Femtosecond pulses with energies in the microjoule range are useful for laser micromachining, surgery, spectroscopy and for other applications. Until recently, the common approach to generating such pulses was to produce low-energy femtosecond pulses in an oscillator, stretch them temporally with a dispersive element (e.g., a prism pair), amplify them and recompress them with another dispersive element. This technique, called chirped-pulse amplification, is successful because the peak power in the stretched pulse is below the damage threshold in the amplifier.An alternative, and significantly less complex, approach to generating these relatively high-energy femtosecond pulses uses mode-locked, diode-pumped solid-state lasers. However, the peak power available from these lasers is limited by nonlinear effects — the so-called B-integral, which increases with the length of the laser rod. Figure 1. The folded resonator had a total length of 9.36 m, corresponding to a mode-locking frequency of 16 MHz. A saturable-absorber mirror mode-locked the laser, and a two-crystal, β-barium-borate Pockels cell (partially) rotated the pulse’s polarization to (partially) dump it from the cavity. Reprinted with permission of Optics Letters. This fact has led a coalition of scientists in Hannover, Germany, to consider thin-disk lasers. They have recently demonstrated what they believe is the first passively modelocked, cavity-dumped thin-disk laser. The laser generated a 1.06-MHz train of cavity-dumped pulses, each having a 680-fs duration and containing 3 μJ.The researchers, associated with Leibniz University and with Laser Zentrum Hannover, used a 110-μm-thick disk of 10-percent-doped Yb:KYW as the gain medium in their 9.36-m-long resonator (Figure 1). They pumped the disk with up to 50 W of 980-nm light from a laser diode, using a parabolic mirror and turning prisms to pass the pump light through the disk 24 times. Their best results, in terms of short pulses with high energy, came when the entire resonator was purged with helium gas, whose nonlinearity is less than that of air.The saturable-absorber mirror at one end of the resonator mode-locked the laser in the single-pulse regime at 16 MHz. After every 15 round-trips of the resonator, the intracavity pulse’s polarization was partially rotated by the Pockels cell, so ∼23 percent of the pulse energy was ejected from the resonator by the thin-film polarizer (Figure 2). By varying the voltage applied to the Pockels cell, the researchers could adjust the percentage of the intracavity pulse that was dumped from the cavity.Figure 2. This plot shows the intracavity intensity at a fixed point inside the resonator. After the intracavity pulse made 15 round-trips of the resonator and built up an energy of ~13 μJ, the Pockels cell rotated its polarization sufficiently so that ~23 percent of its energy reflected off the thin-film polarizer and emerged from the laser.But there are several inherent drawbacks to using thin disks as the gain medium in mode-locked lasers. Even a disk that is wedged and has an antireflection coating produces residual intracavity etalon effects, which tend to discriminate against certain longitudinal modes and thereby reduce the number that are available for locking.In addition, because the thin gain medium is hard against a mirror, the standing waves of all longitudinal modes have almost the same shape inside the disk, and they compete with each other for the available gain. This effect also reduces the number of longitudinal modes available for locking. As a result of these drawbacks, the 680-fs pulses the scientists obtained from their thin-disk laser were significantly longer than the 370-fs pulses they obtained from oscillators with rod-type gain media. Nonetheless, the 3-μJ pulse energies combined with the megahertz repetition frequency represent a new regime for mode-locked, cavity-dumped, diode-pumped solid-state lasers.Optics Letters, June 1, 2007, pp. 1593-1595.