Effect of Two-Step Aging on Cluster Formation in Al­Mg­Si Alloys+

Ken Takata1, Jun Takahashi2, Makoto Saga1, Kohsaku Ushioda3,Akira Hibino4 and Masao Kikuchi5

1Steel Research Laboratories, Nippon Steel & Sumitomo Metal Corporation, Futtsu 293-8511, Japan2Advanced Technology Research Laboratories, Nippon Steel & Sumitomo Metal Corporation, Futtsu 293-8511, Japan3Technical Development Bureau, Nippon Steel & Sumitomo Metal Corporation, Futtsu 293-8511, Japan4Research & Development Division Fukaya Center, UACJ Corporation, Fukaya 366-8511, Japan5Research Center for Steel, Kyushu University, Fukuoka 819-0395, Japan

The change in the state of a Mg­Si cluster with pre-aging at 363K, followed by aging at 303 or 323K, was studied by means of a tensiletest, three-dimensional atom probe (3DAP), and differential scanning calorimetry (DSC) measurements. Mg­Si clusters formed duringisothermal aging (one-step aging) at 363K after solution heat treatment were different from the ones formed at 303 and 323K. Furthermore,during aging at 303 and 323K following pre-aging at 363K (two-step aging), the clusters that were originally formed at 363K (high-temperatureclusters) grew in size and a new type of clusters (low-temperature cluster) were newly formed at 303 and 323K. The increase in yield stress withaging time at 303 and 323K was greater with the pre-aging at 363K than without the pre-aging. The greater increase in the yield strength wasattributed to the growth of the high-temperature clusters formed in the pre-aging and the nucleation and growth of the low-temperature clusters,both of which proceeded during the aging at 303 or 323K. [doi:10.2320/matertrans.M2013388]

(Received October 10, 2013; Accepted March 17, 2014; Published April 25, 2014)

Keywords: aluminum­magnesium­silicon alloy, magnesium­silicon cluster, two-step aging, three-dimensional atom probe, differentialscanning calorimeter, yield stress

1. Introduction

Manufacturers have recently started to use Al­Mg­Sialloys in automotive panels because they exhibit favorableprecipitation hardening after paint-bake treatment. Thesealloys are pre-aged at a temperature below 448K after solid-solution heat-treatment to achieve paint-bake hardening. Atroom temperature prior to stamping and paint-baking, thestate of precipitates in the alloy can be changed and thestrength can be increased. In order to effectively use thesealloys, the precipitation behavior and the change in theirtensile property during aging at room temperature after pre-aging at 363K must be examined and clarified.

Table 1 shows previous reports on the precipitationsequence in Al­Mg­Si alloys at 448K or lower.1­5) It wasobserved that Mg clusters, Si clusters, Mg­Si clusters, andGuinier­Preston (GP) zones were formed at 373K or lower.The GP zones appear as Mg­Si clusters with about 2.0 nmdiameter in size in transmission electron microscopy (TEM)images.4) The Mg/Si ratios in the Mg­Si clusters, GP zones,and ¢AA precipitates were equal to that of the alloycomposition. In contrast, the Mg/Si ratio in ¢A and ¢ was2 : 1.4) In addition, the GP zones were rapidly formed at448K, and the ¢AA were also formed after a longer period ofaging at the same temperature.

Murayama et al. reported that pre-aging at room temper-ature for 6048 ks following solid-solution heat-treatmentretards the formation of ¢AA.3) Furthermore, a distributiondensity of ¢AA formed after aging at 448K, which followedaging at room temperature, turned out to be lower than that of¢AA formed when the specimens were aged at 448K after pre-aging at 343K.4) From these results, they obtained thefollowing conclusions: the GP zone formed during the aging

at 343K grew into ¢AAwhen heat-treated at 448K; the Mg­Siclusters formed at room temperature did not grow into ¢AA;and thus these two kinds of clusters were different fromeach other. Meanwhile, the Mg­Si clusters formed at roomtemperature could no longer grow into ¢AA. Therefore, whenthe alloy was aged at room temperature after pre-aging at363K, one could expect the formation of another type ofclusters that were different from those formed after pre-agingat 363K. We previously reported that there was a differencein the increasing rate of yield strength with aging time atroom temperature depending on whether the specimens werepre-aged at 363K.6,7) Namely, pre-aging at 363K assisted theincrease in yield strength at room temperature. This impliesthat clusters formed during pre-aging at 363K affected notonly the precipitation behavior during subsequent aging atroom temperature but also the change in tensile properties.The purpose of this study is to clarify the effect of the pre-aging on the precipitation behavior and the change in thetensile strength during the subsequent aging at roomtemperature.

It is well known that the size and distribution density ofclusters affects the yield strength. An increase in yieldstrength resulting from the impediment of the dislocationmovement by precipitates, ¦·, can be represented by thefollowing equation:8,9)

�· ¼ MF=ðbLÞ; ð1Þwhere M, F, b, and L are the Taylor factor, the maximuminteraction force between a precipitate and dislocation, theBurgers vector, and the interspacing of the precipitates,respectively. It is considered that eq. (1) holds true even forclusters. Accordingly, a higher density of clusters lowers theL value and thus raises the yield strength; similarly, a largersize of clusters raises the F value and thus raises the yieldstrength. Consequently, a change in yield strength is expectedto reflect the change in size and distribution density of

+This Paper was Originally Published in Japanese in J. Japan Inst. Metals76 (2012) 677­683.

Materials Transactions, Vol. 55, No. 6 (2014) pp. 885 to 891©2014 The Japan Institute of Metals and Materials

clusters. There are two mechanisms for a dislocation toovercome the clusters, depending on the magnitude of F: theby-pass mechanism, leaving the Orowan loops, for F > Gb2

(where G is the shear modulus); the shear mechanism forthe other case. Note that eq. (1) can be applied to bothmechanisms to estimate the increase in the yield strength.

For the purpose of analyzing the states of Mg and Si in Al­Mg­Si alloys, differential scanning calorimetry (DSC) is ahelpful tool. Moreover, the size and distribution density ofthe clusters can also be rigorously measured using a three-dimensional atom probe (3DAP).

In this study, we first studied the changes in the states ofthe clusters in the alloys subjected to isothermal aging (one-step aging) at 303, 323, or 363K, following a solution heat-treatment. Next, we studied tensile properties in the alloyssubjected to two-step aging: pre-aging at 363K, followed byaging at 303 or 323K. The cluster formation behavior wasanalyzed using DSC, 3DAP, and TEM.

2. Experimental

2.1 SpecimensAl­Mg­Si alloys with Mg and Si content of 0.7mass%

each were used as the specimen. Alloy sheets with athickness of 1mm were produced in a factory by casting, hotrolling, and cold rolling. Subsequently, water quenchingfollowing solution heat-treatment for 60 s at 803K using asalt bath in the laboratory was carried out. The water-quenched specimens were stored in liquid nitrogen, exceptwhen they were subjected to aging treatment and tensile tests.Some water-quenched specimens were subjected to isother-mal aging (one-step aging) at 303K in a drying furnace;some were aged isothermally at 323 or 363K in an oil bath.The other specimens were first subjected to pre-aging at363K for 7.2 to 43.2 ks, after which they were aged at 303 or323K in a two-step aging process.

2.2 Methods for tensile testing, TEM observations,3DAP, and DSC

The yield strength was obtained from tensile testsconducted with an Autograph AG-10T testing machine(Shimadzu, Japan). Tests were performed at room temper-ature using Japanese Industrial Standard (JIS) No. 5 speci-mens with a gauge length of 50mm. The tensile direction wasparallel to the rolling direction, and the cross-head speed was1mm/min. The strain rate was 3.33 © 10¹4/s. The yieldstrength was obtained as the value of 0.2% proof stress. TheTEM thin foil was prepared in the usual manner: thespecimen was first chemically polished on both surfaces toobtain a thickness of 80 µm; it was then perforated with twin-jet electropolishing. An HF-2000 TEM (Hitachi Ltd., Japan)was used with an acceleration voltage of 200 kV.

Each size (the number of component atoms) and thedistribution density of clusters were evaluated with theparticle analysis method10) using 3DAP (Oxford Nano-Science Ltd., UK). As no optimum values for the twoparameters of d, the fixed distance, or N, the minimumnumber of atoms for particle analysis have ever beenestablished, the arbitrariness still has to be eliminated. Basedon our past experience, however, we consider d = 1.0 nm andN = 10 the optimum values. Since the detection rate of theatoms was 50%, the number of component atoms was givenfrom twice of the value obtained by the particle analysis. Forthe present measurement, the specimen temperature was 20Kor lower, and the measured number of atoms was 200,000or higher. The data obtained within the analysis volume,7 © 7 © 70 nm, were used after confirming that there wasno significant scatter in the measured data from more thanthree different points in the same specimen. For the DSCmeasurements, a DSC8230 apparatus (Rigaku Co., Japan)was used. The measurement temperatures ranged from roomtemperature to 773K, and the heating rate was fixed at20K/min.

Table 1 Precipitation sequence in Al­Mg­Si alloy.

Aging conditionAlloy, mass% Phase Measurement Ref.

Temperature, K Aging time

448

259.2 ks Al­0.8Mg­0.79Si ¢AA TEM 1)

32.4 ks Al­0.58Mg­0.32Si, Al­0.53Mg­0.71Si ¢AA TEM 2)

10.8 ks Al­0.53Mg­0.71Si ¢AA TEM 3)

1.8 ks Al­0.53Mg­0.71Si GP zone TEM 3)

1.8 ks Al­0.57Mg­0.31Si, Al­0.53Mg­0.66Si GP zone TEM 4)

1.8 ks Al­0.8Mg­0.79Si ¢AA TEM 1)

373

604.8 ks Al­0.95Mg­0.81Si Mg­Si cluster 3DAP 5)

3.6 ks Al­0.95Mg­0.81Si Mg­Si cluster 3DAP 5)

600 s Al­0.95Mg­0.81Si Mg­Si cluster 3DAP 5)

343 57.6 ks Al­0.57Mg­0.31Si, Al­0.53Mg­0.66Si GP zone TEM, 3DAP 4)

RT

7.88 © 104 ks Al­0.95Mg­0.81Si Mg­Si cluster 3DAP 5)

6048 ks Al­0.53Mg­0.71Si Mg­Si cluster, Mg-cluster, Si-cluster APFIM 3)

6048 ks Al­0.57Mg­0.31Si, Al­0.53Mg­0.66Si Mg­Si cluster, Mg-cluster, Si-cluster 3DAP 4)

2419 ks Al­0.95Mg­0.81Si Mg­Si cluster 3DAP 5)

604.8 ks Al­0.95Mg­0.81Si Mg­Si cluster 3DAP 5)

As-quenched Al­0.8Mg­0.79Si Mg-cluster APFIM 1)

K. Takata et al.886

3. Results and Discussions

3.1 Isothermal (one-step) agingFigure 1 shows the DSC curves obtained with as-quenched

specimens and with specimens isothermally aged at 303, 323,and 363K following the solution treatment. Edwards et al.reported that in the DSC curves obtained after solutiontreatment of an Al­0.8mass% Mg­0.79mass% Si alloy,exothermic peaks were detected from a lower temperaturetoward a higher temperature in the sequence of a Mg­Sicluster formation, ¢AA formation, and ¢A formation.1) Theheating rate for their DSC measurements was 5K/min, whichis slower than the rate of 20K/min used in this study; hence,our exothermic and endothermic peak temperatures shouldbe shifted towards higher temperatures compared with theirmeasurements. Nevertheless, the sequence of the exothermicand endothermic reactions must remain unchanged. There-fore, it was assumed that the observed exothermic peaks at363, 548, and 593K in the present DSC curves of theas-quenched specimens indicate a cluster formation, ¢AAformation and ¢A formation, respectively. Regarding theminute endothermic peak at 503K, we considered that theclusters corresponding to the exothermic peak at 363Kdissolved at 503K. In addition, in the present DSCmeasurements, none of the Mg clusters, Si clusters, or Mg­Si clusters was observed separately. However, the later-described results of 3DAP evidently enabled us to presumethat most of the clusters obtained were Mg­Si clusters.Murayama et al. indicated that Mg­Si clusters formed at343K or higher differed from those that formed at 343K orlower.4) Serizawa et al. reported the existence of two types ofclusters and named them Cluster I and Cluster II.5) As later-describing precisely, we also confirmed the two types ofclusters, of which formation temperatures were different.However, the peak temperature of their formation obtainedby DSC measurements in Serizawa’s study differed fromthose obtained in our present study. Hence, in this paper, therespective clusters are referred to as high- and low-temper-ature clusters.

When the specimen aged at 303K for 604.8 ks wascompared with the as-quenched specimen, we discovered that

the exothermic peak at 363K disappeared, and the endother-mic peak at 503K increased. This indicates that thesupersaturated solute Mg and Si atoms in the quenchedstage formed low-temperature clusters while they were agedat 303K for 604.8 ks. Also, when the specimen aged at 323Kfor 604.8 ks was compared with the as-quenched specimen, itcould be seen that the exothermic peak at 363K disappeared,and the endothermic peak at 503K increased. Accordingly,even with aging at 323K, the supersaturated solute Mg andSi atoms formed low-temperature clusters, similar to the caseof aging at 303K. A comparison of specimen aged at 363Kfor 21.6 ks with the as-quenched specimen showed that theexothermic peak at 363K disappeared, the exothermic peakat 548K sharply decreased, and the exothermic peakoriginally located at 593K shifted towards lower temper-atures. When specimens were aged at the same temperaturefor 604.8 ks, the exothermal amount at 548K decreasedfurther. The above results indicate that high-temperatureclusters might have already formed prior to DSC measure-ments when they were aged at 363K, presumably leading tothe decrease in the exothermic peak at 548K. However, it iswell known that aging at 363K permits only a high-temperature cluster formation, not a ¢AA formation.4,5) There-fore, it was deduced that a decrease in the exothermic peak at548K, with pre-aging at 363K, was related to the formationof high-temperature clusters. We consider that once high-temperature clusters were formed, ¢AA formation wasfacilitated by the high-temperature clusters that acted as thenucleus of ¢AA. Consequently, the larger amount of ¢AA hadbeen formed in the heating period during DSC measurementas the time of aging at 363K was prolonged, leading to thedecrease in the exothermic peak at 548K.

Figure 2 shows a 3DAP map of Mg and Si atoms obtainedfor specimens aged at 363K for 21.6 and 2592 ks,respectively. The clusters containing Mg and Si can beclearly seen in Figs. 2(b) and 2(d), respectively. Theobserved clusters were mostly Mg­Si clusters, which areconsidered to be high-temperature clusters. The distributiondensity and the average number of component atoms of high-temperature clusters in the specimen aged for 21.6 and2592 ks were 4 © 1024m¹3 and 28 atoms for 21.6 ks, and4 © 1024m¹3 and 120 atoms for 2592 ks, indicating that onlythe size increased, with almost no change in clusterdistribution density. Accordingly, it was concluded that agingat 363K did not form a new nucleus of Mg­Si clusters;however, aging offered a chance for the high-temperatureclusters, which were formed in an initial period, to grow.Consequently, it was estimated from 3DAP analysis thatthe average interval between clusters was constant atapproximately 10 nm, irrespective of aging time. Figure 3is a high-resolution TEM (HRTEM) image of the specimenaged at 363K for 2592 ks. The clusters mapped by 3DAPin Fig. 2 were not clearly observed by TEM, as shown inFig. 3.

3.2 Two-step agingFigure 4 shows changes in the yield strength of specimens

aged at 303K, after pre-aging at 363K for 7.2, 14.4, 21.6,and 43.2 ks (two-step aging), together with those of a non-pre-aged (one-step aging) specimen.

300 400 500 600 700 800

Hea

t flo

w, Q

Temperature, T/K

As-quenched

303 K-604.8 ks

323 K-604.8 ks

363 K-21.6 ks

363 K-604.8 ks

363 503 548 593

0.05 mW/mg

Exothermic

Endothermic

Fig. 1 Differential scanning calorimetry (DSC) thermograms of specimenssubjected to quenching (as-quenched) and isothermal aging at 303K for604.8 ks, 323K for 604.8 ks, 363K for 21.6 ks, and 363K for 604.8 ksafter solution heat treatment in a salt bath.

Effect of Two-Step Aging on Cluster Formation in Al­Mg­Si Alloys 887

The initial sharp increase in yield strength observed in thenon-pre-aged specimen possibly resulted from rapid nucle-ation of the low-temperature clusters, which occurredconsuming a great deal of quenched-in vacancies. The rapidformation of the clusters should yield a sharp decrease in L ineq. (1), leading to the sharp increase in "·. On the otherhand, it may be inferred that the slow increase following thesharp initial increase in yield strength stemmed from the slowgrowth of the low-temperature clusters.

The initial yield strength in the pre-aged specimensincreased with pre-aging time at 363K. Based on the3DAP maps shown in Fig. 2, we consider that this increase

primarily resulted from the growth of high-temperatureclusters through an increase in F in eq. (1). Furthermore,the increasing rate in yield strength with aging time at 303Kwas higher with the pre-aging at 363K than without the pre-aging. In the pre-aged specimens, the two types of clusters®high- and the low-temperature clusters®are believed to haveevolved during aging at 303K. Meanwhile, in the non-pre-aged specimen, only low-temperature clusters were formed.Therefore, it is postulated that the difference in the increasingrate in yield strength with aging time at 303K was associatedwith the difference in the states of clusters formed duringaging at 303K. Both the high- and low-temperature clustersas well as solute atoms are considered to have contributed tothe yield strength. Since no high-temperature clusters existedin the non-pre-aged specimen, a slight increase in yieldstrength with aging time at 303K can be attributed to thecombined contribution of a softening with the decrease in theamount of solute atoms and a hardening with the growth oflow-temperature cluster, both of which offsets each other. Incontrast, in the pre-aged specimens, high-temperature clustersexisted prior to aging at 303K and the yield strengthincreased more significantly with aging time at 303K,compared to the non-pre-aged specimen. These facts indicatethat in addition to the nucleation and growth of low-temperature clusters, the increase in size of the pre-existedhigh-temperature clusters occurred with increasing agingtime at 303K after pre-aging at 363K.

Figure 5 shows the DSC curves of specimens aged at303K or at 323K after pre-aging at 363K for 21.6 ks. Here,323K is considered to be the temperature where low-temperature clusters were formed as they were during agingat 303K. The DSC curve of the specimen aged at 303K for864 ks, after the formation of high-temperature clusters at363K, shows an increase in the endothermal peak at 503Kand a decrease in the exothermal peak at 548K, compared

(a)

(b)

(c)

(d)

10 nm

Fig. 2 3D atom probe maps of Mg and Si of specimens aged at 363K: (a) after aging for 21.6 ks; (b) same after particle analysis; (c) afteraging for 2592 ks; (d) same after particle analysis. The Mg and Si atoms are depicted by blue and red circles, respectively. In the particleanalysis, d = 1.0 nm and N = 10.

200020

000

1 nm

Fig. 3 A HRTEM image taken at the [001] Al zone axis and a diffractionpattern taken at [001] in a selected area of an alloy specimen aged at363K for 4320 ks. No contrast corresponding to precipitates wasobserved. The dark regions seen in this image was not due to anyprecipitation or reconstruction of the lattice, but appeared to be due to thedifference in thickness of the membrane.

K. Takata et al.888

with that of the specimen without pre-aging at 363K. TheDSC curve of the specimen aged at 303K for 1728 ks showsa further increase in the endothermal peak at 503K and afurther decrease in the exothermal peak at 548K. Theincrease in the endothermal peak at 503K implies an increasein the amount of the low-temperature cluster formationsduring aging at 303K, and the decrease in the exothermalpeak at 548K implies an increase in the amount of the high-temperature clusters. Also, in the DSC curve obtained for

specimens aged at 323K, the endothermal peak at 503Kincreased while the exothermal peak at 548K decreased,which indicates that the same phenomenon took place duringaging at both 303 and 323K.

Figure 6 shows 3DAP maps of Mg and Si atoms of thespecimens aged immediately after pre-aging at 363K for21.6 ks (a), (b) and of specimens aged at 303K for 864 ks (c),(d) and 4320 ks (e), (f ) after pre-aging at 363K for 21.6 ks,respectively; (b), (d), and (f ) show the results of particleanalysis of the 3DAP maps. Particle analysis enabled us toobserve the clusters clearly. Table 2 summarizes the averagevalues of the number of component atoms in the clusters, andtheir distribution density. The analysis revealed that thenumber of component atoms and their distribution density ofclusters increased with aging time at 303K.

Assuming that the specimen aged at 303K following pre-aging at 363K never formed new high-temperature clusters,the increase in the amount of high-temperature clustersobserved by the DSC measurement (Fig. 5) means that thehigh-temperature clusters grew without changing the distri-bution density. Therefore, the increase in distribution densityof clusters revealed by the 3DAP analysis can probably beattributed to the formation of new low-temperature clusters.In addition, the increase in the amount of low-temperatureclusters confirmed by the DSC measurements (Fig. 5)indicates that the low-temperature clusters grew withincreasing aging time at 303K. Consequently it was deducedthat, room-temperature aging after pre-aging at 363K yieldsthe nucleation and growth of the low-temperature clusters aswell as the growth of the high-temperature clusters.

50

100

150

0 2000 4000 6000

Yie

ld s

tren

gth,

σ/M

Pa

Aging time, t/ks

(a)

50

100

150

0 2000 4000 6000Yie

ld s

tren

gth,

σ/M

Pa

Aging time, t/ks

(b)

50

100

150

0 2000 4000 6000Yie

ld s

tren

gth,

σ/M

Pa

Aging time, t/ks

(c)

50

100

150

0 2000 4000 6000Yie

ld s

tren

gth,

σ/M

Pa

Aging time, t/ks

(d)

Fig. 4 Aging curves of yield strength at 303K without and after pre-agingat 363K for (a) 7.2 ks, (b) 14.4 ks, (c) 21.6 ks, and (d) 43.2 ks. The solidand open circles represent data obtained without and with pre-aging,respectively.

300 400 500 600 700 800

Hea

t flo

w, Q

Temperature, T/K

0.05 mW/mg

pre-aged

503 548

323 K-1728 ks

323 K-864 ks

303 K-864 ks

303 K-1728 ks Exothermic

Endothermic

Fig. 5 DSC thermograms of Al­Mg­Si alloy specimens subjected to pre-aging at 363K for 21.6 ks (pre-aged) and pre-aged and isothermally agedat 303K for 864 ks, 323K for 864 ks, 303K for 1728 ks, and 323K for1728 ks after solution heat treatment in a salt bath and quenching.

Table 2 Number of component atoms of Mg­Si clusters and distributiondensity of the clusters obtained through 3DAP analysis.

Aging conditionsNumber of

component atomsDistributiondensity/m¹3

363K for 21.6 ks 28 4 © 1024

363K for 21.6 ks followedby 303K for 864 ks

40 5 © 1024

363K for 21.6 ks followedby 303K for 4320ks

68 7 © 1024

Effect of Two-Step Aging on Cluster Formation in Al­Mg­Si Alloys 889

Based on the above discussion, the growth behavior of theclusters aged at room temperature after pre-aging at 363K isschematically illustrated in Fig. 7. Immediately after pre-

aging, the high-temperature clusters formed, but at the sametime the Mg and Si atoms both remained in the solid solution(Fig. 7(a)). During room-temperature aging following thepre-aging at 363K, the solute Mg and Si atoms are believedto be consumed through the growth of pre-existing high-temperature clusters and the nucleation and growth of thelow-temperature clusters.

4. Conclusions

The cluster formation in Al­0.7mass% Mg­0.7mass% Sialloys during one-step aging (isothermal aging at 303, 323, or363K) or two-step aging (363K aging followed by aging at303K or 323K) was studied by analyzing the changes inyield strength the DSC and 3DAP observation data. Thefollowing findings were obtained:(1) In the case of isothermal aging at 363K, the high-

temperature clusters grew, and the number of compo-nent atoms increased with aging time.

(2) The increasing rate in yield strength with aging timewas higher for the two-step-aged specimens than thatafter the initial sharp increase for the one-step-agedspecimens.

(3) In the case of two-step aging, the high-temperatureclusters formed by pre-aging at 363K grew with aging

(a)

(b)

(c)

(d)

(e)

(f)

10 nm

Fig. 6 3D atom probe maps of Mg and Si of specimens pre-aged at 363K for 21.6 ks: (a) immediately after pre-aging; (b) same afterparticle analysis; (c) after additional aging at 303K for 864 ks; (d) same after particle analysis; (e) after additional aging at 303K for4320 ks; (f ) same after particle analysis. The Mg and Si atoms are depicted by blue and red circles, respectively. In the particle analysis,d = 1.0 nm and N = 10.

High-temperature cluster

Low-temperature cluster

Mg-Si atoms

(a) (b)

Fig. 7 Schematic illustrations of the formation of high- and low-temper-ature clusters in the specimens aged at 303K following the pre-aging at363K for 7.2 ks in Fig. 4: (a) pre-aged for 7.2 ks; (b) during aging at303K after the pre-aging.

K. Takata et al.890

time at 303K, and it was deduced that a different typeof clusters (low-temperature clusters) were alsonucleated and grew. The above findings were verifiedby both the DSC and 3DAP analyses.

REFERENCES

1) G. A. Edwards, K. Stiller, G. L. Dunlop and M. J. Couper: Acta Mater.46 (1998) 3893­3904.

2) N. Maruyama, R. Uemori, N. Hashimoto, M. Saga and M. Kikuchi:Scr. Mater. 36 (1997) 89­93.

3) M. Murayama, K. Hono, M. Saga and M. Kikuchi: Mater. Sci. Eng. A

250 (1998) 127­132.4) M. Murayama and K. Hono: Acta Mater. 47 (1999) 1537­1548.5) A. Serizawa, S. Hirosawa and T. Sato: Metall. Mater. Trans. A 39

(2008) 243­251.6) K. Takata, K. Ushioda and M. Kikuchi: Mater. Sci. Forum 519­521

(2006) 233­238.7) K. Takata, M. Saga, J. Takahashi, K. Ushioda, A. Hibino and M.

Kikuchi: Proc. 11th Int. Conf. Aluminum Alloy, (2008) pp. 961­968.8) A. J. Ardell: Metal. Trans. A 16 (1985) 2131­2165.9) V. Gerold: Dislocations in Solids, vol. 4, (North Holland Publ. Co.,

Amsterdam, 1979) p. 219.10) D. Vaumousse, A. Cerezo and P. J. Warren: Ultramicroscopy 95 (2003)

215­221.

Effect of Two-Step Aging on Cluster Formation in Al­Mg­Si Alloys 891