Entangled photon sources IV: PAPA

Our team is working actively on entanglement distribution via free-space or optical fiber. Although the wavelengths can be different - for fiber, we prefer something in the telecommunications band (1,550 nm or 1,310 nm) whereas for free-space we currently prefer the Near Infrared (about 800 nm), the design principles for entangled photon pair sources based on Spontaneous Parametric Downconversion (SPDC) are very similar. We also prefer to work in polarization, and to build sources around bulk crystals, because there are many degrees of freedom that can be utilised in a bulk material (see the POMO concept below).

When designing instruments, it certainly helps to have an envelope for Size, Weight and Power (SWAP), and we use this to determine which design choice to optimize. As we like to reduce the footprint of the instrument, we often choose to work on co-linear designs, where all three SPDC fields co-propagate in the same direction.

PArallel PAth entangled photon pair sources - PAPA. See our pre-print.

One of the ways to generate entangled photon pairs is to overlap SPDC processes coherently. There are many degrees of freedom, one could use, but one of the most elegant is to use a Mach-Zehnder interferometer, where two paths co-propagate in a single crystal. This design is also called a linear interferometer, but we prefer to use the acronym PAPA to highlight the fact that it uses parallel pump beam paths. This design has been used successfully in a number of experiments, and one of the seminal uses is in loop-hole Bell tests.

However, most of the previous work were locked into using narrow line-width pump lasers, which were limited in power, or else very costly. As a team that is interested in engineering more cost-effective devices, we had to go back to the drawing board.

The main reason why narrow line-width pump lasers are required is that the pump bandwidth can cause distinguishability in the wavefunction of the SPDC photon-pairs, so that one cannot obtain perfectly coherent superposition. Look at Figure 1(a) below. Consider that the pump must first undergo spatial walk-off in a BBO crystal, then one arm must propagate past a wave-plate and finally proceed through the PPKTP nonlinear crystal where SPDC takes place. A broadband pump will pick up distinguishing phase information by the time it reaches the PPKTP crystal.

To overcome, this we need to "pre-compensate" the pump bandwidth. Until recently, we were unable to do this due to lack of precision in available Sellmeier equations for the various bits of nonlinear crystals. In 2019, we successfully managed to track down the correct Sellmeier equations, and implemented correctly the pre-compensation.

Fig. 1

Fig. 1. (a) Schematic of the main source components. The diagonally polarized pump is focused into the center of the PPKTP crystal with a waist size of ωp = 100 μm and a separation between the two paths of 1 mm. (b) Schematic of the splitting and detection setup. The photon-pair is split by its momentum components on the wedge mirror and the polarizers, Ps and Pi can be inserted to analyze the polarization state. Finally, the beam is coupled into single mode fibers and detected using Geiger mode avalanche photo diodes. The resulting SPDC and polarization quality can be observed in Figure 2.

Fig. 2

Fig. 2. (a) SPDC spectrum measured for the broadband pumped source. A typical emission spectrum of the pump is shown in the inset.Uncorrected pair rate as a function of polarizer angle insignal arm for four fixed polarizer settings in the idler arm. The input power was approximately 10 μW.

Despite the broad range of the SPDC photon pairs, over 100 nm, we get very high entanglement correlation, corresponding to 97% visibility. The photon pair rate was observed to be 300,000 pairs/s/mW. The pump power available can be up to 150 mW, potentially producing 45 million pairs in a single spatial mode. Furthermore, if we take into account that we can improve the detection efficiency by using superconducting nanowire detectors, we can anticipate a rate that is in the order of 1E9 pairs per second. This could enable double downlink configurations for entanglement distribution from geo-stationary satellites.

Now, this might be a problem if all photons were sent only to a set of detectors - the probability of multiple-pair events within the detection time window would be very high. However, the broadband nature of the SPDC is of assistance here. One can imagine that there are dense wavelength division multiplexers that could separate the signal/idler photons into small bands, and the detection system need only compare a few channels at each time. This reduces the probability of accidental correlations.