Nobody spots it at the casting machine. The part comes off the die looking clean, dimensions are in spec, and then three weeks later your customer reports that a housing rated for 3 bar fails a pressure test at 1.5 bar — with a fine weep of air through the wall. That is porosity behaving as a leak path, and it is one of the most common reasons die castings get rejected after machining.
There are two honest ways to deal with it: fix the casting process, or seal the pores. This guide walks through the second path — resin impregnation — and explains when it is the right call, when it just hides a problem, and how leak testing fits into the workflow.
Why Die Castings Leak
High-pressure die casting fills the cavity in 30–80 milliseconds with metal moving at 20–60 m/s. That speed entraps air and lubricant decomposition gases in the melt, and as the casting solidifies from the outside in, thick sections can also develop shrinkage voids. Most porosity is harmless — round, dispersed, below 5% of wall volume. The trouble starts when pores connect into a continuous path through the wall, typically where:
- a thin wall meets a thick boss, and shrinkage draws gas inward;
- machining cuts within 0.3–0.5 mm of a pore cluster and opens it to the surface;
- the parting line or overflows trapped gas along a predictable band.
Pressure-tight applications — valve bodies, pump housings, battery enclosures, pneumatic manifolds — cannot tolerate these paths. A casting that passes visual inspection can still weep under 2 bar. That is why pressure-tight die cast parts are almost always specified with a leak test, not just dimensional checks. If your part lives in this world, budget for it when you request a quote from a die casting supplier, because 100% leak testing can add real cost per piece.
Leak Testing: Know What You Are Measuring
Three methods cover most production reality:
- Pressure decay. Pressurize the part (plugged ports), isolate it, watch the pressure drop over 10–60 seconds. Cheap and fast, but it cannot locate the leak and is sensitive to temperature drift. Best for high-volume go/no-go screening.
- Immersion bubble test. Pressurize 1–5 bar (typically 1.5–2× working pressure), dunk the part, look for bubbles. Slower and operator-dependent, but it shows where the leak is — which is exactly what you need when deciding between impregnation and a die revision.
- Helium sniffing or bombing. For small, critical parts (fuel systems, medical fluidics). Detects leaks orders of magnitude smaller than bubble tests, at corresponding cost. Rarely justified for industrial housings.
Set your test pressure realistically. Testing at 5× working pressure sounds rigorous but rejects castings that would never leak in service, and drives you into unnecessary impregnation. A common starting point: test at 1.5–2× maximum working pressure, with a decay or bubble threshold agreed in writing with the customer.
What Impregnation Actually Does
Impregnation forces a liquid thermosetting resin (usually a methacrylate or anaerobic system) into the connected pore network and cures it solid in place. A standard cycle looks like this:
- parts are cleaned and dried (moisture ruins adhesion);
- the vessel draws vacuum to roughly 20–50 kPa absolute for a few minutes, pulling air out of the pores;
- resin is introduced and pressure applied (3–6 bar) to drive it into the pore network;
- pressure is released, parts are rinsed, and the resin cures — typically in hot water at 85–90 °C for 10–20 minutes.
Commercially, impregnation reliably seals pores up to about 0.5 mm and leaves a hard, machinable, permanent seal inside the wall. It survives normal service temperatures (most resins are good to 200 °C briefly, ~150 °C continuous) and does not attack subsequent powder coating or e-coating. One caveat: if the part will be anodized later, discuss sequencing with your finisher — impregnation resin in surface pores can cause color variation, the same family of problems high-silicon alloys already have with anodizing.
When to Impregnate — and When Not To
Impregnation fixes leak paths. It does not restore strength, so it is the wrong answer when porosity is large enough to affect structural integrity. A practical decision frame:
- Leak rate small, pores ≤ ~0.5 mm, walls otherwise sound → impregnate. This is the textbook case, and for many part families (hydraulic covers, gear housings) impregnation is standard production practice, not a rescue measure.
- Leak rates under 2–3% of production → impregnate the rejects selectively and treat the cost as scrap recovery.
- Leak rates climbing above 5–10%, or leaks tracing to one die region repeatedly → the die is the problem. Impregnation becomes a permanent tax on every part. Fix the process instead.
- Porosity large and shallow enough to matter for strength or plating → rework or scrap. Sealing a structural void with resin is false economy, and any plating shop will reject a porous surface anyway.
Fixing It at the Die Instead
Before accepting impregnation as a permanent process step, push your caster on the upstream causes, because most chronic leak paths are die-side:
- Venting — increase vent depth to the material limit (roughly 0.010–0.020 mm for aluminum alloys before flashing) or add vents where bubbles cluster.
- Overflow placement — overflow wells should sit at the last point the metal fills; missing them there guarantees trapped gas at exactly the wrong location.
- Gate speed and shot profile — reducing in-gate velocity from ~50 to ~30 m/s, or switching to a slower slow-shot/fast-shot crossover, cuts gas entrainment measurably. Simulation helps target these changes rather than guessing.
- Local squeeze pins — for shrinkage porosity at a boss, a squeeze pin that closes just after cavity fill can consolidate the thick section and eliminate the void at its source.
The Sequencing Rule Most Programs Get Wrong
Machining opens porosity. A casting that passed leak test as-cast can fail after you machine a mounting face 0.2 mm deep into a pore cluster. So the order matters:
- machine first, leak test second;
- impregnate after machining, not before;
- leak test again after impregnation — 100% if the part is safety-critical, sampling otherwise;
- apply coatings last, after any resin curing is complete.
Skipping the intermediate test hides whether leaks come from the casting or from your machining allowance, and it makes supplier conversations pointless.
Bottom Line
Impregnation is a mature, cheap-per-part way to rescue otherwise good castings from micro-leaks — but it is a seal, not a repair. Use it deliberately: define a leak test with your customer, track failure rates by die region, and treat rising impregnation volume as an instruction to revise vents, overflows, or gate speed. Programs that do this keep impregnation as a small insurance line item. Programs that don't, quietly turn it into the most expensive resin they ever bought.
For parts that must combine pressure tightness with tight-tolerance features, our CNC machining team regularly handles post-casting second operations with porosity-aware fixturing — worth discussing before the die is designed.
