You are on page 1of 7
nature physics ARTICLES De a ae (@ check tor upton Entanglement between distant macroscopic mechanical and spin systems Rodrigo A. Thomas©"5, Michat Parniak®*s, Christoffer Ostfeldt'S, Christoffer B. Moller®"2, Christian Bzerentsen’, Yeghishe Tsaturyan©", Albert Schliesser', Jiirgen Appel ©“, Emil Zeuthen©* and Eugene S. Polzik®"™ ity have thus far hampered the generation of macroscopic hybrid entanglement. Here, we generate an entangled state between the motion of a macroscopic mechanical oscillator and a collective atomic spin oscillator, as witnessed by an Einstein-Podolsky- Rosen variance below the separability limit, 0.83 + 0.02 <1. The mechanical escillator is a millimetre-size dielectric membrane and the spin oscillator is an ensemble of 10° atoms in a magnetic field. Light propagating through the two spatially separated systems generates entanglement because the collective spin plays the role of an effective negative-mass reference frame and provides—under ideal circumstances—a back-action-free subspace; in the experiment, quantum back-action is suppressed cessing’, quantum-enhanced sensing’ and fundamental tests ‘of quantum theory". Hybrid quantum systems often provide novel synergetic functionalities. In particular, entangled slates ‘of motional and spin degrees of freedom have played a prominent role in quantum computing and simulation with trapped ions and atoms’”. There, entanglement between motion and spin s generated by short-range interactions between individual atoms positioned at smicrometre-scale distances, with motional and spin degrees of free- dom associated with the same atoms ‘A very different regime, focused on long-range macroscopic entanglement between the motion of one object and a spin of another, has been proposed in ref. (see als ref. "), The key idea is tolet an atomic spin in a magnetic field act asa negative-mass oscl- Intor enabling coherent quantum back-action (QBA) cancella- tion", thereby permitting travelling ligat to generate entanglement between the two objects. The negative mass idea, which has been itly used in earlier experiments with two atomic ensem- os", has been further developed in refs. "and has become the basis for quantum-mechanict-Iree subspaces™ Negative-mass-enabled instability” and strong coupling" have been recently demonstrated using the coupling of t motional degree of freedom to a spin aystem. In refs.”»", sympathetic cooling of a mechanical oscillator optically coupled to atoms has been shown, ‘The negative-mass reference frame idea hae also been utilized in proposals" by using an auailiary mechanical system and multiple rive tones, In this way, entanglement har been generated between ‘#0 micromechanical oscillators embedded in a common micro- wave cavity”. An approach to mechanical-mechanical entanglement based on single-photon detection was demonstrated in refs." Fates is a key resource for quantum information pro- Here we report an experimental implementation of Einstein Podolsky-Rosen (EPR) entanglement in a hybrid system consist- ing of a mechanical oscillator and a spin oscillator’, as depicted schematically in Fig. 1s. An ouLof-plane vibrational mode of a soft-clamped, highly stressed dielectric membrane”, which is ‘embedded in a free-space optical cavity, constitutes the mechani cal subsystem. The spin subsystem is prepared in a warm ensemble ‘of optically pumped caesium atoms confined in a spin-preserving microcell”, The two oscillators are coupled to an itinerant light field and optically read out in a cascaded fashion, The basic ingredi ‘ents of our hybrid set-up and the principle of QBA evasion using a negative-mass oscillator have been described inf. “The collective macroscopic spin J, = SO"; F, of N= 10" atoms, cach with total angular momentum components (A, #0, #°), is optically pumped in the direction x of the magnetic bias field B I the limit where the magnitude of the mean longitdinal spin J. = [Gi] fa exceeds che traneverse collective spin components J and J the ater ean be mapped tothe harmonic olor vat ahles 0 Je) Ve and Ps Iy/ Vl, sting the canonical ‘commutation relation [Xs,Ps] = i (ref. "). The transverse compo- nents precess around the magnetic field a the Larmor fiequeycy dyeBaccordingtofly = hax), ~ hoy), + (hon /2)(0 +P) rere the first term i a constant offet. Since the optical pumping prepares the collective spin near the energetically highest Zeeman state, the collective spin teal dace a nogativesnass ovclator, thal in, oy <0 (eel °), with & counter rotating trajectory (Fig. Is), The ‘negative mast ter mninology arse by analggy to he standard harmonic osclator Hamstonian Hf moi" /2 ~ P*/(2m) in which the sign ofthe ‘Niels Bohr Institute, University of Copenhagen, Copenhagen, Denmark "Present addres: [CFO - Islut De Clences Fotonigues The Barcelona Irstitute of Science and Technology, Cstellelels, Spain. Present address: Prizes Schoo of Molecular Engineering, Unversity of Chicago, Chiao, I USA. ‘Present aderess: Darsh Fundamental Metrology, Harsholm, Denmari “These authors contributed equaly: Redr ga A. Thomas, Micha Parva, CChristoter Btfeldt, Christotfer8. Maier. e-mail pound ude ae NATURE PHYSICS | OL 1 | FEBRUARY 2021| 228-23] svsirstrecomynaticeyses . ‘ i |e cul p. a crane 2 * aay os Consioning r,s 3 Fig. | Tracking of the EPR oscilator. a A simi schematic ofthe entangled system, consisting of an atomic spin ensemble and a mechanical oscilatorin cavity, separated from the atoms by 21m distance, ané probed by light na cascaded manner Phase spaces and the evolution forthe spin Gi, Pp) ané mechanical (iy, Py) quantum degrees of ‘reedom are shown above the respective systems. The measurement photocurrent is convolved with Wlener iter K(, (approximate envelope shown inthe inst), ta yield 2 conltional trajectory, Quantum phase-space trajectory ofan EPR-entangled oscillator pair along wth deterministic variance ofthe estimate V, = 191.4 0.05 fer!=0 (ed) ane ‘the approximately final condt onal variance of V, = 0.83 + 0.02 att = 110, is (blue). Eveluton ofthe conditional variance V, forthe resorant (rete blue) and far-detunec (green), that fora joint and spectrally separated coscltors, respectively, The cle marks the variance at the end ofthe tesjectery mb, The shaded areas mark the Wo uncertainty of V. sass m determines that of both the potential and kinetic energies, as docs the sign of in Hs ‘Trajectory of an entangled EPR pair Fundamentally, the non-commuting quadratures of motion [55(0,B(0] = #for the individval systems (where j€ (SM) labels spin and mechanics) cannot be known simultaneously with abi- ttaty precision due tothe Heisenberg uncertainty principle; in par tieuat, Var(X + Var[P,]> 1. This limit is enforced by the QBA of the meter field (or example light) on the measured osellator ‘Such amit does not apply toa commuting combination of vari ables sich a [Kur Bon [Oie = £0)/VB (hus Ps)/V 0 that is, the sum of variances is no longer bounded from below. In fact, V = Var(Xeeg] + Var[Pana) <1 (ref, ") implies entanglement between systems § and M, which is analogous to violating the finglesystem limit with the EPR variables, Since the EPR vai- ables describe spatially separated systems, this effective osellator ienon-loca ‘We entangle the two oscillators by a back-action-evading colle- tive position measurement, For matched frequencies, i, = oy, = «> 0, the neative-maes spin oxilators rexponee tothe perturbing ‘pticalfield appenswithaphase oppositetothatofthe postive mast ocilator. The rerlling information written onto the optical meter phase is Pr 3 Xeen(t) = coscotpna(0) + sinotPapy(0), and thas only depends on the inital values of K, and b, inthe absence of damping and intrinsic oscillator noise. Thus, under ideal condi- tions, the oint measurement on a EPR-entangled system produces Pee SS) a noiseless trajectory of one oscillator in the reference frame of the other" In quantum theory, those trajectories arise as the expectation values of the dynamical variables with respect to the conditional quantum state A.(t), that is, incorporating the information con- tained in the measurement record obtained at times <1, Tracking the conditional state evolution is relatively straightforward in the present case of Gaussian states, dynamics and measurements (Supplementary Section C), where (#) is characterized solely by its first and second moments, which may be extracted by linear fil- tering of past measurement outcomes”. Optimal filer funetions are determined from the equations of motion, noise statistics and ‘the input-output relations forthe light fields. The optimal filter that takes into account data from a time period [0, ft estimate, for ‘example, Xena is called the Wiener filter, a version of the Kalman [iter widely used for state estimation’. In the simplest ease, the fil- ter envelope isan exponential with the rate defined by decoherence and readout processes, as pictorally shown inthe inset in Fig. 1s Such exponential filtering has been used in, for example, refs,” From the Wiener filter K, for Stor the conditional quadrature is obtained as [ow Kyl 1.90) @ Xipalt) where i() is the instantancoxs photocurrent gbtained by the homodyne detection of the optical quadrature Pi (0 of the trans: mitted light To obtain the exact Wiener fier we solve the Wiener= Hopf equitions (Supplementary Section C), which involve the crose-corelation C,{0) between the osclator signal Karn and i) 5 wellas C0), the autocorrelation of) ‘The virance of the conditions) sat, the residual uncer tainty in our knowledge about the aystem, is deterministic and even by Var Xeral(t) = VarfXera(t) Xing) = Vaz] Vari Xtpa(t) thats, the diference between the unconditional (steady state) variance Var)Seme) and the, (ensemble) variance ‘of our optimal estimate Var[Xépq(t)] = Jd¥ Kx(—#, Cult) (Gapplementary Section C), We calculate Var[Xeyy] and Cq() ‘onng Sted model parameter, In this mone, « complete tof fecond moments and the fll conditional covariance matix for (tion. Png) ate found. The raw experimental photocurrent i(¢) is ased to obtain the stochastic frst moments, fully defining the Gaussian state. Varfkqrq] contains contributions due to imperfect (QBA cancelation and thermal fhactuatons, ut ifthe correlation of Xeon with Xpy is strong enough, it leads to entanglement ss vwinctsd by VarXam|(t) Qualtatvely speaking, conditioning {uppresses the thera noise We use Wiener filtering to continuously track the EPR oscil Intor (Kirn, Pim) by inferring the conditional expectation val ‘ues Xipq = (Xi — XG)/V2 and Pigg = (Pty + PS)/V2 (optimal weights of M and § variables are determined by the full model presented in Supplementary Section D). Demodlating 1) vith corot and sino, respective, we obtain the conditional system quadratures (Xjyp, Poyg)- describing the rotating frame dynamics of the EPR-entangled system (Fig. 1b). As the condi- tioning progresses, we obtain & mote precise estimate of the con- ditional sytem state, as witnesed by the decteating conditional variance shown in Fig. Ie, In the long-condtioning time limit Taximum information is exttacted ffom past measurements and the shape of the Wiener fiter attains ie steady-state form, K(e 11)“ K(@ ~1), The corresponding steady-state condi tional vrianen thet we observe bru near essa cory 62) is Vo = Var Kenn] + Var Pop] = 08340021, ceriffing ent- Anglementof the spin and snechancs. This can be dtety compared wrth a cae where the frequencies ofthe sprtems are not matched tnd consequently the best value V; ~202 a 0.03 ls above the ns Nees) aaah eed Fig.2| Experimental set-up for hybrid entanglement generation, The local ascilator LO reads out the spin system precessng inthe magnetic ‘eld B wth the quantum sideband fields writen nto the orthogonal ight polarization, After spiting of LO, LO, phace shies by @ clave to LO, ‘s mixed withthe sidebards.Atter projection into a comman pelan2aion, thislightis sen to the mechanical system, which s probed in reflection Final homedye measurement ofthe cascaded hyord ystem is performed with LO, with phase 8.4/2, all-wave pat; 4/4, quarter-wave alate; BS, polarising beam zpiter. See main ert for etal. Inset: mace shape af the ‘mechanical ede under investigation (absolute displacement, linear scale), entanglement limit In the rest ofthe paper, we describe the experi- ‘ment and analysis leading tothe entanglement observation Experimental implementation The layout of the hybrid system is outlined in Fig. 2 (see Supplementary Section A for further details). First, the light inte acta with the collective spin ofa caesium atomic ensemble, contained ina 300 ym > 300 ym x 10 mim glass cell The spin anti-rlaxation coating of the cell”, along with magnetie shielding, provides spin coherence lifetime of T, = 0.7 ms. Light interacts with the spin ensemble in a double-pass con- figuration, thus increasing the lght-spinigetaction strength. The quantum operators of interest, Xy and Pr, are the in-phase and in quadrature vacuum fluctuations ofthe field poavized orthogo- nally to local oscillator 1 (LO,). Light-matter mapping is well described by the Hamiltonian Hig /h x yTs(XsXi, + [sPsP,), ‘which is doge tothe quantum non-demolition (QND) interaction 26, 003, The interaction leads to rotation ofthe input polriza- tion state, with a coupling rate [3/2x = 20 KEdz and the bandwidth Yol2n = 17 ki (Gll-ith at halmaximum) duc to decoher- ce processes, The deviation from the QND interaction leads to light partly exchanging states with the oscilator”™, addition. ally broadening the spin osellator” with rate dyJ2n = 2G, = 1.2 Kifz The spin also couples to sts own effective thermal bath vith the net stochastic force Fs originating from imperfect optical pumping, spin-exchange collisions and projection noite, resulting inthe mean bath occupation n, = 0.8, Afr interaction with the ‘pin systems, quantum light x coupled tothe mechanical oscillator (Gee Methods) “The mechanical osellator i relied in a highly stressed silicon nitide membrane that i¢ 13 nm thick and hae milimetre-cale ‘transverse dimensions, The membrane is periodically patterned, leading tothe emergence ofa phononic bandgap. The sftclamped mechanical mode is an out-of plane, localized centre-of mast vibrational mode (se inset in Fig. 2) witha frequency of ox/2R 211370 Mllz and a qualiy factor of Q = 650 x 10% that i, a natu ral linewidth of yu/2e = 2.1 mle, atexyogenie operating temper- dures, The membrane is placed neat the optical beam waist of a 2.6-mm-long cavity that basa linewidth of x/2x = 4.2 MH, and is strongly ovetcoupled in reflection by 93% ‘The optomechanical system is mounted in a 4 K flow cryostat and optically probed. The effective thermal bath at 10 K acts az a stochastic driving force Fy for the mechanical mode of interest, Light is detuned by /2x = ~0.7 MHz from the cavity resonance, cooling the mechanical mode to near its motional quantum ground state with a mean phonon occupancy of roughly 2. This éynami- ‘al back-action cooling’ broadens the mechanical response to ‘rl? = 3.9 kHz and redshifts its resonance frequency by 1 kHz to (yi2x = 1.369 MHz. The state of the mechanical system is extracted ‘optically ata readout rate of Fy/2 = 15 kHz Homodyne phaze- quadrature measurement ofthe light reflected off the aptomechanical cavity is performed with LO,, The optical transmission between the spin and mechanical systems is» = 0.53, and the final EPR detection efficiency is = 0.77, which includes ‘optical lostes in the path between the hybrid system and the detec tor and the detector quantum efficiency. ‘Model for the hybrid system "To construct the Wiener fiter and deduce the entanglement from the data we derive the inpat-ouiput relations for both systems indi- vidually and forthe bybrid set-up. The “reponse, equation for the individual ov tors is X)= lh + 2VTE AK Pe)), where signi- fies sign(o,), ‘The effetive. Fourier domain’ suscepbity is genie) ol 2 sr) ining, (xing) be ‘namical broadening by = y,~ Yo ~ 2, parameterized in terme lhe readout rte and 1> >= 1 isthe Fourier frequency) Postve dynamical broadening {, > 0 provides beneficial cooling while adding extra QBA noise. ‘The inputcoutput elation for the optical quadratures re ag ayy" probing the individual oscillators is XY =P, + VTi(+K 1)", showing how ¢ # 0 entails ‘the simultaneous mapping of the oscillator response nto both light quadratures (see Supplementary Section B for details). Tn the hybrid experiment light propagates from the spin ensem ble to the mechanics, and the phases ofthe quadratures ave adjusted by tuning the phate @ between LO, and LO, (Fig. 2) such that Kine = VOR, + VTP owhere Rt, to intersystem losses 1 — v. Whenever ¢y # (ya part of the spin response Xs js mapped, into the optical quadrature driving the mechanics, X} 4, + Pt yg this enables non-local dynamical cool- ing of the combined EPR oscillator, a mechanism related to uncon- ditional entanglement generation", The EPR readout impinging on the detector is 2" — JARS + VT= Wy accounting for the finite EPR detection efficiency ‘Combining those relations, we obtain the EPR readout through the phase quadrature of ight jethe vacuumield de BE = BY 4 val/Tic— VETER) A) BY yal — vo Berane + Be Pasa o STritulPse + VOTH oak] — Be Vt) where Pf" is the measurement imprecision noise including shot noe (SN) and broadband spin noise dv to imperfect motional averaging”. The second line of equation (28) contains the uncor- related noise contributions driving the individual subsystems: fntrinsic thermal and ground-state noise F and the extraneous QBA Ki The thermal forces actingon subsystem jare suppressed ducto the ‘dynamical cooling 5, > 0 (contained in x) occurting locally at each subsystem, Additionally, the thermal spin response Fs is farther ashe z Fraquaney 2 (M2) Dlg Messurerontrige St Spine thermal —Theoy Mechanics hemal 7/7 Baceston wore Freguoney zee) Frequency 2 NH) Fig. 3 | Quantum noise spectra ofthe hybrid system. a, Optical phase quadrature power special densities (PSD) far she measurements of the indicualoseilbtors (detuned by 110 KH), in nts of shot nose (SN). “The feature at “.259 MHz i due to laser phase niseb Joint spectrum 2 the EPR system for he spin escilator tuned cose to resonance with the mechanical escilater. Notably the relative (as well as absolute) amount of QBA noiseinche jit signals signtcartly reduced comsared with that ‘ofthe inaividusl oscillators. e, Wiener fie normalized aasolute square amplitude (bie, le axis) and phase (orange, rght axis) or the resonant hybrs ase, The ‘ering procedure dserns the hybrid system signal rm ‘experimental imperections, for example, he las phase nsise peak suppressed by z5/ ue to the non-local dynamical EPR cooling, introducing the eross-susceptblity 7 f (2) = xyfy(2) ~ i2¢sl' ‘The joint QBA term x X;'; in the first line of equation (21) embodies the central physical mechanism of our scheme resulting fiom the following two interfering processes, fist, the spin system produces squeezed amplitude fluctuations X75 ~ (zs/9)%X¢ that ap into the mechanical phase quadrature fesponse according to Tutus second, the spin QBA response Py ~ Pests is subse- quently filtered by the mechanical system according to lz. We remark thal the function yy suppresses near-resonant spectral components in a bandwidth y, with maximal suppression yl, at 2 oy for 5, > 0). Since ru € I this entails trong suppression ofthe spin QBA response, whereas the amplitude squeezing by the spins more moderate 5/7, = 08. ‘The prefactortoX", may be rewritten as Yl eae) X Tytan + Tato highlighting the condition Ty + Pts = 0 fr total broad band QBA cancellation (independent of dynamical broadening), which requires @y,= ~0,- In the case of unmatched intrinsic line- Widths 70 Yop One til needs ay, = —w, to minimize the term. Equation (28) demonstrates that dynamical broadening enhances (QBA suppression substantially via the factors ‘While QBA reduction is necessary to achieve V< 1 itis not suf- ficient, due tothe inevitable presence of ground-state fluctuations contained in F, (equation (25). These thermal fluctuations, along with residual QBA, can be suppressed by the conditional tracking and/or the coherent dynamical cooling mechanisms (local and non-local) discussed above: here we simultaneously employ both types of mechanism. ‘While equation (2) captures all essential aspects of the involved EPR dynamics, certain technical or peripheral effects were eft out for simplicity These include the finite overcoupling of the optical cavity and the option of introducing optical quadrature rotations between the subsystems as well as in the homodyne detection Pee SS) ‘Morgover, equation (28) neglects the phase noise QBA contributions x Pr Piy to the EPR response, which are minor for the parameter regimie considered here, The fll model accounting forall the afore- ‘mentioned effects was employed in analysing the experimental data {Supplementary Section B) Back-action interference ‘The various noise suppression mechanisms of equation (23) mani: fest themselves in the noise spectra of the optical readout P (ig. 3). When the two oscillators are detuned by 110 kttz by chang- ing w, oB, they are essentially probed separately (Fig. 3). However, > 7 V4.1) and fat readout enV jy > 1 here and henceforth®. We have expressed the Pkylm-+ 1) in terms ofthe thermal bath occupancy n. Within the QND model, C,= QBA/TH. An efi seat measurement of single eystem,n= 1 ad C, > oo, can bring the conditional variance to the ground-state value V8! — 1, but aot belove. Ta the idealized hybrid case with matched readout rates Py 1, =F and intsnsic susceptibilities zy = “Zag implying Oya = wig, © 0 and fo = Yo © 7), the tracking Feces the variance fiom the QBA-free unconditional value V, = 1 + 2n (where now n= (4,4 14912) 0 wh fe vem Tar a ay which shows that ideal QBA cancellation removes the lower bound V2 Lsescited with the sngle-osciltor case Ieis deat from equation () that high QBA/TH >> 1 and high ffcencies are imperative for generation of an entangled state @) am Nees) Or aed g kd-oaenet Frequency 22 (2 : 8 i ‘ome turn yoy eh Fig. 4 | Entanglement tuning and optimization, We sweap the resonant (armor frequency a the spin aselator across the mechanical resonance. This tunes the dstingu'shabilty of the two osellators inthe measured light, thueverying the achievable cantonal PR entanglement ab, Spectr wth respective fits (#) and resulting variance for diferent values of the machanical-spin detuning (b).Black curves ina depict the rmecharies-nly spectrum. Dashed verial ines indicate spin frecuene. Grey curve nb depict variances of optimized EPR combinations, ‘aleulated from theoretical spectra given parameter values from the orange point. Excess expermenal noise leads to expected deviations of ‘his theoretical bound ane experimental yalues. ror bars represent uncertainty of ¥, a calculated by the Markov chain Monte Caro” ‘method (see Metheds) For the spectra corresponding te the black points see Supplementary Fig 8 ‘Numerical simulation confirms the crucial role of those factors and shows that with an intermediate coupling stength QBA/TH ~ 1 and lower efficiencies, such asin re," entanglement generation is not possible. Key new experimental features include an enhanced «qality Lact ofthe mechanical oscillator, an enhanced spin-pho- ton interaction due to the double pass and the overall improved eflciency ofthe hybrid set-up (Supplementary Table 1). The later involves both better direct transmission eficiencies, as well as an ‘optimized coupling of the probe beam tothe collective spin mode leading toa substantial appression of the broadband spin noize ‘When applied to experimental data, the Wiener filter not oaly optimally discerns the EPR signal from white measurement noise, bt also rejects other coloured, peaked or cross-correlated noises Figure 5c presents the steady-state, frequency-domain Wiener il ter Ky(Q) x |, de exp(i)Ky() forthe hybrid case of nearly resonant oscillators, 2 takes into account the fll model ofthe joint system along with experimentally measured noie. In the experiment, the conéitional variance is determined by smany factors, sich as optical losses 1 and 7, a¢ well a mismatched Intrinsic linewidths 7uu # 7m tnd readout rates I, Figure 3 presents results for the cate Fy = 11, Whereas in Fig, 36 we have matched am frequencies wy. = “ey in Fig. 4a we present a series of spectta in ‘which lay is swept through the mechanical resonance by tuning, the B field, The resulting V. fora se of sch measurements (Fig. 1b) ‘exhibits a smooth transition between the regimes of entangled and. non-entangled states of the hybrid system, Notably, our system is rather resilient even to quite substantial oscillator detunings an ~ ‘x. The bandwidth of Vas a function of the detuning is affected by the readout rates, here amounting to several oscillator Kinewidths. Conclusion and outlook ‘We have demonstrated entanglement between distant objects in a hybrid system consisting of a mechanical oscillator and an atomic spin ensemble. This consttules a new milestone for hybrid macro- scopic entanglement and for demonstration of noiseless trajectories in the negative-mass reference frame, ‘This enables quantum communication between distant mechan- {cal and atomic systems using for example, teleportation-based pro- tocols, thereby adding a hitherto missing link tothe hybrid systems landscape’ It paves the road towards, for example, an entangle- ‘meat link between an electromechanically coupled superconduct- ing qubit’"' and a distant atomic ensemble quantum memory. Moreover, the disparate entangled objects respond to very differ ‘ent perturbations and thus facilitate measurements of motion and. fields with reduced quantum noise, for example, of-resonant con- tinuous force detection in gravitational-wave interferometers”, and resonant pulsed measurements based on state preparation and retrodiction”. Future work on enhancing entanglement and achieving practical detection of noiseless trajectories of motion will primarily concen. ‘rate on amending experimental imperfections. Specifically, a factor ‘of 3 reduction of the roadband spin noise by better mode-matching fof light to the atomic ensemble, reduction of intersystem optic cal losses down to 10%, improvement of cavity overcoupling to x= 0.98 and improvement ofthe fractional coherent spin read- ‘out ys BY a factor of 3 wil lead to V. = 0.3 (—5 dB) according to ‘our model ‘Online content Any methods, additional references, Nature Research report: ing summaries, source data, extended data, supplementary infor mation, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at htip/doiorg/10.1058) '41567-020-1031-5, Received: 1 April 2020; Accepted: 6 August 2020; Published online: 21 September 2020 References 1. Horodecki, R, Horodecki P, Horodecki, M. Horodecki K. Quastm tntangement Rev Mod Pye 81, 465-542 (2009) 2, Bogen: C'D. Reinhard F & Capper, P. Quant sensing. Rev Mod Phys. 8 o3s002 Go), 4. Chen, ¥ Macroscopic quantum mechan: theory and experimental concepts of epomechancs | Ph B46, 109001 C1), 4. Arpelneyer M. Kipenberg I] & Marquardt F Cavity optomechaics, ‘Ret od. Py 86, 1391-1452 (2010, 5, Kimble Hh quantum interne Natae 453, 1025-1030 (2008. {6 Kise Geta. Quantum technologies wits hybrid seem. Pro, Nal ‘Acad, Si USA 112, 3866-3873 (2015), 7, Ghee Zaller® Quantum eampetatins with cold taped ions. Phys. Fey Let, 74 4081-4094 (1995) 4, Gross Ce Bloch 1 Quantum simulations with ulsacld stam in opt lates Scene 357, 985-1001 (20 9, Brown, KR, Kim, & Monroe, C. Co designing a sale quantum omer with rapped atomic ions p Quant. 2, 16034 (2016. 10, Hammere K, Aspeimeyer M. Poa 5. Zale, P Esabsing Einstein Polosky- Rosen cane’ between anomechanic and tome ‘ensesles Ps Re Le 102, 020501 (2009) Dae) Pee SS) 1. Mangkhova A.D, Rakhuborsky. A, A. & Filip, R Pulsed tom-mechanial quantum non-dewoltion eae np Quart. Inf 6.4 (2920) 12. usar B. Konhehn, A: & Polak F 5. Experimental long ved ‘atnglerent of wo macoscopic objets. Nate 413, 400—405 (2002), 15. Polk E'§.& Hammerer K Lrjectaries whol quntum uncertainties ‘noe ly, 537, ASABD 2), Maller € et al. Quantum back scan evading measurement of motion Sa 4 egutie mae ference frame. Netire 547, 191-198 2017) 15. Trng, M.& Cater CM. Coherent quanta noise cancion for optomechanical sensors Phys.Rev Lett 105, 123601 (2010) 16 Hammeret K, Sorensen, A'S Poa F. 5. Quantum interface beoeen Ihr and stmieeavembles Rey Mod. Pys. 82, 101-1093 (2010) 17 Muschi CA, Pol, BS Cra [1 Dusipatvcly dives entanglement lta macroscopic slic esemblee Py Re #3, 052312 201), KKemter Heal Entanglement generied by dipation and sendy sate ceanglement of wo macroscopic objets. Phys. Rex Let, 107, 040503 201). 19, Stannige Ks Rabl P& Zoller P Drven-dipave preparation of entangled sate sn cacadedqonn-opcl networks, New [hye 14 063014 (2013), 2a Westy, DV, Mosehi CA Hammerer,K Disipalve versus conditional scoeniion of Guerin caangement and spin squeezing, Py Rev. A #7, (s3120 2019, 2. Tang Xe a Unconditional enya estnglement in macroscopic Iybrd systems by eaerent noise cancelation Plyr Rev Lett 12, 103602 (2018), 22 Thang, M.& Caves, C. M. Evading quantum mechanics: engineering 4 cassie subspter wthin «quantum entronment. Phys Rev 2, si0i6 (2012, 28 Keble |, Gerber J A, Dowd, £ & Stamper Kur D.M. Negative mass Sasa ofthe pia sad motion ofan ami gar deven by epi aisy Duckacton. Pye Ren Lett 120, 01360 (2018). 24 Kg. TM et Light medinted strong coupling between + mechani ‘xllator and store spins 1 meter apart. Sienct 368, 1742179 (200), 25 cial A. etal Sympathese cooking of membrane osilor in a hybrid smechani-atml stem. Nat. Manado! 1055-89 (2015) 26, Chugh, Pet a Combined feedback and wap cooing of 4 mechani oslo coupled to alacld atoms New f Phy 20, os3m0 (2018), 20. Tan H, Buchmann, 1, Seok HS LL G Achieving sendy sate cntngiement of emote mietomechanicalaciltrs by cacided cavity coupling Phy. Rev A 87, 02318 QUID, 28 Woolley M1. & Clerk, AA, Two-modeback-action evading mewsurements sn carly optomedhanes. Phy ey A 87, 063836 (201). 28, Gdkelort-Rorpp Fecal Sabledestanglesoat of massive snecanical tuclstos Natur 56, 478-482 (2018). 50 Riednger Ret al. Remote quantum entanglement betwen 80 rramechanicloriltre Netire S56, 4173-47 (018) 3. Lee. KC. ak Entanging macroseopi amanda oom temperatre Seen 334, 1253-1286 (010. 2 Thaturyan, Wy, Burg, A, Polak B.S. & Seles A. Utzacobecent ‘banomechanial resonators wa soft camping and dsipaion dlton [at Nanotech 12, 76-783 (2017), 38. Boreegaue, [el a Saale photonic neiworksteitecture based tn intimal seraging in toe tempers gs Na. Coram 7, 11386 2016) 3M. Holstein T & Primakot 1, Fie dependence ofthe intrinsic domain magnetization of fetromagnet Pye Rev 54, 1094-1113 (190), Doan. LM, Ged, G, Clg] & Zoe, Poseptabiy exten for onlious varable spstess. Pi ev Let 84, 2722-2725 (200). 236, Miller Hbhardl Het a! Quantum tale preparation and macroscopic ‘clanglement in graialional wave detectors Pye. Rey A 2, 2002 (208), Roast M, Maron D, Chen, Jf Seiler A. Observing and veriying the quant trajectory ofa mechanial eronator Pye Ret. Le 123 13601 G01) ‘Wecrorek W et Optunal sae estimation for cat optomedhania spect Piys Rev. at 144, 323601 G13), Wanless Weta. Qusntum aoe lated and eptanglemeat-ausited smagaclometey Ps. Rev Let. 104, 13350 (2010). 40, Balin MEV ct a High guaity at lation contig material for alll slom vapor cli. Opt. Hsp 1 5825-5830 (2010, Geremia. | M Stoo, K & Mabuci 1. Tensor polariabity and

You might also like