High-Temperature Material Lattice Combining Low Thermal

Expansion, High Stiffness and Strength

by



Pamela L. Fetchko






A dissertation proposal submitted in partial fulfillment of the requirements for the
degree of Master of Science
(College of Engineering Department of Materials Science and Engineering)
in the University of Michigan
2008














Report Documentation Page
Form Approved
OMB No. 0704-0188
Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington
VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it
does not display a currently valid OMB control number.
1. REPORT DATE
01 FEB 2009
2. REPORT TYPE
N/A
3. DATES COVERED
-
4. TITLE AND SUBTITLE
High-Temperature Material Lattice Combining Low Thermal Expansion,
High Stiffness and Strength
5a. CONTRACT NUMBER
5b. GRANT NUMBER
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) 5d. PROJECT NUMBER
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)
University of Michigan
8. PERFORMING ORGANIZATION
REPORT NUMBER
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES)
AFIT/ENEL BLDG 16 Rm120 Wright-Patterson AFB, OH 45433-7221
10. SPONSOR/MONITORS ACRONYM(S)
11. SPONSOR/MONITORS REPORT
NUMBER(S)
CI09-0040
12. DISTRIBUTION/AVAILABILITY STATEMENT
Approved for public release, distribution unlimited
13. SUPPLEMENTARY NOTES
The original document contains color images.
14. ABSTRACT
15. SUBJECT TERMS
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF
ABSTRACT
UU
18. NUMBER
OF PAGES
44
19a. NAME OF
RESPONSIBLE PERSON
a. REPORT
unclassified
b. ABSTRACT
unclassified
c. THIS PAGE
unclassified
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std Z39-18

2




















The views expressed in this article are those of the author and do not reflect the official
policy or position of the United States Air Force, Department of Defense, or the U.S.
Government.
























3

Abstract

Low or near zero thermal expansion is a desirable property for structures undergoing high
heat fluxes and thermal stresses. A number of lattice concepts have been investigated to solve
this issue. Unfortunately they do not display high stiffness and strength, are too complicated to
allow for ease of fabrication, or are not proven at high enough temperatures from practical use on
extreme temperature environments. A bi-material lattice that combines low or near zero thermal
expansion with high stiffness over a wide temperature range has been introduced and discussed
in previous research. This concept was effectively modeled and tested using an aluminum alloy,
7075-T6, and a titanium alloy, Ti-6Al-4V. The main objective of this research is to prove this
bi-material lattice concept can be applied to higher temperature combinations of material capable
of 1000
o
C while still maintaining high strength and stiffness. Criteria for materials selection are
presented. Three configuration concepts are analyzed via finite element calculations. Results
show that the bi-material lattice can be used with high temperature materials up to 1000
o
C but
some adjustments in topology are required to achieve that temperature in a Ni Nb metallic
lattice.
















4
Table of Contents

List of Figures..5
List of Tables...6
1. Introduction7
2. Design Analysis...10
2.1. Topology Concept10
2.2. Materials Selection...12
2.3. Modeling/Simulation15
2.4. Cell Geometry...18
3. Results..18
3.1. Dove-tail configuration.18
3.2. Keyhole configuration..19
3.3. Rounded dove-tail configuration..21
4. Conclusions..23
5. Future Work.24
6. References....25
7. Figures..27
8. Appendix A..43




























5
List of Figures
Figure 1: Curved rib diagram similar to one used by Lakes..27
Figure 2: Lakes lattice for thermal expansion via lateral bending displacement...27
Figure 3: Sigmund & Torquatos zero expansion lattice created by topology optimization
methods..27
Figure 4: Steeves et al. topology concept and calculations...28
Figure 5: Schematic illustration of the triangular zero-thermal expansion lattice concept...28
Figure 6: Pin-jointed lattice used in the study by Steeves and co-workers...29
Figure 7: Coefficient of thermal expansion vs. temperature comparison of high temperature
materials.29
Figure 8: Specific Strength vs. Temperature comparison among several alloy classes30
Figure 9: Comparison chart of nickel-cobalt alloys showing yield strength vs temperature30
Figure 10: Comparison chart of nickel-cobalt alloys showing coefficient of thermal expansion
vs. temperature...31
Figure 11: Comparison chart of Haynes 188 and Wah-Chang C-103 showing yield strength vs.
temperature31
Figure 12: Comparison chart of Haynes 188 and Wah-Chang C-103 showing coefficient of
thermal expansion vs. temperature32
Figure 13: Binary phase diagram of nickel-niobium interdiffusion within Haynes 188...32
Figure 14: Boundary conditions applied to the lattice model in ABAQUS simulation....33
Figure 15: Stress-strain curves for Haynes 188 at 1000
o
C33
Figure 16: Stress-strain curve for Wah-Chang C-103 at 1000
o
C..34
Figure 17: Dovetail configuration of lattice cell34
Figure 18: Highest stress concentrations in Haynes 188...35
Figure 19: Highest stress concentrations in Wah-Chang -103..35
Figure 20: Mises stresses in dovetail configuration at 200
o
C36
Figure 21: Overall displacement, U, of dovetail configuration at 1000
o
C36
Figure 22: Keyhole configuration of lattice cell37
Figure 23: Mises stresses in keyhole configuration at 300
o
C37
Figure 24: Overall displacement, U. of keyhole configuration at 1000
o
C.38
Figure 25: Rounded dovetail configuration of lattice cell.....38
Figure 26: Mises stresses for rounded dovetail at 1000
o
C.39
Figure 27: Equivalent plastic strain for rounded dovetail at 1000
o
C.39
Figure 28: Displacement of rounded dove configuration at 1000
o
C.40
Figure 29: Stress comparison of Haynes 188 for three configurations..40
Figure 30: Stress comparison of Wah-Chang C-103 for three configurations..41
Figure 31: Phase fraction diagram of niobium-nickel interdiffusion within Haynes 188.42










6
List of Tables
Table 1: Chemical composition percentage of Wah-Chang C-103...13
Table 2: Chemical composition percentage of Haynes 188......14
Table 3. Material properties for the candidate alloys used in the lattice model15
Table 4: Ramberg-Osgood coefficients for Haynes 188 and Wah-Chang C-103.17
Table 5: Highest stress values for dovetail configuration..18
Table 6: Highest stress values for keyhole configuration..20
Table 7: Highest stress values for rounded dovetail configuration22
Table 8: Chemical composition of comparison alloys...43


























7
1. Introduction
Structural systems that experience extreme temperature variations during service in space
applications [1-3] are subject to high thermal stresses that lead to thermo-mechanical fatigue of
components. As the material is heated and cooled cyclically, the expansion and contraction of
the material results in thermo-mechanical stresses that can eventually lead to failure of the
material and thus the structure. Hypersonic vehicles are of particular interest in this research
because the high speeds at which they operate lead to high thermal loads that in turn induce high
stresses. Since future hypersonic vehicles are intended to be used more than once, they will be
continuously subjected to thermal cyclic loading. This constant loading and unloading can lead
to thermo-mechanical fatigue if an inadequate material is used. Therefore, a low thermal
coefficient is necessary to prevent the occurrence of high thermal stresses within the material as
it heats up. However, high stiffness and strength are also required to support bending moments
and in-plane stresses that can occur in these structures during normal service conditions. Such a
combination of material properties is rarely found in a single material. For instance, metals add
weight to the structure, which is a hindrance for aerospace applications, while ceramics are
brittle and thus limit the amount of strain structures can handle before fracture.
Composite materials have long been studied in order to combine two or more materials
with differing mechanical properties that together could achieve a desired thermal property, such
as lower thermal expansion or greater stiffness. A number of researchers [4-6] have investigated
manipulating the coefficient of thermal expansion of both laminate and particle-reinforced
composites to attain negative and positive values that could exceed a single materials
possibilities. However, zero-coefficient of thermal expansion is difficult to attain with these
types of composites.

8
Lakes [7] investigated a two constituent composite in which each constituent had a
different thermal expansion coefficient and empty space was assumed to exist within the
structure. This empty space allowed the structure itself to experience higher thermal expansion
coefficients than either constituent independently. Depending on the placement of each
constituent in the composite, the overall thermal expansion could be either positive or negative.
This concept was demonstrated using a curved rib, as shown in Figure 1. If the higher thermal
expansion constituent is on the convex side an increase in temperature will cause the rib to curve
further and if it was on the concave side it will tend to straighten.
Lakes proved the importance of including void space in composites in order to alter the
thermal expansion of the composite beyond the rule of mixtures. He used a lattice, as shown in
Figure 2, that relied on lateral bending as demonstrated in the curved rib concept. The only
drawback with this configuration is that the material has low stiffness and strength since the
expansion is based on the bending of its constituents.
In the Lakes configuration, the structure would experience either positive or negative
thermal expansion; Sigmund and Torquato [8-9] furthered the concept by considering a net
thermal expansion of zero in the overall structure. They recognized the need for low thermal
expansion materials for structures experiencing large temperature fluctuations and thus also
considered the stiffness of their configurations. They used a numerical topology optimization
method based on previous work by Sigmund [10-11], Guedes and Kikuchi [12], and Bensoe et al
[13] to determine how each constituent and void spaces should be distributed. This same model
was later used by Sigmund [14] to create two-dimensional and three-dimensional composite
exhibiting a higher bulk modulus and a lower shear modulus than previous composite
configurations. Choosing to use a three constituent composite, two phases and a void phase as in

9
Lakes research, their topology for the zero thermal expansion configuration is shown in Figure
3.
Topology optimization is defined by Patel et al [15] as an iterative process that
determines the best arrangement of a limited volume of structural material within a given spatial
domain so as to obtain optimal mechanical performance of the design concept. His paper offers
a concise comparison of three different types of methods. The topology optimization method
used by Sigmund and Torquato optimizes the distribution of phases in a configuration in order to
achieve desirable thermoelastic properties by solving finite-element problems where material
type is periodically changed within the finite elements until the appropriate overall mechanical
properties are achieved. In this case, material type and density was changed within each of the
finite elements in order to change the strain energies so that uniform strain energy could be
achieved. While it was proven to be successful in configuring optimal structures, it was also
shown that there is a tradeoff between the topology of the structure and the overall bulk modulus
of the material. Since the two materials used should have differing thermal expansion but similar
stiffness values, for a net thermal expansion of zero, the bulk modulus was found to be
approximately 7% of the bulk modulus of the constituents. Such a low bulk modulus means that
the configuration itself has a low resistance to compression and thus low overall stiffness under
compression loading conditions. In addition, the structures that were generated were too
complex to allow for ease of manufacturing of near net shape components.
Steeves [16] and co-workers explored a zero or low expansion lattice concept that was
not only capable of high strength and stiffness but also easy to manufacture. Their approach also
takes into account void spaces within a lattice to acquire the desired thermoelastic properties
while still maintaining the requisite mechanical strength of the of material structure. They

10
accomplish this by using a stretch-dominated lattice as opposed to the bending-dominated lattice
used by Lakes. Steeves lattice concept was demonstrated using aluminum for the high thermal
expansion constituent and titanium for the low thermal expansion constituent. This
configuration was successfully tested up to a temperature of 300
o
C. Since the current research
utilizes the design criteria developed by Steeves et al., their concepts will be discussed in detail
in the topology section below. The goal of this research program is to assess the applicability of
Steeves stretch-dominated lattice for structures that experience high-temperatures reaching
1000
o
C that use niobium for the low thermal expansion constituent and a nickel-cobalt alloy for
the high thermal expansion constituent.

2. Design Analysis
2.1 Topology Concept
Steeves et al. [16] developed a stretch-dominated lattice [17] consisting of two
constituents that make up the lattice structure. Constituent 1 has the lower coefficient of thermal
expansion and makes up a continuous lattice composed of skewed polyhedra. The polyhedra
that make up the continuous lattice are skewed by angle u from an equilateral triangle, Figure 4.
Constituent 2, with the higher thermal expansion coefficient, would make up a discontinuous set
of polyhedra that would be contained within the continuous lattice. In addition, the lattice
consisted of two types of nodes, expansion nodes that are allowed to move as the structure is
heated up and lattice nodes which are points where constituent one is connected to other cells
within the lattice. In this case the expansion nodes will be points of zero expansion within the
lattice. Although there are different configurations that could be used to make up the desired
lattice with varying degrees of filling of the interior of the lattice, a triangular unit cell was used

11
in Steeves et al. work and thus will be used in the current research as well. A schematic
diagram illustrating the above concept with constituents and expansion nodes labeled is shown in
Figure 5.
In the present configuration, the thermal expansion of constituent 2 does not affect the
overall expansion of the lattice and thus the net thermal expansion of the lattice is dependent on
constituent 1 as long as constituent 2 expands isotropically. The length of the unit cell, L, and
the skewness of the polyhedra, 0 , within constituent 1 determine the overall geometry of the
lattice. The skewness of the polyhedra, 0 , can be calculated using the coefficient of thermal
expansion of both materials used and the length of the unit cell as shown in Equation 1. This
value 0 can then be used in conjunction with L to determine the lengths of the two constituents
as shown in Equations 2 and 3. Figure 4 illustrates this concept in further detail (note o is the
desired net thermal expansion of the lattice).
o
o
1
=
1
1
2
o
2
o
1
|
\

|
.
|
sin 20 ( )
1
3
+ tan0
|
\

|
.
|
1
1
2
sin 20
( )
1
3
+ tan0
|
\

|
.
|
(1)
l
1
=
L
2
1+ 3tan0
( )
(2)
l
2
=
L
2cos0
(3)
Like the thermal expansion properties, solely constituent 1, the continuous lattice, will
determine the structural behavior of the lattice. Figure 6 shows an example of the pin-jointed
lattice studied by Steeves et al., where a pin allowing each constituent to move freely without
being mutually constraining attached the two materials. The model structure proved that the
expansion of the lattice as a whole was near zero as the calculations suggested. In addition, the

12
pin-jointed lattice was found to be comparable with the Gibiansky-Torquato bounds [18], which
are narrower than the Rosen-Hashin and Schapery bounds [19-20], at low area fractions. This
experiment allowed for the rotation and elongation of the constituents but did not take into
account the rotational resistance that will be experienced when the two constituents are bonded.
The joints between the two constituents are typically made by press fitting them together , which
may result in a metallurgical bond between the two materials at the point of contact with high
temperature excursions. As the lattice is heated constituent 2 will apply stress to the slowly
expanding constituent 1. This will lead to the in-plane bending of the continuous lattice as it
resists the expansion of constituent 2.
Steeves et al. [16] found that the uniaxial stiffness of the lattice when bonded together is
dependent on the bending stiffness of the constituents. When the net thermal expansion was
equal to zero, the stiffness increased as the constituents increased in thickness allowing for a
constant length. This increase in stiffness can actually cause the skewness angle, 0 , for zero-
expansion to deviate from the theoretical calculations and can possibly lead to yielding due to the
applied bending stresses.

2.2 Material Selection
Steeves et al.

[16] demonstrated the afore-mentioned concept by modeling the lattice
using aluminum as constituent 2 and titanium as constituent 1 at temperatures below 300
o
C. The
current research utilizes this design concept in order to apply it to higher temperatures of at least
1000
o
C. The material property that was of primary interest for this research was the coefficient
of thermal expansion. A combination of materials was needed that have varying coefficients of
thermal expansion to combine to form a composite of low thermal expansion. The coefficients

13
of thermal expansion of some materials [21-23] used in high temperature applications are
compared in Figure 7. In this chart, Haynes 188 and Wah-Chang C-103 are also plotted for
comparison. This shows that individually the two have the highest coefficients of thermal
expansion over the temperature range when compared to other high-temperature materials.
Additionally their coefficients of thermal expansion are different enough to achieve an overall
zero expansion lattice.
In order to reach a minimum temperature of 1000
o
C, the material chosen must retain its
mechanical properties at elevated temperatures. The next mechanical property of interest for this
concept is elevated temperature strength. According to the specific-strength vs. temperature
chart shown in Figure 8, nickel and niobium represent the logical choice as metallic candidate
materials for high-temperature applications.
In particular a nickel-cobalt alloy was chosen as the candidate high coefficient of thermal
expansion material while niobium was chosen for the low coefficient of thermal expansion
constituent. In addition to high-temperature strength, both materials should exhibit high
toughness and ductility. Niobium was chosen not only because it retains its strength at high
temperatures but also because it is lightweight compared to other refractory materials and
therefore an ideal candidate for aerospace applications where weight saving is a key design
consideration. For this analysis, the alloy Wah-Chang C-103 (Wah-Chang, Albany, OR, USA)
will be used.
Table 1: Chemical composition percentage of Wah-Chang C-103 [22]
Nb Hf Ti Zr Ta W
87-87.6 10 0.7-1.3 0.7 0.5 0.5

A comparison of several nickel-cobalt alloys, all Haynes International alloys, was made
to determine the candidate material with superior properties for the high-temperature

14
applications. Nickel-cobalt superalloys exhibit high strength and good creep resistance at high
temperatures, and are available in thin sheet form, thus making them an ideal choice for this
research. A comparison of the 0.2% yield strength vs. temperature variation is shown in Figure 9
and a comparison of the coefficient of thermal expansion vs. temperature is shown in Figure 10
for all the alloys that were considered in this work; detailed data is further provided in Appendix
A. Haynes 188 (Haynes International, Kokomo, IN, USA) alloy was chosen due to its ability to
maintain high yield strengths at elevated temperatures as well as a gradual decrease in the yield
strength with increase in temperature. In addition to yield strength, Haynes 188 has the highest
elastic modulus and coefficient of thermal expansion among the several alloys considered here.
Figure 11 and Figure 12 show a comparison of Haynes 188 and Wah-Chang C-103 for both
0.2% yield strength vs. temperature and coefficient of thermal expansion vs. temperature.
Table 2: Chemical composition (wt%) of Haynes 188 [23]
Co Ni Cr W Fe Mn Si C La B
39 22 22 14 3 1.25 0.35 0.10 0.03 0.015

A phase diagram was constructed using Pandat [24] to determine if any brittle phases will
form at the lattice joints during interdiffusion between Haynes 188 and Wah-Chang C-103. As
niobium replaces nickel within Haynes 188 some brittle phases will form, as shown in Figure 13.
These phases are o (which consists mainly of nickel, cobalt, chromium, and tungsten), o (nickel,
niobium, cobalt), and C14 (niobium, cobalt, nickel). At 1000
o
C, all three brittle phases can
form. While interdiffusion will be limited at this temperature, these phases could nevertheless
influence the behavior of the bi-material joints.
The material properties that were used for modeling are listed in Table 3. It is important
to note that the thermal expansion coefficient listed is the average over the entire temperature

15
range of 20
o
C-1000
o
C, to see how the thermal expansion coefficient averages vary over the
temperature range see Figure 12.
Table 3. Material properties for the candidate alloys used in the lattice model
Density

(g/cm
3
)
Coefficient of
Thermal Expansion
T (C) (10
-6
/C)
Elastic Modulus

T (C) (GPa)
Ultimate Tensile
Strength
T (C) (MPa)
0.2% Yield
Strength
T (C) (MPa)
25-100 11.90 20 232 20 985.0 20 470.0
25-200 12.60 100 225 540 775.0 540 275.0
25-300 13.20 200 217 650 755.0 650 265.0
25-400 13.80 300 209 760 650.0 760 270.0
25-500 14.50 400 201 870 450.0 870 250.0
25-600 15.20 500 193 980 265.0 980 185.0
25-700 15.80 600 184
25-800 16.50 700 176
25-900 17.10 800 169
900 161
Haynes
188
8.98
25-1000 17.90
1000 153
1095 145.0 1095 88.0
93 6.84 20 420.6 20 296.5
204 7.02 540 310.3 540 199.9
427 7.20 650 317.2 650 186.2
649 7.38 760 320.6 760 172.4
871 7.56 870 310.3 870 162.0
982 7.74
Wah-
Chang
C-103
8.85
1093 7.92
20-1000 90
1095 186.2 1095 137.9

2.3 Modeling/Simulation
The finite element analysis was accomplished using the ABAQUS program [25]. Using
the formulas discussed in the topology section and the two materials selected above, the ratio
o
2
/o
1
was calculated to be 2.313 and thus the skewness u was determined analytically to be
26.6
o
. Given these two parameters the length of the unit cell, L, as well as the length of the two
constituent sides, l
1
and l
2
, were calculated. It is important to note that the calculations for the
length of constituent 2 do not take into account the width of constituent 1. For this reason, in

16
order to obtain the true value of l
2
, twice the thickness of constituent 1 must be subtracted from
the value obtained above to account for the thickness at both connection points of constituent 2
to constituent 1.
The two constituents will be bound together using a press fit connection that is modeled
by constraining corresponding nodes in the two materials to have the same displacements. In the
stretch-dominated lattice thus designed a two-dimensional problem was modeled. A quadratic
mesh composed of approximately 1300 8-node generalized plane strain elements was used.
Due to the symmetry of the lattice, only half of the cell was modeled, see Figure 14. This
was done for simplification of the boundary conditions applied. The boundary conditions that
were applied consisted of a pinned point at the center of the triangle at 30
o
(half of an equilateral
angle) that was constrained in the x and the y so that there was no displacement at this point.
Since the pinned point is the center of a unit cell, it is thus a point of zero expansion within the
lattice. The boundary condition at the pinned point also eliminates rigid body displacements
within the model. Furthermore, due to symmetry, displacements along the entire left side of the
cell were constrained to zero in the x direction. This constraint does not inhibit the material from
moving in the y-direction along the edge of the model. These boundary conditions do not inhibit
elongation of either constituent in any way and allow for an accurate simulation to be run. A
detailed schematic diagram of the boundary conditions is shown in Figure 14.
The material is expected to reach its yield strength during heating or cooling and thus
plastic deformation and strain hardening is expected to occur. In ABAQUS the strain hardening
was modeled using the Ramberg-Osgood equation (Equation 4 below) where E is the modulus of
elasticity, o is the stress, o
o
is the yield strength, c is the strain, o is the coefficient of thermal
expansion, and n is the Ramberg-Osgood coefficient.

17
Ec =o +o
o
o
o
|
\

|
.
|
n1
o
(4)
The Ramberg-Osgood equation is used to model stress-strain response in materials that
undergo strain hardening past the materials yield point. The Ramberg-Osgood coefficient, n, is a
constant that depends on the material being considered. In order to calculate the value of n for
Haynes 188 and Wah-Chang C-103 it was assumed that in both materials the fracture strength
equaled the ultimate tensile strength. Using the material data sheets provided for both materials
[22-23], the percent elongation at fracture could determine the strain at fracture when using
Equation 4. This constant could then be fit in over the stress range from yield strength to
ultimate tensile strength to determine the strain experienced at a certain stress and thus the stress-
strain curve of the material. The resulting stress-strain curves for Haynes 188 and Wah-Chang
C-103 at 1000
o
C are shown in Figures 15 and 16 respectively. It is also important to note that
Haynes 188 has a higher yield strength but a lower ultimate tensile strength over the temperature
range of interest.
Table 4: Ramberg-Osgood coefficients for Haynes 188 and Wah-Chang C-103
Material Temperature
(C)
Yield
Strength
(MPa)
Ultimate Tensile
Strength
(MPa)
Percent
Elongation
(%)
Ramberg-Osgood
coefficient
20 296.5 420.6 30 23.61
540 199.9 310.3 21 24.44
650 186.2 317.2 17 19.90
760 172.4 320.6 18 17.31
870 162.0 310.3 19 16.70
Wah-Chang
C-103
1095 137.9 186.2 45 39.60
20 470.0 985.0 56 15.99
540 275.0 775.0 69 11.96
650 265.0 755.0 73 11.86
760 270.0 650.0 70 14.04
870 250.0 450.0 77 21.21
980 185.0 265.0 84 35.64
Haynes 188
1095 88.0 145.0 89 27.15

18
2.4 Cell Geometry
The original configuration that was discussed was the dovetail design used by Steeves et
al. in their experiments. For the materials used in this research; however, it posed a problem,
which will be discussed below, as higher temperatures were reached. In order to counteract this
finding two additional configurations were explored, a keyhole configuration and a rounded
dovetail configuration that was modified from the original dovetail. All three configurations and
their results during simulation will be presented in greater detail in the results section following.

3. Results
3.1 Dovetail Configuration
The first configuration to be discussed is the dovetail configuration originally used by
Steeves et al [15], shown in Figure 17. This configuration allowed for a tight fit between the two
materials since constituent 2 expands at a faster rate than constituent 1. This configuration was
tested in 100
o
C increments from 100
o
C up to 1100
o
C. At each increment the highest Mises
stress within each material was recorded and the results are shown in Table 5 below.
Table 5: Highest stress values for dovetail configuration
Temperature (C) Haynes 188 (MPa) Wah-Chang C-103 (MPa)
100 247 123
200 531 266
300 705 353
400 644 429
500 584 389
600 563 282
700 535 401
800 451 376
900 342 342
1000 252 231
1100 173 260


19
Based on the findings above, the ultimate tensile strength of Haynes 188 was not reached
over the entire temperature range; the same did not hold true for Wah-Chang C-103. At a
temperature of 400
o
C the ultimate tensile strength is approximately 340 MPa however the
stresses experienced within the niobium constituent, constituent 1, are 429 MPa (see Figures 18 -
20). The stresses were concentrated around the expansion nodes as well as the lattice joints
between the niobium constituent, constituent 1, and the nickel-cobalt alloy, constituent 2. The
areas experiencing the highest stress concentrations for Haynes 188 and Wah-Chang C-103 are
shown in Figures 19 and 20, respectively. This indicates that at or around 400
o
C the lattice will
fail in the vicinity of one of the sharp corners of the dovetail design due to fracturing within the
material. For this reason the geometry of constituent 1 must be altered to lower stress
concentrations within the niobium.
It is important to note that while the stresses were exceeded, the lattice did maintain near
zero expansion at the expansion nodes up to 1000
o
C (see Figure 21). This shows that the lattice
concept can be applied to any two materials with a sufficiently different CTE in order to obtain a
zero-expansion lattice.

3.2 Keyhole configuration
To recap, the stresses within constituent 1, the niobium, were the limiting factor in the
dovetail configuration discussed above. The areas experiencing the highest stresses were near
the expansion nodes and the lattice joints. In order to alleviate stress concentrations, a rounded
joint, as opposed to one with sharp angles like the dovetail, was applied (see Figure 22). This
joint will be referred to as a keyhole joint. Once again the lattice cell was tested from 100
o
C

20
1100
o
C in 100
o
C increments. The Mises stresses that develop due to expansion are summarized
in Table 6 below.
Table 6: Highest stress values for keyhole configuration
Temperature (C) Haynes 188 (MPa) Wah-Chang C-103 (MPa)
100 59 140
200 130 309
300 213 423
400 280 372
500 319 348
600 405 338
700 466 389
800 499 376
900 451 303
1000 326 326
1100 183 243

The findings for this configuration differed from the dovetail but with the same overall
result. The ultimate tensile strength of Wah-Chang C-103, 355 MPa, was reached at an earlier
temperature of 300
o
C experiencing a stress of 423 MPa (see Figure 23). Once again the highest
stress concentrations were around the expansion nodes and the bi-material lattice joints. The
main difference between the keyhole and the dovetail configuration is that in the case of the
keyhole the ultimate tensile strength of Haynes 188 was also reached. At approximately 900
o
C
the ultimate tensile strength of Haynes 188 is 400 MPa and the stress experienced by Haynes 188
at the lattice joint was 451 MPa. For this configuration, fracture would occur in both materials at
the lattice joint and in Wah-Chang C-103 around the expansion nodes as well.
The difference in the behavior of the Haynes 188 material in the two joint configurations
can be attributed to strain hardening. In the dovetail configuration the greatest stress seen within
the material occurred at 300
o
C while in the keyhole configuration it occurred at 800
o
C. The
material did not undergo plastic deformation until approximately 500
o
C, unlike the dovetail
where it occurred very early on at around 200
o
C, and thus was not able to undergo strain

21
hardening in time to strengthen the material prior to the ultimate tensile strength of the material
decreasing at higher temperatures. From these results, it can be determined that strain hardening
is instrumental in allowing the lattice joint to survive the expansion of the material at higher
temperatures. Once again it is important to note that while the material would undergo fracture,
the expansion was near zero at the expansion nodes (see Figure 24).

3.3 Rounded dovetail configuration
The keyhole configuration proved that for the Haynes 188 constituent the dovetail joint is
preferred as it allows for plastic deformation and thus strain rate hardening to occur at lower
temperatures. The same problem continued to occur in the Wah-Chang C-103 constituent as the
ultimate tensile strength was reached at lower temperatures. For this reason, the lattice geometry
must be modified to reduce stresses within the Wah-Chang C-103 constituent. The expansion
nodes still experienced high stresses regardless of the joint change so a dovetail configuration for
the joint will be used but the sharp corners at the expansion joints will be rounded on the inside
named a rounded dovetail configuration (see Figure 25). The reason for the joint only being
rounded on the inside, as opposed to both sides of the expansion joint, is because while only a
single cell is being modeled here a continuous lattice will be used in practical applications. A
sharp angle should be made between the two legs of constituent 1 at the zero expansion points in
order to mimic the connection joint between the three cells that will eventually make up a
continuous lattice. Once again the lattice cell was tested from 100
o
C1100
o
C in 100
o
C
increments. The stress findings are summarized in Table 7 below.





22
Table 7: Highest stress values for rounded dovetail configuration
Temperature (C) Haynes 188 (MPa) Wah-Chang C-103 (MPa)
100 254 106
200 547 228
300 713 357
400 644 322
500 583 291
600 563 282
700 534 267
800 449 300
900 351 263
1000 263 241
1100 166 222

The findings for this configuration show that the ultimate tensile strength for both
materials was not reached until 1100
o
C indicating that failure will not occur until that
temperature. The contour plot shown in Figure 26 indicates the areas of high stress within each
material at 1000
o
C. These results prove that the rounded dovetail configuration was successful
in inhibiting the ultimate tensile strength from being reached until much higher temperatures
than the other two configurations. This occurs because the dovetail allows the Haynes 188
constituent to strain harden early on, strengthening the material prior to higher stresses being
induced by the temperature increase as was shown earlier in the dovetail configuration. The
Wah-Chang C-103 constituent experiences lower stresses at both joints because of the decreased
stress concentrations at the edges in this configuration. This occurs because the rounded edges
not only lower stress concentrations but allow the material to bend more readily as the Haynes
188 constituent pushes against the Wah-Chang C-103 constituent at the lattice joints. This
effectively causes lower stresses to be experienced in both the lattice joints and the zero-
expansion joints.
The plastic strains experienced by both constituents occurred at the zero-expansion joints
and the lattice joints as expected (see Figure 27). These were the areas of highest stress and thus

23
the ones that would experience plastic deformation. As for the remaining areas of the cell no
plastic deformation occurred. This was expected, as the stresses within those areas remained
small throughout the temperature increase. Once again, as in the other two configurations, the
displacement at the zero-expansion joints is near zero as shown in Figure 28.
Figure 29 and Figure 30 summarize the findings of all three configurations and show the
highest stress values within each material; the highest stress regions within the material are
indicated in the diagrams next to the plots. The highest stresses observed in the C-103 and the
Haynes 188 occurred at the same location throughout the temperature range 20
o
C 1000
o
C for
all three configurations investigated. In the Haynes 188 peak stresses occurred in the dovetail
connection. The same held true for C-103 with the addition of the expansion nodes as well.

4. Conclusions
A bi-material lattice was modeled using ABAQUS that exhibited low thermal expansion
as well as high stiffness and strength. The lattice is stretch-dominated and thus the expansion is
allowed through the stretching of the constituents rather than bending. This facilitated the high
stiffness of the materials to be maintained.
Using a rounded dovetail configuration the stresses within both materials could be
maintained below the ultimate tensile strength. This was due to strain hardening within the
Haynes 188 constituent and a facilitated bending of the Wah-Chang C-103 constituent. The
displacement of the expansion nodes was proven to be minimal and essentially zero for
calculation purposes.

24
In conclusion, the bi-material lattice design composed of nickel and niobium should be
suitable for thermal excursions up to 1000
o
C, in terms of mechanical integrity and its requisite
thermal and mechanical properties.

5. Future Work
Currently more simulations are underway to determine the effects of temperature cycling
on the lattice model by considering the creep characteristics of the materials used. A single cell
is currently being fabricated to compare and verify the model simulations performed using
ABAQUS. This cell will have a topology that allows for a net thermal expansion of 4 10
-6
/C,
thus enabling the measurement of the displacements. Subsequent to the successful
demonstration of the lattice model by way of the single cell test a full panel will be fabricated
and tested for survivability up to 1000
o
C. In addition, a phase fraction diagram was constructed
based on the interdiffusion of niobium with nickel within the Haynes 188 alloy, Figure 31, that
showed the percent fraction of brittle phases formed within the alloy actually decrease as
temperature increases. It would be of interest to further look into configuring another cell
composed of Haynes 188 and Wah-Chang C-103 that would be capable of temperatures of
1200
o
C and higher to take advantage of this finding. Also of interest would be the design of low
expansion structures that take advantage of the low densities and higher temperature capabilities
of C/C and C/SiC composites.





25
6. References
1 Blankenship, C.P. and Yu, J.C. Structures and Materials Technology for Space
Station. AIAA Paper No. 88-2446
2 Rawal, S. Metal-Matrix Composites for Space Applications. Journal of the Minerals,
Metals and Materials Society 53.4 (2001): 14-17.
3 McManus, H.L. Control of space structure thermal deformation: an overview.
Proceedings of SPIE 1917:1 (1993): 545-554.
4 Kelly, A., Stearn, R.J., and McCartney, L.N. Composite materials of controlled
thermal expansion. Composites Science and Technology 66 (2006): 154-159.
5 Kelly, A., McCartney, L.N., Clegg, W.J., and Stearn, R.J. Controlling thermal
expansion to obtain negative expansivity using laminated composites. Composites
Science and Technology 65 (2005): 47-59.
6 Nadeau, J.C. and Ferrari, M. Effective thermal expansion of heterogeneous materials
with application to low temperature environments. Mechanics of Materials 36 (2004):
201-214.
7 Lakes, R. Cellular solid structures with unbounded thermal expansion. Journal of
Materials Science Letters 15 (1996): 475-477.
8 Sigmund, O. and Torquato, S. Composites with extremal thermal expansion
coefficients. Appl. Phys. Lett. 69.21 (1996): 3202-3205.
9 Sigmund, O. and Torquato, S. Design of Materials with Extreme Thermal Expansion
using a Three-Phase Topology Optimization Method. J. Mech. Phys. Solids 45.6
(1997): 1037-1067.
10 Sigmund, O. Materials with Prescribed Constitutive Parameters: An Inverse
Homogenization Problem. Int. J. Solids Structures 31.17 (1994): 2313-2329.
11 Sigmund, O. Tailoring materials with prescribed elastic properties. Mechanics Of
Materials 20 (1995): 351-368.
12 Guedes, J.M. and Kikuchi, N. Preprocessing and Postprocessing for Materials Based
on the Homogenization Method with Adaptive Finite Element Methods. Computer
Methods in Applied Mechanics and Engineering 83 (1990): 143-198.
13 Bendsoe, M.P., Guedes, J.M., Haber, R.B, Pedersen, P. and Taylor, J.E. An Analytical
Model to Predict Optimal Material Properties in the Context of Optimal Structural
Design. J. Appl. Mech. 61 (1995): 930-937.
14 Sigmund, O. A new class of extremal composites. J. Mech. Phys. Solids 48 (2000):
397-428.
15 Patel, N.M., Tillostson, D., Renaud, J.E, Tovar, A., and Izui, K. Comparitive Study of
Topology Optimization Techniques. AIAA Journal 46.8 (2008): 1963-1975.
16
4
Steeves, C., dos Santos e Lucato, S.L, He, M., Antinucci, E., Huchinson, J>W., and
Evans, A.G. Concepts for structurally robust materials that combine low thermal
expansion with high stiffness. J. Mech. Phys. Solids 55 (2007): 1803-1822.
17 Deshpande, V.S., Ashby, M.F, and Fleck, N.A. Foam Topology Bending versus
Stretching Dominated Architectures. Acta Materialia 49 (2001): 1035-1040.
18 Gibiansky, L.V. and Torquato, S. Thermal Expansion of Isotropic Multiphase
Composites and Polycrystals. J. Mech. Phys. Solids 45.7 (1997): 1223-1252.
19 Rosen, B.W. and Hashin, Z. Effective Thermal Expansion Coefficients and Specific
Heats of Composite Materials. Int. J. Engng. Sci. 8 (1970): 157-173.

26
20 Schapery, R.A. Thermal Expansion Coefficients of Composite Materials Based on
Energy Principles. Journal of Composite Materials 2.3 (1968): 380-404.
21 MatWeb: Material Property Data. www.matweb.com.
22 Wah-Chang. C-103 Data Sheet. Wah-Chang, Albany, OR, USA.
23 Haynes International. Haynes 188 Data Sheet. Haynes International, Kokomo, IN,
USA.
24 CompuTherm LLC. Pandat, 7.0 ed.
25 Simulia. ABAQUS/CAE Users Manual, 6.8 ed.
26 Johnson, S. and Pollock, T. Metallics for Actively Cooled Structures. Presented at
Workshop on Materials and Structures for Hypersonics Flight, 11 July 2007.



















27
7. Figures

Figure 1: Curved rib diagram similar to one used by Lakes

Figure 2: Lakes lattice for thermal expansion via lateral bending displacement [7]


Figure 3: Sigmund & Torquatos zero expansion lattice created by topology optimization methods [8]



28

Figure 4: Steeves et al. topology concept and calculations [16]

Figure 5: Schematic illustration of the triangular zero-thermal expansion lattice concept


29

Figure 6: Pin-jointed lattice used in the study by Steeves and co-workers [16]

Coef f i ci ent of Ther mal Ex pansi on
0
2
4
6
8
10
12
14
16
18
20
100 200 300 400 500 600 700 800 900 1000
Temper at ur e ( C)
Ni Alloy ( Haynes 188)
Nb ( C- 103)
Molybdenum
Tungst en
Tant alum
Goodfellow C/ C composit e

Figure 7: Coefficient of thermal expansion vs. temperature comparison of high temperature materials


30

Figure 8: Specific Strength vs. Temperature comparison among several alloy classes [26]
0. 2% Yield St rengt h
0
100
200
300
400
500
600
700
800
900
1000
0 500 1000 1500
Temperat ure ( C)
Haynes 25
Haynes 188- cold
Haynes 188- hot
Haynes 263
Haynes 282
Haynes 617
Haynes Waspaloy

Figure 9: Comparison chart of nickel-cobalt alloys showing yield strength vs. temperature (note: alloy compositions
available in Appendix A)

31
Coefficient of Thermal Expansion
10
11
12
13
14
15
16
17
18
19
0 200 400 600 800 1000 1200
Temperat ure ( C)
Haynes 25
Haynes 188
Haynes 263
Haynes 282
Haynes 617
Haynes Waspaloy

Figure 10: Comparison chart of nickel-cobalt alloys showing coefficient of thermal expansion vs. temperature
(note: alloy compositions available in Appendix A)

0.2% Yi el d St r engt h
0
50
100
150
200
250
300
350
400
450
500
0 200 400 600 800 1000 1200
Temper at ur e ( C)
Haynes 188
Wah- Chang C- 103

Figure 11: Comparison chart of Haynes 188 and Wah-Chang C-103 showing 0.2% yield strength vs temperature

32
Coef f i ci ent of Ther mal Ex pansi on
6
8
10
12
14
16
18
20
0 200 400 600 800 1000 1200
Temper at ur e ( C)
Haynes 188
Wah- Chang C- 103

Figure 12: Comparison chart of Haynes 188 and Wah-Chang C-103 showing coefficient of thermal expansion vs.
temperature


Figure 13: Binary phase diagram of nickel-niobium interdiffusion within Haynes 188

33


Figure 14: Boundary conditions applied to the lattice model in ABAQUS simulation


Haynes 188
0
50
100
150
200
250
300
0 0. 2 0. 4 0. 6 0. 8 1
St r ai n

Figure 15: Modeled Stress-strain curves for Haynes 188 at 1000
o
C.

34
Wah- Chang C- 103
0
50
100
150
200
250
300
350
0 0. 05 0. 1 0. 15 0. 2 0. 25
St r ai n

Figure 16: Modeled Stress-strain curve for Wah-Chang C-103 at 1000
o
C.

Figure 17: Dovetail configuration of lattice cell

35

Figure 18: Highest stress concentrations in Haynes 188



Figure 19: Highest stress concentrations in Wah-Chang C-103

36

Figure 20: Mises stresses (Pa) in dovetail configuration at 400
o
C


Figure 21: Overall displacement, U, of dovetail configuration at 1000
o
C

37

Figure 22: Keyhole configuration of lattice cell


Figure 23: Mises stresses (Pa) in keyhole configuration at 300
o
C


38

Figure 24: Overall displacement, U, of keyhole configuration at 1000
o
C


Figure 25: Rounded dovetail configuration

39

Figure 26: Mises stresses (Pa) for rounded dovetail at 1000
o
C

Figure 27: Equivalent plastic strain for rounded dovetail at 1000
o
C

40

Figure 28: Displacement of rounded dove configuration at 1000
o
C


Figure 29: Stress comparison of Haynes 188 for three configurations

41

Figure 30: Stress comparison of Wah-Chang C-103 for all three configurations

42

Figure 31: Phase fraction diagram of niobium-nickel interdiffusion within Haynes 188








43
Appendix A: Nickel-Cobalt alloy comparisons
A comparison on six nickel-cobalt alloys, all manufactured by Haynes International, was
done. The chemical composition percentages are given in Table 8 below. The findings indicated
that Haynes 188 was the best alloy for this research as it had excellent thermal expansion,
elasticity, and strength. Due to its excellent properties as well, Haynes 25 could be used in this
experiment; however Haynes 188 is superior.
Table 8: Chemical composition of comparison alloys (wt%)
Ni Co Cr W Mo Ti Al Fe
Haynes 25 10 51 20 5 - - - -
Haynes 188 24 38 23 15 - - - -
Haynes 263 52 20 20 - 6 - - -
Haynes 282 57 10 19.5 - 8.5 2.1 1.5 1.5
Haynes 617 54 12.5 22 - 9 - 1.2 1
Haynes waspalloy 58 13.5 19 - 4.3 3 1.5 2

Elast ic Modulus
100
120
140
160
180
200
220
240
0 200 400 600 800 1000 1200
Temperat ure ( C)
Haynes 25
Haynes 188
Haynes 263
Haynes 282
Haynes 617
Haynes Waspaloy


44
Ult imat e Tensile St rengt h
0
200
400
600
800
1000
1200
1400
0 500 1000
Temperat ure ( C)
Haynes 25
Haynes 188- cold
Haynes 188- hot
Haynes 263
Haynes 282
Haynes 617
Haynes Waspaloy