and March 2011 the master and slave lasers stayed locked and even survived a nearby earthquake of magnitude 5.2 [15]. 5. Conclusions diode Cooling plate Fig. 9. An exploded view of the slave laser assembly. The picture is the cross section of all the elements with exception of the aspheric lens and diode, that are fully shown. The diode is inserted from the back of the mount and locked in place using three screws. After which the microwave cable is inserted and soldered directly to the laser diode. Cooling and temperature stabilisation is provided by a Peltier element and heat sink (both not shown) glued to the back plate. We provide a complete laser system for cold atom experiments including: the master laser, its spectroscopy, and amplification. It is inexpensive, has good long-term stability, and is extremely reliable ideal for complex experimental setups such as cold atom and especially Bose Einstein condensation experiments.

5 114 D. Sahagun et al. / Optics Communications 290 (2013) Acknowledgements This work has been supported by a Marie Curie Excellence Grant of the European Communities Sixth Framework Programme under Contract MEXT-CT and by FONCICyT project number D.S. acknowledges the National Research Foundation and the Ministry of Education, Singapore. V.P. acknowledges that part of this research has been co-financed by the European Social Fund ESF and Greek National Funds through NSRF (Heracleitus II). We would like to thank V. Ladopoulos, G. Gousis, and J. Koutsaidis for invaluable electronics support. Appendix A. Supplementary material Supplementary material (electronics diagrams, optical and mechanical drawings) associated with this article can be found in the online version at [2] M.H. Anderson, J.R. Ensher, M.R. Matthews, C.E. Wieman, E.A. Cornell, Science 269 (5221) (1995) 198. [3] C.C. Bradley, C.A. Sackett, J.J. Tollet, Hulet, Physical Review Letters 75 (9) (1995) [4] K. Davis, M.O. Mewes, N.J. van Druten, D.S. Durfee, D.M. Kurn, W. Ketterle, Physical Review Letters 75 (22) (1995) [5] V. Bolpasi, W. von Klitzing, Review of Scientific Instruments 81 (2010) [6] The companies and products are mentioned for information only. They do not imply any endorsement either by the authors or their institutions. [7] L. Ricci, M. Weidemiiller, T. Esslinger, A. Hemmerich, C. Zimmermann, V. Vuletic, W. Kijnig, T. Hänsch, Optics Communications 95 (1995) [8] C.E. Wieman, L. Hollberg, Review of Scientific Instruments 62 (1) (1991) 1. [9] A.S. Arnold, J.S. Wilson, M.G. Boshier, Review of Scientific Instruments 69 (3) (1998) [10] K.G. Libbrecht, J. Hall, Review of Scientific Instruments 64 (8) (1993) [11] R.A. Valenzuela, L.J. Cimini, R.W. Wilson, K.C. Reichmann, A. Grot, Electronics Letters 24 (12) (1988) 725. [12] T. Ikegami, S. Ohshima, M. Ohtsu, Japanese Journal of Applied Physics 28 (Part 2, No. 10) (1989) L1839. [13] R. Kowalski, S. Root, S.D. Gensemer, P.L. Gould, Review of Scientific Instruments 72 (6) (2001) [14] P.N. Melentiev, M.V. Subbotin, V.I. Balykin, Physics 11 (8) (2001) 891. [15] The earthquake happened on 28/02/2011 before 10:00 a.m., our master laser remained locked. Details about the earthquake can be found in / References [1] D.J. Wineland, R.E. Drullinger, F.L. Walls, Physical Review Letters 40 (1978) 1639.