Interface Effects on the Quasi-Static and Impact Toughness of Discontinuously Reinforced Aluminum Laminates TODD M. OSMAN, HALA A. HASSAN, and JOHN J. LEWANDOWSKI Trilayer laminates consisting of two layers of aluminum alloy 7093 surrounding one layer of 7093/SiC/15p were produced via two roll-bonding techniques as well as by adhesive bonding. The effects of systematic changes in interface characteristics (i.e., weak bond via roll bonding with a thin ductile interlayer material, stronger bond via roll bonding without a thin ductile interlayer, and strongest bond via adhesive bonding with a semibrittle material) on the subsequent laminate toughness was studied. The fracture resistance of the laminates and the constituent materials was examined via instrumented Charpy notched impact testing in the crack-arrester orientation as well as by fracture-toughness testing of bend bars tested in the crack-divider and the crack-arrester orientations. The notched impact resistance of the trilayer crack-arrester laminates was found to be greater than both monolithic 7093/SiC/15p and 7093 samples of similar global thickness. The laminated structure promoted crack arrest, deflection, and large-scale deformation of the unreinforced layers, producing R-curve behavior. The tendency for interface delamination was predicted and confirmed based on recent mechanics-based analyses. The trilayer laminate structures tested in the crack-divider orientation exhibited a greater R-curve than either of the 7093/SiC/15p or 7093 samples tested at similar global thickness. Both types of roll-bonded laminates (i.e., stronger interface and weak interface containing a thin metal interlayer) exhibited a greater enhancement in Charpy impact toughness and mode I fracture toughness than did the adhesively bonded (i.e., semibrittle interface) laminates. These relative improvements in toughness were rationalized by estimating the contributions to energy absorption by the delamination and crack bridging in these systems and by the effects of the interface type on these processes. These results are generally relevant to the performance of these materials under impact and under certain blast loading and penetration situations. DOI: 10.1007/s11661-008-9538-x The Minerals, Metals & Materials Society and ASM International 2008
I.
INTRODUCTION
DISCONTINUOUSLY reinforced aluminum (DRA) materials benefit from an enhanced stiffness due to the incorporation of reinforcement particles such as silicon carbide (SiC) or alumina (Al2O3), in both particulate and whisker form. The addition of the reinforcement, however, degrades the fracture resistance with respect to the unreinforced matrix material, as documented in earlier work.[1–7] In the MB78/SiCp system,[4,5] in which MB78 is a 7XXX P/M aluminum alloy (Al-7Zn-2Mg2Cu-0.14Zr), the increased stiffness achieved by the incorporation of reinforcement material was accompanied by a decrease in tensile ductility from 19 to 3.5 pct and a decrease in initiation toughness, JIC, from 31.5 to TODD M. OSMAN, formerly with U.S. Steel Research, Monroeville, PA 15146, and Graduate Assistant, Case Western Reserve University, is Technical Director, TMS, Warrendale, PA 15086. HALA A. HASSAN, Fulbright Fellow, JOHN J. LEWANDOWSKI, Leonard Case, Jr. Professor, and Director, Center for Mechanical Characterization of Materials, are with Department of Materials Science and Engineering, Case School of Engineering, Case Western Reserve University, Cleveland, OH 44106-7204. HALA A. HASSAN, Assistant Professor, Department of Design and Production Engineering, Ain Shams University, Cairo, Egypt. Contact email:
[email protected]. Manuscript submitted September 10, 2007. Article published online May 8, 2008 METALLURGICAL AND MATERIALS TRANSACTIONS A
7.4 KJ m-2, when the DRA (MB78/SiC/15p) material was compared to the monolithic aluminum alloy (MB78) in the overaged condition. The addition of reinforcement also reduces the growth toughness, as measured by the tearing modulus.[4,5] However, the magnitude of growth toughness appears to be very dependent upon the processing, matrix, alloy, reinforcement level, specimen thickness, and matrix heat treatment. For example, Kamat et al.[7] reported catastrophic crack growth in 2014/Al2O3p and 2024/ Al2O3p systems, while others[4,5] reported some crack growth resistance in powder-processed composites that was dependent upon the reinforcement level, reinforcement size, and matrix heat treatment. The crack resistance (as measured by the tearing modulus, T) decreased from 7.2, for the overaged monolithic MB78 to 0.7, for an overaged MB78/SiC/15p composite containing silicon carbide particles with an average size of 15 lm. In that work, the tearing modulus was also found to be further reduced by decreasing the particle size or increasing the volume fraction of reinforcement.[4,5] There has been considerable research performed on the fracture of DRA materials, including those under dynamic loading conditions,[17] to investigate the role of microstructural features such as particle size, particle VOLUME 39A, AUGUST 2008—1993
distribution, reinforcement level, reinforcement type, and matrix condition.[1–16] While modifications in these intrinsic characteristics serve to improve the fracture resistance of DRA materials, other approaches to increasing fracture resistance are available. For example, extrinsic toughening mechanisms[18] have been proposed[1,6,19] as a possible method of further increasing the damage tolerance of DRA material; this has also been useful in increasing the energy absorption under high-rate-penetration conditions.[20] Laminate structures have been proposed[1,6,19,20] as an extrinsic toughening approach for DRA materials. Early work with laminated materials focused on structures consisting of monolithic metals,[21–27] while recent studies have investigated the behavior of laminates containing both reinforced and monolithic alloys.[1,6,19,20] An initial study[19] of a bilayer laminate tested in the crackdivider orientation demonstrated the enhancement in crack growth resistance afforded by a laminate structure. In that work, the crack growth toughness of a composite increased 250 pct when a layer of 3003 aluminum (i.e., an Al-Mg alloy) was laminated to a layer of MB85/SiC/15p (in the underaged condition).[19] The MB85 is a P/M 2XXX aluminum alloy (Al-3.5Cu1.5Mg-0.4Zr-0.21Mn). The degree of improvement in both notched and unnotched Charpy impact energy was also found to be greatly influenced by the layer thickness in the bilayer laminates tested in the crack-arrester orientation.[19] In addition to the behavior of the individual laminae and the laminate interfacial properties, the laminate architecture remains one of the most important factors controlling the laminate properties. Figure 1 illustrates the loading of a laminated plate product in which the
Fig. 1—Loading scenarios of laminate structures. Drawings are not to scale. 1994—VOLUME 39A, AUGUST 2008
thickness may be 1 to 2 orders of magnitude smaller than the length and width of the plate. If global mode I loading is assumed, crack growth in the final product is most probable in the crack-divider orientation, for both propagation from a free surface and growth from an internal location such as a rivet hole, necessitating toughness testing in this orientation. In contrast, impact damage is most probable in the crack-arrester orientation shown in Figure 1; thus, impact testing in this orientation is conducted presently. The current study was performed in an effort to determine the fracture mechanisms that contribute to the toughness in laminate structures under both notched impact and static global mode I loadings. Laminated DRA composites were fabricated with different interfacial strengths in the plate product form; the fracture resistance of the laminates was compared to that of the individual constituents (i.e., the monolithic aluminum alloy and the conventional DRA material). The effects of changes in the interfacial character on the resulting toughness were rationalized and quantified via the mechanisms of crack growth operating in the different conditions. The fatigue crack growth characteristics of similar laminated materials are summarized elsewhere.[28–31]
II.
EXPERIMENTAL PROCEDURES
The monolithic aluminum alloy used for this study was 7093 (Al-9.0 wt pct Zn-2.2 wt pct Mg-1.5 wt pct Cu-0.14 wt pct Zr-0.10 wt pct Ni), while the DRA material was 7093/SiC/15p (7093 reinforced with 15 vol pct F-600 grade (~10-lm) silicon carbide particles). Both materials were produced via powder metallurgy techniques by Alcoa, Inc., New Kensington, PA.[32] Two laminates were fabricated via roll-bonding techniques at a high temperature with an approximately 60 pct thickness reduction, resulting in plates with dimensions of 10 · 200 · 900 mm. Both laminates consisted of one 5.0-mm layer of the 7093/SiC/15p composite bonded to two 2.5-mm layers of the monolithic 7093 via roll-bonding procedures. The first laminate was a sandwich structure, produced to contain direct contact between the DRA material and the monolithic aluminum layers; the second roll-bonded laminate contained a commercially pure aluminum foil interlayer (~25-lm thick) between the DRA material and the monolithic aluminum layers. Additionally, a 10-mm-thick plate product was produced from the 7093/SiC/15p and the monolithic 7093. After rolling, the two DRA laminates, as well as the 7093 and 7093/SiC/15p plates, were heat treated to the T7E92 condition. This is an overaging treatment that involves a solution heat treatment at 488 C/2 h/cold water quench (CWQ), followed by two-step aging at 120 C/24 h and 150 C/8 h. This thermal treatment results in a 573-MPa yield strength and a 13 pct strain at failure for the 7093, and a 615-MPa yield strength and a 4 pct strain at failure in the X7093/SiC/15p. A third type of laminate was produced via adhesive bonding techniques in which only a 250-lm-thick epoxy METALLURGICAL AND MATERIALS TRANSACTIONS A
adhesive layer (3M AF163-2K) was used to bond the DRA material to the aluminum. Both the 7093 and the 7093/SiC/15p were heat treated to the T7E92 condition prior to adhesive bonding. In this case, two 2.5-mmthick 7093-T7E92 panels were adhesively bonded to one 5.0-mm-thick 7093/SiC/15p-T7E92 panel, producing plates that were approximately 10 · 125 · 250 mm in dimension. The adhesive layer thickness decreased from 250 to 100 lm after pressing. Because mechanical working was not used for the bonding and heat treatment was performed prior to the bonding, residual stresses arising from both deformation processing and heat treatment were kept to a minimum in these laminates. After all fabrication and heat treating were completed, longitudinally oriented specimens from the laminates and from the 7093 and 7093/SiC/15p materials were machined for use in the fracture resistance evaluations detailed here. Additionally, the interfacial regions of each laminate were evaluated using a JEOL* 840A *JEOL is a trademark of Japan Electron Optics Ltd., Tokyo.
scanning electron microscope. Potential service applications for such materials indicate that impact will most likely occur normal to the interfacial planes of the laminates. Instrumented notched Charpy tests were performed in order to simulate this behavior while characterizing the resistance to initiation and propagation in the laminate and control material. Instrumented impact testing permits the acquisition of load and deflection during impact as well as the calculation of the energy absorbed; it has been previously used in related work with laminated materials.[20,21] In the present work, samples were machined from both the laminate and the control material; the resultant load-vs-displacement curves for each material were analyzed to determine the enhancement of the energy absorption capability afforded by laminate structures. In order to determine the sequence of the failure mechanisms in the crack-arrester orientation, threepoint bend tests were also performed on notched Charpy specimens at a slow displacement rate of 0.002 mm/min. In-situ observations of crack propagation were made during testing via optical microscopy. Fracture-toughness tests were then conducted in the crack-divider orientation with global mode I loading. Three-point bend samples were machined from both the laminate and the control material, while testing was conducted in accordance with ASTM E813,[33] with the exception of the utilization of a saw-cut prenotch (radius ~ 125 lm), with a depth of 1780 lm, producing a/w = 0.2, and a remaining ligament of 720 lm between the monolithic alloy and the DRA material, instead of a fatigue precrack. The machined notch was used because of the difficulty of producing a planar crack front during fatigue precracking in such laminates. Previous work[34] has shown that specimens containing notches with root radii smaller than 125 lm produced METALLURGICAL AND MATERIALS TRANSACTIONS A
values for toughness that were essentially equivalent to those produced on the testing of specimens containing a fatigue precrack. One objective of this work was to document the enhanced resistance to fracture of a laminated composite system and the mechanisms influencing crack propagation. Previous work[19] has illustrated the tendency for nonlinear load-vs-displacement behavior in similar laminate systems; therefore, a J-integral analysis was used as a first attempt to quantify fracture resistance. An unloading compliance technique described in ASTM E813-89[33] was initially employed, in an effort to determine fracture toughness as well as crack extension during testing. The J-integral was calculated for each unloading as a function of elastic fracture toughness, as defined in ASTM E399[35] (Eq. [1]), KQ; Poisson’s ratio, m; elastic modulus, E; sample thickness, B; initial ligament size, bo; and the area under the load-vs-load point displacement curve, Apl, as shown in Eq. [1]. The monolithic 7093 and 7093/SiC/15p samples exhibited linear load-vs-displacement behavior; the fracture initiation toughness, KQ, was determined via Eq. [2], where PQ is the load, S thespan length, B the thickness, W the ao a function related to the initial sample width, and f W crack size, ao.[35] Additionally, an initiation toughness, KQ, was also calculated for the laminates at the point in the load-vs-displacement trace at which crack pop-in occurred. KQ ð1 m2 Þ 2Apl ½1 þ J¼ E Bbo KQ ¼
P Q S ao f W BW3=2
½2
Crack extension in the DRA laminate was analyzed via the unloading compliance method of ASTM E81389.[33] Additionally, serial sections were made of the laminate after testing. The near-crack-tip regions and locations closer to the starter notch were analyzed via optical metallographic techniques. Particular attention was focused on the crack profile in both constituents and on the behavior of the interfacial regions during crack growth. As will be documented here, extensive nonplanar crack extension occurred in the DRA laminates. As a consequence, the J-integral analysis outlined in ASTM E813 may not be applicable for quantifying the crack growth resistance in the laminates currently under consideration. A quantification of the fracture resistance after initiation (i.e., the growth toughness) of the laminates was desired; therefore, an alternate measure of fracture toughness was considered. A first approximation of toughness can be made via the consideration of the area under the load-vs-displacement trace during testing. A method that directly uses the area under the load-vs-displacement trace is the equivalent-energy method outlined in ASTM E992,[36] which permits the calculation of the apparent (equivalent-energy) fracture toughness, KEE. Comparisons between the fracture VOLUME 39A, AUGUST 2008—1995
resistance of the laminate and the control materials were then made on the basis of the apparent fracture toughness for each material.
III.
RESULTS
Figure 2(a) shows a secondary electron scanning electron microscopy (SEM) image of the roll-bonded laminate without a foil interlayer (RB laminate); Figure 2(b) shows a backscatter-electron image of the roll-bonded laminate with a foil interlayer (RB/foil laminate). There is intimate contact between the DRA
Fig. 2—(a) Scanning electron micrograph of RB laminate, (b) backscatter-electron image of RB/foil laminate, and (c) scanning electron micrograph of ADH laminate. 1996—VOLUME 39A, AUGUST 2008
material and the monolithic alloy in the RB laminate, while a 25-lm foil interlayer separates the DRA material and the monolithic alloy in the RB/foil laminate in Figure 2(b). Figure 2(c) shows the 100-lm-thick adhesive layer between the DRA material and the monolithic alloy in the adhesively bonded laminate (ADH laminate). A. Impact Behavior: Crack-Arrester Orientation Figure 3 compares the load-vs-displacement traces from instrumented notched Charpy impact tests for 7093-T7E92, 7093/SiC/15p-T7E92, and laminated DRA. In all cases, there was a sharp peak and a rapid decrease in energy absorption after fracture initiation at the maximum load. However, the DRA laminates exhibited increased energy absorption after fracture initiation via a marked increase in the area under the load-vsdisplacement curves. Table I compares the results for each of the conditions (i.e., interfaces) tested. All three laminates indicate improvements in impact resistance over that of 7093-T7E92 and 7093/SiC/15p-T7E92. The RB laminate and the RB/foil laminate, however, exhibited superior fracture resistance when compared to the ADH laminate. This was manifested as a lower fracture initiation resistance (i.e., maximum load) and a lower crack growth resistance (i.e., smaller area under the curve after initiation) for the ADH laminate in comparison to the RB laminate and the RB/foil laminate. In order to better determine the fracture mechanisms operating in the DRA laminates, in-situ observations of crack growth during quasi-static three-point bend testing were made via optical microscopy. The two roll-bonded laminates (i.e., RB and RB/foil) exhibited similar behavior. These laminates exhibited interfacial delamination during testing. Figure 4 displays a load-vs-time trace for the RB laminate and includes observations for a quasistatic three-point bend test in the crack-arrester orientation. The RB laminate sample depicted in Figure 4 was notched such that a 720-lm ligament of the monolithic alloy separated the notch from the DRA material. Delamination at the DRA/metal interface occurred prior to crack growth from the notch; the delamination extended approximately 4.0 mm (6.0 mm for the RB/ foil laminate) to either side of the notch in the aluminum ligament. Upon crack growth from the notch, the crack was arrested at the delaminated interface between the aluminum and DRA layers. Crack propagation in the DRA layer occurred after reinitiation of the crack on the tensile surface of the DRA layer. The crack once again arrested at the second interface between the DRA and aluminum layer via interfacial delamination, as shown in Figure 5, followed by large-scale yielding of the remaining monolithic ligament. This behavior was markedly different than that of the monolithic DRA material, in which fracture localized into a narrow band with little indication of macroscopically ductile behavior, although SEM examination revealed locally ductile fracture of the DRA. Table I also reveals that the impact resistance of the roll-bonded laminates was superior to that of the monolithic alloy, without a loss in loadbearing capacity. METALLURGICAL AND MATERIALS TRANSACTIONS A
Fig. 3—Comparison of the load-vs-deflection curve for notched monolithic alloy (7093-T7E92), notched DRA material (7093/SiC/ 15p-T7E92), and notched laminated DRA tested in the crack-arrester orientation, using instrumented Charpy impact.
Table I. Results from the Instrumented Notched Charpy Tests for 7093T-7E92, 7093/SiC/15p-T7E92, and DRA Laminates (Crack-Arrester Orientation)
Material 7093 7093/SiC/15p RB laminate RB/foil laminate ADH laminate
Maximum Load (KN)
Total Deflection (mm)
Total Energy (J cm-2)
10.2 5.4 10.4 10.7 7.5
1.8 0.9 8.2 13.7 6.4
8.2 2.0 21.5 28.4 9.5
Fig. 4—Load-vs-deflection curve for RB laminate tested in crackarrester orientation under quasi-static three-point bending.
The crack growth behavior of the ADH laminate was significantly different than that for the roll-bonded laminates. Delamination at the adhesive/metal/DRA interface did not occur prior to crack growth from the notch, although crack growth from the notch was arrested at the interface via local delamination. In contrast to the roll-bonded laminates, which exhibited METALLURGICAL AND MATERIALS TRANSACTIONS A
Fig. 5—Interfacial delamination after crack growth through DRA layer in RB DRA laminate tested under quasi-static three-point bending.
extensive delamination, delamination in the ADH laminate sample was localized to a length of approximately 1.0 mm on either side of the main crack. Subsequent crack growth occurred after crack reinitiation in the DRA layer with re-arrest at the second interface between the DRA and monolithic layers, followed by extensive deformation in the aluminum ligament. B. Quasi-Static Fracture Resistance: Crack-Divider Orientation Fracture of both the 7093-T7E92 and 7093/SiC/15pT7E92 samples occurred under approximately linear elastic conditions, with unstable fracture occurring at the maximum load (e.g., 5 KN for 7093 and 1.8 KN for 7093/SiC/10p). In contrast, the DRA laminates displayed distinctly nonlinear load-vs-crack opening displacement traces, with evidence of stable fracture under rising load conditions, as shown in Figures 6(a) through (c). All of the laminate specimens were unloaded prior to complete separation. Fracture initiation, as detected by a change in compliance and load drop, occurred at approximately the same load/toughness for the DRA laminates (denoted by the arrows on Figures 6(a) through (c)) and the DRA material tested separately. However, the postfracture initiation behavior of the laminate was drastically different from that of the DRA material. A first attempt to quantify the increased resistance to crack growth afforded by the laminate structures entailed the construction of a toughness, J, vs crack extension, Da, curve, in which J was calculated via Eq. [1] and Da was calculated from the compliance for a series of unloadings, according to ASTM E813-89.[33] However, a linear fit could not be made for the laminate structure, probably due to the complicated fracture mechanism in the laminate, which will be discussed here. In order to rationalize this apparent inability of the standard methods of data presentation to model the toughness of the laminate structures, the nature of crack growth in the laminate was investigated via a serial sectioning technique, in which micrographs were taken VOLUME 39A, AUGUST 2008—1997
Fig. 6—Load-vs-crack opening displacement for (a) RB laminate, (b) RB/foil laminate, and (c) ADH laminate. (Crack-divider orientation and crack pop-in are marked by arrows.).
of regions normal to the direction of crack growth. Figure 7(a) displays a micrograph of the near-crack-tip region (~4.8 mm from the initial notch) for the RB laminate. A contiguous crack is only evident in part of the DRA layer and does not extend to the DRA/metal interface or into the monolithic layers. Figure 7(b) illustrates a section approximately 2 mm from the initial notch in the same sample shown in Figure 7(a). In this case, interfacial delamination was observed; the contiguous crack was once again present in the DRA layer and did not extend into the monolithic layers. Crack growth in the monolithic layers was only observed at a distance of 1.6 mm from the initial notch; it extended the furthest 1998—VOLUME 39A, AUGUST 2008
Fig. 7—(a) Micrograph from the near (4.8 mm from notch) crack tip region (crack-divider orientation). Arrow marks the extent of cracking in DRA in RB laminate unloaded after significant crack growth and crack opening displacement (COD) and (b) micrograph illustrating interfacial delamination (horizontal arrows) and contiguous cracking in the DRA layer, 2 mm from initial crack without extension in the monolithic layers (crack-divider orientation).
in the central regions of the monolithic layers (i.e., the crack front was nonplanar in each of the monolithic layers). METALLURGICAL AND MATERIALS TRANSACTIONS A
Table II. Toughness Values for 7093-T7E92, 7093/SiC/15pT7E92, and DRA Laminates (Crack-Divider Orientation) Material
Initiation Toughness Apparent Toughness (KQ)* (MPaÆm1/2) (KEE)** (MPaÆm1/2)
7093 7093/SiC/15p RB laminate RB/foil laminate ADH laminate
42.2 23.3 21.0 19.7 19.0
42.2 23.3 41.1 44.1 34.0
*Toughness at crack initiation, as denoted by a load drop and compliance change. **As defined by ASTM E992.[36]
Fig. 8—Comparison of crack growth in (a) RB laminate and RB/foil laminate and (b) ADH laminate (from the serial section analysis shown by dotted lines). Interfacial delamination denoted as vertical lenticular regions in (a) was observed in the former two cases and not in the ADH laminate. The extent of fracture in each constituent (i.e., Al, DRA, or interface) is indicated in each case for different distances from the original notch.
Both roll-bonded laminates exhibited similar crack extension behavior. Cracking in the various constituents accompanied by delamination at the interfaces between the DRA and the monolithic aluminum layers is shown schematically in Figure 8. In the ADH laminate, the crack again extended further in the DRA layer and was bowed (i.e., extended furthest in the center of the DRA layer). In contrast to the RB laminate and the RB/foil laminate, the ADH laminate did not exhibit interfacial delamination prior to or after the contiguous crack in the DRA layer reached the interface, as shown in Figure 8. The crack growth always extended through the adhesive layer into the monolithic layer without interfacial delamination, as shown in Figure 8(b). These observations of nonplanar crack extension invalidate the strict use of a J-integral approach (i.e., ASTM E813) to describe the fracture behavior in such materials. As a result, an equivalent-energy calculation, as outlined in ASTM E992,[36] was performed, to quantify the crack growth resistance of the laminates. Table II displays the apparent fracture toughness, KEE, of the monolithic aluminum, the DRA material, and the DRA laminates. The apparent toughness, KEE, of the METALLURGICAL AND MATERIALS TRANSACTIONS A
monolithic alloy and the DRA material are equivalent to the initiation toughness, KQ, because fracture occurred at the maximum load for each of these materials. In contrast, the laminate structures exhibited an increased load-carrying capability after fracture initiation, as represented by the rising load after crack pop-in in Figures 6(a) through (c). The maximum load sustained in the laminate occurred after substantial crack growth and resulted in an equivalent-energy fracture toughness that was much greater than the initiation toughness. Comparisons on the basis of a KEE value may be made between the DRA laminates, because, in each case, an equivalent volume fraction and layer thickness of the constituents was maintained. A direct comparison between the KEE values for the laminates and the monolithic materials in this study may not be directly made, due to the different amounts and thickness of monolithic aluminum. For example, the DRA laminates contained two 2.5-mm-thick monolithic aluminum layers and one 5.0-mm-thick DRA layer, while both the 7093-T7E92 and 7093/SiC/15p-T7E92 samples tested were 10 mm in thickness. It is noted, however, that for both the monolithic aluminum alloy and the DRA material, KQ equals KEE. In contrast, an improvement in growth resistance for the laminates may be seen in the fact that the KEE values are consistently greater than KQ (i.e., an increased fracture resistance beyond fracture initiation). Table III and Figures 7 and 8 illustrate that the presence of the monolithic aluminum layers in the laminate were effective in retarding catastrophic fracture in the DRA layer. The ADH laminate had the greatest amount of crack extension in both the DRA and monolithic aluminum layers, as compared to the RB and the RB/foil laminates. Although crack growth in the roll-bonded laminates was similar, the RB/foil laminate exhibited a greater extent of interfacial delamination than did the RB laminate, as summarized in Table III.
IV.
DISCUSSION
The fracture resistance of the DRA laminates currently under investigation has been found to be superior to that of the individual constituents at an equivalent thickness. In the case of the specimens tested in the VOLUME 39A, AUGUST 2008—1999
crack-arrester orientation, the degree of improvement in total energy absorbed via lamination was dependent on the type of interface between the layers. While lamination did not significantly change the peak load in the instrumented impact tests, the total deflection was dramatically increased, as shown in Table I. The peak load achieved represents the load required to initiate and propagate fracture though the aluminum layer. Thus, there should not be a significant effect on the peak load; it is expected that this load should be greater than that obtained for the DRA. In-situ monitoring and postmortem analyses of the three-point bend tests revealed Table III.
Summary of the Crack Growth Behavior for Laminates
Nonplanar Crack Front Length (mm) of Largest Crack Extension in Material RB laminate RB/foil laminate ADH laminate
DRA Layer
Monolithic Layer
4.9 4.6 6.8
1.6 1.1 5.0
Length of Delamination (mm) along DRA/Al Interface Distance from Notch (mm) 1.0 2.0 3.0 4.0
to to to to
2.0 3.0 4.0 5.0
RB Laminate
RB/Foil Laminate
ADH Laminate
2.0 2.4 1.5 1.4
2.5 3.1 NA 2.7
none none none none
the sequence of events shown schematically in Figure 9. The large increase in the deflection and toughness of the laminated specimens appears primarily due to the significant increase in the total deflection. The source of these increases in total deflection relates to the interface delamination and the subsequent deformation/ bending of the remaining ligaments/layers. Delamination observed ahead of the crack tip in Figure 9(a) reduces the stress triaxiality ahead of the crack tip, as observed in similar work on other laminate systems,[19,23] and provides the main mechanism for crack arrest shown in Figure 9(b). The remaining sample then acts as an unnotched specimen, with the corresponding increase in fracture resistance associated with the removal of a stress concentration (i.e., the notch). In support of this argument, the unnotched 7093/SiC/15p-T7E92 impact specimens have an impact energy of 5.9 J cm-2 compared to 2.0 J cm-2 for notched specimens. Even after reinitiation in the DRA layer, crack propagation may once again be impeded via delamination at subsequent interfaces, as illustrated in Figure 9(c). Finally, further energy absorption is associated with the bending of the remaining unfractured/ unnotched monolithic aluminum layer, as shown in Figures 3 and 4. The greatest increase in impact energy (and total deflection) was obtained for the RB/foil laminates, which exhibited the greatest degree of delamination, followed by the RB laminate and the ADH laminate, which exhibited successively less interfacial delamination in the impact tests, consistent with these arguments. The enhancement of static mode I toughness in the laminate structures loaded in the crack-divider orientation may also be rationalized by a change in the crack
Fig. 9—Schematic of crack propagation in notched bend samples in the crack-arrester orientation: (a) delamination at interface prior to crack growth from the notch, (b) blunting of crack after crack propagation from the notch, and (c) blunting of crack at the second interface after propagation through the composite layer. 2000—VOLUME 39A, AUGUST 2008
METALLURGICAL AND MATERIALS TRANSACTIONS A
propagation mechanisms from that exhibited by the monolithic aluminum alloys and DRA. Figures 10 and 11 illustrate crack propagation in the crack-divider laminates, based on the serial sectioning analyses. Crack growth was found to be highly nonplanar, as damage extended further in the DRA layer than in the monolithic layers. Similar behavior has been reported by Antolovich et al.,[26] in a ferrous-based laminate, and by Atkins and Mai,[37] in polymeric systems. In addition to the nonplanar crack front, interfacial delamination may occur between the layers, depending on the interface strength. A first approximation of the increase in fracture resistance afforded by a laminate structure may be gained by comparing the crack extension in the DRA layer of the laminate to that exhibited by a 7093/SiC/15p
Fig. 10—Three-dimensional view of crack propagation in the crackdivider orientation. (Only two layers are shown, for clarity).
sample. The DRA layer in the laminate had a thickness (5 mm) that was greater than the critical thickness for plane strain conditions (i.e., 3.6 mm). Under these conditions, a conservative estimate of the stable crack growth present prior to fracture in a 5-mm monolithic 7093/SiC/15p sample would be the size of the plastic zone size at the specimen surface (i.e., approximately 0.50 mm). This is an order of magnitude less than the crack length that was present in the DRA layer in the laminates. As a result, the laminate structures were found to drastically increase the ability of the DRA material to withstand crack growth without catastrophic failure. This increase in fracture resistance is manifested in an improved load-carrying capacity (2750, 3010, and 2490 N for the RB, RB/foil, and ADH laminates, respectively, vs 1860 N for the monolithic DRA material) and an increase in the area under the load-vsdisplacement trace (1.64, 1.78, and 1.30 Nm for the RB, RB/foil, and ADH laminates, respectively, vs 0.25 Nm for the monolithic DRA material). Crack growth did not occur to an equivalent extent in the DRA and monolithic aluminum layers. As a result, there was a substantial ligament of monolithic alloy in the crack wake of the laminated DRA material. Crack bridging can be an important method for producing extrinsic toughening in a laminated structure. As reviewed elsewhere for the case of ductile-phase-toughened ceramics and intermetallics,[38–45] the stress intensity levels at the crack tip may be effectively reduced via the irreversible plastic deformation of ductile ligaments in the crack wake. Model experiments have been performed, to determine the role of a ductile ligament in bridging a crack in a DRA laminate.[46] In that work, an overaged monolithic aluminum (7093) tensile specimen was adhesively bonded across the load line of an overaged MB78/SiC/15p DRA compact tension specimen. Fracture of the DRA material tested without the bonded monolithic tensile specimen occurred without any stable crack growth. By contrast, when the monolithic material was present in the crack wake of the DRA material, stable crack growth occurred to a crack length of 7.0 mm. Tables III and IV show the degree of delamination and the length of the uncracked ligament from the various toughness tests conducted presently. While the global volume fraction of reinforcement is decreased in the DRA laminate (from 15 to 7.5 pct), the improvements in fracture resistance are much greater than those displayed by monolithic DRA materials with comparable reinforcement levels. In the work of Kamat Table IV. Comparison of Remaining Unbroken Ligament in DRA Laminates after Unloading at Crack Opening Displacement of 0.8 mm Amount of Uncracked Material in (mm) Material
Fig. 11—Proposed crack growth mechanism in the crack-divider orientation of roll-bonded laminates illustrating nonplanar crack growth in the DRA layer and interfacial delamination. METALLURGICAL AND MATERIALS TRANSACTIONS A
RB laminate RB/foil laminate ADH laminate
DRA Layer
Monolithic Layers
2.0 3.3 1.0
5.3 6.8 2.8
VOLUME 39A, AUGUST 2008—2001
et al.,[7] very little, if any, stable crack growth was observed in a 2014/Al2O3/5p composite with 15-lm particles. Similarly, recent work[6,20] with the X2080/ SiCp system has shown that the improvement in fracture resistance is not related simply to a decrease in the global reinforcement volume fraction. In particular, the growth resistance and apparent toughness (as measured by KEE) of a laminated DRA material with a global reinforcement level of 12 pct was greater than those of both X2080/SiC/10p and X2080/SiC/20p. In Sections A and B, an attempt is made to quantify the separate effects of interface debonding and crack bridging on the toughness of the different laminates tested presently. A. Quantification of Interfacial Effects The mechanical behavior of the DRA laminates was significantly affected by the interface characteristics in the present work, as illustrated by the differences in the toughness and the degree of interface delamination in each of the tests. Table V displays the important properties for 7093-T7E92, 7093/SiC/15p-T7E92, the epoxy adhesive, and the commercially pure aluminum foil. The RB laminate only consisted of 7093 and 7093/ SiC/15p. The RB/foil laminate contained a 25-lm aluminum foil interlayer between the 7093 and 7093/ SIC/15p layers. This interlayer had a modulus comparable to the 7093 and an intermediate toughness. By contrast, the ADH laminate contained a 100-lm-thick epoxy layer that had a very low modulus and a low toughness between the 7093 and 7093/SiC/15p layers. The conditions for delamination between dissimilar materials, bonded together and loaded in tension parallel to the interface in which a crack in one material approaches the interface, has been treated previously by Kendall[52] as well as by He and Hutchinson.[53] The
analyses reveal that the crack will be deflected along the interface instead of progressing through the next layer, if the following condition is achieved: Gic ðhm Em þ hf Ef Þ 1 ½3 hf E f 4pð1 t2 Þ Gfc where Gic and Gfc are the fracture energies of the interface and uncracked layer, respectively; hm and hf are the thickness of the cracked matrix and uncracked layer, respectively (fiber); Em and Ef are the Young’s moduli of the cracked matrix and uncracked layer, respectively (fiber); and m is Poisson’s ratio (taken equal in both constituents). In the present work, the interface toughness between the Al layer and the DRA in RB and RB/foil samples was estimated based on the delaminated length at the interface between the Al layer and DRA/Al foil (e.g., delaminated length = 4 and 6 mm on both sides around the crack, in RB and RB/foil laminates, respectively), and the energy at the point at which delamination occurred (i.e., Figure 4, point 1). The energy was estimated by calculating the area under the curve at this point. The calculated energy was then normalized by the delamination area at the interface, to estimate the interface toughness, Gic. The calculations summarized in Table VI predict that crack deflection should occur for all of the laminates. For the ADH laminates, Table VI shows that, even using the fracture energy of the adhesive, which is likely higher than that of the interface, the crack should again deflect, as observed experimentally. These results are also consistent with previous work by El-Shabasy[54] on similar adhesive bonded multilayer laminates (2080/SiC/ 20p-2080). The shear strength between the different layers was measured as 6 to 20 MPa. The remaining
Table V. Properties of Laminate Constituent Materials Material 7093 7093/SiC/15p AF163-2K[47] Al foil
E (GPa)
mel
mpl
KQ (MPaÆm1/2)
KIC (MPaÆm1/2)
GC* (KJÆm-2)
72 100 1.1 70
0.33 0.33** 0.34 0.33
0.50 0.26** 0.34 0.50
40 23 N/A 27***
38[50] 13 to 15[50] 1.2 NA
22.9 5.3 1.26 10.3
*Calculated based upon the relationship G = K2/E. **Values taken for a similar DRA material, MB78/SiC/15p-OA.[51] ***Estimated, based upon the tensile properties of a commercially pure aluminum foil[48] and the model proposed by Robinson and Tetelman.[49]
Table VI. Prediction of Crack Deflection Based on Material Properties 2
Laminate Type Gic (KJ/m ) Gfc (KJ/m2 ) Em (GPa) Ef (GPa) hm (mm) hf (mm) RB RB/foil ADH
1.78 1.187 1.26**
5.3 5.3 5.3
70 70 70
100 100 100
2.5 2.5 2.6*
5 5 5
m 0.34 0.34 0.34
Gic/Gfc RHS*** Crack Deflection 0.33 0.22 0.13
0.34 0.34 0.34
yes yes yes
*hm = thickness of Al layer and adhesive layer. **GiC = fracture energy of the adhesive. ***RHS = right-hand side of Eq. [3].
2002—VOLUME 39A, AUGUST 2008
METALLURGICAL AND MATERIALS TRANSACTIONS A
properties and the calculations used to examine the crack deflection criterion are shown in Table VI. Once it is established that delamination should and does occur, it is necessary to estimate the effects of delamination on energy absorption. Controlled interfacial delamination results in an increase in the volume of material that participates in the fracture process and reduces the energy available for further crack growth, due to the energy consumed during the formation of internal free surfaces. For the case of a two-phase material in which a brittle matrix has ductile reinforcements, Deve et al.[44] found that the plastic dissipation, Wd, associated with large-scale delaminations during loading may be represented by Eq. [4], where f is the volume fraction of ductile material, ro is the flow stress of the ductile material, n is the work-hardening coefficient of the ductile material, and d is the debond length. This relationship may be only a lower bound for energy absorption in a system in which both constituents are capable of some degree of plastic deformation.[44,55] Additionally, this equation was derived for a system that contains a continuous matrix completely surrounding the reinforcement (i.e., fibers or particles). Although this model may not be directly applicable to the current system, it is instructive to note that, in a system that relies on debonding for toughening, energy absorption scales directly with the debond length. Wd ¼ 2fro nd
½4
Table VII summarizes the calculated energy dissipation, using Eq. [4] and the measured debond lengths, d, obtained from the serial sectioning experiments summarized earlier. Values for n = 0.08 and ro = 573 MPa were input into Eq. [4] via separate tension experiments on these materials, while f = 0.5 represents the present volume fraction evaluated. As expected, an increased toughness results from an increase in the debond length, as shown in Table VII. Increasing the debond length via the use of foil interlayers (i.e., RB/foil) further increases the Wd converted to the effective toughness, Kd, using the relationship Wd = K2d/E. In the crack-divider orientation, the roll-bonded laminates had a fracture resistance greater than the adhesively bonded laminates. The roll-bonded laminates had comparable bridging ligaments (i.e., uncracked monolithic aluminum). By contrast, the ADH laminate had the smallest bridging Table VII. Calculation of Energy Dissipation (Wd = 2frond) from Debonding and Converted to Toughness Increase, Kd
Laminate Type RB
RB/foil
Debond Length (d) mm
Debond Energy Dissipation Wd (J mm-2)
Effective Toughness Kd (MPaÆm1/2)
1.4 1.5 2.0 2.4 2.5 2.7 3.1
64.2 69 91.7 110 114.6 124 142
67.0 69.5 80.1 87.7 89.6 93.2 99.7
METALLURGICAL AND MATERIALS TRANSACTIONS A
ligament and the greatest amount of crack extension. In the case of the ADH laminate, the interfacial region (100 lm in Figure 5) is much greater than the interfacial region in the RB/foil laminate (i.e., 25 lm in Figure 4(b)) and the RB laminate (essentially zero, Figure 4(a)). The adhesive layer, therefore, may be considered as an additional layer. In crack bridging, the effectiveness of a ligament is directly proportional to the toughness (i.e., yield strength and ductility) of the bridging material. The strength of the epoxy (i.e., 30 MPa) is significantly less than that of the aluminum alloy 7093 (i.e., 573 MPa). As a lower bound approximation, bridging by the low-strength epoxy layer would not offer any significant bridging contribution to toughness. Next, interfacial contributions must be considered. The ADH laminate did not exhibit any interfacial delamination in the crack-divider orientation. The lack of delamination may be related to the relatively high bond strength of the epoxy adhesive, with respect to the fracture toughness. Intrinsically, the epoxy absorbs little energy during fracture (cf. KQ and GC in Table V). However, the relatively strong bond (i.e., in comparison to the roll-bonded laminates) enables rapid crack growth to a long distance in all of the layers of the laminate. This suggests that the crack in the DRA layer would be effectively bridged until cracking began in the monolithic layers. At that point, crack growth would proceed in all of the layers (i.e., the DRA, the monolithic aluminum, and the epoxy adhesive). This is revealed by the rapid increase in the crack opening displacement and the decrease in the load-carrying capacity for the ADH laminate in Figure 6(c) after the maximum load. A greater degree of noncatastrophic fracture was exhibited in the roll-bonded laminates. In both cases, crack growth in the DRA layer was markedly less than that in the ADH laminate. This may be directly related to the occurrence of interfacial delamination and the higher toughness of the bridging layers. Interfacial delamination may occur due to the combination of the lower bond strength and the difference in the plastic Poisson’s ratio between the DRA and monolithic aluminum (Table V), thereby enabling the aluminum layers to plastically contract more than the DRA layer during loading. The energy consumed by the delamination process will reduce the energy available for propagation in both the DRA and monolithic aluminum layers. If crack growth is arrested in the monolithic aluminum layers, the remaining uncracked aluminum will also act as a bridging ligament, the contribution of which is estimated next. B. Quantification of Bridging Effects The bridging effect of uncracked ligaments on increasing the toughness was estimated in two ways. The first way was to simply estimate the difference between the stress intensity in the DRA and Al layers. Figure 6 was used to provide the instantaneous load at the longest crack extension (e.g., P = 2.4 KN for the RB laminate), while Table III summarized the longest VOLUME 39A, AUGUST 2008—2003
Table VIII. Calculations of Shielding via Bridging and Debonding
Laminate Type RB RB/foil ADH
Bridging Length (mm)
Load (KN)
Kb* (MPaÆm1/2)
Kb** (MPaÆm1/2)
Kd (MPaÆm1/2)
Total K*shielding (MPaÆm1/2)
Total K**shielding (MPaÆm1/2)
3.3 3.5 1.8
2.4 2.7 1.5
16.4 20.3 19
37.6 54 33.6
76 94.2 0
92.4 114.3 19.0
113.7 148.2 33.6
Kb*, calculated based on difference between the stress intensity in DRA and Al over this bridging length. Kb**, calculated using the weight function over this bridging length. K*shielding = Kb* + Kd. K**shielding = Kb** + Kd.
crack in each layer. Table VIII summarizes the results of these calculations and is listed as Kb*. The other method estimates the bridging contribution using the weight function method.[56,57] Z Kb ¼ rðxÞhðx; aÞdx ½5 L
where r(x) is the traction as a function of distance x behind the crack tip; h (a,x) is the weight function; and the integration limits are determined by the bridging length (L). A simplified approach to getting the traction r(x) is to assume that is it constant and equals the constrained flow stress (rc) of the metal; the discreteness of the reinforcement is accounted for by multiplying by the volume fraction (f).[56,57] The final form for Kb using this approach is Z Kb ¼ frc hða; xÞdx ½6 L
The weight function for the single-edge notch (three-point bending) bend geometry with a half outer span-to-width ratio greater than unity, as follows:[58] " # rffiffiffiffiffiffi X Am;l: ð a Þl 2 1 xmþ1 w hða; xÞ ¼ : 1 : pffiffiffiffiffiffiffiffiffiffiffi : 1 þ a 3=2 pa 1 xa a m;l ð1 Þ w
½7 In the present work, the constrained flow stress of the Al 7093-T7E92 was almost equal to the yield strength of the material, as reported in the previous work,[59] as rc = 573 MPa and f = 0.5. Table VIII shows the results of the weight function analyses for the bridging effects, listed as Kb** to differentiate it from Kb* calculated earlier. Both bridging calculations indicate that the RB/foil sample provides the highest bridging contribution, while the RB and ADH provide a somewhat lower contribution. However, it is important to note that the total energy increase is a result of both the bridging and the delamination contributions. Table VIII also summarizes the total crack shielding contribution as the sum of the bridging and delamination contributions. As indicated earlier, the ADH samples exhibited no delamination, 2004—VOLUME 39A, AUGUST 2008
while the RB/foil exhibited the greatest amount of delamination. The net effect is the production of the highest shielding in the RB/foil samples, followed by the RB samples, and then the ADH samples, consistent with the ranking of energy absorption in the laminated samples tested presently. Although the magnitude of the shielding terms are somewhat high, compared to the actual toughness measured, this likely results from an overestimation of the properties used for the interface and bridge regions.
V.
CONCLUSIONS
Three DRA laminates were fabricated: two roll bonded and one adhesively bonded. Notched Charpy impact tests were conducted in the crack-arrester orientation; quasi-static fracture-toughness tests were performed in the crack-divider orientation. Significant differences in the fracture resistance and delamination/ debonding in the three laminates were obtained and were explained by considering the fracture mechanisms and methods of energy absorption in such structures. The conditions for interface delamination were examined and estimates were offered for the separate energy enhancements provided by interface debonding and crack bridging. The following conclusions are made. 1. All three DRA laminates displayed improvements in both quasi-static and dynamic fracture resistance, compared to that of a monolithic DRA material. All laminates exhibited resistance curve (R-curve) behavior and noncatastrophic fracture, in comparison to both the monolithic aluminum and the DRA material. 2. The impact resistance in the crack-arrester orientation was greater in the roll-bonded laminates than in the adhesively bonded laminate. The superior fracture resistance of the roll-bonded laminate was accompanied by an increase in the extent of the interfacial delamination and interfacial crack deflection, thereby reducing the driving force for further crack propagation. The tendency for interfacial crack deflection was examined using recent mechanics-based analyses and was found to be consistent with the analyses that predicted that crack deflection should occur in the present materials. METALLURGICAL AND MATERIALS TRANSACTIONS A
3. The fracture resistance in the crack-divider orientation was greatest in the roll-bonded laminates, as compared to the adhesively bonded laminate, consistent with the greater interfacial delamination obtained in the former. The interfacial bond strength that was stronger in the ADH laminate than in the roll-bonded laminates, produced crack growth to a larger distance in all of the layers (i.e., the DRA, monolithic aluminum, and epoxy layers) in the ADH laminate, in comparison to both RB laminates. 4. The roll-bonded laminate manufactured with a foil interlayer exhibited a greater fracture resistance than did the roll-bonded laminate without a foil interlayer. This was accompanied by a larger crack-bridging ligament and a greater energy absorption capacity, associated with a larger extent of interfacial delamination in the former. The separate contributions of interface delamination and bridging to the toughness were estimated and found to reproduce the trends exhibited in all of the laminates tested.
ACKNOWLEDGMENTS Two of the authors (JJL and TMO) thank Alcoa, Inc. and the United States National Science Foundation for partial initial funding of this project under Grant No. NSF-DMR-PYI-89-58326, as well as for permission to publish this work. Useful discussions with Preet M. Singh, formerly of Case Western Reserve University (CWRU), as well as with Tom J. Rodjom, Thomas Elliott, and Ralph R. Sawtell, Alcoa, are acknowledged. HAH acknowledges the support of the United States–Egypt Joint Board on Scientific and Technological Cooperation for the Junior Scientist Development Visit Grant AY 2005/2006 and Fulbright Grant AY 2006/2007 currently at CWRU. This work was also partially supported (HAH and JJL) by Grant No. ONR-N00014-07-1-0583 (project manager, Dr. David Shifler). The authors also acknowledge useful interactions and discussions with Warren H. Hunt.
REFERENCES 1. J.J. Lewandowski, C. Liu, and W.H. Hunt, Jr.: Mater. Sci. Eng., A, 1989, vol. 107, pp. 241–55. 2. P. Mummery and B. Derby: Mater. Sci. Eng., A, 1991, vol. 135, pp. 221–24. 3. T.W. Cline and P. Withers: An Introduction to Metal Matrix Composites, Cambridge University Press, Cambridge, United Kingdom, 1993. 4. M. Manoharan and J.J. Lewandowski: Acta Metall. Mater., 1990, vol. 38, pp. 489–96. 5. M. Manoharan and J.J. Lewandowski: Mater. Sci. Eng., A, 1992, vol. 150, pp. 179–86. 6. W.H. Hunt, Jr., T.M. Osman, and J.J. Lewandowski: JOM, 1993, vol. 45, pp. 30–35. 7. S.V. Kamat, J.P. Hirth, and R. Mehrabian: Acta Metall. Mater., 1989, vol. 37, pp. 2395–402. 8. D.J. Lloyd: Acta Metall. Mater., 1991, vol. 39, pp. 59–71. 9. A. Mortensen: in Fabrication of Particulates Reinforced Metal Composites, J. Masenauve and F.G. Hamel, eds., ASM INTERNATIONAL, Materials Park, OH, 1990, p. 217.
METALLURGICAL AND MATERIALS TRANSACTIONS A
10. T.F. Klimowicz and K.S. Vecchio: in Fundamental Relationships between Microstructure and Mechanical Properties in Metal Matrix Composites, P.K. Liaw and M. Gungor, eds., TMS, Warrendale, PA, 1990, p. 255. 11. T.J.A. Doel, M.H. Loretto, and P. Bowen: Composites, 1993, vol. 24, pp. 270–75. 12. T.J. Downes and J.E. King: Compos., 1993, vol. 24, pp. 276–81. 13. Y. Flom and R.J. Arsenault: Acta Metall., 1989, vol. 37, pp. 2413– 23. 14. I.A. Ibrahim, F.A. Mohamed, and E.J. Lavernia: J. Mater. Sci., 1991, vol. 26, pp. 1137–56. 15. D.L. Davidson: in Metal Matrix Composites: Mechanisms and Properties, R.K. Everett and R.J. Arsenault, eds., Academic Press, Boston, MA, 1991, p. 217. 16. D.L. Davidson: Composites, 1993, vol. 24, pp. 243–54. 17. S.I. Hong, G.T. Gray, and J.J. Lewandowski: Acta Metall. Mater., 1993, vol. 41, pp. 2337–51. 18. R.O. Ritchie: Mater. Sci. Eng., A, 1989, vol. 103, pp. 15–28. 19. L. Yost Ellis and J.J. Lewandowski: Mater. Sci. Eng., A, 1994, vol. 183, pp. 59–67. 20. D.R. Lesuer, C.K. Syn, O.D. Sherby, J. Wadsworth, J.J. Lewandowski, and W.H. Hunt, Jr.: Int. Mater. Rev., 1996, vol. 41, pp. 169–97. 21. S. Lee, J. Wadsworth, and O.D. Sherby: J. Comp. Mater., 1991, vol. 25, pp. 842–53. 22. E.A. Almond, J.D. Embury, and E.S. Wright: Interfaces in Composites, ASTM STP 452, ASTM, Philadelphia, PA, 1969, p. 107. 23. J.D. Embury, N.J. Petch, A.E. Wraith, and E.S. Wright: Trans. Metall. Soc. AIME, 1967, vol. 239, p. 114. 24. R.D. Goolsby: Int. Conf. on Composite Materials, TMS, Warrendale, PA, 1978, p. 941. 25. J.G. Kaufman: J. Basic Eng. Trans. ASME, 1967, vol. 114, pp. 503–07. 26. S.D. Antolovich, K. Kasi, and G.R. Chanani: Proc. 1971 National Symp. Fracture Mechanics, ASTM, Philadelphia, PA, 1971, p. 135. 27. Y. Ohashi, J. Wolfenstine, R. Koch, and O.D. Sherby: Mater. Sci. Eng., A, 1992, vol. 151, pp. 37–44. 28. H.A. Hassan, J.J. Lewandowski, and H.M. Abd El-Latif: Metall. Mater. Trans. A, 2004, vol. 35, pp. 45–52. 29. H.A. Hassan, J.J. Lewandowski, and H.M. Abd El-Latif: Metall. Mater. Trans. A, 2004, vol. 35A, pp. 2291–303. 30. H.A. Hassan, J.J. Lewandowski, and H.M. Abd El-Latif: J. Mater. Sci., 2004, vol. 39, pp. 3063–67. 31. H.A. Hassan and J.J. Lewandowski: Mater. Sci. Technol., 2007, vol. 23 (12), pp. 1505–12. 32. W.H. Hunt, Jr. and T.J. Rodjom: Adv. Powder Metall., 1992, vol. 9, pp. 21–31. 33. ASTM STP E813, ASTM, Philadelphia, PA, 1989. 34. M. Manoharan and J.J. Lewandowski: Int. J. Fract., 1989, vol. 40, pp. R31–R34. 35. ASTM STP E399, ASTM, Philadelphia, PA, 1974. 36. ASTM STP E992, ASTM, Philadelphia, PA, 1984. 37. A.G. Atkins and Y.W. Mai: Elastic and Plastic Fracture: Metals, Polymers, Ceramics, Composites, Biological Materials, John Wiley & Sons, New York, NY, 1988. 38. V.V. Kristic, P.S. Nicholson, and R.G. Hoagland: J. Am. Ceram. Soc., 1981, vol. 64, pp. 499–504. 39. A.G. Evans and R. McMeeking: Acta Metall., 1986, vol. 34, pp. 2435–41. 40. P. Mataga: Acta Metall., 1989, vol. 37, pp. 3349–59. 41. M. Bannister and M.F. Ashby: Acta Metall., 1991, vol. 39, pp. 2575–82. 42. H.C. Cao and A.G. Evans: Acta Metall., 1991, vol. 39, pp. 2997– 3005. 43. J. Cook and J.E. Gordon: Proc. R. Soc. A, 1964, vol. 282, pp. 508– 20. 44. H.E. Deve, G.R. Odette, R. Mehrabian, A.G. Evans, R. Emaliani, and R. Hecht: Acta Metall., 1990, vol. 38, pp. 1491–502. 45. F. Zok, S. Jansson, A.G. Evans, and V. Nardone: Metall. Trans. A, 1991, vol. 22, p. 2107. 46. T.M. Osman, P.M. Singh, and J.J. Lewandowski: Scripta Metall., 1994, vol. 31, pp. 607–12. 47. Structural Adhesive Film AF-163-2K, Aerospace Technical Data, 3M Company, St. Paul, MN, 1986.
VOLUME 39A, AUGUST 2008—2005
48. Aluminum: Properties and Physical Metallurgy, J.E. Hatch, ed., ASM INTERNATIONAL, Materials Park, OH, 1984. 49. J. Robinson and A.S. Tetelman: 3rd Int. Conf. on Fracture, ICF3, Munich (paper II), Pergamon Press, New York, NY, 1973, pp. 421–27. 50. J.J. Lewandowski, C. Liu, and W.H. Hunt, Jr.: Processing and Properties for Powder Metallurgy Composites, P. Kumar, K. Vedula and A. Ritter, eds., TMS, Warrendale, PA, 1987. 51. P.M. Singh and J.J. Lewandowski: Metall. Mater. Trans. A, 1995, vol. 26A, pp. 2911–21. 52. K. Kendall: Proc. R. Soc., 1975, vol. 344A, pp. 287–302. . 53. M.Y. He and J.W. Hutchinson: Int. J. Solids Struct., 1989, vol. 25, pp. 1053–67.
2006—VOLUME 39A, AUGUST 2008
54. A.M.B. El-Shabasy: Sci. Bull. Fac. Eng. Ain Shams Univ., 2004, vol. 39 (2), pp. 367–81. 55. P. Davies: in Advanced Composites, I.K. Partrdge, ed., Elsevier Applied Science, London, 1989, pp. 303–29. 56. D.R. Bloyer, K.T. Venkateswara Rao, and R.O. Ritchie: Metall. Trans A, 1998, vol. 29A, pp. 2483–96. 57. D.R. Bloyer, K.T. Venkateswara Rao, and R.O. Ritchie: Metall. Trans A, 1999, vol. 30A, pp. 633–42. 58. T. Fett and D. Munz: Stress Intensity Factors and Weight Functions for One Dimensional Cracks, Insitut Fur Materialforschung Kernforschungszentrum, Karlsruhe, Germany, 1994. 59. T.M. Osman and J.J. Lewandowski: Metall. Mater. Trans. A, 1996, vol. 27A, pp. 3937–47.
METALLURGICAL AND MATERIALS TRANSACTIONS A