Proceedings of the 13th Automotive Materials Conference: Ceramic Engineering and Science Proceedings, Volume 7, Issue 9/10

131 Pages·1986·8.4 MB·English

13th Automotive Materials Conference
Proceedings of the
13th Automotive Materials Conference
Tariq Quadir and Y. Tien
T.
Conference Chairmen
A Collection of Papers Presented at the
13th Automotive Materials Conference
Sponsored by the
Department of Materials and
Metallurgical Engineering
and
Michigan Section
The American Ceramic Society, Inc.
November 6-7, 1985
University of Michigan
Ann Arbor, Michigan
ISSN 0196-6219
Published by
The American Ceramic Society, Inc.
65 Ceramic Drive
Columbus, Ohio 43214
@The American Ceramic Society, Inc., 1986
i
hecutfve Director & Publisher Editor
W. Paul Holbrook William J. Smothers
Director Publicahions Production Coordinator
of
Linda S. Lakemacher Carl Turner
I
Committee on Publications: Victor A. Greenhut, chair; David W. Johnson,
Jr.; John F. MacDowell; W. Paul Holbrook, ex oficio; Lynn A. Morrison, ex
officio; Liselotte J. Schioler, ex officio; William J. Smothers, ex oficio.
Editorial Advfsory Board: Liselotte J. Schioler, chair; Cameron G. Harman,
Jr., chair-elect; Hamid Hojaji; Hamlin M. Jennings; Waltraud M. Kriven; Ronald
H. Lester; David B. Marshall; Gary L. Messing; John J. Petrovic; William J.
Rice; Thomas F. Root; Robert 0. Russell; James E. Shelby, Jr.; David P.
Stinton; James M. Stubbs, Jr.
Editorial and Subscription offfces: 65 Ceramic Drive, Columbus, Ohio,
43214. Subscription $60 a year; single copies $15 (postage outside $5
US.
additional). Published bimonthly. Printed in the United States of America.
Allow four weeks for address changes. Missing copies will be replaced only if
valid claims are received within four months from date of mailing. Replacements
will not be allowed if the subscriber fails to notify the Society of a change of
address.
CESPDK Vol. 7, 9-10, pp. 1073-1197, 1986
NO.
The American Ceramic Society assumes no responsibility for the state-
ments and opinions advanced by the contributors to its publications, or by the
speakers at its programs.
@Copyright, 1986, by the American Ceramic Society. Permission to photocopy
for personal or internal use beyond limits of Sections 107 and 108 of the
the US.
Copyright is granted by the American Ceramic Sodety for libraries and other
Law
usas regstered with the Copyright Clearance Center, provided that the fee of $2.00
copy of each is paid directly to CCC, 21 Congress Street, Salem, MA
per article
01970.-The fee for articles published before 1986 is $2.00 per copy. This
also
consent does not extend to other kinds of copying, such copying for general
as
distribution, for advertising promotional purposes, or for creating new collective
or
Requests for special permission and reprint requests should be addressed
works.
to the Technical Editor, the American Ceramic sod@ (01%-6219/86 $2.00).
ii
Preface
A very strong interest has developed in the automotive industry in ceramic
materials. One of two prime focal points is the use of ceramics engine
as
components. (The second is ceramics for sensors.) Ceramics are attractive
for use in engines because their presence in selected locations permits higher
operating temperatures and therefore greater fuel efficiencies.
The Michigan Section of the American Ceramic Society and the Depart-
ment of Materials Science and Engineering at The University of Michigan
sponsored a symposium on Promsing of Automotive Ceramics for their 13th
Automotive Materials Conference that was held in late 1985. Approximately
200 engineers attended sessions that included powder preparation and charac-
teristics, fabrication processes, silicon-based ceramics, zirconia-containing cer-
amics, and composite ceramics. Throughout these the role of microstructure
on the properties and performance received particular attention.
The support of the following companies is gratefully acknowledged:
AC Sparkplug Division, GMC Ford Motor Company
Allied Automotive General Motors
Champion Spark Plug Co. Metallurgy Dept.
Chrysler Corp. Physical Chemistry Dept.
Corning Glass Works Harshaw/Filtrol Partnership
Corning Glass-Zircoa Mitsui & Co. (USA)
Detroit Diesel-Allison, GMC NGK-Locke, Inc.
Diamonite Products Sohio Engineering Materials
Dow Chemical Co. TAM Ceramics
T. Y. Tien, L. H. Van Vlack
The University of Michigan
iii
Table of Contents
Some of the High Temperature Performance of
Aspects ........................
Ceramics and Ceramic Composites 1073
A. G. Evans and J. Dalgleish
B.
..........................
Ceramic Component Fabrication 1095
L. Kennard
Fred
.......
The Role Powder Properties in Ceramic Processing.. 1112
of
J. A. Mangels
Processing, Microstructure, Properties Relationships for
..........................
Automotive Structural Ceramics 1122
David W. Richerson
Processing and Properties of Structural Silicon Carbide ....... 1135
T. J. Whalen
Preparation of Silicon Nitride Powders ..................... 1144
Gary Crosbie
M.
Transformation-Toughened Bulk Tetragonal Zirconia: I,
...................
Overview of Development and Properties 1150
T. K. Gupta and C. A. Anderson
Transformation-Toughened Bulk Tetragonal Zirconia: 11,
Mechankal Properties Dependence on Composition. Grain
..................................
Size, and Temperature 1 158
C. A. Anderson and T. K. Gupta
............
Processing of Transformation-Toughened Alumina 1169
Keith Wilfinger and W. Roger Cannon
.........................
Alumina-Sic Whisker Composites 1182
T. N. Tiegs and P. F. Becher
Chromium-Aluminum Hydroxide Synthesis and Sintering of
............................
the Solid Solution CrA,.,O,. 1187
S. Raman, Doremus, and German
V. R. H. R. M.
Role of Also, on Properties of Sintered Si,N,-Y,O,-Al,O,
..............................................
Ceramic 1197
G. Bandyopadhyay, K. French, and A. Pasto
W. E.
V
Ceramic Engineering and Science Proceedings
William J. Smothers
Copyright @The American Ceramic Society, Inc., 1986
Some Aspects of the High Temperature Performance
of Ceramics and Ceramic Composites
A. G. EVANSA ND B. J. DALCLEISH
College of Engineering, Univ. California
Santa Barbara, CA 93106
Ceramics and ceramic composites are subject to creep rupture at eleuated
temperatures. The rupture strain in such materials has been shown to exhibit a mapr
transition. from creep brittleness to creep ductility. The emphasis of the present
article is on the dejinition of microstructures that provide ductility. For this pur-
pose, thefundamental principles inoolued in high temperatureflow andfracture are
reviewed, and physical models of the ductile-to-brittlet ransition are presented. The
mechanicalphenomenoi nwlwd in these considerations include: creep crack growth,
crack blunting. flaw nucleation, and stress corrosion.
Introduction
Ceramics are typically capable of withstanding higher temperatures than other
materials. Hence, the substantial interest in such materials for heat engines,',*
bearing^,^ etc. However, high temperature degradation phenomena exist that in-
fluence performance and reliability. The important degradation processes include:
creep,' creep rupture,S.6f law generation,' diminished toughness,*a nd microstruc-
tural instability.9T he fundamental principles associated with some of degrada-
these
tion phenomena are reviewed, and prospects for counteracting the prevalent
mechanisms are discussed.
The strength of a ceramic typically diminishes at elevated temperatures (Fig.
I), initially owing to the diminished potency of toughening mechanisms**a nd subse-
quently, following the onset of creep." The degradation mechanisms that operate
at the highest temperatures - in the creep regime - are emphasized in this article.
A dominant microstructural consideration with regard to elevated temperature
behavior is the existence of a grain boundary phase.9 Such phases typically re-
main after liquid phase sintering and, frequently, are amorphous and silicate-based.
The second phase constitutes a vehicle for rapid mass transport and dominates the
creep, creep rupture,I3 and oxidation9p roperties, as well as the microstructural
I*
stability. The grain size constitutes another important microstructural parameter,
by virtue of its influence on the diffusion length and on the path density. Amor-
phous phase and grain size effects are thus emphasized in subsequent discussions
of microstructural influences on high temperature properties.
The high temperature phenomenon that, in the broadest sense, has overwhelm-
ing practical significancet is the existence of a rrunsition between creep brittleness
and creep d~ctilirys(.F~i g. 2(A)(B)).F racture in creep ductile regime occurs at
large strains (c 5 0.1, Fig. 2(C)), in excess of allowable strains in typical com-
ponents. Consequently, when creep ductile behavior obtains creep rupture is not
normally a limiting material property. The current article thus emphasizes the
material parameters that govern the brinle-to-ductile transition. However, it is
recognized that this transition may not occur within a practical range in materials
having undesirable microstructures. The emphases regarding microstructural design
would thus differ from those presented in this article. Finally, some preliminary
1073
remarks and speculations regarding the influence of reinforcements, such as
whiskers and fibers, on the high temperature performance are presented.
Creep Ductility
The transition to creep ductility represents, at the simplest level, a competi-
tion between flow and fracture and thus, occurs when the flow stress becomes
smaller than the stress needed to induce the unstable extension of cracks (Fig. 2
(A)). At a more sophisticated level, it is necessary to specify the flow and fracture
characteristics, subject to the imposed loading. Theflow in fine grained materials
is supposedly governed by diffusional creep and can usually be represented by a
viscosity*
kT l3
'
+
= D6 [l DJlDbIfl
where 1 is the grain size, Dv is the lattice diffusivity, 52 the atomic volume and
Db is the diffusion parameter pertinent to either the grain boundary, D&, or the
grain boundary phase, 016,. Some complicating effects occur in very fine grained
materials, involving nonlinearity at low stresses.'* Such effects are not understood,
but are presumed to relate to stressdependent interface-limited phenomena (such
as grain boundary sliding). Nonlinearities are also encountered in liquid phase
sintered systems,15.'6a gain for reasons not yet apparent.
The pertinent fracrure processes are more complex. The fracture parameter
seemingly having the greatest relevance to the brittle-toductile transition is the
threshold stress intensity, Kth, that dictates the onset of crack blunting6 (Fig. 3).
Specifically, at stress intensities below Kth, crack growth is prohibited, whereupon
creep ductility is assured (Fig. 2(A)).A conservative criterion for creep ductility
is thus obtained by applying the inequality
where u is the radius of the largest crack that either pre-exists or may be nucleated
by heterogeneous creep, oxidation, etc. and ad is the design stress. However, it
is also recognized that the permissible creep strain E* must not be exceeded within
the lifetime, f*, resulting in a second criterion,
The inequalities of Eqs. (2) and (3) must both be satisfied in order to assure ade-
quate creep performance. Further progress thus requires appreciation of the creep
crack growth threshold, as well as an understanding of the dominant high
temperature flaws.
In some materials, significant creep crack growth is not encountered before
the ductility transition. For such materials, the critical stress intensity, Kc is pre-
sumed to be relevant fracture parameter, replacing Kth in Eq. (2). Consequently,
the
Kc at elevated temperatures is also afforded consideration.
Creep Crack Growth
Creep Cmck Growth Mechanisms
The basis for comprehending creep crack growth mechanisms is the character
of the crack tip when diffusion operates, at elevated temperatures. At such
temperatures, chemical potential continuity and force equilibrium are demanded
1074
at the crack tip.17 Hence, since cracks are typically intergranular at high
temperature^,'.^.^^ the crack tip must be partially blunt (Fig. 4) in order to satisfy
the equilibrium relations,''
where q is dihedral angle, Yb are the grain boundary and surface energies,
the and Ys
respectively, is the surface curvature at the crack tip and uo is the normal stress
k,
on the grain boundary at the tip intersection. The resultant tip configuration, as
well as the corresponding crack tip field are very different from those associated
with the sharp cracks involved in brittle fracture. Consequently, the conditions for
extension of the crack cannot readily related to the ambient fracture toughness.
be
Instead, the crack growth mechanisms involve the removal of material from the
crack tip region (by diffusion or viscous flow), resulting in the creation of new
crack surface. Two categories of such mechanisms typically dominate: direct ex-
tension mechanisms that entail matter transport over relatively large distance~~~.~*
(Fig. 4(A)), and damage mechanisms that involve small scale mass transport within
a zone directly ahead of the crack tip'9*20(F ig. 4(B)). However, the mechanistic
details are sensitive to various aspects of the microstructure.
Creep crack growth rates in ceramics that exhibit Newtonian behavior typically
satisfy the nondimensional form:
KIuoJL = F(M) (5)
where L is a characteristic length for grain boundary diffusion, and F is a function
of various microstructural features, such as grain size cavity spacing. Typically,
and
both uo and L depend on crack velocity, resulting in non-linear crack growth rates
a =ao(KIK# (6)
where ho and n are material sensitive coefficients. In particular, the magnitude
of n depends sensitively on the dominant mechanism and the choice of boundary
conditions. Selection of conditions that pertain to the actual crack growth problem
of interest is thus a crucial aspect of comparing crack growth measurements with
predictions.
In some materials, especially those containing amorphous phases, intact
ligaments of amorphous material remain behind the crack tip16 (Fig. 5). These
ligaments enforce crack surface tractions that reduce the tip K and thus impede
crack growth. Such wake effects need to be incorporated into generalized models
of creep crack growth. Some of the relevant models and the associated conditions
are described below.
Direct Extension Mechanisms: Direct crack extension involves the mass flow
depicted in Fig. 4(A). The flux within the crack is directed toward the tip, while
the local grain boundary flux occurs away from the tip, causing net removal of
matter from the crack.17,1T8 he deposition of matter onto the grain boundaries is
accommodated by grain displacements normal to the crack plane, resulting in work
done on the system. The work done compensates for the increase in both surface
energy and strain energy, thereby allowing crack extension to proceed with a net
reduction in free energy. Crack growth rate predictions have been performed for
cracks located at a bicrystal boundary, wherein matter deposition is accommodated
elastically. The corresponding viscoelastic behavior pertinent to fine-grained
polycrystals has yet to be evaluated. The importance of grain size is thus, present-
1075
ly, unknown. Nevertheless, the elastic results provide useful insights. The non-
dimensional crack growth rate when matter transport involves surface diffusion
along the crack has the
where E is Young's modulus and the subscript s refers to the surface. The cor-
responding relation when the crack contains amorphous fluid phase that 'wets'
an
the crack surfaces is'"
where co is the equilibrium concentration of solid dissolved in the liquid. These
results clearly indicate the relative role of the mass flow parameters, Db and ~ 1 ,
as well as important effects of the dihedral angle (i.e. of Y//Ye). Furthermore, it
is noted that the crack growth rate is predicted to vary as a nonlinear function of
K, due to the nonlinear relation between crack velocity and the predominant diffu-
sion lengths (e.g., L in (5)).
Eq.
Operation of the above mechanism in polycrystals is restricted by the ability
of cracks to circumvent grain junctions. Specifically, when the crack does not con-
tain a wetting fluid, the dihedral angle, 'Ir, is large and substantial mass flow is
needed to achieve crack extension across a grain junction. Consequently, only the
relatively narrow cracks that obtain at higher velocities extend- b y this me-ch anism.
However, when a wetting fluid is located in the crack, ('Ir 0 or 71 rb/2),
the crack can remain a narrow entity,'" even at low velocities, extend beyond
as and
the grain junction. For this reason, a wetting fluid may be regarded as a prime
source of high temperature stress corrosion.
Materials that contain a continuous amorphous phase may be subject to alter-
an
native direct crack advance mechanism.13 In this instance, an amorphous phase
meniscus at the crack tip (Fig. 6) simply extends along the grain boundary, caus-
ing the crack to grow, and leaving amorphous material on the crack surface. Analysis
of this process has been conducted subject to the conditions: the amorphous phase
is thin, the grain displacements are discretized by the sliding of grain boundaries
ahead of the crack and such displacements are accommodated by viscous creep
of the surrounding solid. Then, crack growth is highly constrained and the crack
growth rate has the forrn,I3
where 6 is now the amorphous phase thickness (the subscripts o and c refer, respec-
tively to the initial value and the value when the grains at the crack tip separate).
Unfortunately, it is not possible to compare (9) with J2q. (8), because of the
Eq.
very different material responses used to derive the results. Nevertheless, it is
noteworthy that the crack velocity in (9) is insensitive to the thickness of the
Eq.
second phase, but strongly dependent on grain size.
,,A
Damage Mechanism: The prevalent mechanism of damage-enhanced crack
growth involves the nucleation and growth of cavities on grain boundaries in a
damage zone ahead of the c r a ~ k '(~Fi.g~. 7~). The stress on the damage zone
motivates growth of the cavities, once nucleated. Consequently, the crack progresses
1076
when the damage coalesces on those grain facets continguous with the crack. The
growth of cavities in the damage zone generally causes displacements that modify
the
the stress field ahead of the crackZo( c.f. Fig. 4(A)). Determination of the crack
growth rates thus requires solution of simultaneous relations for cavity growth
the
rate (as determined by the resultant normal stress) and the stresses (as dictated by
the displacements induced by cavity growth). Such calculations have been con-
ducted for a viscous sold.13*zT0h en, when the damage zone is large (such that
damage growth is relatively unconstrained) the steady-state crack growth rate has
the form
where X is the spacing between cavities in the damage zone. Non-linear behavior
would obtain if All were dependent on crack velocity. Zone size effects also emerge,
and affect the linearity, when the zone size becomes small.I3
Comparison of the above crack growth rate predictions with data has been
achieved by using independent measurements of X and of the damage zone size
obtained, on failed specimens2‘( Fig. 8). However, a full predictive capability does
not exist, because is no fundamental understanding of effects of microstruc-
there the
tue on A. Nevertheless, certain important trends are apparent. In particular, the
importance of the grain size, diffusivity and cavity spacing appear explicitly and
have the expected influence on crack growth rates. When an amorphous phase is
present,20 the velocity increases by 116, as well as by the increase in diffusivity
(DlIDb),
Mechanism Regimes: Various observations and predictions suggest that the
direct extension damage mechanism have differing realms of dominance. Obser-
and
vurions of failed specimensz1h ave revealed that cavitation damage exists on the
fracture surface in the region of slow crack growth (Fig. 9(A)). By contrast, rapid
propagation is accompanied by a facetted fracture surface (Fig. 9 Such obser-
(4).
vations clearly suggest the prevalence of damage mechanisms at the lower crack
velocities. Crack growth modek predict similar features (Fig. 3), the direct
because
extension mechanisms have a larger n (Eq.( 6)). owing to additional velocity depen-
dent parameters (notably, the crack width). This separation of the regimes of
relevance has significant implications for two features of the fracture process: the
crack growth threshold, Kth, and the critical stress intensity factor, Kc, discussed
as
in the subsequent sections.
Effect of Ligumenrs: When intact ligaments remain behind the crack tip, they
exert forces on the crack surface that tend to reduce the tip K and thus diminish
the creep crack growth rate. The general trends can be conceived from a simplified
analysis, depicted in Fig. 5, based on observations by Wiederhorn, er al. I5.l6 The
intact regions exert tractions that depend on the size, I, and viscosity, TI, of the
ligament material. The corresponding opening rate of the crack surface is govern-
ed by the viscosity 7 of the body and the resultant tip K. Hence, by utilizing a
Dugdale analysis, it can be readily demonstrated that the change in K provided
by the intact ligaments has the form,
where d is the spacing between ligaments and x is a constant = 0.1. Then, the
crack growth rate may be related to the applied K,b y combining Eqs. (6) and (1 1)
1077
with
K,=K+AK
to give the relation
The ligaments thus introduce a complex dependence between crack growth rate
stress intensity. Furthermore, strong effects on crack growth rate of the viscosity
and
of the ligament material and ligament size and spacing are apparent. Ligament ef-
may be of considerable importance in the near threshold region and thus,
fects
some understanding of how ligaments form is regarded as an important topic for
future research.
The Threshold Stress Intensity
The considerationso f the preceding sections reveal that the threshold represents
a process that intervenes while crack growth is occurring by a damage mechanism
(Fig. 3). It thus seems appropriate to regard the threshold as a stress intensity level
that inhibits the nucleation of damage in the crack tip region.22F or a viscoelastic
solid, typical of most ceramics, damage inhibition would require that the elastic
stress on the first grain boundary facet (as modified by grain boundary sliding,
at the crack tip) be less than a ‘critical’ stress for cavity nucleation. Indeed, con-
siderations of cavity nucleation rates2*i ndicate that crack growth can be nuclea-
tion limited (Fig. lo), resulting in a relatively abrupt decrease in the crack growth
rate. A nucleation limited threshold thus seems plausible, with the threshold oc-
curring at a stress intensity
where, F(*) = ,//‘~(8/’~/3)[”2’ -3cos~+cos3~]Lor’ 3in, the presence of an amor-
phous phase,
This predicted threshold is larger than values observed experimentally (probably
because of additional stresses induced by grain boundary sliding transients).22N ever-
theless, general trends in Klhwith grain size and surface energy appear to be in
accordance with the limited threshold data available in the literature. Specifically,
the threshold is apparently lower in materials having a fine grain size* and in the
presence of an amorphous phase that both reduces the surface energy pertinent
to damage nucleation, and allows increase in characteristic nucleation dimen-
an the
sion (lie replaces,
PI3).
Comparison of (14) with (2) reveals the explicit influence on the ductile-
Eq. Eq.
to-brittle transition of such parameters the grain size, diffusivity, surface energy,
as
dihedral angle, and amorphous phase content. In particular, amorphous phases
substantially reduce Kth and thus encourage creep brittlene~s’~T.’h~e major re-
maining uncertainty is the flaw size, a. High temperature flaws are discussed in
the following sections.
1078