I I I fIt I

NTIAC-89-1

NONDESTRUCTIVE EVALUATION OF

ADHESIVE BOND QUALITY

STATE-OF-THE-ART REVIEWLnoSwRI Project 17-7958-838

by

G. M. Light

Hegeon Kwun

SOUTHWEST RESEARCH INSTITUTESan Antonio, Texas

i DTIC DELECTE

L12 M

June 1989

NONDE TRUCTIVE TESTING INFORMATION ANALYSIS CENTER

Approved for public release; distribution unlimited

This document was prepared by the Nondcstructive Testing Information Analysis Center (NTIAC).Southwest Research Institute. 6220 Culebra Road, San Antonio. Texas 78284. NTIAC is a full serviceinformation analysis center sponsored by the U.S. Department of Defense, serving the information needsof the Department of Defense, other U.S. Government agencies, and the private sector in the field ofnondestructive testing.

NTIAC is operated under Contract DLA900-84-C-0910 with the Defense Logistics Agency. Technicalaspects of NTIAC operations are monitored by the Office of the Undersecretary of Defense. Research.and Engineering.

I

I I - '

I.

This document was prepared under the sponsorship of the U.S. Department of Defense. Neither the UnitedStates Government nor any person acting on behalf of the United States Government assumes any liabilityresulting from the use or publication of the information contained in this document or warrants that suchLruse or publication of the information contained in this document will be free from privately owned rights.

Approved for public release, distribution unlimited.All rights reserved. This document, or parts thereof, may not be reproduced in any form without written 1permission of the Nondestructive Testing Information Analysis Center. L

Lk

I

I

I NONDESTRUCTIVE EVALUATIONOF ADHESIVE BOND QUALITY

State-of-the-Art Review

SwRI Project 17-7958I

byII

G. M. Light and Hegeon Kwun

SOUTHWEST RESEARCH INSTITUTESan Antonio, Texas

I

June 1989

INONDESTRUCTIVE TESTING INFORMATION ANALYSIS CENTER

Approved for public release; distribution unlimitedIi

89

This document was prepared by the Nondestructive Testing Information Analysis Center (NTIAC),Southwest Research Institute, 6220 Culebra Road, San Antonio, Texas 78284. NTIAC is a fullservice information analysis center sponsored by the U.S. Department of Defense, serving theinformation needs of the Department of Defense, other U.S. Government agencies, and the privatesector, in the field of nondestructive testing.

NTIAC is operated under Contract DLA900-84-C-0910 with the Defense Logistics Agency.Technical aspects of NTIAC operations are monitored by the Office of the Undersecretary ofDefense, Research, and Engineering.

This document was prepared under the sponsorship of the U.S. Department of Defense. Neitherthe United States Government nor any person acting on behalf of the United States Governmentassumes any liability resulting from the use or publication of the information contained in thisdocument or warrants that such use or publication of the information contained in this documentwill be free from privately owned rights.

Approved for public release, distribution unlimited.All rights reserved. This document, or parts thereof, may not be reproduced in any form withoutwritten permission of the Nondestructive Testing Information Analysis Center.

UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAGE

Form Approved

REPORT DOCUMENTATION PAGE OM8No. 0700188

la. REPORT SECURITY CLASSIFICATION lb. RESTRICTIVE MARKINGS

Unclassified

2a. SECURITY CLASSIFICATION AUTHORITY 3. DISTRIBUTION /AVAILABILITY OF REPORT

Approved for public release;2b. DECLASSIFICATION / DOWNGRADING SCHEDULE distribution is unlimited

PERFORMING ORGANIZATION REPORT NUMBER(S) 5. MONITORING ORGANIZATION REPORT NUMBER(S)

NTIAC-89-1

a. NAME OF PERFORMING ORGANIZATION 6b. OFFICE SYMBOL 7a. NAME OF MONITORING ORGANIZATIONNTIAC (If applicable) Office of the Undersecretary of DefenseSouthwest Research Institute Research and Engineering

C. ADDRESS (City, State, and ZIPCode) 7b. ADDRESS (City, State, and ZIP Code)

P.O. Drawer 28510

San Antonio, TX 78284 Washington, DC 20301

a. NAME OF FUNDING/SPONSORING 8b. OFFICE SYMBOL 9. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBER

ORGANIZATION (If applicable)

Defense Logistics Agency DTIC-DF DLA900-84-C-0910

r- ADDRESS (City, Stare, and ZIPCode) 10. SOURCE OF FUNDING NUMBERS

PROGRAM PROJECT TASK WORK UNITCameron Station ELEMENT NO. NO. NO. ACCESSION NO.Alexandria, VA 22304

1. TITLE (Include Security Classification)

Nondestructive Evaluation of Adhesive Bond Quality

2. PERSONAL AUTHOR(S)G. N. Light and Hegeon Kwun

13a. TYPE OF REPORT 13b. TIME COVERED 14. DATE OF REPORT (Year, Month, Day) 15. PAGE COUNT

State-of-the-Art I FROM 8/87 _TO 8/88 June 1989 556. SUPPLEMENTARY NOTATION

7. COSATI CODES 18. SUBJECT TERMS (Continue on reverse if necessary and identify by block number)FIELD GROUP SUB-GROU Nondestructive Testing/ Adhesive Bond Strength/

Adhesive Bonds Sonic Testing

I Adhesive Bond Testing Ultrasonic Testing ,

.\ABSTRACT (Continue on reverse if necessary and identify by block number)

his state-of-the-art report describes the bonding process, the destructive methods used

to measure bond strength, and the various NDE methods that have been evaluated for deter-

mining the quality of a bond. These NDE methods include sonics, ultrasonics, acoustic

emission, nuclear magnetic resonance, x-ray and neutron radiography, optical holography,I and thermography. Each of these methods has shown

some limited success in detecting

debond conditions. At the present time, however, it appears that only the sonic, ultra-

sonic, and nuclear magnetic resonance methods have the potential capability to differ-

entiate qualitatively the grada ions between a good bond and a debond and thus provide

a correlation to bond strength.

20. DISTRIBUTION/AVAILABILITY OF ABSTRACT 21. ABSTRACT SECURITY CLASSIFICATIONCE UNCLASSIFIED UNLIMITED 1:1 SAME AS RPT 0] DTIC USERS

. NAME OF RESPONSIBLE INDIVIDUAL 22b. TELEPHONE (include Area Co e) I22c. OFFICE SYMBOL

Form 1473, JUN 86 Previous editions are obsolete. SECURITY CLASSIFICATION OF THIS PAGE

UNCLASSIFIED

BLOCK #18 CONTINUED

Acoustic Emission TestingNuclear Magnetic ResonanceX-Ray RadiographyNeutron Radiography

Optical HolographyU

ThermgrapI

I*ABSTRACT

Adhesively bonded materials are increasingly being used by industries to obtain stronger, lighterweight, and more durable structures. Many of these structures undergo tremendous environmentalweathering and service stresses; and, consequently, the possibility of bonding failure is a majorindustrial concern, especially where rupture could cause severe harm to human operators. Thenondestructive evaluation (NDE) community has been studying various methods for inspectingbonds before and during service; but, to date, bond strength cannot be directly measured non-destructively. The NDE approach has been used to obtain NDE test parameters (such as ultrasonicattenuation, velocity change, thermal dissipation, and mass absorption) from test samples and thento correlate the parameters with the actual bond strengths measured when these same samples weredestructively assayed. The correlation is considered a qualitative bond evaluation; presently, only

* a destructive assay can supply a quantitative measurement for the strength of a specific bond.

This state-of-the-art report describes the bonding process, the destructive methods used to measurebond strength, and the various NDE methods that have been evaluated for determining the qualityof a bond. These NDE methods include sonics, ultrasonics, acoustic emission, nuclear magneticresonance, x-ray and neutron radiography, optical holography, and thermography. Each of thesemethods has shown some limited success in detecting debond conditions. At the present time,hoeei appears thtonly tesonic, ultrasonic, adnuclear magnetic resonance mtoshv

the potential capability to differentiate qualitatively the gradations between a good bond and adebond and thus provide a correlation to bond strength.

I

I Acoess on ForNTIS GRA&IDTIC TABUna~Dounoed

~Dt t itOn/-Availability Codes

Avilad/orDs Spe tal

U!I

I. . , ,, u i I I I

'I

TABLE OF CONTENTS

Page

A BST RA C T ........................................................... iiLIST O F FIG U RES ..................................................... vLIST O F TA BLES ...................................................... vi

I. CHARACTERISTICS OF ADHESIVE BONDING ......................... I

A . A pplications ................................................... I

1. Advantages and Disadvantages .................................. 12. Typical Fabrication Applications ................................. 13. Types of Adhesive Bonding Structures ............................ 2

3 B. Bonding Process ................................................ 2

1. A dhesive Types ............................................. 22. Sequential Steps ............................................ 43. Bonding M echanism s ......................................... 6

C. Q uality and Strength ............................................. 7D . Failure M echanism s ............................................. 10

II. NDE METHODS AND TECHNIQUES ................................. 11

A . G eneral D iscussion .............................................. 11B. Sonic Techniques ............................................... 12

1. C oin Tap Test .............................................. 122. M echanical Impedance ........................................ 121 3. M echanical Resonance ........................................ 13

C. U ltrasonic Techniques ........................................... 14

3 1. Amplitude-Domain Reflection/Transmission at the Bond Interface ....... 142. Spectral Domain Reflection/Transmission at the Bond Interface ......... 213. Low Frequency ............................................. 254. Rayleigh and Surface Waves ................................... 265. Lamb and Interface W aves ..................................... 266. Stress-W ave Factor .......................................... 307. Horizontally Polarized Shear W aves .............................. 308. Obliquely Incident Ultrasonic W aves ............................. 31

D. Acoustic Emission Techniques ..................................... 31E. Holographic Techniques .......................................... 32F. Radiographic Techniques ......................................... 32G . Therm ography ................................................. 323 H. Nuclear Magnetic Resonance ...................................... 33

iii

I

I

TABLE OF CONTENTS (CONTD)

3 Page

III. QUALITY CONTROL AND INSPECTION NEEDS ....................... 34

i A . M aterials ..................................................... 34

1. A dhesives ................................................. 342. A dherends ................................................. 34

B. Bonding and Curing ............................................ 34C. Inservice Inspection ............................................. 34

IV . SU M M A RY ...................................................... 35

3 V. CONCLUSIONS ................................................... 37

VI. REFERENCES .................................................... 38

IN D E X .............................................................. 47

I

I

II

IU

ii

I

LIST OF FIGURES

1-1 Examples of several joint designs ................................ 3

1-2 Honeycomb sandwich structure ................................... 4

1-3 Schematic of a fiber-reinforced composite laminate structure(m ultidirectional) ............................................ 4

1-4 Typical adhesive bonding process ................................ 5

1-5 Cohesive and adhesive bond failure .............................. 102-1 Force/time records for impacts on debonded and sound regions along

with associated frequency responses .............................. 13

2-2 Logarithm decrement of HT-424 bonded panels versus tensilestrength at 7.2 kH z ........................................... 14

2-3 Pulse-echo techniques ........................................ 153 2-4 Reflected wave, A exp (iwt)K ................................... 16

2-5 Curve showing failure load versus frontwall echo/backwall echo foran aluminum-to-aluminum step-lap joint with a 0.012-inch scotch-weldbondline thickness ........................................... 17

2-6 Correlation between the measured attenuation of the adhesive and thestrength of the adhesive bond ................................... 18

2-7 Correlation between the measured velocity of sound in the adhesive andthe strength of the adhesive bond ................................ 18

2-8 Typical time-domain waveform showing the reflections from theadherent/adhesive (1) and adhesive/adherend (2) interfaces ............ 20

2-9 Typical deconvolved frequency-domain waveform using the3 adherent/adhesive interface reflection as the system response .......... 20

2-10 Experimentally predicted shear strength versus the actual shearstrength in a defect-free specimen (measured destructively) as obtainedusing a RESID network ....................................... 21

2-11 Experimentally predicted bondline thickness versus the actual bondlinethickness (measured by sectioning) as obtained using a RESID network . .. 21

2-12 Experimentally predicted percent of cure versus the actual percentof cure (measured with differential scanning calorimetry, DSC, andthermometric analysis, TMA) as obtained using a RESID network ....... 21

2-13 Plots showing how the spectra from flat-bottom holes (b: 6mm,c: 3mm, d: 1.5mm) differed considerably from the spectrum for thesound area (a) .............................................. 23

2-14 Correlation of ultimate shear strength witl. 1/B ..................... 232-15 Experimental layered geometry configuration used to generate

Rayleigh-wave data. .......................................... 26

vI

3 LI"ST OF FIGURES (CONTD)

3 igure Page

2-16 Relationship between phase velocity of the first two modes of platewaves and shear strength of the adhesive joint in bimetal .............. 28

2-17 Change in the interface wave velocity as a function of curingtime for aged specim ens ....................................... 28

3 2-18 Change in time delay as a function of the aging time ................. 28

2-19 Experimental relationship between the interface wave velocity andthe normalized shear strength ................................... 29

2-20 Dependence of the amplitude of the reflected signal on the timeof adhesive curing for different thicknesses of the adhesive (metal-metalbond) ... .... ... .... ... ... .. .... .. ... .... ... .. ... .. .. .... . 3 1

vIIIIII,IIII

viI

I

3 LIST OF TABLES

i Table Page

1-1 General magnitude of different types of intermolecular forces ........... 73 1-2 List of ASTM standard test methods for measuring strengths ofbonded joints ............................................... 9

2-1 Effects of bond variables on ultrasonic measurements ................. 19

2-2 Ultrasonic features extracted from waveforms ....................... 204-1 Summary of NDE methods and techniques applied to determine

adhesive bond quality and an evaluation of their capabilities ............ 36

ivi

II

II

IiIII

i vii

I3 I. CHARACTERISTICS OF ADHESIVE BONDING

A. Applications 9 Does not permit visual exam-ination of the bonded area; and,

1. Advantages and Diradm ages as a result, quality control andquality assurance are more dif-

Adhesive bonding is the process of ficult.joining materials by surface attachment with

the aid of an adhesive. Because of its Generally restricts upper servicestrength and reliability, adhesive bonding temperature to about 180 de-offers many important advantages over other grees C, althcugh some specialjoining methods such as welding, riveting, and adhesives such as polyquinoxalinesmechanical fastening (1-4). Some advantages are available for limited use toare that it about 370 degrees C.

* Distributes stress more uniformly Requires tight control for proc-and minimizes areas of high stress esses such as surface preparation,concentration and, therefore, material handling, and curing.

permits fabrication of lighter andstronger structures. Degrades under environmental

exposure.* Provides joints with smooth con-

tours, allowing fabrication of more May require long cure times,aerodynamically favorable struc- particularly where high curingtures. temperatures are not used.

9 Allows joining of dissimilar ma te- 2 7)pia Fabrication Applicatnsrials such as metals to composites,rubbers to metals, or metals to Adhesive bonding has been exten-metals. In the latter case, adhe- sively used for fabrication of load-bearingsive bonding minimizes the possi- structural components used in aircraft, heli-bility of electrolytic corrosion copters, rockets, missiles, satellites, and spaceproblems. shuttles (1, 2, 5-7). The primary reasons are

savings in weight and smooth joint contours.* Permits easier fabrication of com- Specific examples of structures constructed by

plex contoured parts. using adhesive bonding are airframes, wings,fuselages, rudders, doors, tail sections, and

* Provides good damping of sound floor panels of aircraft (including helicopters)and vibrations in a structure. and bulkheads, nose-cones, nozzles, and

* Cmotor casings of rockets or missiles.I * Can be used as a seal againstliquids or gases as well as an Additionally, adhesive bonding haselectrical insulator or conductor. been widely used in other industries such as

automotive, electronic, and building construc-e May reduce the fabrication cost of tion (1, 8-1Q). Examples of applications are

a structure. fabrication of brakes and bodyshells of auto-mobiles; fabrication of electrical equipment

On the other hand, adhesive bonding such as motors, transformers, generators, andhas several disadvantages and limitations microwave devices; and construction of tile3 (1-4). Some of these are as follows: floors, roofs, and mobile homes.

III

I

I33. 7)pf of Adhesive Bonding Stures B. Bonding Proes

Types of adhesive bonding structures 1. Adhesive 7pesused in the aircraft and aerospace industriesinclude (1, 4, 11-14): Adhesives used for structural bond-

ing can be typically classified as thermoset-o Metal-to-metal joints ting, thermoplastic, and elastomeric (1-4 5).

Many of the modern structural adhesives are,• Metal-to-composite joints however, hybrids of these three (typically

combinations of thermosetting and thermo-* Metal-to-rubber joints plastic or elastomeric) to optimize strength,

toughness, and high-temperature properties.* Composite-to-composite joints

Thermosetting adhesives are syn-* Honeycomb sandwich structures thetic, organic substances, which can be con-

composed of various combinations verted by chemical reaction into a perma-of honeycomb core materials nently hard and practically infusible solid.(metal, paper, plastic) and facing They are activated by application of heat,materials (metal, composite, catalysts, or a combination of both and are set

paper) permanently by cross-linking reactions. Theyhave a higher modulus of elasticity; greater

o Fiber-reinforced composite lami- resistance to heat, chemicals, and adversenate structures. environments; but lower toughness than ther-

moplastics or elastomerics. The two mostMany joint designs or configurations important thermosetting adhesives are those

(including lap, strap, stiffening, cylindrical, based on epoxy and phenolic.angle, corner, and flange) have been used infabricationof adhesively bonded components. Thermoplastic adhesives are syn-Examples of several joint designs are illus- thetic, organic substances, which are hardtrated in Figure 1-1. A honeycomb sandwich when cold and softer when heated. Commonstructure is a laminar construction consisting thermoplastic resins include polyvinyls, poly-of thin facings bonded to a relatively thick acrylics, and polyethylene. Often employed inlightweight core as depicted in Figure 1-2. metal and plastic bonding, they are not usu-Fiber-reinforced composite laminate structure ally suitable for structural applications--espe-is made by bonding and curing two or more cially if subjected to elevated temperatures.layers of material (called prepreg) together; Certain high-temperature thermoplastic adhe-the prepreg is produced by embedding layers sives such as polyamides, polyimides, andof high-strength, small-diameter (about 10 PEEK (polyetheretherketone) have highmicrons) reinforcing fibers in a polymeric strength and good toughness and have beenmatrix (Figure 1-3). The layers of fibers are used in fabrication of advanced compositestscked in various sequences of orientation to laminates (6-18).produce desired mechanical properties.Stacking sequences are unidirectional or mul- Elastomeric adhesives are thosetidirectional, such as 0 -90 -0, 0 -45 -0, -45 45 organic substances that can be stretched to at45 -45, and 0 -45 -90 -45 -0. least twice their original length without inher-

ent loss of elastic properties. Natural andTypical metals used for adhesive synthetic rubbers such as silicone, nitriles,

bonding structures in aircraft and aerospace polyurethane, and polysulfides belong to thisindustries include aluminum, magnesium, category. They are often used as sealants, butsteel, titanium, stainless steel, and their lack the strength to be used alone for struc-respective alloys. Metal honeycomb cores are tural applications. They are widely used as amostly made of aluminum. Fiber materials modifying agent for thermosetting or thermo-used for reinforcing composite laminates plastic adhesives to improve toughness andinclude carbon/graphite, glass, boron, aramid, peel strength of the system.

I and polyethylene.

2I

II

Simple Simple

Beveled Recessed3 Stepped-Lap Joints

RadiusedS in g le -L a p J o in ts 000

I Single Taper

Simple Double Taper

I Beveled Increased Thickness

Radiused Landed

Double-Lap Joints Scarf Joints

3 Single Variable Adhesive Thickness

Stif Adhesive- Less-Stiff Adhesive

Double 7

I VanableStiff ness Adhesive

Beveled

Pin

I Radiused

Recessed Other Configurations

Strap JointsIFigure 1-1. Examples of several joint designs [Ref. U.S. Army DARCOM Handbook,reproduced by permission of the U.S. Department of Defense]

3

,I

I Adhesive Sheet promote, and speed up the curing reactions;Skin or Face Sheet and to reduce material cost.

Adhesives are available in liquid,r paste, film/tape, or powder form. Liquid

adhesives are applied by brush, spray, roller,or mechanical spreaders. Paste adhesives areapplied by spreading with mechanical toolssuch as a roller or trowel. Since the pasteshave high viscosities, they can be used onvertical surfaces with little danger of drip.Film/tape adhesives are the most common in

structural bonding applications. They offer auniform bondline thickness and are conve-nient for application to flat or slightly curvedsurfaces. Film/tape adhesives are available in

r 1a pure sheet of adhesive or with cloth orFigure 1-2. Honeycomb sandwich paper reinforcement. These film/tape adhe-structure sives are also called "prepregs" and are gener-

ally stored under refrigeration. Film-typeadhesives have a limited storage life com-pared with other forms. Powder adhesives,applied by dusting or spraying, are activated3 by applying heat.

2 Sequential Steps

I A typical adhesive bonding processinvolves preparation of constituent materials(such as adherends, prepregs, adhesives,

X.X: and/or primers), application of adhesives,assembly/lay-up, curing, and final finishing asillustrated in Figure 1-4 (1, 4,. 19). To reducefabrication time, improve reliability, andmaintain a consistent quality, many steps ofthe bonding process have been automated;and the trend will continue (19-21). Eachbonding process, shown in the figure, is brieflydescribed in the following five subsections,

Matrix Resin respectively.

I a. Material Preparation

Figure 1-3. Schematic of a fiber- a

reinforced composite laminate struc- Adherends. Of the processesture (multidirectional) used in preparing materials for bonding, sur-

face treatment of adherends plays the mostimportant role in determining the quality and

Besides the basic resin, an adhesive reliability of the bonded structure. Surfaceconsists of several other substances such as preparation varies depending on the specificsolvents, fillers, catalysts, diluents, hardeners, substrates employed, the design of parts, andantioxidants, or placitizers. These are em- the adhesive or primer used. In general, theployed to control viscosity, reinforcement, process involves the following steps: (1) clean-shrinkage, temperature operating ranges, or ing, (2) rinsing, (3) chemically treating,thermal expansion coefficient; to activate, (4) rinsing, and (5) drying. Before surface

4

I. .= = / a a I I I II

For example, chromic-sulfuric acid [etch

Material Preparation known as FPL (Forest Products Laboratory)adherends (substrates, honeycomb) etch] and phosphoric, chromic, and sulfuric

* adhesives acid anodizing have been used for aluminumprimers (1, 4, 5, 2); nitric-hydrofluoric acid pickling,

chromic acid anodizing, and alkaline peroxideetch for titanium (i, 4, 2-25); and hydro-chloric dcid and nitric-phosphoric etch forsteel (1, 4, 26).

Adhesiveionplhestvo The chemical treatment is fol-

lowed by rinsing in distilled or deionizedwater and then drying in clean air. Each stepof the surface-preparation process is per-

Assembly/Lay-up formed at a specified temperature (typicallyroom temperature to 100 degrees C) and forzII j a specified amount of time (typically a few

minutes to half an hour).

In many cases, the faying surfaceCuring of adherends is further treated by priming or

conversion coating to enhance bonding andretard environmental degradation such ascorrosion or water intrusion. Priming typi-I 1 cally involves spraying a thin coating of pri-

Finishing mer with a thickness of a few to several mi-crons to the faying surfaces and drying andcuring the primer at a specified temperaturefor a predetermined amount of time. Various

Figure 1-4. Typical adhesive bond- liquid or paste adhesives such as epoxy,ing process phenolic, and silicone or coupling agents such

as silane are used as primers.

preparation, parts are prefitted to ensure that Conversion coating consists offaying surfaces have the proper clearance, applying a chemical coating to a desired thick-

ness (1) on the surface of metallic adherendFor cleaning the surface, various after cleaning, rinsing in a chemical solutionmethods are used including solvent, alkaline, such as chromate, and drying (4, 26, 27).ultrasonic, and vapor degreasing; abrading (by Conversion coatings are applied by dipping,sanding or filing); and sandblasting (1, 4). spraying, or brushing. Such coatings are cor-When any chemicals are used in the cleaning rosion resistant and provide a good wettingoperation, the parts are rinsed thoroughly in surface for the adhesive.distilled or deionized water. When abradingor sandblasting is used, the dust particles For preparing metallic honey-created are removed by vacuuming or blasting comb cores, vapor degreasing, solvent clean-with clean filtered air. ing, or light sanding followed by vacuuming or

blasting with clean air are used (1). PlasticChemical treatment is employed and paper cores are normally bondable as

for metal or plastic substrates, excluding received. If contaminated, the core may behoneycomb cores, to make their faying sur- vapor greased. Light sanding followed byface receptive to adhesion. A wide variety of vacuuming or blasting with clean air may alsochemical treatment methods has been used be employed. If evidence of moisture is ob-depending on the substrate material (1, 4). served, the core is dried in an oven.

5I

I3 Adhesives. When film or tape d Curing

adhesives are used, preparation involves typi-cally (1) removing from a storage which is After the assembly, the adhesivegenerally refrigerated, (2) thawing to room bonds are cured according to a predeterminedtemperature, and (3) cutting to the required curingprocedure specifying time-temperature-size and shape. For cutting, mechanical, pressure requirements during the cure cycle.laser, or water-jet cutters are used (2). When For elevated-temperature curing, the heat-upother adhesive forms such as liquid, paste, or and cool-down rates and maximum tempera-powder are used, preparation involves weigh- ture limits are also controlled. The gas re-ing and mixing of base and additive materials leased during curing is removed by vacuumingsuch as catalysts, curing agents, hardeners, to eliminate debonds and porosities in bond-fillers, or placitizers. If the materials were ing. During the cure cycle, pressure is main-stored cold, they must be warmed to room tained over the entire bonded area to obtaintemperature before weighing and mixing. a uniformly thin adhesive bond layer and toI _overcome viscosity of the adhesive at the

Primers. Primers are prepared curing temperature, internal pressure exertedin a manner similar to the way liquid or paste by the release of adhesive solvents or wateradhesives are prepared. vapors, and surface imperfections and dimen-

sional mismatch between the mating sur-b. Adhesive Application faces (1).

Adhesives are applied to the e. Finishingsubstrates within typically a few hours afterthe surface preparation treatment. Various Finishing involves removingmethods such as spray, brush, roller, and excess cured adhesives, trimming and machin-spreader are used for applying adhesives in ing the cured assembly to the desired dimen-liquid, paste, or powder form. Film or tape sions, and sealing and painting the assembly.adhesives are applied by placing them on themating surface of the substrate. Then wrin- 3. Bonding Mechanismskles are worked out, all air-trapped air bub-bles are removed, and the film is trimmed to While many theories of adhesionthe desired size. A tacking paste may be used have been proposed and debated, the detailedto hold the film in place and to prevent slip- mechanisms involved in adhesive bonding orpage or movement during assembly. adhesion of two surfaces are not yet fully

understood. This inability to measure thec. Assembly/Lay-Up intrinsic adhesion forces acting across the

adhesive/adherend interface has hindered theAfter the adhesives are applied, development of a comprehensive adhesion

the components are joined and assembled in theory and the understanding of the detaileda tool or holding fixture. For laminate struc- mechanism (8, 29). No single theory ad-tures, prepregs are laid up in a predetermined vanced so far can explain adhesion in a gen-sequence of fiber alignment. Tools or holding eral all-encompassing way (4). In the follow-fixtures hold the detail components in place, ing paragraphs, a brief description is given ofapply the necessary pressure, and sometimes the four main mechanistic theories proposed:transfer or apply the necessary heat for cur- mechanical interlocking, diffusion, electro-ing. Examples of assembly tools or fixtures static, and adsorption (4, 28-30).are dead weight, vacuum bags, autoclave,Sclamps, and press (1). Provisions for the Mechanical Interlocking Theory. Inescape or breeding of gas released during the mechanical interlocking theory, mechani-curing are made. The assembly/lay-up is cal interlocking or keying of adhesive into thedone in a well controlled atmosphere includ- irregularities of the substrate surface providesing typically 20- to 60-percent relative humid- the adhesion forces. For certain cases such asity, temperature between 18 to 32 degrees C, the adhesion of polymers to textiles and bond-and clean air (4). ing of porous substrates, this mechanism

6

,,III I

appears to be the major source of the intrin- moisture resulting from the use of primers,sic adhesion forces. However, it is believed promoters, or coupling agents.that the frequently observed increase in meas-ured joint strength with mechanical roughen- When comparing the general magni-ing is due to other mechanisms associated tudes of these different types of forces (31),with a resulting change in physical and chemi- the Van der Waals forces are the weakest.cal properties of the surface with roughening. (See Table 1-1.) At small intermolecularseparations, however, dispersion forces are

Diffusion Theory. The diffusion sufficiently strong to provide good bondtheory is based on the mutual diffusion of strengths. For example, at molecular contactpolymer molecules across the interface be- (separation of about 4 to 5 A), the attractivetween two polymers to cause adhesion. This force between two perfectly plane parallelmechanism requires that the adhesive and plates is about 109 to 1011 dynes/cmp(14 tosubstrate polymers be mutually soluble and 1400 Klbs/inch2); at a separation of 10 A, theI the macromolecules have sufficient mobility, force is about 108 to 1010 dynes/cm 2 (1.4 toWhere the solubility of the material is not 140 Klbs/inch 2) (8). Although secondarysimilar or one polymer is highly crosslinked bonding forces alone can provide good jointand crystalline or is above its glass transition strengths at the interface, it is believed thattemperature, interdiffusion is an unlikely the additional presence of primary bondingmechanism. forces is crucial for establishing environ-

_________mentally stable interfaces (32).Electrostatic Theory. In the electro-

static theory, adhesion is caused by attractiveelectrostatic forces which arise from a double Table 1-1layer of electrical charge formed at the inter-face when two different electronic band struc- GENERAL MAGNITUDE OF DIFFERENTtures of adhesive and substrate are in contact. TYPES OF INTERMOLECULAR FORCESThe relevance and relative importance of this [Ref. (31), reproduced by permissionmechanism is still vague, of publisher Marcel Dekker, Inc.]

Adsorption Theory. Adsorption is Type of Force Energy (kcal/mole)the most generally accepted adhesion theory.Adhesion comes from the attractive surface Ionic 140 - 250forces acting between the atoms in the two Covalent 15- 170surfaces when the two are in intimate molecu- Metallic 27 - 83lar contact. The most common forces are Hydrogen bonds Up to 12Van der Waals and hydrogen bond (these are Dispersion Up to 10referred to as secondary bonds); other forces Dipole-induced-dipole Up to 0.5are ionic, covalent, and metallic bonds (theseare referred to as primary bords).

Van der Waals forces are attributedto different effects: (1) dispersion (or Lon- Quality of a bond is related to the func-don) forces arising from internal electron tional usefulness of the bonded assembly.motions which are independent of dipole Although no definition is universally accepted,moments and (2) polar (or Keesom) forces the bond quality generally refers to absencearising from dipole-dipole interactions. The of defects such as debonds, delaminations,hydrogen bond is formed through the attrac- voids, or foreign substances; mechanicaltion between a hydrogen atom and an electro- strengths of the bonded joint; and joint dura-negative atom such as oxygen, nitrogen, or bility in the expected service environment. Afluorine. Primary interfacial bonding may better quality bond generally implies that theoccur in certain circumstances and may be bonded joint has less amount of defects andresponsible for the enhanced joint strength a higher mechanical strength and a longer

* and resistance of the interface to attack by durability.

7

I

I

Bond strength generally fers to the unit * Treatment of adherend surface andload, which is applied in tension, compression, resulting physical and chemical prop-flexure, peel, impact, cleavage, or shear, erties of the surface such as rough-required to break an adhesive assembly with ness, cleanliness, and wettability (25,failure occurring in or near the plane of the 29, 8-42)bond (1, 4). Therefore, bond strength iscalled by various names such as tensile, shear, * Primer type (4346)flexural, peel, impact, and fatigue, dependingon the load type. Standard test methods for * Environmental condition such asmeasuring these strengths in the laboratory humidity, temperature, and exposurewith the use of standard test samples are time during the bonding process (19,available. Table 1-2 lists some of the test 7, 48)standards adopted by the American Societyfor Testing and Materials (ASTM). These Curing parameters such as tempera-standard test methods are destructive and are ture, pressure, heat-up rate, andnot meant to be used for determining the curing time (49-51)bond strength of a joint assembly in service. * Bondline thickness (2, 3)

Strictly speaking, bond strength should Residual stress in the adhesive (4)refer to the interfacial adhesion force per unitacigbetween the adhesive and thearea acigbtenteahsv.n h Presence of defects induced duringadherend. Current technology, however, is fabrication or serviceincapable of measuring such interfacial forces.It has been well known that the bond • Workmanshipstrengths measured per standard test methodsdo not correlate well with the interfacial ad- * Environmental degradation due to,hesion energies theoretically calculated. This for example, moisture, corrosion,poor correlation is attributable to various vibration, shock, impact, temperaturefactors influencing the mechanically measured cycling, fatigue loading, and bacterialjoint strength including the sample geometry degradation (55-60)and loading conditions used, flaw location,and failure mode (failure may not occur at In reality, the only way to determine thethe adhesive-adherend interface). Lack of true strength of a bond is to destructively testappropriate means for measuring interfacial it. Since this is not acceptable, one option isforces is apparently the major reason for the to perform a statistical evaluation by destroy-general use of the mechanically determined ing and quantifying some number of produc-joint strength as the bond strength. Cur- tion bonds. But this, too, is not acceptable torently, no consensus exists on defining quality the manufacturer because it is prohibitivelyin a quantitative way based on defect content expensive. The next best choice is to fabri-and location, mechanical strength, and dura- cate test samples at the same time and underbility data. the same conditions as the production bonds.

Many factors influence the quality of In practice, the fabrication goal is tobond (1, 2, 4, 33-37) including: know that the production bond is as good as

the bond in the test samples that are destruc-* Bonded joint type and geometry tively analyzed to verify the fabrication

process. Generally, the production bonding* Type of adherend material process is assumed to be equivalent to the

process used on the test sample. However,* Adhesive type and composition and many parameters affecting the final strength

of additives of a bond cannot be tightly controlled, so thisassumption may be inappropriate.

8I

Table 1-2

LIST OF ASTM STANDARD TEST METHODSFOR MEASURING STRENGTHS OF BONDED JOINTS

IASTM

Standard Title

i D897-78 Standard Test Method for Tensile Properties of Adhesive Bonds

D903-49 Standard Test Method for Peel or Stripping Strength of AdhesiveBonds

D905-49 Standard Test Method for Strength Properties of Adhesive Bonds inShear by Compression Loading

D950-82 Standard Test Method for Impact Strength of Adhesive Bonds

3 D1002-72 Standard Test Method for Strength Properties of Adhesives in Shearby Tension Loading

D1062-78 Standard Test Method for Cleavage Strength of Metal-to-MetalAdhesive Bonds

D1 184-69 Standard Test Method for Flexural Strength of Adhesive Bonded Lami-nate Assemblies

D2919-84 Standard Test Method for Determining Durability of Adhesive JointsStressed in Shear by Tension Loading

D3163-73 Standard Test Method for Determining the Strength of Adhesivelybonded Rigid Plastic Lap-Shear Joints in Shear by Tension Loading

D3164-73 Standard Test Method for Determining the Strength of AdhesivelyBonded Plastic Lap-Shear Sandwich Joints in Shear by Tension Loading

D3166-73 Standard Test Method for Fatigue Properties of Adhesives in Shear byTension Loading

Nondestructive testing can provide the late the effect of bond interface conditions onopportunity to obtain useful information the NDE data to the actual bond strengthabout the production bond in an acceptable, data obtained destructively so that NDE cancost-effective means. While bond strength provide some information about the quality ofcan only be absolutely measured by a destruc- the bond (relative to the destructively assayedtive test, NDE methods can directly measure test samples). By properly correlating specificsome valuable parameters indicative of the NDE parameters to the results obtainedbond interface condition. This interface does through the destructive assay, a bond qualitygreatly affect the ultimate strength of the gauge could be developed for each type ofbond. Thus, it is the goal of NDE to corre- bond and geometry.

9

I i II I

The premise that NDE is not used to within the adhesive layer or one of the adher-directly measure bond strength, but can only ends, but not adhesively at the adhesive/ad-be used to determine bond quality (relative to herend interface. However, because somebond-strength data obtained destructively) is bonds that fail adhesively can exhibit greatera key point for utilizing the information con- mechanical strength than a similar jointtained in this report. bonded with a weaker adhesive, which fails

cohesively, determining quality based on theD. Failure Mechanisms failure mode is not recommended (4).

Failure of adhesive joints usually results Similarly with adhesion, the detailedfrom the initiation and propagation of flaws mechanisms of joint failure are not wellsuch as cracks, debonds, and delaminations. understood including the mechanisms of de-Depending on the location of the fracture fect initiation and propagation toward thesurface, the failure is generally divided into critical size. Defects such as debonds, cracks,two modes, adhesive and cohesive (1, 4). and delaminations, however, can be initiatedAdhesive failure is fracture at the interface and propagated by fatigue; impact; static andbetween the adhesive and adherend (Fig- dynamic loading; and weakening of adhesive,ure 1-5). Cohesive failure is fracture in the substrate, and the adhesive/adherend inter-adhesive so that a layer of adhesive remains face through aging and environmental degra-on both adherends (Figure 1-5). Generally, dation, or by a combination of these (2, 54,failure occurs in mixed modes and is ex- 7, Li-!4). Microecopic defects preexisting inpressed as a certain percentage cohesive or the material or induced during fabrication dueadhesive. to imperfect processing can also act as the

initiators. The defect propagation and failureThe failure mode of a bond is sometimes mode are dependent on various parameters

used as a criterion for quality of bonding; for including the material type, joint design,example, a good bond would fail cohesively stacking sequence, environmental condition,

type of surface treatment, and quality ofbonding process.

Adhesive Adhesives can be weakened through- -thermal degradation and breakdown, biologi-

Adherend cal attack, or hydrolysis (2, 4, ). Substrateweakening occurs through corrosion of the

Cohesive Adhesive substrate surface or weakening of the oxidelayer on the surface (2, 5, 6, 4, 2, 65).

AdeedMatrix resin of fiber-reinforced composite

materials can be weakened in humid or aque-Adhesive ous environment through corrosion, hydroly-

sis, depolymerization, or microcracking due to

500/ swelling caused by water uptake (6-7.). TheIOCohesive weakening of the interface between the adhe-sive and substrate occurs through desorption

Figure 1-5. Cohesive and adhesive or displacement of the adhesive by the intru-bond failure sion of water or other liquids (2, 4).

III

10I

I

U IT. NDE METHODS AND TECHNIQUES

3 A. General Discussion Thermography can successfully monitorthe variation in thermal conductivity due to

Over the last 20 years, a large amount of defects in the bonded region. ThermographicNDE research and development (R&D) has techniques, such as infrared (IR) scannersbeen conducted to determine how to detect and cameras, liquid crystals, treated paper,and predict the quality of bond in adhesively paints, and phosphors, and thermometricbonded structures (71-131). These methods techniques, such as thermocouples and radi-investigated include sonics, ultrasonics, holog- ometers, were evaluated; but several problemsraphy, radiography, thermography, acoustic were found. First, it is difficult to get suitableemission, and nuclear magnetic resonance. In heat/cold sources; second, high anisotropicthis section, the state of the art for each of materials have different thermal properties inthese methods will be discussed. different directions, which cause data inter-

pretation problems; and, third, metallicSome of these methods have had little bonded structures can only hold temperature

success. In a 1974 general review of the NDE gradients for a short time.of adhesive bonds, Norriss (71) reported onseveral techniques includingx-ray, holography, Zurbrick (73) reported that the mostVacuum cup (Avro bond tester), Resonance critical area for NDE R&D was the substrateFokker Tester, eddy sonic Shurtronic, ultra- surface preparation prior to bonding. Hesonic Sondicator, heat-sensitive coatings and developed an equation for predicting bondpapers, acoustic impedance and emission, and strength based upon surface energy in whichthrough-transmission and pulse-echo UT. all the controlling variables could potentiallyThe most usable methods seemed to be the be measured using NDE. The equation is asFokker Bond tester, Sondicator, through- follows:transmission UT, and acoustic impedance.However, most of these could only detect ESL = Es - EL cosO (1)voids. The preferred method was FokkerBond Tester for sheet-to-sheet joints; and for where Es is the solid-surface energy (in athe core-to-sheet joints, radiography and vacuum), EL is the liquid-surface free energythrough-transmission UT worked best. (in a vacuum), ESL is the solid-liquid interface

surface-free energy (assumed to be the bondIn a more recent report (1984), Teagle adhesive strength), and 0 is the contact angle

(72) reported that no single NDE technique at the three-phase point.solved all problems. His studies found thatradiography could be used to detect damage The bond strength is the energy storedto honeycomb, presence of debris and con- in the adhesive volume at the moment oftaminants, and lack of bond due to absence of fracture. The bond strength is primarily de-adhesive. Conventional UT with 1 to 10 MHz termined by the thickness of the bondline,can be used in the through-transmission mode which greatly affects the energy stored in theto determine attenuation, which gave some bond. The ultimate bond strength is relatedinformation about debonds, delaminations, to the solid-surface energy, liquid-surfaceand porosity; but these defects cannot be energy, thickness of the bond, contact angle,easily distinguished from one another. and strain. Bond quality can be predicted ifApproximately the same information can be bond thickness, contact angle of adhesive-to-obtained with pulse-echo UT. Frequency substrate, and substrate surface-free energyanalysis of reflected waves gives information can be measured nondestructively.about how the waves have interacted with thematerials in the inspection path and can help To measure the variables just listed, NDEdiscriminate between the different bonding techniques including exo-electron emission,problems. ultrasonic gas-phase transmission, electricI

I 11 II

3 field reflectrometry, and light specular reflec- B. Sonic Techniquestance were used by Zurbrick. Substrate sur-face-free energy had the most dramatic linear Sonic techniques refer to inspection proc-effect. The contact angle was less dramatic. esses that utilize mechanical impulses to pro-Effective strain was less than anticipated, and duce low-frequency vibrations. These vibra-the others behaved in a similar manner. tions are usually in the audio range, and theZurbrick's basic premise was that the adhe- propagation of the frequencies is affected bysive bonding is controlled by the surface elec- the mechanical condition (e.g., flaws andtron-energy state. He took light reflections density variation) of the part under inspec-from the surface using a 45-degree incident tion.white light and a detector at 45 degrees. Hisdata indicated that the reflected light was Several sonic inspection techniques haveinversely related to the calculated substrate been used over the years for detecting thesurface-free energy. quality of bond of an adhesively bonded struc-

I ture. These include the coin tap test,Segal et al. (74) handled the problem mechanical impedance, and mechanical reso-

somewhat differently. They divided the prob- nance (75).lem of nondestructively determining adhesivebond integrity into four parts: (1) detection 1. Coin Tap Testof the unbond, (2) prediction of the cohesivequality of the bond, (3) detection of adhesive The coin tap test has been used forfailure, and (4) prediction of a combined many years as a quick yet sensitive techniquecohesive/adhesive quality, to find large debond areas in laminated and

honeycomb structures. This test is usuallyAlmost all NDE methods have been used very subjective, but work by Adams and Caw-

to inspect for debonds; that is, UT, eddy ley (76) has shown that the technique can besonic, radiography (x-ray, neutron, tracers), quantized. They point out that the differencethermography (IR, moving heat source, liquid between tapping a good or a defective regioncrystals, thermochromatic coating), optical is due to the change in force input. The localholography, acoustic emission, liquid pene- structural stiffness is affected by the presencetrants, and microwaves. The consensus of this of the defect, and the force/time characteris-work was that the most accurate method for tics are dependent upon the local structuraldetecting voids, porosity, and excess adhesive impedance. Adams and Cawley found thatwas radiography. Neutron radiography is the impact on a good region had a highergood for bonded metallic structures; for struc- amplitude and a shorter time duration (ortures containing composite, x-ray is more contained higher frequencies), while the im-effective. However, access to both sides of pact on a defective region had a lower ampli-the part is needed. When access to one side tude and a longer duration (or containedonly is available, then optical holography or lower frequencies), as illustrated in Fig-pulse-echo ultrasonics is most useful. ure 2-1. Therefore, by incorporating a force

transducer in the impact hammer, the locationFor detecting debonds, the most common of adhesive defects can be found. The sensi-

NDE method is ultrasonics. This can be done tivity of the technique has not been evaluated.by using tapping, pulse echo, through trans-mission, resonance, acoustic impedance, spec- 2 Mechanical Impedancetrum analysis, logarithmic decrement, acousticholography, and critical angle reflectivity. The mechanical impedance tech-

nique works on the principle that the local-The following subsections discuss the ized impedance of a sample is dependent

work conducted using the various methods upon the defect condition of the sample atand specific techniques. that point (7). The point impedance of a

I12U

I3 5 technique does not work well with a compos-

ite structure.40- TO Over good area

30 3. Mechanical Resonance

p o tA

planar debond was modeled as

PO -plate restrained around the edges by the sur-rounding structure (6). As the frequency of

1excitation increased, the debond resonated.10 At resonance, the impedance over the defec-

tive region decreased, and the response for a0 400 800 1200 1600 given force input increased. If the frequency

Time (As) were close to the fundamental frequency, the50 sresponse amplitude for the defective region

W cwould be much higher than the surrounding-Goodaor region. The layers above a defect were

30 modeled as a disc, and the resonant frequencyS Defect of the disc was directly related to the depth of

the defect and the radius of the defectiveio- region, as well as the square roots of Young's

modulus, the density, and a function of Pois-400 800 1200 1600 2000 son's ratio. The typical operating frequenciesF,, ,ef cy () for instruments using this technology were 20

I to 30 kHz.Figure 2-1. Force/time records for impacts

on debonded and sound regions along with Another mechanical resonanceassociated frequency responses [Ref. (6), method used to evaluate adhesive bonds util-reproduced by permission of Gordon and ized a lower frequency. Low-frequency (-10Breach, Science Publishers, Inc.] Hz) dynamic measurements of the shear

modulus of epoxy phenolic and other adhesivesystems were shown to vary greatly during

structure can be determined by applying a curing and, in fact, to provide a measure offorce and measuring the resultant velocity of the degree to which curing had pro-the structure due to the force. That is, gressed (77). Damping of the vibrations of

adhesive rod specimens also correlated withZ = F/v (2) cure and strength. A plot of the log decre-

ment of decay of free oscillations measuredwhere Z is the local impedance, F is the har- versus bond strength shows that the lower themonic force input to the structure, and v is adhesive bond, the higher the decrement.the resultant velocity of the structure. This is illustrated in Figure 2-2.

I Most commercially available instru- Meyer and Chapman (78) reportedments work between 1 and 10 kHz. As the that the Ford Motor Company has used NDEimpedance of the structure decreases, the to monitor the bond quality of adhesivelyquality of the bond decreases. The impe- joined fiberglass-reinforced plastic (FRP)dance, however, is very dependent upon the since late 1970. Their technique used theflexibility of the base structure. If the struc- low-frequency Automation Industries' Sondi-ture more cator S-2B. They found that the three factorsthe defective region can be higher or lower affecting the strength of FRP were the sub-than that of the good region, depending upon strate, primer, and adhesive. To address thethe input frequency; the test, therefore, is not potential problem with quality assurance,very reliable. Because the sensitivity of this Ford developed three concepts: quality bond-technique is directly related to the contact ing environment, bond-strength profile, andstiffness of the material being tested, the bond merit factor (BMF). The reason Ford

13

.5 bondline can be correlated with destructivelydetermined bond strengths. These results are

Initial Measurement a consequence of the viscoelastic properties of

oItial Mont the adhesive whereby changes in the elasticI * After,-lO Months

.4 Room Temp Cure moduli and damping capacity of the adhesivecan be related to the molecular structure and

D hence the quality of the adhesive.

E C. Ultrasonic Techniques

Ultrasonic techniques refer to inspection.S2 processes that utilize a piezoelectric-element.- B transducer to generate ultrasonic energy into

Za I IB a part. This energy interacts with the part0 D under inspection and is propagated back to.C either the transmitting transducer or a sepa-

rate receiving transducer. The ultrasonic.1 ' - - energy can be transmitted in various modes

A A (e.g., longitudinal, shear, or surface waves)

and at various angles. The condition of thematerial under inspection affects the ampli-

000 tude and frequency of the ultrasonic waves as400 500 600 700 they propagate through the material. In some

Tensile Strength (psi) inspections, onl, the amplitude information isused to evaluate bond quality; and in others,

Figure 2-2. Logarithm decrement of HT-424 only the frequency information is analyzed.bonded panels versus tensile strength at7.2 kHz [Ref. (7), reproduced by permission A tremendous amount of work has beenof the American Society for Nondestructive conducted to evaluate the capability of ultra-Testing (ASNT)] sonic techniques to determine the quality of

adhesive bonds. The work has consisted ofstudying longitudinal, shear, surface, and

came up with the bond merit factor was to interface waves. Techniques include thoseanswer the question: If a partial bond exists that measure the response of the signaland is not detected, is it serious to the prod- reflected from or transmitted through theuct? The equation for the BMF is given by bonded interface region and those that inter-

act directly with the adhesive layer. TheseBMF = (L - C- P/2)/L (3) techniques are described in more detail in the

following paragraphs.where L is the length of the bond inspected,C is the length of completely debonded joints, 1. Amplitude-Domain Reflection/lFrans-and P is the length of any joint found partially mision at the Bond Interfacebonded.

The amplitude-domain reflection/In a review paper, Thompson et transmission techniques attempt to determine

al. (77) pointed out that a number of survey the quality of the bond by evaluating theexperiments have suggested that the adhesive interaction of the ultrasonic beam with thebond quality of honeycomb sandwich panels adhesive boundary interfaces. For the pulse-also can be nondestructively determined by echo mode, as the ultrasonic beam interactsmeasurements of vibrational response. The with the boundary of the adhesive, certaindata showed that both the damping character- reflections should occur. [See Figure 2-3,istics of panel vibrations as a whole and the Ref. (9).] If a good bond exists, little reflec-velocity of the propagation of elastic waves tion should be observed (except due to pucethat travel along the surface and the sample impedance matching). If a defective bond

14I

3 the defect echo yields information on thedefect configuration. Frequency and timefeatures can be used to characterize the de-fect that produces the ultrasonic reflection.

Flaw Detector CRTBond Unbond Work by Adler and Whaley (LI)

showed experimental dependence of spectralTransducer/ Transducer variations within a reflected broadband ultra-

sonic pulse on the size and orientation of thereflector. An analytical model was developedassuming that the interference of the wavesreceived from the edges of the reflectingsurface was responsible for the variations ofthe received frequency spectra. Their work3_. was aimed at developing a method for deter-

Adherend mining the size of arbitrarily oriented flaws ofcrack-like geometry. When an object wasplaced in the path of a sound beam, the

Adhesive,// object produced waves which constructivelyinterfered (added in phase) for frequencies

Adherend given by fn = nv/(2d sin a) (4)

I Figure 2-3. Pulse-echo techniques [Ref. (79), where n is an integer, d is the diameter of thereproduced by permission of the Academic circular object in the path of the ultrasound,Press] v is the velocity of the sound in the medium,

and a is the angle of the transducer relativeto the object.

exists, then reflection would increase because

the impedance mismatch is due not only to Tattersall (82) found that flaws arethe impedances of the adherents and adhe- commonly detected by their ability to reflectsive, but also to the presence of air or other a burst of ultrasound back to the piezoelectricinclusions. The through-transmission tech- transducer. Relying on the existence of anique would interact with the same bound- reflected signal, however, is not adequate toaries, but good bonds would have high trans- detect the difference between a weak and amission, while poor bonds would have good bond because both good and bad bondsdecreased transmission. can reflect ultrasound. He developed a the-

ory that provides expressions for the coeffi-Work by Gericke (80) in 1963 cients of the reflected and transmitted ampli-

Ss.iowed that UT techniques for metals and tude and phase. The formula for the re-other solid materials were of limited value flected amplitude from an adhesively bondedbecause they did not determine the geometry area isof concealed defects when these defects werecomparable in size, or smaller than, the width ZI - Z 2 + iw(Z 1Z2/K)of the UT search beam. Whenever such A = (5)relatively small flaws were involved, UT pulse- ZI + Z 2 + iw(ZIZ 2/K)echo testing at a single frequency yielded flawlocation, but no information about its geom- where Z1, Z2, and K are the ultrasonic impe-etry. If a UT signal containing a wide band dances ol the adherent and adhesive, respec-of frequencies is used (a white source), then tively, and K is a constant. By determiningthe form and spectral energy distribution of the amplitude, the quality of the bond can bethe reflected beam are influenced by the derived from the Argand diagram show ingeometry of the defect. Thus, an analysis of Figure 2-4.

15I

I

3 were cleaned in the same way but without the

Decreasing Adhesion chromic sulfuric acid etch. [The cleaning and(o/K>O) bonding procedures were given in this pa-

_ per WA).

E AO/K +co UT data were collected using 1/4-b Ia -Real Axis inch, 10- and 20-MHz transducers in an

Z1 -Z2 /1 =0 W0/K . -co immersion mode. Several samples were de-ZI , +Z2 structively tested to determine the bond

.P.ese c strength. Their results indicated that a vari-Decreasing Adhesion ation in the interfacial condition of the bond

(0o/K<O) was most noticeable ultrasonically by anamplitude change in the interface echo. Asthe quality of the interfacial bond decreased,the amplitude of the reflection from the adhe-

Figure 2-4. Reflected wave, A exp (iwt)K sive-substrate interface increased.[Ref. (2), reproduced by permission of theAtomic Energy Research Establishment, Har- The 10-MHz data did not correlatewell] well; the 20-MHz data, however, showed

clearly which sample had good surface prepa-ration and which did not. Because some of

Rose and Meyer (83) utilized ultra- the data were unclear, more experimentationsonic immersion and spectroscopic procedures is needed using the model developed by Rosefor predicting and evaluating bond quality in and Meyer.aluminum-to-aluminum step-lap joints madewith Scotch-Weld 2216 structural adhesive In 1979 Chernobelskaya, Kovnovich,joint. The UT tests were conducted using a and Harnik (6) developed a quantitativemedium-focused, 1/4-inch diameter, 10-MHz method of testing adhesive bonded jointstransducer in the immersion mode. The using a high-resolution ultrasonic probe. Thefrontwall echo (FWE) and the backwali echo probe resolved echoes from the two interfaces(BE) were compared. The results showed of the bondline. The method was tested onthat as the ratio of FWE/BE increased, the aluminum-to-aluminum joints bonded withadhesive bond quality increased. In other Metlbond 328 of Narmco Corp. Their meth-words, the adhesive was transmitting the od first tested the unbonded piece and ob-ultrasonic energy through the adhesive when tained a set of echoes from the free surface,it was well bonded (see Figure 2-5). There- and then data were taken from the bondedfore, the echo at the adhesive interface was joint. The effective reflection coefficient ofdecreased relative to the frontwall surface. the interface was derived from the measuredSimilar work has been conducted by Biggiero amplitude ratio, which takes into account theet al. (84). loss in the aluminum. The true reflection

coefficient can be derived from the measuredMeyer and Rose (85) demonstrated transit times, thicknesses, and known densities

the application and use of analytical models of the aluminum and the adhesive accordingin the experimental ultrasonic evaluation of to R = (Z2-Zl)/(Z2+ZI). The ratio of theinterface conditions in an aluminum-to-alumi- effective reflection coefficient to the truenum adhesively bonded system. Their results reflection coefficient was the measure of theshowed that a variation in bond quality due to adhesive bond quality. Cohesive propertiessurface preparation can be detected ultrasoni- are also discussed as being related to thecally through careful inspection and signal- modulus of elasticity and ultrasonic absorp-processing analysis. They worked with 50 tion.step-lap test specimens. The bondline thick-ness was controlled to 0.01 inch. Half of the This model was simple; but it wassamples were cleaned in accordance with only tested on two samples, and the data3 recommended procedures, and the other half stated that the two samples were good. No

16U

ISpectral Analysis

1100x Specimen With .019"

I 000 Bondline Thickness

1 900..~800.

0 Processing Parameterso 700(x Standard Mix and Cure• 'o 600.I =Z" 3Teoetca •5Base/6 Accelerator

x 3 Experimental Standard Cure

400 0] Standard Mix UnderCure

200 A Reference Echo With

No Bond3 100

1 2 3 4 5 6 7 8 9 10 11 12 13

I Figure 2-5. Curve showing failure load versus frontwall echo/backwall echo foran aluminum-to-aluminum step-lap joint with a 0.012-inch scotch-weld bondline3 thickness [Ref. (3), reproduced by permission of ASNT]

destructive tests were conducted on the sam- tern can be made. Alers pointed out that thepies to correlate the findings, bond is really not between the adhesive and

the metal, but between the oxide of the metalAlers, Flynn, and Buckley (7) felt surface and the primer. The oxide/primer

that the basic premise on which ultrasonic interface is usually between 0.02 and 0.2 mi-methods can be used to predict the cohesive cron thick, and NDE test has to target thisand adhesive quality of a metal-to-adhesive thickness.bond is based upon the argument that meas-urable changes in the elastic properties of the Expressions for the various reflectionadhesive or the interface should be associated coefficients were given based upon impe-with changes in the cohesive or adhesive dances. For very thin adhesive layers, thestrength. Their work showed that the cohe- resonance and antiresonance dips becomesive quality of a bond can be predicted from coupled and it is difficult to determine thequantitative measurements of the velocity of thickness and adhesive properties without asound and the attenuation in the adhesive computer.layer.

The cohesive strength of a polymericWhat needed to be addressed was adhesive should be closely correlated with its

the lack of cohesive strength within the bulk bulk physical properties. An improper mix-of the adhesive and the lack of adhesive ture of chemical constituents of an incorrectstrength at the polymer-to-metal interface, curing procedure is apparent in cohesiveWhen adhesive cures, it becomes hard and strength and also modifies the ultrasonicless attenuative to ultrasound. By measuring velocity and attenuation. The Fourier trans-these two parameters, a cure-monitoring sys- form of the adhesive interface reflection is

I. .. . i l I I I I 17

3 taken and divided by the Fourier transform of Segal, Thomas, and Rose (4) alsothe front surface reflection. The minima in used ultrasonic techniques for inspectingthe spectrum is related to the layer thickness, bonded structures. They found that the bestand the depth is related to the attenuation as ultrasonic technique for predicting adhesivea function of frequency. The ultrasonic atten- quality was is high-frequency ultrasonics usinguation showed a "knee-type" curve relation- signal processing, as described by Rose andship with the bond quality (as shown in Fig- Raisch (88). Williams and Zwicke (89) util-ures 2-6 and 2-7). Velocity was linearly ized various ultrasonic resonance methods torelated to bond quality. The resonance fre- inspect adhesively bonded joints. They deter-quency was correlated with bond quality, but mined that neither simple nor more compli-showed that the shift in resonance frequency cated ultrasonic approaches were able towas small for changes in adhesive strength. absolutely assess joint quality.

A good bond is determined by sev-5000 eral variables, and the ultrasonic interaction

20 also depends upon variables that are difficult4000 * to separate. The authors categorized the

15 bond variables into two areas: intrinsic andextrinsic. The intrinsic properties effecting

< cohesive bond strength included degree ofI 20 cure, adhesive chemistry, bondline thickness,2 and voids. The extrinsic properties effecting

1000 5 the adhesive bond strength included surface1 0cleanliness, surface contamination, etchant

10 2 25 30 35 type and time, primer type and application,and wetting properties of the adhesive. They

Attenuation Coeft. Np/ecm found that even if all the intrinsic properties

were within proper tolerances, the bond maystill fail due to extrinsic factors. They in-

Figure 2-6. Correlation between the meas- cluded joint geometry and loading, and theured attenuation of the adhesive and the presence of defects.strength of the adhesive bond [Ref. (87),reproduced by permission of ASNT] A multidisciplinary approach was

required that combined NDE and fracturemechanics to form a basis for a comprehen-

5000 sive quality assurance solution. NDE couldI20 detect defects and measure intrinsic cohesive

4000- and adhesive properties and, in the case

300 *15 where no defects were present, could assessI 3000, bond quality. Fracture mechanics was em-

210 ployed to determine acceptable defect sizesL 200 (to be detected by NDE) for a given set of

a. intrinsic bond properties. The interaction of1000 o ""ultrasonic waves with certain properties of the

bond were placed in a table by the authors.01 .. 0 (See Table 2-1) For example, longitudinal2.0 2.2 2.4 2.6 2.8 (L)-waves were strongly influenced by the

i ocure state and bondline thickness, but did notdo well for chemistry, porosity, etch, defects,or surface condition.

Figure 2-7. Correlation between the meas-ured velocity of sound in the adhesive and the The experimental work was con-strength of the adhesive bond [Ref. (7), ducted on 6061 aluminum using Eastman EA3 reproduced by permission of ASNTJ 9649 adhesive. The L-waves were generated

18

"I

Table 2-1

EFFECTS OF BOND VARIABLES ON ULTRASONIC MEASUREMENTS(S = Strong Dependence; W = Weak Dependence) [Ref. (9), reproduced bypermission of ASNT]

Bond VariablesUltrasonic Cure Bondline SurfaceMeasurements State Chemistr Porosity Thickness Etch Condition Defects

Longitudinal Wave3 Velocity, VL S W S W

VL vs. Wavelength W

Longitudinal WaveHysteresis Atten., ctHL S S W W S

Longitudinal WaveScattering Atten., aSL

'oSL vs. Wavelength S

Shear WaveVelocity, VT S W S S

3 VT vs. Wavelength W

Shear WaveHysteresis Atten., c>HT S S S W S

Shear WaveScattering Atten., '>ST S S S

3 1ST vs. Wavelength S

I with a PVDF transducer which had high in- The RESID algorithm was appliedternal damping and broadband frequency to ultrasonic amplitude data. These ampli-response (1 to 35 MHz). Shear (S)-wave tudes, shown in Figures 2-8 and 2-9, were thework was conducted with an S-wave trans- positive and negative RF amplitudes of theducer using a UTRC proprietary couplant. reflections from the adherent/adhesive inter-The signals from the front and back surfaces face and the positive and negative RF ampli-were digitized and processed. UTRC devel- tudes of the adhesive/adherent interface.oped a pattern-recognition analysis algorithm The features used in the algorithms are listedcalled REcursive Structure IDentification in Table 2-2.(RESID) based upon a theory developed byA. G. Ivakhnenko (20). The algorithm in- Using the developed algorithms,volved five features (S-wave velocity, L-wave experimentally predicted bond properties suchattenuation, S-wave attenuation, L-wave atten- as predicted bond strength, predicted shearuation versus frequency, and S-wave attenu- strength, and predicted percent cure wereation versus frequency). correlated with destructively measured prop-

3 erties, shown in Figures 2-10, 2-11, and 2-12.

19I

1500 1.6 A

C 300 A32A 1.2

1 01 A! -100 V2B

3 2B-3 2iB

-1

F2 '3

0 1 2 3 4 5 0 10 20 30 40 50

ITime, psec Frequency, MHz

Figure 2-8. Typical time-domain waveform Figure 2-9. Typical deconvolved frequency-

showing the reflections from the adherent/ domain waveform using the adherent/adhe-

adhesive (1) and adhesive/adherend (2) inter- sive interface reflection as the system

faces [Ref. (89), reproduced by permission of response [Ref. (89), reproduced by permission3IS T o f A S T

I Table 2-2

I ULTRASONIC FEATURES EXTRACTED FROM WAVEFORMS

I [Ref. (89), reproduced by permission ofteASNT]

I(a) Longitudinal WaFeatures

a Feature Detemined by* Related to

Lp (V2A +V2B)/(VA + V1 ) Total Attenuation. cL

Total Attenuation, aLS F1

VL (Dispersion)

S

VL (Dispersin)

0 5 2Total Attenuation, L

[Iog(1/A 3)'iog(1/A 2)]/logF 2-1ogF 0 J aL = L(F)3 (b) Shear Wave Features

Feature Detem ined by* Related to

ISp (VA+VB/VA+ V1 )Total Attenuation, aT

sS 0 V2 )/(VI A Total Attenuation, aT rSw F0

Longitudinal Velocity, VT

ad 1) Fd VT (Dispersion)

S2 F

VT (Dispersion)

L3 55 1 B B Total Attenuation, T

5 [Iog(1/A 2)-log(1/Al)]/IogFlOgF0 ] aS=

S7 [log(1/A 3)-log(1/A 2)!/[logF2-logF0 l aS = s(F)

*VIA, VFB, etc.; AD, A2 , etc.; and F1, F2, etc. are defined in Figures 2-8 and 2-9.

I20I

SP 'V + I2) ( I + I13 Toa Ate uto ,

U8o aTraining Specimens 100 _ _ _ _ _

0 **Evaluation Specimens

6 0

0 0 C680

004l 2 FP ethfA64

0:2 0V a.

Destructively Measured Shear Strength, psi/1 000 40I 00 I

Figure 2-10. Experimentally predicted shear 30 50 70 90

strength versus the actual shear strength in a % Cure (DSC and TMA)

defect-free specimen (measured destructively)as obtained using a RESID network [Ref. Figure 2-12. Experimentally predicted per-(89), reproduced by permission of ASNT] cent of cure versus the actual percent of cure

(measured with differential scanning calorim-etry, DSC, and thermometric analysis, TMA)

M 9 as obtained using a RESID network [Ref.e= Evaluation Specimens (8), reproduced by permission of ASNT]U" 8 0-, ainSciDs,

00

7 formed a raster scan over the test plate. TheI.- test plate consisted of a sandwich configura-

6 tion and contained curing defects, adhesivecontamination, and voids in the adhesive.I0 D 5They looked at ten different samples. The

- RF waveform was digitized and analyzed on4 by several feature-extraction algorithms. The

*.5 6 ratio of the adhesive layer reflection to theDestructively Measured Bondline Thickness, Mils front-surface reflection was one of the major

features.Figure 2-11. Experimentally predicted bond-line thickness versus the actual bondline thick- 2 Spectral Domain Reflection/lTrans-

ness (measured by sectioning) as obtained mission at the Bond Interfaceusing a RESID network [Ref. (89), repro-duced by permission of ASNTJ The spectral domain reflection/trans-

mission techniques attempt to determine thequality of the bond. They do so by evaluating

Rose, Avioli, and Bilgram (1) devel- the interaction of the frequency content ofoped a combined NDE technique that made the ultrasonic beam with the adhesive bound-use of signal processing, pattern recognition, ary interfaces. Frequency parameters such asand a high-density ultrasonic scan into a hy- center frequency, bandwidth, power spectrum,brid-scheme for assessing the integrity of an and others can be calculated from the Fourieradhesively bonded structure. transform of the reflected or transmitted

t sz ultrasonic beam. These features can be ex-Experimentally, they used a 10-MHz tracted from waveforms obtained from test

transducer in the immersion mode and per- samples that simulate various bond conditions

I21

I

3 and an algorithm using weighted values of the The ultrasonic pulse-echo ratio datafeatures that are best correlated to the bond were obtained using a 15-MHz transducerquality. If a good bond exists, then a certain with a Panametrics 5052PR pulser/receiver.algorithm can be developed. If a defective The Q of the resonance was related to thebond exists, then the algorithm gives a differ- ratio (r) of acoustic impedances in the twoent value, media by

Chang et al. (92) analyzed the fre- Q = 180/(sin' 1(2r/r 2-1) (6)quency spectrum of ultrasonic plane wavestransmitted through a multilayered laminate The Q values were measured at the half-widthstructure at normal incidence to determine of the curves in the frequency spectrum.the amplitude distribution of the frequencycomponents. In supporting the theoretical The UT P-E amplitude ratio did notcalculations, the wave equation was solved to give good correlation, but did show a general-evaluate the displacement field with the ap- ized trend of Al/A2 decreasing as bondpropriate boundary conditions in a six-region strength increased (where A1/A2 is the ratiolaminate. Cavity resonance of the plane of signal amplitude for the top and bottomwaves in the layers produced peaks in the adhesive interfaces). The shear strength wastransmission-frequency spectrum. Experi- found to vary directly with Q-value.ments were conducted using a pair of broad-band acoustical transducers transmitting a Chang, Yee, and Couchman (4)Iapulsed ultrasound centered at 5 MHz through used an ultrasonic-frequency, spectral-analysismultilayer adhesively bonded aluminum plates technique for detecting flaws in compositewith different thicknesses. Resonant peaks in materials. This technique depended on thethe experimental frequency spectra were com- phenomenon of resonance interference ofpared with those theoretically calculated from acoustical waves in materials. When theregions of good bond and debond. material thickness was an integral multiple of

the half wavelength of the sound waves, de-The theory gave expressions for the structive interference of a return echo by

transmitted energy and transmitted flux coef- multiple reflections in the materials producedficient, and the effects of air gap and good antiresonance dips in the frequency spectrumadhesive layers were shown. Also, the effect for the reflected signal.of the adhesive-layer density, adhesive-layersonic velocity, and adhesive sonic impedance Their experimental results showedon the theory were discussed. The experi- that a well-bonded region of the compositemental results looked promising, but more had a very different frequency spectrum fromwork was needed. a delaminated region. They could detect

defects as small as 1.5 mm but were not ableChang, Couchman, Bell, and Gordon to differentiate among different sizes of de-

(23) in 1975 reported that NDE parameters fects. (See Figure 2-13.)evaluated from the ultrasonic spectroscopicmethod could be correlated with adhesive Chang et al. (95) utilized ultrasonicbond strength in multilayered structures. spectroscopy to evaluate the quality of theSelected portions of the reflected or trans- bond. They theorized that in adhesivelymitted RF signals were digitized; a Fourier bonded structures, destructive interference oftransform was performed, and resonance the pulsed sound waves at the boundaries ofpeaks or antiresonance dips in the frequency the bond layer produced spectral informationspectrum characterized the thickness of the characteristic of the bond. The bondlinesubstrate plates and the adhesive gap. The thickness could be determined accuratelyshape and the Q-value of the peaks and dips from the frequency minima in the spectra.determined the bond quality. The test sam- The width of the antiresonance dipped atpies were 2024 aluminum with PL717 adhe- half-power points, and the amplitude of thesive. Each layer was 0.125 inch (0.317 cm) dips were related to the acoustic properties at3 thick. the interfaces of the adhesive layer.

22I

I _ _ _ _ _

i a Good Area b c d

* ~ 25

20 15 15 15

115 10 0 10

2 34 5 6 78 2 3 4 5 6 7 8 2 3 4 56 78 2345878

Frequency [MHz)

Figure 2-13. Plots showing how the spectra from flat-bottom holes (b: 6mm, c: 3mm, d:1.5mm) differed considerably from the spectrum for the sound area (a). [Ref. (94),reproduced by permission of IPC Magazines Ltd.]

They provided an analytical calcu-lation for the sound-wave interference in the ___-_,_

bondline using a notation developed by Brek- co°t, \e ,, .hovskikh (96). This calculation included the 2000 L .... \ ,III..ultrasonic attenuation of the adhesive layerwhich, in effect, decreased the amplitudedifferences between the resonance peaks anddips in the frequency spectrum. Also, the ' ,000acoustic impedance was directly related to the oelastic modulus of the adhesive by A =(E p) 1/ 2 where E was the modulus and p wasthe density.0 I , I I "

The experimental setup used 15- IS ?5

MHz, L-wave pulse echo; a wideband pulser/receiver; and plates of 2024 Al and RB-398 Figure 2-14. Correlation of ultimate shearadhesives. Fabrication processes included low strength with 1/B [Ref. (5), reproduced bycuring temperature, unetched substrate sur- permission of the Institute of Electrical andfaces, adhesive sheet cutout, and insufficient Electronics Engineers, Inc.,0 1976]bonding pressure.

Their test results showed that the Raisch and Rose (88) examined theamplitude ratio correlated well with bond potential of selected ultrasonic signal featuresquality. Also, excellent correlation between and the ability of these features to predictthe reciprocal of the half-power bandwidth performance on an adhesively bonded struc-(1/B) and shear strength was found (see ture. Adhesive, rather than cohesive, prob-Figure 2-14). The amplitude ratio and the lems were considered. They claimed that the1/B had a linear relationship, which provided cohesive strength prediction had been solveda basis for a mathematical model to predict by Flynn (27) using measurements of wavebond quality. speed and attenuation in the bulk adhesive,

23

Ii

Iprovided that adhesive-type failure did not In the tests, a lO-MIz transducer

occur. The cohesive-strength prediction prob- was used. The results using three differentlem was also studied by Yee, Chang, and transducers showed correct prediction of 95,Flynn (28) using ultrasonic spectroscopy 95, and 89 percent (the training set was donemeasurements to characterize the physical with the first transducer). However, the algo-state of the adhesive in a bonded laminate. rithm was still felt to be very much transducer

dependent (especially on the pulse shape). AThe experimental work of Raisch transducer test was developed to evaluate the

and Rose (8) was conducted using 0.25-inch use of different transducers. This deconvolu-(0.63-cm) diameter, 10-MHz transducers at tion basically made all the transducers equalnormal incidence in the immersion mode. in their capability to predict good versus badThe data were acquired, digitized, and then bond.operated in the amplitude-time and ampli-tude-frequency domain. Thirteen features Ultrasonic pulse waveforms reflectedwere extracted for evaluation. Those in the from the interface of bonded joints wereamplitude-time domain included (1) peak-to- digitally recorded by Alers and Elsley (100).peak pressure variation, (2) the activity region Certain frequency-dependent features of the(obtained by using a spline curve containing reflected waves such as the lowest frequencyall the relative maximums of the rectified of resonance and the highest split resonancebond echo), (3) the area enclosed by the frequency were correlated to the mechanicalrectified amplitude-time bond echo within the strength of the bond. The pulse-echo methodactivity region, and (4) the ratio of positive- was used in the immersion mode. The signalsto-negative area enclosed by the amplitude- from the top of the metal surface and fromtime bond echo within the activity region. the interface bonded region could be sepa-The remaining nine features were in the rated. A drawback of looking just at theamplitude-frequency domain. They included amplitude ratios was that the bond strength(1) the maximum amplitude of the Fourier was really determined by a very thin layer atSpectrum, (2) the 6-dB down point on each the metal-to-adhesive interface, and the re-side of the maximum amplitude frequency flection from this critical layer was dominatedpeak, and (3) the number of relative maxi- by the metal-polymer impedance discontinuitymums and minimums occurring within the rather than the structure of the interface.bandwidth region of the spectrum. The fre- Their method was to Fourier transform thequency locations of the maximums and mini- entire signal through the composite structuremums that occurred at lower and higher fre- and to look at the frequency features.quencies relative to the spectrum maximumalso were picked, and then ratios of these The results showed that resonantamplitudes and their associated frequencies frequency was very dependent upon bondlinewere used. Using this technique on 45 speci- thickness. Other frequency-dependent vari-mens yielded an overall 84-percent accuracy ables needed to be evaluated and then corre-of calls. lated to the bond strength.

I Rose and Thomas (99) showed that Thomas and Rose (01) studied theultrasonic C-scan techniques can be used to problem of predicting adhesive bond perfor-detect delaminated bonds. A completely mance for both surface preparation and un-automated ultrasonic inspection system was dercure defects using an ultrasonic experi-developed for predicting bond quality in mental test bed system. The experimentalmetal-to-metal bonded step-lap joints, test bed incorporated ultrasonic and computerResults to date provide a 91-percent reliability equipment necessary to acquire and processfor solving the problem of predicting the data from various types of adhesively bondedadhesive bond performance. The system used test specimens. A set of 154 bond specimenspattern-recognition techniques, especially the was used to develop an algorithm that was 91-nearest neighbor rule and the Fisher linear percent reliable in separating good and poordiscriminator algorithm, which separated the bond samples. A Fisher Linear Discriminantbond data into strong and weak bonds. function was selected by the test bed as the

24I

best pattern-recognition routine for this classi- 6061T6 aluminum and the Dexter Hysonfication problem. adhesive EA 9628. The samples had different

surface preparations and were fabricated inThe features selected were: different humidity level environments. His

conclusion was that the frequency dip inter-9 Peak-to-peak ratio of the echo vals or the successive cutoff frequencies could

and the reference signal-frequency be correlated to the condition of the surfaceshift preparation and the level of humidity under

* Peak frequency which the bond was made.

I Deepest depression frequency 3. Low Frequency(dip frequency)

* Difference between the frequency Thompson, Thompson, and Alersof the dip and the peak frequency (7) reported results of several experiments

that suggested the adhesive bond quality of* Ratio of dip amplitude to peak honeycomb sandwich panels can be nonde-amplitude structively determined by measurements of

* Difference between the first- and vibrational response. The data showed thatsecond-dip frequencies both the damping characteristics of panels d uvibrations and the velocity of the propagation* Standard deviation of the transfer of elastic waves traveling along the surface

function and sample bondline can be correlated withe 6 dB beam width of the first dip. destructively determined bond strengths.

These results were a consequence of the vis-E. A. Lloyd (102) worked with coelastic properties of the adhesive, whereby

A-scan and frequency spectrum data from one changes in the elastic moduli and dampingsubstrate of the adhesively bonded composite. capacity of the adhesive can be related to theUsing the same transducer and equipment, he molecular structure and hence the strength oftook data through the adhesively bonded the adhesive.substrate and compared the two sets of data.The unbond condition was expected to be Data collected on honeycomb atsimilar to the data taken from the substrate.A discussion of the theory was given, which modulus and cohesive strength of the bondbasically said that when the transducer was with the damping of the UT signal. The fre-over the bonded region, a mixed frequency quency and temperature at which the damp-spectrum was obtained, which was a measure ing reached its maximum were interrelatedof the compliance of the adhesive layer. through the expression wt = I where w is 2irf

and t is the relaxation time according to theRecent work conducted by N. R. Arrhenius equation

Joshi (1_3) also utilized the ultrasonic spec-trum obtained from the interface region of t = to e W/ (7)the bond. The method he used was called theChirp-Z Transform (CZT). The CZT is ob- where t is a temperature-independent factor,tained by taking the fourier transform of the W is atemeati nen dent ator,A-scan (time-amplitude data display) and W is the activation energy, k is Boltsmann'sdetermining the location of dips in the fre- constant, and T is the absolute temperature.quency spectrum. These dips are caused by Damping of the vibrations of theinterference between the bond interface lay- adhesive rod specimens also correlated withers. The frequencies of the dips were used cure and strength. They plotted the logas center frequencies in the CZT to expand decrement (decrease) of decay of free oscilla-the frequency resolution. tions measured with bond strength. The data

showed that the lower the adhesive bond, themetric double-lap specimens fabricated from higher the decrement (see Figure 2-2).

25

I l i I II I I

I

Averbukh and Gradinar (104) con- propagated in a solid-fluid-solid guiding struc-ducted work with the Fokker Bond Tester. ture.They placed the piezoelectric transducers ona sheet of material and determined the acous- Thompson et al. (7) also did sometic characteristics. Then the composite mate- surface work and found that the surface-waverial (composed of the first sheet cemented to velocity increases with the adhesive bondanother sheet) was monitored to determine quality.its acoustic characteristics. The compositeproduced a shift in the frequency and acoustic 5. Lamb and Interface Wavesimpedance. The shift in frequency was indi-rectly proportional to the bond quality. The Rokhlin (107) analyzed diffractionBond Tester, however, had some short- of Lamb waves by a finite crack situated oncomings: (1) it required that the adhesive the plane of symmetry of an elastic layer.strength exceed or be equal to the cohesive The surface of the crack and the layer werestrength and (2) it imposed sometimes unat- assumed to be stress free. The field of thetainable technological requirements on the reflected and transmitted waves as well as thecementing. field in the vicinity of the crack were given as

expansions of natural waves of the elasticBudenkov et al. (105) showed that layer. The amplitudes of these waves were

the bond tester and the Stabmeter work only found from exponentially converging infinitewhere the adhesive strength was higher than systems of equations.the cohesive strength. Bond quality can bedetermined by correlating the strength and The interaction of waves at bondedthe characteristic impedance of the glue using interfaces was studied by Jones and Whit-ultrasonics. tier (108). They looked at plane-strain elas-

tic-wave propagation for two dissimilar half-4. Rayleigh and Surface Waves spaces joined together at a plane interface by

an elastic bond. The bond thickness wasStaecker and Wang (106) studied the assumed to be small compared to the wave-

application of acoustic Rayleigh wave propa- length; the existence of interface waves wasgation in a layered medium. A theoretical shown to be governed by a parameter involv-derivation was made for the propagation in a ing bond stiffness and wavelength. The infi-fluid between two solids. Experimentally, they nitely stiff bond case (fully cured) wouldused 30-MHz Rayleigh waves of 2-microsec- reduce to the Stoneley wave data; and theond duration on the exposed region of the infinitely soft bond would produce two Ray-second layer (see Figure 2-15). The wave leigh surface waves, one in each medium.traveled into the bonded region and excitedcharacteristic waves of the layer. Their work They also found that as the bonddemonstrated that acoustic waves could be became stiff, the Rayleigh waves could changeI

Elastic Waves Bound To A Fluid Layer

I. FluidOutput Input

Scotch Tape B Soda-lime Glass A Scotch Tape

~a - Quartz

Figure 2-15. Experimental layered geometry configuration used to generate Rayleigh-wave data [Ref. (106), reproduced by permission of the American Institute of Physics]

26I

I

3 into two interface waves, each involving quality was studied by Pilarski (110). Hismotion of both media. One of these waves, work dealt with two classes of problems:the one with the higher speed, disappeared (1) cases where a step-like change occurred infor a sufficiently stiff bond. With further the acoustic impedance such as bimetals orincrease in bond stiffness, the remaining in- glued metal bonds and (2) cases where theterface wave either went into a Stoneley wave bond was a thin intermediate layer whoseor disappeared, depending on whether or not thickness was much less than the wavelengthStoneley waves were possible. Furthermore, of the ultrasonic wave. The second case waswhen the Rayleigh wave speed in the faster when the layer was approximately equal tomedium was greater than the S-wave speed in the wavelength of the ultrasonic waves.the slower medium, the Rayleigh wave in thefaster medium existed only for zero bond One way to evaluate the adhesivestiffness. As soon as the stiffness became quality of a layered joint is based upon thefinite, the wave disappeared, and only a wave measurements of the pressure coefficient ofsimilar to the Rayleigh wave in the slower the reflected wave for the bond interface.medium may occur. Another way is to use ultrasonic waves propa-

gating parallel to the bond surface. TheseA new technique for analyzing inter- waves are called subsurface waves in the case

facial conditions in completed adhesive bonds of a layer on a base, plate waves in the casewas suggested by Claus and Kline (109). The of one or more solid layers, and interfacemethod was based on the sensitivity of Stone- waves in the case of two elastic half-spacesley waves, which propagated along the bound- (i.e., a solid medium of thickness severalary between dissimilar solid media, to changes times the wavelength of a surface mode).in the material properties of the interface Previous work had used the decay of theregion. Stoneley wave attenuation measured surface, plate, or interface waves in evaluatingafter polishing was found to increase as a the bond quality.function of increasing surface roughness inspecimens of borosilicate crown glass bonded The purpose of Pilanski's paperwith an anaerobic cement to a substrate of (110) was to use velocity measurement of7740 Pyrex mirror glass. waves propagating along the bonded surfaces

to determine the bond quality. The theoryWhile many NDE techniques have considered two types of boundary conditions,

been used to detect large-scale defects such as i.e., welded and adhesively bonded bound-voids or delaminations, no existing single aries. For the welded case, a continuity ofNDE technique can detect "weak bond" condi- displacement and stress occurred. For thetion. This work, therefore, was aimed at adhesively bonded case, a vanishing tangentdeveloping the Stoneley wave-measurement stress can occur. Decrease in adhesive qualitytechnique as a tool to determine the bond yields a decrease in the phase velocity of thequality. The experimental setup utilized sur- surface waves, as shown in Figure 2-16.face acoustic waves (SAW) from an x-cutpiezoelectric crystal mounted on a conven- Experimentally, the phase velocity oftional Rayleigh-angle water-coupled wedge. the ultrasonic waves can be measured usingAs the waves passed into the adhesively bond- critical-angle reflectivity. The change in theed region, they became Stoneley waves, which critical angle is related to the change in veloc-are affected by the boundary condition. The ity which, in turn, is related to the quality ofoutgoing waves were detected with an optical the bond. In general, the higher the phaseinterferometer. In these experiments, the velocity, the greater the bond.Stoneley wave attenuation increased as thesurface roughness of the two adherents in- Most of the previous UT methodscreased. used for determination of quality of adhesive

bonds used 0-degree L-waves, which are notThe use of ultrasonic wave-velocity sensitive to the adhesion properties between

measurements of surface, plate, or interface the adhesive and the adherents. Rokhlinwaves for the evaluation of adhesive bond (111) used guided waves to produce shear

27I

Ibetween the two sets increases.

cI [m/sl for M 1 (See Figures 2-17 and 2-18.)

2400 2600 2800 3000 When the thicknesses of the bondedI I I I substrates are comparable to the wavelength,BO 1010 g1 ,Fe, g2 Lamb waves are used. The dispersion char-

300 Ml M21 -acteristic of Lamb waves can be useful to30 -monitor the slip (liquid) and rigid (bonded)

boundary conditions. During the curing proc-

/2 ess (liquid to solid), the Lamb wave propaga-S 200 - tion mode is transformed with a change in

phase from the mode in the slip condition tothe mode in the rigid condition.

(100 (g=g2 ) / g1- 3

t-g = 0.55 MHz •mm

4100 4300 4500 4700

cf [m/s] for M 21 1-

0.6 - _'

Figure 2-16. Relationship between phase 660.4 /12velocity of the first two modes of plate waves & 04" sand shear strength of the adhesive joint in 'b#.'bimetal [Ref. (110), reproduced by permission 18 ..- aof .6_Q

of ASN • o!cmse*210 .aa00 0 a~ ~A O 4 0.

60 120 180 240

stress on the interface. Interface waves were Time (min)applied for monitoring curing of the adhesive,and Lamb waves were used to study adhesive Figure 2-17. Change in the interface wavejoints of the thin sheets. velocity as a function of curing time for agedII

specimens [Ref. (111), reproduced by permis-An interface wave can be obtained sion of the National Aeronautics and Space

in the adhesive joint by generating a surface Administration (NASA)-Lewis]wave into the lower substrate. The thicknessof the layer must be smaller than the UTwavelength. Bond failure can occur inside theadhesive (called cohesive failure) and between °the adhesive and adherent (called adhesivefailure). Using the interface waves, the fol- g0.2 -lowing can be observed:

" When contact between the adhe- < 0.4

sive and the substrate is not ideal,the interface wave velocity de- 0.6creases. [

* By monitoring two interface 0 4 8 12 16 20 24

waves, one through an unbonded Aging (Days)

region (calibration) and onethrough the region that is adhe- Figure 2-18. Change in time delay as a func-sively curing, as the curing of the tion of the aging time [Ref. (1LL), repro-adhesive cures, the phase velocity duced by permission of NASA-Lewis]

I28I

I

I Data were collected using 0.5-, 1-, face wave produced shear stress only in the1.5-, and 2-MHz transducers on adhesive film.sz film. This worked in the following way. First,that were 80 to 90 microns thick (appioxi- a Rayleigh wave was propagated on the sur-mately 3 to 4 mils). The 1- to 2-MHz data face of the top part. If two plates were putwere good; the 0.5-MHz data, however, did together with a thin layer between them withnot show good results due to edge problems. the layer thickness much smaller than the

Rayleigh wavelength and much larger thanRokhlin. Hefets, and Rosen (112) the hydrodynamic boundary layer, then the

studied a thin adhesive film between two phase velocity of the interface wave (V,) wasbonded adherents and showed that it was smaller than the Rayleigh wave velocity (Vr).capable of localizing the energy of elastic The relationship iswaves in the form of an interface wave. Thephase velocity and transmission losses of the VI/Vr = I - KthBrpo/4par (9)interface wave were measured during thecourse of polymerization of the adhesive, where Br = (ar2 1)1/2, and ar = Kr/K t are theThe phase velocity of the interface wave and wave number for the Rayleigh and shearthe effective shear modulus of the interface waves in the substrates, p is the substratefilm were related to the quality of the adhe- density, and h is half of the layer thickness.sive bond, as shown in Figure 2-19.

Now if the layer thickness were sinai-The thin film located between plates ler than the boundary layer, then the shear

exhibited waveguide properties if the shear displacement was not damped out across themodulus of the film were smaller than that of film thickness and was transmitted from thethe substrates. When the film thickness was lower to the upper substrate without a breakmuch smaller than the wavelength, the inter- in continuity. Thus, as the film solidified, the

phase velocity tended toward the shear veloc-ity. The measured velocity and damping

1 factor of the interface waves could be used tocalculate the complex shear modulus of theinterface film. When there was an absence ofI 0.8 shear bonding between the adhesive and at

C, least one substrate, then velocity of the inter-i face wave was close to the Rayleigh wave

0.6 velocity.

Two 1.2-MHz narrow-band trans-0.4 ducers were used for the tests. The greater

a the change in velocity, the greater was the

0quality. Also, the greater the transmission1 0.2 loss, the greater was the quality.

7' Rokhlin, Hefets, and Rosen (ii-1)

..... ialso studied the waveguide properties of a0.5 1.0 1.5 2.0 2.5 3.0 3.5 thin film separating two elastic half-spaces

possessing a shear modulus higher than theNormalized Velocity Changes shear modulus of the film. It was shown

(AV -L- -1) (at %) analytically and experimentally that such aVr film was capable of localizing the energy of

Figure 2-19. Experimental relationship be- elastic waves near the interface. The velocitytween the interface wave velocity and the of the interface wave was determined by thenormalized shear strength [Ref. (112), repro- shear modulus, density, thickness of the film,duced by permission of the American Insti- and elastic properties of the substrates. Thetute of Physics] velocity of the interface wave was between the

* 29

I

I k,,ieigh and the shear wave in the solid Experimentally, the stress-wave fac-substrates. It was shown that the interface tor measurements were obtained using thewave could be used to estimate the elastic and portable acoustic emission technology (AET)dissipation properties of thin interface layers. AU instrument. The pulsing transducer had

a flat response from 0.1 to 3 MHz. The re-6 Stress-Wave Factor ceiving transducer was resonant at 375 kHz.

r eThe injected pulse was 150V at a rate of 250Vary et al. (1.4-11_6) showed that the pulses/second.

ultraso.nic parameter called the stress-wavefactor correlated with the tensile strength and 7. Horizontally Polarized Shear Wavesinterlaminar shear strength of graphite fibercomposites. Williams and Lampert (117) Horizontally polarized shear (SH)looked at impact damage to composites and waves are generated with special transducerscompared the through-thickness attenuation that emit shear waves at 0 degrees. In orderand the stress-wave factor measurements to to propagate SH waves, a viscous couplant

the degradation. must be applied to effectively couple theshear waves into the part.

Experimentally they used two 0.1 to

3 MHz, broadband through-transmission Yew (112) performed a preliminarytransducers to determine the attenuation study using SH waves on a bonded plate tothrough the damage composite, and a 0.1 to estimate the bonding quality of the adhesive3 MIlz, broadband-transmitting transducer bond. The concept was that the characteris-with a 375-kHz receiving transducer to meas- tics of the SH waves in the bonded structureure the stress-wave factor. The stress-wave were dependent upon the bonding stiffness offactor was defined as the adhesive; thus, an estimation of the ad-

e = g r n (8) hering strength can be made by observing thebehavior change of the SH wave motion.

where g is the accumulation time after which Two measurable modes of SH waves are inthe counter was automatically reset, r was the the structure during the early stages of adhe-transmitter repetition rate, and n was the sive curing, and the amplitude of the second-number of cycle oscillations exceeding a fixed mode SH wave decreases as the adhesivethreshold in the output waveform generated cures and finally disappears after the adhe-by the input pulses. The stress-wave param- sive-curing process is completed.eter characterized the specimen length, whilethe through transmission characterized the In addition, Yew studied the propa-specimen's thickness. They concluded that gation of Stoneley waves in the interface ofboth inspection techniques provided adequate the two bonded solids. It was shown that themeans to determine quality of the composite. attenuation and the propagation speed of

sonic waves in the medium were related toDos Reis, Bergman, and Bucksbee the bonding strength and the stiffness of the

(118) conducted tests to evaluate the adhesive adhesive. A mathematical model for usingbond quality between rubber and steel plates SH waves to determine bond quality wasusing the stress-wave factor measurement described. When the cohesive strength wastechnique. This technique measured the low, the wave behavior of the bond was simi-relative efficiency of energy transmission into lar to that of a free plate. The wavelength ofthe specimen. A UT pulse was injected with the applied SH wave was so long in compari-a transducer mounted on the surface of the son with the plate thickness that a standingspecimen. The number of oscillations higher wave was established along the thickness ofthan a chosen threshold were monitored. The the plate.more internal damage, the greater was theattenuation and the less efficient the energy The experimental tests were con-transmission (lower ringing). The ringing was ducted on aluminum test plates bonded withdetected by another transducer on the sur- DER 332 epoxy resin with a T403 amineface. curing agent and 399 accelerator. The SH

I 30

I

I were generated with a 0.5-MHz transducer. and attenuation in the bulk of the sample.A 600-kHz wave-train frequency, 30 micro- The change in phase velocity was determinedseconds long, was delivered to the strip at a by measuring the phase shift of the RF signal.rep rate of 500 Hz using an Arenburg pulsed Transducers used were 1.3 MHz. The re-oscillator. After the wave train had traveled flected signal amplitudes as a function of cure4 inches (10 cm), it was detected by a 5-MHz, time for three different adhesive thicknessesSH transducer. This set up two modes in the (8, 35, and 75 microns) were plotted, asplate. The second mode signal clearly de- shown in Figure 2-20. The thinnest showedcreased as the adhesive cured, while the first the most dramatic effect, changing approxi-mode stayed relatively constant. One prob- mately 11 dB over the cure cycle, while thelerr with this technique was ensuring proper thickest samples only changed 3 dB over thecoupling of the SH waves. Another issue was cure cycle. That is, the technique worked bestthat the width of the receiver transducer when the ratio of the adhesive layer thicknessshould be shorter than the SH wavelength, to the UT wave velocity was approximately

0.01.

& Obliquely Incident Ulrsonic Waves

ultrasonic energy introduced into a part at -2 - 75 p.

large angles relative to the normal of the part 0 0 o(angles usually above 60 degrees from the -4 _normal). In many cases, the reflection coeffi- C 0 Acient measured at the bond interface peaks at -6 0 35O 35 p

a relatively high or obliquely incident angle. 0 - o

Using obliquely incident waves, -Rokhlin and Marom (120) demonstrated that < 10 0

amplitude changes of the reflected waves o 84during adhesive curing occurred simultane- -12 00 Jously with changes of the wave velocity in the 100 200 300adhesive. The sensitivity of the method toevaluate the interface properties was higher Time (mi)for thinner adhesive-interface films. It wasshown by Kuhn and Lutsch (121) in 1961 that Figure 2-20. Dependence of the amplitude ofthe reflection coefficient for obliquely incident the reflected signal on the time of adhesiveL-waves from a slip interface may be higher curing for different thicknesses of the adhe-in some cases than from a free surface; thus, sive (metal-metal bond) [Ref. (120), repro-the value of the reflection coefficient can be duced by permission of the American Insti-used for evaluating the interface properties. tute of Physics'

3 Bulk L-waves are insensitive to theexistence of a thin liquid layer at the inter- Similar results were obtained byface, which exhibits no shear strength resis- Pilarski and Rose (122) using obliquely inci-tance. To measure properties of the adhesion dent shear waves. They looked at adhesivelybetween the adhesive and the substrate and to bonded aluminum samples and found that themonitor the adhesive cure, shear deformation reflection coefficient as a function of angleof the interface must be induced. The theory through a bonded structure varied as a func-was discussed, and optimum angles for inspec- tion of the bond rigidity.tion were derived.i D. Acoustic Emission Techniques

The experiment was discussed. They

measured (1) the amplitude of the reflected Acoustic emission (AE) techniques utilizesignal at the joint and (2) the L-wave velocity piezoelectric transducers primarily to receive

I* 31

3 high-frequency pulses generated in a material can be used to receive the penetrating radi-by some change within the material such as ation and generate an image.defect initiation or growth.c iGamma-ray or x-ray penetration is af-

Acoustic emission (105) can be used to fected by the condition of the material suchevaluate bond quality. A defective bond pro- as density variations and the presence ofduces a large increase in AE during loading, voids or inclusions. This type of radiographyThis method, however, is not being used in can be used to detect damage to honeycomb,practical applications because it has not been presence of debris and contaminants, and lacksufficiently developed or have been applied of bond due to absence of adhesive (2).for only a few cases. Averbukh and Gradinar(104) collected data on bonded compcz.tes Neutrons are affected more by nuclearand found that the amount of AE as a func- processes than density variations such astion of load varied depending upon the qual- proton-neutron scattering. Most organicity of the bond. adhesives contain between 8-and 12-percent

hydrogen (proton), so neutron radiographyE. Holographic Techniques becomes a viable tool to inspect for adhesive

problems. Dance and Petersen (124), using aHolography is a fairly complex optical californium-252 source, demonstrated that

technique which requires a laser, a test object, neutron radiography could be used to detectand an identical reference object. The laser defects in adhesives.light illuminates both the reference object andobject under test. Any differences in the G. Thermographyreflected light from the object under testcompared to the reference shows up as a Thermographic techniques utilize heatchange in the diffraction pattern. Changes on energy and monitor how this energy interactsthe order of a fraction of a wavelength of the with and travels through the object underlaser light can be perceived (on the order of inspection. The wavelength of this energy ismicroinches). usually in the range of 8 to 20 microns. In

order to detect or image this energy, specialMueller, Gupta, and Keating (123) devel- detectors are needed.

oped a holographic weak-signal-enhancementtechnique (WSET) which enhanced the image Thermography can be separated into twoof the source of a weak signal of interest in methods, passive and active (75). The passivethe presence of an unwanted strong signal method utilizes an external heat source whilewith very little a priori knowledge. The feasi- the active method utilizes internal tempera-bility of the WSET was shown by simulating ture changes when an external force is applieda two-solid-layer acoustic problem on a digital that produces internal frictional heating. Thecomputer. The enhancement method was thermographic method makes use of the dif-based upon the use of the signal field for ference in thermal conductivity of a goodreconstruction and high-pass filtering of some bond and a poor bond. In the poor bond,of the information. there could be a discontinuity at the adhe-

sive/adherend interface which would produceF. Radiographic Techniques a poorer thermal conductor.

Radiographic techniques refer to the use In practice, passive thermography utilizesof penetrating atomic or nuclear particles to an external heat source and an infrared cam-image an object. The most common ways to era (or other imaging system sensitive toobtain penetrating radiation are by using an thermal radiation). Monitoring of thermalisotopic gamma-ray source, x-ray tube, or transients is the most effective method. Mon-neutron source. Each provides radiation to itoring of absolute temperatures is difficultilluminate an object. The radiation passes because of variations in thermal emissivities,through the object under test and is usually nonuniform heating, and many other prob-imaged on film. Electro-optical devices also lems. The defects appear as cooler regions (if

32I

Ia through-transmission technique is used) the defect membrane resonance. Again, how-because of the lower thermal conductivity ever, very little work has been conducted tothrough them. If the camera and the heat determine the capability of active thermog-source are on the same side, then the defects raphy to determine the quality of the bond.will appear hotter because they do not trans-mit the heat as well. In addition, the sensitiv- Recently, a large amount of work hasity of the method decreases as the thermal been conducted using thermographic tech-conductivity and depth of the defect increase. niques to detect the presence of debond.

However, this work was presented as papersWork by McLaughlin et al. (125) and at seminars and technical meetings and has

Reynolds and Wells (126) has shown that not yet been published.thermography can be used to detect delamina-tions and voids. Very little work has been H. Nuclear Magnetic Resonancedone to determine its capability to actuallydetermine the quality of the bond. Nuclear magnetic resonance (NMR) is

primarily used to detect the presence of hy-Active thermography makes use of an drogen, especially in the form of moisture.

externally applied force and an infrared cam- This method has not been used to determineera. For example, by vibrating the surface of bond quality; however, it is believed that worka bonded structure, the defective regions will should be conducted using NMR since one ofexperience a local rise in temperature due to the major causes of bond failure is the inges-frictional heating. These defects can then be tion of moisture. This method requires theobserved using the infrared camera. Work by use of a large magnetic field and an RF exci-Russell and Henneke (127) has shown that ter coil. This method cannot be used fordefects in composite structures can be de- metals, but should be of practical use fortected if the vibrational frequency is that of composite structures.

IIIIIIIII

33I

3 III. QUALITY CONTROL AND INSPECTION NEEDS

A. Materials After surface treatment, the adher-end needs to be checked for the wettability,

1. Adhesives surface roughness, uniformity of the treat-ment, and lack of contaminated area. If

Quality control of the adhesives priming is involved, primer thickness, curingselected for bonding is very important to state, and presence of defects require inspec-obtain bonded joints of desired strength tion. In addition, temperature, humidity, and(1, 4). Testing and inspection are needed to cleanliness of the preparation area must beassure that the physical and chemical prop- monitored and controlled.erties of the adhesives meet the requiredspecifications. Properties requiring controls B. Bonding and Curing

I include mix proportion of the base and addi-tive materials, fiber-resin volume ratio, mix Control of the bonding and curing proc-homogeneity, gelation time, resin content, esses is another key for a successful bond-fiber alignment, moisture content, viscosity, ing (1, 4). During adhesive application, theflow, and tack. thickness and uniformity of the adhesive must

be monitored and controlled. For film-type2 Adherends adhesive applications, the presence of wrin-

kles, trapped air bubbles, or foreign substanceThe quality of incoming adherend such as release paper must be checked. For

details comprising the joint must be inspected honeycomb sandwich structures, measuringand tested to assure their meeting acceptance and controlling the size of the fillet is re-requirements (1, 4). Parameters such as size, quired. Proper assembly and fitting of com-thickness, material type, and general appear- ponents or proper stacking sequence duringance as to distortion, corrosion, or defects lay-up must be monitored; and curing temper-require checking. Careful control of the sub- ature, pressure, and time has to be measuredsequent surface preparation of the adherends and controlled. To control the curing process,is of paramount importance for production of the degree of curing must be monitored.reliable bonded joints (1, 4, 9). All the proc- After the adhesive is cured, the bonded areaesses involved in the surface preparation-- needs to be inspected for defects such ascleaning, rinsing, chemical treatment, drying, debonds and voids.and/or priming--must be monitored and con-trolled. Physical and chemical properties of C. Inservice Inspectionthe materials used in these processes such asprimer, solvent, water, and solution for chemi- The strength of the bonded joints andcal treatment must be tested and inspected components degrades with service throughfor temperature, composition, and contami- development of environmentally inducednants. defects such as cracks, debonds, delamina-

tions, and corrosion. To ensure the structuralintegrity and safety of the bonded assembly,periodic inspections are essential.

IU

U 34

I

IV. SUMMARY

This report has presented a large amount of method to detect large, near-surface delami-information on the subject of adhesive bond- nations.ing. Applications of adhesive bonding in usetoday were cited along with associated advan- Other methods also were evaluated includingtages and disadvantages. A generalized dis- acoustic emission, optical holography, radiog-cussion of the bonding process and the raphy, and thermography. Application ofparameters causing bonding problems were these methods to certain bonding problemsdiscussed as well as failure mechanisms. has shown good success in differentiatingFinally, a brief but thorough coverage of the between a bond and debond, but very littlenondestructive methods utilized in the past to capability in quantifying the quality of theevaluate the condition of the bond were ad- bond. Radiography in reality is used to detect3 dressed. defects such as inclusions and voids in the

adherend/adhesive layer that can cause poorMost of the information in the report deals bond quality. Nuclear magnetic resonancewith the use of NDE methods employed to has some potential for determining bonddate, involving such ultrasonic techniques as: quality in nonmetallic structures by detecting

the presence of moisture in the bond. Little" Through-transmission or pulse-echo reflec- work, however, has been done with this tech-

tion of longitudinal waves with the adher- nique.end/adhesive/adherend layer (in both timeand frequency domain), While many NDE techniques have been

applied to assess bond quality, most of these" Stoneley waves interacting with the adhe- can only determine whether there is a bond

sive layer, (of some unknown quality) or a complete lackof bond (debond). This differentiation

* Rayleigh waves, capability of an NDE technique is indicatedby an X in the bond/debond column in

" Obliquely incident waves, and Table 4-1 on the following page. Othertechniques have shown some promise in being

* Horizontally polarized shear waves, able to determine if a bond is good, average,poor, or unacceptable. These categories are

Work using all these techniques has indicated called bond quality, and the techniques withresults that correlate with bond quality in a this capability or potential to distinguish bondlimited number of tests. quality are listed with an X under the bond

quality column. Some techniques also canAdditional work has been conducted using detect the qualities of the adhesive such assonic waves, and the body of evidence shows adhesive thickness or the presence of inclu-that sonic waves can be used to detect de- sions or porosity in the adhesive layer. Thesebonds. The "tap" technique has been used techniques are listed with an X in the inclu-throughout the industry as a quick and easy sions and thickness columns.

35IIU

35

I

I

I Table 4-1

SUMMARY OF NDE METHODS AND TECHNIQUES APPLIEDTO DETERMINE ADHESIVE BOND QUALITY AND AN

EVALUATION OF THEIR CAPABILITIES

Bond/ BondMethod/Technique Debond Qualily Inclusions Thickness

SonicsCoin Tap X XMechanical Impedance X XMechanical Resonance X X X

UltrasonicsPulse-echo X X XThrough-Transmission X XFrequency Spectrum X ?X*X XRayleigh Waves X ?X* X XInterface Waves X X XStress Wave Factor X ?X* X XHorizontally Polarized

Shear Waves X X X XObliquely Incident Waves X X X

* Acoustic EmissionPassive X XActive X X

Holography X ?X* X

I Radiography X X

Thermography X X

3 Nuclear Magnetic Resonance X X

*Under Bond Quality, ? denotes that these techniques have demonstrated some theoretical

or experimental potential for determining bond quality.

IIII

36I

IV. CONCLUSIONS

Almost every NDE method has been utilized rial interact with the adherend/adhesive inter-at one time or another in an attempt to face, but more work must be done to furtherdevelop a technique to adequately describe evaluate this technique.bond quality in a bonded structure (128, 129).Most of the work to date has centered around Based upon the amount of work conducted toapplication of ultrasonic techniques. The date, the most promising approach for theinteraction of the through-transmitted (or largest variety of bonding inspections is appli-reflected) longitudinal (L) wave with the cation of one of the ultrasonic techniques.adherend/adhesive/adherend interfaces can The information in this report is based uponbe used to easily detect the difference be- multiple case studies, some of which weretween the good bond and the total debond. correlated with destructive testing; the oneThe L-wave, however, does not perturb the point, however, that seems apparent is that noadhesive layer in a way that would provide a one technique was tested on a large variety ofmeans to interrogate the strength of the adhe- different bond types with known bondsive layer or the adherend/adhesive interface, strength.

Other ultrasonic waves such as the Stoneley A program should be conducted using a largeand interface waves directly interact with the number of ultrasonic techniques on a carefullyadhesive layer so that the strength of the produced set of samples that would coveradhesive or adherend/adhesive interface can most bonding situations. The test plan shouldpotentially be measured. Unfortunately, the be coordinated with both NDE as well asapplication requirements for these techniques adhesive bonding experts. After a thoroughmake it almost impossible for field inspec- evaluation, the samples should be destruc-tions. Horizontally polarized shear waves tively tested to determine the actual bondlaunched from the top surface into the mate- strength.

I37III

37I

I3 VI. REFERENCES

1. Cagle, C. V. Adhesive Bonding: Tech- 10. Ray, A. "Adhesives--Classic Member ofniques and Applications. New York: Industrial Society." Chemical Engineer-McGraw-Hill, 1968. ing World. 21, No. 3 (1986), pp. 42-44.

2. Cotter, J. L., and M. G. D. Hockney. 11. Hagemaier, D. J. "Bonded Joints and"Metal Joining with Adhesives." Inter- Non-Destructive Testing: Bonded Hon-national Metallurgical Reviews. 19 eycomb Structures--I." Nondestructive3 (1974), p. 103. Testing. 4 (1971), p. 401.

3. Petri, E. M. "Plastics and Elastomers as 12. Segal, E., and J. L. Rose. "Nondestruc-Adhesives." Handbook of Plastics and tive Testing Techniques for AdhesiveElastomers. Chap. 10. C. A. Harper, Bond Joints." Research Techniques ined. New York: McGraw-Hill, 1975. NDT. Chap. 8, Vol. IV. New York:

Academic Press, 1980.i 4. MIL-HDBK-691B. MilitaryStandardiza-

tion Handbook- Adhesive Bonding. 13. Cook, T. N., and R. A. Prince. Inspec-Washington, D.C.: U.S. Department of tion and Repair of Advanced CompositeDefense, 1987. Airframe Structures for Helicopters.

USAAVRADCOM TR-82-D-20. Strat-5. McMullen, P. "Fibre/Resin Composites ford, Connecticut: Sikorsky Aircraft

for Aircraft Primary Structures: A Division, United Technologies Corp.,Short History, 1936-1984." Composites. 1983.15 (1984), p. 222.

14. Dreger, D. R. "Design Guidelines for6. George, D. L. "An On-Site View of the Joining Advanced Composites." Fabri-

Inspection and Repair of Carbon Fibre cation of Composite Materials. M. M.Composite (CFC) Aircraft Structures." Schwartz, ed. Metals Park, Ohio:British Journal of Nondestructive Testing. American Society for Metals (ASM),II27 (1985), p. 22. 1985, pp. 9-13.

7. Postier, R. A. "Factory Automation for 15. Patrick, R. L., ed. Treatise on AdhesionComposite Structures Manufacturing." and Adhesives: Materials. Vol. 2. NewSAMPE Quarterly. April 1985, pp. 45- York: Marcel Dekker, Inc., 1969.

49.16. Lee, W. J., J. C. Seferis, and D. C. Bon-

8. Moody, I. N., P. A. Fay, and G. D. ner. "Prepreg Processing Science."Suthurst. "Can Adhesives Meet the SAMPE Quarterly. 17, No. 2 (1986), pp.Challenge of Vehicle Bodyshell Con- 58-68.struction?" Sheet Metal Industries. 64,No. 7 (1985), pp. 332-340. 17. Hergenrother, P. M. "Recent Advances

in High Temperature Polymers." Poly-9. Ludbrook, B. D. "Heat Curing Adhe- merJournal. 19 (1987), p. 73.

sives." Proceeding of InternationalConference on Structural Adhesives in 18. Wedgewood, A. R., K. B. Su, and J. A.Engineering, Institution of Mechanical Narin. "Toughness Properties and Serv-Engineers. Conference Publications ice Performance of High Temperature1986-6. London: Mechanical Engineer- Thermoplastics and Their Composites."ing Publications Ltd., 1986. SAMPE Journal. January/February

1988, pp. 41-45.

I38I

I19. Carlsson, G. "How to Maintain a Con- 26. Trawinski, D. L., and R. D. Miles. "A

sistent Quality in Your Bonding Proc- Comparison of Conversion Coating andess." Progress in Advanced Materials and Chemical Etching Surface PreparationsProcesses: Durability, Reliability, and for Steel." Proceeding of 16th NationalQuality Control. G. Bartelds and R. J. SAMPE Technical Conference. Vol. 16.Schliekelmann, eds. Amsterdam: Else- Azusa, California: SAMPE, 1984, pp.vier Science Publishers B.V., 1985, pp. 633-643.I 61-81. 27. Trawinski, D. L., D. K. McNamara, and

20. Roberts, R. W. "Automated Compos- J. D. Venables. "Adhesive Bonding toites Cure Control Implementation: A Conversion Coated Steel Surfaces."Cure Modeling Approach to Automa- SAMPE Quarterly. 15, No. 3 (1984), pp.tion." SAMPE Journal. September/ 6-13.

I October 1987, pp. 28-33.o 28. Huntsberger, J. R. "The Mechanisms of

21. Colton, J. S., J. Baxter, J. Behlendorf, Adhesion." Treatise on Adhesion andT. Halim, B. Harris, G. Kiesler, K. T. Adhesives: Theory. Chap. 4, Vol. 1.Lu, S. Sammons, and W. Savage-Moore. R. L. Patrick, ed. New York: Marcel"The Automation of the Lay-Up and Dekker, Inc., 1966.Consolidation of PEEK/Graphite Fiber

* Composites." SAMPEJournal. Septem- 29. Kinloch, A. J. "Review--The Science ofber/October 1987, pp. 19-25. Adhesion: Part 1, Surface and Inter-

facial Aspects." Journal of Materials* 22. Hennemann, 0. D., and W. Brockmann. Science. 15 (1980), p. 2141.

"Surface Morphology and Its Influenceon Adhesion." Journal of Adhesion. 12 30. Patrick, R. L., ed. Treatise on Adhesion

(1981), p. 297. and Adhesives: Theory. Vol. 1. NewYork: Marcel Dekker, Inc., 1966.

23. Brown, S. R. "An Evaluation of Tita-

nium Bonding Pretreatments with a 31. Good, R. J. "Intermolecular and Inter-We(ge Test Method." Proceedings of atomic Forces." Treatise on Adhesionthe 27th National SAMPE Symposium. andAdhesives: Theory. Chap. 2, Vol. 1.Vol. 27. Azusa, California: Society for R. L. Patrick, ed. New York: Marcelthe Advancement of Material and Proc- Dekker, Inc., 1966.ess Engineering (SAMPE), 1982, pp.363-376. 32. McMahon, P. E., and L. Ying. "Effects

of Fiber/Matrix Interactions on the24. Dezfuly, M. A., C. Vachos, and E. H. Properties of Graphite/Epoxy Compos-

Andrews. "Oxide Morphology and ites." NASA Contractor Report 3607.Adhesive Bonding on Titanium Sur- Houston, Texas: NASA, 1982.faces." Journal of Materials Science. 19(1984), p. 3626. 33. Rose, J. L., and P. A. Meyer. "Ultra-

sonic Procedures for Predicting Adhe-25. Jones, M. T., P. D. Pitcher, P. Poole, sive Bond Strength." Materials Evalu-

and M. H. Stone. "Effect of Surface ation. 31 (1973), p. 109.Pretreatment and Alloy Type on theDurability of Adhesive Bonded Tita- 34. Hart-Smith, L. J. "Further Develop-nium Alloy Joints." Proceedings of Inter- ments in the Design and Analysis ofnational Conference on Structural Adhe- Adhesive-Bonded Structural Joints."sives in Engineering, Institution of Joining of Composite Materials. Ameri-MechanicalEngineers. Conference Publi- can Society for Testing and Materialscations 1986-6. London: Mechanical (ASTM) Special Technical PublicationEngineering Publications Ltd., 1986, pp. (STP) 749. K. T. Kedward, ed. Phila-93-104.

39I

deiphia, Pennsylvania: ASTM, 1981, cal Society (ACS) Symposium Seriespp. 3-31. 322, Chap. 16. Washington, D.C.:

ACS, 1986.35. Venables, J. D. "Review: Adhesion

and Durability of Metal-Polymer 43. Srinath, V. S. "Critical Factors Affect-Bonds." Journal of Materials Science. ing the Stress Durability of Adhesive

.19 (1984), p. 2431. Bonds." Structural Adhesives and Bond-Iing: Theoretical Aspects, Materials &36. Brockmann, W., 0. D. Hennemann, Processes, Apolications/Case Histories.

H. Kollek, and C. Matz. "Adhesion in Papers presented at Structural Adhe-Bonded Aluminum Joints for Aircraft sives and Bonding Conference, MarchConstruction." International Journal of 13-15, 1979. El Segundo, California:Adhesion and Adhesives. 6 (1986), Technology Conferences Associates,p. 115. 1979, pp. 24-47.

37. Munjal, A. K. "Optimization of Design 44. Boerio, F. J., R. G. Dillingham, A. R.Allowables for Composite Structures." Snow, and C. A. Gosselin. "SilaneSAMPE Quarterly. 18, No. 2 (1987), pp. Primers for Improved Hydrothermal18-27. Stability of Adhesive Bonds to Alu-

minum." Proceeding of the American38. Goan, J. C., and S. P. Prosen. "Inter- Chemical Society Polymer Materials:

facial Bonding in Graphite Fiber-Resin Science and Engineering Division.Composites." Interfaces in Composites. Vol. 50, No. 187. Washington, D.C.:ASTM STP 452. Philadelphia, Pennsyl- ACS, 1984, pp. 444-447.vania: ASTM, 1969, pp. 3-26.

45. Liang, F., and P. Dreyfuss. "Durability

39. McMillan, J. C. "Surface Preparation-- of Adhesive Bonds Between Glass orThe Key to Bondment Durability." Metal Substrates and a Polybutadiene-Bonded Joints and Preparation for Bond- Polyurethane," Journal of Applied Poly-ing. Advisory Group for Aerospace merScience. 29 (1984), p. 3147.Research & Development (AGARD)Lecture Series No. 102, North Atlantic 46. Bryson, J. H. "'he Feasibility of Adhe-Treaty Organization, Chap. 7. Ottawa, sively Bonding Preprimed Steel forCanada: AGARD, 1979. Structural Applications." Society of

• Automotive Engineers (SAE) Technical40. Kodali, S. P., R. C. Curley, L. Cottrell, Paper Series No. 870175. Warrendale,

B. M. Ditchek, D. K. McNamara, and Pennsylvania: SAE, 1987.J. D. Venables. "Effect of Rinse WaterpH on Adhesive Bond Durability." 13th 47. Kaelble, D. H. "Surface Energetics andNational SAMPE Technical Conference. Structural Reliability of AdhesiveVol. 13. Azusa, California: SAMPE, Bonded Metal Structures." Adhesion1981, pp. 676-684. and Adsorption of Polymers: Polymer

Science and Technology. Vols. 12A41. Pocius, A. V., C. J. Almer, R. D. Waid, and B. L. H. Lee, ed. New York:

T. H. Wilson, and B. E. Davidian. "In- Plenum Press, 1979, pp. 301-312.vestigation of Variability in the Adhe-sive Bonding Characteristics of 301 48. Hinrichs, R., and J. Thuen. "Environ-Stainless Steel." SAMPE JournaL 20, mental Effects on the Control ofNo. 6 (1984), pp. 11-16. Advanced Composite Materials Proc-

essing." SAMPE Journal. 15, No. 642. Gosselin, C. A. "Effect of Surface Prep- (1979), pp. 12-21.

aration on the Durability of StructuralAdhesive Bonds." Polymeric Materials 49. Quick, S. "Effects of Cure Temper-for Corrosion Control. American Chemi- ature and Pressure on Adhesive Bond

40I

Properties of P2 Etched Aluminum." Pennsylvania: Technomic PublishingProceeding of the 28th National SAMPE Co., 1984.Symposium and Exhibition. Azusa, Cali-fornia: SAMPE, 1983, pp. 1116-1126. 59. Gosselin, C. A. "The Durability of Lap-

Shear Adhesive Bonds Following Accel-50. Pike, R. A., F. P. Lamm, and J. P. erated Environmental Testing." Corro-

Pinto. "Factor Affecting the Processing sion Protection, Resistance and Testing.of Epoxy Film Adhesives III Heat-Up SAE Special Publications SP-612. War-Rate." Journal of Adhesion. 17 (1984), rendale, Pennsylvania: SAE, 1985, pp.p. 51. 15- 23.

51. Purslow, D., and R. Childs. "Autoclave 60. Awa, V. S., J. R. Vala, and M. Jeya-Moulding of Carbon Fiber-Reinforced seelan. "Effect of Impact and FatigueEpoxies." Composites. 17 (1986), Loads on the Strength of Graphite/

I p. 127. Epoxy Composites." Composite Mate-rials: Testing and Design. ASTM STP

52. Wolfe, H. F., C. L. Rupert, and H. S. 893. J. M. Whitney, ed. Philadelphia,Schwartz. "Evaluation of Bonding Pennsylvania: ASTM, 1986, pp. 196-Parameters on Random Fatigue Life of 206.Bonded Aluminum Joints." Journal ofAircraft. 19 (1982), p. 581. 61. Allred, R. E. "The Effect of Temper-

ature and Moisture Content on the53. Chai, H. "Bond Thickness Effect in Flexural Response of Kevlar/ Epoxy

Adhesive Joints and Its Significance for Laminates: Part II, [±45, 0/90] Fila-Model Interlaminar Fracture of Corn- ment Orientation." Environmentalposites." Composite Materials: Testing Effects on Composite Materials. G. S.and Design. ASTM STP 893. J. M. Springer, ed. Vol. 2. Lancaster, Penn-Whitney, ed. Philadelphia, Pennsyl- sylvania: Technomic Publishing Co.,vania: ASTM, 1986, pp. 209-231. 1984, p. 43.

54. Kinloch, A. J. "Review--The Science of 62. Kollek, H., W. Brockmann, and 0. D.Adhesion: Part 2, Mechanics and Hennemann. "Some Aspects of Chem-Mechanisms of Failure." Journal of istry in Adhesion of Aluminum Joints."Materials Science. 17 (1982), p. 617. Progress in Advanced Materials and Proc-

esses: Durability, Reliability, and Quality55. Christian, J. L., W. E. Witzell, and Control. G. Bartelds and R. J. Schliekel-

B. A. Stein. Environmental Effects on mann, eds. Amsterdam: Elsevier Sci-Advanced Composite Materials. ASTM ence Publishers B. V., 1985, pp. 83-94.STP 602. Philadelphia, Pennsylvania:ASTM, 1976. 63. Mall, S., and W.S. Johnson. "Character-

ization of Mode I and Mixed Mode56. Vinson, J. R. Advanced Composite Failure of Adhesive Bonds Between

Materials - Environmental Effects. Composite Adherends." CompositeASTM STP 658. Philadelphia, Pennsyl- Materials. Testing and Design. ASTMvania: ASTM, 1978. STP 893. J. M. Whitney ed. Philadel-

phia, Pennsylvania: ASTM, 1986, pp.57. Reifsnider, K. L. Damage in Composite 322-334.

Materials: Basic Mechanisms, Accumu-lation, Tolerance, and Characterization. 64. Jordan, W. M., and W. L. Bradley.ASTM STP 775. Philadelphia, Pennsyl- "Micromechanisms of Fracture invania: ASTM, 1982. ToughenedGraphite-EpoxyLaminates."

Toughened Composites. ASTM STP58. Springer, G. S. Environmental Effects on 937. N. J. Johnston, ed. Philadelphia,

i Composite Materials. Vol. 2. Lancaster, Pennsylvania: ASTM, 1987, pp. 95-114.

41I

I

65. Fisher, P., and J. J. DeLuccia. "Effects 72. Teagle, R. R. "Recent Advances inof Graphite/Epoxy Composite Mate- Mechanical Impedance Analysis Instru-rials on the Corrosion Behavior of Air- mentation for the Evaluation of Adhe-craft Alloys." Environmental Effects on sive Bonded and Composite Structures."Advanced Composite Materials. ASTM 3rd European Conference on Nondestruc-STP 602. Philadelphia, Pennsylvania: tive Testing. Florence, Italy, October 15-3 ASTM, 1976, p. 50. 18, 1984, pp. 139-162.

66. Kerr, J. R., J. F. Haskins, and B. A. 73. Zurbrick, John R. "Techniques forStein. "Program Definition and Prelim- Nondestructively Characterizing Metal-inary Results of a Long-Term Evalu- ie Substrate Surfaces Prior to Adhesiveation Program of Advanced Composites Bonding." International Advances infor Supersonic Cruise Aircraft Applica- Nondestructive Testing. 5 (1977), pp. 41-tions." Environmental Effects on Ad- 70.vanced Composite Materials. ASTMSTP 602. Philadelphia, Pennsylvania: 74. Segal, E., G. Thomas, and J. L. Rose.ASTM, 1976, p. 3. "Hope for Solving the Adhesive Bond

Nightmare." Proceedings of the 12th67. Hahn, H. T., and R. Y. Kim. "Swelling Symposium on NDE. San Antonio,

of Composite Laminates." Advanced Texas: Nondestructive Testing Informa-Composite Materials--Environmental tion Analysis Center (NTIAC), AprilEffects. ASTM STP 658. J. R. Vinson, 1979, pp. 269-281.ed. Philadelphia, Pennsylvania: ASTM,1977, p. 98. 75. Guyott, C. C. H., P. Cawley, and R. D.

Adams. "The Non-Destructive Testing68. Allen, R. C. "Some Corrosion Mecha- of Adhesively Bonded Structure: A

nisms in Attack of Resin and Resin- Review." Journal of Adhesion. 20Glass Laminates." Polymer Engineering No. 2 (1986), pp. 129-159.and Science. 19 (1979), p. 329.

76. Adams, R. D., and P. Cawley. "The69. Menges, M., and K. Lutterbeck. "Stress Mechanics of the Coin-Tap Method of

Corrosion in Fiber-Reinforced Plastics Non-Destructive Testing." Journal ofin Aqueous Media." Developments in Sound and Vibration. 122, No. 2 (1988),Reinforced Plastics-3. Chap. 4. pp. 299-316.G. Pritchard, ed. London: ElsevierApplied Science, 1984. 77. Thompson, D. 0., R. B. Thompson, and

G. A. Alers. "Nondestructive Measure-70. Sargent, J. P., and K. H. G. Ashbee. ment of Adhesive Bond Strength in

"Adhesive Joints Involving Composites Honeycomb Panels." Materials Evalu-and Laminates: Measurement of ation. April 1974, pp. 81-85.Stresses Caused by Resin Swelling."Environmental Effects on Composite 78. Meyer, F. J., and G. B. Chapman.Materials. Vol. 2. G. S. Springer, ed. "Nondestructive Testing of Bonded FRPLancaster, Pennsylvania: Technomic Assemblies in the Auto Industry."Adhe-Publishing Co., 1984, p. 403. sives Age. April 1980, pp. 21-25.

71. Norriss, T. H. "Nondestructive Testing 79. Segal, E., and J. L. Rose. "Nondestruc-of Bonded Joints, the Control of Adhe- tive Testing Techniques for Adhesivesive Bonding in the Production of Pri- Bond Joints." Research Techniques inmary Aircraft Structures." Nondestruc- NDT. IV (1980), pp. 275-316.tive Testing. December 1974, pp. 335-339. 80. Gericke, 0. R. "Determination of the

Geometry of Hidden Defects by Ultra-i sonic Pulse Analysis Testing." Journal

42I

I

I of the Acoustical Society of America. 35, 89. Williams, R. S., and P. E. Zwicke.No. 3, March 1963, pp. 364-368. "Assessment of Adhesive Properties

Using Pattern Recognition Analysis of81. Adler, L., and H. L. Whaley. "Interfer- Ultrasonic NDE Data." Materials Eval-

ence Effect in a Multifrequency Ultra- uation. 40, March 1982, pp. 312-317.sonic Pulse Echo and Its Application toFlaw Characterization." Journal of the 90. Ivakhnenko, A. G. "Polynomial TheoryAcoustical Society of America. 51 (3 of Complex Systems." IEEE Trans-Part 2) (1972), pp. 881-887. actions on Systems, Man and Cybernetics.

82. Tattersall, H. G. "The Ultrasonic SMC-, No. 4, October 1971.

Pulse-Echo Technique as Applied to 91. Rose, J. L., M. J. Avioli, and R. Bil-Adhesive Testing." Journal of Physics gram. "A Feasibility Study on the NDED. Applied Physics. 6 (1973), pp. 819- of an Adhesively Bonded Metal to832. Metal Bond: An Ultrasonic Pulse Echo

Approach." British Journal of Nonde-83. Rose, Joseph L., and Paul A. Meyer. structive Testing. March 1983, pp. 67-71.

"Ultrasonic Procedure for PredictingAdhesive Bond Strength." Materials 92. Chang, F. H., J. C. Couchman, andEvaluation. June 1973, pp. 109-114. B. G. W. Yee. "Transmission Fre-

quency Spectra of Ultrasonic Waves84. Biggiero, G., G. Canella, and S. Cicchi- Through Multi-Layer Media." Proceed-

nelli. "Ultrasonic Testing for Bond ings of the 1973 Ultrasonic Symposium.Efficiency of Carbon to Stainless Steel New York: The Institute of ElectricalAdhesive Joints." 3rd European Confer- and Electronics Engineers, Inc. (IEEE),ence on Nondestructive Testing. Flor- November 5-7, 1973, pp. 209-215.ence, Italy, October 5-18, 1984, pp. 206-223. 93. Chang, F. H., J. C. Couchman, J. R.

Bell, and D. E. Gordon. "Correlation85. Meyer, Paul A., and Joseph L. Rose. of NDE Parameters with Adhesive

"Ultrasonic Determination of Bond Bond Strength in Multi-Layered Struc-Strength Due to Surface Preparation tures." Proceedings of the 10th Sympo-Variations in Aluminum to Aluminum sium on NDE. San Antonio, Texas:Adhesive Bond System." Journal of NTIAC, April 1975.Adhesion. 8 (1976), pp. 145-153.

94. Chang, F. H., B. G. W. Yee, and J. C.86. Chernobelskaya, T., S. Kovnovich, and Couchman." Spectral Analysis Tech-

E. Harnik. "The Testing of Adhesive- nique of Ultrasonic NDT of AdvancedBonded Joints by a Very High Resolu- Composite Materials." Nondestructivetion Ultrasonic Probe." Journal of Phys- Testing. August 1974, pp. 194-198.I ~ ics D: Applied Physics. 12~ (1979). 95. Chang, Francis H., Paul L. Flynn, David

87. Alers, G., P. L. Flynn, and J. J. Buckley. E. Gordon, and Jerry R. Bell. "Princi-"Ultrasonic Techniques for Measuring pies and Application of Ultrasonic Spec-the Strength of Adhesive Bonds." Mate- troscopy in NDE of Adhesive Bonds."rials Evaluation. 5, No. 4, April 1977, IEEE Transactions on Sonics and Ultra-pp. 77-84. sonics. SU-23, No. 5, September 1976,

pp. 334-338.88. Raisch, J. W., and J. L. Rose. "Com-

puter Controlled Ultrasonic Adhesive 96. Brekhovskikh, L. M. Waves in LayeredBond Evaluation." Materials Evaluation. Media. New York: Academic Press,May 1979, pp. 55-64. 1960, p. 53.

4I 43

I 97. Flynn, P. L. "Cohesive Bond Strength 105. Budenkov, G. A., Yu. V. Volegov, V. A.Prediction." Interdisciplinary Program Pepelyaev, and V. I. Red'ko. "Inspect-for Quantitative Flaw Definitions, Semi- ing the Strength of Glued Joints withannual Report. Thousand Oaks, Califor- Ultrasonic Interference Waves." Sovietnia: Rockwell International Science Journal of Nondestructive Testing. 3,Center (Rockwell), September 4, 1976, March-April 1977, pp. 138-145.to February 1, 1977. (English Translation, January 1978.)

98. Yee, B. G. W., F. H. Chang, and P. L. 106. Staecker, P. W., and W. C. Wang.Flynn. "Signal Processing Methods for "Propagation of Elastic Waves Bound toMeasuring Cohesive Bond Strength." a Fluid Layer Between Two Solids."Interdisciplinary Programfor Quantitative Journal of the Acoustical Society ofFlaw Definitions, Semiannual Report. America. 53, No. 1 (1973), pp. 65-74.Thousand Oaks, California: Rockwell,July 18, 1975, to January 1, 1976. 107. Rokhlin, S. "Diffraction of Lamb

Waves by a Finite Crack in an Elastic99. Rose, J. L., and G. H. Thomas. "The Layer." Journal of the Acoustical Society

Fisher Linear Discriminant Function of America. 67, No. 4, April 1980, pp.for Adhesive Bond Strength Prediction." 1157-1165.British Journal of Nondestructive Testing.21, No. 3, May 1979, pp. 135-139. 108. Jones, J. P., and J. S. Whittier. "Waves

at a Flexibly Bonded Interface." Journal100. Alers, G. A., and R. K. Elsley. "Meas- of Applied Mechanics. December 1967,

I urement of Metal to Adhesive Bond pp. 905-909.Quality Using Digital Signal Analysis."1977 Ultrasonics Symposium Proceedings, 109. Claus, Richard 0., and R. A. Kline.IEEE Catalogue No. 77CH 1264- 1SU. "Adhesive Bondline Interrogation UsingNew York: IEEE, 1977. Stoneley Wave Methods." Journal of

Applied Physics. 50, No. 12, December101. Thomas, Graham H., and Joseph L. 1979, pp. 8066-8069.

Rose. "An Ultrasonic Evaluation andQuality Control Tool for Adhesive 110. Pilarski, A. "Ultrasonic EvaluationBonds." Journal ofAdhesion. 1_0f1980), of the Adhesion Degree in Layeredpp. 293-316. Joints." Material Evaluation. 43, May

1985, pp. 765-770.

102. Lloyd, E. A. "Nondestructive Testing ofBonded Joints, A Case for Testing 111. Rokhlin, S. I. "Adhesive Joint Evalu-Laminated Structures by Wide-Band ation by Ultrasonic Interface and LambUltrasound." Nondestructive Testing. Waves." Proceeding of Analytical Ultra-December 1974, pp. 331-334. sonics in Material Research and Testing.

Cleveland, Ohio: NASA-Lewis, Novem-103. Joshi, N. R. Technical Report to the US. ber 13-14, 1984, pp. 299-310.

Army Materials Technology Laboratory.Contract No. DAAL03-86-D-0001. 112. Rokhlin, S. I., M. Hefets, and M. Ro-Watertown, Massachusetts: U.S. Army sen. "An Ultrasonic Interface-WaveMaterials Technology Laboratory, Sep- Method for Predicting the Strength oftember 1988. Adhesive Bonds." Journal of AppliedU Physics. 52, No. 4, April 1981.

104. Averbukh, I. 1., and V. V. Gradinar.

"Inspecting the Strength Characteristics 113. Rokhlin, S. I., M. Hefets, and M. Ro-of Composite Materials." Soviet Journal sen. "An Elastic Interface Wave Guidedof Nondestructive Testing. 2, No. 3, by a Thin Film Between Two Solids."May-June 1973, pp. 261-264. (English Journal of Applied Physics. 1, No. 7,Translation, March 1974.) July 1980.

* 44

I

I 114. Vary, A., and K. J. Bowles. "Ultrasonic Boundary with Transverse Slip." Jour-Evaluation of the Strength of Unidirec- nal of the Acoustical Society of Americational Graphite-Polyimide Composites." 33 (1961), pp. 949-954.NASA Technical Memorandum TM-73646. Cleveland, Ohio: NASA-Lewis, 122. Pilarski, A., and J. L. Rose. "A Trans-April 1977. verse-Wave Ultrasonic Oblique-lnci-

dence Technique for Interfacial Weak-115. Vary, A., and K. J. Bowles. "Use of an ness Detection in Adhesive Bonds."

Ultrasonic-Acoustic Technique for Non- Journal of Applied Physics. 63, No. 2,destructive Evaluation of Fiber Coin- January 15, 1988.posite Strength." NASA TechnicalMemorandum TM-73813. Cleveland, 123. Mueller, R. K., R. R. Gupta, and P. N.Ohio: NASA-Lewis, February 1978, pp. Keating. "Holographic Weak-Signal185-91. Also published in ASTM Jour- Enhancement Technique." Journal ofnal of Testing and Evaluation. 2, No. 4 Applied Physics. 43, No. 2, February(1979). 1972, pp. 457-462.

I 116. Vary, A., and R. F. Lark. "Correlation 124. Dance, W. E., and D. H. Petersen.of Fiber Composite Tensile Strength "Verification of the Structural Integritywith the Ultrasonic Stress Wave Fac- of Laminated Skin-to-Spar Adhesivetor." NASA Technical Memorandum Bondlines by Neutron Radiography."TM-78846. Cleveland, Ohio: NASA- Journal of Applied Polymer Science.-Lewis, April 1978. Applied Polymer Symposium. 32 (1977),

117. Williams, J. H., and N. R. Lampert. pp. 399-410.

"Ultrasonic Evaluation of Impact Dam- 125. McLaughlin, P. V., E. V. McAssey, andaged Graphite Fiber Composite." R. C. Deitrich. "Nondestructive Exami-Material Evaluation. December 1980, nation of Fibre Composite Structurespp. 68-72. by Thermal Field Techniques." NDT

InternationaL 13 (1980), pp. 56-62d.118. dos Reis, H. L. M., L. A. Bergman,

and J. H. Bucksbee. "Adhesive Bond 126. Reynolds, W. N., and G. M. Wells.Strength Quality Assurance Using the "Video-Compatible Thermography."Acousto-Ultrasonic Technique." British British Journal of Non-destructive Testing.Journal of Nondestructive Testing. 26 (1984), pp. 40-43.November 1986, pp. 357-358.

127. Russell, S. S., and E. G. Henneke.119. Yew, Ching H. "Using Ultrasonic SH "Dynamic Effects During Vibrothermo-

Waves to Estimate the Quality of Ad- graphic NDE of Composites." NDThesive Bonds: A Preliminary Study." International 17 (1984), pp. 19-25.

I Journal of the Acoustical Society ofAmerica. 76, No. 2, August 1984, pp. 128. Cotter, J. L., and M. G. D. Hockney.525-531. "Metal Joining with Adhesives." Inter-

national Metallurgical Reviews. Review120. Rokhlin, S. I., and D. Marom. "Study of 183 (1974), pp. 103-115.

Adhesive Bonds Using Low FrequencyObliquely Incident Ultrasonic Waves." 129. Joiner, Neil. "Adhesive Bond Testing."Journal of the Acoustical Society of Nondestructive Testing-Australia. AprilAmerica. LO, No. 2, August 1986, pp. 1983, pp. 17-23.585-590.

130. Francke, W., and A. W. J. de Gee. "A121. Kuhn, G. J., and A. Lutsch. "Elastic Nondestructive Method for the Meas-

Wave Mode Conversion at a Solid-Solid urement of the Adhesive Bond Strength

II 45

I

U of Thermally Sprayed Nonfused Coat- 131. McBrearty, M., M. Negin, A. Zielensk,ing." Advances in Surface Coating Tech- and M. Weilerstein. "Realtime Nonde-nology, VoL I-Papers. Abington, Cam- structive Evaluation in Ultrasonic Wirebridge CB16AL: The Welding Insti- Bonding." 1983 Ultrasonic Symposium.tute, Abington Hall, 1978, pp. 99-109. New York: IEEE, 1983, pp. 861-865.

4I

II

IIIIIII

I

I

INDEX

Acoustic emission, 11, 12, 30-32, 35, 36Adherend, 5, 6, 8, 10, 20, 32, 34, 35, 37Adhesion, 5-8, 10, 27, 31Adhesive, 1, 2, 4-32, 34-37Adhesive types

elastomeric, 2thermoplastic, 2thermosetting, 2

ASTM standard, 9Bonding, 1, 2, 4-8, 10-13, 16, 23, 29, 30, 34, 35, 37

materials, 8, 34mechanisms, 6quality, 1, 7, 13, 34

strength, 30structures, 2

composite-to-composite, 2honeycomb, 2, 4, 5, 11, 12, 14, 25, 32, 34metal-to-composite, 2metal-to-metal, 2, 9, 24metal-to-rubber, 2

Coin tap, 12, 36Composite (see Bonding, structures)Curing, 1, 2, 4-6, 8, 13, 17, 21, 23, 28, 30, 31, 34Elastomeric (see Adhesive types)Failure mechanisms, 10, 35Holography, 11, 12, 32, 35, 36Honeycomb (see Bonding, structures)Inservice inspection, 34Joint designs, 2, 3Mechanical impedance, 12, 36Mechanical resonance, 12, 13, 36Iresnce,Metal-to-composite (see Bonding, structures)Metal-to-metal (see Bonding, structures)i Metal-to-rubber (see Bonding, structures)

Neutron (see Radiography)Nuclear magnetic resonance, 11, 33, 35, 36Prepreg, 2Quality control, 1, 34Radiography, 11, 12, 32, 35, 36

neutron, 12x-ray, 12, 32

Sonics, 11, 36coin tap (see Coin tap)mechanical impedance (see Mechanical impedance)mechanical resonance (see Mechanical resonance)

Thermography, 11, 12, 32, 33, 35, 36Thermoplastic (see Adhesive types)Thermosetting (see Adhesive types)Thickness, 4, 5, 8, 11, 16-19, 21, 22, 24, 26-31, 34-36

Ii 47

I INDEX (COMM)

I Ultrasonics, 11, 12, 18, 26, 36amplitude-domain, 14attenuation, 11, 17-19, 23, 27, 30, 31Chirp-Z, 25features, 15, 19, 21-25interface waves, 28, 36Lamb waves, 28low frequency, 25obliquely incident, 31, 35, 36pulse-echo, 15, 36Q-value, 22Rayleigh waves, 35, 36shear waves, 36

horizontally polarized, 30, 37spectral domain, 21stress-wave factor, 30surface waves, 14, 26, 27through-transmission, 11, 35, 36velocity, 18

X-ray (see Radiography)

I

I

48