Layer adhesion and anisotropy

An FDM part is not a solid block of plastic. It is a stack of beads laid down hot, each one welded to the one below as it cools. That weld never reaches the strength of the plastic itself, and from that single fact comes the most consequential decision in the whole part: which way you stack the layers. Get the orientation right and a modest part holds; get it wrong and the most carefully modeled geometry splits cleanly along a layer line under the first load. The problem isn't the material — it's the direction you left weak.

The plane between layers is the weak plane

Within a single layer, the nozzle drags one continuous, molten bead: the polymer chains tangle and flow together, so along the bead you're pulling on plastic that's almost continuous. Between layers the story changes. By the time the new bead lands, the layer below has already cooled past the temperature at which chains can diffuse across the boundary. The hot bead reheats it a little, the chains entangle only in the contact zone, and what's left is a partial weld. Partial is the key word for everything that follows.

That bond sits around 40–80% of the strength the part has in the XY plane, and standard PLA usually sits at the low end of that band, around 50–70%. The high end only shows up under favorable conditions — a warm chamber, a material that holds heat well, parameters tuned all the way in — so don't count on it. The conclusion doesn't change: the Z direction, perpendicular to the layers, is always the part's weak plane, and by a wide margin. A bracket that holds 40 kg with the load running along the layers may give way at 25 kg if you rotate that same load so it peels the layers apart. The shape is identical; only the direction the weakness points in has changed.

That's why an FDM part behaves more like a stack of paper — or plywood — than an isotropic solid: tough in the plane of the sheets, easy to separate sheet from sheet. It's the anisotropy introduced in How FDM Shapes Your Design, now seen from the strength side.

Rotate it, don't fill it

This is the mistake that ruins the most parts. When something breaks flat and brittle, the instinct is to crank up the infill. It's almost never the answer. If the fracture came out clean and followed a layer line, the good plastic didn't fail — the interlayer welds came unstitched. And you don't fix that by packing more material into a plane that wasn't doing any work.

The fracture plane hands you the diagnosis. A clean, rough fracture face perpendicular to the Z axis, the bead pattern on display, means the load was falling across the layers — the worst case there is. The fix is geometric, not a parameter. Rotate the part so the layers run along the direction it's being pulled, flexed, or pried in, not across it.

A hook printed in its loading plane pulls along the grain and holds; the same hook printed lying down peels the layers apart on the inside of the curve and breaks at a fraction of the load. An arm printed standing up takes bending across its weakest plane; printed flat on the bed, it carries that same bending through continuous material. Neither the shape nor the material changes — what changes is the direction the weak plane ends up facing. How to pick that orientation — and the conflict with overhangs that almost always shows up — is covered in Orientation and overhangs.

Temperature welds; the fan unstitches

The interlayer bond is a thermal process, so you have knobs to improve it without touching the geometry. The hotter the new bead arrives, the more it reheats the layer below and the better the chains diffuse across the boundary. Within each material's safe window, raising nozzle temperature buys real Z strength: often 10–20% going from the cold to the hot end of PLA's window. It's the difference between welding the layers and barely gluing them.

But that lever has a limit. Past the hot end of the window you stop gaining weld: you start degrading the polymer, increasing ooze and stringing, and — on hotends with a poor heat break — softening the filament inside the heat-break body and risking jams. "Always hotter" isn't the advice; the advice is to work at the hot end of the material's window, not above it.

The opposing factor is cooling. An aggressive part fan solidifies the interface before the weld finishes forming: the new bead never gets time to reheat the one below. So on parts where strength matters, it pays to turn the fan down. Slowing down helps for the same reason: a bead deposited more slowly keeps the interface hot longer.

None of this is free — that's the tuning trade-off. More heat and less fan weld better between layers, but you pay for it in overhangs. A bead that comes out hotter and cools more slowly sags as soon as it juts past the bead below, so overhangs lose definition and the underside comes out rougher. You have to choose per part: on one that'll take load, prioritize the weld and accept the uglier overhang; on a decorative piece with fine overhangs, the other way around.

Thinner layers, better welds

Layer height moves Z strength too, and it's often overlooked. Thinner layers multiply the number of interfaces, but each new bead is thinner and the layer below is closer to the nozzle when it takes the heat, so the overlap between bead and bead is greater and the weld more complete. Very thick layers do the opposite: fewer interfaces, but each with less well-fused area. On parts where the Z direction takes load, dropping from 0.3 to 0.2 mm, or from 0.2 to 0.12 mm, usually buys interlaminar adhesion in exchange for print time.

Don't confuse this with bed adhesion. Bed temperature and the first layer govern whether the part sticks to the glass and doesn't lift during the print — a different phenomenon, covered in How FDM Shapes Your Design. Here we're talking about the weld between layers inside the part, the one that decides whether it holds when you pull on it in Z. They're two problems that share the word "adhesion" and little else.

Every material starts from a different point

Not all plastics weld the same between layers, and that changes the starting point before you touch any parameter.

PETG usually welds better between layers than PLA, but not because it's stiffer — its modulus is in fact noticeably lower than PLA's. The cause is that it extrudes hotter (around 230–250 °C versus PLA's ~200 °C), holds heat longer, and forms a more continuous interface, so its anisotropy tends to be less pronounced. The same viscosity that helps the weld also makes it prone to stringing. PLA welds well and is the most predictable, but its interlayer bond sits in the low band and it's the most brittle when the load falls across the layers.

ABS and ASA need a chamber: without a warm environment, their shrinkage on cooling builds up stresses that separate the layers. And nylon has a problem of its own: moisture. It readily absorbs water from the air, and if you print it wet the absorbed water flashes to vapor as it crosses the melt zone, forming bubbles and voids that wreck the consolidation of each bead. Poorly dried nylon gives porous, Z-weak parts no matter how well the temperature is tuned; drying it isn't optional — it's the first step.

One nuance on the PETG row: "little or no fan" isn't an absolute rule. For the Z strength of a thick, load-bearing wall, yes: the less fan, the better the weld. But PETG usually needs some fan (20–50%) so bridges don't sag, to define overhangs, and to keep stringing under control. It's a trade-off, not a switch: turn it off where strength rules, turn it up where finish and overhangs rule.

Orient first, tune second, anneal almost never

Order matters. Anisotropy is fixed, first and foremost, with geometry: orient the grain so the load runs along the layers, and only then dial in temperature, fan, speed, and layer height to tighten the weld a few points more. Those parameters recover 10–20%; the right orientation can be the difference between the part breaking and holding, because it decides whether the load touches the weak plane at all. No nozzle setting saves a part oriented the wrong way.

That leaves annealing as a last resort, and it's worth treating it as what it is: heating a finished part above its glass transition lets the polymer crystallize further and can raise both stiffness and heat resistance. But that same crystallization makes it more brittle, and on top of that it shrinks and warps the part, typically 1–3% and rarely uniformly, so you can't anneal anything that relies on tight tolerances to fit. It doesn't rescue a badly oriented part: geometry that aims the weak plane into the load still aims it there after annealing.

The honest remedy for anisotropy is always the same: find the failure direction, lay the layers across it, and add material where the stress concentrates. How that material becomes more walls, and how orientation combines with the other parameters that shape the design, is taken up in Print Settings That Matter to Design.