Article pubs.acs.org/Langmuir
Control of Adhesion via Internally Pressurized Subsurface Microchannels Edward Peter Arul† and Animangsu Ghatak*,†,‡ †
Department of Chemical Engineering, Indian Institute of Technology, Kanpur, UP 208016, India INM-Leibniz Institute for New Materials and Saarland University, Campus D 2 2, 66123 Saarbruecken, Germany
‡
ABSTRACT: While pressure sensitive adhesives in general consist of a layer of viscoelastic glue sandwiched between two adherents, we explore here the design of an adhesive embedded with microchannels which remain either open to atmosphere or pressurized to different positive and negative pressures. We subject these layers to indentation by a rigid cylinder such that in addition to adhesion between the indenter and the adhesive surface, the inner walls of the channels too self-adhere; during retraction of the indenter, these surfaces debond, but at a different load, thus resulting in hysteresis. When these channels are pressurized to different extents, the contact areas of various interfaces vary, so also the resultant hysteresis. For experiments with constant depth of indentation, the hysteresis increases and attains maxima at an intermediate value of the internal pressure inside the channels. The hysteresis increases also with the skin thickness of the adhesive over the channels. These results show that subsurface channels in an adhesive allow active manipulation of adhesion over a large range via coupled effect of geometry of channels, their surface characteristics, and the pressure inside.
■
structures,6,15−20 self-adhesion of channel walls, dynamic alteration in channel orientation,5 and so on. Importantly, these mechanisms do not work independently but in unison with others, resulting often in a coupled effect. For example, it has recently been shown that when embedded channels within an adhesive layer are filled with a wetting liquid, the excess interfacial energy of wetting leads to bulging of the adhesive surface but also alters the channel from being circular.21−24 In fact, the channel shape does not remain constant during the course of peeling of an adherent off the adhesive, because peeling itself alters the curvature of the channel cross section which in turn alters the stress field within the material and its ability to arrest the peeling front. The end result is a dynamic coupled effect of geometry and surface energy in the adhesive material, such that the energy required to detach a flexible adherent off this layer enhances over that on a smooth and homogeneous adhesive surface. The question however arises how exactly the internal pressure in the embedded microstructures alters the adhesion strength of the adhesive. Here we have addressed this question by designing adhesives embedded with planar microchannels and by pressurizing them to different extents. We have carried out displacement controlled indentation tests in the vicinity of the embedded channels. Depending on the depth of indentation, the internal surfaces of the walls of channels self-adhere resulting in increase in the effective area of adhesion over and
INTRODUCTION Recent advances in pressure sensitive adhesives deal with variety of novel mechanisms for tuning and enhancing the adhesion strength without compromising the reusability of the adhesive. These mechanisms do not depend upon the bulk viscoelasticity of the material which increases the adhesion strength but at the cost of rendering the adhesive unsuitable for reuse after it has been applied once or twice on a surface. Instead, these mechanisms involve patterning the adhesive surface by generation of microstructures, e.g., pillar, rods, and fibers. Hierarchy in these structures is achieved by generation of an inverted “micro-tree” which essentially mimics the natural adhesives at the feet of several living organisms. Fabrication of the “tree” is, however, not easy, which limits the economic application of such adhesives. A different approach in this context has been to embed subsurface structures within the adhesive layer, e.g., air and liquid filled microchannels,1−4 monolithic stack of channels, channels arranged in different orientations, e.g., curved and straight,5 channels with inner surface of the wall topographically patterned6 and chemically modified with surface active groups,5−10 and so on. The molecular length scales of surface active groups, microscopic dimensions of the topographical pattern on surface, and channel geometry render a hierarchical structure to the adhesive which has been shown to influence adhesion via multiple mechanisms, e.g., alteration of the local modulus or the compliance of the adhesive,11,12 undulation at the surface of the adhesive,13 dynamic alteration of channel geometry14 via change in capillary pressure inside these embedded vessels,4 crack arrest and initiation at the vicinity of the of subsurface © 2012 American Chemical Society
Received: November 23, 2011 Revised: February 6, 2012 Published: February 22, 2012 4339
dx.doi.org/10.1021/la204618u | Langmuir 2012, 28, 4339−4345
Langmuir
Article
indenter and the adhesive surface, and Figure 1c shows the top view of the channel along with the punctured holes.
above that between the indenter and the adhesive with consequent increase in hysteresis. We have shown that the indentation depth at which such self-adhesion occurs between the internal surfaces and the hysteresis5−10 alters with the internal pressure inside the channel; adhesion hysteresis maximizes at an intermediate positive and negative pressure inside these channels.
■
RESULTS AND DISCUSSION Adhesion Hysteresis of a Solvent Extracted PDMS Lens. The effect of solvent extraction on adhesion hysteresis was studied by carrying out JKR experiment5 of a PDMS hemispherical lens on a microscope glass slide coated with a self-assembled monolayer of octadecyltrichlorosilane (OTS) molecules. The PDMS lens of radius of curvature of 1.6 mm was prepared by the method described in ref 8. The work of adhesion in this experiment was found to be Wloading = 38 ± 2 mJ/m2, Wunloading = 57 ± 3 mJ/m2, i.e., adhesion hysteresis W1 = 20 mJ/m2. These experiments were repeated using the same hemispherical lens after solvent extraction as described in ref 8. The work of adhesion was now found to be Wloading = 45 ± 4 mJ/m2, Wunloading = 75 ± 3 mJ/m2, i.e., adhesion hysteresis W1 = 30 mJ/m2, which was somewhat larger than the previous case. Adhesion hysteresis however increased significantly when the same solvent extracted PDMS hemispherical lens was brought in contact with a PDMS layer bonded to a rigid substrate which too was solvent extracted. The adhesion hysteresis, known as “self-adhesion hysteresis,”5−10 was estimated to be W2−2′ = 180 mJ/m2. This enhancement in adhesion hysteresis resulted from hydrogen bonding interaction between two chemically treated contacting surfaces.7,8 Hysteresis of Channel Embedded Adhesive Layers. We incorporate the effect of self-adhesion hysteresis in adhesive layers by embedding within it a rectangular microchannel. The geometry of the channel (width and height) is such that during indentation with a rigid cylinder its walls self-adhere and during withdrawal they separate albeit at different loads, resulting in self-adhesion hysteresis in a complete cycle. The internal structure of the microchannel alters the compliance and effective modulus of the adhesive film such that a noncontinuous variation of load with the indentation depth is not captured by the Johnson−Kendall−Robert theory.5,25 Instead, the adhesion hysteresis is estimated by calculating the area of the shaded portion between the loading and unloading curves followed by dividing it by the maximum area of contact between the rigid indenter and adhesive surface Amax:5,6,16,24 ΔE = ∮ FdΔ/Amax. Curve 2 in Figure 2a shows the typical F vs Δ data for an adhesive layer in which the channel with dimensions hc = 100 μm, 2w = 2 mm, and t = 205 μm was maintained at atmospheric pressure, i.e., P = 0 mmHg. The portion A−B of this curve shows the loading data when the indenter is brought in contact with adhesive surface 1 and is pressed against it. A rectangular contact area appears, the width of which increases with the depth of indentation and the contact load. At this small indentation, the channel wall 1−2 behaves like a freehanging thin membrane supported at two edges; therefore, with increased indentation it first stretches and bends, and eventually the two planar surfaces 2−2′ of the channel adhere forming an interface. With further increase in indentation, the adhesive behaves like a homogeneous layer so that the slope of curve B− C increases to what is expected for a layer without any embedded channel. During unloading of the indenter, the portion C−D of curve 2 nearly superimposes on the loading curve B−C, thus yielding negligible hysteresis. With further unloading, however, the area of contact of the interface 2−2′ decreases as the contacting surfaces begins to separate, albeit at a lower load than at which the contact was formed, thus
■
EXPERIMENT Figure 1a depicts the schematic of the experiment in which an adhesive layer remained strongly bonded to a rigid glass plate.
Figure 1. (a) Schematic of the indentation experiment on adhesive films embedded with pressurized microchannel. A glass hemicylinder was used as the indenter. (b) Optical micrograph of the area of contact between the indenter and the adhesive film. (c) Top view of optical micrograph showing the top view of the microchannel with an air inlet through which the air was pumped with syringe pump and an outlet which was connected to a manometer for measuring the pressure inside the microchannel.
Thin, solvent extracted PDMS layers embedded with channels of desired geometry were used as the model adhesive. The channels were either open to the atmosphere or their internal pressure was maintained at different positive and negative values. A syringe pump (Harvard Apparatus, model number PHD 2000) was used to pump air into the microchannels through an inlet hole punctured through the film into the channel. An outlet to the microchannels was connected to a Utube mercury manometer to measure the pressure. The diameter of the U-tube was 2 orders of magnitude larger than that of the channel. Therefore, any change in volume of microchannel during experiment resulted in insignificant change in height of the liquid head in the manometer. Thus the pressure inside the channel could be maintained constant throughout the loading and unloading experiment. A glass hemicylinder of length 1.2 mm and radius of 1.4 mm was used as the indenter, which was first aligned axi-symmetric with the microchannel and was then brought into contact with the adhesive and pressed against it at a constant rate, 1.5 μm/s, using a motorized nanopositioner (Newport, PC actuator). The contacting load was measured using a weighing balance (Citizen made, 5 digits) interfaced with a computer through RS232. The contact area was viewed using a digital camera mounted on a microscope. Figure 1b shows an optical micrograph of the area of contact between the hemicylindrical 4340
dx.doi.org/10.1021/la204618u | Langmuir 2012, 28, 4339−4345
Langmuir
Article
vs Δ data at 80 mmHg pressure (gauge pressure) inside the channel. Because of this internal pressure, the thin skin 1−2 above the microchannel no longer remains flat but bulges out as shown in Figure 3b, thus generating a positive bulge height δ. Similarly, when the pressure inside the channel is decreased below atmosphere, the skin bulges in as shown by Figure 3c; i.e., it generates a negative δ. How does the bulge height vary with the internal pressure inside the channels? Large deformation of thin plates has been described rigorously by using the Fö ppl−von Kármán equations26 which are quite complex and not amenable to analytical solution except for few simple situations. Nevertheless, two extreme situations can be considered: the film is thin enough that the energy is essentially stored in stretching mode, which results in deflection increasing with applied pressure as26,27 ∼P1/3. The data in Figure 4 indeed show that Figure 2. Force vs displacement plots for the adhesive embedded with rectangular microchannel. Curves 1, 2, and 3 represent load vs displacement data for indentation at three different internal pressures, P = −80, 0, and 80 mmHg, respectively, in channels. The shaded portion is a measure of the adhesion hysteresis.
yielding a “self-adhesion hysteresis”.5−10 Once the interface 2− 2′ completely separates, the unloading curve jumps back to the loading curve at point E following which we observe negligible hysteresis in the E−F portion of curve 2. Here, the width of interface 1 decreases, and eventually the indenter completely separates from the adhesive surface at F. Insignificant hysteresis occurs in the portion BCD and AEF of the load−displacement curve, whereas maximum hysteresis occurs within the portion BDE. The hysteresis was obtained as ΔE = 482 ± 43 mJ/m2, which is a significant enhancement over that of the smooth control adhesive layer without any embedded structure, W2−2′ = 180 mJ/m2. Channels Filled with Air at Positive and Negative Pressure. Importantly, for these adhesive layers hysteresis is comprised of two components: that occurring via adhesion and debonding of the interface 2−2′ and that between the indenter and the adhesive surface. When the channels are filled with air at pressure different from atmospheric, the adhesion and debonding of the 2−2′ interface occurs at different indentation depth and contact loads as compared to that occurring at atmospheric pressure. The curve 3 in Figure 2 represents the F
Figure 4. Maximum bulging height δ0 of thin skin above the rectangular channel plotted as a function of internal pressure P inside the channel. Symbols ○, □, and Δ indicate skin thickness 104, 205, 408 μm, respectively.
for skin thickness t = 104 μm, the maximum bulge height of the deformed skin indeed varies with the internal pressure as δ ∼ P1/3. In the other extreme, in which the bending energy of the elastic layer dominates, the deflection scales linearly with pressure, ∼P. The data for skin thickness, t = 205 and 408 μm, show that δ ∼ P, signifying that deformation of these layers remains intermediate to these two extreme situations. Notice that in addition to the bulging effect, prestress is also developed in the adhesive at the vicinity of the channel because of the
Figure 3. Schematic representation of an adhesive layer embedded with microchannels with different internal pressure. The drawings a−c represent pressure P = Patm, P > Patm, and P < Patm, respectively. 4341
dx.doi.org/10.1021/la204618u | Langmuir 2012, 28, 4339−4345
Langmuir
Article
pressure inside. When the prestress is small, it does not alter the modulus significantly; however, for large enough bulging of the thin elastomeric skin on the channel, its shear modulus tends to increase. The F vs Δ curves 1 and 3 in Figure 2 capture the dual effect of change in topography of the adhesive surface and the consequent change in the local shear modulus of the adhesive layer at the vicinity of the channel. For example, curve 3 shows that the self-adhesion of interface 2−2′ occurs at a load F = 0.114N and indentation depth Δ = 150 μm larger than what is observed with the channel kept at atmospheric and subatmospheric internal pressure, F = 0.068N and Δ = 100 μm, respectively. This data of cyclic loading−unloading experiment at positive pressure differ from that obtained at atmospheric pressure in another way. During withdrawal of the indenter from the adhesive surface, the contact load remains significantly smaller than that during loading all through. As a result, unloading curve C′D′E′F′ does not superimpose at all the loading curve A′B′C′ thus resulting in a large hysteresis, ΔE = 1400 ± 98 mJ/m2, significantly larger than what is obtained at atmospheric pressure as in curve 2 of Figure 2. Notice that only a small portion of this hysteresis corresponds to that observed during separation of the 2−2′ interface; hysteresis essentially occurs during separation of the indenter from the adhesive surface. Similarly curve 1 in Figure 2a shows the load vs displacement data for negative pressure inside the microchannel: P =− 80 mm Hg. Because of the inward bulge of the adhesive film 1−2 at this pressure, the channel wall 2−2′ selfadheres at a smaller load F = 0.023N and smaller indentation depth Δ = 50 μm forming the interface 2−2′. The adhesion hysteresis represented by the shaded area between the loading and unloading data of curve 1 is estimated to be ΔE = 1520 ± 140 mJ/m2, which too is larger than that obtained for adhesive embedded with channels at atmospheric internal pressure. Notice that here the hysteresis predominantly occurs during separation of interface 2−2′. In essence, the adhesion hysteresis increases for both positive and negative pressure inside the channels, while in the former it is the hysteresis at the interface of the indenter and the adhesive that dominates; for the latter, it is the hysteresis at the interface 2−2′ that becomes more prominent. Optimum Pressure for Maximization of Hysteresis. While the curves 1−3 in Figure 2 represent the F vs Δ data for three representative cases, negative, atmospheric, and positive pressure, respectively, the optimum operating pressure for a given set of geometric parameters of the channels and a given maximum depth of indentation Δmax by the indenter is obtained by varying systematically the pressure inside the channel. These sets of experiments show that for a constant Δmax the hysteresis does not increase monotonically with either the positive or the negative pressure inside the channel, because the indentation depth, Δ2−2′, at which the internal walls of the channel can selfadhere to form the interface 2−2′ does not remain constant but varies with change in pressure as also does the area of the interface 2−2′. For positive pressure, Δ2−2′ increases with P, because of increase in bulging height of the thin skin above the channel. In Figure 5A we show the typical data of Δ2−2′ (in μm) as a function of pressure P (in mm Hg) for three different skin thickness: 104, 205, and 408 μm, respectively. For each case, as the pressure exceeds a threshold value, the required indentation depth Δ2−2′ for the channel walls to self-adhere exceeds the maximum indentation depth of 200 μm. The bar chart in the inset of Figure 5A shows these threshold pressures as 90, 135,
Figure 5. Plot of indentation depth, Δ2−2′,for self-adhesion to occur between internal walls of channels as a function of pressure, P. (A) Data represent indentation on an adhesive embedded with channels having dimensions hc = 100 μm and 2w = 2 mm maintained at positive pressure. The symbols Δ, ○, and □ represent data for three different adhesive films with varying skin thickness as t = 104, 205, and 408 μm, respectively. The bar diagram at the inset shows the threshold pressure Pc+ for which the internal surfaces of the channel do not self-adhere for the maximum indentation depth of Δ2−2′ = 200 μm. (B) Plot of Δ2−2′ vs P for negative pressure inside channels. The bar diagram at the inset shows the threshold pressure Pc− for which the internal surfaces of the channel self-adhere without any indentation by the indenter. The solid and dashed lines are a guide to the eye.
150 μm, respectively. For pressure beyond this critical limit, the interface 2−2′ does not at all form; consequently, the hysteresis at interface 1−2 between the indenter and the adhesive remains effective, resulting in significantly diminished hysteresis. Notice that at zero atmospheric pressure the Δ2−2′values do not exactly match the channel height of ∼100 μm, which possibly has to do with shrinkage of cured PDMS network as it is cooled to room temperature from its curing temperature of ∼80 °C. For negative pressure too, the ΔE vs P data in Figure 6 shows that with decrease in pressure, hysteresis first increases until a
Figure 6. Hysteresis ΔE plotted as a function of pressure P within channel of dimensions hc = 100 μm and 2w = 2 mm. The symbols ○, □, and Δ represent the data for skin thickness t = 104, 205, and 408 μm, respectively.
maximum value is reached at an intermediate pressure, beyond which it decreases. Actually, for negative pressure in the channel, the channel wall 1−2 bulges in; as a result, the indenter now requires smaller indentation depth Δ2−2′ and consequently smaller load on indenter, for the internal surfaces 2 and 2′ to contact and adhere. This indentation depth Δ2−2′ decreases with decrease in pressure P inside the channel. For each case, as the pressure decreases below a threshold limit Pc−, Δ2−2′ asymptotically goes to zero implying that beyond this threshold pressure, the internal surfaces of the channel selfadhere to form 2−2′ without any indentation required to be done on the adhesive. In essence, here again, in cyclic loading− 4342
dx.doi.org/10.1021/la204618u | Langmuir 2012, 28, 4339−4345
Langmuir
Article
and the internal pressure P. In the limit of P → 0 and zero indentation load on the indenter, the contact radius can be estimated from JKR theory as r0 = ((256Wle)/(πEe))1/3R2/3. Actual contact radius will however be different from this quantity because of finite pressure inside channel, so that in the limit that the radius of the cylinder R far exceeds (Δ0 + hc), R ≫ (Δ0 + hc), r/R can be expressed as r/R ≈ (r0/R)(1 + (Δ0(P))/hc). In order to estimate, we will consider the specific case when the elastic energy of the skin remains stored as bending energy, and we can then write the following force balance equation
unloading experiment, it is the hysteresis at the indenter− adhesive interface that remains effective thus significantly diminishing the hysteresis. The hysteresis beyond this critical internal pressure corroborates with that for smooth homogeneous film without any embedded microstructures. It is interesting to note that the figure shows that despite the similar general behavior of hysteresis at +ve and −ve pressures, the variation is not symmetric as it depends also upon the geometry of the channel and how it alters with indentation.
■
ANALYSIS OF THE EXPERIMENT In order to rationalize the experimental observation, we attempt to estimate how the area of interfaces 1 and 2−2′ varies with the internal pressure P in the channel. In Figure 7 we present a
D
d4δ dx 4
= P,
0
