Research Collection

Doctoral Thesis

Surface modifications for improved aqueous lubrication under low-contact-pressure conditions

Author(s): Heeb, Raphael Emanuel

Publication Date: 2009

Permanent Link: https://doi.org/10.3929/ethz-a-006001889

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library DISS. ETH No. 18431

Surface Modiﬁcations for Improved Aqueous Lubrication under Low-Contact-Pressure Conditions

A dissertation submitted to ETH ZURICH

for the degree of Doctor of Sciences (Dr. sc. ETH Zurich)

presented by RAPHAEL EMANUEL HEEB Dipl. Werkstoﬀ-Ing. ETH born on January 16, 1980 citizen of Altst¨atten(SG)

accepted on the recommendation of Prof. Dr. Nicholas D. Spencer, examiner Prof. Dr. Seunghwan Lee, co-examiner Prof. Dr. Hugh Spikes, co-examiner

2009

FOR KATHRIN AND MY FAMILY

The scientist is not a person who gives the right answers, he’s one who asks the right questions.

CLAUDE LEVI-STRAUSS´

Abstract

The reduction of the interfacial friction between two surfaces in relative motion is a prerequisite for the proper functioning of many systems, ranging from machine parts to human joints. While the lubrication of two contacting surfaces primarily aims at the reduction of friction and wear, either by a separation of the surfaces by means of a fluid film or by introducing a layer of low shear strength between them, additional requirements such as the environmental compatibility of a lubricant or the energy efficiency of tribological systems have become important during recent years. From this perspective, the substitution of traditional oil-based lubricants with aqueous systems would be highly beneficial for specific tribological systems. Although water cannot significantly increase its inherently low viscosity under pressure, Nature demonstrates that aqueous lubrication is feasible and highly effective under certain circumstances. The objective of this thesis is to investigate surface modifications that promote the reduction of the macroscopic interfacial friction in aqueous environments. With the focus on mild contact-pressure conditions, it was feasible to investigate strongly attached self-assembled monolayers (SAMs) as well as polymer brushes with respect to their aqueous lubricating properties on a macroscopic scale. In the first part of the thesis, the basic surface-chemical as well as structural parameters that were considered important in aqueous lubrication, were investigated by means of different thiol SAMs on gold surfaces. The setup developed for tribological experiments consisted of a conventional pin-on-disk tribometer with a soft elastomeric rather than a rigid slider, which additionally allowed for the ex situ investigation of the tribologically stressed area by means of spectroscopic techniques. The macroscopic sliding friction as experienced from alkanethiol SAMs against an oxi-

vii viii dized and therefore hydrophilic poly(dimethyl siloxane) (ox-PDMS) slider revealed that surface-chemical as well as structural properties of the SAMs are parameters that largely influence the aqueous lubrication performance. The spectroscopic analysis of the sample areas that were exposed to tribological stress further revealed that the SAMs remain intact after pin-on-disk experiments. Based on these preliminary findings, the experimental work was extended in two different directions. Firstly, the aqueous lubrication performance of monolayers formed from hydrophilic PEG thiols with two different molecular weights was compared to that of previously employed alkanethiol SAMs. Subsequently, the solution parameters of the aqueous lubricants, such as pH and ionic strength, were examined with regard to their influence on the lubrication performance of the monolayers. Besides the tribological characterization, changes in conformation as well as in the hydration properties of the monolayers were investigated. These results revealed that hydrophilic, PEG-based monolayers with a high molecular weight serve as promising lubricant additives in various aqueous environments. The acquired knowledge about the parameters influencing the aqueous lubricating properties was also transferred to other technologically important systems. In an attempt to further enhance the lubricity of silicon oxide surfaces in aqueous media, high-surface-density poly(methacrylic acid) (PMAA) brushes were prepared by means of a novel “grafting from” approach. The utilization of an ultraviolet light emitting diode (UV-LED) for the surface-initiated photopolymerization led to high-molecular-weight polyelectrolyte brushes within relatively short irradiation times and rendered tedious cleaning steps unnecessary. Macroscopic pin-on-disk experiments involving PMAA brushes under mild contact-pressure conditions revealed undetectably low friction coefficients and very good long-term stabilities of the longer brushes. Together with the detailed analysis of bulk and surface properties of the employed PDMS tribopairs, this study has systematically investigated the aqueous lubrication performance of different surfaces under low contact-pressure conditions. It was demonstrated that the interfacial friction in an aqueous environment is dependent on surface-chemical and structural properties of both surfaces as well as on their interaction with the lubricant. Zusammenfassung

Die Minimierung der Grenzflächenreibung zwischen zwei sich kontaktierenden Ober- flächen in Relativbewegung ist eine Voraussetzung fürdas einwandfreie Funktion- ieren vieler Systeme, von Maschienenbauteilen bis hin zu menschlichen Gelenken. Währenddie Schmierung zum Ziel hat, die kontaktierenden Oberflächen mittels Schmierfilm oder einer geeigneten Schicht mit niedriger Scherfestigkeit zu separi- eren, sind in den letzten Jahren zusätzliche Anforderungen an Schmiermittel wichtig geworden, zum Beispiel deren Umweltverträglichkeit oder die Energieeffizienz des ganzen tribologischen Systems. Unter diesem Aspekt wäredie Substitution von tra- ditionellen, ölbasiertenSchmiermitteln mit wässrigenSystemen fürspezifische tri- bologische Systeme vorteilhaft. Obwohl Wasser seine geringe Viskositätunter Druck nicht massgeblich erhöhenkann, demonstriert die Natur, dass wässrigeSchmierung unter gewissen Umständensehr effizient ist. Das Ziel dieser Dissertation war die Untersuchung von Oberflächenmodifikationen zur Reduktion der makroskopischen Grenzflächenreibung in wässrigerUmgebung. Da der Fokus auf milde Kontaktdruckbedingungen gerichtet war, wurde die Unter- suchung der wässrigenSchmiereigenschaften von fest adsorbierten, sich selbst or- ganisierenden Monoschichten und bürstenähnlichen Polymeren auf makroskopischer Ebene möglich. Im ersten Teil dieser Dissertation wurden anhand monomolekularer Schichten von Thiolen auf Goldoberflächen diejenigen Parameter untersucht, welche fürdie wässrigeSchmierung als wichtig erachtet wurden. Die dafürentwickelte experi- mentelle Apparatur bestand aus einem herkömmlichen Stift-Scheibe-Tribometer mit einem elastischen anstatt eines harten Stifts, womit im Anschluss an die Reibungs- experimente die spektroskopische Charakterisierung der triologisch beanspruchten

ix x

Fläche möglich wurde. Aus der makroskopischen Gleitreibung zwischen Alkanthiol Monoschichten und oxidierten und deshalb hydrophilen Poly(dimethyl siloxan) (ox- PDMS) Oberflächen wurde deutlich, dass oberflächenchemische sowie strukturelle Eigenschaften der Monoschichten deren wässrigeSchmiereigenschaften massgeblich beeinflussen. Die spektroskopische Analyse der tribologisch beanspruchten Proben- flächen ergab, dass die Monoschichten nach dem Reibungsexperiment intakt blieben.

Basierend auf diesen Resultaten wurden die Experimente in zwei unterschiedliche Richtungen ausgedehnt. Zuerst wurden die Schmiereigenschaften von zwei hydrophilen Poly(ethylenglykol) (PEG) Monoschichten mit unterschiedlichem Moleku- largewicht mit denjenigen der Alkanthiolschichten verglichen und danach wurde untersucht, welchen Einfluss verschiedene wässrigeSchmiermittel auf die Schmier- leistung der verschiedenen Monoschichten haben. Neben der Charakterisierung der tribologischen Eigenschaften wurden die Monoschichten auch auf Unterschiede in deren Konformation sowie in deren Hydrierung untersucht. Diese Resultate haben aufgezeigt, dass hydrophile PEG-basierte Monoschichten mit einem hohen Moleku- largewicht vielversprechende Schmieradditive fürzahlreiche wässrigeUmgebungen darstellen. Im weiteren Verlauf dieser Dissertation wurden die gewonnenen Kenntnisse über die Parameter, welche die wässrigen Schmiereigenschaften beeinflussen, auf weitere technologisch wichtige Materialien übertragen. In einem Versuch, die Schmiereigen- schaften von Silikonoxidoberflächen in einer wässrigenUmgebung zu verbessern, wurden diese mittels Poly(methacrylsäure)Bürstenmit einer hohen Oberflächendich- te modifiziert, wobei eine neue Pfropfmethode angewendet wurde. Durch die Ver- wendung einer Ultraviolet Leuchtdiode (UV-LED) zur oberflächeninitiierten Pho- topolymerisation von Polyelektrolytbürstenwurden hohe Molekulargewichte mit relativ kurzen Belichtungszeiten erreicht. Zudem wurden langwierige Reinigungss- chritte nach der Polymerisation überflüssig. Makroskopische Reibungsexperimente mittels Stift-Scheibe-Tribometer mit Poly(methacrylsäure)Bürstenunter milden Kontaktdruckbedingungen haben zu unmessbar niedrigen Reibungskoeffizienten und zu sehr guter Langzeitstabilitätder langen Bürstengeführt. Zusammen mit einer detaillierten Analyse der Material- und Oberflächeneigen- schaften der verwendeten PDMS Proben, hat diese Arbeit die wässrigenSchmier- eigenschaften verschiedener Oberflächen unter kleinen Kontaktdruckbedingungen untersucht. Es wurde aufgezeigt, dass die Grenzflächenschmierung in wässriger Umgebung von oberflächenchemischen sowie strukturellen Eigenschaften beider Ober- flächen sowie von deren Wechselwirkungen mit dem Schmiermittel abhängigist. Contents

1 Introduction 1 1.1 Aim of the thesis ...... 1 1.2 Tribology ...... 2 1.2.1 Friction ...... 2 1.2.2 Lubrication ...... 3 1.2.3 Wear ...... 7 1.3 Self-assembled monolayers (SAMs) ...... 7 1.3.1 Fundamentals ...... 7 1.3.2 Self-assembled monolayers from thiols on gold ...... 8

1.3.3 Self-assembled monolayers from silanes on SiOx surfaces . . . 11 1.4 Polymer brushes ...... 13 1.4.1 Fundamentals ...... 13 1.4.2 Formation of polymer brushes ...... 14

2 Materials and Methods 17 2.1 Poly(dimethyl siloxane) (PDMS) ...... 17 2.1.1 Properties of PDMS ...... 18 2.1.2 Composition of silicone elastomer kit (Sylgard 184) ...... 19 2.1.3 Preparation of PDMS samples from the Sylgard 184 elastomer kit ...... 19 2.1.4 Bulk extraction of PDMS samples ...... 20 2.1.5 Hydrophilicity of PDMS ...... 21

2.2 SiO2 surfaces ...... 21 2.3 Au(111) surfaces ...... 22

xi xii CONTENTS

2.4 Pin-on-disk tribometry ...... 23 2.5 Polarization-modulation infrared reﬂection- absorption spectroscopy (PM-IRRAS) ...... 24 2.6 Variable angle spectroscopic ellipsometry ...... 25 2.7 Quartz-crystal microbalance with dissipation monitoring (QCM-D) . 26 2.8 Atomic force microscopy (AFM) ...... 27 2.9 Contact-angle measurements ...... 28 2.10 Ultraviolet-visible (UV/Vis) spectroscopy ...... 29 2.11 Surface-initiated graft-polymerization by means of ultraviolet (UV) irradiation ...... 30

3 Macroscopic Testing of Alkanethiol SAMs as Low-Contact-Pressure Aqueous Lubricant Additives 33 3.1 Introduction ...... 33 3.2 Experimental Procedures ...... 35 3.2.1 Self-assembled monolayers (SAMs) ...... 35 3.2.2 Poly(dimethyl siloxane)(PDMS) ...... 36 3.2.3 Water contact angle measurements ...... 37 3.2.4 Ellipsometry ...... 37 3.2.5 Polarization-modulation infrared reflection-absorption spectroscopy (PM-IRRAS) ...... 37 3.2.6 Atomic force microscopy (AFM) ...... 37 3.2.7 Pin-on-disk tribometry ...... 38 3.3 Results and Discussion ...... 39 3.3.1 Initial characterization of the SAM films ...... 39 3.3.2 Nanotribological properties of the SAM films by means of AFM 40 3.3.3 Elastomeric sliding contact on the SAM films on a macroscopic scale ...... 42 3.3.4 PM-IRRAS studies on the sliding track ...... 45 3.3.5 Lubrication mechanisms: The transition from mixed lubrication to soft EHL ...... 48 3.3.6 Lubrication mechanisms: the role of surface chemistry . . . . . 51 3.4 Conclusions ...... 51

4 Inﬂuence of Salt on the Aqueous Lubrication Properties of Ethylene Glycol-based Self-assembled Monolayers (SAMs) 53 4.1 Introduction ...... 53 4.2 Experimental Procedures ...... 56 CONTENTS xiii

4.2.1 Self-assembled monolayers (SAMs) on gold (Au) ...... 56 4.2.2 Poly(dimethyl siloxane) (PDMS) pins ...... 56 4.2.3 Pin-on-disk tribometry ...... 57 4.2.4 Contact angle measurements ...... 57 4.2.5 Variable angle spectroscopic ellipsometry (VASE) ...... 58 4.2.6 Polarization-modulation infrared reflection-absorption spectroscopy (PM-IRRAS) ...... 58 4.2.7 Quartz-crystal microbalance with dissipation monitoring (QCM- D) ...... 58 4.3 Results and Discussion ...... 59 4.3.1 Sample characterization prior to tribological experiments . . . 59 4.3.2 Aqueous lubricating properties in a low-salt environment (HEPES 0) ...... 64 4.3.3 Influence of high salt concentrations on the aqueous lubricating properties of the SAMs: HEPES 0, 1 M NaCl . . . . . 65 4.3.4 Influence of high salt concentrations on the conformation of surface-immobilized PEG SAMs ...... 67 4.4 Conclusion ...... 69

5 Inﬂuence of Solution pH on the Aqueous Lubrication Properties of Thiol Self-assembled Monolayers (SAMs) 71 5.1 Introduction ...... 71 5.2 Experimental Procedures ...... 72 5.2.1 Self-assembled monolayers (SAMs) on gold (Au) ...... 72 5.2.2 Poly(dimethyl siloxane) pins ...... 72 5.2.3 Pin-on-disk tribometry ...... 73 5.2.4 Water contact angle measurements ...... 73 5.2.5 Variable angle spectroscopic ellipsometry (VASE) ...... 73 5.2.6 Polarization-modulation infrared reﬂection-absorption spectroscopy (PM-IRRAS) ...... 74 5.2.7 Quartz-crystal microbalance with dissipation monitoring (QCM- D) ...... 74 5.3 Results and Discussion ...... 74 5.3.1 Sample characterization prior to tribological experiments . . . 74 5.3.2 Pin-on-disk tribometry ...... 75 5.3.3 Control experiments ...... 82 5.4 Conclusions ...... 86 xiv CONTENTS

6 Controlled Growth of Poly(methacrylic acid) (PMAA) Brushes from Silicon Surfaces via UV-LED-initiated Photopolymerization - Synthesis and Aqueous Lubrication Performance 89 6.1 Introduction ...... 89 6.2 Experimental Section ...... 92 6.2.1 Materials ...... 92 6.2.2 Synthesis of silanized photoiniferter, SBDC ...... 92 6.2.3 Ultraviolet-visible (UV-Vis) spectroscopy ...... 93

6.2.4 Vapor deposition of silanized photoiniferter onto Si/SiO2 . . . 93 6.2.5 Controlled radical photopolymerization by means of a UV-LED 94 6.2.6 Characterization of PMAA brushes ...... 95 6.3 Results and Discussion ...... 95 6.3.1 Ultraviolet-visible (UV-Vis) absorption of photoiniferter and monomer ...... 95 6.3.2 Characterization of the photoiniferter-modiﬁed substrates . . . 97 6.3.3 Photopolymerization of methacrylic acid to form poly(methacrylic acid) (PMAA) brushes ...... 97 6.3.4 Investigation of the in situ PMAA brush growth by means of QCM-D ...... 100 6.3.5 Hydrophilicity of PMAA brushes ...... 104 6.3.6 Macroscopic aqueous lubrication properties of PMAA brushes 104 6.4 Conclusions ...... 109

7 Inﬂuence of Bulk Properties of Poly(dimethyl siloxane) (PDMS) Tribopairs on the Aqueous Lubrication Performance 111 7.1 Introduction ...... 111 7.2 Experimental Procedures ...... 112 7.2.1 Preparation of PDMS samples with diﬀerent elasticities . . . . 112 7.2.2 Characterization of PDMS samples ...... 114 7.2.3 Pin-on-disk tribometry ...... 115 7.3 Results ...... 116 7.3.1 Elasticity of PDMS samples ...... 116 7.3.2 Macroscopic tribological properties of PDMS tribopairs . . . . 117 7.3.3 Considerations of the lubrication regimes involved ...... 123 7.3.4 Investigation of selected PDMS tribopairs after tribological stress ...... 126 7.4 Conclusions ...... 129 CONTENTS xv

8 Conclusions and Outlook 133 8.1 Conclusions ...... 133 8.2 Outlook ...... 136

References 139

Acknowledgments 155

Curriculum Vitae 157 xvi CONTENTS CHAPTER 1

Introduction

The first chapter aims to rationalize the work performed in this thesis and it attempts to briefly discuss the scientifically relevant fields, which are tribology and surface modification by means of self-assembled monolayers (SAMs) or polymer brushes.

1.1 Aim of the thesis

The lubrication of tribological contacts by means of a water-based fluid has attracted considerable attention during the last decade, because water represents an environmentally friendly and cost-effective alternative to traditional, oil-based lubricants [1–4]. The reason why aqueous lubricants currently play a minor role in engineering is mostly related to the inherently low viscosity of water as well as to its low pressure coefficient of viscosity, i.e. water cannot significantly increase its viscosity under load and is thus readily squeezed out of the contact area [5]. On the other hand, Nature employs aqueous lubricants very successfully, most prominently in mammalian synovial joints, which exhibit lubrication abilities that have never been achieved artificially. A closer look into this specific tribological system reveals that the bones are covered with articular cartilage - a highly compliant, soft, water- compatible, and complex material [6]. It is envisioned that, by partly mimicking Nature’s approach, it should be feasible to successfully design artificial tribological systems which are compatible with aqueous lubrication.

The aim of this thesis was to develop water-compatible surface modiﬁcations and to subsequently investigate their macroscopic aqueous lubrication performance

1 2 1.2. TRIBOLOGY under low contact pressures. By employing poly(dimethyl siloxane) (PDMS) as a model elastomer for one or both tribological contacts, the contact pressures can be maintained at a low level. In order to make the surfaces susceptible to aqueous lubrication, water has to be attracted to the surface and maintained within the contact area during tribological stress. Besides simple hydrophilization of the tribopair by means of oxidation techniques, which are known to be effective in enhanc- ing aqueous lubrication [3], more sophisticated surface modifications with enhanced water-immobilization capabilities have been explored, such as specific self-assembled monolayers (SAMs) or polymer brushes. In addition to the water-compatibility of the surface modifications, a specific focus was placed on the stability of the films under macroscopic sliding. To guarantee effective lubrication over a prolonged period, the generated films should have a strong affinity for the substrate surface and remain intact during tribostress. Ultimately, the tribological performance of the surface modifications were assessed in aqueous lubricants with different solution properties such as pH and salt concentration. It is hoped that this thesis will provide a set of criteria for the selection of aqueous lubricant additives and it attempts to elucidate the underlying lubricating mechanisms with complementary experimental techniques.

1.2 Tribology

The term tribology is derived from “τριβoς”, the Greek word for rubbing or attri- tion, and is deﬁned as the science and technology of interacting surfaces in relative motion. Tribology embraces friction, wear and lubrication phenomena, which have constituted the basis of innumerable engineering and everyday applications since the existence of humankind. Today, the costs of tribological deﬁciencies are estimated to exceed 1 % of the gross national product of industrial countries [7–9]. Thus, fundamental understanding of the interplay between friction, lubrication and wear processes is crucial for prolonged lifetimes of mechanical devices and ultimately for the reduction of energy consumption.

1.2.1 Friction

The friction force FR is deﬁned as the interfacial force that counteracts relative movement between two bodies. This is well illustrated in Figure 1.1, where an external force F is applied to a block that is in contact with a surface. As long as the applied force F does not exceed the static friction force FRS, the block does not move. If the external force F overcomes the static friction force, i.e. F > FRS, the 1.2. TRIBOLOGY 3 block starts to move in the direction of the applied force and the resulting friction force is then called dynamic FRD. N

F FR

Figure 1.1: The external force F applied to the block which is in contact with a surface, induces an interfacial friction force FR that counteracts relative movement between the two bodies.

The characteristic parameter that describes the frictional properties between two sliding bodies is the coefficient of friction µ, which is defined as the ratio between the friction force FR and the normal force N. For identical material pairings, the dynamic friction coefficient is usually lower than the static friction coefficient.

F µ = R (1.1) N The fundamental laws in tribology date back to Leonardo da Vinci (1452 - 1519) and they were rediscovered by Guillaume Amontons (1663 - 1705), who presented his observations to the Royal Academy of Sciences in Paris, where he formulated the two Amontons’ laws in 1699. In 1785, Charles Augustin de Coulomb discovered the third law, and all of them are of entirely empirical nature (Table 1.1):

Table 1.1: The three fundamental laws in tribology, formulated by Amontons (1. and 2. Law) and Coulomb (3. Law) 1. Law The friction force is proportional to the normal force 2. Law The friction force is independent of the apparent contact area 3. Law The friction force is independent of the sliding velocity

Surprisingly, these laws hold for many tribological contacts, both lubricated and unlubricated, but there are cases in which they become invalid, especially if polymers are utilized as tribological contacts.

1.2.2 Lubrication

In most engineering applications, lubricants are employed in order to introduce a layer of low shear strength between two sliding surfaces. While solid, liquid or 4 1.2. TRIBOLOGY gaseous lubricants are known to reduce the friction as well as wear between surfaces in relative motion, liquids represent the most common materials for lubrication purposes. Figure 1.2 shows the well-known Stribeck curve, which is of empirical nature and displays the variation of the friction coefficient as a function of the quantity ηv/N, where η is the lubricant viscosity, v is the sliding velocity and N is the normal load. From the Stribeck curve it is possible to identify three principal lubrication regimes, i.e. hydrodynamic (HL), elastohydrodynamic (EHL) and boundary (BL) lubrication. These lubrication regimes differ from each other essentially in terms of the lubricant film thickness which separates the sliding surfaces, and consequently in terms of the measured friction forces and the resulting wear. μ BL EHL HL

ηv/N

Figure 1.2: The empirical Stribeck curve describes the variation in the coeﬃcient of friction as a function of ηv/N for lubricated contacts.

Hydrodynamic lubrication (HL)

In this type of lubrication the sliding surfaces are separated by a relatively thick lubricant film in comparison to the asperity height, and the normal load is supported by the hydrodynamic pressure of the lubricant film. In order for the hydrodynamic lubrication to prevail, the contact between the two surfaces needs to be conformal, i.e. the two surfaces have to be (nearly) parallel over a relatively large area. This lubrication regime dominates when sufficiently viscous lubricants are employed under high-sliding-speed conditions and it is sometimes referred to as full-film lubrication. The resulting coefficients of friction are largely determined by the viscous drag of the entrained lubricant and, in the ideal case, wear should be negligible due to the absence of asperity contacts. 1.2. TRIBOLOGY 5

Elastohydrodynamic lubrication (EHL)

If the separating lubricant film between the two surfaces becomes thinner at high loads and/or low sliding speeds, the high contact pressures can cause elastic deformations of the bulk material. Hence, besides the increase in lubricant viscosity with pressure, elastic deformation of the contacts has to be considered in the EHL regime. High local contact pressures and thus considerable elastic deformations are frequently found for non-conforming contacts such as for sphere-on-plane geometries. Depending on the relative role of the two basic parameters in EHL, i.e. elastic deformation and increase in lubricant viscosity under pressure, Hamrock and Dowson have defined four different EHL regimes (Table 1.2) [10].

Table 1.2: The four elastohydrodynamic lubrication (EHL) regimes

EHL regime Elastic deformation Viscosity increase Isoviscous-rigid No No Piezoviscous-rigid No Yes Isoviscous-elastic Yes No Piezoviscous-elastic Yes Yes

The isoviscous-elastic regime usually dominates if at least one of the two contacts consists of a soft material, e.g. a rubber, and it is often referred to as soft EHL. Here, the material deforms elastically under the applied load while the resulting contact pressure remains rather low. Hence, the increase in the lubricant viscosity is negligible in soft EHL. Since only elastic deformations of the contacts have to be considered in this case, the soft EHL regime enables the utilization of low-viscous

ﬂuids as lubricants. With an empirical equation, the minimum ﬁlm thickness hmin for a sphere-on-plane geometry can be expressed as:

0.77 0.65 0−0.44 −0.21 hmin = 2.8 · R · (η · us) · E · w (1.2)

In Equation 1.2, R is the radius of the sphere, η is the viscosity of the lubricant, us is the sliding speed, E’ is the eﬀective elasticity modulus and N is the normal load. It is noted that, according to this equation, the predicted lubricant ﬁlm thickness is that of ideally smooth samples and apparently independent of the surface properties of the contacting materials. In order to take into account the surface roughness of the tribopair, it is useful to introduce the so-called Λ ratio:

Λ = hmin/σ (1.3)

In Equation 1.3, hmin is the minimum ﬁlm thickness from Equation 1.2 and 6 1.2. TRIBOLOGY

q pin disk σ = Ra + Ra is the combined surface roughness of the tribopair. The Λ ratio is often taken as a measure of the predominating lubrication regime for rough surfaces. For Λ ≥ 3, fluid-film lubrication is expected, whereas for 1 ≤ Λ ≤ 3 mixed lubrication is believed to prevail. In cases where the lubricant film thickness does not exceed the surface roughness of the tribopair, i.e. for Λ ≤ 1, boundary lubrication is expected.

Boundary lubrication (BL)

Under very severe sliding conditions, i.e. at high loads and/or low speeds, it is extremely difficult to maintain even thin lubricant films between the contacting surfaces. Thus, in the so-called boundary regime, direct contact between asperities is inevitable and usually accompanied by high friction and wear. Molecules that are able to protect the contacting surfaces under these circumstances are known as boundary lubricants. Suitable molecules should possess a high surface affinity and they ideally form a dense monomolecular layer to protect the surface, for instance a self-assembled monolayer or a polymer brush (Figure 1.3). Boundary lubricants prevent direct asperity contact and they ideally reduce the adhesion between asperities compared to that of the “unprotected” surfaces.

Figure 1.3: Boundary lubricants protect the tribologically stressed surfaces by preventing direct asperity contact in the absence of a fully separating lubricant ﬁlm.

Boundary lubrication typically becomes the predominant lubrication regime if liquids with a low pressure-coefficient of viscosity, such as water, are employed as base lubricants. The build-up of a separating fluid film between the contacts is impeded in this case, because these low-viscous fluids cannot readily support external loads without being squeezed out of the contact area. Here, boundary lubricants enable the generation of a protecting lubricant film. Under these circumstances, the inherent viscosity of the lubricant becomes secondary, because the low shear- strength interface mainly arises from the adsorbed boundary-lubricant layer on the surfaces. The base lubricant itself can contribute to the resulting friction by its incorporation into the boundary lubricant layer to give a “fluid-like” cushion layer. 1.3. SELF-ASSEMBLED MONOLAYERS (SAMS) 7

1.2.3 Wear

The occurrence of wear can be described as a progressive loss of material from a surface, caused by mechanical contact and relative movement of a second counterpart. In general, four wear mechanisms are distinguished; adhesive and abrasive wear, surface disruption as well as reactive wear [11]. The so-called Archard wear equation represents a useful means for relating the wear volume per sliding distance Q to the normal load W and the hardness H of the softer material.

KW Q = (1.4) H In Equation 1.4, K is called wear coefficient and is a dimensionless constant, which is often employed to characterize the severity of wear [9]. The hardness of the material is sometimes integrated into the wear coefficient, and hence, the constant is termed dimensional wear coefficient k=K/H.

1.3 Self-assembled monolayers (SAMs)

1.3.1 Fundamentals

By definition, self-assembled monolayers (SAMs) evolve from adsorption and spontaneous assembly of surfactants on solid surfaces [12]. In general, one functional group within the surface active molecule, often referred to as the “head-group”, has a high affinity for the surface, while the “backbone” extends away from it. The molecular assemblies generated in this way represent a facile tool for tailoring and controlling interfacial properties such as adhesion, wettability or friction. The design flexibility of surface-active molecules, for instance the introduction of functionalized terminating or “tail” groups, allows for the tailoring of surface properties on a molecular scale. Hence, depending on the chemical structure of the adsorbates as well as on the adsorption parameters, monolayers with distinct structural and chemical properties can be generated, as schematically depicted in Figure 1.4.

Figure 1.4: Possible structural conformations of self-assembled monolayers (SAMs) on a substrate surface. 8 1.3. SELF-ASSEMBLED MONOLAYERS (SAMS)

While the ﬁrst self-assembled monolayer system was observed for silanes on glass by Zisman in 1946 [13], the interest of the research community in SAMs only grew tremendously after SAMs from alkanethiols on gold surfaces were discovered by Nuzzo et al. in 1983 [14]. To date, alkanethiolate SAMs on gold surfaces represent the most studied self-assembly systems, while SAMs from silanes are of considerable technological importance, mainly because of their aﬃnity to oxide surfaces. In the following, the self-assembly systems employed in this thesis are discussed in more detail, namely SAMs from thiols on gold as well as from silanes on SiOx surfaces.

1.3.2 Self-assembled monolayers from thiols on gold

A large body of investigations in the ﬁeld of self-assembled monolayers (SAMs) has been carried out with organosulfur species on gold surfaces, among which alkanethiols on Au(111) are probably best characterized. Although organosulfur compounds are also known to react strongly with silver, copper, platinum and other substrates, gold represents the standard surface for thiol-based SAMs. The reason is partly of historical nature and partly due to the inertness of gold, i.e. it does not form a stable oxide layer and it is unreactive towards most chemicals [15]. The mechanism of thiol chemisorption on gold surfaces still remains incompletely understood but it can probably be described as an oxidative addition of the S-H bond to the surface from which hydrogen is eliminated by a reduction process [12]:

1 R − S − H + Au0 → R − S−Au+ · Au0 + H (1.5) n n 2 2 The thiolate-Au bond is of very strong nature (homolytic bond strength ≈ 40 kcal/mol) [16] and thiol SAMs are thus immobilized on gold surfaces in a quasi- covalent manner. While the chemisorption of thiols can be achieved from the gas phase or from solution, the latter method is straightforward and SAMs are conveniently achieved by immersing the substrates into dilute solutions of thiols in an appropriate solvent. The most widely employed solvent for the preparation of thiol SAMs is ethanol, mainly because a wide range of thiols are suﬃciently soluble as ethanolic solutions, because high-purity ethanol is available at low cost and due to its low toxicity [15].

Alkanethiols

The adsorption of alkanethiols on Au(111) surfaces from ≈ 1 mM ethanol solutions was shown to consist of two basic steps [17]: The ﬁrst step is characterized by a very fast initial adsorption, which occurs within seconds to minutes and accounts for 80 1.3. SELF-ASSEMBLED MONOLAYERS (SAMS) 9

- 90% of the final film thickness. The second step in the SAM formation involves a much slower incorporation of additional molecules as well as the ordering of alkyl chains to form a two-dimensional crystal. This step was shown to evolve over several days in order to minimize pinholes as well as conformational defects [15, 18]. On polycrystalline gold surfaces, which predominantly expose the (111) crystal face, the sulfur atoms in the n-alkanethiol adsorbates are found to form a √ √ ( 3x 3)R30◦ overlayer [16]. Specific ordering of the SAMs is a consequence of minimizing the free energy of the adsorbate layer by maximizing the attractive (e.g. van der Waals) interactions between adjacent carbon backbones [15]. Thus, alkanethiol SAMs adopt an all-trans conformation on gold surfaces, with a tilt angle α of nearly 30◦ with respect to the surface normal. An additional rotation about the backbone axis of approximately 50◦ (β) allows for maximum attractive van der Waals interactions of adjacent backbones, as illustrated in Figure 1.5 [15].

Figure 1.5: Conformation of a single chain inside an alkanethiol SAM on a Au(111) surface.

Clearly, the ordering of the SAMs is largely dependent on the chain length of the backbone. It is generally found that alkanethiols with a backbone length below

8 CH2 units form more disordered monolayers due to an insuﬃcient number of attractive van der Waals interactions of adjacent molecules [19]. Additionally, grain boundaries as well as point defects present in polycrystalline substrate materials represent the limiting factor in the ordering of SAMs.

Ethylene glycol (EG)-based thiols

Monolayers of end-grafted poly(ethylene glycol) (PEG) have been shown to exhibit excellent resistance against unspecific protein adsorption on a variety of surfaces. Consequently, molecular layers containing EG functionalities have been widely used as model substrates in biomedical engineering [20–22]. In tribology, PEG-based 10 1.3. SELF-ASSEMBLED MONOLAYERS (SAMS) polymers were found to serve as efficient boundary lubricant additives in aqueous environments [23]. Hence, the investigation of SAMs containing EG moieties has been motivated from several perspectives. The generation of monolayers containing ethylene glycol units can be achieved by different means. On gold surfaces, these molecular layers are most conveniently obtained via thiol functional groups. In order to achieve very dense EG monolayers, the latter units can be linked to alkanethiol backbones, i.e. as functional “tail” groups. While the alkanethiols form a well-ordered SAM, the ethylene glycol units are forced to adopt ordered configurations, accordingly. Another possibility to obtain ethylene glycol-based surface coatings on gold is achieved by the direct adsorption of thiolated oligo- or poly(ethylene glycol) molecules from suitable solvents. The difference in the chemical nature of the EG backbone −(CH2 − CH2 − O)n− compared to alkyl chains −(CH2)n− changes the requirements for a suitable adsorption medium. EG-based thiols are frequently adsorbed from aqueous solvents because of the water solubility of ethylene glycol. In comparison to the all-trans conformation of alkanethiol SAMs, monolayers of oligo(ethylene glycol) (OEG) on gold were found to adopt a helical configuration, which mainly arises from intramolecular hydrogen bonding between water molecules and EG units. In contrast, the all-trans conformation of alkanethiol SAMs stems from intermolecular van der Waals interactions and they are known to possess a higher degree of order compared to OEG SAMs.

“good solvent” “poor solvent” O

O O O O O

O O O O O O

O O O O O

O O O O O O

O O O O O O O O O O O O O O O O O O

O O O

O O O O O O O O

O O O O

O O O O O O

O O

O O O O

O O

O O O O O O O

O O

O O O

O O O O O O O O O O

O O O

O O O O O O

O O O O

O O O O O O O O O O

O O O O O O O O

O O O O O

O O O O O O O O O

O O O O O O O O O O O

Figure 1.6: Diﬀerences in adsorption densities of EG-based SAMs adsorbed from “good” and “poor” solvents.

The adsorption of polymers such as PEG onto a substrate surface generally leads to much less ordered and thus less dense monolayers, mainly because of steric repulsion between the polymer chains. In order to obtain higher surface-grafting densities, PEG molecules are sometimes adsorbed close to cloud-point conditions. Clouding is observed under conditions where the PEG molecules become insoluble, for instance in aqueous solutions of high ionic strength and/or at elevated temperatures. The less solvated and therefore more compact molecules can then adsorb on the surface in a higher density. Conversely, the large radius of gyration of highly hy- 1.3. SELF-ASSEMBLED MONOLAYERS (SAMS) 11 drated molecules adsorbed on a surface prevents other molecules from adsorbing by blocking potential adsorption sites around them, as illustrated in Figure 1.6. PEG monolayers adsorbed from a poor solvent will become solvated under good solvent conditions, i.e. they will extend their chain dimensions by stretching away from the substrate surface. When adsorbed from a good solvent, PEG chains will not extend their dimensions any further, and stretching away from the surface is only observed if the adsorption density is suﬃciently high.

Mixtures of thiols

The coadsorption from solutions containing a mixture of two different thiols represents a straightforward possibility to introduce various functional groups on the surface as well as to manipulate the order in the SAM. The combination of thiol molecules in the mixture can be manifold, i.e. identical or distinct backbone chemistry and chain length, equal or different tail groups. Another important parameter in the preparation of “mixed” SAMs is the selection of the solvent, especially if the thiol mixture contains a polar and a nonpolar compound. From a solution with identical molar fractions of two thiols, the one with the lower solubility in the solvent will represent a higher mole fraction in the final monolayer. Thus, the composition of the monolayer can be adjusted by the molar fractions of each component in the solvent but the latter does not necessarily reflect the molar fraction of the final SAM. Another thermodynamic parameter that influences the SAM composition of two-component systems is the backbone chain length, in that longer chains exhibit a higher stability on the surface due to increased van der Waals interactions to neighboring chains [24]. On the other hand, the adsorption of short-chain thiols is kinetically favored and shorter molecules possess a higher mobility, which is important in ordering of the SAMs. It is believed that “mixed” SAMs can show some phase separation, hence no macroscopic domains have been observed for mixed alkanethiol SAMs of different chain lengths or terminating groups, respectively [24, 25].

1.3.3 Self-assembled monolayers from silanes on SiOx surfaces

Monolayers of organosilicon molecules on oxide surfaces have reached considerable technological importance during the last decades. So-called “silane coupling agents” have found widespread use as adhesion promoters [26, 27]. In contrast to thiol SAMs, the formation of silane monolayers requires an oxide surface with the most prominent substrates being silicon oxide, glass, aluminum oxide, quartz or mica [12]. 12 1.3. SELF-ASSEMBLED MONOLAYERS (SAMS)

Adsorption of silane SAMs - methods and mechanism

The formation of silane SAMs on oxide surfaces can be achieved by adsorption from a liquid or from the vapor phase. For the generation of silane SAMs from solution, there are a number of parameters that need to be controlled, including the type of solvent, water concentration, pH and temperature [28]. From this viewpoint, vapor- phase adsorption appears to be more convenient. However, the adsorption kinetics from the vapor phase might be slower compared to adsorption from solution and not all molecules are eligible for vapor-phase adsorption, especially if they have a high molecular weight. Such molecules are not conveniently vaporized in a suﬃciently high concentration, which can result in incomplete monolayers. Yet, for molecules that are suitable for vapor-phase adsorption, this technique is highly attractive since there are fewer parameters to control during the adsorption and the reported SAMs appear to be well structured and dense [29]. Figure 1.7 illustrates the basic steps involved in the adsorption of alkoxysilanes - together with chlorosilanes the most commonly employed organosilicon derivatives. The ﬁrst step involves the hydrolysis of alkoxy groups (most often trimethoxy- or triethoxy-) to silanols in the presence of water. This water may either come from the surface-bound hydration layer, from the atmosphere or it may be necessary to add water in some cases, to increase the degree of hydrolysis [30]. The subsequent formation of hydrogen bonds with surface silanol groups (step 2 in Figure 1.7) is followed by a condensation reaction in which water is liberated to form covalent siloxane bonds to the substrates. In the case of trimethoxy- or trichloro-terminated silanes, lateral condensation of silanol groups with neighboring chains, i.e. step 4, generates a SAM with a unique stability. Due to steric restrictions it is not possible for all three silanol groups to bind to the surface [31].

R R R R R R R

1 2 3 4

+ H O - H O 2 2

CH O Si OCH HO Si OH 3 3 HO Si OH

OCH OH HO Si OH

3 O Si O Si O Si

H H H H H H H H H O O O O O O O O O O O O O

Figure 1.7: Diﬀerent adsorption steps in the SAM formation from alkoxysilanes on oxide surfaces. 1.4. POLYMER BRUSHES 13

Conformation of silane SAMs

For alkoxy- or chlorosilanes with a trifunctional headgroup, as depicted in Figure 1.7, there is a tendency towards the formation of three-dimensional structures, since all three silanols can condense with hydroxyl moieties. This could prevent the formation of a full monolayer, i.e. partial coverage or multilayers are feasible [32]. It is therefore believed that the water content in the system is of fundamental importance. An overabundance of water was found to induce polycondensation of silane molecules prior to adsorption, whereas a lack of water is known to result in incomplete monolayers [33]. Silane SAMs are thus more prone to defects in comparison to thiol monolayers, unless all adsorption parameters are carefully controlled. A further source for defects inside the monolayer is the low mobility of covalently immobilized silanes, which also leads to the picture of silane SAM formation from isolated islands to which single or groups of molecules couple via condensation of free silanols [12].

1.4 Polymer brushes

1.4.1 Fundamentals

The term “polymer brush” refers to the conformation of densely adsorbed, end- tethered polymer chains on a substrate surface. If the distance d between attachment points becomes smaller than the dimension of unperturbed polymer chains, i.e. smaller than the radius of gyration Rg, the resulting monolayers leave the “mushroom” regime and enter the “brush” regime, where chains stretch away from the interface (Figure 1.8). The formation of a polymer brush does largely depend on the type of solvent. That is, the maximization of polymer-solvent interactions en- hances the brush formation in good solvents, whereas in poor solvents, the chains collapse due to more favorable segment-segment interactions between them. The formation of polymer brushes on a surface can be described by a simple thermodynamic model developed by Alexander and de Gennes [34–36]. Dense packing of undeformed polymer coils on a substrate surface with coil-to-coil distances d smaller than Rg forces the coils to overlap and thus increases the chains’ interaction energy Fint due to contacts between chain segments. In order to avoid such segment-segment interactions, the chains stretch away from the surface, increasing the polymer layer-thickness h. On the other hand, the loss of entropy due to chain stretching increases the elastic free energy Fel of the chains. Hence, the resulting equilibrium thickness of the brush is a balance between the two free-energy contributions Fint and Fel. Minimization of the total free energy with respect to the brush 14 1.4. POLYMER BRUSHES

“mushroom” regime “brush” regime

Rg h

d < R d ≥ R g g

Figure 1.8: Polymer chains attached to an interface adopt different conformations that depend on the distance d between attachment sites. For d ≥ Rg, the monolayer is in the “mushroom” regime, compared to the “brush” regime, where chains adopt a stretched conformation perpendicular to the surface (for d < Rg). height h, results in linear scaling of the latter with the chain length N, regardless of the solvent quality. That is, h ∼ N · σ1/3 in good solvents and h ∼ N · σ1/2 in poor solvents, with σ being the surface coverage. This result is in contrast to the 0.59 radius of gyration of polymer solutions, in which Rg ∼ N for good solvents and 0.50 Rg ∼ N for theta solvents [37, 38]. This theoretical model demonstrates that densely attached polymer chains deform under equilibrium conditions and the thickness of the polymer brush changes linearly with chain length, regardless of the solvent quality. It is noted that this simplified theory does not allow the investigation of more complex properties such as chain conformations or monomer density profiles in detail, but it gives a rough measure of the basic features of polymer brushes [39].

1.4.2 Formation of polymer brushes

In principle, there are two distinct strategies to prepare polymer brushes on a substrate. The ﬁrst method is known as “grafting to” and refers to the adsorption of preformed polymer chains with functional terminating groups. With the second approach, also known as “grafting from”, polymer brushes are prepared by an in situ polymerization starting from the substrate surface. The two methods are schematically illustrated in Figure 1.9, since both approaches were employed to generate polymer brushes in this thesis. In the following, the characteristics of both techniques are discussed separately. 1.4. POLYMER BRUSHES 15

“Grafting to” “Grafting from”

Figure 1.9: “Grafting to” refers to a method in which polymer brushes are generated by the adsorption of preformed polymer molecules. “grafting from” represents an in situ polymerization approach to form polymer brushes .

“Grafting to”

The preparation of polymer brushes via the “grafting to” approach is straightforward and most conveniently achieved by solution adsorption of preformed polymer chains onto a substrate surface. This method is closely related to the generation of self- assembled monolayers (SAMs) on a surface. The possibility to graft polymers with a low polydispersity is a distinct advantage of the “grafting to” approach, however the maximum layer thickness of the resulting brush is rather limited [40]. Initially, the polymer chains adsorb far apart from each other, i.e. at distances d >> Rg, for entropic and energetic reasons. As a certain surface coverage is reached, further polymer chains have to diffuse through the existing polymer layer in order to reach the surface, and hence, against a concentration gradient. This adsorption mechanism has both kinetic and thermodynamic constraints and, depending on the solvent, relatively low grafting densities are achieved with the “grafting to” method [41]. A further drawback is the competition of certain backbone moieties with the terminal functional groups for surface attachment sites. If the polymer chains have a high affinity for the substrate surface, they might block several attachment sites so that the terminal functional groups cannot chemisorb onto the surface, which results in incomplete and only poorly defined polymer films. The selection of suitable solvents can partly overcome this problem but generally, there are several limitations in the generation of polymer brushes via the “grafting to” method.

“Grafting from”

The main characteristic of the “grafting from” approach is that the polymerization takes place in situ, hence the brush formation requires more experimental eﬀort. 16 1.4. POLYMER BRUSHES

In general, the “grafting from” method employs a surface-immobilized initiator, which, in the presence of suitable monomers, induces the polymerization starting from the substrate surface. Since the initiating species are usually low-molecular- weight compounds, they can be adsorbed in a high density on a substrate surface via self-assembly. Thus, it is feasible to grow very dense and chemically more versatile brushes with the “grafting from” approach. Although very long brushes with a dry thickness varying from nm to µm can be obtained with appropriate methods, the polydispersity is usually higher and more difficult to control compared to brushes formed with the “grafting to” approach. However, the possibility to create patterns of polymer brushes by a local stimuli that initiates polymerization as well as the high brush densities that can be achieved make the “grafting from” approach very attractive for the preparation of dense polymer monolayers on a variety of substrates. In recent years, numerous synthetic strategies have been developed in order to generate polymer brushes via “grafting from”. The techniques include thermal, radical or ionic polymerization and can be further classified according to their initiation mechanisms or the type of initiating species involved [41]. In this thesis, polymer brushes were generated by means of radical polymerization of suitable monomers from surface-immobilized initiating species, which were activated by UV irradiation. More details about the specific “grafting from” approaches are given in subsequent chapters. CHAPTER 2

Materials and Methods

This section first describes the materials that were employed as substrates for tribological contacts. The basic properties of the materials utilized for surface modification of the substrates have partly been mentioned in the Introduction, but they will be explicitly described in the corresponding chapters. Furthermore, the main characteristics of the experimental techniques utilized in this thesis are briefly described, while the detailed measurement procedures are given in the individual chapters.

2.1 Poly(dimethyl siloxane) (PDMS)

Crosslinked poly(dimethyl siloxane) (PDMS) served as the model elastomer throughout the thesis, i.e. it was always utilized as one or both tribological contacts to maintain low contact pressures. Native poly(dimethyl siloxane) (PDMS) represents a polymer with an inorganic backbone consisting of alternating silicon and oxygen atoms, with each silicon bearing two methyl (-CH3) groups (Figure 2.1).

CH3 CH3

H3C Si O Si CH3

CH3 CH3 n

Figure 2.1: Chemical structure of poly(dimethyl siloxane) (PDMS)

17 18 2.1. POLY(DIMETHYL SILOXANE) (PDMS)

The introduction of functional groups into PDMS, which partly replace the methyl moieties, as well as the utilization of other functional siloxanes enables the formation of a silicone elastomer via chemical crosslinking. While the network formation can be achieved by diﬀerent means, so-called addition-cure systems, which utilize platinum-catalyzed hydrosilylation reactions, are known to result in the most stable chemical crosslinks and the formation of volatile products is limited [42]. In hydrosilylation reactions, silicon hydride (Si-H) species of one chain add to unsaturated carbon double bonds (Si-C=C) of another chain with the formation of a carbon-carbon single bond, as schematically shown in Figure 2.2.

O O O O

[Pt] H2 H2 H3C Si H + Si CH3 H3C Si C C Si CH3

O H2C O O O

Figure 2.2: Hydrosilylation reaction between two poly(dimethyl siloxane) (PDMS) chains bearing Si-H or Si-C=C functional groups.

2.1.1 Properties of PDMS

Since crosslinked PDMS elastomers exhibit inherently weak mechanical properties, reinforcing ﬁllers such as silica are most often incorporated in commercial silicone rubbers [43, 44]. Despite their inferior mechanical properties compared to other elastomeric materials, PDMS has been widely employed in research as well as in numerous applications in recent years [45–47].

Table 2.1: Characteristic properties of poly(dimethyl siloxane) (PDMS)

Properties of PDMS elastomers High elasticity Low free surface energy Physiological inertness Good thermal and oxidative stability High electrical resistance

One key property of PDMS elastomers is their ease of fabrication, which contributed to their use in microﬂuidics and microcontact printing [2, 46, 48–53]. Prior to crosslinking, PDMS can be cast into suitable molds of almost any desirable shape. Other attractive properties of PDMS are its physiological inertness as well as its good thermal and oxidative stability (Table 2.1). 2.1. POLY(DIMETHYL SILOXANE) (PDMS) 19

2.1.2 Composition of silicone elastomer kit (Sylgard 184)

Throughout this thesis, PDMS elastomers were fabricated from a two-part liquid component silicone kit (Sylgard 184, Dow Corning Corporation, Midland MI, USA). The latter comprises a base (part A) as well as a curing agent (part B), whose components are shown in Table 2.2 and Figure 2.3, respectively [54, 55]. It is apparent that Sylgard 184 consists of numerous compounds, which contribute to the formation of the PDMS elastomer with its distinctive properties. The unsaturated moieties of vinyl-terminated poly(dimethyl siloxane) (1 in Figure 2.3) undergo hydrosilylation reactions with the methylhydrosiloxane moieties of compound 2 in the presence of the platinum catalyst (4), as depicted in Figure 2.2. The Si-H species of 2 are further capable of forming crosslinks with unsaturated carbon moieties of compounds 3, 5 and 7. While 3 and 7 are employed as reinforcing materials (i.e. fillers) to significantly improve the mechanical properties of the final elastomer, tetramethyltetravinylcyclotetrasiloxane (5) acts as a modulator, which inhibits fast crosslinking at room temperature and thus increases potlife. Other constituents of Sylgard 184 include low-molecular weight siloxanes (6) as well as small quantities of volatile cyclic compounds such as xylene (8) and ethylbenzene (9).

Table 2.2: Composition of Sylgard 184, a two-component silicone elastomer kit Compound Part A Part B 1 Vinyl-terminated poly(dimethyl siloxane) X X 2 Methylhydrosiloxane-, dimethylsiloxane polymer X 3 Poly(dimethyl siloxane) micronetwork (VQM 1) X 4 Platinum catalyst X 5 2,4,6,8-Tetramethyltetravinylcyclotetrasiloxane X 6 Tetrakis(trimethylsiloxy)silane X 7 Dimethylvinylated, trimethylated silica ﬁller X X 8 Xylene X X 9 Ethylbenzene X X

2.1.3 Preparation of PDMS samples from the Sylgard 184 elastomer kit

The standard PDMS sample preparation in this thesis was performed according to the instruction manual of the manufacturer.The base and and the curing agent of SYLGARD 184 were thoroughly mixed at a 10:1 ratio by weight. The foams generated during mixing were removed by a gentle vacuum before the mixture was transferred into the master and incubated in an oven (70 °C) overnight. In order to 20 2.1. POLY(DIMETHYL SILOXANE) (PDMS)

Si O H2 H2 Si 1 H2CC C Si O Si C C CH2 6 O Si O H n 2n H 434 Si O Si

2 Si O Si O Si H m n 7 dimethylvinylated and trimethylated silica Si O Si O Si 3 O O O Si O Si H SiH2 O 8 Si

4 [Pt] 9

Si O 5 O Si Si O O Si

Figure 2.3: Chemical structure of the compounds in Sylgard 184 silicone elastomer kit.

prepare PDMS pins with hemispherical ends, a commercial polystyrene cell-culture plate with round-shaped wells (radius = 3 mm, 96 MicroWell Plates , NUNCLON Delta Surface, Denmark) was employed as master. PDMS disks were fabricated by casting the pre-mixed polymer solution into aluminum plates with sphere-shaped wells, whereas the side exposed to the ambient during curing was employed for tribological investigations.

2.1.4 Bulk extraction of PDMS samples

As previously mentioned, PDMS elastomers fabricated from a Sylgard 184 elastomer kit contain considerable amounts of reinforcing ﬁllers as well as some weight percent of low-molecular-weight compounds (Table 2.2 and Figure 2.3). Hence, regardless of the completion of the crosslinking reactions, there will always be uncrosslinked species with a certain mobility inside the network. Since nonpolar solvents are capable of swelling PDMS, it is possible to extract volatile and uncrosslinked low molecular weight species from the bulk [2, 56]. 2.2. SIO2 SURFACES 21

The extraction procedure, which was employed in this thesis comprised the immersion of the PDMS samples into n-hexane for 24 hours. During this period, the solvent was exchanged twice. In order to dry the swollen PDMS samples, they were evacuated in a desiccator for 60 minutes and subsequently stored in ambient for 24 hours before use.

2.1.5 Hydrophilicity of PDMS

Several techniques have been suggested in literature to oxidize the surface of PDMS samples and thus render it hydrophilic (i.e. wettable by water). Among them are radio-frequency (RF) plasma and UV/ozone treatments or exposure to UV irradiation [48, 57–59]. While only air-plasma treatments were carried out for the oxidation of PDMS surfaces throughout this thesis, it is generally agreed that all the mentioned techniques lead to the generation of a silica-like (SiOx) surface layer. Depending on the oxidation parameters, the generated oxide layer can be extremely brittle, i.e. cracks are readily formed as a result of either internal stresses or small external mechanical deformation [60]. It is well known that the hydrophilicity of PDMS is short-lived in ambient, i.e. hydrophobic recovery takes place within hours to days, depending on the parameters of initial surface treatment [47, 56, 60–62]. The mechanism of hydrophobic recovery is still the subject of debate but it is likely to have diﬀerent contributions. Many authors ascribe the loss of surface hydrophilicity to the diﬀusion of low-molecular weight compounds from the bulk to the surface. These uncrosslinked species are either inherently present in PDMS [56] or produced via chain scission during oxidation [61] and they are believed to migrate to the surface via cracks in the silica-like surface layer. Other possible contributions to the hydrophobic recovery of oxidized PDMS surfaces include reorientation of polar groups into the bulk or the condensation of silanols on the surface [59, 63, 64]. The standard procedure for the hydrophilization of PDMS samples in this thesis was performed by means of a RF air-plasma treatment (high power, Harrick Plasma Cleaner/Sterilizer, Ossining, NY, USA) for 60 s at a pressure of 0.1 torr.

2.2 SiO2 surfaces

The silicon substrates employed in this thesis are of the Si (100) type (P/Bor <100>, Si-Mat Silicium Wafers, Germany) and contain a native oxide layer due to the storage under atmospheric conditions. The oxygen in the crystalline SiO2 surfaces occupies a bridging position between two Si atoms and each silicon adopts a tetrahedral 22 2.3. AU(111) SURFACES

conﬁguration with four oxygen atoms in the bulk SiO2 [65]. Due to the omnipresence of water in the atmosphere, silicon oxide is readily hydrolyzed, as shown in Equation 2.1:

Si − O − Si + H2O → 2Si − OH (2.1)

The hydroxylated silica surface displays a highly energetic and thus reactive character. Depending on the type of surface silanols and their lateral distribution,

SiO2 surfaces frequently hold a monolayer of hydrogen-bonded water under ambient conditions [66]. In the absence of surface-bound water, the surface silanols can undergo condensation reactions with neighboring silanols or other molecules. Silicon wafers were employed as substrates for UV-induced surface graft polymerization, since silica serves as an ideal substrate for surface modifications. Firstly, because the native oxide layer on silicon surfaces exhibits a high density of surface silanols, which can serve as covalent linkers, and secondly because the oxidation of PDMS also leads to SiOx surfaces, which are similar to the native oxide layer of silicon. Hence, the surface modifications performed on silicon oxide should also be plausible on oxidized PDMS, while the characterization of thin films on silicon wafers is more convenient.

2.3 Au(111) surfaces

The gold samples in this thesis were employed as 100 nm thick surface coatings on top of soda-lime glass microscope slides (SuperFrost, Menzel-Gläser,Braunschweig, Germany). After cutting the glass substrates into 2.5 x 2.5 cm pieces, they were cleaned by means of ultrasonication in ethanol (4 x 15 minutes, solvent exchanged each time) and subsequent oxygen-plasma-treatment (2 minutes, high power, Har- rick Plasma Cleaner/Sterilizer, Ossining, NY, USA). The clean glass substrates where pre-coated with an adhesive layer of 10 nm of chromium. Polycrystalline Au(111) layers were thermally evaporated from an Au wire (99.99 %, ø 2 mm, Umi- core Materials AG, Liechtenstein) and used as substrates for the generation of thiol self-assembled monolayers (SAMs). All the SAM films employed in this thesis were generated by immersion of the substrates in ethanolic thiol solutions for > 12 h. Prior to the immersion step, the gold substrates were rinsed with pure ethanol and subsequently dried with nitrogen. The high affinity towards sulfur as well as its inertness make gold an ideal substrate, since it does not react with atmospheric oxygen and most other chemicals. The popularity of gold surfaces for the generation of thiol SAMs also stems from 2.4. PIN-ON-DISK TRIBOMETRY 23 the fact that it is easy to generate thin films on various substrates and gold surfaces are conveniently characterized by a number of surface-analytical tools such as surface-sensitive infrared spectroscopy, ellipsometry or quartz crystal microbalance. And hence, SAMs on gold are historically most studied [15].

2.4 Pin-on-disk tribometry

The pin-on-disk tribometer (CSM Instruments SA, Peseux, Switzerland) employed in this thesis represents a macroscopic technique to measure the sliding friction for sphere-on-plane conﬁgurations. The characteristic parts of the instrumental setup are illustrated in Figure 2.4. The pin-bearing measuring arm is loaded by dead weights in order to apply a normal force (N) and it is brought into contact with the disk, which is mounted inside a cup holder. The latter is driven by a motor and the disk inside the cup holder is rotated at a certain velocity. The interfacial friction forces generated upon contact between the pin and the rotating disk are determined via the lateral deﬂection of the measuring arm, which is equipped with a strain gauge. N

F R

Figure 2.4: The central parts of a pin-on-disk setup (left) and the top view of the rotating disk with the relevant forces indicated (right). Interfacial friction forces between pin and disk are detected by the lateral deﬂection of a strain gauge in the measuring arm.

A distinct advantage of a pin-on-disk tribometer is that a wide range of materials can be readily employed as samples since there is nearly no restriction on the shape and the mechanical properties of the samples. The available sliding speed range, typically from mm/s to cm/s, allows for the investigations of several lubrication regimes ranging from ﬂuid ﬁlm to boundary lubrication. Due to the relatively large contact areas encountered in pin-on-disk experiments, it is feasible to perform an ex situ spectroscopic analysis of the sliding track. However, the multiple asperity 2.5. POLARIZATION-MODULATION INFRARED REFLECTION- 24 ABSORPTION SPECTROSCOPY (PM-IRRAS) contacts encountered in all macroscopic approaches lead to very high and uncon- trollable local contact pressure. Hence, pin-on-disk tribometry makes high demands on the stability of surface coatings.

2.5 Polarization-modulation infrared reﬂection- absorption spectroscopy (PM-IRRAS)

Infrared reflection-absorption spectroscopy (IRRAS) is a very sensitive and at best semi-quantitative technique for the investigation of organic thin films on metal surfaces. Infrared spectroscopy is based on the interaction of infrared light (800 - 4000 cm−1) with molecules. By passing infrared light through a sample, the absorbed irradiation at a particular energy induces molecular vibrations. For such infrared absorption to occur, the molecule must change its electric dipole moment via stretching, bending or twisting of chemical bonds. These vibrations are detected at specific energies and represent the essential information that can be obtained from infrared spectroscopy [67]. In the case of IRRAS, p-polarized infrared light is reflected at the sample surface and analyzed by a detector. Since only the normal component of the electric field vector Ez is eligible for interactions with adsorbed species, a grazing angle incidence is necessary for sufficient measurement intensity. Consequently, only molecular vibrations with a dipole moment normal to the sample surface are excited, which is also known as the surface selection rule. The use of a photoelastic frequency modulator (PM) brings about the advantage that no reference measurement is required and thus rules out atmospheric contamination issues, a frequently observed problem on reference samples. Because s-polarized light does not interact with ultrathin layers on metal substrates, PM-IRRAS uses it for the background measurement while the p-polarized light is simultaneously probing the thin film properties of the adlayer. Advantages of the PM-IRRAS technique include the high surface sensitivity without the necessity for ultra-high vacuum conditions and the possibility to collect detailed information about the chemical identity and the orientation of molecules in the sub-monolayer range. Restrictions of surface-sensitive infrared spectroscopy are the difficulties in quantitative analysis as well as the limited lateral resolution due to an increased spot size at grazing angle incidence. The PM-IRRAS experiments in this thesis were performed with a Bruker IFS 66v IR spectrometer (Bruker Optics, Germany) with a photoelastic modulator (PMA 37, Bruker Optics, Germany). The interferogram from the external beam port was polarized with a KRS-5 wire-grid polarizer and a photoelastic modulator. Thereafter, the incident beam was reflected 2.6. VARIABLE ANGLE SPECTROSCOPIC ELLIPSOMETRY 25

at the sample surface at an angle of 80° and detected with an N2(l)-cooled MCT detector. For all measurements, 1024 scans were collected with a 8 cm−1 resolution. The multiplexed interferograms were processed and baseline-corrected with a polynomial using the instrument software (OPUS, Bruker Optics, Germany).

2.6 Variable angle spectroscopic ellipsometry

Ellipsometry represents a very fast and accurate optical technique for determining thin film thicknesses and optical constants. The working principle of the ellipsometer used in this thesis (VASE, J.A. Woollam Co., Inc., Lincoln, NE, USA) is visualized in Figure 2.5. The randomly polarized light from a laser source is first passed through a polarizer and a rotating compensator to generate light of a known polarization state before it interacts with the sample surface. Upon reflection of the incident light beam from the sample, the polarization state of the light changes, i.e. a phase shift as well as a change in the amplitude occurs for both p- and s-components of the polarized light. These phase and amplitude changes are detected by the analyzer as well as on a CCD camera and represent the two characteristic measured quantities in ellipsometry. From these two experimental values, the film thickness and/or the optical constants of an interface can be determined. Since the film thickness and the refractive index are correlated for layers below 50 nm, an assumption of the refractive index is required. The film thickness is then extracted through a model- based analysis, which involves the relations that describe the interaction of light with matter.

D etec tor Polarizer Light source Analyzer Compensator

Sample

Figure 2.5: The basic setup of a variable angle spectroscopic ellipsometer (VASE). Upon polarization, the light is reﬂected at the sample surface and its intensity and changes in polarization are detected.

Key advantages of VASE are that the instrument sensitivity is in the sub- nanometer range and the data acquisition is very fast. However, the lateral resolution is limited to about 200 nm due to the size of the laser spot and the necessity for reﬂective, macroscopically smooth samples represents a drawback of this technique. 2.7. QUARTZ-CRYSTAL MICROBALANCE WITH DISSIPATION MONITORING 26 (QCM-D)

2.7 Quartz-crystal microbalance with dissipation monitoring (QCM-D)

The QCM-D technology is based on frequency and dissipation changes of a quartz crystal resonator upon mass uptake or loss. The essential parts of this highly sensitive measuring device are illustrated in Figure 2.6. An oscillating electric ﬁeld, which is applied between the two metal electrodes evaporated on both sides of the piezoelectric quartz crystal, induces mechanical shear waves. When operated at resonance frequency, the quartz crystal resonator is very sensitive to changes in mass, i.e. the mass ∆m adsorbed on top of the crystal electrode will induce a frequency shift and a change in the dissipated energy. Hence, QCM essentially operates as a balance and for rigid ﬁlms coupled to the metal electrode, the decrease in frequency is proportional to the mass of the adlayer, expressed by the Sauerbrey relation (2.2):

C · ∆f ∆m = − (2.2) n In Equation 2.2, ∆m is the mass change, C is the mass sensitivity, which is 17.7 ng·cm−2·Hz−1 for a 5 MHz resonator in its fundamental mode (n = 1), ∆f is the frequency change and n = 1, 3, 5, 7 is the overtone number. The thickness of the adhering layer can be estimated by dividing its mass by the eﬀective density of the adhering layer [68].

O H

O H O

O O H H

H O O O H O O O O H H

O H

O O O

H O

H O O H O O

O H H

O O O

H H

H O

O H

O H O O

O O O H O

H H H

O H

O O

O H O O O H H H f < f 0 ~ D > D 0

Figure 2.6: Quartz crystal resonator of a QCM-D setup, across which an oscillating electric ﬁeld is applied. When operated at resonance frequency, QCM-D is very sensitive to small changes in the adsorbed mass.

Often, the adsorbing film is not rigid but has viscoelastic properties and consequently, the film does not fully couple to the oscillation of the crystal. In these cases, the Sauerbrey relation becomes invalid. Since “soft” (viscoelastic) films dampen the oscillation of a crystal, the dissipation representing the sum of all energy losses in the system provides a useful measure for the characterization of such films. The measurement of multiple frequencies allows for an extraction of the film properties 2.8. ATOMIC FORCE MICROSCOPY (AFM) 27

(viscosity, elasticity, thickness) by applying a viscoelastic model (Voigt model). The QCM-D experiments in this thesis were performed with a Q-Sense E4 instrument (Q-Sense AB, Sweden). The on-line monitoring of the adsorbed mass as well as the possibility to account for the amount of solvent incorporated in the adhering layer render the QCM-D technology very useful for the investigation of boundary lubricants in aqueous tribology. The possibility to investigate the hydration properties of adlayers makes QCM-D a valuable complementary technique in addition to optical waveguide lightmode spectroscopy (OWLS) or ellipsometry, both of which provide information about the dry mass or the ﬁlm thickness based on optical quantities.

2.8 Atomic force microscopy (AFM)

Atomic force microscopy represents a versatile technique to investigate topographi- cal as well as tribological properties of surfaces at the nanoscale. As schematically depicted in Figure 2.7, the basic set-up of the AFM employed in this thesis (Dimen- sion 3000, Veeco Metrology Group, Santa Barbara, CA, USA) consists of a static sample stage, a piezoscanner to which end a cantilever with a sharp tip is attached and a laser beam that is reflected from the back of the cantilever onto a 4-segment photodiode. As the tip is brought into proximity of the sample surface and scanned across it, the movement of the cantilever, which arises from the interaction of the tip and the surface, is detected by tracing the changes in the position of the reflected laser light on the photodiode detector. Given the spring constant of the cantilever, the voltage generated on the photodiode can be converted into a force. Normal deflections of the cantilever in the z direction are due to normal forces, whereas the torsion of the cantilever arises from friction forces. Due to the large spring constant of the cantilever together with the small dimensions of the tip, the AFM is very sensitive to small forces, e.g. adhesion, electrostatic or van der Waals forces. It should be mentioned that most AFM are constructed in a way that the sample is moving, whereas the AFM tim is stationary during the scanning process. With regard to tribology, AFM represents a very useful technique to study single- asperity contacts, mainly due to the nanometer-scale dimensions of the tip. Hence, the contact pressure experienced in AFM is readily controlled by the applied normal force. A further advantage is the possibility to perform experiments in various environments, for instance under ultrahigh vacuum (UHV), in liquid, and with controlled humidity. In addition to that, the samples do not have to be conductive as in the case of scanning tunneling microscopy (STM). For the investigation of lubri- 28 2.9. CONTACT-ANGLE MEASUREMENTS

y x Laser Piezoscanner

Detector Cantilever

Figure 2.7: Central parts of an atomic force microscope (AFM). The cantilever with its known spring constant and a small tip is probing the surface while its deflections are detected by changes in the laser positions on a photodiode. cation regimes, AFM is less versatile compared to macroscopic approaches since the limited scan speed (up to 0.1 mm/s) does not allow sufficient fluid entrainment into the contact zone between tip and surface, i.e. essentially only boundary lubrication studies are feasible by means of AFM. The nanoscale-frictional properties as measured by AFM in this thesis were determined based on the conventional beam-deflection and four-quadrant photodetector method [69–71]. Here, the interfacial friction force between a sample and a tip is measured as a function of applied load. Under a fixed load, the friction force is obtained by subtracting the lateral force obtained during retrace from that obtained during trace in a ‘friction loop’ [71, 72].

2.9 Contact-angle measurements

The equilibrium position of a liquid drop on an ideally smooth solid surface can be described by means of the Young’s equation, which states that the interfacial tension along the 3-phase-line (solid-liquid-gas) cancels out. As depicted in Figure

2.8, the γxy are the interfacial tensions between two phases, i.e. solid (S), liquid (L) and gas (G) phase and Θ is the static contact angle of the liquid drop, which is conveniently determined by the sessile-drop method. In this thesis a contact-angle goniometer (Ram´e-HartModel-100, Ram´e-HartInstrument Co., Netcong, NJ, USA) was employed to determine the static contact angles, which result from the angle that is enclosed by the surface and the liquid-gas phase tangent along the 3-phase line (Figure 2.8). The static contact angle Θ is very sensitive to the outermost chemical species on a surface, which renders such measurements very useful for the investigation of 2.10. ULTRAVIOLET-VISIBLE (UV/VIS) SPECTROSCOPY 29

γ γ - SL Youngʼs equation: cos θ = SG γ LG γ LG

γ SG θ γ SL

Figure 2.8: The equilibrium position of a liquid drop on an ideally smooth surface is determined by the Young’s equation, which states that the interfacial tension along the solid-liquid-gas interface cancels out at equilibrium.

surface hydrophilicity. Since aqueous lubrication is very closely associated with the wettability of the tribological surfaces by water, static contact angle measurements have proven to be a very fast and simple method for basic surface characterization.

2.10 Ultraviolet-visible (UV/Vis) spectroscopy

The photon energy in the visible and the ultraviolet region of the electromagnetic spectrum is suﬃcient for inducing electronic transitions in molecules. Because the Beer-Lambert law (Equation 2.3) relates the absorbance of a solution to its concentration, UV/Vis spectroscopy is frequently used for concentration determination.

A = −log(I/I0) = · c · L (2.3)

In Equation 2.3, A is the measured absorbance, I0 the light intensity at a given wavelength, I the transmitted intensity, is the extinction coeﬃcient, c the concentration of the absorbing species in solution and L the pathlength through the sample. The extinction coeﬃcient is a molecular property in a given solvent and therefore constant at a particular temperature and pressure. The spectrophotometer employed in this thesis (Cary 1 UV-Vis, Varian Deutschland GmbH, Germany) measures the light intensity I passing through the sample and compares it to a reference I0 intensity for a range of wavelength. The obtained spectra are graphs of light absorbance A versus wavelength and they were employed as an indicative measure for the UV light absorption of the photoinitiator and the monomer, respectively. 2.11. SURFACE-INITIATED GRAFT-POLYMERIZATION 30 BY MEANS OF ULTRAVIOLET (UV) IRRADIATION

2.11 Surface-initiated graft-polymerization by means of ultraviolet (UV) irradiation

The basic feature of all surface-initiated polymerization techniques is the in situ growth of polymeric chains from a substrate surface. These techniques represent the “grafting from” approach described in the introduction, where dense polymeric monolayers can be created under appropriate conditions. The three common steps of surface-initiated grafting methods include the initiation of surface-immobilized functional species, chain propagation and chain termination. Whereas a number of stimuli are possible for the initiation of polymerization reactions, ultraviolet (UV) irradiation was employed as a trigger in this thesis. Figure 2.9 shows the essential steps involved in surface-initiated UV-graft polymerization. hν

1 2 3

Figure 2.9: Surface-initiated UV graft-polymerization in liquid. Upon generation of the surface-initiating species (1), the sample is irradiated with UV light in the presence of suitable monomers (2), leading to a growth of polymer chains from the surface (3).

After the generation of initiating species on the substrate surface, e.g. by adsorption of a photoinitiator (1), the sample is brought into contact with the reactive monomer, which is usually dissolved in a suitable solvent (2). Upon activation of the surface-immobilized initiating species by means of UV irradiation, polymer chains start to grow from the surface via the addition of reactive monomer (3). The polymerization usually stops due to chain-terminating reactions, after the monomer is consumed or if the activation trigger is switched oﬀ. The advantages of surface-initiated polymerization techniques are those mentioned in the introduction for “grafting from” approaches. For a high density of initiating species on the surface, very dense polymer brushes can be obtained under the appropriate experimental conditions. The initiation by means of UV light brings about the advantage that the polymerization is fast and can be carried out at low temperatures and in the presence of moisture. Furthermore, the chain lengths of the 2.11. SURFACE-INITIATED GRAFT-POLYMERIZATION BY MEANS OF ULTRAVIOLET (UV) IRRADIATION 31 surface-attached polymer can be conveniently controlled by means of UV light intensity and irradiation time, respectively, and patterning with adequate photomasks becomes possible. Drawbacks of UV-initiated grafting methods are that the reactions have to be carried out in the absence of radical-scavenging oxygen and that the surface-immobilized initiator has to be selected according to the UV absorbance of the monomer, in order to avoid bulk polymerization. Hence, the intensity as well as the spectral range of the UV source needs to be stringently controlled. The details of the photoinitiating system employed in this thesis are given in chapter 6. 2.11. SURFACE-INITIATED GRAFT-POLYMERIZATION 32 BY MEANS OF ULTRAVIOLET (UV) IRRADIATION CHAPTER 3

Macroscopic Testing of Alkanethiol SAMs as Low-Contact-Pressure Aqueous Lubricant Additives

3.1 Introduction

In the past two decades, self-assembled monolayers (SAMs) generated from alkanethiols and alkylsilanes have been extensively studied as model systems for boundary lubricants [69, 70, 73–93]. Since the structure and chemistry of SAM films can be systematically tailored at the molecular level, the majority of reports have focused on the study of friction and lubricating properties of these systems under nanoscale contact conditions [69, 70, 73–77, 79–82, 86, 88, 90–93]. Technically, this has become possible for two reasons; firstly, SAMs can be readily generated on substrates displaying extremely smooth surfaces, such as monocrystalline gold, mica, and highly polished silicon, and thus boundary lubricant films with ideally smooth morphologies have become available. As a comparison, it is worth noting that more classical boundary lubricants, such as fatty acids and alcohols [94–97], have generally been formed on polished metal/metal oxide surfaces, which display much higher surface roughness. Secondly, as the counterface to these boundary lubricant films, a slider consisting of a single asperity, such as the atomic force microscope (AFM) tip [71, 98], has become available. Thus, the control of contact pressure within the contact area became feasible by simply controlling the external load. This is important in that tribological stress could be applied exclusively onto the SAM films, i.e., without penetrating to the underlying substrate, and thus their tribological properties as boundary-lubricant films can be probed. In contrast, tribological contacts involv-

33 34 3.1. INTRODUCTION ing SAMs probed by more conventional, macroscale approaches, such as pin-on-disk tribometry [78, 83–85, 87, 89], have generally resulted in irreversible damage to the samples, both the SAM films and the supporting substrates, which renders the interpretation of the observed tribological properties far more complicated. Clearly, this is due to the multi-asperity contacts arising from the counterpart roughness, e.g., the macroscale slider (pin), and the resulting uncontrolled, extremely high local contact pressures. The poor mechanical durability of SAMs, and often of the supporting substrates themselves in addition, is a major barrier to fundamental tribological studies of SAM films on a macroscopic scale. This problem, however, can be readily overcome if a highly compliant material, such as a rubber, is employed as the sliding counterpart. Under ordinary tribometer experiment conditions, the apparent contact pressure applied by an elastomer can be easily maintained below the MPa range, and due to its high compliance, the surface asperities of the elastomer can readily flatten out under pressure. Thus, the wear problems arising from uncontrolled asperity contact can be avoided. In fact, this configuration, either “soft” slider on “rigid” track or “rigid” slider on “soft” track, has routinely been employed in tribological studies of elastomers [99–101], especially in soft elastohydrodynamic lubrication (soft EHL) studies [3, 4, 10, 102–108].

(a)

rigid pin

(b) (b) Contact Pressure Contact soft pin (c)

(a) (c)

Sliding Speed

Figure 3.1: A pressure-versus-sliding-speed diagram showing the contact conﬁguration of an elastomeric slider on self-assembled monolayer (SAM) ﬁlms: schematic illustrations for (a) single-asperity contacts on a nanoscopic scale as in AFM (low contact pressure, low speed), (b) multi-asperity contacts on a macroscopic scale as in conventional pin-on-disk tribometry employing rigid sliders (high contact pressure, high speed) and (c) soft contacts on a macroscopic scale by employing an elastomeric slider in pin-on-disk tribometry (low contact pressure, high speed).

As schematically depicted in Figure 3.1, the effect of using an elastomeric slider for pin-on-disk tribometric studies of SAM films is to reduce the contact pressure to a level typically available from nanotribological contacts or lower, while the contact 3.2. EXPERIMENTAL PROCEDURES 35 area and the sliding speeds are maintained at levels typical for macrotribological studies. Thus, the tribological properties of SAMs can be investigated in a wear- less regime, as is the case with nanotribological approaches, yet with the possibility of both macroscopic contact area and high sliding speeds. Furthermore, the contact area is large enough that standard surface-analytical approaches can readily access it to characterize the structural and chemical changes present after tribological stress; currently, no generally available surface-analytical approach can access the contact area generated by an AFM probe other than AFM itself. It is considered that this approach - soft slider on SAM films/rigid substrate - is particularly useful to investigate the role of surface chemistry on the efficacy of lubricant-film formation, which is crucial in the soft elastohydrodynamic lubrication (soft EHL) regime. In recent studies, [3, 104] involving water and poly(dimethyl siloxane) (PDMS) as lubricant and elastomeric tribopair, respectively, it was shown that the formation of a soft EHL film is determined not merely by the bulk mechanical and rheological properties of the tribosystem alone as predicted by classical theory [72, 109], but that the surface-chemical properties of tribopairs also play a significant role; when the surface of the tribopair is insufficiently wetted by the lubricant, the predicted formation of an EHL film is significantly retarded, even if the bulk mechanical properties meet the conditions for the formation of a soft-EHL film. Given the wealth of useful approaches that have been established in self-assembly techniques to modify surface-chemical characteristics, the effects of a broad range of surface-chemical modifications on soft EHL can be conveniently explored. In the present chapter, the aim is to establish this methodology by employing SAM systems whose tribological properties have been well established, allowing to focus on testing the proposed methodology.

3.2 Experimental Procedures

3.2.1 Self-assembled monolayers (SAMs)

Three types of SAM ﬁlms have been prepared by spontaneous adsorption from ethanolic solutions of 11-mercaptoundecanol (HS(CH2)11OH), 1-dodecanethiol

(HS(CH2)11CH3), and 1-hexanethiol (HS(CH2)5CH3) onto polycrystalline gold surfaces deposited on glass. For simplicity, these SAMs are abbreviated as C11OH,

C11CH3, and C5CH3, respectively, throughout this article. A schematic illustration of the SAM ﬁlms on gold substrates generated from these molecules is presented in Figure 3.2 . The alkanethiols (Sigma-Aldrich, Switzerland) were used as received. Ethanol (Fluka, Switzerland) was used as solvent for alkanethiol solutions (1 mM). 36 3.2. EXPERIMENTAL PROCEDURES

Soda-lime glass microscope slides (SuperFrost, Menzel-Gl¨aser,Braunschweig, Ger- many) were cut into 2.5 x 2.5 cm2 pieces and used as substrates for the gold ﬁlms. The glass substrates were cleaned according to the following procedure; (a) ultrasonication in ethanol four times for 15 min - the solvent was exchanged each time

(b) O2-plasma for 2 min (high power, Harrick Plasma Cleaner/Sterilizer, Ossining, NY, USA). After cleaning, the substrates were coated with a 10-nm adhesive layer of chromium, followed by 100-nm layer of gold by thermal evaporation (MED020 coating system, BALTEC, Balzers, Lichtenstein). The SAM ﬁlms were generated by immersion of the substrates in the relevant 1 mM ethanolic alkanethiol solution overnight. Prior to the immersion step, the gold substrates were rinsed with pure ethanol and subsequently dried with nitrogen.

(a) (b) (c)

OH OH OH OH OH OH OH OH CH 3 CH 3 CH 3 CH 3 CH 3 CH 3 CH 3 CH 3

CH 3 CH 3 CH 3 CH 3 CH 3 CH 3 CH 3 CH 3

SSS S SSS S S S S S SSS S SSS S SSS S Au

glass

Figure 3.2: Schematic representation of the alkanethiol SAM ﬁlms on gold substrates used in this study: (a) 11-mercaptoundecanol (HS(CH2)11OH), (b) 1-dodecanethiol (HS(CH2)11CH3) and (c)1-hexanethiol (HS(CH2)5CH3).

3.2.2 Poly(dimethyl siloxane)(PDMS)

Poly(dimethyl siloxane) (PDMS) was employed as an elastomeric counterpart (pin) against the SAM ﬁlms/gold/glass in pin-on-disk tribometry experiments. The PDMS pins were prepared from the Sylgard 184 silicone elastomer kit introduced in chapter 2. In order to prepare PDMS pins with hemispherical ends, a commercial polystyrene cell-culture plate with round-shaped wells (radius = 3 mm, 96 MicroWell Plates , NUNCLON Delta Surface, Denmark) was employed as master. Planar PDMS sheets were also prepared for the characterization of surface hydrophilicity according to the same procedure. The surface of the PDMS pins was hydrophilized by means of air- plasma treatment for 1 min, and thus they are denoted as “ox-PDMS” throughout this thesis. For a control experiment, the PDMS pins were immersed in n-hexane, to extract uncrosslinked monomer species prior to the air-plasma treatment; the solvent, n-hexane, was exchanged twice during 24 h. 3.2. EXPERIMENTAL PROCEDURES 37

3.2.3 Water contact angle measurements

The surface hydrophilicity of the tribopairs, including the SAM ﬁlms and ox-PDMS, was characterized by measuring static water contact angles, θw, by employing the contact-angle goniometer mentioned in chapter 2. All contact-angle measurements were averaged over ﬁve runs.

3.2.4 Ellipsometry

The dry thicknesses of the SAM ﬁlms were determined by variable angle spectroscopic ellipsometry. Measurements were conducted under ambient conditions at three angles of incidence (65°, 70°, and 75°) in the spectral range of 370 - 995 nm. Measurements were ﬁtted with the WVASE32 analysis software using a three-layer model for an organic layer on a gold/glass substrate.

3.2.5 Polarization-modulation infrared reﬂection-absorption spectroscopy (PM-IRRAS)

Polarization-modulation infrared reﬂection-absorption spectra (PM-IRRAS) were recorded with the IR spectrometer introduced in chapter 2. All experiments were performed with a grazing angle of 80°.

3.2.6 Atomic force microscopy (AFM)

Atomic force microscopy was employed to characterize the surface morphological and nanotribological properties of the samples. The surface morphology was char- TM acterized by TappingMode AFM; the Ra values of the SAM ﬁlms were obtained by directly imaging the samples, whereas the Ra values of the PDMS pin were obtained by imaging the polystyrene master. A tapping-mode silicon cantilever and tip

(PointProbe Plus, k = 92 N/m, f0 = 330 kHz, Germany) was used as a probe. Nan- otribological properties of the SAM films were characterized by contact-mode AFM under distilled water (> 18 MΩcm). Based on the conventional beam-deflection and four-quadrant photodetector method [69–71], the interfacial friction force between a sample and a tip was measured as a function of applied load. Under a fixed load, the friction force was obtained by subtracting the lateral force obtained during retrace from that obtained during trace in a “friction loop” [71, 72]. The friction force under a given load was obtained at more than 10 different positions on the samples for statistical evaluation. The scan length was 1 µm and the sliding speed 2 µm/s. This procedure was repeated by varying the applied load. Since the 38 3.2. EXPERIMENTAL PROCEDURES purpose of the AFM nanotribological studies in this work is to provide a reference for the relative order of frictional forces of the SAM films on a nanoscopic contact scale, the normal spring constant value of the commercial silicon nitride cantilevers indicated by the manufacturer (Nanoprobes, Veeco Metrology Group, Santa Bar- bara, CA, USA), kN (0.12 N/m), was employed without further calibration. In order to ensure the valid comparison of the frictional data, however, all measurements were performed with an identical tip/cantilever assembly. The applied load varied between 10 nN and 80 nN. Friction forces were expressed in arbitrary units, as received from the photodiode detector without further conversion. Along with friction measurements, the adhesive properties of the SAM films were also characterized by acquiring force-versus-distance curves using the identical tip/cantilever assembly. Prior to the measurements, the tip/cantilever assembly was immersed in a commercial acidic cleaning solution (Cleaner, COBAS INTEGRA (HCl 300 mM, detergent 1%), Roche, Germany) for 5 min, followed by immersion in distilled water for 10 min, and finally oxygen-plasma cleaning for 30 s.

3.2.7 Pin-on-disk tribometry

The macroscopic-scale tribological properties of the SAM films were characterized by means of pin-on-disk tribometry. PDMS with an end-radius of 3 mm was employed as a standard pin material, but for comparison, a stainless steel ball (DIN 5401- 20 G20, Hydrel AG, Romanshorn, Switzerland) with a radius of 3 mm was also employed as pin. The stainless steel ball was rinsed with ethanol, blown dry with nitrogen, and oxygen-plasma cleaned for 30 s. The raw data for the friction forces were recorded as a function of time (or the number of rotations) over a fixed track, using the software (InstrumX version 2.5A) provided by the manufacturer. All measurements were carried out in distilled water (> 18 MΩcm). The standard protocol for the friction measurements involved the acquisition of µ-versus-speed plots (µ = friction force/load) and friction-versus-load plots. For µ-versus-speed plots, the speed was varied from 0.25 mm/s to 100 mm/s under a fixed load (1 N), and for friction-versus-load plots, the load was varied from 0.5 N to 5 N under two fixed speeds (1 mm/s and 50 mm/s). The number of rotations was 20 for each measurement. The average friction force from the latter half of these (11 - 20th) was obtained to avoid “running-in” effects. For control experiments involving the examination of the sliding track by PM-IRRAS, the number of rotations was extended to 500 under 1 N. 3.3. RESULTS AND DISCUSSION 39

3.3 Results and Discussion

3.3.1 Initial characterization of the SAM ﬁlms

Before the start of the tribological experiments, the SAM ﬁlms formed on thermally evaporated gold substrates were characterized by ellipsometry, water-contact-angle measurements, AFM, and PM-IRRAS. Ellipsometry revealed monolayer thicknesses of 1.57 ± 0.11 Nm for the C11OH, 1.43 ± 0.13 nm for the C11CH3, and 0.47 ±

0.52 nm for the C5CH3 SAM ﬁlms. The refractive index for the organic layers was assumed to be 1.45. The water contact angle, θw, for the C11OH SAM ﬁlm, 20 ±

5°, was substantially lower than those for the C11CH3, 109 ± 2°, and the C5CH3, 106 ± 2 °, films as well as that of bare gold substrate, 70 ± 5°, which indicates the exposure of -OH (C11OH) and -CH3 (C11CH3 and C5CH3) groups on the surfaces of the corresponding SAM films. The average values and the standard deviation of water contact angles were obtained from more than 10 different measurements on three different samples for each SAM. The slightly lower water contact angle of

C5CH3, 106°, than that of C11CH3, 109°, on average, is generally ascribed to more probable exposure of methylene groups (–CH2–) arising from the less well-packed backbone structure of C5CH3 compared to C11CH3 SAMs [109, 110]. The θw values of the ox-PDMS surfaces were < 3°. AFM revealed homogenously distributed grainy features of the gold substrates (diameter: a few tens of nm scale), which are typical morphological characteristics of thermally evaporated polycrystalline gold surfaces on glass or silicon oxide. The morphologies of the SAM ﬁlms on the gold substrates were indistinguishable from those of the substrates, which supports the formation of homogeneous monolayers.

The Ra values obtained from the gold substrates over 1 µm x 1µm, 10 µm x 10 µm, and 50 µm x 50 µm, were 1.3, 1.4, and 1.6 nm, respectively (the error bars are ± 0.60 nm), and those for the three SAM films were virtually identical. In parallel, the topographic images of the master for the PDMS pins were also obtained. Compared with the SAM films on gold substrates, slightly rougher surfaces, including some random features of a few tens of nm in height, were observed. The Ra values obtained over 1 µm x 1µm, 10 µm x 10 µm, and 50 µm x 50 µm, were 1.2, 2.4, and 3.4 nm, respectively (the error bars are ± 1.10 nm). Analysis by PM-IRRAS revealed more detailed information on the chemical and structural features of the SAM films. The asymmetric and symmetric methyl C–H −1 −1 stretches, νa(CH3) at 2962 cm and νs(CH3) at 2877 cm , respectively, and the −1 band at 2936 cm assigned to a Fermi resonance (FR) between the CH3 symmetric stretching mode and overtones of bending modes, appear only for C11CH3 and 40 3.3. RESULTS AND DISCUSSION

C5CH3 SAM ﬁlms, but are absent for C11OH SAM ﬁlms, as expected. The position and peak width of the methylene asymmetric C–H stretching modes are particularly sensitive to the conformational order of the alkyl chains [79, 81, 109–112].

(CH ) (CH ) ! (CH FR) !s 2 !a 2 s 3

C5CH3

C11CH3 Intensity (a.u.)

C OH (CH ) (CH ) 11 !s 3 !a 3

2750 2800 2850 2900 2950 3000 3050 -1 Wavenumber (cm )

Figure 3.3: PM-IRRA spectra obtained from the three alkanethiol SAMs.

As seen in Figure 3.3, the νa(CH2) peak of the C11CH3 film appears at 2919 −1 cm , indicating a highly ordered hydrocarbon backbone structure. The νa(CH2) −1 peak of the C11OH film was observed at a slightly higher value of 2921 cm . In contrast, the νa(CH2) peak of the C5CH3 film is poorly defined and appears much broader than those of the C11CH3 and C11OH SAM films, which indicates that the conformation of the C11CH3 film is very disordered and liquid-like. As a whole, these data verified that the three SAMs displaying (a) varying surface hydrophilicity (C11OH versus C11CH3/C5CH3 SAMs) and (b) varying conformational order of the alkyl chains (C11CH3 versus C5CH3 SAMs) have been successfully generated on polycrystalline gold substrates.

3.3.2 Nanotribological properties of the SAM ﬁlms by means of AFM

The nanotribological properties of the SAM ﬁlms generated from C11OH, C11CH3, and C5CH3 SAMs were characterized through the measurement of frictional forces with a plasma-cleaned silicon nitride tip as a function of load under distilled water. Representative friction-versus-load plots obtained by AFM are presented in Figure 3.4. The force-versus-distance curves obtained from the same three SAM ﬁlms are shown in Figure 3.5. 3.3. RESULTS AND DISCUSSION 41

150

C11OH

C11CH3

C5CH3

100 Friction (a.u.) 50

0 0 20 40 60 80 Normal Load (nN)

Figure 3.4: Friction-force-versus-load plots obtained from the interaction between an AFM tip and the three SAM ﬁlms in distilled water.

In order to ensure a statistically valid comparison, the measurements were repeated several times in varying order using the same tip/cantilever assembly. Al- though minor differences were observed in the frictional and adhesive properties of the samples, the relative order of friction forces of the SAM films were consistently observed as in Figure 3.4: C5CH3 > C11CH3 > C11OH. This order is highly correlated with the order of “pull-off forces” obtained from force-versus-distance curves as shown in Figure 3.5: C5CH3 ≈ C11CH3 > C11OH. As mentioned above, the choice of the three SAMs in this work was motivated by an attempt to employ standard SAM systems, whose tribological properties have been well established and, therefore, the focus was placed on the proposed test methodology. For the CH3-terminated normal alkanethiol or alkylsilane SAM films, it has been well known that longer hydrocarbon chains usually contribute to en- hancing both the packing density and the order of the backbone and, therefore, tend to lower the interfacial frictional properties [73, 74, 80, 82, 90, 92, 93]. While most previous nanotribological studies of these systems have been carried out under ambient conditions, this work has now been able to show the same trend in distilled water: alkyl-chain deformation still appears to be the dominant energy-dissipation mechanism for the two CH3-terminated SAM films.

C11OH and C11CH3 SAMs display similarly well-ordered backbones, are constructed from identical hydrocarbon chains, and yet present diﬀerent terminal groups

(OH– versus CH3–). The tribological properties of the SAM ﬁlms displaying chemically diﬀerent terminal groups are known to be closely associated with their adhesive properties, which are, in turn, dependent on both the chemical identity of the 42 3.3. RESULTS AND DISCUSSION counterface and the medium in which the tribological interaction takes place [74– 77, 79, 81, 86, 88, 91]. Since the oxygen-plasma-treated silicon nitride AFM probe in this work displays hydrophilic surface characteristics (OH-groups), the work of adhesion between two hydrophilic surfaces (the AFM probe versus C11OH) is apparently lower than that involving a hydrophobic surface (the AFM probe versus

C11CH3) in aqueous media [88], as shown in Figure 3.5. The difference in the adhesive forces for these two SAM films with the AFM probe is thus mainly responsible for the difference in the nanotribological properties. Overall, the nanotribological properties of the SAM films revealed by AFM in this work serve as a reference for the intrinsic tribological properties of the films imparted by their structural and chemical features in a wear-less contact regime.

C11OH C CH 20 11 3 C5CH3

Normal force (nN) -10

-20

-30 0 200 400 600 800 1000 Piezo displacement (nm)

Figure 3.5: Force-versus-distance plots obtained from the interaction between an AFM tip and the three SAM ﬁlms in distilled water.

3.3.3 Elastomeric sliding contact on the SAM ﬁlms on a macroscopic scale

The primary interest in employing an elastomeric (PDMS) pin as the sliding partner for these three SAM ﬁlms is to test if the alkyl-chain-dependent (C5CH3 versus

C11CH3 SAMs) and terminal-functional-group-dependent (C11CH3 versus C11OH SAMs) frictional properties, as observed in AFM experiments (Figure 3.4 - 3.5), can be reproduced on a macroscopic scale. Figure 3.6 shows that this is indeed the case; when the friction-versus-load plots were obtained from the sliding contacts between ox-PDMS pins, which are chemically similar to the oxidized silicon nitride AFM tip, and the three SAM ﬁlms in distilled water; the same order of the interfacial 3.3. RESULTS AND DISCUSSION 43

frictional properties, C5CH3 > C11CH3 > C11OH, was measured at 1 mm/s. This trend was consistently observed from the friction-versus-load plot obtained at much higher sliding speed, 50 mm/s, as shown in Figure 3.7.

6 oxPDMS pin, 1 mm/s C11OH C CH 5 11 3 C5CH3

3 Friction (N) 2

0 0 1 2 3 4 5 Load (N)

Figure 3.6: Friction-force-versus-load plots (sliding speed = 1 mm/s), obtained from the tribopair of ox-PDMS/SAM ﬁlms in distilled water.

1.0 oxPDMS pin, 50 mm/s (b)

C11OH

C11CH3 0.8 C5CH3

0.6

Friction (N) 0.4

0.2

0.0 0 1 2 3 4 5 Load (N)

Figure 3.7: Friction-force-versus-load plots (sliding speed = 50 mm/s), obtained from the tribopair of ox-PDMS/SAM ﬁlms in distilled water.

However, the µ values at 50 mm/s are signiﬁcantly lower than those at 1 mm/s for all three SAMs: for instance, 0.008 (50 mm/s) vs. 0.061 (1 mm/s) for the C11OH

film, 0.024 vs. 0.76 for the C11CH3 film, and 0.14 vs. 0.90 for the C5CH3 film, under the load of 5 N. The speed-dependent frictional behavior of the sliding contacts of the ox-PDMS/ SAM films is more clearly seen in the µ-versus-speed plots (load = 1 N), as shown in 44 3.3. RESULTS AND DISCUSSION

Figure 3.8. While the µ values were generally decreasing with increasing speed for all cases, the onset of the decrease in µ appears to be dependent on the hydrophilicity of the SAM films. For the two CH3-terminated SAM films, for instance, the µ values started to decrease from ca. 5 mm/s, reaching 0.04 - 0.05 at the highest speed, whereas the decrease of µ for the C11OH SAM started from the lowest speed, 0.25 mm/s, gradually to the highest speed, 100 mm/s, reaching a lowest value of µ ≈ 0.01. These observations suggest that the lubrication mechanisms in the low- and high-speed regimes might be different, despite the same relative order of the frictional properties of the three SAM films. This issue will be discussed in detail below. When the SAM films were slid against a hydrophilic (oxygen-plasma treated), yet rigid slider, a stainless steel pin, wear of the film and the substrate became so significant that the structural and chemical characteristics of the SAMs were not discernable any more, as shown in Figure 3.9. Due to the apparent damage on the sliding track for each measurement, the sliding track was changed for each speed condition, and the µ values remained constant at ca. 0.15 over the entire speed range for all three samples (two SAM films and the bare gold substrate). Although some previous macrotribological studies using a rigid slider have revealed alkyl- chain-length-dependent frictional properties of SAMs, this distinction was possible only under a load of a few mN [84]. The pin-on-disk tribometer experiments in this work have thus revealed that the frictional properties of the SAM films observed by AFM could be reproduced on a macroscopic contact scale by employing a hydrophilic elastomeric slider.

10 oxPDMS pin

0.1 Coefficient of Friction

C11OH

C11CH3

C5CH3

0.01 0.1 1 10 100 Sliding Speed (mm/s)

Figure 3.8: µ-versus-speed plots (load = 1 N) obtained from the tribopair of ox- PDMS/SAM ﬁlms in distilled water. 3.3. RESULTS AND DISCUSSION 45

10 Au

C11OH

C5CH3

0.1 Coefficient of Friction

0.01 0.1 1 10 100 Sliding Speed (mm/s)

Figure 3.9: µ-versus-speed plots (load = 1 N) obtained from the tribopair of stainless steel/SAM ﬁlms in distilled water.

3.3.4 PM-IRRAS studies on the sliding track

One distinct advantage of the macrotribological versus the nanotribological approach is that the sliding tracks are generally wide enough to be accessed by standard surface-spectroscopic tools. Given that the tribological properties of the three SAM films characterized by pin-on-disk tribometry using an elastomeric slider, ox-PDMS, revealed the same order of frictional properties as were observed by AFM, it is of interest to examine if the SAMs retain their structural integrity following tribological stress. It is noted that most tribological experiments on SAM films by AFM are believed to take place in a wear-less regime. However, no direct spectroscopic evidence is currently available due to the very small contact area involved. In this work, PM-IRRAS was employed to examine the molecular structure of the SAMs following tribological stress applied by the elastomeric slider. For these experiments, the sliding of an ox-PDMS pin against the C11OH and C11CH3 SAM films was extended to 500 rotations at 1 mm/s and 1 N. The PM-IRRA spectra obtained from inside and outside the sliding tracks are compared in Figures 3.10 (A, B) and 3.11 (A, B). In Figure 3.10 (A, B), the spectral region sensitive to the characteristic structural features of the SAM films is shown. Firstly, it is notable that the peak positions for −1 the asymmetric methylene C–H stretch, νa(CH2, 2919 - 2921 cm ), are virtually identical for inside (solid line) and outside (dotted line) the sliding track, which indicates that the SAM films remain intact and retain a nearly well-ordered backbone structure despite the tribological stress. Secondly, the intensity of the asymmetric −1 methyl C–H stretch, νs(CH3, ca. 2962 cm ) has somewhat increased inside the 46 3.3. RESULTS AND DISCUSSION

Intensity (a.u.) C

2750 2800 2850 2900 2950 3000 3050 -1 Wavenumber (cm )

Figure 3.10: PM-IRRA spectra obtained from inside and outside the sliding track of the SAM ﬁlms. The displayed spectral region is sensitive to the characteristic structural features of the SAMs (A: ox-PDMS/ C11OH, B: ox-PDMS/C11CH3, C: ox-PDMS (extracted)/C11OH, D: ox-PDMS (extracted)/C11CH3). The tribostress was applied by generating the sliding contacts between ox-PDMS/SAM ﬁlms for 500 rotations (load = 1 N and sliding speed = 1 mm/s).

sliding track, and this change is more signiﬁcant for the cases of C11CH3 SAMs than of C11OH SAMs. The increase in the intensity of the asymmetric methyl C–H stretch peak coincides with the occurrence of the peaks at 1110 and 1265 cm−1

(arising from Si–O asymmetric stretching and Si–CH3 asymmetric bending modes, respectively [112]) inside the sliding track, as shown in Figure 3.11 (A, B), which suggests that the transfer of uncrosslinked monomer species from the pin to the SAM films might have occurred. In order to test this hypothesis, the same experiments were repeated after extracting uncrosslinked monomer species from the PDMS pin network. As shown in Figure 3.10 (C, D), the PM-IRRA spectra corresponding to the C–H stretching modes, from 2750 cm−1 to 3050 cm−1, obtained from inside and outside the sliding track are almost identical. A slight difference in peak intensity for some cases shown in Figure 3.10 (C, D) was statistically not significant. The occurrence of Si–O–Si peaks in the low wavenumber region, Figure 3.11 (C, D), has also been substantially suppressed for all cases. However, the transfer of a trace amount of the monomer species to C11OH SAMs by ox-PDMS pins was observed to be not completely avoidable. The extraction of the monomer species from the PDMS pins, however, did not influence on the relative frictional properties of the SAM films shown in Figures 3.6 - 3.8. Spectroscopic examination of the sliding track by PM-IRRAS clearly verified that the SAM films retain their ordered structure following the tribostress provided by 3.3. RESULTS AND DISCUSSION 47

C Intensity (a.u.)

1000 1050 1100 1150 1200 1250 1300 -1 Wavenumber (cm )

Figure 3.11: PM-IRRA spectra obtained from inside and outside the sliding track of the SAM ﬁlms. The displayed spectral region is sensitive to the Si–O–Si stretching and to the Si–CH3 bending modes (A: ox-PDMS/ C11OH, B: ox-PDMS/C11CH3, C: ox-PDMS (extracted)/C11OH, D: ox-PDMS (extracted)/C11CH3). The tribostress was applied by generating the sliding contacts between ox-PDMS/SAM ﬁlms for 500 rotations (load = 1 N and sliding speed = 1 mm/s).

the elastomeric slider. The analysis of the sliding track that has received the tribological stress is common practice in macrotribological studies of SAM films [83–85]; however, it was even easier when an elastomer was employed as the sliding partner, because of the larger contact area achieved at the elastomeric interface. It should be emphasized that the spatially resolved PM-IRRAS measurements presented here were simply measured by positioning different regions of the sample in the beam path under normal sampling conditions and, therefore, the spectra presented above as “inside” the wear track do not completely exclude that the beam was sampling some of the area outside the track as well. As a rough estimate, the fraction of the wear track probed by the beam in this measurement, assuming an ideally aligned beam path and sample positioning, is about 80 %. However, the reasonable resolution between inside and outside the sliding track by PM-IRRAS obtained even with such a simple set-up suggests, that this approach holds potential as a powerful research methodology. Although PM-IRRAS was employed mainly to verify the mechanical durability of the SAM films in this work, more diverse and systematic spectroscopic studies associated with tribological contacts are expected; the detection of the monomer transfer following the sliding contact between ox-PDMS/SAM films can be one example. 48 3.3. RESULTS AND DISCUSSION

3.3.5 Lubrication mechanisms: The transition from mixed lubrication to soft EHL

Another important aspect of macrotribometric approaches is that they can easily explore the high-speed regime. This feature is particularly advantageous when the tribological testing is carried out in a liquid medium as in this work; with increasing sliding speed, the entrainment of the liquid into the contact zone and the formation of the fluid film starts to occur, and thus the transition from boundary lubrication to fluid-film lubrication is also observed with increasing speed. Despite its extremely low pressure-coefficient of viscosity, water can also form a fluid-film lubricant layer when one or both sides of the tribopair is comprised of an elastomer, i.e., in the soft-EHL regime (also known as isoviscous-elastic lubrication regime) [3, 4, 10, 102– 108]. The likelihood of the soft-EHL mechanism occurring and the corresponding fluid-film thickness for a given elastomeric tribosystem can be readily predicted according to the theoretical model described by Hamrock and Dowson [103] (later revised by Esfahanian and Hamrock [10]) based on the bulk mechanical properties of the tribopair and the rheological properties of the lubricant. When expressed in terms of the material and measurement parameters, the minimum film thickness is

0.77 0.65 0−0.44 −0.21 hmin = 2.8 · R · (η · us) · E · w (3.1)

where R is the radius of the slider for a sphere-on-plane contact (3 mm), η is the −4 viscosity of the lubricant (9 · 10 Pas), us is the sliding speed, E’ is the reduced 2 2 contact modulus, 1/E’ = ((1 - ν1 )/E1) + ((1 - ν2 )/E2) where ν is the Poisson ratio (EPDMS = 2 MPa, Eglass = 270 GPa, νPDMS = 0.5, νglass = 0.2) and w is the applied load (variable). The calculation of the film thickness was carried out by employing the parameters employed for the tribometer experiments (Figures 3.6 - 3.8). The results are shown in Figures 3.12 and 3.13. Since the film thickness predicted from Equation 3.1 is that for ideally smooth surfaces (Ra = 0), the relative magnitude of the film thickness compared to the surface roughness can be taken into account by estimating the ratio between them, i.e. the Λ ratio (Λ = hmin/σ, where σ is the combined surface roughness of the q pin disk tribopair, Ra + Ra and hmin is the minimum film thickness). This Λ ratio is often employed to estimate the lubrication mechanism for rough surfaces; generally, fluid-film lubrication is expected when Λ ≥ 3 and boundary lubrication is expected when Λ ≤ 1. Consequently, mixed lubrication is expected to dominate when 1 ≤ Λ ≤ 3 [113]. It is noted that due to the larger area of contact for the tribometer experiments in this work (mm-range in contact radius), the Ra values of the sample 3.3. RESULTS AND DISCUSSION 49

h1 h2 L1 L2 100 20

16 10 14

12 !

ratio

(nm) 1 10 min h 8

6 0.1 4

0.01 0 0 1 2 3 4 5 Load (N)

Figure 3.12: The expected minimum ﬁlm thickness and the Λ ratio calculated according to Equation 3.1 for the sliding contacts between ox-PDMS/glass tribopairs in distilled water as a function of load (h1 and L1: sliding speed = 1 mm/s, h2 and L2: sliding speed = 50 mm/s).

1000 20 h L 18

16 100 14

12 ! ratio

(nm) 10 10 min h 8

6 1 4

0.1 0 0.1 1 10 100 Sliding Speed (mm/s)

Figure 3.13: The expected minimum film thickness and the Λ ratio calculated according to Equation 3.1 for the sliding contacts between ox-PDMS/glass tribopairs in distilled water as a function of speed at 1 N load (h = minimum film thickness, L = Λ ratio). surfaces characterized by AFM are extrapolated to the corresponding scale. Figure 3.12 shows that the Λ ratio for the friction-versus-load plots at 1 mm/s ranged from 2.78 (0.5 N) to 1.71 (5 N), whereas those at 50 mm/s ranged from 11.5 (0.5 N) to 7.12 (5 N), which suggests that the sliding contacts of ox-PDMS/SAM films might have occurred in different lubrication regimes in distilled water, i.e. mixed lubrication for 1 mm/s and fluid-film lubrication for 50 mm/s. Although the experimental data for the fluid-film thickness are currently not available, significantly 50 3.3. RESULTS AND DISCUSSION lower µ values observed from the friction-versus-load plots at 50 mm/s (Figure 3.7) compared with those at 1 mm/s (Figure 3.6) are consistent with this argument. The Λ-ratio-versus-speed plots in Figure 3.13 shows that the transition to fluid-film lubrication, i.e., Λ ≥ 3, is expected to occur from ca. 10 mm/s in the case of an applied load of 1N. It is emphasized that for the experiments carried out in distilled water, the occurrence of different lubrication mechanisms at different speeds was only to be expected for elastomeric contacts on the macroscopic scale, such as the ox-PDMS/SAM film tribopairs tested by means of pin-on-disk tribometry. When the same calculation was carried out under the same experimental conditions for the AFM probe/SAM

films tribopair (EAF M tip = 140 GPa, ESAM = 9.3 GPa, νAF Mtip = 0.25, νSAM = 0.35, R = 50 nm, all the parameters quoted from the references [114] and [115]), the expected film thickness is in the range of 10−5 nm, and the Λ ratio is smaller than 2.10 · 10−5 as shown in Figure 3.14. Even if 1000 µm/s, the upper speed limit of an ordinary AFM, is employed for the calculation of the film thickness, the corresponding value is not higher than 1.50 · 10−3 nm and the Λ ratio is also not higher than 1.14 · 10−3, as shown in Figure 3.14. In other words, the transition from boundary lubrication to mixed or fluid-film lubrication is unlikely to occur for typical tribological contacts on SAM films by AFM probes in distilled water.

-2 h1 h2 L1 L2 -2 10 10

-3 -3 10 10 ! ratio -4 -4

(nm) 10 10 min h

-5 -5 10 10

-6 -6 10 10 0 20 40 60 80 100 Load (nN)

Figure 3.14: The expected minimum ﬁlm thickness and the Λ ratio calculated according to Equation 3.1 for the sliding contacts between a silicon nitride AFM probe and the SAM ﬁlm in distilled water as a function of load (h1 and L1: sliding speed = 2 µm/s, h2 and L2: sliding speed = 1000 µm/s). 3.4. CONCLUSIONS 51

3.3.6 Lubrication mechanisms: the role of surface chemistry

The three SAM films employed in this work contributed to the modification of the interfacial frictional properties of the sliding contacts of elastomer/rigid substrates in two ways. Firstly, in the context of activating the fluid-film lubrication mechanism, the SAM films provided different surface hydrophilicities of the sliding track to vary the interfacial hydrophilicity of the tribopair; as addressed in a previous study involving self-mated PDMS sliding contacts in water [3]. The bulk mechanical properties of the tribopairs and the surface roughness, discussed in the previous section, are, in practice, not the sole parameters that determine the activation of the soft-EHL mechanism: The surface chemistry of the tribopairs also plays a very significant role. Although the C11OH SAM is the most hydrophilic of the three SAM

ﬁlms employed in this work with θw ≈ 25°, it is less hydrophilic than ox-PDMS, θw <

3°, and thus the µ values for ox-PDMS/C11OH were observed to be generally higher than those of ox-PDMS/ox-PDMS [3]. Secondly, in the mixed lubrication regime, where the formation of the soft-EHL film is only partly expected (i.e. 1 < Λ < 3), both surface chemistry and the structural integrity of the SAM films influenced the interfacial frictional properties in a similar fashion to that observed in AFM experiments. In terms of surface chemistry, the hydrophilic C11OH SAMs showed lower frictional forces than those observed for the two CH3-terminated SAMs when sliding against an ox-PDMS pin in the low-speed and/or high-load regime; under such conditions, however, this trend was not due to the higher feasibility of forming an aqueous fluid-film, but rather due to the lower work of adhesion for the hydrophilic interface in an aqueous environment, similarly to the AFM experiments shown in

Figure 3.5. In terms of the structural integrity, the C11CH3 SAMs revealed lower frictional forces than the C5CH3 SAMs for all cases, again mirroring the AFM experiments.

3.4 Conclusions

In this chapter, it was demonstrated that the intrinsic frictional properties of three alkanethiol SAM films, previously assessed by nanotribological approaches only, can be manifested on a macroscopic scale by employing an elastomer, PDMS, as the sliding partner in pin-on-disk tribometry. This approach provides unique opportunities to investigate the tribological properties of SAM films that have hitherto not been accessible by either conventional macrotribological or nanotribological approaches. Firstly, while the contact pressure can be maintained low enough to retain the integrity of SAM films, as with nanotribological approaches, high-speed, macroscale 52 3.4. CONCLUSIONS tribological contacts can be achieved by employing a pin-on-disk tribometer with an elastomeric counterface. Secondly, the wide sliding tracks generated in this approach - even wider than those obtained in conventional tribometry using rigid sliders - allow for standard spectroscopic approaches to access the contact area and characterize the influence of the tribological contacts on the SAM films. Thirdly, by running the measurements in a liquid medium, it is possible to induce a range of lubrication regimes, from boundary lubrication to fluid-film lubrication, over the speed range available from an ordinary pin-on-disk tribometer. Given that a large toolbox of self-assembly approaches to controlling surface chemistry has been established, this approach possesses the potential for systematic studies to investigate the influence of surface chemistry on elastomer-based tribological systems. CHAPTER 4

Inﬂuence of Salt on the Aqueous Lubrication Properties of Ethylene Glycol-based Self-assembled Monolayers (SAMs)

4.1 Introduction

End-tethered polymer chains have drawn considerable attention in the past decade as an effective means to achieve ultra-low interfacial friction [1, 116–118]. When the two surfaces sliding past each other are modified with end-grafted polymer chains, the interfacial friction forces can be greatly reduced under “good” solvent conditions. Under these conditions, the tethered polymer chains display a highly extended “brush-like” conformation, acting as a cushioning layer to sustain the externally applied pressure. Thus, the solvent quality plays a critical role in determining the lubricity of such systems; when exposed to solvents with poor solvent quality, the polymer chains adopt a collapsed state, and thus the cushioning effect to withstand the external pressure is correspondingly impaired. Previous studies have shown that the lubricating effect of surface-grafted polymer chains is indeed degraded in liquids with poor solvent quality [119–121]. Lubrication by means of end-tethered polymers is particularly attractive for water-soluble polymers in an aqueous environment, opening the possibility of using water as a lubricant. In this regard, poly(ethylene glycol) (PEG) has been most intensively investigated [3, 117, 120–123], partly due to its unique solubility in water, and partly because of its excellent biocompatibility [20, 21, 124]. Many previous studies involving PEG polymer brushes have shown a great enhancement in the aqueous lubrication properties of various materials [117]. One of the critical

53 54 4.1. INTRODUCTION aqueous parameters that has been known to affect various PEG solution properties is the ionic strength; generally, the addition of a high concentration of electrolytes into aqueous PEG solutions is known to decrease the solubility of PEG by disrupting the unique EG-water structure, i.e. by “salting-out” [125–130] and/or by complexation of PEG with cations [131, 132]. This latter effect is similar to the ion complexation of crown ethers, where the cation is immobilized inside the polyether ring through ion-dipole interactions with ether oxygens [133]. The ability of PEG to form complexes with cations stems from the high flexibility of the polyether chain but the effect is considerably weaker than that observed with cyclic ethers [134]. The “salting-out” effect of PEG is believed to originate from the disrupting of structured water in the vicinity of the chain, which normally accounts for the stabilization of the helical PEG conformation in water. It has been reported that ions disturb the water structure around PEG chains by becoming hydrated [135]. Both effects are known to induce a smaller radius of gyration of PEG chains, a reduced viscosity, and a lower cloud-point temperature [125–132]. In other words, the addition of salts into aqueous PEG solutions results in the degradation of the solvent quality of water toward PEG.

A similar salt effect, due to degrading solvent quality, would be expected for the aqueous lubricating properties of PEG chains that are end-grafted onto surfaces. However, direct experimental correlation between the conformational change and the aqueous lubricating properties of the end-tethered PEG chains induced by the addition of salts has rarely been addressed. The influence of solvent quality on the lubricating properties of surface-grafted PEG chains has been investigated, in a general sense, by employing organic solvents that differ from water in their solvent quality [119–121, 136]. Challenges in obtaining the salt-effect data include the stringent experimental conditions that need to be met to investigate the influence of salts on the conformational change and the aqueous lubricating properties of surface-grafted PEG chains. Firstly, anchoring of PEG chains should be stable enough to allow them to remain immobilized on the surface in the course of the exposure to high-salt aqueous solution. This restricts potential anchoring approaches for PEG chains to those that are immune to changes in electrostatic interactions between anchoring groups of the polymer chains and the substrate, such as covalent bonding. For this reason, for this study PEG chains end-functionalized with thiol groups (–SH) were selected, namely α-methoxy-ω-mercapto poly(ethylene glycol) (m.w. 5000 Da), which can be readily bound onto gold surfaces. For comparison purposes, an oligo-(ethylene glycol) analogue, namely α-methoxy-ω-mercapto hepta(ethylene glycol) (m.w. 356 Da) was also employed to investigate the role of 4.1. INTRODUCTION 55 the ethylene glycol chain length in aqueous lubrication. Additionally, two other hydrophilic surfaces, which are obtained from alkanethiols with hydrophilic end-group chemistries (–OH and –COOH) on gold, have also been employed to represent hydrophilic, yet not “brush-like” layers. Sodium chloride (NaCl) was selected for its practical importance and its monovalence; 1 M was chosen as the concentration of NaCl since previous studies on bulk aqueous PEG solution properties have shown that a noticeable change is expected only in the high-concentration regime [135, 137].

For tribological tests, the tribostress should be mild enough as to not induce desorption of PEG chains from the surface; if grafted PEG chains are desorbed due to tribostress, and then possibly readsorbed, the probed changes in frictional properties induced by changes in solvent quality can be complicated by effects involving the (re)adsorption kinetics of the polymer chains [138]. As is well known, tribological testing with conventional pin-on-disk tribometry generally leads to a rapid and irreversible failure of most adsorbed monomolecular organic films due to high local contact pressures at the multiple asperity contacts [83, 85, 139]. How- ever, by employing a soft rather than a rigid slider in a conventional pin-on-disk tribometer [117], non-destructive probing of the tribological properties of ultra-thin organic films can be readily achieved by virtue of the elastic pin deformation and low contact pressure generated within the contact area. While non-destructive, mild tribocontacts can also be achieved by nanotribological approaches, such as atomic force microscopy (AFM) [82, 89, 93, 140–143] or the surface forces apparatus (SFA) [1, 122], the current approach has the distinct advantage that the regions of the organic films within the contact area can be examined by spectroscopic approaches, due to the large contact area between pin and disk.

The tribometry setup described above, combined with the appropriate PEG- immobilization chemistry, thus enables the investigation of the influence of salts on the aqueous lubrication properties of brush-like PEG layers. It is emphasized that, due to the immobilization of PEG chains by a method that is impervious to changes in the salt concentration and tribostress, the probed frictional properties can be entirely ascribed to the conformational changes of the surface-grafted PEG chains arising from salt effects. In parallel with tribological measurements, the structure and stability of the brush-like PEG layer has been probed by standard surface-analytical tools, including polarization-modulation infrared reflection-absorption spectroscopy (PM-IRRAS), variable angle spectroscopic ellipsometry (VASE), and static contact angle measurements. The structural changes of the adsorbed polymer upon addition of salt have been probed by quartz-crystal microbalance (QCM-D), and can be correlated with the aqueous lubrication properties. 56 4.2. EXPERIMENTAL PROCEDURES

4.2 Experimental Procedures

4.2.1 Self-assembled monolayers (SAMs) on gold (Au)

Self-assembled monolayers (SAMs) on gold substrates were generated through spontaneous adsorption of thiol molecules (Figure 4.1) from absolute ethanol (Fluka, Switzerland). Besides α-methoxy-ω-mercapto poly(ethylene glycol) (m.w. 5000 Da) (1 in Figure 4.1), SAMs from α-methoxy-ω-mercapto hepta(ethylene glycol) (m.w. 356 Da) (2 in Figure 4.1) (Iris Biotech GmbH, Marktredwitz, Germany) as well as from two alkanethiol molecules, 1-mercaptoundecanol (3 in Figure 4.1) and 11- mercaptoundecanoic acid (4 in Figure 4.1) (Sigma-Aldrich, Switzerland) were prepared as reported in chapter 3 [117]. The substrates were immersed into 50 µM (1, 2) or 1 mM (3, 4) ethanolic thiol solutions for > 24 h. As a reference, (1) was also adsorbed from 50 µM ultrapure water (Millipore, 18.2 MΩcm). All chemicals were used as received, without further puriﬁcation. In the following, the tested SAMs are abbreviated as shown in Figure 4.1; α-methoxy-ω-mercapto poly(ethylene glycol) 5000 Da (1) as PEG5k, α-methoxy-ω-mercapto hepta(ethylene glycol) (2) as

EG7, 1-mercaptoundecanol (3) as C11OH, and 11-mercaptoundecanoic acid (4) as

C10COOH.

O 1 HS N O PEG5k H n O 2 HS EG7 7

3 HS OH C11OH

OH 4 HS C10COOH

Figure 4.1: Molecular structure of thiol molecules from which self-assembled monolayers were generated on polycrystalline gold substrates.

4.2.2 Poly(dimethyl siloxane) (PDMS) pins

For macroscopic pin-on-disk experiments, hemispherical poly(dimethyl siloxane) (PDMS) pins were fabricated as described in chapter 2. After fabrication, the pins were extracted and oxidized according to the protocol in chapter 2. The oxidized 4.2. EXPERIMENTAL PROCEDURES 57 pins, referred to as ox-PDMS pins in the following, displayed a water contact angle of < 3° and they were immediately employed in tribological measurements after plasma treatment.

4.2.3 Pin-on-disk tribometry

The macroscopic sliding friction between ox-PDMS pins and SAM-bearing substrates was measured with a pin-on-disk tribometer. All experiments were conducted in two aqueous solutions of varying salt concentration, namely 1 mM 4-(2- Hydroxyethyl)piperazine-1-ethanesulfonic acid (HEPES BioChemika Ultra, Fluka, Switzerland) in pure water (referred to as HEPES 0) and in its equivalents with 1 M sodium chloride (NaCl) (referred to as HEPES 0, 1 M NaCl). The frictional properties of the sliding contacts between the SAM films and ox-PDMS pins were initially characterized over a wide range of speed (from 19 to 0.1 mm/s) under a load of 1 N. For all measurements, the coefficients of friction (µ) were observed to be gradually increasing with decreasing sliding speed, whereas the µ values obtained from all samples were very similar at the highest speed. Consistent with previous studies [3, 117], the relative difference in detailed surface chemical/conformational characteristics was more clearly visualized at slow sliding speeds, i.e. in the mixed/boundary- lubrication regime. For this reason, in this work, the characterization of tribological properties has been focused on the slow-speed regime. The coefficient of friction (µ) was recorded over 10 rotations (6 mm track radius) at eight different sliding speeds within the same track, while the focus is placed on the values obtained from 0.2 mm/s experiments. Although the sliding tracks of some tribopairs were clearly discernable under an optical microscope, infrared spectroscopy confirmed that the SAM films on gold were virtually intact after tribostress. The raw data was acquired with the frequency of one data point per second, but for better visibility, the number of data points was filtered in such a way that 5 data points per rotation are shown in the following plots.

4.2.4 Contact angle measurements

Prior to pin-on-disk measurements, static water contact angles of all SAM ﬁlms were measured. The contact angles were determined immediately after the samples were taken out of the thiol solutions, rinsed with ethanol and blow dried with N2 gas. PEG5k and EG7 SAMs were additionally rinsed with ultra pure water to remove unbound or loosely bound molecules. The static contact angles of at least ﬁve samples with identical surface chemistry were measured and averaged. 58 4.2. EXPERIMENTAL PROCEDURES

4.2.5 Variable angle spectroscopic ellipsometry (VASE)

The film thicknesses of the adsorbed SAMs on Au in a dry state were characterized by ellipsometry at three different angles of incidence (65°, 70°, 75°). The measured data were fitted with the analysis software provided by the manufacturer (WVASE32, L.O.T. Oriel GmbH, Germany) and the spectral range considered was from 370 nm to 995 nm. The dry film thickness of the organic top layer with an assumed refractive index of 1.45 was extracted from the analysis of a three-layer model.

4.2.6 Polarization-modulation infrared reﬂection-absorption spectroscopy (PM-IRRAS)

The structural properties of the different SAMs on gold were analyzed by polarization- modulation infrared reflection-absorption spectroscopy (PM-IRRAS). PM-IRRA spectra of the substrates were first recorded after > 24 h of thiol self-assembly. In order to test the stability of the SAMs in the lubricants (under no tribostress), infrared spectra were further recorded after 2 h immersion of the samples into the two lubricants employed in this work, the immersion time corresponding to the duration of a pin-on-disk measurement.

4.2.7 Quartz-crystal microbalance with dissipation monitoring (QCM-D)

In order to gain specific insight into the conformation of the long-chain PEG thiols (PEG5k) and the related lubrication mechanism, quartz-crystal microbalance experiments with dissipation monitoring (QCM-D) were performed. After 4 h in situ adsorption of PEG5k SAMs from a 50 µM ethanolic solution on the gold- coated quartz crystal, the cell was flushed with ethanol to remove unbound thiol molecules. Subsequently, HEPES 0 was injected in order to mimic the conformation of the PEG5k SAMs in the aqueous lubricant. Thereafter, the low-salt aqueous solution was replaced with HEPES 0, 1 M NaCl in order for variations in mass and structural properties (i.e. hydration) of the films to be followed at higher salt concentrations. 4.3. RESULTS AND DISCUSSION 59

4.3 Results and Discussion

4.3.1 Sample characterization prior to tribological experiments

In order to verify the SAM formation on gold, the samples were characterized by various surface analytical tools prior to pin-on-disk experiments. The average static water contact angles from at least 5 samples are summarized in Table 4.1. They were determined to be 30 ± 1° for PEG5k (adsorbed from both ethanolic and aqueous solution), 69 ± 1° for EG7, 22 ± 2° for C11OH and 25 ± 3° for C10COOH, whereas the bare gold substrates exhibited a water contact angle of 82 ± 2°.

Table 4.1: Static water contact angles and ellipsometric dry thickness of the four thiol SAMs.

SAM CAH2O (°) dELM (nm) PEG5k (EtOH) 30 ± 1 8.22 ± 1.77 PEG5k (H2O) 30 ± 1 2.85 ± 0.26 EG7 69 ± 1 2.06 ± 0.04 C11OH 22 ± 2 1.57 ± 0.11 C10COOH 25 ± 3 1.62 ± 0.08

The distinctively lower water contact angles of the PEG5k, C11OH, and the

C10COOH SAMs compared to the bare gold substrates are consistent with the predominant exposure of hydrophilic moieties on the surface. The EG7 monolayers revealed a signiﬁcantly higher water contact angle compared to their long-chain analogs (PEG5k). It was previously reported that thiolated oligo-(ethylene glycol) SAMs adopt a well-ordered 7/2 helical structure on polycrystalline gold surfaces [144, 145]. The water contact angle observed on EG7 SAMs, which is more than two times higher than that obtained from PEG5k SAMs, could be a consequence of the predominant exposure of terminal methoxy groups to the surface in helically ordered EG7 SAMs compared to their random-coiled, long-chain analogs (PEG5k), which can expose the ether oxygens. The dry ﬁlm thickness of the SAMs on gold substrates measured by ellipsometry was found to be 8.22 ± 1.77 nm for PEG5k (EtOH) and 2.85 ± 0.26 nm for

PEG5k (H2O) samples, 2.06 ± 0.04 nm for EG7, 1.57 ± 0.11 nm for C11OH and 1.62 ± 0.08 nm for C10COOH SAMs, as shown in Table 4.1. The values are averaged over a minimum of 5 samples. Apparently, the dry thickness of the 5 kDa PEG SAMs adsorbed from ethanol is substantially higher compared to the other monolayers with film thickness values between 1.6 and 2.1 nm. Additionally, the PEG5k (EtOH) thickness value shows the largest error bar, which can be attributed 60 4.3. RESULTS AND DISCUSSION to the slightly fluctuating room temperature, which was shown to be very close to the cloud point of 50 µM PEG5k (EtOH) thiol solutions. In contrast, the PEG5k SAMs adsorbed from aqueous solution exhibit an almost three times lower film thickness compared to the monolayers adsorbed from ethanol, which is attributed to the solubility difference of PEG in the two solvents; poorer solubility of PEG in ethanolic solution (i.e. cloud point conditions) tends to induce a smaller radius of gyration of PEG chains in the bulk solution, which in turn facilitates a denser packing on the surface. The EG7 SAMs displayed ellipsometric thickness values that are in good agreement with the results from previous works. Vanderah et al. have calculated the theoretical thickness of oligo-(ethylene glycol) thiols with different structural conformations [144, 145]. According to these authors, the expected all- trans thickness of EG7 SAMs would be 2.85 nm, in contrast to the 7/2 helical conformation with a calculated thickness of 2.27 nm. The results obtained in this chapter, with average thickness values of 2.06 nm, thus suggest that EG7 SAMs exhibit a predominantly helical conformation. According to a model proposed by Sofia et al. [20], dry thickness values obtained from ellipsometry measurements can be converted into an approximate surface grafting density. As shown in the two-dimensional illustration in Figure 4.2, the definition of a unit cell (side lengths L and height dELM ) for each chain attached to the surface, allows for the calculation of the volume V occupied by a single chain through

2 Mw V = L · dELM = (4.1) ρdry · NA

O O O

O O O O z

O O O O O O

O O O O O O O O O O

O O

O O O O O

O O O O O O O O O O O

O O

O O O O O O O O O O O d O O ELM O O O O O O O O

O O O

O O O O

O O O O O

O O O O O O

O O x O Au

Figure 4.2: Two-dimensional schematic of close-packed unit cells, each bearing one polymer chain, for the derivation of the surface grafting density.

2 where L is the surface area occupied by a single chain, dELM is the dry thickness measured by ellipsometry, Mw is the molecular weight of the PEG chains, ρdry is the density of the dry monolayer and NA is the Avogadro’s number. The surface grafting density, i.e. the number of attached chains per unit area, can then be expressed as 4.3. RESULTS AND DISCUSSION 61

1 dELM · ρdry · NA 2 = . (4.2) L Mw In Equations 4.1 and 4.2, a constant value for the dry density of the monolayer

(ρdry) has to be assumed, i.e. close packing of attached chains on the surface is a prerequisite for the validity of this model. Assuming a value of 1 g/cm3 for the density of the dry PEG monolayer, the calculated surface density of the PEG5k SAMs was determined to be 1.02 ± 0.22 chains/nm2. This value is substantially higher than the grafting density of 5 kDa PEG monolayers adsorbed from aqueous solutions (0.35 ± 0.03 chains/nm2), which were shown to be in agreement with previously reported values [22]. The higher dry thickness of the monolayers was attributed to the poor solubility of PEG in ethanol as opposed to aqueous solutions, which tends to induce a smaller radius of gyration of PEG chains in bulk solution and in turn facilitates a denser packing on the surface. This value is still much lower than the calculated close-packed all-trans surface density of PEG5k SAMs on gold (5.88 chains/nm2) [22] indicating a considerable degree of disorder within the PEG5k monolayers. In comparison, the surface density of EG7 SAMs was found to be 3.49 ± 0.06 chains/nm2 and implies a more ordered structure of the short EG-monolayers. Vanderah et al. have calculated the theoretical thickness of oligo-(ethylene glycol) thiols with different structural conformations [144, 145]. According to these authors, the expected all-trans thickness of EG7 SAMs would be 2.85 nm. In contrast, the 7/2 helical conformation of EG7 SAMs would result in a theoretical thickness of 2.27 nm, which is equivalent to 3.84 chains/nm2. The results in this chapter, with average thickness values of 2.06 nm and a corresponding surface grafting density of 3.49 ± 0.06 chains/nm2, thus suggest that EG7 SAMs exhibit a predominant helical conformation. Infrared spectroscopy was employed to characterize the compositional and structural features of the SAM films on gold as well as to study the film stability in the aqueous lubricants. PM-IRRA spectra obtained after thiol adsorption from ethanolic (PEG5k (EtOH), EG7, C11OH and C10COOH) and aqueous (PEG5k (H2O)) solutions revealed the successful formation of self-assembled monolayers from all thiols. Figure 4.3 shows the PM-IRRA spectra that were recorded from the EG-based SAMs and the corresponding assignments of the peaks are summarized in Table 4.2. The spectra are normalized to their ellipsometric thickness to afford a better comparison of the relative intensities in the three spectra. Both PEG5k and EG7 SAMs adsorbed from ethanol exhibit strong bands in the high (3100 - 2700 cm−1) frequency region, arising from the C–H stretching modes, and in the low (1400 - 62 4.3. RESULTS AND DISCUSSION

PEG5k (EtOH) 1118

PEG5k (H 2O) EG7 (EtOH)

2892 PEG5k (EtOH) 964 1022 1145 O

O O O

O O O O

O O O O O O

O O O

O O

O O O O O 1346

O O O O O O O

O O

2981

O O O 2815 O

1242

O O

O O O O

2742 O 1280

O O O

O O O 1199 Au (x 3)

PEG5k (H 2O)

O O O O O O O O O O O O

O O O O O O O O

O O O O O O O

O O O O O O O O O O O O O O O O O O O O O O

O O O O

O O O O O O

O O

O O O O

O O O O O O O O O

O O O O O O O O

O O

O O O

O O O O Au Normalized Intensity

EG7

(x 3) Au

3200 3100 3000 2900 2800 2700 1400 1300 1200 1100 1000 900

Wavenumber (cm -1 )

Figure 4.3: PM-IRRA spectra recorded from PEG5k SAMs, adsorbed from both, ethanol and water, and from EG7 SAMs prior to pin-on-disk tribometry.

−1 900 cm ) frequency regions, due to the CH2 bending and the C–O, C–C stretching −1 modes. The bands at 1346, 1242, and 964 cm are assigned to the –CH2– wagging, twisting and rocking modes, respectively. The strong band at 1118 cm−1 is assigned to the C–O–C stretching mode. The positions of these bands in EG7 SAMs are similar to those reported for the ordered helical structure of the oligo-(ethylene glycol) SAMs [145]. The spectrum obtained from PEG5k monolayers is similar to that from EG7 SAMs, yet with some important differences. In the EG7 case, the C–O–C stretching band appears as a single peak with a maximum at 1118 cm−1, whereas in the PEG5k case, this peak exhibits a strong additional component at 1145 cm−1. The peak at 1118 cm−1 is assigned to a helical conformation of the EG units and the 1145 cm−1 component to a non-helical or trans conformation of the EG units [144–147]. This indicates that EG7 SAMs adopts a highly ordered helical structure, the PEG5k monolayer in contrast possessing a considerable amount of disorder. This is in agreement with the contact angle and ellipsometry data presented above. Further evidence of the difference in the structural ordering of the two monolayers can be seen from the relative intensities of the vibrational modes of the terminal methoxy group. The vibrational modes of the methyl C-H stretching −1 −1 modes at 2981 and 2815 cm and CH3 rocking mode at 1199 cm are observable in the EG7 spectrum while in the PEG5k spectrum these peaks are completely absent. This is again ascribed to the presence of an ordered structure in the EG7 monolayer, exposing the terminal methoxy groups at a well-defined orientation with 4.3. RESULTS AND DISCUSSION 63

Table 4.2: Assignment of peaks from PM-IRRAS spectra recorded for PEG5k and EG7 SAMs. PEG5k (cm−1) EG7 (cm−1) Assignment – 2981 CH3 asymmetric stretching 2889 2892 CH2 symmetric stretching – 2815 CH3 symmetric stretching 2742 2742 CH2 combination 1346 1346 CH2 wagging 1280 1284 CH2 twisting 1242 1242 CH2 twisting – 1199 CH3 rocking 1145 – C–O–C stretching (trans) 1118 1118 C–O–C stretching (helical) – 1022 964 975 CH2 rocking respect to the surface normal, whereas on the PEG5k sample the methoxy groups adopt a random orientation with respect to the surface normal leading to an intensity being too weak to be observed. Reflection-absorption infrared intensities are a consequence of both the surface concentration of the adsorbed species and their relative orientations with respect to the surface. The difference in surface grafting density alone is not sufficient to account for this huge difference in the intensity of these bands. The PEG5k SAMs adsorbed from aqueous solution show considerably reduced absorption bands in the high as well as in the low frequency region. Com- pared to ethanol adsorption, PEG5k (H2O) monolayers exhibit very broad peaks, which is indicative of a more random conformation of the SAM. This is most noticeable from the absence of a strong band at 1118 cm−1, which is assigned to the helical conformation.

Figure 4.4 shows that both alkanethiol SAMs (C11OH and C10COOH) exhibit well-deﬁned peaks around 2920 and 2850 cm−1, which are attributed to the asymmetric νa(CH2) and symmetric νs(CH2) methylene stretching modes, respectively, which is consistent with previous studies [109]. Additionally, the appearance of a −1 peak at 2880 cm (the symmetric methylene stretching mode of the CH2 group which is directly bound to hydroxyl oxygen) exclusively from C11OH and the ap- −1 − pearance of the peaks around 1740 (νC=O(CO2H)) and 1450 cm (νC−O(COO )) exclusively from C10COOH SAMs (data not shown) further support the presence of OH– and COOH– moieties on respective SAM surfaces. Infrared spectroscopic measurements carried out by exposing the samples to the two aqueous lubricants for the same duration of time as the tribological experiments, did not show any noticeable degradation of the SAM ﬁlms on this time scale (data 64 4.3. RESULTS AND DISCUSSION

C OH 11 ν (-(CH ) -) C 10 COOH a 2 n

ν s(-CH 2-OH)

ν s(-(CH 2)n-) Intensity

C11 OH C10 COOH

O OH O OH O OH O OH C C C C

Au Au

3150 3100 3050 3000 2950 2900 2850 2800 2750 -1 Wavenumber (cm )

Figure 4.4: PM-IRRA spectra recorded from C11OH and C10COOH alkanethiol SAMs prior to pin-on-disk tribometry. not shown). However, it should be borne in mind that these ex situ measurements only served to check the stability of the films after exposure to lubricants with high salt concentrations and do not reflect the differences in the structure of the films in solution upon the addition of the salts, as probed by the tribological and QCM-D measurements discussed below. To test such an effect of added salts on the chain conformations, it would require an in situ PM-IRRAS measurement of these films under high salt concentration, which is beyond the scope of the present investigation.

4.3.2 Aqueous lubricating properties in a low-salt environment (HEPES 0)

The lubricating properties of the SAM films were initially characterized in aqueous lubricants possessing virtually no salts (HEPES 0). Figure 4.5 shows the µ versus revolution plots of the SAM-bearing substrates in HEPES 0 (sliding speed = 0.2 mm/s). Compared with bare gold substrates (µ ≈ 0.40), all employed SAMs show a significant reduction in the coefficient of friction in HEPES 0; the relative µ values were observed to be in the order of EG7 ≈ C11OH > C10COOH > PEG5k. This order is not proportional to the water contact angle order of the SAM films (Table 4.1). A distinctive exception are the PEG5k monolayers, which revealed the lowest coefficient of friction (µ ≈ 0.03) among all the measured surfaces. As is well known, this is mainly ascribed to the “brush-like” conformation of the PEG chains in PEG5k SAMs, which allows for interactions with water molecules to most effectively generate a hydration layer at the tribological interface. Due to the 4.3. RESULTS AND DISCUSSION 65

1 HEPES 0

0.1

0.01

Coefficient of Friction PEG5k EG7

C11OH

C10COOH Au 0.001 0 1 2 3 4 5 6 7 8 9 10 Number of Rotations

Figure 4.5: Coefficient of friction versus revolution plots obtained from sliding ox-PDMS pins against the four different thiol SAMs on gold in HEPES 0. (normal load = 1 N, sliding speed = 0.2 mm/s). well-ordered structure and consequent hydrophobic surface characteristics, the EG7 SAMs cannot hold boundary lubricant water films as effectively as PEG5k SAMs.

The difference in the measured friction between the two alkanethiols (C11OH, and C10COOH) is also significant. In HEPES 0, C10COOH SAMs were found to provide a more lubricious interface together with ox-PDMS pins, resulting in µ values of around 0.05. The coefficient of friction of the hydroxyl-terminated alkanethiols (C11OH) was significantly higher (µ ≈ 0.12). While both SAM films show similar water contact angles, by substituting the hydroxyl-terminating group of the alkanethiol SAM with a carboxyl group (–COOH), the resulting µ values decrease significantly. This behavior is attributed to an additional electrostatic repulsion between the partially deprotonated C10COOH SAMs and ox-PDMS pins at pH 5.8 measured in HEPES 0.

4.3.3 Inﬂuence of high salt concentrations on the aqueous lubricating properties of the SAMs: HEPES 0, 1 M NaCl

The inﬂuence of salt concentration on the frictional response of the EG-based SAMs is shown in Figure 4.6. For both PEG5k and EG7 SAMs, a considerable increase in friction is observed in 1 M NaCl solutions (PEG5k: µ ≈ 0.03 → µ ≈ 0.06; EG7: µ ≈ 0.15 → µ ≈ 0.32). The relative increase in the coeﬃcient of friction by a factor of two is similarly 66 4.3. RESULTS AND DISCUSSION

1 Ethylene glycol SAMs

0.1

0.01 Coefficient of Friction PEG5k - HEPES 0 PEG5k - HEPES 0, 1M NaCl EG7 - HEPES 0 EG7 - HEPES 0, 1M NaCl 0.001 0 1 2 3 4 5 6 7 8 9 10 Number of Rotations

Figure 4.6: Comparison of the friction coefficients from ethylene glycol SAMs measured in HEPES 0 and HEPES 0, 1 M NaCl. (normal load = 1 N, sliding speed = 0.2 mm/s). high for both the short- and long-chain SAMs. The higher friction in aqueous solutions with a high salt concentration is ascribed to the “salting-out” effect, which has frequently been reported for PEG molecules in aqueous salt solutions [125–130]. This causes a reduction in hydration at the tribological interface comprised of end- tethered PEG chains, and consequently leads to a degraded lubricating effect as shown in Figure 4.6. For both PEG5k and EG7 SAMs, it is speculated that the interaction of salt with the two monolayers is similar, in that the reduction in hydration leads to a less fluid-like layer. While the highly extended PEG5k SAMs are believed to collapse upon the interaction with salts, it is speculated that the helically ordered EG7 SAMs can only become compressed to a certain extent, since the oligo-(ethylene glycol) SAMs are more densely packed compared to PEG5k films. It has been previously suggested [148] that EG7 SAMs are capable of incorporating water within their structure and “salting-out” and/or cation complexation does occur to some extent.

Figure 4.7 compares the µ values at 0.2 mm/s obtained from C11OH as well as from C10COOH SAMs in low and high ionic strength buffers (HEPES 0 and HEPES 0, 1 M NaCl). Similar to the observations made from EG-based SAMs, the frictional response of both alkanethiol SAMs is significantly higher in solutions with a 1 M salt concentration (C11OH: µ ≈ 0.12 → µ ≈ 0.32; C10COOH: µ ≈ 0.05 → µ ≈ 0.15). The relative increase in the friction coefficients between HEPES 0 and HEPES 0, 1 M NaCl is comparable for both alkanethiol SAMs and higher than the increase measured for EG SAMs, i.e. about a factor of three. The presence of sodium chloride is believed to influence the aqueous lubrication properties of C11OH 4.3. RESULTS AND DISCUSSION 67

1 Alkanethiol SAMs

0.1

0.01 Coefficient of Friction C11OH - HEPES 0

C11OH - HEPES 0, 1M NaCl

C10COOH - HEPES 0

C10COOH - HEPES 0, 1M NaCl 0.001 0 1 2 3 4 5 6 7 8 9 10 Number of Rotations

Figure 4.7: Comparison of the friction coeﬃcients from alkanethiol SAMs measured in HEPES 0 and HEPES 0, 1 M NaCl. (normal load = 1 N, sliding speed = 0.2 mm/s).

and C10COOH SAMs through screening of the hydrophilic terminating groups by hydrated ions, which eventually hinders the build-up of a separating lubricant ﬁlm between the sliding contacts. The interaction of hydrated ions with the hydrocarbon backbone can be excluded due to the high packing density of alkanethiol SAMs on gold. The fact that in 1 M NaCl solutions, the µ values from carboxylic SAMs

(C10COOH) are lower than those measured from C11OH, is believed to originate from less screening of the bulkier carboxylic terminating group by hydrated ions compared to the smaller hydroxyl groups of C11OH SAMs.

4.3.4 Inﬂuence of high salt concentrations on the conformation of surface-immobilized PEG SAMs

In order to understand the inﬂuence of salts on the conformation of end-tethered PEG chains (PEG5k SAM in this work) and subsequently to correlate this with their aqueous lubricating properties, QCM-D measurements were performed to elucidate the interaction of PEG5k monolayers with HEPES 0 and HEPES 0, 1 M NaCl. Additionally, a reference experiment on a blank gold crystal was carried out, since changes in the solvent properties (e.g. density, viscosity) can induce frequency shifts or dissipation changes on the uncoated QCM crystal. Figure 4.8 shows resonance frequency as well as dissipation changes of a PEG5k monolayer that was pre-adsorbed from ethanolic solution. The values measured from the reference crystal are subtracted from the displayed curves, and thus the presented frequency shifts and dissipation changes in Figure 4.8 are solely induced 68 4.3. RESULTS AND DISCUSSION

0 25 Ethanol HEPES 0 HEPES 0, 1 M NaCl HEPES 0 -20 20 -40 Dissipation (10 -60 15

-80

PEG5K - F3 10 -100 PEG5k - D3 -6 ) Frequency Shift (Hz) -120 5 -140

-160 0 220 240 260 280 300 320 340 360 Time (min)

Figure 4.8: Frequency shift and dissipation changes in QCM-D experiments observed for PEG5k SAMs, which were pre-adsorbed onto gold crystals from 50 µM ethanolic solutions. The induced changes occurred upon injection of different solvents. by the PEG5k SAM. Upon replacing ethanol (adsorption medium) with HEPES 0 at 240 minutes, an immediate frequency shift of -35 Hz and a change in dissipation of around 1 · 10−6 is observed for the third overtone of the quartz resonator. It is thus apparent that a significant hydration process is occurring upon injection of the aqueous lubricant. This indicates that a considerably higher amount of solvent is coupled to the adsorbed PEG5k SAM layer in the aqueous environment compared to ethanol. This is consistent with observations in previous studies [120, 121]. The dissipation, which relates to the viscoelasticity of the adsorbed layer, also confirms that the PEG5k monolayers are highly solvated in water, judging from the increase in the dissipated energy upon switching from ethanol to HEPES 0. After both frequency and dissipation signals had reached plateaus in HEPES 0, the high-salt aqueous lubricant (HEPES 0, 1 M NaCl) was injected (at ca. 285 minutes). Both frequency and dissipation changed immediately upon the transition from HEPES 0 (1 mM HEPES) to high salt containing buffer; the effective increase in frequency by 5 - 6 Hz indicates a decrease in the total mass coupled to the crystal in a high-ionic-strength environment. Similarly, the dissipation of the SAM-bearing substrate decreases by 0.6 · 10−6. In order to see whether the frequency shift is due to the loss of adsorbed PEG molecules on the surface or not, the HEPES 0, 1 M NaCl solution was again replaced with HEPES 0 (at ca. 345 minutes), upon which the frequency and dissipation values returned to their initial magnitudes. Thus, the observed changes in QCM-D experiments are fully reversible and therefore solely induced by the high salt concentration. 4.4. CONCLUSION 69

Hence, the conformational change of PEG chains observed in 1 M NaCl solutions are basically similar to those in ethanolic solution in that poorer solvency of a high-salt aqueous solution induces the loss of water within the monolayer and thus a partly collapse of the hydrated PEG5k brush. As mentioned in the introduction, effects due to “salting-out” as well as the complexation of cations with EG units have frequently been observed in various bulk properties of aqueous PEG solutions in high-ionic-strength media [125–128, 130]. The present observation confirms that such effects are at work for surface-tethered PEG5k SAMs. Ultimately, the reduction in hydration of the end-grafted polymer chains is thought to be mainly responsible for the poorer aqueous lubricating properties of PEG5k SAM films in the presence of high concentrations of NaCl. As expected, QCM-D experiments employing EG7 SAMs did not reveal sufficiently high frequency shifts to observe “salting-out” from the oligo-(ethylene glycol) SAMs upon exchange of HEPES 0 with HEPES 0, 1 M NaCl. This behavior is in agreement with the short and highly ordered helical chains of EG7 SAMs on Au, which only allow for a limited number of water- and/or cation-EG interactions.

4.4 Conclusion

It was demonstrated in this chapter that ethylene glycol-based self-assembled monolayers (SAMs) can be successfully employed as aqueous lubricant additives. It was shown that surface chemical as well as structural properties of the SAMs are largely governing the aqueous lubrication performance under low sliding speeds. High molecular weight PEG SAMs were shown to exhibit superior lubricating properties to oligo-(ethylene glycol) and alkanethiol SAMs, respectively. The less ordered PEG5k SAMs displayed a large hydration eﬀect after exposure to aqueous media and the thus created ﬂuid-like interface was shown to greatly reduce the frictional response. In contrast, highly ordered, shorter chain SAMs were found to be less effective in friction reduction, either because of their hydrophobic backbone (C11OH,

C10COOH) or because the interchain spacing did not allow for suﬃcient interaction of solvent molecules with the hydrophilic backbone (EG7). For these densely packed SAMs, the nature of the terminating group did largely inﬂuence the resulting friction. In a high salt environment, alkanethiol SAMs can only interact with hydrated ions at their terminating group, while “salting-out” of PEG5k SAMs was observed above a certain surface-density. Conversely, PEG monolayers with a lower surface-density did not show “salting-out” in 1 M NaCl solutions and yet, no increase in friction was observed. In summary, the aqueous lubrication performance of thiol 70 4.4. CONCLUSION

SAMs on gold was governed by the architecture of the individual thiol molecules and the resulting surface-chemical as well as structural properties of the ﬁlms. The detailed interaction of the SAMs with the lubricant was shown to determine the lubrication performance of each monolayer in aqueous environments of low and high ionic strengths, respectively. CHAPTER 5

Inﬂuence of Solution pH on the Aqueous Lubrication Properties of Thiol Self-assembled Monolayers (SAMs)

5.1 Introduction

In an approach to combine the advantages of macro- and nanoscale techniques, an experimental setup to non-destructively probe the aqueous lubrication performance of thiol self-assembled monolayers (SAMs) on gold was introduced in chapter 3 [117]. This has become feasible by employing a conventional macroscopic pin-on-disk setup with a soft rather than a rigid slider. In addition, by virtue of the large contact area resulting from the elastic deformation of the soft pin, spectroscopic analysis within the sliding track became possible. In chapters 3 and 4, surface-sensitive infrared spectroscopy (PM-IRRAS) conﬁrmed that, as a direct consequence of the reduced contact pressure, the SAMs are left virtually intact during tribostress under the load of 1 N in neutral aqueous lubricants, even at sliding speeds well below 1 mm/s. In the previous chapter, it was demonstrated that the macroscopic aqueous lubrication performance of alkanethiol SAMs as well as of ethylene glycol-based thiol SAMs is sensitive to the ionic strength of the lubricant. In the present chapter, the inﬂuence of the solution pH on the aqueous lubricating properties of the identical thiol SAMs is systematically investigated. In particular, it was observed that an alkaline aqueous solution (pH 12) appears to enhance the lubricating properties of various hydrophilic interfaces. Along with tribological experiments under low contact-pressures and sliding speeds, surface analytical information is obtained by standard surface characterization techniques such as surface-sensitive infrared spec-

71 72 5.2. EXPERIMENTAL PROCEDURES troscopy (PM-IRRAS), variable angle spectroscopic ellipsometry (VASE), quartz- crystal microbalance (QCM-D) and atomic force microscopy (AFM), from which the potential lubrication mechanisms have been hypothesized. Even though an un- equivocal explanation for the enhanced aqueous lubricating properties of hydrophilic interfaces in alkaline solutions has not been obtained at present, these observations are very interesting in that aqueous lubricating properties of hydrophilic interfaces can be further improved by manipulating solution parameters. Although a variety of experimental work was performed, some aspects concerning the inﬂuence of solution pH on the aqueous lubrication properties still remain incompletely understood for the investigated systems. However, a number of questions could be solved and it is expected that further experimental and theoretical work, as suggested in the conclusions, should contribute to a complete picture of the relation between lubrication performance and solution pH.

5.2 Experimental Procedures

5.2.1 Self-assembled monolayers (SAMs) on gold (Au)

Self-assembled monolayers (SAMs) of the thiol molecules described in chapter 4 were generated on gold substrates through spontaneous adsorption from ethanolic solutions. The details of the adsorption protocol can be found in chapter 4, from which the abbreviations of the employed SAMs were adopted, i.e. C11OH

(1-mercaptoundecanol), C10COOH (11-mercaptoundecanoic acid) (Sigma-Aldrich, Switzerland), PEG5k (α-methoxy-ω-mercapto poly(ethylene glycol) 5000 Da), and EG7 (α-methoxy-ω-mercapto hepta(ethylene glycol)) (Iris Biotech GmbH, Mark- tredwitz, Germany).

5.2.2 Poly(dimethyl siloxane) pins

For macroscopic pin-on-disk experiments, hemispherical poly(dimethyl siloxane) (PDMS) pins were fabricated and extracted in n-hexane as previously described in chapter 2. Prior to pin-on-disk experiments, the PDMS pins were air-plasma treated for 60 s at a gas pressure of 0.1 Torr. The oxidized pins, referred to as ox- PDMS pins in the following, displayed a resulting water contact angle of < 3° and were immediately employed in tribological measurements after plasma treatment. 5.2. EXPERIMENTAL PROCEDURES 73

5.2.3 Pin-on-disk tribometry

The macroscopic sliding friction between ox-PDMS pins and SAM-bearing substrates was measured with a pin-on-disk tribometer. The experiments were conducted in aqueous solutions of three distinct pH values, namely in 1 mM 4-(2- Hydroxyethyl)piperazine-1-ethanesulfonic acid (HEPES BioChemika Ultra, Fluka, Switzerland) in pure water (pH = 5.8, referred to as HEPES 0) and in its equivalents adjusted to pH 2 (referred to as HEPES 0, pH 2) and pH 12 (referred to as HEPES 0, pH 12) by addition of hydrochloric acid (HCl) and sodium hydroxide (NaOH), respectively. The coefficient of friction (µ) was recorded over 10 rotations (6 mm track radius) at a fixed sliding speed of 0.2 mm/s. The slow sliding speed was chosen in order to demonstrate the stability of the SAMs during tribostress. Additionally, on the same sliding track as the actual measurement was carried out, a running-in procedure of the ox-PDMS pin was performed prior to each experiment (6 different sliding speeds, 10 rotations each). While optical microscope images of the ox-PDMS pins showed signs of wear after the experiments, infrared spectroscopy confirmed that the SAM films on gold were still present after tribostress.

5.2.4 Water contact angle measurements

As previously described in chapter 4, the static water contact angles of all SAM ﬁlms were measured immediately after the samples were taken out of the thiol solutions, rinsed with ethanol and blown dried with N2 gas. PEG5k and EG7 SAMs were additionally rinsed with ultra pure water to remove unbound hydrophilic molecules. The static contact angles of at least ﬁve samples with identical surface chemistry were measured and averaged.

5.2.5 Variable angle spectroscopic ellipsometry (VASE)

The dry film thickness of the adsorbed SAMs on Au was characterized as described in chapter 4, i.e. by ellipsometry at different angles of incidence (65°, 70°, 75°). The measured data were fitted with the analysis software provided by the manufacturer (WVASE32, L.O.T. Oriel GmbH, Germany) and the spectral range considered was 370 nm to 995 nm. The dry film thickness of the organic top layer with an assumed refractive index of 1.45 was extracted from the analysis of a three-layer model. 74 5.3. RESULTS AND DISCUSSION

5.2.6 Polarization-modulation infrared reﬂection-absorption spectroscopy (PM-IRRAS)

The basic structural properties of the different SAMs determined by polarization- modulation infrared reflection-absorption spectroscopy (PM-IRRAS) were already discussed in chapter 4. On the basis of the experimental details in chapter 3, the sliding tracks of the PEG5k and EG7 SAM-bearing substrates were analyzed by PM-IRRAS after tribological experiments in aqueous lubricants with different solution pH.

5.2.7 Quartz-crystal microbalance with dissipation monitoring (QCM-D)

In order to gain speciﬁc insight into the interaction of the long-chain PEG thiols (PEG5k) with the aqueous lubricants with diﬀerent solution pH, quartz-crystal microbalance experiments with dissipation monitoring (QCM-D) were performed.

Both alkanethiols (C11OH and C10COOH) as well as the EG7 SAMs do not possess the sufficient molecular weights needed for the in situ investigation of the hydration properties by means of QCM-D. And indeed, the measurements did not display detectable changes in resonance frequency and dissipation upon adsorption, and thus the data are not presented. The PEG5k SAMs that were adsorbed onto the gold-coated quartz crystal from a 50 µM ethanolic solution were first exposed to HEPES 0 in order to mimic the conformation of the PEG5k SAMs in the standard aqueous lubricant. Thereafter, the two lubricants with a different solution pH, i.e. HEPES 0, pH 2 or HEPES 0, pH 12 were injected. In order to track whether the differences in mass and/or structural properties (i.e. hydration) of PEG5k SAMs are reversible, the two lubricants with an acidic or alkaline solution pH were replaced with HEPES 0 after stable frequency and dissipation values were obtained.

5.3 Results and Discussion

5.3.1 Sample characterization prior to tribological experiments

In order to verify the SAM formation on gold, the sample surfaces were characterized by surface-analytical tools prior to tribological experiments. These results were discussed in section 4.3.1 of chapter 4. 5.3. RESULTS AND DISCUSSION 75

5.3.2 Pin-on-disk tribometry

After characterization of the four SAM films, their interfacial frictional properties against an elastomeric slider (oxidized, hemispherical PDMS, radius = 3 mm) were investigated with a pin-on-disk tribometer by acquiring the coefficient of friction in three aqueous solutions; (1) low ionic strength aqueous buffer solutions (HEPES 0, pH = 5.8), (2) aqueous buffer solutions with pH 2 (HEPES 0, pH 2), and (3) aqueous buffer solutions with pH 12 (HEPES 0, pH 12). The normal load (1 N) and the sliding speed (0.2 mm/s) were kept constant for all pin-on-disk experiments. It is noted here again, that prior to each measurement, the running-in procedure of the ox-PDMS pin was performed at different sliding speeds inside the same sliding track in which the subsequent experiment was carried out. In the following, the results are presented as µ-versus-revolution plots obtained from identical SAM films under varying aqueous solution parameters (Figures 5.1, 5.2, 5.4, and 5.5); these plots highlight the influence of the aqueous pH on the lubricating properties of each SAM film.

Figure 5.1 compares the µ values obtained from C11OH SAMs in the three aqueous lubricants employed in this chapter. The frictional properties of the C11OH SAMs/ox-PDMS sliding contacts are very similar in near-neutral (HEPES 0) and acidic aqueous solution (HEPES 0, pH 2), i.e. µ ≈ 0.10. Considerably lower friction coeﬃcients were measured in alkaline solution (HEPES 0, pH 12) where

µ ≈ 0.04. For C11OH SAMs, the friction reduction at pH 12 compared to HEPES 0 with a solution pH of 5.8 is believed to partly originate from electrostatic interactions. In alkaline solutions, the oxidized PDMS pins as well as the hydroxyl- terminated SAMs are deprotonated to a certain degree. The resulting repulsion between the two negatively charged surfaces is expected to support the separation of both surfaces and thus facilitates sliding. At pH 2, however, the neutral OH moieties on both the oxidized PDMS and the C11OH surface gave rise to an almost identical lubrication behavior compared to HEPES 0 (pH = 5.8).

For C10COOH SAMs, as shown in Figure 5.2, the variation in solution pH towards lower values (HEPES 0, pH 2) appeared to have a different influence on the frictional properties compared to C11OH SAMs. Experiments performed in near-neutral buffers (HEPES 0) resulted in coefficients of friction of around 0.05, while considerably lower friction (µ ≈ 0.03) was measured in an acidic environment (HEPES 0, pH 2). One possible explanation could be that the degree of ionization for the carboxyl SAMs as well as for the ox-PDMS pins is very low at pH 2. Under these conditions, carboxylic acid-terminated SAMs (C10COOH) are believed to interact more strongly with water molecules, in a sense that one water molecule 76 5.3. RESULTS AND DISCUSSION

C11OH HEPES 0 1 HEPES 0, pH 2 HEPES 0, pH 12

0.1 Coefficient of Friction

0.01 0 1 2 3 4 5 6 7 8 9 10 Number of Rotations

Figure 5.1: µ-versus-revolution plots obtained from C11OH SAMs on gold against ox- PDMS pins in three aqueous lubricants of distinct solution pH. (normal load = 1 N, sliding speed = 0.2 mm/s) can establish two hydrogen bonds with a carboxyl group, which should result in the formation of a more stable, molecular water ﬁlm in contrast to the ionized state.

These characteristics could also explain the improved lubrication of C10COOH

(µ ≈ 0.03) versus C11OH SAMs (µ ≈ 0.10) at pH 2.

C10COOH HEPES 0 HEPES 0, pH 2

0.1 Coefficient of Friction

0.01 0 1 2 3 4 5 6 7 8 9 10 Number of Rotations

Figure 5.2: µ-versus-revolution plots obtained from C10COOH SAMs on gold against ox-PDMS pins in three aqueous lubricants of distinct solution pH. (normal load = 1 N, sliding speed = 0.2 mm/s)

Pin-on-disk data from C10COOH SAMs in an alkaline lubricant could not be collected due to disruption of the samples during the running-in procedure. The reason is believed to stem from the presence of sodium hydroxide (NaOH) in HEPES

5.3. RESULTS AND DISCUSSION 77

O O

C O O O

+ Na

O OH + NaOH O O C C

Figure 5.3: Presumed acid-base reaction between C10COOH SAMs and ox-PDMS pins in an alkaline environment. This condition has repeatedly resulted in the disrupture of pins and disks during the running-in procedure in pin-on-disk experiments performed in HEPES 0, pH 12.

0, pH 12. At pH 12, the carboxyl-terminating groups of the C10COOH monolayers are likely to undergo the acid/base reaction illustrated in Figure 5.3 to result in the sodium acetate. The latter is positively charged and is most likely to strongly interact with the negatively charged ox-PDMS pin at pH 12. The resulting electrostatic bridging between pin and disk is thus believed to be the reason for the high friction and sample damage observed in alkaline solutions. The influence of the aqueous lubricant composition on the lubricating properties is also clearly manifested for EG7 SAMs. As shown in Figure 5.4, extremely low µ values of below 0.02 were observed at alkaline pH (HEPES 0, pH 12), rendering this condition most favorable for the aqueous lubrication of EG7 SAMs. The acidic aqueous solution (HEPES 0, pH 2) and HEPES 0 showed identically high friction (µ ≈ 0.12), with the µ values being almost one order of magnitude higher compared to the pH 12 case. Since it was not evident why the alkaline lubricant is beneficial for the ox-PDMS/EG7 SAM tribopair, e.g. the EG7 monolayer does not carry any charges at pH 12, a number of additional control experiments were performed. These experiments are discussed in the next section. For the PEG5k SAMs, as shown in Figure 5.5, the frictional properties were generally maintained at a very low level (µ ≈ 0.02 - 0.06) for all pH conditions compared to the other sample surfaces. Nevertheless, the experiments in alkaline solutions (HEPES 0, pH 12) again resulted in the lowest friction coefficients (µ ≈ 0.016), whereas the measurements in acidic solutions (HEPES 0, pH 2) resulted in much higher µ values of 0.06. In HEPES 0, medium coefficients of friction were measured, i.e. µ ≈ 0.03. The long-chain PEG5k SAMs did respond to the alkaline environment in a similar way to EG7 monolayers, yet with less friction reduction 78 5.3. RESULTS AND DISCUSSION

1 EG7 HEPES 0 HEPES 0, pH 2 HEPES 0, pH 12

0.1 Coefficient of Friction

0.01 0 1 2 3 4 5 6 7 8 9 10 Number of Rotations

Figure 5.4: µ-versus-revolution data obtained from EG7 SAMs on gold against ox-PDMS pins in three aqueous lubricants of distinct solution pH. (normal load = 1 N, sliding speed = 0.2 mm/s)

PEG5k 0.1

0.01 Coefficient of Friction

HEPES 0 HEPES 0, pH 2 HEPES 0, pH 12 0.001 0 1 2 3 4 5 6 7 8 9 10 Number of Rotations

Figure 5.5: µ-versus-revolution data obtained from PEG5k SAMs on gold against ox- PDMS pins in three aqueous lubricants of distinct solution pH. (normal load = 1 N, sliding speed = 0.2 mm/s) relative to HEPES 0. However, PEG5k SAMs did react much more sensitive to acidic pH conditions compared to EG7 monolayers. In order to investigate the influence of pH on the aqueous lubricating properties of PEG5k SAMs, QCM-D measurements were performed to elucidate the interaction of PEG5k molecules with aqueous lubricants of different solution pH. In the following plots, the response from simultaneously obtained control experiments on a blank gold crystal is subtracted. Thus, the changes in frequency and dissipation can be solely attributed to the direct interaction of the PEG5k SAM with the aqueous 5.3. RESULTS AND DISCUSSION 79 lubricant. The large hydration of PEG5k films upon changing from the adsorption environment (ethanol) to HEPES 0 has already been illustrated in chapter 4 (Figure 4.8, page 68). Upon replacement of HEPES 0 with the alkaline lubricant (HEPES 0, pH 12), the changes of both resonance frequency and dissipation were very small, as shown in Figure 5.6. Because the original frequency and dissipation values were obtained after replacing HEPES 0, pH 12 with HEPES 0, QCM-D experiments revealed that PEG5k SAMs remain stable at alkaline pH. For experiments with the acidic lubricant (HEPES 0, pH 2), almost identical results were obtained, as shown in Figure 5.7. Hence, QCM-D measurements in alkaline as well as in acidic aqueous solution did not reveal significant changes in hydration as well as conformational properties of the PEG5k SAMs.

0 25 HEPES 0 HEPES 0, pH 12 HEPES 0 -20 20 -40 Dissipation (10 -60 15

-80 PEG5K (EtOH) - F3 PEG5k (EtOH) - D3 10 -100 -6 ) Frequency Shift (Hz) -120 5 -140

-160 0 360 380 400 420 440 460 Time (min)

Figure 5.6: Changes in resonance frequency and energy dissipation of a PEG5k SAM after exposure of the monolayer to the alkaline lubricant (HEPES 0, pH 12). QCM-D experiments did not reveal signiﬁcant variations in the adsorbed mass or in structural conformations at pH 12.

+ − Due to the presence of excess (10 mM) hydronium (H3O ) and hydroxyl (OH ) ions at pH 2 and 12, respectively, one could suspect that the aqueous lubricating properties of PEG chains can be influenced by a “salting-out” effect. According to the “salting-out” effectiveness of ions towards PEG [149], OH− is known to be most effective among the common ions in the “salting-out” of PEG from aqueous solutions. One would therefore expect that PEG5k SAMs adopt a more collapsed conformation in the presence of OH− ions, i.e. at alkaline solution pH. In contrast, H+ ions are known to be least effective in the “salting-out” of PEG, and thus, the conformation of PEG chains in an aqueous acidic environment should be virtually unchanged compared to neutral pH conditions. Thus, one would expect higher µ values from PEG5k SAMs in HEPES 0, pH 12 than in the pH 2 solution if the 80 5.3. RESULTS AND DISCUSSION

0 25 HEPES 0 HEPES 0, pH 2 HEPES 0 -20 20 -40 Dissipation (10 -60 15

-80 PEG5K (EtOH) - F3 PEG5k (EtOH) - D3 10 -100 -6 ) Frequency Shift (Hz) -120 5 -140

-160 0 440 460 480 500 520 Time (min)

Figure 5.7: Changes in resonance frequency and energy dissipation of a PEG5k SAM after exposure of the monolayer to the acidic lubricant (HEPES 0, pH 2). QCM-D experiments did not reveal signiﬁcant variations in the adsorbed mass or in structural conformations at pH 2.

“salting-out” effect is the dominating mechanism, as with the case of NaCl shown in chapter 4. However, the experimental observations in this chapter are directly opposite, i.e. the friction forces of PEG5k SAMs are considerably lower in alkaline compared to acidic solutions. In fact, the concentration of hydronium and hydroxyl ions, 10 mM, is too low to induce distinct conformational changes of surface-tethered PEG chains. Most of all, since QCM-D experiments did not reveal different conformations or changes in hydration of the PEG5k SAMs in either lubricant, the “salting-out” effect could be excluded as the dominant mechanism responsible for the changes in the lubricating properties of PEG5k SAMs in HEPES 0, pH 12 and HEPES 0, pH 2. In order to explore the stability of EG7 and PEG5k monolayers under tribological stress, PM-IRRA spectra were recorded before and after pin-on-disk experiments. The sliding track that was subjected to tribological stress could be analyzed by positioning different regions of the sample in the beam path, as described in chapter 3. This method allows for the recording of IR spectra across the sample. However, due to the spot size of the IR beam on the surface, it is difficult to avoid the inclusion of small areas outside the sliding track in the spectrum. Nevertheless, this method is believed to serve as a reasonably accurate and fast tool to make ex situ investigations of the tribologically stressed area within the sliding track. Figures 5.8 and 5.9 show the PM-IRRA spectra recorded from EG7 and PEG5k SAMs prior and after pin-on-disk experiments in the different lubricants. In HEPES 0 and HEPES 0, pH 12, only minor differences between the initial spectra and 5.3. RESULTS AND DISCUSSION 81

Before tribotest EG7 After tribotest

(x 5) HEPES 0

(x 5) HEPES 0, pH 2 Intensity (a.u.)

(x 5) HEPES 0, pH 12

3100 3000 2900 2800 2700 1400 1300 1200 1100 1000 -1 Wavenumber (cm )

Figure 5.8: PM-IRRA spectra recorded from EG7 SAMs prior (solid line) and after (dashed line) pin-on-disk experiments in three diﬀerent lubricants.

Before tribotest PEG5k After tribotest

(x 2.7) HEPES 0

(x 2.7) HEPES 0, pH 2 Intensity (a.u.)

(x 2.7) HEPES 0, pH 12

3100 3000 2900 2800 2700 1400 1300 1200 1100 1000 -1 Wavenumber (cm )

Figure 5.9: PM-IRRA spectra recorded from PEG5k SAMs prior (solid line) and after (dashed line) pin-on-disk experiments in three different lubricants. those recorded after tribostress could be observed for both SAMs. For pin-on- disk experiments performed in the acidic lubricant (HEPES 0, pH 2), however, EG7 and PEG5k monolayers revealed significantly reduced IR intensities inside the sliding track, as well as a certain broadening of distinct peaks. These characteristics are typical for a loss of material and/or for disordered monolayers. While QCM- D experiments from PEG5k monolayers exposed to HEPES 0, pH 2 did not reveal differences in the adsorbed mass or in the conformation of the SAM, IR data indicate that the stability of PEG5k monolayers in an acidic environment is reduced in the presence of tribostress. While the friction measured from PEG5k SAMs was 82 5.3. RESULTS AND DISCUSSION higher in the acidic lubricant compared to HEPES 0, EG7 SAMs exhibited nearly identical friction coefficients in HEPES 0 and in HEPES 0, pH 2, although the IR spectrum of EG7 SAMs also shows reduced intensities after the pin-on-disk measurement in the acidic lubricant. However, in chapter 4 it became evident that PEG5k and EG7 SAMs display significantly different surface grafting densities and as a result, the loss of molecules from either monolayer is not expected to have the same impact on the frictional properties of PEG5k and EG7 SAMs. That is, the removal of only few molecules from the less densely packed PEG5k monolayers might lead to a partial exposure of the underlying substrate and hence to higher friction.

5.3.3 Control experiments

In order to investigate the favorable lubricating properties of all, except C10COOH SAMs in alkaline solution, a number of control experiments were performed. Figure 5.10 shows the µ-versus-sliding-speed plot obtained from ox-PDMS pins against bare gold substrates. In the absence of self-assembled monolayers, it is apparent that measurements in HEPES 0, pH 12 led to considerably lower friction coefficients compared to experiments in HEPES 0, especially towards slower sliding speeds. Therefore, it is speculated that alkaline solutions might have a beneficial influence on the ox-PDMS slider.

Au HEPES 0 HEPES 0, pH 12 1

0.1 Coefficient of Friction

0.01 0.1 1 10 Sliding Speed (mm/s)

Figure 5.10: µ-versus-sliding-speed plot obtained from ox-PDMS pins sliding against bare gold substrates in HEPES 0 and in HEPES 0, pH 12.

The inﬂuence of HEPES 0, pH 12 on the tribological properties of oxidized PDMS surfaces was examined with two additional pin-on-disk experiments. Figure 5.3. RESULTS AND DISCUSSION 83

1 Normal load: 10 N 1. H0, r = 4mm Sliding speed: 0.25 mm/s 2. H0pH12, r = 4mm 3. H0, r = 4mm Coefficient of Friction

0.1 0 5 10 15 20 25 30 Number of Rotations

Figure 5.11: Pin-on-disk data obtained from two ox-PDMS contacts in HEPES 0 and HEPES 0, pH 12. After 30 rotations in HEPES 0 (black circles), the lubricant was exchanged with the alkaline equivalent before another 30 rotations were performed within the same sliding track (empty circles). The last measurement was again carried out in HEPES 0 (grey circles).

5.11 displays the results of a pin-on-disk experiment with two ox-PDMS tribopairs in near-neutral (HEPES 0) and alkaline (HEPES 0, pH 12) aqueous solutions. As indicated in the legend of Figure 5.11, three sliding experiments (30 rotations each) were carried out within the same sliding track at 0.25 mm/s sliding speed and under 10 N normal load. After a running-in procedure (25 mm/s, 30 rotations), the first 30 rotations were performed in HEPES 0, after which the aqueous lubricant was exchanged with its alkaline equivalent (HEPES 0, pH 12). The reduction in the friction coefficient from µ ≈ 0.35 in HEPES 0 to µ ≈ 0.15 in the alkaline lubricant is apparent and occurs immediately. After replacing HEPES 0, pH 12 with the neutral solution, the friction increases steadily until it almost reaches the original µ values obtained from the first measurement in HEPES 0. When the identical experiment was performed with untreated, hydrophobic PDMS contacts, no influence of the alkaline solution on the friction coefficient was detected (Figure 5.12). Hence, the silica-like oxide layer generated by the air-plasma treatment of PDMS is believed to significantly contribute to the low friction coefficients observed from experiments in alkaline aqueous solutions. The influence of solution pH on ox-PDMS surfaces in the absence of tribological stress was analyzed by means of PM-IRRAS experiments. For this purpose, thin PDMS films were spin-coated on top of gold substrates, oxidized by air-plasma treatment and subsequently immersed in water or in its equivalents adjusted to pH 2 (with HCl) or pH 12 (with NaOH). 84 5.3. RESULTS AND DISCUSSION

1 Normal load: 10 N 1.PDMS-PDMS, pH 7, 4mm Sliding speed: 0.25 mm/s 2.PDMS-PDMS, pH 12, 4mm Coefficient of Friction

0.1 0 5 10 15 20 25 30 Number of Rotations

Figure 5.12: Pin-on-disk data obtained from two untreated PDMS contacts in HEPES 0 and HEPES 0, pH 12. After 30 rotations in HEPES 0 (black circles), the lubricant was exchanged with the alkaline equivalent and another 30 rotations were performed within the same sliding track (empty circles). Nearly identical µ values were obtained from both lubricants.

PDMS ox-PDMS ox-PDMS (after 2h immersion)

(x 11) H2O

H O, pH 2 (x 11) 2 Intensity (a.u.)

(x 11) H2O, pH 12

3050 3000 2950 2900 2850 1400 1200 1000 -1 Wavenumber (cm )

Figure 5.13: PM-IRRA spectra obtained from spin-coated PDMS layers (solid black lines) on top of gold substrates after 60 s air-plasma treatment (dashed black lines) and after 2 h immersion into aqueous solutions of diﬀerent pH (solid grey lines).

From Figure 5.13 it is apparent that the oxidation of PDMS leads to a decrease of the intensity bands in the high wavelength region, i.e. at 2965 cm−1 and 2906 cm−1. These bands are assigned to symmetric and asymmetric methyl stretching vibrations

(νs,a(CH3)), respectively, and they are expected to decrease due to the introduction of oxygen species on the surface of crosslinked PDMS [58]. Compared to untreated

PDMS, the oxidized sample shows a slight reduction of the symmetric Si-CH3 defor- −1 mation modes (νs(Si-CH3)) at 1265 cm , which is also attributed to the formation 5.3. RESULTS AND DISCUSSION 85 of surface silanols during plasma treatment. In addition, the ox-PDMS surface exhibits a reduced intensity of the asymmetric Si-O-Si stretching modes (1114 cm−1) as well as a slight shift towards higher wavenumbers in comparison to untreated PDMS. As shown in Figure 5.13, a broad absorption band emerges around 1200 cm−1 after air-plasma treatment. These two characteristics have previously been attributed to the decomposition of PDMS during oxidation via chain scission, as well as to the formation of short hydrocarbon linkages (e.g. -Si-(CH2)2-Si-) between PDMS chains [57, 150]. After immersing the PDMS samples in aqueous solutions of diﬀerent pH values, it is obvious that H2O and its acidic equivalent did not alter the properties of the thin ox-PDMS layer. In contrast, the immersion in an alkaline solution led to an increase of the asymmetric Si-O-Si absorption band around 1114 cm−1 and to a reduced intensity in the 1200 - 1250 cm−1 region compared to the freshly oxidized PDMS sample.

Hence, the bottom spectra (H2O, pH 12) in Figure 5.13 points towards an increase in the number of siloxane bonds in the alkaline environment whereas the hydrocarbon linkages between PDMS chains seem to decrease. It is hypothesized − that hydroxyl ions (OH ) are capable of cleaving -Si-(CH2)2-Si- linkages via the formation of silanol (Si-OH) bonds. The subsequent condensation of such silanols to siloxane bonds could be responsible for the increase of the Si-O-Si stretching band after the immersion of ox-PDMS samples in alkaline solution. On the other hand, it is well known that the exposure of silica surfaces to alkaline solutions results in so-called “wet-chemical etching”, i.e. in the generation of silanol (Si-OH) moieties after cleavage of siloxane bonds [151–153].

Si − O − Si + OH− → Si − OH + Si − O− (5.1)

According to Equation 5.1, a reduced intensity of the asymmetric Si-O-Si stretching band would be expected after the immersion of ox-PDMS samples in an alkaline solution. In Figure 5.13, however, an increase in the number of siloxane bonds was observed, after ox-PDMS surfaces were immersed in aqueous NaOH solution. In the presence of tribological stress, the brittle oxide layer of the ox-PDMS pins is subjected to a certain degree of wear. Under these circumstances, the alkaline solution appears to be able to maintain the hydrophilicity of the ox-PDMS surface via the regeneration of silanol groups from disrupted siloxane bonds. Furthermore, the condensation of silanols to siloxane bonds could lead to a self-healing effect of the brittle SiOx layer of ox-PDMS pins. In summary, the performed control experiments verified the beneficial influence of the alkaline solution on the silica-like surface of ox-PDMS samples. Whereas untreated PDMS samples did not show sig- 86 5.4. CONCLUSIONS nificantly distinct friction from experiments performed in HEPES 0 and HEPES 0, pH 12, the latter solution was found to enhance the lubrication performance of oxidized PDMS samples. The infrared spectra of ox-PDMS samples immersed into alkaline solutions additionally confirmed that HEPES 0, pH 12 can alter the surface properties of ox-PDMS surfaces in the absence of tribological stress.

5.4 Conclusions

In this chapter, the previous investigations on the aqueous lubrication behavior of self-assembled monolayers (SAMs) on gold were extended. By employing the experimental setup introduced in chapter 3 [117], macroscopic pin-on-disk experiments were performed under low contact pressures in aqueous environments of different solution pH. As was previously demonstrated, the contact pressure was easily maintained below 1 MPa when a soft elastomeric slider (ox-PDMS) was utilized, which allowed for the non-destructive tribological testing of the SAMs on polycrystalline gold at low sliding speeds (0.2 mm/s). The solution pH of the lubricant was found to have a considerable impact on the frictional properties of the individual SAMs. Compared to the near-neutral aqueous solution (HEPES 0), the acidic and the alkaline lubricant did not act like poor or good solvents per se. For PEG5k SAMs, for instance, QCM-D experiments revealed that the different lubricants did not induce detectable changes in the hydration and/or conformational properties of the SAMs compared to HEPES 0. On the other hand, ex situ PM-IRRAS experiments inside the sliding track of the EG- based monolayers after tribological stress suggested a relatively inferior stability in HEPES 0, pH 2 compared to experiments in the neutral and the alkaline lubricant. The stability of the SAMs, however, could not account for the generally observed low friction in HEPES 0, pH 12. Therefore, a set of control experiments was performed to analyze the interaction of ox-PDMS pins with the different aqueous lubricants. It was found that the alkaline solution has a beneficial influence on the brittle oxide layer of the PDMS pins. However, since the relative magnitude of the enhanced lubrication effect in the alkaline lubricant was different for each SAM, it was necessary to investigate the detailed interaction of the lubricant with the individual monolayers. Hence, with the experimental work performed in this chapter it was not possible to fully elucidate the effective lubrication of different SAM/ox- PDMS tribopairs in alkaline solution. While a number of questions were approached successfully within this study, some more experimental as well as theoretical work is needed to fully elucidate the underlying lubrication mechanisms for the studied 5.4. CONCLUSIONS 87 tribological systems. It is expected that the answers to the following questions are necessary to complete the picture of the enhanced lubrication performance in alkaline aqueous solutions:

• What are the diﬀerences between the lubrication mechanisms for PEG-based and hydroxyl (OH)-terminated surfaces in alkaline solutions?

• How do surface hydroxyl groups (Si-OH) inﬂuence the intrinsic tribological

properties of SiO2 surfaces?

• Would alkaline lubricants also facilitate the sliding between two hard contacts with a hydrophilic yet not oxidized surface?

In order to answer these questions and to gain a more general insight into the lubricity of alkaline solutions, the following experimental as well as theoretical work is considered to be necessary:

• Utilization of a soft, hydrophilic pin without an oxide layer

• Further variation of the concentration of NaOH and HCl

• Employment of alkaline solutions other than NaOH

• Theoretical simulation of the hydration structure around PEG in the presence of Na+, OH− as well as H+, Cl−

To summarize, the complete understanding of the detailed interaction between the lubricants and the SAM/ox-PDMS tribopairs employed in this work, is expected to provide important information for the design of tribological systems with low frictional properties in an aqueous environment. In terms of EG-based monolayers, previous studies have mainly reported on the degradation of aqueous lubrication properties in poor lubricants, e.g. at high salt concentrations or at elevated temperatures. In this context, the present investigations are quite unique because the alkaline solution was found to enhance the lubrication properties of EG-based monolayers compared to near-neutral aqueous solutions. Even though alkaline solutions are not desirable in many tribosystems, the regeneration of the lubrication mechanism under more favorable solution conditions would be highly attractive, but this presumes a profound knowledge of the interactions between the lubricant and both tribological contacts. 88 5.4. CONCLUSIONS CHAPTER 6

Controlled Growth of Poly(methacrylic acid) (PMAA) Brushes from Silicon Surfaces via UV-LED-initiated Photopolymerization - Synthesis and Aqueous Lubrication Performance

6.1 Introduction

Surface modiﬁcations by means of polymer brushes represent a very attractive tool for the tailoring and control of interfacial properties such as adhesion, friction, wettability or biocompatibility [1, 119, 154]. Two principal experimental approaches are available to generate polymer brushes on solid substrates, namely “grafting to” and “grafting from” [40, 155]. While both approaches have their own unique characteristics, “grafting from” methods are known to be more advantageous for producing a high grafting density of chains, because the polymers are generated in situ from a surface with a high density of initiating species. On the other hand, while “grafting to” approaches are more experimentally straightforward, they are restricted in terms of grafting densities, mainly due to the diﬀusion-limited adsorption of preformed polymers [156]. In recent years, numerous experimental strategies have been developed for the preparation of polymer brushes by means of “grafting from” methods [40, 157, 158]. Among them, controlled radical polymerization (CRP) strategies such as atom transfer radical polymerization (ATRP) [159], nitroxide-mediated polymerization (NMP) [160–162], reversible-addition fragmentation transfer polymerization (RAFT) [163, 164], and photoiniferter-mediated photopolymerization (PMP)

89 90 6.1. INTRODUCTION

[165, 166] have gained considerable attention. The common feature of these methods is that the propagating chain continuously experiences activation-deactivation cycles to maintain a low radical concentration, which in turn minimizes irreversible chain termination. Therefore, CRP methods allow for the precise control of the polymer molecular weight and usually yield polymer brushes with low polydispersity. In addition, the preparation of block copolymer brushes becomes feasible as the active chain ends are typically preserved when a polymerization step is interrupted.

In this work, the photoiniferter-mediated photopolymerization (PMP) method originally developed by Otsu et al. [165, 166] was adapted to prepare poly(methacrylic acid) (PMAA) brushes on silicon surfaces, covered with native SiO2. The PMP method is based on dithiocarbamate derivatives, which act as initiator, transfer agent and terminating species (iniferter). Upon ultraviolet (UV) irradiation, the iniferter dissociates into a highly reactive radical as well as a non-initiating counter radical, which acts as transfer agent and reversibly terminates the propagation reaction. Advantages of the iniferter concept include the facile control of the polymerization reaction by means of irradiation time and UV intensity, the comparatively fast reaction kinetics compared to other CRP methods, as well as the fact that the polymerization can be easily performed at room temperature or below to avoid thermal polymerization of heat-sensitive monomers. Furthermore, the photoiniferter technique is amenable to aqueous media, it is suitable for micropatterning [167, 168] and it does not require any sacriﬁcial initiator in the monomer solution. The latter is signiﬁcant, since it limits the formation of free polymers in the bulk solution, and eliminates the need for extensive cleaning steps after brush formation.

The UV source employed in this work consisted of a recently developed, commercial high-power ultraviolet light-emitting diode (UV-LED) [169–171], and it was chosen for its distinct advantages compared to conventional UV sources. The main advantage of UV-LEDs is their intrinsically narrow emission spectrum. Light emitting diodes consist of a p-n junction formed by two semiconducting materials with a direct band gap. Since the emission wavelength is directly correlated to the band gap of the p-n junction, it is feasible to adjust the emission spectrum of UV-LEDs via doping of appropriate semiconducting materials. Thus, the sharp spectrum of these novel UV-LEDs allows for a specific selection of the wavelength region according to the employed photoinitiating system, and the reactive monomer, respectively. In comparison to conventional mercury lamps, for instance, the necessity of optical filters to block irradiation of undesired regions in the lamp spectrum is greatly reduced. This is advantageous, since such filters also reduce the intensity in the desired wavelength region, leading to very poor energy efficiency and long irradia- 6.1. INTRODUCTION 91 tion times. The narrow spectrum of UV-LEDs is highly beneficial for controlled, surface-initiated polymerization reactions, since the low-wavelength region of common mercury lamps generally induces polymerization of the monomer in the bulk solution, if this part of the spectrum is not blocked with optical filters. As a consequence, free polymer chains can become entangled within the growing polymer brush, which makes their controlled growth difficult and tedious post-cleaning processes unavoidable. The intensity of UV-LEDs can be conveniently adjusted by means of the applied voltage and their compact design allows for very flexible experiments, e.g. under vacuum conditions. UV-LEDs therefore display significant advantages over traditional UV light sources for UV-induced graft polymerization, in addition to their environmental compatibility, durability and cost-effectiveness.

The polymer brushes prepared by means of UV-LED-induced photopolymerization were intended for the reduction of the interfacial friction in an aqueous environment. While covalently attached polymer chains are readily removed during macroscopic sliding friction under high contact pressures [172], the tribological experiments in this chapter aimed at aqueous lubrication under low contact pressures, for medical applications, for example. It was recently shown that strongly attached self-assembled monolayers (SAMs) and polymer brushes, prepared by a “grafting to” approach, represent promising aqueous lubrication additives under mild contact-pressure conditions [172]. While the previously employed surface modifications consisted of neutral, hydrophilic molecules with limited molecular weights and/or grafting densities, the preparation of weak polyelectrolyte brushes by means of the PMP method was believed to further enhance the lubricating ability of polymer brushes in an aqueous environment. In comparison to neutral hydrophilic brushes, polyelectrolyte brushes exhibit a very high osmotic pressure in aqueous environments of low ionic strengths, which renders them highly suitable for water- based lubrication purposes. Under this viewpoint, methacrylic acid (MAA) was chosen as a monomer, since it shows a high affinity towards water and because direct photopolymerization of poly(methacrylic acid) (PMAA) brushes is feasible under carefully controlled reaction conditions. The preparation of poly(methacrylic acid) (PMAA) brushes on Si/SiO2 surfaces was performed with a silane-derivatized dithiocarbamate iniferter developed by de Boer et al [173]. While the preparation of dense polyelectrolyte brushes by means of “grafting to” is difficult, as previously mentioned, a number of polyelectrolyte brushes have previously been synthesized by “grafting from” approaches [154, 155, 174–179]. Since the carboxyl groups of acrylic acid-based monomers are prone to interact with metal catalysts, the synthesis of polyelectrolyte brushes by means of ATRP usually involves a hydrolysis step 92 6.2. EXPERIMENTAL SECTION after the preparation of a neutral brush [154, 177–179]. Therefore, the direct polymerization of polyelectrolyte brushes has typically been limited to either thermally induced approaches that require thorough cleaning steps after the brush synthesis, especially if a sacrificial initiator or free radicals are present in the monomer solution [174, 176], or to the UV-induced photoiniferter approach. In this work, a silane-derivatized dithiocarbamate iniferter was utilized to prepare poly(methacrylic acid) (PMAA) brushes on Si/SiO2 surfaces under UV irradiation. The combination of the photoiniferter-mediated photopolymerization (PMP) with a UV-LED source appears to be ideally suited to the direct preparation of polyelectrolyte brushes with minimal free polymer formation during brush synthesis. Following characterization of the PMAA brushes by means of surface-analytical techniques such as quartz crystal microbalance with dissipation monitoring (QCM- D), spectroscopic ellipsometry and static contact-angle measurements, the PMAA brushes were demonstrated to enhance aqueous lubrication of Si/SiO2 under low- contact-pressure conditions.

6.2 Experimental Section

6.2.1 Materials p-(Chloromethyl)phenyltrimethoxysilane (ABCR, Germany), tetrahydrofuran (THF, 99.5 % extra dry, Acros, Germany), methanol (Fluka, Switzerland), 2-propanol (Fluka, Switzerland), sulfuric acid (95 - 97 %, Sigma-Aldrich, Germany) and hydrogen peroxide (30 wt.-% in water, VWR, Germany) were used as received. Sodium N,N-diethyldithiocarbamate (97 %, Fluka, Switzerland) was recrystallized from methanol. Water was deionized with a GenPure filtration system (18.2 MΩcm, TKA, Switzerland) and methacrylic acid (98 %, Fluka, Switzerland) was first distilled under vacuum and subsequently passed through an inhibitor remover column (Sigma- Aldrich, Germany). The aqueous buffer solution employed for tribological experiments was prepared by adding 1 mM 4-(2-Hydroxyethyl)piperazine-1-ethanesulfonic acid (HEPES, BioChemika Ultra, Fluka, Switzerland) in pure water and the solution pH was adjusted to a value of 7.4 by the addition of sodium hydroxide (NaOH, Fluka, Switzerland). The buffer is abbreviated as HEPES 0 throughout this chapter.

6.2.2 Synthesis of silanized photoiniferter, SBDC

The photoiniferter employed in this work (Figure 6.1) was synthesized according to a previous protocol [173]. Brieﬂy, p-(chloromethyl)phenyltrimethoxysilane 6.2. EXPERIMENTAL SECTION 93

(CMPTMS) and sodium N,N-diethyldithiocarbamate (STC) were dissolved separately in 10 ml of dry THF before the STC solution was added slowly to the dissolved CMPTMS solution. After 3 h of stirring at room temperature, the solution was passed through a glass ﬁlter and THF was evaporated under reduced pressure. Prior to use, the photoiniferter (SBDC) was connected to a vacuum line for further drying (p = 10−2 mbar, 4 h) and stored at -20°C.

OCH3

H2 H3CO Si C S C N

OCH3 S

Figure 6.1: Chemical structure of the synthesized N,N-(Diethylamino)dithiocarbamoyl- benzyl(trimethoxy)silane (SBDC) photoiniferter.

6.2.3 Ultraviolet-visible (UV-Vis) spectroscopy

The UV absorption spectra of both the SBDC photoiniferter and methacrylic acid were measured with the UV-Vis spectrophotometer described in chapter 2. The obtained spectra were taken as a reference for the selection of the spectral UV range in order to ensure eﬀective initiation while avoiding polymerization of the monomer in solution.

6.2.4 Vapor deposition of silanized photoiniferter onto Si/SiO2

Silicon wafers were cut into 20 mm x 20 mm pieces and ultrasonicated in 2-propanol twice for 15 min each time. Surface hydroxyl groups were generated on the silicon substrates by immersing the samples in a solution of concentrated sulfuric acid and

30 wt.-% hydrogen peroxide (H2SO4 :H2O2 = 7 : 3), also known as piranha solution, for 60 min. After rinsing the substrates with copious amounts of deionized water, they were dried with N2 gas and immediately employed for surface modiﬁcation.

In order to deposit the SBDC photoiniferter on the hydroxylated Si/SiO2 substrates from the vapor phase, a 10 µl drop of the photoiniferter was placed in a desiccator, which was evacuated for 60 s with a rotary vane pump to evaporate residual solvent. Thereafter, the freshly cleaned Si/SiO2 substrates were placed around the SBDC drop and the desiccator was evacuated again, this time for 60 min (p ≈ 10−2 mbar). After closing the valve to the vacuum pump, the photoiniferter was allowed to adsorb onto the silicon oxide substrates for > 48 h until atmospheric pressure was reached. Prior to UV-induced polymerization reactions, the 94 6.2. EXPERIMENTAL SECTION photoiniferter-modiﬁed susbstrates were ultrasonicated in toluene for 2 min to remove physisorbed initiator before the initiator layer was characterized by variable angle spectroscopic ellipsometry (VASE) and static contact-angle measurements.

The vapor deposition method was chosen because the formation of a homogeneous SBDC layer, which is essential for the controlled and uniform growth of polymer brushes over large areas, is more readily achievable than by adsorption from solution.

6.2.5 Controlled radical photopolymerization by means of a UV-LED

The grafting solution consisted of 10 vol.-% methacrylic acid (MAA) in distilled water. Prior to polymerization, the solution was degassed in a sealed glass flask by three alternating ultrasonication (5 min) and vacuum (5 min) cycles. The photoiniferter-modified substrates were sealed in a round-bottom borosilicate flask

(DURAN, Schott, Germany) and continuously purged with N2 gas for 5 min. Simul- taneously, the previously degassed monomer solution was purged with N2 gas and then transferred to the sample-containing glass ﬂask via a syringe under a nitrogen atmosphere. The high-power UV-LED (NCSU033A, NICHIA Corporation, Japan) with a narrow emission spectrum at 365 ± 5 nm was mounted onto a printed circuit board (PCB) in series with a 5 Ω resistor and operated with a laboratory power supply. During operation, two axial fans were placed in front of the UV-LED to avoid a heat-related lifetime degradation of the diode. The distance from the LED surface to the sample was typically 25 mm and the intensity at 365 nm was measured with a radiometer (UVX radiometer with UVX-36 sensor, UVP, Upland, CA). The LED-to-sample distance of 25 mm was determined from the size and the angle- dependent light intensity of the UV-LED to ensure a uniform exposure over the whole sample area (20 mm x 20 mm). As previously mentioned, the optical power output of the LED is determined by the forward current. In this work, however, a voltage of 8.5 V was applied to result in a ﬁxed forward current of 500 mA, which is the recommended long-term operation maximum from the manufacturer. After the polymerization reaction, the samples were taken out of the monomer solution, continuously rinsed with water for 60 s, and blown dry with N2 gas before they were characterized by means of surface-analytical techniques. 6.3. RESULTS AND DISCUSSION 95

6.2.6 Characterization of PMAA brushes

The brush thickness in a dry state was determined by means of variable-angle spectroscopic ellipsometry at three different angles of incidence (65°, 70°, 75°). In order to ensure the formation of homogeneous polymer brushes over the whole sample area of 400 mm2, five different spots were measured on each sample, from which the thickness values were determined via the analysis of a three-layer model. The spectral range considered was from 370 nm to 995 nm and the dry film thickness of the PMAA brush layer was assumed to have a refractive index of 1.45. Static water contact angles were determined by the sessile-drop method described in chapter 2 at all stages of the surface modification process, i.e. after cleaning of the SiO2 substrates, after vapor deposition of the photoiniferter and after photopolymerization of the PMAA brushes. The in situ photopolymerization kinetics of PMAA brushes on

SiO2 surfaces were monitored by means of quartz-crystal microbalance experiments with dissipation monitoring (QCM-D). The flexibility of the UV-LED setup allowed for the positioning of the LED at a 20 mm distance from the QCM cell with a transparent quartz glass window. The photoiniferter-modified quartz crystal resonator was immobilized inside the cell and the sample-to-LED distance was maintained at 25 mm. The degassed monomer solution was injected into the QCM cell with a syringe. Macroscopic tribological experiments were performed with the previously employed pin-on-disk tribometer. The sliding partner of the PMAA brush-bearing substrates was a soft, elastomeric poly(dimethyl siloxane) (PDMS) pin, which displayed a hydrophilic SiOx surface layer after a 60 s air-plasma treatment [57, 180]. The basic frictional properties of the PMAA brushes were tested at sliding speeds ranging from 10 mm/s to 0.25 mm/s. For each tribopair, 20 rotations were carried out at six different sliding speeds under a fixed sliding track. The normal load was kept constant at 1 N. In order to analyze the stability of PMAA brush-bearing substrates under tribological stress, long-term pin-on-disk experiments (1000 rotations) were performed under 1 N normal load and at a sliding speed of 1 mm/s.

6.3 Results and Discussion

6.3.1 Ultraviolet-visible (UV-Vis) absorption of photoiniferter and monomer

Figure 6.2 shows the UV-Vis absorption spectra recorded from the SBDC photoiniferter as well as from as-received and cleaned methacrylic acid. While 10 µl of SBDC was dissolved in 1 ml acetone and measured against an acetone reference, 96 6.3. RESULTS AND DISCUSSION

MAA was measured against air. As can be seen from Figure 6.2, significant UV absorption of the inhibitor-free, clean MAA starts around 300 nm, which is significantly lower compared to the UV absorption of as-received, stabilized MAA around 320 nm. In comparison, the SBDC photoiniferter shows an absorption maximum at approximately 340 nm. Importantly, the monomer and the SBDC photoiniferter have distinct regions of UV absorption that do not overlap significantly, which is considered essential for a controlled surface-initiated polymerization process. In order to avoid self-polymerization of the monomer in solution, the spectral region of the UV source has to be matched with the absorption band of the photoiniferter and the monomer, respectively. That is, the UV source should have minimal emission in the region where the monomer shows significant UV absorption. Hence, careful selection of photoiniferter, monomer and UV source are believed to significantly enhance the degree of control over the polymerization reaction. In this respect, the narrow emission spectrum of the 365 nm high-power UV-LED employed in this work is highly advantageous. The indicated emission spectrum of the UV-LED in Figure 6.2 shows that UV-induced polymerization of MAA in solution is unlikely to occur, since the narrow emission spectrum of the UV-LED does not overlap with the UV absorption region of the monomer.

Methacrylic acid (as received) Methacrylic acid (cleaned) SBDC Photoiniferter

UV-LED (365 nm) Absorbance (a.u.)

260 280 300 320 340 360 380 400 420 440 Wavelength (nm)

Figure 6.2: UV-Vis spectra recorded from the SBDC photoiniferter and from the as- received as well as from the cleaned methacrylic acid (MAA) monomer. The emission spectrum of the 365 nm UV-LED is indicated. 6.3. RESULTS AND DISCUSSION 97

6.3.2 Characterization of the photoiniferter-modiﬁed substrates

After vapor deposition and removal of the physisorbed SBDC photoiniferter from the Si/SiO2 substrates, the average static water contact angle was measured to be 68 ± 3°. This value is significantly higher compared to the contact angle of the cleaned SiO2 surfaces (5 ± 3°) and is consistent with the successful formation of a molecular photoiniferter layer on the silicon substrate. The obtained water contact angles are in good agreement with previously reported values that were obtained after SBDC adsorption from solution [181]. In order to verify the homogeneity of the vapor-deposited photoiniferter layers, the ellipsometric thickness was determined at three distinct spots on each sample, and the average value obtained from at least 10 samples was determined to be 0.75 ± 0.06 nm. This value is significantly lower than the theoretical maximum thickness (1.3 nm) calculated by Rahane et al. [181], and suggests that the surface coverage is below that of a full monolayer. However, it has been previously shown that too high a concentration of surface radicals can lead to extensive termination reactions in surface-initiated polymerization (SIP) approaches, thus favoring initiator densities below full-monolayer coverage [182–184]. Furthermore, the employed vapor deposition method can be applied to materials that are not compatible with organic solvents such as polymeric substrates, and the probability of multilayer formation is significantly reduced when adsorbing trifunctional silanes from the vapor phase [29, 185].

6.3.3 Photopolymerization of methacrylic acid to form poly- (methacrylic acid) (PMAA) brushes

The photoinitiated grafting of poly(methacrylic acid) (PMAA) brushes from SBDC- modiﬁed Si/SiO2 substrates was performed with a 365 nm UV intensity of ca. 12 mW/cm2, measured at the sample surface. Figure 6.3 shows the dry ellipsometric thickness of the PMAA brushes obtained from 10 vol.-% monomer solutions as a function of irradiation time. As is visible from the dry thickness of the PMAA brushes, the UV-LED-initiated polymerization method allows for an eﬀective control of the brush thickness with irradiation time. The PMAA brushes reach high dry-thickness values after short irradiation times and with low monomer concentrations. The initial slow growth of the PMAA brushes is followed by a regime, in which the layer thickness increases very rapidly with irradiation time. For exposure times beyond 45 min, however, the brush growth was found to slow down. Similar growth characteristics have been 98 6.3. RESULTS AND DISCUSSION

300

250

200

150

100 Dry Brush Thickness (nm)

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 Irradiation Time (min)

Figure 6.3: Characteristics of the dry PMAA brush thickness as a function of irradiation time. The monomer solution consisted of 10 vol.-% MAA in water and the 365 nm UV- LED intensity at the sample surface was 12 mW/cm2. The lines serve as guide to the eye.

found in earlier investigations using the same [181] or different [175, 182–184] SIP methods. It is generally agreed that, in comparison to controlled/living polymerization reactions in solution, controlled SIP methods do not exhibit a living character [181, 182, 184, 186]. The relatively low concentration of deactivating species, i.e. dithiocarbamyl radicals in this work, facilitates irreversible termination reactions, leading to a loss of reactive, surface-bound radicals. Hence, a saturation of the brush thickness at longer irradiation times is expected eventually. Since previous studies with the identical SBDC photoiniferter showed that the saturation of poly(methyl methacrylate) (PMMA) brush growth occurs significantly faster for higher monomer concentrations [181], aqueous solutions with a fixed monomer concentration (10 vol.-% MAA) were employed in this work, in order to ensure effective control over the PMAA brush thickness with irradiation time.

Many of the attractive properties of polymer brushes are closely related to the molecular weight of the individual polymer chains as well as to their proximity to each other on the surface, i.e. the surface grafting density. According to the simple model introduced in chapter 4 [20], the surface grafting density can be estimated from ellipsometry data.

In Figure 6.4, the previously described model is adapted to closed-packed PMAA chains. Each unit cell with side lengths L and height d bears one single polymer chain, and the volume V occupied by one chain can be calculated from 6.3. RESULTS AND DISCUSSION 99

S S

N O

N C H C N

H S

O S O

S OH O O

O OH O

O OH

O O O

OH OH OH

O OH

OH O O

O OH

OH O O

O O

OH O

OH OH

O O

d OH OH O

OH OH

OH O

OH O O OH

H O

OH O

O H

OH OH

O z

O H

O O O OH

O O OH O OH H O O

Si O Si O Si x SiO 2

Figure 6.4: Two-dimensional model consisting of close-packed unit cells for the calculation of the surface grafting density and/or the molecular weight of surface-tethered polymer chains.

M V = L2 · d = w (6.1) ρdry · NA where L2 is the surface area occupied by a single chain, d is the dry thickness measured by ellipsometry, Mw is the molecular weight of a single chain, ρdry is the density of the dry monolayer and NA is the Avogadro’s number. In order to obtain the surface grafting density of the SBDC photoiniferter layer, for instance, Equation 6.1 can be transformed to

1 dPI · ρdryP I · NA 2 = . (6.2) LPI MwP I

Besides the ellipsometric thickness of the SBDC monolayer (dPI = 0.75 ± 0.06 nm) as well as the molecular weight (MwP I = 317.5 g/mol) of the individual molecules, a constant value has to be assumed for the dry density (ρdryP I ) of the photoiniferter layer. Hence, the validity of this model is restricted to closely packed chains on the surface and to molecules which do not possess a signiﬁcant density gradient along the chain. Assuming a value of 1 g/cm3 for the density of the SBDC monolayer, its surface grafting density was determined to be 1.42 chains/nm2. Based on this data, it is possible to calculate the approximate molecular weight of the PMAA chains after photopolymerization. Provided that all photoiniferter chains induce polymerization and grow with an identical rate, i.e. LPI = LP MAA, and by further assuming 3 a dry density (ρdryP MAA = 1 g/cm ) for the PMAA chains, Equation 6.1 can be 100 6.3. RESULTS AND DISCUSSION transformed to

2 MwP MAA = LP MAA · dP MAA · ρdryP MAA · NA (6.3)

Table 6.1 shows the calculated molecular weights of the PMAA brushes that were obtained from average dry ellipsometry thickness data of the photoiniferter monolayer and the PMAA brushes, respectively.

Table 6.1: Approximate molecular weight of the PMAA brushes prepared in this work based on calculations from Equations 6.1 - 6.3.

UV Irradiation (min) dP MAA (nm) MwP MAA (g/mol) 5 10 4’233 15 20 8’467 30 70 29’633 45 190 80’433 60 240 101’600 75 290 122’767

It is noted again that the accuracy of the calculated molecular weight values depends on the values selected for the dry density of the SBDC and PMAA layers. In general, the molecular weight values of the PMAA brushes in Table 6.1 are probably underestimated since the number of photoiniferter molecules that induce simultaneous polymer growth, is probably less than unity. Nonetheless, this simple model provides a useful approximate upper limit of the surface grafting density and a lower limit of the molecular weight of the PMAA brushes prepared via the UV-LED-induced “grafting from” approach employed in this work.

6.3.4 Investigation of the in situ PMAA brush growth by means of QCM-D

In order to ensure that the monomer solution does not polymerize in the absence of initiator, a blank SiO2 crystal was ﬁrst exposed to 10 vol.-% MAA in water and UV-irradiated for 30 minutes in a QCM-D setup. Figure 6.5 shows an increase of

6 - 7 Hz in the resonance frequency of the third overtone of the SiO2 crystal upon UV exposure. This behavior has been observed previously [187], and is probably attributed to photo-induced noise. After the UV-LED was switched oﬀ after 30 min, the original resonance frequency values were retrieved, conﬁrming that the 365 nm UV irradiation did not induce polymerization of the MAA monomer. 6.3. RESULTS AND DISCUSSION 101

12 Frequency Shift - F3 10 UV-LED off

2 Frequency Shift (Hz) 0

-2 UV-LED on -4 50 60 70 80 90 100 Time (min)

Figure 6.5: Shift in the resonance frequency of the third overtone of a SiO2 QCM crystal in 10 vol.-% MAA solution upon exposure to 365 nm UV-LED irradiation.

The in situ growth characteristics of PMAA brushes by means of QCM-D experiments are shown in Figure 6.6. After the UV-LED was switched on, the resonance frequency (black circles) increased slightly before a continuous decrease was observed. The beginning of the negative frequency shifts is believed to mark the start of PMAA brush polymerization since the decrease in resonance frequencies can be attributed to an increase in the mass that is coupled to the QCM crystal.

1000 UV-LED on Frequency Shift - F3 0 Dissipation - D3 800

2. -1000 1. monomer rinse Dissipation (10 600 -2000

-3000 UV-LED on 400 UV-LED off -6

-4000 )

Frequency Shift (Hz) water pumping 200 -5000 3. 4. -6000 0 UV-LED off

0 50 100 150 200 250 300 350 400 Time (min)

Figure 6.6: Frequency shift (black circles) and dissipation changes (grey circles) observed in QCM-D experiments during surface-initiated, in situ polymerization of PMAA brushes from 10 vol.-% monomer solutions in water. After a ﬁrst polymerization step, the QCM cell was rinsed with fresh monomer solution (after 100 min) before the UV-LED was switched on again (after 290 min) to induce a second polymerization step. After the UV source was switched oﬀ (after 325 min), the cell was rinsed with fresh monomer twice before pure water was continuously pumped through the system in order to remove unbound polymer. 102 6.3. RESULTS AND DISCUSSION

Up to approximately 25 min of brush growth, i.e. until ca. 40 min in Figure 6.6, the frequency of the third overtone of the QCM crystal decreases almost linearly, suggesting a continuous brush growth. Prolonged irradiation was shown to lead to a higher polymerization rate, which is in good agreement with the evolution of the dry PMAA brush thickness in Figure 6.3. In contrast to the growth characteristics of the dry brushes from ellipsometry data, the QCM-D experiment did not show a slower growth beyond 45 min of UV irradiation. These results suggest that a very slight degree of chain transfer occurs after longer irradiation times, since rinsing with fresh monomer solution after the UV-LED was switched off (after 85 min in Figure 6.6) led to a decrease in the mass coupled to the QCM crystal from -2700 Hz after 100 min to -2380 Hz after 290 min in Figure 6.6. In contrast, the dissipation, which represents the sum of all energy losses per oscillation cycle, was almost constant during that period. It is believed that the rinsing of the QCM cell with fresh monomer solution led to the removal of unbound polymer chains which were entangled inside the brush. These chains are likely to evolve from chain transfer reactions and consequently reduce the mobility of individual brush chains. Thus, the removal of entangled chains brings about a loss of mass, i.e. a higher resonance frequency, but also a net increase in dissipation. The viscoelastic properties of polymer brushes, which are reflected in the dissipated energy from QCM-D experiments, are important in view of tribology, since the solvation of the brush leads to a fluid-like cushion layer, which promotes facile sliding.

With the SI-PMP technique employed in this work, the active chain ends should be preserved after the polymerization step. In order to verify that, the UV-LED was switched on again (after 290 min) after stable resonance frequencies were obtained. As visible from the frequency and dissipation shifts after 300 min in Figure 6.6, the photopolymerization of PMAA brushes continues approximately 10 minutes after the UV-LED was switched on for the second time. Compared to the initial step, the second photopolymerization reaction occurs at a significantly faster rate, indicated by the steep decrease of the resonance frequency from -2300 Hz to -6000 Hz within 25 min. After the UV-LED was switched off, the frequency continued to show a slight shift towards lower values, suggesting that the polymerization did not stop completely after the UV source was turned off. This is in contrast to the observations from the first polymerization step and indicates that the concentration of chain terminating species is somewhat lower. It is thus very likely that fewer polymer chains grow much faster compared to the initial polymerization step. This is also reflected in the dissipation curve of the third overtone frequency (grey circles), which only increases for about 290 · 10−6 whereas the first polymerization led to a 6.3. RESULTS AND DISCUSSION 103

560 · 10−6 increase in dissipation. The generation of bulk polymer is very limited during the second polymerization step, as indicated by the small increase in the resonance frequency and decrease in the dissipation, respectively, after rinsing with fresh monomer (rinsing steps 3 and 4 in Figure 6.6). After approximately 375 min, deionized water was continuously pumped through the QCM cell in order to remove potential unanchored PMAA chains as well as to see differences in the hydrational properties of the brush. The fast saturation of both the resonance frequency and the dissipation values suggest that the presence of unbound polymer chains is very limited during the UV-LED-induced photopolymerization method employed in this work. Hence, simple rinsing steps after the polymerization reaction appear to be sufficient to obtain well-defined PMAA brushes. While the frequency shifts towards higher values under a continuous flow of water, the dissipation was found to increase slightly. The higher frequency, indicative of a lower mass coupled to the

QCM crystal, is associated with the exchange of the heavier methacrylic acid (Mw

= 86.10 g/mol) with water (Mw = 18.02 g/mol), rather than with the removal of polymer chains. This behavior was conﬁrmed from the higher dissipation values in the presence of water, indicating that the viscoelastic properties of the PMAA brushes increased in water compared to 10 vol.-% aqueous methacrylic acid solution. This is also attributed to the partial deprotonation of carboxylic acid groups at higher solution pH as well as to the enhanced hydrogen bonding of water with -COOH moieties.

Figure 6.7 shows that thermally induced polymerization of the monomer could be excluded during photopolymerization, since the UV-LED did not induce any temperature increase of the QCM-D setup during the first polymerization step. From the stable temperature around 25 °C set by the temperature control of the instrument (Figure 6.7), as well as from the fact that UV exposure of the monomer itself did not induce polymerization (Figure 6.5), it is suspected that chain transfer to monomer is occurring at later stages of the polymerization. Free polymer formed in the course of polymerization cannot be distinguished in a QCM-D experiment if the chains are entangled inside the brush or if the polymer significantly increases the viscosity of the monomer solution. The rinsing step after switching off the UV-LED, however, showed that free polymer chains are present in solution. Unfortunately, most other studies presenting QCM data do not show such a rinsing step, although this would help to effectively distinguish between surface-tethered and unbound polymer. 104 6.3. RESULTS AND DISCUSSION

25.05

25.03

UV-LED on UV-LED off

25.00 Temperature ( °C) 24.98

24.95 0 10 20 30 40 50 60 70 80 90 100 110 Time (min)

Figure 6.7: Actual temperature of the QCM-D setup during photopolymerization. The UV-LED does not induce any temperature increase in the vicinity of the heat-sensitive instrument.

6.3.5 Hydrophilicity of PMAA brushes

Since the PMAA brushes prepared in this work were intended for aqueous tribology, the water-compatibility of the surfaces is of particular interest. Therefore, the hydrophilicity of the PMAA brushes was measured by means of static contact-angle experiments. In comparison to the photoiniferter-modified substrates with a static water contact angle of 68 ± 3°, the PMAA brush-bearing samples exhibited average contact-angle values of 49 ± 7°, irrespective of the brush lengths. The water contact angles of hydrophilic polymer brushes have previously been found to be finite and nearly independent of the molecular weight, which has been attributed to the fact that the polymer chains can bridge to the solvent vapor interface [188]. Since the active chain ends were preserved during the PMAA polymerization reaction via photoiniferter technique (Figure 6.6), the diethyldithiocarbamate-terminating groups are likely to contribute to the hydrophobic character of the PMAA-modified substrates in air. Nevertheless, it is expected that PMAA brushes can provide a lubricious interface in an aqueous environment because they exhibit a large number of ionizable carboxyls along their backbone.

6.3.6 Macroscopic aqueous lubrication properties of PMAA brushes

If the relative sliding speed between two contacting bodies is low or if the applied normal load is high, the lubricant cannot readily form a separating film but it is 6.3. RESULTS AND DISCUSSION 105 rather squeezed out of the contact area. This behavior is even more pronounced for low-viscosity fluids such as water and normally results in high interfacial friction and wear. In this context, the presence of a protecting polymer brush has been shown to greatly reduce the friction in the boundary-lubrication regime [1, 116, 118]. While direct contact and adhesion between asperities can be avoided for polymer brush-bearing surfaces, the incorporation of lubricant molecules inside the brush additionally creates a fluid-like layer of low shear-strength and thus facilitates sliding. It was previously demonstrated that hydrophilic polymer brushes can effectively reduce the interfacial friction in an aqueous environment under low sliding speed conditions [104, 138]. In those publications, the authors have employed “grafting to” approaches to generate polymer brushes on a variety of substrates. By applying the “grafting from” method described in this chapter, the formation of high-surface- density polyelectrolyte brushes became feasible. The PMAA brushes synthesized in this work are believed to be highly beneficial for aqueous lubrication purposes under low contact pressures. Firstly, because the high surface density of polymer molecules forces the brushes to adopt an extended conformation, and secondly because the dissociation of the carboxyl moieties at higher solution pH increases the swelling of PMAA brushes and thus, a fluid-like cushion layer is formed.

Under high contact pressures, even strongly attached polymer brushes are easily removed during macroscopic sliding experiments. The reduction of the contact pressure by means of a soft rather than a rigid slider is usually sufficient for enabling the polymer brushes to sustain the tribological stress. The macroscopic pin-on-disk data obtained from different PMAA brushes against oxidized PDMS pins in HEPES 0 (normal load = 1 N) are presented in Figure 6.8. For comparison purposes, results from a bare Si/SiO2 sample are included. From the µ-versus-sliding speed plots it is clear that clean, hydrophilic silica surfaces exhibit moderately good aqueous lubrication properties under low contact pressure conditions, especially at relatively high sliding speeds (µ ≈ 0.017 at 10 mm/s). Towards lower sliding speeds, i.e. in the boundary regime, however, these surfaces are not capable of maintaining a sufficiently thick lubricant film, which is manifested in the increase of the coefficient of friction by a factor of two (µ ≈ 0.035 at 0.25 mm/s).

In this respect, the PMAA brush-bearing surfaces were found to serve as very eﬀective boundary lubricant additives. All tested brushes with a dry ellipsometric thickness ranging form 15 - 190 nm showed friction coeﬃcients below the sensitivity limit (µ = 0.005) of the employed macrotribometer at 1 N normal load. This resolution limit is indicated with a dotted line in Figure 6.8 and the plot shows that these low friction values were maintained over the whole sliding speed range. 106 6.3. RESULTS AND DISCUSSION

0.07

SiO2 0.06 PMAA brush (190 nm) PMAA brush (50 nm) PMAA brush (15 nm) 0.05

0.04

0.03

0.02 Coefficient of Friction

0.01

0.00 0.1 1 10 Sliding Speed (mm/s)

Figure 6.8: Pin-on-disk speed-dependence measurements obtained from PMAA brushes and from bare Si/SiO2 substrates against an oxidized PDMS pin in HEPES 0 (normal load: 1 N).

Therefore it was not possible to effectively distinguish the different PMAA brushes from their lubricating properties in HEPES 0. The results suggest that all PMAA brushes were able to create a highly hydrated, fluid-like interface, rendering these surface modifications highly compatible for aqueous lubrication purposes at neutral pH. In order to test the stability of the PMAA-brushes in long-term pin-on-disk experiments, two samples with distinctive brush heights (30 nm and 240 nm dry thickness) were exposed to tribological stress under 1 N normal load and a sliding speed of 1 mm/s for 1000 rotations. The brush-bearing samples were immersed in HEPES 0 for 10 min prior to the pin-on-disk experiment. Figure 6.9 shows that the µ values from the shorter brush reach a maximum of 0.01 during initial rotations before the friction decreases to µ ≈ 0.006 after 200 rotations. The frictional response during that period is explained by the relatively slow complete hydration of the brush. Since the static water contact angles of the dry brushes are rather high, it seems feasible that the hydration of PMAA chains takes some time. Apparently, the immersion of the brushes into the aqueous lubricant prior to pin-on-disk experiments is not sufficient for a complete hydration, but the samples become more hydrated in the presence of tribological stress. Between 200 and 450 rotations, the coefficient of friction from the 30 nm PMAA brush is nearly constant, before a slight but continuous increase in friction could be observed. After 790 rotations, the µ values increase drastically up to µ ≈ 1.0. Hence, the short PMAA brushes did not sustain the tribological stress for 1000 rotations. When the identical long-term pin-on-disk experiment was performed with the 6.3. RESULTS AND DISCUSSION 107

1 1 mm/s, 1 N, HEPES 0 30 nm PMAA brush 240 nm PMAA brush

0.1

0.01 Coefficient of Friction

0.001 0 100 200 300 400 500 600 700 800 900 1000 Number of Rotations

Figure 6.9: Long-term pin-on-disk experiments involving PMAA brushes with two distinct brush heights (30 nm and 240 nm). The shorter brush (gray circles) did not sustain the applied tribological stress (1 N normal load, 1 mm/s sliding speed) for 1000 rotations, while the 240 nm PMAA samples revealed very low friction coefficients over the whole experiment. longer PMAA brush (240 nm dry thickness), the coefficient of friction decreases from initially µ = 0.01 to values below the detection limit of the pin-on-disk tribometer (µ = 0.005) within the first 400 rotations. This behavior was again attributed to a continuous hydration process of the thick PMAA brush, the highest lubricity being expected after maximum swelling. In comparison to the short brush, the friction coefficients from the 240 nm PMAA sample were found to decrease over a longer period of time, which is believed to be due to a prolonged hydration process of brushes with a higher thickness. The next 200 rotations are characterized by undetectably low friction coefficients, after which the µ values seem to increase slightly and stabilize around µ = 0.006 for the remaining 400 rotations. In comparison to the speed-dependence experiments in Figure 6.9, where the undetectably low friction was observed from the initial sliding speed of 10 mm/s, the swelling of both short and long brush, i.e. the fluid entrainment, appears to be significantly slower at a sliding speed of 1 mm/s. Nonetheless, the obtained friction coefficients from both brushes were well below µ = 0.01 over several hundred rotations, with the difference that the 30 nm PMAA brush did not show the superior long-term stability of the 240 nm PMAA samples. In Figure 6.10, a comparison between the frictional properties of a 15 nm PMAA brush and a 3 nm PEG (5000) monolayer is made. Based on the ellipsometric dry thickness values from the 15 nm PMAA brush, its molecular weight was determined to be 6350 g/mol, which is in the range of the employed PEG chains (5000 g/mol). 108 6.3. RESULTS AND DISCUSSION

1 PEG brush (3 nm) PMAA brush (15 nm)

0.1

0.01

Coefficient of Friction 0.001

0.0001 1 10 Sliding Speed (mm/s)

Figure 6.10: Comparison of µ-versus-sliding speed plots between 15 nm PMAA brushes and 3 nm PEG (5000) monolayers, which have been prepared via “grafting from” and “grafting to” methods, respectively. The molecular weight of individual polymer chains was expected to be in the same order of magnitude for both brushes.

The ﬁve times higher dry thickness of the PMAA brush compared to the PEG monolayer can be explained by the diﬀerent preparation methods, i.e. the PEG monolayer was obtained by the adsorption of silanized PEG (5000) molecules on

Si/SiO2 substrates from toluene solution, i.e. by a “grafting to” method. Since PEG-based polymer brushes have previously proven to be very effective aqueous boundary lubricant additives [104, 117, 138], it was interesting to compare them directly with PMAA brushes. Even though the PEG (5000) monolayer showed very low and extremely stable friction coefficients of around 0.05 over the tested speed range (10 - 0.5 mm/s), the values obtained from the 15 nm PMAA brush were again found to be below the detection limit (µ = 0.005) of the tribometer, indicated by the dotted line in Figure 6.10. This also explains the high deviation in the µ values obtained from PMAA brushes compared to those from PEG monolayers, for which friction was one order of magnitude above the sensitivity limit. The superior lubrication properties of PMAA brushes compared to PEG monolayers at neutral solution pH are attributed to a number of factors. Firstly, the PMAA brushes prepared by the “grafting from” method show a significantly higher dry thickness than the PEG monolayers, indicating that the grafting density of PMAA brushes is considerably higher. Secondly, the density of hydrophilic moieties (-COOH) inside the brush is higher for PMAA brushes and their ionization at neutral solution pH is expected to enhance the swelling of polyelectrolyte brushes. The hydrated thickness of the boundary lubricant is also important

6.4. CONCLUSIONS 109

H H

O H

H H

H O

H O H H

O H H

H H H H

H H O

O H

O H H H

H O

O H

H H H H H O H

H O O O

H H

H H H

H H O H O O

H O

H H H

H O

O H H

O H

O H H H

O O

H H

H O

H H O O O

H O O S H

O H H H

S O O H H

S H H

H S

H H

H S O

O H

H H C

C H O

C C

H O O

O N S N

H H N

S H S O S

H S H

H O OH H OH OH O O N

O N H H

O H

O O

O OH O

O O H O

H H H

H H

H O H

H H

O H

O H H O

H H

H H H O O O

O O

H O O

H H H

H O H H H

OH O H

OH H H

O O OH

H H

H H O O O

H H

O H O

H H

O O

H OH

H O O H

O H

H O H O

OH O O O H

H O O

H H OH

H O

H H

O H

O O

H O

H O H O

OH O O

H H

O O

H H

H O

O H O

H H H

O O

H O H

OH O

O O H

H O H OH O H

H H

H O H

O O

H O

H O O

H OH

H O O

H O

H H O H H

H O

O H H

H O O H H O

H H

H O O

O H

H OH OH

O H OH

H H

H O O

O H O

OH O O

O O

H H H

OH O H H

O O

H H O H H

H O H O

H O

O H O O H

H H H H O

H O H

O O

H H H

O O O

O H

H H H

O OH O H H O

O O

O H H H

OH H H O O

OH H

H OH

H O H H

O O H

O H H

H H

H H O

H H

O O

H O

O O H

H H

O H

O OH

H H

H O

O O H

H O H O O

O H H

H O

H H

O O O H

O O H

H H H O O O O

OH H H H

H O

O H H H H

O H

H H

O OH O O

H O O

H H H H O

O O H O

O H O

H H H H O O

O H H

H H H H

O O

H H O

O O

H H OH

H H O O O

O H O O OH

OH O H H

O O H O O H

O H

H O O

O O H

H O H O O

H H O H

H H

O O

H H H H H O

O H

H H O

H O H

H O

H H

O H O O H O O O

H H H

H OH O H H O O

O OH O

OH H H

O H

Si O Si O Si O Si O Si O Si O HO HO HO HO HO SiO HO 2

Figure 6.11: Presumed tribological interface between a highly hydrated PMAA brush and an ox-PDMS slider in an aqueous environment of neutral pH. with regard to the surface roughness of the tribopair. If the hydrated brush length is signiﬁcantly larger than the roughness of the tribopair, direct asperity contact between the sliding surfaces can be avoided, even if only one surface is bearing a polymer brush. A further eﬀect that reduces the friction between PMAA brushes against oxidized PDMS pins results from electrostatic repulsion between the two tribopairs at neutral pH. The schematic in Figure 6.11 illustrates the presumed tribological interface formed by a PMAA brush against an ox-PDMS in an aqueous environment at neutral pH.

6.4 Conclusions

In this chapter, the controlled growth of poly(methacrylic acid) (PMAA) brushes on

Si/SiO2 surfaces via a photoinduced “grafting from” approach was reported. The employment of a relatively novel UV-LED setup allowed for the preparation of polymer brushes with a high dry thickness within comparatively short reaction times and monomer concentrations. By careful selection of monomer, photoinitiating system and UV source, the polymerization of monomer in solution could be suppressed, which rendered tedious cleaning steps of the formed polymer brushes unnecessary. In this context, it is considered that the utilization of UV-LEDs is very advantageous, since the narrow emission spectrum of the LED could be selected in a region where the monomer does not absorb UV irradiation. In conventional mercury arc 110 6.4. CONCLUSIONS

UV lamps, for instance, the lower regions of the emission spectrum have to be filtered in order to avoid polymerization of the monomer in solution. Besides the fact that optical filters often drastically reduce the intensity in the desired wavelength region, such lamps have to be effectively cooled to avoid heating of the monomer solution and associated thermally induced polymerization. After the preparation of the PMAA brushes, their lubrication ability under low contact pressures was tested in a neutral aqueous solution. It was shown that the macroscopic friction between polyelectrolyte brushes of different molecular weights and soft, hydrophilic ox-PDMS pins was below the detection limit of the employed pin-on-disk tribometer over the entire speed range tested. While the PMAA brushes could not be distinguished with µ-versus-sliding speed plots, the long-term stability of short 15 nm PMAA brushes was shown to be inferior to long brushes (240 nm dry thickness). A further comparison between PMAA brushes and PEG monolayers, of which the latter represent well-known aqueous boundary lubricants that are generally adsorbed on surfaces via “grafting to” methods, showed that PMAA brushes are favored in terms of aqueous lubrication properties. Besides higher grafting densities of the PMAA compared to PEG layers, enhanced swelling of the polyelectrolyte brushes in neutral aqueous media and additional electrostatic repulsion against the oxidized PDMS slider are presumably responsible for the significantly decreased frictional response. In summary, it is expected that strongly attached polyelectrolyte brushes hold great potential as effective boundary lubricant additives under low contact pressures, i.e. in a wear-less regime. To this end, the controlled polymerization of dense polymer brushes via the employed photopolymerization method is believed to serve as a versatile tool for the specific fabrication of low-friction surfaces. CHAPTER 7

Inﬂuence of Bulk Properties of Poly(dimethyl siloxane) (PDMS) Tribopairs on the Aqueous Lubrication Performance

7.1 Introduction

Silicone elastomers based on crosslinked poly(dimethyl siloxane) (PDMS) have been widely used in microfluidic devices for nearly a decade [2, 46, 49, 51, 52]. As mentioned in chapter 2, the ease of fabrication as well as the high elasticity, physiological inertness and its electrical resistance have made PDMS a material of choice in numerous research activities and applications [3, 42, 50, 53, 152, 189, 190]. In the context of tribology, elastomeric materials are capable of drastically lowering the contact pressure and are thus appropriate for investigations of the isoviscous-elastic or soft elastohydrodynamic (soft EHL) lubrication regime [10]. In a practical sense, the low contact pressures encountered from soft material pairings allow the employment of low-viscous lubricants such as water. Since water cannot significantly increase its viscosity under pressure, it is readily squeezed out of the contact zone under high contact pressures and water is thus not able to support externally applied loads, which generally results in high friction and wear. However, previous studies have shown that elastomeric materials are compatible with aqueous lubrication via a soft-EHL mechanism, provided that both contacting surfaces exhibit a hydrophilic character [3]. This requirement is additional to the equations that were established for predicting the minimum lubricant film thickness between tribopairs operating in the soft EHL regime [10], which do not account for surface chemical parameters. In order to satisfy the requirement of surface hydrophilicity, crosslinked

111 112 7.2. EXPERIMENTAL PROCEDURES

PDMS serves as an ideal model elastomer, since the generation of a hydrophilic surface is readily achieved through oxidation of the samples by various techniques [48, 58, 61, 191]. The surface layer of oxidized PDMS (ox-PDMS) has been shown to consist of a hydrophilic but brittle silicon oxide (SiOx) layer, which further provides active attachment sites for more complex, chemical surface modifications [58, 192]. The main aim of the work described in this chapter was to systematically investigate the influence of elasticity on the aqueous lubrication properties of PDMS tribopairs. While lower elastic moduli were conveniently obtained by adjusting the ratio of elastomer base and crosslinking agent of the employed Sylgard 184 elastomer kit, stiffer samples were prepared according to a recently developed protocol at IBM Zurich (Rüschlikon, Switzerland) (see below). Macroscopic pin-on-disk experiments were performed with untreated as well as with oxidized PDMS tribopairs in order to gain information on the role of elasticity in the aqueous lubrication performance of hydrophobic and hydrophilic tribopairs, respectively. It was further investigated how uncrosslinked, volatile low-molecular weight species, which were extracted from PDMS tribopairs, influence the frictional response of untreated and oxidized samples. The characterization of different PDMS samples before and after tribological experiments by means of static water contact angles inside and outside the sliding track and scanning electron microscopy (SEM) completed the analysis of the aqueous lubrication properties of PDMS-PDMS tribopairs of different elasticities.

7.2 Experimental Procedures

7.2.1 Preparation of PDMS samples with diﬀerent elasticities

The standard protocol for the fabrication of PDMS samples from the Sylgard 184 silicone elastomer kit is described in chapter 2. In order to obtain samples with a lower elastic modulus, the 10:1 mixing ratio of elastomer base and crosslinking agent was adjusted to 10:0.5. PDMS samples that were prepared according to the standard 10:1 protocol recommended by the manufacturer are denoted as Sylgard 10:1 and samples with a lower amount of crosslinking agent, i.e. with a mixing ratio of 10:0.5, are denoted as Sylgard 10:0.5 throughout this chapter. For the preparation of elastomeric PDMS samples with a higher stiﬀness, a protocol that was previously developed at IBM Zurich was employed [193]. Here, a so-called modulator mixture is added to the base and the curing agent of the Syl- gard 184 elastomer kit. Figure 7.1 shows the two compounds present in the mod- 7.2. EXPERIMENTAL PROCEDURES 113 ulator mixture, namely poly(methylhydrosiloxane) (HMS-992, ABCR GmbH, Karl- sruhe, Germany) and tetramethyltetravinylcyclotetrasiloxane (TVTMCTS, Fluka AG, Buchs, Switzerland).

Si O O Si Si O Si O Si Si O O Si H n Poly(methylhydrosiloxane) 2,4,6,8-Tetramethyltetravinylcyclotetrasiloxane (HMS-992) (TVTMCTS)

Figure 7.1: The two components of the modulator mixture that was employed to increase the crosslinking density of PDMS samples prepared from a Sylgard 184 silicone elastomer kit.

The role of poly(methylhydrosiloxane) is to incorporate additional silicon hydride (Si-H) functionalities, i.e. crosslinking sites, into the mixture, resuting in a higher stiﬀness of the cured PDMS elastomer. The simultaneous addition of modulator (TVTMCTS), which is already present in the curing agent of Sylgard 184 in small amounts, is necessary to inhibit immediate crosslinking at room temperature induced by the highly reactive Pt catalyst.

E'(0) [MPa]

9.0 8.5 28.00 27.00 8.0 26.00 25.00 7.5 24.00 7.0 23.00 22.00 6.5 21.00 20.00 6.0 19.00 18.00 5.5 17.00

5.0 16.00 15.00 4.5 14.00 13.00 4.0 12.00 11.00 3.5 10.00 SiH / vinyl molar ratio in modmix in ratio molar vinyl / SiH 3.0 9.000 8.000 2.5 2.0 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 modulator content [%]

Figure 7.2: Empirical graph developed by IBM Zurich for the preparation of PDMS elastomers with a speciﬁc shear modulus. The latter can be adjusted via the ratio of silicon hydride (Si-H) to vinyl species in the modulator mixture, i.e. by the number of potential crosslinking units.

According to the empirical graph in Figure 7.2, it was possible to predict the 114 7.2. EXPERIMENTAL PROCEDURES shear modulus E’(0) of the cured PDMS elastomer by adjusting the Si-H to vinyl molar ratio in the modulator mixture as well as the total modulator content. It is noted here that for all recipes the weight ratio of the base (part A) to the curing agent plus modulator mixture (part B + modmix) was kept constant at 10:1. On the basis of Figure 7.2, three recipes with distinct elasticities were selected. Table 7.1 lists the composition of the samples as well as the abbreviations that were used throughout the chapter, with the number referring to the expected shear modulus of the cured PDMS elastomers.

Table 7.1: Composition of the three samples that were prepared according to the IBM recipe. Sample Si-H / vinyl ratio Modulator content IBM 9 2.75 0.63 IBM 18 5.70 1.06 IBM 27 8.25 1.40

As previously mentioned in chapter 2, the Sylgard 184 elastomer kit contains several volatile low-molecular weight species, which are not incorporated into the elastomer network. In order to investigate the inﬂuence of such uncrosslinked species on the frictional properties of PDMS tribopairs, the elastomeric samples were extracted in n-hexane according to the protocol described in chapter 2. Pin-on-disk experiments were usually recorded from unextracted and extracted PDMS tribopairs, except for IBM 27 samples, which were frequently damaged during the extraction procedure. The oxidation of extracted as well as unextracted PDMS samples was achieved by means of a RF air-plasma treatment for 60 s at a gas pressure of 0.1 torr, as mentioned in chapter 2. The hydrophilized PDMS surfaces exhibited static water contact angles below 3° and are denoted as ox-PDMS in the following.

7.2.2 Characterization of PDMS samples

The bulk mechanical properties of all PDMS recipes were determined from tensile tests with an Instron Tensile Tester (Universal Tester, Table-Top Model 5565, Darm- stadt, Germany). For this purpose, test specimen with a “dog bone” shape were cut from 1 mm thick PDMS sheets that were casted into polystyrene petri dishes and the Young’s modulus was calculated from the slope in the area of linear elongation (2 - 4 %) in the stress-strain diagram using the instrumental software. Three tensile tests were performed for all PDMS recipes, while the inﬂuence of the extraction procedure on the Young’s modulus was only examined for Sylgard 10:0.5, Sylgard 10:1 and IBM 9 samples. 7.2. EXPERIMENTAL PROCEDURES 115

Prior to tribological experiments by means of pin-on-disk tribometry, the static water contact angles of PDMS disks were determined by the sessile-drop method introduced in chapter 2. The measurement of the contact angles from ox-PDMS surfaces also served as a verification of the oxidation process from air-plasma treatments. Due to the curved geometry of the PDMS pins, their contact angles could not be determined, but it was assumed that the simultaneous oxidation of pin and disk leads to identical hydrophilicities. The large contact areas that are typically formed by two soft tribological contacts, additionally allowed for the verification of the hydrophilicity inside and outside the sliding track after pin-on-disk experiments. This approach was considered useful for the detection of possible oxide-layer degradation induced by tribological stress. Atomic force microscopy (AFM) was employed for roughness measurements of the PDMS samples. The main focus was placed on the pins, whose roughness was expected to be different from the disks, since the pins were cast in a cell-culture plate with round-bottomed wells. Two ox-PDMS pins from recipes with distinctively different elasticities were chosen for a roughness characterization by means of AFM, namely Sylgard 10:1 and IBM 27. In order to overcome the large adhesion between the silicon nitride AFM tip and the ox-PDMS samples, all measurements were carried out in water using the Dimension 3000 AFM setup described in chapter 2 in contact mode. Scanning electron microscopy (SEM) was employed to trace possible changes in the oxide layer of PDMS samples after tribostress. In these experiments, Sylgard 10:1 and IBM 27 samples were investigated, since they represent specimens from two PDMS recipes with significantly different elasticities. Secondary-electron SEM pictures (LEO 1530, ZEISS, Oberkochen, Germany) were recorded from ox-PDMS samples that had previously been coated with approximately 5 nm of platinum. Pictures were obtained by zooming into specific areas of the samples, e.g. inside and outside the sliding track of the disk.

7.2.3 Pin-on-disk tribometry

Macroscopic tribological data were obtained from pin-on-disk experiments in HEPES 0 solution, a low-salt aqueous buffer with a solution pH of 7.4. Spherical PDMS pins with a radius of 3 mm were loaded by dead weights (1 N and 5 N) and subsequently brought into contact with a rotating PDMS disk. With this configuration, coefficient-of-friction versus sliding-speed plots were obtained by running 20 rotations at nine different sliding speeds (100 - 0.25 mm/s) within the same sliding track. After the reversibility of the sliding-speed direction (high → low or low → 116 7.3. RESULTS high) was confirmed from preliminary experiments with three distinct PDMS pairings, all presented plots were obtained by continuously decreasing the sliding speed from 100 mm/s to 0.25 mm/s. Prior to sliding-speed experiments, a running-in procedure (10 mm/s, 100 rotations) was performed within the same sliding track as the actual experiment. For each tribopair, µ-versus-sliding speed data were first obtained at 1 N normal load and after changing the sliding track, an identical data set was obtained at 5 N. While qualitatively identical results were obtained from both normal loads, the experiments performed under 5 N were shown to amplify the results from 1 N measurements and thus, only the 5 N data is presented in the following. The µ values in the corresponding plots represent average values over the last 10 rotations at a specific sliding speed.

7.3 Results

7.3.1 Elasticity of PDMS samples

Figure 7.3 shows the Young’s moduli of the two Sylgard recipes from tensile tests. The reduction of the amount of crosslinking agent in Sylgard 10:0.5 samples led to a three times lower stiﬀness (E = 0.45 MPa) compared to the standard Sylgard 10:1 elastomer (E = 1.40 MPa). The solvent extraction of the PDMS samples did not signiﬁcantly alter the elasticity of materials from either of the Sylgard recipes, as shown in Figure 7.3.

2.0 Unextracted PDMS Extracted PDMS 1.6

1.2

0.8 Young's Modulus (MPa) 0.4

0.0 Sylgard 10 : 0.5 Sylgard 10 : 1

Figure 7.3: Young’s moduli of as-fabricated (unextracted) and extracted PDMS samples prepared with the Sylgard recipe.

The three PDMS samples that were prepared according to the IBM recipe also 7.3. RESULTS 117 exhibited significantly different Young’s moduli (Figure 7.4). While the IBM 9 (E = 1.10 MPa) samples showed a somewhat higher elasticity than the Sylgard 10:1 elastomer, IBM 18 (E = 2.40 MPa) and IBM 27 (E = 4.15 MPa) exhibited considerably higher Young’s moduli. Since the extraction of the IBM 9 samples did not lead to significant changes in elasticity, the Young’s modulus of the other IBM samples were not measured after extraction. The analysis of the bulk mechanical properties of the differently prepared PDMS samples confirmed that it is possible to vary the elasticity of crosslinked PDMS over one order of magnitude by adjusting the crosslinking density.

Unextracted PDMS

2 Young's Modulus (MPa) 1

0 IBM 9 IBM 18 IBM 27

Figure 7.4: Young’s moduli from as-fabricated PDMS samples prepared according to the IBM recipe. The extraction of the IBM 9 in n-hexane did not lead to signiﬁcant changes in the elastic properties of the samples.

7.3.2 Macroscopic tribological properties of PDMS tribopairs

In the following, the pin-on-disk data obtained from Sylgard and IBM recipes are discussed separately, mainly because the IBM recipes were prepared with two additional compounds, i.e. the modulator mixture described above (Figure 7.1). Since surface properties are known to largely inﬂuence the aqueous lubrication performance of PDMS tribopairs [3], the distinction between samples of diﬀerent composition is considered important.

Sylgard recipes

Figure 7.5 shows the pin-on-disk speed-dependence data from as-prepared as well as extracted Sylgard 10:0.5 and Sylgard 10:1 samples at 5 N normal load in 118 7.3. RESULTS

HEPES 0. At the initial sliding speed (100 mm/s), both untreated, hydrophobic samples show identically high friction coefficients (µ ≈ 1). Towards lower speeds, the friction forces from the standard Sylgard 10:1 samples decreased continuously towards µ ≈ 0.3 - 0.4, while the softer PDMS samples (Sylgard 10:0.5) exhibited rather constant µ values of around 1.0 over the entire sliding-speed range. From the magnitude of the measured friction coefficients, it is not expected that a lubricant film has been established between the two PDMS surfaces, also because hydrophobic interactions between the two PDMS contacts do not support the fluid entrainment into the contact zone. In addition, the decrease in friction towards lower sliding speeds, as observed from Sylgard 10:1 tribopairs, is typically observed for unlubricated, dry contacts since the energy dissipation generally decreases at lower sliding speeds. The higher friction coefficients of Sylgard 10:0.5 compared to Sylgard 10:1 samples was expected for hydrophobic tribopairs, because the larger contact area allows for more hydrophobic interactions and thus for more energy dissipation during sliding. As can be seen in Figure 7.5, the extraction of both Sylgard recipes (empty symbols) showed very little effect on the measured friction coefficients. Ex- cept for the Sylgard 10:1 samples at low sliding speeds, where the extraction led to a slightly higher frictional response compared to the unextracted samples, the solvent treatment in n-hexane did not significantly influence the frictional properties of unoxidized PDMS samples prepared from the Sylgard recipe.

0.1 Coefficient of Friction Sylgard 10 : 1, unextracted Sylgard 10 : 1, extracted Sylgard 10 : 0.5, unextracted Sylgard 10 : 0.5, extracted 0.01 1 10 100 Sliding Speed (mm/s)

Figure 7.5: Coeﬃcient of friction versus sliding speed plot obtained from pin-on-disk experiments employing as-prepared and extracted Sylgard 10:1 and Sylgard 10:0.5 PDMS tribopairs (normal load: 5 N, HEPES 0).

In order to investigate the influence of elasticity and solvent extraction on the frictional properties of hydrophilic Sylgard samples, the latter were oxidized by means of air-plasma treatment prior to pin-on-disk experiments. In comparison to 7.3. RESULTS 119 the friction coefficients that were obtained from the unoxidized, hydrophobic tribopairs (Figure 7.5), the corresponding values from hydrophilic, ox-PDMS samples are significantly lower, as displayed in Figure 7.6.

10 Sylgard 10 : 1, oxidized Sylgard 10 : 1, extracted & oxidized Sylgard 10 : 0.5, oxidized Sylgard 10 : 0.5, extracted & oxidized

0.1 Coefficient of Friction

0.01 1 10 100 Sliding Speed (mm/s)

Figure 7.6: Coeﬃcient of friction versus sliding speed plot obtained from pin-on-disk experiments employing oxidized as well as extracted & oxidized Sylgard 10:1 and Sylgard 10:0.5 PDMS tribopairs (normal load: 5 N, HEPES 0).

The changes in the friction coefficient with sliding speed were also different compared to untreated, hydrophobic samples. In the sliding speed range from 100 - 10 mm/s, the µ values from both Sylgard tribopairs were found to decrease slightly before stable friction coefficients were reached. For sliding speeds below 2.5 mm/s, the coefficients of friction from Sylgard 10:1 samples increased from µ ≈ 0.02 to µ ≈ 0.04, the values from Sylgard 10:0.5 tribopairs from µ ≈ 0.08 to µ ≈ 0.13. Such µ-versus-sliding speed curves are commonly observed from lubricated contacts and the changes in the measured friction with sliding speed are associated with transitions between different lubrication regimes. From the shape of the µ-versus-sliding speed plots in Figure 7.6, it was expected that the tested speed range was capable of displaying the transition from the soft elastohydrodynamic (soft EHL) (high speed) to the boundary lubrication (BL) (low speed) regime for the ox-PDMS tribopairs. However, as with the untreated PDMS tribopairs, the oxidized samples with a lower Young’s modulus (Sylgard 10:0.5, E = 0.45 MPa) exhibited considerably higher friction compared to Sylgard 10:1, E = 1.40 MPa). For lubricated contacts operating in the soft EHL regime, this behavior was not expected, since the lubricant film thickness is supposed to be higher under lower contact pressure conditions. Consequently, the softer Sylgard 10:0.5 samples should display lower frictional properties compared to Sylgard 10:1 tribopairs. A further difference between the two Sylgard recipes was observed from the comparison of the lubricating properties 120 7.3. RESULTS from extracted and unextracted ox-PDMS tribopairs. The extraction of Sylgard 10:1 samples prior to air-plasma treatment did not greatly influence the resulting µ values, except at the lowest speeds, where the extracted and oxidized samples showed slightly lower friction coefficients than the unextracted ox-PDMS tribopairs. In contrast, extracted Sylgard 10:0.5 samples showed significantly higher friction coefficients over the whole speed range compared to the unextracted ox-PDMS tribopairs. These observations called for a more detailed investigation of the tribopairs prior to and after tribological stress, whose results will be discussed below.

IBM recipes

The friction coeﬃcients that were measured from unoxidized IBM 9, IBM 18 and IBM 27 samples are very similar to the values obtained from Sylgard recipes (Figure 7.5). The hydrophobic PDMS tribopairs exhibit µ values of around 1 in the high- speed regime, while continuously decreasing coeﬃcients of friction were measured towards lower sliding speeds, resulting in µ ≈ 0.5 at 0.25 mm/s, as shown in Figure 7.7.

10 IBM 9, unextracted IBM 9, extracted IBM 18, unextracted IBM 18, extracted IBM 27, unextracted 1

0.1 Coefficient of Friction

0.01 1 10 100 Sliding Speed (mm/s)

Figure 7.7: Coeﬃcient of friction versus sliding speed plot obtained from pin-on-disk experiments employing as-prepared (IBM 9, IBM 18, IBM 27) and extracted (IBM 9, IBM 18) PDMS tribopairs (normal load: 5 N, HEPES 0). Data from extracted IBM 27 samples could not be obtained, since the extraction in n-hexane led to rupture of the samples during drying.

As with the Sylgard samples, the decreasing friction from IBM PDMS tribopairs towards lower sliding speeds is characteristic of unlubricated contacts, suggesting that no lubricant film was formed between the hydrophobic PDMS tribopairs in HEPES 0. Furthermore, it was not possible to distinguish the PDMS samples with different elasticities from their frictional response, since the µ values were almost 7.3. RESULTS 121 identical for IBM 9, IBM 18 and IBM 27 tribopairs. The extraction in n-hexane to remove uncrosslinked low-molecular weight species did also not influence the lubricating properties of the PDMS samples as can be seen from Figure 7.7. Since the drying procedure after solvent extraction led to the rupture of IBM 27 pins and disks, these data could not be collected.

0.1 IBM 9 MPa, oxidized IBM 18 MPa, oxidized IBM 27 MPa, oxidized Coefficient of Friction 0.01

1 10 100 Sliding Speed (mm/s)

Figure 7.8: Coeﬃcient of friction versus sliding speed plot obtained from pin-on-disk experiments employing oxidized IBM 9, 18 and 27 and PDMS tribopairs (normal load: 5 N, HEPES 0).

The oxidation of the PDMS samples prepared according to the IBM recipe also resulted in a significant friction reduction, as is visible from Figure 7.8. In contrast to the hydrophobic samples, it was possible to distinguish the ox-PDMS IBM tribopairs according to their frictional response in HEPES 0. Interestingly, the reduction in friction again followed the order of increasing stiffness, as previously observed from Sylgard 10:1 and Sylgard 10:0.5 samples. That is, IBM 27 tribopairs with the highest Young’s modulus (E = 4.15 MPa), exhibited the lowest µ values over the entire sliding speed range, i.e. µ ≈ 0.012 - 0.017, followed by IBM 18 (µ ≈ 0.019 - 0.025, E = 2.40 MPa) and IBM 9 (µ ≈ 0.021 - 0.037, E = 1.10 MPa) samples. Further similarities with the Sylgard samples include the onset of increasing friction coefficients towards lower sliding speeds, which was also observed for speeds below 2.5 mm/s. According to the soft EHL theory, one would again expect that stiffer samples exhibit higher friction, because of the limited lubricant film thickness, but the IBM tribopairs also showed the opposite behavior. Unlike for the hydrophobic PDMS samples, the extraction and subsequent oxidation of IBM 9 and IBM 18 tribopairs manifests itself in higher µ values compared to the unextracted ox-PDMS samples (Figures 7.9 and 7.10). While the difference in friction between extracted and as-prepared ox-PDMS IBM 9 tribopairs is clearly 122 7.3. RESULTS

1 IBM 9 MPa, oxidized IBM 9 MPa, extracted & oxidized

0.1 Coefficient of Friction

0.01 1 10 100 Sliding Speed (mm/s)

Figure 7.9: Coeﬃcient of friction versus sliding speed plot obtained from pin-on-disk experiments employing oxidized as well as extracted & oxidized IBM 9 PDMS tribopairs (normal load: 5 N, HEPES 0).

1 IBM 18 MPa, oxidized IBM 18 MPa, extracted & oxidized

0.1 Coefficient of Friction

0.01 1 10 100 Sliding Speed (mm/s)

Figure 7.10: Coefficient of friction versus sliding speed plot obtained from pin-on-disk experiments employing oxidized as well as extracted & oxidized IBM 18 PDMS tribopairs (normal load: 5 N, HEPES 0). visible over the entire sliding speed range (Figure 7.9), both ox-PDMS IBM 18 samples (Figure 7.10) initially showed identical friction coefficients (µ ≈ 0.02). The µ values of unextracted ox-PDMS IBM 18 samples (filled triangles) increased from µ ≈ 0.019 to µ ≈ 0.025 for sliding speeds below 2.5 mm/s, while for the extracted ox-PDMS tribopairs (empty triangles), the increase in µ started at 25 mm/s and led to values of 0.06 at the lowest sliding speed. Summarizing the observations of the effect of extraction of low-molecular weight species from bulk PDMS samples, it can be concluded that the extraction does not 7.3. RESULTS 123 influence the frictional properties of hydrophobic PDMS tribopairs. Conversely, the extracted and oxidized PDMS samples in general showed degraded aqueous lubricating properties compared to as-prepared, ox-PDMS samples. Because significantly higher standard deviations were observed for extracted and oxidized compared to as-prepared ox-PDMS samples, and from the fact that stiffer ox-PDMS samples exhibited enhanced lubrication properties in HEPES 0, the detailed characterization of the different tribopairs after pin-on-disk experiments should be considered.

7.3.3 Considerations of the lubrication regimes involved

As mentioned in the introduction, the isoviscous-elastic, also known as the soft elastohydrodynamic lubrication (soft EHL) regime, is expected to dominate for tribological contacts consisting of materials with a high elasticity. While the elastic deformation of rubbers such as PDMS is signiﬁcant under pressure, the increase in lubricant viscosity is negligible. As mentioned previously, Hamrock and Dowson derived an empirical expression for the calculation of the minimum lubricant ﬁlm thickness for ideally smooth sphere-on-plane geometries, which was later revised by Esfahanian and Hamrock [10]:

0.77 0.65 0−0.44 −0.21 hmin = 2.8 · R · (η · us) · E · w (7.1)

According to Equation 7.1, the fluid-film thickness and hence the lubricity of a given tribosystem should depend on the stiffness of both contacting materials. Given that the lubricant does not significantly increase its viscosity under a certain pressure, as is the case for water, lower friction coefficients are expected from softer samples due to a higher lubricant film thickness. On the other hand, once a sufficiently thick water lubricant film has been established, i.e. one that exceeds the combined roughness of the tribopair, it is not expected that the friction decreases any further. As demonstrated in the previous chapters, another parameter that critically influences the aqueous lubrication behavior is the hydrophilicity. Although Equation 7.1 does not take surface-chemical properties into account, it has been demonstrated that both tribological contacts need to show a sufficiently wettable character to induce soft EHL [3]. With regard to the experimental work in this chapter, it was expected that the hydrophilic ox-PDMS surfaces qualify for soft EHL in the high-speed regime. The calculation of hmin according to Equation 7.1 for all ox-PDMS samples employed in this work (us = 100 and 0.25 mm/s, R = 0.003 m, η = 0.001 Pas, w = 5 N), was taken as a reference for determining the prevailing lubrication regimes. Figure 7.11 124 7.3. RESULTS

displays the corresponding hmin values for the two Sylgard recipes as a function of elasticity, while the friction coeﬃcients are also included for the two sliding speeds.

1 COF (100 mm/s) Sylgard Minimum Film Thickness (100 mm/s) 1000

COF (0.25 mm/s) Minimum Film Thickness (nm) Minimum Film Thickness (0.25 mm/s)

100

0.1

10 Coefficient of Friction

0.01 1 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 Young's Modulus (MPa)

Figure 7.11: Calculated minimum film thickness (empty symbols) as well as measured friction coefficients (filled symbols) from ox-PDMS Sylgard 10:1 (E = 1.40 MPa) and Sylgard 10:0.5 (E = 0.45 MPa) samples for two distinct sliding speeds in HEPES 0 (100 mm/s and 0.25 mm/s).

In Figure 7.11, the minimum lubricant film thickness calculated for Sylgard 10:1 and Sylgard 10:0.5 ox-PDMS tribopairs at sliding speeds of 100 mm/s was found to be between 135 and 225 nm (empty black circles). After considering the surface roughness values of Sylgard 10:1 PDMS pins (Ra = 7.8 nm) and disks (Ra = 0.12 nm), which were determined by AFM measurements (scan area = 25 µm2) prior to tribological stress, soft EHL was expected since the Λ ratio was determined to be between 48 - 80. However, from the µ values that were obtained at 100 mm/s (filled black circles), the softer Sylgard 10:0.5 tribopairs exhibited significantly worse lubricating properties compared to Sylgard 10:1 samples, which is not expected according to the soft EHL theory. The identical behavior was observed at the lowest sliding speed (0.25 mm/s), where the softer samples again showed higher coefficients of friction (filled grey squares) compared to Sylgard 10:1. The minimum film thickness calculations at 0.25 mm/s (empty gray squares) revealed, however, that mixed or boundary lubrication is expected under these conditions, since the lubricant film was determined to be in the range of 3 - 4 nm, i.e. comparable to the roughness of the PDMS samples (Λ ≈ 1.0 - 1.4). It should be mentioned that the actual Λ ratios might be smaller, i.e. boundary lubrication is more likely to prevail, since the roughness of the ox-PDMS samples was determined prior to tribological stress. In summary, the plotted data in Figure 7.11 suggests that Equation 7.1 does not unambiguously predict the actual film thickness for the 7.3. RESULTS 125 investigated ox-PDMS Sylgard tribopairs. The Sylgard 10:1 and especially Syl- gard 10:0.5 ox-PDMS samples exhibit very similar µ values at both sliding speeds, although the predicted lubricant film thickness should be vastly different for experiments at 100 mm/s and 0.25 mm/s, respectively. This indicates that the oxide layer of both samples is subjected to a partial degradation during sliding friction and thus prevents the soft EHL mechanism from predominating.

0.1 1000 COF (100 mm/s) IBM Minimum Film Thickness (100 mm/s) COF (0.25 mm/s) Minimum Film Thickness (nm) Minimum Film Thickness (0.25 mm/s)

100

10 Coefficient of Friction

0.01 1 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Young's Modulus (MPa)

Figure 7.12: Calculated minimum film thickness (empty symbols) as well as measured friction coefficients (filled symbols) from IBM 9 (E = 1.10 MPa), IBM 18 (E = 2.40 MPa) and IBM 27 (E = 4.15 MPa) ox-PDMS samples for two distinct sliding speeds in HEPES 0 (100 mm/s and 0.25 mm/s).

Figure 7.12 shows the characteristics of the minimum film thickness and the friction coefficients at the highest (100 mm/s) and lowest (0.25 mm/s) sliding speed for the three IBM recipes. As previously mentioned, the µ values at sliding speeds of 100 mm/s (filled black triangles) were found to decrease with increasing stiffness of the ox-PDMS tribopairs. Considering that the AFM roughness measurements revealed somewhat lower values for IBM 27 pins (Ra = 4.4 nm) and disks 2 (Ra = 0.23 nm) (scan area = 25 µm ) compared to Sylgard 10:1 samples, the pin- on-disk experiments at 100 mm/s were also expected to lie in the soft EHL regime (Λ ≈ 39 - 70). As with the Sylgard samples, the characteristics of the µ values from the IBM samples as a function of the Young’s modulus are not indicative of the predicted soft EHL mechanism at a sliding speed of 100 mm/s. At the lowest speed (0.25 mm/s), where mixed or boundary lubrication is expected to prevail, since the lubricant film thickness was determined to be in the low nanometer range (Λ ≈ 0.8 - 1.4) (empty gray diamonds), the µ values (filled gray diamonds) show the same trend compared to experiments at 100 mm/s. Since the relative magnitudes of the friction coefficients are only slightly different at low and high speeds, it implies that 126 7.3. RESULTS soft EHL was not the dominating lubrication mechanism for the ox-PDMS IBM tribopairs at both (100 mm/s and 0.25 mm/s) sliding speeds. Although Equation 7.1 suggested two disctinct lubrication mechanisms to be in force for experiments at sliding speeds of 100 mm/s and 0.25 mm/s, the characteristics of the frictional response with the elasticity of the tribopairs was almost identical for Sylgard as well as for the IBM samples at both speeds. Whereas in the soft-EHL regime, higher fluid film thicknesses and thus lower friction coefficients are expected from softer tribopairs, the macroscopic pin-on-disk experiments in this chapter revealed the opposite behavior. In fact, the friction was found to be smaller for stiffer samples, at low as well as at high sliding speeds, indicating that the soft EHL mechanism was not at work for all tested samples up to 100 mm/s. A possible explanation for the observed behavior could be a degradation of the oxide layer of certain PDMS tribopairs during the course of a pin-on-disk experiment. As mentioned above, the hydrophilicity of both tribological contacts is a prerequisite for the soft EHL mechanism to occur in an aqueous environment. Given that the hydrophilicity of ox-PDMS is degraded during sliding friction under 5 N normal load, it seems feasible that the build-up of a lubricant film is already disturbed in the high-speed regime. Hence, to further elucidate the lubrication mechanisms of the experiments performed in this chapter, it was necessary to investigate the PDMS tribopairs in more detail.

7.3.4 Investigation of selected PDMS tribopairs after tribological stress

From the macroscopic pin-on-disk experiments performed in this work, the lowest friction was obtained from the ox-PDMS samples with the highest Young’s modulus, i.e. IBM 27 samples. At the lowest as well as at the highest sliding speed, the coefficients of friction were found to continuously decrease with increasing stiffness of the oxidized PDMS tribopairs. Based on these observations, it was suggestive to investigate selected ox-PDMS tribopairs in more detail. Firstly, the hydrophilicity of different ox-PDMS disks was characterized subsequently to tribological stress. In order to be able to determine the water contact angles in- and outside the contact area, pin-on-disk control experiments were performed under 10 N normal load. The high load was necessary in order to obtain a sufficiently wide sliding track for contact-angle measurements within the tribostressed area. After sliding ox-PDMS pins against disks made of the identical material for 30 rotations at a sliding speed of 0.25 mm/s in HEPES 0, static water contact angles were recorded inside and outside the sliding track of the disk. Prior to tribological 7.3. RESULTS 127 stress, all ox-PDMS disks exhibited water contact angles of < 3°.

70 0.7 CA inside track CA outside track 60 Coefficient of Friction 0.6 Coefficient of Friction 50 0.5

40 0.4

0.3 30

0.2 20 Static Water Contact Angle ( °) 0.1 10

0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Young's Modulus (MPa)

Figure 7.13: Characterization of static water contact angles inside (filled squares) and outside (empty squares) the sliding track of ox-PDMS disks with different elasticities subsequent to pin-on-disk control experiments (normal load: 10 N, sliding speed: 0.25 mm/s, lubricant: HEPES 0). On the right axis, the average µ values after 30 rotations (filled circles) are displayed as a function of sample elasticity.

In Figure 7.13, the static water contact angles measured inside (filled squares) and outside (empty squares) the sliding track of ox-PDMS samples are displayed as a function of their Young’s modulus. On the right axis, the corresponding µ values (filled circles) averaged over 30 rotations, obtained under 10 N normal load and at a sliding speed of 0.25 mm/s, are shown (lubricant: HEPES 0). In contrast to the static water contact angles prior to tribological stress, all tested ox-PDMS samples showed a significantly poorer hydrophilicity in- as well as outside the sliding track. Interestingly, the increase in the water contact angles could be correlated with the Young’s modulus of the PDMS samples. Furthermore, the coefficient of friction was also found to increase along with the contact angles inside and outside the sliding track. Figure 7.14 shows the cross-section of a Sylgard 10:1 sample (E = 1.40 MPa), along which five water droplets were deposited with the contact angle goniometer. It is evident that the areas that were subjected to tribological stress exhibit higher contact angles compared to areas outside the sliding track. 128 7.3. RESULTS

outsideinside outside inside outside

Figure 7.14: Water droplets deposited accross a Sylgard 10:1 ox-PDMS disk after a pin- on-disk experiment (normal load: 10 N, sliding speed: 0.25 mm/s, lubricant: HEPES 0). The two droplets deposited inside the circular sliding track of the disk exhibit signiﬁcantly higher static water contact angles compared to the water droplets outside the tribostressed area.

These control experiments demonstrate that the oxide layer of air-plasma-treated PDMS samples is subjected to degradation in the course of pin-on-disk experiments, despite the relatively low contact pressures. From the fact that even areas that were not exposed to tribological stress showed higher contact angle values, it is speculated that the diffusion of low-molecular weight species from the bulk to the surface is facilitated by a cracked oxide layer, especially because these samples were not extracted prior to oxidation. Therefore, the high water contact angles in areas not exposed to tribological stress are believed to result from hydrophobic recovery induced by contaminants originating from the elastomer network and/or from the atmosphere. To summarize, these control suggested that the oxide layer from stiffer PDMS samples is subjected to less degradation compared to softer tribopairs, which explains why IBM 27 samples with the highest Young’s modulus showed the lowest friction in an aqueous environment. The Sylgard 10:1 and IBM 27 disks that were exposed to pin-on-disk experiments under 10 N normal load for 30 rotations (sliding speed: 0.25 mm/s) were further analyzed with scanning electron microscopy (SEM). As was suspected from tribological as well as from contact-angle data, the oxide layers of the PDMS samples were damaged to different extents. From the SEM images with a magnification of 25’000 in Figure 7.15, it is clear that the oxide layer inside the sliding track of a Sylgard 10:1 disk was much more damaged compared to that of a IBM 27 disk. The worn oxide layer of the softer sample additionally showed a multilayer structure. Once this oxide layer is partly worn away, such silica particles are believed to induce the higher friction observed from soft ox-PDMS tribopairs. In summary, the frictional response from ox-PDMS tribopairs of different elasticities was found to be related to the degradation of the brittle oxide layer during tribological stress in an aqueous environment. While ox-PDMS samples with a higher stiffness showed a greater resistance to wear and consequently lower friction coefficients, the oxide layer from softer PDMS samples degraded much faster under tribological stress, which was manifested by a reduction 7.4. CONCLUSIONS 129

Figure 7.15: SEM images (magniﬁcation: 25’000x) inside the sliding track of ox-PDMS disks prepared according to the Sylgard 10:1 (left, E = 1.40 MPa) and IBM 27 (right, E = 4.15 MPa) recipe, respectively. Both samples were previously employed in pin-on-disk experiments under 10 N normal load for 30 rotations at a sliding speed of 0.25 mm/s (lubricant: HEPES 0). of their surface hydrophilicity and therefore higher interfacial friction in an aqueous solution.

7.4 Conclusions

The experimental work in this chapter was dedicated to exploring the effect of the bulk mechanical properties of PDMS tribopairs as well as the influence of hydrophilicity on the aqueous lubrication properties. The facile preparation of elastomeric samples with different elasticities made from the same base material allowed for the tribological investigation of samples with nearly identical surface properties. This was considered important, since the detailed surface chemical properties are known to critically influence the frictional response in aqueous lubrication. When as-prepared, hydrophobic PDMS tribopairs were employed in macroscopic pin-on- disk experiments, the influence of the bulk mechanical properties on the measured friction in water was not very pronounced. This was attributed to the fact that no ap- preciable lubricant film could be established at the tribological interface, which was reflected in the high friction coefficients measured from all PDMS recipes. For air- plasma-treated and thus hydrophilic ox-PDMS samples, soft EHL was expected to be the prevailing lubrication mechanism over a certain sliding-speed range. Since the predicted minimum lubricant film thickness does not account for surface-chemical properties, it was previously suggested that the surfaces need to be sufficiently hydrophilic in order to evoke soft EHL. Thus, it was considered important that the samples employed in this chapter should possess identical surface properties. Ac- cording to the soft-EHL theory, changes in the elasticity should lead to a different lubricant film thickness and, hence to different µ values for otherwise identical tri- 130 7.4. CONCLUSIONS bosystems. The indirect verification of the predicted changes in the lubricant film thickness was attempted by comparing the frictional response from ox-PDMS tribopairs with different Young’s moduli. The obtained results could not confirm soft EHL as the predominant lubrication mechanism for all ox-PDMS tribopairs employed in this work under 5 N normal load, either at high (100 mm/s) or at low sliding speeds (0.25 mm/s). The friction coefficients were found to decrease with increasing Young’s moduli of the tribopairs, which is contrary to what is expected from the soft-EHL theory. A more detailed analysis of the tribopairs subsequent to tribological stress revealed that the oxide layer of certain samples, which was generated by means of air-plasma treatment, was subjected to degradation during pin-on-disk experiments. Water contact angle data as well as SEM images confirmed that the degradation of the brittle oxide layer on air-plasma treated PDMS samples occurs faster for softer samples, which was found to directly influence the lubricating properties in an aqueous environment. It could therefore not be confirmed if soft EHL is at work for all employed ox-PDMS recipes, mainly because of the instability of their oxide layer and their hydrophilicity. The comparison of ox- PDMS samples from which low-molecular weight species have been extracted prior to air-plasma treatments with as-fabricated and oxidized samples could not confirm that the migration of hydrophobic, uncrosslinked PDMS components to the surface was responsible for the degraded hydrophilicity and the therewith-associated higher friction. Extracted and oxidized PDMS samples exhibited significantly higher µ values compared to their unextracted ox-PDMS analogs. This implies that the oxide layer of previously extracted ox-PDMS samples was less stable compared to the unextracted tribopairs.

Based on the findings in this chapter, it appears to be beneficial to employ stiffer ox-PDMS samples for aqueous lubrication purposes. These samples have exhibited a more stable oxide layer after plasma treatment, whose integrity is essential for maintaining the hydrophilicity of the ox-PDMS tribopairs and the low-frictional properties in an aqueous environment associated therewith. Additional attention should be paid to the oxidation process, i.e. the air-plasma treatment in this thesis. It is expected that the oxidation can be further optimized in order to achieve a sufficiently hydrophilic but low-defect oxide surface with a reduced brittleness. However, controlled oxidation by means of plasma treatment requires stringent control of the process parameters such as pressure, treatment time, power level as well as the positioning of the samples inside the chamber. Regarding the results from the previous chapters of this thesis, the present data suggest that the oxide layer of the ox-PDMS sliders might have been subjected to partial degradation during macro- 7.4. CONCLUSIONS 131 scopic tribological stress. Nonetheless, the comparison to untreated, hydrophobic PDMS tribopairs demonstrated that even after prolonged sliding, the ox-PDMS samples are still capable of effectively reducing the frictional response in an aqueous environment. 132 7.4. CONCLUSIONS CHAPTER 8

Conclusions and Outlook

8.1 Conclusions

The aim of this thesis was to investigate the ability of surface modifications to reduce macroscopic interfacial friction in aqueous environments under mild contact-pressure conditions. In order to achieve the low contact pressures, crosslinked poly(dimethyl siloxane) (PDMS) substrates were chosen as one or both tribological contacts. With the experimental setup introduced in chapter 3, which consisted of a conventional pin-on-disk tribometer with a soft rather than a rigid slider, it was possible to non-destructively investigate the aqueous-lubrication properties of different thiol SAMs on a macroscopic scale. Previous tribological studies involving thiol SAMs on gold have focused on experiments at the micro- and nanoscale only, because high contact pressures experienced in conventional macroscopic testing generally result in fast removal of molecules during sliding friction. In contrast, surface- sensitive infrared spectroscopy (PM-IRRAS) in chapter 3 confirmed that the thiol SAMs remained intact after the tribological stress experienced from an ox-PDMS slider. The ex situ analysis of the tribostressed area by means of PM-IRRAS became possible within the wide sliding track that was generated by the soft ox-PDMS slider. In macroscopic pin-on-disk experiments, it could be demonstrated that surface- chemical as well as structural properties of the SAMs are the two basic parameters that govern the aqueous lubrication performance. While alkanethiol SAMs with a hydrophilic terminating group could significantly reduce the interfacial friction against an oxidized and thus hydrophilic PDMS slider, the frictional response from

133 134 8.1. CONCLUSIONS hydrophobic SAMs was only marginally lower compared to that from bare gold substrates. It was further demonstrated that shorter SAMs with a higher degree of disorder inside the monolayer are not as eﬀective as aqueous lubricant additives compared to well-ordered SAMs.

In chapter 4, the tribological investigations were focused on thiols with a hydrophilic character. In comparison to alkanethiol SAMs, monolayers from poly(ethylene glycol) (PEG) with a molecular weight of 5000 g/mol were shown to further decrease the frictional response in a low-salt aqueous buffer solution. It was again found that the structural properties of the SAMs play an important role in aqueous lubrication, since monolayers from oligo(ethylene glycols) (OEGs) with 7 ethylene glycol units exhibited significantly higher values than the PEG 5000 films. Hence, hydrophilic polymers with high molecular weights were found to further reduce the friction compared to short chain SAMs, even though the PEG-based monolayers showed a considerably lower surface density and thus a higher degree of disorder. In contrast to the well-ordered OEG SAMs, the more disordered PEG 5000 monolayers were highly hydrated, which has been considered very important for successful aqueous lubrication. When pin-on-disk experiments were performed in an aqueous buffer of high ionic strength, it was shown that the frictional response from PEG-based as well as from alkanethiol monolayers becomes higher. For PEG 5000 films, quartz crystal microbalance experiments with dissipation monitoring (QCM- D) revealed that the inferior lubrication was associated with the “salting-out” of surface-immobilized PEG chains, which is consistent with the bulk aqueous PEG solution properties in high-ionic-strength media.

Besides the influence of ionic strength on the aqueous lubricating properties of thiol-based hydrophilic monolayers, the role of solution pH was addressed in chapter 5. It was again demonstrated that surface-chemical as well as structural properties of the monolayers are of fundamental importance for aqueous lubrication under low- sliding-speed conditions. Compared to the near-neutral aqueous buffer, acidic and alkaline aqueous solutions did not inherently degrade the aqueous lubricating properties, as was found for high salt concentrations. In fact, alkaline aqueous solutions were found to generally improve the aqueous lubrication performance of different SAM / ox-PDMS tribopairs. More detailed investigations revealed that PEG 5000 monolayers did not exhibit different hydration properties at high or low solution pH, respectively, while infrared spectra inside the tribostressed area revealed that the stability of the SAMs in the acidic medium is limited. Further control experiments aimed at explaining the role of the ox-PDMS slider. It could be confirmed that an alkaline solution is capable of improving the aqueous lubrication performance of 8.1. CONCLUSIONS 135 hydrophilic oxidized PDMS tribopairs. However, the beneficial effect of the alkaline solution on the tribosystems employed in this thesis could not entirely be attributed to its impact on the brittle oxide layer of PDMS, since the magnitude of the friction reduction compared to the neutral solution was different for each SAM counter surface. Hence, some more experimental as well as theoretical work is needed to be able to fully understand the underlying lubrication mechanisms in alkaline solution.

Based on the conclusions from the experimental work involving different thiol monolayers on gold, the aim was to further improve the aqueous lubrication performance by developing tailor-made surface modifications on technologically important substrate materials. Compared to chapters 3 - 5, where pre-fabricated thiol molecules were adsorbed onto gold surfaces from solution, i.e. by means of a “grafting to” approach, a novel “grafting from” strategy was introduced in chapter 6. The utilized photoiniferter method allowed for a controlled growth of high-density poly(methacrylic acid) (PMAA) brushes initiated by means of ultraviolet (UV) irradiation. By employing a recently developed high-power ultra-violet light-emitting diode (UV-LED), it was shown that it is possible to directly grow polyelectrolyte brushes within relatively short times from aqueous solutions with low monomer concentrations. This approach was shown to have distinct advantages over methods using conventional UV sources, mainly due to the narrow emission spectrum of the UV-LED, which consequently led to much better defined polymer brushes and significantly reduced the number of cleaning steps required following brush preparation. From a tribological standpoint, the PMAA chains have proven to serve as excellent lubricant additives in a neutral aqueous environment. Brushes of distinct molecular weights (i.e. brush lengths) were not distinguishable in their performance by speed-dependent pin-on-disk experiments, which were carried out for comparatively short periods of time (20 rotations at each speed), i.e. the frictional response of all brushes was shown to be below the detection limit of the instrument. Only long-term experiments revealed that longer PMAA brushes are more stable than shorter brushes. The superior lubricating properties of the weak polyelectrolyte brushes compared to PEG brushes prepared via “grafting to” methods arose from the significantly higher surface grafting densities as well as from the higher molecular weights of the individual chains. The dissociation of the carboxyl groups further enhanced the swelling of the PMAA brush at neutral solution pH, which resulted in the generation of a very lubricious interface with a cushion-layer effect.

The experimental work in chapter 7 was dedicated to the bulk properties of PDMS substrates. The preparation of crosslinked PDMS tribopairs according to 136 8.2. OUTLOOK different recipes allowed for an effective control of their Young’s moduli over one order of magnitude. Although hydrophilic ox-PDMS samples with higher stiffness were expected to exhibit inferior aqueous lubricating properties due to a lower thickness of the lubricant film predicted by soft EHL theory, the opposite behavior was observed. The stiffest ox-PDMS tribopairs were shown to have a more stable oxide layer, judging from the analysis of static water contact angles inside and outside the tribostressed area after pin-on-disk experiments. Hence, it could be concluded that the oxide layers on top of PDMS surfaces generated by plasma treatment can be partially damaged by tribostress, and this damage is more pronounced for the PDMS tribopairs with a lower Young’s modulus. In summary, the systematic investigation of the parameters that govern the aqueous lubrication performance led to the development of surface modifications that are capable of drastically reducing the interfacial friction against an ox-PDMS slider in different aqueous environments. It was further demonstrated that strongly attached molecules sustain macroscopic tribological stress under mild contact-pressure conditions. Among the surface modifications investigated in this thesis, the polyelectrolyte brushes prepared by means of a “grafting from” approach were the most promising candidates as aqueous lubricant additives, not least because it was possible to prepare high surface-density polymer brushes with high charge densities.

8.2 Outlook

The surface modifications investigated in this thesis comprised two methods in which polymers are strongly attached to the substrate. In view of tribology, strongly attached lubricant additives are beneficial under low contact-pressure conditions, where the adsorbed molecules remain attached to the surface during tribological stress. While the “grafting to” method employed in this thesis was shown to be straightforward, the UV-LED-initiated “grafting from” approach involved significantly more experimental work. On the other hand, the developed UV grafting method is expected to exhibit enormous potential for the controlled preparation of polymer brushes with specific surface densities, molecular weights and backbone chemistries, respectively. Consequently, the research could be extended in many directions. It would for instance be interesting to have control over the surface grafting density of the polymer brushes. This could be achieved by changing the spatial distribution of the initiating species on the surface, for instance by the co-adsorption of a second, non-initiating molecule. Different initiator densities are believed to 8.2. OUTLOOK 137 be interesting for two reasons. Firstly, one could try to optimize the polymerization process since it is expected that an overabundance of initiating species leads to excessive chain termination reactions and therefore to less well-defined brushes. Secondly, it would be interesting to investigate the conformational as well as the tribological properties of polymer brushes with different grafting densities. For this purpose, macroscopic pin-on-disk as well as QCM-D experiments are believed to be most suitable. The latter method is regarded as particularly useful since the polymerization can be conveniently monitored in situ and subsequently, the hydration properties of the brushes could be analyzed. In this respect, the interaction of polymer brushes with aqueous solutions of different ionic strengths and pH values would be interesting to study. As demonstrated in this thesis, the lubricant properties can significantly influence the tribological performance of SAMs and polymer brushes. It is expected that changes in the brush conformation will influence the frictional properties of the polymer brushes and a correlation between properties on different length scales would contribute to the understanding of the underlying lubrication mechanisms.

Another potential strategy to enhance the aqueous lubricating properties of different materials comprise surface modifications by means of hydrogels. In order to create such highly hydrated polymer networks, the developed “grafting from” method could be slightly tailored. By incorporating a bifunctional crosslinking agent during or after brush synthesis or by post-irradiation of the prepared brushes, the formation of polyelectrolyte hydrogels can be expected. It would be interesting to compare the lubricating properties of hydrogels and polymer brushes made from identical monomers. In addition, the crosslinking of the as-prepared brushes could be beneficial for the stability of the films during tribological stress.

The preparation of polymer brushes with the “grafting from” method described in this thesis is expected to work on other substrate materials apart from silicon oxide. Besides other metal oxides and glass, grafting from oxidized PDMS surfaces should be possible. Because of the incompatibility of various polymeric substrates with organic solvents, the deposition of the photoiniferter was obtained from the vapor phase and the UV grafting was performed in an aqueous medium. Hence, the preparation of polymer brushes from ox-PDMS surfaces should be rather straightforward. It would then be possible to analyze the friction between two ox-PDMS surfaces, where each of which is bearing a polymer brush. While the tribological characterization of modiﬁed PDMS samples is easily possible, it is far more complicated to investigate the thickness as well as the conformational properties of brushes on transparent substrates. Hence, it was necessary to ﬁrst establish the methodol- 138 8.2. OUTLOOK ogy on silicon oxide surfaces, which are compatible with various surface analytical techniques. The directions in which future research on aqueous lubrication under mild contact pressures could go are rather diverse. The experimental work suggested above could be performed with a number of substrate materials, as mentioned, but also with a large variety of monomers. From this perspective, it is expected that the preparation of polymer brushes with a variety of functionalities can serve as a toolbox for tailoring the surface properties of various materials in order to enhance their aqueous lubrication performance as well as further interfacial properties. Bibliography

[1] U. Raviv, S. Giasson, N. Kampf, J. F. Gohy, R. Jerome, and J. Klein. Lubri- cation by charged polymers. Nature, 425(6954):163–165, 2003.

[2] J. N. Lee, C. Park, and G. M. Whitesides. Solvent compatibility of poly(dimethylsiloxane)-based microﬂuidic devices. Analytical Chemistry, 75(23):6544–6554, 2003.

[3] S. Lee and N. D. Spencer. Aqueous lubrication of polymers: Inﬂuence of surface modiﬁcation. Tribology International, 38(11-12):922–930, 2005.

[4] J. de Vicente, J. R. Stokes, and H. A. Spikes. Lubrication properties of non- adsorbing polymer solutions in soft elastohydrodynamic (ehd) contacts. Tri- bology International, 38(5):515–526, 2005.

[5] D. Dowson. History of Tribology. Professional Engineering Publishing, Lon- don, 2nd edition, 1998.

[6] S.N. Scherge and M. Gorb. Biological Micro- and Nanotribology - Nature’s Solutions. NanoSciene and Technology. Springer, 2001.

[7] H. P. Jost. Tribology - Origin and future. Wear, 136(1):1–17, 1990.

[8] H. P. Jost and J. Schoﬁeld. Energy saving through tribology - A techno- economic study. Proceedings of the Institution of Mechanical Engineers, 195(JUN):151–173, 1981.

[9] I.M. Hutchings. Tribology - Friction and Wear of Engineering Materials. Butterworth-Heinemann Ltd, London, 1992.

[10] M. Esfahanian and B. J. Hamrock. Fluid-ﬁlm lubrication regimes revisited. Tribology Transactions, 34(4):628–632, 1991.

[11] D.A. Rigney. Fundamentals of Friction and Wear of Materials. American Society for Metals, 1980.

139 140 BIBLIOGRAPHY

[12] A. Ulman. Formation and structure of self-assembled monolayers. Chemical Reviews, 96(4):1533–1554, 1996.

[13] W. C. Bigelow, D. L. Pickett, and W. A. Zisman. Oleophobic monolayers. 1. Films adsorbed from solution in non-polar liquids. Journal of Colloid Science, 1(6):513–538, 1946.

[14] R. G. Nuzzo and D. L. Allara. Adsorption of bifunctional organic disulﬁdes on gold surfaces. Journal of the American Chemical Society, 105(13):4481–4483, 1983.

[15] J. C. Love, L. A. Estroﬀ, J. K. Kriebel, R. G. Nuzzo, and G. M. Whitesides. Self-assembled monolayers of thiolates on metals as a form of nanotechnology. Chemical Reviews, 105(4):1103–1169, 2005.

[16] L. H. Dubois and R. G. Nuzzo. Synthesis, structure, and properties of model organic-surfaces. Annual Review of Physical Chemistry, 43:437–463, 1992.

[17] C. D. Bain, E. B. Troughton, Y. T. Tao, J. Evall, G. M. Whitesides, and R. G. Nuzzo. Formation of monolayer ﬁlms by the spontaneous assembly of organic thiols from solution onto gold. Journal of the American Chemical Society, 111(1):321–335, 1989.

[18] R. H. Terrill, T. A. Tanzer, and P. W. Bohn. Structural evolution of hexade- canethiol monolayers on gold during assembly: Substrate and concentration dependence of monolayer structure and crystallinity. Langmuir, 14(4):845–854, 1998.

[19] M. D. Porter, T. B. Bright, D. L. Allara, and C. E. D. Chidsey. Spontaneously organized molecular assemblies. 4. Structural characterization of normal-alkyl thiol monolayers on gold by optical ellipsometry, infrared-spectroscopy, and electrochemistry. Journal of the American Chemical Society, 109(12):3559– 3568, 1987.

[20] S. J. Soﬁa, V. Premnath, and E. W. Merrill. Poly(ethylene oxide) grafted to silicon surfaces: Grafting density and protein adsorption. Macromolecules, 31(15):5059–5070, 1998.

[21] G. L. Kenausis, J. Voros, D. L. Elbert, N. P. Huang, R. Hofer, L. Ruiz-Taylor, M. Textor, J. A. Hubbell, and N. D. Spencer. Poly(l-lysine)-g-poly(ethylene glycol) layers on metal oxide surfaces: Attachment mechanism and eﬀects of polymer architecture on resistance to protein adsorption. Journal of Physical Chemistry B, 104(14):3298–3309, 2000.

[22] L. D. Unsworth, Z. Tun, H. Sheardown, and J. L. Brash. Chemisorption of thiolated poly(ethylene oxide) to gold: Surface chain densities measured by ellipsometry and neutron reﬂectometry. Journal of Colloid and Interface Science, 281(1):112–121, 2005. BIBLIOGRAPHY 141

[23] S Lee and N.D. Spencer. Achieving Ultralow Friction by Aqueous, Brush- Assisted Lubrication. Superlubricity. Elsevier B. V., 2007.

[24] C. D. Bain and G. M. Whitesides. Molecular-level control over surface order in self-assembled monolayer ﬁlms of thiols on gold. Science, 240(4848):62–63, 1988.

[25] C. D. Bain and G. M. Whitesides. Formation of 2-component surfaces by the spontaneous assembly of monolayers on gold from solutions containing mixtures of organic thiols. Journal of the American Chemical Society, 110(19):6560–6561, 1988.

[26] E. P. Plueddemann. Adhesion through silane coupling agents. Journal of Adhesion, 2:184–&, 1970.

[27] P. G. Pape and E. P. Plueddemann. Methods for improving the performance of silane coupling agents. Journal of Adhesion Science and Technology, 5(10):831–842, 1991.

[28] F. D. Osterholtz and E. R. Pohl. Kinetics of the hydrolysis and condensation of organofunctional alkoxysilanes - A review. Journal of Adhesion Science and Technology, 6(1):127–149, 1992.

[29] J. Duchet, B. Chabert, J. P. Chapel, J. F. Gerard, J. M. Chovelon, and N. Jaf- frezic Renault. Inﬂuence of the deposition process on the structure of grafted alkylsilane layers. Langmuir, 13(8):2271–2278, 1997.

[30] G. L. Witucki. A silane primer - Chemistry and applications of alkoxy silanes. Journal of Coatings Technology, 65(822):57–60, 1993.

[31] P. Roumeliotis and K. K. Unger. Structure and properties of normal- alkyldimethylsilyl bonded silica reversed-phase packings. Journal of Chro- matography, 149(FEB):211–224, 1978.

[32] A. Y. Fadeev and T. J. McCarthy. Self-assembly is not the only reaction possible between alkyltrichlorosilanes and surfaces: Monomolecular and oligomeric covalently attached layers of dichloro- and trichloroalkylsilanes on silicon. Langmuir, 16(18):7268–7274, 2000.

[33] M. E. McGovern, K. M. R. Kallury, and M. Thompson. Role of solvent on the silanization of glass with octadecyltrichlorosilane. Langmuir, 10(10):3607– 3614, 1994.

[34] S. Alexander. Polymer adsorption on small spheres - Scaling approach. Journal De Physique, 38(8):977–981, 1977.

[35] P. G. Degennes. Scaling theory of polymer adsorption. Journal De Physique, 37(12):1445–1452, 1976.

[36] P. G. Degennes. Conformations of polymers attached to an interface. Macro- molecules, 13(5):1069–1075, 1980. 142 BIBLIOGRAPHY

[37] P. G. Degennes. Scaling Concepts in Polymer Physics. Cornell University Press, Ithaca and London, 1979. [38] G. Jannink and J. Descloizeaux. Polymers in solution. Journal of Physics- Condensed Matter, 2(1):1–24, 1990. [39] S. T. Milner. Polymer brushes. Science, 251(4996):905–914, 1991. [40] B. Zhao and W. J. Brittain. Polymer brushes: Surface-immobilized macromolecules. Progress in Polymer Science, 25(5):677–710, 2000. [41] Ruhe J; Brittain WJ; Advincula RC; Caster KC, editor. Polymer Brushes: Synthesis, Characterization, Applications. Wiley-VCH, 2004. [42] J. M. Lambert. The nature of platinum in silicones for biomedical and health- care use. Journal of Biomedical Materials Research Part B-Applied Biomate- rials, 78B(1):167–180, 2006. [43] A. M. Bueche. Filler reinforcement of silicone rubber. Journal of Polymer Science, 25(109):139–149, 1957. [44] L. Bokobza. Elastomeric composites. I. Silicone composites. Journal of Applied Polymer Science, 93(5):2095–2104, 2004. [45] S. H. Kim, E. A. Cherney, R. Hackam, and K. G. Rutherford. Chemical- changes at the surface of RTV silicone-rubber coatings on insulators during dry-band arcing. IEEE Transactions on Dielectrics and Electrical Insulation, 1(1):106–123, 1994. [46] J. C. McDonald and G. M. Whitesides. Poly(dimethylsiloxane) as a material for fabricating microfluidic devices. Accounts of Chemical Research, 35(7):491– 499, 2002. [47] J. G. Wankowicz, S. M. Gubanski, and W. D. Lampe. Loss and recovery of hydrophobicity on RTV coating surfaces. IEEE Transactions on Dielectrics and Electrical Insulation, 1(4):604–614, 1994. [48] Y. Berdichevsky, J. Khandurina, A. Guttman, and Y. H. Lo. UV/Ozone modification of poly(dimethylsiloxane) microfluidic channels. Sensors and Ac- tuators B-Chemical, 97(2-3):402–408, 2004. [49] D. C. Duffy, J. C. McDonald, O. J. A. Schueller, and G. M. Whitesides. Rapid prototyping of microfluidic systems in poly(dimethylsiloxane). Analyt- ical Chemistry, 70(23):4974–4984, 1998. [50] K. Glasmastar, J. Gold, A. S. Andersson, D. S. Sutherland, and B. Kasemo. Silicone transfer during microcontact printing. Langmuir, 19(13):5475–5483, 2003. [51] J. M. K. Ng, I. Gitlin, A. D. Stroock, and G. M. Whitesides. Components for integrated poly(dimethylsiloxane) microfluidic systems. Electrophoresis, 23(20):3461–3473, 2002. BIBLIOGRAPHY 143

[52] X. Q. Ren, M. Bachman, C. Sims, G. P. Li, and N. Allbritton. Electroosmotic properties of microfluidic channels composed of poly(dimethylsiloxane). Jour- nal of Chromatography B-Analytical Technologies in the Biomedical and Life Sciences, 762(2):117–125, 2001. [53] C. Thibault, C. Severac, A. F. Mingotaud, C. Vieu, and M. Mauzac. Poly(dimethylsiloxane) contamination in microcontact printing and its influence on patterning oligonucleotides. Langmuir, 23(21):10706–10714, 2007. [54] A. Beigbeder. Use of a marine anti-fouling and fouling release coating composition. WO 2008046166, 2007. [55] Dow Corning Corporation, Midland, Michigan MI. Sylgard 184 Material Safety Data Sheet, 2007. [56] A. Toth, I. Bertoti, M. Blazso, G. Banhegyi, A. Bognar, and P. Szaplonczay. Oxidative damage and recovery of silicone-rubber surfaces. 1. X-ray photo- electron spectroscopic study. Journal of Applied Polymer Science, 52(9):1293– 1307, 1994. [57] K. Efimenko, W. E. Wallace, and J. Genzer. Surface modification of sylgard- 184 poly(dimethyl siloxane) networks by ultraviolet and ultraviolet/ozone treatment. Journal of Colloid and Interface Science, 254(2):306–315, 2002. [58] V. M. Graubner, R. Jordan, O. Nuyken, B. Schnyder, T. Lippert, R. Kotz, and A. Wokaun. Photochemical modification of crosslinked poly(dimethylsiloxane) by irradiation at 172 nm. Macromolecules, 37(16):5936–5943, 2004. [59] M. Morra, E. Occhiello, R. Marola, F. Garbassi, P. Humphrey, and D. Johnson. On the aging of oxygen plasma-treated polydimethylsiloxane surfaces. Journal of Colloid and Interface Science, 137(1):11–24, 1990. [60] H. Hillborg and U. W. Gedde. Hydrophobicity recovery of polydimethylsiloxane after exposure to corona discharges. Polymer, 39(10):1991–1998, 1998. [61] J. Kim, M. K. Chaudhury, and M. J. Owen. Hydrophobic recovery of polydimethylsiloxane elastomer exposed to partial electrical discharge. Journal of Colloid and Interface Science, 226(2):231–236, 2000. [62] M. J. Owen and P. J. Smith. Plasma treatment of polydimethylsiloxane. Journal of Adhesion Science and Technology, 8(10):1063–1075, 1994. [63] A. Olah, H. Hillborg, and G. J. Vancso. Hydrophobic recovery of UV/ozone treated poly(dimethylsiloxane): Adhesion studies by contact mechanics and mechanism of surface modification. Applied Surface Science, 239(3-4):410–423, 2005. [64] H. Hillborg and U. W. Gedde. Hydrophobicity changes in silicone rubbers. IEEE Transactions on Dielectrics and Electrical Insulation, 6(5):703–717, 1999. 144 BIBLIOGRAPHY

[65] H. N. Waltenburg and J. T. Yates. Surface-chemistry of silicon. Chemical Reviews, 95(5):1589–1673, 1995.

[66] R.K. Iler. The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface Properties and Biochemistry of Silica. Wiley-Interscience, 1979.

[67] B.H. Stuart. Infrared Spectroscopy: Fundamentals and Applications. Analyti- cal Techniques in the Sciences (AnTS). John Wiley and Sons, Ltd, 2004.

[68] M. Rodahl, F. Hook, A. Krozer, P. Brzezinski, and B. Kasemo. Quartz-crystal microbalance setup for frequency and Q-factor measurements in gaseous and liquid environments. Review of Scientiﬁc Instruments, 66(7):3924–3930, 1995.

[69] R. W. Carpick and M. Salmeron. Scratching the surface: Fundamental investigations of tribology with atomic force microscopy. Chemical Reviews, 97(4):1163–1194, 1997.

[70] G. J. Leggett, N. J. Brewer, and K. S. L. Chonga. Friction force microscopy: Towards quantitative analysis of molecular organisation with nanometre spatial resolution. Physical Chemistry Chemical Physics, 7(6):1107–1120, 2005.

[71] C. M. Mate, G. M. McClelland, R. Erlandsson, and S. Chiang. Atomic-scale friction of a tungsten tip on a graphite surface. Physical Review Letters, 59(17):1942–1945, 1987.

[72] A. Marti, G. Hahner, and N. D. Spencer. Sensitivity of frictional forces to pH on a nanometer scale: A lateral force microscopy study. Langmuir, 11(12):4632–4635, 1995.

[73] B. D. Beake and G. J. Leggett. Variation of frictional forces in air with the compositions of heterogeneous organic surfaces. Langmuir, 16(2):735–739, 2000.

[74] N. J. Brewer, B. D. Beake, and G. J. Leggett. Friction force microscopy of self-assembled monolayers: Inﬂuence of adsorbate alkyl chain length, terminal group chemistry, and scan velocity. Langmuir, 17(6):1970–1974, 2001.

[75] S. C. Clear and P. F. Nealey. Chemical force microscopy study of adhesion and friction between surfaces functionalized with self-assembled monolayers and immersed in solvents. Journal of Colloid and Interface Science, 213(1):238– 250, 1999.

[76] S. C. Clear and P. F. Nealey. Lateral force microscopy study of the frictional behavior of self-assembled monolayers of octadecyltrichlorosilane on silicon/silicon dioxide immersed in n-alcohols. Langmuir, 17(3):720–732, 2001.

[77] C. D. Frisbie, L. F. Rozsnyai, A. Noy, M. S. Wrighton, and C. M. Lieber. Functional-group imaging by chemical force microscopy. Science, 265(5181):2071–2074, 1994. BIBLIOGRAPHY 145

[78] H. Ishida, T. Koga, M. Morita, H. Otsuka, and A. Takahara. Macro- and nanotribological properties of organosilane monolayers prepared by a chemical vapor adsorption method on silicon substrates. Tribology Letters, 19(1):3–8, 2005.

[79] H. I. Kim, T. Koini, T. R. Lee, and S. S. Perry. Systematic studies of the frictional properties of ﬂuorinated monolayers with atomic force microscopy: Comparison of CF3- and CH3-terminated ﬁlms. Langmuir, 13(26):7192–7196, 1997.

[80] S. Lee, Y. S. Shon, R. Colorado, R. L. Guenard, T. R. Lee, and S. S. Perry. The inﬂuence of packing densities and surface order on the frictional properties of alkanethiol self-assembled monolayers (SAMs) on gold: A comparison of SAMs derived from normal and spiroalkanedithiols. Langmuir, 16(5):2220– 2224, 2000.

[81] S. Lee, A. Puck, M. Graupe, R. Colorado, Y. S. Shon, T. R. Lee, and S. S. Perry. Structure, wettability, and frictional properties of phenyl-terminated self-assembled monolayers on gold. Langmuir, 17(23):7364–7370, 2001.

[82] A. Lio, D. H. Charych, and M. Salmeron. Comparative atomic force microscopy study of the chain length dependence of frictional properties of alkanethiols on gold and alkylsilanes on mica. Journal of Physical Chemistry B, 101(19):3800–3805, 1997.

[83] M. Nakano, T. Ishida, T. Numata, Y. Ando, and S. Sasaki. Alkyl chain length eﬀect on tribological behavior of alkanethiol self-assembled monolayers on Au. Japanese Journal Of Applied Physics Part 1-Regular Papers Short Notes & Review Papers, 42(7B):4734–4738, 2003.

[84] M. Nakano, T. Ishida, T. Numata, Y. Ando, and S. Sasaki. Tribological behavior of terphenyl self-assembled monolayer studied by a pin-on-plate method and friction force microscopy. Japanese Journal Of Applied Physics Part 1- Regular Papers Short Notes & Review Papers, 43(7B):4619–4623, 2003.

[85] A. Nichols and S. C. Street. Spectroscopic analysis of the tribological behavior of a model boundary layer lubricant. Analyst, 126(8):1269–1273, 2001.

[86] A. Noy, D. V. Vezenov, and C. M. Lieber. Chemical force microscopy. Annual Review of Materials Science, 27:381–421, 1997.

[87] S. Ren, S. R. Yang, Y. P. Zhao, J. F. Zhou, T. Xu, and W. M. Liu. Friction and wear studies of octadecyltrichlorosilane SAM on silicon. Tribology Letters, 13(4):233–239, 2002.

[88] S. K. Sinniah, A. B. Steel, C. J. Miller, and J. E. ReuttRobey. Solvent exclusion and chemical contrast in scanning force microscopy. Journal of the American Chemical Society, 118(37):8925–8931, 1996. 146 BIBLIOGRAPHY

[89] I. H. Sung and D. E. Kim. Surface damage characteristics of self-assembled monolayers of alkanethiols on metal surfaces. Tribology Letters, 17(4):835–844, 2004.

[90] E. W. van der Vegte, A. Subbotin, G. Hadziioannou, P. R. Ashton, and J. A. Preece. Nanotribological properties of unsymmetrical n-dialkyl sulﬁde monolayers on gold: Eﬀect of chain length on adhesion, friction, and imaging. Lang- muir, 16(7):3249–3256, 2000.

[91] D. V. Vezenov, A. Noy, L. F. Rozsnyai, and C. M. Lieber. Force titrations and ionization state sensitive imaging of functional groups in aqueous solutions by chemical force microscopy. Journal of the American Chemical Society, 119(8):2006–2015, 1997.

[92] S. S. Wong, H. Takano, and M. D. Porter. Mapping orientation diﬀerences of terminal functional groups by friction force microscopy. Analytical Chemistry, 70(24):5209–5212, 1998.

[93] X. D. Xiao, J. Hu, D. H. Charych, and M. Salmeron. Chain length dependence of the frictional properties of alkylsilane molecules self-assembled on mica studied by atomic force microscopy. Langmuir, 12(2):235–237, 1996.

[94] W. Hardy and I. Bircumshaw. Bakerian lecture - Boundary lubrication - Plane surfaces and the limitations of Amontons’ law. Proceedings of the Royal Society of London Series A-Containing Papers of a Mathematical and Physical Character, 108(745):1–27, 1925.

[95] O. Levine and W. A. Zisman. Physical properties of monolayers adsorbed at the solid-air interface. 2. Mechanical durability of aliphatic polar compounds and eﬀect of halogenation. Journal of Physical Chemistry, 61(9):1188–1196, 1957.

[96] O. Levine and W. A. Zisman. Physical properties of monolayers adsorbed at the solid-air interface. 1. Friction and wettability of aliphatic polar compounds and eﬀect of halogenation. Journal of Physical Chemistry, 61(8):1068–1077, 1957.

[97] D. Tabor. Boundary and extreme pressure lubrication. 3. Discussion - Mech- anism of boundary lubrication. Proceedings of the Royal Society of London Series A-Mathematical and Physical Sciences, 212(1111):498–505, 1952.

[98] R. W. Carpick, N. Agrait, D. F. Ogletree, and M. Salmeron. Measurement of interfacial shear (friction) with an ultrahigh vacuum atomic force microscope. Journal of Vacuum Science & Technology B, 14(2):1289–1295, 1995.

[99] A. Schallamach. How does rubber slide. Wear, 17(4):301–&, 1971.

[100] H. R. Brown. Chain pullout and mobility eﬀects in friction and lubrication. Science, 263(5152):1411–1413, 1994. BIBLIOGRAPHY 147

[101] J. Greenwood, D. Tabor, and H. Minshall. Hysteresis losses in rolling and sliding friction. Proceedings of the Royal Society of London Series A-Mathematical and Physical Sciences, 259(129):480–&, 1961.

[102] D. Dowson and Z. M. Jin. Microelastohydrodynamic lubrication of low-elastic- modulus solids on rigid substrates. Journal of Physics D - Applied Physics, 25(1A):A116–A123, 1991.

[103] D. Dowson and B. J. Hamrock. Minimum film thickness in elliptical contacts for different regimes of fluid film lubrication, pages 22–27. Mech. Eng. Publ. Bury St. Edmunds, Suffolk, 1979.

[104] S. Lee, R. Iten, M. Muller, and N. D. Spencer. Inﬂuence of molecular architecture on the adsorption of poly(ethylene oxide)-poly(propylene oxide)- poly(ethylene oxide) on PDMS surfaces and implications for aqueous lubrication. Macromolecules, 37(22):8349–8356, 2004.

[105] J. B. Medley, A. B. Strong, R. M. Pilliar, and E. W. Wong. The breakdown of ﬂuid ﬁlm lubrication in elastic-isoviscous point contacts. Wear, 63(1):25–40, 1980.

[106] D. F. Moore. Elastohydrodynamic transition speed for spheres sliding on lubricated rubber. Wear, 35(1):159–170, 1975.

[107] S. C. Richards and A. D. Roberts. Boundary lubrication of rubber by aqueous surfactant. Journal of Physics D - Applied Physics, 25(1A):A76–A80, 1991.

[108] A. D. Roberts and D. Tabor. Extrusion of liquids between highly elastic solids. Proceedings of the Royal Society of London Series A-Mathematical and Physical Sciences, 325(1562):323–&, 1971.

[109] R. G. Nuzzo, L. H. Dubois, and D. L. Allara. Fundamental-studies of micro- scopic wetting on organic-surfaces. 1. Formation and structural characterization of a self-consistent series of polyfunctional organic monolayers. Journal of the American Chemical Society, 112(2):558–569, 1990.

[110] R. G. Nuzzo, F. A. Fusco, and D. L. Allara. Spontaneously organized molecular assemblies. 3. Preparation and properties of solution adsorbed monolayers of organic disulﬁdes on gold surfaces. Journal of the American Chemical Society, 109(8):2358–2368, 1987.

[111] F. Bensebaa, R. Voicu, L. Huron, T. H. Ellis, and E. Kruus. Kinetics of formation of long-chain n-alkanethiolate monolayers on polycrystalline gold. Langmuir, 13(20):5335–5340, 1997.

[112] K. D. Truong and P. A. Rowntree. Formation of self-assembled butanethiol monolayers on au substrates: Spectroscopic evidence for highly ordered is- land formation in sub-monolayer ﬁlms. Journal of Physical Chemistry, 100(51):19917–19926, 1996. 148 BIBLIOGRAPHY

[113] I. M. Hutchings. Tribology - Friction and Wear of Engineering Materials, Chapter 4, pages 67–70. Butterworth Heinemann, 1992.

[114] E. Barrena, S. Kopta, D. F. Ogletree, D. H. Charych, and M. Salmeron. Relationship between friction and molecular structure: Alkylsilane lubricant ﬁlms under pressure. Physical Review Letters, 82(14):2880–2883, 1999.

[115] T. P. Weihs, Z. Nawaz, S. P. Jarvis, and J. B. Pethica. Limits of imaging resolution for atomic force microscopy of molecules. Applied Physics Letters, 59(27):3536–3538, 1991.

[116] J. Klein. Shear, friction, and lubrication forces between polymer-bearing surfaces. Annual Review of Materials Science, 26:581–612, 1996.

[117] S. Lee and N. D. Spencer. Superlubricity, Chapter Achieving Ultralow Friction by Aqueous, Brush-Assisted Lubrication, pages p–365. Elsevier, 2007.

[118] S. Lee and N. D. Spencer. Materials science - Sweet, hairy, soft, and slippery. Science, 319(5863):575–576, 2008.

[119] F. T. Limpoco, R. C. Advincula, and S. S. Perry. Solvent dependent friction force response of polystyrene brushes prepared by surface initiated polymerization. Langmuir, 23(24):12196–12201, 2007.

[120] M. T. Muller, X. P. Yan, S. W. Lee, S. S. Perry, and N. D. Spencer. Lubrication properties of a brushlike copolymer as a function of the amount of solvent absorbed within the brush. Macromolecules, 38(13):5706–5713, 2005.

[121] M. T. Muller, X. P. Yan, S. W. Lee, S. S. Perry, and N. D. Spencer. Prefer- ential solvation and its eﬀect on the lubrication properties of a surface-bound, brushlike copolymer. Macromolecules, 38(9):3861–3866, 2005.

[122] T. Drobek and N. D. Spencer. Nanotribology of surface-grafted PEG layers in an aqueous environment. Langmuir, 24(4):1484–1488, 2008.

[123] T. Pettersson, A. Naderi, R. Makuska, and P. M. Claesson. Lubrication properties of bottle-brush polyelectrolytes: An AFM study on the eﬀect of side chain and charge density. Langmuir, 24(7):3336–3347, 2008.

[124] L. D. Unsworth, H. Sheardown, and J. L. Brash. Protein-resistant poly(ethylene oxide)-grafted surfaces: Chain density-dependent multiple mechanisms of action. Langmuir, 24(5):1924–1929, 2008.

[125] M. Ataman. Properties of aqueous salt-solutions of poly(ethylene oxide) - Cloud points, theta-temperatures. Colloid and Polymer Science, 265(1):19– 25, 1987.

[126] F. E. Bailey and R. W. Callard. Some properties of poly(ethylene oxide) in aqueous solution. Journal of Applied Polymer Science, 1(1):56–62, 1959. BIBLIOGRAPHY 149

[127] F. E. Bailey and J. V. Koleske. Poly(ethylene oxide). Academic Press: New York, 1976.

[128] E. A. Boucher and P. M. Hines. Eﬀects of inorganic salts on properties of aqueous poly(ethylene oxide) solutions. Journal of Polymer Science Part B- Polymer Physics, 14(12):2241–2251, 1976.

[129] B. Briscoe, P. Luckham, and S. Zhu. Rheological study of poly(ethylene oxide) in aqueous salt solutions at high temperature and pressure. Macromolecules, 29(19):6208–6211, 1996.

[130] R. D. Lundberg, F. E. Bailey, and R. W. Callard. Interactions of inorganic salts with poly(ethylene oxide). Journal of Polymer Science Part A-1-Polymer Chemistry, 4(6PA1):1563–&, 1966.

[131] Y. Masuda and T. Nakanishi. Ion-speciﬁc swelling behavior of poly(ethylene oxide) gel and the correlation to the intrinsic viscosity of the polymer in salt solutions. Colloid and Polymer Science, 280(6):547–553, 2002.

[132] K. Tasaki. Poly(oxyethylene)-cation interactions in aqueous solution: A molecular dynamics study. Computational and Theoretical Polymer Science, 9(3- 4):271–284, 1999.

[133] C. J. Pedersen. Cyclic polyethers and their complexes with metal salts. Journal of the American Chemical Society, 89(26):7017–&, 1967.

[134] H. K. Frensdorﬀ. Stability constants of cyclic polyether complexes with uni- valent cations. Journal of the American Chemical Society, 93(3):600–&, 1971.

[135] E. Florin, R. Kjellander, and J. C. Eriksson. Salt eﬀects on the cloud point of the poly(ethylene oxide) + water-system. Journal of the Chemical Society- Faraday Transactions I, 80:2889–2910, 1984. Part 11.

[136] S. J. Oshea, M. E. Welland, and T. Rayment. An atomic-force microscope study of grafted polymers on mica. Langmuir, 9(7):1826–1835, 1993.

[137] M. Bjorling, G. Karlstrom, and P. Linse. Conformational adaption of poly(ethylene oxide) - A C-13 NMR-study. Journal of Physical Chemistry, 95(17):6706–6709, 1991.

[138] S. Lee, M. Muller, R. Heeb, S. Zurcher, S. Tosatti, M. Heinrich, F. Amstad, S. Pechmann, and N. D. Spencer. Self-healing behavior of a polyelectrolyte- based lubricant additive for aqueous lubrication of oxide materials. Tribology Letters, 24(3):217–223, 2006.

[139] M. Masuko, H. Miyamoto, and A. Suzuki. Tribological characteristics of self- assembled monolayer with siloxane bonding to si surface. Tribology Interna- tional, 40(10–12):1587–1596, 2006. 150 BIBLIOGRAPHY

[140] H. S. Ahn, P. D. Cuong, S. Park, Y. W. Kim, and J. C. Lim. Eﬀect of molecular structure of self-assembled monolayers on their tribological behaviors in nano- and microscales. Wear, 255:819–825, 2003.

[141] N. J. Brewer, B. D. Beake, and G. J. Leggett. Friction force microscopy of self-assembled monolayers: Inﬂuence of adsorbate alkyl chain length, terminal group chemistry, and scan velocity. Langmuir, 17(6):1970–1974, 2001.

[142] Y. S. Shon, S. Lee, R. Colorado, S. S. Perry, and T. R. Lee. Spiroalkanedithiol- based SAMs reveal unique insight into the wettabilities and frictional properties of organic thin ﬁlms. Journal of the American Chemical Society, 122(31):7556–7563, 2000.

[143] X. J. Yang and S. S. Perry. Friction and molecular order of alkanethiol self-assembled monolayers on Au(111) at elevated temperatures measured by atomic force microscopy. Langmuir, 19(15):6135–6139, 2003.

[144] D. J. Vanderah, G. Valincius, and C. W. Meuse. Self-assembled monolayers of methyl 1-thiahexa(ethylene oxide) for the inhibition of protein adsorption. Langmuir, 18(12):4674–4680, 2002.

[145] D. J. Vanderah, J. Arsenault, H. La, R. S. Gates, V. Silin, C. W. Meuse, and G. Valincius. Structural variations and ordering conditions for the self- assembled monolayers of HS(CH2CH2O)(3-6)CH3. Langmuir, 19(9):3752– 3756, 2003.

[146] L. Malysheva, A. Onipko, and B. Liedberg. Ab initio modeling of defect signa- tures in infrared reﬂection-absorption spectra of SAMs exposing methyl- and hydrogen-terminated oligo(ethylene glycols). Journal of Physical Chemistry A, 112(4):728–736, 2008.

[147] S. Tokumitsu, A. Liebich, S. Herrwerth, W. Eck, M. Himmelhaus, and M. Grunze. Grafting of alkanethiol-terminated poly(ethylene glycol) on gold. Langmuir, 18(23):8862–8870, 2002.

[148] S. L. Clark, M. F. Montague, and P. T. Hammond. Ionic eﬀects of sodium chloride on the templated deposition of polyelectrolytes using layer-by-layer ionic assembly. Macromolecules, 30(23):7237–7244, 1997.

[149] R. W. Callard and F. E. Bailey Jr. Some properties of poly(ethylene oxide) in aqueous solution. Journal of Applied Polymer Science, 1(1):56–62, 1959.

[150] A. D. Delman, M. Landy, and B. B. Simms. Photodecomposition of poly- methylsiloxane. Journal of Polymer Science Part A-1-Polymer Chemistry, 7(12PA):3375–3386, 1969.

[151] R. F. Bartholo and T. R. Kozlowsk. Alkali attack of glass surfaces by molten salts. Journal of the American Ceramic Society, 50(2):108–&, 1967. BIBLIOGRAPHY 151

[152] F. Miyaji, M. Iwai, T. Kokubo, and T. Nakamura. Chemical surface treatment of silicone for inducing its bioactivity. Journal of Materials Science-Materials in Medicine, 9(2):61–65, 1998.

[153] H. Zengin, B. Hu, J. A. Siddiqui, and R. M. Ottenbrite. Surface modiﬁcation of glass beads with poly(acrylic acid). Polymers for Advanced Technologies, 17(5):372–378, 2006.

[154] R. Dong, S. Krishnan, B. A. Baird, M. Lindau, and C. K. Ober. Patterned bio- functional poly(acrylic acid) brushes on silicon surfaces. Biomacromolecules, 8(10):3082–3092, 2007.

[155] J. Ruhe, M. Ballauﬀ, M. Biesalski, P. Dziezok, F. Grohn, D. Johannsmann, N. Houbenov, N. Hugenberg, R. Konradi, S. Minko, M. Motornov, R. R. Netz, M. Schmidt, C. Seidel, M. Stamm, T. Stephan, D. Usov, and H. N. Zhang. Polyelectrolytes with Deﬁned Molecular Architecture I, Volume 165 of Advances in Polymer Science, Chapter Polyelectrolyte brushes, pages 79–150. 2004.

[156] J. Ruhe. Polymer Brushes: On the Way to Tailor-Made Surfaces. Polymer Brushes: Synthesis, Characterization, Applications. Wiley-VCH, 2004.

[157] S. Edmondson, V. L. Osborne, and W. T. S. Huck. Polymer brushes via surface-initiated polymerizations. Chemical Society Reviews, 33(1):14–22, 2004.

[158] Y. Tsujii, K. Ohno, S. Yamamoto, A. Goto, and T. Fukuda. Surface-Initiated Polymerization I, Volume 197 of Advances in Polymer Science, Chapter Struc- ture and properties of high-density polymer brushes prepared by surface- initiated living radical polymerization, pages 1–45. 2006.

[159] J. S. Wang and K. Matyjaszewski. Controlled living radical polymerization - Halogen atom-transfer radical polymerization promoted by a Cu(I)Cu(II) redox process. Macromolecules, 28(23):7901–7910, 1995.

[160] D. Benoit, V. Chaplinski, R. Braslau, and C. J. Hawker. Development of a universal alkoxyamine for “living” free radical polymerizations. Journal of the American Chemical Society, 121(16):3904–3920, 1999.

[161] C. J. Hawker, G. G. Barclay, A. Orellana, J. Dao, and W. Devonport. Initiating systems for nitroxide-mediated ”living” free radical polymerizations: Synthesis and evaluation. Macromolecules, 29(16):5245–5254, 1996.

[162] C. J. Hawker, A. W. Bosman, and E. Harth. New polymer synthesis by nitroxide mediated living radical polymerizations. Chemical Reviews, 101(12):3661– 3688, 2001.

[163] J. Chiefari, Y. K. Chong, F. Ercole, J. Krstina, J. Jeﬀery, T. P. T. Le, R. T. A. Mayadunne, G. F. Meijs, C. L. Moad, G. Moad, E. Rizzardo, and S. H. Thang. Living free-radical polymerization by reversible addition-fragmentation chain transfer: The RAFT process. Macromolecules, 31(16):5559–5562, 1998. 152 BIBLIOGRAPHY

[164] G. Moad, E. Rizzardo, and S. H. Thang. Living radical polymerization by the RAFT process. Australian Journal of Chemistry, 58(6):379–410, 2005. [165] T. Otsu and M. Yoshida. Role of initiator-transfer agent-terminator (iniferter) in radical polymerizations - Polymer design by organic disulﬁdes as iniferters. Makromolekulare Chemie-Rapid Communications, 3(2):127–132, 1982. [166] T. Otsu, M. Yoshida, and T. Tazaki. A model for living radical polymerization. Makromolekulare Chemie-Rapid Communications, 3(2):133–140, 1982. [167] N. Luo, A. T. Metters, J. B. Hutchison, C. N. Bowman, and K. S. Anseth. A methacrylated photoiniferter as a chemical basis for microlithography: Micropatterning based on photografting polymerization. Macromolecules, 36(18):6739–6745, 2003. [168] Y. Nakayama and T. Matsuda. Surface macromolecular architectural designs using photo-graft copolymerization based on photochemistry of benzyl n,n- diethyldithiocarbamate. Macromolecules, 29(27):8622–8630, 1996. [169] D. Morita, M. Sano, M. Yamamoto, M. Nonaka, K. Yasutomo, K. Akaishi, S. Nagahama, and T. Mukai. Over 200 mW on 365 nm ultraviolet light emitting diode of GaN-free structure. Physica Status Solidi A-Applied Research, 200(1):114–117, 2003. [170] D. Morita, M. Sano, M. Yamamoto, K. Matoba, K. Yasutomo, K. Akaishi, Y. Kasai, S. Nagahama, and T. Mukai. Quantum Sensing and Nanophotonic Devices, Volume 5359 of Proceedings of the Society of Photo-Optical Instru- mentation Engineers (Spie), Chapter 365-nm ultraviolet light emitting diodes with an output power of over 400 mW, pages 415–421. 2004. [171] D. Morita, M. Yamamoto, K. Akaishi, K. Matoba, K. Yasutomo, Y. Kasai, M. Sano, S. Nagahama, and T. Mukai. Watt-class high-output-power 365 nm ultraviolet light-emitting diodes. Japanese Journal of Applied Physics Part 1-Regular Papers Short Notes & Review Papers, 43(9A):5945–5950, 2004. [172] S. Lee, R. Heeb, N. V. Venkataraman, and N. D. Spencer. Macroscopic tribological testing of alkanethiol self-assembled monolayers (SAMs): Pin-on-disk tribometry with elastomeric sliding contacts. Tribology Letters, 28(3):229–239, 2007. [173] B. de Boer, H. K. Simon, M. P. L. Werts, E. W. van der Vegte, and G. Hadzi- ioannou. “Living” free radical photopolymerization initiated from surface- grafted iniferter monolayers. Macromolecules, 33(2):349–356, 2000. [174] M. Biesalski, D. Johannsmann, and J. Ruhe. Synthesis and swelling behavior of a weak polyacid brush. Journal of Chemical Physics, 117(10):4988–4994, 2002. [175] P. Jain, J. H. Dai, G. L. Baker, and M. L. Bruening. Rapid synthesis of functional polymer brushes by surface-initiated atom transfer radical polymerization of an acidic monomer. Macromolecules, 41(22):8413–8417, 2008. BIBLIOGRAPHY 153

[176] R. Konradi and J. Ruhe. Interaction of poly(methacrylic acid) brushes with metal ions: An infrared investigation. Macromolecules, 37(18):6954–6961, 2004. [177] K. Matyjaszewski, P. J. Miller, N. Shukla, B. Immaraporn, A. Gelman, B. B. Luokala, T. M. Siclovan, G. Kickelbick, T. Vallant, H. Hoffmann, and T. Pakula. Polymers at interfaces: Using atom transfer radical polymerization in the controlled growth of homopolymers and block copolymers from silicon surfaces in the absence of untethered sacrificial initiator. Macromolecules, 32(26):8716–8724, 1999. [178] A. J. Parnell, S. J. Martin, C. C. Dang, M. Geoghegan, R. A. L. Jones, C. J. Crook, J. R. Howse, and A. J. Ryan. Synthesis, characterization and swelling behaviour of poly(methacrylic acid) brushes synthesized using atom transfer radical polymerization. Polymer, 50(4):1005–1014, 2009. [179] T. Wu, P. Gong, I. Szleifer, P. Vlcek, V. Subr, and J. Genzer. Behavior of surface-anchored poly(acrylic acid) brushes with grafting density gradients on solid substrates: 1. Experiment. Macromolecules, 40(24):8756–8764, 2007. [180] J. L. Fritz and M. J. Owen. Hydrophobic recovery of plasma-treated polydimethylsiloxane. Journal of Adhesion, 54(1-2):33–45, 1995. [181] S. B. Rahane, S. M. Kilbey, and A. T. Metters. Kinetics of surface-initiated photoiniferter-mediated photopolymerization. Macromolecules, 38(20):8202– 8210, 2005. [182] J. B. Kim, W. X. Huang, M. D. Miller, G. L. Baker, and M. L. Bruening. Kinetics of surface-initiated atom transfer radical polymerization. Journal of Polymer Science Part A-Polymer Chemistry, 41(3):386–394, 2003. [183] M. Niwa, M. Date, and N. Higashi. In situ photopolymerization of methacrylic acid at a self-assembled xanthate monolayer surface on gold. Formation of poly(methacrylic acid) brushes and their interaction with cytochrome C. Macromolecules, 29(11):3681–3685, 1996. [184] D. Q. Xiao and M. J. Wirth. Kinetics of surface-initiated atom transfer radical polymerization of acrylamide on silica. Macromolecules, 35(8):2919–2925, 2002. [185] G. Y. Jung, Z. Y. Li, W. Wu, Y. Chen, D. L. Olynick, S. Y. Wang, W. M. Tong, and R. S. Williams. Vapor-phase self-assembled monolayer for improved mold release in nanoimprint lithography. Langmuir, 21(4):1158–1161, 2005. [186] S. B. Rahane, S. M. Kilbey, and A. T. Metters. Kinetic modeling of surface-initiated photoiniferter-mediated photopolymerization in presence of tetraethylthiuram disulfide. Macromolecules, 41(24):9612–9618, 2008. [187] Y. Nakayama and T. Matsuda. In situ observation of dithiocarbamate- based surface photograft copolymerization using quartz crystal microbalance. Macromolecules, 32(16):5405–5410, 1999. 154 BIBLIOGRAPHY

[188] M. A. C. Stuart, W. M. de Vos, and F. A. M. Leermakers. Why surfaces modified by flexible polymers often have a finite contact angle for good solvents. Langmuir, 22(4):1722–1728, 2006.

[189] J. H. H. Bongaerts, K. Fourtouni, and J. R. Stokes. Soft-tribology: Lubrica- tion in a compliant PDMS-PDMS contact. 33rd Leeds-Lyon Symposium on Tribology, 2006.

[190] J. H. H. Bongaerts, D. Rossetti, and J. R. Stokes. The lubricating properties of human whole saliva. Tribology Letters, 27(3):277–287, 2007.

[191] M. J. Owen and P. J. Smith. Plasma treatment of polydimethylsiloxane. In- ternational Symposium on Polymer Surface Modiﬁcation - Relevance to Ad- hesion, 1993.

[192] H. Hillborg, J. F. Ankner, U. W. Gedde, G. D. Smith, H. K. Yasuda, and K. Wikstrom. Crosslinked polydimethylsiloxane exposed to oxygen plasma studied by neutron reﬂectometry and other surface speciﬁc techniques. Poly- mer, 41(18):6851–6863, 2000.

[193] H. Schmid and B. Michel. Siloxane polymers for high-resolution, high-accuracy soft lithography. Macromolecules, 33(8):3042–3049, 2000. Acknowledgments

During my PhD thesis, I received great support from numerous people and they all contributed to this work in one or the other way. Herewith, I would like to acknowledge their help and to thank everybody for their efforts. First, I would like to especially thank Prof. Nicholas D. Spencer for having offered me a PhD position in his group. I very much appreciated his helpful advice and especially his companionate leadership. Also, I would like to thank Nic for his patience with my project. A very big thank you goes to Prof. Seunghwan Lee for being a great supervisor. From our first contact, when Seunghwan was supervising my semester project, I appreciated his profound knowledge in tribology and surface science and Seunghwan actually inspired me to do my PhD thesis in this field. I would also like to thank Seunghwan for his great friendship and patience with my project. I would also like to sincerely thank Prof. Hugh Spikes for accepting the invitation as a co-examiner. Thanks also to Prof. Walter Steurer, for representing the department at my PhD examination. ETH Zurich is acknowledged for generously funding this thesis.

Many thanks go to everybody at the Laboratory for Surface Science and Technology (LSST), especially: Dr. Nagaiyanallur V. Venkataraman for his great and friendly support in the field of infrared spectroscopy and self-assembled monolayers. I really appreciated Venky’s good ideas as well as his inputs and co-authorship in two publications. Dr. Lucy Yuxin Clasohm for her fast help with AFM roughness measurements when the time was short towards the end of my thesis. Dr. Erik Reimhult for his help with my last-minute QCM-D experiments as well as with the interpretation of the data. Dr. Tobias Balmer for his advice concerning optics as well as for interesting discussions. Dr. Stefan Zürcher and Dr. Samuele Tosatti, SuSoS AG, for their inputs regarding surface chemistry as well as for generously sponsoring wine at the annual offsite group meetings.

155 156

Torben Gillich for his great help in the chemistry lab. Torben’s inputs in the field of surface-initiated polymerization helped me a lot and I very much appreciated his friendship. Christoph Mayer for his help in the chemistry lab and for organizing nice bar- beques and other social events for the whole group as well as for his friendship. Eva Beurer, Maura Crobu and Filippo Mangolini for being reliable lunch partners as well as for their dinner invitations. Doris Spori for being a great lab manager and for looking after everybody in the E floor. Whitney Hartung for sharing the aqueous lubrication project with me and for numerous discussions. Mathias Rodenstein for his inputs regarding ultraviolet light-emitting diodes (UV-LEDs). I would also like to thank all other members of LSST, for interesting scientific or private discussions, for entertaining coffee breaks, for organizing social events and for making this group very special. Furthermore, I would like to thank the students who worked on my project, Re- bekka Ginés,Lars Fleig and Robert M. Bielecki. Robert continued to help me as a Hilfsassistent in my project and I would like to additionally thank him for his great effort and his friendship.

There were a number of people outside LSST who supported me during my thesis and I would like to also thank them for their help: Dr. Rupert Konradi, BASF, for his expertise in the ﬁeld of surface-initiated polymerization, for general hints and suggestions as well as for the proof-reading of one publication. Dr. Andreas M¨uhlebach, CIBA Specialty Chemicals, for his inputs and discussions regarding photoinitiated surface polymerization. Dr. Heiko Wolf, IBM Research Laboratory Zurich, for his help and suggestions regarding PDMS elastomers. Dr. Kirill Feldman, Polymer Technology Group, for the introduction and the use of the tensile tester. Lorenz Bonderer, Nonmetallic Inorganic Materials Group, for the SEM experiments as well as for his friendship and numerous interesting tennis matches.

A very special thank you goes to my parents who supported me during all these years. Thank you very much for having given me the opportunity to focus on my education and for believing in me. Last but not least, my warmest thank you goes to Kathrin for her love, care and patience during all these years, for showing interest in my research and for continuously motivating me and cheering me up. Curriculum Vitae

Raphael Emanuel Heeb

Date of Birth: January 16, 1980 Nationality: Swiss, Citizen of Altst¨atten(SG) Principality of Liechtenstein Present Address: Almeisliweg 2 CH-9475 Sevelen Switzerland

Education

2005 - 2009 Doctoral student at the Laboratory for Surface Science and Technology (LSST), Department of Materials, ETH Zurich 2000 - 2005 Studies of Material Science at ETH Zurich, Graduation with the degree Dipl. Werkstoﬀ-Ing. ETH 1995 - 1999 High School in Heerbrugg (SG), Matura Typus C

Practical Experience

2006 - 2009 Supervison of three master projects, Laboratory for Surface Science and Technology, ETH Zurich 2005 Temporary research associate with Unilever Corporate Re- search, Colworth, U.K. 2003 Internship at Hilti AG, Schaan, Principality of Liechten- stein 2002 - 2004 Teaching assistant at the Institute for Mechanical Systems, ETH Zurich

157 158

Presentations

2008 The role of surface properties in aqueous lubrication (poster presentation) R. Heeb, S. Lee, N.V. Venkataraman, N.D. Spencer Gordon Research Conference on Tribology, July 6 11, Waterville, Maine, USA

2008 The solid-liquid interface in aqueous lubrication (oral presentation) R. Heeb, S. Lee, N.D. Spencer SAOG-GSSI 24th Annual Meeting, January 25, Fribourg, Switzer- land

2008 Die Rolle der Oberfläche bei wasserbasierter Schmierung (oral presentation) R. Heeb, S. Lee, N.D. Spencer CSM Workshop, February 22, Dübendorf, Switzerland

2008 Surface modiﬁcations for the reduction of friction in water (oral presentation) R. Heeb, S. Lee, N.V. Venkataraman, N.D. Spencer Material Science Colloquium, May 14, ETH Zurich, Switzerland

2007 Soft-elastohydrodynamic lubrication employing alkanethiol self- assembled monolayers (poster presentation) R. Heeb, S. Lee, N.V. Venkataraman, N.D. Spencer International Nanotribology Forum, March 26-30, Hoi An, Vietnam

2006 Surface modiﬁcations of elastomeric materials for aqueous lubrication (poster presentation) R. Heeb, S. Lee, N.D. Spencer Sliding dynamics: the physical and mechanical viewpoints on friction, November 19 24, Lyon, France 159

Publications

2009 R. Heeb, R.M. Bielecki, S. Lee, N.D. Spencer, Room-Temperature, Aqueous-Phase Fabrication of Poly(methacrylic acid) Brushes by UV-LED-Induced, Controlled Radical Polymerization with High Se- lectivity for Surface-Bound Species. Macromolecules, 2009. 42(22): p. 9124-9132.

2009 R. Heeb, S. Lee, N.V. Venkataraman, N.D. Spencer, Inﬂuence of Salt on the Aqueous Lubrication Properties of End-Grafted, Ethy- lene Glycol-based Self-assembled Monolayers (SAMs). ACS Applied Materials and Interfaces, 2009. 1(5): p. 1105-1112.

2007 S. Lee, R. Heeb, N.V. Venkataraman, N.D. Spencer, Macro- scopic Tribological Testing of Alkanethiol Self-assembled Monolayers (SAMs): Pin-on-disk Tribometry with Elastomeric Sliding Contacts. Tribology Letters, 2007. 28(3): p. 229-239.

2006 S. Lee, M. Muller, R. Heeb, S. Zuercher, S. Tosatti, M. Heinrich, F. Amstad, S. Pechmann, N.D. Spencer, Self-healing behavior of a polyelectrolyte-based lubricant additive for aqueous lubrication of oxide materials. Tribology Letters, 2006. 24(3): p. 217-223.