METALLURGY Precipitation strengthening of aluminum alloys ...High-strength aluminum alloys are important for lightweighting vehicles and are extensively used in aircraft and, increasingly,

METALLURGY

Precipitation strengthening ofaluminum alloys by room-temperaturecyclic plasticityWenwen Sun1, Yuman Zhu1, Ross Marceau2, Lingyu Wang1, Qi Zhang1,Xiang Gao1, Christopher Hutchinson1*

High-strength aluminum alloys are important for lightweighting vehicles and areextensively used in aircraft and, increasingly, in automobiles. The highest-strengthaluminum alloys require a series of high-temperature “bakes” (120° to 200°C) toform a high number density of nanoparticles by solid-state precipitation. We foundthat a controlled, room-temperature cyclic deformation is sufficient to continuouslyinject vacancies into the material and to mediate the dynamic precipitation ofa very fine (1- to 2-nanometer) distribution of solute clusters. This results inbetter material strength and elongation properties relative to traditional thermaltreatments, despite a much shorter processing time. The microstructures formedare much more uniform than those characteristic of traditional thermal treatments anddo not exhibit precipitate-free zones. These alloys are therefore likely to be moreresistant to damage.

Solid-state precipitation is one of the mosteffective methods to strengthen engineer-ing alloys. This mechanism is found in allalloy classes but is particularly importantfor aluminumalloys. The high-strength alu-

minum alloys used in planes, trains, and auto-mobiles owe their strength to a fine distributionof precipitates formed by solid-state nucleationand growth. The key to the high strength is theformation of a high number density of closelyspaced precipitates that are able to resist dislo-cation motion.Ever since solid-state precipitation was iden-

tified as a strengthening mechanism for alu-minum alloys in 1906 (1, 2), metallurgists haveused elevated-temperature artificial aging treat-ments (a “bake”) (3) to induce precipitation andobtain the highest-strength states. For precipitatenucleation to proceed, an energy barrier (DG*)must be overcome (4). The driving force forprecipitation increases with increasing solutesupersaturation (Cb – Ceq, where Cb is the bulksolute content and Ceq is the equilibrium solutecontent; Fig. 1A) and consequently, the proba-bility of overcoming the barrier to nucleation[exp(–DG*/kT ), Fig. 1B] increases with decreas-ing temperature. However, precipitates usuallyhave a composition different from that of theoverall alloy, and their formation requires a localchemistry change. This change is mediated bydiffusion (D) and requires the presence of atomicvacancies whose concentration increases withincreasing temperature (Fig. 1B). As a result,themaximumprecipitate nucleation rate (dN/dt,

Fig. 1B) occurs at an elevated temperature, as acompromise between the thermodynamics andkinetics of precipitation. This temperature rangeis where aluminum alloys are processed. Forexample, alloys AA2024, AA6061, andAA7075 areartificially aged for 12 hours at 190°C, 8 hoursat 175°C, and 24 hours at 120°C (5), respectively,to induce the high number density of precip-itates that leads to their maximum strengths[T6 state (6)].Although this is the approach metallurgists

have taken for the past 100 years (3), the elevated-

temperature artificial aging consumes sub-stantial time and energy. The process itself is acompromise because temperature controls boththe driving force for nucleation and the rate ofsolute diffusion. We show that this compromisebetween the thermodynamics and kinetics ofprecipitation can be broken. We demonstratethat aluminum alloys can be processed at roomtemperature and in minutes, while obtainingstrengths and elongations equivalent to or ex-ceeding the thermally treated highest-strengthstate (T6).We apply an alternating stress (Fig. 1C)using an axial, servohydraulic fatigue machineto induce controlled microscale, back-and-forthdislocationmovement at room temperature. Thedragging of “jogs” on the dislocations createsvacancies (7) that facilitate diffusion at roomtemperature in aluminum. The repetitive cyclicdeformation of the material injects vacancies,and the diffusion occurring in the presence of ahigh driving force creates solute aggregates thatconsiderably strengthen the alloy. We refer tothis as cyclic strengthening (CS).We measured the stress-strain curves of CS-

processed, water-quenched (WQ), underaged(UA), and peak aged (PA-T6) states for alumi-numalloys AA2024, AA6061, and AA7075 (Fig. 2,A to C). The WQ state had no precipitation heattreatment. The uniform elongation as a functionof yield strength and ultimate tensile strength(UTS) is summarized in Fig. 2, D to I. We foundthat the thermally treated materials showed theexpected inverse correlation between strengthand elongation. The alloyswe strengthenedusingroom-temperature cycling (Fig. 1C) were just asstrong or slightly stronger than the PA-T6 states,but with 20 to 60% relative improvement inuniform elongation (Fig. 2, D to F). We processedthe CS samples at room temperature and kept

RESEARCH

Sun et al., Science 363, 972–975 (2019) 1 March 2019 1 of 4

1Department of Materials Science and Engineering, MonashUniversity, Clayton, VIC 3800, Australia. 2Institute forFrontier Materials, Deakin University, Geelong, VIC 3126,Australia.*Corresponding author. Email: [email protected]

Fig. 1. Comparison ofconventional precipita-tion treatment andcyclic strengthening(CS) approach.(A) Schematic phasediagram showing an asolid solution at hightemperature and a + btwo-phase field at lowertemperature (Cb, bulksolute concentration;Ceq, equilibrium soluteconcentration in a phase).(B) Schematic illustrationof the nucleation rate(dN/dt) and growth rate(dR/dt) of b phase as afunction of temperature[exp(–DG*/kT), probabilityof overcoming the nuclea-tion barrier; D, solutediffusivity]. (C) Cyclicstrengthening profile.

on April 17, 2020

http://science.sciencem

ag.org/D

ownloaded from

the cycling frequency low to avoid adiabaticheating. The sample temperature did not exceed26°C during stress cycling. Our results illustratethat controlled, room-temperature cyclic plas-ticity can result in high strength and excellentductility. This occurs without any net changein shape and without resorting to the artificialaging process.We found that the microstructures generated

from the CS process are very different from thoseformed by traditional thermal treatments (Fig. 3).The well-known fine distribution of solid-stateprecipitates characterizes the thermally treatedPA microstructures (Fig. 3, A to C). The CS mi-crostructures (Fig. 3, D to F) show some dis-location loops, but few other features can beseen at the 10- to 50-nm scale. The dislocationloops form as a result of the agglomeration ofvacancies injected into the material during theCS process (Fig. 1C). The dislocations strengthenthe alloys, but the low density of dislocationsrelative to precipitates in the thermally treatedalloys requires other microstructural featuresto explain the mechanical response of the CSalloys (Fig. 2).Wemade a low-angle annular dark-field scan-

ning transmission electron microscopy (LAADF-STEM)micrograph for the CS-processed AA2024alloy with atomic resolution to better under-stand the origin of the strengthening (Fig. 4A).Bright contrast in this type of image is the resultof a combination of atomic number contrast andstrain due to atomic deviations from a perfect-ly periodic lattice (8). We found an extremelyfine distribution of 1- to 2-nm solute clusters inthe alloy, which were associated with atomicdisplacements (blue arrows, Fig. 4, B and C).We used high-angle annular dark-field STEM(HAADF-STEM) to reveal the correspondingchemical distributionwithin the clusters (Fig. 4,D to G). We frequently observed clusters com-prising two or more Cu-rich columns (circledbright dots, Fig. 4, E to G). Dull spots in ourimages (yellow arrows) likely include the aggre-gation of vacancies. We should not expect theclusters to be the same as those observed duringconventional natural aging (GP or GPB zones).The clusters formed from the CS process occuras vacancies are continuously pumped into thesystem and the clusters are continuously beingsheared by mobile dislocations during forma-tion. This leads to variations in shape (favoringequiaxed), composition, and vacancy contentof the clusters relative to those observed duringtraditional natural aging. We made atom probereconstructions (Fig. 4, H and I) showing themarked increase in number density of clustersas a result of CS processing. This heterogeneoussolute distribution has a strong effect on dislo-cation motion and drastically strengthens thematerial. We observed similar structures inAA6061 and AA7075 after CS (fig. S5). The cyclicdeformation profile (Fig. 1C) is controlled so asto minimize the trapped dislocation densityintroduced. Consequently, the majority of dislo-cations present aremobile, and their continuousback-and-forth motion limits the opportunities

for heterogeneous nucleation on them to takeplace.The comparatively attractive elongation re-

sulting from the CS treatment (Fig. 2) high-lights important advantages of the CS process.The post-necking elongation behavior of theCS samples is generally better than that of thethermally treated WQ T6 state (Fig. 2, A to C).The CS samples are far better in this respect thanthe alloys air-quenched from the solution treat-ment temperature (8). This change in propertiesis because the micromechanisms of damageare quite different for the two microstructures(Fig. 3). We do not find this to be surprising,because a feature of traditionally precipitate-strengthened alloys is the presence of precipitate-free zones (PFZs) next to grain boundaries (9).The PFZs form from the annihilation of excessvacancies that are trapped (after quenching)during thermal processing, and the PFZsmay beexasperated by the formation of grain boundaryprecipitates (6). This results in a comparativelyweak zone (20 to 200 nm) adjacent to grainboundaries. During deformation, plasticity local-izes in the PFZs and compromises the ability of

the material to resist mechanical damage (elon-gation, fracture, fatigue) (10). The electrochemicalproperties, such as pitting corrosion and stresscorrosion cracking, are also detrimentally affectedby PFZs (11).The CS process does not introduce PFZs into

the alloy. Because the CS process (Fig. 1C) injectsvacancies into the system by dragging jogs ondislocations, and because these preferentiallymove wherever it is easiest, the process spatiallyhomogenizes the local strength at each pointin the microstructure. This creates an extremelyhomogeneousmicrostructure. The post-neckingelongation (Fig. 2) is one manifestation of thishomogeneity, but the ability of the material toresist other forms of damage (both mechanicaland electrochemical) should also be improved.The temperatures where the precipitate nucle-

ation rate passes through a maximum duringthermal processing also correspond to the tem-peratures where the precipitate growth rate isat amaximum (Fig. 1B). Nucleation and growthare always in competition, and this limits themaximum number density of precipitates thatcan form during thermal processing. During the

Sun et al., Science 363, 972–975 (2019) 1 March 2019 2 of 4

Fig. 2. Mechanical response of alloys subjected to traditional thermal treatments andthe CS approach. (A to C) Engineering stress strain curves [WQ, water-quenched; UA,underaged; PA, peak aged (T6); CS to different stresses at various frequencies]. (D to F)Correlation between yield strength and uniform elongation. (G to I) Correlation betweenUTS and uniform elongation. Shadowed regions in (D) to (I) represent the inversecorrelation between strength and elongation.

RESEARCH | REPORTon A

pril 17, 2020

http://science.sciencemag.org/

Dow

nloaded from

CS treatment, because dislocations are movingback and forth, they repeatedly shear soluteaggregations that form and limit their growth.This process allows for the formation of precip-itate number densities and sizes that are not

achievable using thermal processing alone.Indeed, the clusters adopt a steady-state sizeduring cyclic deformation (12). This steady-state size (Rss) can be estimated by consideringthe competition between growth and the shear-

ing by dislocation motion (8):

Rss ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2csWl2

Qf� Cb � Ceq

Cp � Ceq

sð1Þ

Sun et al., Science 363, 972–975 (2019) 1 March 2019 3 of 4

Fig. 3. Microstructures of thermally treated and CS alloys. (A to C) LAADF-STEM images of PA microstructures of AA2024 (A), AA6061 (B),and AA7075 (C). (D to F) LAADF-STEM images of CS microstructures of AA2024 (D), AA6061 (E), and AA7075 (F). The electron beamwas parallel to h001iAl in (A), (B), (D), and (E) and parallel to h110iAl in (C) and (F).

Fig. 4. Solute clusters in AA2024 before and after cyclic strengthening(CS). (A) Atomic-resolution LAADF-STEM image showing clustersin CS-treated sample viewed along h001iAl. Inset: The correspondingselected-area electron diffraction pattern reveals weak diffraction(indicated by arrows) from clusters. (B and C) Solute clusters enlargedfrom (A). Arrows denote atomic displacements from the perfectly periodiclattice. (D to G) Atomic-resolution h001iAl HAADF-STEM images showing

solute clusters in the CS sample. Each bright dot in the atomic-resolutionHAADF-STEM images is a Cu-rich atomic column. The arrows in(D), (F), and (G) indicate atomic columns containing vacancies.(H and I) Atom probe reconstructions showing Mg-Cu clusters containingnine or more solute atoms in samples with identical thermal histories,without (H) and with (I) the CS treatment, respectively. Individualclusters are colored differently.

RESEARCH | REPORTon A

pril 17, 2020

http://science.sciencemag.org/

Dow

nloaded from

where c is a constant ~0.1, s is the stress level,W is the atomic volume, l is the vacancy annihila-tion distance, Qf is the vacancy formation energy(13, 14), and Cp is the precipitate solute content.Using reasonable values (8), one obtains a steady-state size of 1 to 1.5 nm, which agrees well withFig. 4A and Hutchinson et al. (12).Although the CS process works extremely

well for aluminum alloys at room temperature,elevated temperature is likely required to use thisprocessing strategy for higher–melting pointalloys. In order for solute atoms to diffuse throughthe alloy, the atomsmust both find a vacancy andhave enough energy to jump into the vacant site.The CS process solves half this problem by in-jecting vacancies into the alloy.However, the hopinto the vacant site requires thermal activation.Thermal activation for aluminum alloys is noproblem, as room temperature is sufficient toovercome themigration barrier [~0.65 to 0.75 eV(13–15)]. However, the temperature will need tobe higher for higher–melting point metals suchas steels and Ti- and Ni-base superalloys.The CS process we implemented here used

axial tension-compression cycles to ensure excel-lent load control and were performed slowly toavoid adiabatic heating. One of the advantagesof the CS approach relative to traditional thermaltreatment is that it is comparatively fast and can

be made faster (several minutes) using highercycling frequencies. Modified ultrasonics areone obvious processing direction worth explor-ing. Ultrasonic vibration is already known toaccelerate diffusion in the solid state (16, 17),although this is largely due to the temperaturerise during the treatment. Other approachescapable of inducing controlled microscale cyclicplasticity are well worth consideration.

REFERENCES AND NOTES

1. A. Wilm, German patent DRP 244554 (1906).2. A. Wilm, Metallurgie 8, 223 (1911).3. P. D. Merica, R. G. Waltenberg, H. Scott, Heat Treatment of

Duralumin (National Bureau of Standards, 1919).4. D. A. Porter, K. E. Easterling, M. Sherif, Phase Transformations

in Metals and Alloys (CRC Press, 2009).5. C. R. Brooks, in ASM Handbook, Volume 4: Heat Treating

(ASM International, 1991), p. 823.6. I. J. Polmear, D. St John, J. F. Nie, M. Qian, Light Alloys:

Metallurgy of the Light Metals (Butterworth-Heinemann, ed. 5, 2017).7. G. Saada, Physica 27, 657–660 (1961).8. See supplementary materials.9. A. Kelly, R. B. Nicholson, Prog. Mater. Sci. 10, 151–391 (1963).10. A. K. Vasudévan, R. D. Doherty, Acta Metall. 35, 1193–1219

(1987).11. D. E. J. Talbot, J. D. R. Talbot, Corrosion Science and

Technology (CRC Press, ed. 3, 2018).12. C. R. Hutchinson, F. De Geuser, Y. Chen, A. Deschamps,

Acta Mater. 74, 96–109 (2014).13. A. Seeger, D. Wolf, H. Mehrer, Phys. Status Solidi B 48,

481–496 (1971).14. J. Bass, Philos. Mag. 15, 717–730 (1967).

15. T. W. Gullion, M. S. Conradi, Phys. Rev. B 36, 3880–3883 (1987).16. G. I. Eskin, Ultrasound Advances to Metallurgy

(Machinostroeinie, Moscow, 1975).17. O. V. Abramov, High-Intensity Ultrasonics (CRC Press, 1999).

ACKNOWLEDGMENTS

We thank the Monash Centre for Electron Microscopy (MCEM) forthe use of their transmission electron microscopes, DeakinUniversity’s Advanced Characterisation Facility, and A. Taylor ofDeakin University for assistance in performing the atom probeexperiments. Funding: Supported by the Australian GovernmentResearch Training Program Scholarship, Monash InternationalPostgraduate Research Scholarship, and Monash GraduateScholarship (L.W. and Q.Z.) and by Australian Research Councilgrants DP150101577 (C.H.) and DE170100307 (Y.Z.). Authorcontributions: The concept was developed by C.H.; W.S. and C.H.designed the experimental program; W.S. and Q.Z. performedall mechanical tests and designed the CS schedules; L.W. and C.H.derived Eq. 1; Y.Z. performed all atomic-resolution transmissionelectron microscopy; W.S. and R.M. performed the atom probetomography; X.G. performed preliminary conventional transmissionelectron microscopy. Competing interests: Authors declare nocompeting interests; Data and materials availability: All data areavailable in the main text or the supplementary materials.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/363/6430/972/suppl/DC1Materials and MethodsFigs. S1 to S6Tables S1 and S2References (18–24)

12 October 2018; resubmitted 29 December 2018Accepted 4 February 201910.1126/science.aav7086

Sun et al., Science 363, 972–975 (2019) 1 March 2019 4 of 4

RESEARCH | REPORTon A

pril 17, 2020

http://science.sciencemag.org/

Dow

nloaded from

Precipitation strengthening of aluminum alloys by room-temperature cyclic plasticityWenwen Sun, Yuman Zhu, Ross Marceau, Lingyu Wang, Qi Zhang, Xiang Gao and Christopher Hutchinson

DOI: 10.1126/science.aav7086 (6430), 972-975.363Science 

, this issue p. 972Sciencethermal methods. This method should also work for other alloy systems.quickly creates many very fine precipitates that have the same strengthening effect as those characteristic of traditionala processing method that relies on mechanical cycling by pushing and pulling on the alloys at room temperature. This

developedet al.strategy for doing this is through hours of high-temperature cycling to form precipitates in the alloy. Sun High-strength aluminum alloys are important for producing lightweight cars, trains, and airplanes. The traditional

Pushing and pulling for high strength

ARTICLE TOOLS http://science.sciencemag.org/content/363/6430/972

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2019/02/27/363.6430.972.DC1

REFERENCES

http://science.sciencemag.org/content/363/6430/972#BIBLThis article cites 17 articles, 0 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Science. No claim to original U.S. Government WorksCopyright © 2019 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of

on April 17, 2020

http://science.sciencem

ag.org/D

ownloaded from