The quantum secure network (QSN)^[1–10] is a communication paradigm that leverages the principles of quantum mechanics to realize fundamentally secure communication. Unlike classical cryptographic methods, which rely on computational hardness assumptions, QSN utilizes quantum entanglement^[11–13] and quantum superposition^[14–17] to enable secure communication and resist both classical and quantum threats. A key challenge in scaling QSN is the realization of broadband entangled photon-pair sources. Such sources are essential for improving key rates and enabling frequency multiplexing, thereby supporting multi-user quantum communication architectures and advancing the scalability of entanglement-based QSN.

Various nonlinear optical processes have been explored and utilized to generate entangled photon pairs, such as the spontaneous parametric down-conversion (SPDC)^[18–25] process in the second-order nonlinear optical materials like lithium niobate, and the spontaneous four-wave mixing (SFWM) process in the third-order nonlinear optical materials such as optical fiber^[26–29], silicon^[30–34], silicon nitride^[35–39], gallium nitride^[40], aluminum gallium arsenide^[41,42], and silicon carbide^[43]. Among these, sources based on periodically poled lithium niobate (PPLN) have garnered significant attention due to the strong second-order nonlinearity (χ(2)), which facilitates highly efficient SPDC. Quasi-phase-matching allows PPLN sources to be engineered for the broadband spectral range, enhancing their applicability for multi-user QSN configurations. On the other hand, PPLN-based photon pair sources can generate entangled photons at telecom wavelengths, which aligns with existing fiber-optic communication infrastructure, enabling easy integration into current and future QSNs. Despite these advances, generating broadband energy-time entangled photon pairs with high performance remains challenging.

In this study, we present an experimental demonstration of broadband energy-time entangled photon pair generation with a fiber-pigtailed PPLN waveguide. The quantum light source generates entangled photon pairs within a wavelength range of 64 nm in the telecom band through the SPDC process. Entangled photon pairs from eight paired International Telecommunication Union (ITU) channels are selected and filtered using a waveshaper with a filtering bandwidth of 200 GHz. The measured coincidence counts (CCs) of photon pairs from eight paired ITU channels exceed 152.9 kHz with the coincidence-to-accidental ratios (CARs) greater than 260 at a pump power of 51.3 µW. Energy-time entanglement properties are characterized through the two-photon interference measurements, and the visibilities for all two-photon interference curves are larger than 98.13%. These results pave the way for the development of large-scale QSNs in the future.

The experimental setup for generating the broadband entangled photon pairs is shown in Fig. 1(a). In our experiments, the PPLN waveguide (HC Photonics) is fabricated by the reverse proton exchange (RPE) method with a length of 50 mm and a poling period of 17 µm, respectively. The coupling efficiency of the PPLN waveguide at the output port is 86% in our experiments. A tunable continuous-wave (CW) laser at a wavelength of 770.3 nm is used to pump the PPLN waveguide, the power of which is adjusted and monitored by a variable optical attenuator (VOA) and an optical power meter (OPM) with a 99:1 beam splitter (BS), respectively. The polarization of pump light is controlled by a polarization controller (PC) and a polarization beam splitter (PBS) to ensure the polarization mode alignment of the PPLN waveguide. Correlated/entangled photon pairs are then generated through the type-0 SPDC process, where a pump photon is annihilated and the signal and idler photons are spontaneously generated in the telecom band. To maintain the phase-matching condition, the temperature of the PPLN waveguide is controlled by a temperature controller. To reject the residual pump light, an isolator (ISO) at the telecom band is used at the output port of the PPLN waveguide.

To verify the broadband property of the source, we measure the spectrum of a single photon generated from the PPLN waveguide. As shown in Fig. 1(b), the generated photons are selected by a tunable band-pass filter (TBF) before being detected by the superconducting nanowire single-photon detector (SNSPD), and a time-to-digital converter (TDC) is used for single-photon counting. For the measurement of the single photon spectrum, we set the bandwidth of the TBF to 0.65 nm and tune the center wavelength from 1480 to 1620 nm. The measured results of the single-photon spectrum without subtracting the dark counts of SNSPD are shown in Fig. 2(a), which is consistent with the theoretical results^[44] and implies that the correlated/entangled photon pairs can be obtained in a broadband spectrum with a full width at half-maximum (FWHM) of ∼64nm. The Raman photon spectrum is measured by replacing the PPLN waveguide with optical fibers at 770 and 1550 nm, the lengths of which are the same as that of the PPLN waveguide pigtails, as shown in Fig. 2(b). The count rate is consistent with the dark counts of the SNSPD, indicating that the generated noise photons are negligible.

The correlation properties of generated photon pairs are characterized through the measurement setup shown in Fig. 1(c), in which the signal and idler photons are filtered using a waveshaper. In particular, photon pairs from eight paired ITU channels are selected with a filtering bandwidth of 200 GHz, where the channel numbers are listed in Table 1. The signal and idler photons are then detected by the SNSPDs with the detection efficiencies of 82% and 85%, respectively. The TDC is employed to record the counting rates of the signal and idler photons, and the coincidence events between them. The results of correlation properties are shown in Fig. 3. Figure 3(a) shows the single side counts of signal and idler photons from paired ITU channels C44 and C48 under different pump powers. Squares and circles are experimental results, and both single-side counts follow a linear behavior to the pump power. Figure 3(b) shows the CARs under different pump powers, in which a CAR of 1037.7±26.0 is achieved with a pump power of 16.2 µW. As shown in the inset of Fig. 3(b), the CCs and accidental coincidence counts (ACCs) are obtained with a coincidence window of 300 ps. The CARs versus CCs of the photon pairs from eight paired ITU channels are shown in Fig. 3(c), which agrees with the theoretical expectation. For the paired ITU channels C44 and C48, the measured CCs reach 70.8 kHz at a pump power of 16.2 µW. The generation rate of photon pairs and brightness are calculated as 3.9 MHz and 150.4 MHz/(mW nm), respectively, with total losses of 8.7 and 8.6 dB for signal and idler photons. When the pump power reaches 51.3 µW, all the measured CCs of photon pairs from eight paired ITU channels are larger than 152.9 kHz with the CARs exceeding 260, which are summarized in Table 1.

When using a CW pump light, the correlated photon pairs generated in the SPDC process are inherently energy-time entangled, which has been widely applied in multi-user QSNs. In our work, the generated photon pairs are sent to the measurement setup in Fig. 1(d) to characterize the energy-time entanglement through two-photon interference in the Franson interferometer. In Fig. 1(d), the signal and idler photons are first separated through the waveshaper, and then go through two unbalanced Mach–Zehnder interferometers (UMZIs), respectively. In our experiment, the used UMZIs have a relative delay of 625 ps, and the relative phase difference α or β between the long arm and short arm can be adjusted by applying a voltage on the piezoelectric actuator. The photons from the output ports of the UMZIs are detected by the SNSPDs, and the TDC records the corresponding photon counts and CCs.

The two-photon interference results between paired ITU channels C44 and C48 are shown in Fig. 4 at a pump power of 51.3 µW. As shown in Figs. 4(a) and 4(b), the correlation histogram of signal and idler photons after propagating through the UMZIs results in three peaks. Two side peaks represent signal and idler photons taking the different arms (one taking the short arm and the other taking the long arm) of the UMZIs, resulting in these phase-independent coincidence results. The central peak represents signal and idler photons taking the same arm (either the short arm or the long arm) of the UMZIs, which results in the interference of the CC rate depending on the phase of α or β. Two-photon interference curves are shown in Fig. 4(c) when α=−π/2 and 0. The dots are measured data by varying the phase β and post-selecting the central coincidence peak. The lines are the 1000-time fitting results with the Monte Carlo method^[45], showing curve visibilities of 99.33%±0.16% and 99.50%±0.12% without subtracting the ACCs, respectively. The two-photon interferences of other photon pairs from eight paired ITU channels are measured, and the curve visibilities are listed in Table 1, presenting curve visibilities better than 98.13%. The measured results of Franson interference are further utilized to verify the violation of Clauser–Horne–Shimony–Holt (CHSH)–Bell inequality for the energy-time entanglement. The two curves shown in Fig. 4(c) indicate the curve visibilities of 99.33%±0.16% and 99.50%±0.12%, corresponding to the values of the CHSH–Bell inequality of S=2.8095±0.0045 and S=2.8143±0.0034. Thus, the Bell inequalities are violated by 180 and 240 standard deviations, respectively, when α=−π/2 and 0. The Bell inequality violations of entangled photon pairs from eight paired ITU channels are also listed in Table 1, showing violations by at least 33 standard deviations.

In conclusion, we have demonstrated a broadband entangled photon-pair source using a fiber-pigtailed PPLN waveguide, exhibiting high efficiency, spectral brightness, and entanglement fidelity. Based on the SPDC nonlinear optical process, such a light source produces correlated/entangled photon pairs with an FWHM of 64 nm in the telecom band. The correlated/entangled photon pairs are filtered using a waveshaper with a filtering bandwidth of 200 GHz. At a pump power of 16.2 µW, the generation rate of correlated photon pairs from paired ITU channels C44 and C48 reaches 3.9 MHz with a measured CAR of 1037.7±26.0, corresponding to a spectral brightness of 150.4 MHz/(mW nm). When the pump power reaches 51.3 µW, all the measured CCs of photon pairs from eight paired ITU channels are larger than 152.9 kHz with the CARs exceeding 260. Furthermore, the energy-time entanglement properties are characterized through two-photon interference measurements in the Franson interferometer. The interference visibilities of photon pairs from eight paired ITU channels are greater than 98.13%, and the Bell inequality violations surpass 33 standard deviations, confirming the high-quality entanglement of the generated photons. The broadband energy-time entangled photon-pair source would pave the way for the development of large-scale QSNs.