©2011 Liehui Ge All Rights Reserved

Home , Sphaerodactylus, Synthetic setae

LIEHUI GE

SYNTHETIC GECKO ADHESIVES AND ADHESION IN GECKOS

A Dissertation

Presented to

The Graduate Faculty of The University of Akron

In Partial Fulfillment

of the Requirements for the Degree

Doctor of Philosophy

Liehui Ge

May, 2011

ABSTRACT

Geckos’ feet consist of an array of millions of keratin hairs that are hierarchically

split at their ends into hundreds of contact elements called “spatula(e)”. Spatulae make intimate contacts with surface and the attractive van der Waals (vdW) interactions are

strong enough to support up to 100 times the animals’ bodyweight. Tremendous efforts have been made to mimic this adhesion with polymeric materials and carbon nanotubes

(CNT). However, most of these fall short of the performance of geckos. “Contact splitting principle”, based on Johnson–Kendall–Roberts (JKR) theory, predicts that a vertically aligned carbon nanotube array (VA-CNT) will be at least 50 times stronger than gecko feet. Although 160 times higher adhesion was recorded in atomic force microscopy (AFM) measurements, macroscopic VA-CNT patches often show low or even no adhesion to substrates.

The behavior of VA-CNT hairs near the contact interface has been explored using a combination of mechanical, electrical contact resistance, and scanning electron microscopic (SEM) measurements. Instead of making the expected end contacts, carbon nanotubes make significant side-wall contacts that increase with preload. Adhesion of side-wall contact CNTs is determined by the balance of adhesion in the contact region and the bending stiffness of the CNTs, thus a compliant VA-CNT array is required to make adhesive patches.

iii Macroscopic patches of compliant VA-CNT array have been fabricated. Patches

of uniform array have adhesive strength similar to that of geckos (10 N/cm2) on a variety

of substrates and can be removed easily by peeling. When the array is patterned to mimic the hierarchical structures of gecko foot-hairs, strength increases up to four times. VA-

CNT-based gecko adhesives are self-cleaning, non-viscoelasticity and give good strength in vacuum. These properties are desired in robotics, microelectronics, thermal management and outer space operations.

Current theory still cannot completely explain adhesion of gecko feet. A series of experiments have been carried out to measure adhesion at different temperatures using a single protocol with two species of gecko that had been previously studied (G. gecko and

P. dubia). Strong evidence of an effect of temperature was found but the trend was counterintuitive given the thermal biology of geckos and it violated the prediction by van der Waals interactions. Consequently, other factors (e.g., humidity) that could explain the variation in the observed clinging performance were examined. Evidence was found, unexpectedly, that humidity is likely an important determinant of clinging force in geckos.

Both van der Waals and capillary forces fail to explain the shear adhesion data at the whole animal scale. Resolution of this paradox will require examination of the physical and chemical interaction at the interface and particular way in which water interacts with substrate and setae at the nanometer scale.

iv ACKNOWLEDGEMENTS

I owe my deepest gratitude to my advisor, Dr. Ali Dhinojwala, whose

encouragement, patience, support and guidance have enabled me to develop a deep

understanding of the subject of adhesion of gecko feet. He has also been a source of

inspiration and creativity during the course of my study.

I would like to thank my committee members, Dr. Gary R. Hamed, Dr. Li Jia, Dr.

Darrell H. Reneker and Dr. Peter H. Niewiarowski for their valuable comments and suggestions.

I would like to show my gratitude to those who have made contribution to my work. Dr. Lijie Ci, Dr Anubha Goyal, Dr. Sunil Pal from Dr. Pulickel Ajayan’s research group at Rice University and Rensselaer Polytechnic Institute have provided high quality carbon nanotube samples and offered help in developing synthetic gecko adhesives. Dr.

Darrell H. Reneker generously donated his tube furnace when I needed it most. Dr. Peter

H. Niewiarowski and his students Stephanie Lopez, Alyssa Stark have performed a lot of experiments and taken good care of geckos. Dr. Todd Blackledge allowed me to use his equipment. REU interns, Emily Hagen and Rachel Shi, have carried out experiments. I am also indebted to other members of the group for supporting me.

I would also like to thank National Science Foundation for financial support and

The University of Akron for tuition and stipend scholarship .

Lastly, I offer my regards and blessings to all of those who supported me in any respect during the completion of the project.

v Lastly, I offer my regards and blessings to my family for supporting me in any respect during the completion of this dissertation. My wife, Zhiwei Rong and my daughter, Angela Ge have always been good companions. My father, Gangzheng Ge and my stepmother, Xifen Huang have always provided encouragement and support. Last but

not least, I want to thank my late mother Hao Shen whom I have missed most for all the sacrifices she made for where I am today.

vi TABLE OF CONTENTS

Page

LIST OF TABLES ...... ix

LIST OF FIGURES ...... x

CHAPTER

I. INTRODUCTION ...... 1

II. BACKGROUND ...... 9

2.1 Adhesion in Geckos ...... 10

2.2 Modeling of Gecko Adhesion ...... 35

2.3 Design Guidelines for Synthetic Gecko Adhesives ...... 62

2.4 Fabrication and Application of Gecko-Inspired Adhesives ...... 64

III. EXPERIMENTAL ...... 80

3.1 Fabrication of CNT-based Synthetic Gecko Adhesives ...... 80

3.2 Characterizations of Adhesion Properties ...... 83

3.3 Synthetic Gecko Tape ...... 90

3.4 Live Gecko Experiment ...... 95

IV. RESULTS AND DISCUSSION ...... 101

4.1 Adhesion of VA-CNTs...... 101

vii 4.2 Synthetic Gecko Tape ...... 124

4.3 Mechanism of Adhesion in Geckos ...... 135

V. CONCLUSIONS...... 145

5.1 Adhesion of VA-CNTs...... 145

5.2 Mechanism of Adhesion in Geckos ...... 154

BIBLIOGRAPHY ...... 157

viii LIST OF TABLES

Table Page

2.1 Mechanisms of adhesion excluded by experiments...... 21

2.2 Adhesion force of spatulae in dry and wet environment ...... 26

2.3 Pull-off force and scaling efficiency of contact shapes ...... 47

2.4 Adhesion of live Tokay gecko on rough surface ...... 53

2.5 Comparison of polymer-based synthetic gecko adhesives ...... 72

2.6 Comparison of CNT-based synthetic gecko adhesives ...... 76

3.1 Sex, size, and toe pad area of geckos ...... 97

4.1 Average maximal adjusted adhesive force by species and temperature...... 137

ix LIST OF FIGURES

Figure Page

2.1 The hierarchical structures on adhesive toe pads of geckos ...... 11

2.2 TEM images of sections through a single Tokay gecko seta...... 13

2.3 Adhesion force of a single seta ...... 16

2.4 Shear force from sliding setae...... 18

2.5 Adhesion of a single spatula...... 19

2.6 Evidence for van der Waals forces...... 24

2.7 Effect of humidity on adhesion ...... 28

2.8 Effect of temperature on clinging ability ...... 31

2.9 Self-cleaning of gecko feet ...... 34

2.10 Scaling of contact element size and density with body mass ...... 38

2.11 The interaction between two surfaces...... 42

2.12 Peel zone model of spatula ...... 43

2.13 Contact geometries and shapes of contact area ...... 45

x 2.14 Pull-off forces of two different spatulae on rough surface...... 52

2.15 Fractal gecko-hair model ...... 55

2.16 Spring models of hierarchical structures ...... 59

2.17 Hierarchical system of tilted beams ...... 61

2.18 Adhesion design map for different contact shapes ...... 63

2.19 Aarray of polyimide pillars mimicking gecko foot-hairs ...... 66

2.20 SEM images of polypropylene fiber array ...... 67

2.21 Combined lamellae and nanofiber arrays ...... 68

2.22 SEM images of self-cleaning polypropylene fibrillar adhesive ...... 69

2.23 Nanostructures fabricated by sequential molding ...... 70

2.24 Slanted nano-hairs from UV-curable PUA resin ...... 71

2.25 Hierarchical hairs from UV-curable PUA resin ...... 72

2.26 SEM images of CNT-based gecko adhesives ...... 75

2.27 Examples of wall-climbing robots...... 78

3.1 Chemical vapor deposition setup for CNT growth ...... 81

3.2 Characterization of thin-walled CNTs ...... 82

3.3 Images of VA-CNT-based synthetic setae and spatulae ...... 83

xi 3.4 Schematic of friction and adhesion cell ...... 86

3.5 Friction and adhesion cell for SEM imaging...... 88

3.6 Schematic of contact resistance measurement...... 90

3.7 Specimen of synthetic gecko tape for macroscopic force measurement...... 92

3.8 Shear force measurement of synthetic gecko tape ...... 94

3.9 Peel force measurement of synthetic gecko tape...... 95

3.10 Apparatus measuring shear force of live geckos...... 98

4.1 Side view SEM images of VA-CNT arrays ...... 105

4.2 High resolution side view SEM images of VA-CNT array ...... 106

4.3 Normal compression modulus of VA-MWCNT array ...... 108

4.4 Lateral compression modulus of VA-MWCNTarray ...... 109

4.5 High resolution SEM image of periodic buckling of VA-CNT array ...... 111

4.6 Euler and constrained buckling models ...... 113

4.7 High resolution top view SEM images of VA-CNT array...... 115

4.8 Side-wall contact model of CNT ...... 116

4.9 Change in contact area in a normal pull-off test ...... 118

4.10 Change in contact area in a shear pull-off test ...... 120

xii 4.11 Preload dependence of normal pull-off strength ...... 122

4.12 Preload dependence of shear pull-off strength...... 123

4.13 Shear pull-off force for unpatterned synthetic gecko tapes ...... 125

4.14 Shear pull-off force of patterned synthetic gecko tapes vs. area ...... 127

4.15 Shear pull-off force for hierarchically patterned synthetic gecko tapes ...... 128

4.16 Images of CNTs bearing shear load ...... 129

4.17 Carbon nanotube residues on the mica surface ...... 130

4.18 Peel force of synthetic gecko tapes at 45° angles...... 132

4.19 Comparison of viscoelastic tape and synthetic gecko tape...... 133

4.20 Robustness of VA-CNT-based synthetic gecko tape...... 134

4.21 VA-CNT-based synthetic gecko tape tested in vacuum chamber...... 135

4.22 Results of live gecko temperature trials ...... 138

4.23 Results of live gecko humidity trials...... 139

4.24 Combined results of temperature and humidity trials ...... 141

xiii

CHAPTER I

INTRODUCTION

Nature has invented many adhesive systems during millions of years of engineering by evolution. These biological adhesive systems have been optimized for the survival of the species. Many organisms such as bacteria, fungi, marine algae, barnacles, mussels, and terrestrial vertebrates use adhesive polymers [1-23]. Many arthropods such as flies, beetles and spiders, and vertebrates such as lizards possess tarsal adhesive pads densely covered with adhesive hairs [24-32]. Some of these animals even take advantage of both adhesive polymers and hairs. Biological adhesives vary widely in chemistry, structure and capability which cannot be matched by currently available synthetic adhesives. Their remarkable performance under challenging conditions and diversity in chemistry and structure are great examples for developing new synthetic adhesives that offer better performance or satisfy the needs of special applications. For example,

adhesive bonds often fail in water due to poor bonding and corrosive chemistry. A

synthetic adhesive that bonds strongly under water was developed by learning a lesson

from mussels whose adhesive proteins are water resistant and form strong bonds under

water [18-23]. The key compound 3, 4-dihydroxyphenyl-L-alanine (DOPA) has been

successfully extracted and synthesized [19, 33, 34].

Biological adhesives are very complex both in chemistry and morphology, involving a wide range of interactions and components with different functions. However, the mechanisms sometimes are not well understood. To replicate a simple biological

adhesive can be very challenging. The sticky toe pads of geckos are a perfect example of

nature’s complex and fascinating way of manipulating interactions between molecules to

make them stick and unstick to almost any surfaces easily. The mechanism that allows

geckos to climb on a vertical surface or hang from a ceiling with one toe has attracted considerable interest for over two millennia. Recent experimental results favored

intermolecular forces as the origin of the geckos’ ability to defy gravity [35-50]. Each toe

is covered with high aspect ratio tilted hairs called setae which are made of stiff β-keratin.

The ends of these hairs split into nanometer-size contact elements called spatulae. This hierarchical design makes the array of tilted gecko hairs behave like a very soft material,

enabling the nanometer-sized contact elements to make intimate contact with surfaces. At

a close distance of less than a few nanometers, the intermolecular forces between

molecules are attractive, translating into very strong adhesive force at macroscopic length

scales. In fact, Tokay geckos can support forces that are almost forty times of their

bodyweight on vertical surfaces. The debate still remains active whether van der Waals

forces or capillary forces are responsible for adhesion in geckos. Although, van der

Waals forces were confirmed by experiments as the primary mechanism, humidity and

temperature effects were also observed, which cannot be explained by van der Waals

interactions. This suggests that capillary forces also play an important role in adhesion

and capillary forces do not compromise van der Waals forces.

The adhesion in geckos was further investigated at smaller length scales. The

adhesion forces generated from a single seta and spatula were measured by micro-electro-

mechanical sensor and atomic force microscope (AFM), respectively [43, 49, 50].

Interestingly, these tiny hairs generated much higher force than the value estimated by

results from the whole animal. In contrast to friction and sliding between solids, it was

observed that the adhesion force increased during sliding and then reached a maximum

steady value. Furthermore, the maximum force increased with the speed of sliding [51].

These measurements suggest that the geckos’ adhesive was designed with a safety factor

that would save the animal in rare but life-threatening events such as a free fall. Geckos

can recover from a free fall by slapping their foot against a surface when passing by. The

faster the geckos pass the stopping surface, the greater force is generated by the toe pads

to stop the fall.

The above unique properties of the adhesive toe pads of geckos have aroused

extensive interests in theoretical analysis and modeling of geckos’ adhesive hairs [52-67].

The “contact splitting” model has been proposed to explain the adhesion of hairy

structures [52]. By analyzing the Johnson–Kendall–Roberts (JKR) model of a hemisphere

and half space contact, it was found that the total adhesion force is enhanced by a factor

of if the hemisphere is replaced by n smaller self-similar hemispheres. Analyses of other√푛 contact shapes such as flat punch and elastic tape also result in similar enhancement factors of different exponents [67-70]. This clearly suggests that one important role of hairy structures is to enhance adhesion. Indeed, the first generation of gecko-like synthetic adhesives was simply an array of high aspect ratio polyimide pillars [71].

Although the adhesion was only 3 N/cm2, this is in the same order of adhesion force of

geckos. The adhesion of a smooth polyimide film as a control was nearly zero. Several

more complex models with consideration paid to hierarchy have been proposed [60-67].

A fractal model built on self-similar hairs showed that the effective work of adhesion could be greatly increased by adding hierarchical layers. In addition to that, simulation

suggests that the hierarchy is important for adhesion on rough surfaces which is common in natural environment. Hierarchy and tilted stalks of setae could also effectively reduce the modulus of the array by several orders of magnitude, thus making the array sufficiently compliant even it is made from stiff β-keratin [44, 72, 73]. The experimentally measured effective modulus of an array of setae is only ~ 83 kPa, which meets the Dahlquist’s criterion for tackiness [44]. Experiments confirmed that gecko feet stick well to a wide range of rough surfaces [74-78]. Moreover, the geckos’ adhesive is the first known self-cleaning adhesive. It was demonstrated that arrays of setae isolated from Tokay geckos recovered their ability to cling to vertical surfaces by taking only a few steps on clean glass after being dusted with silica particles [79]. The functional properties of geckos’ adhesive are summarized as the following:

• Directional.

• Attaches strongly with minimal preload.

• Detaches quickly and easily.

• Increase in shear force upon sliding.

• Sticks to nearly every material.

• Self-cleaning.

• Anti-self-adhering.

• Non-sticky default state.

• Wear-resistant.

Gecko-like synthetic adhesives (GSAs) have achieved or exceeded the adhesion strength of live geckos but only a few of them are self cleaning and wear resistant.

Contact mechanics models and simulations provided guidelines for the fabrication of GSAs but fabrication of GSAs is a very challenging task, primarily due to the difficulty in making arrays of complex features in small length scales. The requirements for high-performance dry adhesives such as proper size, aspect ratio, tilting angle, shape, hierarchy and materials properties are not easily optimized during fabrication. The first generation of gecko inspired adhesive was an array of micrometer-sized polymer pillars

[71, 80] and its adhesion performance was poor. Recent advances of micro-fabrication have enabled production of complex structures that incorporate more and more design parameters. Smaller size, higher density, tilted fiber, shape of contacting element and hierarchical structure have all been reported to improve adhesion or to enable features such as easy release, self-cleaning.

Polymeric hairs [46, 71, 72, 80-151] and carbon nanotubes (CNT) [152-158] have been most frequently used as materials for making synthetic gecko hairs. Generally, polymer-based adhesives have been fabricated by a top-down approach such as molding with templates, direct lithography. Advanced top-down fabrication techniques have been implemented to fabricate templates with slanted nanometer-size holes and controlled tip shape. Polymer-based methods offer versatile approaches to fabricating gecko-mimicking nano-hairs with tailored geometry (radius, height, shape of tip, angle and hierarchy) and a wide range of material properties (modulus, surface energy). However, the adhesion strength is usually lower than that of geckos and CNT-based adhesives because of the relatively large feature size and poor mechanical strength of polymer materials at very small length scales. Conversely, CNT-based adhesives have been fabricated by a bottom- up approach. Vertically aligned CNT arrays are grown from a catalyst layer deposited on

a substrate by chemical vapor deposition (CVD). CNT-based dry adhesives usually have much higher adhesion strength than geckos and polymer-based adhesives due to the extremely small radius ( 10 nm), high strength and high modulus ( 1 TPa) of CNTs.

However, CNT-based dry∼ adhesives usually require a much larger preload∼ than geckos

and polymer-based gecko adhesives to activate adhesion.

This dissertation will be focusing on mimicking gecko foot-hairs using vertically aligned carbon nanotubes and humidity and temperature dependence of adhesion in geckos which challenge the widely accepted molecular mechanisms.

To mimic gecko foot-hairs, vertically aligned multi-walled carbon nanotube arrays (VA-MWCNT) were chosen. The adhesion between an AFM tip and a VA-CNT array is up to 160 times greater than the value measured from live geckos [152]. The

“contact splitting principle” predicates that the theoretical adhesion strength is at least

500 N/cm2 [153]. In addition, extremely high and controllable aspect ratio, outstanding mechanical strength and tunable flexibility of carbon nanotubes make it an excellent

material for mimicking gecko foot-hairs. However, VA-CNT patches often show low or

even no adhesive strength at macroscopic scale. The mechanism for why CNTs behave so

differently at macroscopic scale was not clear.

To understand the mechanism of adhesion, the behavior of VA-CNT hairs near

the contact interface was explored using a combination of mechanical, electrical contact

resistance measurements, and scanning electron microscopic (SEM) imaging. It was

found that instead of making expected end contacts, carbon nanotubes make significant

side-wall contacts which increase with preload. In side-wall contact geometry, the

adhesion of CNTs is determined by the balance of adhesion in the contact region and the

bending energy of CNTs. As a consequence, a compliant VA-CNT array is required for making adhesive patches and adhesion strength is truly pressure sensitive, because it increases with preload.

Macroscopic patches of compliant VA-CNT array have been fabricated. Patches of uniform array showed adhesive strength similar to geckos (10 N/cm2) on a variety of substrates and was removed easily by peeling. When the array was patterned to mimic the hierarchical structures of gecko foot-hairs, the adhesion strength showed an enhancement of up to four times. VA-CNT-based gecko adhesives also demonstrated self-cleaning, non-viscoelasticity and good strength in a vacuum environment. These are desired for applications in robotics, microelectronics, thermal management and outer space operations. The details of fabrication methods, performance trials, and macroscopic and microscopic measurements will be described in Chapter III and results and discussion will be presented in Chapter V.

Most if not all features of geckos’ adhesive toe pads have been achieved by synthetic gecko adhesives. The synthetics have been used in applications such as wall- climbing robots, transporting objects, supporting weight and conductive adhesives.

Although none of the gecko adhesives have been commercialized, many prototype robots and adhesives have been demonstrated.

As previously mentioned, studies of adhesion in geckos at the toe pad, setal and spatulae level suggest that van der Waals forces are the primary mechanism responsible for adhesion [43, 45, 46, 49]. But temperature and humidity were found to affect adhesion in a very complex way [50, 159-161]. If clinging in geckos is only based on van der Waals forces, clinging capacity should be temperature insensitive over the range of

body temperatures typically experienced by geckos [162]. Adhesion as a key functional

capacity of locomotion in geckos may be free of typical thermal biophysical constraints

experienced by ectotherms in thermally heterogeneous environments. This suggests that

thermal independence of adhesion may have driven the evolution of the geckos’ adhesive

system [162]. Temperature sensitivity of clinging capacity has only been measured separately in two species: Phelsuma dubia [163] and Gekko gecko [162] at whole organism scale. However, contrasting results were obtained between these two separate studies probably due to differences in methodology and species used. A more systematic examination of the clinging ability of Gekko gecko and Phelsuma dubia using a single methodology over the range of 12 ~32 °C at 5 °C intervals will be described in Chapter

III and results will be discussed in Chapter V. Both van der Waals and capillary forces fail to explain the results at the whole animal scale. The adhesion force measurements at single setal scale with variable humidity showed that the modulus of setae decreases with increase in humidity which could explain the humidity dependence of setal adhesion

[161]. Yet those trials were conducted in only one temperature. Adhesion force measurements at the single setal scale in variable temperatures are needed to eliminate the possibility of thermal biophysical constraints (vascular, muscular and skeletal) influencing adhesion. However, no study has been done to evaluated the effect of

temperature and humidity on the adhesion of the seta and spatula.

CHAPTER II

BACKGROUND

Geckos are lizards belonging to the family Gekkonidae. They are found in warm

climates throughout the world. There are an estimated 2000 different species of geckos

worldwide. Many of the species have the ability to climb on vertical and even inverted

surfaces using their adhesive toe pads. Interestingly, geckos do not fall even when their feet slide [164, 165]. Geckos’ toe pads are sticky due to the hierarchical hair-like

microstructures which function like an adhesive but are different from conventional

adhesives such as viscoelastic pressure sensitive adhesive found on Scotch® tape. The

adhesive toe pads of geckos are just one example of adhesives invented by nature. The

geckos’ adhesive system is remarkably complex, involving a wide range of interactions

and components with different functions which are very challenging to replicate. This

chapter will be dedicated to introducing the structure and morphology of geckos’ feet,

adhesion in geckos from the whole organism to a single spatula, molecular mechanisms,

the unique properties of geckos’ adhesive toe pads such as self-cleaning, contact

mechanics analysis, theories and simulation with mechanical models, design principles in

mimicking gecko foot-hairs and finally an overview of the state-of-the-art of polymer-

based and CNT-based synthetic gecko adhesives and their applications in wall-climbing

robots, electronics, space missions and other industries.

2.1 Adhesion in Geckos

The adhesive toe pads of geckos are very complicated structures. They are

composed of multi-level hierarchical structures consisting of scansors, setae, and spatulae

ranging from millimeter to nanometer length scales. Measurements of adhesion have been accessible to all these length scales due to the advancement in experimental tools.

These measurements are crucial to the understanding of the mechanisms for adhesion in geckos. Experimental data obtained from whole organism, setae and spatulae suggested intermolecular forces are the molecular mechanisms of adhesion in geckos. However which intermolecular interactions are dominant has been the topic of active debate. These experiments involve many variables including, geometry, three dimensional orientation, preload, and properties of the surface and foot-hairs, temperature and humidity. Results are sometimes controversial, leading to different conclusions. The measurements of adhesion in geckos and debate about the molecular mechanisms will be reviewed in this section.

2.1.1 Adhesive Hierarchical Structure on Gecko Foot

The hair-like structure of the gecko toe pad was discovered more than a century ago [166-168]. Setae were revealed by optical microscope in the early 20th century [169].

Due to limitation in resolution, the spatulae were finally discovered after the electron microscope was invented [25, 170]. Figure 2.1 shows the hierarchy of these structures starting with a whole animal image of a Tokay Gecko (Figure 2.1A) which is the most studied species of gecko for adhesion due to their large size and availability in pt trade.

Figure 2.1 The hierarchical structures on the adhesive toe pads of geckos shown in the order from macroscopic to microscopic length scales. (A) Tokay Gecko (B) a gecko foot showing scansors. Scanning electron microscope (SEM) images of (C) an array of setae, (D) a single setae, (E and F) an array of spatulae at the distal end of a setae, and (G) a single spatulae showing its triangular end. [39, 174]

Scansors shown in Figure 2.1B are soft ridges about 1~2 mm wide which are covered by

arrays of setae. Sometimes scansors are referred as lamellae in the literature. Setae are derived from the interaction between the oberhautchen and the clear layer of the epidermis. The two layers form the shedding complex and permit natural skin shedding in lizards. Setae consist of a resistant but flexible corneous material largely made of keratin

associated α-proteins and β-proteins [171]. Histochemical and ultrastructural analysis indicated that lipid material is closely associated spatially with maturing setae as they progress to maturation during shedding [171]. The SEM images (Figure 2.1C-E) of setal array show that setae are grouped in the Tokay gecko (see group of four setae shown in

Figure 2.1C). Setae are approximately 5 µm in diameter. The length of Tokay’s setae varies from approximately 50 to 100 µm with the shorter setae positioned in the front of each group. The density of setae is approximately 14,000/mm2 [172]. At the distal end,

each seta branches into bundles of fibrils which are about 100 nm in diameter at the point

of splitting and gradually narrow to about 85 nm immediately prior to flaring out into the

triangular shaped terminal elements [25, 27, 172-174]. These terminals are called

‘spatulae’ because of their resemblance to kitchen spatula. Spatula also refers the terminal and fibril stalk as a whole. The spatulae are 150 to 275 nm across at the far end.

The stalks of the setae are tilted at an angle enabling the spatulae to have intimate contact

with the surface during application.

The internal structure of setae was observed by transmission electron microscopy

(TEM) in the cross-section of a single seta [174]. It is shown in Figure 2.2 that the

fibrillar structure is preserved over the entire length of the setae. The spatulae bearing

fibrils at the distal end of setae are likely to traverse the entire length of the seta. The

proteinaceous fibrils are held together by a matrix-like proteinaceous sheath. X-ray

diffraction patterns show conclusively that the only ordered protein constituent in these

structures exhibits characteristics of β-keratin [174]. Raman microscopy of individual setae, however, clearly indicates the presence of additional protein constituents, some of which may be identified as α-keratins [174]. The presence of both β-keratins and α-

keratins are also supported by electrophoretic analysis and immunological analyses of

proteins in setae [174-176].

Figure 2.2 Transmission electron microscopy (TEM) images of sections through single Tokay gecko seta. (A) before and (B) after terminal branching are shown together with (C) a tangential section through a setal stalk. [174]

2.1.2 Adhesion in Geckos

Common tests methods for measuring the performance of gecko adhesive include

normal pull-off test, shear pull-off test and peel test. In a normal pull-off test, the separating force is perpendicular to the bonded surfaces. In a shear pull-off test, the separating force is parallel to the bonded surfaces. In a peel test, the separating force is applied at an angle. These geometries are clearly defined in macroscopic length scale when measuring an adhesive tape. But for geckos, macroscopic geometry dose not translate to the same geometry at microscopic scale because geckos’ setae and spatulae are compliant and tilted at a certain angle. When geckos attach and detach their foot-pads by uncurling and curling of the toe, the geometry at the length scales of setae and

spatulae is peeling. Normal force is sometimes referred to perpendicular force and shear

force is also referred to parallel force, lateral force, frictional adhesion and friction in the

literature. Here the shear force must be distinguished from friction force because there is

no normal force applied to the adhesive during measurement. Instead, a preload force or

preload is applied before shearing. It is the initial perpendicular force that pushes the

adhesive toward the surface, but preload is not maintained during the pull while there is

always a normal force in friction measurement. It should be noted that preload was found

to be critical in the measurement of single seta and single spatula forces.

2.1.2.1 Adhesion of the Whole Organism

The adhesion of geckos has been measured by many researchers at the whole

organism level on various substrates and under different temperatures and levels of humidity [163, 177-180]. Attempts to understand the effects of temperature and humidity

at the whole organism level will be reviewed in later sections.

Irschick et al. evaluated the clinging ability of 14 seta-bearing lizards, including

the Tokay gecko [177]. Here, clinging ability means the force required to pull a lizard

down on a nearly vertical surface. The measurements were carried out at 22 °C on an acetate sheet supported by a glass plate. The substrate was smooth to avoid the influence of the geckos’ claws. It was found that the two front feet of a Tokay gecko were able to support 20 N of force parallel to the substrate. With an average toe pad area of 227 mm2,

this number is equivalent to approximately 9 N/cm2 adhesion force in the lateral direction.

If averaged by the number of setae, it is equivalent to approximately 6.2 µN shear load

per seta, assuming all setae were engaged. The actual force on load bearing setae could be much greater, however, there is no evidence that all setae are engaged at the same time.

In these experiments, the preload, normal and peel forces were not measured.

Preload at the whole organism level seems to be very small. In this study, geckos adhered

easily to the testing surface with a natural grip. It is difficult to measure preload, normal

and peel force in the whole animal adhesion study because of the unique toe uncurling

attachment and curling detachment locomotion behavior of the geckos.

2.1.2.2 Adhesion of a Single Seta

The adhesion of a single seta was measured by Autumn et al. using a micro- electro-mechanical systems force sensor as a force gauge [43]. The contact geometry was

well controlled and preload, normal and shear forces were measured. The results were

compared to contact mechanics models.

Autumn et al. found that the adhesion of a single seta depends on the preload and three-dimensional orientation between the seta and surface. The shear force of a single seta was measured by pressing it against a surface and then pulling parallel to the surface as illustrated in Figure 2.3A. The setae generated significant shear force when the spatulae were projecting toward the surface as shown Figure 2.3B. This orientation allowed the spatulae to make intimate contact with the surface. The force parallel to the surface increased linearly with the preload and was substantially greater than the force produced by the inactive, non-spatula region at all preloads. When the spatulae were

projecting away from the surface, the setae force also increased with preload, but did not

exceed 0.3 µN. With a preload of 15 µN, the maximum adhesive force of a single seta

averaged 194 ± 25 µN, nearly thirty-fold greater than the value estimated from whole organism measurements. It was hypothesized that the increase in shear force was due to increasing number of spatulae making contact with the surface during small distance sliding of about 5 µm before reaching the maximum force. The orientation of the setae

was also important in detachment. The parallel force was over ten times greater than the force pulling setae away (perpendicular) from the surface. It is shown in Figure 2.3C that

Figure 2.3 Adhesion force of single seta. (A) Shear force of a single seta was measured by pressing it against the surface, then pulling parallel to the surface. Maximum shear force was 194 µN with a preload of 15 µN. (B) Orientation and preload dependence of shear force. The solid line represents setae with spatulae projecting toward the surface. The force increases with preload. The dashed line represents setae with spatulae projecting away the surface. The force produced is much lower. (C) Setae detached at a similar critical angle (30.6 ± 1.8°). [43]

setae detached at a similar angle (30.6 ± 1.8°) when pulled away from the surface. These results are consistent with geckos’ complex motion of toe uncurling during attachment and toe curling during detachment, analogous to removing a piece of tape from a surface.

The results indicate that achieving maximal adhesion requires a small push perpendicular (preload) to the surface, followed by a small parallel drag, which explain the load and direction dependence of adhesion observed at the whole-animal scale [181,

182]. Furthermore, Autumn later referred to this property as the non-sticky default state of setae [42] because they are not sticky unless attached with proper orientation. Unlike conventional adhesives, gecko setae do not have a tacky feel. Fingers can be easily removed after touching geckos’ toe pads. Resistant force can only be sensed by sliding fingers from the base of the toe to the tip. Importantly, gecko setae also do not self-adhere.

In the resting (not bearing any load) state, setal stalks are recurved proximally. When the toes of the gecko are engaged, the setae may become bent under the small preload, flattening the stalks between the toe and the substrate such that their tips point distally.

This small preload and a small dragging displacement of the toe proximally may bring the spatulae uniformly in contact with the substrate thus maximizing their contact area.

It is interesting that the parallel force reached a maximum steady state after the onset of sliding as shown in Figure 2.3A. Gravish et al. revisited the sliding of setae [51].

It was discovered that shear force did not exhibit a decrease at transition from static to kinetic sliding as is expected in ordinary solid material. As shown in Figure 2.4A, both shear force and normal force increased at the onset of sliding and continued to increase with shear speed ranging from 500 nm/s to 158 mm/s. The maximum force at steady state increased with the rate of sliding as shown in Figure 2.4B. This behavior is similar to

sliding behavior of polymers. Macroscopic stick-slip was not observed, but uncorrelated

stick-slip of spatulae was not excluded. This experiment also showed that gecko setae do

not show wear and tear after 30,000 cycles. Wear resistance, and steady shear force

during sliding, may emerge from the stochastic stick-slip of a population of individual fibrils at high resonant frequencies.

Figure 2.4 Shear force from sliding setae. (A) As sliding begins at 300 ms, friction (or shear, solid line) and adhesion forces (or normal dashed line) remain constant and do not exhibit any notable decrease. Negative normal force indicates while positive force indicates compression. (B) Steady-state contact forces increased in magnitude as a function of drag speed. [51]

2.1.3 Adhesion of a Single Spatula

The adhesion force of individual spatulae was measured by atomic force

microscopy [49, 50, 159]. The adhesion force was obtained from the force-distance curve.

The shear and peel force were not measured, so the comparison with a single seta should

be limited to normal pull-off force.

Huber et al. glued a single seta from Tokay geckos to an AFM tip [49]. The seta was then processed in a focused ion beam microscope to cut off most of the spatulae. The seta with four spatulae left was brought in contact with glass substrate under a compressive preload of 90 nN followed by 7 µm lateral sliding at 25 °C and 35% RH as shown in Figure 2.5A. The perpendicular force required to pull the seta off the surface was then measured. There are two distinct peaks and one weak peak in the histogram of all force measurements in Figure 2.5B. The first peak with a mean value of 10.8 ± 1.0 nN was assigned to a single spatula detachment. The second peak at 20.4 ± 1.9 nN was assigned to two spatulae detachment. A weak third peak at 30 nN was considered as the results of three spatulae detaching at the same time. Four-spatula detachment was not observed in the experiments. The reported force values were measured in perpendicular direction. The lateral force values are expected to be at least ten times higher than perpendicular values based on the measurement of single seta [43].

Figure 2.5 Adhesion of a single spatula. (A) The spatula was brought into contact with 90 nN preload followed by 7μm sliding while maintaining the preload. The perpendicular pull-off force was recorded. (B) Histogram of all measurements showing two strong peaks at 10 and 20 nN and a weak peak at 30 nN. The peaks are attributed to the detachment of one, two and three spatula(e), respectively. [49]

The same group reported measurements with changing humidity [159] after Sun

et al. reported dependence of adhesion of a single spatula on humidity [50]. The effect of

humidity will be reviewed in its own section. Experiments with a single spatula were also

done on rough surfaces [164] which will be discussed in the section outlining the effect

of roughness.

2.1.4 Molecular Mechanism of Adhesion in Geckos

The adhesive microstructures of many gecko species and adhesion force of the

whole organism, setae, and spatulae are well documented, but an in-depth understanding

of the molecular mechanisms has been elusive. At least seven possible mechanisms for

gecko adhesion have been discussed over the past 175 years, including secretion of glue,

suction cup, electrostatic attraction, microinterlocking, friction and two intermolecular interactions: van der Waals and capillary forces. [43, 45, 50, 159, 160, 167, 179-181, 183,

184] The first five mechanisms were easily eliminated. The experimental observations that ruled out these mechanisms are listed below in Table 2.1 . The intermolecular forces remain as the most possible mechanism, but controversial experimental results have aroused debate between van der Waals interaction and capillary force.

2.1.4.1 van der Waals Forces

van der Waals force was named after Johannes Diderick van der Waals who won the Noble Prize in physics in 1910. van der Waals forces are defined as the attractive or repulsive forces between molecules other than those due to bond formation or to the

electrostatic interaction of ions or of ionic groups with one another or with neutral

molecules. The term includes: dipole–dipole, dipole-induced dipole and instantaneous induced dipole-induced dipole forces [185]. It is widely used for the totality of

nonspecific attractive intermolecular forces in of gecko adhesion literature.

Table 2.1 Mechanisms of adhesion excluded by experiments

Mechanism Observations

Secretion of glue No glue secreting glands on geckos’ toes [167, 183, 184].

Suction cup Ability to adhere in vacuum [181].

Electrostatic attraction Ability to adhere to metal in ionized air [181].

Ability to adhere to molecularly smooth semiconductor Microinterlocking: surface [43].

Ability to adhere to molecularly smooth inverted surfaces Friction [43]

van der Waals interactions are weak in comparison with covalent, hydrogen, ionic,

and metallic bonds. But, unlike other bonds, van der Waals forces are always present and

attractive over the distance between molecules. The van der Waals force per unit area

between two parallel surfaces, fvdW, is given by

= for < 30 nm 6 퐴 푓푣푑푊 3 퐷 where A is the Hamaker constant and D휋퐷 is the distance between the two surfaces [186].

The Hamaker constant depends on the material properties of both the interacting bodies and the intervening media. Accurate calculation of the Hamaker constant involves the

dielectric constant and optical properties of the interacting materials over the entire

frequency range in the electromagnetic spectrum [186].

Assuming van der Waals forces are dominant in gecko adhesion, the magnitude of

these forces is expected to be dependent on materials properties and separation. Surfaces

with higher dielectric constant have a higher Hamaker constant, resulting in stronger van

der Waals force. The Hamaker constant for materials interacting in dry air is typically ~

1019 J. Altering the dielectric constant of one or both surfaces can alter the Hamaker

constant, which can be as low as one-third the typical value for some polymer-polymer

interactions (e.g. polytetrafluoroethylene, PTFE) and as high as five times this value for

some metal-on-metal interactions. The intervening media usually is dry air with a

dielectric constant close to 1. If the intervening media is a material with a very high

dielectric constant, the Hamaker constant becomes very small, leading to a weak van der

Waals force. In water, the Hamaker constant can be reduced by an order of magnitude. It

is also interesting to consider the possibility that there are molecular layer(s) of water

adsorbed on the surface where van der Waals forces act between the setae and water

layer. In such case, the calculation of The Hamaker constant requires more complicated

model. Nevertheless, the variation in The Hamaker constant is only within an order of

magnitude while distance typically varies in the range of 0.2 ~ 0.4 nm which can cause

variation in van der Waals force by six or more orders of magnitude.

Hiller showed experimentally that adhesion force in geckos is correlated with

surface energy. As a result, intermolecular forces were concluded to be the primary

mechanism responsible for adhesion in geckos [180]. The adhesion force decreased with

increasing water contact angle of the surfaces and the observation that geckos cannot

adhere well to non-polar polytetrafluoroethylene (Teflon®) suggested that the polarity of

the surface is an important factor in the strength of adhesion. Water contact angle alone

cannot determine the contributions of van der Waals and polar interactions because the intermolecular attraction between liquid droplet and a surface is the sum of dispersive

(van der Waals) and polar components.

The adhesion force measurements of a gecko seta on both hydrophilic and hydrophobic surfaces favor van der Waals forces as the dominant mechanism [43, 45, 46].

Autumn et al. separated the effect of polarizability (high dielectric constant, ε) from

polarity (hydrophobicity and hydrophilicity, measured by water contact angle θ) by

measuring adhesion on two polarizable semiconductor surfaces that varied greatly in hydrophobicity [46]. On a gallium arsenide (GaAs) semiconductor surface which is highly hydrophobic (θ = 110°) and highly polarizable (ε = 10.88 ), the parallel force of a

single gecko toe was measured. Parallel force on the strongly hydrophilic (θ = 0°) and

polarizable ( ε = 4.5 ) silicon dioxide (SiO2) semiconductor surface was measured as a control. The perpendicular force of a single gecko seta on hydrophilic SiO2 and

hydrophobic Si (θ = 81.9°) microelectromechanical systems (MEMS) force sensors were

also compared [43, 46]. If capillary adhesive forces dominate, it is expected that the

adhesion forces on the strongly hydrophobic GaAs and Si surfaces will be weak. In

contrast, if van der Waals forces dominate, strong adhesive forces are expected on the

hydrophobic and polarizable GaAs and Si surfaces. In either case strong adhesion to the

hydrophilic SiO2 semiconductor and Si control surfaces are expected. The results shown in Figure 2.6 suggest that van der Waals forces are the primary mechanism. Hiller’s data was re-analyzed by Autumn et al. [45]. A strong correlation between force and adhesion

energy for θ > 60° supports the hypothesis based on van der Waals forces. Although these experiments proved van der Waals forces were dominant, humidity was found to be an important factor for adhesion in geckos [50, 159, 160, 179] and cannot be neglected.

Figure 2.6 Evidence for van der Waals forces as dominant molecular mechanism. (A) Qualitative prediction of adhesion force based on capillary adhesion and van der Waals forces. (B) Adhesion forces of gecko toe on highly polarizable semiconductor wafer surfaces differing in hydrophobicity are similar. Adhesion forces of single seta attaching to highly polarizable MEMS cantilevers differing in hydrophobicity are similar. Both results show there is no significant difference between hydrophobic and hydrophilic surfaces. [46]

Given our current knowledge of van der Waals force in geckos, it can be

hypothesized that the adhesion in geckos is a passive phenomena which is insensitive to

the physical performance of the animal. To test this hypothesis, adhesion trials were

performed under variation of temperatures [162, 163, 179]. The results are controversial

and will be discussed in the effect of temperature section separately.

Assuming van der Waals forces are the dominant mechanism for adhesion in

geckos, the adhesion force of a gecko can be estimated. Typical values of the Hamaker

constant range from 4 × 10−20 to 4 ×10−19 J [186]. The contact area of a spatula is taken to

be 2 × 10−14 m2 [25, 28, 45] and the separation between the spatula and the contact surface is estimated to be about 0.2 ~ 0.6 nm. This equation yields the range of force of a

single spatula to be about 0.2 ~ 53 μN. Using a typical value of The Hamaker constant 1

×10−19 J and a gap distance of 0.3 nm, the result is 4 μN. Even the lowest estimated value

of 0.2 μN is still much higher than the result from single spatula force measurement of ~

10 nN [49, 50, 159]. This suggests that the van der Waals forces and the simple contact

model above are not sufficient to explain the adhesion of a single spatula.

2.1.4.2 Effect of Humidity

Despite many experimental results that support van der Waals forces as the

primary mechanism for adhesion in geckos, humidity has a profound impact on adhesion

and cannot be excluded [43, 46]. At the contact interface, humidity can affect adhesion in

two possible ways: capillary force of a liquid bridge due to condensation of water vapor

in the air [160] or a monolayer of water adsorbed on the surface [159]. A recent study

showed that the changes in the materials properties of setae instead of capillary forces

could explain humidity-enhanced adhesion and van der Waals forces remain the only

empirically supported mechanism of adhesion in geckos [161].

Sun et al. measured adhesion of a single spatula from spiny-tailed house gecko

(Hemidactylus frenatus) with AFM method [50]. They found the measured force varied with humidity as listed in Table 2.2. The adhesion is significantly higher at higher humidity or with hydrophilic surfaces but is greatly reduced in water. The results strongly suggest capillary adhesion under these conditions.

Table 2.2 Adhesion force of spatulae in dry and wet environment [50]

Liquids that wet or have a small contact angle on a surface will spontaneously

condense from vapor at micro-contact site as bulk liquid if the contact site is below

certain critical size [186]. This critical size is determined by Kelvin radius =

푘 1/(1/ + 1/ ) where r1 and r2 are the radii of curvature of the liquid bridge푟 at

1 2 equilibrium.푟 The푟 Kelvin radius is connected with the relative vapor pressure (p/ps) by the

well-known Kelvin equation:

= , ln( / ) 훾 ∙ 푉 푟퐾 푅 ∙ 푇 ∙ 푝 푝푠 26

where γ is the surface tension, R is gas constant, T is absolute temperature, V is molar

volume, p is partial pressure of water vapor, ps is saturation vapor pressure, and p/ps is the relative vapor pressure (relative humidity for water). The surface tension γ of water is

0.074 N/m at 20 °C, leading to a critical van der Waals distance of water of γV/RT = 0.54

nm. Consequently, for 90% relative humidity, Kelvin radius is 10 nm. For 10% humidity,

Kelvin radius is 0.5 nm. At small vapor pressures, the Kelvin radius gets comparable to the dimensions of the molecules and thus the Kelvin equation breaks down. Capillary

force of liquid bridge consists of Laplace force, the surface tension and the solid-to-solid

interaction [160]. The capillary force depends on both the relative humidity and the

hydrophobicity (contact angle) of both spatulae and mating surface. Assuming spatulae

have a water contact angle of 128°, Kim et al. calculated the total adhesion force of a

spatula on hydrophilic and hydrophobic surfaces as the sum of Laplace, surface tension,

and DMT adhesion forces. The effects of humidity for hydrophilic and hydrophobic

surfaces as shown in Figure 2.7A are quite different over the range of 0 ~ 100% humidity.

Regardless of water contact angle, at humidity greater than 90%, the total adhesion forces

reached maximum values, indicating enhancement of adhesion in this region. However,

the formation of liquid bridge between the spatulae and mating surface due to

condensation of water vapor in air occurs only above 90% relative humidity [159]. Below

90% relative humidity Kelvin equation seems to break down.

A monolayer of water, which is always present on surfaces under normal

atmospheric conditions, can significantly influence the attraction between two surfaces.

The thickness of the layer increases with humidity as shown in the inset of Figure 2.7C

[159]. Huber et al. prepared hydrophilic surfaces with 2 nm (N-phil) and 192 nm (T-phil)

thick silicon oxide layers which have similar short-range forces but different long-range forces. These surfaces were coated with a monolayer of hydrophobic molecule

(octadecyltrichorosilane) which does not change long-range forces (N-phob and T-phob).

A glass substrate with an intermediate water contact angle of 58° was also prepared. The perpendicular pull-off force was measured at 25 °C and 52% humidity. The results in

Figure 2.7B show that the adhesion forces decreased significantly with water contact angle of the surfaces. Pull-off force measured when submerged in water is significantly reduced. The adhesion is sensitive to the presence of hydrophobic monolayer (short-range

Figure 2.7 Effect of humidity on adhesion. (A) Trend for adhesion forces predicted by capillarity on surfaces varying from hydrophilic to hydrophobic. (B) Adhesion forces of single spatula on hydrophilic and hydrophobic surfaces. (C) Adhesion forces of single spatula on hydrophilic and hydrophobic surfaces with changing humidity. [159, 160]

force) but not to the thickness of Si oxide layer which indicates the adhesion is determined by the chemical composition of a few layers at the surface. The results agreed with similar measurements by Sun et al.[50].

Huber et al. also measured the adhesion of a single spatula on glass and hydrophobic surfaces with controlled variation in humidity. The results in Figure 2.7C show that the adhesion forces increased in a liner monotonic manner in both cases. The rate of increase on hydrophobic surface is much lower than that on hydrophilic surface.

The differences at very low humidity (< 10%) are small and data for high humidity (>

70%) were not obtained. The results obtained on glass was found to fit the model based on modified Hamaker constant due to monolayer of water adsorbed on the surface. The lower increase on hydrophobic surface was qualitatively explained by smaller adsorption energy. The results on hydrophilic glass seem to agree with capillary model while the increase in adhesion on hydrophobic surface was not predicted.

The effect of humidity on the mechanical properties of setal keratin is also taken into consideration [161]. Puthoff et al. reported that an increase in humidity at room temperature softens setae and increases viscoelastic damping, which increases adhesion.

In addition, the trends with hydrophobic and hydrophilic surfaces are similar.

Consequently, changes in materials properties, not capillary forces, fully explain humidity-enhanced adhesion, and van der Waals forces remain the only empirically supported mechanism of adhesion in geckos. While Peattie et al. found that humidity does not have an effect on the elastic modulus of setae in two gecko species in the relevant range of 16-64% RH [73]. In addition, no effects of temperature and age of seta were established. However, dramatic changes in modulus and adhesion were observed

only at very high humidity (> 70%) by Puthoff et al. which was not measured by Peattie

et al.

2.1.4.3 Effect of Temperature

Studies of gecko adhesion at the toe pad, setal and spatulae level suggest that

intermolecular interactions are responsible for adhesion [43, 45, 46, 49, 50, 159]. If

clinging in geckos is only based on van der Waals forces, clinging capacity should be

temperature insensitive over the range of body temperatures typically experienced by

geckos [162]. Adhesion as a key functional capacity of locomotion in geckos may be free of typical thermal biophysical constraints experienced by ectotherms in thermally heterogeneous environments. It suggests that thermal independence of adhesion may have driven the evolution of the gecko adhesive system [162]. Temperature sensitivity of clinging capacity has only been measured in two species: Phelsuma dubia [163] and

Gekko gecko [162] at whole organism level. However, contrasting results were obtained between these two separate studies as summarized in Figure 2.8. No study has been done to evaluate the effect of temperature on the adhesion of seta and spatula.

The study conducted by Losos measured the clinging ability of Tokay Gecko on plexiglass plate at temperatures 12, 16, 17, 22, 24, 31, 34, 35, 41 °C in a walk-in environmental chamber. Geckos that adhered to plexiglass were tipped gradually at a steady rate and the angles at which they fell off were recorded. It was found that clinging performance was non-linearly dependent on temperature with a peak at 17 °C, as shown in Figure 2.8A. It was assumed that the temperature-independent intermolecular bonds were responsible for geckos’ clinging ability. Temperature-dependence of clinging ability

was attributed to the thermal dependence of geckos’ muscular and vascular system and humidity (not regulated in the trials) effect on the intermolecular bonds.

Figure 2.8 Effect of temperature on clinging ability of geckos. (A) Tokay Gecko and (B) Phelsuma dubia. [162, 163]

Bergmann et al. measured the clinging ability of Phelsuma dubia (a day gecko) in the temperature range of 15 ~ 35 °C at 5 °C interval. In this study, the clinging ability was investigated by pulling the animal’s two front feet backward on a smooth acetate sheet oriented horizontally at a constant rate of 5 cm/sec. The animals’ body temperature was regulated at the desired temperature for one hour and performance trials were conducted at room temperature without control of humidity. The results showed large variation at all five temperatures and there is no specific trend in the clinging ability in

Phelsuma dubia with changing temperature as shown in Figure 2.8B. The authors further suggested that the clinging ability is of passive nature and it does not depend upon the temperature and performance of geckos’ muscular and vascular system.

However, results from these studies are conflicting and difficult to interpret because different methodologies were employed, leaving the question of temperature sensitivity of clinging capacity of geckos unanswered. A more systematic examination of the clinging ability of Gekko gecko and P. dubia using a single methodology over the range of 12 ~ 32 °C at 5 °C intervals will be described in Chapter III and results will

presented and discussed in Chapter IV.

2.1.5 Self-cleaning Properties of Gecko Feet

Conventional adhesive tapes have a layer of soft polymer adhesive material which

is “sticky”'. These tapes lose adhesion with repeated use due to dust contamination.

Unlike conventional adhesives, adhesion of gecko setae seems to be immune to natural

contaminants. Dirt particles such as sand, dust, leaf litter, pollen, and plant waxes which

exist in natural environment are likely to contaminate gecko setae. It is reasonable to

assume that these particles are trapped on the adhesive spatulae without external

disturbance to remove them. Particulate contamination reduces the function of adhesive

pads of some insects [187] and grooming is required in order to restore adhesive function

[27]. However, geckos retain adhesive ability of their setae during the month between

shedding cycles without grooming their feet like beetles or secreting sticky fluids to

remove dust like ants [188] and tree frogs [189]. Adhesives that clean themselves during use without water or chemical are called dry self-cleaning adhesives. Gecko setae are the first known dry self-cleaning adhesive. Other well-known self-cleaning surfaces use water droplets to remove dirt particles. This phenomenon is termed as the “lotus effect”.

Lotus leaf removes dirt particles from its non-adhesive and waxy surface with the help of

water droplets. Self-cleaning by water droplets has been observed in many plant [190]

and animal [191] surfaces.

It was demonstrated by Hansen et al. that the arrays of setae isolated from the

gecko Tokay geckos recovered their ability to cling to vertical surfaces after walking only

a few steps on clean glass after being dusted with 2.5 µm radius silica microspheres [79].

The images of microspheres before and after self-cleaning steps are shown in (Figure

2.9A and B). The extent of recovery is shown in (Figure 2.9C and D). Self-cleaning of isolated setal array and experiments with geckos’ unique toe peeling motion (digital hyperextension) restricted suggest that self-cleaning is an intrinsic property of gecko adhesion.

Contact mechanical models suggest self-cleaning occurs by an energetic non- quilibrium between the adhesive forces attracting a dirt particle to the substrate and those attracting the same particle to one or more spatulae as shown in (Figure 2.9E). When the energy (force) required to remove a contaminant particle from the wall is greater than that required to remove the same particle from a spatula, self-cleaning will occur. For small contaminants (Rp < 0.5 μm), there are not enough spatulae available to adhere to the particle due to dimensional constrains. For larger particles, the curvature of the particle

makes it impossible for enough spatulae to adhere to it. The model also suggests that self-

cleaning requires the surface energy of spatulae to be relatively low (equal to or less than

that of the wall), limiting the spatulae to be made of a hydrophobic material. Hui et al.

argued that in the above hypothesis the particles must be sufficiently soft to achieve good

contact with the wall. This scenario is unrealistic because typical dirt particles are

sufficiently rigid, and thus the adhesion between the particle and surface is nearly zero. It

was proposed that geckos get rid of hard particles by scrubbing them off while walking

[192]. Gecko inspired synthetic adhesives with self-cleaning properties have been successfully fabricated to mimic this important feature of geckos’ feet [154, 157, 193].

Figure 2.9 Self-cleaning of gecko feet. (A) SEM image of seta array after dirtying with microspheres. (B) after five simulated steps. Microspheres are still present, but much less. Recovery of force by self-cleaning for (C) isolated setal arrays on glass after n simulated steps. (D) for live geckos. (E) Model of interactions between N gecko spatulae of radius Rs, a spherical dirt particle of radius Rp, and a planar wall. [79]

2.2 Modeling of Gecko Adhesion

The adhesion in geckos and its mechanism have been studied for a long time. But,

the role of multi-level hierarchical structure has not been fully understood. Intuitively, the

contact area of hairy pad on a smooth substrate is reduced compared to an equal-sized smooth adhesive pad. Assuming adhesion is proportional to contact area [194, 195], the hairy morphology should reduce adhesion. In nature, hairy adhesive organs have evolved independently at least three times in lizards including geckos [28, 177, 196], at least three times in insects [197] and occur in three phylogenetically distant groups of spiders [26,

29]. The abundance of hairy pads suggests that hairy pads represent an optimized design for surface attachment over millions of years of evolution. But, what is the advantage of this convergently developed hierarchical morphology? Contact mechanic models have been used to simulate the adhesion of gecko on smooth and rough surface. The results of

these simulations partially answered the above question and provided guidelines for

engineering of synthetic mimics.

2.2.1 Modeling of Adhesion in Geckos

Contact mechanics theories have been used to understand and model adhesion

mechanisms in geckos. The classical Hertz theory [198] assumes no attractive

interactions between contacting objects. Johnson et al. extended the Hertz theory to

contact between adhesive elastic spheres and developed the Johnson–Kendall–Roberts

(JKR) model [199, 200] in which the size of the contact area is determined via a balance

between elastic and surface energies similar to Griffith’s criterion for crack growth [201]

in an elastic solid. The JKR theory is quite appropriate for modeling contact between

large and soft materials. For stiff contact, the assumption of a crack-like singular field

becomes increasingly inaccurate for small and stiff materials leading to DMT

(Derjaguin–Muller–Toporov) model [202]. JKR theory has been used quite frequently to

model the contact in gecko adhesion

2.2.1.1 JKR Model and Contact Splitting Principle

The contact of the setae and spatulae with a smooth flat surface has been modeled

using continuum theory and highly simplified geometries. For cylinders with a

hemispherical end of radius R adhering to a flat surface, JKR model which considers

attractive interaction predicts the diameter of the contact area:

12 = + 3 + [6 + (3 )] , 3 푅 2 푑 ∗ ∙ �퐹 휋푅훾 휋푅훾퐹 휋푅훾 where R is the radius of the퐸 hemisphere, E* is the reduced modulus, F is the compressive

force and γ is the work of adhesion. This equation also predicts a finite force to separate the cylinder from the surface after the contact is made:

= . 3 푐 Applying this simple geometry to 퐹spatulae2 휋푅훾 contact and using values of R = 100 nm and γ = 10 ~50 mJ/m2, the predicted pull-off force for a gecko spatula is 4.7 ~23.6 nN.

The result matches the range of single spatula force measured by AFM [49, 50] .The model predicts the pull-off force remarkably well, despite its simplicity.

Another important prediction of the JKR model is the enhancement of adhesion

by contact splitting. Again, by using the simple hemispherical end of spatulae adhering to

a flat surface contact geometry and replacing a large spatula with R into n smaller spatulae with radius of Rs, the pull-off force of single spatula scales with R, but the

2 number of contact scales with (R/Rs) . The difference in scaling power law results in a remarkable enhancement factor . The total pull-off force of n smaller spatulae is given by: √푛

= ′ 푐 푐 Thus, a greater number of 퐹(smaller)√푛 ∙contact퐹 elements per pad area increase the overall adhesion [46, 52]. It can also be interpreted that total intimate contact area made by the elements is increased by splitting. This principle is termed “contact splitting principle” and has been applied to explain the correlation of setal density with body size in adhesive hair bearing animals, because larger animals with relatively less available surface area such as geckos require a stronger adhesion per unit attachment area than smaller animals such as insects. This model has been supported by data on the scaling of density of setae with body mass from diverse animals as shown in Figure 2.10A [52]. But a more careful re-analysis of the data on seta density and body mass as shown in Figure

2.10B suggests a large difference in hair density between dry and wet adhesive systems

[203, 204]. The density of wet contacts for insects is much smaller than that of dry contacts for geckos and spiders. Evarcha arcuata seems to be an exception to “contact splitting principle”. The contact density for Evarcha arcuata is similar to other spiders while its bodyweight is much less.

The “contact splitting principle” has several inherent assumptions. Firstly, it is implied that the load (pull-off force) is evenly distributed over all the contact elements

(setae and spatulae) and all elements detach from the surface simultaneously, so that

Figure 2.10 Scaling of contact element size and density with body mass. (A) Terminal elements (circles) in animals with hairy design of attachment pads. Heavier animals exhibit finer adhesion structures. (B) The large difference in hair density between dry and wet adhesive systems. [52, 204]

the total pull-off force of a hairy pad is the sum of the pull-off force of all single setae or spatulae. However, this assumption requires the contact elements to be identical in length and no stress concentration point in the system. Using 40 ~ 200 µN shear force generated by single seta, density of 14400 setae/mm2 and “contact splitting principle”, the total

force generated by a Tokay gecko with 227 mm2 total pad area is 131~654 N which is

much larger than the reported value. Secondly, this assumption will not hold when hairy pads are removed by peeling (curling of toes). During peeling, stress is concentrated at

the edge of crack propagation where only a small number of setae close to the edge line

share the load unevenly [205]. Consequently, the pull-off force in peeling will be much smaller than what is predicted by “contact splitting principle”. Indeed, the pull-off force is much smaller when the setae are pulled at an angle greater than 30° as reported by

Autumn [43].

Both JKR and DMT models predicted infinite adhesion strength as the size of

contacting objects is reduced to zero so that the adhesion can be improved just by making

the contact smaller and smaller. These results are realistic in physics because the adhesion strength cannot exceed the theoretical maximum strength of adhesive interaction. On the other hand, the adhesion strength can approach the theoretical strength for any contact size via shape optimization. Theoretical strength can be achieved by either tweaking the shape of the contact surfaces or by reducing the size of the contact object. But shape effect becomes less important when the size decreases. The other limitation of JKR model and contact splitting is that it does not predict shear force directly. It is reasonable to predict that shear force also is increased by contact splitting, since JKR equation predicts real contact area and shear force is proportional to real contact area. Furthermore, the preload dependence of shear force for single seta could not be explained by this simple model, since it predicts a finite pull-off force regardless of preload.

2.2.1.2 Kendall Peeling Model

Another simple but more realistic contact geometry model considers the shape of spatulae as strips of adhesive tape [49, 206]. When a tape is peeled off from a surface, the

stress is concentrated at the edge of the peeling (crack). Using the approach of Kendall

[207], the peel force at 90° is given by F = γ·w for spatula being pulled perpendicularly,

where w is the width of the tape (spatula ) and γ is the adhesion energy as in JKR model.

This model neglects stored elastic energy due to tape elongation under the tensile load which is very small at 90°. Using a typical value of γ = 10~50 mJ/m2 and w =200 nm,

Kendall peel model predicts a pull-off force for single spatula force in the range of 2~10

nN. This value is remarkably close to measured value for a single spatula [49, 50, 159].

If the angle between stalk of seta or spatula and the surface is not 90°, the Kendall peeling model should include a term for elastic energy:

( ) = + 2 2 1 cos 퐹⁄ 푤 퐹⁄ 푤 퐺 where F is the peeling force, w ℎ퐸is the width− of the휃 tape, h is the thickness of the

tape, E is the elastic modulus of the tape, θ is the peeling angle, and G is the adhesion energy required to fracture a unit area of interface at a peeling angle of 90°.

Federle [204] argued that “effective work of adhesion”, the energy (work) required to bend and stretch the tape or the fibers during detachment, is much greater than the energy needed to create two new interfaces [208, 209]. Due to the discontinuous geometry of adhesive contacts, most of this energy is not transmitted to neighboring fibers behind the crack and is dissipated upon detachment and fibers act as effective

‘crack arresters’ [205, 210, 211]. Chung et al. examined the role of discontinuities in adhesion and crack propagation of patterned adhesive films [211]. These films were made of silicone elastomers and were patterned by lateral, longitudinal or cross incisions.

The adhesion was measured by a displacement-controlled peel experiment. It was found that the crack propagation on these patterned adhesive films is controlled by the depth

and spacing of these simple incision patterns. With the crosswise incisions, the stress

distribution was more uniform and significant enhancement by a factor of 10 ~ 20 in

fracture energy was achieved.

2.2.1.3 Peel Zone Model

A more satisfactory peeling model for spatula was developed by Tian et al. [212]

to incorporate shear forces which are not predicted in the JKR and Kendall peeling model.

This model considers the molecular origin of the shear, friction and adhesion forces from

the van der Waals forces between the two surfaces. For two contacting planar surfaces,

the surface potential Ex along the x direction (parallel to the surfaces) can be described

approximately by an amplitude-varying surface potential (sinusoidal function here for

simplicity as shown in Figure 2.11A):

2 = ,

푥 0 휋푥 퐸 퐸 ∙ 푠푖푛 � 0 � so the shear force is: 푥

2 = = 2 , 푥 0 푓 푑퐸 휋퐸 푥 퐹 − − 0 ∙ 푐표푠 � 휋 0� where x0 is a critical spacing related푑푥 to the atomic푥 lattice, molecular푥 or asperity dimension on the spatulae and substrate. Under zero compressive force, the maximum static friction

max or shear force occurs when Ex = 0, FL = Ff . The interface begins to slide or slip if

max lateral load exceeds Ff . Otherwise, the friction force is simply Ff = FL, and the

max interface remains static. FL < Ff usually is satisfied during gecko attachment and detachment, that the shear force never reaches the ‘‘critical’’ point where the spatula starts to slip.

Figure 2.11 The interaction between two surfaces. (A) The origin of the shear (lateral or friction) force. (B) The Lennard–Jones surface–surface potential in the normal (z direction) to the substrate, which determines the normal attractive or adhesion force. [212]

The normal potential Ez and the normal force FvdW between two surfaces along the z direction are described by Lennard–Jones potential involving an attractive energy

EA and a repulsive energy ER as shown in Figure 2.11B . When two surfaces are pressed together with D < D0, the force FvdW is repulsive. When two surfaces are at D > D0, FvdW

max is attractive, reaching a maximum value of Fn = FvdW (pull-off force). If the pull-off

max force is greater than FvdW , the surfaces spontaneously detach. The repulsive side is usually very steep and can be approximated as a hard wall for D < D0. For two flat surfaces, the maximum attractive force per unit area is:

= . 6 퐴 푓푣푑푊 3 The forces F(θ) pulling through shaft 휋퐷of spatula0 at angle θ are balanced in three regions as shown in Figure 2.12A. (1) The contact region from x = 0 to x1, where distance between the spatula and the surface is at D0 and FvdW =0. Lateral component of F(θ) is balanced by the shear force. (2) The load bearing “peel zone” between x1 and x2 where

the total van der Waals force FvdW of the spatula is balanced by the force F(θ) along the

spatula shaft. (3) for x > x2, the van der Waals force acting on the shaft is too weak

beyond this cut-off distance Dc and can be neglected, thus the tension or pulling force remains constant and equal to F(θ) along the shaft. The trends for normal, lateral force and the total pulling force F(θ) predicted by this model are shown in Figure 2.12B. The region shown by the horizontal shaded band is the range of maximum values for F(θ) before the surfaces detach or slip, determined mainly by maximum static friction force

max max Ff . The condition F(θ) < Ff is clearly satisfied for angles θ greater than ≈ 10°. It also

shows that the pulling force F(θ) can vary by more than two orders of magnitude depending on shaft angle θ. For small θ it is determined mainly by the lateral force while

for large θ it is determined mainly by the adhesion force. The cross-over angle is about

max 45°. When F(θ) > Ff at very small angles, the spatula begin to slide or slip on the

substrate.

Figure 2.12 Peel zone model of spatula. (A) Contact geometry and force balance in contact region, peel zone and non-contact region of a spatula being pulled at angle θ. (B) Angle dependence of normal force component FvdW, the shear force component Ff, max the net pulling force F(θ), and the maximum shear force Ff can be obtained from a spatula pad in contact with a substrate (shaded band). [212]

The advantage of this model is that it predicts force based on the angle at which

the spatula is being pulled and a cross-over point. The disadvantage is that it is still difficult to theoretically predict an accurate value for FvdW of a single spatula due to the

uncertainty of The Hamaker constant, D0, cut-off distance Dc, width b and thickness h of

the spatulae. Larger values of the FvdW could be obtained by using larger values for A, b,

Dc, or a smaller value for Do. To compare with experimental results, the shear force of single seta is estimated to be about 200 ~ 2000 nN by using maximum experimental shear force of a single seta of 200 μN [43] and assuming 100 ~ 1000 spatulae per seta. The result is in agreement with the prediction by peel zone model at low angles. At 90° pulling angle, the model predicts a pull-off force of 16 nN for a single spatula, which is very close to the values of 10 nN measured by Huber et al. [49, 159] and 2–16 nN by Sun et al. [50] in which the spatulae were pulled perpendicularly. However, adhesion forces of spatula have never been measured when pulling angle is varied. The peel zone model can be intuitively demonstrated by peeling a strip of Scotch® at variable peeling angles

[53]. The peel force is much smaller at 90° angle.

2.2.1.4 Effect of Contact Shape and Size

In previous contact-mechanics analyses, it was assumed that the terminal

elements can be modeled as having hemispherical shape. Morphology studies, however,

show that the shape of spatulae differs substantially from hemispheres. The analysis of

the effects of different shapes on the adhesive force in the framework of contact

mechanics was presented by Spokenak et al. [69, 206]. The contact mechanics of single

contacts are extended to “contact splitting” for optimizing adhesion. Figure 2.13 shows

five different contact shapes. Table 2.3 listed the formulae for pull-off force for these shapes.

Contact shape can exert a strong influence on adhesion and efficiency of contact splitting. Big differences arise for comparatively larger contact elements of relatively stiff material. At very small contact sizes (several tens of nanometers) and low modulus

(below 1 MPa) the influence of shape becomes negligible. The best shape for perfectly smooth surfaces is the flat punch followed by the torus and the sphere. However, the flat punch does not provide a ‘robust’ contact due to sensitivity to roughness and dirt particles which damage molecular contact. By comparison, the suction-cup mechanism can only compete at large contact sizes and is inefficient in the sub-micrometer regime.

Figure 2.13 Contact geometries and shapes of contact area: (A) sphere, (B) flat punch/vertical cylinder, (C) torus/horizontal cylinder, (D) suction cup, (E) elastic tape. [206]

The total pull-off force can be improved by splitting the contact into many finer

contacts of identical shape. The benefit of this “contact-splitting” strategy also depends

on the shape. Splitting a macroscopic contact into n contact elements, the total apparent

contact area remains unchanged, but the pull-off force is greatly increased to

= , ′ 푟 푐 푐 where r is a measure of contact splitting퐹 efficiency:푛 퐹 larger values of r result in greater enhancement in pull-off force. Table 2.3 lists the r values for different shapes. It is readily seen that the spherical contact, especially under curvature-invariant conditions, offers the best splitting efficiency (r = 1), followed by the elastic tape (r = 1/2), and the torus (r = 1/3 or 1/2). The flat punch does not respond favorably (r = 1/4). The adhesion of suction cups cannot be improved by contact splitting (r = 0).

Natural attachment systems seem to reflect some of the theoretical findings of this study. The flat punch is hardly found in nature. Very compliant contact elements (e.g. in the grasshopper) have shapes resembling a flat punch, whereas stiffer contact elements exhibit torus like shapes (e.g. in insects) or tape-like structures (e.g. the gecko). The

suction-cup mechanism occurs in nature but the size of the cups never lies below 10 µm and it rapidly loses its advantage when sizes reduce. The torus-like structure, which is found in attachment devices of fly and beetle is a preferred shape as it combines relatively high absolute forces with an appreciable scaling efficiency.

For a given contact area A, the theoretical pull-off force of the fiber is σthA, where

σth is the theoretical strength of adhesion. It is possible to design an optimal shape of the

tip of the fiber to achieve the theoretical pull-off force [70]. This optimal shape has uniform separation of the two contacting bodies over the entire contact region where

Table 2.3 Pull-off force and scaling efficiency of contact shapes [206]

Shape Pull-off force Pc r

3 * Sphere R 1/2, 1 2 − πγ Flat punch/vertical 8 E R 1/4 cylinder ∗ 3 −� π γ Horizontal cylinder and E r * 3 R , r R 1/3, 1/2 3 2∗ 2 torus π γ − π � ≪

Suction cup p R 0 2 ∆ π 4 R sin

Tape peeling 2 1/2 (1 cos −) γ+ +α(1 cos ) Eh 2 γ � − α − α *Invariant radius

stress field is uniform at pull-off. .However, such design tends to be unreliable at the

macroscopic scale because the pull-off force is sensitive to small variations in the tip shape. It is interesting that the shape-insensitive optimal adhesion becomes possible when

the diameter of the fiber is reduced to length scales on the order of 100 nm. As for very

small contact elements the influence of the shape diminishes [70, 206, 213, 214],

hemisphere and elastic tape which have identical scaling efficiency then become efficient

contact geometries. Spatulae in spiders (hemispherical) and the spatulae (elastic tape) in

geckos are the smallest terminal contact elements known. In general, optimal adhesion

could be achieved by a combination of size reduction and shape optimization. The

smaller the size is, the less important the shape is. At large contact sizes, optimal

adhesion could still be achieved if the shape can be manufactured with a sufficiently high

precision.

2.2.1.5 Sliding Friction Model

Setae during sliding exhibit two characteristic behaviors as shown in Figure 2.4:

(1) shear force increases and reaches maximum at steady state without stick-slip, (2)

maximum shear force increases with shear speed. Model based on stick-slip during sliding was proposed to explain the above behavior [51]. It is assumed that setal shafts remain in tension during sliding and are rigid, while the spatulae undergo uncorrelated stick-slip. Spatulae cycle through a stick-slip process governed by two characteristic times: tstick, the average lifetime of a spatula-substrate bond and τ0, the characteristic time

for a spatula to stick again to the substrate after bond rupture which is determined by the

resonant frequency of the spatula shaft. At a low sliding speed, most of the spatulae are

stuck to the substrate because reattachments occur rapidly relative to the speed of sliding.

At high sliding speed, the reattachment becomes slow compared with the motion of the

substrate, resulting in a decline in shear force because fewer spatulae attach

simultaneously as sliding speed increases. So the rate dependence of number of contacts

can be calculated from materials properties and sliding speed. In the sliding experiment,

it was calculated that the reattachment time was much faster than sliding speed.

Logarithmic force-velocity model for atomic stick-slip was used to predict the rate

dependence of the force of each contact. Rate dependence of the total force of a spatulae array can be estimated by combining the above stick-slip model and force-velocity model.

The models suggest that smooth macroscopic kinetic friction is the result of uncorrelated stick-slip of all the discrete microscale contact elements sliding along the surface.

2.2.1.6 Adhesion on Rough Surface

Instead of molecular smooth surfaces, geckos often encounter rough surface in nature. Although, their claw can help them climb on macroscopic rough surface, it will do work when the characteristic length of surface roughness is much smaller than the tip of their claw. Then, it is important to find out how efficient their hairy pads adhere on such rough surface.

It is known that solid surfaces are usually rough and do not spontaneously adhere to each other because the intimate contact area is only a very small fraction of the apparent contact area [215]. Good adhesion can be achieved if at least one of the surfaces is very soft [216]. Adhesive materials, such as pressure sensitive adhesive (PSA), are sufficiently soft and tacky and can adapt to topography of rough substrates and achieve intimate contact. PSAs are fabricated from soft viscoelastic materials that satisfy

Dahlquist’s criterion which states Young’s modulus for tacky materials to be ~ 100·kPa or less at room temperature [217, 218]. This kind of soft and tacky material is typically more susceptible to creep, degradation, contamination and wear and tear during normal use.

Hairy design of geckos’ adhesive pads makes it easy to adapt to the topography of rough or non-flat substrates and achieve intimate contact. Due to flexibility of setae and

spatulae, the arrays of setae actually have a very low effective elastic modulus [44] and behave like soft solids even though they are made of relatively stiff materials [44, 73].

Gecko setae are composed mostly of β-keratin [175, 219]. The elastic modulus measured from single seta is ~1.6GP [73], which clearly indicates it is a relatively stiff material.

The result is similar to that of β-keratin previously measured in birds, which is consistent with the highly conserved the molecular structure of β-keratin across birds and reptiles.

The effective moduli of setal arrays during vertical and + 45° compression (along the natural path of drag of the setae) measured by Autumn et al. were 83 kPa and 86·kPa, respectively [44]. Mechanical modeling predicts that adhesives can be made from stiff materials if they take the form of high aspect ratio fibril array.

The mechanical behavior of setae arrays during compression and relaxation depends on the mode of deformation of individual seta. Bending is a likely mode of deformation of setae and a simple approach to understand this is to model arrays of setae as cantilever beams [205, 208, 220-222]. The cantilever model suggests that the effective modulus of the fibril array decreases with the diameter of the cantilever beam which favors greater fraction of intimate contact area on rough surfaces. The cantilever model also predicts that longer and softer setal shafts and a lower shaft angle will contribute to a reduced effective modulus.

However, surface roughness could vary over several orders of length scale. Not only modulus but also the size of the contact elements relative to the characteristic length scale of rough surface determines total real contact area [208, 221]. To achieve sufficient molecular contact area to a rough substrate, an adhesive must be able to conform to the surface roughness that spans a wide range of length scales. Geckos’ hairy adhesive pads

with hierarchy at different length scales from nanometer sized spatulae to micrometer sized setae, could achieve intimate contact by conforming to the surface roughness profile at those length scales. (1) The underlying tissue, the setae must be compliant enough so that they can adapt to the large and intermediate scale surface profile and (2) terminal elements (spatulae) must be very small and flexible to compensate for smaller scale roughness. If amplitude of surface roughness is large, setae in contact of apexes of surface profile are compressed to allow some setae to reach and make contact with the valleys of surface, so the overall adhesion is reduced because the elastic energy stored in the compressed setae tends to separate the array from the surface. Therefore, good adhesion can only be achieved by making setae very compliant. However, flexibility of setae should be controlled to avoid self-adhering (or self-mating), because softer setae also tend to stick to each other.

Even when setae are compliant enough to follow the intermediate scale roughness, surface roughness smaller than the size of a terminal element can compromise intimate contact and result in poor adhesion. The terminal elements, such as spatulae, are often extremely thin and flexible suggesting that part of the small scale roughness is compensated by the flexibility of the terminal plate [221]. The analysis of mechanics of adhesion of gecko on rough surfaces suggests that the hierarchical structure plays an important role in geckos’ ability to explore large variety of surfaces with roughness at large and small length scales.

The adhesion of spatulae and clinging ability of live Tokay gecko to substrates with different roughness was investigated by Huber et al. [164] Rough surfaces having root mean square amplitude (RMS) values ranging from ~ 20 nm up to ~1.1 µm were

prepared from polyvinylsiloxane templates using epoxy resin. The result for single spatula is shown in Figure 2.14. The attachment ability of live Tokay gecko was tested on five different types of polishing papers with nominal asperity size of 0.3, 1, 3 , 9 and 12

µm. The pull-off forces showed a distinctive minimum between 100 nm and 300 nm

RMS roughness. The results for live gecko are listed in Table 2.4.

The experimental results for single spatulae are confirmed by measurements on live geckos. The animals could adhere perfectly to either very smooth or rough surfaces but slipped more easily on substrates with an intermediate RMS roughness of about 300 nm. This trend has been predicted by Persson et al. for the adhesion of lizards on sandpaper surfaces [221].

Figure 2.14 Pull-off forces of two different spatulae on rough surface. The continuous line is a guide to the eye. [164]

In contrast to the above results, Pugno et al. observed a maximum in geckos’ adhesion with a PMMA surface having an intermediate roughness of RMS = 618 nm and concluded that RMS alone is not sufficient to describe the three-dimensional topology of a complex surface [76]. In addition, the waviness of the surface can alter the adhesion significantly even if the RMS is the same.

Table 2.4 Adhesion of live Tokay gecko on rough surface [164]

asperity size Surface RMS (nm) Observation (µm) Smooth Adhered stably to the ceiling for at Small small glass least 5 min. Polishing Not able to stay; slid on the slope of 0.3 90.0 ± 2.7 papers 135°. Clung to the ceiling for a while, but Polishing toes slowly slid off the substrate and 1 238.4 ± 6.0 papers the contact had to be continuously renewed. Polishing Adhered stably to the ceiling for at 3 1156.7 ± 133.1 papers least 5 min. Polishing Adhered stably to the ceiling for at 9 2453.7 ± 87.2 papers least 5 min. Polishing Adhered stably to the ceiling for at 12 3060.3 ± 207.7 papers least 5 min

2.2.2 Modeling of Hierarchical Structures

Various models for hierarchical structured gecko toe have been proposed to simulate the behavior of attachment, detachment and adhesion on rough surfaces [60-67].

Yao et al. demonstrated theoretically that a self-similar, multilevel fibrillar microstructure could be designed from the bottom (nanoscale) up to achieve flaw-tolerant adhesion at

multiple length scales. The hairy microstructure was also modeled as a strongly

anisotropic material and it was shown that the adhesion force varies with the pulling direction. Bhushan et al. developed a three-level hierarchical spring model to investigate

the effect of surface roughness and fibrillar stiffness on adhesion. Schargott et al. used a

3D tilted beam system to simulate adhesion in geckos.

2.2.2.1 Fractal Gecko Hair Model

A fractal gecko hairs model as illustrated in Figure 2.15(A) assumes self-similar fibrillar structures at multiple hierarchical levels mimicking the ultrastructure of geckos’ spatulae [67]. It was introduced by Yao et al. to show that structural hierarchy plays a key role in adhesion.

In the adhesion of every two immediate levels, a larger fiber with a hairy top surface is in contact with a substrate as illustrated in Figure 2.15 (B). The larger fiber contains a number of thinner fibrils on its top surface resulting in a two-leveled structure.

For these two levels of structures as a whole, the effective work of adhesion for the larger fiber is no longer equal to thermodynamic work of adhesion even though van der Waals forces are the only interaction between these small fibrils and the substrate. While the stress-separation curves for fibrils at the bottom level are described by intermolecular interactions, the separation at the level of the higher fiber is strongly influenced by the elastic properties and geometry of the bottom level fibrils. If the fibrils are long enough, the elastic deformation of the fibrils will make significant contributions to the separation

process and adhesion failure occurs by an abrupt drop in stress near the theoretical

strength of surface interaction. In this way, the strain energy stored in the fiber becomes

part of the “cohesive energy” of the higher level fibrils to be dissipated through dynamic snapping of the thin fibrils. The effective work of adhesion can be increased almost three orders of magnitude.

Figure 2.15 Fractal gecko-hair model. (A) Bottom–up design scheme of a hierarchical fibrillar structure. At each level, the fibers depend on smaller fibrils at lower hierarchical levels as effective “adhesive bonds” with a surface. The fibers themselves act as “adhesive bonds” for larger fibers at higher hierarchical levels. (B) Schematic of a hairy surface containing arrays of fibrillar protrusions contacting a substrate. [67]

Using the fractal gecko hair model, a general repetitive procedure can be formulated to determine the enhancement of “work of adhesion” parameters at all hierarchical levels, starting from the lowest level. Calculation shows that multiple levels of hierarchy may be needed to achieve flaw tolerant adhesion at macroscopic length scales.

Yao et al. also compared results with the observed hierarchical structure of gecko.

Using typical values of material properties for β-keratin, the diameter and length of the first level fiber are 140 nm and 1.37 μm. These values are not far off from the dimensions of spatula hairs of Tokay gecko which is around 100–200 nm wide and 0.5–3 μm long.

The dimension of the second level is around 7.56–11.34 μm wide and 286–491 μm long, which is close to the dimensions of a seta on geckos’ feet which are with about 5 μm in width and 110 μm in length. In addition, calculations based on this fractal model predicts that the number of the lowest level fibrils on the fiber of the second level is around 1539–

2309, which agrees with the observation of 100–1000 spatulae/seta. The density of the second level is 11,494 setae/mm2 for triangular shaped first level. This is also comparable to the observed density of around 14,400 setae/mm2.

The consequence of this fractal model of gecko hair is that adhesion can be further enhanced by adding more levels, but geckos only adopt four levels of hierarchy from spatulae to toe. It is reasonable to ask why gecko has not evolved more hierarchical levels for heavier animals. A possible answer to this question is that the fracture of fibers eventually rises to become the dominant mode of failure at the system level as the adhesion strength is enhanced by introducing more and more levels of hierarchy.

It was shown that strong anisotropy allows the adhesion strength to vary strongly

with the direction. This orientation-dependence of pull-off force enables not only robust attachment in the stiff direction of the material but also easy release by pulling in the soft direction. When pulled in the stiff direction, less elastic energy can be stored in the material, leading to lower energy release rate to drive random crack-like flaws induced by surface roughness. On the other hand, much more elastic energy can be stored in the

material when pulled in the soft direction, especially when the material is strongly

anisotropic, leading to much higher energy release rate. This orientation-dependence was termed as ‘stiff-adhere, soft-release’ principle by these authors.

2.2.2.2 Spring Model

The effect of hierarchy on attachment ability on a random rough surface was also

simulated by Bhushan et al. as illustrated in Figure 2.16 using a multi level spring model

[60, 61, 65]. Numerical results were calculated using parameters based on gecko setae.

The base of the springs and the connecting plate between the levels are assumed to be

rigid. The range of roughness of the surfaces covers roughness for relatively smooth, artificial surfaces to natural, rough surfaces.

The force–distance curves were calculated for one-, two-, and three-level hierarchical systems for contacting surfaces with changing amplitude of roughness σ ranging from σ = 0.01 μm to σ = 10 μm and an applied preload chosen based on the average bodyweight of geckos. The model system is compressed against the rough surface with a preload, and then it is pulled away from the surface. It is very clear from the simulation that the effect of multi-level hierarchy on normal pull-off force is very

limited on relatively smooth surface (σ = 0.01 µm), but the effective work of adhesion is

greatly improved. On intermediate roughness (σ = 0.1 µm and σ = 1 µm), both normal

pull-off and effective work of adhesion are increased with addition of higher hierarchy,

but the effect decays with greater roughness (σ = 10 µm). Adhesion energy stored, adhesion coefficient, the number of contacts per unit length, and the adhesion energy per unit length of the one-, two-, and three-level models were calculated over a wide range of roughness. The adhesion coefficient is defined as the ratio of the pull-off force to the applied preload, which is a measure of the strength of adhesion with respect to the preload. The maximum adhesion coefficient is about 36 at σ = 0.01 μm. This means that a small preload can induce very strong adhesion. However, if σ is increased to 1 μm, the adhesion coefficient for the three-level model drops to 4.7. The adhesion coefficient falls below 1 when σ is greater than 10 μm. This means that the model system yields a smaller force than preload. The adhesion coefficient for the two-level model is lower than that for the three-level model, but there is only a small difference between the adhesion forces of the two- and three-level models because the stiffness of level III for the three-level model is chosen to be higher than those of levels I and II. A less stiff third level could improve adhesion coefficient by 20–30%. The results also showed that the trends in the number of contacts are similar to the trend of the adhesion force. The adhesion energy decreases with an increase of σ. For a smooth surface with σ = 0.01 μm, the adhesion energies for two- and three-level hierarchical models are 2 and 2.4 times larger than that for the one- level model, respectively, but these values decrease rapidly at surfaces with σ greater than

0.05 μm; and in every model the adhesion energy finally decreases to zero at surfaces with σ greater than 10 μm. In the three level system, there is a 2–3 times increase in

Figure 2.16 Spring models of hierarchical structures. One-, two-, and three-level hierarchical spring models for simulating the effect of hierarchical morphology on the interaction of a seta with a rough surface. I, II, and III are hierarchy indexes. lI, lII, and lIII are lengths of the springs, SI is the space between spatulae, kI, kII, and kIII are stiffnesses of the springs. R is the radius of the tip, and h is the distance between the base of the upper spring of each model and the mean line of the rough profile.[60, 61, 65]

adhesion energy when the stiffness of the third level decreases 10 times. It is also shown that as the preload increases, the adhesion force increases up to a certain applied load and then has a constant value, while adhesion energy continues to increase with an increase in the applied load.

However, the measurements of adhesion force of a single gecko seta demonstrated that both perpendicular preload and lateral drag were needed for effective attachment. This hierarchical spring model considers only normal-to-surface deformation and forces.

Additional simulation in lateral direction could make the model more applicable.

2.2.2.3 Hierarchical Tilted Beam Model

Based on the hierarchical structure of gecko toe pads, a three dimensional model for hierarchical layers of tilted beams was constructed by Schargott [66]. As shown in

Figure 2.17, each layer was constructed in such a way that no cohesion (self-mating) between adjacent beams can occur. Simulation based on this three dimensional model showed that the effective modulus for arrays of stiff materials (e.g., E = 4GPa) can be significantly reduced to less than 100 kPa simply by tilting and adding hierarchical layers.

The adhesive force was calculated numerically. For the adhesive contact on stochastically rough surfaces, the maximum adhesion force increases with an increasing number of hierarchical layers. Additional calculations show that the more layers are involved, the more the adhesive properties depend on the texture of the rough surface

Chen et al. investigated the mechanics of reversible adhesion of geckos in terms of the attachment and detachment mechanisms of the hierarchical microstructures on the

scale of toe [62]. At the very bottom level of the hierarchy, because the spatula pads are

Figure 2.17 Hierarchical system of tilted beams, built bottom to top with a reference length scale of ao.[66]

very thin (thickness of approximately 5 nm), they are easily absorbed on to a solid

surface. The de-cohesion process zone size of the spatula pad at the interface is small so

that small spatula pads can make full use of the area for adhesion. There are 10 times greater pull-off forces at slightly less than 30° peeling angle than 90° peeling angle. A

slender hairy structure can do more than providing much higher adhesion energy than the

intrinsic energy associated with van der Waals interaction. It highly magnifies the

difference in adhesion energy between attachment and detachment. Limiting the diameter

of the seta below a critical value ensures uniform stress distribution in the structure. By

uncurling (rolling in) and pulling on the toe for attachment while curling (rolling out) and

peeling the toe for detachment, the difference in adhesion forces between the two states has been further magnified at the scale of the toe. In such a way, the gecko attains an adhesion force much higher than its bodyweight with displacement controlled pulling and a detachment force much lower than its bodyweight with peeling at a large angle.

2.3 Design Guidelines for Synthetic Gecko Adhesives

The results of simulation with contact mechanics models, contact splitting principle, and hierarchical models not only give reasonable explanation to the morphology of biological hairy pad, but also provide the necessary guidelines for developing biomimetic adhesion systems by suggesting optimal dimensions and material properties [213, 222]. These parameters packed into ‘‘adhesion design maps” as shown in

Figure 2.18 include the following:

• Elastic modulus of the fibers and their radius.

• The limit of fiber fracture (blue). For very thin fibers, the adhesion strength will

exceed their theoretical fracture strength.

• The limit of ‘‘ideal contact strength” (red) is given by an ideally fitting contact

between the two surfaces, without necessity for elastic deformation.

• The limit of adaptability (green). Compliance, which is necessary to adapt to

rough surfaces.

• The apparent contact strength σapp, i.e. the pull-off force divided by the apparent

contact area Aapp, is shown as contours superimposed on the maps.

• The limit of condensation or clumping, matting (cyan). The tendency of fibers

sticking to each other, rather than contacting surfaces.

Conode was defined to ensure adaptability. These loci are shown as black dots at the intersections of the green and cyan lines for different values of the fiber aspect ratio λ.

The optimum adhesive is found at the intersection of the conode with the fiber fracture limit, as indicated in the map (red circle). Additionally, the tilting angle with respect to the surface and hierarchy needs to be considered for release mechanism and adhesion on rough surfaces.

Figure 2.18 Adhesion design map for different contact shapes. (A) spherical contact, (B) flat tips, (C) toroidal tips (D) elastic tape at a peel-off angle of 60°. [213, 222]

2.4 Fabrication and Application of Gecko-Inspired Adhesives

Morphological studies of adhesion system in geckos and other hairy pad bearing

species revealed the complexity of these dry adhesives. The task of developing gecko-

inspired adhesives is a very challenging one due to the difficulty of making an array of

complex features at small length scales. Optimal size, aspect ratio, tilted angle, shape,

hierarchy, and materials properties that are required for a high-performance dry adhesive

are not easily achieved during fabrication.

The first generation of gecko inspired adhesives were arrays of micrometer scale polymer pillars [71, 80]. Their adhesion performance was poor. Recent advances of micro-fabrication enabled production of complex structures that incorporate more and

more design parameters. Synthetic adhesives with improved adhesion, easy release, self-

cleaning, and other features have been reported. Polymeric hairs [46, 71, 72, 80-151] and carbon nanotubes [152-158] have been most frequently used as materials for synthetic gecko hairs. Generally polymer-based adhesives have been fabricated by a top-down approach, including cutting of polymer films, drawing polymer fibers from melts and solutions, molding with templates, direct lithography and etching techniques. Advanced top-down fabrication techniques have been implemented to fabricate templates with slanted nanoholes and controlled contact shapes. Polymer-based methods offer versatile approaches to fabricating gecko-mimicking nanohairs with tailored geometry (radius, height, shape of tip angle and hierarchy) and a wide range of material properties

(modulus, surface energy). However, the adhesion strength is usually lower than that of geckos and CNT-based adhesives because of the relatively large feature size and poor mechanical strength of polymer materials at small length scales. The CNT-based

adhesives have been fabricated by a bottom-up approach. Vertically aligned CNT arrays are grown from the catalyst layer deposited on a substrate by chemical vapor deposition

(CVD). CNT-based dry adhesives usually have much higher adhesion strength than

geckos and polymer-based adhesives due to the extremely small radius ( 10 nm) and high modulus ( 1000 GPa) of CNTs. Some exemplary work will be reviewed∼ in this

section. ∼

2.4.1 Polymer-based Adhesives

Autumn et al. fabricated polymer spatulae from polydimethylsiloxane (PDMS) and polyester resin by a simple AFM based molding technique [46, 80]. A mold was made by punching a flat wax surface with an AFM probe having a conical shape and a tip radius of 10 – 20 nm and a height of 15 µm. The punched surface was then filled with polymer. After curing, molded polymer surfaces were detached from the wax by peeling.

Normal pull-off force was 181 ± 9 nN for PDMS spatulae with tip radius of 230 – 440 nm and 294 ± 21 nN for polyester spatulae with tip radius of≈ 350 nm. This work was limited to fabrication of just a few synthetic spatulae.

Geim et al. fabricated a prototype of “gecko tape” [71]. A dense array of flexible

polyimide pillars (Figure 2.19) was made by electron-beam lithography and dry etching in oxygen plasma. Its geometry was optimized to ensure collective adhesion. A relatively large area of 1-cm2 of polyimide pillars fabricated on a piece of silicon wafer sustained

only 0.01 N normal pull of force with a very large preload of ~ 200 N. A relatively high

normal adhesion ( 3 N/cm2) was obtained after the array was transferred to a soft

substrate and pull-off∼ force scaled with tape area. It was concluded that the flexibility of

the substrate was crucial for equal load sharing and achieving a high pull-off force.

However, the polyimide pillar array requires very large preload and is very prone to mechanical failure. In addition, the slow and expensive e-beam lithography process is a major shortcoming of this approach. Interestingly enough, many researchers ignored fundamental differences between normal and shear measurements and did not report shear strength of their synthetic tapes, but they compared normal pull-off with live geckos’ shear value.

Figure 2.19 SEM image of array of polyimide pillars mimicking gecko foot-hairs. The pillars were fabricated by e-beam lithography and dry etching. [71]

Nanomolding methods have been developed to overcome the limitations of direct e-beam lithography. In these methods, various templates having nano-holes such as Si,

SiO2, anodic alumina oxide (AAO), polycarbonate film, are prepared by electron-beam

lithography, photolithography, etching or electrochemical reactions followed by molding

polymers against the substrates to generate nanohairs.

Majidi et al. [96] and Lee et al. [125] fabricated fiber arrays by casting a thin

layer of polypropylene film into a 20 µm thick polycarbonate filter having pore radius of

0.3~2.5 µm [96]. The polycarbonate filter was then removed by dissolving it in a solvent.

The fiber array (Figure 2.20A) demonstrated high interfacial shear strength, but no measurable normal adhesion on smooth surfaces due to the high stiffness of polypropylene. Arrays of angled microfibers were also fabricated from polypropylene as shown in Figure 2.20B and C. The fiber array was processed by rollers that were heated at 50 °C to form angled microfiber arrays. Friction experiments demonstrated that this fibrillar polymer surface exhibits directional adhesion. Sliding of clean glass surfaces against and along the microfiber direction without applying an external normal force produced an apparent shear stress of 0.1 and 4.5 N/cm2, respectively. High density

polyethylene can also be used to fabricate high friction hairs. Lee et al. fabricated HDPE

Figure 2.20 SEM images of polypropylene fiber arrays. (A) an array of 20 µm long, 0.6 µm diameter polypropylene fibers showing high friction. The fibers were fabricated by casting a film on polycarbonate membrane which was etched away later. (B) Top view of angled polypropylene fibers and (C) Side view of angled polypropylene fibers. [96, 125 ]

hairs on HDPE lamellar structures [142] that mimic the function of geckos’ scansors as

shown in Figure 2.21. Experiments showed that nanofiber arrays on lamellae adhered to both planar and nonplanar surfaces and they exhibited 5 times greater shear strength than the arrays without the lamellar supporting structure on a grating substrate with 100 μm peak-to-peak roughness. The observed behavior on nonplanar surfaces was attributed to the high compliance of the lamellar flaps.

Figure 2.21 Combined lamellae and nanofiber arrays. (A) Arrays of lamellar flaps. (B and C) Nanofiber arrays on the thin lamellar flaps.[142]

A self-cleaning, high-aspect-ratio fibrillar adhesive was fabricated by casting a single layer of a 25-μm-thick polypropylene (PP) film into polycarbonate (PC) membrane filter [193]. The contaminated synthetic adhesive recovered about 33% of its shear adhesion after multiple contacts with a clean, dry surface. The fibrillar adhesive shed about 60% of the contaminating microspheres with 1.5 and 2 μm in diameter onto the

glass substrate as shown in Figure 2.22, while some microspheres remained between the

fibers. For larger particles (3 and 5 μm), the adhesive force did not recover.

Jeong et al. presented a two-step technique, a simple method for fabricating

micro/nanoscale hierarchical structures [93, 120]. This lithographic method involves a

sequential application of the molding process in which a uniform polymer-coated surface

Figure 2.22 SEM images of self-cleaning polypropylene fibrillar adhesive. (A) contaminated by gold microspheres. (B) after 30 contacts (simulated steps) on clean glass substrate. [193]

is molded by means of capillary force above the glass transition temperature of the

polymer. The two steps consist of (1) molding using a low-resolution, micro-patterned template to fabricate larger structures and (2) molding using a high-resolution nano- patterned template to fabricate smaller structures on top of the microstructures from the first step. It can be combined with a fiber drawing method to elongate the fibers further.

The structure made by simple molding and hierarchical structures made by two-step molding are shown in Figure 2.23A-F.

Figure 2.23 Nanostructures fabricated by sequential molding. (A) Simple molding, (B and C) molding and nanodrawing process, (D) two-step molding and (E and F) two- step UV-assisted molding. [120]

An improved nano-molding method can be used to fabricate angled nanohairs as shown in Figure 2.24 [137]. The mold is made by an etching technique which offers precise control over size, length, leaning angle and tip shape and allows direct fabrication

of slanted nanohairs from a template containing nanoholes with controlled slanted angle

and tip shape. The resulting angled nanohairs showed a significant increase in the shear

adhesion force of 26.0 N/cm2 in the direction of inclination and much reduced force of

2.2 N/cm2 when pulled∼ against the direction of the inclined angle. In addition to simple

slanted nanohairs, micro-nanoscale combined hierarchical hairs as shown in Figure 2.25

were also fabricated by using an additional step of UV-assisted molding on top of the slanted hairs. These hierarchical nanoscale patterns maintained their adhesive force even on a rough surface where simple nanohairs lost their adhesion strength.

Figure 2.24 Slanted nano-hairs from UV-curable PUA resin. (A) well-defined PUA nanohairs with bulged flat top and slanted angle of 60° with respect to the horizontal plane. (B) top view showing clear spatulae-like flat head. [137]

Table 2.5 below summarizes polymer-based hairy dry adhesives with macroscopic adhesion greater than 3 N/cm2 in at least one test direction. To make an

apple to apple comparison, adhesion measured with a small spherical probe or at small

length scale such as those measured by AFM tip is not listed, because those results

generally cannot be used to predict macroscopic behavior.

Figure 2.25 Hierarchical hairs from UV-curable PUA resin. (A) A titled SEM image of two-level hierarchical PUA hairs formed over a large area. (B and C) Magnified, titled images of second level hairs. [137]

Table 2.5 Comparison of polymer-based synthetic gecko adhesives

Modulus Shear Normal Preload Contact Materials Method Hierarchy Ref Pa N/cm2 N/cm2 N/cm2 surface

PI ~3G e-beam - 3 50 Glass 1 [71]

Molding PMMA ~2.4G nanodrawin 3-4 ~0 - - 1-2 [120] g

Lithograpy PDMS ~2M - 22 0.24 Glass 1 [134] molding

PU 3M lithography 10 5 - Glass 1-2 [143]

PUA 19.8M Molding 11 ~0 0.3 Wafer 1 [139]

PUA 19.8M Molding 26 - 0.3 Wafer 1-2 [137]

PP 1G Molding 2.5 ~0 <0.1 Glass 1 [124]

HDPE 0.9G Molding 4 - - Glass 2 [142]

PI=polyimide, PMMA=polymethylmethacrylate, PDSM=polydimethylsiloxane, PU=polyurethane, PUA=polyurethane acrylate, PP=polypropylene, HDPE=high density polyethylene.

2.4.2 Carbon Nanotube Based Adhesives

Carbon nanotubes have been identified as an excellent material for mimicking

gecko foot-hairs due to its extremely high and controllable aspect ratio, outstanding mechanical strength and tunable flexibility and adhesion. Scaling law predicts that at

nanometer scale, the shape of contact becomes less important and the effect of contact

splitting becomes dominant. The effective modulus of the vertically aligned array is

several orders less than a single nanotube because of the extremely high aspect ratio and

low packing density. Dry adhesives based on vertically aligned multi-walled carbon

nanotubes (VA-MWCNT) [154, 156-158] and vertically aligned single-walled carbon

nanotubes (VA-SWCNT) [155] have been fabricated and shown excellent adhesion in

both shear and normal directions.

Yurdumakan et al. measured the adhesive force of a VA-MWCNT array using an

AFM method [152]. The VA-CNT array was embedded in polymer matrix and the tube

ends were exposed by solvent etching as shown in Figure 2.26A. The adhesion force of ~

70 nN, normalized by the area of the AFM tip, was reported to be 160 times greater than

the shear value measured from live geckos. It is highly possible that actually geometry of

tip-tube contact be side-wall contact even though the AFM tip is dipped perpendicularly

into the CNT array surface. This work shed light on using VA-CNTs as biomimetic

gecko foot-hairs.

The normal adhesion of single carbon nanotubes was examined by Maeno et al.

with gold and SiO2 surfaces using a manipulation technique with a transmission electron microscope [223]. It was found that the normal adhesion is proportional to the cross section of the CNT which indicates the contact geometry is flat-flat contact. The adhesion

of CNT-Au is higher than that of CNT- SiO2 which is consistent with the difference in

The Hamaker constant. The adhesion of a single CNT of 7.5 nm in diameter to SiO2

surface is 18 nN. Using 5 × 1010 tubes/cm2 as the density of CNTs, the normal adhesion

strength of the CNT array is ~ 900 N/cm2. However, the amazing adhesion strength does

not translate to macroscopic scale.

Zhao et al. measured the adhesion of macroscopic patches of VA-MWCNT array

as grown on silicon wafer in both perpendicular and parallel directions [153]. The

2 maximum adhesive strengths were measured to be 11.7 N/cm in the normal direction and

2 about 7.8 N/cm in the shear direction with glass surface. It appeared that the adhesion strengths with the glass surface tend to be higher than with other surfaces, such as gold, parylene, gallium arsenide, etc., under similar preloading conditions. It was speculated that the presence of hydroxyl groups on hydrophilic glass surfaces under ambient conditions may interact to some extent with the likely hydrogen terminated graphitic walls of the MWCNTs. The adhesion strength over repeated cycles decays due to the relatively poor adhesion of MWCNTs to their growth substrate. Significantly improved adhesion strength was achieved by adding molybdenum to the catalyst under layer to enhance the bonding of CNTs to the growth substrate.

Synthetic gecko tape based on hierarchical structured VA-MWCNT arrays was

fabricated by patterning catalyst on the growth substrate [154, 157]. In this design (Figure

2.26B and C), the nanometer sized CNTs make contact with the surface just like spatulae

do and the bundles of CNTs, 50~500 µm wide and 200 ~ 500 µm long, are similar to

gecko setae. Synthetic gecko tape was made by transferring micro-patterned carbon nanotube arrays onto flexible polymer backing tape. The gecko tape can support a shear

stress of 36 N/cm2, nearly four times higher than the gecko foot with a preload in the range of 25 ~ 50 N/cm2, and sticks to a variety of surfaces, including Teflon®. The peel- off and normal adhesion were much lower than the shear force, which satisfies the requirement of gecko adhesive. These carbon nanotube-based synthetic gecko adhesive tapes exhibit self-cleaning properties. They can be cleaned by water, as shown by the leaves of lotus and lady s mantle plants. In addition, the synthetic tapes can also be cleaned by a contact mechanism similar to that exhibited by geckos. After mechanical cleaning, the shear strength recovers up to 90% (and 60% for water-cleaned samples).

The detail of this work will be presented in Chapter III, IV.

Figure 2.26 SEM images of CNT-based gecko adhesives. (A) The VA-MWCNT array embedded in polymer matrix. (B and C) Hierarchical CNT structure mimicking gecko setae and spatulae. (C) Curly ended VA-MWCNT array. (D and E) VA-SWCNT- based adhesive. [152, 154-156]

Qu et al. was able to achieve an adhesion force of ~100 N/cm2 by using VA-

MWCNT arrays with curly ends as shown in Figure 2.26D, which is almost 10 times higher than that of live geckos, and a much stronger shear adhesion force than the normal adhesion force with as grown VA-MWCNT arrays [156]. However, the preload was 2 kg on a 4 × 4 mm2 area ( ~125 N/cm2)which was much higher than the preload used in the

work by Ge et al. Maeno et al. found that a broad distribution of the number of walls

(layers) in the nanotubes favors adhesion and a 44.5 N/cm2 shear strength after applying a

47 N/cm2 preload was reported for a VA-MWCNT array transferred to flexible tape [158].

VA-SWCNT arrays synthesized by a combined plasma-enhanced chemical vapor

deposition and fast heating method as shown in Figure 2.26E and F resulted in maximum

macroscopic adhesive forces of 29 N/cm2 in the normal direction and 16 N/cm2.in the

shear direction [155]. Unlike other CNT-based adhesives, the adhesion strength in the

normal direction was much higher than shear. Adhesion of CNT-based gecko adhesives

is summarized below in Table 2.6.

Table 2.6 Comparison of CNT-based synthetic gecko adhesives

Shear Normal Preload Contact Materials Hierarchy Backing Ref N/cm2 N/cm2 N/cm2 surface

MWCNT 7.8- 11.7 >500 Glass 1 Si [153]

MWCNT 36 ~3 25~50 Mica 1-2 tape [154]

SWCNT 29 16 125 Glass 1 Si [155]

MWCNT 100 19 125 Glass 1 Si [156]

MWCNT 44.5 - 47 Glass 1 PP tape [158]

Although the adhesion of CNTs measured at nanometer scale dose not translate to

macroscopic tape, in general, CNT-based gecko adhesives have higher adhesion strength

than polymer-based adhesives due to outstanding structural properties. CNT-based gecko

adhesives can be made into large areas and the adhesion force scales with the area. The

misunderstanding that exists in review literature that the patterned area of CNT arrays is

usually small ( 1.6 mm2) was potentially a typo or mathematical error. However, a major concern of the∼ CNT-based adhesives is it requires a high preload (25~125 N/cm2)

compared to that of polymer-based adhesives ( < 0.5 N/cm2).

2.4.3 Applications of Gecko Inspired Adhesives

Most features of the geckos’ adhesive toe pads, such as directional adhesion and self-cleaning, have been achieved by synthetic gecko inspired adhesives. These adhesives have been used in wall-climbing robots and as conductive, reusable and self-cleaning adhesives. Because intermolecular interactions are passive mechanisms, gecko adhesives are expected to perform well in vacuum and over a large range of temperature, which make synthetic gecko adhesive suitable for outer space application.

Although none of the gecko adhesives have been commercialized yet, many prototype robots have been demonstrated. A model car tire wrapped with a microfiber array [96] was demonstrated by Ron Fearing’s lab [224]. The high friction force enable the 1/18 scale model car with gecko tires to climb on acrylic sheet at about 60° angle.

This gecko tire only works on smooth surfaces. Stickybot, which resembles the

locomotion of geckos was developed at Stanford University [225]. Stickybot can climb

smooth vertical surfaces such as glass, plastic, and ceramic tile at 4 cm/s. Hierarchical,

compliant structures with directional adhesion were also implemented on robots.

Geckobot and several other types of wall-climbing robots were developed at Metin Sitti’s

lab [226]. Jeong et al. demonstrated transporting LCD glass panel using gecko inspired adhesive [137]. Some of these examples are shown in Figure 2.27.

Figure 2.27 Examples of wall-climbing robots. (A) A model car tire wrapped with microfiber array. (B) Stickybot, a bioinspired robot capable of climbing smooth surfaces. (C) Geckobot, a motor actuated robot inspired by the gecko lizard. [224-226]

CNT-based gecko adhesives have not been used in wall-climbing robots because a much higher preload is usually required. CNT-based gecko adhesives were

demonstrated to form conductive joints [152, 153]. This useful function could be used to

replace soldering in the fabrication of electrical circuits and to enable fast assembly and

disassembly, which could reduce cost in manufacturing as well as recycling. CNT-based

gecko adhesives adhered in vacuum [227] and sustained higher temperature while

maintaining adhesion strength [155] as demonstrated by Qu et al. It suggests useful application in outer space, which requires adhesives to perform in vacuum and sustain unusual temperature fluctuations.

CHAPTER III

EXPERIMENTAL

3.1 Fabrication of CNT-based Synthetic Gecko Adhesives

CNT-based synthetic gecko adhesives were fabricated in the following steps:

1. Electron-beam deposition of catalyst on silicon wafer for CNT growth. To

achieve hierarchically structured synthetic setae, this layer was deposited by

photolithography methods.

2. Growth of VA-CNT array in a tube furnace at desired temperature.

3. Transferring VA-CNT array onto flexible or rigid backings.

The adhesion of CNT-based synthetic gecko adhesives was then measured in both shear, normal and peel geometries at macroscopic and microscopic scales.

3.1.1 Fabrication of VA-CNT Array

The vertically aligned carbon nanotube array was fabricated by chemical vapor deposition at 750 °C on silicon (Si) wafer with a 500 ~ 1000 nm of silicon dioxide (SiO2) top layer using iron as catalyst.

The Si wafers were obtained from University Wafers. The wafer box must be opened and kept in clean room to avoid contamination, so no additional cleaning is needed. If they are found spoiled, the wafers need to be rinsed with acetone, methanol,

and isopropyl alcohol and then blow dried with nitrogen (N2) gas. The polished side of the wafer was used for carbon nanotube growth. A 10 nm thick aluminum (Al) buffer layer was first deposited followed by a 1.5 nm thick iron catalyst layer (which forms nano-sized particles for catalytic growth of carbon nanotubes) on top of the Al layer.

Figure 3.1 shows the chemical vapor deposition (CVD) setup for carbon nanotube growth. It consists of a tube furnace with an alumina or quartz processing tube (inner diameter 45 mm). The Si wafer with catalyst layer was positioned in the middle of the heating zone and then the tube was sealed. Air in the system was purged with Ar/H2 gas mixture (15% H2) with a flow rate of 1300 sccm. The Ar/H2 mixture was also used to carry the vaporized carbon source during the CVD process. The carbon source, ethylene

(with a flow rate of 50-150 sccm) was introduced when the temperature at the heating zone reached 750 °C. A very low concentration of water vapor with dew point of 20 °C was carried to the reaction tube by a fraction of Ar/H2 flow during carbon nanotube growth. The length of carbon nanotubes ranges from 200 µm to 500 µm which is

Figure 3.1 Chemical vapor deposition setup for CNT growth.

controlled by varying the reaction time between 10 and 30 minutes. The alignment of carbon nanotubes produced by this CVD process can be seen under scanning electron microscope (Figure 3.2A). Diameter of the carbon nanotubes was determined by transmission electron microscope (Figure 3.2B). From high-resolution TEM (inset of

Figure 3.2B) images, it can be seen that carbon nanotubes consist of 2-5 walls. The average diameter of the carbon nanotubes was ~ 8 nm, determined from the histogram in

Figure 3.2C.

Figure 3.2 Characterization of thin-walled CNTs. (A) Side view SEM image showing the alignment of carbon nanotube. (B) TEM image showing the diameter of carbon nanotube and high-resolution TEM (inset) image indicates that carbon nanotubes consist of 2-5 walls. (C) The average diameter of the carbon nanotube is ~ 8 nm.

3.1.2 Fabrication of Patterned VA-CNT Arrays

The experimental process for the growth of patterned vertically aligned carbon nanotube arrays involves one extra step: photolithography. A mask was designed with

AutoCAD software and then sent to mask printing service for fabrication. Then a photoresist layer was coated on a SiO2/Si wafer using spin coater. On a mask aligner, the photoresist layer was cured by exposing to UV light through the mask. After developing,

the uncured photoresist was removed and wafer went through the same electron beam deposition process to deposit a layer of catalyst. The cured photoresist was removed and catalyst layer replicated the patterns on the mask. After CVD process, arrays of VA-CNT pillars were fabricated. The images of gecko foot, and SEM micrographs of gecko setae array and the VA-CNT setae are shown in Figure 3.3.

Figure 3.3 Images of VA-CNT-based synthetic setae and spatulae. (A) Optical picture of gecko foot. (B) SEM image of gecko setae. (C and D) Side views (C) and higher- magnification SEM image (D) of the 100-μm setae. (E–H) SEM images of synthetic setae of width 50 (E), 100 (F), 250 (G), and 500 (H) μm.

3.2 Characterizations of Adhesion Properties

Carbon nanotubes have been chosen to construct the synthetic setae and spatulae because recent atomic force microscopy measurements using a silicon probe revealed that vertically aligned carbon nanotubes have strong adhesion with a silicon AFM tip [152].

But large variation in the performance at macroscopic scale is still not well understood.

The adhesion properties of CNT-based synthetic gecko adhesive were investigated at single seta level by using individual bundles of CNT. The synthetic setae were available in width ranging from 50 to 500 µm (Figure 3.3E-H). These measurements were conducted in shear and normal geometries.

3.2.1 Sample Preparation

To address the mechanism by which carbon nanotube hairs are brought into contact with a substrate, a combination of mechanical, electrical measurements and in- situ scanning electron microscopy is used to visualize the contact geometry at the interface between large numbers of VA-CNTs carbon nanotubes and to measure their mechanical properties and adhesion forces on smooth glass or silicon substrates. Single synthetic seta was prepared by transferring patterned VA-CNT array from Si wafer to a 3 mm × 3 mm glass slide with the following steps:

1. A thin layer of freshly mixed Dow Corning Sylgard 184 resin (PDMS) was coated

on glass slide previous treated by plasma cleaner.

2. The PDMS resin was allowed to cure until it was tacky.

3. Then, the tacky PDMS layer on the glass slide was pressed slightly against the

VA-CNT top surface without deforming the vertically aligned carbon nanotube

layer underneath and then was placed in an oven heated to 100 ºC for about 1

hour until the resin cured completely.

4. After curing, the glass slide was peeled off from the wafer with the VA-CNT

pillars transferred onto the glass. Due to the small size of the carbon nanotube

bundles, it was difficult to transfer only one bundle of carbon nanotube.

5. Individual bundles were removed manually with a sharp razor blade under a

stereoscope until only one bundle was left. Only 500 µm × 500 µm CNT setae

were used because it is difficult to isolate a single seta with smaller size.

6. The samples were then mounted on MTS Nano Bionix® system or the friction

cell (Figure 3.4) to measure adhesion and friction responses.

3.2.2 Stress and Strain Measurement

Measurements of mechanical properties of the vertically aligned carbon

nanotubes were performed using MTS Nano Bionix® system with Windows®-based

TestWorks® software. The MTS Nano Bionix® system can measure force of up 500 mN

with 50 nN load resolution and extension or compression of up to 15 mm with 35nm

displacement resolution. The compression stress-strain curves of the single CNT bundle

were obtained by compressing and then retracting the sample. The measurements were

conducted in both parallel and perpendicular directions to the alignment of nanotubes.

Due to the asymmetric nature of the nanotube bundle, significantly different responses

are expected.

The adhesion and friction measurements were also performed using a home-built

friction and adhesion testing device (Figure 3.4) that can apply and measure normal and

shear force. It consists of a sample holder with two degrees of freedom (x and y). There are two pairs of low friction linear bearings perpendicular to each other at the bottom of

the platform to ensure smooth linear motion in x and y direction. In each direction, the

holder is driven by a New Focus picomotor with a resolution of ~30 nm/step. The exact

calibration of the picomoters is done by moving the motor 10,000 steps and measuring the distance travelled using a microscope. There are two sets of steel double cantilever

beams perpendicular to each other which isolate the deflection in x and y direction. The deflection of the cantilever beams is measured by two plate capacitors. The capacitor

Figure 3.4 Schematic of friction and adhesion cell.

has one plate mounted on the cantilever and the other on a stationary post. A triangle

wave AC current is applied to the capacitors. Deflection of the cantilever beam changes

the distance between the two plates of the capacitor resulting in a change in capacitance.

The corresponding changes of voltage on the plate capacitors are measured.

The force-voltage relationship of the adhesion and friction cell is calibrated by

hanging known weights and recording the corresponding changes in voltage of the

capacitors. Two independent capacitors are used to measure the shear and normal force.

The spring constants of the cantilever beams are measured by bringing the surfaces into

contact and then measuring the force as a function of displacement. The calibration

curves are linear in the range of forces used in these experiments.

In the case of shear measurements, the same samples were used in multiple

measurements for low preloads. However, for higher preloads the shear stresses are high

enough to permanently deform the VA-CNT structure during sliding and those arrays

were only used once. The shear data for larger preloads were collected using larger

patches of VA-CNT array (4 mm × 4 mm size).

3.2.3 Electron Microscope Imaging

To image CNT array under normal and shear load, it is critical to visualize the behavior of CNTs when they are brought into contact with a surface. A miniature version of the adhesion and friction cell (Figure 3.5) was custom-built for semi in situ SEM

imaging. It allows side-view observation of the VA-CNT array while it is being compressed or sheared. This cell consists of two flat sample holders which sit on two linear bearings perpendicular to each other. With the holders moving along perpendicular directions, they apply and hold normal and shear load to the sample. The motion is driven manually with two micro-actuators. Due to the complication of implementing electric

motor in SEM chamber, the normal and shear loads were applied outside the chamber in

semi in situ fashion.

To obtain high resolution images when VA-CNTs are brought into contact with a silicon surface, charging must be minimized. Because there is a layer of non-conductive oxide layer on silicon wafer, a thin layer of silver was sputtered on it to prevent charging.

The VA-CNT array (3 mm × 3 mm) was transferred onto smooth glass surface using

double-sided conductive carbon tape and mounted on the vertical sample holder. The free

end was brought into contact with the silicon wafer sputtered with a thin layer of silver.

Figure 3.5 Friction and adhesion cell for SEM imaging.

Top view SEM images were obtained before and after making contact. Average

side contact length was obtained by averaging the lengths of 30 randomly selected

nanotubes from three SEM images taken at random locations of the top surface. The number density was obtained by counting the number of tubes in a box of 2 µm wide and

the same length as the average tube length. The number density before contact was obtained by counting the tube ends in a 2 μm × 2 μm box.

3.2.4 Electrical Contact Resistance Measurement

Intimate contact area or true contact area is a very important parameter in the studies of adhesion and friction phenomena. The intimate contact area can be measured using optical microscopic tools with a transparent substrate. Due to the dimensions of

CNTs, the contact area of CNTs is not accessible to any optical microscopic tools.

Although electron microscopy such as scanning electron microscope (SEM) and transmission electron microscope are widely used in the characterizations of CNTs, the electron beam is blocked by the substrate and the contact area is still not accessible.

Instead of direct imaging of the contact area, the electrical resistance of the contact interface was measured simultaneously with the mechanical stress-strain measurements. It is assumed that resistance of the contact interface is inversely proportional to the contact area, because contact resistance is dominant in the total resistance of single CNT [228].The stress-strain measurements were done using a MTS

Nano Bionix® instrument. As shown in Figure 3.6, the setup was modified to measure the electrical resistance during the loading and unloading cycle and shear force measurement.

A 500 µm × 500 µm VA-CNT sample was transferred on conductive indium tin oxide

(ITO) glass substrate using silver-filled epoxy. The electrical resistance is measured using a Keithley 2700 digital multimeter when the VA-CNTs are brought into contact with another piece of ITO electrode. The total resistance is assumed to be the sum of resistance of the measuring circuit (Rcir), resistance of carbon nanotube bundle (RCNT),

and resistance of the contact interface (Rint). The resistance at strain values of 80-90%

(maximum achieved contact area) was taken as minimum resistance of the system (Rmin ≈

95 Ω). The resistance (Rcir) for the direct ITO-ITO contact was 45 Ω. The resistance

(RCNT) of the CNT bundle is less than Ω9 measured by fixing both ends on copper

electrode using silver filled epoxy. The true contact area A is proportional to (Rmin-Rcir-

RCNT)/(R-Rcir-RCNT).

Figure 3.6 Schematic of contact resistance measurement.

3.3 Synthetic Gecko Tape

The natural gecko foot (Figure 2.1 and Figure 3.3B) has 5-μm-diameter setae arranged in many rows along the footpad, and the SEM image in Figure 3.3 shows 100- to 200-nm-diameter spatulae. The synthetic setae and spatulae fabricated using vertically

aligned carbon nanotubes are shown in the SEM images in Figure 3.3E-H. Figure 3.3C

and D show the picture of carbon nanotubes on a flexible gecko tape to illustrate the similarity in the natural and synthetic hairs. Different sizes of the carbon nanotube

patches ranging from 50 to 500 µm in width (Figure 3.3E-H) have also been fabricated.

The synthetic setae consist of thousands of synthetic spatulae which are aligned

individual carbon nanotubes of average diameter of 8 nm (shown at higher magnification

in Figure 3.3D). The as grown VA-CNT array usually has some contamination from amorphous carbon which adversely affects adhesion of the nanotubes. The amorphous carbon can be oxidized using air plasma. Because the rate of oxidation for amorphous carbon is much higher than that for carbon nanotubes, amorphous carbon are burnt off in just a few minutes. The synthetic gecko adhesive tape was then made by transfer the array of VA-CNT onto a single-side adhesive tape by a slight pressing and then peeling

off from the wafer.

The adhesion of synthetic gecko tapes was measured using shear, normal and peel

geometries. To measure the macroscopic adhesion forces, small areas (usually a 4 mm ×

4 mm sample) of synthetic gecko tape specimens were pressed against a smooth substrate

using a cylindrical roller. In controlled preload experiments, desired weights were placed

on top of rubber block with a 4 mm × 4 mm bottom surface which was used to ensure

uniform distribution of the preload.

In shear geometry, the specimens were pulled in parallel to the mating surface

which is similar to lap shear test of an adhesive joint. The maximum shear load when the

specimens were pulled off from the surface was recorded as shear strength.

3.3.1 Fabrication of Synthetic Gecko Tape

The test specimen as shown in Figure 3.7 was a strip of single sided adhesive

tape of 4 mm wide and 20 mm long, sandwiched between two pieces of plastic sheet of

the same width. There is a 4 mm × 4 mm area covered with pressure sensitive adhesive

on the far left end of the stripe. The arrays of carbon nanotubes were transferred on to the

4 mm × 4 mm area. In a shear test, the load is applied at the right end of the strip.

Several types of mating surfaces were chosen ranging from very hydrophilic to

hydrophobic. Hydrophilic surfaces include mica, glass. Hydrophobic surfaces include

poly(methyl methacrylate), poly(octadecyl acrylate) and Teflon. The results from these

surfaces can be used to verify molecular mechanism. Mica was frequently used because a

clean surface is prepared simply by lifting off the old layer using Scotch® tape.

Figure 3.7 Specimen of synthetic gecko tape for macroscopic force measurement.

3.3.2 Substrate Preparation

The mating surfaces were prepared as following:

Mica: It has a highly-perfect basal cleavage yielding remarkably-thin sheets.

Freshly cleaved mica surface was prepared by lifting off a thin layer of mica sheet with an adhesive tape.

Glass: glass microscope slides were cleaned in a potassium hydroxide/isopropyl solution for at least 30 minutes, followed by rinsing with deionized water. The final

washing was done in water containing a small amount of surfactant, which coats the glass

with a monolayer of surfactant and prevents any further condensation of impurities from

the atmosphere. The glass substrate was plasma-treated for 5 min in oxygen plasma to

remove the surfactant layer before the adhesion measurements.

Poly(methyl methacrylate) and poly(octadecyl acrylate): These polymer films were prepared by spin coating a 100- to 200-nm-thick layer on a silicon substrate. The samples were annealed above the glass transition temperature for poly(methyl methacrylate) and above the melting temperature for poly(octadecyl acrylate) polymer under vacuum.

Teflon: Teflon sample was cleaned for 6 h in a mixture of Nochromix and sulfuric acid. The Teflon sample was rinsed with deionized water and blow dried with nitrogen.

3.3.3 Pure Shear Test

Pure shear tests were conducted as shown in Figure 3.8. A normal preload equivalent to a pressure of 25–50 N/cm2 was applied by a cylindrical roller to deform the

gecko tape and the carbon nanotube structures to achieve good contact between the tape

and the substrates. The specimen was then pulled with a pure shear force at 0.1 N

increments until it was pulled off from the substrate. The maximum load before it was

pulled off was recorded. This geometry has been chosen to compare these synthetic

materials with the force measurements on live geckos that were done by using an almost

vertical geometry (Figure 3.8) by Irschick et al. This is equivalent to dragging the gecko

almost parallel to the surface.

The shear adhesion strength was also measured for Scotch® tape. A series of increasing shear load were applied to patches of 4 mm × 4 mm of Scotch® tape

specimens adhering to a clean glass substrate. Because the adhesive on Scotch® tape is viscoelastic, the specimens were pulled off from the glass plate after a certain amount of time. The time required to pull off the specimen were recorded. As a comparison, a 4 mm

× 4 mm patch of synthetic gecko tape was also tested with a 3.2 N shear load.

Figure 3.8 Shear force measurement of synthetic gecko tape. (A) Geometry used in shear force measurements (B) live gecko clinging force measurement on almost vertical surface.

3.3.4 Peel Test

The force required to peel the gecko tapes at different angles was measured. The setup is shown in Figure 3.9. This peeling measurement is important because geckos detach their footpad from a surface by uncurling its toes. This uncurling action is very similar to peeling a tape off surface. It is necessary because, at those peeling angles, the

peeling forces are much smaller than shear force so that geckos can move effortlessly on a vertical surface by using its foot-hairs repeatedly without any damage. In mimicking the

gecko feet, peeling resistance has to be much smaller in comparison to shear forces.

3.4 Live Gecko Experiment

The adhesion of two gecko species (Phelsuma dubia and Gekko gecko) was

measured with variation in temperature and humidity.

3.4.1 Live Gecko Preparation

10 Phelsuma dubia (Dull Day Gecko) and 16 Gekko gecko (Tokay Gecko) were obtained from California Zoological Supply. Their sex, bodyweight and toe pad area are listed in Table 3.1.

Phelsuma dubia is a diurnal gecko found in Madagascar. These specimens weighed between 4.7–7.3 g. Gekko gecko, is a nocturnal gecko found in Southeast Asia.

These specimens weighed between 54–80 g. All specimens were housed in individual 10 gallon glass tanks with paper towel substrate, maintained within a dedicated animal

research facility at the University of Akron. Each tank was misted with water as well as

provided with a fresh bowl of water daily. G. gecko were fed vitamin dusted crickets 5

times a week whereas the P. dubia were fed vitamin dusted crickets 3 times a week and

the other 2 days they received a fruit supplement. The room was maintained at 25 ± 1°C

and set to a 12 hour photoperiod with UVA/UVB full spectrum lights. Heating tape along

the underside of each tank allowed geckos to thermo-regulate within the range of body temperatures typical for free-ranging geckos (25–35 °C).

3.4.2 Temperature Trials

A walk in environmental chamber was used. It was maintained to within ± 1 °C of five randomly presented experimental temperatures, 12, 17, 22, 27, and 32 °C. Each set of trials was initiated approximately two hours into the species-specific diel-cycle (9am for P. dubia and 9pm for G. gecko). Geckos were placed inside the environmental chamber from 1-2 hours prior to all experimental procedures to allow body temperature equilibration. No more than a single temperature was tested on any given day. As shown in Figure 3.10, a specially designed apparatus was used to hold a glass substrate and a motor driven force gauge (Shimpo FGV-10X) which moved relative to the glass substrate.

Digital output from the force gauge was collected by a LabVIEW® program which can save traces of each adhesion trial. For all trials, values with the substrate in the vertical position were reported. Each gecko was fitted with specially designed harnesses positioned around the pelvis and allowing unimpeded movement of legs and minimizing

Table 3.1 Sex, size, and toe pad area of geckos

Gecko Sex Weight (g) Toe Pad Area ( cm2)

Day 1 F 5.3 0.64 Day 2 F 5.6 0.72 Day 3 M 5.0 0.66 Day 5 M 4.7 0.65 Day 6 F 5.5 0.73 Day 7 F 7.3 0.84 Day 8 F 5.5 0.72 Day 9 F 6.9 0.79 Day 12 M 6.2 1.00 Day 14 M 3.9 0.59 Tokay 1 M 68.8 4.65 Tokay 2 M 59.3 4.95 Tokay 3 M 59.1 4.87 Tokay 4 F 73.2 5.46 Tokay 5 M 54.3 3.93 Tokay 6 M 78.2 5.31 Tokay 7 F 69.2 4.49 Tokay 8 M 79.9 5.27 Tokay 9 F 73.3 4.91 Tokay 10 M 47.8 4.90 Tokay 11 F 42.8 4.06 Tokay 12 F 44.9 4.32 Tokay 14 F 46.4 4.47 Tokay 17 M 43.8 4.57 Tokay 18 M 50.0 5.55 Tokay 19 F 39.2 4.15

Figure 3.10 Apparatus measuring shear force of live geckos. Geckos could be pulled at a constant rate selectable over a wide range of values. All pulls were accomplished with the substrate in a vertical orientation. Maximum force was the highest value recorded between the start of a pull and the point at which all four feet began to slide on the substrate.

abrasion to the geckos’ skin. At the start of each test, the harness was attached to the force gauge which was subsequently driven downward by the motor. Geckos were introduced to the substrate and induced to take step with each foot, providing a standardized posture with all four feet and toe pads fully in contact with the substrate.

Each individual was then pulled parallel to the surface in a smooth motion until all four feet began to slide indicating the end of the test (referred to hereafter as a ‘pull’). Geckos were pulled three times in succession to estimate maximum clinging force. Between each gecko, the glass was cleaned with ethanol and dried before the next set of pulls.

3.4.3 Humidity Trials

Subsequent to completing all the temperature trials, concerns have been raised regarding to the variation in the relative humidity of the environmental chamber at different temperatures. It might be a contributing factor to the variation in estimates of maximum clinging ability. Consequently, a second set of trials was initiated to estimate the effect of humidity at the temperature where recorded clinging forces were maximal for both species (12 °C). Humidity trials followed the protocol described above for temperature trials, except that temperature was maintained constant but relative humidity was maintained at one of four levels, presented in a random order (30, 55, 70, and 80%).

Humidity trials at 12 °C were followed by two trials at 32 °C (at 35 and 80% RH) to test whether temperature and humidity interact. At the completion of every trial, geckos were weighed to the nearest 0.1 g. Before every trial and after the very last trial, the feet of each gecko were scanned with a flat bed scanner so the total adhesive area of the toe pads could be estimated.

3.4.4 Statistical Analyses

Statistical analyses include data from several separate experiments, each with slightly different sample sizes and overall designs. Whenever an analysis involved data from geckos exposed to all combinations of experiment factors (i.e., humidity and temperature), a univariate repeated measures ANOVA was used to test for the statistical significance of effects in the model (assumption of sphericity was never violated).

Clinging force is reported in two forms, raw and “adjusted,” facilitating comparisons using both standardized and raw values. Adjusted clinging forces are the residuals from a linear regression of raw clinging force on toe pad area, providing a measure of clinging force that accounts for variation among individual lizards in toe pad size. Throughout the manuscript, whenever clinging forces are presented as ‘adjusted’ they are standardized as described above; values without the ‘adjusted’ descriptor represent the raw values.

Finally, when means are reported in the text or tables they are accompanied by their

respective standard errors, unless otherwise noted.

100

CHAPTER VI

RESULTS AND DISCUSSION

4.1 Adhesion of VA-CNTs

The “contact splitting principles”, adhesion design maps for fibrillar adhesives,

and many modeling works of gecko-inspired adhesive system suggest that a highly dense array of carbon nanotubes could make a superior synthetic gecko adhesive. If the rough correlation between bodyweight and hair density is extrapolated (Figure 2.10), a density

of ~ 103 hairs/µm2 is required to support the bodyweight of human. The density of VA-

CNT array varies from 1010 ~ 1011 tubes/cm2 which is much higher than polymer hairs.

Other advantages of carbon nanotubes over polymeric systems are their small diameter,

high aspect ratio and mechanical strength. The diameter of multi-walled carbon

nanotubes ranges from 8 ~ 40 nm, and the diameter of single-walled carbon nanotubes is around 1 nm. Individual carbon nanotubes have elastic modulus up to 1 TPa. The modulus can be significantly lower due to defects. At this length scale, the contact between carbon nanotubes and the substrate is considered to be molecular which enables the universal attractive van der Waals interactions. It is almost impossible for polymer systems to achieve the same diameter, aspect ratio and mechanical strength as carbon nanotubes with existing technology.

The adhesion strength of VA-CNT array with diameters ranging from 20–30 nm

and an aerial density of 1010 ~ 1011 tubes/cm2 is estimated to be more than ~ 500 N/cm2

101

by JKR theory if all of the CNTs make end contact [153]. This is 50 times greater than

live gecko. A recent experiment with single CNT showed that the adhesion of CNTs with

7.5 nm diameter to SiO2 surface is 18 nN if the contact is in flat-flat geometry. This force,

if normalized by the cross section area of the tube, is ~ 41000 N/cm2. The normal

adhesion of a multiwall nanotube array with a density of 5.0 × 1010 tubes/cm2 is

estimated to be approximately 900 N/cm2 if all of the CNTs make flat-flat contact [223].

The interaction of carbon nanotubes, either in the form of entangled layer [229] or

vertically aligned array [152], with AFM tip have been measured independently. Both studies confirmed that the adhesion between carbon nanotubes and AFM tip is significant and in the case of vertically aligned array, the adhesion is ~ 1600 N/cm2 which is

normalized by the area of the AFM tip. This value is almost 160 times more than that of

live Tokay gecko. A 1-cm2 synthetic gecko tape should be able to support the bodyweight

of a typical adult. Counter intuitively, the initial attempts to make macroscopic synthetic

gecko adhesive patches from the same carbon nanotubes were not very successful despite

the very exciting numbers from theoretical calculation and AFM measurements of

individual nanotubes. The adhesive forces in both shear and normal direction were very

low or even not measureable. One good example of poor adhesion performance of

macroscopic patch of vertically aligned carbon nanotube array is the work of Cao et

al.[230] These arrays were compressed up to 1000 cycles with very high stress and up to

85% strain. There were no indications of adhesion which is usually represented by a dip

in stress-strain curve. It seems to be a paradox that individual nanotubes adhere strongly

but a macroscopic patch of vertically aligned carbon nanotube array showed non-

measureable adhesion force. This paradox is not unique to carbon nanotubes. Many

102

polymeric arrays also have this problem. Why does “contact splitting principle” fail to extend to synthetic hairs? It is really a hurdle to the fabrication of useful synthetic gecko adhesives.

After a careful re-investigation of the “contact splitting principle” and scaling mechanical models, it is found that the each and every contact elements are assumed to bear the same load, so that the total adhesive force of a hairy pad is the product of the force of a single contact element and the number of elements. This model is built on the end contact geometry in which only the ends of hairs make contact with the surface. To achieve good adhesion, end contact model requires all the contact elements to make molecular contacts simultaneously which is not realistic. Due to the differences in the length hairs, it is almost impossible to ensure an even distribution of load among all contact elements and molecular contact of all the elements with the surface because for shorter elements to make molecular contacts, longer elements must deform and the elastic energy stored tends to separate the shorter one from a flat substrate. The separating force can be very high for a high modulus material such as carbon nanotubes. Furthermore, on a rough surface, it is more difficult to match the profile of the hairy array with the surface profile to ensure molecular contact everywhere. These cases are similar to our daily experiences that two flat substrates do not stick to each other spontaneously. The above factors contributed to a much lower number of elements actually in molecular contact with the surface, resulting in very low adhesion of the vertically aligned carbon nanotube array. In AFM measurement, this problem can be avoided, because the diameter of the

AFM tip is in the same range of diameter of single carbon nanotubes and the tip could make contact with a single nanotube. However, the contact geometry is not known in

103

AFM measurements. It is possible for individual carbon nanotubes to make molecular

side contact with the AFM tip. And the true contact area can be much larger than the end contact geometry. It becomes crucial to find out the contact mode and correlation between adhesion and properties of carbon nanotube arrays.

4.1.1 Visualization of Contact Geometry of VA-CNTs

To address the mechanism by which the carbon nanotube hairs are brought into

contact with the substrate, a combination of mechanical, electrical measurements and in-

situ scanning electron microscopy is used to visualize the contact geometry at the

interface between large numbers of VA-CNTs carbon nanotubes and to measure the their mechanical properties and adhesion forces on smooth glass or silicon substrates.

The changes taking place during compression (parallel to the alignment of CNT) loading-unloading cycle of VA-CNT array were visualized by acquiring side view SEM images of the nanotube array before making contact, during and after compression with the SEM friction cell as illustrated in Figure 3.5. The average diameter of CNTs is 8 nm.

Although in situ force and displacement measurements are preferred and theoretically possible by inclusion of sensors in the friction cell, they were not conducted due to the complications of modifying the vacuum chamber of SEM. Figure 4.1A shows the SEM image of a 500 µm × 500 µm CΝΤ pillar prior to making contact. The nanotube hairs deform under pressure, but the deformation recovers if the strains are less than 5-10%. At higher strains, the structure deforms significantly in response to higher stress (Figure

4.1B) and it does not recover after removing normal load (Figure 4.1C). In some of the

104

pillars, the formation of kinks was observed which did not recover upon unloading

(Figure 4.1C).

Figure 4.1 Side view SEM images of VA-CNT arrays. (A) prior to compression, (B) under compression, and (C) after compression. After removing the normal load, the height of the VA-CNT does not recover.

The deformation of individual nanotubes was visualized with higher magnification SEM images as shown in Figure 4.2. Carbon nanotubes are not necessarily straight and their intrinsic undulations can be observed as shown in Figure 4.2. Before making contact, the ends of carbon nanotubes are relatively well aligned and the

105

difference in length is obvious. When under compression load, these carbon nanotube

ends bend and their side walls make contact with the substrate as shown Figure 4.2B. If compressed with high strains, the bending deformation of carbon nanotubes does not recover after removing normal load as shown in Figure 4.2C. This irreversible deformation of the VA-CNTs is probably due to the mutual adhesion of VA-CNTs which is analogous to that in the capillary induced collapse of VA-CNT. While thin VA-CNTs

(8 nm diameter) deform irreversibly, much thicker VA-CNTs (40 nm diameter) are much stiffer and do recover on unloading, because of the much higher bending energy relative to mutual adhesion.

Figure 4.2 High resolution side view SEM images of VA-CNT array. (A) prior to compression, (B) under compression, and (C) after compression.

106

4.1.2 Stress-strain Curves of VA-CNT Array

The adhesion of VA-CNT array with average diameter of 8 nm was measured using the MTS Nano Bionix® tester in compression mode in contact with glass substrates

at displacement rate of 1 µm/s. The compression stress-strain result of a VA-CNT pillar

of 500 µm × 500 µm in size and 280 µm in height is shown in Figure 4.3 where the

compression and retraction directions are parallel to the carbon nanotube alignment. The stress-strain curve is highly nonlinear and has three distinct regions during compression and more importantly an adhesive jump-out during retraction. In the first regime, at low strains (less than 5%), the modulus of the VA-CNT is very low but variable (0.3 ± 0.2

MPa). This large variability in the apparent modulus for small strains is due to the increase in density of the CNT away from the ends, and due to the experimental difficulty of aligning the two surfaces to make perfect parallel contact. Both of these effects cause all the nanotubes to fully engage with the substrate only when the strain is of the order of

5%, leading to an increase in the apparent modulus. In the second more distinct region

(strain of 5-20%) all the carbon nanotubes are engaged and the modulus is 1.6 ± 0.3 MPa.

In this regime, irreversible changes were observed near the interface where the nanotube

hair undergoes bending and buckling. In the third regime, beyond a critical stress in the

order of 0.2 MPa, a significant change in slope was observed and the modulus drops to

0.25 ± 0.1 MPa. In this regime, the material deforms with a very small increase in stress

due to buckling and bending of the nanotube hairs. The retraction cycle has significant

hysteresis, an adhesive jump-out and the height of the nanotube brush is reduced

permanently.

107

Since these nanotubes are aligned, it is expected that the compression modulus of to be anisotropic. The compression stress-strain result of a VA-CNT pillar of 500 µm ×

500 µm in size and 280 µm in height is shown in Figure 4.4 where the compression and

retraction are perpendicular to the carbon nanotube alignment. When the compressive

force is perpendicular to the alignment direction of the VA-CNTs, the modulus is 90 ± 20

kPa which is much small than the parallel direction. There is also a sudden jump-out

Figure 4.3 Normal compression modulus of vertically aligned multi-walled carbon nanotube array. Stress-strain measurements of a 280 µm tall VA-CNT array parallel to the orientation of the VA-CNT.

108

during retraction cycle as a result of loss of contact between the sides of carbon nanotubes and the surface (glass). This indicates an important role of side contact in adhesion. Furthermore, the stress increases monotonically with deformation and the structure does not recover after unloading, which indicates that adhesion between the carbon nanotubes prevents their recovery.

Figure 4.4 Lateral compression modulus of vertically aligned multi-walled carbon nanotube array. Stress-strain measurements of a 280 µm tall VA-CNT array perpendicular to the orientation of the VA-CNT.

109

The differences in adhesion, stress-strain curve and periodic buckling of the carbon nanotubes in this study and the previous study are striking. The fundamental differences are the average diameter and the number density of the carbon nanotubes. In the previous study, the density of carbon nanotubes is 1 × 1010 carbon nanotubes/cm2,

average diameter is 40 nm and the compression modulus of the array is 50 MPa. The

carbon nanotubes used in this study is much thinner and have average diameter of 8 nm,

density of 7 × 1010 carbon nanotubes/cm2 (measured based on side images), and a

compression modulus of 1.6 ± 0.4 MPa. After normalizing with the number of the density

of carbon nanotubes, the 40 nm and 8 nm diameter individual carbon nanotubes have

average compression modulus of 500 nN/nanotube and 1-2 nN/nanotube, respectively.

The bending rigidity of hollow tubes is = ( )/64, where E is the Young's 4 4 표 푖 modulus, I is the area moment of inertia,퐸퐼 do 퐸휋is the푑 outer− 푑 diameter, and di is the inner

diameter of the tube. The 40nm tubes will be a factor of ≈ 300 times stiffer than the 8 nm diameter tubes, assuming the thickness of the carbon nanotubes to be 3-5 nm. The

experimental result of this ratio from the compression modulus is approximately 250-500.

It is consistent with the theoretical prediction for hollow cylinders.

4.1.3 Constrained Buckling of VA-CNT Array

When the array was compressed, an interesting periodic buckling with a wavelength of ~ 2.5 µm occurred at a critical stress of ~ 0.2 MPa and buckling did not recover after the probe was removed as shown in Figure 4.5. The periodicity of the buckled carbon nanotubes is much larger than their intrinsic undulations. A new model

110

other than Euler buckling is needed to explain this periodic buckling because the buckling wavelength is not a function of the total length of carbon nanotubes.

Figure 4.5 High resolution SEM image of periodic buckling of VA-CNT array. The periodicity of the buckling is much larger than the intrinsic undulations.

From the observations of these SEM images, it is clear that the buckling of these carbon nanotube array is very different from what was reported previously by Cao et al. in which the carbon nanotube array unbuckled and recovered its height upon load release.[230] The mechanical properties of this type of carbon nanotube array are expected to be dramatically different. The critical stress for the 8 nm diameter carbon nanotubes ( ≈ 0.2 MPa) is a factor of approximately 50 times lower than that of 40 nm

diameter carbon nanotubes ( ≈ 12 MPa). Previous attempts to explain the behavior of

111

these arrays rely on the Euler buckling model (Figure 4.6A) for isolated fibers, according to which the buckling stress scales as / [230]. Using E ≈ 1 TPa (an ideal 2 2 11 2푐 upper limit for CNT), N ≈ 10 carbon nanotubes/cm≈ 휋 퐸퐼푁 퐿 , and length of VA-CNTs ≈ 300 µm,

a critical stress of 20 Pa is obtained. This estimate based on the Euler buckling model is

many orders of magnitude smaller than the measured critical stress of 0.2 MPa (Figure

4.3). In addition, these buckling stresses are not a function of the height of the VA-CNTs.

This is because carbon nanotubes are closely packed, so one must account for the

interactions between the VA-CNTs. Indeed, the fibrillar interface behaves more like an orthotropic material. To understand the response of individual tubes, it is assumed that

each one of them is surrounded by a soft material of other VA-CNTs with an effective

modulus G. The lateral compressibility of the VA-CNT array restricts long wavelength

buckling that would occur for isolated nanotubes. The bending stiffness of VA-CNT

prohibits short wavelengths as illustrated in Figure 4.6B. The balance of these two

opposing effects leads to periodic buckling with a wavelength predicted by constrained

buckling equation:

= 2 1 휅 4 휆 휋 � � where κ is the bending rigidity and α is propo훼 rtional to G. For a homogenous,

incompressible, isotropic media, α is given by:

4 = ln( ) 휋퐺 훼 where l is a characteristic length scale of the 푙buckling⁄ 푎 ( the buckling wavelength which

was observed to be on the order a few micrometers for VA-CNT array) and a is a

112

microscopic length of order for the tube radius, which is ~4 nm for CNTs here [231]. For

8 nm diameter CNTs,

4 1.95 ln(625) 휋퐺 훼 ≈ ≈ 퐺 The critical force for single tubes is:

= / 2 2 푐 This equation is similar to that for Euler푓 buckling,휋 휅 휆 except that the length L is replaced by

buckling wavelength λ.

Figure 4.6 Euler and constrained buckling models. (A) Euler buckling. (B) Constrained buckling model where a stiff rod of modulus E is surrounded by a matrix with a modulus G. A compression force initiates buckling. However, a single buckle is inhibited by the surrounding VA-CNTs and this leads to periodic buckling that is not correlated with the height of the VA-CNT array.

Using lateral compression modulus G ≈ 90 ± 20 kPa from measurement (Figure

4.3), buckling wavelength λ ≈ 1.2 µm is predicted by the constrained buckling equation.

The critical force for individual tubes is ~ 0.3 nN and the critical stress for a VA-CNT

113

array having a density of 7 × 1010 carbon nanotubes/cm2 is ~ 0.2MPa. These predications

are very consistent with the values observed in the experiments (λ ≈ 2.5 µm, Figure 4.5

and critical stress of ~ 0.2 MPa in Figure 4.3 ).

4.1.4 Adhesion Analysis of Side Contact Geometry

Further evidence for the role of collective behavior of the VA-CNT in controlling

adhesion and friction is observed in the top view SEM images collected before (Figure

4.7A) and after the shear experiments (Figure 4.7B). Initially, before shear, the carbon nanotubes are mostly straight and form small adhesive clumps. After being brought in contact perpendicular to the surface and shear measurement, significant alignment of the

VA-CNTs with the substrate and a concomitant increase in the number of contacts of carbon nanotubes with the surface occur. Visually, the matted appearance before contact turns to a glossy, shiny appearance after contact which is expected for a transition to a smooth aligned surface. The side view of the nanotube brush in contact with and after shear against a silicon wafer confirms that not only the ends but also the buckles are oriented along the direction of shear.

From the SEM images in Figure 4.2B and Figure 4.7B, it is observed that the increase in contact area is due to side contacts of the carbon nanotubes because they do not recover after making contact. The analysis of the top view SEM images showed that the average length of side contact is around 1.7 ± 0.5 µm. In a simple side-wall contact as illustrated in Figure 4.8, the attractive interactions in the adhesive contact region hold the tube onto the surface, while the elastic energy stored in the bent region tends to separate the tube from the surface. The adhesion energy is proportional to the length of side

114

Figure 4.7 High resolution top view scanning electron micrographs of VA-CNT array. (A) prior to compression and (B) after compression.

contact and the elastic energy is a function of bending rigidity of the tube. If the adhesion

energy is greater than the elastic energy, then the tube can adhere to the surface with a

preload. In the contact region, the van der Waals interaction free energy per unit length w

between carbon nanotube and flat surface is

= 1 , 2 24퐴푑 푤 3 2 where A is the Hamaker constant, l is contact length,퐷 d is the diameter of the CNT and D

is the separation between CNT and substrate. Using typical values of A = 6 × 10−20 J and

-20 -20 D0 = 0.34 nm, for d = 8 nm, w = 3.6 × 10 J/nm and for d = 40 nm, w = 8.0 × 10 J/nm.

115

The van der Waals interaction free energy does not increase much when diameter

increases 5-fold while bending rigidity of CNT:

( ) = 644 4 퐸휋 푑표 − 푑푖 퐸퐼 varies several orders of magnitude. The characteristic length at which adhesion energy

and elastic energy is balanced can be estimated by:

1. 2 퐸퐼 푙 ≈ � � Using EI ≈ 5 × 10-23 N·m2 and w = 3.6 × 10-20푤 J/nm for CNTs having 8 nm diameter and

2 nm wall thickness, a characteristic length of approximately 1.2 µm is obtained. This

value is consistent with the average side contact length of 1.7 ± 0.5 µm observed from

SEM image in Figure 4.7B. Because of the intrinsic undulations of CNTs (Figure 4.2), side-wall contact is not uniform and extra contact length is needed to compensate adhesion. It is reasonable that the observed length is greater than the predicted value.

Figure 4.8 Side-wall contact model of CNT. The CNT modeled as an elastic rod. When compressed, it bends and makes side-wall contact with the surface. The elastic energy stored in the bent tube works against adhesion

116

Interestingly, using EI ≈ 1.6 × 10-20 N·m2 and w = 8 × 10-20 J/nm for CNTs having

40 nm diameter and 5 nm wall thickness, the average characteristic length is around 14

µm. Similarly, for CNTs having 20 nm diameter and 5 nm wall thickness, EI ≈ 2.0 × 10-21

N·m2 and w = 5.6 × 10-20 J/nm for 40 nm diameter CNTs, the average characteristic length is around 6 µm. This explains the difficulty of 20 nm and 40 nm diameter CNT which are much stiffer tubes in achieving adhesion.

4.1.5 Contact Area

It is difficult to determine the change in contact area during a loading-unloading cycle, because the interface is not accessible to microscopy tools. Instead of direct measurement of contact area, electrical resistance was measured in conjunction with stress-strain measurements, because the contact resistance dominates the overall resistance of carbon nanotube placed on two electrodes [228]. Using a simple model of n identical CNTs in a vertical array, the resistance of single CNT is the sum of contact region and non-contact region = , + , . These CNTs are

푖 푖 푛표푛−푐표푛푡푎푐푡 푖 푐표푛푡푎푐푡 connected in parallel, so 푅 푅 푅

, , = + = + , 푅푖 푛표푛−푐표푛푡푎푐푡 푅푖 푐표푛푡푎푐푡 푅푡표푡푎푙 푅푛표푛−푐표푛푡푎푐푡 푅푐표푛푡푎푐푡 푛 푛 where Rcontact is reciprocal to contact area. The setup that is used to measure electrical resistance is shown in Figure 3.6. The raw data of changes in resistance as a function of strain are shown in Figure 4.9A. Figure 4.9B shows the stress-strain curve during compression along with the relative contact area. The reference contact area (resistance)

117

Figure 4.9 Change in contact area in a normal pull-off test. (A) Change in resistance when VA-CNT array is compressed (black) and retracted after bringing the array in contact with ITO glass. (B) Stress (black, loading; red, unloading) and relative contact area (green, loading; blue, unloading) as a function of strain.

118

is obtained by compressing the sample really hard to nearly 90% strain. Interestingly, the

relative contact area increases monotonically as the sample is compressed and follows the

trend of compression stress. Increase in both number of contacts and length of contact could contribute to the increase in contact area. During the retraction cycle, the contact area and the mechanical behavior show significant hysteresis and relative contact area at zero load on retraction curve is about 55% and at least 50% contact area remained just before jump-out. These results confirmed the prediction of side contact analysis. For compliant carbon nanotube array, significant contact area is made by applying a preload.

The adhesion of VA-CNT is expected to vary with preload based on the change in contact area observed here.

The changes in relative contact area were also measured when shear load was applied and the results are shown in Figure 4.10. Figure 4.10A shows that significant

contact area was made after applying a preload of 2.8 × 105 Pa. As shown in Figure

4.10B, the shear force increases uniformly with displacement until the brush slips

intermittently via a characteristic stick-slip motion. The relative contact area decreased

gradually when shear load was increased and it dropped significantly from ~75% at the

onset of the first slip to ~20% after the slip. The CNTs gripped the substrate repeatedly

after each slip and the change in contact area followed every stick-slip events. The 75%

contact area at the onset of the first slip also confirmed the prediction of side contact

analysis. For compliant carbon nanotube array, significant contact area is made by

applying a preload to bearing a shear load. The shear adhesion strength of VA-CNT is

also expected to vary with preload based on the change in contact area observed here.

119

Figure 4.10 Change in contact area in a shear pull-off test. (A) Change in relative contact area when a preload is applied to VA-CNT array. (B) Change in relative contact area when a shear load is increased.

120

4.1.6 Preload Dependence of Adhesion Strength

In addition to increasing the effective stiffness, the cooperative bending and

buckling deformation of the VA-CNTs also affect the adhesion and friction properties of

the brush. When a brush is brought in and out of contact over a typical load-unload cycle,

the response with a preload greater than critical stress is shown in Figure 4.3 and the

response with a preload less than critical stress is shown in Figure 4.11A. In both cases,

there is a sudden jump-out during retraction, indicating loss of adhesive contact (Figure

4.3 and Figure 4.11A). Figure 4.11B shows the normalized pull-off forces as a function

of preload. The normal pull-off strength increases linearly with increase in preload. This

correlation is quite unlike the response of simple adhesive elastic contacts, where the

pull-off force is independent of preload based on contact mechanics models. In Figure

4.2B, the SEM image of the side wall contact of the carbon nanotubes under preload

shows significant bending and buckling. The increase in adhesive contact area at the

interface explains the increase in adhesive forces. The increase in contact area is also

confirmed in contact resistance measurement.

The combination of anisotropy in the mechanical response and the adhesion of the

soft compliant VA-CNT brush also lead to an unusual shear response. A preload was

applied as previously described and then the sample was retracted until the normal load is

nearly zero. These measurements are unlike the traditional friction measurements where

the shear forces are measured as a function of normal load. The shear measurements at

minimal normal load are sensitive to the adhesive forces and hysteresis of the contact

after the preload is applied on the sample. Figure 4.12A shows that the shear force increases uniformly with displacement until the brush slips intermittently via a

121

Figure 4.11 Preload dependence of normal pull-off strength. (A) A typical cycle for adhesion measurements that shows jump-out indicating adhesive contact (≈ 300 µm in height). (B) Normal pull-off force/area scales linearly with preload (units of pressure).

122

Figure 4.12 Preload dependence of shear pull-off strength. (A) A typical shear profile shows the static friction force builds up until it reaches maximum static shear force. The normal load is reduced to nearly zero after applying normal preload for 5 min. Stick-slip happens in the dynamic friction region. (B) The critical shear stress when the sample starts to slide is plotted as a function of preload. The results for smaller preloads (< 50 × 103 Pa) were done with 500 µm× 500 µm size arrays (and 300 µm in height) using the friction cell. The shear stress for larger preloads (> 50 × 103 Pa) was measured using 4 mm × 4 mm (VA-CNT bundles with 500 µm× 500 µm in size and 100 µm in height) samples.

123

characteristic stick-slip motion. In Figure 4.12B, the maximum nominal shear stress is plotted as a function of applied preload. The shear stress increases with the preload suggesting that this is directly correlated with the adhesion of the hairy array.

Interestingly, the shear forces are also 4 times higher than the normal pull-off measurements, and this anisotropy between the shear and adhesion forces is an intrinsic property of fibrillar structures.

The results from contact resistance measurement can explain why both normal and shear adhesion strengths increase with preload. The contact area for these compliant nanohairs is a function of applied load; and because of intrinsic adhesion of compliant carbon nanotubes, the contact area shows significant hysteresis. These correlations between preload and contact area make the adhesion truly controllable. The normal and shear strength can be controlled simply be varying preload.

4.2 Synthetic Gecko Tape

The contact mode and correlation between mechanical properties and the adhesion of CNTs have been answer by microscopic analysis. Fabrication of large area of adhesive patches is an important step in making it useful for applications.

4.2.1 Unpatterned Synthetic Gecko Tape

Unpatterned VA-CNT array was transfer to flexible backing as described in

Chapter 3. To measure the macroscopic adhesion forces, small areas of flexible gecko tapes were pressed against a smooth mica sheet using a cylindrical roller. The actual

124

shear measurements were done under no external normal load. This geometry has been

chosen to compare these synthetic materials with the force measurements on live geckos

that were done by using an almost vertical geometry by Irschick et al.[177] This is equivalent to dragging the gecko almost parallel to the surface. The shear force results for the unpatterned gecko tape of 4 mm × 4 mm in size are shown in Figure 4.13. The forces supported by the synthetic structures are comparable to the live gecko measurements.

These shear forces supported by the live geckos is calculated in the units of force supported by a 0.16-cm2 area (10 N/cm2 × 0.16 cm2). However, the force supported by

0.12-cm2 and 0.25-cm2 patch was 1.2 N and 1 N, respectively. And the shear force does

not increase with the area of the tape. This is a disadvantage, because the weight

supported by the gecko tape cannot scale up by increasing the contact area.

Figure 4.13 Shear pull-off force for unpatterned synthetic gecko tapes on surfaces with various hydrophobicity.

125

Similar to the natural gecko, these synthetic gecko tapes stick to both hydrophobic

and hydrophilic surfaces. Figure 4.13 shows the shear forces supported by the unpatterned gecko tape (0.16 cm2 area) on hydrophilic surfaces such as mica and glass

(water wets both surfaces). It is also shown in Figure 4.13 that the shear stress of a

partially hydrophobic surface which is coated with polymethylmethacrylate that has a

water contact angle of 70–80° and a very hydrophobic surface which is coated with

poly(octadecyl acrylate) comb polymer that has a water contact angle of 110°. The shear

strength is similar for both hydrophobic and hydrophilic surfaces, and this finding

supports the idea that van der Waals forces play an important role in the shear mechanism.

Surprisingly, the measurements on Teflon® surfaces also show large shear strength that is

comparable to that obtained from hydrophilic surfaces, while Tokay Gecko had difficulty

adhering to Teflon®.[47, 180].

4.2.2 Patterned Synthetic Gecko Tape

In order to make synthetic gecko tape that can support larger shear forces,

patterned VA-CNT arrays were designed and fabricated. Discontinuity can enhance

adhesion by distributing stress more uniformly and over a larger area.[211] To test this

idea, a series of 2 × 2, 3 × 3…8 × 8 arrays of 500 µm × 500 µm VA-CNT bundles are

fabricated and shear forces are measured. The results are shown in Figure 4.14. The shear

force scales up with the area of the patches (number of pillars). This result indicates that

micrometer-size textures are required in addition to nanometer-size features of carbon

nanotubes. The role of the micro-sized textures is similar to setae in the hierarchical

structure of gecko foot-hairs which are about 5 µm in diameter. Figure 4.15 shows the

126

measured shear forces for a series of patterned VA-CNT arrays (texture sizes shown in

Figure 3.3D–H) of 0.16 cm2 tape area. Using patterned surfaces with features of 50 – 500

µm wide, a factor of four to seven times higher shear force was obtained as compared

with the unpatterned surfaces of similar area. The shear force supported by the 100- to

500-μm patches is 3.7 N which is two to three times higher than the live Tokay geckos.

The advantages of patterns became less prominent on reducing the patch size to 50 μm

(and 300 μm in height) because a decrease in the ratio of the width to its height makes setae mechanically weak. When a smaller height of the 50-μm setae (200 μm) is used, the

shear force is 5.8 N, a factor of four times higher than live Tokay geckos.

6 ~34 N/cm2 5 ~0.085 N/pillar

2 Shear Load (N) 1

0 0 10 20 30 40 50 60 70

Number of CNT pillars

Figure 4.14 Shear pull-off force of patterned synthetic gecko tapes vs. area. For patterned VA-CNT array, shear force increases proportionally with the size of the patch. And shear adhesion strength is much higher than the non-textured VA-CNT

127

To understand why these carbon nanotubes support such large forces, a series of

optical and SEM images were taken during shear loading and after the cracks have propagated through the sample. In Figure 4.16A, four optical images of the edge of the tape under increasing load are shown. The black regions are the carbon nanotubes sticking out of the edges and making contact with the substrate. On application of the shear load, one can observe the increase in the width of the carbon nanotube edge showing that the tape and the carbon nanotubes are deformed and that the crack propagation is pinned at the interface. The SEM picture of the edge after pressing the tape in contact with the substrate is shown in Figure 4.16B. The stretching of the tape increases with an increase in load, and after reaching a certain critical load a catastrophic rupture is initiated; the failure is cohesive, and carbon nanotube residues are left behind on the mica surface as shown in Figure 4.17A.

Figure 4.15 Shear pull-off force for unpatterned and hierarchically patterned synthetic gecko tapes.

128

No carbon nanotube residue is left behind on the mica substrate when the tapes

are peeled at different angles (in a peeling geometry rather than shear geometry), similar

to how a gecko walks, and the results are discussed later. This cohesive failure indicates

that the interfacial adhesion strength is sufficient to support large shear forces. The 50 to

500 μm patterned surfaces are important in supporting larger shear stress and increasing

the total shear force supported by the tape by increasing the contact area (similar shear

adhesion strength values were obtained for gecko tapes of 0.16 cm2 and 0.25 cm2 area).

Figure 4.16 Images of CNTs bearing shear load. (A) Optical images, (B) SEM micrograph

The SEM image of the carbon nanotube residues remaining on the mica surface is

shown in Figure 4.17B. The aligned and broken strands of the carbon nanotube bundles indicate that there is large energy dissipation in the cohesive failure which supports the side contact model. If the length of the side contact reaches a critical value, it is possible

that the shear adhesion exceeds the tensile strength of the nanotube resulting in rupture of

tubes and cohesive failure. In the case of unpatterned gecko tapes, the failure was

129

interfacial with very little carbon nanotube remaining on the mica surface after peeling.

Interestingly, live gecko hairs seem to be wear free even after being pulled along the

surface.[51]

Figure 4.17 Carbon nanotube residues on mica surface. (A) Optical image. (B) SEM micrograph.

The optical and SEM images show that the 50 to 500 μm patches deform and hinder the crack growth. In fracture mechanics, it has been demonstrated that the resistance to the propagation of cracks is very important in increasing the toughness of the materials. In the case of adhesive tapes, Kendall has shown that the peeling strength of tapes with patches of different stiffness or thickness can be much higher than the tapes with uniform thickness and stiffness. Chaudhury and coworkers have shown that 100 to

200 μm patches on poly(dimethylsiloxane) sheets increases the peeling force by a factor of 10–20 in comparison to a unpatterned poly(dimethylsiloxane) sheet.[211] The role of pattern in all of these cases, including the patterned gecko tape, is to stop, deviate, and reinitiate the crack propagation in comparison to materials with uniform properties (also

130

referred to as the Cook–Gordon mechanism). It is interesting to note that the

enhancement in shear stress in the case of natural gecko, poly(dimethylsiloxane) patterns,

and synthetic gecko tapes are observed for sizes of patterns that are comparable to the

thickness or height of the setae.

Currently, there are several arguments that support the idea of adhesion playing

an important role in shear strength. Firstly, the shear experiments were done under

negligible normal load (just the weight of the tape,≈ 20 mg). Secondly, when the ta pe breaks under load, the crack front (observed by video camera) propagates at speeds >10 m/s, which are expected for adhesive stick to sliding (or breaking) transition. Thirdly, a high pull-off forces were observed using silicon AFM tip to make contact with carbon

nanotube bundles. Finally, normal pull-off loads that increase with preload were observed

for a large range of preload, which has to result from adhesion rather than friction.

4.2.3 Easy Peeling of Synthetic Gecko Tape

The forces required to peel the gecko tapes at different angles were measured and

the results are shown in the Figure 4.18. This peeling measurement is important for two

reasons. First, the peeling resistance has to be much smaller in comparison to shear forces

in mimicking the gecko feet. The gecko moves or gets unstuck from a surface by

uncurling its toes. This is necessary because, at those peeling angles, the peeling forces

are weak and the gecko can effortlessly move on a vertical surface by using its foot-hairs

repeatedly without any damage. Secondly, the peeling force divided by the width of the

tape (F/w) provides an approach to determine the energy of detachment (G). The 500 μm

patterned synthetic tape peels off the mica substrate with an adhesive force/width of only

131

16 N/m at a 45° angle, 20 N/m at a 30° angle, and 96 N/m at a 10° angle. This peeling process at angles >10° does not involve any breaking or transferring of carbon nanotubes on the substrate, and the synthetic gecko tape can be reused many times without damage.

The low peeling forces at a non-zero angle are due to the preexisting crack at the peeling front. A similar effect is also observed if you peel off the viscoelastic tapes from substrates, where the shear forces are much higher than the peeling forces at non-zero angles.

Figure 4.18 Peel force of synthetic gecko tapes at 45° angles.

132

4.2.4 Robustness of Synthetic Gecko Tape

Interestingly, the shear forces supported by the gecko tape are very stable and

time independent as shown in Figure 4.19. The weight supported by the patterned gecko tape is stable for many days (it has been measured for a period of 15 days). In contrast, a

Scotch® adhesive tape shows strong time-dependence as expected for viscoelastic

materials. The viscoelastic tape is initially stronger than the synthetic gecko tape but

quickly loses its adhesive strength and it is outperformed after only a few minutes. If the

gecko foot-hairs were made of viscoelastic adhesives, then the geckos would required

much stronger muscles to move rapidly up the wall and could not remain stationary for

prolonged periods of time.

Figure 4.19 Comparison of viscoelastic tape and synthetic gecko tape.

133

The robustness of VA-CNT-based gecko adhesive tape was tested by repeating stick-unstick cycles. The results of 17 cycles are shown in Figure 4.20. The strength of the tape did not show degradation over repeated uses.

Figure 4.20 Robustness of VA-CNT-based synthetic gecko tape ( 4 mm × 4mm).

4.2.5 Synthetic Gecko Tape Works in Vacuum

Base on the mechanism of van der Waals interactions, the VA-CNT-based gecko tapes are expected to stick in vacuum environment as well as in the air. The adhesion was also tested qualitatively in a vacuum chamber as illustrated in Figure 4.21. The result that it works in vacuum is also an evidence supporting van der Waals forces.

134

Figure 4.21 VA-CNT-based synthetic gecko tape tested in vacuum chamber.

4.3 Mechanism of Adhesion in Geckos

Mimicking gecko foot-hairs with VA-CNT array has achieved significant success.

There are still a lot to be learned from live geckos. A preponderance of evidence suggests

that geckos stick to substrates as a consequence of the formation of a large number of

intimate setal-substrate contacts engaging van der Waals attraction [43, 45, 46]. One prediction about performance that can be derived from this mechanism is that adhesion should be temperature insensitive over a biologically meaningful range. However, the results of previous work [162, 163] have been equivocal on this matter. Such a simple prediction ignores important details of adhesion at the whole organism scale. For example, adhesion may involve more than just the sum of the setal-substrate interactions;

135

contributions of chemistry at the interface, elaborate subcutaneous vascular, muscular and

skeletal elements in adhesion and release are currently completely unexplored.

Presumably, these components would be subject to thermal dependencies typical of ectotherms. A series of experiments measured adhesion at different temperatures using a single protocol with two species (G. gecko and P. dubia). Although evidence of a strong

effect of temperature was found, the direction of the effect was counterintuitive given the

thermal biology of geckos and it violated the prediction given by van der Waals

interactions presumed responsible for adhesion. Consequently, other factors (e.g.,

humidity) that could explain the variation in the observed clinging performance were

examined. Evidence was found, unexpectedly, that humidity is likely an important

determinant of clinging force in the geckos. Implications of these results for

understanding both the factors affecting adhesion in geckos as well as for inferring

mechanisms that underlie such performance are explored below.

4.3.1 Temperature Effect

The body weight and toe pad area of geckos are listed in Table 3.1. Tokay geckos were larger (58.2 ± 6.8 g) and had greater total toe pad areas (4.74 ± 0.98 cm2) than Day

geckos (5.58 ± 0.32 g and 0.73 ± 0.06 cm2, respectively). The results of average adhesive

forces measured at 12, 17, 22, 27 and 32 °C are tabulated in Table 4.1 Tokays were also

able to generate higher maximal total clinging forces than Day geckos (31.2 ± 4.4 N and

5.5 ± 0.5 N, respectively, F1, 15 = 29.9, P < 0.0001), but maximal force per unit toe pad

area was similar between species (7.83 ± 1.72 N/cm2 and 6.34 ± 1.62 N/cm2, Day and

Tokay respectively; F1, 15 = 1.6, P = 0.23). Adjusted maximal clinging force was

136

significantly different across temperatures (F4, 56 = 8.09, P < 0.0001), and there was a

significant difference between species in the effect of temperature on clinging force

(species × temperature F4, 56 = 2.91, P = 0.03) leading to analyze the effect of temperature separately for each species. Although the effect of temperature was significant for both

Day and Tokay geckos (F4, 28 = 28.49, P < 0.0001 and F4, 28 = 5.05, P = 0.003,

respectively), the trend for an increase in clinging ability with decreasing temperature

appeared stronger for Day geckos (Figure 4.22). For example, the clinging ability of

Tokay geckos at the intermediate temperature (22 °C) was close to but not quite

significantly different from clinging ability at the lowest temperature (12 °C; matched

pairs t-Test, t = 2.23, P = 0.056). However clinging force at 12 °C was significantly

higher than at 22 °C for Day geckos (12 °C; matched pairs t-Test, t = 10.78, P < 0.0001).

Table 4.1 Average maximal adjusted adhesive force (N) by species and temperature

Temperature ( °C)

Species 12 17 22 27 32

Day 5.53 ± 0.51 2.11 ± 0.38 1.05 ± 0.40 0.71 ± 0.30 0.79 ± 0.19

Tokay 31.22 ± 4.39 14.37 ± 2.20 20.64 ± 2.87 20.19 ± 3.18 13.05 ± 2.96

4.3.2 Humidity Effect

At 12 °C, clinging ability increased significantly with humidity (Figure 4.23; F3,27

= 4.61, P = 0.01), such that forces measured at the highest humidity (80%) were nearly

twice as high as the forces observed at the lowest humidity (35%). Although there was a

137

trend suggesting the effect of humidity was stronger for Day geckos, there was no

significant difference between species in the rate of change in clinging force with

humidity (species × rh F3,27 = 2.28, P = 0.1). In order to determine if the humidity effect

observed at 12 °C varied with temperature, a set of trials were conducted to measure clinging ability at 32 °C at 35% and 80% RH. Surprisingly, at 32 °C, clinging ability did not vary in response to humidity as it did at 12 °C (Figure 4.23). Instead, the clinging ability of Tokay geckos was significantly higher at 35% RH compared to 80% RH (13.75

± 3.0 N vs. 2.48 ± 2.0 N; F1, 15 = 38.9, P < 0.0001), while the clinging ability of Day geckos did not vary significantly between 35% and 80% RH (0.375 ± 0.2 N vs. 0.327 ±

0.36 N; F1, 15 = 0.05, P = 0.8261). Moreover, at 32 °C clinging ability was among the

lowest measured in any set of trials.

Figure 4.22 Results of live gecko temperature trials. Humidity is not controlled in these trials. Relationship between temperature and body-size corrected adhesion (clinging force [N]; left axis, bars) or relative humidity (%; right axis, solid line).

138

Figure 4.23 Results of live gecko humidity trials. Adhesion at constant temperature of 12 °C and variable relative humidity (RH). Adhesion increased significantly with RH, but slopes were not significantly different. Red lines and symbols show results of 35% and 80% trials at 32 °C designed to test for an interaction of temperature and humidity.

139

4.3.3 Combined Effect of Temperature and Humidity

Restricting attention just to temperature effects (Table 4.1 and Figure 4.23), leads to the conclusion that adhesion is highly temperature sensitive in both species, with greatest adhesion observed at the lowest test temperature. However, RH varied by approximately 15% between the two temperature extremes (12 °C and 32 °C, Figure

4.23). It is important to note that relative humidity was not controlled in previous studies that examined temperature effects on adhesion in these two species, but differences in protocol are intriguing. Bergmann and Irschick [162], who found no evidence of temperature effect on clinging ability of P. dubia heated or cooled geckos in an environmental chamber and then performed clinging tests at room temperature and ambient humidity conditions of their laboratory. Alternatively, Losos [163], performed clinging tests on G. gecko inside a walk-in environmental chamber set to one of 9 different temperatures. As in this design, relative humidity in the latter study presumably varied with temperature, and a strong effect of temperature on clinging ability was demonstrated. Notably, adhesion in the Losos study was maximal at intermediate temperatures (which corresponded to the highest relative humidity). Consideration of both temperature and 12 °C humidity trials (Figure 4.24), in light of the two studies described above, suggests that clinging ability is sensitive to variation in humidity, not temperature. Unfortunately, the humidity response does not appear to be simple. Indeed, the response to variation in humidity at 12 °C is not constant but apparently itself is affected by temperature (Figure 4.23). In other words, clinging ability by geckos on a smooth hydrophilic surface like glass appears to be sensitive to both temperature and humidity.

140

Figure 4.24 Combined results of temperature and humidity trials. Adhesion during variable temperature, uncontrolled humidity trials (black solid lines), and variable relative humidity constant temperature trials (12 °C; red-dotted lines) showing convergence of maximal clinging force with increasing temperature and decreasing humidity.

141

Interestingly, the effect of humidity on gecko adhesion has never been formally investigated at the whole animal scale. However, two studies focusing on elucidating the mechanism underlying the setal-substrate interaction have demonstrated that adhesive force between a single spatula and substrate is humidity dependent (at room temperature)

[50, 159]. Scaling of the effect of humidity at the nano-scale shows an approximately 1.3- fold increase in adhesion between 35% and 80% RH [159], or a 3-fold difference between 0 and 70% RH [50]. This study shows that at the whole animal scale the effect of humidity is quite complex. An approximate doubling of adhesion over 35 to 80% RH at 12 °C has been observed. On the other hand, the adhesion forces are relatively insensitive to humidity at 35 °C. Notably, however, there is now an important gap in experiments at the setal scale: humidity manipulations have only been done at a single temperature.

The strong influence of humidity on the adhesive forces at a single setal and whole animal level also suggests that capillarity forces may play an important role in adhesion. A strong effect of humidity on adhesion forces has also been observed using atomic force microscopy for hydrophilic tips in contact with hydrophilic surfaces.

Alternatively, when one of the materials is hydrophobic, adhesion is humidity independent [232]. This study raises several paradoxes that cannot be resolved with existing data or theories. Firstly, the humidity dependent adhesion forces at 12 °C suggest that setae must be hydrophilic. However, gecko foot surface is superhydrophobic with a water contact angle of 160.9° and water droplets do not wet gecko foot and roll off easily upon tilting [46]. According to the Cassie-Baxter and Wenzel models, surface that exhibit superhydrophobic water contact angles (> 150°) is the result of chemistry and roughness.

142

Secondly, it is not clear why humidity has a strong influence on the adhesion forces only at low temperatures. A capillary model predicts that the adhesion forces at the same humidity should be proportional to absolute temperature [160]. Therefore, the adhesion forces at 35 °C should be only 8% higher than the adhesion forces at 12 °C. These predictions are much smaller than experimental observations. In addition, the capillary model predicts the adhesion forces at 35°C should show similar dependence on humidity.

Both the van der Waals and capillary forces fail to explain the shear adhesion data at the whole animal level. Resolution of this paradox will likely require examination of the particular way in which water interacts with substrate and seta at the nanometer scale. In addition, the adhesion force measurements at a single setal scale are needed at variable temperatures before eliminating the possibility of thermal biophysical constraints influencing the adhesion.

Irrespective of the mechanism, effects of humidity on otherwise “dry adhesive” biological systems may not be limited to gecko toe pads. The explanation involving rearrangement of keratin protein to expose more polar amino acid groups at the surface has also been discussed as a potential mechanism for increasing adhesion. For example, members of the spider infraorder Araneomorphae produce a derived type of silk which adheres to surfaces with greater force as humidity increases [233, 234]. It has been hypothesized that regularly spaced nodes in the silk strands are richer in hydrophilic polar or charged amino acids, which under humid conditions should promote hydroscopic interactions between the nodes and substrates [234] It is unclear whether similar mechanisms might underlie the humidity response of gecko toe pads. However, efforts to characterize the protein structure of setae have revealed that setae have a complex

143

structure of heterogeneous α- and β-keratins [175, 235, 236]. Furthermore, there is evidence that the lateral regions of the β-keratins sequenced in gecko setae are relatively rich in hydrophilic and polar amino acids that could modify adhesive interactions under moderate humidity conditions [237], perhaps in a way analogous to the increased stickiness of nodes in silk of some spiders in the Araneomorphae. Additional methods are needed to probe water at the contact interface and chemistry at the surface of gecko foot.

144

CHAPTER V

CONCLUSIONS

5.1 Adhesion of VA-CNTs

The “contact splitting principle”, adhesion design maps for fibrillar adhesives, and

many modeling works of hairy adhesives such as geckos’ suggest that VA-CNT array has advantages in density, diameter, aspect ratio, and mechanical robustness over polymeric system for making synthetic gecko adhesive. The density of VA-CNT array ranging from

1010 ~ 1011 tubes/cm2 is much higher than polymer hairs. This density is equivalent to

~103 hairs/µm2 which is dense enough to support the bodyweight of a human if the rough

correlation between bodyweight and hair density in Figure 2.10 is extrapolated. The adhesion strength of VA-CNT array with diameters of 20 ~ 30 nm and an aerial density of 1010 ~ 1011 tubes/cm2 is estimated to be more than ~ 500 N/cm2 by JKR theory if all of

the CNTs make end contact [153]. This is 50 times greater than live gecko. A recent

experiment with single CNT showed that the adhesion of CNTs with 7.5 nm diameter to

SiO2 surface is 18 nN and the contact is in flat-flat geometry. This force if normalized by

the cross section area of the tube is ~ 41,000 N/cm2. The normal adhesion of a multiwall

nanotube array with a density of 5.0 × 1010 tubes/cm2 is estimated to be approximately

900 N/cm2 if all of the CNTs make flat-flat contact [223]. The interaction of carbon

nanotubes, either in the form of entangled layer [229] or vertically aligned array [152]

with AFM tip, confirmed that the adhesion between carbon nanotubes and AFM tip is

145

significant. In the case of vertically aligned array, the adhesion is ~ 1600 N/cm2 if normalized by the area of the AFM tip. This value is almost 160 times more than that of live Tokay gecko. A 1-cm2 synthetic gecko tape should be able to support typical bodyweight of an adult. Counter intuitively, the initial attempts to make macroscopic synthetic gecko adhesive patches from the same carbon nanotubes were not very successful despite the very exciting numbers from theoretical calculation and AFM measurements of individual nanotubes. The adhesive forces in both shear and normal direction were very low or even not measureable.

The mechanism why CNTs behave so differently was not clear. To explore the correlation between mechanical properties of carbon nanotubes and their adhesion, vertically aligned arrays of small diameter carbon nanotubes (average diameter≈ 8 nm) have been fabricated by water-vapor assisted CVD process. Their mechanical properties and adhesion behavior were compared with thicker CNTs (40 nm). The differences between 8 nm CNTs in this study and the 40 nm CNTs in previous studies are striking.

To explore the mechanism of adhesion of carbon nanotube hairs, in-situ scanning electron microscopy is used to visualize the contact geometry at the interface between large numbers of VA-CNTs carbon nanotubes. A combination of mechanical and electrical methods is used to measure their mechanical properties and adhesion forces on smooth glass or silicon substrates.

A series of SEM images were acquired when the VA-CNT array was brought into contact with model surfaces. Before making contact, the ends of carbon nanotubes are relatively well aligned and the difference in length is obvious. Contrary to end contact or flat-flat contact, these carbon nanotube ends bend, buckle and make side-wall contact

146

with the substrate when under compression load. If compressed with high strains, the

deformation of carbon nanotubes does not recover even after removing the normal load.

This irreversible deformation of the VA-CNTs is probably due to the mutual adhesion of

VA-CNTs which is analogous to that in the capillary induced collapse of VA-CNT.

While thin VA-CNTs (8 nm diameter) deform irreversibly, much thicker VA-CNTs (40

nm diameter) are much stiffer and do recover on unloading, because of the much higher

bending stiffness relative to mutual adhesion. Similar to thicker VA-CNTs (40 nm

diameter), periodic buckling was also observed during compression, but the buckling

wavelength is much smaller (~ 2.5 µm).

Complete stress-strain curves were obtained for a VA-CNT array of 8 nm

diameter CNTs. Because VA-CNT array is anisotropic, the stress-strain curves were measured in two directions: parallel and perpendicular to the orientation of CNTs. When the compressive force is parallel to the carbon nanotube, the stress-strain curve is highly nonlinear and has three distinct regions during compression and more importantly there is an adhesive jump-out during retraction. In the first regime, at low strains (less than 5%), the modulus of the VA-CNT is very low but variable (0.3 ± 0.2 MPa). In the second more distinct region (strain of 5-20%) all the carbon nanotubes are engaged and the modulus is

1.6 ± 0.3 MPa. In this regime, irreversible changes were observed near the interface where the nanotube hair undergoes bending and buckling. In the third regime, beyond a critical stress in the order of 0.2 MPa, a significant change in slope was observed and the modulus drops to 0.25 ± 0.1 MPa. In this regime, the material deforms with a very small increase in stress due to buckling and bending of the nanotube hairs. The retraction cycle has significant hysteresis, an adhesive jump-out and the height of the nanotube brush is

147

reduced permanently. When the compressive force is perpendicular to the alignment

direction of the VA-CNTs, the modulus is 90 ± 20 kPa, which is much smaller than the

parallel direction. There is also a sudden jump-out during retraction cycle as the result of

loss of contact between the side-walls of carbon nanotubes and the surface (glass). This indicates an important role of side contact in adhesion. Furthermore, the stress increases monotonically with deformation and the structure does not recover after unloading, which indicates that adhesion between the carbon nanotubes prevents their recovery.

In the previous study, the density of carbon nanotubes is 1 × 1010 carbon

nanotubes/cm2, average diameter is 40 nm and the compression modulus of the array is

50 MPa. The carbon nanotubes used in this study are much thinner and have average

diameter of 8 nm, density of 7 × 1010 carbon nanotubes/cm2 (measured based on side

images), and a compression modulus of 1.6 ± 0.4 MPa. After normalizing with the

number of the density of carbon nanotubes, the 40nm tubes is a factor of 250-500 times

stiffer than the 8 nm diameter tubes which is consistent with the theoretical prediction for

hollow tubes.

The critical buckling stress observed for the 8 nm diameter carbon nanotubes is

~0.2 MPa. It is a factor of approximately 50 times lower than that of 40 nm diameter

carbon nanotubes ( ~12 MPa). According to Euler buckling model for isolated fibers, the

critical stress is approximately 20 Pa, which is many orders of magnitude smaller than the

measured critical stress of 0.2 MPa. In addition, these buckling stresses are not a function

of the height of the VA-CNTs. This is because carbon nanotubes are closely packed, so

one must account for the interactions between the VA-CNTs. The fibrillar interface

behaves more like an orthotropic material. The constrained buckling model considers that

148

each carbon nanotube is surrounded by a soft material of other VA-CNTs with an

effective modulus G. The lateral compressibility of the VA-CNT restricts long wavelength buckling and the bending stiffness of VA-CNT prohibits short wavelength.

The balance of these two opposing effects leads to periodic buckling with a wavelength that is the function of bending rigidity and lateral modulus. The bucking wavelength (λ ~

1.2 µm) is predicted by the constrained buckling equation. The critical buckling force of individual tubes fc ~ 0.3 nN is predicted by an equation similar to that for Euler buckling,

except that the length (L) is replaced by buckling wavelength (λ). The critical buckling

stress for a VA-CNT array having a density of 7 × 1010 carbon nanotubes/cm2 is ~

0.2MPa. These predictions are very consistent with the values observed in the experiments (λ ≈ 2.5 µm, Figure 4.5 and critical stress of ~ 0.2 MPa in Figure 4.3).

The cooperative bending and buckling deformation of VA-CNTs also affects the adhesion and friction properties of the array. When the array is brought in and out of contact over a typical cycle of loading and unloading, a sudden pull-out is observed indicating loss of adhesive contact. Interestingly, the normalized pull-off force increases linearly with an increase in preload. This correlation is quite unlike the response of simple adhesive elastic contacts, where the pull-off force is independent of preload based on contact mechanics models.

The VA-CNT brush also shows unusual shear response. Shear adhesion strength was measured with increasing preload. Unlike traditional friction measurements, the shear measurements at minimal normal load are sensitive to the adhesive forces and hysteresis of the contact after the preload is applied on the sample. The shear stress increases with the preload suggesting that this is directly correlated with the adhesion of

149

the CNT array. Interestingly, the shear forces are also a factor of 4 times higher than the

normal pull-off measurements, and this anisotropy between the shear and adhesion forces

is an intrinsic property of fibrillar structures. The SEM images of the side wall of the

CNTs show significant buckling, increased adhesive contact area at the interface after

applying a preload, and contact area change during load-unload cycle; all of which can

explain the increase in adhesive forces.

Further evidence for the role of collective behavior of the VA-CNT in controlling

adhesion and friction is observed in the top view SEM images collected before and after the shear experiments. Initially, before shear, the carbon nanotubes are mostly straight and form small adhesive clumps. After being brought in contact perpendicular to the

surface and shear measurement, significant alignment of the VA-CNTs with the substrate

and a concomitant increase in the number of contacts of carbon nanotubes with the

surface occur. Visually, the matted appearance before contact turns to a glossy, shiny

appearance after contact which is expected for a transition to a smooth aligned surface.

The side view of the nanotube brush in contact with and after shear against a silicon

wafer confirms that not only the ends but also the buckles are oriented along the direction

of shear. By analysis of the top view SEM images, the average length of side contact is

around 1.7 ± 0.5 µm. In a simple side-wall contact, the attractive interactions in the

adhesive contact region hold the tube onto the surface, while the elastic energy stored in

the bent region tends to separate the tube from the surface. The adhesion energy is

proportional to the length of side contact and the elastic energy is a function of bending

rigidity of the tube. The balance of these two opposing effects determines adhesion of the

tube. If the adhesion energy is greater than the elastic energy, then the tube can adhere to

150

the surface with a preload. The bending rigidity scales with d4 while adhesion energy

scales with d0.5. The characteristic length at which adhesion energy and elastic energy is

balanced for 8 nm CNT is estimated to be approximately 1.2 µm. This is reasonably close

to the average side-wall contact length of 1.7 ± 0.5 µm observed from SEM images.

Because of the intrinsic undulations of CNTs, side-wall contact is not uniform and extra contact length is needed to compensate adhesion. It is reasonable that the observed length is greater than the estimated value. Interestingly, the average characteristic length for a 40 nm carbon nanotube is around 14 µm, as this explains the difficulty of thick tubes in achieving high adhesion with much stiffer tubes.

To determine the change in contact area during a loading-unloading cycle, electrical resistance of contact was measured in conjunction with stress-strain measurements. The contact resistance Rcontact is reciprocal to contact area. The relative

contact area increases monotonically as the sample is compressed and follows the trend

of compression stress. During the retraction cycle, the contact area and the mechanical

behavior show significant hysteresis and relative contact area at zero load on retraction is

about 55%. The 50% contact area remained just before jump-out explains why both

normal and shear adhesion strengths increase with preload. The contact area for these

compliant nanohairs is a function of applied load, and because of intrinsic adhesion of

compliant carbon nanotubes, the contact area shows significant hysteresis. These

correlations between preload and contact area make the adhesion truly controllable. Thus

normal and shear strength can be controlled simply be varying preload.

To fabricate large synthetic gecko adhesive patches, unpatterned VA-CNT array was transferred to flexible backing and shear adhesion force was measured. The forces

151

supported by the synthetic structures are comparable to the live Tokay geckos. However,

the shear force does not increase with the area of the tape. This is a disadvantage, because

the weight supported by the gecko tape cannot scale up by increasing the contact area.

Similar to the natural gecko, these synthetic gecko tapes stick to both hydrophobic and

hydrophilic surfaces and strength is similar for both hydrophobic and hydrophilic

surfaces. This result supports the idea that van der Waals forces play an important role in

the shear mechanism. Surprisingly, the measurements on Teflon® surfaces also show

large shear strength that is comparable to that obtained from hydrophilic surfaces, while

Tokay gecko had difficulty adhering to Teflon®.

In order to make synthetic gecko tape that can support larger shear forces,

patterned VA-CNT arrays were designed and fabricated. Patterned surface can enhance

adhesion by distributing stress more uniformly and over a larger area. Indeed, the shear

force scales up with the area of the patches when VA-CNT is patterned and the maximum

shear force is up to 4 times higher than live Tokay gecko. This result indicates that

micrometer-size textures are required in addition to nanometer-size features of carbon nanotubes. The role of the micro-sized textures is similar to setae in the hierarchical structure of gecko foot-hairs.

Optical and SEM images show that the micrometer-size textures deform and hinder the crack growth. After reaching certain critical load, a catastrophic rupture is initiated; the failure is cohesive and carbon nanotube residues are left behind on the mica surface. The aligned and broken strands of the carbon nanotube bundles indicate that there is large energy dissipation in the cohesive failure which also supports the side contact model. If the length of the side contact reaches a critical value, it is possible that

152

the shear adhesion exceeds tensile strength of the nanotube resulting in rupture of tubes

and cohesive failure. In the case of unpatterned gecko tapes, the failure was interfacial

with very little carbon nanotube remaining on the mica surface. Interestingly, live gecko

hairs seem to be wear free even after being pulled along the surface.

In addition, peeling the tape at angles greater than 10° does not involve any

breaking or transferring of the carbon nanotube on the substrate, and the synthetic gecko

tape can be reused many times without damage. This feature is important because peeling

resistance has to be much smaller in comparison to shear forces in mimicking the gecko

feet. The gecko moves or gets unstuck from a surface by uncurling its toes. This is

necessary because, at those peeling angles, the peeling forces are weak and the gecko can

effortlessly move on a vertical surface by using its foot-hairs repeatedly without any damage.

Unlike viscoelastic adhesives, the adhesion strength of CNT-based gecko adhesive is time-independent while Scotch® tape shows strong time-dependence as

expected for viscoelastic materials. It is robust as the strength of the tape does not show degradation over repeated uses. It can also be used in vacuum environment.

In summary, the cooperative nature of the compliant nanohairs controls their anisotropic mechanical properties, adhesion, and friction. The relative contact area for these nanohairs increases with increase in load and is very hysteretic with no real decrease in the actual area of contact until pull-off. This hysteretic behavior of contact area explains the increase in adhesion and friction forces with applied preload.

Understanding the mechanics and deformation of nanoscale fibers such as carbon nanotubes is critical for designing materials such as gecko-inspired adhesives, soft

153

resilient foams, nanocomposites, and shape-memory materials. The dynamic nature of

contact under load made by soft and nanoscale materials, and the difficulty in

determining their footprints underpin the challenges in understanding nanoscale tribology

and adhesion. Hierarchical structures of setae and spatulas found on the gecko foot have

been replicated using aligned multiwalled carbon nanotubes. The multiscale structures

with length scales of micrometers (setae) and nanometers (spatulas) are necessary to

achieve high shear and peeling forces.

5.2 Mechanism of Adhesion in Geckos

An in-depth understanding of the mechanism for the adhesion in geckos is very

important for the design and fabrication of synthetic gecko adhesives. A preponderance

of evidence suggests that geckos stick to substrates as a consequence of the formation of

a large number of intimate setal-substrate contacts engaging van der Waals attraction.

One prediction about performance that can be derived from this mechanism is that adhesion should be temperature insensitive over a biologically meaningful range.

However, the results of previous work are equivocal on this matter. Such a simple prediction ignores important details of adhesion at the whole organism scale. For example, adhesion may involve more than just the sum of the setal-substrate interactions; contributions of chemistry at the interface, elaborate subcutaneous vascular, muscular and skeletal elements in adhesion and release are currently completely unexplored.

Presumably, these components would be subject to thermal dependencies typical of ectotherms. A series of experiment measured adhesion at different temperatures using a single protocol with two species (G. gecko and P. dubia). Although evidence of a strong

154

effect of temperature was found, the direction of the effect was counterintuitive given the thermal biology of geckos and it violated the prediction given by van der Waals interactions presumed responsible for adhesion. Consequently, other factors (e.g., humidity) that could explain the variation in the observed clinging performance were examined. Evidence was found, unexpectedly, that humidity is likely an important determinant of clinging force in the geckos.

The strong influence of humidity on the adhesive forces at a single setal and whole animal level also suggests that capillarity forces may play an important role in adhesion which is expected for hydrophilic-hydrophilic pairs. Alternatively, when one of the materials is hydrophobic, adhesion is humidity independent. This study raises several paradoxes that cannot be resolved with existing data or theories. First, the humidity dependent adhesion forces at 12 °C suggest that setae must be hydrophilic. However, gecko foot surface is superhydrophobic with a water contact angle of 160.9° and water droplets do not wet gecko foot and roll off easily upon tilting. According to the Cassie-

Baxter and Wenzel models, surface that exhibit superhydrophobic water contact angles

( >150°) is the result of chemistry and roughness. Second, it is not clear why humidity has a strong influence on the adhesion forces only at low temperatures. A capillary model predicts that the adhesion forces at the same humidity should be proportional to absolute temperature. Therefore, the adhesion forces at 35 °C should be only 8% higher than the adhesion forces at 12 °C. These predictions are much smaller than experimental observations. In addition, the capillary model predicts the adhesion forces at 35 °C should show similar dependence on humidity. Both the van der Waals and capillary forces fail to explain the shear adhesion data at the whole animal level. Resolution of this paradox will

155

likely require examination of the chemistry at the contact interface and particular way in which water interacts with substrate and seta at the nanometer scale. In addition, the adhesion force measurements at a single setal scale are needed at variable temperatures before eliminating the possibility of thermal biophysical constraints influencing the adhesion.

Irrespective of the mechanism, effects of humidity on otherwise “dry adhesive” biological systems may not be limited to gecko toe pads. The explanation involving rearrangement of keratin protein to expose more polar amino acid groups at the surface has also been discussed as a potential mechanism for increasing adhesion. For example, members of the spider infraorder Araneomorphae produce a derived type of silk which adheres to surfaces with greater force as humidity increases. It has been hypothesized that regularly spaced nodes in the silk strands are richer in hydrophilic polar or charged amino acids, which under humid conditions should promote hydroscopic interactions between the nodes and substrates. It is unclear whether similar mechanisms might underlie the humidity response of gecko toe pads. However, efforts to characterize the protein structure of setae have revealed that setae have a complex structure of heterogeneous α- and β-keratins. Furthermore, there is evidence that the lateral regions of the β-keratins sequenced in gecko setae are relatively rich in hydrophilic and polar amino acids that could modify adhesive interactions under moderate humidity conditions, perhaps in a way analogous to the increased stickiness of nodes in silk of some spiders in the

Araneomorphae. Addition methods are needed to probe water at the contact interface and chemistry at the surface of gecko foot.

156

BIBLIOGRAPHY

1. Labare, M. P.; Guthrie, K.; Weiner, R. M., Polysaccharide exopolymer adhesives from periphytic marine bacteria. J. Adhes. Sci. Technol. 1989, 3, 213-223.

2. Klapper, I.; Rupp, C. J.; Cargo, R.; Purvedorj, B.; Stoodley, P., Viscoelastic fluid description of bacterial biofilm material properties. Biotech. Bioeng. 2002, 80 (3), 289- 296.

3. Sutherland, I., A sticky business. Microbial polysaccharides: current products and future trends. Microbiol. Today 2002, 29, 70-71.

4. Leathers, T. D., Biotechnological production and applications of pullulan. Appl. Microbiol. Biotechnol. 2003, 62 (5), 468-473.

5. Haag, A. P.; Geesey, G. G.; Mittleman, M. W., Bacterially derived wood adhesive. Int. J. Adhes. Adhes. 2006, 26 (3), 177-183.

6. Mendgen, K.; Hahn, M.; Deising, H., Morphogenesis and mechanisms of penetration by plant pathogenic fungi. Annu. Rev. Phytopathol. 1996, 34 (1), 367-386.

7. Hardham, A. R., Cell biology of fungal infection of plants. In The Mycota: biology of the fungal cell, Howard, R. J.; Gow, N. A. R., Eds. Springer: Berlin Heidelberg New York, 2001; Vol. 7, pp 91–123.

8. Osherov, N.; May, G. S., The molecular mechanisms of conidial germination. FEMS Microbiol. Lett. 2001, 199 (2), 153-160.

9. Tucker, S. L.; Talbot, N. J., Surface attachment and pre-penetration stage development by plant pathogenic fungi. Annu. Rev. Phytopathol. 2001, 39, 385-417.

10. Vreeland, V.; Waite, J.; Epstein, L., Polyphenols and oxidases in substratum adhesion by marine algae and mussels. J. Phycol. 1998, 34 (1), 1-8.

11. Walker, G., The biochemical composition of the cement of two barnacle species, Balanus hameri and Balanus crenatus. J. Mar. Biol. Assoc. UK 1972, 52 (429-435).

12. Walker, G., The Adhesion of Barnacles. J. Adhes. 1981, 12 (1), 51 - 58.

13. Yule, A.; Walker, G., Adhesion of the barnacle, Balanus balanoides, to slate surfaces. J. Mar. Biol. Assoc. UK 1984, 64, 147-156.

157

14. Berglin, M.; Larsson, A.; Jonsson, P. R.; Gatenholm, P., The adhesion of the barnacle, Balanus improvisus, to poly(dimethylsiloxane) fouling-release coatings and poly(methyl methacrylate) panels: The effect of barnacle size on strength and failure mode. J. Adhes. Sci. Technol. 2001, 15, 1485-1502.

15. Williams, T. A.; Larsen, J. H., New function for the granular skin glands of the eastern long-toed salamander, Ambystoma macrodactylum columbianum. J. Exp. Zool. 1986, 239 (3), 329-333.

16. Evans, C. M.; Brodie, E. D., Jr., Adhesive Strength of Amphibian Skin Secretions. J. Herpetol. 1994, 28 (4), 499-502.

17. Persson, B. N. J., Wet adhesion with application to tree frog adhesive toe pads and tires. J. Phys.-Condens. Mat. 2007, 19 (37), 376110.

18. Wiegemann, M., Adhesion in blue mussels (Mytilus edulis) and barnacles (genus Balanus): Mechanisms and technical applications. Aquat. Sci. 2005, 67 (2), 166-176.

19. Waite, J.; Andersen, N.; Jewhurst, S.; Sun, C., Mussel adhesion: finding the tricks worth mimicking. J. Adhes. 2005, 81 (3), 297-317.

20. Carrington, E., The ecomechanics of mussel attachment: from molecules to ecosystems. Integr. Comp. Biol. 2002, 42 (4), 846-852.

21. Yu, M.; Hwang, J.; Deming, T. J., Role of l-3,4-dihydroxyphenylalanine in mussel adhesive proteins. J. Am. Chem. Soc. 1999, 121 (24), 5825-5826.

22. Yu, M.; Deming, T. J., Synthetic polypeptide mimics of marine adhesives. Macromolecules 1998, 31 (15), 4739-4745.

23. Waite, J. H., The phylogeny and chemical diversity of quinone-tanned glues and varnishes. Comp. Biochem. Physiol. B 1990, 97 (1), 19-29.

24. Gorb, S.; Gorb, E.; Kastner, V., Scale effects on the attachment pads and friction forces in syrphid flies. J. Exp. Biol. 2001, 204 (8), 1421-1431.

25. Ruibal, R.; Ernst, V., The structure of the digital setae of lizards. J. Morphol. 1965, 117 (3), 271-293.

26. Coddington, J. A.; Levi, H. W., Systematics and evolution of spiders (Araneae). Annu. Rev. Ecol. Syst. 2003, 22 (1), 565-592.

158

27. Stork, N. E., A comparison of the adhesive setae on the feet of lizards and arthropods. J. Nat. Hist. 1983, 17, 829-835.

28. Williams, E. E.; Peterson, J. A., Convergent and alternative designs in the digital adhesive pads of scincid lizards. Science 1982, 215 (4539), 1509-1511.

29. Rovner, J. S., Adhesive hairs in spiders: behavioral functions and hydraulically mediated movement. Symp. Zool. Soc. Lond. 1978 42, 99-108

30. Gorb, S., Attachment devices of insect cuticle. Kluwer Academic: Dordrecht, 2001; p 305.

31. Langer, M. G.; Ruppersberg, J. P.; Gorb, S., Adhesion forces measured at the level of a terminal plate of the fly's seta. Proc. R. Soc. B 2004, 271 (1554), 2209-2215.

32. Varenberg, M.; Gorb, S., A beetle-inspired solution for underwater adhesion. J. R. Soc. Interface 2008, 5 (20), 383-385.

33. Sever, M. J.; Weisser, J. T.; Monahan, J.; Srinivasan, S.; Wilker, J. J., Metal- mediated cross-linking in the generation of a marine-mussel adhesive. Angew. Chem. Int. Ed. 2004, 43 (4), 448-450.

34. Lee, B. P.; Chao, C.-Y.; Nunalee, F. N.; Motan, E.; Shull, K. R.; Messersmith, P. B., Rapid gel formation and adhesion in photocurable and biodegradable block copolymers with high dopa content. Macromolecules 2006, 39 (5), 1740-1748.

35. Autumn, K., Toes unstuck-response. Am. Sci. 2006, 94 (4), 292-293.

36. Autumn, K., Anti-adhesive properties of gecko setae. Integr. Comp. Biol. 2005, 45 (6), 959-959.

37. Autumn, K., How gecko toes stick - The powerful, fantastic adhesive used by geckos is made of nanoscale hairs that engage tiny forces, inspiring envy among human imitators. Am. Sci. 2006, 94 (2), 124-132.

38. Autumn, K., Gecko adhesion: structure, function, and applications. MRS Bull. 2007, 32, 473-478.

39. Autumn, K.; Dittmore, A.; Santos, D.; Spenko, M.; Cutkosky, M., Frictional adhesion: a new angle on gecko attachment. J. Exp. Biol. 2006, 209 (18), 3569-3579.

159

40. Autumn, K.; Gravish, N., Gecko adhesion: evolutionary nanotechnology. Philos. T. R. Soc. A 2007.

41. Autumn, K.; Gravish, N.; Wilkinson, M.; Santos, D.; Spenko, M.; Cutsosky, M., Frictional adhesion of natural and synthetic gecko setal arrays. Integr. Comp. Biol. 2006, 46, E5-E5.

42. Autumn, K.; Hansen, W., Ultrahydrophobicity indicates a non-adhesive default state in gecko setae. J. Comp. Physiol., A 2006, 192 (11), 1205-1212.

43. Autumn, K.; Liang, Y. A.; Hsieh, S. T.; Zesch, W.; Chan, W. P.; Kenny, T. W.; Fearing, R.; Full, R. J., Adhesive force of a single gecko foot-hair. Nature 2000, 405 (6787), 681-685.

44. Autumn, K.; Majidi, C.; Groff, R. E.; Dittmore, A.; Fearing, R., Effective elastic modulus of isolated gecko setal arrays. J. Exp. Biol. 2006, 209 (18), 3558-3568.

45. Autumn, K.; Peattie, A. M., Mechanisms of adhesion in geckos. Integr. Comp. Biol. 2002, 42 (6), 1081-1090.

46. Autumn, K.; Sitti, M.; Liang, Y. C. A.; Peattie, A. M.; Hansen, W. R.; Sponberg, S.; Kenny, T. W.; Fearing, R.; Israelachvili, J. N.; Full, R. J., Evidence for van der Waals adhesion in gecko setae. Proc. Natl. Acad. Sci. USA 2002, 99 (19), 12252-12256.

47. Hiller, U.; Blaschke, R., Zum Haftproblem der Gecko-Füsse. Naturwissenschaften 1967, 54 (13), 344-344.

48. Huber, G. Nanoscale adhesion of individual gecko spatulae explored by atomic force microscopy. University of Stuttgart Max Planck Institute for Metals Research, Stuttgart, Stuttgart, 2006.

49. Huber, G.; Gorb, S. N.; Spolenak, R.; Arzt, E., Resolving the nanoscale adhesion of individual gecko spatulae by atomic force microscopy. Biol. Lett. 2005, 1 (1), 2-4.

50. Sun, W. X.; Neuzil, P.; Kustandi, T. S.; Oh, S.; Samper, V. D., The nature of the gecko lizard adhesive force. Biophys. J. 2005, 89 (2), L14-L17.

51. Gravish, N.; Wilkinson, M.; Sponberg, S.; Parness, A.; Esparza, N.; Soto, D.; Yamaguchi, T.; Broide, M.; Cutkosky, M.; Creton, C.; Autumn, K., Rate-dependent frictional adhesion in natural and synthetic gecko setae. J. R. Soc. Interface 2010, 7 (43), 259-269.

160

52. Arzt, E.; Gorb, S.; Spolenak, R., From micro to nano contacts in biological attachment devices. Proc. Natl. Acad. Sci. USA 2003, 100 (19), 10603-10606.

53. Pesika, N. S.; Tian, Y.; Zhao, B.; Rosenberg, K.; Zeng, H.; McGuiggan, P.; Autumn, K.; Israelachvili, J. N., Peel zone model of tape peeling based on the gecko adhesive system. J. Adhes. 2007, 83 (4), 383 - 401.

54. Russell, A. P.; Higham, T. E., A new angle on clinging in geckos: incline, not substrate, triggers the deployment of the adhesive system. Proc. R. Soc. B 2009, 276 (1673), 3705-3709.

55. Liu, J.; Hui, C. Y.; Shen, L.; Jagota, A., Compliance of a microfibril subjected to shear and normal loads. J. R. Soc. Interface 2008, 5 (26), 1087-1097.

56. Majidi, C., Remarks on formulating an adhesion problem using Euler's elastica (draft). Mech. Res. Commun. 2006, 34 (1), 85-90.

57. Majidi, C., Shear adhesion between an elastica and a rigid flat surface. Mech. Res. Commun. 2009, 36 (3), 369-372.

58. Strus, M. C.; Zalamea, L.; Raman, A.; Pipes, R. B.; Nguyen, C. V.; Stach, E. A., Peeling force spectroscopy: Exposing the adhesive nanomechanics of one-dimensional nanostructure. Nano Lett. 2008, 8 (2), 544-550.

59. Yamaguchi, T.; Gravish, N.; Autumn, K.; Creton, C., Microscopic modeling of the dynamics of frictional adhesion in the gecko attachment system. J. Phys. Chem. B 2008, 113 (12), 3622-3628.

60. Bhushan, B., Adhesion of multi-level hierarchical attachment systems in gecko feet. J. Adhes. Sci. Technol. 2007, 21 (12-13), 1213-1258.

61. Bhushan, B.; Peressadko, A. G.; Kim, T. W., Adhesion analysis of two-level hierarchical morphology in natural attachment systems for 'smart adhesion'. J. Adhes. Sci. Technol. 2006, 20 (13), 1475-1491.

62. Chen, B.; Wu, P. D.; Gao, H., Hierarchical modelling of attachment and detachment mechanisms of gecko toe adhesion. Proc. R. Soc. A 2008, 464 (2094), 1639- 1652.

63. Gao, H.; Wang, X.; Yao, H.; Gorb, S.; Arzt, E., Mechanics of hierarchical adhesion structures of geckos. Mech. Mater. 2005, 37 (2-3), 275-285.

161

64. Hui, C. Y.; Jagota, A.; Shen, L.; Rajan, A.; Glassmaker, N.; Tang, T., Design of bio-inspired fibrillar interfaces for contact and adhesion - theory and experiments. J. Adhes. Sci. Technol. 2007, 21 (12-13), 1259-1280.

65. Kim, T. W.; Bhushan, B., Effect of stiffness of multi-level hierarchical attachment system on adhesion enhancement. Ultramicroscopy 2007, 107 (10-11), 902-912.

66. Schargott, M., A mechanical model of biomimetic adhesive pads with tilted and hierarchical structures. Bioinspiration & Biomimetics 2009, 4 (2), 026002.

67. Yao, H.; Gao, H., Mechanics of robust and releasable adhesion in biology: Bottom-up designed hierarchical structures of gecko. J. Mech. Phys. Solids 2006, 54 (6), 1120-1146.

68. Yao, H.; Gao, H., Mechanical principles of robust and releasable adhesion of gecko. J. Adhes. Sci. Technol. 2007, 21 (12-13), 1185-1212.

69. Gao, H. J.; Ji, B. H.; Jager, I. L.; Arzt, E.; Fratzl, P., Materials become insensitive to flaws at nanoscale: Lessons from nature. Proc. Natl. Acad. Sci. USA 2003, 100 (10), 5597-5600.

70. Gao, H.; Yao, H., Shape insensitive optimal adhesion of nanoscale fibrillar structures. Proc. Natl. Acad. Sci. USA 2004, 101 (21), 7851-6.

71. Geim, A. K.; Dubonos, S. V.; Grigorieva, I. V.; Novoselov, K. S.; Zhukov, A. A.; Shapoval, S. Y., Microfabricated adhesive mimicking gecko foot-hair. Nat. Mater. 2003, 2 (7), 461-463.

72. Schubert, B.; Majidi, C.; Groff, R. E.; Baek, S.; Bush, B.; Maboudian, R.; Fearing, R. S., Towards friction and adhesion from high modulus microfiber arrays. J. Adhes. Sci. Technol. 2007, 21 (12-13), 1297-1315.

73. Peattie, A. M.; Majidi, C.; Corder, A.; Full, R. J., Ancestrally high elastic modulus of gecko setal β-keratin. J. R. Soc. Interface 2007, 4 (17), 1071-1076.

74. Teodorescu, M.; Majidi, C.; Rahnejat, H.; Fearing, R. S. In Effect of surface roughness on adhesion and friction of microfibers in side contact, The STLE/ASME International Joint Tribology Conference, Miami, Florida, Miami, Florida, 2008.

75. Takahashi, K.; Berengueres, J. O. L.; Obata, K. J.; Saito, S., Geckos' foot hair structure and their ability to hang from rough surfaces and move quickly. Int. J. Adhes. Adhes. 2006, 26 (8), 639-643.

162

76. Pugno, N. M.; Lepore, E., Observation of optimal gecko's adhesion on nanorough surfaces. BioSyst. 2008, 94 (3), 218-222.

77. Porwal, P. K.; Hui, C. Y., Strength statistics of adhesive contact between a fibrillar structure and a rough substrate. J. R. Soc. Interface 2008, 5 (21), 441-448.

78. Hui, C. Y.; Glassmaker, N. J.; Jagota, A., How compliance compensates for surface roughness in fibrillar adhesion. J. Adhes. 2005, 81 (7-8), 699-721.

79. Hansen, W. R.; Autumn, K., Evidence for self-cleaning in gecko setae. Proc. Natl. Acad. Sci. USA 2005, 102 (2), 385-9.

80. Sitti, M.; Fearing, R. S., Synthetic gecko foot-hair micro/nano-structures as dry adhesives. J. Adhes. Sci. Technol. 2003, 17 (8), 1055-1073.

81. Campolo, D.; Jones, S. D.; Fearing, R. S. In Fabrication of gecko foot-hair like nano structures and adhesion to random rough surfaces, IEEE Nanotechnology Conference, San Francisco, CA, San Francisco, CA, 2003.

82. Sitti, M.; Fearing, R. S. In Synthetic gecko foot-hair micro/nano-structures for future wall-climbing robots, IEEE International Conference on Robotics and Automation, Taipei, Taiwan, September 2003; Taipei, Taiwan, 2003; pp 1164-1170.

83. Ghatak, A.; Mahadevan, L.; Chung, J. Y.; Chaudhury, M. K.; Shenoy, V., Peeling from a biomimetically patterned thin elastic film. Proc. R. Soc. A 2004, 460 (2049), 2725-2735.

84. Majidi, C.; Groff, R.; Fearing, R. In Clumping and packing of hair arrays manufactured by nanocasting, ASME International Mechanical Engineering Congress & Exposition, Anaheim, California, USA, November 2004; Anaheim, California, USA, 2004.

85. Menon, C.; Murphy, M.; Sitti, M. In Gecko inspired surface climbing robots, IEEE International Conference on Robotics and Biomimetics, Shenyang, China, August 2004; Shenyang, China, 2004; pp 431- 436.

86. Shah, G. J.; Sitti, M. In Modeling and design of biomimetic adhesives inspired by gecko foot-hairs, IEEE International Conference on Robotics and Biomimetics, Shenyang, China, August 22 - 26, 2004; Shenyang, China, 2004; pp 873-878.

87. Daltorio, K. A.; Horchler, A. D.; Gorb, S.; Ritzmann, R. E.; D., Q. R. In A small wall-walking robot with compliant, adhesive feet, IEEE/RSJ International Conference on

163

Intelligent Robots and Systems, Edmonton, Alberta, Canada, August 2005; Horchler, A. D., Ed. Edmonton, Alberta, Canada, 2005; pp 4018-4023.

88. Fearing, R. S. E. C., CA, US), Sitti, Metin (Pittsburgh, PA, US) Adhesive microstructure and method of forming same. 2005.

89. La Spina, G.; Stefanini, C.; Menciassi, A.; Dario, P., A novel technological process for fabricating micro-tips for biomimetic adhesion. J. Micromech. Microeng. 2005, 15 (8), 1576-1587.

90. Northen, M. T.; Turner, K. L., A batch fabricated biomimetic dry adhesive. Nanotechnology 2005, 16 (8), 1159-1166.

91. Choi, P.; Kim, J.; Guo, L. J. In Fabrication of synthetic gecko tape and characterization of adhesion forces, International Conference on Electron, Ion, and Photon Beam Technology and Nanofabrication, Baltimore, May 2006; Baltimore, 2006.

92. Dai, Z. D.; Yu, M.; Gorb, S., Adhesion characteristics of polyurethane for bionic hairy foot. J. Intell. Mater. Syst. Struct. 2006, 17 (8-9), 737-741.

93. Jeong, H. E.; Lee, S. H.; Kim, J. K.; Suh, K. Y., Nanoengineered multiscale hierarchical structures with tailored wetting properties. Langmuir 2006, 22 (4), 1640- 1645.

94. Jeong, H. E.; Lee, S. H.; Kim, P.; Suh, K. Y., Stretched polymer nanohairs by nanodrawing. Nano Lett. 2006, 6 (7), 1508-1513.

95. Kim, S.; Metin, S., Biologically inspired polymer microfibers with spatulate tips as repeatable fibrillar adhesives. Appl. Phys. Lett. 2006, 89 (26), 261911.

96. Majidi, C.; Groff, R. E.; Maeno, Y.; Schubert, B.; Baek, S.; Bush, B.; Maboudian, R.; Gravish, N.; Wilkinson, M.; Autumn, K.; Fearing, R. S., High friction from a stiff polymer using microfiber arrays. Phys. Rev. Lett. 2006, 97 (7), 076103.

97. Northen, M. T.; Turner, K. L., Meso-scale adhesion testing of integrated micro- and nano-scale structures. Sensor. Actuat. A-Phys. 2006, 130, 583-587.

98. Northen, M. T.; Turner, K. L., Batch fabrication and characterization of nanostructures for enhanced adhesion. Curr. Appl. Phys. 2006, 6 (3), 379-383.

99. Yoon, E. S.; Singh, R. A.; Kong, H.; Kim, B.; Kim, D. H.; Jeong, H. E.; Suh, K. Y., Tribological properties of bio-mimetic nano-patterned polymeric surfaces on silicon wafer. Tribol. Lett. 2006, 21 (1), 31-37.

164

100. Aksak, B.; Murphy, M. P.; Sitti, M., Adhesion of biologically inspired vertical and angled polymer microfiber arrays. Langmuir 2007.

101. Aksak, B.; Sitti, M.; Cassell, A.; Li, J.; Meyyappan, M.; Callen, P., Friction of partially embedded vertically aligned carbon nanofibers inside elastomers. Appl. Phys. Lett. 2007, 91 (6), 061906/1-061906/3.

102. Bhushan, B.; Sayer, R. A., Surface characterization and friction of a bio-inspired reversible adhesive tape. Microsyst. Technol. 2007, 13 (1), 71-78.

103. del Campo, A.; Greiner, C.; Alvarez, I.; Arzt, E., Patterned surfaces with pillars with controlled 3D tip geometry mimicking bioattachment devices. Adv. Mater. 2007, 19 (15), 1973-1977.

104. Gorb, S. N.; Sinha, M.; Peressadko, A.; Daltorio, K. A.; Quinn, R. D., Insects did it first: a micropatterned adhesive tape for robotic applications. Bioinspiration & Biomimetics 2007, 2 (4), S117-S125.

105. Greiner, C.; delCampo, A.; Arzt, E., Adhesion of bioinspired micropatterned surfaces: effects of pillar radius, aspect ratio, and preload. Langmuir 2007, 23 (7), 3495- 3502.

106. Kim, D. S.; Lee, H. S.; Lee, J.; Kim, S.; Lee, K.-H.; Moon, W.; Kwon, T. H., Replication of high-aspect-ratio nanopillar array for biomimetic gecko foot-hair prototype by UV nano embossing with anodic aluminum oxide mold. Microsyst. Technol. 2007, 13 (5-6), 601-606.

107. Kim, S.; Aksak, B.; Sitti, M., Enhanced friction of elastomer microfiber adhesives with spatulate tips. Appl. Phys. Lett. 2007, 91 (22), 221913.

108. Kustandi, T. S.; Samper, V. D.; Ng, W. S.; Chong, A. S.; Gao, H., Fabrication of a gecko-like hierarchical fibril array using a bonded porous alumina template. J. Micromech. Microeng. 2007, 17 (10), N75-N81.

109. Kustandi, T. S.; Samper, V. D.; Yi, D. K.; Ng, W. S.; Neuzil, P.; Sun, W., Self- assembled nanoparticles based fabrication of gecko foot-hair-inspired polymer nanofibers. Adv. Funct. Mater. 2007, 17 (13), 2211-2218.

110. Murphy, M. P.; Aksak, B.; Sitti, M., Adhesion and anisotropic friction enhancements of angled heterogeneous micro-fiber arrays with spherical and spatula tips. J. Adhes. Sci. Technol. 2007, 21, 1281-1296.

165

111. Noderer, W. L.; Shen, L.; Vajpayee, S.; Glassmaker, N. J.; Jagota, A.; Hui, C. Y., Enhanced adhesion and compliance of film-terminated fibrillar surfaces. Proc. R. Soc. A 2007, 463 (2086), 2631-2654.

112. Reddy, S.; Arzt, E.; del Campo, A., Bioinspired surfaces with switchable adhesion. Adv. Mater. 2007, 19 (22), 3833-3837.

113. Santos, D.; Spenko, M.; Parness, A.; Kim, S.; Cutkosky, M., Directional adhesion for climbing: theoretical and practical considerations. J. Adhes. Sci. Technol. 2007, 21 (12-13), 1317-1341.

114. Varenberg, M.; Gorb, S., Shearing of fibrillar adhesive microstructure: friction and shear-related changes in pull-off force. J. R. Soc. Interface 2007, 4 (15), 721-725.

115. Yao, H.; Rocca, G. D.; Guduru, P. R.; Gao, H., Adhesion and sliding response of a biologically inspired fibrillar surface: experimental observations. J. R. Soc. Interface 2007.

116. Zhao, A.; Mei, T.; Lin, X.; Ni, L. In Fabrication and characterization of tree-like nanorod arrays for bionic gecko foot-hairs, 7th IEEE International Conference on Nanotechnology, Hong Kong, China, August 2 - 5, 2007; Hong Kong, China, 2007; pp 259-262.

117. Aksak, B.; Murphy, M.; Sitti, M. In Gecko inspired micro-fibrillar adhesives for wall climbing robots on micro/nanoscale rough surfaces, IEEE International Conference on Robotics and Automation, Pasadena, CA, May 2008; Pasadena, CA, 2008.

118. Cheung, E.; Sitti, M., Adhesion of biologically inspired oil-coated polymer micropillars. J. Adhes. Sci. Technol. 2008, 5 (6), 569-589.

119. Cho, W. K.; Choi, I. S., Fabrication of hairy polymeric films inspired by geckos: Wetting and high adhesion properties. Adv. Funct. Mater. 2008, 18 (7), 1089-1096.

120. Jeong, H. E.; Lee, S. H.; Kim, P.; Suh, K. Y., High aspect-ratio polymer nanostructures by tailored capillarity and adhesive force. Colloids Surf. A 2008, 313-314, 359-364.

121. Kim, S.; Kustandi, T. S.; Yi, D. K., Synthesis of artificial polymeric nanopillars for clean and reusable adhesives. J. Nanosci. Nanotechno. 2008, 8, 4779-4782.

122. Kim, S.; Sitti, M.; Jang, J.; Thomas, E. In Fabrication of bio-inspired elastomer nanofiber arrays with spatulate tips using notching effect, 8th IEEE Conference on Nanotechnology, 2008; pp 780-782.

166

123. Kim, S.; Spenko, M.; Trujillo, S.; Heyneman, B.; Santos, D.; Cutkosky, M., Smooth vertical surface climbing with directional adhesion. IEEE Transactions on Robotics 2008, 24 (1), 65-74.

124. Lee, J.; Fearing, R. S.; Komvopoulos, K., Directional adhesion of gecko-inspired angled microfiber arrays. Appl. Phys. Lett. 2008, 93 (19), 191910-3.

125. Lee, J.; Majidi, C.; Schubert, B.; Fearing, R. S., Sliding-induced adhesion of stiff polymer microfibre arrays. I. Macroscale behaviour. J. R. Soc. Interface 2008, 5 (25), 835-844.

126. Lu, G.; Hong, W.; Tong, L.; Bai, H.; Wei, Y.; Shi, G., Drying enhanced adhesion of polythiophene nanotubule arrays on smooth surfaces. ACS Nano 2008, 2 (11), 2342- 2348.

127. Mahdavi, A.; Ferreira, L.; Sundback, C.; Nichol, J. W.; Chan, E. P.; Carter, D. J. D.; Bettinger, C. J.; Patanavanich, S.; Chignozha, L.; Ben-Joseph, E.; Galakatos, A.; Pryor, H.; Pomerantseva, I.; Masiakos, P. T.; Faquin, W.; Zumbuehl, A.; Hong, S.; Borenstein, J.; Vacanti, J.; Langer, R.; Karp, J. M., A biodegradable and biocompatible gecko-inspired tissue adhesive. Proc. Natl. Acad. Sci. USA 2008, 105 (7), 2307-2312.

128. Northen, M. T.; Greiner, C.; Arzt, E.; Turner, K. L., A gecko-inspired reversible adhesive. Adv. Mater. 2008, 20 (20), 3905-3909.

129. Sameoto, D.; Li, Y.; Menon, C., Micromask generation for polymer morphology control: nanohair fabrication for synthetic dry adhesives. Adv. Sci. Tech. 2008, 54, 439- 444.

130. Santos, D.; Heyneman, B.; Kim, S.; Esparza, N.; Cutkosky, M. In Gecko-inspired climbing behaviors on vertical and overhanging surfaces, 2008 IEEE International Conference on Robotics and Automation, Pasadena, CA, USA, May 19-23, 2008; Pasadena, CA, USA, 2008; pp 19-23.

131. Schubert, B.; Lee, J.; Majidi, C.; Fearing, R. S., Sliding-induced adhesion of stiff polymer microfibre arrays. II. Microscale behaviour. J. R. Soc. Interface 2008, 5 (25), 845-853.

132. Boesel, L. F.; Greiner, C.; Arzt, E.; Campo, A. d., Gecko-inspired surfaces: a path to strong and reversible dry adhesives. 2009.

133. Cheung, E.; Sitti, M., Adhesion of biologically inspired polymer microfibers on soft surfaces. Langmuir 2009, 25 (12), 6613-6616.

167

134. Davies, J.; Haq, S.; Hawke, T.; Sargent, J. P., A practical approach to the development of a synthetic Gecko tape. Int. J. Adhes. Adhes. 2009, 29 (4), 380-390.

135. Glass, P.; Chung, H.; Washburn, N. R.; Sitti, M., Enhanced reversible adhesion of dopamine methacrylamide-coated elastomer microfibrillar structures under wet conditions. Langmuir 2009, 25 (12), 6607-6612.

136. Greiner, C.; Arzt, E.; Campo, A. d., Hierarchical gecko-like adhesives. Adv. Mater. 2009, 21 (4), 479-482.

137. Jeong, H. E.; Lee, J.-K.; Kim, H. N.; Moon, S. H.; Suh, K. Y., A nontransferring dry adhesive with hierarchical polymer nanohairs. Proc. Natl. Acad. Sci. USA 2009, 106 (14), 5639-5644.

138. Kim, S.; Sitti, M.; Xie, T.; Xiao, X., Reversible dry micro-fibrillar adhesives with thermally controllable adhesion. Soft Matter 2009, 5 (19), 3689-3693.

139. Kim, T. I.; Jeong, H. E.; Suh, K. Y.; Lee, H. H., Stooped nanohairs: Geometry- controllable, unidirectional, reversible, and robust gecko-like dry adhesive. Adv. Mater. 2009, 21 (22), 2276-2281.

140. Kim, T. I.; Pang, C.; Suh, K. Y., Shape-tunable polymer nanofibrillar structures by oblique electron beam irradiation. Langmuir 2009, 25 (16), 8879-8882.

141. Ko, H.; Lee, J.; Schubert, B. E.; Chueh, Y.-L.; Leu, P. W.; Fearing, R. S.; Javey, A., Hybrid core−shell nanowire forests as self-selective chemical connectors. Nano Lett. 2009, 9 (5), 2054-2058.

142. Lee, J.; Bush, B.; Maboudian, R.; Fearing, R. S., Gecko-inspired combined lamellar and nanofibrillar array for adhesion on nonplanar surface. Langmuir 2009, 25 (21), 12449-12453.

143. Murphy, M. P.; Aksak, B.; Sitti, M., Gecko-inspired directional and controllable adhesion. Small 2009, 5 (2), 170-175.

144. Murphy, M. P.; Kim, S.; Sitti, M., Enhanced adhesion by gecko-inspired hierarchical fibrillar adhesives. ACS Appl. Mater. Interfaces 2009, 1 (4), 849-855.

145. Sameoto, D.; Menon, C., Direct molding of dry adhesives with anisotropic peel strength using an offset lift-off photoresist mold. J. Micromech. Microeng. 2009, 19 (11), 115026.

168

146. Sameoto, D.; Menon, C., A low-cost, high-yield fabrication method for producing optimized biomimetic dry adhesives. J. Micromech. Microeng. 2009, 19 (11), 115002.

147. Chen, S. H.; Mi, C. H., Friction properties of bio-mimetic nano-fibrillar arrays. Chin. Phys. Lett. 2009, 26 (10), 108103.

148. Yanik, M. F., Towards gecko-feet-inspired bandages. 2009, 27 (1), 1-2.

149. Yoon, H.; Jeong, H. E.; Kim, T. I.; Kang, T. J.; Tahk, D.; Char, K.; Suh, K. Y., Adhesion hysteresis of Janus nanopillars fabricated by nanomolding and oblique metal deposition. Nano Today 2009, 4 (5), 385-392.

150. Sitti, M.; Cusick, B.; Aksak, B.; Nese, A.; Lee, H. I.; Dong, H.; Kowalewski, T.; Matyjaszewski, K., Dangling chain elastomers as repeatable fibrillar adhesives. ACS Appl. Mater. Interfaces 2009, 1 (10), 2277-2287.

151. Jeong, H. E.; Kwak, R.; Kim, J. K.; Suh, K. Y., Generation and self-replication of monolithic, dual-scale polymer structures by two-step capillary-force lithography. Small 2008, 4 (11), 1913-1918.

152. Yurdumakan, B.; Raravikar, N. R.; Ajayan, P. M.; Dhinojwala, A., Synthetic gecko foot-hairs from multiwalled carbon nanotubes. Chem. Commun. 2005, (30), 3799- 3801.

153. Zhao, Y.; Tong, T.; Delzeit, L.; Kashani, A.; Meyyappan, M.; Majumdar, A., Interfacial energy and strength of multiwalled-carbon-nanotube-based dry adhesive. J. Vac. Sci. Technol. B 2006, 24 (1), 331-335.

154. Ge, L.; Sethi, S.; Ci, L.; Ajayan, P. M.; Dhinojwala, A., Carbon nanotube-based synthetic gecko tapes. Proc. Natl. Acad. Sci. USA 2007, 104 (26), 10792-10795.

155. Qu, L.; Dai, L., Gecko-foot-mimetic aligned single-walled carbon nanotube dry adhesives with unique electrical and thermal properties. Adv. Mater. 2007, 19 (22), 3844- 3849.

156. Qu, L.; Dai, L.; Stone, M.; Xia, Z.; Wang, Z. L., Carbon nanotube arrays with strong shear binding-on and easy normal lifting-off. Science 2008, 322, 238-242.

157. Sethi, S.; Ge, L.; Ci, L.; Ajayan, P. M.; Dhinojwala, A., Gecko-inspired carbon nanotube-based self-cleaning adhesives. Nano Lett. 2008, 8 (3), 822-825.

158. Maeno, Y.; Nakayama, Y., Geckolike high shear strength by carbon nanotube fiber adhesives. Appl. Phys. Lett. 2009, 94 (1), 012103-3.

169

159. Huber, G.; Mantz, H.; Spolenak, R.; Mecke, K.; Jacobs, K.; Gorb, S. N.; Arzt, E., Evidence for capillarity contributions to gecko adhesion from single spatula nanomechanical measurements. Proc. Natl. Acad. Sci. USA 2005, 102 (45), 16293-16296.

160. Kim, T. W.; Bhushan, B., The adhesion model considering capillarity for gecko attachment system. J. R. Soc. Interface 2008, 5 (20), 319-327.

161. Puthoff, J. B.; Prowse, M. S.; Wilkinson, M.; Autumn, K., Changes in materials properties explain the effects of humidity on gecko adhesion. J. Exp. Biol. 2010, 213 (21), 3699-3704.

162. Bergmann, P. J.; Irschick, D. J., Effects of temperature on maximum clinging ability in a diurnal gecko: Evidence for a passive clinging mechanism? J. Exp. Zool. A Comp. Exp. Biol. 2005, 303A (9), 785-791.

163. Losos, J. B., Thermal sensitivity of sprinting and clinging performance in the Tokay gecko (Gekko gecko). Asiatic Herpetol. Res. 1990, 3, 54-59.

164. Huber, G.; Gorb, S. N.; Hosoda, N.; Spolenak, R.; Arzt, E., Influence of surface roughness on gecko adhesion. Acta Biomater. 2007, 3 (4), 607-610.

165. Jusufi, A.; Goldman, D. I.; Revzen, S.; Full, R. J., Active tails enhance arboreal acrobatics in geckos. Proc. Natl. Acad. Sci. USA 2008, 105 (11), 4215-4219.

166. Cartier, O., Abt. Epidermis Studien über den feineren Bau der Haut bei den Reptilien. I. der Geckotiden. Arb. Zool. Inst. Würzburg 1872, 1, 83-96.

167. Cartier, O., Studien über den feineren Bau der Epidermis bei den Geckotiden. Verh. Würzburger Phys. Med. Ges. 1872, 1, 239–258.

168. Cartier, O., Studien über den feineren Bau der Epidermis bei den Reptilien. II. Abtheilung. Über die achsthumserscheinungen der Oberhaut von Schlangen und Eidechsen bei der Häutung. Arb. Zoo.l Zoot. Inst. Würzburg 1874, 1, 239-258.

169. Schmidt, H. R., Zur Anatomie und Physiologie der Geckopfote. Jena. Z. Naturw. 1904, 39, 551.

170. Altevogt, R., Probleme eines Fußes. Kosmos 1954, 50, 428-430.

171. Alibardi, L.; Edward, D.; Patil, L.; Bouheni, R.; Dhinojwala, A.; Niewiarowski, P. H., Histochemical and Ultrastructural analysis of adhesive setae of the lizards Phelsuma dubia and Anolis carolinensis indicates they contain lipids in addition to keratins. 2010.

170

172. Hermann Schleich, H.; Kästle, W., Ultrastrukturen an Gecko-Zehen (Reptilia: Sauria: Gekkonidae). Amphibia-Reptilia 1986, 7, 141-166.

173. Röll, B., Epidermal fine structure of the toe tips of Sphaerodactylus cinereus (Reptilia, Gekkonidae). J. Zool. 1995, 235 (2), 289-300.

174. Rizzo, N. W.; Gardner, K. H.; Walls, D. J.; Keiper-Hrynko, N. M.; Ganzke, T. S.; Hallahan, D. L., Characterization of the structure and composition of gecko adhesive setae. J. R. Soc. Interface 2006, 1742–5689.

175. Alibardi, L., Ultrastructural autoradiographic and immunocytochemical analysis of setae formation and keratinization in the digital pads of the gecko Hemidactylus turcicus (Gekkonidae, Reptilia). Tissue Cell 2003, 35 (4), 288-296.

176. Sawyer, R. H.; Glenn, T.; French, J. O.; Mays, B.; Shames, R. B.; Barnes, G. L.; Rhodes, W.; Ishikawa, Y., The expression of beta (beta) keratins in the epidermal appendages of reptiles and birds. Am. Zool. 2000, 40 (4), 530-539.

177. Irschick, D. J.; Austin, C. C.; Petren, K.; Fisher, R. N.; Losos, J. B.; Ellers, O., A comparative analysis of clinging ability among pad-bearing lizards. Biol. J. Linn. Soc. 1996, 59 (1), 21-35.

178. Bergmann, P.; Irschick, D. J., Effects of temperature on maximum acceleration, deceleration and power output during vertical running in geckos. J. Exp. Biol. 2006, 209 (8), 1404-1412.

179. Niewiarowski, P. H.; Lopez, S.; Ge, L.; Hagan, E.; Dhinojwala, A., Sticky gecko feet: The role of temperature and humidity. PLoS ONE 2008, 3 (5), e2192.

180. Hiller, U., Untersuchungen zum Feinbau und zur Funktion der Haftborsten von Reptilien. Z. Morph. Tiere 1968, 62 (4), 307-362.

181. Dellit, W. D., Zur Anatomie und Physiologie der Geckozehe. Jena. Z. Naturw. 1934, 68, 613-656.

182. Haase, A., Untersuchungen über den Bau und die Entwicklung der Haftlappen bei den Geckotiden. Arch. Naturgesch. 1900, 66, 321-345.

183. Wagler, J. G., Natürliches System der Amphibien, mit vorangehender Classification der Säugetiere und Vögel Cotta’sche Buchhandlung: München, Stuttgart und Tübingen, 1830.

171

184. Simmermacher, G., Haftapparate bei Wirbeltieren. Zool. Garten. 1884, 25, 289- 301.

185. International Union of Pure and Applied Chemistry, IUPAC compendium of chemical terminology. 2nd ed.; Royal Society of Chemistry: Cambridge, England, 2000.

186. Israelachvili, J. N., Intermolecular and surface forces. 2nd ed.; Academic Press: London ; San Diego, CA, 1992; p xxi, 450 p.

187. Gorb, E. V.; Gorb, S. N., Attachment ability of the beetle Chrysolina fastuosa on various plant surfaces. Entomol. Exp. Appl. 2002, 105 (1), 13-28.

188. Federle, W.; Riehle, M.; Curtis, A. S. G.; Full, R. J., An integrative study of insect adhesion: mechanics and wet adhesion of pretarsal pads in ants. Integr. Comp. Biol. 2002, 42 (6), 1100-1106.

189. Hanna, G.; Jon, W.; Jon Barnes, W. P., Adhesion and detachment of the toe pads of tree frogs. J. Exp. Biol. 1991, 155 (1), 103-125.

190. Neinhuis, C.; Barthlott, W., Characterization and distribution of water-repellent, self-cleaning plant surfaces. Ann. Bot. 1997, 79 (6), 667-677.

191. Baum; Baum, C.; Meyer; Meyer, W.; Stelzer; Stelzer, R.; Fleischer; Fleischer, L. G.; Siebers; Siebers, D., Average nanorough skin surface of the pilot whale (Globicephala melas, Delphinidae): considerations on the self-cleaning abilities based on nanoroughness. Marine Biol. 2002, 140 (3), 653-657.

192. Hui, C. Y.; Shen, L.; Jagota, A.; Autumn, K., Mechanics of anti-fouling or self- cleaning in Geckos. In Proceedings of the Annual Adhesion Society Meeting, 2006.

193. Lee, J.; Fearing, R. S., Contact self-cleaning of synthetic gecko adhesive from polymer microfibers. Langmuir 2008, 24 (19), 10587-10591.

194. Stork, N. E., Experimental analysis of adhesion of chrysolina polita (chrysomelidae: coleoptera) on a variety of surfaces. J. Exp. Biol. 1980, 88 (1), 91-108.

195. Walker, G.; Yulf, A. B.; Ratcliffe, J., The adhesive organ of the blowfly, Calliphora vomitoria: a functional approach (Diptera: Calliphoridae). J. Zool. A 1985, 205 (2), 297-307.

196. Irschick, D. J.; Herrel, A.; Vanhooydonck, B., Whole-organism studies of adhesion in pad-bearing lizards: creative evolutionary solutions to functional problems. J. Comp. Physiol., A 2006, 192 (11), 1169-77.

172

197. Beutel, R. G.; Gorb, S. N., Ultrastructure of attachment specializations of hexapods (Arthropoda): evolutionary patterns inferred from a revised ordinal phylogeny. J. Zool. Syst. Evol. Res. 2001, 39 (4), 177-207.

198. Hertz, H., On the Contact of Rigid Elastic Solids and on Hardness. Macmillan: New York, 1882.

199. Johnson, K. L.; Kendall, K.; Roberts, A. D., Surface energy and the contact of elastic solids. Proc. R. Soc. A 1971, 324 (1558), 301-313.

200. Kendall, K., The adhesion and surface energy of elastic solids. J. Phys. D Appl. Phys. 1971, 4 (8), 1186-1195.

201. Griffith, A. A., The phenomena of rupture and flow in solids. Philos. T. R. Soc. A 1921, 221, 163-198.

202. Derjaguin, B. V.; Muller, V. M.; Toporov, Y. P., Effect of contact deformations on the adhesion of particles. J. Colloid Interface Sci. 1975, 53 (2), 314-326.

203. Kesel, A. B.; Martin, A.; Seidl, T., Adhesion measurements on the attachment devices of the jumping spider Evarcha arcuata. J, Exp. Biol. 2003, 206 (16), 2733-2738.

204. Federle, W., Why are so many adhesive pads hairy? J. Exp. Biol. 2006, 209 (14), 2611-2621.

205. Hui, C. Y.; Glassmaker, N. J.; Tang, T.; Jagota, A., Design of biomimetic fibrillar interfaces: 2. Mechanics of enhanced adhesion. J. R. Soc. Interface 2004, 1 (1), 35-48.

206. Spolenak, R.; Gorb, S.; Gao, H. J.; Arzt, E., Effects of contact shape on the scaling of biological attachments. Proc. R. Soc. A 2005, 461 (2054), 305-319.

207. Kendall, K., Thin-film peeling-the elastic term. J. Phys. D Appl. Phys. 1975, 8 (13), 1449-1452.

208. Persson, B. N. J., On the mechanism of adhesion in biological systems. J. Chem. Phys. 2003, 118 (16), 7614-7621.

209. Gay, C.; Leibler, L., Theory of tackiness. Phys. Rev. Lett. 1999, 82, 936.

210. Jagota, A.; Bennison, S. J., Mechanics of adhesion through a fibrillar microstructure. Integr. Comp. Biol. 2002, 42 (6), 1140-1145.

173

211. Chung, J. Y.; Chaudhury, M. K., Roles of discontinuities in bio-inspired adhesive pads. J. R. Soc. Interface 2005, 2 (2), 55-61.

212. Tian, Y.; Pesika, N.; Zeng, H.; Rosenberg, K.; Zhao, B.; McGuiggan, P.; Autumn, K.; Israelachvili, J., Adhesion and friction in gecko toe attachment and detachment. Proc. Natl. Acad. Sci. USA 2006, 103 (51), 19320-19325.

213. Greiner, C.; Spolenak, R.; Arzt, E., Adhesion design maps for fibrillar adhesives: The effect of shape. Acta Biomater. 2009, 5 (2), 597-606.

214. Wang, S., Size effect in nanoadhesion of fibrillar structure. Chin. Sci. Bull. 2007, 52 (5), 592-595.

215. Fuller, K. N. G.; Tabor, D., The effect of surface roughness on the adhesion of elastic solids. Proc. R. Soc. A 1975, 345 (1642), 327-342.

216. Briggs, G. A. D.; Briscoe, B. J., The effect of surface topography on the adhesion of elastic solids. J. Phys. D Appl. Phys. 1977, 10 (18), 2453-2466.

217. Dahlquist, C. A., Pressure-sensitive adhesives. In Treatise on adhesion and adhesives, Patrick, R. L., Ed. Dekker: New York, 1969; Vol. 2, pp 219-260.

218. Pocius, A. V., Adhesion and adhesives technology: an introduction. 2nd ed.; Hanser: Munich, 2002; p xiii, 279 p.

219. Alibardi, L.; Sawyer, R. H., Immunocytochemical analysis of beta (β) keratins in the epidermis of chelonians, lepidosaurians, and archosaurians. J. Exp. Zool. 2002, 293 (1), 27-38.

220. Glassmaker, N. J.; Himeno, T.; Hui, C. Y.; Kim, J., Design of biomimetic fibrillar interfaces: 1. Making contact. J. R. Soc. Interface 2004, 1 (1), 23-33.

221. Persson, B. N. J.; Gorb, S., The effect of surface roughness on the adhesion of elastic plates with application to biological systems. J. Chem. Phys. 2003, 119 (21), 11437-11444.

222. Spolenak, R.; Gorb, S.; Arzt, E., Adhesion design maps for bio-inspired attachment systems. Acta Biomater. 2005, 1 (1), 5-13.

174

224. http://robotics.eecs.berkeley.edu/~ronf/Gecko/index.html. (accessed 03/11/2010).

225. Sangbae, K.; Spenko, M.; Trujillo, S.; Heyneman, B.; Santos, D.; Cutkosky, M. R., Smooth vertical surface climbing with directional adhesion. Robotics, IEEE Transactions on 2008, 24 (1), 65-74.

226. http://nanolab.me.cmu.edu/projects/waalbots/index.shtml. (accessed 03/11/2010).

227. Ge, L.; Sethi, S.; Ci, L.; Ajayan, P. M.; Dhinojwala, A., Shear resistance of carbon nanotube hierarchical microstructures. In Proceedings of the 31st Annual Meeting of The Adhesion Society, 2008, Ed. Austin, TX, 2008.

228. Wakaya, F.; Katayama, K.; Gamo, K., Contact resistance of multiwall carbon nanotubes. Microelectronic Eng. 2003, 67-68, 853-857.

229. Decossas, S.; Cappello, G.; Poignant, G.; Patrone, L.; Bonnot, A. M.; Comin, F.; Chevrier, J., Interaction forces between carbon nanotubes and an AFM tip. Europhys. Lett. 2001, 53 (6), 742-748.

230. Cao, A.; Dickrell, P.; Sawyer, W.; Ghasemi-Nejhad, M.; Ajayan, P., Super- compressible foam-like carbon nanotube films. Science 2005, 310, 1307-1310.

231. Landau, L. D.; Lifshitz, E. M.; Kosevich, A. M.; Pitaevskii, L. P., Theory of elasticity. 3rd English ed.; Pergamon Press: Oxford Oxfordshire ; New York, 1986; p viii, 187 p.

232. Qian, L.; Tian, F.; Xiao, X., Tribological Properties of Self-Assembled Monolayers and Their Substrates Under Various Humid Environments. Tribol. Lett. 2003, 15 (3), 169-176.

233. Hawthorn, A. C.; Opell, B. D., Evolution of adhesive mechanisms in cribellar spider prey capture thread: evidence for van der Waals and hygroscopic forces. Biol. J. Linn. Soc. 2002, 77 (1), 1-8.

234. Hawthorn, A. C.; Opell, B. D., van der Waals and hygroscopic forces of adhesion generated by spider capture threads. J. Exp. Biol. 2003, 206 (22), 3905-3911.

235. Alibardi, L.; Toni, M.; Valle, L. D., Expression of beta-keratin mRNAs and proline uptake in epidermal cells of growing scales and pad lamellae of gecko lizards. J. Anat. 2007, 211 (1), 104-116.

236. Alibardi, L.; Toni, M., Distribution and characterization of proteins associated with cornification in the epidermis of gecko lizard. Tissue Cell 2005, 37 (6), 423-433.

175

237. Toni, M.; Dalla Valle, L.; Alibardi, L., The Epidermis of Scales in Gecko Lizards Contains Multiple Forms of β-Keratins Including Basic Glycine-Proline-Serine-Rich Proteins. J. Proteome Res. 2007, 6 (5), 1792-1805.

176