CAHBON ANI) GRAPHITE MATRICES IN CARBON-CARBON COMPOSITES

8/11/2019 CAHBON ANI) GRAPHITE MATRICES IN CARBON-CARBON COMPOSITES

1/16

CAHBON ANI) GRAPHITE MATRICES IN CARBON-CARBON COM POSITE S:

AN OVERVIEW OF THEIR FORMATION, STRUCTURE, AND PROPERTIES

Gerald Rellick

Mechanics and Ma terials Technology Cente r. Technology

0

erations

The Aerospace Corporation,

PO.

Box 92957,

Los

Angeles, E A 90009

Keywords: Carbon-carbon composites. Graphitization. Composite properties

WHY CARBON-CARBON COMPOSITES?

Carbo n-carbo n (CIC) composites. so-called because they combine carbon-fiber reinforcem ent in an all-

carbon matrix. can best be viewed as part of the broader category of carbon-fiber-based composites,

a l l

of

which seek

to

utilize the light weight and exceptional streng th and stiff ness of ca rb on libers. However, in CIC.

the structural benefits

of

carbon-fiber reinforcement a re combined with the refractoriness

of

an all-carbon

materials system. mak ing CIC com posites the material

of

choice

for

severe-environment applications. such as

atmo spheric reentry, solid rocke t motor exhausts. and d isc brakes in high -perfor manc e military an d comm er-

cial aircraft. Their dimensional stability. laser hardness, and low outga ssing

also

make them ideal candidates

for various space stru ctural applications.

Such mechanical and refractory properties are not met by the various bulk graphites for t w o reasons:

(1)

graph ites arever y flaw sensitive and . therefore. brittle: and

(2)

graph ites ar e difficult to fabricate in to large

sizes and complex shapes. The se difficulties are hrgely overcome by taking adva ntage of the "two phase prin -

ciple

of

material structure and strength

[I]."

In the classical two-phase mat erials system. or compos ite. a high-s trength . high-m odulu s, discon tinuou s-

reinforcement phase is carried in a low-modulus. contin uous- matrix phase: e.g.. grap hite fibers

i n

a thermo-

plastic-resin matrix. The stre ss in a composite structure having fiber reinforcement that is continuous in

length. is carried in proportio n to the moduli of the constitu ent phases. weighted by their respective volume

fractions. 'merefo re. the much stiffer (highe r-mod ulus) fibers will

y

tlie princip al load bearer s. and the ma-

trix. in addition to havingthetaskof binding togetherthccomposite.will deform under load and distribute the

majority

of

stress

to

the fibers. At the same time. becau se the brittle carbon fibers arc rso/&4 tlie poss ibility

that an individual fiber failure will lead to propagation and catastropbic failure is practically eliminated.

Ano ther major benefit of com posites is that they permit the construc tion of complex geom etries. and

in

such a way that different amou nts of the load-carrying fibers can b e oriente d in specific directions t o accom-

modate the design loads

of

the final structure. Closely assoc iated with this "tailoring" f eatu re of com posites is

that carbon-fiber technology enables exploitation of the exceptional basal-plane stiffness (and strength.

i n

principle, although this is still much far ther from realization) of sp2 bonded carbon atoms-Le., the fibers are

not

isotropic. but rath er have their g raphite bass1 planes oriented preferentially in the fiber axial direction.

For

very-high-temperature carbo n-fiber -com posite applic ations. say. abov e 2000C. even for brief peri-

od s of time, it is necessary to employ a carb on m atrix: however, like the fiber. the carbo n matrix is also brittle.

When fiber-matrix bond ing is very stron g

i n

CIC. brittle fracture is frequently obs erved. The explanation is

that strong bonding permits the development of high crack tip stresses at the fiber-matrix interfac e; cracks

that i nitiate in either fiber or matrix can then propaga te through the com posite. However. if the matrix

or

the

fiber-matrix interface is very weak. or microcracked. then the primary advancing crack can be deflected at

such weakened interfaces

or

cracks . This is the Cook-Gordon theory [2] for streng thening of b rittle solids.

which states. more specifically. that

if

the ratio ofth e adhesive strength of the interface to the general cohesive

strength

of

the solid is in the right range. large increases in the strength and toughness

of

otherwise brittle

solids may re sult. There fore, good fiber strength utilization in a brittle-matrixcomposite like CIC depends on

control of the matrix and interfacial structures.

The objective of this pa per is to provide a brief overview of carb on an d g raph ite matrices in CIC. with an

emphasis

on

recent research

on

some of the more fun dam ental m aterials issues involved

13-71,

Muc h of what

is presented is tak en from our own published work, which has focused

on

understanding how the structure

of

1094


2/16


3/16

In

addition to orientation, another important feature of carbon matrices is their g raphitizab ility, which is

a measure of the eas e in converting the pyrolyzed carbon matrix product into crystalline graphite through

high-temperature heat treatments in the

-2000-3000'C

range. Th e state of graphitiz ation can b e assessed by a

number of techniqu es, the most co mm on of which is X-ray diffraction (XR D). However, in C/ C it is usually

very difficult to resolve th e result ant compo site diffraction response in to the respectiv e fiber and m atrix re-

sponses, because both phases are carbon. A technique to circumvent the samp le volume problem is laser

Raman microprobe spectroscopy (LRMS). Although the interpretation of the Ram an spectra is more ambig-

uous

than with XRD , LRM S permits focusingof avisible-light beam, as small as 1 im in diameter,

on

aregion

of th e specimen while recording the Ram an spec trum, which is active in carbon

[u)] .

Useful struc tural infor-

mation on a local scale ca n b e obtain ed in principle.

O ne majo r difficulty with applyin g LRM S to composite s is that the size of constitue nt phas es is of the

order o f microns. making it necessary to prepare the specim ens for examination using st and ard optical polish-

ing techniques. Such polishing tends to damage the near-surface structures and leaves behind a thin layer of

polishing debris. Since the p robe depth of the optical beam is only abo ut 50 nm

[ZO],

he Raman spectrum

unfortunately becomes a function

of

the preparation technique

[21-231.

A techniqu e we have employed extensively and with good success, and which is an outgrow th of early

work performed at Los Alamos L aboratories [24.25], involves SEM examination of specim ens that have been

polished an d then cathodica lly etched with xenon. When th e carbon struc ture is graphitic, and when the

graphite layer planes are oriented perpendicular to the plane

of

sect ion , we see, typically, a pronounced lamel-

lar

texture,

as

revealed for the inner- and outerm ost CVI layers in the C /C

of

Fig.

3.

Th e lamella r texture is the

result of differential etching rat es of the various microstructural units, the exact nature of which is still not

clear. Th e most likely mechanism is preferential removal a t lower-density, less-o rdered intercrystalline-type

bounda ries th at sep arate regions of good crystalline registry; this is seen very dram atically in highly oriented

pyrolytic graphit es reac ted in oxygen [26 27]. he techniq ue is effective, principally, in distinguis hing broadly

between graphitic and nongraphitic carbon

on

a scale of microns.

Returning to Fig. 3, this particular specimen has the CVI deposition sequence R U S U R L (as determined

separate ly from polarized-light microscopy) and has been heat- treated to 2500C for

1

hr.

The

lamella r tex-

ture of the RL zones indicates their graphitized structure. whereas the absence

of

significant texture in the SL

zone indicates that the SL struc ture is essentially glassy carb on. This observation was confirmed by XRD ,

LRMS , and by selected physical-p roperty meas urem ents [B] he effect of having

a

graphitic and well-ori-

ented matrix is illustrated by the higher therm al conductivities for heat-treated R L composites shown in Fig. 4.

Modulus enhancement is another interesting effect of

a

well-oriented, graphitic matrix (Fig.

5).

For the

particular pseudo-3-D. fe lt-based C/C co mpo site of the figure. there were two CV I densifications . Following

the first, the composite stru ctu rew as heat-treated to 2500'C; the second CV I was left int he as-deposited state

( - loOo-1200"C). Th e relative proportions of the first and secon d CV I varied with each sp ecim en, bu t the total

CV I weights were approximately the sam e. The fiber volume (and weight) fraction was cons tant ( -u) ) for

each composite.

The strong dependence

of

the modulus on the relative proportion

of

heat-treated C VI indicates that the

carbon m atrix can carry a significant fraction of the load, particular ly, in this case, if the structure is heat-

treated to typical graphitization temperatures. Th e modulus-enhancement effect by the m atrix

is

especially

striking in this composite beca use of the use

of low-modulus fibers at fairly low volume fractio ns, However, as

will be seen, this effect is

an

importan t m aterials and processing consideration in all CIC composites.

Coal-Tar

and

Petro

eum

Pitches

Th e second method for CIC densification

is

the use of coal-tar and petroleum pitches. Because they are

thermo plastic, pitches a re used mostly for redensification: Le., furt her densifying of

a

C/C structure that has

been "rigidized" by

an

earlier impregnationldensificationste p (e.g., a resinimpre gnated fabric preform) or

that has sufficient rigidity from the friction between the elements of the woven structure ( e g , 3-D braided

preform).

1096


4/16

Pitches a re unique i n passing through a liquid-crystalline ransformation at tempe ratures between about

350 an d 55OC [29]. In this transform ation, large lamel lar molecules form ed by the reactions of thermal crack-

ing and aromatic polymerization are aligned parallel to form

a n

optically anisotrop ic liquid crystal known as

the carbonaceous mesophase

[30].

T he alignment of the lamellar molecules is the basis for easy thermal gra-

phitizability of the carbonized product. On e

of

the features

of a

meso phase-b ased matrix is high bulk density,

which is achievable because the matrix density can a ppro ach the value for single-crystal graphite , 2.26 g/cm3.

The topic of pitch impregnation and densification of C IC intro duces the subject of densification efficien-

cy, th e most meaningful me asur e of which is volumetric densifica tion efficiency [31]. It is the ratio of thevol-

ume of carbon matrix in a process cycle to the volume of porosity available for densification.

For pitches carbonized a t atmosp heric pressure, coke yields are of the order of

50-60%,

impregnant den-

sities

are

- 1.35 g/cm3. and , as we have noted, densities for pitch-derived matrices ar e -2.2 g/cm3. From these

values we calculate volum etric densificatio n efficiencies

of

only 3 0 4 0 % at atmosp heric pressure [31]. By re-

sorting to so-called hot isostatic-pressure-impregnation-carbonization (HIPIC), to pressures of about

15,000psi, carbon yields of pitches can be increased to almost

90%

[lo]. But even with HIPIC, volumetric

filling is only

55 .

There fore, given

a

preform with initial porosity of 45%, typical for man y 3-D woven struc-

tures, three cycles at maximum

densifico/ion efficiency

would be required to reduce the porosity to 4%. With

curr ent HIP IC pro cedure s, however, it is found tha t at least five cycles at

15,000

psi are requi red to achieve this

sa m e level of porosity. Su ch reduce d efficiency in real systems is the result of forced expulsion of pitch from

the preform as a result of the gas-forming pyrolysis reactions accom panying carbonization.

Clearly, one way

to

incr ease efficiency, for a given weight-based carb on yield, is to select either

an

impreg-

nant o r an HTT that will lower the final matrix den sity. As will be seen in the next sub section, lower-density

carbo n matrices

can

be achieved by using resin precursors that form

a

glassy-carbon-type structure. But,

although this approach fills more of the available space,

it

does

so

with a lower-density carbon matrix, which is

different in struc ture from the higher-density graphitic matrix. Th e trade-offs in pro perties, particularly me-

chanica l, ar e not well under stoo d. We will touch on this topic again in the next subsection.

Appro aches to improving densification efficiency

of

pitch-based matrices without resorting to HIPIC

processing include the use of heat-treated and solvent-extracted pitches [32] and partially transformed (to

meso phase) pitc hes [33,34]. A novel app roa ch, dev eloped by W hite a nd Sheaffer 1351, is to oxidatively stabilize

the m esophase following impregnation and transformation,

an

approach similar to that employed in meso-

phase-fiber stabilization. Th e result is a hardene d mesophase that is resistant to the bloating effects of

pyrolysis gases but that, upon further heat treatment, yields

a

dense, graphitic carbon.

Th e strong orienting effect of the fiber surface

o n

the large lamellar m esopha se molecules is

an

interest-

ing feature ofmesophase formation in C /C composites. This effect was demonstrated by the work of Zimmer

an d Weitz [36], who used p olarized-lig ht microscopy to show that m esopha se molecules near

a

fiber sur face in

a

close-packed fiber bund le always aligned parallel t o the fiber surfac es, even in the pres ence of stron g magnet-

ic fields. S inger and Lewis dem onstrated earlier that m agnetic fields would orient mesophase molecules in

bulk mesophase [37]. Zim mer

and

Weitz showed that mesophase would also orient in matrix-rich regions

within the fiber bundles-i.e.,

at

points far removed from fiber surfaces

[36].

They calculated a magnetic

coherence length of 7 Fm, which correspon ds roughly to the distan ce over which the orien tation effect acts.

Such localized o rientation in the liquid-crystalline state w ould lead o ne to expect the final, graphitized

matrix also to be well oriented in the imm ediate vicinity of he fiber. First observed by Evangelides

[38]

using

SEM

in conjun ction with xenon-ion-etching, suc h

a

matrix sheath effec t is depicted in Fig.

6

in

a

coal-tar-

pitch-densified C/C.

Modulus enhancement in pitch-based C/ C has been widely reported, b ut whether the effect is due to the

matrix or t o an increase in the fiber m odulus, resulting from high-temperature heat-treatment-induced struc-

tura l changes in the fiber, has not been clarified [39]. Th e sheath effect is

also

pronounced in resin-based

car bon matrices, but for d ifferent reasons, which we will examine in the next subsection.

1097


5/16

Matrix microcracking is characteristic of all CICs, bu t it is particularly prevalent in graphitic matrices

because of the comb ination of weak shear planes in polycrystalline graphite an d the the rma l stresses gener-

ated during heat treatm ent (Fig.

7)

[40,41]. Microcracking also has importan t effects

on

the engineering pr op

erties ofC /C materials-particularly the matrix-dominated properties in the unreinforced directions, such as

the interlam inar shear strength and perpendicular-to-ply tensile strength in 2-D C/C lam inates. However, as

mentioned above, such m icrocrack ing app ears

to

improve in- plane flexural and tensile strength, by way of a

Cook-Gordon mechanism [42 45] .

Th e third, and last, class of C/ C impregnant to be discussed is thermoset resins, which are th e basis for

prepreg fabric and tapes, a s noted above; resin systems c an also be used for reimpregnation.

In

addition to

their easy fabricability. therm osets have the a dvantag e of charring-in-place; th at is, although they softe n and

deform on heating, they d o not fuse o r iquefy, and, therefore, no special tools or techniqu es must be employed

to retain the matrix in the composite during pyrolysis.

Therm oset resins are usually highly crosslinked, which makes them resistan t to therm al graphitiz ation n

bukform, even to temperatures of 3000C [5,46]. P henolic resins are currently m ost commonly used for pre-

preg operations, whereas furan -bas ed resins ar e used mo re for reimpregn ating. B oth have char yields typically

in the

5040%

range.

Th e development

of

ultra-high-char-yield resins derived from polymerization of diethy nylbenz ene DEB)

[47-511, usually term ed polyarylacetylenes (PAA) [47], has received mu ch focu s in recent years. T h e stru ctur e

of DEB is illustrated in Fig. 8. along with a synthesis route tha t involves a catalytic cyclotrimer izationprepoly-

merization in methyl ethyl ketone solvent [48.49]. T he cyclotrimerization l iber ates m uch

of

the exothermic

heat of polymerization, thereby allowing safe, controllab le curing. T he principa l appea l of PAAs is th eir ex-

tremely high char yield. From th e average structure, we calculate a theoretical carb on yield of ab out 95 ; in

practice, PAA s can have carbon yields of 90% to 700C. although mor e practical formu lations employing

monofunctional chain terminators to improve flow properties reduce this yield to about

85

[48,49].

Similar to other crosslinked therm osets, PAAs prod uce largely nongra phitizable carbo ns. To extend the

range of matrix struc tures for this fabricable resin system, we have been exploring approa ches t o in siru matrix

catalytic graphitization in C /C in

our

laboratory On e promising approach, by Za ld iv ar et d. [52], has been the

use of boron in the form of a carb oran e compo und. Figure 9a is a plot of room-temperature tensile strength

of

undoped and boron-doped unidirectional C/C s versus HTI ; the strengths a re calculated relative to the fiber

cross-sectional areas on the assum ption that the matrix carries negligible load relative to the fibers. The

strength of the fibe rs in the cured- resin comp osite is taken to be th e value for full stren gth utilization. T he plot

illustrates a number of im port ant features. First, for the und oped syste m, strength exhibits a large decrease as

the comp osite proceed s from cu re to carbonization , owing to conversion of the co mp liant polymer matrix into

a well-bonded, low-strain-to-failure carbo n matrix. Increasin g boron levels lead to incre ased strength utiliza-

tion for the 1100C

H T I

samples: T he undoped specim en behaves a s a monolithic solid and fractures in a

planar-catastrophic mode (Fig. loa); the

5%

B-do ped samples exhibit extensive fiber pullout (Fig. lob), which

indicates a weakened interface. Th e reasons for the weakened interfac e ar e unclear, since X-ray diffraction

revealed no significant difference in graphitization between doped and undoped specimens after this

HIT

At higher H T Is , the use of higher boron levels leads to a red uction i n strength utilization(and an increase

in modulus; Fig. 9b), du e tocatalytic graphitiza tion of the fiber. Fur ther H To f th e undop ed spec imen s beyond

1100C reclaims much of the lost fiber strength, for the reaso ns discussed above. Mo re work is needed to

define the mechanisms by which catalytic graphitiz ation of the matrix affects the properties of C/C.

We recently reported a striking modulus enhancem ent for the same type of 1-D composite studies (Zaldi-

var

el

a/. [7]), using four m esopha se-based fibe rs from D uPon t and PAA resin (Fig. 11). The number in the

fiber designation is the axial tensile mo dulus, i n Mpsi. For

HT

to 2750C. all the composites exhibit sharp

increases in f iber moduli, t o values exceeding 150 Mps i, which is the theoretica l limit

of

the graphite basal-

plane modulus. Since the moduli are calc ulated relative to the original fiber cross -sectional areas, su ch values

1098


6/16

indicate that the com posite mod ulus m ust have significant contributions from the matrix. An example of

a

matrix sheath that may be contributing to the composite modulus is shown in Fig. 12.

This figure brings us to the su bject of stress-induced orientation an d graphitization in otherwise nongra-

ng carbo n m atrices in C / C . While the phenomen on of stress graphitization of hard carb ons ha s been

noted for some time [24,25,53,54], only recently have serious efforts have been made to understand the

physical-mechanical mechan isms involved in matrix orientation and graphitization in C / C [5,6,55]. This topic

isof more than academic interest, because the formation

of

a two-phase m atrix of graphitic and nongraphitic.

oriented and unoriented, zones ca n have a major influence

on

the m echanical properties

of

CIC composites.

To examine preferred orientation in thermosetting-resin-impregnated matrices, cross sections

of

C/C

tows fabricated from Amoco T50 PAN-based fibers and a

PAA

resin were polished, then heat-treated to

2900'C for 1h r and xenon-ion-etched. T he polarized-light microg raph of Fig.

13a

reveals that in addition to

the pronounced lamellar zones, the smoo th-appearing zones-which, by definition, have formed no observ-

able texture with etching-are nevertheless oriented,

as

evidenced by the polarized-light extinction contours

sweeping across the surface

of

the sample

as

the analyzer is rotated.

We conclude tha t even the thickest

(> -20 am matrix regions in this specimen a re oriented. Pronounced optical anisotropy in the matrix for the

same composite heat-treated to onIy 1200 "C is revealed by Fig. 13b.

As

expected, etching produced

no

lamel-

lar texture for this low HTT

The highly localized na ture of the com bination of stress-induced orie ntation an d graphitization is one of

its mo re intere sting features; Le.. all

of

the carbon matrix in the specimen of Fig.

13a

is oriented to some de-

gree, yet only certain discrete reg ions becom e lamellar graphite u pon

HT

to 2900'C.

SEMs

of ion-etched

specimens reveal this localized graphitization more clearly (Fig.

14);

particularly striking

is

the shrinkage of

the matrix away from the fiber, which is

a

result of the volume decrease accomp anying graphitization.

TEM is an extremely effective technique for studying the

local

structure on an even finer scale.

In

the

transverse (Fig. 15a) and longitudinal (Fig. 15b) bright-field images of thin s ections of a T50 fiberlresin-

derived C / C heated to 2750 C. crystallite formation and orienta tion a re evident, particularly in the transverse

section (com pare with Fig. 13a). Selected a rea electron diffraction confirm ed the highly crystalline struc tureof

the interfilament matrix regions [56].

In the SEM of Fig. 16a, we observe a n interesting effect: A t the interstice of five continguous fibers there

is no lamellar formation in the m atrix pocket, except perhaps imme diately adjacent to th e filament surfaces.

This effec t was typically observed in close-packed groups of three t o five fibers. In contr ast, in m ore extensive

matrix regions-for example. those tha t bound two relatively fiber-rich areas, and where the matrix bound-

aries are fairly straight-we observe relatively unimped ed developm ent of lamellar structur e over a dista nce of

several microns (Fig. 16b). Such lamellar development is particularly striking at th e extreme outsid e of the

single-tow specimen s where quite thick

( -

1-2 fiber diame ters) lamell ar zones form (Fig. 16c). In Fig . 16d, we

see that an interruption in the uniformity of the interface between this outer m atrix cru st and the composite

leads to

a

transition from the lamellar

to

nonlamellar structu re.

Further microstructural features not seen in polished specimens are revealed in the SEM of a tensile-

fractu re surfa ce (Fig. 17).

The

lamellar regions in the matrix a re still evident, and th e PAN-based fibers show

their typical fibrillar struct ure. But we now obselve in the matrix both lam ellar and fibrillar textures, the latter

resembling that seen in the T 50 fibers, which a re generally considered to be oriented glassy carbon [46].

Two observations sugges ted to

us

that the key factor in determ ining lamellar-structu re formation in

a C/C

compo site matrix is

a

multiarial deformationof the resin du ring its pyrolysis to carbon. First, consider that, in

normal PAN-fiber manufacture, which leads to a fibrillar structure, the filaments ar e subjected to a uniaxial

tensile Stress durin g oxidation stabilization. However, when carbonized w ithout prior oxidation stabilization,

but in

very

th in

sections, such

as

between the layers of montm orillonite clay, PAN ha s been sh own to yield a

single-crystal structu re following subsequ ent graphitizatio n heat tre atm ent [57 econd, in partially oxidized

(through-the-th ickness) PAN fibers. the unoxidized, fusible core can form lamellar carb on [58].

1099


7/16

In both examples, the m echanical restraints imposed on the PAN during its pyrolysis would be expected

to produce multiaxial deformation. In this critical regime, a number of stresses act at the fiber-matrix inter-

face, assum ing good fiber-matrix adhesio n: First, there is an axial tensile stres s that acts

o n

the matrix; it is

a

consequence of the large matrix pyrolysis shrinka ge, and the high axial modulus an d low axial therm al expa n-

sion of the fiber. This matrix shrinkag e alsogenerates two additional matrix stresses in the plane perpendicu-

lar to the axial direction -a com pressiv e stress, which acts radially, and a tensile stress, which acts circumfer-

entially.

We tested this hypothesis by performing

a

linear elastic plane-strain thermal stress analysis for three

different local fiber-matrix com posite configurations: a clustered arrangement of three fibers and four fibers,

sketche d in Figs. 18a and b, respectively, and a matrix with free boundaries. These three cases correspond

closely to those seen in Figs. 16a-d. Th e material prope rties used for the

PAN

fiber and phenolic-resin matrix

ar e typical values obtaine d from

a

variety

of

source s. Th e mechanical prop erties of the pyrolyzing matrix are

those reported by Fitzer and B urger [59]. The therm al environm ent was a heatup from room tempe rature to

1M)O C.

In the analysis we ar e concerned only with the stresses in th e matrix in the plane p erpendicu lar to the fiber

axis, because th e tensile s tress of the matrix in the fiber directio n at any point i n the matrix is clearly m ore or

less constant at a given temperature owing to the plane-strain consideration. The stresses in the radial-tangen-

tial plane may vary signficantly, dependin g

on

their relative location

to

the fiber. At any point in the matrix,

therefore, we have a state of triaxial stress.

The development of lamellar structure in the matrix was postulated to be favored by two factors:

1)

large value of the maximum tensile stress in the plane, an d (2) a small value of th e ratio of minimum-to-maxi-

mum p rincipal stress in this same plane. Th at is. for a given value of m aximum tens ile stress in the matrix,

lamellar formation

is

favored more when the minimum-to-maximum stress ratio at any location is either small

or negative (Le., compressive). These two parameters may vary with the fiber spacing and boundary condi-

tions, e.g., constrained

or

free edge.

Figure 19a, a plot of principa l stress orientation and relative stress magnitudes, in dicates th at the maxi-

mum stress adjacent to the outside diameter of each fiber is dominate d by hoop tension with a very low leve l of

radial tensile stress; by contr ast, the maximum stress i n the center of the pocket is equal to abo ut one-third that

at the fiber surface, and the minimum (tensile) stress is now significant. From our hypothesis, these two fac-

tors will work in the direction of reduced lamellar formation relative to that at the fiber surface.

The effects of an increa se in the a/r ratio (Fig.

18)

are to decrease the maximum hoop stress

at

the fiber-

matrix bound ary and increase the stress ratio in the pocket region.

In

other words, when the th ree fibers are

mo re closely packe d, the formation of lamellar struc ture at the fiber surface is more favored t han when they

ar e loosely packed; however, within the pocke t,

it

is less favored. Sim ilar results were found for th e four-fiber

case.

We used th e model

of

Fig. 18b

to

make the calculation for the free-boundary condition occurring along

a

straight, resin-rich area; the stress in the matrix along the free boundary is primarily unidirectional. Fig.

ure 19b illustrates that the relative stress magnitude and orientation correlate with the location

of

form atio n of

lamellar structure depicted in Fig. 16c.

In

conclusion, it is seen that th e magn itude and orientatio n

of

the matrix shrinkage stresses during pyrol-

ysis,

as

estim ated by this analysis, ar e consistent with the proposed m odel for stres s orientation and

graphitization.

Much still remains t o be learned about matrix stress graphitization in

C/C:

e.g., the effects

of

fiber type,

fiber volume, matrix precursor. an d high-temperature creep deformation. Equally intriguing is

the possibility

of being able to control C/C properties by controlling the matrix orientation and graphitization behavior.

1100


8/16

SUMMARY

Carbo n-carbon com posites are an exceptional class of high-strength, low-weight refractory materials;

however, effective utilization of th e ca rbo n fiber properties requires a ppropriate selection of th e carb on

or

graphite matrix and processing conditions. Th e matrices may be derived from hydrocarbon gases, coal-tar

and petroleum pitches, and thermosetting resins, and represent a range of structures and properties. Curren t

research is beginning to elucidate how C /C com posite properties may be controlled by controlling the struc-

tures of th e matrix, both in bu lk matrix regions and, more sensitively, at th e crucial fiber-matrix inte rpha se

region.

ACKNOWLEDGMENTS

This pape r reviews selected asp ects of work funded by the Air Force Space Systems Division under C on-

tract No.

FO4701-88-C-0089,

and by the Aerospace Sponsored Research Program. The author wishes to hank

a number

of

coworkers: Rafael Zaldivar. particularly, who, as an A erospace MS and P hD Fellow, is responsi-

ble for much of the work discussed here; Dr. Dick Chang for his effective collaboration

on

portions of this

work; and Paul Adams , Jim Noblet. Ross Kobayashi. Joe Uht. Dick Brose, and C a

Su,

all

of

whom contributed

in significant ways.

REFERENCES

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22

23.

24.

25.

26.

27.

28.

Slayter. G.,

Sci.

A m . .

124

(January

1962).

Cook, J.. and Gordon, J. E., f roc . R. SOC.London

A282,5 08 (1964).

Jortner. J., Effects of Weak Intetfaces on ThermostructuralBehavior

of

C-C

Composites. Office of Naval

Research, Annual R eport, pp.

19-31

(March

1985).

Peebles. L. H., Meyer, R. A,. and Jortne r. J., In Inletfaces

in

Polymer, Ceramic and M etal Ma trir

Compos-

ites, H.

Ishida, Ed., p. 1. Elsevier

(1988).

Zaldivar, R. J . and Rellick, G.

S.,

Some Observations

on

Stress Graphitization in C-C Comp osites,

Carbon

(in press).

Rellick, G.

S.,

Chang, D. J., and Zaldivar. R. J.,

Exf.Abstz ,

20th

Carbon

Cor

(1991),

p.

404.

Zaldivar,

R.

J., Rellick, G.

S.,

and Yang, J., Erf .

Absrz, 20fh Carbon Conf. (1991),

p. 400.

Fitzer. E.,

Carbon 25, 163 (1987).

Fitzer, E., and Huttner, W., J.

fhys.

D;

ppl. fhys. 14.347 (1981).

McAllister, L. E., an d Lach man ,

W,

In Han dbooko f Composites, A. Kelly and

S.

I:Mileiko, Eds., Vol.

4,

p.

109,

Elsevier

(1983).

Hsu, S. E., and Chen, C.

I.,

In Superalloys, Supercomposites and Superceramics. p.

721,

Academic Press

(1989).

Buckley. J. D., Ceram . Bull.

67, 364 (1988).

Strife, J. R.. an d S heehan , J. E.. Ceram. Bull.

67, 369 (1988).

KO, F., Ceram. Bull. 68, 401 (1989).

Bokros, J . C., In Ch em isty and Physics of C arbon, P. L. Walk er, Jr. . Ed ., Vol. 5, p. 1, Dekker, NY (1969).

Kotlensky, W V, In Chemistry and Physics of Carbon,

F

L. Walker, Jr., Ed.,

Vol.

9, p. 173, Dekker, NY

(1973).

Pierson, H. O., and Lieberman, M.

L.,

Carbon 13, 159 (1975).

Granoff, B., Carbon 12. 405 (1974).

Loll,

P.,Delhaes. P., Pacault, A., and Pierre, A,, Carbon 15, 383 (1977).

(a) Tuinstra,

F.,

and Koenig, J.,

I .

Chem. fhys . 53. 1126 (1970); (b)

1.

Compos. Mater. 4, 49 2 (1970).

Miller, T G. Fischbach, D. B., and Macklin, J. M. rr. Abstr, 12th Carbon Conf (1975). p. 105.

Vldano, R., and Fischbach, D. B.,

Err. Abstz, 15th Carbon

Conf

(1981).

p.

408.

Katagairi.

G.,

Ishida,

H..

and Ishitani. A,, Carbon 26,

565

(1988).

Chard,

W

C., Reiswig, R. D., Levinson, L.

S.,

and Baker, 1:

D., Carbon 6, 950 (1968).

Reiswig, R. D., Levinson, L. S.,and O Rou rke, J. A,, Carbon

6, 124 (1 8).

Rodriguez-Reinoso,

E

nd Thrower, P.A,, Carbon 12.

269 (1974).

Rellick, G. S., Thrower, P.A,, an d Walker, P. L., Jr., Carbon

13, 71 (1975).

Rellick, G. S., Ert.Abstc, 20th Carbon Conf.

(1591).

p.

368.

11 1


9/16

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55

56.

57.

58.

59.

arylacetylene-Derived C-C, SAMPE1 in press).

Brown, S. C.. and Reese, H. E.

Proc. lANNAFRNTSMtg. ,

CP IA Pub l. No.496. PatrickAF B, FL(1986).

Zaldivar. R . J., Kobayashi, R.. Rellick, G . S . . and

Yang,

J. M., Carborane-Catalyzed Graphitization in

Polyarylacety lene-Derived C-C.

Carbon (in press), and Err. Ab sf c, 20th Carbon

Conj (1991),

p . 388.

Franklin, R. E..

Proc. R. SOC.London A209,

196 (1951).

Hishiyama, Y., Inagaki, M., Kimura. S., and Yamada. S.,

Carbon

12, 249 (1974).

Kimura.

S .

Tanabe,

Y.,

Takase, N., and Yasuda,

E..

Ipn.

J.

Cbem.

SOC.9,

1479 (1981).

Rellick,

G.

S., and Adams, F M.. Aerospace Sponsored Research Program unpublished results.

Kyotani, T, Sonob e, N., an d Tom ita. A, , Nature 331, 331 (1988).

Joh nson, J . W.. Rose, P.

G

an d Scott, G.. Third Conf on Indust. Carbons and Graphites, SCI, London

(1971). p. 443.

Fitzer, E., and Burger, A,, In Carbon Fibers; Their Compo sifes and Applications 134, The Plastics Inst..

London (1971).

Brooks, J.

D.,

and Taylor, J. H., Nature 206, 697 (1965).

White, J . L., In h o g . Solid State Chem.,J. 0 McCauldin and G. Somarjai, Eds., Val.9, p. 59, Pergamon

Pres s (1975).

Rellick, G. S., Carbon 28, 589 (1990).

Zimm er. J. E., an d Salvador, L. M.. U.S. Patent 4 ,554,024 (19 November 1985).

Briickmann. H., Dr.-Ing Thesis, University

of

Karlsruhe (1979).

Singer, L. S . Lewis, I. C., and Bacon, R..Proceedingsof the JA NN AF Roc ketN oule Technology Subcomil-

fee Meeting, Naval Surface W arfare Center, Silver Spring, M D (17-19 Oct ober 1989).

White. J. L., and She affer , F M.,

Carbon

27,6 97 (1989).

Zimmer. J. E., and Weitz, R. L., Carbon 26, 579 (1988).

Singer, L. S., and Lewis, R. T., f i t . Abstc, 11th Carbon Conf. (1973), p. 207.

Evangelides, J. S.. f i t . AbstK, 12th Carbon Conf. (1975), p. 93.

Jortner, J.. Appendix A in Ref. 3 above.

Jortner, J.,

In

Proc. Army Symp.

on

Solid Mechanics. AM MR C M S 76-2 (1976), p p. 81-97.

Jortner, J., Carbon

24,

603 (1986).

Buch, J. D., Structure-Property Relationships

in

Advanced Carbon Malerials, Report No.

TOR-0079(4726-04)-1,Th e Aerospace Corporation, El Segundo. CA 25 September 1979).

Manocha, L. M., and Bahl, 0. F Carbon 26, 13 (1988).

Manocha, L. M., Yasuda, E., Tanabe,

Y.,

and Kimura, S.,Carbon

26.

333 (1988).

Meyer, R. A ., and Gyetvay.S. R., In Petroleum Derived Carbons, J. Bacha,J. Newman, and J . L. White,

Eds., ACS Symp. Series 303. p. 380, AC S (1986).

Jenkins,

G.,

Kawamura. K., and Ban,

L.

L.,

Proc. R . SOC.London

A327.501(1972).

Jabloner, H., U.S. Patent No. 4,070.333 (24 January 1978).

Barry , W.T. aulin, C.. and Kobayashi, R.,Review

ofPolyaylacetyleneMatrices

for Thin-Walled Co mpos-

ites.

Report No. TR-0089(4935-06)-1.Th e Aerospace Corporation.

El

Segundo. C A 25 Septem ber 1989).

Katzman. H. A.,

Polyaylacetylene Resin Composites.

TR-IM90(5935-06)-1, T he A eros pace Cor pora tion ,

El Segundo, CA ( 2 April 1990).

Zaldivar, R. J.. Rellick, G. S., and Yang. J M.. Pr oces sing Effects on th e Mec hanic al Behavior of Poly-

1102


10/16

Figure

1. Schematic of (a) 3-D block construction and (b)

2-D

plain-weave

fabric (McAllister an d L achm an [lo]).

Figure

2. Polarized-light micrographs showing as-deposited CVI carbon microstructures

of

two specimens. Depo sition seque nce: (a) RUSL; (b) S U R L (Rellick [B] .

5v

Figure 3. Scanning electron micrograph of speci-

men after heat treatment at 25OO'C for 1 hr

[a].

1103


11/16

0.6 0.7

0.8

WEIGHT FRACTIONOF

FIBER

PLUS

FIRST

CVI

Figure

4.

Through-thickness thermal cond uctivity Figure

5.

Composite tensile modulus versus

(at KT) for composite specimens of different CVI

weight fraction o f fiber plus heat-treated CVI

128).

structures a nd processing stages [28].

FIBER

MATRIX

FIBER

2 pm

Figure 6. SEM showing highly aligned coal-tar-

pitch-derived graphite matrix in the interfilament

region of a

CIC

comp osite. Fibers are

AmwoT50

from PAN.

1104


12/16

0.2

mm

Figure

7. Optical micrograph ofcross section of 3-D CIC

compo site densified

with

both pitch and resin and heat-

treated t o 2750C. Note extensive matrix microcrack ing.

DJETHYNYLBENZENE

MONOMER

POLYARYLA CETYLENE (PAA)

PREPOLYMER

A

CARBONIZATION

GRAPHITIZATION COMPOSITES

ARBON-CARBON

looo'c)

(heat, pressure)

PAA POLYMER

COMPOSITES

Figure

8.

Chemical structure and processing

of

PAA -based composites

(Barry el

01 [48]

and Katzman

[49]).

1105


13/16

400

W

LL

z

1000 2000 3000

HEAT.TREATMENT TEMPERATURE

( C)

Y

m

0 1000 2000

3000

50

HEAT-TREATMENT TEMPERATURE

( C)

a) b)

Figure

9.

Plots of tensile (a) strength and (b) fiber modulus of undoped and B-doped

P M 5 0 C/C composites (Zaldivar et

nl. [52] ) .

00

pm

4

b)

Figure 10. Micrographs of fracture surfaces of (a)undoped an d (b) B-doped

PAA-derived

CIC

composites heat-treated to

11OOC

[52 ] .

300

E130

1000 2000

3

HTT

( C)

Figure 11. Moduli of composites

HTI (Zaldivar et a/ [7]).

1106

)o

versus


14/16

a m

Figure

13

Polarized-hght micrographs

of

unidirectlonal C/C com-

posite heat-treated to

(a)

2900'C and (b) 1100C (Rellick et

a/

[6])

0

l m

Figure

12.

Fracture sur-

face

of

E105 composite to

2750'C

H7T.

showing ma-

trix sheath tube (Zaldivar

et n l

[7]).

U

1 I rm

Figure

14.

SEM micrograph

of

PX-7 filament em -

bedded in PAA-derived carbon matrix heat-

treated t o 2750C.

2 r m a) 0.4 pm Ib)

Figure

15. T E M

bright-field images

of

C/C resin-matrix-derived unidirectional composite:

a ) ransverse and (b) longitudinal sections (Rellick and Ad am s [56]).

1107


15/16

.I

I

Figure

16.

SEMs of ion-etched unidirectional C/Cs heat-treated to 2900C

(Rellick et

a/ . [ 6 ] ) .

U

4w

Figure 17. SEM fracture surface

of

T50/SC1008 heat-treated to2900'C [ 6 ] .

1108


16/16

k

h

Figure

18. Schematic

of

the local packing arrangement of (a) three and (b) four fibers.

Shaded ar ea denotes region for which stresses are calculated [6].

e

u l

z 1.0

w

2 0.5

6

Y

0

0 0.5 1.0 1.5 2.0 2.5

r.AXIS

RELATIVE DISTANCE FROM CENTER OF FIBER

a)

r-MIS

RELATIVE DISTANCE FROM CENTER OF FIBER

b)

5

Figure 19. Com puter p lot of the directions and relative magnitudes of the matrix stresse s in

the plane

of

the fiber of various points, relative to Figs. 18a and b

[6].

1109

Documents

CAHBON ANI) GRAPHITE MATRICES IN CARBON-CARBON COMPOSITES