3.1 Tunable DPL

To achieve DPL operation in the USP and NP, a mode-locked fiber oscillator is designed as shown in Figure 1(a). The maximum output power of the mode-locked pulse is 3.2 mW with a repetition rate of 32 MHz. The –3 dB bandwidth $\Delta {\lambda}_\mathrm{pulse}$ and the pulse width, as shown in Figures 2(a) and 2(b), are 0.64 nm and 5.6 ps, respectively, confirming USP generation.

Figure 2 (a) Optical spectrum and (b) autocorrelation trace of the mode-locked seed pulse. (c) Optical spectrum of the laser pulse after PAM1.

Since the generation of the ps/ns DPL relies on first generating the NP, the next step is to generate the NP. As shown Figure 1(b), the USP passes through the 1×2 FC1 with a splitting ratio of 4:6, with 60% of the power coupled into the TPSR for temporal broadening. However, an insertion loss can also be introduced by the TPSR. Therefore, PAM1 is introduced before FC1 to enhance the power and compensate for the loss. The amplified pulse, as shown in Figure 2(c), demonstrates a –3 dB bandwidth of 0.72 nm under 10 mW output power.

The temporal broadening $\Delta {\tau}_{\mathrm{total}}$ induced by the TPSR can be expressed as follows:

$$\begin{align*}\Delta {\tau}_{\mathrm{total}}\approx \mathrm{GVD}\cdot \Delta {\lambda}_\mathrm{pulse},\end{align*}$$

where GVD is the group velocity dispersion introduced by the TPSR and $\Delta {\lambda}_\mathrm{pulse}$ is the input pulse bandwidth^[ Reference Kashyap^20]. In our experiment, the GVD can be continuously adjusted between 500 and 720 ps/nm.

In particular, when the GVD is set to 500, 600 and 650 ps/nm, the corresponding stretched pulse widths are 0.38, 0.42 and 0.47 ns, respectively, as shown in Figures 3(a)–3(c). It can be seen that this dispersion enables controlled temporal broadening of the USP while nearly preserving its temporal profile. Further increasing the GVD to 720 ps/nm results in pulse broadening to 0.53 ns, as shown in Figure 3(d). However, pulse splitting is also observed, demonstrating a deterioration in the pulse temporal profile. To maintain the single-pulse operation, the stretched pulse with duration of 0.42 ns at GVD = 600 ps/nm is adopted for the subsequent amplification. Subsequently, the USP and the NP are combined using a 2×2 FC, generating a DPL sequence.

Figure 3 Typical pulse widths of the USP with (a) 500, (b) 600, (c) 650 and (d) 720 ps/nm GVD introduced by the TPSR.

The temporal interval is another important factor for improving the accuracy in the laser processing applications. Thus, except for the pulse width difference, the adjustment of the temporal interval between the two pulses is also desired. Meanwhile, the tunable temporal interval is primarily achieved by changing the optical path difference between the NP and USP sequences, as shown in Figure 1(b). In particular, an ODL (1.2 ns with 1 ps adjustment accuracy) is inserted into the USP path. Coarse adjustment of the temporal interval, typically with ns precision, is accomplished by varying the fiber length, while fine-tuning at the ps-level resolution is achieved via the ODL. This dual-adjustment mechanism enables a broad range of temporal interval modulation with a high resolution. The maximum tunable temporal interval ranges from –15.6 to +15.6 ns, which is determined by the pulse interval of the seed source. The positive value indicates that the NP precedes the USP, whereas a negative value signifies that the USP leads the NP. The temporal interval range cannot be further enlarged constrained by the repetition rate of the seed source. Figure 4(a) shows temporal profiles of the dual-pulse structure with intervals of –4, 4.2 and 12.5 ns, illustrating the feasibility of flexible control. Furthermore, the repetition rate of the DPL can be tuned from 100 kHz to 32 MHz by varying the settings of the AWG.

Figure 4 DPLs with temporal intervals of –4, 4.2 and 12.5 ns, respectively.

In practical applications, active control of the inter-pulse power ratio is essential, since their divergent nonlinear responses induce efficiency variations in energy extraction, mandating individualized pulse optimization^[ Reference Ahmed, Iqbal and Baig^21]. By adjusting the PA, precise control of the dual-pulse power ratio is achieved while simultaneously enabling the generation of fixed pulse width pulses. Furthermore, independent adjustments of the ODL and PA allow the precise and flexible modulation of both the temporal interval and power ratio. Thus, it can be seen from the above results that the proposed configuration enables the realization of an all-fiber DPL featuring four key tunable characteristics: (1) pulse widths; (2) temporal intervals; (3) power ratios; (4) repetition rates. Notably, since both pulse trains originate from the same seed laser, the system maintains intrinsic coherence without an additional passive synchronous control device. This integrated architecture makes it particularly suitable for subsequent amplification, while serving as a versatile optical source for various applications requiring tailored DPL outputs.

3.2 Main amplifier

Applications such as nano/micro fabrication demand high-power lasers to meet rigorous processing requirements^[ Reference Yang, Wei, Ma, Song, Ma, Zhao, Huang, Zhang, Yang and Wang^22– Reference Li, Yu, Zou, He, Ning, Wu, Chen and Lin^25]. Thus, in this part, the amplification properties of the proposed DPL are investigated. A seed source operating at 32 MHz with an average power of 307 mW has been chosen for subsequent pulse selection and amplification. The power ratio of a DPL output, with 12.9 ps and 0.43 ns pulse widths separated by 4.2 ns temporal intervals, is approximately 1:1 and is preserved for subsequent amplification studies. Figure 5 presents the temporal profile and amplified spectrum of the DPL. The signal-to-noise ratio (SNR) of the DPL is 38.3 dB, as shown in Figure 5(b). After the two pulses are coupled, PAM3 is cascaded with a multi-mode fiber amplifier to provide amplification. After PAM3, the powers of the DPL are quantified as follows: 5.13 W at 16 MHz, 4.96 W at 6.4 MHz and 4.79 W at 3.2 MHz.

Figure 5 (a) The temporal profile and (b) the corresponding spectrum of the DPL.

Figure 6 presents amplified powers and corresponding spectra for DPL operation, NP operation and USP operation, at a repetition rate of 16 MHz. The DPL (blue line) achieves a maximum output power of 1092 W with an optical-to-optical conversion efficiency (η) of 80%. Similarly, the NP (green line) reaches a maximum output power of 1148 W at η = 84.1%. In contrast, the USP (red line) delivers a comparatively low maximum output power of 412 W, with a conversion efficiency of 63.4%. This limitation is mainly due to the onset of pulse instabilities at high pump power levels, which can lead to detrimental effects, including irreversible optical fiber damage.

Figure 6 (a) The output power of the main amplifier versus pump power under DPL, NP and USP operations. (b) Spectra of DPL operation under 136, 538 and 1092 W output. (c) Spectra of NP operation under 54.9, 539 and 1148 W output. (d) Spectra of USP operation under 50, 191.4 and 412 W output.

Comparative analysis reveals significant divergence in amplification characteristics across pulse duration regimes. The USP exhibits efficiency saturation when pulse energy exceeds 7.75 μJ (124 W@16 MHz), manifested as progressively decreasing amplification efficiency. In contrast, both NP and DPL operations maintain linear power scaling behavior. Notably, when DPL pulse energy exceeds 38.75 μJ (620 W@16 MHz), the DPL exhibits reduced amplification efficiency compared to the NP, which may be due to the influence of the USP of the DPL. A comparative spectral analysis of the amplified pulses is conducted to elucidate the underlying physical mechanisms.

Figures 6(b)–6(d) systematically investigate the spectral evolution in the main amplifier under three different operations. Notably, Figure 6(b) reveals severe spectral broadening under DPL operation at 1092 W, primarily driven by strong nonlinear effects, particularly stimulated Raman scattering (SRS). The SNR decreases to only about 22 dB. In contrast, Figure 6(c) demonstrates that the NP operation maintains an exceptional SNR of 40.4 dB even at a higher power of 1148 W, while ASE dominates the noise profile. This result highlights the NP’s remarkable resistance to nonlinear spectral distortion. As for the USP, it exhibits extreme spectral broadening, shown in Figure 6(d) under 412 W operation, with SRS drastically reducing the SNR to just 7.3 dB. Critically, comparative analysis demonstrates that at 1 kW output power, the DPL exhibits significantly suppressed spectral broadening relative to the USP – an effect arising from the NP component, which endows the DPL with intermediate spectral characteristics between the NP and USP.

The pulse width of the NP measured by an oscilloscope is 0.44 ns at an output power of 1148 W, as illustrated in Figure 7(a). Meanwhile, the pulse width of the USP measured by an autocorrelator is 15.9 ps at an output power of 400 W, as shown in Figure 7(b). Under increasing output power, the NP demonstrates remarkable pulse width stability (0.42–0.44 ns), while the USP suffers from substantial pulse width broadening (5.6–15.9 ps).

Figure 7 (a) Pulse width of the NP recorded using an oscilloscope. (b) Autocorrelation trace of the amplified USP.

The observed nonlinear effects originate mainly from the USP’s high peak power in the fiber medium^[ Reference Dai, Qi, Yu, He, Yang and Lu^26, Reference Qi, Gao, Zhang, He, Lu, Yan, Sun, Lin and Yang^27]. The pulse width of the NP (0.44 ns) is approximately 27 times that of the USP (15.9 ps), leading to a proportional reduction in peak power. Thus, it demonstrates that spectral broadening in DPL amplification is primarily governed by the USP component. Furthermore, the broadened DPL spectrum adversely affects amplifier efficiency through nonlinear processes. Spectral broadening triggers rapid nonlinear accumulation, leading to irreversible energy transfer to Raman-shifted wavelengths. This cascade effect ultimately causes signal depletion and degraded pump-to-signal conversion efficiency.

3.2.1 Impact of temporal intervals on optical amplification characteristics

To systematically investigate the influence of temporal intervals on amplification dynamics, the temporal intervals are set to 4.2 and –1.5 ns. As shown in Table 1, at temporal intervals of 4.2 and –1.5 ns, the maximum output powers at a repetition rate of 16 MHz are 1092 and 1089 W, with corresponding η of 80.0% and 79.6%, and spectral SNRs of 22.0 and 20.8 dB, respectively, demonstrating nearly identical amplification performance. Similar amplification characteristics are also observed at a repetition rate of 3.2 MHz.

Table 1 Amplification characteristics under varying temporal intervals.

Owing to the limited temporal resolution of the current detection system, the minimum temporal interval is controlled at 1.5 ns. Consequently, the above conclusion is strictly valid only for temporal intervals exceeding 1.5 ns. However, as temporal intervals decrease into the ps and sub-ns regime, significant nonlinear inter-pulse interactions are expected to emerge, fundamentally altering the amplification dynamics^[ Reference Qi, Wang, Wang, Gao, Zhu, He, Lv, Yang and Lin^28]. Specifically, theoretical studies indicate that a diminishing temporal interval between pulses can lead to pronounced competitive energy extraction mediated by gain and loss dynamics^[ Reference Chen, Lin, Hu, Wang, Liang, Ling, Yang, Guo, Liu, Chen, Wei and Yang^29]. Therefore, elucidating the underlying physical principles of this competitive extraction mechanism warrants further investigation to optimize the performance of USP amplification systems.

3.2.2 Amplification characteristics under low-repetition-rate operation

In response to industrial demands for enhanced performance, this study investigates the DPL amplification at 6.4 and 3.2 MHz repetition rates, which support high-energy pulses. Figure 8 presents the output power, spectra and temporal profiles of the DPL. At 6.4 MHz, the system achieves a maximum output power of 1024 W (η = 75.1%) with a spectral SNR of 26 dB, while at 3.2 MHz, the maximum power reaches 708 W (η = 69.3%) with an SNR of 25 dB, as shown in Figures 8(a) and 8(b). The maximum pulse energy of the DPL is 221 μJ (3.2 MHz@708 W). Meanwhile, Figure 8(a) reveals a linear dependence between the DPL output power and pump power under both repetition rate conditions.

Figure 8 (a) The output power of the DPL and (b) the amplified spectra of the DPL at 6.4 and 3.2 MHz.

As demonstrated in Figure 8(b), the measured spectral SNRs exhibit distinct dependencies on repetition rate and output power. At 6.4 MHz with 1024 W output power, the SNR is quantified as 33.2 dB for ASE and 26 dB for SRS. When operating at 3.2 MHz with 708 W output power, the corresponding SNR decreases to 28.1 dB (ASE) and 25 dB (SRS), respectively. This systematic characterization reveals that lower repetition rates significantly promote ASE generation, as evidenced by the SNR reduction^[ Reference Zheng, Xie, Liu, He, Qi, Li, He, Liu, Lin and Yang^30]. This behavior can be attributed to the reduced energy extraction efficiency under low-repetition-rate operation, which creates favorable conditions for ASE development.