1. Introduction

High average-power, CW, diffraction-limited fiber lasers and amplifiers have been demonstrated with powers of up to a few kW, where they are typically limited by transverse modal instabilities (TMI) [1]. However, when amplifying nanosecond pulses to high average power, the limit is much lower, due to nonlinear effects, such as spontaneous Raman scattering (SRS) and self-phase modulation (SPM), which are generated due to high intensities and long interaction lengths typical of fiber lasers. Increasing the fundamental mode mode-field diameter (MFD) will decrease nonlinearities, but larger MFD typically decreases higher-order mode (HOM) bend-loss, causing a lower TMI threshold and deterioration of the beam quality.

As the core size becomes large, HOM loss drops rapidly in conventional fibers. Consider, for example, a step-index fiber with 0.06 NA operating at 1070 nm. The effective mode-field diameter and bend-induced higher-order mode loss, calculated from a mode solver [2], are shown in Fig. 1, assuming the fiber is coiled to 15 cm bend diameter. The MFD in Fig. 1 is for the fundamental mode, only. For core diameters above 25 µm, the LP₁₁ loss is less than 1 dB/m and the fiber operates as a few-moded fiber.

 

Fig. 1. Effective mode-field diameter and LP₁₁ loss as a function of core diameter for a step-index fiber operating at 1070 nm wavelength and coiled to 15 cm bend diameter.

Download Full Size | PDF

Recently, we reported amplifier results from new Yb-doped fiber with designs for increased HOM loss, allowing for high TMI thresholds [3]. In those fibers, as well as the fibers reported in this work, the HOM loss is calculated from the measured refractive index profile using a mode solver.

In Table 1 below, the baseline properties of the fibers tested in Refs. [3] and [4] are shown. Fiber #1 was a conventional fiber without the HOM suppressing design. In comparison, fiber #2 utilized the newly developed, increased HOM loss design. The MFD of fiber #1, 19.4 µm, was slightly smaller than that of fiber #2, 19.8 µm. In addition, the cladding absorption of fiber #1, 1.4 dB/m, was substantially lower than fiber #2, 1.8 dB/m, resulting in a 9 m operating length for fiber #1, compared to 7 m for fiber #2.

Table 1. Properties of optical fibers tested in this work

View Table

Despite the higher absorption and larger MFD, however, the TMI threshold of fiber #2 was measured to be 5 kW, compared to 3.2 kW in fiber #1. We attribute the increased TMI threshold to the factor 3 increase in the higher-order mode loss in fiber #2, demonstrating the significant advantage provided in the new designs with enhanced HOM suppression.

The new HOM suppressing designs provide scalability in MFD while still maintaining high HOM loss. Consider that in Fig. 1, the HOM loss of a conventional fiber with 30 µm core diameter is three orders of magnitude lower than a fiber with 20 µm core diameter. On the other hand, in Table 1, the HOM loss of a 30 µm core fiber (#5) with the new HOM suppressing design is nearly equal to that of a 20 µm core fiber (#2).

In Refs. [5–7] we showed that the mode-field of such fibers can be scaled up to core sizes as large as 40 µm with MFD up to 26 μm. This paper summarizes, provides additional details and processes, and presents all these results in an orderly manner.

The combination of large MFD and high HOM loss enables all-fiber amplifiers with high pulse energy, high average power, and high peak power, while still maintaining diffraction limited beam quality. Using the 25 µm MFD, we achieved 1.6 mJ, ns pulses, with 1.2 kW average power and 370 kW peak power [6]. With the higher absorption fiber with MFD of 26 μm, we demonstrate 2 mJ pulse energy with peak power of >420 kW [7]. This combination of high average power, high pulse energy, high peak power and diffraction limited beam quality has potential to allow increased throughput in fiber-based micro-machining applications, for example.

The results reported here, to the best of our knowledge, combine the highest pulse energy, highest average power and highest peak power to date, from an all-fiber laser/amplifier with ns pulses and diffraction-limited beam quality. A comparison of our results to previous results reported in the literature, is depicted in the final figure, and discussed there in the text, highlighting the improvements achieved in this work.

In section 2 we describe the experimental setup and procedure. In section 3, the results are presented, and in the final section we summarize and conclude.

2. Experimental setup and procedure

The pulsed amplifier setup is shown in Fig. 2. A seed laser was first preamplified and then launched into the power amplifier based on Yb-doped fiber with increased HOM loss. The experiments primarily used a co-pumped configuration (Fig. 2(a)), but an amplifier using a counter pumped configuration was also tested (Fig. 2(b)).

 

Fig. 2. a) Co-pumped and b) counter-pumped experimental setups. PSC: pump-signal combiner; CLS: cladding light stripper; DM: dichroic mirror; PM: power meter. (c) Seed laser and pre-amplifier configuration for 3 W launch into power amplifier.

Download Full Size | PDF

The seed laser and pre-amplifier configuration is shown in Fig. 2(c). The seed laser was a directly modulated diode laser, amplified in a series of pre-amplifiers with inter-stage isolators. Inter-stage bandpass filters were used to remove amplified spontaneous emission, and narrow the signal bandwidth. Gain fibers in the three pre-amplifiers stages 1,2, and 3 had core diameters of 4, 4, and 10 µm, and lengths of 1, 1, and 2 m, respectively. As described below, initial experiments were done with only two pre-amps with an output of 700 mW. In later experiments, the third stage was added to boost the seed power to 3 W. Figure 2(c) shows the 3W configuration of the seed laser system.

In this study, three different fibers were tested for pulse operation as detailed in Table 1 (#3,5,6). All fibers had a 400 µm cladding diameter. For reference, the fibers reported in [3,4] (#1,2) are also shown in Table 1. The MFDs of the fiber used for pulse testing, ranged from 20.1 µm to 26 μm. The HOM loss of each of the fibers is also given in Table 1, relative to fiber #2 reported in Ref. [4]. Fiber #3 was a relatively early fiber made with the new design, and as such was not as well optimized as later fibers made, resulting in a lower HOM loss relative to fiber #2. In contrast, fibers #4-6 showed HOM losses that were close to what was achieved in Ref. [4] with fiber #2, even though the core size and MFD were substantially larger in fibers #4-6.

Four banks of diode lasers were utilized, each consisting of six ∼65 W, wavelength-locked, 976 nm diodes in 105 µm core fiber. In each bank, the diode lasers were coupled together using a pump combiner (OFS CoolMode 7:1 pump combiner, part number 7000626B) into 230 µm core fiber for pump inputs. The 976 nm, wavelength-locked diode pumps were then coupled into the gain fiber together with the seed laser using a 6 + 1:1 pump-signal combiner with 10 um core signal input fiber, six 230 µm core multi-mode pump input fibers, and a 20/400 output fiber (OFS CoolMode 6 + 1:1 pump-signal combiner, part number 7000665B-05).

The 976 nm cladding absorption shown in Table 1 was measured at high power in an amplifier under load. To measure the high-power, 976 nm cladding absorption, an amplifier was built with the fiber under test. 20 W of CW seed power and 620 W of wavelength-locked, 976 nm diode pump power was launched into the amplifier. The gain fiber length was cut back while monitoring the unabsorbed pump. A fit to the unabsorbed pump vs. fiber length in the linear regime then provided the high-power, 976 nm cladding absorption. (The values of the absorption noted in references 4-6 used a low power measurement method. The absorption as reported here is more relevant to the amplifier in operation at high power. For reference, the small-signal pump absorption at 915 nm of fiber #5 was 0.98 dB/m and fiber #6 was 1.5 dB/m.).

An amplifier was then operated with a 3 W CW seed input @ 1064 nm, to test for the TMI threshold. TMI was monitored by calculating the standard deviation of a photo-diode signal (150 MHz BW), measuring power scattered from the power meter. A sharp increase in the photodiode standard deviation indicated the onset of TMI. A value of 0.3% standard deviation in the photodiode signal was used to set the TMI threshold consistently between different fibers.

The TMI threshold of fiber #5 with 24.8 µm MFD and 2.9 dB/m cladding absorption was 1.8 kW. The TMI threshold of fiber #4 was 2.2 kW, whereas the higher absorption, 26.3 µm MFD fiber (fiber #6) had a TMI threshold of ∼1 kW. The high average power obtained in such a large MFD, high absorption fiber confirms that the increased HOM losses helped increase the TMI threshold.

For the pulsed experiments, a seed laser at 1064 nm with an 8 ns pulse width was preamplified. A cladding light stripper (CLS) was employed at the output of the power amplifier, to remove residual power in the cladding. A dichroic mirror was used to separate any remaining pump power from the output amplified signal.

The first experiments done with fibers #3-5 in Table 1 utilized an angle cleave for end termination of the gain fiber and 0.7 W of seed power. While we were able to obtain high average power with this configuration, numerous fiber fuses occurred, particularly as the gain fibers were cut shorter and average power was increased. To address the fuses, the angle cleave was replaced with an AR-coated end cap in later experiments, and an additional preamplifier was built to boost the seed laser power up to 3 W on launch into the power amplifier. With these improvements, the amplifier could be run at higher output powers, without suffering fiber fuses.

The pulsed seed laser was run at repetition rates between 333 kHz and 4 MHz. At higher rep rate the amplifier was generally more efficient, however the pulse energy was lower. Depending on the fiber and the experiment, the amplifier was either pump power limited, Raman limited, or at times limited due to approaching the TMI threshold for the specific fiber, requiring operation at lower rep-rate, and hence lower average power.

The optimum length of the fiber was determined using a cutback method. The various fibers were tested at lengths from 7 to 2.5 m. For each fiber length, the pump power was increased, the signal power was measured, and the optical spectrum was monitored. The pump power was increased until the extinction ratio between the fundamental signal wavelength peak (1064 nm) and the Raman wavelength peak (1117 nm) was ∼10 dB. Then the fiber was shortened, and the test was repeated.

Pulse energy was determined by the average power divided by the repetition rate, since at such high rep rates, energy meters are unable to directly measure the pulse energy, and the ASE does not have opportunity to develop, due to the short intervals between pulses.

3. Experimental results

3.1 Co-pumping amplifier configuration

Fibers were first measured in an amplifier with a CW seed laser to characterize the TMI threshold. Figure 3 shows the output signal power vs. the pump power for fibers #4 and #5. The TMI thresholds were found to be at a signal power of 1.8 and 2.2 kW, respectively for the 2.9 and 2.4 dB/m fibers. The CW amplifier with the 2.4 dB/m fiber used 5 m of fiber, and the 2.9 dB/m fiber utilized 4 m of fiber.

 

Fig. 3. CW signal power out vs. pump power for two ∼25 μm MFD fibers, up to the TMI limit.

Download Full Size | PDF

Pulse experiments were performed with fibers #3, 5, and 6 from Table 1, with MFDs of approximately 20, 25 and 26 µm respectively. With the 20 µm fiber (#3), we started with a fiber length of L = 5.6 m and performed a cutback measurement down to L = 3.1 m. A summary of the maximum power out and the efficiency at maximum power, as a function of fiber length is shown in Fig. 4(a). As the fiber was shortened, the maximum efficiency decreased, due to an increase in unabsorbed pump. However, the maximum pulse energy increased, due to a decrease in nonlinearity with shorter fiber, allowing us to reach higher pump power. The efficiency is defined as the output power from the amplifier, minus the input seeder power, divided by the pump power measured at the input to the pump/signal combiner.

 

Fig. 4. Maximum pulse energy and efficiency at maximum pulse energy vs. fiber length co-pumped for a) 20 µm MFD fiber (500 kHz). b) 25 µm MFD with seeder at 0.7 W (500 kHz). c) 25 µm fiber w/ 3 W seeder (500 kHz + 1 MHz). d) 26 µm fiber, 3 W seeder (333 kHz).

Download Full Size | PDF

The fiber was further cutback and retested until the fiber was too short to absorb the pump sufficiently, and both maximum power and the efficiency decreased. A maximum average power of 436 W with an efficiency of 61.7% was achieved at a fiber length of L = 3.6 m for the 20 µm fiber. At 500 kHz repetition rate, the corresponding pulse energy was 0.87 mJ.

To further increase the achievable pulse energy, the diameter of the core was increased, resulting in a comparable increase in the pump cladding absorption. Two fibers (4 and 5) with MFDs of ∼24.7 and 24.8 μm and with cladding absorptions of 2.4 and 2.9 dB/m @ 976 nm respectively, were fabricated. For the pulsed work the 2.9 dB/m cladding absorption fiber (#5) was used, as the higher absorption allowed for shorter fiber, and lower nonlinearity. The seeder was operated at 500 kHz and amplified in a preamplifier to a level of 0.7 W, prior to amplification in the power amplifier.

The results are shown in Fig. 4(b). The ∼25 μm fiber in the co-pumping configuration produced 1.24 mJ with an efficiency of 69%. This fiber produced higher pulse energies than the 20 µm fiber, due to the larger MFD and shorter operating length, which provided lower nonlinearities.

As discussed earlier, to further increase the power and reduce the occurrence of fuses, an additional preamp was used with the seed laser, to boost the seed average power into the amplifier to 3 W. In addition, the output angle cleave was replaced with an AR-coated, angled end cap with 6 mm diameter. At a rep-rate of 500 kHz, a pulse energy of 1.6 mJ with 802 W average power, and efficiency of 77.6% was achieved, before the extinction ratio between the 1064 nm signal and the Raman peak at ∼1117 nm reached 10 dB. By operating the seed laser at 1 MHz, an average power of 1194 W was achieved with pulse energy of 1.194 mJ and efficiency of 78%. At 1 MHz, the amplifier was pump power limited. With additional pump power, we expect the average power could approach the TMI limit of 1.8 kW for this fiber. We did not continue to shorten the fiber for the 500 kHz case, since we had already seen in the 0.7 W seeder case (Fig. 4(b)) that shortening the fiber lowered the output pulse energy.

A graph showing the signal power vs. pump power for CW operation and pulsed operation at 500 kHz and 1 MHz rep rates is shown in Fig. 5(a). Each of the four pump banks could provide a maximum of ∼380 W. However, the pump wavelength of the banks was not locked at 976 nm until the pump power was between 180 and 220 W. When using more than one unit, the power of the first unit was increased to ∼180 W, and then additional banks were turned on sequentially in the same manner. When the power of any individual pump unit was <220 W, low efficiency resulted, as the pump wavelength was not locked, resulting in the low efficiency below ∼850 W pump power shown in Fig. 5(a).

 

Fig. 5. a) Signal power out and efficiency vs. pump power at CW, 500 kHz and 1 MHz. CW fiber length: 4.5m, pulsed fiber length: 3.2 m and b) pulse energy vs. pump power at 333 kHz with 26 µm fiber at various fiber lengths.

Download Full Size | PDF

Finally, fiber #6 with a 40 µm core, 26 µm MFD, and 4.5 dB/m, 976 nm cladding absorption was tested. The CW TMI threshold of this fiber was measured to be ∼1000 W in an amplifier using 2.25 m of fiber. For pulsed testing, the fiber length was cut from 3.8 m to 2.5 m with the seed laser repetition rate of 333 kHz. The lower rep rate was employed to avoid reaching the TMI power threshold. The pulse energy as a function of pump power at various fiber lengths is shown in Fig. 5(b). The maximum pulse energy obtained at each fiber length during the cutback procedure and the efficiency at the maximum pulse energy is shown in Fig. 4(d).

At 3.8 to 2.5 m fiber lengths, pulse energies of 1.17 to 2 mJ were obtained, with efficiencies of 81 to 66%, respectively. A maximum pulse energy of 2 mJ with an average power of 660 W was achieved.

The pulse shape from the output of the seed laser is plotted in Fig. 6(a) at the level of 0.7 W and 3W. Figure 6(b) shows the pulse-shapes at 1.6 mJ and 2 mJ as achieved with the 25 µm fiber and the 26 µm fiber as sampled with a Thorlabs FPD310-FC-NIR photodetector with a 1 GHz bandwidth. The undershoot at the end of the pulse is an artifact of the photodetector response. The 8 ns seeder pulse narrowed, due to gain-induced pulse shortening, as the output power was increased. The output pulse length in both cases was ∼4 ns long and the peak powers were ∼370 kW and ∼420 kW for the 25 µm fiber and the 26 µm fiber, respectively. The peak powers were calculated based on the known pulse energy and the integration of the pulse-shape from the photodetector, where only the positive part of the pulse-shape is accounted for.

 

Fig. 6. a) Seed laser pulse shape. b) Pulse shape at pulse energy of 2 mJ (26 µm fiber) and 1.6 mJ (25 µm fiber).

Download Full Size | PDF

Optical spectra measured under various conditions are shown in Fig. 7. The output of the seed laser at 3 W average power is shown in Fig. 7(a) over a broad wavelength range, and zoomed in in the inset. The pre-amp output at 3 W, with a peak at 1064 nm had a 3 dB bandwidth (BW) of 2 nm and a 10 dB BW of 5 nm. The spectrum obtained with the 25 µm MFD fiber at the maximum pulse energies at the two rep-rates is plotted in Fig. 7(b). A widening of the pulse spectra due to SPM is visible, and a Raman peak can be seen at ∼1117 nm, ∼10 dB and ∼13 dB below the 1064 nm peak, for 500 kHz and 1 MHz, respectively.

 

Fig. 7. (a) Seed laser spectrum at 3 W output power. Inset shows the spectrum on a zoomed in scale. (b) Output spectra at maximum pulse energies with 25 µm fiber at 1 MHz (1.2 mJ), and at 500 kHz (1.6 mJ) and with 26 µm fiber (2 mJ). c) Output spectrum from 25 µm fiber at L = 4.4 m as a function of signal power and pulse energy d) Output spectrum from the 25 µm fiber at L = 3.2 m at 500 kHz, as a function of signal power and pulse energy.

Download Full Size | PDF

Also in Fig. 7(b) is the spectrum obtained at the maximum pulse energy level with the 26 µm fiber. A spectral widening due to SPM can be seen. The Raman peak at 1117 nm is buried in the SPM.

Figure 7(c) shows the output spectra obtained with the 25 μm MFD fiber with a non-optimum fiber length of 4.4 m while using a seeder at 0.7 W. At low pump power levels, only the seeder wavelength appeared, while as the pump power was increased, the nonlinearities increasingly impacted the output spectrum. A Raman peak at ∼1117 nm was visible, while SPM widening of over 100 nm occurred. At the highest pulse energies, the Raman peak was washed out by the SPM, as shown in 7(b). When the extinction ratio between the 1064 nm signal and the Raman peak reached ∼10dB, the power increase was stopped.

Figure 7(d) shows the output spectra with the 25 μm fiber at the length of 3.2 m, where the high pulse energy of 1.6 mJ was obtained. There is considerable spectral broadening due to SPM, in addition to the Raman peak near 1120 nm. Comparing 7(c) to 7(d), the Raman scattering is far more significant in the longer fiber lengths, whereas SPM induced broadening becomes more significant in the shorter lengths.

To help quantify the nonlinear broadening for various operating conditions, the results of the spectral measurements are compiled in Fig. 8 in the form of spectral width and extinction ratio between signal and Raman peak as a function of pulse energy for the fibers, repetitions rates, and fiber lengths used in the experiments. Figure 8(a) and (b) plot the 3 dB and 10 dB spectral width respectively, as a function of output pulse energy, quantifying how the spectrum broadens with increasing pulse energy. Figures 8(c) and 8(d) plot the extinction ratio between signal and Raman peak for the 25 µm fiber at 500 kHz and 1 MHz rep rate respectively, for different fiber lengths. As the pulse energy increases, the Raman extinction ratio drops, whereas, the Raman extinction ratio increases for shorter fiber lengths, showing the benefit of decreased nonlinearity.

 

Fig. 8. Spectral width and Raman extinction ratio as a function of pulse energy. (a) 3 dB spectral width as a function of pulse energy. (b) 10 dB spectral width as a function of pulse energy. (c) Raman extinction ratio as a function of pulse energy, in the 25 µm MFD fiber at 500 kHz rep rate, at different fiber lengths. (d) Raman extinction ratio as a function of pulse energy in the 25 µm MFD fiber at 1 MHz rep rate, at different fiber lengths.

Download Full Size | PDF

Beam quality measurements of the amplifier output were performed using a Thorlabs M² measurement system. For CW experiments, the beam quality was measured using a rotating slit beam profiler. For all the fibers tested, the measured M² was < 1.1, up to the onset of TMI. A representative measurement for a CW signal in the 25 µm MFD fiber at 1.4 kW is shown in Fig. 9(a).

 

Fig. 9. Beam quality (M²) measurements. (a) CW measurement at 1.4 kW signal power in the 25 µm MFD fiber. (b) Pulsed measurement at 900 W signal power and 1 MHz rep-rate in the 25 µm MFD fiber. Inset – measured beam profile. (c) Beam quality vs. signal power in the 25 µm fiber up to 1060 W at 4 MHz and at 1 MHz and the 26 µm fiber at 500 kHz.

Download Full Size | PDF

For pulsed signals, a camera-based profiler was used. It was found that the camera-based M² measurements tended to be somewhat noisier than the rotating slit profiler measurements. In addition to higher background noise in the camera profiler, as compared to the rotating slit profiler, the camera profiler was far more sensitive to signal level and saturation. A representative M² measurement for the 25 µm fiber with 3.2 m length, at 1 MHz rep-rate and 900 W signal power is shown in Fig. 9(b). A typical pulsed beam profile measured with the camera is shown in the inset of Fig. 9(b).

A compilation of beam quality measurements is shown in Fig. 9(c). The beam quality of the 25 µm MFD fiber was measured with a length of 3.2 m at 1 MHz and 4 MHz rep-rates up to 1060 W signal average power. The 26 µm fiber was measured at signal output powers of up to 710 W at 500 kHz and a length of 3.5 m. With pulsed operation, the M² values were between 1.1 and 1.2 at all power levels.

3.2 Counter pumped configuration

The 25 μm fiber (#5) was also tested in a counter-pumped configuration. The setup is shown in Fig. 2(b). A custom pump-signal combiner (PSC) was designed for counter pumping the LMA Yb fiber with 32.7/400 common pump/signal fiber and 24.6/400 signal output fiber. The signal insertion loss of the counter-pump PSC was measured to be 0.22 dB.

Counter pumping generally provides lower nonlinearity as high signal power is shifted towards the end of the active fiber, as opposed to co-pumping (Fig. 2(a)), where high power is generated near the entrance side. However, counter-pumping requires additional passive fiber on both sides of the PSC, in which the maximum power propagates, so it is imperative to ensure that this auxiliary fiber be kept as short as possible to minimize nonlinear impairments. In this work, this additional passive fiber length was 0.65 m (total for both sides of the PSC). Finally, in a counter pumped configuration, the high-power signal propagates through the output PSC, which adds additional loss as well as the potential for beam quality deterioration. In fact, we did see the development of an additional mode in the output beam, indicating a poorer beam quality, due to the larger core output fiber, however we did not measure the M² in the counter-pumped case.

Here the active fiber length was tested from 7 m to 2.8 m. The maximum pulse energy and efficiency at maximum pulse energy as a function of the fiber length are shown in Fig. 10, together with the co-pumped results shown previously (Fig. 4(b) for comparison). As the fiber was shortened to L = 3.1 m, the efficiency decreased, as the shorter fiber absorbed less pump power; however, the output energy increased, with decreasing nonlinearity. At 2.8 m, the fiber was too short to absorb sufficient pump power and subsequently, both the energy and the efficiency decreased. The highest pulse energy obtained was 1.2 mJ with an efficiency of 61%.

 

Fig. 10. Maximum pulse energy and efficiency at maximum pulse energy vs. fiber length, with 25 µm MFD fiber and seeder at 0.7 W comparing co- and counter-pumped configuration (500 kHz).

Download Full Size | PDF

Comparing the two architectures, co-pumping produced a higher pulse energy and the efficiency at the maximum energy point was higher. However, the counter-pumped efficiency was much less sensitive to fiber length than in co-pumping. In co-pumping, if the fiber is too long, then the high energy pulse propagates through “unpumped” gain fiber where the pump has already been depleted. In contrast, with counter pumping, the “unpumped” region coincides with the weak pulse at the amplifier input. As a result, in applications where efficiency is an important parameter, counter-pumping could be advantageous. For example, at an efficiency of 77%, counter pumping produces 1 mJ of power, while co-pumping produces only 0.71 mJ at the same efficiency.

4. Summary and conclusions

Large-mode area fibers with increased HOM losses are capable of simultaneously achieving both high TMI thresholds and low nonlinearities. These fibers enabled diffraction limited output beams with an unprecedented combination of pulse energy and high average powers.

In contrast to the first reports of high TMI fibers with high HOM loss fibers in Ref. [3] and Ref. [4] that were operated CW, the fibers designed for high power pulsed operation have both significantly larger MFD and substantially higher cladding absorption to lower the amplifier nonlinearity as much as possible. The HOM losses achieved in these ∼25 µm MFD fibers are roughly comparable to the HOM losses reported in the 20 µm fibers designed for narrow linewidth operation with high SBS thresholds [3,4]. However, while high cladding absorption is important for operating with as short a fiber as possible, higher cladding absorption also places substantially higher demands on the HOM loss due to the higher heat load in shorter fibers. It can be seen in Table 1 that the TMI threshold decreases with increasing cladding absorption. Nevertheless, with cladding absorption levels compatible with pulsed fiber lasers, the achieved HOM loss levels have provided kilowatt average powers in these LMA fibers.

Figure 11 shows measured CW TMI thresholds, maximum pulse energy and maximum average power in pulse operation, obtained as a function of fiber clad absorption. As the absorption increases, the maximum achievable pulse energy increases, since a shorter fiber has lower nonlinearity. 0.9, 1.6, and 2 mJ were produced for 20, 25, and 26 µm respectively. However, as the absorption is increased, the TMI threshold decreases, as shown in Table 1 and in Fig. 11(right axis). As a result, the maximum average power that can be obtained simultaneously decreases. The TMI power threshold and the maximum obtained power, is shown in the same graph, using the right-hand axis. Consequently, the highest pulse energy of 2 mJ obtained with the 26 µm fiber was at a relatively lower rep-rate of 333 kHz so as not to surpass the TMI threshold of ∼1000 W.

 

Fig. 11. Maximum obtained pulse energy, CW power and pulse average power vs. fiber clad absorption at 976 nm.

Download Full Size | PDF

Finally, we summarize the results reported here and place them in the context of previous demonstrations of pulsed Yb-doped fiber amplifiers in the literature. Figure 12 plots the reported results of which we are aware of all-fiber, diffraction-limited, ns pulsed lasers with average power above 200W or pulse energy above 0.2 mJ [5–14]. Under these conditions, the highest power previously demonstrated was from Zhang [8] who reported pulse energies of 0.63 mJ with an average power of 570 W. Su [9] reported an average power of 913 W with ns pulses. However, the pulse energy was only ∼0.090 mJ. 466 W with 0.046 mJ energy was reported by Huang in [10].

 

Fig. 12. All-fiber, diffraction limited, ns reports of pulse energy vs. average power. The repetition rate for each of our data points is noted in the graph.

Download Full Size | PDF

The red triangle, orange squares and green diamonds show the results with 20, 25, and 26 µm MFD fibers with increased HOM loss respectively. These novel fibers achieve record pulse energies, over 3 times higher than previously demonstrated. In addition, the average power from these fibers is over a factor of 2 larger compared to any laser of this type with over 0.1 mJ pulse energy. Note that the higher energy data points in the 26 µm MFD fiber were obtained at lower repetition rates to stay below the TMI threshold of ∼1 kW.

The highest peak power reported in this work was 420 kW. There are reports of all-fiber ns lasers with higher peak powers. In [15], 670 kW was achieved by Meng, but the beam quality (M²= 1.6) was not diffraction limited, and the average power was low (63 W) in comparison to this work. In [16], Cai demonstrated a peak power of 1000 kW, but the beam quality was approximately 50 times diffraction limited. In [17], Fang achieved 700 kW peak power, but the M² was ∼3.5. To the best of our knowledge the LMA fibers with increased HOM loss described in this work have achieved the highest peak power of any all-fiber, diffraction-limited, ns pulsed laser.

In conclusion, increasing the HOM loss of LMA gain fibers opens an entirely new regime of operation with diffraction limited output from fiber lasers that simultaneously operate with very high pulse energy, peak power, and average power. In turn, these new fiber designs could enable new applications in the areas of micromachining, free-space communications, remote sensing and directed energy. With these new fiber designs, we have demonstrated nanosecond pulses with mJ pulse energy, kW average power, hundreds of kW peak power and diffraction limited output.