Dynamically Tunable Dry Adhesion via Subsurface Stiffness Modulation

Transcription

1 University of Nevada, Reno Dynamically Tunable Dry Adhesion via Subsurface Stiffness Modulation A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Mechanical Engineering by Milad Tatari Dr. Wanliang Shan/Thesis Advisor August, 2018

2 Copyright by Milad Tatari 2018 All Rights Reserved

3 THE GRADUATE SCHOOL We recommend that the thesis prepared under our supervision by MILAD TATARI entitled Dynamically Tunable Dry Adhesion via Subsurface Stiffness Modulation be accepted in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Wanliang Shan, Ph.D, Advisor Feifei Fan, Ph.D, Committee Member Behrooz Abbasi, Ph.D, Graduate School Representative David W. Zeh, Ph.D, Dean, Graduate School August, 2018

4 i ABSTRACT Tunable dry adhesion has different applications, including microtransfer printing, climbing robots, pick and place robots and gripping in manufacturing processes. This study investigates a new type of tunable adhesion mechanism in which the adhesion strength can be dynamically and reversibly tuned via subsurface stiffness modulation. In the current study, a novel soft gripper is introduced and demonstrated such that the adhesion strength can be switched between weak and strong states depending on the subsurface stiffness. The low and high stiffness correspond to weak and strong adhesion strengths, respectively. The proposed soft device is constructed out of a soft elastomer, polydimethylsiloxane (PDMS), which is embedded with a tunble stiffness core, Conductive Propylene Based Elastomer (CPBE). Stiffness of the core can be tuned via application of electrical voltage. Activation of the core with electrical voltage reduces the core stiffness and subsequently stiffness of the whole composite post. This reduction in the composite post stiffness results in a change in the stress distribution and required force for delamination at the adhered interface and finally a drop in the effective adhesion strength is achieved. ANSYS software is employed to conduct Finite Element (FE) simulations and derive the pattern of stress distribution at the interface. Pattern of crack initiation and propagation has also been experimentally video recorded for both activated and non-activated composite posts. Both experimental and simulation results show that when the post is not electrically activated (stiff) crack starts from the center and high adhesion strength is achieved. But when the post gets activated (compliant state) the interfacial crack starts from the edge and as a result adhesion strength

5 ii is decreased. The adhesion of the composite posts with a range of dimensions and activation voltages has been characterized and it is shown that the adhesion can be reduced by as much as a factor of 6. As a demonstration for the applications of the proposed soft gripper, a variety of objects with different weights has been manipulated in the supplemental video. The proposed novel tunable dry adhesion mechanism uses subsurface stiffness modulation in which the adhesion strength is completely reversible and does not need a complicated activation mechanism.

6 iii ACKNOWLEDGEMENTS Thank you, Dr. Wanliang Shan, my thesis advisor, for the guidance to complete this project. Dr. Kevin Turner, thank you for all your efforts and help to accomplish this project. I am going to especially thank my parents and my brother for all their supports and efforts throughout my whole life to reach this point. I would like to thank Mr. Amir Mohammadi Nasab for his efforts during the project. I am grateful to my committee members, Dr. Feifei Fan and Dr. Behrooz Abbasi, for reviewing my work and providing excellent feedback. I give special thanks to the members of the Shan Research Group for being friendly and supportive colleagues.

7 iv Table of Contents ABSTRACT... i ACKNOWLEDGEMENTS... iii LIST OF TABLES... Error! Bookmark not defined. LIST OF FIGURES... vi Chapter 1 Introduction Introduction Soft Robotic Grippers Gripping by Actuation Mechanism Gripping by Stiffness Control Gripping by Adhesion Control Dry Adhesion Fabrication Reversible and Tunable Dry Adhesion Chapter 2 Background Introduction Mechanics of Adhesion Tunable and reversible dry Adhesion Tunable stiffness mechanisms Project Chapter 3 Design and Fabrication Introduction Design Fabrication Method Chapter 4 Testing Method and Setup Introduction Experimental Setup Testing Conditions... 43

8 v Chapter 5 Results and Discussion Introduction Simulation Results Experimental Results Discussion Demonstration of Applications Chapter 6 Summary, Conclusions, and Future Work Summary Conclusions Future Work References...59

9 vi LIST OF FIGURES Figure 1-1: Classification of soft robotic gripper technologies for grasping objects with different geometries [2] Figure 1-2: grippers with passive structures. a) contact-driven deformation working principle (left)[2], Fin Ray robotic gripper [16] b) Tendon driven mechanism working principle (left) [2], origami gripper with adjustable stiffness joints for multiple grasp modes [21]... 5 Figure 1-3: a) working principle of FEAs [2]. b) PneuNets [29] c) Fingers actuated by electrohydrodynamics [30]. d) Self-healing polymers [31] Figure 1-4: a) Working principle of Dielectric Elastomer Actuators (DEAs) [2]. b) A soft gripper based on dielectric elastomer structure [37]. c) Working principle of electroactive polymers [2] d) Soft gripper using IPMC fingers [39] Figure 1-5: a) working principle of SMPs [2] b) Soft gripper using bidirectional shape memory polymer [40] c) working principle of SMA material [2] d) soft gripper with an elastomeric figure structure using shape memory alloy [41] Figure 1-6: a) A soft gripper based on granular jamming mechanism [59] b) A soft gripper through a combination of encapsulated LMPA inside a soft material and dielectric elastomer actuator effect [60] Figure 1-7 a) Soft gripper using electroadhesion mechanism can handle a variety of objects[2]. b) Gecko adhesive system from macro to nanostructures c) Dry adhesive system with mushroom shaped microfibers handle different objects [71] Figure 1-8. Contact tips shape of fibrillary pad in animal [75] Figure 1-9. a) Scanning electron microscopy (SEM) images of a) PDMS microfibers fabricated by self-molding photolithographic [79]. b) PMMA nanofibers fabricated by hot-embossing and elongation by mold removal at high temperatures [64], [79] Figure a) SEM images of tilted fibers fabricated by soft molding on SU-8 wafers b) Hierarchal structure fabricated by double molding photolithography Figure Fabrication steps of micropatterns with different 3D tip shape [64], [84] Figure 2-1: The values of atomic bond energy, strain energy and total energy for different crack lengths Figure 2-2: Fracture modes (left to right: opening, in-plane shear, out of plane shear) Figure 2-3: Critical energy release rate vs peeling velocity in different temperatures for PDMS material [90] Figure 2-4: Normalized normal stress vs a shear displacement for a soft composite post with an embedded rigid core [93] Figure 2-5: Scanning electron microscope images of arrays of 35 microns with mushroom tips. Angles: (a) 34º (b) 90º (c,d) 23º [102] Figure 2-6: Tip behavior with different loadings. (a) Original unloaded condition (b) Preload compression. (c) Shearing in the releasing direction (d) Shearing in the gripping direction [102] 24 Figure 2-7: Pull-off forces when magnetic field is off and on. Applicable in dry and wet adhesion [103]... 25

10 Figure 2-8: A) tilting of magnetic pillars to opposite directions when the magnet is approached from opposite side. B) Behavior of pillars under a stronger magnetic field. (Scale bar is 20 microns) [103] Figure 2-9: Pull-off forces of thermally actuated mechanism for different preloads [107] Figure 2-10: Dry adhesion behavior under dry, 100% humidity and immersed in water conditions [109] Figure 2-11: (a) Schematic of an adhesive system with a tunable stiffness backing layer. (b) Adhesion measurement of different states of rigidity of the backing layer [117] Figure 2-12: (a) schematic of a rigid core embedded in a soft shell. (b) Homogenous post (c) composite post when the rigid core is close to the interface [Helen s paper] Figure 2-13: CPBE embedded in PDMS soft shell. After activation with electrical current, the composite post stiffness reversibly softens to 10 % of its original value [122] Figure 2-14: The proposed mechanism consisting of rigidity tunable core embedded in a soft shell can be used as a soft gripper. When the composite post gets activated, it softens and low adhesion is the reason to release the object quickly Figure 3-1: (a) (1) Schematic of the composite post (2) control sample (3) Composite post sample (b) Ratio of the composite post pull-off forces to that of control sample for different values of t/r. As t/r decrease crack is moving from the edge towards the center Figure 3-2: (a) Isometric, front and side views of the schematic of the composite post with a U- shaped tunable rigidity core. (b) Schematic of electrical connection and photograph of a composite post sample, consisting of PDMS with an embedded U-shaped CPBE core and copper wires for electrical connection [AMI] Figure 3-3: Different steps of fabricating a U-shaped tunable rigidity core Figure 3-4: (a) Mold 1 to cast and cure PDMS with an embedded channel (b) Mold 2 which is deeper to fabricate the final post. An acrylic sheet is positioned at the bottom of final post to achieve a very smooth and flat PDMS interface Figure 3-5: Fabrication process of the soft composite post. Note that in Step 3 an acrylic sheet with a smooth surface is placed at the bottom of a longer mold to achieve a smooth PDMS surface at the bottom of the post (i.e., the surface that forms the adhered interface) Figure 4-1: The experimental setup used for all the testing. A 50N load cell was installed for more precise measuring Figure 4-2: Close-up of the testing area showing exposed copper electrodes, fixed glass substrate, and fixture used to attach the post to the Instron Figure 5-1: Finite element-predicted distributions of normal stress on the adhered interface of the composite post. The stresses are normalized by dividing the local stress by the average stress at the adhered interface. Results are shown for one quarter of the contact area of the post; the right and bottom edges of the plots are symmetry planes and the center of the post is located at the origin of the coordinate frame (lower right corner of each plot). Results for a) t/l = 0.35, nonactivated. b) t/l = 0.12, non-activated. c) t/l = 0.011, non-activated. d) t/l = 0.12, activated [128] Figure 5-2: Adhesion forces of the non-activated composite posts with different t/l values. All tests were performed five times. The standard deviation for pure PDMS post adhesion force measurements is 0.03 N [130] vii

11 Figure 5-3: Percent remaining adhesion strength of the composite posts as a function of activation voltages. In all adhesion experiments, the composite post is activated for 2 minutes [130] Figure 5-4: Stiffness of the composite and pure PDMS posts for different activation voltages ( Voltage OFF corresponds to the stiffness measurement after cooling down for each activation, with activation time of 2 mins) [130] Figure 5-5: Measured temperature of the interface between the composite post (t/l = 0.12) and the glass substrate [130] viii

12 1 Chapter 1 Introduction 1.1 Introduction This section will talk about different categories of soft robotic grippers, their principal of operations, advantageous and disadvantageous to manipulate a variety of objects. Then as a unique approach towards soft robotic grippers, dry adhesive mechanisms will be introduced and discussed as the main focus of the current study. Bio-inspired dry adhesion concept and underlying physical phenomena will also be presented. Fabrication of gecko adhesive micro and nanofibers will be shortly reviewed. Tunable and switchable dry adhesion can be employed to handle objects with dimension ranging from micro to macro and potentially be used as soft grippers. A short background on tunable a reversible dry adhesion and its theory will familiarize the readers with the concept of the project before going into details. 1.2 Soft Robotic Grippers Grasping, releasing and manipulation are main functions of humans and animals to do daily tasks. In simple words, it includes picking up, twisting and transferring an object to different directions and whenever needed holds or releases the object. In high-tech industries, there is always a need to manipulate different parts or devices whether in mass production, manufacturing or human-robot interaction tasks [1]. Traditional robotic grippers being used in many manufacturing industries for decades consist of rigid components i.e. joints and link [2]. Actuators (pneumatic, tendon-like) can be placed within the joints or links. Proprioceptive sensors (Hall-effect sensors, encoders, torque sensors,

13 2 etc.) are employed to gather the gripper information such as position and velocity, and exteroceptive sensors (pressure, optical, resistive sensors) can provide some information about the object [2]. Although a lot of novel anthropomorphic gripper designs have been presented so far [2, 3], due to intrinsic rigidity and complexity, they have difficulties in handling soft and fragile objects and also in human-robot interaction [4 7]. Soft and compliant components are considered as a solution to update the traditional grippers (adding partly soft components) [8,9] or replace them by completely soft grippers [10]. Advanced materials, polymers and soft components are being widely studied to design light, simple and versatile grippers to overcome limitations of grippers made of rigid components. Mostly traditional rigid components used in grippers have difficulties in adaptability and flexibility to handle a variety of objects. As an example, if the object geometry and size vary, the gripper needs to be set up again for new working condition. Moreover, contact forces between a rigid gripper and a hard object leads to some shocks and might damage the object. An individual might say that a rubber pad can be a solution to prevent contact damages [2]. Yes that is correct, but this does not solve all the problems. As a consequence, there is a need to develop soft robotic grippers to replace the traditional hard grippers to overcome aforementioned limitations. One of the basic and main components of soft grippers is elastomers that can provide large strains up to 100%. Silicone rubbers are considered as an appropriate choice thanks to their ease of manufacturing and shaping, resistive to temperatures from -55 to 300, low toxicity and robustness [2]. Materials that change their properties in response to stimuli can be widely used in soft robotic grippers to provide different functionalities. These materials

14 3 include shape memory alloys (SMA), shape memory polymers (SMP) and low melting point alloys (LMPA). As an example, stiffness tunablity can significantly help the functionality of a soft gripper to conform to a typical object (soft state) and pick the object up (hard state). Types of stimuli can be an electric or a magnetic field, ph, chemical concentration, humidity, light and materials in granular forms [2].The recent progress in soft robotics [11] [14] leads to a huge improvement in soft gripper technologies and functionalities. Soft gripping technologies can be categorized in three main groups based on their functionalities. Gripping by actuating, by controlling the stiffness and by controlled adhesion are 3 main categories that will be discussed shortly in this chapter and then the 3 rd method which is adhesion control will be addressed in detail as the main goal of this research. Grippers using actuation mechanism works based on bending gripper fingers around a typical object as an individual does to pick up different items in his/her daily life. Grippers based on stiffness control employ tunable rigidity materials. This type of gripper, in the soft state with low actuation force, can grasp fragile objects with complicated geometries and pick them up whenever the gripper gets rigid. Shape memory polymers and low melting point alloys as phase change materials can be potentially used in these types of grippers. Gripping based on controlled adhesion such as dry adhesion (also called Gecko adhesion) and electro adhesion is based on surface forces at the contact interface between the gripper and object. Advantageous of gripping by controlled adhesion is manipulating delicate objects, or objects with flat surfaces, while at the same time it is limited to dry, smooth and clean surfaces. These gripper technologies will be explained in details in this

15 4 chapter. Figure 1-1 qualitatively classifies the performance of these grippers for objects with different geometries. The lighter color shows a worse performance for that specific object geometry. For example, it can be seen that for flat objects controlled adhesion is the best method to employ in soft grippers. Figure 1-1: Classification of soft robotic gripper technologies for grasping objects with different geometries [2] Gripping by Actuation Mechanism Actuation mechanisms in this category of soft grippers include passive structures with external motors, fluidic elastomer actuators, electroactive polymers and shape memory materials. Deformation in passive structures can be achieved via contact between the gripper and the object surface (contact driven) or pulling an embedded cable in the structure (tendon driven mechanism). The structure based on contact-driven deformation bends

16 5 when touching an object surface and grasps the object in response to reaction forces [2]. External motors provide the required actuation forces. Fin Ray grippes which are inspired from fish fins are examples of these types of grippers (Figure 1-2 (a)) [15], [16]. Traditional grippers based on tendon-driven deformation use rigid links, joints and springs [17] [20]. Recently tendon-driven mechanisms consist of soft elastic joints (i.e. Figure 1-2 (b)) [21] [25]. Elastic hinges can be beneficial in returning the actuated fingers to the initial position (stored bending energy). One important limitation of grippers with passive structures is miniaturization to make compact and portable grippers since they mostly use external source of actuation. Figure 1-2: grippers with passive structures. (a) contact-driven deformation working principle (left)[2], Fin Ray robotic gripper [16]. (b) Tendon driven mechanism working principle (left) [2], origami gripper with adjustable stiffness joints for multiple grasp modes [21] FEAs (Fluidic Elastomer Actuators) or SPA (Soft Pneumatic Actuators) have been used as grippers for decades. Advantageous of SPA grippers include ease of fabrication and low cost soft materials [11], [26]. They are also pretty robust. Actuation is achieved when a fluid (liquid or gas) is inserted to the chamber of a deformable elastomer structure.

17 6 Asymmetric geometry causes bending of the actuator due to inflation of the chamber. The generated forces of FEAs are high and proportional to the liquid pressure inside the gripper chamber. Blocked forces of 80 N at 300 kpa and 112 N at 200 kpa are reported in [27], [28]. Figure 1-3 shows the working principle of FEAs and different kinds of soft grippers developed in the literature. Figure 1-3: (a) working principle of FEAs [2]. (b) PneuNets [29] c) Fingers actuated by electrohydrodynamics [30]. (d) Self-healing polymers [31]. Electroactive polymers are polymers that reversibly deform in response to electric stimuli and called EAPs. Dielectric Elastomer Actuators (DEAs) are an important class of EAPs [32], [33]. DEAs consist of a thin elastomer layer (thickness up to a few hundred microns) sandwiched between two flexible and stretchable electrodes. High voltage activation of two electrodes causes electrostatic attraction, squeezing of the soft layer leading to its area expansion [34] [36]. Figure 1-4 (a) and (b) shows the principle of operation and a developed soft gripper based on DEA, respectively [2], [37]. Ionic Polymer-Metal

18 7 Composites) (IPMCs) are another type of electroactive polymers consisting of an electrolyte-swollen polymer layer sandwiched between two metallic layers. When no voltage is applied, anions and cations of the polymer is uniformly distributed and there is no deformation. When a voltage is applied to the electrodes, the cations move toward the cathode and anions toward the anode resulting in a bending deformation [38]. Figure 1-4: a) Working principle of Dielectric Elastomer Actuators (DEAs) [2]. b) A soft gripper based on dielectric elastomer structure [37]. c) Working principle of electroactive polymers [2] d) Soft gripper using IPMC fingers [39]. Shape memory materials in the category of Shape Memory Alloys (SMAs) and Shape Memory Polymers (SMPs) can recover their original shape from a deformed shape in response to a stimulus. SMPs are composed of elastic domains and transition domains. Heating and then cooling of a SMP result in stiffening of transition domain and blocking the deformation of the elastic domain. Now, heating the material again would remove the blocking force form the elastic domain and the SMP can recover the original undeformed

19 8 shape. SMA s shape memory effect happens due to phase change between martensite and austenite phases caused by temperature. Heating the material over the transition temperature would change the phase from martensite to austenite with a higher modulus such that it can recover the original shape. Figure 1-5 shows the working principle of SMPs and SMAs along with some soft grippers developed using these materials. Figure 1-5: (a) working principle of SMPs [2]. (b) Soft gripper using bidirectional shape memory polymer [40]. (c) working principle of SMA material [2]. (d) soft gripper with an elastomeric figure structure using shape memory alloy [41] Gripping by Stiffness Control This class of soft grippers can switch their stiffness between soft and rigid depending on the gripping modes. In the soft state it approaches the object with minimum compression applied to the object appropriate for grasping fragile objects. After grasping an object, the stiffness increases and the gripper goes to the rigid state to hold and move the object. This type of soft grippers consists of granular materials, Low melting point alloys, Electrorheological Fluid and Magnetorheological Fluids and also shape memory materials.

20 9 Change of the pressure between granules can switch the stiffness of the granular structure between soft and rigid states [42]. This stiffness control method is called granular jamming. Vacuuming a box of granules provide rigidity and once the air goes in to the box, the whole structure will be soft. The granular jamming mechanism can provide a rigidity change ratio up to 24 times [43], and it can solidify and liquefy in the order of (s) [44], [45]. Figure 1-6 (a) shows a developed soft gripper using granular jamming mechanism. Low melting point alloys (LMPAs such as Field s metal) are able to change their phases in low temperatures between This property can be used in developing rigidity tunable soft components when a low melting point alloy is embedded in an elastomer and heated up via an embedded soft heater or joule heating [46] [49]. In the solid state, their Young s modulus is in the order of 3-9 GPa. A typical soft gripper using LMPA embedded in PDMS (soft material) is demonstrated in Figure 1-6 (b). Electrorheological and Magnetorheological Fluids can change their stiffness in response to electric and magnetic fields respectively. ER fluids are composed of polarizable particles inside a dielectric fluid like on oil [50]. When exposed to an electric field, these particles form fibrillated chains resisting deformation resulting in an increase in their rigidity [51]. Similarly, when an MR fluid encapsulated in a soft shell is exposed to a magnetic field, an increase in their viscosity leads to a higher stiffness. The advantageous of ER/MR fluids are their response time (less than 10 ms) [51] and the broad range of stiffness change ratio [52] [54]. A couple of different soft grippers have been developed based on ER [55], [56] and MR [57], [58]. Shape memory alloys and shape memory polymers that can change

21 10 their stiffness due to a phase transition can be also included in this category (discussed in the previous section). Figure 1-6: (a) A soft gripper based on granular jamming mechanism [59] (b) A soft gripper through a combination of encapsulated LMPA inside a soft material and dielectric elastomer actuator effect [60] Gripping by Adhesion Control Contact between two surfaces generates shear stress proportional to the applied normal pressure. The generated shear friction force is the key component in soft grippers to provide a holding force [2]. The larger friction force provides the higher pull-off force which is called the adhesion force. Soft grippers based on the adhesion control have unique advantages comparing to previously discussed grippers. In this method normal force to the object in gripping process is much smaller than previous gripping mechanisms for example gripping by actuation, and therefore it can manipulate fragile objects easily. Due to low power requirement they can be made as portable lightweight grippers. Normal adhesion force is beneficial for grasping flat objects (i.e. single-point grasping) while grippers based on actuation have difficulty with it. This type of adhesion is useful for manipulating soft and deformable objects as long as the gripper structure is compliant enough. Controlled adhesion mechanism consists of electroadhesion and dry adhesion.

22 11 Electroadhesion is the electrostatic forces between two surfaces subjected to an electric field. Electrostatic force, also called Coulomb force is the attraction or repulsion of two surfaces due to induced electric charges. Attraction is between positive and negative electric charges and the amount of attraction can be tuned with the amount of electric charges. Applying an electric field to compliant electrodes embedded in a soft dielectric elastomer induce opposite charges in the object resulting in an attraction force for grasping. The electroadhesion is an appropriate grasping method for rough and smooth surfaces [61]. Since a strong electric field is needed to induce electric charges, a few kv voltage is necessary for this gripper technology. Geometry, size of the electrodes and the thickness of dielectric (insulation) layer are important parameters to maximize the required adhesion forces [62], [63]. Figure 1-7 (a) shows a soft gripper introduced by Jun Shintake et al. based on electroadhesion with the ability to manipulate a variety of objects. Geckos are the best climbers in the world. They can climb rough, smooth, inverted or vertical surfaces easily. This climbing ability comes from nanofibers on the tip of their fingers as shown in Figure 1-7 (b). Each fiber alone cannot provide a high force, but a bunch of millions of them can have a high adhesion strength. This concept has been inspired to develop adhesive systems that are durable and low cost. As the underlying mechanism for this type of adhesion, also called dry adhesion, normal pressure of microfibers to the objects surface generate a van der Waals force resulting in a shear stress at the adhered interface [64] [66], [67]. Geckoadhesive systems can be applied to smooth and rough surfaces [68]. They have applications in wall climbing robots [69], wearable adhesive pads [70] for humans and object manipulations [71]. Figure 1-7 (c) shows a

23 12 pattern of mushroomed shaped microfibers on an elastomer membrane has the ability to pick up and hold a variety of objects [72]. Soft grippers with dry adhesion technology have good versatility, applicable to handle a range of objects, mostly rigid with smooth surfaces. For rough surfaces some optimizations are needed on the geometry and material softness of the microfibers [68], [72]. Manipulations of soft and deformable structures using geckoadehsive system might be also challenging. Different classes of smart materials (such as those ones with negative Poisson s ratio) [73] [80]are being widely used in soft robotic grippers. Soft components can also be used as an application to reduce noise and vibration level in automotive industries [81] [83]. Fracture mechanics of a thin rod in a soft medium has also been investigated to predict the rod buckling behavior [84], [85]. Figure 1-7: (a) Soft gripper using electroadhesion mechanism can handle a variety of objects[2]. (b) Gecko adhesive system from macro to nanostructures. (c) Dry adhesive system with mushroom shaped microfibers handle different objects [72]. 1.3 Dry Adhesion Recently, it has been shown that geckos move based on normal and frictional (lateral) adhesion [86] [88]. Tilted fibers of geckos respect to the surface is beneficial for attachment-detachment cycles since normal and frictional adhesion forces are coupled.

24 13 Figure 1-8 shows the shape of the contact tips (circle) in different animals from beetles to geckos [89]. It can be seen that the heavier animals have finer and dense micro structures. Miniaturization of the fibrils is limited by: Fiber Fracture, Ideal Contact Strength, Fiber Condensation and Contact Adaptability [89], [90] Fabrication Figure 1-8: Contact tips shape of fibrillary pad in animal [89]. Inspired by geckos, a pattern of soft microfibers can significantly enhance the dry adhesion or attraction forces between two opponent surfaces. The shape of fibrils include flat, tilted spatula-ended fibrils and hierarchal fibrillary structures. Microfibrillar surfaces are mostly fabricated by casting PDMS (Polydimethylsiloxane) on microfabricated molds made by soft-lithography [91], [92]. Scanning Electron Microscopy (SEM) image of PDMS microfibrils fabricated by soft molding on SU-8 photo-lithographic templates is shown in Figure 1-9 (a). Microfibers have radius of 2.5 µm and height of 20µm. Nanofibrillar surfaces can be shaped by melting a polymer through conformal contact of a nanostructured

25 14 mold with heat and pressure. A SEM image of polymethylmethacrylate (PMMA) nanofiber structure made via PUA mold is shown in Figure 1-9 (b). Figure 1-9: Scanning electron microscopy (SEM) images of (a) PDMS microfibers fabricated by selfmolding photolithographic [93]. (b) PMMA nanofibers fabricated by hot-embossing and elongation by mold removal at high temperatures [64], [93]. A mold of silicon rubber consisting of angled holes is fabricated by soft molding to achieve tilted fibrillary structures [94], [95]. A tilted microfiber structure with a tilting angle of 25 and fiber dimater of 8 µm is shown in Figure 1-10 (a). Hierarchical structures with groups of stacked fibers were obtained by multi step photolithography using SU-8. Figure 1-10 (b) shows a hierarchical structure made with PDMS through double molding technique [96], [97].

26 15 Figure (a) The SEM image of tilted fibers fabricated by soft molding on the SU-8 wafer. (b) Hierarchical structure fabricated by double molding photolithography. Fabrication of fibers with 3D tip shapes can be achieved depending on the various conditions for curing and molding. Shapes of planar, spherical, symmetric, and asymmetric are different shapes of fiber tips to enhance the adhesion strength. Figure 1-11 shows different steps to make planar fibers using simple molding, spherical tip via upside-down curing to make circular shape at the tip, curing with a little pressure towards a flat surface to make mushroom shaped fibers. The fiber can be asymmetric when the flat substrate is tilted and then cured.

27 16 Figure Fabrication steps of micropatterns with different 3D tip shape [64], [98] 1.4 Reversible and Tunable Dry Adhesion Tunable and reversible dry adhesion mechanisms can provide strong adhesion under one set of conditions and weak or reduced adhesion under another. Surfaces with tunable or switchable adhesion have a broad set of applications, ranging from transfer printing of semiconductor elements and climbing robots to gripping in pick-and-place manufacturing [99] [102]. Recently, different methodologies have been proposed to have tunable and dynamically reversible dry adhesion strength. Advantageous and weaknesses of different methods will be discussed in the next chapter and in this study a novel approach will be introduced and presented for dynamically tunable dry adhesion strength up to 6.

28 17 Chapter 2 Background 2.1 Introduction A brief introduction on adhesion fracture mechanics, crack initiation and propagation will be presented in this chapter. Different methods of achieving reversible dry adhesion that have been presented so far, advantageous and disadvantageous of each method will be discussed. Then, in order to overcome limitations of previous methods for tunable and reversible dry adhesion, a novel method will be introduced. We address the methodology of the proposed mechanism. Next chapters cover fabrication, simulation and adhesion measurements Results. 2.2 Mechanics of Adhesion Two adhered interfaces can be detached whenever a crack forms and propagates through the whole interface. In order to have a crack, the applied energy breaks several atomic bonds (brittle materials) and also a part of it dissipates near the crack tip as a plastic deformation (metallic materials). The total energy to have a crack length a is [103]: E bond = 2 (γ s + γ p ) a B 2-1 In which γs is the energy required to break atomic bonds per unit surface area created by the crack. The surface area is a B where a is the crack length and B is the part thickness. As the crack grows, two free surfaces will be appeared such that s in γs stands for it. γp is the energy due to plastic deformation at the crack tip per unit surface area.

29 18 Whenever an external stress is being applied to a part, strain energy is stored in the item. Once a crack forms and propagates, it is shown by Griffith that the stored mechanical strain energy for an infinite plate decreases according to [103]: U = σ2 σ2 V B π a2 2E 2E 2-2 σ is the applied stress, E is the Young s modulus, V is the part volume, B is the part thickness and a is the crack length. The total energy in the system is equal to the summation of Eqs. 2-1 and 2-2: E total = 2 (γ s + γ p )ab σ2 σ2 V B π a2 2E 2E 2-3 Figure 2-1: The values of atomic bond energy, strain energy and total energy for different crack lengths.

30 19 Figure 2-1 shows the atomic bond energy, strain energy and the total energy of the system. For the total energy (red line) it can be seen that for short crack lengths as the crack grows the total energy increases to a maximum value, so an additional energy is needed for crack growth. The crack is stable. However, for larger cracks an increase in crack length leads to decrease in total energy and the crack propagates with no additional energy. In this case the crack is not stable. Taking derivative of the total energy gives us the critical value of the applied stress in which 2 (γ s + γ p ) = G c and Gc is called Griffith Critical Energy Release Rate [103]. Gc is a material property and independent of the applied loads and geometry of the body. For example Gc of glass is around 7 J/m2. Adhesion is bonding of two surfaces and for their separation a crack needs to form and propagate leading to a complete detachment of surfaces. Fracture mechanics is important to help us characterize the failure mechanism of dry adhesion of two surfaces. Figure 2-3 shows 3 basic fracture modes of 2 surfaces. Mode I is the opening mode caused by tension load. Mode 2 is in-plane shear and mode 3 is out of plane shear. Critical energy release rate for a body experiencing mode 1 of fracture is smaller than that of mode 2 and 3 for corresponding loadings [103]. The energy release rate failure criterion states that whenever G Gc, a crack can grow.

31 20 Figure 2-2: Fracture modes (left to right: opening, in-plane shear, out of plane shear) 2.3 Tunable and reversible dry Adhesion When an adhesive pad has the ability to tune its stickiness to an opposing surface between high and low adhesion strength, it is called reversible dry adhesion mechanism. In this research we have only focused on dry adhesion mechanisms not wet ones. The potential applications of reversible dry adhesion mechanisms have been thoroughly discussed in the previous chapter. One strategy to provide adhesion tenability is kinetic control [104] [106]. Figure 2-3 shows the critical energy release rate for different temperatures versus the peeling velocity. Peeling velocity can be used to control the adhesion strength or G.

32 21 Figure 2-3: Critical energy release rate vs peeling velocity in different temperatures for PDMS material [104]. Switching modes of loading is the other way to tune the adhesion strength [106] [109]. Minsky et al. have shown that applying a shear load at the adhered interface can significantly alter the stress distribution leading to a change in crack initiation location (moving from an edge crack to a center crack) [107].

33 22 Figure 2-4: Normalized normal stress vs a shear displacement for a soft composite post with an embedded rigid core [107]. Another method of achieving tunability of the dry adhesion strength is to use fibrillary structures with angled or asymmetric geometries [110] [114]. Asymmetric geometry of fiber tips leads to asymmetric stress distribution at the contact interface [115] due to moment created when the tip is sheared. Figure 2-5 shows SEM images of fiber structures with mushroom tips.

34 23 Figure 2-5: Scanning electron microscope images of arrays of 35 microns with mushroom tips. Angles: (a) 34º (b) 90º (c,d) 23º [116]. Any rotation of the tip with respect to its original shape creates a peeling moment that alters the stress distribution at the tip. Figure 2-6 illustrates the bending behavior of the fiber tip in a compression state and when the fiber has a shearing load in the gripping and releasing directions as well.

35 24 Figure 2-6: Tip behavior with different loadings. (a) Original unloaded condition (b) Preload compression. (c) Shearing in the releasing direction (d) Shearing in the gripping direction [116] Applying a magnetic field to a bioinspired micropattern can be employed to have a reversible dry adhesion mechanism [117] [119]. Under a magnetic field microfibers tilt to different directions leading to a change in the adhesion strength. Figure 2-7 shows pull-off forces of a magnetically driven dry adhesion system when the magnetic field gets off and on consecutively [117].

36 25 Figure 2-7: Pull-off forces when magnetic field is off and on. Applicable in dry and wet adhesion [117] Figure 2-8 illustrates the tilting behavior of fibers in presence of a weak and strong magnetic field. Figure 2-8: (A) tilting of magnetic pillars to opposite directions when the magnet is approached from opposite side. (B) Behavior of pillars under a stronger magnetic field. (Scale bar is 20 microns) [117].

37 26 Some mechanisms have been proposed with thermally controllable adhesion [120] [122]. A thin film micro-fibrillar adhesive made of adhesive polymers and shape memory polymers has been introduced to have reversible dry adhesive actuated by thermally tunable adhesion [121]. This tunablity in adhesion comes from temperature effects and a change in the stiffness of sub-surface shape memory polymer Figure 2-9: Pull-off forces of thermally actuated mechanism for different preloads [121]. Humidity can also be used to control the dry adhesion [123] [126]. It has been shown that adhesion forces enhanced significantly in 100% humidity and diminished when immersed in water [123]. Figure 2-10 shows the adhesion behavior between a setal array and a silica surface for dry, 100% humidity and immersed in water.

38 27 Figure 2-10: Dry adhesion behavior under dry, 100% humidity and immersed in water conditions [123]. Laser excitation (ER-YAG laser) [127], changing the contact area [128] [130], phase change of the backing layer (embedded low melting point alloy) [131], [132] have also been investigated to have tunable and reversible dry adhesion mechanism. Figure 2-11 shows an adhesive system using a tunable rigidity backing layer (crystalbond) that can tune its stiffness and accordingly the adhesion force [131]. It shows that the soft backing layer has the lowest adhesion force comparing to other cases.

39 28 Figure 2-11: (a) Schematic of an adhesive system with a tunable stiffness backing layer. (b) Adhesion measurement of different states of rigidity of the backing layer [131]. These existing methods have been limited by complex fabrication processes [87], [95], [107], [111], [112], [114], [116] [121], [123], slow actuation [120] [126], [131], [132], insufficient change in adhesion between the low and high adhesion states [106], [107], [111], poor reversibility [95], [106], [107], [114], [120], [127] and/or high constraints on the working environment [117] [119], [123] [126], [128], [130]. Recently, it has been shown that enhanced and tunable dry adhesion can be achieved via composite post structures consisting of a stiff core and a compliant shell; the composite structure alters the stress distribution at the adhered interface and the effective adhesion strength [107], [133]. Figure 2-12 shows schematic of a composite post (rigid core embedded in a soft shell) that can significantly enhance the dry adhesion strength.

40 29 Figure 2-12: (a) schematic of a rigid core embedded in a soft shell. (b) Homogenous post (c) composite post when the rigid core is close to the interface [Helen s paper] 2.4 Tunable stiffness mechanisms Previous studies have shown that inserting a rigid core in a soft shell can increase the dry adhesion strength [107], [133]. In this study, we are investigating the effect of rigidity tuning of subsurface on the dry adhesion strength. Regarding the tunable stiffness mechanisms, in the field of soft robotics, there has been some recent work in multifunctional materials with tunable stiffness that enable soft machines with unmatched deformability and adaptability [134]. There are two general approaches to dynamically alter the stiffness of a material. One approach involves the dynamic addition and subtraction of fluids in a composite [134]. As an example, it is inspired by the sea cucumber that a composite made of cellulose nanofibers and a rubber matrix can tune its modulus by a factor of 40 upon exposure to water or isopropanol [134].

41 30 The other method uses external stimuli, such as electric [134] [136] /magnetic fields [119], temperature [136] [138], light [139], or a combination of these, to induce phase change [10], [134], [136], [138], [140], [141]. Both approaches in stiffness tuning have their own advantages and disadvantages, however the latter group, especially electrically-actuated approaches, enable more compact systems and are easier to incorporate since they do not involve bulky supporting facilities such as pneumatic valves. Shan et al. have proposed a polymer with a weight composition of 51/9/40% propylene, ethylene and structured carbon black that can be electrically activated. Electrical activation of the polymer embedded in a soft shell heats up the polymer and whenever it passes the melting point (72.9 ), the Young s modulus drops from 37 MPa to 1.5 MPa. Figure 2-13 shows the stiffness tunablity of the CPBE (Conductive Propylene Based Elastomer) embedded in a soft shell [136].

42 31 Figure 2-13: CPBE embedded in PDMS soft shell. After activation with electrical current, the composite post stiffness reversibly softens to 10 % of its original value [136]. 2.5 Project Here, we introduce a new concept in switchable adhesion that exploits two recent advances: (1) development of core-shell posts that have enhanced dry adhesion due to their composite structure [107], [133] (2) development of conductive propylene-based elastomer (CPBE), which is a polymer-based material that undergoes a significant decrease in stiffness when electrically activated [136]. A composite post with a stiff core and compliant shell has enhanced adhesion relative to a homogenous compliant post as the presence of the stiff core alters the strain energy release rate and the stress distribution at the interface, resulting in a higher effective adhesion strength [107]. Here, the core is made of a material with a

43 32 reversibly tunable stiffness, CPBE [134], [136]. Applying electrical voltage to the CPBE core causes resistive heating of the core and a significant drop in modulus when it is heated to near or above its transition temperature of 72.9 C [140]. The shell of the post is made of PDMS with a modulus of approximately 2.1 MPa [107], while the CPBE core has a Young s modulus of approximately 175 MPa [136] in the non-activated state and a modulus of <1 MPa in the activated state. Thus, in the non-activated state, the post has a stiff core and a compliant shell which leads to a high effective adhesion strength, but once activated, the stiffness of the core becomes comparable to the shell and the adhesion strength is reduced. The adhesion of the post in the non-activated state can be manipulated by altering the position of the CPBE core relative to the interface during fabrication process. In the activated composite posts, adhesion strengths can be controlled through the magnitude and duration of the applied activation voltage. Figure 2-14 illustrates the working principle of the proposed gripper developed based on tunable dry adhesion via subsurface stiffness modulation. When the tunable stiffness composite post is not activated (stiff), it has a high adhesion force and can pick up a typical object (with a flat and smooth surface). In the next step, whenever the composite post gets activated (compliant state), the object will be released due to a low adhesion force in the compliant state.

44 Figure 2-14: The proposed mechanism consisting of rigidity tunable core embedded in a soft shell can be used as a soft gripper. When the composite post gets activated, it softens and low adhesion is the reason to release the object quickly. 33

45 34 Chapter 3 Design and Fabrication 3.1 Introduction In this project, in order to fabricate a tunable stiffness composite post, molding fabrication technique has been employed. The composite post has two main parts: a soft shell made of PDMS and tunable rigidity core made of CPBE material. The stiffness of the CPBE core drops when it is activated with an electrical voltage and its temperature passes After being activated, the stiffness of the composite post would be in the order of that of control sample (pure PDMS post). Since we want to measure dry adhesion forces, a smooth and flat PDMS interface is needed. This is achieved with curing the PDMS interface exposed to an acrylic sheet. It should be mentioned that the distance between the rigid core and the interface is a critical parameter for the dry adhesion forces of the non-activated composite posts. 3.2 Design It has been shown that embedding a rigid core inside a soft shell can increase the adhesion strength [133]. Figure 3-1 shows the schematic of the cylindrical composite post developed by Helen Minsky et al. It has been proved that when Ecore Eshell (~10 ) and 0.1 < t < 1 R (t is the distance between the interface and the rigid core, R is radius of the cylindrical post) the maximum stress location moves form the edge (large t/r ratios) towards the center (for composite posts with 0.1 < t < 0.5). This change in the stress distribution pattern can R increase the adhesion pull-off forces by a factor of as high as 3.

46 35 Figure 3-1: (a) (1) Schematic of the composite post (2) control sample (3) Composite post sample (b) Ratio of the composite post pull-off forces to that of control sample for different values of t/r. As t/r decrease crack is moving from the edge towards the center. Here, in this project the adhesion force is tuned via stiffness modulation of the composite post. To do this, a rigidity tunable material (CPBE) is embedded in a soft shell (PDMS). Applying electrical voltage to the core, heats it up and once the temperature passes 72.9 it melts and gets softened [139]. The developed composite post has a rectangular cross section with dimensions of 3 mm 5 mm. Figure 3-2 shows the schematic and the final composite post sample. Copper wires are attached to the ends of the CPBE core to allow for electrical connection. L and W are the half lengths of the long and short edges, respectively (Figure 3-2 (a)) and the thickness of the PDMS shell near the adhered interface is t (Figure 3-2 (b)). The fabricated posts have overall dimensions of 3 mm (2W) 5 mm (2L). The soft layer thickness, t, is varied from 27.5 µm to 875 µm. The U-shaped CPBE strip has cross-sectional dimensions of 2.4 mm 1.5 mm, a total length of 38 mm, and an electrical resistance of approximately 400.

47 36 Figure 3-2: (a) Isometric, front and side views of the schematic of the composite post with a U-shaped tunable rigidity core. (b) Schematic of electrical connection and photograph of a composite post sample, consisting of PDMS with an embedded U-shaped CPBE core and copper wires for electrical connection [142]. The shell of the post is made of PDMS with a modulus of approximately 2.1 MPa [107], while the CPBE core has a Young s modulus of approximately 175 MPa [136] in the nonactivated state and a modulus of <1 MPa in the activated state. Thus, in the non-activated state, the post has a stiff core and a compliant shell which leads to a high effective adhesion strength, but once activated, the stiffness of the core becomes comparable to the shell and the adhesion strength is reduced. The adhesion of the post in the non-activated state can be manipulated by altering the position of the CPBE core relative to the interface during fabrication and the change in adhesion strength can be controlled through the magnitude and duration of the applied activation voltage.

48 Fabrication Method The tunable stiffness core is made of CPBE, the stiffness of which can be tuned through the application of electric voltage [10], [136]. It has been shown that the CPBE has a glass transition temperature below -60, a melting peak at 72.9 ±0.07, and a crystallization peak at 59.8 ± 0.07 [140]. So when the CPBE temperature reaches 72.9, it melts and after cooling down its stiffness goes back to the original value. To make the U-shaped core, CPBE pellets are compacted in a heat press machine (Carver 4389) at 100 under a pressure of 5 psi for 24 hours to form a CPBE sheet with a thickness of 2.4 mm. A CO2 laser cutter (Epilog Helix 24, Epilog Laser Inc.) is used to pattern the U-shaped strips from the CPBE sheet. Figure 3-3 shows different steps of fabricating a U-shaped CPBE core [136]. Figure 3-3: Different steps of fabricating a U-shaped tunable rigidity core[136].

49 38 Now, the U-shaped CPBE core is ready to be embedded in a soft shell. Molding fabrication technique is used to embed the CPBE core in a PDMS shell. Figure 3-4 shows mold 1 and mold 2 that are used in a layer by layer fabrication technique that will be explained later in this chapter. Mold 1 is the first mold to design a channel to easily place the U-shaped core inside the soft layer derived from mold 1. Second mold is deeper and is designed to position the PDMS layer (derived from the first mold) and CPBE core (inside the channel) for final molding process. It should be mentioned that an acrylic sheet is placed at the bottom of mold 2 such that a smooth and flat surface is achieved at the bottom of final composite post. Figure 3-4: (a) Mold 1 to cast and cure PDMS with an embedded channel (b) Mold 2 which is deeper to fabricate the final post. An acrylic sheet is positioned at the bottom of final post to achieve a very smooth and flat PDMS interface. The composite posts were fabricated with a multi-step additive manufacturing procedure (Figure 3-5). The CPBE core (U-shaped strip) was embedded in the PDMS post. The molds for casting were printed with a 3D printer (Objet24, Stratasys Inc.). PDMS (Sylgard 184, Dow Corning Corporation, Midland, MI), with a 10:1 weight ratio of base elastomer to curing agent, was mixed in a Thinky Mixer (AR-100, THINKY Inc.) for 10 mins. It was then degassed using a vacuum oven (Across International, AccuTemp) for 10 mins. The first mold was filled with PDMS, degassed for 10 mins. The extra PDMS was removed by dragging a glass slide over the surfaces and the PDMS was then cured at 80 for an hour.

50 39 This resulted in a layer of PDMS with a thickness of 1.5 mm with channels defined by the features on the bottom of the mold. The CPBE strip, with copper lead wires attached, was placed inside the channel of the PDMS layer and then the whole assembly was positioned into the second mold. The second mold was slightly longer than the first mold. A smooth acrylic sheet was placed at the bottom to ensure a flat and smooth surface at the bottom of the final PDMS post. Uncured PDMS was poured in the second mold and degassed for 10 mins, additional PDMS was removed, and the composite post was cured in an oven at for 24 hours. A lower curing temperature was used to stay below the melting temperature of the CPBE [140]. Figure 3-5: Fabrication process of the soft composite post. Note that in Step 3 an acrylic sheet with a smooth surface is placed at the bottom of a longer mold to achieve a smooth PDMS surface at the bottom of the post (i.e., the surface that forms the adhered interface). Now, the final composite post is fabricated and ready for adhesion measurement. In the next chapter, experimental setup for adhesion measurement will be discussed.

51 40 Chapter 4 Testing Method and Setup 4.1 Introduction To test reversible tunable dry adhesion of the developed composite posts, a fixture to hold the composite posts was designed and 3D printed to be installed onto the Instron machine. Adhesion measurement was conducted for both activated and non-activate states of the composite posts. For the non-activated state, the post is relatively rigid. While, when the post is activated with an electrical voltage, it gets softened and compliant affecting the composite post dry adhesion strength to an opposing glass substrate. 4.2 Experimental Setup Testing of the adhesion strength of the composite posts was performed on an Instron 5969 with a 50 N load cell installed. A power supply (GW INSTEK GPR-30H100) was used for activation of the composite posts. The PDMS post was inserted and clamped into the test fixture which was attached to the load cell of the Instron while a glass slide (substrate) was installed on the bottom platform. The final testing setup can be seen below in Figure 4-1 & Figure 4-2. Note that the 50 N load cell is connected to the bottom of 50 kn load cell for precise measurement.

52 Figure 4-1: The experimental setup used for all the testing. A 50N load cell was installed for more precise measuring. 41

53 42 Figure 4-2: Close-up of the testing area showing exposed copper electrodes, fixed glass substrate, and fixture used to attach the post to the Instron. Testing parameters for all samples were the same. In a typical adhesion experiment, the composite post was brought into contact with the glass at a speed of 10 µm/s to reach a compressive preload of 2 N. The post was held at this preload for 2 mins and then it was retracted at a speed of 5 µm/s to a compressive preload of 0.6 N. The U-shaped CPBE strip was activated with an electric voltage for 2 mins. Finally, the post was retracted at 50 µm/s. The adhesion force is defined as the peak tensile force measured during retraction.

54 Testing Conditions The adhesion strength of the composite post to a smooth glass substrate was measured as a function of two parameters: (1) the activation voltage of the CPBE strip (2) the t/l ratio of the post. Activation voltage for adhesion measurement was 20, 30 and 40 Volts. t/l values for all posts changed from to L is half of the long edge. Temperature of the adhered interface was also measured by placing a thermocouple at the interface. These results that will be discussed in the next chapter confirm that 2 mins activation time with 20, 30 and 40 V does not damage the post. It should be also mentioned that 3 mins activation time with 40 V voltage is probable to burn polymeric U-shaped core. In the next chapter, a finite element simulation is conducted to find the stress distribution pattern at the interface. The location of maximum stress changes form the edge towards the center for activated and non-activated composite post respectively. All experimental results will also be discussed in the next chapter.

55 44 Chapter 5 Results and Discussion 5.1 Introduction ANSYS software is used to perform Finite Element (FE) modeling and derive the stress distribution pattern at the adhered interface. This stress distribution pattern is helpful to compare it with experimental results and justify the reason that inserting a rigid core causes crack formation at the center instead of the edge. This chapter also discusses all adhesion strength measurement results, pull-off forces and post stiffness for different activation voltages and t/l values. Supplemental video S1 and S2 show the formation of a crack for activated and non-activated posts, respectively. Also as a potential application of the developed reversible dry adhesion mechanism, it is shown in supplemental video S3 that the composite post can pick up an object (stiff state) and whenever needed releases the object (compliant state). At the end, all results will completely be discussed and analyzed. 5.2 Simulation Results Finite element (FE) modeling, using ANSYS, was performed to examine the stress distribution across the adhered interface for composite posts with different geometries. The model geometry matched the experiments, however quarter symmetry was exploited to reduce the model size. The thickness of the PDMS shell close to the interface, t, was varied between 27 µm and 875 µm, to match the t/l ratios that were experimentally investigated. The specific mesh for each model was determined through a convergence study. PDMS and CPBE were assumed to be isotropic, homogeneous, and linearly elastic with Poisson s ratio of 0.49 and 0.4, and Young s modulus of 2.1 MPa and 175 MPa (non-activated), respectively [107], [136].When activated and compliant, ECPBE was assumed to be equal to

56 45 EPDMS. A fixed displacement boundary condition was imposed to the adhered interface of the post. The interface between the PDMS and the CPBE core was assumed to be perfectly bonded. Normal loading was applied by displacing the nodes along the top of the post an equal amount in the z-direction, with the x, y directions free to displace. Figure 5-1 shows the FE-predicted distributions of normal stress at the adhered interface. The exact stress values at the edge of the post are sensitive to the mesh size due to the presence of a singularity. Regardless, the FE results provide an understanding of how the normal stress distribution at the adhered interface changes with the post geometry and the stiffness of the core. At large t/l ratios, the stress is highest at the edge of the post (Figure 5-1 (a), t/l = 0.35). As the t/l value is decreased, the stress at the center increases and the stress at the edge decreases. At a certain t/l value, the peak stress will shift to the center and continue to be there as t/l is decreased (Figure 5-1 (b), t/l = 0.12), similar to the results observed in Minsky et al. [107]. However, if t/l is decreased further, the stress peak moves away from the center toward the edges of the rectangular projection of the CPBE core onto the adhered interface (Figure 5-1). This suggests that the location of the highest normal stress at the adhered interface, for the elastic modulus mismatch range of the materials used in this study (ECPBE/EPDMS ~ 84), can move to a point between the edge and the center of the post bottom surface. As the edge regions are most likely to have defects that facilitate crack initiation at the interface, lower stresses near the edges make it is more difficult for a crack to initiate and, consequently, result in a higher adhesion force. Figure 5-1 (d) shows the normalized stress distribution at the adhered interface for an

57 46 activated and compliant post with t/l=0.12. After activation, the stress is highly localized at the edge and that the normalized stress values are significantly higher. Figure 5-1: Finite element-predicted distributions of normal stress on the adhered interface of the composite post. The stresses are normalized by dividing the local stress by the average stress at the adhered interface. Results are shown for one quarter of the contact area of the post; the right and bottom edges of the plots are symmetry planes and the center of the post is located at the origin of the coordinate frame (lower right corner of each plot). Results for (a) t/l = 0.35, non-activated. (b) t/l = 0.12, non-activated. (c) t/l = 0.011, non-activated. (d) t/l = 0.12, activated [143].

58 Experimental Results The adhesion strength of composite posts with various layer thicknesses, in non-activated and activated (compliant) states, were experimentally measured. All tests were run in displacement control. For cases in which the core was activated, the activation time was fixed at 2 mins and the activation voltage was varied from 20 V, 30 V, to 40V. Activation voltage and duration are important because higher voltage and longer duration can lead to damage of the core material. The voltages and duration used here are chosen such that there is no damage of the core, which will be shown via experimental data later in the paper. Furthermore, the lower voltages and a relatively long activation time of 2 mins were purposefully chosen over higher voltages with shorter activation times in the adhesion characterization measurements in order to minimize the effect of variations in the exact timing of the mechanical experiments on the adhesion measurements. Later in this chapter, we show that higher voltage can be applied for shorter times (<30 s) and that the higher voltages lead to an adhesion change over a relatively short time period. Figure 5-2 (a) shows a comparison of the adhesion forces measured on non-activated composite posts with varying t and the adhesion of a homogenous PDMS post. Posts containing the stiff core have enhanced dry adhesion strength compared to a homogenous PDMS post; this is expected based on previous reports of the adhesion of composite core-shell structures [107]. The adhesion enhancement due to the addition of the core can be attributed to the reduced compliance of the system, as explained in Bartlett et al. [144], as well as the effect of the core on the stress distribution at the interface, which can affect where along the interface (edge or interior) the crack initiates [107]. Fracture mechanics-based models in Minsky et al. [107] account for the effects of both the compliance and the stress distribution

59 48 on effective adhesion strength. The homogenous PDMS post with a rectangular cross section of 3 mm 5 mm had an adhesion force of 0.52 ± 0.03 N, while the composite posts with the same cross-sectional geometry and 0.12 < t/l < 0.15 had adhesion forces between 1.21 ± 0.06 N and 1.56 ± 0.07 N, which is about 2 to 3 enhancement relative to the homogenous PDMS post. The adhesion is higher when the compliant layer at the interface is thinner; for t/l= 0.011, the adhesion force is 2.12 ± 0.14 N, which is > 4 that of the homogenous PDMS post. Figure 5-2: Adhesion forces of the non-activated composite posts with different t/l values. All tests were performed five times. The standard deviation for pure PDMS post adhesion force measurements is 0.03 N [142]. The U-shaped CPBE core is activated (i.e., made compliant) and then allowed to cool to obtain reversible adhesion tuning. Figure 5-3 shows the adhesion strength of the activated case relative to the corresponding non-activated cases in Figure 5-2. More details about the

60 49 adhesion strength measurement procedure can be found in the subsequent experimental section. Results show that the reduction in adhesion is a function of the activation voltages and the t/l, and that, in general, the adhesion strength is reduced more in posts with smaller t/l (Figure 5-3). For 0.12 < t/l < 0.15, the adhesion strength is reduced to between 38 % and 74 % of the non-activated cases, depending on the activation voltages. At t/l of 0.35, the adhesion strength is reduced to between 64 and 77 % of the non-activated values. For the post with the smallest t/l ratio, t/l = 0.011, the adhesion strength reduction is the most significant, and is as low as 17 % of the non-activated values (at 40 V), which is a reduction of approximately 6. Interestingly, the adhesion strength of the t/l = post in the activated and compliant state is even lower than the adhesion strength of the homogenous PDMS post. At 40 V, the CPBE strip is believed to be fully activated, whereas for 20 V and 30 V, the CPBE strip is only partially activated. Note, there is no significant change in adhesion strength values after multiple cycles of activation for all cases and that the adhesion force returns to the un-activated value after cooling. Additional cyclic performance testing was conducted on a sample with t/l = 0.22 for 20 cycles of activation (40 V for 2 mins) and subsequent cooling at room temperature. The adhesion change ratio was measured to be 2.04 ± 0.24 over 20 cycles, indicating no significant change in performance with repeated activation.

61 50 Figure 5-3: Percent remaining adhesion strength of the composite posts as a function of activation voltages. In all adhesion experiments, the composite post is activated for 2 minutes [142]. Experimental measurements also show that the pull-off force and consequently the adhesion strength of the fabricated composite post depends on the t/l ratio and the activation voltage of the U-shaped CPBE strip (Figure 5-2 & Figure 5-3), which is consistent with the presented FE simulations in section 5.2. The stiffness of the composite post under different activation conditions was characterized (Figure 5-4). Here, the stiffness is measured when the post is being retracted from the contact surface and is defined as the slope of the unloading curve prior to pull-off. For all the results in Figure 5-4, one composite sample with a t/l = 0.12 is used, and the sample was activated for 2 mins, cooled down for 10 mins to ensure it returned to room temperature, and activated again to observe the relationship between stiffness and activation voltage. Activation voltages were varied from 5 V to 55 V in increments of 5 V such that the voltage range covers the three voltages used in the adhesion characterization