Mid-infrared laser sources are needed for spectroscopy, imaging, and electron diffraction, but generating them efficiently at high power and broad bandwidth remains difficult. Conventional approaches — optical parametric amplification, difference frequency generation — use nonlinear crystals that limit bandwidth through phase-matching constraints and damage at high intensities. Plasma has no damage threshold and no phase-matching bandwidth limit, but coupling laser energy into plasma efficiently requires a mechanism.
Cascading deceleration in plasma channels provides one (arXiv:2603.20916). When an intense laser pulse propagates through a preformed plasma channel, it undergoes successive redshifts — the photons lose energy to the plasma electrons in a cascading process. Each deceleration step shifts the spectrum further into the infrared. The channel confines the beam transversely, maintaining high intensity over long interaction lengths. The result: over 30% energy conversion efficiency from near-infrared input to mid-infrared output spanning 0.58 to 6.86 micrometers.
The output has a distinctive structure. Each spectral sub-band consists of few-cycle pulses — temporally compressed because the deceleration process compresses the pulse as it redshifts. The bandwidth increases with laser intensity when plasma density is sufficient to support the cascading process. Higher intensity means more deceleration steps, which means broader spectral coverage.
The input requirements are modest: energies compatible with high-repetition-rate laser systems, not the single-shot behemoths typical of plasma experiments. This makes the approach practical rather than merely demonstrated.
The structural insight: the plasma doesn't generate mid-infrared light through a nonlinear optical process in the usual sense. It decelerates near-infrared light into the mid-infrared, one step at a time. Each step is a small frequency shift; the cascade of steps produces the large total shift. The plasma is not a converter but a brake — it slows the photons, and slower photons are lower-frequency photons. The bandwidth comes not from a single broad interaction but from a sequence of narrow ones, each starting where the last ended.