Porosity Acceptance Criteria for Die Castings: X-Ray, Leak Test and Impregnation Specs

Why "no porosity" is not a specification

Every high-pressure die casting contains micro-porosity. The metal is injected at high speed into a steel die, air and volatile residues have to escape through vents in milliseconds, and the solidification shrinkage of the alloy leaves microscopic voids at hot spots. This is true of every HPDC part made anywhere in the world, from any supplier, at any price. A drawing note that forbids porosity outright does not describe a better casting; it describes a rejection that has not happened yet.

What buyers actually need is functional assurance: the part does not leak, does not fail under load, and does not show voids where it matters — on sealing faces, threaded features, structural ribs or visible surfaces. Porosity acceptance criteria are the bridge between that functional need and the foundry's process. Written well, they let the supplier control the process where it counts and quote honestly; written poorly, they push both sides into 100% X-ray of everything, endless rework loops and finger-pointing at FAI.

The three criteria that make porosity verifiable

1. Critical areas, marked on the drawing

Start by marking where porosity actually matters: sealing faces that carry an O-ring, walls adjacent to leak paths, areas that will be welded or machined to a finish, and any region the FEA model flagged as load-bearing. The rest of the casting gets a general grade; the critical areas get a stricter one. A single uniform criterion across the whole part is the most expensive way to buy porosity control, because it forces the tightest requirement onto areas that never see stress.

2. A radiographic grade tied to a reference standard

For aluminum and magnesium die castings, the working reference is ASTM E505 — reference radiographs produced specifically for die castings, with graded classes for gas porosity and shrinkage. A note such as "porosity grade 3 or better per ASTM E505 within the marked sealing zone; grade 5 elsewhere, X-ray per ASTM E94 practice" gives the X-ray operator a physical standard to compare against, not an opinion. For sand and gravity castings the parallel document is ASTM E155, but die casting buyers should anchor to E505 because its discontinuity classes match what HPDC actually produces.

Radiography itself is a sampling decision, not a religion. Common practice grades from one shot per setup on a new program toward statistical sampling as capability data accumulates. Parts whose leak integrity is safety-relevant may stay on 100% X-ray for the life of the program — that is a cost decision the drawing should make explicitly, not a foundry default.

3. A leak test with a method, a pressure and a reject rule

"Leak-tight" is not a criterion. A verifiable note states the method, the medium, the pressure and the pass rule. The three methods in everyday use sit at different cost points:

• Pressure-decay — pressurize, isolate, watch the gauge. Fast and dry; suits high-volume housings with moderate tightness targets.
• Immersion (bubble) test — pressurize and submerge; a visible bubble train pinpoints the leak. Cheap, sensitive enough for many enclosures, but operator-dependent and wet parts need drying.
• Helium mass spectrometry — the sensitive end of the scale, for refrigerant and braking circuits where leak rates far below what bubbles reveal still matter. Highest equipment cost, usually justified on program volume.

The test pressure is normally agreed case by case against the service condition — a common engineering approach is to test above the maximum working pressure by a margin the two parties sign off in the control plan, because an arbitrary multiple either wastes cycle time or lets real leaks through. Whatever the number, it belongs in the PPAP control plan and on the FAI report, not in an email thread.

Impregnation: when to allow it and when to demand it

Vacuum impregnation forces low-viscosity resin into the casting's connected porosity and cures it, sealing leak paths permanently. Done properly to a recognized impregnation practice — MIL-STD-276 is the classic military reference, and commercial resin suppliers publish equivalent process specs — it does not alter the alloy matrix, dimensions within any meaningful tolerance, or mechanical properties. It converts a scrapped casting into a good one.

The strategic question is where impregnation sits in your quality architecture. Two schools exist:

• Impregnation as standard for leak-critical parts — the supplier impregnates every part before the leak bench. Predictable cost per part, near-zero leak escapes, and the foundry's process capability matters less at the leak bench.
• Impregnation as rescue — parts are leak-tested as-cast; only failures are impregnated and retested. Cheaper when porosity is genuinely rare, but it makes impregnation rate a live KPI of process health, which is exactly what you want to watch on a new tool.

Either can be correct. What is not correct is discovering at the PPAP stage that the buyer assumed one model and the supplier quoted the other. State the model on the drawing.

Porosity on machined faces: the criterion everyone forgets

The die skin of a die casting is dense and sound; porosity concentrates beneath it. Machining strips that skin, which is why voids appear on machined faces that X-ray of the as-cast part never flagged. Buyers of machined castings therefore need a surface criterion in addition to the radiographic one. Typically agreed windows include:

• a maximum pore diameter on functional faces (sealing lands, bearing bores), commonly in the fraction-of-a-millimeter range;
• a maximum number of visible pores per unit area on cosmetic or coated surfaces;
• no connected porosity across a sealing land — a single pore of the allowed size is acceptable, a chain of them is not;
• a minimum remaining wall between the machined surface and any internal void, so the part keeps its pressure rating after machining.

The exact numbers are a function of the application and should come out of the DFM conversation, not out of a copied template. What matters is that they exist in writing before the first article is cut, because re-negotiating surface porosity limits after machining capacity is booked is how programs slip.

How a capable supplier attacks the problem upstream

Criteria are the contract; process is what actually delivers the casting. The upstream levers, in the order they should be pulled:

Vacuum assist evacuates the shot sleeve and die cavity before injection, removing most of the entrapped air that becomes gas porosity. For gas-tight parts it is the single highest-leverage investment — our article on vacuum die casting and porosity control covers the mechanics. Simulation before steel is cut predicts where the last metal solidifies, which is where shrinkage porosity will form; our tooling design and mold-flow guide explains how overflows, cooling channels and gate geometry are placed to move that porosity into non-critical areas. Shot-profile discipline — the slow-shot and intensification settings that our maintenance guide on shot sleeve and plunger tip wear touches on — keeps every shot identical, because porosity variance between shots is what forces 100% inspection in the first place.

The causes and mechanisms behind all of this are covered in depth in our companion article, where die casting porosity comes from and how it is managed. This article deliberately stays on the acceptance side: what the buyer writes down, and what evidence the supplier owes.

What OKDieCast provides against these criteria

OKDieCast is the die casting brand of Shenzhen Leijing Technology Co., Ltd., operating a 22,000 m² integrated plant with high-pressure and vacuum die casting, 5-axis CNC machining to ±0.05 mm, leak test benches and X-ray inspection, and full batch traceability reports with every shipment. The site is ISO 9001 and IATF 16949 certified. New programs move from drawing to sample in 7 days, which is early enough for the porosity criteria to be tested on real parts — not discovered after the tool is paid for. Drawings can go to our engineering team for DFM review and a quote within 24 hours.

Frequently asked questions