Relating Quench Sensitivity to Microstructure in 6000 Series Aluminium Alloys*1

Katharina Strobel1, Mark A. Easton1, Lisa Sweet1, Malcolm J. Couper2;*2 and Jian-Feng Nie1

1CAST Co-operative Research Centre, Department of Materials Engineering, Monash University, Victoria 3800, Australia2Rio Tinto Alcan, Pacific Technology Centre, Brisbane, Queensland 4000, Australia

In high strength 6000 series alloys dispersoids that form during heating to the homogenization temperature are used to improve fracturetoughness and suppress grain growth during the extrusion process. However, these dispersoids can act as heterogeneous nucleation sites for non-hardening Mg-Si precipitates if quenching after extrusion is delayed. This leads to a reduced level of Mg and Si in solid solution and hence lowerachievable strength and hardness. This phenomenon is called quench sensitivity. In this study, the hardening response of several 6000 seriesaluminium alloys is related to microstructural features, especially dispersoid density. Therefore, the alloys were quenched at varying rates afterextrusion and age hardened to peak strength. Quench sensitivity is related to dispersoid density as well as to the enthalpy related to theprecipitation of Mg-Si phase measured by DSC. The results suggest that in alloys containing dispersoids, quench sensitivity is primarilydetermined by the number density of dispersoids. However, effects associated with elements in solid solution cannot be ruled out. TEMinvestigations suggest that not only the general reduction of Mg and Si is responsible for the reduced mechanical properties, but that aninhomogeneous distribution of hardening precipitates might be another determining factor. [doi:10.2320/matertrans.L-MZ201111]

(Received October 5, 2010; Accepted February 6, 2011; Published May 1, 2011)

Keywords: 6000 series alloys, quench sensitivity, dispersoid

1. Introduction

Quench sensitivity, the loss of properties due to reducedquench rates after extrusion, is a particular problem in highstrength 6000 series aluminium alloys. While high quenchrates are achievable, they are an added cost and can lead todistortion of extrusions, especially of more complex profiles.Transition metals, especially manganese and chromium, areadded to improve fracture toughness and to inhibit graingrowth during recrystallization. These elements, in combi-nation with iron, form dispersoid phases during heating to thehomogenization temperature.1) If cooling from the extrusiontemperature is delayed, these dispersoid phases act asnucleation sites for coarse, non-hardening Mg-Si precipi-tates.2) This reduces the degree of supersaturation of Mg andSi and hence reduces the alloy’s response to age hardeningtreatments. Consequently, mechanical properties dependstrongly on the quench rate.

Therefore there are two key alloy characteristics thatdetermine the quench sensitivity: the presence of nucleationsites for non-hardening precipitation and the degree ofsupersaturation. However, it is really the presence ofheterogeneous nucleation sites that are problematic as thesupersaturation is also required to drive the precipitationprocess during ageing.3) The suitability of dispersoids asnucleation sites depends on their interface with the matrixwhich is determined by their composition, crystal structureand size. In chromium and manganese containing 6000 seriesalloys dispersoids are isomorphous to Al12Fe3Si andAl15Mn3Si2 with Cr, Mn and Fe substituting for Fe and Mnin these phases respectively.2,4,5) For relatively high chromi-um contents (> 0:3 mass%) the �0-Al13Cr4Si4 phase has beenreported.2) The dispersoid phases identified in the literature

are summarized in Table 1. Due to the large lattice parameter(a ¼ 1:26 nm4)) of the �-phases and dispersoids in the sizerange between 20–500 nm, the interface between the dis-persoids and the matrix is incoherent and hence can provide ahigh energy interface for nucleation of other phases.

The number density of the dispersoids is known to increasethe quench sensitivity of these alloys.2) The number densityis determined by the amount of transition metals and thehomogenization treatment, where short homogenizationtimes at comparably low temperatures lead to high densitiesof fine dispersoids.3) The same dispersoid density is achievedat lower chromium contents compared to manganese, asmanganese is more soluble in both the matrix and primaryintermetallic phases.2)

As 6000 series alloys gain their strength from fine �00-MgSiprecipitates, increased amounts of Mg and Si in the alloy leadto higher strength.6) However, increased supersaturation inalloys containing high amounts of Mg and Si in solid solutionalso provides a greater driving force for nucleation of non-strengthening precipitates during the quench, thus leadingto more quench sensitive alloys. This means that higherstrength alloys will always tend to be more quench sensitive.

While there is understanding of the general relationshipsbetween dispersoid formation and quench sensitivity, a studyof the relationship between dispersoid characteristics and therelative quench sensitivity of various alloys is not available inthe open literature. A quantification of the reasons for thesevariations is likely to provide a path for the development ofless quench sensitive alloys and is consequently the topic ofthis research.

2. Experimental

Four different alloys with compositions in the range ofAA6060, AA6005A, AA6061 and AA6082 respectivelywere investigated in this study (Table 2). The alloys chosenrepresent common 6000 series extrusion alloys with AA6060being considered a low strength alloy and AA6005A,

*1The Paper Contains Partial Overlap with the ICAA12 Proceedings by

USB under the Permission of the Editorial Committee.*2Present address: ARC Centre of Excellence for Design of Light Metals,

Monash University, Victoria 3800, Australia

Materials Transactions, Vol. 52, No. 5 (2011) pp. 914 to 919Special Issue on Aluminium Alloys 2010#2011 The Japan Institute of Light Metals

AA6061 and AA6082 being considered medium to highstrength alloys.

The alloys were direct chill cast into extrusion billets withstandard casting temperatures and cooling rates for thespecific alloys. Homogenization was conducted at 570�C for2 h followed by an air quench. The alloys were extruded,water quenched and straightened by 0.5% strain. In order todetermine quench sensitivity a modified Jominy test, which isdescribed in detail elsewhere,7) was conducted to achieve awide range of cooling rates. Thermocouples were embeddedinto a 600 mm long bar at particular distances to measure thecooling rates. These bars were solution treated (AA6060 at520�C for one hour and AA6005A, AA6061 and AA6082 at550�C for one hour). Following this, the bars were fixed on aholder at one end and 50 mm of the other end was swung intoa tank containing �150 litres of water at room temperature.This method provided cooling rates from 50�C/min for thefixed end to over 2000�C/min for the water quenched end.In a final step the alloys were heat treated to T6 conditionby ageing for 6 h at 190�C for AA6060 and 8 h at 175�C forAA6005A, AA6061 and AA6082. Rockwell F was used toassess the mechanical properties of each of the alloys.

Scanning electron microscopy (SEM) investigationswere undertaken on polished samples using JEOL 7001FFEGSEM in the back scattered electron (BSE) mode. Todetermine the dispersoid size and density high contrast BSEimages were analyzed using Adobe Photoshop ExtendedCS4 Win EULA software. The analysis of SEM images wasfound to give satisfactory results and hence chosen due tosavings in time and effort. Transmission electron microscopy(TEM) was carried out to investigate the composition andstructure of dispersoids in selected alloys using a PhilipsCM20 operated at 200 kV equipped with an EDS detector.

Differential scanning calorimetry (DSC) is widely used tostudy phase transformations in metals. A number of studiese.g.8–11) used DSC qualitatively to study various precipitationreactions. Milkereit et al.12) published a detailed report onusing DSC to measure the precipitation of Mg-Si-phasesduring cooling after solution treatment. DSC was performedin order to compare the precipitation of non-hardening phasesin the four different alloys at a cooling rate, which represents

a forced air quench in an industrial process. It was conductedusing a Perkin Elmer DSC7 in a nitrogen atmosphere.Disc shaped samples, 5 mm in diameter and 2 mm high, wereheated at 50�C/min to 550�C, held for 60 min in order tocompletely dissolve previously precipitated Mg-Si phasesand cooled at a rate of 50�C/min. All runs were corrected bysubtracting the baseline of the DSC.

3. Results

The hardness measured along the extruded and peak agedbars is related to the quench rate (Fig. 1).7) In the fastquenched state the difference between the soft alloy AA6060and the medium to high strength alloys is significant, whereasin the slow cooled state this difference is reduced up to thepoint that AA6082 shows lower hardness than the alloyAA6060. The hardness of AA6060 is almost independent ofthe cooling rate while the hardness of the other investigatedalloys, especially AA6082, is highly dependent on thecooling rate.

In order to be able to compare the quench sensitivitywith other alloy features, quench sensitivity is defined as thedifference in the measured hardness between water quenchedand air cooled samples. The dipsersoid density was deter-mined from SEM micrographs (Fig. 2). AA6082 containedthe greatest number of dispersoids, followed by AA6061 andthen AA6005. The average length of the dipsersoids for thethree alloys was approximately 150 nm. Dispersoids were notobserved in AA6060.

The dispersoid-containing alloys AA6005A, AA6061 andAA6082 show a linear relationship between particle density

Table 1 Composition and Crystal structure Mn- and Cr-containing dispersoids in Al-Mg-Si alloys.

Alloy Type Proposed Dispersoid Phases Crystal Structure Source

AlþMgþ Siþ FeþMn �-Al12(MnFe)3Si when Mn:Fe <1:6 at% bcc, a ¼ 1:26 nm 1, 4)

�-Al15(MnFe)3Si2 sc, a ¼ 1:26 nm

AlþMgþ Siþ FeþMnþ Cr �-Al(MnCrFe)Si cubic, a ¼ 1:26 nm 1)

AlþMgþ Siþ Feþ Cr �-Al(CrFe)Si bcc/sc, a ¼ 1:26 nm 5)

�0-AlCrSi for Cr > 0:3 mass% fcc, a ¼ 1:09 nm

Table 2 Composition (in mass %) of the alloys investigated.

Si Fe Cu Mg Mn Cr

AA6060 0.43 0.18 — 0.45 0.04 —

AA6005A 0.73 0.18 0.12 0.53 0.10 0.06

AA6061 0.62 0.21 0.25 0.87 0.13 0.10

AA6082 1.05 0.15 0.05 0.75 0.60 0.08

Fig. 1 Hardness versus quench rate for the investigated 6000 series

alloys.7)

Relating Quench Sensitivity to Microstructure in 6000 Series Aluminium Alloys 915

and quench sensitivity (Fig. 3), with the extrapolationthrough the origin suggesting that dispersoid density is astrongly dominant factor. Despite dispersoids not beingobserved in AA6060 some quench sensitivity was stillobserved in this alloy and consequently it did not lie on thestraight line.

Figure 4(a) shows a transmission electron micrograph ofa sample of alloy AA6061 after cooling in air and agehardening, highlighting a coarse, elongated non-hardeningprecipitate nucleated on a dispersoid. The strengtheningprecipitates which formed during age hardening are visiblein the micrograph as well. However, in the proximity of thenon-hardening phase there are no hardening precipitates.This precipitate free zone was depleted of Mg and Si as aresult of the formation of the non-hardening phase duringthe slow cooling from the extrusion temperature.

The EDS spectra indicate the elements present in thedispersoid Fig. 4(b) and the non-hardening phase Fig. 4(b).

The dispersoid contains Al, Fe, Mn, Cr and Si whichindicates that it is a cubic �-Al(FeMnCr)Si phase.2) TheEDS spectrum for the elongated, non-hardening phase showsAl, Mg and Si and hence indicates a metastable Mg-Si phase,most likely �0-Mg-Si.

The precipitate free zones could also be observed inAA6082 and AA6005A. Dispersoids in AA6082 were foundto be either the body centered cubic (bcc) �-Al12(FeMn)3Sior the simple cubic (sc) �-Al15(FeMn)3Si2.13)

The amount of phase transformation during cooling fromsolution treatment at 550�C was determined by differentialscanning calorimetry (Fig. 5). The cooling rate was similar tothat in a forced air quench, which is the lowest quench rate inFig. 1. No peak was observed in AA6060. However, alloysAA6005A, AA6061 and AA6082 had significant peaksindicating precipitation of the non-hardening Mg-Si phases.The size of the peaks, reflected in the enthalpy of precip-itation (Table 3), indicates more precipitation occurs inAA6082 than in AA6061 and AA6005A. While the peak ofAA6005A is uniform, both AA6061 and AA6082 show ashoulder during the early stages of precipitation. In Fig. 6the enthalpy measured from the DSC is plotted againstquench sensitivity. The relationship between these twoparameters follows a similar trend as the one betweenquench sensitivity and dispersoid density. However, AA6061seems to diverge from this trend, showing a higher enthalpyof precipitation relative to AA6082 and AA6005A for itsquench sensitivity.

(b)(a)

(c)(d)

Fig. 2 High contrast back scattered electron images (15 kV, 10 mm working distance) showing different dispersoid densities in (a)

AA6005A, (b) AA6061 and (c) AA6082. The measured dispersoid density of the three transition metal containing alloys is shown in (d).

Fig. 3 Relationship between dispersoid density and quench sensitivity in

selected 6000 series alloys (homogenisation 570�C/2 h).

Table 3 Quench sensitivity and enthalpy of precipitation of the inves-

tigated alloys.

AlloyQuench Sensitivity

[HRF]

Enthalpy of

Precipitation [J/g]

Main Peak Onset

Temperature [�C]

AA6060 6.7 0 —

AA6005A 12.2 �2:02� 0:22 378� 2

AA6061 17.8 �6:91� 0:30 391� 1

AA6082 32.7 �9:50� 0:41 394� 2

916 K. Strobel, M. A. Easton, L. Sweet, M. J. Couper and J.-F. Nie

4. Discussion

This work indicates that the dispersoid density is a keyfactor that affects the quench sensitivity of an alloy and thatthe relationship is directly proportional (Fig. 3); at least in

the 6000 series processed under equivalent conditions. It isnoted that the exception is AA6060 which does show somequench sensitivity despite containing virtually no disper-soids. While the data shows such a strong relationship, it isimportant to consider whether this should be the case or not,and whether this can be expected to be universally the case.TEM investigations on an air cooled and age hardenedsample of AA6061 show a precipitate free zone (PFZ)around the coarse precipitate which nucleated on thedispersoid. The precipitation of these non-hardening precip-itates leads not only to a reduction of the total amount of Mgand Si in solid solution but causes zones free of Mg and Si.There is no precipitation of fine, hardening precipitatesoccurring in this PFZ and it is likely that this is the reasonfor the reduction in properties of the alloy. Hence to explainquench sensitivity, the nucleation and growth of the non-hardening precipitates on the dispersoids needs to beconsidered.

The three dispersoid containing alloys AA6005A, AA6061and AA6082 all contain dispersoids that are very similar intheir size and composition (Table 1). While the size is

Fig. 4 Transmission electron micrographs of air cooled and artificially

aged AA6061 showing an elongated, non-hardening phase on a dispersoid

(a) EDS spectrum of (b) the dispersoid and (c) the non-hardening phase.

Fig. 5 DSC traces, taken at a scan rate of 50�C min�1 during cooling after

an imitated solution treatment; (a) AA6060, (b) AA6005A, (c) AA6061

and (d) AA6082.

Fig. 6 Quench sensitivity in relation to the energy measured for the

precipitation of non-hardening phases.

Relating Quench Sensitivity to Microstructure in 6000 Series Aluminium Alloys 917

determined by the homogenization treatment, the composi-tion and crystal structure are determined by the compositionof the alloy, in particular the transition metal content. Fortransition metal contents typical in the alloys investigated(Mn < 0:6 mass%, Cr < 0:3 mass%) all dispersoid phasesare based on the primary intermetallic phases Al12Fe3Si orAl15Mn3Si2. This means that all dispersoids have a cubicunit cell with the cell parameter a ¼ 1:26 nm, so that theinterfacial energy is expected to be very similar. Thepotential of each single dispersoid to act as a nucleationsite is hence the same, particularly since the size of thedispersoids are very similar. Consequently it would beassumed that the nucleation kinetics will be the same in eachof the alloys.

The DSC results (Fig. 5) provide information on thegrowth kinetics and more particularly the degree of thetransformation. There is a strong relationship between theenthalpy measured and the quench sensitivity of the alloy. Itcould be expected that since the dispersoids have more or lessthe same crystal structure and the non-hardening precipitatesare Mg-Si precipitates and the cooling regime is the same,that at each of the dispersoids similar size non-hardeningprecipitates would form leading to a direct relationshipbetween the dispersoid density, the measured enthalpy andthe quench sensitivity of the alloy. This is more or less whatis found, although it appears that the AA6061 alloy has alower measured quench sensitivity than would be expectedfor the measured enthalpy. Hence there appears to be somenuances to the precipitation process that require furtherinvestigation.

Interestingly, the exothermal DSC peaks for AA6082 andAA6061 do not have a Gaussian shape, but show a shoulderat the higher temperature end and the shape of the shoulderdiffers between the alloys. This can be explained byMilkereit’s et al. findings,14) that show two different peaksfor slow cooling rates (0.1–2 K/min). These two peaks wereattributed to two different precipitation reactions. The hightemperature reaction, starting at 490�C for 0.1 K/min andlower temperatures for increasing cooling rates, was reportedto be linked to the formation of the equilibrium Mg2Si phase.Another possibility is precipitation of Si, especially inAA6082. The low temperature reaction, starting at about270�C for 0.1 K/min was linked to the precipitation of themetastable phase �0 or B0. At higher cooling rates these twopeaks start to merge. This explains the shoulders in the peaksfor AA6061 and AA6082, and provides a direction for futurework on understanding the precipitation of the non-hardeningMg-Si phases on the dispersoids. At a grain size of about80 mm in AA6060 the surface area of the grains isapproximately 10 mm2/mm3 whereas the surface area ofdispersoids in the dispersoid containing alloys AA6005A,AA6061 and AA6082 is about one order of magnitudehigher. This indicates why dispersoids dominate the precip-itation and hence the quench sensitivity in these alloysystems.

The other interesting result was that in alloy AA6060,which contained only 0.04 Mn and where no dispersoids wereobserved, it appears that quench sensitivity cannot be relatedto dispersoid density. One explanation for its quenchsensitivity can be drawn using Shuey’s et al. findings,15)

which describe solute loss not only to precipitation ondispersoids but also to precipitation on grain boundaries. Thisleads to the conclusion that in cases where dispersoids arepresent, the precipitation of non-hardening phases occurs onthe dispersoids with their density being the dominant factor,and in the case where no dispersoids are present, precipitationcould occur on other heterogeneous nucleation sites. TheDSC results, however, show that no precipitation occursduring the cooling of AA6060. Therefore, other factors arelikely to be responsible for the loss in hardness. A possibleexplanation lies in the reduced amount of vacancies dueto slower cooling rates. Vacancies are involved in earlyclustering during ageing16) and a lower vacancy concentra-tion will reduce the ageing response of an alloy.

Dispersoids in the different dispersoid containing alloysAA6005A, AA6061 and AA6082 were measured to be ofsimilar size for a given homogenization treatment. As theamount of dispersoid forming elements in each of thesealloys is significantly different it can be concluded that thehomogenization treatment, which was the same for the threealloys determines the size of the dispersoids.

To investigate whether the dispersoid size affects thequench sensitivity a final experiment was performed. Threedifferent homogenization treatments were performed onAA6082 alloy (2 h at 510�C, 2 h at 570�C and 12 h at570�C). Subsequently the alloy was extruded, solutiontreated at 550�C for 1 h, quenched with forced air and coldwater, respectively, and aged (8 h at 175�C). The dispersoidshave previously been characterised13) and it was found thatas the homogenization temperature and time increased thedispersoid size also increased, but the composition andcrystal structure of the dispersoids changed depending on thedispersoid size from sc for dispersoids with a length smallerthan 200 nm to bcc for large dispersoids with a lengthexceeding 200 nm. Despite this the quench sensitivity wasagain linearly related to the dispersoid density. Hence, atleast for this alloy it appears that dispersoid size in theobserved range of 50 to 500 nm has little effect on the quenchsensitivity and consequently it appears that it has little effecton the nucleation kinetics. One of the important clues is that,although the crystal structure of the dispersoids variesslightly, all show a cubic structure with the same unit cellparameter. Therefore it is expected that all dispersoidspresent under the conditions described above have the samepotential as nucleation sites for Mg-Si phases. It is nowimportant to determine the orientation relationships betweenthe dispersoids and the non-hardening precipitates to deter-mine whether this is true.

5. Conclusion

The quench sensitivity of four different 6000 series alloyswas determined. Quench sensitivity was most pronouncedin AA6082, followed by AA6061, AA6005 and AA6060.It was apparent that increasing the alloy content, especiallythe amount of transition metals, increased quench sensit-ivity. Quench sensitivity was found to be directly propor-tional to the dispersoid density formed from the transitionmetals, as dispersoids act as heterogeneous nucleation sitesfor the precipitation of coarse Mg-Si phases. The amount

918 K. Strobel, M. A. Easton, L. Sweet, M. J. Couper and J.-F. Nie

of Mg-Si phases that precipitated on the dispersoids wasdetermined using DSC and it was found that the measuredenthalpy values correlated reasonably well with the quenchsensitivity. In addition to the general reduction of Mg and Siin solid solution, the precipitation on dispersoids leads to aninhomogeneous distribution of hardening precipitates andthe formation of precipitate free zones. The dispersoid freealloy AA6060 showed some quench sensitivity as well. Thisreduction in achievable hardness was found to be not relatedto precipitation of Mg-Si phases during cooling fromextrusion temperature. The DSC results indicated that theremay be two reactions that occur during cooling, especiallyin AA6061 and AA6082 indicating that further work isrequired to understand the precipitation of the non-hardeningphases that leads to quench sensitivity.

Acknowledgments

The authors would like to acknowledge the support of theCAST CRC, established and supported by the AustralianGovernment’s Cooperative Research Centres Programme.Rio Tinto Alcan’s contributions to this project are gratefullyacknowledged. The authors would like to thank Dr. SumingZhu for his help with the TEM work and Dr. Steven Pas forhis huge support with the DSC work. Further acknowledgedis the Monash Centre for Electron Microscopy (MCEM) forproviding support and facilities.

REFERENCES

1) L. Lodgaard and N. Ryum: Mater. Sci. Eng. A 283 (2000) 144.

2) O. Lohne and A. L. Dons: Scandinavian J. Metall. 12 (1983) 34.

3) D. V. Gullotti, J. Crane and W. C. Setzer: Isothermal transformation

characteristics of several 6xxx series alloys, ET Anthology, E. T.

Foundation, Ed., (1977, ET Foundation: Atlanta).

4) J. E. Yoo, A. Shan, I. G. Moon and S. J. Maeng: J. Mater. Sci. 34 (1999)

2679.

5) L. Lodgaard and N. Ryum: Mater. Sci. Technol. 16 (2000) 599.

6) C. D. Marioara: Metall. Mater. Trans. A Phys. Metall. Mater. Sci. 36A

(2005) 691.

7) E. D. Sweet, B. R. Harker, X. Zhang and M. J. Couper: Quench

sensitivity test method for aluminium extrusion alloys, Preceedings

of the Ninth Int. Aluminum Extrusion Technilogy Seminar, E. T.

Foundation, Ed., (2008, ET Foundation: Orlando).

8) N. C. W. Kuijpers, W. H. Kool and S. van der Zwaag: Mater. Sci.

Forum 396–402 (2002) 675.

9) G. A. Edwards, K. Stiller, G. L. Dunlop and M. J. Couper: Acta Mater.

46 (1998) 3893.

10) Y. Birol: J. Thermal Anal. Calorimetry 93 (2008) 977.

11) R. J. Livak and J. M. Papazian: Scr. Metall. 18 (1984) 483.

12) B. Milkereit, O. Kessler and C. Schick: Thermochim. Acta 492 (2009)

73.

13) K. Strobel, E. Sweet, M. Easton, J.-F. Nie and M. Couper: Mater. Sci.

Forum 654–656 (2010) 926.

14) B. Milkereit, C. Schick and O. Kessler: The 12th International

Conference on Aluminium Alloys, S. Kumai, et al., Ed., (The Japan

Institute of Light Metals: Yokohama, 2010).

15) R. T. Shuey and A. J. Den Bakker: Mater. Solution Conf., (ASM

International, 2001) p. 189.

16) K. Yamada, T. Sato and A. Kamio: Mater. Sci. Forum 331–337 (2000)

669.

Relating Quench Sensitivity to Microstructure in 6000 Series Aluminium Alloys 919