10 -- Mid-IR difference-frequency generation in MgO:PPLN¶
A 1064 nm pump (1 uJ, 1 ps) and a 1550 nm signal (10 nJ) generate a 3.39 um idler in 5 mm of periodically poled
MgO:LiNbO3 (type 0, all extraordinary, d33). "output": "idler" makes the mid-IR the main beam from here on; the
signal and pump are kept as by-products.
The overview figure of this example's run (python make_figures.py redraws it; the file is hero.png).
ultrafast run dfg.ufs run.json -o results
ultrafast-plot results
- The QPM period is solved for the nominal wavelengths (30.6 um at 24.5 C) unless
"periodM"is given."model": "averaged"(default) uses deff x 2/pi with the grating phase;"domains"flips the sign of d domain by domain (the integrator lands on every domain wall) -- slower, and it captures the higher QPM orders. - The pump is ~70 % depleted. Back-conversion with group-velocity mismatch carves sharp features into all three
pulses and broadens their spectra, so the crystal is given
"spectralMargin": 5(internal window 5x the input bandwidth instead of 2.5x). If a crystal's spectra reach the edge of its internal window, the engine absorbs that part, books it as "numerical" in the energy budget and warns once it exceeds 0.1 % -- the cue to raise the margin. - Instruments:
period_tuning(QPM period acceptance),temperature_tuning(MgO:LN Sellmeier is temperature dependent: ~50 K FWHM here),conversion.
Cross-check against pyNLO¶
compare_pynlo.py runs the same DFG in pyNLO and in the engine (with pyNLO's
plane-wave coupling and domain-resolved poling) and plots the idler spectra:
python python/tools/get_pynlo.py # once: clones the pinned pyNLO into python/third_party/pynlo
PYNLO_SRC=$(pwd)/python/third_party/pynlo/src python docs/examples/Nonlinear/ppln-dfg/compare_pynlo.py
For the default case (0.5 uJ pump, 3 mm) idler, signal and pump energies agree to ~0.1 % and the spectra overlay
(pynlo_comparison.png); compare_pynlo.py 2e-6 5e-3 (signal gain ~50, 35 % pump depletion) agrees to 0.1-0.5 %. The automated version is python/tests/test_pynlo_dfg.py.