TVE 15060 JUNI Examensarbete 30 hp Juni 2015

Combined Tungsten Disulfide and Graphene Low Friction Thin Film Synthesis and Characterization

Fredrik Johansson

Institutionen för teknikvetenskaper Department of Engineering Sciences Abstract Combined Tungsten Disulfide and Graphene Low Friction Thin Film; Synthesis and Characterization Fredrik Johansson

Teknisk- naturvetenskaplig fakultet UTH-enheten Tungsten disulfide is a proven material as a low friction solid coating. The material is well Besöksadress: characterized and has proven its capabilities the Ångströmlaboratoriet Lägerhyddsvägen 1 last century. Graphene is this centurys most Hus 4, Plan 0 promising material with electrical and mechanical properties. With it the 2D material revolution have Postadress: started. Box 536 751 21 Uppsala In this thesis I present a feasible way to sputter tungsten disulfide on graphene as a substrate with Telefon: little damage to the graphene from energetic 018 – 471 30 03 particles and a straight forward method to quantize

Telefax: the damage before and after deposition. Further I 018 – 471 30 00 investigate compositional changes in the sputtered films depending on processing pressure and how Hemsida: tungsten disulfide film thickness and the amount of http://www.teknat.uu.se/student graphene damage affects the materials low friction capabilities. It is shown that graphene is not a viable substrate for a low friction tungsten disulfide film and that tungsten disulfide is an excellent material for low friction coatings even down too a few nanometers and that the films behavior during load in the friction testing significantly depends on the processing pressure during sputtering.

Handledare: Docent Andreas Lindblad Ämnesgranskare: Professor Urban Wiklund Examinator: Nóra Mazzsi TVE 15060 JUNI Sammanfattning

I många maskiner är det oumbärligt att ha en låg friktion. Det är att två ytor kan röra sig mot varandra med mycket lite motstånd eller tänk dig hur det skulle vara med en motor utan olja. För att erhålla denna låga friktion finns det flera olika sätt, som sades tidigare, används ofta olja som smörjmedel. Men i alla tillämpningar kan inte flytande smörjmedel användas utan ett fast material som ger låg friktion måste användas. En typ av material som används i sådana sammanhang är material som är strukturerade i lager. Dessa material sitter starkt ihop i lagret men de är lätta att få att glida på varandra. Två exempel på dessa material är wolframdisulfid och grafit. Om du har endast ett lager av grafit kallas det grafen (alltså endast ett lager av kolatomer), det är ett material med många framtida användnings områden. I det här arbetet kommer en film av kombinerad grafen och wolframdisulfid att undersökas. Den första svårigheten med detta material är hur det kan tillverkas. En vanlig teknik för att göra filmer av wolframdisulfid är att använda så kallad katodförstoftning, som ofta, och även framgent i detta arbete kallas sputtring (från engelskans sputtering). Sputtring är en teknik där joner kastas mot ett block av det önskade beläggningsmaterialet. När jonerna träffar blocket slås atomer ut och de kommer i sin tur att träffa ytan som man vill belägga. Dessa atomer kommer sedan att bygga upp filmen. Atomerna som färdas i kammaren kommer att ändra riktning varje gång de träffar andra atomer. Om detta sker i luft kommer antalet kollisioner vara så många att inga atomer når ytan, därför sker allt i en vakuumkammare. I detta arbete används en yta av kiseldioxid med ett lager grafen på som beläggningsyta. Precis som det låter så kan ett lager grafen enkelt skadas om atomerna som träffar ytan har för hög energi, det är ju bara ett lager atomer. I arbetet visas att tillföra lagom lite gas i kammaren gör att atomerna bromsas så mycket så att de träffar grafenet med så lite energi att det inte förstörs. I arbetet undersöks hur olika tjocklekar på wolframdisulfiden påverkar friktionen samt vilken effekt ett helt eller trasigt grafenskikt har. Det visas att grafen är ett dåligt material att belägga med wolframdisulfid. Det visas också att filmerna med trasigt grafen har lägre friktion än de där grafenet fortfarande är helt. Detta beror på att den extra gasen som har använts för att skydda grafenet vid beläggningen gör att wolframdisulfid skiktet blir poröst och lätt går sönder. Det visas också att om låg friktion endast behövs ett kort tag, till exempel om man ska föra in något i ett trångt utrymme, så räcker det att belägga med en film som bara är några nanometer tjock, som jämförelse är ett hårstrå ungefär 100 000 nanometer tjockt.

iii Contents

Abstract i

Sammanfattning ii

1 Introduction 2 1.1 Background ...... 2 1.2 Aim of the Thesis ...... 2

2 Theory 3 2.1 Tungsten Disulfide ...... 3 2.2 Graphene ...... 4 2.2.1 Transfer of Graphene ...... 5 2.3 Sputtering ...... 6 2.3.1 Sputter Damage from Energetic Particles ...... 7 2.3.2 Sputtering with Graphene As a Substrate ...... 8 2.4 Tribology ...... 8 2.4.1 Friction ...... 9 2.4.2 Friction in Layered Materials ...... 10 2.5 Analysis Methods ...... 11 2.5.1 Raman Spectroscopy ...... 11 2.5.2 Scanning Electron Microscopy ...... 14 2.5.3 X-ray Photoelectron Spectroscopy ...... 16 2.5.4 Atomic Force Microscopy ...... 17

3 Experimental 19 3.1 Chemical Vapor Deposition of Graphene ...... 19 3.2 Transfer of Graphene ...... 20

3.3 Sputtering of WS2 Films ...... 20 3.4 Analysis ...... 21

4 Results and Discussion 23 4.1 Synthesis of Graphene and the First Results ...... 23

4.2 Characterization of Deposited WS2 Films ...... 25 4.3 Tribology Testing ...... 30 4.4 Characterization of the Wear Tracks ...... 32

5 Conclusions and Outlook 37

iii 6 Acknowledgements 39

Appendix A SEM-images of films A-F as deposited 46

Appendix B XPS surveys spectra of films A-F 47

Appendix C SEM-images of the wear tracks 48

Appendix D SEM-images of the brittle cracks and crumbling 49

Appendix E AFM-images of samples A-G 50

iv List of Figures

2.1 Molecular structure of WS2 ...... 3 2.2 Cross section and top-view of WS2 films ...... 4 2.3 The structure of graphene ...... 5 2.4 The principle of sputtering ...... 6 2.5 The energy distribution of sputtered particle ...... 7 2.6 Schematic representation of the plowing component of friction 10 2.7 The principle for layered solid lubricants ...... 11 2.8 Raman spectrum of single layer graphene ...... 12

2.9 The refractive indices of Si and SiO2 ...... 13 2.10 The wave path through the stack of graphene, SiO2 and Si . . 14 2.11 The calculated contrast of graphene on SiO2 ...... 15 2.12 The electron beam trajectory in the SEM and the excitation volume ...... 16 2.13 The photoelectric effect ...... 17 2.14 XPS spectra, a survey spectrum and a peak fitted, resolved, spectrum of the W 4f region ...... 17 2.15 The principle of the AFM ...... 18

3.1 The homebuilt CVD furnace for growth of graphene . . . . . 19 3.2 The process of transfer of graphene ...... 20 3.3 The pin-on-disk friction testing equipment ...... 22

4.1 Characterization of the first batch of graphene ...... 24 4.2 Contrast simulation of three-layer graphene ...... 24 4.3 Effect on graphene from sputtering ...... 25 4.4 SEM image and EDX result of the films as deposited . . . . . 26 4.5 XPS spectra of the W 4f peaks from samples A through F . . 27 4.6 XPS spectra of the S 2p peaks from samples A through F . . 27 4.7 XPS spectra from the films as deposited ...... 28 4.8 AFM images of the films as deposited ...... 28 4.9 Results from the pin-on-disk friction testing ...... 30

4.10 Raman spectroscopy maps of WS2 and silicon from sample H 31 4.11 SEM images of cracked and crumbled film ...... 32 4.12 SEM images of the wear tracks after friction testing . . . . . 33 4.13 Raman spectroscopy maps of silicone in the wear tracks . . . 34 4.14 Raman spectroscopy maps of the steel balls from the friction testing ...... 35

v 4.15 Resonance of hematite in wear track of sample G ...... 35 4.16 Pyrite and hematite on the ball from sample C ...... 36 4.17 Damaged graphene after friction testing ...... 36

1 1. Introduction

1.1 Background

Both graphene and tungsten disulfide (WS2) are two interesting and well- studied materials, they have applications in everything from electronics to low friction coatings [1, 2, 3, 4, 5]. These materials are commonly used individually, but they have not yet been tried as a combined thin film. A similar material combination, where molybdenum disulfide (MoS2) have been grown by van der Waals epitaxy with CVD (chemical vapor deposition) onto graphene have shown good results [6]. This leads to an interesting question if sputtering, with graphene as a substrate, can yield similar growth. Sputtering with graphene as a substrate has been quite successful for thin metal and metal-oxide films but it has not been done with metal-sulfides [7, 8].

Since both WS2 and graphene has shown promise as low friction coatings the combined film of the materials is interesting to study.

1.2 Aim of the Thesis

The aim for this study is to investigate the synthesis process and properties of a sputtered WS2 film on a graphene substrate. The study is divided into two parts. The first part is an investigation on how WS2 can be sputtered onto a graphene substrate and how the sputtered WS2 film affects the graphene. The second part was the characterization and tribological measurements of the synthesized thin films.

Investigation info the possible van der Waals epitaxy growth of WS2 on the graphene substrate is left out this study, as it requires an extensive transmission electron microscopy study.

2 2. Theory

2.1 Tungsten Disulfide

Tungsten disulfide (WS2) (Tungstenite in its natural mineral form) is a material in the transition metal dichalcogenides (TMDs) family. TMDs have the general formula MeX2 where Me is a transition metal and X is elements in the chalcogen group.

WS2 is a layered material with sheets consisting of a layer of tungsten atoms sandwiched with 2 layers of sulfur atoms in a trigonal prismatic co- ordination as seen in figure 2.1. The sheets are held together with van de Waals forces giving the sheets the possibility to glide on each other, similar to the well known smearing of graphite.

Figure 2.1: The molecule structure of WS2 showing the trigonal prismatic coordi- nation as well as the sandwiched layered structure of the material.

There are many known applications for WS2, these include, as in this thesis, as a low friction coating. Other applications are as thin film solar cells, transistors (as a heterostructure with graphene) and as a catalyst for hydrogen evolution in water splitting. [9, 10, 11]

The sheet of WS2 can grow in different orientations when sputtered and it is known to always nucleate in the {001} planes parallel to the substrate [12].

3 Even though the first layer nucleate parallel to the substrate the preferred direction of the sheets changes after a few layers from [001] to [100] which is perpendicular to the substrate and lead to an almost amorphous structure for the further growth which can be seen in figure 2.2a and b [12]. This effect can be altered if the substrate is heated during deposition, then the structure changes to a dense film without noticeable gaps between the sheets also seen in figure 2.2c and d. [13]

Figure 2.2: a) and b) show a WS2 film deposited at room temperature, a in cross- section and b) in top view, c) and d) show a WS2 film deposited with the substrate at 300 ℃ with c) incross-section and d) in top-view. Reprinted with permission from Ref. 10.

2.2 Graphene

Graphene is a two-dimensional monolayer of sp2 bonded carbon (as seen in figure 2.3) and possesses unique properties, both mechanical, electrical, chemical and thermal. [14, 15, 16, 17] The material have attracted huge interest ever since Geim and Novoselov proved its existence in 2004 for which they were awarded the Nobel prize in Physics 4 years later. There are several ways to produce graphene where growth by chemical vapor deposition (CVD) and mechanical exfoliation from graphite are the most common. Mechanical exfoliation of graphene was the first way to produce graphene and it was done by repeated peeling of the material layer by layer using tape or similar adhesive to the point were only one layer of the material remains [15]. Other mechanical exfoliation processes include the use of blending graphite in a blender (in principle similar to a normal household blender). This technique is performed in a liquid, which results in graphene flakes in a solution. This graphene is suitable for using in graphene

4 Figure 2.3: Shows the hexagonal structured carbon atoms of monolayer graphene. composites and coatings. [18] In the CVD growth process the substrate is heated under vacuum to high temperature (∼1000 ℃) under flow of hydrogen. This to increase the grain size of the copper foil from a few µm to ∼100 µm, which will increase the grain size of the graphene [19, 20]. After the recrystallization of copper, methane is introduced in the chamber and the graphene start to crystal- lize. The typical growth consists of randomly nucleated crystal sites on the Cu-surface, which then continue to grow and after full growth for a full polycrystalline film [21]. The process is self-terminating after the monolayer formation, this is attributed to the low solubility of carbon in copper so precipitations of carbon does not form during the cooling process (which in other systems, for example with nickel substrates, leads to formation of multilayer graphene films) [19]. The mechanical exfoliation process yields larger grain sizes compared to CVD-graphene but the CVD process yields much larger areas of grown graphene. The typical size for exfoliated graphene is up to ∼50 µm but for CVD-graphene a roll to roll process have been shown that produces graphene sheets of 30 inches. [19, 20] This leads to that graphene from different manufacturing techniques is suitable for different applications. In this work the chosen manufacturing method is the CVD-grown graphene, for its versatility in sizes and consis- tency in quality, giving samples that are easily compared.

2.2.1 Transfer of Graphene Mechanically exfoliated graphene can be directly placed or dripped (if in solution) onto the desired substrate whereas CVD-grown graphene is, with its growth on a metal substrate, not directly transferable to the desired sub- strate. As a solution to this problem a straight forward transfer process of

5 CVD grown graphene has been developed which can be adjusted for almost all types of substrates [22, 23, 24]. The process consists of adding a sup- porting material on top of the graphene, wet-etching of the Cu-substrate, direct transfer to the desired substrate and removal of the supporting mate- rial leaving a graphene sheet on the substrate [25]. The process used in this thesis is described more in detail in section 3.2.

2.3 Sputtering

Sputtering is a form of physical vapor deposition in which energetic particles (normally argon ions) are accelerated towards a target (composed of the material that is to be deposited) from which atoms are ejected. [26] These atoms are inelastically scattered and they are ejected from the target surface as seen in 2.4. The sputtered target atoms hit the substrate surface with an energy that depends on the initial ejection energy and the number of scattering event in the fly path. Sputtering is carried out in a vacuum chamber at processing pressures typically between 0.1-10 Pa. The mean free path of the sputtered atoms is a function of atom size and pressure. As an example the mean free path of an oxygen atom at atmospheric pressure is less than 10-6 m and in vacuum (10-5 Torr) it is ∼4.5 m [27]. In the pressure range normal for sputtering, the mean free path is from a few mm to several cm [28].

Target material

Energetic argon ion +

Plasma Sputtered target atoms

Coating

Substrate

Figure 2.4: The principle of sputtering showing the energetic argon ion that scatters the target atoms, which forms the coating on the substrate.

The ionization of the argon gas is achieved through a discharge plasma and it is sustained by the ejected secondary electrons, created from the ions

6 hitting the target, which may ionize argon and thereby keep the plasma process ongoing. The acceleration of the ions towards the target is achieved with a potential on the target, the so-called target potential that typically is several hundred volts. The sputtered atoms will have an initial energy as the distribution that can be seen in figure 2.5 where the mode (the value for maximal probability density in a distribution, e.g. the mean value for a Gaussian distribution) is roughly half of the surface energy barrier [26] (∼ 11.75 eV for W [29]). The mode does not change significantly when the energy of the normal incident ions is increased, but at the same time will the mean energy of the ions increase [30], i.e. the most particles will still have the energy of the mode but the high-energy tail will increase. Number of particles Number Mode

High energy tail

Energy

Figure 2.5: The energy distribution of sputtered particles with the mode and the tail of high-energy particles denoted.

2.3.1 Sputter Damage from Energetic Particles There are 3 types of energetic particles that can damage the substrate, i.e. introducing defects in the forming film or the substrate. These are:

• The high-energy tail of sputtered particles

• Reflected argon

• Negative ions The high energy tail of sputtered particles comes from the energy distribution of the sputtered atoms as seen in figure 2.5 where a number of the sputtered atoms will have an energy that is much higher than for the

7 majority of the atoms. These atoms are not so many and they can thereby quite easily be thermalized, i.e. their energy can be significantly reduced, to the same energy level as the rest of the particles. Reflected argon is argon ions that when they hit the target loses their charge and are scattered in all directions. This will happen to some of the argon ions and the energy of these high, up to several hundred eV, they may only loose a little energy in the reflection event. The reflected argon is not always an unwelcome phenomena as the energy contribution from these can help the sputtered atoms in the growing film to find stable positions i.e. the same effect as heating the substrate. Negative ions are ionized atoms in the chamber, which just like the argon ions, are accelerated by the target potential. Since the negative ions have a negative charge they will be accelerated towards the substrate and not the target (as the argon ions will). They are accelerated with the full target potential (i.e. an energy of ∼300eV) which means that these ions are the most energetic particles that will hit the substrate. The effect of all these energetic particles can be significantly reduced with an increase in processing pressure, either by introduction of more argon gas or a reduction of the pumping of the chamber. With a higher processing pressure, the mean free path of the particles will be shorter and with several scattering events on their path towards the substrate they will loose most of their kinetic energy, i.e. be thermalized.

2.3.2 Sputtering with Graphene As a Substrate As stated in section 2.3.1 sensitive substrates can be damaged by energetic particles in the chamber having energies in the range from a few to sev- eral hundred eV. Graphene is a sensitive substrate that will have defects introduced if the incoming energy is too high. For graphene defects to be introduced, an energy of ∼33 eV [31] is required, meaning that negative ions and reflected argon may damage the graphene. To mitigate this problem, the pressure in the sputter chamber can be increased as shown in [8, 7]. Graphene is interesting as a substrate material since it has been shown that graphene can be used for van der Waals epitaxy growth [6] of graphene analogue materials such as MoS2 and WS2 [32].

2.4 Tribology

Tribology is the science of friction, wear and lubrication, i.e. the interac- tion of surfaces in relative motion. These are aspects that everyone takes into consideration in everyday life but normally doesn’t reflect much about. These everyday things include how the car wheels stick to road while turn- ing, why does the sole of shoes wear off, why do dresser doors squeal etc. When addressing tribology in a more of an engineering perspective it is

8 something that must be taken in account in almost all applications. You can say that tribology is the reason why every moving part eventually stops. This implies that tribology is an important factor in all mechanical design and that the field should be investigated thoroughly.

2.4.1 Friction Friction is the part of tribology that will be investigated in this thesis. Fric- tion is defined as how much force is needed to move an object across the surface and it is normally denoted with the friction coefficient, µ, which is the fraction of the tangential force (FT ) and the normal force FN as in eq 2.1. [33]

F µ = T (2.1) FN This is the case despite the area of the contact surface. Which is one of the classic laws of friction formulated by Amontons in 1699 and modernized by Coulomb in 1785:

The frictional force is proportional to the normal force, i.e. the friction coefficient is independent of load

The coefficient of friction is independent of the nominal contact surface

The coefficient of friction is independent of the sliding speed

These laws implies one more law

The coefficient of friction is independent of the surface topography

This comes from the fact that the friction will come from the parts of the surfaces that actually are in contact. This area is much smaller than the total surface area, which is intuitive when you visualize the nanostructure of the any surface. As said by F.P. Bowden:

“Putting two solids together is rather like turning Switzerland upside down and standing it on Austria - the area of intimate contact will be small”

I.e. the total load will be carried on a number of small contact spots (ai, fi is the force on the respective contact spot) and the real contact surface (Ar) and normal force can be written as in equation 2.2

n n X X Ar = ai FN = fi (2.2) i=1 i=1

9 The real contact surface (Ar) between surfaces can actually be estimated with the normal force and the hardness of the softest material (H) as in equation 2.3.

F A = N (2.3) r H There are two components to coefficient of friction, the adhesion com- ponent (µa) and the plowing component (µp). The adhesion component is dependent on how well a material withstands shear, the force needed to shear a material is the product of the shear stress (τy) and the real con- tact surface. This with equation 2.3 in equation 2.1 leads to the adhesion component as in equation 2.4

τ A τ µ = y r = y (2.4) HAr H This would be the whole friction coefficient if the surfaces were close to perfect but if one of the surfaces has structure, e.g. that one of the surfaces has the form of a ball, then the plowing component will have an effect. The size of the plowing component is dependent on the size of the ball and the penetration depth, i.e. a dependence on applied force as seen in figure 2.6 and equation 2.5.

F T R d F N θ

d

Figure 2.6: A schematic representation of the plowing component with a hard ball on a surface.

4R2 µ = 2θ − sin 2θ
(2.5) p πd2

2.4.2 Friction in Layered Materials One common method to reduce friction is the use of lubricants. This is an addition of material, which can be both liquid or solid, to reduce the

10 shear stress τy of the contact surfaces i.e. the added material is the material which is sheared as can be seen in figure 2.7 which is quite different from the plowing principle in figure 2.6. One common material as solid lubricant is graphite, which with its multilayer structure is easy shear. WS2 is also a layered material and possesses the same qualities as graphite as well as some other properties, e.g. that WS2 can easily be coated to surfaces by sputtering or other coating methods. With the use of solid materials for low friction applications some ma- terials have special abilities, one of these is the formation of tribofilms in the area of material contact. This is the formation of material with desired properties in the contact area. For WS2 this is the formation of a few layers of pristine and crystalline WS2 even if the surface is under-stoichiometric. This arises from the extreme temperatures and pressures in the local contact spots. This tribofilm can also transfer to the other surface in the contact, that film is then called a transfer film. In the case of WS2 films, the transfer film can also consist of crystalline, layered WS2 e.g. in [1] and references therein.

Figure 2.7: The principle for layered solid lubricants with two surfaces moving relative to each other.

2.5 Analysis Methods

2.5.1 Raman Spectroscopy

In Raman spectroscopy the inelastic scattering of light in matter is measured and gives information on the atomic structure of the material. The material is illuminated with monochromatic light and when a photon interacts with an atom it will excite it to a virtual energy state. When the atom relaxes it will emit a photon with the energy of the difference form the virtual state and the ground state [34]. The energy difference between the emitted photon and the incoming photon is called the Raman shift and it is normally presented in reciprocal centimeters (cm-1). The analysis is non-destructive, often requires no sample preparation, can be performed at ambient conditions and is simple to preform [35]. This leads to that Raman spectroscopy is a widely used method for characterization, especially of graphene.

11 2.5.1.1 The Raman Spectrum of Graphene The part of the Raman spectrum of graphene commonly used to characterize the material consists of three distinct peaks as seen in figure 2.8. The G peak is associated to the normal first order Raman scattering process from the carbon atoms in the graphene lattice, the D and 2D peaks originates from a double resonance processes in D between a defect and a phonon and in 2D between two phonons [36]. The G and 2D peaks will thereby be associated to the sp2 oriented carbon in the graphene and the D peak will be relative to the number of defects in the graphene lattice. These attributes lead to easy distinction between mono-, bi- and few-layer graphene [37], as well as an easy measurement of defects concentration and thereby damages of the graphene. A common way to quantify the Raman spectrum of graphene is to use the intensity ratios 2D/G and D/G were the 2D/G relation is a measure on how many layers and how graphene-like the sample is (for monolayer graphene this ratio is normally >2) and the D/G ratio should be as low as possible [7].

2D

G

Intensity a.u. From substrate

D

1400 1600 1800 2000 2200 2400 2600 Raman shift [cm -1 ]

Figure 2.8: Raman spectrum of single layer CVD graphene transferred to Si/SiO2 substrate, with the distinct G and 2D peaks and the defect induced D peak.

2.5.1.2 The Substrate Effect on Graphene Monolayer graphene is, as stated above, very thin. And a sheet of graphene is intuitively not visible to the naked eye or in an optical microscope. But graphene is actually a perfect gray-filter [38] which leads one to assume that if the contrast is right, then graphene could be visible to the naked eye.

When graphene is placed onto SiO2 (as seen in figure 2.11) an optical effect occurs and the contrast between the graphene and the surface increases. This is due to the difference in indices of refraction (ni = n + ik where n

12 is the index of refraction and k is the extinction coefficient [39]) and phase shifts from the change in optical path due to thickness of the SiO2 layer and the wavelength of the incoming photons. The change in contrast as a function of wavelength of the incoming light and the thickness of the SiO2 can be seen in seen in figure 2.11a. The contrast is calculated using equations 2.6 through 2.9. Figure 2.10 describes the wave path in equation 2.9 where the square of the sum of the reflected waves escaping the surface. [40]

The indices of refraction for both SiO2 (n2) [41] and Si (n3) [41] are dependent on the wavelength as shown in figure 2.9. SiO2 has only a real part and it is fitted from experimental data with a 2 term power model whereas silicone has both a real and a complex part which both are interpolated from experimental data. The index of refraction for graphene is assumed as the same as for graphite, which is supported by experimental data [40]. The index of refraction for graphite is also wavelength dependent in the infrared range but in the visual range it can be fitted to 2.6 − 1.3i (n1) [42]. In figure 2.11a the calculated contrast can be seen as a function of SiO2 thickness and wavelength, with the contrast as the colormap.

7 1.54 6 n Si

5 1.52 ) 2 4 1.5 3 k Si Index (Si) Index Index (SiO Index 2 1.48 n SiO 1 2 1.46

250 300 350 400 450 500 550 600 650 700 750 800 λ [nm]

Figure 2.9: Fitted experimental values of the refractive indices of silicon and silicone dioxide used in the contrast calculations for both n and k, observe the two different y-axes

 i(Φ1+Φ2) −i(Φ1−Φ2) −i(Φ1+Φ2) I(n1) = r1e + r2e + r3e

  i(Φ1−Φ2) i(Φ1+Φ2) −i(Φ1−Φ2) + r1r2r3e × e + r1r2e (2.6) −1 2 −i(Φ1+Φ2 i(Φ1−Φ2) + r1r3e + r2r3e

where

13 Air r t t’ r t t’ t t’ r t t t t’ r r r n 1 1 1 2 1 1 2 2 3 1 1 2 2 1 2 3 0

-t r’ r -t t t’ r r 1 1 2 1 2 2 1 3 t Graphene 1 t r t t t’ r -t t t’ r r r n 1 2 1 2 2 3 1 2 2 1 2 3 1

-t t r r t t r r r r 1 2 2 3 1 2 2 2 3 3

SiO2 t t t t r -t t r r r n 1 2 1 2 3 1 2 2 3 3 2

Si t t t 1 2 3 n 3

Figure 2.10: The wave path through the stack of graphene, SiO2 and Si showing both transmitted and reflected waves as well as second reflections in orange.

n0 − n1 n1 − n2 n2 − n3 r1 = , r2 = , r3 = (2.7) n0 + n1 n1 + n2 n2 + n3 are the relative indices of refraction. And

Φ1 = 2πn1d1/λ Φ2 = 2πn2d2/λ (2.8)

are the phase shifts from the change in optical path due to thickness and indices of refraction. The contrast (C, equation 2.9) is defined as the relative intensity of the reflected light in the presence (n1 6= 1) and absence (n1=n0=1, i.e. air) of graphene.

I(n = 1) − I(n ) C = 1 1 (2.9) I(n1 = 1)

2.5.2 Scanning Electron Microscopy As a part of this project films with thickness of ∼5 nm are analyzed and to image them microscopy is used. The first choice in microscopy is to use traditional light optical microscopy (LOM) where reflected of transmitted light in the visual range is used to image the sample. LOM is quick and

14 A B

Air: n 0=1

Graphene: n1=2.6-1.3i d1=0.34nm

SiO 2: n2(λ) d2

Si: : n3(λ)

Figure 2.11: a shows the contrast of graphene on SiO2 as a function of wavelength of the incoming radiation and the thickness of the SiO2-layer with the contrast as a colorbar and b shows the model of the different material layers used in the calculations.

cheap but has a resolution limit of approximately 1 µm and that value is related to the wavelength of the light. Scanning electron microscopy (SEM) uses electrons instead of light to form the image and since electron have a wavelength several magnitudes smaller than light and the SEM thereby has much better resolution than LOM. In the SEM electrons are emitted from an electron source and accelerated and focused through a series of electromagnetic lenses in what is called the electron column as can be seen in figure 2.12. The electron hits surface atoms on the sample and interacts with the same. In this interaction a variety of signals are produced from different depths of the sample, the so- called interaction volume (as seen in figure 2.12). These can be used for information regarding the topography and/or composition of the sample surface. One common mode for imaging is the detection of secondary electrons. During the electron-atom interaction inelasticity scattered electrons are ejected from the sample. These electrons are called secondary electrons and have very low kinetic energy (∼50 eV), which correlates to an escape depth of a few nanometers. The escape depth, i.e. a function of the electrons mean free path, is different for different materials and follows the so called uni- versal curve [43], this gives the contrast in the secondary electron image. The number of ejected secondary electrons depends on the surface, its angle towards the electron detector and the energy of the incoming electron beam, thus giving a 3D-effect in the formed image. The secondary electron thereby makes topological contrast, which can be intuitively interpreted. From deeper in the interaction volume characteristic X-rays are emitted from the electron atom interaction. As the name implies, the X-ray energy is characteristic for the atom that emits the X-ray making it possible to perform energy dispersive X-ray spectroscopy (EDX) in the SEM. With EDX analysis it is possible to make both qualitative and quantitative elemental

15 Electron gun

Anodes

Primary electron beam 1st Condeser lens

Secondary electrons 2nd Condeser lens Backscattered electrons Characteraistic X-ray

Objective lens

Scanning coils

Sample

Figure 2.12: Left: the electron column and the main parts of the SEM and the trajectory of the electron beam. Right: the interaction volume from the electron surface interaction. analysis of the sample.

2.5.3 X-ray Photoelectron Spectroscopy X-ray photoelectron spectroscopy (XPS) is an analytical method devel- oped at Uppsala University by Kai Siegbahn during the 1950’s (Siegbahn referred to the technique as ESCA, Electron Spectroscopy for Elemental Analysis)[44]. As described by Verma [45] XPS is based on the photoelectric effect where X-ray photons interacts with core level electrons of an atom and a photoelectron is emitted (as seen in figure 2.13). The photoelectron has an kinetic energy (Ek) corresponding to the difference of the X-ray photon energy (hν) and the binding energy of the electron to the atomic nucleus (Eb) with a correction for the work function (EW ) which is the energy difference between the Fermi level and the vacuum level. The analysis is performed by measuring the electrons kinetic energy and subsequently calculating the electrons binding energy by using equation 2.10.

Eb = hν − (Ek + EW ) (2.10) Since all electrons has a specific binding energy, which also differs with

16 Photoelectron

ΕV

ΕV

ΕW

ΕF Valance Photon (hν) band

ΕB

Core electron levels

Figure 2.13: The ejection of photoelectrons from an atom, i.e. the photoelectric effect.

the chemical environment, a spectrum of Eb and the intensity count will give information on the composition of the sample as well as the oxidation state and ligands of the atom (as seen in figure 2.14).

W 4f O 1s W in WS x 8000 W in WO 3

6000 W 4f 42 40 38 36 34 32 30 Binding energy [eV]

Counts 4000 S 2p 2000

1400 1200 1000 800 600 400 200 0

Eb [eV]

Figure 2.14: XPS spectra from the 100 nm (75 mTorr chamber pressure) WS2 film, the O 1s, S 2p and W 4f peaks are denoted. Inset shows the high resolved spectra of the W 4f peaks of the same sample showing a curve fit to resolve the sulfide bonded tungsten (lower Eb) and the oxygen bonded tungsten (higher Eb.)

2.5.4 Atomic Force Microscopy Atomic force microscopy (AFM) [46] is a scanning probe microscopy tech- nique where a small tip is moved across the surface and a topographical

17 image with atomic scale resolution is acquired. There are a lot more modes other than the topographical in which the AFM can image and measure but for this thesis only the topological is described.[47] The actually topography is measured by focusing a laser on the can- tilever tip and when the cantilever is bent from the surface of the sample the reflected laser spot will move on a photo analyzer and the movement can then be related to the exact movement of the cantilever, as can be seen in figure 2.15a.

A B Photodetector

Contact mode Laser Repulsion

0 nN

Non-contact mode Attraction Distance

Figure 2.15: In a, the principle of the cantilever in AFM with a light deflection detector. In b, the difference of contact and non-contact mode as regard to forces and distance from the sample surface.

There are two main techniques the AFM operates in, contact and non- contact mode. In the contact mode the tip is touching the surface and moves with the topography by the repulsive force from the surface (as can be seen in the force/deflection diagram in figure 2.15b, this effects the surface since there is mechanical contact between the tip and the surface. In the non- contact mode the tip is held close to the sample surface and when moved across the sample the attractive force (as seen in figure 2.15b) between the tip and the surface will bend the cantilever and this can be related to the topography.

18 3. Experimental

3.1 Chemical Vapor Deposition of Graphene

The CVD growth of graphene was performed in a homebuilt tubular vacuum furnace as seen in figure 3.1. First a 45·75 mm piece of 25 µm thick copper foil (Alfa Aesar, 99.8%) was rolled to a tube and placed in a quartz tube (with a diameter slightly smaller than the quartz tube of the furnace) and placed in the furnace tube. The furnace was subsequently pumped to 20 Pa ◦ and heated to 1000 C under a constant flow of 20 sccm of H2 and amounts of argon, argon is used to stabilize the pressure in the furnace at the desired pressure. The copper foil was then annealed for 30 min (after approximately 15 min the copper foil recrystallizes and form large grains, larger grains will lead to fewer grain boundaries to interfere with the graphene growth). After 30 min the flow of H2 was reduced to 5 sccm and methane (Air Liquide) was introduced at a flow of 30 sccm to a total pressure of 50 Pa to start the growth of the graphene. After the 90 min the gas-flows and the heating was turn off and the furnace was allowed to cool to room temperature. The copper foil with the graphene was then stored in a sealed container not to contaminate it before the transfer.

Figure 3.1: The homebuilt CVD furnace for the growth of graphene showing the quartz tube in which the growth takes place as well as the heating and insulating elements.

19 3.2 Transfer of Graphene

After the growth of the graphene (see section 3.1) the copper foil with the graphene is spin-coated with a thin layer of PMMA-resist (poly methyl methacrylate, Micro Resist Technology) to act as a support structure for the graphene. The coated foil is then cut into the wanted size and placed

floating on the surface of 3M solution of FeCl3 to etch away the copper foil. When the foil is completely etched away, as seen in figure 3.2 (approximately after 15 min) the graphene/PMMA stack is removed from the solution (us- ing a plastic spatula) and washed in deionized water three times. The stack is then placed on the Si/SiO2 substrate and dried for more than 10 min on a hot-plate at 40 oC to remove all the water from under the stack. Subse- quently the hot-plate is turned up to 150 oC and the stack is left to heat treat for 30 min. The substrate is then cooled to room temperature and as the final step the PMMA is carefully dissolved in an acetone bath for 3 times, 10 min each time, and thoroughly blow-dried with N2 between the baths. This leaves large-area monolayer graphene with few defects and homo- geneous quality through the entire film.

Figure 3.2: The process of the transfer of graphene, showing the graphene/PMMA- stack floating in 3 M FeCl3-solution and the different washing steps

3.3 Sputtering of WS2 Films

For the first part of the investigation, several graphene samples were sput- tered with tantalum to identify the sputter damage of the graphene. Tan- talum was chosen since it is an element that is very similar to tungsten and neither tungsten nor WS2 were available for sputtering at the moment. Three different thicknesses of WS2 deposited at two different pressures were chosen to be investigated further for the low friction application. These can be seen in table 3.1.

20 Table 3.1: The sample denominations with deposition pressure and film thickness for each sample.

Sample Deposition pressure [mTorr] Film thickness [nm] A 2.3 5 B 2.3 50 C 2.3 100 D 75 5 E 75 50 F 75 100

G CVD-graphene (no WS2) 0.34 H 2.3 (not on graphene) 111 I 75 (not on graphene) 255

The different materials were deposited in two different sputtering systems and all sputtering was performed by Dr. Tomas Nyberg (department of Solid State Electronics, Angstr¨omLaboratory).˚ The tantalum on graphene experiments were performed in a cryo-pumped custom-built stainless steel vacuum chamber with cylindrical dimensions (height 0.3 m x diameter 0.3 m) equipped with a 2 inch magnetron powered with a constant power of 60 W from an RF power supply (Cesar 1312 from Dressler). The Ar flow was set to 20 sccm. The base pressure (< 5 · 105 Pa) was measured with a hot cathode ionization gauge, while the process- ing pressure was measured with a capacitance gauge. All depositions were carried out on floating substrates with a target to substrate distance of 10 cm.

The WS2 coatings were deposited by non-reactive pulsed DC magnetron sputtering, using a von Ardenne system fitted with three 4-inch magnetrons and a substrate heater. A sintered WS2 target (99.9%) was used. The Ar flow was set to 20 sccm and the target power was 200 W. The sample stage was rotated during the deposition, and no bias was applied to the substrates, which were kept at floating potential. The sputtering time, to reach the desired thicknesses of the films, were determined by extrapolation from depositions at the different processing pressures, the thicknesses were determined by needle profilometry.

3.4 Analysis

Raman spectroscopy was performed using a Renishaw in-Via microscope at 5x and 20x magnification with 532, 633 and 785 nm lasers and gratings of 1800 and 1200 cm−1 both in static and extended modes. For the mapping of the wear tracks and the balls from the friction testing the streamline

21 Figure 3.3: The pin-on-disk testing equipment used for the friction testing. mapping mode was used. The maps were constructed in the Renishaw Wire software (versions 3.4 and 4.1) where the maps are a representation on the

fit between the measured spectra (in each point) and WS2 and Si reference spectra. Scanning electron microscopy (SEM) was performed with a Zeiss 1550 Leo microscope using an inlens detector, working distance of 2-6 mm and an acceleration voltage of 3-5 kV. For energy dispersive X-ray spectroscopy (EDX) the same microscope was used with an 80 mm2 Oxford instrument silicon drift detector, 8mm working distance and 10-20 kV acceleration volt- age. X-ray photoelectron spectroscopy was performed on a PHI Quantum 2000 instrument using an Al-Kα scanning radiation source (1486.7 eV). The samples were not cleaned by sputtering before the analysis since sputtering is known to induce damage to WS2 films [13]. The survey spectra are identified with use the Multipak software and curve fits are performed in IGOR pro (all peak fits were performed by Docent Andreas Lindblad, department of physics and astronomy, Angstr¨omLaboratory).˚ For the atomic force microscopy a Bruker Multimode 8 with Scanasyst in air using peak force error tapping mode. The tribological testing was performed with a commercial pin-on-disk system with a rotating disk sliding against a stationary ball as can be seen in figure 3.3. The tests were conducted for 1000 revolutions with an applied force of 5 N and a sliding speed of 0.3 m/s. The balls used in the experiment were ball bearing steel balls with a diameter of 6 mm.

22 4. Results and Discussion

In this chapter the results from the synthesis and analysis will be presented.

4.1 Synthesis of Graphene and the First Results

The first batch of graphene samples were synthesized and transferred by me. As can be seen in figure 4.1a there where only strips of what was, with Raman spectroscopy, characterized as single layer graphene (as seen in 4.1b), the rest of the surface had a Raman spectra as in 4.1b that possibly could be from a carbon containing compound, but not with a characteristic spectrum. These were then coated with a 5 nm, sputtered, tantalum film. Metal films are normally not transparent for light, but for very thin film some light will actually reflect from the buried interface, in this case the graphene. This means that the damages to the graphene can be measured through the tantalum films. As seen in figure 4.1c were the D/G ratios from before and after sputtering almost the same at high pressures, this indicates that an increase in sputter pressure will minimize the damaging high energetic particles during the sputtering process. The D/G ratio before sputtering changes a bit between the samples and this can be attributed to the fact that it was not single layer graphene that was measured on as described later in the section. These results, although very promising, are but a mere effect of the substrate effect as described in section 2.5.1.2. The LOM image in figure 4.1a shows merely the contrast difference for mono- and few-layer graphene. Graphene, synthesized in the same CVD-reactor as the samples in 4.1a, then and transferred to a Si substrate with a known 300 nm SiO2 surface was then acquired from Mr. Patrik Ahlberg at department of Solid State Electronics (Angstr¨omLaboratory).˚ These samples were also characterized with Raman spectroscopy and they consist, as can be seen in 4.1e, of single layer graphene and that on the entire surface. Samples from Patrik Ahlberg were then used throughout this work, the growth and transfer processes for these are the same as described in sections 3.1 and 3.2. The areas of what was assumed at first to be single-layer graphene are attributed to be areas of three-layer graphene from creases that arises dur- ing the transfer of the graphene. Contrast simulations shows that positive contrast can arise from three layer graphene and no contrast from single layer graphene when the SiO2 layer is about 200 nm thick and a 532 nm laser is used, as can be seen in figure 4.2.

23 Figure 4.1: a) LOM image of the first batch of graphene. b) Raman spectra cor- responding to the highlighted areas in a). c) the ratio of D/G from before and after sputtering of 5 nm tantalum. d) Raman spectra of pristine graphene from the samples acquired from Patrik Ahlberg.

Figure 4.2: Contrast simulation of three-layer graphene, calculated as described in section 2.5.1.2 on page 12 but with a graphene thickness of 1.02 nm (i.e. the thickness of three layers of graphene).

24 4.2 Characterization of Deposited WS2 Films

Before and after deposition all the samples were characterized by Raman spectroscopy to identify how the graphene withheld to the sputtering process (samples E and F were not characterized before sputtering but are assumed to be of equal, consistent quality, as all other graphene samples, a total 22 samples were prepared in this sample series). The main parameters that were regarded were the D/G- and 2D/G ratios (as described in detail in section 2.5.1.1 on page 12), which can be seen in table 4.1. As seen in figure 4.3 the graphene layer were destroyed for sample A through C, where the D/G and 2D/G-ratios are incalculable, but for samples D through F the graphene suffered little damage. Samples E and F shows an increase in damage compared to sample D but this can come from a thickness effect. The laser must, in these films, penetrate a much thicker semi-transparent film compared to sample D. The D/G and 2D/G ratios seen in table 4.1 are the mean from calculations from an excess of 100 spectra, which were acquired over an area of the samples.

Figure 4.3: The effect on the graphene layer from the sputtering process. Show- ing the difference in destruction for different sputtering pressure and different film thicknesses, with the Raman spectrum from as deposited graphene as a reference, the intensity for the reference spectrum is divided by a factor 4 for an easier com- parisons.

The surfaces were also imaged using SEM to determine the coverage of the WS2 film. As can be seen in figure 4.4 a cauliflower-like structure builds up the film is visible in figure 4.4a, the 5 nm film deposited at a pressure of 2.3 mTorr. This can also be seen in samples D and E. The rest of the surfaces do not exhibit the cauliflower structure as distinct as in figure

25 Figure 4.4: a) shows the surface of sample A with the cauliflower-like structure, which builds up the film, visible. b) shows the surface of sample C with the orange line showing the line from the EDX line scan in c).

Table 4.1: The results from Raman spectroscopy, XPS and AFM. Showing the D/G and 2D/G ratios which corresponds to the damage and quality to the graphene (incalculable for samples A through C), the S/W ratio showing the stoichiometry of the films

Before deposition After deposition Sample D/G 2D/G D/G 2D/G A 0.052 2.19 -† -† B 0.093 2.24 -† -† C 0.062 2.18 -† -† D 0.052 2.15 0.20 2.17 E-‡ -‡ 0.46 2.61 F-‡ -‡ 0.44 1.92 † Incalculable since the graphene layer is destroyed ‡ Not measured, assumed to have the approximately the same D/G ratio as samples A through D

26 W 4f W 4f 2.3 mTorr 75 mTorr 100 nm 50 nm 100 nm

C F

W in WSx

0n 5 nm 50 nm W in WS W in WO3 x B E W in WO3 Intensity a.u Intensity a.u 5 nm A D

44 42 40 38 36 34 32 30 42 40 38 36 34 32 30 Binding energy [eV] Binding energy [eV]

Figure 4.5: The resolved spectra of W 4f with peak fits showing tungsten bonded both to sulfur and oxygen. The electron binding energy of W 4f is 33.6 and 31.4 eV.

S 2p 2.3 mTorr S 2p 75 mTorr S 0 m50 nm 100 nm 100 nm

C WS x F 0n 5 nm 50 nm B E Intensity a.u Intensity a.u 5 nm

A D

170 168 166 164 162 160 158 170 168 166 164 162 160 158 156 Binding energy [eV] Binding energy [eV]

Figure 4.6: The resolved spectra of S 2p with peak fits showing sulfur bonded to tungsten and as elemental sulfur. The electron binding energy of S 2p is 163.6 and 162.5 eV.

4.4a, but instead a lot of particles on the surface as seen in figure 4.4b and the AFM images in figure 4.8 (SEM images of all samples can be found in appendix A). These particles were analyzed using EDX, which indicates that the particles consists of tungsten and sulfur as seen in fig 4.4c (the line scan is from the orange line in 4.4b). Since the films are very thin, the signal from W and S were very small and most signal came from the substrate (the large signal from silicon). There are also some indications of tungsten away from the particles as well as sulfur. The difference in the cauliflower structure forming the film is a difference in densities of the films deposited at different processing pressures. During

27 O 1s W 4d C 1s W 4f 3/2 W 4p 1/2 O KLL W 4p W 4s S 2p F 1s N 1s Counts

A S 2s F KLL C KLL Si 2p Si Si 1s Si S LMM

D

1400 1200 1000 800 600 400 200 0 E [eV]

Figure 4.7: Survey spectra from samples A and D (i.e. the 5 nm films) showing all identified peaks as well as positions of the not visible Si 1s and 2p. Auger features are denoted by elements and the letters designating the transition.

Figure 4.8: AFM images of sample A and D showing the particles on the surfaces, these particles vary in size but many of them are in the region of 20-30nm as seen by the scale. A small hole can be found in one of the films, the hole is indicated by the red circle on film D. deposition of the films, the sputtered atoms at the lower processing pressure have a higher kinetic energy giving them a higher probability to find a low- energy position in the forming film. Whereas the sputtered atoms at the higher processing pressure does not have high enough energy to migrate on the surface. This can be seen in the SEM images in figure 4.4 and also in appendix A. The difference in the 5 nm films (samples A and D) is very

28 small but the difference is significant in thicker films.

To get the composition of the films surfaces an XPS analysis was per- formed (survey spectra from all samples can be found in appendix B). As can be seen in figure 4.5 and 4.6 the peak fitted spectra shows both tungsten and sulfur, and both of the bonded in several ways. Pure S 2p peaks are at 162.5 and 163.6 eV and pure W 4f are at 31.4 and 33.6 eV, peaks close to these values are so called shifts which correspond to different bonds e.g. ox- ide or sulfide bonds. For tungsten, the peaks at lower eV are corresponding to a sulfide bond whereas the peaks at higher energies correspond to oxide bonds [13] (which also is a possibility since the films have been exposed to oxygen). For the sulfur the peaks at higher eV corresponds to a metallic bond (i.e. WSx) and the peaks at lower eV are attributed to elemental sul- fur on the surface of the sample [48]. When comparing the energy difference between W with sulfide bond and oxygen bond at the different processing pressures in figure 4.5 the shift between the different contributions is de- creasing with higher pressure (at 2.3 mTorr the shift is 3.5 eV and at 75 mTorr it is 3.125 eV), this shift is attributed to more under-stoichiometric

WO3 at higher processing pressures. This leads to the conclusion that the surface of the films consists of WS2 and both elemental sulfur and WO3-x where an increase in processing pressure leads to under stoichiometric WO3. The literature shows increasingly stoichiometric WS2 at higher processing pressures. [49, 12]

The spectra in figure 4.5 and 4.6 shows that most of the oxygen in the

films is bonded as WO3 and small amount as SO (the small bump at ∼168 eV in the S 2p spectra [13]).

The theory is that the films consist of WS2 and that the surface has a capping layer of elemental sulfur and WO3.

In figure 4.8 AFM images from sample A and D can be seen (AFM images from samples A through G can be found in appendix E). These show no indication of holes in the films apart from a small hole close to one of the

WS2 particles on the surface on sample D (as seen in the inset). Which indicates that the particles either have hit the surface with high energy or that the particles have hindered the sputtering atoms where the hole is in the shadow of the particle. There are also indications of the presence of grain boundaries in the AFM images that also can be seen in the SEM images in figure 4.4.

Combining the SEM, EDX, XPS and AFM results concludes that the

films are full covering, the surface consists of WS2, WO3 and elemental sulfur where the WO3 and elemental sulfur possibly is in a capping layer on top of the film.

29 0.7

0.6

0.5

0.4 µ 0.3

0.2

0.1

0 0 100 200 300 400 500 600 700 800 900 Revolutions 0.7 A 2.3 mTorr 5 nm B 0.6 2.3 mTorr 50 nm C 2.3 mTorr 100 nm 0.5 D 75 mTorr 5 nm 0.4 E 75 mTorr 50 nm µ F 75 mTorr 100 nm 0.3 G CVD graphene 0.2 H 2.3 mTorr without graphene 111 nm 0.1 I 75 mTorr without graphene 255 nm 0 050 100 150 200 250 300 350 400 Revolutions

Figure 4.9: Above can the friction results be seen for the entire test-run; below is a detailed view of the first 400 revolutions for all samples. As can be seen was it only sample H that survived the entire test without increase in µ.

4.3 Tribology Testing

The friction testing on the pin-on-disk setup shows, as can be seen in figure 4.9, that an initial coefficient of friction (µ) of ∼ 0.15 was achieved for sample B and C. This is roughly the same as for the samples without graphene, i.e. sample H and I. Samples A, D, E and F starts of at µ ≈ 0.17. The only samples to survive the entire test of 1000 revolutions were samples C, H and I, although sample H was the only which kept a low coefficient of friction throughout the test, ∼0.08 in the steady state. This can be explained with the formation of a tribofilm of crystalline, layered

WS2 as can be seen in the Raman map of the wear track in figure 4.10. The rest of the samples did not survive the entire test as they after approximately 200 revolutions (500 revolution for sample B) had coefficients of friction of the same level as steel on SiO2, i.e. µ ≈ 0.63 (the value for which all samples converges). Sample A and D, i.e. the 5 nm films, only survived for ∼50 revolutions

30 Figure 4.10: Raman spectroscopy maps showing the formation of WS2 tribofilm and small damages to the film where silicon from the substrate is visible on sample H, the only sample to exhibit low friction throughout the friction test. but these were with a consistent low coefficient of friction meaning that a few nanometers of coating are enough to reach low friction. This can be useful in applications were a low coefficient of friction is needed for a short period of time and where a solid lubricant is needed, e.g. a mechanical component which is to be inserted in a capsule with a tight fit. When regarding the samples up to the point of 400 revolutions the sam- ple without graphene has a lower coefficient of friction at a longer steady state compared to the samples with a buried graphene layer. The films de- posited at 2.3 mTorr have a slightly lower coefficient of friction than the films deposited at 75 mTorr. For the samples with the buried graphene layer sample B and C keeps a lower coefficient of friction for more revo- lutions than sample E and F, i.e. the films deposited at 2.3 mTorr shows better performance than the films deposited at 75 mTorr. One reason for this can be that the 2.3 mTorr films are harder (as seen by the brittle cracks in the film in figure 4.11) than the 75 mTorr films which does not break in a brittle way but more or less crumbles which also can be seen in figure 4.11. This might come from that the 2.3 mTorr films are harder and more denser compared to the 75 mTorr films. One other explanation may be that it is an effect of the adhesion of the WS2-film on SiO2 where the low-damaged graphene layer (deposition at 75 mTorr) actually reduces the adhesion of the film or that the damaged graphene layer (deposition at 2.3 mTorr) increases the adhesion of the film, or a mixture of both effects. Which might not be surprising since a theoretical study has shown that there is low adhesion between graphene and MoS2, an analogue of WS2 [50], this was not discov- ered during the literature study but instead after all experimental work and might have changed the focus of the investigation if discovered earlier. Samples F and I (the two thickest films deposited at 75 mTorr) each shows a dip in the friction coefficient just before the film breaks, for F after 75 revolutions and for I after 325 revolutions. Three explanations comes to mind for this behavior, first it might come from movement of the sample during the test so that it is measurements from new areas of the samples, but this seems unlikely since the coefficient of friction is lower than for the rest of

31 the film in both cases. One other explanation is the formation of a tribofilm consisting if layered WS2 that gives a significant lowering of the friction. And a third explanation is that the films has broken into pieces as in figure 4.11 and the friction that is measured is actually the friction between the film and the piece of film that the ball rests on. There is also a noticeable difference in the lifespan of the two films, where sample F is destroyed after ∼100 revolutions but sample I survives until ∼400 revolutions. This probably arises from the fact that the film in sample I is 2.5 times thicker than the film in sample F.

C F

10 µm 10 µm

Figure 4.11: Shows the difference between the cracked and crumbled films in this example with samples C and F, for C the cracks look brittle and hard and for F it is more of crumpled pieces. Images of all samples can be found in appendix D

We can also se that sample G only survives a handful of revolutions before it is destroyed and the coefficient of friction reaches a stable value of ∼0.6, which is not to surprising since it is only one layer of atoms. Single layer graphene is normally studied in the microscopic scale and for graphene to give low friction in macroscopic environments several studies have used graphene flakes in solution [51] or few-layer graphene [52] where the graphene sheets can slide on each other as in graphite (but still with lower friction than for graphite).

4.4 Characterization of the Wear Tracks

After the friction testing the wear tracks were analyzed with SEM and Ra- man spectroscopy. As can be seen in figure 4.12 the films were severely damaged in all the samples except in sample H, where the film was dam- aged but not in the extent as the other films. There is a big difference in the fracture behavior of the different films. The films deposited at 2.3 mTorr (i.e. samples A, B, C and H) had clean, brittle fractures indicating a hard film whereas the films deposited at 75 mTorr (i.e. samples, D, E, F and I) looked crumbled. This can, as described in section 4.2, be attributed to the difference in density of the films.

32 Figure 4.12: SEM images of the wear tracks on sample C, F, H and I where sample H is the least damaged with the film somewhat still intact.

The difference in the wear of sample C and H is significant and the only difference in the synthesis of the samples is the intermediate layer of damaged graphene in sample C. This is further proof to the statement that graphene is not a suitable substrate for WS2. As can be seen in figure 4.13 and 4.14, the Raman spectroscopy mapping of the wear tracks and the balls all samples shows a strong signal from silicon which reasonable since friction testing shows that the films have been destroyed in all samples except H. Figure 4.10 shows that a tribofilm has been created in sample H. This is also consistent with the friction testing where sample H was the only to have a consistent coefficient of friction throughout the test. As can be seen in the maps of the balls, there has been transfer of silicone from the substrates to the balls. As can be seen in figure 4.15 an unknown substance with a distinct Raman spectrum has formed on the balls from sample G and also D (not shown in the figure). This substance is hematite, i.e. rust [53]. The distinct signal comes from the fact that hematite exhibits resonance when illuminated by a 532 nm laser. The amount of hematite in the sample is actually quite small as can be seen in figure 4.15 since the hematite signal gets smaller with an increase in the laser wavelength (first to 633 nm) and that it disappears completely when illuminated by the 785 nm laser. On the ball from the friction testing on sample C there was another unknown substance found during the mapping. In a few points on the sample a Raman spectrum as shown in figure 4.16

33 Figure 4.13: Maps showing the presence of silicone in the wear tracks of sample A through G and I, map of sample H can be seen in figure 4.10. was found, this has some similarities to the hematite spectrum in figure 4.15, but with a few other peaks. I recognized the spectrum to be similar to a spectrum of pyrite, FeS2. When comparing the peak positions to the know peak position from pyrite and hematite there is a clear match as can be seen in figure 4.16. This leads to the conclusion that it is pyrite and that it have its origin from the contact of iron from the steel balls and the elemental sulfur in the capping layer that have been found on all samples. In figure 4.17 the damage graphene of sample G can be seen. A small par- ticle has hit the surface and the graphene can be seen curled, the graphene is the lighter gray area that looks quite similar to silk.

34 Figure 4.14: Maps showing the presence of silicone in the wear tracks on the balls from the friction testing of sample A through F, H and I, as well as the presence of WS2 on the balls of samples C and H. The ball of sample G did not have either silicon or WS2 present and thereby no map of that sample is presented.

785 nm 633 nm 532 nm Intensity a.u

200 300 400 500 600 700 800 900 1000 Raman shift [cm -1 ]

Figure 4.15: The unknown substance, found to be hematite (i.e. rust) in the wear track of sample G and D and the resonance effect with increasing intensity as the laser wavelength decreases.

35 ×10 4 5

Pyrite Hematite 4

3 Intensity 2

1

0 500 1000 1500 Raman shift [cm -1 ]

Figure 4.16: The Raman spectrum from the unknown substance on the ball from the friction test of sample C with reference of peak position of pyrite and hematite from literature [54, 53].

Figure 4.17: Damaged graphene after friction testing on sample G. The graphene is the darker gray areas and the fluffy, silk-like structure that consists of several layers of graphene; one of these areas is highlighted in red.

36 5. Conclusions and Outlook

The results shown in the thesis show that WS2 is a good material for low friction coatings where the 100 nm WS2 film without graphene have a co- efficient of friction lower than 0.1. The low friction effect is noticeable even when the film is very thin, as with the 5 nm films. These films are though destroyed after only a few revolutions during friction testing. It is also shown that the processing pressure during film deposition strongly influences the films behavior during load in the friction testing. Where films deposited at higher processing pressures are more porous and are peeled off significantly faster than films deposited at lower processing pressures. The porosity of the films also affects the coefficient of friction where the coefficient of friction is lower for films, of equal thickness, deposited at lower processing pressures compared to the higher processing pressure. The damage effects to the films would be interesting to investigate further, possibly by stopping the friction test just before the destruction of the film and investigating the wear tracks.

It is shown that WS2 can be sputtered on graphene with little damage using an increased processing pressure, and that the damage can be quan- tized via Raman spectroscopy even on 100 nm films. That the films are full covering of the surface even down to a film thickness of just 5 nm and that the WS2 component is near stoichiometric close to the surface. The surface of the films are though both oxidized and have a large amount of elemental sulfur which acts as a capping layer. In order to get composi- tional information of the film from under the capping layer some kind of depth profiling is needed, since sputtering depth profiling of WS2 is known to induce damage in the film a HAXPES-studie is the suggested approach for further examination of the film composition.

Further it is shown that graphene is not a viable substrate for WS2 in low friction applications. The graphene layer will reduce the life span, and function of the WS2 whether or not the graphene is pristine or damaged. For future studies of the combined material a good approach is to in- vestigate how temperature affects the composition of the WS2 and if it is achievable to sputter crystalline WS2. It would also be interesting to inves- tigate the electric properties of the graphene after sputter deposition on top for possible electronic applications since the material is not suitable as a low friction coating. Further investigation into the possible van der Waals epitaxy growth of the WS2 on graphene is interesting, for this an extensive diffraction, spectroscopy and microscopy study would be essential possibly using, x-ray

37 diffraction, gracing indices diffraction, high-resolution transmission electron microscopy, electron diffraction, EDX and HAXPES.

38 6. Acknowledgements

First of all I would like to thank my supervisor Andreas Lindblad for the opportunity to do this project and all the help along the way with discus- sions and encouragement. Tomas Nyberg for sputtering of all the films, the many discussions and the basic idea for the sputtering, without you the project would been cut really short. Patrik Ahlberg for teaching me how to grow and transfer CVD-graphene and for the graphene samples for the study. Erik Lewin for the XPS measurements and for reminding me of the difference between Swedish and English. Urban Wiklund and Carl Svanberg for introduction and help with tribology and the tribology laboratory. Hu Li for the introduction to use of the AFM and Linus von Fieandt as my SEM mentor. Mattis Fondell for always finding time to help and support me. Also my friends especially my roommates Micke and Stina, and Tomas for all our lunch discussion.

39 Bibliography

[1] Jill Sundberg. Triboactive Low-Friction Coatings Based on Sulfides and Carbides. PhD thesis, Uppsala University, 2014.

[2] Harald Ski¨oldNyberg. Formation and Function of Low-Friction Tri- bofilms. PhD thesis, Uppsala University, 2014.

[3] Fredrik Gustavsson. Triboactive Component Coatings Tribological Test- ing and Microanalysis of. PhD thesis, Uppsala University, 2013.

[4] Kwang-seop Kim, Hee-jung Lee, Changgu Lee, Seoung-ki Lee, Houk Jang, Jong-hyun Ahn, Jae-hyun Kim, and Hak-joo Lee. CVD Grown Graphene : The Thinnest Solid Lubricant. Materials Science, (6):5107– 5114, 2011.

[5] D. Marchetto, C. Held, F. Hausen, F. W¨ahlisch, M. Dienwiebel, and R. Bennewitz. Friction and wear on single-layer epitaxial graphene in multi-asperity contacts. Tribology Letters, 48:77–82, 2012.

[6] Yumeng Shi, Wu Zhou, Ang-Yu Lu, Wenjing Fang, Yi-Hsien Lee, Allen Long Hsu, Soo Min Kim, Ki Kang Kim, Hui Ying Yang, Lain- Jong Li, Juan-Carlos Idrobo, and Jing Kong. van der Waals Epitaxy of MoS2 Layers Using Graphene as Growth Templates. Nano letters, 12(6):2784–2791, 2012.

[7] C. T. Chen, E. a. Casu, M. Gajek, and S. Raoux. Low-damage high-throughput grazing-angle sputter deposition on graphene. Applied Physics Letters, 103(2013):2011–2016, 2013.

[8] Xue Peng Qiu, Young Jun Shin, Jing Niu, Narayanapillai Kulothun- gasagaran, Gopinadhan Kalon, Caiyu Qiu, Ting Yu, and Hyunsoo Yang. Disorder-free sputtering method on graphene. AIP Advances, 2(2012):0–6, 2012.

[9] K. Ellmer. Preparation routes based on magnetron sputtering for tung- sten disulfide (WS2) films for thin-film solar cells. Physica Status Solidi (B) Basic Research, 245(9):1745–1760, 2008.

[10] Thanasis Georgiou, Rashid Jalil, Branson D Belle, Liam Britnell, Ro- man V Gorbachev, Sergey V Morozov, Yong-Jin Kim, Ali Gholinia, Sarah J Haigh, Oleg Makarovsky, Laurence Eaves, Leonid a Pono- marenko, Andre K Geim, Kostya S Novoselov, and Artem Mishchenko.

40 Vertical field-effect transistor based on graphene-WS2 heterostruc- tures for flexible and transparent electronics. Nature nanotechnology, 8(2):100–3, 2013.

[11] Damien Voiry, Hisato Yamaguchi, Junwen Li, Rafael Silva, Diego C B Alves, Takeshi Fujita, Mingwei Chen, Tewodros Asefa, Vivek B Shenoy, Goki Eda, and Manish Chhowalla. Enhanced catalytic activity in strained chemically exfoliated WS nanosheets for hydrogen evolution. Nature materials, 12(9):850–5, 2013.

[12] V. Weiß, S. Seeger, K. Ellmer, and R. Mientus. Reactive magnetron sputtering of tungsten disulfide (WS2-x) films: Influence of deposition parameters on texture, microstructure, and stoichiometry. Journal of Applied Physics, 101:103502, 2007.

[13] Jill Sundberg, Rebecka Lindblad, Mihaela Gorgoi, H˚akan Rensmo, Ulf Jansson, and Andreas Lindblad. Understanding the effects of sput- ter damage in W-S thin films by HAXPES. Applied Surface Science, 305:203–213, 2014.

[14] I. W. Frank, D. M. Tanenbaum, A. M. van der Zande, and P. L. McEuen. Mechanical properties of suspended graphene sheets. Journal of Vacuum Science & Technology B: Microelectronics and Nanometer Structures, 25(2007):2558, 2007.

[15] K.S. Novoselov, A.K. Geim, S.V. Morozov, D. Jiang, Y. Zhang, S.V. Dubonos, I.V.Grigorieva, A.A. Firsov. Electric Field Effect in Atomi- cally Thin Carbon Films.

[16] D C Elias, R R Nair, T M G Mohiuddin, S V Morozov, P Blake, M P Halsall, A C Ferrari, D W Boukhvalov, M I Katsnelson, A K Geim, and K S Novoselov. Control of Graphene’s Properties by Reversible Hydrogenation: Evidence for Graphane. Science, 323(5914):610–613, January 2009.

[17] Alexander a Balandin, Suchismita Ghosh, Wenzhong Bao, Irene Cal- izo, Desalegne Teweldebrhan, Feng Miao, and Chun Ning Lau. Supe- rior Thermal Conductivity of Single-Layer Graphene 2008. Nano Lett., 8:902–907, 2008.

[18] Keith R Paton, Eswaraiah Varrla, Claudia Backes, Ronan J Smith, Umar Khan, Arlene O’Neill, Conor Boland, Mustafa Lotya, Oana M Istrate, Paul King, Tom Higgins, Sebastian Barwich, Peter May, Pawel Puczkarski, Iftikhar Ahmed, Matthias Moebius, Henrik Pettersson, Ed- mund Long, Jo˜aoCoelho, Sean E O’Brien, Eva K McGuire, Beat- riz Mendoza Sanchez, Georg S Duesberg, Niall McEvoy, Timothy J Pennycook, Clive Downing, Alison Crossley, Valeria Nicolosi, and

41 Jonathan N Coleman. Scalable production of large quantities of defect- free few-layer graphene by shear exfoliation in liquids. Nature materials, 13(June):624–30, 2014. [19] Xuesong Li, Weiwei Cai, Jinho An, Seyoung Kim, Junghyo Nah, Dongx- ing Yang, Richard Piner, Aruna Velamakanni, Inhwa Jung, Emanuel Tutuc, Sanjay K Banerjee, Luigi Colombo, and Rodney S Ruoff. Large- area synthesis of high-quality and uniform graphene films on copper foils. Science (New York, N.Y.), 324(5932):1312–4, June 2009. [20] Sukang Bae, Hyeongkeun Kim, Youngbin Lee, Xiangfan Xu, Jae-Sung Park, Yi Zheng, Jayakumar Balakrishnan, Tian Lei, Hye Ri Kim, Young Il Song, Young-Jin Kim, Kwang S Kim, Barbaros Ozyilmaz, Jong-Hyun Ahn, Byung Hee Hong, and Sumio Iijima. Roll-to-roll pro- duction of 30-inch graphene films for transparent electrodes. Nature nanotechnology, 5(8):574–8, August 2010. [21] Wei Wu, Luis A. Jauregui, Zhihua Su, Zhihong Liu, Jiming Bao, Yong P. Chen, and Qingkai Yu. Growth of single crystal graphene arrays by locally controlling nucleation on polycrystalline Cu using chemical vapor deposition. Advanced Materials, 23:4898–4903, 2011. [22] Xuesong Li, Yanwu Zhu, Weiwei Cai, Mark Borysiak, Boyang Han, David Chen, Richard D. Piner, Luigi Colomba, and Rodney S. Ruoff. Transfer of large-area graphene films for high-performance transparent conductive electrodes. Nano Letters, 9:4359–4363, 2009. [23] William Regan, Nasim Alem, Benjam´ınAlem´an,Baisong Geng, C¸alar Girit, Lorenzo Maserati, Feng Wang, Michael Crommie, and a. Zettl. A direct transfer of layer-area graphene. Applied Physics Letters, 96(2010):1–4, 2010. [24] Keun Soo Kim, Yue Zhao, Houk Jang, Sang Yoon Lee, Jong Min Kim, Kwang S Kim, Jong-Hyun Ahn, Philip Kim, Jae-Young Choi, and Byung Hee Hong. Large-scale pattern growth of graphene films for stretchable transparent electrodes. Nature, 457(7230):706–710, 2009. [25] Si Chen, Zhi Bin Zhang, Laipeng Ma, Patrik Ahlberg, Xindong Gao, Zhijun Qiu, Dongping Wu, Wencai Ren, Hui Ming Cheng, and Shi Li Zhang. A graphene field-effect capacitor sensor in electrolyte. Applied Physics Letters, 101(2012), 2012. [26] Martin, Peter M. Handbook of Deposition Technologies for Films and Coatings: Science, Applications and Technology. William Andrew, In- corporated, 2009. [27] Carl Nordling 1931 and Jonny Osterman¨ 1952. Physics handbook for science and engineering. Studentlitteratur, Lund, 2006.

42 [28] Erik S¨arhammar,Erik Strandberg, Nikolas Martin, and Tomas Nyberg. Sputter rate distribution and compositional variati ons in films sput- tered from elemental and multi-element targets at different pressures. International Journal of Materials Science and Application, Volume 3(2):29–36, 2014.

[29] Xue Yang and Ahmed Hassanein. Atomic scale calculations of tung- sten surface binding energy and beryllium-induced tungsten sputtering. Applied Surface Science, 293:187–190, 2014.

[30] R. V. Stuart and G. K. Wehner. Energy distribution of sputtered Cu atoms. Journal of Applied Physics, 35(6):1819–1824, 1964.

[31] O. Lehtinen, J. Kotakoski, A. V. Krasheninnikov, A. Tolvanen, K. Nordlund, and J. Keinonen. Effects of ion bombardment on a two-dimensional target: Atomistic simulations of graphene irradia- tion. Physical Review B - Condensed Matter and Materials Physics, 81(15):1–4, 2010.

[32] H. S S Ramakrishna Matte, a. Gomathi, Arun K. Manna, Dattatray J. Late, Ranjan Datta, Swapan K. Pati, and C. N R Rao. MoS2 and WS2 analogues of graphene. Angewandte Chemie - International Edition, 49(24):4059–4062, 2010.

[33] Staffan Jacobson 1958 and Sture Hogmark 1943. Tribologi: friktion, sm¨orjningoch n¨otning. Liber utbildning, Stockholm, 2nd edition, 2011.

[34] John R Ferraro, Kazuo Nakamoto, Chris W Brown, and Inc Ebrary. In- troductory Raman spectroscopy. Academic Press, Amsterdam; Boston, 2003.

[35] A (Ado) Jorio. Raman spectroscopy in graphene related systems. Wiley- VCH, Weinheim; Chichester, 2011.

[36] L. M. Malard, M. A. Pimenta, G. Dresselhaus, and M. S. Dressel- haus. Raman spectroscopy in graphene. Physics Reports, 473(5-6):51– 87, 2009.

[37] Y. Y. Wang and et al. Raman Studies of Monolayer Graphene: The Substrare Effect. J. Phys. Chem. C, 112(0001):10637–10640, 2008.

[38] R R Nair, P Blake, A N Grigorenko, K S Novoselov, T J Booth, T Stauber, N M R Peres, and A K Geim. Fine structure constant defines visual transparency of graphene. Science (New York, N.Y.), 320(5881):1308, 2008.

[39] Charles Kittel and Paul McEuen. Introduction to solid state physics. Wiley, Hoboken, N.J, 2005.

43 [40] P. Blake, E. W. Hill, A. H. Castro Neto, K. S. Novoselov, D. Jiang, R. Yang, T. J. Booth, and A. K. Geim. Making graphene visible. Applied Physics Letters, 91(2007):89–92, 2007.

[41] Edward D Palik and Gorachand Ghosh. Handbook of optical constants of solids. Academic Press, San Diego, 1998.

[42] M. Bruna and S. Borini. Optical constants of graphene layers in the visible range. Applied Physics Letters, 94(3), 2009.

[43] M. P. Seah and W. A. Dench. Quantitative electron spectroscopy of surfaces: A standard data base for electron inelastic mean free paths in solids. Surface and Interface Analysis, 1(1):2–11, 1979.

[44] Kai Siegbahn and Carl Nordling. ESCA, atomic, molecular and solid state structure studied by means of electron spectroscopy. Nov. Act. Uppsaliensis, 1967.

[45] H R Verma. Atomic and nuclear analytical methods: XRF, M¨ossbauer, XPS, NAA and ion-beam spectroscopic techniques. Springer, New York; Berlin, 2007.

[46] G. Binnig Quate and C. F. Atomic Force Microscope. Physical Review Letters, 56(9):931, 1986.

[47] Peter Jonathan Eaton and Paul West. Atomic force microscopy. Oxford University Press, Oxford; New York, 2010.

[48] J. S. Zabinski, M. S. Donley, S. V. Prasad, and N. T. McDevitt. Synthe- sis and characterization of tungsten disulphide films grown by pulsed- laser deposition. Journal of Materials Science, 29:4834–4839, 1994.

[49] E S¨arhammar,E Strandberg, J Sundberg, H Nyberg, T Kubart, S Ja- cobson, U Jansson, and T Nyberg. Surface & Coatings Technology Mechanisms for compositional variations of coatings sputtered from a WS 2 target. Surface & Coatings Technology, 252:186–190, 2014.

[50] Yandong Ma, Ying Dai, Meng Guo, Chengwang Niu, and Baibiao Huang. Graphene adhesion on MoS(2) monolayer: An ab initio study. Nanoscale, 3(9):3883–3887, 2011.

[51] Diana Berman, Ali Erdemir, and Anirudha V. Sumant. Graphene: A new emerging lubricant. Materials Today, 17(1):31–42, 2014.

[52] Liang Xu, Tian-Bao Ma, Yuan-Zhong Hu, and Hui Wang. Vanishing stick-slip friction in few-layer graphenes: the thickness effect. Nanotech- nology, 22(28):285708, 2011.

44 [53] M Fondell, F Johansson, M Gorgoi, L Von Fieandt, M Boman, and A Lindblad. Phase control of iron oxides grown in nano-scale structures on FTO and Si ( 100 ): Hematite , maghemite and magnetite. Vaccum, 117:85–90, 2015.

[54] V. Toniazzo, C. Mustin, J. M. Portal, B. Humbert, R. Benoit, and R. Erre. Elemental sulfur at the pyrite surfaces: speciation and quan- tification. Applied Surface Science, 143(1):229–237, 1999.

45 A. SEM-images of films A-F as deposited

46 B. XPS surveys spectra of films A-F F E D C B A

0

W4f

S2p

S2s

5/2

W4d 200

3/2 W4d

C1s

3/2 Ca2p

N1s

3/2

W4p 400

1/2 W4p

O1s W4s

600 F1s Energy [eV] Energy

800

F KLL F O KLL O

1000

C KLL C

1200 S LMM S

1400 Intenisty a.u Intenisty

47 C. SEM-images of the wear tracks

48 D. SEM-images of the brittle cracks and crumbling

49 E. AFM-images of samples A-G

50