Strengthening Mechanisms in Polycrystalline Multimodal

Nickel-Base Superalloys
R.W. KOZAR, A. SUZUKI, W.W. MILLIGAN, J.J. SCHIRRA, M.F. SAVAGE,
and T.M. POLLOCK

Polycrystalline c-c¢ superalloys with varying grain sizes and unimodal, bimodal, or trimodal
distributions of precipitates have been studied. To assess the contributions of specific features of
the microstructure to the overall strength of the material, a model that considers solid-solution
strengthening, Hall–Petch effects, precipitate shearing in the strong and weak pair-coupled
modes, and dislocation bowing between precipitates has been developed and assessed. Cross-
slip-induced hardening of the Ni3Al phase and precipitate size distributions in multimodal
microstructures are also considered. New experimental observations on the contribution of
precipitate shearing to the peak in flow stress at elevated temperatures are presented. Various
alloys having comparable yield strengths were investigated and were found to derive their
strength from different combinations of microconstituents (mechanisms). In all variants of the
microstructure, there is a strong effect of antiphase boundary (APB) energy on strength.
Materials subjected to heat treatments below the c¢ solvus temperature benefit from a strong
Hall–Petch contribution, while supersolvus heat-treated materials gain the majority of their
strength from their resistance to precipitate shearing.

DOI: 10.1007/s11661-009-9858-5

The Minerals, Metals & Materials Society and ASM International 2009

I. INTRODUCTION then required to determine the extent of the improve-

POLYCRYSTALLINE nickel-base superalloys are
capable of maintaining strength in excess of 1 GPa at
temperatures up to 800 C. Because of this, these
superalloys are widely used in aircraft engines and
power generation turbines as turbine blade and disk
materials. The ever-present drive to improve perfor-
mance and efficiency by operating engines at increas-
ingly higher temperatures has led to the continual
development of new alloys that retain their strength at
higher temperatures.
Traditionally, in the design of new superalloys, the
compositions of existing alloys and the specific features
of processing are incrementally changed to achieve
improved properties. Extensive mechanical testing is
R.W. KOZAR, formerly Graduate Student, with the Department of
Materials Science and Engineering, University of Michigan, Ann Arbor,
MI 48109, is Senior Engineer, with Materials Technology, Bettis Atomic
Power Laboratory, West Mifflin, PA 15122. Contact e-mail: rwkozar88
@yahoo.com A. SUZUKI, formerly Post Doctoral Research Fellow,
with the Department of Materials Science and Engineering, University
of Michigan, Ann Arbor, MI 48109, is Metallurgist, with Ceramic and
Metallurgy Technologies, GE Global Research, Niskayuna, NY 12309.
W.W. MILLIGAN, Professor, is with the Department of Materials
Science and Engineering, Michigan Technological University, Hought-
on, MI 49931. J.J. SCHIRRA, Manager, is with the Aircraft Group,
Pratt & Whitney, East Hartford, CT 06108. M.F. SAVAGE, formerly
Staff Engineer, with Materials and Processes Engineering, Pratt &
Whitney, East Hartford, CT 06108, is Manager, Group Strategy and
Development, Pratt & Whitney, East Hartford, CT 06108. T.M.
POLLOCK, L.H. and F.E. Van Vlack Professor, is with Materials
Science and Engineering, University of Michigan, Ann Arbor, MI
48109.
Manuscript submitted September 17, 2004.
Article published online May 28, 2009
ment. Although this process has resulted in many useful
alloys, it is expensive and time consuming; the design
process could benefit from a greater contribution of
process and property modeling.
Commercial polycrystalline superalloys that are used
for turbine disk alloys possess complex microstructures
that are optimized for a range of mechanical proper-
ties.[1–4] Typical premium wrought nickel-base superal-
loys consist of a c matrix that contains up to three
different size distributions (approximately 0.01, 0.1, and
1.0 lm in diameter) of L12 (ordered fcc) c¢ precipitates.
The smaller two populations of precipitates are coherent
with the c matrix, while high-angle boundaries exist
between the micron-sized c¢ precipitates and the c
matrix. Due to the presence of ordered Ni3Al precipi-
tates, the strength of these alloys often increases between
room temperature and 800 C. These nickel-base super-
alloys also have excellent creep properties.[5,6]
Given the presence of ordered precipitates and the
exceptional strength of this class of materials, superal-
loys have been the subject of a range of analytical
models for strengthening.[7–13] A feature of virtually all
of these models is that they consider simple microstruc-
tures in which only one or two strengthening mecha-
nisms are active. These models are not directly
applicable to the complex microstructures present in
the commercial forms of these premium materials, in
which many strengthening mechanisms are active con-
currently. The goal of this research was to quantitatively
assess the influence of a range of microstructural
features on strength across a range of temperatures.
This was accomplished by the development of a model

1588—VOLUME 40A, JULY 2009 METALLURGICAL AND MATERIALS TRANSACTIONS A

that considers all relevant strengthening mechanisms for
a given microstructure, integrating and building on
previous analyses of strengthening in nickel-base alloys.
Transmission electron microscopy (TEM) was used
to investigate the operation of specific mechanisms and
to discern transitions in deformation modes unique to
these multimodal microstructures. Finally, the relation-
ship between specific features of the microstructure and
their overall contribution to strength is assessed with the
model and is compared to experimental data.

II. EXPERIMENTAL MATERIALS

AND PROCEDURES
A. Experimental Materials
The microstructural variation induced by commer-
cially relevant regimes of processing and heat treatment of
the nickel-base alloy IN100 motivated the experiments
and modeling. Experiments on IN100 and an alloy with a
modified chemistry but a microstructure similar to IN100
were used for validation of the yield strength model. The
materials were chemically and microstructurally charac-
terized in detail by Wusatowska-Sarnek et al.[14] using
transmission electron microscopy (TEM), scanning elec-
tron microscopy (SEM), and chemical analysis tech-
niques. For the IN100 alloy, the phase compositions were
measured by electrochemical extraction with inductively
coupled plasma analysis. The volume fractions of the
phases were measured by SEM and TEM analysis and this
information was incorporated into a thermodynamic
database that was subsequently used to calculate the
phase compositions in the modified chemistry alloy. The
compositions of these two alloys along with the compo-
sitions of the c and c¢ phases for IN100[14] are summarized
in Table I. For the modified chemistry alloy, changes in
the c and c¢ compositions were calculated on the basis of
expected (small) differences in partitioning of the elements
to the two phases.
All materials were subjected to standard powder
metallurgy hot compaction, extrusion, and isothermal
forging procedures. Following this, the materials were
subjected to several heat treatments, in order to obtain a
variety of microstructures. The typical heat treatment,
shown schematically in Figure 1 for IN100, consisted of
a subsolvus or supersolvus heat treatment (for IN100,
the c¢ solvus temperature is 1185 C), followed by
stabilization and aging heat treatments. These three heat
treatments establish the final microstructure of the alloy.
Most of the samples in this study were subjected to a
standard subsolvus heat treatment at 1143 C for
Fig. 1—Typical processing for superalloy disk material.
2 hours with a 1-hour stabilization at 982 C, followed
by gas fan cooling and then an 8-hour aging step at
732 C followed by air cooling. Heat treatments that
vary from this will be presented as appropriate.
B. Microstructure
Figures 2(a) through (c), (d) and (e), and (f) and (g)
show the microstructures at different scales of IN100
given the standard subsolvus heat treatment, the IN100
given a supersolvus heat treatment, and the supersol-
vus alloy with a modified chemistry, respectively.
Figures 2(a) and (b) are courtesy of Wusatowska-
Sarnek et al.[14] Figures 2(d) through (g) were prepared
using an etchant consisting of 50 vol pct lactic acid,
30 vol pct HCl, and 20 vol pct HF. Figure 2(c) was
acquired using a PHILIPS* STEM CM12 transmission
*PHILIPS is a trademark of Philips Electronic Instruments Corp.,
Mahwah, NJ.
electron microscope. Foils for the TEM were ground to
a thickness of 0.12 mm in three stages, using 320-, 600-,
and 1200-grit SiC papers sequentially. The samples were
then twin-jet electropolished with a solution of 340 mL
methanol, 50 mL perchloric acid, 45 mL distilled water,
and 65 mL butyl cellusolve at –40 C with a voltage of
20 to 25 V.

Table I. Chemical Composition of Alloys and Their Microconstituents (Weight Percent)

Alloy/Phase Designation Ni Al Cr Co Mo Ti V Fe C Zr B Y
IN100 56.0 4.9 12.3 18.3 3.3 4.3 0.70 0.10 0.06 0.02 0.02 —
c phase, IN100 38.7 2.25 24.5 27.8 5.73 0.93 0.05 — — — — —
c¢ phase, IN100 71.8 7.06 2.59 8.94 1.42 6.97 1.23 — — — — —
Modified alloy 52.9 4.5 14.5 20.0 3.6 3.6 0.70 0.10 0.08 0.02 0.02 0.07
c phase, modified alloy 36.0 1.83 27.2 28.9 5.31 0.66 0.05 — — — — —
c¢ phase, modified alloy 69.7 7.16 3.37 10.7 1.64 6.14 1.34 — — — — —

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 40A, JULY 2009—1589


Fig. 2—Microstructure of (a) through (c) subsolvus IN100, (d) and
(e) supersolvus IN100, and (f) and (g) supersolvus modified chemis-
try alloy. (a) and (b) Courtesy of Wusatowska-Sarnek et al.[14] (c)
through (g) Etched according to Ref. 14.
The subsolvus heat treatment for IN100 results in a
trimodal distribution of precipitates. The microstructure
of subsolvus IN100, shown in Figures 2(a) through (c),
consists of 40 to 45 pct c matrix and 55 to 60 pct c¢
precipitates by volume. The c matrix, appearing as light
gray in Figure 2(a), contains two distributions of
smaller c¢ precipitates; for this sample, the grain size is
3.1 lm. Also appearing in Figure 2(a) as the darker
constituent are large c¢ precipitates, referred to as
primary c¢. This population of Ni3Al is large in size,
due to the fact that it is not solutionized and coarsens
during the subsolvus thermomechanical processing and
final heat treatment. In this sample, these primary c¢
precipitates comprise 20 pct of the microstructure and
1590—VOLUME 40A, JULY 2009
have an average size of 1.43 lm. The primary c¢
precipitates are not isolated and high-angle grain
boundaries often exist between adjacent primary c¢
precipitates.
In Figure 2(b), smaller secondary c¢ precipitates can
be resolved within the c matrix. Secondary c¢ precipi-
tates nucleate and grow during cooling following the
subsolvus heat treatment and in the subsequent stabil-
ization cycle. The sample in Figures 2(a) and (b) had an
average secondary c¢ diameter of 170 nm. Tertiary c¢
precipitates form in the regions between the secondary c¢
precipitates during the cooling from the stabilization
cycle and final lower temperature aging step; they can be
seen in Figure 2(c). Tertiary c¢ precipitates may have an
average size as large as 20 nm, but for the standard heat
treatment shown in Figure 2(c), the average size is
12 nm.
The supersolvus heat treatment of IN100 dissolves the
primary c¢, resulting in a bimodal distribution of
precipitates for a series of heat treatments similar to
those described earlier. Figures 2(d) and (e) shows the
microstructure of IN100 after a supersolvus heat treat-
ment at 1205 C for 2 hours and subsequent stabiliza-
tion and aging at 982 C for 1 hour and 732 C for
8 hours, respectively. As with subsolvus IN100, the
microstructure of supersolvus IN100 consists of 40 pct c
matrix and 60 pct c¢ precipitates by volume, with the c
matrix containing two distributions of c¢ precipitates.
The average grain size of the supersolvus IN100
shown in Figure 2(d) is 34.4 lm, which is an order of
magnitude larger than that of the subsolvus IN100.
Figure 2(d) also illustrates the absence of primary c¢
precipitates in supersolvus IN100. In Figure 2(e), sec-
ondary c¢ precipitates can be resolved within the c
matrix, because they have an average size of 340 nm and
a higher volume fraction than in subsolvus IN100.
Tertiary c¢ also exists in supersolvus IN100; it has a
similar average size but a higher volume fraction in
comparison to subsolvus IN100 (Figure 2(e)).
The microstructure of the modified chemistry alloy is
similar to supersolvus IN100, except that it has a lower
volume fraction of c¢ precipitates (53 pct). The modified
chemistry alloy, shown in Figure 2(g), has an average c
grain size of 26.8 lm, which is smaller than the
supersolvus IN100. Figure 2(g) shows that the second-
ary c¢ precipitates are only 110 nm, considerably smaller
than in the supersolvus IN100 alloy, due in part to the
lower solution heat-treatment temperature.
Three additional unimodal and bimodal microstruc-
tures established by the supersolvus heat treatment of
unmodified IN100 were studied. These were as follows: a
unimodal microstructure of secondary c¢ (mean preci-
pitate diameter of 180 nm) was achieved by supersolvus
heat treatment at 1205 C for 2 hours, followed by
cooling at 25 C/min, then stabilization heat treatment
at 990 C for 1 hour; a bimodal microstructure of
secondary c¢ (mean precipitate diameter of 180 nm) and
tertiary c¢ was achieved by supersolvus heat treatment at
1205 C for 2 hours, followed by cooling at 25 C/min,
then stabilization and aging heat treatments at 990 C
for 1 hour and at 732 C for 8 hours, respectively;
and a bimodal microstructure of secondary c¢ (mean
METALLURGICAL AND MATERIALS TRANSACTIONS A
precipitate diameter of 564 nm) and tertiary c¢ was
achieved by supersolvus heat treatment at 1205 C for
2 hours, followed by cooling at 10 C/min, then stabil-
ization and aging heat treatments at 990 C for 1 hour
and at 732 C for 8 hours, respectively.
C. Mechanical Testing
Tensile testing was performed to measure the yield
strength and strain-rate sensitivity of experimental
materials at temperatures ranging from 22 C to
760 C. Testing was conducted on a screw-type Instron
5582 (Instron, Norwood, MA). Strain measurements
were taken with a linear variable displacement trans-
ducer (LVDT). The test temperature was regulated to
±3 C in a resistance-heated furnace. Cylindrical sam-
ples with a 4.09-mm-diameter gage section were strained
to within the range of 0.2 to 0.5 pct past yielding at a
rate of 5 9 10
5/s; the strain rate was then increased to
5 9 10
4/s. Strain-rate sensitivity was calculated using
the method of Venkadesan et al.[15]
III. STRENGTHENING MECHANISMS IN IN100
When viewing the microstructures in Figure 2, it is
apparent that multiple strengthening modes may be
active in these materials. Possible operative mechanisms
include solid-solution strengthening of the c matrix and
c¢ precipitates, grain size hardening of both the c matrix
and the primary c¢ precipitates, and shearing of each
population of precipitates. In addition, depending on
the size of the c¢ precipitates, the mechanism limiting the
dislocation motion may be weak pair coupling shearing,
strong pair coupling shearing, or cross-slip-induced
hardening at elevated temperatures. Each of these
mechanisms will be discussed in more detail in the
following sections. Bowing between precipitates was
considered but, due the high volume fraction and small
size of the c¢, bowing was not generally operative.
Coherency strengthening was also considered but, in the
alloys being studied, the precipitate-matrix misfit was
near zero, so misfit strengthening was neglected.
In the following sections, analytical expressions that
account for the strengthening due to each of these
individual mechanisms are introduced, the exact forms
of the equations used in the model will be presented and
the mechanism(s) responsible for the strengthening due
to each microconstituent will be discussed.
A. Grain Size Hardening
The most commonly encountered hardening mecha-
nism is that associated with variations in grain size.
Strengthening is well known to be inversely proportional
to the grain size; this phenomenon has been attributed
to the existence of grain boundaries as sources of
dislocations and barriers to deformation. In IN100,
high-angle grain boundaries serving as sources for
dislocations are present both between c-c¢ grains and
between the large primary c¢ precipitates.
METALLURGICAL AND MATERIALS TRANSACTIONS A
According to the Hall–Petch relation,[16] the extent of
grain size hardening can be described by
r ¼ ro þ ky d
1=2
m ½1
where r is the yield strength, ro is the friction stress, ky is
the Hall–Petch coefficient, and dm is the average grain
size. When considering measurements of the Hall–Petch
constant ky for pure Ni and a range of superalloys,[17–19]
a value of 750 MPa lm½ was used in all analyses
reported here.
The form of the equation used in the model was
modified slightly to account for the fact that the
mechanism does not provide strength to the entire
microstructure, but only to the volume fraction f
occupied by the grains of interest. Also, ro was omitted,
because it is accounted for by other mechanisms
considered separately in the model. Including these
modifications, the Hall–Petch contribution is
Dr ¼ f ky d
1=2 ½2a
m
where Dr is the strengthening increment due to grain
size hardening and f is the fraction of the alloy occupied
by either the c-c¢ grains or large, primary c¢ separated by
high-angle boundaries.
In Ni-base superalloys, only a fraction of the primary
c¢ precipitates are in contact with other primary c¢
precipitates; therefore, the Hall–Petch contribution
attributable to primary c¢ is less than Eq. [2a] predicts.
To account for this, several hundred primary c¢ precip-
itates were inspected and the fraction of primary c¢ with
adjacent primary c¢ precipitates was measured. It was
found that 2/3 of primary c¢ had adjacent primary c¢
precipitates. To account for this, Eq. [2a] was modified to
2
Dr ¼ f kyc0 d
1=2
m ½2b
3
for primary c¢ precipitates. This is clearly an approxi-
mate estimate needed to capture the contribution of the
varying primary c¢ size on the overall strength. More
detailed modeling of deformation in this type of duplex
system would be useful, to establish a sound analytical
basis for this aspect of strengthening. In the present
analyses, this contribution to strengthening is small, due
to the low volume fraction of primary c¢.
B. Solid-Solution Strengthening of c Matrix
The theory for solid-solution hardening of the nickel
matrix is based on a model proposed by Gypen and
Deruyttere,[20] which assumes a superposition of
strengthening of individual solutes that individually
have differing potencies.[21–24] Because solute spacing is
proportional to the square root of the concentration, the
resultant strengthening is given by
X  dr pffiffiffiffiffi
Dr ¼ pffiffiffiffiffiffiffi Ci ½3
i
dCi
ffiffiffiffiffi is a strengthening coefficient that reflects the
where pdr
dCi
strengthening potency of each alloying element.
VOLUME 40A, JULY 2009—1591
 Table II. Solid-Solution-Strengthening Constants
for Alloying Additions to the c Matrix (dr/dC1/2
i )

Al Cr Co Mo Ti V
1/2
Constant (MPa/pct ) 225 337 39.4 1015 775 408

By studying the effect of ternary and higher-order

additions to the c nickel matrix, Roth[25] developed
strengthening coefficients dr/dC1/2 i that describe the
contribution of various elements in solution with nickel.
The constants pertinent to strengthening the c matrix in
IN100 are shown in Table II.

C. Strengthening Due to Dislocation Interactions

The c¢ precipitates in IN100 have the L12 (ordered fcc)
crystal structure. When these ordered precipitates are
sheared, ordinary dislocations must travel in pairs so
that order is not destroyed. When a pair of dislocations
shears a c¢ precipitate, there are several mechanisms that
can dictate the interaction, depending upon the size of
the precipitates and the (antiphase boundary) APB
energy (and, thus, the equilibrium spacing of paired
dislocations). For small precipitates (less than approx-
imately 20 nm), the pair of dislocations may not lie
within an individual precipitate (Figure 3(a)). This case
is referred to as weak pair coupling. When the precip-
itates are larger, the trailing dislocation enters a precip-
itate before the leading dislocation exits (Figure 3(b)).
This situation is referred to as strong pair coupling. In
both of these situations, the strengthening increment is
due to the formation of an APB within the precipitate.
Figure 3(c) schematically compares the stress required
to push a pair of dislocations though a precipitate for
each mechanism as a function of precipitate size. The
active mechanism changes from weak pair coupling to
strong pair coupling as the c¢ size increases.
The phenomenon of increasing strength as a function
of temperature is well documented for Ni3Al in its bulk
crystalline form.[26,27] While detailed models continue to
be developed to describe the associated dislocation
mechanisms, the basis of this hardening mechanism
involves thermally activated cross-slip pinning of mobile
dislocations.[28–35] As will be discussed in Section 3, this
cross-slip-induced hardening is apparently active in the
primary c¢ and large secondary c¢ precipitates (>300 nm).
1. Weak pair coupling
According to Brown and Ham,[10] the stress required
to move a dislocation though a random array of
obstacles is
F3=2
s ¼ pffiffiffiffiffiffiffiffiffi ½4
bL 2TL
where s is the shear stress, F is the obstacle strength, b is
the Burgers vector, L is the particle spacing, and TL is
the dislocation line tension.
Equation [4] assumes that precipitates are infinitely
small. To account for the finite size of precipitates, the
center-to-center particle spacing L was replaced with the
Fig. 3—(a) Configuration of dislocations and precipitates during
weak pair coupling, (b) configuration of dislocations and precipitates
during strong pair coupling, and (c) relative strength due to each
mechanism vs precipitate size. Duplicated from Huther et al.[12]
spacing between particles Ls. For infinitely small spher-
ical particles, Ardell[36] proposed that
sffiffiffiffiffiffiffi
8
L¼ ds ½5a
3pf
where f is the volume fraction and ds is the mean pla-
nar diameter. To account for the finite size of the pre-
cipitates, the average planar diameter was subtracted
from L, resulting in the spacing between precipitates:
sffiffiffiffiffiffiffi
8
Ls ¼ ds
ds ½5b
3pf
When dislocations shear ordered, coherent particles,
an APB forms on the slip plane. By multiplying the APB
energy cAPB by the length of the precipitate-opposing
dislocation motion, the mean planar diameter ds, F can
be replaced by cAPBds. Substituting this expression into
Eq. [4] gives
3=2
cAPB 3=2 ds
s¼ pffiffiffiffiffiffiffiffiffi ½6
bLs 2TL
Equation [6] represents the shear stress required to
push the leading dislocation into a precipitate. While the
leading dislocation is creating disorder within the parti-
cle, the trailing dislocation is restoring order and, as a
result, the stress required to move the second dislocation
through a precipitate is not equal to Eq. [6]. To find the

1592—VOLUME 40A, JULY 2009 METALLURGICAL AND MATERIALS TRANSACTIONS A

stress on the second dislocation, an interaction force sinter
must be accounted for, due to the elastic repulsion
between the dislocations. The following set of equations
describes the force balance on each dislocation:
leading dislocation: sbL1 þ sinter bL1
cAPB d1 ¼ 0
½7
trailing dislocation: sbL2
sinter bL2 þ cAPB d2 ¼ 0
where s is the shear stress required to move the dislo-
cation pair and d is the mean length of dislocation ly-
ing inside a precipitate. The subscripts 1 and 2 refer to
the leading and trailing dislocation, respectively. The
interaction force can be eliminated from Eq. [7] by
solving the two equations simultaneously, giving
 
c d1 d2
s ¼ APB
½8
2b L1 L2
In order to comply with Eq. [6],
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
d1 cAPB ds ds
¼
L1 2TL Ls
The second dislocation passes through the material
while remaining essentially straight (Figure 3(a)).
Because of this, L2 is equal to L (Eq. [5a]) and d2 is
the average length of a random line intersecting a
particle of diameter ds, which is pd2s /4L2. Substituting
the expressions for d1/L1 and d2/L2 into Eq. [8] produces
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  !
cAPB cAPB ds ds p ds 2
s¼
½9
2b 2TL Ls 4 L
Equation [9] is the weak pair-coupling equation used
in the present model; it is essentially the same equation
as that presented in Gleiter and Hornbogen[8] and
Brown and Ham,[10] except for the use of interparticle
spacing Ls. Equation [9] indicates that weak pair
coupling is approximately a function of d1/2 s and c3/2
APB.
Weak pair coupling is the active deformation mecha-
nism for precipitates with an average planar diameter less
than approximately 10 to 20 nm, depending on the APB
energy. Because of this, weak pair coupling is generally
only operative when tertiary c¢ precipitates are present.
Figure 4 shows a pair of dislocations gliding between
two larger secondary c¢ precipitates at 1 pct plastic
deformation at 621 C. The spacing between leading
and trailing dislocations is 20 to 30 nm, as indicated by
the white arrows. The inserted TEM dark-field image
shows tertiary c¢ particles between secondary particles.
Because tertiary particles are approximately 10 nm in
diameter, there should be some tertiary particles
between the leading and trailing dislocations.
2. Strong pair coupling
When ordered precipitates reach a critical size, their
resistance to shearing begins to decrease. Huther and
Reppich[12] conducted a theoretical analysis of this
phenomenon. They found that, when a precipitate is
large enough, both dislocations can be within the same
METALLURGICAL AND MATERIALS TRANSACTIONS A
Fig. 4—TEM bright-field image of subsolvus IN100 after 1 pct
plastic deformation at 621 C, showing pairs of dislocations between
secondary c¢ precipitates. Inset is the TEM dark-field image of
subsolvus IN100 before deformation, showing tertiary c¢ particles
between secondary c¢ precipitates.
precipitate; the maximum stress required to push the
dislocations through the precipitate occurs just as the
second partial dislocation reaches the precipitate. A
detailed derivation of the strong pair coupling equation
is presented in Reference 12 and will not be repeated
here. The analysis by Huther and Reppich resulted in
the following expression for the strengthening increment
due to shearing in the strong pair mode:
   1=2
1 Gb 1=2 pds cAPB
Ds ¼ f 0:72w
1 ½10
2 ds wGb2
where G is the shear modulus, ds is the mean planar
diameter, and w is a parameter reflecting the elastic
interaction between paired dislocations.
Several modifications to the model of Eq. [10] were
incorporated, to capture the strengthening mechanisms
in the more complex alloys of this study, as follows.
(a) In the model of Huther and Reppich, the yield
strength was attributed to a single deformation
mechanism. Because of this, the empirical parameter
w was apparently required to account for the effects
of other strengthening mechanisms, such as solid-
solution strengthening. Because the goal of this pres-
ent study was to create a model without any fitting
parameters, w was removed from their equation.
(b) Huther and Reppich assumed that all dislocations
were of a screw character and used a constant line
tension in their derivation, with TL = Gb2/2 incor-
porated into Eq. [10]. In the present study, it was
assumed that dislocations were 60 deg, mixed in
character. Accounting for the character of the
dislocation:[10]
   
Gb2 1
mcos2 h R
TL ¼ ln ½11
4p 1
m ro
where m is the Poisson’s ratio; h is the angle between
the dislocation line and its Burgers vector; R is the
outer cutoff distance, which is approximately the
VOLUME 40A, JULY 2009—1593
 distance to the closest parallel dislocation of the
opposite sign; and ro is the inner cutoff radius, taken
to be b. According to Brown and Ham,[10] R is
approximately equal to the interparticle spacing.
Due to the high volume fraction of precipitates in
the alloys being studied, the interparticle spacing is
generally between 5 and 50 nm. As the precipitates
become larger, a substructure in the adjacent matrix
develops (Figure 4). To account for the combined
effect of precipitate spacing and the presence of
dislocation networks in the matrix, R is taken to be
a constant and is assigned a value of 10 nm.
(c) In the model of Huther and Reppich, the particle
spacing L was taken to be (p/f)1/2 ds/2. As previously
discussed, the present study has adopted Eq. [5a] to
describe the center-to-center precipitate spacing. In
the case of strong pair coupling, interparticle spac-
ing is not described by Eq. [5b] because, at the criti-
cal stress, the length of the dislocation within a
precipitate is not equal to ds. Huther and Reppich[12]
have defined d(x) to be the length of the dislocation
within a precipitate at the critical stress, where
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4T dkAPB p
dðxÞ ¼
1 ½12
kAPB p 2T
Substituting d(x) for ds, Eq. [5b] become r
sffiffiffiffiffiffiffi
8 dislocations into smaller precipitates along with the
Lx ¼ ds
dðxÞ ½13
3pf
Incorporating these changes, the form of the strong
pair coupling equation used in the model is
 1=2
2TL pds cAPB
Ds ¼
1 ½14
pbLx 2TL
Evaluation of Eq. [14] established that strong pair
coupling is approximately a function of d
1/2
s and c1/2
APB,
for the alloys of interest.
Strong pair coupling is the active deformation mech-
anism for c¢ precipitates with an average planar diameter
greater than approximately 10 to 20 nm, again depend-
ing again on the APB energy. This means that strong
pair coupling is generally the active shearing mechanism
for secondary c¢ precipitates. Figure 4 shows pairs of
a=2h110i dislocations in the matrix region between
secondary c¢ precipitates, labeled ‘‘A,’’ indicating that
dislocations must move in pairs to facilitate the shearing
of ordered secondary c¢ precipitates. Finally, it is worth
noting that the effects of plastic constraint[37] were not
considered, due to the fact that the precipitates are
spherical and may eventually be sheared.
3. Cross-slip-induced hardening
Similar to single-phase Ni3Al, many two-phase
superalloys exhibit a peak in flow stress in the temperature
range 500 C to 750 C.[38–43] This suggests that cross-slip
pinning processes operate within the larger c¢ precipitates
1594—VOLUME 40A, JULY 2009
in a manner similar to bulk Ni3Al. Considering that the c¢
precipitates are initially free of dislocations, the operation
of such hardening processes would result in the storage of
dislocations within the Ni3Al. If these processes do not
operate, it is expected that the dislocation pairs would
enter the precipitate, glide uninhibited across the precip-
itate, and exit. Figure 5 shows dislocations accumulating
in a primary c¢ particle after 1 pct strain at 621 C. This
and other similar observations suggest that thermally
activated cross-slip pinning is an additional strengthening
mechanism active in large c¢ precipitates, including
primary and large secondary c¢ precipitates.
It is a challenge to account for cross-slip-induced
hardening in secondary c¢ precipitates, which range in
size from 100 to 900 nm. To study this in more detail,
two unimodal microstructures with mean precipitate
diameters of 180 and 564 nm were established by the
supersolvus heat treatment of unmodified IN100.
Figure 6 shows TEM images of these samples after
1 pct total deformation at 621 C. In the sample with
smaller precipitates, pairs of dislocations are observed
mainly in the c matrix (Figure 6(a)). On the other hand,
in the sample with larger precipitates, the accumula-
tion of dislocations within precipitates is observed
(Figure 6(b)). The fact that the majority of the disloca-
tions observed in Figure 6(a) exist as pairs located in the
matrix indicates the shearing of precipitates without
dislocation trapping by thermally activated cross-slip
pinning. Considering the high stresses required to push
smaller area swept out by the dislocations as they glide
through the particle, the opportunity for a thermally
activated cross-slip pinning event would diminish as the
precipitate size decreases.
Quantitative analyses of the incidence of dislocation
accumulation in the precipitates as a function of
precipitate size were conducted on a large number of
micrographs. A similar volume of precipitate material
was examined at each precipitate size. Figures 7(a) and
(c) show the distribution of the precipitate size, mea-
sured as a mean value of the precipitate widths along
two h100i directions. The size ranges measured were 70
to 300 nm and 250 to 900 nm, respectively. The number
of dislocation segments per particle was counted and
plotted against precipitate size, as shown in Figures 7(b)
and (d). In the sample with small precipitates, the
value is less than 1, regardless of the precipitate size
(Figure 7(b)). However, in the sample with large pre-
cipitates, the value is approximately 2 pairs of disloca-
tions in precipitates smaller than 550 nm, and more than
twice that when the size exceeds 700 nm (Figure 7(d)).
This result indicates that thermally activated cross-slip
pinning is activated much more frequently in precipi-
tates larger than 300 nm and that this is likely to
contribute to the overall strength of the material.
In addition to investigating the incidence of disloca-
tion accumulation in the samples with a unimodal
microstructure with a mean precipitate size of 564 nm,
six samples were tensile tested at temperatures ranging
from room temperature to 760 C, to determine the
temperature dependence of the flow stress at different
levels of offset strain. The flow stress at different levels of
METALLURGICAL AND MATERIALS TRANSACTIONS A
Fig. 5—TEM bright-field image of subsolvus IN100 after 1 pct plastic deformation at 621 C, showing dislocations accumulating in primary c¢.
microstrain was found to increase to a greater degree at
temperatures above 400 C, compared to lower temper-
atures; this indicates that hardening due to thermally
activated cross-slip pinning processes is likely to be
operative (Figure 8).[37]
As mentioned previously, there have been a number
of studies in Ni3Al that attempt to relate detailed
dislocation-hardening phenomena to macroscopic flow
behavior.[28–35] However, there is not yet a comprehen-
sive theory for hardening in these materials. Because
there is an abundance of experimental data pertaining to
cross-slip-induced hardening as a function of tempera-
ture, the strength derived from this mechanism was
quantified with the use of this large experimental data
set reported in the literature. The relationship between
the cross-slip-induced hardening of pure Ni3Al and
temperature was adapted from Mishima,[44] Flinn,[45]
and Davies[46] (Figure 9). These three authors experi-
mentally determined the strength of pure polycrystalline
Ni3Al at temperatures ranging from 73 to 1273 K. The
results of these studies were averaged and were used as
the strength of pure Ni3Al (Figure 9).
Mishima[44,47] and others[43,48–50] also found experi-
mentally that the change in strength due to an alloying
addition to Ni3Al is linearly proportional to the atomic
fraction of that alloying element. Figure 10 shows the
data of Mishima reproduced in Reference 51, demon-
strating how additions of 4 and 8 pct Ti raise the flow
stress in proportion to the titanium content. From a
series of such experiments, Mishima determined con-
stants for the strength increment provided by a wide
range of the alloying additions to Ni3Al. Using the
METALLURGICAL AND MATERIALS TRANSACTIONS A
approach of Mishima, the cross-slip-induced hardening
constants for individual alloying elements in the super-
alloys were assessed and are summarized in Table III.
The cross-slip-induced hardening equation in the
model is " #
X  dr 
Dr ¼ f rðTÞNi3 Al þ Ci ½15
i
dCi
where f is the fraction of precipitates subject to cross-
slip-induced hardening, rðTÞNi3 Al is the strength of pure
Ni3Al as a function of temperature, and Ci is the
concentration of the ith alloying element. In the model,
cross-slip-induced hardening occurs in primary c¢ and
secondary c¢ precipitates larger than 300 nm. An APB
energy of 200 mJ/m2, an intermediate value among
those reported in the literature,[52–55] was used in the
majority of the analyses. As will be discussed later, this
material property, while not well known, has a signif-
icant effect on strength.
4. Bowing
Orowan bowing occurs when the stress required for a
dislocation to bow between precipitates is less than the
stress required for the dislocation to penetrate precip-
itates. The stress above which bowing will occur was
determined by Orowan and is[10]
Gb
s¼ ½16
L
Equation [16] considers infinitely small precipitates. To
account for the finite size of precipitates, the center-to-
VOLUME 40A, JULY 2009—1595
 Fig. 8—Temperature dependence of the flow stress for unimodal
IN100 with 564-nm c¢ precipitates measured at varying levels of mi-
crostrain. The larger increase in flow stress at high temperatures
after larger amounts of microstrain indicate cross-slip pinning is
occurring.

Fig. 6—TEM bright-field images of IN100 with a unimodal micro-

structure after 1 pct total deformation at 621 C. Mean precipitate
size is (a) 180 and (b) 564 nm.

Fig. 9—Experimental data showing the relationship between cross-

slip-induced hardening of binary Ni3Al and test temperature. Data
of Mishima,[44] Flinn,[45] Davies,[46] and the composite curve are used
in the model.

Fig. 7—Histograms of (a) and (c) c¢ precipitate size distribution and

(b) and (d) number of dislocation segments per precipitate in IN100
with a unimodal microstructure after 1 pct total deformation at
621 C. Mean precipitate size is (a) and (b) 180 and (c) and (d) Fig. 10—Experimental data of Mishima[51] demonstrating the linear
564 nm. dependence of the strength of c¢ on the alloying concentration.

1596—VOLUME 40A, JULY 2009 METALLURGICAL AND MATERIALS TRANSACTIONS A

 Table III. Cross-Slip-Induced Hardening Constants
for Alloying Additions to c¢ (dr/dCi)
Al Cr Co Mo Ti V
Constant (MPa/pct) 0 7 0 16.8 25 8.2
center spacing was replaced with the interparticle spac-
ing:
Gb
s¼ ½17
Ls
The stress required for dislocations to bow between
precipitates is inversely proportional to the interparticle
spacing and consequently, bowing typically occurs
between widely spaced precipitates.
IV. MODELING CONSIDERATIONS
To assemble a comprehensive model, several other
factors had to be taken into account. The weak and
strong pair-coupling equations overlook the fact that
precipitates have a distribution of sizes; this may result
in significant error if the precipitates are in the size range
near a mechanism transition. These equations also
disregard the characteristic temperature dependence of
the strengthening phenomena, decreasing their accuracy
at the temperatures most critical to industrial applica-
tions. Additionally, the superimposition of the strength-
ening due to each mechanism must be considered.
A. Precipitate Size Distributions
The nucleation, growth, and coarsening of precipi-
tates typically results in a characteristic precipitate
distribution.[56–59] Parametric studies have revealed that
an error in the strengthening increment associated with
neglecting the size distribution occurs when the average
precipitate size is close to the transition between weak
pair coupling and strong pair coupling. Therefore, a
Lifshitz, Slyozov, and Wagner (LSW) size distribu-
tion[60] was applied to each of the individual populations
of c¢ precipitates such that the predicted strengthening
increment is the weighted average of the strength due to
all likely precipitate sizes within the distribution.
In Figure 11, the yield strength increment due to
strong and weak pair coupling as a function of the mean
planar diameter is plotted for a c¢ population with a
volume fraction of 0.06. For this population, the
transition between weak and strong pair coupling is at
13.25 nm. The bold curve in Figure 11 shows the yield
strength increment due to this population, when an
LSW size distribution is taken into account. In this case,
the omission of a size distribution results in an overes-
timation of the yield strength of approximately 12 MPa.
B. Temperature Dependence of Pair Coupling
The equations prescribed by Brown and Ham and
Huther and Reppich for pair coupling did not include a
METALLURGICAL AND MATERIALS TRANSACTIONS A
Fig. 11—Comparison of the yield strength attributable to a popula-
tion of tertiary c¢ with a volume fraction of 0.06 as a function of the
average diameter. Without using the LSW size distribution, the pair-
coupling equations would overestimate the yield strength by up to
12 MPa. The LSW distribution shown is for a population with an
average diameter of 13.25 nm.
dependence on temperature. However, it might be
expected that the penetration of dislocations into the
precipitates is thermally assisted. The kinetics of ther-
mally activated deformation have been considered in
detail by Kocks, Argon, and Ashby[61] and the relevant
functional form for temperature dependent changes in
flow stress is

@lnr 
Q
¼ ½18
@lnTe_ mkT
where Q is the activation energy, m is the strain-rate
sensitivity, k is the Boltzmann’s constant, e_ is the strain
rate, and T is the temperature.
Due to the complexity of the deformation modes, the
activation energy and strain-rate sensitivity are not
easily derived from theoretical considerations only.
Therefore, strain-rate change experiments were used to
determine the values of these parameters for subsolvus
IN100. According to Reference 61:

@ln_e 
m¼ ½19
@lnrT
Using Eq. [19], m was determined by measuring the
change in flow stress associated with a change in the
strain rate at temperatures ranging from room temper-
ature to 1033 K, as described in Section II–D. The
results of these tests are summarized in Table IV. After
the strain-rate sensitivity was determined at these five
test temperatures, an optimization routine using iSight
software (SIMULIA, Providence, RI) was used to
establish a functional form for strain-rate sensitivity as
a function of temperature.
The activation energy was taken to be a constant, but
a discontinuity was found at approximately 400 C.
At this temperature, Q was found to change from
156 kJ/mol.K below 400 C to 380 kJ/mol.K above
400 C. It is interesting that 400 C corresponds to the
minimum in strength for IN100, and is consistent with
VOLUME 40A, JULY 2009—1597
Table IV. Experimentally Determined Strain-Rate Sensitivity Table V. Values of k from the Literature
and Yield Strength vs Temperature
References Material k
Temperature (K) m Yield Strength (MPa)
61 theoretical calculations 1
295 360.5 1094 66 SS + SiO2 strengthened Cu 1
533 — 1056 67 SS + Al2O3 strengthened Cu-Zn 1.5
894 146.7 1051 62 theoretical calculations 1.5
922 92.3 1057 62 SS + SiO2 strengthened Cu 1.76
977 51.2 1025 18 Ni superalloys with ordered 1.13 to 1.19
1017 22.4 986 coherent c¢

Fig. 13—Experimental and predicted yield strength curves for sub-

Fig. 12—Experimentally determined temperature dependence on pair solvus IN100, assuming different values of k.
coupling.

the onset of thermally activated cross-slip pinning

processes. In Figure 12, the remaining fraction of the
room-temperature yield strength predicted by the pair-
coupling equations is plotted as a function of temper-
ature. This temperature dependence was incorporated
into the yield strength model by multiplying Eqs. [9] and
[14] by the function plotted in Figure 12. In this sense,
the temperature dependence is considered to arise from
the thermally assisted penetration of dislocations into
the precipitates.
C. Superposition
Once the increment of strength due to each mecha-
nism was assessed, it was necessary to consider how
these strengthening increments should be superimposed.
There are several possible methods of superposition,
most of which assume the general form:
X rys ¼ grain size hardening of c matrix (Eq:½2Þ
rktotal ¼ Drki ½20
i
where k can be taken as 1 for linear superposition or 2
for Pythagorean superposition, or any value in between.
There are many examples from the literature in which
k has been assigned various values between 1 and 2.
Some values of k from the literature were summarized in
Reference 62 and have been included in Table V. The
study of Nembach et al.[18] was conducted on nickel-base
superalloys similar to Nimonic 105 that contained a
unimodal distribution of ordered coherent c¢ precipitates
with volume fractions ranging from 0.07 to 0.47. They
found that k was 1.13 to 1.19 for underaged and peak-
aged c¢ and slightly higher for overaged c¢.
In the model, taking k to be significantly different
from 1 resulted in considerable underestimation of the
yield strength. Additionally, varying k essentially acted
as a fitting parameter that could be used to alter the
magnitude of the yield strength vs the temperature
profile. These findings are illustrated in Figure 13, in
which experimental yield strength curves are plotted for
subsolvus IN100 along with predicted yield strength
curves, assuming different values of k. In the absence of
any strong evidence for strong superposition effects, k
was assigned a value of unity. Taking k to be 1, the
contributions to yield strength in the model are then
defined as
þ solid-solution hardening of c matrix (Eq:½3Þ
þ grain size hardening of primary c0 (Eq: [2b])
þ anomalous hardening of primary c0 and
large secondary c0 (Eq:½15Þ
þ (shearing/bowing of secondary c0 -strong pair
coupling (Eqs: [14] and [17])
þ shearing/bowing of tertiary c0 -weak pair
coupling (Eqs: [9] and [17]) fðTÞ ½21

1598—VOLUME 40A, JULY 2009 METALLURGICAL AND MATERIALS TRANSACTIONS A

where Eqs. [9] and [12] are modified to incorporate an
LSW size distribution and f(T) indicates the temperature
dependence of dislocation shearing. Note that Eq. [21]
contains no fitting parameters, although some physical
constants contained in the individual equations have
uncertainty.
A program was written to evaluate Eq. [21] based
upon input from a user or upon direct input from
software used to simulate microstructures. The program
accepts microstructural input in terms of both the size
and volume fraction of each microconstituent and the
chemical composition of the c and c¢ phases. This
program was used to evaluate the model in comparison
to a range of experimental yield strength data.
greater than 300 nm, the shape of the yield-strength-vs-
temperature profile does not agree with the experimental
data.
Figure 16 contains a summary showing the error
between the predicted and experimental yield strength
for all test samples. This graph illustrates good corre-
lation between the predicted and experimental yield
strength for subsolvus IN100. For the supersolvus
IN100, modified subsolvus IN100, and modified chem-
istry alloy, the correlation between the predicted and

V. RESULTS AND DISCUSSION

A. Predicting Yield Strength
A total of approximately 36 tensile tests were con-
ducted on a variety of microstructures and chemistries
over the temperature range 22 C to 760 C. Supersol-
vus IN100, subsolvus IN100, and two different heat
treatments of the chemically modified supersolvus alloy
(Table I) were analyzed and evaluated against the
model. The heat treatments and tensile testing were
performed at Pratt & Whitney (East Hartford, CT) and
Fig. 14—Comparison of predicted and experimental yield strengths
the University of Michigan (Ann Arbor, MI). The vs test temperature for subsolvus IN100 and supersolvus IN100.
microstructural information summarized in Table VI
was experimentally characterized by Wusatowska-Sar-
nek et al.[14]
Figure 14 shows a comparison of the experimental
and predicted yield strength vs temperature for the
unmodified subsolvus IN100 and supersolvus IN100.
Figure 14 demonstrates good agreement between pre-
dicted and experimental results for the unmodified
subsolvus IN100. It also displays a good prediction of
the yield-strength-vs-temperature profile for supersolvus
IN100. It is worth noting the lack of primary c¢ in the
supersolvus alloy due to the heat treatment and the
corresponding higher volume fractions of secondary and
tertiary c¢. The model overpredicts the supersolvus yield
strength; the likely reasons for this high degree of error
(~200 MPa) will be discussed shortly.
Figure 15 shows the effect of the cross-slip-induced
hardening of large secondary c¢ particles on the yield Fig. 15—Disregarding the cross-slip-induced hardening of large sec-
strength of supersolvus IN100. Aside from the effect of ondary c¢ precipitates leads to underestimation of yield strength,
cross-slip-induced hardening in secondary c¢ precipitates especially at high temperatures.

Table VI. Microstructural Characterization of Alloys Used to Evaluate the Yield Strength Model

Modified Alloy Modified Alloy

Subsolvus Modified Supersolvus Supersolvus Heat Supersolvus Heat
IN100 Subsolvus IN100 IN100 Treatment 1 Treatment 2

Phase Vf Size (lm) Vf Size (lm) Vf Size (lm) Vf Size (lm) Vf Size (lm)
c matrix 0.40 4.1 0.40 4.90 0.40 34.4 0.47 26.79 0.47 32
Primary c¢ 0.20 1.20 0.25 1.27 — — — — — —
Secondary c¢ 0.34 0.17 0.32 0.15 0.46 0.34 0.42 0.11 0.39 0.22
Tertiary c¢ 0.06 0.008 0.03 0.021 0.14 0.011 0.11 0.007 0.14 0.011

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 40A, JULY 2009—1599

Fig. 16—Comparison of predicted to experimental yield strength for
nine test microstructures at all test temperatures.
experimental yield strength is not as good, with devia-
tions between the predicted and experimental yield
strengths ranging from 50 to 200 MPa. Possible causes
for the error between the experimental and predicted
results include the following.
(a) Material characterization. Slight errors in materials
characterization, particularly characterization of
the volume fraction and average diameter of ter-
tiary c¢ precipitates, can lead to large errors in the
predicted yield strength. The magnitude of error
likely due to characterization will be discussed in
Section IV–D.
(b) Model assumptions. The theories pertaining to
some strengthening mechanisms are not particu-
larly well developed. While an effort was made to
construct the model using the most logical disloca-
tion-precipitate models, in some cases, the models
may not accurately model the actual dislocation
glide characteristics in these complex microstruc-
tures.
With regard to the strengthening due to strong and
weak pair coupling, the magnitude of strengthening
predicted by the model is similar to that predicted by
some recent two-dimensional dislocation glide mod-
els[63,64] but with somewhat different dependencies on
APB energy and precipitate size.
B. Identification of Dominant Mechanisms
In addition to predicting the overall yield strength, it
is also useful to be able to evaluate the relative
contributions of individual mechanisms and, in turn,
determine which constituents of the microstructure are
responsible for the strength of an alloy. Figure 17(a)
shows the magnitude of strengthening due to each
mechanism for four test alloys. While these alloys have
comparable yield strengths, they each derive their
strength from different mechanisms. Recall that the
strength due to cross-slip-induced hardening may come
from both primary and secondary c¢ precipitates and
that the strength due to strong pair coupling may come
from both secondary and tertiary c¢ precipitates.
1600—VOLUME 40A, JULY 2009
Fig. 17—Even among similar alloys, the source of strength of a
material is highly dependent upon microstructure. These figures
depict the magnitude of strengthening sorted by (a) mechanism and
(b) microconstituent. The variation in the sources of the strength of
an alloy illustrates the impossibility of modeling the yield strength of
these materials using a simple model.
Similarly, Figure 17(b) shows that each alloy owes its
strength to different microconstituents. The c matrix
gains its strength from solid-solution strengthening and
grain size hardening. The primary c¢ derives its strength
from cross-slip-induced hardening, order strengthening,
and grain size hardening. The secondary c¢ obtains its
strength from strong pair coupling and cross-slip-
induced hardening; the tertiary c¢ owes its strength to
weak and strong pair coupling. The input parameters
assumed for all four test alloys summarized in Figure 17
were APB energy = 0.20 J/m2, c ky = 750 MPa lm1/2,
and c¢ ky = 500 MPa lm1/2.
This variation in active strengthening mechanisms
prevents the construction of simple models. Conversely,
mapping the mechanisms (microconstituents) responsi-
ble for strength in a manner outlined here could prove
useful in modifying various features of an alloy. It
should be highlighted that the modification of these
microstructural features and the experimental measure-
ment of their effects on strength can be a time-
consuming process.
The model can also be used to highlight the elements
crucial to an accurate determination of yield strength.
Two model inputs found to be especially important to
METALLURGICAL AND MATERIALS TRANSACTIONS A
yield strength prediction were the APB energy and the
volume fraction of tertiary c¢. Figure 18 shows the
predicted yield-strength-vs-temperature profile for sub-
solvus IN100 from Figure 13 in bold. For this curve, the
values of the APB energy and volume fraction tertiary c¢
are 0.20 J/m2 and 0.06, respectively. Also plotted are the
curves that indicate how the yield-strength-vs-tempera-
ture profile would change if either the APB energy or the
volume fraction of tertiary c¢ is altered.
Decreasing the APB energy from 0.20 to 0.15 J/m2
resulted in a reduction in the predicted yield strength of
85 to 100 MPa, depending on the temperature. Increas-
ing the volume fraction of tertiary c¢ from 0.06 to 0.11
produced a boost in the predicted yield strength ranging
from 115 to 130 MPa, depending on the temperature.
These are significant deviations in the predicted yield
strength, considering that these parameters are difficult
to determine either experimentally or theoretically.[14,65]
From this study, it is clear that detailed measurements
of microstructural parameters are crucial to model
development and that the continued development of
rapid characterization techniques is essential.
C. Variation in Predicted Yield Strength
due to Characterization Error
Regardless of the methods used in materials charac-
terization, there is bound to be some degree of error in
the microstructural quantification and chemical compo-
sition of an alloy. Typical experimental errors were
assigned to each microstructural and chemical model
Fig. 18—Slight changes in the APB energy or volume fraction of ter-
tiary c¢ leads to a large error in yield strength prediction. Both of
these elements of the model are difficult to determine with a high de-
gree of accuracy.
Table VII. Expected Errors in Microstructural and Chemical Composition Input*
Property c Matrix
Volume fraction 0.4 ± 2.5 pct
Size (nm) 4100 ± 25 pct
APB energy (J/m2) not applicable
*All alloying additions to nickel allowed ±5 pct.
METALLURGICAL AND MATERIALS TRANSACTIONS A
input; this variability is summarized in Table VII for
subsolvus IN100. The yield strength was then calculated
thousands of times, while each parameter was allowed
to vary independently but randomly within these limits.
The yield strength was found to vary by 247.6 MPa with
a standard deviation of 39.3, based upon a single set of
input microstructural and chemical data. Figure 19
shows the variation in the yield strength increment due
to tertiary c¢, as a function of the volume fraction of
tertiary c¢. The volume fraction of tertiary c¢ was
allowed to vary from 0.04 to 0.08 and the average
diameter of tertiary c¢ was allowed to vary from 6.4 to
9.6 nm. The characterization of tertiary c¢ to even this
level of accuracy is a very difficult process.[65] This plot
indicates that the yield strength increment due to
tertiary c¢ ranges from 105 to 302 MPa. Even for a
constant volume fraction of tertiary c¢, the yield strength
increment due to tertiary c¢ has a range of approxi-
mately 100 MPa.
D. Testing the Predictive Capability of the Model
Investigation of the strengthening mechanisms active
in multimodal nickel-base superalloys has indicated that
the presence of tertiary c¢ precipitates have a significant
impact on yield strength. To test this theory, supersolvus
IN100 with 180-nm secondary c¢ precipitates with and
without tertiary c¢ was tensile tested to 1 pct total strain
at 621 C. The yield strengths with and without tertiary
c¢ were 1092 and 884 MPa, respectively. Figure 20
shows the presence of tertiary c¢ precipitates between
Fig. 19—Variation in the predicted yield strength increment due to
tertiary c¢, owing to errors in microstructural quantifications as a
function of the measured volume fraction of tertiary c¢. Input param-
eters were allowed to vary within the limits outlined in Table VII.
Primary c¢ Secondary c¢ Tertiary c¢
0.2 ± 30 pct 34 ± 10 pct 0.06 ± 33 pct
1200 ± 5 pct 170 ± 10 pct 8 ± 20 pct
0.2 ± 10 pct 0.2 ± 10 pct 0.2 ± 10 pct
VOLUME 40A, JULY 2009—1601

Fig. 20—Microstructure of the sample shown in Figure 6(a) follow-
ing an aging heat treatment at 732 C for 8 hours, during which ter-
tiary c¢ precipitated.
the secondary c¢ following the aging heat treatment. The
volume fraction of tertiary c¢ was estimated from TEM
micrographs; the difference in the yield strength before
and after the precipitation of tertiary c¢ was predicted to
be 170 ± 50 MPa, in good agreement with the experi-
mental results. This verifies the fact that tertiary c¢
precipitates have a significant influence on the yield
strength of multimodal nickel-base superalloys.
VI. CONCLUSIONS
In this study, the contributions to the strength of
polycrystalline multimodal nickel-base superalloys have
been investigated and captured in a multimechanism-
yielding model. The following conclusions can be made.
1. Modeling that consists of the superposition of
possible strengthening mechanisms in a polycrystal-
line nickel superalloy with multiple distributions of
precipitates is useful for optimization of the micro-
structure and heat-treatment approaches.
2. While APB energies and volume fractions of
tertiary c¢ < 20 nm are difficult to measure experi-
mentally, they have a strong influence on the pre-
dicted yield strength.
3. Dislocation storage in precipitates with radii less
than 300 nm is infrequent; therefore, cross-slip-
induced hardening contributes significantly to the
yield strength when secondary c¢ precipitates
become larger than approximately 300 nm in diam-
eter.
4. Modeling indicates that materials subjected to heat
treatments below the c¢ solvus temperature benefit
from a strong Hall–Petch contribution, while super-
solvus heat-treated materials gain the majority of
their strength from their resistance to precipitate
shearing.
1602—VOLUME 40A, JULY 2009
ACKNOWLEDGMENTS
The authors express their gratitude to Mike Long,
Engineous Software (SIMULIA, Province, RI), for his
assistance with iSight software, and to Dennis Dimi-
duk and Triplicane Parthasarathy for their valuable
discussions. The financial support of the DARPA/
AIM project under Grant No. F005484, sponsored by
Pratt & Whitney, is gratefully acknowledged.
REFERENCES
1. C. Ducrocq, A. Lasalmonie, and Y. Honnrat: in Superalloys, D.N.
Duhl, G. Maurer, S. Antolovich, C. Lund, and S. Reichman, eds.,
TMS, Warrendale, PA, 1988, pp. 63–72.
2. R.H. Caless and D.F. Paulonis: in Superalloys, D.N. Duhl, G.
Maurer, S. Antolovich, C. Lund, and S. Reichman, eds., TMS,
Warrendale, PA, 1988, pp. 101–10.
3. H. Hattori, M. Takekawa, D. Furrer, and R.J. Noel: in Superal-
loys, R.D. Kissinger, D.J. Deye, D.L. Anton, A.D. Cetel, M.V.
Nathal, T.M. Pollock, and D.A. Woodford, eds., TMS, Warren-
dale, PA, 1996, pp. 705–12.
4. T.E. Howson, Jr., W.H. Couts, and J.E. Coyne: in Superalloys,
M. Gell, C.S. Kortovich, R.H. Bricknell, W.B. Kent, and J.F.
Radavich, eds., TMS, Warrendale, PA, 1984, pp. 277–86.
5. F.R.N. Nabarro and H.L. deVilliers: The Physics of Creep: Creep
and Creep-Resistant Alloys, Taylor & Francis, London, 1995.
6. J. Dennison, P.D. Holmes, and B. Wilshire: Mater. Sci. Eng.,
1978, vol. 33, pp. 35–47.
7. A.J. Foreman and M.J. Makin: Philos. Mag. A, 1966, vol. 14,
pp. 911–24.
8. H. Gleiter and E. Hornbogen: Mater. Sci. Eng., 1968, vol. 2,
pp. 285–302.
9. P. Feltham: J. Phys. D, 1968, vol. 1, pp. 303–08.
10. L.M. Brown and R.K. Ham: in Strengthening Methods in Crystals,
A. Kelly and R.B. Nicholson, eds., Elsevier Publishing Co., Ltd.,
Essex, England, 1971, pp. 9–135.
11. J.L. Castagne: J. Phys., 1966, vol. 27, pp. 233–39.
12. W. Huther and B. Reppich: Z. Metallkd., 1978, vol. 69, pp. 628–34.
13. E.J. Lee and A.J. Ardell: in Strength of Metals and Alloys, P.
Haasen, V. Gerold, and G. Kostorz, eds., Pergamon Press, Ltd.,
Oxford, England, 1979, pp. 633–38.
14. A.M. Wusatowska-Sarnek, G. Ghosh, G.B. Olson, M.J. Blackburn,
and M. Aindow: J. Mater. Res., 2003, vol. 18, pp. 2653–63.
15. S. Venkadesan, P. Rodriguez, K.A. Padmanabhan, P.V. Sivapra-
sad, and C. Phaniraj: Mater. Sci. Eng., 1992, vol. A154, pp. 69–74.
16. M. Meyers and K. Chawla: Mechanical Behavior of Materials,
Prentice Hall, Upper Saddle River, NJ, 1999.
17. A.W. Thompson: Acta Metall., 1977, vol. 25, pp. 63–66.
18. S. Schilnzer and E. Nembach: Acta Metall., 1992, vol. 40, pp. 803–13.
19. F. Wallow and E. Nembach: Scripta Metall., 1996, vol. 34,
pp. 499–505.
20. L.A. Gypen and A. Deruyttere: J. Mater. Sci., 1977, vol. 12,
pp. 1028–33.
21. R.L. Fleischer: Acta Metall., 1963, vol. 11, pp. 203–09.
22. N.F. Mott and F.R.N. Nabarro: in 1947 Bristol Conference on
Strength of Solids, Physical Society of London, 1948, p. 1.
23. J. Friedel: Dislocations, Pergamon Press, Oxford, England, 1964.
24. U.F. Kocks: Metall. Trans. A, 1985, vol. 16A, pp. 2109–30.
25. H.A. Roth, C.L. Davis, and R.C. Thomson: Metall. Mater. Trans.
A, 1997, vol. 28A, pp. 1329–35.
26. J.H. Westbrook: Trans. AIME, 1957, vol. 209, pp. 898–904.
27. P.A. Flinn: Trans. AIME, 1960, vol. 218, pp. 145–54.
28. B.H. Kear and H.G.F. Wilsdorf: Trans. TMS-AIME, 1962,
vol. 224, pp. 362–64.
29. O. Veyssiere and G. Saada: in Dislocations in Solids, F.R.N.
Nabarro and M.S. Duesbery, eds., North Holland, Amsterdam,
1996, pp. 253–441.
30. D. Caillard and J.L. Martin: Thermally Activated Mechanisms in
Crystal Plasticity, Pergamon, London, 2003.
31. K.J. Hemker, M.J. Mills, and W.D. Nix: J. Mater. Res., 1992,
vol. 7, pp. 2059–69.
METALLURGICAL AND MATERIALS TRANSACTIONS A
32. B. Viguier, J.L. Martin, and J. Bonneville, in Dislocations in Solids,
F.R.N. Nabarro and M.S. Duesbery, eds., North Holland,
Amsterdam, 2002, p. 459.
33. S.S. Ezz and P.B. Hirsch: Philos. Mag. A, 1994, vol. 69, pp. 105–27.
34. B. Devincre, P. Veyssiere, and G. Saada: Philos. Mag. A, 1999,
vol. 79, pp. 1609–28.
35. V. Paidar, D.P. Pope, and V. Vitek: Acta. Metall., 1984, vol. 32,
pp. 435–48.
36. A.J. Ardell: Metall. Trans. A, 1985, vol. 16A, pp. 2131–66.
37. P.H. Thornton, R.G. Davies, and T.L. Johnston: Metall. Trans.,
1970, vol. 1, pp. 207–18.
38. B.H. Kear and B.J. Piearcey: Trans. TMS-AIME, 1967, vol. 239,
pp. 1209–15.
39. V.K. Sikka and E.A. Loria: in Superalloys, D.N. Duhl, G. Maurer,
S. Antolovich, C. Lund, and S. Reichman, eds., TMS, Warrendale,
PA, 1988, pp. 203–12.
40. C. Monier, C. Bertrand, J.-P. Dallas, M.-F. Trichet, and M.
Cornet: Mater. Sci. Eng., 1994, vol. A188, pp. 133–39.
41. A. Nitz and E. Nembach: Metall. Mater. Trans. A, 1997, vol. 29A,
pp. 799–807.
42. A. Banik and K.A. Green: in Superalloys, T.M. Pollock, R.D.
Kissinger, and R.R. Bowman, eds., 2000, pp. 69–74.
43. M. Dollar and I.M. Bernstein: in Superalloys, D.N. Duhl, G.
Maurer, S. Antolovich, C. Lund, and S. Reichman, eds., TMS,
Warrendale, PA, 1988, pp. 275–84.
44. Y. Mishima, S. Ochiai, N. Hamao, M. Yodogawa, and T. Suzuki:
Trans. Jpn. Inst. Met., 1986, vol. 27, pp. 648–55.
45. P.A. Flinn: Strengthening Mechanisms in Solids, ASM, Metals
Park, OH, 1962, p. 17.
46. R.G. Davies and N.S. Stoloff: Trans. TMS-AIME, 1965, vol. 233,
pp. 714–19.
47. S. Miura, Y. Mishima, and T. Suzuki: J. Jpn. Inst. Met., 1992,
vol. 56, pp. 1214–20.
48. F.E. Haredia and D.P. Pope: High-Temperature Ordered Inter-
metallic Alloys II, Materials Research Society, Pittsburgh, PA,
1987, pp. 213–20.
METALLURGICAL AND MATERIALS TRANSACTIONS A
49. J. Lopez and G.F. Hancock: Phys. Status Solidi A, 1970, vol. 2,
pp. 469–74.
50. K. Aoki and O. Izumi: Phys. Status Solidi A, 1976, vol. 38,
pp. 587–94.
51. C.T. Liu and D.P. Pope: in Intermetallic Compounds, J.H.
Westbrook and R.L. Fleischer, eds., John Wiley & Sons,
New York, NY, 1995.
52. R.J. Taunt and B. Ralph: Philos. Mag., 1974, vol. 30, pp. 1379–94.
53. M. Vittori and A. Mignone: Mater. Sci. Eng., 1985, vol. 71,
pp. 29–37.
54. D.M. Dimiduk, A.W. Thompson, and J.C. Williams: Philos. Mag.
A, 1993, vol. 67 (3), pp. 675–98.
55. T. Kruml, B. Viguier, J. Bonneville, P. Spatig, and J.L. Martin:
MRS Symp. Proc., 1997, vol. 460, pp. 529–34.
56. D. McLean: Met. Sci., 1984, vol. 18, pp. 249–56.
57. P.W. Voorhees: J. Stat. Phys., 1985, vol. 38, pp. 231–53.
58. S.C. Hardy and P.W. Voorhees: Metall. Trans. A, 1988, vol. 19A,
pp. 2713–21.
59. T. Eguchi, Y. Tomokiyo, and S. Matsumura: Phase Transitions,
1987, vol. 8, pp. 213–26.
60. I.M. Lifshitz and V.V. Slyozov: J. Phys. Chem. Solids., 1961,
vol. 19, pp. 35–50.
61. U.F. Kocks, A.S. Argon, and M.F. Ashby: Thermodynamics and
Kinetics of Slip, Pergamon Press, New York, NY, 1975.
62. U. Lagerpusch, V. Mohles, and E. Nembach: Mater. Sci. Eng.,
2001, vols. A319–A321, pp. 176–78.
63. T.A. Parthasarathy, S.I. Rao, and D.M. Dimiduk: Superalloys,
TMS, Warrendale, PA, 2004, pp. 887–96.
64. S.I. Rao, T.A. Parthasarathy, D.M. Dimiduk, and P.M. Hazzle-
dine: Philos. Mag. A, 2006, vol. 86 (4), pp. 215–25.
65. P. Sarosi, G. Viswanathan, D. Whitis, and M. Mills: Superalloys,
TMS, Warrendale, PA, 2004, pp. 989–96.
66. R. Ebeling and M.F. Ashby: Philos. Mag., 1966, vol. 13, pp. 805–
34.
67. P.B. Hirsch and F.J. Humphreys: Proc. R. Soc., 1970, vol. A318,
pp. 45–72.
VOLUME 40A, JULY 2009—1603