40-Year-Old 'Green Problem' Solved: Noisy Modes, Quiet Beam

A green laser whose 200 longitudinal modes each flicker at essentially 100% modulation depth delivers a total beam steady to 0.05% rms (DC to 10 MHz). Thomas Baer of Stanford has finally explained why, in a paper published in Physical Review A — 40 years after his own 1986 theory defined the puzzle as the "green problem" (T. M. Baer, Phys. Rev. A 114, 033503, 2026).

The lasers in question are intracavity-doubled neodymium DPSSLs (diode-pumped solid-state lasers) that pump Ti:sapphire oscillators and frequency combs. Anyone with a Fabry-Pérot etalon can repeat the key measurement: filter the beam to a few modes and each transmitted mode blinks on and off in bursts tens of microseconds long, modulated essentially 100%. Remove the filter, and their sum stays flat.

Why the 1986 model couldn't see it

At Spectra-Physics in the early 1980s, Baer's team built miniature intracavity-doubled lasers to replace argon-ion lasers — and found them wildly noisy, with intensity fluctuations of tens of percent, sometimes chaotic. His 1986 rate-equation model showed the doubling crystal couples longitudinal modes through sum-frequency generation; with two or three modes, they fight, switching on and off irregularly. That became the accepted description.

Then came a cure that made no sense. Lengthen the cavity so hundreds of modes oscillate, and total noise nearly vanishes while individual modes get noisier. The 1986 equations — and the many-mode versions built on them — track only intensity and discard optical phase. For two or three modes that simplification is safe. For 200, it's fatal: whether the summed intensity is constant or beating on a nanosecond scale depends entirely on phase arrangement, and a phase-blind model can't ask the question.

What the crystal does to the phases

The new model follows amplitude and phase of several hundred modes through one round trip at a time — through the gain medium with spatial hole burning, and twice through the nonlinear crystal. The hardware: a 42-cm cavity, a neodymium vanadate gain element pumped by a diode laser, and a lithium triborate (LBO) crystal near an end mirror that double-passes the green on its way out. Baer's workstation is roughly 100,000× faster than his 1986 HP 9816 (8-MHz processor, 128 KB RAM), which makes the problem numerically tractable.

The mechanism: the crystal isn't a one-way converter. Green generated from one IR mode pair converts back into other IR modes, carrying the phase of the pair that produced it. Through that lever the crystal adjusts every mode's phase on every pass — not randomly. A single quantity (mean of intensity squared divided by square of the mean; one for constant intensity, larger for bunched light) cannot increase on a forward pass through the crystal — a theorem the paper proves. The phases move toward the constant-intensity arrangement because it loses the least IR to green conversion, and the crystal finds it within a fraction of a microsecond, far faster than the gain medium responds.

Meanwhile the division of power among modes is only weakly constrained. Modes trade power back and forth on the gain-recovery timescale, each switching fully on and off — but the crystal's coupling ties gains and losses across the whole set, making the exchanges nearly zero-sum, while the fast phase rearrangement keeps the summed waveform smooth. Loud modes, quiet beam.

What changes at the bench

Three practical consequences follow:

• A quiet total doesn't mean quiet modes. Any application that filters the beam spectrally or relies on a subset of modes will see noise the full beam hides.
• Mode count is the design parameter. The empirical 1990s cure — a longer cavity — moves the laser from the few-mode regime where coupling drives competition into the many-mode regime where it drives cancellation. Bill Nighan and John Cole's long-cavity Millennia design, sold in tens of thousands of units, was right for a reason nobody had recognized.
• The crystal is a stabilizer, not just a doubler. Models treating it as nonlinear loss to be minimized miss the phase organization that actually determines whether the laser is quiet.

The result also reframes the device itself: unlike modelocked lasers, 1960s FM lasers, or quantum cascade combs, nothing in the cavity is driven. The crystal passively organizes phases toward near-constant intensity while amplitudes wander — a combination, Baer notes, that has been in commercial production for 30 years without being recognized as a phase-organized laser.

Because the mechanism depends only on many oscillators sharing a quadratic nonlinearity — nothing specific to neodymium or LBO — it may extend to coupled-oscillator systems in the Haken/Kuramoto synchronization tradition, from Josephson arrays to power grids. For instrument builders, the open question is whether specs quoting total-intensity rms noise adequately serve applications that spectrally slice these pumps.