DEFORMATION AND FRACTURE OF P/M

5combination of increased strain-rate sensitivity and decreasedsusceptibility to hydrogen embrittlement at high strain rates.The decrease in tensile ductility with increasing porosity is bothpronounced and technologically important.There appear to be two mechanismswhich contribute to the loss of ductility:(1)a decrease in uniform elon-gation due to the combination of pore growth (a "bulk" porosity effect) and a6porosity-induced decrease in work hardening rate [see Fig. 1Iand (2)porosity-triggered shear instabilities which occur at large strains and are-caused -by planes of high local pore content ("imperfections"). 7 In the latter26case, the analysis of Saje, Pan, and Needlemanmay be used to predict the17total fracture strain of the material; Fig. 2 shows that again good agreementis obtained between theory and experiment.- The Effect of Void/Pore Distributions on Ductile Fracture as Modeled by ArraysofHoles (with Ellen Dubensky)jThe above study shows the pre-existing porosity or strain-induced voidscan introduce planes of weakness into the materials.These "imperfections"can subsequently trigger a shear instability which creates microvoid sheetsand final failure.*Obviously, the distribution of pores/voids (or otherprocessing "defects" such as non-uniform microstructures) is an important anoften the controlling parameter in determing the resistance to tensile fracture.The purpose of this study is to model experimentally the influence ofvoid/pore distributions on ductile fracture.The only previous researchconcerning fracture in a material with a random distribution of voids is atheoretical analysis developed by Melander 27who used computer simulation toapply his theory to two arrays of voids; no experimental verification wasattemted.Thus the present research is also the first aimed at developing a4:A'

LLICE'q*"0.8-0.I:z2.Nii 0.4Eq.20.2002024i6810Percent Initial PorosityFig. 1. The influence of porosity fo on the relative uniformstrain, which is defined as the strain Eu at the onsetof diffuse necking (maximum load) for a porous materialdivided by that of a fully den'se compact. Eq. 2 isCu nl-fo)/l-fo-afo) where n is the strain hardeningexponent and a is a constant determined by the rate ofstrain-induced pore growth 6 .-P, ,.-- *.

1.251.000Ti-6AI-4VThery.2-o* 1.1 f 2-P- P30.75EF*00.5000y0.25.AA000246810% porosityFig. 2.The dependence of strain to fracture -f on porositycontent. The experimental data is compared to theprediction of the continuum imperfection analysis 2 6based on imperfections resulting from planes of highpore content whose area fraction of porosity fp*isrelated to that in the bulk f by either f I.1 fp2/3or fp* - 1.3 fp2 /3 ; see ref. .d- o- .

6sound experimental basis for the theory of ductile fracture as influenced by*void distribution.4In the present study, void distributions are modeled in two dimensions asarrays of holes whose positions are predicted by a random-number generator inPan appropriate computer program; for an example, see Fig. 3.Initially, theexperiments are being conducted on arrays of equi-sized holes in which the (a)*area fraction of holes F, (b) diameter of the holes D, and (c) minimum spacingof the holes Sm is controlled.The study is based on two materials (1100 Aland 7075 Al) of differing work hardening rates which are tested under condi*tions of plane stress vs. plane strain:sheet (plane stress deformation),(a) 1100-0 Al in the form of 1 mmi(b) 7075-T6 Al also as 1 mm sheet and (c)7075-T6 Al plate (-6 mm thick) in which deformation between the holes ispredominantly plane strain.The test matrix shown in Table I indicates thevalues of F, D, and S used in this study.mThree different random arrays aregenerated and tested for each combination of D, Lm, and-specimensH*A total of 72thus are being tested in uniaxial tension; one specimen in eachcondition is gridded for local strain determinations and all specimens arephotographed during testing.Table 1.Experimental Values of Hole Diameter D, Area Fractionof Holes F, and Minimum Inter-hole Spacing S ExaminedmD1.2 tmm442.0 mmSmmm.40.5F(-)am%42.02.5%5.0%() 2.51A5.0%(4.5.F 2.5%5.0%2.5%5.0%

145HOESD O.M.F.MRANSETIA145 HOLES. 0 1.2MM. S 0.5MM.F 5;tRANSE TI A***.'o41170075sheet0i00sheet7075 plate o04O4@ o4 00.RANSETI1C.:::145404 HOLES,0 1 .2MM.S 0 .5MM.47075e1aHOLmeSe.oftnslspcmn7075sheet[ .".-*pspe%-" ' ':' -'-'. %"44 447sheetee0 1.2amnmu spacingM.5tho04teticnsI% , " .F 5%Frcue4'1100 shaee-t.plateshee. ;*oFa, rild5'"**44*4 * 44 *4' ". '. 0 " . . " . 0- %4,444a.adamn.2"diaete" " "H: -""LE2"" " "mm"- " "4,' ". "-4M Lumspain0.5Mmmdrile. -4;. - - . " " " "" " " "" " ' ; " " '. . .4. . .,thrugththcnsof7tesl- 4seien. Fratet4-4. ., "'. -. ,""""-. . . .2' .'- ".-'

3The experiment technique is based on a 2factorial design in which 2levels (values) are chosen for each of 3 factors (parameters) D, F and S , andtests are performed for all possible combinations.e-levels for each parameter.Table 1 shows the " "andAs a measure of experimental error, threedifferent random arrays have been generated for each D, F and Sm combination;Fig. 3 shows examples of three different arrays for a specific set of D, F,and Sconditions.The factorial analysis permits an estimation of the "effects" of the holedistribution parameters D, S , and F as well as their interactive effects(DS, DF, SmF, and DS mF) on the resulting mechanical properties within therange of D, Sm, and F values in Table I. In the present case, the engineeringstrain to failure e and the yield stress afare of primary interest, althoughythe engineering stress at maximum load 0 m has also been examined.siaThe analy-thus provides the mean magnitudes of the coefficients a, b,.h inthe following linear relationship:effao cefoo -2S -F 2S -DF IS F 22 mFF2m2-D2 ma2 bm2DS FUsing the 23 factorial design, the magnitude of a particular coefficient canbe obtained by comparing the magnitude of a property (ef ay, or a ) at them( ) or (-) levels of D, Sm and F as indicated in Table I. For example, theestimated effect of hole diameter D on fracture strain e( ))-(el at D-is:-(ef at D 2.01.2 mm (-)), where ef is the mean fracture strain for thethree tests at each condition.A hole distribution parameter is determined to have a significant effecton a property if the estimated effect of that parameter is sufficiently larger-.-.,"" -"- - " ,-.'. .-'.-'. -',' *'-." :-,. " .*.:.*.,-' .'. .,- .-.,*. .,--* -':-.;;

8than the standard experimental error.This is illustrated in Table II forthe case of fracture strain in the 1100-0 Al sheet.For example, as shown inTable II,Table II.The dependence of fracture strain ef1%) for the range of holedistribution parameters listed belowParameter(s) MeanDSCoefficientabcdefghDSDFmS FDSPm For the range:D 1.2 or 2.0 smmSm Effect 7.85-1.460.92-0.05-0.310.03-0.360.340.5 or 2.0 vim, FStd. Error 0.19% 0.38 0.38 0.38 0.38 0.38 0.38 0.382.5 or 5.0%increasing the hole diameter D in the 1100-0 sheet from 1.2 to 2.0 m atconstant Sand F causes a decrease of ef by 1.46 0.38%.Inspection ofTable II shows that this and the influence of minimum hole spacing Smare theonly parameters which result in significant effects on ductility of the 1100-0Al sheet whereas the area fraction of hole F and interactive effects areinsignificant (compare the magnitudes of the effects with the standard errorsin Table II).It must be emphasized that these conclusions are based on thedata analysis for the range of conditions in Table I.Thus while the areafraction of holes undoubtedly affects fracture strain, its influence from F 2.5 to 5.0% is small.*For a given property, the standard error for coefficients b through h in Eq.I is based on a pooled estimate of the variance among three different random5rays with identical values of D, S , and F. To estimate the pooled variance, the individual variances for eacR D, Sm , and F condition are summed anddivided by the number of degrees of freedom.

9To date the 1100-0 Al sheet [plane stress, high strain hardening] and7075-T6 Al plate [plane strain between holes, low strain hardening] have beentested and analyzedl preliminary data on 7075-T6 sheet is also available.Using the analysis procedures described above, the coefficients in Eq. 1 foref, aEy ,, and Gm have been determined for these two materials and stress-stateconditions and compared to the standard error for that property.Thoseparameters which are statistically significant over the range of D, Sm,and F-values tested are summarized in Table III.Table III.The magnitudes of the coefficients (see Eq. 1) of those holedistribution parameters which are significant in influencingfracture strain ef,yield stress 0 , and stress at maximumload a . The ranges of hole paravters are: D 1.2-2.Omm,S o.5-2.0m, and F2.5-5.0%.mDPropertyMaterial0f)1100 Sheet7075 Sheet7075 Plate-1.5 0.4-0.3 0.10.4 0.20.9 0.41100 Sheet-5.0 1.44.6 1.47075 Sheet7075 Plate-20.8 4.917.2 4.9-30.0 5.6-33.1 4.91100 Sheet7075 Sheet-22.3 5.04.8 1.719.6 5.0-27.9 5.07075 Plate-28.3 7.534.6 7.5-23.6 7.5a (MPa)a(MPa)NSFm0.8 0.2Interactions(0.5 0.2)DS-3.0 1.4The results in Table III indicate several conclusions:(1) With the exception of the 7075 sheet which shows very littleductility in any condition, increasing the minimum hole spacing Sincreasesboth the strength and the ductility of the materials examined thus far.(2) An increase in hole size (from 1.2 to 2.0mi) causes a modest decreasein ductility and a comparatively larger decrease in the yield and failurestrengths. ,b*.* * S. . . . . . . .'.-,.s-m

F.10(3) Increasing the area fraction of holes from 2.5 to 5.0% causes a rapiddecrease in the strength of the 7075-T6 plate and sheet.The ductilities ofall three materials are statistically unaffected in range 2.5-5.0%.(4)Only one interaction term is noted:eff(DSmfor the 1100-0 Al.Specifically, the loss of ductility with decrease S is greater for the 1.2 ummMholes than for the 2.0 -a holes.11(5) As isillustrated in Fig. 3, the fracture path depends primarily onthe stress state between holes and is nearly independent of matrix strainhardening.e: I"Plane-stress" fractures in the 1100 and 7075 sheet follow near-identical paths roughly normal to the maximum principal stress axis.4Incontrast, the "plane-strain" fracture paths tend to be more tortuous with manyligaments oriented -450 to014*OvSome (tentative) inferences may be drawn as regards ductile fracture ofmetals (or failure of panels or plates containing multiple holes).First,minimum hole (void) spacing, or the degree of clustering, exerts a very stronginfluence on the flow and fracture of materials which contain either preexisting porosity or may easily form voids at inclusion.The effect is such"Rothat increasing minimum hole spacing [rendering hole spacings more uniform]usually increases both strength anid ductility, especially when plane straindeformation between holes exists, as in the case of 7075-T6 plate.This isdemonstrated in Fig. 4 in which the yield strength and ductility of panelscontaining equally-spaced, regular arrays of holes are compared to data fromrandom arrays; note the influence of "nearest neighbor spacings".Secondly,hole size appears to have a strong influence on yield and tensile strength;this suggests an importance of the scale of the plastic zones adjacent toboles/voids with regard to the dimensions of the ligament between holes/voids.This is demonstrated by the catostrophic fracture of the 7075-T6 Al plate.

.r -r:.,'.-,,"W.-.-Vtoo*.REGULAR ARRAYSD 1.2mm, F 5%Sque444941'.4 49.5.4444*4**44.yield random 4cdSQUARiindi44*444.4 444424.4444 444944 4 474A th.t DI AGONALwpDfor 4r HEXAGONALaCOMPRISOEGULR BETEENAN RANOM-ARAY aNeaes(MA( DigoaARANeighbor"n (m2.47454. 3 Square 7.5 497 2 RadmRado4i4.AS-o%% nreu44gutiiysfhoeel sreutory0T(MAlf%)duare as.5 utlt*;*.'-,heshwn4oecniuainoholeofVHexago% % .V al*4.aray'""2.0 46 02.8sceai0.5431.Random2.502 2.5.9Hexagoa424404745.D ao aNetilt aNihoplt.Datanforam:are9ando.9::9.*.,. .,

.-.114. specimens after one or two ligaments fail, probably due to a shear instabilitybetween closely-spaced, oriented holes.Confirmation of the above analysis depends on a more complete analysis ofthe results and on extending the range of hole contents.This would obviously"provide a much better indication of the validity of our inferences, especiallywith regard to the influence of the area fraction of holes/voids.Hot Isostatic Pressing (with Barbara Lograsso)The use of hot isostatic pressing (HIP) to compact both powders andcastings to full density has been a couercial practice for more than adecade. Over that time period, the pressure-time-temperature conditionsrequired to achieve a fully dense material have been developed by each industrial user on an empirical basis.Most of the published data currentlyavailable concerns the pressure-time-temperature conditions required toachieve a 100% dense compact.Only the very recent study of Swinkels,-.Wilkinson, Arzt, and Ashby examines HIP over a range of densities and thosedata are only for lead, tin, and PMMA.28 In this study, little attention wasdevoted in that study to the role of interparticle bonding, and theexperimental conditions chosen were such that diffusional sintering, whichresults in rounding of porosity and densification, was ignored.This studyseeks to establish pressure-time-temperature relationships whtch characterizeHIP for a range of metallic powders which exhibit different degrees ofinterparticle bonding.The results test and extend existing theories, andprovide a basis for more efficient utilization of HIP and for understandingHIP-induced effects.This could be particularly important in the consolida-tion of rapidly solidified powders in which low temperatures - short times maybe crucial to retaining the superior properties i herent in such powders.''- - .-'.S'- .",44'N%".D;:.'. .'."2.-.*-'.-0 -."""*.*. " p. """*4. -. -.*". """. . .""".o4

12In this present study, four powder materials (commercially pure titanium,Ti-6A1-4V, nickel, and 316 stainless steel) are being HIP'ed over a range oftemperatures, pressures and times to determine the pressure-temperature-timefinal density relationships. To facilitate the HIP'ing, the powder is beinghermetically sealed in metal cans, copper for T 900 C and a steel for T 900 C.To protect the powder from reacting with the can, the powder is wrapped in atantalum pouch.welded.The filled cans are then outgassed, crimped and the ends areThe compaction is performed in a ASEA Mini-Hipper pressure vesselusing argon gas as the compressing medium for pressures up to 205 MPa and5,otemperatures up to 15000C.The influence of powder shape and size and thedegree of interparticle bonding is also being examined.For the experiments,a factorial design approach is being employed so that relatively few tests areperformed per variable (time, temperature, pressure, powder size).Further-more, this encourages the use of a sequential approach consisting of settingthe initial design, testing, and then reassessing the design as the results ofeach group of data becomes available.Preliminary results for the four metallic powders being examined areshown in Table IV below, but more data is obviously required before pressure-time-temperature-final density relationships may be established.Theinter-particle bonding and densification process can also be qualitativelyexamined on the basis of the associated fracture surfaces.Figure 4 shows acomparison of Ni to Ti-6Al-4V under similar pressure-time-temperature conditions.4The behavior of these two materials can be explained in terms of theoperating mechanisms for densification during HIP'ing:diffusional creep atlow stresses, power-law creep at higher stresses, plastic flow at yet higherstresses as well as diffusional sintering at high temperatures.For example,we may compare the HIP behavior of Ti-6A1-4V to that of Ni at 700 C, by noting7

13that the uch higher flow stress of the titanium alloy at 7000C (-220 MPaverses -60 WeC for Ni) results in a lower densification rate for Ti-6A1-4v.At 800C, the flow stress of Ti-6A1-4V decreases to -80-120MPa which is-omsparable to 316 stainless steel but still greater than -40 MPa for Ni.Despite a similar flow stress at 8000C, Table IV shows that the Ti alloy hasmore rapid densification than the stainless steel.attributed to two effects:This behavior can be(a)Ti alloy has a higher diffusivity than eitherNi,or 316 stainless steel at 8000C, (b) the Ti alloy is able to dissolve itsoxide.The higher diffusivity results in more rapid densification by creep aswell as diffusional sintering as powder particles undergo interparticlebonding [compare Ni to Ti-6A1-4V in Fig. 5].Table IV.Final Densities of Four Metallic Powders Subjectedto the Indicated HIP CTiT-A-V99.1( T 235MPa35MPa1lhr9 hrs1lhr9 hrs82.3%90.0%69.2%70.6%97.810096.4I hr-76.797.899.590.292.672.497.298.485.086.875.0All material Plasma Rotating Electrode Powder (spherical); the 316 S/SITi-6Al-4V, and CP Ti powder particles are in the size range: -70 to 100mesh, and the Ni is -100 to 140 mesh.

Ti 6A1-4VNiIN11*700 0 C700 0 C.1hr1 hrIIL0BO0MPa34M a34800 0CClhr1*34IFig.1 hrMPa34 MPa8000 C8000 C1 hr1 hr103 MPa103 MPa5Scanning electron micrographs of Ni and Ti-6A1-4V powder particlesafter HIPing under the conditions noted.same magnification;note 100im marker.All micrographs are the

14Research is currently underway to determine the degree of interparticlebonding on a comparative basis as in Fig. 5, to extend the range of pressuresand times for the HIP cycles in Table IV, and to separate the densificationdue to sintering from that of diffusional creep.The Effect of Plastic Anisotropy and Grain Size on the Hydroqen Embrittleventof Titanium Sheet(with Dale Gerard)In a previous study, by Bourcier and Koss, 29 the influence of hydrogen onthe ductility of commercially pure (CP) titanium sheet has been investigatedover a range of stress states from uniaxial to equibiaxial tension.Thosedata show that hydrogen embrittlement (HE) of plastically anisotropic Ti sheetdepends on stress state, being most severe in equibiaxial tension. Quantitative metallography indicated that the effect of stress state was related tothe acceleration of (1)void formation due to strain-induced hydride fractureunder equibiaxial tensile deformation and (2) void link-up in equibiaxialtension occuring at a small void density.The comparative ease of voidformation in equibiaxial tension was in turn believed to be a consequence ofthe large degree of plastic anisotropy in the Ti sheet examined in the previousstudy. Thus the objective of this study is to examine the effect of hydrogenembrittlement of titanium sheet having low degree of plastic anisotropy. Insuch sheet material, the maximum principal stress obtained under multiaxialloading conditions should be nearly equal to that in uniaxial tension.It washoped that this would identify a means of reducing the sensitivity of thishydrogen embrittlement process to stress state.A thermomechanical treatment has been developed to break down the textureAs a result, the plastic anisotropy ratios have beenof the titanium shee

and powder metallurgy (P/M) consolidated alloys. The present research is an examination of the effects of pre-existing porosity and of matrix strain hardening on the deformation and fracture of high strength engineering alloys in general and