ADHESION MEASUREMENT OF THINdownloads.hindawi.com/journals/apec/1976/936912.pdf ·...

Electrocomponent Science and Technology1976, Vol. 3, pp. 21-42

(C) Gordon and Breach Science Publishers Ltd., 1976Printed in Great Britain

ADHESION MEASUREMENT OF THIN FILMSK. L. MITTAL

MHVSystem Products Assurance, IBM Corporation, Poughkeepsie, N. Y. 12602, U.S.A.

(Received August 18, 1975)

A critical and comprehensive review of the various methods for the measurement of adhesion of thin films ispresented. Emphasis has been placed on the techniques used for thin films (<lm) but consideration has also beenextended to methods which have proved successful in case of relatively thicker coatings and are potentiallyapplicable in the study of thin films. Most of the methods catalogued in this review have been harnessed inmeasuring the adhesion of thin evaporated or sputtered metallic films on an assortment of substrates; the literaturereferences to the study of thin polymeric films are extremely meager.

The methods of determination of adhesion are discussed under three headings: Nucleation Methods, MechanicalMethods and Miscellaneous Techniques. Furthermore, the mechanical methods are categorized depending upon themode of application of the force to disrupt the interface. Although the emphasis is on the quantitative methods ofmeasuring adhesion, but for completeness qualitative as well as the methods which are still in a state of infancy havebeen included.

Requirements for the ideal adhesion test are outlined and, apparently, there does not exist any test whichfulfils all the virtues. Many of the techniques may not provide absolute quantitative values of adhesion strength butthese can profitably be used to follow relative changes in adhesion strength due to process variables, ageingweathering, etc.

The principle, merits, potentialities, and limitations of each technique are outlined, and the difficultiesassociated with measuring adhesion strengths and their relationship to "basic adhesion" are discussed.

INTRODUCTION

In recent years, the use of thin films (< 1/m) hasproliferated in a variety of electronic, engineering,optical, biomedical, nuclear, space, and otherapplications. This wide spread use is attributed to thefact that, quite often, materials in monolithic formare not suitable for the diverse and specialrequirements, and thin films provide the requisiteanswer. Thin films are used for many and variedpurposes: to provide resistance to abrasion, erosion,corrosion, galling, tarnish, wear, radiation damage, orhigh temperature oxidation; to reduce friction orelectrical resistance, provide lubrication, preventsticking; and also to provide special magnetic ordielectric properties.

Whatever their intended use may be, theproperties, structure, functional characteristics, andperformance all depend, inter alia, on adhesionbetween the coating and the substrate. Below arecataloged some of the prime reasons emphasizing theimportance of adhesion of thin filns.

1) Adhesion is very important in thin filmtechnology because the thin films (usually "(1 um andin some applications of the order of 50 nm) are sofragile that these must be supported by moresubstantial substrates and the degree to which the

film can share the strength of the substrate dependsupon the adhesion between the two.

2) Adhesion is very important in determining thedurability of thin film devices, for example, inmicroelectronic circuits.

3) Adhesion plays an important role in governingthe kinetics of the growth and structure of the films(formed by vacuum deposition, for example), withthe result that performance of thin films is dictatedby adhesion forces. Film structure will be aggregatedwhenever the cohesional energy exceeds the energy ofadhesion. This dependence of film integrity uponadhesion forces is not only important from theviewpoint of performance of such films but it also hasa basic scientific import.

4) The durability and longevity of thin films arelargely dependent upon their adhesion to thesubstrate since this determines the ease of removal.

5) It is highly desirable that films should becapable of being cleaned and polished, particularlywhere these are used for front surface reflectors andit means they should have good adhesion in additionto other factors like resistance to corrosion.

6) Wear is intimately related to the extent ofadhesion of thin film to the. substrates; if theadhesion is poor, the film will wear off quite rapidly.

21

22 K.L. MITTAL

7) Adhesion is a fundamental parameter in surfacechemistry and physics because it depends directly oninteratomic and intermolecular forces. Thin filmsshould provide an ideal situation because it is in suchsystems that conditions of true interfacial contactcan most nearly be attained. For example, a filmdeposited on a clean substrate under high vacuumconditions faithfully contours all surface irregularitiesand imperfections and almost perfect contactbetween the film and the substrate should be readilyobtained.

8) Good adhesion of thin films, used as protectiveovercoats, is very vital for the environmentalprotection (e.g. protection from corrosion) of thesubstrates. If the adhesion is poor, the extent ofdeterioration of the substrates by environmentalfactors (humidity, corrosive gases, etc.) is greatlyaccelerated.

9) The metallization of polymeric materialsreduces considerably the gas permeability of thepolymeric, and this reduction is strongly influencedby the adhesion between the metal and the polymer.

Henderson has discussed the importance ofadhesion in electronics.

Thin films are deposited by a variety of techniqueswhich are well described in many monographs andtechnical papers. The adhesion of thin films isaffected by substrate cleaning, procedure, the rate ofdeposition, the film thickness, the type of substrate,the substrate temperature, the purity of sourcematerial, pressure in the evaporator duringdeposition, glow discharge ambients etc. In order toquantify the effects of these variables, the necessityfor having reliable techniques for measuring adhesionis quite manifest.

So the purpose of this review is to criticallyexamine the various methods of measurement ofadhesion which have been tried successfully, or withpartial success in the study of thin films on varioussubstrates. Let it be pointed out at the outset thatmost of the adhesion measurements have been madeusing evaporated or sputtered metallic films ontoglass, metals, or oxide substrates. Only a meageramount of published work is available on theadhesion measurement of thin polymeric films. Theviscoelastic nature of polymeric films dictates thatdifficulties are inevitable in employing some of thetechniques used in the case of thin metallic films, andthe need to explore the possibilities of other potentialtechniques is obvious. Recently, there has been aresurgence of interest in the application of thinpolymeric films to various substrates, so the necessity

of good and controlled adhesion in such systemscannot be overemphasized.

2 PREVIOUS REVIEWS

Before presenting a comprehensive account of themethods of measurement of adhesion, it is in order totake due cognizance of the reviews published earlieron a similar topic. Weaver in 1958 made a survey ofthe adhesion forces and their influence on thestructure and properties of thin evaporated films andhe has reviewed the practical methods of measuringfilm adhesion and durability. It should be pointed outthat this review has been very restrictive in coverageof the methods of measurement and emphasis hasbeen on the thin evaporated films only. In asubsequent article by Weaver in 1965, again thestudy of adhesion of thin evaporated metal films wasaccentuated and only a few techniques of measuringadhesion were discussed.

Davies and Whittaker4 in 1967 made a compre-hensive survey of the methods of testing the adhesionof metal coatings (such as electroplates) to metals.This review presents a good coverage of the melhodsof measuring adhesion but is mainly confined to metalcoatings deposited primarily by electrolytic means.Some of the techniques of measuring adhesion arenot referred to or are mentioned only in passing.Furthermore, the units in which adhesion values havebeen expressed are not consistent, i.e., widelydifferent units have been used and this hinders aneasy comparison of different methods as used bydifferent investigators. More recently, Williams hasdiscussed the adhesion of thin vacuum depositedfilms (mainly metallic)and he has focussed primarilyon the factors controlling the adhesion of such films;only a peripheral account of the techniques ofmeasuring adhesion is presented.

Bullett and Prosser 6 have presented an exhaustiveexposition of the methods of measuring adhesion buttheir emphasis has been on paints or coatings whichare far thicker (25/m) than the thin films. Some ofthe methods which are viable in the field of paintsand coatings may not be suitable for thin films. Thisis an excellent review for workers in the field ofpaints and coatings, and may not be too pertinent tothin film work because some of the techniquesapplicable for thin films have been excluded.Campbell7 has presented nicely the techniquessuitable for measuring adhesion of thin films, butsome of the techniques have not been discussed inadequate detail. This review should be particularly

ADHESION MEASUREMENT OF THIN FILMS 23

consulted for nucleation methods of determiningadhesion.

More recently, Mittal8 has presented a compre-hensive account of the techniques for measuringadhesion of electrodeposited coatings; many of thesetechniques are applicable to thin films also.Chapmang, and Weaver 0 have discussed summarilysome of the techniques for measuring adhesion ofthin films, and they have pointed out the inherentdifficulties involved in this area.

After this background information, it isappropriate to enumerate the salient features of thepresent review; these are as follows:

1) An attempt gas been made to present ascomprehensive and general account as possible; the main

points of each technique are fully discussed.

2) All the fully developed, partly developed, and the

endeavors. Qualitatively, the word adhesion(unmodified) simply signifies the sticking together oftwo similar or dissimilar materials; if the materials areidentical then the term "autohesion" or "homo-hesion" is applied, and in the case of dissimilarmaterials the preferred term is "heterohesion".

The difficulties arise in prescribing a quantitativedefinition of adhesion, and in devising some suitableexperimental technique that can give numerical valuesof adhesion which can be compared with the valvescalculated assuming various models. In the presentreview, adhesion, is treated in three different forms:basic adhesion; thermodynamic or reversible adhesion;and experimental or practical adhesion. Each of thesemanifestations of adhesion are fully explicated below.

3.1 Basic Adhesion

promising methods in embryonic state have been included. Basic adhesion is related to the nature and strength ofThe merits and demerits of each are fully discussed, the binding forces between two materials in contact

3) The values of adhesion are all expressed in SIunits, i.e., in Newton per squaremeter (N m-2). Theconversion factors from one system of units toanother are given in the Appendix. Other quantitieshave also been expressed in SI units.

4) Various kinds of films and substrates have beendiscussed, i.e., the review is not restricted to anyparticular type of film or substrate.

5) Wherever possible, the thickness and otherimportant parameters of films are explicitlymentioned.

6) As the purpose of this review is to discussexclusively the techniques for measuring adhesion, thetechniques of depositing thin films or the factorsaffecting adhesion of deposited films are notdiscussed.

3 WHAT IS ADHESION?

Before expatiating upon the methods of measuringadhesion, it is imperative to comprehend clearly themeaning of the term "adhesion" and the variouse,xperimental quantities measured as representationsof adhesion.

The term "adhesion" is fraught with semanticdifficulties and it has been defined in a variety ofways. ’ 2 The diversity in the definitions ofadhesion stems from the fact that adhesionphenonmena are encountered in many fields resultingin specific nomenclature and terminology in specific

with each other. These forces could be either primaryvalence type (i.e., ionic, co-valent, co-ordinate,metallic), pseudo primary valence type (i.e., hydrogenbonding), or secondary valence type (i.e., Debye,Keesom and London dispersion forces knowncollectively as Van der Waals forces). But this basicdefinition of adhesion is not very helpful as it is notpossible either to calculate the magnitude or tomeasure such adhesion forces in practical systems. Anelegant approach based upon the molecular modelswas taken by Taylor and Rutzler 3, but it has notbeen extended to a stage where this could be used toquantify adhesion in a system of interest. Relativemagnithde of various forces are summarized byGood,14 as shown in Table 1.

3.2 Thermodynamic or Reversible Adhesion

This particular form of adhesion is defined as thereversible work done in creating a unit area of theinterface between two substances. It is defined asfollows"

WA B "l’A + "[B "l’A Bwhere WA B is the reversible work of adhesion, ’a thespecific surface free energy of substance A, ’B thespecific surface free energy of substance B and ’A B isthe interfacial specific free energy. This definition issimply based upon the change in free energy of thesystem before and after the contact is made betweenA and B. The above definition is not useful in case ofthin films as the various parameters are not known,so WAB cannot be calculated. In cases where

24 K.L. MITTAL

TABLEApproximate ranges of binding energies for varioustypes of interactions. (Reprinted from reference 14,p. 15, by courtesy of Marcel Dekker Inc.)

Energy,kcal/moleType of force

Chemical bondsIonicCovalenta

MetallicIntermolecular force

Hydrogen bondsDipole-dipole

DispersionDipole-induced-dipole

aThe "bottom cutoff" for covalent bonding isarbitrary.

bin principle, there is no minimum energy ofinteraction for intermolecular forces. Minimum valuesare sometimes reported in the literature, but these areusually minimum observed values. Thus it is ratherdifficult to measure an energy of interaction, at anytemperature T, that is very much smaller thanRT NkT, because the average thermal energy permolecule or per degree of freedom is of the order ofkT.

CThe smallest energy of interaction that has beenreported is that between two helium atoms,approximately 0.02 kcal/mole.

140 to 25015 to 17027 to 83

up to12 kcalb

up to 5 kcalup tol0 kcalc

up to 500 cal

both A and B are liquids or one is solid and the otherliquid, the above expression is very commonly used inone form or another to quantify the work ofadhesion.15,16

3.3 Experimental or Practical Adhesion

More common terms for this are "bond strength" or"adhesion strength". Experimentally, adhesion can bemeasured in two ways: (a) in terms of the tbrces,defining the force of adhesion as the maximum forceper unit area exerted when two materials areseparated, and (b) alternatively, in terms of work orenergy, defining the work of adhesion as the workdone in separating or detaching two materials fromone another. Forces of adhesion and work ofadhesion can be related only if assumptions are madeabout the changes in force with distance of separationso that an integration can be performed. In otherwords, W, the work of adhesion is

= ff(x)dxIf the break occurs at the interface, then it is termedadhesive failure, and if it occurs between A or B thenit is cohesive failure. Where no definable interface

exists due to interdiffusion, the measured values willrepresent the weakest plane in the system. It is,therefore, particularly important to locate theposition of separation. In cases where there is no clearcut adhesive failure, the terms "bond strength" or"adhesion strength" should not be used but,unfortunately, there does not exist a proper term torepresent nonadhesive failure; in the field ofadhesives, the term "joint strength" is used torepresent joint failure irrespective of the site offailure.

Assuming there is a true adhesive failure, the mostimportant question is: What is the relationshipbetween the bond strength and basic adhesion?This can be described by the expression,

Bond Strength f (basic adhesion, otherextraneous factors)

These extraneous factors depend upon internalstresses and the technique used for measuring bondstrength. For a given system, the bond strength valuesdiffer when measured by different techniques;furthermore, some of the techniques are operator-dependent, thus compounding the problem further.

After this brief orientation, it is important topoint out that the experimental values of adhesionmay not have direct relevance to the basic adhesion asdefined earlier. For example, if the adhesion of apolymeric film to metal is measured as the workrequired to detach it, then the experimental work ofadhesion includes the "basic adhesion" plus the workspent in other processes, such as the inelasticdeformation of the polymer. In other words, the"basic adhesion" can only be estimated fromexperimental adhesion provided due allowance hasbeen made for other processes.

4 METHODS OF ADHESION MEASUREMENT

Before embarking on the techniques per se, twopoints should be made clear:

a) The practical adhesion may not be a directmeasure of the basic adhesion as defined above.

b) The values of experimental adhesion obtainedby different methods may not be directly compar-able.

The methods of measurement of adhesion of thinfilms can be categorized in a number of ways:

1) Qualitative and quantitative methods.2) Destructive and non-destructive methods.3) Mechanical and non-mechanical methods.


4) Fully developed, partly developed and themethods in inchoate stage.

5) Practical methods and the methods ofacademic interest.

6) Routine and non-routine methods.However, in this review, the methods have been

classified into three categories: (a) nucleationmethods, (b) mechanical methods, and (c) miscellaneoustechniques. Mechanical methods are further divideddepending upon the mode of application of forces todetach the film from the substrate. Miscellaneousmethods comprise the following: Thermal method,X-ray technique, ESR, capacitance method, cathodictreatment, and pulsed laser or electron-beamtechnique.

4.1 Nucleation Methods

These methods depend upon the observation of thekinetics of formation of thin films and these can alsobe labelled as non-destructive methods.

Campbell in a recent review has given a detailedexposition of these methods and this obviates acomprehensive account here; only a capsuledescription is presented below.

In general parlance, adhesion is thought of as amechanical property of the film. However, on anatomic scale, the removal of the film consists ofbreaking the bonds between the individual atoms ofthe film and the substrate so that macroscopicadhesion can be considered as the summation of theindividual atomic forces. In principle, therefore, itshould be possible to relate the adsorption energyf ofa single atom on the substrate, Ea, to the totaladhesion of the film.

It is not difficult to see from above that theadhesion determined by nucleation methodscorresponds to the basic definition of adhesion asdiscussed earlier. Here, only the adsorption forces areconsidered, so the summation of individualadsorption energy of adatoms should be free fromother processes which beset the mechanical methods(Sections 4.2, 4.3). It might be in order to mentionthat nucleation methods are based upon themeasurement of (a) nucleation rate (b) island density(c) critical condensation and (d) residence time of thedepositing atoms comprising the film.

Superficially, the nucleation methods seem simpleand very useful in determining quantitatively

-The adsorption energy of an adatom or single atomadsorbed on a substrate is the work required to reversiblyremove the atom to infinity.

the adhesion of thin films but a deeper lookwill reveal that these methods are fairly compli-cated and are of limited applicability. It isnecessary to have an access to an electron microscopein order to count island densities and this in turnmeans that it must be possible to remove the islandsfrom the substrates without moving them relative toone another. Even if this can be done, it may notalways be possible to make enough observations toseparate the cohesion energy term, Eb, from theadsorption term Ea. There are not many instances inthe literature where the adhesion values determinedby nucleation methods have been compared with thevalues obtained using mechanical techniques. Thisarises partly from the difficulty of making meaningfulquantitative measurements by mechanical methodscoupled with the limited applicability of nucleationmethods. Benjamin and Weaver calculated theadhesion energies of films of aluminum, silver andcadmium on glass substrates, and also of aluminumand silver on single crystal cleavage surfaces ofsodium chloride and potassium chloride. Theyconcluded that the adhesion energy is due to physicaladsorption and can be explained in terms of Van derWaals forces alone. More recently, Chapman 2 hascompared the mechanical stripping measurementswith nucleation results for gold on glass. He foundthat with the lowest stripping speeds (the value ofadhesion depends upon the speed of stripping) used(50 nm/sec), the average adhesion energy was doublethat obtained by nucleation methods, but hementions that it is necessary to work at even lowerstripping speeds to obtain better agreement.A few comments regarding nucleation methods are

in order: (a) These are not the tests for measuringadhesion; the information from this type of work hasled to a better understanding of thin film formationand structures (b) The techniques are not applicableto completed films, (c) Unlike mechanical tests,nucleation studies yield the adsorption energy of asingle adatom, and the validity of using suchadsorption energy to calculate adhesion energy perunit area is questionable on the grounds that the bondenergy in a continuous film will not in general be thesame as that for a single atom.

4.2 Mechanical Methods

All mechanical methods use some means of removingthe film from the substrate. These methods can bebroadly classified into two categories depending uponthe mode of detachment of the film: (a) methodsinvolving detachment normal to the interface, and

26 K.L. MITTAL

(b) methods involving the lateral detachment of thefilm from the interface.

Methods involving detachment normal tothe interface

A force of increasing magnitude is applied to thefilm-substrate interface until the interface isdisrupted. Either the maximum force applied ismeasured, or some other criterion such as area ofdetachment is used to designate adhesion. In thesemethods, an attempt is made to detach the film in adirection normal to the interface, so that the wholearea of the film is detached simultaneously when theinterface is broken. If the whole area is not detachedsimultaneously, then the analysis of the forces ismore complex. The following methods are consideredunder this heading.

4.2.1.1 Direct pull-off (DPO) method Thebasic principle of the method is to attach somekind of pulling device (such as brass pins) tothe back of the film by means of soldering or someadhesives .and then pull the film in the perpendiculardirection using a tensile tester machine (for example,Instron). Belser and Hicklin 19 measured the adhesionof evaporated and sputtered metal films on glass,quartz and glazed-tile surfaces. They soldered headedbrass pins to the metal films using "Cerroseal"(50-50 tin-indium, m.p. 118C) or tin-lead solderand the force to pull the film, covered by the pinhead, perpendicularly from the substrate surface wasmeasured by a spring balance and winch system. Thearea of the pinhead was about 0.05 cm. Similarly,Belser2 o cemented the ends of 0.952cm diameteraluminum cylinders with 0.3175 cm diameter axialholes to films and pulled them from the substrates.Frederick and Ludema21 measured the adhesion ofvapor deposited aluminum films (25 nm thick) onfresh soda-glass by direct tensile method. Afterdeposition of the film, a clean gold sphere of 0.3 cmdiameter was pressed against the aluminum at roomtemperature and then pulled to separate it from glass.Beno 2 has devised a tensile test similar to that ofBelser: o using solid cylinders. Chiang and Ing:3

measured the adhesion of evaporated amorphousselenium films on aluminum coated with oxide layer.They cemented brass blocks to the film and substratesides using epoxy adhesive and then pulled theassembly apart in a commercial tensile tester. Collinset al. 24 measured the adhesion of evaporatedaluminum films (40 to 50 nm) on glass. Contact to thefilm was made by bonding a metal disc to the film

using cold-setting araldite. The force to remove thefilm from the glass was applied either through a springbalance or a Hounsfield Extensometer. RecentlyKuwahara et al., 2s measured the adhesion ofevaporated aluminum films on glass plates by directlypulling off the film from the substrate. Hordon andWright26 have described a vacuum apparatus to measureinterfacial adhesion based on the rupture of the interfaceby applying tensile force.

The direct pull-off method suffers from thefollowing difficulties: (a) simple tensile tests aredifficult to perform, and most of the time theseinvolve a complex mixture of tensile and shear forces,which renders the interpretation of the results moredifficult, (b) alignment must be perfect to insureuniform loading across the interface, (c) such tests arelimited by the strengths of available adhesives orsolders, (d) there is always a possibility of (i) adhesiveor solvent penetrating and affecting the film-substrateinterface (ii) stresses produced during setting ofcement or adhesive (iii) non-uniform stressdistributions or stress concentrations over the contactarea during the pulling process. All of these factorsaffect the fmal adhesion strength measured.

More recently, some of the alignment problemshave been overcome by Jacobsson and Kruse27 bycementing identical cylinders to the film and to thecorresponding uncoated area on the reverse side ofthe substrate, followed by application of a tensileforce normal to the film along the collinear axis ofthe cylinders, as shown schematically in Figure 1.Even with this arrangement, some misalignment couldoccur and they have accounted for the effect of thisand the effect of variation of adhesive thickness.Results are presented for films of zinc sulfide,cryolite, and silver deposited on BK7 glass substrates.The forces of adhesion on ion-bombarded substratesin these cases were found to be 4.3 x 107,5.4 x 107, and 2.3 x 107 Pa respectively.

Apropos, Cramer et al,28 have described a similararrangement for adhesion measurement ofelectodeposits and the effect of the misalignmentof rods or cylinders was derived from the basicformulas for the flexural behavior of beams.

4.2.1.2 Moment or topple method A slightvariation of the direct pull-off method has beeninvestigated in which instead of applying theforce vertically to the interface, it is applied in ahorizontal direction to a rod glued to the film, andthe moment of the force required to break the filmfrom the substrate is a measure of adhesion. Butler 9

has described a simple film adhesion comparator


P2 CEMENT

THIN FILM

SUBSTRATE

CEMENTP1 t

FIGURE Diagram for adhesion test by application of anormal pulling force. The force F is applied at P1 and P2 withthe force aligned through the center C of the substrate.Identical steel cylinders are cemented on the coated topside and uncoated lower side of the substrate. (Fromreference 27)

based upon this principle. Butler points out that thismethod has advantages over the direct pull-offmethod because (a) this arrangement offers lesssubstrate distortion since there would be no resultantoverall force normal to the plane of the substrate and(b) this apparatus would not require such criticalalignment as in the case of normal pull.

In this arrangement, as shown in Figure 2, one endof the rod is cut to form two short legs and theresults are calculated assuming pure tension underone leg and compression below the other leg. Butlerobserved that tensile stresses in excess of 4 x 107 Pacould be obtained for copper films deposited on glasssubstrate. The adhesive used for glueing was Eastman910. Subsequently, the technique has been used byButler et al, 3 o and Stoddart et aL 31 Wieckowski32

has analysed the methods for checking adhesion ofconductive thin films and has concluded that themethod of detachment using a side force seems to bethe best. The objections regarding the use ofadhesives or solders also apply to this modification asdiscussed for the direct pull-off method.

METAL FILM

FIGURE 2 Diagram showing arrangement for the momentor topple method. (From reference 29, Copyright: Instituteof Physics, England)

4.2.1.3 Ultracentrifugal method The direct pull-off method suffers from the drawback that some sortof adhesive or solder is required to attach some deviceto pull the film, which is sometimes highlyundesirable. In the ultracentrifugal technique noadhesive or solder is used, and adhesion is determinedby measuring the force necessary to detach a filmnormal to the surface. As the name suggests, thespecimen is in the form of a rotor which is spun atextremely high speeds to provide the requisitecentrifugal force as shown schematically in Figure 3.At a critical speed of rotation, the coating can nolonger withstand the centrifugal stress and isdetached. Hallworth appears to have been thepioneer of this technique which he described as"whirl test". But the real progress in the design of theequipment has been through the efforts of Beams etal.3 4 who suspended the rotor in a vacuum in a

SUPPORT SOLENOID(BOTTOM HALF)

DRIVE

VACUUM CHAMBER,

DAMPER

TO VACUUM PUMP

IC CORES

EELYSUSPENDEDROTOR

PICKUP COIL

FIGURE 3 Diagram of ultracentrifugal arrangement formeasuring adhesion. (From Handbook of Thin FilmTechnology, L. I. Maissel and R. Clang, Editors). Copyright(1970) McGraw-Hill Book Company. Used with permissionof McGraw-Hill Book Company.

magnetic field and applied a rotating magnetic fieldto spin the rotor. More recently, Dancy3s hasdesigned a modern version of this system with angularspeeds in excess of 80,000 rps. In the past, the mainconcern has been that the rotor should be ferro-magnetic and this limited the study of coatings toonly a few substrates. But with the latest design ofDancy, coatings on nonmagnetic substrates may also beinvestigated with the magnetic support system. The

28 K.L. MITTAL

nonmagnetic specimen must be rendered magneticand this can be accomplished by the inclusion of aferromagnetic rod that is aligned along the axis ofrotation, or by the addition of a ferromagnetic discwhich is cemented to one end of a test rotor. Thepossible stress distribution within the rotor must nowbe calculated considering the influence of this addedbody. In all cases, the maximum speeder of rotation isgoverned by the mechanical properties of the rotor(or the weakest material in a composite rotor) and forany given material can be increased only by reducingthe overall diameter of the rotor. For example, withmild steel the speed would be limited to 30,000 rpsfor 0.470 cm diameter rotor, increasing to 45,000 rpsfor 0.3125 cm diameter. Table II summarizes thebursting speeds and other parameters for sphericalsteel rotors of different diameters.

TABLE II

Bursting-speeds of spherical steel rotors (from reference 34a)

Rotor PeripheralRotor speed speed CentrifugalDiam. (10 (103 acceleration(mm) rev/min) cm/sec) (106 g)

3.97 4,420 96 47.12.38 7,410 92.5 721.59 12,660 105 1430.795 23,160 96.5 2400.521 37,980 104 4280.398 48,000 100 515

In the ultracentrifugal system, the forces on thecoating are given by"

Ar 4rt2n2r2d ( t t 2 )T 1+-+t g r rSr

where T is the hoop stressA is the adhesion in g/cm2

r the radius of the rotor in cmn is the number of revolutions per secondd density of the coating in g/cm3

t thickness of the coating in cmand g is the acceleration due to gravity

( 981 cm/sec2 ).The hoop stress, T, can be eliminated by making

slits in the coating parallel to the axis of rotation or

I-According to Guinness Book of World Records thehighest man-made rotatory speed ever achieved is1,500,000 rps on a steel rotor with a diameter of about0.025 cm, installed in 1961 in Professor Beams’ laboratory atthe University of Virginia, Charlottesville.

in case of hard coatings like chromium, it isautomatically eliminated as shown by Dancy andZavarella. a7 Under the conditions ofT 0, the above equation reduces to: A4r2n2rdt/g. From this equation, it is clear that therotor size will be dictated by the degree of adhesionto be measured. For example, for n 30,000 rps, themaximum measurable adhesion using a 0.470 cmdiameter rotor with 0.0125 cm coating having adensity of 8.3 g/cm3 would be 9.14 x l0 g/cm2 or9.1 x 107 Pa. If the adhesion strength exceeds9.1 x 107 Pa then the above combination of variousdimensions is not satisfactory unless the speed isincreased. If the speed and the rotor radius have to bemaintained, then the increse in either the thickness orthe density of coating will increase the value of A.

Beams et ala4 obtained acceleration of the rotorof 109 times normal gravity, acceleration beinglimited by the bursting strength of the rotor. However,their results deal only with the poorly adherent filmsand the surface was deliberately contaminated beforedeposition of silver films to reduce adhesion. Beams34reports the work of Dancy and Kuhlthau in which theystudied the adhesion of electrodeposited chromium films(0.0254 cm in thickness) on cylindrical steel rotorsfrom 0.3175 to 0.9525 cm in diameter and obtainedan effective adhesion of 1.13 x 108 Pa. If the samemethod were applied to an evaporated chromiumfilm of about 50 nm thickness, and assuming thesame density of the film, then a centrifugalacceleration of 3 x 101 o times normal gravity wouldbe required, which is considerably more than the valueof 109 g obtained by Beams et al.

Other workers have also used the ultracentrifugal tech-nique to measure the adhesion of electrodeposited copperand nickel a8 chromiuma9 as well as organic films40-4 5

In summary, this technique is quite versatile if therotor materials are selected properly, then a variety ofcoatings can be studied. Particularly, with the adventof the solid-state ultracentrifuge, the technique offersincreased promiseas The following comments are inorder: (i) no adhesive or solder is required, (ii) thechoice of the rotor material is important as the extentof adhesion values which can be measured will bedictated by the rotor material, (iii) in polymericcoatings, the viscoelastic properties play an importantrole, 0v) in case of very thin deposits, it may bedifficult to obtain the requisite centrifugal forceas the total force is the product of mass andacceleration, and if mass is small, acceleration mustbe increased and this is limited by the rotormaterial, (v) no systematic study of the effect ofthickness, or the rate of acceleration has been


conducted, and (vi) if the temperature of the rotorrises above the glass transition temperature ofthe polymeric film, then the situation is quitecomplicated. In addition, Huntsberger44 observedthat the values of adhesion obtained by thistechnique are lower than those obtained by othermethods, and this could be explained if the filmexhibited creep leading to a peeling phenomenon,where the load to start detaching the film would befar less than the normal force to throw it off.Reference 44 should be consulted for criticisms ofthis technique. Before leaving the discussion of thistechnique, it should be emphasized that systematicwork needs to be done on the adhesion measurementof thin films, polymeric or metallic, to varioussubstrates using this technique.

4.2.1.3 Ultrasonic method Before discussing thismethod, it may be proper to point out clearly thatthe ultrasonic method has been applied to study theadhesion of relatively thicker paint films; thereferences where thin films have been studied areextremely scanty. An early original paperdescribing this method in the study of adhesion isthat of Moses and Witt4 6. In this method, thenormal force is supplied by the inertia of the coatingsubjected to rapid reversals of motion at ultrasonicfrequencies. They studied the adhesion of polysty-rene and vinyl resin films on metal cylinders. Themethod and apparatus utilize an electrodynamicsystem for producing longitudinal ultrasonicvibrations in a metal cylinder. A film attached to thefree end of the vibrating cylinder separates from themetal when the force due to the acceleration exceedsthe adhesion force at the interface. The acceleratingforce is determined by the frequency and amplitudeof vibrations and by the mass and area of thefilm. The amplitude of vibrations of the rod orcylinder is calculated from the input voltage. Themaximum acceleration is given by 47r2fo a, beingthe fundamental frequency of vibrat.on given by(I/2)v/E[p, 1’ being Young’s modulus, the rod lengthand 0 the density. By knowing the dimensions anddensity of the film one can determine the force ateach reversal. It was calculated that at a frequency of20kHz, the maximum acceleration would be of theorder of 2 x 10 and at 50 kHz, 5 x 10 timesnormal gravity. The mechanical restrictions on thesystem limited the choice of cylinder material tomagnesium or aluminum alloys, and to increase theversatility of the method screw caps were used at theends of the cylinder. These screw caps could be othermetals.

Moses4 7 measured the adhesion of polystyrenefilms deposited from toluene solution onto duralcylinder at a frequency of 23.6 kHz, and the adhesionwas of the order of 4.14 x l0 Pa. The thickness ofthe film was not clearly mentioned.

Although the thickness of the film was notmentioned, it is very clear that he used polymericfilms thicker than gm, thereby he could use arelatively low frequency to separate the thicker filmfrom aluminum metal. As pointed out earlier anacceleration of the order of 5 x 10 times normalgravity is obtained at a frequency of 50 kHz; higheraccelerations may be obtained by working at higherfrequencies which means using piezo-electrictransducers and for fundamental operations of aquartz vibrator the limiting frequency is usually ofthe order of 12 MHz. Using a frequency of 10 MHzcan raise the acceleration to 10 9 times normalgravity, which is of the same order as that obtainedby Beams34 in the ultracentrifugal technique. Morerecently Faure et al,48 fixed the film support to theends of an amplifier of mechanical vibrations, thevibration being induced by a piezoelectric transducer.The films used were granular and fissured silver films.Furthermore, assuming that the frequency of themechanical vibration is much smaller than the naturalfrequency of the silver grains, the order of magnitudeof the adhesion between the grains and the supporthas been estimated. In addition, they have pointedout that the adhesion of continuous films can also bemeasured by this method. On the basis of the abovediscussion, it is apparent that more work using thinfilms is necessary before its usefulness can beestablished. The effective force can be increased byraising the mass of the film (essentially increasing thethickness), but it is a well known fact that theadhesion of vacuum deposited films is affected by thestresses in thick films and cases are known where verythick layers peel spontaneously. So measurements onthick films may give misleading results when theinformation is extrapolated to thin films. However,this is a potential method which can be evolved into aquantitatve technique for measuring adhesion.

Methods based upon the application oflateral stresses for detachment

The primary purpose here is to apply lateral stressesso as to cause the detachment of the film from thesubstrate. The methods most widely used by painttechnologists to apply such lateral stresses are bend,cupping, or impact tests, or by attacking the paint filmby plough or knife tests, or in some instances both

30 K.L. MITTAL

together as in most scratch test variants. Many ofthese tests may lead to cohesive cracking of the filmbut careful inspection often reveals that thecoarseness of crack pattern is directly related toinadequacy of adhesion. Moreover, a film mightappear merely to have cracked but will often peelspontaneously from the cracks within a short time oftesting, indicating that adhesion failure had in factbegun over a substantial area. Before discussingmethods under this heading and which are applicableto thin films, it might be stressed that some of thetechniques might be very useful in the area of paintfilms but might not be valid when studying adhesionof thin films. Only those techniques which haveshown promise for the adhesion of thin films arediscussed below.

4.2.2.1 Scotch tape method This is a relativelyold technique for testing adhesion and is usuallyattributed to Strong49 A pressure sensitive tape ispressed onto the film and then rapidly stripped.Three possibilities arise: (a) the film is completelyremoved from the substrate (b) film is not at allremoved, and (c) the film is partly removed orremoved in patches. So it is apparent that this testis highly qualitative and can be used to screen casesinvolving very poor adhesion from those whereadhesion is appreciable. Strongs himself used thistest to determine the adhesion of evaporatedaluminum films on glass and observed a pronouncedeffect of cleaning of the glass on the degree of adhesionWilliams and Backus employed this test to studythe effects of variables on the adhesion of evaporatedthin metallic films on glass. Thin films of approx-imately 2 nm thickness were deposited on cleanmicroscopic-slide glass and their relative adherencetested by their resistance to removal by means ofstripping with a film of collodion. This way theycould categorize the metal films into various classesdepending upon their ease of removal. Recently Haq,Behrndt and Kobin 2 have used this test to study theadhesion between glass or oxidized silicon substratesand evaporated gold films in combination with anadherent intermediate layer of Ta, Si, Ge, or Cr. Thisway they were able to study the changes in adhesionas a function of relative thickness of Au andunderlayer films as well as of time, environment andtreatment after deposition. As this test is onlyqualitative, they could distinguish complete lifting,partial lifting, and no lifting of the films as shown inFigure 4. It is interesting to notice that they foundthat the adhesion of Au films to glass substrate was a

function of the thickness of the film; while very thinfilms (<10 nm) adhered strongly, complete liftingwas observed for thicknesses above about 17 nm, withpartial lifting for the intermediate thickness. Mattoxand McDonaldS used this method to study theadhesion of sputtered cadmium films on iron andthey could investigate qualitatively the effects of thetechniques of depositing films on the substrate.

Although this test is highly qualitative, it offerscertain advantages: (a) it is a very inexpensive andquick test, so it can be utilized for screening cases of

Au Ge 0.85

7 Au: Ge 1.25

-I!, x x

10 11 12 13

TIME (DAYS)

FIGURE 4 Change of adhesion with time for differentAu-Ge film combinations. (From reference 52).

very poor adhesion from the ones showingappreciable adhesion (b) this test can prove veryvaluable in an exploratory research where effect ofvarious operating variables is studied (c) it can proveuseful in studying the effects of humidity, temper-ature, and other factors on the adhesion between thefilm and the substrate. A few of its inherent weakpoints can be enumerated as: (a) highly qualitative sono numerical values of adhesion can be obtained andthis renders it incomparable to other techniques(b) type of tape, application pressure, and the mannerof stripping affect the final results and this makes itdifficult to compare results obtained by differentworkers, (c) applicable only when the adhesion betweenthe film and the substrate is less than that between thetape and the film; so, it cannot be used in cases wherethe films adhere very tenaciously to the substrate.Beno22 reports that such tests are limited to adhesionof about 5.17 x 10 Pa.


4.2.2.2 Peel test The film can be peeled from thesubstrate in two ways: (a) by directly holding ontothe film, and (b) by applying some sort of backingmaterial to the film and then holding onto thebacking. The actual peel test involves peeling aspecified width of the film, as shown in Figure 5, at aspecified angle, 90 being the most common. In thistest, it is virtually impossible to specify the areainvolved at any instant, so the force used in peelingthereforce has little significance. The results areexpressed as energy or work done per unit area. Sothe results from peeling experiment are not directlycomparable to the results obtained using othertechniques which provide adhesion values in terms offorce per unit area. Furthermore, in order to makeany useful measurement, the film must be completelyremoved from the substrate, which limits theapplicability of this technique to those interfaceswhich exhibit relatively poor adhesion.

Sundhal 4 has studied the adhesion of Ta2 N-TiPd Au films on high purity A12 03 substrates by 90peel tests performed on leads which were thermo-compression bonded to the films. From his findinghe concluded that a necessary requirement for goodadhesion of these films to alumina substrates was thepresence of a minimum amount of Ca and/or Si at thesurface. He selected the 90 peel test because thisseemed to simulate the conditions in the fabricationof reliable thin film circuits. The thickness of thefilms was "1.5/am. Bonds which failed at the interfaceat a force of less than 2.5 kg were consideredsymptomatic of poor adhesion.

Chapman18 has employed peeling technique in thestudy of adhesion of gold films on soda lime. Heterms it the mechanical stripping method which is aquantitative version of the qualitative Scotch tapetest. He has used both supported and unsupportedfilms, consequently he used two different arrangements:(1) In this case a tape is attached to the backof the film. The tape is then attached via a thintrapeze to the winding string. As the strippingproceeds, the tape and the film are removed from thesubstrate. (2) The tape is attached only to the firstpart of the film and then the tape and the film arescribed out, with a razor blade as shown schematic-ally in Figure 6. Stripping then proceeds from thetape-film combination to the unsupported film and itis on the unsupported film only that the measure-ments are made.

Chapman obtained energies of the order of2000 erg/cm2 in the case where backing material wasused; in the case of thicker gold films, which obviatesthe need for backing material, the measured peel

LOAD CELL

NTCoHRRONOU

SUBSTRATE/

FIGURE 5 Diagram of apparatus for stripping or peelingexperiments. (From Handbook of Thin Film Technology,L. I. Maissel and R. Glang, Editors. Copyright: 1970McGraw-Hill Book Company. Used with permission ofMcGraw-Hill Book Company.

SUBSTRATE

FILM

TAPE

TAPE AND FILM CUTTO SHAPE SHOWN

FIGURE 6. Diagram for peeling experiments on unsupportedfilms. (From reference 18.)

32 K.L. MITTAL

energies fell to about 500 erg/cm2 Moreover, thepeel values depend upon the rate of peel, and theresidual gas pressure. All these observations signifythat the measured peel energy consists of twocomponents: (a) true adhesion energy, i.e., the energyof adhesion due to intermolecular interactions, and(b) the energy spent in the inelastic deformation ofthe film material. So unless the work expended in thedeformation of film material is known and accountedfor, one cannot determine the energy of adhesion. Inthe case of polymeric films, in addition to the rate ofpeel, angle of peel, width of peel, the temperature ofpeel are also important in dictating the final measuredpeel energies. Chen and Flavin have discussed themechanics of film peel adhesion and the effects ofnon-elastic behavior of the film are analysed. Poleyand Whitaker 6 have recently employed the dynamicpeel test in understanding the effects of cleaningprocedures, substrates, vacuum treatments, and glowdischarge ambients on the chromium-glass adhesion.

4.2.2.3 Tangential shear or lap shear methodThis technique has not been as popular as the directpull-off method, but in certain cases lap shear testsimulates better the actual application conditions ofthin films, so the results using this test may be moremeaningful. Very recently, Lin57 has measured theadhesion of Au, Cu, and A1 films (of thicknesses from0.05 to 40/.tm) to glass and MgO single crystalsubstrates using this test. For films deposited mainlyat a vacuum of 10 -7 to 10 -8 Torr, the shear stresswas found to be virtually independent of the filmthickness and had mean values of 3 x 105 Pa for Au,8 x 105 Pa for Cu and 1.5 x 106 Pa for A1 films onglass. The measured shear stress is the tangential forceper unit area required to break the bond between thefilm and substrate. The experimental arrangement isshown schematically in Figure 7.A piece of glass microscope slide is bonded to the

upper surface of the film which has been depositedon the desired substrate. The film and the substrateare then clamped rigidly in position and a shear forceapplied parallel to the film by weights attached to athin stainless steel wire passing over low frictionpulley. A normal load of 0.110 to 0.2100 kg is alsopresent to counteract any pivoting tendency aboutthe edge of the test piece. The shear stress values arecalculated from the force required to break the bondbetween the film and the substrate divided by thearea of contact of fast-set adhesive. Lin has pointedout that the most satisfactory and convenient toapply adhesive was the cyanoacrylate monomer(Eastman Kodak 910). Furthermore, he did not

observe any penetration of this adhesive through thefilms. Lin has compiled an extensive table summ-arizing the shear stress values of different films andsubstrates measured for a variety of film depositvariables. A few comments regarding this techniqueare in order: (1) An adhesive is used in this test as inthe direct pull-off or moment method and thus thistechnique suffers from the same weaknesses and

G LASS

NORiALLOAD

SUBSTRATE

FIGURE 7 The lap shear test arrangement for measuringadhesion. The arrow shows the direction of the shear force.(From reference 57, Copyright: The Institute of Physics,England.)

criticisms as enumerated earlier in the discussion ofdirect pull-off method. The undesirable effectsof adhesives are aggravated in the study ofpolymeric films and although Lin did not findany diffusion of the adhesive through themetallic films, this may not be true in polymericfilms, (2) According to Lin, this simple lapshear test has these advantages: (a) it avoidssevere deformation of the substrates, (b) thefilm is gripped over a relatively large area, thus thestress is less concentrated, and (c) it approximates to anominally "pure shear" measurement even thoughstereoscan and optical microscopy observationindicated that peeling is involved after initiation ofshear in some regions. (3) It is important to mentionthat Lin observed that the shear stress values wereabout 100 times smaller than those obtained inmeasurements where the force is applied normal tothe substrate and this could be explained by showingthat the films are more easily removed by a forceparallel to the substrate than by a perpendicular force(the theoretically predicted ratio between the shearand tensile stresses is 1:4).

Dini et al. 58, s 9 have described a ring shear test tomeasure quantitatively the adhesion of electro-deposits, the thickness of these deposits being muchmore than that of thin films. In this method,essentially, a cylindrical rod is coated with separate


rings of electrodeposits of predetermined width. Thisrod is forced through a hardened steel die having ahole larger in diameter than the rod but less than thatof the rod plus the coating. The coating is detachedand the adhesion strength A, determined by theformula, A W/rdt,, where d is the diameter of therod, t is the width of the deposit, and W is the forcerequired for detaching.

Although this technique was not used in the studyof adhesion of thin films, but there appears to be nofundamental objection or difficulty; so it shouldcertainly be tried for the case of thin films. Theremay be some difficulty in getting dies machined toprecise diameters, but it does not appear insurmount-able.

4.2.2.4 Scratch or stylus or scribe method In therecent past this test has become a controversial issueand arguments have been extended both for defenceand attack. Essentially, in this test, shown schematic-ally in Figure 8 a smoothly rounded chrome steel,tungsten carbide, or diamond tip ("0.003 to 0.005 cmradius) is drawn across the film surface, and a verticalload, applied to the point, is gradually increased untilthe film is completely removed resulting in a clearchannel. The critical load at which the clear track isformed is taken as a measure of adhesion. Heavens6 o

and Heavens and Collins61 were the first to use thistest to study quantitatively the adhesion of thinevaporated metallic films on glass and the effect ofchromium interlayers. Weaver and Hill62 studied theincrease in adhesion of aluminum films to glass by thepredeposition of films of chromium. They also usedthe load necessary to strip the film completely as ameasure of adhesion.

The test remained semi-quantitative and nonumerical values of adhesion could be obtained until1960 when Benjamin and Weaver63 analysed thescratch test in detail and derived some simpleequations to convert the critical load into numericalvalues for adhesion. A capsule description of theiranalysis and findings is given below. A detailedanalysis showed that the action of the point alwaysinvolved plastic deformation of the substrate and thisdeformation produces a shearing force at the filmsubstrate interface around the rim of the indentationproduced by the point, and a simple relationshipbetween the applied load and the sheafing force wasdeveloped, so that adhesion could be calculated as ashearing force.

Fx/r2 a_, where a

AUXILIARY WEIGHT

F’- PIVOT/ f";’x ./- COIL ATTACHED

r/////Z’//////////;/J’///J’////////2BEAM-MOUNTED STYLUSACTING ON SPECIMEN ONCOORDINATE TABLE

FIGURE 8 Schematic layout of scratch or stylus test. (Fromreference 70, Copyright: The Institute of Physics, England.)

Where W is the critical load, r is the radius of the tip,and F is the shearing force per unit area due to thedeformation of the surface, a is the radius of thecircle of contact, and P is the indentation hardness ofthe substrate material.

Based upon their studies of adhesion of evaporatedsilver films on various substrates, and of variousevaporated metal films on glass, Benjamin and Weaverfound that (a)in each case, the measured loadbecame constant for film thicknesses exceeding acertain value; this value was close to 80 nm. Forthinner films a slightly smaller load was required andthey surmised that this could be attributable to filmstructure and it was no innate weakness of themethod (b) the load did not depend directly uponthe mechanical properties hardness, elasticity ortensile strength of the film as the load did not varywith the thickness of the film. (c) the load dependedupon the nature of both the film and the substrate,without being directly attributable to the mechanicalproperties of either. From these observations theyconcluded that the load is determined essentially bythe properties of the interface.

The scratch test has since been used for manyyears to show ageing effects in adhesion,6-67 and toinvestigate alloying effects due to diffusion in twolayer metal films.6 a

Weaver and Parkinson68 offered the mostconvincing evidence that the critical load wasdetermined primarily by the adhesion between thefilm and the substrate from their experiments withtwo layer metal film on glass. Gold was depositedon glass which was overlaid with aluminum. Asgold has a relatively poor adhesion to glass, so themeasurements started at low value. Aluminum has amuch higher adhesion to glass and they observed anabrupt increase in adhesion at exactly the stage wherealuminum, in the form of intermetallic with gold,

34 K.L. M ITTAL

< 6o

58nm 105nm 125nm 135nm

t-’-’-l’ -’-0 9 18 27 36 45 54TIME

FIGURE 9 Adhesion changes during annealing of Au-A1thin-film diffusion couples with gold underlayers. Differentthicknesses of gold underlayers are shown in the diagram.(From reference 68.)

appears at the metal-glass interface, as shown inFigure 9. Furthermore, an exact parabolic relation-ship between the gold thickness and the time asmeasured up to the break in the adhesion curve wasfound, which is in accordance with the diffusiontheory. The scratch test enjoyed a substantial degreeof popularity but recently it has been criticized by

3069various groups.Recently Butler, Stoddart and Stuart o employed

the scratch test to study the adhesion of vacuumdeposited In, Sn, Pb, Au, Cu, A1, Ni, Cr and Mo (ofvarious thicknesses up to 3.2/am) on glass usingdiamond and steel styli with tip radii of approx-imately 25/.tm and loadings of up to 0.230 kg. Usingscanning electron and optical interference microscopythey showed that the process of scratch formationwas very complex and varied with the film material,indicating that it was not possible to deduce absolutevalues of adhesion using a simple general theoreticalmodel.

Some of their comments are summarized below:

whether failure occurs first in the film or in thesubstrate or at the interface.

4) Film detachment often occurs at lighter loadsthan required for track clearance which appears todepend on film tearing, film pile-up in front of thestylus, dust, imperfections, etc.

Figure 10 is an electron micrograph of the scratchproduced which shows film becoming detached andforming a small hillock or bubble ahead of the stylusand some detachment and raising of the film oneither side of the actual track. 70 Very recently,Weaver 0 has analysed the behavior depicted inFigure 10, and his theoretical analysis, apparently,offers explanation for all the observable details in themicrograph. Moreover, it modifies and expands thedetails of how the film is removed, but the initial stateis still dependent upon the shearing force under thestylus. Also Weaver has explained why styli ofcertain materials give consistent values of criticalload, while others are not too useful.

1) Their work has shown the unreliability ofsimple optical methods of detecting re-adherencecaused by the sliding stylus. A film can becomedetached before the formation of a cleared track andconversely a film may be thinned to opticaltranslucency without being removed.

2) The form of the track depends on the elasticand plastic deformation of the film and substrate, i.e.,it depends primarily on their hardnesses.

3) There is no preferential failure (i.e., yielding ofa permanent nature) at the film substrate interface:factors such as the size and surface finish of the stylusas well as the thickness of the film can determine

FIGURE 10. Track due to a moving stylus which hascaused film (copper) detachment ahead of the stylus. Theedges of the track contain extruded, raised and folded film.Film thickness 0.24 #m; stylus load 40 g; angle of view 21to film plane. (From reference 70, Copyright: The Instituteof Physics, England.)


Scratch test has been employed by many workersin studying the adhesion of an assortment offilms.S3, 71-75 It is important to point out thatChopra69 found that the critical load (in the caseof polycrystalline Au films deposited on glass)increased nearly linearly with film thickness above200 nm, which suggests that the critcial load isdetermined both by the adhesion at the interface and themechanical strength of the film. Hamersky72, in hisstudies of adhesion of aluminum films deposited onglass slides, concluded that the the accuracy of thescratch test result is dependent on the point radius andonly indirectly on its hardness. The metlaod givesresults which are reproducible using a point radiuslarger than 0.2 mm. Berendsohn7 studied adhesionof a variety of films on different substrates andTable III summarizes selected results from his work.

So far the discussion of the scratch test hascentered around the measurement of critical load, butOroshnik and Croll,73 during their detailed studies ofthe scratch test, found that complete removal of afilm (e.g. aluminum) by a stylus was an infrequentevent. At stylus loadings below (and often above) theso-called "critical load" there was only partialremoval of the film, but not necessarily in relativeproportion to the loading. This led them to developthe concept of "Threshold Adhesion Failure" as anoperational criterion for adhesion measurements. It isdefined as follows:

Threshold Adhesion Failure occurs if, within theboundaries of a scratch and over its cm path,removal of the film from its substrate can bedetected by transmitted light with a microscope(40x magnification)at even one spot no matterhow small.

Measurements embracing this new criterion are madeusing the "up-and-down" or "staircase" method. Thestylus is incrementally loaded for successive scratchesin the load range where threshold failure is expected.Measurements were made largely on 0.5 /am thickvacuum evaporated aluminum films on fused quartzsubstrates using diamond styli with nominal curvatureradii of 44/am. Some of their findings are: (1) Thescratch test can discriminate between stylus loads 0.1g apart, the resolution of the apparatus; (2) MeanThreshold Adhesion Failure loads are relativemeasurements; (3) No tw o styli of any one material(diamond or tungsten carbide) of the same nominaltip radius of curvature yield the same ThresholdAdhesion Failure load values; (4) Any one stylusexhibits its own individual scribing and testingcharacteristics; (5) Each material (e.g. gold as

compared with aluminum) shows a different scratchtest behavior, thereby requiring suitably modifiedmeasurement techniques. The findings of Oroshnikand Croll certainly necessitate further probing intothe complexities and detailed mechanics of thescratch test.

Although there may be some criticism apropos ofconversion of critical loads into quantitative adhesionvalues, yet the technique can be profitably used to(i) Follow the effect of deposition parameters andother experimental variables, and thereby determinethe optimum adhesion parameters. (ii) Detectcoatings of marginal adhesion. (iii) Test an area toosmall for other tests. (iv) Study the effects of aging,weathering, point-to-point variation, i.e., to check thetmiformity of substrate-film adhesion. A wide rangeof adhesion strengths (3 x 107 Pa to 3 x 109 Pa) havebeen measured, which are reproducible to + 5%. Thelower limit can easily be extended in appropriate casesby using a point with a greater tip radius, but theupper limit is imposed by the strength of thesubstrates which tend to crack or splinter at higherloads.

Two recent improvements in the scratch techniqueshould be noted. So for most of the work has beendone on glass substrates because of the difficultiesin observing clear channel on opaque substrates.Greene et al76 have developed a technique which isnot limited by the optical transmission of thesubstrate, and thus extends the utility of the scratchtest to all kinds of substrates. Secondly, in order toobtain the critical load, one has to make many scratchesusing increasing loads, but Ahn77 has indicated thatall the requisite information can be derived by simplymaking one pass with programmed load.

4.2.2.5 Blister method In this test, a fluid- gas78

or liquid79 is injected beneath the coating at thecoating-substrate interface and the hydrostaticpressure is increased until the coating begins todetach (peel away) from the interface. The beginningof the peel is usually indicated by discontinuity onthe pressure-volume plot. Most of the work using thistechnique has been done on relatively thick (25/am)paint films. Dannenberg78 concluded that thetechnique was valid to surface coatings other thanorganic paints. For a recent theoretical analysis ofblister test, the reader should refer to Williams.8 o Noreference could be found in the literature apper-taining to the application of blister method to thinfilms. As the work required to detach metalliccoatings will be considerably more than that for the

36 K.L. MITTAL

X X X X X X X X X X X X X X X X X X

zz-.z.z...


polymeric coatings, only less adherent metallic filmsmay be amenable to this test. Furthermore, the valuesof adhesion strength thus determined will beinfluenced by the thickness, ductility, or brittlenessof the film.

4.3 Abrasion Method

The abrasion resistance of films has received muchattention and the literature can be traced back to1930, but in such studies the main objective was todetermine the durability of the films, not adhesionper se. The abrasion resistance was determined by

81rubbing the surface with emery loaded rubber.However, a careful consideration will reveal that theabrasion resistance not only depends on the hardnessof the film but it is influenced by the adhesion of thefilm to the substrate also.

This method has not gained much popularityand so the references to the use of this techniqueare scarce. Schossberger and Franson8 2 developed anabrasion-resistance test to study the adhesion ofvacuum deposited aluminum films on glass surfaces.They used a stream of fine silicon carbide particlesdropped from a known height to abrade the film andthe removal of the film was monitored by measuringthe electrical resistance of the film. By using thismethod they could detect differences in adhesioncaused by the plating and annealing conditions.Poorest adhesion was found for platings made onunclean surfaces. The optimum substrate temperaturefor good adhesion was determined to be 17 5 C.

The abrasion method, undoubtedly, involvesburnishing as well as stripping action and theseadditional processes can certainly affect the values ofadhesion obtained. In summary, this methodcertainly offers certain advantages but it has notfound wide use as a practical method of obtainingquantitative adhesion values.

4.4 Miscellaneous Techniques

Under this heading are discussed those techniqueswhich, at least at the present time, cannot be labelledas practical tests for measuring adhesion. Some testsare in a state of infancy but offer great potential,whereas others are of academic interest or are tooelaborate. However, for completeness of this review,it is necessary to include these techniques also.

4.4.1 Thermal method The principle of themethod is as follows: when a film is chemicallydissolved from its substrate, the energy liberated isequal to the heat of solution of the film minus the

energy of adhesion between the film and thesubstrate. A microcalorimeters is used to measuresmall heat changes. Apparently, only Chapman84 hasemployed this technique in the study of adhesion ofvacuum deposited metallic films on NaC1 substrate;and no other reference to the actual use of thistechnique could be found in the literature. RecentlyChapman9 has indicated that there are experimentaldifficulties to be overcome in order to render thistechnique successful. As the liberated heat is verysmall, so the microcalorimeter in its present state maynot offer a viable technique. The reader is referred tothe original reference.

4.4.2 X-ray method This is one of thenon-destructive methods of studying adhesionat the film-substrate interface. Chopra69

points out that the techniclue of X-raytopography can be employed to obtain qualitativeinformation on the adhesion of epitaxial filmsdeposited on single-crystal substrates (Si, forexample). This is based upon the observation that thestrains at the film substrate interface as well as thepoor adhesion of the film modify the diffractioncontrast of the substrate and thus provide qualitativeinformation on strain and adhesion.

This method is applicable only in certain situationsand cannot be generalized. Furthermore, nonumerical values of adhesion of even epitaxial filmscan be obtained. However, it can be utilized as acomparative method for the adhesion of epitaxialfilms.

4.4.3 Electron spin resonance method Thedetails of this technique are not available because ofthe lack of published material on this topic.Campbell7 reports on the basis of his privatecommunication with Smith that this is one of thepotential techniques for obtaining comparativemeasurements of adhesion. It appears that like theX-ray method, this might be of limited applicability.

4.4.4 Capacitance method The details of thistechnique are not available to this author but themethod is incorporated in this review with theassumption that one person’s handicap may beanother’s standby. This method has been patented byNekrasov85. The details of this technique are takenfrom Bullett and Prosser6 In this method, an elasticelectrode is placed on the coated metal panel and theabsolute capacitance of the electrode so formed ismeasured at very low and very high frequencies. Theratio of the differences in capacitance to the

38 K.L. MITTAL

capacitance at high frequency is claimed to give aco-efficient characteristic of the adhesion of the film.This co-efficient is an inverse measure of the adhesionof the film.

It is obvious that with the scanty details, it is notsafe to appraise this method with respect to itsapplicability in determinimg quantitatively theadhesion of thin films.

4. 4.5 Cathodic treatment method Davies andWhittaker4 have referred to this relatively oldtechnique. In a sense, it is similar to the blistermethod (see above)with the difference that herehydrogen is used as the fluid to cause blistering. Themethod is based upon the following: The coated partis made the cathode in an electrochemical cell and thehydrogen evolved diffuses through the coating tocollect at the interface and causes blistering. It isobvious that (1) this method has limited applicabilityas the substrate should be metal (2) it is qualitative(3) specific interaction between the film material andhydrogen evolved will affect the results. This methodhas not gained much attention as the originalreferences date back to 1930’s.

4.4.6 Pulsed laser or electron-beam methodRecently, a very interesting technique for themeasurement of adhesion of thin films has beendescribed by Anderson and Goodmana 6. Thetechnique consists basically of generating acompressive pressure wave in the solid object ofinterest. The wave is next transferred by reflectioninto a tensile wave which then stresses the interfaceto be tested. The authors point out that suchtechniques have been used in the past for thedynamic tensile testing of bonds between two solidsbut their application to thin film adhesion was notmentioned. The actual description of the technique isas follows: A pulsed laser or a pulsed electron beammachine is used to deposit energy necessary togenerate the compressive stress waves needed for theactual test. The first step is to prepare a substrate of aconvenient size and then coat the desired area. Thelaser or electron beam machine is then used tobombard the remaining uncoated surface. Thecompressive wave which is generated propagatesthrough the sample and is reflected from the coatedsurface which is left free. When the compressive wavereaches the free surface, it is reflected and invertedinto a tensile wave of approximately the sameamplitude. The amplitude of this resulting tensilewave can be altered by changing the initial energyinput.

The authors point out that the determination ofthe actual adhesion strength involves considerabledifficulties and they have not elaborated upon suchproblems. Furthermore, they have discussed the meritsand demerits of this technique as regards the measure-ment of adhesion of thin films. The main disadvant-age is that the technique is experimentally exactingand could prove prohibitively expensive under certaincircumstances. In other words, the technique cannotbe used generally. In summary, this method certainlyoffers some promise but considerable efforts must bespent to collect some numerical values of adhesion ofthin films before it can be classified as a practicalmethod for measuring adhesion.

5 DISCUSSION AND CONCLUSIONS

Ideally speaking, one would like to measurequantitatively the basic adhesion, i.e., adhesion dueto intermolecular interactions between the film andthe substrate. But as pointed out earlier, theexperimentally measured adhesion strength consistsof basic adhesion plus contribution from extraneoussources. These extraneous sources include theinternal stresses in the film which always decreasethe inherent adhesion strength, and the defects orextraneous processes introduced by the measuringprocedure which may decrease the basic adhesionby introducing unmeasured stresses or may increasethe basic adhesion by reducing the effects of internalstresses.This can be expressed as

Practical adhesion basic adhesion (loss ofadhesion due to internal stress)+ (effects of defects orextraneous processesintroduced by the testprocedure).

Furthermore, the nature and extent of theinterfering processes differ from technique totechnique, with the result that the practical adhesionvalues obtained by different techniques may not bedirectly comparable. In addition, the practicaladhesion is measured either in terms of the force orthe work required to disrupt unit area of theinterface, and unless variation of the force withdistance is known precisely, the values of practicaladhesion expressed in the two different units cannotbe directly relatable.

An ideal adhesion test should be (a) quantitative,(b) reproducible, (c) not very time consuming,(d) easily adaptable to routine testing, (e) relatively


simple to perform, (f) nondestructive, (g) indepen-dent of the film thickness, (h) independent of theoperator’s experience, (i) applicable to all combin-ations of film materials and substrates, (j) valid over awide range of sample sizes, (k) applicable to productsand processes, (1)independent of the manner ofperforming the test. Furthermore, no specialized testequipment should be necessary; but such idealizationis not realized in practice as there is no techniquewhich fulfills even half the above attributes. Howeverthe situation is not so hopeless, as many of thetechniques have shown good promise, if certainconditions are fulfilled; furthermore, the lack of anideal test should be an impetus for devising bettertechniques.

As is generally true, there are protagonists ofeach technique, but the following conclusions caneasily be elicited from this review.

1) If no numerical values of adhesion are requiredthen almost any of the techniques can be used tofollow the changes in adhesion due to surfacetreatments, deposition variables, weathering,corrosive environment, ageing, etc.

2) The nucleation methods are not really the testsfor measuring adhesion; the information derived fromthese has led to a better understanding of filmformation and structure. These methods have limitedapplicability, are not valid for completed films, andrequire quite elaborate experimental set-up.

3) In the techniques where some sort of adhesiveis used to glue the pulling device to pull the film, thefollowing important points should be kept in mind:(1) The cohesive strength of the adhesive should bemore than the adhesion between the film and theadhesive, and between the film and the substrate;(2) the adhesion between the adhesive and the filmshould be more than that between the film and thesubstrate; (3) the adhesive used should not alter theproperties of the film-substrate interface; this mightbe less of a danger in metallic films, but thedifficulties are aggravated when studying polymericfilms.

In consideration of these undesirable aspects ofgluing, ultracentrifugal and ultrasonic techniquesoffer certain advantages. With the new design ofDancy3s where both ferro-magnetic and non-magnetic substrates can be studied, the ultra-centrifugal technique becomes very promising. It isgenerally believed that in this technique, a purelynormal stress is applied to the film, but the situationis not so simple as commented by Huntsberger44 Inthe case of thin films, a high acceleration is required

to provide the requisite ultracentrifugal stress, whichmeans the choice of the rotor material and sizebecomes very critical, as the maximum accelerationwhich can be obtained is dictated by the burstingstrength of the rotor. The new solid-state designshould facilitate the use of this technique, provideddue acceleration can be obtained. The ultrasonictechnique certainly offers some desirable features andusing piezoelectric transducers, one should be able tostudy adhesion of thin films. With suitablemodifications, any substrate or film material can bestudied.

4) Scotch tape test is too qualitative, smalldifferences in adhesion are not discernible, and theinterpretation of the results is quite subjective. Thepeel test is simple, requires modest equipment, butthe analysis of the forces involved is quite complic-ated, and the angle of peel, rate of peel, width of thetest strip, and the viscoelastic properties of the filmare all very important. Furthermore, peel valuesexpressed in force/length cannot be directlycompared with the tensile or shear values which arerecorded in terms of force/area.

Scratch or stylus test certainly offers manyadvantages as it is quick, reproducible, easy toperform, and it has been used quite profitably toinvestigate the effects of deposition variables, ageing,etc. However, the conversion of critical loads intoquantitative values of adhesion may not be asstraigthforward as discussed by Benjamin andWeaver63. Moreover, the findings of Oroshnik andCroll7 that a complete removal of the film at the socalled "critical load" is an infrequent event certainlyshed a different light on the significance of criticalload which has been used so frequently as a measureof adhesion. The scratch test is a very good techniquebut more investigation is needed to understand thedetailed mechanics of scratch formation and thevariables which affect the final results.

There are not many references to the tangentialshear test but Lin 7 has used it very recently and hasclaimed superiority of this method over the directpull-off technique. More adhesion studies using thismethod are certainly needed.

Blister method has been used in case of relativelythicker polymeric coatings, but it should be applicableto thin polymeric films as well. The technique maynot be suitable for the adhesion studies of metallicfilms. Abrasion test has not been popular in recentyears, but the earlier studies certainly engenderconfidence in this method.

5) The tests covered under "Miscellaneous Tech-

40 K.L. MITTAL

niques" are too narrow in scope and only sporadic refer-ences to their use could be found in the literature. Thefollowing points should be noted: (i) some of theseare only qualitative, (ii)in some cases, specializedequipment is necessary, and (iii) some may simplydetect voids or gaps which are manifestation of pooradhesion strength. The thermal method has beendescribed and used by Chapman only84 and lately9he has expressed the experimental difficultiesinvolved in measuring very small heat changes withthe existing microcalorimeters.

In the end, it should be added that in addition toimproving upon the existing techniques or devisingbetter means of measuring adhesion, more work isneeded in these two areas: (i) Cross comparison ofvarious techniques. Various substrate-film combin-ations should be tested for adhesion by differenttechniques. (ii) Adhesion measurement of thinpolymeric films. The time dependent and viscoelasticbehavior of polymeric films may invalidate orcomplicate the usefulness of certain techniques whichare quite valid in the case of metallic films. So adetailed inquiry into the effects of these and othercharacteristics of polymeric films on their adhesionstrength should be conducted.

REFERENCES

1. A.W. Henderson, Adhesion in electronics, in Aspects ofAdhesion, Vol. 5, D. J. Alner, Editor, pp. 86-104, CRCPress, Cleveland, Ohio, 1969.

2. C. Weaver, Adhesion of thin films, Proc. of theFirst International Conference on Vacuum Techniques,Namur, 2, 734-736 (1958), Pergamon Press, London,England, (1960).

3. C. Weaver, The adhesion of metal films to surfaces,Chem. Ind., 370-373, February 27, 1965.

4. D. Davies and J. A. Whittaker, Methods of testing theadhesion of metal coatings to metals, MetallurgicalReviews 12, 15-26 (1967).

5. D. G. Williams, The adhesion of vacuum-deposited films,A Technical Report from Dow Chemical U.S.A., RockyFiats Division, P.O. Box 888, Golden, Colorado. UnderU.S. Atomic Energy Commission Contract AT (29-1)-1106, RFP-1690, July 21, 1972, 17 pages.

6. T. R. Bullett and J. L. Prosser, The measurement ofadhesion, Prog. Org. Coatings 1, 45-73 (1972).

7. D. S. Campbell, Mechanical properties of thin films, inHandbood of Thin Film Technology, L. I. Maissel andR. Glang, Editors, Chapter 12, McGraw Hill BookCompany, N. Y., 1970.

8. K. L. Mittal, A critical appraisal of the methods formeasuring adhesion of electrodeposited coatings, inProperties ofElectrodeposits: Their Measurement andSignificance, R. Sard, H. Leidheiser, Jr., and F. Ogburn,Editors, Chap. 17, pp. 273-306, The Electro-chemicalSociety, Princeton, N.J. (1975).

9. B. N. Chapman, Thin-film adhesion, J. Vac. Sci. Tech. 11106-113 (1974).

10. C. Weaver, Adhesion of thin films, J. Vac. ScL Tech. 12,18-25 (1975).

11. G. Salomon, in Adhesion and Adhesives, Vol. 1, p. 3,R. Houwink and G. Salomon, Editors, ElsevierPublishing Co., New York, 1965.

12. R. J. Good, On the definition of adhesion, J. Adhesion,in press.

13. D. Taylor, Jr. and J. E. Rutzler, Jr., Adhesion usingmolecular models, Ind. Eng. Chem. 50, 928-934 (1958).

14. R. J Good lntermolecular and interatomic forces, inTreatise on Adhesion and Adhesives, Vol. 1, Chap. 2pp. 9-68, R. L. Patrick, Editor, Marcel Dekker, Inc., NewYork, 1967.

15. W. A. Zisman, Relation of the equilibrium contact angleto liquid and solid constitution, in Advances inChemistry Series No. 43, pp. 1-51, American ChemicalSociety, Washington, D.C., 1964.

16. K. L. Mittal, Surface chemical criteria for maximumadhesion and their testing against the publishedexperimental adhesive strength values, in AdhesionScience and Technology, vol. 9A, pp. 129-168, L. H. Lee,Editor, Plenum Press, New York, 1975.

17. P. Benjamin and C. Weaver, Condensation energies formetals on glass and other substrates, Proc. Roy. SocA252, 418-430 (1959).

18. B. N. Chapman, Adhesion of thin metallic films, inAspects ofAdhesion Vol. 6, pp. 4 3-54, D. J. Alner,Editor, CRC Press, Cleveland, Ohio, 1971

19. R. B. Belser and W. H. Hicklin, Simple rapid sputteringapparatus, Rev. Sci. lnstrum. 27, 293-296 (1956).

20. R. B. Belser, Aging study of metal plating on quartzcrystals, Report No. AD48261, Defense DocumentCenter, Wash., D.C., 1954.

21. J. R. Frederick and K. C. Ludema, .Adhesion ofvapor-deposited aluminum to lime-soda glass, Z Appl.Phys. 35,256-257 (1964).

22. R. L. Beno, An adhesion tester for evaporated metalfilms, U.S. Atomic Energy Commission Contract AT(29-1) -1106, R.F.P. -672, January 11, 1966.

23. Y. Chiang and S. W. Ing, Interface and adhesion studieson evaporated selenium on oxide surfaces, J. Vac. ScLTech. 6, 809-814 (1969).

24. L. E. Collins, J. G. Perkins and P. T. Stroud, Effect ofion bombardment on the adhesion of aluminum films onglass, Thin Solid Films.4, 41-45 (1969).

25. K. Kuwahara, T. Nakagawa and K. Kuramasu, Effect ofion-pump evacuation on the adhesion of evaporated thinfilms. Japanese Inst. Metals Trans. 12,218-219 (1971).

26. M. J. Hardon and M. A. Wright, High vacuum adhesiontest apparatus, Rev. Sci. lnstrum. 40, 1017-1021(1969).

27. R. Jacobsson and B. Kruse, Measurement of adhesion ofthin evaporated films on glass substrates by means ofdirect pull off method, Thin Solid Films 15, 71-77(1973).

28. S. D. Cramer, et al., Measuring electrodeposit adhesion,Metal Finishing, 45-50 (March 1970).

29. D.W. Butler, A simple film adhesion comparator, J.Phys. (E) Sci. lnstrum. 8,979-980 (1970).

30. D.W. Butler, C. T. H. Stoddart and P. R. Stuart, Somefactors affecting the adhesion of thin metal films, inAspects ofAdhesion Vol. 6 pp. 53-63, D. J. Alner,Editor, CRC Press, Cleveland, Ohio, 1971.


31. C. T. H. Stoddart, D. R. Clarke and C. J. Robbie, Thinfilm adhesion: Effect of glow discharge on substrate, J.Adhesion 2, 270-278 (1970).

32. K. Wieckowski, Conductive thin films, Pr. Nauk. Inst.Technol. Electron Politekh Wroclaw No. 7,7-16(1972).

33. F. D. Hallworth, Bond testing of silver-plated enginebearings, Automotive and Aviation Ind. 95(2),30- (1946)

34. a) J.W. Beams, Production and use of high centrifugalfields, Science 120, 619-625 (1954).

b) J.W. Beams, J. Breazeale and W. L. Bart, Mechanicalstrength of thin films of metal, Phys. Rev. 100,1657-1661 (1955).

c) J.W. Beams, Study of adhesion by the high-speedrotor technique, Tech. Proc. Amer. ElectroplatersSoc. 43,211-214 (1956).

35. W. H. Dancy, Jr., A solid state ultracentrifuge foradhesion testing of electrodeposits, Technical Report RE701-138, April 1970, 58 pp. AD 752 460, availableNTIS.

36. 19 75 Guiness Book of World Records. Norris McWhirterand Ross McWhirter, Editors, p. 185, Bantam Books,New York, 1975.

37. W. H. Dancy, Jr. and A. Zavarella, Adhesion of chromiumand other deposits to steel, Plating 52, 1009-1016(1965).

38 V.M. Ponizovskii, Yu. G. Svetlov and G. V. Chirkov,Determination of adhesion of copper and nickel coatingson a steel base by means of large centrifugal fields,Zashch Metal 3(4), 515-517 (1967); Chemical Abstracts68, 42493X (1967).

39. V. M. Ponizovskii and G. P. Spelkov, Determination ofthe strength of the adherence of chromium electroplateon steel by a high centrifugal field method, Zavod Lab.40(1), 107-109 (1974); Chemical Abstracts 80, 127398t(1974).

40 A.M. Malloy, W. Soller and A. G. Roberts, Evaluationof adhesion of organic coatings by ultracentrifugal andother methods -Part 1, Paint Oil & Chem. Rev., pp.14-19, August 27,1953.

41. H. Alter and W. Soller, Ultracentrifugal measurement ofthe adhesion of epoxy polymer, Ind. Eng. Chem. 50,922-927 (1958).

42. R. L. Patrick, W. A. Vaughan and C. M. Doede, Theultracentrifuge as an Instrument for testing adhesion,J. Polymer Sci. 28, 11-16 (1958).

43. J. R. Huntsberger, Influence of selective adsorption onadhesion, J. Polymer Sci. 43,581-582 (1960).

44. J. R. Huntsberger, Ultracentrifugal measurement ofadhesion: some limitations, Official Digest 33,635(1961).

45. V. M. Ponizovskii, et al., Determination of the adhesionof coatings to steel.., centrifugal force, LakokrasochMater. Ikh Primen No. 5, 52-54 (1972); ChemicalAbstracts 78, 85996n (1973).

46. S. Moses and R. K. Witt, Evaluation of adhesion byultrasonic vibrations, Ind. Eng. Chem. 41, 2334-2338(1949).

47. S. Moses, The nature of adhesion, Ind. Eng. Chem. 41,2338-2342 (1949).

48. R. Faure, A. Carlan, J. Crebassa, G. Desvousseaux and B.Robrieux, Modification de la structure des couchesminus d’argent soumiser a des vibrations mechaniquesmeasure de l’adhesion, Thin Solid Films 9,329-339(t972).

49. J. Strong, Evaporated aluminum films for astronomicalmirrors, Publ. A.S.P. 46, 18 (1934).

50. J. Strong, On the cleaning of surfaces, Rev. ScL lnstrum.6, 97-98 (1935).

51. R. C. Willians and R. C. Backus, The electron-micrographic structure of shadow-cast films and surfaces,J. Appl. Phys. 20, 98,106 (1949).

5 2. K. E. Haq, K. H. Behrndt and I. Kobin, Adhesionmechanism of gold-underlayer film combinations to oxidesubstrates, Z Vac. Sci. Tech. 6, 148-152 (1969).

53. D. M. Mattox and J. E. McDonald, Interface formationduring thin film deposition, J. Appl. Phys. 34,2493-2494 (1963).

54. R.C. Sundhal, Relationship between substrate surfacechemistry and adhesion of thin films, J. Vac. Sci. Tech.9, 181-185 (1972).

55. W. T. Chen and T. F. Flavin, Mechanics of film adhesion:elastic and elastic-plastic behavior, 1BM Journal ofR&D.16, No. 3,203-213 (1972).

56. N. M. Poley and H. L. Whitaker, Adhesion of chromiumfilms to soda lime glass, J. Vac. ScL Tech. 11, 114-118(1974).

5 7. D. S. Lin, The adhesion of metal films to glass andmagnesium oxide in tangential shear, J. Physics (D):Appl. Phys. 4, 1977-1990 (1971).

5 8. J.W. Dini, J. R. Helms and H. R. Johnson, Ring sheartest for quantitatively measuring adhesion of metaldeposits, Electroplating Metal Finish. 25, 5-11 (1972).

59. J.W. Dini and H. R Johnson, Ring shear adhesion testsvarious deposit substrate combinations, Metal Finish.pp. 44-48 (Aug. 1974).

60. O.S. Heavens, Some factors influencing the adhesion offilms produced by vacuum evaporation, J. Phys. Rad-ium 11,355-360 (1959).

61. O.S. Heavens and L. E. Collins, L’epitaxie dans leslames polycrystallines, J. Phys. Radium 13,658-660(1952).

62. C. Weaver and R. M. Hill, Adhesi6n of evaporatedaluminum films, Phil. Mag. 3, 1402-1410 (1958).

63. P. Belajamin and C. Weaver, Measurement of adhesionof thin films, Proc. Roy. Soc. 254A, 163-176 (1960).

64. P. Benjamin and C. Weaver, Adhesion of metal films toglass, Proc. Roy. Soc. 254A, 177-183 (1960)

65. P. Benjamin and C.Weaver, The adhesion of evaporatedmetal film on glass, Proc. Roy. Soc. 261A, 516-531(1961).

66. P. Benjamin and C. Weaver, The adhesion of metals tocrystal faces, Proc. Roy. Soc. 274A, 267-273 (1963).

67. C. Weaver, Adhesion of high energy surfaces, inAdhesion Fundamentals and Practice, pp. 46-57,Maclaren & Sons Ltd., London, 1969.

68. C. Weaver and D. J. Parkinson, Diffusion in gold-aluminum, Phil. Mag. 22, 377-389 (1970).

69. K. L. Chopra, Thin Film Phenomena, pp. 313-323,McGraw-Hill BoGk Company, N.Y., 1969.

70. D. W. Butler, C. T. H. Stoddart and P. R. Stuart, Thestylus or scratch method for thin film adhesionmeasurement: some observations and comments,J. Physics (D): Appl. Phys. 3,877-883 (1970).

71. M. M. Karnowsky and W. B. Estill, Scratch test formeasuring adherence of thin films to oxide substrates,Rev. Sci. Instrum. 35, 1324-1326 (1964).

72. J. Hamersky, Feststellung optima!.er Parameter beiAdhasionsmessungen unter Anwendung der Ritz-methode, Thin Solid Films 3,263-267 (1969).

42 K.L. MITTAL

73. J. Oroshnik and W. K. Croll, Thin film adhesion testingby the scratch method, paper presented at the SurfaceScience Symposium of the New Mexico Section of theAmerican Vacuum Society, held in Albuquerque, N.M.,April 22, 1970.

74. A. T. English and P.A. Turner, Stability of conductormetallization in corrosive environments, J. ElectronicMaterials 1, 1-14 (1972).

75. O. Berendsohn, Quantitative adhesion tests of vacuumdeposited thin films, J. Testing Evaluation 1, 139-143(1973).

76. J. E. Greene, J. Woodhouse and M. Pestes, A techniquefor detecting critical loads in the scratch test for thinfilm adhesion, Rev. Sci. Instrum. 45,747-749 (1974).

77. J. Ahn, personal communication, 1974.78. E. Hoffman and O. Geogoussis, Measurement of

adhesion of paint films, J. Oil. Color. Chem. Assoc. 42,267-269 (1959).

79. H. Dannenberg, Measurement of adhesion by a blistermethod, J. Appl. Polymer Sci. 5,125-134 (1961).

80. M. L. Williams, The continuum interpretation forfracture and adhesion, J. Appl. Polymer Sci. 13, 29-(1969).

81. L. Holland, Vacuum Deposition of Thin Films,pp. 88-103, Chapman & Hall Ltd., London, 1956.

82. F. Schossberger and K. D. Franson, Adhesion ofevaporated metal films, Vacuum 9, 28-35 (1959).

83. E. Calvet and H. Prat, Recent advances in Micro-calorimetry, Pergamon Press, N.Y., 1963.

84. B. N. Chapman, The adhesion of thin films, Ph.D.Thesis Electrical Eng. Dept. Imperial College, London,1969.

85. M. M. Nekrasov, U.S.S.R. Patent 127063.86. N. C. Anderson and A. Goodman, A technique for

measuring the adhesion of thin films and the dynamictensile strength of bonds, with commercial applications,Technical Report Published from Sandia Laboratories,Albuquerque, New Mexico, March 1972, 7 pages. TheReport is available from NTIS under numberSC-DR-69-320.

Appendix

SOME CONVERSION FACTORS AND UNITSCOMMONLY USED TO EXPRESS ADHESIONSTRENGTH

Pa (Pascal) 10 dyn cm -2

(i.e., N m -2) 1.4504 x 10 -4 psi1.4504 x 10 -2 gf cm -2

1.0197 x 10-2 gf cm-2.0197 x 10 -s kgf cm-2

1.0197 x 10 -1 kgf m-2

1.0197 x 10 -7 kgf mm-210 #bar7.5 x 10 -3 Torr0.9872 x 10 -5 Atmosphere

kg cm-1 5.6 lb. in -1

J rn -2 10 3 erg cm-2N l0 dynTorr mm (Hg)= 133.3 Pa

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttp://www.hindawi.com Volume 2010

RoboticsJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of



RotatingMachinery


Hindawi Publishing Corporation http://www.hindawi.com

Journal ofEngineeringVolume 2014

Submit your manuscripts athttp://www.hindawi.com

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014

The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

SensorsJournal of


Modelling & Simulation in EngineeringHindawi Publishing Corporation http://www.hindawi.com Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks


ADHESION MEASUREMENT OF THINdownloads.hindawi.com/journals/apec/1976/936912.pdf ·...

Documents

Transcript of ADHESION MEASUREMENT OF THINdownloads.hindawi.com/journals/apec/1976/936912.pdf ·...