STP768 Eb.1415051 1 PDF

Copyright by ASTM Int'l (all rights reserved); Tue Jan 17 20:03:14 EST 2012
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.
COMPOSITES FOR
EXTREME ENVIRONMENTS
A symposium
sponsored by ASTM
Committee D-30 on
High Modulus Fibers and
Their Composites
Bal Harbour, Fla., 11 Nov. 1980
ASTM SPECIAL TECHNICAL PUBLICATION 768

N. R. Adsit, General Dynamics/Convair Division,
editor
ASTM Publication Code Number (PCN)

04-768000-33
1916 Race Street, Philadelphia, Pa. 19103

#
Copyright ® by AMERICAN SOCIETY FOR TESTING AND MATERIALS 1982
Library of Congress Catalog Card Number: 81-69769
NOTE
The Society is not responsible, as a body,
for the statements and opinions
advanced in this publication
Printed in Baltimore. Md.

April 1982
Foreword
The symposium on Composites for Extreme Environments was held on 11
Nov. 1980 in Bal Harbour, Fla. ASTM Committee D-30 on High Modulus
Fibers and Their Composites sponsored this symposium. N. R. Adsit of
General Dynamics/Convair Division served as symposium chairman and
edited this publication.
Related
ASTM Publications
Joining of Composite Materials, STP 749 (1981), 04-749000-33
Methods and Models for Predicting Fatigue Crack Growth Under Random
Loading, STP 748 (1981), 04-748000-30
Test Methods and Design Allowables for Fibrous Composites, STP 734
(1981), 04-734000-33
Fractography and Materials Science, STP 733 (1981), 04-733000-30
Fatigue of Fibrous Composite Materials, STP 723 (1981),
04-723000-33
Nondestructive Evaluation and Flaw Criticality for Composite Materials,
STP 696 (1979), 04-696000-33
Composite Materials: Testing and Design (Fifth Conference), STP 674
(1979), 04-674000-33
A Note of Appreciation
to Reviewers
This publication is made possible by the authors and, also, the unheralded
efforts of the reviewers. This body of technical experts whose dedication,
sacrifice of time and effort, and collective wisdom in reviewing the papers
must be acknowledged. The quality level of ASTM publications is a direct
function of their respected opinions. On behalf of ASTM we acknowledge
with appreciation their contribution.
ASTM Committee on Publications
Editorial Staff
Jane B. Wheeler, Managing Editor
Helen M. Hoersch, Senior Associate Editor
Helen P. Mahy, Senior Assistant Editor
Allan S. Kleinberg, Assistant Editor
Virginia M. Barishek, Assistant Editor
Contents
Introduction
PoLYiMiDES—MATERIALS FOR HIGH-TEMPERATURE ENVIRONMENTS
Environmental Effects on Graphite Fiber Reinforced PMR-IS Polyimide—

T. T. SERAFINI AND M. P. HANSON 5
V378A Polyimide Resin—A New Composite Matrix for the 1980's—

L. MCKAGUE 20
Iliermomechanical Characterization of Graphite/Polyimide Composites—

S. C. KUNZ 33
Thermophysical Properties Data on Graphite/Polyimide Composite

Materials—M. D. CAMPBELL AND D. D. BURLEIGH 54
Elastic Properties and Fracture Behavior of Graphite/Polyimide

Composites at Extreme Temperatures—D. P. GARBER, D. H.
MORRIS, AND R. A. EVERETT, JR. 73
ATMOSPHERIC AND EXTROATMOSPHERIC ENVIRONMENTS
Filament Wound Composite Thermal Isolator Structures for Cryogenic

Dewars and Instruments—E. E. MORRIS 95
Space Environmental Effects on Graphite/Epoxy Composites—c. L.

LEUNG 110
Effects of Extreme Aircraft Storage and Flight Environments on Graphite/

Epoxy—P. SHYPRYKEVICH AND W. WOLTER 118
MOISTURE ENVIRONMENTS
Environmental Exposure of Carbon/Epoxy Composite Material

Systems—R. C. GIVLER, J. W. GILLESPIE, JR., AND R. B. PIPES 137
Dynamic Tests of Graphite/Epoxy Composites in Hygrothermal

Environments—L. w. REHFIELD, R. P. BRILEY, AND S. PUTTER 148
Copyright by ASTM Int'l (all rights reserved); Tue Jan 17 20:03:

(PDVSA Los Teques) pursuant to License Agreement. No further
Influence of Quality Control Variables on Failure of Graphite/Epoxy
Under Extreme Moisture Conditions—L. L. CLEMENTS AND
p. R. LEE 161
SUMMARY
Summary 175
Index 177
Copyright by
(PDVSA
STP768-EB/Apr. 1982
Introduction
This volume presents a state-of-the-art review of "Composites for Ex-

treme Environments" by pubhshing the papers presented at a symposium by
the same name held at Bal Harbour, Fla. on 11 Nov. 1980. The papers repre-
sent the latest data and applications of polyimide with use temperatures up
to 316°C. Metal matrix composites which would be useful at extreme envi-
ronments have been excluded from this volume as a result of government re-
strictions on publication of information for these materials.
Although the authors have made considerable progress in generating the
data needed for designing with composites in extreme environments, the
papers also point up the need for more complete studies. Results presented
here can do much to guide future studies on composites.
The first of the three groups of papers covers recent work on polyimide
matrix composites. They present an excellent review of the state-of-the-art
for these materials and other information useful to all composite investi-
gators.
The second set of papers addresses applications of epoxy matrix compos-
ites and is more limited in scope, but clearly shows that composite matrix
can be used in some harsh environments. The conditions considered (for the
space shuttle and for military aircraft) are real and represent areas where the
breakthroughs on composites will be made.
The third set of papers continues the saga of what moisture does to epoxy-
matrix composites. This area is one of great concern and has been covered in
previous ASTM publications. Undoubtedly, it will be the topic of future
symposiums.
N. R. Adsit
General Dynamics/Convair Division, San
Diego, Calif. 92138; symposium chairman
and editor.
Copyright by ASTM Int'l (all rights reserved); Tue Jan 17

Copyright 1982 b y AS FM International www.astm.org
(PDVSA Los Teques) pursuant to License Agreement. No fu
Polyimides—Materials for High-
Temperature Environments
T. T. Serafini^ andM. P. Hanson^
Environmental Effects on Graphite

Fiber Reinforced PMR-15 Polyimide
REFERENCE: Serafini, T. T. and Hanson. M. P., "Environmental Effects on Graphite

Fiber Reinforced PMR-15 Polyimide," Composites for Extreme Environments, ASTM
STP 768, N. R. Adsit, Ed., American Society for Testing and Materials, 1982,
pp. 5-19.
ABSTRACT: Studies were conducted to establish the effects of thermo-oxidative and

hydrothermal exposure on the mechanical properties of T300 graphite fabric rein-
forced PMR-15 composites. The effects of hydrothermal exposure on the mechanical
properties of HTS-2 continuous graphite fiber composites were also investigated. The
thermo-oxidative stability characteristics of T300 fabric and T300 fabric/PMR-I5
composites were determined. Flexural strengths of specimens from composites in the
as-fabricated and environmentally exposed conditions were determined. The useful
lifetime of T300 fabric/PMR-15 composites in air at 316°C was found to be about
100 h. The useful lifetimes in air at 228 and 260°C were determined to be 500 and
1000 h, respectively. Absorbed moisture was found to reduce the elevated tempera-
ture properties of both the T300 fabric and HTS-2 continuous fiber composites. The
moisture effect was found to be reversible.
KEY WORDS: PMR-15 polyimide composites, graphite fabric, thermo-oxidative ex-

posure, hydrothermal exposure, mechanical properties, composite materials
Fiber reinforced polymer matrix composites, particularly those based on

epoxies, are achieving considerable acceptance as engineering materials for
the fabrication of aerospace structural components. However, the maximum
use temperature of fiber reinforced epoxies is limited to about I77°C. Until
the development of PMR polyimides,^ attempts to increase the use tempera-
ture of fiber reinforced composites by utilizing high temperature resistant
polymers as matrix materials invariably met with little success. The commer-
cial availability of prepreg materials based on the PMR polyimide, desig-
nated as PMR-15, now has made it possible to design and fabricate fiber
reinforced polymer matrix composites for use at temperatures up to 316°C,
or nearly twice the use temperature of epoxy-based composites.
' Head, Polymer Matrix Composites and materials engineer, respectively. National Aeronau-
tics and Space Administration, Lewis Research Center, Cleveland, Ohio 44135.
^Serafini, T. T., Delvigs, P., and Lightsey, G. R., "Thermally Stable Polyimides from Solu-
tions of Monomeric Reactants," Journal of Applied Polymer Science, Vol. 16, No. 905, 1972.

6 COMPOSITES FOR EXTREME ENVIRONMENTS
The effects of long-term exposure in air at elevated temperatures (thermo-

oxidative exposure) and the combined effects of absorbed moisture and ele-
vated temperatures (hydrothermal exposure) on composite properties are
two areas of vital concern to the designers of composite structures. Many
studies have been performed to determine the effects of thermo-oxidative
exposure on the properties of PMR-15 reinforced with continuous graphite
fibers. In contrast, studies to determine the effects of thermo-oxidative expo-
sure on the properties of graphite fabric reinforced PMR-15 composites have
not been reported. The area of hydrothermal effects on PMR-15 composites
has not been extensively studied. Although the effects of hydrothermal expo-
sure on the properties of PMR-15 reinforced with continuous Celion 6000
graphite fibers have been investigated,' the effects of hydrothermal exposure
on T300 graphite fabric reinforced PMR-15 have not been reported.
The purpose of this investigation was to determine the effects of thermo-
oxidative and hydrothermal environments on the properties of T300 graph-
ite fabric/PMR-i5 composites. The effects of hydrothermal exposure on the
properties of HTS-2 continuous graphite fiber composites were also deter-
mined. The effects of the thermo-oxidative and hydrothermal environments
on composite properties were established on the basis of changes to the room
temperature and elevated temperature composite flexural strengths and
moduli and interlaminar shear strengths after environmental exposure. The
thermo-oxidative stability characteristics of T300 graphite fabric and T300
graphite fabric/PMR-15 composites also were determined.
Experimental Procedure
Materials
Style 182 fabric woven from Union Carbide T300 graphite fibers and Her-
cules HTS-2 graphite fiber tows were used as reinforcing materials. The T300
fibers consisted of 3000 filaments in a one-ply construction and were sized
with an epoxy compatible sizing. The HTS-2 fibers consisted of 12 000 fila-
ments per tow.
The polyimide resin used in this investigation was the high-temperature
polyimide designated as PMR-15. The monomers used to formulate PMR-15
are shown in Fig. 1. The monomethyl ester of 5-norbornene-2,3-dicarboxylic
acid (NE) and 4,4'-methylenedianiline (MDA) were obtained from commer-
cial sources. The dimethyl ester of 3,3', 4,4'-benzophenonetetracarboxylic
acid (BTDE) was prepared as a 50 weight percent solution by refluxing a
suspension of the corresponding dianhydride in anhydrous methanol for ap-
proximately 2.75 h. The monomer stoichiometry for the PMR-15 solution
'Davis, J. G., Jr.. "High Temperature Resin Matrix Composites for Aerospace Structures."
Selected NASA Research in Composite Materials and Structures. NASA CP 2142. National
Aeronautics and Space Administration. 1980. pp. 143-182.
SERAFINI AND HANSON ON ENVIRONMENTAL EFFECTS
STRUCTURE NAME ABBREVIATION
ax' ' MONOMETHYL ESTER OF 5-NORBORNENE-

2,3-DICARBOXYLIC ACID
NE
(• -OH
0 <) O
" o "
DIMETHYL ESTER OF 3,3'.4.4'- BTDE
MiO-C Y C-OMc BENZOPHENONETETRACARBOXYLIC ACID
0 O
HO-C C-OII
" 2 ' ^ \ / "-"2 \ /'"'2 4,4'-METHYLENE0IANILINE MOA
FIG. 1—Monomers used for PMR-15 polyimide.
was 2NE/3.087MDA/2.087BTDE. The PMR-15 solution was prepared by

dissolving the monomers in a calculated amount of anhydrous methanol to
yield a 50 weight percent solution.
Composite Fabrication and Specimen Preparation

To prepare the unidirectional fiber prepreg tape, the HTS-2 tows were
wound on a drum at approximately 3 turns per centimeter, and impregnated
with a predetermined quantity of resin to provide cured laminates having a
fiber content of 55 to 60 volume percent. The prepregs were air dried to re-
duce the solvent content to approximately 10 percent prior to removal from
the drum. HTS-2 laminates having a thickness of 0.20 cm were prepared by
cutting 7.6 by 20.3 cm plies from the prepreg tape and unidirectionally stack-
ing ten plies between porous Armalon fabric in a preforming mold. The
stacked layup was imidized at 204°C for 1 h under a pressure of approxi-
mately 0.07N/cm^ Compression molding was accomplished by placing the
preform into a matched metal die that had been preheated to 232°C. Follow-
ing a dwell time of 5 min at essentially zero pressure, a mold pressure of
345N/cm^ was applied, and the mold temperature was increased to 288°C at
the rate of 5.6°C/min. Pressure and temperature were maintained for 2 h,
followed by cooling to 204°C before releasing the pressure and removing the
laminate from the die. The cured laminates were postcured in an air circulat-
ing oven in which the temperature was raised from ambient temperature to
288°C at a rate of 2.2°C/min and held at 288°C for 16 h.
The T300 graphite fabric/PMR-15 prepreg was obtained from a commer-
cial source with a resin content of approximately 40 percent by weight. Lam-
inates having thicknesses of 0.14 and 0.28 cm were fabricated using the same
procedure that was employed for the unidirectional HTS-2/PMR-15 lami-
nates. All plies for the fabric laminates were stacked with their warp yarns in
the 0-deg direction.
Composite Environmental Exposure

Coupons (approximately one third of a 7.6 by 20.3 cm laminate) were sub-
jected to either thermo-oxidative or hydrothermal exposure. All of the cou-
pons were cut from essentially void-free laminates as assessed by ultrasonic
C scan. The thermo-oxidative environments were provided by air circulating
ovens. Bleed air was metered into the ovens at a rate of 100 cmVmin. Cou-
pons were periodically removed from the ovens, and allowed to cool to room
temperature in a desiccator before reweighing to determine weight losses.
The hydrothermal environment was accomplished by supporting the lami-
nate coupons in a closed chamber above a water bath held at 82°C so that
condensate formed on the laminate surfaces. The coupons were periodically
removed from the chamber, blotted dry, and then weighed. After saturation
had been attained (no significant increase in coupon weight with increased
exposure time), the coupons were removed from the chamber and sealed in a
vapor-proof container. The conditioned coupons were cut into flexural and
short beam shear specimens using a diamond cutting blade. Flexural speci-
mens were 1.27 cm wide by 6.67 cm long. The short beam shear specimens
were 0.64 cm wide; the lengths of the specimens were selected so as to result
in a shear test span-to-thickness ratio of 4.
Composite Testing
Flexural tests conformed essentially to the ASTM Tests for Flexural Prop-
erties of Plastics and Electrical Insulating Materials [D 790-71 (1978)]. Tests
were made on a 3-point loading fixture with a variable span. Tests were per-
formed using a span-to-thickness ratio of approximately 32. The rate of cen-
ter loading for flexural testing was 0.127 cm/min. Interlaminar shear
strength tests were conducted in accordance with the ASTM Test for Appar-
ent Interlaminar Shear Strength of Parallel Fiber Composites by Short Beam
Method (D 2344-76) using a constant span-to-thickness ratio of 4. For the
elevated temperature tests, the load was applied to the specimens after the
chamber had equilibrated at the test temperature for 10 min. A limited
number of temperature spike tests were performed on moisture-saturated
specimens. For these tests, the load was applied to the specimen immediately
after the specimen was installed in the preheated test fixture.
Results and Discussion

Fiber and Composite Thermo-Oxidative Stability
Figure 2 shows the weight loss characteristics of T300 and HTS-2 graphite
fibers after isothermal exposure in air at 316°C. The data for the HTS-2 fi-
SERAFINI AND HANSON ON ENVIRONMENTAL EFFECTS 9
O T300 FABRIC
O HTS-2 FIBER
FIG. 2—Weight loss of graphite fibers exposed in air at 316°C.
bers were taken from Delvigs, Alston, and Vannucci/ The superior thermo-
oxidative stability of the HTS-2 fibers is clearly evident. After 800 h of expo-
sure in air at 316°C, the weight loss of the HTS-2 fibers was only 4.2 percent,
compared to the 50 percent weight loss exhibited by the T300 fibers after
500 h in air at 316°C. The comparatively poor elevated temperature thermo-
oxidative stability of T300 fibers very likely would be manifested in inferior
composite performance at temperatures approaching 316°C.
The weight loss behavior of T300 graphite fabric/PMR-15 composites as a
function of exposure time is shown in Fig. 3 for composites exposed in air at
260, 288, and 316°C. In contrast to the behavior of the bare unprotected
T300 fibers, which exhibited a weight loss of 50 percent after 500 h of expo-
sure at 316°C, the weight loss of theT300/PMR-15 composites was only 4.8
percent after 500 h at 316°C. The significant increase in the rate of composite
weight loss after about 400 h at 316°C is clearly evident. Similar weight loss
behavior has not been reported previously for graphite fiber reinforced
PMR-15 composites exposed at 316°C. The significantly increased weight
loss rate found in this study for the T300/PMR-15 composites is undoubt-
edly due to the limited thermo-oxidative stability of the T300 fibers at 316°C.
As expected, the T300/PMR-15 composites exhibited improved oxidative
stability at 260 and 288°C. After 1000 h at 288°C and 500 h at 260°C, the com-
posite weight losses were only 4.2 and 0.8 percent, respectively. The results of
these composite weight loss studies indicate that the useful life of
T300/PMR-15 composites at 316°C is likely less than 400 h and is at least
1000 h at 260 and 288°C.
Figure 4 shows the flexural and interlaminar shear properties retention
characteristics of O.I4-cm-thick laminates and the flexural properties reten-
tion characteristics of 0.28-cm-thick laminates. Specimens of both thick-
nesses were exposed at 316°C for various time intervals, and then tested at
"Delvigs. P., Alston, W. B., and Vannucci. R. D., "Effects of Graphite Fiber Stability on the
Properties of PMR Polyimide Composites." NASA TM 79062 and AVRADCOM TR 78-62,
National Aeronautics and Space Administration, May 1979.
10 C O M P O S I T E S FOR EXTREME E N V I R O N M E N T S
EXPOSURE
TEMPERATURES,
°C
316
260
1000
FIG. 3—Weight loss of T300 graphite fabric/PMR-15 composites exposed at elevated

temperatures.
288°C. It can be seen that during exposure at 316°C, the 288°C flexural
strengths and moduli of both laminate thicknesses decreased rapidly, whereas
the interlaminar shear strength of the 0.14-cm-thick laminate decreased
more gradually. After 620 h of exposure at 316°C, the flexural strength and
modulus retention values for the 0.14-cm-thick laminates were 33 and 40
percent, respectively, compared to a retention value of 78 percent for inter-
laminar shear strength. In view of the limited oxidative stability of the T300
fibers and T300/PMR-15 composites at 316°C, the rapid degradation of
flexural properties was expected. However, it was difficult to account for the
large difference between the retention values for the flexural and interlam-
inar shear properties only on the basis of surface degradation. To ascertain if
the difference might have resulted from structural features that were present
in the as-fabricated composites or were introduced during elevated tempera-
ture exposure, cross-sections of as-fabricated and exposed specimens were
examined metallographically.
Figure 5 shows representative photomicrographs of specimens from 0.14-
cm-thick T300 graphite fabric/PMR-15 as-fabricated laminates and from
laminates that had been exposed at 316°C. Figure 5a shows that the as-fabri-
cated laminates were defect free. In Fig. 5/», it can be seen that a few through-
the-surface-ply cracks had developed after 120 h of exposure. Figure 5c
O FLEXURAL STRENGTH
• FLEXURAL MODULUS
A INTERLAMINATE SHEAR
SOLID LINE - 0.14 cm THICK
DASHED LINE-0.28 cm THICK
120^
S 6 0 -
FIG. 4—Property retention ofTSOO graphite fabric/ PMR-15 composites exposed at 316''C and
tested at 288°C.
shows that after 624 h of exposure, the cracks were more numerous. As is
known, specimens subjected to flexural testing fail at one of the surfaces by
either a tensile or compressive failure mode, whereas shear specimens fail at
the neutral plane of the specimen. Hence, it appears that surface cracking as
well as surface degradation (weight loss) are responsible for the large differ-
ence between flexural and shear properties. More importantly, it appears
that the overall degradation which occurs at 316°C limits the useful lifetime
of T300/PMR-15 composites at 316°C to about 100 h.
Figure 6 shows the flexural and shear properties retention characteristics
for 0.14-cm-thick laminates that were exposed and tested at 288°C. As ex-
pected, the composites exhibited improved properties retention characteris-
tics after exposure at 288°C than after exposure at 316°C. For example, the
flexural and interlaminar shear strength retention values were 75 and 96 per-
cent, respectively, after 500 h at 288°C, compared to retention values of 46
and 88 percent for the same properties after 500 h at 316°C. Figure 7 shows
photomicrographs of laminate cross-sections after exposure at 288°C for 500
and 1000 h. It can be seen in Fig. 7a that only a limited number of through-
the-surface cracks had developed after 500 h of exposure at 288°C. Figure 7b
shows that the surface cracking had become extensive after 1000 h exposure.
Based on the results of the 288°C exposure studies, it may be concluded that
the useful lifetime of T300/PMR-15 composites at 288°C is limited to about
500 h.
FIG. 5—Typical photomicrographs ofTSOO graphite fabric/PMR-15 composites after various

thermal-oxidative exposures (X50).
O FLEXURAL STRENGTH
A INTERWMINAR SHEAR
0.14 cm THICK
1000
FIG. 6—Property reterttion ofT300 graphite fabric/PMR-15 composites exposed at 288°C and
tested at 288°C.
Figure 8 shows the mechanical properties retention of 0.14 and 0.28-cm-

thick T300/PMR-15 composites that were exposed at 260°C and tested at
288°C. After 1000 h of exposure at 260°C, both laminate thicknesses exhib-
ited properties retention levels of 90 percent or more. The significant in-
crease in the flexural strength of the 0.14-cm laminates as the exposure time
increased to 500 h cannot be explained. The important point to note is that
the T300/PMR-15 composites exhibited excellent retention of properties
after 1000 h of exposure at 260°C. Figure 9 shows that laminates that had
been subjected to extended exposure at 260°C were almost completely free of
any surface cracks. Thus, it may be concluded that the useful lifetime of
T300/PMR-15 composites at 260°C is at least 1000 h.
Hydrothermal Characteristics
As is known, the absorption of moisture by polymer matrix composites
adversely affects the strength and elastic properties of the composites at ele-
vated temperatures. It has been shown that the reduction in elevated temper-
ature properties is caused by the plasticizing effect of moisture on the matrix
reducing its glass transition temperature, or Tg.' In this study, unidirectional
graphite fiber (HTS-2) and graphite fabric (T300) reinforced PMR-15 com-
posites were exposed to moisture until saturation had been achieved. To as-
sess the effect of moisture on laminate properties, the flexural strengths, mod-
uli, and interlaminar shear strengths of the saturated materials were
determined at various.temperatures.
'Browning, C. E,, "The Mechanics of Elevated Temperature Property Losses in High Per-
formance Structural Epoxy Resin Matrix Materials After Exposure to High Humidity Envi-
ronments," AFML TR-76-153, Air Force Materials Laboratory, March 1977.
•<«^- -i-«-'r
^^^^mS'XM^
(a) 500 hr AT 2 K ° C.
-^•-v»?""7T«l
^tlt- f"??
'••*i^>5»ÂiSi^
(b) 1000 hr AT 2 8 ^ C.
FIG. 1—Typical photomicrographs ofTiOO graphite fabric/PMR-15 composites after various
thermal-oxidative exposures {X50).
O FLEXURAL STRENGTH
A INTERLAMINAR SHEAR
0.14 cm THICK
1201—
1000
FIG. 8—Properly retention ofTSOO graphite fabric/PMR-15 composites exposed at 260° C and
tested at 288°C.
Figure 10 shows a comparison of the properties of moisture saturated

(wet) HTS-2/PMR-15 laminates having a thickness of 0.20 cm to baseline
properties (no moisture exposure) as a function of test temperature. The wet
flexural strengths and moduli were only slightly lower than the baseline
properties up to 204°C. However, at 260 and 316°C, the wet flexural
strengths and moduli were 10 to 20 percent lower than the baseline proper-
ties. The wet interlaminar shear strengths of the laminates were 10 to 18 per-
cent lower than the baseline properties at all test temperatures.
Figure 10 also points up some of the uncertainties associated with deter-
mining realistic properties of wet laminates at elevated temperatures. The
words "realistic properties" refer to properties of the wet laminates with
minimal desorption of moisture diiring elevated temperature testing. Data,
indicated by the tailed symbols, are shown in Fig. 10 for the flexural
strengths and moduli of wet specimens that were subjected to a temperature
spike. As described earlier in the experimental procedure section of this re-
port, in a temperature-spike test, the load was applied to the specimen im-
mediately after it had been installed in the heated test fixture. Temperature
measurements have shown that the temperature of the test specimen ap-
proaches the test fixture temperature before the specimen fails. Figure 10
shows that the flexural strengths and moduli of the temperature-spiked wet
specimens are significantly lower than the flexural properties of wet speci-
mens that were equilibrated at the test temperature for 10 min prior to apply-
ing the load. These results clearly show that the absorption of moisture by
PMR-15 composites is reversible. The lower flexural properties of the tem-
perature-spiked specimens undoubtedly reflect the Tg lowering effect of
moisture on the PMR-15 matrix. The Tg of a temperature-spiked moisture-
saturated unreinforced epoxy resin was found to be approximately 100°C
lower than its Tg in the dry condition.' A corresponding reduction in the Tg
of moisture saturated PMR-15 could lower the Tg of fiber reinforced
(a)5(»hrAT26rf'C.
'MSmtî
,^«^^^^^#f^
(btlOOOhr AT26CPC.
FIG. 9—Typical photomicrographs ofTiOO graphite fabric/PMR-15 composites after various

thermal-oxidative exposures (X50).
WET
BASELINE
TAIlfD SYMBOLS
TEMP SPIKE
180)(lfl.-
160
120
o.
22x10^
<l
- 0- TT-^
a
3Q:-
50 150 250
TEMPERATURE. "C
FIG. 10—Comparison of hydroihermally exposed HTS/PMR-15 composite properties to base-

line properties.
PMR-15 composites from 336°C in the dry state to perhaps 236°C.' Thus,
the much greater reduction in flexural properties of temperature-spiked spec-
imens, especially at the higher test temperatures (for example, reductions of
49 and 25 percent, respectively, for flexural strength and modulus at 316°C,
compared to reductions of 12 and 6 percent for the same properties at 204°C)
indicate that in the temperature-spike tests, the specimens were tested above
or near the Tg of wet HTS-2/PMR-15 composites. In contrast to the behav-
ior found for flexural properties, no discernible differences were observed
between the interlaminar shear strengths of temperature spiked specimens
and temperature equilibrated specimens. A possible explanation for the con-
trasting behavior is as follows.
It is well known that the exposure of a laminate plate to moisture estab-
lishes a moisture gradient through the laminate thickness. Also, as was dis-
cussed in the Fiber and Composite Thermo-Oxidative Stability section of
this paper, shear failure of an interlaminar shear specimen occurs at the neu-
tral, or midplane of the specimen. It is possible that the moisture concentra-
' Serafini, T. T. and Delvigs, P. in Proceedings of the 1978 International Conference on Compos-
ite Materials, B. Noton, Ed., American Institute of Mining, Metallurgical, and Petroleum Engi-
neers, New York, 1978, pp. 1320-1329.
O WET
BASELINE
a TAItfO SYMBOLS
TEMP SPIKE
120x10-
80
ccz 40
J \ \ I
12xlC^
— O-
^11 1 ^_
50 150 250
TEMPERATURE, °C
FIG. 11—Comparison of hydrothermally exposed T300fabnc/PMR-I5 composite properties to

baseline properties.
tion at the neutral plane was not altered by subjecting the specimen to either
the 10 min equilibration period or to the temperature spike; hence, wet spec-
imens tested in either manner failed at the same load. It also should be re-
called that failure of a flexural specimen occurs at one of its surfaces. Appar-
ently, during temperature-spike testing of flexural specimens, the moisture
concentration of the surface plies remained at a higher level than the mois-
ture concentrations of the surface plies of specimens which had been equili-
brated for 10 min at elevated temperature. The higher moisture concentra-
tion in the surface plies of temperature-spiked specimens would have caused
them to undergo compressive failures at lower loads than was observed for
specimens which had been equilibrated at the elevated temperatures prior to
loading.
Figure 11 compares the properties of wet 0.14-cm-thick T300 fabric rein-
forced PMR-15 laminates with baseline properties as a function of test
temperature. Because the trends observed for this composite system are sim-
ilar to those observed for wet HTS-2/PMR-I5 laminates tested at elevated
temperatures, the points discussed for wet HTS-2/PMR-15 are also perti-
nent to T300/PMR-15; accordingly, they will not be repeated.
In summary, the results of this study on the effects of hydrothermal expo-
sure on the properties of graphite fiber reinforced PMR-15 composites indi-
cate that absorbed moisture causes a reduction of Tg, which is reflected in
lower properties at elevated temperatures. It must be noted, however, that
the absorption of moisture is reversible.
Conclusions
Based on the results obtained in this investigation, the following conclu-
sions can be drawn.
1. The useful lifetimes of T300 graphite fabric reinforced PMR-15 compos-
ites exposed in air at 316 and 288°C are approximately 100 and 500 h, respec-
tively. The useful lifetime in air at 260°C is at least 1000 h.
2. Absorbed moisture reduces the elevated temperature properties of
graphite fiber reinforced PMR-15 composites.
3. Temperature-spike testing of moisture saturated composites provides a
more realistic assessment of the effect of absorbed moisture on the elevated
temperature properties of composites.
4. The absorption of moisture by graphite fiber reinforced PMR-15 com-
posites is reversible.
L. McKague^
V378A Polyimide Resin—A New

Composite Matrix for the 1980's
REFERENCE: McKague, L., "V378A Polyimide Resin—A New Composite Matrix for
the 1980's," Composites for Extreme Environments, ASTM STP 768, N. R. Adsit, Ed.,
American Society for Testing and Materials, 1982, pp. 20-32.
ABSTRACT: During 1979, General Dynamics conducted an extensive screening pro-

gram to evaluate new graphite composite materials. V378A, a new modified bis-
maleimide resin developed by U.S. Polymeric, has met the predetermined screening
criteria for harsh aircraft environments [recurrent service to 177°C (350°F) following
prolonged exposure to 75 percent mean relative humidity]. During the screening
procedure, environmental, mechanical, and processing requirements were examined.
In this report, these criteria, in addition to the V378A results, are reviewed and
compared with results for 5208 resin, a widely used epoxy resin. The comparison
shows that V378A resin represents an important advancement in composite properties
and processibility with widespread application in the manufacture of composite
materials.
KEY WORDS: composites, graphite/polyimide. flex properties, tension properties,

environmental effects, thick laminate processing
High-temperature epoxy resins have been investigated widely and used in

composite materials for more than a decade. Currently, epoxy matrix com-
posites are being used extensively for primary aircraft structures, and they
are completely adequate for these applications.
For several years, however. General Dynamics has been seeking a material
with improved characteristics in two fundamental areas: improved processi-
bility into large-area thick laminates—tail and wing skins with thicknesses
ranging from 50 to 150 plies; and greater environmental margin—higher
glass transition temperature (Tg) after saturation in 75 percent relative
humidity.
Graphite/epoxy materials generally are considered to be the most easily
processed class of composite materials. Epoxies evolve no reaction volatiles,
and today's hot-melt systems contain no solvents. Nevertheless, porosity is
found commonly in thick graphite/epoxy laminates. Unfortunately, the
porosity is often bad enough to cause scrappage of a laminate.
'Engineering specialist senior. General Dynamics, Fort Worth, Tex. 76101.
20

Copyright® 1982 b y AS FM International www.astm.org
McKAGUE ON NEW COMPOSITE MATRIX 21
Extensive work on this problem has been conducted at Genera! Dynamics

with the support of Narmco. This work has indicated that atmospheric mois-
ture absorbed by the prepreg is a significant causative factor of thick lami-
nate porosity.^
Atmospheric moisture absorbed by cured epoxy laminates also harms
their mechanical properties at elevated temperatures by lowering Tg. The Tg
of epoxy saturated in a mean relative humidity of 75 to 80 percent is lowered
from 246°C (475°F) to about 125 to I30°C (260 to 270°F).'''' While these
conditions still exceed current requirements, even a small increase in super-
sonic performance could leave some new aircraft without adequate thermal
margin for use of graphite/epoxy structures.
Almost four years ago, General Dynamics began seeking a material with
improved processibility and environmental margin. For the first two years,
this involved an interactive program between General Dynamics and several
material suppliers to develop and evaluate new resins. In early 1979, General
Dynamics began a second phase of work, an intensified Independent Re-
search and Development (IRAD) effort to screen and evaluate graphite pre-
preg systems. The general criteria required a material with mechanical
properties like those of graphite/epoxy at room temperature, but with im-
proved values at 177°C (350°F) after saturating in 75 percent relative humid-
ity. Improved processibility also was required.
Eight materials were tested against the screening criteria. Of these mate-
rials, six were new formulations. When screening was completed, one mate-
rial, a bis-maleimide modification developed by U.S. Polymeric, met every
environmental and mechanical screening requirement. Improved processing
characteristics of the new resin also have been verified.
This new resin has been designated by U.S. Polymeric as V378A,' a major
advancement in composite matrix materials for advanced aircraft. This
paper describes the screening program, its results and some important thick-
laminate processibility characteristics of V378A.
Screening Program
For screening evaluations, eight graphite composite prepreg materials
were obtained from four companies. The following materials were obtained:
F-178, a bis-maleimide; CPI-2272, a bis-maleimide; X5231, a modified bis-
^Hinrichs, R. and Thuen, J., Society of Aerospace Material and Process Engineers Journal.
Vol. 15, No. 6, Nov./Dec. 1979, pp. 12-21.
'Browning, C. E., Husman, G. E., and Whitney, J. M. in Composite Materials: Testing and
Design (Fourth Conference), ASTMSTP 617, American Society for Testing and Materials, 1977,
pp. 2-20.
^McKague, E. L., Reynolds, J. D., and Halkias, J. E., Journal of Applied Polymer Science,
Vol. 22, 1978, pp. 1643-1654.
'Street, S. W., The 1980's—Payoff Decade for Advanced Materials, Vol. 25, Society for the
Advancement of Material and Processes Engineering, 1980, pp. 366-375.
tnaleimide; X108-43A, a modified bis-maleimide; X108-55B, a modified bis-

maleimide; V378A, a modified bis-maleimide; XE-716-3, a modified epoxy;
and XE-716-5, a modified epoxy. V378A, XE-716-3, and XE-716-5 were
reinforced with T300-6K, and all others were reinforced with Celion 6000.
The XE-716-5 material was dropped from the screening program relatively
early, and a second submittal of X108-55B with T300 fiber reinforcement
took its place.
Each of the materials was tested according to a plan aimed at evaluating
environmental and mechanical characteristics. Initially, processing aspects
were investigated only to the extent necessary to make good-quality lami-
nates for the evaluation tests. More extensive investigations of processibility
were planned for each material passing the property screening requirements.
The properties screening plan is illustrated in Fig. 1. The plan's first step
involved establishing cure and bleeding procedures. For each material, the
supplier's recommendations were followed, with changes made only as nec-
essary to achieve a porosity-free laminate with a cured resin content of 27 to
30 percent by weight. Almost every change involved only simple adjustments
in the quantity of bleeder material.
Two types of cure cycles satisfied all screened materials. One was a
straight-up cycle involving heating to 179°C (355°F) without an intermediate
temperature hold. The other cycle involved an intermediate temperature
hold. Figure 2 identifies the cycle details for the various materials.
The plan's second step—development of a postcure cycle—entailed estab-
lishing the highest glass transition possible within six hours without causing
matrix degradation or cracking. Table 1 identifies the postcure cycle used for
each material.
After a good-quality, well-cured laminate was made for each material,
preparations were undertaken for an evaluative screening process. Specific
ENVIRONMENTAL EFFECTS
• Thtrmal Aginfl
• ThtrmalSptkas
• Glut Trantttion
FLEX a SMEAR
POSTCURE CYCLE
145 TENSION
STRAIN ENERGY RELEASE RATE
FIG. 1—Properties screening plan.
OAPPLV VACUUM
OAPPLY.6MPaAN0
VENT VACUUM
HEATfNG RATE • 3>C/MII)
HOURS HOURS
FIG. 2—Cure cycles for screened materials.
criteria were established for each step of the evaluation, as summarized in

Table 2.
Concurrent evaluations of environmental resistance and mechanical prop-
erties were conducted. The environmental resistance evaluations involved
tests of three characteristics: Tg, thermal aging, and thermal spike effects.
For the first characteristic, a target for minimum dry Tg was set at 288°C
(550°F). After saturating in 75 percent relative humidity, wet Tg was tar-
geted to exceed 177°C (350°F).
The requirement for the second characteristic, thermal aging, was to verify
that the high-temperature portion of a lifelong supersonic service spectrum
would not physically degrade the matrix. The requirement specified that no
cracks be formed in (0/±45/0), laminates after 25 h at 171°C (340°F), fol-
lowed by 264 h at 135°C (275°F). Inspection was by means of X60
photomicrographs.
Tests concerning the third environmental characteristic, thermal spikes,
also were conducted. The objective of these tests was to show that ten twice-
weekly thermal spikes to 177°C (350°F) would not cause the material to gain
TABLE 1—Postcure cycles for screened materials.
Resin Postcure Cycle
F-178 6 h at 246°C (475°F)

CPI-2272 1.5 h at 204°C (400°F)
X5231 4 h at 232'=C (450°F)
XI08-43 A 1 h at 274°C (525°F)
X108-55B 4 h at 246°C (475°F) and 274°C (525°F)
V378A 4 h at 246°C (475''F) and 288°C (SSCF)
XE-716-3 1 h at 232°C (450''F)
XE-716-5 1 h at 218°C (425°F)
TABLE 2—General and specific screening criteria.
Processibility
Self-adhesion tack
Cure with present equipment (maximum 190°C, 0.7 MPa)
Seven days out-life at 27''C
Unrestrained postcure (maximum 343°C, 6 h) resulting in crack-free (0/±45/0), laminates
General Properties
Room-temperature mechanical propertiesrties like epoxy
lative humidity (RH
177°C service capability after 75% relative (RH)
Specific Target Criteria
Environmental
Thermal aging; no cracks formed in (0/±45/0), laminates after 25 h at 171°C plus 264 h at
135°C
Thermal spikes: maximum weight gain of 0.15% when exposed intermittently to 75% RH and
177°C spikes
Glass transition temperature (Tg):
minimum dry = 288''C
minimum wet (75% RH) = 177°C
Flexure and shear
Dry 0° flexure
minimum RT flex strength = 1724 MPa
minimum strength retention at 177°C = 80% of 24°C value
minimum RT modulus = 127 GPa
Short beam shear
minimum dry at 24°C = 90 MPa
minimum dry at 177''C = 69 MPa
±45 tension
Minimum initial wet (75% RH) modulus at 177°C = 10.3 GPa
Minimum wet (75% RH) ultimate stress at 177''C = 73 MPa
Minimum wet (75% RH) ultimate strain at 177°C = 2.3%
Critical strain energy release rate (G,c): minimum value of 87 5 J / m ' at 24°C
over 0.15 percent more weight than unspiked controls in the same 75 percent
relative humidity chamber. Meeting this crite:rion would demonstrate a
material's resistance to matrix cracking caused by moisture and thermal
transients.
To help verify that the dry mechanical properties of each candidate were
like those of graphite/epoxy, tests of 0-deg flexure and short beam shear
were conducted. The minimum requirement for room temperature (RT) 0-
deg flexure strength was set at 1724 MPa (250 ksi), with 80 percent of the RT
value maintained at 177°C (350°F). The minimum flexural modulus was set
at 127 GPa (18.5 Msi). Minimum requirements for short beam shear strength
were 90 MPa (13 ksi) at RT and 69 MPa (10 ksi) at 177°C (350°F).
To evaluate the combined effects of moisture and temperature on resin-
dominated mechanical behavior, tensile stress-strain behavior of ±45 cou-
pons was determined at 177°C (350°F) following saturation in 75 percent
relative humidity. Minimum requirements were an initial modulus of 10.3
GPa (1.5 Msi), an ultimate strength of 73 MPa (10.6 ksi), and an ultimate
strain of 2.3 percent.
Critical strain energy release rate (G/c) also was determined to assess each
material's resistance to delamination and flaw growth. The minimum re-
quirement was set at 87.5 J/m^ (0.5 in -Ib/in^), which represents approxi-
mately 90 percent of the value typical of T300/5208 graphite/epoxy.
Screening Results
All screening tests were conducted in the General Dynamics-Fort Worth
Division's engineering chemistry laboratory. These tests showed that proces-
sibility of all screening materials was adequate to make thin laminates (8 to
16 plies) of acceptable quality for evaluation. The laminates were all free of
porosity and cracks, and cured resin contents were all in the 27 to 30 percent
range.
All materials but two were successfully postcured and thermally aged
without forming matrix cracks. However, during postcure, X108-43A and
CPI-2272 formed matrix cracks easily visible at X60. The CPI-2272 material
was particularly unstable, forming many matrix cracks during the thermal
aging tests. In fact, this tendency to crack was a major factor in the estab-
lishment of the material's postcure cycle. The cracks in X108-43 A caused by
postcuring did not appear to worsen during thermal aging.
Three materials, XE-716-3, XE-716-5, and X5231, failed to meet the ther-
mal spike requirement. Specimens of these three materials gained over 0.15
percent more weight due to ten thermal spikes in 177°C (350°F) oil than un-
spiked specimens in the same 75 percent relative humidity chamber. All
other materials gained only 0.08 to 0.12 percent more weight in 75 percent rel-
ative humidity after ten thermal spikes.
Dry Tg for all materials but two ranged from 260 to 270°C (500 to 520°F).
V378A and the XI08-43A materials each exceeded the 288°C (550°F) target
by 5 to 10°C (9 to 18°F). Testing for wet Tg (saturated in 75 percent relative
humidity prior to test) was suspended because the moisture diffusion rate of
the polyimides was much more rapid than that of epoxies, due to higher
permeability. This higher diffusion rate caused too much specimen drying
during the thermomechanical analysis (TMA) testing.
Mechanical evaluations began with tests of 0-deg flexure and short beam
shear properties. Each material met or exceeded the RT 0-deg-flex-strength
target of 1724 MPa (250 ksi). All but XE-716-5 met or exceeded the min-
imum modulus requirement. In general, the polyimides showed good reten-
tion of strength at 177°C (350°F). All but the X108-55B polyimide met or ex-
ceeded the 80 percent retention requirement. The epoxies did not meet the
retention requirement. Short beam shear strength requirements were met by
all materials except X108-43A and X108-55B with C6000 fibers.
Only one material, V378A, met every requirement for ±45 tension behav-
ior. Figure 3 shows bar graphs of the I77°C (350°F) results for each material
in terms of initial modulus, ultimate strength, and ultimate strain. For com-
—'
[^TARGET
1
-
i'
< ,
i
_
s
t
_
1i 5
E
i
1 -
I ULTIMATE STRENGTH I
H ULTIMATE STRAIN V
FIG. 3—±45 initial tension property comparisons at liyc {350°F) after saturating in 75 per-
cent relative humidity.
FIG. A—Stress-strain comparisons ofT300-6K/V37SA and T300-3K/5208 at lirC (,350"?).

75 percent relative humidity saturated.
ISO
12S
IM
J . _1
! «- -
t5"
SO - -
T
2t| 'B"
<
1
s
a 1 a 1I a a
E Sl l l
FIG. 5—Comparison of critical strain energy release rates.
parison, the figure also shows the performance of the 5208 epoxy resin. Fig-
ure 4 shows a comparison of the wet 177°C (350°F) stress-strain behavior of
T300-6K/V378A and T300-3K/5208, illustrating the degree of environmental
improvement achieved by V378A. These tests, all performed after saturating
in 75 percent relative humidity, involved rapid heating of the gage section
with hot air blowers. Weight measurements conducted before and after heat-
ing verified that this procedure helped to assure retention of moisture during
the test. Specimens were coupons 2.54 by 22.9 cm (1 by 9 in.) loaded at 2.5
mm/min (0.1 in./min), and each had one transverse and two axial strain
gages. Each specimen was cut from a (±45)2S laminate.
The target for Gic of 0-deg lamina was met only by V378A and XE-716-3.
However, tests of C6000/X108-55B initially gave high but erratic results.
Two additional panels of this material were made and tested, and each pro-
duced consistent but below-target results. Figure 5 summarizes the Gu test
results, which were obtained with a nontapered, double cantilevered beam
test procedure developed at the Fort Worth Division.
A review of all of the screening results was conducted. Of the eight mate-
rials evaluated, it was determined that only T300-6K/V378A graphite/
polyimide material met all of the established criteria. Each of the other mate-
rials failed at least two of the screening requirements.
Thick Laminate Processibility

Once the V378A material passed environmental and mechanical screen-
ing, tests of thick laminate processibility were begun. Early tests were con-
ducted using prepreg with resin content of about 36 percent by weight. At-
tempts to bleed excess resin from 60-ply laminates were entirely unsuccessful.
Each cured laminate had a resin content gradient through the thickness that
typically ranged from 26 to 28 percent near the bag surface, to almost 36 per-
cent at the center. Putting half of the bleeder at the tool surface did not solve
80 too 120 140 160 110

TEWKRATURE l«CI
FIG. 6—Viscosity comparison ofV378A and 5208 resins.
the problem of high resin at the center, nor did reductions in laminate thick-
ness to 40 plies.
To investigate the cause of this flow problem. General Dynamics re-
quested Rheometric test data on V378A from U.S. Polymeric. As illustrated
in Fig. 6, the data show that the viscosity of V378A is much higher than the
normal 5208 epoxy purchased by General Dynamics. In addition, a short
flow period was indicated by Fisher-Johns "gel time" data, as shown in Fig.
7. To combat the effects of these characteristics, a hold in the cure cycle was
established near and on the low-temperature side of the point of minimum
viscosity, in order to allow the longest time for resin flow. However, this
cycle did not result in sufficient bleeding, even with 40-ply laminates.
General Dynamics approached a solution to the bleeding problem in two
different ways. Experiments were conducted to determine the feasibility of
100
l«
1
<9
10
in 110
TENKRATUDE I t l
FIG. 7—Gel time versus temperature for V378A resin.
prebleeding or incrementally bleeding the material in 20-ply stacks. In addi-

tion, we requested U.S. Polymeric to develop a genuine net resin prepreg (27
to 30 percent by weight). Both approaches were successful. Problems asso-
ciated with prebleeding and incremental bleeding led to the decision to work
with the net resin prepreg, which offers several significant advantages.
Because of V378A's viscosity and thickening rate characteristics, a
cure cycle was evolved which incorporates a low-temperature dwell
(80°C = 175°F) of 2 h. The autoclave cycle is completed with 4 h at 179°C
(355°F). Pressure (586 kPa = 85 psi) is applied before heating begins to as-
sure compaction, which occurs during the 80°C (176°F) dwell. Separator fab-
ric, perforated film, and dry glass vent materials are used in bagging the net
resin material, but no bleeder material as such is used. Flexural tests con-
firmed that the properties developed in this cycle are equivalent to the origi-
nal straight-up cure cycle.
After the V378A cure cycle, illustrated in Fig. 8, a freestanding, oven post-
cure takes place for 4 h at 245°C (475°F) followed by 1 h at 288°C (550°F).
Application of pressure prior to heating greatly simplifies the cure cycle.
With epoxies, debate continues about the correct time to apply pressure dur-
ing the cure cycle. Development and use of adaptive controls have been
discussed widely to allow batch-to-batch and part-to-part variances in pres-
sure application. All of this debate stems from the tendency to form voids/
porosity in thick graphite/epoxy laminates.
From the beginning of our work with V378A, thick laminates have been
made consistently without voids/porosity. Although the application of pres-
sure prior to heating is a valuable cure cycle simplification, it does not ex-
plain why thick V378A laminates contain no voids/porosity.
Porosity is believed generally to result from the formation of gas bubbles
around various nucleation sites. Gas bubbles can nucleate and grow only
TOSTCURE 4 MRS AT 2460C + 1 HR AT 288°C
150 /
/
100 / OAPPLV VACUUM \

OAI>PLr0.6MPa
/ • VENT VACUUM
• REMOVE PRESSURE
HEATING RATE »1.7°cyMIII
/ \
/ \
0 1
! S 1 > 1
FIG. %—Cure cycle for thick V378A laminates.
when the pressure inside the bubble exceeds the sum of the autoclave pres-
sure and the restraint provided by the resin-to-gas surface tension;* that is
P ^ 7 : 5 ) > Po + 6ag,iT,Xydr, (1)
where
Pg = gas bubble pressure,
Po = autoclave pressure,
Ogi = resin-gas surface tension,
5max = maximum nucleation site size,
T = temperature,
S = gas concentration, and
X = degree of resin advancement.
Based upon Eq 1, the low-temperature dwell in the V378A cure cycle may be
the major factor in avoiding bubble growth. Because the 80°C (176°F) dwell
is 50°C (122°F) lower than the epoxy-cycle dwell temperature, the driving
force to enlarge a bubble is reduced significantly.
FIG. 9—Machine-laid laminate.
'Carter, H. G., General Dynamics, Fort Worth, Tex., private communication, Feb. 1980.
Resin viscosity may be another important factor in avoiding bubble

growth. As was shown in Fig. 6, V378A's viscosity is much greater than that
of epoxy. Since surface tension generally increases <vith viscosity, the more
viscous V378A provides greater restraint to bubble growth.
Extensive experience with graphite/epoxy has shown that the occurrence
of voids/porosity has a strong statistical correlation to high humidity during
layup. With T300-6K/V378A, thick laminates consistently have been made
without voids/porosity even when laid up in relative humidities as high as 73
percent. Two such laminates, developed by the Manufacturing Technology
Department at the Fort Worth Division, were hand-laid skins for a single-
cell wing root box. The skins were approximately 40.6 by 147 cm (16 by 58
in.) and were 68 and 89 plies thick.
Two large laminates also have been successfully machine-laid using net-
resin T300-6K/V378A prepreg. Laminate thicknesses ranged from 5 to 52
plies. Ultrasonic inspection and destructive examinations showed both lami-
nates to be free of voids/porosity. Figure 9 is a photograph of one of the
laminates.
Even with net resin prepreg (resin contents as low as 27 percent), V378A
has enough tack to make machine-laying feasible. However, tack is lost
rather quickly upon exposure to air. Because of this tack loss, the machine-
laying and hand-laying operations appear to constitute the greatest challenge
to the manufacturing effort.
Concluding Remarks
U.S. Polymeric's T300-6K/V378A, a graphite-fiber-reinforced modified
bis-maleimide, has met all of the screening criteria established by General
Dynamics for an improved composite material in advanced aircraft applica-
tions. When both V378A and epoxy have been saturated in 75 percent rela-
tive humidity, V378A has better retention of its mechanical properties upon
exposure to 177°C (350°F) heat. The dry Tg of V378A measured by ther-
momechanical analysis (TMA) is approximately 300°C (570°F) when post-
cured 4 h at 245°C (475°F), followed by another hour at 288°C (550°F). De-
spite this high Tg, the material exhibits a ±45 tension strain-to-failure,
comparable to that of epoxies.
Of equal or greater importance, the T300-6K/V378A material offers a
significant improvement over epoxies in terms of thick-laminate processibil-
ity. Laminates in the 60 to 90-ply thickness range can be cured without voids/
porosity even when laid up in high humidity. This provides major cost bene-
fits through reduction of component rejections and scrappage. Furthermore,
V378A has emerged as the first practical true-net-resin graphite/thermoset
prepreg, thereby providing further manufacturing cost savings through the
avoidance of bleeder materials and associated labor charges.
Because of these processing advantages and improved properties, T300-
6K/V378A represents a significant advancement in composite materials for

the advanced aircraft manufacturing requirements of the 1980's.
A cknowledgments
The materials screening and processing investigations reported in this
paper were funded entirely by General Dynamics as an Independent Re-
search and Development (IRAD) task. The author acknowledges the work
and ideas of J. D. Reynolds, M. S. Williams, and J. H. Fruit with thanks and
appreciation.
S. C. Kmz^
Thermomechanical Characterization
of Graphlte/Polyimide Composites
REFERENCE: Kunz, S. C , "Thermomechanical Characterization of Graphite/Polyimide

Composites," Composites for Extreme Environments, ASTMSTP 768, N. R. Adsit, Ed.,
ABSTRACT: The stiffness, strength, and shear properties of three polyimide resins
(NR-150B2, PMR-15, and CPl-2237) combined with three different moduli graphite
fibers (C-6000, F-5A, and GY-70) have been determined in the temperature range 20
to 371°C (68 to 700°F). Flexural test results from the various unidirectional composites
show that stiffness retention with increasing temperature is affected only by the thermal
integrity of the polyimide matrix. No loss in modulus occurs up to 316°C (600°F) for
the PMR-15 and CPl-2237 based composites [T, = 377°C (710°F)] or to 260°C (500°F)
for the NR-150B2 based material [Tg = 349°C (660°F)], with any of the three fibers. In
contrast, both flexure and shear strengths show fiber-dependent behavior, irrespective
of the matrix resin, with increasing temperature. The higher modulus fiber composites
(F-5A, GY-70) undergo little strength change up to 343°C (650°F). Composite strengths
of the lower modulus fibers (C-6000), however, degrade by as much as 50 percent over
the same temperature range. Thermal-oxidative stability of the various graphite fibers,
and its effect on interfacial strength degradation, are considered primary causes for the
fiber-type dominated strength behavior. In general, strength retention appears directly
related to degree of graphitization (modulus) of the fibers. The accumulated mechani-
cal property data, some previously unknown, are correlated with microstructural fea-
tures such as fiber-matrix adhesion, porosity, and processing defects. This study of a
family of graphite/polyimide composites provides a basis for selecting a high tempera-
ture, structural weight-saving material.
KEY WORDS: graphite/polyimide composites, thermomechanical properties, graphite

fibers, thermal-oxidative stability, thermal degradation, fiber-matrix adhesion, porosity
Replacement of structural metal with a high-performance composite in an

elevated temperature application may be limited by mechanical and physical
property retention of the fibers and matrix. Developments in resin chemistry
and processing techniques within the last decade have promoted graphite
fiber reinforced polyimide resins to being realistic candidate materials for
use in the 177 to 371°C (350 to 700°F) range. For example, glass transition
temperatures (Tg) above 427°C (800°F) have been achieved in a graphite/
PMR-15 polyimide system [7],^ and qualification of a composite jet engine
'Member technical staff, Sandia National Laboratories, Albuquerque, N. Mex. 87185.

^The italic numbers in brackets refer to the list of references appended to this paper.
33

nozzle flap constructed using NR-150B2, a lower Tg [343°C (650°F)] polyimide

resin, has progressed to inflight evaluation [2]. Although the primary attrac-
tion of polyimide resin-based composites is their ability to withstand much
higher temperatures than the more conventional reinforced epoxy resins,
their ability to retain ambient-temperature mechanical properties during
both short and prolonged elevated temperature exposure has emerged as
strongly dependent on the graphite fiber reinforcement. Specifically, the
thermo-oxidative stability of graphite fibers, which varies with the fiber type
(for example, with modulus, degree of graphitization), largely determines the
high temperature stability of the composites [3-7].
Fiber-dependent high temperature behavior has been observed in a variety
of graphite-polyimide systems [1,3-7,9-13] and is supported also by studies
of carbon fibers themselves [3,5,8]. However, the diversity of composite sys-
tems examined, often under varying test conditions, makes difficult an eval-
uation of the trends and mechanisms which link fiber and composite proper-
ties. This paper attempts such an evaluation by comparing mechanical
property data obtained from graphite/polyimide composites selected to con-
tain (a) like fibers in different polyimide resins and (b) different fiber types in
the same resin. In this way, the extent to which the various types of fibers de-
grade or affect property retention can be systematically deduced. Matrix be-
havior at high temperature and the role of porosity and processing defects
also can be screened. Fracture surface characterization, in particular, yields
information on composite failure modes: as revealed, for example, by the na-
ture of the fiber-matrix interface. Such methodical comparison of composite
properties with microstructural features is designed to provide a basis for fu-
ture selection of specialized composite materials.
Materials
Three types of graphite fibers with a range of modulus (E) values were
selected to produce composite laminates: Celion C-6000 [E = 234
GPa(34 X 10* psi); Celanese Structural Composites], Fortafil F-5A
[£ = 331 GPa(48 X 10* psi); Great Lakes Carbon Corp.], and Celion GY-70
[£ = 517 GPa(75 X 10* psi); Celanese]. The three polyimides chosen as ma-
trices were NR-150B2 (Du Pont Co.) and two versions of PMR resins, PMR-15
(Composites Horizons) and CPI-2237 (Ferro Corp.), prepared by the indi-
vidual composite prepreggers based on the original NASA-Lewis formulation
[10]. Thesd fibers and resins were incorporated into prepreg material, and
subsequently into unidirectional composite laminates by commercial manu-
facturers in the combinations shown in Table 1. In the instances where fiber
sizing was necessary to ease prepreg fabrication (for example, C-6000/
NR-150B2), the only sizing used was the matrix polyimide resin itself; in no
case were fibers sized with epoxy resin.
All eight graphite/polyimide systems in Table 1 were compression molded.
Although the exact cure cycle for the NR-150B2 material manufactured by
KUNZ ON GRAPHITE/POLYIMIDE COMPOSITES 35
m
"O <N 00 O Tf VO O
o O o —^ O O O
-H -H -H +1 -H +1 -H
^^ 0^ ? V~i r^ r-; OO 0\
fn f*i Tt rn rS TT •—
r^ • * (N rt O ro rn o
ri ^ O —' rs O —: ^
-H -H -H 41 +1 +1 -H HH
—1 oo o fS oo ^ 1^ ro
u o —J wS (3\ vS VO rn P^
1
<N
o o o oo
3 GO OO
m
oo
o o o o o o o o o o
0
-H »0 <N CS - ^ —
O O
OS "O 0 \ t
O
- ^ —.
0
s
<1 o> <n pvj ^ ON vo s o -H r-> f n o \ ^
m ( S ^ »n
00 •<1- ON NO
r s »n r-( s o f s • n <N <n (N »n c s >ri (N »n (N W1
p - 00 »n ON ^ ^ Q r - o \ - ^ -H Q NO 00 ON v ^
f*^ m "O 00 r f »n 00 —« r - -H r ^ S t ^ O 1^ - ^
f*! r ^ f*^ t ^ r ^ r ^ m r- m r-
f*^ \ 0 f*^ \ 0 f/l \ 0
I
0 o o o
w w S X X X
I u
4) -—V « ^—^
•^
00 >
• ••« •
00 >
•2 a -2 5 -s
,.
V}
U
M
V
M
M
U
M
(A
U
tfl 0 o o
o c -a c -o a a. o. a
u c4 u a c E E E
E
I o
O
00
u "O 4> "O
f- 5 I- S W
u V] u 1/3
O
c
U
0
c
U
\
o
c
U
\
o
c O
d.
0 s-
o
u
•
o 0 0 o
a, 1 0 '1 0 .§ N
N u
2
u •c • 0c
(-a \ := ^ .-= o 0 o
X X X
s-i & l = >
asdss
u da d 0
u
0
a
o
Q. a
O
o
U
0
1
£ U J3 U §
£ £ u U
E
o
E
o
u
E
o
U LJU i
b
>-
< "r
V >-
6 o u o
< •^ • o o •
o
a
E
o
u
OL.
Hamilton Standard is proprietary, that used by Fiberite Corp. as an accep-

tance criterion for prepreg materials supplied to Hamilton Standard can be
generalized as follows: full vacuum staging for 90 min at 130°C (266°F),
pressing at 4.13 MPa (600 psi) for 20 h during slow heat-up to 316°C (600°F),
and subsequent cooling under pressure at room temperature. This is fol-
lowed by a postcure, with the part unrestrained and in air, consisting of:
heating for 45 min from room temperature to I66°C (330°F), a 42 h heat-up
to 343°C (650°F), and a 24 h hold at 343°C (650°F). In comparison, the
procedure used by Composites Horizons for GY-70/NR-150B2 laminates is
30 min staging at 149°C (300°F), followed by a 30 min compression molding
under 6.89 MPa (1000 psi) pressure at 427°C (800°F), with no postcure.
Fabrication of PMR-15 laminates at Composites Horizons consisted of
vacuum bag staging for 30 min in a low vacuum at 204°C (400°F), compres-
sion molding for 2 h under 6.89 MPa (1000 psi) pressure at 288°C (550°F),
postcuring for 16 h at 316°C (600°F), and finally, subjecting the part to a 16 h
thermal treatment at 371°C (700°F). Similarly, the recommended Ferro Corp.
procedure for CPI-2237 composites is to precure (stage) for 1 h at 204°C
(400°F) in a low vacuum, followed by a slow heat-up to 232°C (450°F) under
0.21 MPa (30 psi) pressure, and then cure by applying full pressure [1.38 to
3.45 MPa (200 to 500 psi)] for 2 h at 329°C (625°F).
Glass transition temperatures (7^) of the eight composite systems were de-
termined by thermomechanical analysis (TMA). Testing was done in a nitro-
gen atmosphere with a heating rate of 20°C/min (68°F/min) using a 5 g (0.18
oz) probe in an expansion mode. The single Tg value shown for each compos-
ite type in Table I is accompanied by the temperature range (AT) over which
the transition occurred.
Fiber and void contents of the composites, indicated as volume fractions,
are listed also in Table 1. Measurements of fiber weight in each composite
were made using a hot sulfuric acid chemical digestion technique [14]. Values
of fiber volume fraction were then calculated using measured specimen vol-
umes together with fiber densities of 1.76 g/cm^ (0.063 lb/in.') for C-6000,
1.80 g/cm' (0.065 lb/in.') for F-5A, and 1.96 g/cm' (0.071 lb/in.') for GY-70.
A density of 1.4 g/cm' (0.051 lb/in.') for the pure resin was used to calculate
the resin volume fraction. The void content was taken as unity minus the fiber
and resin contents. Fourdeteminations were made for each type of composite.
Testing
All of the composite laminates had unidirectional (0°) fiber orientation
and were obtained from the manufacturers as 2.54 mm (0.1 in.) thick panels.
The desired mechanical properties to be measured were: fiexural modulus, E,
fiexural strength, a/, and apparent interlaminar shear strength, T. Test spec-
imens were machined from the panels in accordance with the ASTM Tests for
Fiexural Properties of Plastics and Electrical Insulating Materials [D 790-71
(1978)] for the flexural modulus and flexural strength determinations, and
with the ASTM Test for Apparent Interlaminar Shear Strength of Parallel
Fiber Composites by Short Beam Method (D 2344-76) for the shear strength.
Span-to-depth ratios (where specimen depth equaled panel thickness) of 40:1,
16:1, and 4:1 were used for measuring flexural modulus, flexural strength,
and short beam shear strength, respectively.
The three types of flexural tests were performed in a floor model Instron at
room temperature [20 to 22°C (68 to 72°F)] and at the intervals 121°C
(250°F), 204°C (400°F), 260°C (500°F), 288°C (550°F), 316°C (600°F), 343°C
(650°F), and 371°C (700°F). Elevated temperature testing was done in an
oven, in air environment, which could be maintained within ±1°C (±2°F).
The specimens were allowed approximately 15 min to thermally equilibrate
after the desired oven temperature had been reached. A minimum of five test
specimens were used to determine a given property at a given temperature.

Composite Mechanical Properties
Flexural modulus results are shown separately in Fig. 1 for each of the
three polyimide matrix resins. Excellent stiffness retention is apparent up to
260°C (500°F) for the NR-150B2 composites in a. At temperatures approach-
ing the onset of the glass transition range, >288°C (550°F), the modulus
values begin to fall off gradually. Both PMR composites retain their room
temperature stiffness to higher temperatures, showing no significant loss up
0 100 200 300 400 100 " f C 300 400 200 300
NR 150B2 \\ PMR 15 .%T CPI 2237

• GY 70
-\~r "\
-•-.
-*—r»^.
./- \ '9 -~ ^"
100 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 700 100 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 700 100 200300400500600 700
(a) (b) (c)

FIG. 1—Flexural modulus dependence on temperature of graphile/polyimide composites, shown
according to matrix resin type. Data symbols for the composite systems are referred to composite
properties in Table 1.
to 316°C (600°F). The CPI-2237 material in c undergoes an abrupt decrease

in modulus between 316°C (600°F) and 343°C (650°F), whereas the PMR-15
composites in b all show an inflection point in modulus near 316°C (600°F)
(discussed later) before gradually tailing off.
A number of trends are apparent from the three plots. For a given polyimide
matrix, the shapes and transition points in the E versus Tcurves are identical
regardless of fiber type. The relative magnitude of composite stiffness is de-
termined by the fiber type (modulus) as expected, giving rise to the stacking
of curves shown in Fig. lb. In general, the maximum use temperature for zero
stiffness loss coincides with the onset Tg of the resin and is independent of the
fiber. Also, it is clear by comparing the modulus behavior for different com-
posites of a given fiber that the type of matrix resin dictates the composite
temperature response.
The curve shapes for the three different matrix type composites can be re-
lated to the resins themselves. NR-150B2 is an aromatic condensation poly-
imide, linear, amorphous, and essentially uncrosslinked, with a reported Tg
between 350°C (662°F) and 371°C (700°F) [7]. As a thermoplastic, its char-
acteristic of softening near the Tg reduces its ability to transfer load when
used as a matrix material. This is reflected in Fig. la by the stiffness decline
near 288°C (550°F), the apparent onset of the glass transition. In contrast,
both PMR-15 and CPI-2237 are addition-type polyimides, which typically
undergo a thermal transition near 340°C (644°F) related to an addition
crosslinking reaction [11]. This can account for the higher Tg values of the
PMR composites (Table 1), and hence higher temperature stiffness retention
in Figs, lb and c. The Tg also coincides with the£-7curve inflections in all
three PMR-15 based composites which suggests a postcure crosslinking reac-
tion during 316°C (600°F) testing, prior to softening at 371°C (700°F). The
absence of similar behavior in CPI-2237 composites may indicate an already
0 100 200 400 0 100 y ^ 300 400 0 100 200 300 «U
1OO2ra30O40OS0OeoO7QO 100 2 0 0 3 r o « X ) 5 0 0 0 0 0 7 0 0 100 200300400500000700
(a) (b) (c)

FIG. 2—Unnotchedflexural strength dependence on temperature ofgraphite/polyimide compos-
ites, shown according to matrix resin type.
100 200 300 400 0 100 T'^ 300 400 0 100 200 300
NR150B2 ig«?? PMR15 :^sr CPIg237 :i^

• QY70
•» "
100 2 0 0 3 0 0 4 0 0 5 0 0 0 0 0 7 0 0 100 2 0 0 3 0 0 4 0 0 5 0 0 0 0 0 7 0 0 100 200300400500600700

T°F
(a) (b) (c)
FIG. 3—Short beam shear strength dependence on temperature of graphite/polyimide compos-
ites, shown according to matrix resin type.
fully crosslinked state, or a significant deviation in chemical structure from

PMR-15 upon which it is based.
The flexural strength results in Fig. 2 are in marked contrast to the matrix-
dominated modulus behavior. Comparison between the three resins shows a
radical loss in composite strength for the lowest modulus fibers, but an in-
crease or small decrease for the intermediate and high modulus fibers, be-
tween 20°C (68°F) and 343°C (650°F). Specifically, all C-6000 composites,
regardless of resin type, undergo a 50 percent reduction in strength; the
strengths of F-5A and GY-70 composites increase by up to 20 percent in
both PMR polyimides, but decrease by the same amount in NR-150B2.
These fiber-type dominated strength trends are repeated by the shear
strength results in Fig. 3. A 50 percent loss in shear strength occurs between
room temperature and 316°C (600°F) for C-6000 fiber composites, while
shear strengths of F-5A and GY-70 composites typically decrease by only 30
and 20 percent, respectively. As described in detail later, the flexural and
shear strength changes with temperature reflect changes in fiber properties
and fiber-matrix adhesion.
Fiber-Composite Stability Interactions

In both Figs. 2 and 3, the flexural and shear strength versus temperature
curves for each matrix type lie stacked consistently in order of increasing
fiber strength or, conversely, in order of decreasing fiber modulus. Also, the
large discrepancy between initial (room temperature) strength values for the
different fiber types begins to shrink considerably above 204°C (400°F) and
approaches insignificance near 316°C (600°F). These observations strongly
suggest that the thermal stability of the composites is related directly to that
of the reinforcing fiber. This was the conclusion of numerous other studies
[3-6,8,10,13]. The oxidation resistance of graphite fibers, determined by

weight loss during isothermal aging in air, was found to depend on the fiber
type and, furthermore, to dictate composite property retention at elevated
temperature.
Two parameters in particular appear instrumental to the thermo-oxidative
stability of graphite fibers: carbon content and processing-induced contami-
nants, such as the alkali metals sodium and potassium [3,5]. The very high
carbon contents (>99 percent) of both GY-70 and F-5A, shown in Table 2,
reflect their high processing temperatures [>2027°C (3680°F)] and indicate
an extremely inert graphitic nature as well as a near-complete conversion to
carbon during pyrolysis. These account for the high moduli and reported
low sodium and potassium levels remaining in the fiber [5]. In contrast,
C-6000 fibers are processed at a lower temperature [~1227°C (2240°F)], and
as a result have a lower carbon content (93 percent) and modulus. They also
have an appreciable amount of sodium on the fiber surface [3]. The desirabil-
ity of low concentrations of sodium and potassium stems from the ability of
these elements to act as catalysts for the oxidation of carbon in air. The re-
ported trend that a high sodium content lowers fiber stability [5] is shown by
the weight loss variation with carbon content in Table 2. C-6000 fibers un-
dergo significantly greater oxidation than the high modulus types, under
simulated conditions of both long-term aging [700 h at 316°C (600°F)] and a
short-term thermal excursion [24 h at 371°C (700°F)]. Thus, graphite fiber
behavior at elevated temperature can be generalized as: the higher the modu-
lus (or carbon content), the greater the thermo-oxidative stability of the fiber.
Table 2 lists percentage changes in composite strength properties between
room temperature and 316°C (600°F) and 343°C (650°F). These clearly reflect
the fiber-stability/fiber-type dependence just described. For example, flexural
strength values for both high modulus fibers (GY-70, F-5A) either decrease
or increase by approximately 20 percent (the latter effect will be described
later). Flexural strengths of C-6000 composites decreased by more than twice
that amount. The same trend is even more apparent in the progression of
approximate 20 to 30 to 50 percent losses in shear strength, in order of de-
creasing fiber modulus, regardless of matrix type. Although three-point-bend
flexural strength measurements reflect to a certain extent the shear response
of the matrix and fiber-matrix interface, measured shear strength values (r)
are a direct, if qualitative, indication of the fiber-to-matrix bond strength.
Thus, even though the most highly graphitized fibers produce the lowest inter-
laminar shear strengths at room temperature [IS], their superior thermal-
oxidative resistance could contribute to the interfacial strength retention at
316°C (600°F) compared to the C-6000 material.
The potential for fiber degradation with temperature, which may translate
to loss in interfacial shear strength, is present during any of the following: (a)
fabrication of the fiber from its precursor (pyrolysis of polyacrylonitrile); (b)
processing of the composite laminate; and (c) elevated temperature aging/
c^ m ^H 1 ^ W-) r-
o *- cs (s rn >n '<*' -^
II I I I I
o
a.
E
o
U
_ \o «o 00 r-«oo
(N '— CN > 0 TJ- •*
++ + 1 11
-^••B o <on
g <ri m
Z i& « S eu
Z a, U
o js U C
•J •^ o o W1 ( ^ 0>
^ tn c~ o -^ —
^
V
j= U tu
i-S
r^ ff^ vo
-I
n
<
8 8
a
o
X
M
<
o PL,
testing of the laminate. As mentioned previously, the level of contaminants

(for example, sodium and potassium) which originate in the precursor mate-
rial is directly related to the degree of graphitization, such that the higher
modulus fibers contain the lowest levels. Thus, the presence of potential oxi-
dation catalysts is determined during the original fiber forming operation.
Furthermore, carbon fibers are traditionally surface treated to promote sub-
sequent matrix bonding in a composite [75]. This consists of an oxidation
process which increases both the surface area of the fiber and its wettability
by the resin, but, more iqiportantly, produces oxidized carbon functional
groups on the fiber surface [3,15]. The reaction of these with chemical groups
in the resin is the mechanism which leads to enhanced fiber-matrix adhesion.
The extent to which a carbon fiber can be oxidized during surface treatment,
that is, its thermal-oxidative stability, dictates its residual surface oxygen
concentration, and therefore its adhesion characteristics [3]. Thus, composites
containing fibers which are inherently more stable (oxidation resistant) are
expected to exhibit the lower shear strengths.
This phenomenon explains the shear strength versus temperature trends in
Fig. 3. Thermal exposure of GY-70 and F-5A composites, either during lam-
inating or testing, induces little oxidative degradation in the stable fibers.
C-6000 composites, however, appear to undergo continuous oxidative attack
(weight loss) at elevated test temperatures due to inherent oxidative suscep-
tibility of the fibers. This probably is aggravated further by the presence of
sodium and potassium, which could account for variations in slope (loss rate)
of the T-T curves between high and low level contaminant fibers.
The laminating temperature (for example, cure, postcure) of the composite
influences the extent of fiber-matrix chemical interaction, and hence the initial
(ambient temperature) shear strength. Reported data of C-6000/PMR-15
laminates cured at 427°C (800°F) having lower room temperature shear
strengths than those cured at 316°C (600°F) [7] may be an example of severe
fiber oxidation during cure resulting in reduced bonding. Effects of different
laminating temperatures for the different matrix type composites (for exam-
ple, NR-150B2 versus PMR) on the fibers, and hence shear strength, cannot
be distinguished from the data in Fig. 3. For example, that identical fibers in
three different matrix resins yield a range of room temperature values may
be due to differing cure temperature effects (for example, degradation) on
the fibers and interface, but also could reflect variations in bond strength due
solely to the matrix. The resin softening temperature (Tg) and the occurrence
of post-curing reactions are two matrix variables which would influence
resin shear responses, and could contribute to slope differences in the flexural
strength and shear strength versus temperature curves. In particular, the
flexural strength increases (Fig. 2) of both PMR type composites can be at-
tributed to a postcure (cross-linking) reaction during elevated temperature
testing [5,70,77] which, as expected, is not seen in the thermoplastic NR-150B2
composites. However, comparison of Figs. 2 and 3 indicates that fiber degra-
dation effects override any resin postcure contribution to the interface-

dominated shear values.
Structure-Property Interactions
Fracture Surface Characterization—Evidence to support the postulated ef-
fects of fiber/interface degradation and resin postcure on elevated tempera-
ture properties can be obtained from fracture surfaces of flexural test speci-
mens. Figure 4 shows scanning electron micrographs comparing room
temperature and 343°C (650°F) fracture behavior of three fiber types in a
PMR-15 matrix (see Fig. 2). An obvious feature in Fig. 4a is that the pulled-
out C-6000 fiber lengths protruding from the surfaces are significantly longer
for a specimen tested at 343°C (650°F) than at 22°C (72°F). This suggests,
according to composite fracture theory [16], that either a toughness increase
or an interface breakdown occurs with increasing temperature. Comparison
of areas (A) under the actual load (/*)-deflection (d) curves for the respective
composites indicates a marked drop in toughness at the high temperature.
Thus, the greater lengths of exposed fibers point to a loss in fiber-matrix ad-
hesion, in agreement with the experimental shear data. In contrast, the P-d
curves for both F-5A and GY-70 composites in Figs. 4b and 4c, respectively,
FIG. 4a—Comparison of room temperature and 343°C (dSVF) fracture behavior of PMR-15
composites containing C-6000 fibers. The pulled-out fiber lengths on the fracture surfaces indicate
fiber-matrix bond integrity: areas (A) under the load {P)-deflection (d) curves reflect composite
fracture toughness.
F 5A / PMR 15
(b)
72 F 650 F
•
1/
P /
/A, / A,
Aa-161%A,
FIG. 4b—Comparison of room temperature and 343°C (6S(fF) fracture behavior of PMR-15
composites containing F-SA fibers. The pulled-out fiber lengths on the fracture surfaces indicate
fiber-matrix bond integrity; areas (A) under the load (P)-defiection (d) curves reflect composite
fracture toughness.
suggest a significant toughness increase with temperature with little or no in-

crease in pull-out length. Here, the greater energy to fracture is consistent
with the observed rise in flexural strength with temperature due to postcuring
reactions (Fig. 2) and is attributed to greater frictional forces exerted by the
matrix on the fiber surfaces. That F-5A fiber forms a better bond than GY-70
with PMR-15 resin is apparent from the shear data (Fig. 3), and is supported
further by its higher Ajio A\ ratio in Fig. 4.
Similarly, examination of NR-150B2 fracture surfaces reveals features,
shown in Fig. 5, which also are consistent with the mechanical property data.
The C-6000 material in Fig. 5a again shows longer pull-out lengths and lower
apparent toughness at 316°C (600°F) than at room temperature. The GY-70
composite in Fig. 5b, however, requires lower fracture energy at high
temperature with no apparent change in pulled-out fiber length. In this case,
the poor fiber-matrix interlocking observable in the photomicrographs at
both temperatures cannot be compensated for by a postcure reaction, due to
the thermoplastic nature of the NR-150B2 polyimide matrix.
In summary, fiber-type dominated strength retention of graphite/polyimide
composites at high temperatures is confirmed fractographically. C-6000 fiber
QY70/PI^15
(c)
Aj»120% A,
FIG. 4c—Comparison of room temperature and 343°C (650f'F) fracture behavior ofPMR-15
composites containing GY-70 fibers. The pulled-out fiber lengths on the fracture surfaces indicate
fiber-matrix bond integrity; areas (A) under the load (Vydeflection (d) curves reflect composite
fracture toughness.
composites undergo an obvious loss of fiber-matrix adhesion, regardless of

matrix resin type. Poor fiber adhesion, which is evident also as a failure
mechanism in F-5A and GY-TO fiber composites, can be masked in the me-
chanical data by the strengthening effect of postcure crosslinking reactions
in a PMR type matrix.
Characterization of Microstructure—Decreasing interlaminar shear
strength with increasing void content is typical behavior for fiber reinforced
resins, and occurs in graphite/polyimide composites as well [77]. Aside from
optimizing mechanical properties, low void contents are particularly impor-
tant in these materials to protect the graphite fibers from oxidative attack.
Voids, especially if in contact with the fibers, provide paths for air to reach
the fibers, allowing the postulated oxidation reaction to occur [3]. Diffusion
of oxygen through the matrix, enhanced by the presence of voids, also can
contribute to fiber and interfacial degradation [3,6].
Measured void contents of the laminates listed in Table 1 are typically be-
tween 3 and 5 percent. These values reflect some error from the experimental
d—
A,-83% A,
FIG. 5a—Comparison of room temperature and 316°C (600^1^ fracture behavior ofNR-150B2
composites containing C-6000 fibers.
A2*81%A,
FIG. 5b—Comparison of room temperature andilS'C (fiOCPF) fracture behavior ofNR-150B2
composites containing GY-70fibers.
chemical digestion technique, and from potentially significant variations be-

tween true and reportedfiberdensities [3] used in their calculation. Therefore,
they are considered only as approximate, but serve to point out the consider-
able void content magnitude. Porosity levels can be evaluated also from mi-
crographs of polished sections shown in Fig. 6 for all the laminates. Calcu-
lated porosities generally correlate well with those observed. Discrepancies
NR 150B2
f-ISOfim-l
(a)
PMR 15
(b)
FIG. 6b—Optical micrographs of PMR-15 polyimide resin based composites. Note porosity lev-
els and uniformity of fiber bundle compaction and fiber distribution.
CR 2237
l_150^m-f
(C)
FIG. 6c—Optical micrographs of CPI-2237 polyimide resin based composites. Note porosity
levels and uniformity of fiber bundle compaction and fiber distribution.
do exist, namely in GY-70/NR-150B2 (ii), GY-70/PMR-15, and F-5A/

PMR-15 composites, where the measured void contents of 0, 3.7, and 3.7
percent, respectively, appear inconsistent with the micrographs. An important
consideration, by way of explanation, is that considerable variation is found
in laminate compaction (density) and uniformity of fiber distribution, par-
ticularly in difficult-to-process material such as NR-150B2 composites. This
in turn produces fluctuations in the determined void content, by either chem-
ical digestion or microscopic techniques, depending on which section of a
laminate is used for examination. A typical example in Fig. 7 shows an opti-
cal micrograph (a) and a scanning electron micrograph (b) taken from the
same GY-70/NR-I50B2 (ii) laminate as in Fig. 6, but from a different region.
Both illustrate the extent of porosity within and between fiber bundles, of
nonuniform fiber distribution, and poor interply compaction, as a result of
processing.
Correlation of laminate void contents (Table 1) with the respective shear
strength losses (Table 2) generally indicates a trend of increasing strength
degradation with increasing porosity level. C-6000 fiber composites, which
undergo the greatest loss in shear strength at high temperature, are also the
most porous. The porosity level is seen here as a contributing factor in fiber
' KlSOZim-. -Ti^t;^
FIG. 7—Optical (a) and scanning electron (b) micrographs showing voids within and between
fiber bundles in GY-70/NR-150B2 (ii) composite.
FIG. 8—Optical micrograph of interlaminar shear crack in F-5A/PMR-15 composite tested in

short beam shear at 316°C (60(fF). Note that the failure occurs in a path along the fiber-matrix
interfaces.
degradation. Similarly, the greater strength retention of GY-70 composites

may reflect a lower void content as well as greater fiber stability. An example
of an interlaminar shear failure is shown in Fig. 8. The crack preferentially
follows the fiber-matrix interface, illustrating greater matrix shear strength
than fiber-matrix adhesion.
A number of processing-induced features, and other anomalies, are re-
vealed by scanning electron examination of the composite fracture surfaces.
An example of the possible magnitude of a void is shown in Fig. 9. Surfaces
of fibers lining the perimeter of the void, roughly six fiber widths in diameter,
are seen to be exposed and vulnerable to any species present in the void (for
FIG. 9—Scanning electron micrograph of fracture surface showing magnitude of void in

F-5A/PMR-15 composite.
FIG. 10—Anomalous center-cored C-6000 fibers in NR-150B2 resin composite revealed by

scanning electron microscopy offracture surface. Note ends of normal fibers to the left.
example, oxygen, moisture). Fig. 10 shows cored C-6000 fibers which were
found clustered among normal ones in an NR-150B2 composite. Their origin
and effect on mechanical properties are not known, but they serve to illustrate
one of the many variations reported [5] in graphite fiber uniformity. Finally,
an example of a nonoptimum microstructure, typically associated with pro-
cessing ultrahigh performance materials, is shown in Fig. 11. Incorporation
of GY-70 fibers into an NR-150B2 resin results in extensive fiber-to-fiber
contact due to the bilobal nature of the fibers. Relatively little fiber-matrix
interlocking is observed, compared, for example, to C-6000 fibers in PMR-15
in Fig. 4a, revealing the characteristic poor adhesion property of GY-70
fibers.
FIG. 11 —Scanning electron micrographs of G Y- 70/NR- 150B2 composite fracture surface. Note
the frequent occurrence offiber-to-fiber packing contact and poor fiber-matrix adhesion.
Conclusions
Modulus retention to temperatures between 260°C (500°F) and 316°C
(600°F) is not a limiting factor for structural application of graphite/
polyimide composites. Flexural and shear strengths at 316°C (600°F), how-
ever, are moderately to severely reduced from room temperature values, de-
pending on the fiber type. The highest modulus fibers appear to be the most
oxidation resistant, and produce the most thermally stable composites. Con-
versely, the lower modulus fibers suffer thermal degradation, made apparent
by a predominantly interfacial composite failure mode at the higher temper-
atures. Despite thermal losses in both flexural and shear strength, C-6000
composites retain consistently higher values than other higher modulus
composites, at any given temperature. The presence of voids, significant in
number and size, is expected to contribute to fiber oxidative attack as well as
long-term oxygen and moisture absorption. The family of graphite/polyimide
composites investigated shows potential for both stiffness and strength criti-
cal applications, providing their thermal stability and micromechanical
properties are considered in design.
Acknowledgments
This work was supported by the U. S. Department of Energy (DOE) under
Contract DE-ACO4-76-DP00789. G. E. VanTine, J. E. Perea, and N. H. Hall
performed the experiments for this study. Technical review of the manuscript
by A. K. Miller, K. Hirschbuehler, and F. P. Gerstle is gratefully
acknowledged.
References
[7] Petker, I. and Stern, B. A., 23rd National SAMPE Symposium and Exhibition. Society for
the Advancement of Material and Process Engineering, May 1978, p. 775.
[2] Lottridge, L., 25th National SAMPE Symposium and Exhibition, Society for the Advance-
ment of Material and Process Engineering, May 1980, p. 765.
[i] Gibbs, H. H., Wendt, R. C , and Wilson, F. C , Polymer Engineering and Science, Vol. 19,
No. 5, April 1979, p. 342.
[<| Delvigs, P., Alston, W. B., and Vannucci, R. D., 24th National SAMPE Symposium and
Exhibition, Book 2, Society for the Advancement of Material and Process Engineering,
May 1979, p. 1053.
[5] McMahon, P. E., 23rd National SAMPE Symposium and Exhibition, Society for the
Advancement of Material and Process Engineering, April 1978, p. 150.
[5] Scola, D. A., 22nd National SAMPE Symposium and Exhibition, Society for the Advance-
ment of Material and Process Engineering, April 1977, p. 238.
[7] Gibbs, H. H., 21st National SAMPE Symposium and Exhibition, Society for the Advance-
ment of Material and Process Engineering, April 1976, p. 592.
[S] Sheppard, C. H., "Development and Demonstration of Manufacturing Processes for
Fabricating NTS Graphite/PMR-15 Polyimide Structural Elements," First Quarterly Prog-
ress Report, Nov. 1977, Contract NASl-15009 of Boeing Aerospace Co., Seattle, Wash.
98124; National Aeronautics and Space Administration, Langley Research Center,
Hampton, Va. 23665.
[9] Sheppard, C. H., Hoggatt, J. T., and Hunter, A. B., llth National SAMPE Technical
Conference, Society for the Advancement of Material and Process Engineering, Nov. 1979,
p. 40.
[iO] Serafmi, T. T., "Processable High Temperature Resistant Polymer Matrix Materials,"
NASA Technical Memorandum 71682, 1975.
[11] Serafmi, T. T., "Status Review of PMR Polyimides," NASA Technical Memorandum
79039, 1979.
[12] Cavano, P. J., "Resin/Graphite Fiber Composites," NASA Report CR-134727, Dec. 1974.
[13] Cavano, P. J., "Second Generation PMR Polyimide/Fiber Composites," NASA Report
CR-159666, Oct. 1979.
[14] Haynes, W. M. and Tolbert, T. L., Journal of Composite Materials, Vol. 3, Oct. 1969, p. 709.
[IS] Goan, J. C. and Prosen, S. P., Interfaces in Composites, ASTM STP 452, American Society
for Testing and Materials, 1969, p. 3-26.
[16] Kelly, A., Proceedings, The Royal Society, London, Vol. A 319, 1970, p. 95.
[17] Vannucci, R. D., "Effect of Processing Parameters on Autoclaved PMR Polyimide
Composites," NASA Technical Memorandum 73701, 1977.
M. D. Campbell^ and D. D. Burleigh^
Thermophysical Properties Data

on Graphite/Polyimide
Composite Materials
REFERENCE: Campbell, M. D. and Burleigh, D. D., "Thermophysical Properties

Data on Graphite/Polyimide Composite Materials," Composites for Extreme Environ-
ments, ASTM STP 768. N. R. Adsit, Ed., American Society for Testing and Materials,
1982, pp. 54-72.
ABSTRACT: Measurements of the thermal properties of HTS/NR 150B2 and

HTS/PMR 15 laminates were made over the temperature range 116 to 588 K (—250 to
600°F). Properties investigated included thermal conductivity, thermal expansion,
specific heat, and emittance. Layups of [0]n and [0, 45,90, 135], were measured in two
directions for both materials. The quality of both composite systems was relatively
poor. This was typical of polyimide composites at the time. It was expected that, for a
given layup and direction, results would be comparable for the two materials. This was
the case, with some differences attributable to laminate quality rather than differences
in basic materials properties. Erratic variations in thermal expansion measurements
from the same specimen were observed at times. Although these were tentatively at-
tributed to laminate quality, additional work is necessary to fully understand this
behavior.
KEY WORDS: composite materials, polyimide, graphite fiber, thermal expansion,

thermal conductivity, specific heat, emittance
Graphite/polyimide composites are being considered for some aerospace

applications where lightweight, high-strength, high-temperature materials
are required; specifically, in the aft body flap and cargo bay doors of the
space shuttle and in the advanced supersonic transport (AST). These compos-
ites also could replace titanium in some applications. Thermophysical
properties are of interest for the thermal design of structures built with
graphite/polyimide composites.
The object of this program was to generate data on two specific graphite/
polyimide systems being considered for use, and secondarily, to supply data
which would be representative of graphite/polyimides in general, for use in
preliminary design.
The measurements were performed using test methods regularly employed
' Physicist, General Dynamics Convair Division, San Diego, Calif. 92138.
54

CAMPBELL AND BURLEIGH ON THERMOPHYSICAL PROPERTIES 55
TABLE 1—Test panel properties.
Nominal Resin Tensile

Total Thickness, Content, Strength,
Material Layup Plies mm wt % MN/m^
HTS/PMR 15 (UNI) [0],8 18 3.2 35.1 1 263.0

HTS/PMR 15 (ISO) [0,45,90,135]2S 16 2.8 32.9 460.0
HTS/NR 150B2 (UNI) [0]24 24 3.2 22.2 1 551.0
HTS/NR 150B2 (ISO) [0,45,90,I35]82 24 3.2 23.1 548.0
for the characterization of composites with similar properties. They are

based on ASTM standards with modifications, as necessary, to permit accu-
rate evaluation of the peculiar properties of these materials.
Test Materials
Measurements were made on two materials, each in a [0,45,90,135]g quasi-
isotropic layup and in a unidirectional layup.
The HTS/PMR 15 panels were furnished to the program by the National
Aeronautics and Space Administration (NASA) Langley Research Center.
The HTS/NR 150B2 panels were fabricated by General Dynamics Convair
Division. As a check of laminate quality, resin content and tensile strength
were measured for both layups of both materials. Results are shown in
Table 1.
S* K • « • • ^ ^'W h« .434 (MÂJPMR lSl9OtrapctL2«01)
> i .»^' '. ^S^'i^'? S^'^'f **:rX7'?Ssî'^""-^' .t.>"V'r.\-;;.-
!asiiMie««
FIG. 1—Test laminate cross-sections. (X20).
Figure 1 shows cross-sections of both layups of both materials at X20. These

all were taken from thermal expansion specimens after testing. Both materials
showed extensive microcracking in the quasi-isotropic layup. Specimens of
untested material (no thermal cycle exposure) also were examined and ap-
peared similar. Void content of the HTS/PMR 15 panels was low, while void
content was high in the HTS/NR 150B2 panels. Compared to laminates
made with epoxies, individual plies in both materials wandered considerably.
The ply alignment is much worse in the PMR 15 composites. This is attrib-
uted to relatively poor impregnation of the yarn by the matrix in the pre-
preg; that is, the resin is lying on the surface of the yarn rather than being
homogeneously distributed among the fibers.
The cracks seen in Fig. 1 are more numerous in the crossplied material,
probably due to stresses created by the curing process.
Table 2 shows the test matrix.
Test Methods
Thermal Conductivity
Due to the high anisotropy of the graphite/polyimide composites, two dif-
ferent methods were used in measuring the conductivity. For the X (0° fiber
direction) direction, the cut-bar (comparative) method was used (Fig. 2).
Measurements in the Y (90° fiber direction) and Z (through-thickness) direc-
tions were made on a guarded-hotplate (absolute) apparatus covered by the
ASTM Test for Steady-State Thermal Transmission Properties by Means of
the Guarded Hot Plate (C 177-76) (Fig. 3).
Cut-Bar Method
In the cut-strip/cut-bar technique (Fig. 2), a test stack is built up from al-
ternate pieces of a standard material and the unknown specimen. One end of
the stack has an electrical heater; the other end has an electrical heater plate
built on top of a heat sink which may be cooled with liquid nitrogen (LN2),
water, or air. The use of these cooling media coupled with the use of various
insulating plates between the plate heater and the heat sink allow us to vary
the mean test temperature from 100 to 600 K. Two thermocouples are
TABLE;I—Test matrix.
Thermal Thermal Specific

Layup Conductivity Expansion Heat Emittance
X Y Z X Y Z
UNI I 1 2 2
ISO 1 ... 1 2 ... 2 2 2
NOTE—A similar set was tested for both HTS/PMR 15 and HTS/NR 150B2. Numbers indicate
specimen quantities. Temperature extremes are: 116 and 588 K.
RADUTKM SWELDS
FIG. 2—Cut-Strip thermal conductivity lest apparatus.
mounted in each segment of the stack and the entire assembly is radiation
shielded. A bell jar is placed over the assembly and tests may be run in vacuum
or a variety of gaseous environments.
Measurements are made by establishing the desired mean temperature and
a temperature differential of 10 to 25 K across the stack with the two heaters.
When equilibrium is reached, the temperature differences across the specimen
and the standard references are recorded. Thermal conductivity then is cal-
culated using the relationship:
K, = Kr ATs/AT, • X./X, • As/A,
where
Kx = thermal conductivity of unknown,
Ks = thermal conductivity of standard,
FIG. 3—Guarded hot plate thermal conductivity test apparatus.
AJ, = temperature difference across standards (average of two),

ATx = temperature difference across unknown,
Xs = distance between thermocouples on standards (average of two),
Xx = distance between thermocouples on unknown.
As = cross-sectional area of standards (average of two), and
Ax = cross-sectional area of unknown.
Standards are chosen to produce AT values comparable to those across
the unknowns to minimize errors due to unequal heat losses. Standards are
machined from material certified by the National Bureau of Standards (NBS).
Guarded Hot Plate Method

In measurements of conductivity through the fiber planes or in the y-
direction of a unidirectional laminate, we use a guarded hot plate apparatus
(Fig. 3). This is an absolute method that measures thermal conductivity
directly and is covered in principle by ASTM Method C 177-76. The configu-
ration was developed by General Dynamics and is an extension of the cryo-
genic apparatus described by General Dynamics in ASTM STP 411.^
In this method, two thin, concentric electrical heaters are sandwiched be-
tween two test panels. The test panels are approximately 101.6 mm (4 in.) in
diameter with the center 50.8 mm (2 in.) diameter used as the test section.
The specimen faces and heaters are thermocoupled so their temperatures
may be monitored. The specimens are in contact with cooling plates (heat
sinks) at their outer faces. The temperatures of the heaters and cooling plates
can be adjusted to bring the specimens to the desired mean temperature and
AT. The power to the guarded heater is adjusted to eliminate lateral gra-
dients and to ensure that all power supplied to the center heater flows nor-
mally to the specimen planes. The cooling plates are cooled with LN2, water,
or air, and the insulation between outer heaters and the cooling plates is ad-
justed to permit the apparatus to be used from 100 to 600 K.
At equilibrium, the electrical power to the center heater and the average
hot and cold face temperatures are recorded. Thermal conductivity, K, is
calculated from the relationship
K=Ci {EI) t/A AT
where
Ci = geometry and unit factor,
(£7) = center heater power,
t = average center-section thickness,
A = effective center-section area, and
AT = temperature difference between hot and cold faces.
Haskins, J. F. in Thermal-Conductivity Measurements of Insulating Materials at Cryogenic

Temperatures, ASTM STP 411, American Society for Testing and Materials, 1967, pp. 3-12.
For the Z direction measurements, 100-mm (4-in.) test discs were cut from
the as-received sheet without disturbing the surfaces. For the Y direction
measurements, 32 strips approximately 100 by 4 mm (4 X 5/32 in.) were cut,
with the 4 mm (5/32 in.) dimensions in the Y direction. These were then bonded
with NR 150B2 resin to form a plate with the laminate Y direction oriented
through the thickness. The test discs were then cut from these plates. The
resin film has a thermal conductivity similar to the matrix in the laminates
and amounted to less than 1 percent of the total area; therefore, it should not
significantly influence the test results.
Thermal conductivity data were taken at mean temperatures of approxi-
mately 116,180,245,310,375,440,505, and 569 K. Accuracy of both methods
is within ±10 percent.
Thermal Expansion
All expansion measurements were made using a modified Leitz push-rod
dilatometer covered in principle by the ASTM Test for Linear Thermal
Expansion of Rigid Solids with a Vitreous Silica Dilatometer [E 228-71
(1979)]. All specimens were vacuum-dried prior to test to eliminate the pos-
sibility of length changes resulting from moisture content changes during the
expansion test. The specimen was maintained in a dry helium atmosphere
while under test. The apparatus was calibrated using an NBS-certified fused-
silica standard. Accuracy over the full temperature range is better than
±7 X 10"' m/m. This is a mean coefficient of thermal expansion (CTE) un-
certainty of ±0.016 X 10'' m/m/K over 472 K. Data were taken in the fol-
lowing temperature sequence: 297, 116, 176, 236, 297, 347, 397, 447, 497,
547, 588, 297, 116, 588, and 297 K.
Specific Heat
Specific heats were measured using a DuPont 910 Differential Scanning
Calorimeter. Using an NBS-certified synthetic sapphire standard, uncertain-
ties were less than ± 2 percent. A test specimen approximately 1 X 6 mm in
diameter was used. Test specimens were vacuum-dried prior to test. For this
program, data were taken and specific heats of each specimen calculated at
116, 180,245, 310,375,440,505, and 588 K, and plotted against temperature.
Emittance
Total hemispherical emittance was obtained by measuring spectral reflec-
tivity and using X -I- ex = 1, where X = spectral reflectivity, the fraction of
light being reflected, and ex = spectral emissivity, the fraction of light being
emitted.
Spectral reflectivity measurements were made at ambient from 2.5 to 30 jitm
using an ellipsometer. The emittance, e, was then calculated at ambient and
588 K.
Test Results
Test results for each specimen were plotted against mean specimen
temperature and best-fit quadratics drawn through the data. These curves
are shown in Fig. 4 through 7. Representative data are in Table 3.
Since the fiber was the same in both cases and the resins were generically
identical, it was expected that both materials would have nearly the same
thermal conductivities. This was the case in the directions where the matrix
dominated; that is, Y direction of unidirectional laminates (Fig. S) and Z di-
rection for isotropic laminates (Fig. 7). Conductivity in the former was
slightly higher, the result of fibers in the prepreg lamina crossing each other
and providing higher conduction paths across the ply than those provided by
the matrix alone.
Results differed significantly, however, for the directions where the fiber
properties dominate. Near room temperature, values for the HTS/PMR 15
specimens were higher than the HTS/NR 150B2 specimens by a factor of 2
to 3 in the X direction of unidirectional (Fig. 4) and isotropic (Fig. 6) lami-
nates. The reason for this is not clear. In these measurements the fibers
should be carrying most of the heat and in both cases the fibers are the same
type (HTS). The resins are different but are generically the same; their con-
ductivities should be almost identical. This should be a very small contribu-
tion to the overall conductivity in any case. If there were significant differences
between the fibers used for the two different composites, it would be expected
that the thermal expansion and tensile strength data would also be different.
This is not so, however.
The only observed differences between the two composites are the relatively
high porosity of HTS/NR 150B2 and the higher resin content of
j 1 .:
0
-
-
HTS/PMR15
HTS/NR 150B2
1 ; I 1 r
]
1 ;
j i y^ •
B/" y "i j
!
1 • I
'
V^
•i
so 100 ISO 200 350 300 3S0 400 4S0 500 S50 SOO
TEMPERfllURE IKI
FIG. 4—Thermal conductivity, UNI X comparison (cut bar).
1 1 -
1 - HTS/PMR15
0 - HTS/NR1S0B2
' ^--^
1 ^^-^ a ^^
a
-
i
50 100 ISO 20C 250 300 350 400 450 500 550 600
TEMPERATURE IK I
FIG. 5—Thermal conductivity. UNI Y comparison {hot plate).
- HTS./PMR15
0 - HTi>/NR150e2
* •
y
L/\ ^
/"
1 i
SO 100 150 200 250 300 3S0 400 450 SX 550 600
TEllPERflTURE IKl
FIG. b—Thermal conductivity, ISO X comparison (cut bar).
ij • HT >/PMR15
0 - HT »/NR150 B2
r:=»T«
:^
! „ - -
:^S^
--TT-r-n --n-r-n
50 100 150 2X 250 30C 350 400 450 SOO 550 600
TEMPERRTURE IK)
FIG. 7—Thermal conductivity, ISO Z comparison (hot plate).
•o •o • o
E i4
•o • t~
o »
•* • —
•* • •»
d • d
— ^ \ o ô <ri (N
C8 O c
*~ C"
o o o
\ V \ ^
c <N ^ w-1 «t ,
r^ «ri o r^ I
u « C ; OO 00 m ^ ,
d d o p,
U o I I I T.
B
a
S
o
. <r> . o\ . o
.-1
ea
<•
H
• QO
J= «
^^ 3
•o
c0
u
N N
Z Z Z O O O O
en W5
D D
I
fS <N
lo
a:
o
• ^
"O
mi
OB
o
"^
i 03
o
Su oi:
S OS as :s 0£
c 0. Z
•V z •X NZ
• ^ 0. z
12 X
H M
H
(/}
H
X X X X X
X X
HTS/PMR 15. Voids in the matrix and lack of matrix-fiber contact would be
expected to affect the measurements of unidirectional laminates in the Y di-
rection, and the voids should reduce the Z direction conductivity of isotropic
laminates by a percentage near their volume percentage in the specimen.
These measurements were in fairly good agreement, however, which indicates
that the difference in void and resin content do not have a significant effect
on the conductivity in these directions.
In measuring conductivity in the X-direction, it is important that the heat
be carried uniformly throughout the cross-section. For this to occur, the heat
must be introduced uniformly into the end of the specimen; that is, into the
ends of the fibers. This is difficult to ensure. For all fibers to be involved
equally in the heat transfer, the matrix must convey heat among the fibers.
The matrix conductivity is low and if it is further reduced by a high void vol-
ume, then little heat may be transferred to those fibers insulated from the
heat source. This would result in the total area of the cross-section involved
in heat transfer being considerably smaller than anticipated, and would yield
a proportionally lower number.
It is assumed, then, that the differences observed are caused by faulty heat
transfer in the HTS/NR 150 composite due to discontinuities and that the
higher values measured for HTS/PMR 15 are more representative of both
materials.
Thermal Expansion
As a graphite fiber-reinforced organic matrix composite is heated, the
negative CTE fiber tends to contract, while the positive CTE matrix tends to
expand. The resulting length of the composite is dependent on many factors,
most of which are subject to variations from laminate to laminate and from
point to point within a laminate.
CTE's (Al/1/degree) are not constant. In fact, they may not even vary
smoothly. For this reason, it is not appropriate to quote a value for the CTE
of a composite except when referring to a narrow, specific temperature range.
It is more appropriate to present data graphically in the form of Al/1 ver-
sus temperature so that the thermal strain from point to point can be seen
(Figs. 8 through 23). The lines connecting the individual Al points are straight
lines and do not necessarily represent specimen behavior between the dis-
crete points.
For comparison purposes, CTE values for each specimen are given in
Table 3. Since the results often varied from cycle to cycle, they are only rep-
resentative values to indicate the gross behavior of the specimen.
Unidirectional—The combination of the negative CTE fiber and the posi-
tive CTE resin produced very low expansion in the X direction of both uni-
directional materials. Here, the full extent of thefibersis exhibited. During the
initial thermal cycle both materials had slope sign changes from minus to
— 1
lO
2*"^^-^ i.ri3
^-^—f.
^^T" r"^'^
1
J'
) 100 200 300 400 Sm 600 TOO
TEMP K L2196 N2523 1 2 - 9 - 7 7
FIG. i—nermal expansion. HTS/PMR 15, UNI. X.
plus in the 400 to 450 K range. Slopes ranged from approximately —0.4 X 10"*
m/m/K to +6 X 10'' m/m/K for the PMR 15 specimens, and from - 5 X 10"'
m/m/K to + 0.5 X 10"' m/m/K for the NR 150 specimens.
Both materials showed a tendency to change length on succesive thermal
cycles. Differences between values at 300 K were as great as 2 X 10"* m/m
for the NR 150 specimens (Figs. 10 and 11). With the exception of the final
cycle on specimen L2499, this scatter was roughly half for the PMR 15 spec-
imens (Figs. 8 and 9). This indicates that a permanent length change has
taken place which could be caused by microcracking, post-curing of the
resin, or a complex stress relaxation.
In the Y direction of the unidirectional laminates where the matrix CTE
dominates, the matrix differences are evident (Figs. 12 through 15). CTE's
? « - ^ _ ^ A
t.S"
'•^•fe-J * Jl 10 II
IZ.K
~**
~~-~~^_^
It
•d -'
0 100 200 3Cn iOO 500 600 TOO
TEMP K L2499 N2526 1 - 1 0 - 7 8
FIG. ^—Thermal expansion. HTS/PMR 15, UNI, X.
iV
" • - ^ • ^
13"^^^^ ^"^
IZ,I&
^ - ^ • .
• . . .
• • • '
.... .... ....

TD1P K L2895 N3038 5-17-79
FIG. 10—Thermal expansion. HTS/NR 150B2. UNI. X.
2 ' J, L
1,5
-1 0
t'"*'^ 7 - o
-1 5:^^ M •^ s
a: -I
1?
I-.
^ -2 It'
300 300 400 500 BCD

TEMP K L28g9 N3044 5-23-79
FIG. W—Thermal expansion, HTS/NR 1S0B2, UNI, X.
ISO-]
100-
50- ?# =•
2 7
i-S"
0-
_^^^^%
3
<i
-100-
200 300 400 500 Sm
TEMP K L2502 N2554 1-27-78
FIG. \2—Thermal expansion. HTS/PMR IS. UNI. Y.
(.S
«.,/<5
^
H»
i^^"^^
^s^
X
0 too 2t)0 300 400 SCO 600 TOO
TEMP K L2503 N2558 1 - 3 1 - 7 8
FIG. U—Thermal expansion, HTS/PMR 15. UNI. Y.
15
1'
HJ. 7
i.J.IS
1 3 ^ ^ ^ *<
0 100 200 300 400 500 600 TOO
lEdP K L2B97 N3011 5-16-79

FIG. U—Thermal expansion. HTS/NR 150B2. UNI. Y.
lab-
(5
lA
•^^r f^
• ^ — i * ^ ^^^ >,^.I2.H(,
175-
0 100 3 0 0 9 0 0 4 0 0 5 0 0 6 0 0 7 0 3 1
TEMP K L29e9 HS0i2 5-21-79

FIG. is—Thermal expansion. HTS/NR 150B2. UNI. Y.
were approximately 24 X 10'' m/m/K for the PMR 15 and 16 X 10"* m/m/K
for the NR 150 specimens, respectively.
Scatter in the Al data for the PMR 15 specimens was nearly twice as great
as for the NR 150 specimens, which had lower resin content. It is possible
that the high void content prevents microcracking.
Quasi-isolropic—The longitudinal (X) CTE of a quasi-isotropic laminate
should be more positive than that of a unidirectional laminate of the same
fiber and matrix. This was observed for both materials, with the CTE of the
quasi-isotropic being positive over the entire temperature range, rather than
negative at the low end and slightly positive at the high end as were the uni-
directional layups (Figs. 8 through 11 and 16 through 19).
Expansion of the PMR 15 specimens was considerably greater than that
of the NR 150B2 specimens, with the former yielding slopes in the order of
/
A
^
— 2
T
' |>
5
1 3^
^^* /
//
^ * \ y ^
13
/
lU
/
0 100 200 300 400 soo em TOO aco

TEMP K L2500 N2530 1 " 1 7 - 7 8
FIG. Xd—Thermal expansion, HTS/PMR 15. ISO. X.
/•^ ,s
'•*
^/
3^
' >
/
i
laT^ . •- «
"
6-
a 100 200 300 100 500 soo TOO aoo
FIG. n—Thermal expansion. HTS/PMR 15. ISO. X.
IJ
II
^ 2
K
9
A -5
-— 1
2
H
/ ^
11 ^ -— ^
Ifc
y
100 200 300 100 500 600 7C0 aao
FIG. \%—Thermal expansion, HTS/NR 150B2. ISO. X.
? 2-
J
'<^ "
5*/^
- fr-J
^
i
.-> >^ y
J
_—-—Sirf
:J^
lb
"
0 lOO 200 300 4X SOO EOO TOO
TEMP K L2893 N3017 5-25-79

FIG. 19—Thermal expansion, HTS/NR 150B2, ISO, X.
0.8 to 1.7 X 10"' m/m/K and the latter 0.3 to 1.2 X 10"* m/m/K. These
values are indicative of the higher resin content in the PMR 15 specimens;
the difference in the expansion values is of the same order of magnitude as
the difference in resin content.
Both materials exhibited a large amount of scatter in Al/1 over the thermal
cycles.
Net length changes at ambient temperatures were approximately 2 X 10"*
m/m for the NR 150B2 and 4 X 10"" m/m for the PMR 15. Both NR 150B2
specimens produced about the same resuhs. Results for the two PMR 15
specimens were different from each other both in shapes of the Al/1 curves
and total change in length from 113 to 588 K.
It is assumed generally that the CTE's in the Y and Z directions of a uni-
directional laminate are the same since no fibers run in either of these direc-
tions. In reality, the lack of perfect alignment of the fibers in the preprcg
creates some fiber influence in the Y direction of unidirectional laminates

and a lower CTE. This was the case for both materials, with the exception of
the very low values for one HTS/PMR 15 isotropic Z specimen, L2548 (Figs.
21 through 23). Microscopic examination of this specimen revealed it was de-
laminated in some areas, had excessively nonparallel plies, and high void
content. The latter two defects would tend to minimize the contribution of
the matrix, and hence lower the CTE.
Specific Heat
Specific heat results for both materials are plotted in Fig. 24. The curves
are best-fit quadratics. Representative data are in Table 3.
Although the fiber is the same in both materials (HTS), the resins are suf-
l<
Aji<
IS
\%.\\
l'^
1)
I&
200 300 Ida 500 600

TEMP K L2504 N2559 2-2-78
FIG. 20—Thermal expansion. HTS/PMR 15. ISO. Z.
It
•'y 10
•-*5=r-i^ A \ ^ 9
f^~~7
200 300 400 500 600

TEMP K L2548 N2563 2-8-78
FIG. 2\—Thermal expansion. HTS/PMR 25. ISO. Z.
_ ^^
13
200 300 100 SOO 60O

TEMP K L2928 N3077 7-6-79
FIG. 11—Thermal expansion. HTS/NR 150B2. ISO, Z.
IJ"
- 100 /
// „
/j^ ^
T
, S
11.H,*
2*^—'
1}
200 300 400 500 600

TEMP K L2929 N307B 7-9-79
FIG. n—Thermal expansion, HTS/NR 150B2, ISO, Z.
ficiently different in structure that a difference in their specific heats could be

expected. There were also significant differences in resin content which
would lead to higher specific heat values for the HTS/PMR 15 specimens. As
the data show, average specific heat of the PMR 15 specimens was consistently
higher than that for the NR 150B2.
Emittance
Emittance results for both materials are given in Table 3. In all cases,
values for the HTS/NR 150B2 specimens were slightly higher than those for
the HTS/PMR 15 specimens. Emittance data at 589 K, for both materials,
are approximately 3 percent higher than those at 300 K.
: 1 1
a - HTS/PMH15
0 - HTS/NR150B2
''X
5er^ ic 0 i> 0 X XI 2S0 3( 0 3 >0 41 0 < >0 5 0 5" 0 6C

TEnPERHTURE IKl
FIG. 2A—Specific heat comparison (ISO).
Conclusions
Thermal conductivities of HTS/PMR 15 and HTS/NR 150B2 are very
similar in matrix-dominated directions (Z and Y directions of isotropic
laminates).
Significant differences between the two materials were observed in fiber-
dominated directions (X direction of both unidirectional and isotropic lami-
nates). It is suspected that the differences resulted from the higher void con-
tent of the HTS/NR 150B2 affecting the accuracy of the measurement. The
values were expected to be the same as those for HTS/PMR 15.
Thermal Expansion
Scatter was high and reproducibility was poor for both materials, leading
to the conclusion that laminate quality was not optimum in either case.
Higher expansion of the PMR 15 laminates in all except the X direction of
the unidirectional laminate (where the fibers dominate) was assumed to be
the result of higher resin content.
Specific Heat
Differences between specific heats for the two materials were expected,
due to differences in structures of the resins and in laminate resin contents.
Values for the HTS/PMR 15 specimens were consistently higher than those
for the HTS/NR 150B2 specimens, with the difference being approximately
4 percent at 300 K. This correlates with the higher resin content of these spec-
imens since the specific heat of the resin is higher than that of the fiber.
Emitiance
Values for the HTS/NR 150B2 specimens were approximately 4 percent
higher than those for HTS/PMR 15. In both cases, 589 K results were ap-
proximately 3 percent higher than the results at 300 K.
Acknowledgments
The work was sponsored by NASA Langley Research Center under
Contract No. NAS-1-15103, monitored by R. K. Clark.
D. P. Garber,^ D. H. Morris.^ andR. A. Everett, Jr?
Elastic Properties and Fracture

Behavior of Graphite/Polyimide
Composites at Extreme Temperatures
REFERENCE: Garber, D. P., Morris, D. H., and Everett, R. A., Jr., "Elastic Proper-
ties and Fracture Behavior of Graphite/Polyimide Composites at Extreme Tempera-
tures," Compositesfor Extreme Environments, ASTMSTP 768, N. R. Adsit, Ed., Amer-
ican Society for Testing and Materials, 1982, pp. 73-91.
ABSTRACT: The influence of elevated and cryogenic temperatures on the elastic mod-
uli and fracture strengths of several Celion 6000/PMR-15 laminates was measured.
Tests were run at -157, 24 (room temperature), and 316°C (-250, 75, and 600°F).
Both unnotched and notched laminates were tested. Several failure criteria, developed
to predict the uniaxial fracture strength of epoxy laminates, were used to predict the
fracture strength of polyimide laminates.
Lamina elastic moduli were measured at each temperature by testing unnotched
[0]8, [90]i, and [±45]2i laminates. The measured values were used with classical lami-
nate theory to predict the elastic constants in [0/45/90/—45],, [0/45/90/-45]2.,
[45/0/-45/0],, and [45/90/-45/90], laminates. With few exceptions, the predictions
agreed with the moduli measured experimentally. As for the ultimate tensile strength,
although the 8-ply and 16-ply quasi-isotropic laminates were about equally strong at
elevated temperature, their respective strengths diverged at the lower temperatures.
The 8-ply laminates lost strength as the temperature decreased, whereas the 16-ply
laminates became stronger.
The notched laminates had layups of [±45]2., [0/45/90/-45],, and [45/0/-45/0]..
The measured moduli, the ultimate strengths, and the point stress or average stress cri-
terion of Nuismer and Whitney were combined to calculate the characteristic lengths
associated with each criterion. Characteristic lengths were compared to determine the
effect of temperature.
KEY WORDS: composites, graphite/polyimide, temperature effects, notched strength
Graphite/polyimide is a composite material being developed to extend the

useful temperature range of composite materials for such applications as jet
engines, supersonic cruise aircraft, and space shuttle structures. Toward this
' Materials engineer, Kentron International, Inc., NASA Langley Research Center, Hampton,
Va. 23665.
Âssociate professor. Department of Engineering Science and Mechanics, Virginia Polytechnic
Institute and State University, Blacksburg, Va. 24061.
' Aerospace engineer. Structures Laboratory, U.S. Army Research and Technology Labora-
tories (AVRADCOM), NASA Langley Research Center, Hampton, Va. 23665.
73

end, the Composites for Advanced Space Transportation Systems (CASTS)

program at the National Aeronautics and Space Administration (NASA)
Langley Research Center was established to develop a graphite/polyimide
composite that could be used for structural application in the space shuttle
environment. All structures, whether they are in an air or space environ-
ment, must be fastened to other structures. Since this fastening often results
in holes being introduced into the structure, the so-called notch behavior of
the basic material must be understood. The primary objective of this work
was to develop an understanding of the notch behavior of a graphite/poly-
imide composite over the temperature range that would be experienced by
part of the space shuttle structure.
In order to characterize the notch behavior of the graphite/polyimide
material studied in this program, both unnotched and notched tension tests
were conducted. The data reported in this paper are the results of the first
phase of a two-phase program. The first phase covers tests on notched spec-
imens that are 102 mm (4 in.) wide or smaller. The second phase of the pro-
gram, although not reported here, investigates the fracture behavior of test
specimens that are up to 305 mm (12 in.) wide.
Materials and Specimens

The material used for this study consisted of Celion 6000 graphitefibersin
a matrix of PMR-15 polyimide. Specimens were cut from panels of [Ojg,
[90],, [±45]2., [±45]4., [0/45/90/-45]2„ [0/45/90/-45],, [45/0/-45/0],,
and [45/90/—45/90], laminates. Ply thiclcnesses in the cured laminates ranged
between 0.127 mm (0.005 in.) and 0.152 mm (0.006 in.). Fiber volume frac-
tions varied from 54 to 67 percent.
Unnotched tension specimens were 25.4 mm (1 in.) wide by 406 mm (16
in.) long. Fracture specimens were 406 mm (16 in.) long with widths ranging
from 19 mm (0.75 in.) to 102 mm (4 in.). Fracture specimens had center cir-
cular holes that ranged between 1.6 mm (1/16 in.) and 25.4 mm (1 in.) in
diameter.
Test Procedures
Tension and fracture tests were performed at 316, 24, and —157°C (600,
75, and —250°F). An environmental chamber, which was small enough to
permit the specimens to be gripped outside the chamber, was used to control
the test temperature. The maximum variation in temperature was ±5°C
(±9°F) at elevated and cryogenic temperatures. Heat was provided by elec-
trical resistance elements, and cooling was provided by liquid nitrogen.
All tests were performed at a constant cross-head speed. For tension tests
the cross-head speed was 0.0008 mm/s (0.002 in./min), and for fracture tests
it was 0.002 mm/s (0.005 in./min). For all experiments the test section within
the environmental chamber was 203 mm (8 in.). It took approximately 1 h to
GARBER ET AL ON GRAPHITE/POLYIMIDE COMPOSITES 75
reach 316°C (600°F), and 15 min to reach -157°C (-250°F). Specimens were
allowed to soak at temperature for 10 to 15 min before being tested. Loads
were measured with a load cell that was thermally isolated from the test
specimens.
Average longitudinal and transverse strains for tension tests were measured
by foil strain gages that retained accuracy over the range of temperatures in
this study. The accuracy and repeatability of the gages for determining elas-
tic modulus has been established by Chapman [1].* Except where noted, ul-
timate strains were not measured due to gage failures. Strains were not mea-
sured for the fracture tests.
Results
The resuhs of the unnotched tension tests are summarized in Tables 1 and
2. Three replicate tests were performed at each temperature for each lami-
TABLE 1—Laminate material properties.
Unnotched
Specimen Tensile
Temperature, Thickness, Strength, £x, G,2,
Laminate deg C (deg F) mm (in.) MPa (ksi) GPa (Msi) Vxy GPa (Msi)
[0]> 316 1.14 1430 132° 0.333"

(600) (0.045) (208) (19.1)
24 1.12 1450 128° 0.367'
(75) (0.044) (210) (18.6)
-157 1.12 1520 140° 0.364'
(-250) (0.044) (221) (20.3)
[90], 316 1.12 17.7 5.93' 0.024
(600) (0.044) (2.57) (.860)
24 1.17 27.7 8.55' 0.040
(75) (0.046) (4.02) (1.24)
-157 1.14 31.8 10.5' 0.049
(-250) (0.045) (4.61) (1.52)
[±45]2. 316 1.30 92.4 8.48 0.813 2.10
(600) (0.051) (13.4) (1.23) (0.304)
24 1.17 121 18.5 0.776 4.67
(75) (0.046) (17.6) (2.68) (0.678)
-157 1.17 133 22.3 0.686 6.61
(-250) (0.046) (19.3) (3.23) (0.959)
d d
[±45]4. 316 2.31 95.1 0.885
(600) (0.091) (13.8)
24 2.31 132 17.2 0.771 4.76
(75) (0.091) (19.1) (2.49) (0.690)
-157 2.24 145 22.3 0.737 6.41
(-250) (0.088) (21.0) (3.23) (0.930)
'En.
'Ell.
''Calibration error.
*The italic numbers in brackets refer to the list of references appended to this paper.
2 g ?; g
u o. O
_ VO _, —
r^ •*
''I
• O o
a
X
U4
I III " ^ f^
S
l^—'
Tf
^ "O —' >r»
o ' 5 ^ OO
<N - " <N —
ill oS-5r5'nS^S'ô=>o-:SrS»!o°i».o
-^ (-^ "^ r^ -^ (-i <N , i *^ Q "^ « ^^ iri "^ (
I I ^ I I I I I I
^
? I
.1
nate, except where noted. Results presented in this paper are average values
of the replicate tests. In most cases, data values were within a few percent of
the average values presented in this paper. The principal exception occurred
in the data for the [90]8 laminate. For this laminate, the tensile strength at
elevated temperature varied almost 30 percent from the average. As men-
tioned previously, it was not possible to determine ultimate tensile strains in
many cases, due to failure of the gages from either adhesive failure or gage
tearing at high strains.
Lamina constants £11, £22, and 1*12 were determined from tests of the unidi-
rectional laminates, and Gn was determined from tests of the [±45] lami-
nates using Rosen's method [2]. The results appear in Table 1. Laminate
constants for unnotched tension specimens were determined experimentally
for the other laminates, and compared with laminate analysis calculations
based on experimental lamina constants (Table 2). As would be expected,
good agreement was obtained.
The results of the notched tension tests are summarized in Table 3. Three
replicate tests were performed at each temperature for each laminate, except
where noted. Results presented in this paper are average values of the repli-
cate tests. The test data were correlated using the Nuismer and Whitney
point stress and average stress criteria [i]. Each criterion yields a characteris-
tic distance, which appears in the table. More will be said about this distance
in a later section.
Discussion
Unnotched Tension Tests

In order to see more clearly the effect of temperature on elastic properties,
data also are presented in graphic form. The laminates can be divided
roughly into matrix- and fiber-dominated laminates. It can be seen from
Figs. 1 and 2 that both axial stiffness and tensile strength are nearly inde-
pendent of the temperature for the fiber-dominated laminates. An anomaly
in the tensile strength of the 8-ply and 16-ply quasi-isotropic laminates is ap-
parent in Fig. 2. The laminates were approximately equal in strength at ele-
vated temperature. The 8-ply laminate, however, became weaker with de-
creasing temperature, whereas the 16-ply laminate became stronger. The
situation is quite different for matrix-dominated laminates. As seen in Figs. 3
and 4, both axial stiffness and tensile strength increase with decreasing
temperature.
The effect of temperature on Poisson's ratio is somewhat different, as
shown in Fig. 5. For comparison purposes, the laminates may be divided
into three groups. The laminates containing 45-deg plies but no 90-deg plies
exhibit high values of Poisson's ratio, with the ratio increasing with increas-
ing temperature. Laminates containing 90-deg plies but no 0-deg plies ex-
TABLE 3—Notched tension test results.
Notched
Specimen Specimen Hole Tensile Characteristic Distance
Temperature, Width, Thickness, Diameter, Strength,
deg C (deg F) mm (in.) mm (in.) mm (in.) MPa (ksi) do mm (in.) flo mm (in.)
0/45/90/-45].
316 100 1.09 25.4 199 3.07 7.47
(600) (3.94) (0.043) (1.00) (28.8) (0.121) (0.294)
100 1.09 9.53 281 2.57 7.82
(3,94) (0.043) (0.375) (40.7) (0.101) (0.308)
100 1.07 4.76 297 1.50 4.80
(3.93) (0.042) (0.188) (43.1) (0.059) (0.189)
100 1.09 1.59 342^ 0.79 3.18
(3.94) (0.043) (0.063) (49.6) (0.031) (0.125)
24 100 1.09 25.4 175 3.07 7.47
(75) (3.95) (0.043) (1.00) (25.4) (0.121) (0.294)
100 1.12 9.53 234 2.21 6.38
(3.95) (0.044) (0.375) (34.0) (0.087) (0.251)
100 1.07 4,76 274 1.70 5.82
(3.95) (0.042) (0.188) (39.7) (0.067) (0.229)
100 1.07 1.59 310* 0.86 3.78
(3.93) (0.042) (0.063) (44.9) (0.034) (0,149)
50.8 1.17 4.76 259 1.47 4,72
(2.00) (0.046) (0.188) (37.6) (0.058) (0.186)
25.4 1.14 4.76 228 1.12 3.20
(1.00) (0.045) (0.188) (33.0) (0.044) (0.126)
19.1 1.14 4.76 222 1.14 3.30
(0.752) (0.045) (0.188) (32.2) (0.045) (0.130)
-157 100 1.07 25.4 217 8.10 26.4
(-250) (3.94) (0.042) (1.00) (31.3) (0.319) (1.04)
100 1.09 9.53 279* 6.10 29.2
(3.95) (0.043) (0.375) (40.5) (0.240) (1.15)
100 1.12 4.76 290° 3.56 18.9
(3.95) (0.044) (0.188) (42.0) (0.140) (0.744)
100 1.09 1.59 324-^ 12.24 424'
(3.93) (0.043) (0.063) (47.0) (0.482) (16.7)
[45/0/-45/0].
316 63.2 1.17 25.4 298 3.02 8.08
(600) (2.49) (0.046) (1.00) (43.2) (0.119) (0.318)
63.2 1.17 9.53 476 2.49 8.31
(2.49) (0.046) (0.375) (69.0) (0.098) (0.327)
63.2 1.09 4.76 503 1.37 4.80
(2.49) (0.043) (0.188) (72.9) (0.054) (0.189)
63.2 1.09 1.59 684' 2.36 20.7'
(2.49) (0.043) (0.063) (99.2) (0.093) (0.815)
24 63.2 1.17 25.4 274 2.64 6.81
(75) (2.49) (0.046) (1.00) (39.7) (0.104) (0.268)
63.5 1.14 9.53 423 1.96 6.07
(2.50) (0.045) (0.375) (61.4) (0.077) (0.239)
63.2 1.09 4.76 467 1.22 4.09
(2.49) (0.043) (0.188) (67.7) (0.048) (0.161)
63.2 1.12 1.59 542 0.66 2.62
(2.49) (0.044) (0.063) (78.6) (0.026) (0.103)
50.8 1.09 4.76 465 1.24 4.09
TABLE 3 Continued
Notched
deg C (deg F) mm (in.) mm (in.) mm (in.) MPa (ksi) rf„ mm (in.) Oo mm (in.)
(2.00) (0.043) (0.188) (67.5) (0.049) (0.161)

25.4 1.04 4.76 411 0.94 2.87
(1.00) (0.041) (0.188) (59.6) (0.037) (0.113)
18.9 1.07 4.76 403'' 0.99 3.05
(0.746) (0.042) (0.188) (58.5) (0.039) (0.120)
-157 63.5 1.17 25.4 lit 2.69 6.93
(-250) (2.50) (0.046) (1.00) (40.2) (0.106) (0.273)
63.5 1.17 9.53 414' 1.83 5.49
(2.50) (0.046) (0.375) (60.0) (0.072) (0.216)
63.5 1.12 4.76 441 1.04 3.23
(2.50) (0.044) (0.188) (63.9) (0.041) (0.127)
63.5 1.07 1.59 546' 0.66 2.67
(2.50) (0.042) (0.063) (79.2) (0.026) (0.105)
[±45]2,
316 100 86.2 e e
1.17 25.4
(600) (3.95) (0.046) (1.00) (12.5)
100 1.27 9.53 104
(3.95) (0.050) (0.375) (15.0)
100 1.22 4.76 114
(3.95) (0.048) (0.188) (16.5)
100 1.22 1.59 115
(3.95) (0.048) (0.063) (16.7)
24 100 1.14 25.4 118 e e
(75) (3.95) (0.045) (1.00) (17.2)
100 1.24 9.53 127
(3.95) (0.049) (0.375) (18.4)
100 1.22 4.76 151
(3.95) (0.048) (0.188) (21.9)
100 1.27 1.59 160
(3.95) (0.050) (0.063) (23.2)
50.8 1.24 4.76 136
(2.00) (0.049) (0.188) (19.7)
25.4 1.22 4.76 116
(1.00) (0.048) (0.188) (16.8)
18.9 1.24 4.76 103
(0.746) (0.049) (0.188) (14.9)
-157 100 1.22 25.4 125 e e
(-250) (3.96) (0.048) (1.00) (18.2)
100 1.12 9.53 151
(3.95) (0.044) (0.375) (21.9)
100 1.24 4.76 174
(3.95) (0.049) (0.188) (25.3)
100 1.24 1.59 174
(3.95) (0.049) (0.063) (25.2)
[0/45/90/-45] .
316 63.5 2.36 4.76 328 1.85 6.68
(600) (2.50) (0.093) (0.188) (47.5) (0.073) (0.263)
24 63.2 2,41 4.76 319 1 40 4. 37
TABLE 3 Continued
Notched
deg C (deg F) mm (in.) mm (in.) mm (in.) MPa (ksi) dc mm (in.) a^ mm (in.)
(75) (2.49) (0.095) (0.188) (46.2) (0.055) (0.172)

-157 63.5 2.34 4.76 345 1.22 3.58
(-250) (2.50) (0.092) (0.188) (50.0) (0.048) (0.141)
"One specimen.
*Not used in plot.
'Two specimeiis.
''Four specimens.
'Nuismer-Whitney model not applicable.
-'ÂU specimens failed at grip.
'Two specimens failed at grip.
* One specimen failed at grip.
hibit low values of Poisson's ratio, with no significant effect of temperature.

The remaining laminates have intermediate values of Poisson's ratio which
closely resemble the values for some isotropic metallic materials. For these
laminates there is no significant effect of temperature.
Due to a lack of data for the Celion 6000/PMR-15 material, it is not pos-
sible to make comparisons with the work of other researchers.
-200 200 600

150 —r—
20
100 15
TENSILE
MODULUS, Msl
GPa [i)5/n/-i|5/0]. 10
50 [o/is/goz-iB],
[0/II5/90/-15] 2s
-200 0 200 HOO

TEMPERATURE, °C
FIG. 1—Axial stiffness versus temperature for fiber-dominated laminates.
°F
-200 200 600
2000
1 1 1 1
1500
tOlg " 200
TENSILE
STRENGTH, 1000 - - ksl
MPa
[W/OZ-HS/Olj
100
500 ____[0/i|5/90/-H5]2s
u—^— "
_ o — — •—
-
[0/l(5/90/-45]5
0 1 1 , 1 1 1
-200 0 200 too
TEMPERATURE, °C
FIG. 2—Tensile strength versus temperature for ftber-dominated laminates.
2 Msl
-200 0 200 100

Temperature, °C
FIG. 3—Axial stiffness versus temperature for matrix-dominated laminates.
°F
-200 200 500 30
200 1 • ' • - " . - , , _ — , . ,
-——_____U5/90/-U5/90]5
150-
TENSILE
STRENGTH, 100
MPa
^"""^^^^^^^^^^^^^S^ 15 ksl
[^^§f8
50
1 1 1 r , 0
-200 0 200 ilOO
TEMPERATURE, °C
FIG. 4—Tensile strength versus temperature for matrix-dominated laminates.
1.0
.6
POISSON'S
RATIO
/[0/i45/90/-'45]5
[0/i45/90/-'*5],
[it5/90/-i(5/90]^
[90]o
-200 0 200 100

TEMPERATURE, °C
FIG. 5—Poisson's ratio versus temperature.
Notched Tension Tests

The effect of temperature on notched strength will be examined first. Fig-
ure 6 indicates a slight temperature effect for the quasi-isotropic laminate.
Higher strengths were obtained at elevated and cryogenic temperatures than
at room temperature. On the other hand, the strength of unnotched quasi-
isotropic laminates decreased with decreasing temperature (Fig. 2). The
notched strength of this laminate at cryogenic temperature should be viewed
with caution, as approximately half of the specimens failed at the grips.
Due to the large number of grip failures at —I57°C (—250°F) for the quasi-
isotropic laminate, all tests of the [45/0/—45/0], laminate were conducted
on specimens 64 mm (2.5 in.) wide. Figure 7 shows only a slight temperature
effect for this laminate, except for the smallest hole (one of three specimens
did not fail at the hole nor at the grip). Cryogenic and room temperature
strengths were nearly identical with a slightly higher strength at elevated
temperature. Similar trends were noted for the unnotched laminates (Fig. 2).
As with the quasi-isotropic laminate, the effect of hole size on strength is
very noticeable.
The data for the [±45]2i laminate, as shown in Fig. 8^ should be viewed
with caution. Many of the specimens showed extensive ply pullout after fail-
ure, making it impossible to determine with any degree of certainty whether
failure initiated at the hold, the grip, or the region of high thermal gradient
-200 200 600

too
300
- 10
NOTCHED
STRENGTH, 200
ksl
MPa
HOLE D1A>1ETER
o 1.6 mm (1/16 In) 20
o M.8 iTin (3/lb In)
100
6. 9.5 mm (3/8 In)
O 25.t mm (1 In)
-200 0 200 too

TEMPERATURE, °C
FIG. 6—Notched strength versus temperature for the [0/45/90/—4S], laminates, width = 102
mm (4.0 in.).
500
SO
100-
NOTCHED
STRENGTH, ksl
NPO
1)0
HOLE DIAMETER
20C- 1.6 nri ( 1 / 1 6 I n )
i|,8 nm (3/16 In)
9.5 inn (3/8 In)
25.1 m (1 In)
-200 2!5!r too
TEMPERATURE, °C
FIG. 1—Notched strength versus temperature for the [45/0/-45/0], laminates, width = 65.4
mm (2.5 in.).
-200 200 600

200
150
20
NOTCHED
STRENGTH, 100 HOLE DIAMETER ksl
HPO 0 1.6ran(1/16 In)
a 1.8 mm (3/16 In)
a 9,5ran(3/8 In) 10
50 O 25.1)ran(1 In)
-200 0 200 100

TEMPERATURE, ° C
FIG. 8—Notched strength versus temperature for the [±4S]2s laminates, width = 102 mm (4.0
in.).
1,0 2,0 3,0 H.O

500 1 1 1 1 1 1 1 1 - 70
^ CtS/OZ-W/OIj
400
56
300 [0/15/90/-H5]5
NOTCHED
- 42
STRENGTH, ksl
HPa
200-
28
[tl.5]2s
n
100- 14
1 1
1 1 1 J 1 1 1
20 40 60 100
WIDTH, mn
FIG. 9—Notched strength versus width at room temperature for a fixed/law size at 4.8 mm
(3/16 in.).
where the specimens passed through the oven wall. This suggests that [±45]
specimens should be tested in a different manner.
Despite the uncertainty as to the location of failure initiation, the data
shown in Fig. 8 clearly indicate the existence of a notch effect which is
temperature dependent. As would be expected for a matrix-dominated lami-
nate, both strength and the apparent notch effect decrease with increasing
temperature.
Figure 9 shows the effect of width on notched strength for afixedflaw size
and temperature. For all three laminates there is little width effect for plates
wider than S1 mm (2 in.); that is, plates with a hole diameter to width ratio
less than 0.094 may be treated as infinite plates. As mentioned previously, no
tests were conducted on 102 mm (4 in.) plates for the [45/0/—45/0]s lam-
inate. It should be noted that all data points in Fig. 9 are the average of three
identical tests.
Failure Model
Figures 10 and 11 compare the experimental notched strengths with
strengths predicted by the Nuismer and Whitney point stress and average
stress models [3]. The point stress model states that failure of a notched ten-
sion specimen will occur when the stress at some distance, do, ahead of the
notch equals the unnotched tensile strength. In equation form, this failure
criterion is
ON
-^=2[2 + C' + 3C- (KT' - 3)(5C* - 74')]-'
Oo
1.0,-
^ 316°C (600°F)
o 2i)°C (75°F)
O -157°C (-250°F)
NOTCHED
TENSILE
STRENGTH
UNNOTCHED
TENSILE ^ yO^ = 5,8 m (0.23 in)
STRENGTH ,t - CURVES FITTED ONLY
TO DATA AT 2lt°C
AND 316°C
_L.
0.1 0.2 0.3 0.1
HOLE DIAMETER
WIDTH
FIG. 10—Comparison of experimental and theoreticalfailure strengths for the [0/45/90/—45],

laminates, width = 102 mm (4.0 in.).
where { = R/{R + do), ON" is the failure stress for a notched infinite plate, Oo
is the unnotched tensile strength, KT° is the orthotropic stress concentration
factor for an infinite width plate, and R is hole radius. The infinite plate
notched strength is found by multiplying the experimentally determined
notched strength for a finite width plate by a finite width correction factor
1.0 £, 316°C (600°F)
o 2H°C (75°F)
D -157°C (-250°F)
NOTCHED
TENSILE
STRENGTH
UNNOTCHED ,ag = 1.6 mm (0.18 In)
TENSILE
STRENGTH
0.1 0.2 0.3 0.4

HOLE DIAMETER
WIDTH
FIG. 11—Comparison of experimental and theoretical failure strengths for the [45/0/—45/0],
laminates, width = 64 mm (25 in.).
The average stress model states that failure of a notched specimen will
occur when the average stress over some distance ahead of the notch, Oo,
equals the unnotched tensile strength. This failure criterion may be written
^ ^ = 2(1 - 0[2 -C-^ + (KT" - 3)(l^ - C')]''

Oo
where { = R/{R + Oo)- The distance do and a<, are called characteristic
distances.
It is apparent from Fig. 10 that one value offloor do gives fair predictions
of the notched strength of the [0/45/90/—45], laminate for both room and
elevated temperature. These characteristic distances were found using the
appropriate failure criterion, the unnotched strength, and the experimentally
determined notched strengths. The values of Oo and do are average values for
all data at the two temperatures and various hole sizes. The average stress
criterion gives slightly better results than the point stress criterion. Average
values of ao and do also were determined from the cryogenic temperature
data for this laminate. Due to uncertainty in data from both notched and
unnotched tests, however, curves for these values are not shown in the figure.
As shown in Fig. 11, the notched strengths of [45/0/—45/0], laminates at
all three temperatures can be predicted fairly well using one value of Oo or do.
Again, the average stress criterion gives a slightly better fit.
However, as shown in Fig. 12, the fracture data for the matrix-dominated
[±45] laminate cannot be correlated using the Nuismer-Whitney model [5].
For this laminate, notched strengths greater than unnotched strength were
1.5r
e D
1.0
NOTCHED
TENSILE
STRENGTH
UNNOTCHED A 316°C (600°F)
TENSILE
STRENGTH g 5 o 2«°C (75°F)
D-157°C (-250°F)
0.1 0.2 OTi UTI

HOLE DIAMETER
WIDTH
FIG. 12—Comparison of experimental and theoretical failure strengths for the [±4S\i, lami-
nates, width = 102 mm (4.10 in.).
measured. Similar results have been found for a graphite/epoxy laminate at

room temperature [5].
Figures 13 to 15 depict characteristic distance as a function of tempera-
ture. Figure 13 shows the behavior of a,, and do for a 16-ply quasi-isotropic
laminate for three temperatures. Also shown is the variation of Oo and do
with temperature for the [45/0/—45/0]i laminate. The characteristic distances
show nearly linear response with temperature for both laminates.
As shown in Fig. 14, the characteristic distance-temperature behavior for
the 8-ply quasi-isotropic laminate is different from the 16-ply behavior
shown in Fig. 13. As illustrated in Fig. 14, the elevated and room tempera-
ture value of ao and do are practically identical, but characteristic distances at
cryogenic temperature are much higher. This difference in behavior may be
due to two factors.
First, due to a large number of grip failures, the average values of a^ and do
at cryogenic temperature, as shown in Fig. 14, may not be as accurate as the
values at room and elevated temperature. However, it is felt that the number
of tests at a given temperature does not explain adequately the large differ-
ence in flo and do for low temperature as compared to the other two
temperatures.
Second, the differences in Oo and do shown in Fig. 14 may be due to the ex-
treme sensitivity of the characteristic distance calculation of the ratio of
notched to unnotched strength. Slight changes in the ratio atT/oo cause a<, and
AVG, PT,
STRESS STRESS
• oCltS/OZ-IS/Olj. H = en mil (2,5 In)
• D [0/115/90/-1)5],, , W = 102 mm (11.0 In)
CHARACTERISTIC
DISTANCE, H
mn
0 200
TEMPERATURE, °C
FIG. 13—Characteristic distance versus temperature for the [45/0/~45/0], and [0/45/-
90/-45]2, laminates.
-200 200 600

30
1.0
20
CHARACTERISTIC
DISTANCE,
nm
0.5 In.
10
^ a
-200 0 200 100

TEMPERATURE, °C
FIG. 14—Characteristic distance versus temperature for the [0/45/90/—45], laminates,

width = 102 mm (4 in.).
AVG. PT.
STRESS STRESS
• o [W/OZ-aS/Olj, W= 6i| rm ( 2 . 5 In)
• ° C0/i45/90/-'(5]2s' W = 61 mm (2.5 In)

4 £i [0/')5/90/-H5]5, W = 102 mm (1.0 In)
• O NUISMER AND WHITNEY (REF. 3)
CHARACTERISTIC
DISTANCE, t
-200 0 200 100

TEMPERATURE, °C
FIG. 15—Characteristic distance versus temperature for the fiber-dominated laminates.
do to change dramatically. The noticeable decrease in the unnotched strength

of 8-ply quasi-isotropic laminate with decreasing temperature (Fig. 3) ap-
pears to be responsible for the unusually high values which were calculated
for the characteristic distance at cryogenic temperature, as shown in Fig. 14.
In order to reduce the effect of an apparent unnotched strength anomaly
on the characteristic distance calculations, the Oo and do values were recalcu-
lated for the 8-ply quasi-isotropic laminate using the unnotched strength of
the 16-ply laminate for normalization. This procedure appeared to be justi-
fied in view of the exceptional trend in the unnotched tensile strength data
for the 8-ply quasi-isotropic laminate. The decrease in strength with decreas-
ing temperature apparent in Fig. 3 is significantly different from the behav-
ior of the other fiber-dominated laminates and from the behavior of the ma-
trix-dominated laminates shown in Fig. 4. The modified results are shown in
Fig. 15, along with all other point and average stress criteria data. All the
data in this figure show an approximately linear relationship with tempera-
ture. The data are in fair agreement with the room temperature data for sim-
ilar laminates of graphite/epoxy as reported by Nuismer and Whitney [5].
Although not shown in Fig. 15, the room temperature values of Oo are similar
to those found by Morris and Hahn [4,5] for similar laminates of graphite/
epoxy.
Since the Celion 6000/PMR-15 composite was in the early stages of devel-
opment, no attempt was made to obtain a better fit to the data shown in
Figs. 10 and 11. However, the mathematically simple models of Nuismer and
Whitney appear to predict the notched strengths of the fiber-dominated lam-
inates studied. A more accurate prediction for various temperatures awaits
further development of the Celion 6000/PMR-15 material such that more
consistent values of strength are obtainable.
Summary and Conclusions

The results presented here indicate that elastic modulus and unnotched
tensile strength for the fiber-dominated laminates are not highly temperature
dependent. However, the corresponding moduli and strengths for the ma-
trix-dominated laminates are quite sensitive to changes in temperature.
Notched tensile strength is dependent on both hole size and temperature,
regardless of the type of laminate. Notched plates with hole diameter to plate
width ratios less than 0.094 may be treated as infinitely wide plates, at least
at room temperature.
The Nuismer-Whitney point stress and average stress models [i] give fair
predictions of notched strength for the [0/45/90/—45]8 laminate. Due to un-
certainty in the cryogenic data resulting from the large number of grip fail-
ures, little confidence can be placed in the values for a<, and do determined for
this temperature. The available data, however, tend to indicate that ao and do
are temperature dependent. Due to the relative insensitivity of the strength
calculations to small changes in characteristic distance, it is felt that one

value of Oo or do may be sufficient for room and elevated temperatures, as ev-
idenced by Fig. 10.
The temperature dependence of Oo or do for the [45/0/—45/0]. laminate is
much less than for the quasi-isotropic laminate. In fact, one value of Oo or do,
along with the corresponding Nuismer-Whitney model, results in fair predic-
tions of the notched strength of the [45/0/—45/0], laminates for all three
temperatures. Since this study was conducted during the early stages of de-
velopment of the Celion 6000/PMR-15 material, no attempt was made to de-
termine the functional form of the temperature dependence of the character-
istic lengths.
The Nuismer-Whitney models are of no value when predicting the notched
strength of the [±45]2. laminate, regardless of temperature.
References
[/] Chapman, A. J., "Graphite/Polyimide Tension Tests at Elevated and Cryogenic Tempera-
tures," NASA Conference Publication 2079, National Aeronautics and Space Administra-
tion, Langley Research Center, Hampton, Va., 1979.
[2] Rosen, B. W., Journal of Composite Materials, Vol. 6, Oct. 1971, pp. 552-554.
[3] Nuismer, R. J. and Whitney, J. M., in Fracture Mechanics of Composites, ASTM STP 593,
[4] Morris, D. H. and Hahn, H. T., Journal of Composite Materials, Vol. 11, April 1977, pp.
124-138.
[5] Yeow, Y. T., Morris, D. H., and Brinson, H. F., Experimental Mechanics, Vol. 19, No. 1,
Jan. 1979, pp. 1-8.
[5] Morris, D. H. and Hahn, H. T., in Composite Materials: Testing and Design (Fourth Confer-
ence), ASTM STP 617, American Society for Testing and Materials, 1977, pp. 5-17.
Atmospheric and Extroatmospheric
Environments
E. E. Morris^
Filament Wound Composite Thermal

Isolator Structures for Cryogenic
Dewars and Instruments
REFERENCE: Morris, E. E., "Filament Wound Composite Tliermal Isolator Structures

for Cryogenic Dewars and Instruments," Composites for Extreme Environments, ASTM
STP 768. N. R. Adsit, Ed., American Society for Testing and Materials, 1982, pp.
95-109.
ABSTRACT: Studies showing higli tensile strength, low thermal conductivity, and ade-
quate fatigue strength capabilities in conjunction with low resin outgassing properties
of S-901 Tiber glass with SCI REZ 080 and 081 epoxy resins has resulted in use of fila-
ment wound tension straps, struts, and conical shells as thermal isolators in several
high-performance cryogenic applications. These thermal isolator structures and their
use in the following cryogenic systems are discussed in this paper: hydrogen and oxy-
gen dewars for space shuttle, helium tank for the infra-red astronomy satellite, space-
craft refrigerators, and infrared telescope. Mechanical and thermo-physical properties
of the composite laminates are presented.
KEY WORDS: composite materials, structural composites, filament wound composite,

cryogenic properties, thermal conductivity, glass fiber, epoxy resin, tensile strength, fa-
tigue strength, thermal conductivity, outgassing
Composite materials are replacing metals in many aerospace and commer-

cial applications. During the 1960's much effort was directed at developing
glass filament wound composites for missile motor cases and pressure ves-
sels. Later, with the development of (a) boron, graphite and Kevlar fila-
ments, {b) advanced matrix systems, and (c) improved manufacturing meth-
ods and capacity, the advanced composite industry emerged and now is
supplying many primary and secondary structures for military, space, and
commercial applications.
Much of the early work with glass filament wound pressure vessels, spon-
sored primarily by the National Aeronautics and Space Administration
(NASA)-Lewis Research Center, was aimed at development of cryogenic
fluid containment tanks for future flight vehicles. These programs resulted in
some of the first data on performance of glass/epoxy structures from room
'Senior vice-president, Structural Composites Industries, Inc., Azusa, Calif., 91702.
95

to cryogenic temperatures. The data clearly show significant increases in

strength and modest increases in modulus at low temperatures. Subse-
quently, S-glass/epoxy filament wound composite material in several forms
has been extensively characterized from 394 K (250°F) to 4 K (-452°F) to
provide mechanical and thermo-physical properties of the composite lami-
nates. Comprehensive reviews of data from these programs have been pub-
lished by Kasen.^'
The results of these studies convinced designers to consider and select
glass/epoxy composites for cryogenic service where low thermal conductiv-
ity could be used in conjunction with high mechanical properties to produce
thermal isolation structures for spacecraft hardware. Over the period 1970 to
1980, thermal isolators were developed for several high-performance cry-
ogenic systems, which will be described here. The isolators consist of fila-
ment wound tension straps, struts, and conical shells. The principal applica-
tions have been in spacecraft cryogenic hydrogen, oxygen, and helium
dewars and in cryogenic coolers and refrigerators. In conjunction with the
hardware development and production, significant design and characteriza-
tion data have been acquired and are summarized in this paper.
Thermal Isolator Straps for Cryogenic Dewars and Instruments
Space Shuttle Power Reactant Storage Assembly (PRSA)

The space shuttle cryogenic PRSA tanks provide storage of oxygen and
hydrogen used to generate the orbiter's electrical power and make-up oxygen
for the crew compartment. Each tank is a dewar consisting of two concentric
spheres, with the oxygen tank having an outside diameter of 99 cm (39 in.)
and the hydrogen tank having an outside diameter of 119 cm (47 in.); Fig. 1
shows schematically the arrangement of the oxygen tank.
The inner sphere, or pressure vessel, contains the stored fluid at cryogenic
temperatures and supercritical pressures. The outer sphere allows mainte-
nance of a high vacuum between the two spheres for thermal insulation.
Twelve filament wound glass/epoxy matrix composite straps are used to
support the pressure vessel from the outer shell; they are located within the
vacuum space between the two spheres. These straps are able to minimize
heat conduction from outside the tank into the stored cryogen, because of
the inherent low conductivity of the glass/epoxy composite and the low
cross-sectional area required to support the loaded pressure vessel. All the
composite straps are preloaded to as high as 907 kg (2000 lb) to prevent com-
pression loading during space shuttle launch vibration and ±5 g constant ac-
^Kasen, M. B., in Composite Reliability, ASTM STP 580. American Society for Testing and
Materials, 1975, pp. 586-611.
'Kasen, M. B., "Mechanical and Thermal Properties of Filamentary-Reinforced Structural
Composites at Cryogenic Temperatures—I: Glass-Reinforced Composites," Cryogenics, Vol.
25, No. 4, 1975.
MORRIS ON CRYOGENIC DEWARS AND INSTRUMENTS 97
OXYGEN TANK
DENSITY PROBE-
HEATER ASSEMBLY-
- PRESSURE
VESSEL
, SUPPORTS
QUANTITYA
r^^^^^TiT J^ [&L^^^ \
SIGNAL \
CONDITIONER^
GIRTH RING- l ^ / M I A C - I O N PUMP
VACUUM PINCH O F F ^ \ FLUID INTERFACE-
>—MOUNTING TRUNNION
FIG. 1—Space shuttle PRSA oxygen dewar.
celeration loads. Total static supported weight is about 408 kg (900 lb) for
the oxygen tank and 82 kg (180 lb) for the hydrogen tank.
The straps, made from unidirectional windings of Owens-Corning S-901
glass fiber roving impregnated with SCI REZ 081 epoxy resin, are shown in
Fig. 2. They have high strength-to-density and strength-to-thermal conduc-
tivity ratios at ambient and cryogenic temperatures. Oxygen pressure vessel
thermal isolator support straps measure 25.4 cm (10 in.) long with 3.18-cm
(1.25-in.) -diameter load pin, and 0.26-cm (0.10-in.) -thick by 0.89-cm (0.35-
in.) -wide legs. The hydrogen tank straps are 30.5 cm (12 in.) long with 3.18-cm
(1.25-in.) -diameter load pin and 0.20-cm (0.08-in.) -thick by 0.76-cm (0.30-in.)
-wide legs.
The strap and inner pressure vessel support assembly is shown in Fig. 3;
the outer vacuum shell is added to complete the dewar assembly. Qualifica-
tion testing of the PRSA assemblies has been completed successfully, includ-
ing critical vibration and thermal tests. So effective is the combined insulat-
ing characteristics of the hydrogen tank that it would take 63 days for the 20
K (—423°F) hydrogen to evaporate in a filled PRSA tank.
Infra-Red Astronomy Satellite (IRAS) Helium Tank

The IRAS is an earth-orbiting instrument to be used for mapping of in-
frared sources to learn more about the origins of the universe. The heart of
the IRAS is a 60-cm (24-in.) infrared telescope, which is supercooled to 2 K
(—456°F) to give it unprecedented sensitivity to celestial infrared radiation
IIHI
FIG. 2—Space shuttle PRSA cryogenic tank support straps.
(IR). The low temperature reduces thermal background noise to low levels so
that the IR detectors are more sensitive. Telescope cooling is achieved using
cryogenic super critical helium stored in a toroidal main cryogen tank, which
is in turn supported with S-901 glass/SCI REZ 080 resin unidirectional
thermal isolation straps. Figure 4 shows the general arrangement of IRAS,
and Fig. 5 shows the straps with metal load pin spools installed. Nine 30.5-
cm (12-in.)-long straps are used in the IRAS to support the cryogen tank and
its contents, and to provide thermal isolation sufficient for a one-year mis-
sion in space with active helium cooling. The IRAS recently passed its vibra-
tion qualification testing successfully.
Cryogenic Coolers and Refrigerators

S-901 glass/SCI REZ 080filamentwound straps-similar to those described
for the space shuttle PRSA and for the IRAS, but of smaller size, have been
used for satellite instruments. Table 1 lists some typical sizes. Some of the
straps are smaller than a paper clip, with extremely small cross section, but
have very high breaking strength. Figure 6 shows some of these small parts.
Applications include the radiative cryogenic cooler for the Geostationary
Operational Environmental Satellite's (GOES) visible infrared spin-scan ra-
diometer atmospheric sounder sensor, which measures temperature and
moisture in the earth's atmosphere and takes cloud cover pictures. The
GOES-3 instrument transmitted pictures of the Mount St. Helens volcano
eruption. Similar instruments are in the Pioneer 2 spacecraft to take pictures
wllian EMU
Thermal
Isolator
Straps
FIG. 4—IRAS telescope system configuration.
of Saturn and its rings. The near infrared mapping spectrometer of the Gali-
leo spacecraft to Jupiter also will use small filament wound straps to hold an
infrared detector array and to align it. One side of the array is a black radia-
tor, radiating into space to achieve very low array temperatures (70 to 80 K)
for maximum resolution and efficiency. The straps limit heat flow to the ra-
diator to permit low temperature. Another similar current application of the
S-901 glass/SCI REZ 080 straps is in the thematic mapper instrument of the
Landsat D spacecraft.
Another important class of instruments utilizing the filament wound
thermal isolators is cryogenic refrigerators, typified by the High Energy As-
tronomy Observatory (HEAO) cryogenic ammonia and methane refrigera-
tor systems. These are employed in the gamma ray spectrometer to maintain
the necessary low temperature.
m-
FIG. 5—IRAS cryogenic tank support straps.
TABLE 1—Summary of sizes of suspension straps fabricated.
Typical
Total Inside Pin Diameter, Width, Thickness, Cross-Sectional Breaking Load,
Length, in. in. in. in. Area, in.^ lb
27.00 0.85 0.225 0.173 0.0779 12 000

12.000 0.990 0.800 0.058 0.0928 20 000
11.040 1.25 0.386 0.080 0.06176 15 000
10.040 1.25 0.386 0.080 0.06176 13 300
10.500 0.50 0.440 0.055 0.0484 9 600
8.750 1.25 0.350 0.100 0.0700 15 700
8.750 1.25 0.350 0.100 0.0700 14 800
8.325 1.13 0.410 0.210 0.1722 26 800
6.500 0.50 0.600 0.055 0.0660 13 000
5.725 0.90 0.220 0.140 0.0616 10-800
4.625 0.20 0.040 0.023 0.00184 425
3.500 0.50 0.605 0.055 0.0666 13 000
3.310 0.53 0.168 0.150 0.0504 6 400
3.090 0.25 0.080 0.070 0.0112 1 750
2.830 0.53 0.168 0.073 0.0245 4600
2.581 0.28 0.100 0.040 0.0080 1 600
2.250 0.25 0.080 0.035 0.0056 1 240
1.846 0.28 0.040 0.025 0.0020 425
1.557 0.14 0.020 0.012 0.00048 120
1.275 0.14 0.020 0.012 0.00048 120
1.030 0.28 0.040 0.023 0.00184 400
0.955 0.20 0.040 0.023 0.00184 400
NOTE—Conversion factors: 1 in. = 2.54 cm,; 1 in.^ = 6.452 cm^ 1 lb = 0.454 kg.
^vy 3 HHI
Â^
A' iTâ'l-^'MiTîiP"
FIG. 6—Small support straps for cryogenic instruments.
Thermal Isolator Shells and Struts for Cryogenic Instruments

Another configuration of thermal isolator support structure made from S-
901 glass/SCI REZ 081 resin consists of nested conical shells, with integral
metal attach rings, as shown in Fig. 7. This structure is designed to support a
cryogenically cooled infrared telescope. Whereas the straps can support only
tensile loads, the conical shells can take tension, compression, and bending.
Filament wound struts, cylindrical in shape with integral metal end attach-
ments, have been produced for spacecraft cryogenic use in applications
where the isolators must sustain tensile and compression loads.
FIG. 7—Thermal isolator shells for cryogenic instrument.
Cryogenic Composite Strap Materials and Fabrication

Glass Fibers and Resins
Representative properties of S-901 glass filaments and S-901 glass/SCI
REZ 080 or 081 epoxy resin unidirectional wound composites are shown in
Table 2. S-901 glass fiber is produced by Owens-Corning Fiberglass as con-
tinuous roving for high-performance structural applications; it is purchased
under specification MIL-R-60346. The roving is preimpregnated with a
proprietary epoxy resin system, either SCI REZ 080 or SCI REZ 081, before
being used for strap fabrication. The specific resin systems, compounded and
processed by SCI, were selected for their high strength, exceptional interlam-
inar shear properties, and serviceability at temperatures ranging from cry-
ogenic to 121°C (250°F). Fiber-to-resin ratio is controlled carefully to insure
part-to-part uniformity and reproducibility. Delivered strength characteris-
tics are a function of strap fabrication and geometry, as will be discussed
later.
Filament Winding and Cure

Straps are fabricated on precision hard tooling, designed to permit fila-
ment winding of the parts under optimum tension levels to result in a band
configuration with high fiber content, low void content, proper width and
thicknesses, and close tolerance net dimensions. Fiber and resin contents in
the composite are about 70 and 30 v/o, respectively; void content is less than
1 v/o. Final resin cure is at 149 to 177°C (300 to 350°F).
TABLE 2—Typical S-901 glass/epoxy uniaxial filament-wound composite material properties.
Filament
Ultimate strength, psi 650 000
Elastic modulus, psi 12.4 X 10'
Density, lb/in.' 0.088
Composite
Filament fraction in composite, volume percent 67
Density, lb/in' 0.072
Longitudinal modulus, psi
350° F 8.3 X 10'
75°F 8.3 X 10'
-320°F 9.1 X 10'
Longitudinal tensile ultimate strength, psi
350°F 174 000
75°F 220 000
-320° F 275 000
Coefficient of thermal expansion, in/in, °F
75 to -320°F 1.6 X 10"'
75 to 350°F 1.4 X 10"'
Note—Conversion factors: 1 psi = 703.1 kg/m'; 1 lb/in.' = 27 680 kg/m'; °C = 5/9 (°F - 32);
1 in. = 2.54 cm.
Acceptance Testing
In addition to dimensional inspection, straps often are given acceptance
tests, which include proof loading and spring constant determination. In
some applications, the spring constants strap-to-strap must be within a very
close tolerance range, and straps are segregated into matched sets. Some
straps, for example, the PRSA straps, are strain gaged on each leg and each
part is individually calibrated as a spring; strain gage output, extensometer
output, and load pin-to-load pin displacement are measured as a function of
load, and must be within specified limits. Usually, lot acceptance witness
specimens are subjected to destructive fatigue and static testing in order to
certify a production lot.
Strap Design and Performance Data
Static Strength
Delivered strap static strength at room temperature is a function of strap
geometry, specifically the load pin diameter-to-leg thickness ratio, D/t. A
preliminary design curve for nominal strap strength at ambient temperature
is shown in Fig. 8, based on more than 100 tests. Delivered strength is higher
for "thinner" composites, while "thick" composites result in lower strength,
as might be expected.
Composite strength increases considerably in going from ambient to cry-
ogenic temperatures (about 25 to 35 percent). However, most applications
Strap Bend-Diameter to Leg-Thickness Ratio, D / t
FIG. 8—Static tensile strength of S-glass/epoxy suspension straps at ambient temperature.
with which we are familiar have a "warm" strap end, so that the higher
strength associated with cryogenic temperature exposure is not used to ad-
vantage in hardware design.
Modulus and Spring Constant

Straps have a linear load-deflection relationship to ultimate strength
failure.
The theoretical unidirectional tensile modulus of S-901 glass/epoxy resin
composite is 5624 X 10* kg/m^ (8 X lO' psi). However, for the strap, the ap-
parent modulus of elasticity has to be evaluated with consideration to strap
geometry. Strain gages and extensometers measure only point and gage
length strain, respectively, without regard to elongation in the curved end
sections during loading. Therefore, the effective gage length, Lp, is assumed
to be one half the band perimeter, as follows
, _ (2L + irD)
^P- ~2
in which L is the straight leg length and D is the load pin diameter. This value
is then used in the classic expression for the modulus of elasticity
AL/Lp
in which P is the tensile load, A is twice the cross section area of one leg (2
IVt), and AL is the axial deflection between load pin centers. From this rela-
tionship, it is obvious that the straps have an apparent modulus less than
theoretical, with short straps having large pin diameters yielding the lowest
values.
Since the straps are preloaded in tension to prescribed levels in each spe-
cific application, precise knowledge of their spring constant is essential.
Strap spring constant (K) is defined by the relationship
K = P/AL
As previously mentioned, normal strap acceptance test procedure is to
measure spring constant for each strap produced over the load range of
interest.
Precise thermal conductivity measurements have been made on S-901
glass/SCI REZ 080 straps by the National Bureau of Standards over the
temperature range 2 to 280 K (—456 to 45°F). Results are shown in Fig. 9*.
* Hust, J. G. and Arvidson, J. M., "Thermal Conductivity of Glass Fiber/Epoxy Composite
Support Bands for Cryogenic Dewars," Report 275.03-78-2, National Bureau of Standards,
Feb. 1978.
100 —1—r—rr'TTT— -T 1 1' 1 T-r 1 1 1 1 1

f\ ^'
slO-»
"- /
•
SO
/ -
60
- / -
^ 40 - / ^ ,
1
3
^
_ ^ y^ •
>
- -
-0 /
0
U
H 10 - / -
R
- / -
6
y -
4 7 ,, 1 1 1 J 1 1 1
, , 1 1 1 1 1 1 1 1
,
ar "1 (, 8 10 . 20 40 60 80 100 200 400
FIG. 9—Thermal conductivity of composite bands.
Outgassing
Standard evaluations performed to assess vacuum compatibility of com-
posite materials for spacecraft applications include determination of vacuum
weight loss and volatile condensable material (VCM). These tests have been
established as a screening procedure for nonmetallics considered for use in
NASA spacecraft. The criteria used for selection are that a maximum of 1.0
percent weight loss and a maximum of 0.10 percent VCM are permitted as a
result of a 24 h pulverized material bakeout at 125°C (257°F) in a
<6.6661 X 10'' Pa (<5 X 10"' torr) test chamber with VCM collected on a
25°C (77°F) substrate.
The Jet Propulsion Laboratory performed tests of SCI REZ 080 and SCI
REZ 081 epoxy following their procedure for vacuum outgassing of poly-
mers (micro-VCM technique) with acceptable results, as shown in Table 3.
TABLE 3—Outgassing test data for S-90I glass/SCI REZ 080

and SCI REZ 081 epoxy resin composites
Material
SCI REZ 080 SCI REZ 081
Manufacturer SCI SCI

Mix ratio A/R A/R
Cure A/R A/R
Weight loss, percent 1.07, 0.95 0.89, 0.93
VMC, percent 0.00, 0.00 0.00, 0.00
Deposit very, very faint film none
TTTTTT] 1 I I I llll[ 1 I I lllll{ 1 I I llll|{ 1 I I I I IIT
) • NBS Study O Hofrr. OlsiTi (H.-l. <)
Mran. R - O.OS
1.1 l l l l l I I I I Hill I I I I mil I I I

FatiRuc C y c l e s , N
FIG. 10—Room temperature fatigue data.
Fatigue
Ambient and cryogenic temperature fatigue tests have been performed on
thermal isolator suspension straps under several programs. Results obtained
are similar to characteristics determined for glass/epoxy flat laminates.
The National Bureau of Standards (NBS) conducted fatigue tests at room
temperature and 4 K (-452°F) on SCI fabricated S-901 glass/SCI REZ 080
filament wound support straps.'' Room temperature fatigue data are shown
in Fig. 10, where the maximum fatigue stress is plotted as a function of the
static strength. Fatigue data obtained at liquid helium temperature are added
to the room temperature data in Fig. 11. These data clearly demonstrate a su-
perior fatigue resistance at 4 K (—452°F), about an order of magnitude
greater than at 295 K (72°F). Although a reliable S-N curve at 4 K (-452°F)
could not be constructed without more extensive data, it appears that the
stress level required for failure at a specified number of cycles may approach
twice the value required at room temperature. The improvement in fatigue
resistance at 4 K (—452°F) exceeds that which might be predicted based on
scaling the fatigue stress levels to account for the increase in static strength
alone.
' Tobler, R. L. and Read, D. T., "Fatigue Resistance of a Uniaxial S-Glass/Epoxy Composite
at Room and Liquid Helium Temperatures," National Bureau of Standards, 1976.
' Hofer, K. E. and Olsen, E. M., "An Investigation of Fatigue and Creep Properties of Glass
Reinforced Plastics for Primary Aircraft Structures," Report M6104, Illinois Institute of Tech-
nology Research Institute, April 1967.
I iiiii|—I I Iiiiii|—I I Iiiiii{—I I I Ii i i i | — I I II nil
295«K, Mean
"I M i n i I I I Mini I I I iiiiil il I I I I III!
Fatigue Cycles. N
FIG. 11—i K fatigue results.
-1—I—I I I 111'l 1—I I I 11 I I I I I I 11 i i j 1—I—I I I I I I
Q PRSA Strap H^ R = 0.43
Q PRSA Strap O , R = 0.28
Figure 10. R = 0,1
QQ O
I I , I ...il .J I I I 11 I I I I I 1111 I I I I 1111
Fatigue Cyclee. N
FIG. 12—Room temperature fatigue data.
Typical ambient temperature tension-tension axial fatigue test results for

some space shuttle PRSA support straps are shown in Fig. 12. For compari-
son purposes, the NBS data of Fig. 11 are included.
Conclusion
Materials, processes, and design data have been developed for composite
thermal isolators, and these cryogenic structures have been used successfully
for a range of sophisticated spacecraft applications. With increasing use and
cost of cryogens, new trends are expected in the use of composites with cryo-
gens for industrial products, in addition to space applications.
C. L. Leung^
Space Environmental Effects on

Graphite/Epoxy Composites
REFERENCE: Leung, C. L., "Space Environmental Effects on Graphite/Epoxy

Composites," Composites for Extreme Environments, ASTMSTP 768, N. R. Adsit, Ed.,
ABSTRACT: In a study of T300/934 graphite/epoxy composite, which forms the

composite skin of the space shuttle cargo bay doors, the following composite proper-
ties were investigated as a function of increasing gamma-radiation exposure dosages:
directional moisture absorption/desorption kinetics, glass transition temperatures, in-
terlaminar shear strength and damping peaks (tan 8). Irradiation conditions were:
open to atmosphere and vacuum, ambient and high temperature (lOCC).
Four dosage levels were investigated: 4 X 10', 8 X 10', 1.4 X lO', and 3.2 X 10* rads,
the final dosage level being equivalent to a three-year orbit life. Directional moisture
diffusion studies indicate anomalies in the three diffusing axes (translaminar, transfi-
brous, and interlaminar). Interlaminar shear strengths showed an initial increase, fol-
lowed by a decrease, as the dosages increased. Although the glass transition tempera-
tures remained essentially unchanged, the peak heights of the tan 6 measurements also
showed an initial increase, and then decreased as radiation dosages increased.
KEY WORDS: space environments, gamma radiation, moisture diffusion, dynamic

properties, interlaminar shear strength, composites
Space environments can have several effects on composite materials. The

vacuum will act to dry out such materials, adversely affecting dimensional
stability. Surface coatings may delay, but never stop, the process of degas-
sing. If the composite materials are not coated and exposed to the sun, solar
ultraviolet radiation can adversely affect surface properties. However, the
greatest effect from space environments is a decrease in the mechanical
properties due to high-energy radiations.
A basic parameter that determines the radiative degradation of organic
materials and their composites is the type of radiation involved in the proc-
ess. Radiations composed of photons or particles with higher energies than
those encountered in binding electron orbitals are called high-energy radia-
tions, for example, y-rays and high-energy electron beams. The essential fea-
ture of high-energy irradiation of an organic polymer is that, on passing
through matter, the radiation deposits energy by the interaction with bond-
' Technical staff member, Rockwell International Science Center, Thousand Oaks, Calif. 91360.
110

Copyright® 1982 b y A S T M International www.astm.org
LEUNG ON SPACE ENVIRONMENTAL EFFECTS 111
ing electrons [7].^ These may acquire sufficient energy to be removed from
the macromolecular chain to form ionic systems. These electrons also may
be excited to higher nonbonding energy levels, forming radicals. These ionic
and radical species then may react with the surrounding materials to form
new chemical species.
Selection of Space Environmental Effects

The primary interaction of high-energy radiation with organic matter is
through the physical processes that directly involve the atomic nuclei or sur-
rounding electrons, and is independent of the nature of the chemical bonding
or structure of the system. In organic systems, the primary effect of energy
absorption from either gamma rays, neutrons, or high-energy electron beam
is that energetic electrons are produced randomly throughout the material,
and the majority of the property changes are caused by this internal electron
bombardment. As long as the radiation has energies that are high when
compared to the bonding energies and excitation or ionization potentials of
the chemical systems under investigation, the choice of radiation source
seems to depend mainly on its penetrating power through the organic matter.
In other words, the extent of radiative degradation and the type of active in-
terrhediates depends on the energy of the incident radiation. Although high-
energy electron beams can be more energetic than 7-rays, they are not as
penetrating. Radioactive cobalt (Co-60), which has a half-life of 5.3 years, is
used widely as a source of 7 radiation, since beams of this material have a
penetration of at least several inches. High-energy electron beams from elec-
tron accelerators are absorbed mainly on the surface of the material (<0.1 mil
penetration depth), although the total surface dosage can be quite high (be-
tween 10' to 10'^ rads). Since in many applications total thickness of the
composite components can be quite high, the penetration power of 7 radia-
tion seems more desirable in radiative degradation studies.
The equilibrium temperatures of spacecraft components depend on the
ratio of solar absorbance (a,) to thermal emittance (e) for the spacecraft sur-
face. For a typical geosynchronous orbit, this a,/e ratio produces tempera-
tures from approximately 20 to 50°C. Internal sources of heat (primarily due
to electronics) can extend this temperature range to approximately 100°C.
While transients during eclipse conditions can lower the surface temperature
to less than —60°C, the heat capacity of the spacecraft and the internal sources
of heat generally keep internal temperatures above ICC. Therefore, the be-
havior of structural materials over the temperature range of 20 to 100°C is of
paramount importance in space applications.
The vacuum of space varies from ~10"* torr (at a 100-km ahitude) to less
than 10"'" torr (at altitudes greater than 1000 km). Even at a 200-km altitude,
the molecular mean path, heat transfer rates, and degassing rates are such
^ The italic numbers in brackets refer to the list of references appended to this paper.
that the bulk properties of the composites will be essentially equal to that at
a lOOO-km altitude. Therefore, the vacuum chosen (10"' to 10"' torr) in the
laboratory is sufficiently small that reducing it even more would have essen-
tially no effect on results obtained during the irradiation exposure.
When cured, epoxy resins are above average in radiation resistance in
comparison with other thermosetting resins [2]; this is very likely due to their
rigidity and aromatic content. Of itself, the chemical composition of a par-
ticular resin influences its resistance to radiative degradation. For example,
when irradiated with -y radiation, glycidyl amine resins cured with aromatic
diamine-based hardener were found to show little deterioration of flexural
strength at doses up to 10'° rads, while the glycidyl ether/aromatic diamine
system shows a 50 percent reduction in flexural strength at about 4 X lO'
rads [5].
The presence of graphite fibers is likely to affect the photochemical behav-
ior of the matrix material. Besides being conductors for free electrons, graphite
fibers may transfer the absorbed radiation energy back into the polymer,
thereby promoting degradation. Graphites and carbons as a class are more
susceptible to property changes from irradiation than other inorganic ceramic
oxides, such as beryllium oxide, alumina, and silica. It has been reported [2]
that the chemical reactivity of graphite is increased in the presence of radia-
tion, mainly by oxidation in the presence of carbon dioxide, oxygen, water
vapor or all three. These oxidation products may consist of reactive species
which are retained on the fiber surface and then interact with the matrix poly-
mer, affecting the bonding between the fiber-matrix interface.
Experimental
A 48-ply unidirectionally reinforced composite panel (Fiberite 934 rein-
forced with Union Carbide T300 graphite fibers) was fabricated and cured
by standard production procedures. The cured composite has a fiber volume
Vf = 0.60 and a volume fraction of voids Vy < 0.01.
For short beam interlaminar shear tests, specimens 0.015 m long and 0.004
m wide were cut with a diamond cutoff wheel to a thickness of 0.002 m.
Thin plate specimens for moisture absorption measurements were cut to
provide dimensions of length by width by thickness of (20 by 5 by 1) X 10"' m.
Figure 1 details the specimen orientation which exposes the interlaminar,
transfibrous, and translaminar faces, respectively, as the major exposed
areas (80 percent of total area) in moisture absorption measurements [6].
In studying the dynamic mechanical properties of the composites by the
Rheovibron, fiexural specimen geometry was used, as discussed by Massa
and coworkers [7,8]. This specimen geometry recently has been successfully
used in studying environmental effects on the mechanical properties of
graphite/epoxy composites [P, 10]. A schematic of the flexural specimen as-
Interlaminar surface (i - 3)
Transfi&rous surface {i » 1)
TA
fT TransTaminar surfaca ( i = 2}
FIG. 1—Specimen orientation and geometry for kinetics of absorption in interlaminar, transfi-
brous, and translaminar directions of uniaxial reinforced composite (layers represent plies and dots
fiber ends).
sembly is shown in Fig. 2. The specimens, in the form of rods, were ma-
chined to about 0.0015 m in diameter.
Specimens for interlaminar shear, dynamic mechanical, and moisture dif-
fusion measurements were inserted in small vials. For irradiation under
vacuum, the vials were evacuated to about 10"' torr, and then sealed. High
temperature irradiation was achieved by placing the vials inside a thermo-
statted dewar which was regulated at 100°C.
Irradiation was achieved by a Co-60 source having a flux rate of 2.57 X lO'
rads/h. For both the evacuated and open vials, four different dosage levels
were used: 4.4 X lO', 8.8 X 10\ 1.4 X 10*, 3.2 X 10* rads. Both room
temperature and 100°C were used during irradiation. The highest dosage level
(3.2 X 10* rads) is equivalent to the total radiation received in three years in
a geosynchronous orbit.
After the appropriate radiation dosages were received, the vials were
broken, and the specimens were stored in a dry desiccator until use.
Interlaminar shear (short beam) strengths were measured on a three-point
loading device. Moisture diffusion measurements were measured by immersing
the specimens in water at 75°C, and periodically weighing the moisture up-
take. Glass transition and tan 6 measurements were obtained on a Rheovibron
FIG. 2—Schematic representation of the Rheovibron flexural specimen assembly.
Dynamic Viscoelastometer (Model RHEO-200), scanning from a temperature

range of 20 to about 300°C.

Moisture Absorption
The results of the moisture diffusion analysis of the specimens as a function
of radiation dosage and irradiation conditions are tabulated in Table 1. Also
shown in the table are the directional diffusion constants of the baseline (not-
irradiated) specimens. Table 1 shows that, as expected for unirradiated spec-
imens, transfibrous diffusion coefficients are higher than translaminar and
interlaminar diffusion coefficients. However, y radiation produced different
effects on the directional diffusion coefficients of the composites. While the
translaminar and interlaminar diffusion coefficients increased with the radi-
ation dosages, the transfibrous diffusion coefficients remained almost un-
TABLE I—Moisture diffusion of irradiated 934/T300 composites at 75°C.

oC Diffusion Coefficient (10" 11 „ 2 / . ^
m /s)
0 4.4 X 10' 8.8 X 10' 1.4 X 10' 3.2 X 10'
Temperature, rads rads rads rads rads
Transfibrous 25 0.96 1.03 1.22 0.94 0.83

Translaminar 25 0.24 0.39 0.40 0.71 0.68
Interlaminar 25 0.22 0.28 0.43 0.47 0.58
Transfibrous 25,V 0.97 0.75 0.68 0.56
Translaminar 25,V 0.24 0.69 0.58 0.57
Interlaminar 25,V 0.22 0.56 0.59 0.60
Transfibrous too 0.96 0.90 0.87 0.91 0.90
Translaminar 100 0.24 0.42 0.46 0.55 0.58
Interlaminar 100 0.22 0.43 0.45 0.48 0.50
Transfibrous lOO.V 0.96 0.95 0.87 0.99 0.85
Translaminar 100,V 0.24 0.49 0.48 0.49 0.53
Interlaminar 100,V 0.22 0.44 0.44 0.51 0.56
°V = in vacuum.
"T—I—I—\—I—I—r "1—\—I—r 1 r "I—r

O lOO'coPEtg
• 100°C VACUUM
O 25°C OPEN
* 2S°C VACUUM
D BASELINE
RADS (X10'|
FIG. 3—Loss tangent {tan S) of composites irradiated with y radiation.
changed and even decreased slightly. Reference to Fig. 1 shows that the major
diffusion paths for the interlaminar and translaminar surfaces are through
the matrix resin, and that for the transfibrous surface it is through the fiber-
matrix interface. Therefore, the data in Table 1 suggest that while y radiation
damaged the matrix resin (by microcracking, perhaps), it improved the
matrix-fiber interaction, and thus lowered the moisture diffusivity along the
matrix-fiber interface.
Dynamic Mechanical Measurement

Results for the Rheovibron thermal scans forflexuraldamping loss tangent
(tan 8 = G"/G') are shown in Fig. 3, and the glass transition temperatures,
taken as the maximum in the tan 8 versus temperature plot (tan 6max), are
shown in Table 2. It can be seen that the glass transition temperatures of the
specimens generally decreased from a baseline value of 260°C to around
225°C, the decrease being within the range reported in the literature [5].
However, a much more interesting observation is shown in Fig. 3: the loss
tangent of the composites initially decreased, then increased again as total
radiation exposure increased.
TABLE 2—Glass transition temperatures in irradiated 934/T300 composites, °C.
Total Dosage, Open Vacuum Open Vacuum

rads RT RT 100=0 100°C
0 260
4.4 X 10' 231 230 236 235
8.8 X 10' 230 230 239 230
1.4 X 10' 234 235 240 238
3.2 X 10' 225 238 237
The flexural geometry of the specimen, as shown in Fig. 2, is unique in

that it is very similar to that of a three-point interlaminar shear strength con-
figuration, so that the measured loss tangent at the glass transition region
can be viewed as the combined contributions due to the matrix resin and the
fiber-matrix interface, tan d being a measure of the internal friction or ratio
of energy dissipation within the material. However, as pointed out by Kenyon
and Neilsen [11], additional postcuring (either by heat, or in this case, by y
radiation) of the matrix resin would have increased the magnitude of the tan
8 peaks instead of the decreasing tan 8 values as shown in Fig. 3. Therefore,
it is speculated that exposure of graphite/epoxy composites to y radiation
probably improves the fiber/matrix interfacial bonding (or interaction) ini-
tially. Indeed, Raff and Subramanian [12] reported that the strength of joints
between stainless steel or copper and epoxy resins can be increased up to 300
percent by y radiation.
Interlaminar Shear Strength

Short beam interlaminar shear strengths of the specimens are shown in
Fig. 4. Each individual data point is the average of ten specimens. It can be
seen that exposure of the specimens to y radiation improved the inter-
laminar shear strength of the composites initially, but the shear strength
started to decrease to the original baseline values at about 3 X 10' rads, sim-
ilar to a three-year service life in space orbit. The results of Figs. 3 and 4 do
indeed correlate with each other: as the fiber/matrix interfacial interaction
improved, loss tangent (as measured by flexural dynamic Rheovibron tech-
_I 1 1 I 1 I L — I 1 1 i 1—
to 12 14 ie IS 20 22 24 20 2> » 32
RADS (Xio'l
FIG. 4—Interlaminar shear strength of composites irradiated with y radiation.
nique) decreased, and therefore the interlaminar shear strength improved.

Long-term exposure ultimately becomes detrimental, as shown by the in-
crease in loss tangent and decrease in interlaminar shear strength.
Summary and Conclusion

This paper describes the effect of 7 radiation exposure in 934/T300 com-
posites. The results indicate that for this graphite/epoxy system, the mois-
ture sensitivity, dynamic mechanical, and shear strength generally will be
comparable to that of the unexposed system after about three years of serv-
ice (about 3 X 10' rads). Indeed, the performance of the material may even
be improved upon initial operation in space environments.
A cknowledgments
This research was performed under Rockwell International's Independent
Research and Development Program. The technical assistance of G. Lind-
berg and L. Bivins is greatly appreciated.
References
[/] Kusy, R. P., Macromolecules, Vol. 9, 1976, p. 791.
[2] King, R. W., Broadway, N. J., Mayer, R. A., and Palinchak, S. in Effects of Radiation on
Materials and Components, J. F. Kircher and R. W. Bowan, Eds., Rheinhold, New York,
1964, Chapter 3.
[3] Morgan, J. T. and Stapleton, G. B., "Post Irradiation Mechanical Properties of Radiation
Resistance Cast Epoxy Resin Systems," Report RL-75-136, Science Research Council,
Chilton, England, July 1975.
[4] Brown, G. L., Thomasson, J. F., and Kurland, R. M., National Society for the Advance-
ment of Material and Process Engineering Symposium Exhibition Proceedings, Vol. 24,
No. 2, 1979, pp. 1021-1031.
[J] Haskins, J. F. and Holmes, R. D., "Advanced Composites Design Data for Spacecraft
Structural Applications," AFML-TR-79-4208, Air Force Materials Laboratory, March
1980.
[5] Leung, C. C , Dynes, P. T., and Kaelble, D. H. in Nondestructive Evaluation and Flaw Crit-
icalityfor Composite Materials, ASTM STP 696, American Society for Testing and Mate-
rials, 1979, pp. 298-315.
[7] Massa, D. J., Journal of Applied Physics, Vol. 44, No. 6, 1973, pp. 2595-2600.
[8] Massa, D. J., Flick, J. R., and Petrie, S. E. B., American Chemical Society, Division of
Organic Coatings and Plastics Chemistry, Preprints, Vol. 35, 1975, pp. 371-376.
[9] Kaelble, D. H., "Nondestructive Test for Strength Degradation in Composites, First Year
Report," Contract F33615-74-C-5180, Advanced Research Projects Agency, Arlington,
Va., and Air Force Materials Laboratory, Wright Patterson Air Force Base, Ohio, Sept.
1975, pp. 232-250.
[70] Dynes, P. J. and Kaelble; D. H., Journal of Applied Polymer Science, Vol. 22, 1978, pp.
837-840.
[11] Kenyon, A. S. and Nielson, L. E., Journal of Macromolecular Science, Chemistry, Vol. A3,
No. 2, 1969, pp. 275-295.
[12] Raff, R. A. F. and Subramanian, R. B., Journal of Polymer Science, Vol. 15, 1971, pp.
1535-1537.
p. Shyprykevich^ and W. Walter^
Effects of Extreme Aircraft Storage

and Flight Environments on
Graphlte/Epoxy
REFERENCE: Shyprykevich, P. and Wolter, W., "Effects of Extreme Aircraft Storage

and Flight Environments on Graphite/Epoxy," Composites for Extreme Environments.
ASTM STP 768, N. R. Adsit, Ed., American Society for Testing and Materials, 1982,
pp. 118-134.
ABSTRACT: The ability of graphite/epoxy composites to absorb moisture from the

environment and the consequent degradation of resin-dependent strength properties is
well understood under conditions of constant temperature and humidity. However, if
the environment consists of high humidity exposure with intermittent high tempera-
ture excursions, anomalous diffusion behavior occurs accompanied by further reduc-
tions in strength. This paper reports the results obtained from a test program in which
graphite/epoxy specimens were exposed to a tropical environment representative of
extreme aircraft runway storage and frequent I27°C temperature excursions of long
duration representative of supersonic flights.
The tests were conducted in a specially developed apparatus which held specimens
in chains of five each with accompanying travelers while exposing the specimens to
prescribed daily environmental conditions. The duration of the real-time simulation
test was IS months. After various intervals of exposure, weight data were gathered
using accompanying travelers, and residual static tests were performed at room
temperature and 127°C. The test results for the extreme environment are compared to
test results obtained for specimens exposed to the same flight profile, but temperate
runway storage. The results indicate that the moisture gains of the extreme environ-
ment specimens are much greater than what is thought possible to be held by the resin.
Since the static compressive strength is reduced with increases in moisture content, the
degradation in compressive strength is greater than what was expected.
A semiempirical transport model based on Fickian diffusion and hygrothermal his-
tory dependent material properties was developed to account for changes in absorptiv-
ity characteristics as a function of glass transition temperature (T,) exceedances. The
test data for model development were obtained from an accelerated environment test
which attempted to find economic alternatives to real-time exposure tests. The mois-
ture predictions with the new model gave reasonable correlation with real-time test
observations.
KEY WORDS: composite materials, graphite/epoxy, moisture absorption, strength

properties, glass transition temperature exceedance
' Project engineer. Advanced Composites, Grumman Aerospace Corp., Bethpage, N.Y. 11714.
^Technical specialist. Structural Mechanics, Grumman Aerospace Corp., Bethpage, N.Y.
11714.
118

SHYPRYKEVICH AND WOLTER ON AIRCRAFT STORAGE 119
TENSION - TYPE I AND III GRAPHITE/EPOXV
COMPRESSION - TYPE II AND IV LAMINATE
FIBERGLASS/EPOXY
GRIPPERPADt4)
BONDED TO SPECIMEN
^^^ I FIBER ORIENTATION:

3/16 IN. DIA Tl HI-LOK-"^
INSTALLED WET TYPE I |0/.«5«)ls
TYPE II (0/145/01 s
TYPE III IO/!4S/9a2/I45/0l f,
TYPE IV |0/t45/02/:45/0)s
FIG. 1—Specimen configuration.
Real-time test data were generated by simulating runway storage at

Mather Air Force Base and Andersen Air Force Base, temperate and tropi-
cal environments, respectively, and the B-1 flight profile. The two bases were
selected from world-wide weather data and U. S. Air Force (USAF) aircraft
(bombers) basing information.' The tropical environment when combined
with a weekly supersonic flight profile containing a 10-min hold at 127°C is
considered extreme for the epoxy based material, since the exceedances of
the glass transition temperature (Tg) were frequent with significant penetra-
tion through the thickness of the specimens. The simulated operational envir-
onment is also extreme for the given aircraft. It was assumed that the aircraft
spends all its service life in the tropics and flies a supersonic mission once a
week.
The approach was to measure the effects of the extreme environment using
specimens for tensile and compressive strength determinations and accom-
panying travelers for moisture tracking. Comparisons also were available to
specimens exposed to temperate and to accelerated environments.
Specimen Design and Fabrication

There were four types of specimens varying in thickness and fiber orienta-
tion; however, the overall specimen configuration remained constant as
shown in Fig. 1. The unloaded bolt in the specimen test section induced fail-
ures in the test section, provided stabilization, and, at the same time, was rep-
resentative of actual aircraft usage without undue complexity. Specimens
'Whiteside, J. B. et al, "Environmental Sensitivity of Advanced Composites, Vol. 1, Envir-

onmental Definition," AFWAL-TR-80-3076, Air Force Wright Aeronautical Laboratories,
Wright-Patterson Air Force Base, Ohio, Dec. 1980.
were 8-ply, Type I (0/±45/90)., Type II, (0/±45/0)., and 16-ply, Type III,
[(0/±45/90)s]2, and Type IV [(0/±45/0)s]2. Types I and III were used to
monitor static tensile strength. Types II and IV monitored static compressive
strength. The tension specimens were quasi-isotropic laminates which have
been widely used in other generic studies. The compressive specimens were
similar to the tension specimens, except that the 90-deg plies were replaced
by 0-deg plies to induce matrix-type failure in compression.
Hercules AS/3501-5A graphite/epoxy prepreg was the material, and all
panels were made from the same batch. The laminates were cured using a 1 h
hold at 107°C under vacuum plus 585 kPa, followed by 1 h at 177°C under
585 kPa pressure. The laminates were cooled down slowly to minimize micro-
cracking. The panels were then oven postcured for 8 h at 177°C.
The cured panels were ultrasonically and dimensionally inspected. Next,
the panels were sawed into plates sufficiently large to produce 15 to 25 test
specimens and the same number of 5.0 by 2.4 cm associated travelers. These
travelers were cut from the specimens after the fiberglass tabs were bonded
to the specimens. The edges of the specimen (two sides) and travelers (four
sides) were ground to final dimensions. The fabrication process was com-
pleted by drilling holes in both the specimens and travelers, and locally spot-
facing the travelers for accurate thickness measurements. After dimensional
and visual inspection and installation of the boh, each specimen and the as-
sociated traveler were packaged in sealed bags.
Test Procedures
The static tests were performed on a universal testing machine of the
constant-rate-of-head movement type. The test load was applied in accord-
ance with the standard practice of 0.13 cm/min. Strain measurements were
made on a representative number of specimens using two resistance strain
gages adhesively bonded in back-to-back fashion on the faces of the speci-
mens, centered between the tab and the center hole.
The tension tests were performed using the methods currently practiced by
industry and described in the U. S. Air Force Guide for Advanced Compos-
ites. The specimens were gripped and an axial load applied until failure
occurred.
The compressive tests were performed using an Illinois Institute of Tech-
nology Research Institute (IITRI) compression tool modified to accomodate
the 2.54-cm-wide specimens. The fixture incorporates a gripping arrange-
ment similar to (reversed) that used for tension testing, thereby introducing
load from the grip to the specimen by shear rather than by end loading. Buck-
ling was observed by strain gage reading during initial tests of Type II (8-ply)
specimens. Therefore, a Teflon-faced fixture for supporting this specimen
along its edges was designed, fabricated, and used. The unloaded bolt in the
center of the specimen for both 8 and 16-ply specimens was used as a contact
WET BULB
THEBMISTOB 30 T/C SPECIMENS
INEACHMFR
f VIEW SIDE VIEW
FIG. 2—Environmental conditioning system.
point for midpoint stabilization, thus effectively reducing the unsupported

length by a factor of two.
The environmental tests were performed in the apparatus shown schemat-
ically in Fig. 2. Environmental units each containing specimens in chains of
five were charged with 60 specimens each, and cycled through daily profiles
(one for each month) simulating the B-1 horizontal stabilizer upper surface
of a white aircraft parked outdoors. The daily variation for each mbnth is
given.' For the real-time test this information was turned into control data
for the test apparatus.
A set of control cams was required to program a typical daily cycle for
each month. The temperate environment has distinct seasonal variations,
which required four cams. One cam sufficed for the tropics, which do not
experience any significant seasonal variations. The temperate environment
was started with a daily cycle typical of January. These conditions corre-
spond to a yearly average of approximately 26°C, 85 percent relative humid-
ity and 16°C, 68 percent relative humidity for the tropical and temperate lo-
cations, respectively.
Approximately once each week a supersonic flight temperature profile
typical of B-1 aircraft was imposed on the specimens. The idealized flight en-
vironment is shown in Fig. 3. In testing, the terrain-following segment was
somewhat simplified. Before and after each flight, travelers were weighed
carefully and the weight recorded.
Unfortunately, at the end of testing, it was discovered that copper and sil-
icon deposited on the travelers in a random fashion, invalidating all inter-
mediate measurements. It was possible, however, to reconstruct the moisture
gain history using final weight measurements of specimens which were re-
moved from the environmental units after 2, 5, 8, 12, and 15 months expo-
sure. The travelers were dried until they reached an equilibrium weight level.
The difference between this weight and the initial dry weight was assumed to
be the weight of the extraneous deposit and was subtracted from the final
weight measurement of the traveler. The corrected weight then was used to
calculate the moisture gain of the traveler.
FIG. 3—Flight thermal profile.
Real-Time Test Results
Moisture Uptake
The test moisture gains are plotted in Figs. 4 and 5 for the temperate and
tropical storage conditions. Figures 4 and 5 also include analytical weight
gain predictions based on "virgin" (constant) material diffusion characteris-
tics. The virgin material predictions were calculated by a finite-difference
formulation of Fickian diffusion equation* with equilibrium moisture con-
ANALYSIS - TROPICAL LOCATION
ANALYSIS - TEMPERATE LOCATION
EXPOSURE. MONTHS
FIG. 4—Moisture gain history of 8-ply graphite/epoxy.
'Schneider, P. J., Conduction Heat Tranter, Addison-Wesley, Reading, Mass., 1955.
EXPOSURE, MONTHS
FIG, "5—Moisture gain history of 16-ply graphite/epoxy.
tent related to relative humidity by

relative humidity V '
1 percent
100
where Av = 1.86, and diffusivity related to temperature, T (°K), by
-6095 \
D = Dv exp I ———I mm^/s
where Z)„ = 11.657. The boundary conditions were daily variations for each
month.' The band in the analytical predictions is the result of the drying ef-
fect of the weekly supersonic flight. The seasonal effect is evident for the
temperate environment. When the analysis was continued beyond the
15-month duration shown in Figs. 4 and 5 the predicted equilibrium mois-
ture contents were 1.17 and 0.52 percent for the tropical and temperate loca-
tions, respectively.
For the tropics (solid symbols in Figs. 4 and 5) the average moisture gains
were much greater than the predicted initial analysis. Furthermore, equilib-
rium moisture level was not reached, even for the 8-ply specimens, making
moisture levels predictions for 20-year aircraft life uncertain. The divergence
of the analysis from the test data occurred quite early, before the end of the
2-month period. The test moisture gains (2.5 to 3.0 percent) after 15 months
of tropical exposure were much greater than the 1.86 percent saturation level
at 100 percent relative humidity assumed for this material system.
For the temperate environment, moisture gains of the specimens are
shown in Figs. 4 and 5 by open symbols. A moisture equilibrium level was
reached at about 1 percent for the 8-ply specimens. This is approximately
double the value that was predicted by analysis. Some of the discrepancy be-
tween test and analysis was attributed to the malfunctions of the wet bulb
sensor in the test apparatus. For example, between days 160 and 185 several
exposures to 100 percent relative humidity at temperatures of 10 to 24°C (the
equivalent of rainy days) caused a noticeable increase in moisture absorbed
in the specimens. Two flights were applied during this period, and the re-
sponse of the specimens was affected by this experience. However, the corre-
lation between test and analysis for the temperate environment is much better
than for the tropical environment. The moisture uptake trends between test
and analysis are similar including the effect of seasonal variations of relative
humidity in the exposure cycle, especially noticeable for the 8-ply specimens.
Residual Static Strength Tests

After various intervals of real time exposure, residual static tests were per-
formed at room temperature and 127°C. The data are plotted in Figs. 6 and 7
in terms of exposure time, with gross failing stress as the static strength pa-
rameter for the two static test temperatures. The tensile strength (Fig. 6) was
found to be insensitive to exposure time for either tropical or temperate
runway storage locations. All tension specimens failed across the hole, as ex-
pected. The compressive strength was reduced, especially at 127°C, as shown
in Fig. 7. The failure mode of Type II specimens was local instability. For
Type IV specimens, the predominant failure mode was compressive laminate
failure. For the two Types IV low points (8 and 15 month tropical exposure
at 127°C), the failure mode changed to that of the Type II specimens. The
two types of failure modes are reproduced schematically in Fig. 8. The com-
pressive failure was usually across the hole, and may have been initiated by
nearby delaminations. The failure mode typical of Type II specimens was
more of a dislocation or slippage, and usually occurred away from the hole.
0 127 C I STATIC, DHV, AVE OF 15 SPECIMENS
O HT I OPEN TEMPERATE |
a 127 C I SOLID TROPICAL I
|TYPEII|| B " - t*5. 5°)sJ2
• 4 6 8 10 12 14 16 I 4 6 8 10 12 14 16
EXPOSURE TIME. MONTHS EXPOSURE T I M E . MONTHS
FIG. 6—Tensile stalk strength as a function of exposure time.
STATIC, D R Y , A V E OF 15
HT I OPEN TEMPERATE
127 C I SOLID TROPICAL
« E 400
GROSS
FAILING
STRESS,
MPa
•0 2 4 6 a 10 12 14 16 0 2 4 6 8 10 12 14
EXPOSURE TIME, MONTHS EXPOSURE TIME, MONTHS
FIG. 7—Compressive static strength as a function of exposure time.
Compression Tests at Room Temperature—Compressive residual strength

data are replotted against moisture content in Figs. 9 through 12. The static
strength of each specimen and its corresponding moisture content is plotted.
The room temperature data are shown in Fig. 9 for 8-ply specimens, and in
Fig. 10 for 16-ply specimens. Results show that moisture content is a viable
parameter to characterize the compressive strength. The decrease in strength
is monotonic with moisture content, regardless of whether the specimens
were exposed to the tropical or temperate environments. Beyond 1 percent
moisture content, further increase in moisture content did not appreciably
decrease static strength. It should be noted that the temperate environment is
approximately equivalent to a background relative humidity of 68 percent. If
virgin diffusion properties together with weekly drying effects due to flights
are used, the equilibrium moisture content would equal 0.52 percent. Sim-
FIBERGLASSTABS
T Y P I C A L F A I L U R E MODE
OF TYPE IV SPECIMENS
T Y P I C A L F A I L U R E MODE OF TYPE I I SPECIMENS A N D TYPE I V

SPECIMENS A T 127°C A N D > 1.2% MOISTURE
FIG. S—Failure modes under axial compressive load.
(0, +45, OL
o KH
t
o <P
O REALTIME TESTS, TEMPERATE

• REAL TIME TESTS. TROPICAL
b-O-J *
03 04 05 08 10 12 16 IS ?0 32 34 36 38 30
MOISTURE CONTENT, % WEIGHT
FIG. 9—Compressive static strength as a function of moisture content, Type 11 specimen, RT.
ilarly, for the tropical environment, the equivalent background relative hu-
midity of 85 percent and flights would result in a moisture content of 1.17
percent.
The effect of real-time exposure with temperature spiking on room
temperature static residual compressive strength also can be deduced from
Figs. 9 and 10. In this case, the real-time data are compared to the precondi-
tioned data (square symbol).
Preconditioning specimens by constant background humidity and tempera-
ture is a simple laboratory method of obtaining desired moisture content
levels in graphite/epoxy laminates. Exposure times can be accelerated by in-
creasing ambient temperature and relative humidity. The specimens here
were preconditioned by two time-sequenced humidity levels at the same con-
c e o , +45, 0) l
00 <W,5
o %
O REAL TIME TESTS, TEMPERATE

• REAL TIME TESTS, TROPICAL
!L AVGANOSPRE AD Of PRECONDITIONED SPECIMENS

'^f' (I5SPECIMENSI
0 0 2 04 0.6 0.8 10 1? \4 16 IB 30 23 34 36 ?8 30
MOISTUHE CONTENT I % W E I G H T I
FIG. 10—Compressive static strength as a function of moisture content. Type JV specimen. RT.
(0, +45, 0 ) ,
O REAL TIME TESTS, TEMPERATE

• REAL TIME TESTS, TROPICAL
prflp-l
04 06 OB 12 10 14 16 1.8 20 22 24 26 2.B
MOISTURE CONTENT, %WEIGHT
FIG. 11—Compressive static strength as a function of moisture content, Type 11 specimen,

12TC.
Slant temperature in order to simulate as closely as possible the distribution

of moisture through the thickness of the specimen exposed to real-time stor-
age conditions. For the 8-ply specimens, conditioning consisted of 100 per-
cent relative humidity, followed by a 44 percent relative humidity for the
0.52 percent target moisture and a 76 percent relative humidity for the 1.17
percent target moisture. During the entire exposure of 10 days, a tempera-
ture of 77°C was maintained. The same humidity time sequence was used for
the 16-ply specimens; the background temperature, however, was 43°C. The
total time of conditioning was 17 days for the 0.52 percent target moisture
C(0. ± « , 0)^32
; 350
4-'
FIG. 12—Compressive static strength as a function of moisture content. Type IV specimen,
nrc.
and 30 days for the 1.17 percent target moisture. The length of time at each
humidity level varied with specimen thickness and target moisture.
Figure 9 compares 8-ply static strength results of preconditioned speci-
mens with simulated real-time exposed specimens. It can be seen that at low
moisture levels (0.5 percent) temperature spiking had no effect, but at higher
moisture levels (1.2 percent) spiking reduced residual strength. These results
are reasonable, since spiking does not exceed the Tg at low moisture levels,
and no changes in material properties are expected. For the 16-ply specimens
(Fig. 10), the reduction of strength due to spiking at high moisture contents
is not as apparent as for the 8-ply specimens. This is probably due to the dif-
ferences in failure modes between the 8 and 16-ply specimens.
Compression Tests at 127°C—The. 8-ply 127°C residual strength data are
shown in Fig. 11. The static strength decreases monotonically with moisture
content without an apparent lower strength limit. The four low points be-
tween 2.6 and 2.8 percent moisture content represent the longest exposure
time, 15 months. Thus, it appears that exposure time or number of flights
with consequent Tg exceedances contribute to strength degradation.
Two strength levels are apparent for the 16-ply specimens (Fig. 12). The
two levels represent two distinct failure modes as revealed by examination of
failed specimens. As previously discussed, at low moisture contents with or
without spiking, the specimens failed by compressive crushing; at moisture
contents beyond 1.4 percent, the failure mode is local instability. The cause
of the shift in failure mode could be resin softening caused by moisture con-
tent, exposure time combined with spiking, or a synergistic effect of both.
Increase of moisture beyond 1.4 percent did not decrease further the residual
compressive strength.
By comparing the residual strengths of preconditioned specimens to real-
time specimens at the same moisture contents, the validity of simple acceler-
ated environmental experiments to simulate real-time exposure can be as-
sessed. Figures 11 and 12 show that strengths at low moisture content were
unaffected by whichever means the given moisture level was reached. At an
intermediate moisture content of 1.2 percent the strength of preconditioned
specimens appeared to be lower. This is the opposite of what was observed
for room temperature 8-ply specimens, as shown in Fig. 9. At high moisture
contents, the preconditioning method fails, since these moisture contents
cannot be obtained without temperature spiking and subsequent apparent
modification of the material diffusion and absorptivity characteristics.
It is difficult to separate the effects of moisture content and temperature
spiking on residual compressive strength. It may be possible in the future to
use a physical parameter such as Tg exceedance to characterize the complex
effects of real-time exposure. The number of T, exceedances are not uniform
through the thickness of the specimen; hence, a ply-by-ply strength analysis
will be required. This in itself is not intractable. However, the definition and
effects of Tg exceedances are not yet well characterized.
0.2 0.4 0.6 0.8 .4 1.6 Ia 2.0 2.2 2.4 2.6 2B

MOISTURE CONTENT, % WT
FIG. 13—Glass transition temperature as a function of moisture content for graphite/epoxy.
Moisture Transport Model

Model Development
To obtain better correlation with laboratory observations, the Fickian dif-
fusion analysis with virgin material properties was repeated by introducing
variable diffusion and absorptivity coefficients. Results of constant back-
ground conditioning tests have shown that the diffusion and absorptivity
coefficients irreversibly increased whenever the "thermal spike" exceeded
the Tg, as defined in Figure 13.' The data on which this curve is based were
from Footnote 5.
Since the finite difference method was used to numerically integrate the
one-dimensional Fickian diffusion equations with respect to time, it was
possible to divide the laminate thickness into finite layers. At each time step,
every layer is tested to determine if Tg has been exceeded for each layer. If an
exceedance has occurred, the diffusivity and absorption coefficient for that
layer could be permanently altered. Each temperature excursion where the
temperature exceeds Tg and then drops below Tg again was counted as one
exceedance.
Since the numerical integration procedure used the forward difference
scheme, the normal convergence criteria in choosing the time interval must
be observed. In addition, in this analysis, care must be taken that the interval
is small enough to represent accurately the rapidly changing humidity and
temperature conditions as the heat pulse approaches Tg. The time increment
also must be small enough to accommodate what is essentially an instan-
taneous increase in diffusivity and moisture content when Tg is exceeded.
The size of the layers into which the slab is divided also must be considered
Delasi, R. J., Whiteside, J. B., and Wolter, W., "Effects of Varying Hygrothermal Envi-
ronments on Moisture Absorption in Epoxy Composites" in Fibrous Composites in Structural
Design: Proceedings of a Conference on Fibrous Composites in Structural Design, E. M. Lenoe, D.
W. Oplinger, and J. J. Burke, Eds., San Diego, Calif., Nov. 1978.
carefully so that a sufficiently fine division is obtained in regions where steep

moisture gradients exist, such as near the exposed surfaces.
The increase in diffusivity and absorption coefficients were expressed as
ratios with respect to the virgin or unspiked values by the following empirical
relationships
1
Ds/D V = C, ( 1 - +1
BiN + T)
1
As/A V = C2 ( 1 - +1
B2N + T)
where
Ds = value of diffusivity after A^ exceedances of Tg,
Dv = value of diffusivity before any exceedances of Tg,
B\, B2, C\, C2, = numerical coefficients,
N= number of exceedances of Tg,
As = value of absorption coefficient after A^ exceedances of Tg,
and
/4 V = value of absorption coefficient before any exceedances
ofTg.
The values of 5i, B2, Ci, and C2 were manipulated to try to obtain the best
correlation with accelerated environmental test data shown in Fig. 14. This
test attempted to reproduce the effects of the real-time tropical exposure by
maintaining a background of 85 percent relative humidity whenever the flight
profile, identical to real-time tests, was not applied. The test was performed
EXPOSURE TIME, DAYS
FIG. \4—Prediction versus accelerated test data for 8-ply specimens.
10 20 30 40 50 60 70
N - NUMBER OF EXCEEDANCES OF GLASS TFÂNSITION TEMPE TUHE
FIG. 15—Relationship between thermal spiking and moisture absorptivity for graphite/epoxy.
in the same apparatus as the real-time tests, except that the flight profile was
applied daily during the five-day workweek and the 85 percent relative hu-
midity was maintained by on-ofi' conditioning at a constant 60°C. On-off
conditioning to produce 85 percent relative humidity consisted of turning on
steam for 2 h of a 3-h total cycle. Because of the high background relative
humidity and peak flight temperature (127°C), enough T, exceedances
would occur to duplicate the effects of tropical exposure.
The two main criteria which judged the adequacy of the correlation were
the match of the predicted moisture level with the average test data and the
magnitudes of excursions due to thermal spikes. The test apparatus recorded
temperature and relative humidity histories were used as boundary conditions
for the analytical predictions, except that the peak temperature of the thermal
spike was the average of four specimens' thermocouple readings. The values
of coefficients which gave the best correlation were Bi = Ci = 0, Bi = 0.03,
and Ci = 8.5.
It was not necessary to vary the diffusion coefficient, and only the absorp-
tivity was changed as a result of Tg exceedance. This is somewhat surprising.
The Tg exceedances create microcracks in the matrix and matrix/fiber inter-
face, and one would expect the rate of moisture absorption to change with Tg
exceedance. It should be noted that the chosen values of coefficients are not
unique; a different set of coefficients with nonzero Bi and Ci could provide
an equally good correlation. The functional relationship is shown graphically
in Fig. 15. The correlation between the test data and analysis is excellent for
the 8-ply specimens shown in Fig. 14. For comparison, the analysis with con-
stant absorptivity coefficient also is shown in Fig. 14. As can be seen, the
- TROPICAL EXPOSURE
• TEMPERATE EXPOSURE
TEST MEASUREMENTS. TEMPERATE EXPOSURE

TEST MEASUREMENTS. TROPICAL EXPOSURE
EXPOSURE TIME. MONTHS
FIG. 16—Comparison of predictions and real-time test data for 8-ply specimens.
predictions using that analysis begin to diverge from the test data after about
30 days, and at 60 days have reached a possible equilibrium state not at all
consistent with the test data.
Correlation With Real-Time Data

The developed semiempirical model was applied to predict moisture uptake
under simulated temperate and tropical storage conditions of Figs. 4 and 5.
The boundary conditions were modified from idealized data to actual test
apparatus temperatures and relative humidities. Thus all irregularities due to
equipment malfunctions were included in the analytical model.
The correlation for the temperate location, shown in Figs. 16 and 17, is
reasonable. The drop in moisture content after six months is due to seasonal
- TROPICAL EXPOSURE
_ _ . - TEMPERATE EXPOSURE
O TEST MEASUREMENTS, TEMPERATE EXPOSURE

• TEST MEASUREMENTS, TROPICAL EXPOSURE
EXPOSURE TIME. MONTHS
FIG. 17—Comparison of predictions and real-time test data for 16-ply specimens.
variations: it is drier in the summer than in the winter. Only two exceedances
of Tg were noted by the analytical model, all due to equipment malfunctions.
A more severe test of the analytical model was provided by the tropical lo-
cation real-time test data. The correlation for the 8-ply specimens (Fig. 16) is
quite good. For the 16-ply specimens (Fig. 17) the predicted moisture gains
are lower than the test data. Better tuning of the model could result in better
correlation. However, it is not surprising that a transport model based on
diffusion, even with history-dependent properties, does not correlate per-
fectly since increases in absorptivity may be due to increased surface area
and capillarity. Another source of error is due to the fact that the empirical
coefficients of the model are based on test duration of 60 days and as such
may not be applicable to tests of longer duration. However, the model is
considerably more accurate than predictions based on virgin material moisture
absorption behavior, as shown in Figs. 16 and 17.
Conclusions
Real-time tests simulating a tropical environment with 127°C flight
temperatures, have demonstrated that graphite/epoxy material properties
are degraded beyond what was expected and shown for temperate exposure.
The degradation has been traced to exceedances of Tg and is manifested by
substantial increases in absorptivity of the material and consequent further
reductions in matrix-dependent strength properties.
To account for the increases in the material diffusion/absorption proper-
ties, a semiempirical model was developed based on Fickian moisture trans-
port, whereby the material absorptivity was increased according to the
number of Tg exceedances. Predictions using the adjusted properties matched
test results much better. A more refined analysis, perhaps accounting for the
amount of Tg exceedance, or other factors, is indicated. More work is needed
to understand the mechanisms of moisture transport, but the analysis per-
formed here points to a possible direction for material with Tg exceedance
history.
Exposure to extreme environment (ground storage with weekly supersonic
flight profile) does not reduce fiber-dominated properties, as evidenced by
residual static tensile strength. Static compressive strength is degraded by
such exposure, especially at 127°C. The strength reduction is related directly
to moisture content, irrespective whether the water is held by resin or in micro-
cracks. The largest compressive strength reduction (50 percent from dry
properties) occurred at 127°C, where for the thicker 16-ply specimens, resin
loss of capability for load transfer caused the failure mode to change from
overall laminate compressive failure to local instability.
Preconditioning or other accelerated environmental tests, if they are to be
successful in duplicating real-time extreme environment, must include means
to exceed Tg. The semiempirical Fickian diffusion model described in this
paper would be of great help in planning such experiments.
A cknowledgments
The authors wish to acknowledge the support of Grumman's R. Price and
P. Palter for conducting the tests. The authors also would like to thank J. B.
Whiteside for advice and suggestions. This program was funded partially
by Air Force Flight Dynamics Laboratory (AFFDL) Contract No.
F33615-76-C-5234, monitored by E. Demuts, and Advanced Development
Departments' in-house funding.
Moisture Environments
R. C. Givler,^ J. W. Gillespie, Jr.,^ andR. B. Pipes^
Environmental Exposure of
Carbon/Epoxy Composite
l\/laterial Systems
REFERENCE: Givler, R. C, Gillespie, J. W., Jr., and Pipes, R. B., "Environmental

Exposure of Carbon/Epoxy Composite Material Systems," Composites for Extreme
Environments, ASTM STP 768, N. R. Adsit, Ed., American Society for Testing and
Materials, 1982, pp. 137-147.
ABSTRACT: The effect of potentially adverse environmental conditions on composite

material systems is investigated. Hybrid glass-carbon/epoxy laminates were thermally
cycled between —54 and +177''C, and residual compressive and flexural properties de-
termined. Results indicate that extended thermal cycling had only negligible effect on
laminate compressive or flexural strength. Observed cracks in the 90-deg and adjacent
±4S-deg plies were observed and their existence predicted by laminate analysis, which
showed thermal residual stresses to exceed the intrinsic material strength in those plies.
Thermal fatigue loading increased crack density in the 90-deg and ±4S-deg laminae
eight-fold, subsequent to 1000 cycles.
KEY WORDS: thermal cycling, graphite/epoxy, hybrid, transverse cracking, flexure,

compression, crack density, composites
Each layer of a laminated composite is anisotropic in thermal expansion.

The lamina deformations due to thermal loading are controlled largely by
the fibers in the direction parallel to the filaments, whereas in the direction
transverse to the fibers the coefficient of thermal expansion is dominated by
the matrix properties. Varying the fiber orientation angle between adjacent
plies of the laminate causes a mismatch in the coefficient of thermal expan-
sion among laminae. Hence, as the laminate undergoes uniform deformation
due to a temperature change, stresses are introduced which may vary from
ply to ply.
Interest in the cyclic thermal residual stresses of hybrid composite material
systems has prompted this investigation to determine the effect upon me-
chanical properties. It was also of interest to determine the effect of this
thermal fatigue loading on material degradation in the form of cracks or inter-
' Graduate student, graduate student, and professor, respectively. Department of Mechanical
and Aerospace Engineering, Center for Composite Materials, University of Delaware, Newark,
Del. 19711.
137

laminar disbonds. The proposed laminate configuration was a 32-ply glass/

epoxy (120 weave/3501-6), carbon/epoxy (AS/3501-6) system with the fol-
lowing stacking sequence
[0GL5/0/45/-45/0/45/90/-45/0/45/-45/0]S
The laminate stacking sequence is shown schematically in Fig. 1. As an added

note, the surface plies of glass serve as an insulation layer between the carbon
core and a bonded aluminum substrate. The particular sequence of orienta-
tions within the carbon core is typical of the laminate configuration seen in
many applications. Here, the interaction between numerous ply orientations
may be observed as a result of the cyclic thermal stresses.
Thermal Cycling Apparatus

The temperature excursion for cycling the laminates ranged from 177 to
—54°C. In addition to this large temperature variation, the laminates were
subjected to a maximum of lOOO thermal cycles. To satisfy these criteria, an
automated thermal cycling unit was fabricated.
A cutaway, top-section view of the thermal cycling apparatus is shown in
Fig. 2. The apparatus consists primarily of two distinct chambers separated
by a pivoting door. The chamber on the left was maintained at the elevated
temperature of 177°C with the use of a conventional heating coil. The low
temperature of —54°C was achieved in the adjacent chamber by exposing the
specimens to a gaseous, liquid nitrogen environment. Actual cycling of the
composite laminates was facilitated by clipping the specimens to the rotating
(120 weave/3501-6)
(AS/3501-6)
symmetric
(Cb /0/45/-45/0/45/90/-45/0/45/-45/0) c
5
FIG. 1—Laminate stacking sequence.
GIVLER ET AL ON COMPOSITE MATERIAL SYSTEMS 139
KEY
o.
TEMPERATURE
SENSOR
THERMAL INSULATING MATERIAL
L N , GAS INLET
FIG. 2—Top view of thermal cycling apparatus.
door; as the door rotates to the dotted configuration (Fig. 2), the specimens
were transported from the heated chamber to the cooling chamber and, as
was the case in the undashed configuration, the chambers are thermal sealed
from one another.
A temperature sensor mounted in the vicinity of the specimens, in con-
junction with a Koenig-Pretempco temperature controller, regulated the
thermal cycle history. When the desired specimen temperature was achieved
in either chamber, the door rotated through an angle of ±90-deg, thus expos-
ing the specimens to the other chamber. In this manner, the laminates auto-
matically were thermally cycled.
Figure 3 illustrates a typical thermal cycling history with an average cy-
cling period of 37 min. The plotted temperature variation is that of the en-
vironment, with a maximum temperature gradient of 0.67°C/s. To deter-
mine the nature of the spatial temperature variation through the thickness of
the laminates, the Biot (Bi) number was calculated. It may be recalled that
the Bi number is the ratio of internal resistance to external resistance of heat
transfer, and its value was found to be
(1)
k
where
h = coefficient of heat transfer,
/ = thickness of the laminate, and
k = thermal conductivity of the laminate.
The implication of Eq 1 assures a uniform temperature distribution through-
out the laminate despite the varying surface temperatures. Thus, changes in
200 -I 1 r 1 — —1 1 1[ - • -f— —1 1 1 r-
190
n n
v
100 -
1 -
I
u
i 90
a
u
a.
Z
u
0
0.67-C/MC i
-90
THERMAL CYCLING
HISTORY
-100 t 1 1 1 1 1 1 1 1 ' • 1 1 1
20 30 4 0 90 60 70 ao 90 100 110 120 130 140 190
TIME , MINUTES
FIG. 3—Environmental exposure of carbon/epoxy composite material system.
surface temperatures are propagated rapidly inward, and stresses due to

thermal gradients through the thickness may be neglected.
Experimental Program
The experimental program was divided into two segments. The first phase,
described here, was an investigation of the residual composite flexural
strength subsequent to thermal cycling. Five specimens were thermally
cycled for 0, 1, 10, 100, and 1000 cycles, respectively. Upon completion of the
thermal cycling, the specimens were tested in three-point bending to failure
at 177°C. The flexure specimen dimensions were 1.27 by 7.6 cm, while the
nominal 32-ply laminate thickness was 0.40 cm. The loading nose was ap-
plied adjacent to the glass cloth plies of the laminate shown in Fig. 1. Reac-
tion end points for the flexural test fixture were separated by a distance of
5.08 cm, yielding a span-to-depth ratio of 12.7.
Results of the residual flexural response of the cycled laminates are illus-
trated in Fig. 4. Each data point corresponds to the average value from five
tests, and the error bands represent one standard deviation in the experimen-
tal data. For purposes of comparison, the actual uncycled flexural strength
was recorded to be 542 MN/m^. The trend in the data indicates the normal-
izedflexuralstrength increases slightly with thermal cycling. This anomalous
behavior possibly may be explained in terms of the post-curing of the resin
I 1
1.25
(>
t
1
• 1
.90
ALL TESTS AT I 7 7 ' C
•
LJOG ( T H E R M A L CYCLES)
ACTUAL STRENGTH
UNCYCUD STRENGTH
FIG. 4—Effect of thermal cycling onflexural strength.
phase during thermal cycling with the effect of a slight increase of the glass
transition temperature of the matrix. The presence of transverse cracking (to
be discussed subsequently) offered no degradation in flexural strength, since
the thermal cracks were found only in the 90-deg and 45-deg layers, which
contribute little to laminate flexural strength.
The remaining phase of the experimental program examined the effect of
thermal cycling on compressive strength of the composite laminate. Com-
pression data for the laminate of Fig. 1 was generated via the Illinois Institute
of Technology Research Institute (IITRI) compression test and recorded at
room temperature. Standard specimen dimensions were maintained, which
yielded a gage section of 12.7 by 4.0 mm. All fractures appeared to be com-
pressive in nature as local instabilities in the test section were not observed.
The results of the compressive test may be examined in Fig. 5, where it is
shown that normalized compressive strength was not affected by thermal cy-
cling. Again, each data point of Fig. 5 represents the average value from five
tests, while the error bands signify one standard deviation in the data. It
should be noted that the uncycled strength corresponds to the strength of the
thermally uncycled laminates (653 MN/m^), which have been subjected to
one-half thermal cycle, that of curing the laminate. This strength value was
used for normalization purposes.
Perhaps the most informative observation to be made from this experi-
mental investigation was the presence of transverse cracking in the 90-deg
M } }
.50
NOTE: UNCYCUDSTRENOTHISTHATOF
THE CURED LAMINATE (WHICH
CORRESPONDS TO ONE - H A L F
CYCLE.)
I 2 3
LOG (THERMAL CYCLES)
ACTUAL STRENGTH
UNCYCLEOSTRENSTH
FIG. 5—Effect of thermal cycling on compressive strength.
and adjacent ±45-deg plies. Figure 6 illustrates a representative crack in the

90-deg ply that is shown to extend through the entire ply thickness. The rela-
tionship between crack density and thermal cycles is provided by Fig. 7. In
constructing the definition of crack density, all visible, transverse cracks
under X100 magnification were counted per linear dimension regardless of
their length. In contrast, other authors may employ slightly different defini-
tions for crack density; only transverse cracks extending the total ply thick-
ness are counted. Several points concerning Fig. 7 are noteworthy. First, it is
noted that the curve does not intersect the zero ordinate. This indicates the
presence of cracks even at zero cycles or, alternatively, the curing of the lam-
inate alone produced 2 to 3 cracks per centimetre. Also, it is seen that the
crack density increases sharply with the logarithm of thermal cycles. It has
been suggested that this curve should asymptotically approach a certain lim-
iting value of crack density known as the crack saturation value. Clearly, the
results presented here do not indicate such a limiting value, and the reason
may lie in the various definitions of crack density, discussed previously. The
cracks were first observed in the 90-deg laminae, but as the cycling continued
they grew into adjacent ±45-deg laminae.
Theoretical Analysis
Laminated plate analysis was employed to predict the stress state within
the composite laminate. Special concern was given to determining correctly
FIG. 6—Typical transverse crack extending through 90-deg laminae.
the stresses in the 90-deg and adjacent ±45-deg pHes to justify the presence
of transverse cracks in these pHes. The usual notation for a laminated plate
analysis is given in Fig. 8, (which is redrawn from Vinson and Chou^) and
the constitutive relation describing the laminate response is given in Eq 2
(2)
IM] LÎijJLJ
where
Aij = Y.(Qv)k(hk-hk-i) (3fl)
Bij= l/2E(eff)t(A*'-A*-i') (36)

^Vinson, J. R. and Chou, T. W., Composite Materials and Their Use in Structures, Halsted
Press, New York, 1975.
f. 15 -
0 1 2 3 4
LOG (THERMAL. CYCLES)
FIG. 7—Effect of thermal cycling on crack density.
Dij=l/3îQy)k{hk'-hk-i') (3c)
and
Qv = in''[Q][n (4)
Here, f is defined to be the transformation matrix of the stress tensor be-
tween rotated reference frames, while ^* transforms the engineering strain
components accordingly.'
Since the desired temperature excursion was quite large, it is necessary to
include the temperature-dependent material property variation. Moduli vari-
ation (both shear and elongational) were taken as linear functions of
temperature. The specific numerical values for the carbon/epoxy system were
referenced in the Air Force design guide,* while suppliers' data were used for
temperature-dependent properties in the glass weave material system. These
property variations are incorporated into the analysis in the definition of the
'Ashton, J. E. et al. Primer on Composite Materials: Analysis, Technomic Publishing Co.,

Westport, Conn., 1969, Chapter 2.
* Advanced Composites Design Guide, 3rd ed., 3rd revision. Air Force Flight Dynamics
Laboratory, Dayton, Ohio, 1977.
Ik
z*
- , K , "4
•
1^^^ .-
''o
^' X-<i)- ^<r'^^<P^
®
.^ <S> ^y<^^^
^^^<^
^ ®
ffl yy^
y
FIG. 8—Nomenclature for laminated plate analysis.
Qij terms, namely
(5fl)
(1 — J'12»'2l)
EziiT)
222 (56)
(1 — J'12»'2l)
Qn — ^21 — . (5c)
(1 — I'l2J'2l)
Qi6 = GniT) (5d)

For balanced, symmetric laminates subject only to thermal loadings, Eq 2
takes the form
W = M][etotai-jr a(T)dtj (6)
where the coefficient of thermal expansion also has been permitted to vary
linearly with temperature. Owing to this linear temperature variation, Eq 6
may be rewritten simply as
{N} = [>I] [etoui - a (r-jA ( ^ - ô)] (7)
Using Eq 7, one now may predict the stresses that arise from the thermal
excursion of the laminate specimens. The stress-free temperature was taken
to be 177°C so that thermal residual stresses resulted as the temperature de-
creased from this point. Figure 9 shows the variation of thermally induced
stresses in the 90-deg carbon/epoxy ply as predicted by Eq 7. Predicted
stresses have been normalized by the respective intrinsic material strengths,
also taken to vary with temperature.* The stresses superscripted with an aster-
isk denote the improved prediction of the stresses by allowing the coefficient
of thermal expansion to vary with temperature, as in Eq 6. Indeed, Fig. 9
-50 0 50 100
temperature (°C)
FIG. 9—Analytically predicted thermal stresses in the 90-deg carbon/epoxy laminae.
shows the transverse residual stresses to exceed those of the inherent material
strength in the 90-deg carbon/epoxy ply only when a is allowed to vary with
temperature.
Likewise, examining Fig. 10 in a similar manner would lead to the predic-
tion of transverse cracks to appear in the adjacent ±45-deg carbon/epoxy
plies. The stresses in the remaining glass and graphite plies were far below
the ultimate material strengths, and no thermally induced cracks were ob-
served in these plies.
Conclusions
It was found that thermal cycling of the hybrid
[OGL5/O/45/-45/O/45/9O/-45/O/45/-45/O].
laminate produced cracks in the 90-deg and adjacent ±4S-deg laminae. Lam-
inate analysis verified that indeed the thermal stresses induced by thermal
cycling exceeded the intrinsic strength of the 90-deg and adjacent ±45-deg
laminae. An average of 3 cracks per centimetre were observed in the un-
cycled (control) specimens due to curing alone. Crack densities were ob-
served to increase with thermal cycling to a maximum of 26 cracks per cen-
50 100
temperature K)
FIG. 10—Analytically predicted thermal stresses in the ±45-deg carbon/epoxy laminae.
timetre subsequent to 1000 thermal cycles. It was expected that the interface
between the glass and carbon plies might be a site for delamination, due to
the gross mismatch of corresponding coefficients of thermal expansion.
However, this was not the case, as no glass/epoxy or carbon/epoxy dis-
bonds were observed subsequent to thermal cycling.
Compressive strength, measured at 21°C, revealed no significant change
after 1000 thermal cycles; thus, it appears thermal cycling has no effect upon
uniaxial compressive strengths for the laminate investigated.
Flexural strength, measured at 177°C, increased 29 percent after 1000
thermal cycles. A conservative argument would state that flexural strength
experienced no degradation due to thermal cycling.
L. W. Rehfield.^ R. P. Briley.^ and S. Putter^
Dynamic Tests of Graphite/Epoxy

Composites in Hygrothermal
Environments
REFERENCE: Rehfield, L. W., Briley, R. P., and Putter, S., "Dynamic Tests of Graph-
ite/Epoxy Composites in Hygrothermal Environments," Composites for Extreme Envi-
ronments. ASTM STP 768, N. R. Adsit, Ed., American Society for Testing and Mate-
rials, 1982, pp. 148-160.
ABSTRACT: The objective of this research is to experimentally determine the effects

of hostile, hygrothermal environments on the dynamic behavior of graphite/epoxy
composites.
This objective has been accomplished by performing flexural vibration tests on
Narmco 5208/T3O0 beam specimens of four distinct layups. Dynamic behavior in the
dry, room temperature state is contrasted with five other environmental states, includ-
ing moisture saturation and temperatures up to 93°C (200°F). The properties deter-
mined are fundamental natural frequency and damping. The former can be directly re-
lated to an effective dynamic modulus in flexure. These properties reflect the
consequences of plasticization of the epoxy matrix material in hygrothermal environ-
ments. The experimental results indicate that matrix controlled modes of response are
the most affected. Generally damping increases and stiffness decreases with increasing
hygrothermal conditioning. For most applications, stiffness reductions are the domi-
nant concern; the experimentally determined dynamic reductions are consistent with
known static ones.
KEY WORDS: hygrothermal effects, composite materials, structural dynamics, envi-

ronmental effects
Epoxy resins used as matrix materials in structural composites experience

degradation of mechanical properties due to moisture absorption and ele-
vated temperature environments. Considerable effort has been devoted to
documenting the extent of the degradation for simple layup configurations
and states of stress, for several popular material systems and for static and
fatigue loadings. The present work, by contrast, is concerned with dynamic
behavior. The objective is to establish a data base to facilitate the use of
graphite/epoxy composites for dynamic applications in hostile, hygrother-
mal environments.
' Professor and visiting scholar, respectively. School of Aerospace Engineering, Georgia Insti-
tute of Technology, Atlanta, Ga. 30332.
^Senior engineer, McDonnell Douglas Astronautics Co., St. Louis, Mo. 63166.
148

REHFIELD ET AL ON DYNAMIC TESTS OF GRAPHITE/EPOXY 149
An early study of hygrothermal effects on dynamic structural behavior of

epoxy matrix composite beams in bending is reported in Ref 1.' This work
shows the contrast between two limiting reference conditions—dry at room
temperature [25°C (77°F)] and moisture saturation at 93°C (200°F). Changes
in fundamental natural frequency, an overall, practical measure of stiffness,
and damping were determined experimentally. The results indicate that stiff-
ness is reversibly lowered in matrix controlled modes of deformation and
that damping characteristics are altered in both fiber and matrix controlled
modes due to environmental conditioning. These property changes are the
consequences of plasticization of the epoxy matrix. This type of information
is of particular importance in applications pertaining to acoustically driven
vibration.
The present work has its origin in the research of Ref 1. The investigation
is expanded to include four new environmental states and an additional type
of composite specimen layup. Improved environmental control is achieved
during the tests and a better testing technique is employed. A new, additional
feature that is included in the present paper is a rational means for adjusting
experimentally determined damping coefficients for extraneous influences.
Supplementary experiments have been conducted which permit damping
contributions due to air and accelerometer cable motion to be estimated
simply. Adjusted damping data are presented also, therefore, to facilitate
application of the results.
The effects of frequency on dynamic behavior are not part of the work re-
ported herein. This subject is currently under investigation, and will be dis-
cussed in future publications. The intent here is to compare the response of
composite beams under identical conditions of excitation but differing states
of moisture conditioning and temperature. The viewpoint adopted is that the
changes in behavior of a given structure due to hygrothermal environments
are of direct, practical interest to the designer.
Experimental Approach
Composite Test Specimens
Three composite graphite/epoxy panels of distinct layup have been manu-
factured by McDonnell Douglas Astronautics Co.-St. Louis from Narmco
5208/T300 unidirectional type. The layups are [0], [±45] and [0, -f-45, 90,
—45, 0, -l-45]s. They are each 12-plies thick and symmetric, and will be re-
ferred to as Type A, B, and C, respectively. Types A and B are limiting types
of fundamental interest, while C is representative of practical vehicle skin
panels. An approximate fiber volume fraction of 61.5 percent has been de-
termined by analysis of typical specimens. The specimens used in the earlier
study [7] came from these same panels also.
'The italic numbers in brackets refer to the list of references appended to this paper.
Beam specimens were cut from the panels for environmental conditioning
and testing. [90] specimens are obtained from the Type A panel by cutting
normal to the fiber direction; these will be designated Type D specimens.
This type of specimen, added for this study, is also of fundamental interest.
Six specimens of each type were prepared. They are nominally 2.54 cm (1.00
in.) wide by 20.32 cm (8 in.) long. While actual thicknesses have been mea-
sured, a nominal thickness of 0.0137 cm (0.0054 in.) per ply has been used
throughout to analyze the test data; this value is based upon manufacturing
experience at the St. Louis McDonnell Douglas facility. Thickness mea-
surements performed also confirm this as a reasonable value.
Overview of the Experimental Program

The underlying philosophy of the experimental program is the same as
that stated in Ref. 1. The intent of the tests is to provide information for as-
sessing the nature and extent to which environmental conditioning influ-
ences dynamic behavior of composite beams in bending. In order to do this,
tests have been conducted that correspond to the following environmental
states: (a) 25°C (77°F), dry; (b) 25°C (77°F), moisture saturated; (c) 60°C
(140°F), moisture saturated; (d) 82°C (180°F), moisture saturated; (e) 93°C
(200°F), moisture saturated; and (/) 82°C (180°F), dry. The properties de-
termined are fundamental natural frequency and damping.
The composite specimens require considerable handling as moisture con-
tent is determined by weighing them. They are oven dried for two days at
60°C (140°F), weighed, and vibration tested in cantilever fashion at room
temperature [25°C (77°F)]. This is followed by conditioning in a constant
temperature water bath at 82°C (180°F) until moisture saturation (approxi-
mately 80 days). The specimens are weighed and vibration tested again in a
Blue M environmental chamber at 82°C (180°F) and 95 percent relative hu-
midity. Each specimen is weighed also after vibration testing in order to de-
termine moisture loss during handling and testing. The specimens then are
returned to the water bath at 60°C (140°F), with the procedure repeated at
the new temperature after conditions stabilize. The 93°C (200°F) data are
taken from Ref 1 and were obtained using different specimens from the same
parent panels. After all moisture saturated tests are completed, the speci-
mens are redried for testing at 82°C (180°F). In addition, a selected number
of redried specimens are retested at room temperature. In all cases, redried
specimen data agree within normal limits with original data. This is inter-
preted to mean that no damage occurred during the course of conditioning
and testing.
Since environmental effects are to be determined, testing in a vacuum
chamber at room temperature—the usual means of determining damping—
could not be used. An exception could have been made for the room temper-
ature reference state. It was decided to use a uniform procedure for testing
throughout, however, in order to facilitate comparisons and to eliminate

procedure-related influences.
In addition to the intrinsic material-related damping that is of interest,
there are parasitic damping influences due primarily to air and accelerometer
cable motion that contribute to measured values. Supplementary experi-
ments have been conducted which permit rational estimates of these ex-
traneous effects. These experiments are discussed, along with the data ad-
justment procedures, in the discussion of results.
Testing Technique
The vibration testing technique has been dictated by the need for simplic-
ity, repeatability and preservation of the environmental conditions. In order
to preserve the environmental states achieved for the composites by condi-
tioning in the water bath to the maximum extent, the excitation technique se-
lected for vibration testing is a transient one. The advantage obtained is a
short duration test. The beams are mounted in cantilever fashion in a fix-
ture with prescribed tip deflection. The tip suddenly is released, producing
transient response of the specimen that is dominated by the fundamental
mode of vibration. The fixture with a mounted beam specimen is shown in
Fig. 1. Response is sensed by an accelerometer mounted near the specimen
tip. A schematic of the experimental setup for the tests appears in Fig. 2.
The accelerometer signal is filtered through a low pass filter to minimize
the higher mode frequency content. The data are processed by means of a
Fourier analyzer. The analog to digital conversion capability of the analyzer
is all that is utilized. The low pass filter also serves to minimize aliasing er-
rors in the analog to digital conversion process. The digitized accelerometer
output is printed out, and the fundamental frequency and corresponding
damping coefficient determined by simple analyses. Frequency is determined
simply by counting the number of peak values contained in a suitably chosen
time interval. Values for the damping coefficient are obtained by the loga-
rithmic decay method.
Each test of an individual specimen in each environmental condition has
been repeated. This results in 12 data values (2 tests on each of 6 specimens)
for each environmental condition and for each of the 4 specimen types.
The large number of data values corresponding to each type of specimen
and environmental condition have been obtained to minimize the effects of
two factors. One is the variability found in even the highest quality compos-
ite material; it exceeds that of conventional "homogeneous" materials such
as metals. The second factor is associated with the difficulty of achieving
consistent clamping conditions in each of the tests. Elapsed time savings and
simplicity are achieved with cantilever tests; these advantages are believed to
outweigh the uncertainties associated with such things as uniform clamping
FIG. 1—Vibration test fixture with a mounted beam specimen.
pressures and swelling strains. All these factors are dealt with in a practical
manner by repetition.
The prescribed initial tip deflections correspond to maximum strains at
the beams' fixed ends which are far less than the ultimate strains for the
material. For specimen Types A andC, a tip deflection of 0.34 cm (0.134 in.)
has been imposed. A value of 0.41 cm (0.161 in.) is applied to specimen
Types B and D. These selections insure that no damage results from the vi-
bration tests themselves. This is verified indirectly by repeatability of the test
results.

Remarks on the Presentation of Results
A large number of tests have been conducted and a considerable volume
of data has been taken. This information includes weight data recorded at a
number of times throughout the testing history. The presentation of all of
this information is secondary in this context. The intent here is to delineate
the essential features of the dynamic behavior. Consequently, only results
BLUEM FIXED DISPLACEMENT

ENVIRONMENTAL TRIGGER MECHANISM
TEST CHAMBER
ACCELEROMETER
SPECIMEN
—I Ji.
CHARGE LOW PASS

AMPLIFIER FILTER OSCILLOSCOPE
FOURIER
PRINTER
ANALYZER
FIG. 2—Schematic diagram of the experimental setup for the vibration tests.
pertinent to this objective are presented. A complete compilation of all re-

sults appears in Ref 2.
For reference, at 82°C (180°F), the mean weight gain due to moisture ab-
sorption for all specimens was 1.52 percent for saturation. The mean weight
loss which occurred during handling and testing at 82°C (180°F) was 0.08
percent. Therefore, a close approximation to in situ tests has been achieved.
All pertinent experimental data appear in Tables 1 and 2. A quantity Ef,
the effective dynamic modulus in simple flexure, has been determined from
the following equation
^-^Nf-7)^f-ferik (1)
TABLE 1—Mean dynamic modulus data.
Effective Dynamic Modulus in Simple

Flexure for Composite Specimens, GN/m^
(Standard Deviation as Percentage of the Mean)
Dry, 25°C 25°C 25°C 60°C 82°C 93°C 82°C

Specimen Testing (77°F) (77°) (140°F) (180°F) (200°F) (180°F)
Type Frequency, Hz Dry Saturated Saturated Saturated Saturated Dry
A 54 113.1 115.8 118.0 119.1 121.4 114.5

(6.2) (7.2) (6.1) (4.9) (8.2) (6.8)
B 25 24.3 23.8 22.7 22.2 19.2 23.1
(2.3) (2.5) (2.2) (2.1) (2.0) (2.1)
C 44 74.8 78.0 78.7 78.7 69.5 75.9
(3.2) (1.8) (2.6) (2.9) (1.9) (2.5)
D 27 9.8 9.5 9.0 8.6 9.6
(6.3) (5.3) (5.3) (5.0) (4.5)
TABLE 2—Mean experimentally determined damping coefficients.
Damping Coefficient, percent of critical value

(Standard Deviation as Percentage of the Mean)
Dry, 25°C 25°C 25°C 60°C 82°C 93°C 82°C

Specimen Testing (77°F) (77=) (140°F) (180°F) (200°F) (180°F)
A 54 0.455 0.594 0.625 0.554 0.586 0.415

(4.3) (11.9) (7.0) (7.5) (11.8) (6.1)
B 25 0.788 0.939 0.937 0.823 0.594 0.706
(4.8) (11.0) (4.7) (4.7) (6.1) (5.7)
C 44 0.546 0.673 0.601 0.644 0.362 0.543
(5.0) (6.1) (6.4) (6.7) (5.0) (3.0)
D 27 0.885 0.798 1.071 0.897 0.707
(7.6) (5.4) (6.0) (4.0) (7.1)
L is the cantilever beam length, / i s the second moment of the cross sectional
area, m is the mass per unit length of the beam, and/i is the measured fun-
damental frequency in cycles per second. M, is an effective concentrated tip
mass that includes the effects of the accelerometer and accelerometer cable;
/, is the corresponding mass moment of inertia of the effective concentrated
tip mass about the beam neutral axis. The above equation is derived from a
Rayleigh quotient calculation for a simple cantilever beam with a tip mass.
The approximate vibration mode shape used is
^{X) = 1 - cos ^ (2)
A" is a coordinate measured from the fixed end of the beam. Damping is re-
ported as values for the damping coefficient as a percentage of the critical
values.
The values presented in Tables 1 and 2 are the means of 12 independent
experimentally determined values. Below each entry in parenthesis is the cor-
responding standard deviation expressed as a percent of the mean. This is a
convenient and concise way of presenting the experimental results.
Experimental Results
Table 1 contains the mean values of E/ calculated from measured mass,
geometric and fundamental frequency data for the specimens. Type A spec-
imens show a slight progressive increase in stiffness due to hygrothermal
conditioning; flexural response of these specimens is fiber controlled, so no
significant degradation was expected. Static data presented for AS/3501-5
graphite/epoxy from tension tests in the fiber direction [3,4] reflect a similar
slight increase in stiffness with hygrothermal conditioning. Type B and D
specimens show a decrease in stiffness with increased hygrothermal condi-
tioning. Response of these specimens is matrix controlled, and this behavior

is confirmed by the static data from Ref 3 and 4. The stiffness reduction for
Type B specimens was shown previously to be reversible in the extreme case
considered [i]. Type D specimen tests at 93°C (200°F), saturated could not
be conducted; this type of specimen is too fragile to survive testing and han-
dling at this condition.
Type C specimens behave in a "mixed" manner. A slight increase in stiff-
ness, which suggests that fiber controlled influences are dominant, is found
up to 82°C (I80°F), saturated. A stiffness reduction at 93°C (200°F), satu-
rated suggests that matrix controlled influences are more important at this
temperature. A desirable property of a practical laminate would be fiber
controlled flexural response; consequently, from a practical standpoint, the
trend reversal issue is of considerable interest.
The values of Ey for Type A and B specimens differ from the modulus
values measured by Walter [5] in static tension tests of specimens cut from
the same panels. The static tensile modulus of Type A specimens is 148.24
GN/m^ (21.5 X 10* psi), while that of Type B specimens is 18.62 GN/m^
(2.7 X 10' psi). The differences between static tensile and dynamic flexural
stiffnesses may be attributable to several physical factors, including the ap-
proximate nature of Eq 1. Similar discrepancies also have been reported by
Dugundji [6]. While the reasons for these discrepancies are not known, they
are not the primary concern in the present context. The dry static values are
in the normal range for this material system. Consequently, the results pre-
sented should be representative of what would be obtained if the tests were
to be duplicated by others.
Mean experimentally determined values of damping coefficient appear in
Table 2 along with the corresponding values for standard deviation. These
values include parasitic extraneous contributions. Consequently, before dis-
cussing damping, a means of adjusting the data to compensate for these ef-
fects is presented.
Adjustments to Damping Data

One of the original, but unstated, objectives of this research was to estab-
lish the technology for dynamic hygrothermal testing that could be applied
to structural elements, components, and full scale structures such as aircraft.
The choice of instrumentation and testing technique reflect this orientation.
This has been achieved successfully, as the quality and repeatability of the
data in Tables 1 and 2 indicate. For material damping determination, how-
ever, the influence of extraneous factors on the measured damping coeffi-
cients must be established. Supplementary tests have been conducted to iso-
late parasitic damping due to air and accelerometer cable motion.
Air damping associated with the vibration of cantilever beams has been
studied thoroughly by Baker, Woolam, and Young [7]. For small ampli-
tudes, aerodynamic damping is proportional to amplitude. The appropriate

relationship is
Â is the damping coefficient contribution due to air, pA is the air density, ps

is the specimen density, A is the amplitude of the tip deflection, and h is the
specimen thickness. Kisa. proportionality factor which has been found from
experiments.
In order to determine K, experiments have been conducted without using
an accelerometer on a 2024-T3 aluminum beam and composite beams of
each type. The beams have been tested using electromagnetic noncontacting
B & K Instruments MM0002 transducers for excitation and response deter-
mination. These are resonance tests in the fundamental mode. Damping is
determined by suspending excitation and observing the decay of the re-
sponse. Tests are conducted at several amplitudes. Damping versus ampli-
tude plots are constructed. Extrapolation to zero amplitude yields intrinsic
specimen damping. The amplitude-dependent portion is attributed to air.
This technique is quite effective. Verification tests with the aluminum beam
show that the specimen and air damping values are the proper magnitude
and agree well with values from Ref 7.
Air damping contributions determined from some of the supplementary
tests appear in Fig. 3. Results obtained in dry air at 25°C (77°F) and in mois-
ture saturated air at 82°C (180°F) are shown. All types of composite speci-
mens behave in the same way, thus indicating that there are no significant
frequency-dependent influences present. Data from Ref Zand from these in-
dependent experiments suggest that K = 102 in Eq 3 is a value that provides
good damping estimates.
The other extraneous contribution to damping is due to parasite effects as-
sociated with accelerometer cable motion. These effects may be partially at-
tributable to aerodynamics and partially due to mechanical deformation of
the cable. Estimates of this contribution have been obtained by conducting
experiments with accelerometer and cable mounted. Tests at various ampli-
tudes for each type of specimen have been performed. Cable-related damp-
ing data show relatively high scatter from one mounting to another. Conse-
quently, twelve independent test sequences were performed and the data
averaged for each specimen type. This damping contribution appears to be
frequency dependent as well. Therefore, corrections for each type of speci-
men have been determined. These corrections are given in Table 3.
Results of typical tests on a Type B specimen are shown in Fig. 4. In the am-
plitude range of concern {A/h > 0.2), the damping data with the accelerome-
ter and cable mounted tend to indicate a parallel trend with data for a clean
configuration. The average differences between values at corresponding tip
<
O
K
UJ
D _
Q lij
li
O cc
o o
t- u.
z o
u z
o —
o
z
a.
s
<
0.2 0.4 0.6 0.8
TIP DEFLECTION/THICKNESS
FIG. 3—Air damping contribution determination.
deflections are used to establish the entries in Table 3. Cable damping has
been found to be insensitive to environmental conditions.
The damping coefficient for a specimen, ^s, is determined by adjusting the
measured damping coefficient, [,M, according to the following equation
CS^CM-{CA + CC) (4)
CA is obtained from Eq 3, Cc from Table 3, and average values of {M from
Table 2. Average specimen damping coefficients obtained in this manner are
presented in Table 4. This is a more useful form for the data.
TABLE 3—Corrections for cable damping.
Dry, 25='C Damping Coefficient Contribution

Specimen Testing Due to Cable,
Type Frequency, Hz percent of critical value
A 54 0.227
B 25 0.193
C 44 0.265
D 27 0.176
CLEAN CONFIGURATION
0,3
_1_
0 02 0.4 0.6
_!_
0.8 10
TIP DEFLECTION/THICKNESS
FIG. 4—Damping tests on a Type B composite specimen in dry air at 25°C {77°F).
Remarks on Damping
Air and cable damping contributions are large. The former must be toler-
ated in environmental tests. The latter is present because the reliability and
suitability of transient response determination by accelerometer was desired.
Their magnitude tends to overshadow the intrinsic specimen damping in
fiber controlled modes of response for Type A and C specimens. The ad-
justed values of damping coefficient, therefore, should be viewed in this
light.
Some observations on the results in Table 4 deserve mention. First, except
for the 25°C (77°F), wet Type D specimen value, at the same temperature,
damping increases with moisture saturation for all specimen types.
Secondly, all specimen types show an immediate increase in damping with
hygrothermal conditioning. A reversal of trend occurs for Type B, C, and D
TABLE 4—Mean adjusted specimen damping coefficients.
Damping Coefficient, percent of critical value
Dry, 25°C 25°C 25°C 60°C 82°C 93°C 82°C

Specimen Testing (77°F) (77°F) (140°F) (180°F) (200°F) (180°F)
A 54 0.062 0.202 0.259 0.211 0.263 0.075

B 25 0.391 0.547 0.576 0.496 0.286 0.375
C 44 0.113 0.242 0.197 0.265 0.002 0.164
D 27 0.504 0.423 0.726 0.582 0.393
specimens, however. Type B and C specimens experience a reduction in

damping for the 93°C (200°F), saturated state; this transition should be
compared to the Type C specimen stiffness reduction noted earlier. A rever-
sal of trend for Type D specimens is apparent at 82°C (180°F), saturated,
however. In view of the approximate nature of the data correction process,
perhaps it is hazardous to attempt to draw firm conclusions regarding these
apparent trend reversals.
The previously mentioned data clarify the behavior of the composite spec-
imens considerably. At the time Ref 1 was written, only extreme environ-
mental data corresponding to room temperature [25°C (77°F)], dry and
93°C (200°F), saturated were available. Obviously, these data alone are mis-
leading. The data at intermediate temperatures are needed to fully grasp the
significance of hygrothermal effects.
The present authors conjecture that the "rise and fall" behavior exhibited
by £/for the Type C specimens and by the damping coefficient may be asso-
ciated with the glass transition of the epoxy matrix. It would appear that a
practical glass transition temperature occurs for the moisture saturated ma-
trix material somewhere between 82°C (180°F) and 93°C (200°F). While a
glass transition temperature below 93°C (200°F) seems somewhat lower than
values normally associated with this material system in a saturated state, the
data clearly indicate a transition that is fully consistent with this hypothesis
[8,9}. Glass transition temperature is decreased, of course, with moisture ab-
sorption. No independent glass transition temperature experiments have
been performed, so this hypothesis remains unsubstantiated by additional
evidence. Results presented in Ref 10 however, tend to support this
hypothesis.
Concluding Remarks
Considerable attention has been given to test conditions and testing tech-
nique. Since environmental effects are determined, vibration testing in a vac-
uum chamber could not be used. Damping contributions due to air and in-
strumentation wiring had to be evaluated in supplementary tests in order to
correct the experimental data obtained in the hot, moist environments.
The experimental results, after correction for the parasitic damping con-
tributions, produced no startling, unexpected findings. Matrix controlled
modes of response are the most affected by hygrothermal conditioning, and
stiffness reduction is likely to be the problem of most practical concern.
Since air and cable damping contributions are large, the adjustment
procedure, although rational, is not fully satisfactory. In particular. Type A
and C specimen adjusted damping coefficient values appear to be generally
higher than expected in the hot, moist cases. In connection with the current
experiments in progress to study frequency effects, specific attention will be
given to improving the means for determining damping. In particular, the
tests are being conducted using the electromagnetic noncontacting transduc-

ers for excitation and response determination. These transducers are emi-
nently suited to the small beams and virtually eliminate parasitic damping
contributions, although they cannot be used for larger scale experiments.
A cknowledgments
This work was sponsored by the U. S. Air Force Office of Scientific Re-
search under Grant AFOSR-73-2479 and Contract No. F49620-78-C-0085.
The composite test specimens were provided by the McDonnell Douglas
Corp.
References
[1] Maymon, G., Briley, R. P., and Rehfield, L. W. in Advanced Composite Materials—Envi-
ronmental Effects, ASTM STP 658, American Society for Testing and Materials, 1978, pp.
221-233.
[2] Briley, R. P., "Effects of Environmental Conditions on the Dynamic Behavior of Resin
Matrix Composites," Ph.D. thesis, Georgia Institute of Technology, Atlanta, Ga., 1981.
[3] Verette, R. M. and Labor, J. D., "Structural Criteria for Advanced Composites,"
AFFDL-TR-76-142, Vol. 1, Air Force Flight Dynamics Laboratory, Wright-Patterson
Air Force Base, Ohio, 1976.
[4] Browning, C. E., Husman, G. E., and Whitney, J. M. in Composite Materials: Testing and
Design (Fourth Conference), ASTM STP 617, American Society for Testing and Materials,
1977, pp. 481-496.
[5] Walter, G. A., "Special Problem Report for AE 8500," Georgia Institute of Technology,
Atlanta, Ga., 1976.
[6] Dugundji, J. in Proceedings of the Sixth Annual Mechanics of Composites Review, M.
Knight, Ed., AFWAL-TR-81-4001, Air Force Wright Aeronautical Laboratories, Wright-
Patterson Air Force Base, Ohio, 1981, pp. 70-73.
[7] Baker, W. E., Woolam, W. E., and Young, D., InternationalJournal of Mechanical Sci-
ences, Vol. 9, 1967, pp. 743-766.
[8] Lazan, B. J. in Structural Damping, J. E. Ruzicka, Ed., American Society of Mechanical
Engineers, New York, 1959, pp. 1-34.
[9] Lazan, B. J., Damping of Materials and Members in Structural Mechanics, Pergamon Press,
Elmsford, N. Y., 1968.
[/O] Kibler, K. G., "Time-Dependent Environmental Behavior of Graphite/Epoxy Compos-
ites," AFWAL-TR-80-4052, Air Force Wright Aeronautical Laboratories, Wright-Patter-
son Air Force Base, Ohio, 1980.
L. L. Clements and P. R. Lee
Influence of Quality Control Variables

on Failure of Graphite/Epoxy Under
Extreme Moisture Conditions
REFERENCE: Clements, L. L. and Lee, P. R., "Influences of Quality Control Variables

on Failure of Graphite/Epoxy Under Extreme Moisture Conditions," Composites for Ex-
treme Environments, ASTM STP 768, N. R. Adsit, Ed., American Society for Testing
and Materials, 1982, pp. 161-171.
ABSTRACT: Tension tests on (0°)8 T300/5208 graphite/epoxy composites were per-

formed to determine the influence of various quality control variables on failure
strength as a function of moisture and moderate temperatures. The extremely high-
and low-moisture contents investigated were found to have less effect upon properties
than did temperature or the quality control variables of specimen flaws and prepreg
batch-to-batch variations. In particular, specimen flaws were found to reduce drasti-
cally the predicted strength of the composite, whereas specimens from different
batches of prepreg displayed differences in strength as a function of temperature and
extreme moisture exposure. The findings illustrate the need for careful specimen prep-
aration, studies of flaw sensitivity, and careful quality control in any study of compos-
ite materials.
KEY WORDS: composite materials, graphite/epoxy composites, tensile strength, en-

vironmental tests, moisture, quality control
The use of composite materials in commercial aircraft primary structures

is hindered by the absence of convincingly reliable techniques for predicting
composite durability under actual service conditions. Development of such
techniques is complicated by the fact that significant changes in composite
durability can occur not only at extreme temperatures, but also at moderate
temperatures due to extreme moisture contents. This problem is being ad-
dressed at the National Aeronautics and Space Administration (NASA)-
Ames Research Center in a program investigating the mechanisms of defor-
mation, strength degradation, and failure of graphite/epoxy composites.
' Research scientist. Advanced Research and Applications Corp., Sunnyvale, Calif. 94086.
Current affiliation: associate professor. Materials Engineering Department, San Jose State Uni-
versity, San Jose, Calif. 95192.
^Foothill/DeAnza Community College District in cooperation with NASA-Ames Research
Center, Moffett Field, Calif. 94035. Current affiliation: Chemical Systems Division, United
Technologies, Sunnyvale, Calif. 94086.
161

Copyright 1982 b y A S T M International www.astm.org
TABLE 1—Physical and mechanical properties specified for WYP-30-1/0

(zero twist) grade of Thornel 300 graphite fiber.'
Physical properties
Filaments/fiber bundle 3000
Twist none
Filament density* 1.73 Mg/m'
Filament equivalent diameter* 6.9 urn
Mechanical properties
Minimum tensile strength 2660 MPa (385 ksi)
Average tensile modulus 200 to 240 GPa (32 to 35 msi)
Minimum average strain to failure 1.1%
"Dale Black, NARMCO Materials, Inc., Costa Mesa, Calif., June 1979,
'Not part of specification. Taken from Union Carbide Corporation product literature.
The portion of that work to be reported here involves an assessment of the

influence of various quality-control variables—specifically prepreg batch,
cure conditions, and specimen quality—on the effect of moisture and mod-
erate temperature on the tensile properties of (0°)8 Thornel 300/NARMCO
5208 graphite/epoxy composites.
Experimental Procedure
Materials
The "T300/5208" graphite/epoxy composite was fabricated from prepreg
tape manufactured by NARMCO Materials, Inc., from Union Carbide Cor-
poration's Thornel 300 graphite fiber, and NARMCO's 5208 epoxy resin.
Table 1 gives the physical and mechanical properties of the WYP-30-1/0
(zero twist) grade of Thornel 300 fiber used in the prepreg. The NARMCO
5208 epoxy resin is one of several commercial epoxies based on the
TGDDM-DDS system, that is, the main constituents are tetraglycidyl 4,4'-
diaminodiphenyl methane epoxy (such as Ciba Geigy MY-720) cured with
4,4'-diaminodiphenyl sulfone (such as Ciba Geigy Eporal). The 5208 system
contains about 90 parts per hundred (pph) by weight of TGDDM, about 24
pph DDS, and about 10 pph of glycidyl ether of a bisphenol-A novolac
epoxy (Celanese SU-8).'
Specimen Fabrication
The specimens used in this study were fabricated for NASA-Ames by an
outside vendor. Large (approximately 1 m^) panels of the suitable lamination
sequence were prepared from one of two different batches of 0.3-m-wide
prepreg tape. These panels were cured in an autoclave held for 1/2 h at
' May, C. A., "Exploratory Development of Chemical Quality Assurance and Composition of
Epoxy Formulations," Technical Report AFML-TR-76-112, Air Force Materials Laboratory,
Dayton, Ohio, 1976.
CLEMENTS AND LEE ON QUALITY CONTROL VARIABLES 163
135°C and then 2 h at 180°C, under 700 kPa pressure. The average volume
percent fiber from these panels was determined to be 64.6 percent, with a
range of 64.3 to 64.8 percent. Next, vendor-fabricated tabs made from 0/90-
deg fiber glass fabric and epoxy resin were bonded to the panels. The tab ad-
hesive used for the panels made from prepreg Batch A was FM-143 adhesive
by 3M, cured for 1 h at 125°C and 50 psig. For the panels made from prepreg
Batch B, an unknown but reportedly comparable adhesive was used. Speci-
mens then were cut from the panels using a dry carborundum cut-off wheel.
The nominal configuration of these specimens was 12.7 mm wide, 1.2 mm
thick, with a gage length of 127 mm, and 60-mm-long fiber glass tabs.
The as-received specimens were found to suffer from numerous fabrica-
tion defects. Two problems were particularly troubling: extensive torn fibers
stuck out from the cut sides, and numerous edge delaminations extended
into the cut sides of specimens; and the composite itself, or the composite
plus the tabs, tended to be bowed. Because of our concern over these defects
and because of reproducibility problems in our preliminary results, most of
the specimens used in this study were subjected to extensive screening and
rework. This procedure involved screening the specimens and rejecting any
specimens that exhibited obvious bow or other irreparable defects. The spec-
imens were then reworked by wet-polishing sufficient material off both cut
sides to remove all detectable (at X30) torn or delaminated material. This
screening and rework procedure was done to bring all specimens up to the
quality outlined in the ASTM Test for Tensile Properties of Fiber-Resin
Composites (D 3039-76). The resulting width of these polished-edge speci-
mens ranged from 10 to 12 mm. The properties of unpolished = edge spec-
imens were also investigated. These specimens were screened for bow and for
edge defects. Only specimens without visually obvious edge delaminations
were retained.
In order to answer questions about the possible influence of degree of cure
upon 0-deg properties, some of the specimens were given a postcure of 2 h at
200°C, followed by a slow oven cool.
Environmental Conditioning
As-received specimens contained 0.15 to 0.45 percent moisture. Specimens
destined for mechanical testing were first dried in a vacuum desiccator at
100°C for 7 days, then held for at least 2 days under vacuum at room
temperature. Weight gain/loss studies (on dummy specimens, without tabs,
yet taken from the same panels) confirmed that this was sufficient time to en-
sure complete moisture removal from the specimens. Specimens to be tested
in the "dry" condition were left in a room-temperature vacuum desiccator
until being tested at the appropriate temperature and <5 percent relative
humidity. All specimens to be tested in the "wet" condition were placed,
after drying, in an environmental chamber at 60°C and approximately 100
percent relative humidity for at least 60 days. This process produced essen-
tially complete moisture saturation/ Work by Adamson, however, has

shown that the so-called "reverse thermal effect" can produce an even higher
moisture content than "normal" saturation in similar graphite/epoxy com-
posites.' This reverse thermal effect occurs when specimens are first satu-
rated, or nearly saturated, with water at a given temperature and then are
placed in water (or high humidity) at a lower temperature. Thus, in order to
increase moisture content to a true extreme, wet specimens were finally held
at room temperature and at essentially 100 percent humidity for at least 45
days before testing. Weight gain from the dummy specimens confirmed such
an additional increase in moisture content: specimens conditioned for at
least 60 days at 60°C and 100 percent relative humidity contained 1.57 ±0.23
percent of water (by weight), whereas specimens further conditioned for 45
days at room temperature and 100 percent relative humidity contained 2.02
±0.13 percent water.* All wet specimens tested in the study were subjected to
this additional moisture exposure. In addition, all specimens were held at
room temperature and —100 percent relative humidity until being tested at
the appropriate temperature and —100 percent relative humidity.
Mechanical Testing
Tension tests to failure were performed inside an environmental chamber
using a 10 000-kg-capacity servo-hydraulic mechanical testing machine. The
tensile grips were mounted to an alignment device consisting of a three-post
ball-bearing die set. This alignment device was mounted in turn to the hy-
draulic actuator and, through a universal joint, to the load cell and load
frame. Longitudinal strain was measured using an axial strain-gage exten-
someter; for many of the tests, transverse strain also was measured using a
diametral extensometer.
All tests were done at a constant elongation rate which resulted in an ac-
tual strain rate of 3 X 10"'i"'. Time to failure (at about 1 percent strain) was
approximately 5-1/2 min.
Experimental Matrix
The experimental conditions studied were as follows.
1. Temperature: 25 and 96°C.

2. Moisture content: dry — 0 percent [tested at < 5 percent relative humid-
ity at 25°C (77°F), < 2 percent relative humidity at 96°C]; wet = 2 percent
(tested at —100 percent relative humidity).
3. Prepreg batch: Batches A and B.
* Lifshitz, J. M., "Strain Rate, Temperature, and Humidity Influences on Strength and Mod-
uli of a Graphite/Epoxy Composite," to be published in Composites Technology Review, Vol. 4,
No. 1, Spring 1982.
'Adamson, M. J., Journal of Materials Science, Vol. 15, 1980, p. 1736.
' All limits given in this paper are 95 percent confidence limits, based on the "t" test.
4. Cure condition: not postcured, cured as received; postcured 2 h at

200°C.
5. Specimen quality: polished edges = specimens screened for bow and
with cut sides wet polished to remove damage; unpolished edges = speci-
mens screened for bow and for visually obvious damage to cut sides.
Nearly every permutation of conditions was studied. However, in the case

of unpolished-edge specimens, only specimens from Batch A, mostly not-
postcured, were considered.

Our concern with the effect of prepreg batch upon properties resulted
from some early findings in this study. Anomalous results from some early
tests were traced to specimens prepared from prepreg Batch B. A micro-
scopic investigation of tested and untested Batch B specimens and of the
prepreg itself was performed. From optical microscopy, scanning electron
microscopy (SEM), and consultation with NARMCO and Union Carbide,
we concluded that there were indeed some differences between Batch B and
other prepreg such as Batch A. Some of these differences are illustrated in
Fig. 1. A photomicrograph of a laminate made from "normal" prepreg (such
as Batch A) is shown in Fig. la. The cut and polished filament ends reflect
light very effectively and thus appear light-colored. A photomicrograph of a
laminate made from prepreg Batch B is shown in Fig. lb. In this laminate,
there are light and dark areas, with the transition between such areas occur-
ring both between layers and within a single layer. In the dark areas, as can
be seen from the magnified view of Fig. Ic, the individual filaments have
been damaged. Figure Ic also illustrates that the dark areas are frequently
connected with individual fiber bundles (3000 filaments to a bundle). The
obvious conclusion that the filaments somehow are degraded in these areas
was refuted by careful polishing and SEM work, which showed that the
epoxy matrix in these areas somehow is altered and weakened. Unless ex-
treme care is exercised, the apparent filament degradation actually occurs
during metallographic preparation. The altered epoxy matrix allows the fil-
aments to move around and thus be damaged during polishing. Since the al-
tered epoxy occurs within and around individual fiber bundles, the epoxy
probably has reacted to some surface effect or contaminant on some of the
fibers used to make up the prepreg. We suspect that this effect is related to
another problem encountered on a few occasions by other T300/5208 users,
and labeled "zebra tow" by them.' In this case, some of the surface tows
(fiber bundles) of the composite panels tended to pull loose from the panel
when the peel ply was removed. We have noted some such surface features
on specimens made from our Batch B prepreg.
'Dale Black, NARMCO Materials, Inc., Costa Mesa, Calif., private communication, Dec.
1979.
100*«n—<*j f*—
10|BM—«4 I"* • '
FIG. 1—Photomicrographs of T300/520S laminates from "normal" {Batch A) and "anoma-

lous" (Batch B) prepreg. (a) Batch A, (b) Batch B, and (c) Batch B: anomalous fiber bundle among
normal bundles.
Figure 2 is a plot of axial elastic modulus, f u , as a function of tempera-

ture and moisture content at two strain levels. As can be seen from this fig-
ure, Eu (as determined from the slope of the stress-strain curve) is statisti-
cally significantly higher at an axial strain, en, of 0.5 percent than at
€11 = 0.1 percent. Other researchers have also observed this increase in mod-
ulus with strain.*'* Figure 2 also illustrates that the increase holds true for
all combinations of moisture and temperature. One possible explanation for
(121 t ^ i ^
(6)T •
(17)
(ll)I^
(12) M1°
IITTJ
- • a -0% MOISTURE-
• o -2% MOISTURE
1 1 1 1
FIG. 2—Axial elastic modulus, Bu as a function of temperature at two different axial strains,
en, levels and moisture contents for Batch A specimens. {Error bar shows 95 percent confidence lim-
its. Numbers in parentheses are numbers of specimens.)
the phenomenon is that curved filaments in the composite straighten with in-
creased strain so that more filaments carry the applied load/ Another possi-
ble explanation is that there may be a strain-induced improvement in orien-
tation of the covalently bonded carbon platelets within the individual
filaments.'
Other findings on the behavior of f u were as follows.
1. There is no statistically significant effect of the cure condition or spec-

imen quality on £ii.
2, There may be an influence of prepreg batch upon En- The mean for
Batch B is systematically lower than that for Batch A. For example
£ii (0.1 percent en) £ n (0.5 percent en)
Batch A 129.5 ± 1.9 GPa 139.5 ± 2 . 5 GPa (Af= 12)'°

25°C, wet
Batch B 127.3 ± 1.6 GPa 137.1 ± 1.9 GPa (N = 9)
In this case, the mean of the Batch A specimens is greater than that of Batch
B at the 95 percent confidence level.
3. There is a statistically significant temperature-induced increase in En
for dry specimens only. (See Fig. 2.) Since the axial thermal-expansion coef-
ficient of the fiber is negative (that is, the fiber contracts axially with in-
creased temperature), the increase in En is not surprising. However, it is not
clear why no such effect is seen for wet specimens.
4. There may be an effect of moisture upon £11, but such an effect is not
systematic. At 25°C and 0.1 percent strain, the mean £11 for the wet speci-
'Sendeckyj, G. P., Richardson, M. D., and Pappas, J. E. in Composite Reliability. ASTMSTP
580, American Society for Testing and Materials, 1976, pp. 528-546.
' M . K. Towne, Union Carbide Corp., Cleveland, Ohio, private communication, Oct. 1979.
^"N = number of specimens entering into statistics.
mens is higher than that for the dry specimens to better than 95 percent con-
fidence. However, at 96°C and a strain of 0.5 percent, the mean £ii for the
wet specimens is lower than that for the dry specimens to better than 99 per-
cent confidence.
Determination of the major Poisson's ratio, v\2, was difficult since the di-
ametral extensometer slipped under many conditions. It was not possible to
get any good data on wet specimens, but the data obtained indicate no influ-
ence of any of the other variables upon vn- In particular, there appears to be
no influence of strain upon vn. The mean value of vn was 0.33 ± 0.01
{N= 14).
A very important finding of this study was the magnitude of the influence
of specimen quality upon 0-deg strength. Figure 3 compares 0-deg strength
as a function of temperature for polished-edge and unpolished-edge Batch A
specimens. As can be seen, the strength is significantly increased by the im-
proved specimen quality resulting from polished edges. It is also interesting
to compare these results with strengths of specimens previously rejected as
having irreparable defects:
0-deg Strength
Polished edge 1542 ± 89MPa(Af=9)

Batch A, 25°C, dry Unpolished edge 1333 ± 79MPa(A'=8)
Rejects 1313 ±71 MPa(iV= 12)
The 0-deg strength of the good unpolished edge is no better than that of the
reject specimens. This finding confirms that the specimen preparation
procedure outlined in ASTM Method D 3039-76 is not overly conservative in
this case.
From Fig. 3, it also is seen that edge polishing produced a change in
strength behavior as a function of temperature. The 0-deg tensile strength of
dry polished-edge specimens increased significantly with an increase in
1800
• D -0% MOISTURE
• O -2% MOISTURE
FIG. 3—0-deg tensile strength as a function of temperature for polished-edge and unpolished-
edge Batch A specimens at two moisture contents.
temperature, whereas the small increase in mean strength for dry unpolished-
edge specimens is not statistically significant. Moisture content, on the other
hand, had no statistically significant effect upon strength for Batch A speci-
mens whether they were edge polished or not. Unfortunately, we were unable
to get reliable strength data at 96°C wet because of end-tab failures. This was
because the additional moisture content we produced using the reverse
thermal effect, when combined with elevated temperature, caused tab failure
to occur before composite failure. (For Batch A specimens, the tabs them-
selves failed prior to composite failure, but for Batch B specimens, the tab
adhesive failed at very low loads.) Thus, our observations on the effect of
moisture content on strength are valid only at 25°C.
These findings are somewhat in disagreement with those of Lifshitz.'* Lif-
shitz studied unpolished-edge specimens and reported an effect of moisture
content on strength at both 25 and 96''C. Lifshitz's specimens, however, were
conditioned at 60°C and —100 percent relative humidity only. We suspect
that the additional moisture content we induced using the reverse thermal ef-
fect produced a degradation that cancelled any strength increase resulting
from nominally wet conditions alone. Lifshitz also reported a statistically
significant increase in strength with temperature for his unpolished-edge
specimens. However, his data compared only "room" and "wet" (without
reverse thermal effect) moisture conditions, whereas ours compares dry spec-
imens only.
Figure 4 illustrates the effect of prepreg batch upon 0-deg strength. The
mean strength at 25°C for wet Batch B specimens is less than that for dry
Batch B specimens at the 90 percent confidence level (but not at the 95 per-
cent confidence level). Far more clearly significant is the difference in
strength between Batch A and Batch B specimens at 96°C. As can be seen
from Fig. 4, the Batch B specimens are much weaker at this temperature. As
we have explained, we unfortunately were unable to fail the composite itself
at 96°C wet.
Scanning Electron Microscopy

In an attempt to understand the mechanisms that have produced the
property changes described above, we have initiated scanning electron mi-
croscopy studies on the failure surfaces of the composite specimens. Figure 5
shows typical failure surfaces for dry, polished-edge, not-postcured speci-
mens. One specimen from each Batch A and Batch B was tested at 25 and
96°C. As can be seen from Fig. 5 the appearance of the individual broken fil-
ament is the same regardless of the temperature or batch. In fact, this holds
true for typical failures at all conditions. Furthermore, there are few single
filaments "pulled out" of any of the failure surfaces, and the amount of
epoxy remaining on the surface of the filaments indicates that the interfacial
bond in all cases is reasonably strong.
£ 1600
S
• a -0% MOISTURE
• O - 2 % MOISTURE
60
T, C
FIG. 4—0-deg tensile strength as a function of temperature for polished-edge Batch A and Batch
B specimens at two moisture contents.
It is difficult, however, to pick out differences in the specimens which

might explain the observed differences in properties. Obtaining such an ex-
planation from the failure surfaces is complicated by a number of factors.
First, the elastic recoil at failure induces significant secondary damage,
which complicates failure-surface analysis. Second, there is a great deal of
specimen-to-specimen variability at any one condition. Third, even for a
FIG. 5—Scanning electron micrographs of failure surfaces of dry, not-postcured, polished-edge

specimens, (a) Batch A, tested at 250''C, (b) Batch B, tested at 25''C, (c) Batch A, tested at 96''C,
and (d) Batch B, tested at 96°C.
given specimen, the differences in appearance between different areas are

great. Thus, no single set of micrographs such as is shown in Fig. 5 can repre-
sent the specimens.
One difference, however, appears again and again regardless of area or
specimen. The matrix in the specimens tested at 96°C shows less of the
smooth, cupped, and glass-like signs of conchoidal failure than does the ma-
trix of specimens tested at 25°C. Unfortunately, this is equally true for Batch
A and Batch B specimens, so it does not explain the observed strength differ-
ences between specimens of the two batches. We hope that further micros-
copy will clarify these observations.
Conclusions
Our results demonstrate that the tensile properties of T300/5208 graph-
ite/epoxy composites are affected by various quality-control variables. Per-
haps the most important finding is the large effect of specimen quality
(preexisting torn fibers and edge delaminations, even if they are not easily
detectable) on strength. Another important finding is that strength and its
variation with temperature and moisture content can be influenced by pre-
preg batch-to-batch variations. These findings lead us to make three points
about studies of mechanical properties of graphite/epoxy. First of all, the
importance of careful specimen preparation technique cannot be overem-
phasized. The procedure outlined in ASTM Method D 3039-76 often is com-
promised in practice, but it is clear from our work that such a rigorous
procedure is justified. Second, the influence of preexisting flaws upon prop-
erties should be an integral part of any study of composite properties. And
third, the influence of such quality-control variables should be established,
or the variable itself carefully controlled, in any composite destined for ac-
tual service exposure.
Acknowledgment
Work for this project was performed at NASA-Ames Research Center,
Materials Science and Applications Office, Moffett Field, Calif., under Con-
tract NAS2-9989.
Summary
STP768-EB/Apr. 1982
Summary
The two most important findings at this conference were not what the
experimenters set out to prove. Shyprykevich and Wolter found that the ac-
celerated tests used to predict real time behavior of graphite/epoxy compos-
ites understated the effects. Meanwhile, Clements found what we already ex-
pected—that the quality of material and its preparation can have a larger
effect on the end properties than the deliberate exposures to harsh environ-
ments. The conclusion one would draw from this is that more studies of the
real life exposures are needed and that all investigators must take great care
in preparation of specimens and accurately report their test conditions.
The papers in this book have been divided into three sections. The first
deals with polyimide matrix composites, while the second and third sections
are concerned with epoxy matrix composites. Government restrictions pre-
vented the presentation of any work on metal matrix composites, even
though they are useful at some of the extreme conditions.
The polyimide matrix is useful at elevated temperatures up to 316°C
(600°F), while epoxies generally are useful to only half that temperature. The
results obtained by Serafini and Hanson clearly show that PMR-15 has a
useful life at the elevated temperature of up to 1000 h. The second paper, by
McKague, shows that a second class of polyimides is available. This class of
materials is more workable by people experienced with epoxies. The proper-
ties are superior to those of the epoxy composite at intermediate tempera-
tures [177°C (350°F)]. This makes them attractive for application on high
performance aircraft.
Kunz found that the more graphitic fibers (high modulus, lower strength)
were more stable in the polyimide resin composites. This confirms the find-
ing of Serafini and Hanson that the quality of the laminate (voids) had an ef-
fect on the susceptibility of the laminate to oxidation. The attempts to mea-
sure thermal conductivity and thermal expansion by Campbell and Burleigh
also were handicapped by the quality of the laminate.
The mechanical properties of similar polyimide matrix composites were
measured by Garber, Morris, and Everett. They found that the laminate
strength and modulus of fiber-dominated composites were not strong func-
tions of temperature in the range —157 to 316°C. The matrix-dominated
175
Copyright by ASTM Int'l

Copyright' 1982 b y A S T M International
(all
www.astm.org
rights reserved); Tue Jan 17
composites' properties were affected by temperature and, one has to suspect,

quality of the laminate.
The second set of papers is application oriented. The work by Morris
shows that glass/epoxy composites have particular applications. The very
low heat flow down glass/epoxy straps and rods makes them excellent for
holding cryogenic storage tanks on earth or in space. Leung's work reviewed
the use of a graphite/epoxy on the space shuttle. He found that the expo-
sures which the materials will experience will not adversely affect their prop-
erties, a fact that should comfort the astronaut who will ride in the shuttle. If
these conclusions were encouraging, the findings of Shyprykevich and Wal-
ter were not. They found that the accelerated tests on graphite/epoxy cou-
pons underestimated the real-time moisture gain and loss of properties.
Their conclusions would lead one to be tempted to park the shuttle in orbit
rather than on a runway in the tropics.
The last three papers are a collection of information on other effects.
Givler, Gillespie, and Pipes found little effect of thermal cycling of graphite/
epoxy between moderate temperatures (—54 to -l-177°C). Rehfield, Briley,
and Putter studied the effect of moisture on the dynamic properties, while
Clements and Lee studied their effect on static properties.
A^. R. Adsit
General Dynamics/Convair Division, San
Diego, Calif. 92138; symposium chairman
and editor.
STP768-EB/Apr. 1982
Index
Failure, 85
Fatigue, 107
Absorbed moisture, 19
Fiber-matrix adhesion, 42
Adhesive, 163
Fickian diffusion, 129, 133
Aircraft, 118-134, 161
Filament wound composite, 95-109
AS carbon fibers, 120, 138, 154
Flaw, 85, 171
Autoclave, 29
Flexure properties, 8, 10,38, 52, 141,
147
B Fortafil fiber, 34
Beams {See Sandwich beams) Fracture, 43, 74, 87
Gamma (7) radiation, 100, 111, 116

Celion carbon fibers, 6, 22, 34, 74,91
Glass fiber, 96, 103, 138
Compression, 96, 120-134, 142, 147
Glass transition temperature {See Tg)
CPI-2237 resin, 34
GY70 carbon fiber, 34
Crack, 11,25, 56, 143
Crack density, 144, 146
H
Cryogenic properties, 83, 90, 95-109
C-scan, 8 Heat capacity, 111
Heat conductivity, 96
D HTS carbon fibers, 6, 55, 71
Hybrid, 137, 146
Damping, 149-160 Hydrothermal, 13
Degradation, 11 Exposure, 8, 148-160
Delamination, 163, 171
Diffusion, 114, 123, 130 I
Dynamic properties, 113, 115, 148-
160 Infra-red, 97
Interlaminar shear strength, 8-19,
39, 52, 113, 116
Elastic properties, 77, 90

Emittance, 59, 70, 72, 111 Long term exposure, 6
M
Fabrication of composites, 6, 22, 36, Mechanical properties, 11, 24, 96,
119, 162 148
177
Copyright' 1982 b y A S T M International www.astm.org
Microcracks, 56 Scanning electron microscope, 50,

Modulus, 13, 24, 37, 78, 105, 144, 165, 169, 170
153-160,167 SCI REZ080 resin, 103
Moisture Shear modulus (See Modulus)
Absorption, 52, 114, 148, 153 Short beam shear (See Interlaminar
Concentration, 18 shear strength)
Content, 150, 164, 166, 169 Space environments, 110-117
Diffusion, 113 Space shuttle, 96
Gain, 122-134 Specific heat, 59, 69, 71
Stacking sequence, 138
N Stiffness (See Modulus)
Strain-energy, 25, 27
Net resin, 29
Strain gage, 75, 104, 120, 164
Nondestructive testing {See C-scan)
Structural dynamics, 148-160
Notched strength, 83, 90
Swelling, 152
NR-150B2 resin, 34, 55, 71
O
Tensile properties, 26, 74, 76, 86,
Outgassing, 106
124, 133, 168, 170
Oxidation resistance, 40, 52
Tg, 15,21,33,36, 115, 119, 129, 133,
141, 159
Thermal aging, 23
Photomicrograph, 11 Conductivity, 56, 60, 71, 105
PMR-15 polyimide, 6-19, 33, 55, 71, Cycle, 138-147
74,91 Degradation, 40
Poisson's ratio, 77, 168 Expansion, 59, 63, 71, 137, 145
Gravimetric analysis (TGA), 162
Isolate, 96, 102, 109
Spike, 8-19, 23-32, 126, 129, 131
Quality control, 161-171
Stress, 138, 146
Thermomechanical properties, 25, 33-
R
53
Radiative degradation, 111 Thermooxidative exposure, 6
Real time, 119, 124, 132 Stability, 10, 40, 42
Relative humidity, 150 Thermophysical, 54-72, 96
Residual stress, 137, 145 Thick laminate processing, 27
Resin Transverse cracking, 141, 146
934, 112 T 300 carbon fibers, 8-19, 22, 112,
3501, 120, 138, 154 149, 162
5208, 149, 162
U
Unnotched specimens, 74
Sandwich beam, 150-160 Strength, 90
INDEX 179
xr Ml r^n r/;i Weight loss, 9, 40

Vacuum, 111, 150, 163
Vibration test, 97, 151-160 Y
Void content, 29, 45, 52, 63, 103, 112
V378A polyimide, 20-32 Young's modulus (See Modulus)

STP768 Eb.1415051 1 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

STP768 Eb.1415051 1 PDF

Uploaded by

Copyright:

Available Formats

Copyright by ASTM Int'l (all rights reserved); Tue Jan 17 20:03:14 EST 2012

ASTM SPECIAL TECHNICAL PUBLICATION 768

ASTM Publication Code Number (PCN)

1916 Race Street, Philadelphia, Pa. 19103

Printed in Baltimore. Md.

ASTM Committee on Publications

PoLYiMiDES—MATERIALS FOR HIGH-TEMPERATURE ENVIRONMENTS

Environmental Effects on Graphite Fiber Reinforced PMR-IS Polyimide—

V378A Polyimide Resin—A New Composite Matrix for the 1980's—

Iliermomechanical Characterization of Graphite/Polyimide Composites—

Thermophysical Properties Data on Graphite/Polyimide Composite

Elastic Properties and Fracture Behavior of Graphite/Polyimide

ATMOSPHERIC AND EXTROATMOSPHERIC ENVIRONMENTS

Filament Wound Composite Thermal Isolator Structures for Cryogenic

Space Environmental Effects on Graphite/Epoxy Composites—c. L.

Effects of Extreme Aircraft Storage and Flight Environments on Graphite/

Environmental Exposure of Carbon/Epoxy Composite Material

Dynamic Tests of Graphite/Epoxy Composites in Hygrothermal

Copyright by ASTM Int'l (all rights reserved); Tue Jan 17 20:03:

This volume presents a state-of-the-art review of "Composites for Ex-

Copyright by ASTM Int'l (all rights reserved); Tue Jan 17

Environmental Effects on Graphite

REFERENCE: Serafini, T. T. and Hanson. M. P., "Environmental Effects on Graphite

ABSTRACT: Studies were conducted to establish the effects of thermo-oxidative and

KEY WORDS: PMR-15 polyimide composites, graphite fabric, thermo-oxidative ex-

Fiber reinforced polymer matrix composites, particularly those based on

Copyright by ASTM Int'l (all rights reserved); Tue Jan 17

The effects of long-term exposure in air at elevated temperatures (thermo-

STRUCTURE NAME ABBREVIATION

ax' ' MONOMETHYL ESTER OF 5-NORBORNENE-

" 2 ' ^ \ / "-"2 \ /'"'2 4,4'-METHYLENE0IANILINE MOA

FIG. 1—Monomers used for PMR-15 polyimide.

was 2NE/3.087MDA/2.087BTDE. The PMR-15 solution was prepared by

Composite Fabrication and Specimen Preparation

Composite Environmental Exposure

Results and Discussion

FIG. 2—Weight loss of graphite fibers exposed in air at 316°C.

FIG. 3—Weight loss of T300 graphite fabric/PMR-15 composites exposed at elevated

FIG. 5—Typical photomicrographs ofTSOO graphite fabric/PMR-15 composites after various

Figure 8 shows the mechanical properties retention of 0.14 and 0.28-cm-

Figure 10 shows a comparison of the properties of moisture saturated

FIG. 9—Typical photomicrographs ofTiOO graphite fabric/PMR-15 composites after various

FIG. 10—Comparison of hydroihermally exposed HTS/PMR-15 composite properties to base-

FIG. 11—Comparison of hydrothermally exposed T300fabnc/PMR-I5 composite properties to

V378A Polyimide Resin—A New

ABSTRACT: During 1979, General Dynamics conducted an extensive screening pro-

KEY WORDS: composites, graphite/polyimide. flex properties, tension properties,

High-temperature epoxy resins have been investigated widely and used in

'Engineering specialist senior. General Dynamics, Fort Worth, Tex. 76101.

Copyright by ASTM Int'l (all rights reserved); Tue Jan 17

Extensive work on this problem has been conducted at Genera! Dynamics

tnaleimide; X108-43A, a modified bis-maleimide; X108-55B, a modified bis-

STRAIN ENERGY RELEASE RATE

FIG. 1—Properties screening plan.

FIG. 2—Cure cycles for screened materials.

criteria were established for each step of the evaluation, as summarized in

TABLE 1—Postcure cycles for screened materials.

Resin Postcure Cycle

F-178 6 h at 246°C (475°F)

TABLE 2—General and specific screening criteria.

FIG. A—Stress-strain comparisons ofT300-6K/V37SA and T300-3K/5208 at lirC (,350"?).

Thick Laminate Processibility

80 too 120 140 160 110