Graduate heses and Dissertations
Iowa State University Capstones, heses and
Dissertations
2011
Improving the phase stability and oxidation
resistance of β-NiAl
Travis Michael Brammer
Iowa State University
Follow this and additional works at: htps://lib.dr.iastate.edu/etd
Part of the Materials Science and Engineering Commons
Recommended Citation
Brammer, Travis Michael, "Improving the phase stability and oxidation resistance of β-NiAl" (2011). Graduate heses and Dissertations.
10189.
htps://lib.dr.iastate.edu/etd/10189
his hesis is brought to you for free and open access by the Iowa State University Capstones, heses and Dissertations at Iowa State University Digital
Repository. It has been accepted for inclusion in Graduate heses and Dissertations by an authorized administrator of Iowa State University Digital
Repository. For more information, please contact digirep@iastate.edu.
Improving the phase stability and oxidation resistance of β-NiAl
by
Travis Michael Brammer
A thesis submitted to the graduate faculty
in partial fulfillment of the requirements for the degree of
MASTER OF SCIENCE
Major: Materials Science and Engineering
Program of Study Committee:
Mufit Akinc, Co-major Professor
Matthew J. Kramer, Co-major Professor
Iver Anderson
Iowa State University
Ames, Iowa
2011
Copyright © Travis Michael Brammer, 2011. All rights reserved.
ii
TABLE OF CONTENTS
CHAPTER 1: GENERAL INTRODUCTION
Thesis Organization
Motivation
Literature Review
References
1
1
1
2
14
CHAPTER 2: A MULTI-STAGE HEIRARCHICAL APPROACH
TO ALLOY DESIGN
Abstract
Introduction
Experimental Procedure
Results and Discussion
Conclusions
Acknowledgement
References
17
17
18
20
24
31
32
32
CHAPTER 3: EFFECT OF PLATINUM GROUP METAL
SUBSTITUTIONS AND HAFNIUM ADDITION ON OXIDATION
RESISTANCE OF β-NiAl
Abstract
Introduction
Experimental Procedure
Results and Discussion
Conclusions
Acknowledgement
References
34
34
35
36
38
55
55
56
CHAPTER 4: ROLE OF GRAIN SIZE ON THE OXIDATION
RESISTANCE OF β-NiAl
Abstract
Introduction
Experimental Procedure
Results and Discussion
Conclusions
Acknowledgement
References
58
58
59
60
62
75
76
76
CHAPTER 5: EFFECT OF PLATINUM GROUP METALS ON THE
MELTING TEMPERATURE OF β-NiAl
Abstract
Introduction
Experimental Procedure
79
79
80
80
iii
Results and Discussion
Conclusions
Acknowledgement
References
81
86
87
87
CHAPTER 6: GENERAL CONCLUSIONS
88
ACKNOWLEDGEMENTS
90
1
CHAPTER 1:
GENERAL INTRODUCTION
Thesis Organization
This thesis is written in an alternate format. The thesis is composed of a general
introduction, four original manuscripts, and a general conclusion. References
cited within each chapter are located immediately after that section. In addition,
figures and tables are numbered independently within each chapter.
The general introduction focuses on the driving force behind this research, and
gives an overview of previous work done on nickel-based superalloys . Chapter
2 describes the preliminary experiments and how those experiments guided the
rest of the thesis work.
Chapter 3 deals specifically with the oxidation
performance of platinum group metal (PGM) and hafnium modifications to β-NiAl
intermetallic.
Chapter 4 investigates the role of grain size on the oxidation
resistance of NiAl based alloys.
Chapter 5 focuses on the role of melting
temperature on the oxidation resistance of NiAl based alloys.
Chapter 6
summarizes the important results of this study.
Motivation
High temperature alloys are essential to many industries that require a stable
material to perform in harsh oxidative environments. Many of these alloys are
suited for specific applications such as jet engine turbine blades where most
other materials would either melt or oxidize and crumble (1). These alloys must
have a high melting temperature, excellent oxidation resistance, good creep
resistance, and decent fracture toughness to be successfully used in such
environments. The discovery of Ni based superalloys in the 1940s revolutionized
the high temperature alloy industry and there has been continued development of
these alloys since their advent (2). These materials are capable of operating in
2
oxidative environments in the presence of combustion gases, water vapor and at
temperatures around 1050°C. Demands for increased fuel efficiency, however,
has highlighted the need for materials that can be used under similar
atmospheres and at temperatures in excess of 1200°C. The current Ni based
superalloys are restricted to lower temperatures due to the presence of a number
of low melting phases that result in softening of the alloys above 1000°C.
Therefore, recent research has been aimed at exploring and developing newer
alloy systems that can meet the escalating requirements. This thesis comprises a
part of such an effort. The motivation of this work is to develop a novel high
temperature alloy system that shows improved performance at higher
temperatures than the currently employed alloys. The desired alloy should be in
accordance with the requirements established in the National Energy Technology
Laboratory (NETL) FutureGen program having an operating temperature around
1300°C.
Literature Review
High Temperature Alloys
Initially stainless steels were used for high temperature applications such as jet
engines in WWI planes (4), but these alloys were limited by their melting
temperature. Later, scores of alloys have been discovered and developed for use
in high temperature environments. Cobalt, nickel, and iron based superalloys
were developed in the 1940s with a large alloy constituent being either chromium
or aluminum to act as the reservoir for a passivating layer. These alloys have
been improved for high temperature oxidative environments through various
processing innovations. But when the highest temperatures (1200-1350°C) must
be achieved and strength is the consideration, then nickel-based superalloys are
the materials of choice.
Nickel based superalloys can be used to a higher
3
fraction of their melting points than just about any other commercially available
materials (1).
Refractory metal silicides have been produced that form a protective silica layer
when exposed to temperatures up to 1700°C (3). Curre ntly MoSi2 is used as a
heating element in furnaces. Yet these alloys are quite dense and show poor
fracture toughness at ambient temperatures.
As shown in Figure 1, increases in operating temperature for superalloys have
increased over the past seven decades and this improvement relies heavily on
the microstructure of the alloy.
conventional casting.
Wrought alloys were improved through
The creep resistance was then improved through
directional solidification followed by single crystals of the alloys. Decreasing the
number of grains in the alloy significantly reduces the creep tendencies at high
temperature.
Figure 1: Evolution of the high temperature capability of the superalloys (2)
4
Cooling Systems and Coatings
Gas turbine engines typically utilize Ni based superalloys which are composed of
a number of low melting phases with melting points ranging from 1250-1450°C.
They are applied in combustion gas environments where temperatures can
exceed 1480°C for brief times, making them susceptible t o softening and
environmental damage due to oxidation (1, 2, 5-11).
Cooling systems and
coatings are employed to decrease the temperature the superalloy sees during
operation. Figure 2 below shows a schematic of the layers within the top surface
of a current superalloy system. The development timeline for these two systems
is displayed in Figure 3 and shows that cooling systems were first utilized in the
1960s while the thermal barrier coatings (TBCs) emerging in the 1990s.
Figure 2:
system (12)
Schematic representation of a multi-layer thermal barrier coating
5
Figure 3: Evolution of cooling systems and coatings for turbine engines (2)
Current combustor and turbine engine components must be actively cooled with
air from the compressor. An obvious approach to maximize efficiency is to
reduce the use of cooling air. One way to accomplish this is to design more
effective heat transfer geometries. This can be achieved by drilling holes through
the component to bleed in air, which cools the combustion gas path surface (2,
13).
However, more significant benefits can be obtained by the application of thin
thermally insulating layers known as thermal barrier coatings or TBCs (12).
TBCs protect turbine engine components by acting as thermal insulators
between the base metal and the hot gases to which they are exposed. Thermal
barrier coatings based on zirconium oxide with a couple of orders of magnitude
lower TCE than the base metal have been used for nearly three decades to
extend the operational lifetime of aircraft turbines and their components operating
in hostile thermal and corrosive environments (14, 15).
Generally, TBCs are attached to the superalloy via a bond coat layer as shown in
Figure 2. The bond coat is an interlayer between the metal substrate and the
ceramic coating. The bond coat produces its own thermally grown oxide (TGO).
NiAl and its derivatives have been used for more than 30 years as bond coats in
TBCs due to their excellent oxidation and corrosion resistance and their ability to
6
maintain their structural integrity during service. The specific properties of NiAl
that relate to high temperature applications are described next.
Properties of NiAl
The first recorded reference to the compound NiAI appeared in a phase diagram
study published in 1908, where the unusually high melting temperature of this
phase was noted (16). Subsequent research has identified NiAl as an ordered
intermetallic material which possess low density, good oxidation resistance, and
metal-like thermal conductivity. These properties have made NiAI an attractive
material for a wide range of engineering applications, including electronic
metallization in advanced semiconductor heterostructures, high temperature
environmental coatings, surface catalysts, and high-current vacuum circuit
breakers. However, lately there has been a strong interest to understand and
develop alloys based on NiAI for high temperature applications in advanced
propulsion systems and turbine generators. The following sections will give a
brief overview of the general properties of NiAl.
Phase Equilibria and Crystal Structure:
The binary Ni-Al phase diagram is shown in Figure 4 and it reveals that there are
five intermetallic phases in this system: NiAl3, Ni2Al3, NiAl, Ni5Al3, and Ni3Al. NiAl
exhibits a wide range of compositional deviations from stoichiometry that is
accommodated by the substitution of nickel atoms onto aluminum sites in nickelrich compositions and by the formation of vacancies on the nickel sites in
aluminum-rich compositions.
7
Figure 4: Phase diagram of Ni-Al (17)
NiAl has a B2 CsCl type crystal structure which means that the Ni and Al atoms
form interpenetrating simple cubic sublattices which has an overall crystal
structure of body centered cubic. As shown in Figure 5, one type of atom locates
at 0, 0, 0 and the other at 1/2, 1/2, 1/2. The lattice parameter of the stoichiometric
composition at room temperature is 0.2887 nm (18).
Figure 5: Illustration of the B2 crystal structure for NiAl (19)
8
Melting Temperature
Stoichiometric β-NiAl has a melting temperature of 1638°C as shown in Figure 4.
Such a high melting temperature indicates strong interatomic bonding and is
necessary for high temperature applications. Current Ni based superalloys have
phases that start melting at 1250°C and soften above 1 000°C which creates an
upper limit of use for any ultra-high temperature applications. Also a higher
melting temperature would result in a lower homologous temperature at the
surface of the alloy which means the alloy could run at hotter temperatures.
Young’s Modulus
NiAl exhibits a high Young’s modulus over a wide range of temperatures as
shown in Figure 6. This indicates that the intermetallic retains its stiffness even
at high temperatures which ensures minimal flexing and distortion of parts under
load. This high stiffness can be attributed to the partially nonmetallic bonding of
the intermetallic as well as the high melting temperature.
Figure 6: Elastic modulus of NiAl compared with steel and TiAl (3)
9
Creep Behavior
Highly creep-resistant materials are desirable to maintain the aerodynamic shape
of the airfoil and a tight clearance between the blade tip and the seal, which
controls the cycle efficiency of the turbine engine. Typical creep requirements for
a Ni based superalloys include stresses up to 200 MPa at temperatures as high
as 1000°C for over 300 h. NiAl on the other hand do es not have useful creep
strength above 50% of its absolute melting temperature (~680°C) due to the BCC
derivative crystal structure.
However, addition of 1% Hf to the NiAl greatly
improves the creep resistance making it on par with current single crystal Ni
based superalloy creep behaviors (20).
Density
Current nickel-based superalloys have densities of up to 9 g/cm3 (6). In 1985 it
was estimated that superalloys accounted for 50% of the total weight of an
aircraft (1). . Density can also be important in gas turbines where increased
density can result in increased stress on mating components. Improvement in
alloy capability at the expense of density increase is not desirable. Modern cast
nickel-based superalloys tend to have densities in the high end of the density
range. Stoichiometric NiAl has an intrinsic density of 5.9 g/cm3 which is about
two thirds that of current Ni based superalloys (16, 19).
Thermal Conductivity
The thermal conductivity of NiAl was measured to be 76 Wm-1K-1 (20). This is 48 times greater than the Ni based superalloys. The relatively high conductivity
would help to quickly dissipate the heat that the part sees which would result in
lower metal temperature or would require less cooling air.
10
Strength and Fracture Toughness
A turbine airfoil is subjected to many stresses such as centrifugal stress, thermal
stresses from temperature gradients, and vibrational excitation.
A typical Ni
based superalloys require tensile yield strength of approximately 700-1000 MPa
at room temperature and up to 500 MPa at 800°C (16) . Polycrystalline NiAl has
a room temperature yield strength of around 200 MPa and only 100 MPa at
800°C as shown in Figure 7.
Figure 7: Temperature dependence of tensile yield strength of NiAl alloys (20)
For polycrystalline NiAl the fracture toughness has been measured to be about 5
MPa•m1/2 at ambient and 22 MPa•m1/2 at 430°C (21) with the most dramatic
increase in fracture toughness occurring at the ductile to brittle transition
temperature (DBTT). The fracture toughness values are essentially independent
of grain size, stoichiometry, and processing technique as shown in Figure 8.
Fracture of NiAl at room temperature occurs by transgranular cleavage without
any stable crack growth (22).
This unstable crack growth at ambient
temperatures is one of the main reasons that this alloy has not risen to replace
11
the current Ni based superalloys which have a much higher typical room
temperature fracture toughness around 50 MPa•m1/2 (20).
Figure 8: Fracture toughness of NiAl as a function of temperature (19)
The DBTT for stoichiometric NiAl is 300°C which then incr eases the thermally
activated slip systems allowing for improved ductility and fracture toughness.
Although the transition temperature is relatively low, NiAl alloys still have
unfavorable brittle characteristics at ambient temperatures. Figure 9 shows the
tensile elongation as a function of temperature with the first signs of ductility
occurring at the DBTT.
Figure 9: Ductile to brittle transition for NiAl (23)
12
Coefficient of Thermal Expansion
The coefficient of thermal expansion (CTE) is important for structural applications
since thermal stresses depend directly on the magnitude of the CTE. The CTE
for NiAl is very close to Ni based superalloys (24). The stable protective oxide
that forms on NiAl based alloys is Al2O3. Differences in the CTE of the substrate
and the oxide can cause thermal stresses to build up at the interface which
eventually cause the oxide to spall off of the surface. A substrate and oxide with
similar CTE is preferred for applications where many thermal cycles are
necessary.
18
CTE (ppm/°C)
16
14
12
10
8
6
500
700
900
1100
1300
1500
Temperature (°C)
Figure 9: Coefficient of thermal expansion of NiAl and Al2O3 (25)
Oxidation
The Pilling-Bedworth ratio of oxides relates the volume of the oxide formed to the
volume of the metal consumed. If the ratio is less than unity, the oxide will be in
tension and will not be adherent and will spall off. If the ratio is more than unity,
the oxide layer will be in compression and will uniformly cover the metal surface
and be protective. If the ratio is much greater than unity, there is risk of too much
compressive stresses being built up and the oxide might crack and will not be
protective. The Pilling-Bedworth ratio for the alumina scale that the NiAl forms is
13
1.38 (26). This means that the scale formed by NiAl is uniform and protective to
the underlying metal which acts as a diffusion barrier for continued oxidation.
The alumina undergoes several phase transitions as it is heated before reaching
its thermodynamically stable α-Al2O3 forms. The phase transition temperatures
along with the stable temperature ranges for various forms are shown in Figure
10. The α-Al2O3 is the most stable and serves as a protective scale.
Figure 10: Temperature ranges and transition temperatures of the aluminas (27)
14
References
1) H. Okamoto, Al-Ni (aluminum-nickel), Journal of Phase Equilibria and
Diffusion, Vol. 25, No. 4, pp. 394, 2004
2) R.D. Noebe, R.R. Bowman, and M.V. Nathal, Physical and mechanical
properties of the B2 compound NiAI, International Materials Reviews, Vol. 38 No.
4, pp. 193-232, 1993
3) S.M. Meier, D.K. Gupta, K.D. Sheffler, Ceramic Thermal Barrier Coatings for
Commercial Gas Turbine Engines. JOM, Vol. 43, No. 3, pp. 50-53, 1991
4) R.L. Jones, Thermal Barrier Coatings, in Metallurgical and Ceramic Protective
Coatings, Chapman & Hall: London, pp. 194-235, 1996
5) R.C. Reed, Superalloys, Cambridge: Cambridge University Press, 2006
6) M.J. Donatchie Jr., S.J. Donatchie, Superalloys: A Technical Guide (2nd Ed.),
Materials Park, OH: ASM International, pp. 439, 2002
7) C.T. Sims, W.C. Hagel, The Superalloys, John Wiley and Sons: New York,
1972
8) C.T. Sims, N.S. Stoloff, W.C. Hagel, Superalloys II, John Wiley and Sons: New
York, pp. 615, 1987
9) N. Birks, G.H. Meier, F.S. Pettit, High-Temperature Oxidation of Metals (2nd
Ed.), Cambridge, UK: Cambridge University Press, 2006
10) P. Kofstad, High Temperature Corrosion, New York: Elsevier Applied
Science, 1988
11) S. Bose, High Temperature Coatings, Burlington, MA: ButterworthHeinemann, 2007
12) A.C. Vieria Coelho, G.A. Rocha, P.S. Santos, H.S. Santos, P.K. Kiyohara,
Specific surface area and structures of aluminas from fibrillar pseudoboehmite,
Revista Materia, Vol. 13, No. 2, 2008
13) D.B. Miracle, The physical and mechanical properties of NiAl, Acta
Metallurgica et Materialia, Vo. 41, pp. 649, 1993
15
14) M. Bestor, Investigation of the effect of hafnium on the properties of β-NiAl
bond coats deposited on Ni-based superalloys, PhD dissertation, University of
Alabama, 2010
15) A. Russell, K. Lee, Structure-Property Relations in Nonferrous Metals, John
Wiley & Sons, Hoboken, NJ, 2005
16) J.A. Haynes, B.A. Pint, W.D. Porter, I.G. Wright, Comparison of thermal
expansion and oxidation behaviors of various high-temperature coating materials
and superalloys, Materials at high temperatures, Vol. 21, No. 2, pp. 87-94, 2004
17) H. Cobb, History of stainless steel, ASM International, Materials Park, OH,
2010
18) F.O. Soechting, A design perspective on thermal barrier coatings.
Proceedings of the thermal barrier coating workshop, Cleveland, OH: NASA
Lewis Research Center, 1995
19) S.M. Meier, D.K. Gupta, The evolution of thermal barrier coatings in gas
turbine engine applications, Journal of Engineering for Gas Turbines and Power,
Vol. 116, No. 1, pp. 250-257, 1994
20) H.W. Grunling, W. Mannsmann, Plasma sprayed thermal barrier coatings for
industrial gas turbines: morphology, processing and properties, Journal De
Physique IV, Vol. 3, No. 11, pp. 903-912, 1993
21) S. Reuss, H. Vehoff, Temperature dependence of the fracture toughness of
single phase and two phase intermetallics, Scripta Metallurgica et Materialia, Vol.
24, pp. 1021-1026, 1990
22) R.R. Bowman, R.D. Noebe, R. Darolia, in 2nd Annual HITEMP Review,
NASA CP-10039, pp. 41-47, 1989
23) J.J. Lewandowski, G.M. Michal, I.E. Locci, J.D. Rigney, in Alloy phase
stability and design, Materials Research Society CP, pp. 341-347, 1991
24) D.B. Miracle, R. Darolia, NiAl and its alloys, Structural Applications of
Intermetallic Compounds, John Wiley & Sons, 2000
16
25) S. Takizawa, S. Miura, T. Mohri, Structural stability of NiAl with the L10
structure and local lattice distortion in the Ni3Al alloy around excess Al atoms,
Intermetallics, Vol. 13, No. 11, pp. 1137-1140, 2005
26) R.L. Wachtell, in Investigation of various properties of NiAl, Technical Report
52-291, Wright Air Development Center, 1952
27) V. Raghavan, Material Science and Engineering: A first course (4th Ed.), Jay
Print Pack Private Ltd., New Delhi, India, 1998
17
CHAPTER 2:
A MULTI-STAGE HEIRARCHICAL APPROACH TO
ALLOY DESIGN
P.K. Ray, T. Brammer, Y.Y. Ye, M. Akinc and M.J. Kramer
Ames Laboratory and Department of Materials Science and Engineering
Iowa State University, Ames, IA 50011
(Published in JOM, Vol.62, pp. 25-29, 2010)
Abstract
A multi-stage hierarchical sieving approach based on a combination of semiempirical and ab initio calculations along with selected experimental studies were
used to down-select potential alloy systems for ultra-high temperature
applications. This approach indicates that the Mo-Ni-Al system has potential for
applications at the target temperatures of 1200-1300°C . The Mo was selected for
its high melting temperature, room temperature toughness and creep resistance
while the NiAl is a reservoir for the Al2O3 passivating scale. Microstructures
based on casting and powder processing of the Mo-Ni-Al alloys was studied.
Oxidation behavior of the Mo-Ni-Al alloys at 1100 and 1200°C in dry air was
determined and those alloys with ≤ 20 at% Mo were shown to be superior to the
T2 (Mo5SiB2).
Furthermore, the calculations predicted that small amounts of
platinum group metals (PGM) Pd, Ir and Rh additions would increase the melting
temperature without forming detrimental intermetallic phases, in which resulted in
improved oxidation stability of the NiAl phase.
18
Introduction
Superalloys are the workhorse materials for land based gas turbine blades,
which are required to operate under extremely harsh combustion environment
while retaining their mechanical integrity. Operating temperatures for Ni-based
superalloys have been greatly increased by application of the thermal barrier
coating along with a suitable Ni-Al based bond coat. Commercial turbine blades
are routinely operated at temperatures up to 1150°C (1). Even higher efficiencies
such as envisioned in the DOE NETL’s FutureGen gas turbine program will
require alloys that can function at temperatures above melting point of today’s
commercial alloys. Hence, design of ultra high temperature alloys will require a
major breakthrough. Refractory metal silicides in the metal-T2-A15 phase field
region can operate at high temperatures and retain their strength (2-5). However,
oxidation resistance of these alloys is significantly reduced. Alloys in the T1-T2A15 or disilicides of refractory metals offer improved oxidation resistance (6).
However, the presence of only brittle intermetallics in the phase assemblage
renders them unusable in practice.
A significant increase in operating temperatures is not likely to be achieved by
tweaking current Ni-based alloy compositions. A trial-and-error based approach
towards the discovery of new materials is tedious; yet we lack the numerical tools
to efficiently and accurately predict new phases and their properties. Potential
phase space is enormous. For instance, a 4-component Ni based system, with all
the elements selected from the transition metal elements, yields 3,654 possible
combinations, and adding only one more constituent expands the number of
combinations to 23,751. Considering that many commercial high temperature
alloys can have in excess of 8 constituents, a systematic investigation of even a
fraction of the potential phase space through computational thermodynamics or
ab initio methods are unrealistic. Hence using rapid sieving techniques in an
alloy design project is essential. Faster but lower accuracy approaches to quickly
19
screen out the least favorable combinations followed by more rigorous methods
to narrow the potential phase space offers an efficient alloy design approach.
Fortunately, we do have a few clues to pursue for developing fast screening
methods. For instance, the melting temperature of a material can be taken as
one of the guidelines for a prospective high temperature alloy. The melting
temperature of a compound, in turn, correlates well with the interatomic bond
strength and its formation enthalpy (7-9). It has been demonstrated that the
formation enthalpy of multicomponent system can be estimated with a
reasonable degree of accuracy and extremely low computational cost using an
extended Miedema approach (10). However, melting temperature is not the only
requirement; only certain elements are known to promote the formation of
protective oxide scales, most notably, Al, Cr and Si. Therefore, additional
screening criteria include developing alloys comprising of a phase that will
promote the formation of a slow growing protective oxide scale. Due to the low
stability of chromia at the target operating temperatures, 1350°C, and
susceptibility of a silica based scales to moisture in coal combustion byproducts,
we are left with an alumina forming alloy (11). Alloys based on a mix of refractory
metals (RM) and Ni-Al are believed to be the best candidates. The refractory
metal forms the backbone of the alloy, providing toughness at low temperatures
and creep resistance at high temperatures. The β-NiAl (B2 structure, SG 221), in
turn, acts as a reservoir for the passivating alumina scale. However, alloying
additions to the NiAl are required in order to boost its melting temperature (Tm =
1640°C).
The primary focus of this study was to evaluate a series of alloy combinations
narrowed down by the guidelines mentioned above, using semi-empirical, as well
as ab initio calculations for their potential as a high temperature (T~1300°C),
oxidation and fracture resistant material. Select alloy compositions predicted by
20
the computational tools were then synthesized, characterized and tested for their
high temperature oxidative stability as the first goalpost in the alloy development
process.
Experimental Procedure
Theoretical calculations
The initial cut through the prospective phase space was performed using semiempirical calculations based on an extended Miedema approach (10). Briefly,
this approach is based on using an energy minimization scheme by optimizing
the compositions of the constituent binary systems under appropriate mass
balance constraints. Our modified Miedema model was used for estimating the
difference in formation enthalpies between Ni-Al and TM-Al (TM = transition
metal). Twenty six elements from the transition metal block were tested over the
entire composition range of their ternaries with Ni and Al. Three considerations
were given for the selection of the TM : <1> high formation enthalpy, <2>
absence of intermetallics with Ni or Al over a significant composition range and
<3> a cubic crystal form to insure some ductility at lower temperatures. Using
these criteria, Mo was clearly the best candidate. Molybdenum retains its bcc
structure up to its melting temperature and has a limited solubility with Al or Ni.
While not possessing good oxidation stability in of itself, its low diffusivity and low
solubility with NiAl suggests that a graded microstructure can be designed to
minimize susceptibility for oxidation.
The next step in alloy design was to search for ternary addition to the NiAl to
boost its melting temperature. As mentioned above, the ternary addition to the
NiAl should not only raise the melting temperature but should be retained in solid
solution with the NiAl. Again we used our modified Miedema model to downselect the most promising candidates based on their formation enthalpies. Of the
late-transition elements, only Hf, Y, Zr, Nb, Pd, Rh and Ir showed the highest
21
formation enthalpies with NiAl and a smaller affinity for Mo. It is critical that the
ternary addition to NiAl do not exhibit high heat of mixing with Mo. At this point,
more accurate calculations were required.
Ab initio methods are preferred at this stage. Unlike embedded atom methods,
interatomic potentials do not have to be developed or in the case of
computational thermodynamics, developing databases are not necessary. The
disadvantage of ab initio methods is in instances where the crystal structure and
its atomic decoration are not known so that a large number of prospective
structures must be calculated. It should be stressed that unless specifically
determined, ab initio provides only the most enthalpically stable structure (0 K).
Entropically stable structures or highly complex low symmetry systems are very
difficult to identify with this approach. In this particular system we suspected that
the small platinum group metal (PGM) additions will not result in degeneracy of
the β-NiAl, which was readily confirmed experimentally. Therefore the number of
calculations required was small.
The ab initio calculations were carried out using the Vienna ab initio Simulation
Package (VASP) (12-14) to get the thermodynamically stable structure of B2
phase in the Ni1-xAlTMx and NiAl1-xTMx systems using a 54-atom supercell, where
TM represents the transition metal element selected by the Miedema’s model.
The density functional calculations are performed using a plane-wave pseudopotential representation, with ultra-soft pseudo-potentials (13) for all species and
with a plane-wave energy cutoff of 300 eV. The k-point sampling was chosen to
converge all of the total energies to an accuracy of 2 meV/atom. The k-point grid
used for structural relaxation was 6×6×6, chosen according to the MonkhorstPack scheme (15), and symmetry reduced to the irreducible Brillouin zone.
22
Experimental methods
Based on our calculations, the Mo-Ni-Al system was selected as a viable
candidate for the base alloy. A number of compositions in this system were
tested for their resistance to high temperature oxidation. The testing
temperatures were restricted to 1200°C due to massive scal e spallation at higher
temperatures. This pointed to a need for improving the NiAl phase with selected
transition metal additions as indicated by the calculations above.
Mo-Ni-Al alloys
Mo-Ni-Al alloys are known to exhibit a two phase (bcc Mo + β-NiAl)
microstructure over a limited range of compositions (16). All the experiments in
the present study were carried out in this phase field. Higher phase fraction of
Molybdenum would be detrimental to the oxidation resistance; hence the phase
fraction of Mo was kept below 35 at% in all cases. The alloys were synthesized
using powder metallurgical techniques as well as casting. The powder metallurgy
involved mixing the pure metal powders (~ 10 µm) in a mixer/shaker (SPEX
8000, SPEX CertiPrep Inc, Metuchen, NJ) for 10 minutes followed by compaction
of the powders by dry pressing at a pressure of 3500 psi (24 MPa). The
cylindrical pellets thus produced had a diameter of 10 mm and a height of 20
mm. The pellets were then sintered at 1750°C for 45 m inutes. The pellets
synthesized by this route had nominal compositions of Mo = 15, 20, 25 and 35
at%, with the balance in each case being equiatomic NiAl. Oxidation coupons
having a diameter of 10 mm and a thickness of 1 mm were cut from the
cylindrical samples and subjected to interrupted oxidation tests at 1100°C which
is the maximum temperature that current superalloys are exposed to in practice.
Oxidation tests with multiple compositions allowed us to study the effect of Mo
phase fraction and determine an appropriate composition range where the
oxidation resistance was reasonably optimized.
23
The composition of the drop cast alloy (Mo20Ni40Al40) was decided based on the
oxidation behavior of the sintered alloys. Drop cast samples were prepared from
pelletized elemental powder mixture of Mo (Alfa Aesar, 99.5% purity), Ni (MPC,
99.6% purity) and Al (Alfa Aesar, 99.8% purity) which were arc-melted. The
alloys were re-melted thrice to achieve a greater degree of homogenization
followed by drop-casting in order to obtain a cylindrical rod. The drop-cast alloy
was subject to interrupted oxidation tests over a range of temperatures from
1000°C to 1200°C.
In this test, the samples were exposed to the target temperatures for a total of 20
hours with occasional interruptions for mass measurements at ambient
temperature. The testing temperatures were deliberately kept at 1200°C or lower
since our work on oxidation of pure nickel aluminides at temperatures above
1200°C (discussed below) has shown massive scale spallation. M icrostructures
of the as-prepared alloys as well as the oxidized coupons were studied using a
JEOL 5910LV scanning electron microscope (JEOL, Tokyo, Japan) at an
accelerating voltage of 20 kV.
Nickel Aluminides with TM additions
The Ni-Al-TM alloys were produced from pieces cut from pure bulk metal sheets
obtained from the Materials Preparation Center at Ames Laboratory, having a
purity of 99.7% or more. Pure Ni and Al were first arc-melted together in an argon
atmosphere to form β-NiAl. The ab initio calculations suggested that the
transition metals have a preference for the Ni site in the B2 structure. Alloys with
compositions Ni50-xAl50TMx, (x = 3, 6 and 9) were then arc-melted in an argon
atmosphere.
The
samples
were
re-melted
thrice
to
achieve
better
homogenization before drop-casting. The rods were annealed at 1300°C in an
argon atmosphere for 6 hours to ensure homogeneity. SEM was used to
characterize the microstructure while phase analysis was done using x-ray
24
diffraction (XRD) on a Philips PANalytical x-ray diffractometer (Panalytical,
Almelo, Netherlands) with a Bragg-Brentano geometry and Cu Kα1 radiation (λ =
1.54056 Å). The x-ray data was Rietveld refined using the GSAS software in
order to estimate the lattice parameters of the phases as well as determine site
preference for different elements (17). Of the transition metals studied, it was
observed that alloys with Platinum group metal (PGM) modifications formed
single phase alloys while the other alloys showed a eutectic microstructure.
Hence, further studies were carried out only on the PGM modified alloys.
Oxidation coupons were prepared as described above, but more aggressive
testing conditions, 1300°C for two hours and ambient f or 30 min, were employed
to determine if the PGM improved the oxidation stability of NiAl reservoir phase.
Results and Discussion
Oxidation behavior of the Mo-Ni-Al alloys
Samples processed by liquid phase sintering showed the influence of alloy
composition on the oxidation behavior. Alloys under 20 at% Mo, with equiatomic
NiAl showed fairly good oxidation resistance, while the alloys with a higher
volume fraction of Mo (e.g. Mo35Ni32.5Al32.5) failed to form a passivating oxide
layer (Figure 1). The strong composition dependence of oxidation is due to Mo’s
deleterious effect on the oxidation resistance. Minimizing the inter-connected Mo
network is required. Continuity of Mo phase depends on its volume fraction and
the size of the Mo grains. Higher the volume fraction of the Mo phase, and larger
the grain size, larger is the probability of connectivity and resulting in poorer
oxidation resistance.
2
Specific Mass change (mg/cm )
25
0
-20
-40
15 at%
20 at%
25 at%
35 at%
-60
-80
-100
-120
-140
0
5
10
15
20
Time (hours)
Figure 1: Interrupted flowing air oxidation of liquid phase sintered alloys of
different Mo content at 1100°C.
Figure 2a shows a low magnification microstructure of the oxidized Mo20Ni40Al40
cast alloy at 1200°C. It can be seen that the alloy has developed a continuous
alumina scale of about 5 m thickness, albeit not uniform, across the length of
the sample. A closer look at the oxide scale at higher magnification (Figure 2b)
shows that the scale thickness fluctuations is due to presence of long Mo
dendrites growing into the alloy.
26
Figure 2: (a) Low magnification image of the oxidized alloy. The alumina scale
appears to have formed continuously on the surface. (b) Higher magnification
image of the oxidized alloy showing the non-uniformity of scale thickness
primarily as a result of presence of Mo phase at the surface. Mo gets oxidized
and the resultant MoO3 volatalizes allowing oxygen to penetrate relatively larger
distances into the alloy.
It has been widely observed in case of nickel based alloys that the initial
oxidation product of NiAl is the formation of NiO (18). Subsequently, NiO is
reduced by Al resulting in the formation of Ni and Al2O3. Typically such a
27
reduction results in linear rather than parabolic kinetics. Aluminum can also react
with atmospheric oxygen to form Al2O3 which should exhibit parabolic oxidation
kinetics. Additionally, NiO and Al2O3 can react to form a NiAl2O4 spinel (19). All
the reactions involving Ni and Al result in a slow mass gain, whereas the
oxidation of Mo and the subsequent volatilization of MoO3 results in a rapid mass
loss (4). Consequently, the overall oxidation rate is a sum of mass gain due to
oxidation of Ni and Al and loss due to volatilization of MoO3. Figure 3 shows the
oxidation kinetics during the interrupted oxidation process for the Mo20Ni40Al40
alloy at 1200°C. As a baseline comparison the steady stat e mass loss value for
the T2-Mo5SiB2 compound is shown on the same figure for comparison. The
preliminary data indicates that the oxidation behavior of Mo-Ni-Al alloy is
0
2
Mass change (mg/cm )
comparable to the refractory metal silicides at fairly elevated temperatures.
Mo20Ni40Al40
-1
Mo5SiB2 (steady state value)
-2
-3
-7
0
2
4
6
8
10
12
14
Time (hours)
Figure 3: Interrupted flowing air oxidation of an arc-melted and drop cast alloy
(Mo20Ni40Al40) at 1200°C
Oxidation studies were limited to a temperature of 1200°C because the alumina
scale formed on nickel aluminides is prone to spallation above 1200°C. This can
be seen clearly in Figure 4, which shows the cyclic oxidation response of the
28
pure nickel aluminide at 1300°C. The next section will discuss the role of alloying
additions to increase the melting temperature and the oxidation behavior of these
allows.
Oxidation behavior of the nickel aluminide
X-ray diffraction studies on the nickel aluminides with Zr, Y, Hf and Nb additions
indicated a two phase mixture. The PGM additions were shown to form a single
phase with the same crystal system as the β-NiAl. A monotonic and
approximately linear increase in the lattice parameter was observed with
increasing PGM additions. Subsequent Rietveld refinement confirmed that PGM
substitutions preferred the Ni site, which is in accordance with the ab initio
calculations (Figure 4).
Formation Enthalpy (kJ/mol)
-52
NiAl
Rh substituting Al
Ir substituting Al
Pd substituting Al
Rh substituting Ni
Ir substituting Ni
Pd substituting Ni
-56
-60
-64
-68
-72
0
1
2
3
4
5
6
7
8
9 10 11 12
Substitution Level (at%)
Figure 4: Variation of formation enthalpy of β-NiAl with PGM additions.
29
Figure 4 shows the variation of formation enthalpy of PGM substituted β-NiAl. It
can be seen that substitution of PGM elements for Ni is energetically favorable.
Also, controlled additions result in a monotonic change in formation enthalpy.
This suggests that these PGM additions might be beneficial for obtaining an
increase in the melting temperature of the nickel aluminide. The other transition
metals that showed large negative formation enthalpies with the nickel-aluminide
included Hf, Y, Zr, and Nb. All of these additions were experimentally found to
produce deep eutectics. A high formation enthalpy can indicate high stability of
both the solid and the liquid phase. The phase selection is decided by the relative
stability of these two phases. Miedema’s model is a structure-less model; hence
it can’t be used to predict which of these two phases will be more stable.
Formation of a eutectic suggests that the liquid phase is relatively more stable
compared to a single phase intermetallic with the same composition in case of
these alloying additions. Hence, further oxidation studies were carried out with
only the PGM additions, since they were readily soluble in the β-NiAl.
Figure 5 shows the oxidation kinetics for the three PGM substitutions (all 6 at%)
to the β-NiAl compared against undoped β-NiAl as a benchmark. It can clearly be
seen that the Pd substitution performed worse than the β-NiAl during oxidation
testing at 1300°C. The oxide spallation of the Pd samp le was noticeable early on
during the test. The spallation flakes were large and discernable with the naked
eye, just like the β-NiAl spallation. The Rh sample performed better than the
base alloy NiAl, but worse than the alloy doped with Ir. The sample with Rh
showed an adherent oxide scale initially, but after five hours oxide spallation was
observed. From the oxidation resistance point of view, Ir substitution to β-NiAl
appears to be superior. No sign of spallation was evident at times shorter than 20
h for alloys with 9 at% substitution. In all cases it should be noted that the best
performing alloy at the end of 20 hours was always the alloy with highest fraction
of the ternary substitution for all of the elements tested.
2
Specific Mass Change (mg/cm )
30
1.00
0.75
Ir
0.50
0.25
Rh
0.00
-0.25
-0.50
-0.75
Pd
-1.00
0
5
NiAl
10
15
20
Time (hours)
Figure 5: Cyclic oxidation of β-NiAl with and without PGM substitutions.
Figure 6a-d shows the oxidation microstructures of the baseline NiAl along with
each ternary substitution set at 6 at%. These samples were oxidized for 24 hours
at 1300°C and then examined to better assess the evolu tion of the oxide scale
formation. Figure 4 clearly displays that neither the benchmark β-NiAl alloy nor
the β-NiAl doped with Pd additions show an adherent oxide layer. Since both
alloys had relatively significant mass loss, the surface would be devoid of a
continuous oxide scale. The Rh doped β-NiAl showed an overall mass gain as
well as signs of spallation. Looking at the oxide microstructure, it can be seen
that the oxide layer is fairly continuous with a thickness of approximately 10 µm.
The Ir addition showed an almost continuous oxide scale with a thickness of 8
microns. The only discontinuities were observed at the corners of the specimen
where stresses in the oxide layer would be highest.
31
Figure 6: Microstructures of oxidized alloys after oxidation at 1300°C for 24
hours. (a) β-NiAl; and β-NiAl with 6 at% (b) Rh, i.e. Ni44Al50Rh6 ; (c) Ir, i.e.
Ni44Al50Ir6; (d) Pd i.e. Ni44Al50Pd6
Conclusions
We have shown that a hierarchical approach to alloy design starting with a less
accurate but fast tool, followed by a more accurate but time consuming
computational tools can be an efficient and effective means of down-selecting
high temperature alloys. Our modified Miedema model is particularly effective in
identifying regions of high (negative) formation enthalpies, indicative of high
melting temperatures.
Using the criteria of refractory base metal as the
backbone of the alloy and identifying the NiAl as the most promising reservoir
compound for alumina scale former, we demonstrated that Mo-Ni-Al-(PGM)
shows good promise as a high temperature alloy. To further stabilize the
reservoir compound, the most enthalpically stable ternary additions were
determined. The most promising of the Ni-Al-(PGM) alloys based on ab initio
32
calculations were synthesized and their oxidation behavior was studied. It was
found that a few alloys in this system had enough potential to warrant further
investigation of these materials for ultra-high temperature applications. The Mo
phase fraction and its grain size play a critical role in the oxidation behavior of
these alloys. Future work would be focused on developing PGM substituted MoNi-Al alloys in the two phase bcc-Mo (ss) + β-NiAl phase fields.
Acknowledgement
This work was supported by the DOE-FE (AMR program) through Ames
Laboratory contract no. DE-AC02-07CH11358 through Iowa State University.
References
1) R. C. Reed, The superalloys: Fundamentals and applications, Cambridge
University Press, Cambridge, 2006.
2) M. Meyer, M. Kramer, M. Akinc, Advanced Materials 8 (1996) 85-88.
3) M. K. Meyer, M. J. Kramer, M. Akinca, Intermetallics 4 (1996) 273-281.
4) M. K. Meyer, M. Akinc, Journal of the American Ceramic Society 79 (1996)
938-944.
5) J. R. Nicholls, MRS Bulletin 28 (2003) 659-670.
6) A. J. Thom, E. Summers, M. Akinc, Intermetallics 10 (2002) 555-570.
7) J. H. Rose, J. Ferrante, J. R. Smith, Physical Review Letters 47 (1981) 675.
8) Li, P. Wu, Chemistry of Materials 14 (2002) 4833-4836.
9) C. Li, J. Lim Hoe, P. Wu, Journal of Physics and Chemistry of Solids 64 (2003)
201-212.
10) P. K. Ray, M. Akinc, M. J. Kramer, Journal of Alloys and Compounds 489
357-361.
11) J. E. Croll, G. R. Wallwork, Oxidation of Metals 4 (1972) 121-140.
12) G. Kresse, J. Furthmüller, Physical Review B 54 (1996) 11169.
13) G. Kresse, J. Furthmüller, Computational Materials Science 6 (1996) 15-50.
33
14) G. Kresse, J. Hafner, Physical Review B 47 (1993) 558.
15) H. J. Monkhorst, J. D. Pack, Physical Review B 13 (1976) 5188.
16) X. Lu, Y. Cui, Z. Jin, Metallurgical and Materials Transactions A 30 (1999)
1785-1795.
17) B. H. Toby, Journal of Applied Crystallography 34 (2001) 210-213.
18) G. R. Wallwork, Reports on Progress in Physics 39 (1976) 401-485.
19) X. Zhao, I. P. Shapiro, P. Xiao, Surface and Coatings Technology 202 (2008)
2905-2916.
34
CHAPTER 3:
EFFECT OF PLATINUM GROUP METAL SUBSTITUTIONS
AND HAFNIUM ADDITIONS ON THE OXIDATION
RESISTANCE OF β-NiAl
T. Brammer, M. Akinc and M.J. Kramer
Ames Laboratory and Department of Materials Science and Engineering
Iowa State University, Ames, IA-50011
(To be submitted to Corrosion Science Journal)
Abstract
High temperature oxidation studies were performed on β-NiAl along with modified
NiAl alloys using Platinum Group Metal (PGM) substitutions in the range of 3-9
at%. Both isothermal and cyclic tests were carried out in the temperature range
of 1150-1300°C using dry flowing air.
The PGM that showed the biggest
improvement in oxidation resistance to NiAl was iridium which was shown to
slightly decrease the oxidation growth rate as well as create an oxide layer that
was more adherent to the surface.
Alloys that had up to 0.1 at% hafnium
addition were also tested in the same manner for comparison. Hafnium additions
showed a greater effect of slowing the oxidation growth rate and still afforded
good adherence of the oxide scale to the surface. The most oxidatively resistant
sample tested was the 6-9 at% iridium containing NiAl alloys with addition of 0.05
at% hafnium.
35
Introduction
Many of the current land based gas turbine blades utilize a nickel-based
superalloy, which have two major phases, γ-Ni and the L12 Ni3Al. Although these
alloys have good oxidation resistance and excellent fracture toughness, they
have reached their limit with respect to operating temperature.
The melting
temperature of Ni3Al is 1363°C, while that for Ni is 1450°C. These tur bine blades
must employ the use of cooling systems along with a thermal barrier coating
(TBC) to shield the base alloy from the extreme gas temperatures (1). Even
then, these alloys have an upper use temperature of 1100ºC (1). A higher
combustion temperature results in improved Carnot engine efficiency (2); hence
alloys that can be used at higher operating temperatures while retaining their
oxidation resistance and mechanical strength are highly desirable.
The base alloy selected was the intermetallic β-NiAl which has a high melting
temperature and inherently good oxidation resistance (5).
Its high melting
temperature of 1640°C allows it to operate at higher combustion temperatures
yielding greater turbine engine efficiencies. An extended Miedema model is
employed as a rapid sieving tool to determine which third element substitution
predicts an increase the enthalpy of formation of the base alloy which also
manifests itself with an increase in melting temperature (4).
The extended
Miedema model calculations predicted that PGM substitutions will lead to
reduction in formation enthalpy relative to the base alloy. Ab initio calculations
were then carried out for the systems identified as potential candidate alloys by
the extended Miedema approach. Alloys were also screened for their ability to
form a protective oxide scale when subjected to high temperature environments.
Although the operating temperatures are a point of interest, the main subject of
focus of this report is the oxidation resistance of ternary alloys based on β-NiAl.
36
In this paper, we report the isothermal and cyclic oxidation behavior of β-NiAl
with PGM substitutions and with Hf additions. The PGM alloying additions were
expected to increase the melting temperature of NiAl (3) as well as improve the
oxidation resistance (6-9).
Alloys were subjected to both isothermal and cyclic
oxidation tests. Wherever applicable, the rate constants were estimated from the
isothermal oxidation plots. The scale adherence and relative protection afforded
to the underlying base alloy were assessed from the cyclic oxidation runs. Both
types of tests are critical to characterize the growth kinetics of the oxide and
stability of the protective oxide scale. Additionally, microscopic examination of the
oxide scales was carried out to assess the scale thickness and microstructure.
Experimental Procedure
Materials
The alloys were synthesized from pieces cut from pure bulk metal sheets
obtained from the Materials Preparation Center (USDOE, Ames Laboratory,
Ames, Iowa) having a purity of 99.7% or better. Pure Ni and Al were first arcmelted together on a chilled copper hearth in an argon atmosphere using a nonconsumable tungsten electrode. This initial melting ensured formation of β-NiAl.
The alloying additions were then melted along with the β-NiAl synthesized
according to the aforementioned steps. The nominal compositions were prepared
according to Ni50-xAl50PGMx (x=0, 3, 6, and 9 at %). The alloys containing Hf
were produced by using pieces of Hf modified Ni sheet metal instead of pure Ni.
After synthesis of desired composition by arc-melting, the samples were turned
over and re-melted a minimum of three times in the argon atmosphere to achieve
better bulk homogenization. After re-melting the samples were drop-cast into a
10-mm-diameter cylinder using the arc-melting furnace. These rods were then
annealed in an inert atmosphere at 1300°C for 6 hour s to achieve a better
compositional homogeneity.
37
The annealed rod was sectioned into 1 mm thick discs using a circular precision
saw. The samples to be tested in the cyclic furnace were then mechanically
polished using wet 600 grit SiC paper. A 1-mm-diameter hole was drilled into
samples for the isothermal testing and suspended vertical furnace tube by a
sapphire wire. All samples were washed with methanol before testing.
Oxidation testing and characterization
Isothermal oxidation at 1150 and 1200°C was carried ou t using a high
temperature oxidation testing rig fitted with a Mellen SC13R 1.25 inch vertical
tube furnace (Concord, New Hampshire) and a Cahn 2000 recording
electrobalance (Madison, Wisconsin) having an accuracy of 0.1 mg .
The
sample was suspended from the balance on a sapphire wire hook. Dry air was
passed through the system at a rate of 60 mL/min. An Ionmaster fan was used
to remove the static electricity in the system to ensure a higher signal to noise
ratio in the measurements. The furnace was turned on and set to the desired
dwelling temperature using a heating rate of 15°C/mi n and then held at the
temperature for 100 hours.
Since the Mellen furnace could not maintain
operating temperatures as high as 1300°C, a different furnace was used for
experiments at 1300°C. Isothermal oxidation at 1300° C was carried out in a
Thermo Scientific Lindberg Blue M 1500 (Asheville, North Carolina) horizontal
open tube furnace by passing dry air at a rate of 200 mL/min. The masses of the
coupons were recorded before and after the oxidation test for each alloy. After
each 100 hour test, the samples were removed from the furnace fore
microstructural characterization. Each test was conducted twice for each of the
alloys shown with the average being reported.
Cyclic oxidation at 1150°C was carried out using an auto mated vertical Suga
open tube furnace (Hokkaido, Japan).
Each cycle consisted of one hour at
testing temperature followed by a half hour at ambient temperature at which time
38
the sample weighed.
The samples were removed from furnace after five
hundred cycles for microstructural characterization.
Due to the number of
samples that needed to be tested and the availability of the furnace, the samples
analyzed at higher temperatures were tested in a different furnace.
Cyclic
oxidation at 1200 and 1300°C was manually carried out using a Thermo
Scientific Lindberg Blue M 1500 (Asheville, North Carolina) horizontal open tube
furnace. Each cycle consisted of two hours at the desired testing temperature
followed by half an hour at ambient temperature at which time the mass
measurements were recorded. The samples were removed from the furnace
after one hundred hours at the testing temperature to characterize the scale
microstructure.
After oxidation tests, each sample was gold sputtered and then copper plated.
The copper coating helps to retain the oxide scale during the polishing process.
After copper plating each sample was vertically mounted using epoxy resin. The
samples were then cut so that the middle of the oxidized sample was visible at
the surface. This surface was then polished on a polishing wheel down to a
0.3 m surface finish. The polished samples were then analyzed using a JEOL
5910Lv scanning electron microscope (SEM) using an accelerating voltage of 20
kV.
Results and Discussion
Isothermal Oxidation at 1150°C
Figures 1 and 2 show the oxide scales of the alloys after 100 hours of testing at
1150°C. The copper plating and the sample surface are labeled in the figures
with the oxide scale being the dark line between these two regions. All
compositions are given in atomic percent with the hafnium amount being 0.05
at% unless otherwise noted.
39
Copper
Sample
Figure 1: Microstructures of oxidized alloys after isothermal oxidation at 1150°C
for 100 hours (a) β-NiAl, (b) NiAl+Hf
PGM additions alone were found to be inadequate to sufficiently form a
protective oxide scale at higher temperatures (Table 1), so hafnium was
introduced into these alloy systems in the early stages of alloy development. The
micrographs shown in Figure 1 indicate that the hafnium addition has a beneficial
effect at 1150°C.
Compare β-NiAl (Figure 1a) to β-NiAl+Hf (Figure 1b) and
notice that without Hf addition the NiAl exhibits an adherent but non-uniform and
relatively thick oxide scale which was measured to be about 15 m at the thickest
point. The same alloy with hafnium addition exhibits a very thin, continuous and
uniform oxide layer that was measured to be 2 m at the thickest point. The
hafnium modified alloy appears to be more protective than the former because a
thinner oxide scale contains less residual stress than a thick oxide scale (10-13)
and would retain the oxide scale more readily.
At 1150°C, the PGM modification seems to have a neglig ible effect when
compared to the hafnium effect (Figure 2). The sample shown in Figure 2a
contains PGM with no hafnium and shows a scale that is more planar and slightly
thinner than the NiAl shown in Figure 1a. All PGM+Hf alloys show similar scales
that are continuous and have relatively uniform thicknesses of 1-2
m.
At
seemingly random intervals, the oxide scale protrudes into the sample and
creates small pockets of thicker oxide. Yang referred to these pockets as “oxide
pegs” and they are believed to anchor the oxide to the substrate creating a more
40
adherent oxide scale (14). The white spots noticed at the bottom of these oxide
pegs (Figure 2a and 2b) were found to be regions high in hafnium. It is plausible
that the small additions of hafnium allow the formation of these pegs to anchor
the oxide to the substrate alloy. These oxide scales are comparable to the oxide
scale formed by NiAl+Hf alloy shown in Figure 1b.
Copper
Sample
Figure 2: Microstructures of alloys after isothermal oxidation at 1150°C for 100
hours (a) Ni-50Al-9Ir, (b) Ni-50Al-6Rh+Hf, (c) Ni-50Al-6Ir+Hf, (d) Ni-50Al-6Pd+Hf
Figure 3 shows a graphical representation of the calculated parabolic rate
constants. The rate constants provide a clear and concise way of representing
and comparing the oxide scale growth rates (15, 16).
Within the NiAl alloy
systems, the dominating factor of Al203 scale growth is primarily oxygen diffusion
through the scale to the alloy surface followed by the upward diffusion of
aluminum to the oxide scale (17-20). A faster growing scale would increase in
thickness rapidly until the residual stresses within the oxide layer could no longer
be contained (10-13). At this point the oxide begins to form cracks to relieve the
built up stresses and eventually breaks off from the metal surface.
These rate constants were calculated using data between 50 and 100 hours to
ensure that the transient oxidation from θ-Al2O3 to α-Al2O3 had completed and
was not contributing to irregular mass changes since θ-Al2O3 grows at a different
41
rate than α-Al2O3 (5, 6, 21, 22).
All of the isothermal oxidation curves fit a
2
-4 -1
Kp (mg cm s )
parabola with R2 ≥0.85.
1.E-08
1.E-07
1.E-06
1.E-05
1.E-04
1.E-03
NiAl
9% Ir
NiAl+Hf
6%Pd+Hf
6%Rh+Hf
6%Ir+Hf
1.E-02
1.E-01
1.E+00
Figure 3:
The parabolic rate constants of oxide growth during isothermal
oxidation at 1150°C for 100 hours. Note the y scale is chosen such that the
higher the bar, the lower the oxidation rate.
The baseline alloy β-NiAl shows the highest oxidation rate. 9% Ir substitution
was used as a reference to show how PGM modification affects the oxide growth
rate. The largest substitution of iridium was tested to show the highest effect.
This alloy showed a slightly slower oxidation rate than the baseline β-NiAl. The
hafnium modified NiAl alloy showed a significantly slower growing oxide than the
baseline NiAl.
It is understood from this data that of the two modifications,
adding hafnium is more effective than substituting PGMs at forming a slower
growing oxide. The alloys that produced the slowest growing oxides at 1150°C
were the ones that contained both PGM and hafnium modifications. Combining
the iridium with the hafnium modification in the same alloy produced the slowest
growing oxide scale.
42
Isothermal Oxidation at 1200°C
Figure 4 shows the oxide scale morphology of the alloys after testing at 1200°C
for 100 hours. Compared to Figures 1 and 2, these samples appear to have
oxide scales that are non-uniform and thicker. At 1200°C the NiAl alloy produced
an oxide scale that spalled off in one large piece during retrieval of the specimen
from the test rig. This is evidenced in the micrograph of Figure 4a by the missing
oxide layer. The other PGM+Hf modified alloys are shown to have consistent
oxidation resistance. The oxide scales at 1200°C are ab out 5-6 m as compared
to 1-2 m at 1150°C. Overall, the PGM+Hf alloys are not show ing signs of scale
spallation.
Copper
Sample
Figure 4:
Microstructures of oxidized samples after isothermal oxidation at
1200°C for 100 hours (a) β-NiAl, (b) Ni-50Al-6Rh+Hf, (c) Ni-50Al-6Ir+Hf, (d) Ni50Al-6Pd+Hf
Figure 5 shows a graphical representation of the measured parabolic rate
constants of the alloys at 1200°C. These rate constants w ere calculated using
data between the testing hours of 50 and 100.
All the isothermal oxidation
2
curves at this temperature fit a parabola with R ≥0.93 or higher. The graph is
shown as a logarithmic scale starting at 1 and decreasing along the y-axis.
Although the scale of the NiAl alloy spalled after the test (Figure 4a), no
43
spallation was noticed during the test and the parabolic rate is assumed to be
correct.
1.E-08
2
-4 -1
Kp (mg cm s )
1.E-07
NiAl
9% Ir
1.E-06
NiAl+Hf
1.E-05
6%Pd+Hf
1.E-04
6%Ir+Hf
6%Rh+Hf
1.E-03
1.E-02
1.E-01
1.E+00
Figure 5: The parabolic rate constants of oxide growth after isothermal oxidation
at 1200°C for 100 hours. Note the y scale is chosen such t hat the higher the bar,
the lower the oxidation rate.
As demonstrated previously in Figure 3, the oxidation growth rates appear to be
similar in trend for the PGM+Hf modified alloys and the NiAl+Hf alloy. Again the
iridium does show a slight decrease in the oxide growth rate when compared to
the baseline NiAl. The hafnium addition has a significant effect of slowing the
growth rate. It should be noted that although they remain similar to each other,
the oxidation rates did increase by an order of magnitude compared to the rate
constants at 1150°C as expected.
Isothermal Oxidation at 1300°C
Samples with varying levels of substitution for the Ni in β-NiAl base alloy were
tested for oxidation resistance at 1300°C. The thermal shock at the end of the
cycle could, in part, account for the spallation of the scale and therefore mass
loss for some of the samples as shown in Table 1.
44
Table 1: Specific mass change (mg/cm2) after isothermal oxidation at 1300°C for
24 hours
Alloy
Ni-50Al-3X
Ni-50Al-6X
Ni-50Al-9X
X=Pd
-1.58
-1.54
-1.55
X=Rh
-1.73
-1.08
1.42
X=Ir
-0.47
0.89
0.86
For reference, the baseline β-NiAl had a mass change of -1.75 mg/cm2
As seen in Table 1, the palladium substitutions have negligible effect on the
oxidation resistance of NiAl. Rh and Ir substitutions help improve the oxidation
resistance, with higher substitutions leading to better oxidation resistance.
Photographs of the specimens after the 24 hour oxidation test are displayed in
Figure 6. It can be seen in these pictures that most of the oxide scale had fallen
off of the surfaces of the β-NiAl and Pd substituted alloys (Figure 6a and 6b),
leaving behind bulk alloy with small flecks of oxide scale. On the surface of the Ir
and Rh containing samples (Figure 6c and 6d) no bare metal was visible. Each
sample has a 10 mm diameter.
45
Metal
Oxide
Figure 6: Photographs of the surface of the samples after isothermal oxidation at
1300°C for 24 hours. (a) Ni-50Al, (b) Ni-50Al-9Pd, ( c) Ni-50Al-9Ir, (d) Ni-50Al9Rh
Figure 7 shows the oxide morphology of the oxidized samples containing 6 at%
of each PGM after 24 hours of testing at 1300°C. The dark band between the
copper plating and the sample is the oxide scale. Two of the images (Figure 7a
and 7b) do not show this oxide scale, presumably because the oxide had spalled
off, leaving behind new sample surface without a protective oxide.
46
Copper
Sample
Figure 7: Microstructures of oxidized alloys after isothermal oxidation at
1300°C for 24 hours (a) β-NiAl, (b) Ni-50Al-6Pd, (c) Ni-50Al-6Ir, (d) Ni-50Al-6Rh
The oxidized microstructure of NiAl (Figure 7a) is almost identical to the oxidized
microstructure of the Pd modified alloy (Figure 7b). As shown in the specific
mass changes in Table 1, both of these alloys lost mass. These micrographs
give evidence that the mass loss was due to the spallation of the oxide layer
formed during oxidation testing. The oxide scale in the Rh modified alloy, while
present, is not continuous (Figure 7d). The thickness of the remaining oxide was
found to be about 10 m. The Ir modified alloy shows a continuous oxide scale
(Figure 7c), with a thickness of approximately 8 m.
Small amounts of hafnium additions to similar alloys have produced a more
adherent and protective oxide scale (23-27). Hence further tests were carried
out by adding Hf in minor amounts to the Ni-50Al-6Rh alloy. Addition of as low
as 0.05 at% Hf resulted in a mass gain of 0.46 mg/cm2 while the same alloy
without the Hf addition had a mass loss of 1.08 mg/cm2 (See Table 1). Higher Hf
additions showed similar mass gains, i.e. 0.1, 0.25, 0.5 at% Hf addition resulted
in 0.47, 0.48, 0.54 mg/cm2 mass gain respectively.
Hafnium addition was shown to result in forming an adherent and continuous
oxide scale as shown in Figure 8 compared to discontinuous scale observed the
47
same alloy without Hf addition (see Figure 7d). Therefore, the Hf additions were
limited to 0.05 at% Hf for the alloys studied going forward.
Figure 8: Microstructure of oxidized Ni-50Al-6Rh+0.05Hf alloy after isothermal
oxidation at 1300°C for 24 hours
Cyclic Oxidation at 1150°C
Figure 9 shows the specific mass change as a function of oxidation cycles for 0,
3, 6, 9 at% PGM substitutions (Figure 9a, 9b, 9c respectively) as well as alloy
with 9 at% PGM and 0.05 at% Hf (Figure 9d). Specific mass change for β-NiAl is
also shown as a baseline for comparison.
Mass loss on any sample with
oxidation implies spallation of oxide scale. This continual oxide spallation process
would eventually deteriorate the sample beyond use after many cycles.
Spallation in these cyclic experiments is assumed to result from the stresses
caused by the thermal expansion mismatch between the alloy and oxide scale
(27-29).
1.2
2
Specific Mass Change (mg/cm )
2
Specific Mass Change (mg/cm )
48
0.8
0.4
0.0
-0.4
Rh
Ir
Pd
NiAl
-0.8
-1.2
0
100
200
300
400
500
1.2
0.8
0.4
0.0
-0.4
-0.8
-1.2
0
100
)
2
0.8
0.4
0.0
-0.4
-0.8
-1.2
200
300
Cycles
400
500
Specific Mass Change (mg/cm
2
Specific Mass Change (mg/cm )
1.2
100
300
400
500
Cycles
Cycles
0
200
1.0
0.5
0.0
NiAl
NiAl+Hf
6Rh+Hf
6Ir+Hf
6Pd+Hf
-0.5
0
100
200
300
400
Cycles
Figure 9: Mass change per surface area plots for alloys during cyclic oxidation at
1150°C for 500 cycles (a) 3% PGM, (b) 6% PGM, (c) 9% PG M, (d) 6% PGM+Hf
The graph in Figure 9a shows that with the exception of Ir, the PGM substitutions
at 3% have no beneficial effect to the cyclic oxidation resistance.
At 6%
substitution (Figure 9b) the iridium containing alloy shows no oxidation spallation
over the entire 500 cycle test at 1150°C. The rhodiu m containing alloy also
showed improved oxidation resistance, with possibly slight spallation as indicated
by small scatter in mass change. The palladium modified alloy performs very
500
49
similarly to the baseline NiAl alloy and shows mass loss after 100 cycles. Alloys
containing 9% of the PGM (Figure 9c) show very similar results to that of the 6%
modifications. Here the rhodium containing alloy again shows a slight decrease
in mass change at 500 cycles. Finally, alloys containing 6% PGM substitution
along with Hf modifications (Figure 9d) appear to have the best cyclic oxidation
resistance. The oxide grows very slowly as seen by the minimal mass gain over
long periods of time and the oxide adheres well to the surface due to the oxide
pegs seen before in Figure 2.
Here the PGM+Hf alloys show very similar
oxidation resistance. Some of these alloys can be compared to previous work
done by Leyens et al. in which low substitution amounts of iridium outperformed
low substitution amounts of palladium during cyclic oxidation testing at 1150°C
(8). In this work, he also noticed that none of the precious metal modifications
improved oxidation resistance as effectively as Hf doping.
Figure 10 shows the oxide scales of the cyclic tested samples at 1150°C for 500
cycles. Alloy compositions are designated on each micrograph. The black band
represents the oxide scale with copper plating above. No hafnium is shown in
these images as all hafnium containing alloys show a very continuous, thin (34 m), and adherent oxide scale.
50
Ni-50Al-3Rh
Ni-50Al-3Ir
Ni-50Al-3Pd
Ni-50Al-6Rh
Ni-50Al-6Ir
Ni-50Al-6Pd
Ni-50Al-9Rh
Ni-50Al-9Ir
Ni-50Al-9Pd
Figure 10: Microstructures of oxidized alloys after cyclic oxidation at 1150°C for
500 cycles. Samples are labeled within the figure with the copper plating being
on top.
Comparison of the micrographs for several types of PGM substituted alloys at
three substitution levels indicates that all 3% PGM substituted alloys showed a
non-uniform and undulating oxidation layer indicative of scale spallation. Of the
6% PGM substituted alloys; iridium exhibited a uniform, continuous scale with a
thickness of 8 m. Increasing the PGM content from 6 to 9 % has resulted in
slight improvement in oxidation resistance. Among the three PGMs, palladium
was the least effective showing mass loss and spallation at all substitution levels
as indicated in Figure 9.
Cyclic Oxidation at 1200°C
Figure 11 shows the specific mass change for alloys with 3, 6 and 9% PGM
(Figure 11 a, b, and c respectively) along with 6% PGM and 0.05% Hf addition at
1200°C.
2
Mass Change (mg/cm )
2
1
Specific Mass Change (mg/cm )
51
0
-1
-2
NiAl
Rh
Ir
Pd
-3
-4
0
20
40
60
80
100
1
0
-1
-2
-3
-4
0
20
2
Specific Mass Change (mg/cm )
2
Specific Mass Change (mg/cm )
1
0
-1
-2
-3
-4
20
40
60
60
80
100
Furnace Hours
Furnace Hours
0
40
80
100
1.0
0.5
0.0
NiAl
NiAl+Hf
6Rh+Hf
6Ir+Hf
6Pd+Hf
-0.5
0
20
Furnace Hours
40
60
80
Furnace Hours
Figure 11: Mass change per surface area plots for alloys during cyclic oxidation
at 1200°C for 100 hours (a) 3% PGM, (b) 6% PGM, (c) 9% PGM, (d) 6%
PGM+Hf
The trends observed for specific mass change at 1200°C is similar to that at
1150°C. The two PGMs that showed to produce a prote ctive oxide scale at this
temperature are iridium and rhodium containing alloys.
Clearly, addition of
100
52
hafnium prevents the spallation for all the samples including the benchmark βNiAl (Figure 11d).
Figure 12 shows cross-sections of the oxidation scales after cyclic testing at
1200°C. As demonstrated before in Figure 8, the all oys that continue to form a
protective oxide scale are the higher substitution amounts of both iridium and
rhodium. Palladium formed an oxide scale that spalled off within all substitution
amounts tested. The hafnium modified alloys all showed a thin, adherent, and
continuous oxide scale.
Ni-50Al-3Rh
Ni-50Al-3Ir
Ni-50Al-3Pd
Ni-50Al-6Rh
Ni-50Al-6Ir
Ni-50Al-6Pd
Ni-50Al-9Rh
Ni-50Al-9Ir
Ni-50Al-9Pd
Figure 12:
Microstructures after cyclic oxidation at 1200°C for 100 hours at
testing temperature.
Samples are labeled within the figure with the copper
plating being on top.
The 3% iridium containing alloy (Figure 12) shows a thicker oxide scale than the
3% rhodium containing alloy yet the specific mass change plot shows the
rhodium alloy to have gained more mass. This can be explained due to the
missing oxide at other locations on the sample surface of the 3% Ir sample. The
image displayed in Figure 12 is of a region that retained the oxide. The rhodium
containing alloys in Figure 12 seem to show improved scale morphology when
53
compared to those in Figure 10. This apparent discrepancy can be explained
primarily due to the number of cycles to which the alloys were exposed.
Previously the alloys were tested for 500 cycles at 1150°C, whereas in this test
the cycles were reduced to about 20 at 1200°C. The change in temperature and
number of cycles accounts for the more planar looking oxide on the 1200°C
rhodium containing alloys in Figure 12.
Cyclic Oxidation at 1300°C
Figure 13 shows the specific mass change curves for varying amounts of PGM
substituted alloys.
Figure 13 shows results from the most aggressive test
conducted on these alloys. All of the 3% PGM alloys showed signs of spallation
within the first few hours during this test. The 6% PGM substituted alloys all
showed the effects of scale spallation within the first ten hours. The 9% PGM
substituted alloys performed slightly better and did not show signs of spallation
until after 20 hours. In the 6 and 9% PGM alloys it is noticed that the iridium
containing alloy is the last to spall off its oxide scale.
Both the palladium
containing alloy along with the baseline β-NiAl showed poor oxidation resistance
compared to the Ir and Rh substituted alloys. Again it can be seen that the
hafnium along with the PGM substitutions give rise to the best oxidation resistant
alloys. Of the PGM+Hf modified alloys, the PGM that gives rise to the most
desired oxidation resistance is iridium.
2
2
Specific Mass Change (mg/cm )
2
Specific Mass Change (mg/cm )
54
0
-2
-4
-6
-8
Rh
Ir
Pd
NiAl
-10
-12
0
20
40
60
80
100
2
0
-2
-4
-6
-8
-10
-12
0
20
80
100
80
100
2
2
Specific Mass Change (mg/cm )
2
60
Time (hours)
Time (hours)
Specific Mass Change (mg/cm )
40
0
-2
-4
-6
-8
-10
-12
0
20
40
60
Time (hours)
80
100
1.0
0.5
0.0
NiAl
NiAl+Hf
6Rh+Hf
6Ir+Hf
6Pd+Hf
-0.5
0
20
40
60
Time (hours)
Figure 13: Mass change per surface area plots for alloys during cyclic oxidation
at 1300°C for 100 hours (a) 3% PGM, (b) 6% PGM, (c) 9% PGM, (d) 6%
PGM+Hf
Cyclic oxidation results show a continuing trend that both iridium and to a lesser
degree rhodium substitution to β-NiAl have a beneficial effect on the oxidation
resistance at high temperatures. Palladium additions continued to show little or
no effect on oxidation when compared to the β-NiAl alloy. It is unknown at this
time why palladium additions would have little effect on the scale adherence
55
during cyclic oxidation while its neighbor on the periodic table, Rh, would have a
significant effect. Grain size effects on oxidation of similar alloys have been
noted previously (14). Other changes such as the thermal expansion coefficient
could be playing a role here as well.
Conclusions
Of the three types of PGM substitutions to β-NiAl, iridium and rhodium showed
significant improvement in oxidation resistance at all temperatures tested.
Iridium slowed the growth of the oxide scale in the isothermal tests. This may
also have contributed to the formation of a more adherent and protective oxide
scale during the cyclic oxidation. The iridium substituted alloys were always the
last of the PGMs tested to show signs of scale spallation.
In general, the
oxidation resistance increased as the amount of PGM increased.
It was shown that hafnium improves the oxidation resistance of both isothermal
and cyclic oxidation environments in the temperature range tested. Analysis of
the isothermal oxidation results gives evidence that hafnium addition as low as
0.05 at% slows the oxidation rate giving rise to a thinner oxide scale. Analysis of
the cyclic oxidation results gives evidence that hafnium additions forms an oxide
that adheres better to the metal.
Combining the iridium substitutions along with the hafnium addition gives rise to
an alloy with a slow growing, continuous, planar, and adherent oxide layer and
hence superior oxidation resistance for both isothermal and cyclic environments.
Acknowledgement
This work was supported by the DOE-FE (AMR program) through Ames
Laboratory contract no. DE-AC02-07CH11358 through Iowa State University.
56
References
1) M. Donachie, S. Donachie, Superalloys: A Technical Guide (2nd Ed.), pp. 8,
309-314, 2002
2) R. Reed, The Superalloys: Fundamentals and Applications, pp. 9-10, 2006
3) P.K. Ray, et al., Journal of Metals, Vol. 62, pp. 25–29, 2010
4) P.K. Ray, et al., Journal of Alloys and Compounds, Vol. 489, pp. 357-361.
5) J. Balmain, et al., Materials Science and Engineering, Vol. A224, pp. 87-100,
1997
6) Y. Cadoret, et al., Oxidation of Metals, Vol. 64, 2005
7) M.J. Li, et al., Surface and Coatings Technology, Vol. 167, pp. 106-111, 2003
8) C. Leyens, et al., Surface and Coatings Technology, Vol. 133-134, pp. 15-22,
2000
9) G. Fisher, et al., Surface and Coatings Technology, Vol. 113, pp. 259-267,
1999
10) W.J. Quadakkers, et al., Surface Coatings Technology, 1999
11) H.M. Tawancy, et al., Scripta Metallurgica et Materialia, Vol. 33, pp. 14311438, 1995
12) W.J. Brindley, R. Miller, Surface and Coatings Technology, Vol. 43, pp. 446457, 1990
13) E. Basuki, et al., Materials Science and Engineering, Vol. A224, pp. 27-32,
1997
14) S. Yang, et al., Intermetallics, Vol. 9, pp. 741-744, 2001
15) S. Mrowec, et al., Oxidation of Metals, Vol. 8, 1974
16) D. Monceau, B. Pieraggi, Oxidation of Metals, Vol. 50, pp. 477, 1998
17) B.A. Pint, et al., Solid State Ionics, Vol. 78, pp. 99-107, 1995
18) B.A. Pint, et al., Oxidation of Metals, Vol. 39, pp. 167, 1993
19) J. Jedlinski, S. Mrowec, Materials Science Engineering, Vol. 87, pp. 281,
1987
20) F.H. Stott, Materials Science Forum, Vol. 251-254, pp. 19-32, 1997
57
21) T.F. An, et al., Oxidation of Metals, Vol. 54, pp. 301-316, 2000
22) S. Xiaofeng, et al., International Journal of Materials and Product
Technology,
23) B.A. Pint, Surface and Coatings Technology, Vol. 188-189, pp. 71-78, 2004
24) B.A. Pint, et al., Materials Science and Engineering, Vol. A245, pp. 201-211,
1998
25) B.A. Pint, Oxidation of Metals, Vol. 49, pp. 531–560, 1998
26) B.A. Pint, et al., Materials at High Temperature, Vol. 17, pp. 165–171, 2000
27) J.A. Haynes, et al., Materials at High Temperature, Vol. 21, pp. 87-94, 2004
28) Y.S. Touloukian, et al., Thermophysical Properties of Matter, Vol. 13, pp.
176, 1977
29) J.R. Nicholls, Materials at High Temperature, Vol. 14, pp. 289-306, 1997
58
CHAPTER 4:
ROLE OF GRAIN SIZE ON THE OXIDATION RESISTANCE
OF β-NiAl
T. Brammer, M. Akinc and M.J. Kramer
Ames Laboratory and Department of Materials Science and Engineering
Iowa State University, Ames, IA 50011
(To be submitted to Intermetallics)
Abstract
It has been shown in previous work that hafnium and certain platinum group
metals (PGMs) have the ability to improve the oxidation resistance of NiAl alloys.
This study focuses on the role that each of these modifications play in grain size
of the alloys and how grain size can affect the oxidation performance of the
alloys. It was seen that iridium and rhodium additions constrain the grain size
during casting by forming PGM rich dendrites. The size of the dendrites then
limits the average grain size of the alloy.
During annealing at 1300°C, the
dendrites diffuse during the first few hours and the grains start to grow at a
similar rate compared to an unmodified NiAl alloy. It is therefore understood that
PGMs play only a slight role in hindering grain growth.
However, the grain
growth of alloys containing 0.05 at% hafnium is severely diminished.
It is
believed to be due to hafnium acting as pinning agents at the grain boundaries
although this could not be verified due to the small quantity of Hf.
59
Introduction
Nickel aluminide, β-NiAl is of great interest for high temperature applications due
to its combination of low density (5.86 g/cm3), high melting point (1638°C), high
modulus (240 GPa), high thermal conductivity (76 W/mK), and in particular, its
excellent oxidation resistance (1,2,3). β-NiAl forms a protective Al2O3 scale upon
exposure to oxidizing atmosphere at high temperature. However, the Al2O3 scale
formed on cast β-NiAl alloy can suffer from thermal spallation due to poor
adhesion between the scale and the substrate (4). Previous studies by Yang et
al. concerning isothermal oxidation at 1000°C of ca st NiAl alloys and a sputtered
NiAl coating showed that the sample with the smaller grain size restricted void
formation at the scale/substrate interface which was more adherent (5,6). The
NiAl sample with the smaller grain size accelerated the phase transformation
from metastable θ-Al2O3 scale, which has a high growth rate, to the more stable
α-Al2O3. This accelerated phase transformation is believed to be due to large
number of nucleation sites brought about by the large number of grain
boundaries (7).
Previous work done on the oxidation resistance of NiAl alloys showed that some
PGM substitutions along with minute additions of hafnium improved the oxidation
resistance at high temperatures (8).
These modifications were shown to
decrease the growth rate of the oxide scale during isothermal oxidation testing at
1150 and 1200°C. The modifications also formed an oxide layer that showed
improved adhesion of the scale to the substrate in the temperature range tested
up to 1300°C. The oxides that formed were thinner, more planar, and more
continuous along the alloy surface.
It was desired to know how these
modifications were able to improve the oxidation resistance. Grain size analysis
of the alloys was completed to determine if differences in grain size could
account for the improved oxidation resistance.
60
In this article, we report the role that synthesis technique and chemical
modification of β-NiAl (namely PGM substitutions and hafnium additions) play on
the grain size of β-NiAl alloys. Furthermore, we studied the possible correlation
between the grain size of the alloys and the oxidation resistance observed. It is
primarily desired to understand if grain size plays a role in the oxidation
resistance of these alloys.
If it does, can the improvement in the oxidation
resistance be accounted for solely due to this effect?
Experimental Procedure
Materials
In order to isolate the effect of chemical modification from the role of grain size,
two samples having the same chemical composition but different grain sizes
were prepared. For melt-spun samples with a fine grain size, a NiAl button was
produced by arc-melting nickel and aluminum chunks in an argon atmosphere on
a copper hearth. The alloy was placed into a graphite crucible which is
transferred to a melt-spinner with a Pillar induction heating coil and the entire
system was heated up to 1720°C in He atmosphere at 16 kPa. The molten NiAl
was injected onto the edge of a chilled copper disk with a wheel speed of 20 m/s.
The NiAl ribbon was then collected and annealed in an inert atmosphere at
1300°C for 6 hours. Strips of the ribbon were then used for oxidation testing.
The cast alloys were produced from Al and Ni pieces cut from bulk metal sheets
obtained from the Materials Preparation Center at Ames Laboratory, having a
purity of 99.7% or more. Pure Ni and Al were arc-melted together in the desired
ratio on a chilled copper hearth in an argon atmosphere using a non-consumable
tungsten electrode to form β-NiAl alloy. The alloying additions were then melted
along with the β-NiAl synthesized according to the aforementioned steps. The
nominal compositions were prepared according to Ni50-xAl50PGMx (x=0, 3, 6, and
9 at%) with the PGMs being either Ir, Rh, or Pd. The alloys containing hafnium
61
were produced using pieces of Hf modified Ni sheet metal instead of pure Ni.
After initial arc-melting, the samples were turned over and re-melted a minimum
of three times in an argon atmosphere to achieve better homogenization before
drop-casting. After re-melting the samples were drop-cast into 10 mm-diameter
cylindrical rods. They were then annealed in an inert atmosphere at 1300°C for
varying time intervals (t=0, 6, 40, 70, and 106 hours).
Characterization
The cast and melt-spun alloys were annealed for 6 hours at 1300°C before
oxidation runs. Isothermal oxidation at 1150 and 1200°C was carried out using a
high temperature oxidation testing rig fitted with a Mellen SC13R 1.25 inch
vertical tube furnace (Concord, New Hampshire) and a Cahn 2000 recording
electrobalance (Madison, Wisconsin) having an accuracy of 1 g. The sample
was suspended from the balance on a sapphire wire hook. Dry air was passed
through the system at a rate of 60 mL/min.
An Ionmaster fan was used to
remove the static electricity in the system to ensure a higher signal to noise ratio
in the measurements. The furnace was turned on and set to the desired dwelling
temperature using a heating rate of 15°C/min and th en held at the temperature
for 100 hours.
In this study an AmRay 1845FE scanning electron microscope (SEM) utilizing an
EDAX-TSL Delphi 2.5 Geneses model EBSD-EDX was used for orientation
imaging microscopy (OIM). Epoxy mounted samples were polished according to
standard metallurgical methods followed by electro-polishing for 8 seconds in a
solution of 1 part 70% nitric acid 2 parts methanol at room temperature at 30
volts to remove smearing effects of mechanical polishing on the sample surface,
as well as etch the alloy to bring out the grain boundary relief. The specimen
was then lightly polished with colloidal silica to remove any adherent reaction
products. The specimens were analyzed at a 70 º from horizontal geometry. The
62
resulting reflected Kikuchi patterns were analyzed using TSL version 5.02
software (TexSEM Laboratories, Inc.
Draper, UT).
A hexagonal pixel step
pattern was used with step size small enough to assure at least 10 pixels per
grain at a rate of 4 patterns per second. The resulting solutions and Hough filter
solutions were recorded at each pixel for later evaluation. Average grain size
was measured using the TSL software by estimating an ellipse to best fit each
grain.
Results and Discussion
Before oxidation testing, the average grain size along with standard deviation
was determined for both the NiAl alloys. The results are given below in Table 1.
Table 1: Average grain size of NiAl
Synthesis
Grain Size ( m)
St. Dev.
Cast
635
54
Melt-spun
126
9
Table 1 shows a significant difference in grain size of these samples having been
annealed at 1300°C for 6 hours. To ensure that the melt-spun ribbons were
single phase and did not form any undesired phases due to the rapid cooling, xray diffraction patterns were obtained for melt-spun and cast alloys after the 6
hours of annealing. XRD patterns are shown in Figure 1. The pattern from the
melt-spun alloy was averaged between the free side and the wheel side of the
ribbon. Both samples exhibit single phase β-NiAl with no unidentified peaks.
63
Figure 1: X-ray analysis of cast and melt-spun NiAl
Parabolic growth rate constants of cast and melt-spun NiAl at 1150°C and
1200°C are calculated and presented in Figure 2.
T he mass change data
between 50 and 100 hours was used to ensure the transition from θ-Al2O3 to αAl2O3 had completed and was not contributing to irregular mass changes since θAl2O3 grows faster than α-Al2O3 (9-12). All of the oxidation curves fit a parabola
with a line fitting R2 value of 0.97 or better.
The graphs are shown as a
logarithmic scale starting at 1 and decreasing along the y-axis. Therefore, the
taller bars in the chart have a slower growing oxide. It is noticed that the smaller
grained specimen formed a slower growing oxide when compared to the larger
grained specimen.
64
1.E-06
2
-4 -1
Kp (mg cm s )
1.E-07
1.E-05
1.E-04
Figure 2: The calculated oxide growth rates between the hours of 50-100 for
both cast and melt spun NiAl at 1150°C and 1200°C
In Figure 2 the melt spun samples had a slower growing oxide scale than the
cast samples.
As expected based upon diffusion principles the same alloys
subjected to the lower temperature showed a slower growing scale when
compared to testing at the higher temperature. Therefore, the same alloy with no
compositional change shows a consistent trend in oxide growth rate dependent
upon the synthesis technique. Finer grain size has been shown to increase the
aluminum diffusion to the oxide metal interface (13). This allows for fast shortcircuit diffusion of aluminum and creates an alumina scale that is compact and
therefore reduces further growth rate of the oxide scale.
The mechanism by which the improvement in oxidation resistance brought about
by iridium, rhodium, and hafnium addition is unclear (8). In order to assess the
role of Pd, Ir, and Rh substitution, the grain size of the PGM substituted cast and
annealed samples were analyzed by OIM employing TSL software. The average
grain sizes are displayed in Figure 3a. Figure 3b shows the role of Ir and Rh
additions on the grain size the alloy on a finer scale.
65
Grain Size ( m) .
2500
NiAl
Rh
Ir
Pd
2000
1500
1000
500
0
0
2
4
6
8
10
At% PGM
300
Rh
Ir
Grain Size ( m)
.
250
200
150
100
50
0
0
2
4
6
8
10
At% PGM
Figure 3: Average grain sizes for (a) NiAl and PGM modified alloys (b) Ir and Rh
containing alloys
66
In Figure 3a, it can be seen that compared to the baseline NiAl, the palladium
addition increases the average grain size at 3% but the grain size decreases with
more Pd addition. For Ir and Rh, substitution of these elements decreases the
grain size. It seems Rh is more effective at lower concentrations than Ir but at the
highest concentration of 9 at% Ir is more effective than Rh (120 vs. 160
m).
Images obtained from NiAl and the 3% PGM modified samples during OIM
analysis are displayed in Figure 4.
67
Figure 4: Images captured during OIM analysis of (a) NiAl, (b) Ni-50Al-3Rh, (c)
Ni-50Al-3Ir, (d) Ni-50Al-3Pd
Figure 4a shows anisotropic grains that are several mm long and approximately
0.2 mm wide that seem to radiate inward to the center of the casting which is
towards the bottom left-hand corner of the image. Figure 4b shows the cast NiAl
alloy containing 3% rhodium. Compared to the baseline NiAl, the grains are
much smaller and isotropic in shape.
containing 3% iridium.
Figure 4c shows the cast NiAl alloy
Here the grains are smaller than NiAl and as in Rh
substitution, they are isotropic in shape. Figure 4d shows the 3% palladium
modified NiAl alloy in which the grains seem to radiate inward to the center of the
casting which is in the bottom left-hand side of the image. The grains appear to
be similar in length to the NiAl cast alloy, but instead of being anisotropic with
high aspect ratio, these grains show lower aspect ratio with a large average grain
size. All of the alloys show random grain orientation as indicated by the stark
contrast differences between neighboring grains.
The columnar grains that are indicative of NiAl as well as the Pd substituted
alloys are created due to the heat flow of the alloy and the hottest part of the
system being the melt. Therefore the crystals grow inward from the interface of
the mold with the crystals that are most favorably lined up with the direction heat
flow growing faster.
The equiaxed grains formed by the Ir and Rh substituted
alloys are created by the heat flow of the system where the crystals are the
hottest part. The crystals reject heat and solute during solidification. Therefore
all the grains are equiaxed and they should be smaller due to the higher
nucleation rate in an undercooled system. These two different types of heat
flows cause the change between the grain structures seen in the different alloy
systems.
68
Figure 5: Solid/Liquid interface morphology and temperature distribution (14)
Confirmation of this mechanism that causes different grain sizes and shapes in
the alloys upon casting was found during SEM analysis of the alloys. The SEM
backscattered image of Ni-50Al-9Ir is shown in Figure 6 after a colloidal silica
etching. It is clear from this low magnification image that equiaxed dendrites are
forming during the casting process and these dendrites limit the size of the
grains. Discrete looking dendrites as shown in Figure 6a were only observed in
the as-cast samples of iridium and rhodium containing alloys. No dendrites were
observed in the palladium containing alloys which showed to have large grains.
As it is shown in Figure 6b, all of the dendrites have diffused within six hours of
annealing with only small residual chemical heterogeneity remaining as indicated
by the slightly brighter and darker patches.
Therefore the dendrites formed
during casting were not thermodynamically stable and they diffused during
annealing to reduce the chemical heterogeneity.
69
Figure 6: SEM backscattered images of (a) as-cast Ni-50Al-9Ir, (b) Ni-50Al-9Ir
annealed for 6 hours at 1300°C
The composition at the center of the dendrites was determined using EDS to
contain approximately 20 at% Ir. The surrounding darker regions were found to
contain about 2 at% Ir. Since dendrites were forming and the compositional
heterogeneity was relatively large, x-ray analysis was carried out on this sample
to ensure that it was single phase. The x-ray analysis results of as-cast and
annealed 9 at% Ir samples are shown in Figure 7.
70
Intensity (arb)
As-cast
6hr anneal
20
40
60
80
100
120
140
Angle (2θ)
Figure 7: X-ray analysis of as-cast and annealed Ni-50Al-9Ir
Both the as-cast and annealed samples appear to be single phase according to
the x-ray data obtained. Therefore the dendrites shown in Figure 6a and 6b,
although they are rich in iridium, remain continuous in crystal structure
throughout the sample.
The x-ray pattern for the as-cast sample appears to have broader peaks than the
annealed sample. The increased iridium content in the dendritic regions would be
expected to strain the lattice and cause the lattice parameter increase to account
for the larger atom size. Iridium, which has an atomic radius of 180 pm, is
substituted on a nickel site which has an atomic radius of 149 pm (15). This
increase in the atomic size would also increase the lattice parameters and would
cause the Bragg diffraction angle to widen as seen in Figure 7. The measured
lattice parameters for PGM substituted NiAl alloys are shown in Figure 8. The xray data for each alloy was obtained and Rietveld refinement was used to
71
determine the lattice parameter of each alloy. The measured lattice parameter
increases with increasing PGM substitution. The two PGM substitutions that
increase the lattice parameter the most are iridium and rhodium and they are
also the two PGMs that cause the formation of the dendrites.
Lattice Parameter (Å)
2.93
2.92
2.91
X=Rh
X=Ir
2.90
NiAl
X=Pd
2.89
3
6
9
Composition (at% X)
Figure 8: Measured lattice parameters for PGM substituted NiAl alloys
The possible solidification process causing the Ir rich dendrites to form is due to
the increased melting temperature of Ir substituted NiAl.
The first metal to
solidify in the casting would be the metal containing the highest amount of Ir.
The next enveloping layer to solidify would be slightly lower in Ir. This would
continue to have layers of less and less Ir until the last molten metal solidified.
This solidification process can be seen in Figure 9 and it explains how the
dendrite lengths can limit the average grain size of the alloy. Both Ir and Rh
substituted NiAl alloys show these dendrites.
72
Figure 9: Schematic representation of dendrite formation during solidification
(16) (a) nucleation of dendrites, (b) growth of dendrites, (c) complete
solidification, (d) final grain size constrained by dendrites
The discrete and vivid looking PGM rich dendrites were not observed in any of
the annealed samples. It is inferred that the dendrites diffuse during the 6 hours
of annealing at 1300°C in order to equilibrate the composition and approach
thermodynamic equilibrium as indicated in Figure 5c and 5d. Given that the
dendrites diffuse during annealing and the dendrites limit the size of the grains
during casting, annealing would possibly result in significant grain growth.
Samples of as-cast and annealed NiAl, Ni-50Al-9Ir, and Ni-50Al-9Ir+Hf were
analyzed using OIM to determine if the grains coarsened during annealing at
1300°C. The average grain sizes of the alloys duri ng the annealing test are
displayed in Figure 10.
73
1350
Average Grain Size ( m) .
.
9%Ir
1200
9%Ir+Hf
1050
NiAl
900
750
600
450
300
150
0
0
20
40
60
80
Annealing Time (hours)
100
120
Figure 10: Grain growth of NiAl, Ni-50Al-9Ir and Ni-50Al-9Ir+Hf during annealing
at 1300°C
It is clear in Figure 10 that the grains of the cast alloy NiAl as well as the alloy
containing iridium grew readily at 1300°C. Therefo re, although Ir substitution
reduces the average grain size during casting it does little to stop the grain
growth during annealing.
Although coarse grains are desired for high
temperature creep resistance (17), they were shown in Figure 2 to increase the
oxide growth rate. The alloy containing hafnium on the other hand effectively
hindered the grain growth at 1300°C. Not even smal l amounts of dispersed
alumina particles are able to stop grain growth in NiAl (18). It is believed that the
hafnium is preferentially segregated to the grain boundaries where it can act as a
Zener pinning agent (19, 20). Within the current scope of the study, presence of
such a small quantity of Hf at the grain boundaries could not be ascertained.
OIM images of the grains for both of these alloys were captured after 6 and 106
hours of annealing. They are shown in Figure 11 to demonstrate the retarding
effect of 0.05 at% hafnium on the grain growth. After 6 hours of annealing, both
alloys have very similar grain sizes. After an additional 100 hours of annealing at
74
1300°C, the average grain size area of the alloy wi thout Hf grew by 40-fold while
the alloy with Hf grew in average grain size area by only 3-fold.
Figure 11: OIM images of the grains of (a) Ni-50Al-9Ir annealed for 6 hours, (b)
Ni-50Al-9Ir+Hf annealed for 6 hours, (c) Ni-50Al-9Ir annealed for 106 hours, (d)
Ni-50Al-9Ir+Hf annealed for 106 hours
75
Assuming the grain growth shown in Figures 10 and 11 are indicative of ideal
grain growth, the data was used to calculate the temperature dependent grain
growth constant which is given by:
2
d 2 − d o = kt
where k is the temperature dependent growth constant, d is the grain size at time
t, and do is the initial grain size.
The grain growth constants for the three alloys shown in Figure 10 are shown
below in Table 2.
Table 2: Temperature dependent grain growth constants
k ( m2/hr)
R2
Ni-50Al
10,528
0.999
Ni-50Al-9Ir
6,198
0.986
327
0.957
Alloy
Ni-50Al-9Ir+Hf
Conclusions
Comparing the same composition of an as-cast and melt-spun sample, the meltspun sample had a much finer grain size and produced a more compact slower
growing oxide scale which correlates well with Yang’s work (6). It was shown in
Figure 10 that the average grain size of an unmodified NiAl alloy increases
drastically during annealing at 1300°C.
Of the PGM modifications to NiAl, both the Ir and Rh additions constrained the
grain size by forming PGM rich dendrites during the casting process. Both Ir and
Rh have been noted to improve the oxidation resistance of NiAl by forming a
slower growing and more adherent oxide scale (8). The PGM rich dendrites
76
were still single phase with the rest of the alloy, but they did cause slight
broadening of the peaks in the x-ray diffraction plot. Pd modifications did not
form dendrites and did not show any improvement in oxidation resistance when
compared to NiAl (8). The PGM rich dendrites diffused after short annealing
times at 1300°C. As the dendrites diffused, the gr ains grew at a similar rate to
the unmodified NiAl alloy indicating that they do little to slow grain boundary
movement. It is unsure at this point whether Ir and Rh substitutions only improve
the oxidation resistance due to the grain size role or if there are other effects that
also play a factor in improved oxidation resistance.
Adding 0.05 at% Hf to the alloys effectively hindered grain growth at 1300°C. It
is believed that Hf segregates to the grain boundaries and effectively pins the
grain boundaries so that grain coarsening is severely diminished. Hf has shown
to improve the oxidation resistance of NiAl alloys by forming a slower growing
and more adherent oxide scale (8).
Acknowledgement
This work was supported by the DOE-FE (AMR program) through Ames
Laboratory contract no. DE-AC02-07CH11358 through Iowa State University.
References
1) R.D. Noebe, R.R. Bowman, M.V. Nathal, Review of the physical and
mechanical properties of the B2 compound NiAl, International Materials Review,
Vol. 38, pp. 193-232, 1993
2) D.B. Miracle, R. Darolia, Intermetallic Compounds: Principles and Practice,
Vol. 2, Chichester: John Wiley & Sons, pp. 55-74, 1995
3) C.T. Liu, J.O. Stiegler, F.H. Froes, Metals handbook (10th Ed.), Vol. 2, ASMInternational, Metals Park, pp. 913–942, 1990
77
4) J.L Smialek, Oxide morphology and spalling model for NiAl, Metallurgical
Transactions A, Vol. 9, No. 3, pp. 309-320, 1978
5) S.L. Yang, F.H. Wang, Y. Niu, W.T. Wu: Isothermal oxidation of ß-NiAl alloy
and sputtered coating at 1000ºC in air. Material Science Forum, Vol. 369-372,
pp. 361-368, 2001
6) S.L. Yang, F.H. Wang, W.T. Wu, Effect of microcrystallization on the cyclic
oxidation behavior of β-NiAl intermetallics at 1000 °C in air, Intermetall ics, Vol. 9,
pp. 741-744, 2001
7) M.W. Brumm, H.J. Grabke, The oxidation behaviour of NiAl – Phase
transformations in the alumina scale during oxidation of NiAl and NiAl-Cr alloys,
Corrosion Science, Vol. 33, No. 11, pp. 1667-1690, 1992
8) T. Brammer, M. Akinc, M.J. Kramer, Effect of platinum group metal
substitutions and hafnium additions on the oxidation resistance of β-NiAl, to be
submitted to Intermetallics, 2011
9) J. Balmain, M.K. Loudjani, A.M. Huntz, Microstructural and diffusional aspects
of the growth of alumina scales on β-NiAl, Materials Science and Engineering A,
Vol. A224, pp. 87-100, 1997
10) Y. Cadoret, D. Monceau, M.P. Bacos, P. Josso, V. Maurice, P. Marcus,
Effect of platinum on the growth rate of the oxide scale formed on cast nickel
aluminide intermetallic alloys, Oxidation of Metals, Vol. 64, No. 3-4, 2005
11) T.F. An, H.R. Guan, X.F. Sun, Z.O. Hu, Effect of the θ-α-Al2O3 transformation
in scales on the oxidation behavior of a nickel-base superalloy with an aluminide
diffusion coating, Oxidation of Metals, Vol. 54, No. 3-4, pp. 301-316, 2000
12) M. Li, X.F. Sun, H.R. Guan, X. Jiang, Z.O. Hu, Oxidation behavior of Pdmodified aluminide coating at high temperature, Journal of Material Science and
Technology, Vol. 19, No. 3, 2003
13) G.C. Rybicki, J.L. Smialek, Effect of the θ to α-Al2O3 transformation on the
oxidation behavior of ß-NiAl+Zr, Oxidation of Metals, Vol. 31, pp. 275-304, 1989
78
14) W. Kurz, D.J. Fisher, Fundamentals of Solidification, Trans Tech
Publications, USA, pp. 11, 1984
15) E. Clementi, D.L. Raimondi, W.P. Reinhardt, Journal of Chemical Physics,
Vol. 47, pp. 1300, 1967
16) K.R. Trethewey and J. Chamberlain, Corrosion for Students of Science and
Engineering (2nd Ed.), John Wiley Inc., New York, pp. 23-45, 1995
17) J.D. Whittenberger, R. Ray, S.C. Jha, Influence of grain size on the creep
behavior or HfC – dispersed NiAl, Materials Science and Engineering, Vol. A151,
pp. 137-146, 1992
18) M.M. Moshksar, H. Doty, R. Abbaschian, Grain growth in NiAl – Al203 in situ
composites, Intermetallics, Vol. 5, Issue 5, pp. 393-399, 1997
19) B.A. Pint, Experimental observations in support of the dynamic-segregation
theory to explain the reactive-element effect, Oxidation of Metals, Vol. 45, pp. 137, 1996
20) M.A. Bestor, R.L. Martens, R.A. Holler, M.L. Weaver, Influences of annealing
and hafnium concentration on the microstructures of sputter deposited β-NiAl
coatings on superalloy substrates, Intermetallics, Vol. 18, pp. 2159-2168, 2010
79
CHAPTER 5:
EFFECT OF PLATINUM GROUP METALS ON THE
MELTING TEMPERATURE OF β-NiAl
T. Brammer, M. Akinc and M.J. Kramer
Ames Laboratory and Department of Materials Science and Engineering
Iowa State University, Ames, IA-50011
(To be submitted to Intermetallics)
Abstract
The Rose-Ferrante relation was used to calculate the melting temperatures of
platinum group metal (PGM) substituted β-NiAl.
Experimental testing was
completed of the same alloy systems and there is agreement between the
calculated and experimental results. Iridium and rhodium substitutions increase
the melting temperature of β-NiAl while palladium substitutions decrease overall
the melting temperature. At 12% substitution amounts, iridium increases the
melting temperature of NiAl by 150°C, rhodium by 13 0°C, and palladium
decreases it by 50°C. The melting temperature tren ds directly correlate to the
oxidation resistance observed in previous studies (1,2). This seems to indicate
that the alloy with the lower homologous melting temperature performs better
during high temperature oxidation testing.
80
Introduction
β-NiAl has been studied as a good candidate for implementation in high
temperature environments (3). The alloy has good inherent oxidation resistance
at high temperatures by forming a protective alumina scale (4).
This scale
shields the underlying metal from further oxidation. NiAl can also be used at high
temperatures due to its high melting temperature of 1638°C (5). NiAl based
alloys could be used at even higher temperatures if the alloy were modified to
increase the melting temperature.
The elemental modification to NiAl must
create a solid solution with NiAl to increase the melting temperature while not
adversely affecting the oxidation resistance.
It has been noticed in previous work that certain PGM substitutions to β-NiAl
improve the oxidation resistance at high temperatures (1). It was desired to
understand if the PGM substitutions had an effect on the melting temperature of
the alloy and determine if the melting temperatures directly correlated to the
oxidation resistance of the alloy. Given the same conditions, an alloy with a
higher melting temperature would have a lower homologous melting temperature.
Having a lower homologous melting temperature would ensure that the alloy
would have reduced grain growth at the high temperatures, and grain size has
been directly correlated to oxidation resistance (6).
Experimental Procedure
Materials
The alloys were produced from pieces cut from pure bulk metal sheets obtained
from the Materials Preparation Center at Ames Laboratory, having a purity of
99.7% or more. Pure Ni and Al were first arc-melted together on a chilled copper
hearth in an argon atmosphere using a non-consumable tungsten electrode.
This initial melting ensured that β-NiAl would form. The alloying additions were
then melted along with the β-NiAl synthesized according to the aforementioned
81
steps. The nominal compositions were prepared according to Ni50-xAl50PGMx
(x=0, 3, 6, 9, and 12 at%). After initial arc-melting, the samples were turned over
and re-melted a minimum of three times in an argon atmosphere to achieve
better homogenization before drop-casting. After re-melting the samples were
drop-cast into 10mm diameter cylinders.
These cylindrical rods were then
annealed in an inert atmosphere at 1300°C for 6 hou rs to achieve a better
compositional homogeneity. A small sample size of about 0.2 grams was cut
from the cylinder to be used for melting temperature studies.
Testing
Melting temperature testing was carried out using a Linseis L81/2400 high
temperature combined differential thermal analysis (DTA) and thermogravimetric
analysis (TGA) instrument (Princeton, NJ). Argon was allowed to flow through
the system at a rate of 4 L/hr to ensure no oxygen build up during testing. The
sample was placed in a tungsten crucible with an alumina liner. The furnace
uses a temperature ramping rate of 15°C/minute. Th e instrument uses C-type
thermocouple wires and transmits the voltage difference between the sample and
an empty crucible to a program on the computer. The data is then read using a
software program called STA Measurement.
The melting temperature was
calculated based upon the average of the onset of melting and crystallization.
Results and Discussion
The melting temperature of an alloy can be calculated with reasonable accuracy
(14.5%) using the Rose-Ferrante relation (7,8).
The Rose-Ferrante relation
utilizes the universal binding energy theory to create a universal model for
calculating melting temperatures.
The model directly correlates the melting
temperature of an alloy to the cohesive energy of the alloy. This relationship is
shown in Equation 1.
82
Rose-Ferrante Relation for Pure Metal:
Tm = 0.03 E
c
kB
(Eq. 1)
Where Tm=melting temperature, Ec=cohesive energy, kB=Boltzmann’s constant
The prefactor 0.03 is used in the case of pure metals. For binary compounds
where both metals have the same crystal structure, the prefactor changes to
0.0302. This small change can account for switching from a single metal system
to a binary metal system. It is therefore assumed that it will not change much for
ternary alloys with the same crystal structure either. Therefore it is assumed that
the Rose-Ferrante relation is a decent model for estimating melting
temperatures.
The calculated melting temperatures were based upon cohesive energy of the
alloy as shown in Equation 1. The cohesive energy of an alloy is the energy
required to break the atoms of the solid into isolated atomic species (9). The
cohesive energy of the alloy can therefore be thought of as the total energy of the
bonds within the alloy. The cohesive energy of the alloy is therefore related to
the individual cohesive energies of the pure metals and the formation enthalpy of
the alloy (10). This relationship is displayed in Equation 2.
Cohesive Energy of Ternary Metal Alloy:
c
c
c
E c = x1E1 + x2 E2 + x3 E3 − H f
(Eq. 2)
where Ec= cohesive energy of the alloy, x1=atomic fraction of metal 1,
E1c=cohesive energy of metal 1, Hf=formation enthalpy of alloy
The cohesive energies of the individual metals were found in C. Kittle’s book (9)
and they are reproduced in Table 1 for comparison. Since the PGMs are being
83
substituted onto the nickel site, their cohesive energies would be replacing Ni in
the amount of the desired substitution. If the PGM has a higher cohesive energy
than nickel, the cohesive energy of the overall alloy would increase and therefore
the calculated melting temperature would also increase. Iridium and rhodium
have cohesive energies that are higher than nickel, while palladium has a lower
cohesive energy.
Table 1: Cohesive Energies of Pure Metals
Metal
Cohesive Energy (kJ/mol)
Al
Ni
Rh
Ir
Pd
327
428
554
670
376
The other constituent needing to be accounted for before the melting
temperature can be calculated is the formation enthalpy of the alloy.
The
formation enthalpies of PGM substituted NiAl alloys with nominal compositions of
Ni50-xAl50PGMx (x = 0, 2, 4, 7, and 11 at%) were calculated using an extended
Miedema model (10). The formation enthalpy of the alloys is shown in Figure 1.
As it can be seen, all of the PGM substitutions create a NiAl alloy with a more
negative formation enthalpy. The higher the amount of PGM substitution, the
more negative the formation enthalpy of the alloy becomes. Since there is a
negative sign in front of the formation enthalpy constituent in Equation 2, a more
negative formation enthalpy value would denote an alloy with a higher melting
temperature. Of the PGM substitutions, the one giving rise to the most negative
formation enthalpy is rhodium followed by iridium and finally palladium.
84
-60
NiAl
-62
Ηf (kJ/mol)
-64
-66
-68
X=Pd
-70
X=Ir
-72
X=Rh
-74
0
2
4
6
8
10
12
Atomic% X
Figure 1: Calculated formation enthalpies for PGM substituted NiAl alloys
Now combining the data from the individual cohesive energies and the calculated
formation enthalpies, the calculated melting temperature can be determined for
each of the desired PGM substituted NiAl alloys. The results are displayed in
Figure 2.
X=Ir
1840
Calculated Tm (°C)
1820
1800
X=Rh
1780
1760
1740
1720
X=Pd
1700
NiAl
1680
0
2
4
6
8
10
12
Atomic% X
Figure 2: Calculated melting temperatures for PGM substituted NiAl alloys
85
Figure 2 shows iridium to have the highest calculated melting temperature based
upon the Rose-Ferrante relation. Iridium is the highest because it has a much
higher cohesive energy than the nickel atoms it’s substituting for. Also iridium
substituted alloys were calculated to have formation enthalpies which are more
negative than binary NiAl.
Both of these constituents indicate that iridium
substitutions will increase the melting temperature of NiAl. Rhodium likewise
should produce an alloy with an increased melting temperature when compared
to NiAl.
Conversely, substituting palladium for nickel in NiAl is expected to
decrease the overall melting temperature of the alloy. This reduction in melting
temperature is primarily due to the lower cohesive energy of palladium when
compared to the nickel it’s substituted for.
The experimental melting temperatures of the alloys were determined using a
high-temperature DTA.
The resulting melting temperatures from each of the
produced alloys are displayed in Figure 3. As it can be seen in the figure, the
model used for calculations correctly predicted the general melting temperature
trends when compared to the calculated melting temperatures in Figure 2.
Substituting iridium creates the alloy with the highest melting temperature
showing an increase from the baseline NiAl of 150°C at 12 at% substitution.
Substituting rhodium increases the melting temperature about 130°C. Higher
substitution amounts of iridium and rhodium continue to increase the melting
temperature of the alloy. Palladium substitutions decrease the overall melting
temperature of the NiAl alloy with higher substitution amounts continuing to
decrease the melting temperature. At 12 at% substitution, palladium decreased
the melting temperature of NiAl by about 50°C.
86
1800
X=Ir
Experimental Tm (°C)
1780
1760
X=Rh
1740
1720
1700
1680
1660
1640
NiAl
1620
X=Pd
1600
0
2
4
6
8
10
12
14
Atomic% X
Figure 3: Experimental melting temperatures for PGM substituted NiAl alloys
One noticeable difference is that all of the experimental melting points seem to
be about 50°C lower than the calculated melting tem peratures which are within
the 14.5% stated accuracy of the Rose-Ferrante model. This discrepancy in the
calculations could arise due to the fact that an extended Miedema model was
used instead of the traditional binary Miedema model.
As it was shown in
Equation 1, the prefactor changes slightly when switching from pure metals to
binary systems. The change in this prefactor from binary systems to ternary
solutions has not been documented and could not properly be accounted for.
Conclusions
The calculations based on the Rose-Ferrante model produced relatively accurate
melting temperatures that correlated closely to the measured melting
temperatures. Substituting iridium and rhodium into the NiAl alloy increases the
overall melting temperature of the alloy.
The more substitution amount, the
higher the melting point of the alloy. NiAl substituted with 12 at% Ir showed a Tm
gain of 150°C and 12% Rh substitution showed a T m gain of 130°C. Palladium
87
substitutions decreased the overall melting temperature of the alloy by 50°C at
12 at% substitution.
Substitutions that increase the overall melting temperature of the alloy also
decrease the homologous melting temperature for that alloy. Decreasing the
homologous melting temperature would lead to reduced grain growth at high
temperatures, and grain growth has been shown to affect the oxidation
resistance of these alloys (6).
Acknowledgement
This work was supported by the DOE-FE (AMR program) through Ames
Laboratory contract no. DE-AC02-07CH11358 through Iowa State University.
References
1) P.K. Ray, et al., Journal of Metals, Vol. 62, pp. 25–29, 2010
2) T. Brammer, et al., Oxidation Paper to be communicated to Corrosion Science
Journal
3) D.B. Miracle, R. Darolia, Intermetallic Compounds, Vol. 3, pp. 55-74, 2000
4) H.J. Grabke, M. Schutze, Oxidation of Intermetallics, pp. 79-84, 1998
5) C.T. Liu, J.O. Stiegler, Science, Vol. 226, pp. 636-642, 1984
6) T. Brammer, et al., Grain Size Paper to be communicated to Intermetallics
7) P. Wu, et al., Journal of Physics and Chemistry of Solids, Vol. 64, pp. 201-212,
2003
8) J.H. Rose, et al., Physics Review Letter, Vol. 47, pp. 675, 1981
9) C. Kittel, Introduction to Solid State Physics (7th Ed.), pp. 55-57, 2001
10) P.K. Ray, et al., Journal of Alloys and Compounds, Vol. 489, pp. 357-361,
2010
88
CHAPTER 6:
GENERAL CONCLUSIONS
Alloys based on NiAl offer significant potential payoffs as structural materials in
gas turbine applications due to a unique range of physical and mechanical
properties.
Alloying additions to NiAl could be used to further improve the
pertinent properties that currently limit this system from replacing Ni based
superalloys. Modifications to NiAl were explored to increase the phase stability
and oxidation resistance which would allow these alloys to be used at even
higher temperatures yielding greater efficiencies.
The extended Miedema model was an effective tool that screened all of the
potential phase space for ternary substitutions to NiAl and found the few potential
systems worth further investigation.
After production of the alloys it was
determined that Ir, Rh, and Pd were the top candidates for substitution on Ni site
up to 12 at%.
The melting temperature of NiAl could be increased as much as 150°C with 12
at% Ir and 130°C with 12 at% Rh substitution. Pall adium on the other hand
decreased the melting temperature by 50°C at the 12 at% substitution level.
The grain size was found to have a profound influence on the oxidation
resistance. Both Ir and Rh substitutions resulted in finer grain sizes compared to
Pd substitutions or base NiAl. The grain size increased drastically during high
temperature annealing with the PGM substitutions hindering grain growth only
slightly.
However, the addition of 0.05 at% Hf limited the grain growth
dramatically during high temperature annealing.
NiAl inherently has respectable oxidation resistance up to 1100°C. It was found
through experimental testing that both Ir and Rh substitutions improve the
89
oxidation resistance of NiAl at ultra-high temperatures with Ir performing the best.
Both PGM substitutions decreased the growth rate as well as forming a more
adherent oxide scale. Pd substitutions appeared to have a negligible effect to
the oxidation resistance of NiAl. Hafnium addition of 0.05 at% was found to
decrease the oxidation rate as well as increase the scale adherence.
The
combination of both Ir substitution (6-9 at%) and Hf addition (0.05 at%) produced
the alloy with the best oxidation resistance.
Although improvements in phase stability and oxidation resistance have been
made to the NiAl system, more development and testing are still needed. Two
major issues yet to be resolved are the low fracture toughness at ambient
temperatures and low creep resistance at elevated temperatures. Efforts are
underway to improve both of these properties by adding a second phase
refractory metal, namely molybdenum.
90
ACKNOWLEDGEMENTS
This work was performed at Ames Laboratory, which is operated for the US
Department of Energy (DOE) by Iowa State University. This work was supported
by the DOE-Fossil Energy (Advanced Materials Research Program) through
Ames Laboratory contract no. DEAC02-07CH11358.
Special thanks to my parents, David and Sharla Brammer, and to my fiancée,
Katie Brown, for their support and encouragement throughout my academic
career.
I would also like to thank Drs. Mufit Akinc and Matthew Kramer, my co-major
professors, for giving patient, insightful and thoughtful advice throughout my
research experience. This work could not be completed without their instruction
and criticisms. Their contribution to this work is greatly appreciated.
Thanks to my colleague Dr. Pratik Ray for useful discussions and advice
throughout the work. His efforts are greatly appreciated.
Thanks to Fran Laabs for discussions on the OIM grain size analysis work.
Thanks to Mattew Besser for providing zirconia and alumina crucible liners.
Thanks to Arne Swanson for his work in drop-casting the alloys.
Thank to Dr. Zhihong Tang for help during the cyclic oxidation tests.
Thanks to my lab mate Kevin Severs for providing a healthy atmosphere in the
labs as well as his criticism and discussions.
Thanks to other members of the ceramic processing research group for their
discussion.
Finally thanks to my graduate committee, Dr. Mufit Akinc, Dr. Matthew Kramer,
and Dr. Iver Anderson.