With high efficiency and exceptional beam quality, fiber lasers hold great promise. The 1-µm wavelength makes it possible for them to replace both diode- and lamp-pumped Nd:YAG lasers in the micromachining, automotive and biomedical markets.
Qihong Lou in his laboratory with the coherent beam combination experimental setup.
Fiber lasers with up to kW-level output power are now technologically sophisticated, and fiber-based laser systems for industrial applications have hit the market. These lasers are being developed not only for continuous-wave and long-pulsed-mode operation but also in the picosecond pulse mode.
Although kW-level systems were realized in the lab several years ago, it has taken a while for them to make the leap into industry for several reasons, including diode bar lifetime and diode price. However, diode-pumped fiber lasers offer many advantages, including high efficiency, stable emission, a long lifetime and low energy consumption. Their key parameters are described as follows.
Besides output power, the most relevant parameter of a laser system is the beam quality, which is characterized by the beam propagation factor M2. Beam quality is defined as the product of second intensity moments, measured in the near and far field. For double cladding high-power fiber lasers, the M2 = 1.1-1.3, with a core diameter of less than 10 µm. For large core double cladding with a core diameter between 20 and 40 µm, the M2 is increased to 1.5-3.0.
For many industry applications, the beam parameter product (BPP) roughly holds for the 86.5 percent power content values. It is given as:
BPP= M2 • λ/π , where λ is light wavelength.
The efficiency is an important parameter for the running cost and the investment. Of interest is the plug efficiency, which is defined as:
Plug efficiency=laser output power/total electrical input power
The total input electric power includes the power for the cooling units, the control system and the efficiency of the power supply. For a fiber laser system, it is around 20 to 30 percent—which is roughly 10 times higher than the efficiency of a lamp-pumped solid state laser system.
Fluorescence spectrum of China-made double cladding fiber.
Lamp-pumped laser systems are limited by the lifetime of the lamp, which is about 500 hours. In the production line, laser systems are running 20 hours each day. Lamps are replaced every three weeks to ensure safety—which is not so much a problem of cost, but rather a loss of production time. Diode-pumped fiber lasers have longer lifetimes compared with lamp-pumped solid state lasers.
This article reports on the development of fiber lasers, with a special focus on the work we have done at the Shanghai Institute of Optics and Fine Mechanics (SIOM).
Kilowatt-level continuous wave fiber lasers
Optical fibers play an important role in transferring the optical signal; they also act as host medium in fiber lasers and amplifiers when the rare-earth elements are doped into the fiber core. Neodymium (Nd3+) and erbium (Er3+) are the most common rare-earth elements used as dopants. Recently, many research organizations have achieved very high efficiencies, diffraction-limited beam quality, and output powers of more than 1 kW with ytterbium-doped double-clad fiber lasers.
The most common rare-earth dopant in silica glass fiber is ytterbium (Yb3+), which is becoming the geometry of choice for high-power fiber lasers. The laser transition is 2F5/2 to 2F7/2 with the terminal level 623 cm-1 above the ground state. The thermal energy at room temperature is 200 cm-1; therefore, the terminal state is thermally populated—which makes Yb3+ a quasi-three-level system.
By comparison, the terminal laser level in Nd3+ is about 2,000 cm-1 above the ground state. Pumping of Yb-doped glass fiber around 915 nm, 945 nm or 975 nm produces the smallest amount of heating compared to any other major laser system. Actually, the pump radiation in this material generates only about one-third of the heat compared to Nd-doped glass laser.
Characteristics of two types of China-made double cladding fiber.
The fractional thermal loading is around 11 percent for a Yb-doped glass laser pumped at 945 nm and 32 percent for an Nd-doped glass laser pumped at 808 nm. This substantially reduced thermal dissipation is the result of a very small energy difference between the photons of the pump and laser radiation. This quantum defect, or Stokes shift, is 9 percent in Yb-doped glass versus 24 percent in Nd-doped glass.
The thermal load generated in a laser medium is of primary concern for high-power applications. The reduced thermal heat load can potentially lead to higher-than-average power systems with better beam quality than is possible with an Nd-doped material.
The simplest and most efficient way to pump double-clad fibers with high-power pump sources is to couple the beam into the inner cladding of double-clad fiber through its ends—the so-called end-pumping scheme. In this configuration, a large inner cladding is required in order to accommodate the large pump beam of the high-power laser diode source, which is either coupled to the double-clad fiber with bulk optics or fiber optics.
The bulk optics approach, in which the pump light is launched into the end of the fiber, is often used in laboratory applications. A major drawback is that one or both of the fiber ends are obstructed by the bulk optics that are used to launch the pump light. In addition, this approach lacks scalability (one fiber has only two ends) and is difficult to implement in a compact and rugged manner.
Fiber laser with one-end pumping using spatial filter.
Kilowatt-level continuous wave fiber laser system. By using this high power fiber laser system with 200 W output power (left), one can weld two steel plates with 2 mm thickness (right).
However, most double-clad fibers made to date use a low-index polymer as the outer clad material in order to achieve the desired high numerical aperture (NA≈0.3–0.45). These polymers have much poorer thermal stability than glass. In high-power applications, the polymer near the inner-outer clad interface can easily burn or gradually degrade during the high-power pump.
Due to the possibility of faulty steps in beam shaping and assembling, the high-power collimated pumping beam is not so good for pumping the double-clad fiber directly. A spatial filter can be used to improve the beam quality of the high-power pump light. A special aspheric lens has been designed and fabricated to inject the pump light into the inner cladding with high coupling efficiency.
The focus spot and the cone angle of the pump light match well with the corresponding parameters of the double-clad fiber (Chin. Phys. Lett., 21, 1083); they benefit from the diffraction-limited performance of the aspheric lens, as well as the optical spatial filter. And the pump light can be safely and efficiently coupled into the inner cladding. In 2004, we found that the maximum laser output power of 20-m double-clad fiber was more than 200 W at 1.1 µm for one-end pumping, with a slope efficiency of more than 69 percent.
By 2006, with two ends pumping configuration, the laser output power had increased to 1,050 W with a pump power of 1,500 W.
Q-switched pulsed fiber laser
In recent years, the laser community has shown increasing interest in Q-switched fiber lasers, due to their advantages of compactness, high efficiency and high spatial beam quality. Such Q-switched double-clad fiber lasers with short, high-energy pulses are preferred in applications such as range finding, remote sensing, medicine and laser processing. Several acousto-optic Q-switched double-clad fiber lasers have been reported over the past few years (Opt. Lett. 32, 2774 and Opt. Lett. 23, 454 and 1683).
In previous work, researchers have used the acousto-optic Q-switch as a traditional loss modulator of cavity in fiber laser. Typically, one can obtain pulses with long-pulse width (several hundred nanoseconds) and lower repetition rate because of the fiber laser configuration.
In order to obtain shorter pulses and higher repetition rates, scientists will need to consider high pump power and high doping density fiber. At SIOM, we reported a novel acousto-optic Q-switching operation mechanism. Meanwhile, another research group demonstrated stable compressed pulses (35 ns) at a repetition rate of 1-50 kHz with lower absorbed pump power (Opt. Lett., 25, 37).
The Q-switch Yb-doped double-clad (YDDC) fiber laser. Diagram of the Q-switch YDDC laser setup (top) and profiles of the regular Q-switched pulse (bottom, left) and compressed Q-switched pulse (bottom, right).
By using a configuration of the Yb-doped double-clad fiber laser cavity with a Fabry-Perot cavity, researchers have obtained conventional repetition-rate-stabilized Q-switched pulses when the gate time width of the acousto-optic modulator (AOM) is several microseconds. The pulse width ranges from 600 ns to near a microsecond at the repetition frequency of 50 kHz. When the gate time width of the AOM is regulated to several hundred nanoseconds, interesting phenomena appear. The pulse width is sharply compressed with a stable repetition rate and high-peak power.
When the absorbed pump power is 5W, a 20-µJ pulse with a sub-40-ns pulse width is achieved at the repetition rate of 50 kHz. As far as we know, this pulse width appears to be one of the shortest ever reported in the Q-switched double-clad fiber laser at 50 kHz. The stability of the pulse-to-pulse was found to be above 90 percent. The output beam quality factor M2 is about 2. When another 4-m-long fiber is used, similar phenomena are observed and a 36-µJ, 58-ns pulse is obtained at 50 kHz.
During the short gate time, the Q-switched pulse does not have enough time to build into a full Q-switched pulse profile. Accordingly, we observed a short fall time of pulse with our experiment. However, the small pulse is not terminated because a part of the inversion population remains along the fiber when the AOM is closed. During the time, the fiber acts as an amplifier and continues supplying the energy for the “seed” pulse. With the optimal gating time, a higher power peak pulse can be achieved after its double-pass amplifier along the fiber.
The coherent combination of four beams
The output power of the single fiber laser has been rapidly improved and now exceeds kilowatt magnitude. However, the ultimate output is limited by nonlinear effects such as stimulated Raman scattering and stimulated Brillouin scattering. Beam combination is an effective geometry that can improve output power with excellent beam quality. Many researchers have brought up various techniques of beam combination.
The MOPA (master oscillator and power amplifier) system makes use of active phase correction; it involves complicated interferometric detection and phase modulation of each element of a fiber laser array. The self-organization of a multicore fiber array and the all-fiber coherent beam combining can’t actually avoid the ultimate power limitation of a single fiber, and the fabrication of a multicore fiber involves very complicated processing.
Phase locking 2 x 2 fiber laser array with a self-imaging resonator and a spatial filter.
Because the principle of the self-Fourier resonator is the same as the Talbot cavity, the resonator involves complicated and rapidly evolving field amplitudes and wavefronts, and it can’t operate in single-mode. By using Vernier-Michelson configuration, the whole system becomes increasingly complicated and difficult to align because the beam splitter must be increased with the increasing numbers of laser elements.
The wavelength beam combination (WBC) approach, which uses a dispersive element to combine all the beams in both near- and far-fields, requires each fiber laser working at a slightly different wavelength, and the linewidth of the element must be as narrow as possible.
For various operation mediums such as Nd:YVO4 and Nd:YAG, researchers have demonstrated the coherent addition of laser arrays by using a self-image resonator (Opt. Lett. 21, 1996; Appl. Phys. Lett. 84, 302512-14; and Appl. Phys. Lett. 85, 4837). With this method, the phase correction of fiber lasers is passive and is realized by means of a self-adjusting process of the resonance frequencies of the fiber lasers’ array in order to adapt to changes in the optical path lengths. Thus, the system is quite compact and steady.
Furthermore, this method does not need polarization controlling of each fiber laser beam, because a polarization eigen state can always be found in the two elements of the system, regardless of their relative orientation (Appl. Phys. Lett. 85, 4837). Using this method, one can couple a number of fiber lasers with different lengths into a common self-imaging resonator with a spatial filter for phase locking. The configuration should be improved to achieve a high-power level for the thermal tolerance of spatial filters and a high fill-factor of the individual beam in the array.
By using the self-imaging resonator with a spatial filter, we have demonstrated the phase-locking of two fiber lasers with different fiber lengths. The patterns of the beam profile at the output mirror exhibit high-contrast interference lobes. Even if the individual laser optic length changes, these lobes are still in a state of relative stability. For a pump power of 126 W, a coherent output power of 26 W is obtained, and the corresponding coherent power combination efficiency is as high as 92 percent.
Beam profile at the collimators of the near field. Calculated beam profiles (a) at the collimators, in the far field for coherent Gaussian beams with various phase difference between adjacent elements, (b) when ΔΦ=π, and (c) when ΔΦ=0. The dotted lines mark the position of the spatial filter for mode control. The experimental beam profiles are also shown for (d) the near field and the far field with (e) out-of-phase mode and (f) in-phase mode.
The self-imaging resonator consists of four identical small plano-convex lenses at the plane M1, a convergent lens (L) acts as Fourier transform lens in this system, and a flat semitransparent output mirror (M2) is used with 50 percent transmission at 1,080 nm. Four 975-nm diode lasers serve as a pump source in the phase-locking experiment. Four fiber lasers with different lengths use the end-pumping configuration. The four collimated beams are placed symmetrically about the resonator optical axis on the front focal planes of Fourier transform lens L. The semitransparent output mirror M2 is set on the back focal panes of the lens.
The beam profiles at the front focal plane (M1) and at the output plane (M2) of the system are related to each other through a Fourier transform. Thus, a beam profile symmetrically about the optical axis reproduces itself after a round-trip in the resonator. Two counter-propagating waves are coupled. We examine the beam profile of the output at the output mirror (M2) by means of a CCD camera and a laser beam analyzer.
The computer-simulated beam profiles at the collimators and at the far field for various phase differences between adjacent elements are shown in parts (a), (b) and (c) of the figure above. Part (d) shows the beam profile at the collimators of the near field. From the phase control standpoint, a fiber laser array is different from other single gain elements with fixed length. The length of the fiber changes as the temperature changes, so it is impossible to ensure equal lengths in the fiber laser array.
There isn’t a fixed phase difference of elements for a fiber laser when it is running freely without a spatial filter; thus, the beam profile exhibits low-contrast lobes that are constantly moving irregularly. The phenomena result from the competition of the modes. In order to stabilize the phase, we have placed a spatial filter at the mirror M2 for low loss of the needed mode and for the high loss of the other modes. We have observed the beam profiles of the in-phase mode in this experiment and the full width of the central lobe is 0.3 mrad in (e) and (f).
When the whole system operates under the high power condition, the physical properties of the spatial filter do not cause any changes (glowing, breaking, melting and so on); this means that the filter may tolerate the high power. There is no doubt that the coherent output power can be increased greatly by the same method if we optimize the parameters of the resonator and the fiber laser array.
[Qihong Lou, Jun Zhou, Bing He and Hongming Zhao are with the Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai, China.]
References and Resources
>> S. Ménard et al. Opt. Lett. 21, 1996-8 (1996).
>> B. Rossi. Laser Focus World 33, 143 (1997).
>> Z.J. Chen et al. Opt. Lett. 23, 454 (1998).
>> H.L. Offerhaus et al. Opt. Lett. 23, 1683 (1998).
>> J.A. Alavarer-Chavez et al. Opt. Lett. 25, 37 (2000).
>> E.J. Bochove et al. Opt. Lett. 28, 1200 (2003).
>> D. Sabourdy et al. Opt. Express 11, 87 (2003).
>> J. Zhou et al. Chin. Phys. Lett. 21, 1083 (2004).
>> Y. Zhou et al. Appl. Phys. Lett. 84, 3025-27 (2004).
>> L. Liu et al. Appl. Phys. Lett. 85, 4837 (2004).
>> C.J. Corcoran and F. Durville. Appl. Phys. Lett. 86, 201118 (2005).
>> B. He et al. Opt. Express 14, 2721, (2006).
>> M. Fridman et al. Opt. Lett. 32, 790 (2007).
>> T.H. Loftus et al. Opt. Lett. 32, 349 (2007).
>> H. Zhao et al. Opt. Lett. 32, 2774 (2007).