Researchers at the French aerospace research agency Onera (Toulouse, France) have made a significant breakthrough in the field of mid-IR (8 to 12 µm) optical parametric sources by developing a new phase-matching scenario in nonlinear optical materials.1, 2 Dubbed Fresnel phase matching, the technique relies on the phase change introduced by total internal reflection, and allows the use of common nonbirefringent materials (gallium arsenide, zinc selenide, and gallium phosphide, for example) for the production of widely tunable mid-IR coherent light (see figure).
The research was spurred by the need for a versatile IR light source to be used for counterterrorism and other defense necessities (such as detection of explosives, toxic gases, and IR countermeasures for aircraft), as well as detecting pollution. "Many aggressive chemical agents display specific absorption features in the mid-infrared (particularly the fingerprint region between 8 and 12 µm), which allows a convenient way to detect them with tunable sources," says Emmanuel Rosencher, one of the researchers. "It explains why so much effort has been dedicated to the fabrication of such sources for so many years."
While quantum-cascade lasers can serve as such a source, their restricted tunable ranges, commonly covering a few hundred nanometers, can be limiting. Moreover, only a small quantity of energy is delivered in each pulse (typically 10 nJ at room temperature), making them poor candidates for long-distance remote sensing.
On the other hand, sources based on optical parametric interaction (three-wave mixing) are widely tunable and can deliver high-power pulses. In difference-frequency generation (DFG), two near-IR laser beams are mixed in a nonlinear crystal, yielding a coherent wave at the difference frequency. Here, the tunability of the final source is linked to the tunability of the near-IR ones. In optical parametric generation, a laser beam spontaneously splits in a nonlinear crystal into two beams of other frequencies (which add up to the original frequency). Here, the tunability is limited to the transparency range of the materials, which can be very large.
Fresnel phase shift
In both cases, making the three waves interact efficiently in the nonlinear crystal calls for phase matching. Dispersion causes the three waves to propagate in the crystal with different phase velocities; as a result, the power transfer between the three waves grows quadratically with the distance in the crystal, up to the so-called coherence length. Beyond this length, power will flow back in the opposite direction; consequently, power periodically flows back from the mid-IR beams to the near-IR pumping beams, leading to a low conversion efficiency.
In quasi-phase-matching, the crystallographic orientation of the nonlinear crystal is inverted every coherence length. This idea has been put to work in ferroelectric materials that are poled by a high electric field (for example, periodically poled lithium niobate). However, there are no ferroelectric materials transparent beyond 4.5 µm. In contrast, the phase-matching technique developed by the Onera researchers efficiently produces wavelengths beyond 7 µm.
One of the new devices is based on a common gallium arsenide wafer. The sides of the wafers are polished with a slanted face so that the three waves can be introduced from the side and trapped by total internal reflection in the wafer, zigzagging through the structure. The thickness of the wafer is chosen such that, given the reflection angle, the distance traveled between two reflections is close but not exactly equal to the coherence length. When totally reflected at the wafer surfaces, the three waves are submitted to a Fresnel phase shift that can be easily tuned by adjusting the angle, the polarization, and frequency of the waves. When properly tuned, the DFG power now constructively builds up at each total internal reflection. Because the thickness of the wafer is not a precise parameter, the device is simple to construct and tune.
In one experiment, two waves (p and s polarized, respectively) tunable between 1.9 and 2.3 µm were produced by a conventional optical parametric oscillator. The two waves were introduced into a 400-µm-thick gallium arsenide wafer with an end-slant angle of 28°. The wafer was "epiready" (meaning it had a surface almost atomically smooth) so that the reflectivity at each bounce was very close to one (typically 99.5%). The DFG signal was tuned between 8 and 15 µm through a simultaneous tuning of the OPO waves and the angle of incidence (see figure). The light underwent up to 66 bounces. The output energy was in the few hundreds of nJ range for input power in the mJ range. More recently, the researchers have obtained up to 10 µJ of mid-IR light with this method.
- R. Haidar, et al., Appl. Phys. Lett. 82, 1167 (2003).
- R. Haidar, et al., Appl. Phys. Lett. 83, 1506 (2003).