1. Introduction

It is expected that in future quantum communication applications such as multi-party quantum cryptography [1] and distributed quantum computing [2], distantly located users would need to share and consume quantum entanglement as a resource [3]. However, as the total amount of entanglement shared among users cannot be created nor increased with local operations and classical communications (LOCC) [4], entanglement sharing must definitely involve some means of entanglement distribution over a physical channel. This implies that in addition to a classical communication channel, an entanglement distribution channel must also be established among users. Today, distributing entanglement over very long distances is still beyond reach due to the difficulty of implementing quantum repeaters [5, 6]. However, a local-area entanglement distribution fiber network covering a distance of the order of 100 km seems realizable in the near future. We envision that in this network, a centrally located entanglement distributor playing the role of a service provider, would create highly-entangled photon-pairs and distribute them via fiber-optic transmission lines to users within the network according to users’ demand [7].

There has already been a number of proposals on telecom-band entangled photon-pair sources based on spontaneous parametric down-conversion (SPDC) or spontaneous four-wave mixing (SFWM) [8–16], and recently, a few groups have reported entanglement distribution over 100 km of optical fiber, using either time-bin-entangled or polarization-entangled photon-pairs [17–20]. From a telecommunication engineering point of view, it would be desirable if the available transmission bandwidth of the fiber-optic transmission lines can be fully utilized for entanglement distribution. However, most of the entangled photon-pair sources studied so far have exhibited a relatively narrow bandwidth as compared to the available transmission bandwidth of optical fiber. For fiber-based entangled photon-pair sources, the bandwidth is seriously limited by spontaneous Raman scattering [21, 22], while sources based on quasi-phase-matched (QPM) crystals or waveguides are usually narrowband because of long device lengths and/or operations far from the degeneracy wavelength [15, 16].

 

Fig. 1. (color online) Concept of multi-channel wavelength-multiplexed entanglement distribution. An entanglement distributor playing the role of a service provider uses a single broadband source of entangled photon-pairs to distribute entanglement to application users. AWG: arrayed waveguide grating

Download Full Size | PDF

With narrowband sources, a service provider would have to wavelength-multiplex a large number of them before entanglement distribution in order to fully utilize the available transmission bandwidth, and this is not cost-effective. The concept of a single broadband source suitable for wavelength-multiplexed entanglement distribution has not been pursued so far. It is obvious that this broadband source must produce highly-entangled photon-pairs for all wavelength channels simultaneously, and we emphasize here that this condition is not necessarily satisfied for existing sources. For wavelength-demultiplexing of distributed photons into independent wavelength channels, arrayed waveguide gratings (AWGs) can be used, as in classical wavelength-division-multiplexed (WDM) systems [23]. Figure 1 shows the concept of wavelength-multiplexed entanglement distribution using just one single broadband source.

In this work, we demonstrate experimentally for the first time a telecom-band polarization-entangled photon-pair source that is well-suited for wavelength-multiplexed entanglement distribution. Our source is specifically based on a very short, type-0, MgO-doped periodically-poled lithium niobate (PPLN) waveguide operating near degeneracy, and so it has a very broad SPDC bandwidth. We show that our source is potentially capable of simultaneously supporting up to forty-four independent wavelength channels.

This paper is organized as follows. In Section 2, we describe the experiment. In Section 3, we present experimental results followed by a discussion. Section 4 concludes this paper.

2. Experiment

Figure 2 shows the experimental setup. The source is based on a 1-mm-long, type-0, MgO-doped PPLN waveguide (HC Photonics) placed at the center of a polarization-diversity loop without any form of temperature control [14]. For our application, we would like all photons to be produced in the telecom-band, and so we have selected a PPLN waveguide that has a degeneracy wavelength at 1550 nm. Photon-pair production near the degeneracy wavelength gives a broad SPDC bandwidth. Furthermore, the short length of the selected waveguide ensures a very broad phase-matching bandwidth for the SPDC process at room temperature.

The pump laser incident at the polarization beam-splitter (PBS) is diagonally-polarized so that it is split into two orthogonally polarized components by the PBS. The polarization-maintaining fiber (PMF) forming a loop has a 90-degree-twist such that the same polarization-mode of the PPLN waveguide is pumped bi-directionally. The photon-pairs that are produced within the waveguide then combine at the same PBS resulting in a polarization-entangled state, denoted by |H〉_s|H〉_i+e ^iθ|V〉_s|V〉_i, where the subscripts s and i denote signal and idler, respectively. θ is an unknown but constant phase, which we compensate using a half-wave plate (HWP) at the receiver side in order to obtain the maximally-entangled state |Φ⁺≡|H〉_s|H〉_i+|V〉_s|V〉_i. The filter at the exit of the PBS is a piece of mirror that is coated anti-reflective (AR) at the pump wavelength (775 nm) and highly reflective (HR) at the telecom-band. It should be emphasized that the PBS is a custom-made device specifically designed to operate at both the pump wavelength and the entire telecom-band. The dichroic mirror (DM) is optically coated for high transmittivity from 1500 nm to 1550 nm and high reflectivity from 1550 nm to 1585 nm, in order to separate the signal and idler photons, as well as to enable broadband output from the source. The quarter-wave plates (QWPs) and polarizers (POLs) placed before two InGaAs single-photon counter modules (SPCMs, id Quantique) are used to realize the sixteen polarization-settings needed for quantum state tomography. The HWP at the signal channel compensates the unknown phase, as already mentioned. Density matrices are reconstructed from measured number of coincidence counts in 100 seconds (with accidental coincidence counts included) using the maximum likelihood method described in [24]. Further details of the experimental setup can be found in [14].

 

Fig. 2. (color online) Schematic of the experiment. Red arrows show input pump laser. Black arrows show telecom-band photon-pairs created in the PPLN waveguide. ATT: attenuator, BPF: band-pass filter, CW: continuous-wave, DM: dichroic mirror, HWP: half-wave plate, PBS: polarization beam-splitter, PC: polarization controller, PMF: polarization-maintaining fiber, POL: polarizer, PPLN: periodically-poled lithium niobate waveguide, QWP: quarter-wave plate, SMF: single-mode fiber, SPCM: single-photon counter module

Download Full Size | PDF

 

Fig. 3. (color online) Each individual channel consists of a signal channel and a corresponding idler channel that are entangled in polarization. Signal and idler wavelengths shown are determined for a pump wavelength at 776 nm.

Download Full Size | PDF

In order for the source to be suitable as a wavelength-multiplexed source of polarization-entangled photon-pairs, two-photon interference fringe visibility and entanglement fidelity of the generated photon-pairs must be sufficiently high across the wavelength range of interest. Using two band-pass filters (BPFs) having 60 GHz pass-band (equivalent to 0.5 nm bandwidth at 1580 nm) and tunable from 1525 to 1580 nm, we can effectively wavelength-demultiplex the generated photon-pairs into forty-four independent channels, and measure the two-photon interference fringe visibility and entanglement fidelity for each individual channel. As illustrated in Fig. 3, each individual channel consists of a signal channel and an idler channel whose wavelengths are related by the energy conservation law, λ^-1 _pump=λ^-1 _j,signal+λ-1 _j,idler, where λ_j,signal and λ_j,idler λ denote the signal and idler wavelength, respectively, for Channel j, and λ_pump denotes the pump wavelength. For convenience, we have fixed Channel 1’s signal wavelength to 1525.0 nm. It is then simple to obtain the signal wavelengths of the other channels since adjacent channels are separated by 60 GHz. The corresponding idler wavelengths can be found easily from energy conservation law as the pump wavelength is known.

3. Results and Discussions

 

Fig. 4. (a) Two-photon interference fringe visibility and (b) entanglement fidelity of the generated photon-pairs versus photon-pair generation rate for Channel 27. The theoretical curve assumes a thermally distributed photon-pair number.

Download Full Size | PDF

First we look at the dependence of entanglement quality on photon-pair generation rate (in units of per pump pulse) when the pump laser is pulsed (from a Ti:Sapphire femtosecond laser). Figure 4 shows data taken for Channel 27 (signal and idler wavelengths are 1537.2 nm and 1567.1 nm, respectively). As shown in Fig. 4(a), the two-photon interference fringe visibility (average value calculated from HH, HV, VH and VV coincidence count rates) decreases with increasing photon-pair generation rate. The entanglement fidelity also decreases with increasing photon-pair generation rate, as shown in Fig. 4(b). This is due to an increased number of accidental coincidence counts as the result of a larger number of multiple-pairs emitted at higher pump powers [14]. This observation suggests that the photon-pair generation rate for a single channel cannot be raised to a high level where the effects of multiple-pair emission become significant. The theoretical curve in Fig. 4(a) is obtained using the theoretical model described in [7], assuming a thermally distributed photon-pair number.

We next compare pulsed and continuous-wave (cw) pumping. The cw pump is obtained from an external cavity laser (wavelength 775 nm). Figure 5 shows experimental results for selected channels. It is important to note that the source was optimized for one channel and left untouched during all subsequent measurements. As shown in Fig. 5, entanglement fidelity does not differ much across the wavelength range of interest for both cw pumping and pulse pumping. The slight decrease in entanglement fidelity observed at Channel 1 and Channel 44 is probably due to higher loss of the dichroic mirror at those wavelengths. We also found that for comparable level of coincidence count rates, cw pumping gives much higher entanglement fidelity than pulse pumping.

 

Fig. 5. Entanglement fidelity of selected channels, for both cw pumping (open circles) and pulse pumping (filled circles). The lines are included to aid visualization.

Download Full Size | PDF

This result can be understood as follows. The pulsed pump’s spectral full-width-half-maximum (FWHM) is 3 nm, which is much broader than the wavelength-demultiplexing filter bandwidth. This causes a significant fraction of the photon-pairs produced by the broadband pump being filtered by the narrowband filters into uncorrelated photons, leading to an increase in the number of accidental coincidence counts, which in turn, degrades the entanglement fidelity. In contrast, a cw pump does not have this problem. However, it should be noted that in general, pulse pumping is preferred over cw pumping in a practical entanglement distribution system because it is easier to synchronize timing for pulse pumping. Our result thus suggests that the quality of the entangled photon-pairs generated from our setup could be improved by using wavelength-demultiplexing filters having a broader bandwidth to reduce the number of uncorrelated photons (but this leads to fewer wavelength channels) or by using a narrowband pulsed laser instead for pumping. The optimal wavelength-demultiplexing filter bandwidth is closely related to the pump spectral width and will be the subject of future study.

As for the operating bandwidth of the source, although the SPDC bandwidth is expected to cover hundreds of nm (which is beyond our measurement capability now), we are currently limited by the tunability of the narrowband wavelength-demultiplexing filters. A second limitation would be the cut-off wavelengths of the dual-band PBS and the dichroic mirror, as determined by the optical coating. These limitations, however, could be overcome in future. Our preliminary results thus show that the concept of a single broadband source suitable for multiple-channel wavelength-multiplexed entanglement distribution is indeed a feasible one.

4. Conclusion

In conclusion, we have proposed a broadband source of telecom-band polarization-entangled photon-pairs that is well-suited for multi-channel wavelength-multiplexed entanglement distribution over optical fiber. By using a very short PPLN waveguide, together with broadband optically coated polarization-beam-splitter and dichroic mirror, we have shown that polarization-entangled photon-pairs can be produced over a broad bandwidth in the telecom-band. Using a pair of tunable wavelength-demultiplexing filters having 60 GHz pass-band, we have found that the proposed source is potentially capable of simultaneously supporting up to forty-four independent wavelength channels. We have also suggested ways to improve the current setup, and we expect the proposed broadband source to play an important role in future wavelength-multiplexed entanglement distribution fiber networks.