1. Introduction

Q-switched thulium fiber lasers emitting at the eye-safe and water absorption region of 1.9-2 μm are useful in many applications such as remote sensing, LIDAR and medicine. Among all of the known Q-switching technologies, passive Q-switching in an all-fiber laser system that has no internal air gap has the advantages of negligible Fresnel reflection between the components and low dissipation losses due to the alignment-free and maintenance-free nature of the laser. These characteristics permit an efficient Q-switching operation capable of employing a high population hold-off ratio that leads to a desirable short pulse width and high pulse energy. Passive Q-switching in a thulium all-fiber laser has been previously achieved through the use of a holmium-doped fiber as a saturable absorber (SA) [1,2] as well as through nonlinear stimulated Brillouin scattering (SBS) [3]. Holmium-doped crystals have long been used to Q-switch bulk thulium lasers [4], although thus far, holmium fiber remains the only fiber-type SA Q-switch reported for thulium fiber lasers. In 2013, Tang et al. [3] proposed an intriguing SBS Q-switching effect that was induced by intense intra-cavity focusing using a large mode-field-area (MFA) ratio between the gain fiber and a 50-m passive fiber with an ultrahigh numerical aperture (UHNA). Although sharp pulsing with peak powers in the kilowatt range was achieved, stabilizing the output pulse energy seemed challenging due to the lack of control of the SBS strength and the emission wavelength. The technique of inducing SBS in a UHNA fiber is similar to the mode-field mismatch method proposed in our earlier works [5–7]. It was first demonstrated in [5] that passive Q-switching can be achieved in a standing-wave all-fiber cavity in which the gain medium and SA Q-switch are fibers with the same dopants but with a large MFA difference. In theory, a higher intensity through the SA can be created through the use of a relatively smaller MFA of the SA, which also provides faster absorption bleaching in the SA than gain saturation in the gain fiber. Such a lensless intra-cavity focusing effect can be further enhanced by managing the relative position of the SA in the resonator and optimizing the loss ratio between the SA and the output coupler [7]. Based on the MFA mismatch method, peak powers in the kilowatt range have been achieved in Yb³⁺-doped all-fiber laser systems [8,9]. Although thulium fiber has been verified to be a potential candidate for high-power laser materials, the demonstration of an all-thulium all-fiber passively Q-switched laser using the MFA mismatch method has never been reported.

The energy transition between the ³F₄ and ³H₆ levels of a Tm³⁺-doped fiber corresponds to an absorption band from 1.5 to 2.0 μm and an emission band from 1.6 to 2.1 μm. The Stokes shift between the absorption and emission bands provides an intra-band pump window from 1.55 to 1.6 μm for an efficient quasi 4-level thulium laser emitting at 1.9-2.1 μm. To produce sharp Q-switched pulses, a short resonator length is essential and can be minimized by core-pumping a highly doped gain fiber. In this work, we demonstrate an all-fiber thulium laser that was core-pumped at 1590 nm and passively Q-switched at 1900 nm using the MFA mismatch method. The emitting wavelength of 1900 nm was chosen based on the following factors: this wavelength is near the water absorption peak of 1.93 μm, the modest absorption strength at 1900 nm favors a short SA fiber length, and most importantly, Q-switching at 1900 nm can lead to a switched gain at 2000 nm on the SA fiber sufficiently large to reset the SA fiber back to its initial status. Stable pulsing with a pulse energy of 12 μJ and a pulse width of 160 ns was obtained in a 250-cm Q-switched resonator. However, for convenience and flexibility during the experiment, passive (undoped) fiber of total length 2 m was employed in the resonator. A pulse duration of less than 40 ns was achievable if the resonator length was reduced to 60 cm.

2. Experiments

The 1900-nm passively Q-switched all-fiber laser is schematically depicted in Fig. 1. The system was core-pumped by a lab-made 1590-nm fiber laser providing a CW pump power of up to 4.5 W. Two thulium gain fibers with different Tm³⁺ doping concentrations, SM-TDF-10P/130-HE and SM-TSF-9/125, and one thulium SA fiber, SM-TSF-5/125, were obtained from Nufern, Inc. and used in the study. The three fibers are hereafter referred to as TF_GH, TF_GL and TF_SA, respectively. The emission and absorption cross sections of Tm³⁺ at the working wavelengths [10] and the specifications of the fibers are listed in Table 1. The lengths of the gain fibers, TF_GH and TF_GL, were cut to be 20 and 390 cm, respectively, to provide the same initial absorption of approximately 42 dB at the pump wavelength, λ_p. It was calculated that with both TF_GH and TF_GL pumped to reach a one-trip gain of 8 dB at 1900 nm, the pump absorption was saturated to be approximately 30 dB, indicating a pump efficiency of nearly 99.9% and negligible pump power entering the SA fiber.

 

Fig. 1 Schematic of the 1900-nm passively Q-switched all-fiber thulium laser core-pumped at 1590 nm.

Download Full Size | PDF

Table 1. Characteristics of the employed thulium doped fibers. The emission and absorption cross sections of Tm³⁺ fiber are from [10]. The initial absorptions were calculated and calibrated accordingly by the mode field confinements at λ_p and λ_Q.

View Table

By cleaving the fiber between the gain fiber and the 2000-nm HR FBG, as illustrated in Fig. 1, the 1900-nm CW output powers were measured and compared for TF_GH and TF_GL, as shown in Fig. 2(a). Because both gain fibers have the same coupling, quantum, and pump efficiencies, they exhibited similarly high slope efficiencies of near 70%. However, the pump threshold of the heavily thulium-doped TF_GH was 720 mW, much higher than the 280 mW threshold observed when using TF_GL. The relatively high pump threshold of the TF_GH fiber was primarily attributed to the cross relaxation of the Tm³⁺ population density N₂ of the ³F₄ level and partially attributed to its relatively larger mode field diameter. The cross-relaxation rate of N₂ can be calculated from the expression 2 × k₂₁₂₄ (N₂)² [11–13], where the coefficient k₂₁₂₄ describes the energy transfer process, (³F₄, ³F₄) → (³H₄, ³H₆). Using a value of k₂₁₂₄ = 1.5 × 10⁻¹⁷ cm³⋅sec⁻¹ [12,13], the cross-relaxation rate becomes comparable to the decay rate due to spontaneous emission when the pumped N₂ population reaches a value of half of the TF_GH fiber doping concentration. Namely, although a core-pumped heavily-doped thulium fiber favors a short resonator length and sharp Q-switched pulses, the increase in the pump threshold with increasing Tm³⁺ doping concentration imposes a limitation on the laser compactness. Figure 2(b) shows the average output power of the 1900-nm passively Q-switched all-fiber laser. In spite of the high pump threshold of the TF_GH fiber, the larger MFA ratio between the TF_GH and the TF_SA fibers resulted in a better slope efficiency for the pulsed operation due to the lower percentage of energy required for bleaching the TF_SA.

 

Fig. 2 The average output powers of a 1900-nm (a) CW and (b) passively Q-switched thulium all-fiber laser by core-pumping a heavily thulium-doped gain fiber (TF_GH, blue) and a relatively lightly thulium-doped gain fiber (TF_GH, red) at 1590 nm. The CW thresholds, TH_CWH and TH_CWL, were 0.72 and 0.28 W, respectively, and the Q-switching thresholds, TH_H1, TH_H, TH_L1, and TH_L, were 1.25, 0.82, 0.51, and 0.33 W, respectively.

Download Full Size | PDF

During pulsed operation, bleaching the TF_SA SA fiber at 1900 nm by a Q-switched pulse actually converts the TF_SA fiber to a gain medium at 2000 nm, i.e., gain-switching the coupled 2000-nm resonator. A subsequent gain-switched pulse bleaches the TF_SA SA fiber at 2000 nm and resets the TF_SA fiber to act as a Q-switch for the next 1900-nm pulse. The initial absorption loss of the TF_SA SA fiber was designed to be 22.2 dB at 1900 nm. Thus, for this laser system, a one-trip threshold gain of at least 23.7 dB was required to initiate the first Q-switched pulse. However, following the first Q-switched pulse, the SA loss at 1900 nm reset by a gain-switched pulse of 2000 nm, denoted by L_sa, was much less than the initial absorption loss of 22.2 dB, resulting in an actually lower lasing threshold for subsequent Q-switched pulses. It is evident from Fig. 2(b) that for the TF_GH gain fiber, there was one pump threshold TH_H1 for the first Q-switched pulse and one lower pump threshold TH_H for subsequent Q-switched pulses. Similarly, TH_L1 and TH_L denote the two thresholds when employing the TF_GL gain fiber. For sequential Q- and G-switching, we have derived the relation between the SA loss L_sa at 1900 nm (dB) and the switched gain G_s at 2000 nm (dB) based on Eq. (5) in [6]:

The lengths of the 1900-nm Q-switched resonators employing the TF_GH and TF_GH gain fibers were approximately 2.5 and 6 meters, producing pulse widths of 160 and 400 ns (FWHM), as shown in Figs. 3(a) and 3(b), respectively. The 1900- and 2000-nm pulse durations were steady and independent of the applied pump power. The Q- and G-switched pulses were detected by an EOT, Inc. ET5000 InGaAs photodetector and monitored on an Agilent MegaZoom 300-MHz oscilloscope. It should be noted that the 2.5-meter resonator contained thulium fibers of only 50 cm, i.e., 20 cm of TF_GH and 30 cm of TF_SA. Therefore, a pulse width of less than 40 ns should have been achievable if the resonator was greatly shortened by removing most of the passive (undoped) fiber in the system. The peak-to-peak duration between the 1900- and 2000-nm pulses, Δt_p, was related to the switched gain G_s and the length of the G-switched resonator. It was found that a short G-switched resonator could lead to strong interference between the 1900- and 2000-nm pulses as well as unbalance and instability in the system. To prevent such interference, the peak-to-peak duration Δt_p was designed to be 0.8 μs by extending the G-switched resonator length with an extra 5 meters of passive single-mode fiber located between the 1900-nm output coupler and the final 2000-nm HR FBG. Correspondingly, the duration of 2000-nm G-switched pulse was about 250 ns in the extended resonator length of total 6.5 meters.

 

Fig. 3 Passively Q- and G-switched pulses produced using (a) the heavily thulium-doped gain fiber TF_GH and (b) the relatively lightly thulium-doped gain fiber TF_GL.

Download Full Size | PDF

In addition, strong sub-nanosecond pulses contained in a gain-switched pulse were often observed, despite the use of a high-Q 2000-nm resonator confined with two fiber Bragg gratings (FBGs) of more than 99.5% reflectivity. Such sharp self-pulsing has been reported in the literature and was attributed to the strong absorption of the ³H₅ level and the corresponding short lifetime of 7 ns [12]. The 2000-nm output pulse energy was relatively small, and it was not practical to directly measure the average power of the G-switched pulses by separating the 2000-nm pulses from the Q-switched ones using a commercial WDM or a bulk band-pass filter. Nevertheless, the average output power of the 2000-nm G-switched pulses was determined to be approximately 0.5% that of the 1900-nm Q-switched pulses by the use of a monochromator coupled with a PbS photodetector. The spectrum of the pulsed output was measured and is shown in Fig. 4.

 

Fig. 4 Output spectrum of the passively Q- and gain-switched all-fiber thulium laser system that employed the heavily thulium-doped gain fiber TF_GH.

Download Full Size | PDF

The dependence of the repetition rate on the applied 1590-nm pump power and a typical pulse train at 25 kHz observed when employing the TF_GH fiber are shown in Figs. 5(a) and 5(b), respectively. The 1590-nm pump source was an Er-Yb doped fiber laser pumped by a multimode 915-nm LD. To avoid time jitter caused by the pump fluctuation and to acquire a consistent result, the LD was air cooled and the pulse repetition rate was measured at the LD surface temperature of 27°C. Based on the linear relationships observed in Figs. 2(b) and 5(a), the estimated pulse energies appear to be independent of the pump power, and they were calculated to be 12 and 9 μJ when employing the TF_GH and TF_GH gain fibers, respectively. Their peak powers were accordingly calculated to be approximately 75 and 22 W. The comparatively higher pulse energy achieved with the TF_GH gain fiber resulted from the larger MFA ratio between the TF_GH and TF_SA fibers, and the shorter pulse width was attributed to the shorter resonator length and the larger MFA ratio as well.

 

Fig. 5 (a) Dependence of the pulse repetition rates on the applied pump power (blue dots for TF_GH and red dots for TF_GL). (b) A typical pulse train at 25 kHz observed when using the TF_GH gain fiber.

Download Full Size | PDF

3. Conclusion

We have demonstrated a 1900-nm passively Q-switched thulium all-fiber laser using the MFA mismatch method. One heavily thulium-doped fiber TF_GH and one relatively lightly thulium-doped fiber TF_GL were evaluated as the gain fibers, which were core-pumped at 1590 nm. Both the TF_GL and TF_GL fibers exhibited high slope efficiencies of nearly 70% for CW lasing. However, much higher pump thresholds for both CW and pulsed operation were observed for the TF_GL fiber, which was attributed to the cross relaxation of the dense population N₂ of the ³F₄ level. The laser system was designed to work with high population hold-off ratios for the Q-switched and gain-switched resonators, which ensured repeatable SA bleaching at 1900 and 2000 nm and stable pulse characteristics. Through the use of the TF_GH fiber in a 250-cm resonator, the Q-switched pulse characteristics such as the pulse energy and the pulse duration were a steady 12 μJ and 160 ns, respectively, and were independent of the applied pump power. The pulse repetition rate appeared to increase linearly with the pump power, up to 50.7 kHz at the maximum pump power of 4.5 W. The Q-switched resonator consisted of 50 cm of thulium fiber in total and nearly 200 cm of passive fiber. By removing most of the undoped fiber in the system, a pulse width of less than 40 ns and a corresponding peak power of greater than 300 W were achievable.