Nonclassical Interference and Entanglement Generation Using a Photonic Crystal Fiber Pair Photon Source

We demonstrate two key components for optical quantum information processing: a bright source of heralded single photons; and a bright source of entangled photon pairs. A pair of pump photons produces a correlated pair of photons at widely spaced wavelengths (583 nm and 900 nm), via a ${\chi }^{(3)}$ four-wave mixing process. We demonstrate nonclassical interference between heralded photons from independent sources with a visibility of 95% (after correction for background), and an entangled photon pair source, with a fidelity of 89% with a Bell state.

Figure 1

Interference experiment involving two heralded single photons from separate sources. (a) A picosecond Ti:sapphire laser emitting 1.5 ps pulses with a repetition rate of 82 Mhz is split in a $50:50$ beam splitter and then pumps two fibers (PCF 1 and 2, each 12 cm long) with variable delay $\delta t$. The pair photon outputs are collimated using aspheric lenses and then separated into signal and idler arms using dichroic mirrors (DM) centered at 700 nm. The signal beams are then launched into single mode fibers to guarantee optimal spatial mode overlap on the $50:50$ fiber beam splitter. A polarization controller (PC) and 0.28 nm FWHM bandpass filter (583 nm center wavelength) further ensured the indistinguishability of the signal photons. A slightly wider (2 nm FWHM) filter is used on the long wavelength (900 nm) idler beams mainly to minimize losses. These beams are also collected in single mode fiber and idler detections are used to herald signal detections in a fourfold coincidence circuit. (b) Fourfold coincidences measured as a function of delay time between the heralded photons $\delta t$ clearly showing the expected dip around zero delay with a width corresponding to the coherence length of the filtered photons ($l_c\approx 1.2\mathrm{mm}$ for 0.28 nm bandwidth at 583 nm). Using 1 mW of pump power per fiber, we obtain 1700 pair coincidences/s per fiber and 0.04 fourfold coincidences/s. (c) Increasing the pump power to 8 mW per fiber we see a net 80 fourfold coincidences per second and a high visibility dip after correction for background. Note that in both experiments there is no contribution from any other process such as Raman scattering or fluorescence.

Figure 2

(a) Schematic of the entangled photon pair source. The same pump laser as in the previous interference experiment is used in this setup. The laser pulse polarization is here rotated to 45° in a half-wave plate. The pump pulses are then split in the PBS and launched into each end of the PCF. The fiber is twisted so that both pump directions are polarized along the same axis and identical phase matching conditions hold. The photon pairs thus exit from the PBS in the same port as the pumping beam and a notch mirror is used to separate them from the input pump beam. The entangled photons are launched into single mode fiber and narrow-band filtered (0.28 and 2 nm in the signal and idler arms, respectively) to ensure indistinguishability between the $|HH⟩$ and $|VV⟩$ components and the small static phase shift between them is compensated in the pump beam by adjusting the Soleil-Babinet compensator (S-B). We use a quarter- and half-wave plate followed by a PBS to analyze the various polarization correlations needed to reconstruct the state [26]. (b) Tomographic reconstruction of the real and imaginary parts of the density matrix of the generated state. This clearly shows that what we produce is very close to the maximally entangled state $|HH⟩+|VV⟩$.