Arrangement of this Thesis

The thesis is arranged as following: the principles and the characteristic of gas discharge will be described in Chapter 2. In Chapter 3, the fabrication process of proposed device is described. Besides, the measurement equipments used to characterize the device are illustrated. In Chapter 4, the results including ignition voltage, voltage margin and the glow images operating in neon and argon gas are presented and discussed. The conclusion of the thesis and the future work are given in Chapter 5.

Chapter 2

Principle

2.1 Introduction

A better concept of micro-plasma devices can be gained by considering more conventional plasma phenomena. This chapter provides some background information on describing the behavior of gas discharges. Some background is given on breakdown in gas and glow discharges. Also, the importance of Paschen’s law is explored and its effect on the operation of gas discharges at this thesis is discussed.

2.2 Plasma

Generally, the term “plasma” is used to describe a partially or completely ionized gas containing electrons, ions and neutrals. Although there is always a small degree of ionization in any gas, a stricter definition of the plasma is “a quasi-neutral gas of charged and neutral particles which exhibits collective behavior” [31]. Quasi neutrality refers to the characteristic that positive and negative space charges balance in a given volume such that overall, the plasma is considered to be electrically neutral.

The collective behavior of the plasma is result from the Coulomb forces that are long range and cause remote regions to interact with one another.

In Fig. 2.1, Micro-plasmas show a new class of the plasma whose properties fall somewhere between those of glow discharge and arcs. These two types of processing plasma are characterized to low-pressure glow discharges and high-pressure arcs.

However, the low electron temperature and non-equilibrium make them more similar to glow discharges. As a result, they are often referred to as a “high pressure glow discharges”.

Fig. 2.1. Space and laboratory plasmas classified by their electron temperature, Te, and charged particle density, n [32].

2.3 Gas Discharges

2.3.1 Gas Breakdown and Paschen’s Law

The phenomenon of gas breakdown is considered as the transition from insulating state to conducting state. The associated voltage required to cause the transition is referred to as the breakdown voltage. Current flows between the electrodes via the movement of charged particles (ions and electrons) to and from the electrodes. When the electrons are energetic enough to dissociate the gas through impact to create ions, formation of a gas discharge occurs. The average electron temperature, Te, is generally less than breakdown energy required for ionization.

Ionization processes are possible because electrons have a distribution of energies [32]. If we assume a Maxwellian distribution at the average electron temperature, Te, the electron energy distribution function will be of the form [32]:

( ) exp( )

e

g T

ε = ε − ε (2.1)

where ε is the electron energy. From the graphical depiction of Eq. 2.1 in Fig 2.2, the distribution shows a maximum corresponding to Te/2 and a tail at higher energies.

Electrons in the high-energy distribution are responsible for ionization processes that create and sustain the plasma. For non-Maxwellian plasmas, the tail of the distribution may be higher or lower than that shown in Fig. 2.2 because of electron heating or collision processes. If the plasma is to be sustained after breakdown, there must be a mechanism for the generation of electrodes and ions to replace the ones lost to the electrodes.

Fig. 2.2. Electron energy distribution in a weakly ionized gas, assuming a Maxwellian distribution at the average electron temperature, Te.

Fig. 2.3. The diagram of the plane parallel electrode for producing a direct current (DC) glow discharge.

The simplest configuration employed for striking a gas discharge is plane parallel electrode, a cathode and an anode, separated by a distance of d, as illustrated in Fig. 2.3. The interelectrode space is filled with a noble gas at a pressure of p. The volume processes are defined by collisions between energetic electrons and neutrals that from an ion, and as a result, another electron. Thus the change in the number of electrons along an element of length dx, is proportional to the number of electrons ne

at x. This equation and its solution are given by :

e e

dn =αn dx (2.2)

( ) (0) exp( )

e e

n x =n αx (2.3)

where α is an ionization coefficient which is known as the first Townsend coefficient.

By integrating between x=0 and x=d we obtain for the number of electrons nd at d or the current id at d : irradiation of the cathode. When an energetic ion strikes the cathode, electrons are liberated from the cathode surface as a result of secondary electron emission, γ, defined as the probability of a secondary electron generation on the cathode by an ion impact. An electron generated from the cathode produces exp(αd) 1− positive ions in the interelectrode space. We neglected the electron losses here due to recombination and attachment to electronegative molecules. Electron–ion

recombination was neglected because the ionization degree is very low during the breakdown; attachment processes important in electronegative gases will be especially discussed below. All the exp(αd) 1− positive ions produced in the gap per electron move back to the cathode, and altogether strike out γ[exp(αd) 1]− electrons from the cathode in the process of secondary electron emission. The current in the gap is non-self-sustained as long as γ[exp(αd) 1]− is less than one because positive ions generated by electron avalanche must produce at least one electron to start a new avalanche. As soon as the electric field becomes high enough the transition to self-sustained current (the breakdown) takes place. Thus, the simplest breakdown condition in the gap can be expressed as :

[exp( d) 1] 1

γ α − = (2.5)

It is possible to derive relations for the breakdown electric field based on Eq. 2.5 by rewriting the ionization coefficient α, relating the similarity parameters α/p and E/p.

exp( )

/ A B

p E p

α = − (2.6)

where A and B are published values for different operating characteristics, E is the electric field and p is the noble gas pressure. Combination of Eq. 2.5 and 2.6 gives the following convenient formula for the breakdown voltage :

ln ln( )

This relationship is referred as Paschen’s law. The law essentially states that the breakdown characteristics of a gap are a function (generally not linear) of the product of the gas pressure and the gap length [33]. The experimental Paschen’s curves for different gases are illustrated in Fig. 2.4. These curves have a minimum voltage point, corresponding to the easiest breakdown conditions. In the case where pd is small, low pressure or narrow gaps, the number of ionizing collisions is minimized through a lack of available neutrals or a short path length over which a collision can occur.

Therefore, the curve rises rapidly on the side left of the minimum. The right side of the curve scales almost linearly with pd because the probability that an electron will undergo an ionizing collision is still high.

Fig. 2.4. The experimental Paschen’s curves for different gases.

2.3.2 Glow Discharges

The name of glow discharge is the phenomenon that plasma is luminous and depends on the geometry of the electrodes and the vessel, the gas used, the electrode material. When breakdown occurs, the gas suddenly becomes conductive. The current and voltage in the gas discharge vary resulting changing supply voltage. These parameters can be measured to obtain a voltage-current (V-I) characteristic for a discharge as illustrated in Fig. 2.5. In addition, current is limited through the discharge by using an external ballast resistor.

Fig. 2.5. Voltage-current (V-I) characteristic for a discharge.

Three general regions can be identified on the diagram above, the dark discharge (Townsend region), the glow discharge and the arc discharge. Because the discharge remains invisible to the eye except for corona discharge and the breakdown itself, the regime between A and E on the voltage-current characteristic is termed a dark

discharge. The electric field sweeps out the ions and electrons created by ionization from background radiation during the background ionization. The ions and electrons migrate to the electrodes in the applied electric field producing a weak electric current.

Increasing voltage results in an increasing fraction of these ions and electrons. If the voltage between the electrodes is increased far enough, eventually all the available electrons and ions are swept away, and the current saturates. The regime is named the saturation region because the current remain constant while the voltage is increased.

After the voltage across the saturation region, the current will rise exponentially. The electric field is now high enough so the electrons initially present in the gas can acquire enough energy before reaching the anode to ionize a neutral atom. As the electric field becomes even stronger, the secondary electron may also ionize another neutral atom leading to an avalanche of electron and ion production. The regime of exponentially increasing current is called the Townsend discharge. The Corona discharge occurs in the regime of Townsend discharge because of high electric field near sharp points, edges, or wires in gases prior to electrical breakdown. If the coronal currents are high enough, corona discharge can be technically “glow discharges”, visible to the eye. For low currents, the entire corona is dark, as appropriate for the dark discharge. After the breakdown, the gas enters the normal glow regime, which the voltage is almost independent of the current over several orders of magnitude in the discharge current. The electrode current density is independent of the total current in this regime. This means that the plasma is in contact with only a small part of the cathode surface at low currents. As the current is increased, the fraction of the cathode occupied by the plasma increases until plasma covers the entire cathode surface.

When the cathode is completely covered by the plasma, the current is increased by increasing the voltage in order to force the cathode current density above its natural value and provide the desired current. The regime which has the positive slope of

current-voltage curve is so-called the abnormal discharge. The discharge maintains itself at considerably lower currents and current densities and only then makes a transition back to Townsend discharge regime. Further increase in the current will result in a sudden drop in voltage and the discharge will undergo a glow-to-arc transition. After transition of the discharge is an arc discharge where the discharge voltage decreases as the current increases, until large currents are achieved at a low voltage, and after that the voltage increases slowly as the current increases.

The glow can be produced by applying a potential difference in a gas between two electrodes. The potential drops rapidly close to the cathode, vary slowly in the plasma, and change again close to the anode. The voltage distribution in a dc glow discharge process is shown in Fig. 2.6. The electric fields between the gaps are restricted to sheath at each of electrodes. The sheath fields are such as to repel electrons trying to reach either electrode. Electrons originating at the cathode will be accelerated, collide, transfer energy, leave by diffusion and recombination, slow by the anode and get transferred into the outside circuit. The luminous glow is produced because the electrons have enough energy to generate visible light by excitation collisions. Since there is a continuous loss of electrons, there must be an equal degree of ionization going on to maintain the steady state. The energy is being continuously transferred out of the discharge and hence the energy balance must be satisfied also.

Simplistically, the electrons absorb energy from the field, accelerate, ionize some atoms, and the process becomes continuous. Additional electrons are produced by secondary emission from the cathode. These are very important to maintain a sustainable discharge.

Fig. 2.6. The potential distribution in a dc glow discharge process.

Fig. 2.7. The spatial distribution of (a) the luminous regions in a typical discharge and (b) the electric field.

Unlike Townsend or arc discharge, the glow discharge is characterized by distinct regions with large variations in electric field [34]. The approximate appearance of the different luminous regions is shown in Fig. 2.7(a). Also shown for reference is the associated potential distribution between cathode and anode in Fig.

2.7(b). Three basic regions are described below, the cathode region, the glow regions and the anode region. The length of the cathode region is from the cathode surface to the boundary of the negative glow. Most of the potential drop across the cathode and anode occur between the cathode and the negative glow. The energies of electrons accelerated in this region are high enough to produce ionization and avalanching in the regions to the right of the right of the negative glow. The electrons are accelerated from the cathode to Aston Dark Space which is a thin regime with a strong electrical field. This regime has a negative space charge, meaning that stray initial electrons together with the secondary electrons from the cathode outnumber the ions in this regime. The electrons are too low density and/or energy to excite the gas, so it appears dark. The next regime to right of Aston dark space is cathode glow. The electrons are energetic enough to excite the neutral atoms they collide with. In addition, the cathode glow has a relatively high ion density. The cathode dark space that is to the right of the cathode glow has moderate electric field, a positive space charge and a relatively high ion density. These fine-structured regions make up what is known as Crooke’s dark space which is commonly referred to as the cathode fall or sheath. The brightest intensity of the entire discharge names negative glow. The negative glow has relatively low electric field, long, compared to the cathode glow and is the most intense on the cathode side. Electrons carry almost the entire current in the negative glow region. Electrons that have been accelerated in the cathode region to high speeds produce ionization, and slower electrons that have had inelastic collisions already produce excitations. These slower electrons are responsible for the negative glow. The

electron number density in the negative glow is about 10¹⁶ electrons/m³. Before reaching the positive column, the electrons lose their energy through more collisions in the Faraday dark space. The electron density decreases by recombination and diffusion to the walls. Furthermore, the net space charge is very low and the electric field is small. The uniform electric field in the positive column continually accelerates the electrons, which, in turn, undergo decelerating collisions. In the regime, small electric field is typically 1 V/cm. The electric field is just large enough to maintain the degree of ionization at its cathode end. The electron number density is about 10¹⁵ to 10¹⁶ electrons/m³ in the positive column, and the electron temperature of 1 to 2 eV. A thin sheath also exists near the anode with a similar structure to the cathode fall made-up of the anode glow and anode dark space. The voltage drop in this region is significantly smaller than near the cathode and plays little role in the discharge dynamics. Anode glow is slightly brighter than positive column. This is the boundary of the anode sheath. The anode dark space between the anode glow and the anode is the anode sheath. It has negative space charge due to electrons traveling from the positive column to the anode. There is a higher electric field than the positive column.

The anode pulls electrons out of the positive column and acts like a Langmuir probe in electron saturation.

2.4 Penning Effect

A gas mixture consists of a rare gas containing impurity atoms possibly at very low concentrations. The impurity atoms have an ionization potential Vion which is lower than or equal to the metastable potential Vmeta of the minority nobles gas. The Penning effect in a gas mixture is the ionization by charge transfer (charge exchange)

during collision between a metastable atom and a neutral atom which decreases the average energy to form an ion pair, e.g.

Ne^∗+Ar→Ne Ar+ ⁺+e (2.8)

In a glow discharge, the Penning effect results in an increase of the ionization coefficient (Townsend first coefficient). In other hand, the breakdown potential and the cathode fall potential are decreased. The ionization and metastable energies of the rare gas are listed in Table 2.1.

Table 2.1 The ionization and metastable energies of the rare gas.

Ionization and Metastable energies

He 24.6 19.8, 20.6

Ne 21.6 16.6, 16.7

Ar 15.8 11.5, 11.7

Kr 14 9.9, 10.5

Xe 12.1 8.3, 9.4

Chapter 3

Fabrication and Measurement Instruments

3.1 Introduction

The interdigitated electrode micro-plasma device will be demonstrated in this chapter. The embodiment including the fabrication processes, technologies and instruments which are available to develop the structure of interdigitated electrode micro-plasma device will be described in the following sections. First, the semiconductor process including wet bench, furnace, spin coating, exposure, develop, sputter, lift-off and evaporate will be used. Besides, the features of the fabricated device were measured by typical semiconductor measurement systems, such as optical microscope, atomic force microscope (AFM), scanning electron microscope (SEM). In addition, the performance, such as ignition voltage, voltage margin and glow images were characterized in a specific vacuum system including charge-couple device (CCD) camera and pulse DC controller. The mentioned instruments mentioned above will be illustrated in this chapter.

3.2 Semiconductor Fabrication Process

The detail fabrication processes are listed below and the flow chart is shown in Fig. 3.1.

Metal Deposition

Barrier Rib Formation Substrate Cleaning

Silicon Oxide Growth

UV Exposure

Lift-Off

Dielectric Layer Deposition

Fabricated Device

Fig. 3.1. The flow chart of fabrication process for micro-plasma device (a) substrate cleaning (b) silicon oxide growth (c) metal deposition (d) dielectric layer deposition (e) barrier rib formation.

a. Substrate Cleaning:

First of all, silicon wafers were cleaned by RCA clean in wet bench, shown in Fig. 3.2. The RCA clean is the industry standard for removing contaminants from silicon wafer. Werner Kern developed the basic procedure at RCA (Radio Corporation of America) laboratories in 1960’s. As shown in Table 3.1, the main purpose for this procedure is removing organic residue and films from silicon wafers. The decontamination works based on sequential oxidative desorption and complexing with H2O-NH4OH-H2O2 (called “standard clean-1”, SC-1) at 75~85 °C. A second standard clean (SC-2) is often used H₂O-HCl-H₂O₂ at 75~85 °Cto further clean the surface.

SC-1 is used to remove the organic residues from silicon wafer. In the process, it oxidizes the silicon and has a thin oxide on the surface of the wafer which should be removed is a pure silicon surface is desired.

Fig. 3.2. The photograph of Wet Bench for silicon cleaning.

Table 3.1 The RCA clean procedure.

RCA Clean Steps

1. DI water rinse, 5 min

2. H²SO4 : H2O2 = 3:1 Organic Clean

3. DI water rinse, 5 min

4. HF : H²O = 1:100 Chemical oxide Strip

5. DI water rinse, 5 min

6. NH⁴OH : H2O2 : H2O = 1:4:20 (SC-1), 10 min (75~85 °C) Particle Clean 7. DI water rinse, 5 min

8. HCl : H²O2 : H2O = 1:1:6 (SC-2), 10 min (75~85 °C) Ionic Clean 9. DI water rinse, 5 min

10. HF : H²O = 1:100 Chemical oxide Strip

11. DI water rinse, 5 min

12. Spinner Dry wafers

b. Silicon Oxide Growth:

Because silicon wafer is a semiconductor, the electrode formed on the surface will influence the discharge properties without a buffer layer. In order to avoid discharge between electrode and silicon wafer, a quite thick silicon oxide (about 1 µm) was grown on the surface as a buffer layer. The buffer layer was grown at 1100 °C using both thermal wet and dry oxide by Furnace, shown in Fig. 3.3.

Fig. 3.3. The photograph of Furnace for thermal oxide.

Table 3.2 The growth of 1 µm silicon oxide for Si (100).

Table 3.2 The growth of 1 µm silicon oxide for Si (100).