摘要： In range resolved differential absorption LIDAR for gas concentration measurement, an attractive solution would be mid-infrared (MIR) micro-pulse LIDAR with photon counting detection. Single-photon detection in MIR is, however, difficult because of the low photon energy. Both superconducting nanowire detectors and low bandgap APDs need cryogenic cooling and are still limited to wavelengths shorter than approximately 2.5 μm. On the other hand, silicon-based single photon avalanche diodes (Si-SPAD) for the visible and near infrared (NIR) wavelength range operate at room temperature with desirable properties such as good timing resolution, low dark count rate and short dead-time. Nonlinear up-conversion through sum frequency generation (SFG) can be used to transfer single-photons from MIR to wavelengths detectable by a Si-SPAD. The appropriate SFG up-conversion is obtained by mixing with visible or NIR photons in a nonlinear crystal, either in a single pass configuration [1], or inside a laser cavity [2]. In the latter case, the high intra-cavity power enables efficient conversion. Of particular interest is to exploit the quasi-phase matching (QPM) scheme. The crystals can then be engineered for colinear interaction at the required wavelengths and bandwidth. A typical 10 mm long QPM crystal has a MIR bandwidth of about 10 nm only, which with proper filtering gives a low background noise. In this work, a periodically poled Rb-doped KTiOPO4 (PPRKTP) crystal was used to convert pulses at 2.4 μm to 737 nm by intra-cavity mixing with 1,064 nm from a Nd:YVO4 laser. A sketch of the cavity and the detection scheme can be seen in Fig. 1 (a). The cavity consists of three mirrors and the Nd:YVO4 laser crystal which also acts as the input coupler. The plane mirrors have high transmission for the MIR and the generated wavelength. The Fresnel reflections at the surfaces of the uncoated PPRKTP crystal were calculated to be 8.6%, which corresponds to a 30% loss per round-trip. The 1.064 nm beam had a 1/e2 radius of 150 μm and a power of about 10 W inside the crystal. The MIR pulses had a pulse energy in the order of 1 nJ, a duration of around 60 fs and a repetition rate of 1 kHz. They were focused down into the PPRKTP crystal and the light leaving the cavity was filtered to remove the other wavelengths and a Si-SPAD was used to detect the 737 nm pulses. Even though the 737 nm pulses were attenuated by at least 50 dB, the Si-SPAD detector was fully saturated, i.e. every single pulse of 2.4 μm radiation triggered the detector. The quantum efficiency of the detector was 27 % and the dead-time was 77 ns. This indicates that substantially weaker pulses could be detected with this method. The pulses were detected with a timing resolution of 140 ps, which corresponds to a spatial resolution of 20 mm for LIDAR. The detector response when the 1.064 nm laser was either on or off can be seen in Fig. 1 (b). The temporal resolution was mainly determined by the trigger jitter in the electronics for the Si-SPAD. Nonetheless, this is, to our knowledge, the best timing resolution for detection of pulses in the MIR region by up-converting the MIR photon inside a cavity.