1. Introduction

In 1978, the photosensitivity of germanium doped fiber was found by Hill in the Canadian communications research center, and then the first fiber Bragg grating (FBG) was fabricated by the standing wave formed by bidirectional propagation of 488 nm argon ion laser beams [1]. With the rapid development of fiber components fabrication technique, fiber gratings have become an important passive optical component that have been widely applied in the fields of fiber lasers, fiber communications and fiber sensors so far. In 1990, the tilted fiber Bragg grating (TFBG) model was initially proposed by Meltz [2]. The grating plane in a TFBG is no longer perpendicular to the fiber axial, which means there is a certain inclination angle, and further enhances the coupling of forward propagating core modes and cladding modes. Due to its easily fabrication without many adjustments to the system for normal FBGs, special mode coupling characteristics and unlimited fiber types, in the past two decades TFBGs have been intensively studied and widely applied to the fields of gain-flattened erbium-doped fiber amplifiers [3,4], wavelength division multiplexing (WDM) components [5], polarization dependent devices [6,7], and various sensors [8,9]. Among them, spectrum filters are the most basic applications of TFBGs. A relevant report has tried to combine the chirped fiber Bragg gratings (CFBGs) with TFBGs to fabricate a tunable broadband rejection filter [10]. A good filtering functions over a large range of bandwidths (more than 100 nm), together with a low insertion loss (less than 1 dB) and a negligible back-reflection (lower than −20 dB) was achieved, which is useful for suppression of SRS in fiber lasers and amplifiers.

During the last two decades, with the fast development of double-clad fibers manufacture technique and pump laser brightness, the output power of fiber lasers has experienced an outstanding increase [11–13]. However, the extensive research shows that the stimulated Raman scattering is one of the main limits for further power scaling and reliability of fiber laser systems. Up to now, many methods have been proposed to suppress the stimulated Raman scattering in fiber system, such as the use of large-mode-area (LMA) fibers [14], spectrally selective fibers [15,16], or lumped spectral filters [17,18] like long-period gratings (LPGs). The large-mode-area fiber was used to suppress SRS in the early time, but limited by the material and manufacturing technologies. Spectrally selective fibers usually possess highly complex designs, and also limited by the maximum fiber core size that can be employed. LPGs have good filtering properties by coupling the Raman light from the core mode to the cladding-mode [18], but instabilities resulting from numerous cross-sensitivities limit its application. Similar to LPGs, TFBGs can couple the forward propagating core modes to backward-propagating cladding modes, which makes it possible to work as a rejection filter for Raman suppression. Compared to LPGs, CTFBGs have a continuous broadband spectral profile, a better stability and an adjustable wavelength range which can be changed in different tilt angle to meet various requirements. But so far, there is only a United States patent propose to use TFBGs to filter out the Raman scattered light in a high power laser system [19]. And no experimental research about it has been reported.

In this paper, we design and fabricate a CTFBG and use it to suppress the stimulated Raman scattering in a fiber amplifier for the first time. The filtering center wavelength of the CTFBG is designed to match the peak Raman wavelength of the 1090 nm fiber laser. And the CTFBG is fabricated by the method of rotating the phase mask. A fiber amplifier system is established to test the CTFBG, and a maximum suppression ratio about 25 dB is achieved. Test results demonstrate that CTFBGs are useful for further fiber laser power scaling.

2. Theories and simulations

As shown in Fig. 1, like common FBG, TFBG possess a periodic refractive index modulation along the fiber axis, but the tilt angle between fiber cross section and grating plane leads more complex mode coupling. Thus, core and cladding mode coupling may also occur in TFBG, including the mode coupling between forward-propagating core mode and counter-propagating cladding mode, and between forward-propagating core mode and radiation mode, in which the fiber diameter is assumed to be infinite [20]. As a result, we can see a number of discrete resonances in the short wavelength range corresponding to core-cladding mode coupling in the transmission spectrum of a TFBG.

 

Fig. 1 Structure diagram of a TFBG.

Download Full Size | PDF

For TFBGs, due to the tilted grating plane, the grating period along the fiber axis Λ is different from that perpendicular to grating plane Λ_g, and their relationship can be expressed as Λ = Λ_g/cosθ, where θ represents the tilted angle. So the Bragg resonance wavelength for TFBGs [20] can be expressed as λ = 2n_coreΛ_g/cosθ, where n_core is refractive index of core mode. Further, due to the existence of the tilt angle, forward-propagating core mode could be coupled to counter-propagating cladding mode, and the cladding mode resonance wavelength can be expressed as λ_cl,i = (n_core + n_cl,i)Λ_g/cosθ, where n_cl,i is the effective refractive index of the ith cladding mode. Chirped gratings have a non-uniform period along their length, in which a series of peaks at adjacent wavelengths will be reflected and overlap with each other, resulting in a broadband spectrum. So Λ_g changes along the length and the broadened coupling bandwidths of the ith cladding mode can be expressed as Δλ_cl,i = (n_core + n_cl,i)(Λ_g,max-Λ_g,min)/cosθ.

According to the theoretical analysis, it is possible to inscribe a CTFBG as a rejection filter. To design a suitable CTFBG for suppression of SRS in fiber lasers, we simulate the transmission spectrum of a CTFBG with different tilt angles, index modulation amplitudes and chirp rates, as shown in Fig. 2. All simulations are done in the single-mode fiber HI1060 made by Corning, and using phase masks with a period of 792 nm. Figure 2(a) shows the spectrum of TFBG and CTFBG, in which the tilt angle θ = 4°. A series of discrete resonances are shown in the spectrum of TFBG. With chirping, they overlap each other as a smooth envelope, which accords well with the theoretical analysis. Figure 2(b) shows the spectrum of CTFBG in different tilt angles, with chirp rate 2 nm/cm. With the θ get bigger, more cladding mode resonances show up, and the filtering center moves to shorter wavelength, meanwhile the rejection depth decreases. This characteristic is vital for the design of a CTFBG, which related to the width of the Raman gain spectrum of the fiber used in a fiber laser or amplifier. Figure 2(c) shows how the index modulation amplitudes impact the spectrum of CTFBGs with chirp rate 2nm/cm and tilt angle θ = 4°. Obviously, the index modulation amplitude is a very important parameter for the rejection depth of the transmission spectrum. Practically, the maximum index modulation amplitude that can be achieved depends on the photo-sensitivity of fibers and the hydrogen loading pressure. As shown in Fig. 2(d), we also simulate CTFBGs with different chirp rates, and the tilt angle θ = 4°. To clearly show the pure effect of the chirp rate on the transmission spectrum, all the four curves are set to nearly have the same rejection depth by setting different index modulation amplitudes for different chirp rates, namely 0.0008, 0.0009, 0.0011, and 0.0016 for 0.8, 5, 10, and 20 nm/cm respectively. According well with the theoretical analysis, the more chirping is added, the broader spectrum is got, meanwhile, the harder to get deep.

 

Fig. 2 Calculated transmission spectrum of (a) TFBG and CTFBG with tilt angle 4°, (b) CTFBGs in different tilt angles, (c) CTFBGs with different index modulation amplitudes, (d) CTFBGs with different chirp rates.

Download Full Size | PDF

According to the simulations, we can know that the broadband filtering spectrum can be obtained by increasing the tilt angle or the chirp rate of the phase mask. However, a too large tilt angle or chirp rate will make it more difficult to inscribe for a deeper rejection. We have to make a balance among the tilt angle, the chirp rate and the rejection deep to achieve the best effect of Raman suppression. Based on our simulations, we choose 4° tilt angle and 0.4 nm/cm chirp rate. When the phase mask has a period of 792 nm, the center wavelength of the filtering span is near 1145 nm, which well meets the Stokes wavelength of 1090 nm laser.

3. Fabrication and measurements

Figure 3 shows the inscription system. The excimer laser (COMPexPro110, made by Coherent Corporation, using KrF) produces the 248 nm UV light. Two mirrors are used to adjust the height of light path. After a diaphragm and a collimation system, the light spot has a better energy distribution. Finally, the light is focused on a chirped phase mask by a cylindrical lens. The chirped phase mask and the hydrogen-loaded fiber are mounted on two six-dimension holders respectively, and fixed with a tiny gap about 10 μm, which requires a low coherence demand for the ultraviolet laser source. In the process of apodization and compensation, an amplitude mask must be scanned to contain its whole shape. As for tilting, we keep the phase mask and the fiber perpendicular to the incident inscribing beam, and only rotate the phase mask around the axis of the light beam, causing an angle of θ between phase mask grooves and the fiber, as shown in Fig. 3(b). And the real tilt angle of the grating plane in the core can be given by θ_T = π/2-arctan(n_uvtanθ)⁻¹ [6], where n_uv is the core refractive index of the fiber at 248 nm, and for HI1060 fiber it is about 1.452.

 

Fig. 3 (a) Fabrication system based on phase mask and (b) configurations used to “tilt” the grating planes inside the fiber.

Download Full Size | PDF

In the process of inscription, the laser produces 248 nm UV light with a single pulse energy 100 mJ at a frequency of 30 Hz. According to the simulation results, a linearly chirped phase mask with a period of 792 nm and a chirp rate of 0.4 nm/cm is used. Figure 4 shows the spectrum of the CTFBG after annealing. Its 3 dB bandwidth is about 17.3 nm, and the central depth of the cladding mode envelop is lower than −13 dB. Obviously the Bragg resonance can be seen in both transmission and reflection spectrum. The resonances caused by mode coupling do not show up in the reflection spectrum because the power carried by these modes gets stripped away. The closest dip to the Bragg resonance, stronger than others on the short wavelength side, is called “ghost” mode resonance because it actually consists of the superposition of several low-order cladding modes while sharing some properties of the Bragg resonance. Its insertion loss at 1090 nm is measured to be about 0.25 dB by the standard cut-off method.

 

Fig. 4 The measured spectrum of the CTFBG fabricated for the following experiments.

Download Full Size | PDF

4. Experiments and results

Figure 5 shows the experimental setup for the suppression of the SRS in a fiber amplifier. The seed laser oscillator delivers 1090 nm light, pumped by 976 nm laser diodes. To lower the Raman threshold in our experiments, 200 meters and 150 meters HI1060 fibers are used at the end of the seed and the amplifier respectively. The output fiber end is mounted on an adjustable stage without any movement before or after inserting the CTFBG to the system. We detect the total output power and the spectrum of the light scattering from the output fiber end cut in an angle of 8°.

 

Fig. 5 Experimental setup for the suppression of the SRS in fiber amplifier.

Download Full Size | PDF

Figure 6(a) shows the changing spectrums of the output as the pump power increases without CTFBG, and the seed power is 11.73 W measured at the end of the 150-meter fiber. The Stokes light near 1145nm starts to be seen at pump power 13.1 W, then increases rapidly at higher pump power. The output spectrums with the CTFBG made by us are shown in Fig. 6(b) and the measured seed power is 10.81 W, which is lower than the previous value mainly due to the splicing loss. It can been seen that the Stokes light starts to be observed at a higher pump power of 28.2 W with the CTFBG, which means a larger Raman threshold is achieved. And the Raman signal is strong suppressed at higher pump power. Figure 6(c) shows the real suppression spectrum subtracting the output spectrum without CTFBG from that with CTFBG at pump power 36.3W, and a nearly 25 dB suppression ratio is achieved, which is much larger than the measured depth of the cladding mode envelop of the CTFBG. It is because that the SRS intensity is not linear to the pump power beyond the Raman threshold. The slope efficiencies with and without the CTFBG are calculated as shown in Fig. 6(d). As the pump power increases, a downward trend in the slope efficiency without CTFBG can be obviously observed while no change with CTFBG, which means a promotion in efficiency.

 

Fig. 6 Changing spectrum of the output as the pump power increases (a) without CTFBG and (b) with CTFBG, (c) real suppression spectrum corresponding to experimental data at pump power 36.3W, (d) comparison of the slope efficiency with and without the CTFBG.

Download Full Size | PDF

5. Conclusions

CTFBGs have good optical spectrum filtering characteristics. According to the Raman gain properties in silica fibers, a CTFBG was designed and fabricated in this paper, and has been used to suppress the stimulated Raman in fiber amplifiers experimentally for the first time. A maximum Raman suppress ratio of nearly 25 dB was achieved by only one CTFBG. Experimental results demonstrate that CTFBG could be employed as an effective component for Raman suppression, which is significant for further power scaling in the high power oscillators and amplifiers. Further, for a more satisfying effect of suppression, a chain of CTFBGs can be concatenated one after the other in the future.