The two defects most often confused are gas porosity, which shows as round smooth-walled voids 0.05-0.5 mm across caused by entrained air, and shrinkage porosity, which shows as jagged interconnected voids at the thermal centre of sections thicker than about 6 mm. Gas porosity is fixed by slow-shot velocity, venting, melt degassing and vacuum assist; shrinkage porosity is fixed by thinning the section, keeping wall variation under 2:1, raising intensification pressure toward 100 MPa and cooling the hot spot. Fill defects such as cold shut and misrun come from a metal front that reached 20-30% solid before the cavity filled, and are fixed with die temperature of 180-280 °C, gate velocity of 30-50 m/s and shorter flow paths.
A customer sent back four rejected housings with a note saying “porosity”. Three had gas porosity from a slow-shot velocity 40% too high. The fourth had shrinkage porosity in a 9 mm boss that no process change would ever fix. Same word on the rejection tag, opposite root causes, and one of them was a design problem reported as a process problem.
That is the recurring theme of defect work here. The symptom is rarely specific enough to identify the cause, and the fix is often on the drawing rather than the machine.
Diagnostic table
| Defect | Appearance | Root cause | Design fix | Process fix |
|---|---|---|---|---|
| Gas porosity | Round smooth voids 0.05-0.5 mm below the surface | Entrained air in sleeve or cavity, or dissolved hydrogen | Thin the wall, remove blind pockets at last-fill points | Tune slow-shot velocity, degas, enlarge vents and overflows, vacuum assist |
| Shrinkage porosity | Jagged dendritic voids at the thermal centre of a thick section | 6% contraction with no liquid feed path | Thin the section, variation under 2:1, core bosses, use ribs | Raise intensification, extend dwell, cool the hot spot, re-gate |
| Cold shut (cold lap) | Seam or crease where two flow fronts met, weak in bending | Fronts met below fusion temperature | Increase local wall, shorten flow path, do not split the flow | Raise die and metal temperature, faster gate, add an overflow |
| Misrun (short fill) | Incomplete part, rounded thin edges furthest from the gate | Front reached 20-30% solid before fill completed | Thicken section, flow-length-to-wall under 100, move the gate | Raise temperatures, more shot volume, faster fast shot, better venting |
| Flash | Thin fin at the parting line, slides or ejector pins | Die opened under injection load, or worn parting faces | Reduce projected area, keep the parting line planar | Higher tonnage, lower intensification, resurface faces |
| Sink mark | Shallow depression opposite a rib, boss or junction | Local mass shrinks inward, pulling the skin | Rib 60-80% of wall, boss base under 2× wall, core it | Raise intensification, cool the hot spot, lower metal temperature |
| Blister | Raised dome appearing after bake, heat treatment or service heat | Subsurface gas expands as the skin softens | Thin the wall, remove trapped-air pockets | Reduce entrained air, vacuum assist, lower bake, never solution treat |
| Soldering (die sticking) | Torn patch on the casting, aluminum build-up on the die | Aluminum chemically welding to die steel | Add draft, fillet the area, avoid thin cores in the hot zone | Fe 0.8-1.1%, cool the spot, slower gate, PVD-coat the insert |
| Drag mark | Parallel scores running in the ejection direction | Insufficient draft, die damage or galling | Draft 1-2° external, 2-3° internal, pin near the drag | Polish in the draw direction, repair damage, balance ejection |
| Flow lines and streaking | Faint swirls or matte and glossy patches, no dimensional effect | Uneven flow front, non-uniform die surface and spray | Move cosmetic faces off the gate and last-fill zone, add texture | Stabilise spray, raise die temperature, re-gate |
| Heat checking print-through | Fine raised network of lines, worsening over tool life | Thermal fatigue cracks in the die printing onto parts | Fillet corners at 0.5× wall, min 1 mm, no thin die blades | Control die temperature, avoid spray shock, refurbish on schedule |
| Warping and distortion | Out of flat or twisted, often worse after machining | Uneven cooling, unbalanced ejection, released residual stress | Symmetrical ribbing, uniform walls, cast datum pads | Balance ejection, longer dwell, cooling fixture, T5 relief |
| Inclusions and hard spots | Gritty bright particles when machining, chipped inserts | Oxide films, refractory debris, iron-silicon intermetallics | Avoid machining into thick section centres | Filter the melt, Fe under 1.1%, clean furnace, no long holding |
| Cracks and hot tearing | Irregular cracks at corners, rib junctions or thin webs | Shrinks against rigid steel and tears while semi-solid | Fillet corners, avoid restrained geometry, blend sections | Lower metal temperature, even die temperature, eject earlier |
Porosity: two defects sharing one name
More scrap comes from treating these as one problem than from any other mistake.
Gas porosity
Round voids with smooth, shiny walls, typically 0.05 to 0.5 mm across, appearing anywhere in the casting and usually first seen on a machined face.
Most of it is air mechanically entrained during injection: metal in the shot sleeve breaks into a wave and folds air into itself, or the cavity cannot vent fast enough and the last air is compressed into the metal at the final fill point. A smaller share is hydrogen, which dissolves in molten aluminum and drops out of solution as it freezes.
Slow-shot velocity is the largest process lever, and its optimum is a specific value for a given sleeve diameter and fill fraction, not “as slow as possible”. Beyond that: melt degassing, larger vents, more overflows, and vacuum assist to 50 to 150 mbar, a trade-off covered in high pressure die casting explained.
Shrinkage porosity
Jagged, branching, interconnected voids with rough walls at the thermal centre of the thickest local section, frequently paired with a sink mark on the nearest outer surface.
Aluminum contracts roughly 6% by volume as it solidifies. Wherever the last liquid in a region freezes with no path back to pressurised metal, that contraction comes out of the metal itself as a void. It is a feeding problem, not a gas problem.
In most cases this is a design defect. Thin the section, keep wall variation under 2:1, core out solid bosses, and replace mass with ribs. Sections above about 6 mm hold porosity at their centre however the process is set, which is why the design guide treats section thickness as the primary variable. Process work helps at the margin: intensification toward 100 MPa, longer dwell, local cooling, and re-gating.
Telling them apart
| Observation | Gas porosity | Shrinkage porosity |
|---|---|---|
| Pore shape | Round, smooth walled | Jagged, dendritic, branching |
| Pore size | 0.05-0.5 mm, fairly uniform | Variable, often larger, interconnected |
| Location | Anywhere, including thin walls | Thermal centre of the thickest section |
| Correlation with geometry | Weak | Strong, always the heavy section |
| Associated sink mark | No | Often |
| Response to higher intensification | Small | Significant |
| Response to slow-shot tuning | Significant | None |
The fastest field diagnosis: gas porosity moves around shot to shot, while shrinkage porosity sits at the same coordinates on every part, because the thermal centre of a boss does not wander.
Fill defects
Cold shut and cold lap
The metal front does not advance as a single wave. It splits around cores, ribs and holes, and the branches meet again downstream. If both have cooled below the temperature at which they can fuse, they meet without welding and leave a seam.
The consequence is worse than the appearance. A cold shut is an internal crack with a clean interface, so the part can lose most of its bending strength across that line while looking almost acceptable, and one near a sealing surface is a direct leak path.
Design fixes are geometric: more wall along the affected path, shorter flow distance, and no features that split the flow far from the gate. Process fixes are thermal and kinetic: die and metal temperature toward the upper ends of 180-280 °C and 660-700 °C, gate velocity within 30 to 50 m/s, and an overflow where the fronts meet.
Misrun and short fill
A misrun is a cold shut that ran out of time entirely, leaving the part incomplete with rounded, thinned edges furthest from the gate. Check the ratio of flow length to local wall thickness first: under 60 is comfortable, up to 100 workable, above 100 is designing in a misrun. A 1.5 mm wall reaches roughly 90 to 150 mm from its gate. Venting is as likely a cause as temperature, because a cavity that cannot exhaust its air presents back-pressure and the front stops.
Flow lines and cosmetic streaking
Faint swirls or alternating matte and glossy regions with no dimensional consequence, caused by an uneven flow front plus variation in die surface and release agent coverage. Some visual variation is inherent at Ra 1.6 to 3.2 µm, so handle it honestly at quotation: move cosmetic requirements away from the gate and last-fill regions, specify a shot-blast or textured finish, or coat the part through surface finishing.
Pressure and parting line defects
Flash
A thin fin at the parting line is normal at 0.05 to 0.15 mm and is trimmed off. It becomes a defect when it is heavy, appears at slides and ejector pins, or grows across a run. Heavy flash means separating force exceeded clamping force and the die opened slightly, which is a tonnage error or an intensification pressure raised to chase porosity. Growing flash means worn parting faces or platen parallelism drifting.
Sink marks
A shallow depression on an outer surface, always directly opposite a rib, boss or thick junction. The mechanism matches shrinkage porosity, except the shrinkage pulls the soft skin inward instead of opening a void, and a part often has both at the same location. Because it is a mass problem the design fix is the reliable one: ribs at 60 to 80% of the adjoining wall, boss base under twice the wall, cored bosses, and a relief pocket behind a visible face.
Blisters
A raised dome that was not there when the part shipped, appearing after a paint bake at 180 to 200 °C, heat treatment, or hot service.
The gas pocket was always there, just below the surface behind a thin skin, and heating expands it while softening the aluminum until the skin domes. This is the direct reason die castings are never heat treated to T6, and why a blister appearing in the paint shop is an injection defect rather than a coating defect.
Die condition defects
Soldering and die sticking
Molten aluminum chemically welds itself to the die steel. The casting comes out with a torn patch and the die carries a matching build-up, usually near the gate where metal is hottest and fastest.
Iron content is the central control. Aluminum dissolves iron, and a melt already carrying 0.8 to 1.1% iron is far less aggressive toward the die than one at 0.5%. Above roughly 1.1%, ductility falls and coarse intermetallics form, so the window is narrow and needs verifying per lot. The other controls are local die cooling, lower gate velocity, adequate draft and fillets, consistent release agent coverage, and PVD-coated inserts where a location solders repeatedly.
Drag marks
Parallel score lines running in the ejection direction, because the casting was dragged across die steel it was still gripping. Causes in order of frequency: insufficient draft, die surface damage or galling, polishing marks running across the draw direction, and unbalanced ejection. The design fix is draft, 1 to 2° external and 2 to 3° internal, plus an ejector pin near the area that drags.
Heat checking printed on parts
A fine network of interconnected raised lines, absent on early parts and progressively worse across the tool’s life. The die is cracking: its surface cycles between 180-280 °C and near 700 °C every shot, and thermal fatigue opens a craze pattern at the sharpest internal corners and around the gate.
No process fix reverses it, only measures that delay it: stable die temperature, no thermal shock from over-spray, and scheduled refurbishing. The design fix is decided long before production: fillet every internal corner at 0.5 times the wall, never below 1 mm, and the die reaches the upper end of the 80,000 to 150,000 shot range.
Dimensional and metal quality defects
Warping and distortion
Parts out of flat or twisted, sometimes on ejection and sometimes only after machining. Three mechanisms produce it: uneven cooling, unbalanced ejection bending the part while it is soft, and residual stress released when a cut removes material.
The third surprises people, because the casting measured flat before machining. Stress relieve with a T5 cycle at 200 to 250 °C, cut in two passes so a rough pass releases stress and a finish pass sets the dimension, and design cast datum pads so the fixture avoids drafted surfaces. Fixture strategy is covered in machining after casting.
Inclusions and hard spots
Usually discovered on the machining line: an insert chips, tool life collapses, or a bright hard particle appears in a bore. The causes are oxide films folded into the metal, refractory debris, and coarse iron-silicon intermetallics formed when iron runs high or metal is held too long. This is almost entirely a melt management defect, controlled by filtration, degassing, iron below 1.1% and clean furnace practice. Separately, high-silicon alloys are abrasive, so assume shorter tool life in ADC12 than in A380.
Cracks and hot tearing
Irregular cracks at sharp corners, rib-to-wall junctions and thin webs, often first seen after trimming. The casting contracts as it cools while the steel around it does not, and if the geometry prevents free contraction the part tears while still partly liquid at the grain boundaries. Design controls: fillet the corners, blend section changes, and avoid geometry that restrains shrinkage, such as a thin web spanning two heavy features. Process controls: lower metal temperature, uniform die temperature, earlier ejection, balanced ejection force.
Detection methods
| Method | Finds | Coverage | Notes |
|---|---|---|---|
| Visual to a lit standard | Flash, cold shut, misrun, drag, soldering, sink, streaking, blisters | 100% typical | Needs a zoned drawing defining cosmetic faces |
| Dimensional, CMM and gauges | Warping, distortion, dimensional drift | First article plus SPC | Measure against declared cast datums |
| Dye penetrant | Surface cracks, cold shuts, hot tears | Sampling or 100% | For surface-connected defects |
| Sectioning and metallography | Gas versus shrinkage porosity, wall thickness | Destructive sampling | Reference for root-cause work |
| X-ray radiography | Internal porosity, inclusions, core shift | Sampling, or 100% real-time | Grade to ASTM E505 levels 1 to 4 |
| Leak test | Interconnected porosity across a pressure wall | 100% on sealed parts | Air-under-water or decay at 0.5-2 bar |
| Helium mass spectrometry | Very small leak paths | 100% on critical seals | Down to 1 × 10⁻⁵ mbar·L/s |
| Machining exposure | Subsurface porosity at the cut face | Inherent to the operation | Where most porosity is found |
The lesson from that last row: run a machining trial on first-article castings early. Porosity that only appears when a face is cut open is the most common late surprise in a die casting program, and far cheaper to find on ten prototypes than a shipped lot. Our inspection equipment and documentation practice is on the quality page.
Vacuum resin impregnation seals interconnected leak paths and is an accepted production process on pressure-tight housings, but it does not restore mechanical properties or fix a structural void. Where sealing is central, A360 with its excellent pressure tightness beats sealing A380 after the fact.
Write acceptance criteria before you need them
Most defect disputes are definition disputes. A drawing specifying the following converts an argument into a measurement:
- A surface zone map marking cosmetic, functional and non-critical faces.
- Maximum pore size and total pore area per machined face, for example no pore over 0.8 mm and none within 1 mm of a sealing edge.
- X-ray grade and level, for example ASTM E505 level 2 in a defined region.
- Leak rate limit and test pressure, if the part seals.
- Allowable flash, parting line witness and ejector pin witness heights.
- Whether impregnation is permitted, and whether it must be reported.
Agree these at DFM rather than after first articles arrive. Every one is quotable, and a supplier who knows the standard will engineer to it.
Our engineers review geometry for every mechanism above before tooling is cut, as part of the DFM report accompanying each aluminum die casting quotation. If you have parts showing these symptoms now, send photographs, the location and the local wall thickness through contact for an assessment of whether it is a process problem we can tune out or a geometry problem.
