1. Introduction

Random lasers (RLs), firstly reported by Ambartsumyan et al. in 1966 [1], refer to a class of lasers in which there is no a classical resonator but with randomly distributed scattering centers. The typical configuration forms of RLs are the powders [2–4] and laser dyes with nanoparticles [5]. The simplicity of realization of those RLs provides them potential advantages over conventional lasers, however some typical drawbacks restricts their development, such as requirement of a pulse laser pump with high peak power, low efficiency and, especially, no control over the spectral properties of emission. To confine and orient the light emission of RLs, the random fiber lasers (RFLs) as a new breed of lasers were proposed and demonstrated experimentally, such as producing the random distributed feedback (R-DFB) by inserting the gain solution with random scattering particles into a hollow core optical fiber [6], by using strong scattering from a number of different fiber Bragg gratings (FBGs) placed in a random way along fiber axial direction [7,8] or by introducing random phase errors in a single long FBG [9]. RFLs have attracted much attention due to their advantages of narrow linewidth, tunable spectral capability, simple configuration and low cost. Especially, in recent years, a new type of RFL using R-DFB from the weak Rayleigh backscattering (RBS) in an ultra-long single-mode fiber (SMF) has been studied extensively [10–22]. The first report of RFL using such kind of R-DFB was based on the gain deriving from the stimulated Raman scattering (SRS) in silica fibers [20], and relative studies followed continuously [19,21,23–36]. Two other gain mechanisms including the amplification of Stokes waves in stimulated Brillouin scattering (SBS) [11,37,38] and the emission from rare-earth-doped fibers [10,12,15] were also introduced into the RFLs successively. Hybrid gain mechanism for RFLs was reported as well [11,22,39–41].

Since the modeless characteristics in RFLs induced by the R-DFB effect can significantly suppress the interactions among multi-lasing wavelengths [10,12–14,16–19,39,42,43], especially for the serious wavelength competition existing in the homogenous gain medium of rare-earth-doped fibers [10,12,39], and in view of the inherent merits of RFLs aforementioned, the multiwavelength RFLs (MW-RFLs) have aroused much attention owing to the potential applications in wavelength-division-multiplexing optical communication systems, optical fiber sensing, instrument testing, microwave photonics and high-resolution spectroscopy [17,19,44–46]. In recent years, many MW-RFLs have been reported by combining with the comb-filters such as cascaded FBGs [13,23,32], all-fiber Lyot filter [16,39], Sagnac loop interferometer [10,12,14,43], waveshaper programmable filter [42], and Fabry-Pérot (F-P) cavity combined with a special Mach-Zehnder interferometer [18]. However, in contrast, the gain efficiency of SRS is low and a pump laser power of the order of several watts is generally required, which makes the SRS-based RFLs complicated and relatively expensive; The gain bandwidth of 1^st order SBS is too narrow, and it is pretty difficult to achieve multiwavelength lasing by stimulating higher order Stokes waves using 1^st order random laser [17]. Considering that an erbium-doped fiber (EDF) has a gain bandwidth of >50 nm and the pump threshold of an EDF-based RFL can be less than 20 mW [15], the EDF is a good candidate as a gain medium for MW-RFL for many special applications. Up till now, only the multiwavelength random EDF lasers (MW-REDFLs) using a fiber Sagnac loop as the comb-filter have been demonstrated as seen in [10,12]. But, the power difference of lasing channels was serious and the lasing slope efficiency was low in [12], and the laser system was complicated and lasing slope efficiency was rather low induced by the four-wave mixing effect in [10], which are far from satisfactory. Superimposed fiber-Bragg-gratings (SI-FBG) has been shown to be used as a low-cost comb-filter of narrowband, compact, easy to fabricate, and flexible to design [47–50]. However, to the best of the authors’ knowledge, a MW-REDFL using a SI-FBG as the comb-filter or a SI-FBG with more than three FBGs used in a fiber laser has not yet been reported in literatures.

In this paper, we propose and experimentally demonstrate a high performance MW-REDFL with six lasing wavelength-channels. This is what we believe the first report about a RFL based on a half-open linear cavity where both ends are formed by one reflecting comb-filter mirror of six-superimposed fiber-Bragg-gratings (6SI-FBG) and 2 km long SMF for providing R-DFB from weak RBS. The MW-REDFL is with both of the high spectrum and output power stabilities which are validated by the measured little maximum peak power difference (d_p) among six lasing channels, little peak power fluctuation (f_p) and little wavelength fluctuation (f_λ) in each channel over time, little output power fluctuation over time and low relative intensity noise (RIN). Besides, the high performance of total output power versus pump power is also obtained by means of the simple and efficient laser cavity design of the MW-REDFL.

2. Experimental setup and principle

The configuration of the proposed MW-REDFL is illustrated in Fig. 1(a), which is simply composed of a 7 m long EDF pumped by a 980 nm laser diode (LD, CONNET FIBER OPTICS, VLSS-980) through a 980/1550 nm wavelength division multiplexer (WDM), a 2 km long SMF, one 6SI-FBG based reflecting comb-filter and an isolator. The EDF used is home-made, using a customized modified-chemical-vapor-deposition system and a customized specialty optical fiber drawing tower, with a mode field diameter of ∼5.4 µm at 1550 nm, an absorption of ∼3 dB/m at 980 nm and an absorption of ∼6 dB/m at 1550 nm. The 2 km long SMF (Corning Cor., SMF-28) is used to provide the R-DFB via the RBS effect with a scattering coefficient of ∼4.5×10⁻⁵ km⁻¹.

 

Fig. 1. (a) Experimental configuration; 6SI-FBG: six-superimposed fiber-Bragg-gratings; EDF: erbium-doped fiber; LD: laser diode; WDM: wavelength division multiplexer; the 2 km SMF-28 fiber placed in a foam box for reducing of environmental vibrations. (b) Reflection and transmission spectra of 6SI-FBG; Inset: zoomed-in of one reflection channel to show details. (c) Measurement system; ① optical spectra measurement; ② radio frequency (RF) spectra measurement; ③ optical power output measurement; ④ relative intensity noise (RIN) measurement; OSA: optical spectrum analyzer; PD: photodetector; ESA: electrical spectrum analyzer; OPM: optical power meter.

Download Full Size | PDF

The 6SI-FBG reflector with a length of 2 cm was formed by six narrow-band FBGs inscribed in turn at the exact same fiber position of a section of hydrogen-loaded germanium-doped Corning SMF-28 fiber by using six uniform phase masks with different periods, based on the phase mask method with a KrF excimer laser emitting ultraviolet light at 248 nm. The reflection and transmission spectra of the 6SI-FBG measured using an EDF amplifier (EDFA) and a Yokogawa AQ6370D optical spectrum analyzer (OSA) with a resolution of 0.02 nm and a data sampling interval of 0.001 nm are shown both in Fig. 1(b). The center wavelengths of six reflection peaks are 1544.269 nm (channel 1), 1546.281 nm (channel 2), 1548.255 nm (channel 3), 1550.267 nm (channel 4), 1552.317 nm (channel 5) and 1554.242 nm (channel 6) with the 3-dB bandwidths of 0.126 nm, 0.142 nm, 0.158 nm, 0.160 nm, 0.152 nm and 0.150 nm respectively, and the reflectivity at every reflection peak is larger than 90% (depth >10 dB for every dip), which make the 6SI-FBG an excellent comb-filter and an outstanding multiwavelength lasing cavity mirror. Inset shows the zoomed-in of one reflection channel, and two obvious sidelobes on both sides of the main reflection peak, might induced mainly by the imperfect apodization of the FBG and the overlapping inscription of FBGs, are seen. The useless end-face of the 6SI-FBG is fusion spliced with an angled FC/APC fiber connector to eliminate the unwanted Fresnel reflection. It should be noted that, compared with using an array of closely spaced FBGs, the 6-SIFBG as a reflector of laser cavity is more compact, more stable, easier for packaging, and more flexible for adjusting or tuning. The isolator is used to minimize the end-face Fresnel reflection of the laser output fiber port and other back-reflections. Figure 1(c) shows the measurement system of all experiments, including the optical spectra measurement using the AQ6370D OSA, the radio frequency (RF) beating spectra measurement using an 18 GHz photodetector (PD) and an electrical spectrum analyzer (ESA, Keysight Cor., N9010A), the optical power output measurement using an optical power meter (OPM, Joinwit Cor., JW3209), and the RIN measurement using a 400 MHz PD (Thorlabs, PDB470C), an oscilloscope (Tektronix, TDS2024C) and the N9010A ESA.

The mechanism of the MW-REDFL can be described as follows. The amplified spontaneous emission (ASE) light comes from the EDF pumped by the 980 nm LD and arrives at the reflector of 6SI-FBG to be filtered. Thanks to the high reflectivity and narrowband of 6SI-FBG at each channel of six ones, the seed light with strong intensities at six wavelength-channels, for the multiwavelength lasing of the MW-REDFL, is generated and subsequently amplified by stimulated emission of radiation in the EDF again. Then, after the amplified light backscattered by the R-DFB via the RBS effect in the 2 km SMF-28 fiber, one propagating loop is achieved. Finally, as long as the pump power is sufficient enough to overcome losses when the signal light at each channel travels forth and back for every single loop in the half-open cavity where both ends are provided by the mirror of 6SI-FBG and the R-DFB of the 2 km SMF, the six-wavelength lasing can be achieved. Note that, for reducing of environmental vibrations, we placed the fiber spool of the 2 km SMF in a foam box as shown in Fig. 1(a).

The gain coefficients of the 7 m long EDF at all of the six center wavelengths of the 6SI-FBG are 37.07 dB/W, 36.60 d B/W, 36.17dB/W, 35.69 dB/W, 35.26 dB/W and 35.00 dB/W respectively, calculated using the measured reflection spectrum of the 6SI-FBG and the single-pass amplification spectrum of the reflection, with a pump power of 250 mW, as shown in Fig. 2. The passive insertion loss of the 6SI-FBG was measured to be approximately ≤0.50 dB in a wavelength-range of 1539∼1559 nm. In Fig. 2, the ASE spectrum is also given by the dotted-line, and a gain variation of ∼3 dB from 1543 nm to 1555 nm can be seen. Additionally, from the single-pass amplification spectrum of the 6SI-FBG’s reflection, we can see a maximum power difference of 4.46 dB for the six amplified channels, which provides a guidance for the better equalization of the 6SI-FBG in the future after single-pass amplification and subsequently contributes to power-equalized six-wavelength lasing for the MW-REDFL.

 

Fig. 2. Reflection spectrum of 6SI-FBG (red dashed-line) and one-time amplification spectrum of 6SI-FBG’s reflection (blue solid-line), measured with the ASE spectrum of the 7 m long EDF pumped by a power of 250 mW. ASE spectrum of the pumped EDF also given by the magenta dotted-line.

Download Full Size | PDF

3. Results and discussions

Due to the weak wavelength competition benefiting from the R-DFB effect in a MW-REDFL, the simultaneously lasing of six wavelengths can be obtained easily under a low pump power without the need of any adjusting of the RFL’s elements. The evolution of the lasing output spectrum with the increase of pump power was investigated firstly using the OSA with a resolution of 0.02 nm and a data sampling interval of 0.002 nm as shown in Fig. 3. As can be seen in Fig. 3(a), when the pump power is set at only 30 mW, the laser operation with six wavelength-channels centered at 1544.330 nm, 1546.360 nm, 1548.366 nm, 1550.328 nm, 1552.380 nm and 1554.316 nm respectively, consistent well with the six reflecting center wavelengths of the 6SI-FBG, is obtained, although the optical signal to noise ratio (OSNR) is only no more than 30 dB for all of the lasing channels and the d_p of six lasing channels is as high as 6.160 dB. Note that the little wavelength deviation between the lasing channels and the corresponding reflecting center wavelengths of the 6SI-FBG is mainly induced by the recoating process of grating zone for protection using an optical fiber coater (Fujikura, FSR-02) and the room temperature fluctuation. However, with the pump power increasing from 30 mW to 250 mW as shown in Fig. 3(a) to (f), the OSNR for each lasing channel becomes higher and higher, and the d_p of six lasing channels becomes lower and lower. Under the pump power of 250 mW, the OSNR is as high as 40 dB, and the d_p is as low as 1.263 dB as marked in Fig. 3(f). Note that, due to the high ASE noise in a half-open cavity RFL, the OSNR of such kind of RFL’s output is generally lower than that of a common fiber laser with a traditional oscillating cavity, in which an OSNR of >60 dB is easy to achieve. It also can be obviously seen from Fig. 3(a) to (f) that there are many emission lines and unstable random spikes in every lasing channel, which is mainly attributed to the too weak RBS at low signal power, the Stokes lines of SBS generated in the long SMF-28 fiber and the environmental disturbances. The occurred SBS effect can be verified by measuring the RF beating spectrum of the laser output, as shown with the red solid-line in Fig. 4, using the self-homodyne technique by detecting the laser output with the 18 GHz PD before connecting its output to the ESA. An obvious beating signal at the frequency of 10.87 GHz can be seen, which is consistent with the Stokes frequency shift of SBS at C-band. For comparison, the beating spectrum of laser output was also measured when the foam box was removed, as shown as the blue dashed-line. As can be seen, except for the SBS induced beating signal, some other obvious frequency noises were also captured by the ESA, which inferred are mainly from the beating of the unstable lasing frequency lines induced by the environmental disturbances. The insets of Fig. 3(a) to (f) show in detail the zoomed-in spectra of the fifth lasing channel at ∼1552.380 nm under different pump power, in each of which the randomly lasing spikes and two sidelobes are clearly seen. Additionally, with the pump power increasing from 30 mW to 250 mW, the spectrum becomes more stable and the bandwidth of lasing channel becomes wider, which may be mainly attributed to four aspects as follows. With pump power increasing, i) the RBS becomes stronger, which enhances the laser oscillation to make it more stable; ii) the nonlinear spectral broadening, resulting from nonlinear interaction such as cross-phase modulation and frequency mixing, takes effect [10]; iii) the generated SBS Stokes waves in each lasing channel amplified twice by the EDF via reflection of the 6SI-FBG are with considerable intensity and output with the random laser together to broaden the lasing channel bandwidth; iv) the lower-order SBS Stokes wave components in each lasing channel exceed threshold and excite higher-order SBS lines, and overlapping effect of multi-order SBS lines broadens the bandwidth of each lasing channel and makes it relatively stable [12]. Therefore, the laser emission and spectrum evolution behaviors of the MW-REDFL’s output are contributed together by the R-DFB of the RBS effect, the nonlinear interaction and the amplified and cascaded SBS generated in the long SMF-28 fiber. Note that the sidelobes on both sides of each lasing channel in Fig. 3 directly resulted from the sidelobes of the 6SI-FBG as shown in Fig. 1(b).

 

Fig. 3. Lasing spectra at different pump powers of (a) 30 mW, (b) 50 mW, (c) 100 mW, (d) 150 mW, (e) 200 mW and (f) 250 mW. The inset in each figure is the enlarged spectrum of the fifth lasing channel centered at ∼1552.380 nm. d_p: maximum peak power difference.

Download Full Size | PDF

 

Fig. 4. Typical RF beating spectrum of laser output at 50 mW pump power measured by the self-homodyne method with (red solid-line) and without (blue dotted-line) the foam box respectively.

Download Full Size | PDF

The output stability of the MW-REDFL was then investigated in detail. i) At the pump power of 250 mW, as a typical case, we studied the optical spectrum stability by measuring multiple repeated OSA scans and analyzing the experimental data. Figure 5(a) and (b) respectively show the five spectra of six-wavelength lasing repeated at ∼20 min intervals and the f_p of six lasing channels in the 80 min time span, and a stable multiwavelength operation of the MW-REDFL is demonstrated by both of the little d_p among six lasing channels in every OSA scan and the little f_p for each lasing channel in the 80 min. As given in Fig. 5(a) and (b), the maximum of d_p measured is only 0.930 dB and the maximum of f_p measured is only 1.101 dB, which powerfully validates the mitigation effect of R-DFB to the strong wavelength-competition existing in the homogeneous broadening gain medium of EDF. ii) We also measured the wavelength variations (v_λ) of six-lasing channels as a function of pump power from 30 mW to 300 mW and the wavelength fluctuations (f_λ) of six lasing channels in a time span of >1.5 h (∼95 min) at a constant pomp power of 250 mW, using the OSA with a resolution of 0.02 nm and a data sampling interval of 0.003 nm, as shown in Fig. 6(a) and (b) respectively. Note that due to the numerous random spikes in each lasing channel we used the OSA’s average-mode of 100-times for each single spectrum measurement and read the center wavelength of each channel. As can been seen in Fig. 6(a), with the increasing of pump power, a wavelength “red-shift” for each lasing channel occurs and the maximal variation for six lasing channels is 0.052 nm, which are mainly induced by the thermal effect of the 6SI-FBG with the increasing of optical intensity inside the laser cavity. However, a random wavelength fluctuation for each lasing channel in an experimental time of >1.5 h is observed (see Fig. 6(b)) and the maximal value is as small as 0.032 nm, which are mainly induced by the influences of ambient vibration and temperature fluctuation (1∼2°C) to the 6SI-FBG. iii) Additionally, Fig. 7 gives the 3-dB bandwidths of six lasing channels as a function of pump power, measured also with OSA’s average-mode, in which a slow increasing trend for each lasing channel can be seen, with the maximum variation of 0.068 nm. Note that the random fluctuation mixed with the increasing trend for each channel is mainly from the thermal effect of the 6SI-FBG, the vibration and the ambient temperature fluctuation. iv) The stability of six-wavelength lasing output power was studied by monitoring the total output power at the pump power of 100 mW and 200 mW respectively using the OPM for 240 minutes as shown in Fig. 8. It can be seen that the power fluctuation at both of the pump powers is ≤0.05 dB, showing an excellent output-power stability for the proposed MW-REDFL. Note that, according to the above experimental data, the output power fluctuation is much less than the f_p for each lasing channel, which indicates that the power fluctuation mainly occurred among six lasing channels while the total output power maintained a high stabilized value under a fixed pump power. v) The RIN of the WM-REDFL was characterized by the RIN measurement system. Figure 9(a) and (b) show the measured RIN spectra at the pump powers of 100 mW and 200 mW respectively with the ESA frequency range setting of 0∼1 MHz and resolution bandwidth (RBW) setting of 100 Hz. Meanwhile in the inset-1 of Fig. 9(a) and inset-2 of Fig. 9(b) the RIN spectra in a smaller frequency range of 0∼100 kHz are also given. In comparison, a RIN spectrum of a low noise commercial DFB laser (HAN’S RAYPRO SENSING product, Model: RP-MP-10-0100-02, a 50 kHz linewidth single-frequency laser at 1550 nm with ultra-low RIN) was also plotted in Fig. 8(a) and (b) using the red dashed-line. As can been seen, at frequencies lower than 100 kHz the RIN spectra of the proposed MW-REDFL almost overlap with that of the commercial DFB laser. At frequencies larger than 100 kHz, the RINs almost reach a constant level of −125 dB/Hz for 100 mW pump power and −120 dB/Hz for 200 mW pump power respectively, which validate again the high power stability of six-wavelength lasing output of the MW-REDFL. Note that the optical power input into the PD must be adjusted to an appropriate level through a variable optical attenuator for each measurement.

 

Fig. 5. (a) Five times repeated OSA scans at ∼20 min intervals under the pump power of 250 mW; (b) Peak power fluctuations of six lasing channels in the 80 min time span. d_p: maximum peak power difference; f_p: peak power fluctuation over time.

Download Full Size | PDF

 

Fig. 6. (a) Wavelength variations of six lasing channels versus the pump power; (b) Wavelength stabilities of six lasing channels at 250 mW pump power in a time span of >1.5 h. v_λ: wavelength variation over pump power; f_λ: wavelength fluctuation over time.

Download Full Size | PDF

 

Fig. 7. Variations of 3-dB bandwidths of the six lasing channels versus the pump power. v_b: variation of 3-dB bandwidth over pump power.

Download Full Size | PDF

 

Fig. 8. Measured results of output power stability at 100 mW and 200 mW pump power respectively.

Download Full Size | PDF

 

Fig. 9. Relative intensity noise (RIN) spectra (blue solid-line) of lasing output measured in the frequency range of 0∼1.0 MHz at (a) 100 mW and (b) 200 mW pump powers with RBW of 100 Hz. Inset-1 of (a) and inset-2 of (b) respectively displaying the same measurement in a frequency range of 0∼100 kHz. For comparison, the RIN spectrum (red dashed-line) of a commercial DFB laser also plotted.

Download Full Size | PDF

The total output power as a function of pump power was also investigated as shown in Fig. 10. With the exception of the first two experimental data points, the others are linearly fitted with an adjusted (adj.) R-square coefficient as large as 0.99992 and a slope efficiency as high as 10.99%, indicating an outstanding output power performance of the MW-REDFL. Besides, it can be seen that the threshold of the fiber laser is as low as 23.51 mW, verifying the efficient laser cavity design of the proposed MW-REDFL.

 

Fig. 10. Total output power variation versus the pump power.

Download Full Size | PDF

Although the 2-km long SMF used as the R-DFB medium is sufficient for obtaining outstanding laser output of six-wavelength operation for the proposed MW-REDFL, the SMF length may not be the best choice. Figure 11 shows the comparison of output spectra when different lengths of SMF used as the R-DFB medium, under a 250 mW pump power for demonstration, measured by the OSA with a typical scan for each fiber-length used. As can been seen, for different lengths of R-DFB medium, stable six-wavelength lasing can be guaranteed for all the cases. Additionally, we can see from Fig. 11 and Table 1, obtained from Fig. 11, that almost a longer SMF can bring a higher OSNR but a larger d_p and more severe unstable lasing spikes. In Table 1, OSNR_min denotes the minimum OSNR among six lasing channels. Therefore, under a certain pump power for the MW-REDFL, it may have a trade-off of preferred SMF-length as the R-DFB medium to obtain an acceptable optical spectrum, such as using 2∼5 km SMF at the pump power of 250 mW.

 

Fig. 11. Comparison of lasing spectra using different lengths of SMF as the R-DFB medium at 250 mW pump power.

Download Full Size | PDF

Table 1. d_p and OSNR_min among six lasing channels versus different lengths of SMF.

View Table

Note that the generated wavelengths and the number of lasing channels of the MW-REDFL are strictly associated with the reflection wavelengths of the SI-FBG, which may be modified by the combination of applying different stretching forces to the photo-sensitive fibers and using phase masks with different periods. However, for a desired reflectivity, practically there will be a limitation on the maximum number of FBGs that is possible to write in a same fiber region without cross-disturbances, and, besides, many times repeated inscribing of FBGs in a same region might damage the fiber eventually. Additionally, the femtosecond inscription method [32,51] may be an alternatively promising technique for achieving greater flexibility to obtain any SI-FBG with desired reflections.

4. Conclusions

We proposed and experimentally demonstrated a high stable MW-REDFL based on a half-open cavity structure with a reflecting comb filter of 6SI-FBG as a mirror and a 2 km long SMF-28 fiber as R-DFB medium for the first time. Six lasing wavelength-channels at about 1544.330 nm, 1546.360 nm, 1548.366 nm, 1550.328 nm, 1552.380 nm and 1554.316 nm were achieved, and the weak competition among six channels was verified by the measured stable lasing spectra. The high stability of laser output was proved by the d_p of ≤0.930 dB among all six lasing channels in every OSA scan and f_p ≤1.101 dB for each lasing channel in a 80 min time span, the f_λ≤0.032 nm of six-lasing channels in an experimental time of >1.5 h, the output power fluctuation of ≤0.05 dB measured in 4 h at both of the pump power of 100 mW and 200 mW, and the RINs of −125 dB/Hz and −120 dB/Hz for the pump powers of 100 mW and 200 mW, respectively at the frequencies larger than 100 kHz. The wavelength and 3-dB bandwidth variations of six-lasing channels with the increase of pump power were also investigated in detail. Besides, the MW-REDFL was of a low pump threshold of 23.51 mW, a high slope efficiency of 10.99% and an excellent linear relationship between output power and pump power. The performances of the proposed MW-REDFL can be improved further with good temperature compensation and vibration isolation packaging for practical applications.