Pump–probe without a delay stage: how asynchronous optical sampling works

In a classic pump–probe experiment, one pulse excites the sample and a second, delayed pulse measures what happened. The delay is created geometrically: the probe beam travels a slightly longer path, and a mirror on a translation stage changes that path length. One millimetre of extra travel is about 3.3 picoseconds of delay.

What the delay stage costs

The arithmetic is unforgiving. Reaching a nanosecond of delay requires roughly 150 mm of stage travel in a double-pass geometry — a large, expensive stage. Reaching ten nanoseconds is impractical for most laboratories.

Speed is the second cost. The stage has to physically move to each delay point, settle, and be measured. A full scan takes seconds to minutes, which limits how many scans you can average, which sets your signal-to-noise floor.

Then there is alignment. As the stage travels, beam pointing changes slightly. Over a long scan that produces a systematic artefact that looks like real dynamics and is very hard to distinguish from it.

The ASOPS idea

Asynchronous optical sampling gets rid of the stage entirely by using two pulse trains with slightly different repetition rates.

Suppose the pump laser runs at repetition rate f and the probe at f + Δf, where Δf is small compared to f. Pulse pair number one arrives coincident. By the time the next pulses arrive, the probe has drifted by a small amount relative to the pump — because its period is slightly shorter. The pair after that drifts a little further. The delay between pump and probe therefore sweeps automatically, one small step per laser period, until it has walked through the entire interval between pulses and the pattern repeats.

Three consequences follow directly.

The scan window is 1/f. The delay sweeps across the full inter-pulse interval. At 60 MHz that is roughly 17 nanoseconds of optical delay per scan — a range no practical mechanical stage reaches.

The scan rate is Δf. The full delay window is traversed Δf times per second. At a kilohertz offset that is a thousand complete scans a second, against seconds per scan for a stage. Averaging that was previously unaffordable becomes routine.

Time is magnified by f/Δf. Femtosecond-scale dynamics arrive at the detector stretched onto a microsecond timebase, so the detection electronics never need femtosecond bandwidth. This is what makes the whole approach practical.

The problem ASOPS creates, and how a shared cavity solves it

The delay axis is only as accurate as the stability of Δf. If the two lasers drift relative to each other, the time axis stretches and compresses unpredictably, and averaging over many scans smears the result instead of improving it.

The conventional answer is to lock the two lasers together with fast feedback electronics — which reintroduces the complexity, cost and failure modes that removing the stage was supposed to eliminate.

The alternative is to generate both pulse trains inside a single shared cavity. Because both combs experience the same mechanical vibrations, the same thermal drift and the same environmental noise, their fluctuations are highly correlated and largely cancel in the difference. The relative timing is passively stable, without an active lock. That is the architecture K2 Photonics builds around.

Where ASOPS fits, and where it does not

It fits when the delay range you need is long — carrier recombination on nanosecond timescales, acoustic phonon echoes, THz time-domain traces, thermal transport — or when you need many averages quickly, or when the setup has to be compact and stay aligned without a specialist.

It fits less well when your dynamics are complete within a few picoseconds and you need the very finest time resolution available. There, the time step is set by Δf/f relative to the pulse duration, and a short-range mechanical scan may still be the better instrument.

The honest summary: ASOPS trades away nothing except moving parts, but the advantage is largest exactly where the mechanical approach hurts most — long delays and heavy averaging.

Paeonia represents K2 Photonics across Asia. See the K2-ASOPS system.