Drilling a two hundred micrometre hole in a stainless hypotube is not a smaller version of cutting sheet steel. A working guide to micromachining physics, heat affected zone control and how to write the tender so you get a machine that works.
A workshop in Tejgaon quotes a job for a row of two hundred micrometre holes along a stainless tube wall, wins it on price, then finds that the smallest drill it can hold true breaks on the third hole and that the burr inside cannot be reached. The job goes to a laser, and the conversation changes from cutting speed to something else entirely: how much metal around each hole was melted, and whether the customer will accept it.
In sheet metal cutting you tune four things: power, feed rate, assist gas pressure and focus position. Everything scales, and the laser is the expensive part. In micromachining the controlling variables are fluence relative to the material's ablation threshold, pulse duration, repetition rate and focused spot size. The laser stops being the limiting factor surprisingly early. What decides whether you hold the tolerance is the motion system, the beam delivery, the vision alignment and how the part is held.
Every thermal laser process leaves a border around the cut where the material was melted and resolidified, oxidised, or metallurgically altered without melting. In practice that shows up as a recast layer, dross on the exit side, microcracks, a burr, an oxide colour band and, in hardened steels, a softened or embrittled edge. This is why medical device and electronics customers write a maximum heat affected zone into the drawing while a fabrication shop cutting brackets has never heard of it.
The reasons are specific. A recast layer on a stent strut or a needle is a crack initiation site under cyclic loading. On surgical stainless, disturbing the passive chromium oxide film creates a corrosion site that will fail a salt spray test or an implant biocompatibility assessment. On a flexible circuit, charring at the cut edge changes the dielectric behaviour and creates a leakage path. In each case the part passes visual inspection and fails qualification.
Absorption at the working wavelength decides most of the process. Copper, gold, silver and aluminium reflect strongly in the near infrared, which is why fine work on those metals often moves to green or ultraviolet. Polymers absorb ultraviolet photochemically, breaking bonds directly, which is how a UV laser produces a clean polymer edge where a CO2 laser leaves a melted, rolled lip. Glass, sapphire and diamond are either transparent to the usual wavelengths or extremely hard, and need a specialised process rather than more power.
| Material | The difficulty | Usual approach |
|---|---|---|
| Stainless steel, nitinol, titanium | Recast layer and passive film damage on medical parts | Short pulse, low fluence, inert assist gas, electropolish afterwards |
| Copper, gold, aluminium | High reflectivity in the near infrared | Shorter wavelength, or very high peak power to reach threshold |
| Thermoplastics and films | Melting, rolled edges, thermal distortion | Ultraviolet for cold, clean edges; CO2 for speed where edge quality allows |
| Printed and flexible circuits | Charring, delamination, damage to layers below | Ultraviolet, tight fluence control, layer-selective processing |
| Glass, sapphire, ceramics | Brittle fracture and chipping at the edge | Ultrashort pulses, or a filamentation and separation process |
| Diamond and superhard materials | Nothing mechanical will touch it economically | Dedicated hard material laser cutting systems |
| Silicon and compound semiconductors | Chipping, debris and damage to adjacent structures | Ultraviolet or ultrashort, with a controlled kerf and cleaning step |
Marking runs from batch codes on packaging to permanent identification on surgical instruments. The distinction that matters technically is between engraving, which removes material, and annealing, which heats the surface just enough to grow a coloured oxide without breaking it. For stainless steel instruments that will be autoclaved thousands of times, an annealed mark is the safer choice, because an engraved mark creates a crevice that holds moisture and residue. Fibre marking platforms such as the Coherent PowerLine FL series handle metals, and sealed CO2 systems such as the PowerLine C handle organics, packaging films, wood and paper.
There are three strategies, and the choice is a straight trade between quality and cycle time. Percussion drilling fires repeated pulses on one spot, which is fastest and gives the most taper and recast. Trepanning cuts around the circumference and produces a rounder, cleaner hole. Helical drilling spirals the beam down through the material and gives the best wall quality and the least taper, and takes the longest. Aspect ratio, the depth divided by the diameter, is the number that determines which of these is even feasible.
The applications are more common in Bangladesh than people assume: side holes and eyes in cannulae and needles, which is what a system such as the Coherent NA needle drilling platform is built for, plus spinneret and nozzle plates for textiles, filter and mesh apertures, and dosing orifices.
At this scale, kerf width and edge quality replace cutting speed as the headline figures. Tube cutting is its own discipline, with the beam entering a curved wall and the far wall needing protection from the transmitted beam, which is why dedicated tube systems such as StarCut Tube exist for stent and hypotube work. Monocrystalline diamond work on the eM15 platform and superhard cutting on the Q60 address materials where mechanical machining is impossible or economically absurd, while precision systems such as ExactCut 430 cover the general fine cutting case.
Being blunt about this saves money. Laser micromachining is not cheaper than stamping, punching or wire EDM for a simple geometry in high volume, and is often slower per part. What it gives you is access to features and materials that the alternatives cannot reach, and freedom from tooling.
The most common procurement failure here is a tender written around a model number and a power rating. It invites a challenge from a competitor offering something nominally equivalent, and it delivers a machine that satisfies every word of the specification and cannot make your part. Write the requirement around measurable acceptance criteria on a witness component instead.