Experimental investigation of an electronegative cylindrical capacitively coupled geometrically asymmetric plasma discharge with an axisymmetric magnetic field Open Access

A stable electropositive argon plasma mixed with a small amount of reactive gas such as oxygen has a wide range of applications in the semiconductor industry for processes such as silicon wafer etching,^1–5 thin film deposition,^6–8 negative ion sources,^9–12 and formation of oxide film.^13,14 The admixture of reactive gases not only provides necessary radicals for surface processes but also influences the plasma parameters. The control of parameters such as density and temperature is essential for optimizing the deposition and etching process to get the film of the desired structure and morphology. Therefore, studying the nature and properties of plasma formed by the mixture of electropositive and electronegative gas is an important area of research. All these applications are influenced by the negative ion concentration in the discharge. Therefore, it is worthwhile to investigate the plasma parameters and the negative ion density in such discharges. Various authors studied negative ion production in argon–oxygen discharge. Shindo and Kawai measured the negative ion density by using a Langmuir probe and ion acoustic wave (IAW) method in the argon + oxygen plasma in an Electron Cyclotron Resonance (ECR) discharge chamber and measured the negative ion fraction (α = n₋/n_e) of the order of 0.3. They have also reported that α strongly depends on the gas mixture ratio and has the optimum value when the oxygen percentage in the discharge is about 15%.¹⁵ Katsch et al. studied the negative ion fraction in pulsed inductively excited plasma. They have found that a high negative ion fraction occurs at low pressure and high input power.¹⁶ Gudmundsson et al. experimentally investigated the plasma parameters in the argon–oxygen plasma and compared the measured parameters with the values obtained by using the volume-averaged global model assuming Maxwellian electron energy distribution function (EEDF).¹⁷ Gudmundsson and Thorsteinsson also employed a volume-averaged global model to study the dissociation process of oxygen molecules in the argon–oxygen discharge. They found that the dissociation fraction of the oxygen molecule increases with the increase in the argon content in the discharge.¹⁸ Kitajima et al. reported the increased growth rate of SiO₂ when a stable gas like argon or krypton is mixed with a small amount of oxygen. They observed a twofold increase in the oxygen metastable state density with the dilution of the argon gas.⁴ Other notable works related to the argon plasma diluted with oxygen gas are done by Takechi and Lieberman,¹ Lee and Lieberman,¹⁹ Wang et al.,²⁰ Bradley et al.,²¹ Worsley et al.,⁵ etc.

It is well known that low-pressure radio frequency (RF) discharge produces higher plasma density at lower RF power as compared to DC discharges. Therefore, capacitively coupled low-pressure RF discharges have various applications in microelectronic fabrication in solar cells for etching and deposition of thin films,^22–26 semiconductor industry,^27–31 biomedical applications,^32–34 space technology,^32,33 etc. Generally, molecular gas mixed with novel gas is essential in optimizing the etching and deposition rate in film deposition and plasma processing. A mixture of two or more gases is used in the plasma processing chamber to achieve the desired morphology of the deposited film. In order to improve the efficiency of such processes, a good understanding of the plasma system is necessary. A simple CCP discharge system consists of two parallel plates, in which one is RF-powered and the other is grounded. Substrate is generally placed in the powered electrode for plasma processing.^22,23 Various works have been done to enhance the efficiency of such devices by independently controlling the flux and energy of the impinging ions. Some of the techniques adopted are using dual or multiple frequency RF sources,^35–44 tailored current and voltage waveforms,^45–52 and operating the system in the pulsed power mode.^53–55 It is observed that higher plasma density is produced by applying the higher RF, which increases the ion flux to the substrate. However, with the increased driving frequency, the plasma uniformity decreases, reducing such system's efficiency. To overcome this problem, Harvey et al. used multi-tile electrodes at very high frequencies,⁵⁶ Chabert et al. used shaped electrode and dielectric lens,⁵⁷ and Schmidt et al. used a lens-shaped circular electrode to improve the plasma uniformity.⁵⁸ The performance of a capacitively coupled plasma source can also be enhanced by applying the static magnetic field perpendicular to the static electric field. The applied magnetic field reduces the radial loss of plasma to the wall, thereby enhancing the density. Apart from enhancing the performance of the CCP, the applied magnetic field provides the interaction mechanism between the RF field and plasma. Hence, the external magnetic field provides additional scope to control the properties of the discharge in order to make it suitable for particular applications.^59–62 In magnetically enhanced reactive ion etching (MERIE), the transverse magnetic field is applied to the capacitively coupled plasma sources to increase the plasma density by reducing the plasma loss across the field lines.⁶¹ You and Chang⁶³ and Lee et al.⁶⁴ studied the transition of electron kinetic properties from non-local to local by applying a static magnetic field to the CCP discharges. Thus, the magnetic field acts as the controlling parameter to tune the plasma parameters such as plasma density and temperature in the discharge to create a region of uniform ion flux suitable for plasma processing. It implies that applying a static magnetic field in the CCP discharges can control the power transfer mechanism. Trieschmann et al.⁶⁵ and Yang et al.^66,67 demonstrated that the application of an asymmetric magnetic field to the capacitively coupled discharge, DC self-bias as well as asymmetric plasma response can be activated even in the geometrically and electrically symmetric system. This phenomenon is known as the magnetic asymmetric effect. This effect can independently control the ion flux and energy. Thus, in the presence of a transverse magnetic field, many physical processes such as E × B drift, axial variation of plasma parameters, mode transition, and power dissipation can occur, influencing the performance and efficiency of the CCP discharge source for industrial applications.

However, the E × B drift in the case of a parallel plate CCP generates the non-uniform plasma density and electron temperature profile.⁶⁸ This configuration affects the performance of parallel plate CCP discharge for industrial applications. Cylindrical electrode (CE) CCP having axisymmetric magnetic fields can increase the plasma density and uniformity by radially confining the electrons. Such a system can be designed using two co-axial cylinders, while an axisymmetric magnetic field can be generated using electromagnets.^69,70

Considering the applications of stable electropositive argon plasma diluted with a small amount of reactive electronegative gas such as oxygen in the microelectronics and semiconductor industry, and the advantages of cylindrical electrode CCP with axisymmetric magnetic field, in the present study, we have studied the negative ion production in argon + oxygen discharge in geometrically asymmetric cylindrical capacitively coupled RF plasma source. By varying the magnetic field strength from 0 to 8 mT, we have studied the variation of plasma parameters such as density (electron, positive, and negative ion), negative ion fraction, and electron temperature with the changes of discharge parameters such as applied RF power and axial magnetic field. In this experiment, we have used RF compensated Langmuir probe to measure the negative ion fraction in the bulk plasma. Langmuir probe has various advantages. It can provide spatially resolved data and reach to the remote locations inside the electrode. Moreover, Langmuir probe is a cost-effective diagnostic that can be easily fabricated in-house.

In this work, we have investigated the possibility of using CCP cylindrical electrode plasma source with an axisymmetric magnetic field for the production of negative ion density argon oxygen mixed plasma. The paper is structured as follows: Sec. II describes the cylindrical electrode CCP plasma source with an axisymmetric magnetic field and the Langmuir probe diagnostics technique adopted for measuring negative ion fraction. Section III is devoted to the results and discussion. Finally, the paper is concluded with the summarization and the outcome of the studies in Sec. IV.

Figure 1 shows the schematic of the plasma chamber, diagnostic setup, and the electrode assembly. The system consists of a large vacuum chamber of 31 cm diameter and 120 cm length. The chamber is evacuated using a Pfeiffer make 700 l/s turbo molecular pump (TMP) backed by a rotary pump up to the base pressure of 2 × 10⁻⁵ mbar. Inside the vacuum chamber, a cylindrical electrode (CE) of 20 cm in length and an inner diameter of 24 cm is placed, which is powered by a 13.56 MHz RF power supply and matching network. At both ends of the powered electrode, there are two annular rings of 19 cm inner and 23 cm outer diameter. Annular rings are followed by two grids, G₁ and G₂, on both ends of the powered electrodes. Grids G₁ and G₂ are electrically isolated using insulator blocks. Grid G₁ is grounded and is a fine grid, whereas grid G₂ is coarse and is electrically floated. The whole electrode assembly is isolated using the ceramic blocks. An axial magnetic field is produced by using two electromagnets. The two electromagnets are kept at 27 cm apart from each other. The electromagnets are powered by a DC power supply having a maximum current rating of 60 A. A more detailed description of the setup and the variation of the axial magnetic field with the changes in the value of direct current applied to the electromagnet is given in Refs. 69 and 70.

The explained electrode assembly is kept in the uniform axial magnetic field region. An RF-compensated Langmuir probe is inserted radially at the axis of the chamber through the small hole, cut at the center of the cylindrical electrode. Pure argon plasma and argon oxygen mixed plasma with varying oxygen percentages from 10% to 30% produced inside the cylindrical electrode at the working pressure of 1 × 10⁻³ mbar. The density variation of different species (electron, positive, and negative ion density) as well as electron temperature and negative ion fraction with the applied magnetic field for different values of RF power at the different concentrations of oxygen gas is investigated in this work.

Plasma parameters are measured using an RF-compensated Langmuir probe described in our previous works.^69,70 The effect of fundamental frequency (13.56 MHz) and its first harmonics (27.12 MHz) is compensated by using a resonant filtering technique. In this method, the resonant circuit is made by connecting self-resonating inductors to the probe tip, which provides high impedance for fundamental (180 kΩ) and first harmonic (25 kΩ). The main probe tip is made up of tungsten wire of 0.2 mm diameter and 5 mm length. A floating auxiliary probe with area greater than the main probe tip is connected to the main probe by using a 1 nF capacitor. The auxiliary probe reduces input RF impedance across the probe sheath. I–V characteristics are obtained under different operating conditions by driving the main probe with an amplified voltage sweep from high voltage amplifier (WMA-300, Falco Systems) and a unity gain isolation amplifier circuit using AD215 isolation amplifier. Orbital Motion Limited (OML) theory is used to calculate the plasma density. In the present study, the magnetic field strength is varied from 0 to 8 mT, which correspond to electron gyro radius from ∼0.05 to 0.4 cm and is larger than the probe radius (0.01 cm). Additionally, the Langmuir probe is inserted across the field lines to reduce the influence of field lines on the plasma by the magnetic field. Therefore, the influence of magnetic field on measured I–V is minimal, and a non-magnetized approximations are valid for the estimation of plasma parameters. The electron temperature is calculated from the logarithmic slope of I–V curve. Plasma potential is derived from either the first or double derivative of the I–V curve.

The chamber is evacuated up to the base pressure of ∼2 × 10⁻⁵ mbar using a turbo molecular pump backed by a rotary pump. All the experiments reported in this work are performed at the 1 × 10⁻³ mbar working pressure. At first, pure argon is injected into the chamber, then subsequently 90% argon and 10% oxygen, 80% argon and 20% oxygen, and finally 70% argon and 30% oxygen are injected into the chamber. The error bars for the plasma parameters are calculated by repeating the experiment at the same discharge conditions. The errors in the electron temperature are then determined by taking multiple sets of linear fit of the semi-log plot, followed by a standard deviation for all the values. Similarly, the technique is repeated for the ion saturation current for the error estimation in the density values.

Figure 2(a) shows the variation of mean positive ion mass [M₊(X)] calculated by using Eq. (3) in argon–oxygen mixed plasma at 20% oxygen concentration in the discharge, and for different values of applied RF power. It is observed that the mean ion mass lies approximately between 45 and 55 × 10⁻²⁷kg. The dominant positive ions present in the discharge are Ar⁺, O⁺, and O₂⁺ ions. Thus, this method estimates the reasonable values of positive ions present in the discharge. Figure 2(b) shows the variation of the mean ion mass with the increasing oxygen concentration from 10% to 30% in the argon–oxygen discharge at the externally applied axial magnetic field of 4 mT. It is observed that mean ion mass decreases slowly and steadily with the increasing oxygen concentration in the discharge. With the increasing oxygen gas, the concentration of O⁺ and O₂⁺ increases and the concentration of Ar⁺ decreases, and Ar⁺ ion is heavier as compared to O⁺ and O₂⁺ ions, the mean ion mass remains approximately similar trend with the externally applied magnetic field at the given RF power and oxygen concentration and decreases steadily with the increasing oxygen gas concentration. However, a small fluctuation in the value of mean ion mass is observed, which may be due to the application of RF power to the electrodes and also due to the limitation of diagnostics. This calculated value of mean ion mass is substituted in Eq. (1) for the calculation of negative to positive ion density ratio (β).

Figures 3(a) and 3(b), respectively, show the variation of electron density and electron temperature with the applied axial magnetic field for 10% oxygen in the discharge. For a low (25–75 W) operating power, the measured plasma density is higher of the order of 10¹⁶ m⁻³ due to enhanced power coupling efficiency in RF discharge. For an unmagnetized case (B = 0), the plasma density varies from ∼8.5 × 10¹⁵ to ∼1.3 × 10¹⁶ m⁻³, with ∼50% rise, when the applied RF power to the electrode is increased from 25 to 75 W. As magnetic field strength increases, the electron density shows a transition from a higher value (below 4 mT) to a lower value (above 4 mT). At low RF powers (25 and 50 W), the transition is sudden, whereas a smooth transition is observed at 75-W RF power.

A transition in the electron density and temperature is related to the Larmor radius compared to the sheath thickness. Figure 4 shows a plot of variation in the sheath thickness and Larmor radius with the increasing value of the applied magnetic field strength for 10% oxygen at 75 W applied RF power. In this plot, Child–Langmuir formula $Rsh=2λDs32 V0Te32$ is used to calculate the sheath thickness.²² Here, λ_Ds is the Debye length and V₀ is the electrode voltage measured using a commercial I–V sensor (Impedans Ltd.) as shown in Fig. 1.

At no magnetic field, the Larmor radius is infinite, so the electrons produced near the sheath are free to diffuse axially. Therefore, the central plasma density is higher. At 1–3 mT of the magnetic field, the Larmor radius is higher than the sheath thickness; hence, the peripheral electrons diffuse to the axial region of the discharge tube, thereby increasing the electron density at the axis of the cylindrical electrode. However, when the magnetic field increases beyond 4 mT, the Larmor radius decreases and becomes lower than the sheath thickness. As a result, the higher energy electrons are confined near the cylinder's radial edge,⁷⁰ decreasing the central plasma density, as shown in Figs. 3 and 5. An approximately similar variation of Larmor radius and sheath thickness is observed for other values of magnetic field and oxygen concentration.

In the present discharge system, the electric field is in the radial direction and the magnetic field is in the axial direction; therefore, E × B drift is produced in the azimuthal direction with the drift velocity given by $vBE=E/B$⁠. The closed E × B drift enhances ionization at the radial edge due to collision. However, with the increase in the magnetic field at the given power, the drift velocity (⁠ $vBE$⁠) and hence the enhanced ionization due to collision decrease. This is another factor responsible for the decrease in plasma density with the increase in the magnetic field at a fixed RF power.

For 1–∼4 eV electron temperatures, collisional energy (ε_c) lost per electron–ion pair created for Ar is less than O, and ε_c of for O₂ is higher as compared to that of O and Ar as shown in Fig. 6. Oxygen gas primarily exists in the molecular form. For 10%, 20%, and 30% oxygen, argon is the dominant gas in the discharge. A slight increase in oxygen gas in argon discharge increases the values of A_eff and $εT$⁠; therefore, as per the power balance equation [Eq. (6)], there is a small decrease in electron density. In other words, we can say, that as oxygen is a molecular gas, higher energy is required for ionization as compared to the argon gas. Therefore, electron density decreases as the oxygen concentration is increased in the argon discharge.

Figure 5(b) shows the variation of plasma temperature with the axial magnetic field at 75 W RF power at 10%, 20%, and 30% oxygen in the discharge. The electron temperature lies between 1 and 2 eV for the applied magnetic field from 0 to 3 mT, but as the magnetic field is increased above 4 mT, the temperature increases to 3–4 eV. For 10% and 20% oxygen in the discharge, the variation of electron temperature with the applied magnetic field is comparable. Since the electron temperature is measured at the axis of the chamber near the inside of the cylindrical electrode, the applied RF influences it; therefore, some small fluctuation in the electron temperature is observed. The fluctuation of the electron temperature may be also possibly due to the limitation of Langmuir probe diagnostics. However, at 30% oxygen, the electron temperature is relatively stable and is slightly lower than at 10% and 20% oxygen. According to the particle balance model, by substituting the value of μ_B, we get $KTe=Kiz2deff2m+(X)$⁠. The ionization rate constant K_iz for Ar, O, and O₂ is approximately similar for electron temperature between 1 and 3 eV.⁷³ The effective plasma length d_eff decreases with the increasing oxygen concentration.⁷⁶ Also, mean positive ion mass [M₊(X)] decreases slightly with the oxygen concentration. Due to the slight decrease in d_eff and M₊(X) with the increasing oxygen concentration, the value of electron temperature decreases with the oxygen concentration in the discharge.

Figure 7(a) shows the plot of the variation of plasma potential with the axial magnetic field for 20% of oxygen in the discharge. It is observed that the plasma potential is very high when the applied RF power is 25 W as compared to 50 and 75 W, and the potential decreases moderately with the axially applied magnetic field. There is a higher loss of electrons to the walls at lower RF power due to lower electron density and/or higher temperature. In order to reduce this loss and to maintain the quasineutrality in the plasma volume, the plasma potential is higher. However, when the axial magnetic field is increased, the plasma losses to the wall decrease due to the radial confinement of electrons. As a result, the plasma potential decreases gradually with the axial magnetic field. However, for 50 and 75 W RF power, a decrease in the plasma potential with the axial magnetic field is not substantial, when compared to 25 W RF power. Figure 7(b) shows the variation of plasma potential as a function of the axial magnetic field at 75 W applied RF power for 10%, 20%, and 30% oxygen in the discharge. An interesting observation is that, at the lower value of the axial magnetic field below 4 mT, the plasma potential for 10% oxygen is highest and decreases subsequently for 20% and 30% oxygen in the discharge. However, at the magnetic field above 4 mT, the trend completely reverses. The variation of the trend of plasma potential with the changes in the oxygen concentration in the discharge is due to changes in the composition of the ions, i.e., Ar⁺, O⁺, O₂⁺ escaping to the reactor's walls, as well as due to the changes in the values of negative ion fraction with the changes in the oxygen percentage.

Figure 8 shows the variation of the difference between plasma potential and floating potential (V_P–V_F) as the function of applied magnetic field strength when varied from 0 to 8 mT for the applied power of 20% oxygen at 25, 50, and 75 W. V_P–V_F follows approximately the similar trends as the electron temperature [Fig. 3(b)], i.e., lower in the magnetic field regime below 4 mT and increasing thereafter. This is due to the fact that V_P–V_F is proportional to the electron temperature; however, in highly electronegative plasma, there is considerable deviation of V_P–V_F from the electron temperature. However, in our case, the plasma is weakly electronegative due to a lower percentage of oxygen in the discharge and thus V_P–V_F follows the same trend as electron temperature. These results show that the modification of sheath due to the presence of negative ions will have negligible influence on the estimated plasma parameters.

Figures 9(a)–9(c) show the plot of the variation of positive ion, electron, and negative ion density with the axial magnetic field at 75 W of applied RF power when the oxygen concentration in the argon discharge is varied from 10% to 30%. In all three cases, the variation of the densities with the axial magnetic field follows similar trends. The electron density closely follows positive ion density, whereas negative ion density is much lower. In the lower magnetic field regime (below 4 mT), the gap between the positive ion density and the electron density is more, which shows that the negative ion density is higher to maintain quasineutrality. This is due to lower electron temperature in this range [Figs. 3(b) and 5(b)] as the negative ions are generally formed by the dissociative attachment of low energy electrons to the metastable oxygen molecule, i.e., $e+O2M→O−+O$⁠. However, for the magnetic field higher than 4 mT, the electron density comes very close to the plasma density, indicating the lower negative ions in this region. However, at higher oxygen concentrations in the discharge, the density of all the charged particles is lower; the reason for this effect is already explained above. Therefore, to estimate the qualitative variation of negative ions in the discharge with the oxygen concentration, it is necessary to study the variation of negative ion fraction (⁠ $α=n−/ne$⁠).

Figure 10 shows the variation of negative ion fraction (α) with the axial magnetic field at 75 W of applied RF power when the oxygen concentration in the discharge is varied from 10% to 30%. The negative ion fraction is higher at the lower values of the applied axial magnetic field, as can be seen in Fig. 10. However, we have observed that the slightly higher negative ion fraction at the lower concentration of oxygen gas (10%) for some magnetic field strength. This is due to the fact that an optimum concentration of oxygen produces higher dissociation of oxygen molecules, thus changing the ratio of atomic to molecular oxygen density. Shindo et al. also reported that the optimum value of α in microwave discharges in the argon–oxygen mixture at the oxygen concentration of about 15%.¹⁵ Therefore, it is reasonable to expand the present study to include the mechanism of the production of negative ions theoretically as well as the effects of the axial magnetic field in detail in future studies.

This work investigates plasma density including negative ions, plasma potential, and electron temperature in electronegative argon–oxygen discharge produced in a cylindrical capacitively coupled plasma discharge with an axisymmetric magnetic field. The operating conditions such as applied RF power (25–75 W), oxygen gas concentration (10%–30%), and magnetic field strength (0–8 mT) are varied. The negative ion density is measured using the RF-compensated Langmuir probe from the saturation current ratio method when a small quantity of oxygen gas is added to the pure argon discharge. The experimental results show an increase (∼50%) and a decrease (∼40%) in the plasma density with a rise in RF power and oxygen concentration, respectively. It is observed that the density of different species remains constant below 4 mT and thereafter decreases drastically. On the other hand, the electron temperature and plasma potential with respect to floating show the opposite trend. This observed behavior is due to the confinement of the high-energy electrons at the radial edge at a higher magnetic field since Larmor radius decreases below sheath thickness. Plasma potential remains higher at low RF power in order to reduce the loss of electrons to the walls and to maintain the quasineutrality in the plasma bulk. In the lower magnetic field regime, i.e., below 4 mT, the gap between the positive ion density and the electron density is higher, indicating a higher negative ion density. Whereas for the higher magnetic field regime (above 4 mT), the gap between the positive ion and the electron density decreases, indicating a decrease in the negative ion density. Negative ion fraction is higher at lower magnetic fields and decreases with the increase in the magnetic field. We have observed a slightly higher negative ion fraction at 10% concentration of oxygen gas in the mixture. It shows that α is strongly dependent on the gas mixture and that optimizing the composition of the gases is necessary for optimal negative ion fraction. Thus, in such discharge systems, the axial magnetic field acts as the controlling parameter to vary species density (positive ion, electron, and negative ions) and temperature as required for plasma processing applications. In conclusion, this experiment provides the ranges of magnetic field strengths for plasma confinement in cylindrical CCP discharges with the axisymmetric magnetic field, i.e., the field regime in which the density of the different species including negative ions is higher in such discharges.

This work was supported by the Department of Atomic Energy, Government of India, and Science and Engineering Research Board (SERB), Core Research Grant No. CRG/2021/003536.

The authors have no conflicts to disclose.

Swati Dahiya: Methodology (equal); Resources (lead); Writing – review & editing (supporting). Narayan Sharma: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (lead); Writing – review & editing (equal). Shivani Geete: Data curation (equal); Formal analysis (equal). Sarveshwar Sharma: Conceptualization (equal); Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal). Nishant Sirse: Conceptualization (equal); Data curation (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Shantanu Karkari: Conceptualization (equal); Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available within the article.