Two-Step Die Motion for Die Quenching of AA2024Aluminum Alloy Billet on Servo Press

Jae-Yeol Jeon+1, Ryo Matsumoto and Hiroshi Utsunomiya+2

Division of Materials and Manufacturing Science, Graduate School of Engineering,Osaka University, Suita 565-0871, Japan

The authors reported that die quenching of a cylindrical AA2024 aluminum alloy billet less than 9mm in height was feasible on a servopress. However, it was also found that the reduction in height was limited less than 5% due to partial melting. In order to enhance thedeformability in single operation, the two-step die motion is proposed. A cylindrical billet was heated to 823K and transferred to the press. Thenthe billet was uniaxially compressed with ¦h/h0 = 5%, and further held between the dies for cooling. After sandwiching for 8 s, the billet with aheight of h1 = 7.6mm was further compressed with a reduction in height (¦h/h1) of 2 or 5% at lower temperature. The die quenching processwith the two-step die motion leads to increase the total reduction in height to 10%. It is confirmed that super-saturated solid solution successfullyformed at the 1st step is maintained in the 2nd step. It is found that the peak hardness of the two-step processed billet is higher than that of theone-step processed billet, and that the precipitation kinetics in artificial aging is accelerated by the two-step motion.[doi:10.2320/matertrans.L-M2014806]

(Received January 21, 2014; Accepted February 25, 2014; Published April 11, 2014)

Keywords: die quenching, AA2024, precipitation, servo press, two-step die motion, strain aging

1. Introduction

In automotive industries, there is a rapid growth in demandof lightweight components such as aluminum alloy part toimprove fuel consumption. At present, the required strengthis achieved by precipitation hardening of age-hardenablealuminum alloys after forming, solution heat treatment (SHT)and quenching. Though the ductility recovers and uniformmicrostructure develops in the solution heat treatment, itoften results in thermal distortion of formed components. Inorder to suppress thermal distortion of formed componentsand to improve its strength and to form desirable shapedmicrostructure, die quenching processes were studied inrecent years.1­4) In these processes, first the aluminum alloy isheated to dissolve the precipitates within the primary ¡-Almatrix. Then, forming and quenching are simultaneouslyperformed. In other words, the material is formed as well ascooled by the dies at once. Then, the processed aluminumalloy is artificially aged for precipitation hardening. Thus,this technique is able to manufacture stronger products withcomplicated shapes through less number of processes withless thermal distortion compared with the conventional T6treatment. Most of past papers were related to sheet metalsdue to easier rapid cooling because automotive bodies wereconsidered for applications.1­3) In previous study, the authorsinvestigated the feasibility on die quenching of a cylindricalAA2024 aluminum alloy billet after the two different SHTtemperatures using highly conductive WC­20mass%Codies.5) Die quenching was feasible after SHT at 823K underlimited conditions, while not feasible after SHT at 773K dueto precipitation caused during slower cooling. The authorsalso reported that the billet height should be less than 9mmfor die quenching.6) It was also found, however, the reductionin height (¦h/h0) was very limited less than 5% (A billetheight of h0 = 8mm) due to crack formation on side surface.

Wang et al. also reported the lower ductility of an AA2024-T3 aluminum alloy at die quenching process.2) It wassupposed that cracks were formed by partial melting atgrain-boundary triple junctions.5,6) It could be a seriousdrawback for industrial applications because thick orcomplicated parts are not die-quenchable.

Thus, the objective of this study is that in order to applyhigher reduction in single operation, the two-step die motionis proposed. After the one-step die quenching, furthercompression is applied without opening dies. Therefore, thesecond compression can be regarded as pre-straining process.The effectiveness of the two-step motion is investigated. Inaddition, the aging behavior of the die-quenched billet ismade clear.

2. Two-Step Die Motion

In general, the age-hardenable aluminum alloy billet issolution heat-treated followed by the water quenching asshown in Fig. 1(a). On the other hand, in the one-step processfor die quenching of age-hardenable aluminum alloy billet,fully solutionized billet is plastically deformed beforeprecipitation start, simultaneously rapidly cooled down tobelow the nose of the CCT diagram by sandwiching betweenthe conductive dies to form super-saturated solid solution(SSSS) as shown in Fig. 1(b). In the two-step processproposed by the authors, after the first compression furthercold compression is applied. In detail, after the firstcompression as Fig. 1(b), then straightly the die-quenchedbillet is further cold worked in single operation with diemotion control as shown in Fig. 1(c).

The expected advantages of the proposing two-stepprocess are classified into two main points. First, it willenhance the deformability in total during the die quenchingprocess. It widens shaping and forming applications. Basedon the previous studies,5,6) if a billet was compressed over 6%in height, cracks were initiated at voids caused by partialmelting at grain-boundary triple junctions and that were

+1Graduate Student, Osaka University+2Corresponding author, E-mail: uts@mat.eng.osaka-u.ac.jp

Materials Transactions, Vol. 55, No. 5 (2014) pp. 818 to 826©2014 The Japan Institute of Light Metals

subjected to tensile stress. So the deformability may beimproved if further compression is applied sufficiently belowthe solidus temperature.

Second, it will accelerate the precipitation kinetics andincrease the strength during isothermal artificial aging. Age-hardening is essential because it is the final stage in thedevelopment of the properties of heat-treatable alloys suchas AA2xxx, AA6xxx and AA7xxx, which is the controlledby decomposition of the SSSS to form finely dispersedprecipitates. It is well known that deformation has asubstantial effect on the precipitation behavior.7,8) Gomieroet al. reported that the role of dislocations generally results inacceleration of the precipitation kinetics and coarsening of

the precipitates.9) Yassar et al. reported that pre-strainingbefore aging accelerates precipitation due to decreasedactivation energy for the growth of precipitates.10) Kim et al.found that pre-ECAP solution treatment combined with post-ECAP aging is very effective to enhance the strength ofAA6061 aluminum alloy.11) These suggest that the combinedeffect of the two-step process which consists of firstcompression of die quenching and second cold compressionmay be useful. Using a servo press, the pre-straining beforeaging does not require further process because the mainstraining and the pre-straining can be conducted in singleoperation if die motion is controlled. So strain aging can beutilized without decreasing productivity or increasing cost.Therefore, servo press is an advantageous machine becausepress ram motion can be controlled flexibly to increase thestrength with strain aging.12)

3. Experimental Procedures

The material used was a cylindrical AA2024-T4 aluminumalloy billet. The chemical composition of this alloy is givenin Table 1. Billets with a diameter of 16mm and a height ofh0 = 8mm were machined from a round bar of 20mm indiameter. The Vickers hardness of the as-received billet was124HV. The forming was conducted on a 450 kN servo press.The tool arrangement for the die quenching is shown inFig. 2. In this hot forming and die quenching (HFQ) process,first the billets were solution heat-treated at 823K for 1.8 ksin an electric furnace under air and transferred to the presswithin a few seconds. The billets were loosely supported bycopper wires between the dies to avoid contact with lower diebefore the HFQ process. Cemented tungsten carbide dies(WC­20mass%Co) were used. The thermal conductivity ofthe die was approximately 70W·m¹1·K¹1, which was muchlarger than that of tool steels.13) The surface roughnesses ofthe dies were 0.02­0.04 µmRa. The dies were kept at roomtemperature before the HFQ process. Then, the billet wascompressed by 5% in height (¦h/h0) during the first motionbecause the maximum reduction by the one-step die motionwas ¦h/h0 = 5% due to partial melting in previousstudies.5,6) The deformation duration was 0.238 s for thebillet with ¦h/h0 = 5%. The corresponding mean strain ratewas 0.21 s¹1 for the billet with ¦h/h0 = 5%. The compressedbillet with a height of h1 = 7.6mm was further held betweenthe dies for quenching; sandwiching duration by dies,td.q. = 8 s. And then straightly the second reduction in height(¦h/h1) of 2 or 5% was applied without opening gapbetween the dies. As the billet temperature was cooled in thefirst step, the second step was cold compression. Therefore,the total reduction was ¦h/h1 = 7% (1 ¹ 0.95 © 0.98 =0.07) or ¦h/h1 = 10% (1 ¹ 0.95 © 0.95 = 0.1). The defor-mation durations were 0.216 s for the billet with ¦h/h1 = 2%

Time

Tem

pera

ture

Precipitation

(a)

Time

Tem

pera

ture

Precipitation

(b)

Time

Tem

pera

ture

Precipitation

(c)

Fig. 1 Schematic illustration of the conventional process for the waterquenching (a), the one-step process for die quenching (b), and the two-step process for die quenching (c) imposed on the CCT diagram.

Table 1 Chemical composition of the alloy utilized in this study [in mass%].

Element Si Fe Cu Mn Mg Cr Zn Ti Zr+TiOthers

AlEach Total

AA2024 0.16 0.19 4.8 0.59 1.6 0.06 0.02 0.01 0.08 0.05 0.15 Re

Two-Step Die Motion for Die Quenching of AA2024 Aluminum Alloy Billet on Servo Press 819

and 0.238 s for the billet with ¦h/h1 = 5%, respectively. Thecorresponding mean strain rates were 0.09 s¹1 for the billetwith ¦h/h1 = 2% and 0.21 s¹1 for the billet with ¦h/h1 =5%. The further compressed billets were again held betweenthe dies for 1 s to avoid the elastic recovery. Then, the billetwas ejected from the bottom dead center of the press andcooled down to room temperature in air as shown in Fig. 3.Subsequently, the billet was artificially aged at 463K in an oilbath. The properties of the die-quenched billets by the two-step die motion were compared with those of the billets bythe one-step die motion, the water quenching (WQ) and theair cooling (AC) without forming. Vickers hardness test wasperformed on a longitudinal section with load of 0.49N andholding duration of 15 s. To clarify that die quenching wassuccessfully conducted without precipitation hardening, theprocessed billets at the center were subjected to thermogravimetric-differential thermal Analysis (TG-DTA). Thespecimens of 3mm in diameter with less than 10mg ofweight were machined from the processed billets. TG-DTAanalysis under nitrogen atmosphere was performed from 503to 603K with the heating rate of 10K/min with 10mg Al2O3

(99.99%) as a reference.

4. Results

4.1 DeformabilityBy the one-step die motion the height of the billet can be

compressed without cracks by less than 5% in height. Abovethe reduction, cracks occur along the thickness direction onbulged side surface during the HFQ process as shown inFig. 4. In other words, by the one-step die motion thedeformability of the alloy in HFQ process is rather limited by

the intergranular fracture.5,6) On the other hand, the billet iscompressed by less than 10% in height by the two-step diemotion without defects. It means that in the case of the two-step, the total reduction in height (¦h/h1) increases up to10% as expected. The mechanism of fracture during the HFQprocess by the two-step die motion will be discussed laterwhen the total reduction is over 11% in height (¦h/h1).Interestingly, the appearances of the as-HFQed billets by theone-step and the two-step die motion over 6% reduction inheight show that more vertical profile around the end surfacesthan that of the lower reduction which just shows slightlybulged side surface with decreasing reduction. The verticalprofile implies that the cooling rate is higher near the endsurfaces so that the deformation is concentrated around thecenter. The billets show similar vertical profiles over 6%reduction in height.

4.2 Temperature changeThe measured temperature changes of the WQ, the AC and

the HFQ by the one-step and the two-step die motion arecompared in Fig. 5. The temperature was measured by aK-type thermocouple welded at the center of side surface. Inthe case of the one-step shown in Fig. 5(a), the cooling rateof the WQ is the fastest as expected. The temperaturedecreases to below 400K within one second. The next is theHFQ with ¦h/h0 = 5% and that with ¦h/h0 = 2% in order.The cooling rate of the higher reduction (¦h/h0 = 5%) isfaster than that of the lower (¦h/h0 = 2%) until the ejectionfrom the press. It is found that higher reduction is effective toincrease the cooling rate. This may be due to the fact thatmore heat is transferred from the billet to the dies due tohigher pressure and higher contact ratio on the interface whenthe reduction is higher. The continuous cooling time (CCT)diagram of AA2024 in literature14) is super-imposed onFig. 5. The cooling curves of the WQ, the HFQ with ¦h/h0 = 2% and that with ¦h/h0 = 5% are located below thenose of the precipitation-start curve. It is supposed that diequenching of the billet was successfully carried out by theHFQ as well as the WQ without precipitation. On the otherhand, the cooling curve of the AC is placed above the nose so

Billet

Dies

T.C.

Thermo-couple

Supporting wire

Fig. 2 The tool arrangement for the die quenching on a servo press.

1st Holding(td.q.= 8 s)

1st Compression

Pos

ition

of r

am

Time Bottom dead center

Top dead center

2nd Holding(td.q.= 1 s)

2nd Compression

Fig. 3 Schematic illustration of press ram motion in forming with diequenching on a servo press.

ΔΔh/h0 = 2% Δh/h0 = 5% Δh/h0 = 6%

10mm

Cracks

(a)

FD

10mm

Δh/h1 = 7% Δh/h1 = 10% Δh/h1 = 11%

Cracks

(b)

FD

Fig. 4 Photographs of the as-HFQed billets with different reduction inheight. (a) One-step and (b) Two-step.

J.-Y. Jeon, R. Matsumoto and H. Utsunomiya820

that precipitation takes place. The longer dotted line imposedon the CCT diagram is an imaginally curve because theprecipitation of HFQ may be accelerated by the strainintroduced.

In the case of the two-step shown in Fig. 5(b), the coolingrates of the two HFQ processes show almost the same as thatof the one-step with ¦h/h0 = 5% in Fig. 5(a) because theHFQ process just before the second compression is the sameas that of ¦h/h0 = 5% by the one-step die motion. Andtemperatures just slightly increase at the second compressiondue to the heat generation by plastic work. It means that coldcompression was conducted at the second step after sufficientcooling. The two cooling curves of the HFQ are also belowthe nose of the precipitation-start curve. Therefore, it issupposed that super-saturated solid solution is maintainedthrough the two-step process. The longer dotted line is alsoimposed on the CCT diagram because the precipitation ofHFQ by the two-step die motion may be more accelerated byadditional strain introduced at the second step.

4.3 Hardness changeThe hardnesses at the center of the billet after cooling are

compared in Fig. 6. Vickers hardnesses of the as-WQed billetand the as-ACed billet are 107 and 117HV, respectively. Inthe case of the one-step, Vickers hardnesses of the as-HFQedbillets with ¦h/h0 = 2% and that with ¦h/h0 = 5% are 108and 109HV, respectively. The hardness of the as-WQed billetshows the lowest. On the other hand, the hardness of theas-ACed billet is the highest. This is due to precipitationhardening occurred during cooling. The two as-HFQed billetsshow similar hardness as the as-WQed billet. The hardnessesof the as-HFQed billets well account for successful formationof super-saturated solid solution (SSSS) as the waterquenching. These hardness changes well correspond withthe reported precipitation curve shown in Fig. 5(a).

In the case of the two-step, Vickers hardnesses of theas-HFQed billet with ¦h/h1 = 7% and that with ¦h/h1 =10% are 114 and 121HV, respectively. The hardnessess ofthe two as-HFQed billets are higher than that of the one-stepprocessed billets and increase with reduction at the secondstep. The reason is work hardening caused by the secondstep.

4.4 DTA analysisFigure 7 shows the results of TG-DTA analysis. It is

known that major strengthening phase of AA2024 alloy is SAprecipitate formed around 553­563K.15) In the case of theone-step shown in Fig. 7(a), the peak temperatures of the as-WQed billet, the as-ACed billet, the as-HFQed billet with¦h/h0 = 2% and that with ¦h/h0 = 5% are 554, 578, 546and 542K, respectively. Those of the two as-HFQed billetsare even lower than the as-WQed billet due to strain aging.The peak temperature of the as-HFQed billet with ¦h/h0 =5% is the lowest. The peak temperature of the as-ACed billetis far higher than that of the as-WQed billet. It means that theprecipitates were well formed during the cooling.

In the case of the two-step shown in Fig. 7(b), the peaktemperatures of the as-HFQed billet with ¦h/h1 = 7% andthat with ¦h/h1 = 10% are 530 and 528K, respectively. Thepeak temperatures of the two as-HFQed billets are muchlower than the two as-HFQed billets by the one-step diemotion. In addition, the peak temperature is lower whenhigher reduction is applied. Therefore, it is found that

10-1 100 101 102 103

300

400

500

600

700

800

900

Tem

per

atu

re, T

/ K

Time, t / s

[14]

AC

WQ

Δh/h0 = 5%

Δh/h0 = 2%

(a)

Air-coolingCompression Sandwiching

by dies Air-cooling

10-1 100 101 102 103

300

400

500

600

700

800

900(b)

Time, t / s

Tem

per

atu

re, T

/ K

Δh/h1 = 7%

Δh/h1 = 10%

[14]

Air-cooling1st Compression

Air-cooling

Sandwiching by dies

2nd Compression

Sandwiching by dies

Fig. 5 Measured temperature changes of the WQ, the AC and the HFQimposed on the CCT diagram of AA2024. (a) One-step and (b) Two-step.

80

100

120

140

160

= 2%AC

Vic

kers

har

dn

ess,

HV

WQ Δh/h0 Δh/h0= 5%

Δh/h1= 7%

Δh/h1= 10%

One-step Two-step

Fig. 6 Hardnesses at the center of the billet after cooling of the as-WQedbillet, the as-ACed billet and the as-HFQed billets by the one-step diemotion and the two-step die motion.

Two-Step Die Motion for Die Quenching of AA2024 Aluminum Alloy Billet on Servo Press 821

successful die quenching is maintained during the two-stepdie motion. And the billet is work-hardened in the secondstep.

4.5 Aging behaviorHardnesses at the center of the as-quenched and the peak-

aged billets are summarized in Fig. 8. Hardness changesduring isothermal aging at 463K are shown in Fig. 9. Fromthe hardness of the as-quenched state, the hardnesses of allthe billets increase with increasing aging time to their peaks.The peak aging time of the WQed billet and the ACed billetare 46.8 and 86.4 ks, respectively and hardness of that billetsare 149 and 141HV, respectively. The WQed billet showsearlier precipitation kinetics and higher hardness as expecteddue to its higher cooling rate. In the case of the one-step,hardnesses of the two HFQed billets increase rapidly thanthat of the WQed and the ACed billets in turn. The hardnessincrease of the HFQed billet with the ¦h/h0 = 5% is earlierthan that of the ¦h/h0 = 2%. The peak aging time of theHFQed billet with ¦h/h0 = 5% is at 32.4 ks and that with¦h/h0 = 2% is at 36.0 ks. The peak hardness of the HFQedbillet with ¦h/h0 = 5% is 145HVand that with ¦h/h0 = 2%is 141HV. It is found that faster precipitation kinetics andhigher hardness are achieved with increasing reduction in the

second step. However, the peak hardness of the billet with¦h/h0 = 5% is lower than that of the WQed billet. It issupposed that the number of heterogeneous nucleation siteson dislocations for precipitates in the HFQed billet with¦h/h0 = 5% are higher than that of the WQed billet.

In the case of the two-step, from the hardness of the as-quenched state, the hardnesses of the two HFQed billetsincrease with increasing aging time to their peaks. Especiallythe two HFQed billets by the two-step die motion increaserapidly than that of the billet with ¦h/h0 = 5% by the one-step die motion. The peak aging time of the HFQed billetwith ¦h/h1 = 10% is at 21.6 ks and that with ¦h/h1 = 7%is 25.2 ks. The peak hardness of the HFQed billet with¦h/h1 = 10% is 150HV and that with ¦h/h1 = 7% is146HV. The peak hardness of the HFQed billet with ¦h/h1 = 10% shows comparable value with the WQed billet andthe peak aging time is the earliest among all the conditionsdue to the second strain introduced during the HFQ process.It is also found that the HFQ process by the two-step diemotion more accelerates the precipitation kinetics andincreases the peak hardness than that by the one-step diemotion.

4.6 Hardness distributionHardness distributions through the billet height processed

Temperature, T / K

En

do

ther

mal

Exo

ther

mal

Hea

t fl

ow

/ n

Vs-1

Δh/h1 = 7%

Δh/h1 = 10%

(b)50

520 540 560 580 600

520 540 560 580 600

(a)

Temperature, T / K

Hea

t fl

ow

/ n

Vs-1

Exo

ther

mal

E

nd

oth

erm

al

AC

WQ

Δh/h0 = 2%

Δh/h0 = 5%

50

Fig. 7 TG-DTA results at the center of the as-WQed billet, the as-ACedbillet and the as-HFQed billets at the heating rate of 10K/min. (a) One-step and (b) Two-step.

80

100

120

140

160

180

21.6 ks25.2 ks32.4 ks36.0 ks86.4 ks

Vic

kers

har

dn

ess,

HV

= 2%Δh/h0WQ AC

= 5%Δh/h0 Δh/h1

= 7%Δh/h1= 10%

46.8 ks

One-step Two-step Peak-agedAs-quenched

Peak aging time

Fig. 8 Hardnesses at the center of the as-quenched and the peak-agedbillets.

0.1 1 10 10080

90

100

110

120

130

140

150

160

170

180

Aging time, t / ks

Vic

kers

har

dn

ess,

HV

----

----

----

----

----

----

--

AQ

WQACΔh/h0 = 2%Δh/h0 = 5%

Δh/h1 = 7%Δh/h1 = 10%

Fig. 9 Changes in hardnesses at the center of billets during isothermalaging at 463K.

J.-Y. Jeon, R. Matsumoto and H. Utsunomiya822

by the one-step die motion and the two-step die motion areshown in Fig. 10. The hardness was measured on longi-tudinal section through the height of billet. In the case of theone-step shown in Fig. 10(a), the as-HFQed billet with¦h/h0 = 2% and that with ¦h/h0 = 5% show uniformhardness distributions through the height. The as-WQedand the as-ACed billets also show relatively uniformdistributions. In the peak-aged state, the two HFQed billetsshow no significant distributions in hardness. It means thatAA2024-T4 aluminum alloy billet with h0 = 8mm isuniformly die-quenched due to high cooling rate and highthermal conductivity of the alloy. It is found that the uniformhardness distributions of the as-HFQed billets were also keptafter the peak-aging.

On the other hand, in the case of the two-step shown inFig. 10(b), the as-HFQed billet with ¦h/h1 = 7% and thatwith ¦h/h1 = 10% show curved hardness distributions,while the two as-HFQed billets by the one-step die motionshow uniform hardness distributions through the height. Thereason of the curved distributions with higher hardness nearthe center is attributed to the work hardening introduced

during additional compression is mainly concentrated at thecenter. Hardness distribution of the as-HFQed billet with¦h/h1 = 10% toward the center is wider than that of¦h/h1 = 7% as shown in Fig. 10(b). It is thought that theamount of work hardening near the center is greater due tohigher amount of reduction and relatively higher temperatureat the center during the process. In the peak-aged state, thetwo HFQed billets show slightly curved distributions inhardness. It is thought that strain accumulation near the centeris relieved during the artificially aging.

5. Discussions

It is confirmed that the HFQ process using the proposedtwo-step die motion is effective to increase the total reductionin height compared with the one-step die motion undercertain conditions. It is proved that super-saturated solidsolution is successfully maintained through the two-step diemotion without precipitation as well as the one-step diemotion and even the WQ. It is confirmed by the fact that thecooling curves of the HFQ by the two-step die motion arealmost the same as that of the HFQ with ¦h/h0 = 5% by theone-step die motion. Those are below the nose of the reportedprecipitation curve results in super-saturated solid solution asthe WQ.

The hardnesses of the as-HFQed billets by the two-step diemotion are higher than that of the billets by the one-step diemotion, while those are very similar to that of the as-WQedbillet. This is partly due to work hardening introducedby additional compression of the second step. The workhardening is verified as a restoration process by the TG-DTAanalysis. Hence, higher hardnesses of the as-HFQed billetsby the two-step die motion are caused by work hardening.The peak aging time and hardness of the HFQed billet with¦h/h0 = 5% processed by the one-step die motion are32.4 ks and 145HV, respectively, while the peak aging timeand hardness of the HFQed billet with ¦h/h1 = 10%processed by the two-step die motion are 21.6 ks and150HV, respectively. It is found that the two-step die motionaccelerates the precipitation kinetics and increases the peakhardness more rather than that of the one-step die motion dueto strain aging of higher strain introduced at the second step.

5.1 Improvement in deformabilityThe total reduction in height increases from the billet with

¦h/h0 = 5% of the one-step to ¦h/h1 = 10% by the two-step die motion. Above the reduction, cracks occur on bulgedside surface of the billet. In previous study, the HFQ processover 6% reduction in height (¦h/h0) by the one-step diemotion leads to cracks occurred by intergranular fracture dueto partial melting at grain-boundary triple junction eventhough the SHT temperature at 823K was closed to thesolidus temperature at 775K in an AA2024 aluminumalloy.5,6) To investigate the formation of cracks, SEM imagesof fractured surface on the billet with ¦h/h1 = 11% by thetwo-step die motion are shown in Fig. 11 and compared thebillet with ¦h/h0 = 6% by the one-step die motion. Toolmarks formed in machining were observed in the horizontaldirection. In images of high magnification, intergranularfracture is also observed on the as-HFQed billet by the two-

80

90

100

110

120

130

140

150

160

170

180(a)

Vic

kers

har

dn

ess,

HV

Distance from center of billet / mm

AQ

PA

BottomTop Center

WQACΔh/h0 = 2%Δh/h0 = 5%

0 1 2 3

-3 -2 -1

-3 -2 -1

0 1 2 380

90

100

110

120

130

140

150

160

170

180(b)

Distance from center of billet / mm

Vic

kers

har

dn

ess,

HV

AQ

PA

BottomTop Center

Δh/h1 = 7%Δh/h1 = 10%

Fig. 10 Hardness distributions through the billet height. (a) One-step and(b) Two-step.

Two-Step Die Motion for Die Quenching of AA2024 Aluminum Alloy Billet on Servo Press 823

step die motion as Fig. 11(b). This phenomenon is not usualin aluminum alloys because ductile fractured surface withdimples are generally observed after deformation even at R.T.Also temperatures of the two HFQ by the two-step diemotion at the second compression are below 320K near roomtemperature as shown in Fig. 5(b). To verify the mechanismof the intergranular fracture during the HFQ process by thetwo-step die motion, cross-sections of the same two speci-mens are shown in Fig. 12. Traces of liquid phases appeared

along grain boundaries are observed on the billet of the two-step die motion. It is supposed that voids are formed at grain-boundary triple junctions and propagate along grain bounda-ries leading to intergranular fracture especially on bulged sidesurface of the billet as a result of tensile stress as shown inFig. 11. The number of voids on the billet by the two-step diemotion is less than that of the one-step die motion. This maybe due to forming temperature difference. Second compres-sion is conducted at low temperature without liquid phase in

20μm500

20500

Low magnification High magnification(a)

(b)

μmμm

μm

Fig. 11 SEM images of fractured surface on the bulged side surface of the as-HFQed billets. (a) One-step (¦h/h0 = 6%) and (b) Two-step(¦h/h1 = 11%).

50 5

50

Low magnification High magnification

Liquid phase

Cavitation

5

Trace of Liquid phase

Cavitation

(a)

(b)μm μm

μmμm

Fig. 12 SEM images for voids at grain junctions on the cross-section of the as-HFQed billets. (a) One-step (¦h/h0 = 6%) and (b) Two-step (¦h/h1 = 11%).

J.-Y. Jeon, R. Matsumoto and H. Utsunomiya824

the two-step die motion. In other words, voids are closed bythe compression at the second step.

5.2 Hardness distributionsIn hardness distributions through the height, the two

as-HFQed billets show curved hardness distributions withhigher hardnesses near the center, while the billets by theone-step die motion show uniform hardness distributionsthrough the height. To reveal this phenomenon, TG-DTAanalysis was carried out at the surface and at the center of theas-HFQed billet with ¦h/h1 = 7% and that with ¦h/h1 =10% by the two-step die motion to compare kinetics ofprecipitation as shown in Fig. 13. The peak temperatures atthe surface of the as-HFQed billet with ¦h/h1 = 7% and thatwith ¦h/h1 = 10% are 536 and 534K, respectively, andthose of the center are 530 and 528K, respectively. The peaksat the center of the two as-HFQed billets are lower than thatof the surface. It is supposed that work hardening is moreconcentrated at the center rather than the surface due to thetemperature distribution and frictional constraint. Sufficientholding time may be useful to decrease temperature gradientand to obtain uniform hardness distributions. Better lubrica-tion maybe also effective. Also, the peak temperature is lowerwhen higher reduction is applied. Therefore, hardness nearthe center of the as-HFQed billet with ¦h/h1 = 10% ishigher than that of ¦h/h1 = 7% due to higher amount ofwork hardening as shown in Fig. 10(b).

5.3 Aging behaviorIt is well known that the hardness and the strength of

the alloy are affected by the precipitate morphology anddistribution developed during aging. The precipitationsequence of AA2024 aluminum alloy is reported inliterature16) as ¡ssss ¼ GPB zones ¼ SA ¼ S (CuMg2Al),where GPB zones stands for Guinier Preston Bagaryatskyzones and is an atomic arrangement of Cu and Mg atoms on{100}¡ of the Al matrix. For AA2024 aluminum alloy in theT6 state, initial strengthening is caused by GPB zones. At thestage of the peak aging, it is thought that the metastable phaseof SA, which is normally nano-orders with needle- or plate-like shape,15) is precipitated since it is the major strengthen-

ing precipitate in a commercial AA2024 aluminum alloy.After the peak aging, the hardness decreases by the phasetransformation of the strengthening precipitate to the equi-librium precipitate. Transformation to the equilibriumprecipitate occurs coupled with growth of larger precipitatesat the expense of finer ones as Ostwald ripening, reducingstrength.

It is found that the HFQ process accelerates theprecipitation kinetics and increases the peak hardness of anAA2024 aluminum alloy billet with h0 = 8mm after SHT at823K by strain aging due to higher dislocation densitiesintroduced during the process. The possible mechanism forthe acceleration is heterogeneous nucleation of precipitatesfrom segregated solutes on dislocations.9,17) In addition, it isfound that more effective acceleration and strengthening bythe two-step die motion is suggested to be linked with thehigher dislocation accumulation rate in the solutionizedmatrix and presence of higher density of fine precipitates inthe aged matrix.

6. Applications

The two-step process combined die quenching and pre-straining in single operation using die-motion control on aservo press has been proposed in this study. It shows that theprocess can improve its deformability in die quenching ofage-hardenable aluminum alloys. It leads the acceleration ofthe precipitation kinetics and increase the peak hardnesswithout adding another process. Therefore, the process canimprove the productivity in industries. This novel process isrealized by die motion control on a servo press.

However, the two-step process may not be always effectiveif applications to other alloys of higher deformability areconsidered. It would be better to apply only the one-stepprocess. In applications of the two-step process, such anAA2024 aluminum alloy billet is one of the most applicablematerials. In previous study, the authors pointed out that theHFQ process must be initiated above the precipitation-starttemperature of approximately 760K to avoid precipitationstart. Therefore, the HFQ process was conducted after theSHT at 823K to consider temperature drop caused by timeinterval for transfer from the electric furnace to the pressalthough the recommended SHT temperature for a com-mercial AA2024 aluminum alloy is approximately 763­773K.5,18) However, the higher SHT temperature causedvoids formation at grain-boundary triple junctions due topartial melting, then leading to intergranular fracture alonggrain boundaries when reduction applied over 6% in height.Therefore, only small reduction less than 5% in height wasapplicable to avoid intergranular fracture, and then thecompressed billet was further held between the dies forquenching. The temperature after the first-step die quenchingis far below the solidus temperature close to room temper-ature so that further compression is sufficiently applied tothe billet up to 5% in height without any defects. Thus,the two-step process for die quenching is appropriate forsuch an AA2024 aluminum alloy billet because theimproved deformability can be obtained using microstructurecontrol by die motion control in single operation on a servopress.

520 540 560 580 600

Temperature, T / K

En

do

ther

mal

Exo

ther

mal

Hea

t fl

ow

/ n

Vs-1

50

Surface

Center

Δh/h1 = 7%

Δh/h1 = 10%

Δh/h0 = 7%

Δh/h0 = 10%

Fig. 13 TG-DTA results at the surface and at the center of the HFQedbillets by the two-step die motion at the heating rate of 10K/min.

Two-Step Die Motion for Die Quenching of AA2024 Aluminum Alloy Billet on Servo Press 825

In addition, servo press with flexible ram motion has beendeveloped to lead new forming processes such as reducingfriction in sheet forging, suppressing defects during extrusionagainst counter tool, supplying liquid lubricant throughinternal channel into the hole during extrusion, reducingspringback by holding of dies at the bottom dead center, andso on.19­22) It is notable that die motion control on servo presscan be a metallurgical means for microstructure control andproperty improvement though they have not been studiedwidely.

7. Conclusions

Die quenching of a cylindrical AA2024 aluminum alloybillet was feasible on a servo press by sandwiching withdies of WC­20mass%Co. However, the reduction waslimited less than 5% in height. In order to apply higherreduction, just after die quenching further compression wasconducted straightly on the same machine (two-step diemotion). Following remarks have been drawn in thisstudy:(1) The low deformability due to partial melting at grain-

boundary triple junctions is improved using the two-step die motion. The total reduction increases from thebillet with ¦h/h0 = 5% of the one-step to ¦h/h1 =10% by the two-step die motion.

(2) Temperature change and TG-DTA results show thatthe two-step die motion does not cause precipitationhardening during the process. The hardness is highernear the center of the die-quenched billets by the two-step die motion. The distribution reflects work harden-ing introduced during the second step mainly concen-trated at the center.

(3) As the pre-straining is realized in the second step, thepeak aging time and hardness of the HFQed billet with¦h/h1 = 10% by the two-step die motion are 21.6 ksand 150HV, respectively, while those by the one-stepdie motion are 32.4 ks and 145HV. Therefore, the two-step die motion is more effective to accelerate theprecipitation kinetics and increase the peak hardnessthan that of the one-step die motion.

Acknowledgment

The authors are grateful to the Light Metal EducationalFoundation, Inc. for material supply.

REFERENCES

1) R. P. Garrett, J. Lin and T. A. Dean: Adv. Mater. Res. 6­8 (2005) 673­680.2) L. Wang, M. Strangwood, D. Balint, J. Lin and T. A. Dean: Mater. Eng.

Sci. A 528 (2011) 2648­2656.3) M. S. Mohamed, A. D. Foster, J. Lin, D. S. Balint and T. A. Dean: Int.

J. Mach Tools. Manuf. 53 (2012) 27­38.4) T. Maeno, H. Fujii, K. Mori and M. Sato: Steel Res. Int. 81 (2010) 354­

357.5) J. Y. Jeon, R. Matsumoto and H. Utsunomiya: Proc. THERMEC’2013.

in press.6) J. Y. Jeon, R. Matsumoto and H. Utsunomiya: J. Mater. Process.

Technol. (2013) submitted.7) A. Deschamps, F. Livet and Y. Bréchet: Acta Mater. 47 (1999) 281­292.8) H. L. Lee, W. H. Lu and S. Chan: Scr. Metall. Mater. 25 (1991) 2165­

2170.9) P. Gomiero, A. Reeves, A. Pierre, F. Bley, F. Livet and H. Vichery:

Proc. 4th Int. Conf. on Aluminum Alloys, ed. by T. H. Sanders andE. A. Starke, (1994) pp. 644­651.

10) R. S. Yassar, D. P. Field and H. Weiland: Metall. Mater. Trans. A 36(2005) 2059­2065.

11) W. J. Kim, J. K. Kim, T. Y. Park, S. I. Hong, D. I. Kim and Y. S. Kim:Metall. Mater. Trans. A 33 (2002) 3155­3164.

12) K. Osakada, K. Mori, T. Altan and P. Groche: CIRP Annals-Manuf.Technol. 60 (2011) 651­672.

13) R. Matsumoto, Y. Osumi and H. Utsunomiya: Steel Res. Int. (2012)Special Edition, 207­210.

14) Th. Herding, O. Kessler, F. Hoffmann and P. Mayr: Mater. Sci. Forum.396 (2002) 869­874.

15) H. C. Shin, N. J. Ho and J. C. Huang: Metall. Mater. Trans. A 27 (1996)2479­2494.

16) O. Novelo-Peralta, I. A. Figueroa, G. Lara-Rodriguez and G. Gonzalez:Mater. Chem. Phys. 130 (2011) 431­436.

17) Y. Birol: Scr. Mater. 52 (2005) 169­173.18) Product and Manufacture of Aluminum Alloy, Japan Inst. Light Metals,

Tokyo, (1991) p. 283.19) T. Maeno, K. Osakada and K. Mori: J. JSTP 50 (2009) 951­955.20) K. Osakada, R. Matsumoto, M. Otsu and S. Hanami: CIRP Annals-

Manuf. Technol. 54 (2005) 245­248.21) R. Matsumoto, S. Sawa, H. Utsunomiya and K. Osakada: CIRPAnnals-

Manuf. Technol. 60 (2011) 315­318.22) K. Mori, K. Akita and Y. Abe: Int. J. Mach. Tools Manuf. 47 (2007)

321­325.

J.-Y. Jeon, R. Matsumoto and H. Utsunomiya826