1. Introduction

Q-switching operation is a useful technique employed in laser systems to produce short and high-intensity light pulses. There has been of great interest in Q-switched fiber lasers because of the advantages, including high efficiency, flexibility, compactness and high spatial beam quality. A Q-switched fiber laser with a smaller core diameter could directly provide a high intensity-density output of 0.1-10 MW/cm² without external focusing components. Such a high intensity density is quite useful for micromachining, nonlinear optics studies and biomedical applications. Like most of the conventional bulk Q-switched lasers, Q-switched fiber lasers have been realized using similar active [1, 2] and passive [3-5] bulk Q-switches. These fiber lasers contain free-space sections in the resonators, and require sophisticated techniques of alignment for pump coupling and in-and-out light coupling between fibers, Q-switches and mirrors. Large-core double-cladding (DC) gain fibers and multimode laser diodes of several wattages are often used to ease the difficulty of pump coupling, and achieve great gain of tens -hundreds dB that can suitably compensate the fiber coupling loss.

Lately an increasing attention has been drawn upon the all-fiber Q-switched lasers that require no alignments and have inherently low cavity losses. An all-fiber laser is generally a single-mode system pumped with a single-mode laser diode of hundreds of milliwatts. The laser is composed of commercial fiber components as WDMs, FBGs, fibers and fiber-pigtailed components of low insertion losses. Although an all-fiber system has limited CW output power for the small core diameters and the low-power LD pump, pulses with comparably high intensity densities can still be achieved in an all-fiber Q-switched laser with a relatively low price.

A few all-fiber actively Q-switched lasers have been realized using acousto-optic modulators [6-9], piezoelectric (PZT) actuators [10, 11] and magnetostrictive transducers [12]. A more compact and economic pulsed laser can be obtained using a saturable absorber Q-switch (SAQS). Nevertheless, because of the difficulty of finding fiber-type SAQSes, most of the literatures regarding all-fiber SAQS lasers were works of theoretical modeling and simulation [13, 14]. A low efficient and yet stable self-pulsed all-fiber erbium laser was reported in 2004 [15]. The mechanism of self Q-fluctuation was attributed to the thermal lensing effect resulting from the excited state absorption of erbium. An all-fiber SAQS ytterbium laser was reported by Foriadi et al. in 2005 [16]. In that study, a DC Yb³⁺-doped fiber was pumped by a 15W 976-nm LD and passively Q-switched at 1085 nm with a Sm³⁺-doped SAQS fiber. Peak pulse power of 30 W with pulse width of 0.65 μs was obtained at the repetition rate of 130 kHz.

In this paper, we propose a simple scheme of an all-fiber SAQS ring laser and establish the location-dependent rate equations to keep track of the non-uniform distributions of the resonant photons, the gain population inversion in the gain fiber N_g and the absorption population is the SAQS fiber, N_a. The design was numerically and experimentally demonstrated using single-mode Er³⁺-doped fiber at the emission wavelength of 1550 nm. More than 90% extraction efficiency of N_g by a Q-switched pulse was obtained. It is important to note that the proposed design is applicable for self Q-switching performances of all the 3-level laser materials that can also serve as 2-level saturable absorbers at the emission wavelength.

2. Modeling and Simulation

Figure 1 shows the schematic design of an all-fiber SAQS ring laser. The gain region z_g is for the gain fiber from z_g1 to z_g2, and the absorber region z_a for the absorber fiber from z_a1 to z_a2. The circulator and the FBG determine the roundtrip direction of lasing from z_g1 to z_g2, through the circulator, z_a1, z^a2, reflected by the FBG, back to z_a1, through the circulator, the WDM, and then starting from z_g1 again. In the later experiment, the gain and the absorber were the same type of erbium-doped fiber. The 980-nm pump is coupled by the WDM into the gain fiber and blocked by the circulator. Thus, the absorber fiber is left unpumped. It is reasonable to expect that the photon density in the absorber fiber should be on average twice (or more than twice) that in the gain fiber. The more intense photon density would result in a fast bleaching of the absorber, and then lead to a passive Q-switching performance.

 

Fig. 1. Schematic design of a passively Q-switched all-fiber ring laser

Download Full Size | PDF

In most cases of dealing with laser dynamics, an assumption is commonly made in the rate equations that the photons, the gain population, the absorption population and the cavity losses are distributed uniformly in the resonator [17]. This assumption is usually appropriate for standing-wave lasers. However, it is evidently not applicable for the proposed ring laser. To take account of non-uniformity, the photon densities, line densities of gain and absorption populations are defined to be functions of z location, indicated in Table 1.

Table 1. Definitions of the variables and parameters in simulation

View Table

In simulation, time is scaled by t=k×t_r, where t_r is the roundtrip transit time and k is a positive integer. In every time slice, N_g(z) and N_a(z) are solved iteratively by the rate equations:

The pump rate and the population relaxation are negligible in the Q-switching duration and ignored in the equations. In every time step, once N_g(z,k) and N_a(z,k) are obtained, n(z,k) is calculated accordingly along the propagation direction by

The simulation starts from the threshold condition when the gain is equal to the total loss. Before lasing (t<0), the pump intensity I_p(z_g) and the corresponding N_g(z_g) can be numerically solved for the initial condition:

The other initial variables and parameters are n_g(z_g1,0)=1×10², A=1.26×10^-7 cm², t_r=18 ns, σ_g=σ_a=5×10^-21 cm² and N_T=1.38×10¹⁹ cm^-3. The parameters are based on the characteristics of the erbium fiber employed in the experiment. The transmission losses of 0.6 dB are assumed (i.e. α₁=α₂=0.1382 correspondingly). The reflection loss of the FBG output coupler is 0.6 dB. The length of the absorber is 10 cm and that of the gain fiber is 70 cm. The simulation result is shown in Fig. 2. A passively Q-switched pulse has peak power of 1.72 W and pulse duration of 198 ns. If the output beam diameter is 10 μm, an output intensity density of 2.2 MW/cm² is obtained. Three temporal points t_a, t_b and t_c are chosen for later observation, when n_ai(z_a2, t_b) reaches the pulse peak, and n_ai(z_a2, t_a)=n_ai(z_a2, t_c)=n_ai(z_a2, t_b)/100. Figure 3 shows N_g(z_g) and N_a(z_a) at the moments t_a, t_b and t_c, normalized by the initial N_g(z_g1,0) and N_a(z_a2,0). It is clear that the SAQS fiber is saturated faster than the gain fiber, and well bleached at the end of the pulse. The saturation intensity of an erbium fiber with a core diameter of 4 μm is about 0.5 mW at 1550 nm. The inner peak intensity through the SAQS is 25 W. Thus the SAQS is about 10% saturated at the observation point t_a when the passing intensity reaches 0.25 W, about 90% saturated with 25 W and completely saturated after the pulse. The extraction efficiency of the total gain population (N_g integral over z_g) can be calculated to be 0.93, indicating a highly efficient Q-switching performance.

 

Fig. 2. Pulse simulation of a passively Q-switched all-fiber ring laser.

Download Full Size | PDF

 

Fig. 3. Saturations of N_g(z) and N_a(z) by the pulse, observed at the beginning, the top and the end of the pulse. The time points are marked with a, b, c in Fig.2.

Download Full Size | PDF

In the simulation, the pump is ignored during the 20-μs Q-switching action, and a Q-switched pulse occurred passively. After Q-switching, the recovering time of N_g is determined by the relaxation lifetime τ_g2, and that of N_a is by τ_a2. For achieving next Q-switching and preventing from free-running lasing, Na needs to recover faster than N_g. Thus when τ_g2 is close to or smaller than τ_g2, modulated pump is required for sequentially Q-switched pulses.

3. Experiment

The experiment was arranged as depicted in Fig. 1. The erbium fiber used for the gain and the SAQS in the experiment had an absorption loss of 30 dB/m at 1550 nm, and a core diameter of 4μm. The length of the absorber was 10 cm and that of the gain fiber was 70 cm. The reflectivity of the FBG output coupler was about 90% at 1550nm with a bandwidth less than 0.3 nm. The total roundtrip length of the ring resonator was about 360 cm. All the applied characteristics were similar to the parameters used in the previous simulation. The laser output was detected by a 175-ps fast EOT ET-3000 InGaAs photodetector and plotted using an Agilent 300-MHz MegaZoom oscilloscope. The output power was measured using a Newport 818-IR power detector that was assembled with an integrating sphere and an energy meter.

A light-modulated 980-nm single-mode LD was employed as a pump source. The pump intensity I_p was modulated near the threshold I_pth, alternating between 0.7×I_pth and 1.05×I_pth. The applied I_p was between 5-7 mW. The low-level pump less than the Q-switching threshold could slow down the recovering of N_g, prevent free-running lasing, and give time for full recovery of N_a. The relaxation lifetime of erbium τ₂ is about 10 ms. By experience, it was better that the duration of the low-level pump in modulation was longer than ~3τ2, giving rise to the limitation of the pulse repetition rate near 30Hz.

Steady sequentially Q-switched pulses were obtained, as shown in Fig. 4. The modulation rate was set to be 25 Hz. Each pulse appeared about 12 ms after the rising edge of the pump. The pulse had pulse energy of 0.37 μJ and pulse FWHM of 218 ns. The peak pulse power of 1.69 W was achieved, in very good agreement with the simulation result. Pulse power degraded quickly with a modulation frequency over 30 Hz. Increasing the high pump level (> 1.05×I_pth) would shift the pulse closer to the rising edge of the pump. The duty-cycle should be reduced accordingly for avoiding free-running lasing. However, it could not increase the pulse power, energy and repetition rate. In addition, the Q-switched pulses were similar with the low-level pump in the range of 0~0.7×I_pth, and Q-switching performance disappeared and free-running lasing occurred when the low-level pump power was larger than 0.7 × I_pth. The average pump power could be much saved by using On/Off pump (i.e. making the low-level pump zero) and increasing the upper-level pump intensity with a shorter duty cycle. In other words, better pump efficiency could be expected if pulsed pump was employed. Please do not mistake the proposed SAQSing mechanism for gain switching by pulsed pump. As demonstrated in the previous simulation, the pump factor is ignored in Eq. (2), and the Q-switching mechanism is passively performed by quick bleaching of N_a.

 

Fig. 4. A steady output of sequentially SAQSed pulses. The square waveform was the driving current applied on the 980-nm pump laser diode. The measured pulse energy was 0.37 μJ, and the pulse duration was 218 ns.

Download Full Size | PDF

4. Conclusion

We have proposed a simple design for all-fiber saturable absorber Q-switched lasers, and numerically and experimentally demonstrated the high efficiency of Q-switching performance using single-mode erbium-doped fibers. A set of location-dependent rate equations was established for modeling the Q-switching dynamics and the time variations of n(z), N_g(z), and N_a(z). In the experiment, peak pulse power of 1.69 W with pulse FWHM of 218 ns was achieved with 7-mW pump power of a 980-nm LD. The experimental data was in solid agreement with the simulation result. The proposed laser scheme is compact, economic, efficient and applicable for various laser materials.