11月18日是什么星座
Materials Science and Engineering A329–331(2002)621–630
Mechanisms and effect of microstructure on creep of TiAl-bad
alloys
S.Karthikeyan a,*,G.B.Viswanathan a,P.I.Gouma b,Vijay K.Vasudevan c,
Y-W.Kim d,M.J.Mills a
a Department of Materials Science and Engineering,The Ohio State Uni6ersity,477Watts Hall,2041College Road,Columbus,
OH43210,USA
b Department of Materials Science and Engineering,State Uni6ersity of New York,Stony Brook,NY11794-2275,USA
c Department of Materials Science an
d Engineering,Uni6ersity of Cincinnati,497Rhodes hall,PO Box210012,Cincinnati,OH45221,USA
d Uni6ersal Energy Systems,4401Dayton-Xenia Road,Dayton,OH45432,USA
Abstract
Transmission Electron microscopy studies on crept samples of an equiaxed Ti–48Al alloy deformed to strains near the minimum strain rate show a microstructure dominated by unit1/2[110]type dislocations.The dislocations are pinned by jogs of varying heights.The jogged-screw model is adopted,where the rate controlling step is assumed to be the non-conrvative dragging of the jogs along the length of the screw dislocations.The prence of tall jogs and the existence of a stress-dependent upper bound to the height of tall jogs which can be dragged have been incorporated into the model.The modifications lead to excellent agreement with experimental data.The evolution of the creep curve is also qualitatively predicted.In contrast,lamellar structures show a highly inhomogeneous deformation behavior with dislocation activity in lamellae above a critical thickness and negligible activity below that limit.This limit is related to the minimum stress required to cau channeling of dislocations.The obrvation of jogged gments in the thicker lamellae suggests that a modification of the jogged-screw velocity law could be ud by incorporating an effective stress approach.Qualitative analysis of a probable method of evaluating the creep rate is discusd. Finally,microstructural changes during the aging process in K5and K5SC alloys have also been studied.Aging caus the dissolution of metastable h2.The effect of Si and C on the aging behavior a
nd precipitation of the silicide and carbide particles during aging and following creep is discusd.The results suggest that microstructural stability is critically important in order to achieve the highest possible creep strengths.©2002Elvier Science B.V.All rights rerved.
Keywords:Titanium aluminide;Equiaxed;Lamellar;Jogged screw model;Carbides
/locate/ma
1.Introduction
Ordered intermetallic compounds of the Ti–Al sys-
tem have been the subject of extensive rearch during
the last decade or so due to their potential application
as high temperature structural materials.Of special
technological interest are alloys of this system having a
two-pha microstructure(k and h2),which posss a
very high specific stiffness,good oxidation resistance up
取消定时关机命令
to900°C,and high temperature strength( 600MPa
at600°C)[1].Due to their excellent specific stiffness,
TiAl alloys have potential applications in aircraft en-
gines in locations where clearances are crucial,such as
frames,al supports and cas.Their high stiffness
also shifts vibration frequencies upward,which is usu-
ally beneficial for structural components.The other
perceived us are as high-pressure compressor stators,
入职周年祝福low-pressure turbine blades,transition duct supports in
the aerospace industry,and turbocharger wheels and
combustion valves in the automobile industry.g-TiAl
appears to be capable of substituting for superalloys in
certain applications with substantial weight saving and
minimal redesign.
The understanding of creep behavior in the materi-
als is of utmost importance in view of their potential
for high temperature applications.Even though signifi-
cant progress has come about in the understanding of
low temperature deformation behavior of g-TiAl,and
despite the significant body of work done in the area of *Corresponding author.Tel.:+1-614-688-3409;fax:+1-614-292-
1537.
0921-5093/02/$-e front matter©2002Elvier Science B.V.All rights rerved.
PII:S0921-5093(01)01659-8
S.Karthikeyan et al./Materials Science and Engineering A329–331(2002)621–630 622
creep deformation,the prent understanding of creep mechanisms,and the effect of alloying and microstruc-ture,is very limited[2].Clearly a difficulty is that creep is influenced by numerous microstructural parameters [27]and many of the investigations in g-TiAl have been made on different structures.Widely-reported values of creep activation energies in equiaxed structures similar to tho for lf diffusion,and stress exponents in the range4–6,both suggest an underlying creep mecha-nism involving recovery of dislocations by climb[2–10]. However,the abnce of subgrain formation(in the minimum creep rate regime)suggests that the power law behavior is quite different from that of a pure metal.In this paper,we summarize a propod model for creep in g-TiAl which is bad on a modification of the jogged screw model.The justification for this model is prented,along with a comparison of the model with measured creep data.
Two-pha TiAl alloys can have three broad cate-gories of microstructures:equiaxed,duplex and fully lamellar.It is now established that lamellar microstruc-tures,consisting of alternating g-TiAl and a2-Ti
3Al laths,have a far superior creep resistance when com-pared to the equiaxed and duplex microstructures.Sev-eral explanations have been cited for the superior creep resistance of the fully lamellar structures,such as the prence of interlocked lamellar boundaries[2,5,16], inhomogeneous deformation characteristics of lamellar structures[2,17],composite behavior and reduction of the slip distance due to the prence of the interlamellar boundaries[2,18–20].Deformation modes including twinning[2,9,11,12],dynamic recrystallization[8,13], and interface sliding[4,14,15,32]have also been re-ported.So it becomes imperative to understand the exact nature of creep strengthening in the fully lamellar structures.In this paper,we also comment on the possible sources of the superior creep strength of lamel-lar structures.
Modifications to the alloy chemistry have been at-tempted in order to maintain a balance between the ductility and high temperature creep resistance.Such modifications by the addition of substitution elements like Cr,Nb and Mo,and interstitial elements such as C and Si has met with some success.The effect of Cr is to enhance room temperature ductility[21],while Nb and W em to reduce creep rates becau of their low diffusivities[1,2,22].The effect of interstitial elements is more dramatic[22,23].C has been reported to solute strengthen by atmosphere drag effects at low concentra-tions[22].The primary mechanism of strengthening upon the addition of C and Si,however,
ems to be precipitation strengthening due to carbide and silicide formation[19,21–24].As discusd in this paper,intrin-sically related to the process of precipitation is the dissolution of h2which takes place during aging.The microstructural changes during aging or exposure to creep conditions must be considered in order to develop a more complete understanding of creep strengthening in the complex alloy compositions.
2.Material preparation
The creep results and microstructural evidence de-scribed in this paper were obtained from four different ts of materials:binary Ti–48Al in both equiaxed and lamellar structures,and a‘K5’[28,29]alloy with and without C and Si additions.The binary alloy had a nominal composition of Ti–47.86Al–0.116O–0.016N–0.041C–0.076H.Cylindrical blanks of the forged alloy were heat treated at1473K in the(h+k)two-pha region just above the eutectoid temperature and fol-lowed by a stabilization treatment at1173K for6h. This heat treatment yields a near-gamma microstruc-ture for which the volume fraction of h2is small and k grains are in the equiaxed morphology with a grain size of50m m.The samples will be referred to as Ti–48Al(EQ)henceforth.Blanks of similar composition (Ti–48Al)were heat treated at1658K(above the a transus)to ensure a fully lamellar structure.The samples were also given a similar stabilization treat-ment at1173K for6h.The sam
ples will be referred to as Ti–48Al(FL)henceforth.Cylindrical creep sam-ples were prepared from the blanks and tested in tension at two temperatures,768and815°C.
The nominal composition of the K5alloy is Ti–46.5Al–2.0Cr–3.0Nb–0.2W.The K5SC alloy has the C and Si as interstitial elements and the nominal composi-tion is Ti–46.0Al–1.8Cr–3.0Nb–0.2W–0.1C–0.2Si. The alloys were isothermally forged to90%.Blanks of K5and K5SC were given a heat treatment involving solution treatment(in the single pha a phafield)at 1633K for5min and furnace cooling to1473K, followed by air cooling to room temperature.The microstructure obtained thus is fully lamellar.An aging treatment of1173K for24h was given to both K5and K5SC samples.Microstructures were studied,both be-fore and after the aging treatment.Parallelepiped sam-ples were machined from the aged samples and creep tested in compression at760°C.
Stress increment tests and monotonic tests were per-formed to obtain the stress dependence of the minimum creep rate.Thin foils for transmission electron mi-croscopy obrvations were prepared from discs c-tioned normal to the stress axis.The foils were thinned using a twin jet electro-polisher using a solution consist-ing of65%ethanol,30%butan-1-ol and5%perchloric acid,at a voltage of20V,current of30mA and temperature of−40°C.Obrvations on the mi-crostructures were conducted on a Philips CM200 transmission electron microscope operated at200kV. Conventional T
EM techniques were employed.Cen-
S.Karthikeyan et al./Materials Science and Engineering A329–331(2002)621–630623
tered darkfield images using h2reflections were ud to image the h2laths and the measurements of the lath spacing were done by tilting the sample so as to get the lamellar interfaces edge-on(the interfaces are parallel to the electron beam).A line intercept method was ud for measuring the lamellar thickness distribution and the h2volume fraction.
3.Creep behavior
The strain rate versus strain curves for the Ti–48Al (equiaxed and FL)as well as that for K5and K5SC have been shown in Fig.1.Examination of the curves reveals that the materials exhibit a normal primary transient,a brief minimum creep regime,followed by an extended tertiary regime(even for samples tested in compression).It is evident that the Ti-48Al(FL)is more creep resistant than the Ti–48Al(EQ)structure,while the K5alloys containing Cr,Nb and W em to have a much better creep respon when compared to the binary alloy.As discusd later,it is important to recognize that the lamellar spacing is significantlyfiner for the K5ries alloys than for the binary ca.The aged K5SC alloy ems to have the best creep respon, with minimum creep rates an order of magnitude lower
than that of K5.Creep tests carried out on unaged K5SC samples show a much poorer creep respon when compared to K5SC aged or even K5aged.In the discussion to follow,we will attempt to shed light on the relative ranking of the different alloys and mi-crostructural forms.
For limited stress ranges,the creep respon in g-TiAl,both in the equiaxed and fully lamellar condi-tions,ems to obey a power law type Dorn equation relating the strain rate to the stress as
m=A·|n·exp(−Q/RT),(1) where A is a constant,Q is the creep activation energy, R is the gas constant and T,the temperature.In Fig.2, it can be en that the equiaxed structures exhibit a stress exponent n of 5.Similar results have been widely reported[2,4–10].The stress exponents for fully lamellar structures however em to be clearly lower thanfive( 3)at lower stress and significantly larger thanfive( 8)at higher stress.Higher stress expo-nent values for lamellar structures has been frequently reported[4,17,18].
The next ction will prent TEM obrvations of deformed microstructures in both equiaxed and lamel-lar forms,and relates the obrvations to possible creep mechanisms.We then attempt to rationalize the obrved stress dependencies and their variation with microstructure.
Fig.1.The strain rate versus strain curves for various fully lamellar structures and an equiaxed alloy a
re shown.The fully lamellar structures have lower minimum creep rates than the equiaxed ca.The effect of C/Si addition is clearly en by comparing the aged states of the KS and K5SC alloys.
S .Karthikeyan et al ./Materials Science and Engineering A 329–331(2002)621–630
624Fig.2.Strain rate versus stress curves for equiaxed and fully lamellar Ti –48A1alloy and for the fully lamellar KS alloy.Note the lower stress exponents at lower stress and higher stress exponent at higher stress for the lamellar structures.
tion and appear to be frequently pinned along their lengths,as can be en in Fig.3.The gments on either side of the pinning points are en to be bowed-out,forming local cusps along the length of the dislocations.The average spacing between the apparent pinning points is 200nm.Tilting experiments in the TEM have con firmed that the cusps are frequently associ-ated with jogs on the screw dislocations [10].
The obrvation of cusped screw gments in the equiaxed structures,and the general abnce of sub-grains,suggests that creep rate may be controlled by the non-conrvative motion of jogs along the length of the screw dislocations [10].The basic premi of this model is that jogs are formed on screw gments by the collision of two migrating kinks formed on two differ-ent glide planes.As the screw gment advances,kinks would tend to collect at the jogs causing them to grow.Subquent motion of the screw dislocation then requires the non-conrvative dragging of the jogs (if lateral,glide motion is dif ficult).A balance between the work done by the applied force and the chemical force exerted by the vacancies produced by such non conr-vative motion of jogs gives ri to the velocity of such jogged screw dislocations [10]:6s =
4y D s h
·[exp ~V l
hkT
−1],
电烙铁怎么用
(2)
where D s is the lf diffusion coef ficient,h is the jog
height,~is the applied shear stress,V is the atomic volume,l is the spacing between the jogs,k is the Boltzman ’s constant and T ,the temperature.As the jogs grow taller,they could reach a critical height h j ,above which the oppositely signed,near edge gments attached to the top and bottom of the jog can bypass each other [10]:h d =
Gb
8y (1−w )~
,
(3)
where G is the shear modulus,b is the Burgers vector and w ,Poisson ’s ratio.So jogs taller than h j can be assumed to act as a source now,rather than being dragged along with the screw dislocation.Taking into account the stress dependence of h j ,dislocation density z (through Taylor ’s expression)and the velocity,the jogged screw model can be suitable modi fied to give an expression for the minimum creep rate.k =
y D s
气度
南通工贸技师学院d
~ 2
小企业管理制度
桃花髻exp
~V l
4i h d kT
−1n
,
(4)
where h is the Taylor factor which controls the disloca-tion density,and i is a parameter that characterizes the average jog height.
Fig.4shows the agreement between the model and the experimental data for equiaxed Ti –48Al.The pri-mary and extended tertiary transients obrved may originate from the evolution of jog height as a function
Fig.3.TEM micrograph showing the deformation structure in crept equiaxed Ti –48A1.Screw gments are en with numerous pinning points.The gments on either side of the pinning points are signi fi-cantly bowed.
4.Microstructural obrvations and discussion
4.1.Equiaxed microstructure
The deformation microstructure in the equiaxed Ti –48Al alloy,at creep strains corresponding with the minimum creep rate,is dominated by 1/2[110]type dislocations.There is little tendency for subgrain for-mation as might be expected for an n 5behavior.Similar dislocation structures have been frequently re-ported to dominate creep microstructures [8–10].The dislocations tend to be elongated in the screw orienta-
S .Karthikeyan et al ./Materials Science and Engineering A 329–331(2002)621–630625
of strain.At the beginning of the deformation,intrinsic sources active on loading cau screw dislocations to develop small jogs.As the average height of the jogs is small,it is relatively easy to drag them,leading to a large initial strain rate.As the strain increas,the average jog height also increas which leads to a signi ficant drop in the strain rate.Eventually as the jogs grow even taller,they start acting as sources and there is an increa in the total dislocation density.The new dislocations created would have a distribution of jog heights which is skewed towards short heights.This would lead to an increa in strain rate in the tertiary stage.Dlouhy and co-workers have offered a different explanation for the form of the transient as being related to the build-up of compatibility stress between grains,and the ont of twinning,due to the dif ficulty in activating c -component dislocations [31]
4.2.Lamellar microstructures
It has been suggested previously that creep in the fully lamellar structures is controlled by the soft mode
of deformation [2,4,14,15].Soft mode deformation takes place by shear parallel to the lamellar interfaces.The mean free path of the mobile dislocations in the soft mode is related to the domain or
colony size,which is two to three orders of magnitude larger than the lamellar spacing,making this slip mode relatively weak.Soft mode deformation,particularly in the lamellar interface,has been propod as the primary mode of creep deformation [4,14,15,32].If this were the ca,then a finer lamellar spacing would result in more interfaces and hence a larger creep rate.Even though there is some evidence that there is an optimum lamel-lar spacing and too fine lamellar structures could cau weakening [25],a majority of work done in this area suggests that a finer lamellar structure yields lower creep rates,suggesting that other mechanisms may be dominant [5,19,20].Twinning has also been frequently propod as a creep mechanism [4,11,12,14,30],though it ems prevalent only at larger stress and strain levels [4,14].
Microstructural obrvations show minimal twinning in our samples (at minimum creep rate)and creep data supports the view that reducing the lamellar spacing reduces the creep rates.In light of this,it appears that dislocation activity within the laths may be highly signi ficant in controlling the creep respon.This is supported by microstructural evidence of dislocation activity (1/2[110]type dislocations)within the laths.However,deformation is inhomogeneous,with smaller dislocation densities in the thinner laths and signi fi-cantly larger densities in the thicker laths (Fig.5).Such inhomogeneity in deformation microstructure,and its possible role in increasing the stress exponents in lamel-lar structures,has been suggested earlier [17].Lamellar interfaces of both types (k /k and k /h 2)appear to effec-tively constrain deformation to individual k -laths,with little evidence for direct transmission under creep conditions.
The thinner lamellae have hard-mode dislocations channeling through them,while the thicker lamellae have near-screw dislocations that are often cusped in con figurations similar to tho in equiaxed structures (Fig.6).The stress to move the channeling disloca-tions through a thin,capped film,must overcome a bowing stress which Nix has provided an expression for [26]:~b =
Gb 4yu ·ln u
b
,(5)
where u is the lamellar spacing.The effective stress on the channeling dislocations would be given by ~applied −~b .As the lamellar thickness decreas,the effective stress available to propagate dislocations decreas.The fully lamellar structure often has a distribution of lamellar thickness.So for a given applied stress,lamel-lae thinner than a critical cut off u c would not experi-
Fig.4.Comparison of predicted and measured creep rates for two temperatures using the modi fied jog-screw model for equiaxed struc-tures.
Fig.5.TEM micrograph showing the deformation structure in crept,
lamellar Ti –48A1.Deformation microstructure is inhomogeneous with signi ficantly larger dislocation density in the thicker laths.
