Effect of Broaching on High-Temperature Fatigue Behaviorin Notched Specimens of INCONEL 718

T. CONNOLLEY, M.J. STARINK, and P.A.S. REED

Notches were machined in specimens of INCONEL 718 by a broaching process, where differingbroaching runs led to differing extents of subsurface deformation and surface roughness. Fatigue testswere carried out at 600 °C with a trapezoidal loading waveform at 0.25 Hz. The broaching processthat led to the more severe subsurface deformation (but lower surface roughness) showed the worstfatigue performance. Analysis of total strain amplitude in the notch root with the aid of an elasto-plastic finite-element (FE) model showed that the work hardening related to the subsurface deform-ation caused by the different broaching can account for the difference in fatigue lives. Differences ininitiation and growth behavior were seen for the two broached finishes as well as for broached andsubsequently polished samples. These differences are discussed in terms of a change in crack growthinitiation and growth mechanisms due to the presence of the work-hardened layer.

I. INTRODUCTION

IN the broaching process, a cutting tool with multipleshaped teeth is drawn through a hole or over a surface toremove material by axial cutting.[1] Figure 1 shows a schematicdiagram of a broaching tool and its application in the machin-ing of a U-notch specimen of the type used in this study (Fig-ure 2). With appropriate tool design, complex profiles canbe machined with the single pass of a tool. In gas turbines,the blades are usually attached to the rims of turbine discsusing a fir-tree root fixing and the complex notch geometryin the rim is commonly produced by broaching. The stressconcentration effect of the notch means that fatigue initiationin this region is more likely. One type of in-service damageto turbine discs that must therefore be considered is low-cyclefatigue (LCF) cracking in the fir tree root fixtures.

For the present work, the INCONEL* 718 (IN718) nickel-

known about high-temperature crack initiation and short crackgrowth in this alloy. Various studies on a range of super-alloys have identified high-temperature crack initiation at slipbands,[13,14] inclusions,[15] precracked carbide particles,[16] andpreferentially oxidized carbides.[17–20] Volume expansion ofoxidizing primary carbides was identified as a mechanism ofenvironmentally assisted crack propagation in notch-rupturetests on IN718.[21] The role of oxidized primary carbides infatigue crack initiation has also been studied.[22–25]

Additionally, there is an interest in understanding crackbehavior in notched geometries such as the fir-tree root fix-ings in a turbine disc rim. The natural initiation and growthof short cracks in U-notch IN718 specimens has been stud-ied at 600 °C in air and reported in detail elsewhere by theauthors.[25] Fatigue tests were conducted under load controlusing a 1-1-1-1 trapezoidal waveform, on specimens withas-broached and polished U notches. Multisite crack initiationoccurred in the notch root. Many of the cracks initiated atbulgelike features formed by volume expansion of oxidiz-ing (Nb,Ti)C particles. In unstressed samples, oxidation of(Nb,Ti)C particles occurred readily, producing characteristicsurface eruptions. Scanning electron microscopy on metal-lographic sections revealed some subsurface (Nb,Ti)Coxidation and localized matrix deformation around oxidizedparticles. A mechanism for crack initiation by subsurface car-bide expansion during oxidation was proposed. Surface shortcrack growth rates in the notch root of polished specimenswere measured using an acetate replica technique. Observedshort crack growth rates were approximately constant acrossa wide range of crack lengths. However, there was a transi-tion to rapid, accelerating crack growth once cracks reachedseveral hundred microns in length. This rapid propagation inthe latter stages of the fatigue life was assisted by crack coa-lescence. Polishing the U notch to remove broaching marksresulted in a pronounced increase in fatigue life.

The purpose of this article is to compare fatigue crack behav-ior in two batches of IN718 specimens with broached Unotches, where separate broaching runs led to different initialnotch roughness and levels of subsurface deformation. Thisincludes study of the natural initiation and propagation char-acteristics of short cracks in broached notched IN718 speci-mens at 600 °C in air (conditions similar to those experienced

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 35A, MARCH 2004—771

T. CONNOLLEY, Research Engineer, is with the National Centre for Bio-medical Engineering Science, National University of Ireland, Galway, Ireland.M.J. STARINK and P.A.S. REED, Readers, are with the Materials Group,School of Engineering Sciences, University of Southampton, Southampton,SO17 1BJ, United Kingdom. Contact e-mail: pasr1@soton.ac.uk

Manuscript submitted April 8, 2003.

*INCONEL is a trademark of INCO Alloys International, Huntington,WV.

based superalloy was chosen due to its current and projectedwidespread use for gas turbine discs. The majority of pub-lished LCF studies on this alloy have concentrated on theeffects of temperature and environment on long crack propa-gation rates. This has included studies of mechanisms ofgrain-boundary oxidation and its influence on fatigue crackpropagation.[2,3,4] Dwell times at maximum or minimum loadwere generally found to increase crack propagation rates,with the dwell permitting the time-dependent oxide embrit-tlement of grain boundaries.[5–11] Most of these studies wereon long cracks using compact tension (CT) specimens, butthe hold time effect has also been observed for preinitiatedcracks in U-notch specimens tested in bending.[12]

Compared to the wealth of knowledge on IN718 obtainedfrom laboratory long crack fatigue testing, relatively little is

03-186A-4.qxd 2/7/04 2:53 PM Page 771

772—VOLUME 35A, MARCH 2004 METALLURGICAL AND MATERIALS TRANSACTIONS A

*JEOL is a trademark of Japan Electron Optics Ltd., Tokyo.

Fig. 1—Schematic diagram of the broaching process used to produce theU-notches. A similar process is used for the fir tree root fixings in gasturbine disc rims.

Fig. 2—U-notch specimen geometry.

in the fir tree root fixings of a gas turbine disc). Crack initia-tion and growth in the polished notch root (thus removing anysurface roughness effects) has also been compared for the twobatches of specimens.

II. MATERIAL AND SPECIMENCHARACTERIZATION

Two batches of single-edge U-notch specimens, labeledbatch A and batch B, were extracted from two IN718 turbinedisc forgings produced from the same cast of material. Eachforging was solution treated at 955 °C for 1 hour followedby an air cool, then aged at 720 °C for 8 hours, cooled at50 °C/h to 620 °C, aged at 650 °C for a further 8 hours, andfinished with an air cool. The composition of the disc mater-ial was B—0.003 wt pct, C—0.031 wt pct, N—0.006 wtpct, O—0.00006 wt pct, Mg-0.002 wt pct, Al—0.46 wt pct,Si—0.08 wt pct, P—0.01 wt pct, S—0.0004 wt pct, Ti—1.02 wt pct, Cr—18.0 wt pct, Mn—0.06 wt pct, Fe—18.6wt pct, Co—0.43 wt pct, and bal. Ni. The U-notch speci-

men geometry is shown in Figure 2. The dimensions ofthe U notch were chosen to give a stress concentrationfactor similar to a blade fir tree root fixing in a turbine discrim. Finite-element (FE) modeling of the specimen pre-dicted an elastic stress concentration factor of 2.08. Notcheswere machined in each batch using a broaching processsimilar to that used for machining fir tree root fixings inproduction discs. The broaching process produced longmachining marks parallel to the notch axis. A Talysurf*

*Talysurf is a trademark of Taylor Hobson, Ltd., Leicester, UK.

*Instron is a trademark of Instron Corporation, MA, USA.

instrument was used to obtain the following measures ofsurface roughness in the notch roots:

Ra � average deviation from the mean surface, andRv � maximum valley depth below the mean surface

detected in a measurement run.

Three measurement runs (t1 through t3) were made per-pendicular to the notch axis per specimen, spaced at 5-mmintervals and acquired using a data length of 1.04 mm and acut-off length of 0.08 mm—the longest data length possiblewithout risk of damage to the Talysurf stylus. Metallographicsections were taken to assess any microstructural changesthat may have occurred as a result of the broaching process.General microsections were also taken to compare themicrostructure in the two specimen batches. Metallographicsamples for optical microscopy and scanning electron micro-scope (SEM) examination were prepared using standard tech-niques. Specimens were etched using Nimonic etch (10 mLHNO3, 50 mL HCl, 40 mL H2O, and 2.5 g CuCl2). Opticalmicroscopy combined with image analysis was used todetermine grain size and primary carbide volume fraction. AJEOL* JSM 6400 SEM was used for fractography and further

metallography. The identity of phases was confirmed byenergy-dispersive X-ray spectrometry (EDS). Microhardnessmeasurements were performed on cross sections to estab-lish if surface layers exhibited signs of work hardening. Anindent small enough to be entirely within limited deforma-tion layers, but with a test mass large enough to give a truerepresentation of hardness, was required. In each case, 20near-surface measurements were performed using a test massof 200 g. Indents were located 20 �m below the notch surface.

III. FATIGUE TEST METHODS

Three-point bend fatigue tests were conducted in air at600 °C under load control, using a 1-1-1-1 trapezoidalwaveform (Figure 3). Specimens were heated inside a high-temperature chamber mounted on an Instron* 8501 servo-

hydraulic testing frame. The temperature was monitored andcontrolled via an R-type thermocouple attached to the frontface of each specimen. For all tests, the recorded temperaturewas within �2 °C of the target temperature. Tests were per-formed at a load ratio R � 0.1, and nominal maximum stresses�max from 750 to 1000 MPa were employed. Applied loads

03-186A-4.qxd 2/7/04 2:53 PM Page 772

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 35A, MARCH 2004—773

Fig. 3—Trapezoidal waveform (1-x-1–1) used in fatigue testing, where x �dwell in seconds at maximum load. For all tests, x � 1.

Table I. Mean Surface Roughness Data for As-BroachedNotch Surfaces*

Roughness PolishedMeasurements Batch A Batch B Specimen

Mean Ra 0.1607 0.4011 0.0053 (0.0004)(�m) (0.0202) (0.0803)

Mean Rv 0.7960 1.7574 0.0258 (0.0086)(�m) (0.1567) (0.4369)

Maximum Rv (�m) 2.1582 3.1157 0.0357

*Standard deviations are given in parentheses.

Fig. 4—Surface roughness measurements in U-notch roots. Values of Ra

for batch A and batch B.

were calculated to give the desired nominal value of �max in thenet specimen cross section (i.e., ignoring the stress concentratingeffect of the notch). Elastic-plastic FE modeling of the U-notchspecimen was performed to predict the strain range in the notchroot for the different nominal applied stress levels. In terms ofthe individual total strain components, the equivalent strainrange was calculated as follows:

[1]

Each strain component included elastic and plastic contri-butions. The strain amplitude is given by

[2]

The majority of tests were performed on specimens with notchesin the as-broached condition, to represent the surface finishpresent in a production turbine disc. These tests were run tofailure, then examined in an SEM to identify the crack initia-tion sites. A second set of tests was performed on specimenswith notch surfaces mechanically polished to a 1-�m diamondfinish. To minimize notch enlargement, specimens were securedin a specially designed jig and polished using soft 4-mm-diameter � 8-mm-long cylindrical dental felts mounted in apillar drill. Selected polished notch tests were interrupted inorder to take cellulose acetate replicas from the notch root.

IV. RESULTS

A. U-Notch Surface Roughness and MicrostructureComparison

Mean Ra and Rv data for batch A and batch B are given inTable I. Batch A data are the averages from 15 specimens.Batch B data are the averages from 22 specimens. The polishedspecimen data are from a batch B specimen that had receivedthe standard polishing procedure, where care was taken to uselow polishing forces to prevent smearing or residual stressgeneration, specimens were inspected under an optical micro-scope and in an SEM after polishing. The authors considerany differences in surface roughness between polished speci-mens from Batch A or B to be minimal. Maximum Rv valuesare also listed, to give an indication of the maximum broachingmark depth in each batch. The data show that the notches in

�amp ���eq

2

(��zz � ��xx)2 �

3

2 1(�gxy)

2 � (�gyz)2 � (�gxz)

2 2 D1/2

��eq �12

3 C(��xx � ��yy)

2 � (��yy � ��zz)2 �

batch B specimens were rougher than those in batch A. Thiscan be seen in the plot of Ra values for individual specimensdisplayed in Figure 4. Polishing of the U notches produceda large reduction in surface roughness.

Metallographic sections through the notches of as-broachedspecimens (Figure 5) revealed a difference in the subsurfacemicrostructure of the notch between batch A and batch B. Inbatch A, a subsurface layer was observed, which extended allthe way around the surface of the U notch and had the appear-ance of a zone of heavy localized deformation (Figure 5(a)).The thickness of the layer was between 10 and 16 �m in thenotch root and up to 40 �m on the notch sides. No deform-ation layer was observed on the other machined surfaces ofthe batch A specimens, only the broached notch surface. Inbatch B, a deformation layer was not observed in the broachednotch (Figure 5(b)) or on the other machined surfaces. Thedeformation layer was still present in batch A specimens afterU-notch polishing. The layer’s apparent depth in the notchroot was reduced to approximately 7 to 10 �m. Away fromthe notched region, both batches had a nonequiaxed grainstructure with larger elongated grains (diameter approximately20 to 70 �m) surrounded by extensive regions of small, moreequiaxed grains (diameter approximately 5 to 10 �m). Thelarger grains were elongated in the tangential and radial direc-tions, as would be expected for forged discs. Table II givesthe mean linear intercept grain sizes in the three disc forgingdirections, showing no significant difference in grain sizebetween batches A and B.

03-186A-4.qxd 2/7/04 2:53 PM Page 773

774—VOLUME 35A, MARCH 2004 METALLURGICAL AND MATERIALS TRANSACTIONS A

Table III. Vickers Microhardness Results

Batch A Batch B

Location Mean Vickers Microhardness Standard Deviation Mean Vickers Microhardness Standard Deviation

Near surface 454 23 371 15Bulk material 394 8 391 9

Table II. Mean Linear Intercept Grain Sizes, Disc Material

Batch A Batch B

Mean MeanLinear Standard Linear Standard

Intercept Deviation* Intercept Deviation*Direction (�m) (�m) (�m) (�m)

Tangential 10.8 1.4 9.6 0.6Radial 10.8 0.4 11.5 0.8Axial 6.3 0.6 6.7 0.8

*Note the standard deviation is between the different areas assessed.

results were 0.37 pct for batch A and 0.33 pct for batch B,with very similar size distributions exhibited for both batches.It has been shown[26] that the amount of carbides and nitridesis directly determined by the C and N content of the alloys,and that the measured volume fraction of carbides andnitrides is consistent with the C and N content of the presentIN718 alloy.

Microhardness indents on a batch A notch microsectionwere located in the deformation layer on the notch sides.These were compared with measurements made on bulkmaterial several millimeters below the surface, and withmeasurements made in similar locations on a batch B speci-men. The hardness data (Table III) revealed differences inhardness on the surface layers between the batches. It isnoted that if we assume a Gaussian distribution of hard-ness on the measured locations, the limit of 95 pct confi-dence in measured average is 2�D/ , with N the numberof hardness indentations made (N � 20). This means con-fidence limits are in the range of 4 to 10 and the differ-ences in average hardness are clearly significant. The meanmicrohardness in the deformed surface layer of the batchA specimen was approximately 15 pct higher than that forthe bulk material, indicating that the material had been workhardened by the broaching process. This may be an under-estimate, due to the effect on hardness readings of reducedconstraint near the specimen surface. No evidence of workhardening was found in the batch B specimen. The appa-rently softer batch B near-surface hardness relative to thebulk value was probably due to reduced constraint near thespecimen surface.

B. Fatigue Lives

Fatigue results for the as-broached conditions in eachbatch are displayed as stress-life and strain-life curves inFigures 6(a) and (b). In Figure 6(b), lifetimes are plottedagainst the strain amplitude predicted by Eqs. [1] and [2].The strain-life data for each batch have been fitted to theCoffin–Manson relationship:

[3]

where

�� � strain range,�f � fatigue strength coefficient ( � the uniaxial true

fracture stress),E � elastic modulus,Nf � number of cycles to failure,�f � fatigue ductility coefficient (material-dependent),

and1,2 � material-dependent constants.

The parameters of the fitted curves for each batch aregiven in Table IV. The fit is an empirical one since �f, �f,

��

2� asf ¿

Eb(2Nf)

l1 � � ¿f (2Nf)l2

1N

(a)

(b)

Fig. 5—Optical micrographs of sections through U-notch root. (a) BatchA specimen, showing presence of a deformation layer at the surface. (b)Batch B specimen, with no deformation layer in the notch root. The highersurface roughness of the batch B specimen can be seen.

Grain boundaries were decorated with particles of vari-ous morphologies, which EDS revealed to be the � (Ni3Nb)phase. Other phases observed using optical microscopy andan SEM were primary carbides and nitrides. The combinedarea fraction of primary carbides and nitrides was deter-mined by image analysis on polished microsections. The

03-186A-4.qxd 2/7/04 2:54 PM Page 774

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 35A, MARCH 2004—775

(a)

(b)

Fig. 6—Stress and strain-life curves for as-broached U-notch specimens(1-1-1-1 trapezoidal waveform, 600 °C, in air).

Table IV. Coffin–Manson Fit Parameters

FittingParameterValue Batch A Batch B Comments

E (MPa) 1.68 � 105 1.68 � 105 held constant�f (MPa) 580 658 allowed to vary in fit1 �0.01 �0.01 held constant�f 0.339702 0.445076 allowed to vary in fit2 �0.71371 �0.74398 allowed to vary in fit

Fig. 7—Strain-life data for all U-notch tests performed at 600 °C, using a1-1-1-1 trapezoidal waveform. The data is fitted with Coffin–Mansontype expressions for strain-life data.

and 2 are all allowed to vary, but the raw data and theresulting fits do indicate a difference in material fatiguestrength/ductility between batches A and B

It is apparent in Figure 6 that batch B exhibited longerfatigue lives than batch A when tested under the same testconditions. In Figure 7, the fatigue lifetimes of the pol-ished notch roots in both batches can be compared with theperformance of the as-broached notch roots. Uninterruptedas well as interrupted polished tests are compared. For unin-terrupted tests on batch A specimens, polished notches pro-duced longer lifetimes to fracture than as-broached notchesunder the same test conditions. The beneficial effect of notchpolishing was also observed for batch B specimens testedat �max � 885 MPa. However, a batch B polished notch

test performed at �max � 790 MPa had a shorter life than as-broached batch B specimens. Fractographic examination ofthis test showed that a crack had initiated at a subsurface(Ti,Nb)N particle.

C. Initiation Behavior

After fracture, the as-broached specimens were examinedin an SEM to characterize the crack initiation sites. Multisitecrack initiation occurred in the notch root. Several typesof crack initiation site in air were identified in the twobatches:

1. bulgelike features associated with the oxidation of sub-surface primary carbides (Figure 8(a)),

2. surface eruptions formed by the oxidation of primary car-bides (Figure 8(b)),

3 initiation at (Ti,Nb)N particles (Figure 8(c)),4. small bulgelike features not associated with the oxida-

tion of subsurface primary carbides (Figure 8(d)), and5. sites with no obvious surface eruption, bulge, or nitride

particle.

Additional information about the crack initiation processin air was obtained from the interrupted tests on polishedspecimens. Examination of surface replicas taken during theinterrupted tests showed that cracks began to initiate withinthe first 10 pct of the overall fatigue life. The evolution ofthe number of cracks present in the notch root as tests pro-gressed is shown in Figure 9 for �max � 750 and 790 MPain batch A. For both tests, the majority of cracks initiatedat bulgelike features, with the remainder starting at surfaceeruptions or at sites with no discernable eruption or bulge(termed intrinsic initiation). A common feature in both testswas an incubation period at the beginning of the test in whichno cracks were observed. After the initial incubation period,the number of cracks increased at a steady rate. The initia-tion rate then decreased. Eventually, major coalescence ofcracks led to rapid reduction in the number of individualcracks observed in the notch root.

03-186A-4.qxd 2/7/04 2:54 PM Page 775

776—VOLUME 35A, MARCH 2004 METALLURGICAL AND MATERIALS TRANSACTIONS A

(a) (b)

(c) (d)

Fig. 8—Examples of short crack initiation sites observed in the notch root of batch A and batch B specimens tested in air. (a) A crack initiated from abulged site, commonly seen in the notch root in batch A. (b) A surface eruption, with some deformation visible initiating a crack also in batch A. (c) Ini-tiation in batch B at a subsurface (Ti,Nb)N particle (rectangular shaped particle about 15 �m below surface). (d) Small bulgelike features, but not associ-ated with an oxidized (Nb,Ti)C particle (SEM, Secondary Electron Imaging (SEI)).

For as-broached specimens, it was observed that the totalnumber of initiation sites per specimen was much lowerfor batch B than batch A tested under the same conditions.It was also observed that unlike batch A, the most numerouscrack initiation sites were not oxidized primary carbides(eruptions or bulgelike sites). Instead, the most numeroussites were the sites with no apparent feature on them. Some-times, small bulgelike features were observed, but they werenot associated with subsurface (Nb,Ti)C particles. The EDSanalysis of these bulges showed there was some slight enrich-ment in Nb relative to the matrix, but not the very highenrichment in Nb indicating the presence of an (Nb,Ti)Cparticle. Examples of these bulges without associated particlesare shown in Figure 8(d). These small bulges were not evi-dent elsewhere on the notch surface, unlike the prominentbulges associated with (Nb,Ti)C carbide particles observedon batch A specimens. Where an (Nb,Ti)C particle waspresent at an initiation site in a batch B specimen, this waseasily confirmed by the high Nb signal detected using EDS.Occasionally, initiation at (Ti,Nb)N particles was observed,as shown in Figure 8(c). The (Ti,Nb)N particles in the mater-ial were known to be Ti rich, and this was used to confirmthe identity of the particles at initiation sites.

The evolution of the number of cracks present in the notchroot as tests progressed is shown in Figure 10 for two testscarried out at �max � 885 MPa in batch B. For the test rep-resented in Figure 10(a), the dominant crack initiated at asurface eruption and was first observed after 1000 cycles.This was the only crack to grow significantly. Most of theother cracks that initiated did so late in the fatigue life aheadof the main crack tip and in the residual ligaments. In Fig-ure 10(b), the dominant and only significant crack was firstobserved after 1000 cycles at a (Ti,Nb)N particle, and again,the majority of cracks initiated late in the fatigue life aheadof the main crack tips of the dominant crack. This was dif-ferent from the behavior seen in batch A at �max � 750 and790 MPa, where cracks initiated and grew from several sitesand later coalesced to form the final dominant crack.

The difference in initiation characteristics between thebatch A and B tests can be seen in Figures 11(a) and (b),where the total number of cracks observed is plotted againstthe number of cycles (a) or percentage of life (b). In all fourtests, crack initiation rates were low at first, up to 5000cycles. The batch A tests at �max � 750 and 790 MPa thenexhibited crack initiation at a fairly constant rate beforereaching a plateau, with a decrease in crack numbers at the

03-186A-4.qxd 2/7/04 2:54 PM Page 776

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 35A, MARCH 2004—777

(a)

(b)

Fig. 9—Quantification of number of cracks observed on replicas from pol-ished batch A U-notch specimens (trapezoidal waveform 1-1-1-1, 600 °C,in air): (a) �max � 750 MPa, and (b) �max � 790 MPa.

(a)

(b)

Fig. 10—Quantification of number of cracks observed on replicas frompolished batch B U-notch specimens (trapezoidal waveform 1-1-1-1, 600 °C,in air) �max � 885 MPa: (a) test 1 and (b) test 2.

end of the life due to coalescence. However, the two batchB �max � 885 MPa tests showed very rapid crack initiationwith no plateau and no significant crack coalescence, fol-lowed shortly by specimen fracture. Comparing crack initi-ation behavior plotted against percentage of fatigue life,the difference in initiation behavior between the lower stresslevels (�max � 750 and 790 MPa) and �max � 885 MPa isalso apparent. The lives of the tests at 750 and 790 MPatherefore appeared to be initiation dominated, whereas thetests at 885 MPa were propagation dominated.

D. Crack Propagation Paths

Striking differences were observed in the propagation char-acteristics of cracks between the two batches. In batch A,extremely straight, apparently transgranular crack growth wasobserved for all cracks, except where the tips underwent localdeflection during crack-crack interactions. An example of thisremarkably straight crack growth can be seen in Figure 12,where an acetate replica of a crack initiating from a bulge canbe seen. The fatigue fracture surface produced by such crackgrowth can be seen in Figure 13, showing a transition from

very flat transgranular fracture to mixed transgranular/inter-granular fracture as the crack propagates into the depth of thespecimen; again, a subsurface bulge associated with anenlarged particle ((Nb,Ti)C) can be seen at the initiation site.In the batch B specimens, more crack deflections wereobserved (most easily seen in the polished specimens). Fig-ure 13 shows an example of such a crack, which in this casehas initiated from a TiN particle; the crack path exhibits manymore deflections with apparent mixed transgranular/inter-granular propagation, and it appears to become more inter-granular as the crack becomes longer. There is clear evidenceof slip band formation in adjacent grains, which was notobserved in the polished tests carried out in batch A.

Fatigue life results showed that polished specimens usu-ally had longer lifetimes than as-broached specimens testedunder the same conditions. Since the broaching marks wereparallel to the notch root and hence the direction of surfacecrack propagation, it was possible that the broaching markswere accelerating crack initiation or crack propagation,resulting in shorter fatigue lives than polished specimens. Inthe batch A tests, the cracks observed were straight and par-allel to the notch axis on both as-broached and polished speci-mens. Hence, it was not possible to determine conclusively

03-186A-4.qxd 2/7/04 2:54 PM Page 777

778—VOLUME 35A, MARCH 2004 METALLURGICAL AND MATERIALS TRANSACTIONS A

(a)

(b)

Fig. 11—Comparison of crack initiation behavior for tests on polishedU-notch specimens at different stress levels (a) in terms of cycles and (b)in terms of percentage of life to fracture. The higher stress level data (885MPa) is from batch B samples, and the two lower stress levels are datafrom batch A samples.

Fig. 12—Crack A from test A600-790-1-PR-01 after 15,000 cycles (�max �790 MPa, batch A), showing asymmetry about the initiation point (opticalmicroscopy from acetate replica).

Fig. 13—SEM image of fracture surface showing transition from very flat,transgranular fracture to mixed transgranular-intergranular fracture in batchA specimen. Note the bulge above the (Nb,Ti)C particle at the crack ini-tiation site.

if the broaching marks affected crack propagation. However,for the batch B tests, there was some evidence that the broach-ing marks were influencing crack propagation. The SEMfractography was performed on uninterrupted run to frac-ture tests performed under the same conditions, comparingan as-broached and a polished notch. Comparisons of thesurface crack path in the notch root near the main crack ini-tiation site are shown in Figure 15. In the as-broached notch,the surface crack path was essentially straight and parallel tothe broaching marks, whereas in the polished notch, the crackpath deviated along slip bands or grain boundaries.

V. DISCUSSION

Both discs were produced consecutively from the samecast of material using the same forging and heat treatmentprocess, so one would expect a similar microstructure to result.This is borne out by the results of metallography on the twobatches of specimens, showing them to have similar grainsizes, similar area fractions of primary carbides and nitrides,and similar size distributions of primary carbides and nitrides.

Fig. 14—Crack Y from test B600-885-1-PR-02 after 4000 cycles (�max �885 MPa, batch B), showing crack deflections. Crack propagation appearedto be mixed transgranular-intergranular (optical microscopy from acetatereplica).

03-186A-4.qxd 2/7/04 2:54 PM Page 778

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 35A, MARCH 2004—779

(a)

(b)

Fig. 15—Comparison of surface crack fracture paths in notch root of batchB specimens at �max � 885 MPa. Batch B specimens (a) as-broached and(b) polished (SEM, SEI).

However, two differences between the as-received batches ofU-notch specimens were surface roughness and the presenceof what appeared to be a work-hardened deformation layerin the notch surface of batch A specimens. Both these obser-vations point to differences in the broaching process used tomachine the U-notches. It is believed that the deformationlayer in batch A specimens arose from use of a blunt tool,causing local deformation of the material rather than cuttingit cleanly. These two differences, and especially the work-hardened layer, are thought to be the cause of the differentcrack initiation and crack growth behavior of the two batches,as will be discussed subsequently.

Previous work by the authors on batch A specimens[25]

demonstrated that primary carbides at or close to the surfaceof IN718 oxidized at elevated temperatures, and there wasa substantial volume expansion associated with oxidation.

This was observed in unstressed specimens and U-notchfatigue specimens exposed at 550 °C and 600 °C in air. Thebatch A U-notch tests in particular exhibited a link betweenprimary carbide oxidation, localized matrix deformation, andfatigue crack initiation. Intense planar slip in batch B speci-mens (the specimens without a work-hardened layer) wasevident from the slip bands visible on polished U-notch sur-faces at �max � 790 and 885 MPa. Compared to batch Aspecimens, a greater proportion of fatigue cracks initiatedintrinsically, rather than at oxidized primary carbides. Theseobservations are thought to be linked; if intense slip bandsdevelop in a material, then these can lead to crack initiationin a variety of ways, including slip-step formation producinglocal notches on the fatigue surface or at stress concentra-tions arising from slip band interaction with grain bound-aries or carbides/nitrides. Formation of intense slip bandsand crack initiation along them is promoted by cyclic soft-ening, with dislocations preferentially gliding along the softerband rather than in the harder surrounding regions.[27,28]

IN718 is sensitive to the environment at elevated tempera-tures and grain-boundary embrittlement by oxidation canoccur.[7,29,8] Impingement of slip bands on grain boundariesproduces local stress concentrations at the boundary. Theslip bands can also act as fast diffusion paths for diffusionof oxygen, accelerating grain-boundary oxidation. Hence,stress concentrations at a grain boundary combined withgrain-boundary embrittlement may initiate cracks. In thecase of the U-notch specimens, crack initiation appeared tobe predominantly transgranular, so grain-boundary embrit-tlement was not a major mechanism of crack initiation inthis case. However, some intergranular crack propagationwas observed on fracture surfaces, so the environment wasaffecting crack propagation paths.

A few bulges were observed at batch B initiation sites,which were not associated with oxidized primary carbides.These features may have been an artefact of the broachingprocess, or small protrusions of material—possible evidenceof intense slip band formation—and were of a similar scaleto slip band extrusions observed elsewhere.[28] Note that inas-broached specimens, the roughness of the U-notch maskedthe appearance of slip bands on the notch surface. Protrusionswhere slip bands intersect a free surface may act as crackinitiators due to the local stress concentrating effect of thestep formed. Some intrinsic initiation was observed in batchA specimens as well as in batch B specimens, though intenseslip bands were not seen in the polished notch roots ofbatch A specimens. However, there may still have been crackinitiation in the batch A notch roots along favorably orientedslip lines, and machining marks on as-broached specimenscould also have acted as crack initiators.

The bulges above oxidized carbides seen at crack initiationsites in the batch A U-notch fatigue tests are evidence forextensive plastic deformation in the matrix around oxidiz-ing subsurface primary carbides, as predicted by the misfitcalculations in previous work.[25] For a mechanism of fatiguecrack initiation in batch A U-notch specimens, it is proposedthat the misfit strains due to carbide oxidation were super-imposed on the elastic-plastic stress-strain field due to exter-nal loading. This created local strains high enough to causefracture of the matrix in the vicinity of an oxidizing particle,hence initiating a fatigue crack. The plastic deformation inthe notch root and the formation of slip bands may have

03-186A-4.qxd 2/7/04 2:54 PM Page 779

780—VOLUME 35A, MARCH 2004 METALLURGICAL AND MATERIALS TRANSACTIONS A

Fig. 16—Modified batch A strain-life curve for as-broached U-notch spec-imens, waveform 1-1-1-1, 600 °C, air.

promoted stress-assisted oxygen diffusion, causing morerapid oxidation of the subsurface primary carbides thanthat observed for unstressed thermal exposure specimens.Once a crack initiated, the crack itself provided anotherfast oxygen diffusion route to a subsurface particle. The sur-face cracks observed experimentally at oxidized subsurfacecarbides were very sharp, probably resulting in a high stressintensity at the crack tip, which drove further propagation.

Matrix deformation was seen around some surface erup-tions formed by oxidized (Nb,Ti)C carbide particles. It wasoriginally thought that initiation at these surface oxide erup-tions might be due to the micronotch effect of a cavitybeneath the erupted particle. However, the observation ofmatrix deformation associated with some surface eruptionsin thermal exposure specimens and on the U-notch fatiguespecimens suggests that it was the additional matrix straincaused by expansion mismatch that assisted crack initiation.Matrix deformation, combined with plastic strain accumu-lation from the fatigue loading, could have caused initiationfrom those surface eruptions, which produced sufficientmisfit strains in the matrix. Initiation from surface eruptionswas observed in batch A and batch B specimens, thoughmany surface eruptions in both batches of material did notinitiate cracks, so that only those eruptions that producedhigh enough misfit strains acted as crack initiators.

The high incidence of bulges at crack initiation sites inbatch A specimens suggests that expansion of oxidizing sub-surface primary carbides was a significant mechanism offatigue crack initiation in the notch root. Certainly, for batchA specimens, it was the oxidized sub-surface carbides, ratherthan oxide eruptions in general, that were the more numeroustype of crack initiation site in U-notch specimens tested inair. This was not the case for batch B specimens, in whichinitiation at oxidized subsurface carbides was infrequent. Thedifference in crack initiation behavior between the two batchesof specimens when tested under the same conditions mustarise from the different roughness of the broached U-notchsurfaces or the local microstructure in the notch root.

Surface roughness will affect crack initiation behavior.Generally, rougher surfaces result in earlier crack initiationand shorter fatigue lives, due to the local stress concentrationeffect of surface features such as scratches and machiningmarks. In a study by Andrews et al.[30] on two microstruc-tural variants of IN718, it was found that machining marks,rather than microstructure, were controlling crack initiationlife in double edge notch specimens at 650 °C. In this study,if surface roughness was the critical factor, batch B wouldbe expected to exhibit worse fatigue performance than batchA. However, the as-broached batch B specimens had longerlives, as shown in Figure 6, and generally fewer initiationsites than batch A specimens tested under the same condi-tions. For polished batch B specimens, the total number ofinitiation sites was also lower than in batch A. Surface rough-ness alone does explain the difference in initiation behaviorbetween batch A and batch B.

The other difference between the notches in batch A andbatch B was the presence of a deformation layer in batch Asamples. This was up to 40-�m thick on the notch sides andapproximately 10- to 16-�m thick in the notch root. Thelayer was not observed in batch B specimens. Vickers micro-hardness measurements indicated that the surface deforma-tion layer was approximately 15 pct harder than the bulk

material in batch A specimens. Cahoon et al.[31] related yieldstress to hardness using the formula

[4]

where

�y � yield stress in MPa,Hv � Vickers hardness,Q � a constant, andn � work-hardening exponent.

From Eq. [4], it follows that providing n is constant, fora 15 pct increase in hardness, a 15 pct increase in yield stressis predicted. This higher localized yield stress may have influ-enced crack initiation behavior. Considering a purely con-tinuum approach, it is possible to estimate how the higherlocal yield stress affected the local strain amplitude, and hencethe strain-life characteristics of batch A specimens. Ideally,this should have been done by modifying the FE model totake into account a higher local yield stress at the notch sur-face, but this was not possible using the existing mesh or theuser-defined constitutive model. Instead, a simplified approachwas adopted, using the existing FE model outputs of elasticand plastic strain components at the notch root center. A 15pct increase in yield stress will increase the elastic strainbefore yielding occurs by the same factor, so the elastic com-ponents of strain in the notch root at �max were increased bya factor of 15 pct to reflect this. It was not possible to estab-lish the effect of an increased yield stress on plastic straincomponents, so these were assumed to be unchanged. Thisassumption was considered reasonable, since most of thereversible strain in the notch root was due to the elastic con-tribution. The total strain amplitudes �amp for batch A testswere then recalculated using Eqs. [1] and [2]. The resultingadjusted strain-life curve for batch A is compared with theoriginal batch B results in Figure 16, and the appropriateCoffin–Manson fit data are compared in Table V. With totalstrain amplitudes adjusted to account for pre-existing coldwork hardening in the notch root, the batch A results nowlie close to those obtained for batch B specimens. Hence,an increased local yield stress and total strain amplitude in

sy � 9.81aHn

3b(Q)(n)

03-186A-4.qxd 2/7/04 2:54 PM Page 780

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 35A, MARCH 2004—781

Table V. Coffin–Manson Fit Parameters

Fitting Parameter Values Batch A Batch B Batch A (Hardened) Comments

E (MPa) 1.68 � 105 1.68 � 105 1.68 � 105 held constant�f (MPa) 580 658 580 allowed to vary in fit1 �0.01 �0.01 �0.01 held constant�f 0.339702 0.445076 0.010305 allowed to vary in fit2 �0.71371 �0.74398 �0.2767 allowed to vary in fit

the notch root could explain the difference in the strain-lifebehavior between batch A and batch B specimens, in the pre-sent continuum approach. It should be noted that in this analy-sis we are considering the local area of material in the notchroot, and the empirical Coffin–Manson fits to the data areconsistent with the broaching in batch A having a detrimentaleffect on the fatigue strength/ductility of the material in thenotch root. Fatigue life appears to be dominated by initia-tion/short growth; therefore, the condition of the material inthe localized area of the notch root is clearly important.

In considering a more microstructural explanation of therole of the surface layer, it is noted that the higher yieldstress of the work-hardened layer will have led to greaterstresses around misfitting particles. For example, with a15 pct increase in �y, the maximum value of the predictedhoop stress calculated from the model presented in Reference25 increases from 327 to 376 MPa. For a carbide particleoxidizing in the deformation layer of batch A, these highermisfit stresses combined with the external loading may haveenhanced crack initiation and early crack growth. This isconsistent with the observation of many cracks initiating atoxidized carbides in batch A, but only a few in batch B.An additional effect on crack initiation and propagation inbatch A may have resulted from reduced ductility in thework-hardened layer. In a few cases in batch B, crack initia-tion at primary nitride particles was observed. This wasprobably due to slip band impingement on the particles,producing stress concentrations at the particle-matrix inter-face, which resulted in crack initiation either by crackingof the particle or decohesion of the particle-matrix interface.This mechanism of crack initiation has been observed inother superalloys containing primary carbide and nitride par-ticles, such as INCOLOY 908.*[32] An interesting experiment

A specimens tested at the lower stress levels �max � 750 MPaor �max � 790 MPa, and the high stress level batch B testsat �max � 885 MPa. For the batch A tests, there was an initialincubation period at the start in which no cracks wereobserved. The initially slow rate of crack nucleation in bothtests is indicative of some form of incubation period nec-essary for crack initiation. Some of the first cracks observedinitiated at bulges due to subsurface primary carbide oxi-dation, so the incubation period for initiation is linked to thetime required for oxidation of the particles. The time at thetest temperature of 600 °C was between 2 and 6 hours—afactor of 10 less than the time required for subsurfaceoxidation in unstressed thermal exposure samples.[25] Thisimplies a strong effect of fatigue cycling on the kinetics ofoxidation, probably caused by enhanced oxygen transportalong slip bands. The first particles to oxidize and initiatecracks may have been those favorably situated for fast oxygendiffusion to the particles, either being close to the surfaceor on a grain boundary. As already discussed, the deformationlayer present in the notch root of batch A specimens mayhave promoted faster diffusion.

Following the initially slow rate of crack initiation, therewas a transition to behavior in which the number of cracksappearing increased at a steady rate, which was similar inboth tests. The factors controlling this rate are thought to bethe time required for oxidation of carbides in different partsof the specimen and the time required for accumulation ofsufficient cyclic damage in the material to initiate furthercracks. The number of primary carbides that are potentialcrack initiation sites will depend upon the number densityof primary carbides in the material, their distribution rela-tive to the notch surface, and their distribution relative topreferential oxygen diffusion pathways such as grain bound-aries. The local stress-strain environment around potentialcrack initiation sites will vary, depending on the positionof the site in the notch root and shielding by cracks that havealready initiated. The number of cracks initiated eventuallyreached a plateau, which can be attributed to a combinationof exhaustion of potential crack initiation sites, reductionin the local strain at initiation sites due to shielding by neigh-boring cracks, and crack coalescence.

In batch B tests, the development of crack initiation overtime was significantly different from the batch A tests per-formed at lower stress levels. Figure 11 shows that fewcracks initiated until the latter stages of the fatigue life. Asdescribed earlier, the fatigue failure was controlled by ini-tiation and growth of a single crack, with most of the othercracks initiating in the stress-strain field ahead of the tip ofthe growing crack or in the residual ligaments. One possiblecause of the difference between batch A and batch B testswas the absence of a work-hardened surface layer in batchB. This may have reduced the propensity for crack initiation

*INCOLOY is a trademark of INCO Alloys International, Huntington,WV.

to sort out the role of subsurface oxidation vs surface condi-tion as the primary initiation mechanism would be to imparta compressive stress on the immediate surface layer by shotpeening and then perform equivalent fatigue tests.

The quantitative data for the number of crack initiationsites show that there were more initiation sites in batch Aspecimens than in batch B specimens tested under the sameconditions. As already discussed, this may be linked to thedeformation layer present in batch A specimens. Generalobservations on interrupted U-notch tests indicated that cracksfirst initiated within the first 25 pct of the overall fatiguelife. More detailed crack initiation data were obtained fromthe interrupted replica tests on polished U-notch specimens,as shown in Figures 11(a) and (b). Comparing these plotsof number of cracks vs number of cycles, it can be seen thatthere were differences in initiation kinetics between batch

03-186A-4.qxd 2/7/04 2:54 PM Page 781

782—VOLUME 35A, MARCH 2004 METALLURGICAL AND MATERIALS TRANSACTIONS A

at oxidized carbides compared to batch A specimens. Anotherpossible reason for reduced crack initiation was the muchhigher stress level used for batch B. This led to a shorterfatigue life, allowing less time for subsurface carbide oxida-tion and hence limited initiation by an oxidation mismatchstrain mechanism. The formation of persistent slip bandswas enhanced by the greater strain range in the notch rootat �max � 885 MPa, and the absence of a surface deformationlayer may also have promoted persistent slip band formation.This permitted rapid propagation of the dominant surfacecrack along slip bands or grain boundaries, reducing thepotential for crack initiation elsewhere due to shielding bythe main crack, except in the high stress regions ahead ofthe crack tips in the residual ligaments.

One feature that may have affected crack propagation wasthe presence of broaching marks in the notch roots. Themarks were straight and parallel to the notch axis and there-fore favorably oriented to influence crack propagation. Inbatch A specimens, cracks in the notch root were straightand parallel to the notch axis in both as-broached and pol-ished specimens, so a conclusive link between broachingmarks and crack propagation path was not apparent. In batchB specimens, a clear difference between microscopic crackpaths was apparent when as-broached and polished speci-mens tested under the same conditions were compared. Foran as-broached notch, the surface crack path was essentiallystraight and parallel to the broaching marks, whereas in apolished notch, the crack path deviated along slip bands orgrain boundaries, as shown in Figure 15. This differencemay help to explain why polished specimens generally hadlonger fatigue lives than as-broached ones. In as-broachedspecimens, the broaching marks could have a micronotcheffect, providing a favorable pathway for surface crackgrowth parallel to the notch root, so that short crack growthrates were faster in as-broached specimens than in polishedspecimens. In the polished U-notch of batch B specimens,intense slip bands developed. Cracks were observed to prop-agate along slip bands or along grain boundaries. Some inter-granular propagation in IN718 is expected at 600 °C in air,because of oxidation embrittlement of grain boundaries.[5,8,30]

In batch B U-notches, it is proposed that surface crack prop-agation of the crack along slip bands and grain boundariesproduced a longer, more tortuous crack path, resulting inpartial shielding of the crack from the applied stresses andhence a lower crack propagation rate. In batch A specimens,removal of the broaching marks may have reduced the localcrack driving force by eliminating the stress concentratingeffect of the broaching marks, even though the actual crackpath was unaffected. Removal of some of the surface work-hardened layer by polishing could also have acted to reducesurface crack propagation rates in batch A.

Another difference between the two batches is the absenceof slip bands on polished notch surfaces of the batch thathad the deformed surface layer (batch A). Ghonem andco-workers[5,33] found that prestraining IN718 at room tem-perature produced more homogeneous slip during subsequentfatigue testing at 650 °C in air. Their observation is relevantto the case of the U-notch roots because it demonstrates thatprior deformation at room temperature can influence sub-sequent high-temperature slip behavior. If the deformationlayer in batch A did result in more homogeneous slip at600 °C, as suggested by the work of Ghonem and co-

workers, this would explain the absence of slip bands onpolished notch surfaces. We think that the reduced propensityof slip band formation is caused by the forest of dislocationsintroduced by room-temperature deformation (by prestrainingintroduced by broaching with a blunt tool) interacting withmoving dislocations, thus hampering deformation by longrange slip over a limited set of slip planes, i.e., by dispersingslip. In batch A U-notch tests, surface cracks were straightand transgranular, which is also consistent with the theoryof Ghonem and co-workers that homogeneous slip reducesintergranular cracking. It was observed that below the notchsurface, there was a transition to mixed transgranular-intergranular cracking of the material. The depth at whichthis occurred was 30 to 50 �m in batch A specimens, greaterthan the visible depth of the surface deformation layer, butperhaps at the boundary of the region of material in whichhomogeneous slip was enhanced. However, reduced fatiguelives in batch A compared to batch B are indicative of adetrimental effect of prestraining. Substantial work hardeningwill also have reduced the ductility of the material withinthe deformation layer, possibly reducing its resistance tocrack initiation and propagation.

VI. CONCLUSIONS

The fatigue behavior of IN718 U-notch specimens wasfound to be different in two batches of specimens that hadsimilar microstructures in terms of grain size, primary carbidecontent, and the presence of grain-boundary � phase. How-ever, batch A specimens were found to have a work-hardeneddeformation layer in the broached notch, and a lower notchsurface roughness than batch B. Batch B specimens exhibitedsubstantially longer fatigue lifetimes than batch A under thesame experimental conditions. It has been shown that anincrease in yield stress in the batch A deformation layercould account for the difference in strain-life behaviorbetween the two batches. Fatigue lifetimes were affected bynotch surface roughness. Polishing the notch had a beneficialeffect on fatigue life in both batches.

For batch A specimens, it was the oxidized subsurfacecarbides, rather than oxide eruptions in general, that werethe more numerous type of crack initiation site in U-notchspecimens tested in air. For batch B specimens, initiationat oxidized subsurface carbides was infrequent and fewerinitiation sites were observed under the same conditions,despite the rougher notch surface. The deformation layer inbatch A affected crack initiation and propagation behavior,possibly by influencing the degree of slip homogeneity. Slipbands were not observed on the surface of polished notchesin batch A, but were in batch B, suggesting more hetero-geneous slip in batch B.

Crack initiation kinetics varied with applied stress. At thelower stress levels of �max � 750 or 790 MPa in batch A,there was an initial incubation period at the start in whichno cracks were observed. Following the initially slow rate ofcrack initiation, there was a transition to behavior in whichthe number of cracks appearing increased at a steady rate.The factors controlling this rate are thought to be the timerequired for oxidation of carbides in different parts of thespecimen and the time required for accumulation of suffi-cient cyclic damage in the material to initiate further cracks.

03-186A-4.qxd 2/7/04 2:54 PM Page 782

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 35A, MARCH 2004—783

The number of cracks initiated eventually reached a plateau,attributed to a combination of exhaustion of potential crackinitiation sites, reduction in the local strain at initiation sitesdue to shielding by neighboring cracks, and crack coales-cence. At �max � 885 MPa in batch B specimens, fatigue fail-ure was controlled by initiation and growth of a single crack,with most of the other cracks initiating in the stress-strainfield ahead of the tip of the growing crack or in the residualligaments. One possible cause of the difference between batchA and batch B tests was the much higher stress level usedfor batch B. This led to a shorter fatigue life, allowing lesstime for initiation by primary carbide oxidation.

In polished notch roots, the surface cracks in batch A wereall remarkably straight, with no deflections except thoseoccurring during crack coalescence. This may be related tothe deformation layer observed in the batch A notch roots.Slip bands were observed on the batch B polished notch sur-face, but not in the equivalent tests performed in batch A.The absence of slip bands in batch A may have been dueto extensive work hardening in the surface deformation layer,which prevented easy dislocation glide and the formation ofslip bands.

ACKNOWLEDGMENTS

This work was funded by the United Kingdom Engineeringand Physical Sciences Research Council and ALSTOMPower. The authors thank Dr. R. Jakeman, Dr. S.J. Moss,and M. Hughes, ALSTOM Power, for valuable technicaldiscussions and provision of the IN718 material.

REFERENCES1. ASM Handbook, 10th ed., vol. 16, Machining, ASM INTERNA-

TIONAL, Metals Park, OH, 1995, pp. 194-211.2. S.P. Lynch, T.C. Radtke, B.J. Wicks, and R.T. Byrnes: Fatigue Fract.

Eng. Mater. Struct., 1994, vol. 17, pp. 313-25.3. A. Andrieu, R. Molins, H. Ghonem, and A. Pineau: Mater. Sci. Eng.,

1992, vol. A154, pp 21-28.4. E. Andrieu, G. Hochstetter, R. Molins, and A. Pineau: Proc. 3rd Int.

Symp. on Superalloys 718, 625, 706 and Various Derivatives, TMS,Warrendale, PA, 1994, pp. 619-30.

5. H. Ghonem and D. Zheng: Metall. Trans. A, 1992, vol. 23A, pp. 3067-72.6. P. Shahinian and K. Sadananda: Trans. ASME J. Eng. Mater. Tech-

nol., 1979, vol. 101, pp. 224-30.

7. H. Ghonem, T. Nicholas, and A. Pineau: Fatigue Fract. Eng. Mater.Struct., 1993, vol. 16, pp. 565-76.

8. H. Ghonem, T. Nicholas, and A. Pineau: Fatigue Fract. Eng. Mater.Struct., 1993, vol. 16, pp. 577-90.

9. J.P. Pedron and A. Pineau: Mater. Sci. Eng., 1982, vol. 56,pp. 143-56.

10. E. Andrieu, R. Cozar, and A. Pineau: Proc. Superalloy 718—Metal-lurgy & Applications, TMS, Warrendale, PA, 1989, pp. 241-47.

11. L.A. James: Eng. Fract. Mech., 1986, vol. 25, pp. 305-14.12. M.R. Bache, W.J. Evans, and M.C. Hardy: Int. J. Fatigue, 1999, vol. 21,

Suppl. S69-S77.13. D. Fournier and A. Pineau: Metall. Trans. A, 1977, vol. 8A,

pp. 1095-1105.14. J.C. Healy, L. Grabowski, and C.J. Beevers: Int. J. Fatigue, 1991,

vol. 13, pp. 133-38.15. T. Denda, P.L. Bretz, and J.K. Tien: Metall. Trans. A, 1992, vol. 23A,

pp. 519-26.16. G.R. Leverant and M. Gell: Trans. TMS-AIME, 1969, vol. 245,

p. 1167.17. J. Reuchet and L. Remy: Mater. Sci. Eng., 1983, vol. 58, pp. 19-32.18. J. Reuchet and L. Remy: Mater. Sci. Eng., 1983, vol. 58, pp. 33-42.19. M. Reger and L. Remy: Mater. Sci. Eng., 1988, vol. A101, pp. 47-54.20. M. Reger and L. Remy: Mater. Sci. Eng., 1988, vol. A101, pp. 55-63.21. G. Sjoberg, N.-G. Ingesten, and R.G. Carlson: Proc. 2nd Int. Symp.

on Superalloy 718, 625 & Various Derivatives, TMS, Warrendale, PA,1991, p. 603.

22. P.A.S. Reed, F. Hachette, D. Thakar, T. Connolley, and M.J. Starink:Proc. 8th Int. Conf. on Mechanical Behaviour of Materials (ICM8),F. Ellyin and J.W. Provan, eds., Fleming Printing Ltd., Victoria, BC,Canada, 1999, vol. 1, pp. 418-23.

23. T. Connolley, P.A.S. Reed, and M.J. Starink: Proc. 5th Charles ParsonsTurbine Conf., Institute of Materials, London, 2000, pp. 982-98.

24. T. Connolley, M.J. Starink, and P.A.S. Reed: Proc. 9th Int. Symp. onSuperalloys, T.M. Pollock, R.D. Kissinger, R.R. Bowman, K.A. Green,M. McLean, S. Olson, and J.J. Scirra, eds., TMS, Warrendale, PA,2000, pp. 435-44.

25. T. Connolley, P.A.S. Reed, and M.J. Starink: Mater. Sci. Eng., 2003,vol. A340, pp. 130-45.

26. T. Connolley: Ph.D. Thesis, University of Southampton, Southampton,2001.

27. H.F. Merrick: Metall. Trans., 1974, vol. 5, pp. 891-97.28. J.B. Clark and A.J. McEvily: Acta Metall., 1964, vol. 12, pp. 1359-72.29. S. Floreen and R.H. Kane: Fat. Eng. Mater. Struct., 1980, vol. 2,

pp. 401-12.30. R.G. Andrews, A.K. Koul, and P. Au: Proc. 2nd Int. Symp. on Super-

alloy 718, 625 & Various Derivatives, E.A. Loria ed., TMS, Warrendale,PA, 1991, pp. 943-54.

31. J.R. Cahoon, W.H. Broughton, and A.R. Kutzak: Metall. Trans., 1971,vol. 2, pp. 1979-83.

32. Z. Mei, C.R. Krenn, and J.W. Morris: Metall. Mater. Trans. A, 1995,vol. 26A, pp. 2063-73.

33. D. Zheng, A. Rosenberger, and H. Ghonem: Mater. Sci. Eng., 1993,vol. A161, pp. 13-21.

03-186A-4.qxd 2/7/04 2:54 PM Page 783