Laser users requiring multiwatt outputs from mode-locked lasers have been forced to resort to bulk lasers (i.e., those based on rods or slabs) or on complex and expensive master oscillator power amplifier fiber configurations. Recently, however, investigators at the University of Arizona’s College of Optical Sciences in Tucson demonstrated a simple, short, all-fiber oscillator capable of generating as much as 2.4 W of average power at a 95-MHz mode-locking frequency. The key was a gain medium of soft phosphate-glass fiber, which can support much higher doping levels than other fibers. The investigators previously showed that an average power of 1 W/cm of fiber can be obtained from heavily doped phosphate-glass fiber. They now have used a 20-cm length of such fiber as the gain medium of a mode-locked laser. The short length finesses one of the most vexing issues in the design of high-power fiber lasers. Problematic nonlinear effects, such as stimulated Brillouin scattering, stimulated Raman scattering and four-wave mixing, increase with the product of power density in the fiber and of the fiber’s length. Most attempts to overcome these problems have involved increasing the size of the fiber core so that the power density is decreased. An alternative, upon which the Arizona researchers seized, is to shorten the fiber. This approach is enabled by the high doping possible in phosphate-glass fiber. Figure 1. The length of the resonator — from the 100 percent mirror on the left to the semiconductor saturable absorber mirror at right — was approximately 1 m, corresponding to a mode-locking frequency of 95 MHz. Images ©OSA. They spliced normal single-mode fibers to both ends of the doped fiber and side-pumped it with up to 22 W of 975-nm radiation from fiber-pigtailed diode lasers (Figure 1). The laser resonator was formed by a 100 percent reflecting mirror deposited on one end of the single-mode fiber and a semiconductor saturable absorber mirror (SESAM) at the other end. One of the difficulties in designing the laser was reducing the power density sufficiently at the SESAM; had the latter simply been butt-coupled to the single-mode fiber, the saturation intensity would have been achieved quickly and the absorber would have opened prematurely, allowing a very long pulse to oscillate in the resonator.Two design features minimized this. First, the investigators coupled 80 percent of the circulating power out of the resonator with a fused coupler so that the power that reached the SESAM was only a quarter of the output power. Second, they adiabatically tapered the fiber leading to the mirror structure so that the beam expanded by a factor of four (i.e., its area increased by a factor of 16) on it. When they replaced the SESAM with a broadband reflector, they observed 2.5 W of 1.56-µm output power from the 22 W of pump power. With the SESAM in place, they saw stable mode-locking from 400 mW up to 2.4 W of output. The spectral width of the output was ∼5 nm, and the duration of the presumably Gaussian pulses varied from 8 ps at the lowest output power to 44 ps at the highest (Figure 2). Figure 2. The linear dependence of the pulse duration on output power resulted from the saturation dynamics of the semiconductor saturable absorber mirror.This dependence of pulse duration on output power resulted from the saturation dynamic of the SESAM. It narrows the intracavity pulse — and increases the power density incident on itself — until it reaches a power density that saturates the absorption. Once the absorption is completely gone, no further shortening occurs. In other words, the saturable absorber clamps the peak power to a given value and adjusts the pulse duration to maintain that peak power. In this case, the clamped value for peak power was 540 W. Breck Hitz Optics Letters, March 1, 2006, pp. 592-594.