Faculteit der Natuurwetenschappen, Wiskunde en Informatica

Master Analytical Chemistry

Research on the surface interaction of metastable helium and ions for the removal of carbon layers on optical devices

Marij Stoevenbelt

Research on the surface interaction of metastable helium and ions for the removal of carbon layers on optical devices

Master research thesis by

M. Stoevenbelt

Analytical chemistry

Supervisor

Dr. W. Th. Kok

Daily supervisor

Ing. N. Koster

This report reflects the work of my master research at TNO in the department of semiconductor equipment in Delft, The Netherlands.

Table of content

Preface1

Abstract2

Samenvatting3

1.Introduction4

1.1Scientific question4

1.2Plasma4

1.3Plasma deposition5

1.3.1Chemical vapour deposition5

1.3.2Physical vapour deposition5

1.4Plasma cleaning5

1.5Plasma Analysis and Test Setup (PATS)6

1.6Metastable helium7

2.Materials and methods10

2.1Introduction10

2.2Langmuir probe10

2.3Optical spectrometer11

2.4Metastable helium sensor11

2.5Used substances and samples12

2.6Operating conditions12

2.6.1Langmuir probe12

2.6.2Spectrometer14

3.Experimental part15

3.1Literature search15

3.2Setting up the system15

3.2.1Set up15

3.2.2Results15

3.2.2.1Langmuir probe15

3.2.2.2The spectrometer17

3.3Getting to know the equipment17

3.3.1Set up17

3.3.2Results17

3.4Plasma dependence on power and pressure parameters19

3.4.1Set up19

3.4.2Results20

3.4.2.1Power dependence20

3.4.2.2Pressure dependence21

3.5Differences at different positions23

3.5.1Set up23

3.5.2Results24

3.6Difference in cleaning rate25

3.6.1Set up25

3.6.2Results26

3.7Sensor for helium metastables27

3.7.1Set up27

3.7.2Results27

3.8Metastable helium or not30

3.8.1Set up30

3.8.2Results32

4.Conclusions33

5.Recommendations34

6.Acknowledgements35

7.References36

Appendicesi

Appendix 1i

Appendix 2iv

Appendix 3vii

Appendix 4xi

Appendix 5xiv

List of abbreviations

Arargon

CVDchemical vapour deposition

CWcontinuous wave

EMelectromagnetic

EMCelectromagnetic compatibility

Hhydrogen

Hehelium

He*metastable helium

I-Vcurrent-voltage

N.A.not applicable

Neelectron density

Ni+ion density

O2oxygen

O2/Aroxygen/argon

PATSplasma analysis test setup

PVDphysical vapour deposition

RFradio frequency

SMIRPshielded microwave induced remote plasma

Teelectron temperature

Vffloating plasma potential

Vpplasma potential

Research on the surface interaction of metastable helium plasma Marij Stoevenbelt

Preface

Plasma surface cleaning is becoming more and more popular in the field of physics and industry (such as solar and semicon). In some industries it is very important to have materials with a clean surface and a high degree of cleanlyness. The materials that have to be clean are often very sensitive to heat load and electromagnetic fields as well. Plasma is an interesting option because of the possibility to create plasma at moderate temperatures and atmospheric pressure. For this purpose TNO developed the SMIRP source, the shielded microwave induced remote plasma. This microwave source is capable of creating plasma at a relatively low temperature.

The research question asked in this thesis is about plasma cleaning using helium gas. It is found experimentally that helium cleans just as well or better than for instance hydrogen. Not understood is how this works and why it works. The explanation seems to be that the cleaning is done by metastable helium, but this is never proven without doubt. Sputtering is proven to work but a more complex mechanism is assumed. The purpose of this research study is to investigate whether the cleaning is really done by metastable helium or if it is done by helium ions.

In order to investigate this, a small literature research is done first to get an overview of what is happening in this field of expertise. After the theoretical part it is needed to get to know the system and all the apparatus attached to it, including installing some of the equipment. Some introduction experiments will be done, to know how the equipment reacts as well as to see how different types of plasma behave inside the chamber.

To conclude the practical time in this internship, more complex experiments will be done as to determine whether the cleaning is done by metastable helium or by helium ions. Besides this, the optimal conditions such as the distance of the sample to the source are investigated.

In chapter 1 an introduction is given to the chemistry behind plasma cleaning, the plasma cleaner and the approach to the scientific question. After this an explanation of the equipment installed on the vacuum chamber is given in chapter 2. In chapter 3 the experimental setup and results are given. The last two chapters (4 and 5) contain some conclusions and remarks regarding future research.

Abstract

Being able to understand why mild helium plasma cleaning works as well as hydrogen plasma cleaning is important to TNO since it can help them to advice clients about the right cleaning tool and process for their equipment. Helium can play an important role in this since it is an inert and relatively safe gas, preventing a number of problems observed with other cleaning methods. The plasma analysis test setup (PATS) is equipped with a Langmuir probe and a spectrometer for the purpose of investigation. During the small literature search starting this internship it is found that many researchers observe the effects of mild helium cleaning, but they do not have an explanation for it.

At the end of the research period the conclusion had to be drawn that there was no answer yet on the original research question as to whether the plasma cleaning is done by metastable helium or by helium ions. However, some very interesting aspects of helium cleaning came up.

During the experiments with the spectrometer it became clear that there are definitely differences between the different types of plasma. In helium plasma the different transitions to the two metastable states showed differences as well for different circumstances, indicating that they are a good tool to tune the plasma in a way that one species dominates. This can be important in future cleaning rate experiments.

The cleaning rate experiments showed that using continuous wave plasma it is possible to clean samples even at a further distance from the source. They also showed that the sample facing away from the source is cleaned as well, this proves no line of site is necessary for plasma cleaning when using helium plasma. This is an indication that there are long living species present, which might be metastable helium atoms. Although the sensor as constructed to measure the metastables was not working properly, it showed potential for the future.

Samenvatting

Voor TNO is het belangrijk om te begrijpen waarom helium plasma net zo goed lijkt te reinigen als bijvoorbeeld waterstof plasma, zodat ze hun klanten kunnen adviseren over de methode die gebruikt moet worden. Helium plasma kan hier een belangrijke rol in spelen aangezien helium van zichzelf zachter is, waardoor minder schade aan het oppervlak van het te reinigen voorwerp ontstaat. De Plasma Analysis Test Setup (PATS) is uitgerust met een Langmuir probe en een spectrometer om plasma’s te kunnen observeren. Tijdens de kleine literatuurstudie die in de opdracht was verwerkt, is gebleken dat veel onderzoekers zien dat helium plasma inderdaad goed cleaned, maar niemand kan uitleggen waarom het zo is.

Uiteindelijk is aan het eind van de stage de conclusie getrokken dat er nog geen antwoord is gevonden op de oorspronkelijke onderzoeksvraag of helium reiniging nu met behulp van metastabielen of met ionen wordt gedaan. Er zijn echter wel een aantal interessante aspecten van helium reiniging ontdekt.

Tijdens spectrometer experimenten werd duidelijk dat er wel degelijk verschillen bestaan tussen de verschillende typen plasma. Bij helium plasma lieten de ratio’s voor de verschillende transities naar de twee metastabiele staten, verschillen zien bij veranderende omstandigheden. Dit impliceert dat er een goede manier is om plasma te optimaliseren zodat een van de twee metastabiele toestanden overheerst. Voor toekomstige experimenten kan dit van groot belang zijn.

De experimenten met betrekking tot de cleaning rates laten zien dat het bij gebruik van continuous wave mogelijk is om oppervlakken te reinigen die verder van de bron af staan. Ook hebben ze laten zien dat het mogelijk is om te reinigen zonder dat er een zichtlijn met de bron is. Dit is een indicatie dat er reactieve atomen of moleculen aanwezig zijn met een lange levensduur. De sensor, gebouwd om de hoeveelheid metastabielen te meten, werkte helaas niet zoals gewenst, maar liet wel mogelijkheden voor de toekomst zien.

1. Introduction

An overview of the theoretical background is provided on the principals and techniques used in this research project. The first part of this chapter covers the general information about plasma and the possibilities when using it. The second half of the chapter contains some information about the specific equipment used during the research study.

1.1 Scientific question

Plasma cleaning is usually done by reactive gases such as hydrogen, oxygen or by sputtering using inert gases such as argon. An experiment was done using helium plasma and although not expected it was found that it had a high cleaning rate for samples coated with carbon. Investigation revealed that helium is capable of cleaning surfaces as well as hydrogen, but with less harmful side effects. Sputtering is not expected since the plasma has a low energy, but, then what is causing the cleaning of the surface? The theory is that neutral helium atoms with an high internal energy (so-called metastables) are capable of forming short lasting volatile species with the contaminants on the surface or breaking the bonds connecting it to the surface. The advantage of soft cleaning is that there is less damage to the surface. The mechanism behind the helium cleaning is not yet fully understood and needs more research. The search for answers to this question is described in this report.

1.2 Plasma

Plasma is considered the fourth state of matter (see figure 1). Plasma is a partially ionised gas, in which gas interactions are dominated by charged species like ions, anions and electrons. Like gas, plasma does not have a definite shape or a definite volume. The presence of a non-negligible number of charged particles makes the plasma electrically conducting so it responds strongly to electromagnetic fields. Depending on the type of atoms, the proportion between the number of atoms ionised and neutral and the energy of the particles, there are many different types of plasma; each having their own properties.[endnoteRef:1] Plasma can be created in many different ways, as long as there is enough energy added to the gas. [1: http://en.wikipedia.org/wiki/Plasma_(physics), visited November 14th 2009.]

(SolidLiquidGasPlasmaAdd energyH2O (s)H2O (l)H2O (g)H,H2,H+,e-,H2-O,O2,O3,O-,O2-H2O)

Figure 1 States of matter

Figure 1 shows how states of matter are reached by adding enough energy, finally resulting in plasma. When it reaches plasma state it is transformed into a wide variety of species like molecules, atoms, ions or anions having high internal energies. Plasma can be used for cleaning as well as for deposition. In section 1.3 and 1.4 both processes will be addressed.

1.3 Plasma deposition

Plasma deposition involves the formation of a thin coating at the substrate surface. Plasma deposition can be done in two different processes; chemical vapour deposition (CVD), involving a reactive gas binding molecules to the surface, and sputtering/physical vapour deposition (PVD), which involves the sputtering of the target material and condensing it on a surface.

1.3.1 Chemical vapour deposition

In CVD the chemical that is to be deposited is added to the gas used to create the plasma. During plasma activity radicals and ions of the chemical are created and are directed towards the target surface. At the surface the radicals and ions bind to the chemicals present in the surface layer.[endnoteRef:2] [2: A.C. Jones & M.L. Hitchman, Chemical Vapour Deposition; Precursors, Processes and Applications, The Royal Society of Chemistry, Cambridge, 2009.]

1.3.2 Physical vapour deposition

The other plasma deposition method is PVD that mainly consists of deposition technologies in which material is released from a source and transferred to the substrate. The two most important technologies are evaporation and sputtering.[endnoteRef:3] [3: D.M. Mattox, Handbook of physical vapor deposition (PVD) processing, Film formation, adhesion, Surface preparation and contamination control, Noyes publications, New York, 1998.]

Using evaporation, the sample is placed inside a vacuum chamber after which the temperature is increased until the point is reached where it starts to boil. The molecules which are released condense on all other surfaces inside the chamber.

In case of sputtering, the material is released from the source at much lower temperature than evaporation. The substrate is placed in a vacuum chamber with the source material, named a target, and plasma is activated. The ions created are accelerated towards the target, causing atoms of the source material to be ejected and condense on all surfaces including the substrate.

1.4 Plasma cleaning

Plasma cleaning involves the removal of impurities and contaminants from surfaces through the use of energetic plasma created from gaseous species. Plasma cleaning can be divided in two processes, chemical cleaning and physical cleaning. In case of a reactive gas, chemical reactions are possible at the surface. When using an inert gas to create plasma, the excited ions can collide with a surface and remove a small amount of material, this is mechanical cleaning .[endnoteRef:4] Different types of plasma cleaning are pictured in figure 2 using hydrogen (H) as the plasma gas. [4: http://www.yieldengineering.com/default.asp?page=264, visited November 13th 2009.]

(VolatilisationSoft sputteringSputtering)

Figure 2 Different types of plasma cleaning processes in H2 plasma environment

There are different types of processes to perform plasma cleaning:

· Volatilisation; A radical having little kinetic energy can arrive at the surface. The radical can bond to the element present at the surface. This can lead to the formation of a new compound such as methane in figure 2. The new compound is then released from the surface. This is a very mild method of removing contamination.

· Soft (chemical) sputtering; A radical or ion can have enough energy to break bonds of the contaminant with the surface. When the radical arrives at the surface, the transferred energy is used to remove the bond of a contaminant and the surface. This is still a mild method of plasma cleaning.

· Sputtering; An ion having a high kinetic energy can knock other atoms off the surface, including atoms from the substrate. This is a very aggressive cleaning method which can lead to damage of the surface.

Helium is an inert gas, so it should have no abilities to bond to other atoms. The cleaning processed are therefore expected to be soft sputtering or sputtering. In earlier experiments however, it is noted that helium is also capable of soft cleaning, which could indicate volatilisation. Volatilisation would mean that helium is capable of making a short living bond to another atom. This could be possible with metastable helium.

1.5 Plasma Analysis and Test Setup (PATS)

The Plasma Analysis and Test Setup (PATS) system (see figure 3) is a dedicated plasma cleaner for research on plasma processes and cleaning recipes or other applications involving low pressure plasmas. The main focus is aimed at cleaning delicate surfaces like mirrors, electronics and thin layers. For this a special type of plasma called Shielded Microwave Induced Remote Plasma (SMIRP) is developed at TNO. SMIRP has the great advantage that is relatively cold and has no electromagnetic (EM) fields outside the plasma formation region; hence it is possible to clean electromagnetic compatibility (EMC) sensitive components. The cleaning is performed by volatilisation and soft sputtering and not, in contrast with other types of cleaning, by sputtering. Depending on the contaminant to be cleaned an appropriate gas can be chosen.

At this moment 4 microwave (SMIRP) sources and a radio frequency (RF) source are mounted onto the PATS. The microwave sources can be operated in continuous wave as well as pulsed mode. There are cages put over the microwave sources (as a type of Faraday cage) to keep the plasma creation at one place where it can be observed and manipulated. During continuous mode the sources are continuously coupling power into the gas. When pulsed mode is used the effective power introduced is lower, depending on the duty cycle (Tpulse/period) with which the microwave power is coupled into the gas. During the experiments only one source was used to limit the amount of variables.

Figure 3 The PATS system.

The PATS is equipped with analytical equipment like a Langmuir probe, UV/VIS spectrometer, mass spectrometer and an ellipsometer. These are used to study the plasma processes and surface interactions. Besides cleaning, the PATS can also be used for surface modification of polymers and studies involving sterilization and disinfection of surfaces for medical, food and pharmaceutical applications. The latter possibilities are beyond the scope of this thesis and will not be explained here.

1.6 Metastable helium

Since helium is an inert mono-atomic gas, collisions between atoms and electrons are perfectly elastic. When helium is partially ionised, some electrons are not bound to the nucleus, resulting in a plasma in which the free charge carriers (electrons and ions) create a good electrically conductive environment. If electrons are then accelerated under a suitable potential, the velocity (and thus, the energy) of the electrons continually increase. If an atom received enough energy to become excited it is possible that this atom relaxes to a state from where it is not possible to relax to the ground state directly, a metastable radical is formed . The metastables can have energies of 19.8 or 20.6 electron volts.[endnoteRef:5] These energies depend on the spin of the two electrons in the excited state. When they are oriented antiparallel (singlet or para state) the energy is 20.6 eV and when the electrons are oriented parallel (triplet or ortho state) the energy is 19.8 eV. [5: http://www.chromatography-online.org/topics/helium.html, visited November 13th 2009.]

The theoretical lifetime for metastable helium is quite different from the normal lifetime of excited species in a plasma. Normally an excited species has a lifetime in the order of 10-9 seconds, the triplet state has a lifetime of 2*10-2 seconds and the singlet state has a lifetime of 4200 seconds[endnoteRef:6]. Taking into account this long lifetime and the long mean free path (the average distance covered by a particle between successive impacts) of helium in a vacuum the reactivity of helium plasma might be explained. When the helium atoms cannot notice each other within the chamber they are likely to encounter something else first. This means there is a relatively large amount of metastables present. [6: W. Sesselmann, B, Woratschek, J. Kuppers, G. Ertl, H. Haberland, Interaction of metastable noble gas atoms with transition-metal surfaces: Resonance ionization and auger neutralization, Physical review B, 1987, 35(4), 1547-1559.]

Helium can be considered to be metastable when either the spin of the electrons prevents a relaxation to the groundstate or when the dipole is not capable of relaxing to the ground state.

Metastable excited helium states have a long lifetime since the transition to the ground state is a quantum mechanically forbidden transition by dipole selection rules. Metastable helium has to be excited to a higher energy level, or have enough internal energy to reverse a spin orbit, to be able to relax to the ground state.[endnoteRef:7] Different helium states with their energy levels are given in figure 4. The red circles represent the metastable states, since from there only indirect, in the case of para helium (left side), or no relaxation, in case of the ortho helium (right side), to the ground state is possible. The most important emission lines for this project are some of the singlet and triplet lines. The following transitions are the ones of the most interest; [7: J. H. Hetherington, The Spectrum of Helium and Calcium, E. Lansing MI, 2000, 1, 16.]

Singlets;3p → 2s 502 nm

3d → 2p 668 nm

3s → 2p 728 nm

Triplets;3p → 2s 389 nm

3d → 2p 588 nm

32 → 2p 707 nm

The 3p → 2s transitions are the most interesting since it involves a direct relaxation to the metastable states. The measurement of these emission lines gives information about the quantity of metastable helium present.

Also the ratios of the same singlet and triplet transition are important since they can give a measure for the amount of singlets present versus the amount of triplets present. In the end the metastables wanted are those who express the largest cleaning rates and the lowest heat load or damage to the surface. The ratios between singlet and triplet transitions can help to determine the optimal cleaning settings.

Figure 4 helium energy levels and transitions[endnoteRef:8] [8: J.J. Brehm, W.J. Mullin, Introduction to the structure of matter, a course in modern physics, John Wiley & Sons, New York, 1989.]

The states with a red circle are the metastable states. From here no direct relaxation to the ground state is possible. The transitions from 2p → 2s are not measurable with the spectrometer used during this research although they could give extra information about singlet life time.

2. Materials and methods2.1 Introduction

As described earlier, a Langmuir probe and an optical spectrometer are used during the research. Added to this standard analytical equipment is a sensor to detect the amount of metastable helium. All equipment is introduced first, after which a short part follows about the gases, substances and samples used. The last part of the chapter contains the primary settings for all equipment.

2.2 Langmuir probe

The Langmuir probe (figure 5) can be used to determine a number of characteristics of a certain plasma. The tip of the probe is inserted into the plasma and a varying potential is applied to this tip. Electrons and ions will be either repelled or attracted to the tip, resulting in a typical I-V (current-voltage) characteristic. Using the obtained I-V characteristic, the ion density, electron density, electron temperature, the ion flux, floating potential and the plasma potential can be calculated. Langmuir probes are routinely used to determine the plasma parameters in areas as diverse as: low pressure plasmas for materials processing, the design of ion sources and new plasma chambers, and edge plasmas in fusion devices.

Figure 5 View of a Langmuir probe.

In the I-V curve three different regions can be distinguished, as can be seen from figure 6. At a high negative voltage, all electrons are effectively repelled and the probe attracts only ions, so this is called the ion saturation region. At a high positive voltage, ions are repelled effectively and only electrons are attracted by the probe, so this is called the electron saturation region. The region in between is called the electron saturation region.[endnoteRef:9] [9: Hiden analytical, Plasma Diagnostics Introduction to Langmuir Probes, Technical Information Sheet 531.]

In the Plasma Diagnostics Introduction to Langmuir Probes, Technical Information Sheet 5318 formulas are given which the software program uses to calculate the characteristics of the plasma.

Figure 6 Typical I-V current

2.3 Optical spectrometer

With optical spectroscopy the wavelengths of emitted light by the transitions of atoms, molecules or ions from an excited state to a lower energy state can be examined. Each element emits and absorbs a characteristic set of discrete wavelengths according to changes in its electronic structure, by observing these wavelengths the elemental composition of the gas or plasma can be determined.

Using an optical spectrometer enables observation of the emission lines which are specific for helium ions or (metastable) helium atoms. Using these results, an estimation can be made as to which states are dominant in the plasma.

By coupling energy into a gas, some of the atoms or molecules become excited. Excited states have a certain lifetime, after which the atom will relax to a lower energy state. When they do, they emit a photon at a unique wavelength, corresponding to the energy differences, ΔE = hν = hc/λ, h=Planck constant, ν= frequency, c= speed of light and λ= wavelength. The brightness of the emission lines can give a lot of information about the abundance of the amount of atoms or ions in a certain state.

2.4 Metastable helium sensor

There are several possibilities to measure the amount of metastable helium in a plasma but according to Miura and Hopwood[endnoteRef:10] the most practical one is to build a metastable helium (He*) sensor (see figure 7). The measurable current released by the plasma is composed of several components, the electron current (Ie), the ion current (Ii), the photon current (Iᵞ) and the metastable current (IHe*). In order for the sensor to work, a shielding mechanism has to be applied to shield the sensor from all the currents except for the metastable current which leads to the following Itot= IHe*. This means only neutrals are captured on the surface of the sensor, composed of both atoms and metastable helium atoms. When the sensor is not shielded, the signal is dominated by ions and electrons hitting the surface of the sensor. In figure 7 the mechanism of the sensor is depicted. It is fabricated by creating an outer probe of 1x1 cm aluminium sheet with a hole in the centre. The hole is 3 mm in diameter. The backside of the outer probe is covered with kapton tape which does not conduct electricity. The inner probe is made of stainless steel and is only exposed to the environment through the hole in the outer probe, the backside is again covered with kapton tape to prevent the conduction of electricity. [10: Naoto Miura & Jeffrey Hopwood, Metastable helium density probe for remote plasmas, Review of scientific instruments, 2009, 80, 113502-1 until 113502-5.]

Potentials can be applied to shield the probe from electrons and ions. A negative voltage is put on the outer probe, shielding the sensor from positive ions. A positive voltage is applied to the inner probe, thus shielding the sensor from electrons. He* has enough internal energy to free an electron from the surface, so by measuring the resulting current on the inner probe the amount of He* can be measured.

Figure 7 Schematic overview of the helium metastable sensor10

2.5 Used substances and samples

For plasma analysis the need for chemicals is very low. Cleaning using a ultrasonic bath is needed because everything mounted into the chamber has to be very clean. The parts that need cleaning are submerged in ethanol and cleaned. Also alcoholic cleaning tissues (70 % isopropyl alcohol) are used for this purpose.

The gases used to create the plasma (hydrogen, oxygen, argon and helium) are 5.0 (99.999%) pure. The gas flows are kept constant by the equipment controlling the vacuum chamber.

The samples used during the course of these experiments are round quartz glasses of 1 inch diameter. The samples are coated on one side with a carbon layer, either 20 nm or 125 nm thick.

2.6 Operating conditions2.6.1 Langmuir probe

The Langmuir probe used is a commercially available probe purchased from Hiden Analytical. The default settings for the Langmuir probe are mainly the probe settings. The default settings are depicted in figure 8 and 9. followed by an explanation.

Figure 8 Langmuir sweep settings

The settings in figure 8 can be explained as follows;

Potential ramp; the voltage is swept from -50V to +50V, using a 0.1V interval between each step. Range; the gain is set to 100 mA in order to prevent a shutdown when a strong peak is observed. The probe impedance is set on 4.9 Ohms, which limits the chance of electrical break through. The tip is cleaned before each new scan by means of ion bombarding by applying 100V to it for 200 ms. Since we use continuous wave mode, no trigger signal is used and gate timing is therefore disabled.

Figure 9 Probe positioning of the Langmuir probe

The settings in figure 9 are relatively simple, the standard probe position (park position) is 35 cm from the starting position. This means the probe is inserted in the chamber for about 20 cm.

2.6.2 Spectrometer

The initial spectrometer used during the research period is a commercially available Andor Mechelle (Lot Oriel) spectrometer with an ICCD detector attached to it, type ME5000- iStar DH734. This detector has a spectral range of 200 to 975 nm with a resolution of 4 pixels at full width half max. The Andor Mechelle spectrograph is based on a grating principal and patented optical design, which gives extremely low cross-talk, equally spaced order separation and maximum resolution. The spectrograph offers a simultaneously recorded wavelength range from UV to NIR with very high resolution and no overlapping wavelengths, thus enabling for high resolution real time measurements. The accompanying software is capable of detecting which elements are present is the spectrum.[endnoteRef:11] The spectrometer camera can be cooled. [11: Andor Technology, Mechelle manual, users guide, 2008.]

A second type of spectrometer used during the experiments is a commercially available CCD minispectrometer from Hamamatsu, model TM-UV/VIS C10082CAH. This type has a spectral range of 200 to 800 nm with a resolution of 1 nm.

The spectrometer is capable of scanning in dark and in reference mode, so the spectra can be corrected for background noise or used to compare different spectra.

The spectrometer used is a very simple model only capable of taking a scan and saving it in excel extension. This file can then be opened in excel to be studied further.[endnoteRef:12] [12: Hamamatsu Photonics, HARDWARE INSTRUCTION MANUAL, K29-B60419B, version 1.2.]

3. Experimental part3.1 Literature search

Before the practical part started a small literature search was done. The overall conclusion derived from the found literature is described in this section. The full result of the literature study is described in appendix 1.

The overall conclusion that can be drawn from the papers found is that a lot of researchers observed the effect of helium cleaning, but nobody has determined the exact nature of the mechanism involved. However there are some other useful conclusions to be drawn from the articles, which are given here.

One article suggested that the cleaning by helium is done by metastable helium radicals reacting with the surface, thereby reducing the surface impurities.

When helium plasma is ignited in a fusion reactor (which is a high energetic plasma) the following is observed; Hydrogen rapidly appears when the discharge is started and slowly falls off, disappearing completely when the discharge is stopped. The main source for this hydrogen suggested by the author is adsorbed water but according to some other authors this could also be released from the vessel wall. The following is also observed; Hydrogen and helium are both present during plasma cleaning, even if they are not present in the gas used to produce the plasma.

Accumulation of helium by the vessel wall is influenced by the wall material itself. Graphite holds less helium than stainless steel, resulting in a lower release during plasma treatment.

No correlation is found between helium pressure within a stainless steel vessel and the gas species of the main plasma. During experiments a net loss of helium particles is observed and the main source of helium desorption is the stainless steel wall.

Non-oxygen plasma leads to higher oxygen and nitrogen levels on the surface being cleaned. Storage has distinct effects on the surface composition, which is depending on the plasma type used.

3.2 Setting up the system3.2.1 Set up

Before the plasma analysis test setup (PATS) could be used to observe plasma and its characteristics, the machine and all the equipment attached to it had to be functioning correctly. Since the system was new, most of the equipment needed to be installed and configured for the first time. The optimal settings had to be determined during the experiments.

3.2.2 Results3.2.2.1 Langmuir probe

While setting up the system, some difficulties arose. First of all the Langmuir probe was not working properly, since the probe could not be moved into the vacuum chamber. After contact with the supplier, the problem was solved. However, the results of the first measurements were not as expected. Instead of a curve looking like the curve in figure 6, a deviating curve as shown in figure 10 was observed.

Figure 10 The first Langmuir results

Since this is not the best curve to perform calculations on, improvements and/or optimalisations were needed. Cleaning of the tip before each measurement helped to improve the signal by a factor 2. This was not enough however, since the noise scaled along with the signal. After doing a fourier transformation (appendix 2) it became clear that there was another signal interfering with the measurement. There were a few other possible sources for the problems encountered. These problems were addressed one at a time to determine the cause of the bad signal. The possible causes were the power supply, the plasma source and the type of plasma creation.

The power supply did not seem to be the problem since getting the power from another source only led to a worse signal (appendix 2).

Using an RF source instead of a shielded microwave source did not help either, which we expected, since the RF source is not shielded.

When trying to create plasma using the continuous wave mode of the microwave source instead of the pulsed mode, a definite improvement of the signal was observed. The problem occurred during pulsed mode of the microwave source. Measuring the plasma using the continuous wave mode resulted in a scan looking like a typical curve (see figure 11). The figures of all the experiments are described in appendix 2.

Figure 11 I/V curve using the continuous wave mode

3.2.2.2 The spectrometer

The spectrometer needed to be attached to the vacuum chamber as well. The apparatus had to be calibrated first. The Andor spectrometer showed no response at all when measurements were done. After some testing, the suspicion arose that the image intensifier was not working properly and the spectrometer had to be sent to the manufacturer. A spectrometer was needed, so another spectrometer had to be arranged. A simple Hamamatsu spectrometer from another department was borrowed for the duration of the repairs.

3.3 Getting to know the equipment3.3.1 Set up

The first experiment was also meant to get a better understanding of the behaviour of different types of plasma. The plasma was activated first, to pre-clean the vacuum chamber. After the conditioning, the plasma was ignited again and scans were made using the Langmuir probe and the spectrometer. The Langmuir probe provided information about the electron temperature, the plasma voltage etcetera and the spectrometer provided information about the gas species present in the chamber by means of the emission lines.

This experiment was performed using helium, hydrogen and oxygen/argon (50/50) plasma. Afterwards the results were compared for the different types of plasma.

3.3.2 Results

Using the Langmuir probe and the spectrometer, measurements were done with helium, hydrogen and oxygen/ argon plasmas. The measurements were performed with the standard operating mode. This meant an input power of 350 W and a pressure inside the chamber of 0,5 mbar. A Langmuir probe graph taken for helium plasma is shown in figure 12, and a spectrometer graph for helium plasma is shown in figure 13. The full results including hydrogen and oxygen/argon plasma are described in appendix 3.

Figure 12 Langmuir probe signal helium plasma

Figure 13 Spectrometer results from the helium plasma

In figure 13, among others, the peaks resulting from direct relaxation to metastable helium states are visible. These lines are at 502 nm for the singlets and 389 nm for the triplets. The other transitions for the singlets are visible at 668 nm and 728 nm, and the triplet transitions are visible at 588 nm and 707 nm (see figure 4).

3.4 Plasma dependence on power and pressure parameters3.4.1 Set up

The pressure inside the chamber and the power input into the system could affect the electron density inside the chamber. To test this, the plasma parameters and the emission lines were monitored using the Langmuir probe and the spectrometer at different input powers and pressures. This experiment was done using helium, hydrogen and oxygen/argon plasma. The input powers used were 350, 500 and 750 W at continuous wave. The pressures used were 0,2, 0,5 and 1 mbar. The standard settings were 350W input power at a pressure of 0,5 mbar.

When the input power was changed, the pressure was kept constant and when the pressure was changed, the input power was kept constant. The electron density for the different types of plasma was measured using the Langmuir probe. The Langmuir probe was positioned about 20 cm inside the chamber above the plasma source used for the experiments. The fibre from the spectrometer was placed in front of the antenna of the source, thereby observing the plasma through the middle of the plasma cloud.

Using the Langmuir probe not only the electron density was determined but also the electron temperature, the plasma potential and the ion density. It is also of importance to see if there are differences between the characteristics of the plasma when either the input power or the pressure is changed.

Using the spectrometer any possible changes in emission lines were observed for the different types of plasma at different pressures and input powers. The figure regarding He plasma at standard settings is depicted in figure 14, but for means of comparison only tables with peak heights at each setting are described in this chapter.

3.4.2 Results3.4.2.1 Power dependence

The most important characteristics are summarised in this section, the rest is described in appendix 3. The results for the plasma potential (Vp), electron temperature (Te), ion density (Ni+) and the electron density (Ne) are depicted in figure 14 a, b, c and d respectively for the different input powers.

Figure 14 a) plasma potential, b) electron temperature, c) ion density, d)electron density.

As can be seen from figure 14 the trends are not the same for each type of plasma. The most interesting result is the trend for the plasma potential (Vp), since the trend for hydrogen (green trace) and oxygen/argon (blue trace) is an increase while helium (red trace) stays more or less the same as shown in figure 14a. This could indicate that helium plasma is saturated with excited species at a lower power input than hydrogen and oxygen/argon plasma. The electron temperature (Te in figure 14b) for oxygen/argon plasma follows a decreasing trend while helium plasma follows the same trend as for the plasma potential. This indicates a difference in behaviour under different circumstances.

Regarding the ion density (Ni+) and electron density (Ne) shown in figure 14c and d respectively, the same contradiction can be noted. Where the trend for oxygen/argon contradict each other, the values for helium stay more or less the same and seem to be highly similar.

When the I-V characteristics could not be determined from the curve a value is missing.

When zoomed in more peaks are visible in the spectrum obtained by the spectrometer, but they are not significant for the reviewing of the results so they are not named in this report. The peak heights of the most intense peaks are given in appendix 3. The most important results however, are the singlet and triplet metastable lines, and from those the 502 nm and 389 nm are the most interesting ones. The results for the singlet and triplet transitions are depicted in table 1, including the ratio of the singlet and triplet line set associated with the same transition. These ratios might give an insight as to the metastables present in the plasma at different input powers.

Table 1 Spectrometer results helium plasma triplet (t) vs singlet (s) emission, results in counts

Emission line

350 W

500 W

700 W

502 nm (s)

8528

9840

11081

389 nm (t)

24273

26684

29410

Ratio

2,85

2,71

2,65

668 nm (s)

15198

18057

20580

588 nm (t)

50466

58643

64668

Ratio

3,32

3,25

3,14

728 nm (s)

2155

2367

2627

707 nm (t)

10965

11913

13154

Ratio

5,09

5,03

5,01

The ratios are calculated by dividing the peak height of the triplet by the peak height of the singlet. When comparing the ratios for the direct relaxation to the metastable state (502 and 389 nm, 3p → 2s transition), a decrease is visible for increasing power input. This decrease is also visible for the 3d → 2p transition (668 and 588 nm), but is not so obvious for the 3s → 2p transition (728 and 707 nm). This means there is a difference in equilibrium between singlet and triplet state at different input powers, resulting in an ability to tune the wanted plasma conditions.

Overall there is an increase in absolute signal when increasing the power input. This is also visible for hydrogen and oxygen/argon plasma. The increase in signal is similar for all plasmas.

Combining figure 15 and table 1 results in the following observations. The plasma potential for helium remains the same as well as the ion density, but the electron density is decreasing. The singlet emission lines increase faster than the triplet emission lines with increasing power input. For a higher amount of the preferred metastables a higher input power is required.

3.4.2.2 Pressure dependence

The most important characteristics are summarised in this section, the rest is described in appendix 3. The results for the plasma potential (Vp), electron temperature (Te), ion density (Ni+) and the electron density (Ne) are depicted in figure 15 a, b, c and d respectively for the different pressures.

Figure 15 a) plasma potential, b) electron temperature, c) ion density, d) electron density

As can be seen from figure 15 the trends are not the same for each type of plasma. The most interesting result are the trends for helium plasma (red trace), since the trend is decreasing for all parameters with increasing pressure (including the ones depicted in appendix 3). This indicates that the characteristics of helium plasma change when changing the pressure inside the vacuum chamber. This trend is not visible for the other types of plasma. This indicates that the behaviour observed with helium plasma is unique for helium. This also implies an ability to create a plasma with certain characteristics.

When the I-V characteristics could not be determined from the curve a value is missing.

When zoomed in more peaks are visible in the spectrum obtained with the spectrometer, but they are not significant for the reviewing of the results so they are not named in this report. The peak heights of the most intense peaks are given in appendix 3. The most important results however, are the singlet and triplet metastable lines. These results are given in table 2, including the ratio of the complementing singlet and triplet line. These ratios might give an insight as to the metastables present in the plasma at different pressures.

Table 2 Spectrometer results helium plasma triplet (t) vs singlet (s) emission, results in counts

Emission line

0,2 mbar

0,5 mbar

1 mbar

502 nm (s)

7187

8528

8752

389 nm (t)

19281

24273

25943

Ratio

2,68

2,85

2,96

668 nm (s)

8510

15198

17879

588 nm (t)

30295

50465

61947

Ratio

3,56

3,32

3,46

728 nm (s)

1990

2155

2292

707 nm (t)

8243

10965

12776

Ratio

4,14

5,09

5,57

The ratios are calculated by dividing the peak height of the triplet by the peak height of the singlet. For the transition directly to the metastable state (502 and 389 nm) an increase in ratio is visible, indicating a faster increase of the triplet lines comparing to the singlet transition. This implies a higher amount of relatively short living metastables at a higher pressure.

For helium plasma there is an overall increase in absolute signal when increasing the pressure. For hydrogen and oxygen/argon plasma however, there is a decrease in signal with increasing pressure. This is another indicator of different behaviour of different types of plasma.

Combining figure 16 and table 2 results in the following observation. The electron density as well as the relative amount of singlets decrease with increasing pressure. For better metastable results a lower pressure is required.

3.5 Differences at different positions3.5.1 Set up

There can be a difference in electron density, temperature and ion density and therefore in activity at different positions inside the Faraday cage and the whole chamber. To check this, the Langmuir probe was inserted into the Faraday cage or placed above it as depicted in figure 16, between brackets is the distance to the microwave source. At each position, Langmuir measurements were done to evaluate differences. Using the spectrometer, at the same positions measurements were done to observe (if any) notable changes in the emissions at the respective positions.

1. at half cage height above the cage (14 cm)

2. just above the cage (10,5 cm)

3. in the upper part of the cage (8 cm)

4. at half height of the cage (5 cm)

Figure 16 The measurement positions around the Faraday cage

3.5.2 Results

The line of sight for the spectrometer is blocked on the other side of the cage, this to ensure an ending line of sight. The results in this paragraph only contain the results for helium plasma. Full results for all plasmas are given in appendix 4. With the Langmuir probe only the first three positions are measured, since the last two positions are outside the plasma creation area and are expected to give similar results. Using the spectrometer the 3rd position is not measured. Instead a measurement is done through the antenna of the microwave source, giving a line of sight through the quartz glass separating the vacuum from the atmospheric pressure. In figure 17 and table 3 the results from the measurements using the spectrometer and the Langmuir probe are depicted. The results for the spectrometer are noted as the ratios between singlet and triplet lines to observe any changes in composition of the plasma.

Figure 17 a) plasma potential, b) electron temperature, c) ion density, d) electron density.

As can be seen from figure 18 the trends are not the same for each type of plasma. The most interesting result is the trend helium plasma (red trace), since the trend for all parameters (including those in appendix 4) is decreasing. This implies a change in plasma characteristics when moving away from the plasma creation area. These characteristics change in a different manner for different types of plasma.

The ion density decreases for all plasma types with increasing distance from the plasma creation area.

Table 3 Spectrometer results helium plasma triplet (t) vs singlet (s) emission, results in counts

Emission line

Through the antenna

middle of the cage

Just outside the cage

Half height above the cage

502 nm (s)

64543

21900

13189

9656

389 nm (t)

64433

47838

15542

10948

Ratio

1,00

2,18

1,18

1,13

668 nm (s)

64418

45977

19918

14922

588 nm (t)

64300

64503

47290

34913

Ratio

1,00

1,40

2,37

2,34

728 nm (s)

31100

10852

7328

6362

707 nm (t)

64436

53771

20891

16063

Ratio

2,07

4,95

2,85

2,52

The results from the measurement through the antenna are not compared to the other positions, since these measurements are from a different line of sight and therefore not directly comparable to the other positions.

The ratio is decreasing for the transition directly to the metastable state when moving away from the plasma creation area. This means that when plasma cannot be created anymore, the triplet state decay faster than the singlet states. This leads to the observation that the lifetime of the singlet metastable state is longer than for the triplet metastable state.

The ratios for the indirect relaxations to the metastable state contradict each other so no conclusive observations can be made from these data. Inside the cage the ratio for the 3s → 2p transition (728 and 707 nm) is higher than the ratio for the 3d → 2p transition.

These results indicate a difference in characteristics of the plasma at different positions inside the chamber.

There seems to be a step in ratios when measuring inside and outside the cage. Outside the cage the ratios seem to remain more or less the same, while all absolute intensities decrease. The intensities for the emission lines of hydrogen and oxygen/argon plasma (appendix 4) decrease outside the cage as well, this decrease is much steeper than for helium plasma however. This indicates a difference in decay within the vacuum chamber for different types of plasma, which might be related to the metastables present in helium plasma.

3.6 Difference in cleaning rate3.6.1 Set up

There have been cleaning rate experiments performed on another device, but no cleaning rate experiments have been done in the PATS yet. Cleaning rates are a good indicator on how efficient the cleaning process is. To observe the cleaning rates and also the differences in cleaning related to the orientation of the sample, two samples were placed inside the chamber at the same time. The samples were on one side covered with a carbonlayer and the other side was plane quartz.

One sample was placed facing towards the source and one sample was placed facing away from the source. There were a few reasons for choosing these orientations. A first reason was the long lifetime of metastable helium, which allows the metastables to fly through the chamber long enough to reach the backside of the sample. A second reason was that metastable helium can only relax to the ground state by colliding with another atom, that is not helium. So, although the mean free path (the average distance covered by a particle between successive impacts) is long, most of the collisions metastable helium (He*) has will not result in a high loss of He* and will only deflect metastables in its path. This means that a sample facing away from the source can be reached by the metastables through collisions with other helium atoms. It was expected that the sample facing towards the source is cleaning faster than the sample facing away from the source.

3.6.2 Results

The first cleaning rate experiment on the PATS is conducted using two quartz plates where one side is coated with 125 nm carbon. The microwave source is operated at 350W input power in continuous wave (CW) mode, at 0,5 mbar pressure. The samples, which have a round shape, are positioned about 1 cm above the cage to which the plasma creation area is limited. In figure 18 the placement of the samples can be observed.

Figure 18 The placement of the carbon samples relative to the source

The result was a cleaning rate of 0,66 nm/hour for the sample facing the source, for the sample facing away from the source a rate of 1,34 nm/hour was observed (see table 3). The result is not as expected so the experiment is repeated with samples containing 20 nm of carbon using two different settings and at two different distances from the cage. First to repeat the initial experiment and another setting to clean as is normally done by hydrogen cleaning: 1000 W input power at a 10% duty cycle, which means the plasma source is operational 10% of the time and the average power input is 100 W. The results are given in table 4.

Table 4 Cleaning rate results in nm/hour

settings

Carbon layer (nm)

Distance to the cage (cm)

facing towards the source

facing away from the source

CW, 350 W

125

20

20

1

1

10

0,66

5,21

2,65

1,34

1,89

2,32

pulsed, 1000 W, 10% duty

20

20

1

10

1,01

6,96

The first repeating (with the same settings but with a 20 nm sample) experiment using continuous wave (CW) shows higher cleaning rates when facing towards from the source, which is in contradiction to the first experiment. The second experiment at a greater distance shows comparable cleaning on both sides.

The samples exposed to the pulsed plasma also show differences. The 1 cm, pulsed plasma exposure again showed considerable higher cleaning rates for the sample facing away from the source, while for the sample further away no cleaning rate was detected. This result is puzzling, however it showed that reactive species can reach the backside of these samples. The reason why the surface facing the source is cleaned with a lower cleaning rate is still not understood.

At a distance of 10 cm a significantly lower cleaning rate is observed for pulsed plasma compared to continuous wave plasma. The question that comes from this result is the following; How far do ions and metastables travel with a certain type of plasma? These results might indicate that the travelling distance in a pulsed plasma is much lower than in a continuous wave plasma.

3.7 Sensor for helium metastables3.7.1 Set up

The sensor as described in section 2.4 has been constructed. The distance of the sensor relative to the cage containing the plasma creation area can be varied from 0,5 cm to 45 cm away from the cage. The maximum distance is limited by the wall of the chamber. The plasma type and pressure were different than those in the article of Miura and Hopwood10, so this experiment was used to search for the optimal settings by which Ii=0, Ie=0, Iᵞ=0 and Itot= IHe*. The potentials put on the probes were +25V on the outer probe and -35V on the inner probe in an attempt to tune the photon current (Iᵞ). At a certain distance from the source the signal should not change anymore, since there only the photon current should be detected. The plasma creation area is then so far away that no ions etc. should be able to reach the probe surface. The position of the sensor relative to the source was also varied from looking directly at the source to completely looking away from it and facing the side to reduce the amount of photons reaching the sensor surface. According to Miura and Hopwood the signal should become a flat line when moved far enough away from the source.

The second part of the experiment was to change the voltages on the inner and outer probe in order to sweep the settings and observe the differences. According to Miura and Hopwood10, there should be a limit as to where the values change significantly. The potential on the outer probe was changed from 0 V until +50V and the potential on the inner probe is changed from -50 V to +15V. According to Miura and Hopwood there will be a flat part when the inner potential is below -10V in combination with a positive potential on the outer probe since in that region all ions and electrons are repelled effectively.

3.7.2 Results

During this research it is not yet accomplished to make Itot equal to IHe*. The results that are accomplished are described in this section. The measurements for the first part of the experiment are done with the sensor facing the source, facing away from the source and facing sideways relative to the source. The positions are depicted in figure 19.

1. facing towards and away from the source

2. facing sideways

Figure 19 The position of the sensor relative to the cage

When the sensor is facing away from the source or sideways there is no direct line of sight, thus preventing any direct light on the sensor. The distance to the source is also a little larger when facing sideways, but the measurements are taken at the same relative point for every situation. The graphical results for the measurements are given in figure 20, the full results are given in appendix 5.

Figure 20 He* sensor results at moving distance from the source

When zooming in on the results further away from the source (see appendix 5) it is observed that the expected flattening of the signal is not present, the signal keeps decreasing. At a 5 to 15 cm distance from the source the curve shows a change in behaviour. This could indicate a difference is gas-composition at different positions inside the chamber. This behaviour is not an anomaly since the change in behaviour is also observed when the distance is scanned in reverse. It can also be noted that the signal starts at a lower point when the sensor is facing away from the source or sideways. This can be explained by the fact that the sensor in reality is further away from the cage when facing sideways than the other two positions (see figure 21) and this difference in distance becomes less important when moving further away. This does not explain why the signal is lower when the sensor is facing away from the source. Another explanation is therefore that the sensor was cleaner when these experiments were started.

This lead to the thought that the stabilisation of the sensor could be of influence on the results obtained. Every time the vacuum chamber is opened air which holds water and carbon is introduced into the chamber. To check whether it takes time to get a stable signal, the chamber is opened for a while and then closed again. After the plasma is ignited and the potentials are applied to the sensor, the signal is followed for a while. The results are depicted in figure 21.

Figure 21 Stabilisation of the He* sensor

The signal is not stable within a few minutes as can be noted from figure 22. An explanation for the curve depicted is that in the first 6 minutes the water at the surface of the sensor is being removed and after this the cleaning of the surface slowly starts. The signal is decreasing in such a slow manner that after about 1000 seconds an accurate measurement can be performed.

The results of the ion and electron shielding experiment are depicted in figure 22 and 23. In figure 22 the results of the measurement at 1 cm distance from the cage and facing the source are depicted and in figure 23 the results of the measurement at 1 cm distance from the cage and facing away from the source are depicted. The results from the measurements at a distance of 6 cm from the cage are given in appendix 5.

Figure 22 voltage sweep facing the source at 1 cm distance

Figure 23 voltage sweep facing away from the source at 1 cm distance

The most important observation from the figures above is the fact that although there is not a flat line below -10V on the inner probe, it is observed that at a positive inner probe voltage and an increasing positive outer probe voltage the measured current becomes lower. When a negative inner probe voltage is applied and an increasing outer probe voltage the measured current becomes higher. This behaviour is noted for both distances.

This means that either the sensor is too big for the environment inside this vacuum chamber or the potentials applied to the probes are not large enough.

When this experiment is to be repeated the following result (figure 24) will be the wished result.

Figure 24. The wished result for the ion and electron shielding experiment

3.8 Metastable helium or not3.8.1 Set up

The most important question to be answered was whether there are metastable helium atoms present and if they are responsible for (part of) the cleaning that is observed. First of all, to determine if there are metastables present a set up must be made that excludes everything else. To exclude electrons and ions, the same principle as the He* sensor could be used. When all electrons and ions are repelled only neutrals can reach the centre. To test this, a box was created. This box has two integrated grids which can be kept at a certain potential (see figure 25). Appropriately chosen potentials should shield the sample from any electrons and ions. The surface of the sample was grounded to prevent charging. In figure 26 the position of the box is depicted.

The cleaning rate results from section 3.6.2 suggested that cleaning species might be able to reach the bottom of the box and cause carbon removal (cleaning) of the sample.

Figure 25 Schematic view of the box

(sample position) (box)

Figure 26 the box to check for metastable helium

3.8.2 Results

A sample is put on the bottom of the box (see figure 27), the carbon surface of the sample was grounded to prevent charging and the box is mounted into the chamber as in figure 26. In table 5 the results of the conducted experiments are given.

Figure 27 the alignment of the sample in the box

Table 5 cleaning results from the box

sample

Pressure

Remark

plasma mode

cleaning rate (nm/hour)

4562

0,1 mbar

Pump down over night

pulsed, 1000 W, 10% duty

4563

0,1 mbar

N.A.

continuous wave, 1000 W

4564-1

0,5 mbar

N.A.

continuous wave, 1000 W

4564-2

0,5 mbar

Pump down over night

continuous wave, 1000 W

4564-3

0,1 mbar

cage removed

continuous wave, 1000 W

The conditions for sample 4564-3 were different than for the other samples. When the cage is removed from the plasma source the whole vacuum chamber becomes a plasma creation area.

At first the grids were connected to earth to measure base line cleaning. However, no significant baseline cleaning has been observed at varying settings. The method was abandoned due to lack of time.

4. Conclusions

The overall conclusion is that a definitive answer about the cleaning mechanisms of metastable helium cannot be given at this point. There are some conclusions possible though and they are shown here.

There are definitely differences between different types of plasma, which include the electron density and electron temperature. The reactions to a change in settings or position within the vacuum chamber are different for each type of plasma. For metastable helium the singlet and the triplet states behave different as well when the settings are changed. This leads to the conclusion that the plasma characteristics can be tuned and the spectrometer is an appropriate way of monitoring them.

Spectrometer results in the power input and pressure dependency experiments also proved that the preferred singlet metastable is most abundant at higher input powers and a low pressure. These ratios are a good way of monitoring the processes inside the vacuum chamber.

The emission lines of helium decrease much slower when moving the spectrometer away from the plasma source than for hydrogen and oxygen/argon. This indicates a slow decay of excited helium gas, therefore excited species can travel further away from the source than other plasma types.

The cleaning experiments proved that cleaning is possible for samples without a direct line of sight. This leads to the conclusion that there are species with a lifetime long enough to reach the backside of something placed inside the chamber.

5. Recommendations

It will remain difficult to distinguish between metastable cleaning and ion cleaning. There are some recommendations for future research on this topic.

The cleaning properties of the system should be investigated and optimised. For this purpose different gases such as helium, hydrogen and argon can be used. This ensures the use of both a reactive gas (hydrogen) and an inert gas (argon) to compare the results of helium with. Different settings have to be investigated as well, like the pressure inside the chamber, the power input and the way of coupling power into the source, the location of the samples and the orientation of the samples.

The possibilities for a metastable helium sensor (as in section 3.7) should be reinvestigated at typical PATS settings, in order to design, construct and calibrate a new sensor. When this is successful the experiments in section 3.7 can be repeated to observe the metastable helium currents. The sensor can help to create an understanding of the behaviour of metastables at different positions in the chamber and relative to the plasma source.

The Andor spectrometer should be attached to the PATS because of its large wavelength range and sensitivity. Then it might be possible to see the metastables as well as the ions and if they can be measured the ratios can be compared for different plasma conditions. This can also be coupled to the transition probabilities to accurately determine the amounts present.

When a good and proven method to measure the amount of ions and metastables (singlet and triplet separately) is created successfully, an attempt can be made to couple the cleaning rate and heat load to these amounts. The cleaning efficiency can then be observed per species and they can be tuned.

Finally an attempt can be made to understand the question as to why the cleaning rate is sometimes larger for the sample facing away from the source than for the sample facing the source.

6. Acknowledgements

I learned a lot from the things encountered during this project. It made me aware of some of the principles of vacuum technologies and plasma techniques, since they are completely different from those normally encountered by me. I have also learned about equipment problem solving, which was also new to me, since before all the equipment I worked with had a service contract and no manual troubleshooting was required. The most important thing I have learned though has to do with critically looking at experiments and results. This is however a process that is not finished and I have some things to learn about this. After some difficulties in the beginning, the project came to a proper end. A lot of the insights I gained are a result of the guidance at TNO by René Koops, Norbert Koster, Edwin te Sligte, Peter Bussink and also Timo Huijser, who always took the time to answer my questions. There are a few other persons to thank at TNO since they made it possible for me to go to TNO cheerful every day. First of all my roommate Elfi van Zeijl for all her fun and interesting stories and Anshella Ramdin for her friendship. Besides this Roland van Vliet deserves a thank you as well since he made it possible to do something about my dental problem during the time I was part of his department. I would like to thank the rest of the department for the kind and warm welcome as well as the friendly time I experienced during my time at TNO.

Besides the guidance at TNO, I would like to thank Wim Kok from the University of Amsterdam for the guidance, support and opportunity to pursue the internship I wished to do.

7. References

9

Appendices

Appendix 1

Observations obtained from literature:

Overall conclusions:

The overall conclusion that can be drawn from the papers found is that a lot of researchers observed the effect of helium cleaning, but nobody has determined the exact nature of the mechanism involved. However there are some other useful conclusions to be drawn from the articles, which are given here.

One article suggested that the cleaning by helium is done by metastable helium radicals reacting with the surface, thereby reducing the surface impurities.

When helium plasma is ignited in a fusion reactor (which is a high energetic plasma) the following is observed; Hydrogen rapidly appears when the discharge is started and slowly falls off, disappearing completely when the discharge is stopped. The main source for this hydrogen suggested by the author is adsorbed water but according to some other authors this could also be released from the vessel wall. The following is also observed; Hydrogen and helium are both present during plasma cleaning, even if they are not present in the gas used to produce the plasma.

Accumulation of helium by the vessel wall is influenced by the wall material itself. Graphite holds less helium than stainless steel, resulting in a lower release during plasma treatment.

No correlation is found between helium pressure within a stainless steel vessel and the gas species of the main plasma. During experiments a net loss of helium particles is observed and the main source of helium desorption is the stainless steel wall.

Non-oxygen plasma leads to higher oxygen and nitrogen levels on the surface being cleaned. Storage has distinct effects on the surface composition, which is depending on the plasma type used.

Relevant quotes from articles including reference;

1. Shortening the pump-down time can be done by discharge cleaning. When applied just after the start of the pump-down, the pumping time can be reduced by one fourth. The final pressure is not changed by this procedure.

2. When activating helium plasma a strong emission line of neutral hydrogen is observed, although no hydrogen is introduced into the vacuum. When the plasma is fully operational the intensities of the emission lines of helium are increased as well. In the final stage of the discharge the emission lines of helium are rather strong. The same can be seen for the other situation, when hydrogen plasma is used, emission lines for helium are visible.

3. A SEM image after helium sputtering of the surface of a tungsten sample shows no blistering or bubbles. From EDX analysis it was found that the deposition layer on the surface contains trace amounts of iron (1.3 at.%). Oxygen was not detected. Using a quadrupole the release of helium was followed. It was revealed that some amounts of helium were incorporated into the deposition layer though helium is an inert gas. The quadrupole was used to investigate the presence of hydrogen during the helium plasma discharge. The measurements revealed the following;

· When the plasma discharge begins, hydrogen rapidly appears and then slowly falls off.

· When the discharge was turned off, hydrogen disappears immediately.

· The main source of the hydrogen is guessed to be adsorbed water.

· The trapping mechanism of helium and hydrogen in tungsten might be the same when looking at respectively helium and hydrogen plasma. The release curves look exactly the same. The hydrogen curve in helium plasma however looks different. This is believed to be due to the differences in concentration distribution in the layer.

· The H/W ratio in hydrogen plasma is estimated around 0.15. The He/W and H/W ratio in helium plasma is estimated around 0.08 and 0.075 respectively. (He+H)/W ratio is 0.155, which is very close to the value of H/W in hydrogen plasma. When hydrogen emission during helium plasma discharge can be repressed, the values might be equal.

· Hydrogen isotope and helium retention in the layer is considered to depend on ion flux, tungsten atom flux, incident energy and temperature. Therefore, there is a possibility that some amounts of the layers formed in the reactor contain a large amount of hydrogen isotopes and helium.

The erosion rate of tungsten by helium plasma sputtering was 1.8 times larger than that by hydrogen plasma sputtering. This is due to the difference in weight between He+ and H2+.

4. When comparing accumulation in stainless steel and in graphite, differences can be observed for helium as well as for hydrogen. When increasing the temperature using a stainless steel wall no change in accumulation is visible, but by changing to graphite the accumulation of hydrogen is cut in half and helium is even reduced to 1/3. This results at high temperatures in a reduction of helium release to 1/6. The results indicate that stainless steel accumulates helium gas several times more compared with graphite by plasma irradiation under helium glow discharge cleaning (GDC), which suggests that the helium behaviour during plasma confinement experiment in LHD originates from the stainless steel used as the first wall of the vacuum vessel.

5. By using helium gas, radicals are introduced on the surface which react and thereby reduce the impurities on the surface.

6. Nitrogen and oxygen tend to associate to form amide groups on polymer surfaces.

7. The effect of storage on the oxygen and nitrogen concentration using different plasmas;

· Polystyrene, helium plasma; A slight increase in nitrogen (2,5 instead of 2%) and a marked increase in oxygen concentration (14 instead of 10%) is seen after storage.

· Polystyrene, 33% nitrogen/ 67% hydrogen plasma; The nitrogen concentration decreased slightly over time (9 instead of 15%), while the oxygen concentration increased (15 instead of 9%). The oxygen is present in single, double and multi bonds.

· Polystyrene, 10% oxygen/ 90% helium; There seems to be an apparent loss of oxygen (13 instead of 16%) and a slight increase in nitrogen (1 instead of 0,5%).

· Polystyrene, 10% oxygen/ 90% helium followed by 33% nitrogen/ 67% hydrogen; A loss of nitrogen (8 instead of 15%) and an increase in oxygen concentration is seen (13 instead of 8%).

· Polystyrene, 33% nitrogen/ 67% hydrogen followed by 10% oxygen/ 90% helium; The oxygen concentration is initial a little higher than for the previous experiment (14%), but is stable during storage, the nitrogen concentration appears to decrease (6 instead of 12%).

· Polyethyleneterephthalate, 10% oxygen/ 90% helium followed by 33% nitrogen/ 67% hydrogen; The oxygen concentration is stable but high at 25%, the nitrogen concentration seems to decrease after storage (7 instead of 15%).

· Polyethyleneterephthalate, 33% nitrogen/ 67% hydrogen followed by 10% oxygen/ 90% helium; The oxygen concentration decreases a little bit to 25%, the nitrogen concentration decreases as well (9 instead of 11%).

For polystyrene, oxygen being present in the plasma leads to higher initial concentration, but the levels decrease over time. When there is no oxygen present the opposite happens. The concentration nitrogen seems to behave the same but to a lesser extent.

8. At the prevailing plasma densities, the effects of metastable states of helium on the line intensities would be overestimated if the metastable states are assumed to be in a quasisteady state (i.e. the population determined entirely by the ground-state populations).

9. Since the minimum peak-to-peak RF voltages lie around 200 V it is easiest in helium gas to achieve breakdown. As the voltage and the power are increased, an increase of the light emission intensity is observed, followed by an expansion of the plasma volume. Generally, mixtures of gases require a higher voltage to ignite and sustain the plasma, the same goes for air and mixtures including air. Pure helium is characterised by the lowest excitation temperature at a given power input. Slight temperature decrease is observed at higher power levels. Regarding the electron temperature the same behaviour is observed in low pressure discharges.

10. No correlation is found between helium pressure and the gas species of the main plasma experiment. During the main experiments, a net loss of helium particles is observed. There is a difference between the amount of atoms calculated and measured, which is mostly explained by implantation in the wall. In helium plasma experiments, about half of the inlet helium is implanted into the wall, even though the wall is saturated with helium atoms by the He GDC. The main source of helium gas desorption in the LHD is also the stainless steel wall, since the area of stainless steel is much larger than that of the graphite.

Literature references:

1. K. Akaishi et al. 1997, production of ultrahigh vacuum by helium glow discharge cleaning in an unbaked vacuum chamber, Vacuum, 48, no. 7-9, 767-770.

2. M. Goto et al. 2003, Determination of the hydrogen and helium ion densities in the initial and final stages of a plasma in the Large Helical Device by optical spectroscopy, , Physics of Plasmas, vol. 10, no. 5, 1402-1410.

3. K. Katayama et al. 2007, Helium and hydrogen trapping in tungsten deposition layers formed by helium plasma sputtering, Fusion engineering and design, 82, 1645-1650.

4. Y. Kubota et al. 2003, Investigation of the trapped helium and hydrogen ions in plasma facing materials for LHD using thermal desorption spectrometer and alternating glow discharge cleanings, Journal of nuclear materials, 313-316, 239-244.

5. S. Marais et al. 2005, Unsaturated polyester composites reinforced with flax fibers: effect of cold plasma and autoclave treatments on mechanical and permeation properties, Composites Part A: Applied science and Manufacturing, 36 issue 7, 975-986.

6. R.W. Paynter 1998, XPS studies of the modification of polystyrene and polyethyleneterephthalate surfaces by oxygen and nitrogen plasmas, surface and interface analysis, 26, 674-681.

7. R. W. Paynter 2000, XPS studies of the ageing of plasma treated polymer surfaces, Surface and interface analysis 29, 56-64.

8. R. Prakash et al. 2005, Characterization of helium discharge cleaning plasmas in ADITYA tokamak using collisional-radiative model code, Journal of Applied Physics, 97, 043301-1-043301-7.

9. E. Stoffels et al. 2002, Plasma needle: a non-destructive atmospheric plasma source for fine surface treatment of (bio) materials, Plasma Sources Science and Technology, 11, 383-388.

10. H. Suzuki et al. 2003, Behavior of helium gas in the LHD vacuum chamber, Journal of nuclear materials, 313-316, 297-301.

Appendix 2

Setting up the system

Fourier transformation of a scan using the microwave source;

A moving average is put over the data first to exclude random noise as much as possible. After this a Fourier transformation (standard tool in the data analysis tool) is performed on each scan separately. The scans are then put in a plot and compared. 10 scans are taken but putting all scans in one plot is confusing, so 5 scans are put in a plot at the same time (figure 28 and 29).

Fig. 28 The Fourier signal from scan 1 to 5

Fig. 29 The Fourier signal from scan 6 to 10

From these results it is clear that there is some kind of interference. The distance between the peaks is the same and the signal decreases by the same amount with each peak. Since the sampling speed of the Langmuir probe is not exactly known, the frequency cannot be calculated but the voltage change is known so the frequency is set as 1/V.

Then the power supply was tried. In the manual it is mentioned that the power for the computer as well as the probe have to be drawn from the same phase. However, the signal is not very good so another power supply for only the computer as well as all equipment is tried. In figure 30 the normal situation is given and in figure 31 the result from the different power supply is depicted. There is only one trace in figure 31 since the result for both attempts gave the same result.

Fig. 30 Langmuir results using the normal power supply

Fig. 31 Langmuir result using another power supply

The signal only gets worse and a repeating wave becomes visible. It is therefore not a good idea to switch to another power supply. Everything is connected to the original power supply again.

Until now the microwave source was used in a pulsed mode, but it is also possible to use the microwave source in continuous wave mode or to use an RF source to create plasma. Using all modes a scan is made and they are plotted in figure 32 to see the possible differences.

Fig. 32 the different plasma modes compared

The continuous wave mode gives the best results since there are no interferences and the power put into the system is handled the most efficient. The signal is also much higher, which makes calculations easier for the software.

Appendix 3

plasma dependence on power and pressure parameters

Power dependence

The other parameters determined by the Langmuir probe are the floating plasma potential (Vf), the electron temperature (Te) in eV and the ion flux. These results are given in figure 33.

Figure 33 a) floating plasma potential, b) electron temperature (eV), c) ion flux at different input powers

The most interesting result depicted in figure 33 is the fact that the floating potential for hydrogen (green trace) is positive, while the potential for helium (red trace) and oxygen/argon (blue trace) is predominately negative.

When the I-V characteristics could not be determined from the curve a value is missing.

The results for the spectrometer measurements are given in table 6, 7 and 8 for respectively helium, hydrogen and oxygen/ argon plasma. From these tables ratios can be calculated as well as absolute differences can be seen. Using this data an interpretation can be made as to the changes resulting from a change in environmental characteristics.

Table 6 Peak heights helium plasma

Peak (nm)

350 W

500 W

700 W

319

3065

3444

3824

389

24273

26684

29410

447

11576

13459

15113

492

3311

3835

4295

502

8528

9840

11081

588

50466

58643

64668

668

14420

18057

20580

707

10965

11913

13154

728

2155

2367

2627

Table 7 Peak heights hydrogen plasma

Peak (nm)

350 W

500 W

700 W

434

5086

5737

7037

486

21118

23653

30496

656

64664

64666

64666

Table 8 Peak heights oxygen/ argon plasma

Peak (nm)

350 W

500 W

700 W

697

4412

5852

5950

707

6123

7186

7305

737

9042

12248

12481

750

25977

32861

33194

763

25923

34982

35700

772

8161

10946

11149

777

20025

30265

31747

795

7590

10166

10392

801

8762

11787

12023

811

38245

52315

53475

826

4583

6026

6148

840

10183

13700

13916

842

15496

20952

21372

For all types of plasma an increase in signal can be observed with increasing power input.

Pressure dependence

The other parameters determined by the Langmuir probe are the floating plasma potential (Vf), the electron temperature (Te) in eV and the ion flux. These results are given in figure 34.

Figure 34 a) floating plasma potential, b) electron temperature (eV), c) ion flux at different pressures

The most interesting result depicted in figure 34 is the fact that for helium (red trace) all characteristics show a decrease with increasing pressure.

When the I-V characteristics could not be determined from the curve a value is missing.

The results for the spectrometer measurements are given in table 9, 10 and 11 for respectively helium, hydrogen and oxygen/ argon plasma. From these tables ratios can be calculated as well as absolute differences can be seen. Using this data an interpretation can be made as to the changes resulting from a change in environmental characteristics.

Table 9 Peak heights helium plasma

Peak (nm)

0,2 mbar

0,5 mbar

1 mbar

319

2949

3065

3127

389

19281

24273

25943

447

9085

11576

12345

492

2982

3311

3376

502

7187

8528

8752

588

30295

50465

61947

668

8510

15198

17879

707

8243

10965

12776

728

1990

2155

2292

Table 10 Peak heights hydrogen plasma

Peak (nm)

0,2 mbar

0,5 mbar

1 mbar

434

5292

5086

3479

486

19351

21118

12315

656

53708

64664

40805

Table 11 Peak heights oxygen/ argon plasma

Peak (nm)

0,2 mbar

0,5 mbar

1 mbar

697

5090

4412

913

707

6261

6123

566

737

10528

9042

1336

750

23896

25977

4466

763

30639

25923

3071

772

9588

8161

1260

777

26489

20025

1254

795

8886

7590

1189

801

10647

8762

1074

811

49771

38245

3582

826

5248

4583

929

840

11508

10183

1424

842

18418

15496

1909

From table 9 to 11 it can be derived that for helium the intensity increases with increasing pressure, while the opposite goes for oxygen/argon and hydrogen. This means that the gases react different on a change in pressure inside the chamber.

Appendix 4

Differences at different positions

The other parameters determined by the Langmuir probe are the floating plasma potential (Vf), the electron temperature (Te) in eV and the ion flux. These results are given in figure 35.

Figure 35 a) floating plasma potential, b) electron temperature (eV), c) ion flux

The biggest observation to be made from these results is the fact that the floating potential decreases a little bit when moving away from the source although still inside the cage, but the electron temperature decreases rapidly within the same step.

In table 12 to 14 the results are given for the peak heights, determined at different positions of the spectrometer relative to the cage containing the plasma. When observing the plasma through the antenna, the fiber was placed on the outside of the vacuum chamber.

Table 12 Peak heights helium plasma

Peak (nm)

Through the antenna

Middle of the cage

Just outside the cage

Half height above the cage

319

27734

5742

3907

3749

389

64433

47838

15542

10948

447

64607

25093

13234

10241

492

33277

8321

6727

5860

502

64543

21900

13189

9656

588

64300

64503

47290

34913

668

64418

45977

19918

14922

707

64436

53771

20891

16063

728

31100

10852

7328

6362

Table 13 Peak heights hydrogen plasma

Peak (nm)

Through the antenna

Middle of the cage

Just outsi