Reactive processing to form in-situ nickel aluminide microcomposites

MISSING IMAGE

Material Information

Title:
Reactive processing to form in-situ nickel aluminide microcomposites
Physical Description:
vii, 212 leaves : ill., photos ; 29 cm.
Language:
English
Creator:
Doty, Herbert William
Publication Date:

Subjects

Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1994.
Bibliography:
Includes bibliographical references (leaves 200-211).
Statement of Responsibility:
by Herbert William Doty.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002018865
notis - AKK6301
oclc - 32769189
System ID:
AA00003621:00001

Full Text










REACTIVE PROCESSING TO FORM IN-SITU NICKEL ALUMINIDE
MICROCOMPOSITES

















By

HERBERT WILLIAM DOTY


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1994
































Copyright 1994

by

Herbert William Doty













ACKNOWLEDGEMENTS

The completion of this work could not have been accomplished without the help,

support and encouragement of many individuals. The following need special mention for

their contributions. First I want to express my sincere gratitude to Dr. Reza Abbaschian, my

supervisory committee chairman, for his guidance throughout this work. I am especially

thankful for his ability to expand my vision when I became discouraged. I would also like to

thank the rest of my committee members, Dr. M.J. Kaufman, Dr. F. Ebrahimi, Dr. B.

Moudgil and Dr. A. Lush, for their critiques and valuable comments, in addition to

encouragement and advice.

I would also like to thank my colleagues, Dr. Leon Shaw, Atul Gokhale and Dennis

Gadonniex for our fruitful discussions on many aspects of this research. Special thanks are

given to Wayne Acree and Richard Crockett for hours of help with electron microscopy in

addition to helping me keep my sense of humor. Thanks also are due to Tony Gallucci for

preparation of the photographs for this manuscript.

Finally I am deeply indebted to my family for their love, encouragement and support,

especially my parents and my sons, Andy and Mike, who sacrificed so much during this time.

Lastly, I could not have survived without the inspiration, commitment and selfless support of

my loving wife, Cindy, who typed this manuscript and was a source of endless

encouragement.













TABLE OF CONTENTS

page

ACKNOWLEDGEMENTS ........................................ iii

ABSTRACT .................................................. vi

CHAPTERS

1 INTRODUCTION .................... ............... 1

1.1 Need for a High-Temperature Structural Material ............ 2
1.2 Intermetallics .................... .. .. ........... 5
1.3 Composites ..................................... 6
1.4 Summary ............ ........................... 7

2 BACKGROUND/LITERATURE REVIEW .................. .. 9

2.1 Physical and Chemical Properties of NiAl ................. 9
2.2 Mechanical Properties of NiAl ........................ 12
2.3 Fracture ...................................... 17
2.4 Toughening NiAl ................................. 18
2.5 Composites ................ .................... 22
2.6 Intermetallic Composite Fabrication ..................... 38
2.7 Self-Propagating High-Temperature Synthesis ................ 40
2.8 Development of Composites via RHC Processes .............. 49

3 EXPERIMENTAL PROCEDURE ......................... 52

3.1 Materials ............ ......................... 52
3.2 Oxidation .................. ......... ......... 53
3.3 Composite Fabrication .............................. 57
3.4 Differential Thermal Analysis ........................ 61
3.5 Annealing/Heat Treating ............................ 62
3.6 Analysis Techniques ......... ..................... 63
3.7 Fracture Toughness Measurement ...................... 67

4 RESULTS AND DISCUSSION: PROCESSING ................. 70

4.1 Oxidation .................................... 70
Oxidation of Nickel Powder ........ ................ 71







Oxidation of Aluminum Powder ................ .. .. ..78
Summary .................................. 92
4.2 Reactive Hot Compaction Process Development ........... 94
Differential Thermal Analysis ....... ............... 95
Hot Pressing Parameters ......................... 117
Summary .................. ................. 128
Discussion: Reaction Mechanisms .................. 129

5 CHARACTERIZATION OF THE COMPOSITES ............... 139

M icrostructure ............. ...................... 139
Transmission Electron Microscopy ....................... 155
Alumina Distribution .......... ...... ................ 156
Grain Growth .......... .................. ............ 156
Composition ...................... ............... 160
Fracture Toughness ................... ............... 167
Mechanical Behavior .......... ...... ................ 181
Fracture Surface Analysis ....... ......... ........... 183
Discussion: Toughening .......... ....... .............. 192

6 SUMMARY AND CONCLUSIONS ...................... 196

REFERENCES ........................ ................ ...... 200

BIOGRAPHICAL SKETCH ................... .................... 212














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


REACTIVE PROCESSING TO FORM IN-SITU NICKEL ALUMINIDE
MICROCOMPOSITES

By


Herbert W. Doty

August, 1994

Chairman: Dr. Reza Abbaschian
Major Department: Materials Science and Engineering

A process has been developed to produce alumina (A1203) reinforcements in-situ in

nickel aluminide (NiAl). The process involved reactively hot pressing pretreated elemental

nickel and aluminum powders to form an interconnected network of A1203 and an

interpenetrating NiAl matrix. Four-point bend tests of chevron-notched specimens indicate an

increase in room temperature fracture toughness with volume fraction of alumina up to a peak

of 14 MPaV/m at 18 volume percent, followed by a slight drop-off at 25 volume percent, the

highest level in the study.

The reaction sequence was studied via differential thermal analysis (DTA) and by

interrupting the process at various points throughout the cycle. The elemental powders were

found to form NiAl at temperatures well below the melting temperature of aluminum.

However, when oxidized powders were reacted, the aluminum melted prior to the initiation of

the intermetallic-formation reaction which was delayed until 8800C. It was found that







reacting the oxidized powders under pressure reverted the initiation of the intermetallic-

formation reaction to below the melting temperature of aluminum.

The low-temperature reaction was found to result in the development of a three

dimensional network of interconnected A1203 reinforcements in a NiAl matrix. Oxidizing the

Al powders prior to hot pressing resulted in the formation of a rod-like network of alumina,

while oxidizing the Ni powders produced a tubular alumina network. The oxidation of

aluminum was determined to be the critical step for the development of the proper alumina

morphology and interfacial properties necessary to improve the fracture toughness.

Specifically, the relative humidity in the oxidizing atmosphere controlled spallation from the

Al powders, reducing the occurrence of isolated alumina inclusions in the matrix which were

found to severely degrade fracture toughness, especially with weak NiAl-A1203 interfaces.

A physical model is presented which describes the effect of oxidation pretreatment on

the reaction and the resultant alumina morphology. In addition, the role of pretreatment in

controlling bonding at the NiAl-A1203 interface was shown to result in composites with

improved room temperature fracture toughness.













CHAPTER 1
INTRODUCTION


The search for a material or family of materials to serve as high temperature

structural materials for the next generation of jet engines has generated renewed interest in

intermetallic compounds or intermetallics. Many of these materials have been shown to

exhibit some of the properties necessary to perform adequately in the more hostile conditions

anticipated. For example, operating temperatures 25 to 50 percent greater than current

designs may become commonplace. Additionally, applied stresses will increase due to higher

engine speeds and compression. These factors demand materials with useful mechanical

properties at temperatures approaching 2000 K in an oxidizing environment. The desire to

increase operating temperatures is driven by the competing demands of increased performance

and reduced fuel consumption, both highly desired goals. For the current engine

technologies, it is commonly accepted that performance and fuel consumption are in direct

competition. Enhancement of one results in an accompanying deterioration of the other.

Thus, a general advancement of turbine engine technology is necessary to achieve the

combined goal of increased performance plus lower fuel consumption. To achieve the double

goal, it is necessary to design engines that operate at substantially higher peak and operating

temperatures [Bac92, Lar92]. This thinking is driving many present research programs to

develop future turbine engines to operate at significantly higher combustion temperatures.







2

1.1 Need for a High-Temperature Structural Material



The alloys incorporated in current engines consist of the Ni-based and Co-based

superalloys. These are complex alloys which have been developed nearly to their full

potential during the past 50 years. Evidence is given by the fact that they are routinely

expected to perform at 85% of their melting temperature. Consequently, these alloys are

unsuitable in even modest temperature increases unless other design changes are made to

protect and internally cool them.

For example, two solutions immediately present themselves. First, the current

superalloys can be maintained in high combustion temperature engines provided sufficient

internal cooling is furnished. This is similar to the use of aluminum alloys in internal

combustion engines for automotive applications where average combustion temperatures can

exceed 24000F, well above the melting point of alloys in contact with the combustion

chamber. The difficulties with this approach are four-fold. First, the maximum temperature

difference between the combustion temperature and the alloy temperature is limited by a

combination of the thermal conductivity of the alloy, heat transfer from the combustion gases

to the component and from the component to the coolant, hot spots and coatings. Coatings

are currently used on some superalloy components to provide protection from interaction with

the environment. Increasing combustion temperatures will accelerate substrate/coating

interactions and an imposed temperature gradient will increase thermally induced stresses at

the substrate/coating interface, possibly leading to degradation of the interface and/or spelling.

Secondly, any interruption to the coolant supply can raise the temperature of the

component very rapidly. This can cause permanent damage and lead to catastrophic failure.

Thus, the design of an intercooling system must be fail-safe.







3

Third, the development of internal stresses generated by the temperature gradient

across the component can become significant. As the temperature gradient increases, this

stress can become a significant fraction of the yield stress of the material, thus lowering the

maximum allowable externally applied load (this can be alleviated by use of Functionally

Gradient Materials, but these are currently in the early stages of development [Koi92,

Zha93]).

Finally, intercooling is already successfully utilized in today's jet engines. The ability

of new designs to provide additional heat removal necessitated by raising the combustion

temperature will present further challenges to designers.

Another approach to accommodate a significant increase in the combustion

temperature is to replace the materials from which certain high-temperature components are

made by other materials which can be operated at the new, significantly higher, temperatures.

The materials available for this application fall into three general categories, refractory metals

and their alloys, ceramics and intermetallics. Table 1.1 compares some of the critically

important properties for high-temperature structural materials of selected examples from each

group.

The critical properties necessary for high-temperature utilization of these materials are

as follows:

Retention of adequate mechanical properties at operating temperatures. This is

accommodated by high melting temperature, high creep resistance and gradual loss of yield

strength with temperature.

High strength to mass ratio. Indicated by low density, this property is essential to

improved performance. Low density also reduces centrifugal stresses on rotating parts

thereby decreasing the effective load during the operation.










Table 1.1: Comparison of Properties of Potential High-Temperature Structural Materials


Class Material Melting Density Room
Point (g/cm3) Temperature
(C) Fracture
Toughness
(MPa/m)
Metals Inconel 718 1400 8.1 24.0

tool steel 1500 7.8 98.0

Ceramics A1203 2015 3.9 3.9

SiC 2500 3.1 3.0

Si3N4 1900 3.3 4.1
Intermetallics MoSi2 2020 6.3 4.0

NbAl3 1680 4.6 2.5

NiAl 1682 5.88 6.0


Resistance to environmental degradation. The material must not be susceptible to

attack by highly corrosive combustion gases present during operation. Protective coatings can

be utilized, but they give rise to new engineering challenges.

Damage tolerance. The integrity of parts can in no way be compromised by normal

handling during fabrication, handling, assembly or operation.

Reviewing Table I with these properties in mind would eliminate all three

classifications of materials. The refractory metals can be stricken based on density and

environmental stability, and both ceramics and intermetallics possesses inadequate damage

tolerance.

As a result, many research programs have been initiated to develop methods to correct

these deficiencies without adversely affecting the remaining attractive material properties.







5

Much work has been done to increase the damage tolerance of both advanced ceramics and

intermetallics.



1.2 Intermetallics



Intermetallics possess some properties similar to ceramics, such as high modulus and

brittleness and some properties similar to metals, such as high thermal and electrical

conductivity. The relatively high thermal conductivity of intermetallics is seen as

advantageous over the thermally insulating ceramic materials in that it promotes a reduction in

thermal gradients in an operating engine. Thus "hot spots" are reduced or eliminated which

reduces peak material temperature.

In general, the ceramics provide greater oxidation resistance than intermetallics.

Noted exceptions are the various aluminides and silicides, which develop their oxidation

resistance by initially reacting with the environment to form a protective layer of A1203 and

SiO2, respectively. Once formed, these surface layers provide protection against additional

oxidation by retarding the diffusion of oxygen atoms to the metal atoms at the

intermetallic/surface layer interface.

In this respect, ceramics lose their advantage over the aluminides and silicides. On

the other hand, the protective layer on silicides, SiO, is subject to allotropic transformations

and softening above 1250C. Both of these attributes of silica render it inferior to alumina as

a protective surface coating in a thermally cycled, mechanically stressed application. The

volume changes can lead to cracking and spelling of the protective coating, exposing new

surfaces, increasing the penetration of corrosion products and softening which lead to







6

excessive creep of the coating. Furthermore, silica creep leads to nonuniform dimensional

changes and accelerated oxidation of thin regions.

Thus, the aluminides inherently possess the least deficiencies as a group and represent

the most promise for the near term development of higher temperature structural materials.

To successfully produce higher temperature structural materials from the aluminides,

two aspects of the mechanical behavior, which are common to many intermetallics, must be

overcome. Many investigators have reported the low room temperature damage tolerance

[Ant89, Cah91, Fle89, Mun88, Sha92] and rapid loss of yield strength at elevated temperature

[Cah91, Dar91, Des89, Sha92]. More detail on this behavior is provided in Chapter 2, for

nickel aluminide or NiAl.

Approaches to overcome detrimental mechanical behavior of aluminides range from

thermo-mechanical treatments which provide more mobile dislocations [Brz93, Hac92] to

alloying which incorporates stable phases with a sufficient number of active slip systems

[Mir93], to dispersion strengthening which increases the fracture stress of the material

[Loc90], to compositing with various reinforcements which exploit one or more of a variety

of toughening and strengthening mechanisms [Bar93, Rig92]. Each of these approaches has

met with limited success and work continues in all areas to provide both a fundamental

understanding of the processes at work and to develop practical materials for use in the near

future.



1.3 Composites



In simplest terms, composites are mechanical combinations of two or more materials,

specifically combined to exploit some individual properties of both materials. The usual







7

result being a composite with properties intermediate between the two individual materials. A

more detailed analysis, however, shows the properties of the composite to be controlled by

the properties of the matrix/reinforcement interface, the geometry and volume fraction of the

reinforcements and the type and geometry of loading, in addition to the properties of the

individual materials. Chapter 2 will provide a detailed analysis of the many interactions

between these parameters to illuminate the potential benefits and drawbacks available to affect

properties of the composite. Finally, it is hoped that much of the progress attained in the

other approaches to ameliorate the inadequate mechanical behavior of intermetallics, such as

alloying, may be applied to the matrix material of composites to provide a synergistic effect.



1.4 Summary



In the present study, we look at the effects of microcompositing the stoichiometric

intermetallic, NiAl. NiAl is chosen for its excellent combination of oxidation resistance, low

density, simple B2 structure and relatively high melting point (see Table I). Thus, NiAl is a

sufficient higher temperature structural material in all aspects except for mechanical behavior.

Compositing is seen as a method to develop useful mechanical properties without imposing

detrimental effects on the other, currently benign, properties.

Specifically, this study looks to increase the room temperature damage tolerance of

NiAl via microcompositing with an in-situ produced alumina network. The emphasis on

microcompositing stems from the fact that thermal cycling of composites will induce residual

stresses in the matrix adjacent to the reinforcement as well as in the reinforcement itself. This

residual stress, regardless of sense (tensile or compressive), will decrease the propagation of

matrix cracks via mechanisms described later. However, difficulty arises when higher cooling







8

rates are imposed on systems with large thermal expansion coefficient mismatch, Aa, such

that the magnitude of the residual stress can exceed the fracture stress of the matrix.

Nevertheless, it can be shown that below a certain reinforcement diameter, Aa, does

not cause matrix cracking and therefore the composites can be cycled at fairly high

cooling/heating rates without detriment. In addition, the composites formed with the small

diameter reinforcements will have an effective thermal expansion coefficient between that of

the matrix and the microreinforcement. This "microcomposite" can now be reinforced with

large, ductile, macro-reinforcements to produce a hybrid composite with a reduced Aa to

eliminate matrix cracking. Finally, a proper morphology as well as the volume fraction of the

reinforcement phase can be utilized to achieve higher damage tolerance composites.














CHAPTER 2
BACKGROUND/LITERATURE REVIEW



2.1 Physical and Chemical Properties of NiAl



NiAl is an ordered intermetallic which possesses good oxidation resistance, high

melting temperature, low density and metal-like electrical and thermal conductivity. In

addition to these attractive engineering properties, NiAl has a simple B2 crystal structure,

strongly ordered lattice, wide solubility range, anisotropic elastic behavior and a variable

defect structure.

This rich array of properties has made NiAl an actively studied compound for over 25

years, not only as a high-temperature structural material, but also for applications as diverse

as high-temperature oxidation-resistant coatings, catalysts, high-current circuit breakers and

electronic metallizations in advanced semiconductor heterostructures. Two recent

comprehensive reviews [Mir93, Noe92], of the literature available on NiAl provide exhaustive

coverage from the wide spectrum of research which has been carried out on NiAl. The

current survey shall briefly review only those properties deemed important to the use of NiAl

as a high-temperature structural material.



Structure

The Ni-Al equilibrium phase diagram is given in Figure 2.1. NiAl is an ordered

intermetallic with the simple B2 CsC1 structure (shown in Figure 2.2). Ordering is







1800

1600

1400

S1200 NiAl (Ni

1000

soo

600
600 NiSA13 <

300 (Al) ._.. .
z z z

0 10 20 30 40 50 60 70 80 90 100
A Atomic Percent Ni N i

Figure 2.1. Ni Al Equilibrium Phase Diagram.










SNi


Al


Figure 2.2. NiAI B2 CsCl Crystal Structure.







11

maintained up to the congruent melting temperature of 1955 K. NiAI has the highest melting

temperature in the Ni-Al system and is the most stable phase based on the large negative heat

of formation of -72 kJ/mol at the stoichiometric composition. The B2 structure is stable from

45 to nearly 60 atomic percent Ni and combined with the high thermodynamic stability makes

NiAl fairly easy to fabricate by numerous processes from bridgeman single crystal growth to

self-propagating high-temperature synthesis.



Oxidation Resistance

NiAl alloys have long been utilized to produce a variety of oxidation-resistant coatings

for Ni-based superalloys [Hip90, Mit92]. A combination of easy formation and slow growth

of a protective A1203 scale provides the basis for this use. Doychak et al. [Doy89a], studied

the transient oxides formed on NiAl single crystals at 800C and 1100C. They found that

oxides of all metals formed initially by nucleation and growth along the surface. Eventually a

complete healing layer of Al203 forms beneath the existing scale. Polymorphic forms of

A1203 form during transient stage, but are rapidly (10 hr. at 8000C, 0.1 hr. at 11000C)

replaced by only Al1203 regardless of crystallographic orientation. Tracer diffusion data

[You86] and scale morphologies [Doy89] indicate that the transition oxides grow primarily by

outward cation diffusion. At higher temperatures, the a-Al203 initially grows primarily by

outward Al diffusion [Pre92] but later, large contributions come from inward oxygen

diffusion, mainly along short-circuit diffusion paths [Jed91]. At a given temperature, the

parabolic growth rate of 0A1203 is 1-2 orders of magnitude greater than a-Al203.

Cyclic oxidation testing appears much more severe than isothermal oxidation, due in

large part to the significantly different thermal expansion coefficients (CTE). CTE mismatch

induced stresses cause spelling, which occurs randomly over the surface exposing fresh metal.







12

This leads to short cyclic life times since the Al consumption rate is inversely proportional to

the oxide thickness and spelling tends to maintain thin oxide layers. Barrett [Bar88] found

that small additions (0.1 at. %) of Zr significantly reduce spelling and slightly reduce A1203

growth rates for isothermal conditions.

In general, alumina forming alloys such as NiAl exhibit the most oxygen resistant

behavior over the broadest range of temperatures of metallic and intermetallic materials.



Conductivity

Thermal conductivity is of primary importance in NiAl as a high-temperature

structural material. Its thermal conductivity is from four to eight times that of Ni-based

superalloys, depending on composition and temperature [Dar91]. This high conductivity helps

to reduce temperature gradients in parts fabricated from NiAl during engine operation,

resulting in reduced thermal stress (and fatigue) and effectively lowering the peak operating

temperature of the part. Electrical conductivity of NiAl is similar to metals.



2.2 Mechanical Properties of NiAI



The mechanical behavior of NiAl has been extensively studied [Bak91, Hah89,

Tak93]. Apparent discrepancies appear in the literature that attest to the influences of subtle

variations in defect structure on the mechanical behavior of NiAl. This review will be limited

to pertinent topics which illustrate the current understanding of the lack of room temperature

damage tolerance and the rapid loss of strength at elevated temperature of polycrystalline

binary NiAI. A subsequent section highlights the various approaches which show differing

degrees of success in improving the mechanical behavior of NiAl.









Strength and Ductility

NiAI generally fractures in a brittle intergranular fashion at low temperatures and

exhibits a sharp brittle-to-ductile transition (BDT) at 475-673 K, depending on composition,

loading rate, [Veh92], presence of interstitials [Geo90] and orientation [Tak93]. Just above

the BDT anomalously large tensile elongations of greater than 120% were observed for "soft"

single crystals before decreasing to around 50% at higher temperatures due to the onset of

necking. The anomalously large elongations have been attributed to a balance between work

hardening due to glide and relaxation due to climb resulting in a high necking resistance

[Tak93].

The yield stress of polycrystalline stoichiometric NiAl is reported to be independent of

temperature from room temperature to = 800 K [Noe91a] and to drop rapidly above 900 K.

In general, the flow stress of NiAl is similar to that of BCC metals, exhibiting a strong

temperature dependence at low temperatures that is attributed to a large Peierls stress. At

intermediate temperatures there is a plateau where yield stress is mildly dependent or

independent of temperature followed by a rapid drop-off at elevated temperatures.

Recent studies indicate room temperature ductility of about 1-2 percent, while the

yield strength varies from 115 to 235 MPa [Geo90, Hah89, Pas68, Roz57]. The differences

in the reported strengths may serve to illustrate the high sensitivity of the flow properties of

NiAl to composition, strain rate, grain size, cooling rate, surface finish and fabrication

conditions. Since most studies fail to report one or more of these important variables, direct

comparison of the results is possible only in a superficial manner. Nevertheless,

generalizations concerning the effect of these variables can be made.

Composition. Deviations from stoichiometry are accommodated by the substitution of

Ni atoms onto Al sites in nickel-rich compositions (antisite defects) and by the formation of







14

vacancies on the nickel sites in aluminum-rich compositions [Bra37]. These antisite and

vacancy point defects have been shown to locally order such that no defect has a similar

defect as a nearest neighbor [Geo81], this leads to clusters of Ni2Al and Ni2AI3, respectively

[Liu81].

Perfect Ni:Al stoichiometry of 1:1 is associated with a minimum in strength and a

maximum in ductility. Westbrook [Wes56] first reported this effect in microhardness testing

of arc-melted buttons and related the behavior to the defect structures given by Bradley and

Tayler [Bra37]. This behavior was also shown to apply to yield strength by Vedule and

Khadkikar [Ved90]. A higher slope of the yield stress versus composition curve on the Al-

rich side of the stoichiometric composition indicates a higher degree of strengthening by

vacancy defects compared to antisite defects. The strong influence of defect structure on the

mechanical properties of NiAl appears to hold the key to obtaining (and maintaining) useful

room temperature properties.

Interestingly, the room temperature fracture toughness of NiAl was shown to be

insensitive to deviations from stoichiometry of 50 60 atomic percent Ni [Kum89].

Grain size. Schulson [Sch81] has suggested that there exists a critical grain size, dc,

below which tensile ductility may be imparted upon brittle intermetallic polycrystals. This is

based on the idea that the stress necessary to propagate cracks decreases with increasing grain

size due to greater stress concentrations caused by larger dislocation pile-ups at the grain

boundaries. If it is assumed that the stress necessary to nucleate a microcrack is relatively

constant, it can be seen that in very large-grained material the stress necessary for crack

propagation is less than that required for nucleation. Thus, once formed, the crack propagates

catastrophically. Conversely, in very small-grained material, the propagation stress is greater

than the nucleation stress, thus propagation occurs only after work hardening or plastic flow







15

has occurred. Assuming the length of the microcrack is proportional to the grain size, cracks

nucleate at the yield strength and there is no interaction among cracks, the critical grain size

is given by:


de=(YK-k)2 (2.1)




where Y is a geometrical parameter = 1, Kjc is the plane strain fracture toughness and k, and

Ti are the empirical Petch-Hall parameters which are measures of the effectiveness with which

dislocation slip is impeded by grain boundaries and the resistance to dislocation slip by the

lattice, respectively. This model is supported by Noebe et al. [Noe91a], indicating that the

critical grain size of NiAI is about 1 tm and that for a grain size of 0.1 am, about 5%

elongation can be expected. In addition, the effect of grain size appears to increase as the

deviation from stoichiometry is increased [Noe92] creating an interactive effect to further

complicate the issue.

Conventional processing techniques produce NiAl grains generally > 10 Am, thus

investigations are underway to develop processes, such as mechanical alloying and inert gas

condensation to develop submicron to microcrystalline grain sizes [Boh91, Dym92, Hwa91].



Strain Aging

Recent work by Hack, et al. [Hac92] has shown a profound effect of heat treatment

and cooling rate on the room temperature fracture toughness of NiAl. They found that rapid

cooling from 400C followed by a variety of thermal treatments resulted in vastly improved

toughnesses up to 16.7 MPaV/m for 400C held for 12 hours, followed by air cooling at room

temperature. This phenomenon is believed to be related to the defect structure of the NiAl







16

lattice. Apparently, the specific heat treatments employed result in the formation of excess

mobile dislocations, imparting some degree of plasticity on the single crystals studied. This

result and the reported loss of toughness by a 1.5 hour hold at 200C indicate the occurrence

of dynamic strain aging [Brz93]. Another indication of this phenomenon comes from the

observation of serrations in stress-strain curves in certain temperature/strain rate regimes

[Lau65, Ved87].

The occurrence of irregularities in stress-strain response indicates unstable plastic flow

and has come to be called the Portevin-LeChatelier effect [Ree92]. This appears to indicate a

strong interaction between impurity atoms and dislocations during deformation. At

intermediate temperatures and strain rates, dislocation atmospheres alternately become pinned

by impurities, then as the strain accumulates, they break free, causing a rapid drop in the

applied load. Thus, higher velocity dislocations allow localized softening, resulting in a lower

yield stress until the dislocations again become trapped and the material begins to harden to

another upper yield point. This cycle can be repeated numerous times during deformation.

Important consequences of dynamic strain aging (DSA) include: (1) the temperature

interval at which DSA occurs depends on strain rate (increasing the strain rate raises both the

upper and lower temperature limits); (2) during DSA, the yield stress tends to become

independent of temperature (and flow stress, conversely tends to be independent of strain rate,

i.e., a=A(E)n); and (3) the work hardening rate can become abnormally high and strain rate

and temperature dependent. The various DSA phenomena do not appear to the same degree

in all metals; however they appear to be more of a rule than the exception. NiAl appears to

behave in a similar fashion [Brz93, Cot91, Hac92, Ree92]. The occurrence of DSA in NiAl

serves to explain some of the apparent discrepancies in the NiAl literature, since very few







17

studies make a detailed study of the defect substructures and the manner in which they change

during the deformation process.



Anisotropy

Single crystal studies have shown quite a large degree of anisotropy with changing

crystallographic direction [Was66]. This large degree of anisotropy results in what are

referred to as "hard" ([100]) and "soft" ([111]) orientations. The yield strength ratio

[100]/[111] =6 and measurable tensile ductility is not observed in the "hard" orientation.

The anisotropy results from the highly directional bonding between unlike atoms

along <111> directions which also results in the operative slip systems [Bal66, Hah89,

Pas68, Sto66, Was91]. The primary slip system in NiAl is generally reported as

<001 > {110}. For polycrystalline NiAl, this results in an insufficient number (three) of

independent slip systems to allow general plastic flow, as five independent systems are

required to produce the arbitrary shape change of each grain in order to maintain contact with

adjacent grains during deformation [Mis28].



2.3 Fracture



The typical room temperature fracture mode of polycrystalline NiAl in tension is

intergranular (IG) at the stoichiometric composition [Hah89, Raj89] and transgranular (TG) in

off-stoichiometric compositions [Nag91]. A transition to TG cleavage in stoichiometric alloys

has been reported at 673 K [Hah89] and 873 K [Raj89]. Cleavage is also reported for coarse-

grained stoichiometric NiAl at room temperature at a high strain rate [Gua60].







18

Intergranular failure is a direct result of the availability of only three independent slip

systems. Shape incompatibility of adjacent grains cause microcracking at the boundaries

which coalesce to lead to catastrophic failure. Transgranular cleavage in off-stoichiometric

NiAl, on the other hand, indicates that the grain boundaries are as strong as the matrix. Thus

efforts to increase room temperature ductility of polycrystalline NiAl via grain boundary

strengthening seems fruitless. The room temperature mode I fracture toughness of NiAl, Kic,

is generally reported to be 4-6 MPa/m [Kum92, Nag91] for a wide range of grain sizes. Kic

increases to about 10 MPaV/m at 623 K for as-cast material, and to 50 MPa/m at 673 K for

zone refined material [Rev90]. The fracture energy of unnotched polycrystalline NiAI is

given as 0.8 J/cm2 [Nar90].



2.4 Toughening NiAl



There have been numerous studies undertaken in the attempt to understand the

mechanisms) responsible for the lack of room temperature damage tolerance of NiAl, and

thus, to develop counter measures. Many have been at least partially successful. The

following paragraphs detail some of these successes to show the various methods employed

and the degree of success as well as the limitations.

The techniques can be classified as metallurgical, mechanical or a combination of the

two. Metallurgical approaches generally attempt to enhance ductility by providing plasticity

through the introduction of sufficient mobile dislocations and/or the activation of additional

slip systems. For example, Brzeski, et al. report significant increases in the room

temperature fracture toughness of single crystal NiAl to 16.7 MPaV/m [Brz93, Hac92]. As

mentioned above, this is accomplished by rapid cooling from a rather low (400C) annealing







19

temperature. Mechanical means to increase damage tolerance are concerned with inhibiting

crack growth. Although the end results are similar, the approaches and thought processes can

be quite distinct. Finally, it is believed that the optimum solution may be some synergistic

combination of metallurgical and mechanical approach.



Metallurgical Processes

The previous section summarized the current understanding of the metallurgy of NiAl.

The lack of sufficient slip systems and mobile dislocations explain the lack of plasticity during

deformation of NiAl. The varied approaches to alleviate these deficiencies include alloying

[Abr89, Cot91, Cot93, Dar91, Dar92, Kay91, Whi89], heat treating [Brz93, Hac92] and

thermomechanical processing [Mar91, Mar92, Whi89].

The use of ternary alloying additions to affect slip behavior has largely been

unsuccessful [Cot93, Noe92]. The rationale involves the use of alloying additions which may

lower the ordering energy of NiAl, thus making < 111> slip easier. Calculations based on

interatomic potentials [Hon89, Hon91] indicate that Cr, Mn and V are good choices for

promoting <111> slip. These models indicate that reductions in APB energy of 70 percent

are possible, but require alloying additions of at least 17 percent, which is well above the

solubility limits of Cr (2%) and V (12%) and very little work with Mn is cited [Noe92]. The

addition of small amounts of Fe, Ga and Mo have been reported to increase the room

temperature tensile ductility of NiAl up to 7 fold [Dar92]. The reason for these increases in

damage tolerance are presently unknown; however, a number of possible explanations have

been suggested. These include (1) gettering of interstitials, (2) slip homogenization,

(3) enhancement of dislocation sources, (4) better casting quality, (5) dislocation-solute







20

interactions and (6) reduction in the point defect concentration or distribution [Dar91]. It was

noted that < 100 > {110} slip was not affected by these additions [Dar91].

The use of other alloying additions provide solid solution strengthening which

increases the yield stress, thus lowering ductility by reducing the magnitude of the fracture

stress relative to the yield stress. The use of alloying additions to stabilize a second phase or

to allow precipitation will be covered below under "Mechanical Toughening."

The appearance of limited ductility in thermo-mechanically treated NiAl, such as hot

extrusion, is attributed to the preferred alignment of a < 111> or < 110> texture along the

extrusion axis [Bie92, Kha90]. Thus, the tensile properties will depend on the orientation of

the tensile axis with respect to the extrusion axis. The decrease in yield strength of NiAl due

to the prior application of hydrostatic pressure has been related to the creation of mobile

dislocations [Mar92]. The occurrence of extensive ductility in NiAl testing in tension with

superimposed hydrostatic pressure of 500 MPa is due to the pressure induced suppression of

damage accumulation [Mar91].



Mechanical Processes

The mechanisms which describe the processes operating to provide "mechanical

toughening" are described in a later section under "Composites." The current discussion

illustrates examples of results obtainable via the various techniques.

Generally, the mechanical methods are utilized to increase damage tolerance by

inhibiting crack growth, although for some mechanical means, such as ductile phase

toughening or dispersion strengthening, a secondary affect arises which may provide a greater

toughening effect. Namely, the presence of numerous boundaries can provide sites for the

production of mobile dislocations which can impart plasticity on the alloy regardless of the







21

crack arresting properties of the second phase particles. (This mechanism is known as "slip

transfer" [Noe91]).

Johnson et al. [Joh92], report K, fracture toughness values of 15.5, 13.2, and

30.7 MPavm for the directionally solidified eutectic systems NiAl-12Mo, NiAl-15Mo and

NiAl-40V, respectively. Unfortunately, they also reported an inverse relationship between

room temperature fracture toughness and creep. Kaysser, et al. [Kay91] likewise report a

large increase in toughness by ternary additions to powder metallurgy (PM) NiAl alloys.

Their data show an increase in toughness to 15.4 MPaV/m for hot pressed and hot isostatically

pressed (HIPed) NiAl-5Nb and 14.5 MPav/m for NiAl-5Ti plus increased creep resistance for

both alloys. Russell, et al. achieved room temperature fracture toughnesses as high as

25 MPaVm in two-phase Ni-Al-Co alloys [Rus89].

If the second phase is brittle and strongly bonded to NiAl, the toughness can be less

than that of binary NiAl [Rev90]. Although dispersion strengthened alloys with HfC [Whi90]

are shown to increase the compressive strength, toughness was not evaluated.

NiAl matrix composites with brittle reinforcements have been successfully produced,

resulting in only moderate increases in toughness [Bar93, Kum92, Rig92]. For example,

Rigney et al. [Rig92], report up to a 50% increase in fracture toughness to 9.0 MPaV/m for

hot pressed, prealloyed NiAI reinforced with 10 volume percent, 4.2 'm diameter TiB2

particles. Kumar, et al. [Kum92] showed an increase in fracture toughness of NiAI reinforced

with A1203 whiskers, up to 9 MPaV'm at 15 volume percent. The toughness remained at

about 9 MPaVm up to 25 v/o, the highest level cited in the study.







22

2.5 Composites



General

In the past three decades a new class of materials, known as composite materials, has

emerged. Composite materials or, simply, composites can be generally defined as materials

having two or more inherently different constituents, which may or may not be in

thermodynamic equilibrium. On this basis, composite materials can be arbitrarily classified

into two distinct groups, (1) natural or in-situ composites and (2) artificial composites.

Natural composites are often directionally transformed polyphase materials, such as

directionally solidified eutectic alloys. In this case, the reinforcement phase is generally in

equilibrium with the matrix phase. Rowe, et al. [Row92] produced microlaminate composites

of Nb3Al-Nb, (Nb,Ti)3Al-Nb(Ti,Al), (Nb,Ti)2Al-Nb(Ti,Al) and Cr2Nb-Nb via high rate

magnetron sputtering with low interstitial contents. All these composites were found to be

stable at elevated temperatures except for (Nb,Ti)2Al-Nb(Ti,Al) which formed the

intermetallic phases (Nb,Ti)3Al-(Nb,Ti)2Al. The remaining composites exhibited significant

toughening with room temperature fracture strength of 475, 670 and 720 MPa, respectively.

In addition to work on natural composites, a large effort is underway to develop

artificial composites for high temperature applications. Here, the reinforcements are

mechanically blended with the matrix and as such, are generally not in equilibrium. This may

necessitate a diffusion barrier coating at the reinforcement/matrix interface. But at the same

time, artificial composites allow greater freedom of choice for reinforcement material, thus,

providing for the intelligent selection based on property matching to produce a composite with

optimum mechanical behavior for a given application.







23

The current study has developed a composite system which is neither strictly natural,

nor strictly artificial, but exhibits characteristics of both. In NiAl-AI203 composites, the

constituents are not in strict chemical equilibrium, but for engineering purposes they do

display "kinetic" equilibrium, characterized by virtually no interdiffusion at the fabrication

and service temperatures expected.

Composites are fabricated in many diverse forms. Many properties of these

composites have been used to develop classification systems. One basic classification divides

the various composites by the shape of the reinforcing phase. This usually results in two

broad categories.

Particle reinforced. Particles are mixed into the matrix and the strengthening

mechanism depends on the size and distribution of the particles.

Fiber reinforced. Fibers are included in the matrix and become the dominant load-

bearing component. The fibers may be continuous, chopped or assembled into a three-

dimensional array called a preform.

Within each of these categories, composites are further divided into classes based on

the bulk properties of the constituents. The composite can be classified as either ductile-

ductile, ductile-brittle, brittle-ductile or brittle-brittle, referring to the properties of the fiber

and matrix, respectively. High-temperature intermetallic alloys are generally characterized as

brittle at room temperature and studies utilizing both brittle [Mac93, Kum92] and ductile

[Mat89, Lu90] reinforcements have been made. The toughening mechanisms detailed below

generally favor the use of ductile reinforcements, and although brittle fibers have been shown

to improve toughness [Kum92], the mechanism is not fully understood.

The third widespread classification system for composites is manufacturing method.

Each method imparts unique characteristics on the composite and carries a certain set of







24

advantages and limitations. Some of the matrix/fiber systems favor certain fabrication

processes over others as depicted below.



Composite Toughening

Much of the theoretical treatment of the mechanics of composite materials is drawn

from linear theory of elasticity for homogeneous solids [Chr79]. For brittle intermetallics,

this assumption with regard to behavior is particularly useful, especially in comparison with

the behavior of polymeric or metallic materials. The use of ductile reinforcements provides a

larger degree of deviation from linear elastic response than brittle reinforcements, although

for smaller volume fractions the difference can be accommodated. Smaller volume fractions

are defined where the state of strain in any one particle in the composite body under

homogeneous boundary conditions is not affected by all the other particles [Has83].



Fracture Mechanics

The mechanistic descriptions of fracture toughness began with Griffith and his work

on the fracture of glass [Gri21]. He related the stress necessary to break the glass, OL, to the

size of the flaw, a, and energy of the newly created surface, 7, to the modulus by the

relation:


,L=(2Ey)12 (2.2)
ira



Since glass has no measurable room temperature plasticity, fracture is purely a result

of breaking atomic bonds. This was extended to deformable materials by Irwin [Irw48] to

include the energy absorbed by plastic processes in the creation of the fracture surface, i.e.,







25

an effective surface energy. Plasticity at the crack tip serves to effectively blunt the crack,

spreading the singular peak in the stress distribution near the crack tip over some finite area.

A more plastic material spreads the stress over a larger area, necessitating a larger externally

applied stress to advance the crack.

For brittle materials, the crack is assumed to be sharp. The relation between stress

state near the crack tip, the size of the existing crack and the externally applied load is given

by:


K=O( a)Wa T (2.3)
L



where K is the stress intensity factor in MPaVm, 3 is a dimensionless geometric factor, a is

the characteristic crack length (m), L is a geometric factor describing the cracked specimen

and a is the applied stress in MPa [Bro88].

The externally applied load causes catastrophic crack propagation at the fracture

stress, a,, of the material, which is a function of crack size. At af, the stress intensity factor

becomes the critical stress intensity factor, Ke, which is commonly referred to as the

material's fracture toughness. Although, Kc is given as a material property, it is not

invariant. Loading rate [Mun69, Sau90], sample and loading geometry [Mun80a, New84]

and environmental effects [Mun69] are known to cause variation in experimentally determined

values of Kc. Thus, fracture toughness measurement method will influence the absolute value

obtained and accordingly, direct comparisons of results from different testing methods is not

possible. As a result, fracture toughness values are more correctly referred to as "estimates"

which give an indication of the expected material behavior under narrowly defined conditions.









Toughening Mechanisms

From fracture mechanics, we find that there are essentially two fundamental

approaches to "toughening" brittle solids. The first is to accept the brittleness of the material

and control the flaw size and distribution. This can be accomplished through process controls

which reduce product variation, and/or through a general reduction in the scale of the

microstructure, e.g., reduced grain size. This approach is limited, both technologically by

current state of the art processing capabilities, and theoretically by the size of quantum

defects, vacancies and dislocations. The second approach is to extrinsically create

microstructural features which increase the fracture resistance of the material so that it

becomes insensitive to flaw size [Eva90]. The latter approach has the advantage in that

damage during processing and in service will be much less detrimental to product reliability

than the former, in addition to no theoretical upper limit on fracture toughness, per se.

As a crack grows through a particulate-reinforced composite, many events occur that

are potentially related to the fracture toughness and each of these either make it easier or

more difficult for the crack to extend. In Figure 2.3, the events which have been related to

the resistance to crack growth are identified. Each of these events absorb energy and thus

require more work to be done by the external load.

The following paragraphs describe the important toughening mechanisms utilized to

make brittle matrix composites more damage tolerant. Although an attempt has been made to

rank the various mechanisms according to the magnitude of the effect on toughness, it must

be noted that the method used to express the toughness will change that ranking somewhat.

For this study, the K, fracture toughness based on the peak load obtained in four-point bend

of a chevron-notched sample is the standard. Thus, toughening mechanisms which emphasize

work of fracture, at the expense of peak load carrying ability will be penalized more harshly.














I

S ()
I


O -0

SBridging Process
Zone Zone



Figure 2.3. Schematic of energy absorbing microstructural features in composite materials
including: (a) fracture of reinforcement; (b) interracial separation; (c) crack deflection;
(d) crack bridging; (e) crack arrest; (f) multiple matrix cracking; (g) transformation
toughening and (h) tortuous crack path increases surface area.







28

Xiao [Xia92] illustrates this by comparing the toughness of MoSi2-Nb laminate composites

with strongly bonded interfaces to laminates with interfacial coatings which cause weak

bonding. The uncoated composites clearly demonstrated the highest peak load, whereas ZrO2

coated Nb shows the highest work of rupture in the study, measured as the area under the

load-displacement curve.

Table 2.1 lists the major energy dissipation mechanisms exploited in the design of

brittle matrix composites. The dominant mechanism to control the fracture process depends

on the size, shape and distribution of the reinforcement phase relative to the geometry of the

advancing crack as well as individual properties of the materials and the interfaces. Therefore

it is recognized [Xia94] that this dominant mechanism can and will change as a crack

propagates past the reinforcements. For this reason, the model for an advancing crack is

usually divided into the process zone and the bridging zone [Eva90]. These zones are

indicated in Figure 2.3 and Table 2.1.

Fiber Pull Out. Fiber pullout requires some debonding of the interface and is usually

associated with weakly bonded interfaces. The main contribution to energy dissipation in

such a system results from the frictional forces which must be overcome in order to pull the

fiber out from the matrix [Eva89, Eva90]. Thus increased interfacial roughness and residual

compressive circumferential stresses in the matrix tend to increase the amount of energy

exhausted to pull by a given increment [Eva89].

Crack bridging. Crack bridging, although related to fiber pullout, actually represents

a separate energy dissipation mechanism. Bridging occurs in the wake of an advancing crack

[Eva89, Eva90, Mat89, Xia93]. Figure 2.4 shows the increasing elongation of the ductile

filaments near an advancing crack. The degree of interfacial debonding determines the "gage

length" of reinforcement available to dissipate energy. The gage length also influences the










Table 2.1: Dominant brittle matrix composite toughening mechanisms.
Mechanism Brittle Ductile Zone
reinforcements reinforcements

Fiber Pullout V/ B

Crack Bridging / / B

Residual Stress / / P,B

Crack Blunting / P

Interfacial / / P,B
Debonding

Multiple Matrix / / P
Cracking

Microcracking / / P

Transformation P
Toughening
*not dependent on reinforcements, B= bridging zone, P=process zone.


degree of necking during failure of the ductile ligament in the crack wake [Mat89]. The work

hardening coefficient controls whether ductile deformation occurs over the entire length or

becomes concentrated in one area to produce necking. Thus, a combination of ductile

ligament and interfacial properties determines the degree of energy dissipation through crack

bridging [Eva89a].

Residual stress. The existence of some residual stress in the microstructure, usually

due to the effects of thermal processing and the mismatch in thermal expansion rates between

the matrix and the reinforcement not only influences the magnitude of all other toughening

mechanisms, its presence also leads to two new mechanisms for energy dissipation. Figure

2.5 schematically shows these two mechanisms. In Figure 2.5(a) the residual circumferential

matrix stress is tensile (ama(ix > reinforcement), which causes the crack to be attracted to the

reinforcement by providing a lower energy path out of the original plane of propagation.







30

Thus, the crack is trapped by nearby reinforcements and other energy dissipation mechanisms

are forced to activate more frequently than would occur had the crack continued to travel in

the original plane. Figure 2.5(b) shows the opposite stress state (a a < are~orcemen). In this

case, the crack is deflected out of plane by the increased stress which would be necessary to

overcome the residual state of compression near the reinforcement. Thus, the crack "avoids"

the reinforcement and in the process dissipates additional energy. The magnitude of energy

dissipation depends on the level of residual stress but not the sense [Sig88]. The residual

stress is determined by the thermal expansion mismatch (Aa) and the temperature interval

(AT), whereas the radius of the residual stress atmosphere is determined by the radius of the

reinforcement. Larger reinforcements create larger atmospheres (see "Matrix Cracking,"

below).

Crack blunting. Crack blunting by ductile reinforcements is shown in Figure 2.6. If

both the interfacial bond and the matrix are relatively strong, the crack can be forced directly

into the reinforcement. Once the crack tip breeches the interface, a plastic zone ahead of the

crack forms in the ductile reinforcement. This process dissipates energy through dislocation

creation and motion and eventually via microvoid formation and coalescence.

Interfacial debonding. Figures 2.3 and 2.4 show several possibilities for the

dissipation of fracture energy by interfacial debonding in systems with relatively weak

interfaces. The main difference between them is the timing of their initiation (ahead of the

crack, at the tip or in the wake) which depends on the relative properties of the constituents.

Due to changes in the loading geometry with crack growth and the presence of inherent flaws,

the crack may reinitiate on the far side of the reinforcement in a new plane, usually parallel to

the original crack growth plane.





























Figure 2.4. Schematic of fiber reinforced composite toughening mechanisms. Shown are (a)
crack initiation at interface in process zone, (b) crack front debonding, (c) microcracking,
(d) wake debonding, (e) fiber pullout (sliding) and (f) bridging.








S< Y > Cg,


Matrix


Crack
Growth
Direction


Reinforcement


(a) (b)
Figure 2.5. Effects of residual stress on the direction of crack propagation near a
reinforcement are (a) attracted to reinforcement due to tensile residual stress in matrix and
(b) crack deflection due to compressive residual stress in matrix.

























S(e)


(a) I(d)













Figure 2.6. Schematic of crack blunting energy dissipation mechanism. Crack may be
subject to (a) arrest at interface, (b) plastic zone in reinforcement increases effective crack tip
radius, (c) interfacial separation, (d) crack reinitiates on opposite side of reinforcement
and (e) microcracking.







33

Multiple matrix cracking. When the interface is relatively strong, the crack has been

seen to form multiple branches ahead of the reinforcement [Xia94]. The cracks dissipate

energy by reducing the stress intensity at the crack tip and also provide a longer effective

"gage length" for the bridging of ductile reinforcement.

Microcracking. This mechanism generally occurs in the "process zone" ahead of the

advancing crack tip. The occurrence of microcracking effectively dissipates energy similar to

a plastic zone, although the actual mechanism is quite different (cracks are not mobile).

Instead of dislocations, very fine cracks are generated in the process zone volume, which both

dissipate energy by creating new surfaces and reduce the effective stress intensity.

Transformation toughening. This mechanism occurs due to a stress-induced

microstructural change which forms a product of higher volume than the original

microstructure, such as martinsitic transformation. This transformation begins ahead of the

crack tip, requiring a larger stress intensity to propagate a crack and finishes in the crack

wake, creating a crack closure force which tends to reduce the effect of the far-field stress

crack opening.



Matrix Cracking

In brittle matrix composites, highly localized stresses can arise as the body is cooled

from the fabrication temperature due to the difference in thermal expansion rates between the

matrix and the reinforcement. In addition, stress magnification at the reinforcement/matrix

interface can occur due to the effect of differences in elastic properties when an external load

is applied to the body [Dav68]. These effects can be additive or subtractive depending on the

relative values of the thermal expansion coefficients and elastic constants of the matrix and the

reinforcement.







34

Matrix cracking will occur when the net local stress causes the local stress intensity

factor, K, to exceed the critical stress intensity, Kc, for some preexisting flaw in the vicinity

of the reinforcement [Eva74]. For composites which contain reinforcements with a modulus

greater than that of the matrix (e.g., A1203 reinforced NiAl) the tangential stresses in the

matrix at the interface normal to the applied stress are tensile but less than qa, the applied far-

field stress [Eva74]. Parallel to a,, the stresses are tensile near the surface and become

compressive further into the matrix. Thus, for the specific case of A1203 reinforced NiAI, we

may neglect the stress concentration due to elastic mismatch.

For an elastic spherical inclusion in an isotropic, infinite, linear elastic matrix, the

stress distribution in the matrix due to thermal expansion mismatch is given by [Eva74]:


ar (a-am)(f-) R (2.4)
S(l+vm)/2E+(1 -2v)/E, r




(am-a)(Tf-T) R3
S2[(1 +v)/2E (+(1-2),)/E] r



where ar and aoo are the residual stresses in the radial and tangential directions, respectively,

amr, 7m,rn and Em., are the expansion coefficients, Poisson's ratio and Young's modulus for the

matrix and reinforcement, respectively. T, is the fabrication temperature and T is the ambient

temperature. R is the inclusion radius and r is the distance from the center of the inclusion.

The model is expanded by Fenner [Fen86] to include non-dilute systems where

increasing volume fraction of the reinforcement causes impingement of these stress fields,







35

such that stress-free positions no longer exist at distances r> > R. Thus,


a=C [VK-R2/(r+R2)] (2.6)




ar,=C [Vr+R2/(r+R 2)] (2.7)




E
where C=[AaATEm]/ [1+(1-p)][V,(1-Em/Er)+-3] (2.8)
Er



where v is Poisson's ratio for the composite, usually the matrix value used.

Thus, the maximum stress occurs at the interface and is independent of the

reinforcement diameter. The effect of larger reinforcements is therefore a decrease in the rate

at which the stress drops as a function of r. The likely deleterious effect on fracture behavior

results if cracks develop from preexisting flaws due to these stresses near the interface. The

model predicts a large population of matrix cracks near larger reinforcements due to the

gradual decrease in the stress field away from the interface and a finite probability of matrix

cracking near small reinforcements. An important result of the diminishing residual stress

implies that matrix cracks which form in the radial direction should arrest at some finite

distance into the matrix where K < Kc. These and circumferential cracks due to tangential

tensile stresses become the predominant preexisting flaws in the composite and serve to reduce

the maximum applied stress which the composite can reliably sustain.

Contrary to the predictions of these models, it has been established [Dav68, Eva74,

Ito81, Lan76, Lu91] that a critical reinforcement size exists below which matrix cracking is

suppressed. Although the precise mechanism for crack suppression is not understood, an







36

empirical coefficient, 9?, has been developed which seems to allow a priori determination of

the critical reinforcement diameter [Lu91]. This non-dimensional quantity is defined by

[Lu91]:


9?=R(Er/K )2 (2.9)



where R is the reinforcement size, Km the matrix toughness, Em the matrix modulus and eT is

the misfit strain, given by:

T
e, = AadT (2.10)







with (T To) being the processing temperature range and Aa the difference in thermal

expansion coefficient between the matrix and the reinforcement.

For this coefficient, it is found that matrix cracking does not occur for 9 < 9c.

This critical value of 9? depends on the volume fraction, f, and aspect ratio of the

reinforcements, the ratio of elastic moduli and the interfacial friction coefficient by [Lu91]:


9Sc=3(1-v,)/f (2.11)



for composites which exploit fiber pull out as a toughening mechanism (i.e., weakly bonded

interfaces).

The important results of this model indicate that a strong interfacial bond generates a

critical parameter 9c on the order of 3 6 times as large as a weakly bonded interface for the

same matrix and reinforcement. The spatial arrangement of the fibers in the matrix was







37

shown to affect the incidence of matrix cracking with a cubic arrangement exhibiting a higher

tendency than a hexagonal arrangement [Lu91]. This may be due to the distribution of

"nearest neighbors" where a greater number of nearest neighbors will reduce the intensity of

stress variation as a function of angular orientation around each fiber, providing a greater

uniformity in stress distribution, thereby spreading out the stress over a greater area for a

given reinforcement diameter and volume fraction. Finally, it was suggested that all forms of

matrix cracking will be suppressed if 9? is kept below a value of unity for any

matrix/reinforcement combination [Lu91].

Matrix cracking has also been shown to develop in a less obvious way in NiAl-A103

composites. Misra [Mis93] found in A1203 fiber reinforced FeCrAlY superalloy, that a weak

interface caused separation of the fiber from the matrix which has a larger thermal expansion

coefficient. During high temperature excursions, aluminum from the matrix reacts with

oxygen from the environment in the gap which forms at the fiber/matrix interface, effectively

increasing the diameter of the fiber. Upon cooling the matrix attempts to contract to its

original shape, but is prevented by the now larger fiber, thus magnifying the CTE mismatch

stresses at the interface and leading to matrix cracking in composites which were originally

crack-free. Thus, interfacial separation during thermal cycling must be prevented in these

composite systems to ensure long-term integrity. This can be accomplished by increasing the

strength of the interface, or possibly through incorporation of smaller diameter reinforcements

which reduce interface separation via the mechanism described above to eliminate single cycle

matrix cracking.







38

2.6 Intermetallic Composite Fabrication



The various processes developed to fabricate synthetic intermetallic matrix composites

can be separated into two groups. The first group is characterized as a casting process. Thus

the intermetallic material is first melted, then infiltrated into a preform of the reinforcement

material. The preform is usually preheated to enhance the flow of the molten material and to

help the intermetallic to "wet" the fiber surfaces. The driving force for infiltration is usually

provided by a vacuum, pressurized gas, gravity, electromagnetic levitation or any combination

of the above. The main disadvantages of this molten processing result from the high melting

temperatures involved, i.e., very few refractories can withstand these temperatures for any

length of time and if they can, there is an enhanced reactivity between the molten intermetallic

and the refractory material. This signals a definite advantage for electromagnetic levitation,

which is currently in the early stages of application. Therefore, we are currently limited to

the lower melting, less reactive intermetallics for liquid infiltration processes.

The second group of fabrication methods is powder processing. The methods

developed to fabricate intermetallic matrix composites via powder metallurgy (PM) procedures

are characterized by the application of pressure and heat for a definite length of time to the

powder/reinforcement mixture. The powders can be prealloyed to the desired composition or

a mixture of elemental powders in the stoichiometric ratio which react during processing to

form the matrix.

During reactive processing, a transient liquid may form if one of the constituents is a

low melting metal, such as aluminum or silicon, or if an intermediate low temperature

eutectic exists in the system. This has been utilized to form many of the aluminides [Dot94]

and usually results in lower processing temperatures and pressures to achieve full







39

densification. Liquid Phase and solid state sintering are usually carried out in a hot press or

hot isostatic press and some variation of atmospheres have been reported [Alm91, Ger89]

although vacuum seems to be preferred.

The reinforcements can be random or aligned. The fabrication of continuous fiber-

reinforced intermetallic matrix composites via the foil/fiber/foil techniques has been shown as

one of the most effective approaches. However, the foil is produced by rolling and the high-

temperature intermetallics are far too brittle to be rolled to a thin foil. Thus PM techniques

seem uniquely suited to the fabrication of IMCs.

Continuous fiber reinforced MoSi2 has been produced by the powder coated filament

technique [Lu91]. In this process, a filament is coated by drawing through a slurry of powder

mixed with a binder. After drying, the coated filaments are stacked and HIPed to burn out of

the binder and densify the matrix. This technique has also been applied to the fabrication of

W-Re reinforced TaTiAl2 and Al203 reinforced TiAl [Lu91]. There are, however, several

drawbacks to this technique; (1) handling of the coated fibers (2) low deposition rates (3)

controlling fiber spacing and (4) controlling matrix volume fraction.

Other PM techniques developed to produce brittle matrix composites include the

powder cloth process [Mac91] and the cold spraying technique [Yan91]. In the powder cloth

process, the matrix powder is combined with a fugitive binder into a flexible cloth-like sheet.

These mats are stacked alternately with continuous fiber mats and consolidated via hot

pressing or HIPing. In the cold spraying technique, the powder/binder slurry is deposited

onto the fiber mat to form a flexible tape.

A relatively new process to consolidate prealloyed powders is explosive compaction,

or shockwave synthesis [Bow88]. In this process the cold compact is consolidated when the

force from an explosive detonation is transferred to the cold compact and as the shockwave







40

propagates through the powders, very high frictional forces are generated which cause a

welding of the powders.

A final method, patented by Okazaki, et al. [Oka90] is electric discharge compaction.

The powders are assembled into a copper die, then a capacitor is discharged through the

powder. The large surface resistance of the powders causes interfacial melting and thus a

small pressure on the copper die consolidates the material.

The processes which utilize prealloyed powders are generally slower and require

higher input temperatures (or energies) for a corresponding degree of consolidation, whereas

the reactive processes exploit the thermodynamic properties of the reactants to produce heat

and aid in the consolidation process. Prealloyed powders are not readily available for many

compositions being considered as candidate materials and when they are, they are usually

much more expensive than the corresponding elemental powders. Therefore, the use of

reactive sintering is gaining widespread use. The various parameters available through

reactive sintering to control the reaction and thus the microstructure will be described in detail

for the B-NiAl system below.



2.7 Self-Propagating High-Temperature Synthesis



Self-propagating high-temperature synthesis (SHS) is a general processing technique in

which reactants, usually elemental constituents, when ignited, spontaneously transform to

products due to the exothermic heat of formation. Both reactants and products are condensed

phases. Several other terms are also applied to this process, such as gasless combustion

synthesis, self-propagating combustion and self-sustaining synthesis. SHS has received

considerable attention as an alternative to conventional powder metallurgy (PM) and ceramic






41

processing due to several advantages. These potential advantages include lower energy

consumption resulting from less heat input and shorter processing times, lower capital

investment resulting from process simplicity, and higher product purity due to less reaction

with the processing environment. The steps involved in an SHS process, schematically shown

in Figure 2.7, are elemental powder preparation, cold compaction, and ignition/combustion.

At this point, the formation of the product composition is complete although there are

generally unacceptably high levels of porosity.

The study of utilizing exothermic solid-solid reactions for the production of refractory

compounds such as oxides, nitrides and borides has been reported by former Soviet

researchers for over two decades [Fra85]. Others have reported the successful fabrication of

intermetallics such as MoSi2 [Bha92, Dee92] other silicides such as MosSi3, Ti5Si3 and ZrSi

and various aluminides, notably NbAl3 [Bar90, Lu90a], NiAl [Alm93], Ni3Al [Ger89] and

TiAl [Mol90].

The SHS reaction is characterized by the relationships between the melting and boiling

temperatures of the reactants and the adiabatic temperature of the reaction [Sub92]. The

adiabatic temperature is the peak temperature to which the products are raised under adiabatic

conditions due to the heat evolved by the exothermic reaction. This heat drives the

propagation of the combustion wave by heating the adjacent reactants. Merzhanov [Sub92]

devised a classification system to describe SHS reaction mechanism by the state of the

reactants (solid, liquid or gas) at the reaction adiabatic temperature. When the adiabatic

temperature lies between the melting point of the reactants, the molten reactant spreads at a

high rate throughout the compact, resulting in the highest velocity of combustion. This is also

known as liquid phase sintering (LPS). Complete solid state combustion occurs when the
















00

* 0
* ,
0.a

(a)


Figure 2.7. Typical steps in self-propagating high-temperature synthesis process are (a) blend
powders; (b) cold compaction to controlled green density; (c) ignition; (d) combustion wave
propagates through body; (e) final product.







43

adiabatic temperature is less than both reactant melting temperatures, resulting in the lowest

combustion velocities.

SHS composites are formed by either mechanically adding the reinforcement to the

reactant mixture or when the product consists of two or more materials formed from the

reactants. Both of these are the basis of specific processes which have been developed to

produce certain IMCs which are described in detail below.

In pure SHS, the reaction rate and the state of the product are determined by the

thermodynamics and kinetics of the combustion reaction. In many cases, this occurs in an

explosive manner leading to a product which, although it is of the desired composition, the

physical form has been degraded by excessive heating, or it contains large amounts of

porosity and must be further processed to generate the desired densification and physical

form.



Reactive Hot Compaction

Reactive hot compaction (RHC) is the name given to a group of processes developed

from SHS which make use of external variables to control the reaction so that a more useful

product is formed. RHC is a volumetric combustion process as the reactants are heated

uniformly within a die to the reaction temperature. Thus, the reaction initiates at many sites

throughout the mixture of reactants. The application of pressure is utilized to aid

densification and the addition of prereacted product or an inert material, such as A1203

particles has been shown to slow the reaction by absorbing some of the heat of formation of

the product [Alm91, Dot93].

Figure 2.8 shows a schematic RHC process cycle as applied to NiAl and NbAl3. The

important variables used to control the reaction rate and consolidation are: the shape and size







44

distribution of the reactants, the shape and size distribution of the inert particles (if any),

green density of the compact, heating rate, maximum temperature, the timing and magnitude

of the externally applied load and the atmosphere. As a subgroup of SHS, RHC shares many

features in common; the reactants are blended and usually cold compacted prior to initiating

the reaction. The degree of cold compaction determined as a percentage of theoretical density

can have a great impact on the reaction rate and sequence. Deevi [Dee92] has reported that

for Mo + 2Si powders, increasing the green density of the cold compact from 51% to 59%

of theoretical resulted in the possibility of forming MoSi2 directly without any of the

intermediate Mo5Si3 phase. It appears that increasing the density improves the interfacial

contact between the reactant particles, providing more sites for the initiation of the reaction,

decreasing the distances required for mass transport, thus decreasing compositional fluctuation

in the product. On the other hand, higher interparticle contact also increases the effective

heat transfer of the powder mass, leading to faster conduction of the heat of formation from

the interface. This can lead to loss of the self-propagating characteristic in some RHC

systems [Lu90a].

The effect of particle size on the structure of the product during RHC can be quite

dramatic. In producing NbAI3, Lu [Lu90] found that by increasing the Nb particle size

distribution from < 10 im to 10-30 tm, the as-reacted microstructure contained a second

phase of Nb2Al with a central Nb core, whereas the smaller Nb particles produced only the 2

phase NbAl3-Nb2Al mixture. Subsequent annealing eliminated the Nb cores via a diffusion

controlled process and increased the volume fraction of Nb2AI.

Figure 2.9 shows the effect of heating rate on the temperatures and peak heights on

DTA scans during RHC of NbAI3 [Lu90]. Increasing the heating rate is seen to initially shift

the initiation of the formation reaction to higher temperatures, then it drops back somewhat.



















50 MPa

1250 C






I


Furnace Cool
--- 0


Time (minutes)


49 MPa
/ 350----------
/ 1350 C
- .


20 80
Time, min.


Figure 2.8. Schematic of hot press cycle to form (a) NiAl and (b) NbAI3 by reactive hot
compaction and typical microstructures. Difference in the application of pressure is due to
the absence of liquid phase in processing NiAl.







46
















20


30 "C/mln.
U
Is

"2/ to C /mln



I-


E 5





0 250 500 750 1000 1250 1500
Temperature C





Figure 2.9. DTA profiles showing the effect of increasing heating rate or Nb and Al
elemental powder mixtures.







47

This is probably the result of two competing processes; at very low heating rates, intermediate

products have time to form via diffusion controlled solid-solid reactions, as the heating rate is

raised, the initiation is shifted to higher temperatures due to the inertia of the system but

coincidentally there is less time for intermediate product formation and thus reducing that

barrier to the main reaction initiation. At some heating rate, here between 200C/min. and

300C/min., the intermediate reaction product becomes negligible eliminating this barrier to

the reaction initiation. Also, increasing the heating rate increases the peak temperature.

Consequently, the synthesis reaction is accelerated by higher diffusivity and shorter diffusion

distances [Bos88, Kis57, Phi87].

Philpot, et al. [Phi87] studied the effect of heating rate on the combustion synthesis of

nickel aluminides. They found two exotherms at very low heating rates, whose temperatures

increased with heating rates; at higher rates (> 20C/min.) only one exotherm occurred. In

both cases, the first (or only) peak occurred below the melting point of Al, indicating a solid-

solid reaction. In all cases where two peaks were observed, the highest temperature of the

first peak was always below the initiation temperature of the second peak. They concluded

that when the heat generated by the first reaction is sufficiently high, it initiates the second

(liquid phase) reaction and thus changes the system from two peaks to one, which

incorporates both reactions.

The effects of externally applied pressure is related to the green density effect, as

mentioned above. In addition, the timing of the pressure application relative to the reaction

sequence plays an important role in the densification of the product. It seems that the

application of the pressure during the transient reactions is the most appropriate, as earlier

pressurization might lead to premature or incomplete reactions, whereas pressurization after







48

the reactions would only be hot deformation of a porous mass, requiring further solid-state

diffusion to achieve full density.

Transient liquid phase sintering, or liquid phase sintering (LPS) is a variation of RHC

in which one of the reactants melts, or the system goes through a low temperature eutectic at

temperature below the ignition temperature during heat up. This molten constituent easily

flows throughout the compact by capillary forces and fills in many of the inter-particle voids

aiding in densification and increasing the reaction rate. Anton [Ant88] also reports the

elimination of severe fiber damage normally associated with the consolidation of brittle fiber

reinforced IMCs when LPS techniques are employed. In this process, external pressure was

applied to the system at the melting temperature of Al in FP Alumina/TaAl,, prior to the

initiation of the intermetallic-forming reaction, followed by a further homogenizing anneal at

12000C, presumably still under pressure. This reportedly resulted in a fully dense

microstructure. Lu [Lu90a] on the other hand, waited until after the reaction was complete to

apply external pressure to RHC Nb/NbAl3 composites and reported less then 2% porosity in

the microstructure. In DRC [Dot93] the pressure is applied at room temperature prior to

heating. This eliminates Al melting since the increased contact causes the NiAl-formation

reaction to initiate at 6000C, and results in a fully dense microstructure after holding at

8000C for 30 minutes. In this reaction sequence, the microstructure is not homogeneous but

contains a mixture of Ni3Al and off stoichiometry NiAl. This is homogenized by a 30 minute

anneal at 12000C.



Solid-Solid Reactions

Solid-solid reactions occur when the adiabatic temperature is below the melting

temperatures of the constituents. Thus, the reaction initiates and propagates throughout the







49

compact without the formation of a separate liquid phase, although the intense localized

heating caused by highly exothermic intermetallic reactions can give rise to a local fusion

zone which is quickly cooled by conduction. The presence of local fusion zones is the

probable mechanism allowing 100% densification in very short processing times and is the

subject of the present research. This system normally forms by LPS as the Al melts below

the reaction temperature, but the application of pressure is shown to initiate the reaction prior

to Al melting.

In-situ synthesis of MoSi2-SiC composites via solid state displacement reactions

[Hen92] and co-synthesis [Alm93] have been recently reported. The former makes use of the

reaction Mo2C + Si -> MoSi2 + SiC, while the latter utilizes the elemental powders to

produce the same phases.

Mechanical alloying is a variation of the solid-solid reactive processing of

intermetallics. In this process elemental powders are mixed at very high energy levels in a

ball mixer. The impact and grinding of the balls on the powder provides the energy to initiate

the reaction sporadically during the process and to reduce the particle size of the product.

Bieler, et al. [Bie92] reported mechanically alloying Ni and Al powder in liquid nitrogen to

eliminate the heat of reaction (cryomilling) and reported the formation of NiAl-AIN

nanocomposites via reaction synthesis.



2.8 Development of Composites via RHC Processes



Each alloy/reinforcement system presents a unique set of thermodynamic and kinetic

factors. For successful intermetallic matrix composites to be fabricated via RHC, the process

must be optimized based upon the specifics of the application. These parameters not only







50

control the reaction sequence and rate but also the density and the microstructure of the

resulting IMC. Many of the IMC systems under development are artificial composites, and

thus a state of non-equilibrium likely exists between matrix and reinforcement. It is therefore

necessary to control the interaction between the reinforcement and the matrix via application

of a diffusion barrier coating at the interface. The interfacial coating must perform at least 3

functions. First, it must prevent the chemical interaction between the matrix and the

reinforcement throughout the temperature range of exposure, both during manufacture and in

service. The coating must also provide the proper degree of bonding required by the

composite to aid strengthening and/or toughening since the interface has been shown to

control fracture mechanisms [Eva89, Mat89, Xia93]. Finally, the coating must also withstand

and transfer the stresses induced by differential expansion rates of the matrix and

reinforcement due to thermal cycling and mechanical loading without degradation.

An example of such an interface coating during RHC was reported for ductile filament

reinforced NiAl [Dot93] and NbAl3 [Lu90a]. In both of these cases, the ductile reinforcement

(Nb) was preoxidized to form a surface layer of Nb205. Subsequent elevated temperature

processing converted the Nb205 to A1203 prior to any discernable matrix/reinforcement

interaction. Although the results were similar, the processing routes, necessitated by the

individual systems were different. For NbAl3, heating caused the melting of Al which spreads

throughout the remaining Nb powder, providing ample opportunity for both transient liquid

phase sintering to aid densification and direct contact between liquid Al and the oxidized Nb

surfaces. This allowed a uniform A1203 interface to develop, followed by diffusion controlled

completion of the conversion from Nb205 to A1203. In NiAl on the other hand, the synthesis

reaction initiates prior to the melting of aluminum and proceeds by generating intense

localized heat which may cause a very small local molten zone briefly as the reaction







51

progresses before the heat is conducted away. It seems that the wide solubility range of NiAl

allows easy transfer of Al atoms to the Nb reinforcement to reduce the NbO, to A1203.













CHAPTER 3
EXPERIMENTAL PROCEDURE


The reactive hot compaction (RHC) process has been adapted to produce in-situ

networks of alumina reinforcements in a NiAl matrix. The process begins with an oxidation

pretreatment of elemental nickel or aluminum powder. The powders are then blended in the

stoichiometric ratio and diluted with 20 percent prealloyed NiAI powder. The powder

mixture is then placed in a boron nitride-lined graphite die and hot pressed.

During hot pressing, the powder specimen is put under-50 MPa pressure at room

temperature as the chamber is evacuated. When the pressure reaches 104 torr, the heating

coil is energized to begin heating. As the powder heats up, the pressure on the sample is

maintained at 50 MPa. The specimen is held at 12500C for 25 minutes and then the pressure

on the specimen is removed. After 5 additional minutes dwell time, the specimen is cooled at

5C/minute to room temperature.

The following describe the analytical techniques used to develop this process and to

characterize the resulting microcomposites.



3.1 Materials



The starting materials for all the composites consisted of elemental Nickel (Ni),

elemental Aluminum (Al) and prereacted NiAl powders. The elemental Ni powder, shown in

Figure 3.1 was supplied by Alfa Products. The average particle diameter was 2.2 3.0 tm







53

and contained >99.9% Ni, <0.2% C (analyses are given in weight percent based on the

metallic content of the powders). The aluminum powders, supplied by Valiment, Inc., were

of two size distributions, averaging 4.0 /m and 11.5 /m diameter. These He atomized

spherical particles are shown in Figure 3.2 and contained 99.7% Al, with Si as the major

impurity element. Finally, the NiAl prereacted powder was supplied by Xform, Incorporated

and is shown in Figure 3.3. This powder was reactively atomized in an inert atmosphere by a

proprietary process and formed particles containing 48 52 atomic percent Al. The nominal

composition was 50 a/o +/- 0.5 Al and was screened to < 10 jm diameter.



3.2 Oxidation



Oxidation of the powders was carried out in a tube furnace with both ends open to the

laboratory atmosphere for Ni and NiAl powders. For the Al powder, however, it was

determined that a higher moisture content was required to form a coherent surface oxide

coating. Thus, the set-up, shown in Figure 3.4, was devised whereby compressed air was

slowly bubbled (3.0 SCFH) through water in a filtering flask and subsequently introduced into

one end of the tube furnace. The flask was heated on a hot plate to raise the moisture content

of the furnace atmosphere to controlled, higher levels.

The furnace was preheated to the desired temperature and the powder was placed into

unglazed alumina boats. The boats were preheated to prevent thermal cracking then inserted

into the hot zone of the furnace.

Initially, the powders oxidized non-uniformly as evidenced by a change in color of the

Ni powder. Since the oxide growth rate is inversely proportional to the oxide thickness, this

process becomes self-regulating, ensuring even oxidation throughout the powder.













































Figure 3.1. Elemental Ni powder, as-received.

























































Figure 3.2. Elemental Al powder, as-received (a) 4.0 tm and (b) 11.5 /m average diameter.













































Figure 3.3. Prereacted NiAl powder, as-received.







57

Two precautions were necessary to give uniform alumina distribution in the final

composite microstructure. First, the Ni oxidation reaction is highly exothermic. Therefore,

if sufficient cooling is not provided via air circulation through the furnace, the powder tends

to heat up and accelerate the reaction, thus giving uncontrollable levels of NiO on the powder

and, consequently, uncontrollable levels of A1203 in the composite. This is accomplished by

maintaining air circulation and reducing the mass of Ni powder. A powder depth of 2 mm

was found to be workable, reducing the risk of uncontrolled reaction, while providing for

reasonable production rates.

The second processing difficulty comes from the tendency of both the Ni and Al

powders to agglomerate during the oxidation treatment. These can result in large

inhomogeneities in the NiAl composition of the hot pressed composite leading to a two or

three-phase structure. Preliminary findings show these powder clusters to be oxidized to the

same degree as the loose powder, thus allowing for them to be broken down prior to blending

with the other powders. For the bulk of the present study, the agglomerates were sieved out

with a 100 mesh screen and discarded.



3.3 Composite Fabrication



All composite specimens were fabricated by weighing the powders in the proper

proportions to produce a product of the desired composition. When pretreated powders were

utilized, the ratios were altered to reflect the expected amount of alumina such that the NiAl

matrix composition was 50 atomic percent aluminum +2 a/o. For the 38.1 mm diameter die,

a total weight of 38 grams would produce a specimen approximately 6 mm thick. Thus,

surface irregularities could be ground off while allowing for the 5.1 mm four-point bend












































Figure 3.4. Tube furnace set up for powder preoxidation treatment.









specimen height.

The powders were blended for one hour on a roller mixer, utilizing three steel balls,

3/8 inch in diameter to aid in breaking up of powder lumps. After mixing, the steel balls

were removed and the powder was sieved through a #100 screen (150 Am opening) to remove

any large agglomerations.



Hot Press

The weighed powder was placed in a boron nitride-lined graphite die (Figure 3.5).

The green density of the loose powder was increased to 55 60% of theoretical by gently

tapping the powder-containing die assembly on a hard surface for approximately 2 minutes

(tap density). The upper graphite punch was then fitted into the die and the whole assembly

was loaded into a Centorr vacuum hot press, as in Figure 3.6. The hot press ram was

controlled by manual hydraulic pump fitted with a pressure gage. The die assembly was

loaded to 50 MPa followed by evacuation of the hot press chamber.

When the pressure in the chamber reached < 103 torr, the ram pressure was adjusted

back to 50 MPa to compensate for any drop during the approximately 30 minute evacuation

interval.

At this point, the radio frequency (RF) motor generator was energized and power sent

to the hot press coil to begin heating the die assembly.

The temperature was monitored via optical pyrometer sighted through a quartz access

window in the hot press chamber. Several tests were run utilizing type K thermocouples

inserted into the graphite die to obtain more precise temperature measurement at low

temperatures and for optical pyrometer calibration. Temperature correspondence was less

than + 10C.





Figure 3.5. Schematic of graphite die setup.


(h)


Figure 3.6. Placement of die in induction coil of hot press. Parts shown include: (a) hot press
ram; (b) spacer; (c) graphite die assembly; (d) quartz retainer; (e) pyrometer sight hole;
(f) RF coil; (g) graphite base and (h) vacuum chamber.


F High Strength Graphite
B Grafoil
0 Specimen








61

3.4 Differential Thermal Analysis



The reaction sequence and the effect of the external variables, such as heating rate and

preload on the reaction kinetics were explored by differential thermal analysis (DTA) and by

interrupting the hot press cycle at various points to evaluate the microstructural evolution of

the specimens.

DTA and thermogravimetric analysis (TGA) were performed simultaneously in a

Harrop model ST-736 analyzer on powder blends which consisted of the following:

1. All powders, no oxidation treatment;

2. Al powder oxidized 200 hr. in dry air, the remaining untreated;

3. Al powder oxidized 200 hr. in moist air, the remaining untreated;

4. Ni powder oxidized 20 min. in moist air, the remaining untreated

note: All blends contain 20 % prealloyed NiAl powder, unless specifically

stated otherwise.

The heating rates were 5, 10, and 400C per minute and the preload was simulated by

cold pressing the powder blend in a steel die to 10 and 50 MPaV/m and holding for one

minute. The green compact was subsequently transferred to an alumina crucible and loaded

into the Harrop DTA/TGA and tested at the desired heating rate. Zero preload was tested by

carefully scooping the powder blend into the crucible and striking it off even with the top of

the crucible, then loading the crucible into the DTA furnace.

A baseline was collected for each heating rate by testing two alumina standards

simultaneously, allowing the background temperature differential due to the instrument to be

subtracted from the experimental results. This allows easier identification of endotherms and

exotherms associated with reaction of the powders.







62

All tests were run in an argon atmosphere flowing at 5.0 SCFH, unless otherwise

noted and the temperature difference (AT) and weight of the specimen were digitally recorded

as a function of specimen temperature.



3.5 Annealing/Heat Treating



To ensure complete homogenization and to compare the various composites based on

the lowest NiAl fracture toughness state, samples from each composite were annealed at

12000C for 100 hours in oxygen-gettered argon in a tube furnace. The samples were cooled

at 3 C/min. to room temperature. These samples were subsequently four-point bend tested

for comparison to as-hot pressed specimens.

Heat treated specimens were solutionized at 12000C, cooled at 30C/min. to 5000C,

then quenched. Air quenched specimens were dumped from the alumina boat onto a firebrick

and allowed to cool to room temperature, whereas, water quenched samples were taken from

the furnace and immediately dumped into a tank of water. These procedures resulted in

cooling rates of approximately 158 and 57000C/minute for air quenching and water

quenching, respectively, compared to 5 C/min. in the hot press and 3C/min. for furnace

cooling. In addition, several specimens were quenched directly from 12000C both in air and

into water.







63

3.6 Analysis Techniques



Metallography

All metallographic specimens were electric discharge machined (EDMed) from the hot

press disks and mounted in bakelite. After grinding to 600 grit on wet SiC paper, each

sample was diamond polished with 6 /m then 1 /m diamond paste. Finally, the specimens

were polished on a Buehler Vibromet 2 using 0.025 /m colloidal silica.

Metallographic specimens were subsequently analyzed in the as-polished condition, or

carbon coated with 100-200 A C via an evaporative coating technique or etched with Kroll's

Reagent to reveal the grain boundaries or deep etched with saturated molybolic acid for 20 -

25 minutes to reveal the alumina network. The compositions of the etches are given in Table

3.1.

Transmission electron microscopy (TEM) specimens were produced by EDMing a

3 mm diameter rod from a four-point bend specimen, then cutting 1 2 mm thick slices from

the 3 mm rods with a low speed diamond saw. These disks were ground to 100 150 Jtm

thickness on 600 grit SiC paper then dimpled to 15 20 /m. The disks were finally ion

milled to penetration with a gun angle of 150. After penetration the ion gun was adjusted to

12 and the disk was milled for an additional five minutes.

The various microstructure features were analyzed by a combination of optical (Nikon

Epiphot), SEM (Jeol JSM-35CF), electron microprobe (Jeol Superprobe 733), TEM (Jeol

JEM-200CX), X-ray diffraction (Philips APD 3720), Auger electron spectroscopy (Perkin

Elmer PHI 660 Scanning Auger Multiprobe) and secondary ion mass spectrometry (Perkin

Elmer PHI 6600 SIMS System).










Table 3.1: Etches for Metallographic Specimens


Image Analysis

An estimate of the volume fraction of the alumina in each specimen was established

by determining the area fraction in a total of 10 fields on a Cambridge Instruments

Quantimet 520 image analysis system. Hilliard [Hil68] has shown that the average area

fraction measured in a random test plane is an unbiased estimate of the volume fraction of an

arbitrarily shaped second phase constituent in a multi-phase material. That a photometric

system can be effectively utilized to make this measurement depends on a high degree of

uniformity in the contrast between the measured phase and the rest of the microstructure. For

the present system, backscattered electron images (BEI) taken on the SEM provide the ideal

consistency in contrast required. Thus, for consistency, each specimen was sectioned at the

same position and subsequently polished following identical procedures. Finally, the

specimens were carbon coated and 5 separate fields were photographed at 1500x in

backscattered mode. Each photomicrograph gave 2 fields of analysis on the Quantimet. The

volume fractions reported are the sum of the ten area totals of alumina divided by the total

area analyzed. This estimate is repeatable to +/- 10% of the reported value.


Etch Name Kroll's Reagent Saturated Molybolic Acid

Purpose To reveal grain structure To dissolve NiAl matrix to
reveal Al2O3 network structure

Formula 7ml HF + 14ml HNO3 + 100gm MoO3 + 50ml HF +
100ml HO 150ml HO

Technique Swab for 20-30 seconds, Immerse in etchant in ultrasonic
rinse with water, then cleaner for 20-25 minutes, rinse
methanol, then blow dry. with water, then methanol, then
blow dry.










Density Measurement

The density for each composite was measured and compared to the density calculated

from the rule of mixtures from both the volume fraction of alumina and the thermal expansion

coefficient. Both of the latter calculated values assumed a two component system comprised

of a mixture of pure a-Al203 in pure stoichiometric (/-NiAl.

The experimentally determined density was established by the Archimedes method. In

this method, the sample is placed in boiling water for 10 minutes, cooled then weighed in

distilled water (s.g.= 1.0). Then the samples were dried overnight at 105C and subsequently

cooled to room temperature in a desiccator and weighed in air. The apparent density is

calculated from the following expression:


wt .
s.g.= (3.1)
wt. -wt
air water



Dilatometry

Following fracture toughness testing, a sample was prepared by grinding the ends of

one fractured specimen to produce a rectangular bar of length 10.0 +/.05 mm. The ends

were polished to 600 grit on wet SiC paper. Thus, each specimen had dimensions of 10.0

mm x 5.1 mm x 3.8 mm. Fully annealed (100 hr. at 12000C) samples were utilized to

reduce the influence of residual stress and compositional homogenization on the measured

expansion.

The thermal expansion behavior of these specimens was characterized in a Theta

Industries, Dilatronic@ II research differential dilatometer. This system utilized the difference

in expansion between an unknown and reference specimen. In this work a sapphire crystal

was used as the reference. A primary advantage of differential dilatometry is that a single test








66

can suffice to determine the difference in expansion between the two materials, as well as the

absolute expansion of the unknown. Comparison to a standard increases the accuracy by

reducing the variations due to thermal gradients due to high heating rates.

For differential dilatometry, the specimen and the standard were placed in the hot

zone with a thermocouple between them. Parallel solid alumina push rods contacted both

specimens with 20 25 gm force. The opposite ends of these rods contacted a linear variable

differential transducer (LVDT) located in the water cooled lead. During the test, argon is

flowed through the specimen chamber as the furnace temperature was ramped at 3C/minute.

The expansion of the specimens (and the hot end of the alumina rods) caused a voltage to be

generated by the LVDT. This signal is stored to a computer for subsequent analysis.

The expansion of the specimen was recorded for heating and cooling at 3 C/minute.

The peak furnace temperature was 1100"C. The data determined by this test included the

absolute expansion coefficient, in ppm (mm/mm) and the average thermal expansion

coefficient, a, in ppm/C (mm/mm"C x 106).



Grain Size

The average diameter of the NiAl grains were estimated by the line intercept method.

The microstructure was etched with Kroll's Reagent and viewed through an optical

microscope at 500x. A reticulated glass slide with an etched line 250[km long was inserted

into the microscope optical system. The number of times this line intercepted grain

boundaries was counted for 20 random orientations along the polished plane of the specimen.

Thus, the grain size (G.S.) in microns was calculated according to the lineal analysis first

proposed by Roiswal [Roi98]:
G.S.=20 x250/n (3.2)

where n is the total number of grain boundaries intercepted for all 20 measurements.







67

3.7 Fracture Toughness Measurement



Four-point bend test fracture specimens were EDMed from the as-hot pressed disks

and ground to the dimensions given in Figure 3.8 on wet SiC paper to 240 grit surface finish.

Chevron notches were introduced as shown in Figure 3.8 by carefully making two cuts with a

diamond watering blade. Finally, all specimens were ultrasonically cleaned in acetone for ten

minutes prior to testing.

The bend tests were conducted on an Instron model 1125 testing machine at constant

cross-head speeds of .002, .005 and .02 inches per minute at room temperature in air. The

inner and outer spans, 10 and 20 mm, respectively, were applied to the specimen via SiC

roller fixtures which were free to tilt and rotate to eliminate any lateral forces during the

application of the testing load.

The peak load, P,, from the load vs. displacement curves was utilized to estimate

the fracture toughness, Km., from Munz [Mun80] by:


K =- Y (3.3)



where B and W are the width and height of the bar, respectively and y is the minimum

value of the dimensionless stress intensity factor coefficient as a function of relative crack

length for the particular specimen used. Y was determined using a straight through

crack specimen as [Sra76]:


S,-S Sl-S2 ~,-
Y .=(3.08 +5o+8.33ao)( 2)[1+0.007( 2) z[ 0] (3.4)
W W2 1-a0







68
where a =a, a= aW SI is the outer span (20 mm), S2 is the inner span (10 mm). a. and
W W

a, are defined in Figure 3.7.

This allows the direct calculation of fracture toughness from the peak load and the

specimen geometry.





















Hot Pressed Disk
38 mm diameter


Figure 3.7. Four-point bend specimen showing location within hot pressed disk and chevron
notch dimensions.













CHAPTER 4
RESULTS AND DISCUSSION: PROCESSING



Reactive hot compaction was utilized to fabricate NiAl and NiAl matrix composites

reinforced with alumina. The process consisted of blending the elemental powders and hot

pressing under pressure. In this work, both monolithic NiAl and NiAl-Al203 composites were

fabricated. The former was produced from as-received aluminum and nickel powders and

contained a small amount of A1203 as impurity particles produced from the reduction of

oxides present on the powder surfaces. The latter were synthesized by first pretreating one of

the elemental powders in an oxidizing atmosphere, then reacting as before. Several A1203

reinforcement morphologies, including an interconnected network were produced which were

dependent on specific conditions prevailing during the preoxidation treatment. It was shown

that control of the reinforcement morphology as well as the interfacial bond strength was

possible by controlling the oxidation conditions which enabled the production of toughened

NiAl-A1203 composites.



4.1 Oxidation



Throughout this study, all powders were oxidized in air at elevated temperatures. The

complex effects of the oxidation treatments were evaluated in a number of different ways

including physical appearance, X-ray diffraction, volume fraction/morphology of alumina in

the resulting composites, weight gain and infrared spectroscopy.







71

No attempt was made to characterize the oxide layer on the surfaces of the as-received

powders. However, Valimet, Inc. has measured the alumina surface on the Al powder at 30

A thick and they have shown this thickness to be stable for at least five years. This was

assumed to be the starting condition whereas the surface oxide on the nickel powder is

assumed to be negligible.



Oxidation of Nickel Powder

It was found that due to the exothermic nature of the reaction Ni + O NiO,

circulation of the furnace atmosphere was necessary to prevent overheating of the powder

mass, which caused uncontrollable oxidation of the entire mass to NiO. To prevent the build-

up of hot pockets during oxidation, the Ni powder was spread evenly across the alumina boats

and the depth was kept below 2 mm to ensure the circulating furnace atmosphere would

provide adequate cooling.

During the oxidation process, the small "peaks" in the powder darkened first, as they

were first to reach the furnace temperature, followed by a change to a more uniform brown

color. With more exposure the color changed to black, then steel blue, then gray-green and

finally to a dark olive green in the most heavily oxidized powders. This change in color was

indicative of a changing of the reflective properties of the powder surface due to the growing

oxide shell. The shorter time intervals of 2 and 5 minutes resulted in uneven oxidation of the

mass of powder.

Oxidation of Ni was found to be much more controllable than Al mainly due to the

formation of only one stable oxidation product, and thus lent itself to a more detailed analysis.

The oxidation behavior of Ni indicates it to be in a class known as pyrophoric metals

which are characterized by a high heat of oxidation and a tendency towards linear oxidation.







72

In addition, the oxidation rate is inversely proportional to oxide thickness. This provides a

self-regulating mechanism which insures even oxidation throughout the powder mass.

Physical appearance. Figure 4.1 shows the physical appearance of the powder

surfaces in the "as-received" and "oxidized" conditions. The oxidation treatment obviously

caused the formation of a lower density oxide surface which has hidden many of the former

surface details. The oxide coating appears to completely cover the powder surfaces and

adhere to the surface well.

X-ray diffraction. The x-ray diffraction spectra from nickel powders are given in

Figure 4.2. In this figure, the intensity of the diffracted beam is shown as a function of

diffraction angle, 20. The top and bottom spectra are JCPDS [Joi85] standards for NiO and

Ni, respectively. The measured spectra, beginning with the bottom were produced from as-

received Ni powder, Ni oxidized for 20 minutes at 5000C, 525C and 550C, respectively.

Increasing the oxidation temperature to 5000C, 525C and 550C results in a very rapid rise

in the relative intensities of the NiO/Ni peaks. This indicates a rapid rise in the volume of

NiO and is supported by the large quantity of A1203 formed in the composite from oxidized

nickel powder. In addition, the effect of exposure time at a given temperature resulted in

rapid growth of an NiO shell. This is presented in more detail in the mass gain studies,

below.



Oxidation kinetics

In order to evaluate the oxidation kinetics of the Ni powder, a weight gain study on

samples of the powder was undertaken. The oxidizing conditions for the nickel powder

consisted of convection driven air flow through a three zone tube furnace held at the

temperature indicated. The center zone was reserved for the alumina boats containing the





















































Figure 4.1. SEM micrographs of Ni powders in both the as-received (a) & (b) and oxidized
(c) & (d) conditions.























































Figure 4.1--continued.


t,~FF''










60.0
40.0 N
20.0
20.0 L.._.._.. .._ _,----- ----.
0.0 20.0 40.0 60.0 80.0 100.0









1.62 L5500C

1.28

0.981 II
S 5250C
0.72

0.50

0.32 5000C

0.18-

0.081

0.02 As-received


0.0 20.0 40.0 60.0 80.0 100.0


100.0 Ni NICKEL. SYN
80.0 4- 850
60.0
40.0
20.0 _
0.0 20.0 40.0 60.0 80.0 100.0

Figure 4.2. X-ray diffraction spectra for Ni powder. The top and bottom spectra are JCPD
standards for for NiO and Ni, respectively. The center spectra correspond to Ni powder as
received and oxidized for 20 minutes at 5000C, 525C, 550C.







76

powders. The partial pressure of water in the atmosphere was estimated to be 0.061 atm. and

the NiO surface appeared to adhere well and no evidence of spelling was encountered in

subsequent handling.

Figure 4.3 shows the average weight gain, Am(g/cm2), calculated from the mass gain

data by assuming the nickel powder consisted of spherical particles equal to the average

powder diameter of 2.6 /m, as a function of exposure time at three different oxidation

temperatures. In this regime, the oxidation shows a paralinear relationship for all three

temperatures. Paralinear oxidation is a common example of a combination of two or more

relationships between weight gain and time on the same curve [Kub62]. In this case, the

nickel powder begins to oxidize parabolically then it continues in a linear fashion. It is

common for a clean metal surface to oxidize rapidly in the beginning, usually in a logarithmic

or inverse logarithmic fashion, until a stable film 20 50 A in thickness forms where the

oxidation rate drops rapidly. For the purposes of this work we are primarily interested in the

oxidation kinetics in the slow growth region which is linear, thus:


Am=kt (4.1)



where Am is the weight gain and k is a constant. These results indicate linear oxidation rates

for nickel powder of k5oo=4.04 x 10-5 g/cm2min, k525=8.96 x 105 g/cm2min and k550= 1.43 x

104 g/cm2min for oxidation at 500, 525 and 550C, respectively. For the analysis, the

starting powders were assumed to consist of spherical nickel particles of 2.6 /m diameter (the

average of the size distribution range provided with the suppliers' certificate of analysis).

However, Figure 4.1 clearly shows a much more irregular surface such that the surface area

was understated resulting in an overestimation of the oxidation rate.










1200



1000



CE 800
CE



400
200
c0
(3 600
,I)
U)

400



200



0


. I
S 500 C A
U 525 C
- A 550 C
/


/-

/
A / ..,"

I,,,
../ ., .y.:-- ---o


Time, Minutes


Figure 4.3. Plot of weight gain as a function of oxidation time for Ni powder oxidized in air
at 5000C, 5250C and 550C.









Oxidation of Aluminum Powder

Figure 4.4 presents the results from the DTA and TGA of the Al powder in a low

moisture content air up to the melting point of the powder. The DTA curve exhibits two

important peaks. The broad exotherm, indicated by the positive peak centered around 575C,

is due to the increasing reactivity of the powder with oxygen in the air. The negative peak,

beginning at 6600C, indicates an endothermic reaction and corresponds to the melting of

aluminum. The TGA curve shows small weight gain initiating around 4000C and increasing

up through the melting point. The peak in the DTA exotherm, near 575C, corresponds to

the maximum slope in the weight-gain curve and is interpreted as the maximum oxidation rate

and was thus chosen as the temperature for the bulk of the aluminum oxidation studies.

Some sintering of the powders occurred during the oxidation processes and various

techniques were employed to minimize this effect. A simple screening process after the

oxidation treatment was used to eliminate large clusters of particles from the powder blend

which otherwise would have led to large compositional inhomogeneities and the formation of

secondary phases in the microstructure during hot pressing.

Physical Appearance. Figure 4.5 shows aluminum powder in the as-received

condition. The powders exhibited a smooth spherical surface and a large distribution in size.

At high magnification, the presence of tiny particles, less than 0.1 ltm in diameter were seen

adhering to the powder surfaces.

In comparison, in Figure 4.6, the surface of aluminum powder oxidized in the "dry"

atmosphere is presented. In these, the surfaces show spallation has occurred, as indicated by

missing sections of surface from the spheres. These fine alumina flecks were carried through

the subsequent processing steps and later will be shown to be detrimental to the fracture

toughness of the composite. In contrast, when the oxidation was carried out in an atmosphere































- ~~-4-


TGA


\J DTA


2 :33 2 .3 33 .3C. 5S3 530 700
Toroorcture (*C)


BG 90 '"120 "ont
ODAPont


Figure 4.4. Differential thermal analysis (DTA) and thermogravimetric analysis (TGA) of Al
powder in air (pHO2). Exotherm at 575C apparently caused by reaction of the powder with
oxygen and moisture in the air.





















































(b)

Figure 4.5. SEM micrographs of Al powders in the as-received condition at (a) 2,000x and
(b) 20,000x magnification.







81





































Figure 4.6. SEM micrograph of Al powder oxidized in "dry" air (pH20=0.011) atmosphere
at 5750C showing excessive spelling of the oxide.








82

containing moisture, the oxide layer appeared to adhere well to the surface of the powder, as

shown in Figure 4.7 for aluminum powder oxidized in air at 5750C and 26.5% relative

humidity.

For the aluminum, a low moisture atmosphere resulted in considerable damage to the

oxide coating and thus the aluminum powders were oxidized at the four moisture levels

described in Table 4.1 to evaluate this effect. The estimated values of partial pressure in the

table are based on heating the inlet atmosphere with a known moisture content to the

oxidizing temperature of 575 C. Details of the effect of moisture are given later.




Table 4.1: Estimated oxidizing atmosphere moisture content
for the oxidizing conditions used.


Oxidizing dry air dry air bubbled moist air dry air
Conditions through cold bubbled
water through
boiling water

pH20 (atm) .011 .037 .061 .5*
at 5750C
*Order of magnitude estimate at furnace inlet. Actual pressure could not be calculated
due to condensation between flask and furnace.




X-ray diffraction. The X-ray diffraction spectra from Al powders oxidized in a dry

atmosphere and a humid atmosphere are given in Figure 4.8. From these, it is discernable

that the Al oxidizes only to y-Al203, and no other phases occur above the detection limit for

this technique. Since significant differences in mechanical properties of the subsequent

composites were found, Fourier Transform Infrared Spectroscopy (FTIR) was used to analyze

the surface oxides in more detail. These results will be presented later in this section.







83





















(a)











KM









(b)

Figure 4.7. Aluminum powders oxidized in air at 575C and pH20=0.061 at (a) 2,000x and
(b) 20,000x magnification.










100.0 A1203
80.0 11- 517
60.0i
40.0
20.0 ,20. ._ 1

0.0 20.0 40.0 60.0 80.0 100.0


0.81

0.64'

0.49-

0.36-

0.25 -

0.161

0.09-


Moist Atmosphere


20.0


40.0


60.0


100,0


80.0


0.04'

S Dry Atmosphere


0.0 20.0 40.0 60.0 80.0 100.0
0.0 20.0 40.0 60.0 80.0 100.0


100.01
80.0
60.0
40.0
20.0

0.0 20.0 40.0


Al ALUMINUM
4- 787


60.0


80.0


100.0


Figure 4.8. X-ray diffraction spectra for Al powder. Top and bottom spectra are JCPD
standards corresponding to --Al203 and aluminum, respectively. Measured spectra are for Al
powder oxidized in air at 575C with water partial pressure of (a) 0.011 and (b) 0.061 atm.


~___ ~







85

Weight gain. In order to examine the effects of isothermal oxidation of aluminum,

samples of the powder were oxidized at 5500C. The powder was initially weighed at room

temperature, then weighed again after soaking at 5500C for 2, 5 and 10 minutes. It was

found that the mass dropped for the first 5 minutes, then began to increase. This was due to

surface moisture on the powders which averaged 2.26 x 105 g/cm2 for three powder samples.

The minimum mass, obtained after 5 minutes was used to determine the weight gain data

shown in Figure 4.9. In this figure, two curves are presented which show the increase in

mass for aluminum powders oxidized in air with water partial pressures of 0.011 atm. and

0.037. The curves show very little effect of increasing the moisture content on the weight

gain kinetics in this range. Both powders exhibited rapid weight gain within the first 1.5

hours, then the rate of increase dropped off significantly. This is typical behavior for a

material which produces a protective oxide. It should be emphasized that the complex nature

of the catalytic effect of moisture on the oxidation, as discussed below, dictates only a

qualitative description of the oxidation or oxy-hydration of the aluminum in the present study.



Oxidation kinetics

Numerous previous studies have been dedicated to this endeavor and these results will

be referred to in subsequent discussion [Bec67, Bre70, Eld88]. For the current study, the

extent of oxidation of aluminum powder was measured as the volume fraction of alumina

formed in the composite produced by reacting preoxidized Al powder with as-received Ni and

NiAl powders. The contribution of the latter powders was assumed to be constant.

Figure 4.10 shows the effects of oxidizing time at temperature on the alumina "growth

rate" as measured by volume percent alumina in the final product. The alumina content

appears linear in this figure up to about 17 volume percent, then levels off. This data cannot




















3.0

i- Dry Air
2.5 o Humid Air


2.0 o o


1.5


1.0


0.5


0 .0 i I I I I I I I
0 10 20 30 40 50 60 70 80 90 100 110


Time (hours)





Figure 4.9. Oxidation of aluminum powder in air at 5500C with water partial pressures of
(a) 0.011 and (b) 0.037 atm.
















20


16


12


8


4 -



0 25 50 75 100 125 150 175 200

Time (hours)





Figure 4.10. Plot of percent alumina formed in composites as a function of oxidation time at
5500C. Water partial pressure was 0.061 atm.







88

be directly compared to results from typical oxidation studies due to the measurement

technique used to establish the increase in alumina. Typically, mass gain per unit surface area

or oxide thickness are measured to track oxidation rate, but in the present study this is

confounded with a larger addition of Al powder to maintain the stoichiometric composition of

the matrix. Thus, these results must be viewed with caution.

Effect of moisture. The nature of the effect of moisture on the oxidation of aluminum

and the hydration of oxidation products has been an on-going concern since the early days of

the commercial aluminum industry. A comprehensive review is given in [Alw76].

The effect of room-temperature hydration of the oxides with water was illustrated by a

series of three experiments. In the first test, the aluminum powders were oxidized in a dry

atmosphere and were referred to as the "dry" condition. In the second, the powders were

oxidized in a humid atmosphere and subsequently dried at 575 C for 10 hours and were

termed the "dehydrated" condition. Finally, in the last experiment, the powders were used in

the as-received condition. Figure 4.11 shows the effect of a 0.5 weight percent H20 addition

to powders subject to the various prior oxidizing conditions. In the case of the dehydrated

powder, the powder was divided into three groups to be hot pressed. In the first group, the

"dehydrated" Al was blended with elemental Ni and 20% prereacted NiAl and hot pressed

without any hydration treatment. In the second, the powders were blended, then 0.5 weight

percent H20 was added. This mixture was blended for one hour and immediately hot

pressed. In the final group the mixture was blended with water for one hour and then aged

for 48 hours before hot pressing. In the case of the dry and as-received aluminum powders,

specimens were hot pressed only in the untreated and the rehydrated + aged conditions. Two

important observations can be made from this data. First, the powder/water mixture required

an aging period before the effect on volume percent alumina was fully realized. This is seen

































Dehydrated


2 Untreated
0 Hydrated
o Hydrated + Aged








J7A ,


As-Received


Figure 4.11. Shows the effects of adding 0.5 weight percent distilled water to powder
mixtures on volume fraction of alumina produced. Aluminum powders were oxidized in dry
air (pH20=0.5) then dehydrated, and in the as-received condition.


30

25

20

15

10

5

0


Dry Air


Ir


II







90

in the data for the powder that was dehydrated (Figure 4.11). The sample which was hot

pressed about 1 hour after the water was mixed in with the powder, resulted in only a one

percent increase in alumina. Whereas, for the rehydrated and aged sample, for which the

mixture was allowed to age for 48 hours after mixing, the volume percent increased 23

volume percent further. This occurred for the oxidized as well as the as-received powders

and indicates a diffusion controlled process was responsible adsorption of the moisture into

the powder surfaces. In Chapter 5, it will be shown that the same degree of change is

experienced in fracture toughness when the powders are subjected to this hydration treatment.

In the foregoing discussion, the rehydration was performed on the Al + Ni + NiAl

powder mixture. Thus, the low temperature effect of moisture could not be isolated.

However, a series of experiments in which the aluminum powder was hydrated and aged, then

blended with the Ni and NiAl resulted in no difference in fracture toughness when the

powders were hot pressed from 30 minutes to 7 days after blending. Thus, it appears that the

aging reactions take place on the Al powders independent of the presence of either Ni or

NiAl.

Fourier transform infrared spectroscopy. In order to establish chemical and structural

differences in the oxidized surfaces, fourier transform infrared spectroscopy (FTIR) was

performed on samples of the oxidized aluminum powders.

Infrared (IR) spectra contain qualitative information about the nature of chemical

bonds present in the specimen [Hai67]. In FTIR, the frequency of the incident radiation is

varied over the time of exposure. The ratio of transmitted intensity to incident intensity is

plotted versus frequency or wave number to give the spectrum. Two such spectra are given

in Figure 4.12 for aluminum powders oxidized in dry air (a) and humid air (b). These results

show virtually identical spectra for wavenumbers of 4000 to 1000 cm-'. However, at lower









Oxidized in Dry Air
(a)


Oxidized in Moist Air
(b)


t600 3o00 300 200oo 2400 2600
WAVENUMBER


1600oo 100oo 00 00oo


Figure 4.12. FTIR spectra produced in oxidized aluminum powders (a) oxidized in dry air
(pH20=0.011) and (b) oxidized in moist air (pH20=0.061).







92

values, the spectra become quite different. At low levels of wavenumber, the spectra are

usually considered to reflect changes in the structure of the material, rather than the stretching

and rotation of bonds normally associated with higher wavenumbers. The absorption peak at

935 cm' in Figure 4.12(b) has been assigned to a symmetric stretch vibration of the Al -OH

bond in the surface hydroxide monolayer [Fir91]. That study did not extend the spectrum

below 700 cm' and therefore did not pick up the peak at 569 cm-' in Figure 4.12(a). The

general shape of these spectra compare most closely with those given for boehmite (y-

AlOOH) and gibbsite (y-AI(OH)3) in [Wef87] for Figures 4.12(a) and 4.12(b), respectively.

This identification corresponds favorably to the conditions used in the oxidation of the

powders, but exact identification requires the production of a known specimen for comparison

[Hai67].

Comparison of these spectra to published hydrated aluminum spectra indicates the

oxidation products formed on the powders in this study were likely complex structures,

consisting of several different reaction products. Complete identification of these products is

left for future studies in which the hydroxide state can also be optimized to the properties of

the composites produced.



Summary

The oxidation of nickel powder was found to produce uniform, adherent layers of

NiO on the surfaces of the Ni powders, which resulted from the inverse relation of growth

rate to oxide thickness during the early stages of oxide growth. Regions covered by a thicker

oxide shell oxidized more slowly than less heavily oxidized regions, thus producing a self-

regulating mechanism to control the uniformity of the oxide shell. The oxidation rates were

paralinear in the temperature/time regimes studied. The exothermic nature of the oxidation







93

reaction required sufficient cooling to maintain isothermal conditions during the oxidation

treatment, otherwise anomolously high oxidation rates were encountered.

It was found that during the high-temperature oxidation of aluminum, the humidity

level in the oxidizing atmosphere affected the adherence of the oxide scale to the metal

substrate. At low levels (pH20= 0.011), significant spelling was noted. At higher levels of

humidity (pH20 0.067), spelling was reduced significantly. This indicates the formation of

different hydroxide structures which depend on the humidity level in the oxidizing

atmosphere. The structural effect causing variations in the scaling behavior of the oxides and

the existence of different peaks in the IR spectra clearly indicate the formation of different

oxidation products by changing the water content of the atmosphere. Alwitt [Alw76] reports

results of several studies that show different structures formed for the high-temperature (above

the critical point of water) oxidation of aluminum in steam. These were explained by

variations in the kinetics of formation for the various oxides in relation to heating rate and

temperature, the effect of moisture was assumed to be negligible.

The hydration of alumina at low temperatures has been studied extensively [Alw76,

Bec67, Fir91, Wef87]. The moisture, when added to "dry" powder was found to require an

aging of 48 hours to significantly increase the "oxidation rate" as determined by the volume

fraction of alumina produced in the composites. Aging affected the alumina content when the

water was in contact with the aluminum powder, but not the nickel powder. Thus, it was

concluded that the important surface conditioning by water takes place on the aluminum

powder surface.

The effect of moisture was found to be three-fold. First, the moisture content of the

oxidizing atmosphere regulated the adherence of the oxide film to the powder particles,

secondly the room temperature hydration of the oxidized powders influenced the volume