All-optical-fiber polarization-based quantum logic gate

We report an experimental demonstration of a quantum controlled-NOT gate constructed entirely in optical fiber and operating on polarization-encoded single-photon qubits. We operated this gate using two heralded optical fiber single-photon sources and found an average logical fidelity of 90% and an average process fidelity of $0.83\le \overline{F}\le 0.91$. On the basis of a simple model we are able to conclude that imperfections are primarily due to the photon sources, meaning that the gate itself works with very high fidelity. Such all fiber quantum information processing will likely have important applications in future quantum networks.

Figure 1

(Color online) A two-photon quantum CNOT gate. (a) The setup involves conversion from polarization encoding to spatial mode encoding via the use of polarizing beamsplitters (PBSs) in the control c and target t optical modes, and half wave plates (HWPs) to make all polarizations the same, followed by a nonclassical interference at the central beamsplitter (BS) with a reflectivity equal to 1/3. The other two 1/3 beamsplitters serve to adjust the output amplitudes while 1/2 beamsplitters perform Hadamard operations on the target mode. The dotted lines indicates a phase shift on reflection [15]. The beams are then recombined at polarizing beamsplitters, creating two interferometers. Success of the gate is conditional upon detecting a photon at each of the target and control outputs, which occurs with probability of 1/9. (b) The PPBS simplification of the gate.

Figure 2

(Color online) An all-optical fiber quantum CNOT gate operated with heralded photonic crystal fiber single-photon sources. A ps 708 nm Ti:sapphire laser pumps two PCFs creating a nondegenerate pair of photons at 583 and 900 nm. These are separated at dichroic mirrors (DMs) and pass through interference filters F1 and F2 with bandwidths of 0.2 and 0.8 nm, respectively. Detection of the 583 nm photons heralds the arrival of the 900 nm control c and target t photons at the inputs to the CNOT gate. HWPs are used to create logical and diagonal input states. The gate consists of three PPFCs with the reflectivities for horizontal and vertical photons as shown. The polarizations of the idler photons are then analyzed using a HWP and a PBS cube for each arm. Note that the Hadamard operations before and after the gate are integrated into the encoding and analysis wave plates. Also inserted here are quarter-wave plates (QWPs) which may be tilted to correct for any phase accumulated through the gate. All four photons are then detected using Perkin Elmer silicon avalanche photodiodes (APDs) and sent to an electronic fourfold coincidence counting circuit for analysis.

Figure 3

(Color online) Truth table results for the gate operating in two orthogonal bases: (a) the computational $\left(ZZ\right)$ basis and (b) the diagonal $\left(XX\right)$ basis. We show (i) the ideal operation, (ii) the experimental data, and (iii) the results from the model.

Figure 4

A model for CNOT gate operation. The additional modes (dashed lines) are used to model mode mismatch at the central beamsplitter. The beamsplitters are given reflectivities ${\eta }_i$. The dotted-line side of the beamsplitters indicates the side at which a $\pi$ phase shift occurs upon the incident photon.