Processing, compositional range, and mechanical behavior of the Mo₅Si₃C intermetallic compound


Material Information

Processing, compositional range, and mechanical behavior of the Mo₅Si₃C intermetallic compound
Physical Description:
vii, 149 leaves : ill. ; 29 cm.
Ross, Eli N., 1972-
Publication Date:


Subjects / Keywords:
Materials Science and Engineering thesis, Ph.D   ( lcsh )
Dissertations, Academic -- Materials Science and Engineering -- UF   ( lcsh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1999.
Includes bibliographical references (leaves 136-148).
Statement of Responsibility:
by Eli N. Ross.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 030478709
oclc - 42584433
System ID:

This item is only available as the following downloads:

Full Text







For Mom, Dad, Barbara, Carl, and Nera.


Like most endeavors requiring significant effort, the research and writing of a

dissertation involves the support and assistance of a large number of people. To begin

with, I am very grateful to Andre Costa e Silva for introducing me to the basics of research,

while always maintaining his patience and good humor. To Paul Eason and Luisa Amelia

Dempere, I am indebted for their many useful discussions and team-approach to problem

solving. For his ability to take on equipment and computer problems, willingness to help

and positive attitude, Ryan Kaufman deserves recognition. I am also appreciative of Dr.

Fereshteh Ebrahimi, Dr. Jian Hu, and Dr. Ozer Unal, who provided invaluable assistance

during the mechanical testing portion of this work. The assistance of Wayne Acree at the

Major Analytical Instrumentation Center was important to the materials characterization

efforts in this study.

I consider myself very fortunate to have worked under the direction of Dr. Michael

Kaufman during my time at Florida-his inquisitive nature, approachability, and broad

knowledge base have served me very well during this research effort. I am also glad to

have forged a friendship with Dr. Kaufman, as have many of his present and former

students. I am also grateful to the other members of my graduate committee for adding

their perspectives and guidance toward producing a focused research project. I would also

like to thank Jim Cotton for his encouragement and pragmatism during our discussions.

Finally and most importantly, the support and love given to me by my friends and family

members has been an integral part of my graduate studies, without which none of this

would have been possible.



ACKNOW LEDGEM ENTS ........................................ .......................................iii

A B ST R A C T ............................................................................ ............................. vi


1 IN TRO D U CTION ................................................................ ........................ 1

2 LITERATURE REVIEW ................................... ..... .......................4

2.1 Drivers and Materials Requirements for High-Temperature Applications ............4
2.2 Binary Silicide Compounds for High-Temperature Applications ......................6
2.2.1 M olybdenum Disilicide ..................................... ...................... 7 Processing of MoSi2 and its composites ........................................... 9 Low-temperature mechanical behavior .......................................... 14 Elevated-temperature mechanical behavior .................................. 15 Carbon additions to MoSi2 ................................... ............... 17
2.2.2 MsSi, Compounds ................................................ ....................... 18
2.3 Ternary Silicide Compounds ................................... ............................. 21
2.3.1 Mo-Si-Al System ................................................. 23
2.3.2 M o-Si-B System .................................. ...... ................................. 24
2.3.3 Mo-Si-C System and the MosSi3C Phase .............................................24 Thermodynamics of the Mo-Si-C system ....................................... 24 Atomic arrangement and crystal structure of MoSi3C ................... 27 Processing, microstructures and properties of MoSi3C ....................27

3 PROCESSING AND COMPOSITIONAL RANGE ISSUES ................................33

3.1 Introduction .................................................................. ........................ 33
3.2 M materials and Experimental Procedures ................................................... 33
3.2.1 Raw M materials ............................. ...... .............................. 33
3.2.2 Processing Routes .................... ..................................... 34 Arc m elting ........................................................... 35 Powder mixing and hot pressing ............................... .......... 35 Mechanical alloying + hot pressing ................................................ 37
3.2.3 Materials Characterization and Analysis .............................................. 41 Physical properties .................................. ......................41 M icrostructural analysis ............................................................... 42 Phase identification and compositional analysis .............................44
3.3 Results and Discussion ...........................................................48
3.3.1 Processing ...................... ................................... 48 Compositions studied .................... ..............................48 Arc m elting ................................... ................................... 49 Powder processing methods .......................... ....................... 52
3.3.2 Compositional Range of MosSiC ...................... ........................ 68 Width of MosSi3C single-phase field ............................................. 70 Height of MosSi3C single-phase field .............................................. 72 Heat treatment ........................ ............................. 72 Conclusions ..................... ...................................76
3.4 Summary and Conclusions .........................................................77


4.1 Introduction .............................................. ............................................ 82
4.2 Experimental Procedure ............................................................. 83
4.2.1 Sample Characterization and Analysis ....................................................83
4.2.2 H ardness Testing ............................................. ........................ ... 84
4.2.3 Indentation Fracture Toughness Measurements .................................... 84
4.2.4 Elevated Temperature Compression Testing ........................................... 85
4.2.5 Four-Point Bend Testing ................................... ........................ 88
4.3 Results and Discussion .................................................................. ...........90
4.3.1 Microstructural Characterization of Test Specimens .................................90
4.3.2 Hardness and Fracture Toughness Measurements ................................... 93
4.3.3 High-Temperature Deformation Behavior in Compression ..................... 97
4.3.4 Four-Point Flexure Testing of MoSiC ............................................... 111
4.3.5 Mechanical Property Comparisons with Other Silicide Materials ........... 116
4.3.6 Effectiveness of Alloying to Improve Ductility .................................. 118
4.4 Summary and Conclusions ................................................. 122

5 GENERAL DISCUSSION AND CONCLUSIONS ......................................... 124



B SELECTED COMPRESSIVE STRESS-STRAIN CURVES .............................. 132

REFEREN CES ......................................................................... ..................... 136

BIOGRAPHICAL SKETCH ................................................................ 149

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



Eli N. Ross

August 1999

Chairman: Michael J. Kaufman
Major Department: Materials Science and Engineering

The intermetallic compound MosSi3C was studied in order to examine optimal

processing methods, elevated temperature mechanical behavior, and the influence of crystal

symmetry on the potential for improved room temperature ductility/toughness in ternary

silicides. During the course of the investigation, samples were produced using non-

consumable arc melting and vacuum hot pressing of either blended or mechanically alloyed

powders. The most significant challenge to the production of single-phase Mo5Si3C was

the significantly narrower range of homogeneity for the ternary compound than originally

reported. The primary reason for this discrepancy is thought to be the use of x-ray

diffraction (XRD) as the sole means of phase identification in the original study, whereas

the current work augmented XRD with other microstructural analysis techniques.

Further complicating the processing of MosSi3C was the occurrence of composition

shifts during powder processing. These shifts in stoichiometry were correlated to observed

microstructural features and likely the result of thermodynamically favored reactions

between silica present in the starting powders and carbon. Because of these complications,

materials for mechanical testing typically contained between one and 6 volume percent of

phases other than MoSi3C, with samples having the nominal composition Mo4.8Si3Co.7

closest to single phase.

An average microhardness value of 13.2 GPa was recorded for the ternary phase at

room temperature, and an indentation fracture toughness of approximately 2.5 MPa-m'~

was determined. In the temperature range 1000 to 13000C, materials tested in compression

displayed properties that were highly strain rate and grain size dependent, indicating the

influence of boundary-controlled deformation mechanisms. The behavior of samples tested

in four-point bending at 12000C showed similar behavior, with larger grain size materials

failing after limited deformation while finer grain size materials deformed readily.

Examination of deformed specimens using transmission electron microscopy

indicated that slip in MoSi3C occurs through motion of c[0001]-type dislocations at

temperatures above 10000C. This is consistent with the observed lack of improvement in

the toughness or ductility of MoSSi3C, resulting from any positive contributions due to

crystal symmetry being nullified by the intrinsically high resistance of the complex

MosSi3C crystal lattice to the generation and motion of dislocations.


In recent years, there has been renewed interest in silicide compounds,MoSi2 in

particular, as possible high-temperature structural materials. In the case of MoSi,, this is

largely due to its high melting point (20200C) [1], moderate density (6.24 g/cm3) [2],

measurable deformation at elevated temperatures [3] and excellent oxidation and corrosion

resistance to -17000C due to the presence of an adherent, passivating silica scale [4].

These properties are direct analogues to several of the primary materials requirements for

future generations of gas-turbine engines, namely increased operating temperatures,

reduction of rotating weight and environmental resistance. However, like the majority of

intermetallic materials investigated as potential replacements for current superalloys, MoSi,-

based materials suffer from inadequate low-temperature fracture toughness and poor

formability. In addition, silicides typically show a drop in strength at elevated temperatures

and poor resistance to creep deformation

In all likelihood, the mechanical property shortcomings of MoSi2 have origins both

intrinsic and extrinsic. For example, the high bond strengths and low-symmetry crystal

structure (C 1lb, tetragonal) of MoSi, significantly limit the potential for ductility at low

temperatures [5], while the presence of a glassy SiO2 phase in powder processed MoSi,

may act as fracture initiation sites at ambient temperatures and adversely affect strength at

elevated temperatures because of grain refinement and viscous flow at high temperatures

[6]. Because of this, researchers have attempted to address these issues through

microstructural control, compositing, and alloying additions [7].

Due to the high melting point of MoSi2, traditional casting and forming processes

have for the most part been abandoned in favor of powder-based methods. Techniques

such as handling of siliceous powders in inert atmospheres to suppress silica formation,

etching of powders to remove SiO2, and reduction of silica through alloying additions have

all been attempted as ways to control the presence of silica in MoSi2 microstructures [8, 9].

The effectiveness of these approaches has varied, especially given the difficulty in isolating

the detrimental effects of silica from other possible mechanisms at work. Some success in

improving the high temperature properties of MoSi2 has been achieved through use of

reinforcing phases; SiC has been widely utilized in both conventional and in situ

compositing techniques in long fiber, whisker and particulate forms [3]. Ternary alloying

additions (e.g., Al, B, Cr, Ge, Hf, Nb, Re) to MoSi, have been studied as a possible way

to improve mechanical properties through solid solution strengthening and second phase

precipitation effects, although property improvements have been minimal [10].

In a few intermetallic systems, it has been shown that macro-alloying to produce

ternary and higher-order compounds having higher symmetry (e.g., cubic, hexagonal)

crystal structures results in a significant increase in room-temperature ductility compared

with the parent binary compound [11]. Phase stability studies in Mo-Si-X systems have

shown the presence of a number of ternary silicide phases, many of which exhibit higher

symmetry crystal structures than tetragonal MoSi2. Included among these phases is the

hexagonal D88 compound Mo5Si3C. As the only ternary phase identified in the Mo-Si-C

system, MosSi3C has attracted recent interest since it is the third component in in situ
processed MoSi2-SiC composites [12, 13]. Very little is known about the physical and

mechanical properties of ternary hexagonal silicide compounds such as MosSi3C, including

whether enough mobile, low energy dislocations are present at low temperatures to make

the potential activation of additional slip systems meaningful. Other than selected ternary

isotherms, little information exists regarding the processing of these compounds. In

processing approaches which take advantage of solid-solution strengthening, eutectic

transformations, or displacement reactions, the importance of having accurate

thermodynamic data is vital, especially since many desirable phase combinations exist

adjacent to phase regions of undesirable, low melting (e.g., Si) phases [14].

From the standpoint of investigating the relationship between higher-symmetry

crystal structures and mechanical properties in ternary silicides [15-17] and clarifying the

mechanical behavior of the third component of in situ processed MoSi2-SiC composites

[12], the MosSi3C phase seems a reasonable choice for further study. In addition, it may

be possible during the course of the study to clarify uncertainties as to the actual

compositional range of the ternary silicide compound [18, 19].

The initial stages of this project were concerned with optimizing methods for the

processing of single-phase MosSi3C. While it was possible to screen out methods that

were obviously unsuited to the production of samples for mechanical testing, process

optimization was greatly complicated by confusion surrounding the size of the MosSi3C

single phase field. A systematic reexamination of compositions near the MosSi3C phase

field was used to determine the stoichiometry which had the highest potential for yielding

single-phase materials. After refinement of the best combinations of processing and

composition for producing near-single phase MoSi3C over a range of grain sizes, an

assessment of baseline mechanical properties for MoSi3C-based materials was conducted

and these properties compared with those of other candidate high-temperature materials.

In light of the sizeable number of interrelated process, composition, and property variables

inherent in this materials system, an emphasis was placed on such issues as sample quality

control, process monitoring, extensive documentation, and efficient design of experiments.


2.1 Drivers and Materials Requirements for High-Temperature Applications

The search for materials systems capable of withstanding higher temperatures and

extreme environments while maintaining strength is one of the foremost technical

challenges to continued improvements in performance, efficiency, and reliability of gas

turbine engines for aerospace and power generation applications. Representative materials

requirements and properties of interest for these high-temperature applications are given in

Table 2.1.

Table 2.1 Materials and process requirements for high-temperature materials in gas
turbine engines [20-22].

Property/Issue Relevance
High melting point Higher operating temperatures
Strength retention at high Rotating, load bearing critical
temperatures structures
High elastic modulus Resist centrifugal stresses, thermal
Creep resistance Increased operating life, reliability
Environmental resistance Combustion products, water
vapor, cyclic exposure
Low, high cycle fatigue Increased operating life, less
resistance downtime
High thermal Reduced cooling requirements;
conductivity, low efficient design, coating
thermal expansion compatibility
Low density Rotating parts, power/weight
Toughness, impact Design flexibility, foreign object
resistance damage, low temp. = most of
service life
Ease of forming, Use existing processes, increased
machining, joining yield
Cost Benefits must justify development

Nickel-based superalloys are currently the material of choice for the most

demanding of gas turbine engine applications; alloy optimization and development of

single-crystal technologies along with active cooling and thermal barrier coatings have led

to significant increases in engine operating temperatures [23]. An increase in turbine

engine operating temperature has as a direct result an increase in specific power output

while at the same time leading to a decrease in fuel consumption [21]. In spite of the

documented success of superalloy systems, current superalloys operate within -1000C of

their incipient melting temperatures [23], thus in order for increases in engine operating

temperatures to continue, new materials systems need to be considered.

Potential materials systems to replace superalloys include refractory metals,

advanced ceramics, and intermetallics. Refractory metals maintain strength to very high

temperatures (-1600C) and can be processed to have acceptable ductility and toughness,

but oxidize readily above 5000C and specific properties suffer because of high material

densities [24]. Ceramics are attractive because of high strength at temperature, excellent

creep resistance, low density, and usually good oxidation resistance, but a lack of ductility

and poor electrical conductivity makes forming very difficult and their generally low

fracture toughness values limit design options [25].

Intermetallics, mainly aluminides and silicides, exhibit good oxidation resistance,

low specific gravity, high strength and modulus, good thermal conductivity, and can, in

some cases, be formed using traditional techniques such as casting and forging [11]. In

addition, they often have sufficient electrical conductivity to be machined using electrical-

discharge methods [26]. However, achieving a balance of properties in a particular

material or even an alloy system has proven to be very difficult. Intermetallics are also

generally plagued by low impact resistance, high notch and strain rate sensitivity, and poor

ductility at room temperature [27]. Because of their potentially attractive properties for

high-temperature structural applications, considerable effort has gone into the study and

development of materials such as nickel, iron, and titanium aluminides, Laves phases, and

silicide compounds [11, 28-30].

2.2 Binary Silicide Compounds for High-Temperature Applications

Over 130 different binary silicide compounds have been identified and characterized

[3] with roughly 75 percent being transition metal silicides. Most of these compounds

show mixed covalent-metallic bond character, maintain ordered structures until melting,

and exist as line compounds having little compositional variation. Current applications for

silicides include heating elements, coatings (Figure 2.1), thermal protection systems, diesel

engine glow plugs, infrared detectors, and electrical contacts for improved microelectronic

device performance [7, 26, 31].

Figure 2.1 Silicide coating on Nb-alloy (C-103) afterburner flaps of Pratt &Whitney F-
100 engine [32].

2.2.1 Molybdenum Disilicide

Of the refractory metal silicides, MoSi, has generated by far the most interest for

high-temperature applications, primarily because of its excellent oxidation resistance which

extends to 17000C [33]. In addition to oxidation resistance, MoSi2 also exhibits a high

melting point (2020*C) [1] and a density of 6.24 g/cm3 [2], roughly 25 % less than a

typical Ni-based superalloy. In the absence of impurities, MoSi2 maintains the tetragonal

Cl1b structure (Figure 2.2) up to its melting point. Molybdenum disilicide has seen
application in various forms for almost 90 years [3], with the most well-known being the

use of MoSi2 as the primary component of oxidation-resistant Kanthal Super heating

elements (Figure 2.3).

Research by W.A. Maxwell and coworkers in the early 1950s was probably the

first to consider MoSi,-based compounds for use as high-temperature structural materials

[34, 35]. Maxwell's work noted that monolithic MoSi2 exhibits several undesirable

features, namely a decrease in strength and poor creep resistance at high temperatures and

lack of ductility at room temperature (RT). In fact Maxwell and others [36, 37]

investigated doping MoSi, with carbon and the use of A1203 as a potential reinforcement for

MoSi2 as early as 1953. However, at that time, with gas turbine engine development still
in its early stages, the primary technological demands of aerospace could readily be

satisfied with traditional metallic materials. This, combined with a limited design base for

use of brittle materials in structural applications, is a probable explanation for the limited

amount of work on MoSi, from the time of Maxwell and his contemporaries until about the

mid-1980s [3].
By the end of the 1980s, promising work in identifying toughening mechanisms in

structural ceramics [38] combined with a recognized need for increased operating

temperatures and lower weight in aerospace applications [21] brought about a renewed

interest in MoSi, as a potential replacement for current superalloy materials. From Table

2.2, MoSi, compares fairly well with other intermetallics and ceramics being developed to

ct 4 E


Mo oz"

Figure 2.2 Tetragonal Cl1 1 (tl6) crystal structure of MoSi2 having six atoms per unit cell

Binder (clay, glass)

I i 'Z

Figure 2.3 Microstructure of Kanthal Super heating element (1000x) showing two phase
matrix covered by protective silica layer [40]

replace current superalloys. However, the same problems Maxwell first observed of low

toughness at ambient temperatures and poor creep resistance at elevated temperatures

remain as challenges to the realization of MoSi2-based materials for structural applications.

Table 2.2 Comparison of physical and mechanical properties of MoSi2 with other current
and potential high temperature materials. Where scatter existed in the literature, values
were averaged. Thermal conductivity (k) is given in W/m K at 1000 K, coefficient of
thermal expansion (CTE) data is in ppm/K, elastic modulus (E) is in GPa, toughness
values (K,,) are in MPa-mla, and yield strength (YS) is given in MPa. Values for Nb alloy
and superalloy are typical for these metals, [3, 7, 11, 41-47].

Material Tm Density Oxidation k CTE E K, Y.S.
(C) (g/cc) Limit (C) (RT) (RT) (1000C)
MoSi, 2020 6.24 1700 35 8.1 440 3 300
NiAl 1647 5.86 1227 70 15 294 5 35
TiA 1440 3.9 900 22 12 162 25 150
TiAl 1602 4.2 652 7 10 145 10 40
SiC 25005 3.18 1600 60 5 480 3.5 >2000
AlO, 2050 4 >1600 7 10 380 5 1000
Nb-alloy 2400 8.6 600 45 8 90 30 300
superalloy 1350 8.4 920 12 14 206 50 300 Processing of MoSi_ and its composites

Processing techniques for polycrystalline MoSi2-based materials can be grouped

into either powder-based or melt-based routes. Of the powder processing routes, hot

pressing (HP) is the most commonly used for the production of both monolithic MoSi, and

composites [48]. Starting powders for HP can be either elemental or pre-alloyed in nature,

with the latter usually requiring higher consolidation temperatures ; typically pressing

temperatures range from 1500"C to 18000C in either vacuum or inert gas atmospheres [49].

When high consolidation temperatures (-18000C) are used, weight losses on the order of

20% have been reported for carbon-doped MoSi2 materials [50], the result of a lower-

melting ternary Mo-Si-C eutectic being expelled from the die. Another contributing factor

to weight losses at high temperatures, especially during vacuum HP, is the high vapor

pressure of silicon over MoSi, (Figure 2.4); for this reason, high-temperature consolidation

is often done in an inert gas environment [13].

A variation on HP is hot isostatic pressing (HIP), which typically results in higher

density, higher purity MoSi2 materials [51], although it is a more complex process than

HP. To vary the scale of MoSi, microstructures produced by HP or HIP, options include,

(1) changing the particle size of the major element or compound in the starting powders

[52], (2) affecting grain growth and sintering kinetics by varying pressing temperature and

time [49], and (3) refinement of the MoSi, matrix by introduction of second phase particles

which retard grain growth by a Zener drag mechanism [53]. Reactive sintering techniques

such as reactive HP/HIP or self-propagating high-temperature synthesis (SHS) take

advantage of the thermal energy generated by exothermic reactions between starting

powders, which in some cases dramatically reduces processing times [54]. Residual

porosity and limited microstructural control are limitations of typical reactive sintering

processes [48]. Another method successfully used to produce MoSi2 composites is low-

pressure plasma-spray deposition, which typically results in very fine microstructures [55].

Mechanical alloying (MA), high-energy impact milling of powders, has been used to

synthesize sub-micron silicide powders, often with simultaneous solid-state phase

formation (Figure 2.5). These features allow for the consolidation of MoSi2 and its

composites at lower temperatures with the resultant materials having highly refined

microstructures [56-58]. One concern common to all variants of MA is contamination of

the powder charge by the milling vial and balls [48]; the severity of impurity pickup during

MA is primarily a function of the relative hardness of the powder and milling media and the

solubility of the impurity elements in the powder constituents.

For all powder processing routes, oxygen contamination of the powders is a key

concern. Elemental Si or silicide powders exposed to the atmosphere rapidly oxidize to

form SiO, on their surfaces; in the absence of process modifications, 5 10 Vol.% of the




o 20

1650 1700 1750 1800 1850 1900
Temperature (*C)

Figure 2.4 Plot of equilibrium dissociation pressure of Si over MoSi2, adapted from
Searcy and Tharp [59].

Figure 2.5 Schematic representation of the mechanical alloying process for MoSi,
wherein repeated cold-welding and fracture processes lead to the formation of the silicide
from its constituent powders [48].

oxide will be retained in the consolidated materials (Figure 2.6), typically as pockets of

amorphous silica at the grain boundaries and triple points [34, 35, 60, 61]. The amount of

silica present in a silicide material is a strong function of starting particle size of the

siliceous powders; fine particles having large surface area typically have the highest

percentage of oxide, meaning that control of atmosphere is crucial during processes that

result in very fine powders, such as MA or plasma-spray [48, 49]. Minimizing the amount

of silica in MoSi2-based materials is important primarily from the standpoint of strength at

high temperatures, where the glassy silica is thought to soften and accelerate strength

degradation through grain boundary sliding [62].

Figure 2.6 Light optical micrograph of MoSi, under polarized light, revealing
intergranular silica particles (white spots). Silica is incorporated into powder processed
MoSi2 from oxides resident on the starting powders [9].

In spite of some disagreement over the exact role of silica in elevated temperature

deformation, it is generally agreed that it is an undesirable feature in consolidated MoSi2.

Accordingly, a large amount of effort has focused on ways to remove the oxide phase from

MoSi2, including powder etching [35], inert-gas handling [9] and intentional additions of

elements such as Al, C, Zr, or rare earth elements which reduce silica [34, 58, 63-66].

Table 2.3 shows typical oxygen content values for MoSi2 powders and consolidated parts,

indicating a wide variation depending on powder size and processing parameters. The

effects of silica on mechanical properties will be discussed in greater detail below.

Table 2.3 Oxygen content in MoSi2 powders and consolidated material [63].

Material Form Oxygen Source/Investigator
MoSi2 powder, 45-75 pm 0.3 Cerac _
MoSi2 powder, <45 gm 2.02 0.4 Alfa AESAR7
MoSi2, MA powder, <1 im 4.07 0.016 Jayashankar
MoSi, + SiC, in situ hot-pressed 0.167 Jayashankar
MoSi, + SiC, blended hot-pressed 1.71 0.05 Jayashankar
MoSi, hot isostatically- 0.16-0.22 Rockwell
glove box processing pressed
MoSi2, MA+ hot-pressed 0.1 Los Alamos
Glove box processing

Various fiber, whisker, and particulate reinforcements have been combined with

MoSi, using variants on the aforementioned powder processing techniques. Ductile
reinforcements include Nb [67-69], Ta [67] and W [68]; complications associated with

refractory metal reinforcements include the need for diffusion barrier coatings on the metal

because of their thermodynamic incompatibility with MoSi, [48, 64, 68] and the reduced

oxidation resistance of the composites [48]. A number of oxide and non-oxide ceramics, in

fiber, whisker, and particulate form, have also been used as reinforcing phases for MoSi,

and other silicides [3, 48,49, 70]. One common way of producing MoSi2-ceramic

composites are so-called in situ processing routes which take advantage of solid-state

displacement reactions [71] to produce fine, homogeneously distributed, and

thermodynamically stable reinforcements including SiC [13, 72] and TiB2 [73]. The

processing and properties of in situ MoSi-SiC composites will be discussed in more detail

Of the melt-based methods, the simplest is non-consumable arc melting, which has

been used extensively in alloying studies involving MoSi2 [10, 74, 75]. Another process

of note is the patented exothermic dispersion or XDTM process, wherein ceramic

particulates are precipitated within in a MoSi2-based master alloy; composites produced via

XDTM have the advantage of being inherently stable thermodynamically and can contain

high volume fractions of uniformly distributed ceramic, in excess of 40 Vol.% [76, 77].

Two other melt-based silicide composite processing methods are in situ melt processing

[78] and eutectic solidification [79, 80]. Low-temperature mechanical behavior

Studies of deformation mechanisms in MoSi, single crystals have indicated the

following possible slip systems and temperature ranges: (001)<100> (200C) [81];

{ 110<111>, {0131'/2<331> (900 15000C) [82]; 111) <110> (12000C) [83]; (011)

<100> (RT), ( 110 '/2<111> (T > 6000C) [84]. Even in the temperature regimes in which

multiple slip systems are operative in MoSi2, it has been shown that at most four of these

represent independent slip mechanisms [84]. It was shown by von Mises [85] that for a

single crystal to undergo an arbitrary shape change, a minimum of five independent slip
mechanisms must be operative; by extension, the deformation of a collection of randomly

oriented single-crystals (i.e., polycrystalline material) also requires five independent slip

systems [86]. Note that the von Mises criterion is a necessary, but not sufficient condition,

since deformation also requires the availability of mobile dislocations [27].
At ambient temperatures, polycrystalline MoSi, fractures in a brittle manner with

zero ductility and exhibits low fracture toughness (-3 MPa.m'2) [7, 87, 88]. This
behavior is common to many intermetallics, which typically have: (1) strong atomic

bonding; (2) ordered atomic arrangements and large unit cells; and (3) low-symmetry

crystal structures [27]. These factors combine to greatly increase the difficulty of plastic

deformation; the first two result in increased energy and complexity of dislocations (broad

cores), reducing their generation and subsequent mobility. Along with bond strength, high

bond directionality is also associated with brittleness; using a valence-force-field approach,

it was recently determined that the bond directionality in the C11b (MoSi2) structure was

stronger than in either the C40 (CrSi,) or C54 (TiSi2) structures [89]. Low crystal

symmetry generally results in the number of operative slip systems being insufficient to

satisfy the von Mises criterion [20, 27, 86].

For MoSi2, noticeable permanent deformation is possible only at temperatures

(>1000C) where thermal activation can increase dislocation density and mobility [27, 41]

or produce apparent plastic deformation through time-dependent deformation mechanisms

[90]. For powder processed silicides, it has been proposed that silica present in MoSi,

(Figure 2.6) may have further adverse effects on RT fracture toughness and ductility,

although by what mechanism is not clear [9].
Phase stability studies of MoSi2 with additions such as Ta, Ti, Nb and Cr have

indicated the possibility producing multi-phase silicides, although these materials did not

show any improvement in fracture toughness over monolithic MoSi, [14, 91, 92].

Removal of silica through carbon additions provides some increase in toughness at high

temperatures, but has shown little effect on ambient temperature properties [66]. Ductile

reinforcements such as Nb [67-69], Ta [67] and W [68] in filament, particulate and wire
result in modest toughness improvements. Although reinforcing of MoSi, with SiC, ZrO2,

or p Si3N4 has resulted in RT fracture toughness values as high as 8.5 MPa-m"2 [3, 26,

48, 93] this is still well below the range of some aluminides (e.g., Ti, (Al, Nb) 30

MPaom" [11]) and superalloys. Very limited information exists on the fatigue properties of

MoSi2-based materials, although as might be expected from its low toughness, fatigue
crack growth rates increase rapidly with increasing stress intensity [6]. Elevated-temperature mechanical behavior
Monolithic MoSi, is generally regarded to exhibit three regimes of mechanical

behavior: (1) strong and brittle below ~ 10000C; (2) strong and ductile between -1000 and

-12500C; and (3) weak and ductile above 12500C [41]. The onset of regime (2) is

associated with the brittle-to-ductile transition temperature (BDTT), although there exists

considerable dispute regarding the definition of the BDTT for MoSi2, and its reported value

ranges from 950 to 1400C [94]. Much of this confusion relates to the presence of silica in

MoSi2 as a result of powder processing and how its softening might influence definition of

the BDTT [94] as well as lead to a decrease in the elevated-temperature strength and creep

resistance of MoSi, [48].

For most single-crystal and coarser grain silicides (>20 ipm) tested at temperatures

above 10000C, stress exponents (n) of around 3 are observed, consistent with dislocation

glide-controlled deformation [62]. Materials with finer grain sizes (-5 mn) typically have n

less than 2, which is often associated with a combination of viscous/diffusional flow and

dislocation mechanisms; it is believed that the softening of glassy silica acts a lubricant to

further enhance grain boundary sliding in the finer grain materials, whose fine scale already

predisposes them to boundary-related mechanisms [6]. For example, the creep rate of a 20

tim grain size monolithic MoSi, is more than three orders of magnitudes lower than a 5 pm
grain size material at 12000C [6]. When the deformation behavior of fine grain silica-free

and silica-containing materials of equivalent grain size was compared [63], the silica-free

material did show higher n values than the silica-containing MoSi2, indicating the

deleterious effects of the oxide on creep strength. However, both materials displayed

similar trends of decreasing n with increasing temperature, while a 40 pm silica-free sample

showed the opposite trend, indicating that grain size may actually be a more important

factor than silica content in controlling the high temperature deformation of silicides.

Temperature and strain rate effects on the compressive strength of a coarse (-40 pm) grain,

low-oxygen MoSi2 sample are presented in Figure 2.7 (a).

To further improve the elevated-temperature properties of MoSi, a variety of

reinforcing phases have been examined; viable reinforcements for high-temperature

applications need to have desirable mechanical properties, as well as thermodynamic

compatibility and minimal thermal expansion (CTE) mismatch with the matrix. Of potential

reinforcement materials, SiC has proven to be one of the most popular because of its

thermodynamic compatibility with MoSi2, its negligible effect on matrix oxidation and its

availability in a wide range of forms [41, 48]. However, SiC does display a rather large

CTE mismatch with MoSi2 (see Table 2.2) resulting in thermal strains and matrix cracking

[41] upon cooling. Even so, significant improvements in yield strength, particularly above

1300C, have been reported with both particulate and whisker SiC reinforcements (Figure

2.7 (b)) [3, 77, 91].
Improvements in creep resistance over monolithic MoSi, have also been observed

(Figure 4.8 (a)) with C additions to form SiC and remove silica, with dual mechanisms of

composite strengthening and elimination of the viscous silica [48, 62]. However, as

described above, the effects of silica on creep properties are still somewhat unclear, with

some low-silica MoSi,-based materials showing stress exponents close to unity, although

usually at finer grain sizes [95]. Some of the highest creep resistance reported for MoSi2-

based materials is that of an in situ MoSiz-SiC composite (Figure 4.8 (b)) containing only 5

Vol.% reinforcement, resultant from carbon addition just sufficient to reduce all silica; this

small fraction of SiC still allows for sufficient grain growth during consolidation to obtain a

40 gim grain size, and it is this combination of low silica and large grain size which leads to

the excellent creep strength of this material. Carbon additions to MoSiz
Since carbon additions to MoSi2 are the most common route of silica removal and

are also one common method for producing in situ composites of MoSi,-SiC it is

worthwhile to examine this method more closely. The concept of adding C to MoSi, was
recognized early on [35, 96] as an effective way to reduce the siliceous grain boundary

phase through reactions of the form [65]:
SiO2 + C= Si(g) + CO(g) (2.1)

SiO2 + 3C= SiC+2CO(g) (2.2)

SiO2 +2C=SiC+CO,(g) (2.3)

As an extension of carbon additions to MoSi,, it seems logical that the concept of solid state

displacement reactions [71, 97] could be used to produce a SiC reinforcement phase using

in situ methods. Compared with traditional composite processing schemes, in situ

reactions can often be initiated at lower temperatures and the resultant materials have

interfaces which are inherently stable thermodynamically [64, 97]. Using an exothermal-

dispersion (XDTM) technique, Aiken [76, 77] produced MoSi,+30 Vol.% SiC composites

with equiaxed reinforcement morphology, reduced oxygen content compared with

unreinforced MoSi,, and increased strength at high temperatures. Jayashankar [58, 63]

used a combination of carbon additions, MA and a carbothermal reduction step to eliminate

silica and produce SiC reinforcements. Microstructures in this work contained three

phases-low-oxygen MoSi2, uniformly distributed SiC and the ternary phase MosSi3C,

which contained iron presumably picked up from the steel milling media.
Another approach to producing in situ MoSi2+SiC composites was that of Henager

and collaborators [12, 72] who hot pressed compacts of Mo2C + Si at 13500C/1700C and

modeled their microstructural evolution from diffusion and interfacial reaction

considerations. This in situ approach led to a fine-grain (1-2 gtm) MoSi, matrix with -30

Vol.% of sub-micron SiC and MosSi3C. As expected, an improvement in strength is seen
for these MoSi2+SiC composites although the authors note that the lack of information on

the mechanical properties of the MoSi3C phase complicates analysis of the composite
mechanical behavior.

2.2.2 M-Si. Compounds

Because of their relationship to the MosSi3C compound which is the focus of this

study, available information about the compounds MosSi, and TiSSi, will now be
reviewed. Having the nearest stoichiometry to Mo5Si3C of the three binary silicides of

molybdenum, MosSi3 has a tetragonal D8,(tI32) crystal structure [98], a density of 8.24

g/cm', and melts congruently at 21800C. From the Mo-Si binary diagram [1], MosSi3 has a

homogeneity range of 37 to 40 At.% Si at 12000C.

o YSO 10
0 rl|it)
YSo I.:.

11 Io 100 1 300 1350 1400 1450

Figure 2.7 (a) Compressive flow behavior of high-purity, 40 pm grain size MoSi2 [9].
(b) Bend strength versus temperature for two MoSi2-based composites, along with other
intermetallics and the single-crystal superalloy MAR M247 for comparison [3].

to' ., .. fb/ / ''

-, ....... *S/i!
..1n >i. i oK: io *y / f .
IO io3if i i." Mr / /

0' X*0

t 10' ,- I'


1 S, 15RT

I 10"

40 O0 I0100 100
StrHe. MPS

Figure 2.8 (a) Creep data for monolithic MoSi2 and several MoSi2 composites [95]. (b)
Comparison plot showing creep data for several MoSi2-SiC composites and values for
ceramic-matrix composites [63].

Because of its reduced silicon activity, MoSi, does not from a protective, adherent

silica scale like MoSi2 and as such has poor oxidation resistance; doping with less than 2

Wt.% boron seems to improve the oxidation resistance of Mo0Si3, with the proposed

mechanism being boron-aided viscous sintering to close porosity in the scale formed during

intermediate temperature oxidation [99]. Little is known of the mechanical properties of

MoSi3, other than elastic constants of single-crystal material [100]; Mo5Si3 has been
reported to show superior creep resistance [101] compared with MoSi2.

The TisSi, compound melts at 21300C, has a density of 4.32 g/cm3, and solidifies

in the hexagonal D8, (hP16) crystal structure. One reason given for the study of TiSi3

was to examine its oxidation resistance, expected to be controlled by a mixed silica-rutile

scale; in fact, the reported oxidation behavior of Ti5Si, was similar to titanium in that a

transition from parabolic to linear oxidation kinetics occurs as the external TiO2 scale spalls

off. Doping with less than 3 Wt. % carbon considerably slowed the oxidation process,

perhaps as a result of the formation of an underlying continuous silica scale [102]. In

addition, because of the presence of a Ti-TiSSi, eutectic at 13500C it is possible to produce

directionally solidified eutectic composites capable of 3.5 % plastic strain in compression at

room temperature without cracking [80].

A feature reported for both Mo5Si3 and TiSi3 is the presence of transgranular

microcracks upon cooling from the processing temperature; for MosSi,, samples in one

study having average grain sizes greater than 15 gtm exhibited microcracking [102], while a

grain size-dependent microcrack density was also reported for hot pressed TiSSi, [103,

104]. The root cause of this thermally induced microcracking was in both cases proposed

to be an anisotropy in coefficients of thermal expansion leading to microfracture upon

cooling in accord with the models of Evans [105]. The critical facet size 1, for the onset of

microfracture caused by CTE anisotropy, which is related to a critical grain diameter, is

5.2(1 + v)2 g (2.4)
-o = 2 (2.4)

showing a strong dependence on the degree of CTE anisotropy (Aa) and temperature

gradient (AT); alternatively, it can be shown that the stress intensity factor increases with

increasing grain size for materials with anisotropic expansion coefficients [105].
Indeed, single crystal work on Mo5Si3 indicates a significant degree of thermal

expansion anisotropy [100], a/ac 2.2. Based on this degree of CTE anisotropy in

MoSSi,, tensile stresses near the grain boundaries as high as 1.8 GPa were predicted for the
cooling of a polycrystalline material from the melting point. The hexagonal Ti5Si3

compound has a CTE anisotropy of aa, = 2.3 [104]; doping TiSSi, with 0.85 formula

unit of carbon results in a reduction in the CTE anisotropy. In the case of undoped TiSi,,
CTE anisotropy has been predicted to cause thermally induced cracking at grain sizes

greater than 4 gtm [103].

2.3 Ternary Silicide Compounds

Compared to other areas of silicide development, information on alloying additions

and associated ternary silicide phases is somewhat limited. Early efforts in this area were
focused on development of M-Si-X ternary isotherms, such as the work of Nowotny [18,

106, 107], Brukl [108], and others [109, 110]. With the recent interest in silicide

materials for possible structural applications, ternary isotherms were used in studies of

alloying additions to MoSi2 [91, 92], composite design [12, 58, 77, 97, 111-113],

multiphase silicides [7, 14, 114-116], doping to improve oxidation resistance [101, 102],

and synthesis of ternary silicide compounds and composites [10, 74, 112, 117-120].
One reason for investigating ternary silicide compounds is based on the premise that

either their altered bonding arrangements or, in some cases, higher symmetry crystal

structures may allow for improvement upon the poor RT ductility and toughness of binary

silicides [74]. Macro-alloying has been used in A3X intermetallic systems to stabilize

higher symmetry crystal structures with resultant improvements in RT ductility. For

example, binary Co3V is brittle at ambient temperatures and has a complex hexagonal

(hP24) structure up to 10300C, where it transforms to a ordered-fcc L12 (cP4) structure.

By alloying with Fe, a (Co,Fe)3V compound having the L12 structure can be stabilized to

room temperature for iron concentrations in excess of about 11 At.% [121]. This increase

in crystal symmetry results in the activation of more slip systems, which in turn leads to an

increase in tensile ductility from 1% for Co3V to 40% for (Co, Fe)3V [11]. However, this

sort of crystal symmetry change through compositional control does not always correspond

to improvements in ductility, as is the case in AI3X compounds where alloying additions

can stabilize the L,1 structure, yet these cubic compounds still have only minimal ductility,

probably the result of a residual directionality in bonding limiting dislocation mobility and

resulting in an intrinsically low resistance to fracture [27, 122, 123].

In the case of Mo-Si-X compounds (Figure 2.9), the potential clearly exists to

synthesize cubic or hexagonal ternary silicide phases through macro-alloying. However, it

is unclear what effects these transitions to higher symmetry might have on the mechanical

properties of these compounds. Compared to the stabilized L12 compounds such as

(Co,Fe)3V that exhibit enhanced ductility, MoSi2 and other silicide compounds have in

general more covalent (Si-Si, Mo-Si) atomic bonding [33], and larger, more complex unit

cells. These factors make it less likely that alloying of silicides to increase their symmetry

can induce significant improvements in RT mechanical properties, although preliminary

studies and modeling efforts have touted the "disembrittlement" of MoSi2 through micro-

alloying [124] and the more "metallic" nature of some ternary silicides [75]. In addition to

possible ductility improvements, it has been speculated that ternary silicide compounds may

exhibit other interesting properties, either alone or in equilibrium with a metallic solid

solution [7, 114].

tetragonal 54%


monoclinc j

Figure 2.9 Distribution of crystal structure classes for 64 known Mo-Si-X ternary
compounds (X=AI, B, C, Co, Cr, Er, Fe, Hf, Nb, Ni, Re, Ta, Ti, U, V, W, Zr) [125].

2.3.1 Mo-Si-Al System

The Mo-Si-Al system has been studied for the purposes of determining phase

equilibria [106, 108], possible MoSi, deoxidation routes [126], and synthesis of the

substitutional Mo(Si, Al)2 (C40) compound [127]. Because of conflicting Mo-Si-Al

isotherms and obvious inaccuracies, a revised 14000C isotherm has been created [128]

(Figure 2.10).

It has been proposed that the hexagonal crystal structure of the C40 compound

could lead to RT ductility improvements [10]; however, recent single crystal work [129]

indicates that the compound has high critical resolved shear stresses for all slip systems and

no additional deformation mechanisms compared with MoSi2. Another area of study with

regard to the C40 compound was the potential for improved high-temperature oxidation

resistance due to a mixed alumina-silica oxide scale; results in this area have been mixed

[130, 131] and it has recently been discovered that carbon contamination during powder

processing can lead to degradation of the C40 compound at room temperature [39].

2.3.2 Mo-Si-B System

Because boron has been shown to enhance the oxidation resistance of Mo-rich Mo-

Si alloys, the Mo-Si-B system has been investigated for its phase equilibria [116, 132,

133], its oxidation resistance as a function of composition [99], the possible production of

in situ composites [77, 127], and the microstructural evolution of Mo-T2 alloys [114],

where T2 is the Mo5SiB, compound shown in Figure 2.11.

A rather unique feature of this system is the equilibrium between the molybdenum

solid solution and the ternary silicide phase (T2). In theory, this should make possible the

synthesis of alloys having a continuous ductile matrix (Mo) with the silicide phase(s)

hopefully lending increased oxidation resistance. However, as with other refractory metal-

silicide alloys [79], achieving microstructural optimization is not trivial [114].

2.3.3 Mo-Si-C System and the Mo SiC Phase

As previously discussed, the Mo-Si-C system is of interest from the standpoint of

removing silica from MoSi2 by carbon additions and producing MoSi2-SiC composites

through displacement reactions. Thermodynamics of the Mo-Si-C system

Initial work in the Mo-Si-C system was performed by Nowotny and coworkers

[18], who generated a 16000C isotherm (Figure 2.12) for the system. This isotherm has

several important features: equilibrium between SiC, MoSi2, and MoSi3C (T); the presence

of Mo3Si2, which is now recognized to be Mo5Si,, and MoC, which is not stable at 16000C

according to the currently accepted Mo-C binary diagram [134]; and a range of single-phase

stability for Mo5Si3C spanning Mo:Si ratios of approximately 1.2 to 2.0 and carbon

contents ranging from 6 to 15 At.%. The authors justified this broad compositional range

on the basis of variable filling of the atom sites for molybdenum and carbon [18].

Al Atomic %

Figure 2.10 Revised Mo-Si-Al 14000C isotherm [128].

$* mov Psyirn fs5 Si

Figure 2.11 Proposed isotherm for Mo-Si-B system at 16000C. Note the
TI(Mo,(Si,B)3) and T2 (MoSiB2) phases [132].

The following formula was proposed for the compositional range of the Mo.Si,C,,

phase (so designated to indicate its variable composition),

Mo6(Six, Mo,-x)4(C,,Si-,)6 (2.5)
where x varies from 0.1 to 0.55 and y varies from 0.15 to 0.4. The authors also noted

variations in axial c/a ratio from 0.695 for Mo-rich compositions to 0.687 for Mo-lean

compositions in this study, which was interpreted as indicating a broad range of

homogeneity for the ternary phase.

Renewed interest in the Mo-Si-C system for carbothermal deoxidation and

production of in situ MoSi2 composites in the late 1980s drew attention to the

aforementioned phase equilibria of Nowotny as well as the more recent work of van Loo

[19] and others [135]. Using XRD and electron microprobe (EPMA) analysis of arc

melted specimens and diffusion couples, van Loo and coworkers proposed a 1200C

isotherm (Figure 2.13) of the Mo-Si-C system.

Two principal differences exist between the Nowotny and van Loo isotherms: (1)

van Loo shows MoSi, and SiC in equilibrium with MoSi3, not MoSi3C and (2) the region

of MoSi3C single-phase stability is significantly reduced on the van Loo isotherm. Issue

(1) was addressed by Costa e Silva and Kaufman [136], and it was shown through

thermodynamic calculations and systematic experimental investigation of phase relations,

that Mo.SiC is in equilibrium with MoSi2 and SiC at both 16000C and 12000C, i.e., the

form of the Mo-Si-C isotherm as given by Nowotny (Figure 2.12) is correct. However,

recent work [12, 120] has tended to support the conclusion of a significantly smaller

homogeneity range for MosSi3C than that reported by Nowotny. A similarly reduced

compositional range was recently reported for the T1 phase in the Mo-Si-B system (Figure

2.11) [115]. The original Mo-Si-C phase equilibria study also yielded a liquidus projection

for the Mo-Si-C system (Figure 2.14), indicating a melting point for the ternary phase

above 20000C. Atomic arrangement and crystal structure of MoQSibC

The structure of the Mo5Si3C phase was investigated in detail by Parthe and

collaborators [137] using neutron diffraction. It was determined previously that the ternary

phase had an MnsSi3-type (hP16, D8g) hexagonal crystal structure, but at the time, there

was still question as to how the carbon atoms were arranged in the lattice. Four possible

structure proposals for Mo5Si3C were outlined by Parthe:

1. Partial occupation of silicon 6(g,) positions by carbon atoms with Si atoms
occupying former Mo positions (4(d));

2. Carbon atoms substituting on former Mo positions;

3. Carbon completely occupying octahedral interstitial sites (0,0,0 and 0,0,1/2);

4. Octahedral sites filled variably by C atoms and variable occupancy of 4(d) positions
with Mo atoms.
By examining x-ray diffraction (XRD) and neutron diffraction data, Parthe

determined that proposal (4) is correct and can account for the broad range of homogeneity

of MoSi3C reported by Nowotny [18]. The corresponding structure (Figure 2.15) has the

P6,/mcm space group with 6 Mo atoms in 6(g,) sites with x=0.24, 3.6 Mo atoms in 4(d)

sites, 6 Si atoms in 6(g,,) with x=0.6, and 1.2 carbon atoms in the 2(b) sites. The values

for the lattice parameters a and c are 0.728 nm and 0.505 nm, respectively [137]. Processing, microstructures and properties of Mo Si&C

Despite the relatively large number of studies concerning the Mo-Si-C system, only

three studies (excluding the current work) were found directly applicable to the processing

and properties of MosSi3C-based materials. The first is the original phase diagram study of

Nowotny et al. [18], for which materials were hot pressed from elemental Mo, Si, and C

powders followed by heat treatment for 12 h at 16000C. The only microstructure presented

in the study is shown in Figure 2.16; it is difficult to discern much from this micrograph,

although the sample is quite porous and appears to have features or phases near several

grain boundary triple points. In this study, density and Vickers microhardness values of

8.0 g/cm3 and 14.3 GPa (50g load), respectively, were reported for MosSiC.

Si / 4 Mo
MoSl2 Mo3SI2 M0381
Atomic %

Figure 2.12 Accepted 16000C Mo-Si-C isotherm at [18]. The ternary silicide Mo5Si3C is
designated by T, and is shown to be in equilibrium with SiC and MoSi2 at this temperature.
Note what is now accepted to be Mo5Si, is labeled "Mo3Si2" and the presence of MoC, in
contrast to the now accepted decomposition of MoC into Mo2C above 16000C [134].

Si / I N/ Mo
MoSi2 MosSi3 MoaSi
Atomic %

Figure 2.13 Isotherm of Mo-Si-C system at 12000C [19]. Note the existence of a three
phase equilibrium between MoSi2, SiC, and MoSi3 along with the significantly smaller
MoSSi3C single phase field.

Figure 2.14 Liquidus projection for the Mo-Si-C system [18].

Figure 2.15 Schematic of the D8, crystal structure of MosSi3C, showing carbon atoms
occupying the octahedral interstitial sites of Mo in 6(g) [137].

Quite recently, Suzuki and Niihara [120] investigated MoSSi3C-based compounds

as potential high temperature structural materials. Starting with Mo, Si, and C powders (all
< 10 mun) wet ball-billed in acetone for 48 h followed by hot pressing at 15000C using 30

MPa pressure, samples were produced at a nominal composition corresponding to

Mo56,,Si36,C7,. The authors report single phase materials at this composition on the basis
of XRD and SEM characterization, although their "single phase" structures (Figure 2.17)

contain roughly 4 Vol.% of what is claimed to be SiO2. This is somewhat puzzling based

on the amount of carbon (1.3 Wt. %) which should be more than sufficient to reduce any

silica present on the powders through reactions like those shown in Equations 2.1-2.3. A

possible explanation is that the fine starting powders had increased oxygen content owing

to their high surface area, and the amount of carbon added was insufficient to fully reduce

the silica present. In fact, the authors report a slight carbon gain during hot pressing,

again inconsistent with the operation of most silica reduction mechanisms. For nearly-

single phase MoSi3C materials, the authors report an average density of 7.58 g/cm3

Figure 2.16 Microstructure of ternary MosSi3C phase from Nowotny et al. [18]. The
sample has been etched, explaining the rough surface. The authors report the black regions
to be sintering porosity.

(theoretical density 7.9 g/cm3), average RT 3-point bend strength of 430 MPa, a Vickers

hardness of 12.1 GPa, and a fracture toughness of 3.4 MPa-ml'.


Figure 2.17 Scanning electron micrograph of a MosSi3C sample hot pressed for 2 h at
15000C. Dark globular regions are reported to be silica [120].

A final study was that of Thom and coworkers [102] who explored the possibility

of doping MosSi3 with B and C, primarily to improve oxidation resistance; they noted

carbon additions did not have any beneficial effects on oxidation properties, as a sample

with nominal composition MosSi3C,, underwent rapid MoO3 volatilization and weight loss

in air at 10000C. For samples HIP consolidated to 80 % density, Vickers hardness

values increased with carbon content up to a maximum of 12.8 GPa for one formula unit of

carbon (i.e., the idealized MosSi3C composition).


3.1 Introduction

The primary goal of the processing optimization portion of this study was to

determine what combination of processing method and composition yielded the best

material with respect to such features as single-phase microstructure, sample homogeneity,

gas and sintering porosity, macro- and microcracking, and oxide inclusions. These

microstructural features were deemed important based primarily on their potential effect on

the mechanical properties and deformation behavior of MoSi3C-based materials. Potential

processing schemes were also assessed on the basis of their repeatability, ease of scale-up,

and relative complexity. In order to investigate the effects of grain size on the mechanical

behavior of Mo5Si3C, it was also necessary to determine which processing routes allowed

for adequate control of the scale of the microstructure. Because of the previously

mentioned uncertainty regarding the compositional range of MoSi3C, it was also of interest

to re-examine phase stability and equilibria issues in the near-Mo5Si3C region of the Mo-Si-

C system.

3.2 Materials and Experimental Procedures

3.2.1 Raw Materials

Choice of raw materials form, size, and purity (Table 3.1) varied with processing

route and were limited in most cases to those which were produced commercially. When

commercially-produced materials were used, every attempt was made to obtain information

from the vendor as to the impurity content and type, particle size and size distribution, and

lot number so that the effect of starting materials as a process variable could be properly


Table 3.1 Raw materials used in the production of Mo5Si3C-containing samples.

Material Form Purity Size Source Lot Number
Carbon pieces 99.9 2-3 mm POCO
(graphite) Graphite, Inc.
Molybdenum rod 99.95 3 mm Alfa AESAR H23E06
Silicon lump 99.9999 2-10 mm Alfa AESAR 19294
Carbon powder 99 <50 pm Alfa AESAR G07E05
(graphite) ____________ __
Molybdenum powder 99.9 45-100 tm Cerac X17221
Molybdenum powder 99.9 <45 pm Cerac X 1672
Molybdenum powder 99.7 3-7 .tm Alfa AESAR D27B25
Silicon powder 99.999 45-150 nim Cerac 125889-C-1W
Silicon powder 99.9999 63-75 .m Alfa AESAR 20728
Silicon powder 99.999 <45 pm Alfa AESAR B20G04
Mo2 powder 99+ Cerac 19913
Mo2C powder 99.5 <45 m Cerac 117807-A-1-2
MoSi2 powder 99.995 <150 Im Cerac X16146-1
Mo5Si3 powder 99.5 5.4 gm Cerac 133457-A-1
MoSi3 powder 99.5 <45 Im Cerac 142793-A-(1-4)
MosSi3 powder 99.5 <45 pm Cerac 153888-A-(1-3)

3.2.2 Processing Routes
Based on previous work in our laboratories [13, 58, 64, 126] and others [48, 70,

103], the following processing routes were thought to have the most potential for the
reproducible production of MosSi3C-containing samples: (1) non-consumable arc melting;
(2) mixing of powders followed by hot consolidation; and (3) mechanical alloying of
elemental and/or compound powders followed by hot consolidation. Arc melting

Arc melting was done with a non-consumable thoriated tungsten electrode in a

Centorr model 5 Bell Jar arc melting furnace powered by a Miller Gold Starm SS 400 DC

welding power source. At the beginning of each melting session, the chamber and water-

cooled copper hearth were thoroughly cleaned and 25 g of melt stock consisting of

molybdenum rod, silicon chunks, and graphite pieces were arranged in the melting

receptacles within the bell jar. Before melting, the Mo pieces were ground with 400 grit

SiC paper to remove any oxide present on the surface. The bell jar was evacuated to 13.3

Pa followed by back-filling to -50 kPa with ultra high purity argon (< 0.5 ppm total

hydrocarbons); this procedure was repeated a minimum of three times to ensure a clean

melting environment. In order to scavenge any remaining oxygen, titanium getter buttons

were melted just prior to applying the arc to the melt stock. The arc power was controlled

using a foot pedal with typical currents on the order of 250 A.

In order to enhance the amount of carbon going into solution while reducing the

potential for silicon losses through vaporization, melting was done in two stages, wherein

the molybdenum rod and graphite pieces were melted separately and then brought into

contact with the silicon chunks and re-melted, drawing in the silicon as it melted. In both

melting steps, samples were melted and flipped over a minimum of five times to enhance

homogenization; whenever necessary between melting runs, the chamber was opened and

cleaned with a rotary wire brush and rinsed with acetone. In order to quantify weight

losses during arc melting, samples were weighed before and after each melting step. Powder mixing and hot pressing

The majority of the samples in this study were processed by blending commercially

available powders followed by hot pressing. For this processing method, various

combinations of (Table 3.1) molybdenum, Mo2C, MosSi,, silicon, and carbon powders

were measured out and then blended for 2 h on a tumbler ball mill. Ball milling was

conducted in 250 mL polyethylene vials with ZrO2 milling media at a 3.5:1 (by weight)

ball-to-milling charge ratio. After ball milling, the powders were loaded into high-density

POCO HPD-1 graphite tooling (Figure 3.1) consisting of three parts: (1) a cylindrical die

with an inner cavity; (2) a top punch to which is loaded by the hydraulic actuator on the hot

press; and (3) a shorter bottom punch which holds the powders inside the die. For the

processing study and phase stability portion of this work, graphite dies having an inner

diameter of 16 mm (5/8") were used, while larger, 32 mm (1.25") inner diameter dies were

used for synthesis of samples for mechanical testing. In all cases, the die surfaces were

coated with hexagonal boron nitride spray to provide lubrication at processing temperatures

and to aid in removal of the consolidated specimens. As an additional aid in the removal of

consolidated samples and to minimize sticking of the powders to the die and ram surfaces,

0.13 mm thick GRAFOIL flexible graphite paper was used to line the die inside diameter

as well as the top and bottom punch/powder contact surfaces.

After preparation and charging of the graphite die with the blended powders (-10 g

for the smaller dies, -45 g for the larger dies), the assembly was placed inside a Centorr

Model 600 vacuum hot press (Figure 3.2), which was used to consolidate all powder-

processed samples in this study. The Model 600 allows for application of uniaxial pressure

through a hydraulic actuator and ram in a vacuum (less than 0.13 Pa) or inert gas

atmosphere. Heating inside the water-cooled stainless steel chamber was accomplished by

induction heating coils powered by a 15 kW TEK Specialties, Inc. MG-15 induction power

supply capable of sustaining temperatures in excess of 1700C for the setup used. A quartz

susceptor lined with graphite felt insulation surrounded the die and punch assembly, a

graphite spacer block, and two 1.3 cm thick squares of refractory brick, stacked in the

order shown in Figure 3.1. A two-color Capintec Ratio-Scope 8 optical pyrometer was

used for temperature monitoring in the hot press; when focused through a viewport in the

chamber, the pyrometer measures the temperature at the surface of the graphite die with an

accuracy of approximately 200C over its operating range (6500C to 22000C).

Variations on a two-stage HP approach (Figure 3.3) were used to produce blended

powder samples, with an initial hold under vacuum at temperatures ranging from 1350 to

15000C followed by a densification step under argon at 16500C or 17000C. In some cases,

uniaxial pressure was applied to the sample during the initial heating and low-temperature

hold cycle (pre-load). Typical heating rates were on the order of 50C/min, while cooling

rates ranged from 16 to 25 C/min. After densification, pressure was removed and the

sample cooled either by reducing the power completely and allowing the sample to furnace

cool, or by ramping down the power on the hot press incrementally for 1 h followed by

furnace cooling. For all hot press runs, data on temperature and pressure were recorded at

prescribed intervals; several actual processing profiles are presented in Appendix A. After

cooling and removal of the sample from the die using a mounting press, the sample

dimensions and weight were recorded and approximately 1 mm was ground off of each

surface to remove the adhered graphite foil and any reaction zone.

For heat treatment of the consolidated samples, a Centorr model M60 with Ta

heating elements was used. Heat treatment was for 12 h at 16000C under high-purity argon

followed by furnace cooling at approximately 15 C/min. Mechanical allowing + hot pressing

The starting powders for mechanical alloying (MA) consisted of molybdenum or

MoSSi, mixed with silicon and carbon. Initially, the powders were blended in polyethylene

bottles on a tumbler ball mill using ZrO2 milling media at a ball-to-charge ratio of 3.5:1 by

weight. After blending for approximately 1 h, the powder charge was weighed and placed

in a glass vial; the glass vial and a hardened steel grinding vial containing 0.6 cm diameter

hardened steel balls were then placed in a glove bag. The glove bag was flushed with

argon several times before final inflation and sealing, at which point the powders were

transferred to the grinding vial. After tightly sealing the grinding vial, it was removed from

the glove bag and placed on a SPEX 8000 Mixer/Mill where it was milled for 15 to 18 h.

Hydraulic ram

Top punch
Graphite die
-j" Sample
o Bottom punch
a "Graphite spacer

o Insulating bricks

Induction coil

Figure 3.1 Schematic of setup for hot pressing of Mo5Si3C-based materials. The die and
punch assembly diagramed on the right side of the figure was placed within the quartz
susceptor and surrounded by graphite felt insulation.



Figure 3.2 Centorr Model 600 uniaxial vacuum hot press with Captintec Ratio-Scope 8
two-color pyrometer.

Figure 3.3 Schematic representation of the two-step hot pressing process showing
heating and hold stage under vacuum (1) and the high-temperature consolidation step under
argon (2). Also indicated are two possible pressure application routes, with or without pre-

Powder loadings varied from 10 to 25 g at a constant 2:1 weight ratio of grinding

media to charge. For several milling runs, 0.5 g stearic acid (C8,HO) was added to the

milling charge and a mixture of stearic acid and methanol was used to coat the milling vial

and balls to minimize powder caking and agglomeration. After milling, the powders were

removed from the milling vial and placed back on the tumbler ball mill for several hours to

break up any large agglomerates which may have formed during the MA process.

After ball milling and MA, the powders were loaded into graphite dies and hot

pressed in a manner similar to that already described. For the MA samples, two different

hot pressing (HP) schemes were employed, a single-stage process and a two-stage

process. In the first process, samples were pressed at 1400C for 1 h at a uniaxial pressure

of 47 MPa in a vacuum of less than 0.13 Pa. Samples processed using the two-stage

method were initially held at 15000C for 0.5 h without applied pressure in vacuum,

followed by a densification step of 1 h at 16500C under 47 MPa pressure in an argon

overpressure of roughly 50 kPa.

3.2.3 Materials Characterization and Analysis

To examine the effects of process variables and target compositions on sample

microstructures and to assess the effectiveness of various processing routes, both starting

materials and consolidated samples were subjected to a number of standard materials

characterization and analysis techniques. Physical properties

In order to verify the stated powder size and distribution of commercially obtained

starting powders and also to examine the effects of various mixing and milling techniques

on powder size, particle size analysis (PSA) was performed. In preparation for PSA,

approximately Ig of powder was mixed in a water suspension and stored in a 50 mL

polyethylene bottle. The powder suspension was loaded into a receptacle filled with de-

ionized water in a Coulter LS 230 Particle Size Analyzer which uses a laser-optical

measurement system to measure particle size. To reduce the potential for agglomeration,

powders were also analyzed after addition of a Trinton X-100 wetting agent.

For both arc melted and hot pressed specimens, weight changes during processing

were monitored to aid in the evaluation of processing techniques and parameters. The solid

pieces comprising the raw materials of arc-melted samples were weighed out using a

Denver Instrument Co. S-110 electronic balance having a resolution of 1 mg; the weight of

the arc melted button after final melting was recorded after cleaning with acetone and

drying. In some cases, arc melted buttons were weighed in between melting stages as

well. The weight of powders charged into dies for hot pressing was recorded prior to

pressing and compared with the weight of consolidated samples. Sticking of powder to the

graphite dies and reaction of samples with the graphite foil liner complicated accurate

determination of post-HP weight; typically a sample was measured as removed from the die

and again after grinding off the residual graphite foil, giving a range of high and low

consolidated sample weights.

To estimate the degree of densification in powder-processed samples, the apparent

density was measured by a method derived from ASTM Designation C 134-95 [138].

After sectioning a small cube of material from the center of the sample using a Buehler"

Isomet diamond saw with a Buehler low-concentration diamond watering blade, the cube

was ultrasonically cleaned in acetone and dried. The mass of the dry sample was measured

on the electronic balance and recorded. Using an INOX digital caliper having 10 Km

resolution, the length, width and height of the sample were measured a minimum of five

times. Using the appropriate geometrical relations, the volume of the cube was determined.

Knowing the mass and the volume of the specimen, an apparent bulk density (p) was

determined using the following:

p =m/V (3.1)

where m is the mass of the sample in g and V is the volume of the sample in cm3. Microstructural analysis

Metallographic preparation. Prior to examination of sample microstructures,

specimens were prepared using standard metallographic techniques. After sectioning of the

sample using a diamond saw, the specimen was mounted in 2.54 cm diameter mounts of

either quick-setting epoxy or a phenolic resin. The mounted sample was then wet ground

to a flat, smooth surface using a rotating grinding wheel and a series of silicon carbide

grinding papers (180 to 800 grit). Any remaining scratches were removed by diamond

polishing on Buehler TEXMET" polishing cloths with 6 gm and 1 gm water-based

diamond slurry. As a final polishing step and to accentuate any phase relief in the samples,

samples were mounted into holders and polished for 20 minutes on a Buehler Vibromet

2 vibratory polisher using napless nylon cloth and a 0.2 gm silica suspension. After

grinding and polishing, samples were ultrasonically cleaned in a mixture of water and

ultrasonic cleaning solution, rinsed with methanol, and dried using a hot air blower.

Light optical microscopy. Initial examination of microstructural features such as

porosity, secondary phases, cracking, and grain size was conducted using a Nikon

Epiphot inverted metallograph equipped with a high-intensity light source and a Polaroid"

film back. The metallograph has four magnifying lenses coupled with a 10x objective and

is capable of 50 to 1000x magnification; filters and polarizers including differential

interference contrast and Nomarski diffraction enhance the available imaging options.

Light optical microscopy (LOM) was also used to check for scratches, beveling,

discoloration, and other undesirable metallographic artifacts. The optically anisotropic

hexagonal Mo5Si3C phase displayed enhanced grain contrast without etching, simplifying

determination of the mean grain diameter.

Scanning electron microscopy. For more detailed microstructural examination,

scanning electron microscopy (SEM) was performed on polished specimens using JEOL

35CF, JEOL 6400, and JEOL 733 electron microscopes. Strips of conductive carbon paint

were painted on specimens prior to insertion into the microscope to minimize charging

effects. At a typical accelerating voltage of 12 kV, both secondary electron and

backscattered electron images at both low and high magnification were recorded with a

Polaroid* film back.

Quantitative image analysis. Images of sample microstructures collected from LOM

and SEM were analyzed for quantitative information on three primary microstructural

features of interest: (1) phase volume fraction; (2) volume fraction of porosity; and (3)

determination of grain size. The first step in the image analysis procedure was digital

capture of the images using a flatbed scanner followed by manipulation and enhancement of

images in Adobe PhotoshopM. The digitally enhanced image in PICT format was then

transferred to NIH Image Ver. 1.61 image analysis software (distributed as freeware); NIH

Image is capable of performing automated areal analysis for determination of phase and

porosity volume fraction and aids in the calculation of average grain size using linear

intercept methods.

For areal analysis in NIH Image, the feature of interest is converted to pure black

against a white background by making the image binary. Next, the scale is set based upon

the micron bar imprinted on the SEM image, or in the case of LOM images, a Vickers

microhardness indent of known size was used to calibrate the scale. The area of interest

was then measured, followed by analysis of total phase/particle area; dividing the area of

particles/phase by the total measurement area yields the area fraction which correlates to the

volume percent of a particular feature. This measurement process was repeated at least five

times per image and a minimum of five images were used per sample to obtain average

phase and porosity volume fractions.

To determine mean grain size, NIH Image was used with a line intercept method

similar to that described in ASTM E 112-85 [139]. Using a minimum of five 50 mm test

lines per region of interest, converted to the proper scale using the previously described

methods, an average number of grain intercepts per test line length was determined from

which the average grain diameter was derived. Although grain size analysis was carried

out on both SEM and LOM images, the latter was preferred due to the better grain contrast

achieved through the use of a polarizing filter in the optical microscope. Phase identification and compositional analysis

X-Ray diffraction. For obtaining crystallographic information and phase

identification (phase ID), x-ray data was collected using a Phillips APD 3720 x-ray

diffraction (XRD) system. Samples for XRD were either in powder form, as pre-

processed powders or comminuted consolidated specimens, or solid 1 cm x 1 cm x 1 mm

thick slices from consolidated samples. Powder XRD samples were prepared by mounting

<75 pm powders to an x-ray slide with double-sided tape or a suspension of amyl-acetate

colloidian. Solid specimens were sectioned from bulk samples with a diamond saw and

affixed to x-ray slides using crystal mounting adhesive or double-sided tape; to maintain

proper spatial orientation with respect to the x-ray source and detector, a dummy of the

same height as the solid sample was made from microscope cover slides and tape and

affixed to the x-ray slide. Using a Cu Ka x-ray source at a wavelength of 0.154056 nm,
data was collected from 10 to 1000 20 at 30 20/min with an intensity range of 1000
counts/sec full-scale.
After filtering the raw x-ray spectra to obtain the peaks present in the sample, the
data were analyzed to determine the phases present and the lattice parameters of the

MoSi3C phase. For phase identification, a standard method [140] was used, comparing
interplanar spacings (d-spacings) and relative peak intensities in the x-ray data with d-
spacing and intensity information taken from the JCPDS (Joint Committee on Powder
Diffraction Standards) database on phases likely to be present in the sample. A positive
phase ID for a given peak was arbitrarily defined to be a difference in d-spacings between
experimental data and JCPDS file of -0.0005 nm for large d values and -0.0002 nm for
smaller d-spacings; a reasonable correlation between expected and observed relative
intensities was also deemed necessary for a good match. A slightly modified procedure
was used for identification of MosSi3C because of uncertainties in the published x-ray data
resulting from its variable composition; observed x-ray peaks were compared with a
combination of JCPDS powder diffraction files 8-429 (deleted) and 43-1199, the data of
Parthe et al. [137], and simulated diffraction data obtained from Virtual Laboratories
Desktop Microscopist software application. Typically the eight highest intensity peaks
were identified first to determine major phase constituents) followed by identification of
successively less-intense peaks.
Lattice parameter determination for the Mo5Si3C phase was done by a simplified
regression technique using the relationship between lattice parameters (a, c), d-spacing (d),
and planar indices (hkl) for hexagonal crystals,

d= 4(h + k (3.2)
4(h2 +hk +k
a3 2 + C2

Using the observed interplanar spacings (dob) for (hk0) and (001) planes, baseline values

were determined for a and c, respectively. For the remaining (hkl), a and c values were

calculated by inserting the baseline a and c values (from (hkM) and (001) peaks) into Eq.

3.2 and varying these values incrementally until the calculated d-spacing matched db, for

the given plane within 0.0001 nm.

Wavelength-dispersive spectrometry. Compositional information and phase ID

using wavelength-dispersive spectrometry (WDS) was done on a JEOL 733 electron probe

x-ray microanalyzer (EPMA) operating at an accelerating voltage of 12 kV with a spatial x-

ray resolution of ~ 1 im. Standards for elements of interest (Table 3.2) were used to set

up an automated quantitative analysis routine and their calibration was re-checked at the

outset of each EPMA session. After generation of x-ray spectra from a given sample, a

PROZA quantitative correction algorithm was run to obtain compositional information in

atomic and weight percent. For a given region on a polished sample, the typical procedure

was to record x-ray spectra from a minimum of three distinct points per phase present; by

spreading the electron beam to 50 gm, information on bulk composition was obtained. In

addition to electron probe microanalysis, x-ray dot maps were generated by selecting the

appropriate channel on the image selector corresponding to the spectrometer crystal

positioned appropriately for the element of interest.

Table 3.2 Elements measured and standards used for WDS analysis in the EPMA.

Element X-Ray Spectra Emission Energy Standard
Mo Lao 2.52 molybdenum
Si Ko 1.838 silicon
C Ka 0.283 SiC
0 Ka 0.531 quartz
Fe Kao 7.112 iron

Light element analysis. Measurements of minor amounts of elements with Z < 10

using WDS is characterized by low characteristic x-ray intensities, low peak-to-background

ratios, interference from heavier elements, peak shifting, and surface contamination. These

factors contribute to a minimum detection limit of 0.1 to 1 Wt.% for elements with Z < 10

[141]. Because of this, combustion-based light element analysis was performed at LECO

Technical Services Laboratory (St. Joseph, MI49085) on several hot pressed samples for

more accurate determination of bulk carbon, oxygen, and nitrogen contents. Carbon

measurements were done on a LECO CS-444 instrument in which Ig specimens are

ignited by an induction furnace and the combustion stream is analyzed by infrared (IR)

absorption. A LECO TC-436 nitrogen/oxygen determinator was used for measuring

nitrogen and oxygen contents; solid (-2 g) samples were combusted by inert gas fusion

with nitrogen measured by a thermal conductivity cell and oxygen detected through an IR

absorption cell.

Transmission electron microscopy. Transmission electron microscopy (TEM) was

performed on as hot-pressed samples as well as specimens having undergone high-

temperature compressive deformation. Analysis was conducted on thin foils in JEOL

200CX and Phillips 420 transmission electron microscopes, operating at 200 kV and 120

kV, respectively. Samples for TEM were initially prepared by sectioning -400 pim thick

slices from 3 mm diameter cylinders using a low-speed diamond saw. Two to three slices

were then mounted with crystal mounting wax to a mounting peg and placed in an

adjustable-height Gatan disc grinder; samples were ground through a series of SiC grinding

papers from 400 to 1200 grit in 25-50 gm increments until a thickness of approximately

150 gim was attained. A Dimpler model P300 rotary thinning tool with 6 imn diamond

slurry and a 20 g force offset was used to reduce the foil thickness at the center to

approximately 20 lm. Dimpling was followed by ion milling on a Gatan Model 600 dual

ion mill at 6 kV and 1 mA with an initial milling angle of 180, adjusted to 120 for about 1 h

after initial foil perforation. Through this procedure, the area adjacent to the hole in the foil

was thinned to a degree sufficient for electron penetration.

Output from TEM analysis included bright-field and dark-field images along with

selected-area and convergent-beam diffraction patterns. Resultant diffraction patterns were

indexed by standard techniques [142] to determine lattice constants and phase identity.

3.3 Results and Discussion

3.3.1 Processing

Consistent with the primary project goal of examining the mechanical behavior of

Mo,SiC, the processing of Mo5Si3C-based materials was undertaken within the context of
what combination of processing routes and process variables allowed for the production of

nearly single-phase, fully dense, crack-free samples of various grain sizes. It was

recognized from the outset that achieving all goals in terms of material homogeneity and

quality could be quite challenging, thus an emphasis was placed on optimizing

microstructures as much as possible with respect to sample homogeneity, integrity, and

microstructural scale while attempting to quantify and correlate undesirable sample features

when they appeared. In addition, every attempt was made to assess processing routes and

compositions in a systematic fashion so that the effects of specific process variables and

stoichiometry changes could be delineated. Complementary to this systematic, iterative

approach was an emphasis on quality and process monitoring during all processing stages

and an effort to develop a repeatable set of procedures for each process thereby reducing

potential variability. Compositions studied

An initial step in the processing study of MosSi3C was the identification of target

compositions for arc melting and hot pressing. As previously mentioned, a significant

degree of stoichiometric variation for MosSi3C is reported in the literature [18], indicating

that single-phase MosSiC is stable over a range of compositions (Figure 3.4). As a way to

examine potential compositional effects on mechanical properties, it was decided initially to

produce samples of two compositions (1 and 2 in Table 3.3) in the MoSiC single-phase

field having different amounts of carbon but the same Mo:Si ratio. However, over the

course of the project, a significantly expanded number of compositions was produced,

primarily because of difficulties in achieving single-phase MosSi3C, which will be

discussed in more detail below. Arc melting

Non-consumable arc melting (AM) was considered as a potential processing

method for MoSi3C based on its reputation as a quick, efficient way to melt high-melting

point metals for alloy development. One primary advantage of arc melting over powder

metallurgy-based methods is the much higher degree of sample cleanliness possible

through arc melting. A controlled, high purity atmosphere is used during arc melting in

addition to oxygen-gettering titanium buttons, and the solid melt stock are all expected to

lead to lower contamination levels than present in their powder-processed counterparts.

Compositions produced through arc melting were I and 2 (Table 3.3), containing 13.8 and

11.4 At. % carbon, respectively.

An as-cast microstructure for one of these alloys is shown in Figure 3.5 and clearly

displays a multiphase structure. This type of microstructure was observed for all arc-

melted specimens, and XRD indicated that the major phase was MoSi3, not MosSi3C,

meaning that the samples were not fully equilibrated and the carbon had not dissolved

completely. This is consistent with observations during the melting process of the graphite

pieces used as the carbon source not dissolving under the arc, but rather floating on the

melt button even after multiple re-melts. Similarly, the Mo-carbide visible in Figure 3.5 is

likely a residual unmelted Mo-C master alloy.

C: 58Mo-29Si-13C Mo:Si Ratio
A: 48Mo-41Si-llC + A: 1.17
** S C Mo2C B: 1.50
MoSiC (T) C: 2.00
D: 1.76
+ MoSi D: 60Mo-34Si-6C .. E: 1.24
Mo Si 3 "-)
E: 52Mo-42Si-6C ..

T+MosSi3 ,.

Figure 3.4 Close up of MoSi3C single phase region taken from the accepted Mo-Si-C
16000C isotherm (Fig. 2.10) by Nowotny et al. [18]. The compositions for the five
terminal points (A-E) of the single phase field were estimated from Eq. 2.4 given for the
compositional variation of MosSi3C in [18].

Table 3.3 Compositional information for samples produced via arc melting (AM),
powder blending followed by hot pressing (PBHP), and mechanical alloying followed by
hot pressing (MAHP). All compositions shown are nominal (as measured).
Comp. Mo Si C(At. Mo:Si Processing Total No.
ID (At.%) (At.%) %) Ratio Routes Samples
1 53.5 32.7 13.8 1.64 AM 5
2 55 33.6 11.4 1.64 AM 4
7 54.5 33 16 1.65 MAHP, AM 10
10 55.5 33.5 11 1.66 PBHP 3
11 59 35.6 5.4 1.66 PBHP 1
13 52.3 37 10.7 1.41 PBHP 2
15 57.2 34.7 8.1 1.65 PBHP 2
C05 59.4 35.65 5 1.67 PBHP 1
C75 57.9 34.6 7.5 1.67 PBHP 1
C8B 56.2 35.8 8 1.57 PBHP 1
C10 56.3 33.7 10 1.67 PBHP 2
C10B 55.3 34.7 10 1.59 PBHP, 19
C125 54.7 32.8 21.5 1.67 PBHP 1
C15 53.2 31.8 15 1.67 PBHP 1

Figure 3.5 Back-scattered electron (BSE) image of CN 2 sample produced by arc
melting. Phases displaying dark contrast are SiC, while the lighter-contrast necklace of
features are MoC. The majority phase constituent is MoSi,.

The lack of homogeneity in the samples points to the difficulty of combining

elements having a broad range (Si-1685 K, Mo-2890 K, C-4100 K) of melting

points-even with the two-step melting process used, attempts to homogenize through

further re-melting would have resulted in excessive losses due to silicon vaporization. Heat

treatments under a controlled atmosphere are another common approach to alloy

homogenization, but the extremely slow decomposition kinetics reported for molybdenum

carbides [137] would have resulted in excessive heat treat times (hundreds of hours) and

therefore, this approach was rejected as impractical for the purposes of this study.

Arc melting presented further complications with respect to producing samples

having sufficient integrity for mechanical testing as all specimens contained significant

shrinkage porosity and pervasive macrocracking after solidification (Figure 3.6). The

massive cracking observed is probably a result of the large temperature gradient between

the top of the arc melted button most recently under the welding arc and the bottom of the

button in contact with the copper chill plate. This gradient likely generates thermal stresses

that cannot readily be accommodated by the material; the hexagonal D8, structure of

MosSi3C may result in additional susceptibility to thermally-induced cracking because of

probable thermal expansion anisotropy, seen in other non-cubic intermetallics such as the

isostructural Ti5Si3 compound [103, 104].

Figure 3.6 Scanning electron micrograph of AM sample of CN 7 showing significant
cracking in the as-cast microstructure. The lighter major phase is MosSi3, the dark gray
elongated phase is MoSi2 and the black regions are SiC and porosity.

A further impediment to the effective use of AM as a means to produce mechanical

test specimens is the lack of grain size control inherent in the process. Typical matrix grain

sizes of arc melted samples are in excess of 100 pm; to vary this reproducibly with the

current setup would be very difficult given the low degree of heat input control and the

basically fixed cooling rate imposed by the water cooled copper hearth. Traditional

methods of refining and/or modifying the as-cast structure by thermo-mechanical

processing have been successfully applied to aluminides [143], although it is doubtful

these methods would be feasible for the heavily-cracked silicide buttons produced in this


33.1.3 Powder processing methods

Given the difficulties encountered during arc melting, processing approaches based

on the pressure-assisted sintering of powders were evaluated for production of MosSi3C-

based materials. The primary reason for considering methods based on powder mixing

(PM) was the large degree of flexibility in starting materials, with parameters such as form

(elemental or compound), purity, powder size, and size distribution all having the potential

to be tailored for a given sample. Varying starting powder size is an obvious way to affect

changes in microstructural scale [52], and the more controllable heating and cooling rates

available through hot pressing is a further aid to grain size control. Another potential

advantage to PM techniques is that by optimizing process parameters, near-fully dense and

crack-free bodies can be produced. A final reason to consider powder processing for

MoSSi3C-based materials was the large experience base in our research group in powder
processing of silicide compounds [13, 58, 73, 126].

The two-stage hot pressing scheme used throughout this study was arrived at based

on the work of Henager et al. [72] and Jayashankar [63] on processing of in situ

composites in the Mo-Si-C systems. In both cases, these researchers used a lower

temperature hold followed by a higher temperature densification step. The initial hold is

necessary to allow the volatile reaction products of carbon and siliceous powders to leave

the sample; it is important that this step be performed under vacuuum with only minimal

application of mechanical pressure so as to not entrap gas in the sample (Figure 3.7). The

densification step is carried out at higher temperature to accelerate the diffusional processes

necessary for powder consolidation; this step is done under an applied pressure of 50 MPa

to further assist densification while the HP chamber is backfilled with argon to minimize

potential silicon losses due to the high vapor pressure of Si at temperatures above 16000C

(Figure 2.4).

Powder mixing + hot pressing. The compositions prepared using this method

made up the majority of those studied in this work (Table 3.3) and covered a range of

carbon contents from 5 to 16 At.% and variation in Mo:Si ratio of 1.41 to 1.67. As seen in

Table 3.4, elemental and compound powders covering a broad range of powder sizes were

used as raw materials.

'. ; "

'. .

... *. .

.w *" ; ." ""' .

Figure 3.7 Scanning electron micrograph illustrating the effects of applied mechanical
pressure on trapping gaseous reaction products in the form of gas porosity (dark spots).
Sample of C10B composition, hot pressed at 13500C for 2 h in vacuum (<0.13 Pa) under
applied pressure of 47 MPa followed by 1 h at 17000C in argon atmosphere (-50 kPa).

Powders were blended prior to hot pressing in a tumbler ball mill for 2 h with the

idea of breaking up agglomerates and producing a more uniform powder mixture prior to

hot pressing. The effects of this process are shown in Figure 3.8, where the number of

agglomerates equal to or larger than 200 pm is reduced by more than 60 percent after ball

Table 3.4 Powder size information for starting powders used in this study. All powder
statistics were measured on a Coulter LS230 particle size analyzer without surfactant

Powder Reported Mean Mode Standard 50% Vol. 90% Vol.
Particle Particle (gim) Deviation Less than Less than
Size (im) Size (pm) (im) (Im) (im)
C <50 45.5 19.8 64.2 18.4 138.7
Si 45-150 115.1 140.1 54.1 120.2 183.9
Si <45 40.8 80.1 36.1 16.8 91.4
Mo 45-100 69.5 66.4 10.1 68.4 82.4
Mo 3-7 19.1 14.9 21.8 13.9 30.5
MoC <45 71.9 127.6 68.5 77.3 164.1
MoSi, 5.4 13.2 10.3 10.5 10 28.4
MoSi, <45 6.3 5.4 4.9 5.1 12.7

milling (Figure 3.6 (b)); a slight reduction in mean particle size (107.3 to 94 imn) is seen as

well. Discrepancies were noted in several cases between the particle sizes quoted by the

raw materials vendors and those measured through particle size analysis; for example the

Mo, Mo2C, and MoSi, (5.4 tm reported) in Table 3.4 are all larger than expected,

indicating a propensity for agglomeration.

Several representative microstructures of samples processed through blending of

powders and hot pressing are shown in Figures 3.9 and 3.10. From these micrographs, it

is clear that significant variation in phase type and volume fraction is possible depending on

starting materials and composition. Phase identification for the samples shown was

accomplished through a combination of x-ray diffraction and WDS. Comparing the phases

present in these samples to those expected from their starting composition gives the first

indication of some combination of a discrepancy with the reported size of the single-phase

field for Mo5SiC (Fig. 3.4), compositional shifts occurring during processing, and non-

equilibrium microstructures. Compositions (CN) 11 (Fig. 3.9 (a)) and C15 (Fig. 3.10 (b))

have carbon contents that should position them on the bottom and top, respectively, of the

single-phase field and would be expected to contain at most a few volume percent of

secondary phases, which is not the case. The nearest single-phase sample (Fig. 3.10 (a))

is that processed at CN C10B, having the nominal composition 55.3Mo-34.7Si-10C.

Issues concerning the actual compositional range of MoSSi3C will be discussed in more

detail below. Phase stability issues aside, samples processed through powder mixing and

hot pressing do exhibit the expected refinement in structure (Figure 3.11), control of

microstructural scale, and reduction of macro-cracking compared with the arc-melted


Transmission electron microscopy (TEM) of a thin foil of CN C10B (Figure 3.12)

reveals further information about phases present and their crystal structures. As seen in the

bright-field image (BF) (Fig. 3.12 (a)), a two-phase microstructure of thinner (light

contrast) and thicker (dark contrast) regions is present in this region. A selected-area

Particle Diameter (m)


Different al Volume
10 3-




1 -

oil. __i I ..l __

0.04 0.1 0.4 1 2 4 6 10 20 40 100 200 400 1000
Particle Diameter (pm)

Figure 3.8 (a) Particle size distribution for the as-received Mo, Si, and C powders. (b)
Particle size distribution for same powders after 2 h ball milling showing a decreased
number of agglomerates 200 and slight reduction in Volumean particle size.



0,04 0.1 0.4 1 2 4 6 10 20 40 100 200 400 1000
Particle Diameter (pm)
Figure 3.8 (a) Particle size distribution for the as-received Mo, Si, and C powders. (b)
Particle size distribution for same powders after 2 h ball milling showing a decreased
number of agglomerates > 200 prm and slight reduction in mean particle size.

Figure 3.9 (a) Scanning electron microscope (SEM) image of hot pressed sample of CN
11. Phase displaying light contrast is MoSSi3C, while darker phase is MoSSi3, and very
fine, lighter phase is Mo2C. (b) SEM image of CN 15, produced from a blend of MoC,
Mo, and MoSi2. Matrix is MoSi3C, light gray platelets are Mo2C, and black dots are likely
a mixture of porosity and SiO, from the MoSi, powder precursor.

& *'" ^tr "

(a) .


Figure 3.10 (a) Micrograph from SEM of composition C10B. Matrix is Mo5Si3C with
fine SiC encircling gas porosity (linear features). (b) Micrograph of CN C15. Matrix is
MosSi3C, dark regions are carbon-rich, and light particles are MoC.

diffraction (SAD) pattern from a "thicker" region (Fig. 3.12 (b)) is consistent with a

hexagonal crystal structure oriented along the [0001] beam direction; indexing of this

pattern along with a B = [1120] SAD (not shown) yield lattice parameters of a = 0.733 nm

and c =0.509 nm. These calculated parameters are close to those reported by JCPDS PDF

43-1199 for Mo,4 Si, C06 of a=0.7292 nm and c=0.5043 nm as well those of a = 0.7286

nm and c = 0.5046 nm from a previous XRD and neutron diffraction study [137].
For the brighter (i.e., thinner), faulted regions of Figure 3.12 (a), the SAD patterns

(e.g., Figure 3.12(c)) can be indexed as cubic with a lattice parameter of a -0.437 nm,

which is close to that of 0.436 nm reported for cubic P-SiC [58]. Furthermore, the faulted

appearance of the SiC platelets in the BF image results in streaking and extra diffraction

spots (twin spots) as seen in the [011] SAD pattern. Twinning along 111) in p-SiC is not

uncommon and has been reported for both vapor-deposited material [144] and carbides

resulting from melt processing [78]. The relatively large fraction of SiC observed during

TEM analysis is somewhat surprising considering the nearly single-phase nature of the

bulk sample (Fig. 3.10 (a)), although preferential thinning near the aforementioned SiC-

pore regions is a plausible explanation.

Figure 3.11 Images of powder processed samples showing grain contrast under
polarized light. Light optical micrograph on left is of composition C10B; average grain
diameter calculated through line intercept methods is 16.8 2.1 pm. Dark regions are
porosity. Optical micrograph on right is of C10B with an average grain diameter of 5.6
0.5 lim.

Mechanical allowing + hot pressing. Mechanical alloying (MA) of powders

followed by hot pressing was explored as a variant of powder blending methods.

Attractive features of MA include refined particle sizes (< 1 um), lower processing

temperatures/reduced times, and the ability to synthesize phases during the high energy MA

process [49]. Starting compositions of MA samples included CN 7 and C1OB. As

anticipated, starting powders of MoSi3 (<45 um), Si (<45 pm) and C (< 50 pim) did show

more than a 60 % reduction in mean particle size after a 15 h alloying cycle (Figure 3.13).



Figure 3.12 (a) Bright field (BF) image of Mo Si3C + SiC region. The dark areas are the
ternary phase, while the light, faulted areas are p-SiC. (b) Selected area diffraction pattern
(SADP) taken from dark phase, showing the hexagonal D8, structure of Mo5Si3C. (c)
SADP (B=[011]) of light phase where streaking lines indicates twinning along { 11) in
cubic P-SiC. The m subscripts indicate matrix reflections, while the t subscripts denote
twin reflections.

During some MA runs, significant sticking and caking of powders to the steel

milling media and vial was observed leading to large agglomerates (-500 Jim) and reduced

powder yield. To counteract caking, -0.2 g of stearic acid flakes were added to the powder

charge, the milling media, and milling vial surfaces; the effectiveness of stearic acid is

shown in Figure 3.14, which shows the particle size distributions for identical powders

and MA runs with and without stearic acid additions.

Microstructures of consolidated MA specimens (Figure 3.15) are basically fully

dense and exhibit a matrix grain size typically less than 10 tm, with many grains in the 1

upm range. This fine scale complicated compositional analysis using EPMA, as many of the
regions of interest were on the order of the spatial resolution of WDS (~ 1 pm), creating

problems with interference and overlap between phases.

Elevated oxygen contents due to the increased surface area of the fine MA powders

[9, 56] and contamination from milling media [39] are both concerns when using MA

processes. In fact, samples produced through hot pressing of MA powders did show

increased oxygen content in the form of ~ 2-6 Vol.% amorphous silica particles (Figure

3.16), a feature not present in blended powder samples of the same composition.

Conflicting results in the literature regarding oxygen pickup during MA of silicide materials

[49] indicate the importance of MA process variables. For the materials produced in this

study, the amount of silica present in the consolidated bodies probably would have been

reduced by process modifications such as adding excess carbon to more fully reduce the

silica present and minimizing powder exposure to air after milling by handling the powders

in an inert atmosphere (e.g., glove bag). The reactions between silica and carbon occurring

during processing will be outlined in more detail below.

One feature of the silica present in these MoSi3C-based materials is that while there

are concentrations of the amorphous phase along grain boundaries, there are also many

instances of silica appearing within grains (Figure 3.15) in contrast to much of the data on

powder-processed MoSi2 where SiO2 is typically reported as present primarily at grain

boundaries and triple points [58, 87]. The silica that is present along grain boundaries and

at triple points in the Mo5Si3C-containing materials may act to stabilize the refined structure

by limiting grain growth due to a Zener drag mechanism [61].





B 2.E
E -






0 2-

Particle Diameter (pm)

Particle Diameter (pm)

Figure 3.13 Comparison of particle size distribution before (a) and after (b) 15 h of
mechanical alloying of a 25 g powder charge at a milling media:charge ratio of 2:1.

Differential Volume

I4 tI h Iearic j id

004 01 0.4 1 2 4 6 10 20 40 100 200 1000
Particle Diameter (pm)

Figure 3.14 Comparison plot of particle size distribution of identical starting powders
after 15 h MA run with and without stearic acid additions. Note the bimodal distribution
for the powders milled without stearic acid and the large number of particles with diameters
greater than 20 tm.

Figure 3.15 Micrograph (BSE image) of sample having composition C10B produced
from mechanically alloyed powders. Matrix phase showing grain contrast is MosSi3C,
while minor phase with bright contrast is MoC, and sub-micron black particles are either
silica or porosity.

Figure 3.16 Bright-field image of MA sample (CN C10B) showing presence of
amorphous silica particle near triple point (A).

Though less obvious than the presence of silica, slight contamination from the steel

milling balls and vial in the form of iron was detected from WDS analysis of some

mechanically alloyed specimens, typically on the order of 2500 ppm in the MoSSi3C phase.

It has been reported that iron partitions readily to the MosSiC phase [58] in MA Mo-Si-C

materials, although the Fe content reported in the ternary phase is significantly higher than

that observed in the present study. It is unclear whether the minor iron contamination

might effect phase formation or mechanical behavior of Mo5Si3C-containing materials as Fe

likely substitutes on Mo-lattice sites.

Carbon-silica reaction issues. With most powder processing methods, the potential

exists for the retention of residual phases within consolidated bodies from either intentional

(e.g., sintering aids) or unintentional (e.g., oxides on powder surfaces, contamination)

sources. As previously discussed, the presence of oxide scales on silicon-containing

starting materials leads to the presence of discrete amorphous silica within powder-

processed silicide materials. With samples that also contain carbon as either a contaminant

or, as in this study, alloying element, reactions inevitably occur during processing that lead

to silica reduction with concomitant loss of carbon and production of solid and/or gaseous


In addition to carbon loss, the silica present on the starting powders leads to an

over-estimation of the amount of Si-containing constituent proportional to the percentage of

oxygen in the starting siliceous powder. The net result of these two factors is a depletion in

Si and C with respect to the nominal composition of a Mo-Si-C alloy, pushing the final

composition toward the Mo-rich corner of the Mo-Si-C isotherm. Microprobe analysis and

phase distributions of powder-processed samples typically do indicate a slight increase in

the Mo:Si ratio compared to starting compositions, consistent with compositional shifts

toward the Mo-rich comer. Because the questionable accuracy of light element analysis

using WDS, especially for minor amounts [141], light element analysis using

combustion/IR techniques was used to quantify the bulk carbon and oxygen content of

several hot pressed specimens. Calculations based on the results of light element analysis

show an average depletion of carbon of 1.4 At. % during processing and an average

oxygen content of 1544 ppm for the consolidated samples; without using glove box

handling or powder etching, this oxygen content is still lower than those typically reported

for consolidated silicide materials (Table 2.3); this testifies to the effectiveness of carbon as

a catalyst for silica removal. On average, approximately 16 % of the mass of carbon

initially in the powder compacts is lost due to reactions with the siliceous components.
Calculations based on a single mechanism (Equation 3.3) assuming complete

reaction of an amount of silica in the starting powders equivalent to 2 Wt. % oxygen (from

Table 2.3), give an estimated carbon loss in the range of 0.5 to 4 At. %. The magnitude of

the estimated loss varies depending on choice of starting materials. It was assumed that

different forms of siliceous materials would have equivalent oxygen contents, although this

is probably not the case. This explains the large shift predicted (4 At. % C) for MosSi, + C

samples since the weight fraction of the Si-containing compound (and thus oxygen) is

much higher than for samples containing elemental silicon, which probably contains more

silica than MosSi3 for equivalent surface areas due to the higher activity of Si. In these

calculations, it is also assumed that all SiO2 in the starting powders is in intimate contact

with carbon, and that no additional carbon enters the powder compacts by diffusion from

the graphite paper liners surrounding the compacts during hot pressing. In spite of these

uncertainties, the calculated depletion range does encompass the experimentally-observed

1.4 At. % loss, lending some credibility to the calculations.

Evidence of a deoxidation reaction of the type

SiO2 +3C= SiC+2CO(g) (3.3)

is seen in Figure 3.17, where a network of P-SiC platelets surrounds a pore, presumably

the result of gaseous evolution, in an otherwise single phase material (C10B). These

pore/SiC regions tend to be linear in appearance (in cross section) with their long axis

perpendicular to the hot pressing direction, indicative of a morphology change occurring

during the densification step. Unlike the majority of gas and sintering porosity existing

prior to densification, these SiC/pore regions do no fully close out during densification,

likely because of the SiC boundary layer acting as a mechanical and diffusion barrier to

pressure-assisted close out of the pore.

Rather than forming these SiC/pore structures, it might be more desirable to

produce only gaseous reaction products of the silica reduction during the initial stages of

hot pressing and then minimize the subsequent gas porosity through pressure-assisted

sintering. However, the SiC/pore defects as in Fig. 3.14 appeared in a number of hot

pressed samples of different compositions, indicating that the favored silica reduction

mechanisms are those that result in the formation of SiC. Recall that in addition to

Equation 3.3, another possible silica reduction reaction is,

SiO2 + C= SiO(g)+ CO(g) (3.4)

indicating the potential for the removal of SiO2 with only gaseous byproducts. A plausible

explanation for SiC-forming reactions (e.g., Eq. 3.3) predominating in hot-pressed

samples is shown in Figure 3.18, in which the free energy changes associated with three

possible SiO2-carbon reactions are plotted as a function of temperature. Over the range of

temperatures used in processing, the free energy change of Eq. 3.3 is the smallest, meaning

that it is the favored reaction from a thermodynamic standpoint; above ~15100C, the AG for

Equation 3.3 is negative indicating a driving force for the reduction of silica and the

production of SiC and carbon monoxide gas. The AG values for the other two possible

reactions remain positive and thus it is doubtful that their reaction products would be

favored under equilibrium conditions. The calculations assume the activity of carbon is

constant throughout the reaction which is not actually the case as carbon is consumed. In

fact it has been reported that decreasing carbon activity increases the amount of SiO(g)

produced during the reduction of SiO2; at 1507 OC, all carbon is consumed to form SiC and

only at temperatures in excess of 17000C does the increased Psio alter the products of the

reduction reaction from SiC to free silicon [145]. It has been proposed that lower partial

pressures of oxygen and a reducing atmosphere, as the case with vacuum hot pressing in

graphite dies, shift these free energy curves to lower temperatures, implying the reduction

of silica during hot pressing probably begins to occur below 1510 OC [146].

In several cases (e.g., CN 7), intentional additions of excess carbon and silicon

were added in hopes of correcting for the inevitable compositional shifts caused by silica-

carbon reactions. However, it proved to be difficult to gauge the effectiveness of these

attempts for several reasons. First of all, since accurate information on starting oxygen

content of silicon or silicide powders was lacking, assumptions were made in order to

estimate the additions needed. Accurate quantitative measurement of carbon was not

readily available at our facilities, so systematic measurements of carbon losses in samples

with intentional additions was not possible. Finally, uncertainties about the size of the

MoSi3C single-phase field made it difficult to determine what phases and phase amounts
should be present for given nominal and target compositions. In addition to adding excess

carbon, another route explored to possibly achieve tighter control over stoichiometry was

the use of larger Si powders (-115 gtm) because of their decreased surface area for

oxidation and assumed lower levels of oxygen. However, the impact of larger silicon

powders on composition shifting were inconclusive and the consolidated bodies tended to

be less dense that those produced from finer powders.

3.3.2 Compositional Range of Mo!SiC

In spite of producing samples at a number of compositions within the reported

MosSi3C single-phase region using different processing routes, nearly all of these materials
contained at least five volume percent of other phases. Though complicated somewhat by

issues such as homogenization difficulties with arc melted specimens, contamination during

MA processes, and compositional shifts during hot pressing, it is proposed that the most

significant obstacle to producing single phase MoSi3C is a much narrower compositional

range for this phase than originally thought. To more thoroughly address this issue, it was


Figure 3.17 Image from SEM of pore surrounded by P-SiC in hot pressed sample. This
feature is thought to be the result of reactions between carbon in the sample and SiO2 on the
starting powders.

-4-s02 + C = SO(g) CO(g
-.. 12 + 3C = SIC 2CO(g)
S 02 2C = SIC+ C02(g)

100 ---

1350 1400 1450 1500 1800 1650 1700


Figure 3.18 Plot of equilibrium free energy versus temperature for three possible
reactions between carbon and silica. Plot generated from HSC Chemistry software
package and the thermodynamic database contained therein. The trend with temperature
predicts that only reaction (2) has sufficient driving force to spontaneously generate its
products at temperatures below 1700C.

decided to systematically reassess the range of compositional variation for the MosSi3C

phase by producing samples through blending of powders and HP, focusing in one set of

samples on the range of Mo:Si ratios (width) and in the other on carbon content (height) of

the single-phase field as shown in Fig. 3.4.
A summary of phase identification and composition information from XRD,

EPMA, and light element analysis for several samples representative of those used to

examine the compositional range of Mo5Si3C is presented in Table 3.5 and Table 3.6. Width of Mo SijC single-phase field
To investigate the range of Mo:Si ratios over which MosSi3C existed as a single

phase, starting powders of Mo2C and Si or MosSi3, Si, and C were blended and hot

pressed using the standard two step pressing method. As shown in Figure 3.19, samples

having average MoSi3C Mo:Si ratios (Table 3.5) of 1.60 and 1.69 as measured by WDS

were not solely MoSSi3C as would be expected from Figure 3.4. These samples contained

roughly equal fractions of secondary phases, placing them in the MosSi3C + SiC + MoSi2

(CN 13) and MosSi3C + Mo2C (CN 10) phase fields on opposing sides of the MoSi3C

single phase region. Composition C8B has an average Mo:Si ratio in the MoSi3C matrix

of 1.65 and is nearer to being single phase than either CN 10 or 13, although it still

contains 5.7 Vol.% of the MoSi3 phase.

In the work of Nowotny et al. [18], it was reported that going from Mo-lean (Mo:Si

-1.2) to Mo-rich (Mo:Si -2.0) compositions across the MosSi3C field, the axial ratio (c/a)

of MoSSi3C increased from 0.687 to 0.695. A change in lattice parameters across a single-

phase field is not unusual [147], although in the case of MosSi3C other researchers [12, 19,

120] have observed a much smaller compositional range over which this lattice parameter

variation occurs. In the present work, the c/a ratio did increase with increasing Mo content

from 0.689 to 0.694, but over a Mo:Si ratio range of approximately 0.1, which is
significantly less than the -0.8 given by Nowotny et al [18]. For samples in this study

Table 3.5 Summary of data from XRD and WDS pertinent to the examination of the
compositional variation of Mo5Si3C.

CN XRD Phase c/a Mo Si C Mo:Si
ID Ratio (At. %) (At. %) (At. %) Ratio
Major Mo5Si'C Matrix Matrix Matrix Matrix
13 MosSi3C 0.689 60.2 1.6 37.6 1.5 2.1 +2.3 1.60
C8B Mo-sSiC, 0.687 61.2 1.7 37.1 0.9 1.7 2.6 1.65
Mo Si, +0.01
10 MosSi3C, 0.694 59.1 1.3 34.9 1 6.0 1.7 1.69
Mo,C _0.06
C75 MosSiC, 0.694 60.2 1.6 36.7 1.0 3.0 0.98 1.64
MoSi, 0.09
C10 MoSiC, 0.695 60.7 1.1 35.9 0.72 3.5 1.2 1.69
Mo,C 0.05
C125 MoS3C, 0.695 58.0 1.6 37.2 1.4 4.8 0.95 1.56
Mo,C 0.1

Table 3.6 Light element analysis and phase fractions from image analysis for samples in
compositional range study.

CN C (Wt. %) Equivalent Oxygen (ppm) Second Phase Vol. %
Bulk Analysis At. % C Bulk Analysis
13 6.1 SiC, trace MoSi,
C8B 5.7 MoSi,
10 1.77 0.04 9.82 0.2 558 25 7.7 MoC, trace C
C75 1.12 0.01 6.22 0.05 2450 1670 22.7MoSi,, 1.8 Mo,C
C10 1.49 +0.02 8.25 +0.1 897 163 3 Mo,C
C125 2.02 0.12 11.24 0.6 2274 1551 2.7 MoC, 1.8 C

with higher Mo contents (Mo/Si > 1.7), the MoSiC c/a ratio approached a constant value

of 0.695, indicating that the composition of the ternary phase had reached a terminal value

as overall compositions pushed further into two- or three- phase fields. Height of Mo Si'C single-phase field

Samples for examining the variation of Mo5Si3C carbon content were produced

from blended and hot pressed powders of MosSi3 and carbon. Starting with a Mo:Si ratio

of 1.67 dictated by the MosSi3 stoichiometry (Figure 3.20), samples having final bulk

carbon contents of 6.2 At. % (CN C75, Figure 3.20 (c)) and 11.2 At. % (CN C125,

Figure 3.20 (a)) contain MoSi3C + MoSSi3 (22.7 Vol. %) + Mo2C (1.8 Vol. %) and

Mo5Si3C + Mo2C (2.7 Vol. %) + C (1.8 Vol. %), respectively. Note the aforementioned
shifts in composition toward the Mo-rich corer of the Mo-Si-C isotherm after hot

pressing. Even with the variations in stoichiometry presumably due to C-SiO2 reactions,

samples such as these having Mo:Si ratios near 1.67 are predicted to remain single-phase

MoSi3C for carbon contents from 6 to 15 At. % (Fig. 3.4) which is obviously not the
case. One area in which there is agreement with the original study [18] and more recent

work [120] is the minimal variation in the Mo5Si3C c/a ratio with carbon content, from

0.694 to 0.695 at overall carbon contents of 5 and 15 At. % C, respectively. Heat treatment

A possible explanation [120] for the much smaller homogeneity range for Mo5Si3C

reported in the current work and others [12, 19, 120] is that these studies used processing

and/or equilibration temperatures other than 1600C, the temperature for the original

isotherm in Figure 2.12. It is plausible that shifts in phase boundaries could occur at some

intermediate temperature (i.e., T < 16000C). In addition, samples produced through hot

pressing may not have been fully equilibrated. To investigate these issues, several samples

were heat treated at 16000C and the resultant microstructures characterized by SEM and

WDS. As seen in Fig. 3.21 (b), a sample initially containing Mo2C and C in a Mo5Si3C


MoSi3C + MoC + C

Figure 3.19. Schematic representation of samples used for the width estimation of the
MoSiC single-phase field. Both CN 13 (a) and 10 (b) have starting (x) and final (*)
Mo:Si ratios within the range for Mo SiC (T) predicted by Nowotny et al. [18], while the
phases present in the microstructures indicate that these samples are actually in the adjacent
three-phase fields T+ SiC + MoSi2 and T + Mo2C + C. The phase displaying dark contrast
in (a) is SiC (MoSi, is not clearly discernable in this micrograph), while lighter, raised
phase in (b) is Mo2C and very fine black particles are free carbon.

Mo5Si3C + Mo2C + C

MoSiC + MoC

MosSi3C + MosSi, + Mo2C

Figure 3.20 Schematic representation of samples used to examine the height of the
MoSiC single-phase field. Samples are compositions C125 (a), C10 (b) and C75 (c),
hot-pressed blends of MosSi3 + C, SEI. All three samples exhibit multiphase
microstructures at final (*) carbon amounts that fall within the Mo5Si3C field according to

M 1i/i/f
r_-_- 7 N"

matrix shows little difference in phase type, distribution or scale of microstructure after 12h

at 16000C.

Microprobe analysis is inconclusive for this sample, with the heat-treated specimen

showing reduced carbon content in the Mo5Si3C matrix as well as in the remaining carbide

phase compared with the as-hot-pressed sample, but not to a significant degree. While this

may indicate that the as hot pressed sample may not have been fully reacted, and is

consistent with the sluggish decomposition kinetics reported for Mo-carbide phases [137],

it seems unlikely that extended heat treatments would produce a single-phase MosSi3C

microstructure (i.e., the as-hot-pressed structure is close to equilibrium). The relatively

large scale of the carbide phase relative to the matrix is probably a result of retaining

undissolved Mo2C from the starting Mo2C+Mo+Si powder mixtures.

In their work on diffusion paths and reaction mechanisms in in situ MoSi2-SiC

composites, Henager et al. [12] describe the precipitation of fine SiC into MoSi3C that

becomes supersaturated with carbon while van Loo et al. [19] propose that carbon rapidly

diffuses from MosSi3C in contact with Mo forming Mo2C and MoSi,. These results point

to the possibility that the solubility of carbon in Mo5Si3C may change with temperature and

that phases may be precipitated from supersaturated Mo5Si3C upon cooling due to the

reduced solubility range at lower temperatures (e.g., Figure 2.13). However, the majority

of the microstructures produced in the current work are inconsistent with solid-state

precipitation of phases from Mo5Si3C due to decreasing solubility in the ternary phase with

decreasing temperature. The limited phase redistribution during heat treatment at 1600C

and the general observation that the secondary phases in these materials are typically

coarser (2 10 lim) and less uniformly distributed than what would be expected from second

phase formation via precipitation [148], indicate that the majority of the phase formation

seen in MosSi3C-based materials occurs at the hot pressing temperatures (1350-17000C)

and not during cooling.


(a) (b)
Figure 3.21 (a) Secondary electron image of HP sample of CN 10, showing a MosSi3C
matrix, with Mo2C (light phase) and a small amount of C (dark phase). (b) Secondary
electron image of same sample after heat treatment at 16000C for 12 h, showing Mo5Si3C
matrix, with decreased Mo2C (light phase) and increased amount of free carbon (dark

It is possible that the MosSi3C phase boundaries may be reduced somewhat at lower

temperatures. However, based on the numerous samples produced in this study at

compositions encompassing those given for single-phase MosSi3C which resulted in

multiphase structures, the size of the field reported at 16000C (Figure 3.4) must be a

significant overestimation. It is recognized that heat treatment and equilibration of

MoSSi3C-based samples with varying initial second phase constituents at several different
temperatures would provide a better understanding of the issues surrounding the variation

of the Mo5Si3C single-phase field size with temperature. Conclusions

One possible explanation for discrepancies between the large compositional range

of MoSi3C reported by Nowotny et al.[18] and more recent work, including this study, is

the reliance on x-ray diffraction techniques for phase identification and chemical analysis in

the original work. For example, in the current study a sample comprised of three-phases

(Figure 3.19 (a)) yielded an x-ray diffraction pattern which was indexed completely to

single-phase MosSi3C (JCPDS PDF 43-1199) with no detectable SiC or MoSi2 (Figure

3.22). This demonstrates one of the primary drawbacks of relying solely on XRD for

determination of phase diagram boundaries in cases where phase amounts and/or low

intensity x-ray patterns from second phases may be below the detection limit of XRD


These results indicate the importance of verifying existing thermodynamic data

when attempting to produce in situ composites in multi-component systems or trying to

synthesize single-phase compounds, an issue addressed in several ternary silicide systems

[14, 136]. To confirm the phase boundary locations in a given multi-component system,

careful microstructural examination and analysis is needed to complement phase

identification from XRD. In fact, a microstructure purported to be single-phase MosSi3C is

presented by Nowotny et al. [18] (Figure 2.15), although the high magnification chosen,

sintering porosity, and poor image quality make it difficult to say much about the single-

phase nature of this particular composition.

Based on the height and width data of the MoSi3C field obtained in this study and

other work in the Mo-Si-C system [12, 19, 120], a revised schematic version of the

Nowotny et al. Mo-Si-C isotherm in the near-Mo5Si3C region is presented in Figure 3.23.

This modified section shows the much smaller MoSi3C single-phase region, as well as

proper designation for MoSi,, and Mo2C (not MoC) present at 16000C in agreement with

the accepted Mo-C binary diagram [134]. It is postulated that the actual compositional

range of the ternary phase may be even narrower than presented in this revised diagram,

perhaps approaching a line compound.

3.4 Summary and Conclusions

Despite complications caused by uncertainties in the compositional range of

MoSi3C and challenges related to process optimization, nearly single-phase samples of

MoSi3C (Fig. 3.10 (a)) were produced by hot pressing blended powders at a nominal
starting composition of 55.3Mo-34.7Si-10C. Three main processing routes were

MosSi3C A


0.25 A

0.16 A

0.09 AA



20.0 40.0 60.0 80.0 100.0
Degrees (20)
Figure 3.22 Pattern from XRD of sample CN 13 indexed as single-phase MosSi3C based
on JCPDS PDF 43-1199. Comparison with micrograph (Figure 3.19 (a)) showing definite
multiphase microstructure indicates the hazards of relying solely on XRD for phase

examined in this study, arc melting, powder blending followed by hot pressing, and hot

pressing of mechanically alloyed powders.

While extended heat treatments might serve to homogenize arc melted samples, arc

melting was deemed unsuitable because of difficulties incorporating carbon into the melt,

concerns about Si loss, and thermally induced cracking. Arc melting of brittle materials

susceptible to thermal shock is expected to be an ineffective means of producing samples

for purposes other than phase equilibria and alloy screening studies.

Through hot consolidation of blended or MA powders, it was possible to produce

MosSi3C-containing specimens with varying compositions, phase distribution, and
microstructural scale. Powder processing routes also alleviate the macrocracking and

massive inhomogeneity observed in arc-melted specimens; however compositional control

and contamination issues pose other challenges. Shifts in composition during the course

of processing toward the Mo-rich corer of the Mo-Si-C isotherm were observed and are

likely the result of reactions between silica on the starting Si-containing powders and

carbon. Estimation of these shifts and attempts to correct for them were complicated by the

uncertainties in oxygen content of the starting powders and the variation of oxygen content

with type of siliceous compound.

A further result of reactions between carbon and silica was the formation of SiC-

pore features in many of the hot pressed samples; the formation of silicon carbide and

evolution of CO (g) was shown to be the most thermodynamically favored of potential

silica reduction reactions. The presence of these reaction products points out one of the

realities of powder processing silicide materials-other than circumventing powder

oxidation through complex pre-processing such as powder etching or glove bag processing

[9], one has to accept either the presence of amorphous silica particles in the consolidated

samples or the byproducts of silica reduction by intentional deoxidant additions [64].

CN 10B CN C125


Figure 3.23 Representation of revised portion of Mo-Si-C 16000C isotherm near the
Mo5Si3C single-phase field. The significantly reduced MosSiC compositional range is
shown along with the proper designation for Mo5Si, and the removal of phase fields
involving MoC, which is not stable at 16000C. The stoichiometry best approximating the T
region is Mo4 Si3Co.7. Scanning electron micrographs surrounding the section show
typical microstructures of compositions approximately as indicated on the diagram. For
micrographs (13), (10B), (10), and (8B) marker = 100 .m; for (C125), (C10), and (C75)
marker = 10 gm.


Phase identification through XRD and EPMA and quantitative compositional

information from WDS and light element analysis of samples produced over a range of

starting compositions indicates that the compositional range for Mo5Si3C reported [18] on

the accepted 16000C Mo-Si-C isotherm is most likely a significant overestimation. The

actual compositional range of the ternary compound probably encompasses only a few

percent carbon at basically a fixed Mo:Si ratio. This obviously poses significant challenges

to the production of single-phase microstructures especially in light of shifts in carbon and

silicon content during powder processing. Given the ongoing interest in Mo-Si-X systems

[26, 115, 116, 124, 149-154] for which the bulk of thermodynamic data and phase

diagram information was developed over 40 years ago, it seems wise to verify relevant

compositional ranges and phase equilibria before embarking on costly processing

optimization or property determination studies.


4.1 Introduction

One of the primary reasons for studying ternary silicide compounds such as

Mo5Si3C is to investigate what effects, if any, their higher-symmetry crystal structures have
on mechanical behavior, and how their mechanical properties compare with lower-

symmetry binary compounds such as MoSi2 and MosSi3. Ideally, comparisons would be

made between high purity single crystal materials, allowing for fundamental examination of

issues such as elastic constants, operative deformation mechanisms, dislocation

generation/mobility, and slip character. Unfortunately, production of homogenous single

crystal silicides using float zone or Czochralski techniques is not trivial due to the

aforementioned high vapor pressure of silicon [87]; losses of silicon and composition

gradients would be especially problematic when trying to synthesize phases such as

Mo5Si3C having a narrow range of single-phase stability.

Polycrystalline materials are easier to synthesize, but greatly complicate assessment

of intrinsic mechanical behavior and comparisons with other materials, particularly at

elevated temperatures where grain size/grain boundary effects can dominate deformation.

Because of this, it is desirable to be able to prepare samples at several different grain sizes.

As detailed in the previous section, production of dense single-phase MosSi3C samples was

greatly complicated by discrepancies in the composition range of the phase as well as

processing difficulties; nonetheless, dense, nearly-single phase specimens were produced

at coarse (-20 pm) and fine (-5 tjm) grain sizes. For the purposes of this study, it was

decided that an examination of the mechanical behavior of these optimized samples at low

and high temperatures would serve to establish some initial property values and possible

deformation mechanisms for the ternary phase. In addition, comparing the mechanical

properties of MosSi3C with other silicides would shed some light on the effectiveness of

alloying to enhance deformability through increased crystal symmetry.

4.2 Experimental Procedure

4.2.1 Sample Characterization and Analysis

Samples for mechanical testing were produced from either hot-pressed blended

powders or hot-pressed mechanically alloyed powders; as-hot pressed materials were disks

either 16 mm or 32 mm in diameter, with the larger size preferred since it yields more

mechanical testing samples per hot press run and is also large enough in diameter to allow

for machining of standard 25 mm long bend test bars. Material stock used for mechanical

property evaluations was characterized before testing so that the subsequent mechanical

behavior could be correlated to sample microstructure and processing route. For all

specimens characterized and tested, a minimum of 1 mm was ground from the as-hot

pressed sample surfaces to completely remove the residual graphite foil. Unless otherwise

noted, the experimental procedure for each characterization method was identical to that

described in the previous chapter. For a first check of phases present, x-ray diffraction

data was collected from a solid slice of the sample. A section of the testing stock was

mounted in epoxy, ground flat, and polished to a 1 pm finish for microstructural analysis

on the light optical and scanning electron microscopes, as well as phase identification and

quantitative chemical analysis using WDS. From optical and scanning electron

micrographs, the average grain size of specimens was determined using standard line

intercept methods; image analysis was also used to determine Vol.% porosity and

secondary phases. The bulk density of as-hot pressed samples was also measured using

the previously described techniques (section

4.2.2 Hardness Testing

Hardness measurements were done at room temperature using a bench mounted

Buehler Micromet 2 Microhardness Tester with a Vickers pyramidal diamond indenter at

applied loads up to 1 kgf for a 15 sec dwell time. A calibrated eyepiece equipped with a

micrometer was used to measure the lengths of the two perpendicular diagonals. A

minimum of 15 indents were used to obtain average hardness values for a given sample; to

avoid overlapping stress fields, care was taken to position indents such that their minimum

separation was a minimum of two indents including any crack zones. To determine

hardness values (HV) from indent lengths, the following expression was used,
2P(0.9272) (4.1)
HV= (4.1)
where D is the average length of the two diagonals, and P is the applied load in kgf.

4.2.3 Indentation Fracture Toughness Measurements

Fracture toughness of the MoSi3C compound was measured using indentation

microfracture techniques; this method of fracture toughness testing was chosen for its

simplicity and amenability to measurements on small samples. Five hardness indents were

made at each of three different loads, and from each indent, hardness values were

determined from Equation 4.1. Also measured for each indent was the length from the

center of the indent to the end of the linear crack emanating from the comers of the indent,

the induced crack length (c,) [155]. Using the following expression,
K, =A(EIH)"PIc'/12 (4.2)

where A and n are constants with values [156] of 0.016 and 0.5, E the elastic modulus of

the material, and H the Vickers hardness. Because of limited data in the literature,

toughness values were calculated with E for both MosSi3 (323 GPa [157]) and MoSSi3C

(285 GPa [120]).

Values of Kc were derived from the slope of a linear fit to a P versus co"3 plot.

For each indent, values for co were only recorded if they represented a well-defined crack

pattern, without excessive chipping or crack branching; as a measure of validity of the

crack system, satisfaction of the criterion co. 2a [156], where 2a is the average diagonal

length of the indent (D in Equation 4.1), was verified for each indent.

4.2.4 Elevated Temperature Compression Testing

To examine the flow behavior of MoSi3C and its dependence on temperature,

strain rate, and grain size, compression testing was done over a temperature range of

1000C to 13000C (approximately 0.5 to 0.66 T,). Cylindrical (3 mm diameter x 5 mm

height) compression specimens were cut as -3.3 mm diameter right cylinders on a Brother

HS 100 wire electro-discharge machine (EDM) from hot pressed disks either 31 mm in

diameter by 7 mm thick or 16 mm diameter by 8 mm thick. Compression cylinders were

machined such that the long axis (5 mm) was parallel to the pressing direction. The extra

0.3 mm tolerance in test specimen cylinder diameter was to compensate for subsequent

grinding off of the EDM re-cast layer.

After machining, specimen surfaces were ground using SiC papers ranging from

240 to 800 grit; to ensure flat, parallel contact surfaces, a Starrett V-Block sample grinding

holder was used to sequentially grind the ends of the specimens down in -0.05 mm steps.

The sides of the samples were ground down by rotating them manually on the grinding

media. After grinding, a rotary diamond polishing wheel and 6 tni diamond slurry were

used for final surface preparation. To estimate the degree to which the top and bottom

contact surfaces were parallel, measurements of the cylinder height were made at three

different points along the surfaces, and the typical height deviation was within 10 pnm.

An MTS Model 810 Materials Test System under servo-hydraulic control was used

for compression testing. This system has a stainless steel jacketed, water cooled furnace

chamber and OXYGON furnace. Using resistance-heated molybdenum elements under

purified argon, the furnace is capable of sustaining temperatures to 15000C. Two

thermocouples placed approximately 2.5 cm from either side of the sample were used to

monitor the temperature in the hot zone. Silicon carbide platens resting on graphite push

rods connect the hot zone to the load train; the graphite push rods thread into stainless steel

actuator fittings at the top and bottom of the load train.

After coating sample contact surfaces with boron nitride high-temperature lubricant,

the specimen was placed between the platens and a slight load (- 5 kg) applied to stabilize

the setup. The chamber was then purged and backfilled with purified argon several times,

and left in the backfilled condition under flowing argon for the remainder of the test.

During sample heating at typical heating rates of 160C/min, the MTS was switched to load

control to protect the specimen from damage due to expansion of the test fixtures. After a

sufficient soak time at test temperature (-45 min) to allow for thermal equilibrium between

the specimen and platens, the MTS was returned to displacement control for the remainder

of testing. For all testing, a 2000 lb (907 kg) load cell was used with a 12.7 mm full

scale displacement card.

Measurement of strain during testing was obtained indirectly from the displacement

output of the servo-hydraulic low voltage displacement transducer (LVDT); the use of more

accurate techniques, such as a hang-down extensometer or strain gauges, were made

impractical by the small dimensions of the test setup and design of the furnace system.

Load and stroke (displacement) data during testing was displayed graphically on a Hewlett

Packard plotter with the raw data buffered and collected using LabTech Notebook software

application. The data collection period used depended on the projected length of the test

from 0.5 to 15 s.

For strain rate change tests, a minimum of three percent steady state deformation

after yield was used as the criteria for changing strain rates. For the multiple strain rate

tests, specimens were deformed out to strains approaching 0.30 (engineering). For both

single and multiple strain rate testing, the viewport in the furnace chamber was used to

monitor the specimen for evidence of barreling or off-axis deformation/shearing and if

either of these features was observed, the test was halted. Care was taken to minimize the

likelihood of these undesirable deformation artifacts through judicious system alignment,

good lubrication between specimen contact surfaces and platens, and careful sample

preparation to ensure flat and parallel end surfaces. For both coarse and fine-grained

samples, a combination of single and multiple strain rate tests covering strain rates (e) of

1x10'5 to 9x10-4 S-a were conducted at 1000, 1100, 1200, and 1300C.

To correct the observed displacement values for machine compliance, load-

displacement plots in the elastic regime were obtained with the SiC platens pressing against

each other. Compliance data was obtained for 1000, 1150, 1200, and 1400C at effective

displacement rates corresponding to E of 1xl04 and 1x105 s-' for 5 mm tall specimens.

From the reciprocal of the slope of the linear elastic load-displacement plots a machine

compliance parameter as a function of load was determined at each temperature and strain

rate of interest. After collection of raw load-stroke (position of crosshead relative to

original position) data for a given test, the appropriate compliance correction was used to

subtract the machine compliance from the raw displacement yielding an effective

displacement more closely matching the actual specimen deformation.

To calculate true stress and true strain values from the load and displacement

(compliance-corrected) output, the engineering stress (r,,) and strain (e,,) were first

calculated using the following expressions,
aRg P/Ao (4.3)
=- (4.4)

where P is the applied load, A0 is the original cross-sectional area of the specimen, Al is the

compliance-corrected displacement, and 10 is the original height of the compression

cylinder. From these expressions, it is possible to derive [86] the following expressions

for true strain (E,) and stress (q,) which are better metrics of actual specimen deformation

than their engineering counterparts, especially at high strains,

e,=ln(,g +1) (4.5)

o, = ,,,g (l- ,) (4.6)

For single-strain rate tests, the flow stress was determined using the standard 0.2 % offset

convention [86]; for multiple strain rate tests, a value from the steady-state region of the

stress-strain curve was used as the flow stress

In order to examine flow behavior and strain rate sensitivity from strain rate change

testing it is necessary to determine the true strain rate (e,) for a given imposed crosshead

displacement rate, d [63]. This correction is necessary because the nominal displacement

rates programmed into the controller before testing use the initial length of the sample in

rate calculations which is only valid for the strain rate used at the outset of a given

compression test. Defining the initial true strain rate as ,,, which is equivalent to d /10, one

can calculate e, from:

e, = o. (4.7)
1 + eng

To gain more insight into the high-temperature deformation mechanisms of

MosSi3C, examination of post-deformation microstructures was conducted in the Phillips
420 transmission electron microscope at 120 kV using thin foils from several samples

deformed an average of 8 %. Sample preparation techniques were similar to those

previously described with a few minor modifications. First of all, slices for subsequent

thinning were taken from the center of the compression specimen gage length to minimize

the possibility of obtaining regions from the low-strain regions near the ends of the

specimen [86]. In some cases it was necessary to attach the partially thinned foil using

crystal mounting wax to a 100 ptm thick copper supporting grid 3 mm in diameter

containing a 1 mm x 2 mm oval cutout; this was necessary during the final stages of

manual thinning using the Gatan disc grinder to prevent sample breakage and further edge

erosion prior to dimpling and ion milling.

4.2.5 Four-Point Bend Testing

In order to compare the flow and fracture behavior under compression to that in a

case involving a tensile loading component, several four-point flexure tests were performed

on Mo5Si3C-containing samples. Testing was conducted at Ames Laboratory (Ames, IA

50011) using an MTS servo-hydraulic test system with a Centorr Testorr furnace

attachment in a flowing nitrogen atmosphere. Rectangular parallelepiped specimens were

tested at room temperature and 12000C; for the 12000C testing, the average heating rate was

21C/min. Sample preparation was in accordance with ASTM Designation C1211-92

[158]. Sample configuration A from this standard was slightly modified to allow for easier

sample handling; the sample dimensions used were b=3 mm, d=2 mm, and L= 25 mm,

where b is the width, d the depth, and L, the length of the specimen. The direction of hot

pressing was parallel to the 2 mm (depth) dimension of the specimens which also was the

loading direction.

Bend-test specimens were initially cut from 31 mm diameter hot pressed disks

using a low speed diamond saw and the surfaces ground with 240 grit SiC paper to remove

any irregularities introduced during sectioning. Next, the end surfaces were ground flat

using the V-Block and 180 grit SiC paper until the sample was 25 mm in length. Samples

were then attached to a cylindrical 3.18 cm diameter stainless steel flat using crystal

mounting wax. The flat was inserted into a stainless steel ring 2.54 mm tall with a 3.18

mm inner diameter and the height of the flat adjusted using a set screw until the surfaces of

the test samples were just above the surface of the ring. By using the ring and flat with a

series of SiC grinding papers from 240 to 600 grit it was possible to achieve the desired

sample dimensions while maintaining the four longitudinal surfaces flat and parallel to

within 0.02 mm. As a final step, a 450 chamfer was ground into the four longitudinal

edges to remove any chipping or flaws along the edges that might otherwise cause

premature failure of the specimens.

Both room temperature and elevated temperature testing was done using SiC bend

fixtures having a lower support span of 20 mm and a loading span of 10 mm with 1mm

diameter SiC rollers used as contact points. Samples were tested at a constant crosshead

displacement rate of 0.14 mm/min, which is equivalent to a nominal strain rate of 7.2 x 105

s-' for the sample dimensions used. Displacement measurements were taken from both

strain gauges affixed to the hydraulic ram and the LVDT. To calculate the strength (S) in

four-point bending, the following expression was used,
3-P -L
S= -- (4.8)

where Pis the load at failure and L the length of the support span. For calculation of the

outer fiber stress, a similar equation was used,
3P(L-s) 49
ao= 2bd2 (

where P is the load and s the loading span.

Fracture surfaces of specimens which failed during bend testing were examined in

the JEOL JSM 6400 scanning electron microscope at 15 kV. Both low and high

magnification examination of the surfaces was done, with an emphasis on obtaining

information about fracture mode, fracture path, and possible initiation sites.

4.3 Results and Discussion

4.3.1 Microstructural Characterization of Test Specimens

From each hot-pressed sample used as stock for mechanical test specimens, a

section of material was removed and set aside for characterization so that the observed

mechanical behavior could be correlated with the starting microstructures and processing

variables. The majority of materials used for mechanical testing were similar in that they

were close to single-phase MoSi3C having less than 6.1 volume percent total second phase

constituent. Samples also exhibited variations in grain size, porosity, and composition

(Table 4.1). For samples processed by powder blending and hot pressing, the average

grain size varied from about 14 to 23 pim, while the hot-pressed, mechanically-alloyed

samples had average grain sizes of 6 to 8 gm.

One major difference observed between the samples produced from blended

powders and those synthesized from MA powders was the presence of amorphous silica

particles in the MA samples (Figure 3.15, 3.16). Silica was typically visible in MA

specimens as networks of globular particles -0.1 im in diameter concentrated near grain

boundaries and triple points; in no instance were non-crystalline phases detected during

TEM examination of coarser-grain blended powder samples. The most logical explanation

for the presence of silica in the consolidated MA samples is the increased surface area of the

starting powders leading to significant and rapid oxidation of the powders after exposure to

air. In fact, one batch of MA powders had such a rapid and exothermic oxidation upon

exposure to air that they exhibited small-scale combustion. Because of the relatively large

sample volume, the amount of free carbon in many of the silica-containing regions was

probably insufficient to completely reduce the glassy phase during hot pressing. Evidence

of a large composition shift as a result of C-SiO2 reactions is seen particularly in sample

LD6, where the shift toward the Mo-rich region of the Mo-Si-C isotherm was significant

enough to move the final composition into the MosSi3C + Mo2C + MosSi3 phase field.

Based on work on other silicides containing silica [9,35,159,160], it was expected

that the deformation behavior of polycrystalline silica-containing MoSi3C would be

affected, particularly at elevated temperatures. One of the main goals for the processing of

samples for mechanical property evaluation was to be able to synthesize nearly-single

MosSi3C over a range of grain sizes. Initially it was hoped that blended powder specimens

having a grain size of approximately 10 to 20 .m would act as the lower end of the grain

size range while samples produced from coarser (-95 gm) starting powders would yield

coarse grained (>40 jim) specimens. However, 32 mm diameter samples processed from

these coarser powders displayed large cracks on the outer surfaces of the hot-pressed disks

making them unsuitable for mechanical testing. Microcracks within consolidated samples

were previously observed, although large surface cracks were not observed for smaller

diameter specimens of any starting powder size.

The most plausible explanation for cracking of these larger grain specimens

involves the previously mentioned issue of anisotropy in the coefficients of thermal

Table 4.1 Composition and microstructural characterization information for materials
used for mechanical test specimens. Composition number (CN) corresponds to those
assigned in Table 3.3. The MoSi3C phase is designated by T; PBHP indicates powder
blending and hot pressing, MAHP indicates mechanical alloying followed by hot pressing.
Under "Type of Testing", C corresponds to compression testing and B corresponds to
four-point bending.

Sample CN Proc. Phase ID Grain Size, Type of Notes
ID Route (Vol. %) Avg. (itm) Testing
13IS2 13 PBHP T, SiC (6.1) 13.7 1.4 C
53C10 C10 PBHP T, MoC (3) 14.6 1.6 C
C10B C10B PBHP T 22.5 2.7 C SiC/pores
< 1 Vol. %
LD C10B PBHP T 16.8 2 B > 1.2 %
_________ __ porosity
LD6 C10B MAHP T, Mo,Si (18.2), 8.3 0.9 C, B 6.1 Vol. %
Mo,C (2.3) silica
LD8 10B MAHP T, Mo2C (0.4) 5.6 0.5 C 2.1 Vol. %

expansion of the Mo5Si3C compound. Recall that CTE information for MoSi3C is lacking,

although one could reasonably infer an anisotropic CTE for the ternary compound based on

the reported CTE anisotropy for the isostructural TiSi3 compound [103] as well as the

adjacent binary silicide MosSi3 [157]. Recall also from the relationship (e.g., Equation

2.4) between grain size, temperature gradient and thermal stresses upon cooling, the

increase in stress associated with large grain size and temperature gradients. Based upon

these factors, the 32 mm diameter, coarser grain specimens would be expected to have a

higher propensity for thermally-induced cracking upon cooling than would smaller

diameter, finer grain samples. As further evidence supporting CTE anisotropy in MoSi3C,

it was observed that the density of microcracks tended to scale with the matrix grain size,

with the finest grain size materials (e.g., LD8 in Table 4.1) basically crack-free after hot

pressing, even in the largest samples.

The noticeable cracking in the large-grained samples as well as the limited

availability of coarser starting powders were obstacles to production of large volume, large

grained materials for mechanical testing. Accordingly, the decision was reached to use

samples processed by mechanical alloying and HP as the lower end of the grain sizes

studied, and use the 10-20 pm grain size materials as the upper end (Table 4.1).

4.3.2 Hardness and Fracture Toughness Measurements

Hardness values for several of the materials produced in this study are presented in

Table 4.2. The average hardness of the materials tested was 13.2 GPa (1340 HV) which is

between the two values given in the literature for MosSi3C of 12.1 GPa (1230 HV) [120]

and 14.3 GPa (1460 HV) [18]; the range of hardness values reported for the ternary phase

may be partially due to the different loads used, 10 kg for [120] and 50 g for [18]. The

hardness is marginally higher for the finer grain size materials in Table 4.2, although the

reasons for this are unclear. It is possible that elastic anisotropy of the hexagonal Mo5Si3C

phase could be partially responsible for the variation in hardness with grain size, with local

variations in elastic moduli with grain orientation. For coarse grain samples, the typical

indent diagonal was twice the average grain diameter, while for the fine grain materials the

average diagonal crossed more than seven grains, implying that elastic anisotropy effects

would be more pronounced in the large grain material since each indent sampled fewer

grains. Further evidence for elastic anisotropy was the slight asymmetry of some indents,

presumably due to local variations in elastic constants from grain to grain.

As shown in Figure 4.1, indentation caused cracks to radiate from the corners of the

hardness indent into the matrix; in this particular sample, some cracks branch out from the

long edges of the indent, i.e., not only from the four covers. The position of the indent

relative to a particular set of grains and variations in grain size and shape may account for

this observation; the crack emanating from the bottom corer of the indent is straight and

well-defined as it runs through a larger grain, while the more diffuse transgranular and

intergranular cracking seen at the right corner of the indent is a in a region with a higher

concentration of smaller grains which provide a lower-energy crack path.

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EU5MNH8QN_NKO7N4 INGEST_TIME 2013-09-28T02:25:25Z PACKAGE AA00014253_00001