Scaling Back Heat Treatment of Aluminum 319 Without Changing Material Properties

Robert Mackay & Glenn Byczynski

Recent global events have shown the importance of onshoring some of the current casting supply to mitigate production disruptions. Thus, further optimization of the current casting and heat treat processes will help metalcasting in North America be cost-competitive with offshore sourcing. The heat treat regime is one of the processes where there can be significant optimization with further research. This is particularly true with Cu-bearing Al-Si casting alloys such as the 319 type.

The 319-type aluminum alloy is one of the more widely used casting alloys for propulsion applications in the automotive industry, and it is often given a T7 heat treatment when the highest material properties and dimensional stability are needed.

Finding a way to scale back the solution treatment stage while achieving the same material properties and dimensional stability of a full solution treatment would lower energy cost, increase capacity without capital investment of the solution furnace, and reduce the CO2 footprint of the over-all heat treat process. 

In this study conducted by a major supplier of automotive components, the cast structure and its response/behavior to a T7 treatment were studied for the two alloys shown in Table 1. 

Experimental Procedure

For brevity, the overall experimental description for the casting and heat treatment of precision sand cast components is shown via flow diagrams in Figures 1a (Process Flow 1-without solution stage) and 1b (Process Flow 2-with the solution stage).

The metal composition range and reduced pressure test (RPT) threshold required for Alloy 2 are shown in Table 2. The Fe concentration, which has a maximum limit of 0.6 wt.% (based on most automotive applications using this alloy type), was kept between 0.48 wt.% to 0.52 wt.%. This is important as the Fe concentration in both Alloys 1 and 2 predominately forms the ß-Al5FeSi, α-Al8Fe2Si, α -Al12Fe3Si2, α -Al15(Fe,Cr,Mn) 3Si2, and δ-Al4FeSi6. As the Fe concentration increases from residual to 0.60 wt.%, the growth of the β-Al5FeSi and α -Al15(Fe,Cr,Mn)3Si2 phase increases exponentially in volume fraction. As a result, having a wide range of Fe concentrations in the castings produced means that the comparative mechanical response due to heat treat development may have an elevated degree of scattered mechanical properties. To avoid this additional impact, which would make this heat treat study potentially less conclusive, the Fe range was maintained within the 0.48 wt.% to 0.52 wt.% range mentioned.

As mentioned, the Sr target was maintained within the range of 140 to 170 ppm and added via an Al-10 wt.% Sr master alloy rod. The Sr additions are made for two reasons: (1) They change the Si morphology from platelet to fibrous, and (2) they suppress the size of gross shrinkage formations, which can result in pressure tightness issues.
A total of 390 castings were made using the Nemak-Cosworth process for this study (requiring 31,200 lbs. of liquid Alloy 2). 

Prior to casting riser removal, sand was removed via a thermal sand removal (TSR) furnace (485 ± 5C/905 ±9F), which has a travel time of 6.5 hours with exposure to the 485C (905F) target temperature for 2.5 hours (first solution stage), followed by a forced air quench at 85C/minute (153F/min.). After the riser and core removal process in Process Flow 2, the castings underwent either a 4.5-hour or 7.5-hour solution treatment also at 485 ± 5C (905 ± 9F)—second solution stage, followed by another quench at 90C/minute (162F/min.), which was eliminated in Process Flow 1.

Step 4—artificial aging—was conducted in a continuous flow furnace, not a batch furnace. Temperature gradients were kept within a range of 3.5C (6.3F) to ensure repeatability. Total throughput time in the artificial age furnace was 5.5 hours.

Once the prescribed heat treat regime was completed, both tensile and permanent mold test bars were extracted from the castings.

Table 3 contains a list of the full casting protocol outlining the specific details of the number of castings made for each group and the temperature used for both the TSR solution treat and overage. Trials 10 and 11 were conducted using Process Flow 2 but with a shortened solution stage (4.5 hours from 7.5 hours).

Tensile Testing Protocol

One tensile test sample was extracted from each casting for a total of 390 samples. The microstructure condition at the reduced gauge section of each bar is λ2 = 20.0 ± 0.30 μm and area fraction porosity > 0.10% max (sample check was n = 20). The Nemak Engineering Centre’s Metallurgical Lab was used to conduct all tensile tests on a UTS (SFM–60) tensile machine at a strain rate of 8 x 10-4 (mm/mm)/second. A strain gauge extensometer was attached to the tensile test samples for the measurement of elongation. After the test was conducted, the yield stress (0.2% offset), tensile stress (UTS), and plastic elongation (PE) were measured as defined in the ASTM B 557 specification. 

Dimensional Growth Stability Test Protocol

Dimensional stability testing is used widely to gauge heat-treated cast alloys’ susceptibility to grow after sustained elevated temperatures in the field (e.g., operating powertrain components). This test is critical, as engine blocks will see temperatures ranging from 170C to 190C (338-374F), and e-motor housings can see temperatures as high as 150C (302F).

The test samples are fabricated in the form of cylinders with dimensions 150mm long x 15mm in diameter and checked using a precision coordinate measuring machine (CMM) to confirm dimensions for baseline values. The test cylinders are then placed in a furnace at a temperature of 200C (392F) for a total of 100 hours. After thermal processing, the samples are measured on the same precision CMM to establish the change in length, which is then expressed as a percentage. Fifteen samples are used, with one dimensional growth sample coming from one casting. Multiple samples are not taken from a single casting because if there was a problem with chemistry, heat treatment, or variation in microstructure, it may skew the average value used to generate a permanent growth test result. The approach of using the one-dimensional growth sample from one casting makes it easier to identify outliers and to eliminate that sample from a pooled average with standard deviation calculation. The value cited in this work is the pooled average value (10 samples minimum allowing for up to five outliers) plus 3σ statistical determination.


Once all castings completed the process flows outlined in Figures 1a (Trials 1 to 6) and 1b (Trials 7 to 9) the resulting tensile mechanical property response was plotted as an average of 30 tensile results with ± σ in Figures 2-4. Dimensional stability test results are in Figure 5.

Figures 2-4 show specifically the YS, UTS, and PE plots for the nine different heat treat conditions studied. Best-fitted lines with R2 values were also plotted for the conditions that had secondary solution (Process Flow 2) and those that did not (Process Flow 1). The R2 values seem to range between 0.99 to 0.97 for YS and UTS, indicating a very strong trendline. The R2 values for the best-fitted lines in Figure 4 are slightly wider, ranging from 0.98 to 0.83. This wider range in R2 can be explained by the fact that PE values will have slightly more scatter than for YS or UTS.

The red and green arrows in Figures 2-4 represent an extrapolation between the best-fitted lines that delineate Process Flow 1 and 2. In Figure 4, the red arrow extends from Trial 5 (Process Flow 1) to Trial 9 (Process Flow 2). 
Another 30 castings were then manufactured and heat treated using the Process Flow 2, except now the solution stage was set at 4.5 hours but aged at 240C/464F (Trial 10). Referring to Figures 2, 3, and 4, the YS, UTS, and PE values of Trial 10 are sandwiched in between the original plots for Process Flow 1 and 2, confirming that vertical extrapolation should be possible.

To further the presumption that this extrapolation method is valid, the permanent growth testing from the two heat treat conditions (Trials 5 and 9) highlighted in Figures 2, 3, and 4 could be performed. Figure 5 shows the value of the permanent growth after 100 hours of soak time at 200C (392F) from cast sections from Process Flow 2 (Trial 9—solution treat and 240C (464F) artificial age) and repeated two times from castings that were used in Process Flow 1 (Trial 5 —no external solution treat and 255C (491F) artificial age) and found there was extremely close range in percentage growth plus 3σ (e.g. 0.033% to 0.036%).

The final test to verify this extrapolation process was to run another 30 castings with Process Flow 1 (Trial 5) and another 30 castings with Process Flow 2 (Trial 9), and then plot the mechanical property data along with a -3σ calculation. Figures 6a, b, and c show that a reasonable repeat of the observation seen in Figures 2, 3, and 4 occurred.


The process outlined in this work was meant to show it’s possible to significantly curtail the thermal processing required for Cu-bearing Al-Si alloys cast in sand. This will become increasingly important to support the North American metalcasting industries’ ability to compete with offshore sourcing in terms of price and delivery timelines, and potentially support customers’ requirement to implement a CO2 reduction program associated with the parts they receive from the casting supplier.

Removing the secondary solution stage of a TSR + T7 thermal profile can achieve the same mechanical properties and permanent growth test results by adjusting the age temperature. From the plots in Figures 3, 4, and 5 using the extrapolation method process, developers can help identify what the new age temperature could be if the total solution treatment is reduced from 10 hours to 2.5 hours. 

This may work for other Al-Si-Cu alloys but will most likely not be successful in other 3XX series alloys.

Based on the research of this and previous studies, the only microstructural factors that truly change for 319-type alloy during the solution stage is the elimination of dendrite coring, gradual increase in Cu content of those dendrites, a steep reduction in dislocation density, and a slight reduction of the Al2Cu volume fraction within the inter-dendritic regions. While previous research indicates this occurs specifically for Alloy 2, a relationship between magnitude of these changes with increasing solution treatment time exists. A significant argument can be made when comparing the impact of the test castings in this study having 2.5 hours (TSR) of total solution time, or 2.5 hours (TSR) + 7.5 hours (solution) = 10 hours total, will require a different age temperature to achieve the same results.

It is assumed the test castings with 10 total hours at solution will have more dissolved Cu and a lower dislocation density in the σ-Al dendrite cells than the test castings with the 2.5-hour solution stage within the TSR cycle. Coring is less of a factor because previous research showed this was eliminated within the first 30 minutes of actual solution time.

In the extrapolation conducted on Figures 2, 3 and 4, it was concluded that using only a 2.5-hour solution stage required an age of 255C (491F) in order to get the same dimensional growth and mechanical property result as the same castings solution treated for a total of 10 hours and then aged at 240C/464F. (Again, both age regimes are fixed at 5.5 hours in the same age furnace). Using a higher age temperature in Process Flow 1 to get the same results as in Process Flow 2 implies that the lower Cu content of the dendrites and possibly lower dislocation density means that if Process Flow 1 used 240C (464F) instead of 255C(491F), then YS would rise from 238 MPa to 275 MPa (Figure 2), but the elongation drops from 2% down to 1.5% (Figure 4). This implies that with Process Flow 1, the casting structure is possibly closer to peak age hardness, and to achieve the exact same overaged mechanical properties and dimensional stability, the 255C (491F) age temperature must be used to bring the casting structure closer to the same degree of the overaged condition. In castings that had a total solution time of 10 hours, there would be a compounding effect of the lower volume fraction of undissolved Al2Cu blocky and eutectic phases within inter-dendritic regions. However, this presumes that slight changes in the volume fraction of Al2Cu may be a minor factor compared to the impact of coring removal, Cu content of dendrites, and crystallographic defects.

Finally, the permanent growth samples from Process Flow 1 (Trial 5−2.5-hour solution with 255C/491F age) and 2 (Trial 9−10-hour solution with 240C/464F age) are essentially identical, suggesting that the potential of further aging and resulting permanent growth would be similar because they both were nearly at the same point with respect to peak hardness.

Warranty Impacts

An automotive customer should require any heat treat optimization to have no change to the component’s fit, form, and functionality. The warranty impacts due to a failed e-motor housing or an engine block are significant. These components cannot be replaced while leaving the rest of the propulsion system intact. Full propulsion replacements can range from $7,000 to $12,000 in warranty charges. The examples of where a heat treat regime can impact a warranty claim are as follows:

1. Changing the plastic elongation and YS means that burr formation susceptibility changes, and this can raise a concern about a burr displacing into a casting cavity and becoming wedged, ending up in an assembly. This becomes compounded when a casting is dual- or triple-sourced, which means that a monitoring process may need to be implemented (higher quality control cost) to account for some castings burring more than others due to a supplier’s optimized heat treat processing.

2. The YS (design parameter), UTS, and PE values are used in stress modeling to assess safety factors that gauge potential tensile fracture points in the casting, either under loading conditions or in a vehicular collision. As shown by the authors previously, changing PE, for example, means that these safety factor locations potentially become susceptible to fracture.

3. Having a differing value of permanent growth can prove disastrous when the casting in question is part of an automotive assembly. If an optimized heat treat used had a permanent growth value that was different from the original heat treat regime, this could lead to warranty issues. This would not be seen in assembly because the elevated service environments would not have been experienced until after the vehicle was sold to the customer and driven for several miles/kilometers.

The approach outlined in this work shows that developing mechanical property plots as shown in Figures 2, 3 and 4, optimizations can be employed with the aim of delivering the same casting quality but at lower cost, faster throughput, and reduction of CO2.