1. Introduction

High speed tuning of Fiber Bragg gratings (FBG) can be exploited in a number of fields, such as Q-switching, cavity dumping and pulse picking in a fiber laser, quickly addressing a sensor head or readjusting an optical filter. Various mechanisms have been exploited for tuning FBGs, such as bending beams [1], stretching with piezoelectric stack [2] and using magnetic fields [3]. Common to these methods are a relatively low tuning speed (microseconds) and mechanical durability issues. In recent years, there has been an increasing interest in exploiting strain transverse to the fiber [4–8]. The transverse stress distribution has been previously measured in a number of FBG sensor elements in high birefringence (HiBi) fibers [4], in multi-core fibers [5], in micro structured fibers [6] and in side-hole fibers [7,8]. The birefringence induced by the external load leads to both a shift and an increase in the splitting of the FBG peak associated with the two orthogonal polarization modes. This can be used to gauge the magnitude and the orientation of the transverse strain field. However, to the best of our knowledge there is no published report on fast tuning FBGs in these special fibers till now.

In previous work, we studied the applications of microstructured fibers with internal electrodes for controlling the light in the fiber. Besides electro-optical switching and modulation, nanosecond polarization switching has recently been demonstrated [9]. Increased birefringence results from the application of high-voltage electrical pulses to an internal electrode that heats up and expands. The refractive index difference between fast and slow axis of the fiber with electrodes causes polarization change, and can lead to 90° rotation with proper input polarization alignment.

In this Paper, we present how the high-speed expansion of metal electrodes affects the wavelength of a FBG written in the core of the two-hole fiber. The grating is an excellent tool to follow the evolution of the refractive index experienced by light in the polarization parallel (P) and orthogonal (S) to the direction of the holes. Besides revealing interesting physics, the nanosecond wavelength switching obtained here has potential application in Q-switching fiber lasers.

2. Experiment

The 125 μm thick silicate two-hole fiber, drawn at Acreo FiberLab, had characteristics similar to those of the standard telecom fiber (core diameter 8 μm and Δn = 0.0056) but with two ∼28 μm-diameter holes running parallel to the core separated from it by 14 μm (see Fig. 1). Pieces of fiber were filled with few-centimeter-long (typically 10 cm) metal columns that occupied the entire cross-section of the holes. Eutectic alloys BiSn (melting temperature 137°C) were pumped into the fibers in the molten state and used at room temperature as solid electrodes [10]. The length of the alloy-filled holes and the core-hole separation were such that insertion loss was experimentally measured to be < 0.2 dB and the polarization-dependent loss < 0.1 dB. Both ends of the metal-filled fiber were free from metal to allow for convenient low-loss splicing.

 

Fig. 1. Cross section of the metal-filled fiber used here (SEM picture).

Download Full Size | PDF

The metal-filled fibers were H₂-loaded for 2 weeks at 150 bar, stripped from the acrylate coating and exposed to ultraviolet (UV) in a FBG writing set-up. Frequency doubled Ar⁺ laser was used as a source of UV radiation. In the work reported here, we describe results obtained from a 4 cm-long Hamming-apodized FBG written along the middle section of metal-filled fibers. The grating was annealed in the oven at 100 °C for 12 hours.

After side polishing the fiber to electrically access one of the two internal electrodes at two points separated by 7 cm, a 20 μm thick gold-plated tungsten wire was inserted by ∼5 mm into the electrode while melting the alloy locally to ensure a robustly bonded electrical connection. Then, the pieces of fiber were mounted on an aluminum (Al) substrate and covered with heat conductive, electrically isolating epoxy. Heat dissipation into the substrate allows switching the devices at rates of a few kHz without significant drift of polarization or wavelength. The resistance of the components was typically ∼50 Ω for a BiSn device with 7 cm active length. This allowed for simple impedance matching between the device and the coaxial cable transporting the high voltage pulses. Electrical SMA contacts were used and a 0.1 Ω resistive probe was connected in series with the ∼50 Ω load for monitoring purposes.

A pulse generator consisting of a high voltage direct current (DC) power supply and a high speed switch was employed for the driving pulses. The switch here was a low repetition rate spark gap circuit delivering 2 ns rise-time current pulses with the maximum switching voltage 3 kV and duration 10-100 ns determined by the cable length used. The energy applied to the electrode is a few mJ at most and causes a temperature rise of only a few degrees (typically < 10 °C). The light source used was a tunable external cavity diode laser (TLS) with linewidth < 1 pm, and synchronized with an optical spectrum analyzer (OSA) for high resolution wavelength measurements. Different input polarization states were selected with a polarization controller (PC). A circulator, 3-dB coupler, high-speed photodiode and oscilloscope completed the experimental set-up, schematically illustrated in Fig. 2.

 

Fig. 2. (Color online) Experiment set-up. TLS: tunable laser source; PC: polarization controller; OSA: optical spectrum analyzer.

Download Full Size | PDF

3. Results

Due to the intrinsic geometrical birefriengence of the two-hole fiber, the grating has two reflection peaks at different wavelength corresponding to the different effective refractive indices for the two orthogonal polarization modes in this fiber [4]. The reflection peak at the shorter wavelength takes place in the fast axis mode (S polarization), the one at the longer wavelength in the slow axis mode (P polarization) [11]. When the metal-filled fiber device was put in a temperature-controllable oven, the two internal electrodes were heated and expanded. Consequently, the refractive indices for both polarizations and the birefringence increased as the temperature increased. The peak Bragg wavelengths associated with S and P polarization shifted to longer wavelengths and the splitting increased, as shown in Fig. 3. The insets show the reflected intensity as a function of wavelength at ∼26 °C and at ∼69 °C, measured for the two orthogonal input polarization states that match the eigenstates of the fiber. From room temperature to 69 °C, the Bragg wavelength separation for the two polarizations increased from 26 pm to 87 pm. The 3-dB bandwidth and the reflectivity of each peak remained about 30pm and 70%, respectively. Close to room temperature, the index difference inferred from Δn = n×Δλ/λ was ∼2.6×10^-5, approximately one order of magnitude lower than that of conventional HiBi fiber. Such a state can also be reached by heating the fiber by passing DC current, instead of using an oven as in the characterization measurement here.

One can note from Fig. 3 that the measured temperature sensitivity (∼37.6 pm/°C) is significantly larger than expected for a HiBi FBG fabricated in Panda fiber (∼16.5 pm/°C) [12]. As discussed in Ref [13], increased temperature sensitivity can be obtained when the fiber is embedded in an environment which exhibits a large dL / dT. An estimate can be made, for a ∼53 °C temperature increase, of the wavelength shift due to dn/ dT (0.65 nm), due to the length expansion of a glass fiber (0.045 nm), due to the alloy dilation (0.024 nm) and due to the expansion of the Al substrate (2.1 nm). Here it is assumed that the alloy expansion is restricted by the hard glass fiber (area ratio 1:9 and Young’s modulus ratio 1:6) and that the substrate is perfectly attached to the fiber. The total wavelength shift under these conditions is calculated to be 2.8 nm, which exceeds the measured value from Fig. 3 of ∼2.0 nm. The lower value measured is probably caused by slippage between the substrate and the fiber [13].

 

Fig. 3. (Color online) Temperature dependence of Bragg resonance for S and P polarization in a steady-state situation.

Download Full Size | PDF

 

Fig. 4. Typical current pulse used in the dynamic experiment. Imperfect matching into 50 Ω causes the small step at 100 ns and the undershoot.

Download Full Size | PDF

Dynamic measurements were performed for the S and P polarization states by tuning the TLS at the top of the grating reflection peak before the application of the current pulse and studying the drop in reflectivity as a function of time following the electrical excitation. Alternatively, the TLS was tuned away from the Bragg peak and the increase of reflectivity was measured on the oscilloscope trace following application of the current pulse. Intermediate (partial reflection) cases were also studied. Electrically driven FBG switching experiments were carried out at room temperature. Figure 4 illustrates a typical current pulse used here with the pulse width ∼ 60 ns and the amplitude ∼18 A at low-repetition rate (∼20 Hz). The small step on the right flank and the undershoot observed after the main current pulse are caused by electrical mismatch.

 

Fig. 5. Full switching off-on is accomplished with 29 ns risetime. Inset, the time evolution of signal in microseconds.

Download Full Size | PDF

 

Fig. 6. Full switching on-off is accomplished with 29 ns falltime. Inset, the time evolution of signal in microseconds.

Download Full Size | PDF

Close to full off-on switching with a response time of ∼29 ns was achieved when the TLS was set at the short wavelength side of the S polarization, as illustrated in Fig. 5. The increase of the signal implies that the Bragg peak is shifted to shorter wavelengths following electrical pulse excitation. The return of the Bragg peak to longer wavelengths is followed on a microsecond time scale in the inset of Fig. 5. As heat spreads from the metal electrode into the glass fiber and eventually dissipates, the pressure wave dies and the reflection peak returns to its original spectral position. The process takes tens/hundreds of microseconds. The decay of the reflected signal seen in the inset of Fig. 5 is deceivingly simple, since the probing laser wavelength does not follow the spectral evolution of the Bragg peak once the overlap disappears. Heating, however, red-shifts the grating wavelength.

Close to full on-off switching was also obtained with a response time of 29 ns, as shown in Fig. 6, by setting the TLS wavelength to match FBG peak of the S polarization. The long term recovery of the grating, shown in the inset of Fig. 6, is not monotonic. The reflected signal remains null as long as the overlap between the probing laser source and the FBG displaced to shorter wavelengths is zero. Since the FBG has a FWHM of 30 pm, full switching implies displacement > 20 pm. The time is typically a few microseconds. After this relatively short time, the peak reflectance wavelength overlaps once again the probing laser wavelength. The signal reflected by the FBG then returns to the maximum amplitude. The heat from the metal electrode finally reaches the core of the fiber, red-shifting the Bragg wavelength, and this leads to a second reduction of the reflected signal. This time, however, the FBG peak lies on the long-wavelength side of the laser source. The time scale involved in the heat component is slower. The device can be said to return to room temperature after a delay close to 2 ms. It may be noticed from the inset of Fig. 6, that the second (negative) peak in the reflected signal is not as deep as the first one, where 100% switching is achieved. This depth is a measure of the remaining overlap between the probing laser wavelength and the red-shifted FBG peak. From the relative amplitudes of the peaks, a thermal shift of +17 pm can be estimated for both polarization states, which corresponds to a ∼1.3 °C rise after each high current pulse. Here, the Bragg wavelength dependence on the Al substrate expansion due to temperature is not considered because of the large thermal mass of the substrate and the short duration time and low repetition rate of the current pulse. The alloy was heated by ΔT ∼10 °C with a current pulse of ∼18 Amps over ∼60 ns (with heat deposited ∼1 mJ). Assuming for simplicity that the energy spreads uniformly into the fiber and taking into account typical values for the specific heat of the alloy and glass, a temperature increase in the core of ∼1.6 °C is estimated, which is consistent with the value above.

The dynamic experiments repeatedly showed for the device tested that the largest wavelength shift during the application of the electrical pulse was to shorter wavelengths (typically -27 pm for 100% switching) for the S polarization, while for the P polarization the wavelength shift was positive, but much smaller (typically +5 pm). Similar behavior has also been reported in a HiBi side-hole fiber [14]. One qualitative model to explain the FBG shift to shorter wavelengths is to consider that when the core is compressed in the direction of the holes, it expands in the orthogonal direction and thus the refractive index reduces. Simulations of the effect using the elasto-optical tensor of glass confirm that the compression of the core in one direction leads to a weak increase of the refractive index in the direction of the force, and a stronger decrease of the refractive index for the orthogonal direction.

4. Conclusion

A FBG was written in a two-hole fiber with internal alloy electrodes, which has intrinsic geometrical birefriengence. Nanosecond high current pulses heated the metal electrode and caused metal expansion rapidly. In consequence, birefringence was increased and the gratings was tuned with a response time of 29 ns, limited by the finite time needed to deposit heat and the inductance of the alloy electrode. The fast risetime switching results from the increased pressure, while heat transport to the core is slow and determines the recovery time of the device after every drive pulse. Previous durability tests were carried out with devices pulsed 10⁹ times without degradation in performance and high repetition rates (> 10 kHz) can be obtained by running the device above room temperature [9]. It is believed that the short length, low loss, all-spliced high-speed devices described here can find useful application for Q-switching fiber lasers.