Figure 1

A comparison between previous setups and our setup. (a) The previously used setup for entangling four and six photons in a GHZ state. Three $\beta \mathrm{\text{−}}{\mathrm{BaB}}_2{\mathrm{O}}_4$ (BBO) crystals generate from a single pump pulse three entangled photon pairs in six spatial modes. Delay lines synchronize the projection of a photon from each pair onto a PBS. Six analyzing wave plates (WP), six PBSs, and twelve detectors are required to measure the state. To produce eight-photon entanglement, this setup needs additional components. (b) Our resource efficient setup. Only a single crystal generates pairs from many pump pulses. The pairs are projected onto a large entangled state on a single PBS and occupy two spatial modes and additional temporal modes. The same setup is applicable to entanglement generation of any photon number. (c) The two multiphoton entangled graph states that are possible to obtain without fast polarization rotations: a growing star shaped GHZ state and a connected branched chain (which for three pairs is an $H$-graph state). For two-photon pairs, both states are identical. Numbered circles mark photons and their creation order, where dimmed circles represent possible future states of four and five fused pairs. Connecting lines mark the entangling operations that define the quantum graph state.Reuse & Permissions

Figure 2

Measured amplitude histogram of the generated four-photon GHZ state. Two opposite amplitudes occur on average 65 times more often than any other amplitude. Data were accumulated over 2000 sec. We used photon pairs in the $|{\psi }^{+}⟩$ Bell state. The results are equivalent to using a $|{\varphi }^{+}⟩$ state, up to local polarization rotations.Reuse & Permissions

Figure 3

Coherence of the GHZ state. The sum of all even amplitudes (black squares) and the sum of all odd amplitudes (red circles) at a rotated polarization basis vs the relative delay between the short and the long paths. At zero delay, the even amplitudes interfere destructively, and the odd amplitudes interfere constructively. When delay is introduced, the two GHZ amplitudes are temporally distinguishable and interference is lost. The interference visibility is $69.5\pm 0.8\%$. Data were accumulated over 480 sec per point. Errors are calculated assuming Poisson statistics.Reuse & Permissions

Figure 4

Fourfold coincidence counts of the 16 individual amplitudes of four photons at the plus-minus 45° (PM) linear polarization basis. At zero delay, counts from all amplitudes where P appears an odd number of times (light green background) increase while counts from all even amplitudes (light orange background) decrease. Figure 3 is composed of these scans.Reuse & Permissions