the effect of thermomechanical processing on the tensile

The Effect of Thermomechanical Processing on the Tensile,Fatigue, and Creep Behavior of Magnesium Alloy AM60

Z. CHEN, J. HUANG, R.F. DECKER, S.E. LEBEAU, L.R. WALKER, O.B. CAVIN,T.R. WATKINS, and C.J. BOEHLERT

Tensile, fatigue, fracture toughness, and creep experiments were performed on a commerciallyavailable magnesium-aluminum alloy (AM60) after three processing treatments: (1) as-THI-XOMOLDED (as-molded), (2) THIXOMOLDED then thermomechanically processed(TTMP), and (3) THIXOMOLDED then TTMP then annealed (annealed). The TTMP pro-cedure resulted in a significantly reduced grain size and a tensile yield strength greater than twicethat of the as-molded material without a debit in elongation to failure (ef). The as-moldedmaterial exhibited the lowest strength, while the annealed material exhibited an intermediatestrength but the highest ef (>1 pct). The TTMP and annealed materials exhibited fracturetoughness values almost twice that of the as-molded material. The as-molded material exhibitedthe lowest fatigue threshold values and the lowest fatigue resistance. The annealed materialexhibited the greatest fatigue resistance, and this was suggested to be related to its balance oftensile strength and ductility. The fatigue lives of each material were similar at both roomtemperature (RT) and 423 K (150 �C). The tensile-creep behavior was evaluated for appliedstresses ranging between 20 and 75 MPa and temperatures between 373 and 473 K (100 and200 �C). During both the fatigue and creep experiments, cracking preferentially occurred atgrain boundaries. Overall, the results indicate that thermomechanical processing of AM60dramatically improves the tensile, fracture toughness, and fatigue behavior, making this alloyattractive for structural applications. The reduced creep resistance after thermomechanicalprocessing offers an opportunity for further research and development.

DOI: 10.1007/s11661-010-0478-x� The Minerals, Metals & Materials Society and ASM International 2010

I. INTRODUCTION

MAGNESIUM (Mg) alloys are lightweight structuralmaterials that can exhibit high specific strengths. As theyare approximately 75 pct lighter than steel and approx-imately 35 pct lighter than Al alloys, Mg alloys areconsidered to have promising applications in many fieldswhere weight reduction is important, including armor,portable electronic devices, biomedical devices, automo-biles, and aerospace components.[1–4] THIXOMOLDING*

is a semisolid injection molding method that originatedfrom the semisolidmetal casting developed in the 1970s.[2]

Compared with traditional die-casting, THIXOMOLD-ING has many advantages, such as decreased turbulenceand lower temperature in the molding process, which canlead to lower porosity and an associated increase inelongation to failure (ef), as well as lower shrinkage and,thus, higher dimensional stability.[1–3] The THIXOMOL-DEDmicrostructure can be altered through postprocess-ing treatments to tailor the mechanical properties forspecific applications. This study focused on the effect ofthermomechanical processing on the ambient and ele-vated-temperature [423 K (150 �C)] tensile, fatigue, andcreep properties of THIXOMOLDED magnesium-alu-minum alloy (AM60). In addition, room temperature(RT) fracture toughness experiments were conducted. Allthe experiments were conducted in a direction parallelwith the molding and rolling plane.

II. EXPERIMENTAL

The measured bulk chemical composition of theas-molded AM60 material is shown in Table I. TheAM60 alloy was THIXOMOLDED into plates thatwere 3.2-mm thick,[5] and this was considered theas-molded condition. Some of the plates were thenthermomechanically processed (TTMP) in order toinduce dynamic recrystallization, and this represented

Z. CHEN, PhD Student, and C.J. BOEHLERT, AssociateProfessor, are with Michigan State University, East Lansing, MI48824. Contact e-mail: [email protected] J. HUANG, TechnicalDirector, R.F. DECKER, Chief Technical Officer, and S.E. LeBEAU,President, are with Thixomat, Incorporated, Ann Arbor, MI 48108.L.R. WALKER, Senior Technical Staff in the Microscopy Group, andT.R. WATKINS, Senior Research Staff Member, are with theMaterials Science and Technology Division, Oak Ridge NationalLaboratory, Oak Ridge, TN 37831. O.B. CAVIN, Research Associate,is with the Materials Science and Technology Division, Oak RidgeNational Laboratory, and also Assistant Research Professor with theCenter for Materials Processing Department of Materials ScienceEngineering, University of Tennessee, Knoxville, TN 37996.

Manuscript submitted February 25, 2010.Article published online December 24, 2010

*THIXOMOLDING and THIXOMOLDED are trademarks ofThixomat, Ann Arbor, MI 48108.

1386—VOLUME 42A, MAY 2011 METALLURGICAL AND MATERIALS TRANSACTIONS A

the TTMP condition. Several of the TTMP plateswere then annealed, and this represented the annealedcondition.

Metallographic samples were diamond cut from thedifferently processed materials and mounted in an epoxyresin. The three sample orientations (face, longitudinal,and transverse, Figure 1) were characterized. They werethen polished using silicon carbide paper and diamondpaste to a final finish of 0.25 lm using ethanol as alubricant. Microstructure analysis was performed usingscanning electron microscopy (SEM) and transmissionelectron microscopy (TEM). SEM analysis was per-formed using either a CamScan44FE field (Waterbeach,Cambridgeshire, UK) emission scanning electron micro-scope or a JEOL** 6500 field emission scanning electron

microscope or a JEOL 7500F scanning electron micro-scope. TEM analysis was performed using a Tecnai 20transmission electron microscope at 200 kV. The TEMfoils were prepared by polishing the samples to athickness between 50 and 100 lm, then thinning themusing a model 1010 Fischione (Export, PA) ion miller/polisher at 3 keV and 3 to 5 mA at a 15 deg incidentangle until perforation at RT. The volume fraction ofthe phases was measured using Image J (Bethesda, MD)software on several backscattered electron (BSE) SEMimages taken of each orientation of the plates. Fracturesurfaces from failed samples were examined using SEM.Microprobe analysis was conducted on mounted andpolished samples using a JEOL JXA-8200 wavelengthdispersive/energy dispersive (WD/ED) combined micro-analyzer operated at 10 kV and a beam current of 46 nAto determine the weight percentages of Mg, Al, Mn, andSi in the individual phases. The low accelerating voltagewas chosen to minimize the overlap between the matrixand precipitate phases. The minimum level of detectableSi using the technique employed was estimated to be0.01 wt pct, while for Mg, Al, and Mn, the minimumdetectable level was estimated to be 0.1 to 0.2 wt pct.X-ray diffraction (XRD) 2h scans were collected from 30to 80 deg using a rotating anode powder-texture-stress

goniometer (Scintag, Inc., Sunnyvale, CA) with copperKa radiation. XRD pole figures on a 5 by 5 deg gridwere acquired using the same goniometer with a copperKa radiation over a tilt (chi) range from 0 to 75 deg.Tensile tests were performed at RT and 423 K

(150 �C) on each of the studied alloys with a strain rateof 10�3 s�1. All tests were performed along the longitu-dinal direction of the plates. Before the test, samples wereelectrodischarge machined (EDM) into a flat dogboneshape, then polished to remove the EDM recast layersusing 600 grit silicon carbide paper. The load wascontrolled using a servohydraulic testing system,[6] andthe strain was measured by an alumina-arm extensom-eter attached to the gage section of the sample. Thefatigue tests were performed at RT and 423 K (150 �C)in an air environment using the same servohydraulictesting system. The stress ratio was R = 0.1 and thefrequency was 5 Hz. The materials were tested underseveral applied stress levels, with the maximum stressranging from 50 to 175 MPa. A minimum of three testswere performed for each condition and runout wasconsidered to be one million cycles.Fracture toughness experiments were also performed

at RT using the same servohydraulic system. Single edgenotched tension (SENT) samples were EDM machined,and the geometry of the sample is shown in Figure 2,where width W = 12.5 mm, crack length a = 7 mm,thickness B = 3.2 mm for the as-molded material, andB = 1.6 mm for the TTMP and annealed materials.The longitudinal direction of the specimen was parallelto the longitudinal direction of the plates. The specimensurfaces were polished using silicon carbide paper toreduce the effects of surface defects. No fatigue pre-cracking was performed and the initial precrack (intro-duced through EDM) length a was measured usingdigital imaging. The samples were loaded in tension at a75 N/s loading rate for the TTMP and annealedmaterials and at 150 N/s for the as-molded material.The crack opening displacement was measured using anextensometer attached to the specimen. Two experi-ments were performed for each material.Conventional tensile creep tests were performed using

vertical load frames manufactured by Applied TestSystem, Incorporated (Butler, PA). Flat dogbone-shaped samples, containing a gage length of 25 mmand a gage width of 12 mm, were machined by EDMand polished to remove the recast layers before testing.Test temperatures ranged between 373 and 473 K (100and 200 �C), and targeted test temperatures weremaintained within ±3 K. Applied stresses rangedbetween 20 and 75 MPa. Strain was measured anddocumented throughout the test, by a linear variable

Fig. 1—Schematic illustrating the orientations of the plate.

Fig. 2—Illustration of the single edge notched tension sample.

Table I. Chemical Composition in Weight Percent of the

Studied AM60 Alloy

Al Mn Zn Fe Si Ni Be Mg

6.29 0.28 0.05 0.001 0.02 0.0009 0.0007 93.34

**JEOL is a trademark of Japan Electron Optics Ltd., Tokyo.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 42A, MAY 2011—1387

differential transformer on a high-temperature exten-someter. After a minimum strain rate was achieved,either the load or temperature was increased or the testwas stopped. Upon termination of the creep experi-ments, the samples remained under load while thetemperature was decreased in order to maintain thedeformed state of the sample. TEM and SEM wereperformed on the deformed creep samples.

In-situ tensile-creep experiments were conducted onthe annealed material. Flat dogbone-shaped samples,with gage dimensions of 3-mm width by 2.5-mm thick-ness by 10-mm length, were EDM cut. The specimenswere glued to a metallic platen and polished to a 1-lmfinish using an automatic polishing machine and ethanolas a polishing lubricant. These experiments were per-formed using a screw-driven tensile stage (built by ErnestF. Fullam, Inc., Clifton Park, NY) placed inside an Xl-30field emission gun (FEG) FEI SEM (FEI WorldwideCorporate Headquarters, Hillsboro, OR). Temperaturewas controlled using a constant-voltage power supply toa 6-mm-diameter tungsten-based heater located justbelow the gage section of the sample. An open-bath,closed-loop chiller was used to circulate distilled water atRT through copper tubes to prevent the tensile stagefrom overheating. A fine-gage K-type thermocouple wasspot-welded to the gage section of each sample. After thesample’s gage-section temperature reached the desiredcreep temperature, a 30-minute period was given tostabilize the thermal stress prior to applying load. Theload, which was measured using a 4448 N load cell, wasapplied at 3.7 N/s until reaching the desired creep stress.The tests were considered constant load where the stressfluctuation varied ±3 MPa. The displacement dataacquired during the experiments included that of thesample as well as the gripping fixtures. Thus, thedisplacement values reported do not represent the soledisplacement of the reduced gage section of the sample.Secondary electron SEM images were taken beforeloading and at periodic displacements throughout thecreep experiments without interrupting the experiment.The pressure in the SEM chamber never exceeded 10�6

torr, and therefore, oxidation did not detrimentally affectthe SEM imaging. Further details of this apparatus andtesting technique can be found elsewhere.[7–9]

III. RESULTS

A. Microstructure

Figures 3 and 4 illustrate BSE SEM images of the as-molded, TTMP, and annealed materials, while Figure 5illustrates bright-field TEM images of these materials.The dark matrix phase was the Mg-rich phase (a), whichhas a hexagonal structure, and this encompassed morethan 90 pct of the microstructure by volume. Themedium-contrast phase, which typically formed at thegrain boundaries of the a phase in all three materialconditions, was considered to be the Mg17Al12 bphase,[10] which has a cubic structure, and encompassedapproximately 6 to 10 pct of the microstructure byvolume. It is pertinent to note that after the annealing

treatment, the volume fraction of this phase decreasedfrom 0.10 to 0.06 and, therefore, it appeared that thephase elements tended to go into solid solution, as hasbeen observed previously.[11] The bright precipitateswere present throughout the microstructure in smallquantities. The measured volume fraction of theseprecipitates was approximately 1 pct in each material.These precipitates, which ranged in size from 50 nm to10 lm, were Mn rich and considered to be the Mn5Al8

Fig. 3—Backscattered electron images of the (a) as-molded,(b) TTMP, and (c) annealed materials. Note the lower volumefraction of the Mg17Al12 grain boundary phase present in the(c) annealed material.


with the hexagonal structure observed previously.[11–13]

Table II lists the average volume percents of the studiedmaterials, as measured from the BSE SEM images.

Table III lists the microprobe data for the differentphases. It shows that the matrix a phase was enriched

with Mg, and the composition of the brightest phase wasconsistent with Mn5Al8. It is noted that all the reportedmeasurements for the Mn5Al8 phase were taken from

Fig. 4—Backscattered electron images of the annealed material illus-trating the difference between the fine Mn5Al8 precipitates (bright)and the Mg17Al12 (gray) phase.

Fig. 5—Representative bright-field TEM images of the (a) as-mol-ded, (b) TTMP, and (c) annealed materials. The fine dark precipi-tates in these images were suggested to be the Mn5Al8 phase.


the largest precipitates, which were on the order of10 lm in diameter. However, previous research on anAM50 alloy showed that nanoparticles were also presentwith this composition.[11] Thus, it is likely that thenanoparticles in the current work (dark precipitates inFigures 5(b) and (c)) were also Mn5Al8. The measure-ments of the grain boundary phase were not included,because they did not show consistency with Mg17Al12,and this was suggested to be a result of the interactionvolume encompassing both this grain-boundary phaseand the matrix a phase.

For the as-molded sample, the average equiaxed graindiameter was approximately 10 lm. However, there wasa bimodal grain size distribution as some of the a solidparticles remained solid in the thixomolder during theTHIXOMOLDING operation, and these resulted ingrains significantly larger than 10 lm. The grain in theright-hand side of Figure 3(a) represents a grain thatformed from such a particle. For the TTMP sample, thegrain size distribution was more homogeneous than thatfor the as-molded condition, and the average grain sizewas estimated to be less than 5 lm. For the annealedsample, the grain size was similar to that of the TTMPmaterial. As mentioned earlier, the annealing treatmentreduced the amount of the Mg17Al12 grain boundaryphase in the material, as there was much less contrastapparent near the a-phase grain boundaries.

Figure 6 illustrates the XRD intensity vs 2h plots forthe face of the as-molded, TTMP, and annealed plates.For the as-molded condition, there was a strong peakfor the (10-11) plane and lower peaks for the (0002) and(10-10) planes. This is representative of a material with arelatively random texture. For the TTMP and annealedconditions, the peak for the (0002) basal plane wassignificantly higher than all the other peaks. Althoughnot shown, this was verified by the XRD pole figures ofthe AM60 alloys. Thus, the TTMP process resulted in atexture in which the basal planes were parallel with theplate face, which is common in worked Mg alloys.[14,15]

The basal plane texture increased slightly in the

annealed material (Figure 6). The texture along withthe finer grain size associated with the TTMP andannealed materials resulted in a significant enhancementin the tensile and fatigue strengths, as will be presentedbelow.

B. Properties and Deformation Behavior

1. TensionThe average RT and 423 K (150 �C) tensile properties

along the longitudinal direction of the plates, includingyield strength (YS), ef, and ultimate tensile strength(UTS), are listed in Table IV. At RT, the UTS of theTTMP material was approximately 1.8 times that of theas-molded material, and the UTS of the annealedmaterial was approximately 1.5 times that of the as-molded material. The ef of the as-molded and TTMPalloys was approximately 5 pct, while the annealedmaterial showed a significant increase in ef (>19 pct). At423 K (150 �C), the UTS of the TTMP material wasapproximately 1.6 times that of the as-molded material,and the UTS of the annealed material was approxi-mately 1.3 times that of the as-molded material. The423 K (150 �C) ef of the as-molded and TTMP materialswere similar, while the annealed alloy showed a signif-icantly greater ef. The greater ef values exhibited by theannealed material could be rationalized if the annealingtreatment resulted in a lower dislocation density com-pared to that for the TTMP material. This would also beconsistent with the lower strength exhibited by theannealed material compared to the TTMP material.Compared to RT, each of the studied materials exhib-ited a lower YS and UTS, but a significantly higher ef at423 K (150 �C).Based on the stress-strain relationship, RT strain

hardening exponents were measured and the followingvalues were obtained: n = 0.229 for the as-moldedcondition, n = 0.059 for the TTMP condition, andn = 0.183 for the annealed condition. This shows thedrastically different strain hardening behavior for thethree conditions, where the TTMP exhibited the greateststrain hardening and also the greatest UTS value. Thismay be expected since this material was the mostseverely worked without a subsequent anneal. The as-molded condition exhibited the least strain hardeningand the lowest UTS value.Figure 7 illustrates the fracture surfaces of represen-

tative RT tensile samples. Ductile dimples were evidenton the annealed materials (Figure 7(c)) and less obviousin the as-molded and TTMP materials.

Table II. Average Phase Volume Percents of the Studied

AM60 Alloys (Standard Deviation in Parentheses)

Matrix (a) Mn5Al8 Mg17Al12

As molded 88.7 (1.8) 1.2 (0.3) 10.0 (1.7)TTMP 88.4 (1.4) 1.2 (0.2) 10.4 (1.3)Annealed 92.1 (1.7) 1.1 (0.3) 6.8 (1.9)

Table III. Phase Compositions in Weight Percent (Atomic Percent is in Parentheses)

AM60 Alloy Condition Phase Si Al Mn Mg

As molded matrix (a) 0 (0) 5.8 (5.3) 0 (0) 94.1 (94.7)Mn5Al8 0.7 (0.9) 38.8 (54.6) 57.8 (39.2) 2.7 (5.3)

TTMP matrix (a) 0 (0) 5.5 (5.0) 0 (0) 94.4 (95.0)Mn5Al8 0.6 (0.9) 39.1 (55.5) 59.7 (42.1) 1.2 (1.6)

Annealed matrix (a) 0 (0) 5.4 (4.9) 0 (0) 94.6 (95.1)Mn5Al8 0.7 (1.0) 38.5 (53.8) 56.7 (39.0) 4.0 (6.2)


C. Fatigue

The fatigue maximum applied stress vs cycles-to-failure (Nf) plots are illustrated in Figure 8. At each ofthe given maximum stress levels, the annealed materialexhibited the largest Nf value at both RT and 423 K(150 �C), and the as-molded material exhibited thelowest average Nf values. Thus, the annealed materialexhibited the greatest fatigue resistance, and the reasonfor this was considered to be related to its combinationof tensile strength and ductility. At both RT and 423 K(150 �C), the fatigue limit was suggested to be between50 and 75 MPa for the as-molded material, between 100and 125 MPa for the TTMP material, and between 125and 150 MPa for the annealed material.Figure 9 illustrates SEM images of the polished

subsurface face section of the 423 K (150 �C) fatigue-deformed samples. Grain boundary cracking was ob-served for this material, which has also been notedpreviously during in-situ experiments of die-castAM60.[16] Dislocations formed within the a-phasematrix for the annealed material (Figure 10). Thedislocations appeared to bow around the fine Mn5Al8particles. Thus, these fine particles were expected to havestrengthened the a-matrix, by acting as a barrier todislocation motion. No phase instability was observedduring the RT or elevated-temperature fatigue expo-sures. Surface crack initiation was principally observedfor each of the AM60 materials. Figure 11 illustrates asurface crack initiation site for an annealed sample. TheEDS data gathered at this location indicated thatunusually high levels of oxygen and silicon were present.Thus, this crack nucleation site may have been at aninclusion.

D. Fracture toughness

The fracture toughness was calculated according tothe following equation:[17]

fa

W

� �¼ KQB

ffiffiffiffiffiW

p

PQ

where, for the SENT geometry,

fa

W

� �¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2 tan pa

2W

pcos pa

2W

� 0:752þ 2:02a

W

� �þ 0:37 1� sin

pa2W

� �3� �

Table IV. Average RT and 423 K (150 �C) TensileProperties

0.2 Pct YS(MPa)

ef(Pct)

UTS(MPa)

As molded, RT 131 5.7 204TTMP, RT 331 3.8 367Annealed, RT 227 19.4 302As molded, 423 K (150 �C) 119 7.1 186TTMP, 423 K (150 �C) 261 9.0 297Annealed, 423 K (150 �C) 194 >27.4 250

Fig. 6—XRD intensity vs 2h plots for the face section of the AM60:(a) as-molded, (b) TTMP, and (c) annealed plates.


The term PQ is the critical load when fractureoccurred and was determined from the load-displace-ment relationship according to ASTM E 399.[18]

Table V summarizes the fracture toughness data.Although the Pmax/PQ ratio exceeded 1.10, the condi-tional fracture toughness KQ was calculated based onthe PQ value. The KQ value was then used to estimatethe minimum thickness of the sample necessary for a

valid plane-strain fracture toughness test, based on therelationship suggested by ASTM E 399:

B; a;W� a>2:5KQ

rys

� �2

As shown in Table V, the actual sample thickness wassignificantly smaller than that necessary for achievingplane-strain conditions.

E. Creep

The creep behavior resembled that typical for puremetals indicating three stages of creep. Creep strain vstime curves for the studied AM60 alloys tested undertwo different applied stress levels at 423 K (150 �C) areshown in Figure 12. The minimum creep rate vs stressplots for the studied AM60 alloys are shown inFigure 13. The creep stress exponents (n) can becalculated based on these plots, and the n values at423 K (150 �C) were 4.9, 5.8, and 4.5 for the as-molded,TTMP, and annealed materials, respectively. Figure 14

Fig. 8—Maximum applied stress vs cycles-to-failure curves for thestudied AM60 alloys at (a) RT and (b) 423 K (150 �C).

Fig. 7—Secondary electron images of RT tensile fracture surfaces ofAM60 (a) as-molded, (b) TTMP, and (c) annealed samples.


shows the temperature dependence of the minimumcreep rate under 20 MPa applied stress, and the appar-ent activation energy (Qapp) was calculated based on thisplot. The plot showed a two-stage linear relationship.Between 373 and 423 K (100 and 150 �C), the Qapp

values for the as-molded, TTMP, and annealed mate-rials were 48, 64, and 67 kJ/mol, respectively. Between423 and 473 K (150 and 200 �C), the Qapp values for theas-molded, TTMP, and annealed materials were 128,174, and 126 kJ/mol, respectively. Similar to thatexhibited in fatigue, the grain boundaries serve as cracknucleation sites in creep (Figure 15).The in-situ creep experiments highlighted the surface

grain boundary cracking (Figures 16 through 18). Itappeared that the grain boundary cracking was associ-ated with grain boundary sliding as surface fiducialscratches were jogged at grain boundary locations(Figure 17(b)). Thus, grain boundary cracking mayhave been accommodating grain boundary sliding.

IV. DISCUSSION

A. Microstructure

As noted in Table II, the volume fraction of theMg17Al12 phase was slightly smaller in the annealedmaterials compared with the other materials. It is likelythat dissolution of this phase occurred during the ele-vated-temperature annealing treatment and, therefore,the elements redistributed in solid solution within the aphase, as observed previously.[11] It is noted that theannealing temperature was considerably higher than thecreep temperatures, and redistribution of the Mg17Al12grain boundary phase did not occur during the creepexposures (i.e., the same volume fractions of theMg17Al12phase were present before and after the creep experi-ments). Thus, the AM60 microstructure is considered tobe stable for application temperatures of 423 K (150 �C)and below.

Fig. 10—Bright-field TEM image of an annealed sample fatigue de-formed at RT and a maximum applied stress of 175 MPa. It exhib-ited 36,183 cycles to failure. The Mn5Al8 precipitates appeared to bebarriers to dislocation motion.

Fig. 9—Backscattered electron images of 423 K (150 �C) fatiguedeformed AM60 samples: (a) as-molded, with maximum appliedstress of 75 MPa and 422,532 cycles to failure; (b) TTMP, with max-imum applied stress of 150 MPa and 30,182 cycles to failure; and(c) annealed, with maximum applied stress of 150 MPa and 67,079cycles to failure.


B. Tension

Although the studied materials exhibited lower tensilestrengths compared with most other commonly usedstructural alloys such as Ti alloys, Al alloys, and Nialloys, their specific strength was relatively high due totheir low density. At RT, the specific UTS values for theTTMP and annealed materials were 204 and 168 kNm/kg,

respectively. For reference, the specific UTS was178 kNm/kg for Ti-25Al-17Nb,[19] 67 kNm/kg for Alalloy 2024,[20] and 116 kNm/kg for Ni-Cr alloy 625.[21]

This suggests that this light-metal alloy comparesfavorably with other structural alloys on a density-normalized basis, which is important for the automotive,aerospace, sports and recreation, biomedical, elec-tronic, and other industries. Other stronger Mg-based

Fig. 11—(a) Low- and (b) high-magnification secondary electronimages of an annealed fatigue sample deformed at 423 K (150 �C)and a maximum applied stress of 150 MPa. It failed after 32,747cycles. The fatigue crack initiation was observed at the top surfaceof the specimen.

Table V. Summary of the Fracture Toughness Measurements

Condition Specimen PQ (N) Pmax (N) Pmax/PQ f (a/w) KQ (MPam0.5)Bmin for ValidKIC (mm)

As-molded A 1128 2276 2.01 4.61 15.0 32.6B 1322 2295 1.74 4.64 17.5 44.7

TTMP A 1111 1288 1.16 4.76 31.8 23.1B 1012 1328 1.31 4.61 27.9 17.8

Annealed A 1012 1614 1.59 4.59 27.4 36.4B 1021 1596 1.56 4.66 28.1 38.2

Fig. 12—Creep strain vs time plot of the studied materials at 423 K(150 �C): (a) r = 20 MPa and (b) r = 75 MPa.


alloys are also beginning to be considered for under-going the same TMP conditions, and these alloysalso have potential for implementation in structuralapplications.

C. Fatigue

The increased temperature did not significantly affectNf, and each processing treatment performed on theas-molded material resulted in greater fatigue strength.A previous study on a die-cast AM60 alloy showed thatthe fatigue limit was about 73 MPa.[22] This work hasshown that through TMP of THIXOMOLDED AM60,the fatigue limit can be almost doubled. This improve-ment is of similar magnitude as that involved in tensilestrengthening, and it suggests that the endurance limit:-UTS ratio in Mg alloys can be greater than 0.4. Thisratio is higher than that for Al alloys and approachesthat for Ti alloys. A balance between damage tolerance

Fig. 13—Minimum creep rate vs stress plot at T = 423 K (150 �C)for the studied AM60 alloys. The creep stress exponent (n) valuesare determined from the plot.

Fig. 14—Ln minimum creep rate vs reciprocal of temperature plotfor the studied AM60 alloys at r = 20 MPa. The apparent creepactivation energy (Qapp) values are determined from the plot.

Fig. 15—Backscattered electron images of the subsurface for creep-tested samples: (a) as-molded sample, tested at 20 MPa, 373 to473 K (100 to 200 �C), and total strain was 2.3 pct; (b) TTMP sam-ple, tested at 75 MPa, 423 K (150 �C), and total strain was 4.8 pct;and (c) annealed sample, tested at 20 MPa, 373 to 473 K (100 to200 �C), and total strain was 10.5 pct.


(ductility) and tensile strength appears to be an impor-tant consideration for maintaining high Nf values andfatigue thresholds for Mg alloys in the low-cycle fatigueregime.

D. Fracture Toughness

According to Table V, the SENT sample thicknesseswere well below those required for valid plane-strainfracture toughness condition. Thus, the KQ valuesshould not be interpreted as plane-strain fracturetoughness KIC. The KQ values for the TTMP andannealed materials were similar, and they were roughlydouble that of the as-molded materials. Thus, the TMPtreatment significantly improves the fracture toughness,and this was considered to be related to the significantlyhigher tensile strength achieved in the TTMP materialand the significantly high ef value achieved inthe annealed material as compared to the as-moldedmaterial.

E. Creep

The as-molded material exhibited the lowest creepstrain rates (Figures 12 through 14). One reason for thismight be that the TTMP and annealed materials bothhad smaller grain size than the as-molded material. Inaddition, the TTMP and annealed materials exhibited atexture in which the basal plane was parallel with thelongitudinal plate direction. This orientation favorsa slip, which could contribute to the enhanced creeprate. The TTMP material exhibited a lower minimumcreep rate than the annealed material at applied stressesof 20 and 50 MPa. However, this was not the case atan applied stress of 75 MPa. The reason for thiscrossover is not clear and will be the subject of furtherinvestigation.The measured creep exponents imply that dislocation

climb is the dominant mechanism controlling thesecondary creep rate. Similar creep exponents have beenreported for QE22 (Mg-2Ag-2Nd),[23] Mg-Y alloys,[24]

Mg-Zn-Zr alloys,[25] and die-cast AZ91D and AS21.[26]

Fig. 16—Secondary electron images of an annealed sample taken during an in-situ creep test at 50 MPa and 423 K (150 �C). The displacementswere (a) 0.168 mm, (b) 0.328 mm, (c) 0.579 mm, (d) 0.757 mm, (e) 0.945 mm, (f) 1.351 mm, (g) 1.730 mm, (h) 1.981 mm, and (i) 2.553 mm. Asignificant amount of grain boundary cracking was evident.


In both ingot and die-cast AZ91, creep mechanismsbased on dislocation motion (on basal and nonbasalplanes) were proposed,[27,28] where the ingot exhibited acreep rate one order of magnitude lower than the die-castalloy, which was proposed to be due to the larger grainsize in the ingot microstructure. During creep of Mg-5.6Y-0.04Zn (wt pct) (Mg-1.6 mol pct Y-0.015 mol pctZn), bowed out dislocations were observed to trailstraight dislocation segments parallel to the trace ofbasal planes.[24] The bowed-out dislocations were mov-ing on prismatic planes. The deformation observations inthe current work suggest grain boundary sliding alsocontributed significantly to the strain rates (Figures 15through 18), where grain boundary cracking may havebeen accommodating grain boundary sliding. Thus,these two creep deformation processes may be competingand the measured activation energy suggests that tem-perature may have an influence on this competition.

For temperatures above 423 K (150 �C), the activa-tion energy resembled that for lattice self-diffusion,[29]

while for temperatures less than 423 K (150 �C), theactivations were roughly half that for lattice self-diffusion. This suggests that grain boundary diffusiondominates at lower temperatures. The measured Qapp

values in low-temperature regime were in good agree-ment with previous measurements of particle-strength-ened Mg alloys.[25,26] Dargusch and Dunlop[26] reportedQapp values between 36 and 44 kJ/mol for creep ofAZ91D and AS21 and related it to the grain boundarysliding mechanism promoted by discontinuous precipi-tation of Mg17Al12 for which the activation energy is30 kJ/mol.[30] For the high-temperature creep regime[423 to 473 K (150 to 200 �C)], the measured Qapp

values were close to the 125 kJ/mol measured for high-temperature creep of pure Mg, and in the range of 92 to135 kJ/mol reported for the activation energy fordislocation glide in basal planes in pure Mg, which is

Fig. 18—(a) Low-magnification and (b) high-magnification second-ary electron images of the surface of an annealed sample after in-situcreep testing at 423 K (150 �C) and 75 MPa. The final displacementwas 3.127 mm. A significant amount of grain boundary cracking wasevident.

Fig. 17—(a) Low-magnification and (b) high-magnification second-ary electron images of the surface of an annealed sample after in-situcreep testing at 423 K (150 �C) and 50 MPa. The final creep dis-placement was 2.558 mm. Note this is the same sample as depictedin Fig. 16 and the fiducial scratches indicate offsets at a-phase grainboundaries, as indicated by the arrows in (b).


also the self-diffusion activation energy in the Mglattice.[29] In addition, others have measured Qapp valuesbetween 120 and 143 kJ/mol in the high-temperatureregime.[31] The grain boundary cracking and grainboundary sliding observations were made at the transi-tion temperature (423 K (150 �C)) between the high-temperature and low-temperature regimes. It is expectedthat the equiaxed a grain size is an important micro-structural feature, especially at temperatures of 423 K(150 �C) and below. Thus, the refinement caused by theTMP process would be expected to decrease the creepresistance. Thus, grain size, which did not significantlyinfluence the fatigue behavior, appeared to have agreater influence on the creep behavior. This is a strongconsideration for the implementation of TMP THIXO-MOLDED Mg alloys in creep-driven applications, suchas automotive engine applications, which are subjectedto elevated temperatures.

V. SUMMARY AND CONCLUSIONS

1. The TTMP treatment performed on the as-moldedAM60 alloy significantly reduced the grain size andresulted in a texture in which the (0002) basal planewas parallel with the plate face.

2. The TTMP process performed on the as-moldedAM60 alloy resulted in a significantly increasedstrength without a debit in ef. The annealed mate-rial exhibited an intermediate strength but the high-est ef. At 423 K (150 �C), each material exhibited alower strength but a higher ef than at RT.

3. The TTMP and annealed materials exhibited greaterfatigue strength both at RT and 423 K (150 �C) com-pared with the as-molded material. The annealedmaterial exhibited the greatest fatigue resistance, andthis was suggested to be related to its balance of ten-sile strength and ductility. No significant decrease inNf values occurred due to the increase in temperaturefrom RT to 423 K (150 �C). This work has shownthat through thermomechanical treatment of THI-XOMOLDED AM60, the fatigue limit can be almostdoubled. Surface fatigue crack initiation was ob-served in each AM60 condition.

4. Through thermomechanical treatment of AM60, thefracture toughness can be almost doubled as well.

5. Both the texture and finer grain size of the TTMPand annealed materials were suggested to beresponsible for the increased tensile and fatiguestrength compared with the as-molded material.Thus, thermomechanical processing of THIXO-MOLDED Mg alloys is a viable means to manufac-ture structural Mg alloys with exceptional tensileand fatigue properties as well as fracture toughness.

6. The creep resistance of the as-molded material wassuperior to the TTMP materials. The creep experi-ments indicated cracking preferentially occurred atgrain boundaries and grain boundary sliding. Thus,grain size was expected to be an important micro-structural parameter, and this partially explainswhy the creep resistance of the as-molded materialwas superior to that for the TTMP materials.

ACKNOWLEDGMENTS

This research was conducted, in part, through theOak Ridge National Laboratory’s High TemperatureMaterials Laboratory User Program, which is spon-sored by the United States Department of Energy, Of-fice of Energy Efficiency, and Renewable Energy,Vehicle Technologies Program, and through the OakRidge National Laboratory’s SHaRE User Facility,which is sponsored by the Division of Scientific UserFacilities, Office of Basic Energy Sciences, UnitedStates Department of Energy. A portion of this workwas supported by the Faculty and Student Teams(FAST) Program, which is a cooperative program be-tween the Department of Energy Office of Science andthe National Science Foundation. The authors aregrateful to Dr. Camden Hubbard, Oak Ridge NationalLaboratory, for assisting with the XRD characteriza-tion. The authors are also grateful to Messrs. BryanKuhr and Alex Ritter, Michigan State University, fortheir technical assistance with the SEM, XRD, andin-situ deformation characterization. This manuscripthas been authored by UT-Battelle, LLC, under Con-tract No. DEAC05-00OR22725 with the U.S. Depart-ment of Energy. The United States Governmentretains and the publisher, by accepting the article forpublication, acknowledges that the United States Gov-ernment retains a non-exclusive, paid-up, irrevocable,world-wide license to publish or reproduce the pub-lished form of this manuscript, or allow others to doso, for United States Government purposes.

REFERENCES1. S.E. LeBeau, M.W. Walukas, and R.F. Decker: Thixomat, Inc.,

Ann Arbor, MI; paper presented at Eurofond Congr., LaRochelle,France, 2004.

2. D.M. Walukas, R.F. Decker, R.E. Vining, and R.D. Carnahan:Report, Thixomat, Inc., Ann Arbor, MI.

3. R.F. Decker and S.E. LeBeau: Adv. Mater. Process., 2008,vol. 166, pp. 28–29.

4. S. LeBeau and D.M. Walukas: Thixomat, Inc., Ann Arbor, MI;paper presented at Plastics in Portable and Wireless Electronics,Mesa, AZ, 1999.

5. R.F. Decker, R.D. Carnahan, E. Babij, J. Mihelich, G. Spalding,and L. Thompson: Adv. Mater. Process., 1996, vol. 149, pp. 41–42.

6. G.A. Hartman and S.M. Russ: Metal Matrix Composites: Testing,Analysis and Failure Modes, ASTM, Philadelphia, PA, 1989,pp. 43–53.

7. C.J. Boehlert, C.J. Cowen, S. Tamirisakandala, D.J. McEldowney,and D.B. Miracle: Scripta Mater., 2006, vol. 55, pp. 465–68.

8. C.J. Cowen and C.J. Boehlert: Metall. Mater. Trans. A, 2007,vol. 38A, pp. 26–34.

9. J.P. Quast and C.J. Boehlert: Metall. Mater. Trans. A, 2007,vol. 38A, pp. 529–36.

10. C. Yan, R.X. Bai, Y.T. Gu, and W.J. Ma: J. Achiev. Mater.Manufact. Eng., 2008, vol. 31, pp. 398–401.

11. V.Y. Gertsman, J. Li, S. Xu, J.P. Thomson, and M. Sahoo:Metall.Mater. Trans. A, 2005, vol. 36A, pp. 1989–97.

12. O. Kulyasova, R. Islamgaliev, B. Mingler, and M. Zehetbauer:Mater. Sci. Eng., A, 2009, vol. 503, pp. 176–80.

13. R.M. Wang, A. Eliezer, and E.M. Gutman: Mater. Sci. Eng., A,2003, vol. 355, pp. 201–07.

14. A.D. Rollett and S.I. Wright: Texture and Anisotropy, CambridgeUniversity Press, Cambridge, United Kingdom, 1998, pp. 179–239.


15. G. Garces, E. Onorbe, P. Perez, I.A. Denks, and P. Adeva: Mater.Sci. Eng., A, 2009, vol. 523, pp. 21–26.

16. G. Chadha, J.E. Allison, and J.W. Jones: Metall. Mater. Trans. A,2007, vol. 38A, pp. 286–97.

17. T.L. Anderson: Fracture Mechanics: Fundamentals and Applica-tions, CRC Press, Inc, Boca Raton, FL, 1991, p. 709.

18. Annual Book of ASTM Standards, ASTM, Philadelphia, PA, 2000,vol. 03.01, pp. 431–61.

19. C.J. Boehlert, B.S. Majumdar, S. Krishnpurthy, and D.B.Miracle: Metall. Mater. Trans. A, 1997, vol. 28A, pp. 309–23.

20. N. Hort, H. Dieringa, S.T. Kumar, and K.U. Kainer: Magn.Technol.: Metall., Design Data, Applications, Springer, New York,NY, 2006, p. 317.

21. W. Betteridge: Nickel and Its Alloys, Halsted Press, New York,NY, 1984, p. 91.

22. S.A. Khan, Y. Miyashita, Y. Mutoh, and T. Koike: Mater. Sci.Eng., A, 2008, vol. 498, pp. 377–83.

23. R.L. Mordike and P. Lukac: Proc. 3rd Int. Magnesium Conf., TheInstitute of Metals, London, 1997, pp. 419–29.

24. K. Maruyama, M. Suzuki, and H. Sato: Metall. Mater. Trans. A,2002, vol. 33A, pp. 875–82.

25. C.J. Boehlert: J. Mater. Sci., 2007, vol. 42, pp. 3675–84.26. M.S. Dargusch and G.L. Dunlop: Magnesium Alloys and Their

Applications, Werkstoff-Informationsgesellschaft, Frankfurt,Germany, 1998, pp. 277–82.

27. M. Regev, E. Aghion, A. Rosen, and M. Bamberger: Mater. Sci.Eng., A, 1998, vol. 252, pp. 6–16.

28. M. Regev, E. Aghion, and A. Rosen: Mater. Sci. Eng., A, 1997,vol. 234, pp. 123–26.

29. G.E. Dieter: Mechanical Metallurgy, McGraw-Hill, New York,NY, 1986, p. 432.

30. H. Uchida and T. Shinya: J. Jpn. Inst. Light Met., 1995, vol. 45,p. 572.

31. S.S. Vagarali and T.G. Langdon: Acta Metall., 1982, vol. 30,pp. 1157–70.


the effect of thermomechanical processing on the tensile

Documents