tack - espci paris · 2011. 2. 7. · traditionally, tack properties of a psa have been correlated...

1

TACK

Costantino CretonLaboratoire PCSM

Ecole Supérieure de Physique et Chimie Industrielle10, Rue Vauquelin

75231 Paris Cedex 05France

Email: [email protected]

Pascale FabreCentre de Recherche Paul Pascal

Av. Dr Schweitzer33600 PESSAC

FranceEmail : fabre @ crpp.u-bordeaux.fr

to appear in: Comprehensive Adhesion Science, VOl II, The Mechanics of Adhesion, Rheology of Adhesivesand strength of adhesive bonds

last revised: May 10th 2001

2

Table of contents

I. INTRODUCTION ..................................................................................................... 4

II. EXPERIMENTAL METHODS................................................................................. 4

III. ANALYSIS OF THE BONDING AND DEBONDING MECHANISMS.................... 5

III.1 Bonding process................................................................................................ 6

III.2 Debonding mechanisms ................................................................................... 6III.2.1 Initiation of failure: Cavitation ................................................................................................................ 6III.2.2 Foam formation ........................................................................................................................................ 7III.2.3 Fibrillation and failure............................................................................................................................. 8III.2.4 Effect of geometry..................................................................................................................................... 8

IV. MOLECULAR STRUCTURE AND ADHESIVE PROPERTIES ............................ 9

IV.1. Influence of molecular parameters: nature of the monomers .................... 10IV.1.1. Glass transition temperature Tg............................................................................................................. 10IV.1.2.Monomer polarity ................................................................................................................................... 11

IV.2. Influence of molecular parameters : characteristics of the chain.............. 11IV.2.1. Molecular weight of the chain Mn.......................................................................................................... 11IV.2.2. Molecular weight between entanglements Me ....................................................................................... 12IV.2.3. Chemical crosslinks and molecular weight between crosslinks Mc ....................................................... 12

IV.3. Influence of the organization of the chains at a nanometric scale ............ 13IV.3.1. Block copolymers................................................................................................................................... 13IV.3.2. Change in structure through phase transition ....................................................................................... 14IV.3.3. Heterogeneous particles ........................................................................................................................ 15

IV.4. Role of other components ............................................................................. 15IV.4.1. Tackifiers ............................................................................................................................................... 15IV.4.2. Surfactants ............................................................................................................................................. 16

V. INFLUENCE OF EXPERIMENTAL PARAMETERS............................................ 16

V.1. Velocity, temperature, time and pressure of contact ................................... 16V.1.1. Time-temperature superposition principle.............................................................................................. 16V.1.2. Effect of changing the temperature ......................................................................................................... 17V.1.3. Effect of changing the debonding velocity: cohesive to adhesive transition ........................................... 17V.1.4. Role of contact time and pressure ........................................................................................................... 18

VI. SURFACE EFFECTS.......................................................................................... 20

VI.1. Adhesion on low Gc surfaces......................................................................... 20

VI.2. Effect of surface roughness .......................................................................... 21

VII. CONCLUSIONS................................................................................................. 22

3

VIII. FURTHER READING........................................................................................ 23

IX. REFERENCES.................................................................................................... 23

4

I. INTRODUCTION

In order to be effective, an adhesive must possessboth liquid properties, to wet the surface when thebond is formed, and solid properties, to sustain acertain level of stress during the process ofdebonding. Structural adhesives accomplish this bya chemical reaction, typically a polymerization,which transforms a liquid mixture of oligomers intoa crosslinked polymer. For pressure-sensitive-adhesives (PSA) however this transition must occurwithout any change in temperature or chemicalreaction. This property is called tack and gives PSAthe ability to form a bond of measurable strength bysimple contact with a surface. It gives PSA theireasy and safe handling, since the adhesive can beapplied to the surface without the use of anysolvent, dispersant or heat source.Since a PSA must have some amount of tack to beconsidered as such, it is essential to understandwhat are the minimum requirements in terms ofmolecular structure of the components and in termsof formulation for a material to be tacky. However,because tackiness is a complex and not yetcompletely understood mechanism, it is relativelydifficult to establish simple relevant criteria for agood adhesive material.Traditionally, tack properties of a PSA have beencorrelated to their linear rheological behavior, suchas elastic and loss modulus1-3. While this type ofphenomenological analysis provides many clues forthe practical design of PSA, it is intrinsicallylimited by the fact that a tack experiment involveslarge strains and transient behaviors of the PSA,which cannot be easily predicted by either viscosity(shear, elongational) or any other small strainsteady-state dynamical property. The simpleobservation of the debonding of a PSA tape from asolid reflects the complexity of the phenomena atwork : final rupture often occurs through theformation of a fibrillar structure 4,5 and measuredtack energies are much larger than thethermodynamic work of adhesion Wa characterizingthe reversible formation of chemical bonds at theinterface.Moreover, to consider the adhesive alone is notsufficient to predict its behavior in a situation wheresurface effects can be important : the occurrence ofbubbles or fibrils is not only a matter of theviscoelastic properties of the adhesive but dependsalso on the characteristics of the surface of theadherent : roughness, surface tension, and on thethickness of the adhesive film. In fact, surface andbulk effects are coupled and it would thus be moreaccurate to consider adhesive/substrate pairs thanadhesives and substrates separately.Despite these difficulties, recent experimental andtheoretical work focusing on the microscopic

mechanisms taking place during the debonding ofthe PSA from the substrate, have greatly enhancedour understanding of tackiness.

II. EXPERIMENTAL METHODS

Since by definition a tacky material must be stickyto the touch, all standardized testing methods oftackiness seek to reproduce in one way or anotherthe test of a thumb being brought in contact andsubsequently removed from the adhesive surface.The main experimental methods to quantify tackcan be divided into two categories: methods whichprovide essentially a single number are designed tobe very close to the application and mainly aimed atquick comparisons between materials. Amongthose, described schematically on figure 1, are looptack, rolling ball or rolling cylinder tack. In thosemethods, typically the contact time, debondingvelocity, applied pressure, are reasonablyreproducible but cannot be independentlycontrolled. At the other extreme, probe methods aremore difficult to implement (although standardinstruments are commercially available) butprovide much more information and allow thecontrol of the main experimental parametersindependently. Because these tests are moreinformative they will be the focus of the rest of thispaper.

a b

c

adhesive adhesive

adhesive

Figure 1: Schematic of the different methods forthe evaluation of tack properties. a) probe tack ; b)rolling ball tack, c) loop tack

All probe methods are based on the physicalprinciple described on figure 2. A probe, with a flator spherical tip, is brought in contact with theadhesive film, kept in contact for a given time andunder a given average compressive pressure, andthen removed at a constant velocity Vdeb. The resultof the test is a force vs. displacement curve of theadhesive film in tension.

5

Figure 2: Schematic of a probe tack test

Several variations of the test exist. Historically, thefirst referenced probe test is the Polyken probe tacktest developed by Hammond6. In this case theprobe has a flat tip and is upside down. Thecompressive force is controlled by lifting a weightover which the adhesive is deposited. In this earlyversion, only the maximum force of the debondingcurve was recorded and taken as a measure of tack.An improvement over this methodology wasdeveloped by Zosel7, who used a displacementcontrolled instrument and pointed out theimportance of considering the complete debondingcurve rather than simply the maximum in tensileforce. He also used for the first time in-situ opticalobservations during the debonding of the probe,demonstrating that good adhesives are able to formbridging fibrils between the probe and thesubstrate4. However it is only recently that thesequence of microscopic processes occurringduring the debonding of a flat-ended probe from asoft adhesive were elucidated, again thanks to in-situ optical observations and measurements8,9.In parallel to this development, other groups havebeen using spherical tip probes to test tackiness.Using a spherical tip has two importantconsequences. It solves the practical problem ofgood alignment between the probe and the filmwhich gave rise to poor reproducibility of theresults10 and it allows, at least in the early stages ofthe debonding process to study quantitatively themotion of a circular crack and to use the energyrelease rate concepts at the edge of the contactzone11 to measure a value of Gc, the critical energyrelease rate. However, this latter approach, oftencalled the JKR method, is limited to relativelyelastic PSA as discussed in detail in chapter X ofthis book. For soft and highly viscoelastic systems,using a flat probe has the advantage of applying auniform displacement field to the layer facilitatingthe analysis of the fibrillation process.This chapter will therefore focus on results obtainedwith the flat probe with a particular emphasis on the

recent theoretical and experimental developmentswhich have shed some light on the microscopicmechanisms of debonding.

III. ANALYSIS OF THE BONDING

AND DEBONDING MECHANISMS

The tackiness of a specific adhesive is dependent onits ability to bond under light pressure and shortcontact time, while forming a fibrillar structureupon debonding from the substrate. It is thennatural to test tackiness with a quick bonding anddebonding test such as those described in theintroduction.Unfortunately such a test applies a rathercomplicated loading and unloading cycle to ahighly deformable material. Therefore, amicroscopic analysis of the sequence of eventsoccurring during a tack test is necessary in order toattempt a detailed interpretation of a tack curve.Rather than presenting an exhaustive review of themost recent theories, we have chosen to presenthere a rather phenomenological picture which,while it leaves certain aspects unexplained, remainsconsistent with experimental results.

σ

σ

σ

σ

ε

ε

ε

ε

(lmax-l0)/l0d

lmax

Figure 3: Schematic of the deformationmechanisms taking place during a probe tack testand corresponding images of the stages ofdebonding. Images from 12.

Starting from a flat probe tack test such as thatdescribed on figure 2, the sequence of events can bebroken down into four main events8:

Compression

Contact

Debonding

Force

Time

��

σmax

εmax

W

6

Bonding to the surface of the adherent(compression)First stage of debonding : Initiation of the failureprocess through the formation of cavities or cracksat the interface or in the bulk (tension, smalldeformation)Second stage of debonding : Formation of a foamedstructure of cavities elongated in the directionnormal to the plane of the adhesive film (tension,large deformation)Final stage of debonding: Separation of the twosurfaces either by failure of the fibrils (cohesivefailure, i.e. some adhesive remains on bothsurfaces) or by detachment of the foot of the fibrilsfrom the surface (adhesive failure, i.e. there is noadhesive left on the probe surface). Note that werefer here to a visually observable presence orabsence of adhesive on the surface. Surface analysistechniques like XPS almost always find moleculartraces of adhesive on the adherent's surface.The debonding part of this sequence of events isshown on figure 3 in parallel with a stress-strainmeasurement. While the exact sequence of eventsdepends also on the geometry of the test andthickness of the adhesive layer as discussed insection III.2.4, the general features of a stress-straincurve obtained in a probe test of a PSA, remain thesame and will be characterized typically with threeparameters defined on figure 2: a maximum stressσmax, a maximum extension εmax, and a work ofseparation W, defined as the integral under thestress-strain curve. This work of separation shouldnot be confused with the thermodynamic work ofadhesion Wa, an interfacial equilibrium parametercalculated from the values of surface and interfacialenergies or with Gc, characterizing an energydissipated during the propagation of a crack andwhich will be discussed in more detail in sectionIII.2.2.

III.1 Bonding process

The requirements for a good bonding to the surfaceof the adherent have been discussed by Dahlquistsome years ago and these requirements are stillwidely used in the industry today13. They specifythat a PSA needs to have a tensile elastic modulusE' at 1 Hz lower than 0.1 MPa to bond properly tothe surface. More recently this criterion has beenrationalized in terms of the contact between a roughsurface against an elastic plane14,15. The key resultof this study is that one expects a good molecularcontact under zero pressure when the surface forcesexactly balance the elastic energy cost involved indeforming the adhesive film to conform to therough surface. In terms of elastic modulus of theadhesive this can be written as:

< 2/3

2/1

ζRWE a (1)

where R represents the average radius of curvatureof the asperities of the model surface and ζrepresents the average amplitude of the roughness.For R ~50 µm, ζ ~2 µm and Wa ~ 50 mJ/m2, onefinds a threshold elastic modulus of the order of 0.1MPa.

III.2 Debonding mechanisms

A bonding model can predict whether the bondingstage will occur properly but is not sufficient toknow the level of energy dissipation which willoccur upon debonding (for example water wouldeasily pass the test but is not a useful PSA).Therefore, we will now consider the details of thedebonding process.Normally after the contact is established during thebonding stage, a tensile force is applied to theadhesive film until failure and debonding occurs. Auseful PSA will require a much larger amount ofmechanical work to break the contact than to formit and this work is in particular done against thedeformation of bridging fibrils which can extendseveral times the initial thickness of the film asshown on figure 34,8,16,17.

III.2.1 Initiation of failure: Cavitation

Since the adhesive material is rubbery, it deforms atnearly constant volume and the formation of fibrilscan only occur through the prior formation of voidsbetween them. Depending on the experimentalsystem, these voids can first appear at the interfacebetween the adhesive and the adherent or in thebulk of the adhesive layer but they invariablyexpand in the bulk of the adhesive layer asillustrated on figure 38,18.The occurrence of this cavitation process can bereadily understood by noting that the elastic tensilemodulus E of a typical PSA is about four to fiveorders of magnitude lower than its bulkcompressive modulus. A mechanical analysis of thegrowth of an existing cavity in an elastic rubbershows that in such a medium, a preexisting cavity ispredicted to grow in an unstable manner if theapplied hydrostatic tensile pressure exceeds thetensile modulus E of the adhesive19,20. Thisexpansion condition can be roughly written as:

σ > E (2)

7

If the nucleation of cavities is indeed responsiblefor the decrease in the tensile force on figure 2,according to equation 2 one expects the measuredσmax to be directly related to the elastic modulus Eof the adhesive and therefore to obey time-temperature superposition. This is indeed confirmedby experiments showing that for several PSA onsteel surfaces, σmax at a given reduced debondingrate is directly proportional to E’ (the value of theelastic modulus measured with steady stateoscillatory shear measurements) at an equivalentreduced average deformation rate8,18. This result,shown on figure 4, implies therefore that, providedthat Dahlquist’s criterion is satisfied, the larger theelastic modulus of the PSA, the higher its tack forceon a high energy surface. This statement mayhowever no longer be true for low energy surfacesas discussed in section VI.1.

0.0124

0.12

4

12

4

10

MPa

-4 -2 0 2 4log [aTdεεεε/dt (s-1)]

-4 -2 0 2 4

log (f*aT)

E ’

σmax

Figure 4: σmax and E’ as a function of the reducedshear frequency aT f = ω/2π, where aT is the WLFshift factor, or of the reduced strain rate aTdε/dt.(!) σmax ; (---) E’. Data from 8.

Moreover, it should be kept in mind that such asimplistic model for the nucleation of cavitieswould predict a simultaneous expansion of allexisting cavities at the same identical appliedhydrostatic stress. This is contrary to experimentalresults which show that cavities appear sequentiallyat a range of applied stresses. Furthermore, adifficult outstanding question is that of the nature ofthe defects able to expand into cavities. Gay andLeibler argued that, for rough surfaces or roughadhesives, cavities should expand from defectsconsisting of air bubbles trapped at the interfaceduring the bonding process16. This may be true incertain cases but cavities are just as easily nucleatedon smooth surfaces and in the bulk. Better criteriafor the expansion of a cavity, considering theexistence of defects, are currently developed toobtain a quantitative prediction of the maximumtack stress21.

III.2.2 Foam formation

Until now the viscoelastic properties of theadhesive and the surface properties of the substrate(except for its roughness) have not played asignificant role in controlling the mechanisms.They will be essential however in the subsequentprocess, i.e. the evolution of individually expandedcavities into an elongated microfoam structure.Although the formation of the foam is a rathercomplicated process it can be approximatelycharacterized by two important parameters: the cellsize d in the plane of the adhesive film and themaximum extension lmax of the cell walls in thedirection normal to the plane of the adhesive film asshown on figure 3.The final size of the cells d of the foam will clearlyplay a role in the macroscopic stress sustained bythe walls. This final size of the cells will becontrolled by three parameters which arecharacteristic of the behavior of the adhesive and ofthe interface21-23. Two of them are bulkparameters, the unrelaxed elastic modulus of theadhesive Eo (typically at high frequency) and theaverage Deborah number at which the experimentis being conducted. This Deborah number isdefined as the product of the initial macroscopicstrain rate of the test Vdeb/ho, by a relevantrelaxation time of the adhesive τ:

De = τVdeb/ho (3)

where ho is the initial thickness of the film and Vdebis the probe velocity. At high values of De, oneassumes that the adhesive behaves essentiallyelastically with a modulus Eo, while at low valuesof De, significant relaxation of the stresses canoccur within the time frame of the experiment.The third parameter is interfacial: It is the criticalenergy release rate Gc characteristic of anadherent/adhesive pair. It can be approximatelydefined as:

G c = Go (1+ aTφ( a! )) (4)

where Go is the energy release rate extrapolated at avanishing crack velocity a! , aT is the time-temperature shift factor and φ is a bulk dissipativefunction which depends on debonding rate24,25. Gccharacterizes the amount of energy dissipated by apropagating crack at the interface between the PSAand the surface. This parameter is widely used tocharacterize the adhesion between a crosslinkedelastomer and a hard surface25,26 and, with someprecautions, can also be used for viscoelasticPSA's11.

8

Coming back to the foam formation, the simplestcase is that of high values of Gc and high Deborahnumbers. In this case, the characteristic lateraldimension of the cells d will only be controlled bythe elastic modulus E and the thickness of the layerho:

2/1

≈

EKhd o (5)

where K is the bulk modulus of the adhesive21,22.The distance d is representative of the length overwhich the stress is relaxed by the expansion of thecavity. Forgetting about numerical constants, thismeans that one expects the lateral dimension ofthese cells to be directly related to the thickness ofthe film and inversely proportional to the squareroot of the shear modulus (the bulk modulus K doesnot vary much from one soft adhesive to anotherand is of the order of 1 GPa).If the polymer can relax during the test (De < 1),the cavities can expand laterally causing what isanalogous to a dewetting of the sample as shown onfigure 523.

A B

Figure 5. Comparison of the observed cavities in apoly(2-ethylhexyl acrylate) adhesive debonded : A)at 10 micron/s and B) at 100 µm/s. Images from 8.

III.2.3 Fibrillation and failure

Assuming that the cavities have expanded laterallyas much as they can without coalescing, the laststage of the debonding process will start with thevertical elongation of the walls between cells. Thismechanism implies that, at the molecular scale,there is a progressive orientation of the polymerchains in the direction of traction27,28. There is aninteresting analogy between this process and theformation of craze fibrils in glassy and semi-crystalline polymers. However while craze fibrilsgrow in length only by drawing fresh material froma reservoir of unoriented polymer29, the situationis less clear for PSA fibrils where some of theextension is a result of fibril creep and some is due

to the drawing of unoriented polymer from the foot.The respective weight of these two mechanismswill depend on the rheological properties of theadhesive in elongation: a weakly strain hardeningadhesive will favor fibril creep, while fibrils formedby a strongly strain-hardening adhesive will growby drawing of unoriented polymer from the foot.Once most of the polymer chains are well-oriented,the stress on the fibrils can increase again, causingeither an instability and a fracture of the fibrilsthemselves (macroscopically a cohesive fracture) orthe detachment of the foot of the fibril from thesurface of the adherent (macroscopically anadhesive fracture).The occurrence of one or the other of theseprocesses will depend on a delicate balancebetween the tensile properties of the adhesive andthe interfacial parameter Gc. Despite the fact thatthe level of stress and the maximum extension thatthese fibrils will achieve often controls the amountof work necessary to debond the adhesive (theexternal work done during this process cansometimes represent up to 80% of the practicaldebonding energy), no quantitative analyticaltreatment of this extension and fracture processexists for such highly non linear materials.Numerical methods have however been successfulin predicting at least the extensional behavior if notthe point of fracture30,31.

III.2.4 Effect of geometry

While the above section oversimplifies thedebonding process by separating it into individualstages which are not really independent from eachother, it is however a first step towards a betterunderstanding of the critical parameters controllingtackiness. The three stages of debonding describedearlier, assumed that cavitation was the firstmechanism in the failure process. While this is truefor useful PSA, it is worthwhile to briefly considerthe limits where this may no longer be true. When atensile stress is applied to a confined layer ofarbitrary elastic modulus, one can envision twolimiting cases: a very hard adhesive will notcavitate but form a crack (and the probe tack willbecome the butt joint geometry) and a simple liquidwill form a single filament (and this experiment iscalled a squeeze-flow test).In order to understand in which conditionscavitation is likely to occur, it is necessary toconsider the effect of the coupling between theexperimental geometry and the mechanical andinterfacial properties of the adhesive on the failuremechanisms.The results of fracture tests of adhesive bonds arealmost never independent of the experimentalgeometry because the presence of the interface withits discontinuity in elastic properties ensures that

9

the stress field at the interface depends on both theexternal loading and the elastic properties mismatchas discussed in chapters on hard adhesives.However soft adhesives have the addedcomplication to dissipate energy, not only in arestricted plastic zone near the interface, but over alarge volume, often the entire volume of thesample. This means that there is a very strongcoupling between the boundary conditions of thetest (thickness of the layer, size of the probe andstiffness of the probe) and the observeddeformation mechanisms.In a recent study, Crosby et al.32 have discussed thedifferent possible initial failure mechanisms of athin adhesive elastic layer in a probe test and haveextracted two geometrical parameters which couplewith the two material parameters E and Gc: thedegree of confinement of the adhesive layer(represented by the ratio of a lateral dimension overa thickness of the layer) and a characteristic ratiobetween the size of a preexisting internal flaw acand a lateral dimension of the system a. Theydistinguished among three main types of initialfailure: bulk cavitation, internal crack and externalcrack as shown on figure 6.

1) Cavitation in the bulk

2) Internal crack

3) External crack

crackcrack

probe

F

uprobe

Figure 6: Schematic of the initial failuremechanisms of the elastic layer. Top views (left)and side views (right). Arrows indicate the directionof expansion of the cracks or cavities. In the case ofbulk cavitation, the nucleation can be at theinterface or in the bulk.

When the confinement is high, elastic instabilitiessuch as cavitation in the layer are strongly favored,the more so of course if the elastic modulus of theadhesive is low and the interfacial energy releaserate Gc is high. On the other hand, weak adhesion, ahigher elastic modulus and a lesser degree ofconfinement all favor crack propagation as shownon figure 7.

��

��

��

10-7

10-6

10-5

10-4

10-3

10-2

10-1

100

101

10-2 10-1 100 101 102 103

a/h

ExternalCrack

BulkCavitation

InternalCrack

/EaGGGGcccc

1

3

2

Figure 7: Deformation map with Gc/Ea as afunction of a/h,. Points 1 : low value of Gc/E ,transition from external crack to internal crack withincreasing confinement. Points 2 : intermediatevalue of Gc/E, transition from external crack tocavitation with increasing confinement. Points 3 :high values of Gc/E, always failure by cavitation.Case 1 is typical of a crosslinked rubber on steel orof a block copolymer-based PSA on release paper,case 2 is typical of an acrylic PSA on release paperand case 3 is typical of a PSA on high energysurfaces.

IV. MOLECULAR STRUCTURE AND

ADHESIVE PROPERTIES

A polymer (or copolymer) chain is characterized bythe nature and distribution of its constitutivemonomers along the chain, and molecularparameters such as its number average molecularweight Mn and polydispersity Mw/Mn, its averagemolecular weight between entanglements Me and itsdegree of branching. These factors have animportant effect on the bulk properties of thematerial, such as the glass transition temperature Tgor the large scale organization. On the other hand,the molecular parameters also determine therheological behavior, expressed by the differentmoduli and characteristic relaxation times for small(monomer) or large (chain or a part of a chain)scale motions. In Figure 8, is represented the typicaltensile modulus E of a high molecular weightpolymer with a narrow molecular weightdistribution. At a given temperature, for relaxationtimes shorter than a characteristic time te, orfrequencies of motion larger than 1/te, the chain isunable to relax at a large scale : the high frequencyvalue of the modulus is related to the nature of themonomers and their local mobility. Between te and

10

td, the chain is relaxed on a typical scalecorresponding to the distance betweenentanglements : one observes a plateau region forthe modulus Eo, where its value is inverselyproportional to the average molecular weightbetween entanglements Me. Finally, at times largerthan td, the polymer flows like a liquid : in thisrange, the value of the modulus is related to thepolymer molecular weight, and its degree ofbranching33,34.

Log tte td

E(t)

Eo

Figure 8: Relaxation modulus E(t) as a function oftime at a fixed temperature for a high molecularweight polymer with a narrow molecular weightdistribution.

It is helpful, in order to understand the adhesivebehavior of a PSA, to compare, for the sametemperature, the typical experimental times orfrequencies involved in a quick tack experiment(contact time, separation rate) to the characteristicrelaxation times defined above. For instance, whenthe bond formation step is realized in a time shorterthan te, the very high modulus of the material willprevent it from making a good contact, thus leadingto a poor adhesive behavior. In a similar way, theseparation rate has to be within a range wherefibrils can form in order to obtain large dissipationand thus a good adhesive behavior. In practicalsituations the separation rate and temperature areusually imposed by the specific use that is beingmade of the adhesive, so that one needs to play withthe molecular weight of the polymer, its degree ofbranching or crosslinking and the monomer frictioncoefficient in order to modify its relaxation timesand thus its viscoelastic losses during thefibrillation stage. Thus, even if not sufficient byitself, the knowledge of the linear viscoelasticproperties E’ and E'' versus frequency curve gives astrong indication of the suitability of a givenmaterial for adhesive purposes.As a first approximation, a suitable molecularstructure for a tacky material could be described asa nearly uncrosslinked network : a low plateaumodulus will give a high compliance of the layerand therefore a good contact with the surface, andcrosslinks and entanglements will give thenecessary cohesive strength to form stable fibrils atdebonding frequencies. In practice, this is often

realized by combining two main ingredients: apartially crosslinked, low Tg high molecular weightcomponent and a high Tg, low molecular weightcomponent called a tackifier, but as we will see inthe following, many types of structures can lead tothe right balance of properties.

IV.1. Influence of molecularparameters: nature of themonomers

IV.1.1. Glass transition temperature Tg

Since the characteristic relaxation times of the base-polymer in the PSA are always decreasing withtemperature, one would expect the properties of aPSA to be also monotonically dependent ontemperature : experimentally however one observesthe existence of a fairly sharp optimum of the tackproperties of a material with temperature. This isillustrated in Figure 9, for various polyacrylates35and will be discussed in more detail in sectionV.1.2. More generally, it is admitted that, for thePSA properties of an adhesive to be optimal, its Tgshould lie somewhere between 70°C and 50°Cbelow the use temperature for an acrylate or naturalrubber based adhesive and between 30 and 50°Cbelow the usage temperature for an SBC (styrenicblock copolymer) based adhesive. This temperaturecorresponds to a balance between an elasticmodulus E lower than 105 Pa (Dahlquist's criterion)and a high level of dissipation upon debonding. It isin fact one of the roles of a tackifying resin, or of acomonomer in an acrylic PSA, to raise the Tg of thesystem when the Tg of the polymer alone is too low.

100

80

60

40

20

0

W (J

/m2 )

806040200-20-40

T (°C)Figure 9: Debonding energy W as a function oftemperature for probe tack tests of different acrylicpolymers. (!) Poly(2-ethylhexyl acrylate) ; (")Poly (n-butyl acrylate) ; (#) Poly (isobutyl

11

acrylate) ; ( ) Poly (ethyl acrylate) ; ( ) Poly(methyl acrylate). Contact time 0.02 s. Data from35.

IV.1.2.Monomer polarity

Since it was noticed that the incorporation of polarmonomers such as acrylic acid could enhance thecohesive and adhesive strength of a material,several studies have been devoted to the role playedby the polarity of the monomer in the adhesionprocess.The influence of incorporating polar monomers iscomplex : it both modifies the surface tension andthe bulk properties of the adhesive. However thesetwo effects are not generally dominant at the sametime. In tack experiments, the most visible changebrought about by acrylic acid is a sharp increase inthe long relaxation times of the polymer36. Asshown on figure 10, this causes a shift in thecharacteristic debonding rate at which the transitionfrom cohesive failure of the fibril to adhesivefailure is observed18.

25

20

15

10

5

0

ε max

10-1 100 101 102 103 104 105 106

aTVdeb

25

20

15

10

5

0

ε max

10-1 100 101 102 103 104 105 106

aTVdeb

cohesive adhesive

cohesive

Figure 10: Maximum extension of the fibrils εmax as a function of reduced debonding rate aTVdeb. a)for the PnBA with 2.5% acrylic acid; b) for thePnBA without acrylic acid. T = 23°C, tc = 1 s. (!)

cohesive failure ; (") adhesive failure. Data from18.

However in peel tests, which typically involve longcontact times, the dominant effect is reversed:acrylic acid moieties can slowly migrate to theinterface with the substrate and cause a significantincrease of the interactions which can switch thefracture mode back to cohesive37. More generally,the presence of monomers of different polaritiescan lead to specific time-dependent effects relatedto the kinetics of diffusion or reorientation of thedifferent species at the surface during theexperimental time. Finally, the presence of acrylicacid can suppress the time-temperature equivalencegenerally observed for adhesive tests of softpolymers8,38.

IV.2. Influence of molecularparameters : characteristics of thechain

IV.2.1. Molecular weight of the chain Mn

Commercial PSA's have typically a very broadmolecular weight distribution or alternatively a verybroad distribution of terminal relaxation times. Thispolydispersity is essential to obtain good PSAproperties.In order to understand why this is the case, it isuseful to examine the results obtained withpolymers with a narrow or very narrow molecularweight distribution18,35,39.

14

12

10

8

6

4

2

0

W (J

/m2 )

103 104 105 106 107

Molecular weight (g/mole)

Figure 11: Adhesion energy W as a function of theweight average molecular weight for apolyisobutylene with a low polydispersity. Datafrom 35.

As shown on figure 11 for a polyisobutylene, if thecomparison is made for the same experimentalconditions, reasonable tackiness is only obtained ina relatively narrow range of molecular weights.This can be understood in the following way: the

12

increase of Mn increases the viscosity η, since η ∝Mn

3.4, or, in terms of characteristic relaxation times,the terminal relaxation time τd. In a tackexperiment, this increase in viscosity leads to alarger cohesive strength of the material, which inturn causes a better stability of the PSA fibrils, oncethey are fully formed, and thus to a larger value ofεmax. On the other hand, the formation of the fibrilsfrom the initially formed cavities requires a certainamount of flow so if the terminal relaxation timeand the viscosity increase too much the fibrillarstructure is never formed. As a general guide,experiments conducted at room temperature andrelatively high debonding rates show a maximumtack energy at a molecular weight which isapproximately 5-10 times the average molecularweight between entanglements Me.When considering the molecular weight effects, oneshould nevertheless remember the influence ofexperimental parameters : the optimum molecularweight will depend strongly on the rate andtemperature of the test with high temperatures andlow debonding rates shifting the optimum towardshigher molecular weights18.Conversely, the temperature window of optimumtack for a given molecular weight will be rathernarrow (approximately two decades in frequency or10°C based on figure 10). For the more commoncase of polymers which are not monodisperse buthave a broad molecular weight distribution, theresulting broad distribution of terminal relaxationtimes circumvents this problem allowing both fibrilformation and fibril stability for a range ofexperimental conditions. As a result, an optimum intack energy with Mn is still observed but becomesalso broader and is shifted to higher molecularweights relative to the monodisperse case.

IV.2.2. Molecular weight betweenentanglements Me

Entanglements are crucial in the behavior of PSA.Clearly unentangled polymers do not work well asPSA since they do not have enough cohesivestrength to form stable fibrils. On the other hand theterminal relaxation time of an entangled polymer τdis dependent on (Mn/Me)3 from the reptation modeland plays a role in controlling the elongation of thefibrils as described in the preceding section.Additionally, the average molecular weightbetween entanglements plays also a major role inthe early stages of the debonding process since itcontrols the elastic modulus in the plateau region.Indeed, since tack tests are usually conducted in afrequency range where E is in its plateau region(where E' = Eo ∝ 1/Me) a change in Me has a directconsequence on the value of E' at the debondingfrequency.

For polymers with molecular weights much largerthan Me, a transition in debonding mechanism isobserved for Me larger than approximately 104

g/mol40. At low values of Me, adhesive failure bycrack propagation is observed while for high Mepolymers, failure by cavitation and fibrillation isobserved. This transition can be understood by apurely mechanical argument. The critical stress forcavitation in the bulk to occur is proportional to theelastic modulus E' of the polymer at the testingfrequency. Therefore, when E' decreases, thecritical stress for cavitation decreases, and becomeseventually smaller than the Gc for crack propagationat the interface32 41. Therefore, for low Me, theinitial debonding is by crack propagation andcannot then evolve towards a fibrillar structurewhile for high Me, the debonding occurs throughcavitation and fibrillation and a larger amount ofenergy is dissipated in the process. This transitionfrom fibrillation to homogeneous deformation,which can also be concomitant with a transitionfrom cohesive to adhesive failure, is normallyaccompanied by a drop in tack energy. Zoselpointed out that this critical value of Mecorresponds in fact to the Dahlquist criterion for theelastic modulus E, below which a material is notconsidered tacky : E=105 Pa. Me plays thus a veryimportant role on the debonding stage in a PSA,since it influences the ability for fibrillation, thetype of rupture, and consequently the debondingenergy.

IV.2.3. Chemical crosslinks and molecularweight between crosslinks Mc

Experiments on a series of PDMS adhesives, haveshown that the tack energy is maximum, for adegree of crosslinking which is slightly above thegel point42. Below the gel point, the crosslinks'main effect is to increase the terminal relaxationtime of the polymer. The presence of long brancheswill in particular play a large role on theelongational flow properties of the polymer andincrease fibril stability. This will lead in turn to ahigher value of εmax and a higher tack energy.Above the gel point, additional crosslinking willinitially have an effect on the strain hardeningwhich occurs in extension. Significant strainhardening in extension will occur in the polymer foran increasingly lower level of strain, causing earlydetachment of the fibrils and a lower measured εmax.This effect is illustrated on figure 12 on a series ofacrylic polymers crosslinked beyond the gelpoint40. One should note that the peak, and theheight of the plateau, remain unchanged implyingthat the viscous dissipation in the fibrils and theplateau modulus Eo are much less affected by amodification of the crosslink structure. By

13

extrapolating these measurements, one can arguethat fibrillation should be suppressed altogetherwhen the crosslink density becomes equal to theentanglement density. Following the map of figure6, a change in mechanism from cavitation to crackpropagation has occurred, mainly caused by adecrease in the interfacial dissipative term Gc thistime.

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

Stre

ss (M

Pa)

1086420Strain

Mc/Me = 2 3 5

uncrosslinked

Figure 12: Stress-strain curves of PnBA on steel,UV crosslinked to different degrees as indicated bythe ratio Mc/Me. Data from 40.

Therefore, in a similar way as entanglements do,crosslinks decrease the ability of the material toflow, and eventually lead to a transition to failureby crack propagation and coalescence, and thus to alower tack energy. However, crosslinks arepermanent and completely prevent largedeformations; this is not the case for entanglements,provided that the deformation occurs at sufficientlylow rates.Crosslinking is widely used in the PSA industry totune the properties of removable adhesives byreducing εmax. An example of that type of effect, isthe development of PSA for trauma free removalpatches, where in-situ irradiation leads tocrosslinking, and thus to highly reduced peelingforces for the adhesive43.

IV.3. Influence of the organization ofthe chains at a nanometric scale

The main polymers used as PSA can be divided in afew large classes according to their chemicalstructures : natural rubber based, styrenic blockcopolymer based (mainly triblocks and diblocks ofstyrene and butadiene or isoprene), acrylics and fora smaller part, silicones. From a microstructuralpoint of view, however, the classification should bedifferent. As already mentioned, a pressure-sensitive-adhesive requires both a good shearresistance and some capability to flow in order tofunction properly. The resistance to shear isnormally achieved by introducing crosslinks:

chemical ones as in natural rubber or acrylics, butalso physical ones. In the case of stronglyimmiscible comonomers, for example, anddepending on the chain statistics, the architectureand monomer composition of the polymer chaincan lead to microphase separation so that themicroscopic domains act as physical crosslinks forthe system : this is the case for styrenic blockcopolymers. In certain cases, in addition, there canbe a long range ordered structure, such as a lamellarstacking. All these structures lead to differentbehaviors as far as tack is concerned. We list belowseveral types of systems, as examples of differenttypes of phase separated structures, although theirimportance from an industrial point of view is notnecessarily comparable.

IV.3.1. Block copolymers

Block copolymers with incompatibleblocks which are able to microphase separate aregood candidates for PSA properties. Indeed, blendsof ABA triblocks and AB diblocks, where therubbery midblock of the ABA is the majority phaseand the glassy endblocks self organize in hardspherical domains and form physical crosslinks, arewidely used as base polymers for PSA. The actualadhesives are always compounded with a lowmolecular weight tackifier resin able to swell therubbery phase and dilute the entanglement network.Linear styrene-rubber-styrene copolymers, withrubber being isoprene, butadiene,ethylene/propylene or ethylene/butylene, are themost widely used block copolymers in thiscategory.This class of PSA has unique properties which arerelated to their molecular superstructure. Indeed,compared to chemical crosslinks, physicalcrosslinks present several advantages, the first onebeing that they are reversible with temperature, thusleading to a large viscosity decrease above the glasstransition temperature of the endblocks : this makesthese systems very suitable for hot melt processing,an increasingly popular processing method due toenvironmental regulations. A second advantage isthat the crosslink density is naturally fixed by thecomposition of the system, provided that it is atequilibrium, and thus easier to control thanchemical crosslinking. Finally, the physicalcrosslinks provided by the hard domains are quitefixed under low stresses, giving very good creepproperties, but can be broken and reformed at highstresses, which is essential for the formation of thefibrillar structure.As shown schematically on figure 13, one candistinguish two types of configuration for thetriblock molecules: if the two endblocks of amolecule are incorporated in separate harddomains, the triblock will be described as a bridge.

14

If alternatively both endblocks are incorporated inthe same hard domain, the triblock will bedescribed as a hairpin molecule.Although this process is not fully understood, it isbelieved that the number of bridge moleculesbetween hard domains controls the large strainbehavior. An increase in the amount of tackifyingresin in a pure triblock copolymer causes a changein the ratio of bridge to hairpin molecules, andconsequently a change in the value of εmax as shownon figure 14.

Bridge molecule

Hairpin molecule

Figure 13: Schematic of the molecular structure ofstyrenic block copolymers showing the hairpin andbridge configurations.

1.2

1.0

0.8

0.6

0.4

0.2

0.0

Stre

ss (

MP

a)

6543210

ε

40% 50% 60%

Figure 14: Stress-strain curves of an SIS + resinPSA on steel44 with three different amounts ofresin in wt%. The base polymer was Vector 4311(Exxon) and the resin was Regalite R101(Hercules). Vdeb = 10 µm/s, T = 40°C.

Similarly to the case of simpler homogeneoussystems discussed in section IV.2 the rheologicalproperties of these PSA will be essential incontrolling their adhesive behavior. However themodifications of molecular structure to obtain theserheological properties will be different. Forexample the apparent plateau modulus E' will nolonger be controlled by Me, but by the molecularweight of the elastomeric midblock and by theamount of resin which is incorporated. The terminal

relaxation time controlling fibril extension can besomewhat tuned by replacing some of the triblockin the adhesive by double the amount of diblockwith one half the molecular weight, effectivelymodifying the ratio of hairpin to bridges. It isimportant to note that this ratio can effectively bemodified by the processing conditions (hot-melt,solution cast). In a recent work, Flanigan et al45showed that the adhesive properties of acrylictriblock copolymers cast from two differentsolvents could be very different, even though theirstructure as determined by X-ray scattering was thesame, i.e. the hard domains size and spacing wereidentical.Other studies on the properties of block copolymerswith more complicated architectures exist althoughnot all of them consider tack. Some studiescomparing properties of radial versus simple blockarchitecture allow to compare the effect of chemicalversus physical crosslinks in the case of structuredsystems. The data shows that there is a decrease ofmelt viscosity together with better adhesiveproperties for star copolymers46. This is related to apoint that was discussed earlier : the differentarchitectures of the molecules lead to differentphysical crosslinking densities. In the same spirit,block copolymers with four different arms, twopolyisoprene, and two polystyrene-b-poly(ethylene/butylene) arms were studied, and ledto a better combination of shear strength and meltviscosity for adhesive applications47, compared tothe conventional linear SIS and SEBS triblocks.

IV.3.2. Change in structure through phasetransition

Previously discussed systems are always phase-separated at application temperatures and undergo achange in properties at high temperatures fromelastomeric tacky to liquid for processing reasons :this change occurs over a wide temperature range.It is however possible to tune the molecularstructure to obtain a transition from solid non-tackyto elastomeric tacky for a given temperatureindependent of debonding rate, by using athermodynamic phase transition.Recent experiments48 have compared the tackproperties of a given system below and above thephase transition between an organized smecticphase and an isotropic phase. The system used wasa side-chain fluorinated copolymer, with two typesof pendant groups both able to crystallize. Belowthe transition, in the smectic phase, the systempossesses no tack due to its high elastic modulusthat prevents the formation of a good bond with thesurface. Above the transition, a soft, phaseseparated region with possibly a continuousnetwork of crystalline domains, allows the system

15

both to achieve a good bond for short contact timesand extended cavitation followed by largedeformations. This type of system presents theinteresting feature to have a very abrupt non tacky-tacky transition at a temperature which can inprinciple be varied at will by changing theirmonomer composition.

IV.3.3. Heterogeneous particles

There are very few studies published on PSA filmsobtained from heterogeneous latexes: indeed, mostPSA used in the form of latexes, such as styrene-butadiene or acrylic systems, are randomcopolymers. Although emulsion polymerizationdoes not allow in general as much control over themolecular structure and therefore over adhesiveproperties, the use of solvent-free adhesives toreplace their solvent-based counterparts isincreasingly important, due to the recentenvironmental regulations. One example where thestructure of the final adhesive film can be finelytuned at a scale of a few tenths of a nanometer is .given by a recent study49 which compares the tackproperties of two acrylate copolymer samplessynthesized either with a batch or continuous feedprocess, leading to particles of differentheterogeneities. The tack properties of the twosystems are markedly different, thus rising thequestion of the role of structural heterogeneities atthe particle scale. Such heterogeneities may, inprinciple, be probed by AFM methods, effectivelyperforming a “nano-tack” experiment on isolatedparticles or on monolayers of latex particles50: thispowerful tool could help to understand thecorrelation between the macroscopic properties ofan adhesive and its microscopic response.

IV.4. Role of other components

A formulated adhesive contains in general, inaddition to the base polymer, some small moleculeadditives: a tackifier, and some fillers, plasticizersand antioxidants. Additionally if the adhesive hasbeen obtained by emulsion polymerization itcontains some surfactants. Tackifiers andplasticizers are added to tune the viscoelasticproperties of the adhesive, while surfactants aregenerally unwanted residues of the polymerizationprocess. Again, much technology, most of itproprietary, is involved in these formulationparameters and we only address here some genericpoints based on what is available in the openliterature.

IV.4.1. Tackifiers

The role of the tackifier is to adjust theviscoelastic properties of the system. It typicallyconsists in short chain polymers of molecular massbetween 300 and 3000 g/mole, with a softeningtemperature between 60°C and 115°C, dependingon the adhesive. The tackifying resin has a dualrole: increasing the Tg of the system, whichincreases the viscoelastic losses at high frequencies,and decreasing the modulus at the low frequenciesthat are important at the bond formation stage51,52.The decrease in elastic modulus in the plateauregion can be interpreted as a dilution of theentanglement structure. For a given Tg of thetackifier, there is an optimum weight fraction in aPSA: at low tackifier contents, the plateau modulusis too high to satisfy Dahlquist's criterion and athigh tackifier contents, the Tg of the PSA becomestoo high and again poor bonding occurs. A simple way to determine the optimum amount ofresin is to determine the amount which gives theminimum value of the elastic modulus of thesystem as shown on figure 1535.

10

8

6

4

2

0

W (J

/m2 )

100806040200%Resin

1.2

1.0

0.8

0.6

0.4

0.2

0.0

E (MPa)

Figure 15 : Elastic modulus E in a relaxationexperiment ($) and adhesion energy W in a probetack test on steel (!) as a function of resin contentfor a natural rubber/glycerol ester of hydrogenatedresin blend. Data from 35.

Of course, in order to be able to modify theviscoelastic properties of a system, and to changethe entanglement density, a tackifier needs to bemiscible with the adhesive, which implies that ithas to be chemically adapted to the adhesive52-56.This is realized via different families of tackifiers,depending on the nature of the adhesive compound.Giving an exhaustive list of tackifiers together withtheir compatibilities with adhesives would bebeyond the scope of this review, since new familiesof tackifiers with better compatibility appear on aregular basis57 but the interested reader can refer tomore technologically oriented texts for furtherinformation58.

16

IV.4.2. Surfactants

For emulsion adhesives, the presence andnature of surfactants is also an important parameter.Their action is twofold: by diffusing to the surfaceof the film, they may change its properties and itsability to form a good contact, and by plasticizingthe polymer they can change its bulk properties.However while some studies have appeared on theireffect on peel properties59,60, not much is knownon their influence on tack.

V. INFLUENCE OF EXPERIMENTAL

PARAMETERS

V.1. Velocity, temperature, time andpressure of contact

In a probe tack test, several experimentalparameters can be varied independently, such as thepressure p exerted by the probe on the surface ofthe adhesive during the compressive stage, theduration of the contact stage called contact time tc,or the separation rate of the probe Vdeb. Since achange in experimental parameters can sometimeslead to a change in fracture mechanism, theevolution of the adhesion energy with theseparameters is in general complex, and deserves tobe described a little further. When the debondingrate Vdeb is varied, the characteristic strain ratesapplied to the adhesive layer are changedaccordingly, which in turn modifies its viscoelasticresponse. The response of the material, whichdepends on its spectrum of relaxation times, is alsoa function of the temperature of the sample. Thecontact time tc and pressure p can control the size ofthe real contact area, but also the degree ofrelaxation of the adhesive when the debondingstarts. Effectively, this means that the initialcondition of the debonding part of the test willdepend on the applied pressure and contact time inthe compressive part of the test. Depending on thebulk properties of the material, or on the features ofthe probe-adhesive interface such as roughness, thisdifference in initial condition can have a negligibleeffect (typically for soft PSA on smooth surfaces)or a large one (for stiffer PSA on rough surfaces).

V.1.1. Time-temperature superpositionprinciple

The time-temperature (t-T) superpositionprinciple is based on the idea that when a polymeris deformed, a change in the characteristic

deformation rate is equivalent to a change intemperature. In the context of tack tests, one canassume that the characteristic deformation rate isthe Vdeb/ho so that an increase in Vdeb would beequivalent for example to a decrease intemperature. This t-T superposition, which is verywidely used in linear rheology33 where smalldeformations are applied, can apparently also beused for the large deformations typical of tackexperiments35. However a necessary condition forthis t-T equivalence to apply is that no change infracture mechanism should be induced by a changein the initial condition of the test as defined in V.1.If this change in initial condition causes aqualitative change in the mechanisms involved inthe debonding stage at the microscopic level, therate dependence of the adhesion energy has noreason to follow the t-T principle. As an exampleshown on figure 16, a series of tack experiments8 ata contact time of 60 seconds obey very well the t-Tsuperposition while the same series of experimentsat a contact time of 1 second does not. In this case,the short contact time did not allow the fullrelaxation of the adhesive at the lower temperature,causing therefore a change in the initial condition.Experimentally this change had a moderate effecton the cavitation process, but triggered a significantchange in the later stage fibril formation process asshown on figure 16. Consequently, a master curvefor σmax could be easily built but not so for W asshown on figure 16b and c. This type of effect ismore dominant when the surface is rough and thetime of contact does not allow the adhesive to fullyrelax.

17

0.8

0.6

0.4

0.2

0.0

σ (M

Pa)

121086420

ε

a

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

σ cav (

MPa

)

6420 log(aTVdeb)

20 °C 50 °C 0 °C -20 °C

b

500

400

300

200

100

0

W (J

/m2 )

76543210-1-2 log (aTVdeb)

20 °C 50 °C 0 °C -20 °C

c

Figure 16: a) Stress-strain curves of an acrylateadhesive on steel at two identical values of reduceddebonding velocity. (----) Vdeb = 100µm/s, T = 0°C;(- - -) Vdeb = 1000 µm/s, T = 20°C. b) Maximumstress σmax vs. reduced debonding rate and, c)adhesion energy W vs. reduced debonding rate forthe same adhesive. Note that the low temperaturedata for W do not follow t-T superposition. Datafrom 8.

Another interesting example where t-Tsuperposition fails is that of a change oftemperature leading to a thermodynamic phasetransition in the material (melting of a crystallinephase) and therefore to a change of structure48.In both examples, an experimental parameter hascaused a change in the initial condition of thedebonding test which, in turn, modified themicroscopic separation mechanisms .

V.1.2. Effect of changing the temperature

As explained earlier, the temperature of the test isimportant since it modifies the relaxation times ofthe polymer and therefore the rheological behaviorof the material : an increase of temperature near theTg of the polymer decreases its elastic modulus,thus leading to a better contact area and a largeradhesion energy. On the other hand, whentemperature is too high, the viscosity of the systembecomes very low, decreasing the adhesion energyagain. Therefore a maximum in tack as a functionof temperature is usually observed at approximatelyTg + 50°C as described earlier and shown on figure17. Although a proper formulation can extend theuseful temperature range of a PSA or modify thetemperature difference between the Tg and themaximum tack, it is extremely difficult to obtain amaterial which retains tackiness over a temperaturerange in excess of 50°C.

120

100

80

60

40

20

0

Adhe

sion

ene

rgy

(J/m

2 )

806040200-20-40

Temperature (°C)

No fibrillation Fibrillation

Figure 17: Adhesion energy W as a function oftemperature for a PEHA adhesive on steel. Thefinal separation is always interfacial but a cleartransition between non-fibrillar and fibrillardebonding is observed at T = -10°C. The glasstransition of the adhesive is around -55°C. Datafrom35

A more subtle effect of temperature is alsodisplayed on figure 17 where experiments35 showa rapid increase of the adhesion energy withtemperature, related to a change in the rupturemechanism from interfacial fracture by crackpropagation to failure by fibrillation and debondingof the fibrils. The interpretation of the authors isthat the temperature change is responsible for amodification of the adhesive-probe contact area,which in turn results into different types of rupturemechanisms in the two temperature ranges.

V.1.3. Effect of changing the debondingvelocity: cohesive to adhesive transition

18

The strain rate of an adhesive sample in a probe-tack test is related to the debonding velocity of theprobe, however not in a straightforward way.Indeed, the separation stage induces veryheterogeneous shear and elongation flows in thesample32. A common approximation, bypassing allthese considerations, is however to consider that, atthe beginning of the separation process at least, thesample is homogeneously deformed at a frequencyclose to Vdeb/ho, where ho is the sample thickness.For a monodisperse linear polymer with well-defined relaxation times, its response can be wellpredicted by a Deborah number18. In the regime ofcohesive failure (low De), the adhesion energy W iscontrolled by the value of εmax which continuouslyincreases with De. In this regime σmax does not varymuch with De. On the other hand, in the regime ofadhesive failure (high De), εmax is always low andW is mainly controlled by σmax which increaseswith De.In this case, a sharp drop of tack energy occurs atthe transition from cohesive to adhesive failure, i.e.for De = τVdeb/h between 100 and 1000. Such atransition is completely analogous to that observedby others in peel tests61,62 and corresponds to achange in the initiation mechanism of failure asdiscussed in section III.2.4.

14

12

10

8

6

4

2

0

W (J

/m2 )

2.01.51.00.50.0

Vdeb (mm/s)

Figure 18: Adhesion energy W as a function of Vdebfor a spherical glass indenter (JKR geometry) on ahigh molecular weight PDMS polymer melt. (%)cohesive fibrillar fracture ; (!) adhesive fracture.Data from 63.

A similar change in mechanism with increasingvelocity occurs for the debonding of a glassspherical indenter from a high-molecular weightpolydimethylsiloxane (PDMS). In this case thedebonding occurs in a situation of low confinement(value of a/h is low) with no cavitation. As shownin figure 1863, the adhesion energy is increasingwith velocity in the cohesive regime, where theseparation occurs by bulk yielding and througheventual rupture of a single large column ofpolymer. If Vdeb is further increased, adhesionenergy drops abruptly then slowly decreases with

Vdeb and reaches an asymptotic behavior. In thisregime the separation occurs through thepropagation of an external radial crack. A similar behavior has been observed for acrylicsystems when a flat probe was used (high value ofa/h). As shown on figure 19, the adhesion energyslowly increased in the fibrillation regime, anddropped when the fibril formation wassuppressed40.

100

80

60

40

20

0

W (J

/m2 )

12 4 6 8

102 4 6 8

1002 4 6 8

1000

Vdeb (mm/s)

Figure 19: Adhesion energy W as a function of Vdebfor a PEHA-based adhesive on steel at T = 20°Cand 1 s. contact time. Transition from fibrillar (%)to non-fibrillar (!) fracture. Data from 40.

These three experiments present two similarfeatures : an increase of adhesion energy with V inthe fibrillar regime, whether there is “one” or manyfibrils, and the drop in adhesion energy when goingfrom a fibrillar to a non-fibrillar adhesiveseparation process, which corresponds to a changein initial fracture mechanism from bulk yield tocrack propagation. This implies that when thedeformation rate is high enough to preventsignificant relaxation processes in the polymer,fibril formation can be suppressed and the tackenergy drops.It should be noted that while the transition fromfibrillar to non-fibrillar fracture appears to be verygeneral, it is not necessarily concomitant with achange from cohesive to adhesive failure. It wasreported to be concomitant for linear polymers18,but not for crosslinked or highly branchedpolymers8,40.

V.1.4. Role of contact time and pressure

The adhesion energy versus contact time is plottedon figure 20a for the case of an acrylic adhesive onsteel40. The variation in W observed withincreasing contact times was initially attributed tothe increase of the real contact area. However morerecent and more detailed results have shown thatthe increase in the adhesion energy was also due toa larger value of εmax as can be seen on Fig 20b40.

19

One can envision therefore two separate effects ofthe contact time. For very short contact times, thereal area of contact may really be affected, givingtherefore a lower value of σmax. However forintermediate contact times, it is the fibrillationprocess which is mainly affected and its effect canbe seen more on W than on σmax.

500

400

300

200

100

0

W (

J/m

2)

0.01 0.1 1 10 100

Contact time (s)

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0121086420

0.02 s0 .1 s 1 s 10 s

100 s

Figure 20: a) W as a function of contact time for aPnBA adhesive on steel; b) Corresponding stress-strain curves for selected contact times. Data from40.

Presumably it is not the real contact area which isimportant in this case but rather the degree ofrelaxation of the adhesive during the contact time.If the relaxation times of the adhesive are of thesame order of magnitude than the time of contact,one expects large differences in the initial stressstate of the adhesive layer depending on the contacttime. These differences could then lead to adifferent response of the adhesive during thedebonding stage as shown on figure 20b. Oneshould note that, in principle one expects W tobecome independent of tc for long contact times.Although this effect has been reported35,experiments do not always show a plateau withinthe limits of the experimentally accessible range oftc of a few hundred seconds.Additionally, in some cases the degree of roughnessof the probe may lead to trapped air andinhomogeneous stress fields which are also boundto vary with the time of contact16. Experimentally,in the regime where relaxation times of theadhesive are important, adhesion is always lowerand the time of contact has always a more marked

effect for rough surfaces than for smoothones40,64,65.In the same way and for the same reasons, a plateauin adhesion energy is observed for large contactpressures35 but few systematic studies have beenpublished for viscoelastic systems where relaxationprocesses during the time of contact can play amajor role.V.2. Effect of geometry

Most of the results which are discussed in thisreview were obtained with flat probe tack tests onthin adhesive films (20-100 µm typically). Whilethis geometry has several advantages for theanalysis of fundamental properties of PSA, it is byno means the only one that can be used and oneshould be careful to understand clearly whatfeatures of a tack curve are due to the material andwhich ones are due to the specific experimentalgeometry.As discussed in section III.2.4. the experimentalgeometry and particularly the degree ofconfinement of the adhesive layer can have animportant effect on the initial failure mechanismswhich are observed.Luckily for users, most PSA have properties ofcritical energy release rate Gc and tensile modulus Ein a range which makes them relatively insensitiveto the degree of confinement. Therefore tests donewith spherical or flat indenters give relativelysimilar results.

However for very soft (liquid-like) and very hard(solid –like) PSA, testing tack with the sphericalprobe, which typically applies a much lesser degreeof confinement to adhesive film, may lead tosignificantly different failure mechanisms32. Threefurther comments should be added concerning therole of the geometry:The role of the geometry will be dominant only inthe first stages of the debonding process since oncea fibrillar structure is formed, the stress-strain curveis essentially representative of parallel tensile testsand εmax at least, should be rather insensitive to theinitial geometry.The stiffness of the experimental apparatus willalso play a role in the deformation processes of theadhesive film during debonding. In probe tests, thestiffness of the apparatus is usually much largerthan the stiffness of the film. However in a peel testthis is no longer true and it is well known thatsignificantly different results can be obtained withdifferent backings. This will also hold in probe testsif a tack test is performed on a PSA on its backing.If the backing is soft and elastic it will act as anadditional reservoir of elastic energy while if it isviscoelastic, it will strongly alter the rate andcontact time dependence of the tack test of theadhesive. Therefore probe tests of adhesives should,whenever possible, be conducted without a backing.

20

Finally, the notion of confinement also applies topeel tests; in this case the confinement level may begiven by the ratio between the width of the tape andthe thickness of the film. Since this ratio is usuallylarge, results on peel tests should be correlated withflat probe tests but experimental confirmation ofthis statement is presently not available in theliterature.

VI. SURFACE EFFECTS

PSA are normally designed to stick to most surfacesand to be therefore rather insensitive to the natureof the surface of the adherent. This is why, in theprevious sections, we have considered mainly therheological properties of the adhesive rather thanthe interfacial chemistry. However based on thetheoretical arguments set forth in section III, thereare two major causes of poor adhesion of a PSA toa surface: low Gc and poor contact.The first case is related to the arguments given insection VI.1: if the interfacial Gc (static but alsodynamic) is too low, once cavities are nucleated atthe interface, they can easily propagate, coalesceand debonding occurs without any fibril formation.The second case is discussed in section VI.2 andoccurs when the elastic modulus of the PSA is toohigh (Dahlquist’s criterion is not met) or when thesurface is too rough for the adhesive to conform toit during the short contact time.Before we proceed, a word of caution is necessaryhere in terms of our use of the parameter Gc. Infracture mechanics, Gc has the meaning of anenergy per unit area necessary for a crack toadvance. For elastic elastomers, Gc is a uniquefunction of the velocity of the advancing crack andcan be determined with a fracture mechanics testsuch as the JKR test25. Unfortunately, theindependent determination of Gc by the samemethod is very difficult for viscoelasticmaterials66,67 since it will depend on the history-dependent degree of relaxation of the adhesive.However as a phenomenological parameterassociated with the amount of energy dissipated bythe advance of a crack front (per unit area), it canbe very useful and simplify the description of themechanisms.

VI.1. Adhesion on low Gc surfaces

Several studies have been undertaken to investigatethe adhesion of model PSA on low energy surfacesor more generally on low Gc surface7,68. Examplesof such surfaces are silicone rubbers, commonlyused for release coatings, polyethylene,

polypropylene and polycarbamates. These earlyresults were somewhat contradictory and did notprovide any explanation of the observeddependence on surface tension. Generally, tackdecreased markedly when the surface energy of theadherent decreased well below that of the adhesive.More recently peel experiments showed that forsoft adhesives, it is not necessarily the surfacetension of the adherent which is the controllingparameter but rather the resistance to shear of theinterface69. The cause for this effect, at themolecular level, remains however a controversialissue so we will focus here on a more macroscopicdescription of the tack experiment of a PSA on alow Gc surface. This description should be kept inmind when interpreting experimental data obtainedon such surfaces.

The essential difference between a low energysurface and a high energy surface can be visualizedthrough the comparative analysis of probe tack testsand has been recently discussed41. For a givenexperimental geometry, the key parametercontrolling the behavior of the adhesive is the ratioof the energy release rate over the modulus Gc/E.Three different cases can be observed as a functionof Gc/E:For high Gc/E, the initial mechanism of failure iscavitation and fibrillation. The maximum extensionof the fibrils is not controlled by the surface but bythe rheological behavior of the adhesive inelongation. This is the standard case for a PSA onsteel or glass.

For intermediate values of Gc/E, the initial failuremechanism can still be cavitation but the maximumextension of the fibrils is controlled by the surface.An example of this situation is given on figure 21:the measured σmax is identical for both surfaces butεmax is very different. A more detailed analysis ofthe debonding mechanisms reveals that theinitiation of failure occurs at about the same levelof stress, but the lateral propagation of the existingdefects is much faster for the low Gc situation. Ifthis lateral propagation is fast enough, it preventsany growth of cavities in the bulk of the adhesiveand therefore the formation of the elongated foamof bridging fibrils. What happens is rather acoalescence of adjoining cracks and globaldebonding of the adhesive film from the probe atrelatively low levels of deformation.

21

0.6

0.5

0.4

0.3

0.2

0.1

0.0

σ (M

Pa)

1086420

ε

Figure 21: Stress-strain curves of a soft adhesive(high and intermediate Gc/E) on steel (solid line)and for a low Gc surface-adherent pair (dashed line).Vdeb = 100 µm/s. Data from 41.

Finally for very low values of Gc/E, the initiationmechanism is no longer cavitation but internalcrack propagation. This can occur at lower levels ofstress so that both σmax and εmax are significantlydecreased. An example of that situation is given onfigure 22. This very different initiation mechanismis also clearly apparent from the video images ofthe debonding process shown on figure 22b.

0.6

0.5

0.4

0.3

0.2

0.1

0.0

σ (M

Pa)

1086420

ε

1 cm

1 cm

500 µm

Figure 22: a) Stress-strain curves of the same hardelastic adhesive on steel (solid line) and on a low Gcsurface-adherent pair (dashed line). b) Videocaptures of the debonding process for both surfaces.Vdeb = 1 µm/s. Note the relatively small cavitiesobserved for the steel surface and the largeirregularly shaped interfacial cracks observed forthe low energy surface. Data from 41.

This description is simple and yet very general. Itcan explain why transitions from interfacial

separation to fibrillation are observed by changingVdeb, the surface roughness or surface chemistry orthe contact time. In each one of these cases, thechange in the experimental parameter had an effectboth on E and on Gc. However the magnitude of thiseffect is generally very different and sometimes inopposite directions resulting in a very large changein Gc/E.

VI.2. Effect of surface roughness

The effect of surface roughness is in many waysmuch more complicated, and it is impossible atpresent to give trends generally valid for all PSA-surface pairs. It has been discussed theoreticallyfirst for perfectly elastic systems70 and then formore viscoelastic systems15. Experiments haveshown consistently that for PSA (unlike for othertypes of adhesives) the presence of a high level ofsurface roughness was detrimental for adhesion.

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

σ σσσ (M

Pa)

1.21.00.80.60.40.20.0εεεε

A

B

C

D

E

0.15

0.10

0.05

0.00

1.21.00.80.6

A B

C D E

Figure 23: Stress-strain curves of an acrylic PSAon rough surfaces with a different averageamplitude Ra of the roughness but with the samewavelength. A : Ra =11 nm ; B : Ra =23nm ;C : Ra=46nm ; D : Ra =114nm ; E : Ra =148nm. Vdeb = 30µm/s; T = 20 °C. Data from 71.

This, it was argued, was due to the limitation of thereal surface of contact due to the asperities14,15. Asdiscussed briefly in section III.1, when an elasticlayer is brought in contact with a rough hardsurface, it tries to comply to the topography of thesurface, but if the balance between the amplitude ofthe roughness and the elastic modulus does notcomply to equation 1, contact may be incomplete.This first simplistic picture is however inconsistentwith the experimental observation that even for lowmodulus PSA, an effect of surface roughness isobserved. This effect is however magnified whenthe PSA has a relatively high elastic modulus.A more realistic view may be that the presence of asurface roughness creates an inhomogeneous strainfield around the surface and creates therefore

22

pockets of residual tensile stress which will becomepreferential nucleation sites for cavities. Recentmore systematic experiments71 have shown that theamplitude of the surface roughness had a directeffect on the level of stress at which the cavitiesappeared as shown on figure 23.On the other hand, experiments with triblock basedadhesives have shown that if nucleation is notaffected, (for example at high temperature), thepropagation of the cracks can be greatly affected bythe roughness65. In this regime, as shown on figure24, σmax is hardly affected but the fibril formationcan be completely suppressed on smooth surfaces,by a rapid lateral propagation of nucleating cavitiesin an analogous way as what is observed for theintermediate Gc/E case discussed in the precedingsection.

0.8

0.6

0.4

0.2

0.0

σ (

MP

a)

43210

ε

Figure 24: Stress-strain curves for an SIS + 50 wt% resin adhesive on a rough (dashed line) and on asmooth (solid line) steel surface44. T = 60°C; Vdeb =100 µm/s.

VII. CONCLUSIONS

In this chapter we have attempted to review, boththe main material properties which are required fora PSA to be tacky and how experimentalparameters affect the practical evaluation oftackiness. Although it must be clear from theprevious sections that the bonding and debondingof a PSA from a surface is a complicated process,some important parameters have been identified:The elastic modulus E’ obtained from small strainoscillatory rheological measurements in the linearviscoelastic regime.The maximum extension of the adhesive in a tensiletest εmax or its elongational rheological properties.The spectrum of relaxation times of the adhesive.

E’ should be in a window of 0.01 to 0.1 MPa.Above that level, proper bonding and fibrilformation are reduced and below that level,viscoelastic dissipation during the debondingprocess will be too low. εmax controls the maximum

extension of the fibrils and therefore plays a majorrole in the measured debonding energy. A relativelysmall εmax is typically desirable for removable PSAwhile a larger value is often characteristic of semi-permanent ones. A reasonable idea of the value ofεmax can in principle be obtained by acharacterization of the adhesive in elongationaldeformation.Finally a broad spectrum of relaxation times isnecessary to ensure both quick bonding (shortrelaxation times) and reasonable fibril stability(long relaxation times) as well as to impart a broadtemperature window of use.Additionally to these important material parametersof the PSA, the surface can also play a role incontrolling tackiness in certain cases. Roughsurfaces can prevent proper bonding and, byforming defects, initiate failure during debonding,both effects reducing tack. Low energy surfaces canalso influence tack by preventing fibril formationand the relevant parameter to predict whether thiswill occur or not is the ratio of the critical energyrelease rate Gc over the elastic modulus E.

23

VIII. FURTHER READING

Hammond, F. H. (1989). Tack. Handbook ofpressure sensitive adhesive technology. D. Satas.New York, Van Nostrand Reinhold. 1: 38-60.

Zosel, A. (1992). “Fracture energy and tack ofpressure sensitive adhesives.” Advances in PressureSensitive Adhesive Technology 1: 92-127.

Creton, C. (1997). Materials Science of Pressure-Sensitive-Adhesives. Processing of Polymers. H. E.H. Meijer. Weinheim, VCH. 18: 707-741.

Gay, C. and L. Leibler (1999). “On Stickiness.”Physics Today: 47-52.

IX. REFERENCES

(1) Chang, E. P. J. Adhes. 1991, 34, 189-200.(2) Chang, E. P. J. Adhes. 1997, 60, 233-248.(3) Tse, M. F.; Jacob, L. J. Adhes. 1996, 56,79-95.(4) Zosel, A. J. Adhes. 1989, 30, 135-149.(5) Urahama, Y. J. Adhes. 1989, 31, 47-58.(6) Hammond, F. H. ASTM Spec. Tech. Publ.1964, 360, 123-134.(7) Zosel, A. Colloid Polym. Sci. 1985, 263,541-553.(8) Lakrout, H.; Sergot, P.; Creton, C. J.Adhes. 1999, 69, 307-359.(9) Tordjeman, P.; Papon, E.; Villenave, J.-J.J.Polym.Sci. B Polym.Phys. 2000, 38, 1201-1208.(10) Chuang, H. K.; Chiu, C.; Paniagua, R.Adh. Age 1997, 18-23.(11) Crosby, A. J.; Shull, K. R. J.Polym.Sci. BPolym.Phys. 1999, 37, 3455-3472.(12) Lakrout, H., PhD. Thesis, Université Paris7, 1998(13) Dahlquist, C. A.Pressure-Sensitiveadhesives; Patrick, R. L., Ed.; Dekker, 1969; Vol. 2,pp 219-260.(14) Creton, C.; Leibler, L. J.Polym.Sci. BPolym.Phys. 1996, 34, 545-554.(15) Hui, C. Y.; Lin, Y. Y.; Baney, J. M.J.Polym.Sci. B Polym.Phys. 2000, 38, 1485-1495.(16) Gay, C.; Leibler.L. Phys. Rev. Lett. 1999,82, 936-939.(17) Gay, C.; Leibler, L. Physics Today 1999,47-52.(18) Lakrout, H.; Creton, C.; Ahn, D.; Shull, K.R., submitted to Macromolecules(19) Gent, A. N.; Lindley, P. B. Proc. Roy. Soc.London, A 1958, 249 A, 195-205.(20) Williams, M. L.; Schapery, R. A. Int. J.Fract. Mech. 1965, 1, 64-71.(21) Chikina, I.; Gay, C. Phys. Rev. Lett. 2000,85, 4546-4549.

(22) Lin, Y. Y.; Hui, C. Y.; Conway, H. D.J.Polym.Sci. B Polym.Phys. 2000, 38, 2769-2784.(23) Creton, C.; Lakrout, H. J.Polym.Sci. BPolym.Phys. 2000, 38, 965-979.(24) Gent, A. N.; Schultz, J. J. Adhes. 1972, 3,281-294.(25) Maugis, D.; Barquins, M. J.Phys.D:Appl.Phys. 1978, 11, 1989-2023.(26) Barquins, M.; Maugis, D. J. Adhes. 1981,13, 53-65.(27) Creton, C.Materials Science of Pressure-Sensitive-Adhesives; 1st ed.; Meijer, H. E. H., Ed.;VCH: Weinheim, 1997; Vol. 18, pp 707-741.(28) Good, R. J.; Gupta, R. K. J. Adhes. 1988,26, 13-36.(29) Baljon, A. R. C.; Robbins, M. O. Science1996, 271, 482-483.(30) McKinley, G. H.; Hassager, O. J. Rheol.1999, 43, 1195-1212.(31) Gersappe, D.; Robbins, M. O. Europhys.Lett. 1999, 48, 150-155.(32) Crosby, A. J.; Shull, K. R.; Lakrout, H.;Creton, C. J.Appl.Phys. 2000, 88, 2956-2966.(33) Ferry, J. D. Viscoelastic Properties ofPolymers; 3rd ed.; Wiley:, 1980; Vol. 1.(34) Doi, M.; Edwards, S. F. The theory ofpolymer dynamics; Oxford:, 1986.(35) Zosel, A. Adv. Pressure Sensitive Adhes.Technol. 1992, 1, 92-127.(36) Ahn, D.; Shull, K. R. Langmuir 1998, 14,3637-3645.(37) Aubrey, D. W.; Ginosatis, S. J. Adhes.1981, 12, 189-198.(38) Chan, H.; Howard, G. J. J. Adhes. 1978, 9,279-301.(39) Krenceski, M. A.; Johnson, J. F. Polym.Eng. Sci. 1989, 29, 36-43.(40) Zosel, A. Int. J. Adhes. Adh. 1998, 18,265-271.(41) Creton, C.; Hooker, J. C.; Shull, K. R.,Langmuir in press(42) Zosel, A. J. Adhes. 1991, 34, 201-209.(43) Webster, I. Int. J. Adhes. Adh. 1999, 19,29-34.(44) Hooker, J. C.; Creton, C. 2000.(45) Flanigan, C. M.; Crosby, A. J.; Shull, K.R. Macromolecules 1999, 32, 7251-7262.(46) Komatsuzaki, S. Tappi Hot MeltSymposium 1999, 171-178.(47) Spence, B. A.; Higgins, J. W. Journal ofAdhesion Sealant Council 1997, 237-255.(48) de Crevoisier, G.; Fabre, P.; Corpart, J.-M.; Leibler, L. Science 1999, 285, 1246-1249.(49) Aymonier-Marçais, A.; Papon, E.;Villenave, J. J.; Tordjeman, P.; Pirri, R.; Gérard, P.,Proceedings of Euradh 2000; Lyon; 2000, pp 170-175.(50) Portigliatti, M.; Hervet, H.; Léger, L. C.R.Acad.Sci. Paris, IV 2000, 1, 1187-1196.

24

(51) Aubrey, D. W.; Sherriff, M. Journal ofPolymer Science, Polymer Chemistry Edition1978, 16, 2631-2643.(52) Aubrey, D. W.; Sherriff, M. Journal ofPolymer Science, Polymer Chemistry Edition1980, 18, 2597-2608.(53) Class, J. B.; Chu, S. G. J. Appl. Polym. Sci.1985, 30, 805-814.(54) Class, J. B.; Chu, S. G. J. Appl. Polym. Sci.1985, 30, 825-842.(55) Class, J. B.; Chu, S. G. J. Appl. Polym. Sci.1985, 30, 815-824.(56) Kim, H. J.; Mizumachi, H. J. Adhes. 1995,49, 113-132.(57) Tancrede, J. M. Journal of AdhesionSealant Council 1998, (Spring), 13-27.(58) Satas, D. Handbook of Pressure SensitiveAdhesive Technology; 2nd ed.; Satas, D., Ed.; VanNostrand Reinhold: New York, 1989; Vol. 1, pp940.(59) Zosel, A.; Schuler, B. J. Adhes. 1999, 70,179-195.(60) Delgado, J.; Kinning, D. J.; Lu, Y. Y.;Tran, T. V., Proceedings of 19th Annual conferenceof the Adhesion Society; 1996, pp 342-345.(61) Gent, A. N.; Petrich, R. P. Proc. Roy. Soc.London, A 1969, 310, 433-448.(62) Benyahia, L.; Verdier, C.; Piau, J. M. J.Adhes. 1997, 62, 45-73.(63) Ondarçuhu, T. J. Phys. II 1997, 7, 1893-1916.(64) Zosel, A. J.Adhesion Sci.Tech. 1997, 11,1447-1457.(65) Hooker, J. C.; Creton, C.; Tordjeman, P.;Shull, K. R., Proceedings of 22nd Annual Meetingof the Adhesion Society; Panama City; 1999, pp415-417.(66) Lin, Y. Y.; Hui, C. Y.; Baney, J. M. J.Phys. D: Appl. Phys. 1999, 32, 2250-2260.(67) Baney J.M.; C.Y., H. J.Appl.Phys. 1999,86, 4232-4241.(68) Toyama, M.; Ito, T.; Nukatsuka, H.; Ikeda,M. Journal of Applied Polymer Science` 1973, 17,3495-3502.(69) Zhang Newby, B.-M.; Chaudhury, M. K.;Brown, H. R. Science 1995, 269, 1407-1409.(70) Fuller, K. N. G.; Tabor, D. Proc. Roy. Soc.London, A 1975, A345, 327-342.(71) Chiche, A.; Pareige, P.; Creton, C. C.R.Acad.Sci. Paris, IV 2000, 1, 1197-1204.

tack - espci paris · 2011. 2. 7. · traditionally, tack properties of a psa have been correlated...

Documents