Sample Production for the Graphene Nanogap Junction

(1)

Sample Production for the

Graphene Nanogap Junction

THESIS

submitted in partial fulfillment of the requirements for the degree of

BACHELOR OF SCIENCE

in

PHYSICS

Author : R.J.G. van Weelden

Student ID : s1528408

Supervisor : Prof. dr. J.M. van Ruitenbeek Dr. F. Galli

2ndcorrector : Dr. M.P. Allan

(2)

(3)

Sample Production for the

Graphene Nanogap Junction

R.J.G. van Weelden

Huygens-Kamerlingh Onnes Laboratory, Leiden University P.O. Box 9500, 2300 RA Leiden, The Netherlands

July 6, 2018

Abstract

In this thesis, we describe a new sample production method for the graphene nanogap junction of the AMC group. This production method

features a new, standardised, sample layout with integrated metal contacts and a back gate. A large part of the sample production is outsourced commercially. This leads to mass production yielding more reliable samples. We have also introduced some new pieces of equipment

that make the rest of the sample production both easier and more efficient. Additionally, we have proposed a design for a new sample

holder. This will allow us to perform the experiment in cryogenic environments.

(4)

(5)

List of Figures

1.1 Nucleotide detection using graphene as described by Heerema

et al. [1]. 2

3.1 The sample holders of the modified STM. 10

3.2 Voltage on the Z piezo while approaching [2]. 11

4.1 Our new sample design. 15

4.2 Side profile of silicon after anisotropic etching by TMAOH [3] 16 4.3 Side profile of silicon after anisotropic etching by TMAOH,

showing an undercut of width W [3] 17

4.4 AFM images of the surfaces of the metal pads. 18 4.5 Height profile of the edge of the contact pad. 18 4.6 Height profile of the edge of the back gate. 18

4.7 The breaking tool we developed. 19

4.8 Close up of the broken edge. 20

4.9 AFM image of the broken edge. 20

4.10 Top view of the new substrate holder. 21

4.11 Unification of the holder, the Si, and the auxiliary piece. 22 4.12 Side view of the substrate holder, indicating the drain grooves. 22

4.13 Grooves for tweezers. 23

4.14 The design for our glass funnel. 25

4.15 Holder, chips, and funnel combined. 25

4.16 Side view of proposed sample holder design. 26 4.17 Top view of proposed sample holder design. 27 B.1 The two orientations of the graphene in the funnel. 40

(8)

Chapter

1

Introduction

In the world of molecular electronics, the constant search for smaller com-ponents has led to an increase in interest in the transportation properties of single molecules. This knowledge finds particular application in the field of DNA sequencing. As described by Yang and Jiang [4], there are nu-merous problems with our current sequencing technologies. Oftentimes, the techniques used are very costly, very slow, or simply inaccurate. In recent years, a new and very promising technology has been developed: nanopore-based DNA sequencing.

Graphene has proven to be a very useful material in this field. More conventional electrode materials, like gold, consist of atoms that are larger than the size of single nucleotides (around 0.34nm [5]). Graphene, how-ever, is made up of carbon atoms, which are typically smaller than 0.2nm [6]. This, in combination with graphene’s electrical transport properties, makes it exceptionally suitable for the identification of single nucleotides [7]. Heerema et al. [1] describe four techniques for DNA sequencing using graphene, these techniques are illustrated in figure 1.1.

Arjmandi-Tash et al. [8] argue that graphene nanogap technologies, as seen in figure 1.1.b have the potential to yield more accurate measure-ments than nanopore technologies, even though the fabrication is more complicated. In the Jan van Ruitenbeek research group, a modified Scan-ning Tunnelling Microscope (STM) is used to make a graphene-graphene junction with a controllable gap size. The method of measurement is based on the one portrayed in figure 1.1.b.

The main challenge for this group is to improve the fabrication process of the samples, as it has a low rate of success. Afterwards, the experiment should be expanded to low temperature measurements and liquid envi-ronments.

(9)

2 Introduction

Figure 1.1: Nucleotide detection using graphene as described by Heerema et al.

[1]. a) The ionic current through a nanopore in the graphene membrane is used to determine the nucleotide in the nanopore. b) Variation in the tunneling current between graphene electrodes is used to identify passing nucleotides. c) The cur-rent through the graphene layer varies due to the presence of a nucleotide in the nanopore. d) Physical adsorption of nucleotides on graphene affects the current through the graphene layer.

2

(10)

Chapter

2

Theory

2.1 Graphene

Graphene is a 2D material consisting of only carbon atoms. The carbon atoms are arranged in a hexagonal, honeycomb-like, configuration. Graphene is very useful for our experiments because of how thin it is, its excellent conducting properties and its rigidity near the edges. Graphene’s elec-tron density has been found to be up to 2×1011 cm−2 with mobilities of 200, 000 cm2V−1s−1by Bolotin et al. [9]

A large disadvantage is that, while it is very strong for its thickness, macroscopic forces can easily damage the graphene. A single human error can cause significant damage and even render a sample useless. Hence, one of the main challenges when working with graphene is limiting the amount of (direct) human interaction with it.

2.2 STM

The main principle of the STM is the quantum tunneling effect. In the pres-ence of a potential barrier, an electron’s wavefunction does not come to a sudden stop, but decays exponentially inside the barrier. For sufficiently short distances (of the order of ˚Angstr ¨oms), this effect is noticeable. When a potential difference between the two sides (in the form of a bias voltage, for example) is applied, there will be a net current between the electrodes. [10]

(11)

cur-4 Theory

rent constant: at the setpoint current. The feedback regulates the Z piezo voltage and thereby adjusts the gap size as needed.

Current-Voltage (IV) spectroscopy is used to determine the dependence of the tunneling current on the bias voltage. While the bias voltage is being swept, the feedback loop has to be turned off to prevent it from interfering with the measurements. The theoretical prediction of the current density as a function of bias voltage is given by equation 2.1 [11].

J = e 2πhs φ0−eV 2 exp −4πs h √ 2m r φ0−eV 2 !! (2.1) − φ0+ eV 2 exp −4πs h √ 2m r φ0+ eV 2 !!!

with J the current density, e the elementary charge, h the Planck con-stant, V the bias voltage, m the mass of an electron, s the width of the gap between the electrodes, and φ0 the height of the potential

(symmet-ric) barrier. φ0 is approximately the same as the work function, which is

the amount of energy required to move an electron from a material, to the vacuum.

Current-distance (Iz) spectroscopy determines the relation between cur-rent and gap size. By changing the Z piezo voltage, the gap size is altered and the current is measured. This should yield an exponential relation between current and distance. If there is a linear Iz characteristic, that in-dicates there is some point contact due to a sample crash or residuals on the edge of the samples.

2.3 SEM

The Scanning Electron Microscope (SEM) is used in this experiment to check the sample edges. The SEM works by firing a beam of energetic electrons on a sample. This beam excites secondary electrons inside the material that are observed. The SEM at Leiden University Huygens build-ing has a resolution of up to a few nanometers.

2.4 AFM

In this experiment, we make use of the Atomic Force Microscope (AFM) to image the sample edges and contact pads. The AFM senses the short 4

(12)

2.4 AFM 5

range Van der Waals forces between the AFM sharp probe and the sample. It uses this information to produce a high resolution image of the sample surface.

The images we present were acquired in ’AC mode’, also called ’tap-ping mode’. AC AFM scans the sample surface while vibrating the can-tilever at its resonance frequency. The AFM probe is situated at the end of the cantilever. Essentially, in AC AFM the sharp probe repeatedly touches the sample surface and responds to the Pauli repulsion caused by overlap-ping electron orbitals. [12] This is controlled by a feedback system that is a combination of software and electronics, which aims to keep the tapping force constant.

The feedback aims to keep the tapping force, and thereby the vibra-tion amplitude, constant by adjusting the probe-sample distance. Using a laser, that is reflected by the cantilever, the position of the cantilever can be measured. From this, the height of the sample can be deduced.

In this experiment, we used the ”JPK NanoWizard NW3 Ultra Speed” AFM in the LION AFM facility at the Huygens Laboratory. The cantilever used is the Oltespa Opus with a nominal frequency of 70 kHz and a nom-inal stiffness of 2 N/m.

(13)

(14)

Chapter

3

Methodology

3.1 Setup

3.1.1 The modified STM

Contrary to a regular STM with a sample and a tip, the modified STM has two sample holders. The samples are graphene sheets deposited on silicon/silicon oxide substrates.

The sample on the ’tip’ side is moved using 6 shear piezo elements and is referred to as the Z piezo. The other sample can move in the x and y directions, also using shear piezos. This is referred to as the XY piezo.

The samples are rotated with respect to each other, both along the z axis, and along the x axis. The rotation along the x axis is to prevent any contact between the two Si/SiO2 substrates. If the Si substrates were to

get into contact, it might influence the tunnel junction. The rotation along the z axis is to minimise the contact between the two sheets of graphene. This allows us to shrink the junction to atomic size.

There is a bias voltage over the two graphene samples, and at sufficient distance, a tunneling current can be measured. The tunneling current is sent through a current preamplifier (the Femto), which is typically set to a gain of 108. The Femto converts the current to a voltage which is sent to the controller. This means a measured voltage of 1 V corresponds to a current of 10 nA.

(15)

8 Methodology

3.1.2 The RHK controller

For the control of the STM, we use an RHK controller: the RHK SPM 100. This controller regulates the voltages sent to the XY and Z piezos, as well as the bias voltage. This controller also contains the feedback system that is used in probing for the tunneling current. The feedback system also keeps the current constant when in tunneling mode. This constant current is referred to as the setpoint current.

3.2 Method

3.2.1 Sample preparation

The samples are prepared by Amadeo Bellunato, from the Gregory Schnei-der group. Since a large part of the project is to improve the sample prepa-ration, much of this may change during the project. Because of its rel-evance to the project, we will describe the process of preparation com-pletely.

The production of the samples consists of many steps, all of which have a significant probability of failure, mainly due to human interaction with the samples. We start with preprepared 1×1 cm pieces of copper with a mono-atomic layer of graphene on top, processed by Chemical Vapour Deposition (CVD). We coat these samples with a solution of polycarbon-ate (PCA) in chloroform (1.5 mass%) As an artefact of the CVD process, there is also some graphene on the bottom of the Cu plates. However, this graphene is not a fine sheet and, as such, is not useful for our purposes. So, first, we will need to remove the useless graphene from the Cu. We do this by blasting the lower side of the Cu plate with H2plasma. The Cu plates

are always stored in a small vacuum chamber, to prevent oxidisation of the Cu.

Next, we will have to make the silicon substrates for the graphene. For this, we use undoped Si wafers (with a 250 nm layer of SiO2). By

scratching the edge with a diamond knife and applying pressure on the cut, the Si will break straight along the crystalline edge. The Si plates are cut to a size of roughly 7.5 by 3.0 mm and placed on a tiny stainless steel base with clamps that hold the Si in place. The edge of the substrates is then checked under an (optical) microscope to see if they are sufficiently smooth. Afterwards, the Si is cleaned with acetone, isopropanol (IPA), ethanol and water.

Now, the graphene can be placed onto the Si. Using a razor, the Cu 8

(16)

3.2 Method 9

(and graphene) is cut to size by hand. The pieces are then placed into a petri dish with 0.5 M ammonium persulfite for a few minutes to etch away the Cu. After only the graphene and the polymer are left, they are put into water to remove any residuals of the ammonium persulfite.

Before transfer, the Si substrates need to be cleaned. For this, the pi-ranha solution is used. This is a 3:1, H2SO4:H2O2 solution. The substrate

is left in the piranha for about 5 minutes. Note that piranha is a very ag-gressive and dangerous substance. All preparation and use of piranha should be done inside a fume hood and never be done alone. Simply mix-ing the two substances produces a lot of heat and the piranha will need to settle down for a few minutes.

When the substrate and the graphene are both clean, it is time to trans-fer them on top of each other. For this, we make use of a special guiding setup using a needle that can be moved in the x, y and z directions very precisely with differential screws. The Si substrates are placed into a petri dish filled with milliQ (ultra-pure) water. The graphene is put onto the water and the needle is attached to the polymer.

While the water is slowly pumped away, the needle slowly guides the graphene to the correct location. Note that the Si used must be hy-drophilic. If it is not, and thereby hydrophobic, the fast movement of water away from the Si might cause the graphene to be dragged to an unfavourable position. Old Si becomes hydrophobic, but it is easily made hydrophilic again via an O2plasma treatment.

Afterwards, the junction is completed by plasma etching the graphene bridging the gap between the two Si substrates, again using H2 plasma.

The polymer coating is then removed with chloroform, and cleaning with methanol and IPA.

Again, the samples are checked under the optical microscope, but the only way to really check their usefulness is by trying them out in the STM. Obviously, there are many steps in this process that have a large prob-ability of failure. Often, the main problem is that the steps rely on human interaction with the samples. The most natural step in trying to improve upon this fabrication process is to try and limit the amount of direct hu-man interaction with the samples.

3.2.2 Aligning the samples

The graphene samples are held in the STM sample holders: plastic holders with little springs in them to clamp the samples tight. The holders are shown in figure 3.1. These sample holders can be rotated around a central

(17)

10 Methodology

axis for alignment of the graphene sheets. A Cu wire is attached to the Si substrate and the graphene with a bit of silver paste. The wire is then also attached to a contact at the back of the sample holder.

A-A ( 3 : 1 ) A A 1 /1 3th angle pen WorkInProgress 29-6-2018 sasha 3 : 1 mm

Filepath:P:\My Documents\QMO Ruitenbeek\grafeen point contact\Gijs\sasha.iam

LEIDEN UNIVERSITY PROPRIETARY THIS DOCUMENT CONTAINS CONFIDENTIAL PROPRIETARY INFORMATION THAT IS PROPERTY OF LEIDEN UNIVERSITY DO NOT DISCLOSE TO OR DUPLICATE FOR OTHERS EXCEPT AS AUTHORIZED BY LEIDEN UNIVERSITY

NEN-ISO 2768- NEN-ISO 1101 NEN-EN-ISO 1302

Dimensional tolerancing: Geometrical tolerancing: Surface roughness: Material: Designed by: Design status: Release date: Group:

Qty: Partname:

Project: Scale: Units: Sheet: Leiden institute of physics

Fine mechanical department Niels Bohrweg 2 2333 CA Leiden Netherlands L ei d en u n iv er si ty

FMD

Free rotation with specia

l tool

17

Figure 3.1:The sample holders of the modified STM. The dovetail ends slide into

pieces that are attached to the XY piezo and the slider.

3.3 Measuring technique

In order to detect tunneling current, the two electrodes must be very close to each other. Because the range of the piezo elements is about±100 nm, we cannot simply use them for large displacements of the samples. To remedy this, we make use of ’stick-slip’ motion, shown in figure 3.2.

By extending the piezos to near maximum and suddenly reducing the voltage to zero, the piezos can overcome static friction when they snap back. Making use of this principle, it is possible to use the piezos to move the samples indefinitely, in principle.

10

(18)

3.3 Measuring technique 11

22

Figure 3.7: Coarse motion with slip-stick. (a) first “stick” then “slip” phase, tip moves towards the sample. (b) first “slip” then “stick” phase, tip moves away from the sample.

Figure 3.6: voltage of the Z piezo elements during approach. (a) Stick-slip is initiated. (b) The slider sticks on the piezo elements (in an ideal case). (c) The slider slips, therefore the position of the slider is closer to the sample (in the ideal case). Probe ramp to check for tunneling current is initiated. (d) The tunneling current during the probing did not exceed the set point within the voltage range of the piezo elements. (e) The slider is retracted again a new slip-stick is performed again, this process continues until the tunneling current exceeds the set point: (f). Then the tip is retracted: (g).

Figure 3.2: Voltage on the Z piezo while approaching [2]. a) Stick-slip procedure

is started. b) The piezo moves forward and the slider sticks. c) The piezo rapidly moves back to its earlier position, overcoming static friction. The slider slips. A new probing cycle is initiated. d) The tunnellnig current did not exceed the setpoint and the piezo has reached the end of its voltage range. e) The slider is retracted and a new stick-slip procedure is initiated. This is repeated until the tunnelling current exceeds the set point. f) The setpoint current is reached, the tip is retracted. g) The tip is retracted until it is manually released. When released, the feedback will keep the current around the setpoint.

Once in tunneling distance, the gap size can be controlled very pre-cisely, allowing for current-distance (Iz) spectroscopy. Also, by sweeping the bias voltage, instead of the distance, a current-voltage (IV) characteris-tic can be made.

(19)

(20)

Chapter

4

Results

The preparation of the samples is currently the main limiting factor in the graphene tunnel junction. Because of the high failure rate, many samples end up not being useful. As such, time and money go to waste and the subsequent research can build only on a very limited number of measure-ments.

The preparation method clearly needs a lot of improvement to make it easier, more reliable, and faster. This will in turn lead to a larger number of useful samples, leading to research based on better statistics.

Moreover, expanding the experiment to cryogenic environments will allow us to learn more about the fundamental aspects of the system. This could give information on the characteristics of the junction in the pres-ence of certain molecules. Alternatively, this part of the experiment is necessary for verifying the suspicion of the presence of a chain of single carbon atoms, brought about after graphene-graphene contact [13].

4.1 Overview

We have decided to outsource a large part of the sample production by or-dering diced (sawed) wafers instead of cutting samples to size ourselves, as described in section 4.2. This saves us a lot of time and effort and yields more regularly and precisely dimensioned Si pieces. This demands that the rest of the sample production process is more precise as well. In sec-tion 4.2.3, we discuss the fabricasec-tion of a tool that allows us to break our silicon at a precisely defined point. In order to better protect our Si sub-strates before the transfer of graphene, we have designed a new holder

(21)

14 Results

with additional precautions, presented in section 4.3. We have also devel-oped a new graphene transfer method, as described in section 4.4, using a self-stabilising system. Section 4.5 discusses the design of a new sample holder for the samples.

4.2 New samples

There are four main decisions that started the development of a new de-sign for the samples. First of all, we decided to replace the silver paste contact with a metal pad on the Si. Secondly, we wanted a back gate to be placed on the Si. Thirdly, we wanted to chemically etch a groove to break the Si on, as it can be more precisely placed than a scratch made by hand. Lastly, we wanted the samples to be more regularly and precisely dimensioned to minimise variance in sample quality.

This part of the sample production has been carried out by Hugo Schelle-vis, who processed the wafers at the Else Kooi Lab (EKL) in Delft. The dic-ing of the wafers was also done in Delft, at the Kavli Institute of Nanoscience.

4.2.1 Sample layout

The contact made with silver paste is difficult to reproduce exactly. Also, the silver paste is unfit for organic solvent environments. Since future ap-plications will demand the applicability in such environments, we should find an alternative. We therefore decided it would be useful to use contact pads made of gold.

If we were to put two of these pads on the silicon, we could measure the contact resistance of the metal-graphene interface. That would allow us to more accurately determine the resistance of the tunnelling junction. Having two contact pads on the Si also grants us the ability to perform four-point measurements.

Because of the machine properties at EKL, it was unfavourable for us to use gold as the material for our contact pads. Instead, we decided to use aluminium. During the Al deposition, 0.5 mass% Si will evaporate into the Al. To prevent this from happening, we use saturated Al that already contains 1 mass% Si. The problem with using aluminium contact pads is that Al has a natural oxide layer. This oxide is isolating and thus we need to prevent it from forming.

For this, we coat the Al/Si with titanium nitride. TiN is a strong in-dustrial coating that has minimal oxidation in environments with temper-atures up to 250°C [14]. It is also useful in later phases where we use O2

14

(22)

4.2 New samples 15

plasma for cleaning, as it has minimal (and well-known) oxidation in those conditions as well [15].

In addition to the two contact pads, we wanted to add a back gate to our samples. The function of this back gate is to change the electrical transport properties of the graphene. The back gate is a metal pad that is in direct contact with the Si. By putting a voltage on the back gate, and per extension the Si, we electrostatically influence the charge carrier density of the graphene. Note that the use of a back gate forces us to use doped Si, as undoped Si is an isolator. Also, we will need to locally etch away the silicon oxide layer, because the back gate requires contact with the Si.

3th angle

pen WorkInProgress 28-6-2018

LEIDEN UNIVERSITY PROPRIETARY THIS DOCUMENT CONTAINS CONFIDENTIAL PROPRIETARY INFORMATION THAT IS PROPERTY OF LEIDEN UNIVERSITY DO NOT DISCLOSE TO OR DUPLICATE FOR OTHERS EXCEPT AS AUTHORIZED BY

Generic

Qty: Partname:

Leiden institute of physics

si ty 1 2,5 0,5 0,5 0 ,5 0 ,5 0,5 0 ,7 5 1, 5 ₃ 5 0 ,5

Figure 4.1:Our new sample design. All measurements are in mm.

As shown in figure 4.1, we chose sample dimensions of 5×3 mm. We wanted the contact pads to be very large, to have a high chance of getting a good contact with the graphene. The pads stop short of the edge at 1 mm distance, to avoid damage of the graphene due to height differences.

Assume a pad height of 500 nm. If we consider the thickness of the 15

(23)

16 Results

graphene layer to be in the order of ˚Angstr ¨oms, we see the height differ-ence between the Si edge and Al pad is over a thousand times the thick-ness of graphene. In order to avoid a steep angle of the graphene at the edge, leading to instability and internal stresses, the distance between the Si edge and the contact must be sufficiently large. A distance of 1 mm from the edge, means the horizontal distance the graphene goes is 2000 times larger than the vertical distance it must travel. This leaves the graphene with enough space to counteract this high climb.

The most important part of our silicon substrates is the edge. The man-ually placed cut was often inaccurately placed, yielding too large or too small samples, and was sometimes too slanted to produce a straight break-ing line. Obviously, it is preferable to have the cut be machined into the Si in some way to improve accuracy. We opted for a chemical etch of the Si, opting for TMAOH etching. TMAOH is an anisotropic silicon etchant that creates a V-shaped groove in a well-defined way, as shown in figure 4.2.

Because it is not possible to ensure a straight etch along the entire width of the Si, we have chosen to have the etched groove extend from the side for 0.75 mm. The reason for choosing TMAOH instead of KOH, a similar yet faster etchant, is that SiO2is a mask for TMAOH, but not for KOH. By

choosing TMAOH, the production process is simpler as we do not need to apply and remove the extra mask.

Figure 4.2:Side profile of silicon after anisotropic etching by TMAOH [3]

Because there is a non-trivial undercut at the Si/SiO2, we cannot make

this groove very deep. The undercut is shown in figure 4.3. Upon breaking the chips, the overhanging SiO2can shatter and ruin the edge.

Addition-ally, the groove gets wider as it gets deeper, using up more space of our samples. On the other hand, if the groove is too shallow, the chips will not break cleanly.

It is very difficult, if not impossible, to accurately find an optimum for groove depth. However, an estimate can be made for a groove that will yield reasonable results. We opted for a groove of 20 µm deep, which corresponds roughly to 3.5 to 4% of the Si thickness, depending on the specific thickness of the oxide layer.

Because we outsourced the production process up to this point, the only time we have to break the Si, is when applying the final break. This 16

(24)

4.2 New samples 17

Figure 4.3: Side profile of silicon after anisotropic etching by TMAOH, showing

an undercut of width W [3]

severely limits the amount of critical steps. Even though the diced edges are not clean, even on micrometer scale, we can afford to do this because the inner edge where the graphene goes is the only one that has to be sharp.

4.2.2 Production process

In this section we describe, in short, the steps taken by Hugo Schellevis in the production process. The log book of the process can be found in appendix A.

The wafer used is a p-type wafer (containing more free holes than free electrons), with a single polished side. After cleaning, the alignment marks for the machines are etched into the wafer. Then, the wafer is stripped of its oxide layer, so it can be reapplied in a controlled manner, yielding an oxide layer of well-known thickness. After removing the ox-ide at the positions of the back gates and the breaking grooves, we etch into the Si with TMAOH. Subsequently, a layer of metal (675 nm Al/1%Si plus 40 nm TiN) is applied to the entire wafer.

In figure 4.4, AFM images are shown of the surfaces of the metal pads. The edges of the metal pads and their height profiles are shown in figures 4.5 and 4.6.

Next, a mask is applied and the metal is etched away, leaving behind the contacts and 20 µm wide lines between the chips that serve as align-ment for the dicing procedure. Note that this leaves the samples slightly smaller at the end than we initially intended. The width of the saw used in dicing is 50 µm. Therefore, the chips have a width of:

(25)

18 Results 4 µm 99 nm 0 10 20 30 40 50 60 70 80 90

(a) AFM image of the contact

pad. 4 µm 138 nm 0 20 30 40 50 60 70 80 90 100 110 120 130

(b)AFM image of the back gate.

Figure 4.4:AFM images of the surfaces of the metal pads.

4 µm 1 692 nm 0 100 200 300 400 500 600

(a)AFM image of the edge of the contact

pad.

(b)Height profile along line 1 in

figure 4.5a.

Figure 4.5:Height profile of the edge of the contact pad.

4 µm 1 1.0 µm 0.0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

(a)Height profile of the back gate.

(b)Height profile along line 1 in

figure 4.6a.

Figure 4.6:Height profile of the edge of the back gate.

18

(26)

4.2 New samples 19

Afterwards, a protective photoresist layer is applied and the wafer is sent off to be diced. This final protective layer prevents ’sawdust’, from the dicing, and other contaminants from soiling the surface of our Si chips.

After delivery, we further process the chips at the Huygens Laboratory. First, we must remove the protective photoresist layer. This is done in the clean room. We start by flushing the chips with acetone, which removes the bulk of the photoresist. Immediately after, we flush the chips with IPA to prevent the acetone from evaporating and leaving behind extra poly-mer residues. We then place the chips in an IPA bath for a few minutes. Afterwards, we used an O2 plasma (P ≥ 100W, t ≥ 30s) to clean off any

residue. After removing the photoresist, we can break the chips.

4.2.3 Breaking tool

The delivered chips consist of two Si pieces with contacts. Therefore, to get from these chips to full samples, we first need to break them in two along the etched groove. Because the groove is so small, only 20 µm wide, we cannot afford to freehand this. As a solution to this, we designed and manufactured a tool that we can use to break the chips precisely at the groove. The breaking is done by pressing down at the edges of the chip with the fingers (use gloves). Using tweezers to break the chips damages the sides and backs of the chips substantially. The grooves allow us to pick up the Si pieces safely with tweezers.

A ( 4 : 1 ) A 30 R 0,5 0 ,5 11 9 Silicium Chip

Figure 4.7: The breaking tool we developed. The Si chips are positioned with

the etched groove above the semicircle and broken by pressing down with the fingers.

(27)

20 Results

In figure 4.8, an AFM image of the broken edge is shown. Figure 4.9 shows an image taken with a larger scanning window, showing us more of the silicon surface. In these images, we achieved a resolution of 4 nm per pixel. Thereby, we can see the edge is fairly sharp at the atomic level.

4 µm

1.00 µm

0.00 0.50

Figure 4.8: Close up of the broken edge, visible as the dark strip. The bumps on

the Si surface are dust and polymer residues.

500 nm 235 nm 0 40 60 80 100 120 140 160 180 200 220

Figure 4.9: AFM image of the broken edge. The light strip in the bottom left is

an artefact of the ”Polynomial Background” function in the data acquisition. We speculate that the stripes on the silicon are probably due to either the polishing or the etching process.

20

(28)

4.3 New substrate holders

The substrate holder is the part that holds the Si pieces during the piranha cleaning, graphene transfer, and H2 plasma etching. This used to be a

simple stainless steel plate with clamps and a hole in the middle for the H2plasma to go through.

There were numerous problems with this part. First, while putting the Si in the holder, there was a very high probability of the substrates crashing and the edges being ruined. This is because the Si pieces were free to move in all directions. To counteract this, we made a new substrate holder that limits the lateral movement of the Si pieces by placing them in

grooves, illustrated in figure 4.10. A ( 2 : 1 )

A

3th angle

LEIDEN UNIVERSITY PROPRIETARY THIS DOCUMENT CONTAINS CONFIDENTIAL PROPRIETARY INFORMATION THAT IS PROPERTY OF LEIDEN UNIVERSITY DO NOT DISCLOSE TO OR DUPLICATE FOR

Dimensional tolerancing: Geometrical tolerancing: Surface roughness: Material:

Designed by: Design status: Release date: Group:

Qty: Partname:

si

ty

Drain

space for tweezers space for sample

see through

Figure 4.10:Top view of the new substrate holder.

There is also an auxiliary piece that is inserted from the bottom that protects the Si pieces from crashing into each other. Because of its trian-gular shape, if the Si crashes into the auxiliary piece, only the bottom will touch it. This way, the edge at the top is protected from damage. Since the top part of the edge is directly shielding the graphene during the plasma etching, this is the part that must remain undamaged.

This auxiliary piece demanded the sample holder be placed on an ele-vation. Therefore, we soldered two stainless steel bars on the sides under the holder so that it can fit the auxiliary piece under it, as shown in figure 4.11.

Another problem with the old holder is in the transfer of graphene. Because the old holder was flat at the bottom, there was no way for the water inside the central hole to escape during pumping. This meant there was a water residue at the point where the graphene was to be placed. Therefore, the pumping had to be interrupted and redirected to the hole manually, opening a window for loss of control on the graphene position-ing. In order to avoid such direct human interaction in the middle of the transfer process, we added grooves at the bottom of the holder for the water to flow away. This drain is shown in figure 4.12.

(29)

22 Results A-A ( 1 : 1 ) B ( 4 : 1 ) A A B 1 /1 3th angle pen WorkInProgress 28-6-2018 Assembly1 1 : 1 mm

Filepath:P:\My Documents\QMO Ruitenbeek\grafeen point contact\Assembly1.iam

Qty: Partname:

FMD

Figure 4.11:Unification of the holder, the Si, and the auxiliary piece.

A ( 2 : 1 ) A 1 /1 3th angle pen WorkInProgress 29-6-2018 Assembly1 1 : 1 mm

Filepath:P:\My Documents\QMO Ruitenbeek\grafeen point contact\Assembly1.iam

Qty: Partname:

FMD

Drain

see through

Figure 4.12:Side view of the substrate holder, indicating the drain grooves.

22

(30)

This new holder introduced a new problem: once inside, the samples were hard to reach and it became very difficult to get them out. For this, we made vertical grooves at the backsides of the grooves for the Si, as illustrated in 4.13. These grooves allow us to take out our samples with a pair of tweezers.

A ( 2 : 1 ) A

3th angle

LEIDEN UNIVERSITY PROPRIETARY THIS DOCUMENT CONTAINS CONFIDENTIAL PROPRIETARY INFORMATION THAT IS PROPERTY OF LEIDEN UNIVERSITY DO NOT DISCLOSE TO OR DUPLICATE FOR OTHERS EXCEPT AS AUTHORIZED BY

Qty: Partname:

si

ty

Drain

see through

Figure 4.13: Grooves for tweezers. Useful when putting the Si in and taking it

out.

4.3.1 Mounting the substrates

Mounting the Si substrates in the new holders is fairly straightforward. There are, however, a few things that demand special attention.

Firstly, it is important to dry the substrate holder thoroughly after clean-ing. We recommend cleaning in acetone for five minutes, using a sonicator. Flushing with IPA and distilled water afterwards gets rid of any residues. For drying, we use an N2gun.

(31)

24 Results

While it may seem unnecessary to dry off the water before placing the holder in more water, it certainly is not. Water residues on the holder and the springs will attract the hydrophilic Si, which will stick quite firmly in often undesirable spots on the holder.

When clamping the Si down with the springs, it is important to not tighten them too firmly. Large forces could damage the silicon or the metal pads on them. In addition, the springs do not apply a completely uniform force along the width of the Si. Because of this, the spring will rotate the Si slightly when pressing down. This is not necessarily problematic, but could become an issue when clamping down too much. Therefore, an optimum must be found for which the rotation of the chip is acceptable, while also being held firmly enough.

For ease of use and increased precision, we recommend the use of neg-ative action tweezers for mounting the Si chips. Obviously, it is necessary to use carbon tip tweezers to avoid damaging the Si and the metal pads.

4.4 Graphene transfer

The graphene transfer method using the needle requires constant direct human interaction. This is the most obvious point for improvement and hence our focus lies here. We made a design for a funnel that should auto-matically centre the graphene through capillary forces. If the tube is made of a hydrophilic material, it should repel the hydrophobic polymer, due to the surface tension of water. This effect is described by Vella and Ma-hadevan [16]. Our first design was made of stainless steel, but it did not have the intended effect, as the polymer stuck to the sides of the funnel. We therefore made a second design, shown in figure 4.14, that is made of glass.

As can be seen, this funnel has extra holes drilled into it to allow it to rest on the new Si holder. These holes also help in aligning the funnel above the Si substrates. In figure 4.15 a unification of the holder and the funnel is shown.

Experimentally, we observe that our graphene/PCA rotates by 90° with respect to the funnel shape. This means that the short side of the graphene rotates towards the funnel’s long side or, more appropriately, the long side of the graphene rotates away from the funnel’s long side. As such, in its current state, the funnel does not add much to the safe transfer of graphene.

24

(32)

4.4 Graphene transfer 25 A-A ( 1 : 1 ) B-B ( 1 : 1 ) A A B B 1 /1 3th angle pen WorkInProgress 27-6-2018 trechter 1 : 1 mm

Filepath:P:\My Documents\QMO Ruitenbeek\grafeen point contact\trechter2.ipt

Glass

Qty: Partname:

FMD

30 4 28 4 7 2 8

Figure 4.14:The design for our glass funnel. All measurements are in mm.

A-A ( 1 : 1 ) _{B ( 2 : 1 )} C ( 2 : 1 ) A

A

B C

LEIDEN UNIVERSITY PROPRIETARY

Figure 4.15:Holder, chips, and funnel combined.

(33)

26 Results

4.5 New sample holders

We also made advances towards new sample holders. Because cryogenic experiments require specific equipment, it is impossible to simply cool down the STM we have been working with so far. Therefore, we need to alter the STM for cryogenic experiments to be able to hold the new sam-ples. For this, we need to increase the distance between the STM tip and sample holders by at least 3 mm.

It is convenient for the sample holder to have integrated contacts, so there will be no need for wire bonding on the pads. Instead, the sample can just be slid in and the contacts are formed automatically. In order for this to happen, we have decided to place three wires (500 µm diameter) in a piece of isolating material, visible in figure 4.16.

1mm=10u

2.4 3.4 5.0 2.0 1.0 1.0 0.44 Pressure point Leaf spring 1.6

Figure 4.16:Side view of our proposed sample holder design. All measurements

are in mm.

A leaf spring, attached to the bottom of the holder, presses the sam-ple up against the top plate. The leaf spring has a small dent, facing up-wards. This dent serves as the pressure point of the spring and is situated at the centre of mass of the triangle spanned by the three contact points, as 26

(34)

shown in figure 4.17. This ensures contact is made at all three pads. We want the Si samples, especially the relevant edges, to remain in place rigidly. The two main noise factors in this regard are thermal drift and acoustic vibrations from the environment. Of these two, the sam-ple holder’s shape can only counteract acoustic vibrations directly. This acoustic noise can manifest in two ways: vibrations in the sample holder and in the Si itself. However, Si has a very high Young’s modulus [17] and therefore its suspension in the sample holder is much more critical than the internal vibrations. Therefore, it is desirable to pack the sample as tightly as possible in the holder, while also leaving room to align it properly with the contacts.

Because the samples are rotated around two axes, their bottoms and sides are relatively far away from each other. Therefore, we can extend the holder along the sides and the bottom to add more stability to the samples. As shown in figures 4.16 and 4.17, the holder extends to 0.1 mm short of the sample edge.

4.0 0.4 Pressure point 1.0 1.0 0.5 2.5 0.625 1.875 1.5 4.9 2.5

Figure 4.17: Top view of our proposed sample holder design. All measurements

(35)

28 Results

At the back of the holder, behind the sample, there is another piece. This piece not only adds structural stability by connecting the top and bottom parts of the sample holder, it also stops the sample from going too far into the holder. This stopper is situated 0.1 mm behind the back edge of the sample, when in the design position.

28

(36)

Chapter

5

Discussion

5.1 Stripping recipe

There are many ways to strip the resist off of the Si chips. While there are numerous effective methods, a clear analysis of risk, cost, and result must be made. The most preferred methods for stripping resists often involve chemicals such as fuming (100%) HNO3or NMP (1-methyl-2-pyrrolidone)

[18]. Using these chemicals in the clean room at the Huygens Laboratory requires a lot of preparation, as they are not in stock by default. This makes it preferable to stick to the acetone-IPA-O2recipe described in section 4.2.2.

We have tried a number of different stripping recipes, but we only var-ied the plasma cleaning power and time. A more systematic approach to optimising the stripping process is needed if there is reason to believe it will lead to considerably better results in graphene-metal contact. We currently do not have enough data to draw any conclusion on stripping recipes.

5.2 Graphene etching time

There is much room for optimisation of the etching time of the graphene. A too short etching time might not fully remove the graphene bridging the gap and leave graphene droop hanging over the edges of the Si. This could cause problems during tunnelling. Because of the voltage over the samples, electrostatic attraction could pull such droop on one sample to-wards the other sample, which could come into contact with the graphene

(37)

30 Discussion

there. This would form a new graphene bridge that could be difficult to pull apart.

On the other hand, if we etch too long, we affect the polymer too much. The H2 plasma distorts the inner structure of the polymer. This leaves

carbon residues on our samples that dissolve very poorly in any solvents known to us.

5.3 New sample holders

Our proposed design for the sample holders features a leaf spring that pushes the sample up against the contact. However, for the stability of the samples it is preferable to have the contacts be pressed down upon the sample. This does make fabrication of the holders quite a bit more difficult and requires further attention.

5.4 Graphene transfer

As shown in section 4.4, the glass funnel we designed did not have the intended effect. The design does seem to exhibit its intended working principle, but in an unexpected and undesirable way. We speculate, how-ever, that this is merely an artefact of the specific shape of the funnel and not necessarily an indication that this kind of design does not work.

Therefore, we propose an alternative design, in the hope that it will be tested in the future. On the basis of a qualitative argument and an analytical approximation, we suggest a future design.

The qualitative argument is the most intuitive and so we will start with that. We have observed the graphene/PCA to move so its short edges face the funnel’s long ones. We expect this tendency to remain if we change the specific dimensions of the funnel. The problem was that the funnel’s long edges were perpendicular to where we wanted the short edges of the graphene/PCA to face.

If we were to widen the funnel, so that its width becomes larger than its length, the graphene/PCA should, in accordance with its earlier be-haviour, align its short edges with the funnel’s long edges. Then, the fun-nel’s long edges are parallel to where we want the graphene/PCA short side to face, and thus it will be aligned appropriately. Alternatively, we could make the funnel hole square at 7×7 mm. This would mean the graphene/PCA will have no preferred direction. Because the funnel will retain its stabilising property, the alignment should work out.

30

(38)

In addition to this qualitative reasoning, we will support our recom-mendation with a theoretical argument. We assume that the graphene/PCA floating in the glass funnel can be described using the equations of energy and force found by Vella and Mahadevan [16] in paragraph V.B for the force between two interacting spherical particles. The full derivation is found in appendix B.

We consider the energy of the following two situations. In the parallel position, the graphene/PCA is aligned parallel to the funnel hole. In the orthogonal alignment, it is rotated 90° with respect to the funnel hole. By solving a crude approximation for the energy of the two states, we find a good measure of the funnel width to be between 5 and 7 mm.

On the basis of both the qualitative reasoning and quantitative approx-imation, we therefore recommend testing of the glass funnel with a square hole of 7×7 mm at the bottom.

(39)

(40)

Bibliography

[1] S. J. Heerema and C. Dekker, Graphene nanodevices

for DNA sequencing, NATURE NANOTECHNOLOGY

www.nature.com/naturenanotechnology 11 (2016).

[2] M. Simons, Development of an STM with combined displacement and scan unit, (2016).

[3] K. Westra, KOH and TMAH Etching of Bulk Silicon Recipes, Tricks, What is Possible, and What is Impossible, (2010).

[4] N. Yang and X. Jiang, Nanocarbons for DNA sequencing: A review, Car-bon 115, 293 (2017).

[5] A. S. M.H.F. Wilkins and H. Wilson, Molecular Structure of Deoxypen-tose Nucleic Acids, Nature 171, 738 (1953).

[6] A. Bondi, van der Waals Volumes and Radii, THE JOURNAL OF PHYS-ICAL CHEMISTRY 68, 441 (1964).

[7] N. Agra¨ıt, A. L. Yeyati, and J. M. van Ruitenbeek, Quantum properties of atomic-sized conductors, Physics Reports 377, 81 (2003).

[8] H. Arjmandi-Tash, L. A. Belyaeva, and G. F. Schneider, Single molecule detection with graphene and other two-dimensional materials: nanopores and beyond, Chemical Society Reviews 45, 476 (2016).

[9] K. Bolotin, K. Sikes, Z. Jiang, M. Klima, G. Fudenberg, J. Hone, P. Kim, and H. Stormer, Ultrahigh electron mobility in suspended graphene, Solid State Communications 146, 351 (2008).

[10] G. Binnig and H. Rohrer, Scanning tunneling microscopy, Surface Sci-ence 126, 236 (1983).

(41)

34 BIBLIOGRAPHY

[11] K. N. Kanneworff, The viability of single nucleotide detection using a graphene nanogap, (2017).

[12] AZoNano and Bruker Nano Surfaces, Fundamentals of Contact Mode and Tapping Mode Atomic Force Microscopy, 2012.

[13] A. Bellunato, S. D. Vrbica, C. Sabater, E. W. De Vos, R. Fermin, K. N. Kanneworff, F. Galli, J. M. Van Ruitenbeek, and G. F. Schneider, Dynamic Tunneling Junctions at the Atomic Intersection of Two Twisted Graphene Edges.

[14] Y. Yin, L. Hang, S. Zhang, and X. Bui, Thermal oxidation properties of titanium nitride and titanium-aluminum nitride materials - A perspective for high temperature air-stable solar selective absorber applications, Thin Solid Films 515, 2829 (2007).

[15] C. Jim´enez, J. Perri`ere, C. Palacio, J. Enard, and J. Albella, Transfor-mation of titanium nitride in oxygen plasma, Thin Solid Films 228, 247 (1993).

[16] D. Vella and L. Mahadevan, The ’Cheerios effect’, (2007).

[17] M. A. Hopcroft, W. D. Nix, and T. W. Kenny, What is the Young’s Modu-lus of Silicon?, JOURNAL OF MICROELECTROMECHANICAL SYS-TEMS 19 (2010).

[18] MicroChemicals GmbH, Photoresist Removal, 2013.

34

(42)

Appendix

A

Production Log Book

Flowchart: Standaard Metallisatie: vRuitenbeek Starting material: 2 wafer, P-type, 1-5 Ohm.cm, SSP

1. Cleaning: Fuming HNO3 + Boiling HNO3, QDR till 5MΩ.cm, Rinse&Dryer TT=1, Lithografie: FWAM

2. Coating SPR3012: EVG 120: Co 3012 Zerolayer(=1,4mu)

3. Exposure ASML Stepper: Comurk mask, 120mJ/cm2, Litho/FWAM 4. Development: EVG120: Dev SP

5. Dry Etching120nmSi: Omega201: urk npd,t=40sec

6. Cleaning: Tepla#1 + Fuming HNO3 + Boiling HNO3, QDR till 5MΩ.cm, Rinse&Dryer

TT=2, Oxidatie:

7. Remove native oxide: 0,5%HF dip: Hydrofoob 8. 300nm Oxidatie D1: VarOxid1000C, t= use

9. Measurement Oxide Thickness: SP Leitz#Thermal oxide, 5ptsDox= 272,4+/-1,8nm

(43)

36 Production Log Book

10. Coating Spr3012: EVG 120: Co 3012 1,4mu 11. Exposure: EVG420: TMAH-mask, expt= 6sec

12. Development: EVG120: Dev SP

13. Manual SPR3012 resist op FWAM aanbrengen +1 min uitbakken@115C 14. Dry etching oxide: Drytek: StdOxide,t= 33sec

15. Cleaning: Tepla#1 + Fuming HNO3 + Boiling HNO3, QDR till 5MΩ.cm, Rinse&Dryer

16. Dip wet etching: remove native oxide: 0,5%HF dip, 30sec

17. Selective Si etching: SALAB:33minTMAOH@85C,Visuele Controle (VC) niet goed, +15 min extra TMAOH@85C, VC niet OK, +15 min extra TMAOH@85C, VCOK! Remark: Deeltjes zichtbaar!

18. Cleaning: Fuming HNO3, QDR till 5MΩ.cm, Rinse&Dryer, VC -deeltjes weg

TT=4, Metallisatie

19. Dip wet etching: remove native oxide: 0,5%HF dip, 30sec

20. Sigma204: 675nmAl/1%Si@350C + 40nmTiN@350C(extra 2 dummy wafers meegenomen voor stap 24, tussen elke wafer een extra wafer voor Ti targetclean om voor iedere process wafer dezelfde conditie te krijgen)

TT=5, Lithografie: Metal

21. Spray Coater: Pos resist + Ass. EKL,Resist dikte: 5,5 +/- 0,5 mu (a) EVG120: Only HMDS 50 sec; aanbrengen van een primer (b) EVG101: Spray coater: #HP 1000mbar 2ml 4 layers

(c) Bake 1min, @115C

(d) EVG101: Spray coater: #HP 1000mbar 2ml 4 layers (e) Bake 1min, @115C

(f) EVG101: Spray coater: #HP 1000mbar 2ml 4 layers (g) Bake 5min, @115C

36

(44)

37

(h) Wait for 30 min

22. Exposure: EVG420: Metal-mask,exp t= 180 sec (a) Wait for 30 min

23. Manual Development: AZ400K,dilute 1:2(water), t=6min

24. Omega# etch TiN and Al/Si partly,etch recipe: Al poly 3 (a) Step 1: 40 sec, endpoint rising

(b) Step2: 45 sec, bulk anisotropic Al/Si etching, still 10-20% Al/Si left

(c) Step 3: 40 sec, isotropic Al/Si etching, landing on oxide

25. Wet Al etch PES, remove Al/Si, Step 24c etst geen Si precipitaten, dus:

(a) 10 sec wet Poly etch dip (Hier moet wat oxide zijn ge¨etst, Poly etch bevat HF: oxide dikte niet gemeten)

(b) VC: geen precipitaten zichtbaar, V groef niet aangetast.

26. Cleaning: Tepla#1 + Fuming HNO3 for metals, QDR till 5MΩ.cm, Rinse&Dryer

TT=5, Zagen

27. Spray Coater: Pos resistResist dikte: 5,5 +/- 0,5 mu (a) Manual HMDS: 10 min

(b) EVG101: Spray coater: #HP 1000mbar 2ml 8 layers (c) Bake 1min, @115C

(d) EVG101: Spray coater: #HP 1000mbar 2ml 8 layers (e) Bake 5min, @115C

(45)

(46)

Appendix

B

Deduction of Optimal Funnel Size

Vella and Mahadevan define the following constants for a spherical par-ticle with radius R. LC ≡

p

γ/ρg is the capillary length, which gives the

length scale over which interactions occur. Here, γ is the surface tension at the air-water interface, ρ is the density of water, and g is the gravitational acceleration. This gives a measure for the Bond number B ≡ R2/L2_C, and subsequently the non-dimensional resultant weight of the particlesΣ.

They then find for the energy:

E(l) = −2πγR2B2Σ2K0 l LC (B.1) where K0the modified Bessel function of the second kind of order 0.

Because F= −dE_dl, they find it to be:

F(l) = −2πγRB5/2Σ2K1 l LC (B.2) with K1the modified Bessel function of the second kind of order 1.

Let us now consider the funnel with length lf and width wf, such that

its width is orthogonal to the samples in the holder and consider a piece of graphene/PCA floating in the funnel with length lgand width wg. Lastly,

let us consider the following two situations, illustrated in figure B.1. In the parallel position, the graphene/PCA is aligned parallel to the funnel (width parallel to width and length parallel to length). In the orthogonal alignment, it is rotated 90° (width parallel to length and length parallel to width).

(47)

40 Deduction of Optimal Funnel Size

Figure B.1:The two orientations of the graphene in the funnel. Left is the parallel,

right is the orthogonal orientation.

Assume the energy, contributed to the total, per side to be:

E =laE(l) (B.3)

with la the length of the graphene/PCA side that we want to know the

energy of. For simplicity, we ignore the contribution of the corners. This gives us a simple expression for the energy of both states:

Ek =wgE _w f −wg 2 +lgE _l f −lg 2 =wgK0 _w f −wg 2LC +lgK0 _l f −lg 2LC (B.4) E⊥ =wgE _l f −wg 2 +lgE _w f −lg 2 =wgK0 _l f −wg 2LC +lgK0 _w f −lg 2LC (B.5)

In the equations above, we have left out the factor two for the number of sides and the constants (−2πγR2B2Σ2) in E(l), because we only want to look at the difference between the two energy levels. We want Ekto be the

system’s preferred energy state as it is then aligned with our Si. Therefore, we want the condition Ek < E⊥to hold.

When we plug in the appropriate values for γ, ρ, and g at 20°C, we find that LC = 2.728mm. Using wg = 2.0mm, lg = 5.0mm, and lf = 7.0mm,

this gives us: 40

(48)

41 Ek =2.0K0 _w f −2.0 5.456 +5.0K0(0.37) =2.0K0 _w f −2.0 5.456 +6.0 (B.6) E⊥ =2.0K0(0.93) +5.0K0 _w f −5.0 5.456 =0.94+5.0K0 _w f −5.0 5.456 (B.7)

This leaves us with the inequality: 2.0K0 _w f −2.0 5.456 +5.06 <5.0K0 _w f −5.0 5.456 (B.8) Using a simple python script, shown in appendix C, we find that this gives wf ∈ [5.01, 6.97]. The energy difference of the two states indeed goes

to 0 as wf goes to 7. Additionally, the difference between the forces acting

(49)

(50)

Appendix

C

Python script

import numpy as np import scipy.special x = np.arange(0,10,0.01) E par = 2 * scipy.special.kn(0,(x-2)/5.456) + 5.06 E perp = 5 * scipy.special.kn(0,(x-5)/5.456) for i in range(len(x)):

if not (np.isnan(E par[i]) or np.isinf(E par[i]) or np.isnan(E perp[i]) or np.isinf(E perp[i]):

if E par[i] ¡ E perp[i]:

(51)

(52)

Acknowledgements

During this project, I have been helped by a number of people. Without them, I could not have come as far as I have in these last months.

First of all, I would like to thank R´emi Claessen, who worked on the same setup and with whom I shared a large part of the project. His refreshing insights and immeasurable patience with an unruly STM have been a huge help to me. Next, I would like to thank prof. dr. Jan van Ruitenbeek for granting me this opportunity and for placing the trust in me, to work on this project despite the absence of a daily supervisor. I want to thank dr. Frederica Galli for the time she spent helping me with the STM and the AFM. I would like to thank Hugo Schellevis for the work he did for us and for always being available for discussions. I am very grateful to Christiaan Pen from the FMD, for the discussions on the designs, for turning our ideas into reality, and for supplying the images of the models for my thesis. I would like to thank Kim Akius for always being available for questions and for his patience with my silly mistakes when handling acids. Lastly, I would like to thank MSc. Sasha Vrbica, for her help with the STM setup, MSc. Amedeo Bellunato, for teaching me how to make the samples, and Pauline van Deursen for helping me out in the lab during the tests of the new preparation method.

Sample Production for the Graphene Nanogap Junction

Sample Production for the

Graphene Nanogap Junction

Sample Production for the

Graphene Nanogap Junction

R.J.G. van Weelden

Abstract

Contents

List of Figures

Chapter

1

Introduction

Chapter

2

Theory

2.1

Graphene

2.2

STM

2.3

SEM

2.4

AFM

Chapter

3

Methodology

3.1

Setup

3.1.1

The modified STM

3.1.2

The RHK controller

3.2

Method

3.2.1

Sample preparation

3.2.2

Aligning the samples

FMD

3.3

Measuring technique

Chapter

4

Results

4.1

Overview

4.2

New samples

4.2.1

Sample layout

4.2.2

Production process

4.2.3

Breaking tool

4.3

New substrate holders

FMD

FMD

4.3.1

Mounting the substrates

4.4

Graphene transfer

FMD

4.5

New sample holders

1mm=10u

Chapter

5

Discussion

5.1

Stripping recipe

5.2

Graphene etching time

5.3

New sample holders

5.4

Graphene transfer

Bibliography

Appendix

A