You are on page 1of 10

How Does Tack Depend on Time of Contact

and Contact Pressure?

COSTANTINO CRETON’* and LUDWIK LElBLER*

’Laboratoire de Physico-Chimie Structurale et Macromol6culaire, URA CNRS 278, ESPCI, 10 Rue Vauquelin, 75231
Paris C6dex 05, France; ’Laboratoire de Physico-Chimie Thborique, URA CNRS 1382, ESPCI, 10 Rue Vauquelin,
75231 Paris C6dex 05, France

SYNOPSIS

The tack of polymer melts on rigid substrates under conditions of short contact times and
low pressures is examined. The substrate is modeled as a random rough surface with a
distribution of asperities heights. The true contact area between the adhesive and the
substrate is calculated for a given total load and elastic modulus of the substrate. The
dependence of tack on contact time is accounted for by introducing the relaxation of the
adhesive through a time-dependent elastic modulus. For relatively high pressures the tack
is predicted to scale with 1/E so that for short contact times, tc, the tack is predicted to
scale with (&/T~)”’, where T, is the entanglement time. For lower pressures this simple
scaling law is no longer valid and we predict a complex variation of tack with contact time
and molecular parameters. 0 1996 John Wiley & Sons, Inc.
Keywords: adhesion pressure-sensitive adhesive contact mechanics theory contact
time pressure tack

INTRODUCTION However, even when debonding conditions are


kept the same, the measured energy of adhesion may
It is a common experience to touch a sticky sub- depend on bonding conditions such as the applied
stance with a finger. When removing the finger, one pressure and time of contact under p r e ~ s u r e .This
~.~
measures tack. The question is then how to quantify typically occurs at low pressures and/or short con-
the feeling of tackiness. And in particular what is a tact times.
good reproducible experiment to measure this effect? We refer to this regime, where adhesion is de-
Clearly two stages have to be distinguished the pendent on these bonding conditions, as “tack.” The
bonding process (applying the finger) and the de- goal of this article is to understand the dependence
bonding stage (removing the finger). of tack on the pressure and contact time and to relate
The relevant parameter to measure tack may be quantitatively our results to the molecular structure
the energy dissipated in the process of debonding. of the adhesive and the roughness of the substrate
This energy can be several orders of magnitude to bring a unifying approach to the interpretation
larger than the thermodynamic work of adhesion, of the experimental results.
W ,resulting from the rupture of the Van der Waals This problem of how tack depends on contact
bonds a t the interface.’-3 For polymer melts and time and contact pressure is important in many ap-
elastomers, the viscoelastic effects or the cavitation plications, namely for the design of pressure-sen-
and fibrillation of the adhesive are responsible for sitive adhesives (PSA). In practice, the tackiness is
a large energy dissipation and strong adhesion. tested by the following procedure: the adhesive film
Hence, this energy is dependent on both the tem- is deposited on a rigid backing and a stainless steel
perature and the debonding rate. probe is brought into contact with the adhesive,
maintained in contact under a constant pressure
* To whom correspondence should he addressed.
(typically i04 Pa) for a given time (typically 1 s),
Journal of Polymer Science: Part B: Polymer Physics, Vol. 34,545-554 (1996)
and finally removed a t a constant velocity. The
0 1996 John Wiley & Sons, Inc. CCC 0887-6266/96/030545- 10 maximum force necessary to debond the adhesive is

545
546 CRETON AND LEIBLER

measured a s a function of the time of contact and


contact pressure. Typical results are shown in Figure loo0 3
l a and b for two different pressure-sensitive adhe-
sives on stainless steeL6
Although such a qualitative behavior has long
been known, it seems that no unified picture has
emerged in the literature about what controls quan-
titatively the contact time and pressure dependence
of t a ~ k . In
~ , certain
~ conditions tack is reportedly
nearly independent of contact pressure whereas in
others it is strongly dependent and the same is true
of the effect of the time of contact. 0.1I I I I I I I
-40 -20 0 20 40 60 80
The short contact times may be the origin of very
subtle effects. As a n illustration, consider the tem- T ("C)
perature dependence of the adhesion energy for two Figure 2. Probe tack results showing the adhesion en-
different times of contact, a very short and very long ergy G under short time, low pressure conditions (for t,
one (Fig. 2). Clearly, for short times of contact, the = 1.6 X lo2 s and p = 8 X lo3 Pa) and the maximum
adhesion energy, G , shows a pronounced maximum, adhesion energy (for t, = 1000 s and high pressure). The
polymer is a polyisobutylene (data from ref. 5).

approximately a t 50°C above the glass transition


temperature of the adhesive. This maximum is sup-
pressed, however, when the time of contact is in-
- creased.
N

.
E
z_
T h e explanations found in the literature t o ac-
m count for the decrease in adhesion a t short contact
E times and low pressures are all based on the same
;;
general argument of the decrease in real area of con-
tact between the adhesive and the probe. The rea-
sons given for this decrease in contact area vary,
however, a t least in the terminology, and one can
0.1 1 10 100 find lack of viscous flow;5 incomplete wetting;' too
time of contact 1s) high elastic modulus of the adhesive.l0." This last
effect also has been evoked for the more complicated
case of self-adhesion of polymers where interdiffu-
sion of chains can occur." In a typical tack experi-
ment, the experimental time, which is of the order
of a second, is much shorter than the terminal re-
laxation time of the polymers used for PSAs so that
there cannot be any substantial amount of viscous
flow.
We propose t o consider the adhesive as a visco-
elastic material with a time-dependent modulus and
t o treat the bonding process in the framework of
contact mechanics. More specifically, we model the
substrate as a random rough surface with asperities
lo3 lo4 lo5
of constant radius of curvature and a distribution
Pressure (Nlm ') of asperity heights relative to a reference plane. This
Figure 1. Probe tack results showing the maximum problem of contact of a random rough surface on a n
stress as a function of the time of contact (a) and contact elastic half-plane has been treated p r e v i ~ u s l y ' ~for
,'~
pressure (b) for two commercial pressure-sensitive ad- the case of adhesion of rubber on a rough substrate
hesives. The contact pressure is lo4Pa in (a) and the time but was never applied to the specific case of the
of contact is 1 s in (b) (data from ref. 6). measurement of tack for which viscoelastic effects
TACK OF POLYMER MELTS 547

large pressures relative to those due to Van der


Waals forces (Hertzian case), and then for smaller
pressures where surface forces become important.
In this latter case we follow closely the seminal work
of Fuller and Tabor.14 Within this framework, it is
t<O O<t<t, t > t,
easy to incorporate the characteristics of the ad-
Figure 3. Schematics of the probe tack experiment. hesive and to examine the role of the contact time
and contact pressure.
are important. In particular, what characterizes the
tack measurement are the short time of contact and
even shorter debonding time. The central idea of LARGE PRESSURES
our approach is that the bond formation between
the viscoelastic half-plane and the rough substrate We first consider the case of large pressures as de-
occurs under conditions where elastic contact me- fined in the preceding paragraph. A further as-
chanics is applicable. The instantaneous area of sumption is that the time of contact between the
contact, A ( p , t,), can then be calculated as a function adhesive and the substrate is much shorter than the
of the time of contact, t,, applied pressure, p , and terminal relaxation time of the adhesive so that no
molecular characteristics of the viscoelastic adhe- viscous flow takes place. Under those conditions,
sive. the process of adhesion can be seen as shown sche-
We argue that this area is the relevant quantity matically in Figure 3.
to be considered for the debonding process because In a first stage, the adhesive film is brought into
no substantial reduction in contact area (dewetting) contact with the substrate at a certain rate. This rate
can occur during the very rapid release of the applied being fast, the film acts as if it was elastic. Then a
pressure. This important statement is based on the constant pressure is maintained for a time t,. During
seminal work of Maugis and Barquins,15 who con- this time under constant pressure the melt relaxes
sidered the case of the contact between an elasto- and the film behaves as if its modulus was lower. The
meric sphere and a plane and who found that the contact area at t = t, is therefore larger than at t =
simple release of the applied pressure does not cause 0 and is governed by the viscoelastic relaxation mod-
the contact area to change very rapidly. Only the ulus, E ( t,) , which could be obtained experimentally
application of a negative pressure will cause rapid in a creep experiment after a time t,.
debonding. Alternatively, this amounts to saying One way to represent the contact between two
that the released elastic energy upon removal of the surfaces is to consider one of the surfaces as a model
pressure is much smaller than the viscoelastic losses rough surface. To calculate the true area of contact
required for debonding at normal operating speeds. between our model rough surface and a flat surface,
Therefore, when comparing tack measurements it is useful to consider first the contact of a single
made at the same debonding rate, we argue that the asperity and then introduce the distribution.
tack should be proportional to the true area of con- If the problem is treated as an elastic problem,
tact established during the bonding step. the area of contact between one asperity and the
In this context, an important issue to discuss per- flat surface can be calculated by contact mechanics
tains to the details of the debonding process. Our and is given by the classic Hertz solution if surface
goal here is to set the framework for a better un- forces are neglected.16
derstanding of the bond formation and its influence As described in Figure 4,when a sphere is brought
on tack in the regime where it is contact limited. In in contact with a flat plane under a force, P , the
our analysis of the influence of the bonding process
we implicitly assume that the debonding conditions P I

are kept constant and that the increase in area of


contact does not have any effect on the occurrence
of cavitation and fibrillation in the adhesive during
debonding nor does it cause a transition from ad-
hesive to cohesive separation. The validity of this
assumption and some further possible developments
will be discussed in the concluding remarks.
The next paragraphs will consider in more detail
-
I 2a I

Figure 4. Schematics of the contact between a spherical


how to calculate this true area of contact, first for cap and a plane.
548 CRETON AND LEIBLER

relationships between the radius of contact, a , the A model for a random rough surface has been
applied force P , and the vertical displacement, 6, proposed by Greenwood and Williamson, l3 who
are given by represented the surface as a series of asperities of
constant radius of curvature, R , but with a distri-
bution of heights relative to a mean plane, and a
density of asperities per unit surface, 2. If a flat
plane is brought in contact with such a surface, when
their two reference planes are separated by a dis-
These results are valid provided the modulus of the
tance, d , there will be contact with any asperity
sphere and the plane are very different so that the
whose height was originally greater than d . The
modulus E in eq. ( 2 ) is that of the more deformable
probability of this contact is given by
material. In our case, we describe in the following
sections the physical picture of the hard asperity r x
indenting in a soft elastic plane. The opposite sit- prob(z > d ) = J cp(z) dz (3)
uation (i.e., a soft asperity on a hard plane) would d

give exactly the same results provided that the mod-


ulus of the asperity was used in the calculations. and the total number of such contacts will be
A real rough surface has a distribution of asper-
ities heights and radii of curvature of the asperities.
The type of the distribution depends on the specific n = N r cp(z) dz (4)
mechanical surface treatment and is discussed in
more detail by Johnson.lfi If the surface has been
obtained by sand-blasting it is nearly random and where N is the total number of asperities.
the distribution of asperities heights follows a The total area of contact can be found by com-
Gaussian distribution, as shown in Figure 5. Al- bining eqs. ( 1) and (4) to give
though further surface treatment such as polishing
will modify the randomness of the surface, it is a
Atot = N R cp(z)(z - d ) dz (5)
good starting point.

200pm

L I I 1
I -4 -2 0 2 4
Height in pm

Figure 5. Schematics of a real rough surface and height distribution of asperities for a
sand-blasted surface of aluminium plotted in a normal probability scale. The y-axis on a
normal probability graph is scaled numerically so that a Gaussian distribution of asperities
will fall on a straight line (data from ref. 16).
TACK OF POLYMER MELTS 549

whereas the total load, PtOt,


is given by combining
eqs. ( 2 ) and (4)

Although the distribution of heights of random


surfaces is better approximated by a Gaussian, the
I I
I
I
I
I
I

'
)

Int
significant results for high pressures can be obtained zt? zd
I

analytically by using an exponential distribution and I I

this approach is chosen here. The distribution cp( z ) Rouse f plateau f terminal
is then given by I

Figure 6. Young's modulus E of a high molecular weight


monodisperse polymer as a function of time in a creep
experiment.

With this distribution, the total area of contact and where N , is the number of monomers in a strand
the total load are given by c0
between two entanglements, denotes the monomer
friction coefficient, and b is the monomer length.
A,,, = naR (8) The disentanglement time T d scales as
P,,, = nEa3/'R'/'. (9)

From these two expressions, one can easily obtain


the relationship between load and area:
with Z being the polymerization index. For short
time scales (t, < T,), the relaxation occurs on short
distance scales and is due to Rouse-like motions. As
a result, the modulus decreases according to a simple
which gives the well-known Greenwood result that power law:
the area of contact and the applied pressure are pro-
portional to each other.13The important result that
should be emphasized in our context is that for a
given surface, A is inversely proportional to the
elastic modulus E and proportional to the applied where EE) denotes the plateau modulus. For inter-
pressure. In our problem of tack, the longer the con- mediate times (7,< t, < T d ) , the modulus practically
tact time, the more relaxed is the PSA and the lower does not change. Relaxation requires the disentan-
its modulus E ( t,) . glement of chains from their original tubes and times
This result can then be used to develop a molec- longer than T d . Thus, only at contact times longer
ular interpretation of the tack properties. The sim- than the disentanglement time T d does the adhesive
plest but instructive case is that of a monodisperse deform by viscous flow so that E decreases again
high molecular weight polymer above its glass tran- with t.
sition temperature. Its modulus E ( t , ) vs. log t, is It is important to consider these molecular ar-
shown schematically in Figure 6. The "softening" guments in the context of a tack experiment. We
of the melt is related to the relaxation of deformed propose that the relevant transition for PSAs is from
chains. The entanglements impose topological con- the Rouse domain to the plateau region and is con-
straints on the motion of chains and control the trolled by 7,. Combining eqs. (10) and (13),a t times
relaxation processes. According to the reptation shorter than T,, the area of contact A,,, is expected
picture, there are two characteristic time scales that to increase as t:/' and to level off when t, > 7,.The
define three different regime~.'~ entanglement time 7,can be measured directly from
The entanglement time, T,, is defined as dynamical mechanical spectroscopy. Furthermore,
the crossover contact time at which tack levels off
should be strongly dependent on temperature as it
scales with the monomer friction coefficient lo.On
550 CRETON AND LEIBLER

the other hand, this crossover time should be in- true area of contact and the load as a function of
dependent of the molecular weight of the chains the indentation. First, the radius of the area of con-
provided they are entangled. tact and the force of contact can no longer be ex-
This result seems consistent with experiments of plicitly expressed as a function of the indentation 6
Zosel on the tack as a function of temperature of but need to be evaluated numerically. Second, in
two samples of polyisobutylene of different molec- this low force regime the results are sensitive to the
ular weight^.^ The critical temperature below which tail end of the distribution of asperities and the more
tack starts to decrease is independent of molecular realistic Gaussian distribution must be used rather
weight, implying therefore that the entanglement than the simpler exponential one.
time T, rather than the terminal relaxation time Td We recall here briefly the approach of Fuller and
is controlling the loss of tack. Tabor.'* The first step in the calculation is to express
A viscous flow explanation would imply, on the 6 and a 3 as a function of P . If a , P , and 6 are nor
other hand, that the transition between good and malized by a,,, Po, and tio,as defined in eqs. ( 14)
bad tack is controlled by the disentanglement time: ( 1 6 ) ,one obtains
a t t, > T d the adhesive flows and tack is good whereas
a t shorter times there is not enough viscous flow t o
ensure a good contact. If this hypothesis was true,
the critical contact time for good tack would have a
strong molecular weight dependence. and

SMALL APPLIED PRESSURES

Clearly, the universal law presented in the preceding


From these analytic expressions, we can extract
section cannot be true when surface forces become
numerically the ratios P / P o andA/Ao as a function
of the order of the external applied force. of (6/6,). This finally gives, in normalized param-
The situation for the contact of a single asperity eters, for the average area and the average force per
when surface forces are important has been analyzed
asperity:
by Johnson et al. in their classic J K R approximation
for the contact of two soft solids." The surface forces
appear as a n apparent additional force on the sur- At''
-- -
N
A, = a; [a( $)
face, which would give the same contact area using
the Hertzian analysis. The J K R approximation gives
in particular a nonzero contact area, even under zero X exp( - 21 ( h + A ) 2

.(
applied force. This area, Ao, and the corresponding
radius of the contact area, ao, are approximately
given by P, Po
- -
ptot
N
- N

$)
X ex,( - 51 ( h + A)'
where W is the thermodynamic work of adhesion.
where h = d / a , A = 6/a, A, = &,/a, 9 is obtained
The corresponding apparent Hertz load is given by

Po - WR (15)
numerically from eq. ( 1 7 ) and @ is obtained nu-
merically by combining eqs. 17 and 18.
The evaluation of the integral (20) a s a function
of A, and h gives the average force P I per asperity
and the resulting indentation is given by for a given normalized separation h between the
plane of the adhesive and the midplane of the rough
substrate. Conversely, the separation h can be ob-
tained for a given average force P1.
Once h is known, the average area of contact per
T h e quantitative analysis of the J K R situation asperity Al for a force P I can then be obtained from
brings two complications in the calculation of the eq. ( 1 9 ) .
TACK OF POLYMER MELTS 551

Within this framework, it is now interesting t o h = 0, E as a Function of PI


examine more quantitatively certain experimental
After considering the case where the applied force
conditions and in particular the transition from the
is zero, it is interesting t o examine how does E*
regime where the true area of contact is controlled
evolve when the applied force is increased. In the
by surface forces and the regime where i t is con-
limiting case of high pressures, the Hertz situation
trolled by the external applied force.
is recovered and as A l cc P, / E , one expects E * t o
increase linearly with the applied pressure with a
h = 0 and P1 = 0 slope proportional t o ( R /cr ) 1'2.
This corresponds to the situation where perfect
contact can be attained between the two surfaces
A , ( € ) for Different Values of P l / W R
under no external applied force. Although this sit-
uation is not realistic, as it assumes t h a t the plane In Figure 7 we consider the case where the contact
is already in contact with the substrate, it is nev- area is calculated as a function of the elastic modulus
ertheless interesting to give a lower bound for the for different values of the nominal force per asperity.
critical modulus for perfect contact. Numerically, T h e average area of contact per asperity Al is nor-
the critical modulus is obtained by determining for malized by A *, defined a s the result of integral ( 19)
which value of A, one obtains complete contact (i.e., for h = 0 (i.e., the maximum area of contact within
h = 0 ) for zero applied pressure ( P1= 0 ) . Replacing the framework of our model). What is apparent from
the critical value of A, = 6,/a in eq. ( 1 7 ) , this gives the curves is t h a t whereas a t zero applied force, the
for the critical modulus E* at P1= 0 decrease of A l with E is nearly exponential, as soon
as PI > 0, the curve must have a limiting behavior
a t high moduli of Al cc 1/ E matching the Hertzian
result. This curve could be used to predict the con-
tact area a s a function of the time of contact t,
This result could also have been obtained in the sin- through the relaxation of the modulus E = E ( t,) .
gle asperity case by a physical argument that pro- In this regime the dependence of the contact area
vides more insight. T h e excess free energy of the on the time of contact will not follow a universal
system due to the indentation 6 of the sphere is law but will depend on the relative importance of
given, under zero applied load, by surface forces and elastic forces, and this is consis-
tent with experimental results that show a variety
AFt,, N -WR6 +E
(Y
- a3

where ( 6 / a ) ' is the elastic deformation that takes


of functional dependencies of the tack with the time
of contact.

place over a volume of the order of a 3 . T h e radius 1


of contact a is roughly equal t o % so that by sub-
stituting a in eq. ( 2 2 ) , minimizing F,,, relative to 6,
and finally setting 6 = (for complete contact on
average), one obtains eq. ( 21 ) . 0.1

It is worthwhile t o do some numerical estimates


of what this critical modulus should be. Taking W
= 50 m J / m 2 , R = 100 pm, and cr = 2 pm typical of
0.01
a machined metal surface, one obtains E* N 0.2
MPa. The physical meaning of this value is to put
a lower bound below which, regardless of the applied
pressure, there will be good contact. It is remarkable
0.001
that for a realistic degree of roughness it is in good
agreement with what is experimentally known a s E/E*
the Dahlquist's criterion ( i.e., the critical modulus Figure 7 . True normalized area of contact AJA* as a
above which there is substantial loss of tack). T h e function of the normalized elastic modulus E/E* for var-
value of E* is, however, very sensitive to the nominal ious values of the average force per asperity P I /WR. The
roughness, which should provide a n easy way t o test relationship between the nominal pressure p and the av-
these conclusions experimentally. erage pressure per asperity PI is given by: p = P,/Rc.
552 CRETON AND LEIBLER

.
-
.,
E
01 L3
15

0.01 2 2
3 4 5 6 7 8 9 3 4 5 6 789 0.001 0.01 0.1 1 10 100
0.1 1 10
P,MR t, (s)
Figure 8. True normalized area of contact AJA* as a Figure 9. Probe tack results showing the adhesion en-
function of the normalized force per asperity P,/ WR for ergy G as a function of the time of contact t, for a nominal
various values of the elastic modulus E/E*. The relation- pressure p = 2 X lo5 Pa (data from ref. 4).
ship between the nominal pressure p and the average
pressure per asperity P1 is given by: p = PJRo.
x W / IJ.Numerically, if we take the same values as
in the previous section, this gives p * = 0.025 MPa
A,(P) for Different Values of €/€* or for a 1 cm2 surface, a force of 2.5N.
The physical interpretation ofp* is the nominal
Another natural extension of the calculation is to
pressure on the sample below which surface forces
investigate the dependence of A l with the applied
become important and must be incorporated in the
force for a range of different moduli. In this case a s
calculation.
well, the limiting behavior a t large values of E/E*
Zose14 measured the tack G a s a function of the
must be Hertzian but when E is closer to E*, the
contact time t, of a sample of polyethylhexyl ac-
slope of the curve of Al/A* decreases and becomes
rylate on a stainless steel probe under a pressure
closer to A l cc Pi’*, as illustrated in Figure 8.
of 0.2 M P a and of the contact pressure p with a
fixed contact time of 1 s. These results, shown in
Figures 9 and 10, allow us t o check for the self-
C O M P A R I S O N WITH EXPERIENCE
consistency of our arguments. In the time-depen-
dence experiment Zosel found that G increased with
After obtaining the trends and scaling laws in the
tEl2a t low values oft, and reached a constant value
different regimes for the true area of contact as a
for t, > 1s. In the pressure-dependence experiment,
function of the applied force and the time of contact,
it is interesting to make some more quantitative
2
comparisons with experimental results to test the 0 1
validity of the model. The difficulty lies in the lack
of quantitative simultaneous measurements of both
tackiness and roughness. We must therefore rely on
estimates of roughnesses for typical probes and ad-
hesives. First, we need to estimate the areal density
of asperities. The main results contained in eqs. ( 19)
and ( 2 0 ) give the average area per asperity A l and
the average force per asperity P I as a function of a
mean separation between the adhesive and the sub-
strate. If the average radius R and the mean asperity
height u are known, the areal density of asperities 10
0.01 0.1 1 10
L: is not a n independent variable but can be esti-
mated roughly. The average diameter of a n asperity p (MPa)
is of the order of 6 so that the areal density of Figure 10. Probe tack results showing the adhesion
asperities is of the order of 1/Ra. Therefore, a n av- energy G as a function of the contact pressure for a contact
erage force P, = WR corresponds to a pressure p* time of 1 s (data from ref. 4).
TACK OF POLYMER MELTS 553

he found an increase of G with p1/2and a leveling


off f o r p > 0.3 MPa.
If we take p* = 0.025 MPa as estimated above,
lo 2 loo0

the experimental contact pressure of Zosel, p = 0.2


MPa, is about eight times greater than p * . We can n
then extract the predicted functional dependence of
the contact area with E from Figure 7 with PI
= 8 W R . In this regime we expect the area to vary
with 1 / E and therefore, from the arguments pro-
posed in the Large Pressures Section, the tack
should increase with t E / 2 , in very good agreement
with experimental data.
The question is whether the time scale at which T (“C)
the saturation occurs is consistent with our predic-
Figure 11. Probe tack adhesion energy G (0)and elastic
tions. As we do not have precise information on the compliance 1 / E (-) for polyisobutylene as a function of
frequency dependence of the modulus, we cannot temperature. In the contact-limited regime, G cc 1 / E and
compare directly the saturation time and 7,. How- when 1/E reaches a plateau, the tack is controlled by
ever, a value of 7, of about l s seems quite plausible the viscoelastic losses and decreases with T (data from
for poly (ethylhexyl acrylate) at room temperature. ref. 10).
The pressure dependence of tack, as shown in
Figure 10, is harder to compare quantitatively to our
that we are in a regime where the modulus is much
model in the absence of detailed roughness data. We
higher than E* and the area of contact decreases as
have shown that the apparent exponent of the tack
1 / E , as predicted by our model.
dependence on pressure varies with the ratio E / E * .
This critical modulus E* depends itself both on the
average height ~7and on the average radius of the
asperities R . CONCLUDING REMARKS
The experimentally determined exponent N 3
would correspond from Figure 8 to E / E * N 2.5. Our model is in very good qualitative agreement with
Therefore, the tack should become independent of the available experimental results. In particular, the
pressure at a pressure higher than p c N 3 W / u. Typ- scaling laws successfully predict all the trends of the
ically, for a value of csof a few microns, p c should be influence of the pressure and of the time of contact
of the order of 104-105 Pa, which is slightly lower in the regimes where adhesion is contact limited.
than the experimentally determined value from Fig- Roughness effects seem to be qualitatively well ac-
ure 10. counted for in our picture as well. The more quan-
Another confirmation of the influence of applied titative comparisons based on molecular arguments
pressure, Hammond’s results in Figure Ib show ex- would require simultaneous measurements of stress
ponents from 0.7 to 0.4 when the adhesive goes from relaxation and tack on well-characterized substrates
rough to smooth. These results should be compared and adhesives. In particular, it will be interesting to
with Figure 8 showing the effect of the contact pres- study how the temperature does affect the pressure
sure at a given modulus and are consistent with a and time dependence of tack.
decrease in the exponent when the contact surface We assumed in our analysis that the measured
is made smoother. tack (force or energy) was directly proportional to
A different experimental observation, which the true area of contact. This assumption can only
reinforces the present analysis, is the dependence be valid if two conditions are fulfilled:
of the measured tack on temperature. Experimen-
tally, tack goes through a maximum with tempera- 1. the debonding conditions, such as tempera-
t ~ r e The
. ~ high temperature side of the maximum ture and debonding rate, are kept constant,
is controlled by the viscoelastic losses whereas the 2. the variation in contact area does not affect
low temperature side may be controlled by the area the debonding mechanism.
of contact. Zosel measured, on the same systems,
the tack and the elastic modulus at a time corre- Although the first condition is easy to fulfill by a
sponding to the time of contact t,. His results, shown proper choice of experimental conditions, the second
in Figure 11, are consistent with W cc 1 / E , implying one implies that there will be no transition from
554 CRETON AND LEIBLER

adhesive to cohesive failure or from nonfibrillar to REFERENCES A ND N O T E S


fibrillar fracture when the area of contact is in-
creased. Although we do not have observations con- 1. E. H. Andrews and A. J . Kinloch, Proc. R. SOC.Lond.,
firming these points, the available experimental re- A, 332, 385 (1973).
2. E. H. Andrews and A. J. Kinloch, Proc. R. SOC.Land.,
sults are consistent with our hypothesis that the
A , 3 3 2 , 4 0 1 (1973).
measured tack is proportional to the contact area
3. A. N. Gent and J. Schultz, J . Adhes., 3, 281 (1972).
and do not show evidence of a sharp transition in 4. A. Zosel, Adu. Pressure Sensitive Adhes. Technol., 1,
the tack force or energy a t a given value of the area 92 (1992).
of contact. 5. A. Zosel, Colloid Polym. Sci., 263, 541 (1985).
I t is well known that many PSAs cavitate and 6. F. H. Hammond, A S T M Spec. Tech. Publ., 360,123
form fibrils when they are debonded from the ad- ( 1964). Although the standard experiment of probe
hesive. The detailed process of fibrillation is likely tack, as defined by the ASTM measures a force, one
to depend on the initial contact area but is not nec- can wonder if the energy of debonding would not be
essarily proportional t o it. However, the fibrillar de- a more relevant parameter than the maximum force
bonding normally occurs in a regime where the necessary to debond. Still, when comparing different
adhesives, both maximum force and dissipated energy
adhesion is not contact limited but rather controlled
measurements give similar relative results as long as
by the amount of viscoelastic losses taking place
the debonding mechanisms are the same.
during the large deformation involved in the fibril- 7. A. N. Gent and H. J. Kim, Rubber Chem. Tech., 63,
lation process. 613 (1990).
We assumed implicitly that the specific chemical 8. P. G. de Gennes, C. R. Acad. Sci. Paris, 11, 312, 1415
interactions between the adhesive and the substrate, (1991).
which would affect the thermodynamic energy of 9. M. Toyama, T. Ito, and H. Moriguchi, J . Appl. Polym.
adhesion W, were independent of the time of contact Sci., 14, 2039 (1970).
and pressure, Although this is not true for many 10. C. A. Dahlquist, in Treatise on Adhesion and Adhe-
pressure-sensitive adhesives, the increase in adhe- sives, R. L. Patrick (ed.), Dekker, New York, 1969,2,
sion due to W is generally occurring over a much 219.
11. G. Kraus, K. W. Rollmann, and R. A. Gray, J. Adhes.,
longer time scale than the viscoelastic effects in-
10, 221 ( 1979).
vestigated in the present model so that we feel that
12. J. N. AnandandH. J. Karam, J. Adhes., 1,16 (1969).
it is a reasonable assumption to take W as a con- 13. J. A. Greenwood and J. B. P. Williamson, Proc. R.
stant. SOC.Lond., A , A295, 300 (1966).
As pointed out in the Introduction, the removal 14. K. N. G. Fuller and D. Tabor, Proc. R. SOC.Lond., A ,
of the force upon debonding is extremely fast in a A345,327 (1975).
typical tack experiment. When a negative pressure 15. D. Maugis and M. Barquins, J . Phys. D: Appl. Phys.,
is applied and the two surfaces are still in contact, 11, 1989 (1978).
the debonding process can be represented as the si- 16. K. L. Johnson, Contact Mechanics, 1st ed., Cambridge
multaneous propagation of small circular cracks University Press, Cambridge, 1985.
around each asperity a t a certain speed. The multiple 17. M. Doi and S. F. Edwards, The Theory of Polymer
Dynamics, Oxford University Press, Oxford, 1986.
crack picture had already been suggested by de
18. K. L. Johnson, K. Kendall, and A. D. Roberts, Proc.
Gennes' on a flat surface. T o treat the problem of R. SOC.Lond., A , A324, 301 (1971).
debonding quantitatively, it seems necessary to 19. C. Creton, H. R. Brown, and K. R. Shull, Macromol-
consider a rough surface and to incorporate existing ecules, 27, 3174 (1994).
experimental ' 'J'and results on crack 20. P. G. de Gennes, C. R. Acad. Sci. Paris, II, 307,1949
propagation in viscoelastic media. ( 1988).
21. C. Y . Hui, D. B. Xu, a n d E . J. Kramer, J . Appl. Phys.,
We are grateful to Denis Melot and Michel Glotin (Elf- 72,3294 (1992).
ATOCHEM) for stimulating discussions on PSAs.
Received April 27, 1995
Revised September 22, 1995
Accepted September 27, 1995

You might also like