6966 Langmuir 2007, 23, 6966-6974

Adhesion Enhancement through Micropatterning at

Polydimethylsiloxane-Acrylic Adhesive Interfaces
M. Lamblet,‡,⊥ E. Verneuil,†,§ T. Vilmin,‡,| A. Buguin,† P. Silberzan,† and L. Léger*,‡
Physico-Chimie Curie, UMR 168, CNRS - Institut Curie, 11 rue Pierre et Marie Curie, 75231 Paris,
Cedex 05, France, and Laboratoire de Physique des Fluides Organisés, FRE 2844, CNRS - Collège de
France, 11 Place Marcelin Berthelot, 75231 Paris, Cedex 05, France

ReceiVed October 23, 2006. In Final Form: February 16, 2007

Adhesion at polydimethylsiloxane (PDMS)-acrylic adhesive interfaces is shown to be enhanced through

micropatterning of the PDMS substrate. By varying the geometry of the patterns (groves and hexagonal arrays of pillars
of micrometer sizes, obtained through soft lithography techniques) and comparing rigid and deformable substrates,
the respective roles of the geometry and the size and flexibility of the pattern features on the level of adhesion have
been analyzed. For cylindrical pillars, two regimes are clearly identified: for a relatively low aspect ratio (h/r < 3,
with h and r, respectively, the height and the radius of the pillars), soft patterned substrates are more efficient than
rigid ones at increasing adhesion, pointing out the role of the elastic energy associated with the deformation of the
pattern that is lost when the adhesive detaches from the substrate. Using scaling laws, the predominant contribution
to that elastic energy can be further identified: deformation of the substrate underlying the pillars for h/r < 1.6 or
bending of the pillars for h/r > 1.6.; for a high aspect ratio (h/r > 3), only rigid patterned substrates enhance adhesion,
then the only possible contribution to energy dissipation comes from the enhanced viscoelastic losses associated with
the pattern that induce modifications of the strain field within the adhesive layer. Soft, high aspect ratio patterns lose
their efficiency even if still bent under the effect of the peel forces. This is because when bent, some of the pillars
touch each other and remain stuck together, lying flat on the surface after the passage of the peel front. The bending
elastic energy of the pillars (which is still lost) is then balanced by the corresponding gain in surface energy of the
substrate in the peeled region. These systematic experiments demonstrate that the ability of the patterned surface to
be deformed plays a crucial role in enhancing adhesion and allow us to propose a way to fine tune the level of adhesion
at PDMS-acrylic adhesive interfaces, independently of the chemistry of the adhesive.

Introduction inspired by the amazing aptitude of some insects and lizards,

The precise tuning of the adhesion properties is of great interest
for a number of applications, especially when one deals with
weak adhesion. Substrates made of polydimethylsiloxane (PDMS)
have many advantages as antiadhesive coatings because of their
low surface energy and their weak chemical reactivity. Their
adhesion with acrylic tapes, however, is too low for most usual
applications. Therefore the design of PDMS surfaces with tailored
adhesive properties remains a real technical challenge. A
commonly followed path relies on chemical modification1-3 of
the antiadhesive coating, which usually leads to an adhesion
enhancement that is strongly dependent on the chemical nature
of the adhesive. An alternative and potentially more universal
solution based on microstructuration has been proposed and has
started to be investigated recently by several authors.4-7 It is
* Corresponding author. Current address: Laboratoire de Physique des
Solides, UMR 8502 CNRS - Université Paris Sud, Bâtiment 510, 91405
Orsay, France. E-mail: leger@lps.u-psud.fr. Tel: + (33) 1 69 15 56 45. Fax:
+ (33) 69 15 60 86.
† CNRS - Institut Curie.
§ Current address: Fluides, Automatique et Systèmes Thermiques,
Bâtiment 502, 91405 Orsay, France.
| Current address: Laboratoire de Physico-Chimie Théorique, ESPCI, 10
rue Vauquelin, 75231 Paris, Cedex 05, France.
‡ CNRS - Collège de France.
⊥ Current address: Physico-Chimie Curie, UMR 168, CNRS-Institut
Curie, 11 rue Pierre et Marie Curie, 75231 Paris, Cedex 05, France.
(1) Owen, M. J.; Jones, J. D. Silicone Release Coating. In The Polymeric
Materials Encyclopedia; Salamone, J. C., Ed.; CRC Press: Boca Raton, FL,
1996.
(2) Amouroux, N.; Petit, J.; Léger, L. Langmuir 2001, 17, 6510.
(3) Gordon, G. V.; Schmidt, R. G. J. Adhes. 2000, 72, 133.
(4) Gathak, A.; Mahadevan, L.; Chung, J. Y.; Chaudhury, M. K.; Shenoy, V.
Proc. R. Soc. London, Ser. A 2004, 460, 2725.
such as geckos, to adhere reversibly to many substrates using
micro and nanoscale high aspect ratio structures located on the
tips of their fingers, usually referred to as fibrillar structures or
fibrils (or setae).8-10 For geckos, the dominant mechanism is
still under debate. This lizard exhibits a dry adhesion system,
and both van der Waals forces9 and capillary forces10 have
been discussed. The ability of geckos to climb on various
substrates is then supposed to be governed by a complex interplay
between the mechanical properties of the fibrils, their hydro-
philicity, and their geometry. More recently, the rapid
switching between gecko foot attachment and detachment has
been analyzed theoretically on the basis of a tape model taking
into account the geometry of the fibrillar structure and the
macroscopic action of the gecko toes.11 Thus, we have a lot to
learn from nature, and fibrillar structures seem to be good
candidates to tune adhesion. This assumption has started to be
systematically studied, both theoretically and experimentally, in
the past few years.
(5) Geim, A. K.; Dubonov, S. V.; Grigorieva, I. V.; Novoselov, K. S.; Zhukov,
A. A.; Shapoval, S. Y. Nat. Mater. 2003, 2, 461.
(6) Peressadko, A.; Gorb, S. N. J. Adhes. 2004, 80, 247.
(7) Hui, C. Y; Glassmaker, N. J.; Tang, T.; Jagota, A. J. R. Soc. Interface 2004,
10.1098.
(8) Arzt, E.; Gorb, S.; Spolenak, R. Proc. Natl. Acad. Sci. U.S.A. 2003, 16,
10603.
(9) Autumm, K.; Yiching, A.; Liang, S.; Hsieh, T.; Zesch, W.; Chan, W. P.;
Kenny, T. W.; Fearing, R.; Full, R. J. Nature 2000, 405, 681.
(10) Huber, G.; Mantz, H.; Spolenak, R.; Meckr, K.; Jacobs, K.; Gorb, S. N.;
Arzt, E. Proc. Natl. Acad. Sci. U.S.A. 2005, 45, 16293.
(11) Tian, Y.; Pesika, N.; Zeng, H.; Rosenberg, K.; Zhao, B.; McGuiggan, P.;
Autumn, K.; Israelachvili, J. Proc. Natl. Acad. Sci. U.S.A. 2006, 51, 19320.
(12) Autumm, K.; Sitti, M.; Liang, Y. A.; Peattie, A. M.; Hansen, W. R.;
Sponberg, S.; Kenny, T. W.; Fearing, R., Israelachvili, J. N.; Full, R. J. Proc. Natl.
Acad. Sci. U.S.A. 2002, 99, 12252.

10.1021/la063104h CCC: $37.00 © 2007 American Chemical Society

Published on Web 05/19/2007
Adhesion Enhancement through Micropatterning
From a theoretical point of view, Arzt et al.8 have shown,
using the Johnson-Kendall-Roberts (JKR) theory of adhesive
contacts, that splitting up one contact into many smaller
subcontacts increases adhesion. Persson et al.13,14 have shown
that the effective elastic modulus of a fibrillar structure is much
smaller than that of the corresponding bulk material. As a result,
a fibrillar structure is expected to be highly deformable, which
should help in forming good contact. This is, of course, of
fundamental importance for adhesion on both smooth and rough
substrates. Later, Jagota et al.15 have shown that the work required
to separate a fibrillar structure from a substrate is larger than that
of the same smooth material because the elastic strain energy
stored in the fibrils, when they deform, is lost during pull-off.
Hui et al.,7 for their part, have shown that in the case of fibrillar
structures the stress concentration at the crack tip is redistributed
over a zone described by a characteristic length that is significantly
greater than the cross-sectional dimensions of the fibrils. Within
this zone, the fibrils are under equal load-sharing conditions.
Consequently, the failure of the interface involves a simultaneous
failure of all fibrils inside this zone, which is quite different from
the usual crack propagation for which stress concentration favors
a sequential failure of fibrils starting with the fibrils closest to
the crack tip.
Recently, a number of experiments have been performed to
show how fibrillar structures could enhance adhesion. For
example, Ghatak et al.4 have studied the adhesion properties
between an incision-patterned PDMS elastomer layer and a
flexible plate, pulling the plate to initiate a crack at the plate-
elastomer interface. The crack propagates in an intermittent
manner. The authors mentioned that multiple crack arrest and
initiation on such substrates should result in extra dissipation of
the elastic energy as observed for the fracture of soft elastomers.18
If two-dimensional textured surfaces (similar to a chocolate bar)
are used and if the length scale of these patterns is small enough
(typically smaller than the decay length of the stress applied to
the elastomer), then a large enhancement of the interfacial fracture
toughness is observed. Geim et al.,5 using substrates made of
polyimide hairs posts (micrometer size and interdistance to mimic
Gecko finger tips) supported by double-stick viscoelastic tape,
measured high pull-off forces. Similar experiments have been
performed and analyzed by Hui et al.7 They established that
without the double-stick viscoelastic tape the adhesion is not
increased by the fibrillar structure, suggesting that the enhanced
adhesion measured by Geim et al.5 was due to dissipation in the
viscoelastic layer. Peressadko et al.6 have also shown, through
tack experiments, that the tenacity (force per actual contact area)
of a structured surface made of polyvinylsiloxane was higher
than that of an unstructured one. Finally, Crosby et al.19 have
demonstrated through JKR experiments that adhesion between
glass and PDMS substrate patterned by low aspect ratio cylindrical
posts could be altered from 20 to 400% of the value of
conventional adhesion descriptors for nonpatterned interfaces.
Different local separation processes at the interface were observed,
and general relationships between material properties, pattern
length scales, and adhesion were established, depending on the
characteristic sizes of the array (typical value are a few
micrometers for the height and 20 to 500 µm for the post radius
and edge-to-edge spacing).
(13) Persson, B. N. J.; Gorb, S. J. Chem. Phys. 2003, 119, 11437.
(14) Persson, B. N. J. J. Chem. Phys. 2003, 118, 7614.
(15) Jagota, A.; Bennison, S. J. Integr. Comp. Biol. 2002, 42, 1140.
(16) Hui, C. Y.; Glassmaker, N. J.; Jagota, A. J. Adhes. 2005, 81, 699.
(17) Glassmaker, N. J.; Jagota, A.; Hui, C. Y; Kim, J. J. R. Soc. Interface 2004,
10.1098
(18) Lake, G. J.; Thomas, A. G. Proc. R. Soc. London, Ser. A 1967, 300, 108.
(19) Crosby, A. J.; Hageman, M.; Duncan, A. Langmuir 2005, 21, 11738.
Langmuir, Vol. 23, No. 13, 2007 6967
All of these findings are indeed along the lines of early
experiments from Fuller and Tabor,2020 who measured the rolling
resistance of unvulcanized rubber on a rigid substrate. They
observed that adhesion was enhanced on rough substrates, a
result interpreted using two major assumptions:
- The viscoelastic properties of the rubber are essential to
forming an intimate contact with a rough surface (as a result of
stress relaxation).
- The substrate roughness leads to the formation of isolated
contact regions during peeling. The final stages of separation
involve only isolated still-adhering zones whose associated elastic
energy, built up during peeling, is lost when the contact is broken.
It is thus now well known that fibrillar structures are involved
in the formation of intimate contact with a rough substrates
because they are more deformable than bulk smooth material.
Moreover, different mechanisms of energy dissipation can lead
to enhanced adhesion on fibrillar substrates either because of the
stretching and deformation of the fibrils themselves or because
of an overall modification of the stress field both inside the
fibrillar zone and in the underlying material. These arguments,
however, still remain qualitative, and the relative balance between
the different contributions still has to be elucidated.
In this article, we investigate the adhesion properties between
micropatterned PDMS elastomer surfaces and a commercial
acrylic adhesive tape. We have varied the shape and the
geometrical characteristics of the pattern in order to gain deeper
fundamental insight into the role of patterning on adhesion by
identifying the relevant parameters of the pattern that control the
modulation of adhesion. To uncouple the effects on adhesion of
the deformation of the patterned substrate from the modification
of the dissipation within the acrylic adhesive associated with the
modification of the shape of the interface due to patterning,
comparative experiments on rigid polyurethane patterned sub-
strates with the same geometrical characteristics as the soft PDMS
elastomer patterned surfaces have been developed.
Experimental Section
Fabrication of the Patterned Substrates. Patterned PDMS
substrates, either lamellae or micropillars, were produced by classical
molding techniques21 using an etched silicon wafer as a mold. This
mold was obtained with standard photolithographic techniques: a
thin layer of positive photoresist (AZ 5206, Clariant) spin coated
onto a silicon wafer was exposed to UV light through a quartz/
chrome photomask (Compugraphics) decorated with the desired
pattern. After development, the bare parts of the wafer (corresponding
to dots or lines) were etched by deep reactive ion etching (DRIE).
The remaining photoresist was then removed to give a bare silicon
wafer carved with an array of patterns with various depths and shapes
(cylindrical holes or grooves). This silicon wafer was silanized (after
plasma activation) with tridecafluoro-trichlorosilane in the vapor
phase to ease the next demolding step. PDMS replicas were obtained
by pouring a 1 mm thick layer of liquid silicone PDMS prepolymer
(Sylgard 184, Dow Corning), curing at 65 °C for 24 h, and finally
peeling off the crosslinked PDMS elastomer from the mold. Rigid
patterned substrates were made with a UV photocurable polyurethane-
based adhesive (type J-91, Summer Optical). They were prepared
by pouring this optical cement on silanized, patterned PDMS
elastomers replicas and by peeling off the PDMS mold after exposure
to UV light (10 min) to cross link the cement. The patterned PDMS
elastomer replicas of the initial wafer were obtained by first making
a negative replica of the wafer using the method previously described.
This process was repeated a second time to yield an exact PDMS
copy of the initial wafer. To adjust the surface properties of the
cement, in particular, to target adhesive properties versus the acrylic
(20) Fuller, K. N. G.; Tabor, D. Proc. R. Soc. London, Ser. A 1975, 345, 327.
(21) Xiaa, Y.; Whitdesides, G. M. Angew. Chem., Int. Ed. 1998, 550.
6968 Langmuir, Vol. 23, No. 13, 2007
Figure 1. Two examples of PDMS substrates patterned with micropillars (SEM images). The pillar diameter is about 2 µm, the spatial period
is 4 µm, and the heights are 0.5 µm (left) and 3.8 µm (right). Both scale bars are 2 µm.
Figure 2. AFM image of a PDMS surface patterned with
micropillars. The measured diameter is about 2 µm, the period is
4 µm, and the height is 0.3 µm.
Table 1. Height of the Pillars on Patterned Substrates Made of
PDMS or Optical Cement
height (µm) of height (µm) of
optical cement PDMS
0.5 0,3
0.8 0.7
1.6 1.8
3.3 3.3
3.8 3.8
tape similar to those of the PDMS elastomer, the rigid patterned
substrate was further oxidized in an oxygen plasma, immersed in
a PDMS polymer (20 000 g/mol, I ) 1.3), held at 80 °C for 12 h,
and finally rinsed with toluene to remove all PDMS chains not
attached to the surface. The thickness of the adsorbed PDMS layer
was then about 10 nm thick (as estimated to be similar to what is
measured by ellipsometry for the same PDMS chains adsorbed on
a silicon wafer).
All analyzed substrates (PDMS or polyurethane) consisted of two
or three microtextured squares (1 cm × 1 cm) separated by a smooth
nonpatterned region of the same material (width 0.5 cm) and supported
by a microscope slide.
Characterization of the Patterned Elastomers. The pillar
patterns have been characterized either by scanning electronic
microscopy, SEM (Figure 1), or atomic force microscopy, AFM
(Figure 2). The characteristics of the patterns, pillar radius r, spatial
period i (hexagonal pattern), and height h are given in Table 1. In
all experiments presented below, r and i were kept constant and
equal, respectively, to 1 and 4 µm, and the aspect ratio h/r was varied
from 0.3 to 3.8 by varying the pillar height.
Lamblet et al.
Table 2. Sizes of the PDMS Lamellaea
sample width (µm) period (µm) height (µm)
1 4.5 15 7.5
2 4.5 25 7.5
3 20 25 7.5
4 20 30 7.5
a
The exact height was measured by optical microscopy.
The PDMS elastomer patterns with lamellae have been character-
ized by optical microscopy, imaging a section of the sample normal
to the lamellae. Their dimensions (height and width) and their spatial
period are given in Table 2.
Acrylic Adhesive. The acrylic adhesive was a commercial tape
(3M 600). Its width b is 19 mm. The thickness of the adhesive layer
is 17 µm, and that of the backing is 40 µm.
Material Properties. The storage modulus E′ and the loss modulus
E′′ of all materials have been measured using dynamic shear
experiments at 24 °C and at a frequency of 0.1 Hz. For the PDMS
elastomers, E′ ) 2 MPa and E′′ ) 0.07 MPa. The adhesive tape is
much softer with E′ ) 0.02 MPa and E′′) 0.005 MPa. The optical
cement is much harder, with a Young’s modulus of 1.6 GPa (Summer
Optical information).
Peel Test. The peel force F has been measured with a homemade
apparatus (Figure 1 in Supporting Information), which allows us to
peel a sample in 90° geometry at a chosen velocity V (ranging from
500 nm/s to 1 mm/s) and to visualize the peel front through an
inverted optical microscope. The sample was illuminated perpen-
dicularly to its plane with an optical fiber. The peel energy G was
deduced from the peel force F by the well-known relation G ) F/b,
with b being the width of the tape. Prior to each experiment, the
adhesive tape was put into contact with the substrate, and a load
corresponding to a pressure of 5 × 104 Pa was applied to the assembly
for 24 h to ensure that the adhesive fills the whole space between
the micropatterns (which can be easily checked optically because
of the important difference in the refractive index between air and
adhesive). To obtain a direct estimate of the effect of patterning on
the adhesion energy, whatever the possible fluctuations of the peel
energy on the corresponding nonpatterned substrate (which may
result from small differences in the curing or demolding process),
the peel forces on both the patterned part Fp and on the smooth part
Fb of the substrate have been systematically measured on each sample.
The error bar on the G values presented here is about 15%.
Results
PDMS Lamellae. The simplest geometry tested consists of
a regular array of equally spaced linear rectangular PDMS
Adhesion Enhancement through Micropatterning Langmuir, Vol. 23, No. 13, 2007 6969

Figure 3. Image of a PDMS substrate patterned by lamellae obtained

by optical microscopy (300 µm × 230 µm).

Figure 5. Variation of the logarithm of peel energy G with the log-

arithm of peel velocity V for an acrylic adhesive peeled off a PDMS
elastomer patterned with lamellae of different geometries (Table 1).
Values for the smooth, bare elastomer are also presented (9).

Figure 4. Peel front (observed from the setup described in Figure

1 in Supporting Information) advancing from bottom to top as the
adhesive is peeled off a PDMS elastomer patterned with regularly
spaced lamellae at a peel velocity of 1 µm/s. The lamellae width is
about 4.5 µm, the height is 7.5 µm, and the period is 15 µm. The
scale bar is 50 µm. As the peel front passes, the lines deform, and
bubbles (shown by the arrow) are nucleated in the middle of the
fingerlike instabilities.
lamellae. The adhesive is peeled off of the sample with the peel
front parallel to the lamellae (Figure 4). One can observe two
main features: First, as the peel front passes, the lamellae deform
and bend locally to follow the peel front, Second, the detachment
of the adhesive proceeds through the nucleation of bubbles in
the middle of the fingerlike instabilities of the peel front. Once
the peel experiment is over, the whole pattern recovers its original
shape and remains undamaged. As shown in Figure 5, the peel
energy G is slightly increased on the patterned substrates as Figure 6. Enhancement of peel energy ∆G ) Gp - Gb with peel
compared to the smooth elastomer substrates (solid squares). velocity V for different patterns of lamellae (Table 1).
Moreover, its variation with peel velocity V is different for both
types of substrates. G appears to be independent of V for the PDMS Micropillars. Data Analysis. Figure 7 shows the peel
smooth elastomer, whereas G increases with V on the patterned force as a function of the peel front displacement while peeling
substrates. As one can clearly see in Figure 6, the enhancement the adhesive on a PDMS elastomer with two regions: one is
in G, as estimated through ∆G ) Fp/b - Fb/b, is larger for smal- patterned with micropillars (right side), and the other is smooth
ler periods or smaller widths of the lamellae (patterns 1-3). (left side). The peel force is significantly increased on the patterned
When the adhesive is peeled off with the peel front perpendicular part of the substrate (0.12 N) compared to that on the smooth
to the lines, then the increase in peel energy appears to be much one (0.06 N). Because the adhesive has invaded the channels
smaller and the lamellae are barely deformed close to the peel between the pillars, the real area of contact is larger than the
front (not shown). Of course, because of the patterning, the true apparent one. Knowing the dimensions of the pillars, this surface
surface of the interface is increased in the patterned region com- enhancement can easily be estimated. We assume that the pillars
pared to that in the smooth one, automatically leading to an ap- are perfect cylinders and that the pattern remains nondeformed
parent increase in adhesion if G is deduced through G ) F/b. inside the contact
Knowing the geometry of the pattern, one can easily correct for
that increased area of contact. When this is performed for lamellar S - S0 ∆S 4πrh
patterns, the remaining increase in adhesion hardly overcomes ) )
the error bar in the measurement of the peel energy. Visual obser- S0 S x3i2
vation of the lamellar patterns is thus useful to give trends (the
more deformable the lamellae, the higher the peel force) but is where S is the contact area on the patterned elastomer, S0 is the
not efficient enough to allow for a detailed quantitative inves- contact area on the smooth elastomer, r is the radius of the pillars,
tigation. h is their height, and i is the period of the pillar lattice. Assuming
6970 Langmuir, Vol. 23, No. 13, 2007 Lamblet et al.

Figure 7. Peel force versus peel front displacement when peeling
the adhesive tape on a PDMS sample at 6 µm/s. One can distinguish
two parts in the plot: the left side with a low value of the peel force
corresponding to the smooth elastomer and the right side with a
higher value of the force corresponding to the elastomer patterned
with pillars (height 0.7 µm).
Figure 8. Variation of the logarithm of peel energy G with the
logarithm of peel velocity V while peeling elastomer samples
patterned with micropillars of different heights. Values for the bare
elastomer are also represented.

a peel energy that is similar on both patterned and smooth

elastomers, the peel force Fex expected on the patterned elastomer
(and enhanced compared to that on the smooth substrate as a
result of the increased contact area) can be estimated from the
value of Fb measured on a smooth elastomer:

Fex ) Fb ( ∆S
S0
+1 )
However, this estimated Fex value (0.08 N) is much lower than
the peel force actually measured on the patterned elastomer (0.12
N). Thus, the observed enhancement in peel force reported in
Figure 7 cannot be accounted for by a simple renormalization
of the surface of contact, and we can conclude that patterning
results in an enhanced adhesive energy, Gp, that is greater than
Gb, the adhesive energy on the smooth substrate. To quantify
this enhancement, we chose to calculate Gp from the peel force
Fp measured on the patterned part of the substrate using the Figure 9. Enhancement of the peel energy measured on a PDMS
elastomer patterned with micropillars versus the pillar height h for
following equation (which takes into account the surface different peel velocities.
enhancement due to patterning):
eventually relaxes back abruptly to its initial position when the
Fp
Gp ) adhesive is detached from the top of the pillars (Figure 10 and
b
∆S
S0(+1 ) movie 1 in Supporting Information). The PDMS substrates remain
undamaged and can even be reused. However, for h > 3.3 µm
(about 3 times the pillar radius), ∆Gp drops. One can notice that
The enhancement in adhesion energy is then defined by ∆Gp ) long pillars tend to bend, overlap, and remain bent flat on the
Gp - Gb. surface after peeling, especially for the pillars located at the
Gp(V) CurVes. The peel energy Gp versus the peel velocity V trough of the fingerlike patterns of the peel front (Figure 11).
is shown in Figure 8 for a series of PDMS substrates patterned ObserVations of the Peel Front. On both Figures 10 and 11,
with micropillars of various heights. We selected substrates of the fingerlike deformations of the peel front have a regular shape
similar Gb. As observed with lamellar patterns, the variation of with a well-defined wavelength. This wavelength is independent
Gp with V differs between patterned and smooth elastomer of the pillar height. By comparing the shape of the peel front on
substrates: for smooth substrates, Gp is independent of V, whereas bare and patterned zones of the PDMS substrates (one example
for patterned substrates, Gp increases with V and eventually is given in Figure 12), one finds that, first, the fingerlike
saturates. Again, this suggests different origins for energy oscillations of the peel front appear more periodic on the patterned
dissipation. part than on the bare one and second, the length of the finger-like
∆Gp(h) CurVes. As shown in Figure 9, the enhancement in the pattern decreases as the peel velocity increases (see Figure 13
peel energy due to the patterning of the elastomer, ∆Gp, first for details) until the instability disappears and the peel front
increases with the height of the pillars. As the peel front passes, becomes a straight line above a given velocity. The length of the
each pillar appears to be strongly deformed and stretched and fingers is larger on the patterned part than that on the bare part
Adhesion Enhancement through Micropatterning
Figure 10. Picture of the peel front advancing from bottom to top
while the adhesive tape is peeled off at 6 µm/s from a PDMS elastomer
sample patterned with micropillars (height 0.8 µm). The size of the
image is 100 µm × 130 µm.
Figure 11. Picture of the peel front progressing from bottom to top
while the adhesive tape is peeled off at 6 µm/s from a PDMS elastomer
sample patterned with micropillars (height 3.8 µm). The size of the
image is 380 µm × 500 µm. At the edge of the fingerlike patterns
of the peel front, the pillars (appearing dark) remain flat on the
surface after the adhesive has been detached from the substrate.
at a given velocity. Also, as shown in Figure 14, this length
increases with the pillar height for h < 3.3 µm and drops above
h ) 3.3 µm. One can notice that the length of the fingers follows
the same evolution with the pillar height as does ∆Gp, and third,
the velocity, above which the peel front remains a straight line,
is higher for the patterned part of the substrate than for the smooth
one.
Taking these observations all together, the presence of the
pillars induces a better-defined fingering deformation of the peel
front, and the length of the fingers evolves with pillar height as
the enhancement of the peel energy does.
Different mechanisms could be responsible for the observed
enhancement in the peel energy: energy dissipation in the
adhesive due to the more complicated shape of the interface
leading to a modified field of deformation inside this acrylic
adhesive (compared to the situation with a smooth elastomer),
dissipation due to the elastic deformation of PDMS elastomer
(either in the underlying PDMS film or in the micropillars
themselves), and dissipation in the adhesive due to the final very
rapid detachment of the adhesive from the top of the deformed
pillars. To discriminate between these contributions, similar
experiments have been performed on substrates patterned
with rigid micropillars (polyurethane) that do not deform under
peeling.
Polyurethane Micropillars. Gp(V) CurVes. As shown in Figure
15, the slope of the G(V) curves is affected only for h > 3 µm.
Langmuir, Vol. 23, No. 13, 2007 6971
Figure 12. Pictures of the peel front progressing from bottom to
top while the adhesive tape is peeled off (at 6, 24, and 60 µm/s) a
PDMS elastomer sample patterned with micropillars (height 1.8
µm). (Left) bare part. (Right) patterned part. The size of the image
is 380 µm × 500 µm. The length of the fingerlike pattern decreases
as the peel velocity increases. On the patterned part, this length is
larger, and the shape of the instability is much more regular.
∆Gp(h) CurVes. In Figure 16, the enhancement in the peel
energy due to the patterning of rigid substrates is reported as a
function of the pillar height. One can easily see that the rigid
pillars are efficient only in enhancing the adhesive strength for
pillars that are long enough (h > 3 µm). This is quite different
from the flexible pillars that were losing their efficiency in that
range of h values.
Discussion
The first important factor that affects the aptitude of a patterned
surface to enhance adhesion seems to be the deformability of the
features formed on the surface of the substrate. Indeed, the
adhesion enhancement observed with the lamellae depended on
their ability to be deformed under the effect of peel forces: the
thinner the lamellae, the higher the adhesion, and the trend was
clear even if the adhesion enhancement remained quite weak
with the lamellar geometry. With the more deformable PDMS
micropillars, a significant adhesion enhancement, ∆Gp, was
clearly measured, in addition to trivial effects of enhanced surface
contact, and was increasing with pillar height. Accordingly, at
the peel front, the long micropillars were observed to be strongly
deformed before they detached from the adhesive to recover
their initial state. It appears clear that the elastic energy paid to
deform the micropillars was lost in the process, in a way quite
similar to the mechanism proposed by Lake and Thomas18 to
describe the rupture of cross-linked elastomers or the mechanism
proposed by Jagota et al.15 to describe adhesion through fibrillar
microstructures.
To understand how the micropillars, in the case of PDMS
substrates, was affecting the adhesion energy, we have modeled
the elastic energy of deformation of the patterned substrates as
6972 Langmuir, Vol. 23, No. 13, 2007
Figure 13. (Left) definition of lp, the length of the fingerlike pattern, on an image of the peel front (380 µm × 500 µm). (Right) variation
of lp with the peel velocity for patterned substrates.
Figure 14. Variation of lp, the length of the fingerlike pattern, with
pillar height.
a function of the aspect ratio of the micropillars using scaling
arguments and assuming that the pillars were acting independently
of each other. Several contributions can be distinguished: each
micropillar is bent and stretched, and the substrate itself can be
deformed under the effect of the local peel force transmitted by
the micropillar. The elastic bending energy Eb stored by one
pillar submitted to a force F can be evaluated22 as Eb ) 2F2h3/
3πEr4 (E is the Young’s modulus of the material), whereas the
elastic contribution Es associated with its stretching is Es ) F2h/
2πEr2. To determine the contribution of the elastic deformations
in the bulk PDMS elastomer underneath one pillar, one can
consider the classical Boussinesq problem where a cylindrical
punch indents an elastic medium. The force exerted on the punch
is then related to the deformation by δ ) 2F/3rE, and the resulting
energy is Ed ) 8πF2/27Er. Variations of these three contributions
with respect to the pillar height are shown in Figure 17. This
approach clearly predicts that the higher the pillars, the larger
the enhancement of adhesion. Moreover, two regimes can be
(22) Landau, L.; Lifshitz, E. In Théorie de l’EÄ lasticité, 2nd ed.; M.I.R.: Moscow,
1967; p 107.
Lamblet et al.
Figure 15. Variation of the logarithm of the peel energy G with
the logarithm of the peel velocity V while peeling the adhesive tape
on rigid samples (made of optical polyurethane cement) patterned
with micropillars of different heights. Values for the bare sample
are also represented.
distinguished depending on the value of h. For h < 1.6r, the
elastic energy stored in the system is dominated by the substrate
deformation. This is in good agreement with our observations:
for small h ) 0.3 or 0.7 µm (we recall again that all of the above
presented experiments have been performed with a fixed
micropillar radius of r ) 1 µm), the micropillars are only slightly
deformed, but they transmit stresses to the bulk elastomer
underlying the pillars. The resulting deformation of the substrate
relaxes back after the peel front has gone, and the corresponding
elastic energy is lost. For h > 1.6r, we predict that the dominant
contribution to the elastic energy is related to the bending of the
pillars. This bending contribution increases quickly with pillar
height, as observed experimentally. For our system, the bending
energy of the pillars overcomes the two others contributions by
1 order of magnitude for h g 3.5 µm. For long micropillars, a
buckling instability of the pillars is expected under the pressure
exerted on the system to form the contact between the adhesive
and the substrate. This buckling instability has not been observed
Adhesion Enhancement through Micropatterning
Figure 16. Enhancement of the peel energy measured on rigid
patterned substrates (made of optical polyurethane cement) versus
the pillars height h for different peel velocities.
for our samples even for the largest height investigated. However,
in the range of height above 3.5 µm, another effect comes into
play: above h ) 3.8 µm, the micropillars were observed to lose
their efficiency in enhancing adhesion. For this same height,
they started to no longer be able to relax to their original state
after the peel front had gone. Some micropillars remained
irreversibly bent down, sticking to the surface, especially along
the fingers formed in the adhesive by the periodic distortion of
the peel front, as shown in Figure 11. When such is the case, the
system can no longer be described in terms of independent pillars.
The above-developed picture (stretching of the micropillars
followed by their detachment one after the other, thus acting
independently of each other with respect to energy losses) no
longer holds. The gain in surface energy resulting from a final
state in which some pillars stick all together then needs to be
taken into account. This additional term (opposite in sign to the
elastic energy of deformation of the pillars that is lost when the
peel front passes) is not easy to estimate because it depends on
the exact way in which the micropillars remain connected to
each other. However, it clearly contributes to decrease the
adhesion enhancement. There is thus an optimum height for the
Figure 17. Contributions to the peel energy versus the pillar height h as the adhesive is peeled off the PDMS elastomer substrate patterned
with micropillars. Only dissipation inside the elastomer is taken into account and evaluated by scaling laws.
Langmuir, Vol. 23, No. 13, 2007 6973
deformable pillars to enhance adhesion (between 3.3 and 3.8 µm
for the geometry that we have investigated). The simple
description in terms of elastic deformable pillars independent of
each other should hold below a limiting height hc. This limiting
height hc should depend on the spacing between pillars (another
parameter of the pattern that we have just started to explore).
When it holds, the description we have developed shows that the
important parameter controlling which contribution (bending of
the pillars or deformation of the underlying substrate) dominates
the elastic energy of the patterned substrate lost during peeling
is not the height of the pillars but their aspect ratio. Increasing
the pillar height improves the enhancement of adhesion, but this
is limited: for very long pillars (or very small spacing between
pillars), interactions between neighboring pillars induce a partial
collapse of the pillars after peeling, which corresponds to a surface
energy that is lower than that of the substrate in contact with air
and in the initial configuration of the pattern. The corresponding
gain in surface energy partially compensates for the loss of energy
associated with the bending of the micropillars, and a noticeable
decrease in adhesion enhancement is measured.
For the non-deformable pillars, no elastic contribution due to
the deformation of the pillars is expected. Indeed, experiments
show that the rigid pillars with h < 3 µm did not produce noticeable
adhesion enhancement. We then face a new question: what is
the origin of the observed enhanced adhesion in the case of long
rigid pillars? The only possibility comes from the acrylic adhesive
side of the contact. The adhesive layer is a viscoelastic material.
Patterning the substrate should affect the viscoelastic losses within
the adhesive layer because the whole strain field in the vicinity
of the interface should be modified. Obviously, the depth of the
layer in the adhesive inside which the deformations are affected
by the patterning depends on the geometry of the pattern and,
in the case of micropillars, on their height. It is then reasonable
to assume that the enhancement of adhesion energy that we have
observed for rigid micropillars with h > 3 µm is due to a change
in the deformation field inside the adhesive layer associated with
the patterning of the substrate.
The next open question is then to evaluate whether such a
contribution to the enhancement of adhesion due to the adhesive
layer was not already present in the case of deformable substrates
and superimposed on the contribution of the elastic deformations
in the PDMS substrate. One way to try to answer this question
is to analyze the velocity effects because the viscoelastic losses
6974 Langmuir, Vol. 23, No. 13, 2007
Figure 18. (Right) comparison of pictures of the peel front (advancing from bottom to top at V ) 6 µm/s) while the adhesive tape is peeled
off a PDMS elastomer sample patterned with soft (top picture) and hard (bottom picture) micropillars (h ) 1.6 µm). The size of the images
is 380 µm × 500 µm. (Left) schematic side view of peeling and the definition of length L.
inside the adhesive are expected to be highly velocity-dependent,
whereas elastic contributions are not. We have established that
the velocity dependences were indeed quite different for smooth
or patterned substrates with moderately long micropillars, either
deformable or rigid, at a fixed adhesive layer. Through precise
observations of the peel front, we have some additional
experimental indications that kinetics effects show up for
deformable moderately long micropillars. This is shown in Figure
18. The length L that defines the highly deformed region of the
adhesive near the peel front is much larger for soft than for hard
patterned substrates under the same peel conditions, suggesting
different dissipation inside the adhesive. In the case of deformable
substrates, the rapid detachment of the adhesive from the top of
the pillars can lead to dissipation processes that are quite different
from what happens in the case of rigid substrates where the
adhesive detaches more progressively. The complexity of the
situation would require a finite element numerical analysis, but
this is beyond the scope of the present article. We think, however,
that the systematic comparison between deformable and non-
deformable patterned substrates having the same geometry is a
way to guide such modeling. This should help in understanding
how both sides of the interface are coupled in their deformations
and dissipations, which is a difficult and unsolved question in
most adhesive assemblies. Also, very simple experiments could
be performed with a series of adhesives of various viscoelastic
properties, which would certainly allow one to test and identify
the underlying mechanisms.
Conclusions
By investigating in a systematic manner the efficiency of
surface patterning in enhancing the adhesive strength at PDMS-
acrylic adhesive interfaces and by comparing, at fixed geometry,
rigid and deformable patterned surfaces (having similar adhesion
energies on the smooth part of the substrate), we have shown that
soft deformable micrometric patterns are quite efficient at
enhancing adhesion. For soft arrays of micropillars, using simple
scaling arguments, several regimes have been identified, de-
Lamblet et al.
pending on the aspect ratio of the pillars, with either a dominant
role of the deformations in the underlying bulk elastomer film
through the peel force locally transmitted by the pillars (h/r <
1.6) or of the bending energy of the micropillar themselves,
which is lost when the peel front passes (h/r > 1.6). For higher
structures (h/r > 3), soft patterns are altered when the contact is
broken. The corresponding gain in surface energy largely
compensates for the energy loss associated with elastic deforma-
tion. Rigid patterns, for which there is no elastic contribution to
the adhesion, then become more efficient than soft ones, probably
because of enhanced dissipation inside the adhesive itself. The
present set of experiments backs up the idea that the ability of
the patterned surface to be deformed plays a crucial role in
enhancing adhesion, similarly to what is seen, for example, in
the case of geckos. This is quite similar in spirit to what has been
reported by Crosby et al.,19 even if the present investigation
strongly differs in the range of pattern sizes. (Our pillars have
a 1 µm radius and the contact is 1 cm wide, whereas Crosby et
al.19 were using 50 times wider cylinders and the contact covered
only a few cylinders.) Our investigations show that, by varying
the size of the pattern and/or the mechanical properties of the
substrate, it is possible to fine tune the level of adhesion at PDMS-
acrylic adhesive interfaces, independently of the details of the
chemistry of the adhesive.
Acknowledgment. This work was financially supported by
Rhodia Silicone. We thank Elie Raphael and Robert H. Austin
for fruitful discussions and suggestions as well as Huges Bodiguel
and Christian Fretigny for providing access to their inverted
AFM. The DRIE process has been performed at the Cornell
NanoScale Facility.
Supporting Information Available: Setup of peel test experi-
ment (homemade apparatus). Movie of the PDMS elastomer sample
patterned by pillars (h ) 1.8 µm) peeled off at 6 µm/s. This material
is available free of charge via the Internet at http://pubs.acs.org.
LA063104H