Dynamics of epitaxial graphene growth and adsorptions of cobalt 3

This system is used for preparation of clean surface, growth of thin films and also in-situ characterisations.. Once sample is loaded onto the transfer arm inside FEL, the load-lock door

Research Methods and Procedures

Advances in physics follow advances in instrumentation,

once the instruments are made, many people will make the discoveries

– Abraham Pais, 1986 Inward Bound: Of Matter and Force in the Physical World

At the beginning of this chapter, the ultra-high vacuum (UHV) system

used for this research programme is introduced This system is used for

preparation of clean surface, growth of thin films and also in-situ

characterisations Procedures used to prepare clean graphite and silicon

carbide surfaces are followed immediately The details of the growth

conditions and deposition techniques i.e electron beam evaporation are

also provided This is followed by descriptions on main characterisation

techniques used i.e X-ray photoelectron spectroscopy (XPS), scanning

tunneling microscopy (STM) and atomic force microscopy (AFM) where

the theory, operation and interpretation of data are included This chapter

ends with explanations on sampling procedures used to collect statistics

from STM and AFM results

3.1 The UHV system

Ideally terminated surface will reconstruct (semiconductor) or relax (metals) to reduce the dangling bonds density on the surface However, this does not assure the surface is completely free of dangling bonds In air, it will oxidise and their well-ordered surface structure will be destroyed Even inert surfaces such as graphite and gold are not spared from weak physical adsorption of particles and other gaseous species These contaminants prevent any efforts in studying the true dynamic reflected by this surface when exposed to vapour of film materials Hence, it becomes necessary to work under a strictly clean environment during the duration of experiment Carrying out the experiments under vacuum conditions is one of the ways to accomplish this Gas kinetic theory (see Appendix 1 for more details) predicts that pressure better than 1.0 x 10-10mbar (ultra-high vacuum, UHV regime) is needed to prevent contamination up to 8 hours (typical duration for adsorption studies experiment) UHV also permits techniques such as low energy electron diffraction (LEED) and XPS to function without scattering from stray gaseous molecules At 1 x 10-10 mbar, the mean free path of stray molecules is 9 x 106 m which far exceeded the diameter of any UHV chambers (30 to 60 cm)

3.1.1 Layout of UHV system used

All in-situ STM experiments were done at Surface Science Laboratory at Physics Department, National University of Singapore (NUS) while all the in-situ and ex-situ

XPS experiments were done at Institute of Materials Research and Engineering (IMRE)

The STM system is designed by Omicron GmbH while the XPS system is designed by

VG Scientific The STM system will be described in greater details since 70% of the work were carried out here

The STM system has three chambers (likewise for the XPS system) as seen in Fig 3.1 i.e a fast-entry lock (FEL), a preparation chamber (PREP) and an analysis chamber (ANA) All three chambers are separated by hand operated gate valve Only one sample can be loaded at a time FEL serves as the entry and exit point to the UHV system To introduce sample into the FEL, the FEL is vented using 99.999% nitrogen Once sample

is loaded onto the transfer arm inside FEL, the load-lock door is closed and FEL is immediately pumped down using turbo pump station After 30 minutes, the pressure will

Fig 3.1 Schematic drawing of UHV system at Surface Science Laboratory made by Omicron GmbH that consists of three chambers i.e fast-entry lock, preparation chamber and analysis chamber A scanning tunneling microscope is housed here, attached to analysis chamber Other ports are not drawn

Analysis

chamber

(ANA)

Preparation chamber (PREP)

Rotary pump

Rotary pump

Manipulator

on bellow

normally achieve mid 10-9 mbar The gate valve between FEL and PREP is opened to transfer sample into PREP and is immediately closed back after sample transfer is completed This loading procedure prevents the PREP from direct exposure to atmosphere Due to frequent venting, the base pressure of FEL is often at 10-9 mbar, one order higher than PREP

All the growth techniques such as home-made Si source and electron-beam evaporator are mounted on the PREP Preparation of clean surface and growth of materials are carried out here The PREP is pumped by a turbo pump that is supported on

a rotary pump and also an ion pump Only ion pump is used most of the time since its operation is cleaner than turbo pump which may experience contamination from oil-based rotary pump Turbo pump is used when (i) inert gas is needed to be pumped away (ion pump has low efficiency for inert gas), (ii) as an additional support when pressure rises to high 10-8 mbar during sample preparation or (iii) during system baking where ion pump is not suitable for pressure 10-5 and above (operating at these pressures reduces its life expectancy) Due to repeatable sample cleaning and exposure to FEL, the base pressure

of PREP (5.0 x 10-10 mbar) is slightly poorer than ANA (1.1 x 10-10 mbar) The ANA is supported by an ion pump only All the characterisation work is carried out here i.e STM (or XPS for system at IMRE) For the XPS system, the ANA has a layer of mu-metal to shield electrons emitted from sample from stray magnetic fields other than the electric fields created by the lens system of the XPS The pressure of PREP and ANA are separated from one another using gate valves installed between them The gate valve also

isolates the ANA from contamination during in-situ cleaning of substrate in PREP

3.1.2 Baking procedures

To attain 10-10 mbar or better, the chamber must first be baked to increase the outgassing rate (removal of gas molecules) of the chamber wall Baking is also needed for effective removal of moisture trapped in the system Baking is usually done every time after the system is vented for installation or repair work Typical baking conditions are wrapping the whole UHV system and kept at 150oC for more than 24 hours while continually pumped by turbo or ion pumps When the system is cooled after baking completed, high vacuum is generated due to decrease of outgassing rate The base pressure of the system depends on the outgassing rate of the system and pumping efficiency of the pumps Since most of the UHV components fitted with filament operate

at temperature higher than 150oC, a separate degassing procedure is needed This is normally carried out when the chamber is still hot from baking to minimise recontamination from gaseous molecules onto the chamber wall Degas often starts with components with bigger filament such as titanium sublimation pump (TSP) and ion gauges since they release more residual gaseous and followed by smaller filament components such as X-ray gun and manipulators Degas of growth sources such as silicon (Si) and e-beam evaporator are normally done as the last step

3.1.3 Sample transfer, heating and current feedthroughs

As seen from Fig 3.1, the sample is maneuvered between the chambers using transfer arms and manipulators The transfer arm provides linear movement in one direction parallel to it and also rotation, while the manipulator provides linear translation

in all three x, y, z directions and also rotation A sample stage is mounted on this manipulator for preparation of clean surface and growth studies Current and voltage are applied to the stage via feedthrough mounted on the atmospheric side There are two ways that a sample can be heated as shown in Fig 3.2a The first method is known as direct heating (DH) where current flows through the sample and can be varied to control temperature This heating is suitable for doped semiconducting materials The second method is known as resistive heating (RH) where current flows through a coiled filament sit directly underneath the sample This filament, wrapped by ceramic beads, forms the base of the sample stage The heat from the filament is used to heat up the sample This method is suitable for metals or insulators The temperature is read using infrared pyrometer with emissivity set to 0.68 Figure 3.2b shows the power-temperature calibration for silicon carbide, 6H-SiC(0001) heated using DH

Fig 3.2 (a) Schematic of direct (DH) and resistive heating (RH) facilities for sample preparation.The sample slot and actual position of sample holder (dotted line) are also shown For DH, the current flows via terminal 1 and 3 This circuit is completed if the sample is present where the sample holder’s contact bar will be in physical contact with manipulator’s contact brush Withoutthe sample, the circuit becomes open For RH, current flows via terminal 2 and 3 Filament coilbelow the sample is used to heat up the sample; and (b) shows the power-temperature calibration

of silicon carbide (SiC) heated using DH The emissivity of pyrometer is set to 0.68

(a)

0 200 400 600 800 1000 1200 1400

1

2 3

Contact bar Sample holder

3.2 Preparation of clean surfaces

3.2.1 Preparation of highly oriented pyrolytic graphite

Graphite substrates in the form of highly oriented pyrolytic graphite (HOPG) were purchased from Goodfellow Cambridge Limited Similar to graphite with AB (Bernel) stacking, these substrates are polycrystalline with average domain size of 1m diameter Each substrate was cut into 10mm x 5mm x 1mm to fit onto a sample holder From literature, there are two surface cleaning methods which are also compared in this work

The first method involves cleaning the HOPG ex-situ [1,2] Since the graphite sheets

along the z-direction are held together by weak physical forces, a fresh surface can be easily generated by peeling off first few top layers using adhesive tape This substrate was then immediately mounted onto a sample holder and loaded into the FEL and PREP Without any further treatment, this substrate was used directly for experiment The

second method repeats the ex-situ preparation method However, after transferred into the

PREP, the substrate was also annealed [3,4] Due to low resistivity nature of HOPG (graphite is semimetal), direct heating is not suitable to achieve high temperature Hence, the substrate was prepared using resistive heating as described earlier The substrate was mounted onto a sample holder with a window that helps minimised heat blocking (see Fig 3.3) As the filament is heated up, the temperature of substrate that sits directly above it also increases (Fig 3.3d) The increment of temperature was carefully controlled to maintain pressure below low 10-8 mbar To clean HOPG, the substrate was held at 600oC until pressure recovers near the base pressure, a sign that outgassing from substrate is completed This takes approximately 60 to 90 minutes

3.2.2 Preparation of (6 3x63)R30 o /6H-SiC(0001)

6H-SiC(0001) substrate (on-axis, produced by CREE Inc.) was used to prepare rich (63x63)R30o phase (hereafter 63) The wafer was diced into 10 mm x 2 mm dice size to fit the sample holder Prior loading to UHV chamber, the wafer was submerged inside isopropanol (IPA) solution and sonicated for 15 minutes and followed by acetone

C-by another 15 minutes Due to the wide band gap nature of SiC, very high voltage was needed to drive current through the substrate and the voltage is above the limit of the power supplies in our laboratory Instead, a doped Si substrate was stacked underneath the SiC substrate and current was passed through it to heat up the SiC Prior to that, the Si

Fig 3.3 Resistive heating (RH) sample holder (a) Parts of resistive heating sample holderconsist of one sample plate with window, four sets of stud and nut and two pieces of tantalum (Ta) metal strip The window at the center allows direct heat transfer from underneath to the sample, (b) a sample is placed on top of sample plate, (c) a sample is placed at the center of the window and secured by a pair of screw-tighten Ta metal strips and (d) cross-section of the heating stage with a RH sample holder mounted with a HOPG substrate

x4

(d)

~2mm HOPG

HOPG

Glowing filament

substrate was cleaned in similar manner as SiC and together with SiC, they were mounted onto a DH sample holder (see Fig 3.4) Sample mounting procedures and descriptions of current flow are provided in Fig 3.4

The substrates were degassed for approximately 8 hours at 500oC as shown in Fig 3.5, after which slow heating process took place by increasing the current slowly while maintaining pressure below mid 10-9 mbar To remove surface native oxides, Si deposition was carried out at 850oC for 10 minutes where the oxides will desorb as volatile SiOx species This was followed by annealing without Si flux for another 15 minutes to allow surface gains stability and flashes off excess Si Next, the temperature was gradually increased systematically to 950, 1050 and 1150oC (20 minutes at each temperature) At these temperatures Si starts to desorb from the surface and generates other phases i.e Si-rich (3x3)R30o (between 950oC and 1050oC) and C-rich (63x63)R30o (1150oC) [5,6,7,8] For preparation of C-rich 63 surface, the annealing

Fig 3.4 (a) A blank direct heating sample holder The tantalum (Ta) strip on the left is in contact with the molybdenum (Mo) base plate via Mo studs and washers underneath The Tastrip on the right and the contact bar are insulated from the Mo base plate using ceramic washers The two Ta strips will only be in electrically contact with each other if a sample is put

across as shown in (b) Current, I flows to the metal contact bar, through the sample, to the Ta

strip on the left and then the Mo base plate; and (c) ceramic top plate is fixed for further insulation and to secure the sample onto the holder before loading into FEL

I

Ceramic washer

Mo washer

Contact bar

Ta strip Connector between

contact bar and strip

Ceramic top plate SiC/Si

process completes at temperatures between 1100oC and 1150oC The surface was brought

to cool gradually by decreasing the current slowly

3.2.3 Preparation of graphene on 6H-SiC(0001)

Fig 3.5 Procedures for preparation of clean 63 on 6H-SiC(0001)

0 200 400 600 800 1000 1200 1400

Fig 3.6 Procedures for preparation of graphene on 6H-SiC(0001)

0 200 400 600 800 1000 1200 1400

To prepare epitaxial graphene on 6H-SiC(0001), the initial procedures described

in Section 3.2.2 to prepare 63 were repeated and followed by annealing to temperatures above 1300oC (see Fig 3.6) The coverage of the graphene monolayer can be control using time and temperature In order to follow the graphitisation process systematically, various partially graphitised SiC surfaces were prepared The surfaces were first annealed

at the onset temperature of graphene formation i.e 1200oC for 20 minutes before cooled

down Submonolayer of Cobalt (Co) was evaporated via an in-situ e-beam evaporator Co

was used as tracer to differentiate 63 and graphene surface due to its different adsorption behaviour on these two surfaces After characterised by STM, clean surfaces can be recovered by desorbing the physisorbed Co at ~1000ºC The graphitisation was continued

by increasing the temperatures by 30-50oC and annealed for 20 minutes The sequence above was repeated for a few temperatures until the surfaces are completely graphitised at

1300oC For kinetic studies, the surfaces were also annealed as a function of time for each temperature

3.3 Growth via electron beam evaporation

Omicron’s EFM-3 e-beam evaporator and power supply was used as Co source for all growth and deposition experiments This evaporator uses electron beam as source

of heating to reach the melting temperature of Co evaporant Current is applied through a filament located in front of the evaporant (in the form of metal rod with 2mm in diameter)

as shown in Fig 3.7 This evaporant is connected to high tension As the heated filament emits electrons, the electrons are accelerated to the tip of the evaporant (approximately 700-900V is applied to the evaporant) and the tip of the evaporant is slowly heated up

The current is normally increased slowly to increase the intensity of the electrons and once the melting temperature of the evaporant is reached, atoms at the tip are vapourised and atomic flux will be generated The Co flux can be control via filament current and voltage applied to the evaporant and this allows deposition at different arrival rates to be possible

3.3.1 Calibration of Co flux

The Co flux was calibrated using quartz crystal microbalance (QCM) and STM 100/MF thickness monitor manufactured by Sycon Instruments, Inc The QCM is a mass sensor with extremely high sensitivity (1Å thickness resolution) and has detection limits

in submonolayer regime It consists of an exposed piezoelectric quartz crystal sandwiched

Fig 3.7 Design of an Omicron’s EFM-3 e-beam evaporator High voltage is applied to the Co rod to attract electrons emitted from filament

UHV

vapourised

Co atomssubstrate

HV & linear motion

Filament (+/-), thermocouple

to flux monitorwater cooling

W filament

e-beam heated Co rod

vacuum seal

between two gold electrodes A drive unit is used to monitor the resonant frequency, fo of

the oscillating quartz (typically at 6 MHz) As materials are deposited on this quartz, its fo

is changed by the additional load which is detected at 4 cycles per second and converted

to thickness by the drive unit

Figure 3.8a shows the thickness of Co film increase linearly with the quartz exposure time The Co thickness is read out from a monitor display with parameters of Co

such as density pre-set into the monitor Two different powers, P (emission current*voltage) applied to the Co rod are shown as examples The emission current represents the electron intensity emitted from filament Arrival fluxes, F (monolayer per

unit time) extracted from the slopes are 1.6x10-2 ML/s (8.1 Watt) and 6.8x10-4 ML/s (6.7 Watt) respectively The linearity of the curves in Fig 3.8a shows the stability of the EFM-

3 evaporator for a period up to 2 hours Once the flux is calibrated, the coverage of Co exposed to a surface at any given time can be calculated Figure 3.8b shows calibrated Co

flux versus P where the flux is obtained in the same manner as in Fig 3.8a Figure 3.8b shows another important characteristic of the evaporator where F is a linear function of the evaporator’s power Fitting Fig 3.8b gives F = 0.0088P-0.0567 Hence for any power used, F can be estimated based on the relation above As seen in Fig 3.8b, the curve does

not start at origin (0,0) as 6.5 Watt is needed to heat up the Co rod from room temperature

to its melting point Hence for materials with higher melting point such as platinum (Pt), higher power is expected to generate Pt flux

As aforementioned, the density is pre-set into the QCM’s monitor assuming Co forms hexagonal closed packed (hcp) structure The thickness of Co is calculated using

volume= mass /density relation where mass is deduced from the change of f o Most of the

time, the deposited Co on QCM is not perfect single crystal (hence lower density) and the pre-assumed density causes the thickness (and flux) to be overestimated

When the QCM is not available, a second calibration method using STM is employed Using a STM image of submonolayer Co adsorbed on flat graphite with area

of A nm2, a background equivalent to the uncovered graphite is selected and any height above it (due to Co mass) will be integrated and extracted as Co volume The deposition

is carried out at room temperature to obtain good sticking coefficient Flux, F is estimated

by dividing the volume with volume of 1 ML (i.e area A*0.2nm) and deposition time t

The 0.2nm used here is the interlayer spacing of hcp Co Since Co does not form hcp structures but 3D clusters on graphite, this method also overestimate the Co coverage

0.000 0.005 0.010 0.015 0.020 0.025

3.3.2 Growth of Co on HOPG, (6 3x63)R30 o

and graphene

Co was evaporated on HOPG, 63 and graphene using Omicron’s e-beam evaporator Various fluxes from 10-3 to 10-2 ML/s were used for growth at room temperature Higher fluxes of 10-2 ML/s were used for growth at elevated temperatures A shutter that attached at the front of the EFM’s aperture was used to block Co flux whenever the growth needed to be interrupted for coverage-dependent studies (for e.g

morphology and interaction) using in-situ STM or XPS Besides growth studies, Co were also deposited at room temperature on ex-situ and in-situ prepared HOPG to examine the

influence of contaminants, if any Co were also employed as tracer to follow the transformation of 63 to graphene as it has different adsorption on these surfaces

3.4 Main Characterisation Techniques

3.4.1 X-ray Photoelectron Spectroscopy (XPS)

The X-ray photoelectron spectroscopy (XPS) was invented in 1969 by K Siegbahn [9] It is conceptualised based on Einstein’s theory on photoelectric effect

[10,11] where the photon and its energy, hv has to be absorbed as a packet in any adsorption process Hence, as long as hv is known, measuring the kinetic energy, KE of

the photoejected electron (photoelectron) will give us the binding energy, BE of that electron as given in Eq (3.1) below:

where  is the work function of the material Since then the core level BE of various elements has been extensively studied using XPS and compiled into handbooks [12] The ejection is assumed a three-step process i.e (i) absorption where an electron is excited by

a packet of the incoming photons This causes it to be ejected from its location as illustrated in Fig 3.9, (ii) ejected electron travels and may collide with lattice en route to the sample surface, and (iii) the electron escapes from the surface into the vacuum Due to collision in second step, all the XPS spectra have two components i.e (i) the main peaks contributed from photoelectrons that escaped without inelastic collision and (ii) spectra background contributed from inelastic collision where the kinetic energy of photoelectrons lost at varying amount Only the main peaks reflect the density of states (DOS) of the material (see Fig 3.9) and can be used to interpret the XPS data

(ii) Binding energy and chemical shift

XPS is also used to resolve the electronic states of elements when two or more elements form a molecule or solid For reactive system, valence electrons from participating elements will combine and form new density of states near the Fermi edge

BE of core-level electrons are affected due to different screening of core electrons from its nucleus in the new environment The shift of BE from the BE of pure element, BE (or chemical shift) is reflected by the change of KE, KE according to Eq (3.1) i.e.:

element X Its core-level electrons will experience stronger attraction from its nucleus and hence a shift to higher binding energy will be observed by XPS Likewise, the more electronegative element will have an effectively larger electron cloud and its core-level electron will experience less attraction from its nucleus and a shift to lower binding energy will be observed For e.g when Si is bonded to a more electronegative C in 6H-

SiC(0001), Si 2p will shift by +2.7 eV (from 99.3 eV to 102 eV) while C 1s will have a

lower binding energy and shifted by -1.4 eV (from 285 eV to 283.6 eV) [13]

As aforementioned, although XPS often display the spectra in binding energy format, it is the kinetic energy of the photoelectrons that are actually measured by XPS The following section discusses the instrumentation of a XPS system

(iii) Instrumentation and acquisition procedures

All the XPS experiments were carried out at IMRE, Singapore using a VG Scientific’s 220i Escalab-XL spectrometer and a monochromatic Al Kα source Figure 3.10a shows the schematic diagram of this XPS system that consists of twin anode X-ray source (gun), a quartz crystal (monochromator), a spectrometer and a user interface Both the X-ray gun and spectrometer are controlled using user-interface The spectrometer itself consists of lens, concentric hemispherical analyser and a six discrete electron detectors/ multipliers (channeltrons) The twin anode X-ray gun allows user to select between Al or Mg anode (1486.6 eV for Al Kα and 1253.6 eV for Mg Kα) Only Al source can be monochromatised via the [1010 ] plane of quartz crystal that has lattice spacing of 0.425 nm (coincides with Al Kα wavelength of 0.834 nm) The monochromatic source

provides better resolution (0.3 eV) than non-monochromatic source (0.85 eV), smaller spot size but lower intensity The low intensity can be compensated by increasing the acceptance angle of the analyser to cover broader range of photoelectron take-off angle The acceptance angle is controlled via a lens system inside the analyser tube as shown in Fig 3.10a

Power supplies and electronics

The concentric hemispherical analyser is built up of two metal hemispheres that are arranged such that their centre of curvature coincides with one another as seen in Fig 3.10b Photoelectrons travel through the lens system and enter the hemispheres via an entrance slit However, only electrons with certain KE can reach the other end of the

analyser i.e the detector This KE must match a user-defined Pass Energy (PE) This PE

is created inside the analyser via potentials applied to the two hemispheres with the outer hemisphere’s potential, Vout more negative than the inner hemisphere, Vin An electric field will be created with the PE forms the center pathway between two hemispheres as shown in inset of Fig 3.10 The potential difference between Vin and Vout makes sure that only electrons with very small energy spread from PE are allowed in the gap of the hemisphere without any collision These energies spread are normally 1% of the PE i.e if

PE of 20 eV is selected, ΔPE is 0.2 eV Electrons that travel faster than PE+ΔE will impinge on the outer hemisphere while electrons travel slower than PE-ΔE will impinge

on the inner hemisphere There are two modes which the lens and analyser can be

operated In this work, all acquisitions were done using Constant Analyser Energy (CAE)

mode where the analyser is held at a constant PE throughout the experiment The lens system is left to do all the work where it will scan a range of potential via a set of voltages applied to the lens according to a manufacturer’s prescription known as transmission functions (TF) While ramping through a series of potentials in steps, electrons of different KE are retarded with a same potential simultaneously Only electrons with energy matches the PE after retardation will pass through the analyser and be detected Table 3.1 shows an example how XPS can be used to detect and identify the BE of

photoelectrons emitted from C 1s of graphite

Table 3.1: Detection and binding energy computation for C 1s photoelectrons excited by Al Kαsource

Excitation source, hv (pre-select via software) 1486.6 eV

Kinetic energy, KE of electrons 1198 eV

Pass energy, PE of analyser 100 eV

Scanning potential at which electron detected 1098 eV

Analyser work function,  (pre-set by user) 4.2 eV

Binding energy, BE of electrons (as displayed) 1486.6 – (100 + 1098) – 4.2 = 284.4 eV

Each element produces a unique fingerprint of spectrum according to their electronic levels and they are usually checked against XPS handbook for material identification The resolution of the analyser is given as ΔE/PE Since PE is constant in the CAE mode, the analyser resolution is constant to any of the electron KE before retardation Hence this mode is suitable for peak area quantification Nevertheless the resolution can be improved by adjusting the PE For high resolution scan where electronic states needed to be identified, low PE of 20 eV is typically used in this work while for

survey scan, PE of 100 eV is used The effect of PE on resolution of Ag 3d peaks

acquired under a series of pass energies is shown in Fig 3.11 [14]

The second mode which not used in this work is Constant Retard Ration (CRR) This CRR mode run with constant ratio defined as KE/PE (set by user) Then electrons

are slowed down according to this ratio by scanning the lens system and constantly adjusting the PE of analyser For e.g., if CRR of 10 is set, 1000 eV electrons need to be

slowed down to PE= 100 eV in order to be detected This mode gives constant resolving power but analyser resolution varies with kinetic energies, which is not suitable if quantification is needed For e.g if the CRR is 10, the counts per second for 1000 eV