1 Introduction

Passively Q-switched fiber lasers are attractive for many applications owing to their small form factor, excellent beam quality, and high efficiency. Previous efforts using Cr^4+:YAG as a saturable absorber (SA) have generated 25 μJ, 2.5 ns pulses, but require the use of free-space optics to couple into and out of fiber [1]. Bulk SA’s can be removed by using unpumped rare-earth doped fibers as the saturable absorbing elements. Passive Q-switching using Yb³⁺-doped gain fiber and Sm³⁺-doped fiber as the SA element has produced 19 μJ 650 ns pulses [2]. Similarly, an Er³⁺-doped fiber laser using unpumped Er³⁺-doped fiber as the SA and a Yb³⁺-doped fiber laser using unpumped Yb³⁺-doped fiber as the saturating element have produced 8 μJ 80 ns and 2.8 μJ 280 ns pulses, respectively [3,4]. However, the unfavorably low absorption and slow switching time of rare-earth doped SA fibers in the commonly used ASE bands of Yb³⁺ and Er³⁺ reduce the available unsaturated gain in the pumped fiber, leading to longer, less energetic pulses. Conversely, a quickly saturating SA with large absorption cross-section at the lasing wavelength produces pulses with higher energy and shorter duration, utilizing the large gain from the unsaturated gain fiber at the onset of Q-switching. We have effectively achieved faster switching times by coupling the ASE from a pumped large mode area (LMA) Yb³⁺-doped fiber to an unpumped, single-mode Yb³⁺-doped SA fiber. The comparatively high intensity of the ASE in the single-mode SA fiber bleaches the absorption, thereby making the SA fiber transparent, before the onset of gain depletion in the LMA gain fiber, resulting in short, energetic pulses.

In a previous study [5], we numerically demonstrated that pulse energies approaching 500 μJ with pulse durations as short as 13 ns at 1030 nm could be achieved by using a large mode area (LMA) Yb³⁺-doped gain fiber and SA fiber pair in an all fiber architecture. In this scheme, a large mode area (LMA) double-clad Yb³⁺-doped gain fiber is adiabatically tapered to match the mode field diameter (MFD) of an unpumped, Yb³⁺-doped single mode fiber SA. The LMA gain fiber is cladding pumped at 915 nm or 975 nm with the use of a pump combiner, while the cladding of the single-mode SA is mode-stripped to prevent the pump light from inverting the Yb³⁺ ions. The laser cavity is formed by 4% Fresnel reflection from the flat-cleaved LMA Yb³⁺-doped gain fiber and a highly reflecting (HR) fiber Bragg grating (FBG) spliced to the SA fiber. During initial stage of pumping, amplified spontaneous emission (ASE) generated in the LMA gain fiber propagates to the doped single-mode SA where it is strongly absorbed until a critical level of ASE power is reached, at which point the SA bleaches and becomes transparent at the lasing wavelength. The smaller MFD of the single-mode fiber SA enhances the intensity of the ASE relative to that in the LMA gain fiber, causing the SA to bleach quickly once the ASE power needed for transparency is reached. Conversely, the comparatively low intensity of ASE in the LMA gain fiber preserves large unsaturated gain. Consequently, a high level of unsaturated gain builds in the laser cavity prior to bleaching of the SA, producing a short pulse with high energy. Thus, introducing MFD scaling between the LMA gain fiber and the single-mode SA effectively increases the absorption cross-section of the SA, which in turn holds off larger gain at the onset of Q-switching.

We report an experimental demonstration of a 40 μJ 79 ns Q-switched oscillator at 1026 nm using a collimated free-space optical taper to couple between an LMA Yb³⁺-doped fiber and an unpumped single-mode fiber SA. We further demonstrate amplification to 0.40 mJ using an all-fiber master-oscillator power-amplifier (MOPA) system. In the oscillator, ASE was coupled between the disparate fibers using a 1.73: 1 telescope to nominally match their respective MFD’s to the spot sizes of the focused collimated beam. We adopted the free-space link to diagnose potential issues related to each component of the Q-switching laser cavity. This also afforded advantages from an experimental standpoint in that parameters such as SA length and wavelength could be varied easily without having to re-splice the SA fiber to a tapered fiber mode field adaptor or replace the fiber Bragg gratings. The principle purpose of this study is to map the parameter space encompassing wavelength, pulse pump time, and SA fiber length within the window of operation for a passively Q-switched LMA Yb³⁺-doped fiber with a single mode Yb³⁺-doped SA absorber. However, it is important to note that the free-space link can be easily replaced with well-developed, commercially available tapered fiber mode field adaptors for real applications.

The experimental set-up is described in Section 2 followed by the experimental results for different parameter studies in Section 2.1 – 2.3. Section 2.1 examines the dependence of pulse width and energy on wavelength for a fixed SA length and pump time. Energy and pulse duration are measured against pump time for a fixed SA length and wavelength in Section 2.2. Finally, the effect of SA length on energy and pulse width is investigated for fixed wavelength and pump time in Section 2.3. While no single study is exhaustive, collectively the set provides a roadmap for maximizing pulse energy and reducing pulse duration. A comparison of the experimental results to numerical simulations using conditions identical to those of the individual experiments is detailed in Section 2.4. In Section 2.5 we report energy scaling in a master oscillator power amplifier (MOPA) configuration using a LMA Yb³⁺-doped fiber amplifier pumped at 975 nm. Finally, a discussion and summary of our results, including observations concerning experimental trends, is given in Section 3 followed by concluding remarks in Section 4.

2. Experimental Set-up and measurements

The experimental set-up for the tapered Q-switched fiber oscillator is depicted in Fig. 1 . A 30 cm piece of double clad (LMA) nLight YB1200 20/125 DC fiber (20 µm diameter core, 125 µm cladding, nominal 1200 dB/m absorption in core at 976 nm) is end-pumped by four IPG PDL-30 915 nm 100 mm fiber-coupled diode pumps spliced to a 6 by 1, 100 µm fused fiber bundle (Oz Optics). The total pump power available was 70 W. The ASE exiting the gain fiber opposite the pump side is collimated by an aspheric lens with a 8 mm focal length and launched into a unpumped nLight Yb1200 6/125 fiber (6 µm diameter core, 125 µm cladding) via reflection from two successive 980 nm short pass dichroic mirrors to remove unabsorbed 915 nm light and an aspheric coupling lens with a 4.6 mm focal length. The left-hand side of the LMA gain fiber is flat-cleaved to provide a 3.5% output coupler, while the righ-hand side is angle-cleaved with a 12 degree angle to prevent spontaneous lasing prior to Q-switching. Despite taking this precaution we found that it was necessary to keep the length of the LMA gain fiber below 35 cm to keep the unsaturated gain small enough to avoid spurious lasing. The ASE is coupled out of and back into the YB1200 6/125 single-mode fiber with the aid of a second 11 mm aspheric coupling lens. A 1200 grooves/mm bulk diffraction grating tuned the operating wavelength. The laser cavity is formed by the diffraction grating and the 3.5% Fresnel reflection provided by the flat cleave on the pump side of the LMA gain fiber. The MFD of the fundamental mode of the LMA gain fiber is approximately 14 µm, where as the MFD of the single-mode saturable absorber is 6.1 µm, resulting in a mode area scaling of 5.4: 1. The fiber was bent to an approximate radius of 9 cm to favor oscillation of the fundamental mode.

 

Fig. 1 Experimental set-up of tapered Q-switched fiber oscillator.

Download Full Size | PDF

The beam quality was measured using a second moment fitting routine built into the software for a DataRay WinCam - UCD12 infrared camera / M2DU translation stage system. The measured beam quality exiting the fiber oscillator from the LMA fiber was M² < 1.1. The coupling efficiency of the free-space taper from the LMA gain fiber to the single-mode saturable absorber under non-lasing conditions was experimentally determined to be 60%. A portion of the output was sampled using an 8° wedge optic to split the train of pulses into two beams to measure the temporal and spectral profiles. The spectra were recorded using a fiber-coupled ANDO AQ6317B optical spectrum analyzer set to a resolution of 0.05 nm, while the temporal profiles were measured using a Thorlabs DET10A silicon detector with a 1 ns rise time interfaced to a 2 GHz Tektronix DPO 7254 oscilloscope. Pulse energies were measured with Coherent J10MB-LE and J25MT energy detectors.

We identified three important parameters impacting laser performance: the SA fiber length, the pumping time, and the wavelength. Three sets of experiments were performed to understand the impact of these parameters on laser performance. The first experiment examined the dependence of pulse energy and pulse width on the saturable absorber length, for a fixed wavelength, gain fiber length, and pumping time. In the second experiment, the pulse energies and pulse widths for a fixed saturable absorber length and pumping time were measured against wavelength. Finally, the dependences of pulse energy and pulse width on pumping time for fixed SA fiber length and wavelength were investigated. As discussed in the next section, conditions which favored high absorption in the SA fiber (shorter wavelengths, longer saturable absorbers lengths) and short pump times yielded pulses with the highest energies and shortest durations. Energy scaling up to 400 µJ was demonstrated by injecting the output pulses of the passively Q-switched fiber oscillator into a 1.0 m YB1200 20/125 fiber amplifier end-pumped by a 975nm fiber-coupled diode. All experiments were conducted at 50 Hz to allow adequate time (20 ms) for residual inversion remaining in the SA fiber after each Q-switched pulse to relax back to the ground state. It is important to note that one can expedite relaxation of the excited ions of the SA fiber back to the ground state by forming an additional long-wavelength cavity surrounding the SA fiber [5]. This additional long wavelength cavity enables higher repetition rates without altering the pulse energy or the pulse duration.

Our previous paper [5] noted that excessively long pump duration may cause an after-pulse event. If the pump is not turned down after the giant pulse event, the passively Q-switching laser cavity generates smaller pulses as the gain fiber continues to accumulate energy. These small energy after-pulses are not desirable in many applications requiring single-shot performance, such as remote detection and laser machining. Therefore, for such applications, a pulsed pumping scheme with suitable pumping duration is needed to avoid after-pulses. For every data point reported in the following plots the pump power was first increased until Q-switched pulses were observed within the set pump duration. The pump power was then incrementally increased until just before the onset of a second, weaker after-pulse. This procedure yielded single Q-switched pulses per pumping event with the highest energies and shortest pulse durations for any given combination of SA length, pump time, and wavelength.

2.1 Pulse energy and pulse width vs. saturable absorber length

To examine the dependence of laser performance on saturable absorber length, the grating was tuned to 1026 nm and 31.4 cm of YB1200 20/125 DC fiber was pumped by quasi continuous wave (QCW) pulses at 50Hz, with a pump on-time of 100 µs, from four simultaneously triggered 915 nm fiber-coupled laser diodes. The measured pulse energies and FWHM pulse widths for various SA fiber lengths are plotted in Fig. 2(a) . The corresponding temporal profiles of the pulses are plotted in Fig. 2(b). We find that the pulse energy increases while the pulse width decreases with increasing saturable absorber length. Specifically, the pulse energy increased from 5.5 μJ to 39.5 μJ while the pulse width decreased from 342 ns to 79 ns as the length of the YB1200 6/125 single-mode SA fiber increased from 3.3 cm to 41.7 cm. Again, for these data points, the pump power was optimized to produce only a single pulse with the highest pulse energy. Specifically, the launched (peak) pump power changed from 29 W for the 3.3 cm SA fiber to 43 W for the 41.7 cm SA fiber. The corresponding absorbed 915 nm pump power required to Q-switch the device increases linearly from 12.2 W to 17.7 W. Such a result is expected since longer SA fibers require higher ASE intensities to become transparent. In addition, longer SA fibers increase the hold-off time before the onset of Q-switching, allowing for larger accumulated inversion in the gain fiber which in turn reduces the energy extraction time and, thus, the pulse width. In Fig. 2(a) it is apparent that the pulse energy at 1026 nm starts saturating, or rolling over, for SA fiber lengths in excess of 30 cm. We attribute this to the onset of spurious lasing, which starts depleting the gain. Indeed, SA fiber lengths in excess of 41.7 cm could not see the giant pulse event, due to an earlier spurious lasing without a saturated (transparent) SA. This observation implies further reduction of the reflectivity at the angle-cleaved gain fiber facet would produce output pulses with higher energies and shorter pulse widths. On the other hand, it will be shown in Section 2.2 that increasing the pump power and decreasing the pump time promotes higher inversion at the onset of Q-switching, producing pulses with higher energies and a shorter pulse durations. It is also noteworthy that for SA lengths above 40 cm the splice joint of the SA fiber endcap facing the tapered cavity would routinely shatter, indicating that the gain in the gain fiber becomes excessively high and triggers a spurious lasing event from the non-zero reflection at the angle-cleaved gain fiber end.

 

Fig. 2 (a) Pulse energy, FWHM pulse width vs. YB1200 6/125 fiber SA length. 30 cm YB1200 20/125 gain fiber, 50 Hz rep. rate. (b) Pulse temporal profile for different SA lengths. 22 cm Yb1200 20/125 gain fiber, 50 Hz rep. rate.

Download Full Size | PDF

We observed 40 μJ, 79 ns pulses at 1026nm using 41.7 cm of YB1200 6/125 fiber as the SA and 100 µs-long, 915 nm QCW pump pulses, with 43 W launched pump power and a 17.7 W absorbed pump power. These results match the pulse width reported in [3] using mode-mismatched Er^3+-doped fibers as the SA and gain fibers, and exceed the energy by a factor of five. Similarly, for passively Q-switched Yb³⁺-doped fiber lasers operating in the 1020 nm– 1090 nm band using Yb³⁺-doped fiber as the saturable absorber, these results are a five-fold improvement in pulse duration and a forty-fold improvement in pulse energy [4]. The 460 W of peak power reported here is therefore more than two orders of magnitude larger than that reported in [4] using an Yb³⁺-doped fiber as a saturable absorber. Further advances in pulse width and energy may be possible by increasing the SA fiber length, decreasing the pumping time, and increasing the pump power.

2.2 Pulse energy and pulse width vs. wavelength

The second parameter investigated was wavelength for fixed SA fiber length and pumping time. The wavelength was tuned by rotating the diffraction grating depicted in Fig. 1 about the vertical axis. In Fig. 3(a) the pulse energy and (FWHM) pulse width are plotted against wavelength for a fixed saturable absorber length of 22 cm and a pump time of 100 µs. Similarly, the pulse width and corresponding absorbed 915 nm pump power versus wavelength is shown in Fig. 3(b). As readily observed in Fig. 3(a), the pulse energy increases and the pulse width decreases as the wavelength is tuned from 1040 nm to 1020 nm. This is commensurate with a ~4 fold increase in the absorption cross-section of the Yb³⁺-doped saturable absorber fiber over the same wavelength span [6]. This correlation can intuitively be understood in that shorter wavelengths with larger absorption cross sections can hold off gain more effectively than longer wavelengths with smaller absorption cross sections, allowing for higher gain in the actively pumped gain fiber prior to Q-switching. Please note that a higher gain, coupled with a relatively short pump time (100 µs), produces shorter pulses with higher energies. This statement is supported by Fig. 3(b) where it is observed that higher levels of absorbed pump power are required to switch the SA fiber at shorter wavelengths where the absorption is high, producing comparatively shorter pulses with larger energies. Conversely, at longer wavelengths where the absorption cross-section is smaller, lower levels of inversion in the SA fiber are needed for transparency and, thus, for switching the cavity to a high-Q state. This in turn prevents build up of high gain in the pumped LMA fiber prior to Q-switching, yielding relatively long weak pulses. Again, Fig. 3(b) supports this analysis at longer wavelengths. Figure 4 shows the measured spectrum of a 1025.5 nm pulse sampled on an ANDO fiber-coupled optical spectrum analyzer with a resolution of 0.05 nm. The 0.50 nm bandwidth observed at this wavelength was typical of other recorded wavelengths and well above transform- limited.

 

Fig. 3 (a). FWHM pulse width, pulse energy vs. wavelength. (b) FWHM pulse width, absorbed 915nm pump power vs. wavelength. 30 cm YB1200 20/125 DC fiber, 22 cm YB1200 6/125 saturable absorber fiber. Q-switch rep. rate = 50 Hz.

Download Full Size | PDF

 

Fig. 4 Spectral profile of 1025.5nm, 140 ns pulse. 30 cm YB1200 20/125 DC fiber, 22 cm YB1200 6/125 saturable absorber fiber. Q-switch rep. rate = 50 Hz.

Download Full Size | PDF

As observed in Fig. 3(a), the FWHM pulse width decreased from 365 ns to 96 ns as the wavelength was tuned from 1041nm to 1020nm, commensurate with an increase in pulse energy from 5 μJ to 20 μJ. While no definitive trend in the pulse width can be ascertained at the shortest wavelengths (1025 nm – 1020 nm), it is apparent that the pulse energy begins to roll over below 1025 nm. This is caused by the sharp reduction in the emission cross-section below 1025 nm (relative to the emission peak at 1030 nm) coupled with a steep increase in the absorption cross section. Any improvement in the capacity of the saturable absorber to hold off the gain at shorter wavelengths is nullified by the smaller net gain of the LMA fiber below 1025 nm.

Nevertheless, reducing the pump time and increasing the pump power yields higher energy, shorter pulses by achieving higher inversion in the gain fiber before the ASE saturates the single-mode saturable absorber fiber. However, the available power of the 915 nm pump diodes and the minimum pump time requirements of the diode driver used limit further improvements in pulse energy and pulse width.

2.3 Pulse energy and pulse width vs. pump time

Finally, the pulse energy and duration were measured against pumping time for fixed SA fiber length and wavelength. In Fig. 5 the pulse energy and pulse width versus pumping time are plotted for a 1030 nm passively Q-switched laser using a 30 cm YB1200 20/125 LMA gain fiber and a 14 cm of YB1200 6/125 SA fiber. The pump power is adjusted for each set pump-on time to maximize the single pulse energy. Shorter pumping times would have been possible with higher available pump power. The FWHM pulse width decreases steadily from 235 ns to 154 ns as the 915 nm pumping time is reduced from 300 µs to 60 µs. Conversely, the energy increases monotonically over the same the range of pumping times. Although not shown in Fig. 5, it should be noted that the absorbed pump power needed for Q-switching increases from 5.6 W for a pump time of 300 µs to 22.4 W for a pump time of 60 µs, implying that slightly more pump energy is required for longer pumping times (1.68 mJ for 300 µs vs. 1.34 mJ for 60 µs).

 

Fig. 5 Pulse energy, FWHM pulse width vs. 915 nm pump time. 30cm YB1200 20/125 DC fiber. 14 cm YB1200 6/125 saturable absorber fiber. Rep. rate = 50Hz. Pump power decreases from 22.4 W to 5.6 W as pump time increases from 60 μs to 300 μs.

Download Full Size | PDF

There are two mechanisms for the gain depletion: (1) spontaneous emission and (2) ASE build-up from spurious reflections. Longer pumping time promote spontaneous radiation decay. It is apparent that the gain of a fiber pumped with a high power short period QCW pulse would suffer less from depletion due to spontaneous emission. Also, the non-zero reflection from the angle-cleaved gain fiber end may facilitate gain depletion before the onset of Q-switching. Nonetheless, it is apparent that higher pump powers are required for shorter pumping times as expected for a passively Q-switching process where a critical level of stored energy and ASE is needed to trigger (bleach) the saturable absorber. Unlike the wavelength-dependent measurements reported in Section 2.2, there is no evidence of rollover in either the pump energy or pulse width for pump times as low as 60 µs. This implies that higher energy, shorter pulses may be achieved using higher pump powers and shorter pump times than those reported here. Thus, reducing the pump time would appear to be an effective means of simultaneously improving the pulse energy, reducing the pulse width, and increasing the available rep. rate.

2.4 Comparison to numerical simulations

In order to compare experimental results with theory, we performed a numerical simulation based on a model presented in our previous paper [5]. The details of the simulator can be found in the same reference. The doping concentration of both the gain fiber and the SA fiber (nLight Yb1200) were estimated by measuring the small signal pump absorption in the cladding. The result was Yb³⁺-ion concentration of 1.07 × 10¹⁹ ions/cm³. We adopted the same emission and absorption cross-section spectra used in [5]. The lifetime of the Yb³⁺-ion was assumed to be 830 µs.

To compare numerical simulations with experimental results, we simulated the measurement shown in Fig. 2(a) where the SA fiber length was varied for fixed pumping time and wavelength. For this simulation, we performed a least square fit for two parameters: the coupling efficiency between the gain fiber and the SA fiber and the reflectivity from the diffraction grating reflector. We obtained the best fit between the simulation and the measured data for the pulse energy and the pulse width, for various SA fiber lengths, with a 43.3% coupling efficiency between two fibers and 55% for the diffraction grating’s reflectivity. Clearly the optimum coupling efficiency 43.3% deviates from the measured value of 60%. Still it may be possible that the sensitive alignment of the core-to-core free-space coupling may have drifted during the two experiments. Figures 6(a) and 6(b) show the comparison between the experimental and the simulation data. Both the energy and the pulse duration show excellent agreement between the simulation and the measured data. However, when the SA fiber length becomes long, the simulated energy is clearly larger and the simulated pulse duration is clearly shorter than the experimental measurements. We believe this deviation may have been caused by gain depletion as explained in Section 2.2. A long SA fiber forces higher inversion in the LMA gain fiber, the latter of which can then build up a significant ASE via spurious reflection from the non-ideal non-zero reflection of the angle-cleaved gain fiber end. This depletes the gain, resulting in less energy and longer pulse duration. Nevertheless, the agreement between the simulation and the experiment is excellent, demonstrating that our theory and the numerical simulator presented in [5] is quite accurate.

 

Fig. 6 Comparison of simulation and experiment (a) Energy vs. SA fiber length. (b) Pulse width vs. SA length.

Download Full Size | PDF

We performed a second study using our simulator to determine the importance of the coupling efficiency between the gain and the SA fiber. For this, we fixed the saturable absorber length to 35.7 cm. All other parameters are the same as those reported in the previous simulation. The result shown in Fig. 7 clearly shows that output energy is significantly impacted by the coupling efficiency. Poor coupling efficiency induces loss in the laser cavity. This is particularly problematic for Q-switched lasers since the output energy is directly related to the unsaturated gain at the onset of Q-switching and to the power injected into the gain fiber. Thus, the coupling loss between the SA fiber and LMA gain fiber reduces the power launched into the highly inverted LMA fiber and lowers the pulse energy. Figure 7 clearly shows the proportionality between the coupling efficiency and the output energy. This illustrates the importance of coupling efficiency between the gain and the SA fibers. The method used in [3] and [4] utilizes a simple mismatched coupling between two fibers that significantly reduces the output energy.

 

Fig. 7 Simulated parameter study for output energy vs. coupling efficiency between gain and SA fiber.

Download Full Size | PDF

2.5 Amplification of pulses

Pulse energy levels were improved by injecting the Q-switched pulse train from the passively Q-switched oscillator into an end-pumped 1.0 m YB1200 20/125 DC fiber amplifier pulse-pumped at 975 nm. A schematic of the experimental set-up is given in Fig. 8 . 1030 nm, 125 ns, pulses produced by the passively Q-switched Yb³⁺-doped fiber oscillator were launched into the DC fiber amplifier and isolated from the seed source by a bulk Faraday rotator isolators. The input energy after passing through the isolator was 15.5 μJ. Although neither the LMA gain fiber nor the single mode saturable absorber were polarizing maintaining (PM) it was experimentally observed that the pulses produced by the Q-switched seed laser were polarized with an extinction ration of >10 dB. This favorable PM property could be explained by the polarization-dependent reflectivity of the diffraction grating reflection. The standard deviation in energy measured over 1000 pulses after passage through the isolators was less than 0.25% which, as will be shown, resulted in negligible shot-to-shot variation in energy of the amplified pulses. The YB1200 20/125 DC gain fiber was pumped with 1 ms, 975 nm QCW pulses from a wavelength stabilized 40 W QPC Inc. BrightLock 100 mm fiber-coupled diode laser at 50 Hz. The measured coupling efficiency into the 125 mm double-clad gain fiber was 90%. To build inversion prior to the arrival of the pulse, the LMA gain fiber was pumped 890 µs before the arrival of each pulse into the fiber amplifier. The energy, temporal profiles, and spectral profiles of the pulses were measured as described in Section 2.1.

 

Fig. 8 Experimental set-up of Q-switched fiber MOPA, using a fiber SA.

Download Full Size | PDF

In Fig. 9 the amplified pulse energy is plotted against the launched 975 nm power. Below 46 μJ of output energy and 10 W of launched pump power, the energy rises quasi-exponentially, indicating that the input pulse energy is below the saturation fluence of the LMA amplifier. Between 46 μJ and 346 μJ (9.0 to 21.5 W launched pump power), the inversion near the input side of the fiber amplifier becomes large enough to amplify the pulses to energies suitable for depleting the gain in the remainder of the fiber. Consequently, the energy vs. pump power is linear in this regime. Finally, above 346 μJ, or 21.5 W of launched pump power, the extractable energy of the amplifier fiber is limited, resulting in energy roll-off. Work is in progress to optimize the extractable energy by adjusting the amplifier fiber length.

 

Fig. 9 Amplified pulse energy vs. launched 975 nm power. 1.00 m YB1200 20/125 DC fiber amplifier. Input pulse energy and duration = 15 µJ and 125 ns, respectively. Rep. rate = 50 Hz.

Download Full Size | PDF

3 Discussion

From the results of Sections 2.1, 2.2, and 2.3, there appear to be three parameters which can be tuned to increase pulse energy, decease pulse width, and therefore increase peak power in passively Q-switched Yb³⁺-doped fiber lasers utilizing a doped SA fiber. Two of the parameters, the SA fiber length and the operating wavelength, are similar in that the energy-to-pulse width ratio can be improved by tuning both to favor higher absorption in the SA fiber. Shorter wavelengths, particularly below 1040 nm, are strongly absorbed by the ³F_7/2 – ³F_5/2 transition in Yb³⁺-ions in fused silica, and thus can better hold off the gain in the LMA Yb³⁺-doped fiber than longer wavelengths for a given SA fiber length. Similarly, for a given wavelength the absorption can be increased by lengthening the SA fiber, again improving the capacity of the SA to hold off of the gain. It is important to note that the SA fiber length can be manipulated to improve the energy-to-pulse width ratio at longer wavelengths where the absorption cross-section is comparatively small. As an example, the energy-to-pulse width ratio of the 5 μJ, 365 ns, 1041 nm pulses reported in Fig. 3(a) for a 22 cm SA could be markedly improved by simply increasing the SA fiber length to improve absorption at 1041nm. As such, high energy (~40 μJ), short duration (~80 ns) pulses should be achievable over the commonly used 1020 nm – 1100 nm spectrum of Yb³⁺-doped fused silica fibers. The third parameter - pumping time - is particularly intriguing because, in contrast to the SA fiber length and operating wavelength, the energy-to-pulse width ratio does not appear to saturate or roll over as the pump time decreases (Fig. 3(a)), at least for pump times as low as 60 µs and a SA fiber length of 14 cm. This suggests that further improvements in pulse energy and pulse width are possible by using high power diodes with short pumping times.

Finally, the MFD ratio of the LMA gain and single-mode SA fibers can be altered to favor high energy, short pulses. As discussed in Section 1, the switching time of the SA fiber depends on the intensity, and therefore the mode field area, of ASE propagating in the core, with higher intensities producing faster switching times and shorter pulses with higher energies. Also, large unsaturated gain requires a large core diameter in the LMA gain fiber to reduce the intensity of ASE propagating in the core. Therefore, the pulse energy-to-pulse width ratio could be improved by increasing the MFD of the LMA gain fiber, decreasing the MFD of the SA fiber, or doing both, with care being taken that the intracavity peak power does not exceed the threshold for Stimulated Brillion Scattering in the single mode SA fiber. Although not performed in this study, an investigation of the dependence of the pulse behavior on the ratio of the mode field areas of the gain and SA fibers would be a worthy and informative effort of future research.

It is important to note that the free-space optics used in this set of experiments can be simply replaced with a commercially available fiber components, incorporating all the key elements into a monolithic, all-fiber architecture. Specifically, free-space end-pumping could be replaced by a fused fiber bundle, the free-space mode field adaptor by a commercially available fiber taper, and the bulk diffraction grating by a fiber Bragg grating, making for a compact, light weight, easily deployable, robust fiber laser system. Furthermore, the fiber amplifier described in Section 2.5 could be spliced directly to the passively Q-switched fiber laser with a fiber-coupled isolator between the two stages to make a 400 μJ all-fiber MOPA system with no aligning optics.

As noted in Section 2, all measurements were performed at 50 Hz to allow adequate time for any residual inversion in the SA after the Q-switching event to revert back to the ground state. The low duty cycle is necessitated by the relatively long upper state lifetime (~800 μs) of the ²F_5/2 manifold in Yb³⁺-doped fused silica. However, 50 Hz is not an upper bound for the rep. rate that can be achieved in a passively Q-switched Yb³⁺-doped fiber. As noted in [5] and [7], excess inversion in the SA fiber after Q-switching can be quickly removed by incorporating an additional laser cavity resonate at longer wavelengths which surrounds the SA fiber. It should be noted that the additional cavity should be optically separated from the primary cavity through WDMs. Otherwise, the gain in the LMA gain fiber may directly interact with the longer wavelength reflector from this additional cavity. This efficient relaxing process for the inverted SA fiber is facilitated by the high emission-to-absorption cross-section of Yb³⁺-ion at 1100 nm. Our previous numerical study [5] suggested that 200 kHz rep. rates are possible using this scheme to remove excess inversion in the SA fiber. Such configuration is under progress and will be published elsewhere.

4. Conclusion

We have demonstrated tunable passive Q-switching of a Yb³⁺-doped fiber laser using a large mode area Yb³⁺-doped fiber as the gain medium and un-pumped, single-mode Yb³⁺-doped fiber as the saturable absorber. Under optimal conditions for wavelength, SA length, and pump time we demonstrated 40 μJ 79 ns pulses at 1026 nm. We successfully generated pulses from 1020 nm to 1040 nm and noted that further improvements in bandwidth are possible over the commonly used 1020 nm - 1090 nm ASE band of Yb³⁺-doped fused silica by increasing the SA length. Previously, it was believed that the doped fiber SA would not produce short, high-energy pulses since the absorption cross-section in fused silica is small relative to conventional bulk SAs. Operating at longer wavelengths, which allows energy scaling through the cladding-pumping technique, was deemed to be even more difficult due to the unfavorably small absorption cross-section at longer wavelengths. We overcame this difficulty by scaling the mode field diameters between the LMA gain and single mode SA fibers. This, in turn, enabled the generation of cladding-pumped high-energy, short pulses in an all-fiber passively Q-switch configuration.

We conducted a parametric study of the SA fiber length, lasing wavelength, and pump time to determine critical trends in performance as these parameters were varied. Our findings support the well-understood importance of using a strongly absorbing, quickly bleaching SA as the switching element. We constructed such an SA by using relatively long SA fiber lengths to compensate for the low absorption-to-emission cross section ratio of Yb³⁺-doped fiber in the 1020 nm – 1090 nm band and utilizing mode scaling between the gain and the SA fibers, to quickly bleach the SA fiber. We found that for a given SA fiber length, shorter wavelengths yielded pulses with higher energies and shorter durations relative to longer wavelengths, but that comparable high energy short pulses are possible at longer wavelengths by simply increasing the SA fiber length. Similarly, fast pump times produced more energetic pulses with shorter pulse widths than longer pump times, with no apparent roll over in energy or pulse duration for pump times down to 60 μs. While all measurements were conducted at 50 Hz to allow for complete relaxation of the inversion between pulses, we predict that repetition rates in excess of 200 kHz can be achieve through the use of an additional long wavelength (1100 nm) laser cavity surrounding the SA fiber to efficiently relax the SA ions between successive pulses.