- Parametric Oscillator Produces 9- to 10-µm Output
Researchers at the University of Southampton in the UK have produced what they believe are the longest wavelengths ever generated in a synchronously pumped optical parametric oscillator (OPO). Based on their success to date, they believe that the device eventually may be tunable across an 8- to 12-µm range while producing hundreds of milliwatts of output.
The primary advantage of a synchronously pumped OPO over one that's continuously pumped is the high peak power available in synchronous pumping. High peak powers allow efficient nonlinear conversion while keeping average power -- and associated thermal effects -- to a minimum.
The researchers pumped their CdSe-based OPO with a periodically poled lithium-niobate (PPLN) OPO, which was in turn pumped with a mode-locked Nd:YLF oscillator-amplifier (see figure). The CdSe oscillator was tuned from 9.1 to 9.7 µm by tuning the wavelength of the diffraction-grating-tuned PPLN oscillator, rather than by changing the phase-matching condition via the CdSe crystal. The required 1.85- to 1.97-µm output of the PPLN oscillator was generated in a single period of the PPLN crystal because the oscillator was running so close to degeneracy. The PPLN oscillator produced up to 800 mW of average power in 5-ps pulses.
The mode-locked Nd:YLF oscillator-amplifier at the bottom of the figure pumps the grating-tuned lithium-niobate OPO in the middle, which in turn pumps the 9-µm, CdSe OPO at the top.
The CdSe OPO oscillated in a ring resonator to minimize the round-trip loss of the CdSe crystal and its antireflection-coated faces. Dispersion in the 20-mm-long crystal caused some temporal walk-off among the signal, idler and pump pulses, but even the worst case (the temporal walk-off of the idler from the pump) was less than the 5-ps pump-pulse duration. The CdSe crystal was oriented for noncritical phase matching, so there was no angular walk-off.
The researchers calculated that resonant signal wavelength (2.3 to 2.5 µm) intensities in the CdSe crystal were as high as 2.2 X 1012 W/m2, yet they observed no signs of optical damage to the crystal or its coatings after tens of hours of operation. Moreover, when they chopped the pump beam to a 50 percent duty cycle, they observed a 50 percent decrease in the 9-µm idler output, indicating a complete absence of any deleterious thermal effects in the CdSe crystal. From this, they concluded that much higher outputs are possible, perhaps as high as hundreds of milliwatts of average power.
Tuning of the nonresonant idler output was limited to 9.1 to 9.7 µm by the limited reflectivity of available mirrors and by the water-vapor absorption of the signal wavelength. Working from published Sellmeier equations, however, the researchers calculated that the OPO could be tunable from 8 to 12 µm under ideal conditions.
The scientists calculated that the average 9-µm power inside the CdSe crystal reached 70 mW, although they observed only approximately 10 mW in the output beam. The difference between the two power levels is due almost entirely to transmission losses through the optical elements and to reflection at the crystal's surfaces, which were antireflection- coated at the pump wavelength (1.85 to 1.97 µm) and at the signal wavelength (2.3 to 2.5 µm), but not at the idler wavelength.
The researchers did not measure the quality (M2) of the 9-µm output beam because it was artificially degraded by the concave output mirror. They plan to redesign the OPO with a flat output mirror and then to make M2 measurements. They also did not directly measure the duration of the pulses in the output beam because they did not have suitable autocorrelation components for the 9-µm wavelength. They did, however, measure the spectral width of the output pulses to be 26 nm, which they say is consistent with a bandwidth-limited pulse duration of 5 ps.
MORE FROM PHOTONICS MEDIA