Kilowatt-class, all-fiber amplifiers for beam combining

For industrial and directed energy (DE) applications, fiber lasers have the advantages of reduced size, weight, and power consumption compared with their bulk solid-state and chemical counterparts. However, while multimode fiber lasers with powers as high as 100kW are now commercially available,¹ these have poor beam quality. They are suitable for material processing but not DE applications, which require fiber lasers with near-diffraction-limited beam quality, and are much more challenging to build.

To scale overall power and brightness, developers have pursued multiple fiber laser beam-combining techniques, which can be broadly categorized into incoherent and coherent combining. In one subset of incoherent beam combining (IBC), an array of lasers is superimposed in the far field, but without control of the relative spectra or phases of different elements. Although there have been successful demonstrations of such beam combining,² for DE applications this approach is limited to propagation ranges on the order of a few kilometers. Spectral beam combining (SBC) is another subset of IBC, where lasers of different wavelengths are spatially overlapped—using (for example) a multi-layer dielectric grating³—to create a single beam of multiple colors. SBC has the advantage of not requiring optical path length matching and phase control, but can be limited by the laser gain bandwidth. Alternatively, in coherent beam combining (CBC) all channels operate at the same wavelength but require optical path length matching and active phase control. Both SBC and CBC are therefore well suited for longer-range DE applications.

Regardless of which approach is used, the output power from individual all-fiber amplifiers (i.e., no free-space optics) should be as high as possible, while also being spectrally narrow. A primary limitation to high-power, narrow-linewidth fiber lasers is stimulated Brillouin scattering (SBS), which causes light to be scattered in the opposite direction to the incoming beam. It is possible to overcome this problem using SBS-suppression techniques, such as spectral linewidth broadening, which we can achieve by phase modulation. One way to do this is by radio-frequency modulation, where suppression occurs when the phase variation is shorter than the phonon lifetime. Using this approach, developers recently demonstrated narrow-linewidth, kW-class, ytterbium (Yb)-doped all-fiber amplifiers, and showed beam combining of amplifiers modulated at the gigahertz level for both SBC and CBC. Based on this premise, we have presented a theoretical study of SBS in fibers seeded with phase-modulated light.⁴ Using a time-dependent model, we investigated different modulation schemes in an attempt to achieve maximum SBS suppression at the smallest possible effective spectral linewidth.

One standard technique for phase modulation is to drive an electro-optic crystal with a white noise source (WNS). Specifically, WNS creates a continuum spectrum with a line shape dependent on the voltage and RF filtering. Notably, there are examples of using WNS modulation to attain kW-class, all-fiber amplifiers at >10GHz linewidths.⁵ However, despite promising results, scaling to multi-kW levels has proven difficult, with broader linewidths (>20GHz) envisioned at 2kW. Such linewidths may hinder efficient combining because of the added path length complexities in CBC, where the linewidth is inversely proportional to the temporal coherence. In SBC, broader linewidths reduce the number of channels that can be combined, and can also degrade the beam quality.

To further scale amplifier power for efficient combining, we have theoretically and experimentally investigated pseudo-random bit sequence (PRBS) modulation. A PRBS pattern is typically denoted as 2ⁿ−1 and contains every possible combination of n number of binary bits, except the null pattern. In contrast to WNS, PRBS generates a discrete optical power spectrum with features that are functions of the modulation frequency and pattern length. Moreover, phase shift keying with pseudo-random patterns can provide significant SBS suppression in long-distance optical communication systems.⁶ Similarly, we have shown that PRBS modulation can deliver considerable linewidth reduction in kW-class fiber lasers.⁷

To test our theory, we built an Yb-doped all-fiber amplifier to investigate power scaling using PRBS-modulated signals. Accordingly, at a modulation frequency of 3GHz, the 2⁵−1 PRBS pattern enabled scaling to 1170W of output power at the SBS threshold (see Figure 1). This represents a significant reduction in linewidth for kilowatt all-fiber amplifiers. More importantly, at these levels of SBS suppression, multi-kW (2–3kW) fiber amplifiers at beam-combinable linewidths (∼10GHz) are highly feasible. Furthermore, as shown in Figure 2, PRBS phase modulation eliminates the need for absolute path length matching.⁸

Figure 1. Signal power and reflectivity (the latter is a measure of the strength of the stimulated Brillouin scattering process) versus pump power for a 3GHz, 1170W fiber amplifier. Inset: M²beam quality and beam profile. Notably, we achieved an optical efficiency of 83% and an M² value of <1.2. Although further power scaling can be achieved by increasing the modulation frequency, we were limited by the power-handling capability of our pump combiner.

Figure 2. Recoherence measurements, where recoherence distance can be controlled by modifying the pattern repetition rate. Here, we used a 2³-1pseudo-random bit sequence 5GHz modulated pattern. By increasing the path length difference to an integer multiple of the spacing between the discrete spectral lines, we could restore coherence.

Recently, we extended PRBS coherent beam combining to multi-kW levels. As such, we have demonstrated CBC of five PRBS-modulated 1.2kW all-fiber amplifiers with a diffractive optical element (DOE). Specifically, with the 1×5 DOE, we achieved a total combined output power of 5kW with 82% combining efficiency (see Figure 3). Furthermore, we attained near-diffraction-limited beam quality with an M² of 1.06 (see Figure 3, inset). We believe that, with improvements to our experiments, we may achieve combining efficiencies approaching 90%.

Figure 3. Coherently combined output power versus input power. We used a diffractive optical element to overlap the beams. We achieved 5kW of power with 82% combining efficiency and excellent beam quality with M²=1.06(shown in inset).

In summary, improved SBS suppression achieved through PRBS modulation can have a significant impact on the beam combining of kW-class all-fiber lasers. To investigate this, we have demonstrated a 1.17kW amplifier at a modulation frequency of 3GHz, which represents a significant reduction in spectral linewidth for kilowatt all-fiber amplifiers. More recently, we demonstrated multi-kW class CBC of PRBS-modulated fiber amplifiers. We achieved a total combined output power of 5kW, with excellent beam quality.

In future work, we will seek to continue increasing the number of amplifiers combined for SBC and CBS architectures. We will also aim to scale the output power of narrow-linewidth (beam-combinable) single amplifiers, with the objective of attaining the multi-kW regime.