J O U R N A L O F M A T E R I A L S S C I E N C E 3 6 (2 0 0 1 ) 461 468

The sequence of precipitation in 339aluminum castings

R. K. MISHRA, G. W. SMITH, W. J. BAXTER, A. K. SACHDEV, V. FRANETOVICGeneral Motors Research and Development Center, Warren, Michigan, USA

The precipitation sequences in direct-quenched from the die (DQD) and solution-treated(SOL) 339 aluminum have been determined by a combination of differential scanningcalorimetry (DSC) and transmission electron microscopy (TEM). DSC scans for the alloy inboth conditions exhibit two distinct exothermic peaks, each associated with a uniqueprecipitate. The peak temperatures for precipitation in the DQD and SOL alloys differ byonly a few degrees. TEM of samples heated to the lower temperature peak shows that thefirst precipitate to form in the DQD alloy is S0 (Al2CuMg), whereas in the SOL alloy it is fl 0

(Mg2Si). The principal precipitate associated with the higher temperature peak in both DQDand SOL alloys is Si. The DSC peak temperature identifies the specific precipitate in 339 Al,but the peak area is not a reliable measure of precipitate density. Nano-indentation of thedendrites shows that the strength provided by the precipitates increases in the sequenceSi

T AB L E I Concentrations of primary alloying elements in 339aluminum castings

Total Concentration Dendrite Concentration(wt %) (wt %)

Sample Cu Mg Si Cu Mg Si

DQD 1.04 0.99 12 0.46 0.30 1SOL 1.01 0.78 12 0.97 0.48 1

to study the sequence of precipitation which would oc-cur during a T5 heat treatment, and (ii) solutionizedand quenched material (SOL), to follow the precipita-tion sequence during a T6 temper.

2. Sample preparationThe DQD samples were cut from a casting which hadbeen quenched in water after removal from the die andthen stored in a freezer at74 C prior to sample prepa-ration. The SOL samples were cut from an air-cooledcasting and subsequently solutionized. During fabrica-tion of samples for the DSC experiments, precautionswere taken to minimize exposure to ambient temper-atures. DQD samples were removed from the freezerfor three brief intervals: (i) initial cutting to rods ofsquare cross section; (ii) machining to cylinders with adiameter of 6 mm; and (iii) slicing discs 2 mm thick.After each step, the samples were returned to the freezerwhere they remained until just prior to the DSC exper-iments. Preparation of the SOL samples was simplerbecause all machining was carried out prior to solutiontreatment. The samples were solutionized at 510 Cfor 3 1/4 hours, quenched in water, then immediatelyplaced in the freezer where they remained until minutesbefore the DSC experiments.

The two castings had the same nominal composition(12Si/1Mg/1Cu weight percent), but the total Mg con-tent of the SOL sample was about 20% lower than thatof the DQD casting (Table I). However, more signifi-cant, from the viewpoint of this precipitation study, arethe concentrations of solutes retained in the dendrites.These were measured with an electron microprobe andare also listed in Table I.

3. CalorimetryA Perkin-Elmer DSC7 calorimeter was operated in itstemperature-scanned mode to measure the temperaturedependence of dQ=dt , the rate of heat absorption oremission by the sample. Such a plot has a baseline pro-portional to the specific heat of the sample with su-perimposed endothermic and exothermic peaks due todissolution and precipitation respectively [31, 32].

Typical DSC thermograms for DQD and SOL 339aluminum at a heating rate of 20 C/min are shownin Fig. 1. In each case several exotherms are visible.As discussed above, we can assign the lowest tem-perature peak(s) to Guinier-Preston (GP) zone forma-tion and the two at higher temperature to precipitationevents. The existence of the GP zone peaks indicatesthat the samples had not been greatly affected by theirbrief exposure to ambient temperatures during speci-men preparation. It is noteworthy that the GP peaks

Figure 1 DSC plots of dQ=dt versus temperature for DQD and SOL 339aluminum at a temperature scan rate of 20 C/min. The labels indicate theprecipitation peaks to which the DSC was scanned to prepare samplesfor TEM and hardness studies (see text). The two curves are shiftedvertically to avoid overlapping.

in the two thermograms differ somewhat in appear-ance, presumably reflecting the different thermal histo-ries (SOL vs. DQD). These peaks are followed by anendotherm which we attribute to dissolution of the GPzones prior to the exothermic precipitation processes.It is the latter which are examined in detail in the nextsection.

3. Transmission electron microscopySpecimens for TEM were prepared in the calorime-ter by heating (at 20 C/min) to either the first or sec-ond precipitation peak, after which they were imme-diately cooled and stored in the freezer. (Zhen et al.[29] have used a similar technique.) Four specimenswere prepared in this manner: two each for the DQDand SOL castings. The two peaks for the DQD alloyare labeled in Fig. 1 as DQD-I and DQD-II; those forthe SOL alloy as SOL-I and SOL-II. These specimenswere thinned by mechanical polishing, followed by ionmilling. They were then examined in a Philips EM430scanning transmission electron microscope operatingat 300 kV and fitted with a Noran X-ray detector. Theprecipitates were identified by selected area electrondiffraction, and X-ray microanalysis for multiple spec-imen orientations.

3.1. Precipitates in DQD 339 aluminum3.1.1. Peak DQD-IThis sample contains primarily thin rod-shaped precip-itates about 100 nm long with an aspect ratio larger than10, oriented along or close to the h100i axes of the Alfcc lattice. When viewed along one of the h100i axesof the matrix, three variants of the precipitates are vis-ible in the micrograph of Fig. 2. The precipitates arecoherent with the matrix and have a diffraction pattern(Fig. 2b) characteristic of the ternary S0 (Al2CuMg)phase [40]. Since no other precipitate phase is presentin this sample, it is clear that peak DQD-I correspondsto the formation of the S0 phase.

462

Figure 2 (a) Bright field transmission electron micrograph and (b) corresponding selected area diffraction pattern ([100] zone axis) showing rodshaped S0 precipitates in the DQD sample heated to peak DQD-I in Fig. 1 above. The rods parallel to the [100] zone axis appear as dots in the end-onview while the other variants are parallel to the [001] and [010] directions of the Al matrix.

3.1.2. Peak DQD-IIThe microstructure of this sample (Fig. 3) contains twodifferent phases: (i) rod-shaped S0 phases as discussedabove, and (ii) numerous spherical precipitates of Si,as confirmed by selected area diffraction (SAD) andx-ray microanalysis. Thus the second DSC peak is at-tributed to the precipitation of the Si particles. (Detailedanalysis of Si precipitates in 339 Al can be found in ref-erence 41.) The density of the S0 precipitates in Fig. 3is about one third of that found in Fig. 2, indicating thata large fraction of the S0 precipitates have re-dissolvedas the sample is heated from the DQD-I temperature tothe DQD-II temperature. Thus the second DSC peak isattributed to formation of the Si phase.

3.2. Precipitates in solutionized339 aluminum

3.2.1. Peak SOL-IThe microstructure of this sample (Fig. 4) consists ofnumerous needle-shaped precipitates dispersed in the

fcc Al matrix. These precipitates are50 nm long, havean aspect ratio >10, and lie along the [100] axes. Thedimensions of the precipitates and their diffraction pat-tern confirm that the needles arefl 0 (Mg2Si) precipitates[42] and not fl 00 or fl precipitates. Since no other phaseis present in this sample, the SOL-I peak is attributedto the precipitation of the fl 0 phase.

3.2.2. Peak SOL-IIThis sample contains only Si precipitates dispersed inthe Al matrix [41] (Fig. 5). There is now no evidence ofany Mg2Si phase, indicating that all the Mg2Si precipi-tates, which had formed during the temperature scanthrough Peak SOL-I, have re-dissolved. The size of theSi precipitates ranges from 50 to 60 nm.

4. Dendrite hardnessSome specimens were polished and the hardnessesof the dendrites were measured with a Nanoindenter.A three-sided pyramidal (Berkovitz) diamond was

463

Figure 3 Bright field image for the DQD sample heated to peak DQD-II in Fig. 1. Besides the S0 precipitates as in Fig. 2, there are small round Siprecipitates dispersed in the sample. The corresponding SAD and dark field images of Si are shown in (b).

applied with a load of 7.5 mN (0.75 g) to ten differ-ent dendrites in each specimen. The indentations weretypically 0.5 m deep and 4 m wide, positionedin the central region of each dendrite without interfer-ence from silicon particles or intermetallics. The ther-mal histories of these specimens (prior to polishing) arelisted in Table II together with their measured dendritehardnesses. The as-quenched SOL and DQD specimenswere measured after 10 hours exposure to room tem-perature. (This exposure may have resulted in GP-zone

T AB L E I I Effect of precipitation on dendrite hardness

Thermal History Hardness (GPa)

As DQD 0.91 0.05DQD and Heated to Peak DQD-I 1.13 0.09DQD and Heated to Peak DQD-II 0.95 0.10DQD and Heated to 350 C 0.81 0.05As SOL 0.90 0.05SOL and Heated to Peak SOL-I 1.24 0.07SOL and Heated to Peak SOL-II 1.10 0.06SOL and Heated to 350 C 0.88 0.02

formation, but was unlikely to have produced signifi-cant precipitation.) Four specimens (two each of SOLand DQD) were heated in the calorimeter to either thefirst or second precipitation peak, and thus had the samethermal histories as the TEM specimens. In addition,dendrite hardnesses were measured for SOL and DQDspecimens scanned in the DSC to 350 C (as in Fig. 1).

Initially the SOL and DQD specimens have the samehardness, but after precipitation has occurred, the SOLsample is harder than the DQD. This difference is main-tained even after exposure to 350 C. For both startingconditions, the first precipitation peak corresponds tothe highest hardness. Thus, heating to the second peaksubstantially decreases hardness, which is further re-duced after the brief (1 min) excursion to 350 C.

5. Discussion5.1. The precipitatesThe combined DSC/TEM studies have clearly identi-fied the precipitation sequence in 339 aluminum forboth the DQD and SOL conditions. In each case

464

Figure 4 Bright field image of a SOL sample heated to the first DSC peak (SOL-I in Fig. 1) showing small fl 0 precipitates, The dots in the image arefl 0 precipitates viewed end on. The sample has been oriented to weaken the diffraction contrast and dislocations.

there are two pronounced precipitation peaks in theDSC curves; the peak temperatures (at a scan rateof 20 C/min) and the precipitates are summarized inTable III. In the DQD alloy the lower temperature DSCpeak at 247 C corresponds to the formation of theternary S0 (Al2CuMg) precipitate. In the solutionized

TAB L E I I I Precipitation parameters for DQD and SOL 339aluminum

Peak Peak T (C) Precipitate 1Q (J/g)

DQD-I 247 1.0 S0 (Al2CuMg) 3.67DQD-II 295 1.3 Si 3.55SOL-I 240 2.5 fl 0 (Mg2Si) 3.52SOL-II 289 1.0 Si 2.04

Peak temperatures, T , measured at a scan rate of 20 C/min, are theaverages for several samples.1Q values were derived from 2-gaussianapproximations to DSC precipitation peaks (see Figs 4 and 5).

alloy the peak at 240 C corresponds to the formationof the binary fl 0 precipitate. Apparently the first precip-itate to form is controlled by Mg-Cu clusters [43, 44]in the DQD alloy, but by Mg-Si clusters in the SOLalloy. Heating to the second peak dissolves some of theS0 phase in the DQD alloy but all of the fl 0 phase in theSOL alloy. At this temperature (290 C) Si is precipi-tated in both the SOL and DQD alloys.

In both the DQD and SOL alloys, the higher hard-ness is provided by the precipitate formed at the lowertemperature peak, namely S0 and fl 0 respectively. Forthe solutionized alloy the reason is very clear from theTEM micrographs: the fl 0 precipitates in Fig. 4 are co-herent and more closely spaced than the incoherent Siparticles in Fig. 5. Similarly in the case of the DQDalloy, the coherent S0 precipitates in Fig. 2 are more ef-fective in blocking dislocation motion than the mixtureof S0 and incoherent Si precipitates in Fig. 3.

465

Figure 5 Bright field micrograph and selected area diffraction showing the dispersion of Si particles in the SOL sample heated to the second DSCpeak (SOL-II in Fig. 1). There are no other phases present in this sample.

5.2. DSC peaksThe above detailed knowledge of the precipitation pro-cesses can only be acquired by the time consuming pro-cedures of TEM, which are not appropriate for routineevaluation of castings. Thus an important question iswhether the relatively simple and rapid measurementof the DSC peak structure can provide a unique sig-nature to identify each precipitate and determine itsamount. In this regard, the appropriate parameters arethe peak temperature and the area of each peak (i.e.,the amount of heat released 1Q). To estimate the lat-ter quantity, the DSC peaks are represented as the sumof two gaussians, as indicated by the dotted curves inFigs 6 and 7. This approximation matches the peaks forthe SOL sample (Fig. 7), but it appears that the highertemperature peak of the DQD sample (Fig. 6) over-laps a third smaller peak at 340 C. The 1Q valuesderived from the gaussian fits are listed in Table III.

A crucial test is to compare the SOL samples fl 0peak at 240 C with that for S0 in the DQD sample at247 C. Although the peak temperature difference is

Figure 6 DSC plot of dQ=dt versus temperature for DQD 339 aluminumat a temperature scan rate of 20 C/min. The dotted curve is the sum oftwo gaussian curves fitted to the data in order to estimate the amount ofheat release associated with each peak (see text).

466

Figure 7 DSC plot of dQ=dt versus temperature for SOL 339 aluminumat a temperature scan rate of 20 C/min. The dotted curve is a 2-gaussianfit to the data.

only 7 C, it is still somewhat larger than the specimen-to-specimen variability (Table III). Therefore, we con-clude that the formation of the fl 0, S0, and Si precipitatesin 339 aluminum alloy can be identified on the basis ofthe peak temperature.

In principle the value of 1Q for each peak will de-pend upon the specific precipitate and its concentra-tion, which in turn would be determined by the amountof solute available in the aluminum dendrites and thespecific precipitate phase. However, this is not alwaysborne out in practice. For example, the higher tempera-ture Si peak in the SOL specimens substantially smallerthan that in the DQD sample despite the fact that theSi particle concentrations are essentially the same (seeFigs 3 and 5). This is thought to be due to the fact thatin the SOL alloy the fl 0 dissolves completely as Si pre-cipitates, resulting in an endotherm superimposed onthe Si precipitation exotherm. The discrepancy is notas sever for the DQD sample, where the S0 is more sta-ble and much is retained during Si formation (Fig. 3).Thus we conclude that in this case the value of 1Q isnot an appropriate measure of precipitate density.

6. ConclusionsBased on the results presented here, the following con-clusions may be drawn:

1. The sequence of precipitation in 339 aluminumdepends upon the thermal history of the alloy.

2. In the solutionized alloy, fl 0 forms first, then dis-solves and is replaced by Si.

3. In a casting directly quenched from the die, S0forms first, then partially dissolves as Si forms.

4. The DSC peak temperature identifies the specificprecipitate for 339 Al.

5. The peak area (i.e.,1Q) is not a reliable measureof precipitate density.

AcknowledgementsThe authors thank Coleman Jones for providing the339 Al castings, and Spud Willett, Dick Hall, andDusanka Radovic for technical assistance.

References1. K . H I R A N O and H. I W A S A K I , Trans. Jap. Inst. Met. 5 (1964)

162.2. J . M. P A P A Z I A N , Metall. Trans. A12 (1981) 269.3. M. V A N R O O Y A N , J . A . S I N T E M A A R T E N S D I J K and

E . J . M I T T E M E I J E R , ibid. A19 (1988) 2433.4. C . A N T O N I O N E , F . M A R I N O and G. R I O N T I O N O , Mater.

Chem. Phys. 20 (1988) 13.5. S . A B I S and G. D O N Z E L L I , J. Mater. Sci. Letters 7 (1988) 51.6. J . M. P A P A Z I A N , Metall. Trans. A19 (1988) 2945.7. A . K . J E N A , A. K. G U P T A and M. C. C H A T U R V E D I ,

Acta. Metall. 37 (1989) 885.8. M. C . C H A T U R V E D I , A . K . G U P T A and A. K. J E N A ,

Mater. Sci. Eng. A110 (1989) 187.9. M. V A N R O O Y A N and E. J . M I T T E M E I J E R , Metall. Trans.

A20 (1989) 1207.10. I . D U T T A and D. L . B O U R E L L , Mater. Sci. Eng. A112 (1989)

67.11. C . B A D I N I , F . M A R I N O and A. T O M A S I , Mater. Chem. Phys.

25 (1990) 57.12. A . -M. Z A H R A and C. Y. Z A H R A , J. Thermal Anal. 36 (1990)

1465.13. C . B A D I N I , F . M A R I N O and A. T O M A S I , Mater. Sci. Eng.

A136 (1991) 99.14. Idem., J. Mater. Sci. 26 (1991) 6279.15. P . A P P E N D I N O , C . B A D I N I , F . M A R I N O and A. T O M A S I ,

Mater. Sci. Eng. A135 (1991) 275.16. M. J . S T A R I N K and P . V A N M O U R I N K , Metall. Trans A20

(1991) 665.17. M. J . S T A R I N K , V. J O O R I S and P . V A N M O U R I N K , in

Metal Matrix Composites - Processing, Microstructure and Prop-erties, edited by N. Hansen et al. (Risoj National Lab, Roskilde,Denmark, 1991) p. 675.

18. H . -L . L E E , W.-H. L U and S . L . - I . C H A N , Scripta Metall.Mater. 25 (1991) 2165.

19. I . D U T T A , S . M. A L L E N and J . L H A F L E Y , Metall. Trans.A22 (1991) 2553.

20. I . D U T T A and S . M. A L L E N , J. Mater. Sci. Letters 10 (1991)323.

21. M. J . S T A R I N K and P . V A N M O U R I N K , Mater. Sci. Eng.A156 (1992) 183.

22. T . S . K I M , T . H . K I M , K. H. O H and H. I . L E E , J. Mater.Sci. 27 (1992) 2599.

23. I . D U T T A , C . P . H A R P E R and G. D U T T A , Metall. Mater.Trans. A25 (1994) 1591.

24. C . G A R C I A-C O R D O V I L L A , E . L O U I S , J . N A R C I S O andA. P A M I E S , Mater. Sci. Eng. A189 (1994) 219.

25. M. P . T H O M A S and J . E . K I N G , J. Mater. Sci. 29 (1994) 5272.26. P . R A T C H E V , B . V E R L I N D E N and P . V A N H O U T E , Scripta

Metall. Mater. 30 (1994) 599.27. C . B A D I N I , F . M A R I N O and E . V E R N E , Mater. Sci. Eng.

A191 (1995) 185.28. I . N . A . O G U O C H A and S . Y A N N A C O P O U L O S , ibid. A231

(1997) 25.29. L . Z H E N , W. D. F E I , S . B . K A N G and H. W. K I M , J. Mater.

Sci. 32 (1997) 1895.30. M. J . S T A R I N K and A.-M. Z A H R A , Phil. Mag. 76 (1997)

701.31. G . W. S M I T H , Thermochimica Acta 313 (1998) 27.32. Idem., ibid. 317 (1998) 7.33. H . E . K I S S I N G E R , Anal. Chem. 29 (1957) 1702.34. T . O Z A W A , J. Thermal Anal. 2 (1970) 301.35. R . L . T H A K U R , in Advances in Nucleation and Crystallization

of Glasses, edited by L. L. Hench and S. W. Freiman (Amer. Ceram.Soc., Columbus, OH, 1972) p. 166.

36. J . A . A U G I S and J . E . B E N N E T T , J. Therm. Anal. 13 (1978)283.

37. E . J . M I T T E M E I J E R , L . C H E N G , P . J . V A N D E RS C H A A F , C . M. B R A K M A N and B. M. K O R E V A A R ,Metall. Trans. A19 (1988) 925.

38. H . Y I N N O N and D. R . U H L M A N N , J. Non-Crystalline Solids54 (1983) 253.

39. J . S E S T A K , in Comprehensive Analytical Chemistry, Vol. XII,Part D, edited by G. Svehla (Elsevier, Amsterdam, 1984) Ch. 9,p. 212.

467

40. A . K . G U P T A , P . G A U N T and M. C. C H A T U R V E D I , Phil.Mag. 55 (1987) 375.

41. V . R A D M I L O V I C and R. K. M I S H R A , Scripta Met, to bepublished.

42. P . H I R S C H , A. H O W I E , R . B . N I C H O L S O N , D. W.P A S H L E Y and M. J . W H E L A N , in Electron Microscopyof Thin Crystals (Krieger Publishing, Huntington, NY, 1977)p. 151.

43. G . T H O M A S and M. J . W H E L A N , Phil. Mag. 4 (1959) 511.44. V . R A D M I L O V I C , G . T H O M A S , G. J . S H I F L E T and E . A.

S T A R K E , J R., Scripta Metall 23 (1989) 1141.

Received 31 August 1999and accepted 22 February 2000

468