Until the development of periodically poled nonlinear crystals with high, broadband parametric gain, mode-locking an optical parametric oscillator (OPO) was an unlikely proposition at best. Recently, however, a collaboration among several institutions in France produced what the participants believe is the first actively mode-locked quasi-continuously pumped OPO. They think the work may offer new possibilities in the development of compact and versatile ultrafast optical sources. In a mode-locked laser, an intracavity modulator whose frequency matches the cavity’s round-trip frequency imposes a loss on photons that don’t pass through the modulator during its minimum loss. As a result, the photons bunch up into a pulse that passes through the modulator during its minimum loss. This intracavity circulating pulse builds up until it saturates the gain. Because the amount of energy pumped into the population inversion during one round-trip of the cavity is the same whether or not the modulator is operating, the total number of photons required to saturate the gain is the same in either case. In other words, the effect of the modulator is to bunch all of the photons into a pulse that passes through the modulator during its minimum loss. Such a mode-locked laser can be analyzed mathematically by requiring self-consistency of the intracavity pulse during one trip around the resonator, taking into account the gain medium, the modulator and any other intracavity elements.Figure 1. The cavity length was extended with several folding mirrors to accommodate the frequency of the available acousto-optic modulator. Images ©OSA.Unlike the population inversion in a mode-locked laser, the gain medium in an OPO — i.e., the nonlinear crystal — cannot store energy. Thus, the photon “bunching up” effect in mode-locked lasers is not possible, and OPOs historically have not been considered suitable candidates for mode-locking. However, the recent development of periodically poled nonlinear crystals with ultrahigh parametric gain has made it possible to achieve the threshold for oscillation without photon bunching.Figure 2. The 700-ps mode-locked pulses were separated by the 8.4-ns round-trip transit time of the 2.5-m resonator.The scientists from the Centre National de la Recherche Scientifique’s Laboratoire Aimé Cotton in Orsay and from the Office National d’Études et de Recherche Aérospatiales in Palaiseau arranged their OPO resonator in a long, bow-tie configuration (Figure 1). They selected the ~2.5-m resonator length to accommodate an available 120-MHz acousto-optic modulator and operated the 532-nm-pumped OPO at or near degeneracy to maximize the parametric gain. For the nonlinear crystal, they used a 0.5 × 3 × 30mm MgO-doped periodically poled LiNbO3 crystal from HC Photonics Corp. of Hsinchu, Taiwan.The OPO reached threshold at a surprisingly low pump power of 1.6 W and produced 700-ps pulses (Figure 2). The investigators measured the time evolution of the pump, depleted pump and signal power for both the mode-locked and free-running cases (Figure 3). These plots reveal the fundamental nature of a mode-locked OPO.Figure 3. The pump power was steadily converted to the signal wavelength when the optical parametric oscillator (OPO) was not mode-locked (a). It oscillated at the mode-locking frequency when the OPO was mode-locked (b).When the OPO was not mode-locked, the signal power increased smoothly as a function of time by depleting the pump power in the nonlinear crystal (Figure 3a). (Notice that the curve for the signal is the inverse of that for the depleted pump.) When the OPO was mode-locked, the power oscillated between the signal and the depleted pump (Figure 3b). In other words, the modulator chopped the intracavity power, which varied from the free-running level in the pulse, to a much lower level outside the pulse.Although the observed pulse length was 700 ps, the model based on round-trip self-consistency of the pulse predicts 200 ps. The scientists do not fully understand this anomaly and are working to resolve the matter. Optics Letters, April 1, 2006, pp. 972