We present a novel all-fiber ultrahigh-repetition-rate pulse (UHRP) source based on ultrafast pulse-stimulated dissipative four-wave mixing (FWM). By injecting an ultrafast seed pulse into a dissipative ring cavity equipped with a spectral shaper, a comb-like nonlinear response is generated. The high peak power of the seed pulse reaches the FWM threshold, stimulating a 0.275 THz pulse with an output power of 0.5 W. The gain and spectral shaper in the fiber ring cavity form a dissipative system that modifies the initial field both temporally and spectrally, ensuring UHRP stability even after the pulse is turned off.
【AIGC One Sentence Reading】:We developed a 0.275 THz UHRP source using ultrafast pulse-stimulated FWM in a dissipative fiber ring cavity, achieving 0.5 W output power.
【AIGC Short Abstract】:An all-fiber ultrahigh-repetition-rate pulse source is developed using ultrafast pulse-stimulated dissipative four-wave mixing. It generates a 0.275 THz pulse with 0.5 W output power, achieved by injecting a high-peak-power seed pulse into a dissipative ring cavity. The system ensures pulse stability through gain and spectral shaping.
Note: This section is automatically generated by AI . The website and platform operators shall not be liable for any commercial or legal consequences arising from your use of AI generated content on this website. Please be aware of this.
Ultrahigh-repetition-rate pulses (UHRPs) are widely applied in spectral imaging[1], optical waveform synthesis[2], and astronomical spectrometer calibration[3]. When pulse rates approach 100 GHz, these sources become particularly important in precision measurements[4] and high-capacity optical communication[5]. Dissipative optical systems serve as an excellent platform for studying UHRPs[6,7]. Passively driven fiber ring cavities exhibit dissipative behavior due to constant external energy injection, despite lacking an internal gain component. Pulses formed within such resonators maintain stable operation even with a low-quality factor[8]. Fiber ring resonators with spectral filters function effectively as pulse shapers, enabling ultrafast pulse shaping[8]. This feature enables filter-induced instability to generate UHRPs at the sub-terahertz scale[9]. Filtering effects allow customizing repetition rates to meet specific requirements[10,11]. As more miniaturized passive drive resonators, microcavities have also gained attention for generating UHRPs[11,12]. Kippenberg et al. demonstrated temporal solitons in optical microcavities using continuous wave sweeping to achieve the detuning required for four-wave mixing (FWM)[6]. To increase UHRP power, researchers innovatively propose using pulse fields as the driving source[13,14]. This involves two key adjustments: tuning the pump frequency to the microcavity resonance to trigger modulation instability or dissipative FWM, and aligning the temporal period between the pump and microcavity for efficient operation.
Fiber lasers mode-locked using dissipative FWM[15] have been proposed as an effective approach for UHRP formation. Pasquazi et al. demonstrated a 200 GHz mode-locked laser with a microring cavity, terming the mode-locking approach as filter-driven FWM[16]. Key components for UHRP generation include multiwavelength filters and high-nonlinearity elements like microcavities[17], Fabry–Perot filters[13,18,19], Mach–Zehnder interferometers (MZIs)[20], and microring fibers[21,22]. Fiber-based comb filters provide benefits such as lower cost and reduced coupling losses between fiber and silicon/silica waveguides[23]. An adjustable MZI was employed to achieve variable repetition pulse trains in passively mode-locked fiber lasers, enabling UHRPs with GHz-range tunable repetition rates based on FWM[24]. Kbashi et al. showed that a UHRP mode-locked fiber laser using an MZI filter can serve as a sensor, offering significant advantages over other photonic strain sensing technologies[25]. In 2020, Xu et al. presented a novel 144.3 GHz pulse operation using a fiber-based comb filter[26]. This approach utilizes the nonlinear-polarization-rotation (NPR) effect, enabled by the polarization characteristics of microring devices. The technique, known as NPR-stimulated dissipative FWM, aims to boost intracavity power to enhance FWM mode-locking. The success of these UHRP schemes raises the question of whether new startup methods can effectively trigger UHRP excitation. We compared previously reported UHRP power levels shown in Table 1. The results suggest significant potential for achieving high average output power and UHRPs using the MZI filter.
Table 1. Comparison of Different Schemes of UHRP Fiber Lasers
Table 1. Comparison of Different Schemes of UHRP Fiber Lasers
Filter
Dispersion
Rate (GHz)
Pump (mW)
Output (mW)
High-nonlinear
Raman fiber laser[27,28]
negative
160
—
926
High-nonlinear
Fabry–Perot filter[29]
—
640
—
2.5
MZI
(Dual pump)[30]
negative
1000
800
50
High-nonlinear microring[16]
negative
200
—
15.4
High-nonlinear microfiber[26]
negative
144.3
390
6.21
Fiber loop[31]
negative
0.9
920
4.5
High-nonlinear microfiber[32]
negative
106.7
200
—
Fiber loop[33]
normal
280
2000
50
This study
normal
275
—
500
In this paper, we combined pulse-driven cavities with fiber lasers to develop a novel approach for generating UHRPs. An ultrafast seed pulse was employed to initiate a pulse field in a dissipative ring cavity, increasing intracavity power and stimulating dissipative FWM. An MZI acted as the comb filter, maintaining the all-fiber structure of the dissipative ring cavity and facilitating the production of 0.275 THz UHRPs. After modifying the laser cavity structure, the output power of the UHRPs reached up to 0.5 W. We compared the power of previously reported UHRPs, as shown in Table 1. This work demonstrates the highest average power for UHRPs without high-nonlinear components in the cavity in a 1.5 µm normal dispersion fiber laser based on the dissipative FWM effect, to the best of our knowledge. Additionally, once established, the UHRPs could sustain themselves in the dissipative ring cavity even after the ultrafast seed pulse was switched off.
Sign up for Chinese Optics Letters TOC Get the latest issue of Advanced Photonics delivered right to you!Sign up now
2. Schematic Diagram and Experimental Setup
The schematic diagram of the UHRP fiber ring cavity is illustrated in Fig. 1. It consists of a dissipative cavity for shaping laser pulses and an external seed pulse field. The seed pulse, characterized by its ultra-short duration, was switched off once a comb spectrum was formed. The dissipative cavity incorporates a gain component, a spectral shaper, and passive fibers. The spectral shaper generates a comb-shaped modulation field in the spectral domain. The comb spacing (Fs) is on the order of hundreds of gigahertz, whereas the fundamental repetition rate (F) is in the megahertz range. Mode matching related to this setup will be addressed in the discussion section.
Figure 1.Principle of the ultrafast pulse-stimulated UHRP. T0, period of the seed pulse; F, repetition rate of the seed pulse; Fs, repetition rate of the UHRP; FF and T, repetition rate and period of the dissipative cavity.
The corresponding experimental setup is shown in Fig. 2, where an external ultrafast laser injects an auxiliary field into the cavity. An erbium-ytterbium co-doped amplifier (EYDFA) module provided gain within the dissipative cavity. A segment of dispersion-compensating fiber (DCF) with a dispersion of was used to normalize the cavity dispersion and create a dispersion-managed system. A precisely prepared MZI[34] acted as the spectral shaper in the ring cavity. We tested the polarization effects introduced by the MZI using a polarization analyzer. The polarization-dependent loss of MZI was about 4 dB, which may affect pulse stability, signal-to-noise ratio, and pulse energy. To mitigate these effects, we adjusted the pump power to ensure consistent pulse energy and a high signal-to-noise ratio. The total cavity length, including the pigtails of the passive components, DCF, and fibers in the EYDFA module, was approximately 75.5 m. Two output couplers (OCs) with coupling ratios of 99/1 and 50/50 were used: one to output 1% of the signal to detect the UHRP, and the other to provide an input port for the ultrafast pulse. A polarizer was used to induce a saturation absorption effect. An optical spectrum analyzer with a 0.05 nm minimum resolution and a frequency-resolved optical gating (FROG) were used to monitor the laser outputs simultaneously. , DCF, the EYDFA module, and were connected by fiber connectors. Two test points labeled A were set: one before the DCF and one after the MZI, as shown in Fig. 2.
Figure 2.Experimental setup of the UHRP fiber ring cavity. DCF, dispersion-compensating fiber; EYDFA, erbium-ytterbium co-doped amplifier; OC, output coupler; FROG, frequency-resolved optical gating; OSA, optical spectrum analyzer.
The seed pulse was characterized in both the spectral and time domains, as illustrated in Fig. 3. At a 10 mW seed pulse power, it exhibited a spectral width of approximately 12 nm and a span of 35 nm, as shown by the blue line in Fig. 3(a). The autocorrelation trace revealed a pulse width of 1.17 ps [white line in Fig. 3(b)], indicating its ultrafast nature, as shown in Fig. 3(b). To characterize the transmission features of the MZI, we tested the time-domain and frequency-domain characteristics at Point B. The resulting transmission curve showed a comb-like shape with a spectral spacing of 2.3 nm [Fig. 3(c)], corresponding to a frequency interval of 0.275 THz. The FROG traces indicated that the seed pulse had undergone interference. The spectral modulation induced by the MZI filtering was evident in Fig. 3(d).
Figure 3.Characteristics of the seed pulse and MZI filter. (a) Spectra of the ultrafast pulse at Point A; (b) FROG trace of the seed pulse (inset, autocorrelation trace); (c) transmission curve and (d) FROG trace of the pulse at Point B.
We injected the seed pulse at Point A and detected the pulse information after the EYDFA and MZI at Point B [Fig. 2(b)] under different gain settings. The spectrum profile at Point B appeared to broaden when the EYDFA gain was set to 4000 mA (approximately 32.4 dBm) compared to 3500 mA (approximately 31.5 dBm), with a central wavelength of 1563.7 nm as shown in Fig. 3(c). Corresponding autocorrelation and FROG traces in Fig. 3(d) indicated that the seed pulse had undergone broadening and division due to the effects of DCF, EYDFA, and MZI filtering, which is consistent with our initial expectations. However, the spectral modulation induced by the MZI filtering was obvious.
We first set the pumping current of the EYDFA to 3500 mA (gain about 31.5 dB) and injected the seed pulse. The average power of the seed pulse was increased from 2 to 8 mW. The spectral evolution at the output port of is shown in Fig. 4(a). The spectra had a broad pedestal with a comb shape at 2 mW. A dominant tooth was located at the wavelength of 1567.35 nm, as shown in Fig. 4(a). The linewidth was about 0.45 nm with a resolution of 0.05 nm. Then the average power of the seed pulse was increased to 4, 6, and 8 mW. The intensities of the teeth beside the dominant one increased. The teeth were not stable as the average power increased (see the Visualization 1). This instability was because the interval of the external seed pulse did not align precisely with the cavity period, and an equivalent drift source emerged. As the seed power increased, modes in the cavity became competitive. There was only one main tooth in the spectral domain at powers of 2, 4, and 6 mW, and the tooth widths were 0.43, 0.7, and 0.4 nm, respectively, while another tooth was stimulated as the seed pulse power was increased to 8 mW, as shown in Fig. 4(a). Two main teeth formed, and the tooth widths decreased to 0.28 and 0.35 nm, respectively. The corresponding central wavelengths for the four states are marked in Fig. 4(a). Then the seed pulse power was slowly decreased to zero, as the Visualization 1 shows. The spectrum after the seed pulse is turned off is shown in Fig. 4(b). It can be seen that the pedestal disappeared, leaving a comb-shaped spectrum. We recorded four spectra every 5 min as shown in Fig. 4(c). After quantifying the intensity fluctuation of the teeth, we found the peaks had a smaller intensity fluctuation of about 0.4 dB, while the lower peaks had a larger fluctuation of about 1.7 dB. This fluctuation was mainly caused by the interaction between pulses in the cavity. To analyze the relationship between the time domain and spectral domain, we used an FROG device to measure both the spectrogram and the autocorrelation trace of the pulse waveform. The characteristics in the time domain after the seed pulse was OFF demonstrated the formation of UHRP, as shown in Fig. 4(d). The pulse separation was tested to be 3.63 ps, corresponding to a 0.275 THz repetition rate. The pulse separation corresponded well with the comb spectrum of the MZI filter, according to the equation in Ref. [34]. The pulse width was about 2.08 ps, assuming a Gaussian profile.
Figure 4.Ultrafast pulse-stimulated UHRPs. (a) The spectra at different seed pulse powers; (b) the spectrum of the UHRP when the ultrafast seed pulse is switched off (see the Visualization 1); (c) the recording of UHRP spectra every 5 min, with corresponding parts (d) detailing the FROG and the autocorrelation trace, respectively.
Figure 4(a) illustrates that, as the injected pulse power increases, sidebands around the dominant peaks become stimulated, confirming the feasibility of ultrafast pulse-stimulated UHRP. To explore more comprehensive features of UHRP comb formation and directly demonstrate ultrafast pulse stimulation, we set the EYDFA current to 4000 mA and injected a seed pulse averaging 10 mW. We recorded the spectrum at the port with the seed pulse ON (dotted line) and OFF (solid line), as shown in Fig. 5(a). The spectrum exhibited both a pedestal and a comb-shaped envelope when the seed pulse was ON, whereas the pedestal disappeared when the seed pulse was OFF. Additionally, we recorded the corresponding FROG traces to analyze the time-domain features, as shown in Fig. 5(b). The pulse interval was maintained at 3.63 ps, and the output power at the port was 0.5 W. We believe the output power of the UHRPs in this system can be further increased in two ways: One way is to improve the cavity design by removing and to reduce cavity loss. The MZI was composed of two 3 dB OCs. In this scenario, the seed pulse will be injected through the input port of the MZI, and the signal will be detected from the MZI output port after the seed pulse is switched off. The other way involves enhancing pumping efficiency: maximize the gain within the tolerable limits of the intracavity components to augment the signal power, or utilize gain fibers with higher gain coefficients to enhance the amplification capability of the gain module.
Figure 5.Diagram of the ultrafast pulse-stimulated UHRP. (a) Spectra when the seed pulses are ON (in blue) and OFF (in red), respectively; (b) presentation of the corresponding temporal characterization through FROG trace analysis.
We simulated the formation process of the UHRP in a dissipative fiber ring cavity. The model consists of two components: The first is an external source with an ultrafast temporal envelope that serves as a seed pulse; the second is the main cavity, which includes SMFs, a gain fiber, a DCF, OCs, and a comb filter, as depicted in Fig. 6. The simulations were conducted using the fast Fourier transform method to solve the Ginzburg–Landau equation[35]: where represents the electric field envelope, denotes the propagation distance, and is the local time of the pulse. The parameters and are the group velocity dispersion and the Kerr nonlinear coefficient, while g and Ω stand for the gain and gain bandwidth, respectively.
Figure 6.Simulation model of the ultrafast pulse-stimulated UHRP fiber laser. SMF, single-mode fiber; OC, optical coupler.
The gain fiber can be approximated by the following equation:
A more detailed explanation of the parameters in equations is available in Ref. [31]. Epulse is the pulse energy and Esat is the gain saturation energy. here was set as . The spectral response of the filter transmission can be expressed aswhere is the center angular frequency and is the period of the comb tooth. As an example, the inset of Fig. 7 shows the transmission spectrum for the filter. We also considered a saturable saturation absorption effect induced by the polarizer in the dissipative fiber ring cavity: where is the transmissivity, is the modulation depth, is the saturation power of the pulse, and Ssat is the nonsaturable loss.
Figure 7.Spectra and temporal pulses at different cavity locations: (a) and (b) seed pulses; (c) and (d) pulses after the comb filter.
In the simulations, periodic boundary conditions were applied to model the cyclic nature of the optical pulses in the fiber laser cavity. A pulse wave at the end of a fiber is seamlessly connected to the pulse wave at the beginning of the next fiber, effectively simulating the closed-loop nature of the laser cavity. At the output position, reflective boundary conditions were used. Here, a 50:50 OC and a 99:1 OC were used to output and return the wave to the cavity as shown in Fig. 6. The number of spatial grid points was 212, and the time window was 100 ps. The parameters used in the simulation are shown as follows.
The lengths of , , , , and were 5, 24.5, 5, 5.5, and 3 m, respectively. The SMFs were with a of and a of . The DCF had a length of 28 m, with a of and a of . The gain fiber had a length of 4.5 m, with a of and a of . We set to 0.6, to 3.9 W, and to 0.373 W. The seed pulse possessed a pulse width of 1.17 ps and a spectral width of about 15 nm. was set at 10 pJ. As illustrated in the schematic of Fig. 7, when the ultrafast seed pulse was injected into the fiber ring cavity, the spectrum initially broadened due to the DCF and was subsequently amplified by the gain component. Figures 7(a) and 7(b) show the temporal profile and spectrum of the seed pulse at a peak power of 20 W. During the first circulation, the seed pulse was modulated by the comb filter, as illustrated in Figs. 7(c) and 7(d). The comb spacing was set to 2.3 nm, corresponding to a repetition rate of 0.275 THz for 1.5 µm pulses.
Figure 8(a) shows the simulated UHRP formation spectrum, which closely matches the experimental results in Fig. 4(b). Within the dissipative ring cavity, the pulse field undergoes amplification, division, and shaping due to the combined effects of gain, loss, dispersion, and nonlinearity. As the seed pulse passes through the gain fiber, it splits. The filter then modulates the seed pulse spectrum into a comb shape in the spectral domain, as shown in Figs. 7(c) and 8(b). This modulation creates the initial conditions for the FWM effect. As the seed pulse evolves in the cavity, the shaping effects further stabilize the spectrum, as depicted in Fig. 8(b). The dynamic operation in the time domain can be divided into three steps. In Step I, a high-peak-power seed pulse stimulates the pulse field in the cavity, leading to splitting due to self-phase modulation instability. The continuous energy supply from the gain element, coupled with nonlinear loss from the MZI, induces dissipative characteristics in the ring cavity. The pulse undergoes further division due to nonlinear effects, gain, and the influence of a comb, leading to pulse modulation (see the inset of Step I). During Step II, modulation begins to stabilize. Over time, multiple pulses evolve into stable UHRPs with a uniform spacing of 3.63 ps that corresponds to the spectral spacing in Fig. 8(a), as illustrated in Step III. Pulse evolution can also be assessed by observing energy changes across different round trips. Initially, the high peak power rapidly decreases, followed by small amplitude oscillations. This demonstrates the rapid formation of dissipative FWM and the modulation of the pulse by the dissipative cavity during laser operation, as shown in Fig. 8(d). We show the energy evolution versus round trips, where the high power quickly drops to an average level. After 100 round trips, the energy of the pulses begins oscillating. As seen in Fig. 8(c), multiple pulses are periodically and completely distributed within the cavity at the 100th round trip. This energy oscillation indicates that the multiples are adjusted by the dissipative FWM process.
Figure 8.Characteristics of the pulse evolution: (a) UHRP spectrum; (b) pulse evolution in the frequency domain; (c) pulse evolution in the time domain (Insets illustrate three steps from the split state to the UHRP state); (d) energy evolution (The inset shows energy oscillation).
The comb spacing is on the order of hundreds of gigahertz, while the fundamental repetition rate is in the megahertz scale, as shown in Fig. 9. Generally, they should meet the following condition to form a stable mode-locking pulse[31]: where is the comb spacing of the spectral shaper, is the fundamental repetition rate of the ring cavity, and is an integer that represents the number of circulating pulses in the ring cavity. Here, is larger than , meaning that there are a huge number of longitudinal modes in one comb tooth of the spectral shaper. The ultrafast pulse modes and resonant frequency provided by the spectral shaper were easily matched. Technically, the length difference of the fiber-based MZI can be controlled within 10 µm using fused taper technology[34], which can correspondingly produce 20 THz repetition rate pulses. However, in theory, the upper limit of the repetition rate is also influenced by the gain bandwidth of the EYDFA. In this work, the gain bandwidth was about 50 nm. Assuming there were two teeth, the corresponding repetition rate was about 3.125 THz.
Figure 9.Relationship of the rates between MZI spacing and the cavity fundamental rate.
We highlighted two key steps in UHRP formation: the ultrafast seed pulse-driven ring cavity and the dissipative ring cavity after the seed pulse is turned off. In the first step, the ultrafast pulse serves as a pump. The EYDFA provides gains to compensate for the losses introduced by the OCs and the MZI. After the ultrafast pulse stimulates the optical field, the dissipative ring cavity generates a fiber laser. The ultrafast seed pulses serve as the initial conditions, leading to a process distinct from that of a passively mode-locked laser, which uses noise as the starting condition. We refer to this mechanism as the ultrafast pulse-stimulated FWM effect. From the comparison before and after the seed pulse is switched off in Fig. 5(a) and the Visualization 1, we observe that, when the comb teeth form, the seed pulse contributes a pedestal spectrum and acts as a perturbation due to its mismatched rate with the dissipative cavity. Since FWM is a phase-sensitive process, it fixes the phase of the mixing modes relative to each other, thus locking the modes and producing a pulse train with a period of in the time domain. After the seed pulse is switched off, the dissipative cavity establishes a new balance between gain and loss. The finite bandwidth and sufficient gain from the EYDFA module and MZI filter ensure that only the desired modes are amplified, preventing the growth of unwanted modes due to the dissipative nature of the cavity. This process prevents pulse decay. The ultrafast pulse-stimulated FWM method facilitates the initiation of a cavity with a large OC to achieve a high-energy UHRP.
In conclusion, this study presents a novel method for generating UHRPs in a dissipative ring cavity. Unlike passively driven fiber ring cavities that require constant pumping and high levels of nonlinear components to sustain dissipative properties and induce the dissipative FWM effect, our method utilizes the high peak power of ultrafast pulses. This approach creates a pulse field that significantly boosts the intracavity power instantly upon injection. By integrating a gain module and a spectral shaper, our system effectively prevents pulse splitting, promotes interaction, and maintains a uniform distribution within the cavity, thus stimulating dissipative FWM mode-locking. We call this new mechanism the “ultrafast ignition-stimulated dissipative FWM effect.” We have shown through both numerical simulations and experiments that this technique consistently produces 0.275 THz UHRPs. These pulses remain stable even after the ultrafast seed pulse is switched off. Additionally, the average power of these UHRPs is 0.5 W. To our knowledge, this represents the highest power achieved in 1.5 µm normal fiber lasers with repetition rates over 200 GHz based on a spectral shaper without using high nonlinear components. Our work provides a scalable solution to address challenges in optical communication, spectral imaging, and other fields.