1. Introduction

Q-switching of fiber lasers is an established technique to generate short pulses with high peak power and has many applications, such as optical time-domain reflectometry, material processing, remote sensing and nonlinear optical frequency conversion [1,2]. All-fiber devices for Q-switching are advantageous due to their alignment-free characteristics and lower cavity loss. Although passive Q-switching usually involves a simple laser cavity and produces high-power short duration pulses [3–10], the repetition frequency and the pulse timing are often uncontrolled. In contrast, active Q-switching uses external means to determine the loss of the laser cavity and allows for more control over the characteristics of the output pulses. Therefore, actively Q-switched lasers are the usual choice for industrial applications.

The most common scheme for producing high power pulses from a fiber laser is the master oscillator power amplifier (MOPA) configuration, where a semiconductor laser pulse seeds a fiber power amplifier [11]. This technique is very powerful and is widely used. However, the MOPA configuration also has its weak points, such as the need for a few amplifying stages and very careful isolation of the seed laser from feedback. This motivates the search for a simpler Q-switching scheme that allows for the direct generation of a high-power short duration pulse by a fiber laser. This could potentially eliminate the need for some of the amplifier stages and a sensitive semiconductor laser diode seed. In the last years, several successful all-fiber approaches based on different modulation techniques have been reported, such as all-fiber intensity modulators [12], all-fiber acousto-optic attenuators [13–15], and a microsphere resonator [16]. Q-switching of all-fiber lasers has been accomplished by tuning fiber Bragg grating (FBG) using acousto-optic modulators [17–19], piezoelectric actuators [20,21], magnetostrictive transducers [22], and temperature controllers [23]. Problems with these methods include mechanical relaxation, hysteresis, slow rise- and falltime and low extinction ratio.

In our previous work, a microstructured fiber with electrically driven internal electrodes has been employed to cavity dump an all-fiber erbium-doped ring laser [24]. The internal electrode fiber component allows for polarization rotation of light guided in the core with a risetime <10 ns [25]. These silica-fiber based devices are simple to use, are spliced in a conventional way to the fiber laser and introduce low loss. Nanosecond gating with two consecutive switching actions separated by an adjustable nanosecond delay was also demonstrated [26]. In the present paper, we exploit the electrically controlled polarization switch intra-cavity for Q-switching a linear all-fiber laser, generating pulses of nanosecond duration and tens of watts of peak power. The use of a nanosecond optical switch with ON and OFF functions opens the possibility of designing novel and improved cavity arrangements.

2. Polarization switch

Active Q-switching is performed by an all-fiber nanosecond polarization switch described in detail in previous work [24–26]. This component is based on a microstructured fiber with four holes running parallel to the core. The fiber is 125 μm in diameter with a core diameter of 8 μm. The device is manufactured to be single mode at 1.5 μm and compatible with standard telecommunication components. The four 28-μm diameter holes are filled with BiSn alloy over a length of ~7 cm. The electrodes exhibit a static resistance ~47 Ω. Two of these metal-filled holes placed at a 90° relative angle (as illustrated in Fig. 1(a) ) are electrically connected to a driving source through 50 Ω coaxial cables. The insertion loss measured with these components is very low (≤0.2 dB) and the polarization-dependent loss <0.1 dB. Both ends of the metal-filled fiber are free from metal to allow for convenient low-loss splicing.

 

Fig. 1 (a) Cross section of the microstructured fiber used here (SEM picture). (b) Typical optical response of the polarization switch placed between crossed polarizers.

Download Full Size | PDF

Driving a short high voltage (HV) pulse through one electrode results in heat generation, causing the metal to expand rapidly. This mechanically stresses the core and induces birefringence. The polarization of the guided light then rotates in nanoseconds, which is used for rapid ON-switching. The switching ON is much faster than the thermal relaxation attributed to the heat dissipating in the fiber and surroundings, which takes place on a scale of a few 100 μs. In order to switch OFF without waiting for heat dissipation, another HV pulse is applied to the second electrode. The birefringence introduced previously is counteracted for when the second electrode is driven. Therefore, the polarization is rotated back, causing switching OFF after an electrically controllable delay time.

The electrical signal to drive the polarization switch is generated by a ~5-ns risetime CMOS switch driven by a conventional function generator and biased with a DC power supply to <2 kV. Figure 1(b) shows the typical optical response of the polarization switch with rise- and falltime of ~5 ns. The voltages, full widths at half maximum (FWHMs) and frequencies of the applied HV pulses are: 1350 V, 22 ns, 100 Hz; and 1300V, 22 ns, 100 Hz, respectively. The delay time between the electrical ON switch and OFF switch pulses is ~500 ns. The delay between activation times of the two electrodes can be varied electrically from several nanoseconds to one millisecond. With the correct polarization adjustment, the extinction ratio is measured to be better than 25 dB, limited by the dynamic range of our measurement. Acoustical oscillations [25,27] appear ~75 ns after the fast switching due to imperfect damping to the substrate during packaging of the device used here. This component can be operated at rates ~kHz without significant polarization drift at room temperature.

3. Q-switching

The actively Q-switched all-fiber laser setup is depicted in Fig. 2 . The diode pump available here is a single transversal mode 976 nm laser diode suitable for core pumping. The maximum available pump power after three spliced 0.98/1.55 µm WDM stages for isolation is ~450 mW. A 0.9-m long piece of erbium-doped fiber (EDF) (core diameter 8 µm and 44 dB/m absorption at 978 nm) provides the gain. The 1550 nm port of WDM2 is used for intracavity power monitoring. The linear cavity is formed by a pair of Hamming-apodized high-reflectivity (HR) FBGs. The lengths, central wavelengths, FWHMs and reflectivities for FBG1 and FBG2 are: 7 cm, 1544.80nm, 108 pm, 99.99%; and 2 cm, 1544.75 nm, 504 pm, 99.99%, respectively. The lasing wavelength is determined by the peak wavelength of FBG1. FBG2 is spliced to the polarization switch with a separation of a few millimeters. Polarization controller 1 (PC1) is adjusted to maximize the horizontally polarized light in the cavity, while PC2 is employed as a λ/4 plate. A fiber polarization splitter directs the vertically polarized light to port V, which is also the output port. Power meters, 1 GHz photodiodes connected to an oscilloscope and an optical spectrum analyzer are used to characterize the performance of the fiber laser. The cavity length is ~8.5 m, corresponding to a roundtrip time of ~85 ns. Compared with the ring cavity [24,28], only half the electrical energy is required for polarization switching because the light passes twice through the polarization switch.

 

Fig. 2 Schematic of the all-fiber actively Q-switched erbium fiber laser setup with a polarization switch based on a four-hole fiber with internal electrodes.

Download Full Size | PDF

The Q-switched laser system works in the following way. First the polarization switch is inactive. The cavity is in its OFF-state by properly adjusting PC2 to behave as a static λ/4 waveplate. All radiation from the doped fiber that is reflected by FBG2 is removed from the cavity through port V of the polarization splitter. No feedback is provided and therefore lasing cannot start, building up population inversion in the EDF. By applying a short HV pulse to electrode 1, the polarization switch introduces a λ/4-shift and the Q-factor of the cavity is switched from low to high. The two 100% FBG reflectors, the all-fiber cavity with low loss components and the very fast risetime of the optical switch ensure that a short laser pulse develops in the cavity, caused by pulse distortion due to gain saturation. Circulation of the optical flux is allowed for a few roundtrips. During this period, all optical flux is kept inside the cavity. After a controllable delay time electrode 2 is driven, so that the polarization switch induces a λ/4-shift in the opposite direction. As a consequence, the cavity is switched OFF quickly. All the optical flux is once again extracted from port V of the polarization splitter. By monitoring the intracavity flux, it is possible to adjust the delay to maximize the energy extraction.

Figure 3 displays typical Q-switched pulses and corresponding intracavity flux for launched pump powers 115 mW and 400 mW, at a switching rate 100 Hz. Figure 3(a) displays a Q-switched output pulse with 50 W peak power, <6 ns FWHM and ~0.3 μJ pulse energy. Both the rise- and falltime are less than 5 ns. The voltage and FWHM duration of the applied HV pulses are 1030 V and 38 ns, and 965 V and 40 ns, respectively. The electrical delay time is ~500 ns, i.e. 6 roundtrips, the time taken for the highest intensity pulse to develop in the cavity. The output pulse generated is clean and short, in spite of the intracavity flux being noisy. It is found that the subpulses seen in the inset of Fig. 3(a) are not random, since they reappear after a roundtrip. The limited dynamic range of the measuring system prevents the observation of the pulse growth in the small signal gain regime, when the amplitude of the pulse increases by many orders of magnitude. In the inset of Fig. 3(a), only the optical flux in the last two roundtrips is seen above the noise level, when the gain is saturated. It should be noted that the intracavity flux and the output signal are not measured with the same time reference because of the different fiber lengths involved, but the main peak seen in the inset is also the main peak of the extracted pulse. In Fig. 3(b), a Q-switched pulse with 50 W peak power, ~12 ns FWHM and ~1 μJ pulse energy is obtained when launching 400 mW pump power. The rise- and falltime are ~3.7 ns and ~18.5 ns. The applied HV pulses are 1350 V and 22 ns, and 1300 V and 22 ns. Here, the electrical delay time is adjusted to ~250 ns, since the intracavity signal starts to decrease after 3 roundtrips. The system is sufficiently versatile to accommodate for this adjustment with the turn of a knob. A comparison of Fig. 3(b) and its inset shows the amplification of the main peak by ~8 times in the last roundtrip, while the low amplitude signal trailing the main pulse does not experience amplification, due to gain saturation. The subpulses are attributed to the acoustic oscillations of the polarization switch as shown in Fig. 1(b).

 

Fig. 3 Output power of the Q-switched pulses for a pump power (a) 115 mW and (b) 400 mW. The insets are the corresponding intracavity signals.

Download Full Size | PDF

The fast risetime of the polarization switch (<10 ns) produces an abrupt switching flank. In both examples illustrated in Fig. 3 the pulse duration (~10 ns) is much shorter than the cavity roundtrip time (85 ns). Hence, the configuration employed here allows for the use of long pieces of amplifying fiber, since the pulse duration is not directly linked to the roundtrip time of the laser. This is in contrast to most actively Q-switched fiber lasers, where the pulse duration is associated to the roundtrip time of the cavity. Here, the output of the laser is a short duration single pulse although laser oscillation is allowed during some roundtrips (e.g., Fig. 3(a)). This is a consequence of the polarization switch working initially as a Q-switcher to enable the formation of a pulse and later as a cavity dumper to remove all flux from the cavity. The good extinction ratio of the polarization switch prevents the leakage of the flux during the pulse build-up phase.

4. Conclusion

An actively Q-switched all-fiber laser system with an integrated polarization switch based on an electrically controlled microstructured fiber is reported. The maximum pulse energy measured in the pulses is ~1.0 μJ with about 12 ns duration when launching 400 mW pump power and switching at 100 Hz. Pulse durations are much shorter than the cavity length. The EDF chosen as gain medium to simplify the characterization of the Q-switched fiber laser can be readily replaced by Yb-doped fiber, where higher powers should be produced. Cladding pumping does not change the physical mechanisms described here, and should also lead to the generation of higher power signals.