Laser cut edge quality is the specification nobody writes down and everybody argues about. A part arrives, the edges are rough, the holes are 0.2 mm tight, there is dross on the underside, and the weld shop refuses the batch. None of that is a laser problem or a supplier problem. It is a specification problem. What follows sets out what edge quality actually consists of, what moves it, and how to call it out on a drawing so the argument happens before the parts are cut rather than after. Cutting sits at the front of the nine-step fabrication sequence, and the edge it leaves is inherited by every step after it.
What "edge quality" actually means
"Good edge" is not one property. Several independent characteristics describe a thermally cut edge, and a drawing that says only "clean cut" defines none of them.
- Perpendicularity or angular tolerance — how far the cut face leans away from square. On thicker plate this is usually the characteristic that fails first.
- Surface roughness — the striations, or ripples, on the cut face. Roughness grows with thickness.
- Kerf width — the material the beam removes. This decides whether your hole comes out on size.
- Dross — resolidified molten material stuck to the underside, and spatter stuck to the top.
- Heat-affected zone — the band next to the cut whose metallurgy or surface chemistry has changed.
- Top-edge rounding — a slight melt-round at the entry edge, common at pierce points.
Thermal cutting standards such as ISO 9013 exist precisely to give these characteristics names and tolerance ranges so that we can specify them. We reference the standard by name and expect the customer's drawing to name the acceptance level — we do not invent clause numbers on a quote.
This guide covers how to specify a laser cut edge. It does not cover whether to cut or stamp in the first place; that is an economics question answered in laser cutting versus stamping.
Kerf width: why it decides whether your hole fits
The beam does not cut a line, it removes a slot. That slot is the kerf, and on a part with holes the kerf is the difference between a hole that fits a shaft and a hole that does not.
What moves the kerf
Kerf is not a constant. It moves with:
| What changes | What it does to the kerf |
|---|---|
| Material thickness | Wider kerf and more taper as thickness rises |
| Material type | Stainless and aluminium behave differently from mild steel at the same thickness — see the material selection guide |
| Nozzle diameter and condition | A worn or wrong nozzle widens and destabilises the kerf |
| Focal position | Off-focus cutting widens the kerf and roughens the face |
| Assist gas and pressure | Changes melt ejection and therefore kerf cleanliness |
| Cutting speed | Too fast leaves dross; too slow widens the kerf and grows the heat-affected zone |
Two consequences for design
Holes are sized by the program, not by the drawing. A hole called out at 10 mm is produced by cutting a path that offsets the kerf. If your supplier changes nozzle or material lot and the kerf moves, the hole moves. The fix is to call out hole tolerance as a finished-feature requirement, so it is checked on the part rather than trusted in the program.
Taper means the hole is not cylindrical. The beam is conical, so the kerf is wider at the entry face than at the exit. On thin sheet you can ignore it; on anything above about 6 mm, measure both faces before you decide the hole is wrong. For a clearance hole this rarely matters. For a press-fit, a threaded insert swage hole, or a sealing face, it can matter a great deal — and it is the reason some features should be drilled or reamed after cutting rather than cut to size.
Our tolerance guide covers how to write feature tolerances that survive this.
Dross: three problems with one name
Dross arrives on a non-conformance report as one word. On the sheet it shows up three different ways, and only one of them is cheap to fix.
| Type | Where it is | What it looks like | Usual cause | Usual remedy |
|---|---|---|---|---|
| Bottom dross, loosely attached | Underside, along the cut | Beads that flick off with a fingernail | Speed too high, gas pressure too low, focus too high | Parameter correction; often a fast fix |
| Bottom dross, hard and bonded | Underside, along the cut | Thin, tenacious layer needing grinding | Energy too high, wrong gas, material with high viscosity oxide | Parameter change plus mechanical removal |
| Top spatter | Top face, around the cut and pierce | Fine spherical droplets | Pierce parameters, dirty or oxidised material surface | Pierce tuning, surface condition, protective film |
The reason to distinguish them is cost. Loose dross is a parameter problem and costs nothing to fix on the next sheet. Hard dross is often a material-and-parameter interaction and may need an agreed secondary operation, which is a line item on the quote and a question worth asking before you commit.
Note that dross removal is not the same operation as deburring. Two defects, two causes, and a drawing that says only "deburr" leaves the dross unspecified — see the deburring guide.
Assist gas: the decision that reaches furthest
Assist gas gets chosen for cutting speed. It should be chosen for whatever happens to the part next.
| Gas | Mechanism | Edge appearance | Downstream consequence | Trade-off |
|---|---|---|---|---|
| Oxygen | Exothermic — the reaction adds heat | Dark oxide film on the cut face | Oxide must be removed before welding, painting or powder coating; on stainless it damages corrosion resistance | Fast and cheap to cut; adds a cleaning operation |
| Nitrogen | Inert — melts and blows out | Bright, oxide-free face | Weld-ready and generally coating-ready as cut | Higher gas cost and consumption; slower on thick sections |
| Air | Cheapest; not inert — the oxygen in air still reacts with the melt | Oxidised, variable | Usually needs cleaning; acceptable for non-critical parts | Lowest gas cost; quality varies with humidity and compressor condition |
This is the first question we ask when we quote a cutting job: what happens to the edge next? A batch cut with oxygen on mild steel can be dimensionally perfect and still be rejected by a paint line, because the oxide layer is still there. The cost difference between nitrogen and oxygen cutting is small compared with the cost of a secondary cleaning operation, a rejected coating batch, or a weld that spatter-tests badly. More on the welding side in the welding guide.
For stainless, the consequences are sharper. Heat tint and free iron on a cut edge are a corrosion risk, and the edge has to be treated. That is covered in the stainless passivation and weld cleanup guide.
What thickness does to tolerance
Tolerance is achievable, not universal. It depends on the machine, the power, the material, the fixturing and how much heat is already in the sheet. The bands below are indicative for modern fibre laser cutting of mild steel in good condition. Treat them as the start of a conversation and ask your cutting partner for their capability sheet rather than quoting this table at them.
| Thickness | Indicative dimensional band | Note |
|---|---|---|
| Up to 1 mm | About ±0.1 mm on a stable nest | Distortion usually limits accuracy before the beam does |
| 1–3 mm | ±0.1 to ±0.2 mm | Best combination of tolerance, edge quality and speed |
| 3–6 mm | ±0.2 to ±0.3 mm; taper measurable | Check angular tolerance if the edge is a sealing or bearing face |
| 6–12 mm | ±0.3 to ±0.5 mm; visible striations | Roughness and perpendicularity should be specified explicitly |
| Above 12 mm | Wider still; perpendicularity is the constraint | Confirm per job — shops on 6 kW stop around here, shops with 12 kW and above go considerably further, and plasma or waterjet may quote better |
These are dimensional bands we quote against on our own machines, in mild steel, in good condition. They are not ISO 9013 acceptance levels and they are not a promise about anybody else's laser. Ask for the capability sheet and compare.
Two further realities.
Thin sheet fails by distortion, not by cutting. On 0.8 mm material the beam is far more accurate than the sheet is stable. Heat input, nesting order and part release sequence decide the final geometry — and nesting decisions interact with yield, as set out in the nesting and material yield guide.
Thick plate fails by angular tolerance. The cut face leans. If the edge has to seat, seal or bear, say so on the drawing, because squareness is not implied by a dimensional tolerance.
Heat-affected zone and what it kills
The band next to the cut is not the same material as the sheet. What that costs you depends on the alloy.
| Material | What the heat does | Why it matters |
|---|---|---|
| Mild steel | Thin hardened or oxidised band | Affects coating adhesion and weld behaviour at the edge |
| Stainless steel | Heat tint, chromium-depleted surface | Local loss of corrosion resistance; needs removal |
| Aluminium | Oxide and altered temper locally | Affects appearance after anodising and weld quality — see the anodizing guide |
| Coated or pre-finished sheet | Coating burn-back at the edge | Visible edge damage; may need a design reveal |
For most parts the heat-affected zone is narrow enough to ignore. It stops being ignorable on stainless in a corrosive environment, on parts that will be anodised, and on anything welded immediately after cutting without cleaning. The surface finishing guide covers what has to happen to the edge before coating.
Microjoints, lead-ins and skeleton handling
A part that finishes cutting drops into the skeleton — the scrap grid left behind — and that is where a lot of good edges get damaged. What happens before and after the cut matters as much as the parameters do.
Before the cut
- Microjoints — small uncut bridges that hold the part in place. They leave a nub on removal. Put them on a non-cosmetic edge, on an edge that gets deburred anyway, or on a face that will be machined.
- Pierce points — piercing leaves a small crater and spatter. Pierce in scrap, not on a visible face, and lead in from outside the finished profile.
- Nesting order — cutting adjacent contours in a bad order puts heat into an already-finished edge. Common-line cutting removes the skeleton between two parts but means both inherit the same shared edge, dross and all.
One job we remember: a brushed stainless front panel for a laboratory instrument, rejected for a visible nub on the top edge. The proposed fix was an extra deburr pass. Moving the microjoint to the bottom flange, where the fold hid it, cost nothing and removed the operation entirely.
After the cut
- Part release — parts sliding across a hot skeleton scratch. It is a handling problem, not a cutting problem, and it is worth asking how a supplier manages it on cosmetic parts.
- Stacking — stacked parts rub. Interleaving is cheap; re-polishing is not.
Eight DFM rules
- Call out hole size as a finished feature, not as a program dimension.
- Do not cut press-fit or insert-swage holes to final size in thick plate — taper will bite you. Ream or drill.
- Specify angular tolerance on any edge that seals, seats or bears.
- Choose the assist gas by what happens next, not by cutting speed.
- Say whether the edge will be welded, painted or left as cut. It changes the process.
- Place pierce points and microjoints off cosmetic faces.
- Ask what happens to the part after the cut — release, handling and stacking.
- Request one first-article part and measure the edge features you actually care about, rather than approving on dimensional layout alone. The first article inspection guide sets out how to structure that, and the QC and inspection guide covers where to check it in the process.
Paste-ready drawing callout
CUT EDGE REQUIREMENT:
Process ..................... fibre laser
Assist gas .................. nitrogen / oxygen / air
Edge condition .............. as cut / deburred / cleaned
Dross ....................... none on sealing faces; removable elsewhere
Angular tolerance ........... ___ (only where the edge seats or seals)
Roughness ................... ___ (only where specified)
Holes ....................... finished-feature tolerance +/- ___ mm
Post-cut .................... degrease / passivate / none
NOTES:
Pierce and microjoint locations: off cosmetic faces.
No secondary grinding implied unless stated above.Frequently asked questions
What is a good kerf width for laser cutting?
There is no good value, only a value that suits the material and thickness. Kerf widens with thickness and moves with nozzle condition, focal position, assist gas and speed. What matters for design is that we compensate the kerf in the program, so specify holes as finished features with a tolerance rather than as program dimensions.
Why does my laser cut part have dross on the underside?
Usually cutting speed, assist gas pressure or focal position is off. Loosely attached beads are normally a parameter fix and cost nothing to correct on the next sheet. Hard, tightly bonded dross is often a material-and-parameter interaction and may need an agreed secondary removal operation.
Does laser cutting leave a heat-affected zone?
Yes, though on a fibre laser it is narrow. It matters on stainless, where heat tint reduces local corrosion resistance; on parts that will be anodised; and on anything welded without cleaning. On mild steel that is simply painted it is rarely significant.
Should I specify oxygen or nitrogen cutting?
Choose by what happens to the part afterwards. Oxygen cutting is faster and cheaper but leaves an oxide film that must be removed before welding, painting or powder coating. Nitrogen gives a bright, generally coating-ready edge at higher gas cost.
Why are my laser cut holes slightly tight or tapered?
We produce holes by offsetting the kerf, so a change in kerf moves the hole. Taper comes from the conical beam and grows with thickness. For clearance holes this rarely matters; for press fits, insert swage holes or sealing faces, specify a finished-feature tolerance or plan a secondary operation.
Can laser cut edges be welded directly?
It depends on the assist gas and the material. Nitrogen-cut edges on mild steel are generally weld-ready. Oxygen-cut edges carry an oxide film that should be removed first, and on stainless the heat tint should be cleaned off before welding to protect corrosion resistance.
The short version
Three things decide whether a cut edge argument happens: whether the hole is called out as a finished feature, whether the assist gas was chosen for what happens next, and whether angular tolerance is specified on the edges that seat. Put all three on the drawing and the argument stops happening.
Where we fit
We cut, form, weld and finish sheet metal parts, and we would rather argue about edge quality at the quoting stage than after the first batch. Send us the drawing with the edge requirements marked — including the ones that feel too obvious to bother writing down — and we will tell you what the process can hold, what needs a secondary operation, and what should be changed on the print. Free CAD and DFM review, and we will cut you a first article to measure before we run the batch.