Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading...

10
r ~!i~. -?. w -, t!"2t ELSEVIER Materials Scienceand Engineering A200 (1995) 130-139 A Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition Brian Sanders a, Shankar Mall b aMaterials Directorate, Wright Laboratory (WL/MLLN), Wright Patterson AFB, OH 45433, USA ~dir Force Institute of Technology, Wright Patterson AFB, OH 45433, USA Abstract The fatigue response of an eight-ply, unidirectional, titanium-based metal-matrix composite (MMC) (SCS-6/Ti-15-3) was investigated at elevated temperature (427 °C) using a hybrid strain-controlled loading mode. This hybrid control mode did not allow the thin MMC specimen to experience any compressive stress and, thus, prevented buckling. All fatigue testing was conducted at a constant strain rate of 0.2% s-]. Damage mechanisms were systematically identified for the cases when loading was parallel or perpendicular to the fiber direction. When the fibers were parallel to the loading direction, the dominant damage mechanism was either fiber fracture or matrix cracking. Matrix creep occurred at all levels of strain, and matrix plasticity was observed when the strain level was greater than 0.55%. When loading was perpendicular to the fiber direction, the fiber-matrix interracial damage was the dominant damage mechanism. The severity of this damage varied depending upon the maximum strain level. Matrix cracks also had a critical effect on the fatigue response when the maximum strain level was greater than 0.35%. Plastic deformation in the matrix material occurred for strain levels greater than 0.23%, and matrix creep was a key factor at all strain levels. Fatigue-life diagrams along with dominant deformation and damage mechanisms were established for both cases and are compared with previous studies. Keywords: Isothermal fatigue; Titanium matrix composite; Strain 1. Introduction Many components in future aerospace applications will require the use of advanced materials capable of withstanding severe thermomechanical environments. One class of materials that has been identified for use in these environments is titanium-based metal-matrix composites (MMCs). MMCs with the unidirectional configuration are planned for many applications, espe- cially in the immediate future, such as components of the next generation of gas turbine engines (i.e. rings, disks, etc.). Further, the unidirectional laminate is the building block for multidirectional laminates. It is therefore desirable to thoroughly characterize the longi- tudinal and transverse response of unidirectional tita- nium-matrix composites at elevated temperatures and under a variety of loading conditions. Several investigations have been performed to char- acterize the monotonic and fatigue response of tita- nium-based MMCs [1-6]. Majumdar and Newaz [7] conducted a detailed study of the initiation of damage 0921-5093/95/$09.50 © 1995 -- Elsevier ScienceS.A. All rights reserved SSDI 0921-5093(95)07003-6 and deformation in the unidirectional SCS-6/Ti-15-3 MMC under monotonic loading conditions. A few studies have been performed on the SCS-6/Ti-15-3 sys- tem to understand the damage mechanisms and changes in the material properties under thermal fatigue conditions [8,9]. Other studies have characterized the fatigue response of MMCs under isothermal and ther- momechanical loading conditions [10-13]. The empha- sis of these previous investigations has been directed towards characterizing the macroscopic response (i.e. variations in strain and modulus) and trends in fatigue lives. Most of these studies have been performed using the load-controlled mode. However, materials in most engineering applications are usually subjected to a strain-controlled fatigue condition [14], but there have only been a few investigations studying the fatigue response of MMCs under this control mode [15-17]. This paper provides an in-depth investigation of the damage mechanisms observed in a unidirectional tita- nium-matrix composite subjected to fatigue under the strain-controlled mode at 427 °C. The SCS-6/Ti-15-3

Transcript of Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading...

Page 1: Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition

r

~ ! i ~ . -?. w -, t!"2t

E L S E V I E R Materials Science and Engineering A200 (1995) 130-139 A

Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition

Brian Sanders a, Shankar Mall b

aMaterials Directorate, Wright Laboratory (WL/MLLN), Wright Patterson AFB, OH 45433, USA ~dir Force Institute of Technology, Wright Patterson AFB, OH 45433, USA

Abstract

The fatigue response of an eight-ply, unidirectional, titanium-based metal-matrix composite (MMC) (SCS-6/Ti-15-3) was investigated at elevated temperature (427 °C) using a hybrid strain-controlled loading mode. This hybrid control mode did not allow the thin MMC specimen to experience any compressive stress and, thus, prevented buckling. All fatigue testing was conducted at a constant strain rate of 0.2% s-] . Damage mechanisms were systematically identified for the cases when loading was parallel or perpendicular to the fiber direction. When the fibers were parallel to the loading direction, the dominant damage mechanism was either fiber fracture or matrix cracking. Matrix creep occurred at all levels of strain, and matrix plasticity was observed when the strain level was greater than 0.55%. When loading was perpendicular to the fiber direction, the fiber-matrix interracial damage was the dominant damage mechanism. The severity of this damage varied depending upon the maximum strain level. Matrix cracks also had a critical effect on the fatigue response when the maximum strain level was greater than 0.35%. Plastic deformation in the matrix material occurred for strain levels greater than 0.23%, and matrix creep was a key factor at all strain levels. Fatigue-life diagrams along with dominant deformation and damage mechanisms were established for both cases and are compared with previous studies.

Keywords: Isothermal fatigue; Titanium matrix composite; Strain

1. Introduction

Many components in future aerospace applications will require the use of advanced materials capable of withstanding severe thermomechanical environments. One class of materials that has been identified for use in these environments is titanium-based metal-matrix composites (MMCs). MMCs with the unidirectional configuration are planned for many applications, espe- cially in the immediate future, such as components of the next generation of gas turbine engines (i.e. rings, disks, etc.). Further, the unidirectional laminate is the building block for multidirectional laminates. It is therefore desirable to thoroughly characterize the longi- tudinal and transverse response of unidirectional tita- nium-matrix composites at elevated temperatures and under a variety of loading conditions.

Several investigations have been performed to char- acterize the monotonic and fatigue response of tita- nium-based MMCs [1-6]. Majumdar and Newaz [7] conducted a detailed study of the initiation of damage

0921-5093/95/$09.50 © 1995 -- Elsevier Science S.A. All rights reserved SSDI 0921-5093(95)07003-6

and deformation in the unidirectional SCS-6/Ti-15-3 MMC under monotonic loading conditions. A few studies have been performed on the SCS-6/Ti-15-3 sys- tem to understand the damage mechanisms and changes in the material properties under thermal fatigue conditions [8,9]. Other studies have characterized the fatigue response of MMCs under isothermal and ther- momechanical loading conditions [10-13]. The empha- sis of these previous investigations has been directed towards characterizing the macroscopic response (i.e. variations in strain and modulus) and trends in fatigue lives. Most of these studies have been performed using the load-controlled mode. However, materials in most engineering applications are usually subjected to a strain-controlled fatigue condition [14], but there have only been a few investigations studying the fatigue response of MMCs under this control mode [15-17].

This paper provides an in-depth investigation of the damage mechanisms observed in a unidirectional tita- nium-matrix composite subjected to fatigue under the strain-controlled mode at 427 °C. The SCS-6/Ti-15-3

Page 2: Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition

B. Sanders, S. Mall / Materials" Science and Engineering A200 (1995) 130-139 131

MMC was selected as a model material for this investi- gation because it has the fundamental characteristics of titanium-based MMC systems (i.e. high-strength fibers, a large mismatch between the matrix and fiber co- efficient of thermal expansion). Through systematic testing, the initiation and propagation of damage was identified for two cases: (1) fibers parallel to the loading direction, and (2) fibers perpendicular to the loading direction. These microscopic observations were related to the fatigue life to obtain a clear understanding of the fatigue behavior of this MMC under the strain-con- trolled mode.

2. Materials and experiments

The MMC used in this investigation was a unidirec- tional, eight-ply laminate consisting of Ti-15V-3Cr 3A1-3Sn (weight percent) titanium alloy reinforced with continuous silicon carbide fibers (SCS-6). This composite is frequently referred to as the SCS-6/Ti-15-3 MMC. The MMC plate (305 mm x 305 m m x 1.55 mm) was supplied by Textron Specialty Materials Divi- sion, Textron Inc., Lowell, MA. It was produced by hot isostatic pressing (hipping) alternating layers of fibers (142 pm diameter) and thin matrix foil. The as-received plate was inspected for internal fiber damage, gross matrix cracking, ply delamination, and incomplete con- solidation using ultrasonic immersion through trans- mission testing. No damage was evident from this evaluation. However, as will be shown shortly, some damage was present in the as-received material (i.e. radial cracks in the fiber-matrix interface region). The average fiber volume fraction was measured to be 36%. A 16-ply laminate of Ti-15-3 foils (neat matrix) was manufactured in the same manner as the MMC plate. Specimens from this plate were prepared and heat treated in the same manner as the MMC specimens, which is described next.

Rectangular specimens, with nominal dimensions of 152.4 mm x 12.7 mm, were cut from the plate using a low-speed diamond wheel. The specimens were then heat treated at 700 °C for 24 h in an argon atmosphere to stabilize the microstructure of the metastable matrix material [1,18]. The specimen edges were then polished to remove any damage from cutting. There was no attempt to remove cut fibers from the edges of the specimen.

Fatigue testing was conducted on a servo-hydraulic test machine. This machine was configured with water- cooled, hydraulic grips. A quartz rod, air-cooled, exten- someter was used to measure the displacement over the gauge length of 12.7 ram. The elevated temperature (427 °C) was controlled by two 1000 W, tungsten filament, water-cooled, parabolic strip lamps (one on each side of the specimen). These lamps were regulated

by a temperature-control system with feedback ob- tained through k-type thermocouples spot welded in the 25.4 mm heat zone of the specimen. The temperature gradient within the heat zone was nominally less than 10 °C.

Pollack and Johnson showed that composite com- pressive stresses will develop when a MMC is subjected to fatigue under the strain-controlled mode at elevated temperatures [13]. These compressive stresses may re- sult in buckling of the thin specimens available for testing. Buckling may be prevented by using either anti-buckling guides or thicker specimens. The use of thicker specimens was not possible since only a MMC laminate with eight plies was available. The preliminary tests with an anti-buckling guide, which was similar to one used in previous studies on fiber-reinforced poly- meric composites, showed that fatigue failure occurred in the grip area of the specimen [19]. Therefore, a special strain-controlled test technique was used. This technique was similar to one used by Bartolotta and Brindley [16].

The special test technique did not allow the minimum stress to be less than zero. This was accomplished by increasing the minimum strain by a small increment whenever the minimum stress reached a zero value. This results in a decreasing strain ratio Re (e~min/~ax) as shown in Fig. 1. A personal computer-based feedback system was developed to monitor the stress level and adjust the triangular waveform, which was produced by a MTS 810 microprofiler (i.e. a function generator) and, thus, to control the test machine. This method will be referred to as the hybrid strain-controlled mode in this paper. Using this control technique, the specimen was prevented from buckling while still being able to study the behavior of the MMC under the strain-con- trolled loading condition.

All of the fatigue tests, for both orientations and the neat material, were performed using this hybrid method. Although the frequency varied owing to a

Time

Fig. 1. Hybrid strain-controlled mode.

Page 3: Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition

132 B. Sanders, S. Mall / Mater&ls Science and Engineering A200 (1995) 130-139

changing strain amplitude, the tests were all conducted using a constant strain rate of 0.002 mm mm ~ s under isothermal conditions (427 °C). Fatigue tests were conducted at maximum strain levels ranging from 0.4% to 0.77% for the 0 ° laminate and 0.1% to 0.45% for the 90 ° laminate with an initial Re of 0.05.

The post-test specimen preparation involved section- ing the test specimens for microscopic evaluation and was followed by optical and scanning electron mi- croscopy examination. All samples sectioned for micro- scopic evaluation were taken from the 12.7 mm gage section. Most of the specimens were etched to reveal specific features of the microstructure and damage mechanisms. Two etchants were used in this study. Krolls etchant was used to highlight the microstructure and damage mechanisms, and a different etchant was used when it was desired to observe slip bands in the matrix material. In this case, the specimen was sub- jected to an additional heat treatment at 427 °C for 24 h in an argon atmosphere prior to sectioning and polishing. This additional heat treatment precipitated the a-phase along the slip bands. A solution of 3% ammonium bifluoride and 97% distilled water was then used. This solution preferentially attacks the a-phase, making the slip bands easier to identify [20].

Depleted Zone

(a)

3. Results and discussion

In this section a brief review of the condition of the pre-tested MMC and stress-strain response of the MMC is presented first. A more detailed discussion of the fatigue response can be found elsewhere [17,21]. This is then followed by a detailed discussion of the deformation and damage mechanisms which were eval- uated from the cyclic stress-strain response and micro- scopic examinations. Thereafter, the information is combined with the measured fatigue life to characterize the fatigue response of the tested MMC. The longitudi- nal and transverse laminates (i.e. fibers parallel and perpendicular to the loading direction) are presented separately.

3. I. Untested M M C

Fig. 2(a) shows the condition of the heat-treated SCS-6/Ti-15-3 MMC. A large a-phase was observed at the grain boundaries. A fine a-phase was observed within the grains, and a depleted zone existed around the fibers. This depleted zone has been shown to be softer than the surrounding matrix material [18].

In general, the integrity of the plate was good. Upon closer examination, however, some processing damage (i.e. owing to the hipping procedure) was observed. Fig. 2(b) shows a cross-section of the aged composite at a location where the fibers were very close. Radial cracks,

(b)

Fig. 2. (a) Microstructure of the SCS-6/Ti-15-3 MMC aged at 700 °C for 24 h, and (b) processing damage in the aged SCS-6/Ti-15-3 MMC.

on the order of 5-50 /~m, were observed in the inter- phase zone and matrix material. MacKay showed that cracks in the interphase zone were typically present in the as-received SCS-6/Ti-15-3 MMC when the fiber spacing was less than 60/zm [22]. It has been shown in previous studies that such damage is due to high-tensile hoop stresses that develop in the matrix and interphase zone when cooling the MMC from the processing tem- perature [9,23,24].

3.2. S t ress -s t ra& h&tory

3.2. I. Longitudinal Fig. 3(a) shows the representative stress-strain re-

sponse for the 0 ° laminate (i.e. fibers parallel to the loading direction). To aid in mapping the initiation and progression of damage and deformation, the response over the fatigue life was partitioned into two stages. These will be referred to as Stage IL and Stage IIL for the 0 ° composite. These stages were defined by a change in the stress-strain response which is discussed next.

Inelastic deformation of the matrix material was the primary mechanism that effected the response of the laminate in Stage IL. The non-linear response observed in the first cycle stress-strain curve occurred whenever the maximum strain level was greater than 0.55%. It

Page 4: Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition

B. Sanders, S. Mall / Materials Science and Engineering A200 (1995) 130-139 133

has been shown that this is the result of matrix plasti- cally [7,21]. The subsequent stress-strain response in this stage was linear, but a significant reduction in the maximum stress occurred. Sanders and Mall showed that this was owing to matrix creep [21]. These observations are further supported by the fact that the modulus remained constant in Stage IL (see caption to Fig. 3).

This basic stress-strain response was the same when the maximum strain was greater than 0.73%, but the specimens tested above this strain level fractured in this stage. When the maximum strain was less then 0.73% the maximum stress relaxed to a constant value. No further reduction in stress was then observed for a period of time. This time period was dependent on the maximum strain level.

After a period of cycling to a constant maximum stress, the response changed and entered a second stage. In this second stage, which is defined as Stage IIL, a non-linear behavior in the stress strain response was again observed. This behavior was observed whenever the maximum strain was less than 0.73%. Unlike the response in Stage IL, a simultaneous reduction in the maximum stress and the modulus was observed (see legend). This indicates that a significant amount of damage had occurred in this stage.

1200

1000

8o0

600

40O

200

0 (a) 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007

S t r a i n (mm/mm)

220

200

180

160

140

120

~ 80

6O

4O

2O

(b) 0 0.001 0.002 0.003 0.004 0.005

Stra in (ram/ram)

F i g . 3 R e p r e s e n t a t i v e s t r e s s - s t r a i n r e s p o n s e o f t h e u n i d i r e c t i o n a l

S C S - 6 / T i - 1 5 - 3 M M C a t 427 ° C : (a ) 0 ° l a m i n a t e f o r G~ax = 0 .6%; a n d

(b) 90 ° l a m i n a t e f o r cm~,× = 0 .4%.

3.2.2. Transverse Fig. 3(b) shows the representative stress-strain re-

sponse when the loading was perpendicular to the fiber direction. As in the previous case, the response was partitioned into two stages (Stage IT and Stage IIT) based on changes in the stress-strain response. In Stage IT, the response was dominated by fiber-matrix inter- face damage. The majority of this damage occurred during the loading portion of the first cycle when the maximum strain was greater than 0.23%. This damage causes the knee in the stress-strain curve [25]. On the other hand, Sanders [17] showed that most of the interface damage occurred during cycling when the strain level was less than 0.23%.

In Stage IIT, there was a rapid drop in the modulus, which was accompanied by a simultaneous reduction in the stress. This suggests that additional damage (other than the f iber-matrix interface damage mentioned above) had started to accumulate. When the maximum strain was greater than 0.35%, this damage caused the specimen to fracture into two pieces. On the other hand, the damage was arrested and the laminate stress reduced to a near zero value when the maximum strain was less than 0.35% (i.e. the specimen did not fracture). Finally, it has been shown that matrix creep can also account for some of this reduction in stress [15,17].

3.3. Damage mechanisms

3.3.1. Longitudinal As mentioned above, the 0 ° composite failed in Stage

IL (Fig. 3(a)) when the maximum strain was greater than 0.73%. Fig. 4(a) shows the fracture surface of a specimen that was subjected to cyclic loading at a maximum strain level of 0.75%. The fracture surface was dominated by fiber pull-out and ductile failure of the matrix material. Fig. 4(b) shows a representative cross-section of the matrix between two fractured fibers. The dimple patterns in the matrix represent the coales- cence of micro-voids, which is characteristic of a ductile fracture. There was no evidence of matrix fatigue dam- age (i.e. striations) on this fracture surface or the cross- sections of the specimens tested in this study. This does not preclude the existence of small pockets of matrix cracks, but it does suggest that the failure was domi- nated first by fractured fibers and then by fracture of the matrix owing to a tensile overload. Hence, this shows that fiber fracture was the dominant failure mode for specimens tested at a strain level greater than 0.73%.

Specimens that were tested at a maximum strain level less than 0.73% showed a significant decrease in both the stress and modulus later on in the fatigue life. This was categorized as Stage IIL in the fatigue response (Fig. 3(a)). Fig. 5(a) shows the first layer of fibers from the surface of the specimen cycled at a maximum strain of 0.6%. Several long matrix cracks were observed. These cracks tended to progress transversely through

Page 5: Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition

134 B. Sanders, S. Mall / Materials Science and Engineering A200 (1995) 130-139

(a)

Ductile Fracture

(b)

Fig. 4, Typical fracture surface of a 0 ° laminate that failed in Stage IL (i.e. t~, x > 0.73%): (a) extensive fiber pullout; and (b) ductile matrix fracture.

[26]. A possible explanation of why they were observed in this study is the higher strain rate used in this investigation compared with these previous studies. Fi- nally, no slip bands were observed in the specimens cycled at a maximum strain level less than 0.55%, which suggests that the initiation of plastic deformation oc- curred at a strain level of approximately 0.55%.

Some fatigue tests were stopped at intermediate cy- cles in Stage IL. The specimens from these tests were then sectioned for a microscopic evaluation. This aided in characterizing the progression of damage and defor- mation events. The first test was run until a 5% reduc- tion in the maximum stress was measured. No fiber, matrix, or fiber-matrix interface damage was observed. A second specimen was subjected to cyclic loading until the stress stabilized. Fig. 7 shows the condition of this specimen. The only damage observed at this point in the fatigue life was a few fiber cracks. The cracks were located both close to and far away from the moly weave.

Nicholas and Ahmad [27] showed that the opening of the individual fiber cracks observed in Fig. 7 would have to be of the order of several fiber diameters for this damage to significantly affect the stiffness. This correlates well with the findings in this study which showed that there was no change in the slope of the stress-strain curve in spite of this fiber damage. Also, Sanders and Mall correctly predicted the fatigue re- sponse of the composite in Stage IL assuming only matrix inelastic deformation (i.e. creep and plasticity) was occurring [21]. This clearly shows that the first stage of the fatigue response was dominated by matrix deformation and not fiber damage or matrix cracking.

the specimen and did not cause the fibers to fracture. This is a result of a weak fiber-matrix interface that promotes the conditions for crack bridging. Severe fiber-matrix interface damage also occurred, as shown in a magnified view (Fig. 5(b)). This damage is the result of oxidation of the interphase zone and the cracks propagating longitudinally along the fiber-ma- trix interface. These damage mechanisms partly explain the change in the specimen response observed in Stage IIL of the stress-strain history (Fig. 3(a)), and show that matrix cracking was the dominant damage mecha- nism when the maximum strain was less than 0.73%.

It was mentioned above that a non-linear stress- strain response was observed on the first loading cycle when the strain level exceeded 0.55%. Fig. 6 shows the condition of the matrix material in the specimen that was subjected to fatigue at a maximum strain level of 0.6%. Matrix slip bands were observed in several grains. This provides the physical evidence that the matrix material had deformed plastically at this strain level. Slip bands have not been observed in other investiga- tions of the SCS-6/Ti-15-3 MMC at this temperature

3.3.2. Transverse Fig. 8 shows a typical cross-section of the specimen

cycled at a maximum strain level of 0.4% until failure. Several matrix cracks were observed perpendicular to the loading direction. It is interesting to note that the matrix cracks propagated out of the interphase region at a location other than 90 ° to the loading direction. This was typical of the cracks which formed during fatigue. The fatigue cracks eventually coalesced into a single crack, which led to fracture of the specimen. This implies that matrix cracking was the dominant damage mechanism that eventually led to specimen fracture, in spite of the initiation and propagation of the fiber-ma- trix interfacial damage. This was the case whenever the maximum strain level exceeded 0.35%.

Specimen fracture did not occur when specimens were tested at maximum strain levels less than 0.35%, but damage was still present. This damage was in the form of fiber-matrix interfacial damage and, in some cases, small matrix cracks. Fig. 9(a) shows a typical cross-section of the specimen that was tested with a maximum strain level of 0.25%. Only a few matrix

Page 6: Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition

B. Sanders, S. Mall/Materials Science and Engineering A200 (1995) 130-139 135

$

Loading Direction

$

, Fiber

Matrix

(a)

(b)

Fig. 5. Typical cross-section of a 0 ° laminate that exhibited a Stage IIL response (i.e. ema x < 0.73%): (a) matrix cracking/fiber bridging; and (b) f iber-matr ix interface damage.

cracks can be observed, unlike the cases when the maximum strain was greater than 0.35%. In such cases, the fatigue damage was arrested and the maximum stress dropped to a near-zero value. Thus, the specimen never fractured. It will be discussed later in the paper how the fatigue life was defined in these cases.

Fig. 9(b) shows a typical damaged area for the specimen cycled at maximum strain level of 0.25%. Slip bands in the matrix material were observed emanating out of the f iber-matrix interface regions where the fibers are closest together. Matrix cracks were typically observed propagating nearby f iber-matrix interphase

Loading Direction

$

Slip Bands

Fig. 7. Typical cross-section of a 0 ° laminate after steady-state stress Fig. 6. Matrix slip bands in a 0 ° laminate for em~x = 0.6%. is achieved in Stage IL (i.e. ema × < 0.73%).

Page 7: Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition

136 B. Sanders, S. Mall/Materials Science and Engineering A200 (1995) 130-139

T Loading Direction

,L

Fiber-Matrix Interface Damage

$ Loading Direction

+

Fig. 8. Typical cross-section of a 90 ° laminate that fractured in Stage IIL (i.e. er~,x > 0.35).

cracks and along these slip bands. This shows that the matrix cracks progressed where some localized work hardening of the matrix material had occurred [7]. No slip bands were observed in specimens which were loaded at strain levels less that 0.25%. This implies that plastic deformation of the matrix initiated around a strain level of 0.25%.

Loading Direction

6----ff

Fil M; In! D~

l.,c Di

(a)

SIJ

M Cr

(b)

Fig. 9. Typical cross-section of a 90 ° laminate for 0.23% < c~a x < 0.35% showing: (a) matrix cracks; and (b) matrix cracks, fiber matrix interface damage, and slipbands.

Fig. 10. Typical cross-section of a 90 ° laminate for era, x < 0.23% showing fiber matrix interface damage.

Fatigue damage was limited to fiber-matrix interfa- cial damage when the maximum strain during fatigue was less than 0.23%. Fig. 10 shows a damaged fiber- matrix interface in a specimen which was cycled at a maximum strain level of 0.1%. There was a gradual drop in modulus during the cycling of this specimen [17]. This observation along with a micromechanics- based analysis showed that this interface damage devel- oped during cyclic loading when the maximum strain was less than 0.23% [17]. This is opposed to the sudden damage which occurred primarily during the first cycle at higher values of strain (i.e. above 0.23%). Finally, specimens tested at these strain levels showed fewer damaged interfaces than those tested at the higher strain levels. This type of variation in the amount of damaged fiber-matrix interfaces has been observed in other stud- ies of the SCS-6/Ti-15-3 MMC [1].

3.4. Fatigue-life diagrams

The information obtained from the microscopic eval- uation was combined with the fatigue-life data to de- velop fatigue-life diagrams. For this purpose, the fatigue life was plotted as a function of the maximum applied strain. It was also plotted on a strain-range basis for comparison with previous studies conducted under the load-controlled mode. The damage mechanisms found from the microscopic observations were then used to partition the fatigue curves into three regions to show the relationships between the fatigue life and damage mechanisms. This approach is similar to that suggested by Talreja for polymer-matrix composites [28]. These regions should not be confused with the stages defined earlier. Stages referred to changes in the macroscopic response, while regions were used in this study to group specimens that exhibited similar damage mechanisms (i.e. fiber fracture or matrix cracking).

As mentioned earlier, the specimen fracture does not always occur in the strain-controlled mode. For exam- ple, the 0 ° laminate did not fracture into two pieces

Page 8: Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition

B. Sanders, S. Mall / Materials Science and Engineering A200 (1995) 130-139 137

when the maximum strain level was less than 0.73%, and the 90 ° laminate did not fracture when the maxi- mum strain level was less than 0.35%. Other failure criteria must therefore be defined in addition to speci- men fracture [15,16]. When fracture did not occur, the fatigue life of the 0 ° laminate was conservatively defined to be when a simultaneous reduction in the maximum stress and modulus occurred. This corresponds to the point in the fatigue life where damage dominated the macroscopic response, which was identified as Stage IIL in Fig. 3(a). For the 90 ° laminate and the neat matrix specimens tested in this study, the fatigue life was defined by a 75% reduction in the maximum stress when the specimen did not fracture. This criterion was se- lected so that the fatigue lives measured under the hybrid control method could be compared with the fatigue lives of the same materials tested in a previous study which did not use the hybrid control mode [15].

3.4.1. Longitudinal Fig. l l(a) shows the fatigue diagram with three re-

gions for the 0 ° MMC based on a maximum strain basis. In Region I, the fatigue life was dominated by failure of the fibers. The large scatter band in this region can be attributed to the statistical nature of the fiber strength, fiber spacing, and matrix ductility [28]. In Region II, matrix cracking was the dominant damage

(a)

A

C .m

E

1

0.9!

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0 10

. . . . . . v . , , H , I . . . . . . . i . . . . . . . . i . . . . . . . . i . . . . . . . . i . . . . . . . . i . . . . . . .

R e g i o n I eglo: Temp (°C)

[~ R e f l 427 R e ~ i o n I I I Ref 29 427

• ~ n t S t ~ y 42v,,,, ........ , ........ , ,, l0 t 102 103 104 l0 s 106

N 10

1 . . . . . . . . i . . . . . . . . i . . . . . . . . r . . . . . . . . ~ . . . . . . . . i . . . . .

R e g i o n I

i o n l I

II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . "~ Temp Control /~ [

(°C) Mode

I Present Study 427 Strain Ref 29 21 Load Ref6 300 Load Region I I I

x Ref 30 427 Load [ ] Ref6 h550 Load

10 102 103 ]04 l0 s 106 107

(b) N

Fig. 11. Fat igue- l i fe curves for a 0 ° lamina te : (a) s t ra in cont ro l only; and (b) load vs. s t ra in control .

mechanism. It was observed that the fatigue life in Region II had less scatter than in Region I. This is an area which may be of interest to designers because the effect of damage on the fatigue life shows some pre- dictability in Region II, as opposed to Region I where the failure is more of a statistical nature. Thus, relation- ships between matrix cracking and the fatigue life may be developed to predict the life of a MMC in this region. Finally, Region III was defined as the matrix fatigue limit. In this region, there will be no fatigue damage in the matrix, or at least, the fatigue damage does not propagate to cause the specimen failure [28], In this study, this fatigue limit was defined as the maximum strain below which failure would not occur over a range of one to ten million cycles, and this was conservatively estimated to be at a maximum strain of 0.3% [21].

Fig. l l(b) shows the three region fatigue diagram, as a function of strain range, for both the strain-control mode (present study) and load-control mode from pre- vious studies. The fatigue lives for both control modes were plotted against the strain range measured at their half-life. As previously described, Region I was defined as the fiber-dominated failure mode. A large scatter in the data was observed in this region, The data from the load-control fatigue tests show a large scatter above a strain range of 0.6%. This suggests that damage in these specimens was also dominated by fiber fractures. Below a strain range of 0.6% (Region II) the fatigue lives fall within a narrow scatter band [7]. The data obtained from the strain-controlled fatigue tests fell well within this scatter band. This indicates that the fatigue life of the 0 ° laminate has a strong deterministic dependence on strain range in this region.

3.4.2. Transverse Figure 12a shows the three-region fatigue diagram of

the 90 ° laminate as a function of the maximum applied strain. The data obtained from fatigue tests conducted on the neat matrix in this study and data from [90]8 of the same MMC from a previous study [15] are also shown. All of these results are for the strain-controlled mode. The effect of reinforcement can be observed by comparing the fatigue lives of the composite to that of the neat matrix at the similar strain levels. The com- posite fatigue life was much shorter than that of the neat matrix at higher levels of strain (i.e. in Region I). On the other hand, the fatigue life of the composite in Region II was in good agreement with its counterpart from the neat matrix. This suggests that the fatigue life in this region, as defined in this study, has a strong dependence on matrix creep. In fact, Gayda and Gabb [15] reported that a 50% reduction in the maximum composite stress could occur without any appreciable signs of damage. Region III was defined where no fiber-matrix interface damage would occur. There were no fatigue tests conducted in this region nor was any

Page 9: Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition

138 B. Sanders, S. Mall/Materials Science and Engineering A200 (1995) 130 I39

g o.1

0.01 1 0

(a)

1

. . . . . . . . i . . . . . . . . i . . . . . . . . r

Region I • • , • . .

Temp (oc) Region I l l

• Present Study 427 5 Ref 15 427 • Present Study (Ti-15-3) 427

i i i i i l l r l 3 i , i I I I I I L 4 i i i i i , l , I 5

1 0 1 0 1 0

N

Region I EL

Region II ~ ~ ~e 0.1

Temp Control

P . . . . t Study 427 egion IIl Ref 15 427 Ref 15 427 Load

0 . 0 ] , ' , , , J , , i I , s l l , t r [ , , , , , , , t l I , , , t , , ,

l 0 t 1 0 2 1 0 3 1 0 4 1 0

(b) n

Fig. 12. Fatigue-life diagrams for: (a) a 90 ° laminate and Ti-15-3 neat matrix-strain control only; and (b) a 90 ° laminate-load vs. strain control.

longitudinal and transverse responses were studied. The chronology of the damage and deformation mecha- nisms were systematically identified along with estab- lishment of the applied strain versus fatigue life relationships. The following conclusions can be drawn from this study.

(1) When loading was parallel to the fiber direction, the sequence of damage and deformatior~ was: (i) ma- trix plasticity when the maximum strain was greater than 0.55%; (ii) matrix creep; and then (iii) fracture of fibers occurred when the maximum strain was greater than 0.73%, or matrix cracks developed when the max- imum strain was less than 0.73%.

(2) When loading was perpendicular to the fiber direction, the sequence of damage and deformation was: (i) fiber-matrix interfacial damage initiated and progressed during cycling when the maximum strain was greater than 0.1%; (ii) matrix plasticity and crack- ing occurred when the maximum strain was greater that 0.23%; (iii) specimen failure occurred owing to a propa- gation of these cracks when the maximum strain was greater than 0.35%, or these cracks were arrested when the maximum strain was less than 0.35% and matrix creep became the dominant mechanism controlling the fatigue response.

(3) Similar fatigue lives were observed for the MMC tested under the load- and strain-controlled modes when compared on a strain-range basis.

data available in the literature. This point was approxi- mated based on the experiments and the micromechan- ical analysis conducted by Sanders [17] to be at a maximum strain level of 0.05%.

Fig. 12(b) shows the fatigue-life diagram on a strain- range basis comparing the results from the strain and load-control modes. As in the case of the 0 ° laminate, the strain ranges used in Fig. 12(b) were measured at the specimen half life. It can be observed that the scatter band in Region II was narrow (i.e. less than a factor of two as shown by the solid lines). This suggests that the damage and deformation mechanisms were independent of the control mode for most of the fatigue life. Then once matrix cracks began to propagate, the specimen fractured quickly in the load-control mode. This suggests that an evolutionary law, relating the different damage and deformation mechanisms, can be developed and applied to predict the fatigue response of the MMC for either control mode.

4. Conclusions

Fatigue behavior of the unidirectional SCS-6/Ti-I 5-3 MMC tested in fatigue was investigated under the strain-controlled loading mode at 427 °C. Both the

Acknowledgment

The authors wish to acknowledge the support of the Air Force Office of Scientific Research (AFOSR).

References

[1] B.A. Lerch and J.F. Saltsman, ASTM STP, 1156 (1993) 161- 175.

[2] B.A. Lerch, NASA ConJerence Publication 10051, 35-I-8, NASA, Lewis Research Center, Cleveland, OH.

[3] J.T. Roush, S. Mall and W.H. Vaught, Composite Sci. Technol., in press.

[4] S. Mall and J.J. Schubbe, Composite Sci. Technol., 50 (1994) 49 57.

[5] W.S. Johnson, S.J. Lubowinski and A.L. Highsmith, ASTM STP, 1080 (1990) 193-218.

[6] T.P. Gabb, J. Gayda and R.A. MacKay, J. Composite Mater., 24 (1990) 667-686.

[7] B.S. Majumdar and G.M. Newaz, Philosoph. Mag., 66(2) (1992) 187-212.

[8] G.M. Newaz and B.S. Majumdar , J. Eng. Mater. Technol., ASME, submitted.

[9] S. Mall and P.G. Ermer, J. Composite Mater., 25 (1991) 1668 1686.

[10] T.P. Gabb, J.B.A. Gayda and G.R. Lerch, HalJbrd Scripta Metall., 25 (1991) 2879 2884.

[11] K.A. Hart and S. Mall, ASMEJ. Eng. Mater. Teehnol., in press.

Page 10: Isothermal fatigue behavior of a titanium matrix composite under a hybrid strain-controlled loading condition

B. Sanders, S. Mall/Materials Science and Engineering A200 (1995) 130-139 139

[12] S. Mall and B. Portner, ASME J. Eng. Mater. Technol., 114(4) (1992) 409-415.

[13] W.D. Pollock and W.S. Johnson, NASA Tech. Memo-102699, NASA, Langley Research Center, Hampton, VA, 1990.

[14] M.R. Mitchell, Fatigue and Microstructure, A SM Mater. Sci. Sem., (1978) 385-437.

[15] J. Gayda and T.P. Gabb, NASA Tech Memo-103686, NASA, Lewis Research Center, Cleveland, OH.

[16] P.A. Bartolotta and P.K. Brindley, NASA Tech. Memo-103157, NASA, Lewis Research Center, Cleveland, OH.

[17] Brian P. Sanders, Ph.D. Dissertation, AFIT/DS/AA/93-D90, 1993, AFIT, Wright Patterson AFB, OH.

[18] B.A. Lerch, T.P. Gabb and T.A. MacKay, NASA Tech. Paper- 2970, NASA, Lewis Research Center, Cleveland, OH.

[19] Donald F. Adams, Seminar Notes, Composite Materials Re- search Group, University of Wyoming, 1991.

[20] B.A. Lerch, NASA Tech Memo 103760, NASA, Lewis Research Center, Cleveland, OH.

[21] B.P. Sanders and S. Mall, J. Composites Teehnol. Res., •6(4) (1994) 304-313.

[22] R.A. MacKay, Scripta Metall. Mater., 24 (1990) 167-172. [23] B.J. Sullivan and K.W. Buesking, 21st Carbon ConJl, Buffalo,

NY, 1993. [24] C.A. Bigelow, J. Composites Teehnol. Res., 14(4) (1992) 211-

220. [25] D.D. Robertson and S. Mall, J. Composites Teehnol. Res., 14(1)

(1992) 3 11. [26] Brad Lerch, private communication. [27] T. Nicholas and J. Abroad, Composites Sci. Technol., submitted. [28] R. Talreja, Fatigue of Composite Materials', Technomic Publish-

ing Company, 1987. [29] B.S. Majumdar and G.M. Newaz, NASA CR 191181, NASA,

Lewis Research Center, Cleveland, OH. [30] M.G. Castelli, R. Ellis and P.A. Bartolotta, NASA Tech.

Memo 103171, NASA, Lewis Research Center, Cleveland, OH.