ebook img

Spin Wave Diffraction Control and Read-out with a Quantum Memory for Light PDF

3.6 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Spin Wave Diffraction Control and Read-out with a Quantum Memory for Light

Spin Wave Diffraction Control and Read-out with a Quantum Memory for Light Gabriel H´etet1 and David Gu´ery-Odelin2,3 1 Laboratoire Pierre Aigrain, Ecole Normale Sup´erieure-PSL Research University, CNRS, Universit´e Pierre et Marie Curie-Sorbonne Universit´es, Universit´e Paris Diderot-Sorbonne Paris Cit´e, 24 rue Lhomond, 75231 Paris Cedex 05, France 2 Universit´e de Toulouse ; UPS ; Laboratoire Collisions Agr´egats R´eactivit´e, IRSAMC ; F-31062 Toulouse, France and 3 CNRS ; UMR 5589 ; F-31062 Toulouse, France Aschemeforcontrolandread-outofdiffractedspinswavestopropagatinglightfieldsispresented. Diffraction is obtained via sinusoidally varying lights shifts and ideal one-to-one mapping to light is realized using a gradient echo quantum memory. We also show that dynamical control of the diffracted spin waves spatial orders can be implemented to realize a quantum pulse sequencer for temporal modes that have high time-bandwidth products. Full numerical solutions suggest that bothco-propagatingandcouterpropagatinglightshiftgeometriescanbeused,makingtheproposal 5 applicable to hot and cold atomic vapours as well as solid state systems with two-level atoms. 1 0 PACSnumbers: 2 n Macroscopic spin superpositions in atomic ensembles a J have been studied for decades and have found use in 2π/kR many areas of physics. They are essential for quantum a) + - 0 6 -2 2 optical information storage and retrieval [1] and for sen- θ -1 1 ] sitive magnetometry [2]. Spin superpositions containing h quantized excitations are prepared either by performing Input Lbiegahmt sshift - + Retrieved p probe spatial orders quantum non-demolition measurements on the scattered - t light [3–6], or by directly mapping quantum states of n a light [7–10] onto the atoms. Using appropriate schemes b) i) + - c) (x105) u furthermore enables the actual shape of a spin wave to 1600 i) ii) iii) -3-2 0 23 q be controlled to a high degree. Andr´e et al. [11] pro- - + 1200 -5-4 -1 1 45 [ posed a scheme to perform Bragg scattering of admix- ii) 1 turesofspinandlightcomponentswithcounterpropagat- θ k 800 v ing lasers and electromagnetically induced transparency 400 4 (EIT).Twocounterpropagatingpolaritonmodesarecou- - + iii) 9 pled, enabling light fields to be localised within the 0 11 medium with far reaching prospects in quantum non- + z - 5 15 25 35 linear optics [12] and quantum simulations [13, 14]. A Time (t) 0 signatureofpolaritonicBraggdiffractionwasobservedin . 1 the fluorescence induced by the stationary light [15] but FIG.1: a)Gradientechomemory(GEM)schemewithalon- 0 observing the diffracted modes is currently challenging gitudinaldiffractiongratingappliedduringstorage. b)i)Cre- 5 due to the drastic requirements on the atomic ensemble ation of a narrow k-space distribution of spin waves by map- 1 temperature and the lack of direct read-out procedure. ping a pulse of light in the memory. ii) Diffraction of the : v Thesemethodsarealsonotapplicablefortwo-levelatom polaritonbyapplyingapairofintensedetunedlaserswithan Xi based quantum memories [16, 17]. angleθforatimeτ. iii)Thedetunedlasersareturnedoffand theread-outofthediffractionpatternisdonebyflippingthe r An alternative for realising Bragg scattering of spin gradient sign. c) 2D-map of the atomic coherence amplitude a waves is through the use of a controlled inhomogeneous ink-spaceasafunctionoftime,demonstratingKapitza-Dirac broadening, and a prominant technique is the gradient diffraction of the spin coherence wave. echo memory (GEM) [18, 19]. GEM was already shown to allow light storage at the quantum level with two [16] and three atomic levels [20], to demonstrate efficient specific durations of the grating imprint, the coupling of pulse re-ordering [21] and spectral manipulation of light the spin wave to the output field is efficiently canceled pulses that have large time-bandwidth products [22]. and that diffraction can actually be temporally reversed Here,wedemonstratethatGEMcanbeusedinconjunc- by displacing the grating nodes. These two effects to- tion with a grating to enable efficient coherent mapping gether realize a two-level atom coherent pulse sequencer between diffracted modes and propagating light fields. forlightpulsesthathavelargetime-bandwidthproducts. For this, we use a theory that includes all spatial orders To perform longitudinal diffraction of a spin wave one inscatteringtheoryandshowthatone-to-onemappingis creates a spin wave with longitudinal momentum spread possibleeveninaroomtemperaturevapourcell, beyond ∆kbywritingaweaklaserfieldexcitationinanoptically the secular approximation [23]. Last, we show that for dense atomic ensemble. Using GEM, an effective two- 2 level atomic ensemble is first subjected to a controlled to the atoms for a time τ yields the following polariton linearly varying gradient η(t)z as depicted in Figure 1. evolution After an input pulse with temporal width ∆t enters the +∞ medium, polaritonic modes with increasing momentum (cid:88) Ψ(k,t +τ) = inJ (ντ)Ψ(k+nk ,t ), (3) k over time are created. When the gradient is turned 0 n R 0 off, polaritons remain trapped in the ensemble with a n=−∞ longitudinal momentum extension ∆k = η∆t. Here, we where L is the sample length and J is the n-th order suppose that the effective two-level atoms form a quasi n Bessel function of the first kind. We thus expect the ini- one-dimensional ensemble along a direction z. The cor- tial wave to be splitted into multiple spatial orders with respondingoperatorsareσj (z,t ),where|1(cid:105)and|2(cid:105)are 12 0 a weight given by the Bessel functions. iii) The grat- the two atomic states and j labels atoms within a slice ing is finally turned off at a time t = 20 and the gradi- with length δz. The locally averaged coherence opera- ent is turned on with a negative slope to complete the tors read σ (z,t ) = 1 (cid:80)Nz σj (z,t ), where N is 12 0 Nz j=1 12 0 z gradient echo protocol. The result of numerical simula- the number of atoms in a volume given by the cross sec- tionsisdisplayedontherightpanelofFigure1-c),show- tion area of the interacting light field over a slice δz. We ing the evolution of |Ψ(k,t)|2 as a function of rescaled now suppose that the atomic superpositions do not de- space and time. For the grating parameters, we choose phase over the course of the experiment. In particular, ν =ν/g =2, k =k l =150×105 which, with g =3 0 R R 0 this means that we can ignore the Langevin operators kHz, corresponds to a grating step of 16 mm. We also and treat the operators σ12 as c-numbers. use η = 7.106 and an optical depth 2πN2/η = 5.4. Nu- Torealizediffractionofthespinwave,apairofintense merical simulations show that in the mapping phase the pulses with an angle difference θ, with a Rabi frequency polariton evolves to higher k until the gradient is turned Ω and detuning ∆ from the two-level atomic transition off. When the grating is turned on Bragg diffraction is arethenappliedtotheatomicsamplepreparedinaspin clearly observed. The zeroth order mode is depleted and superposition. If the detuning is larger than the decay higher order modes are populated as a function of time, rate from the excited state, the atoms will thus inter- as expected from Eq. (3). The light-shift δ(z,t) thus act with a standing wave that produces a sinusoidally leads to an evolution of the collective spin wave that is varying light shift δ(z,t) = ν(t)cos(kRz) on the atomic analogous to the Kapitza-Dirac diffraction [25]. coherences, where ν(t) = Ω(t)2/∆. k is related to the R At t = 20 the grating is turned off and the gradient angle between the two lasers that generate the optical turned on. As expected, the light field is reemitted at grating and must be greater than ∆k [32]. In rescaled k = 0. One specific feature of GEM here is that, as is time and space units (normalized by l = c/g and 1/g 0 0 manifest in Fig. 1-c), the energy of the spin wave modes respectively where g is the vacuum Rabi frequency in 0 is released from the memory at times T = T +nk /η, 0 R the light mode and c the vacuum speed of light), the wherendenotesthemodeorderandT thetimeatwhich 0 Maxwell-Bloch equations read: the zeroth order mode comes out. The polariton spatial modes are here in the positive range before switching ∂ √ E = i Nσ so they are all released from the memory with a period ∂z 12 T = k /η = 2.2. In typical diffraction grating experi- ∂ (cid:104) (cid:105) √ R ments with light, electrons [25], degenerate atomic gases σ = i η(t)z+ν(t)cos(k z) σ +i NE, (1) 12 R 12 ∂t [26, 27] or molecules [28], the field onto which a periodic phase profile is imprinted is let free to expand in two- with η = ηl/g , ν = ν/g and N is the atomic density. 0 0 dimensions to realize the equivalent of a Fourier trans- TransformingtheseequationsintothespatialFourierdo- formation. This operation maps the local phase onto a main, one gets an evolution equation for the spin wave transverse profile that displays the discrete set of spatial inside the sample [24] modes. We demonstrated here that GEM exactly real- izes such a Fourier transform but in the time domain, (cid:16)∂ ∂ N(cid:17) −η(t) +i Ψ(k,t) allowing efficient read-out of the diffracted longitudinal ∂t ∂k k modes in this time-of-flight like sequence. (cid:2) (cid:3) =ν(t) Ψ(k+kR,t) + Ψ(k−kR,t) , (2) Wenowturntothedemonstrationofaquantumpulse sequencer. At this point, we shall distinguish two op- where Ψ = kE +Nσ . In this picture, absorption or erating regimes for diffraction read-out: the regime I 12 emissionoflightoccurswhenthepolaritonreachesk =0, where n(cid:48)k > k = η(t − t ) and regime II where R 0 0 in i.e. when the phase matching condition is met. n(cid:48)k < η(t −t ), where n(cid:48) = (cid:100)ντ(cid:101) is the number of R 0 in Letusnowconsiderthetime-of-flight-likesequencede- populated modes for a grating duration τ. In regime I, picted in Fig. 1-b). i) A pulse of light is sent to the sam- n(cid:48) spatial modes exactly occupy both the positive and ple at t = 1.5 with a positive gradient slope and with negative k-values. In regime II, n(cid:48) spatial modes only in ν = 0. ii) The gradient is turned off and the grating occupy the positive k-space, as per Figure 1-b). turned on at a time t = 15. Assuming that the optical Simulations are now realised with the sequence shown 0 fieldisfullytransferedtotheatoms,applyingthegrating in the inset Figure 2-a) and in the regime I, reached by 3 amplitudes of stationary atoms. It is therefore in princi- a) plereversibleprovidedtheystayinapurestatethrough- i) 1 y 1Input Output outthecouplingtothegrating. Spatialmoderefocussing ii) z val efficienc000...468 Normalized probe energy i) iτi) iii)WgWgirrviiaatt)tthhiinnoggut ctpoahcnaosble,eoaafscohsnhieeovweodnf tqfihugeiutaerersmi2ms-bpo)lfy:tuhfseoinlglgorwaatiinsnpggindi-sieffcsrhhaoicfttlieiokdne,bpytrhoπe- e at a time t = 11 to exactly cancel the phase acquired iii) etri0.2 τs 0 5 Tim10e 15 20 by each cohserence. This can readily be done experimen- R tallyusingacousto-opticmodulators. AsshownFig. 2-b) 0 iv) 0 1 2 3 4 5 6 7 8 9 10 (rightpanel)theatomiccoherenceevolutionindeedrefo- Grating duration ντ cusses so that the zeroth order spatial mode is recovered withunitefficiency,demonstratingthereversibilityofthe b) Δϕ=0 (x105) diffraction process. i) I i) t=ts ii) The effects presented Fig 2 can also be observed using 2000 othermemoryprotocols. Forinstancewithhomogeneous three-levelΛatomicmemoryschemessuchasEIT[29]or k 0 Ramancoupling [30], the gratingcan beturnedon when ii) I Δϕ=π the control fields are off, that is when the optical field is transferred to the atomic coherences. The GEM con- −2000 veyor belt in k-space is however essential for diffraction Δϕ imaging. Furthermore,thelasttwopropertiesofthepre- 8 9 10 11 12 1 3 sented spin wave diffraction have direct implications for z Time t precision manipulation of optical pulses and single pho- tonsusingGEM.ItwasalreadyshownthatGEMenables FIG. 2: a) Simulations of the gradient echo memory (GEM) efficientstorageofpulsesthathavelargetime-bandwidth scheme with a diffraction grating applied during storage. productswithouttherequirementforhighopticaldepths Light read-out efficiency as a function of grating duration [24]. Using a Raman transition and a time dependent τ. Inset : Input and output probes with and without the coupling to the ground state, pulse re-ordering was also grating. b) Refocussing of the diffraction pattern. At a time demonstrated[21]. Forinstance,inputpulseswereshown t=t ,thegratingpositionisdisplacedbyλ/2byshiftingone s to come out first-in-first-out (FIFO) by turning off the of the arms phase by π (left panel), producing rephasing of coupling to the ground state at the first k = 0 crossing thediffractionpatterntothezerothorderspatialmode(right andbyswitchingthegradienttwiceusingaΛ-GEM.Ex- panel). tending these effects to two-level atomic memories is im- portantasitwouldenlargetherangeofmaterialsystems where random access memories can be implemented. A simply increasing the grating momentum by a factor of combination of diffraction intensity and refocussing con- 3.3 and reducing the gradient by the same factor. Fig- trol in the regime I actually provides the ingredients to ure 2-a) shows the retrieval efficiency of the zeroth order do so very efficiently as we now show. mode as a function of grating duration in a time window t = 18±3. As the duration of the interaction with the The idea is shown Fig. 3-a). Three pulses are input gratingisincreased, thezerothordermodegetsdepleted to the GEM memory. At t = 10, one turns on a grat- leading to a drop in retrieval efficiency. This is because ing on for a time τs such that the zeroth order mode is the polariton keeps the energy in higher order spatial completely depleted, that is at the first node of the ze- modesthatremaintrappedintheatomicensemble. The rothorderBesselfunction. Thegradientisthenreversed. first positive spatial mode will only be phase matched Since the zeroth order mode is depleted, when the k =0 to the optical field a time k /η after the zeroth order condition is reached, there is no emission of light and R mode is emitted. The energy of the zeroth order mode all the pulse energy is mapped onto higher order spatial releasedfromthememoryfollowsthezerothorderBessel modes. One can however retrieve this energy back into function, meaning that the energy is coherently sloshing the zeroth order mode by turning off the gradient, ap- back and forth between stationary spatial modes inside plying the grating again during a time τs but with one theensemblebeforebeingoutput,asignatureofthewave arm of the grating shifted by π. The key point is that, natureofthespincoherences. Remarkably,alltheenergy duringtherefocussing,thefirstordermodesdonotcross remains trapped inside the sample at a time τ =2.4 cor- k =0. Turning the gradient on again therefore realizes a s responding to the first node of the zeroth order Bessel similarFIFOread-outasintheΛ-GEM:thepulsescome function. Aswewillshow,thisoffersanewusefuldegree out without time-reversal but now using only two-level of freedom for sequencing light pulses. atoms and this purely dispersive diffraction control. Another important ingredient for pulse reordering is One can even re-arrange the order of pulses. For in- refocussing of the spatial modes. The presented diffrac- stance,onecanmakeasequenceofpulses1-2-3comeout tionoriginatesfromasuperpositionoftheelectrondipole in an order 3-1-2 using a similar combination of diffrac- 4 the temperature should be low enough so that atomic a) b) motion does not wash out the higher order modes of First in-first out (FIFO) Pulse re-ordering the spin wave. This is especially important when one Probe intensity0000....24168 12 3 12 3 0000....12468 1 2 3 3 1 2 uiinnsgdeeswecdiollufibnnetdeortn-hpatrthopeaasogsruadmteirningogffihaeulgdnrsdartweindhg’esrseotfetphneaonfgor4ma0t0eitnnegrmss.paWancde- 0 Mangez bon, c’est bien 5 10 15 20 25 30 35t 00 5 10 20 25 30 35 t a storage time of 0.1 ms, atomic temperatures of less k than 1 µK are required for the atoms to feel the valleys FIFO of the standing wave and thus for the atomic coherences to be diffracted, favouring quasi-copropagating geome- Holding tries where magneto-optical traps or even warm vapours can be used. Conversely, provided the light grating does t t notinfluenceatomicmotion,thisdiffractioneffectcanbe used as a very sensitive quantum thermometer for ultra- cold atoms. The narrow two-level atomic transitions in (x105) rare earth doped crystals [16] are also ideal for the effect η 500 to be observed experimentally. The sinusoidal detuning -500 ν 0 π 0 π can be realized simply by using a light shift applied di- τ τ t τ τ t rectly close to the optical transition and the M˝ossbauer s s s s Diffracting Rephasing effect prevents the standing wave to induce atomic mo- tion. The light fields should only be tuned so that stim- FIG.3: Re-orderingofpulsesusingtwolevelatoms. Fromtop ulated and spontaneous emission are negligible. Finally, to bottom : probe intensity as a function of time, evolution being able to control the two-level atom coupling to a of the energy of the polariton in k-space and gradient slope given mode in rare earth doped materials can be useful andgratingamplitudesusedinthenumericalsimulations. a) for turning delay lines into memories without the noise First in first out scheme. b) Pulse reordering. induced by π-pulses [17, 31] thus widening the range of materials available for quantum memory applications. In conclusion, we present a mechanism akin to diffrac- tion and refocussing techniques. This is demonstrated tion of optical fields or electron motion, where coher- Fig. 3-b). First, the third pulse is left to come out after ent superpositions of spins are diffracted and efficiently a standard GEM sequence. Then, the FIFO sequence is read-out. Applying far-detuned laser pulses along an implementedtoreorderthepulses1and2. Thegradient array of atomic spin coherences is shown to generate combinedwithsinusoidalshiftsisthereforeusefulforen- an atomic diffraction pattern in momentum space that gineering quantum memories and specifically to control can be rephased, engineered to control the coupling to pulsesoflightinatomicensemblesthatlackaΛscheme. light and detected via a gradient echo mapping to opti- Let us now discuss how the above mentioned effects calfields. Besidestheproposedcoherentpulsesequencer can be demonstrated using current technologies. Cold for two-level atom memories, the very high sensitivity gases ofalkali atomscanbe usedtostore lightefficiently of diffraction to atomic temperature has implications for using GEM [20] but due to the short lifetime of their ex- tunable and non-invasive cold atom thermometry. cited states, one shall use the Λ-GEM [24]. The phase gratingcanthenbeinducedbyapairofcircularlypolar- We would like to acknowledge useful discussions with ized lasers tuned to the transition used to Raman cou- B.C.Buchler,P.K.LamandfinancialsupportfromANR ple in order to induce a differential light shift to the retour post-doctorants SMEQUI (ANR-13-PDOC-0024- ground states. With cold atoms as a medium however, 01). [1] Alexander I. Lvovsky, Barry C. Sanders, and Wolfgang [7] M. D. Eisaman, A. Andre, F. Massou, M. Fleischhauer, Tittel. Nat Photon, 3(12):706–714, 12 2009. A. S. Zibrov, and M. D. Lukin. Nature, 438(7069):837– [2] Dmitry Budker and Michael Romalis. Nat Phys, 841, 12 2005. 3(4):227–234, 04 2007. [8] J.Hald,J.L.Sørensen,C.Schori,andE.S.Polzik.Phys. [3] L.-M.Duanetal. Nature,414(6862):413–418,November Rev. Lett., 83:1319–1322, Aug 1999. 2001. [9] Ju¨rgen Appel, Eden Figueroa, Dmitry Korystov, [4] A. Kuzmich, L. Mandel, and N. P. Bigelow. Phys. Rev. M. Lobino, and A. I. Lvovsky. Phys. Rev. Lett., Lett., 85:1594–1597, Aug 2000. 100:093602, Mar 2008. [5] R. J. Sewell, M. Napolitano, N. Behbood, G. Colangelo, [10] Kazuhito Honda, Daisuke Akamatsu, Manabu Arikawa, andM.W.Mitchell. Nat Photon,7(7):517–520,072013. Yoshihiko Yokoi, Keiichirou Akiba, Satoshi Nagatsuka, [6] Florian Haas, Jurgen Volz, Roger Gehr, Jakob Reichel, TakahitoTanimura,AkiraFurusawa,andMikioKozuma. and J´eroˆme Est`eve. Science, 2014. Phys. Rev. Lett., 100:093601, Mar 2008. 5 [11] A. Andr´e and M. D. Lukin. Phys. Rev. Lett., 89:143602, A, 80:013818, Jul 2009. Sep 2002. [24] G. H´etet, J. J. Longdell, M. J. Sellars, P. K. Lam, and [12] A. Andr´e, M. Bajcsy, A. S. Zibrov, and M. D. Lukin. B. C. Buchler. Phys. Rev. Lett., 101(20):203601, Nov Phys. Rev. Lett., 94:063902, Feb 2005. 2008. [13] J.Otterbach,R.G.Unanyan,andM.Fleischhauer.Phys. [25] Daniel L. Freimund, Kayvan Aflatooni, and Herman Rev. Lett., 102:063602, Feb 2009. Batelaan. Nature, 413(6852):142–143, 09 2001. [14] J.Otterbach,J.Ruseckas,R.G.Unanyan,G.Juzeliu¯nas, [26] C.M.Fabre,P.Cheiney,G.L.Gattobigio,F.Vermersch, and M. Fleischhauer. Phys. Rev. Lett., 104:033903, Jan S.Faure,R.Mathevet,T.Lahaye,andD.Gu´ery-Odelin. 2010. Phys. Rev. Lett., 107:230401, Nov 2011. [15] M. Bajcsy, A. S. Zibrov, and M. D. Lukin. Nature, [27] Peter J. Martin, Bruce G. Oldaker, Andrew H. Miklich, 426(6967):638–641, December 2003. and David E. Pritchard. Phys. Rev. Lett., 60:515–518, [16] Morgan P. Hedges, Jevon J. Longdell, Yongmin Li, and Feb 1988. MatthewJ.Sellars. Efficientquantummemoryforlight. [28] Stefan Gerlich, Sandra Eibenberger, Mathias Tomandl, Nature, 465(7301):1052–1056, June 2010. Stefan Nimmrichter, Klaus Hornberger, Paul J. Fagan, [17] MBonarota,JDajczgewand,ALouchet-Chauvet,J-LLe Jens Tu¨xen, Marcel Mayor, and Markus Arndt. Nat Gou¨et, and T Chaneli`ere. Laser Physics, 24(9):094003, Commun, 2:263, 04 2011. 2014. [29] AlexeyV.Gorshkov,AxelAndr´e,MikhailD.Lukin,and [18] A. L. Alexander, J. J. Longdell, M. J. Sellars, and N. B. Anders S. Sørensen. Phys. Rev. A, 76:033805, Sep 2007. Manson. Phys. Rev. Lett., 96:043602, Feb 2006. [30] K. F. Reim, P. Michelberger, K. C. Lee, J. Nunn, [19] G. H´etet, J. J. Longdell, A. L. Alexander, P. K. Lam, N. K. Langford, and I. A. Walmsley. Phys. Rev. Lett., and M. J. Sellars. Phys. Rev. Lett., 100(2):023601, Jan 107:053603, Jul 2011. 2008. [31] Mikael Afzelius, Imam Usmani, Atia Amari, Bj¨orn Lau- [20] M. Hosseini, G. Campbell, B. M. Sparkes, P. K. Lam, ritzen, Andreas Walther, Christoph Simon, Nicolas San- and B. C. Buchler. Nat Phys, 7(10):794–798, October gouard, Jiˇr´ı Mina´ˇr, Hugues de Riedmatten, Nicolas 2011. Gisin, and Stefan Kro¨ll. Phys. Rev. Lett., 104:040503, [21] MahdiHosseini,BenM.Sparkes,GabrielHetet,JevonJ. Jan 2010. Longdell, Ping Koy Lam, and Ben C. Buchler. Nature, [32] For instance, if the laser fields are quasi co-propagating, 461(7261):241–245, Sep 2009. aslightangleθ≈1degreesbetweenthemyieldsafringe [22] B.M.Sparkes,M.Hosseini,C.Cairns,D.Higginbottom, spacing d = 2π/k = 2λ/θ2, which, with laser fields in R G. T. Campbell, P. K. Lam, and B. C. Buchler. Phys. the visible range, can yield grating periods ranging from Rev. X, 2:021011, Jun 2012. hundreds of micrometers to centimeters. [23] Gor Nikoghosyan and Michael Fleischhauer. Phys. Rev.

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.