Micromachining With Femtosecond Laser: How It Works and Where It Fits
A femtosecond pulse removes material before heat can spread into the part. That single fact decides which features belong on a laser and which belong on a mill. This page covers the mechanism, the practical limits, and the shop-floor trade-offs for engineers specifying micro features.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
Why micromachining with femtosecond laser runs cold
A femtosecond is 10⁻¹⁵ seconds. In that window a focused pulse reaches the workpiece, ionizes the surface, and turns a thin layer into plasma before the surrounding lattice has time to conduct heat away. The energy never spreads. Material leaves as vapor and small clusters, not as melt that resolidifies along the cut edge.
Compare that with a nanosecond pulse, which is about a thousand times longer. Heat has time to diffuse roughly a micrometer into the bulk. You get a recast layer, a heat-affected zone, and microcracks in brittle materials. On a 50 μm wide slot that damage zone can be a meaningful fraction of the feature itself.
The practical result is a cut edge that needs no deburring and no post-machining etch. For stents, fuel injector nozzles, and thin ceramic substrates, that edge quality is the reason the process exists. The trade-off is throughput: each pulse removes only nanometers of depth, so the tool path has to be fast and repeatable.
Peak power matters more than average power here. A pulse of a few hundred femtoseconds at modest energy per pulse still reaches intensities high enough to ionize any solid, which is why the process works on metals, ceramics, polymers, and diamond with the same optical setup.
Which materials and features actually suit the process
The cold-ablation mechanism is material-agnostic in principle, but the economics are not. Hard, brittle, or heat-sensitive materials gain the most: silicon carbide, alumina, sapphire, PEEK, polyimide, nitinol, and tungsten carbide. These are the parts where a nanosecond laser leaves cracks or where a 0.5 mm end mill simply cannot reach.
Metals are a different story. On 6061 aluminum or 304 stainless, a 0.3 mm end mill held in a 5-axis machine reaches ±0.005 mm and Ra 0.8–1.6 μm without any special setup. If your smallest feature is 0.4 mm wide and 1 mm deep, laser micromachining adds cost and cycle time for no gain in function.
Geometry drives the decision more than material. Through-holes under 100 μm, tapered nozzles, blind slots with 5:1 aspect ratio, and cuts in 50 μm thick foil are laser territory. Pockets with square internal corners, threaded holes, and surfaces that need a specific Ra callout belong on a mill.
One more boundary: depth. Femtosecond ablation is a surface process. Cutting through 3 mm of steel by laser takes many passes and drifts in taper. For anything thicker than roughly 1 mm, a mill or wire EDM holds the wall straightness far more predictably.
What tolerance and surface finish you can hold
Positional accuracy on a good femtosecond setup lands around ±2–5 μm on thin parts, with feature widths down to 10–20 μm depending on optics and wavelength. That is finer than most CNC work, but it is a local number. It applies to the laser-cut feature, not to the whole part envelope.
Surface finish inside a laser-cut wall is typically Ra 0.5–1.5 μm, with some striation from the pulse overlap. It is clean and oxide-free but not mirror. If the drawing calls for Ra 0.2–0.8 μm on a sealing face, that face still gets a finishing pass on a machine, whether laser or CNC cut it first.
Taper is the number engineers underestimate. A femtosecond cut through 200 μm of stainless may show 2–5° of wall taper because the beam focus narrows with depth. On a 50 μm hole that is a 20 percent diameter difference top to bottom. If the hole has to be cylindrical, plan a trepanning path or accept the taper in the design.
Thermal drift is small but not zero. Over a long run, stage and optics heat up. On features under 20 μm, we re-datum the part every few hundred cycles rather than trusting a single fixture offset for the full batch.
How the laser and the mill divide the work
Most production parts we see are hybrids. The blank is milled to envelope, datums are established, and the micro features are laser cut in a second operation. This keeps the laser doing what only it can do and keeps the mill doing what it is fast at.
The order matters. Laser first, then mill, means the mill sees a part with a heat-clean edge and can use the laser cut as a reference. Mill first, then laser, means the micro features land on an already-stable part with known datums. For thin foil, laser first is almost always right because clamping a 50 μm sheet for milling is difficult.
Inspection is the other half of the plan. Micro features cannot be checked with calipers. We use optical CMM and vision measurement, and we inspect 100 percent of parts before shipment on micro work. Raw material certificates and final reports go out on request.
Cleaning is not optional. Ablation debris redeposits as nanoparticles on and around the cut. An ultrasonic bath in DI water plus a plasma or CO₂ clean removes it. Skip that step and the part may pass dimensional inspection while failing a particle or adhesion spec downstream.
Cost, throughput, and when not to use a laser
A femtosecond source costs far more per watt than a fiber or nanosecond laser, and the material removal rate is low. On a 100 μm hole in 100 μm stainless foil, the cycle may be a few seconds. On a 500 μm deep slot, it can be minutes per part. That cycle time is what you are buying.
So the break-even is feature size, not part size. If the smallest feature is above roughly 0.3 mm, CNC milling wins on cost and often on lead time. If the feature is below 0.1 mm, in a brittle material, or surrounded by heat-sensitive geometry, laser wins because the alternative is a damaged part or no part at all.
Volume changes the answer too. For a single prototype with one micro hole, a laser job shop may be faster than building a fixture. For a 10,000-part run of a simple micro slot, a stamped or molded process will beat both, provided the tolerance allows it.
We quote no minimum order quantity, from one prototype to 10,000+ part runs, and we will tell you when the laser route is not worth the money. Quotation and free DFM analysis come back within 12 hours, and production can start within 24 hours once the design is released.
Femtosecond laser vs CNC milling for micro features
Use the smallest feature and the material as the first filter.
| Criterion | Femtosecond laser | CNC milling |
|---|---|---|
| Smallest feature | 10–100 μm | 0.3 mm and up |
| Best material fit | Ceramics, PEEK, nitinol, foil | Aluminum, stainless, steel, brass |
| Heat-affected zone | Negligible | None, mechanical cut |
| Typical tolerance | ±2–5 μm local | ±0.005 mm |
| Wall taper | 2–5° on thick cuts | Straight walls |
| Surface finish | Ra 0.5–1.5 μm | Ra 0.2–3.2 μm by pass |
| Throughput | Low, seconds to minutes per feature | High on accessible geometry |
| Best use | Micro holes, slots, thin foil | Envelopes, datums, threads, pockets |
The call we would make
If the smallest feature is under 0.1 mm, sits in a brittle or heat-sensitive material, or is cut in foil under 200 μm, use the laser. If it is 0.3 mm or larger in metal, mill it. For hybrid parts, rough and datum the blank on a 5-axis machine, then laser the micro features in a second op.
Questions engineers ask before releasing a design
Can a femtosecond laser cut through 2 mm of stainless steel?
It can, but it is slow and the wall taper grows with depth. Above roughly 1 mm, wire EDM or milling holds wall straightness and cost far better.
We usually recommend laser only for the through-features that a mill cannot reach, and machine the rest of the profile conventionally.
Do micro laser cuts need deburring or etching afterward?
No recast or burr forms at the edge, so no mechanical deburring is required on most materials. Ablation debris does redeposit on the surface.
An ultrasonic clean in DI water removes it. On parts with adhesion or particle specs, we add a plasma clean after the bath.
What is the smallest hole you can hold to a diameter tolerance?
Feature widths down to 10–20 μm are possible on thin material with the right optics, and positional accuracy lands around ±2–5 μm.
Below 20 μm, taper and debris become the limiting factors more than the beam itself. Send the drawing and we will confirm what is repeatable.
Can the laser process be used on assembled or coated parts?
Not on assemblies. The beam needs a clear optical path to each feature, and clamping a multi-part stack shifts the focus.
Coated parts are possible if the coating is thin and the ablation threshold differs enough from the substrate. We test a coupon first.
How do you inspect 10–20 μm features?
Calipers and pin gauges are not usable at that scale. We measure with optical CMM and vision systems, and we inspect 100 percent before shipment on micro work.
Inspection reports are available on request, along with raw material certificates.
Is the process suitable for medical or automotive production?
Yes. Nitinol stent features and injector nozzle holes are the classic applications, and both need a clean, crack-free edge.
Our quality system is certified to ISO 9001:2015, IATF 16949:2016, and ISO 13485:2016, which covers both automotive and medical device work.
Send the drawing and we will tell you which process fits
Upload your part and we return a quotation with free DFM analysis within 12 hours. If the laser route is not worth the cost, we say so.
12-hour quote100% inspectionNo minimum order quantityNDA on request