A three step laser stabilization scheme for excitation to Rydberg levels in 85Rb L A M Johnson,1 H O Majeed,1 and B T H Varcoe1 School of Physics and Astronomy, University of Leeds, Leeds, LS2 9JT, UK (Dated: 13 January 2011) We demonstrate a three step laser stabilization scheme for excitation to nP and nF Rydberg states in 85Rb, with all three lasers stabilized using active feedback to independent Rb vapor cells. The setup allows sta- bilization to the Rydberg states 36P -70P and 33F -90F , with the only limiting factor being the 3/2 3/2 7/2 7/2 available third step laser power. We study the scheme by monitoring the three laser frequencies simultane- ouslyagainstaself-referencedopticalfrequencycomb. Thethirdsteplaser,lockedtotheRydbergtransition, displays an Allan deviation of 30kHz over 1 second and <80kHz over 1 hour. The scheme is very robust and affordable, and it would be ideal for carrying out a range of quantum information experiments. 1 1 PACS numbers: 42.62.Fi, 32.80.Ee 0 2 I. INTRODUCTION over the two discussed above in that it uses only conven- n tional IR diode lasers, with no need for second harmonic a J Long term laser stabilization to atomic levels is use- generation. Thismakesthelasersmoreuser-friendlyand 2 fulformakingprecisionmeasurementsinspectroscopy1,2 affordable. Also, the addition of a third laser step allows 1 and, for the case of Rydberg levels, is critical for mod- excitation to the nF series of Rydberg states. In this ern Rydberg atom cavity QED proposals3,4. In both work our frequency comb gives us a unique opportunity ] to assess the behavior of each of the stabilized lasers in h cases long term frequency stability is of importance to the three step scheme simultaneously. p maintainahighfrequencyaccuracy. Ithasrecentlybeen - demonstrated that the detection of Rydberg states in a t n thermal vapor cell can offer greatly reduced sensitivity II. APPARATUS a to stray electric fields5–7. Not only this but purely opti- u cal detection makes these experiments much simpler to q operate than conventional field ionization type experi- [ mentswhichuseabeamortrappingapparatus. Another 1 clear advantage is the simplicity of acquiring analogue v spectroscopic Rydberg signals for laser stabilization ap- 6 plications, by using an ordinary photo diode. 2 4 Recent work has demonstrated laser stabilization to 2 thesetypesofRydbergsignal,detectedinRbvaporcells, . using different types of excitation scheme8,9. In Ref. 8 a 1 Vschemeisexploited,wherea297nmfrequencydoubled 0 1 dyelaserisusedtoexcitetheUVtransitiondirectlyfrom 1 the ground state to Rydberg level, whilst a 780nm laser : is used for detection on a strong D1 cycling transition, v i using an electron shelving scheme. This single step laser X excitation allows excitation to nP Rydberg states. In r Ref. 9 a two step ladder scheme is used, where a 780nm a laser excites a D1 transition and acts as a weak probe in an EIT system. A 480nm frequency doubled diode laser then acts as a strong pump beam on the 5P3/2 to FIG. 1. The experimental setup used for stabilizing to Ry- Rydberg level transition. This two step laser excitation dberg levels. The 780nm and 776nm lasers are locked to allows EIT signals to be detected for nS and nD Ryd- separate Rb reference cells with active feedback to the laser berg states. In this work we excite Rydberg states in piezo and injection current. The 1260nm laser is frequency 85Rb using a three step laser excitation scheme, the de- modulated via the injection current and lock-in detection of tection method is discussed in previous work6,7. The thefirststepabsorptioniscarriedouttoderiveanerrorsignal forfrequencystabilization. The1260nmlaserislockedusing three step level system consists of a 780nm transition active feedback to the laser piezo. Single mode optical fibers 5S F = 3 to 5P F = 4, a 776nm transition 5P 1/2 3/2 3/2 transport laser light from all three lasers to three frequency F = 4 to 5D F = 5 and finally a 1260nm transi- 5/2 comb beat detection units. tion 5D to nL . Rydberg excitations are detected via 5/2 J the reduced absorption of the first step laser on a photo diode. This three photon scheme may be advantageous Figure 1 shows a schematic of the experimental ar- 2 rangement. The optical setup for Rydberg detection is the frequency comb are all referenced to a Rb frequency the same as that described in our previous work7. One standard, disciplinedbyaGPSreceiver. Whenrequired, modification has been made in order to use the 1260nm the comb system allows laser frequencies to be measured laser light in our micromaser experiment after travers- with an absolute accuracy of ∼10−12. ing the cell: The introduction and removal of the third steplaserlightfromthesetupisachievedusingdielectric coated mirrors with polished rear faces. These mirrors III. RESULTS reflect only in the 780nm region and allow the 3mW of 1260nmlaserpowertoberemovedfromthestabilization As expected, we find that lower n states give Rydberg schemeandsenttoourcryostatviaanopticalfiber. The locking signals with a much larger signal to noise ratio, weak coupling of the third step transition ensures there due to stronger coupling with the third step laser. Also, is negligible attenuation of this laser on passing through the nF signals have a larger amplitude in comparison 7/2 the vapor cell. to the nP states for equal laser powers, generally by 3/2 The first step laser is stabilized using a separate Rb a factor of two. With this setup it was possible to sta- reference cell using a polarization spectroscopy scheme, bilize to the Rydberg states 36P -70P and 33F - 3/2 3/2 7/2 which allows stabilization to the center of a spectral line 90F . For locking to higher n states the only limiting 7/2 without the need for frequency modulation10. Active factor was our available third step laser power of 3mW. feedbackforthislaserlockissuppliedviathelasercavity We found that a 15MHz modulation amplitude gave an piezo and the diode injection current. error signal with excellent signal to noise ratio for lock- The second step error signal is derived from the re- ing. Thismodulationwouldideallybeappliedexternally duced first step absorption, in the same manner as with an EOM for example, to leave the 1260nm laser the Rydberg detection. The first and second steps co- linewidthintactforexperiments. Asimpler,andperhaps propagate through another separate Rb reference cell. more effective option, would be to modulate the atoms Whenthefirststepisstabilizeditselectsonlyzeroveloc- usingasolenoidaroundthecell. Inthiscaseadither-free ityatomsfromthiscell. Whenthesecondstepisscanned Zeeman lock could be used11. over the 5P F = 4 to 5D F = 5 transition we see 3/2 5/2 a Doppler free peak in the first step absorption, corre- sponding to reduced absorption as atoms are removed 0.3 from the strong 780nm cycling transition. By adding a 0.2 smallfrequencymodulationtothesecondsteplaser,and monitoring the first step absorption via a lock-in ampli- V) 0.1 fier, an error signal is extracted for stabilization to the al ( n top of this peak. Feedback for this lock is also supplied sig 0 viaTtohdeelraisveeracnaveitryroprieszigonaanldfotrhsetdabioidliezaitnijoencttioonthcuerRreyndt-. Error −0.1 −0.2 berg levels we add a frequency modulation to the third step laser via the injection current, with a modulation −0.3 amplitude of 15MHz and frequency of 90kHz. Detec- −400 −200 0 200 400 tion of the first step absorption is carried out using a Third step laser detuning (MHz) lock-in amplifier, with a time constant of 100µs. Ac- tive feedback for this lock is supplied via the laser cavity piezo. For all three stabilization schemes, the error sig- FIG.2. Theerrorsignalusedtostabilizethethirdsteplaser nals are sent through Proportional Integral Differential to the center of the the 50F transition. 7/2 (PID) controllers and then to the laser heads for feed- back. Aproportionofthelightfromeachofthelasersissent Thelockingsignalsthatwereusedforlockingthethird to individual comb beat detection units where the laser steplasertypicallyhadagradientof10mV/MHz. Figure light is mixed with light from a self referenced optical 2displaysthelockingsignalusedforlockingtothe50F 7/2 frequency comb. The frequency comb spectrum is cen- Rydberg level. We also chose to study the lock to the teredat1500nm. Secondharmonicgenerationandspec- 63P level as it forms the lower level of the microwave 3/2 tralamplificationisusedtosupplycomblightat780nm, transition which is used in our micromaser experiment. 776nmand1260nm. Ineachbeatdetectionunitthebeat Weanalyzedthestabilityofthesystemusingthecomb note between the laser and a predetermined comb line is counter readings. Figure 3 shows the computed Allan detected with a pin photo diode at around 20MHz. The deviations for all three laser steps, from a 5000 second beat notes are subsequently amplified and filtered before data set, when the third step laser was locked on to the being counted with frequency counters. All three fre- 50F Rydberg level. It can be seen that the Allan de- 7/2 quencycountersaresynchronizedtotakereadingssimul- viation of the third step laser is below 80kHz for time taneously. The counters and stabilization electronics for scales upto 103 seconds, and is below 30kHz over one 3 second. We found the Allan deviation of the third step of0.6×smallerat1260nm. Thereforefluctuationsofthe laser when locked to the 63P transition was compa- first step laser are suppressed by a factor of 0.6× on the 3/2 rable and stayed below 150kHz for time scales upto 103 third. seconds. These results demonstrate that the Rydberg Interestingly, if these measurements were translated atoms are very well isolated from the environment, espe- to absolute optical frequencies they would far surpass cially as this vapor cell is unshielded from magnetic and the accuracy of those from the best optical Rydberg electric fields. spectroscopy7,12, and would give an accuracy compara- ble to relative millimeter wave measurements13. In this study we were unable to be confident of the absolute ac- curacyofthesecondandthirdstepcounterreadingsdue 150 Laser 1 to observed counting errors of upto 1MHz for these fre- Laser 2 quency modulated beat notes, see for example Ref. 14. z) Laser 3 H Futureimprovementsinthesetupshouldbeabletoover- k n (100 come this. o ati vi e D n 50 IV. CONCLUSION a All We have used a frequency comb to demonstrate the stability of a three step laser excitation scheme for exci- 100 0 101 102 103 tation to nP and nF Rydberg states. The results show τ (s) that the third step laser in the ladder scheme can have a muchhigherfrequencystabilitythanthefirststepdueto FIG. 3. The Allan deviation of the three laser locks when themismatchinwavelengths. Theresultsalsoshowthat lockedtotheirrespectivecells. Thisdatawascomputedfrom the second laser lock does not compromise the stability a 5000s data set. The third step laser was locked to the of the third in this scheme. This locking scheme will 50F transition. be of great use in experiments that require nP and nF 7/2 Rydberg states. The setup is very easy to construct and maintainduetothefactthatalllasersareordinarydiode As one would expect, we find that frequency fluctu- lasers, with no need for frequency doubling. Our results ations are correlated between each of the laser steps: show that excellent stabilities are achievable with these Thefirststeplaserselectsanarrow6MHzvelocityrange cell based Rydberg state locks, in this scheme we were of atoms from the thermal distribution, with which the able to lock on to the line center of a Rydberg nF state, other two lasers interact. We also used the frequency with deviations of <80kHz over an hour time period. combtostudythebehaviorofthesystemtolaserdetun- 1J.Ye,S.Swartz,P.Jungner, andJ.L.Hall,Opt.Lett.21,1280 ings. For the first two steps, which have similar wave- (1996). lengths, weseeanalmostone-to-onecorrelationbetween 2G. P. Barwood, P. Gill, and W. R. C. Rowley, Appl. Phys. B the first step detuning and the shift of the second step 53,142(1991). signal, this occurs due to the close Doppler matching of 3M.L.Jones,G.J.Wilkes, andB.T.H.Varcoe,J.Phys.B:At. these two steps, with a wavelength ratio of 1.01. This Mol.Opt.Phys.42,145501(2009). 4P.J.BlytheandB.T.H.Varcoe,New.J.Phys.8,231(2006). relationship is not clear from figure 3, where the Allan 5A.K.Mohapatra,T.R.Jackson, andC.S.Adams,Phys.Rev. deviation for the second step lock is considerably higher Lett.98,113003(2007). than the first step, across all time scales. This can be 6P. Thoumany, T. Germann, T. H¨ansch, G. Stania, L. Urbonas, associated with different feedback responses for the two andT.Becker,J.Mod.Opt.56,2055(2009). locks. Surprisingly we have also observed that the sec- 7L. A. M. Johnson, H. O. Majeed, B. Sanguinetti, Th. Becker, andB.T.H.Varcoe,New.J.Phys.12,063028(2010). ond step detuning from the observed second step signal 8P.Thoumany,T.H¨ansch,G.Stania,L.Urbonas, andT.Becker, causes only very small shifts of the observed Rydberg Opt.Lett.34,1621(2009). signal frequency by ∼0.1× the detuning. Therefore the 9R.P.Abel,A.K.Mohapatra,M.G.Bason,J.D.Pritchard,K. second laser plays only a small part in the third laser’s J.Weatherill,U.Raitzsch, andC.S.Adams,Appl.Phys.Lett. 94,071107(2009). stability. For this reason, we see in figure 3 that the sta- 10C. P. Pearman, C. S. Adams, S. G. Cox, P. F. Griffin, D. A. bilityofthethirdsteplaserisnotcompromisedbythatof Smith, andI.G.Hughes, J.Phys.B:At.Mol.Opt.Phys.35, the poorer second step lock. Finally, figure 3 also clearly 5141(2002). shows that the third step fluctuations are considerably 11A.Weis,andS.Derler,Appl.Opt.27,2662(1988). lower than for the first step. This isan advantage of this 12B.Sanguinetti,H.O.Majeed,M.L.Jones, andB.T.H.Varcoe, J.Phys.B:At.Mol.Opt.Phys.42,165004(2009). particular excitation scheme and is a consequence of the 13J. Han, Y. Jamil, D. V. L. Norum, P. J. Tanner, and T. F. wavelengthmismatchbetweentheselasers. Detuningthe Gallagher,Phys.Rev.A.74,054502(2006). firststeplasercausesashiftinthevelocityselectionfrom 14Won-KyuLee,Dae-SuYee, andHoS.Suh,Appl.Opt.46,930 zero, this velocity has a Doppler shift which is a factor (2007).