Cold flakes are large polycrystalline inclusions generated when a solidified skin forms on the inside wall of a cold-chamber shot sleeve, is sheared off by the advancing plunger, and injected into the die cavity as a non-bonded chunk [S2].
The defect is specific to cold-chamber high pressure die casting of alloys such as Al-Si and Mg-Al-RE (AE44), and it is mechanically distinct from externally solidified crystals (ESCs), which nucleate as free-floating equiaxed dendrites in the melt [S2][S3].
Mechanism: how a sleeve wall skin becomes a flake
Molten metal poured into the shot sleeve loses heat to the steel wall, which is typically held below 200 °C while the melt sits at 600–680 °C for aluminum HPDC, so a coherent solidified layer nucleates and grows on the bore surface within seconds of fill [S1][S3].
During the fast shot, the plunger shears that skin from the wall and ejects it; because die lubricant is normally present on the sleeve bore, the flake's flat side rarely metallurgically bonds to the surrounding matrix and persists as an unbonded interface in the casting [S2]. Studies on magnesium AE44 (Mg-4Al-4RE-0.2Mn, wt%) confirm that high pre-solidification fractions produce both ESCs and cold flakes, and that the flakes induce large pockets of positive macrosegregation in the final part [S2].
Cold flakes versus cold shuts versus ESCs
The three terms are routinely confused on the shop floor, but they originate at different points in the cycle and call for different countermeasures: cold flakes form on the cold-chamber sleeve wall before injection [S2][S7]; cold shuts form inside the die cavity when two melt fronts meet after partial solidification, producing fine, thin foliation cleavages on level surfaces or as subsurface cold lap inclusions revealed only by machining [S4]; and ESCs are single-crystal equiaxed dendrite fragments nucleated in the bulk melt, fragmented by the turbulent fast shot [S2].
A practical decision table: cold flake source = shot sleeve (remedy: hotter sleeve, hotter melt, less dwell); cold shut source = die cavity filling (remedy: faster intensification, better venting, corrected slow-to-fast changeover, higher die temperature) [S4][S6]. Both can be triggered by inadequate venting, insufficient die cooling, or wrong slow-shot to fast-shot changeover, which is why root-cause analysis must locate the defect origin before tuning setpoints [S6].
Mechanical penalty in cast parts
Quantitative work on Al-Si alloys shows that cold flakes reduce ductility and ultimate tensile strength (UTS) and increase the scatter band of those properties, while leaving yield stress largely unaffected, because the non-bonded flake interface acts as a premature crack initiator under tensile load [S2].
In cold-chamber HPDC of magnesium AE44, samples with a high fraction of pre-solidification containing both ESCs and cold flakes show a small decrease in yield stress (driven by the larger microstructural length scale of α-Mg) but a substantial drop in UTS and elongation, attributed to the non-bonded flake-matrix interface rather than to ESCs themselves [S2]. The implication is that foundries chasing tensile certification for structural Mg or Al HPDC parts cannot treat the sleeve as a passive container; it is a controlled solidification reactor whose thermal state directly drives the reject rate.
Process levers to suppress flake formation
The literature points to two broad control strategies: raise the bulk thermal budget so less skin forms on the bore, and shorten the dwell time so any skin that does form stays thin [S2]. Levers under the first strategy include higher melt superheat, higher fill fraction, higher sleeve and plunger-tip temperature, and a thicker biscuit that keeps the plunger tip away from the bore [S2].
Levers under the second strategy include reducing the fill-to-fast-shot delay, reducing the slow-shot stroke, and avoiding lubricant over-application, since heavy oil films both insulate unevenly and guarantee the unbonded interface that defines a cold flake [S2][S4]. Related defect-control practice, including venting and intensification timing for cold-shut avoidance, is summarized in industry defect-handling notes for cold-chamber HPDC cells [S6].
Numerical and experimental detection
Maeda and Nomura's 2006 numerical study on Al-alloy cold-chamber HPDC simulated the growth of the solidified layer on the sleeve wall and its entrainment into the cavity, framing cold flakes as a moving-boundary problem solvable by CAE coupling of thermal and fluid fields, an approach that remains the basis of modern sleeve-solidification modules [S3].
More recent work links that framework to alloy-specific microstructural outcomes: the Yu, Zhan and Gourlay 2024 study on AE44 used a grain-morphology map of α-Mg to attribute the change in eutectic Al11RE3 morphology directly to the cooling rate experienced inside the sleeve skin versus inside the die cavity, giving process engineers a way to estimate flake severity from as-cast metallography rather than from destructive tensile screening alone [S2]. For shop-floor use, the same principles inform how shot blasting machine surface-prep cycles are specified for cast coupons used in metallographic flake counting.
Selection criteria: when to treat the defect as a sleeve problem
A defect is most likely a cold flake rather than a cold shut when the inclusion is a flat-faced polycrystalline chunk found anywhere in the casting cross-section, when the affected area shows a non-bonded interface under microscopy, and when the flake's chemistry matches the bulk alloy with no oxide entrapment typical of a cavity-side cold lap [S2][S4].
Conversely, treat it as a cavity-side cold shut when the defect appears only at thick level surfaces, as thin foliation cleavages on pressure-die-cast skin, or as a subsurface inclusion with its own oxide skin revealed by machining [S4]. The differentiation matters because the corrective action diverges: sleeve-side remedies target melt and sleeve temperature plus dwell, while cavity-side remedies target venting, gate and overflow design, intensification pressure, and the slow-to-fast changeover point [S4][S6].
Foundry implementation checklist
Track the sleeve wall temperature at three axial positions, not just the thermocouple near the pour hole, because the flake-forming skin is most likely to nucleate at the mid-stroke region where dwell time is longest; cap melt-to-injection dwell at a value consistent with the alloy's solidification range, with AE44-class Mg and 3xx-series Al typically tolerating only a few seconds of contact before measurable skin growth [S2].
Audit lubricant volume per cycle, since a heavy film on the bore both insulates unevenly and locks in the non-bonded interface that defines the defect [S2]. For broader process context on how sleeve dwell interacts with upstream melt handling and downstream shot parameters, see this field note on aluminum can sheet supply tightens as recycled content targets diverge in 2026, which discusses the same melt-handling discipline in a different process chain. Routine metallographic cross-sections through the biscuit and overflow regions remain the cheapest in-house flake counter, and a rising count is a leading indicator that sleeve or melt setpoints have drifted before UTS data expose the same drift in QA [S2][S3].