Influence of temperature on cyclic stress response and fracture behavior of aluminum alloy 6061

Pergamon

Engineering Fracture Mechanics Vol. 56, No. 4, pp. 531-550, 1997

PII: S0013-7944(96)00102-6

© 1997 Elsevier Science Ltd Printed in Great Britain. All rights reserced 0Ol3-7944/97 $17.o0 + o.0o

INFLUENCE OF TEMPERATURE ON CYCLIC STRESS RESPONSE AND FRACTURE BEHAVIOR OF ALUMINUM ALLOY 6061 T. S. SRIVATSAN, S. SRIRAM and C. DANIELS Department of Mechanical Engineering, The University of Akron, Ohio 44325-3903, Akron, U.S.A. Abstract--The cyclic stress response and fracture characteristics of aluminum alloy 6061 was studied at different temperatures. The specimens were cyclically deformed using tension-compression loading under total strain-amplitude control. The alloy showed evidence of softening at all test temperatures. The degree of cyclic softening was observed to increase with an increase in test temperature. The presence of shearable matrix precipitates in the alloy results in a local decrease in resistance to dislocation movement, thereby causing a progressive loss of strengthening contribution. At the elevated temperatures, localized oxidation and embrittlement at the grain boundaries are promoted by the applied cyclic stress and play an important role in accelerating crack initiation and subsequent crack propagation. The fracture behavior of the alloy is discussed in terms of competing influences of intrinsic microstructural effects, deformation characteristics arising from a combination of mechanical and microstructural contributions, plastic strain amplitude and concomitant response stress, and test temperature. © 1997 Elsevier Science Ltd. All rights reserved.

1. INTRODUCTION A MATERIALproperty that is essential in designing and developing microstructures which have enhanced fatigue resistance is cyclic stress-strain response (CSSR). This term is used to describe the relationship between flow stress and cyclic strain, and is a useful aid in understanding the cyclic fatigue process. The cyclic stress-strain behavior of pure metals and their alloy counterparts have, during the last two decades, been the subject of several intensive investigations [115]. The cyclic fatigue process comprises the following discrete, yet related, phenomena involving

[161: (1) Cyclic plastic deformation of the material, which conditions it for crack (damage) initiation. (2) Initiation of one or more microscopic cracks (damage) and related early crack growth. (3) Growth and coalescence of the microscopic cracks (damage) to form one or more macroscopic cracks. (4) Propagation of the macroscopic cracks (damage) through the elastic and plastic regimes. (5) Final catastrophic failure. Plastic deformation at the crack tip accompanies all stages of propagation, irrespective of the test temperature and environment. Therefore, cyclic plastic deformation, which refers to the stage of early permanent deformation during fatigue, is important to both nucleation and propagation of cracks because the mechanisms governing these processes are related to the conjoint and mutually interactive influences of microstructure, test temperature, applied stress and nature of environment. Microstructure controls the deformation mode and, thereby, the subsequent crack (damage) initiation and propagation behavior. In order to transfer information generated in the laboratory to actual service conditions, a clear understanding of the deformation, damage and fracture mechanisms is essential. Over the years, the aluminum producers have successfully developed a multitude of aluminum alloys and tempers to meet the stringent demands of the aerospace and automotive industries. In fact, the sustained success achieved with the family of aluminum alloys and triedand-true design concept has provided much reluctance to the adoption of alternative materials [17, 18]. These alloys provide specific combinations of monotonic strength, stiffness, ductility, toughness, fatigue crack initiation resistance and fabricability. The relatively high strength-toweight ratio coupled with low cost, good surface finish, corrosion resistance and easy availability 531

532

T.S. SRIVATSANet aL

in a variety of forms, are additional factors that have made the family of aluminum alloys an ideal choice for a spectrum of critical and non-critical engineering applications. During the last three decades, a considerable number of experimental research studies has been undertaken to understand and improve the monotonic properties, cyclic fatigue resistance, fracture toughness and overall fracture behavior of these alloys through microstructural control achieved through alloy design and innovative processing techniques. However, only a few studies have been performed to comprehensively understand the conjoint influence of elevated temperature and microstructure on cyclic stress response, cyclic strain resistance, low-cycle fatigue life and fracture characteristics of aluminum alloys [19, 20]. Interest in cyclic stress and strain resistance, at elevated temperatures, is attributed to the usefulness of such information in understanding: (a) the mechanical working of metals, (b) damage initiation and damage propagation characteristics and (c) developing optimum microstructures for load-critical, fatigue-sensitive and temperaturecritical applications. In particular, an essential prerequisite for better understanding of the cyclic stress response characteristics, cyclic strain resistance, fatigue life and fracture behavior at elevated temperatures is an accurate determination of the rate-controlling damage processes that influence cyclic deformation under the appropriate combination of experimental (mechanical) conditions, temperature, environment and metallurgical condition (microstructure) of the material. Exposure to elevated temperatures can promote many thermally-assisted reactions such as: (l) formation and coarsening of the strengthening precipitates, and (2) a gradual deterioration of the microstructure and the development of microscopic defects. Super-position of fully-reversed strain cycling can exacerbate the kinetics of these reactions, which can jeopardize the long-term integrity of the component during service. Information regarding the high temperature low-cycle fatigue (LCF) behavior of aluminum alloys is relatively limited [19-21]. The preponderance of research studies have focussed on understanding the room-temperature fatigue and fracture behavior [22-29]. The objective of the present paper is to document temperature influences on cyclic stress response characteristics and fracture behavior of aluminum alloy 6061. The mechanisms of damage that influence cyclic deformation during fully-reversed total strain amplitude controlled cycling, at elevated temperatures, are examined. To provide a comprehensive understanding of the mechanisms of damage, the low-cycle fatigue (LCF) failure process is discussed in terms of the specific role of concurrent and mutually competitive influences of cyclic strain amplitude and resultant response stress, intrinsic microstructural effects, deformation characteristics of the microstructure, test temperature and macroscopic aspects of fracture. 2. MATERIAL AND EXPERIMENTAL TECHNIQUES 2.1. Material

The aluminum alloy 6061 used in this study was a product manufactured and provided by the Aluminum Company of America (ALCOA). The nominal chemical composition (in weight percent) of the alloy is given in Table 1. The alloy was investigated in the T651 condition (solution heat-treated, quenched, stress relieved and artificially aged to maximum strength). 2.2 Specimen preparation

Fatigue test specimens were precision machined from the wrought plate with the stress axis parallel to the longitudinal direction (rolling direction). The specimens were smooth and cylindrical in the gage section, which measured 6.25 mm in diameter and 25 mm in length. The lengthto-diameter ratio of the fatigue test specimen was chosen so as to ensure that it would n o t buckle under fully-reversed cyclic straining. To minimize the effects of surface irregularities and Table 1. Nominal alloy composition of aluminium alloy 6061 (in weight percent) Mg

Si

Mn

Cr

Fe

A1

1.00

0.60

0.28

0.20

Traces

Balance

Aluminumalloy 6061

533

ShortTransverse Direction (ST) Rolling Plane " ~ LongTransverse (LT) ~ Direction

~

~

l

r

Longitudinal, R°lling ection (RD)

"

] J Transverse -'~ ~ " Plane

Longitudinal Plane

Fig. 1. Triplanar optical micrograph showingthe grain structure of aluminum alloy 6061-T651 along the three orthogonaldirections.

534

T.S. SRIVATSAN et al.

(a)

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

(b)

Fig. 2. Optical micrograph showing the distribution of coarse constituent and intermediate size particles in: (a) the matrix, (b) along the grain boundary regions.

Aluminum alloy 6061

Fig. 8. Optical micrograph showing deformation structures in sample cyclically deformed at 150°C.

(a)

(b)

J

Fig. 9. Scanning electron micrograph showing intergranular cracking in sample cyclically deformed at 150'C: (a) along grain boundaries, (b) at grain boundary triple .junctions.

535

536

T.S. SRIVATSAN

et al.

Fig. 12. Scanning electron micrographs of the low-cycle fatigue fracture surface of sample deformed at ambient temperature, Nf = 1313 cycles, showing: (a) featureless crack initiation region, (b) microscopic cracks and voids, (c) region of early crack growth, (d) region of overload.

Aluminum alloy 6061

Fig. 13. Scanning electron micrographs of the low-cycle fatigue fracture surface of sample deformed at 75'C, Nf = 2224 cycles, showing: (a) region of crack initiation, (b) high magnification of (a) showing fine microscopic cracks and voids, (c) region of overload showing voids and dimples.

537

538

T.S. SRIVATSAN et al.

Fig. 14. Scanning electron micrographs of the low-cycle fatigue fracture surface of sample deformed at 150°C, Nr = 1117 cycles, showing: (a) region of crack initiation, (b) region of early crack growth showing microscopic cracks, (c) high magnification of (b), (d) region of overload.

Aluminumalloy6061

539

finish, final surface preparation was achieved by mechanically polishing the entire gage section of the test specimens using progressively finer grades of silicon carbide impregnated emery paper to remove all circumferential scratches and surface machining marks. 2.3. Mechanical testing The cyclic fatigue tests were performed on a fully-automated, closed-loop servohydraulic structural test machine (INSTRON) equipped with a 10 000 kg (98 kN) load cell. The load train was aligned prior to the initiation of testing to minimize specimen bending. The cyclic fatigue tests were conducted in the axial total strain-amplitude control mode under fully-reversed, push-pull, tension-compression loading. For each individual test the test machine was programmed to maintain a constant nominal strain rate of 0.001 s-~. The test command signal (strain function) exhibited a triangular waveform and the mean strain was zero. All cyclic fatigue tests were initiated in tension. A 12.7 mm gage length clip-on extensometer was attached to the test specimen at the gage section, using metal springs, to monitor the total strain amplitude during fully-reversed, strain-controlled fatigue tests such that the magnitude of negative strain equals the magnitude of positive strain (RE = E m i n / E m a x - 1). The controlled variable is total strain amplitude (&T/2). The extensometer was calibrated prior to the initiation of LCF testing. The tests were conducted in controlled laboratory air environment (RH = 55%) at both ambient (27°C) and elevated temperatures. The elevated temperatures chosen were 75°C, 100°C, 125°C and 150°C. The maximum test temperature (150°C) corresponding to the temperature at which the alloy was artificially aged to peak-strength. The elevated temperature tests were performed using an environmental chamber system unit (INSTRON Model 3111). The temperature was precisely controlled with the aid of a temperature controller linked to a thermocouple fixed to the surface of the specimen at the gage section. The elevated temperatures were within 2°C of the set-point over the test duration and the temperature gradient along the length of the test specimen was 2°C. Ambient temperature varied from 25°C to 27°C with a maximum 2°C variation during any given test. For the elevated temperature tests the specimens were soaked or exposed to the temperature for 30 mins prior to the initiation of testing. The stress-strain hysteresis loops were recorded on an X-Y recorder equipped with a pen plotter. The plastic strain range (&p) and the cyclic stress range (Sa) were periodically provided as a print out by the Control Console unit of the mechanical test machine. The strain-amplitude controlled low-cycle fatigue (LCF) tests were performed at a constant total strain amplitude (&T/2 = 0.60%). The tensile and compressive stresses were determined from the tips of the monitored hysteresis loops. The number of cycles to failure or separation is taken as the fatigue life (Nf). 2.4. Microstructural evaluation and damage analysis The initial microstructure of the as-received 6061-T651 material was characterized by optical microscopy after the standard metallography preparation technique. Fracture surfaces of the deformed low-cycle fatigue test specimens were comprehensively examined in a scanning electron microscope (SEM) to: (1) determine the macroscopic fracture mode and (2) characterize the fine-scale topography and microscopic mechanisms governing low-cycle fatigue fracture. The distinction between macroscopic and microscopic fracture mechanisms is based on the magnification level at which the observations are made. Samples for SEM observation were obtained from the deformed fatigue specimens by sectioning parallel to the fracture surface. 3. RESULTS AND DISCUSSION

3.1. Microstructure The microstructure of the as-received 6061-T651 alloy is shown in Fig. 1 as a triplanar optical micrograph illustrating the grain structure of the material in the three orthogonal directions of the wrought (rolled) plate. This material is fully recrystallized with fairly large recrystallized grains. No attempt was made in this study to determine the average grain dimension. The preEFM 56/4~C

540

T.S. SRIVATSAN et al.

sence of impurity elements in the alloy results in the precipitation of a high volume fraction of coarse insoluble iron- and silicon-rich constituents during casting. These large particles have been reported to be compounds of A17Cu2Fe and AII2(FeMn)3Si [30 and 31]. Their size ranges from 2 to 10 pm. The presence of the grain refining element chromium results in intermediate size dispersoids, which precipitate during ingot pre-heat and homogenization treatment[32]. The coarse constituent particles and insoluble magnesium-rich phase (A12CuMg) were stratified and distributed along the rolling direction of the wrought plate. At frequent intervals, clustering or agglomeration of the coarse constituent and intermediate size particles was observed [Fig. 2(a)]. The particles were also found decorating the grain boundaries [Fig. 2(b)]. Strengthening in the alloy is due to the magnesium silicide (Mg2Si) phase, which is the primary hardening precipitate formed during artificial aging of the alloy. A ratio of magnesium to available silicon of 1.7:1 ensures that all of the solute is contained in the Mg2Si phase[33]. The excess silicon in the alloy, over and above the amount required for the formation of the ordered Mg2Si phase, is deposited at the grain boundaries as elemental silicon.

3.2. Cyclic stress response The manner in which stress response varies with cycles (N) and plastic strain amplitude (&p/2), at a given temperature, is an important feature of the LCF process. The cyclic stress response curves, which were determined by monitoring the stress response (cyclic stress amplitude: 6a/2) during fully-reversed total strain amplitude-controlled fatigue, provide useful information pertaining to the mechanical stability of the material. Mechanical stability of the fine strengthening precipitates, and the coarse and intermediate size particles during fully-reversed cyclic straining coupled with an intrinsic ability of the material to distribute the plastic strain over the microstructural volume are two important factors controlling the stress response characteristics of the material, at a given test temperature. The stress response curve is a plot of cyclic stress amplitude (6tr/2) as a function of number of cycles (N), at a fixed total strain amplitude (&T/2), and illustrates the path by which the material arrives at its final flow stress level. The stress response curves of alloy 6061-T651 are shown in Fig. 3. It is interesting to note that at all test temperatures the material shows continuous softening to failure from the onset of fully-reversed cyclic straining. At a constant total strain amplitude (lET/2 = 0.60%), the overall degree of softening increased with an increase in test temperature, i.e. the softening at 150°C was greater than the softening observed at 100°C,

325 I~ET/2:0.60%

6061-T651

1

I--

• T-75 C [ ] T.125 C [ ] T=150 C 175

. . . . . . . .

10 0

i

. . . . . . . .

10 1

i

. . . . . . . .

10 2

i

10 3

CYCLES (N) Fig. 3. Cyclicstress

response

curvesfor alloy 6061-T651.

•

•

, , , , ,

10

4

Aluminumalloy 6061

541

AEr/2 = 0 , 6 0 %

325 1

6061.T651

J~ o--o

o

•--.--.--

""0

\

=" ~.

o r-27c • T-7S C r'l T-125C

I ] I 100 , 0

=Ix

o~

N~&,

, 20

, 40

, 60

, 80

\ ? 100

120

PERCENTAGE LIFE (N/Nf) Fig. 4. Effect of test temperature on cyclic stress response of alloy 6061-T651.

and the softening observed at 100°C was more pronounced than the softening observed at temperatures of 75°C and 27°C. Since fatigue life, for a given microstructural condition, is strongly dependent on both test temperature and experimental conditions (cyclic strain amplitude), a comparison of the stress response behavior, over the range of temperatures and at a constant total strain amplitude of 0.60%, is shown in Fig. 4. The stress amplitude (&r/2) during a LCF test is plotted as a function of fraction of life (N/Nf) to make comparison easier. It is evident from this figure that the degree of softening increases with temperature and is far more pronounced or noticeable at the higher test temperatures. The variation of normalized stress (trN/trn) with number of cycles (N) [Fig. 5] confirms the observation of an increase in softening with increase in test temperature. At a given temperature, the softening occurred by a progressive decrease in stress for both the tension and compressive parts of a LCF test (Fig. 6). This is easily distinguishable from the final crack1.050

6061-T651

~r/2 = 0 . 6 0 %

I

0.925

N =Z

[

O T=27G

O ;¢

•

T-75 C

I'1 "i"-125C •

0.800

T-150 C •

00

,

•

" ' ' " 1

10 1

•

•

'

" " " ' 1

'

10 2

'

' "

" ' " 1

'

"

'

' ' " '

10 3

CYCLES (N) Fig. 5. Variation of normalized stress with fatigue cycles for alloy 6061-T651.

10 4

542

T.S. SRIVATSAN et al.

325

At,t / 2 : 0 . 6 0 %

6061-T651 (T-27G)

A Cl ft.

=i

o~ u~ uu np. el)

f

250

O Termlon • Comlxes~on 175

. . . . . . . .

00

i

. . . . . . . .

10 1

i

. . . . . . . .

10 2

i

. . . . . . . .

10 3

10 4

CYCLES (N) 325

A~,2 =o.0o%

6o61 . TSS1 (T-75C)

A

.= 250

0 •

Tension Comweuion

175 ........ , 10 0 101

........

, ........ , ........ 10 2 10 3 10 4

CYCLES (N) 325 A6r/Z = 0 . 6 0 %

A ¢1 ft.

6061-T651 (T-150C)

250 '

I u) u} uJ fig }u~

175 O Ten=ion $

1 ~

. . . . . . . .

10 0

~

1

Compreulon

i

101

. . . . . . . .

!

. . . . . . . .

10 2

i

10 3

10 4

CYCLES (N)

Fig, 6. Stress response curves showing the variation of tensile stress and compressive stress with cycles at a fixed total strain amplitude: (a) 27°C, (b) 75°C and (e) 150°C.

Aluminum alloy 6061

543

related softening, which causes an asymmetric decrease in load for the tensile part of the cycle. The cyclic stress response curve can be considered to consist of three distinct stages: (a) rapid initial softening, (b) followed by progressive softening for most of fatigue life, and then towards the end of the fatigue test, (c) a rapid decrease of stress culminating in failure. The rapid decrease in stress is ascribed to the formation and presence of a number of microscopic cracks, their growth through the microstructure and eventual coalescence to form one or more macroscopic cracks. The softening effect is exacerbated by the concurrent growth of the macroscopic and fine microscopic cracks through the microstructure coupled with progressive destruction of the strengthening precipitates by the dislocation substructure. The cyclic plasticity, which is responsible for damage during fully-reversed cyclic straining, increases with test temperature as shown in Fig. 7. The increase in plastic strain amplitude (CSep/2)with temperature follows an exponential trend. 3.3. Mechanisms governing cyclic deformation and damage The intrinsic micromechanisms responsible for the stress amplitude decrease and fatigue softening are dependent on the independent or conjoint influences of microstructure, cyclic strain range and test temperature. The dominant strengthening mechanism for this aluminum alloy is precipitation hardening. Various mechanisms have been presented in the literature to explain the observed softening. The two key mechanisms which are widely accepted are[34, 35]: (1) Disordering of the strengthening precipitates by "mechanical scrambling". (2) Reduction in size of the strengthening precipitate due to repeated shearing. During fully-reversed cyclic straining the to-and-fro motion of dislocations through the precipitate lattice causes a "scrambling" of the atoms of the strengthening precipitates (Mg2Si). The general structure of the precipitate becomes disordered and the probability of dislocation cutting (shearing) creating different atom pairs is reduced. Disruption of the precipitate structure results in a progressive loss of contributions to strengthening with concomitant softening and the localization and/or accumulation of strain. The strain localization is either sufficient to cause structure disordering or becomes sufficient when accumulated over cycles. When the strengthening precipitates are sheared and aided by the to-and-fro motion of dislocations it causes the weakened plane to continue to slip and deformation becomes localized. At a given strain amplitude and concomitant response stress the softening becomes pronounced when the strengthening precipitates have been sheared on a sufficient number of planes to result

0.6 ~eT/2=0.60 %

P~

6061-T651

0.5

,= Z

0.4 O Im

-I O.

0,3

"

20

°

"

"

I

50

'

'

"

"

I

80

"

"

'

'

I

110

"

"

"

'

I

140

'

'

"

"

l

•

'

'

170

TEMPERATURE (C) Fig. 7. Variation of cyclic plastic strain amplitude with test temperature.

'

200

544

T.S. SRIVATSANet al.

in a progressive degradation of the original precipitation hardened microstructure within the persistent slip bands. Also, destruction of the strengthening precipitates causes the dislocations to pass on the active slip plane and the local work hardening capability is progressively reduced. The slip bands are, therefore, intrinsically softer than the alloy, in its undeformed condition, primarily because hardening contributions due to the second-phase particles are lost [36, 37]. Thus, a grain containing slip bands can be regarded as a composite material consisting of a strong matrix containing thin planar lamellae of softer material. Such a two-phase composite material has been proposed previously to describe the cyclic stress-strain curve of polycrystalline materials [38]. Since the two phases experience or are subjected to the same far-field cyclic stress, the respective strains generated are radically different. This results in severe deformation in the slip bands, with the matrix subjected to only elastic deformation. In the cyclically deformed material, the overall distribution of strain appears to be inhomogeneous, i.e. some areas of the specimen show evidence of extensive deformation (Fig. 8), while other areas are limited to no deformation. The concentration of slip in specific bands results in inhomogeneous deformation (Fig. 8). The localization of irreversible plastic strain within slip bands results in the build-up of dislocations at isolated points along the grain boundaries and at grain boundary triple junctions. The associated high stress concentrations due to the dislocation pile-up open intergranular cracks that propagate without absorbing much energy (Fig. 9). Let N be the number of dislocations that pass on the slip plane from the initiation of cyclic deformation until the end of local slip. The stress concentration at the grain boundary associated with the pile-up of N dislocations is T* = Nze, where ze is the resolved shear stress. When ~* exceeds the stress required for crack initiation at the grain boundaries, intergranular fracture occurs [36]. Presence of intergranular cracks increases with test temperature with a concomitant reduction in low-cycle fatigue life. Also, impingement of the slip bands on the grain boundary particles causes decohesion of their interface resulting in the initiation of microscopic intergranular cracks. These observations are consistent with those reported in the literature on the detrimental influence of stress concentrations associated with planar slip band deformation. Planar slip band impingement has been observed to cause intergranular cracking during different modes of deformation in several polycrystalline nickel-base alloys [39-44]. Hence, it is rationalized that the increased softening observed in alloy 6061-T651, at the higher test temperatures, results from the impingement of slip bands on grain boundaries and grain boundary particles, and the resultant formation of microscopic cracks. Also, the repeated shearing of the Mg2Si strengthening precipitates in slip bands sets up a preferential path for the to-and-fro motion of dislocations during continued deformation of the material. Repeated shearing of the precipitates leads to a reduction in their size to an extent that they offer little or no resistance to the movement of dislocations. Consequently, the stress necessary to shear the precipitates during repeated cycles is reduced, thus offering a resistance-free or precipitate-free path for the movement of dislocations. This phenomenon results in precipitate-free deformation bands and is largely responsible for the observed cyclic softening. The mechanical destruction of the precipitates and associated reduction in their size coupled with the presence and growth of microscopic and macroscopic cracks are viable explanations for the observed cyclic softening in alloy 6061-T651. It has for a long time been recognized that low-cycle fatigue (LCF) life is governed both by the initiation of fatigue cracks (damage) and by Stage II propagation. The initiation stage also consists of Stage I crack propagation along crystallographic slip planes, while Stage II propagation occurs, on an average, in a direction normal to the applied cyclic stress. The present study suggests that, at the given total-strain amplitude of 0.60%, the reduction in fatigue life at the higher test temperatures is mainly due to a decrease in the number of cycles spent in the initiation of fatigue cracks. Possible mechanisms contributing to the "damage" at elevated temperatures are summarized in Fig. 10. It is assumed that the 6061-T651 material has an intrinsic ability to sustain a certain amount of "damage" prior to failure. Such "damage" can be the result of any one or combination of these effects[45]: (1) deformation debris associated with the motion of dislocations, (2) metallurgical changes associated with coarsening and formation of new phases, and (3) interactions between (a) and (b).

Aluminum alloy 6061

545

Overall, the damage tolerance capability of the material depends on the damage accumulation mechanism. Other competing mechanisms contributing to damage during fully-reversed strain cycling are oxidation and embrittlement. The principal strengthening precipitates Mg2Si are ordered and thermodynamically stable for prolonged periods of time at a particular test temperature, and up to test temperatures that are much higher than the test temperatures used in this study. Considering the limited time duration of each LCF test, in comparison with the time actually required to bring about metallurgical transformations through the development of new phases and coarsening of the precipitate particles, this damage mechanism can be ruled out. Of the mechanisms of creep which can be unduly damaging when compared to the other deformation modes, the wave shape effect resulting from cycling with unbalanced hysteresis loops and the intrinsic micromechanisms which promote and enhance grain boundary degradation are important and need to be considered. Since aluminum alloy 6061 was subjected to balanced hysteresis loop cycling, i.e. equal ramp rates for each leg, damage from creep due to the effect of unbalanced cycling does not arise. Creep damage due to enhanced grain boundary degradation is also ruled out since fractographic observations of the samples deformed to failure at the different temperatures showed that the degradation in cyclic endurance at the higher test temperatures did not occur as a result of a drastic change in fracture mode from predominantly ductile transgranular to brittle intergranular failure. The occurrence of plastic strain inhomogeneity and aided by grain boundary particles is believed to be responsible for the intergranular cracking that is observed (Fig. 9). Besides, the short period of time associated with each total strain amplitude-controlled fatigue test, at the elevated temperatures, was insufficient to cause any significant time-dependent damage due to creep. The detrimental influences of water vapor during cyclic fatigue at ambient temperature and in a moist air environment are fairly well documented [46-51]. However, in elevated-temperature applications, the effect of oxidation is of primary concern and environmental damage due to this effect under balanced hysteresis loop cycling has been observed to be significant in several independent studies [52-54]. Oxidation is a time-dependent process and the two important parameters that control the rate of oxidation are: (a) availability of oxygen and (b) solid-state diffusion rates. The normal processes of crack initiation and growth that occur at ambient temperature (27°C) are accelerated in the hostile laboratory air environment at the higher test temperatures. The freshly exposed surfaces that are produced by local plastic deformation are rapidly oxidized and aid in localizing the strain, which results in further rupturing and localization thereby, reducing the time period normally associated with crack initiation. At times, the oxide completely bridges the surface cracks, which results in an increase in the intrinsic loadbearing capability of the specimen. Of the several different possibilities, the following factors are considered instrumental in accelerating fatigue damage and degrading the cyclic fatigue life of alloy 6061-T651 in the hostile laboratory air environment: (1) Absorption of oxygen from the environment and oxidation at and around the immediate vicinity of the crack tip. (2) Segregation of reactive elements in the environments, such as moisture and oxygen, to grain boundaries ahead of the propagating crack tip. (3) The inherent chemical complexity of grain boundaries which results in preferential oxidation along them and along the slip bands. Grain boundary oxidation ahead of an advancing or growing crack considerably weakens the boundary and promotes intergranular grain boundary crack propagation. (4) Cracking of the oxide at the crack tip under the influence of applied cyclic stress and strain, and the subsequent formation of oxide at the freshly exposed metal contribute to enhancing crack growth and a concomitant reduction in cyclic fatigue life. At the elevated temperatures, nucleating mechanisms arising from metal-environment interactions are significantly enhanced in the environment of laboratory air. Oxide formation and subsequent rupturing contribute to an increase in crack length, and hence a local increase in crack propagation rate with a concomitant reduction in cyclic endurance. A reduction in cyclic fatigue life arising from increased softening also results from deformation-environment interactions or embrittlement. A progressive embrittlement of the bound-

546

T.S. SRIVATSANet ~

al.

l~Nmm

• Taq~Dmes oCm~

s r,s m m s ~ o l~llnO s l r o s s - lnims slral~ o s m m s , s m ~ emmm~lJw!

• ~

o m~lallm o ~ m

~

c, i o o ~zlm sire

•

~

• ~lale

Fig. 10. Schematicshowingthe important variablesduring high temperature low-cyclefatigue. ary is favored since the microstructure of the alloy is such that the planar motion of dislocations during plastic deformation is not interrupted by the Mg2Si strengthening precipitates in the microstructure. As a result, innumerable edge dislocations pile-up at the grain boundaries, including those on the surface, and promote stress concentrations at the locations of these pileups. Oxygen, moisture and other reactive elements present in laboratory air, when absorbed onto the surface, reduce the surface energy, i.e. the energy required to form a new surface, resulting in a local strength that is lower than the stress required to initiate a crack. It seems plausible that under the influence of externally applied stress, the oxygen and/or other reactive elements diffuse rapidly along the grain boundaries to other locations of dislocation pile-up causing internal cracks to form. The embrittlement process is a complex interaction of precipitate structure, microstructure (grain size), mechanical properties: such as strength, nature of environment and test temperature. The embrittlement mechanism, exemplified in Fig. 11 shows a planar slip band intersecting with an oxidized grain boundary. The embrittlement of the grain boundary is a key factor, which contributes to the reduction in fatigue resistance experienced at the elevated temperatures together with the damage caused by cyclic plastic strain. The enhancement of both oxidation and embrittlement by means of the applied cyclic stress can result from any one or mutually competitive influences of the following factors: (1) (2) (3) (4)

Disruption of surface integrity by the emergence of dislocations. An increase in chemical potential of the atoms in a shear stress field. An enhancement of local atomic mobility as a direct result of dislocation concentrations. Expansion of the crack in tension, which provides room for the formation of oxides.

547

Aluminum alloy 6061

~amm

grala i m a m 7 -,,,¢

-.../

'g..

MogalI~ ~lo-w

s ~ ~,ad

Fig. 11. Schematic of the embrittlement mechanism showing the intersection of a persistent slip band with an oxidized grain boundary.

3.4. Cyclic fracture behavior The low-cycle fatigue fracture surface features of alloy 6061 were examined at low magnification to identify the fatigue and overload (final fracture) regions, and at higher magnifications in the fatigue region to identify regions of crack initiation and early crack growth. Samples for SEM were obtained from the deformed fatigue specimens by sectioning parallel to the fracture surface. Fracture surfaces revealed near-similar topographies at the different temperatures. Representative fractographs are shown in Figs 12-14. 3.4.1. Ambient temperature (27°C). The fracture surface of the specimen deformed at total strain amplitude of 0.60%, on a macroscopic scale, was essentially normal, i.e. at about 90 °, to the rolling plane and the major stress axis. Fracture was bimodal comprising regions of transgranular and intergranular failure. The region of crack initiation was essentially featureless [Fig. 12(a)]. High magnification observations of the crack initiation region revealed fine microscopic cracks along the grain boundaries and numerous voids, of varying size, distributed randomly through the surface [Fig. 12(b)]. The region of early crack growth, i.e. the region between crack initiation and overload surface, revealed pronounced cracking along the grain boundaries and isolated pockets of shallow near-equiaxed shaped dimples [Fig. 12(c)]. The intergranular cracks were parallel to the major stress axis. The tendency towards localized non-homogeneous (planar) deformation and the termination of the dislocation bands at grain boundaries results in the localization of strain with concomitant stress concentrations in these regions. This is exacerbated by the presence of undissolved iron-rich and silicon-rich intermetallics and the insoluble magnesium-rich constituent phases at and near the grain boundary regions, and is a plausible explanation for the observed intergranular fracture mode. The observed non-linearity (small deflections and bifurcations) of the microscopic and macroscopic cracks is attributed to slip planarity. The overload region revealed voids of varying size and shallow dimples [Fig. 12(d)].

548

T.S. SRIVATSANet al.

3.4.2. Temperature: 75°C. Fatigue fracture at this temperature revealed distinct regions of crack initiation and overload. The crack initiation region was relatively smooth and comprised numerous voids of varying size distributed uniformly through the surface [Fig. 13(a)]. Numerous fine microscopic cracks were observed in the direction parallel to the major stress axis [Fig. 13(b)]. High magnification observations also revealed voids of varying sizes and shapes [Fig. 13(c)] randomly distributed through the fracture surface. The overload region revealed a bimodal distribution of voids and dimples, features reminiscent of classical ductile failure [Fig. 13(c)]. 3.4.3. Temperature: 150°C. On a macroscopic scale fatigue fracture occurred along a plane inclined at 45 ° with respect to the stress axis, following a plane of maximum macroscopic shear stress. However, the fracture surface was relatively rough when viewed at the microscopic level. The region of crack initiation was transgranular and essentially featureless [Fig. 14(a)]. The region between crack initiation and overload revealed numerous fine microscopic cracks and a bimodal distribution of voids [Fig. 14(b)]. High magnification observations in this region revealed microscopic cracking along the grain boundary regions [Fig. 14(c)]. The overload region revealed numerous voids of varying size and shape [Fig. 14(d)]. The coarse iron-rich and silicon-rich constituent particles and traces of magnesium-rich insoluble phases tend to either crack easily on account of their intrinsic brittleness or decohere at their interfaces [55-57]. Interfacial strength is a dominant factor in the nucleation of the microscopic voids. Other factors which exert an influence on void initiation and growth are [57, 58]: (a) size of the second-phase particles, (b) particle shape, (c) location and volume fraction, and (d) intrinsic strength of the particle. Void initiation at the coarse constituent and other large second-phase particles in the microstructure is also influenced by local stress and strain levels, and the deformation mode. The presence of numerous coarse constituent particles and insoluble phase coupled with a matrix microstructure which promotes localized inhomogeneous deformation facilitates the nucleation of voids at relatively low to moderate stresses. In fact, void initiation at a coarse second-phase particle occurs when the elastic energy in the particle exceeds the surface energy of the newly-formed void surfaces [56, 57]. Simultaneously, the intermediate size chromium-containing dispersoids present in the microstructure decrease the energy required to propagate the crack by initiating voids that coalesce to form void sheets, thereby, linking the microcracks initiated at the grain boundary regions and the coarse constituent particles. The larger voids, which are created by fracture of the coarse and intermediate size constituent particles, coalesce by impingement. However, the more widely separated voids coalesce by the formation of void sheets. The halves of the fine microvoids are the shallow dimples visible on the fatigue fracture surface of alloy 6061-T651.

4. CONCLUSIONS Based on a comprehensive study of the influence of test temperature on cyclic stress response and fracture behavior of aluminum alloy 6061, the following are the key observations: (1) The microstructure of the as-received alloy was fully recrystallized with large recrystallized grains. Numerous coarse constituent particles were observed stratified and distributed along the rolling direction of the wrought plate. At frequent intervals clustering or agglomeration of the particles was observed. (2) The cyclic stress response of the alloy containing fine matrix precipitates revealed continuous softening to failure at all temperatures. This is aided by the growth and coalescence of the microscopic cracks to form macroscopic cracks and thus a progressive decrease in stress carrying capability or softening of the structure. The overall degree of softening increased with increase in test temperature. (3) The mechanism largely responsible for the decrease in stress amplitude during fullyreversed strain amplitude-controlled cycling is a progressive deterioration of strengthening contribution to the matrix caused by the interaction of dislocations with the precipitates. The inhomogeneous deformation resulting from dislocation-precipitate interactions results in the localization of strain at the grain boundary regions and the concomitant intergranular deformation. (4) The short period of time associated with each low-cycle fatigue test when compared to

Aluminum alloy 6061

549

the actual time required to cause any time-dependent damage rules out the possibility of creep as a contributing factor. At the elevated temperatures localized oxidation and embrittlement are two major factors contributing to cyclic fatigue damage. The formation of oxides, at the higher test temperatures, and their repeated rupturing during fully-reversed cyclic straining contributes to a local increase in crack length and enhancing the rate of crack propagation and thus softening. Embrittlement results from an interaction between several competing mechanisms involving intrinsic microstructural features, strength of material and nature of the environment. The synergistic influence of these mechanisms on cyclic response is exacerbated by the applied stress. (5) Cyclic fracture morphology was essentially similar at the different temperatures. Fracture was bimodal comprising transgranular and intergranular regions with the cracks along the grain boundaries. The microscopic crack path, through the microstructure, was tortuous with evidence of deflections during propagation. The overload region comprised numerous voids and shallow dimples, features reminiscent of locally ductile failure.

Acknowledgements--The authors graciously thank the State of Ohio: Board of Regents and The University of Akron for partially supporting this research study.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38.

Kuhlmann-Wilsdorf, D. and Laird, C., Materials Science and Engineering, 1980, 46, 209. Figueroa, J. C. and Laird, C., Acta Metallurgica, 1981, 29, 1679. Figueroa, J. C., Bhat, S. P., De La Veaux, R., Murzenski, S. and Laird, C., Acta Metallurgica, 1981, 29, 1667. Mughrabi, H., Herz, K. and Stark, X., International Journal of Fracture, 1981, 17, 193. Laird, C., Metallurgical Treatises, ed. J. K. Tien and J. F. Knott. TMS-AIME, 1981, p. 505. Jia-Kuen Lee and Laird, C., Materials Science and Engineering, 1982, 54, 39. Hansen, N. and Ralph, B., Acta Metallurgica, 1982, 30, 411. Schmidt, C. G. and Miller, A. K., Acta Metallurgica, 1982, 30, 615. Gonzales, G. and Laird, C., Metallurgical Transactions, 1983, 14A, 2507. Yan, B. D. and Laird, C., Acta Metallurgica, 1985, 33(11), 2023. Witmer, D. E., Farrington, G. C. and Laird, C., Acta Metallurgica, 1987, 35(7), 1895. Horibe, S. and Laird, C., Acta Metallurgica, 1987, 35(7), 1919. Bonda, N. R., Pope, D. P. and Laird, C., Acta Metallurgica, 1987, 35(9), 2371. Lerch, B. A. and Gerold, V., Metallurgical Transactions, 1987, 18A, 2135. Fritzemeier, L. G. and Tien, J. K., Acta Metallurgica, 1988, 36(2), 275. Fine, M. E. and Ritchie, R. O., in Fatigue and Microstructure, ed. M. Meshii. American Society for Metals, Metals Park, Ohio, U.S.A., 1979, p. 245. Staley, J. T. and Lege, D. J., Journal de Physique, 1993, IV(3), 179. Staley, J. T., Aluminum alloys and composites. In Encylopedia of Physical Science, Vol. 1, 1992, p. 591. Srivatsan, T. S., International Journal of Fatigue, 1988, 10(2), 91. Srivatsan, T. S., Yamaguchi, K. and Starke, E. A. Jr, Materials Science and Engineering, 1986, 83, 87-107. Srivatsan, T. S., Vasudevan, V. K. and Veeraraghavan, D., Cyclic strain resistance and fracture behavior of aluminum alloy 7055. Engineering Fracture Mechanics (submitted). Sanders, R. E. Jr and Starke, E. A. Jr, Materials Science and Engineering, 1977, 28, 53. Sanders, R. E. Jr and Starke, E. A. Jr, Metallurgical Transactions, 1978, 9A, 1087. Sanders, T. H. Jr and Starke, E. A. Jr, Metallurgical Transactions, 1976; 7A, 1107-1117. Srivatsan, T. S. and Coyne, E. J. Jr, International Journal of Fatigue, 1986, 8(4), 201. Srivatsan, T. S., International Journal of Fatigue, 1991, 13(4), 313-321. Srivatsan, T. S., Lanning, D. and Soni, K. K., International Journal of Fatigue, 1993, 15(3), 231-242. Srivatsan, T. S., Hoff, T. and Prakash, A., Engineering Fracture Mechanics, 1991, 40(2), 297-309. Srivatsan, T. S. and Lanning, D. Jr, Engineering Fracture Mechanics, 1992, 42(5), 877-892. Vasudevan, A. K., Doherty, R. D. and Suresh, S., in Aluminum Alloys: Contemporary Research and Applications, Vol. 31, ed. A. K. Vasudevan and R. D. Doherty. Treatise in Materials Science and Technology, Academic Press, 1989, p. 446. Starke, E. A., Jr, In Aluminum Alloys: Contemporary Research and Applications, Vol. 31, ed. A. K. Vasudevan and R. D. Doherty. Treatise in Materials Science and Technology, Academic Press, 1989, p. 35. Starke, E. A., Jr, and Lutjering, G., in Fatigue and Microstructure, ed. M. Meshii. American Society for Metals, 1979, p. 205. Srivatsan, T. S., Meyers, C. W. and Berry, J. T., Journal of Testing and Evaluation, 1987, 15, 196. Calabrese, C. and Laird, C., Materials Science and Engineering, 1974, 13, 141-150. Calabrese, C. and Laird, C., Materials Science and Engineering, 1974, 13, 159-170. Duva, J. M., Daeubler, M. A., Starke, E. A. Jr and Luetjering, G., Acta Metallurgica, 1988, 36(3), 585. Baxter, W. J. and McKinney, T. R., Metallurgical Transactions, 1988, 19A, 83. Pedessen, O. B., Rasmussen, K. V. and Winter, A. T., Acta Metallurgica, 1982, 30, 57.

550

T.S. SRIVATSAN et aL

39. Bhanu Shankara Rao, K., Seetharaman, V., Mannan, S. L. and Rodriquez, P., High Temperature Material Processes, 1986, 7, 63. 40. Clavel, M. and Pineau, A., Scripta Metallurgica, 1982, 16, 361. 41. Nazmy, M. V. and Duerig, T. W., Scripta Metallurgica, 1982, 16, 331. 42. Shiozawn, K. and Weertman, J. R., Scripta Metallurgica, 1981, 15, 1241. 43. Kukuchi, M. and Weertman, J. R., Scripta Metallurgica, 1980, 14, 797. 44. Valsan, M., Sastry, D. H., Bhanu Shankara Rao, K. and Mannan, S. L., Metallurgical Transactions, 1994, 25A, 159. 45. Antolovich, S. D., in La Fatigue des Materiaux et des Structures, ed. C. Bathias and J. P. Bailon. Mailoine, Paris, 1980, p. 465. 46. Gough, J. H. and Sopwith, D. G., Journal Institute of Metals, 1946, 72, 465. 47. Thompson, N., Wadsworth, N. J. and Louat, N., Philosophical Magazine, 1956, 1, 113. 48. Wadsworth, N. J. and Hutchings, J., Philosophical Magazine, 1969, 3, 1154. 49. Achter, M. R., in .4STM STP 415, American Society for Testing and Materials, Philadelphia, PA, U.S.A., 1967, p. 181. 50. Cook, R. H. and Skelton, R. P., International Metals Reviews, 1974, 19, p. 199. 51. Srivatsan, T. S. and Sudarshan, T. S., Journal of Materials Science, 1988, 23, 1521-1533. 52. Manson, S. S. and Zab, R., in Environmental Degradation of Engineering Materials, ed. M. R. Louthan, Jr and R. P. McNitt. Virginia Polytechnic Institute Press, Blacksburg, VA, U.S.A., 1977, p. 754. 53. Fournier, D. and Pineau, A., Metallurgical Transactions, 1977, 8A, 1095. 54. Prager, M. and Sines, G., Embrittlement of precipitation hardenable nickel-base alloys, ASME Paper No. 71, American Society of Mechanical Engineers, 1971. 55. Gurland, J. and Plateau, J., Transactions ASM, 1963, 56, 442. 56. Van Stone, R. H. and Psioda, J. A., Metallurgical Transactions, 1975, 6A, 672. 57. Van Stone, R. H., Cox, T. B., Low, J. R. Jr and Psioda, J. A., International Metals Reviews, 1985, 30, 157. 58. Argon, A. S., lm, J. and Safoglu, R., Metallurgical Transactions, 1975, 6A, 825.

(Received 9 June 1995)