A part comes off the plate looking excellent. It goes into service, takes a modest load, and separates cleanly along a single horizontal line — not shattered, not deformed, just come apart where two layers were supposed to be one.
Read the fracture face first
Break the failed piece further and look at the broken surface in good light.
A flat, matte plane where you can still see the individual bead outlines means the layers never welded. The beads are intact; the joint between them failed. That is this page.
A rough, irregular break that tears across beads rather than between them means the plastic itself failed. The welds were fine and the material was brittle, which points at filament that had already degraded or at a material being used outside what it can do.
A split that appeared during printing rather than in service is cracking and delamination — same physics, different trigger, and it needs the chamber addressed rather than the profile.
What a weld between layers actually is
When a hot bead lands on the layer below, polymer chains from each side have to migrate across the boundary and entangle with each other. That process needs the interface above a threshold temperature, and it needs time at that temperature. Strength builds continuously with both; there is no moment at which the weld is simply "done".
The only heat available is what the new bead carries with it. Nothing reheats the layer below except that contact. So every variable that matters is really a variable about heat: how hot the plastic was, how much of it arrived, how fast it was taken away again.
That framing predicts something people find surprising — the same profile produces a strong big part and a weak small one. On a part with a few square centimetres per layer, the nozzle returns before the previous pass has cooled. On a slender part it returns to plastic that went cold seconds ago.
The lever most people never touch
Print orientation changes interlayer strength more than every slicer setting on this page combined, because it decides whether the load crosses the welds at all. A hook, a bracket or a lever loaded along the beads is many times stronger than the same geometry loaded across them.
Before adjusting anything, look at how the part stands on the plate and ask where the force will go. Rotating a part ninety degrees is free, instant, and frequently ends the investigation.
A test protocol that gets a real answer
Guessing at temperature does not work, because the useful figure is specific to the spool, the hotend and the part.
- Model or download a small test coupon that fails in a predictable place — a thin tab, printed flat, is enough.
- Print three copies at the bottom, middle and top of the material's stated nozzle range.
- Break each one by hand or in a vice, in the same way, and rank them.
- Reprint the winner with the part-cooling fan halved, and compare again.
An evening of this replaces months of guesswork, and the answer is usually further up the temperature range than the default profile assumes.
The rest of the ranking
- Part cooling. The single most over-applied setting on small parts. Cooling improves overhangs and surface finish, and it directly steals weld strength.
- Layer height relative to nozzle. A layer taller than about three-quarters of the nozzle diameter gives the new bead less contact area and less pressure against the layer below. On a 0.4 mm nozzle, 0.3 mm layers are near that limit and 0.2 mm layers weld noticeably better.
- Line width. Widening the extrusion slightly increases the bonded area at each interface, at no cost in print time.
- Moisture. Water breaks polymer chains in the melt for polyamides, copolyesters and polycarbonate. A dried spool is not a marginal improvement in those materials; it is the difference between a part and a souvenir.
What does not help
Adding infill. Past about a third, extra density adds mass and print time and very little strength, because the load is carried by the perimeters. Another wall costs less and does more.
Annealing is not a general answer either. In the materials where it raises strength, it does so by changing crystallinity, and geometry moves with it — so treat it as a manufacturing step to be designed for, not as a rescue for a print that came out weak.
Materials worth knowing about
Silk and matte PLA both carry additives that improve appearance and reduce interlayer strength; they are display materials rather than functional ones. Polycarbonate and ABS weld poorly in a cold room, and reward a warm chamber more than a hotter nozzle. Straight PLA, printed hot with the fan turned down, welds better than most people expect — its weakness is heat resistance, not layer bonding.