We present the experimental observation of polarization entanglement for three spatially separated photons. Such states of more than two entangled particles, known as GHZ states, play a crucial role in fundamental tests of quantum mechanics versus local realism and in many quantum information and quantum computation schemes. Our experimental arrangement is such that we start with two pairs of entangled photons and register one photon in a way that any information as to which pair it belongs to is erased. The registered events at the detectors for the remaining three photons then exhibit the desired GHZ correlations.

Bell's theorem states that certain statistical correlations predicted by quantum physics for measurements on two-particle systems cannot be understood within a realistic picture based on local properties of each individual particle-even if the two particles are separated by large distances. Einstein, Podolsky and Rosen first recognized the fundamental significance of these quantum correlations (termed 'entanglement' by Schrodinger) and the two-particle quantum predictions have found ever-increasing experimental support. A more striking conflict between quantum mechanical and local realistic predictions (for perfect correlations) has been discovered; but experimental verification has been difficult, as it requires entanglement between at least three particles. Here we report experimental confirmation of this conflict, using our recently developed method to observe three-photon entanglement, or 'Greenberger-Horne-Zeilinger' (GHZ) states. The results of three specific experiments, involving measurements of polarization correlations between three photons, lead to predictions for a fourth experiment; quantum physical predictions are mutually contradictory with expectations based on local realism. We find the results of the fourth experiment to be in agreement with the quantum prediction and in striking conflict with local realism.

We experimentally demonstrate observation of highly pure four-photon GHZ entanglement produced by parametric down-conversion and a projective measurement. At the same time this also demonstrates teleportation of entanglement with very high purity. Not only does the achieved high visibility enable various novel tests of quantum nonlocality, it also opens the possibility to experimentally investigate various quantum computation and communication schemes with linear optics. Our technique can in principle be used to produce entanglement of arbitrarily high order or, equivalently, teleportation and entanglement swapping over multiple stages.Entanglement is not only the essence of quantum mechanics as suggested by Erwin Schrödinger [1], but is also at the basis of nearly all quantum information protocols such as quantum cryptography, quantum teleportation and quantum computation [2]. While entanglement of two qubits is routine in the laboratory, entanglement of three photons [3] with high quality has only recently been experimentally realized [4] and used to experimentally demonstrate the extreme contradiction between local realism and quantum mechanics [5] in so-called GHZ states. In a parallel development entanglement of the quantum states of three atoms [6] or four qubits in ions [7] has been demonstrated, yet in all these cases the quality of the entangled states still needs to be significantly improved in order to be useful for tests of quantum mechanics or in quantum information schemes.A similar situation is found in the recent teleportation experiments [8][9][10][11]. To verify the nonlocal character of teleportation, two conditions must be satisfied in any experiment. On the one hand, one has to demonstrate that a genuinely unknown state (in the optimal case, a qubit which itself is still entangled to another one) is teleported [12], on the other hand a high experimental visibility is necessary in order to exclude local hidden variable models (LHV) [13][14][15][16]. The so-called entanglement swapping experiment [10] is the only one to date that demonstrates the teleportation of a genuinely unknown state. However, since its observed visibility was lower than 71%, one could in principle still doubt the nonlocal feature of teleportation [13].In this letter we report on an experiment that not only demonstrates the observation of four-photon entanglement but also shows high-fidelity entanglement swapping, thus proving the nonlocal character of quantum teleportation. Both features are not only important for performing novel fundamental experiments to test quantum mechanics or to demonstrate its counter-intuitive features, but also to expand our toolbox for quantum computation and quantum communication.Our technique of observing four-photon GHZ entanglement uses two independently created photon pairs (Fig. 1). Suppose that the two pairs are in the statewhich is a tensor product of two polarization entangled photon pairs. Here |H (|V ) indicates the state of a horizontally (vertically) polarized photon. Principle fo...

Braunstein and Kimble observe correctly that, in the Innsbruck experiment, one does not always observe a teleported photon conditioned on a coincidence recording at the Bellstate analyser. In their opinion, this affects the fidelity of the experiment, but we believe, in contrast, that it has no significance, and that when a teleported photon appears, it has all the properties required by the teleportation protocol. These properties can never be achieved by "abandoning teleportation altogether and transmitting randomly selected polarization states" as Braunstein and Kimble suggest. The fact that there will be events where no teleported photons are created merely effects the efficiency of the experiment. This suggests that the measure of fidelity used by Braunstein and Kimble is unsuitable for our experiment During the detection of the teleported photons, no selection was performed based on the properties of these photons. Therefore, no a posteriori measurement in the usual sense as a selective measurement was performed. The detection of the teleported photon could have been avoided altogether if we had used a more expensive detector, p, that could distinguish between one-and two-photon absorption. The inability of our teleportation experiment to perform such refined detections does not, however, imply that "a teleported state can never emerge as a freely propagating state...". Braunstein and Kimble do not, therefore, reveal a principle flaw in our teleportation procedure, but merely address a non-trivial practical question.

No abstract