In this study, a novel inductively coupled plasma (ICP) system is proposed. It comprises a segmented dielectric window and a metal frame. For the proposed ICP system, a thin window can be designed, thereby compensating for the power loss caused by the metal frame. The proposed ICP system has two potential advantages: it can enhance the controllability of the gas flow field and it can reduce the capacitive power coupling. These characteristics enable the superior uniformity and reliable operation of ICP systems for semiconductor processes. The characteristics of the proposed ICP system are investigated using three-dimensional fluid self-consistent plasma simulations and experiments. The proposed ICP system exhibits performance similar to that of the conventional ICP system currently used in etching and deposition processes.

In semiconductor processing, such as three-dimensional (3D) vertical-NAND channel etch,1,2 dry chemical etching for gate-all-around structure formation,3–5 atomic layer etching,6,7 or deposition,8–10 miniaturization, high integration, performance, and consistency are required. To meet these demands, good basic process parameters need to be ensured, such as etch uniformity, selectivity, rate, and profile. Inductively coupled plasma (ICP) is a plasma that enables discharge at low pressures in the range of several mTorr to several tens of mTorr and has a density of 1016–1017 m−3, and it is used in the semiconductor etching process. Most ICP equipment used in semiconductor processing employs a planar antenna. Structurally, the planar antenna ICP comprises a dielectric window between the antenna coil and plasma. For equipment targeting a 300 mm wafer, the diameter of the chamber needs to be large and the thickness of the dielectric window needs to be designed to be ∼30 mm to withstand the pressure difference between the vacuum inside the chamber and the atmospheric pressure. Owing to the material limitations of the dielectric window, the shower-head-type gas injection system used for gas flow field control cannot be used. Therefore, to form a separate gas input system on the side of the chamber, an additional structure in the form of a toroidal ring on the upper part of the chamber or an additional structure in the center of the window is used. Therefore, the gas distribution of ICP is nonuniform and adversely affects the ICP process uniformity.

To easily control the ICP gas flow field and improve the ICP process uniformity, this study proposes a novel ICP system with a segmented dielectric window. In the proposed system, a metal frame supports the segmented dielectric window to withstand the pressure difference between the vacuum chamber and atmosphere. Owing to the metal frame, the system has the following advantages: a hole-array-type gas input device similar to the shower-head-type gas injection system can be installed on the metal frame and it functions as a Faraday shield. Additionally, the thickness of the dielectric window can be reduced. The power transfer efficiency increases with decreasing dielectric window thickness due to the increase in the coupling coefficient between the antenna coil and the plasma. In contrast, in the region where the metal frame exists, the time-varying induced magnetic field generated by the antenna coil current generates an induced electric field. The induced electric field results in a diamagnetic surface current on the metal frame surface, and the electromagnetic field does not penetrate more than the skin depth. Hence, the most important aspect of the system is the reduction in the amount of radio frequency (RF) power loss due to the generation of the metal frame surface current, which is compensated by the increase in the power transfer efficiency because of the thin dielectric window thickness. Moreover, some of the induced metal frame surface current flows along the rear surface to the backside-exposed plasma. Here, the power transfer through secondary inductive coupling is enabled because the plasma sheath can act like an RF dielectric window. However, compared to a conventional ICP with the same window thickness, the power transfer efficiency from the antenna coil to the plasma is inevitably low for the proposed ICP. In this study, the plasma characteristics of a segmented dielectric window ICP (SDWICP) and the conventional ICP are comparatively analyzed via 3D electromagnetic fluid simulations and experiments.

The increase in the power transmission efficiency of SDWICP and ICP as well as the power loss due to the metal frame with respect to the change in the dielectric window thickness because of the secondary inductive coupling component of the eddy current induced on the bottom side of the metal frame is investigated through 3D electromagnetic fluid simulations (COMSOL Multiphysics). Figure 1 displays the basic model of the SDWICP and ICP systems used in the simulation. The system has a diameter of 500 mm and a height of 380 mm. The chuck has a diameter of 340 mm and a height of 50 mm. Figures 2(a) and 2(b) exhibit the dielectric window and metal frame of SDWICP and the dielectric window and toroidal-shape-type gas injection frame of ICP, respectively. Figure 2(c) shows the schematic of the antenna coil. In SDWICP, the dielectric window, which is composed of alumina (Al2O3) and has a relative permittivity of 9.5, is divided into 12 pieces with an inner diameter of 120 mm and an outer diameter of 440 mm. The metal frame may bend due to the pressure difference between the vacuum inside the chamber and the external atmospheric pressure. Thus, a circular region is placed in the middle of the metal frame to prevent bending. The metal frame is composed of aluminum with an electrical conductivity of 3.774 × 107 S/m. The ratio of the area of the dielectric window to that of the metal frame is 4:1. The antenna coils above the window have a 3-turn and 2-branch spiral, yielding a circular current path. For SDWICP, gas is injected into the plasma bulk from the 20 holes on the bottom side of the metal frame and exhausted through the bottom of the chamber. However, for ICP, gas is injected into the plasma edge from the toroidal-shaped structure at the top of the chamber. To generate plasma RF, 13.56 MHz power is applied to the antenna coils. The initial temperature of the reactor and gas temperature are assumed to be 300 K; the gas mass flow is 100 sccm; and the pressure is 20 mTorr. Argon gas is used as the working gas. In the simulations, electron, argon neutral (Ar), argon resonance state (Arr), argon metastable state (Ars), argon excited state of 4p (Arp), and argon ion (Ar+) are considered; the reactions of these species are given in Table I. The 3D simulation model employed herein couples three physical models: the electromagnetic field, plasma, and laminar flow models. The electron density, mean electron momentum, and mean electron energy are solved using the particle conservation and energy conservation equations. The momentum conservation equation for electrons and ions is used by the drift–diffusion approximation. The electron density and electron energy density are expressed using Eqs. (1) and (2), respectively,11,12
net+Γe=Re,
(1)
nεt+Γϵ+EΓe=Sen,
(2)
where ne and nɛ are the electron density and electron energy density, respectively, Re is either a source or sink of electrons, and Sen is the energy loss or gain due to inelastic collisions. In Eqs. (1) and (2), the flux of electrons Γe and the flux of electron energy Γε are approximated as follows:
Γe=μeEneDene,
(3)
Γϵ=μϵEnϵDϵnϵ,
(4)
where μe and μϵ are the electron and electron energy mobilities, respectively. De and Dϵ are the diffusivity of the electron and electron energy, respectively. The electric field E and potential V required for the flux calculation are obtained using the Poisson equation as follows:
ϵ0V=ρ,
(5)
E=V.
(6)
FIG. 1.

Schematic of ICP and SDWICP.

FIG. 1.

Schematic of ICP and SDWICP.

Close modal
FIG. 2.

Schematic of the dielectric window; gas injection layout of (a) SDWICP and (b) ICP; and (c) antenna coil design. The blue region represents the dielectric region; the gray region represents the metal, and the yellow regions represent the gas injection holes.

FIG. 2.

Schematic of the dielectric window; gas injection layout of (a) SDWICP and (b) ICP; and (c) antenna coil design. The blue region represents the dielectric region; the gray region represents the metal, and the yellow regions represent the gas injection holes.

Close modal
TABLE I.

Reactions for argon discharge. All reactions shown in Table I are obtained from Ref. 16. Reproduced with permission from Cha et al., J. Phys. D. 54, 165205 (2021). Copyright 2021, Institute of Physics (IOP).

No.ReactionRate coefficient (m3 s−1)
e + Ar → e + Ar 3.9 × 10−13 exp(−4.6/Te
e + Ar → e + Ars 5×1015Te0.74exp11.56/Te 
e + Ar → e + Arr 5×1015Te0.74exp11.56/Te 
e + Ar → e + Arp 1.4×1014Te0.71exp13.2/Te 
e + Ar → e + e + Ar+ 2.34×1014Te0.59exp(17.44/Te) 
e + Ars → e + Arr 2 × 10−13 
e + Ars → e + Arp 8.9×1013Te0.51exp1.59/Te 
e + Ars → e + e + Ar+ 6.8×1015Te0.67exp4.2/Te 
e + Ars → e + Ar 4.3×1016Te0.74 
10 e + Arr → e + Ar 4.3×1016Te0.74 
11 e + Arr → e + Ars 3 × 10−16 
12 e + Arr → e + Arp 8.9×1013Te0.51exp1.59/Te 
13 e + Arr → e + e + Ar+ 6.8×1015Te0.67exp4.2/Te 
14 e + Arp → e + Arr 3×1013Te0.51 
15 e + Arp → e + Ars 3×1013Te0.51 
16 e + Arp → e + e + Ar+ 1.8×1013Te0.61exp2.61/Te 
17 e + Arp → e + Ar 3.9×1016Te0.71 
18 Ars + Ars → Ar + Ar 2 × 10−13 
19 Ars + Ars → e + Ar + Ar+ 6.4 × 10−16 
20 Ars + Arr → e + Ar + Ar+ 2.1 × 10−15 
21 Ar + Ars → Ar + Ar 2.1 × 10−21 
22 Arp + Arp → e + Ar + Ar+ 5 × 10−16 
23 Arr → Ar + hν 5 × 106 
24 Arp → Ar + hν 3.2 × 107 
25 Arp → Ars + hν 3 × 104 
26 Arp → Arr + hν 3 × 104 
No.ReactionRate coefficient (m3 s−1)
e + Ar → e + Ar 3.9 × 10−13 exp(−4.6/Te
e + Ar → e + Ars 5×1015Te0.74exp11.56/Te 
e + Ar → e + Arr 5×1015Te0.74exp11.56/Te 
e + Ar → e + Arp 1.4×1014Te0.71exp13.2/Te 
e + Ar → e + e + Ar+ 2.34×1014Te0.59exp(17.44/Te) 
e + Ars → e + Arr 2 × 10−13 
e + Ars → e + Arp 8.9×1013Te0.51exp1.59/Te 
e + Ars → e + e + Ar+ 6.8×1015Te0.67exp4.2/Te 
e + Ars → e + Ar 4.3×1016Te0.74 
10 e + Arr → e + Ar 4.3×1016Te0.74 
11 e + Arr → e + Ars 3 × 10−16 
12 e + Arr → e + Arp 8.9×1013Te0.51exp1.59/Te 
13 e + Arr → e + e + Ar+ 6.8×1015Te0.67exp4.2/Te 
14 e + Arp → e + Arr 3×1013Te0.51 
15 e + Arp → e + Ars 3×1013Te0.51 
16 e + Arp → e + e + Ar+ 1.8×1013Te0.61exp2.61/Te 
17 e + Arp → e + Ar 3.9×1016Te0.71 
18 Ars + Ars → Ar + Ar 2 × 10−13 
19 Ars + Ars → e + Ar + Ar+ 6.4 × 10−16 
20 Ars + Arr → e + Ar + Ar+ 2.1 × 10−15 
21 Ar + Ars → Ar + Ar 2.1 × 10−21 
22 Arp + Arp → e + Ar + Ar+ 5 × 10−16 
23 Arr → Ar + hν 5 × 106 
24 Arp → Ar + hν 3.2 × 107 
25 Arp → Ars + hν 3 × 104 
26 Arp → Arr + hν 3 × 104 
Assuming the collision cross section to be a Maxwell distribution, the reactions because of the electron impact are determined as the electron distribution function. The electromagnetic field analysis solves the Ampere law in the form of a vector potential as follows. Since the ICP inductive mode (H-mode) is the basic assumption, the curl-free electric field and displacement current component are not included in the calculation,
××A=μ0Jc+σE.
(7)
Here, μ0 is the free space permeability and Jc denotes the external antenna coil current. The induced current component in the plasma region σE is obtained using the induced electric field E=jωA, which is expressed by Faraday’s law and plasma conductivity,11,
σplasma=e2nemeνm+jω,
(8)
where e, ne, me, νm, and ω are the electron charge, electron density, electron mass, electron collision frequency, and driving frequency of the RF current, respectively. The magnetic field boundary conditions assume that the antenna region boundary and plasma chamber wall are conductors (antimagnetic properties). The following impedance boundary conditions are used for the metal frame where the antenna surface and window are placed,13,14
μ0μrϵ0ϵrjσ/ωn̂×H+n̂En̂=n̂Esn̂Es.
(9)
This boundary condition implies that the magnetic field discontinuity on the conductor surface is determined by the surface current induced in the conductor. Here, Es is the source electric field and n̂ is a unit vector that indicates the interface direction. Through this boundary condition, the surface current and power consumption in the antenna coil and metal frame as well as the power input to the plasma can be solved without placing a very thin skin layer in the calculation region. The argon gas flow condition is applied according to the conservation equation of mass and momentum based on the compressible Navier–Stokes equation,15 
ρu=0,
(10)
ρut+uu=pI+νu+uT23νuI,
(11)
where p is the pressure, T is the absolute temperature, ν is the dynamic viscosity of the fluid, and I is an identity matrix.

Figure 3 displays the power transmitted to the plasma based on the thickness of the dielectric window of SDWICP and ICP. In this simulation, the RF current is 50 A. Generally, the control variable of the equipment in semiconductor processing is RF power and not RF current. However, to maintain the RF input power at a constant value in the simulation, the input power density must be integrated and compared with the desired input power before adjusting the source current. When this process is repeated, the calculation time in the 3D simulation increases; therefore, this study uses RF current and not RF power as the control variable. As mentioned earlier, SDWICP has properties that enable a significant reduction in the dielectric window thickness compared to the conventional ICP; therefore, improving the power transfer efficiency by decreasing the window thickness is important in equipment design. An SDWICP system with an ∼15 cm-thick dielectric window can achieve a similar level of efficiency as the ICP dry etching equipment that is currently used in semiconductor manufacturing with an ∼25–30 mm-thick window.

FIG. 3.

Input power as a function of the dielectric window thickness.

FIG. 3.

Input power as a function of the dielectric window thickness.

Close modal
Figures 4(a)4(f) show the electron density distribution, electron temperature, and plasma potential for SDWICP and ICP systems with similar efficiencies. The electron density distribution and electron temperature vary little between the two systems. The electron density and temperature are primarily determined by the energy and particle conservations, respectively. In addition, the electron temperature depends on the pressure; the electron density is proportional to the input power. Figure 5(a) shows the cross section cut along the antenna coil in the theta axis. In this structure, the directions of the RF current of the antenna coil and metal frame are parallel. The dielectric window is present between the metal frames. The time-varying magnetic flux density generated by the RF current of the antenna coil can be expressed by the following Ampere law:
×B=××A=μ0Jc+Dt.
(12)
FIG. 4.

Plasma potential of SDWICP (top) and ICP (bottom) at 20 mTorr and 50 A. ICP has a dielectric thickness of 25 mm, and SDWICP has a dielectric thickness of 15 mm.

FIG. 4.

Plasma potential of SDWICP (top) and ICP (bottom) at 20 mTorr and 50 A. ICP has a dielectric thickness of 25 mm, and SDWICP has a dielectric thickness of 15 mm.

Close modal
FIG. 5.

(a) Cross section cut along the antenna coil in the theta axis. (b) Distribution and direction of the induced current generated in metal frame facing the antenna coil. (c) Distribution and direction of the induced current generated in the metal frame facing the plasma.

FIG. 5.

(a) Cross section cut along the antenna coil in the theta axis. (b) Distribution and direction of the induced current generated in metal frame facing the antenna coil. (c) Distribution and direction of the induced current generated in the metal frame facing the plasma.

Close modal

Figure 5 shows the distribution and direction of the induced current generated in the metal frame facing the antenna coil (b) and facing the plasma (c). The surface current that forms on the metal frame facing the plasma flows in a counterclockwise direction. This direction is the same as that of the RF current flowing in the original antenna coil. The induced electric field is generated again using this current. SDWICP and ICP exhibit almost the same characteristics, except for the power loss caused by the area of the metal frame, which accounts for 20% of the ICP dielectric window. Power transmission through the dielectric window is an important power transmission path. Therefore, in simple calculations, SDWICP with a dielectric thickness of 15 mm and ICP with a dielectric thickness of 25 mm have a difference in power efficiency of ∼20% of the area of the metal frame. However, as shown in Fig. 3, the power absorbed by the plasma differs by ∼10%. The plasma formation is affected by the secondary inductive coupling caused by the surface current of the bottom surface of the metal frame.

Figure 6 displays the schematics of the experimental setup, which comprises an antenna coil, a segmented dielectric window, and a metal frame. The plasma is diagnosed using an RF-compensated single Langmuir probe.17 Tungsten wire is used as the probe tip, with a diameter of 0.1 mm and length of 5 mm. The Langmuir probe schematic is shown in Fig. 7 and is compensated using an RF choke coil and a capacitor to reduce the RF effect. In this system, the gas exhaust line is located on the left side of the chamber and has an asymmetrical gas flow line. To effectively examine the difference in the plasma characteristics of the gas flow field between SDWICP and ICP, experiments are conducted in an environment with a symmetrical structure. In addition, the properties are remarkable when fluorine-based molecular gases are employed in etching processes rather than gases with a good diffusivity, such as argon. However, since this study examines the basic plasma characteristics of SDWICP and ICP, argon gas is used in the system.

FIG. 6.

Schematic of the experimental setup.

FIG. 6.

Schematic of the experimental setup.

Close modal
FIG. 7.

Schematic of the RF-compensated single Langmuir probe.

FIG. 7.

Schematic of the RF-compensated single Langmuir probe.

Close modal

Figures 8(a)8(c) illustrate the electron temperature, electron density, and plasma potential of SDWICP and ICP with respect to the pressure and input power at the center point of the chamber. SDWICP and ICP exhibit similar trends. The simulation results show that the power transfer efficiency of SDWICP is higher than that of ICP under corresponding experimental conditions (the window thickness of SDWICP and ICP are 15 and 25 mm, respectively). However, according to the experimental results, the electron density of ICP is higher than that of SDWICP. This difference between the simulation and experimental results is observed because the fluid ICP simulation does not consider collisionless heating.

FIG. 8.

(a) Electron temperature, (b) electron density, and (c) plasma potential with input power and pressure at the center of the chamber. (d) Electron density distribution in the radial direction at 1 kW input power. ICP has a dielectric thickness of 25 mm, and SDWICP has a dielectric thickness of 15 mm.

FIG. 8.

(a) Electron temperature, (b) electron density, and (c) plasma potential with input power and pressure at the center of the chamber. (d) Electron density distribution in the radial direction at 1 kW input power. ICP has a dielectric thickness of 25 mm, and SDWICP has a dielectric thickness of 15 mm.

Close modal

At the 5 mTorr condition, the electron density of SDWICP is not significantly different from that of ICP. The main heating mechanism of ICP is heating due to collision, but collisionless heating dominates under low-pressure conditions.11,18,19 Collisionless heating is caused by inhomogeneous RF electric fields in the inductive mode (H-mode) of ICP. Because of the structure of SDWICP, the dielectric window and metal frame are repeatedly present in the azimuthal direction. The fields generated under the dielectric window and metal frame are inhomogeneous. It is speculated that power absorption by collisionless heating is enhanced in SDWICP and exhibits a relatively higher electron density at lower pressures. For both SDWICP and ICP, the electron temperature decreases with increasing pressure and power. At lower input powers, the electron temperature of SDWICP is lower than that of ICP. The metal frame of SDWICP operates as a Faraday shield. The low electron temperature near the dielectric window in SDWICP probably stems from the suppression of the capacitive power coupling.

Figure 8(d) shows the electron density distribution in the radial direction of SDWICP and ICP. The electron density of ICP is slightly higher than that of SDWICP. For ICP, the maximum and minimum values of density in the radial direction are 2.8 × 1017 and 2.41 × 1017 m−3, respectively, and for SDWICP, they are 2.01 × 1017 and 1.87 × 1017 m−3, respectively. The nonuniformity of the electron density is expressed as follows:
NonUniformity%=nmaxnminnmax+nmin×100.
(13)

The nonuniformity of ICP and SDWICP is calculated to be 7.48% and 3.6%, respectively. The nonuniformity of SDWICP is lower than that of ICP. This difference in nonuniformity exists even when the same antenna coil is used, and the gas injection system of SDWICP affects the plasma uniformity. Argon has a good diffusivity; therefore, the influence of the gas flow field is relatively low. Nevertheless, given these data, the use of molecular gas results in a considerable difference between SDWICP and ICP in terms of plasma uniformity.

In this study, ICP with a segmented dielectric window, considering its application as semiconductor etching equipment, was proposed and its basic characteristics were investigated through self-consistent plasma simulations and experiments. Owing to the RF power loss caused by the metal frame, SDWICP exhibits a lower electron density than ICP. However, owing to the metal frame, a thin dielectric window thickness can be designed, which sufficiently compensates for the RF power loss. Additionally, the RF power loss is compensated by the increase in the collisionless heating due to the nonuniformity of the field and the secondary inductive coupling caused by the eddy currents induced in the metal frame. The experimental results showed that SDWICP and ICP exhibit similar electron densities at the low pressure of 5 mTorr. When the pressure was increased to 10 and 20 mTorr, the difference in the electron densities between SDWICP and ICP increased. However, when the pressure was 20 mTorr, the electron density difference between the two systems was not high (∼25%). The plasma uniformity showed that SDWICP has better characteristics than ICP. The difference in plasma uniformity between SDWICP and ICP stems from the uniformity of the flow field. In SDWICP, the control of the flow field through the gas hole array of the metal frame improves the plasma uniformity. Considering that the etching process is mainly performed at very low pressures of several mTorr, the electron densities of SDWICP and ICP are similar and the electron temperature of SDWICP is lower than that of ICP. Therefore, SDWICP is suitable for etching processes.

This study was supported by Samsung Electronics Co., Ltd., under Grant No. IO221006-02716-01 and a Korean Institute for Advancement of Technology (KIAT) grant funded by the Korean Government (Grant No. P0012451, The Competency Development Program for Industry Specialist).

The authors have no conflicts to disclose.

Sang-Woo Kim: Conceptualization (equal); Data curation (lead); Formal analysis (equal); investigation (equal); Methodology (equal); Software (lead); Validation (lead); Visualization (lead); Writing – original draft (lead). Ju-Hong Cha: Investigation (equal); Methodology (equal). Sung-Hyeon Jung: Data curation (support); Investigation (support); Writing - review & editing (support). SeungBo Shim: Funding acquisition (equal). Chang-Ho Kim: Funding acquisition (equal). Ho-Jun Lee: Conceptualization (equal); Investigation (equal); Project administration (lead); Supervision (lead); Writing – original draft (supporting); Writing – review & editing (lead).

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

1.
T.
Reiter
,
X.
Klemenschits
, and
L.
Filipovic
,
Solid-State Electron.
192
,
108261
(
2022
).
2.
S.
Wang
et al,
ECS Trans.
92
,
27
(
2019
).
3.
E.
Eustache
et al,
Semicond. Sci. Technol.
36
,
065018
(
2021
).
4.
A.
Agarwal
and
M. J.
Kushner
,
J. Vac. Sci. Technol., A
27
,
37
(
2009
).
5.
J. J.
Gu
,
H.
Wu
,
Y.
Liu
,
A. T.
Neal
,
R. G.
Gordon
, and
P. D.
Ye
,
IEEE Electron Device Lett.
33
,
967
(
2012
).
6.
C. M.
Huard
,
S. J.
Lanham
, and
M. J.
Kushner
,
J. Phys. D: Appl. Phys.
51
,
155201
(
2018
).
7.
S. S.
Kaler
,
Q.
Lou
,
V. M.
Donnelly
, and
D. J.
Economou
,
J. Phys. D: Appl. Phys.
50
,
234001
(
2017
).
8.
A.
Haider
,
P.
Deminskyi
,
T. M.
Khan
,
H.
Eren
, and
N.
Biyikli
,
J. Phys. Chem. C
120
,
26393
(
2016
).
9.
M.
Hirayama
,
A.
Teramoto
, and
S.
Sugawa
,
J. Vac. Sci. Technol., A
38
,
032408
(
2020
).
10.
C.
Vallée
,
M.
Bonvalot
,
S.
Belahcen
et al,
J. Vac. Sci. Technol., A
38
,
033007
(
2020
).
11.
M. A.
Lieberman
and
A. J.
Lichtenberg
,
Principles of Plasma Discharges and Materials Processing
,
2nd ed.
(
Wiley Interscience
,
New York
,
2005
).
12.
G. J. M.
Hagelaar
and
L. C.
Pitchford
,
Plasma Sources Sci. Technol.
14
,
722
(
2005
).
13.
AC/DC Module User’s Guide, COMSOL Multiphysics v. 6.0,
COMSOL AB
,
Stockholm, Sweden
,
2021
.
14.
T. B. A.
Senior
,
Appl. Sci. Res.
8
,
418
436
(
1960
).
15.
D. J.
Acheson
,
Elementary Fluid Dynamics
(
Oxford University Press
,
1990
), p.
205
.
16.
J.-H.
Cha
et al,
J. Phys. D
54
,
165205
(
2021
).
17.
F. F.
Chen
,
Phys. Plasmas
8
,
3029
(
2001
).
18.
C.-W.
Chung
and
H.-Y.
Chang
,
Appl. Phys. Lett.
80
,
1725
(
2002
).
19.
S.-H.
Seo
,
C.-W.
Chung
, and
H.-Y.
Chang
,
Surf. Coat. Technol.
131
,
1
(
2000
).