Capacitively coupled radio frequency discharge

2.2 Capacitively coupled radio frequency discharge

The electrodes must be conductive in order to sustain the DC glow discharge. If one or both electrodes are non-conductive, e.g. when glow discharge is used for the spectrochemical analysis of non-conducting materials or the deposition of dielectric films, where insulating materials are attached to the electrode surfaces and the electrode is charged due to the accumulation of positive and negative charges, the glow discharge will be extinguished. To solve this problem, an alternating voltage can be applied between the electrodes, so that each electrode can act as an anode and a cathode, and charges accumulated during the positive half-cycle of the voltage will be partially neutralized during the negative half-cycle.

Generally, the voltage frequency used for the alternating voltage is in the radio frequency range (1 kHz–103 kHz, with the most common frequency being 13.56 MHz). Strictly speaking, capacitively coupled discharge can also be generated by other voltage frequencies, so it is more appropriate to call it AC discharge. In addition, the frequency should be high enough so that the half period is shorter than the time it takes for the insulator to be fully charged. Otherwise, there will be a series of short-lived discharges with the electrodes successively taking opposite polarities, instead of a quasi-continuous discharge. It can be calculated that the discharge can stay continuous when the applied voltage frequency is greater than 100 kHz. In practice, many RF glow discharge processes occur at 13.56 MHz. This is because the frequency is among the radio spectrum reserved for industrial scientific and medical purposes by the International Telecommunication Union, and therefore it will not interfere with communications when it propagates a certain amount of energy.

It is worth noting that capacitive coupling refers to a way of coupling the input power into the discharge, i.e., using two electrodes and their sheaths to form a capacitor. As will be shown later, RF power can also be coupled and discharged by other methods.

At typical radio frequencies, electrons and ions behave completely differently, which can be explained by their mass difference. An electron, with its smaller mass, can keep up with the instantaneous electric field changes generated by the RF voltage. In fact, the natural frequency of electrons, the so-called electron plasma frequency, is:

Here   is expressed in. When the electron density changes from, the plasma frequency ranges from Hz, which is much higher than 13.56 MHz. If the voltage frequency is smaller than the ion plasma frequency, the ions can keep up with the change of the electric field in the sheath. As the frequency and mass of ion plasma are inversely proportional, electrons can keep up with the change of electric field at typical radio frequency, while ions can only follow the time-averaged electric field.

Another important aspect of capacitively coupled RF discharge is the self-bias effect, which is also caused by the difference between the masses of electron and ions. A self-bias, or DC bias is formed when (a) the two electrodes are differently sized, and (b) when a coupling capacitor is present between the RF power supply and the electrode, or when the electrodes is non-conductive (as it can be regarded as a capacitor). When a certain voltage (e.g. a square wave) is applied to the capacitor formed by the electrodes (see Fig. 3), the plasma voltage will initially have the same value as the voltage applied. When the applied voltage is initially positive, as shown in Fig. 3, electrons will be accelerated toward the electrode. Thus, the capacitor will be charged rapidly by the electron current, and the plasma voltage will drop. After half a cycle, when the polarity of the applied voltage changes, the plasma voltage changes by the same amount, i.e., twice the amplitude of the applied voltage. By this point, the capacitor has been charged by ion current, and the plasma voltage will drop but this second drop is less pronounced, as the much lower ion mobility results in lower ion flux. At the next half-cycle, the applied potential, and hence also the voltage over the plasma, again changes polarity. The plasma voltage drops faster now, because the capacitor is fully charged again by to the electron flux. This process repeats itself, until the capacitor is finally sufficiently negatively charged so that the ion and electron fluxes, integrated over one RF cycle, are equal to each other. Finally, a time-averaged negative DC bias (indicated by dashed lines in Fig. 3) is induced between the RF power electrodes. This phenomenon also occurs in the grounded electrode, but it has very limited impact. Fig. 4 shows a typical sinusoidal voltage with a frequency of 13.56 MHz and its corresponding DC bias.

Fig. 3 The generation process of self-biasduring capacitive coupling RF discharge in the case of a rectangular pulse:

(a) Voltages applied across the electrodes

(b) Voltage over discharge as a function oftime (dashed lines) and self-bias

Fig. 4 The solid line represents thevoltage between both electrodes, and the dashed line shows the resulting DCbias at the RF-powered electrode.

Due to the negative DC bias, ions continue toaccelerate toward the RF power electrode, which will cause sputtering of RFelectrode material. In fact, capacitively coupled RF discharge is similar to DCglow discharge with a similar subdivision in different regions, similaroperating conditions, and similar processes occurring in the plasma. This isespecially true with the mode of γ regime, where secondary electron emissionand the ionization due to accelerated electrons in the RF-sheath, are theprimary sustaining mechanisms in the discharge. Among the analyticalcapacitively coupled RF discharge processes, a typical mode is the discharge. In thiscase, the air pressure is very high (several hundred Pa), so is the voltage(amplitude can reach 1 kV), and the RF power electrode is much smaller than thegrounding electrode, which leads to a large self-bias voltage (usually onlyabout 80kV lower than the amplitude of the RF voltage). Another mode producedin capacitively coupled RF discharge is discharge. In thismode, the main mechanism involves the ionization process caused by electrons inthe whole plasma. Electrons gains energy from the RF electric field oscillation(i.e. the expansion and contraction of sheath). This is known as ohmic heating.This discharge mode usually occurs at low pressure and voltage. In addition, whenthe electric field is large, electrons can be heated in plasma (i.e., overallohmic heating). This happens with electronegative gas or very long and narrowdischarge tube, where radiation loss due to bipolar diffusion is crucial. Themechanism is similar to DC glow discharge with positive column.

Transitionbetween the α and γ regime is the subject of many theoretical, modeling and experimental studies. α modeis adopted from Townsend’s first ionization coefficient, and the γ coefficient from the ion-in duced secondary electron emission from a target, respectively.

In plasma processing applications, the simplest case of a capacitive coupling radio frequency discharge system, also known as “RFdiode”, consists of a vacuum chamber with two plane electrodes placed several centimeters apart. The substrate is usually placed on one of the electrodes.The driving voltage is typically within 100-1000 V. The air pressure range is 1–100Pa, and the electron density (i.e., plasma density) is . Therefore, both the pressure and plasma density are lower than those of most analytical RF discharges (where pressure is a few hundred Pa and plasma density is approximately ).

2.3 Pulsed glow discharge

In addition to using RF voltage for glow discharge, discrete pulsed voltage can also be applied, usually with pulse width of millisecond or microsecond. At the same average power, compared with DC glow discharge, pulse discharge has stronger instantaneous sputtering, ionization and excitation processes, providing higher efficiency (for analytical spectrochemistry, it means higher sensitivity can be obtained) This is because this process occurs at higher peak voltage and current, and the basic plasma phenomena of excitation and ionization are highly and nonlinearly dependent on electric field strength. Whereas the early analytical investigations dealt primarily with millisecond pulsed glow discharges, more recent work has focused mainly on microsecond discharges, where even higher peak voltages and currents, and hence better sensitivities, can be obtained.

Typical analytical microsecond pulses are 2 kV in amplitude, 10 μs in pulse width, and 200 Hz in pulse repetition frequency. The resulting peak current and power are around 1A and 2 kW, respectively. Therefore, the typical duty cycle is very short, i.e., the ratio of pulse-on period to pulse-off period is very small. This means that the average electrical power is rather low, so that the sample will not excessively be heated. Moreover, the overall rate of sputtering will be low, so that thin films can be analyzed.

For the same reason, also in the semiconductor industry, pulsed power operation has emerged as a promising technique for reducing charge-induced damage and etch profile distortion, which is associated with continuous discharges. Compared with RF technology, another advantage of pulsed DC technology is the simplicity of the coarsening method, which is a result of its reduced impedance matching network and electromagnetic interference problems, and the lower of power supplies for larger reactors. Typical operating conditions for pulsed technology plasmas involve discharge pulse durations in the order of 100 μs. The peak voltage is in the order of 500 V and the pressure is approximately 100 Pa. Also, the reactors are typically much larger, in the order of several m3. Common applications include plasma nitriding of steel and deposition of hard coating.

As far as basic plasma processes are concerned, a pulsed glow discharge is very similar to a DC glow discharge, i.e. it can be considered as a short DC glow discharge, followed by a generally longer afterglow, in which the discharge burns out before the next pulse starts. It should be pointed out that during pulse discharge, the phenomenon of non-thermal balance is facilitated, and the gas temperature is lower than the electron temperature without excessive heating. There is also non-chemical equilibrium, because ionization and recombination occur on a different time-scale.

2.4 Atmospheric pressure glow discharges (APGDs)

As mentioned above, glow discharge can operate over a very wide air pressure regime. The typical pressure range is approximately 100 Pa. Operation at higher (even atmospheric) pressure is, however, possible, but it leads easily to gas and cathode heating and arcing. According to the similarity of classical theories, it is possible to increase the gas pressure (p) if the linear dimension of the device (D) is decreased, the product pD being one parameter kept constant. Miniaturized discharge devices are, therefore, expected to be able to generate glow discharges at atmospheric (or even higher) pressure. In work by Schoenbach et al. and Stark and Schoenbach, an atmospheric pressure micro-discharge in hollow-cathode geometry was developed. The hollow cathode diameter was typically 100–200 mm. In Czerfalvi et al. and Mezei et al., a small discharge at atmospheric pressure was operated with an electrolyte as cathode. Eijkel et al. reported a helium DC glow discharge on a microchip at atmospheric pressure, which serves as a molecular optical emission detector for gas chromatography (for example, for methane, hexane, etc.). Typical dimensions are 1-2 mm in length and several hundred μm in width and height, leading to a typical plasma chamber volume of 50-180 nl. Also, capacitively coupled RF discharges at atmospheric pressure have been used for quite a while by Blades et al. and by Sturgeon and co-workers for analytical applications, and they have also been reported by a group in Romania.

Besides reducing the characteristic length of the discharge chamber, stable atmospheric pressure glow discharges (APGDs) used for technological applications have also been operated when other conditions are satisfied with respect to the structure of the electrodes, the carrier gas and the frequency of the applied voltage. Typically, in APGDs, at least one of the electrodes is covered with a dielectric, and the discharge operates at alternating voltages. For example, helium gives rise to a stable homogeneous glow discharge, whereas nitrogen, oxygen and argon easily cause the transition into a filamentary glow discharge. However, by changing the electrode configuration, it is still possible to let them operate in a homogenous glow discharge regime.

Fig.5

Fig. 5 shows a schematic picture of a typical APGD used for plasma polymerization. The glow discharge is generated between two parallel electrodes, which are covered by a dielectric layer (e.g. alumina). A gas flow, consisting in this case of plasma polymerization of specific monomers and helium as the carrier gas, is led through the discharge. An alternating voltage of 20 kV is applied with a frequency between 1 and 30 kHz. The distance between the electrodes is typically a few millimeters.

The main advantage of APGDs is the absence of vacuum conditions, which greatly reduces the cost and complexity of the glow discharge operation. Moreover, materials with a high vapor pressure, such as rubber, textiles and biomaterials can be treated more easily.

Typical known applications include the surface modification of materials (e.g. enhancing the wettability of polymers used for paints and glues), the sterilization of surfaces (e.g. sterilization of micro-organisms on surfaces in the healthcare industry), plasma polymerization, the production of ozone, etc.

2.5 Dielectric barrier discharges (DBDs)

Strongly related to the APGDs are the dielectric barrier discharges (DBDs), historically also called “silent discharges”. They also operate at approximately atmospheric pressure (typically 0.1–1 atm). An AC voltage with an amplitude of 1–100 kV and a frequency of a few Hz to a few MHz is applied to the discharge, and a dielectric layer (made of glass, quartz, ceramic material or polymers) is again placed between the electrodes. The inter-electrode distance varies from 0.1 mm (in plasma displays), over approximately 1 mm (in ozone generators) to several cm (in lasers).

DBDs are different from APGDsmainly in that the latter are generally homogeneous across the electrodes andare characterized by only one current pulse per half cycle, whereas the DBDstypically consist of microdischarge filaments of nanosecond duration (hence,with many current pulses per half cycle). In fact, this subdivision is ratherartificial, because the same electrode configuration can give rise to an APGDor a DBD, depending on the discharge conditions and the discharge gas. Thereare even indications that the homogeneous APGD is not really homogeneous, butsimply has a homogeneous or diffuse light emission. Therefore, an APGD can alsobe considered as a diffuse DBD.

Fig.6

Two basic configurations of DBDs can be distinguished. The volume discharge (VD) consists of two parallel plates (see Fig. 6a). The microdischarge takes place in narrow channels which cross the discharge gap and are generally randomly distributed over the electrode surface. The number of microdischarges per period is proportional to the amplitude of the voltage. The surface discharge (SD) consists of a number of surface electrodes on a dielectric layer and a counter electrode on its reverse side (see Fig. 6b). There is no clearly defined discharge gap. The so-called micro-discharges are, in this case, rather individual discharge steps that take place in a thin layer on the dielectric surface and can be considered homogeneous over a certain distance. An increase in the voltage now leads to an enlargement of the discharge area on the dielectric. There are also combinations of these two basic configurations, e.g. a co-planar arrangement, such as those used in plasma display panels, or a packed bed reactor, such as those used in certain plasma chemical reactors.

The nanosecond duration of DBDs is caused by a charge build-up at the dielectric surface, within a few ns after breakdown. Indeed, this reduces the electric field intensity of the microdischarge to such an extent that the charge current at this position is interrupted. Because of the short duration and the limited charge transport and energy dissipation, this normally results in little gas heating. Hence, in a microdischarge, a large fraction of the electron energy can be utilized for exciting atoms or molecules in the background gas, thus initiating chemical reactions and/or emission of radiation. This explains the great interest in DBDs for many applications.

In 1857, Siemens already used this type of discharge for the generation of ozone from air or oxygen. Today, these silent discharge ozonizers have become very effective tools, and a large number of ozone installations are being used worldwide for water treatment. Other applications are the pumping of lasers, the generation of excimer radiation in the UV and VUV spectral regions, the production of methanol from methane/oxygen, various thin-film deposition processes, the remediation of exhaust gases and for plasma display panels. Recently, the use of a DBD in analytical spectrometry has been reported, i.e. as a microchip plasma for diode laser atomic absorption spectrometry of excited chlorine and fluorine in noble gases and in air/noble gas mixtures. It has been demonstrated that the DBD has excellent dissociation capability for molecular species, such as

2.6 Corona discharges

Beside a glow discharge with two electrodes, there is another type of pulsed DC discharge, with the cathode in the form of a wire. A high negative voltage (in the case of a negative corona discharge) is applied to the wire-cathode, and the discharge operates at atmospheric pressure. The name “corona discharge” arises from the fact that the discharge appears as a crown-shaped glow pattern around the cathode.

The mechanism of the negative corona discharge is similar to that of a DC glow discharge. The positive ions are accelerated towards the wire, and cause secondary electron emission. The electrons are accelerated into the plasma. This is called a streamer, which includes a moving front of high-energy electrons followed by a tail of lower energy electrons. The high-energy electrons give rise to inelastic collisions with the heavy particles, e.g. ionization, excitation, dissociation. Hence, radicals can be formed, which can crack larger molecules in collisions. Therefore, there is a clear distinction between the electron kinetics and the heavy particle kinetics of interest for the applications. The separation between them is not accomplished here in space, but in time. The corona discharge is also in a strongly non-equilibrium state with respect to both temperatures and chemical properties. The main reason is the short time-scale of the pulses. If the source was not pulsed, there would be a build-up heat, giving rise to thermal emission and to a transition into an arc discharge close to equilibrium.

It should be mentioned that, besides the negative corona discharge, there also are a positive corona discharge, where the wire has a positive voltage, hence acting as anode.

Applications of the corona discharge include flue gas cleaning, disposal of volatile compounds that escape from paints, water purification, etc. Dust particles are removed from the gas or liquid by the attachment of electrons from the discharge to the dust particles. The latter becomes negatively charged and will separated from a gas or liquid.

2.8 Low-pressure, high-density plasmas

In recent years, a number of low-pressure, high-density plasma discharges have been developed, mainly as alternatives to capacitive RF discharges (RF diodes) in etching and deposition applications. Admittedly, one of the disadvantages of RF diodes is that voltage and current cannot be independently controlled, and hence, the ion-bombardment flux and bombarding energy cannot be varied independently of each other, except when applying different frequencies. But applying various frequencies is not always very practical. Hence, for a reasonable ion flux, sheath voltages at the driven electrode must be high. For wafers placed on the driven electrode, this can result in undesirable damage due to the high bombarding energies. Furthermore, the combination of low ion flux and high ion energy leads to a relatively narrow process window for many applications. The low process rates resulting from the limited ion flux in RF diodes often mandates multi-wafer or batch processing, with the consequent loss of wafer-to-wafer reproducibility. To overcome these problems, the mean ion bombarding energy should be controllable independently of the ion and neutral fluxes. Some control over ion-bombarding energy can be achieved by placing the wafer on the undriven electrode, and independently biasing this electrode with a second RF source. Although these so-called RF triode systems are in use, processing rates are still low at low pressures and sputtering contamination is an issue. Various magnetically enhanced RF diodes and triodes have also been developed to increase the plasma density, and hence the ion fluxes. However, as mentioned above, the latter may not have good uniformity, limiting their suitability for plasma processing applications.

The new generation of low-pressure, high-density plasma sources are characterized, as predicted by the name, by lower pressures (typically 0.1–10 Pa) and by higher plasma densities (typically at these low-pressure conditions), and consequently by higher ion fluxes than capacitive RF discharges of similar pressures. In addition, a common feature is that the RF or microwave power is coupled to the plasma across a dielectric window, rather than by direct connection to an electrode in the plasma. This is the key to achieving low voltages across all plasma sheaths at electrode and wall surface. DC voltages, and hence also ion acceleration energies, are typically only 20–30 V at all surfaces. To control the ion energy, the electrode on which the wafer is placed can be independently driven by a capacitively coupled rf source. Hence, independent control of the ion/radical fluxes and the ion-bombarding energy is possible.

Although the need for low pressures, high fluxes and controllable ion energies has motivated high-density source development in recent years, there are still many issues that need to be resolved. A critical issue is achieving the required process uniformity over large wafer surfaces. In contrast to the nearly one-dimensional geometry of typical RF diodes, high-density cylindrical sources are essentially two-dimensional. Hence, plasma formation and transport are inherently radially non-uniform. Another critical issue is efficient power transfer across the dielectric windows over a wide range of plasma operating conditions. Degradation of and deposition on the window can also lead to irreproducible source behavior and the need for frequent, costly cleaning cycles. Finally, the low-pressure operation leads to severe pumping requirements for high deposition or etching rates, and hence the need for large, expensive vacuum pumps. Figs.8–11 present the four most important high-density plasma sources. The common features of power transfer across dielectric windows and separate bias supply at the wafer electrode are clearly illustrated. Moreover, it shows the common idea of separate plasma creation chamber and process chamber. However, the sources differ significantly in the means by which power is coupled to the plasma. In principle, some of these discharges could also be classified in other sections, e.g. the electron cyclotron resonance (ECR) discharge mentioned in Section 2.7. This is because it uses a magnetic field, and it is also involved in matters covered in Section 2.9. However, we have chosen the present classification, because of their common growing importance in plasma processing applications.