Table Of Content

Experimental generation of complex noisy photonic entanglement Krzysztof Dobek,1,2 Michal Karpin´ski,3 Rafal Demkowicz-Dobrzan´ski,3 Konrad Banaszek,1,3 and Pawel Horodecki4 1Institute of Physics, Nicolaus Copernicus University, ul. Grudziadzka 5/7, 87-100 Torun´, Poland 2Faculty of Physics, Adam Mickiewicz University, ul. Umultowska 85, 61-614 Poznan´, Poland 3Faculty of Physics, University of Warsaw, Hoz˙a 69, 00-681 Warsaw, Poland 4Faculty of Applied Physics and Mathematics, Technical University of Gdan´sk, 80-952 Gdan´sk, Poland (Dated: November10, 2011) Wedescribeanexperimentalsetupbasedonspontaneousparametricdown-conversiontoproduce multiplephotonpairsinmaximallyentangledpolarizationstatesusinganarrangementoftwotype-I nonlinearcrystals. Byintroducingcorrelatedpolarizationnoiseinthepathsofthegeneratedphotons one can prepare mixed entangled states whose properties illustrate fundamental results obtained 1 recently in quantum information theory, in particular those concerning bound entanglement and 1 privacy. 0 2 v I. INTRODUCTION used this approach to produce two specific examples of o four-photon mixed entangled states. One of them is the N Smolinstate[5],whosefirstexperimentalgenerationcar- Recent theoretical studies of noisy entanglement re- ried out by Amselem and Bourennane [6] illustrated the 9 sultedindiscoveriesofinterestingphenomenathatoccur delicate nature of the phenomenon of bound entangle- in higher-dimensional systems. One prominent exam- ] ple is provided by bound-entangled states that cannot ment in the presence of experimental imperfections [7]. h Thisinitialworkwassubsequentlyfollowedbythedevel- p be created by local operations and classical communica- opmentofmorerobustpreparationtechniques[8,9]. Our - tion (LOCC), but at the same time cannot be brought t resultsindicatethattheproblemsencounteredinRef.[6] n into a pure maximally entangledformunder LOCC con- are rather generic and do not depend on the specifics of a straints[1]. Anotherprofoundobservationisthatcertain the optical setup. The second example presented here u states can be used to generate the cryptographic key at q is a private state, for which the secure key contents is rates that exceed distillable entanglement [2]. These ef- [ strictly higher than the amount of entanglement distill- fects reveal the highly complex nature of noisy entan- able in the asymptotic regime. These information theo- 1 glement in higher dimensions. Experimental studies of reticpropertieshavebeenverifiedexperimentallyforthe v such entanglement are important for two main reasons. 7 first time in [10]. Here we report an experimental illus- On the fundamental side, it is always vital to confirm 9 tration of a recent theoretical result which links privacy theoretical predictions in actual physical systems. From 2 to incompatibility with local realistic theories [11]. Let a more practical perspective, quantum entanglement is 2 usnotethatnon-trivialformsofnoisyentanglementhave . an essential resource in implementing a wide range of 1 been recently observed also in a system of trapped ions quantum-enhanced protocols for communication, sens- 1 subjected to decoherence induced by spontaneous decay ing, etc. However, its generation and distribution are 1 [12]. usually affected by imperfections which deteriorate the 1 This paper is organized as follows. First, we present : quality of the available resource. Therefore one needs v the experimentalsetup in Sec.II. The examples ofnoisy tools to characterize relevant properties of noisy entan- i entangledstatesalongwithdetailsoftheirgenerationare X glement and to utilize it for practical purposes in an op- describedinSec.III. Resultsoftheircharacterizationare r timal way. a reviewedinSec.IV. Finally, Sec.Vconcludes the paper. In this paper we present an experimental scheme that can be used to produce a wide range of noisy entangled multiphoton states. Our approach exploits the geometric symmetry of the popular scheme to gen- II. EXPERIMENTAL SETUP erate polarization-entangled photon pairs via spontaneous parametric down-conversion (SPDC) realized in Owing to a simple setup and room-temperature op- two type-I crystals [3]. As we describe below, the axial eration, SPDC has been the tool of choice to generate symmetry of this scheme allows one to collect simulta- multiphotonentangledstates. The basic ingredientis an neously several entangled photon pairs with little addi- arrangementtoproduceonemaximallyentangledphoton tionaleffort. Itisworthnotingthatrecentlythe symme- pair. A very popular configuration is based on a single tryoftype-ISPDChasalsobeenexploitedinapractical type-II crystal, with photon pairs collected from the in- methodto generatehigh-dimensionalpathentanglement tersection points of two emission cones with orthogonal of two photons [4]. polarizations [13]. Reflecting the pump pulse and send- Inordertopreparenoisyentangledstatesinthepolar- ing it back to the same crystal allows one to generate ization degree of freedom, the generated photons can be two pairs. This approach,usedto demonstrate quantum subjected to variable correlated birefringence. We have teleportation [14], has been highly refined in subsequent 2 trainof180fspumppulseswiththespectrumcenteredat 390 nm, obtained by frequency doubling in a 1 mm long lithium triborate crystal the output from a Ti:sapphire oscillator(CoherentChameleonUltra),whichresultedin 200mW averagepower. The pumpbeamwasfocusedto a 70 µm diameter spot in the pair of beta-barium borate crystals. The SPDC emission was collimated with a 20 cm focal length lens, sending the generated photons along parallel paths. The idea of this arrangement issimilartothatusedrecentlytogeneratemultipathen- tanglement of photon pairs [4]. In the constructed setup, it was possible to introduce various types of polarization noise by inserting in the paths of one or two photons birefringent elements (half- andquarter-waveplates)withvariableorientations. Two Soleil-Babinet compensators were placed in the pump beam and the path of one of the down-converted photons to control relative phases between the horizontal and verticalcomponents for both the produced pairs. In experiments where photon polarizations were measured individually the photons were filtered through 10 nm interference filters and coupled into single-mode fibers, which delivered them to polarization analyzers shown Figure 1. (a) A schematic of the experimental setup to pro- in Fig. 1(b). Each analyzer consisted of a quarter- and duce multiple polarization-entangled photon pairs. Boxes la- halfwaveplatefollowedbyaWollastonpolarizingbeam- belled with T are polarization analyzers detailed in (b). The splitter,whoseoutputportswerecoupledintomultimode boxNrepresentspolarizationnoiseintroducedbysetsofwave fibers. These fibers guided the photons to avalanche plates shown for the Smolin state (c) and the private state photodiodes operated in the Geiger mode (Perkin-Elmer (d). D, Soleil-Babinet compensator; XX, down-conversion SPCM-AQR). The signals from the detectors were fi- crystsls;IF,interferencefilter;SMF,single-modefiber;QWP, quarter-wave plate; HWP, half-wave plate; PBS, polarizing nally processed using a coincidence circuit with a 6 ns beam splitter; APD, avalanche photodiode. window programmed in an field-programmable gate ar- ray (FPGA) board. All the waveplates were mounted on motorized rotation stages and the whole experiment experiments [15]. An alternative is to send the pump was controlled using a dedicated LabView (National In- pulse through a sequence of type-II crystals [16], which struments) application. Typical count rates in the setup enables generation of three or more pairs at once. Fur- were of the order of 105 Hz for single counts, 104 Hz for thermore,thealignmentcanbe optimizedindependently two-fold coincidences for detectors monitoring the same for each crystal. photon pair, and 2 Hz for four-fold coincidences. Another configuration to produce photon pairs in a maximally entangled polarizationstate utilizes SPDC in twotype-Icrystalswhoseopticalaxesarealignedinper- III. GENERATION OF NOISY STATES pendicularplanes,thusproducingorthogonallypolarized photons [3]. Because the down-converted photons are Weemployedthesetupdescribedintheprecedingsec- generated as ordinary rays, they emerge on an axially tion to generate and characterize noisy four-qubit states symmetric cone. With a suitably adjusted polarization that illustrate interesting phenomena occurring in the of the pump beam both the crystals produce photon theory of high-dimensional entanglement. The first ex- pairs with the same probability. If the photons gener- ample was the Smolin state [5] of four qubits ABAB . atedinthe firstandthe secondcrystalareindistinguish- ′ ′ Itis definedas anequally weightedstatisticalmixture of able apart from their polarization, the source produces four components, each corresponding to both the pairs a maximally entangled polarization state. An attractive prepared in the same Bell state: feature of this configuration is that owing to the axial symmetry of the type-I process, several entangled qubit pairs canbe collectedatonce fromdifferent locations on 1 ̺ˆ = φ φ φ φ the down-conversion cone. This provides a convenient Smolin 4 | +iABh +|⊗| +iA′B′h +| source of multiple photon pairs from a single set of crys- +(cid:0) φ φ φ φ tals without a need to redirect the pump beam. | −iABh −|⊗| −iA′B′h −| + ψ ψ ψ ψ We implemented the above idea in a setup shown | +iABh +|⊗| +iA′B′h +| + ψ ψ ψ ψ . (1) schematically in Fig. 1. The pump beam was a 78 MHz | −iABh −|⊗| −iA′B′h −| (cid:1) 3 We denoted here the Bell states as: The second example of noisy entanglement which we produced experimentally was a four qubit state 1 φ = HH VV | ±i √2 | i±| i 1 ψ = 1 (cid:0)HV VH (cid:1), (2) ̺ˆprivate = 4 |φ−iABhφ−|⊗|ψ−iA′B′hψ−| | ±i √2 | i±| i +(cid:0) ψ ψ φ φ (cid:0) (cid:1) | +iABh +|⊗| +iA′B′h +| where H and V stand for the horizontal and the verti- +|ψ+iABhψ+|⊗|ψ+iA′B′hψ+| icnavlaproialanrtizwaittihonrersepspecetcttivoealyn.yTpheermSmutoaltinionstaotfein̺ˆdSmivoildinuaisl +|ψ+iABhψ+|⊗|φ−iA′B′hφ−| . (5) qubits, which can be seen most easily from its represen- This state has nontrivial properties in the cont(cid:1)ext of tation in terms of Pauli matrices σˆµ: quantum key distribution and therefore we refer to it as the private state. Let us assume that Alice and Bob are 1 respectively in possession of qubits AA and BB . Sup- ̺ˆSmolin = 16 1ˆ1A⊗1ˆ1B ⊗1ˆ1A′ ⊗1ˆ1B′ posethattheymeasurequbitsAandB′inthe eig′enbasis of the σˆy operator composed of two vectors ¯0 and ¯1 | i | i defined as + σˆµ σˆµ σˆµ σˆµ , (3) µ=Xx,y,z A⊗ B ⊗ A′ ⊗ B′! υ¯ = 1 0 +i( 1)υ 1 , υ =0,1. (6) | i √2 | i − | i where we used the standard notation (cid:0) (cid:1) The reduced density matrix of the qubits A and B ex- σˆx = H V + V H | ih | | ih | pressed in this basis takes the form σˆy =i V H H V (4) | ih |−| ih | σˆz =|(cid:0)HihH|−|VihV|. (cid:1) TrA′B′(̺ˆprivate)= 12 |¯0¯0iABh¯0¯0|+|¯1¯1iABh¯1¯1| If we consider a partition of the four qubits into (cid:0)1 ¯0¯0 ¯1¯1 + ¯1¯1 ¯0¯0(cid:1) . (7) two subsystems, the first one comprising just one qubit AB AB − 4 | i h | | i h | and the second one remaining three, e.g. A:BAB , the ′ ′ (cid:0) (cid:1) Smolinstateexhibitsbipartiteentanglement. Indeed,the It is clearly seen that measurements performed by Al- Bell state measurement applied by the second party to ice and Bob on the qubits A and B in the σˆy eigenba- qubits AB allowsher to bring the qubits AB to a max- sis yield equiprobable and perfectly correlated results 0 ′ ′ imally entangled state without communication with the or 1. One may ask whether these results form a secure other party. On the other hand, the state is separable cryptographic key. If Alice and Bob had access only to with respect to any partition into two pairs of qubits, qubits A and B this would not be the case, as the mag- which follows immediately from the construction of the nitude ofthe off-diagonalelementsinEq.(7)is lessthan stateasastatisticalmixturespecifiedinEq.(1)andper- 12, which means that TrA′B′(̺ˆprivate) is not maximally mutational invariance. This implies that if each qubit is entangled. However, the presence of qubits A′ and B′ in hands of a separate party and all the parties are re- turns out to guarantee perfect security. As discussed in strictedto LOCC manipulations only, then it is impossi- [2], these two additional qubits serve as the shield sub- ble to distill any entanglement. Thus the Smolin state is systems preventing an eavesdropper from accessing any an example of bound entanglement [1]. information about measurement results. In this context, To generate the Smolin state in our experimental AandBareoftenreferredtoaskeysubsystems. Interest- setup, we started from two maximally entangled pairs ingly,thefour-qubitstate̺ˆprivatehasdistillableentangle- of photons AB and AB . With suitable settings of the mentstrictlylessthanone,whichmeansthatasymptotic ′ ′ Soleil-Babinet compensators, both the pairs were pre- conversionintoasmallernumberofmaximallyentangled paredinthesamestate φ . Thiscanbeconvertedinto singlet pairs is not the optimal way to generate a cryp- + | i theSmolinstatebyapplyingrandomlytoonequbitfrom tographic key from ̺ˆprivate. eachpair,chosento be B and B , equiprobabletransfor- The state ̺ˆ can be generated using our source ′ private mations1ˆ1B⊗1ˆ1B′, σˆBx ⊗σˆBx′, σˆBy ⊗σˆBy′,orσˆBz ⊗σˆBz′. Dif- of multiple photon pairs using an arrangement of wave ferently from Ref. [6], we realized these transformations platesshowninFig.1(d). ThephotonB issenttrougha bysendingthephotonsB andB throughthesamesetof half-wave plate which realizes either σˆx or σˆz, while the ′ three waveplates, as depicted in Fig. 1(c). Two quarter- photonB travelsthroughasetofthreewaveplatesthat ′ waveplatesareequivalenteithertotheidentitytransfor- can be set to implement identity or any Pauli operator. mation or a half-wave plate depending on whether their Applyingthesetransformationinacorrelatedmanneras faassutiatxabeslyaorerimenutteudahllaylfp-waraavleleplloarteperrepaelinzdesicσûxlaorr.σˆFzu.rtThheirs, pσˆrBziv⊗atσêByst′,atσêBx. I⊗ts1ˆ1gBe′n,eσˆraBxti⊗onσâBxn′d, aenxdpeσˆrBixm⊗enσˆtaBzl′ayniaelldyssisthoef allowedustoimplementallfourtransformationsrequired information-theoretic properties has been first reported to generate the Smolin state, since σˆy =iσˆxσˆz. in [10]. 4 IV. EXPERIMENTAL CHARACTERIZATION a Weperformedafulltomographicreconstructionofthe four-qubit state by sending individual photons to polar- 0.130 ization analyzers and measuring all 81 combinations of projectionsintheeigenbasesoftheoperatorsσˆx,σˆy,and 0.104 σˆz. For each combination, four-fold coincidences were recorded over an interval of approximately 1 hour, re- 0.078 sulting in the total time of an experimental run equal to about81hours. Theorientationofwaveplatesintroduc- 0.052 ingpolarizationnoisewaschangedat30sintervals. The counting circuit was put on hold during the operation 0.026 |VVVV> omapfraofImatnceseoltodyterourd1rre0sear,lrstttoeohertaeantcetaFhistntiPtgvtiGehmteAhpee.orbopwoceeaardvardeutiropwenlaatostofepstcrhoowelglhersiaeccttmhudmptao,etowdake.taoapIdrnpoerpctoothxeriidds- 0|H.0H0H0H> |VHHV> |HVVH> VV> |HHHH|V>HH|VH>VVH> V events composed of pairs of two-fold coincidences trig- |V gered between paths AB and AB , but not necessarily ′ ′ within the time window of the same pulse. Specifically, b after registering a two-fold coincidence for one combination of paths (either AB or AB ) the counting circuit waitedforatwo-foldcoincidence′be′tweendetectorsmon- 0.140 itoring the other combination, and recording the result as a four-fold event. Additional two-fold coincidences 0.112 involving the first combination of paths that occurred duringthewaitingwindowwereignored. Thisprocedure 0.084 allowed us to collect four-fold events at a rate that was approximately half the rate of producing single photon 0.056 pairs, reducing the overallmeasurement time to approximately 3 hours. In this case, the limiting factor was the 0.028 |VVVV> suersptursatineInitnee.egdsFatoonihgffde.ttth2hfhaeeeswtgmpeerponiprvterooarecrtasieetezdenesdudttaresdtrteoeawntrtaeseetirctieobyonneusmfsotserratdueatcrgaittecoesedf.suvflefrDloroieramfxyttaphttehehcreoieSmlelmpxeecrpnootetlepiarnd--l 0|H.0H0H0H> |VHHV> |HVVH> |VVVV> |HHHH|V>HH|VH>VVH> imentaldatausingthe maximumlikelihoodmethod [17– 19]. Theseexperimentalresultscanbecomparedagainst Figure 2. Absolute values of the elements of reconstructed theidealizedstatesbycalculatingthecorresondingfideli- density matrices for (a) the Smolin state and (b) the private ties, defined in general for two states ̺ând ̺ˆ′ as state. F(̺ˆ,̺ˆ′)=Tr √̺̺′√̺ . (8) tion value of an entanglement witness given by [6] (cid:18)q (cid:19) We obtained values F = 0.923 0.002 for the Smolin Wˆ =1ˆ1⊗4+(σˆx)⊗4+(σˆy)⊗4+(σˆz)⊗4 (9) ± state and F = 0.971 0.001 for the private state. The experimental Smolin s±tate fidelity is comparable to that whichwasfoundtobe equalto Wˆ = 1.43 0.02,ver- h i − ± obtained in two other experiments aiming at generating ifying the nonclassical character of the generated state. ̺ˆSmolin inafour-photonsystem[6,9]whichsuggeststhat Wealsocalculatedeigenvaluesofthepartiallytransposed itmaybeatthelimitofwhatcanbeachievedinatypical density matrix with respect to three possible partitions realization with standard optical elements. The higher intotwopairsofqubits. TheresultsarepresentedinTa- fidelityfortheprivatestatemaybeattributedtothefact ble I. Itisseenthatforeachpartitionsomeofthe eigen- that in this case the polarization noise was introduced values are negative. A similar feature was also present withfewerbirefringentelements,thusreducingtheeffects in the first experimental generation of the Smolin state of their imperfections. reported in [6]. The reason behind the occurrence of For the Smolin state, we used polarization measure- negative eigenvalues is that the theoretical Smolin state ments on individual photons to determine the expecta- ̺ˆ is located exactly on the boundary of positive Smolin 5 locality requirement) to a form in which the key subsys- TABLEI.Eigenvaluesofthepartiallytransposeddensityma- tems exhibit correlations violating local realism. In our trixcharacterizingtheexperimentallygeneratedSmolinstate, case, it is easy to see that for the ideal state ̺ˆ de- obtainedforthreepossiblepartitionsintotwopairsofqubits. private fined in Eq. (5) no local operations are necessary. This is because the reduced density matrix of the qubits AB, Theory AB:A′B′ AB′:A′B AA′:BB′ given explicitly by 0.250 0.229 0.228 0.229 0.250 0.216 0.216 0.217 1 3 0.250 0.214 0.215 0.213 TrA′B′(̺ˆprivate)= 4|φ−iABhφ−|+ 4|ψ+iABhψ+|, (10) 0.250 0.202 0.202 0.204 is a statistical mixture of two Bell states with unequal 0.000 0.036 0.034 0.034 weights. With a suitable choice of projective measure- 0.000 0.026 0.029 0.029 ments, any such a mixture violates the Clauser-Horne- 0.000 0.024 0.025 0.023 Shimony-Holt(CHSH)inequality,whichfollowsfromthe 0.000 0.022 0.023 0.022 set of necessary and sufficient conditions derived in [20]. 0.000 0.016 0.016 0.015 In order to test the CHSH inequality we performed 0.000 0.011 0.011 0.012 polarizationmeasurements on the key photons in coinci- 0.000 0.008 0.009 0.009 dencewithdetectorsmonitoringtheshieldphotons. The 0.000 −0.005 −0.007 −0.006 CHSH inequality can be written as 0.000 −0.003 0.005 0.004 0.000 0.003 −0.004 −0.003 2 2, (11) − ≤B ≤ 0.000 0.001 −0.002 −0.002 0.000 0.000 0.001 0.001 where =C(a;b)+C(a;b)+C(a;b ) C(a;b) (12) ′ ′ ′ ′ B − partial transposition states. Non-ideal implementation is a combination of four correlation functions of the polarization noise, imperfect alignment of the polarization analyzers, and the statistical uncertainty of C(a;b)= (a σˆ) (b σˆ) (13) · ⊗ · the measured density matrix may therefore easily produce residualentanglementinthe reconstructeddatafor forBlochvectorsa,a,b,(cid:10)andb thatdefin(cid:11)ethemeasure- ′ ′ any partition. This problem canbe solvedby generating ment bases. For the specific state given in Eq. (10), the a mixture of ̺ˆSmolin and a completely mixed four-qubit maximum violation of the CHSH inequality is obtained state [8, 9] which for a suitable choice of relativeweights for the choice of vectors: demonstratesthephenomenonofboundentanglementin an experimentally robust way. 0 0 0 0 tieEsxopfetrhimeesntatatel c̺ˆharactehraizsabtieoennodfestchreibperdiviancyRepfr.o[p1e0r]-. a=1, a′ =0, b= √25 , b′ =√25, private The reconstruction of information-theoretic quantities 0 1 1 1     −√5 √5 from experimental data was found to be very sensitive        (14) to statistical uncertainties due to highly non-linear de- resulting in the value of the CHSH combination equal pendence on the elements of the density matrix. The to = √5 2.236. Polarization measurements per- B ≈ statisticaldistributionsforthequantitiesofinterestwere formed on the key photons in these bases yielded the obtainedbyevaluatingthemonindividualdensitymatri- result = 2.12 0.01, which is clearly above the limit B ± cesthatformedanensembleconsistentwithexperimental permitted by local realistic theories. It is interesting to data. The resultsdemonstratedastatisticallysignificant note that the reduced density matrix of the key qubits separation between the distillable entanglement and the AB has been found in Ref. [10] to contain no distillable key contents for the generated state. key due to experimental imperfections. Thus the viola- Private states exhibit other specifically nonclassical tion of the CHSH inequality turns out to be a more ro- properties. In particular, Augusiak et al. [11] have re- bust way to detect quantum correlations contained this cently presented a general theoretical proof that perfect two-qubitstate. Thisisunderstandable,sincethe CHSH privacy implies incompatibility with local realistic theo- combinationcanbewrittenasasinglequantummechan- ries. This motivated us to test whether the experimen- ical observable, while the calculation of the key contents tally generated state, despite its non-ideal privacy, can requireshighly nonlinearprocessingof the reconstructed be used to demonstrate a violation of Bell’s inequalities. density matrix. This problem may be alleviated by the The proof presented in Ref. [11] is based on an observa- developmentofmoreefficientmethodstocharacterizethe tion that any private state can be brought by local op- amount of distillable key based on a single or a few ob- erations (without classical communication to satisfy the servables, such as those recently presented in [21]. 6 izations of individual photons in suitably selected bases. 160 Ifmeasurementsarerestrictedtothisclass,theproduced -1 s)]140 pdhomot,oen.gp.afriresqucaenncby,epcroorvriedleadtetdhainttohtehmeroddeaglrsetersucotfufrreeoe-f 0 0120 thepairisindependentofthepolarizationstate[22]. For 5 nts [(100 tdhiteiocnonisfigsautriastfiieodnobvaesredreolantitvweolytlyapreg-eIbcarynsdtwalisd,ththsisofctohne- u co 80 generated photons, which follows from the symmetry of s n the type-I down-conversionprocess [23]. This allows one o ot 60 to avoid heavy spectral filtering and consequently offers h p increased four-fold coincidence rates, which are notori- 4- 40 ously low in most multiphoton experiments. However, 20 if independently generated photons are to be interfered, it is necessary to ensure their spectral indistinguishabil- 0 -12 -10 -8 -6 -4 -2 0 2 4 6 8 ity. This requirement can be verified with the help of 2 the Hong-Ou-Mandeltwo-photoninterferenceeffect [24]. Delay [(cid:215)10 fs] Wecarriedoutapreliminarytestbyproducingtwopairs Figure3. TheHong-Ou-Mandelinterferencedipbetweenher- in a single crystal and interfering photons from differ- aldedphotonsfromtwoindependentpairsgeneratedinasin- ent pairs in an event-ready manner [25] using a single- gle beta-barium borate crystal. The interfered photons were mode fiber optic directional coupler with a 50:50 split- transmitted through 2nm interference filters. The delay was tingratio. Whentheinterferedphotonsweretransmitted introducedbytranslatingoneofthecollimatorscouplingpho- through 2 nm bandwidth interference filters, 79% depth tonsintoasingle-modefiber. Theexperimentaldata(squares of the Hong-Ou-Mandel dip was observed, as shown in witherrorbars)arefittedwithaGaussianprofile(solidline). Fig. 3. This allows for some optimism about using the described source in more sophisticated experiments uti- lizing multiphoton interference effects. V. CONCLUSIONS AND OUTLOOK In conclusion, we presented an arrangement to collect ACKNOWLEDGMENTS photon pairs in maximally entangled polarization states from a single set of two type-I down-conversioncrystals. We wish to acknowledge insightful discussions with Applicationofcorrelatednoiseintroducedusingrotating Czesl aw Radzewicz and Wojciech Wasilewski. This re- wave plates enabled us to produce noisy entangled four- search was supported by FP7 FET projects CORNER photon states that illustrated fundamental results of the (contract no. 213681) and Q-ESSENCE (contract no. entanglement theory. 248095), and the Foundation for Polish Science TEAM Characterizationofinformation-theoreticpropertiesof project cofinanced by the EU European Regional Devel- the generated states was performed by measuring polar- opment Fund. [1] M. Horodecki, P. Horodecki, and R. Horodecki, Phys. 030501 (2011) Rev.Lett. 80, 5239 (1998). [11] R. Augusiak, D. Cavalcanti, G. Prettico, and A. Ac´ın, [2] K.Horodecki,M.Horodecki,P.Horodecki,J.Oppenheim Phys. Rev.Lett. 104, 230401 (2010). Phys.Rev.Lett. 94, 160502 (2005). [12] J. T. Barreiro, P. Schindler, O. Gu¨hne, T. Monz, M. [3] P. G. Kwiat, E. Waks, A. G. White, I. Appelbaum, and Chwalla, C.F.Roos,M.Hennrich,andR.Blatt,Nature P.H. Eberhard, Phys.Rev. A 60, R773 (1999). Physics 6, 943 (2010). [4] A. Rossi, G. Vallone, A. Chiuri, F. De Martini, and P. [13] P.G. Kwiat, K. Mattle, H. Weinfurter, and A. Zeilinger, Mataloni, Phys. Rev.Lett. 102, 153902 (2009). Phys. Rev.Lett. 75, 4337 (1995). [5] J.A. Smolin, Phys. Rev.A 63, 032306 (2001). [14] D.Bouwmeester, J.-W.Pan,K.Mattle, M.Eibl,H.We- [6] E. Amselem and M. Bourennane, Nature Physics 5, 748 infurter, and A.Zeilinger, Nature390, 575 (1997). (2009). [15] J.-W. Pan, M. Daniell, S. Gasparoni, G. Weihs, A. [7] J. Lavoie, R. Kaltenbaek, M. Piani, and K. J. Resch, Zeilinger Phys. Rev.Lett. 86, 4435 (2001). NaturePhysics 6, 827 (2010). [16] C.-Y. Lu, X.-Q. Zhou, O. Gu¨hne, W.-B. Gao, J. Zhang, [8] H.Kampermann,D.Bruß,X.Peng,andD.Suter,Phys. Z.-S. Yuan, A. Goebel, T. Yang, and J.-W. Pan, Nature Rev.A 81, 040304(R) (2010). Physics 3, 91 (2007). [9] J. Lavoie, R. Kaltenbaek, M. Piani, and K. J. Resch, [17] Z. Hradil, Phys.Rev.A 55, R1561 (1997). Phys.Rev.Lett. 105, 130501 (2010). [18] K. Banaszek, G. M. D’Ariano, M. G. A. Paris, and M. [10] K. Dobek, M. Karpin´ski, R. Demkowicz-Dobrzan´ski, F. Sacchi, Phys.Rev.A 61, 010304 (1999). K. Banaszek, and P. Horodecki, Phys. Rev. Lett. 106, 7 [19] D. F. V. James, P. G. Kwiat, W. J. Munro, and A. G. [22] A.B. U’Ren, K. Banaszek, and I.A. Walmsley, Quant. White, Phys.Rev.A 64, 052312 (2001). Inf. Comput. 3, 480 (2003). [20] R. Horodecki, P. Horodecki, and M. Horodecki, Phys. [23] A. Dragan, Phys. Rev.A 70, 053814 (2004). Lett.A 200, 340 (1995). [24] C.K. Hong, Z.Y. Ou, L. Mandel, Phys. Rev. Lett. 59, [21] K. Banaszek, K. Horodecki, and P. Horodecki, preprint 2044 (1987). arXiv:1109.2486. [25] M.Z˙ukowski,A.Zeilinger,M.A.Horne,andA.K.Ekert Phys. Rev.Lett. 71, 4287 (1993)