A simple solution for producing nanosecond laser pulses can be obtained using gain-switched fiber lasers. In this paper, we present an optimized single stage gain-switched ytterbium-doped fiber laser. Three fiber lengths were tested to show the impact of length on the laser output pulse. A pulse as short as 28 ns at 1.4 kW peak power and a maximum peak power of nearly 2 kW at 41 ns pulse duration was achieved. The laser possess a linear polarized output, very good beam quality of M2 < 1.1, and a spectral bandwidth of 0.11 nm.
© 2014 Optical Society of America
Pulse lasers in nanosecond regime are still very interesting for different applications. It is because such kinds of system are relative simple and robust, a very important for the industrial application. Most of nanosecond pulsed fiber lasers are based either on seed laser diode direct modulation, which requires a master oscillator power amplifier (MOPA) configuration, or a Q-switching approach. The latter is usually based on acousto-optic modulators , but can also be realized through alternative techniques such as the photoelastic effect and cavity dumping [2, 3]. However, in some cases, such as those in which a very high peak power is not required, a simpler design based on gain switching can provide the same functionality. In the most commonly used 1 µm wavelength range, pulse duration is one of the biggest obstacles for the implementation of the fiber laser gain switched approach. This is because in this approach, it is difficult to reach the desired 20−50 ns range appropriate for typical industrial applications such as microprocessing.
In general it is typical for laser operation that when the pump power is switched on, an oscillatory response of laser output is obtained. Usually, these relaxation oscillations consist of a series of spikes having amplitudes gradually decreasing with time to a steady state value. The desired gain switched approach utilizes only the first spike by switching off the pump power when the laser pulse (first spike) reaches its maximum and thus reducing the gain for subsequent oscillations. The pump switching must be fast and the pulse powerful enough to provide adequate pumping rate for the maximum peak of the relaxation oscillation. The laser pulse parameters dependence on the pump pulse energy are shown in . Therefore, this technique allows us to produce laser pulses in the nanosecond range without additional elements inside the resonator . Combining the technique with high power fiber lasers, which have superb thermal power management characteristics, form factor, stability, and pulse quality, a very robust pulsed laser system can be designed.
Gain switching has been implemented with nearly all rare earth-doped optical fibers from neodymium , erbium-ytterbium co-doped systems , thulium , and holmium . The shortest pulses were achieved in thulium fibers  for 2 μm laser light and were reported to be 10 ns. Thulium and holmium ion-based fiber lasers both have the disadvantage of a relatively complex pump sources. This limitation results in multistage and more complex systems  utilizing Q-switched fiber lasers in order to realize pulsed pumping. A variety of pump sources in the form of high power laser diodes are available for ytterbium-doped fibers. Most papers [12–15] reports on multi stage laser based on MOPA configuration to provide the appropriate peak power with relatively low pump pulse power (tens of watts) of the main oscillator and pulse lengths of over 100 ns. An improvement was made in an all-fiber designed ytterbium-doped gain-switched MOPA , in which pulses as short as 66 ns, with peak power of 700 W, were achieved from the master oscillator. Adding the amplifier single polarization narrow bandwidth output is reported with 160 ns pulse width and 1.4 kW peak power.
In this paper we demonstrate that with adequate optimization, the same or even higher peak power can be obtained within the oscillator itself and therefore using only a single stage laser. Our approach also enables the generation of much shorter pulses.
2. Experimental setup
The gain-switched single stage fiber laser system in the present work is shown in Fig. 1(a). This system is based on a polarization maintaining (PM) microstructured double clad Yb3+-doped active fiber shown in Fig. 1(b). The fiber was pumped by free space coupled 976 nm pumping diodes. Adequate temperature control was needed to match the pump spectra to the maximum of the ytterbium absorption peak. The out-coupler represents one end of the fiber that was straight cleaved to achieve 3.5% Fresnel reflection. At the opposite end of the resonator, selective feedback was provided by a polarizer and a wavelength selective reflective element. The parameters of the active fiber and resonator are shown in Table 1. Lengths (L) of the fiber used in the experiment were 53 cm, 120 cm, and 180 cm which correspond to the low signal absorptions 4.2 dB, 9.6 dB and 14.4 dB respectively.
A small part of the output laser power was detected using a fast photodiode to provide the signal for a gain switch feedback loop. The function of this loop is to control the pump pulse duration in order to enable only one pulse from the relaxation oscillation to appear and consequently to ensure stable laser operation. The repetition rate through all the experiment was kept constant at 50 kHz.
3. Description of gain switching
To theoretically predict the laser pulse characteristics and onset at given operational parameters, the mechanism of relaxation oscillations must be described, usually through rate equations . A discrete laser model using only differential rate equations in time to describe time and spatial dependencies presented in  can be adapted to this experimental setup. This approach divides the active fiber into several segments and uses an expanded homogenous model, which introduces segment coupling terms (), to calculate the laser photon change. Here τc = dz/c denotes the segment time of flight. Equations describing laser photons (ϕ) and number of excited ions (Ni) in each segment can then be written as:
The rate of pumping is marked as w(t,ϕ). Positive and negative sign superscripts of photon numbers describe their propagation in forward and backward directions, respectively. For a complete laser description, we must include boundary conditions for the photon equations:Fig. 2 for a 120 cm long fiber at 1030 nm laser wavelength. The parameters used in the model are presented in Table 1. The origin of the time scale of the measured and modeled result is set at the beginning of both pumping pulses. The modeled laser pulse has 4% late onset and 2% higher peak power in regard to measured one. However, in general, we obtained very good agreement between theoretical predictions and experimental results.
With existing fibers, the laser pulse duration and peak power can be changed by changing the length of the active fiber/resonator and by changing the pump peak power . As presented in  and , the approximate dependencies of laser pulse parameters on pump pulse power (Pp) are:
4. Results and discussion
Experimental results for three different fiber lengths are shown in Fig. 3. Solid lines in Figs. 3(a) and 3(b), represent fits of the basic approximate relations introduced in Eqs. (4) and (5), respectively. The relation between pulse duration and pump pulse power shows good agreement with the inverse square root dependence. Furthermore, the relation between the laser pulse peak power and the pump pulse power shows good agreement with a linear dependence. The shortest pulse was achieved at the highest pump pulse power and for the shortest active fiber. It is important to note that in Fig. 3, the horizontal axis represents pump pulse power coupled into the fiber. However, laser dynamics directly depends on the absorbed pump pulse power. The absorbed pump pulse power is always lower in our experiment than the power coupled into the fiber. This affected the optical to optical efficiency which was up to 50%. The actual difference between these two quantities depends on absorption and therefore on the fiber length.
This fact can help to explain the dependence of the laser peak power on the pump pulse power and fiber length, as shown in Fig. 3(b). If working at the constant input pump energy two competing effects are present. First the laser pulse peak power increases with the fiber length because of higher absorption and therefore more energy is available for each particular pulse till maximum absorption of the pump light is achieved. Second, the laser pulse duration also increases and thus its peak power must decrease to retain the energy balance. Providing the adaptive pump pulse duration the later can be avoided and peak power is kept the same as can be seen in Fig. 3(b). Here the measurements for a 120 cm and 180 cm long fiber differ in pulse duration but the peak power is almost the same.
However, in the present work, our goal was to achieve a pulse with duration of approximately 30 ns. Thus, we mainly focused on 0.5 m long fibers. We were able to achieve reliable and stable operation at a laser wavelength up to 1050 nm, while the best results were obtained at 1030 nm. The comparison between two wavelengths 1030 nm and 1050 nm was made experimentally and by using the numerical model. Both results are shown in Fig. 4 for a laser pulse duration and peak power at which the model (solid line) and measured data show good agreement.
For the shortest fiber length studied, a laser wavelength of 1050 nm produced longer pulses because of the smaller gain at this wavelength (i.e. smaller stimulated emission cross section). At wavelengths above 1050 nm, the gain is even lower and leads to unstable laser operation. For this short fiber laser, a pulse duration as low as 28 ns at a peak power of 1.4 kW was achieved at 1030 nm, as shown in Fig. 4(c).
The fiber laser with the pulse parameters described here can be used in micro-processing applications. In addition to pulse width and peak power, polarization, beam quality, and pulse-to-pulse energy stability are important parameters.
Fiber architecture also offered single mode operation. The measured beam quality in terms of M2 was less than 1.1, shown in Fig. 5. In addition, the spectral bandwidth and polarization extinction ratio of the laser output were measured, with results of Δλ = 0.11 nm and more than 15 dB, respectively.
Pulse-to-pulse stability was determined at two different pump pulse powers. The pulse trains are shown in Figs. 6(a) and 6(b) for pump pulse powers of 120 W and 320 W, respectively. In Figs. 6(c) and 6(d), every twentieth pulse is shown for each pulse train for comparison. The standard deviation of pulse energy is approximately 1% for both cases.
In this paper, a gain-switched laser system is presented for operation primarily at a laser wavelength of approximately 1030 nm. With the optimization of operational parameters, 28 ns pulses at a peak power of 1.4 kW were achieved from a 53 cm long microstructured polarization maintaining ytterbium-doped fiber. To our knowledge, this is the shortest pulse duration achieved in gain-switched ytterbium fiber lasers. The peak power can be increased to almost 2 kW using longer active fibers, but this leads to longer pulses (41 ns). The output beam was almost diffraction limited with M2 < 1.1. Furthermore, the microstructured polarization maintaining active fiber allowed a linear polarization of output laser light with polarization contrast greater than 15 dB. The stability of the laser was very good with a pulse-to-pulse energy stability of approximately ± 1%.
We thank NTK photonics for providing the active fibers used in this research. This work was supported by the L2-4174 project funded by the Slovenian research agency (ARRS).
References and links
1. M. Laurila, J. Saby, T. T. Alkeskjold, L. Scolari, B. Cocquelin, F. Salin, J. Broeng, and J. Lægsgaard, “Q-switching and efficient harmonic generation from a single-mode LMA photonic bandgap rod fiber laser,” Opt. Express 19(11), 10824–10833 (2011). [CrossRef] [PubMed]
4. M. Giesberts, J. Geiger, M. Traub, and H.-D. Hoffmann, “Novel design of a gain-switched diode-pumped fiber laser,” Proc. SPIE 7195, 71952P (2009). [CrossRef]
6. K. Hattori and T. Kitagawa, “Gain switching of waveguide laser based on Nd-doped silica planar lightwave circuit pumped by laser diodes,” IEEE Photon. Technol. Lett. 4(9), 973–975 (1992). [CrossRef]
11. Y. Tang, F. Li, and J. Xu, “High peak-power gain-switched Tm doped fiber laser,” IEEE Photon. Technol. Lett. 23(13), 893–895 (2011). [CrossRef]
12. S. Maryashin, A. Unt, and V. P. Gapontsev, “10-mJ pulse energy and 200 W average power Yb-doped fiber laser,” in Fiber Lasers III, (SPIE, 2006), 61020O61021–61020O61025.
13. T.-Y. Tsai, Y.-C. Fang, H.-M. Huang, H.-X. Tsao, and S.-T. Lin, “Saturable absorber Q- and gain-switched all-Yb3+ all-fiber laser at 976 and 1064 nm,” Opt. Express 18(23), 23523–23528 (2010). [CrossRef] [PubMed]
14. C. Larsen, D. Noordegraaf, P. M. W. Skovgaard, K. P. Hansen, K. E. Mattsson, and O. Bang, “Gain-switched CW fiber laser for improved supercontinuum generation in a PCF,” Opt. Express 19(16), 14883–14891 (2011). [CrossRef] [PubMed]
15. Y. Sintov, M. Katz, P. Blau, Y. Glick, E. Lebiush, and Y. Nafcha, “A frequency doubled gain switched Yb3+-doped fiber laser,” Proc. SPIE 7195, 719529 (2009). [CrossRef]
16. C. Larsen, M. Giesberts, S. Nyga, O. Fitzau, B. Jungbluth, H. D. Hoffmann, and O. Bang, “Gain-switched all-fiber laser with narrow bandwidth,” Opt. Express 21(10), 12302–12308 (2013). [CrossRef] [PubMed]
17. V. Agrež, F. Bammer, B. Podobnik, and R. Petkovšek, “Influence of the retardation of the multiplexing element in a dual channel Q-switched laser,” Appl. Phys. B 112(1), 73–81 (2013). [CrossRef]
19. D. G. Carlson, “Dynamics of a repetitively pump-pulsed Nd:YAG Laser,” J. Appl. Phys. 39(9), 4369–4374 (1968). [CrossRef]