Composite Failure Analysis
Composite Failure Analysis
DOT/FAA/CT-91/23
VOL II, Part 1 i'l li'1 I t
R.J. Kar
Northrop Corporation
One Northrop Avenue
Hawthorne, California 90250-3277
0
February 1992 U.S. ipcrtnwf
1990
Final Report for Period January 1987 - October
0
Approved for public release; distribution unlimited.
MATERIALS DI RECTORATE
WRICHT LABORATORY 2.
D T IC ? " TF
AIR FORCE SYSTEMS COMMAND
WRIGHT-PATTERSON AIR FORCE BASE, OH 45433-6533 AAYO 11992
arid
92-12348
.- IlII:IIhI,,lI IIj/l
III:/lInIlI;III:
NOTICE
When Government drawings, specifications, or other data are used for any purpose other
than in connection with a definitely Government-related procurement, the United States
Government incurs no responsibility or any obligation whatsoever. The fact that the Government
may have formulated or in any way supplied the said drawings, specifications, or other data, is not
to be regarded by implication, or otherwise in any manner construed, as licensing the holder, or
any other person or corporation; or as conveying any rights or permission to manufacture, use, or
sell any patented invention that may in any way be related thereto.
This report is releasable to the National Technical Information Service (NTIS). At NTIS,
it will be available to the general public, including foreign nations.
This technical report has been reviewed and is approved for publication.
If your address has changed, if you wish to be removed from our mailing list, or if the
addressee is no longer employed by your organization please notify WI/MLSA, WPAFB, OH
45433-6533 to help us maintain a current mailing list.
Copies of this report should not be returned unless return is required by security
considerations, contractual obligations, or notice on a specific document.
REPORT DOCUMENTATION PAGE 1 Form Approved
jOM NoO0104-0188
0
Ub)1rCeportirq hwdeenflor, thi, _^!Lmon of Atformation ii estirmAted tc a.efage I NIOWoe%,pr (uding'h
e r<PO
: t "r.,(t~fl - tn-
I d,!a -we
gaiherr'.g and d., a neendedand Completing ano te'e-noq the :ollecuon Qf ,nf nat~c, Sendri
hanr,~te reg-d~rig t?'. bc-rrdo~t aoett thnl
.
ti,
coI~cto-of
fon,,to *nI~ding wggesiions for .eduong this bnnrden t ian ~ o edn3ts~r"
Davishighwa-y. Suite 1204. A~lingtonVA 22202-4302. and to the OffiCe of M.anagement And Bu~dgetPapt-aCfKPr ~ ~ nOea~~o
lP.O : 00 0,98) N."
n~rs 2I efro
tn, i.C i0503
qtr'
R.J. Kar
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION
REPORT NUMBER
Northrop Corporation
Aircraft Division
One Northrop Avenue
Hawthorne, California 90250-3277
9. SPONSORING'MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING/ MONITORING
AGENCY REPORT NUM5ESR
Wright Laboratory (WL/MLSA)
Materials Directorate WI.-TR-91-4032,
Wright-Patterson AFB, Ohio 45433-6533 DOT/FAA/CT-91-23
Volume 11 - Part 1
11. SUPPLEMENTARY NOTES
-
FEDERAL AVIATION ADMINISTRATION TECHNICAL CENTER
Additional Funding/Sponsorship U. S. DEPARTMENTr OF TRANSPORTATION
Provided By: ATLPNTIC CITY, NEW JERSEY 08405
12a.DISRIBTIO
SATEINT12b. / AAILAILIY DISTRIBUTION CODE
The objective of this program was to create a comprehensive handbook for use in con-i
ducting failure analysis investigations on failed composite structure. This programi
builds upon previous efforts as documented in the "Compendium of Post-Failure
Analysis Techniques for Composite Materials," AFWAL-TR-86-4137. The purpose of
creating this handbook was to document the techniques, the fractographic and
material property data and case history studies currently being utilized in the
analysis of failed composite structure. The major tasks on this programi included:
(1) procedural guidelines for field investigation techniques; (2) an expanded
fractographic data base tor carbon/epoxy materials tested under known conditions,
(3) a fractographic data base for resin based composite materials other than carbon!
epoxy; (4) fractographic documentation of composite material and processing defects;
(5) docuinentatioi. of fracture characte~-istics in adhesive arid mechanical joint
failures; (6) compilation of material property data for composite materials; and
(7) documentation of case histories recently conducted on failed composite structure
17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT
OF REPORT OF THIS PAGE OF ABSTRACT
UNCLASSIFIED UNCLASSIFIED UNCLASSIFIED
NSN 7540-01-280-5500 Stafrdard Formo 2913'Rev 2-89)
2-38 0- 1 r102 1 .i
SUMMARY
The objective of this program was to develop a comprehensive handbook for failure analyses of
fiber-reinforced composites. The program objectives were accomplished through technical tasks that
resulted in the compilation of a reference manual for evaluating failed composite structures.
A field handling logic network was prepared for on-site handling of composites during accident
investigations. Procedural guidelines weze developed from inputs provided by key field personnel from
several government agencies, and from the results of tests performed in-house at Northrop. Several
current and new fractographic techniques were evaluated to identify methods for initiation site
determination and failure sequence identification in failed composite specimens. MacrophotogTaphy,
ply-sectioning, and photographic methods were determined to be valuable supplemental techniques but
could not directly provide initiation site/fracture propagation directiGn when used alone. The
microchemical analysis technique of Fourier Transform Infrared Spectroscopy was determined to be
useful in contaminant failure investigations but will require development of a database of chemical
"signatures."
Northrop expanded the fractographic database originally developed by the Boeing Company for
AS4/3501-6 graphite/epoxy (Gr/Ep) under Air Force Contract No. F33615-84-C-5010 to include the
effects of load, manufacturing, processing, and environmental variables on simple interlaminar and
translaminar test coupons. It was determined that applied load was the principal parameter that altered
the fracture surface characteristics in Gr/Ep. Material form and processing variables indirectly affected
the fracture characteristics in that these caused localized variations in applied load, thereby altering
fractographic features. No significant effects of environment on fracture surface features were deter-
mined. The fractographic database also included documentation of manufacturing and processing defects
that occur in Gr/Ep. The flaws were characterized using optical microscopy, and macrophotography
techniques.
Failure modes in adhesively bonded Gr!Ep and graphite/bismaleimide (Gr/BMI) specimens were
also characterized. Variations in ply thickness, orientation, and loading were carried out to develop mixed
cohesive-adhesive, and singular cohesive or adhesive failures. It was determined that specimen geometry,
lap/strap ratios, and test load played roles in controlling fracture surface characteristics. Fracture
characteristics in the failed adherends served as indicators of fracture direction in mixed and total
adhesive failure modes. The crack directions could not be readily determined in pure cohesive joint
failures.
A test matrix was developed for characterizing the six different failure modes in mechanically
joined composite structures. A computer code entitled SAMCJ (Strength Analysis of Multifastened
Composite Joints), previously developed by Northrop for the USAF, was run to develop the matrix for
quasi-isotropic ASR'3501-6 Gr/Ep joined with titanium "Hi-Lok" tension or shear-type flush head
iii
fasteners. Failure tests and fractographic evaluation were carried out on the specimens. It was determined
that the failure modes wei e a function of applied load, specimen, and fastener geometries.
Detailed in-plane shear tests were also carried out for Gr/Ep. This failure mode was characterized
by the occurrence of hackles on fractured resin and tension fracture characteristics on fractured fiber
ends. Processing variables did not significantly alter the f- Xcture surface characteristics for Gr/Ep tested
under in-plane shear. The information gained from the Northrop and Boeing Gr/Ep studies was used in
initiating a fractographic database for other material systems. The material systems chosen were kevlar
49/3501-6 epoxy (K/Ep), AS4 graphite/5250-3 bismaleimide (Gr/BMI), and AS4 graphite/APC-2 PEEK
thermoplastic (Gr/PEEK). Testing and fractographic evaluation were carried out. for baseline and several
variable conditions. The results for these systems indicated that the type of resin and fiber played strong
roles in controlling the resulting fracture surface characteristics. As for Gr/Ep, envi.'onment and
processing variables did not significantly alter fracture characteristics.
Ncrthrop reviewed formats previously used for reporting metallic and composite fractography
and failure analysis data. Based on an assessment of existing report schemes, Northrop proposed three
data formats for (1) reporting fractographic data, (2) failure analysis information, and (3) organization of
the Composite Failure Analysis Handbook. These were subsequently approved by the Air Force with
minor modifications.
Northrop compiled material properties on current and near-term composite structural materials.
Literature searches were carried out on government and commercial databases for product information
anti properties. Propeities obtained were incorporated into database files using a personal computer. The
data were organized into tabular formats for reporting in the Handbook. The properties for several
classes of fiber, prepreg, and laminates were compiled and organized into the Handbook.
Under an engineering services agreement between Northrop and the university of Utah,
Professor Wiilard Bascom of the University of Utah performed a literature search and made on-site visits
to several government agencies to gather information on composite fractography and failure analysis that
may have been performed at these agencies. No other information was found other than that previously
reported by Boeing. Dr. Bascom also reviewed stress analysis methods and failure micromechanisms for
use in failure analysis investigations. A new failure criterion developed by Dr. Richard Christensen of
Lawrence Livermore Laboratories was determined to be of utility in composite failure investigations.
Verification of the composite failure analysis logic system was perfbrmed through evaluation of
several failed structural items provided by the Air Force. The structural items represented "real-world"
configurations and included (1) a vertical stabilizer, (2) a horizontal torque box assembly, (3) a canopy
support fitting, and (4) two simple components. All the results are presented as case histories in the
Handbook.
As part of the verification process, two simp!e Gr!Ep structures containing intentional defects
were fabricated and tested to failure under controlled laboratory conditions. The failed specimens an(d
related test documentation were shipped to the Air Force for subsequent evaluation by the Boeing
Company.
The Composite railure Analysis Handbook is divided into two volumes. Volume I is the
Program Overview. Volume II comprises the Technical Handbook, and is divided into three parts. Part i
describes all the techniques and procedures for performing composite failure analysis. Part 2 represents
iv
an atlas of fractographs. Part 3 is a compilation ot case histories of investig-itio.as performed by
Northrop, Boeing- and General Electric.
In summary, Northrop has ach;eved the objective of produchig a Handbook containing all the
known techniques, procedures, sample data, and reference suppo-t~ng data for performing post-failure
analysis of fiber-reinforced coamposite structures.
Aceeason lFor
ITIS GRA&I
DTIC TAB 0
Uuannounced Q3
Justificatlon
By
Di.stribut lor)/
Avallbility Codes
, ,t Ds ,indh/or0
IAvSpeo Ia].
0
FOREWORD
Vii
tABLE OF CONTENTS
Section Page
ix
TABLE OF CONTENTS (Continued)
Section Page
x
TABLE OF CONTENTS (Continued)
Section Page
11 G LO SSA RY ........................................................... 11 -1
0
xi
TABLE OF CONTENTS (Continued)
Appendix Page
xm-
xii m
LIST OF ILLUSTRATIONS
Figure Page
2-1 Rockwell Flaw Criticality Study Defect. 1. External Delamination .............. 2-5
2-2 Rockwell Flaw Criticality Studly Defect 2: Blister, In.ernal Delamination ........ 2-5
2-3 Rockwell Flaw Criticality Study Defect 3: Oversized Hole ..................... 2-6
2-4 Rockwell Flaw Criticality Study Defect 4: Hole Exit Side Broken Fibers
B teakout .................... ........................................ 2-6
2-5 Rockwell Flaw Criticality Study Defect 5: Tearout in Countersink .............. 2-7
2-6 Rockwell Flaw Cnticality Study Defect 7. Resin-Starved Laminate
(Bare Fibers) .... ..................................................... 2-7
2-7 Rockwell Flaw Ci iticality Study Defect 8: Resin-Ricii Areaat Arrow
(Dark Areas Are Voids) .................................................. 2-8
0--. ....... .....
*tl.AAt
C......u .
wSt~tJ ..TN
..
t ri.
~
V
.~
V;4.
D -
*. Vod
7I.........00.Prs
'Void ............. 2 -,
2-9 Rockweli Flaw Criticality Study Defect 10: Scratch, Handling Damage .......... 2-9
2-10 Re kwell Flaw Ciiticality Study 1eleect 11: De.i, Handling Damage ............ 2-9
2-11 Rockwell Flaw Criticality Stady Defect 12: Fiber Breakaway From Impact
Surface .............................................................. 2-10
2-12 Rockwell Flaw Criticality Study Defect 13: Splintering, Edge Delamination ...... 2-10
2-13 Rockwell Flaw Criticality Study Defect !4- Overtorqued Fastener .............. 2-11
2-14 Rockwell Flaw Criticality Study Defect 16: Edge Notch, Crack ................. 2-11
2-15 Rockvw-X r law Criticality Study Defect 17: Corner Not.h ..................... 2-12
2-16 Rockwell Flaw Criticality Study Defect 18: Mislocated Hole - Not Repaired ..... 2-12
2-17 Rockwell Flaw Criticality Study Defects 20 and 21: Marcelledt Fibers (Arrow)
and W rinkles .......................... ............................... 2-13
0
2-18 Rockwell Flaw Criticality Study Defect 26: Wrong Material, Correct Layup l 112T. 2-13
2-19 Rockwell Flaw Criticality Study Defect 27: Misoriented Ply From Correct Layup
I 45/90/02/ 45/02/i 4 5 /01S ............................................. 2-14
2-20 Rockwell Flaw Criticality Study Defect 28: Ply Overlap ....................... 2-15
2-21 Rockwell Flaw Criticality Study Defect 29: Ply Undh.lap, Gap ................. 2-15
2-22 Rockwell Flaw Criticality Study Defect 31: Improper Fastener Seating .......... 2-16
2-23 Rockwell Flaw Criticality Study Defect 33: 8 Hole ........................... 2-16
xiii
LIST OF ILLUSTRATIONS (Continued)
Figure Page
2-24 Rockwell Flaw Criticality Study Defect 35: Off-Axis Drilled Hole ............... 2-17
2-25 Rockwell Flaw Criticality Study Defect 36: Countersink on Wrong Side of
Lam inate ............................................................. 2-17
2.26 Rockwell Flaw Criticality Study Defect 39: Burned Drilled Hole ................ 2-18
2-27 Rockwell Flaw Criticality Study Defect 41: Undersized Fastener ............... 2-18
2-28 Rockwell Flaw Criticality Study Defect 42: Dent, Hidden Fiber Breakage ........ 2-19
4-1 Field Handling Logic Network for Composite Parts .......................... 4-2
4-2 Visual Macroscopic Fracture Surface Features .............................. 4-5
xiv
LIST OF ILLUSTRATIONS (Continued)
Figure Page
6-10 Cross-Sectional Diagram of the DSC Cell Used by DuPont DSC Module ......... 6-15
6-11 DSC Thermogram for 3501-6 Resin ....................................... 6-16
6-12 Sample Fingerprinting of Various Systems by Way of X-Ray Fluorescence ....... 6-17
6-13 Pyrolysis Gas Chromatography of Two Difficult Resin Systems ................ 6-19
6-14 Gas Chromatography of DDS, Diaminodiphenyl Sulfone ...................... 6-20
6-15 IR Comparison of Resin ................................................. 6-21
6-16 DRIFTS of Graphite-Epoxy (AS/3501-5) Composite Before and After Thermal
A ging ................................................................ 6-22
6-17 Glass Transition Temperature Determination - TMA Penetration Test Setup 6-23
6-18 Glass Transition Temperature Determination - TMA Penetration Test
M easurem ent .......................................................... 6-23
6-19 Glass Transition Temperature Determination - TMA Flexure Test Setup ...... 6-24
6-20 Glass Transition Temperature Determination - TMA Flexure Test
M e~a ure ienL .......................................................... 6-24
6-21 Determination of Tg by Expansion Me,.hod ................................ 6-25
6-22 Typical DMA Plot for Cured Epoxy Showing Glass Transition Determination . .. 6-25
6-23 Typical DSC Plot for Cured Epcxy Illustrating Determination of Tg and Heat
of Additional Reaction ..................................... ........... 6-26
6-24 DMA Loss Modulus Versus Temperature Showing Effect of Undercure for
Epoxy Resin ........................................................... 6-27
6-25 SEM Micrograph of a Teflon-Contaminated Tension Specimen ................ 6-29
6-26 SEM Micrograph of a Frekote Release Agent Contaminated Tension
Specim en (200X) ....................................................... 6-30
6-27 SEM Micrograph of a Frekote Release Agent Contaminated Tension
Specim en (2,OOOX) ....................... ............................. 6-30
6-28 SEM Micrographs of a Typical Uncontaminated Fracture Specimen ............ 6-31
6-29 SEM Micrographs of Foreign Particle Inclusions illustrating the Characteristic
Fracture Surface Asperity and Radiating Fracture Lines ..................... 6-32
6-30 Initial Inspection and Determination of a Foreign Material Contamination ...... 6-34
6-31 Logic Network for Chemical Analysis of Foreign Material Contamination of a
Composite Fracture Surface ... ......................................... 6-35
6-32 Talc Powder Contamination Along a Bond Line ............................. 6-36
6-33 Typical XPS Spectrnm of an Epoxy Resin Vth Elements Identified ............ 6-38
xv
LIST OF ILLUSTRATIONS (Continued)
Figute Page
xvi
LIST OF ILLUSTRATIONS (Continued)
Figure Page
10-i Analysis Collection and Tracking System (FACTS) Data Input Sheet ........... 10-2
10-2 Nondestructive Examination Data Input Sheet .............................. 10-3
10-3 Materials Characterization Data Input Sheet ............................... 10-4
10-4 Materials Characterization Data Input Sheet (Figure Attachment) ............. 10-5
10-5 Fractography Macroscopic Analysis Data Input Sheet ........................ 10-6
10-6 Fractography Macroscopic Analysis Data Input Sheet (Photo Attachment) ....... 10-7
xvii
LIST OF ILLUSTRATIONS (Continued)
Figure Page
""viii
0
LIST OF TABLES
Table Page
xix
LIST OF TABLES (Continued)
Table Page
xx
0
SECTION 1
Advanced composites are rapidly emerging as a primary material for use in near-term and
next-generation aircraft as they provide greater structural efficiency at lower weight than equivalent
metallic components. Based on trends to date, the next generation of military aircraft could contain as
much as 65 percent of their structural weight in advanced composite materials.
As composite materials continue to be developed and incorpcrated into airframe structures,
needs have arisen for solving problems associated with their use. Composite structures can and will
prematurely fail due to gross manufacturing defects, design errors, or severe in-service damage. Needs
exist for a systematic compilation of failure analy is techniques, procedures, and supporting
fractographic data - in handbook form - that can be used by experienced laboratory personnel,
working in consultation with field investigators, to diagnose the cause for premature component failure
and to make recommendations for preventing similar failures.
The goal of this Composite Failure Analysis Handbook is to provide a guide for conducting
post-failure analysis of fiber-reinforced composite structures. Tt forms a compilation of the procedures,
techniques, and sample data required to conduct analyses of composite structures. Volume II of this
report is the Technical Handbook and it has been divided into three parts.
Part 1 of the Handbook is a summary of the procedures and techniques used to perform failure
analyses of composites. This part is largely comprised of the results of work performed by Boeing under
Air Force Contracts F33615-84-C-501(, and F33615-86-C-5071 and reported by Boeing in AFWAL
TR-86-4137 and WRDC-TR-89-4055. Northrop expanded this summary under Air Force Contract
F33615-87-C-5212 to inc!ude (1) procedures and techniques that have recently developed, (2) documen-
tation of manufacturing and processing defects that occur in graphite/epoxy, and (3) a compilation of
chemical and mechanical properties of current and near-term composite strucl~ral materials. The
sequence that a laboratory investigator should follow during failure analysis of a fractured composite
component (failure analysis logic network), and procedural guidelines for on-site handling and
examination of composites during accident investigations (field handling logic network) have also been
updated and expanded. Laboratory failure analysis techniques such as nondestructive testing, materials
characterization, stress analysis methods, anc the art of fractography are discussed.
Parts 2 and 3 (the Atlas of Fractographs and the Case Histories, respectively) of the Technical
Handbook are discussed in the Summary (Page iii). Additional information has been presented in
Introduction and Purpose sections cf each of these parts.
The Handbook has been designed to be a living document that. can be updated readily. This
work reports the results of six years of fundamental work that has been sponsored by the United States
0 Air Force (USAF) and the Federal Aviation Administration (FAA).
1-1
SECTION 2
SOURCES OF FAILURE
Composites can and will fail, due to service or manufacturing defects and material
inhomogeneities. In general, the sources of failure in composites can be broadly classified as:
1. Design errors
2. Materials and process discrepancies
3. Anomalous service conditions
Unlike homogeneous materials, most modern composite materials are fabricated by laminating
together a large number of relatively thin gage woven or unidirectional plies. For structures made from
such materials it is important to recognize that engineering properties, and hence the component's
ability to operate without failure, depend upon the correct nunt ber, sequence, and orientation of plies
being used to make up the laminate. In general, the significance of errors depends upon the particular
material in question, as well as the magnitude of mistake in terms of the overall laminate construction.
For example, an overall off-axis rotation of 5 degrees can reduce the ultimate compression strength of a
unidirectional graphite epoxy laminate by as much 54 percent. In contrast, a 5-degree rotation of only
one ply out of a 30-ply laminate probably would produce less than a 2-percent decrease in ultimate
strength. From this standpoint it is critical that the impact of identified discrepancies be identified
accurately and taken into consideration prior to being reported or to being identified as a significant
contributor to the cause of failure.
Errors in layup can also have significant effects not directly related to those engineering
properties considered as part of typical design. The materials coefficient of thermal expansion
represents one such property, where variations in stacking sequence can produce significant amounts of
panel warpage or internal residual stresses. These internal stresses have been found to cause damage
such as gross delamination and matrix cracking within the laminate plies. Such damage, while not
always directly responsible for failure, may in many cases constitute one of several contributory
conditions resulting in premature failure.
2-1
component examined. However, as a general guideline, critical parameters which should be given
consideration in the event disciepancies are detc'ied should include:
1. Young's Modulus
2. Basic lamiiate strength (tension, compression, shear)
3. Notch sensitivity (iension and compression)
4. Buckling stability
5. Intenal thern-al stress or residual stress conditions
6. Alterations in environmental susceptibility.
2-2
examinations should identify any gross deficiencies that could have significant effect on material
properties or the magnitude of local stresses within the part.
Composite components can fail when operated under anomalous service conditions. Any variable
that significantly affects ., mechanical response of the laminate or the chemical properties of the
composite constituents can bt xpected to lead to premature failure. Some of the environmental variables
that can be considered deleterious to the service life of composite components include:
1. Temperature: Operation of composite components at temperatures above those for which the
component was designed
2. Chemical Environments: Exposure of composite components to aggressive chemical
environments
3. Abnormal Load Situations: Subjecting the component to abnormal loads for which the
component was not designed, such as low energy-impact damage.
2.4 EXAMPLES
This subsection presents examples of manufacturing, materials, and processing defects that
occur in Graphite-Epo.Y (Gr!Ep). These defects have been previously identified in a study termed the
Rockwell Flaw Criticality Study (Reference 1) as defects that could affect the service life of composite
structural components. These defects and flaws can occur either during fabrication/manufacture of the
composite component, or duiing subsequent service. A list of these defects is provided in Table 2-1. The
corresponding Rockwell study defect number and the stage at which these occur are also given.
Figures 2-1 through 2-28 are photogTaphs illustrating the defects.
2-3
Table 2-1. Rockwell Flaw Criticality Study Defects
2-4
Figure 2-71. Rockwell Fldw Criticality Study Detect 7: External Delamination
Figure 2-2. Rockwell Flaw Ctiticality Study Defect 2: Blister, nternal Delamination
2--5
Figure 2-3, Rockwell Flaw Criticality Study Defect 3: Oversized Hole
Figure 2-4. Rockwell Flaw Criticality Study Defect 4: Hole Exit Side Broken Fibers Breakout
2-6
0
Figure 2-6. Rockwell Flaw Criticality Study Defect 7: Resin-Starved Laminate (Bare Fibers)
2-7
Figure 2-7, Rockwell Flaw Criticality Study Defect 8.: Resin-Rich Area
at Arrow (Dark Areas Are Voids)
Fig ure 2-8. Rockwell Flaw Criticality Study Defect 9: Excessive Porosity, Voids
Figure 2-9. Rockwell Flaw Criticality Study Defect 10: Scratch, Handling Damage
Figure 2- 10. Rockwell Flaw Criticality Study Defect 11: Dent, Handling Damage
2-9
Figure 2- 11. Roc.A.well Flaw Criticality Study Dflecnt 12: Fiber Breakawaly From Impact Surface
Figure 2-12. Rockwell Flaw Criticality Study Defect 13. Splintering, Edge Delamination
2-10
0
m'p CRACKS
Figure2-13. Rockwell Flaw CriticalityStudy Defect 14: OvertorquedFastener
0
2-11
Figure2-15. Rockwell Flaw CriticalityStudy Defect 7: CornerNotch
2..
.__-12 ." |. -
Figure 2-16. Rockwell Flaw Criticality Study Defect 18. Mislocated Hole - Not Repaired
n1
4-__. -
Figure 2-17 RockwellFlaw Critics/lwt Study Defects 20 and 27: Marcelled Fibers (Arrow) and
Wrinkles
m .. .. - - . I=
. ---I
Figure 2-178. Rockwell Flaw Criticality Study Defect 26: Wrong Material, Correct Layup /0IndT
2-13
S-~14' 14
A 'NN
2!d
"' V I .-
O f 70
4k Iw + 0 0 I +; 1
P 17 ~c%
2-1
Figure 2-27. Rockwel(l
raw UOderlap, a
Criticality Study Defect 29: Ply
2-15
Figure 2-22. Rlockwell Flaw Ctiticality Study Defect 31: Improper Fastener Seating
2-16
2.24. Rockwefl Flaw Critic-alityStudy Defect 35: Off-Axis DrilledHole
-ire
Figure 2-25. Rockwell Flaw CriticalityStudy Defect 36: Countersinkon Wrong Side of Laminate
2-17
Figure2-26. Rockwell Flaw CriticalityStudy Defect 39: Burned DrilledHoie
2-18
Figure 2-28 Rockwell Flaw Criticality Study Defect 42- Dent, Hidden Fiber Breakage
2-19
0
SECTION 3
This section reviews the failure analysis logic network (FALN) flowcharts and supporting data
tables developed to provide investigators with guidelines delineating a logical sequence of investigative
operations. These flowcharts are designed to identify the analytical tools necessary for composite failure
analysis. The charts, when combined with decision gates, allow an accurate, cost-effective, and timely
determination of the cause of component failure. Such guidelines have not been available previously due
to the relative complexity of composite materials and their fairly recent use in industry, particularly for
primary structures. In developing these guidelines, specific objectives were incorporated to provide
investigators with a logical sequence that:
1. Considers a variety of potential causes including design or fabrication enrors, anomalies, and
in-service or testing damage
In addition to the general FALN, several specific FALNs are provided for the areas of
nondestructive testing, materials characterizaticn, fractography, and stress analysis. Both types of
FALNs are described in the following paragraphs.
Identifying the logical sequence of steps for doing a post-failure analysis is often a complex and
difficult process. Sufficient information must be gathered and evaluated so that the cause of fracture
may be determined from positive supportive evidence rather than simply by a process of elimination. In
many instances, the development of a coherent set of positive evidence is complex since: (1) numerous
potential causes exist and (2) multiple contributory factors may be involved. In order to identify
accurately the cause of fracture, each of these potential contributory factors must be taken into
consideration. Identifying the cause of fracture and related contributory factors without examining
every conceivable cause requires an organized investigation sequence. Developing this plan of attack is
further cc ..plhcated by the fact that many investigative steps may be destructive to remaining evidence,
and thus may preclude further critical analysis. Consequently, adequate documentation to record
existing evidence, as well as the logical flow of information from one analysis to another, must be
considered.
3-1
To assist investigators in developing a logical plan of attack, an orgonizational framework wAs
developed. This framework is based upon well established procedures utilized in the failure analysis of
metallic structures, modified to meet the specific analytical requirements for fiber reinforced composite
structures. This framework considers each major failure category, potential interrelationships, and the
prevention of premature destruction of evidence. Because composite materials differ in many respects
from metals, the specific operations involved were modified to address those characteristics specific to
composites. This proven framework consists of five basic investigative operations arranged around
intermediate decision points. The approach is aimed at simplifying and streamlining -he number and
complexity of analytical steps involved, usually lowering the overall costs of investigations. The five
major investigative operations are:
An overall diagram, showing the simplified application sequence of these five operations, is
illustrated in Figure 3-1. The sequence initially encourages the use of simple, inexpensive procedures,
such as nondestructive examinations, followed by progressively more detailed procedures such as
material verification and stress analysis. Through comprehensive hands-on application and evaluations,
a siff-iificant amount of detail has been added to the origiinal inve~tig-dtive fi-aniewurk. The expanded
vemion, shown in figure 3-2, establishes a more detailed and accurate path for investigators. It
delineates most of the widely used techniques that may be used and the required decisions involved in
carryipg out a postmortem analysis.
Each initial examination is directed toward identifying items of significance early in the
investigation. Through an iterative process, the number of steps can be minimized and future efforts
concentrated on items of interest. During the initial stages of investigation, background information on
material, fabrication, design, loads, environment, and service or test history is collected and reviewed with
the intention of identifying areas of concern. This process helps develop a familiarity with the component,
its operation, and its service environment. Nondestructive inspections are then performed to identify and
further delineate the extent and nature of nonvisible fracture or damage. The data are documented for later
reference. This operation establishes the groundwork to plan more detailed examinations and helps in
selection of specimens that may require destructive sectioning. After nondestructive examination, the part
is evaluated for confor-mity with engineering, material, and process specifications. This includes such items
as material verification, ply layup, and degree of cure.
Detailed fractographic and stress examinations are the next analytical steps. These operations iden-
tify more specific details and assess their significance. Typically, fractographic examinations are used princi-
pally to identify the origin an-. load conditions involved in failure. In many cases, the main benefit of fracture
examination is the identification of material defects oraanomalies. As such factor.s are identified, sufficient
information may be developed to identify either a specific cause for failure or a point of interest (that is, an
origin) for further analyses. After inputs from fractographic analysis have been developed, strcss analysis
may be performed to evaluate stress states, out-of-compliance conditions, or the critical nature of identified
3-2
defects. In cases where questions may remain, additional specialized tests or further, more detailed, stress
analyses may be required from the stress specialist to model previously indeterminate conditions.
FIELD
L INVESTIGAYION
SELECTION
OFSPECIMENS
FOR LAB
iNVESTI-
FRACTURED SPECIMEN
SUBMITTED TO
LABORATORY
IS HISTORY AS NO
YEISVPSTIDATIONOOR
UO
NANNANNUEC ECESAR
TESTS AS
*EXORIIIN
DEVLO SPCS
NONDEETS
DEFIN UTV
I
SI
EVRNMENT
*XA
* DAMAGE
LOVCAT=
2(.TYE1
~~ON
ON.RERE
TAP PAKGIGO
_ED-
______ m~uID
COMPOND~rS
FiurSIM . Detailed
0CAETIONvetgtv rmwr Pg f2
I
ICIN TGSA-BE. ON 3-4I
S~' ISa slatstt~l
04
Fiue3- DtaldInvsigtv Frmwrk(ae 0f2
40-
SECTION 4
FIELD GUIDELINES
This section describes the procedural guidelines that have been developed for field investigators
to use for handling and selecting fractured components for subsequent laboratory investigations. These
guidelines have beer developed based on inputs and suggestions from several DOD agencies as well as
representatives from the National Transportation Safety Board (NTSB) and Federal Aviation
Administration (FAA).
Figure 4-1 shows the field handling logic network which defines the recommended safety steps
and associated handling, packaging and shipping procedures that are being suggested to field
investigators. Failed parts that have not been damaged by fire proceed to laboratory failure analysis.
The failure analysis report ultimately goes to the Accident Investigation Board. A description of several
of the issues used in this network follows.
4. 4 UAVA~D
Alih6'11%tl~
4 AVIONeICS HPU.ARD$ ..
Fibers may be released during airplane accidents which can be a health hazard. Primary release
occurs during post-crash aircraft fires, although some release may occur during an on-site crash
investigation, during transportation of components to a safe-area, or during scrap disposal. Graphite
fibers act as skin, eye, and lung irritants in a manner similar to exposure to moderate/heavy amounts of
glass fibers. Except for skin irritation, no evidence exists that any serious effects could result from
graphite fibers being imbedded in human skin.
With regard to other health threats associated with carbon fibers, the National Institute for
Occupational Safety and Health (NIOSH) has determined that fiber particles, including carbon fibers,
fiberglass, and asbestos, will not cause malignant disease if they exceed 3.5 microns. The average size of
carbon fibers currently in use in composite structures ranges between 4 and 7 microns. Based on these
observations, it is generally believed that carbon fiber-based composite panels fractured in aircraft
crashes but not subjected to fire are biologically benign and would constitute no more hazard than
fractured aluminum aircraft pieces.
4-1
Identification of_________
Fracture Locations L-P-378 Plastic Sheet
(VisualiNOE) Fieand MIL-T-?2085,
No ireTypelItTape
Fackaging of
_____________ Non -Fire- Damaged
Components
Sectining On-Site W~reckage Documentation
(Crbid~e Saw)
4
Tags/Labes on Debris Z________
MIL-C-16 173,
Tansportation of Debris Grade 4 Fixant and
Cleaning of
Fracture4
F rMIL-B-131, Class 3
Barrier Packaging
Classificatior of Parts Material and Tape
"*NDE1
"*Macro/SEM
*Chemical Methods
" Stess Ar.alysis
hSelection* :
Fractured
off________
Board
Ba'.kground Investigation
Information
"*Seiice Record
"*Drawings Report and
"*SpecificationsSpcmn
T90-3118vA
4-2
3. For aircraft mishaps where carbon fibers are released by fire:
a. Firefighters and rescue personnel should be the only personnel in the immediate vicinity of
the burning/smoking wreckage.
b. Personnel should be prevented from approaching the cr h site and should be restricted
from assembling down-wind of the fire at the crash site.
c. Once the fire is completely out and the wreckage has cooled, all carbon fiber should be
sprayed down with a fixant to contain the release of carbon fibers.
d. The area should be roped offas soon as possible and a single entry/exit point be established
to the mishap site.
4. For aircraft mishaps where no carbon fibers are released by fire:
a. Leather palmed gloves should be worn.
b. Carbon fiber panels can be handled similarly to handling of aluminum panels.
5. All personnel involved with crash/fire-damaged composite parts should be provided with a
suitable shower facility before going off duty to preclude injury from loose fibers.
3. Safety officers should ensure that the following items are readily available at all operating sites
or included in premishap kits:
0
4-3
4.5 ON-SITE CRASH/WRECKAGE RECONSTRUCTION AND HANDLING
1. Once F~xant has contained carbon fiber material, the use of industrial dust masks and gloves is
considered sufficient for work around the crash site where large amounts of carbon fiber mate-
rial are not being stirred up.
2. Complete documentation of all debris at the crash site should be carried out using conventional
photographic equipment. Aerial photography of the crash site to include a "global" perspective
of the cr'ash investigation is also useful.
3. Inspection of the crash-damaged aircraft to identify crash/fire-damaged composite parts
should be carried out. Classification of all composite components into fire-damaged and
non-fire-damaged is also useful. Tagging and labeling of all debris should be carried out,
preferably at the crash site or prior to transport to an accident reconstruction area.
4. Sectioning of craslh!iire-damaged components for further engineering investigation should be
performed using carbide saws and mono-goggle protection. Sectioning should be performed in
areas well away from visible fracture and areas that contain internal damage as determined by
nondestructive tests (coin-tap, portable ultrasonic, X-ray, etc).
5. Crash/fire-damaged parts which require laboratory evaluation and/or repairshould have fibers
contained by wrapping the affected area with 0.006 inch thick plastic sheet (MIL Specification
L-P-378) and taped in place with aircraft preservation tape (MIL-T-22085, Type II tape).
6. Crash/fire-damaged parts which do not require evaluation as part of the accident investiga
and/or are to be scrapped should have fibers con ained (to ensure that fibers are immobilizeu_
by using Corrosion Preventative Compound, MIL-C-16173, Grade 4 spray applied as a fixant
material.
7. Transport of the wreckage to a "safe-area" for accident/wreckage reconstruction should be
carried out as soon as possible. During the accident investigation/repair disposition operations,
the crash-damaged aircraft should be in an enclosed area not subject to the elements of
weather. This prevents degradation of the tape/plastic sheet fiber containment system and
precludes the spread of loose fibers.
4-4
1 1. Composite material that is not required for investigation should be disposed of at an approved
hazardous material waste site.
As much as possible, cleaning should be avoided for subsequent laboratory examination of fracture
surfaces. However, under extreme requirements, conventional cleaning methods such as ultrasonic rinsing
in soap or alcohol used in metal fractography can usually be successfully used for fiber/resin composite
material without damaging fracture details.
Failures in composites can be described in terms of the failure mechanism exhibited on trans-
laminar, interlaminar, and intralaminar fracture types. The first evaluation method available to define and
differentiate between these fracture types is visual macroscopic. The ability to define fracture types at the
macroscopic level can often be the most valuable capability for many investigators, particularly for those
performing field investigations. When examining a failed composite structure, the investigator must assess
the nature and direction of the applied losd, identify the significance and time of fracture, and select
portions of the structure for laboratory analyses. Visual examination alone can often provide sufficient
information to answer these questions. Howwever, this extremely valuable capability is very much in its
infancy compared to the metals field. Figure 4-2 presents a brief overview of the relationships that various
investigators have observed between fracture mode/load conditions and macroscopic fracture surface
* features.
4-5
SECTION 5
In the broadest sense, nondestructive evaluation (NDE) includes any examinat" in that assesses
material integrity without damaging or destroying a component. It is used fGr a variety of tasks such as
in-process quality control, in situ test monitoring, and fleet service inspection. For failure analysis,
nondestructive evaluations are useful for identifying the conditions of invisible fracture for
documentation and for planning subsequent destructive evaluations.
Following preliminary visual and macroscopic analysis cf the failed component, invisible
damage can be identified and evaluated by various techniques outlined below. Figure 5-1 and Tables 5-1
and 5-2 present a summary of the various methods commonly used for failure analyses. Note that the
methods are listed in the order of preference for evaluating and defining each defect condition. The
NDE FALN is structured such that inspections involve progressively more detail of the damage
condition, including;
1. initial plan view inspections
2. Detailed plan view inspections
3. Through-thickness inspections.
It should be noted that many of the failure analyses investigations do not require detail beyond
the initial plan view inspections. Usually failure of the part denotes that some fairly extensive fracture,
often visible, has occurred. The main responsibility of NDE inspection is to define the damage region
around the primary fracture. More detailed analyses, if performed, are usually confined to other regions
on the component, away from the principal fracture region, so as to identify other sites of damage or
contributory defects.
Evaluation Plan
5I.1
NONV:SIBLE
DAMAGE
SUSPECTED 0
ISMORE PLAN NO
VIEW DETAIL
hNEEDED?
YES . O
NO
THPU-THICKNESS iDENTIFICATION
PLAN VIEW IDENTIFICATION IS TR-AND DOCUMENTATION OF
~AND
DOCUMENTATION OF + TIKES YS DMG
IDAMAGE 00 LOCATION o Pulse echo
9 F
idio-opaque penetrant NEEDEDI St ero -rdography
J Neutron radiography . Teomrdography
5-2
9 Table 5-1. FailureAnalysis Techniques - Nondestructive Evaluation
Thru- Measures ultrasonic sound Determines size and o Plan view documentaticn of failure
transmission attenuation location of nonvisible 0 Plan view assessment of part quality
ultrasonic (TTU) * C-Scan--plan view damage, defects. & Planning analyses
* C-Scan presentation fracture in plan view
Pulse ultrasonic Measures ultrasonic sound Determines size and * Plan view documentation of failure
"* B-Scan reflection location of damage. * 1 hru-thickness view documentation of
"* C-Scan * B-Scan-thru- defects, fracturein failure
transmission view both a plan and thru- a Plan view assessment of part quality
presentation thickness view * Thru-thickness assessment of part qudlity
0 C-Scan - plan view * Planning analyses
presentation
Reasonance bond Measures mechanical Determines size and * Determining of size and location of part
tester resonance changes caused location of nonvisible damage
"* Bondacscope by defects, meter or CRT damage e Method can be used when only one side is
2100 display accessible
"* Sondicator
"* Acoustic flaw
detector
"* MIA 3000
X-ray Measures X-ray 0 Determines size and o Plan-view documentation of failure
radiography attenuation plan view location of * Plan view assessment of part quality and
(tomography) presentation trans~amimar defects
fractures and radio- * Planning analyses
opaque defects -
plan view
presentation
* Delamnation size
and location
determined with
radio-opaque
penetrant
0 Thru-thickness
position determ ined
by stereo-
radiography or
X-ray tomography
5
5.3-
Cc LU I
u -o
w 5n (ci
(- Li
Ow
02za
Z(ci) < P)
0 -i
L1J
L 5
u L 0L
a:u.( D
< (6
a: (-)
I- L
cr L
) w
() d
Li i LU
Q o cow
<. <5 (5
< cc Lii (
LA.
Q) 0W<Li .4I
02 a- 0-eL -- e w_ 0
a:MM
z w 0 10 >
'2J 0J 0 C) T 0
U) X <cr d( -- 1->
< C5 -
-A >t.0-
-W Lii 0 Ld <w <
0< 0 CS :z
ccuiZ
to Id d - ~ x C
<: .: <U C(5 >i CC
0
q) < LU
CL 0U 0 Li L
z dNL w wI
I~Ll
LU - oI 2
C) w z z<0~
a: z 0- (z
)
-4
5LL: nz< I
* Initial Plan View Inspections
For preliminary inspections, the major emphasis is to determine the basic outline of the damage
regions such that part breakdown and sectioning can be performed without destruction of evidence or
to minimize the sectioning damage if a repair scheme is considered. Plan view analyses such as
ultrasonics, through transmission ultrasonic (TTU) or pulse echo, and radiography are by far the most
versatile and encompassing techniques for overall determination of the basic outline of the damage
region. Commonly, this coarse damage assessment is necessary for field inspection prior to part
breakdown, sectioning, and subsequent detailed NDE techniques performed in the laboratory. In the
field, pulse echo is the preferred method, particularly desirable in conditions where only one side of the
structure is accessible. When this initial inspection can be performed in the laboratory on a fairly flat
panel, the TTU C-scan method is by far the most preferable method due to its ability to provide a full
scale, plan view, hard copy record of the defect conditions that are aligned normal to the interrogating
beam (delaminations). Defects aligned parallel to the beam direction (translamir.ar cracking) do not
often create appreciable or detectable attenuation, and thus should be examined by X-ray methods.
When more plan view details are required, enhanced X-ray and neutron radiography should be
used. Each technique can identify both translaminar and delamination damage, although the X-ray
technique requires a free edge or surface-intersecting damage so that the penetrant can be introduced.
However, when such a surface defect is present, enhanced radiography is probably the single most
sensitive and accurate inspection technique for composite structures. Neutron radiography, on the other
hand, can be used where metal structure overlays composite material, since neutrons are not as
attenuated by metal as X-rays and are relatively sensitive to polymeric materials containing hydrogen.
This method, however; has not been proven to generate radiographs that exhibit much contrast or
resolution. Another available technique ;.3 eddy current, in which small translaminar cracks can be
identified without the requirement of a free edge.
Through-Thickness Identification
0 with some fundamental theory with regard to evaluating fractured composite structures.
5-5
5.1 ULTRASONIC METHODS
In the TTU method, an ultrasonic transducer is placed on one side of the material and emits an
acoustic pulse. The pulse travels through the material and is received by a second transducer located on
the other side of the material. These transducers are placed in axial alignment so that their common
axis is perpendicular to the surfice of the specimen. With this placement, the amount of energy
transmitted through the material is maximized and can be monitored easily as a function of position
when the material is scanned by the transducers. For a C-scan, the entire surface is inspected by
moving the transducer in a series of closely spaced traverses with a mechanical system. Most
mechanical systems only allow planar scanning motions of flat or circular symmetric shapes. Water jet
techniques have been developed which allow inspection of parts too large to be immersed in a tank.
Current technology exists for robot controlled manipulators which track complex surface geometries,
5-6
* but such systems are not yet commonplace. Since the speed of the test is limited primarily by the speed
of the scanning, several arrayed transducers are often used for large scale inspection tasks to reduce
scadn times; however, for most failure analyses a single trapsducer can inspect a large part in a few
hours.
The transmitted sound can then be evaluated and broken down into several sublevels, or grey
scales, with each lev!3l equivalent to a certain amount of attenuation. This energy loss ca-n be related to
either voltage or decibels (dB). Each of these sublevels ca-n then be assigned numbers or colors and
graphically presented as a plan view of the part. Regions of attenuation greater than a standard, such as
in the 6 to 18 dB range, indicate the presence of significant damage conditions that reflect the energy of
the sound beam. Through the use of real-time computer monitoring of the attenuation, a map of the
sound transmission relative to the part geomL try can be produced by a plotter in which an image is
formed by burning the surface layers of an ink-impregnated conductive paper. Figure 5-2 presents such
a map, with a delamnination identified in the area of high attenuation. The numbers denote a range of
dB sound loss, -with the larger and darker numbers indicating more attenuation.
nA..L
a... ,aa .a...,.:.,..aI~t~ .ttt.....
. ......
a .J
* .t..,.a..~a,..-.aaI................
*..........a......
In 'I*TU, the sound attenuation results from three sources: viscoelastic effects, geometric
dispersion due to material anisotropy, and dispersion due to geometric internal damage. By proper
se~ection of the sound frequency, the attenuation due to delamination and cracks can be maximize] and
the attenuation due to material viscoelasticity and heterogeneity can be minimized. The use of 1 MI-z
has been recognized to provide the best transmittance since. it has a fairly long wavelength and thus ia.
less susceptible to scattering from smaller structural details, particularly for parts with honeycomb
core. When increased sensitivity for smaller details is desired, the use of smaller wavelengths should be
5-7
used; however, frequencies in this range (5 to 15 MHz) do not transmit through honeycomb structure
and require a more critical alignment of the two transducers.
The transporting medium is usually water, which provides a uniform coupler to transmit the
sound waves between the transducers and the specimen. This requirement for a water coupler basically
limits the 'PTU inspections to the laboratory, but a few portable units are available. The specimen is
either immersed or water jets at the transducers supply a stream of the coupling agent. With composite
failures, surface damage in the form of edge delaminations or translaminar cracking is often present.
This surface damage, if extensive, can allow water to penetrate into the cracks. Since the attenuation is
commonly due to air at the crack impeding the transmittance of sound, the water penetration can
displace the air and eliminate, or significantly reduce, the attenuation at these defects. Hence, special
precautions are required to prevent the intrusion of water into these areas, particularly for those
spezimens where a contaminant is suspected and water would be very undesirable. Normally, the open
surface cracks are edge sealed with adhesive tape to inhibit entry of water.
Inspection of the TTU C-scan plan view records can provide a. full scale assessment of the major
defect conditions. Usually this inspection method is adequate to define the general outline of the
delamination, particularly surrounding the major damage region that is visible. Although the C-scan
method is best used to define delaminations, much smaller defects such as porosity can also be
identified in extreme cases. This technique is limited by four factors: (1) both sides of the material must
be accessible; (2) the depth of defects within the laminate cannot be determined; (3) extreme variations
in thickness cannot be evaluated at the same time; and (4) defects aligned parallel to the incident beam
are not easily identifiable ,such as translaminar cracking). Where a more detailed inspection of the
damage is needed or when there is access to only one side of the part, or when laboratory analysis is not
possible, the pulse-echo method should be employed.
In the pulse-echo method, a single iransducer transmits and receives the acoustic pulse. The
transducer emits a gated pulse through the material, which is reflected by the far side of the part and
then is detected by the transducer again. Since the full dynamic range of the receiver is available to
amplify any backscattered acoustic energy, this technique can be made quite sensitive to subtle defect
conditions. The reflections from the front and back surface provide known time-related endpoints sr,
that the depth of the defect can be determined by its time function. A potential disadvantage of this
method is that flaws one ply away from the front or back surface can be hidden by the reflections from
the surfaces. This problem can be alleviated by properly adjusting the instrument to distinguish
between these reflections, in combination with using a de!ay line transducer. Additionally, it is necessary
to record the returned echo trace and section it at variots periods of time in order to have an a- curate
representation of the location of the flaws. Breaking down 'he echo trace allows the differentiation and
separation between closely arranged flaws and prevents the investigator from mistaking several small
flaws as a single large one.
For use in the C-scan for1mat, the inspection is usually performed either in a water bath or by
using columns of water sprayed upon the surface of the specimen. The water serves as the coupler and
delay line for the ultrasonic signals. Information is generally recorded in which the signal levels at each
(and time or depth) are printed or displayed as the transducer is moved over the specimen. The C-scan
5-8
pulse echo is a plan view, two-dimensional image of the internal structure of a material. With gating of
the amplitude-based digital signal, imaging of defects can be identified, related to the position within
the thickness of the material, as shown in Figure 5-3. The use of a combination of two gating zones can
allow the differentiation of delaminations near the front surface (light) and the back surface (dark).
surface
Start pulse Front echo
surface
echo
Back surface
gated
I I
L- J
r
I !
gated region
center
Plan view
Figure 5-3. Pulse-Echo UltrasonicC-Scan Ush g Time-Domain Gating Zones To Iderntify Damage States
of Impact
Short pulse (shock wave) methods can also produce a considerable amount of information,
although it is limited to linear plotting of the data instead of the two-dimensional C-scan. In A-scan, the
reflected pulses can be real-time displayed on a cathode ray tube or can be permanently recorded. By
comparing the reflected pulse information from a region of damage with an area containing no damag(-
(often a calibration sample), the depth of the defect condition can be fairly accurately determined.
Similar to C-scan, this method produces an image delineating the reflections between the front and
back surfaces. By taking several parallel passes of the transducer over the part, a better feeling of the
three-dimensional geometry of the defect can be obtained, as shown in Figure 5-4.
While the A-scan provides data regarding all reflections through the thickness, the B-scan
indicates only the first echo after entering the surface. It is therefore incapable of displaying second and
higher multiple reflections, as the other two methods can. The B-scan is somewhat similar to a
transverse cross-section, in that it provides a record of the depth location on a line across the specimen
surface. By taking several of these scans, they also can be combined to provide a more complete view of
the through-thickness damage in the material. An example of a B-scan is presented in Figure 5-5,
illustrating the extent of damage in a laminate at several locations along the surface of the part. For
5-9
Delamination In
center plies
Part
A__
A-- Scan line
9 9
Front Back
surface surface
field or initial inspections, the A-scan method using a cathode ray tube for imaging can be performed
without a permanent record of the damage. For instance, in early NDE investigations in large failed
structures with visible damage and delamination, hand-held pulse echo can be used to define quickly the
perimeter of the delamination damage. In this case the couplant is usually a light grease, oil, or
commercially available gel.
5-10
Specimen
Sound beam-
incident angle !o / /
-Transducer
B-scan 4[
SimulatdLriae -,
T defect
Sample
P L__J thick~ness
Delamination
Figure 5-5. Pulse-Echo Ultrasonic B-Scan Figure 5-6. Ultrasonic Detection of Simulated
of a Delamination Region Crack in Laminate Plate
by the transmitter, and is continuously monitored by the receiving transducer. With this system,
delaminations create bath a phase and amplitude change, allowing their detection. Because the
ultrasonic signal is angled with respect to the surface, this technique has enhanced sensitivity to cracks
oriented in the through-the-thickness. direction. A drawback is that each ply must be scanned at an
* specific angle of incidence to maximize the signal amplitude and properly determine the location of the
microcracks. A reference standard is also required that duplicates the materials and layup of the
laminate being inspected.
X-ray techniques can be used to detect broken structure and large subsurface fractures when
there is a displacement of members. Extensive cracks in which surface displacement has not occurred
may not be detectable. Defects such as cr'.shed core and fractures associated with impact damage lend
themselves to radiographic evaluation. In addition to detecting cracks, th., investigator can learn
valuable information regarding the internal structure of the component, particularly if there are major
differences in construction such as stiffeners and honeycomb core. Water in the core, if extensive, can
also be detected. For failure analysis, the use of X-ray is most often considered as a secondary method
to ultrasonics, although it possesses many advanitages when faced with damage located parallel to the
incident interrogating beam such as translaminar cracking.
5-11
form a latent image. The film is processed to form a visible image called a radiograph. The radiograph is
then evaluated for information regarding the extent and nature of the defect conditions, if they exist.
Although conventional radiography readily detects through-thickness fractures, it does not
-always present conclusive results. The information obtained from radiographic inspections are affected
by the size and ;:rientation of the defect relative to the incident beam. Defects presented normal to the
beam result in insufficient changes in density so that interlamin-ir defects such as delaminations and
porosity are not detected by conventional radiography. Due to these limitations, the technique has been
significanw.iy improved by the use of some form of radio-opaque penetrant, which produces probably the
single most sensitive inspection technique for detecting cracks in composites that are surface related.
The foiowing pa .-agraphs focus on this valuable technique.
The procedures for making radiographs of defects and damage in composites differs from the
conventional ones in that an X-ray opaque penetrant is used to enhance the damage. The penetrant
proviesignificant improvement in contrast between the damage and the intact composite, as shown
iih Figire 5-7. Various penet"ants have been used, with tetrabromoethane (TBE) being the first solution
evaluate". This penetrant was found to be highly toxic and carcinogenic so its use was discontinued.
Diiudcbiktne (DIB) was used for a sl...t time, but was also found to be a toxic organic halide, was
expensive, and had a short shelf life. The enhancement chemical that has proven nontoxic and was
relatively inexnens:ve is zinc iodide (ZnI,'), used with an alcohol carrier solution. The carrier solutions
are as follows:
Isorropyl Solution MEK Solution
Zinc Iodide - - 60 gnams Zinc ludide - 60 grams
Water - 10 milliliters (ml) MEK - 250 ml
Isopropyl Aicnhol - 10 ml
Kodak "Photoflo" - i mi
The isopropyl solution requires about 30 minutes after application for the penetrant to reach
the end of the damage. This time is ;educed o about ten minutes for the methyl ethyl ketone (MEK)
solution. After saturation, excess penetrant should be removed from the open surfaces with absorbent
material. These penetrant materials should be used with caution since they can potentially damage or
obscure the fine microstructura- frpctilre details and prevent detection of contaminants. Por fairly
chemicaliy stable matrix systems such as epoxies, the use of these renetrants has not created any
undesirable effects such as damage t3 the fractu: surface dptails, particularly when removed with a
clean solution of the primary solvent carrier. For othier material systems such as thermoplastics, the
effects have not been evaluated, and thus sriou!d be spct tested prioi to applicatiop to the fracture
surfaces.
Various X-ray films -tanl~e us2i, although a high rtsolution, single-coat film such as Kodak Type
R industrial, low speed, fine grain film gives the best contrast and resolutioi. Doubk-'.oat film should
be avoided since it is exposed on both emulsions, leading to a doubie image and loss of resolution.
5-12
Figure 5-7. Penetrant-Enhanced X-Ray Image of Edge Delamination
X-ray units emitting soft X-rays are recommended, with a small spot size in the range of 1.5
mm by 1.5 mm or less, with an inherent filtration of 1.0 mm beryllium equivalent or less. The generator
tube should be capable of producing a minimum of 20 kV at 2 mA. The low operating voltages produce
soft X-rays that provide resolution of structural details within the laminate, such as porosity and fiber
spacing irregularities. For applications which require positioning the X-ray tubes at tight locations, the
O use of an end anode side emission is recommended. The optimum exposure times are those that
produce high resolution negatives from which prints can be made. 'Inese exposure times are shorter
than those used for direct viewing. It should be noted that it is difficult to obtain prints of radiographs
that w'ill reproduce adequately by normal printing methods and therefore reporting an.i presentation of
the results are more difficult and somewhat lacking.
Inspection of the films often requires an expert to differentiate artifacts from actual (lamage.
The interpretation requires an understanding of how the penetrant affects the X-ray beam anml how it.
enters the damaged specimen. Regions containing penetrant appear darker than regions containing no
(damage or defects. Regions containing no penetrant have a uniform grayness, especially on structur-es
that have small thickness variations, since comp~osites have relatively low radiographic scattering jr
absorbence clue to the elements present. Cracks such as matrix or translaminar ,:racking appear as
long, narrow, dark lines. The interpretation of artifacts corresponding to (delaminatiens is usually more
difficult. The opening dlisplacement of the crack is the greatest at the edge and the least. at hte end of
the delamination, andl therefore one might expect a visible lightening of the image from the free tdge
towardl the crack tip. While this change in grey level hohlds true at the extreme ends of the crack, the
situation at the central region is such that the capillary forces are not strong enough to hohl the
penetrant This condition results in a central boundary region that (does not stop the X-rays andl lten
appears light and undamaged.
5-13 [
(usually about 7 to 15 degrces). The part is then allowed to remain at the center of the path of the
X-rays and is also centered in the radiograph. The depth of the damage can then be identified, and
overlaying damage that might be masked with a single plan view can be differentiated. With the aid of a
stereo viewer, the defects nearest the X-ray source have the largest relative displacement and the
farthest defects are the least displaced.
This techniqte is commonl- used in metals and has provided satisfactory results for fabric
laminates, particularly for locations around small, localized geometric variances such as fastener holes
and edge radii. Ed~y current testing involves small, hand-held surface probes that produce an
alternating magnetic field. This field is generated by an alternating current test instrument coil. This
alternating expanding and collapsing current induces a magnetic eddy current in the specimen. The
interaction of this magnetic field with the test instrument varies as the internal flaws and fractures are
encountered. The use of -his nstrument is basically limited to solid laminates that are conductive and
have appreciable magnetic permeability. This method relies on the conductivity of the carbon fiber
which is, at best, limited. Fioure 5-8 presents a typical inspection procedure using a probe to detect
subsurface damage in fabric lamit.ates.
Edge replication has proven itself to be an accurate technique for documenting the state of
damage in thin laminate sections. It is a dircet application of the replication technique used for
transmission electron microscopy (TEM) specimen preparation. An acetate film that has been softened
with acetate solution is firmly applied to the edge, then allowed to dry. The replica can then be
shadowed to enhance the surface features. Tlv result is a mirror image of the edge that can be
examined at higher magnification to assess invisible damage. Cracks such as the translarniinar cracks
for the 90-degree plies and edge delarninations in the 0-degree plies can be readily identified and
highlighted by shadowing.
5-14
'04.
Figure 5-8. Typical Inspection Procedure Using Probe to Detect Subsurface Damage
in Fabric Laminates
-15
0
SECTION 6
MATERIALS CHARACTERIZATION
6-1
Most of these conditions are likely to be rare. However, since they require little effort to
examine and have such a critical influence on failure, their inclusion in any complete analysis is a 0
necessity. Configuration and materials characterization examinations are included in the overall FALN
illustrated in Figure 6-1. Because of their potential impact on subsequent operations, these
examinations have been incorporated in the early stages of analysis. In many cases, identification of a
major discrepancy (such as using the wrong material), may sufficiently explain the cause of fracture,
thus expediting and lowering the cost of the overall investigation, in addition to providing the rationale
for appropriate corrective actions.
FRACTURED SPECIMEN
DNO
tr
w
\ ~EXPLA!N
I FAILUIRE/
NONDESTRUCTIVE CO
EXAMINATION I O
+ TESESTS ESSA RY
MATE RIALNCONFIGURATI (IF REQUIRED)
VERIFICANTION
REUR TO EARLIER
YESAS OFTANALYSIRY
NO ./ ADbu'TE ES.,--.-ICAUSEOFRATE
<ENVRONTMEEN
C! STAGE ANALYSI
AF
S IF ACDEQTATIUEL
" CONTAMWNAN IS
" DEFECTS
" DAMAGE
6-2
In support of the overa!l FALN, a detailed flow chart for configuration and materials
characterization was developed. This chart, illustrated in Figure 6-2, provides a detailed guideline for
investigators to evaluate critical items such as configuration, ply layup, degree of cure, and the type of
resin system used. This FALN employs relatively simple analytical techniques. More complex and difficult
analyses occur at later stages of investigation. Table 6-1 provides a detailed matrix summary of material
characterization techniques.
CHEC'MI D.
ESS1CANT IDAY T s N
DR ETE ST tg E
XPlE SAN
OIMIN$ION5
(ONFIGUATION.
FTC YES
M($UT
R I6A$O
TRANS,710P
TEMPEftATuCENE OCRDI~uA NL'.
"* I U
*DMA hDA
NO
in Figure 6-2, one of the more detailed areas of a configuration and materials characterization
study is illustrated. Once a thermoset resin material has cured, ascertaining its identity (such as vendor or
specific resin formulation family) poses a relatively difficult problem. This is because of the generally
similar chemical structure of cross-linked epoxy systems and the tendency of cured resins to resistyielding
their identities by typical chemical analyses methods. However, since many of the currently used epoxies
exhibit nearly identical properties (within a cure-temperature family), identification of the exact resin
system used may not be as crucial as identifying a material with its proper cure-temperature family, for
example 121C (250"F) or 177"C (350'F). Through the combined use of Tg and residual heat of reaction
measurements, the particular cure-temperature family of most materials can be established with relative
confidence.
For example, when fully cured, most epoxies lack any appreciable residual heat of reaction and
the Tg's are at or slightly above their original cure temperature. When minimum heat of reaction is
observed, the measured Tg gives a fairly good indication of the cure temperature family of material used.
However, in cases where significant heat of meaction still exists, and the measured Tg is well below the
specified cure temperature, the indication is that the proper prepreg material was used but an
6-3
Table 61. Failure Analysis Techniques for Materiails C 1laracterization
Ith
SX-ray fluorescence Measures X-ray Determines sulfur convent Hardener content
-g fluorescence spectra
Th r on ca ia e srsm trilD tr ie ls rniin d niyg n rlrsnsse
U Analysis ermal-mnechanical
(TMA) I es,,onse
inisufficient. degree of cure was generated during processing. In waking these analyses, it is Aimportant to
keep in mind that undercure can arise from several causes -- only one of which may be an improijer cure
procedure. Alterations in the prepreg formulation can reduce reaction rates (luring cure. Therefore, when
a sufficient degree of upidercure is detected, it is generally recommended that add~itional tests be carricd
out on the original prepreg (from that specific lot, if possible) to verify its chemistry.
In some analyses, ;dentifying the specific resin system used may be important; however, the inves-
tig-ator should be aware that analytical capabilities are only partially developed, and are. in their infancy for
mare recent resin formulations and systems. Sufficient differences exist in basic formulet ions so that
relatively simple tests may be used to identify dlifferent systems, but for the most pait, the ability to
identify resin systems within the same emperatuz-e-cure family is relatively difficult.
6-4
O 6.1 LAMINATE LAYUP AND PLY ORIENTATION ANALYSES
Unlike homogeneous materials, most modern composite materials are fabricated by laminating
together a large number of relatively thin gage woven or unidirectional plies. For structures made from
such materials it is important to recognize that engineering properties, and hence the component's ability
to operate without failure, depend upon the correct number, sequence, and orientation of plies being used
to make up the laminate. In general, the significance of errors depends upon the particular material in
question as well as the magnitude of mistake in terms of the overall laminate construction. For example, an
overall off-axis rotation of 5 degrees can reduce the ultimate compression strength of a unidirectional
graphite epoxy laminate by as much 54 percent. In contrast, a 5-degree rotation of only one p~y out of a
30-ply laminate probably would produce less than a 2-percent decrease in ultimate st iength. From this
standpoint it is critical that the impact of identified discrepancies be identified accurately and taken into
consideration prior to their reporting or possible identification as a significant contributor to the cause of
failure.
Errors in layup can also have significant effects not directly related to those engineering proper-
ties considered as part of typical design. The materials' coefficient of thermal expansion represents one
such property, where variations in stacking sequence can produce significant amounts of panel warpage or
internal residual stresses. These internal stresses h. ve been found to cause damage such as gross
delamination and matrix cracking within the laminate plies. Such damage, while not always directly
responsible for failure, may in many cases constitute one of several contributoi-y conditions resulting in
premature failure.
For continuous fiber reinforced composite materials, the exact design properties which should be
checked as part of a ply layup analysis will depend strongly upon the requirements of the specific component
examined. However, as a general guideline, critical parameters which should be given consideration in the
event discrepancies are detected should include:
I. Young's Modulus
2. Basic laminate strength (tension, compression, shear)
3. Notch sensitivity (tension and compression)
4. Buckling stability
5. Internal thermal stress or residual stress conditions
6. Alterations in environmental susceptibility.
Some of the most common errors in ply layup include:
1. Missing or additional plies
2. Improper angular orientation
3. Improper ply type, grade, or style.
Several relatively straightforward methods exist for analyzing the orientation, stacking
sequence, and number of plies making up a composite structure. Optical nmtallography is the most
obvious and direct technique. Adapted from the microstructural analysis of netuds, this technique
involves the optical examination of finely j,olished cross sections using a standard reflected-light optical
microscope. Other methods, although not as quick or simple, includ,, image analysis anl radiography
Specific details of each of these methods are presented in the following iparagraphs.
6-5
Similar to standard metalluigical failure analysis procedures, selection of areas for examination
generally constitutes the critical step involved in understanding internal material characteristics. For most
failures, detailed examinations should be carried out in all areas of the part representative of the typical
structure. For example, in a skin panel with attached stringers, a thorough analysis should examine both
the skin as well as the stringer constructions regardless of whether the failure occurred in only one of these
primary components, such as the skin. Reductions in modulus due to a layup error of one component can
often produca higher loads in an adjacent part of the panel resulting in premature failure at a location not
related to the original error. Generally, as a final step, metallographically prepared cross-sections should
also be taken through, or adjacent to, the origin area of failure to examine for local discrepancies and
ensure material uniformity.
Metallographic Preparation. The basic preparation procedures for composites are the same as for
metals. Depending upon the size of the section, mounting in a supportive resin may be necessary to provide
a stable base for sanding and polishing. Resins requiring elevated temperature for cure or forming, such as
bakelite, should in general be avoided since many may approach or exceed the material's original cure
temperatuie and thereby introduce additional damage such as embrittlement, creep, or microcracking.
For most applications room temperature cure systems such as two-part epoxies or methyl methacrylate
resins work well.
Sanding should proceed from a relatively coarse paper (120 grit) to the finer grits (600). Typically
wet-or-dry sandpapers are used during this procedure with a substantial amount of water being applied at
all times to the surface being sanded to remove debris and prevent heat damage. For subsequent opera-
tions it is critical that the sanded surface produced is flat to ensure full surface contact during the final
polishing operations. For most composite materials, final polishing is best accomplished by utilizing a
high-speed polishing wheel covered with stretched silk (or other extremely low nap material such as nylon,
in combination with either diamond paste or an equivalent alumina polishing compound. Final polishing
usually does not require a large number of steps; usually 15 micron followed by 3 micron is adequate.
Resuits are further improved by fairly heavy downward pressure and counter-rotating the specimen to the
direction of wheel rotation.
6-6
Following polishing, etching is generally not required due to the distinct optical differences
which exist between most fibers and the surrounding matrix systems. An exception to this is fiberglass
where a dilute hydroformic or hydrofluoric acid etch may be necessary to enhance fiber visibility.
Plies. The orientation, number, and stacking sequence of plies making up the section of interest
can be generally identified by optical microscopy at magnifications ranging from 50X to 400X. At the lower
magnifications, individual plies can be identified by pronounced alterations in overall reflectivity due to
the fiber orientation of each ply, or by the existence of a linear interfacial region of reduced fiber density
(matrix-rich layer) between each ply (see Figure 6-3). Determination of the precise angular orientation of
each ply requires higher magnification examinations. As illustrated in Figure 6-4, fibers intersecting the
polished surface at an angle appear oval in shape. With this method, large differences between ply
orientations are easily seen by large variances in the elliptical shape of the cross-sectioned fibers. In
determining and reporting ply orientations, it is important to reference the measured angle against the
component's defined 0-degree axis to prevent confusion and maintain consistency with other analyses such
as stress analysis. Commonly, most sections are taken at 0, 45, or 90 degrees to the defined 0-degree axis,
with a vertical section plane. This vertical section does not, however, allow one to distinguish between
certain ply orientations such as +45 and -45 degree plies. A recent technique has been developed which
provides differentiation of all ply angularities with a single cross-sectional p!3nar view. In this method, the
cross-section is prepared such that the polished sectional plane is at a 45-degree angle through the
thickness of the part. This technique is presented in Figure 6-5. Small differences in fiber orientation, such
as fiber wiviiies5, uk" be du:erlizued by lileazw-ing Chu izteV'e:tud filuer le gti, a.d uilieizinglple
trigonometric calculations. For such measurements, the sectional plane should be vertical so that fibers
from the primary reference plies are normal to the polished surface. This will result in the reference plies
appearing circular and wi!l allow measurement of nominal fiber diameter.
Through the use of fairly low cost image analysis systems, both ply and individual fiber
orientations can be determined. Through optical imaging of the polished cross-sections, the intersectional
fiber length or the aspect ratio for each fiber is measured. Simple trigonometric calculations wilh these
data are then used to determine overall ply orientation in each region of view or the average and worst case
fiber waviness. Fiber waviness can be critical to certain product forms and applications like unidirectional
or filament wound structures in which fiber nestling and tow movement occurs. One advantage of image
analysis is that an automated or statistical evaluation of material integrity is available. Since fiber
resolution and differentiation from the surrounding matrix resin is required, high magnification (400X or
more) and various enhancement techniques such as etching (for fiberglass) or staining may be necessary.
inage analyses can also be used to determine the relative percentage of voids in the laminate.
Either by cross-sectioning or by evaluation of the fracture su-faces, a quantitative assessment of the
planar void content can be performed. By taking several measurements along parallel planes, a fairly
accurate determination of the void shape, distribution, and percent can be made.
6-7
N-..
7#
__~-~- _ -
N.(e
AIN
Figure 6-3. Cross Section of a Laminate With Figure 6-4. Identification of Ply Orientationby
Easliy DifherendatiedPiy Separation Fiber End Shape and Ovality
A nondestructive technique is available in which a plan view image of ply orientations can be
determined. This involves the use of radiographically opaque tracer yarns, such that X-ray imaging can
detect these doped fiber bundles. For tape prepregs, the tracer yarns are placed in the 0-degree
direction parallel to the fiber orientation. For fabric, the yarns are most often placed in the warp
direction. This technique is most useful for thin laminates, where the preciSe number of plies and
orientation can be determined. Thick laminates make it difficult to identify the number of plies,
although anomalous conditions such as ply orientation shift and ply dropoffs and pickups are easily
detected. Figure 6-6 presents a radiogTaph in which the tracer yarns are evident.
6-8
-45de +45 dsg
0 deg To
909 de
Figue
Crss6-.
Sc~o
Poishd
at45-dg Anle Trouh th Thiknes ofThePr
6t %-9
N./
Figure 6-6. Radiograph in Which the Tracer Yarns Are Evident for Determir-atiori of the Piy Orientation in
a Th,r, Laminare
6-11
Generally, uncured material identification techniques are seldom used in failure analysis because
the prepreg materials used to fabricate the failed part are not available to analysis. The most commonly
used techniques, when the prepreg is available, are HPLC and IR spectroscopy. HPLC and IR are widely
utilized in the aerospace industry to "fingerprint" the prepreg resin fonnulation. Standard HPLC
chromatograms and IR spectra exist for all qualified materials. These standards provide a quick check of
resin formulation. DSC is less frequently used for material identification. Although DSC can identify
reaction types, specific resin system formulations are difficult. X-ray fluorescence is used to detect sulfonyl
groups to verify the amount of diaminodiphenyl sulfone (DDS) hardener present. Atomic absorption
spectroscopy is utilized to detect boron which verifies the level of BF 3 catalyst present. X-ray fluorescence
and atomic absorption spectroscopy are usually used only if anomalies are detected in 11PLC, IR, or DSC.
High Pressure Liquid Chromatography. HPLC is a common technique for chemical character-
ization of composite matrices. HPLC separates individual resin components according to their size or
solubility using high efficiency, microparticulate columns.
HPLC test procedures are well established in the aerospace industry because of the widespread
usage of HPLC for "fingerprint" inspection of resin formulation. Resin samples are extracted using an
HPLC grade solvent such as 63:37 CH 3 CN:H 2 0. A schematic of a HPLC test setup and instrument
parameters are shown in Figure 6-7. The extracted sample is injected into the column and analyzed
using an ultraviolet detector. A typical HPLC chromatograph is shown in Figure 6-8. Components can
be identified by comparing the peak positions on the chromatograms to a reference standard. Peak
areas can be used to determine relative concentrations.
Sample
Recorder
and
Integrator
Mobile phase
Instrument parameters
Column: microbondapak C 18 waters
Mobile phase. premixed 63% C:-l 3 CNJ37% H 2 0
Flow rate. 1 5 mL/mln
Injection volume- 10 piL
Detection: UV 220 nm
Attenuation: 0.2 AUFS
"hart speed. 1 cm/min (0 5 in/mmln
Figure 6-1. HPLC Test Setup Schematic and Typical Instrument Parameters
6-12
Benzaldehyde
"DS MY 720
Reaction product
response
t I I I
5 10 15 20
Elution time. mim.
SFigure 6-8. Chromatogramof Narmco Resin Matrix Showing Major Peak Assignment
6-13
!nfrared Spectroscopy. IR is an organic chemical analysis tool which is extremely sensitive to
the overall chemical makeup of a complex chemical formulation. IR measures the absorption of incident
infrared radiation, aad is used to verify epoxy resin formulation. This technique is well suited for
characterizing commercial epoxy resins which often contain several components such as aromatic epoxy
main, reactive epoxy diluents for flow control, aliphatic epoxy flexiblizers, elastomers, curing agents,
catalysts, or accelerators. IR can accurately fingerprint the total resin formulation by detectin- the
absorbed frequencies of infrared light associated with the vibrations of specific molecular functional
groups. The degree of absorbance is proportional to the amount of functionality in the resin
formulation.
Because IR is able to fingerprint a given formulation accurately, it is used as a quality control
iool to assure that. lot-to-lot variations do not exceed certain limits. IR would also be useful to
determine the mapufacturer of the material when unidentified material is encountered. The typical
application cequires that the resin is extracted from the prepreg with reagent grade acetone at room
temperature. The extract :s then allowed to dry onto a salt pellet for IR analysis. The resultant
spectrum is then compared to a standard epoxy spectrum acquired from the supplier or or, a previous
batch of matenal. The spectrum is examined for contaminants (additional peaks) or gross formulation
changes (changes in the relative peak ratios). An example of a standard epoxy resin spectrum is shown
in Figure 6-9. If a deviation frjm the standard spectrum is observed, then each subcomponent should be
quantified and compared to standard formulation quantities. Quantitative methods lor the curing agent
and catalyst are considered below.
100
940
V 360
30 T
20--, , .
T9-1I (
3.300 3600 J400 3200 300W 28o)0 2600 2200 2000 1900 1800 1700 1600 1500 1400 1300 1200 1100 1000 900 800 700 -
Wavelength
T90-31/, IrA
A quantitative method for determining the amount of DDS curing agent and carbonyl epoxy has
been developed for the 350i resin system (Reference 2). This involves constructing a working
calibration curve for DDS conc-.ntrations in solution versus JR absorbence for the sulfone peak followed
by a measurement of the absorbance of the unknown concentration of DDS in an epoxy prepreg extract.
An alternate quali-ative approach would rely on comparing the relative change in the ratio between
these peaks as the conoentratio- varied.
6-14
Differential Scanning Calorimetry. DSC is a thermal analysis technique which measures heat
flow into or out of a sample by measuring the differential heat (Pnergy) necessary to keep the sample at
the same temperature as a reference sam:ple. In the case of a curing epoxy system, DSC measures the
heat released during the exothermic cure reactions as the specimen is heated at a constant heating rate.
DSC can be used to identify or characterize uncured prepreg. The cure of typical aerospace
epoxy systems consists of several chemical reactions with different reaction rates and activation
energies which cause certain cure reactions to be favored over certain temperature ranges. The reaction
types occurring in a particular system depend on the formulation of the system. DSC can be used to
separate the different cure reactions to help identify a particular resin system.
A typical DSC cell is shown in Figure 6-10. The sample pan and reference pan are placed i:. the
cell. An inert atmosphere of nitrogen is then introduced into the test chamber. Heat is transferred to
the pans through the sample platforms and hcet flow is monitored by thermocouples. Generally, prepreg
sample sizes of approximately 10 mg are used for DSC experiments depending on hdie size of the sample
pan used. Low heating rates of about 2 to 5C/lrinute are best to separate the various reaction peaks.
Heat flow versus temperature or versus time at constant heating rate from ambient to 3001C is
measured. An interface with a comn'uter or microprocessor for experiment control, data acqdisition,
and data analysis provides greater accuracy and more informatior from each experiment.
Lid Cove
blockr~ngSample
Sampleplatform
Heat er---"pa
junctiong
Sive Li phrae
Figure 6-10. Cross-SectionalDiagram of the DSC Cell Used by DuPont DSC Module
A USC thermogram for the Hercules 3501-6 resin system is shown in Figure 6-11. Exotherm
peaks are observed at two temperatures (149C and 213 0 C). The lower temperature peak is attributed
to the catalyzed cure reactions while the higher temperature peak is attributed to the uncatalyzed cure
reactions. The total heat of reaction (the area between the thermogram peaks and baseline Ti to T4),
the relative peak size, and reaction temperature can identify resin constituents. Comparison to a
standard thermogram can identify a particular resin system.
6-15
21300
( T3540)
HEATUP RATE
(3900 F /
(J2)\/
(212'F) (T4)
The advantages of DSG formulation verification are the small sample size an(1 the experimental
ease. T he disadvantage is the need for standard thermogm-ms and background information fox system
identification.
6-16
the boron concentration. A boron calibration curve is established using a certified standard. 'Typical
working range of concentrations for boron is 400 to 4,000 Al per ml of solution.
100
0 Plot normalized to Fiberite 934 at 103% peak height
Relative 1,1
sulfur
peak
height J--
* percentage
J L ---
NARMCO 5208 HERCULE! 3501 Fl8E!IITE 934
Once a inatf-rial has cured, ascertaining its identity poses a relatively difficult problem arising
from the generally similar chemical structure of cross-linked thermosetting systems and the tendency
of the cured resins to resist yielding their identit ies 4y typical analytical methods. However, since many
of the currently used epoxies exhibit nearly identical properties (within a cure-temperature family),
identification of the exact resin system used may not be as critical as identifying a material with its
proper cure- temperature family. Through the combined use of determinations of Tg and residual heat
of reaction measurements, the particular cure-temperature family can be determined.
6-17
In some isolated cases during post-failure analysis investigations, identifying the specific resin
system used to fabricate the failed part may be in,portant; however, the investigator shouid he aware
that extensive capabi!ities in this area are only partially developed. Sufficient differences exist iii basic
formulation so that relatively simple tests, such as XRF, may be used to identify different systems. For
the most part, the ability 1-o identify resin systems within the same cure-temperature family is difficult.
Two of the most promising techniques are pyrolysis gas chromatography (PGC) and diffuse reflejtance
(FTIR) spectroscopy These techniques are discussed in the following paragraphs.
Pyrolysis Gas Chromatography. The PGC method is used to determine epoxy matrix
chemical formulation. Pyrolysis is the vaporization and degradation of die matrix resin under
controlled heating. The resulting decomposed products pass through a gas chromatograph which
separates the components by physical or chemical parameters such as molecauar size or polarity.
6-18
*NARMCO 5208
345F A7 5F/-,n
CONCENTRATION
1 2 3 4 5 6 7 k! 9 10 11 1" 13 14 15 16 17 18 19 0
T..A E, mn
*FIBERITE 934
3450F AT 5F/rain
SIt
e ~ CON^-F-1TR ATtCN
31 2 4 G 7 9 M .10 1 1 1 3 14 IS 16 17 18 19 20 21
"TIME. -- ,
The degTee of cure of a therrnosetting polymeric composite maitrix is a critical variable affecting
the performance of the composite, esj-ecially in elevated ter:',wrature or moist environments-
Insufficient cure may occur due to lower than required cure temperatures, shorter than raquired cure
times, or errors in resin formrulation or chemistry. Errors in resin tormulation or chemistry are rare
herause of the stringent specification requirements, vendor certification, :,nd receiving inspection
testing cmmon throughout the aerospace industry. However, because conditions ot undercure can
significantly impact structural integrity andi can be ral)idly mdni easily chucked, their analysis should lIe
included ,1s a routine step in most failure analyses. 1Merilfing the (legree (Of cure is analogous to checking
6-19
0
H02238S' N ,
Z (N
U U
sou
80 Z 0-ii-0 0
NormalizedN
peak 32?
heights 60o-
0i./
Z
60J
40 -\U2 1 302 /A
20
the correct heat treat condition on fractured metal components, a procedure considIeredl routine in
metals failure analysis.
As described in the materials characte~rization FALN contained in the beginning of Section 6.
verifying the state of cure of a fractured component normally represents the first stage of mcaterials
characterization analysis. There are several reasons for performing such analyses (luring the initial
stages of an investigation. Perhaps the most significant reason is that sucit analayses can be p-trformed
quickly and easily and produce results which may have a significant impact (in downstream investigpitive
operatioas.
In most cases, evaluation of the dlegree of. cure of a failedi component is hest ad'lressed by it
two-st~age process. In the first stage, Tg analysis establishes whotth, r a lowe-(r than expectedl uegr,.'- of
cure may exist. These analyses (Paragraph 6.3.1) are generally relatively easy to perform and~ I.hereforr
impose a relative~ly simall ivtpact on the amount of effort necossary to carry-out ans nvest i~rat io: . As
ith evaluating the Leat treat of metakl, most of the unitt the apirr --e OI~rimol pro-vessing has Lren
performed, arid simiflarly for composites, the corr-ect (legTce of cure W. . .t~lictt'd and further analyses
wAil not be necessary. lowevei, if the Tg of the laminate and at subhseq~ ot ret-est of a dIried (desiccated)
laminate does not match that anticipatted for the material used, t he secon~d stage analyses
(Paradgraph 6.3.2) should be perforined. ibis involves (leterminlma ion of the e-xi'nZ of arireiited miat erial
- can be perforin.2d to (valuate if tius anomialy represents a condiit ion of uandercure. Here it should be
noted that a low Tg may arise both from ina'lcquate. processing cr resin formulat ion virrors, as, well a,,
the accidental tise of a wrong :Pa~tArial much as a 250,~F curing prepirte in pl~acc of a :350'F curing
6-201
.I . . . i
I '.
SU Epoxude funcional
.H - t CH
*I I *i . . . .. I I
F 6- 1. I" C i"e"
. ... , ,V.
'v "rT I
maeil' degre of cur "a prsete at th beinn ofScin6
6.4.
* Gls 7nlto Teprtr Analysis " Epoxde group O~sappears
prepreg. A detailed description of the logical sequence o? steps (F'ALN) inivolved in assessing the
materials' degree of cure w-as presentedl at the beg'inning of Sectiron 6.
The temperature at which an amorphous polymer reversibly changes from a brittle, glassy state
to a more flexible, rubbery state is The glass transition temperature. In the case of therrnosetting
polymers this transition temperature directly reflects the dg'ree of cure achieved by the material. This
transition tempei-ature is manifested as a chango in slope of the curve ebtained by plotting any of the
primary thermodynamic propperties against temperature. This change in slope falls in the same
temperature range as that in which mechanical softening occurs. A variety of t#4r:ijniques ,xiSt lby which
the point of this softening can b,! determined. The most commo)n are thermal analysis tests such as
6-21
C,-
I J5 1'150F
0N H
Z U)
0 0
7e4121 C (25 1 (3 ) (52 24 3"7
Waelativerf CM I )( C .4
RIFS
ofGrahit-Epxy
Figue6-6- (S130 15) ompoiteBefre nd AterThemalgin
Ahesorbecanicl anlyi
000 A an yai mcaia nryS!DA, n S.TA stems
th inardiplcmet
rlaiv f ua brtz
probe ince in ihth a pes (otc
rotc. Du) n
tehniqume.chanclay~.TAisapesz o ersigh
dmlcnerntof temeaslr (wi.The generaltihighenitivity
terlatibetweinea displaempenatur ofah quart.proe (whiceh isncontaolet withthe shamplte surfrc.ourding
cabr(hciscontrolled)haig
eodi ge canerevecdaaln thcurmalcy. Forlnoet dimpesioa changiesto exhfibied
Intercept
OQuarti penetration
probe Probe
displace-
ment
Penetration
specimen
iTg
Penetration M-Ihod Temperature
Figure6. 17. Glass Transition Temperature Figure 6-18. Glass Transition Temperature
Determination - TMA Determination- TMA
PenetrationTest Setup PenetrationTest Measurement
Flexure. In this test mode, a rectangular ;ample 0.1 inch in width by 0.25 to 0.50 inch in length by 2
to 4 plies in thickness is placed in three p )int flexural stress within the test chamber as shown in
Figure 6-19. With this configuration the probe measures beam deflection as a function of
temperature. Again, as the transition temperature is reached the resin softens, the beam
(specimen) deflects, and ain >flection in the slope of the thermogram being generated occurs.
Nominally, the heat-up rate is approximately 20'C/minute with a probe load of 10 grams. A
thermogram of a 177"C (350'F) epoxy system is presented in Figure 6-20, showing the tangents to
the slope. The Tg is defined as the temperature at which the tangents intercept. Care should also be
taken not to remove specimens from the laminate near a surface which has a different chemistry or
cure temperature material attached. This includes adhesively bonded regions, potted areas, or
surfaces which have had surface prep fabric such as peel-ply.
Expansion. This test is run in much the same man er as penetration, except a larger diameter
probe is utilized to prevent penetration into the sample at the point of softening. As the specimen is
heated, the resin is free to expand in the transverse or lateral direction. Since a distinct increase in
the rate ofthermal expansion normally occurs at the glass transition, the ti ermograri generated by
this technique typically provides an additional method by which to determine the materials' glass
transition (Figure 6-21). As in the penetiation method, '.i ement of the probe directly on carbon
fibers may interfere with accurate Tg determination.
6-23
I ISample
tube -- Slope
S~int6rcept -
"Quartz
"A flexure PIobe
prcbe dispiace-
ment
Flexuro
speclm3n 1 - T _
Fiexure Temperature
support
Figure 6-19. Glass Transition Temperature Figure 6-20. Glass Transition Temperature
Determination - TMA Flexure Determination - TMA Flexure
Test Settp lest Measurement
Of the three methods discussed above., the preferred niethod is the flexural technique.
Generally, the flexural test method produces sharper, more distinct inflections and only requires a very
small sample. The penetration and expansion methods are used less often because of the interference of
the carbon fibers with accurate Tg determination.
Dynamic Mechanical Analysis. High performance epoxy systems used in composite matriceb
exhibit a viscoelastic response to applied mechanical strains. DMA techniques are a useful means of
determining the viscoelastic nature of epoxy matrices. The glass transition temperature can be
determined from DMA expe, inents.
DMA involves the application of a sinusoidally oscillating tensile, torsional, or flexural strain to
a sample and measuring the response versus increasing temperature from -150'C to 400'C. Heating
rates vary from 1VC per minute to 10'C per minute depending on the sample size. Data are presented
in terms of storage modulus (elastic component), loss modulus (viscous component), and tangent delta
(ratio of loss modulus to storage modulus) versus test temperature.
Typical DMA data are shown in Figure 6-22. The glass transition temperature can be
determined as the peak temperature ef the high temperature transition in tangent delta or loss modulus
or as the knee of the storage modulus curve.
DMA is typically not used wherc ,iiiy glass transition temperature measurement is required.
Larger sample sizes and more difficult sample preparatwn and test methods limit the utility of DMA
6-24
transition point - 1011
STg
80- 0 5 -- O 9 G -- I Tg ,ta n b )
Expansionr
5"- Penetration "
0 101I0-
70
Relative Expansion DyreS.' .
probe 65 ocm
displace-
ment 60 I ., ,/
1 tan
50-
45. I I I I I I I I I__I__ _ _ _____
0 40 80 120 160 200 240 -100 -20 6.3 140 220 300
Temperature. OC Temperature. 0(
for glass transition measurement. Since carbon fibers tend to diminish the viscoelastic transition
magnitudes, samples should be tested in a matrix dominated direction if possible.
A typical DSC thermogram for a cured epoxy is shown in Figure 6-23. The glass transition will
appear as a change in the slope of the DSC thermogram. Determination of glass transition temperature
using DSC is sometimes difficult because of the exotherm peaks due to additional reaction during the
DSC test.
6-25
3.6 k
3.2 ~- 000
J7' TC
- AHA
2.8 -
A
A
A
A
A
Heat
flow. 2.4 A
mnW
2.0 - T
Tg
1.6
1.2
I I II
160 180 200 220 240 260
Temperature. C
Figure 6-23. Typical DSC Plot for Cured Epoxy IllustratingDeterminationof Tg and Heat of Additional
Reaction
Solvent extraction, with or without subsequent analysis using IR spectroscopy, may also be used to
indicaie extent of unreacted material. Excessive handling and analysis difficulty discourage these
techniques.
Differential Scanning Calorimetry. DSC can be used to measure the heat of residual reaction in
a partialy cured epoxy matrix. Experimental procedures as described at the beginniig of Section 6 are
used on cured laminates. The heat of residual reaction is the area between the DSC thermogram and a
baseline drawn from the reaction onset temperature, Ts, to the reaction completion temperature, Tc, as
shown in Figitre 6-23. The presence of mjeasurable heat of residual reaction indicates some undercure.
6-26
* The extent of unreacted material can be calculated by dividing the heat of residual reaction by the total
heat of reaction of the prepreg.
Dynamic Mechanical Analysis. DMA can be used to detect undercure and moisture
degradation of cured laminates. Experimental procedures as described earlier are used. An undercured
laminate w.ll exhibit a double high temperature peak in tangent delta and 'oss modulus plus an initial
loss with subsequent recovei., of storage modulus as shown in Figure 6-24. The initial softening
(increase in tangent delta, loss modulus; decrease in storage modulus) is attributed to the glass
transition temperature of the original undercured material. As the test temperature increases, the
matrix undergoes further cure allowing a recovery of properties. The final scftening is due to the glass
transition temperature of the fully cured material. Laminates with absorbed moisture exhibit a peak in
tangent delta and loss modulus at approximately 100'C due to the increased activity of moisture at this
temperature. The presence of this peak indicates significant absorbed moisture in the laminate.
2
Pa iIbftin )
7
1.4x10 - 42031)
Legend: Undelcu- ed
- - - --- -- 25'16 hardener
5
o tt2 % hardener '/"- Fully
1451)
17 cured
1
. 5 1125e
*
xlO (2145)
10 0t IGOI 20 00
I
Ix10O - (145)
Frequency 1 Hz
Figure 6-24. DMA Loss Modulus Versus Temperature Showing Effect of Undercurefor Epoxy Resin
Solvent Extraction/Infrared Spectroscopy. This- method can be used to remove low molecular
weight (uncured or partially cured) polymer from the surrounding cured matrix. A solvent, such as
dimethyl acetamide (DMAc) or reagent grade acetone may be used to extract the uncured segments
from a thin sample laminate. By weighing a solvent extracted and desiccated (moisture free) sample, the
percent of unreacted material can be calculated. The extracted sample can be analyzed using IR
spectroscopy. The test procedure is identical to that described in Paragraph 3.1. The unreacted
functional groups can be identified by the peaks in the IR spectra. The extent of reaction can be
determined from the peak intensities by comparison to standard prepreg spectra.
6-27
6.5 CURED MATERIAL CONTAMINATION ANALYSIS
Cured composites may contain defects due to contamination during manufacturing and
assembly of composites structures. Foreign material contamination present in the cured composite is
readily identified by an array of traditional chemical analysis tools. This paragraph addresses the
methods for selecting the technique and interpreting the data. Emphasis is also placed on showing how
the contaminant affects the fracture morphology and the ultimate integrity of the failed component.
The essence of a successful material contamination analysis is: (1) recognizing the anomalous
fracture morphology, (2) exactly documenting and interpreting the anomalous features, (3) chemically
identifying the material contamination, (4) establishing that the detected contamination is not an
artifact of specimen handling or post-failure inservice exposures, (5) identifying possible sources of
contamination, and (6) evaluating the criticality of the contaminant as related to crack growth and
initiation.
Foreign material defects lead to lower crack propagation energies and a corresponding typical
fracture morphology. Contamination analysis is initiated when anomalous crack propagation or foreign
material inclusions are detected on the fracture surface during the course of the FALN execution. While
performing optical microscopy or scanning electron microscopy (SEM) examination of the fracture
surfaces, the specific fracture morphology allows the analyst to classify the contamination as either a
particulate or weak boundary layer contamination. This step is crucial t,' selecting the steps needed ,
chemically identify the contamination and to relate the analysis to the overall failure analysis !'!,!
location of the origin. It is also crucial to be able to recognize artifacts created by sample preparation or
due to post-failure exposure to contamination sources. The following definitions will serve to classify
contamination types.
In this case, inhomogeneous resin fracture will be observed with inclusion particles on the
fracture plane with the optical or electron microscope. The particulate significantly reduces the net
cross-sectional load bearing area leading to lowered properties such as reduc:d interlaminar shear and
tensile strength. Although composites with their fiber reinforced construction are less sensitive to small
localized defects, large particulates have been found to initiate failure. An individual particle defect, if
located at a critical location, may also act as a crack initiation site with corresponding fracture lines
radiating from the defect.
The presence of weak boundary surfaces account for the majority of crack initiation and
subsequent component failures due to contamination. The weak boundary layer acts as a barrier to
adhesive bonding between:
1. Adjacent plies
2. Dissimilar materials or microconstituents such as fiber and resin or honeycomb core and
laminate
3. Adhesively bonded subcomponents such as stringers and skin.
6-28
ad The weak boundary layer is formed by a chemical compound whic'h interferes with the wetting
and adhesion mechanism between the epoxy adhesive layers. Interfering chemicals such as
hydrocarbons, fluorocarbons, and silicones are common in composite manufacturing as mold release
compounds and parting agents. During the fracture event, the weak boundary layer provides a low
energy crack propagation pathway through the structure. The fracture surfaces exhibit localized regions
of smooth and featureless mating fracture surfaces, as opposed to normal fractures, where there is
significant resin microflow deformation, hackle formation, and secondary cracking. When stress
analysis has indicated that a composite structure failed at a fraction of the design limit strength, and
delaminations are present, weak boundary contamination layers should be suspected.
As illustrated in Figures 6-25, 6-26, and 6-27, contamination caused by Teflon paddle and
Frekote release agent along the interlaminar ply interface exhibit inhomogeneous resin fracture s
compared to a typical uncontaminated baseline fracture specimen llustrated in Figure 6-29. A seccnd
example of an isolated foreign particle inclusion, in Figure 6-29, illustrates the distinctive fracture
surface asperity and crack initiation lines radiating from the particle. However, isolated particles will
not significantly degrade the structural integrity of the composite because the fibers already act as
primary sites of crack initiation. As a result, low delamination strengths are expected and incorporated
into the designs. Particulate contamination along these directions can be tolerated in moderate amounts
without significantly affecting the performance.
1- Mechanically induced
crack direction
6-29
4-.* Mechanically induc'!d
crack direction
Dry
-a-r. r.,.. nvi yea. Ccoa)-l Tes1- SVclfer '0'
Figure 6-26. SEAv, AvIiuyogap, i0a &AiteJ n'-,aswc AycI Cw~,,IcIIu IV,.,~,op.aiIh 1z
Figure 6-27. SEM Micrograph of a rrekote Release Agent Contaminated Ten'sion Specimen (2, OOOXJ
6-30
91
AM
AM
6-31
Figure 6-29. SEM Aicrogr-iphs of Foreign Particle Inclusions illustrating the Characteristic Fractilre
Surface Asperity and Radiating f-racture Lines
320
surfaceSpecial care must be taken to avoid confusing the post-failure surface debris on the fracture
surface with foreign particle inclusions in the resin. By definition, the surface debris will not initiate
radiating fracture lines. This is especially a problem where the fracture surface has had a long
environmental exposure or has experienced in-service cyclic loading conditions. Caution should also be
exercised when extensive mechanical cutting was required to extract the specimen from the failed
component. Often the particle shape can lead to identification of the contamination source where the
particulate contains fractured carbon fibers and the source could be traced to machining debris.
Another concern in examining a resin system with particles on the fracture surface is that the resin
may have intentional particle additions (or phases) which improve properties or manufacturing
characteristics. If there is a suspicion that the resin system may have been modified in this way, the
analyst must prepare a ib fractured specimen for SEM examination as a reference fracture surface.
The flow diagrmn in Figure 6-30 reviews the sequence of steps taken to perform initial
inspections and to classify the contamination type. The next sequence of steps involve selection of the
appropriate analytical techniques to make an accurate chemical identification of the contaminant.
Direct chemical analysis probes can be very decisive in providing a chemical identification in a cost
efficient manner. The flow diagram in Figure 6-31 for analytical work reflects this emphasis. A number
of analytical techniques are recommended in order of decreasing information per unit analysis time for
each application. There are specialists for each analytical technique whose expertise in instrumentation
and spectroscopy will be required. The following paragraphs on analytical techniques are designed to
inform the investig-ator of the -pecial terminology used for these techniques and highlight special
concerns which arise during a composite failure analysis. The section concludes with an applied example
of contamination analysis during a failure analysis.
The scanning electron microscope and the electron probe microanalyzer (EPMA), commonly
known as the electron microprobe, are powerful tools for identification of contaminant particles on a
fracture surface. This paragraph covers the elemental detection capabilities of the SEM and EPMA for
foreign particle detection. The microscopic image formation is covered in this section. These
instruments are used when a quick qualitative identification is needed and provides information
regarding selection of more detailed methods to be performed later.
The SEM and EPMA instruments are equipped with X-ray detectors to monitor the X-ray
emission spectrum generated when a finely focused electron beam impinges on a microscopic feature of
interest. Electron beam induced X-ray emission produces a spectrum unique to each element present
allowing direct determination of particle composition. The analysis volume is determined by the
electron beam energy and the density of the specimen surface. Typically, the analysis volume is 0.5 to 2
micrometers in diameter. If the X-rays are measured with an energy detector, then the analysis is
termed energy dispersive spectroscopy (EDS). If the X-rays are measured with a wavelength detector,
then the process is termed wavelength dispersive spectroscopy (WDS). Each method has its own unique
advantages. Conventional EDS can acquire a spectrum for all the elements from sodium to uranium
simultaneously in 30 seconds to a minute. (Uitrathirn window EDS detectors can detect elements to the
boron range.) Because of the limited energy resolution of the EDS detector, there are a number of
spectral peak overlaps. For example, aluminhum and bromine peaks occur at 1.48 KeV WDS has
6-33
Are there
crack No
Yes
Optical and SEM
examination of
fracture morphology
I Which
contamination
type?
Examine cross
sections of the
fracture and related
areas to show
location and extent
of contamination
a
Ch emical ralysis I
extremeiy high signal-to-noise sensitivity and resolution. WDS can also detect boron, carbon, )xygen,
and fluorine. The disadvantages are that only a single wavelength at a time can be detected, requiring
many minutes to scan even a limited range of elements. Of concern to composites failure analysis is the
ability to use WDS for light element detection. A partial list to assist the analyst in identifying elements
by EDS and WDS is given in Table 6-2. A SEM micrograph and EDS spectrum talc contamination at a
bondline is shown in Figure 6-32. The SEM reveals the sheet-like crystalline morphology and the EDS
spectrum exhibits the primary elements typical of talc powders, magnesium, and silicon, along with
other particulate impurities.
In using EDS or WDS several concerns should be mentioned. First, the composite specimen is
introduced into an instrument vacuum chamber at pressure ranging from 1X10- 8 torn to 1X10- 4 torr. If
the contaminant under investigation is volatile in the instrument vacuum chamber, there is a possibility
that important elemental information will be lost before analysis. Second, these techniques were
originally developed as techniques to study conducting metal specimens. Polymers which are
6-34
0 InChemical analysis
No 'INo
SX-ray dfrcto i
]nfrared spectroscopyl
j IeFTIR
*Diffuse reflectanceJ
Further surface
sensitivity of light Yes
<element detection >
"N.ne eded,7 ~Surface ana!,siss:
No 9 X-ray
photoelectron
spectroscopy
e Auger electron
spectroscopy
I Secondary ion
mass spectroscopy
I1Return to FAL N
nonconductive present a specimen charge buildup problem which is reduced by vapor depositing a
conductive coating on the specimen. Typically, a very thin 5 nanometer film of gold or other noble metal
alloy is used. These films provide interfering peaks in the X-ray spectrum which may mask elements of
interest. When analyzing composite contamination, it is sometimes useful to coat these surfaces with
conductive carbon since carbon is to be expected and will not be detected in the EDS spectrum. There is
a significant literature base to suggest the electron beam can induce degradation and thermal effects to
polymers. The degradation can be minimized by using low electron beam accelerating voltages and low
current densities. Since WDS usually requires very high beam currents, this detection method should be
used as quickly as possible to reduce the detrimental beam damage effects.
6-35
Table 6-2. X-Ray Emission Lines, Partial Listing
EDSX-aenris V
WDS X-ray wavelengths (K~y). nm
B6.76 Na 1.04 1.07
C4.47 Mg 1.25 1.30
N3.16 Al 1.49 1.56
02.36 SI 1,74 1.84
IF1.83 P
Cl
I 2.01
2.6?
2.13
2.86
Cl2.31 2.4
(Rel. 5)
(Al
2000
() Cnunts loco
Au
Au
50 SL +Pd F
500d[ P
0 1 2 3 4 5 6 7 8 9 10(
6-36
* 6.5.4 X-Ray Photoelectron1 Spectroscopy
BE = hv-KE-0-Vcharge (Eq. 1)
where:
X li/Si(E.2
Element x atom% = hI!SXX100 (Eq. 2)
where:
S = relative sensitivity factor
Ix = ubscrved intensities
x = each element
6-37
Table 6-3. Approximae XPS Peak Positions and Emission Probabilitiesfor Common Elements
Relative emission
Element Binding probability (cross
energy section for aluminum
X-rays) (Ref. 4)
01
. I0 C
Binding energy, eV
Figure6-33. Typical XPS Spectrum of an Epoxy Resin With Elements Identified
6-38
When analyzing a composite fracture surface for foreign material contaminants, low surface
energy compounds or releasing agents are high on the list of potential contaminants. These compounds
are typically hydrocarbon oils, fluorocarbons, or organic silicone oils. Table 6-5 summarizes a few rules
of thumb in interpreting the XPS data for these compounds. Some materials which could be
contamination sources are as follows:
1. Molding tool release compounds and sprays
2. Parting films and breather fabrics
3. Prepreg backing papers and films
4. Solvent impurity residues
5. Vacuum pump oils
6. Machining debris
7. Plastic resin sweeps
8. Gloves
9. Vacuum bags
10. Molding compounds, vacuum bag sealants.
JR spectroscopy is the primary technique for identifying the compound or chemical family in
which a foreign organic contaminant belongs. Infrared spectroscopy is much less sensitive than XPS or
EDS/WDS but provides more chemical information. Spccial techniques have been developed to apply IR
6-39
Table 6-5. Ruleq of Thumb for XPS Identification of Typical Release Agents
F--
e Hydrocarbon * Excessively high carbon levels on the surfau(. compared with a cohesive resin fracture
0 Confirm by FTIR analysis of the residue frjm i. olvent rinst ci the fracture surface
* Fwuorocarbon 0 Carbon-shifted peak In the range 288.0 to 292.0 eV; fluorine peak at 689.0 eV
* Ascertain whether there is a fluorocarbon additive to the resin
e Compare with XPS spectrum of known fluorocarbon release agents or parting films used in
manufacturing the part
* Silicone 0 Silicon peak ot 102.0 eV: silicon Auger parameter In ;he range 1708 5 to 1709 5 eV
e Where there a,'e significant s 'con levels. the carbon-to-oxygen ratio will exceed the resin value
*Because some inorganic minerals exhibit the same "PS shifts, verify that there are no inorg;.nic
mineral particulates )n the fracture surface vltn SEM/EDS
* Ascertain whether there is a silicon additive to the resin
* Confirm by FTIR analysis of the residue from a solvent rive of the fracture surface
* Compare ,vlth XPS spectrum of known silicon release agorits used in manufacturing the
component
to the weak boundary layer and the particulate contamination problem. These methods are addressed
in the following discussion.
Organic polymer molecules absorb infrared radiation at frequencies which relate to the
chemical bonds within the molecule. The physics of this process can be attributed to a mechanism
where the infrared light is absorbed at discrete frequencies associated with characteristic molecular
vibrations within the molecule. This principle forms the basis of infrared spectroscopy as an analytical
technique to characterize polymer resins. Each polymer molecule absorbs infrared light in a distinctive
pattern which can be used to fingerprint the molecule. Figure 6-34 illustrates a typical IR spectrum of
an epoxy spectrum. In a contamination analysis where one is investigating an unknown identificat; ,n, it
it ueful to compare the spectrum of the unknown with infrared spectra of model compounds published
in an atlas or stored in a database. Ihfrared spectra are usually plotted as intensity versus wavenumber
(CM-1), wavelength (micrometers or microns), or frequency (s-i).
/CH?-CH-CH2l
-N
\ Epoxy 0
vibration
l
~~l ..
k- "- Alip h a t icC
I,
ttH st retch mod
h
C-H stretc- I V
mdeformation--
Aromatic A--iPara
-i substtution on
6-40
Intensity can be expressed as percent transmittance or absorbance as shown in the following
equ,-tions.
Percent Transmittance
__I
%T - (Eq. 3)
Io
where:
%T - percent transmittance
I = selected radiation
lo = incident radiation
Absorbance
where:
A = absorbance
I = detected radiation
lo = incident radiation
Recently, several developments in the field of IR analysis have allowed the technique to become
more practical for solid materials involved with composite failure analysis and contamination analysis.
The first development is the use of FTIR instruments. It is orders of magnitude greater in sensitivity
than conventional IR instruments. It is almost a necessity to select the FTIR instrument for sensitive
cont iination detection. The FTIR sensitivity can be used to complement the XPS in weak boundary
layer identification. The analysis can be approached from several directions. First, a fracture surface
can be rinsed with an ultrapure solvent, such as methylene chloride, which is allowed to evaporate on a
salt pellet (the standard specimen holder). The residue can then be analyzed as in conventional FTIR
transmittance analysis. Often there is enough material on a contaminated surface to be detected and
the increased chemical information increases the reliability of the contamination source identification.
If a large amount of particulate contamination can be isolated, then particles can be dispersed into a
salt and then consolidated into a pellet. Often a reasonable spectrum can be obtained.
Reflectance methods are desirable in order to avoid the use of removal methods described
above. The attenuated total reflectance (ATR) method is implemented or, an FTIR instrument which
uses an infrared transparent crystal with a high refractive index. The IR radiation is then caused to
propagnte through the crystal and the near surface region of a specimen. The depth of penetration is
typically 0.5 to 2 micrometers. The method has a limnited application to fracture surface, since it
requires a fairly smooth and flat surface to obtain a spectrum. Diffuse reflectance, an older reflectance
method, is coming back into use on the FTIR instruments. This method relies on the detection of
scattered IR radiation directly from a solid surface. The technique has seer limited application for
analysis of rough surfaces which are usually ercountered in composite fractures.
6-41
6.5.6 X-Ray Diffraction
X-ray diffraction (XRD) techniques are used for a wide variety of crystal structure
determinations. In composite failure analysis, XRD is primarily used for crystalline foreign material
particulates which can be isolated from a fracture surface. The cylindrical Debye-Scherer or Gandofii
camera is commonly employed for crystal phases in identification of p)articles. The basis for X-ray
diffraction is given by Braggs law:
A = 2d Sin 0 (Eq. 5)
where:
X-rays diffracted from a particle strike a film from which the diffraction angle theta is
measured and consequently the crystal lattice spacing is determined. A tabulated index of lattice planes
(Reference 6) can be searched for compound identification.
Secondary ion mass spectroscopy (SIMS) can be used to obtain several information levels about
the surface and near surface region of a solid material. SIMS is used in composite failure analysis of
weak boundary layers when XPS cannot provide sufficient sensitivity or chemical information. SIMS is
based on energetic ion beam impact (or sputtering) on a surface with subsequent ionization and removal
of the surface atoms. The ejected surface ions are then detected with a mass spectrometer. The plotted
spectrum of intensity versus atomic mass units (amu) can be interpreted to identify the elements or
molecular fragments of the parent molecule. A number of SIMS instruments exist to perform specific
functions, the simplest of which (often the most frequently encountered) requires very high sputter
removal rates to provide sufficient secondary ions to the detector. These instruments operating in the
dynamic SIMS mode can be used for detecting extremely low 'evels of elements as a function of depth
into the surface. Due to the very high current densities involved, this method (toes not work well on
nonconducting composite surfaces due to specimen charging. Static SIMS applications use very low
sputter removal rates to characterize the surface molecular layer. Future improvements ii this method
to allow routine polymer characterization may find significant uses to complement XPS analysis of
composites. The SIMS microprobe, a variation of the static SIMS instrunient, has an inmaging capability
which can be used to characterize particulates and map chemical inhomogeneities across the surface.
Auger electron spectroscopy (AES) is a surface analysis technique based on electron beam
induced Auger electron emission, characteristic energy for each element, and escape from the surface in
a manner analogous to XPS. In fact, many surface analysis instruments are designed to perform AES
and XPS with the same spectrometer. Since AES requires a conducting specimen, its applications are
6-42
be useful in
limited in composite analysis. However, since the carbon fibers are conductive, AES can
resin. The
determining whether an exposed carbon fiber failed at the fiber matrix interface or in the
and the surrounding resin (which may
analysis is based on the elemental differences between the fiber
Figure 6-35. Usually
be a sizing resin different from the nominal epoxy resin composition) as shown in
path from the fiber to the sample holder.
preparation methods are required to provide a conductive
A secondarily bonded composite structure was returned from service after visual inspection
a laminated
revealed surface cracks on the outer painted skin surface. The composite construction had
When the composite
skin bonded to a honeycomb core. The bonding adhesive was a FM-300 adhesive.
buckling away from the
part was sectioned, it was noted that the paint cracking resulted from the skin
revealed
honeycomb. An optical cross section is shown in Figure 6-36. Optical microscopic examination
of mating
the FM.300 adhesive separated along the centerline of the adhesive layer. SEM micrographs
lack any significant evidence of
separation surfaces are shown in Figure 6-37. The smooth surfaces
occurred. The
fracture features or resin deformation, indicating tbit a low energy adhcsive failure had
The
adhesively separated surface followed the contoured boundary between two plies of adhesive.
fibers of one ply made an
mating surfaces are perfect replicas of each other. Even the protruding scrim
impression on the other ply without showing any evidence of bonding.
The XPS spectra of the debu.ided surfaces shown in Figure 6-38 revealed a significant amount
of silic,,, fluorine, and carbon in addition to the expected composition of the adhesive. XPS peak shifts
indicate the presence of silicone, fluorocarbon, and hydrocarbon chemical functional groups. The next
step was to rinse the fracture surfaces with methylene chloride solvent and allow the soluble residues
from the surfaces to precipitate on a salt pellet. The residue was then analyzed with the FTIR where
hydrocarbon oils were identified in the spectrum. It was important to remember that these surfaces
may have experienced a significant exposure to contaminants from the service environment. Because
the SEM micrographs implied the weak boun iry layer was present during the curing of the adhesive
by the presence of scrim fiber impressions on the mating adhesive plies, the contaminant must have
been present during manufacturing and curing of the bondline.
Using this assumption, areas of the composite part with similarities of location and
construction (the disbond occurred adjacent to a core dropomf were examined for bondline integrity. To
ensure that no secondary contamination would occur, the part was cut with hand sheais (cleaned in a
solvent) which produces very little particulate damage. The part was also cut in the analytical lab well
away from fabrication and machining areas where contaminants can be easily introduced. The result
was that the adhesive bondline was easily separated by hand after being cut out with the shears.
Separation occurred in a similar manner at the interface between the two adhesive plies and the mating
surfaces were smooth and featureless. The advantage of finding these weak boundaries is that the
specimen can be separated and introduced into the XPS spectrometer within minutes, thereby ensuring
that. any contaminant found is directly related to the original debonding. Figure 6-39 shows the XPS
spectra in which the primary contaminant is silicon with the appropriate shift expected for a :.ilicone
release agent. Subsequent FTIR analysis of the soluble residues rinsed from the surface did not detect
any anomalous functional groups, which implies the contamination is very minute.
6-43
8-
7e
5i
Oxygen
DN/DE 4 ..
Carbon
2*
1 t-
o 200 400 600 800 1000 1200 1400 1600 1800 2000
Kinotic onorgy. eV
(a)
7-
DN/DE
4 ,,
Nitrogen
0 _ Silicon
_ _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ m
Carbon
0 200 400 600 800 1000 1200 1400 1600 1800 2300
Kinetic energy. eV
(b)
Figure 6-35. AES Spectra of (a) Carbon Fiber and (b) Exposed CarbonFiber on a Tensile Fracture
63-44
Ply 2I
figure 6-36. Optical Micrograph in Cross Section of a Laminate Skin to Core Buckling Separation
6;-45
ESCA survey SF- 11.700. 2.925 DAT: 580
7 - , ------ 0
6
44
3- C
1
SI Br
-100
90 -00 -700 -600 -500 -400 -300 200 -100 0
Binding energy. eV
61
4~
N(E)/E 0 4
010
lfr
-1000 -900 -800 -700 -00 -500 -400 --300 -2uu -100 0
Binding energy, eV
6-46
ESCA SURVEY SF= 15.600 3.900 DAT= 8.70
* . .
6-
Laminate skin side
5t
4
C
N(E)IE
3
0I
2 .S
cii
-1000 -900 -800 -700 -600 -500 -400 -300 -200 -100 0
Binding enurgy. *V
5C
S C
IS
-1000 -900 -600 -700 -600 -500 -400 -300 -200 -100 0
O Binding energy. eV
Figure 6-39. XPS Spectra of FM-300 Ply Surfaces Separatedby Hand in the Laboratory
6-47
SECTION 7
Different mechanical tests are used to create fracture surfaces in simple test coupons. The six
mechanical tests used are:
1, Interlaminar Mode I Tension (DCB)
2. Interlaminar Mode II Shear (ENF)
3. Translaminar Tension (N4PTT)
4. Translaminar Compression (N4PTC)
5. Rail-Shear
6. Compression After Impact CAI)
Interlaminar Mode I tension fractures are produced using a double cantilever beam specimen
geometry as shown in Figure 7-1. In this test, interlaminar tension conditions are generated at the
specimen midplane by deflecting two beam halves at one end of the specimen. This is made possible by
a release film (FSP) insert. The specimen configuration is illustrated in Figure 7-2. Special fixtures are
used for the specimen to allow free-pin rotation at the beam end and mechanical grip attachment. The
specimen grip fixtures, which are of a triangle grip configuration, are wedged into the crack tip formed
by the FEP insert as shown in Figure 7-3. This results in mouth opening displacement at the beam end
prior to mechanical testing. Any precrack observed during clamping is marked on the specimen edge,
The specimen is loaded under deflection control on a Mechanical Testing System servohydraulic load
frame with the rate of cross-head deflection adjusted during the test to produce a relatively constant
rate of crack extension. In general, crack extension occurs in a progressive manner, with observed crack
growth rates of about 1.27 to 2.54 cm (0.5 to 1.0 in.)/min. Cross-head deflection rates range from 0.25
to 0.51 cm (0.1 to 0.2 in.)/min.
Interlaminar Mode II shear fractures are produced using a modifieu end-notched flexural
specimen geometry as shown in Figure 7-4. Tests are carried out using a cantilever geometry fixture.
The uncracked end cannot rotate or move vertically, but can move horizontally. Such movement
prevents the introduction of extraneous (vertical) loads as the beam shortens under deflection. With
this geometry the top surface is loaded in pure compression, while the bottom surface is in pure tension
"resultingin pure shear at the crack tip.
7-1
Figure 7-1. Double-4antifever Beam Specimen Geometry
0.0.5 in
7-2
GRiPS
C'rack propaga;ion during testing typically occurs by rapid extension of the crack along the
specimen midplane. In general, such rapid growth extends over 60 percent of the test span during the
first crack extension, and then is followed by several increments of slower, more observable growth.
Rapid crack growth precludes the full monitoring of the direction of cracking. However, (luring those
7-3
periods of slow, stable growth, crack extension is observed in a direction away from the FEP crack
starter region, and toward the cantilever beam support. fixture.
The conditions for this test are similar to the translaminar tension except the tram--laminar
compression is produced by four-point beam loading a Chevron-notched specimen with the notch
located transverse to the beam's upper compression surface as shown in Figure 7-6.
7-4
Figure7-6. ANotched-Bend-Bar (Compression)Specimen Geometry
7-5
COMPRESSION ON LOAD
1-00
90
0, 450
STRAIN f
ROSET'-E 00
LOCATED
CENTRALLY ON FIBER DIRECTION O--
SPECIMEN [2
2 6.00
-I
-
2
11
2.00
0.50 DIA
HOLE (TYP)
0.50 R
0.060 0.125---1]"
0.60 0.60 THICK IN
SYMMETRICAL
1.75 LAY-UP
3.00
COMPRESSION ON LOAD
NOTE: ALL DIMENSIONS IN INCHES
532 08
7.6
O LOAD
SPHERICAL HEAD
DIA SPECIMEN
BOLTS A EV- RAIL GUIDES
1
S~ADHESIVE
'1
FRICTIONLESS -- SPECIMEN
SURFACE - RAILS
BA RS _ -- ARAIL GUID ES
BAR GUIDES--_0
SPHERICAL BASE
LOAD
532 09
7-7
-TEST SPECIMEN
RUBBER TIP
g GUIDING PIN
3 in X 5 in CUT-OUT
".. ,LAMP
SUPPORT BASE
7-8
0
SECTIOf 8
FRACTOGRAPHY
Fractography the science of studying fracture surfaces, is the key not only to the determination
of the sequence of fracture events that took place during tLe failure process but also for the
identification of the state of stress that existed at the time of fracture (tension, com;iression, and
shear). Other factors such as environment, mateiial defects, or other material anomalies that may have
contributed to crack initiation, growth, and ultimate failure can also be evaluated using the tools
presented in the following paragraphs.
Due to their laminated construction and high level ef anisotropy, particularly on a localized
scale, failures in composite materials tend to be extremely complex in appearance as well as in
mechanism. The actual failure process and fundamental fracture mechanisms in this class of materials
are not sufficiently well understood at this time. However, many of the analytical methods presently
used for fractography of composite structures have been previously developed for use in failure analysis
of metallic materials, and therefore lave a fairly sound fundamental basis. Efforts within the past 10 or
so years in the area of composite fractography have been directed toward modifying these well
developed metals techniques, which include:
1. Visual and optical macroscopy
2. Optical microscopy
3. Scanning electron microscopy
4. Transmission electron microscopy.
Each of these analytica! techniques provide the investigator with a significant amount of
information regarding the fracture by examination of the morphology and other topographical features
related to the fracture process. As a general rule, all failure analyses should involve each of the above
techniques (with the exception of TEM) in the order of increasing magnification. Each investigation
requires a combination of the above methods, dependent on variables such as size, time available, and
type of fracture. Fractography, as an investigative post-failure analysis method, involves the use of the
various techniques listed above in conjunction with the interplreation of the results as related to the
fracture process. These paragraphs describe in detail the techniques used to identify the morphological
fracture features. The physical principles involved in each technique are presented, including comments
on how these principles can be used to control image type and the type of features observed in 'he
fractographs. Specific examples of how to obtain suitable fractographs and gather data related to
propagation type (interlaminar versus translaminar), crack propagation mode (tension versus shear),
crack propagation relative to the fibers, and other divisions are illustrated.
In Section 4, the interpretive methods and specific examples of how the fractographic data
I relates to fracture mode, crack growth, and other faults are presented.
8-1
In addition to identifying the fracture morphology, the other primary rcsponsibility of the
investigator is to document the key features relevant to the determination of the cause of failure. This
involves photo documentation for reports or presentations, as well as later analyses, and can often
require extra effort during an investigation so that the photos convey the message to nontechnical
personnel. For example, a significant portion of the analyses can be performe(1 using optical microscopy.
and although the image can be evaluated during use of the instrument, the details of resin fracture are
often too fine and their fracture topography too rough (at high magnification) to be documented on
optical photomicrographs. In this case, the SEM is required to document these features, although it
may not be required to analyze resin fracture details. The two techniques are complimentary to one
another. The value and methodology of both photo documentation and the identification of the fiacture
morphology are presented and considerations relative to specimen preparation prior to analyses are
provided.
In fractographic analyses, primary emphasis is placed on:
1. Locating the origin of the failure
Determining the direction of crack propagation is one of the most significant concerns in
fractography. The recommended iechnique for crack mapping uses the lowest magnification capable of
performing the job. This recommendation is made because one of the fundamental problems in detailed
microscopy of large fractures is that there is an extremely limited perspect.ive on how the area being
examined relates to the part as a whole. The situation is similar to the o0l adage, "one can't see the
forest for the trees." With a limited perspective, it is often possible to improperly characterize the
direction, mode, or load state at fracture. By eixxphasizing the use of lower magnifications for eariy
investigations, the FALN imposes a sense of perspective on later, higher magnification inspections.
The chemical analysis of fiature surface features may be required to determine the sequence of
fracture events. By carrying out these examinations after identifying the fracture origin, emphasis can
be placed on any anomalies encountered. Table 8-1 and Table 8-2 summarize the techniques used in
fractography.
8.-2
FRACTURED
SPECIMEN
O O
U ~ N R A CT U RE fS
0
I - R O 0G
A . N
t Mic;oscopyASNT([
MacroscopPROPAGATION>O
CO~rAMINANROPCONATIIO
ANNOMYLYS
Qp0~ Opicalomcrocop
0* SEM
~ ~
0micrsroyoi-----e ~ naye
Optca
0O EE N
ANOMALIE
Table 8-.1. Failure Analysis Techniques - F-ractography
0 Detemination of origin
0 identification and characterization of
select conditions
Slack -scatter Microscopy performed by lnwermfimiat* magnification 0 Documentation as a function of atomic
electron imaging back .refillcted of fra cturewsrface and number
rnicoscopy primary beam electrons defect *ack-reflectad
g~aneratntd by a rastered electrons are sensitive to a identification and characterization of
eltictron beam atomic number and can be defect conditions
used to distinguish surf-ice
dletails as a function of
atomic number
The following paragraphs describe the failure analysis methodology associated with fractoog-
raphy, the decision points, and the use of most of the analytical tools, all combined with the background
reasoning that led to the sequencing and structure of the fractograpliy FALN.
Beginning with a fitiled structural component, the initial responsibility of the investigator is to
document the fracture surfaces, both by extensive photodocumentation andl by visual and optical
rnacroscopy. At this important stage., while the entire part. or structure is intact, significant critical
information may be gained through assessment of the overall types of fracture, their locations, and
obvious anomalous production or service conditions, all oF' which give an investigator a feeling of the
primary loading events, sequence, and other contributory c'ond~itionls that occiurred p~rior to or during
8-4
Table 8-2. FaIlureAnalysis Techniques - FractureSurface Material and Chemical Characterization
Secondary ion mass Qualitative surface chemical Determines elemental and 0 identification of contaminant
spectroscopy (SIMS) analysis of surfece monolayer chemical state of surface monolayers
atoms removed by ion monolayers; repeated
spu"tering operation allows elements
to be profiled versus depth
failure. Dependent upon the size and location of the component, further fractographic analysis may be
delayed for nondestructive analysis such as hand-heid pulse-echo ultrasonics and part breakdown
(fastener removal) into separate components such as skin and spar. As the separated pieces become
available during breakdown, the investigator should identify, define, and document the individual
fracture types such as translaminar or interlaminar. Visual and optical macroscopy should be employed
for these analyses.
Following definition of fracture types, and more detailed NDE examinations of the damage
regions, the most labor intensive and critical portion of the fractographic examinations involve the use
of detailed macroscopic and microscopic analysis. These analyses require the investigator to use
specialized fractographic techniques to determine the fracture directions, modes, and environmental
conditions at fracture. For translaminar fractures, the majority of the analyses involve fiber fracture
details, while for interlaminar fractures, resin fracture analyses predominate. Emphasis should be
placed upon performing as much examination as possible at the lower magnification ranges, thereby
increasing confidence and reliabirity in the determination of crack directions, modes, and any
anomalous conditions which may exist.
If determination of the fracture origin is made, through the use of detailed crack mapping or
visual and macroscopic aralyses, the next step is to assess whether or not an anomalous condition
exists in the origin zone. If such a defect is evident or suspected, such as adhesive separation which has
a general lack of fracture features, commonly associated with contamination, then further examinations
like surface chemical analysis may be required.
8-5
8.1 CARE, HANDLING, AND PROTECTION OF FRACTURE SURFACES
The preservation of the physical evidence should be viewed as one of the most important
responsibilities for the investigator. Education on proper handling and protection prior to any
fractographic examination is strongly recommended for anyone dealing in fractures either in the field or
in the laboratory. When a fracture occurs, and there is even a slight chance that it will be submitted for
laboratory examination, several important steps must be followed so that maximum information can be
obtained. The important factor is prevention of damage that might preclude the use of various
analytical methods. Most of the procedures described are independent of the particular fractographic
techniques being employed.
The care and handling procedures that must be followed are based on general methods that
have been proven for metals and modified for application to composite structures. These procedures
rely on the fact that the fracture surface contains an enormous amount of valuable information and
that anything done to obliterate or alter this information may obstrnct important information related
to the fracture. The damage that can occur can be separated into two types, either mechanical or
chemical.
This type may arise from several sources, including allowing the fracture to impact other
obj.ct.s. This can occur during the fracture process itself (for example, departure from the aircraft and
subsequent ground impact), during removal of a fractured portion from the remainder of the structure,
during transportation to the laboratory, or by careless accidents such as dropping the romponent. P~or
composites, the consequences of damage can be detrimental to the pror.,:r determination of the cause of
failure, since quite often the post-failure damage cannot be differentiated from that actually created
during the failure event. Notably different than metallic structures, fractures created during rapid
loading, such as impact from dropping, often cannot be differentiated from fractures created under
slower loading which might be encountered during part service. Crack growth created (luring handling
such as peeling delamination surfaces apart by hand are especially undesirable as they are often
impossible to differentiate from delaminations created during the fracture event.
The fracture can usually be protected during shipment by properly crating the structure, if it is
large and intact, so that motion of the component will not create rubbing or translation of the mating
fracture surfaces. Where possible, particularly for components that have translaminar fractures, the
amount of motion-induced damage can be eliminated by total separation of the fracture halves prior to
shipment. Parts with only delamination damage can usually be left intact for shipment. To prevent
contamination or fluid ingression such as rain, the structure should be wrapped with a clean plastic
sheet. Smaller components or test coupons should be placed in zip lock bags or other airtight bagging to
maintain the level of absorbed moisture and more completely protect the structuie from contaminants.
If necessary, the fracture can be protected during shipment by taping cloth or cotton to the surface, as
long as loosely adhering material, commonly associated with the critical fracture regions, is not
dislodged.
Tonching the fracture with the fingers or rubbing it should definitely be avoided, as this can
leave oils or other contaminants on the fracture surface which may alter or prevent surface chemical
8-6
* analysis results. One of the most common sources of fracture surface damage is when two halves of a
fracture are fit together. This accomplishes nothing important and always results in some microscopic
damage to the fracture surface, and thus should be avoided.
This type of damage to fractures can be prevented in a number of ways, each using common
sense and requiring an awareness of the necessity of protecting the surface from chemical
contamination or degradation. Since the identification of a foreign material (such as release agent)
present on a fracture surface may be critical in the overall interpretation of the cause of fracture, all
chemical protection schemes such as plastic coatings or desiccants are to be avoided. If such a condition
exists where a substance must be cleaned from the part prior to shipment, such as fire retardant foams,
the most desirable solvent is plenty of clean water.
Usually some sort of cleaning of the fracture surface is required prior to examiniation, since
small pieces of fibers and matrix are present due to either the fracture process or mechanical cutting
operations. However, to prevent potential loss of contamination evidence, cleaning should be avoided
early in the investigations if possible. Numerous cleaning procedures have been attempted, each suited
for the extent and type of deposit, or the preference of a particular laboratory. The most commonly
employed cleaning techniques, in order of least-to-most damage Qr destruction of evidence, are:
1. Dry air blast: This method will remove many loosely adhering materials. If possible, canned air
for laboratory use is preferred over the piped-in compressed air which can often contain oils
and moisture. A soft artist's brush will assist in the removal, but extreme care must be
exercised to see that no damage is done to the fracture surface.
2. Detergent wash: This method can effectively clean most specimens for high magnification
investigations. A large, soft artist's brush can be used to apply a mild, water-base detergent
(such as dishwashing detergent), followed by a thorough rinse in lukewarm tap water and
dryirg by air blast. Stubborn deposits of particulate dust, which can occur on cyclically loaded
parts, can be removed by ultrasonic agitation while the detergent solution is present.
Ultrasonic durations of only a few seconds are usually satisfactory although up to 15 minutes
have been required to remove physically imbedded dust particles.
3. Chemical solvents: Chemical solvents should be avoided and used only when other techniques
do not remove the tenacious deposits or chemical coatings. Prior to using chemical solvents on
critical fracture surface regions, these solvents such as acetone, methyl ethyl ketone (MEK) or
alcohols should be trial tested on fracture surfaces of the same material to check for
degradation or loss of morphological details. Soak durations should be kept to a minimum.
These solvents may also be used with ultrasonic agitation, which may shorten the soak
duration quite significantly. This should be followed by detergent wash, rinse, and dry as above.
8-7
8.2 PHOTO MACROGRAPHY
The 35mm single lens reflex (SLR) cameras offer ease of use with small size and, as stated
above, offer a particular advantage for field work and color photography. The 35mm SLR provides a
quick, "see what you get" feature. With the closeup "macro" lenses, magnifications less than 1X can be
obtained. The major disadvantages are that they offer a small viewing port which provides a limited
assessment of the lighting and focus, the small negative may not always be adequate for enlargement.;
with high speed/large grain film, and the tilt and swing functions of the lens and film planes are not
available as in the view cameras.
Low magnification stereo macroscopes with attached cameras are often required to provide a
light optical view and documentation of the fracture features at magnification ranges beyond those
8-8
* possible for the 35mm and view cameras. These systems also permit three dimensional viewing at
magnifications up to and beyond the lower limits offered by the light optical microscopes. The practical
ranges of magnification for these systems range from 5X to 80X. These systems often have the
additional advantage that detailed macroscopic examinations of the fracture surface details and origin
regions can be performed, while allowing documentation of the features identified.
Optical microscopy has proven itself as a most critical tool for failure analyses, for the
examination of both fracture surfaces and cross sections. For cross sections, the optical microscope
remains the most important technique available. The analyses of cross sections provide ir.sight into the
microstructural features related to construction, crack propagation, and defect conditions. For fracture
surfaces, particularly the delamination surfaces, the optical microscope is possibly the best technique,
and at the least should be used in conjunction with the SEM and TEM, rather than as a substitute. The
fractography of delaminations by optical microscopy can provide information regarding the crack
growth direction, loading conditions at fracture, the origin locations, and anomalous conditions related
to the origin. Undoubtedly this information is considered paramount to the determination of the cause
and sequence of failure and thus should be required for all investigations of delamination surfaces.
Optical microscopes are available from a wide variety of sources and range considerably it cost
and capability. Reflected light is the illumination mode used for composites. These microscopes are
classified as upright or inverted, relative to the location of the stage with respect to the objective lenses.
The upright microscope is the preferred type for examination of fracture surfaces, as the fractures
surface does not contact the stage which can potentially damage the surface. The inverted stage,
however, can accept an extremely large specimen, whereas the upright stage is limited to approximately
6 inches by 6 inches maximum size.
The bench type microscope is the least expensive type and often can provide all of the
capabilities required to perform the investigations during fractographic analyses of composites. Various
metallographs, although usually more expensive, provide more flexibility and resolution. These can
range from simple to full-scale research units, with assorted illumination modes, objectives, light.
sources, hot stages, and other features.
8.3.2 Illumination
A variety of light sources for optical microscopy are available. The low voltage tungsten
filament lamp is most often used on bench microscopes and has adequate light intensity for most
observation and photography requirements. Where more light intensity is required for photo
documentation, xenon arc sources (intens:ty adjusted by neutral density filters) and tungsten halogen
filament lamps (adjusted by filters or electrical current) are the most common.
8-9
8.3.3 Diaphragms
Two diaphragms are available within all systems to provide improved image quality. A field
diaphragm is placed near the light source to minimize internal reflections and glare within the
microscope. This diaphragm is stopped down to the edge of the field of view, while not impairing
examination or photography. A second diaphragm, the aperture, is placed in the light path just before
the vertical illuminator. Opening or closing this diaphragm alters the amount and the cone shape of
light, varying the contrast, sharpness, and depth of field. As magnification is increased, the aperture
should be stopped down. At a given magnification, closing the aperture increases contrast and depth of
field while reducing the image sharpness. A general rule of thumb for aperture setting on rough fracture
surfaces, which require maximum available depth of field, is to stop down the aperture until there is a
noticeable decrease in image quality and then open it slightly to eliminate most of the aberration.
This forms the primary image and is therefore the most important component of the optical
microscop)e. The apochromatic and piano type objectives prodide the highest degree of correction for
aberrations, produce the best results, and reduce eye strain. There are long-distance-working objectives
which are particularly valuable for examination of fracture surfaces with rough topography, and are
usually only necessary for objectives in the upper ranges of magnification such as 20 to 60X. Other
desirable features for fractographic examinations are parfocal lenses (maintains focus when objectives
are changed) and spring loaded lenses (moves when contacted with specimen to reduce damage to
lenses).
Reflected light illumination mode requires a relatively flat surface due to depth of field
limitations and therefore specimen preparation is important to provide the best situation for
examination and documentation. Specimen preparation for cross sections are presented earlier, with
descriptions of the steps for cutting, mounting, and polishing. Specimen preparation for fracture surface
examinations involve cutting the desired fracture region from the remaining structure followed by
cleaning.
Optical fractography is by far the most efficient and cost-effective method for examination of
delamination surfaces. Since the specimen setup and examination times are very short, an enormous
amount of fracture surface is covered with this technique. As a result, a reliable and accurate
determination of the typical and representative fracture surface features is obtained in a relatively
quick fashion. Translaminar fractures, on the other hand, are too rough and the features (fiber ends)
are too fine for optical microscopy Low magnification inspections can be used on delaminations to
verify the plane of fracture and the location of crack growth features such as beach marks and
transverse cracking. More detailed, higher magnification inspections provide resolution of the fine
matrix resin fracture features. These features are used to identify the direction of crack growth (by
evaluating river marks), the fracture mode (tension versus shear), and indications of contamination and
8-10
* environmental extremes. Bright Feld illumination, a stopped-down aperture, and long working distance
objective lenses provide the best combination for examination at high magnifications required for crack
mapping and identification of the fracture modes. Crack mapping methods and interpretation of the
fracture features are presented in Part 2, Volume II, Section 1. The features found with the optical
microscope, even though they are visible through the eyepiece, are often too small to document with
photographic film. In these situations where photomicrographs are desired, the SEM is required.
3. Origin examinations
4. Inspections of interfacial conditions
Several illumination methods are available for cross-section analyses, with bright-field being the
most widely used. Dark-field or oblique illumination provides an excellent image contrast for
differentiating surfac topographical features such as microracks and phase interfaces. Polarized light
can be used to enhance differences between ply orientations for easier ply count and orientation
analyses.
Since its relatively recent origin, SEM has found a wide range of applications in failure analysis,
materials research and development, and quality control. Fractography is probably the most popular
application of the SEM. The three dimensional appearance of SEM fractographs, the large depth of
focus, large magnification range, and simple specimen preparation with direct inspection make the SEM
a versatile and indispensable tool in failure studies and fracture mechanism research. This unique
instrument offers possibilities for image formation of fractured parts that are usually easy to interpret
and provides clear photomicrographs of rough surfaces as well as polished cross sections. The
development of an assortment of related capabilities such as stereo viewing, quantitative microchemical
* analyses, in situ fracture studies and image formation that is easy to interpret all contribute to the
value of this investigative and research tool. Energy dispersive X-ray (EDX) analysis equipment is
8-11
routinely attached to the SEM, providing semiquantitative and, in favorable situations, quantitative
analysis of composition from a small volume. For composites, EDX analysis is usually only required for
contamination analysis.
The SEM is capable of magnifications from about 5X to 250,OOOX, although the majority of
composite fractogr-aphy analyses do not go beyond 20,OOOX. SEM is normally a resolution of
approximately 100 angstroms. The depth of field is about 300 times that of the light optical microscope,
providing an excellent three-dimensional view of the specimen at focal depths of over 1000 microns at
1O0X and 5 microns at 20,OOOX. Specimens can be tilted up to approximately 70 degrees to the incident
beam, while maintaining focus over most of the surface. The working depth ranges from 8mm to about
25mm, allowing an extremely rough surface (such as with protruding fibers) to be examined. The
specimen size is usually limited by the size of the chamber door. The maximum size for the latest SEM
equipment is approximately 6 inches by 6 inches, with limitations in tilt and thickness at this size.
Usually specimens are much smaller (0.5 to 1.0 inch square), so that thicker specimens and maximum
tilting may be allowed. Since the specimen size is limited, very large specimens must be partially
destroyed. This particular limitation indicates that lower magnifications and less destructive inspection
methods such as optical macroscopy and microscopy should be employed prior to SEM analyses.
Proper specimen preparation is basically simple and requires only a small amount of time. The
preparation sequence is usually as follows:
1. Specimen selection: Commonly visual or optical inspections of the fracture surfaces to define
areas of interest for analyses or documentation
2. Specimen cutting/removal: Involves mechanical methods such as abrasive cutoff machines or
hand saws
3. Specimen cleaning: Used to remove dust particulate (from the fracture and cutting processes)
and oils, etc., as discussed earlier; cleaning and coating should not be performed if
contamination is a concern
4. Specimen mounting. Involves adhering the specimen to a mounting stub to secure it and to
provide a conductive path from the specimer to the SEM chamber
5. Specimen coating- Usually required for nonconducting materials such as composites when high
KeVs are used.
Good imaging requires careful specimen preparation, such that electrical charging, electron
beam damage, and outgasing of volatiles are minimized. Composites, due to their laminated
construction and organic structure, are subject to all of these problems if the proper preparation
procedures are not followed. Because of this, the latter two preparation steps (mounting and coating)
are discussed in detail below.
To prevent charging, conductive adhesives such as silver, carbon, or aluminum containing glues
are used to bond the specimen to the stub. Drying times are on the order of 30 minutes, although time
can be significantly reduced with the aid of a warm air blower. In order to maintain proper specimene
identification, labeling the stub with a scribe is often recommended.
8-12
Nonconducting materials such as composites are usually coated to prevent electron charging,
caused by a buildup of a space charge region of absorbed electrons. This charging appears dark at the
site of absorbed electrons and deflects the incident beam, leading to image distortion and significantly
changes the emission of secondary electrons so that the surrounding region appears washed out or
over-bright. Charging can be prevented by operating at low acceleration voltages (less than 5 KeV) or
applying conductive coatings. The latter method is preferred by most laboratories, since the
magnification ranges do not ever approach conditions where the coating would affect microstructural
resolution. It should be noted that non-coated low KeV analyses offer shorter preparation times and do
not cover or obstruct surface contaminations, but do usually require an image enhancement system to
properly process the limited image output.
The coating layer must be thin as feasible to minimize the possibility of obscuring fine details
and thick enough to provide a conductive path for the impinging electrons. The minimum thickness is
dependent upon the general roughness of the specimen surface, and may range from 10 nanometers for
relatively flat specimens to 50 nanometers for extremely rough topographies. Carbon, copper
aluminum, platinum, palladium, silver, gold, or gold-palladium (Au-Pd) are applied by high vacuum
vapor deposition or sputtering. The gold-palladium sputtered is the preferred method, since sputtered
layers provide better adhesion and more even condensation, and it provides the finest microstructure
and lowest possibility of obscuring fine details. A combination that seems to work well for most
fractures is to sputter coat approximately 20 nm of Au-Pd using the DC-pulse mode for five minutes
after backfilling the vacuum chamber to 4.0 to 7.0 Pascal (30 to 50 millitorr) with argon.
It should be noted that a nondestructive method for SEM examination is available for
0SIuations where destructive cutting examination is not possible or the fracture surface may not be
removable from the structure. In these situations, ? two stage acetate replica may be produced with the
same techniques described for TEM analyses. The only difference is that the specimen must be made
conductive, and therefore requires an additional layer of vapor deposited gold or similar material.
The basic features of the SEM are presented in Figure 8-2. This instrument is a combination of
electron-optical, vacuum, and electronic control devices for impinging a narrow beam of electrons from
a heated cathode and focused by a system of magnets to a pinpoint spot on the surface of the specimen.
An image is gained from collecting, modifying, and displaying the resulting emission from the
specimen's surface. The cathode, or filament, is usually tungsten. Accelera -n voltages range from
1,000 to 50,000 volts, although the usual range for composites are 2 to 30 KeV A more effective and
longer life electron source is lanthanum hexaboride (LaB 6). This new cathode requires more care and
warm-up time, a better vacuum, and higher initial cost. To generate the required vacuum, a diffusion
pump or a turbomolecular pump is used.
The electron beam scans the specimen similarly to the way a cathode ray tube is used to image
a scanning raster on a television screen. A scanning generator controls the current to the scanning coils,
which, in turn, deflect the incident beam along closely spaced lines. The magnification is controlled by
_0
8-13
L Gun
0~ Condenser 1
0 ~ Condanser 2
d Scan coils
SObjective
Secondary
electron detector
varying the current in the deflection coils. The electron beam impacts the specimen surface, and the
electrons that return from the specimen surface are collected by a detector. Amplification of this
electron signal is required for inmaging the scanning cathode ray tube.
Whten the primary electron beam is impinged upon the specimen, electrons and other radiations
are emitted from the surface that can be used to produce images and to microchemically determine the
elements present. Figure 8-3 shows the types of emissions radiated from the surface and the
relationship to the type of detection modes available. The most common detection modes for composite
materials are: (1) the secondary electron (SE) mode, (2) the backscatter electron (BSE) mode, (3) the
pseudo backscatter electron (PBE) mode, and (4) X-ray spectroscopy These are used routinely with a
sound understanding of the types of information available and the limitations of each detection mode.
Each of these detection techniques is described below except X-ray spectroscopy, which is presented in
Section 6.
Although both secondary and backscatter electrons are used for fractography, the secondary
electrons are usually the preferred source since they offer better resolution, provide an abundaiit signal,
permit viewing areas of the specimen that are not in direct line with the detector, and provide a better
three dimensional effect due to the shadowing loss of electrons at topographical features. There are
situations where it is necessary to sacrifice resolution for improved image contrast and differentiation
of features or phases by their atomic number. In such a case, backscatter electrons provide image
contrast in specimens which are especially smooth or are viewed at low magnifications.
8-14
Incident / - Incident
electron electron beam
beam Zone of
electron Specimen
Auger e- emission -surface
excited XR X-ray
fluorescence detector _ Zone of
Z
-Specimen characteristic
X-ray generation
(a) Useful Signals Generated by an Electron Beam (b) Pattern of Zones in a Specimen That Are
Sources of Signals Generated by an
Incident Electron Beam
Secondary electrons are produced by interaction of the primary electrons with the atoms in the
first 1 to 10 nm, resulting in emission of the loosely bound atomic electrons. The energy spectrum of
these secondary electrons is independent of the energy of the incident beam and material type. The
energy spectrum is fairly limited in range, with a pronounced maximum at approximately 3 eV. As a
general rule, all electrons below 5u eV are considered to he secondary electrons and electrons with inuret
energy are in the backscatter range. The primary factor for secondary yield is due to topographical
features such as small variations in the surface angle. Since the secondary are emitted from the top few
nanometers, the envelope of the excited and emitted electron volume moves closer to the surface when
the beam contacts the surface at an angle, thus increasing secondary electron yield, and therefore
brightness on the visual image.
As shown in Figure 8-3, the backscatter electrons are produced by single large-angle or multiple
small-angle elastic scattering of the primary electrons as they impact the atoms from 0.1 to 1.0 jim
beneath the surface of the specimen. Different than secondary electrons, the energy distribution of the
backscatter electrons depend directly on the energy of the primary electrons and the atomic number of
the material. In a similar manner, although with a more pronounced effect, surface inclination of the
specimen provides an accentuation of the topographical features of the specimen, In specimens with
high atomic numbers, which have larger atomic sizes, a larger percentage of the electrons are
backscattered from atoms closer to the surface, with little change in energy. Therefore, the yield and
thus the brightness are increased with materials that have higher atomic numbers. For those specimens
which have a very smooth topography, the use of pseudo-backscatter electron is employed. This involves
using the secondary electron detector with gating of the alc #ed electron energies for those electrons
with more than 50 eV This detection mode does not "see" the secondary electrons that have energies
around 3 eV. Figure 8-4 illustrates the use of secondary, backscatter, and pseudo-backscatter for the
imaging of a composite fracture surface.
Optimization of imaging can enhance the image for analysis and documentation. Some
components of a scanning electron microscope have their own characteristics of resolution and noise,
which determine the image quality, however, most instrument parameters are fixed by the designer to
8-15
Backscatter electron detector image Note
increased contrast but lack of general feeling of
3-.1
Object tilt probably provides the greatest effect on the overall image and thus should be
optimized. Specimen tilt, which causes the beam penetration depth and scattering to vary, results in
contrast between topographical features, similar to side lighting in optical macrosccpy or shadowing for
the TEM. The effect of tilting is more pronounced for the backscatter mode than the secondary
emission mode. Since tilting can be used to enhance subtle features on smooth fractures, the use of high
tilts is often required for examination of fatigue features, as shown in Figure 8-5, where fatigue
striations are not visible until tilts beyond 30 degrees are employed. Similarly, the use of tilt often
provides a quite different perspective of the fracture morphology such as hackles, as shown in
Figure 8-6.
The aperture size and working distance determine the depth of focus. For rough surfaces at low
magnifications, a small aperture and large working distance are selected. For high magnifications, a
short working distance, small aperture, and high lens currents iutusbe used Lu iiiimize the spot size.
Interlaminar Frictures. The basic fracture types and the general SEM instrument optimization
for each type are presented belcw-.
1. Crack mode and direction of propagation: Magnification in the range of 400X to 2,OO0X is used
to inspect the direction of river marks, resin microflow, and tilt of hackles. For most
observations, a tilt of approximately 30 degrees in the SE detection mode appears to be best.
2. Fatigue: Where this crack growth mode is suspected, high tilts beyond 60 degrees are often
required, with viewing angle perpendicular to the striation. For instance, Mode II shear
striations usually appear w;thin the fiber matrix separation region, and therefore the viewing
angle should be running parallel to the direction of fibers in a tape material. Pseudo backscatter
and backscatter detection modes have been used successfully to identify and provide
accentuated contrast to striations that are not easily detected using the seconda'y electron
detector.
3. Disbond or contamination: For areas suspected of contamination or regions suspected of
having a thin coating with different atomic number, the use of backscatter is invaluable.
Usually a tilt is not required for this detection method.
Translamlnar Fractures. For these fractures which have a very rough topography, the use of
small apertures is desirable to increase the depth of field. Increasing the working distance, also provides
increased depth of field, although the resolution is degraded at extremely long distances and higher
magnifications. Higher accelerating voltages do not provide any additional benefit since the secondary
8-17
65-dog tilt 5000X
A F L A . a a
'A F' C- S FF
8-19
electron emission is independent of the beam energy. For most of these investigations, the magniflhation
range is from 5X (for macroscopic view to differentiate between compression or tension) to 2,5( .X (to
identify the individual fiber end fracture morphology). Translaminar fractures provide an en( mous
amount of contrast and topography changes. Accordingly, the SE detection mode is most useful since it
is least sensitive to roughness extremes.
The basic specimen used for TEM analyses is the two-stage replica. Direct examination of the
fracture surface can be made with the other techniques discussed above, but replicas offer the unique 0
capability of transposing topographical information of composite delaminations to a high-fidelity
facsimile that can be conveniently handled and transported, and readily examined in the TEM, SEM, or
light optical microscope. The use of replication has significant value when the fracture surface of the
failed part cannot be transported to the laboratory. Additional value is the ability to generate several
replicas of the fracture surface, all of which may be destructively analyzed to determine the physical
features relative to the determination of the cause of failure.
The initial cleaning of the composite fracture surface was covered in the beginning of Section 8.
This usually involves detergent wash with ultrasonic agitation to displace particulate that may be
physically impacted into the fracture surfaces during fatigue cyclic loading. The final stage of cleaning
usually consists of successively applying and stripping several acetate tape replicas. The thin acetate
tape is first wetted with a small amount of acetate solution, allowed to soften, and then applied to the
fracture surface which also has some acetated solution applied to the area. Hand pressure is maintained
without moving for at least a full minute. Following full drying of the replica, it is carefully removed to
limit the damage from lifting fibers from the fracture surface. Usually two replicas are required to
properly clean the fracture surface.
For most composite laminates, the matrix material can withstand short periods of contact with
solvent, even those within their own functional chemical groups. This is particularly true for chemically
stable epoxy resins that are thermosetting and are highly cross-linked. For these resin systems, the use
of the most common replication materials, acetate film and acetone, do not exhibit damage of even the
8-20
* finest fracture details. Other matrix materials, particularly those that are thermoplastic and are not
cross-linked, should be spot tested in an area in the replication technique.
Following the generation of several clean replicas, they are then prepared in the same manner
that is used for metals, in which the plastic replica is first shadowed with a high Z (atomic number)
material such as germanium, and then coated with carbon. Figure 8-7 illustrates the two stage
replication technique. For fatigue striation evaluation and maximum enhancement of the subtle features
involved, the direction of germanium shadowing should be parallel to the direction of anticipated
fatigue crack growth.
Acetate
tape
Specimen
__ Acetate
- shadow
Carbon
Shadowing the Acetate shadow
TEM Image
These replicas are then cut to size, for fit to the copper grids, and then floated in an acetone
solution to remove the plastic portion from the replica. This stage is the most difficult and often very
frustrating. Due to the fiber reinforcements, the replicas have a pronounced tendency to curl along the
ridges formed by the fibers. Laminate tape fractures are extremely sensitive to this problem,
particularly when the fracture involves only unidirectional fiber orientation. Specialized methods can be
attempted, which include:
1. Dilute the acetone solution with distilled water to reduce the chemical mixingand curling, thus
reducing the possibility of tearing and breakup.
2. Provide more than one carbon coating, each at different angles, to provide a more complete
cove ing of the specimen. This increases the thickness which tends to reduce the curling.
3. WhVen the above methods fail, scribe the specimen in the direction perpendicular to the fiber
direction, creating a crosshatched effect, also reducing the chance of curling or breakup. The
crosshatching has been successfully performed using a hot blade, attached to the tip of a small
soldering gun, so that the damage is lessened by the smooth cutting action on the plastic film.
The cuts should be made very close together, as close as 0.01 to 0.02 inches, with very light
pressure to the coated side of the replica.
8-21
4. Place the small replica specimen between two copper grids. (There are butterfly types for this
purpose.) The plastic may be removed overnight by acetone va1por. (There are instruments
sold for this purpose.)
The TEM shown in Figure 8-8 contains an electron source, or filament, that emits a stream of
electrons into a vacuum chamber. The filament is held at high accelerating potential relative to the
grounded anode beneath it. The electrons pass through the hole located in the center of the anode and
then through a condenser lens which consists of a electrically variable magnet. The condenser lens
focuses the beam on the specimen. The electrons are either reflected, absorbed, or transmitted. The
electrons which are transmitted through the specimen are allowed to pass through three magnetic
lenses, forming a succession of three images, each magnified in turn to yield the intended overall
magnification. The range of magnification is from 200X to about 25,OOOX for two-stage replicas;
however the limit of use for composite laminate resin is about 10,OOOX.
The quality of the TEM image is affected iy magnification, image intensity, image contrast, and
the resolution obtainable. For increased resolution, higher magnification and better focus is necessary.
For increased contrast, the objective-lens aperture may be varied, lower KeV's can be used, or the
specimen can be tilted. Specimen tilt is usually the best method for identifying the faint striations
present in the resin fracture regicies. Figure 8-9 presents fiactogiaphs of a specimen which had beenr
subjected to a compression-compression fatigue cyclic loading. The faint striations are barely visible and
lie perpendicular to the localized crack growth directions. River marks and resin microflow are also
evident. The river marks and resin microflow features can be used to identify the localized direction so
that the tilting of the specimen can be made on the correct plane to maximize the enhancement of the
striations. Figure 8-10 illustrates the striations evident at the fiber/matrix interface region for crack lap
shear specimens (80 percent interlaminar Mode II shear). Tilts between 15 and 30 degrees were
required :"or creating enough contrast and striation enhancement for the fractograph examples
presented in this paragraph.
08
8-22
* LGun
IZ z Condenser 1
Condenser 2
Specimen
~ Objective
ZZ intermediate
Z~v2J Projector
/\
-/ Screen
Electron Microscope
Figure 8-8. Basic Features of the Transmission
0
8-23
- Resin fracture
microplanes
7050X
Fiber
. j Striations
Striations " "
j8-2
13,500X
Note: Striations are an indication of fatlg~e growth.
8-24
Fiber
Resin
Note the curved strlatlor's equally spaced along the fiber surface
Figure 8- 10. TEM Phoromicrograph of Shaioniii Featdutes furni a Cr1ackCA-L-C SIfear Specim'ren
8-25
SECTION 9
STRESS ANALYSIS
The purpose of stress analysis in the context of this compendium is to provide information
which may lead to defining the cause of failure damage in a composite structural component. Although
other methods of analysis may identify the origin, direction, r.nd mode of crack propagation, stress
analysis most often provides a quantitative explanation for the cause of failure initiation.
9-1
The stress analysis required to investigate part failure can be relatively complex, but from the
point of view of the failure analyst, the process involved can be divided into two discrete tasks, thus
making the review of available literature most straightforward. The two tasks are:
1. Assessment of the component strength in the as-fabricated condition
2. Assessment of component residual strength consideri:ig flaws, damage, and subcritical cracks.
Table 9-1 illustrates the relationship between the methods identified in the literature and these
two tasks. The first task, assessing the part strength in the as-fabricated condition, evaluates stresses at
a greater level of detail than during initial design. In most cases, this assessment focuses on the crack
origin, with special attention to discrepant manufacturing and production conditions. Principal
techniques at this level of assessment, likely to be of value, include individual-ply, point, and
average-failure criteria, laminate strength criteria, and the Damaged Zone Model.
The individual-ply failure criterion examines stresses at the individual ply level and establishes
the onset of first-ply failure. The chief benefit of this analysis method to failed structures is its ability to
predict the point of initial failure on a microscale. This ability is particularly significant since most
initial design analyses consider stresses on a gross average basis. However, even though there is
potential value in this methodology, some drawbacks exist. The most significant of these is that to
examine edge-effect stresses of individual-ply failure criterion requires a knowledge of plan view stress
distributions (surface stress flow) and detailed finite-element grids. For organizations with experts in
stress analysivs, neither of these factors is a limitation. More significantly, however, the individui-ply
failure criterion predicts failure initiation, not catastrophic failure. It also exhibits significant
inaccuracies when applied to predicting strengths of muitidirectional laminates. Consequently, failure
onset cannot be meaningfully determined based on the individual-ply failure criterion. As a result,
prediction of the stress required for catastrophic failure is more likely to be calculated by gross area
stress calculations with fine, finite-element grid structures placed in the area of failure origin.
Other considerations in estimated gross area stress are the point and the average-failure stress
criteria and the Damaged Zone Model. As described in the works of Daniel (Reference 7), Mikulas
(Reference 8), or Aronsson (Reference 9) on laminate failure criteria, the stress at failure can be
determined for any hole, given a characteristic parameter for the particular material and layup. Based
on results reported in these researchers' works, this method appears to work relatively well, its only
apparent constraints being that it requires empirical measurement of the characteristic parameter and
knowledge of the plan view stresses for the area of interest. In actual application, neither of these
constraints should be significant; however, additional concerns that must be addressed include the
effects of environment, absorbed moisture, and resin formulation on the validity of a single value used
as a characteristic material property
Table 9-1 depicts the second major task involved in the stress analysis of a failed component -
assessing its residual strength. Considerable investigation has been done that can be directly translated
to the analysis of through-thickness flaws such as cracks emanating from holes. Either the Damaged
Zone Model or point-stress failure criterion appear to work well for predicting the onset of crack
instability. The limit in applying either of these methodologies lies in the ability of disciplines such as
fractography to identify and define the size of original damage. Given this measurement, establishing
9-2
Table 9-1. Stress Analysis Methods
9-3
the point of stress necessary to cause failure requires only that values for the characteristic parameter
be known. The literature shows that these values have been measured for a variety of layups. However,
these characteristic material properties may vary with resin system and environment, making the
calculation of residual strength more difficult.
Methods of evaluating the criticality of interlaminar defects has been maturing rapidly in the
last five years. To accompany this, researchers have identified characteristic surface morphologies for
pure Mode I or Mode II crack growth. Crack growth directions under Mode I loading can be determined
quite readily from the surface microfeatures. These emerging technologies in the area of delamination
provide the failure analyst with useful tools for determining the cause of a structural failure.
In summary, reviewing the stress analysis literature revealed several techniques for assessing
the strength of failed components either as-fabricated, or with through-thickness damage or fracture.
These techniques, their value and their limitations are summarized in Table 9-1. Some of these
techniques are somewhat inaccurate, or may require measured properties not readily available; however,
at the very least, the techniques define the applicable methodologies and their attributes.
Sress Analysis Failure Analysis Logic Network. Defining the cause of component failure
requires the accurate understanding of the loads and stresses involved as a part functions. In metals,
this understanding is commonplace. Fracture mechanic calculations are usually carried out during
initial design stages as well as after component failure. Stress analyses are run after failure to review
iiiiatl design stresses in more detail, to evaluate configurational and material errors, and to determine
if damage (cracking) was incurred in service or during maintenance. The logical investigative sequence
for metals and composites is sirnilar. Figure 9-1 illustrates the detailed stress analysis FALN based upon
existing metals procedures and inputs from experts in the area of composite materials stress analysis.
Stress analysis of a failed component is carried out at three different levels: initial design
review, a structural level, and a lamina level.
At the initial design review, the analyst's objective is to verify initial design assumptions and
calculations with respect to the available service history and the location of fracture. This review
establishes what analyses were performed and their adequacy.
The next level of investigation encompasses the most critical stage. At the structural level,
inputs from the main FALN are incorporated, and detailed strains and stability at fracture origin area
are defined. Since most components are designed to gross average strain and stability criteria, the
informatiGn from this analysis usually provides adequate detail to understand the cause of failure. For
the analyst, investigations at this level may involve detailed finitc-element modeling, depending spon
the level of initial analyses.
Investigation of stress at the lamina level is unique to composite materials. Because of the
highly anisotropic nature of laminated composites, internal stresses can exist on a ply-by-ply level, and
individual ply stresses and failure predictions can be handled with varying degrees of detail. The
simplest level of investigation employs laminate theory combined with relatively simple failure criteria.
However, since individual ply failure does not usually constitute catastrophic fracture, additional
iterative analyses are required. These analyses can be performed with a variety of computer programs
9-4
INITIAL DESIGU REVIEW STRESS ANALYSIS: STRUCTURAL LEVEL
------------------------------- I--------
*------------------------------------------r----------------------
aINPUTS FROM MAIN ANALYSIS METHODS
CONSIDERED IN DESIGN
DOES YI EVALUTES
VERIFICATION
REQUIRED 7
* DEVELOP AND
PRFORM TESTS
---------------------------------------------------------------- ----------------
STRESS ANALYSIS: LAMINA LEVEL
I-.------------------------------------------------a
* DEEiNE LAMINATE ANALYSIS METHODS0:
YES Is NOI STRAINS ON I
MICROSCOPICJPLY * Lamninate theory
CONDITION LEVEL * Finite element (local
CRITICAL')j model of mater~ial)
* Micromechanics
* CRIERIAANALYSIS METHODS:
-------------------------------
9-5
(Figure 9-1); however, using such programs requires a detailed understanding of the methods used, their
accuracy, and most importantly, their limitations.
Stress analysis as depicted by the Stress Analysis FALN (see Table 9-2) is conlucted at one or
more of the following levels of complexity depending on the type of damage and economic
considerations: (1) initial , ign review, (2) a structural level analysis, and (3) a lamina or
microstructural level analysis. Throughout the investigation the failure analyst must communica'j with
the stress analyst to understand how the structure was loaded in the damage region (for example,
tension, compression, shear, flexure).
SOS Point stress analysis of AFFDL tech memo CDC 6600 Advanced composite analysis
laminated composites 'rBC-74-107 July
for: 1974
"* lhglane loads
"o Moments
"* Temperature
"* Transverse shear
STAGS Stress analysis and Lockheed COC 6600 Plotting of buckling mode shapes
panel stability document.
evalLations Structural
Anaiysis of
General Shells.
Volume 3. Dec
1975
VIPASA Natural frequencies of COSMIC file CDC 6600 Compression loaded stiffened graphite-epoxy
loaded structures ISCL panel%
(Vibration * Critical buckling Doc ID 00
and loads 17437 Jan 1973
instabilityof 0 Thin. flat NASA TMX.73914
plate assem- rectangular plates May 1976
blies 0 Thermal stress
including
shear and
anisotrophy)
During the initial design review, the stress analyst's objective is to verify initial design
assumptions and calculations with respect to the available loading history, the location of fracture, and
9-6
* environmental conditions. This review establishes what analyses were performed during original design
and evaluates their adequacy.
The next level of investigation encompasses the most critical analysis. At the structural level
th analyst's understanding of the loads and stresses derived from the initial design review will help
evaluate the significance of inputs or anomalies from the main FALN (for example, load types or errors
in number of plies or ply orientation). It may be determined that these defects significantly affect the
stresses or strengths -n the damaged region. In this case, parts of the initial stress analysis would be
repeated to assess the impact of the anomalies at the origin of failure. At the structural level, the stress
analyst employs finite elements and analytical models to compare the gross average strain to strength
critical strain and stability conditions. This information, incorporating the effect of anomalies, usually
provides adequate detail to understand the cause of failure. Redesigns at this level of analysis would be
focused towards reducing gross average design strain to account for anomalies discovered during the
initial design review.
If the cause of failure has not been adequately explained in terms of gross average strains,
stress analysis at the microstructurai or lamina level is required. This is often the case when failure may
have initiated at a design detail such as a hole, edge, or other stress concentrator. At the lamina level,
highly refined finite element meshes in the region of failui-e initiation are employed to determine
detailed three-dimen ional strain distributions. Loads or displacements applied to the finite element
meshes are derived from the stress analysis at the structural level. This detailed type of analysis,
'.though not often done during the initial design, provides information needed to evaluate interlaminar
* normal and shear stress concentrations at free edges. Fracture mechanics, coupled with finite element
analysis, have been successful at predicting the onset of interlaminar crack growth (see Reference 24
and Reference 25). Redesigns, resulting from lamina level studies, may be required to eliminate or
reduce the effect of design details that cause stress concentrations.
The thermoelastic properties of the anisotropic, or more specifically, orthotropic plies or lamina can
be predicted from the properties of the fiber and matrix constituents with micromechanics. Lamination
theory is then employed to calculate the thermoelastic properties of a group of lamina bonded together into
a laminate. References (13, 26, 27 and 28) provide lucid descriptions of the limitations and value of
lamination theory. Stress analysis of fiber reinforced polymers is quite different than that of metals for the
following reasons:
1. Lamina stiffness in the fiber direct ion is typically greater than 10 times the tiffness transverse
to the fibers.
2. Lamina strength in the fiber direction is generally greater than 30 times the strength
transverse to the fibers.
1
3. The differences in the stiffness coefficic nts between plies v.-' in a laminate cause interlaminar
stresses.
4. The differences in the hygrothermal expansion coefficients betweer, fiber and matrix within a
ply and between plies within a laminate may lead to significant residual stresses due to changes
in tcmperature or moisture content.
9-7
Lamination theory can be used to determine the strains and stresses in a composite structure
without considering interlaminar stresses. Finite element, finite difference and analytical methods
beyond the scope of this text have been used to evaluate interlaininar stresses. However, the failure
analyst must recognize that these stresses exist in composite structures and that the resistance to
interlaminar crack growth is hundrels of times less than the resistance to transply crack growth. This
result is expected since the fracture of fibers requires far greater energy than that needed to propagate
an interlaminar crack in which matrix fracture dominates.
The following paragraphs provide the failure analyst with an overview of the techniques used to
determine the stresses in laminated composites. These tools valy widely in their ease of application and
accuracy of results. In addition, it is intended to familiarize the reader with design details,
manufacturing and processing defects, and other considerations which must be applied to composite
failure analysis.
Paragraph 9.1.3 deals with predicting the strength of unnotched multi-directional laminates
with and without edge effects. Criteria for predicting failure of an individual ply are discussed and then
applied to predicting laminate strength. Paragraph 9.2 addresses the influence of ply thickness and
orientation on transverse cracking and delamination. Paragraph 9.3 describes some of the approaches
for predicting reductions in strength caused by inplane stress concentrators such as cutouts and notches
are described. Other strength reductions incorporated into design such as environmental effects and
impact damage are also discussed. Semi-empirical fracture mechanics and stress based approaches are
discussed with respect to their ease of use and generality of application. Paragraph 9.4 discusses, in
more detail, design details causing interlaminar stress concentrations (unique to composites), such as
holes and free edges. This is important because interlaminar stresses cause delamination to grow under
fatigue or static loading leading to significant reductions in compressive strength. Paragraph 9.5 is
designed to familiarize the reader with some of the extrinsic factors (fcr example manufacturing
defects) which may reduce the strength of laminated composites. The stress analyst would then evaluate
the significance of these factors or anomalies with respect to the cause of failure.
Methods for predicting the strength of laminates composed of plies at various angles are needed
to allow designers to orient the fibers in the load bearing and stiffness critical directions. The methods
are semi-empirical in that they rely on measured strengths in the principal material directions for
calibration. The theories are focused toward the strength of an orthotropic laminate under in-plane
loading. Laminate level analysis is always based on ply level analysis. This is fundamental to the concept
of lamination theor).
At this level of analysis the failure of an individual ply is predicted in terms of the strengths in
the principal material directions and an appropriate failure criterion. The overview presented here is
drawn from the excellent discussions given in References 13, 14, and 26.
9-8
Maximum Stress and Maximum Strain Theories. The maximum stress theory states that
fracture occurs when the stress in any of the principal laminate orientations exceeds its respective
strength. This criterion defines a failure envelope described by the following equations:
C, < X fora, > 0and uot < X' foroa < 0 (Eq. 6)
and
02 < Y' foro2 > (Jand la21 < Y' fora 2 < 0 (Eq. 7)
and
where:
"For in-plane loading of an off-axis ply, the principal stresses can be calculated by the
transformation equations below and then substituted into the failure criteria in the previous equations.
where:
ni cos
n = sin
An analogous failure criterion in terms of strains (see Reference 13) generates strength
predictions in close agreement with the maximum stress theory. Figure 9-2 illustrates data taken by
Tsai (Reference 29) that shows there are significant discrepancies between theoretical strength
predictions based on the maximum stress failure criteria and experimental data for glass/epoxy. This is
expected since interactions between stress components are not accounted for.
9-9
150 4-
100
70
40
ox
ksi 20
0 15 30 45 60 75 90
e
TsaI-HIll Theory. A strength criterion based on von Mises' isotropic yield criterion was 0
proposed by Hil! (Reference 30) for anisotropic materials. For biaxial loading and plane stress
conditions, lainina failure would occur when,
.g2, 1 U,
2 G2+ o 2S,2 _ 1 ' q 12)
X )2 + (Eq.
where:
X = X' when oi is positive
X = XC when ol is negative
Y = Yt when a2 is positive
Y = YCwhen 02 is negative
Xt', Xe '= tension and compression strength along the fiber direction
yt, Y' = tension and compression strength transverse to the fiber direction
9-10
1501
100
70
40
ksi
10 ompression
7 Tension
4 i - - r
0 is 30 45 60 75 90
0
It should be noted that the Tsai..Hill criterion provides for interactions between stresses and
yields a smooth curve when compared to the maximum stress or strain criteria. However, von Mises
isotropic yield criterion and hence the Tsai-Hill theory is related to distortional energy as opposed to
dilatational (volume change) energy. The disadvantage to Tsai-Hill's criterion is that biaxial loading of
orthotropic materials always causes distortional and dilatational energy. Henice failur, may not be
directly related to distortional energy as it was for isotropic materials. A more general strength
criterion including additional interaction terms is discussed in the next paragraph.
Tsal-Wu Quadratic Interaction Failure Criterion. The Tsai-Wu strength criterion reduces to the
equation below for inplane loading of a thin (plane stress conditions) orthotropic ply.
2
Fhu, 2 + Fa 2 + F6a 6 2 + 2F1 2 o~a 2 + Fxa, + F2 zu + F6 0 6 1 (Eq. 14)
and
1 1
F2 1 y1
=00 F6 = 0
9-11 - -
F 0
1
F2 2 = 1_
where Xt, X', yt', Y, and S have the same meanings as denoted previously.
F1 2 , which represents the interaction between normal stresses, must be determined by
performing a biaxial stress test. Since this test is relatively complicated, it has been recommended in
References 13 and 26 to use:
where:
FP =0-.5
It should also be emphasized that for uniaxial loading of a unidirectional lamina the failure
strength is insensitive to values of F12 within the stability limits in the following equation.
I 1
1 1 F2)2 < F12 < (Ft F2
-(FW 2
This insensitivity is demonstrated in Figure 9-4 where predicted strengths are in excellent
agreement with experimental data for boron/epoxy (Reference 31). Figure 9-4 also demonstrates that
for uniaxial loading there is little difference between the Tsai-Hill and Tsai-Wu criteria. Although the
Tsai-Wu strength theory is more complicated, the well founded mathematical operations of tensor
theory can be used to transform the strength parameters ii these equations. This is important since it
leads to straightforward computer implementation. References 13 and 26 provide invaluable discussions
and examples on the Tsai-Wu strength theories in terms of stress and an analogous strain criteria.
Christensen has recently developed a new failure criterion for continuous fiber composites that
represents a major departure from traditional analyses (Reference 32). This theory was developed from
an effort to extend conventional laminate theory to thick composite sections. To include out-of-plane
stresses that are present in thick laminates, Christensen postulated the simplifying assumption that the
out-of-plane stresses in a laminate were independeit of the orientation of the fibers. In doing so,
Christensen reported that the failure criterion consisted of one in which fiber-dominated failure could
be separated from matrix failure, and matrix-interface failure. Thus fiber failure and matrix failure
could be treated as two separate events. A more detailed summary of this theory has been presented in
Volume I, Appendix A, of this report.
Lamina level strength theories presented in these paragraphs differ primarily in the number of
empirically determined coefficients used in the curve fitting equations. The relatively simple maximum
stress or strain criteria indicate the failhae mode while the others do not. However, the Tsai-Hill and
O
9-12
200
150
Ox Legend:
ksl 1 - Tsai-Wu (F,, -0.58x10 1 0 )
50 0 Experimental data
0
0 15 30 45 60 75 90
Tsai-Wu strength criteria are recognized as useful design tools. Predicting strength using the strain
based analogies of the theories presented here is a generally accepted practice in the aircraft industry.
This results from the fact that design allowables are given in terms of strain. The Ian ina strength
theories in this paragraph will be applied to predicting laminate strength in the next paragraph.
First Ply Failure. At some point during loading of a multidirectional laminate, the ply or group
of plies with the lowest strength will start accumulating damage. This event, often described as first ply
failure, can be predicted using the lamina level theories presented in this section. Figure 9-5 shows a
comparison between predicted strengths for plies in a uniaxially loaded multidirectional graphite/epoxy
laminate based on the quadratic interaction criterion (Reference 33). It is seen that first ply failure
(that is, the 90-degree plies) occurs at a lower load level than catastrophic laminate failure. This is
expected, since load shedding from the damaged 90-degree plies to the rest of the plies continues until
the laminate cannot carry additional !oad. In this case, a more accurate, but nonconservative, prediction
of the strength would be based on the strength of the 0-degree plies as shown in Figure 9-5. The first
ply failure envelope for a multidirectional laminate is the intersection of the failure envelopes for each
ply angle in the laminate. This is shown schematically in Figure 9-6 based on the quadratic interaction
criterion and ply strength data from (Reference 13).
Ply Discount Methods. First ply failure is quite conservative because the initial damage in a
multidirectional laminate is cracks running parallel to the fibers. These cracks are modeled by reducing
the matrix modulus of the cracked ply group. Micromechanics (References 13 and 26) can then be used
9-13
1000
T300/5208 graphite-epoxy
G [01o/1/0Js
0 [0,'90ij ]s
800
600
Strength. MPo! -
0 deg
400
4%
200 -j"-
g0 deg
0 I II I I I
0 15 30 45 60 75 90
to calculate the reduced transverse and shear modulus of the plies. In the next step, lamination theory
is employed Lo predict the redistribation of loads within the laminate. Loads are then ,'eapplied
incre qnta]Iy until compressive or tensile fiber failure occurs. Since the fibers carry most of the load,
ths point often corresponds to the peak load or strength of the laminate. Ply discount methods
inccrporating the effect of hygrothermal stresses are being used with limited success by designers.
It should be noted that these methods are based en lamination theory which does viot accourt
for the interlaminar stress concentrations at the free edges. Transverse cracks also cause interlaminar
stiess concentrations which may lead to delamination and significant reductions in laminate strength.
Two or three dimensional finite element modeling, discussed in the next paragraph, is one of the
methods employed to investigate these microstructurfd failure modes.
9-14
001
-0.01
-0.02
(45)
Finite Element Modeling. Incorporating the effects of free edge interlaminar stress concentra-
tions with environmental and cure stress considerations represents the next level of complexity in
predicting failure onset using individual-ply failure criteria. For the failure analyst, these works are
useful in that they embody considerations likely to be necessary with real-world structures. Both
Crossman and Herakovich (References 15 and 16, respectively) observed that significant variations in
stresses and strains can occur at the free edge of specimens, Both authors show the magnitude of these
stresses for relatively simple layups and specimen geometries. Generally, the most significant stresses
axe those developed near the specimen's free edge. In analyzing these stresses, both authors employed
two dimensional finite-element grids arnanged along the specimen cross-section. Using these grids
illustrates the degree of complexity involved in determining the interlarninar stresses (or strains) in
microscale with such design details. As noted by Crossman, particularly large gradients can occur in
both Z and Y directions near the free edge (Figure 9-7). Regarding the application of individual-ply
failure criteria, the large increase in oa,u, and r." stresses near the edge of the specimen are
particularly significant, since laminate theory methods would have ignored these increases.
Furthermore, Crossman and Herakovich predicted that stresses would be fither influenced by
internal cure stress, test temperature, and conditions of moisture absorption or desorption (Figure 9-8).
In Herakovich's vork, the relationship of these stresses to failure prediction were considered. Since his
finite-element model examined stresses along three dimensions, it was necessary to develop a full
three-dimensional failure criterion. Herakovich used the tensor polynominal criteria advanced by
Tsai-Wu, in which strength tensors are given in terms of material principal strengths. As illustrated in
Figure 9-9, Herakovich predicted that the onset of failure depends quite strongly on edge stresses,
9-15
particularly for small ( 0) laminate angles. While not verified against actual test data, these results,
when considered with Crossman's, clearly indicate that both edge and environmental stress effects must
be carefully evaluated in predicting the onset of failure for whichever individual-ply criteria are
employed.
206850 20685 0
(00) (30)
S.......... I . 2h
17237 5 z = h 17237 5
(2 5) I - Z
2 3h 5) Z
,3790
(20,
o
h
3
(2 0)
I0
.s 4
I-
o
9'
Z.0
103A2 b 16h 10
3h5
01 5)
.F- O 7 (10O)
b t
68950 68950
. -Z *3h x -
(0)
-.
L
-3447 5
1 34475 ,. Z h
-
-68950__________..... 68950 __.._-- _ , .
( II
-I)
0 2 4 6 8 0 2 4 b 8
09__
9-16
1 % I
\ % ~ ~Mochanical I " Ii
Therma1
le N
II1
z / // / III
b I I ,
Figure 9-8. Through-Thickness Tensor Polynomial Distributionsfor Curing Stresses and Stresses at the
FirstFailure (+/-) Laminates
9-17
75
60
0.9
45
0.7
z
Tensor Polynomial
0.5
H- -0
b T
I.
0.3
30
0.1
100
0.0 02 0.4 06 08 10
Ylb (Rot 16)
9-18_-
Perhaps the farthest advancement of first-ply failure criteria as a method of prediction has been
made by Chamis (Reference 10). In attempting to predict the defect growth and damage of composite
materials subjected to load, Chamis developed an integrated computer program called CODSTRAN.
Within this program, a detailed finite-element grid is constructed and evaluated for failure using both
individual first-ply failure criteria and laminate-level fracture criteria. In the case of individual-ply
failures, Chamis' program incorporates both general quadratic and modified distortional energy (Von
Mises') criteria to predict the occurrence of failure for each clement of the overall finite element model.
What is unique about Chamis' work, however, is that failed-ply elements are eliminated, and the
analysis process is reiterated. As a result, CODSTRAN describes the sequence of events leading to
failure, and an approximate prediction of the load at failure.
The failure modes of composites are far more complex than those discussed in these
paragraphs. At the ply-level, Tsai-Wu's quadratic interaction failure criterion seems to provide
sufficiently accurate results for unidirectionai laminate strength under biaxial loading. Prediction of
multidirectional laminate strength is quite complex since free-edge interlaminar stress concentrations
and hygrothermal stresses must be considered. Finite elemenirs and interactive ply discount methods
have been used with limited success.
It can be seen from the discussions above that computers play an important role in the stress
analysis of composites. Lami.atior- theory predictions of stiffness and the laminate strehng-h theouile
discussed here have been implemented on microcomputers (that is, References 26 and 34).
The methods presented in the following p-agMphs can provide criteria for selecting the
laminate geometry providing optimium strength in the load bearing directions. However, predictions of
laminate strength, at this time, are not quantitatively accurate. Thus, the designer must rely on coupon
or full scale tests to determine the actual strength for a particular laminate geometry. This requires
large and expensive databases to design structural composite parts. If needed, the failure analyst should
consult designers or stress analysts to find out what methods were used to establish the allowables. The
overview presented in this section is intended to familiarize the reader with some of the considerations,
many of which are unique to composites, that must be applied to pr'edicting laminated composite
strength.
In the previous paragraph, it was noted that transverse cracking and free-edge delamination
reduce laminate strength. In this paragraph, the influence of ply thickness and orientation on these
microstructural failure modes is discussed. It is intended to provide the reader with an introduction to
microstructural failure features unique to laminated composites.
In the previous paragraphs, the strain required to cause cracks parallel to the fibers (transverse
matrix cracking) was assumed equal to the transverse failure strain of a 90-degree ply. However, it has
been shown by numerous researchers (References 35 to 39) that the strain to cracking of an off-axis ply
g-roup depends upon its thickness. Fiaggs and Kural (Reference b5) have clearly demonstrated that the
in situ strain to cracking decreases as the thickness of an off-axis ply group (30-, GO- or 90-deg-rec)
9-19
increases (see Figure 9-10). The fact that the in situ strength of a 90-degree ply is greater than that of a
laminate composed only of 90-degree plies has been attributed to the constraint provided by the stiffer
surrounding ply groups Flaggs (Reference 39) used a fracture mechanics approach combined with 2-D
shear lag analysis to successfully model the in situ strain to cracking as shown in Figure 9-10. This
model has also been applied successfully to predicting in situ cracking of 30- and 60-degree ply groups
(Reference 39).
12
Legend:
1.0 -- '- Experimental data
0 20 shear lag
7 90) 16 stra:n cutoff
08
Strain to
cracking
06-
0.2 I ofived
7 8
1 2 3 4 5 6
n (Rof. 39)
Figure 9-10. Comparisonof Experimentally Derived In Situ Lamina Elastic Strains at Onset of Matrix
Cracking With 2D Shear Lag Model Predictionsfor (0/90) T300/934 Laminate Family
With this type of analysis, the first ply failure predictions of laminate strength could be made
more accurate. To accomplish this, Flaggs' model would be used to predict the in situ strain to cracking
of the off-axis ply groups in a multidirectional laminate. Then these strains would be used as the strain
to failure transverse to the fibers in the lamina level failure models. This represents an effort to
integrate the prediction of microstructural failure with gross laminate failure features.
The methods described in Section 9 are commonly used to predict the strength of an unnotched
laminate (that is, without in-plane stress concentrations). However, composite structural components
may contain cutouts, fastener holes, or impact damagc which reduce the in-plane strength below that of
the virgin laminate. These laminates must be designed "a priori" such that the residual strength will be
greater by some safety factor than the operating stresses. Some of the more popular ap)proaches applied
to predicting residual strength are reviewed in the following paragraphs. 0
9-20
Most of the analytic techniques for predicting residual strength employ semi-empirical
approaches similar to the fracture toughness methods commonly used with metal structures. These
methods typically involve the experimental determination of an intrinsic material property related to
crack growth - such as material fracture toughness. This value, when considered with the size and
geometry of the flaw, allows the stress at the fracture to be calculated.
The anisotropy of composites complicates the analyses considerably. Critical stress intensity
factors used in metals analysis are independent of the direction of crack growth. However, the
translaminar toughness of composites may be hundreds of times greater than the interlaminar
toughness. This is expected since breaking fibers is a much higher energy fracture mode than matrix
cracking. Often, the interaction of these two failure modes occurs as in impact damage. In this case the
assessment of residual strength becomes so complicated that the designer must rely on empirical data.
However, when the translaminar and interlaniinar modes are acting independently, the rapidly maturing
fracture mechanics approach for composites are being used successfully
LAYUP (DEGREES)
KC 943 23 35 1622
(Ret 20)
9-21
T-.is point of ir,,tability can be defined by using the equation:
Y geometric factor related to the crack length and location within component being
examined
o, = stress at in.tability
a -
a'Y U'y= Co 2- +a
*Y 2
x a so (1 + E 2 ) (2 + E22)
0 ao
(Ret. 7. Reprnleo w:!h permission frOm B Ht Noion et al. =ICCM II. Tre Metaliu:gica'
Society. 420 Commonwealth Drive. War.enoale. Pennsyivan a 15C86.)
Figure 9-11. Strength Reduction of Uniaxially Loaded Plate With CircularHole According to
Average Stress Criterion
9-22
do =characteristic length dimension
S 1 (af . 3 (a" 4
do 2 2x 2 ~2.)
(Ref. 8)
Figure 9-172. Strength Reduction of Uniaxially Loaded Hole According to Point Failure Stress Criteria
.0 ~ Experimental
- Point Stress criterion (d 0 =1mm 10 04 in 1)
0.2
00025 0.50
Hole Radius, (
(Rof 7. Repr.-iieO Wihnpermiwson Irom B.R Noton ei ai'. ICCM U."The Meiaiiurg.cai
Soc-ely, 420 GornrcnWealth Drive, Wa'renda!e, Pennsy~vania 150o6.)
Figure 9-13. Strength Reductions as a Function of the Hole Radius for (0/ + 451-45190 deg) Graphite-
Epoxy Plates With Circular Holes Under Uniaxial Tensile Loading
The more general Damaged Zone Model (discussed by Aronsson in (Reference 9) requires only
basic laminate p~roperties (strength and stiffness) and the apparent fracture enerV' to predict the
fracture behavior of brittle and ductile notched three point bend specimens.
In this model, the damaged (as opposed to a crack in metals) zone which develops in composites
around a crack or hole is modeled by a crack with cohesive stresses acting on its surfaces. As the
applied load is increased, the damage grows. This is approximnatedl by reducing the cohiesive forces on
* ~the cracked region in the model. The Damaged Zone Model (DZM) and the Point Stress Criteria (PSC)
(Reference 9) accurately predlict the failure load for brittle and (ductile matrix three-point b~endl
9-23
specimens. The real utility of the DZM lies in its ability to analyze complex geometries. The PSC, which
relies on exact calculations of the stress distribution around a flaw, is limited to very simple flaw
geometries such as holes. Tensile failure loads have been predicted by the DZM to within 10 percent of
the experimental values for laminates with the hole geometries in Figure 9-14 (Reference 52). The DZM
should be a tool for the aerospace industry that can be used for predicting the residual strength of
laminates with cutouts.
S+ __ --
(Ref. 52)
Figure 9-14. Hole GeometriesAnalyzed With the DamagedZone Modei (DZM)
Of particular concern to the failure analyst, however, is the ability to predict failure onset for
flaw conditions such as through-cracks or impact damage. In this area, Mikulas (Reference 8) examined
the applicability of the point stress failure criteria; for the case of a through-crack, stress as a function
of distafnce fron. the crack was expressed as:
a _ X
(Jy _ _2 (E q . 1 8 )
U. the square root of X - (2)(
where:
x= X 1+a
2
X distance along the x-axis, away from the crack
a crack length
cs . stress at infinite distance from crack
9-24
As a result, Mikulas indicated that the residual strength for a given crack of size a can be
determined if X is set equal to do.
With respect to impact damage, Mikulas reported that the characteristic length d" depends or
the toughness of the resin system examined. As illustrated in Figure 9-15, a relatively good correlation
exists for tough resin chemistries, but not for brittle, delamination-prone resin systems. These
observations indicate that the accuracy of failure predictions will depend strongly on the resin system
used, and the configuration of damage examined. An excellent discussion of impact damage with respect
to failure modes and the effects of resin toughness has been given by Starnes, Williams, and Rhodes in
References 19, 53, 54, and 55. Their studies have clearly shown that:
1. Tough resins reduce the size of the damage zone caused by impact
2. Several graphite/epoxy systems with varying toughness exhibited similar residual post-impact
compression strength for the same damage zone size
3. The dominant failure mechanisms causing post-impact compression failure are delamination
and shear crippling.
0-014
0 012
0 0o~
010
0 010 dox 1 mm 10 04 in)
0 008 000Impact
Circulart tt
"tough^ resinls) hoe.t
Failure strain 13 cm
0.006 5 12 i)
0 004
o I --
0 02 04 06 08 o
AMW
(Ret 8)
Figure 9-15. Effect of Impact Damage on the Compressive Strength of a Quasi-Isotropic Laminate
While the works discussed earlier in this section examined the residual strength of the
laminates with translaminar through-thickness flaws, other investigators have considerf-d tihte criticeality
of interlaminar defects. This paragraaph has been included to acquaint the readelr with some of the
9-25
significant results and tests applicable to assessing the criticality of interlaminar defects in composites.
Delaminations may grow and initiate component failure due to ;udden loss of stiffness and strength.
Fractography can be used to detect delauaination and interlaminar crack growth directions, but the
methods discussed here can be employed to evaiuate the criticality of delaminations.
As a result of processing and service condi ms, delaminalions may be introduced into
composite structures. Fatigue and static loading of laminated composite structure may initiate
delaminations near interlaminar stress concentrations. Some of the common design features found in
composite structures such as bolted joints, ply drops, and cutouts contain interlaminar stress
concentrations (see Figure 9-16). In addition, multiple delaminations represent. characteristic
post-impact damage. Local instability of a delaminated subregion in composite structures under
compressive loading precipitates out-of-plane deformations and may lead to subsequent crack growth.
Under these circumstances, it has been well docwnented that structural strength and stiffness
reductions are significant (References 57 to 65).
Free edge Cutout (I.e., hole) Ply drop Bonded joint Polted joint
(Ref. 5-)
In the presence of interlaminar stress singularities, the fracture mechanics approach is now
frequently used to assess defect criticality in composites. Double cantilever beam (DCB), Mode I, and
end notched flexure (ENF) Mode II specimens, as shown in Figure 9-17 (a and b), are being used to
evaluate the pure mode critical strain energy release rates. Strain energy release rates are utilized in
these pure mode tests because G is a physically well defined quantity experimentally measurable with
compliance calibration techniques- But, in general, cracks in composite structures are subjected to all
three modes of loading at the crack tip as shown in Figure 9-18.
To study mixed-mode crack growth criteria, the imbedded through-width delamination (ITWD)
and cracked lap shear (CLS) specimens shown in Figure v-19 havc been used by a number of
researchers.
9-26
* The opening moment and the eccentricity in load path at the crack tip of the CLS and !TWD specimens
are the mechanisms causing interlaminar normal (Mode 1) and shear (Mode 11) stress concerntrations
(see F~igure 9-20).
L 1L-
P/2 (a) ENF Specimen P.?
Mode IIId
Mod, Mode 11
9-27
(a) Crack Lap Shear (CLS) Specimen
M0
InterlminarInterlaminar
NormalShear stress
Figure 9-20. Crack Tip Loading Mechanisms Causing IntprlammnarNormal and Shear Stress
Conc9intrations
9-28
The strain energy release rate methods for evaluating the criticality of delaminations assume
that G, is a material property. This type of analysis is based on the equation below presented by Griffith
(Reference 66).
2 (Eq. 19)
G -P dC
2b da
where:
b = width of crack
C = compliance
a = crack length
P = load
For a single mode of loading at the crack tip, the load on the specimen, P, is increased until G
reaches its critical value G, at the onset of crack growth. In this case, the failure criterion can be stated
as crack growth occurs when G reaches G,. Once G, has been measured, the fracture load, P,, can be
found for any other crack length or geometry by evaluating the change in compliance with respect to
crack length (dC/da). Then, dCida can be adequately determined analytically for simple pure Mode I or
Mode II geometries such as the DCB and ENF, respectively; however, finite element methods combined
with the virtual crack closure analysis (Reference 67) are required to evaluate dC/da in more
complicated structural applications. These types of complex geometries often have more than one mode
of loading present at the crack tip (that is, mixed mode). In this case the interaction of modes must be
considered, thus invalidating the simple crack growth criterion mertioned above. The mixed mode
fracture criterion becomes more complicated since the critical value of the Mode II interlaminar
toughness of composites can range from one to ten dimes that in Mode I as shown in Table 9-4.
However it has been recently shown by Rothschilds (Reference 25) and Johnso" (Refeience 68) that the
linear mixed-mode crack growth criterion, in the following equation, provides accurate predictions of
the critical loads at the onset of mixed mode crack growth.
Table 9-4. Gic and Guc Values ObtainedFrom DCB and ENF Testing Reported in Literature
*1 kJ/m 2
= 5.71 In lb/in 2 .
9-29
G, Gp, ,
6-,+ 2!1 = (Eq. 20)
With this criterion, the criticality of inter!aminar cracks can be evaluated in complex
mixed-mode geometries. It shouid be noted that the criterion in the above equation reduces to the
desired result in the case of pure mode loading. It is also seen that the mixed-mode criterion above
requires the pure Mode I, Mode II, and or Mode III critical strain energy release rates. Fracture testing
exhibits rate dependence, material nonlinearity, and subcritical crack growth. To be consistent. these
effects must be considered when using the results of pure mode fracture testing in the analysis of
fracture in 3tructural components.
The previous paragraph discussed methods used to evaluate the criticality of interlaminar
cracks under quasi-static loading. Some excellent investigations on interlaminar fatigue crack growth
were performed by Wilkins (Reference 69) and Russell (Reference 70). They have shown that a power
law relation exists between the change in G during a fatigue cycle and the crack growth rate. Table 9-5
shows the crack growth -ate equations in Mode I and Mode II for AS-1/3501-6 (Gr/Ep). Similar
equations for other materials could be used in conjunction with the methods for predicting G (discussed
above) to evaluate the criticality of interiaminar fatigue in laminated composites. Figure 9-21 shows
rather sorprisingly that over a certain range of cyclic Mode II crack loading, the tougher thermoplastic
AS4iPEEK has a higner ciacir growth rate than the relatively brittle epoxy. The characteristic surface
iaicrofeatures iave been documented in a thorough investigation of composite interlaminar crack
growth by Russell (Reference 70). Failure analysts can use this information to characterize the loading
(static or fatigue) ard interlaminar stress state (Mode I or Mode II) in the region of failure initiation.
Recreating the leads at the onset of failure may illuminate the cause of failure.
GTH Is the cyclic strain energy releaqe rate corresponding to a threshold crack growth
rate of 2.54x10- 6 mm/cycle.
9-30
o 10_,
,,r
10-3
LL
10-5
103
10 10102 4 6 8 2 2 4 6 8 3
2
AGjl (JIm ) (RA =.-1)
Fig=-c
.21. Mode II Fatigue
While the stre igth reductions due to holes, cutouts, and impact damage are incorporated into
design, manufacturing defects are not accounted for. Some of the typical manufacturing defects that
may cause strength reductions are listed in Table 9-6. This chart also lists the mechanical properties
likely to be affected by the defects and some of the methods and data needed to evaluate the criticality
of the flaw. Although the list in Table 9-6 is incomplete, it provides the failure analyst with an
understanding of some causes and effects related to typical manufacturing defects.
9-31
Table 9-6. ManufacturingDefects
Defect Type of delect Cause t~ecban.zat eltod used to Data needed t0
properltes ovaijate criticality perlornt analysis
at deteels
Lam'nate Strength. stittness. Larrsnation theor y.
t
Smatt error tn p y fabrication Layup tectin qua and hnygcothermal laninate fracture
angle defect stabilty criterion, finite
element
porosity
ferianztr shear
Empirical evalua-
Poor impregnation Interlarnnar tough- arnr.natron strength and or
porosity
(I.e. porsit Processing error of resin Into fiber ness--susceptibte theory finite ,nterian....ar tcugh.
tow to transverse elements crack ness versus per-
cracking growfh ctiterion centag,
porosity of
Lamrnration theory.
Lamlrate f.iite elements. Ilipl3ie arid uiter-
incorrect materlai fabrication Layup technique All laminate fracture I.mnnar fracrure
defects critericn crack to.gr.ness
growth criterion
9-32
SECTION 10
This section presents formats for reporting failure analysis and fractographic data, with the
intent that eventually all the information can be incorporated into a computerized database.
-0
10-1
FAILURE ANALYSiS COLLECTION
AND TRACKING SYSTEM (FACTS)
DATA, INPUT SHEET
OPERATOR: DATE: i_
REPORT NUMBER: DESIGN DRAWING PART NAME&NU,18ER: _..-
NUMBER OF LANDINGS-
BACKGROur'D/INORMATION:
LOCATION OF DAMAGE: ___
ENVIRONMENTAL CONCERNS:
(OTHERS):
DATA: ___
ANALYSES CONDUCTED
RESULTS:
RECOMMENDATIONS:
KEYWORDS: ,,_
Figure 10-7. Figure Analysis Collection and Tracking System (FACTS) Data Input Sheet
] (-2
.1NON-DESTRUCTIVE EXAMINATION
DATA INPUT SHEET
OPERATOR. - DATE
PART NAME/NUMBER:
MATERIALS & CONSTRUCTION:
LOCATION OF ANALYSIS.
SUPPORTIVE DATA:
KEYWORDS.
1 0_3::
MATERIALS CHARACTFRIZATION
DATAINPUT SHEET
OPERATOR. DATE
PART NAMEjNUMBER.
MATERIALSjS PE C IFICATIONS
SPECIFICATION REQUIREMENTS:
CURE TEMP
FIBER;RESIN DE,.SITIES
VERIFICATION DATA
"* Tg DETERMINATION-
INSTRUMENTATION_
RESULTS. _
RESULTS
INSTRUME NTATION
RESULTS
"* SPECIALIZED ANALYSES METHODS USED (HPLC. CPC. DS(. SL'RF ANALYSIS. ET C )
RESULTS
KEYWVC)RDS
0
Figure 1G-3. Maerials CharacterizationData Input Sheet
0
1.,-4
MATERIALS CHARACTERIZATION
DATA INPUT SHEET
(FIGURE ATTACHMENT)
10-5
FRACTOGRAPHY
MACROSCOPIC ANALYSIS
DATA INPUT SHEET
OPERATOR DATE.
PART NAMEiNUrABER:
MATERIAL:
VISUAL OBSE RVATIONS.
KEYWORDS.
i1
FRA("TOGRAPHY MACROSCOPIC ANALYSIS
DATA INPUT SHEET
(PHOTO ATTACHMENT)
MAG
MAGNIFICATION:
COMMENTS:
10-7
FRACTOCRAPHY
MICROSCOPIC ANALYSIS
0
DATA INPUT SHEET
OPERATOR: DATE_
PART NAME/NUMBER.
MATERIAL
RESIN/FIBER SYSTEM.
LAYUP:
MICROSCOPIC OBSERVATIONS.
KEYWORDS:
10-8
FRACTO6RAPHY MICROSCOPIC ANALYSIS
OATA INPUT SHEET
(PHOTO ATTACHMENT)
0 Optical photomicrograph
* Low-Mag photomicrograph
* SEM
* TEM
* STEM
* High-Mae photomicrograph
"* SEM
"* TEM
"* STEM
MAG
MAGNIFICATION:
COMMENTS:
10-9
STRESS ANALYSIS
DATA INPUT SHEET
"* RESULISiCOMMENTS.___
* RESULTS: __
KEYWORDS: _
l0-10
STRESS ANALYSIS
DATA INPUT SHEET
(DIAGRAM ATTACHMENT)
COMMENTS.
10-] 1
1.0 INTRODUCTION
2.0 MATERIAL
11
10-12 e_
1.0 ABSTRACT
2.0 BACKGROUND
- Part Identification
- Manufacturing History
- Service History
- Field Information Relatin-g To Fracture
- DetectIion of Problem
4.0 CONCLUSIONS
5.0 RECOMMENDATIONS
10-!3
SECTION 11
GLOSSARY
A-Basis. The mec aical property value above which at least 99 percent of the population of
values is expected to fall, with a confidence of 95 percent.
A-Stage. An early stage in the polymerization reaction of a theiinosetting resin in which the
material is still soluble and fusible.
Adhesive. A substance capable of holding two materials together by surface attachment.
Structural adhesives produce attachments capable of transmitting significant structural loads.
Advanced Composites. A family of high-performance materials that consists of a high-
strength, high-modulus continuous fiber system embedded within an essentially homogeneous matrix
material.
Advanced Filaments. Continuous filaments made from high-strength, high-modulus materials
for use as u .unstituent in advaunced comnposites.
cottlzeutzLL
Angle-Ply Laminate. A laminate in which the fiber orientations in successive plies alternate
between "+" and "-" with respect to the global reference coordinates.
Anlsotropic. Not isotropic; having mechanical and/or physical properties which vary with
orientation at any point within the material.
Ar&mid Fibers. A class of aromatic polyamide fibers, presently including Kevlar, Kevlar 29,
Kevlar 49, and Nomex, all products of the DuPont Company.
Autoclave. A closed vessel that produces an environment of gas pressure, with or without heat.
B-Basis. The mechanical property value above which at least 90 percent of the population of
values is expected to fall, with a confidence of 95 percent.
B-Stage. An intermediate stage in the reaction of a thermosetting resin in which the material
softens when heated and swells when ini contact with certain solvents, but (toes not entirely fuse or
dissolve. Materials are usually precured to this stage to facilitate handling and lay-up prior to final cure.
Bagging. An operation consisting of thermocouple placement, bleeder cloth, and blanket
arrangement over a laminate iay-up, placement of a caul sheet (if required), installation of vacuum hag
and vacuum lines, and sealing of the vacuum bag.
Balanced Laminate. A laminate in which laminae at angles other than 0 degree and 90 degrees
occur omy in i--- pairs (not necessarily adjacent to one another). Every ' +" lamina has a corresponding'
0 "lamina.
11-1 .
Batch (or Lot). In gene.-al. a- quantity of material formned using the same process and having
identical characteristics throughout. A batch of prep-eg tape is produced from at single batch of miat~rix
material. The prerreg tape batch is not necessarily produced at one time, but all sub-batches are
p~rodlucedl in the same equipmient undler identical conditions.
Binders. Either liquid emuilsions )r piulverized resin solids5, directlyv app~liedl to virgrin blown
fibers to form miat prodlucts.
Bleeder Cloth. A nonstnuctural layer of matci ial usedl in the rrianufactuv e of comp)osite parts to
allow the escape of excess gas antd resin (luring cure. The bleeder cloth i.; removed after the la~minate
layul) is cured, and is not part of the final laminated part.
Bond. The adhesion of one surface to another, %%ith or without the use of an adhlesive ats a
bondling agent.
Braid. Yarn diat is interlaced by a process siniilar to that usedl in tile weavirng of fabrics;
available in flat, round, or tubular forins; often referred to as "sleeving' or 'i6raided fabric."
Broadgoods. A term loosely applied to prepreg material greater than 12 inchies ill width,
usually furnished by suppliers in continuous rolls, ?applied to both co!imated unaxial tape and woven
fabric prepregs.
Carbon fibers. Fibers produced via oy-rolvtic de:,Ta-'stion of .wnjt)i(-( c organic fibers, rayon, or
polyacrylonitrile (PAN), which have about 52-991/' carbon c, .,tmnt. The fibers have moduli greater than.
or equal to 70 GPa (10 million Dsi).
C-Stage. The fival stage of curing of a therinosetting resin in which the mzd;erial IS infusible
andl insoluble in common solvents. Fully cured thermnosets are in this stage.
Cable. Yarn that is phed more than once; yklrn made 1-y Olying two or monre previously plied
yarns.
Caul Plate. Smooth metal plate, free cf surface diefects, the same shape ats a toruposite layup;
used to transmit normal pressure during the curing process and to provide a smooth surface on the
finished laminate.
Chemical Size. A surface finish applied to the fiber that c(,ntains some chemical cons~ituents
other than water.
Cocuring. The process of curing a composite laminate an,! siniultaneously bonding it to some
other prepared surface (luring the same cure cycle.
Collimated. Rendered par-Allel.
Composite Material. A material composed of two or more different constituent materials.
Current structural composites consist of any combinwition of fibeirs, whisker,,, andl pmiticdes in a
common mn-atrix mnaterial. Composite materials may be classified als follows:
1. Fibrous composites - Fibers in a matrix
2. Lamninated composites - Laiyers of the same or Vac-iOns materials
3. Particulate composites - Particles in at matrix
4. Hybrid composites - Composed of more than one kind o~flhr mat rix imalteia sys,'em.
11-2
Consolidated Monolayer. A form of metal matrix -reply in which the fiber system and the
encapsulating metallic matrix material are fully or partially consolidatet.. into a stable lamnina.
Menolayers are combinedl into orieiitedl stacks which are then further consolidlate(] into a finished
laminate.
Constituent Class. A group of fiber or matrix constituents of the same generic chemnica, type
or family; e.-., graphite is a generic fiber class and epoxy is a generic matrix class.
Continuous Filament Yarn. A.yarn formed by twisting two or more continuous filaments into a
single, continuous strand.
Core. A sairdwich- filler material, generally cellular in nature, resembling a honeycomb and
referred to by that name. The core material may be paper, nylon, phenolic, aluminumn foil, etc.
Coupling Agent. That part of a sizing or finish which is desig-ned to provide a bond between
the reinforcement and the matrix material.
Cure. To change the properties of a thermnosetting resin irreversibly by chemical reaction
which may be condensation, ring closure, or addition. Cure may be accomplished bty the additinn of
curing (cross-linkirg) agents, with or without catalyst, and with or without hceat.
Debond or Liisbond. An adhesive failure or separation along a bonded interface between two
adlierends.
Cebulk (Densificalton). Tuie compacting of plies and removal of air pockets from a laminiate
lay-up prior to curing. This is generally done in stages, under full vacuum, under ambient or elevated
0 temperature conditions.
Delamination. A separation between adjacent layers of materal in a laminate.
Drape. The ability of tape and broadgoods to 2:onform to a contoured shape.
End. An individual warp yarn, fiber, thread, munofilament, or roving.
Fabtic (Woven Fabric). A generic material construction manufactured by ijiterlacing two yarns
at right angles. T[he warp is the system of yarn or threads running tengthwise iri the fabric, and the fill
or filling is the systein of yarn or threads runiiling crosswise in the fabri~c. Warp is also reft-rred to ats
"warp ends" or simply "ends." Filling is also referred to as "filling picks," "weft," or "pi(+S."
Fiber. A single homnogenous strand of makterial, essentially one-dimensional in he macrmol-
chanical sense, used as a principal constituent in advanced co~nposites because of its high axial strengtill
arid modlulus.
Fiber Content. The amount of' fiber p)re-sent in ai cornplosite, UNImAmllx expmressed its voitumte
fracti'on or weight fra:tio.. of th,, composite.
Fiber Direct~on. The' orientation or alignmeitt of the longitudinal axis of #),(i.fibePr wilIih rvspect
to a selec~ed global reference axis (X).
Fiber System. The type 4f arrangement of the fiber constituent of an adva1nced oliflpot'~ih'.
Examples of fiber systems are collimatted filament-lts or filia1 lriwt Yarns, Woven fabm-ic, randoinmly oriented
short-fiber ribbons, randloin fiber trats, whiskers, dtc.
Filament. A fiber characterized by extreme length, such that there is normally no filament end
within a part except at geometric discontinuities. Filaments are used in filamentary composites and are
also used in filament winding processes, which require long continuous strands.
Filamentary Composite. One form of advanced composites in which the fiber constituent
consists of continuous filaments, Specifically, a filamentary composite is a laminate composed of a
number of laminae, each of which consists of a nonwoven, parallel, unidirectional planar array of
filaments (or filament yarns) embedded in the selected matrix material.
Fill. Yarn oriented at right angles to the warp in the woven fabric.
Finish. A materiai with which filaments are treated, containing a coupling agent to improve the
bond between the filament surface and the resin matrix in a composite material. A finish often contains
ingredients which provide high-temperature lubricity to the filament surface preventing abrasive
damage during handling, and a binder wltich promotes strand integrity and facilitates packing of the
filaments.
Flame-Sprayed Tape. A metal matrix preply form in which the fiber system is held in place on
a foil sheet of matrix alloy by a metallic flame-spray deposit.
Flash. Excess material which forms at the parting line of a mold or a die, or that which is
extruded from a closed mold.
Foamed Plastics. Resins in sponge form, flexible or rigid, with closed or interconnected cells
that vary in density from that of the solid parent resin to 2 pounds/cubic foot; good heat barriers, and
the fiber compressive strength of rigid foams makes them useful as core materials for sandwich
constructions. Also, a chemical cellular plastic whose structure is produced by gases generated from the
chemical interaction of its constituents.
Fugitive Binder. A resinous material used irt the fabrication of metal matrix preplies to hold
the fiber system in place on the metallic foil sheet during shipping, storage, handling, and layup. During
laminate consolidation, the fugitive binder decomposes and the products completely vaporize.
Gel. A semisolid system consisting of a network of solid aggregates in which liquid is held; the
initial jelly-like solid phase that develops during the formation of a resin from a liquid.
Gel Coat. A quick-setting resin used in molding processes to provide an improved surface for
the composite. It is applied to the mold after the mold-release agent.
Graphite Fibers. A group of carbon fibers with a carbon content of 99 percent and high
modulus values (greater than 20 GPa or 3 million psi).
Green Tape. A metal matrix preply form in which the fiber system is held in place on a foil
sheet of matrix alloy by a resinous fugitive binder.
0
11-4 .
Homogeneous Material. A material of uniform composition throughout with no internal
physical boundaries; a material whose properties are constant at every point within the material (but
not necessarily with respect to directional coordinates).
Hybrid Laminate. A laminate composed of laminae of two or more composite material systems
(intraply or interply).
Impregnate. To apply resin onto fibers or fabrics by one of several processes: hot melt, solution
coat, or hand lay-up.
Integral Composite Structure. A composite structure in which several structural elements,
which would conventionally be assembled together by bonding or mechanical fastening after separate
fabrication, are instead laid up and cured as a single, complex, continuous structure.
Interlaminar. A descriptive term referring to a location between adjacent laminae.
Intralamlnar. A descriptive term referring to a location entirely within a single lamina without
reference to any adjacent laminae,
Isotropic. Having properties that are uniform in all directions at any point within the material
(independent of the axis of testing).
Mold Release Agent. A lubricant applied to the mold surface to facilitate the release of the
molded article.
Molding. The process of forming a polymer or a composite into a solid mass of prescribed
shape and size by the application of pressure and heat.
11-5
Plastic. An organic substance of large molecula," weight which is solid in its finished state and,
at some stage during its manufacture or its processing into a finished article, can be shaped by flow.
Plastic!zer. A material of lower molecular weight which, when added to a polymer, separates
the m-lecular chains and results in a lowering of the polymer glass transition temperature (with
concomitant property degradation).
Plied Yarn. An assembly of previously twisted yarns.
Polymer. A high molecular weight organic material, natural or synthetic, formed by the linking
together of a large number of repeating chemical units (monomers). When two or more nomomers are
involved, the product is called a copolymer.
Porosity. A distribution of trapped pockets of air, gas, or vacuum within a solid material,
usually expressed as a percentage of the total nonsolid volume to the total volume (solid plus nonsolid)
of a unit quantity of material. Also referred to as voids or microvoids.
Postcure. After the initial cure, an additional elevated temperature exposure, usually above the
cure temperature and without pressure, to improve final properties arid/or complete the cure. In certain
resins, complete cure and ultimate mechanical properties are attained only by exposure of the cured
resin to higher temperatures than those of curing.
Pot Life (of a resin). The length of time that a resin system retains viscosity that is low enough
to be used in processing.
Precursor. Organic material from which carbon fibers are prepared via pyrolysis. Polyacryloni-
trile (PAN), rayon, and pitch are commonly used precursor materials.
Preply. A lamina in the "raw material" stage, as furnished by a materials supplier, including all
of the fiber system placed in position relative to all or part of the required matrix material. An organic
matrix preply is called a prepreg. Metal matrix preplies include green tape, ilame-sprayed tape, and
consolidated monolayers.
Prepreg, Preimpregnated. A mat, a fabric, nonwoven material, or a ro,'ing, with resin usually
advanced to the B-stage (when it is a thermoset resin), ready for lay-up and cure.
Quasi-isotropic Laminate. A laminate that approximates isotropic mechanical behavior on a
macro-scale.
Reinforced Plastics. Composite materials which have a polymer matrix and a modulus that is
less than 20 GPa (3 million psi).
Resin. A solid, semisolid, or pseudosolid organic material which has an indefinite aT1d often
high molecular weight, exhibits a tendency to flow when subjected to stress, usually has a softening or
melting range, and usually fractures conchoidally. Most resins are polymers.
Roving. An assembly of two or more strands, without a twist.
Sandwich Construction. A structural panel concept consisting in its simplest form of two
relatively thin, parallel sheets (face sheets) of structural material bonded !c and separated by a
relatively thick, lightweight, core.
1i-6
Scrim Cloth (also calied glass cloth, carrier). A reinforcing prepreg fabric woven into an open
mesh construction, used in the processing of a tape or any other B-staged material to facilitate
handling.
Secondary Bonding. The joining together of two or more already-cured composite parts by the
process of adhesive bonding, during which the only chemical or thermal reaction is the curing of the
adhesive itself.
Shelf Life. The length of time a material can be stored under specified environmental
conditions and meet all applicable specification requirements and/or remain suitable for its intended
function.
Sizing. Chemical compounds which, when applied to manufactured filaments, create a loose
bond between the filaments, and provide the desired handling and processing properties.
Strand. A primary bundle of untwisted filaments.
Symmetric Laminate. A laminate in which the ply properties (material and geometric) and
fiber orientation are symmetrical about the laminate midplane (midplane symmetry).
Tack. Stickiness of a prepreg or an adhesive.
Tape. Narrow fabric or material that ranges in width from 1/4 to 12 inches.
Thermoplastic Material. A material that is capable of being repeatedly softened by an increase
in the temperature and hardened by a decrease in the temperature with no accompanying chemical
change.
Thermoset Material. A material which becomes substantially infusible and insoluble when
cured by the application of heat or by chemical means.
Tow. A loose, untwisted bundle of filaments.
Typical Basis. A "typical basis" property value is the average value. No statistical assurance is
associated with this value.
Unidirectional Laminate. A laminate with nonwoven prepreg plies that are all laid up in the
same direction.
Vacuum Bag Molding. A fabrication process in which the lay-up is cured under pressure
generated by drawing a vacuum in the space between the lay-up and a flexible sheet that is placed over
it and sealed at the edges.
Veriflim. A nonadhesive curing film used to check the bond line thickness of mating parts prior
to adhesive bonding.
Void. A physical discontinuity in the form of a gaseous pocket, occurring within a material or
part.
Volatiles. Materials in the sizing or the resin formulation in a prepreg which are released as a
vapor at temperatures near or above room temperature.
Warp. The longitudinally oriented yarn in a woven fabric.
11-7
Whisker. A single filamentary crystal. Whisker diameter- range from 1 to 25 microns, with
length to diameter (aspect) ratios between 100 and 14,000.
X-Axis. The axis in the plane of a laminate ,.hich is used as the 0 degree reference for
designating the fiber orientation of a lamina.
XY Plane. The reference laminate midplane that is parallel to the plane of the laminae.
Y-Axis. The axis in the plane of the laminate which is perpendicular to the X-axis.
Yarn. A yarn is produced by twisting and plying either strands of fiber or continuous filaments
in a form suitable for weaving.
Z-Axis. The reference axis normal to the plane of the laminate.
11-8
SECTION 12
REFERENCES
1. Altman, J., et al., "Advanced Composite Serviceability Program," AFWAL TR-80-4092, July 1980.
2. Sewell, T.A., "Chemical Composition Processing for Air Force/Navy Advanced Composite Matrix
Material," AFWAL TR-81-4180, 1982.
3. Carpenter, J.F, "Test Program Evaluation of Hercules 3501-6 Resin," N00019-77-C-0155, May 1978.
4. Young, PR., Stein, B.A., and Chang, A.C., "Resin Characterization in Cured Graphite F:her
Reinforced Composites Using Diffuse Reflectance-FTIR," 28th National SAMPE Symposium and
Exhibition, Vol. 28, 1983, pp. 824-837.
5. Handbook of Chemistry and Physics, 65th edition, R.C. Weast (ed.), CRC Press Inc., Boca Raton,
Florida, 1985, p. E-130.
6. Ratner, B.D., McElroy, B.J., in Spectroscopy in the Biomedical Sciences, R.M. Gendreau (ed.), CRC
Press Inc., Boca Raton, Florida, 1986, pp. 107-140.
@ 7. Daniel, I.M., "The Behavior of Uniaxial Loaded Graphite/Epoxy Plates With Holes," Proceedings of
the Second International Conference on Composite Materials, published by the Metallurgical
Society of AIME, 1978, pp. 1019-1034.
8. Mikulas, M.M., "Failure Prediction Techniques for Compression-Loaded Composite Laminates
With Holes," Selected NASA Research in Composite Materials and Structures, NASA Conference
Publication 2142, August 1980.
9. Aronsson, C.G., and Bachklund, J., "Damage Mechanics Analysis of Matrix Effects in Notched
Laminates," Composite Materials: Fatigue and Fracture, ASTM STP 907, H.T. Hahn (ed.), ASTM,
Philadelphia, Pennsylvania, 1986, pp. 134-157.
10. Chaynis, C.C., and Smitin, G.T, "CODSTRAN: Composite Durability Structural Analysis,"
NASA-TM-79070, 1978.
11. Craddock, J.N., and Champagne, PJ., "A Comparison of Failure Criteria for Laminate Composite
Materials," Structures, Structural Dynamics and Materials Conference, American Institute of
Aeronautics and Astronautics, 1982, pp. 268--278.
12. McLaughlin, PV, and Dasgupta, A., "BILAM: A Composite Laminate Failure Analysis Code Using
Bilinear Stress Strain Approximations," UCRL-15371, October 1980.
13. Tsai, S.W, and Hahn, H.T, Introduction to Composite Materials, Technomic Pub. Co, Inc., Westport,
Connecticut, 1980.
S 14. Jones, R.M., Mechanics of Composite Ma'erials, Scripta Book Co., Washington, D.C., 1975.
12-1
15. Crossman, FW., "Analysis of Free Edge Induced Failure of Composite Laminates," First USA-USSR
Symposium on Fracture of Composite Materials, Sijthoit and Noordhoff International Publishers,
1979, pp. 291-302.
16. Herakovich, C.T., "On Failure Modes in Finite Width Angle Ply Laminates," Advances in Composite
Materials; Proceedings of the Third International Conference on Composite Materials, Vol 1,
Pergamon Press, 1980, pp. 425-435.
17. Wu, E.M., "Failure Analysis of Composites VtAh Stress Gradients," First USA-USSR Symposium on
Fracture of Composite Materials, Sijthoff and Noordhoff International Publishers, 1979, pp. 63-76.
18. Wilson, D.W, Gillespie, J.W, York, J.L, and Pipes, R.B., "Failure Analysis of Composite Bolted
Joints," NASA-CR-16372, July 1980.
19. Starnes, J.H., Jr., Rhodes, M.D., and Williams, J.G., "Effect of Impact Damage and Holes on the
Compressive Strength of a Graphite-Epoxy Laminate," ASTGM STP 696, 1979, pp. 145-171.
20. Bathias, C., Esnault, R., and Pellas, J., "Application of Fracture Mechanics to Graphite
Fibre-Reinforced Composites," Composites, Vol. 12, July 1981, pp. 195-200.
21. McGarry, F.J., Mandell, J.F, and Wang, S.S., "Fracture of Fiber-Reinforced Composites," Polymer
Engineering and Science, Vol. 16, No. 9, September 1976, pp. 609-613.
22. Awerbuch, J., et al., "Determination Characteristics and Failure Modes of Notched Graphite
Polyimide Composites at Room and Elevated Temperatures," NASA-CR-159375, August 1980.
23. O'Brien, T.K., 'Analysis of Local Delaminations and Their Influence on Composite Laminate
Behavior," NASA-TM-85728, Ja-nuary 1984.
24. Whitcomb, J.D., "Analysis of Instability Related Growth of a Through-Width Delamination,"
NASA-TM-86301, September 1984.
25. Rothschilds, R.J., "Instability-Related Delamination Growth in Thermoset and Thermoplastic
Composites," Presented at ASTM Composite Materials Testing and Design: Eighth Symposium,
Charleston, South Carolina, April 1986.
26. Tsai, S.W, Composites Design - 1986, Think Composites, Dayton, Ohio, 1986.
27. Halpin, J.C., Primer on Composite Materials, Technomic Pub. Co., Inc., Lancaster, Pennsylvania,
1984.
28. Ashton, J.E., and Whitney, J.M., Theory of Laminated Plastics, Technomic Pub. Co., Inc., Stanford,
Connecticut, 1970.
29. Tsai, S.W, "Strength Theories of Filamentary Structures," in Fundamental Aspects of Fiber
Reinforced Plastic Composites, R.T. Schwartz an-.d H.S. Scharts (eds.), Wiley Interscience, New York,
1968, pp. 3-11.
30. Tsai, S.W, and Wu, E.M.. "A General Theory of Strength for Anisotropic Materials," J. Composite
Materials, January 1971, pp. 58-80.
31. Pipes, R.B., and Cole, B.W, "On Off-Axis Strength Tests for Anisotropic Materials," J. Composite
Materials, April 1973, pp. 246-256.
12-2
* 32. Christensen, R.M., "Tenser Transformations and Failure Criteria for the Analysis of Fiber
Composite Materials," J. Composite Materials, September 1988, pp. 874-897.
33. Kim, R.Y., "In-Plane Tensile Strengt. Multidirectional Composite Laminates," Technical Report
UDR-TR-81-84, University of Dayton Research Institute, 1981.
34. Gillespie, J.W, Jr., Microcomputer Software, Center for Composites Manufacturing, Science, and
Engineering, Newark, Delaware, 1984.
35. Flaggs, D.L., and Kural, M.H., "Experimental Determination of the In Situ Transverse Lamina
Strength in Graphite/Epoxy Laminates," J. Composite Materials, Vol. 16, March 1982, pp. 103-116.
36. Parvizi, A., Garrett, K.W, and Bailey, J.E., "Constrained Cracking in Glass Fibre-reinforced Epoxy
Cross-ply laminates," J. Materials Science, 13, 1978, pp. 195-201.
37. Bader, M.G., Bailey, J.E., Curtis, PT., and Parvizi, A., "The Mechanisms of Initiation and
Development of Damage in Multi-Axial Fibre-Reinforced Plastic Laminates," ICCM3, Vol. 3,
Cambridge, England, August 1979, pp. 227-239.
38. Bailey, J.E., Cu-tis, PT., and Parizi, A., "On the Transverse Cracking and Longitudinal Splitting
Behavior of Glass and Carbon Fibre Reinforced Epoxy Cross Ply Laminates and the Effect of Poison
and Thermally Generated Stain," Proc. R. Soc. Lond., 366, 1979, pp. 599-623.
39. Flaggs, D.L., "Prediction of Tensile Matrix Failure in Composite Laminates," J. Composite
Materials, Vol. 19, January 1985, pp. 29-50.
40. Lagace, PA., "Notch Sensitivity and Stacking Sequence of Laminated Composites," Composite
Materials: Testing and Design (Seventh Conference), ASTM STP 893, J.M. Whitney (ed.), ASTM,
Philadelphia, Pennsylvania, 1986, pp. 161-176.
41. Wu, E.M., "Fracture Mechanics of Anisotropic Plates," in Composite Materials Workshop, S.W Tsai,
J.C. Halpin, and N.J. Pagano (eds.), Technomic Pub. Co., Inc., Stanford, Connecticut, 1968,
pp. 20-43.
42. Waddoups, M.E., Eisemann, J.R., and Kaminski, B.E., "Macroscopic Fracture Mechanics of
Advanced Composite Materials," J. Composite Materials, Vol. 5, 1971, pp. 446--454.
43. Whitney, J.M., and Nuismer, R.J., "Stress Fracture Criteria for Laminated Composites Containing
Stress Concentrations," J. Composite Materials, Vol. 8, 1974, pp. 253-265.
44. Nuismer, R.J., and Whitney, J.M., "Uniaxial Failure of Composite Laminates Containing Stress
Concentrations," in Fracture Mechanics of Composites, ASTM STP 593, American Society of
Testing and Materials, 1975, pp. 117-142.
45. Karlak, R.E, "Hole Effects in a Related Series of Symmetrical Laminates," in Proceedings of Failure
Modes in Composites, Vol. IV., The Metallurgical Society of AIME, Chicago, Illinois, 1977,
pp. 105-117.
46. Pipes, R.B., Wetherhold, R.C., and Gillespie, J.W, Jrl, "Notched Strength of Composite Materials,"
J. Composite Materials, Vol. 12, 1979, pp. 148-160.
* 47. Pipes, R.B., Gillespie, J.W, Jr., and Wetherhold, R.C., "Superposition of the Notched Strength of
Composite Laminates," Polymer Engineering and Science, Vol. 19, No. 16, 1979, pp. 1151-1155.
12-3
48. Pipes, R.B., Wetherhold, R.C., and Gillespie, J.W, Jr., "Macroscopic Fracture of Fibrous
Composites," Materials Science and Engineerin , Vol. 45, 1980, pp. 247-253.
49. Lin, K.Y., "Fracture of Filamentary Composite Materials," PhD Dissertation, Departmcnt of
Aeronautics and Astronautics, M.I.T., Cambridge, Massachusetts, January 1976.
50. Mar., J.W, and Lin, K.Y., "Fracture Mechanics Correlatihn for Tensile Failure of Filamentary
Composites With Holes," J. of Aircraft, Vol. 14, No. 7, July i977, pp. 703-704.
51. Poe, C.C., Jr.. and Sova, J.A., "Fracture Toughness of Boron/Aluminum Laminates With Various
Proportions of 0* and t 450 Plies," NASA Technical Paper 1707, November 1980.
52. Aronsson, C.G., "Tensile Fracture of Composite Laminates With Holes and Cracks," The Royal
Institute of Technology, Report No. 84-5, Stockholm, Sweden, 1984.
53. Williams, J.G., and Rhodes, M.D., "Effect of Resin on the Impact-Damage Tolerance of
Graphite-Epoxy Laminates," ASTM STP 787, 1983, pp. 450-480.
54. Starnes, J.H., Jr., and Williams, J.G., "Failure Characteristics of Graphite-Epoxy Structural
Components Loaded in Compression" Mechanics of Composite Materials - Recent Advances,
Hashin and Herakovich (eds.), Pergamon Press, 1983.
55. Williams, J.G., "Effect of Impact Damage and Open Holes on the Compression Strength of Tough
Resin/High Strain Fiber Laminates," in 'rough Composite Materials - Recent Developments,
Noyes Publications, Park Ridge, New Jersey, 1985.
56. Wilkins, D.J., "A Preliminary Damage Tolerance Methodology for Composite Structures," in Failure
Analysis and Mechanisms of Failure of Fibrous Composite Structures, NASA CP 2278, 1983,
pp. 67-93.
57. Gillespie, J.W. Jr., and Pipes, R.B., "Compressive Strength of Composite Laminates With
Interlaminar Defects," Composite Structures, Vol. 2, 1984, pp. 49-69.
58. Webster, J.D., "Flaw Criticality of Circular Disbond Defects in Composite Laminates," CCM-81-03,
Center for Composite Materials, University of Delaware, 1981.
59. Ashizawa, M., "Fast Interiaminar Fracture of a Compressively Loaded Composite Containing a
Defect," Douglas Paper 6994, January 1981.
60. Ashizawa, M., "Improving Damage Tolerance of Laminated Composites Through the Use of New
Tough Resins," Douglas Paper 7250, January 1983.
61. Ramkumar, R.L., "Performancr' of a Quantitative Study of Instability Related Delamination
Growth," NASA Contractor Report 166046, March 1983.
62. Ramkumar, R.L., Kulkarni, S.W, and Pipes, R.B., "Definition and Modeling of Critical Flaws in
Graphite Fiber Reinforced Epoxy Resin Matrix Composite Materials," Naval Air Development
Center Report No. NADC-76228-30, January 1978.
63. Chatterjee, S.N., Hashin, Z., and Pipes, R.B., "Definition and Modeling of Critical Flaws in Graphite
Fiber Reinforced Resin Matrix Composite Materials," Naval Air Development Center Report
No. NADC 77278-30, August 1979.
12-4
* 64. Chatterjee, S.N., and Pipes, R.B., "Composite Defect Significance," Procecdings of the Mechanics of
Composites Review Meeting, Dayton, Ohio, October 1982.
65. Yin, WL., Sallam, S.N., and Simitses, G.J., "Ultimate Axial Load Capacity of a Delaminated
Beam-Plate," AIAA Journal, Vol. 24, No. 1, January 1986.
66. Griffith, A.A., "The Phenomena of Rupture and Flow in Solids," Phil. Trans. Royal Soc. of London,
A221, 1921.
67. Rybicki, EF., and Kanninea, M.F, "A Finite Element Calculation of Stress Intensity Factors by a
Modified Crack Closure Integral," Engineering Fracture Mechanics, Vol. 9, No. 4, 1977, pp. 931-938.
68. Johnson, WS., and Mang-algiri, PD., "Influence of the Resin on Interlaminar Mixed-Mode Fracture,"
NASA 'PM 87571, July 1985.
69. Wilkins, D.J., "A Comparison of the Delamination and Environmental Resistance of a Graphite/
Bismaleimide," NAV-GD-0037, General Dynamics, Fort Worth, Texas, September 1981.
70. Russell, A.M., and Street, K.N., "The Effect of Matrix Toughness on Delamination: Static and
Fatigue Fracture Under Mode II Shear Loading of Graphite Fiber Composites," presented at
NASA/ASTM Symposium on Toughened Composites, Houston, Texas, March 13-15, 1985.
12-5
A-i
Table A- 1. Keywords, Sources and Abstracts in Literature Search of DTIC, NASA and PlasticsCenter
Data Bases
PEEK/GRAPHITE NASA 86
BISMALEIMIDEIGRAPHITE NASA 28
0-
0-
A-2
Table A-2. Comme;cial FiberProperties
A-3
Table A-3. Propertiesof Carbon/EpoxyPrepreg (Page 1 of 2)
Preprey Type/
Menufacturer Product Number OrnenLellon Preprog Cure Resin Content (%)
AMERC.;AN
C'ANAMID CYCOM 907/T 300 UNID4TIECIONAL 40.L 3
CYCOM 9I91 300 F- 34 WEAVE 45_.2
CYCOM 985.'CELION UN'DInlCTrIOENA. TAPE 145 37 2.
CYOOM D85/CEUON 7 MILP.AIIJ WEAVE 40A 2
CYCOM Q&5/T.300 UN:ODIEC'TINAL TAPE 3- 3
CYCOM 9A/rT.300 FP134 WEAVE 42 .2
CYCOM 980/T.300 YARN F 134 WEAVE 42# 2
CYVVOM 980f-300 UNIDIRECTIONAL TAPE 37. 3
A-4
Table A-3. Properties of Carbon/Epoxy Prepreg (Page 2 of 2)
15 5 (50OPS-,350 F) 1 6 (70 F)
12 5 (50PSi;250 F) 15 .
24 7 (50PS:/500 F) 20
16.7 (50 PSi/S00 F) 23
18*: 8(50 PS,/350F) 10 . I15 3 (KO F)
10 61573 (350 F)
18 7 (50PSI/350 F)
1578 (50 PSI/350 F) 10 . !. 3 (35.0F)
1.0 6a 15 3 (35WF)
15 ;a (50 PSI/350 F)
2 3. 14 (27S F)
5.-19 (50 PSI;?75F) . -14
!5 45 (50OPSI/350 F) 2 a (40 F) 14
155 (100 P51/350 F) 1.5 a 21 18.5 (350F)
2 (2250 F) 13(25o F)
7 (I5PSIf250F)
15 110OMtN1325 F) . 420
4 (325SF)
4 15 (15 PSI/324 F)
1 1 (I5 M!N"?1S F) 10-2S (350 F)
162 (IS PSI1/3WF)
0 1 (10 MINi2S F) 7 66 (250OF)
75 10 MIN/i5psl/250F)
09 (10OMIN/?SO F) M-EDIUM .. 12 9 (25-0F)
08 (IOMIN/ISPSI/250F)
1.5 (10 MIN)275 F . 4-2C (27SF)
5 -20 !15 PSI/VS5 Fl .
F) 4-20 (275F)
4-15 (10 PSI/VS F) 1-5 (10 MINJ27C3 ..
8 (250 F)
10 (10 MIN/S PSI/250 F) 2 (10 MIN/250 F) .
15 370 14 12 10 3 -7 (350F)
15 . 370.L14 12 10 6 - Q (W50F)
IS . 370.+14 12 IC 3 -7 [350 F)
'S5 370 ;14 12 10 6 -12 (3510F)
1IS 19378 12 10 3.-7 (350 F)
1.5 - 138 12 10 6- 12 (350F)
IS5 148 12 14 7. 14 (250 F)
i 1. I
12 10 6- 12 (350 F)
1 146 12 10 3- 7 (350 F)
1 . 146 12 14 7.-14 (250 F)
1S5 1 46 12 14 7.-14 (2-SOF)
1 . . 164 12 10 12 -32 1350F)
1 150 12 10 a-12 [3W0F)
1 . . 150 1?10 6-I2 (350 F)
1 150 12 10 3 -7 (350 F)
I ISO 12 10 3 -7 (350OF)
1 5 12 0 6- 12
IS5 75-200 12 7 8-20 (350F)
I1 . 355-385 12 7 8-20 1350F)
1.0 - 150 12 14 2.5 (350F)
1 150 12 17 19 (350 F)
1S 145 12 14 5 -12 (250 F)
1 150 1? 10 3 -7 1350F)
1 140 12 10 6 -12 (350 F)
1 ~150 1? 10 8 -12 (350 F)
1S 185* a 1 10 3 -? (350 F)
15 185 a 12 10 6 -12 (35DF)
15 . 185.t8 12 to 12-32 (35oF)
IS5 370 +14 12 10 12-3? (350F)
15 . - 370,.14 12 10 12 -32 (350 F)
IS5 193 ; 1? 10 Q2-32 (350F)
IS5 2607 10 12 10 3.-7 (350 F)
VS . 2807 `10 12 10 - 12 (350 F)
1.5 . 280 10 12 10 Q2- 32 (350 F)
- 5 I12 on
20 - . 1.5
* . 193
A-5
Table A-4. Propertiesof Glass/Epoxy Prepreg (Page 1 of 2)
Prepreg Type/
Manufacturer Product Number Orientuton PReln Content (%) Resin Flow (%)
AMfERICAN
CYANAM4IO CYCOM 5 143 35-40 20 .4 (50 PSI/325 F)
C tCOM 5105 - 16-40 10-20
CY-".RNNiS4 REGULAR GEL.GPA.E C/104 75 *.5 507 (200 PSI/3ao F)
CYTRON 5104 REGUIAR GEL-GRADE C/108 6&55 44 . 7 (200 PS/340 F)
CYTRON 5104 REGULAR GEL-GRADE C/ 112 60s 5 40_.7 00P;,3,40F)
CY 5104 REGULAR GEL-GRADE C/I 13 60. 5 40-_7 ( 00PS;/340 F)
CYTRON 5104 REGULAR GEL-GRADE C/1BI 55_. 5 35+7 (200 PSi/3*o F)
CYTRON 5104 MEDIUM GEL-GRADE 0/104 75 + 5 45- 7 (200 PSI/340 F)
CYTRON 5104 MEDIUM GEL-GRA0 0/106 65,8 5 40 7 (200 PSI/340 F)
CYT'RJN 51D4 MEDIUM GEL-GRADE O/112 60. 5 38 ;7 (200 P-I/340 F)
c O 1O4 MEDIUM GEL-GRADC D0113 S0I 5 36+ 7 (200 PSJ/340 F)
CYTRON 5104 MEDiUM GEL-.GR.AE D/116 56 5 32 . 7 (200 PSI/340 F)
CYTRON 5150B 104 75 5 505 rd00Ps/340F)
55 305 (" PSI/340 F)
CYTON 5`1508 10o
112 6 5 305 reOo PS1/340F)
CYTRN 51500
113 500. 30+5 (20 PS.1/34o F)
CVYTON 51506
CYT"RON 51506 116 5284 3
225 5 (200 PSI/340 F)
CYCOM 965/E.GLASS 7781 &-H SATIN 36. 2 20 7 (M PSj/30F)
CYCOM 9W/F-GLASS 7781 Al 100 3417 2 20 5 (50 PS0/350 F)
CYCOM 512/E-GLASS 7501 35-45 15-25 (50 PSI/325 F)
CYCOM 513,4 7581 38.2 134.7 (77 F)
A-6
Table A-4. Propertiesof Glass/Epoxy Prepreg(Page2 of 2)
Votatie Coflmna (%) Tack of"Ap Shft LUfe(dars) Out Time (days a 77 F) Gel Time (minutes)
2.2 (0 F)
W160 30
2 (,OF)
e16 10 3-6 (275F)
2 180 (40F) 14
1 160 14 8.t2 (325 F)
12
1.2 9Q2 rm
0
1 4 (27SF)
2.0
05
0.5
05
1.5
A-7
Table A-5. Properties of Key/ar/Epoxy Prepreg (Page 1 of 2)
Preprea Type/
Manufacturer Product Number Orientaton Resin Content () Resin Flow(%
A-8
Table A-5. Propertiesof Kevlar/Epoxy Prepreg (Page2 of 2)
3 (40F) 10-1,_
2
3 (401F)
2 (IOMRN1325F)
a (OF) 10 17 3 (2WOF)
1(1
6 (OF) 10 1713 (250F)
1.0
20
1.0 - 10 F) 15L3 (350F)
a (0F) 115+3 (350 F)
1.0
1.4 3 (40 F) 2.0o(325 F)
8.5 (325 F)
1.14 (10 MIN/325 F)
18 (325F)
10 (10 MIN/32S F)
18 (32SF)
1.2 (I MIN/325 F)
MEDIUM 4-10 (350F)
3 (IOM'N/325F) MEDIUM
2.0 (1OMIN/325F)
3.5 (MS1F)
1.1 ('OMIN/250F) MEDIUM
3.5 (250 F)
1.1 (IOMIN/250F) MEDIUM
MEDIUM 3.3 (250 F)
1.4 (IOMIN/250F)
11.3 10.08 (250 F)
3.8 (250 F)
MEDIUM 8 (260 F)
2 (15 MN/260 F) MEDIUM
0.8
21-23 10 4 4 (250 F)
3 (- F) - -
9 (250 F)
1.5 (lOMIN/25OR -
I (8M.N/325F) 24.3
27.9
I (8 MIN/235 F)
1 (1SMIN/325F) 16 (350F)
16 (350 F)
I (15MiN/325F)
:2.4 (325 F)
1.2
2 (8 MINt275 F) 2.7 (275 F)
2 (8 MIN/275 F)-
4-10 (275F)
2 (SMIN/275F)
4-10 (275F)
2 (8 MIN/275F)
I (IOU1IN/250F) 10 (250F)
A-9
Table A-6. Propertiesof Glass/PolyimidePrepreg (Page 1 of 2)
Prepreq Tyne!
Manufacturer Product Number Orientatoln Lamin1e Cure Resin Content (%)
F114 25-45
HE.XCEL
A-10
Table A-6. Propertiesof Glass/PolyimidePrepreg (Page 2 of 2)
8.7 (5 MIN/6OWF) 3 10
20-2 (IOMIN/15PSI/350F) 3 7 30 (300 F)
2030 (IriMINJtSPS11350F) 12-16 ((0 MIN/350F) 3 7-10
11-17 (IOMIN/i15 SI/350F) 8-8.5 (10 MIN/350F) 3 7-10
NOT APPLICABLE 9 f 3 (5 MIN/eOo F) 3 10
17.5 (10IIN/200PSI/370 F 0.37 IOMIN/370F) 00
0-15 3
A-11
Table A-7. Properties of Carbon/Epoxy Laminates (Page 1 of 2)
L-I-1i.l "Mingo17Type/ Long -in L77q
Me77ti7oIu8 P1.ducINum0.1 Pib~t po.Curt 01I.tA101on (ko1/ ik.11 '0141
* 07 ;.90
5 Z0741NA0 '6000 1610703
67 lses
5 MIX 91000 - 0300.
as '60 1 70383 '65 200
'00 7?90
* 03 11 SOS 0000 WOO
* 02 0bowSB 0000
ON - 0
IQ 1- 8730-0 10 10 700
1. 0000133
- 0385
10 700 000 00. 0901.7 -
70.OB - 077R!7T7000 bMW .00
u
12* * 0n;4000Z074AL 37n 7MW0 0= .
. - 7n40ZOlTAL
I. '7270M 79200*
- - 790718IOOndTAL
S 20800 2790
& -
7000
loa0700 1300 00703 0 00"
:27. 226 - 033
08 2u0 - 083 22WO00 3MW 17300 *00150 00070
* nOZNTI
2 00S00n 7 00IMOCW 0200
02. 09 775 *00L0 7373 0010 700M0 7240 070 aMW
79 Igo . 708
ee 0000 7a 000 'M 7700 79000 -
* . - 'S38S 707770000
- 74380 240010
. 838
a I 33720 40401)-
9. sw 4=0
200 23000
as 003 - a08
So3M 300 003Q 760 33490
* 00? 3138 1
3000790
a, 74 0911090208T9 7MWO 9700 7..100 777100 030
* . . 79 36M0 .000
30 qM 32000
a12 s
21000 2300
'775 IQ 27000
17 2000 220W
I7 27700 27100
4M02 700
0000 7004W0
00.
62 am 70
* a0s 76000
00 05083 7MW 1M
70 1 , 3WAP30714gas 206000 70 - -
77028= 30073000
110sIsO 20=0 23000
7MW 2700-0
* ~* ~ io3383
7732700
S"
I79955 7901800
'$00000900
251M0000 07
11036 177300 70000 00702 0077
07080
SS 10 7000
2MW 7000000
79008 MWio 0
10 9,505 09500
100 0958S 7MW
10000s
95 74s
95 2MWD am
* .79 aM 20= 00000 00
'9 7M 37Q
0000 000 007
* 00 07193 17MW 7780 7M 0 000 0 0'3
700 018 9030 7Mo
0W 10200 003 00M 01
' 00 09389 OW 7M0W am0 000
.
a00083S 26aDO
2w
-'.7500 20900 20000
79 1160 9Z 9 7800 7MOW MW0 70000 MW
1M 7 =17000 7011 7770 007770
790 70 20 70=0 7700 0078
309 70
i- 340200 ?am0 0 875
274 785
2207270 251
20000 72370 2a.00
007b2
00100 - -
LI - U1-n.e.-. SM .w.e
I517 Z1,dC..
?a . 77 3w1
60 . 3800 ja 4.fl
3100 ROD00
to 056
64.WAh~5ON7AL. SI " 4300
75.0 g0 40 30 300D 5000
45 34 3400
30 33 2000 2500
- 75.V770f20710 6.00
-t v4.H-IOR20TAI. - 506006
40
-35 .200 450 .Q
03,000M
123 WO370
624 3500 WOO360
574 .3700
62 6064400 3600
63 W1OO
57
* 300
,6 2610 320
5754 330w
so 42 40Ole*07
GO 3700 3600 W
6768 3600 3's 3730D
so 4 -3000 3400D 3400
757 31,00 5*60450
Go6 - 2 134000'3 30
,67. 2030 070 3I00sa a
S0.6 3150i40
704 3A00 320 3400
57~so . 6000033
15
30 3400 39D0
5. 3 400 l400 -. 00
60 35 000 3200 30
675 .8 0 .00 - 3400
0,
74455X 31,300 3700
656 -2 3450 3400
760 035 1300 3M20 4070
41O4U0
04 - 2"47
Is 2- .110 3Aa400
45? 2S &W40200
.. 0 60 7736
8
94 :460.0
54 4r,5 A-1520S
7able A-9. Properties of Key/ar/Epoxy Laminates (Page I of 2)
UFS -Ullirnate Flexrurel Strength USS - UtIlmCJOShoarStf.ngth FM.- FlexurarlModulus Long - LongiludInaJ
IllS -Ultimate Tensile Stioingth ISS - Inlarlenriner
Sheef Strength TM - Tensile Modulus Trans - Tlensoerse
UCS - U111111141
CornpresIvo Strength CM - Cornprerrrelve Yodulue
A-16
Table A-9. Propertiesof Kevlar/Epoxy Laminates (Page2 of 2)
80 17 3300 3300
75 23 4000 3500
70 25 7.4 3800 4400
67 23 7.7 3700 4700
406 4.2 4110 4400
480.0 OSSS 4200 4300
49.0 52 340 4400
07 20 4.8 2.3 .3500 3700
70 23 5.4 2.7 3900 3600
a1 27 - 3000 A200 3700
75 24 2800 41003500
A-1?
Table A- 10. Propertiesof Glass/PolyimideLaminates (Page 1 of 2)
Laminate Type/
Llanu/ecturer Product Number Fiber Reain Orlentation
UFS - Ulmalne Flexuralsivength USS - Ultlmale Shear Slrenglh FM - Flexural Modulus Long - LongitudInaJ
UTS - Uftlmaile Tensile Strength ISS - Intedamlnwa Sheow Strength TM - Tensile Modulus Trans Transverse
L-CS - UP!rnpje CompresovNe Sltrngth CM - Comnpresslve Modulus
A-18
Table A- 10. Propertiesof Glass/PaolyimideLaminates (Page2 of 2)
96.9 7.55
84.2 63.4 70.7 5.07 4370 3920 4230
HIGH PRESSURE 85.C 57.0 3200
VACUUM BAG 83.5 50.0 2800
103 82.0 78.4 10.6 4230 4450
84.0 79.8 3.11 31510 3W40
VACUUM CURE 79.2 60.9 50.8 10s1 HORIZONTAL 3.85 3350 3300 4470
AUTOCLVE 81.8 61.2 63.1 10 68 HOFIZONTAL 3.15 3500 4040 4670
01
A-19
Table A-i11. Properties of Carbon/Polyimide Laminates (Page I of 2)
UNIDIRECTIONAL 264 -
GENERAL DYNAMICS NR15062 --
284
NR15OB2SSX UNIDRECTiONAL -
USS - Ult1mate Sheow Strength FMA- Flerrur*J Modulus IFS - Tensil Faflure Strain
UPS Ultimate Flexrural Sirsnglh
IS8 - Interlarnlnex Shear Strength TWO-I ensle Modulus Long - LorwltudineI
UTS - Ultimele Tensale Strength
- Ultimate ComrpreseelvStrength
UJCS CM - Comrasstwa Modulus Trants - Transversee
A-20
Table A- 1i. Propertiesof Carbon/PolyimideLaminates (Page2 of 2)
14.5
12.1 SS
7.5
* 19100 20000
77 5.6 C 99M 10400 960
17.9 HOI.ZONrAL
9.0 16.3 HOFUZONTAL 19600 21800 1400 0.01047
10.8 HORZO'TAL - 195W
W8.8 7.0 HORIZONTAL Rw 8100
A-21
Table A-12. Properties of AblativelPhenolic Prepreg (Page 1 of 2)
Ptepma Type/
Manufacturer Product Number Grt~lenloil Resin Content () Resin flow()
A-22
Table A- 72. Propertiesof Ablative/Phenolic Prepreg (Page2 of 2)
o0 (1OMIN/325F) 3 (40F) 30
4 (10 MIN/325 F) 3 (4OF) 30
4 3 (40 F) 30
6 (10 MN/323 F) 3 (40F) 30
4 (15 MIN/325 F) 3 (40 F) 30
7-13 (10MIN/325F) 3 (40F1 30
6-13 (t0MIN/32.F)} 3 (40F) 30
5 (t0 MINJl.32 F) 3 (40F) 30
2-6 (W0MIN/325 F) 3 (40F) 30
2-6 1OMIN/325 F) 3 (40F) 30
9.0 3(40 F) 30
4.? (10 MIN/32,?F) 3 (40F) 30
4-7 (10 MIN/32 0F) 3 (40F) 30
2-7 (10 MIN/32 F) 3 (40F) 30
6 l
(1C MIN/325fim 3 (40F) 30
5 (10 MIN/32 9F) 3 (40F) 7
6.0-9.0
A-23
Table A- 13. Properties of Ablative/Phenolic Laminates (Page 1of 2)
CORPORATION
FERRO~ SILICA/PHENOLh: CA-2204 48 FABRIC
SIUICA/PHENOLIC CA-2221 48 FABRIC
SIUICA/PHENOLIC CA-2221 MC 48 FABRIC
SIUCA/PHIENOUC CA-2230 48 FABRIC
SIUICA/PHENOUIC CA-2230 MC 48 FABRIC
ASBESTOS/ lIENOLIC CA-2212 APO 40 RAYBESTOS-MANHAT FAN
ASBESTOS/PtiENOUC CA-222 RPO 40 RAYBESTOS-MANHATTAN -
ASBESTOS/PHENCUC CA-2249 RPD AORAYBErSTOS-MANHATTAN
GRAPHI-TE/PHENOUC CA-8207 FABRIC
GRAPHITEPHENCIUC CA-8"0 FABRIC
CARBONJPHENOUC: CA-8214
CARBON/PHEKROUC CA-822 FABRIC
CARBON/PHENOUC; CA-821XI FABRIC
QUARTZ/PHENOUC CA-22'0 ASTROOLJARTZ 570 FABRIC
OUARTZ/PHENOUC_ CA-2227/581 PABRIC
CARBON/PHENOUC CA-8217 FABRIC
A-24
Table A- 13. Propertiesof Ablative/PhenolicLaminates (Page2 of 2)
A-25
Table A- 14. Propertiesof Glass/PhenolicLaminates (Page I of 2)
UFS
Laminate Type/ Long
Manulacturer Product Number Fiber Resin Orientation Laminate Cure (Nei)
AMERICAN CYANAMID 6101 2 7501 (Al 100 FINISH) CYCOM 6101 2 FABRIC 730
6102 181 CYCOM 6102 FABRIC 73-0
A--
A-26
Table A- 14. Propertiesof Glass/PhenolicLam~nates (Page2 of 2)
A-27
Table A- 15. Propertiesof GlasslPolyesterLaminates (,Page 1 of 2)
Laminate Type/
Manuftcturer Product Number Fiber Resin OrientAtlon
A-28 -
Table A- 15. Propertiesof Glass/PolyesterLaminates (Page 2 of 2)
A-29
APPENDIX B
This appendix lists a compilation of chemicai and mechanical properties of composite materials
compiled by Boeing under Air Force Contract F,33615-86-C-5071.
0
B-i
Table B- 1. MechanicalPropertiesof Hercules Fibers
Carbon content. % 92 94 94 94 94 -
3
Density. g/cm 1 80 1.80 1.80 1.83 1 73 1 78
Typical epoxy corn posite ASI AS2 AS4 A56 1M6 IM7
properties at RT
F6ber volume. 62 62 62 62 62 62
B-2
4800 MPa 300 GPa
(43 5 msi)
(696.7 kis)
-Strength
4600
(667.6)
' Modulus
4400 - 280
(40 6)
(638.6)
4200
(609 6)
4000 260
(580 6) (37 7)
3600 _ 240
(522 5) (34 8)
3400
(493 5)
3200 1 - 220
(4644) ' (319)
I
3000
(4354)
2 0
280020
2800 (290)
(406 4) "(
Fiber type
3
(Fiber density, g/cm )
0
B-3
3000 MPa 190 GPa
(435 4 ksi) (27 6 msi)
180
2800 Strength
(406 4) ;-- Modlus (261).
2600 170
(377.4) (247)
Tensile Tensile
Strength 240U 160 modulus
(348.3) (23 2)
2200 1S0
(319.3) (21 8)
2000 140
(290.3) -- (203)
1800 130
(261-2) - - '-(189)
AS1 AS2 AS4 AS6 IM6 IM7
1-4
120
100
B0-1
Strength-to-density 60o.y 0~ C i
ratio .
0 82.0 i
0
40 =
00 60
2.0
Composite tyxe
(36.3)
200 0
(29 0) Boron/epoxy
150
(218)
0o
Tensile
modulus 100
(14 5)
Kevlar 49/epoxy 0
50 GPa 5-glass/epoxy
(7 3 msi)
E.glassrepoxy
0 1f 1 , I I I I 1,
1000MPa
(145 1 kM) 1200 1400 1600
(174.2' (203 2) (232 2)
0* Tensile strength
B-5
E U, 2
N %
Cf
CI io
0 l LA 0
EN N , N , U
!! %D 0v 0l '0
"0 N
40N , U
-1 m m Ln N
00 M
40
r.. ~ k N
V,~. 0 '
P4 ^
~E0 -n LM
.61 0% u10 l C
N Q
W 0 1
E %0 IWO N UMU
00 kn %nU 0 1
00 0c 0co0
0. 0
1.0 0 no 0 0%,
tC = t; 2 0 t! . o
0Vic . 0 0 !. r 0v
4,c ft c
m 3. Cl
O . M. oN P % r V P %A
0 N
B-6
0
160
(23.2)
(4)
140
(20 3)
(5)
120 (3)
(17.4)
Tensile
modulus
100
(14.5)
so (1) (2)
(11 6) (D
60 GPa
(8 7 ms,)
(7) (6)
0 (D a I t I I 1 1
200 MPa 400 600 800 1000 1200 1400
(29.0 ksi) (58 1) (87 1) (116 1) (145 1) (174.2) (203 2)
Tensile strength
B-7
'0 co V% %D-
-Q U-m '00
0O m 0' m 0 0 m
0 CDLA a 0 vN 0 mInM
v 44
Z 0 O 'M 4 m 40 0
4 4o
00 N
ON ON ~ so ". io ~ m'
o- Cm ' 0 - CL mNC
-10 m ~ N Go
CL em N4 0 UE 4 . N
'a
- .5
Q'.0'.N NO m m
B-8
60
(087)
HM
40
(05S8)
IMLS e IMHS
Tens,ie PMR
modu Iu',
2 0GPa
(0 29 ms,) LM
Polyinmde
00 f L I LI 1 I1 I
Legend
LM = Low modulus
IMLS , Interrned-ate modulus lovv strength
IMHS Intermed-ate modulus high strength
HM - High modulus
B
B-9
120
100
Ii
soj
E
Strength-to-Density Vz .7
Ratio
(Normalized) E
60 ., G"
'8 "E
40
E cII -S
20 -
Matrixtype
Polyester Thermoset 1.1 -1.4 1.3 - 4 1(0-2 -06) 40-85 (6- 12)
Acetal Thermoplastic 1.4 35(05) 70(10)
B-10
100
(14.5)
90
(1 3)
90 ," -Modulus
(13 1) _ (1 2)
0
E
(11.6) w. Cc (10)
E Z
0U 0
80 0 E
E, i (0
(10.2) 2, (0 673)
gc3 -i-
L" Z
-. (058)
(073)
60 XI 50
I~L --- j
30 - - - 20
(44) (029)
20 I i 20
1
(2 9) (0 15)
10 NAN
(11.5 ksi) _ i I j OGPa
(0 MS,)
B-1 1
IVI
SI 0
* I I
I "-
Stress corn prisi e I
I I I
I I-
Strain
composttei
Stage I - elastic deformation Of both fibers
and matrix
Stage 11 -elastic deformationof fibers, plastic
deformation of matr~x
Stage III - plastic deformation of both fibers
and matrix
Stage IV - failure of both fibers and matrix
FigureB-9. DeformationStages of Fiber,Matrix, and Composite
In a typical composite
B-12
Table B-7. Propertiesof Kevlar and Glas- Fibers
Specific gravity. y 2 60 2 5 1 44 1 44
1 In a typical composite
0
Table B-8. Tensile Strength of Neat Resins
MPa ksi MPa ksi MPa ksi MPa ksi MPa ksi MPa ksi
B-13
0 2 0e
3:EStN
VtxI~j i II I
Mpa~uI
Vqe:uo06#
I
v 06kNOHAPW~U)~2W
GE
LI7T~ME 5811JaH
06OP'qn:j uo~un
m
) A k)P!W Iu Cut u) u W'
,PVJua
iLO 6 V YO
I~f
NI I S
B-14a
Table B-9. Young's Modulus of Neat Resins
Fibredux 914 - epoxy 40 058 31 045 32 04b 2.1 031 07 010 03 004
Hercules 2220-1 -epoxy 30 043 3 1 045 2.6 038 2 1 030 2.2 032 10 0 15
B-15
0.7
00
0%
C5 (D 0 0 0
DSVZS ~ T ~ 030%N
[I LOS lSatflJHeI
L0
t-OZZl $ajn)JaH
Z OS~ ajlnJaH
I o6t, apiqjeDuo~uri
L06V )LJk piweueuA) ue'awV
* L6O~
$alfl)JO
tOE Z
sajaanNa
- -.
Vt067pijpul0
Voi 16DDp.u.A rlai
-E I - 7 -
co 0 0 N0
IN a 10 CI c 0 0
C
'In I-
00)
C0 In
CD C- In In
01 0% 0 0 In 10 N 10 1
CD 10 o0 00 0- Q%
~0 0 0 0 I Iw
0
20 n I
En In
"0' 6' 0 6 I n 11
1 10 Q' 0,
1
CD o w0 V0 m** V%
O 0 n 0 002 4
I :7 0'
(D C) C.4 In In1D1
0 -
0.) 1 Go 10
I-,0
0-0
-~~~ - ~~~
'DE ao
0 0 0 0
0. _____________________B__17_
Table B- 11. Propertiesof CommercialCarbon Fibers
Soo
(72 6)
Celion
GY-70
400
(58 1)
Tensile
modulus
300
(43 5)
Ceehon WM6
3000
T-300
O O AS6 ---
G 0 AS4
AS1
200 GPa
(29U in)
1000 MPa 3000 5000
(1 5 1 ksi) (4354) (725 7)
Tensile strength
B-18
120
10.0
4so
E 5,
CP
60 00
'A;
Fiber type
FigureB- 13. Strength-to-DensityRatio of Commercial
Carbon Fibers
B-19
0- L -
00 -Z
~0
c - o ca O
~~f*C o~-
qo
in
I..
q)
91
.CU ' ~ ~ I0 %
04
00
a, 0 c0 40 . a,
W,
0
US GP 92 4-9
'N B-20D