11
International Journal of Fatigue 27 (2005) 143–153 www.elsevier.com/locate/ijfatigue Effects of low-cycle fatigue on bending properties and fracture toughness of un-HIP’ed Ti–47Al–2Cr–2Nb–1B intermetallic Francesco Marino a, , A. Rebuffo b , F. Sorrentino c a Dipartimento di Scienza dei Materiali e Ingegneria Chimica, Politecnico di Torino, corso Duca degli Abruzzi 24, 10129 Torino, Italy b PHITEC INGEGNERIA S.r.l., corso Susa 299/A, 10098 Rivoli, Torino, Italy c OSVAT S.r.l., 10040 Piobesi, Torino, Italy Received 4 December 2003; received in revised form 19 May 2004; accepted 9 June 2004 Abstract Ingots of c-TiAl base intermetallic alloy were cast in a metallic mould preheated at three different temperatures in order to obtain three different near-fully-lamellar microstructures. Four-point bending strength, bending strain to fracture, fracture tough- ness, hardness and low cycle bending fatigue strength ðr max Þ L of near-fully-lamellar Ti–47Al–2Cr–2Nb–1B with different colony sizes and interlamellar spacings were measured on un-HIP’ed test bars. The effects of low-cycle fatigue, 2 10 5 alternate bending cycles (R ¼ 0:1), at high stress intensity levels, i.e. the fatigue limit, on mechanical behaviour of the three different microstructures were evaluated. The data reported and discussed in this paper show that bending fatigue ‘‘training’’ enhances the mechanical behaviour of this material, mainly through the appearance of a plastic deformation during the bending test. The effects of the microstructures on bending deformation, fracture behaviour and fatigue resistance were discussed as well. # 2004 Elsevier Ltd. All rights reserved. 1. Introduction Fully lamellar gamma-titanium aluminides are under extensive research because of their low density, adequate high temperature properties and oxidation resistance. In order to promote their structural applica- tions, the mechanical fatigue behaviour has to be stud- ied in detail. Fully lamellar (FL) and duplex (DP) are the micro- structural morphologies responsible for the most inter- esting mechanical behaviour of gamma-titanium aluminides: the fully lamellar microstructure shows the highest fracture toughness and fatigue crack growth resistance, whereas the duplex structures display improved tensile properties and better resistance to fatigue crack nucleation [1–3]. The mechanical properties of FL c-TiAl strongly depend on composition and microstructure. It has been shown that the grain size or colony size influences frac- ture toughness and tensile ductility of FL materials in an opposite manner, resulting in an inverse relationship between fracture toughness and ductility [1,4–7]. Fati- gue life and fatigue crack propagation resistance increase with the lamellar colony size, whereas finer lamellar microstructure is more resistant to crack nucleation [8–14]. The aim of this study is mainly the investigation of the influence of bending fatigue on residual mechanical properties. Thus, three Ti– 47Al–2Cr–2Nb–1B ingots with three different near- fully-lamellar (NFL) microstructures were prepared by vacuum induction melting (VIM) in a ceramic crucible, permanent mould casting and heat treatment without hot-isostatic-pressing (HIP) processing. Specific interest was given to the microstructural fea- tures influencing mechanical strength and ductility, fracture toughness and fatigue resistance of NFL c-TiAl A serious barrier to the industrial use of this class of materials comes from its high cost, also due to HIP treatment usually performed to reduce porosity. A proper understanding of the behaviour of this material under mechanical fatigue in the presence of micro- defects may well promote its structural employment in Corresponding author. Fax: +39-011-5644699. E-mail address: [email protected] (F. Marino). 0142-1123/$ - see front matter # 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijfatigue.2004.06.006

Effects of low-cycle fatigue on bending properties and fracture toughness of un-HIP’ed Ti–47Al–2Cr–2Nb–1B intermetallic

Embed Size (px)

Citation preview

� Corresponding author. Fax: +39-011-5

E-mail address: francesco.marino@poli

0142-1123/$ - see front matter # 2004 Els

doi:10.1016/j.ijfatigue.2004.06.006

644699.

to.it (F. Marino).

evier Ltd. All rights reserved.

International Journal of Fatigue 27 (2005) 143–153

www.elsevier.com/locate/ijfatigue

Effects of low-cycle fatigue on bending properties and fracturetoughness of un-HIP’ed Ti–47Al–2Cr–2Nb–1B intermetallic

Francesco Marino a,�, A. Rebuffo b, F. Sorrentino c

a Dipartimento di Scienza dei Materiali e Ingegneria Chimica, Politecnico di Torino, corso Duca degli Abruzzi 24, 10129 Torino, Italyb PHITEC INGEGNERIA S.r.l., corso Susa 299/A, 10098 Rivoli, Torino, Italy

c OSVAT S.r.l., 10040 Piobesi, Torino, Italy

Received 4 December 2003; received in revised form 19 May 2004; accepted 9 June 2004

Abstract

Ingots of c-TiAl base intermetallic alloy were cast in a metallic mould preheated at three different temperatures in order toobtain three different near-fully-lamellar microstructures. Four-point bending strength, bending strain to fracture, fracture tough-ness, hardness and low cycle bending fatigue strength ðrmaxÞL of near-fully-lamellar Ti–47Al–2Cr–2Nb–1B with different colony

sizes and interlamellar spacings were measured on un-HIP’ed test bars. The effects of low-cycle fatigue, 2 � 105 alternate bendingcycles (R ¼ 0:1), at high stress intensity levels, i.e. the fatigue limit, on mechanical behaviour of the three different microstructureswere evaluated. The data reported and discussed in this paper show that bending fatigue ‘‘training’’ enhances the mechanicalbehaviour of this material, mainly through the appearance of a plastic deformation during the bending test. The effects of themicrostructures on bending deformation, fracture behaviour and fatigue resistance were discussed as well.# 2004 Elsevier Ltd. All rights reserved.

1. Introduction

Fully lamellar gamma-titanium aluminides are underextensive research because of their low density,adequate high temperature properties and oxidationresistance. In order to promote their structural applica-tions, the mechanical fatigue behaviour has to be stud-ied in detail.

Fully lamellar (FL) and duplex (DP) are the micro-structural morphologies responsible for the most inter-esting mechanical behaviour of gamma-titaniumaluminides: the fully lamellar microstructure shows thehighest fracture toughness and fatigue crack growthresistance, whereas the duplex structures displayimproved tensile properties and better resistance tofatigue crack nucleation [1–3].

The mechanical properties of FL c-TiAl stronglydepend on composition and microstructure. It has beenshown that the grain size or colony size influences frac-ture toughness and tensile ductility of FL materials in

an opposite manner, resulting in an inverse relationshipbetween fracture toughness and ductility [1,4–7]. Fati-gue life and fatigue crack propagation resistanceincrease with the lamellar colony size, whereas finerlamellar microstructure is more resistant to cracknucleation [8–14]. The aim of this study is mainly theinvestigation of the influence of bending fatigueon residual mechanical properties. Thus, three Ti–47Al–2Cr–2Nb–1B ingots with three different near-fully-lamellar (NFL) microstructures were prepared byvacuum induction melting (VIM) in a ceramic crucible,permanent mould casting and heat treatment withouthot-isostatic-pressing (HIP) processing.

Specific interest was given to the microstructural fea-tures influencing mechanical strength and ductility,fracture toughness and fatigue resistance of NFLc-TiAl

A serious barrier to the industrial use of this class ofmaterials comes from its high cost, also due to HIPtreatment usually performed to reduce porosity. Aproper understanding of the behaviour of this materialunder mechanical fatigue in the presence of micro-defects may well promote its structural employment in

144 F. Marino et al. / International Journal of Fatigue 27 (2005) 143–153

un-HIP’ed condition, resulting in decreased productioncosts. Following this philosophy, the defect toleranceof titanium aluminides is under investigation [15,16].

In this work, HIP treatment was not utilized and theintegrity of all the samples was checked by radio-graphic testing.

2. Experimental procedures

Three ingots, 20 � 60 � 90 mm, of nominal compo-sition (at.%) Ti–47Al–2Cr–2Nb–1B with three differentNFL microstructures were cast in a metallic mouldafter induction melting of the pure elements (99.9%) ina zirconia crucible coated with plasma-sprayed yttria[17]. Boron was added, as a lamellar structure stabilizer[18], in the form of TiB2 powder (99.9%). The processwas carried out under an argon atmosphere. To reducecontamination of the melt, the crucible was coated withan intermetallic layer obtained from a previous cycle ofmelting with the same material.

The actual composition of the three ingots is indi-cated in Table 1.

The aim was to obtain three different NFL micro-structures, therefore the copper mould was preheatedto 300, 500 and 600

vC to cast ingots no. 1, no. 2 and

no. 3, respectively.The three ingots were heat treated below the eutec-

toidic temperature in a sealed quartz tube filled withargon: 4 h at 980

vC—air cooling to room temperature

ðRTÞ þ 8 h at 700vC—furnace cooling to RT þ 24 h

at 815vC—air cooling to RT [19]. This annealing treat-

ment was chosen to avoid any transformation in theNFL microstructure obtained in the as-cast condition.It was previously selected for the improvement in frac-ture toughness [17].

Small bars (3 � 8 � 50 mm) were cut from each ingotby spark erosion for mechanical testing. Samples werepolished using SiC papers.

As reported in the Introduction, the ingots were notsubmitted to HIP treatment. The samples, taken fromthe sound part of the ingot, i.e. only rejecting the area

close to the shrinkage cone, were selected by radio-graphic examinations performed according to ASTM E1742-95 standard. Image quality indicator (IQI) con-forming to Fig. 1 page 4 of the E 1742-95 standard(step-hole, MIL type) was placed on Ti–6Al–4V blocks0.10–0.12 in. thick. Quality level obtained is 2-2T.

Image quality determined as above is a conventionaldetermination. In practice, on the radiographic imageof a block having the same thickness of the specimen, acylindrical discontinuity with 0.02 in. diameter and0.0024 in. height (2% of 0.12 in.) is detectable. Thesame is to be expected of the specimen image, placedon the same radiograph.

According to this test, 5% of the bars was rejected.Mechanical tests were performed at RT. Four-point

bending strength (modulus of rupture, MOR) was mea-sured utilizing a servohydraulic machine equipped witha load cell of 5 kN using a crosshead rate of 5 mm/min.Outer and inner spans of 40 and 20 mm, respectively,were used. Fracture toughness was evaluated by three-point bending tests of Single Edge Precracked Beam(SEPB) [20] specimens (3 � 4 � 25 mm) using a span of16 mm and a crosshead rate of 2 mm/min, according toASTM E 399 standards. Hardness (HV) was measuredby a Vickers indentor using a 10 kg maximum load. Allthe reported values represent the average of at least fivetests. The spread of the results for MOR and fracturetoughness measured values are shown in Table 2.

Fatigue experiments were carried out in four-pointcyclic bending mode with spans of 40 and 20 mm,applying a 10 Hz frequency sinusoidal wave and astress ratio R ¼ rmax=rmin ¼ 0:1. Fatigue tests wereperformed using a servohydraulic machine with a loadcell of 5 kN and specimens of 3 � 8 � 50 mm; low cyclebending fatigue strength ðrmaxÞL was statistically

determined at 2 � 105 cycles with 80% reliability,applying the ‘‘stair case’’ method.

In order to examine the effects of low-cycle fatigue athigh stress levels on mechanical behaviour of NFL c-TiAl, the residual flexural strength, fracture toughnessand hardness of the three different microstructures weremeasured on those specimens which had survived fatigue

Table 1

Chemical composition and microstructural parameters of Ti–47Al–2Cr–2Nb–1B ingots after heat treatment

Ingots

Microstructure C olony size

(lm)

I

(

nterlamellar spacing

nm)

S

(

ize of equiassic c grains

lm)

No. 1

NFL 2 00–500 2 50 5 –10

No. 2

NFL 5 00–1000 4 00 5 –25

No. 3

NFL 7 00–1200 5 00 5 –30

Chemical composition

(wt%)

T

i Al C r N b B O H N Z r þ Y

No. 1 B

alance 32.71 2 .83 4 .57 0 .26 0 .097 0 .001 0 .002 < 0.1

No. 2 B

alance 32.75 2 .78 4 .59 0 .27 0 .096 0 .001 0 .002 < 0.1

No. 3 B

alance 32.81 2 .81 4 .49 0 .26 0 .099 0 .001 0 .002 < 0.1

F. Marino et al. / International Journal of Fatigue 27 (2005) 143–153 145

testing. Microstructural features, crack paths and fracture

surfaces were observed using both optical microscopy

(OM) and scanning electron microscopy (SEM).As each of the three ingots is characterized by a

particular microstructure, at times they will be ident-

ified as microstructure nos. 1–3.

3. Results

The optical micrographs reported in Fig. 1 refer to

ingot no. 1 (a); ingot no. 2 (b); ingot no. 3 (c) and they

are representative of the heat treated microstructure,

which consist of lamellar colonies and a small amount

of fine c grains. In particular Fig. 1(c), due to the lar-

ger grain size, cannot show one complete grain at the

same magnification of (a) and (b). The NFL micro-

structures of the three ingots are different in lamellar

colony or grain size (GS) and interlamellar spacing, as

well as in the amount and size of equiaxed c phase as

indicated in Table 1.Fig. 1(d), ingot no. 2 in the as-cast condition, shows

a phase with an elongated ribbon morphology and a

gently curving, sinuous growing habit which is present

Fig. 1. Optical micrographs of heat treated ingot (a) no. 1; (b) no. 2; (c) no. 3 and (d) as-cast ingot no. 2

Table 2

Mechanical properties after heat treatment

I

ngot no. 1 Ingot no. 2 Ingot no. 3

B

efore fatigue After fatigue� B efore fatigue After fatigue� B efore fatigue A fter fatigue�

HV (100 N)-as cast 4

00 – 3 80 – 3 70 –

HV (100 N)-after heat treatment 3

90 390 3 35 335 3 20 3 20

MOR (MPa) 6

25 � 20 700 � 15 4 80 � 30 575 � 25 4 20 � 38 5 60 � 33

Crosshead displacement to failure (mm) 0

.37 0.52 0 .34 0.48 0 .3 0 .46

KIc (MPa m1/2) 3

2 � 0:5 33 � 0:5 1 9 � 0:5 21 � 0:5 1 3 � 0:5 1 6 � 0:5

ðrmaxÞL 4

40 4 25 3 75

ðrmaxÞL=MOR 0

.70 0 .88 0 .89

MOR=MOR� 1

.12 1 .20 1 .33

� Mechanical properties were measured after 2 � 105 four-point bending cycles at ðrmaxÞL with R ¼ 0:1 and f ¼ 10 Hz; ðrmaxÞL is the four-

point alternate bending fatigue strength after 2 � 105 cycles with R ¼ 0:1 and f ¼ 10 Hz.

146 F. Marino et al. / International Journal of Fatigue 27 (2005) 143–153

in all the ingots. This phase is randomly distributed

and it was not observed in B-free alloys of similar com-

position. X ray diffraction (XRD) analysis suggests

that the ribbon phase is TiB2, as also previously indi-

cated [21,22].No appreciable microstructural changes were

observed after the heat treatment. Only slight coarsen-

ing of lamellar colonies and of c phase, accompanied

by the growth of a few c grains along the colony

boundaries, can be observed in Fig. 1(a)–(c).Mechanical properties of the heat-treated material

are reported in Table 2. Four-point bending strength

was first measured. The load vs. displacement curve for

a specimen obtained from ingot no. 3 is shown in

Fig. 2. The curve is representative of the bending

behaviour of all the ingots. As indicated in Fig. 2, none

of the tested specimens show evidence of diffuse plastic

deformation.Fracture toughness tests show an average KIc value

of 32 � 0:5 MPa m1=2 for ingot no. 1, whereas the low-

est KIc value was displayed by the specimens with

microstructure no. 3 (KIc ffi 13 � 0:5 MPa m1=2). Trans-

verse optical micrographs of cracks and SEM fracture

surfaces are shown in Fig. 3. The SEM micrograph

reported in Fig. 3(a) shows evidence of interlamellar

fracture, whereas Fig. 3(b) testifies translamellar frac-

ture and many intergranular secondary cracks are also

evident. Fig. 3(b) refers to the ingot with microstruc-

ture no. 3 showing the presence, in hard orientation in

relation to the crack plane, of debonded lamellae with

small cleavage facets. Fracture surfaces of the fine

microstructure of ingot no. 1, Fig. 3(c), appear rougherdue to crack deflection activation.

As shown in Table 2, ingot no. 1 displays a low-cyclefatigue strength ðrmaxÞL of about 440 MPa; hence, itsspecimens which survived fatigue testing were cycled inbending mode at a ðrmaxÞL=MOR ratio equal to 0.70.

Ingots nos. 2 and 3 display lower fatigue strength andtheir ðrmaxÞL=MOR ratio is almost equal to 0.90. Thus,

even though fatigue strength of ingot no. 1 is the high-est, the specimens with microstructures nos. 2 and 3reach higher ðrmaxÞL=MOR ratios than the finer micro-

structure no. 1. Fatigue testing shows that alternatebending lifetime is greatly reduced at stress levels justabove ðrmaxÞL; in fact, the S–N curves, Fig. 4, are fairly

flat.Fractographic observations on the specimens

obtained from ingots nos. 1–3 which failed during fati-gue tests (Fig. 5(a)) indicate that fracture starts at weakinterfaces by decohesion of lamellae parallel to thefracture surface (area A) and then propagates tortu-ously between the grains and translamellarly (Fig. 5(b)).Many secondary cracks depart perpendicularly fromfracture surfaces (Fig. 5(c)) and are arrested in theirpropagation by the unfavourable lamellae orientations(Fig. 5(d)). Debonded lamellae on secondary crack sur-faces are also discernible in Fig. 5(e). Fig. 5(f) provesthe existence of small zones (area B) characterized bysome plastic deformation, typical of fatigue cycling.Secondary cracks at grain boundaries are alsoobserved.

As already mentioned above, in order to investigatethe effects of low-cycle fatigue at high stress levels onmechanical behaviour of FL c-TiAl, the residual flex-ural strength, fracture toughness and hardness weremeasured on specimens which survived the fatigue test-ing. The mechanical properties after 2 � 105 alternatebending cycles at rmax equal to the ðrmaxÞL are indi-

cated in Table 2. An increase in flexural strength andelongation to failure can be observed for all samples.Conversely, no increase in hardness was observed. Thisbehaviour testifies the singular influence of fatigue‘‘training’’ at ðrmaxÞL on mechanical properties of the

cast ingots. The load vs. displacement curve reported inFig. 6, which is representative of bending behaviour of

all the cast ingots after 2 � 105 alternate bending cyclesat the correspondent ðrmaxÞL=MOR ratio, points out

that specimens show plastic strain to fracture, nonexist-ent before fatigue ‘‘training’’.

The ductility after fatigue ‘‘training’’ is confirmed bythe increased fracture toughness (Table 2).

The analysis of fracture surfaces confirms anincreased ductility of lamellar colonies and a largeramount of translamellar fracture. In fact, many sec-ondary cracks depart perpendicularly from fracturesurfaces (Fig. 7). The cracks are arrested in their

Fig. 2. Four-point bending stress vs. displacement curve shown by a

specimen from ingot no. 3 and representative of the bending behav-

iour of all the ingots.

F. Marino et al. / International Journal of Fatigue 27 (2005) 143–153 147

S–N curves for microstructures of ingots nos. 1–3; specimens were tested at a rate of 10 Hz with R

Fig. 4. ¼ 0:1.

our-point bending fracture surfaces (SEM) show: (a) interlamellar decohesion; and (b) mostly translamellar fracture with

Fig. 3. F secondary

cracks between colony boundaries in specimens from ingot no. 3; (c) translamellar fracture surface in specimens with microstructure no. 1, which

contains fine size colonies, thus yielding quite a rough fracture surface.

148 F. Marino et al. / International Journal of Fatigue 27 (2005) 143–153

propagation by deformation and delamination of thelamellae in unfavourable orientations: Fig. 8(a) showshow the deformation of a lamella can arrest the trans-lamellar propagation of secondary cracks. As depictedin Fig. 8(a) and (b), secondary crack surfaces aresmooth and debonded lamellae show limited or nodebonded facets.

4. Discussion

According to the results reported in Results, it can

be pointed out that the RT mechanical behaviour of

NFL c-TiAl is strongly dependent on microstructure as

well as on deformation history. Such aspects will be

discussed here in detail.

. 5. Fracture surface of specimens from ingots no. 2 (a–c) and no. 3 (d–f ) after four-point alternate bending fatigue failu

Fig re.

F. Marino et al. / International Journal of Fatigue 27 (2005) 143–153 149

4.1. Mechanical properties of the ‘‘virgin’’ material

Four-point bending strength and fracture toughness

results for the annealed material (Table 2) confirm that

mechanical properties of fully lamellar Ti–47Al–2Cr–

2Nb–1B are strongly affected by the average grain size

and interlamellar spacing.As observed in the load vs. crosshead displacement

curves (Fig. 2), the alloys investigated in this work

show no macroscopic plastic strain to failure, this

probably being related to the elevated yield stress,

almost equal to flexural strength.

According to other investigators [1,4], ductility andflexural strength of the three ingots increase with thedecreasing of grain size and interlamellar spacing. Thisbehaviour supports the theory regarding the existenceof a Hall-Petch type relationship between yield strengthand lamellar colony size as well as interlamellar spa-cing in FL microstructures. This also explains whymicrostructure no. 1 shows the highest bending proper-ties. In spite of relatively low bending ductility, ingotno. 1 shows an excellent combination of strength(625 MPa) and fracture toughness (32 MPa m1/2). Thisoccurs because, as reported by Kim [1], the effect of GSon fracture toughness is in competition with lamellarspacing: KIc increases with lamellar GS up to about500 lm, but then decreases for larger grain size.

As reported by other researchers [1,23], the highbending properties and fracture toughness measured oningot no. 1 confirm the increase of bending strengthand fracture toughness with the decreasing of inter-lamellar spacing.

In spite of lower fracture toughness and four-pointbending strength, samples nos. 2 and 3, characterizedby coarser lamellar colonies and lamellae, still haveflexural strength higher than many alloys of the samecomposition [1,23,24]. The superior crack growthresistance of coarse lamellar microstructures may beseen as questionable due to the scale of the microstruc-ture in relation to the sample size employed in this andother several studies [25] (2–6 lamellar colonies existacross the thin section of the samples and true poly-cristalline constraint is not developed). Furthermore, asstated by Kim [1], KIC might be closely related to theintrinsic ductility of the lamellar structure because thecrack-tip plastic zone should encompass only one orpart of a grain.

Optical and SEM observations of fracture surfacesof the process zone of precracked specimens indicatethat mechanical strength and fracture toughness alsodepend on lamellar laths orientation in relation to theapplied stress and crack orientation. Such observationsconfirm previous investigations presented in literature[1–4]: toughness of lamellar structures in hard orienta-tions increases with the decreasing of lamellar spacingto levels much higher than in easy orientations.

An example of crack-tip shielding zone ahead of themain crack is shown in Fig. 9(a). This optical micro-graph reports the crack growth through the formationof microcracks ahead of the main one; during crackpropagation, the unbroken links between secondarycracks shrink, leading to the bridging by shear liga-ments in the crack wake. The bridging by crack wakeligaments, shown in Fig. 9(b), could be regarded as oneof the main toughening mechanisms active in FL c-TiAl alloys. According to others [25], OM crack pathinvestigation has also depicted that coarse lamellarmicrostructures nos. 2 and 3 exhibit relatively larger

Fig. 6. Four-point bending stress vs. displacement curve representa-

tive of bending behaviour of the microstructure of the cast ingots

after alternate bending fatigue ‘‘training’’ at rmax ¼ ðrmaxÞL.

. Fracture surface of ingot no. 2: deformation and decohe

Fig. 7 sion

of lamellae observed on fracture surfaces of four-point bending tested

specimens after fatigue ‘‘training’’.

150 F. Marino et al. / International Journal of Fatigue 27 (2005) 143–153

size crack wake ligaments than microstructure no. 1.On the other hand, the finer FL microstructures dis-play a larger number of bridging ligaments.

SEM analysis of fracture surfaces on bended speci-mens also shows that fracture toughness of fully lamel-lar titanium aluminides is related to plasticdeformation ahead of the crack tip before crack propa-gation.

Fracture surfaces are rough, especially in specimenswith finer microstructure (no. 1), whereas the lamellaeof coarser structures show small or no cleavage facetsas a result of their higher intrinsic ductility. Therefore,as stated by other researchers [1,4,5], when GS is largerthan the crack-tip plastic zone, the intrinsic ductility ofthe lamellar structure contributes to crack growtharrest and ligaments formation between secondarycracks ahead of the main crack and in the crack wake.

Scanning electron micrographs reported in the pre-vious section testify that NFL c-TiAl exhibits delami-

nation or interlamellar fracture and translamellar

fracture, depending on lamellar orientation with

respect to the crack path or the applied stress axis.

When the crack tip encounters grains containing laths

at hard orientations with respect to the crack plane, a

higher resistance to the crack growth occurs and the

fracture features appear similar to those shown in

Fig. 3(b). Conversely, when the crack tip encounters

lamellar grains containing laths parallel to the crack

plane, crack propagates along the weak interface;

therefore, flat regions (labelled A) in Fig. 3(a) can be

observed on fracture surfaces.On the basis of the results obtained, it is possible to

point out that the fracture process in lamellar micro-

structure has important composite-like features, owing

to the presence of the hard a2 and of the relatively soft

c phases in the lamellar grains which gives the lamellar

colony a large intrinsic ductility [26].

ctographic features of bend test specimens from ingots nos. 2 and 3 after fatigue ‘‘training’’: (a) thin sheet toughening;

Fig. 8. Fra (b) rounded

secondary crack tip.

nt bending fracture surfaces (SEM) report (a) and (b) a transversal view of an incomplete crack showing bridgin

Fig. 9. Four-poi g by crack-wake

ligaments.

F. Marino et al. / International Journal of Fatigue 27 (2005) 143–153 151

4.2. Low-cycle fatigue behaviour

Fatigue behaviour is also strongly dependent onmicrostructural parameters such as GS, interlamellarspacing and lamellae orientation.

FL titanium aluminides exhibit a fatigue limit some-times approaching and exceeding 90% of ultimatestrength [4,7,27] and cyclic hardening under cyclicstrain at a constant amplitude [27,28].

Some studies [6,7] report a low-cycle fatigue limitequal or higher than yield stress. Ingots nos. 2 and 3display a bending fatigue limit after 2 � 105 cycleswhich is 90% of flexural strength. Since the bendingyield strength of the investigated ingots approachesflexural strength, the fatigue data reported in the pre-vious section indicate for FL c-TiAl alloys a low-cyclefatigue limit lower than the yield stress, as previouslyindicated [6–10]. The lower ðrmaxÞL=MOR ratio (equal

to 0.70) after 2 � 105 alternate bending cycles, shownby the specimens with microstructure no. 1, testifiesthat resistance to crack propagation decreases with theaverage colony size; in spite of this, ingot no. 1 showsthe highest bending fatigue strength, 440 MPa, becauseof its superior bending properties.

On the other hand, specimens from ingots nos. 2 and3 show higher ðrmaxÞL=MOR ratios, equal to about 0.9,than specimens with microstructure no. 1 because oftheir coarser lamellar colonies. These data point to theexistence of a large average colony size which corre-sponds to the maximum ðrmaxÞL=MOR ratio for a

given composition and microstructure; thus, the resist-ance to crack propagation decreases with a furtherincrease of lamellar colony size.

The S–N curves (Fig. 4) are fairly flat with a lowvalue of stress life exponent. Some previous experi-ments [6] revealed that the shape of the S–N curve ofFL c-TiAl in the low-cycle fatigue range (LCF,

N < 106) can be quite different from that in the high-cycle fatigue (HCF) regime (>106). Therefore, extrapol-ation of the stress life from the LCF regime to theHCF regime could result in an overestimation of thefatigue life. In this work, the threshold value of 106

cycles has been considered for LCF and HCF regimebecause many authors [6] report LCF fatigue data forc-TiAl up to 106, even if the classical value is 104

cycles.The flat S–N curves shown in Fig. 4 are probably a

consequence of the steep da=dN curve shown by c-TiAl intermetallics [2,11–13,25] and this may confirmthe existence of a large crack growth threshold even ifthis hypothesis has sometimes been rejected [7,11].

According to fatigue resistance data reported bymany investigators [2,9,11], the elevated low-cycle fati-gue strength shown by the FL ingots with coarse col-ony size can be attributed to their larger resistance to

large crack propagation: in FL microstructures smallcracks can easily nucleate between lamellae at appliedstresses below the large-crack growth threshold (DKth),especially in favourably oriented lamellar colonies, bydecohesion of interlamellar slip bands; the interlamellarnucleated cracks can easily propagate up to the colonyboundary where they remain arrested until stresses areincreased to a level that allows further fatigue crackgrowth into the neighbouring colonies. Resistance tofatigue large-crack propagation should thereforeincrease with grain size along with fatigue strength, asthe S–N curve shown in Fig. 4 confirms: the flattestcurves correspond to the two coarser grained micro-structures (nos. 2 and 3).

Fractographic observations after alternate-bendingfatigue failure indicate that the microcracks have nucle-ated and grown in a brittle manner. As shown inFig. 5(a), the fracture seems to start from the specimensurface by decohesion of lamellae parallel to the frac-ture surface and then propagates tortuously betweengrains and lamellae (Fig. 5(b)). SEM analysis alsoshows that the crack path propagates between weakinterfaces whereas a high resistance to the crack growthis offered by grains containing lamellae at hard orienta-tions [9,13] (Fig. 5(a)).

Fracture surfaces for fatigue tested specimens aredifferent from those shown after four-point bendingfailure: the presence of debonded lamellae with largercleavage facets on secondary crack surfaces, Fig. 5(b),indicates a decreased thin sheet toughening. Fig. 5(c)shows the existence of small zones in the compression-stressed parts of the specimens, testifying evident plas-tic deformation typical of fatigue cycling in metallicmaterials.

According to fracture surfaces morphology, crackpath observations on fatigue tested specimens suggestthat crack-tip shielding mechanisms, such as crackdeflection, roughness-induced closure and shear liga-ment formation between secondary cracks, enhancefatigue crack growth resistance as well as fracturetoughness.

Based on these results and confirmed by the LCFdata, an increase in fatigue crack resistance of FLalloys is expected with the increase of lamellar colonysize in consideration of the enhancement of crack-tipshielding mechanisms. Such behaviour of the lamellarstructures has also been reported by other investigators[2,6,25].

4.3. Mechanical properties of the fatigue ‘‘trained’’material

In accordance with the outcome of the presentresearch, the term ‘‘trained’’ material is used whenreferring to those samples of c-TiAl where some lamel-lae piles, reciprocally freed by the formation of a thin

152 F. Marino et al. / International Journal of Fatigue 27 (2005) 143–153

microcracks network due to fatigue stress, may, within

limits, slide reciprocally because of semicoherent inter-

faces between the a2 and c lamellae. This argument

seems to fall in line with previous literature [29]. In this

condition, the mechanical behaviour of the material

can be interpreted as a multilayered-microcomposite

[26,30]. As a consequence not only toughening mechan-

isms, like microcracking, but also twin-sheet toughen-

ing are activated by low-cycle fatigue.Said behaviour may supply an explanation for the

experimental data regarding the increased values of Kc

and MOR after fatigue.The residual flexural strength, Vickers hardness and

fracture toughness of the different microstructures

investigated after training were measured after 2 � 105

alternate bending cycles at r ¼ ðrmaxÞL. Other research-

ers [6,7,14,28] have indicated only cyclic hardening of

lamellar c-TiAl under stress cycling at a constant

amplitude and a fatigue strength higher than yield

strength but data reported in Table 2 show that fatigue

‘‘training’’ at ðrmaxÞL increases ductility, fracture

toughness as well as flexural strength of FL Ti–47Al–

2Cr–2Nb–1B.Macroscopic values of hardness do not increase, tes-

tifying that the improvement of mechanical properties

for the material investigated could be related to the

enhancement of fracture toughness.As indicated in the previous section, the load vs.

crosshead displacement curves obtained after 2 � 105

bending cycles at r ¼ ðrmaxÞL show that FL micro-

structures possess relatively remarkable bending plastic

strain to failure, while their behaviour is simply linear-

elastic before fatigue ‘‘training’’. The analysis of frac-

ture surfaces confirms the beneficial effect of low-cycle

fatigue cycling at ðrmaxÞL on mechanical properties of

the FL microstructures, showing an increased intrinsic

ductility of lamellar colonies especially for ingots nos. 2

and 3. The increased fracture toughness values and the

macroscopic ductility of the fatigued specimens can be

attributed to the increase of intrinsic ductility of lamel-

lar colonies. The larger ductility ahead of the crack tip

may arise from lamellae twin toughening due to alter-

nate bending cycles and/or to a possible fatigue-acti-

vated toughening mechanism at lamellae interfaces

(Fig. 8). Accordingly, the micrograph shown in

Fig. 8(a) exhibits larger thin sheet toughening and

translamellar fracture. Many secondary cracks depart

perpendicularly from fracture surfaces (Fig. 8) being

arrested in their propagation by deformation and

delamination of lamellae in unfavourable orientations

(Fig. 8(a)).Secondary-crack surfaces are smooth with few

debonded lamellae: this suggests an increased thin sheet

toughening mechanism. Rounded secondary crack tips

shown in Fig. 8(b) may contribute to increasingstrength and local ductility.

Still according to experimental evidence, the higherthe maximum applied stress, normalized by the ulti-mate flexural strength during fatigue cycling, the largerthe improvement of mechanical properties of NFLmicrostructures. Thus, toughening mechanisms can beactivated by the high stress intensity levels of bendingcycles at rmax ¼ ðrmaxÞL, whereas the quantitativeenhancement of bending properties, MOR and plasticstrain of NFL microstructures depends on their ðrmaxÞL

=MOR ratio.

5. Conclusions

The results of the mechanical tests performed testifythat even in the presence of size-selected microdefects,it is possible to define a fatigue limit which allows thestructural employment of un-HIP’ed c-TiAl compo-nents.

The present study confirms the strong influence ofmicrostructure on fatigue behaviour of NFL c-TiAlbase intermetallics, in this case un-HIP’ed Ti–47Al–2Cr–2Nb–1B.

Some general features can be summarized as follows:

1. The ðrmaxÞL=MOR ratio increases with the grain size.2. MOR and fracture toughness increase after fatigue

‘‘training’’.3. Appearance of a macroscopic plastic behaviour in

r–e plot after fatigue.4. The larger the ðrmaxÞL=MOR ratio applied to each

microstructure during fatigue ‘‘training’’, the largerthe improvement of the mentioned mechanicalproperties.

5. Unchanged values of hardness before and after fati-gue.

On the base of the remarks reported thus far, inorder to explain the enhancement of after-fatiguemechanical features, it is feasible to put forward aninterpretation in terms of activation of tougheningmechanisms resulting from low-cycle fatigue.

References

[1] Kim YW. Effects of microstructure on the deformation and

fracture of gamma-TiAl alloys. Mater Sci Eng 1995;A 192/193:

519–33.

[2] Gnanamoorthy R, Mutoh Y, Mizuhara Y. Fatigue crack growth

behavior of equiaxed, duplex and lamellar microstructure

gamma-base titanium aluminides. Intermetallics 1996;4:525–32.

[3] Recina V, Karlsson B. Tensile properties and microstructure of

Ti–48Al–2W–0.5Si gamma-titanium aluminide at temperatures

between room temperature and 800vC. Mater Sci Technol

1999;15:57–66.

F. Marino et al. / International Journal of Fatigue 27 (2005) 143–153 153

[4] Chan KS. Understanding fracture toughness in gamma TiAl.

JOM 1992;44(5):30–8.

[5] Yamaguchi M, Inui HS, Yokoshima K, Kishida D, Johnson R.

Recent progress in our understanding of deformation and frac-

ture of two-phase and single-phase Ti–Al alloys. Mater Sci Eng

1996;A213:25–31.

[6] Chan KS. The fatigue resistance of TiAl-based alloys. JOM

1997;49(7):53–8.

[7] Kim YW. Ordered intermetallic alloys, part III: gamma titanium

aluminides. JOM 1994;46(7):30–40.

[8] Dowling Jr WE, Worth BD, Allison JE, Jones JW. In: Kim YW,

et al., editors. Microstructure/property relationships in titanium

aluminides and alloys. Warrendale, PA, USA: TMS; 1991.

p. 123–133.

[9] Chan KS, Shih DS. Fundamental aspects of fatigue and fracture

in a TiAl sheet alloy. Metall Mater Trans A 1998;29A(1):73–87.

[10] Shih DS, Huang SC, Scarr GK, Jang H, Chesnutt JC. In: Kim

YW, et al., editors. Microstructure/property relationships in

titanium aluminides and alloys. Warrendale, PA, USA: TMS;

1991. p. 135–148.

[11] Kruzic J, Campbell JP, Ritchie RO. Fatigue-crack propagation

in gamma-based titanium aluminide alloys at large and small

crack sizes. Proceedings of the MRS Fall Meeting, Boston,

USA, 552, 1998. p. 2.3.1–2.3.7.

[12] Kwai S, Chan KS, Donald S, Shih DS. Fatigue and fracture

behavior of a fine-grained lamellar TiAl alloy. Metall Mater

Trans A 1997;28A(1):79–90.

[13] Choi C, Lee Y, Lee CS. Microstructural influence on the fatigue

crack propagation of a gamma-TiAl alloy. Scripta Mater

1997;36(7):821–7.

[14] Soboyejo WO, Aswath PB, Mercer C. Cyclic deformation and

fracture mechanisms in a gamma-based titanium aluminide

intermetallic. In: Soboyejo WO, et al., editors. Fatigue and frac-

ture of ordered intermetallic materials, II. Warrendale, PA,

USA: TMS; 1995. p. 57–84.

[15] Lerch BA. Effect of defects on the fatigue life of gamma-TiAl.

HITEMP Rev 1999;2:1–11 NASA CP-1999-208915.

[16] Nazmy M, Staubli M, Onofrio G, Lupinc V. Surface defect toler-

ance of a cast TiAl alloy in fatigue. Scripta Mater 2001;45:787–92.

[17] Rebuffo A. Comportamento meccanico di un materiale inter-

metallico per la realizzazione di valvole destinate a motori endo-

termici. Thesis, Politecnico di Torino, Torino, Italy, 1997.

[18] Larsen DE, Kampe S, Christodoulou L. Microstructure mor-

phology effects on fracture toughness of cast gamma titanium

aluminides. Proceedings of the MRS Symposium, 194, 1990.

p. 285–292.

[19] Soboyejo WO, Schwartz DS, Sastry SML. An investigation of

the fracture behavior of gamma-based titanium aluminides:

effects of annealing in the alfa þ gamma and alfa2 þ gamma

phase fields. Metall Trans A 1992;23A:2039–59.

[20] Pancheri P, Bosetti P, Dal Maschio R, Sglavo VM. Production

of sharp cracks in ceramic materials by three point bending of

sandwiched specimens. Eng Fract Mech 1997;59(4):445–7.

[21] Schwartz DS, Shih DS. The influence of boron addition in cast

Ti–47Al–2Cr–2Nb–0.8B. Proceedings of the MRS Symposium,

Pittsburgh, PA, USA, 364, 1995. p. 787–791.

[22] Godfrey AB, Loretto MH. The nature of complex precipitates

associated with the addition of boron to a gamma-based

titanium aluminide. Intermetallics 1996;4:47–53.

[23] Liu CT, Schneibel JH, Maziasz PJ, Wright JL, Easton DS. Ten-

sile properties and fracture toughness of TiAl alloys with con-

trolled microstructures. Intermetallics 1996;4:429–40.

[24] Huang SC, Hall HL. Plastic deformation and fracture of binary

TiAl-base alloys. Metall Trans A 1991;22A:419–27.

[25] Campbell JP, Venkateswara Rao KT, Ritchie RO. The effect of

microstructure on fracture toughness and fatigue crack growth

in c-titanium aluminide based intermetallics. Metall Mater Trans

A 1999;30A:563–77.

[26] Dogan B, Schoneich D, Schwalbe KH, Wagner R. Fatigue pre-

cracking and fracture toughness testing of TiAl intermetallics.

Intermetallics 1996;4:61–9.

[27] Umakoshi Y, Yasuda HY, Nakano T. Plastic anisotropy and

fatigue of TiAl PST crystals: a review. Intermetallics 1996;4:

S65–S75.

[28] Umakoshi Y, Yasuda HY, Nakano T. Plastic anisotropy and

fracture behavior of cyclically deformed TiAl polysynthetically

twinned crystals. Mater Sci Eng 1995;A192/193:511–7.

[29] Bowen P, Rogers NJ, James AW. Fracture and fatigue of cast

gamma TiAl based aluminides. Proceedings of the Conference

on Gamma Titanium Aluminides. Las Vegas, NV, February.

Warrendale, PA, USA: TMS; 1995. p. 849–865.

[30] Kuang JP, Harding RA, Campbell J. Microstructures and mech-

anical properties of an investment cast gamma titanium alumi-

nide. Mater Sci Technol 1999;15:840–50.