The first layer and bed adhesion

Watch enough prints fail and you'll notice almost all of them fail in the same place: the first layer. If that layer doesn't sit flat and stay stuck, nothing you build on top will come out right — a corner peels, the part shifts, and four hours later you're scraping a tangle of stringy plastic off the bed. The first layer is the foundation, and like any foundation it decides whether everything above it holds. But there's a second effect almost nobody pins on the first layer, and it matters just as much as adhesion: that same layer, laid down wrong, moves your dimensions. It fattens the base, closes up the holes near the start, and leaves you with a part that stuck perfectly yet still doesn't fit.

Why the first layer decides adhesion

Every layer but the first prints onto plastic: hot bead welding to a bead that's already cooling. The first one has no plastic underneath — it prints onto glass, onto PEI-coated steel, or onto a textured sheet, a surface that molten plastic won't grip on its own. So the printer falls back on a trick: it brings the nozzle to less than one layer height above the bed and squashes that first bead against it (the squish). It spreads the bead flat, prints it hotter and slower, and keys the plastic mechanically into the texture of the surface while the plastic is still soft.

That squish is what sticks. Too high, and the beads come out as round threads that barely graze each other: little contact with the bed, little welding between lines, and the part holds on borrowed time until a corner lets go. Too low, and the nozzle scrapes, starves the flow, leaves gaps, and drags a ridge around the whole base. The sweet spot is narrow, and you can spot it by eye: squashed, wide beads, just touching their neighbors, with no ridges or grooves between them.

For that squish to be the same across the whole footprint, the bed has to be leveled: the nozzle at the same distance from the surface at every point. A tilted bed gives you a gorgeous first layer in one corner and none in the opposite one, because one side has squish to spare and the other leaves the bead round and unstuck. Leveling — manual or with an automatic probe — is the printer's job, not the designer's, but it's the first thing to check when a part always peels on the same side.

The squish that sticks is the one that deforms your base

Here's the trap most people never see coming. The squash you need to make the part stick is exactly the mechanism that deforms its base. When you compress that first bead below its layer height, the leftover plastic doesn't vanish: it squeezes out sideways. The nozzle leaves a smaller vertical gap than that volume of plastic needs, so it bulges out past the contour. That ridge around the base is the elephant's foot: the part swells a few tenths right at the start and recovers its nominal dimension a few layers up.

The effect always runs the same way. On an outer contour, the elephant's foot fattens it: the base comes out bigger than you drew it. In a vertical hole it does the opposite — the plastic bulges into the bore and pinches the mouth of the hole over the first few tenths of height, leaving a tighter neck than the rest of the bore. This isn't a dimensional drift across the whole part — it vanishes the moment the nozzle stops squashing — but a local defect of a few tenths, concentrated where the part meets the bed.

And it gets worse: the harder you squash to stick better, the more it bulges and the more elephant's foot you get. Adhesion and accuracy pull in opposite directions. Bed temperature matters here too: a bed that's too hot keeps the first layers soft for longer, so they settle and spread under the weight of the plastic above, exaggerating the ridge. That's why you fight elephant's foot with three levers at once: raise the z-offset a touch; drop the bed temperature on the first layers once the part has taken hold; and turn on the slicer's elephant's-foot compensation, which deliberately shrinks the first few millimeters of the contour to give it back its dimension.

Where the dimension is critical, chamfer the base

The elephant's foot is local and predictable, so the cleanest defense is geometric: give that ridge somewhere to escape without eating into your dimension. A 0.2 to 0.4 mm chamfer on the bottom edge of the part — a bevel that pulls material away right where the plastic was going to bulge — makes the elephant's foot fill that void instead of spilling past the nominal contour. The extra material still bulges; it just does so into a volume you'd already hollowed out on purpose, and the part recovers its size a few tenths higher. Size the bevel to the defect you measure, no more: a one-millimeter chamfer removes far more material from the fit than the elephant's foot ever adds.

This really matters when the base is a mating surface: the start of a shaft that enters a hole, the face of a part that seats in a pocket, the edge of a lid that centers against a lip. Without a chamfer, those swollen first tenths are exactly the ones that touch first and bind the assembly. With the chamfer, effective contact begins above the ridge, in plastic that came out on size.

Keep in mind that a chamfer printed upright comes out stepped, an artifact of the layer height itself. Its effect also competes with the elephant's foot in the same zone — you don't need a big bevel, just enough to swallow the ridge. And it's complementary to the slicer's compensation, not an alternative: the chamfer corrects the geometry you control; the compensation corrects what the process adds. But the compensation is a blind horizontal shrink of the contour on the first layers — overshoot the value (typically 0.1 to 0.2 mm) and you trim the footprint that actually sticks, weakening the very adhesion everything else here is trying to protect. Exactly how the squish moves the dimensions of holes and pegs — and by how much — is covered in Holes, pegs and first-layer squish.

Brim, raft, and bed temperature vs. warping

When a part won't stand on its own, the slicer has two ways to add contact with the bed, and each solves a different problem.

A brim is a flat skirt of extra perimeters, printed in the same first layer, fused to the contour of the part and spread outward over the bed. It adds gripping area right at the edge — which is where parts peel — and then pulls off by hand without leaving a mark. It's the right tool for the vast majority of adhesion problems: small footprints, sharp corners, and above all tall, narrow parts, which have a tiny footprint and a high center of mass. On a part like that, the moving nozzle exerts a torque on a minimal base and can topple it mid-print; a wide brim widens that base of support and anchors it.

A raft is a full platform that prints first, with the part sitting on top of it, separated by a thin gap. It weighs more, takes longer, burns plastic, and leaves the underside rough, but it forgives a poorly leveled bed and spreads out the stress of contraction, so it grips surfaces the part can't anchor directly. What decides whether it works is the air gap between raft and part: too narrow and they weld together, impossible to separate; too wide and the part rests on scattered points with an ugly underside. Reach for it only when a brim isn't enough.

But the biggest cause of peeling corners isn't solved by a brim alone: it's warping, and you fight it earlier with bed temperature. As it cools, plastic contracts, and that pull concentrates in the corners, which curl upward and lift off the bed. If the first layer doesn't grip hard enough, warping wins and the corner peels. The first line of defense is to give the plastic a surface that keeps it hot and adhered: bed at around 50–60 °C for PLA and 70–85 °C for PETG. In PLA warping is marginal and the extra heat buys no appreciable adhesion, so stay at the low end of the range; raising it only worsens the elephant's foot. PETG is the opposite, and it comes with a warning: it sticks too well to smooth PEI and can tear the coating off the bed when you remove the part. Here the fix isn't more heat but a separator layer — glue stick or hairspray — acting as a release agent, and a slightly higher z-offset.

Geometry does the rest. Orient the part so its widest flat face rests against the bed: more contact area gives more total grip, though bear in mind a large footprint also accumulates more contraction than a compact one. And round the corners of the footprint where you can: a sharp corner cools unevenly and concentrates the shrinkage right at its tip; a generous radius spreads it out and stays stuck to the bed. A fillet at the base is real adhesion geometry, not styling.

Design a part that wants to stay on the bed — flat face down, stable base, rounded corners and a chamfered base where the dimension demands it — and the first layer will almost solve itself. All of this is a direct consequence of how the machine lays plastic down; if you want the full mental model behind all this, it's in How FDM Shapes Your Design.