Pulse duration: what the material does with the energy before it can spread

At 1064 nm, copper absorbs the beam within about 13 nm of its surface; only about 4 % of the light is absorbed at all, the rest reflects. The question is what happens to that 4 % in the time the pulse lasts. Copper’s thermal diffusivity is 117 mm²/s, so in one nanosecond heat travels roughly 700 nm — fifty times deeper than the light went. A nanosecond pulse therefore melts a volume far larger than the skin it heated, and material leaves as melt pushed out by vapour pressure. The melt that does not leave freezes as recast and burr, and the heat-affected zone scales with the square root of the pulse duration and the diffusivity.

At 200 fs the same copper moves heat only about 10 nm, less than the optical depth. And 200 fs is shorter than the time electrons take to hand their energy to the lattice, a few picoseconds in most metals. The energy is deposited before conduction can begin, and above a threshold fluence the heated skin leaves as vapour and plasma. In drilling stainless steel, a recast layer of about 25 µm with nanosecond pulses became unmeasurable with picosecond pulses in one published comparison; in electrode-edge work, clearing a 20 µm margin took a 1 ns green source where 250 ns NIR pulses could not hold it.

Two cases. First, when the feature or the layer you are working in is thinner than the heat would travel in a longer pulse. If the layer you must not touch is 12 µm below the surface and softens at 130 °C, the micrometre of heat travel a 100 ns pulse produces in stainless is one number; what a train of such pulses leaves behind is another. Do the arithmetic before you pick a source. Second, when the material is transparent at every wavelength you have: fused silica, sapphire, borosilicate, the wide-bandgap wafers. The absorption explorer shows these as gaps: linear absorption is too weak to place. Ultrashort pulses reach the intensities at which the material absorbs nonlinearly, two or more photons at once, and that is the only way a transparent material couples at all.

Short pulses cost throughput. Specific removal rate — cubic millimetres per minute per watt — is lower for ultrashort pulses than for nanosecond pulses on most metals, and the efficient operating point is narrow: ablation depth per pulse grows only logarithmically with fluence, and removal per unit energy peaks at a fluence of about e² (7.4) times the ablation threshold, falling either side. Picosecond sources typically sit between femtosecond and nanosecond on both cost and rate. So the answer is a trial on your material that reports edge quality and rate together, not a regime picked from a chart.

Reverse the copper case. Under a CO₂ beam, PET absorbs over about 50 µm at 10.6 µm and 18 µm at 9.4 µm; moisture in board absorbs over about 12 µm. Their thermal diffusivity is around 0.1 mm²/s — a thousandth of copper’s — so even in a full millisecond of dwell heat has travelled only about 20 µm — the same order as the depth the light penetrated, not the fifty-fold excess of the copper case. The absorbing depth simply heats through. The pulse shape barely matters. What matters is the energy per unit length of cut (power divided by speed) and the wavelength, which sets the depth. That is why CO₂ sources for film and board are continuous-wave or modulated in microseconds, not pulsed in nanoseconds, and why the wavelength explorer, not this guide, decides most packaging processes.

Edge quality on these materials is then thermal management rather than pulse physics. In one published polypropylene study, the reported heat-affected zone across both sides of the groove, at 100 mm/s under a 30 W continuous beam, was about 300 µm, shrinking with speed and with a heat-sinking base under the sheet. Speed, power and support are the controls, and every LasX packaging machine’s recipe sets all three.

For most metals the electron–phonon coupling time is one to five picoseconds. A 10 ps pulse is at the boundary; a 200 fs or 800 fs pulse is well inside it. Both count as heat-confined on the map, and both avoid the melt-dominated removal of nanosecond pulses. The differences show up one level down: measured on steel, the depth heated per pulse was about 12–15 nm at 0.2 ps against about 6 nm at 10 ps, with the ablation threshold and incubation behaviour shifting accordingly. Removal efficiency, burst-mode behaviour and how much heat is left behind per pulse all differ by a factor of two or so rather than an order of magnitude.

Those factors of two decide real parts: a 25 µm feature in a laminate, or a coated electrode where the coating and the foil ablate at different thresholds. A chart cannot show them. The PhotonX lab runs both regimes, and the trial report shows which one held the specification at the required rate, with micrographs.

Polyimide absorbs 355 nm within about 0.3 µm of its surface. Twenty nanoseconds of conduction in polyimide is about 0.08 µm. So a nanosecond UV pulse produces the geometry of a heat-confined process without an ultrashort source: the energy stays within a third of a micrometre of where it landed. Part of the removal is also photochemical — the photon energy at 355 nm is enough to break bonds in the polymer directly — so some material leaves before it has heated at all. Medical laminates and flexible circuits behave similarly where their absorption edge sits in the UV.

CO₂ cuts the same polyimide by volume heating over about 15 µm of depth, and for a slit or a through-cut in a packaging film that is the right tool. The choice is about the feature: a 25 µm channel or a via next to a copper trace wants the smaller thermal footprint, and the explorer shows why — the penetration depth at 355 nm is fifty times shorter than at 10.6 µm for the same film.

Everything above is single-pulse physics. Production is a train of pulses, and heat confinement describes each one, not the sum. In stainless steel roughly 0.4 of the energy of an ultrashort pulse has been measured to remain in the part as heat; at megahertz repetition rates the next pulse arrives before that heat has diffused away from the spot, and the baseline temperature climbs pulse by pulse. Above a threshold surface temperature — around 600 °C in the same stainless work — the surface bumps and melts under pulses that would each, alone, have left it clean. The published criterion relates the maximum average power a process will tolerate to the repetition rate, the spot and the material’s diffusivity.

The practical consequence is that pulse placement matters as much as pulse duration: scanner speed, spot overlap, burst timing and the spacing between passes set how much heat is waiting when the next pulse lands. This is what the timing inside every LaserSharp® module is for. Proton λ controls individual pulses up to 10 MHz with 10 ns laser pulse timing resolution, so pulse spacing is set on the scale of the material’s cooling time rather than the scanner’s convenience, and vision registration keeps the placement where the recipe put it.

A kiss-cut through a label face and adhesive that stops on the liner, or an antenna ablated from a foil without scorching the paper under it, is the same comparison applied at a boundary. Three lengths: the top layer’s optical penetration depth at the chosen wavelength, the heat travel during the pulse, and the layer thickness. The wavelength does most of the work — pick the line the top layer absorbs strongly and the layer beneath weakly, and the energy stops near the boundary on its own. Pulse duration then sets how far the heat leaks past it.

The adhesives page has the sharpest example. A silicone adhesive absorbs about thirty-five times more strongly near 9.4 µm than at 10.6 µm (37× at the 9.36 µm data point), because the short CO₂ line sits on its Si–O band; an acrylic adhesive over a silicone-coated PET liner presents a different pair of depths at each line. Reading those two values off the explorer, and the diffusion lengths off the map, tells you which CO₂ line to run before any sample is cut — and the sample then confirms it.

Pick a wavelength; set a pulse duration with the slider or the marked values; add up to six materials. Each material draws as a solid line — how far heat travels during a pulse of that duration — and a dashed horizontal — how deep the light is absorbed at that wavelength. Where the solid line sits below the dashed level, the material heats through its absorbing depth (volume heating). Where it sits far above, conduction dominates. Left of the shaded strip, the pulse ends before the lattice has heated. A material with no dashed line is transparent at that wavelength for practical purposes; ultrashort pulses would need to couple nonlinearly. Enter a feature or layer thickness to see which durations keep the heat travel inside it.

The map classifies; it does not predict a kerf width, a heat-affected zone, a speed or a fluence for a construction. Those come out of a trial on the actual material, and they are what the PhotonX process report contains.