← Notes

When the bond fails, the mold was probably too cold

This is where the series has been heading. Three notes on how the bond works, where it is used, and what it is made of, all to arrive at the one that matters on a shop floor: the part came out, the plastic peels off the metal, and someone has to say why. Insert-molded metal-resin parts fail in a way that is easy to misread, because the failure is invisible until you pull on it, and by then the shot is long gone.

The single most useful idea for reading these failures is that bond strength is a product, not a sum. It is the quality of the surface treatment multiplied by how completely the resin filled it. Multiply, because if either one is zero, the joint is zero — a perfect treatment molded cold gives you nothing, and a perfect molding onto a dead surface gives you nothing. That framing splits every bond failure into two questions with very different answers: was the surface good when it arrived, and did I fill it. You own one of those. You can only inspect the other.

The two failures that came in the box

Two of the three ways a bond fails are decided before the insert reaches your press, and no setting recovers them.

The first is the treatment itself — an under-etched or mis-processed surface that never had the pore structure to grip. The second is more common and more frustrating: a good surface that was spoiled in handling. The nanopores that do the gripping are tens of nanometres across; a fingerprint, an oil film, condensed moisture, or simply too many days on a shelf will contaminate or degrade them. Treated inserts have a shelf life and a handling spec for exactly this reason, and a batch that sat too long or got touched with bare hands can look identical to a good one and bond like a stranger.

Neither of these is yours to fix at the press, and the important discipline is to not try. Cranking the process to force a bond onto a bad surface makes a part that passes the pull test today and fails in the field, which is worse than a part that fails now. The press's job with an upstream failure is to detect it and send it back — confirm the treatment and the handling, quarantine the suspect lot, and do not mold through it.

There is a tell that separates the two worlds, and it comes straight from part one. A good bond fails in the plastic: force it apart and the resin tears and stays behind on the metal. A bond that fails clean — the resin lifts off and the metal underneath looks bare and shiny — never wet into the surface at all. Clean separation points upstream, to treatment or contamination. Resin left on the metal, but not enough strength, points at you: the surface was good and you did not fill it. Learn to read the fracture face and you have already narrowed the cause in half.

Which defects are the press's, and which came in the box Insert-molding defects sorted into two columns. Fixable at the press: a weak bond with resin left on the metal points to infiltration, so raise mold temperature, injection speed and packing; sink and flow marks around the insert mean a cold insert chilling the melt, so raise mold temperature and preheat the insert; a weld line landing on the joint is moved with a gate change. Send back upstream: a weak bond that separates to bare clean metal is a treatment or handling failure; flash points to insert dimensions or a damaged shut-off; an insert that shifts or deforms is a fixturing problem. The fracture face decides the first split: resin on the metal is the press's problem, bare metal is the box's. A part that must seal is proven with a leak test, not a pull test. Read the defect back to its cause Fix at the press Weak bond, resin left on the metal infiltration — raise mold temp, speed, packing the surface was good; you did not fill it Sink & flow marks around the insert cold insert chilling the melt raise mold temp, preheat the insert Weld line landing on the joint two flow fronts meeting at the interface move it with a gate change Send back upstream Weak bond, bare clean metal treatment or handling failure never wet in — do not mold through it Flash insert dimensions or damaged shut-off incoming inspection / tooling Insert shifts or deforms support missing, one-sided pressure fixturing / gating The fracture face decides the split: resin on the metal is yours; bare metal came in the box. And if the part must seal, prove it with a leak test, not a pull test — a seal does not average.

Mold temperature, and why you measure it instead of setting it

When the surface is good and the bond is still weak, the cause is almost always the same, and it is the most under-used lever on the machine: the mold was too cold.

Here is the physics, and it is worth holding onto because everything else follows from it. When the melt hits the metal, it starts to freeze against it immediately, forming a solid skin. The patent behind this whole field states the failure in one line: the injected resin solidifies before it enters the fine recesses of the treated surface. Those recesses are tens of nanometres wide. The skin that freezes against a cold wall is orders of magnitude thicker than that. So the moment a skin forms at the interface, the pores below it are sealed off with air still in them, and no amount of pressure afterward pushes resin through a frozen lid. The bond is decided in the instant the melt arrives, and mold temperature is the one lever that buys the melt time to stay liquid at the wall long enough to flow in before it sets.

This is why the resins in this family are all the fast-freezing, semi-crystalline ones — PPS, PA, PBT, PET, glass-filled — and why they are so unforgiving. They crystallize quickly against cold steel. The datasheet mold temperature for a glass-filled PPS bonding grade sits around 120°C, and here is the trap: that is a setpoint, and the setpoint is not what the melt feels. The temperature that matters is the actual cavity-surface temperature where the resin meets the insert, and measured with a surface pyrometer that number is routinely lower than the controller reads. The working rule from the floor is to measure, not assume, and to hold the measured surface at 150°C or higher for these parts. The patent goes further still and describes preheating the metal insert itself above 200°C — the same idea taken to its limit, keeping the interface hot enough that the resin stays fluid until it has rooted.

There is a practical trap in how you reach that number. Most shops raise mold temperature with a water or oil unit circulating fluid through the tool, and set to 150°C that way, a probe on the cavity face often reads under 100°C — the circulating fluid cannot carry enough heat to the surface against the losses. Hitting a measured 150°C at the steel usually takes cartridge heaters inserted directly into the mold, not a higher number on the controller. And 150°C is a floor, not a target: for these joints hotter is better, and holding the measured surface at 180°C or above is fine and often helps the resin find the pores.

If you change one thing on a weak-bond part with a good insert, change this, and confirm it with a thermocouple on the steel rather than a number on a screen. It is also the change shops resist most, because a hotter mold means a longer cooling time and a slower cycle. That trade — cycle time against bond integrity — is the real decision hiding inside most of these failures.

Why a cold mold starves the bond Two cross-sections of resin meeting a treated metal surface. Against a cold mold, a solid skin freezes at the interface the instant the melt arrives and seals the nanopores with air still trapped inside, so the resin never enters and the bond is weak. Against a hot mold, the interface stays molten long enough for the resin to flow down into the pores and freeze there as anchors, giving a strong bond. Mold temperature is the lever that buys that time, and it is measured at the steel, above 150 degrees, not read off the setpoint. The bond is decided in the instant the melt arrives cold mold — skin freezes first resin a solid skin seals the pore mouths metal — pores sealed, air trapped, unfilled pores sealed → weak bond hot mold — interface stays molten resin metal — resin flows in and freezes as anchors pores filled → strong anchor Mold temperature buys the melt time to fill before it freezes. Measure it at the steel — 150°C or above for these resins — not off the setpoint.

Speed, pressure, and hold

Mold temperature buys time; the other three levers use it.

Injection speed works the opposite way from intuition. Filling faster gives the flow front less time to freeze against the wall on its way to the joint, so it arrives hotter and more fluid and with more of its skin still molten. On a part fighting a weak bond, a faster fill often helps for the same reason a hotter mold does — it keeps the interface alive.

Pressure and hold do the actual driving. Once the melt is against the treated surface and still molten, packing pressure is what forces it down into the pores, and holding that pressure while the interface freezes is what keeps it there until it sets. Too little pack, or a hold that releases before the skin has solidified, and the resin relaxes back out of the structure it was starting to fill. The instinct to read this as "push harder" is a trap, though: past a point, more pressure and more speed mean more shear and more heat, and you begin degrading the resin and the fiber-to-polymer bond at the very interface you are trying to build. The goal is not force. It is to keep the melt molten and fill the pores — hot enough, fast enough, packed and held long enough — not to hammer a cold shot into a surface that has already sealed.

The defects that are not the bond

Not every insert-molding defect is a bond-strength problem, and the surface finish tells you about the ones that are not.

Sink marks and flow marks in the resin right around the insert usually mean the metal is chilling the melt locally — a cold insert acting as a heat sink, pulling the surface down as the material around it freezes unevenly. The answer is the same family as before: raise the mold temperature, and preheat the insert so it is not the coldest thing in the cavity. Flash, by contrast, is rarely about temperature; it points to the insert's dimensions or a damaged shut-off, which is an incoming-inspection and tooling problem, not a process one. An insert that shifts or deforms during the shot is a fixturing and gating problem — the support was not there, or the melt hit it one-sided and pushed it. These are worth separating out precisely because the reflex — reach for the process — is wrong for them.

And then there is the failure that hides, the one part two warned about. When the part's job is to seal, a bond that is merely weak is not the worst outcome; a bond that leaks is. A joint can pull-test fine and still have a local void, a discontinuity in the treatment, or a weld line running across the interface where two flow fronts met — any of which is a path for water. A seal does not average. It is why these parts are checked with air or helium leak tests and not just mechanical pulls, and why a weld line that lands on the joint is worth moving with a gate change before it ever becomes a leak. The pull test asks whether the joint is strong. The leak test asks whether it is continuous, and for a sealed part that is the question that ships or scraps it.

The order of operations

So a bond failure has a sequence, and the value is in running it in order.

First, confirm the insert. Read the fracture face: clean metal means the surface or its handling failed, and the part goes back upstream, not through a hotter mold. Resin left behind but weak means the surface was good and the fill was not — now it is yours. Then, in order of cost and impact: raise the mold temperature and verify it with a probe on the steel, not the setpoint, at 150°C or above for these resins. Raise injection speed. Increase pack and hold enough to drive the resin into the pores and keep it there while it freezes, without tipping into the shear that degrades it. And if the part must seal, prove it with a leak test, not a pull. The sequence matters because the first step decides whether any of the others can possibly help — and most of the wasted shifts on these parts come from skipping it and molding harder into a surface that was never going to hold.

That is the series. The bond is nano-scale geometry, filled in the first seconds against cold metal, on a surface someone else prepared — invisible physics that decides whether a phone survives a drop, a battery cover stays sealed, a hydrogen tank holds. Most of it happens away from the press. But the last variable, the one that turns a good surface into a good joint or wastes it, is the press, and it comes down to keeping the melt alive at the metal long enough to fill what it was given.

Mold Doctor takes a photo of the defect and your process settings and estimates the likely causes and what to adjust. For an insert-molded metal-resin part, tell it the resin is glass-filled and the insert is treated metal, and the questions it already asks — mold temperature, fill speed, packing — are the same ones this note has been about, pointed at the smallest and most demanding joint injection molding makes.

JINSIMLABS.COM