1. Introduction

The laser sources with high brightness and high average power are widely used in many fields, such as laser radar, free space communications, directed energy, and industrial machining. Coherent beam combining of multiple fiber lasers with moderate output power and good beam quality has been expected to be an effective way to obtain a laser source of this kind [1–3], since fiber lasers offer the advantages of compactness, high efficiency, good beam quality and low running cost, and their limited output from a single fiber laser can be overcome by coherently combining multiple fiber lasers. In recent years, various techniques or mechanisms have been studied to obtain efficient coherent beam combining [4–30], and these techniques can be categorized into two broad classes [2]: collinear interferometric summation methods and aperture-filling methods. For the first class, the component laser beams are coherently combined inside the cavity due to the mutual injection coupling or self-organization process, and the coherent array usually owns only one output port (the others are designed as high loss ports), thus the single combined output is mainly extracted from the common port and are obviously coherent in both near and far fields. However, if the common output port is a fiber port [4–15], the ultimate power limitations of a single fiber laser still can not be avoided; if the beams is combined in the common external cavity and extracted from output coupling mirrors [16–19], scaling the array to a large number of fiber lasers to obtain a high brightness source is a quite difficult job. For the second class, the phase locking array emits a bundle of laser beams with some spaces in the near field, and the beams are coherently added in the spatial overlapped area. If the filling factor is high enough, most of the energy will be concentrated on the central lobe and its brightness will be maximized in the far field [12, 20–28]. Generally speaking, when aperture-filling methods are used to obtain efficient coherent combining in the far field, active phase controlling or mutual injection coupling measures need to be adopted. Whereas, active phase controlling technique requires complicated phase detection and phase modulation for each element of the array, when the number of combining elements is large and output power is high, the scaling difficulty will increase remarkably [25–27]. A simple and effective method, to obtain a phase locking array, may be introducing the mutual injection coupling among the component lasers, and the schemes reported in Refs [12, 28] are designed under this principle.

In this paper, we propose a phase locking array of multiple fiber lasers with an all-fiber coupling loop in terms of this principle. Thanks to the special designed all-fiber coupling loop, it not only provides a common ring cavity for mode selection and stabilization of component lasers, but also provides a common reference cavity for mutual injection coupling among them. Since the phase-locked lasers are extracted from multiple fiber ports, compared with the tree structure with one common output port [4–15], the thermal management and expandability of our array are much more convenient. Moreover, the array is an all-fiber arrangement, thus the robustness and reliability are improved compared to the array with an external cavity [13, 19–24]. As a proof of principle, we have demonstrated efficient phase locking of three fiber lasers employing the all-fiber coupling loop, and the output characteristics and far field interference patterns are investigated, which all have proved the feasibility of the configuration.

2. Experimental setup

The experimental setup is schematically presented in Fig. 1. Three component fiber lasers all employ the typical linear resonator which is formed by a fiber Bragg grating (FBG) and 4% Fresnel reflection at the output cleaved ends, and their Bragg center wavelengths are all close to 1550nm. The gain fibers are single-mode Er-doped fibers (EDF), and their lengths are 9m, 10m and 11.6m. Three 2×2 polarization insensitive fiber couplers (PIFC) are connected with the output ports of three individual fiber lasers, and their coupling ratio are all 80/20, i.e. 80% is extracted by the flat-cleaved fiber collimator (FC), and 20% is coupled into the fiber loop as the signal of mutual injection. Three isolators (ISO) are added in the fiber loop to keep the coupling laser beam circulating in a single direction in the loop. Three fiber collimators are placed in a trigonal array, and the spaces among them are nearly 5mm. A positive lens with the focal length of 50cm is used to make the output discrete beams convergent, and an infrared CCD is placed on the rear focal plane of the lens to record the interference patterns.

 

Fig. 1. Experimental setup of phase locking array of three fiber lasers with an all-fiber coupling loop (marked in blue line). LD, 980nm laser diode; WDM, wavelength division multiplexer.

Download Full Size | PDF

The key point of the scheme is that an all-fiber coupling loop is introduced into the array, and the common loop is made up of multiple PIFCs. To our knowledge, the PIFCs are widely used in the coherent combining of multiple fiber lasers, and usually they are connected to a structure of tree or ladder, but they are firstly configured like this. The loop presented here is indeed a passive ring reference cavity, which is shared by all the component lasers of the array, and its own set of eigen modes can be coupled with all the component lasers. Considering the pigtail lengths of WDM, PIFC, ISO and FC, the path lengths of three linear cavities are nearly 12m, 13.2m and 14.7m, and the round trip path length of the reference cavity is nearly 17m. Therefore, their longitudinal mode spaces are narrow, and it is easy for them to find common modes to resonate. To improve the stability of the array, three ISOs are introduced to restrain the parasitic cavities in the array. Before adding the ISOs, there are lots of useless parasitic cavities in the array, e.g. the cavity between FBG₁ and FC₂, FBG₁ and FC₃, etc; and it is quite difficult for all the cavities to find common modes to resonate as a result of Vernier effect. While after adding the ISOs, just three linear cavities and their common reference cavity exist in the array, thus it is easy for them to find common modes, and the experimental results have showed that the ISOs are quite necessary to keep the phase-locking array operate stably.

3. Results and discussion

The far field interference patterns of the output lasers are recorded by laser beam analyzer (Spiricon, LBA-PC300), which are shown in Fig. 2. Figure 2(a)–2(d) are the interference patterns of laser1+2+3(phase locking array), laser1+3, laser2+3 and laser1+2, respectively, when three pump power all are 100mw. Figure 2(b)–2(d) are recorded when one of the output is blocked by a piece of cardboard in the near field. Figure 2(e) and 2(f) are the calculated far field interference pattern and profile when three equal amplitude in-phase Gaussian beams are placed in the same way as our phase locking array. The visibility recorded in Fig. 2(a) is nearly 0.5, and corresponding degree of coherence is slightly larger than this value for unequal output power of component lasers. The calculated visibility and degree of coherence all are 1, and the calculated profile shows that an inferior maximum exists between two maxima, which do not quite clearly emerge in Fig. 2(a). The unapparent inferior maximum is due to the fringe patterns drifting slowly and the visibility is not very good. The slowly drifting can be owing to without special measures are taken for spatial mode selection, i.e. no spatial filter is employed to bring loss difference between the in-phase mode and out-of-phase mode. The decrease of visibility and degree of coherence can be attributed to the polarization perturbations, since without any polarization controlling measures are taken in this array, and precisely controlling the polarization states of each component of an array is necessary to obtain stable and high-contrast patterns. The large number of speckles appeared in Fig. 2(a) is due to the poor filling factor in the near field, and improving the aperture-filling can reduce these speckles and concentrate more energy in the central lobe. In conclusion, we believe that the experimental results are basically agreement with our theoretical calculations, and the three-cell fiber laser array and any two of them all operate, at least partially operate in phase locking states.

 

Fig. 2. The interference patterns of output lasers recorded by LBA. The truncation appeared in Fig. 2(a) and 2(b) is owing to the saturation of the infrared CCD. In fact, to avoid obvious saturation of the CCD, for its dynamic response range is limited, Fig. 2(a) is recorded when the attenuating system is much stronger than Fig. 2(b)–2(d) are recorded. The unit of coordinates in Fig. 2(f) is the waist radius of Gaussian beam, and its size in our fiber laser is nearly 0.25mm after the collimator.

Download Full Size | PDF

The coherent output characteristics of the phase locking array are researched through the optical spectrum analyzer (OSA, Agilent 86142A) and power meter (ILX Lightwave, FPM 8210H). The recorded spectrums and output power of laser1, laser2, laser3 and phase locking array are shown in Fig. 3(a). Since we adopt the bare fiber patchcord of OSA to accept output power of the array, just part of power can be coupled into the bare fiber due to its small dimension and low numerical aperture (NA), thus we have proportionally magnified the spectrum in terms of the measured output by power meter for the purpose of comparison. The spectrums and output power of component lasers are obtained when the coupling loop is removed from the array. According the data recorded by OSA, we have found that laser1, laser2 and laser3 all operate alone at different lasing wavelength of 1550.06 nm, 1550.17nm and 1550.23nm, respectively. After the coupling loop is introduced into the array, i.e. when mutual injection locking occurs among them, the array operates at 1550.19nm, which is between the lasing wavelength of laser2 and laser3, thus we believe that three lasers nearly operate on the same longitudinal modes, and the mutual injection locking makes the array self-adjust and oscillate on the common longitudinal modes. The output power of component lasers are checked by the power meter, when the pump power all are 100mw, the output of laser1, laser2 and laser3 are 30mw, 35.6 mw and 39.4mw, respectively, and the output power of the array is 94mw and without evident perturbations is observed in half an hour, thus the combining efficiency is close to 90% and the output power is stable.

 

Fig. 3. (a) The typical spectrums of laser1, laser2, laser3 and phase locking array, which are shown in color red, blue, green and black. This Fig. is obtained by superposing four original Figs. together in one coordinates for the purpose of comparison and saving space of the paper. (b) The equivalent sketch of the experimental setup shown in Fig. 1.

Download Full Size | PDF

The phase locking process can be generally explained by the mutual injection locking [30] or the self-organized mechanism [31]. An equivalent sketch which can reveal the connectivity more directly is presented in Fig. 3(b), and it is helpful to understand the mechanism in details. The common reference cavity is actually a fiber ring resonate filter for wavelength selection and stabilization, it has its own set of eigen modes. Each linear cavity also has their own set of eigen modes, and these modes which also exist in the reference cavity will be selectively excited as supermodes of the composite cavity. The phase locking array can be considered a multi-arm composite cavity with different arm lengths, and the modes satisfying the resonance conditions (i.e. constructive interference after one circulation) both in the common reference cavity and three linear cavities have the lowest loss, i.e. their threshold are the lowest, and they will be selectively excited as supermodes of the array. Supposing the cavity lengths of the common ring cavity and the linear cavities are L and L_i (i is the number of fiber laser), the resonance phase conditions are kL=2nπ and kL_i=m_iπ, k=2π/λ, λ is the oscillating wavelength, n, m_i are integers, and the resonance amplitude conditions can be generally summarized as the laser gain is larger than the total loss of the cavity. When the optical path length of one of the components varies, a new set of mutual modes have to emerge to satisfy the resonance conditions again. In other words, each component laser of the array can adjust its frequency and phase automatically to form mutual modes even though there exist some differences in frequency and phase when they operate alone. The self-adjust processes can not always occur efficiently unless some conditions are satisfied [23], it works best in the laser array with broad gain bandwidth, long and unequal cavity lengths and low-Q resonator. In our phase locking fiber laser array, these conditions are basically satisfied.

4. Conclusions

We have proposed and demonstrated a novel phase locking method of fiber laser array by an all-fiber coupling loop, and the interference patterns and output spectrums characteristics are investigated in details, which all provide convictive reasons to prove the occurrence of phase locking. Furthermore, the array can be conveniently scaled up to more elements with higher output power to obtain a high brightness laser source in the far field, and the scaling scheme is shown in Fig. 4. Compared with the tree or similar configurations using fiber couplers, the heat dissipation of our array is evidently improved, but the essential phase locking mechanism is unchanged, it still need to select the common modes to oscillate in the composite cavity, thus the limit of coherent combining mentioned in Ref [32] is not removed in our array. However, I believe the number of scalable components may not be limited so strictly. On one hand their estimate is done under too much hypothesis and ignores the possibility of mutual adjustment of frequency and phase; on the other hand I believe the self-organization theory based on nonlinear coupling [31] is superior to the longitudinal mode overlapped model [32] in explaining the phase locking phenomena. In all, how many elements can be coherently and efficiently combined need to be verified through further theoretical and experimental study.

 

Fig. 4. The scaling scheme of multiple fiber lasers.

Download Full Size | PDF

Although the scheme is quite easy to implement and owns lots of advantages, there still exist some problems, such as the interference patterns are not very stable and the visibility is not good. The slowly transverse move of far field pattern can attribute to no special measures are taken in spatial mode selection, and only phase self-adjust is not enough to obtain a stable in-phase output. The low visibility is due to the uncontrolled polarization state of component laser, which can expect to be solved by adding some polarization controllers (PC) or using polarization-maintaining fiber elements in the component lasers. Moreover, when the output power of component laser is high, the 2×2 fiber couplers and ISOs which could bear high power density are needed to be specially designed, or finding similar elements to replace them to form the coupling loop. The coupling ratio of the fiber coupler also need to be optimized according to the cavity loss and combining efficiency of the array, the mechanism of mutual injection locking or self-organization, the possibility and details of nonlinear coupling, power scalability, also need to be included in our future work.