1. Introduction

High-power and high-energy pulsed fiber laser sources in the picosecond regime are of great interest for a variety of applications ranging from material processing, such as pulsed laser deposition, ablation and micromachining [1–3], to nonlinear frequency conversion employing techniques such as optical parametric oscillation and second harmonic generation [4,5].

Various approaches have been used to perform power scaling of picosecond laser sources. For instance, a MOPA seeded by a mode-locked fiber laser has previously been demonstrated to deliver 3-μJ pulses at an average output power of 150 W and a peak power of 1 MW; however it required free-space coupling between stages and the amplifiers were comprised of rod-type photonic crystal fibers with 100-μm core diameter [6]. On the other hand, a gain-switched-diode-seeded MOPA, with a 30μm-core-diameter Yb³⁺-doped silica fiber (YDF) in the final-stage amplifier, has been shown to deliver a maximum average output power of 200 W at 214 MHz; however the pulse energy generated was only around 0.9 μJ, with a corresponding peak power of about 30 kW [7]. A similar MOPA, which was used for pumping an optical parametric oscillator (OPO), could provide 100-ps pulses at 1064 nm with pulse energy up to around 8 μJ at a maximum average output power of 62 W [4]. However, further energy scaling with this system was primarily limited by the early onset of SRS due to the long length (15.7 m) of active fibers in the MOPA chain, including a 5.7-m-long final-stage power amplifier. Other approaches for power scaling include chirped-pulse amplification (CPA) and divided-pulse amplification, which are more complicated than direct amplification as they require extra components. While various high-power sources have been demonstrated, sometimes the range of their applications might be limited. For example, a 100-W CPA picosecond source running at 1-MHz burst rate [8] could potentially be useful in the material processing industry; however the 240-nm 3-dB spectral bandwidth makes this system unsuitable for nonlinear frequency conversion.

An improvement in energy scaling of picosecond pulses while maintaining a narrow spectral linewidth, high peak power and high average power would open a door to more scientific and technological opportunities. Furthermore, a stable, compact and robust laser source in a fiberized configuration built with shorter and smaller-core commercial fibers, which can be coiled and spliced together, would greatly enhance its practicality; especially in applications that would benefit from stable pulse qualities over an extended period of time.

In particular, besides direct application to material processing, we are interested in building a high-energy and high-power source geared towards pumping nonlinear frequency conversion devices to further expand the range of its applications. For example, the high-energy pulses could be converted into the mid-infrared (MIR) regime via an OPO, and then be used for resonant-infrared pulsed-laser deposition (RIR-PLD) [9]. Specifically, we will target high-energy MIR pulses at wavelengths between 2.5 and 4 μm, covering the vibrational resonance of a wide range of important functional groups, including the OH-stretch and the CH-stretch. Based on the group-velocity mismatch between the pump and idler waves, the associated acceptance bandwidth is estimated to be ~1 nm for quasi phase matching at 152°C in a 4-cm-long periodically-poled lithium niobate (PPLN) crystal. Therefore, it is important to design the MOPA system such that the 3-dB spectral bandwidth of the delivered pulses stays within 1 nm. Moreover, for efficient nonlinear frequency conversion, a polarization-maintaining (PM) and near-diffraction-limited beam is desired from our source.

Here we demonstrate a stable, compact, gain-switched-diode-seeded Yb³⁺-MOPA system, in a nearly-all-fiberized configuration with minimum free-space optics, capable of generating narrow-linewidth picosecond pulses with high pulse energy, high average power and high peak power in a diffraction-limited output beam with single polarization. There are several advantages in the design of this system. First, the system employs a simple direct-amplification configuration comprising spliced and coiled commercial conventional YDFs with core diameters of ≤ 25 μm and a total length of only 6.6 m. Second, the seed wavelength is chosen such that it resides at the blue edge of the ytterbium gain band in order to minimize the build-up of short-wavelength amplified spontaneous emission (ASE) as well as to allow the lengths of the amplifiers to be kept short. As a result of the reduced length, nonlinear spectral broadening due to self-phase modulation (SPM) and nonlinear energy transfer due to SRS can be minimized. Third, the length of each stage is optimized experimentally to minimize ASE and nonlinearity while ensuring high pump efficiency. Fourth, pulse energies approaching 18 μJ and average powers at the 100-W level are simultaneously extracted. Finally, the stability and robustness of the system indicate its potential for practical use, in addition to a proof-of-concept demonstration.

Our system provides pulse energy up to 17.7 μJ with a 97-W maximum average power (excluding ASE) at 5.47 MHz. With a measured 90-ps pulse duration, this corresponds to a 197-kW peak power. For bandwidth-dependent applications such as pumping nonlinear frequency conversion devices, this system is capable of delivering 13.8-μJ pulse energy well confined within a stable 3-dB spectral bandwidth of 0.87 nm. This bandwidth enables the system to potentially generate high-energy MIR pulses at 2.5 – 4 μm by pumping a PPLN-based OPO [4].

We have explored the limits of pulse energy and power scaling with the system until SRS becomes significant (i.e. spectral intensity of the SRS peak exceeds 1% of that of the signal peak).

2. System configuration

As depicted in Fig. 1, the picosecond MOPA consists of a PM-fiber-pigtailed gain-switched laser diode as the seed, followed by a 4-stage YDF-amplifier chain. A 1030-nm Fabry-Perot laser diode (Oclaro LC96A1030-20R), gain-switched by a train of RF pulses operating at a repetition rate of 87.5MHz, is self-seeded by feedback from a uniform fiber Bragg grating with a reflectivity of 12.5% to generate 120-ps, 3.77-pJ pulses at 1034.5 nm with a 3-dB spectral bandwidth of 0.026 nm. The optimization of amplifier lengths was performed experimentally to achieve enough gain to seed the following stage and to extract high power in the final stage, while limiting the ASE build-up and the SPM broadening, and assuring high pump efficiency at each stage. The pump absorption is ~95% in the final-stage power amplifier and ASE is suppressed by at least 25 dB below the signal throughout the entire pre-amplifier chain.

 

Fig. 1 PM picosecond Yb³⁺-fiber MOPA system, seeded by a gain-switched laser diode. Residual pump light in the power amplifier is stripped in the tapered section.

Download Full Size | PDF

Three pre-amplifiers are used to realize enough signal power to seed the final-stage amplifier. The first-stage pre-amplifier consists of an 85-cm-long PM-YDF of 5-μm core diameter and 130-μm cladding diameter. It is forward core-pumped by a 975-nm single-mode laser diode to amplify the pulses to 17 mW. A fiber pigtailed electro-optic modulator (EOM) (Photline NIR-MX-LN-10) is then used as a pulse picker to reduce the repetition frequency to 5.47 MHz. We choose to use a pulse picker rather than running the seed at 5.47 MHz because this significantly reduces the feedback cavity length for the gain-switched diode laser and allows more flexibility in tuning the operating repetition rate of the output pulses for our future work. A total loss of 17 dB was measured resulting from the 5-dB insertion loss of the EOM as well as a factor of 16 in the reduction of the operating frequency. This relatively high loss dictated the use of an additional pre-amplifier. The second-stage pre-amplifier, having a similar physical configuration to the first-stage pre-amplifier brings the average power to 12 mW, ensuring adequate seeding of the third-stage pre-amplifier. The third pre-amplifier is composed of a 2.5-m-long PM-YDF that has a core diameter of 10 μm with an NA of 0.075 and a cladding diameter of 125 μm with an NA of 0.46. This fiber is forward cladding-pumped by a 975-nm multi-mode pigtailed laser diode to scale the average output power to around 80 mW. The pulses are then fed into the final-stage amplifier. Fiberized isolators are used to prevent backward ASE leakage between stages.

The final-stage power amplifier comprises a 2.5-m-long commercial large-mode-area (LMA) PM-YDF (Nufern PLMA-YDF-25/250-VIII), which has a core diameter of 25 μm with an NA of 0.06 and a cladding diameter of 250 μm with an NA of 0.46. A silica end-cap is spliced to the output end of the fiber to reduce the risk of facet damage. Although this LMA fiber can support two spatial modes, namely LP₀₁ and LP₁₁, an in-house-made tapered section is used to preferentially excite the fundamental mode, while a fiber-coiling diameter of 80 mm is used to enhance the leakage loss for the LP₁₁ mode in order to maintain single-mode operation while maintaining a low loss for the LP₀₁ mode. An array of 975-nm multi-mode laser diodes are then combined through a 7x1 pump combiner to free-space backward-pump this LMA PM-YDF. A pump coupling efficiency of 95.5% has been achieved. To increase the robustness and stability of the amplifier, a water cooling system is implemented to minimize the risk of thermal damage to the LMA fiber as well as to reduce thermal drift of the pump wavelength. The LMA fiber is put in a water-cooled assembly made of aluminum plates with ~1 mm of stripped fiber extending out for pump launch.

3. Experimental results

The average output power (excluding ASE) of the MOPA system is plotted against the launched pump power of the final-stage amplifier in Fig. 2, showing a slope efficiency of 79%. An overall energy gain of 66.7 dB is achieved through the MOPA system, reaching 17.7 μJ. This is based on the 97-W maximum average power calculated by taking into account the contribution due to ASE, estimated from the spectrum (Fig. 3(b)) measured at the total output power of 109 W. At this maximum operating condition, an output polarization extinction ratio (PER) of > 14 dB and a diffraction-limited beam quality of M² ≤ 1.07 are measured. A fluctuation of only ~1% in the output power is observed, with no change in the spectrum over 30 minutes of continuous operation at the maximum power.

 

Fig. 2 Average output power (excluding ASE) versus launched pump power of the final-stage power amplifier.

Download Full Size | PDF

 

Fig. 3 (a) Spectra (resolution = 0.01 nm) measured after the seed, after the pre-amplifier chain and at different average output power levels of the MOPA system. (b) Spectra (resolution = 0.5 nm) measured at different stages of the MOPA system when the pulse energy reaches 17.7 μJ. (c) Spectral evolution (resolution = 0.5 nm) when power and energy scaling are performed.

Download Full Size | PDF

The designed gain of the pre-amplifier chain provides a good suppression of nonlinearity before the pulses enter the power amplifier. This means that SPM only takes place in the final stage of the MOPA, leading to a broadening of the 3-dB spectral bandwidth from 0.026 nm to 0.87 nm when the power and energy are scaled up. This effect is delineated in Fig. 3(a); the spectral measurement was performed with an optical spectrum analyzer (OSA) (ANDO AQ6317B) at 0.01-nm resolution. The spectral profile of the seed pulse is included in the figure as a reference. It can be seen that the spectral broadening becomes apparent when the average output power exceeds 19 W, corresponding to a pulse energy of 3.5 µJ. When the pulse energy reaches 17.7 μJ at average output power of 97 W, the 3-dB spectral bandwidth of the signal broadens to 0.87 nm. From the spectral broadening, the peak nonlinear phase shift accrued in the final amplifier stage is estimated to be 10π [10]. Figure 3(b) plots the spectral profiles at 0.5-nm OSA resolution, showing the change in the measured optical signal-to-noise ratio (OSNR) at different stages of the MOPA chain when the system is delivering the maximum pulse energy of 17.7 μJ. The OSNR decreases from 42 dB at the seed stage to 25 dB after the 3rd pre-amplifier, and finally to 18 dB after the power amplifier at the maximum operating condition.

An exploration of the limits on the capability of pulse energy and power scaling with the MOPA system has been performed. The associated spectral evolution shown in Fig. 3(c) clearly indicates a rapid growth of the SRS peak spectral intensity as the power scaled up. When the average output power reaches 70 W, the SRS peak at ~1082 nm starts to become apparent and lies at 30 dB below the signal peak. However, as the output power is increased from 70 W to 97 W (i.e. a factor of 1.38), a rapid 10-fold growth of the SRS peak spectral intensity is observed, reaching to 20 dB below the signal peak. At this point, the 0.87-nm 3-dB spectral bandwidth of the pulses, carrying 13.8-μJ pulse energy, sits well within the targeted 1-nm acceptance bandwidth for MIR generation via a PPLN-based OPO. Further energy scaling with the current system is primarily limited by the rapid growth in SRS level.

In the temporal domain, a 90-ps pulse width is measured at the output of the MOPA system, as illustrated in Fig. 4 (solid red line). This implies a peak power of 197 kW for the 17.7-μJ pulses at a repetition rate of 5.47 MHz. By using a spectral filter (Semrock LP02-1064RU-25) to exclude the wavelengths beyond 1050 nm, it is found that the measured pulse width increases to 124 ps, which is comparable to the 120-ps duration of the seed pulse. All the pulse widths are measured at the full width at half-maximum using a 20-GHz digital communication analyzer (Agilent HP 83480A) and a 32-GHz-bandwidth detector (Agilent HP 83440D).

 

Fig. 4 Temporal pulse profiles before and after amplification.

Download Full Size | PDF

4. Summary

In summary, we have presented a stable, compact and robust gain-switched-diode-seeded MOPA system capable of producing high-energy, high-peak-power and high-average-power picosecond pulses with narrow linewidth in a diffraction-limited and single-polarization output beam. It employs simple direct amplification through commercial conventional YDFs with core diameters of ≤ 25 μm and a total length of 6.6 m. An overall energy gain of 66.7 dB is achieved, reaching pulse energies of 17.7 μJ at 5.47-MHz repetition frequency, with an average output power of 97 W (excluding ASE). A maximum peak power of 197 kW is demonstrated. The measured 3-dB spectral bandwidth of 0.87 nm contains 13.8-μJ pulse energy, making this MOPA an attractive high-energy source for bandwidth-dependent applications. The PER and M² are measured to be > 14 dB and ≤ 1.07 respectively at the maximum operating condition. Stability tests show a power fluctuation of only ~1% over 30 minutes of continuous operation at the maximum output power, with no change in the spectral and temporal domains. By providing pulse energies approaching 18 μJ simultaneously with 100-W-level average power at MHz repetition rates in a diffraction-limited and single-polarization beam, combined with narrow spectral linewidth and 200-kW-level peak power, this compact, stable and robust picosecond source should be attractive to diverse sectors for a variety of applications, especially in high-throughput material processing and for nonlinear frequency conversion to generate high-energy pulses in other spectral regions. In particular, it is a promising source for generating high-energy MIR pulses by pumping a PPLN-based OPO, which is currently under investigation.