Homogenizing a Nickel-Bad Superalloy:Thermodynamic and Kinetic Simulation and Experimental Results
PAUL D.JABLONSKI and CHRISTOPHER J.COWEN
If the chemical inhomogeneity profile is known a priori ,kinetic modeling software such as diffusion-controlled transformations (DICTRA)can be ud to model the homogenization kinetics of an alloy.In this study,the Scheil module within the Thermo-Calc software was ud to predict the as-cast gregation prent within the Ni-bad superalloy Nimonic 105.The gregation profiles were read into DICTRA to refine the homogenization heat treatment of this alloy.The thermodynamic and kinetic modeling of the computationally predicted heat treat-ment and microstructure,and subquent experimental verification on a real casting of Nimonic 105,are prented.
DOI:10.1007/s11663-009-9227-1
ÓThe Minerals,Metals &Materials Society and ASM International 2009
I.INTRODUCTION
M ANY
Ni-bad superalloy castings and ingots
are given a homogenization heat treatment prior to further processing or hot working in order to evenly distribute the alloying elements throughout the micro-structure.[1À3]Solute inhomogenieties can adverly affect the corrosion/oxidation resistance,strength (u of alloy additions),rvice temperature (resulting from artificially lowering the melting temperature in interden-dritic regions),hot workability (from grain boundary liquation or cracking),and induce formation of unde-sired topologically clo-packed phas.Alloys that are ud in the cast form are not subjected to hot working,which can very effectively redistribute solute atoms during wrought processing.[4]Thus,the homogenization heat treatment of cast Ni-ba superalloys is especially critical and can be viewed as the last ‘‘processing’’step the component receives before being put into rvice.In the ca of solid-solution-strengthened alloys,the homogenization heat treatment is the last chance the metallurgist has to optimize the microstructure of the alloy for the required rvice conditions it will experience.Solid solution strengthening in Ni-bad superalloys is typically accomplished by alloying with refractory elements such as W and Mo.Due to their large atomic size,high melting points,and hence sluggish diffusion kinetics,the refractory elements will typically be the most difficult elements to homogenize.For cond-pha-strengthened alloys that require aging heat treatments to induce precipitation reactions (for example,gamma-prime-strengthened alloys),the
homogenization heat treatment is also critical,becau the success of the homogenization heat treatment directly dictates the distribution of the fine dispersion of cond-pha strengthening precipitates that are developed in the microstructure during aging.[5,6]
Parameters for homogenization heat treatments for newly developed alloys are defined by trial and error,past practice on alloys of similar composition (parent alloys),or extensive experimental lab work.[2,3,6]If a sample of the cast microstructure can be obtained,the gregation across a dendrite arm can be profiled and the homogenization modeled with thermodynamic mod-eling software such as diffusion-controlled transforma-tions (DICTRA).[7]Such a homogenization treatment can be important,especially for Ni-bad superalloys that are designed to be slow diffusing for high-temperature mechanical and microstructural stability.While a tradi-tional brute force approach to diffusion calculations may work to design a homogenization heat treatment for simple alloys,this approach is not practical in the ca of Ni-bad superalloys.Typical Ni-bad superal-loys can contain in the neighborhood of 10to 15alloying elements,in addition to stringent maximum limits on tramp elements,such as P and S.
II.EXPERIMENTAL
In this study,the Scheil module of the thermodynamic modeling software Thermo-Calc [8]was ud to predict the extent of chemical microgregation prent within the face-centered-cubic (fcc)austenitic matrix of the Ni-bad superalloy Nimonic 105.The target and measured chemistries of the Nimonic 105alloy cast in this study are available in Table I .Bad upon the chemical gregation predicted in the fcc pha,the kinetic modeling software DICTRA [9]was ud to design an optimized homogenization heat treatment that would fully equalize the concentration profile of each alloying element within the dendrites of the austenitic matrix.
U.S.GOVERNMENT WORK
NOT PROTECTED BY U.S.COPYRIGHT
中国地势特点
PAUL D.JABLONSKI,Metallurgist,is with the National Energy Technology Laboratory,United States Department of Energy,Albany,OR 97321.CHRISTOPHER J.COWEN,Metallurgist,is with the National Energy Technology Laboratory,United States Department of Energy,and Parsons Corporation,South Park,PA 15129.Contact wen@v Manuscript submitted November 6,2008.Article published online March 4,2009.如何画水果
182—VOLUME 40B,APRIL 2009
METALLURGICAL AND MATERIALS TRANSACTIONS B
辰时五行属什么The Scheil module and Ni databa[10]of the Thermo-Calc software was ud to predict the nonequilibrium solidification range of the alloy Nimonic105.The Scheil module is an implementation of the Scheil–Gulliver model,[11,12]which assumes that diffusion occurs infi-nitely fast within the liquid pha and that there is no diffusion in the solid phas that form.The analytical Scheil equation is commonly derived for an idealized system in which the liquidus and solidus are linear with
respect to composition.If this is true,then C s/C l=k, where C s is the composition of the solid and C l is the composition of the liquid at a given temperature.When an incremental amount of solid(df s)forms,then C s df s solute transfers from the liquid to the solid.As a result, the incremental change in the liquid composition is given by
dC l¼ðC lÀC sÞdf s=ð1Àf sÞ½1 This equation can be integrated to describe the compo-sition of the liquid as a function of fraction solid(f s) formed:
C l¼C oð1Àf sÞðkÀ1Þ½2 In the more complicated ca,in which the liquidus and solidus curves are not linear,one would need to describe C s and C l analytically and integrate Eq.[1]accordingly. The Scheil module within Thermo-Calc calculates
D f s, D C s,and D C l in a ur-lected increment of temperature until a predetermined amount of liquid remains.Fast diffusing elements such as B,C,N,and O can be allowed to back-diffu within the solid,but this was not considered in the simulation.The default results obtained from the Scheil simulation for a given alloy are the amount of solid pha formed vs solidification temperature.Not only does this show how the solidifi-cation quence progress,it also shows the phas that develop(many are nonequilibrium),as well as the solidification temperature range.After the data are obtained,the amount of each alloying element in a given pha prent can be determined.In the ca of this alloy, the weight fraction of each of the alloying elements in the fcc pha was calculated as a function of temperature. The fcc pha is the austenitic matrix of alloy105and is also the pha that solidifiesfirst from the melt.There-fore,the chemical gregation that occurs across a condary dendrite arm of this pha constitutes the undesired microgregation that needs to be eliminated through an appropriate homogenization heat treatment. Metallographic measurements on the casting showed that the condary dendrite arm spacing(SDAS)varied between47and96l m.From the measurements,a value of50l m(approximately one-half the maximum SDAS)was ud as a conrvative estimate of the required diffusion distance.The weight fraction of fcc pha was scaled to this distance and read into the DICTRA[13]software,along with the chemistry profiles across the fcc pha dendrites.
An approximately7kg heat of Nimonic105was formulated from high-purity raw materials and melted and cast in a vacuum induction furnace.The melt was cast with50°C superheat above the Thermo-Calc predicted liquidus temperature of1344°C into a100-mm-diameter round graphite mold.The mold was placed in a condary container with loo-packed sand around the outside;the sand height was above thefinal height of the solidified ingot in the mold to produce a slow cooling rate in order to mimic the cooling rate of larger sized castings of the same alloy.
The microstructure of the casting was evaluated through standard metallographic techniques,including ctioning,grinding,polishing,and etching followed by optical microscopy.The SDAS was measured using a line intercept method.A sample from the casting was subjected to the homogenization heat treatment devel-oped using the techniques outlined here.A side-by-side comparison of the as-cast and homogenized specimens was provided by mounting them together in the same metallographic mount,thus providing identical metal-lographic preparation conditions.
III.RESULTS AND DISCUSSION
The equilibrium and nonequilibrium solidification behaviors predicted for the Nimonic105alloy through the Scheil simulation performed in this study are shown in Figure1.The initial formation temp
eratures for each of the predicted phas are given in Table II as a corollary to the graphical information prented in Figure1.The Scheil module provided the temperature at which the last liquid solidified,sometimes referred to as the incipient melt point(T IMP).For this alloy, T IMP=1142°C,which is137°C below the equilibrium solidus temperature of1279°C.Any heat treatment must be carried out below this temperature in order to
Table I.Target and Measured Chemistry(in Weight
Percent)of the Nimonic105Alloy Cast for This Study Nimonic
105C Cr Mo Co Al Ti Mn Si B Target0.1514.85520 4.7 1.10.50.50.05 Measured0.1614.61 5.0220.04 4.43 1.10.510.51
0.05
Fig.1—Equilibrium and Scheil predicted solidification ranges for the
Nimonic105alloy.
METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME40B,APRIL2009—183
avoid localized melting of the interdendritic regions in the as-cast condition.Figure 2shows the calculated amount of Ti in the fcc pha as a function of temperature,and this type of calculation was performed for each alloying element.The solidification progression predicted by the Scheil simulation is directly correlated to the shape of the curve prented in Figure 2,as would be expected.As the fcc pha solidifies from the melt,Ti leaves the liquid to form the fcc pha,and its amount within the fcc pha increas.At 1312°C,the amount of Ti in the fcc pha begins to decrea,which corresponds to the Ti leaving the fcc pha to form Ti-rich MC type carbides as oppod to remaining in solution in the fcc pha.At 1250°C,the carbides (both MC and M 6C)are predicted to lo their stability,and a reincorporation of the Ti into the fcc pha is obrved.
The solute gregation obrved from the DICTRA simulations was significant.Taking the ratio of the chemistry of the fcc pha as it progress through solidification to that of the equilibrium fcc pha composition at 1142°C (shown in Table III ),which is
the terminal solidification temperature from the Scheil simulation,allowed prentation of the gregation ratio of each of the solute elements shown in Figure 3.It can be en from Figure 3that some elements have a gregation ratio over 4.The solute chemistry profiles do not vary smoothly in many cas.For example,there is a sharp change in the Ti profile at about 50pct solid formed.Molybdenum shows a shift at about 90pct solid formed.As described previously for Ti,the shifts occur whenever an additionally condary pha forms that involves an element significantly (e.g.,MC carbide forms at about 50pct solid,involving Ti,and the l pha,involving Mo,forms at 90pct solid).
Since the calculated T IMP was 1142°C,1100°C was chon as the temperature to simulate heat-treatment times of 0,10,40,and 80ks (0, 2.8,11.1,and 22.2hours).It was obrved that the refractory element Mo was not homogenized even after 80ks at 1100°C (Figure 4).However,after a short time at 1100°C,say,10ks,significant changes in the chemical profile have occurred such that a new T IMP may be calculated.It was found that,after 10ks at 1100°C,the T IMP was calculated to increa to 1275°C.Taking into account that the majority of the commercial heat-treatment furnaces top out at about 1200°C,this temperature was chon as the stepped-up homogenization temperature.As such,an additional homogenization simulation was performed:1100°C/10ks +1200°C/balance time.Si
g-nificant improvement of the alloy homogeneity was predicted even after only 30ks (8.33hours)at 1200°C after the initial 10ks at 1100°C (Figure 5).Bad on the predictions,a ction of the casting was subjected to the following two-step heat treatment:heat room temperature (RT)to 1100°C at 10°C/min,hold 3hours;heat to 1200°C at 10°C/min,hold 9hours;gas-fan cool to RT.A comparison of the as-cast and
Table II.Equilibrium and Scheil (Nonequilibrium)Formation Temperatures of the Constituent Phas of the Nimonic
105Alloy
Pha Equilibrium Formation Temperature,°C
Scheil Formation Temperature,°C家规家训图片
Fcc
13441344MC carbide 13091312M 3B 2boride 12891279M 23C 6carbide N/A 1277M 6C carbide N/A 1250l pha N/A 1148G pha
N/A
1146
N/A indicates a nonequilibrium
pha.
Fig.2—Calculated amount of Ti in the fcc pha as a function of temperature.
Table III.Calculated Chemistry of the Fcc Pha upon Final
Solidification at 1142°C Fcc Pha
C水晶功效
Cr
Mo
Co
Al
Ti
Mn
Si
At 1144°C 0.003217.68.9619.52 3.08 1.24 2.27
2.27
Fig.3—Normalized Scheil predicted gregation across a dendrite (from center to edge).
184—VOLUME 40B,APRIL 2009
卫星星历METALLURGICAL AND MATERIALS TRANSACTIONS B
solution-annealed microstructures is shown in Figure 6.Significant coring is apparent in the as-cast specimen,as evidenced by the dark (dendrite core)and light (inter-dendritic)etching regions.In contrast to this,the homogenized sample responded uniformly to the etch.Note that the dendrite outlines are still apparent in the homogenized sample.This is becau the equilib-rium eutectic products remained throughout the heat treatment.This microstructural comparison provides a qualitative confirmation of the heat treatment’s effectiveness.
It should be noted that the Scheil calculations were performed without allowing back diffusion to occur
within the solid phas.This approach yields the most restrictive criteria on chemical gregation and predic-tion of T IMP as a starting point for homogenization.In a production situation,such a conrvative approach is preferred.Nonequilibrium phas (such as l pha)were predicted along with the equilibrium MC carbides.All cond phas were ignored in the solute redistribution calculation,becau they primarily reside interdentritic-ally and the diffusion of interest occurs within the dendrite.It is acknowledged that,as the nonequilibrium phas dissolve,a localized enrichment in
solute will occur (such as Mo in the ca of l pha).However,this enrichment will be localized to the region where the pha solidified initially.What is being treated in this approach is a much larger global issue in the cast material:gregation across the dendrites,which con-stitute nearly 97wt pct of the cast material.The as-cast profiles of the gamma prime forming elements,Al and Ti,are somewhat synergistic in this alloy:high Al is accompanied by low Ti and vice versa,but not on a one-to-one basis to counter the inhomogeneity of the precipitation strengthening elements.Chromium gre-gates to the interdentritic regions,as does the Si and Mn—the latter two displaying the most
pronounced
Fig.4—Weight percent Mo as a function of distance (m )across a dendrite (from center to edge)for the following time quences at 1100°C:0,10,40,and 80
ks.
Fig.5—Weight percent Mo as a function of distance (m )across a dendrite (from center to edge)for the following time quences at 1100°C:0and 10ks;and 1100°C/10ks +1200°C/30and 70
ks.
Fig.6—(a )As-cast and (b )homogenized microstructures of Nimonic 105.The lack of contrast due to coring within the dendrites of the heat-treated sample is evidence of a successful homogenization.
METALLURGICAL AND MATERIALS TRANSACTIONS B
VOLUME 40B,APRIL 2009—185
gregation(over4times the nominal composition)—and is problematic for the weldability of the as-cast material(perhaps leading to interdendritic cracking). The solid solution strengthening element,Mo,ranked third in terms of degree of gregation,with a greater than2times enrichment in the interdendritic region of the as-cast alloy.
Molybdenum is expected to be one of the most difficult of the solid solution strengthening elements to homogenize in this alloy,and the isothermally heat-treated material was not obrved to be homogenous in its Mo profile even after80ks at1100°C.Furthermore, this time was considered to be excessively long.How-ever,it was obrved that a significant redistribution of all the alloying elements had occurred in thefirst10ks of heat treatment,suggesting that there might also be a conco
mitant increa in T IMP.The resulting dendrite tip chemistries obtained after annealing for10ks at 1100°C were ud to estimate a new T IMP of1275°C. With this information in hand,it was felt that the heat-treatment temperature could confidently be raid fol-lowing partial homogenization to increa the mobility of the solute atoms.The resulting two-step heat treat-ment was predicted to produce esntially smooth solute profiles across the dendrites.This was proved qualita-tively with obrvations in the optical microscope,which showed a microstructure for the Nimonic105alloy that was homogeneous and free of nonequilibrium phas following the two-step homogenization heat treatment.
IV.CONCLUSIONS
In summary,the Scheil module was ud to make a conrvative estimate of the T IMP and initial microg-regation across an austenitic fcc dendrite.The initial T IMP and compositional profiles were ud in DICTRA to simulate a heat treatment of the as-cast material. Initial simulations showed a lack of complete homog-enization even after80ks at1100°C.However,signif-icant solute redistribution and an increa in T IMP were obtained after10ks at1100°C,which suggested a two-step heat treatment to be appropriate.Application of this two-step heat treatment in the laboratory was confirmed to homogenize the cast microstructure of the Nimonic105alloy.
ACKNOWLEDGMENTS
The authors acknowledge the assistance of Ed Argetsinger(casting),Paul Danielson(metallography), as well as Paul Mason,Zi-Kui Liu,Qing Chen,and John A gren for helpful suggestions with the Thermo-Calc and DICTRA software.This work was per-formed as part of the United States DOE/OCDO USC Steam Turbine Consortium.Special thanks are given to R.Viswanathan,R.Purgert,P.Rawls,B.Romanowski, and M.Marrocco for their leadership and support of the program.
REFERENCES
1.E.W.Ross and C.T.Sims:in Superalloys II,C.T.Sims,N.S.
花开的过程Stoloff,and W.C.Hagel,eds.,John Wiley&Sons,New York,NY, 1987,pp.122–31.
2.J.P.Collier,A.O.Selius,and J.K.Tien:in Superalloys,R.Duhl,
G.Maurer,S.Antolovich,C.Lund,and S.Reichman,eds.,The
Metallurgical Society,Warrendale,PA,1988,pp.43–52.
3.S.A.Loewenkamp,J.F.Radavich,and T.Kelly:in Superalloys,
R.Duhl,G.Maurer,S.Antolovich,C.Lund,and S.Reichman, eds.,The Metallurgical Society,Warrendale,PA,1988,pp.53–62.
4.W.H.Couts and T.E.Howson:in Superalloys II,C.T.Sims,N.S.
Stoloff,and W.C.Hagel,eds.,John Wiley&Sons,New York,NY, 1987,pp.449–55.
植物保护5.X.Guo,H.Fu,and J.Sun:Metall.Trans.A,1997,vol.28A,p.997.
6.M.D.Trexler,B.C.Church,and T.H.Sanders:Scripta Mater.,
2006,vol.55,p.561.
7.H.E.Lippard, C.E.Campbell,T.Bjo Rklind,U.Borggren,
P.Kellgren,V.P.Dravid,and G.B.Olson:Metall.Trans.B,1998, vol.29B,p.205.
8.Thermo-Calc,version R,Royal Institute of Technology,Stock-
holm,2007.
9.DICTRA,version23,Royal Institute of Technology,Stockholm,
2007.
10.Ni-DATA,version7,Sente Software Ltd.,Surrey Technology
Centre,United Kingdom,2005.
11.E.Scheil:Z.Metallkd.,1942,vol.34,pp.70–72.
12.G.H.Gulliver:J.Inst.Metall.,1913,vol.9,pp.120–57.
13.Mobility Databa,Royal Institute of Technology,Stockholm,
2007.
186—VOLUME40B,APRIL2009METALLURGICAL AND MATERIALS TRANSACTIONS B
