Compact Solid-State Source Is Tunable from Green to Red
Continuously tunable femtosecond pulses are useful in many applications.
Tunable femtosecond pulses in the visible spectral region are valuable in numerous applications, including spectroscopy, frequency metrology and confocal microscopy, but the lasers that produce these pulses tend to be bulky, temperamental and expensive. Recently, however, scientists at Universität Konstanz in Germany demonstrated a compact and relatively straightforward solid-state system that produces milliwatts of average power in femtosecond pulses that are continuously tunable from 520 to 700 nm.
The first major component of the system was an amplified, mode-locked erbium-doped fiber laser that generated 1.55-μm, 65-fs pulses with a repetition frequency of 107.7 MHz and average power of 300 mW. The final component was a periodically poled MgO:LiNbO3 crystal that converted the incoming infrared radiation to visible light (Figure 1). But the investigators took advantage of several novel techniques to enable tunability of their source and to maximize its efficiency.
Figure 1. The tunable, visible femtosecond source comprised a mode-locked, amplified Er:fiber laser system, a highly nonlinear fiber and a periodically poled lithium niobate crystal. Images ©OSA.
To produce a wavelength-tunable infrared signal for subsequent frequency doubling in the LiNbO3 crystal, the scientists injected the 1.55-μm pulses from the laser into a polarization-maintaining, highly nonlinear, dispersion-shifted fiber whose zero-dispersion wavelength was 1.52 μm (Figure 1). The near-overlap of the laser and zero-dispersion wavelengths maximized phase matching within the fiber, and the high peak intensities in the fiber led to a parametric generation of two mutually coherent frequency components. They could vary the spacing between the two components up to as much as 100 THz by adjusting the prechirp on the injected pulse with a pair of silicon prisms. They selected the shorter-wavelength component for subsequent frequency doubling in the LiNbO3 crystal.
The pulses in this shorter-wavelength component had a duration as short as 13 fs after passing through a prism-based pulse compressor. Their average power was ~30 mW, and the researchers could tune them from 1.05 to 1.40 μm by adjusting the prechirp of the pulse injected into the nonlinear fiber.
Figure 2. The crystal’s poling periodicity, and, hence, its phase-matching wavelength, could be altered by translating it sideways; i.e., in the Y-direction.
But merely illuminating a nonlinear crystal with tunable infrared radiation doesn’t result in tunable visible radiation because the crystal’s poling period allows phase matching for only one wavelength. The scientists finessed this problem by using a “fan-out” poling geometry (Figure 2). The poling period of the crystal, and, hence, its phase-matched wavelength, was adjustable by translating the crystal sideways. Thus, if they translated the crystal and tuned the infrared input simultaneously, they could produce continuously tunable second-harmonic radiation in the visible.
But not efficiently. They calculated that, although the second-harmonic acceptance bandwidth was only several nanometers, the bandwidth of the incident infrared radiation was more than 100 nm. In other words, only radiation in a very narrow stripe at the very center of the fundamental’s spectrum would be frequency-doubled.
Figure 3. The bandwidth for second-harmonic generation (f + f = 2f) was very small, so only a small fraction of the infrared light incident on the nonlinear crystal could be doubled into the visible. However, by phase matching both second-harmonic and sum-frequency generation [(f - f0) + (f + f0) = 2f], the scientists converted the entire input spectrum to a narrowband visible output.
They overcame this problem by realizing that sum-frequency generation could be phase-matched along with second-harmonic generation and at nearly the same efficiency if the separation of the two sum-frequency components were within the 100-nm bandwidth of the infrared radiation (Figure 3). Hence, they were able to convert the entire infrared spectrum to a narrowband visible output.
Figure 4. The spectrum of infrared radiation (orange) was uniformly diminished when the nonlinear crystal was inserted (brown). Photons from both wings of the infrared spectrum were combined by sum-frequency generation to produce narrowband visible output (green).
The spectrum of the incoming infrared pulses showed a uniform amplitude reduction when sum-frequency and second-harmonic generation were phase-matched (Figure 4). There was no depression at the center of the spectrum, strongly indicating that the visible light was produced by both processes and not by second-harmonic generation alone.
Optics Letters, April 15, 2006, pp. 1148-1150.
MORE FROM PHOTONICS MEDIA