METALLURGICAL AND MATERIALS TRANSACTIONS A
	ABSTRACTS Volume 26A, No. 12, December 1995 This Month Featuring: The 1994 Edward DeMille Campbell Memorial Lecture; Symposium on Creep and Fatigue of Metal Matrix Composites; Welding and Joining. View December 1995 Table of Contents.

THE 1994 EDWARD DEMILLE CAMPBELL MEMORIAL LECTURE

SYMPOSIUM OF CREEP AND FATIGUE OF METAL MATRIX COMPOSITES

In recognition of this growing area of research, a three-day, six-session symposium on the Creep and Fatigue of Metal Matrix Composites was held in San Francisco as part of the 1994 Annual Meeting of TMS. Papers were presented on topics ranging from the modeling of creep and fatigue behavior to the most recent experimental studies of creep, cyclic deformation and thermal fatigue, crack initiation and fatigue crack growth. Although the majority of presentations focused on particle reinforced alloys, continuous fiber composites and in-situ lamellar composites were also represented. The papers presented in this issue of Metallurgical and Materials Transactions represent a significant fraction, and a representative cross-section of the papers presented at the symposium.

The symposium was sponsored by the Joint TMS-SMD/ASM-MSD Composite Materials Committee and the counsel and assistance of members of this committee are gratefully acknowledged. We also wish to acknowledge our colleagues who participated in this symposium and contributed manuscripts for publication. The efforts of Prof. David Laughlin, Editor, and Ms. Dora Moscatello, Production Editor, Metallurgical and Materials Transactions, to incorporate manuscripts from this symposium into the normal review process are appreciated.

A Finite Element Model of the Effects of Primary Creep in an Al-SiC Metal Matrix Composite
STEVEN L. ATKINS AND JEFFREY C. GIBELING
A two dimensional axisymmetric finite element model has been developed to study the creep behavior of a high-temperature aluminum alloy matrix (alloy 8009) reinforced with 11 vol pct silicon carbide particulate. Because primary creep represents a significant portion of the total creep strain for this matrix alloy, the emphasis of the present investigation is on the influence of primary creep on the high-temperature behavior of the composite. The base alloy and composite are prepared by rapid solidification processing, resulting in a very fine grain size and the absence of precipitates that may complicate modeling of the composite. Because the matrix microstructure is unaffected by the presence of the SiC particulate, this material is particularly well suited to continuum finite element modeling. Stress contours, strain contours, and creep curves are presented for the model. While the final distribution of stresses and strains is unaffected by the inclusion of primary creep, the overall creep response of the model reveals a significant primary strain transient. The effects of true primary creep are more significant than the primary-like transient introduced by the redistribution of stresses after loading. Examination of the stress contours indicates that the matrix axial and shear components become less uniform while the effective stress becomes more homogeneous as creep progresses and that the distribution of stresses do not change significantly with time after the strain rate reaches a steady state. These results also confirm that load transfer from the matrix to reinforcement occurs primarily through the shear stress. It is concluded that inclusion of matrix primary creep is essential to obtaining accurate representations of the creep response of metal matrix composites.

Micromechanics Effects in Creep of Metal-Matrix Composites
L.C. DAVIS, AND J.E. ALLISON
The creep of metal-matrix composites is analyzed by finite element techniques. An axisymmetric unit-cell model with spherical reinforcing particles is used. Parameters appropriate to TiC particles in a precipitation-hardened (2219) Al matrix are chosen. The effects of matrix plasticity and residual stresses on the creep of the composite are calculated. We confirm (1) that the steady-state rate is independent of the particle elastic moduli and the matrix elastic and plastic properties, (2) that the ratio of composite to matrix steady-state rates depends only on the volume traction and geometry of the reinforcing phase, and (3) that this ratio can be determined from a calculation of the stress-strain relation for the geometrically identical composite (same phase volume and geometry) with rigid particles in the appropriate power-law hardening matrix. The values of steady-state creep are compared to experimental ones (Krajewski et. al.). Continuum mechanics predictions give a larger reduction of the composite creep relative to the unreinforced material than measured, suggesting that the effective creep rate of the matrix is larger than in unreinforced precipitation-hardened Al due to changes in microstructure, dislocation density, or creep mechanism. Changes in matrix creep properties are also suggested by the comparison of calculated and measured creep strain rates in the primary creep regime, where significantly different time dependencies are found. It is found that creep calculations performed for a time independent matrix creep law can be transformed to obtain the creep for a time-dependent creep law.

An Approach to the Design of Composites for Service at Elevated and Nonsteady Temperatures
KARIM F. ELFISHAWY AND GLENN S. DAEHN
The language and ideas developed in the continuum mechanics area of ratcheting and shakedown can be applied to the analysis and design of composites that are subjected to both external and thermal loading. Specifically, the diagrams first proposed by Bree and previously applied to composites by Jansson and Leckie can be of value in designing composites that will remain dimensionally stable under stress and temperature cycling. Three simple cases are analyzed analytically and from these, it seems apparent that the resulting composite behavior maps (CBMs) have a characteristic shape. This shape can be conservatively estimated by the linear connection of points that correspond to initial yield and interconnected plasticity under pure mechanical and pure thermal loading. Both experimental and analytical methods of determining these points are discussed. Finally, good agreement between a CBM produced by simplified analytical procedures and a full three-dimensional (3D) finite element analysis is shown.

The Effect of Particle Reinforcement on the Creep Behavior of Single-Phase Aluminum
P.E. KRAJEWSKI, J.W. JONES, and J.E. ALLISON
The effect of TiC particle reinforcement on the creep behavior of Al (99.8) and Al-1.5Mg is investigated in the temperature range of 150 °C to 250 °C. The dislocation structure developed during creep is characterized in these materials. The addition of TiC increases creep resistance in both alloys. In pure aluminum, the presence of 15 vol pct TiC leads to a factor of 400 to 40,000 increase in creep resistance. The creep strengthening observed in Al/TiC/15p is substantially greater than the direct strengthening predicted by continuum models. Traditional methods for explaining creep strengthening in particle-reinforced materials (e.g., threshold stress, constant structure, and dislocation density) are unable to account for the increase in creep resistance. The creep hardening rate (h) is found to be 100 times higher in Al/TiC/15p than in unreinforced Al. When incorporated into a recovery creep model, this increase in h can explain the reduction in creep rate in Al/TiC/15p. Particle reinforcement affects creep hardening, and thus creep rate, by altering the equilibrium dislocation substructure that forms during steady-state creep. The nonequilibrium structure generates internal stresses which lower the rate of dislocation glide. The strengthening observed by adding TiC to Al-1.5Mg is much smaller than that found in the pure aluminum materials and is consistent with the amount of strengthening predicted by continuum models. These results show that while both direct (continuum) and indirect strengthening occur in particle-reinforced aluminum alloys, the ratio of indirect to direct strengthening is strongly influenced by the operative matrix strengthening mechanisms.

Creep Strengthening in a Discontinuous SiC-Al Composite
KYUNG-TAE PARK AND FARGHALLI A. MOHAMED
High-temperature strengthening mechanisms in discontinuous metal matrix composites were examined by performing a close comparison between the creep behavior of 30 vol pct SiC-6061 Al and that of its matrix alloy, 6061 Al. Both materials were prepared by powder metallurgy techniques. The experimental data show that the creep behavior of the composite is similar to that of the alloy in regard to the high apparent stress exponent and its variation with the applied stress and the strong temperature dependence of creep rate. By contrast, the data reveal that there are two main differences in creep behavior between the composite and the alloy: the creep rates of the composite are more than one order of magnitude slower than those of the alloy, and the activation energy for creep in the composite is higher than that in the alloy. Analysis of the experimental data indicates that these similarities and differences in creep behavior can be explained in terms of two independent strengthening processes that are related to (a) the existence of a temperature-dependent threshold stress for creep, ₀, in both materials and (b) the occurrence of temperature dependent load transfer from the creeping matrix (6061 Al) to the reinforcement (SiC). This finding is illustrated by two results. First, the high apparent activation energies for creep in the composite are corrected to a value near the true activation energy for creep in the unreinforced alloy (160 kJ/mole) by considering the temperature dependence of the shear modulus, the threshold stress, and the load transfer. Second, the normalized creep data of the composite fall very close to those of the alloy when the contribution of load transfer to composite strengthening is incorporated in a creep power law in which the applied stress is replaced by the effective stress, the stress exponent, n, equals 5, and the true activation energy for creep in the composite, Q_c, is equal to that in the alloy.

Mechanical Behavior and Failure Micromechanics of Al/Al₂O₃ Composites under Cyclic Deformation
P. POZA AND J. LLORCA
The mechanical behavior under fully reversed cyclic deformation was determined through the incremental step method for two Al alloys reinforced with 15 vol pct Al₂O₃ particulates in the naturally aged and peak-aged conditions. The composites exhibited cyclic strain hardening in all cases, but the hardening was more pronounced in the naturally aged condition. This behavior was reflected by the stress-strain curves in monotonic tension and in fatigue, and the cyclic strain-hardening coefficient was about twice the monotonic one for both materials and tempers. The tensile and cyclic strengths of the materials were very similar, and the dominant failure mechanism under both loading conditions was particulate fracture, which was very localized around the fracture region in fatigue, but was spread along the specimen length in monotonic tension. In addition, a few Al₂O₃ particulates were broken in compression during cyclic de formation. The final fracture micromechanism was the growth and coalescence of voids in the matrix from broken ceramic particulates. This last stage in the fracture process was fast and started when a critical volume fraction of broken reinforcements (between 30 and 45 pct) was reached in a given section of the specimen.

The Effect of Matrix Microstructure on Cyclic Response and Fatigue Behavior of Particle-Reinforced 2219 Aluminum: Part I. Room Temperature Behavior
G.M. VYLETEL, J.E. ALLISON, and D.C. VAN AKEN
The low-cycle and high-cycle fatigue behavior and cyclic response of naturally aged and overaged 2219/TiC/15p and unreinforced 2219 Al were investigated using plastic strain-controlled and stress-controlled testing. In addition, the influence of grain size on the particle-reinforced materials was examined. In both reinforced and unreinforced materials, the naturally aged conditions were cyclically unstable, exhibiting an initial hardening behavior followed by an extended region of cyclic stability and ultimately a softening region. The overaged reinforced material was cyclically stable for the plastic strains examined, while the overaged unreinforced material exhibited cyclic hardening at plastic strains greater than 2.5 X 10^-4. Decreasing grain size of particle-reinforced materials modestly increased the cyclic flow stress of both naturally aged and overaged materials. Reinforced and unreinforced materials exhibited similar fatigue life behaviors; however, the reinforced and unreinforced naturally aged materials had superior fatigue lives in comparison to the overaged materials. Grain size had no effect on the fatigue life behavior of the particle-reinforced materials. The fatigue lives were strongly influenced by the presence of clusters of TiC particles and exogenous Al₃ Ti intermetallics.

The Effect of Matrix Microstructure on Cyclic Response and Fatigue Behavior of Particle-Reinforced 2219 Aluminum: Part II. Behavior at 150 °C
G.M. VYLETEL, D.C. VAN AKEN, and J.E. ALLISON
The 150 °C cyclic response of peak-aged and overaged 2219/TiC/15p and 2219 Al was examined using fully reversed plastic strain-controlled testing. The cyclic response of peak-aged and overaged particle-reinforced materials showed extensive cyclic softening. This softening began at the commencement of cycling and continued until failure. At a plastic strain below 5 X 10^-3, the unreinforced materials did not show evidence of cyclic softening until approximately 30 pct of the life was consumed. In addition, the degree of cyclic softening ( ) was significantly lower in the unreinforced microstructures. The cyclic softening in both reinforced and unreinforced materials was attributed to the decomposition of the strengthening precipitates. The extent of the precipitate decomposition was much greater in the composite materials due to the increased levels of local plastic strain in the matrix caused by constrained deformation near the TiC particles.

Comparison of the Growth of Individual and Average Microcracks in the Fatigue of Al-SiC Composites
E.Y. CHEN, L. LAWSON, and M. MESHII
In reliability predictions, the driving force for crack growth is often taken to be proportional to the product of the applied stress and crack length to the one-half power as in the Paris equation. On the scale of microcracks, the Paris equation appears to persist, at least in the short term, for the growth of individual microcracks. Yet, when evolving distributions of microcrack lengths are analyzed, the fact that the lognormal distribution best fits those of the smaller crack lengths suggests that the growth of all microcracks cannot be described by the Paris equation. These aspects of the microcrack growth problem are explored in this study of what controls the growth behavior of fatigue microcracks on the surface of smooth specimens of a 2124 Al-20 vol pct SiCw composite. Experimental observations of individual microcrack growth are reported and correlated with the distribution data. A Monte Carlo simulation constructed to replicate the observed distributions is presented to assist quality assurance inspection scheduling and provide insight from a probabilistic standpoint. This work indicates that conventional notions of long crack growth cannot be applied directly to microcracks due to the intermittent pattern of their growth. The growth behavior of the averaged crack, however, can be described by a single Paris equation when arrest and coalescence are separately considered. As a result, for reliability purposes, it appears feasible to base microcrack growth predictions on the growth of the average crack and the dispersion of the experimental distribution.

Short Crack Growth Behavior in a Particulate-Reinforced Aluminum Alloy Composite
CHINGSHEN LI AND F. ELLYIN
Short and long crack propagation behaviors in a coarse Al₂O₃ particulate-reinforced 6061 aluminum alloy composite (Al₂O₃/6061Al) are investigated and compared under different ranges of tensile-compressive cyclic stress. It is found that short cracks up to 400 um in length propagate in a shear-dominant mode at maximum cyclic stress level below the fatigue limit until they are permanently trapped by the surrounding particles. The microstructure sensitivity of short crack growth in the composite decreases as the short crack length and/or applied stress range increase. The characteristics of short cracks and the mechanisms of short crack trapping by particles in the material are discussed.

Acoustic Emission during Fatigue Crack Propagation in SiC Particle Reinforced Al Matrix Composites
A. NIKLAS, L. FROYEN, M. WEVERS, AND L. DELAEY
The acoustic emission (AE) behavior during fatigue propagation in aluminum 6061 and aluminum 6061 matrix composites containing 5, 10, and 20 wt pct SiC particle reinforcement was investigated under tension-tension fatigue loading. The purpose of this investigation was to monitor fatigue crack propagation by the AE technique and to identify the source(s) of AE. Most of the AEs detected were observed at the top of the load cycles. The cumulative number of AE events was found to correspond closely to the fatigue crack growth and to increase with increasing SiC content. Fractographic studies revealed an increasing number of fractured particles and to a lesser extent decohered particles on the fatigue fracture surface as the crack propagation rate (e.g., K) or the SiC content was increased.

Effect of Thermal Residual Stresses on Fatigue Crack Opening and Propagation Behavior in an Al/SiCp Metal Matrix Composite
M.E. FITZPATRICK, M.T. HUTCHINGS, J.E. KING, D.M. KNOWLES, and P.J. WITHERS
The effects of a thermal residual stress field on fatigue crack growth in a silicon carbide particle-reinforced aluminum alloy have been measured. Stress fields were introduced into plates of material by means of a quench from a solution heat-treatment temperature. Measurements using neutron diffraction have shown that this introduces an approximately parabolic stress field into the plates, varying from compressive at the surfaces to tensile in the center. Long fatigue cracks were grown in specimens cut from as-quenched plates and in specimens which were given a stress-relieving overaging heat treatment prior to testing. Crack closure levels for these cracks were determined as a function of the position of the crack tip in the residual stress field, and these are shown to differ between as-quenched and stress-relieved samples. By monitoring the compliance of the specimens during fatigue cycling, the degree to which the residual stresses close the crack has been evaluated.

Fatigue Crack Growth Behavior of Composites
A.K. VASUDEVAN AND K. SADANANDA
Fatigue crack growth data of discontinuously reinforced composites published in the literature has been re-evaluated using the two-parametric approach developed by the authors. The use of these two parameters involves K and K_max, as the driving forces, which are required simultaneously for fatigue crack growth to occur. These two parameters are intrinsic to fatigue deformation process. The first parameter is related to the degree of cyclic plasticity that results in fatigue damage near the crack tip, and the second (the peak stress) is required for initiating the local fracture processes in the fatigue-damaged region. Both driving forces are required simultaneously and have to exceed some critical minima for crack advancement. Thus, there are two fatigue thresholds instead of one as is normally assumed. However, in a given region, depending on the material and its crack-tip environment, one or the other parameter controls the growth behavior. Thus, normally for all materials (including composites) below a certain critical R ratio, fatigue crack growth is K_max controlled. Above the critical R ratio, it is K controlled. Although one parameter is the controlling factor in a given regime, both driving forces are needed to complete the fatigue description. This two-parametric approach is valid not only at the thresholds, but also at the higher crack growth rates. An understanding of fatigue process, then, requires a systematic evaluation of how these two driving forces vary with the reinforcement size, shape, volume fraction, and distribution, along with other material properties of the constituent phases, such as the interfaces. Finally, this article discusses some of the possible mechanisms and their effects on the two driving forces, using the limited available data in the literature.

An Evaluation of Fiber-Reinforced Titanium Matrix Composites for Advanced High-Temperature Aerospace Applications
JAMES M. LARSEN, STEPHAN M. RUSS, and J. W. JONES
The current capabilities of continuous silicon-carbide fiber-reinforced titanium matrix composites (TMCs) are reviewed with respect to application needs and compared to the capabilities of conventional high-temperature monolithic alloys and aluminides. In particular, the properties of a first-generation titanium aluminide composite, SCS-6/Ti-24Al-11Nb, and a second-generation metastable beta alloy composite, SCS-6/TIMETAL 21S, are compared with the nickel-base superalloy IN100, the high-temperature titanium alloy Ti-1100, and a relatively new titanium aluminide alloy. Emphasis is given to life-limiting cyclic and monotonic properties and to the influence of time-dependent deformation and environmental effects on these properties. The composite materials offer a wide range of performance capabilities, depending on laminate architecture. In many instances, unidirectional composites exhibit outstanding properties, although the same materials loaded transverse to the fiber direction typically exhibit very poor properties, primarily due to the weak fiber/matrix interface. Depending on the specific mechanical property under consideration, composite cross-ply laminates often show no improvement over the capability of conventional monolithic materials. Thus, it is essential that these composite materials be tailored to achieve a balance of properties suitable to the specific application needs if these materials are to be attractive candidates to replace more conventional materials.

Materials Characterization of Silicon Carbide Reinforced Titanium (Ti/SCS-6) Metal Matrix Composites: Part I. Tensile and Fatigue Behavior
P.K. LIAW, E.S. DIAZ, K.T. CHIANG, and D.H. LOH
Flexural fatigue behavior was investigated on titanium (Ti-15V-3Cr) metal matrix composites reinforced with cross-ply, continuous silicon carbide (SiC) fibers. The titanium composites had an eight-ply (0, 90, +45, -45 deg) symmetric layup. Fatigue life was found to be sensitive to fiber layup sequence. Increasing the test temperature from 24 °C to 427 °C decreased fatigue life. Interface debonding and matrix and fiber fracture were characteristic of tensile behavior regardless of test temperature. In the tensile fracture process, interface debonding between SiC and the graphite coating and between the graphite coating and the carbon core could occur. A greater amount of coating degradation at 427 °C than at 24 °C reduced the Ti/SiC interface bonding integrity, which resulted in lower tensile properties at 427 °C. During tensile testing, a crack could initiate from the debonded Ti/SiC interface and extend to the debonded interface of the neighboring fiber. The crack tended to propagate through the matrix and the interface. Dimpled fracture was the prime mode of matrix fracture. During fatigue testing, four stages of flexural deflection behavior were observed. The deflection at stage I increased slightly with fatigue cycling, while that at stage II increased significantly with cycling. Interestingly, the deflection at stage III increased negligibly with fatigue cycling. Stage IV was associated with final failure, and the deflection increased abruptly. Interface debonding, matrix cracking, and fiber bridging were identified as the prime modes of fatigue mechanisms. To a lesser extent, fiber fracture was observed during fatigue. However, fiber fracture was believed to occur near the final stage of fatigue failure. In fatigued specimens, facet-type fracture appearance was characteristic of matrix fracture morphology. Theoretical modeling of the fatigue behavior of Ti/SCS-6 composites is presented in Part II of this series of articles.

Materials Characterization of Silicon Carbide Reinforced Titanium (Ti/SCS-6) Metal Matrix Composites: Part II. Theoretical Modeling of Fatigue Behavior
K.T. CHIANG, D.H. LOH, P.K. LIAW, and E.S. DIAZ
Flexural fatigue behavior was investigated on titanium (Ti-15V-3Cr) metal matrix composites reinforced with cross-ply, continuous silicon carbide (SiC) fibers. The titanium composites had an eight-ply (0, 90, +45, -45 deg) symmetric layup. Mechanistic investigation of the fatigue behavior is presented in Part I of this series. In Part II, theoretical modeling of the fatigue behavior was performed using finite element techniques to predict the four stages of fatigue deflection behavior. On the basis of the mechanistic understanding, the fiber and matrix fracture sequence was simulated from ply to ply in finite element modeling. The predicted fatigue deflection behavior was found to be in good agreement with the experimental results. Furthermore, it has been shown that the matrix crack initiation starts in the 90 deg ply first, which is in agreement with the experimental observation. Under the same loading condition, the stress in the 90 deg ply of the transverse specimen is greater than that of the longitudinal specimen. This trend explains why the longitudinal specimen has a longer fatigue life than the transverse specimen, as observed in Part I.

Modeling the Minimum Creep Rate of Discontinuous Lamellar-Reinforced Composites
MICHAEL F. BARTHOLOMEUSZ AND JOHN A. WERT
An analytical model has been developed to predict the creep rate of discontinuous lamellar-reinforced composites in which both phases plastically deform. The model incorporates effects associated with lamellar orientation relative to the uniaxial stress axis. For modest to large differences between matrix and reinforcement creep rates, lamellar aspect ratio has a significant impact on composite creep rate. For a prescribed reinforcing phase volume fraction, microstructural inhomogeneity can have a pronounced effect on composite creep properties. In the case of uniaxially aligned rigid lamellar-reinforced composites, an inhomogeneous distribution of reinforcing lamellae in the microstructure substantially increases the composite creep rate. Model results demonstrate that there is no significant improvement in creep resistance for aligned fiber-reinforced composites compared to aligned lamellar-reinforced composites, unless the reinforcing phase is essentially nondeforming relative to the matrix phase.

Creep-Fatigue Life Prediction of In Situ Composite Solders
C.G. KUO, S.M.L. SASTRY, AND K.L. JERINA
Eutectic tin-lead solder alloys subjected to cyclic loading at room temperature experience creep-fatigue interactions due to high homologous temperature. Intermetallic reinforcements of Ni₃Sn₄ and Cu₆Sn₅ are incorporated into eutectic tin-lead alloy by rapid solidification processes to form in situ composite solders. In this study, the in situ composite solders were subjected to combined creep and fatigue deformation at room temperature. Under cyclic deformation, the dominant damage mechanism of in situ composite solders is proposed to be growth of cavities. A constrained cavity growth model is applied to predict creep-fatigue life by taking into account the tensile loading component as well as the compressive loading component when reversed processes can occur. An algorithm to calculate cavity growth in each fatigue cycle is used to predict the number of fatigue cycles to failure, based on a critical cavity size of failure. Calculated lives are compared to experimental data under several fatigue histories, which include fully reversed stress-controlled fatigue, zero-tension stress-controlled fatigue, stress-controlled fatigue with tension hold time, fully reversed strain-controlled fatigue, and zero-tension strain controlled fatigue. The model predicts the creep-fatigue lives within a factor of 2 with the incorporation of an appropriate compressive healing factor in most cases. Discrepancy between calculated lives and experimental results is discussed.

WELDING AND JOINING