LETTER. Squeezed light from a silicon micromechanical resonator

Transcription

1 oi:.38/nture237 Squeeze light from silion miromehnil resontor Amir H. Sfvi-Neini,2 *, Simon Grölher,2 *, Jeff T. Hill,2 *, Jsper Chn, Mrkus Aspelmeyer 3 & Oskr Pinter,2,4 Monitoring mehnil ojet s motion, even with the gentle touh of light, funmentlly lters its ynmis. The experimentl mnifesttion of this si priniple of quntum mehnis, its link to the quntum nture of light n the extension of quntum mesurement to the mrosopi relm hve ll reeive extensive ttention over the pst hlf-entury,2. The use of squeeze light, with quntum flututions elow tht of the vuum fiel, ws propose nerly three ees go 3 s mens of reuing the optil re-out noise in preision fore mesurements. Conversely, it hs lso een propose tht ontinuous mesurement of mirror s position with light my itself give rise to squeeze light 4,5. Suh squeeze-light genertion hs reently een emonstrte in system of ultrol gs-phse toms 6 whose entre-of-mss motion is nlogous to the motion of mirror. Here we esrie the ontinuous position mesurement of soli-stte, optomehnil system frite from silion mirohip n omprising miromehnil resontor ouple to nnophotoni vity. Lser light sent into the vity is use to mesure the flututions in the position of the mehnil resontor t mesurement rte omprle to its resonne frequeny n greter thn its therml eoherene rte. Despite the mehnil resontor s highly exite therml stte ( 4 phonons), we oserve, through homoyne etetion, squeezing of the reflete light s flutution spetrum t level per ent elow tht of vuum noise over nwith of few meghertz roun the mehnil resonne frequeny of 28 meghertz. With further evie improvements, on-hip squeezing t signifint levels shoul e possile, mking suh integrte mirosle evies well suite for preision metrology pplitions. The genertion of sttes of light with flututions elow tht of vuum hs een of gret theoretil interest sine the 97s 3,7 9. Erly experimentl work emonstrte squeezing of few per ent elow the vuum noise level in lrge vriety of ifferent nonliner systems, suh s neutrl toms in vity, optil fires n rystls with ulk optil nonlinerities 2,3. Moern experiments emonstrte squeezing of lmost 3 B (ref. 4). Initil reserh ws minly pursue s strtegy to mitigte the effets of shot-noise, the mnifesttion of vuum noise in the intensity etetion of light, given the possiility of improve optil ommunition 7 n etter sensitivity in grvittionl-wve etetors 3,8. In reent yers, in ition to eing use in grvittionl-wve etetors 5, squeeze light hs enhne metrology in more pplie settings 6,7. The vuum flututions rising from the quntum nture of light etermine our ility to resolve mehnil motion optilly, n set limits on the perturtion use y the t of mesurement 8.A system well suite to stuying quntum mesurement experimentlly is tht of vity optomehnis, in whih n optil vity s resonne frequeny is esigne to e sensitive to the position of mehnil system. By monitoring the phse n intensity of the reflete light from suh vity, ontinuous mesurement of mehnil isplement n e me. Systems operting on this simple priniple hve een relize in vriety of experimentl settings, suh s in lrge-sle lser grvittionl-wve interferometers 9, mirowve iruits with eletromehnil elements 2, soli-stte mehnil elements 2 23 n ultrol gs-phse toms 6,24 integrte with or omprising Fry Pérot vities, n on-hip nnophotoni vities sensitive to mehnil eformtions 25,26. The nonil vity-optomehnil system onsists of n optil vity resonne tht is ispersively ouple to the position of mehnil resonne. The Hmiltonin esriing the intertion etween light n mehnis is H int ~Bg ^ {^^x x zpf, where ^x~x zpf (^ { z^) is p the mehnil position, x zpf ~ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi B=2m eff v m is the zero-point flutution mplitue, v m is the mehnil resonne frequeny, m eff is the effetive motionl mss of the resontor, g is the frequeny shift of the optil resonne for mehnil mplitue of x zpf, B is Plnk s onstnt ivie y 2p, ^ n ^ { re respetively the nnihiltion n retion opertors for optil exittions, n ^ n ^ { re the nlogous opertors for mehnil exittions. The optil vity ey rte, k, is the loss rte of photons from the vity n the rte t whih optil vuum flututions re ouple into the optil resonne 27. Similrly, the mehnil mping rte, i, is the rte t whih therml th flututions ouple to the mehnil system. In ll experimentl reliztions of soli-stte optomehnis so fr, inluing tht presente here, the optomehnil oupling rte, g, hs een muh smller thn k. As suh, without strong oherent rive, the intertion of the vuum flututions with the mehnis is negligile. Uner the effet of oherent lser rive, the vity is populte with men intrvity photon numer Æn æ, n we ponsier the optil flututions out the lssil stey stte, ^? ffiffiffiffiffiffiffiffi hn i z^. This moifies the optomehnil intertion, resulting in liner oupling etween the flututions of the intrvity optil fiel, ^X ~^z^ {, n the position flututions of the mehnil system, ^x: H int ~BG^X ^x x zpf. The pprmetri liner oupling ours t n effetive intertion rte of G: ffiffiffiffiffiffiffiffi hn i g, n the mehnil motion is ouple to the intrvity optil fiel t rte of C mes ; 4G 2 /k. Through this intertion, the intensity flututions of the vuum fiel, ^X (in) h~ (t), entering the vity imprt fore on the mehnil system: ^F BA (t)~ B p ffiffiffiffiffiffiffiffiffiffiffi C mes x zpf ^X (in) h~ (t) ðþ This rition-pressure shot-noise (RPSN) fore hs previously een mesure in n ultrol tomi gs 24 n, more reently, on mrosopi silion nitrie nnomemrne 28. The mehnil motion is in turn reore in the phse of the light leving the vity: ^X (out) h (t)~{^x (in) h (t){2 pffiffiffiffiffiffiffiffiffiffiffi C mes x zpf ^x(t) sin (h) ð2þ Here ^X (j) h ~^ je {ih z^ { j eih, ^ in n ^ out re respetively the opertors of the input n reflete optil fiels from the vity, n h is the qurture ngle with h 5 n h 5 p/2 orresponing respetively to the intensity n phse qurtures. In suh mesurement, the optil vity hs the role of the position etetor, mesuring the oservle ^x t rte C mes, n the RPSN imposes mesurement k-tion fore on the mehnil system 2. In ition to this k-tion noise, therml Kvli Nnosiene Institute n Thoms J. Wtson, Sr, Lortory of Applie Physis, Cliforni Institute of Tehnology, Psen, Cliforni 925, USA. 2 Institute for Quntum Informtion n Mtter, Cliforni Institute of Tehnology, Psen, Cliforni 925, USA. 3 Vienn Center for Quntum Siene n Tehnology, Fulty of Physis, University of Vienn, A-9 Wien, Austri. 4 Mx Plnk Institute for the Siene of Light, Günther-Shrowsky-Strße /Blg 24, D-958 Erlngen, Germny. *These uthors ontriute eqully to this work. 8 AUGUST 23 VOL 5 NATURE 85

2 RESEARCH LETTER flututions from the th lso rive the mehnil motion, with their mgnitues eoming omprle s C mes pprohes the thermliztion rte, C therml (v): i n(v), wheren(v) is the therml th oupny. Formlly, the output noise power spetrl ensity (PSD) of the homoyne etetor photourrent, I, normlize to the shot-noise level is foun y tking the Fourier trnsform of the utoorreltion of eqution (2): S out II (v)~z 4C mes x 2 zpf S xx sin 2 (h)z B 2 Refx mg sin (2h) where S xx (v) is the noise PSD of the mehnil position flututions of the resontor n x m (v)~(m eff (v 2 m {v2 {i i v m )) { is the mehnil suseptiility hrterizing the response of the mehnil system to n pplie fore. The PSD S xx (v) ontins noise stemming from oupling to the therml th, quntum k-tion noise from the light fiel n ny other tehnil lser noise riving the mehnis (Supplementry Informtion). The three terms in S out II (v) in eqution (3), from left to right, re ue to shot-noise, mehnil position flututions n the ross-orreltion etween the k-tion noise fore n mehnil position flututions. Only the thir term n hve negtive noise PSD n give rise to squeezing. The primry hurle to oserving suh squeezing, s in mny quntum mesurements, is strongly oupling to preferre etetion hnnel while simultneously minimizing unwnte environmentl perturtions. Most relevnt to the work presente here re frequenies etune from resonne, v m. j ; v m 2 vj? i, for whih the pproximte output noise PSD is S out II (vw) <z 2C mes v m n(v) Q m ð{ os (2h) Þz sin (2h) where we hve ssume tht the mehnil position flututions re preominntly ue to therml th oupling (t high optil power, k-tion heting n lser noise lso ontriute to the moe oupny). In this se, flututions from the therml th limit ppreile squeezing of the optil proe fiel to regime in whih n(v)=q m =. This requirement is equivlent to hving Q m v m > k B T =B, where k B is Boltzmnn s onstnt n T is the th temperture. Appreile squeezing lso requires etuning jj> ðn=q m Þv m n orresponing mesurement rte lrger thn this etuning. Therefore, in the presene of therml th noise, squeezing over signifint spetrl nwith requires not only lrge oopertivity (C ; C mes / i ) etween the optil n mehnil omponents, s relize in reent ooling experiments 2,26,29, ut the more stringent requirement tht the effetive mesurement k-tion fore e omprle to ll fores ting on the mehnil resontor, inluing the elsti restoring fore of the mehnil struture. To meet the requirements of strong mesurement n effiient etetion, we esigne zipper-style optomehnil vity 25 with novel integrte wveguie oupler frite from the 22-nm-thik silion evie lyer of silion-on-insultor mirohip (Fig. ). The in-plne ifferentil motion of the two ems t funmentl frequeny of v m /2p 5 28 MHz strongly moultes the o-lolize funmentl optil resonne of the vity with theoretil vuum oupling rte of g /2p 5 MHz. As shown in Fig., we use silion wveguie with high-refletivity photoni rystl en-mirror to exite n ollet light effiiently from the optil vity. Light from the silion wveguie is ouple to single-moe optil fire using n optil fire tper n omintion of iti moe oupling n trnsformtion. The experimentl set-up use to hrterize the zipper vity system n mesure the optomehnil squeezing of light is shown in Fig. 2. The silion smple is ple in ontinuous-flow 4 He ryostt with ol-finger temperture of K. A nrrown lser em is use to proe the optomehnil system n mesure the mehnil ð3þ motion of the zipper vity. A wvelength sn of the reflete signl from the vity is plotte in Fig. 2, showing n optil resonne with linewith k/2p GHz t wvelength of l 5,54 nm. Ineffiienies in the olletion n etetion of light result in itionl unorrelte shot-noise in the mesure signl n n reue the squeezing to unetetle levels. For the evie stuie here, the vity oupling effiieny, orresponing to the perentge of photons sent into the vity whih re reflete, is etermine to e g k The fire-to-hip oupling effiieny is mesure t g CP 5.9. A homoyne etetion sheme llows for high-effiieny etetion of ritrry qurtures of the optil signl fiel. Chrteriztion n optimiztion of the effiieny of the entire optil signl pth n homoyne etetion system yiele n overll set-up effiieny of g set-up 5.48, orresponing to totl signl etetion effiieny of g tot 5 g set-up g k Figure 2 shows the noise spetrum of the therml motion of the mehnil resontor otine y setting the lser frequeny ner the vity resonne n tuning the reltive lol-osilltor phse of the homoyne etetor, h lok, to mesure the qurture of the reflete signl in whih mehnil motion is imprinte. The mehnil spetrum shows the in-plne ifferentil moe of interest t v m / 2p 5 28 MHz, s well s severl other more wekly ouple mehnil resonnes of the nnoems n oupling wveguie. A highresolution, nrrown spetrum of the in-plne ifferentil moe is inset in Fig. 2, n shows linewith of i /2p 5 72 Hz, orresponing to mehnil Q-ftor of Q m The vuum oupling rte of the in-plne ifferentil moe, mesure from the etuning epenene of the optil spring shift n mping, is etermine to e g /2p 5 75 khz, in goo greement with theory. From the lirtion of the noise power uner the Lorentzin istriution in Fig. 2, the in-plne ifferentil moe is foun to thermlize t low optil proe power to temperture of T < 6 K, orresponing to therml phonon oupny of n(v m )<:2 4. This yiels rtio 5 μm μm 2 Figure Optomehnil evie., Snning eletron mirosope imge of wveguie-ouple zipper optomehnil vity. The wveguie with is itilly tpere long its length n termintes with photoni rystl mirror next to the vity. The tpering of the wveguie llows for effiient input output oupling n the photoni rystl termintion mkes the oupling to the vity single sie. Two zipper vities re ouple ove n elow the wveguie, eh with slightly ifferent optil resonne frequeny, llowing them to e seprtely resse., Left: lose-up of the oupling region etween one of the vities n the wveguie. Right: finite-element metho (FEM) simultion of the vity fiel leking into the wveguie (log sle). Note tht the fiel oes not lek into the mirror region of the wveguie., Top: FEM simultion showing the in-plne eletril fiel of the funmentl optil vity moe. Bottom: FEM simultion of the isplement of the funmentl in-plne ifferentil moe of the struture with frequeny v m / 2p 5 28 MHz. The mehnil motion, moifying the gp etween the ems, shifts the optil vity frequeny, leing to optomehnil oupling. 86 NATURE VOL 5 8 AUGUST 23

3 RESEARCH Normlize refletion λ,54 nm λ-meter,54.,54.9,54.7 Wvelength (nm) Δlok.44κ κ/2π = 3.42 GHz Normlize etuning, Δ/κ IM IM 2 3 ~ 88 MHz 6 2 Figure 2 Experimentl set-up n evie hrteriztion., The optil signl is erive from n externl-vity ioe lser n is sent into tpere optil fire insie liqui-helium ryostt where the silion smple is oole to T < 6 K. The fire tper is use to ouple light evnesently into the silion optomehnil evie. The optil refletion from the evie is ollete y the sme fire tper n sent to lne homoyne etetor (BHD) for hrteriztion. For further etils of the experimentl set-up, see Methos Summry. The effiienies of the irultor, swith n BHD re respetively enote g 23, g 3H n g HD. AOM, ousto-opti moultor; EDFA, eriumope fire mplifier; ENA, eletroni network nlyser;, fire polriztion ontroller; FS, fire strether; IM, intensity moultor; l-meter, wvemeter; LPF, low-pss filter; PD, photoetetor; PM, power meter; RSA, rel-time spetrum nlyser; VC, vrile oupler; VOA, vrile optil ttenutor., Top: reflete signl from the optil vity t low optil power (Æn æ < ; linewith, k/2p GHz). Bottom: high-power (Æn æ 5 79) PSD (Bm Hz ) FS AOM LPF VOA BHD 2 2 (ω ω m )/2π (khz) VC RSA η 3H = 92% η HD = 66% e BHD PSD (pw Hz ) BHD reltive PSD η 23 = 88% η CP = 9% Tper Liqui-He ryostt EDFA RSA ENA PD PM Drk urrent LO power (mw) PD2 reflete signl, showing the vity-lser etuning (she line) loke to uring squeezing mesurements., Homoyne noise PSD of the reflete signl showing the trnsue therml Brownin motion of the zipper vity t T 5 6 K (green urve; Æn æ 5 8). The re urve is the shot-noise level n the lk urve is the etetor s rk noise (in the sene of light input). Inset, lose-up of the funmentl in-plne ifferentil mehnil moe of the zipper vity (fitte y lue urve; linewith, i /2p 5 72 Hz)., Men vlue of the PSD of the BHD s funtion of the lol-osilltor (LO) power (signl loke). The fille t point inites the lol-osilltor power use in the squeezing mesurements. The re n she lk urves orrespon to liner fit to the t n the level of the etetor rk urrent, respetively. e, Noise PSD s funtion of h lok (rnging from (green) to p (re)) with the signl etune fr off resonne t D/k < 3, referene to the noise level with the signl loke (lue). 4 n(v m )=Q m <:72, whih is well within the regime where squeezing is possile. To stuy urtely the noise properties of the reflete optil signl from the vity, we mke series of mesurements to hrterize our lser n etetion set-up. Figure 2 shows the mesure noise PSD of the lne homoyne etetor for v < v m s funtion of lol-osilltor power, initing liner epenene on power n negligile (=.%) e noise ove shot-noise. In the mesure squeezing t isusse elow, lol-osilltor power of 3 mw is use. Clirtion of the lser intensity n frequeny noise over the frequeny rnge of interest (v/2p 5 4 MHz) is esrie in Supplementry Informtion. The lser intensity noise is mesure to e ominte y shot-noise over this frequeny rnge, n the lser frequeny noise is mesure to e roughly flt in frequeny t level of S vv < r 2 Hz, whih ontriutes insignifintly to the etete noise floor. Figure 2e shows the mesure homoyne etetor noise level normlize to vuum noise for reflete lser light fr-etune from the optil vity resonne. The smll (, 6.5%) evition in the mesure noise level ouns the systemti unertinty in the etetor gin versus qurture is point s well s the ontriution to the mesure noise from optil elements other thn the silion vity. Mesurements of the noise in the reflete optil signl from the vity s funtion of qurture ngle, frequeny n signl power re shown in Figs 3 n 4. These mesurements re performe for lser light on resonne with the optil vity n for input signl powers vrying from 252 nw to 3.99 mw, with the mximum signl power orresponing to n verge intrvity photon numer of Æn æ 5 3,53. The lser is set t the pproprite vity etuning for eh signl power y snning the wvelength ross the vity resonne while monitoring the refletion, n then stepping the lser frequeny towrs the vity from the long-wvelength sie until the refletion mthes the level tht orrespons to etuning of D lok /k This proues shift of w 5 ( )p in the mesure qurture ngle from the on-resonne onition of equtions () (3) (Methos Summry). In Fig. 3, we plot the theoretilly preite n mesure noise PSDs versus qurture ngle for signl power orresponing to Æn æ 5 79 photons. Eh qurture spetrum is the verge of 5 tres, n fter every other spetrum the signl rm is loke n the shot-noise PSD is mesure. The shot-noise level, orresponing to optil vuum on the signl rm, is use to normlize the spetr. At ertin qurture ngles, n for frequenies few meghertz roun the mehnil resonne frequeny, we fin tht the light reflete from the zipper vity shows noise PSD elow tht of vuum. The ensity plot of the theoretilly preite noise PSD (Fig. 3) shows the expete wien squeezing ue to the strong optomehnil oupling in these evies, s well s hnge in the phse ngle where squeezing is oserve elow n ove the mehnil frequeny. This hnge is ue to the hnge in sign of the mehnil suseptiility n the orresponing hnge in phse of the mehnil response to 8 AUGUST 23 VOL 5 NATURE 87

4 RESEARCH LETTER Figure 3 Optomehnil squeezing of light., Theoretil moel. Density plot of the preite reflete-signl noise PSD, s mesure on lne homoyne etetor n normlize to shot-noise, for simplifie moel of the optomehnil system (Supplementry Informtion). Ares with noise elow shot-noise re shown in lue shes on liner sle. Ares with noise ove shot-noise re shown in ornge shes on log sle. The soli white line is ontour elineting noise ove n elow shot-noise., Experimentl t. Density plot of the mesure reflete-signl noise PSD for Æn æ 5 79, normlize to the mesure shot-noise level., Slie of the mesure ensity plot in tken t h lok /p 5.23., Slie of the mesure ensity plot in tken t h lok /p 5.6. In n, the lk urve orrespons to the mesure t slie extrte from. The rk lue tres re severl mesurements of the shot-noise level (verge shown in light lue). Also shown is moel of the squeezing in the sene of therml noise (ornge urve), the sme moel with ohmi therml noise of the mehnil moe inlue (green) n full noise moel inluing itionl phenomenologil noise soures (re urve). The vertil white she lines in n inite the t slies shown in n. RPSN. The mesure noise PSD ensity plot (Fig. 3) shows the presene of severl other mehnil noise peks n reue squeezing nwith, yet the overll phse- n frequeny-epenent hrteristis of the squeezing roun the strongly ouple in-plne mehnil moe re lerly present. In prtiulr, Fig. 3 n Fig. 3 show two slies of the noise PSD ensity plot in whih the region of squeezing hnges from eing elow the mehnil resonne frequeny to eing ove it. In Fig. 4, we show the mesure noise PSD s funtion of qurture ngle for frequeny slie t v/2p MHz of the t shown in Fig. 3. The mesure squeezing n nti-squeezing re seen to e smller n, respetively, lrger thn expete from moel of the optomehnil vity without therml noise. We lso plot, in Fig. 4, the mximum mesure n moelle squeezing s funtions of signl power. The simple theory preits squeezing level tht monotonilly inreses with signl power, wheres the mesure mximum squeezing sturtes t level of % elow the shot-noise t n intrvity power orresponing to Æn æ 5,984 photons. The error in the squeezing is ominte y the unertinty in the linerity of the etetor gin (6.5%) n the vrine of the mesure shot-noise level (6.%). To unerstn the proesses tht limit the nwith n mgnitue of the mesure squeezing, we plot in Fig. 4 the noise PSD for phse qurtures tht mximize (left plot) n minimize (right plot) the trnsution of the mehnil moe pek. Along with the mesure t, we lso plot the estimte noise ue to phse noise of the signl lser, n tht for moel of single mehnil moe ouple to therml th t T 5 6 K. Low-frequeny noise in the motion qurture shows n v 2 frequeny epenene onsistent with struturl mping effets 3, ut is muh lrger thn tht of the single-moe noise moel. Noise in the qurture tht minimizes trnsue motion is orers of mgnitue lrger thn the noise preite y the single-moe moel n lser phse noise, n shows n v 2/2 frequeny epenene. The optil power epenene of the low-frequeny noise in the motion qurture inites tht optil sorption hets the struture to T < 3 K t the highest mesure powers. As etile in Supplementry Informtion, the re urve in eh of the plots in Fig. 4 shows full noise moel inorporting struturl mping noise from higher-frequeny mehnil moes, optil sorption heting n phenomenologil v 2/2 noise term. (shot-noise sutrte) ω ω > ω m 2, 2, Intrvity photon numer, n ω /2 ω < ω m =.8 = Figure 4 Spetrl n power epenene of noise., Mesure lne homoyne noise power of the reflete signl t v/2p MHz (fille irles) versus qurture ngle (D lok /k 5.44 n Æn æ 5 79). The green n re urves orrespon to the single-moe n full noise moels, respetively. Theornge urve represents moel inluing the response of the mehnil moe in the sene of therml noise, tht is, when riven y RPSN only, n the she lue urve shows the therml noise omponent. Inset, lose-up of oxe region., Mesure minimum noise PSD normlize to shot-noise (fille irles) versus Æn æ.left: mximum squeezing for v, v m ; right: mximum squeezing for v. v m.also shown re the single-moe noise moel (green urve) n the full noise moel (re urve)., Blne homoyne noise PSD of the reflete vity signl for D lok / k 5.52 n Æn æ 5 3,53. Left: phse qurture orresponing to mximum trnsution of mehnil motion; right: phse qurture orresponing to minimum trnsution of mehnil motion. In eh plot, the lk urve is the mesure t with the shot-noise level sutrte. Also shown re the moelle lser phse noise (she rown urve), the noise ontriution from single mehnil moe (she lue urve) n the full noise moel (re urve). 88 NATURE VOL 5 8 AUGUST 23

5 RESEARCH These moels inite tht the urrently otinle levels of squeezing re limite y the therml noise of higher-orer mehnil moes. These mesurements show tht y refleting light off thin-film mehnil resontor unergoing lrge-mplitue therml motion, light tht is in ertin respets quieter thn vuum n e otine. In ontrst to previous work with ultrol gs-phse toms 6, whih use nrrown tomi resonne n operte in short ursts owing to the tomi trp lifetime, the soli-stte evies in this work llow for stey-stte squeezing over lmost MHz of nwith n t optil frequenies ompletely tilorle through geometri esign. The moest level of squeezing, limite y therml noise n struturl mping effets in the urrent evies, my lso e sustntilly improve y inresing the mehnil Q-ftor. Mesurements of similr silion evies with ifferent surfe tretments hve yiele mehnil Q-ftors s high s (Supplementry Informtion), whih, for the optil power levels use here, shoul enle on-hip squeezing 6 B elow shot-noise. Given the integrility of these mirohip evies, optil extrtion ineffiienies n e voie y sening the squeeze light generte y one evie iretly into seon proe evie. For exmple, suh n on-hip squeezer n etetor oul form the sis of quntum-enhne miromehnil isplement n fore sensor 7. More generlly, we expet these sorts of evie to enle future experiments involving feek n strong mesurement of the ynmis of mehnil system. METHODS SUMMARY Experimentl set-up. A tunle externl-vity ioe lser, tively loke to wvemeter, is use to generte strong lol osilltor n the mesurement signl. Frtions of the lol osilltor n input signls re split off n etete, using intensity moultors to stilize their power levels. Fire polriztion ontrollers just the polriztion of the lol osilltor n signl. A vrile optil ttenutor is use to set the signl power n n ousto-opti moultor is use to generte tone for lirtion (Supplementry Informtion). The reflete signl from the vity is seprte using irultor, n is swithe etween one of three etetion pths: one ontins power meter for power lirtion, one ontins photoetetor (PD) for spetrosopy of the vity n one ontins n eriumope fire mplifier for mesurement of the mehnil spetrum on rel-time spetrum nlyser or network nlyser. Squeezing of the reflete vity signl is mesure on fourth pth ontining vrile oupler, where the signl is reomine with the lol osilltor n etete on lne homoyne etetor. Homoyne phse ngle. The reltive phse etween the lol osilltor n the signl is etermine from the low-pss-filtere omponent of the signl etete using lne homoyne etetor, n set using fire strether. Loking the level of this signl etermines the phse ngle etween the lol osilltor n the reflete signl, whih we ll h lok. This ngle iffers from the phse h etween the light input to the vity n the lol osilltor, ut is relte to it through the phse response of the vity: h lok 5 h 2 w, where w(d)~arg½{k e =(idzk=2) Š n k e is the extrinsi vity oupling rte. Cvity lok. A lok point slightly re etune from resonne is hosen to voi instilities of the system resulting from thermo-optil effets. The lser is loke to this frequeny using wvemeter with frequeny resolution of 6.5k. Drift of the optil vity resonne over single noise spetrum mesurement (orer of minutes) is foun to e negligile. An estimte of the vrine of D lok is etermine from the epenene of the trnsution of the mehnil motion on the qurture phse, whih inites tht from one lok to nother D lok /k Reeive 25 Ferury; epte 6 My 23.. Brginsky, V. & Khlili, F. Quntum Mesurements (Cmrige Univ. Press, 995). 2. Clerk, A. A., Devoret, M. H., Girvin, S. M., Mrqurt, F. & Shoelkopf, R. J. Introution to quntum noise, mesurement, n mplifition. Rev. Mo. Phys. 82, (2). 3. Cves, C. M. Quntum-mehnil noise in n interferometer. Phys. Rev. D 23, (98). 4. Fre, C. et l. Quntum-noisereutionusing vitywith movle mirror. Phys. Rev. A 49, (994). 5. Mnini, S. & Tomesi, P. Quntum noise reution y rition pressure. Phys. Rev. A 49, (994). 6. Brooks, D. W. C. et l. Non-lssillight generteyquntum-noise-rivenvity optomehnis. Nture 488, (22). 7. Yuen, H. P. Two-photon oherent sttes of the rition fiel. Phys. Rev. A 3, (976). 8. Hollenhorst, J. N. Quntum limits on resonnt-mss grvittionl-rition etetors. Phys. Rev. D 9, (979). 9. Wlls, D. F. Squeeze sttes of light. Nture 36, 4 46 (983).. Slusher, R. E., Hollerg, L. W., Yurke, B., Mertz, J. C. & Vlley, J. F. Oservtion of squeeze sttes generte y four-wve mixing in n optil vity. Phys. Rev. Lett. 55, (985).. Shely, R. M., Levenson, M. D., Perlmutter, S. H., DeVoe, R. G. & Wlls, D. F. Bron prmetri emplifition of quntum noise in n optil fier. Phys. Rev. Lett. 57, (986). 2. Wu, L.-A., Kimle, H. J., Hll, J. L. & Wu, H. Genertion of squeeze sttes y prmetri own onversion. Phys. Rev. Lett. 57, (986). 3. Grngier, P., Slusher, R. E., Yurke, B. & LPort, A. Squeeze-light-enhne polriztion interferometer. Phys. Rev. Lett. 59, (987). 4. Mehmet, M. et l. Squeeze light t 55 nm with quntum noise reution of 2.3 B. Opt. Express 9, (2). 5. The LIGO Sientifi Collortion. A grvittionl wve oservtory operting eyon the quntum shot-noise limit. Nture Phys. 7, (2). 6. Tylor, M. A. et l. Biologil mesurement eyon the quntum limit. Nture Photon. 7, (23). 7. Hoff, U. B. et l. Quntum-enhne miromehnil isplement sensitivity. Opt. Lett. 38, (23). 8. Cves, C. M. Quntum-mehnil rition-pressure flututions in n interferometer. Phys. Rev. Lett. 45, (98). 9. Aott, B. et l. Oservtion of kilogrm-sle osilltor ner its quntum groun stte. N. J. Phys., 7332 (29). 2. Teufel, J. D. et l. Sien ooling of miromehnil motion to the quntum groun stte. Nture 475, (2). 2. Gign, S. et l. Self-ooling of miromirror y rition pressure. Nture 444, 67 7 (26). 22. Arizet, O., Cohon, P.-F., Brint, T., Pinr, M. & Heimnn, A. Rition-pressure ooling n optomehnil instility of miromirror. Nture 444, 7 74 (26). 23. Coritt, T., Ottwy, D., Innerhofer, E., Pel, J. & Mvlvl, N. Mesurement of rition-pressure-inue optomehnil ynmis in suspene Fry- Perot vity. Phys. Rev. A 74, 282 (26). 24. Murh, K. W., Moore, K. L., Gupt, S. & Stmper-Kurn, D. M. Oservtion of quntum-mesurement ktion with n ultrol tomi gs. Nture Phys. 4, (28). 25. Eihenfiel, M., Cmho, R., Chn, J., Vhl, K. J. & Pinter, O. A piogrm- n nnometre-sle photoni-rystl optomehnil vity. Nture 459, (29). 26. Chn, J. et l. Lser ooling of nnomehnilosilltorintoits quntumgroun stte. Nture 478, (2). 27. Griner, C. W. & Collett, M. J. Input n output in mpe quntum systems: quntum stohsti ifferentil equtions n the mster eqution. Phys. Rev. A 3, (985). 28. Pury, T. P., Peterson, R. W. & Regl, C. A. Oservtion of rition pressure shot noise on mrosopi ojet. Siene 339, 8 84 (23). 29. Verhgen, E., Deléglise, S., Weis, S., Shliesser, A. & Kippenerg, T. J. Quntumoherentoupling of mehnilosilltortonoptilvity moe. Nture 482, (22). 3. Liu, Y. T. & Thorne, K. S. Thermoelsti noise n homogeneous therml noise in finite size grvittionl-wve test msses. Phys. Rev. D 62, 222 (2). Supplementry Informtion is ville in the online version of the pper. Aknowlegements We woul like to thnk K. Hmmerer n A. A. Clerk for isussions. This work ws supporte y the DARPA/MTO ORCHID progrmme through grnt from the AFOSR; the Institute for Quntum Informtion n Mtter, n NSF Physis Frontiers Center with support of the Goron n Betty Moore Fountion; the Vienn Siene n Tehnology Fun WWTF; the Europen Commission, through IP SIQS n iquoems; n the Europen Reserh Counil. A.H.S.-N. n J.C. grtefully knowlege support from NSERC. S.G. knowleges support from the Europen Commission through Mrie Curie Fellowship. Author Contriutions A.H.S.-N., S.G. n M.A. esigne the experiment. A.H.S.-N., S.G., J.C. n J.T.H. esigne n frite the evie, n performe the mesurements. A.H.S.-N., S.G., J.T.H. n O.P. performe the nlysis n moelling of the t. All uthors were involve in writing n eiting the pper. Author Informtion Reprints n permissions informtion is ville t The uthors elre no ompeting finnil interests. Reers re welome to omment on the online version of the pper. Corresponene n requests for mterils shoul e resse to O.P. (opinter@lteh.eu). 8 AUGUST 23 VOL 5 NATURE 89