Monitoring an ultrafast process means watching the plume or the sound

There is no melt pool to image. What can be measured is the ablation plume's emission and the acoustic signature of each pulse.

By LasersNews Desk·13 Jul 2026·2 min read

Detailed close-up of a golden microprocessor chip, ideal for technology concepts. — Photo by Jeremy Waterhouse on Pexels

Welding process monitoring has a convenient target: a melt pool and a keyhole, both observable and both correlating with weld quality. Ultrafast micromachining offers no equivalent. Material leaves as a plume of atoms, ions and particles, and the process is over in femtoseconds.

Monitoring therefore watches what the process emits rather than what it forms.

Optical emission

The ablation plume emits light with a spectrum characteristic of the material being removed. Monitoring that spectrum enables layer discrimination: when a scribe passes from one material into another, the emission changes, which is directly useful for the selectivity problems in thin film processing.

Emission intensity also correlates with removal rate, which allows detection of a beam that has drifted out of focus or lost power.

Acoustic emission

Each pulse generates a pressure wave. Microphones or contact sensors detect it, and the signal amplitude relates to how much material was removed.

Acoustic monitoring is cheap and robust, and it works where optical access is awkward. Its limitation is that it integrates over an area rather than localising, and it is sensitive to machine noise.

Imaging

High-speed imaging of the plume is used in process development but rarely in production: the data rate is high, interpretation is difficult, and the useful information is usually available more cheaply from emission or acoustics.

What monitoring achieves

Depth control in layered materials, through emission change at layer transitions. This is the most established production use.

Drift detection. Focus drift, power loss and beam degradation all change emission and acoustics before they produce visibly bad parts.

Fault detection. A blocked nozzle, an exhausted assist gas supply or a missing part produce clear signal changes.

What it does not achieve

Direct measurement of the finished geometry. Depth, taper, edge quality and surface roughness are not measured by watching the plume; they require metrology on the part.

Production processes therefore combine monitoring for stability and fault detection with periodic dimensional metrology for verification — which is the same division that applies in additive manufacturing, and for the same reason: the signal correlates with the process, not with the specification.

This article was produced by the LasersNews AI desk and reviewed by our editors.