betaFeedback

WorksPhysics and Astronomy

Nonlinear skin effect regime when a radio frequency electromagnetic field penetrates into a background plasma

Abstract

Two-dimensional, electromagnetic particle-in-cell simulations are employed to study particle kinetics and power deposition in the skin layer when a Radio Frequency (RF) electromagnetic field penetrates into a background plasma. We identify a new regime at low frequency (∼MHz\sim\mathrm{MHz}) and low pressure, where the motion of electrons can be highly nonlinear in the skin region. Through most of the RF cycle, the electrons are trapped in the effective potential formed by the vector and electrostatic potentials, with energy deposition being small and magnetic moment μμ no longer being an adiabatic invariant. However, for a brief period around the null of the oscillating magnetic field, the electrons get detrapped, causing a jet-like current penetrating into the bulk plasma. During these brief periods, the power deposition becomes high, exhibiting a periodic burst nature. Based on kinetic theory, we provide analytical expressions for the plasma current and energy deposition in the new regime. A criterion for transition between the newly identified low-frequency, periodic-burst regime and the usual anomalous non-local skin effect regime is proposed and verified.

Introduction—The nonlinear interaction between a time-varying Radio Frequency (RF) electromagnetic field and the background plasma has long been a topic of great interest, both in low temperature plasmas [1, 2, 3, 4, 5, 6, 7, 8] and space plasmas [9, 10, 11, 12, 13, 14, 15]. Studying the physics behind the skin effect in this interaction is crucial for understanding laboratory plasma generation systems [1, 16], such as Capacitively Coupled Plasma (CCP) [17, 18, 19, 20, 21, 22, 23, 24, 25] and Inductively Coupled Plasma (ICP) discharges [26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37]. The most classical skin effect in plasma refers to the condensation of electron current near the surface [1]. When an electromagnetic field penetrates into a background plasma, the classical skin depth is δ∼c/ωpe\delta\sim c/\omega_{pe}, where cc is the speed of light and ωpe\omega_{pe} is the electron plasma frequency [1, 38]. This theory assumes a local interaction between electromagnetic field and the plasma, describing the plasma skin behavior well at high gas pressure [39, 40, 41, 42]. Driven by experimental demands to achieve atomic-scale precision, plasma generation systems tend to operate at low gas pressure (few mTorr) [3, 19, 21, 23, 43, 44, 45, 46, 47, 48, 49, 50, 51], making it crucial to understand the underlying electron dynamics and the associated skin effects [52, 53].

For low-pressure, high-frequency (f=13.56f = 13.56 MHz or higher) plasma generation systems, the skin effect was found to be of anomalous type [3, 7, 44, 45, 46, 47, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75], meaning that the skin depth, δ\delta, is determined by non-local kinetics as δ∼(vthc2/ωωpe2)1/3\delta\sim(v_{\mathrm{th}}c^2/\omega\omega_{pe}^2)^{1/3} instead of δ∼c/ωpe\delta\sim c/\omega_{pe}, where vthv_{\mathrm{th}} is electron thermal velocity, ω\omega is the driving frequency. Electron heating in the anomalous regime is mainly due to wave-particle resonance of low-energy electrons, which can be treated similarly to the Landau damping [30, 32, 52, 58, 76, 77, 78, 79]. The anomalous regime was also called the non-local regime because over one RF period the electrons travel a significant distance d∼vth/ω2+ν2>δd \sim v_{\mathrm{th}/\sqrt{\omega^2+\nu^2}} > \delta [58], where ν\nu is the collision frequency. As a result, the plasma current is a non-local function of the RF electric field. Multiple modeling studies have been performed for the plasma generation systems in this regime, utilizing the non-local kinetic approach [30, 52, 80, 81, 82, 83, 84, 85] to predict the spatial distribution of the plasma current and power deposition. At lower frequency (∼1\sim1MHz), however, it is shown in experiments that the skin effect in plasma generation systems manifests behavior of nonlinear type [4] rather than anomalous type, which is featured by the second harmonics in electron current and electrostatic potential [3, 4, 8, 46].

However, due to the lack of self-consistent kinetic 2D modeling, the physics of the nonlinear skin effect at low frequency [3, 4] is not sufficiently understood, because in experiments it is difficult to perform detailed measurements of plasma properties and electromagnetic fields within the skin layer and correspondingly deduce particle kinetics [7, 54]. Although it was previously shown that at low frequencies the particle dynamics could become highly nonlinear [86, 87], those studies could not be applied here because the RF magnetic field and electrostatic potential need to be determined self-consistently. Existing kinetic simulations did not focus on the nonlinear skin effect at low RF frequency [81, 88, 89, 90, 91, 92].

We demonstrate for the first time that a strongly nonlinear skin effect regime with periodic bursty energy deposition presents at low RF frequency. The electrons are trapped in effective potential, with magnetic moment μ\mu no longer being an adiabatic invariant. Detrapping occurs when the RF magnetic field passes through zero, and electrons quickly escape the area near the coil to form a current jet, causing periodic bursts in the energy deposition. This nonlinear skin effect causes a higher third harmonics in the induced electron current. A kinetic theory is developed and predicts well the plasma current in this bursty regime. To our knowledge, this work presents the first comprehensive 2D PIC simulations for the nonlinear skin effect in the interaction between plasma and RF electromagnetic field.

Method – We take ICP discharge system as an example to study the nonlinear skin effects. We implemented the electromagnetic Darwin scheme to EDIPIC-2D [49, 50, 93, 94, 95, 96, 97, 98].

The simulations are performed in Cartesian geometry. Figure 1 (aFigure 2 and (b2) depict the simulation domain. The rectangular chamber dimensions are Dx×Dy=80 mm×80 mmD_x \times D_y = 80\ \mathrm{mm} \times80\ \mathrm{mm}, with a dielectric slab located at 70 mm<y<80 mm70\ \mathrm{mm} < y < 80\ \mathrm{mm} and antennas positioned within it (black rectangles showing cross-section). The plasma occupies the rest of the simulation domain. Because the code implements a direct-implicit time advance [23, 88, 93, 99], we set the cell size to be Δx=0.33 mm\Delta x = 0.33\ \mathrm{mm}, resulting in overall grid dimensions of Nx×Ny=240×240N_x \times N_y = 240 \times240. The antenna currents are 180∘180^\circ out of phase, oscillating with the frequency f=ω/2πf = \omega/2\pi in the range of 1 to 10 MHz and the amplitude Icoil=60I_{\mathrm{coil}} = 60 to 280 A. The coils create an RF electromagnetic field that penetrates into the background plasma. The amplitude of the coil current is relatively high because it is necessary to sustain the plasma. Similar values have been used in experiments [3, 46, 100]. The domain is surrounded by conductive boundaries with secondary emission neglected. Initially, the system contains Argon gas at the pressure p=5 mTorrp = 5\ \mathrm{mTorr} and a plasma with a uniform density Ne=Ni=1017 m−3N_e = N_i = 10^{17}\ \mathrm{m}^{-3}, electron temperature Te=2.0 eVT_e = 2.0\ \mathrm{eV}, and ion temperature Ti=0.03 eVT_i = 0.03\ \mathrm{eV}. 1000 macro-particles per cell are specified initially for all species. Each simulation is analyzed after reaching a steady state.

Time evolution of physical quantities and steady-state profiles

Figure 1. Time evolution of physical quantities and steady-state profiles. Panel (a) is for the low frequency case f=1MHzf = 1\mathrm{MHz} and panel (b) is for f=10MHzf = 10\mathrm{MHz} case. Subplots (a1) and (b1) show time evolution of the ion density and electron temperature. The probes are placed at the center, shown by black stars in (a3) and (b3). Subplots (a2) and (b2) show 2D maps of the magnetic field strength with superimposed magenta curves tracing the field lines. Black rectangles are the cross-sections of the coil wires, and the black horizontal line is the plasma facing boundary of the dielectric. Subplots (a2) and (b2) are plotted at the phase when the RF magnetic field is at the maximum. Subplots (a3)-(b4) are the time-averaged (denoted by ⟨...⟩\langle...\rangle) ion density and energy deposition.

Color maps of current density, particle trajectories, and particle velocity and drift velocity components for 1 MHz and 10 MHz cases

Figure 2. Subplots (a1) and (b1) show the color maps of the current density vs yy and tt; the domain cross sections are taken along the black dashed lines in Figure 1 (a4) and (b4), respectively. Subplots (a2) and (b2) show representative particle trajectories, respectively, for the 1 MHz and 10 MHz cases over the time intervals marked in subplots (a3) and (b3). The red arrows in (a2) designate the counteracting electrostatic and Lorentz forces acting upon the particle in the 1 MHz case. Subplots (a3) and (b3) show the time evolution of the particle velocity and the drift velocity components, with v∇Bv_{\nabla B} being the drift due to the magnetic field gradient, vcurvBv_{\mathrm{curv}B} being curvature drift, and vEy×Bv_{E_y \times B} being Ey×BE_y \times B drift. The vector potential, expressed as qA/meqA/m_e, and the canonical momentum, vz+qA/mev_z + qA/m_e for 1 MHz case are traced by the green and light blue lines.

**Results–**We first focus on two cases, with IcoilI_{\mathrm{coil}} at 130 A and the frequencies of f=1 MHzf = 1\ \mathrm{MHz} and f=10 MHzf = 10\ \mathrm{MHz}. Figure 1 shows the time histories and spatial profiles of the plasma for these cases. The time evolution of the ion density and electron temperature shown in Figure 1 (aFigure 1 and (b1) indicates that the simulations have reached a steady state. Subplots (a2) and (b2) show the magnetic field lines with superimposed color-maps of the field strength. The magnetic field amplitude is almost the same between the low- and the high-frequency cases because it depends mainly on the coil current (B∝IcoilB \propto I_{\mathrm{coil}}). Comparing the ion density profiles between Figure 1 (aFigure 3 and (b3), we see a significant difference. This is due to a comparable difference in energy deposition between (a4) and (b4), because Ez⋅Jze∝ω2IcoilE_z \cdot J_{ze} \propto\omega^2 I_{\mathrm{coil}}.

Figure 3. Comparing PIC simulations with theory for (a): 1 MHz case and (b): 10 MHz case. Subplots (a1) and (b1) show the electron current JzeJ_{ze} and subplots (a2) and (b2) show the energy deposition rate. The comparison is made at the locations denoted by black stars in Figure 1 (a4) and (b4). For our kinetic theory, we use Eq. (2) with ky=2πny/Lyk_y = 2\pi n_y/L_y with nyn_y from 1 to 210 (number of cells occupied by plasma) to compute Jze,1J_{ze,1}. The fluid approach is from Ref. [31] and the previous non-local kinetic calculation is from Ref. [82].

A close examination of the two cases shown in Figure 2 reveals two notable distinctions. First, a pronounced jet-like structure of the electron current is observed in the time history of the low-frequency case (see Figure 2 (aFigure 1). The jet-like structure propagates into the plasma at around t=29.5 μst = 29.5\ \mathrm{\mu s}, with a speed close to the electron thermal velocity vthv_{\mathrm{th}}. The penetration of electron current into plasma in Figure 2 (bFigure 1, however, is due to anomalous skin effect and shows a clear qualitative difference from Figure 2 (aFigure 1. The second distinction is in the representative particle trajectories shown in Figure 2 (aFigure 2 and (b2) for low- and high-frequency cases. For low-frequency case, electrons are trapped in effective potential (see End Matter)

with a bounce-radius (∼2\sim2 mm) smaller than the skin depth (∼10\sim10 mm). The electrons are, however, not trapped for the high-frequency case. To track the electrons, we randomly sampled 400 particle trajectories originating from x=65.0x = 65.0 mm and y=65.7y = 65.7 mm (the location marked by the black stars in Figure 1 (aFigure 4 and (b4)) at t=29.2 μst = 29.2\ \mu\mathrm{s}. We choose this location and time because in the 1 MHz case, most new electrons are produced by ionization near the coil at this phase of the RF period (see Figure 3 in our accompanied paper [101]). A movie for this particle is available in the supplemental material [102]. This particle behavior is in significant contrast to the findings of Ref. [103], where the Larmor radius was assumed to be much larger than the skin depth. It is evident from Figure 2 (aFigure 2 that electrons are trapped near the coil region. The oscillations of velocities shown in (a3) indicate that the electron undergoes a cyclotron motion in y−zy-z plane and bounces parallel to the magnetic field in xx direction. It is important to note that the electrons’ yy locations remain almost unchanged. This is due to the plasma-generated ambipolar electric field [104], EyE_y, counterbalancing the Lorentz force v⃗z×B⃗\vec{v}_z \times\vec{B} and making the time-averaged net force on the electrons in yy direction vanish. Therefore, the bounce invariant I=∫vydyI = \int v_y\mathrm{d}y instead of μ=mev⊥2/2B\mu= m_e v_\perp^2/2B is an adiabatic invariant for the trapped electrons (see End Matter). Figure 2 (aFigure 3 further shows that the particle motion in zz direction is due to particle drifts, with Ey×BE_y \times B drift being dominant [101]. As the RF magnetic field is π/2\pi/2 out of phase with the inductive electric field and it decreases as the electric field increases, the electrons eventually detrap from effective potential near the phase when the magnetic field becomes small: the electrons move into the plasma interior at around t=29.5 μst = 29.5\ \mu\mathrm{s}, forming a jet-like structure shown in Figure 2 (aFigure 1. During this process, the electron’s canonical momentum mevz+qAm_e v_z + qA is conserved (denoted by the light blue line in Figure 2 (aFigure 3), where AA is the vector potential. A sample test particle in the f=10f = 10 MHz case, however, travels over the entire simulation domain, interacting with the coil field in a transient way, as described in multiple previous works [86, 87].

Figure 4. The solid lines show, in a steady state, (a) energy deposition rates, averaged over the entire simulation domain and RF period, (b) time-averaged ion density at the probe location denoted by the black star in Figure 1 (aFigure 4, and (c) time-averaged electron temperature at the probe location denoted by the black star in Figure 1 (aFigure 3, for different values of the coil current and driving frequency. The dashed lines denote the onset threshold for the periodic burst regime, which follows the scaling law of Ne∼Icoil3/ωTeN_e \sim I_{\mathrm{coil}}^3/\omega T_e according to Eq. (5). Each dashed line denotes the boundary for the solid line of the same color in (b). The relevant magnetic field and temperature data for calculating the dashed lines is evaluated at the black stars in Figure 1 (aFigure 4 and (b4). The red and blue stars in (b) denote the transition points between the two regimes.

Theory–As shown in Figure 2 (aFigure 3 and justified in the End Matter, electron motion can be treated as a cyclotron rotation in y−zy-z plane around a guiding center drifting in zz direction. To obtain an analytical expression for electron current, the electron distribution function ff is split into a background Maxwellian f0f_0 as function of local electron density and temperature; and f1f_1, a perturbation to it. We use the assumption of: 1) The maximum electron bounce frequency (in effective potential) Ωmax⁡≫ω\Omega_{\max} \gg\omega. Indeed, for 1 MHz case Ωmax⁡∼108 s−1\Omega_{\max} \sim10^8\ \mathrm{s}^{-1} and ω=6.28×106 s−1\omega= 6.28 \times10^6\ \mathrm{s}^{-1}. 2) We only consider a 1D spatial variation, specifically along the black

dashed line in Figure 1 (a4). Similar assumptions have been made in the pioneering paper of Tuszewski [31]. The theory considers RF magnetic field and electrostatic potential, and is the first to predict current that can be compared with PIC simulations when nonlinear skin effect dominates.

The kinetic equation for the perturbed electron distribution function f1f_1 is

∂f1∂t+v⃗⋅∂f1∂r⃗+qm(v⃗×B⃗+E⃗sc)⋅∂f1∂v⃗=qm∂A⃗∂t⋅∂f0∂v⃗,(1)\frac{\partial f_1}{\partial t}+\vec{v}\cdot\frac{\partial f_1}{\partial\vec{r}}+\frac{q}{m}\left(\vec{v}\times\vec{B}+\vec{E}_{sc}\right)\cdot\frac{\partial f_1}{\partial\vec{v}}=\frac{q}{m}\frac{\partial\vec{A}}{\partial t}\cdot\frac{\partial f_0}{\partial\vec{v}}, \tag*{(1)}

where A⃗\vec{A} is the vector potential of the EM field and EscE_{sc} is the longitudinal electric field EyE_y resulting from charge separation. Note that we have E⃗sc≫∂A⃗/∂t\vec{E}_{sc}\gg\partial\vec{A}/\partial t based on simulations. Essentially, we consider the case where the RF magnetic field strongly alters the particle trajectory. This is the opposite case to kinetic theories describing the high frequency case [82]. Electron-neutral collision is neglected because the collision is weak νen=3.2×106 s−1<ω≪Ωmax⁡\nu_{en}=3.2\times10^{6}\,\mathrm{s}^{-1}<\omega\ll\Omega_{\max} and does not affect our results significantly. The full derivation is presented in our accompanied paper [101]. The total current in zz direction is [105, 106]

J⃗ze=∫dv⃗(v⃗z,d+v⃗⊥)f0+∫dv⃗(v⃗z,d+v⃗⊥)f1=J⃗ze,0+J⃗ze,1,(2)\vec{J}_{ze}=\int d\vec{v}(\vec{v}_{z,d}+\vec{v}_{\perp})f_0+\int d\vec{v}(\vec{v}_{z,d}+\vec{v}_{\perp})f_1=\vec{J}_{ze,0}+\vec{J}_{ze,1}, \tag*{(2)}

where v⃗z,d\vec{v}_{z,d} is the drift velocity in zz direction and v⃗⊥\vec{v}_{\perp} is the perpendicular gyro-velocity. Upon integrating Eq. (1) along the unperturbed electron trajectory, we obtain an analytical expression for f^1\hat{f}_1 (the Fourier transform of f1f_1). The approximation to the electron current Jze,1J_{ze,1} obtained by integrating f^1\hat{f}_1 is

Jze,1(t)=Re{∑kyJ^ze,1(t)eikyy}=Re{∑ky[ωq2NeA^(t)Tevd′vFωe−ξI0(ξ)−iωq2NeA^(t)Te(vd′+vF)kyvth2Ωωe−ξ(I0(ξ)−I0′(ξ))−ωq2NeA^(t)Tevth2ωe−ξ(2ξI0(ξ)−2ξI0′(ξ))]eikyy},(3)\begin{aligned} J_{ze,1}(t)&=Re\left\{\sum_{k_y}\hat{J}_{ze,1}(t)e^{ik_y y}\right\}\\ &=Re\left\{\sum_{k_y}\left[\frac{\omega q^2N_e\hat{A}(t)}{T_e}v'_d\frac{v_F}{\omega}e^{-\xi}I_0(\xi)-i\frac{\omega q^2N_e\hat{A}(t)}{T_e}\frac{(v'_d+v_F)k_yv_{\mathrm{th}}^2}{\Omega\omega}e^{-\xi}\left(I_0(\xi)-I'_0(\xi)\right)\right.\right.\\ &\qquad\left.\left.-\frac{\omega q^2N_e\hat{A}(t)}{T_e}\frac{v_{\mathrm{th}}^2}{\omega}e^{-\xi}\left(2\xi I_0(\xi)-2\xi I'_0(\xi)\right)\right]e^{ik_y y}\right\}, \tag*{(3)} \end{aligned}

where Re{…}Re\{\ldots\} denotes the real part, J^ze,1\hat{J}_{ze,1} and A^\hat{A} denote the Fourier components, NeN_e and TeT_e are the time-averaged local electron density and temperature, both from definition of f0f_0. Also, Ω(t)≈qB(t)/me\Omega(t)\approx qB(t)/m_e, vF≈−⟨F⟩/qB(t)+Ey/B(t)v_F\approx-\langle F\rangle/qB(t)+E_y/B(t) is the total electron drift velocity (where E×BE\times B drift is dominant), ⟨F⟩=−⟨me(v⊥2/2+v∥2)⟩∇⊥ln⁡B\langle F\rangle=-\langle m_e(v_\perp^2/2+v_\parallel^2)\rangle\nabla_\perp\ln B is the average force (over particles) due to the magnetic field gradient and curvature, v⊥v_\perp and v∥v_\parallel are the gyro-velocity components in cylindrical coordinates, I0I_0 is the 00’s order modified Bessel function of the first kind (the prime on it denotes derivative), ξ=(kyvth/Ω(t))2\xi= (k_yv_{\mathrm{th}}/\Omega(t))^2, where kyk_y is the argument of the Fourier component of AA, and the diamagnetic drift velocity is vd′≈Te/qB(t)dln⁡Ne/dyv'_d \approx T_e/qB(t)d\ln N_e/dy. The first term in Eq. (3) is due to the drift v⃗z,d\vec{v}_{z,d}, whereas the second and third terms represent contributions from v⃗⊥\vec{v}_{\perp}. The guiding center current Jze,0J_{ze,0} related to f0f_0 comes mainly from Ey×BE_y \times B drift

Jze,0(t)=qNevF′(1−exp⁡(−12v⊥,tr2vth2)),(4)J_{ze,0}(t) = qN_ev'_F\left(1-\exp\left(-\frac{1}{2}\frac{v_{\perp,\mathrm{tr}}^2}{v_{\mathrm{th}}^2}\right)\right), \tag*{(4)}

where v⊥,tr=∣qB(t)∣δ/2mev_{\perp,\mathrm{tr}} = |qB(t)|\delta/2m_e is the maximum perpendicular velocity for the electrons to be trapped by the magnetic field, since only the trapped electrons in f0f_0 contribute to the current. And vF′=Ey/B(t)v'_F = E_y/B(t). In Eq. (4) we have neglected the diamagnetic current since it is much smaller than the E×BE \times B current.

Our case is more complicated than the standard gyrokinetic treatment [106, 107, 108] in two aspects. First, only the electrons are magnetized, so the current from electron Ey×BE_y \times B drift dominates and cannot be canceled with that of the ions. Second, the guiding magnetic field is time-varying and the drift velocity vFv_F is comparable to or larger than the thermal velocity vthv_{\mathrm{th}}, making the terms associated with vFv_F dominant. Because of this, the contribution from f1f_1 to the current is not negligible. Equation (3) for Jze,1J_{ze,1} contains Bessel functions. Those represent the Finite Larmor Radius (FLR) effect. Physically, the FLR effect occurs because when a particle undergoes gyro-motion and a field gradient is present, the particle tends to spend more time on the low-field side than the high-field side [107, 109, 110, 111]. Effectively, the field acting on the electrons will be reduced, hence reducing the current. Therefore, Jze,1J_{ze,1} takes the sign opposite to that of Jze,0J_{ze,0} and acts to reduce the current. The dominance of the RF magnetic field on particle motion significantly changes the electron current (see simplified formula Eq. (12) in Ref. [101]), indicating the presence of a new physical regime. With the gradual increase of coil current, the plasma density increases and the electron trajectory changes to the classical type as seen in Figure 2 (bFigure 3, with all drifts becoming negligible and the electron current taking the form in existing kinetic theory

J^ze,old=ωq2me∫vzω−kyvy∂f0∂v⃗⋅A^ dv⃗\hat{J}_{ze,\mathrm{old}} = \frac{\omega q^2}{m_e}\int\frac{v_z}{\omega-k_yv_y}\frac{\partial f_0}{\partial\vec{v}}\cdot\hat{A}\,d\vec{v}

[30, 82].

Figure 3 compares the time evolution of JzeJ_{ze} and energy deposition between the theoretical predictions and the PIC simulations. For our theory (black) and the Tuszewski theory (magenta dashed, fluid model) [31], we evaluate the electric field and other relevant quantities. at the location marked by the black stars in Figure 1 (a4) and (b4). We then calculate the electron current from Eq. (2) and the conductivity given in Ref. [31], respectively. To carry out a calculation based on the previous non-local kinetic theory developed for high frequency ICP (blue), we employ the code provided in Ref. [82] to perform a 2D simulation of the entire system, and then extract the electric field and electron current at the same location marked in Figure 1 (a4), (b4). This existing kinetic theory assumes uniform plasma density. We see that the fluid theory does not predict the current and the energy deposition rate, whereas

the previous kinetic theory performs well only for f=10f = 10 MHz case. At 1 MHz, our theory displays a much better agreement for both the current and energy deposition, while the other theories predict energy deposition that deviates from PIC simulations by at least a factor of 5. The electron current for f=1f = 1 MHz shows a higher third harmonic component (Fig. 15 in [101]). Physically, this is due to the coupling between second harmonic electron motion in yy direction (due to potential, see Fig. 37 in [101]) and RF magnetic field at fundamental frequency. We identify the new regime dominated by periodic bursty energy deposition as the “periodic burst regime”. When averaging over the simulation domain, the jet-like current contributes around 70% to the total energy deposition.

Parameter Scan—We perform a parameter scan over IcoilI_{\mathrm{coil}} and ω\omega to identify the onset boundary for periodic burst regime. Figure 4 shows the average energy deposition, ion density, and electron temperature for different values of the driving waveform parameters. We see that at lower frequencies there is a distinct jump in energy deposition. A similar jump is observed for the ion density. This transition occurs because in the periodic burst regime, the electric field and current are nearly out of phase during most of the RF period, since the strong RF magnetic field makes plasma response nearly local. Only in the phase of electron jet will the energy deposition become significant. Therefore, the energy deposition becomes lower than in the anomalous skin effect regime, making the electron density also much lower. Based on this understanding, the criterion for transition between the bursty regime and the anomalous non-local regime is estimated by (see also End Matter)

2mevth∣qB∣=(vthc2ωωpe2)1/3.(5)2\frac{m_{\mathrm{e}}v_{\mathrm{th}}}{\lvert qB\rvert} = \left(\frac{v_{\mathrm{th}}c^2}{\omega\omega_{\mathrm{pe}}^2}\right)^{1/3}. \tag*{(5)}

When the diameter of the electron gyro-circle is larger than the skin depth, the electrons are unlikely to undergo a full gyro-cycle, and are therefore no longer being confined by the effective potential. Based on Eq. (5), an estimate of the plasma density at the threshold of the periodic burst regime is obtained. The cases with different driving frequencies will have different transition values for the plasma density, as indicated in Figure 4 (b) by the dashed lines. One expects to obtain the periodic burst regime in the region located below the dashed lines. A reasonable agreement is found, indicating the validity of Eq. (5) for finding the boundary between the two regimes.

Conclusions—We identify a new nonlinear skin effect regime when an RF electromagnetic field penetrates into a background plasma, where the energy deposition property manifests a periodic burst nature. In this regime, the electrons are trapped in effective potential, with μ\mu no longer being the adiabatic invariant. When the RF magnetic field becomes weak, the electrons detrap, forming a jet-like current propagating quickly into the plasma and accounting for a sizable fraction of the energy deposition. As a result, the electron current in this regime shows higher harmonics instead of just a sinusoidal shape with basic frequency. A kinetic theory is proposed to estimate the electron current, showing good agreement with PIC simulations. The condition for the transition from anomalous skin effect regime (un-trapped electrons) to the periodic burst regime is identified. A significant jump in the plasma density is observed during such transition, similar to what was observed in the E-H mode transition [34, 100, 112, 113, 114]. We propose an estimate for current density (Eq. (2)) and a criterion for transition to the periodic burst regime (Eq. (5)) that can be verified in future experiments.

Acknowledgement– The authors thank Dr. Edward Startsev for checking the derivations in this paper. The authors thank Dr. Sarvesh Sharma for the discussions of high frequency harmonics. The work was partially supported by the National Natural Science Foundation of China (Grant No. 12305223) and the National Natural Science Foundation of Guangdong Province (Grant No. 2023A1515010762). This research was also supported in part by the U.S. Department of Energy, Office of Fusion Energy Science under Contract No. DE-AC02-09CH11466.

End Matter: Electron motion in a 1D model of low-frequency ICP skin layer with account for strong RF magnetic field and electrostatic potential

In this End Matter, we show that the electron trajectory in the low frequency 1MHz case is not simply affected by the RF magnetic field, but by the effective potential formed by the combination of RF magnetic field and electrostatic potential. To better understand the representative electron trajectory shown in Figure 2 (a), we turn to a simplified 1D model of electron motion along the vertical dashed line in Figure 1 (aFigure 4. Note that the electron motion parallel to magnetic field is not considered because the bounce parallel to the magnetic field occurs with lower frequency, and is not expected to significantly affect the electron current. Here, the simulation data shows that the vector potential can be approximated as Az(y,t)=Amax⁡exp⁡(−y/δ)cos⁡(ωt)A_z(y,t)=A_{\max}\exp(-y/\delta)\cos(\omega t) (where δ\delta is the skin depth). The RF magnetic field in the xx direction is Bx(y,t)=∂Az/∂yB_x(y,t)=\partial A_z/\partial y and the RF inductive electric field in the zz direction is Ez(y,t)=−∂Az/∂tE_z(y,t)=-\partial A_z/\partial t. The electron equation of motion reads

medvydt=qvzBx−q∂Φ∂y,medvzdt=qEz−qvyBx,(6)m_e\frac{d v_y}{d t}=qv_zB_x-q\frac{\partial\Phi}{\partial y},\qquad m_e\frac{d v_z}{d t}=qE_z-qv_yB_x, \tag*{(6)}

where Φ\Phi is the in-plane electrostatic potential of the ambipolar electric field Ey(y,t)E_y(y,t) generated by plasma. In this model, the electrons move in the y−zy-z plane under the influence of both the electrostatic and inductive fields. They do not undergo pure cyclotron motion with only the RF magnetic field Bx(y,t)B_x(y,t). Since the canonical momentum mevz+qAz=constm_ev_z+qA_z=const is seen to be conserved, the above equations reduce to

d2ydt2=qme∂Az∂y(vz0−qme(Az−Az0))−qme∂Φ∂y≡−qme∂U∂y,(7)\frac{d^2y}{dt^2}=\frac{q}{m_e}\frac{\partial A_z}{\partial y}\left(v_{z0}-\frac{q}{m_e}(A_z-A_{z0})\right)-\frac{q}{m_e}\frac{\partial\Phi}{\partial y}\equiv-\frac{q}{m_e}\frac{\partial U}{\partial y}, \tag*{(7)}

where Az0A_{z0} is the vector potential at the initial time and position of the electron being considered and the effective potential, U(y,t)U(y,t), has been introduced as

U=qAz22−(qAz0+mevz0)Az+Φ.(8)U=q\frac{A_z^2}{2}-(qA_{z0}+m_ev_{z0})A_z+\Phi. \tag*{(8)}

Figure 5 shows the time evolution of the effective potential and two representations of a trajectory obtained by numerically solving Eq. (7). The electrostatic potential and the vector potential waveforms entering Eq. (7) are taken from the actual PIC simulation results. It is evident that the basic properties of the trajectory shown in Figure 2 (a) are reproduced by the 1D trajectory shown in Figure 5 (b): in both cases the electrons originating near the coil are trapped in the effective potential well, undergoing a cyclotron motion in the y−zy-z plane, while drifting in zz direction. The maximum drift velocity calculated in the 1D model is around 4.8×1064.8 \times10^{6} m/s (see Figure 5 (b)), nearly the same as the maximum drift velocity in Figure 2 (aFigure 3. The bounce frequency Ω=qU′′/me\Omega= \sqrt{qU^{\prime\prime}/m_{\mathrm{e}}} in the effective potential well is nearly the same as the local electron cyclotron frequency ΩB=qB/me\Omega_{\mathrm{B}} = qB/m_{\mathrm{e}} (see discussion in Ref. [101]). As we can see in Figure 5 (a) and (b), the turning points of the bounce motion stay almost unchanged over time, so the average electron flux in yy direction vanishes in the lowest order. As a result, because the electron essentially take a bounce motion in the effective potential, the real adiabatic invariant of the trapped electron is the adiabatic invariant associated with the bounce motion I=∫vy dyI = \int v_y \, \mathrm{d}y instead of the magnetic moment μ=mev⊥2/2B\mu= m_{\mathrm{e}}v_{\perp}^{2}/2B. Therefore, because the turning points of the bounce motion stay almost unchanged (see Figure 5 (a)), the amplitude of vyv_y oscillations stays almost unchanged as well. Remarkably, the phase portrait is not that different from a simple cyclotron motion in vy−vzv_y-v_z phase space (see Figure 5 (c)). This justifies the simplification of using the cyclotron motion approximation applied in deriving the analytical expression for the electron current.

Three subplots showing the normalized effective potential and electron trajectory

Figure 5. Subplot (a) shows the time evolution of the normalized effective potential UU, with the horizontal dashed lines denoting the electron energy level and the vertical dashed lines locating the turning points of an orbit trapped in the effective potential well. Note at t=29.5μst = 29.5\mu\mathrm{s} the electron becomes un-trapped and escapes the skin layer towards the bulk plasma. Subplots (b) and (c) represent the trajectory of an electron with initial velocity vy0=−0.1qAmax⁡/mev_{y0} = -0.1qA_{\max}/m_e, vz0=0v_{z0} = 0, and initial position y0=66mmy_0 = 66\mathrm{mm}. The velocity is normalized by eAmax⁡/me=7×106 m/seA_{\max}/m_e = 7 \times10^{6}\ \mathrm{m/s}.

Now we move to consider the trapping condition for the electrons in the effective potential well. The trapping condition is

E=12mevy2+qU≤min⁡[qU(y=70mm),0],(9)E = \frac{1}{2}m_{\mathrm{e}}v_y^2 + qU \le\min\left[qU(y = 70\mathrm{mm}), 0\right], \tag*{(9)}

which means that electrons neither become lost to the wall nor escape to the bulk plasma. Since the thermal velocity is only about 0.15qAmax⁡/me0.15qA_{\max}/m_{\mathrm{e}}, most electrons are deeply trapped during most of the RF period. However, starting from t=29.5μst = 29.5\mu\mathrm{s} (see Figure 5 (a)), the potential well no longer forms and the electrons start being lost to the bulk plasma. It is these electrons which eventually form the jet-like current identified in the simulations. To obtain a simpler analytical expression for the de-trapping condition, we simplify Eq. (9) by requiring only the left turning point to exist (meaning that electrons do not go to the bulk plasma),

12mevy02+12q2Az2−(qAz0+mevz0)qAz+qΦ<0.(10)\frac{1}{2}m_e v_{y0}^{2}+\frac{1}{2}q^{2}A_z^{2}-(qA_{z0}+m_e v_{z0})qA_z+q\Phi<0. \tag*{(10)}

Then, approximating the electrostatic potential as qΦ=q2Az2/4meq\Phi= q^{2}A_z^{2}/4m_e (justified in Fig. 35 of our accompanied paper [101]), we have approximately

(δρe)2+32δρe−1>0,(11)\left(\frac{\delta}{\rho_e}\right)^2+\frac{3}{2}\frac{\delta}{\rho_e}-1>0, \tag*{(11)}

where ρe=mevth/∣qBx∣\rho_e = m_e v_{th}/\lvert qB_x\rvert and vy0=vz0=vthv_{y0}=v_{z0}=v_{th} is assumed, which becomes

vth≈vz0<∣qAz∣me12≈∣qBx∣δ2me,(12)v_{th}\approx v_{z0}<\frac{\lvert qA_z\rvert}{m_e}\frac{1}{2}\approx\frac{\lvert qB_x\rvert\delta}{2m_e}, \tag*{(12)}

which is exactly Eq. (5) in the main text.

In summary, the electrons are not only affected by the RF magnetic field, but also by the electrostatic potential in x−yx-y plane. The combination of them forms an effective potential that can trap the electrons. The adiabatic invariant is no longer magnetic momentum μ\mu, indicating that it is an essentially new physical process. This is why we always say “trap” and “detrap” in the main text.

References

  1. [1]M. A. Lieberman and A. J. Lichtenberg, MRS Bulletin 30, 899 (1994).
  2. [2]V. Kolobov and V. Godyak, IEEE Transactions on Plasma Science 23, 503 (1995).
  3. [3]V. A. Godyak, R. B. Piejak, B. M. Alexandrovich, and V. I. Kolobov, Physics of Plasmas 6, 1804 (1999).
  4. [4]V. A. Godyak, R. B. Piejak, and B. M. Alexandrovich, Phys. Rev. Lett. 83, 1610 (1999).
  5. [5]A. I. Smolyakov, V. Godyak, and Y. Tyshetskiy, Physics of Plasmas 8, 3857 (2001).
  6. [6]I. D. Kaganovich, Phys. Rev. Lett. 89, 2650006 (2002).
  7. [7]V. I. Kolobov and V. A. Godyak, Plasma Sources Science and Technology 26, 075013 (2017).
  8. [8]V. Kolobov and V. Godyak, Physics of Plasmas 26, 060601 (2019).
  9. [9]P. H. Yoon, Geophysical Research Letters 38 (2011), https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2011GL047893.
  10. [10]X. Tao, J. Bortnik, J. Albert, R. Thorne, and W. Li, Journal of Atmospheric and Solar-Terrestrial Physics 99, 67 (2013), dynamics of the Complex Geospace System.
  11. [11]R. L. Stenzel, J. M. Urrutia, and K. D. Strohmaier, Phys. Rev. Lett. 96, 095004 (2006).
  12. [12]A. Hasegawa and L. Chen, Phys. Rev. Lett. 36, 1362 (1976).
  13. [13]Y. Lin, J. R. Johnson, and X. Wang, Phys. Rev. Lett. 109, 125003 (2012).
  14. [14]M. Hoshino, Phys. Rev. Lett. 108, 135003 (2012).
  15. [15]M. Hoshino, Phys. Rev. Lett. 114, 061101 (2015).
  16. [16]P. Chabert and N. Braithwaite, Physics of Radio-Frequency Plasmas (Cambridge University Press, 2011).DOI
  17. [17]M. M. Turner, Phys. Rev. Lett. 75, 1312 (1995).
  18. [18]M. M. Turner, Journal of Physics D: Applied Physics 42, 194008 (2009).
  19. [19]E. Kawamura, M. Lieberman, and A. Lichtenberg, Physics of plasmas 13 (2006).
  20. [20]T. Mussenbrock, R. P. Brinkmann, M. A. Lieberman, A. J. Lichtenberg, and E. Kawamura, Phys. Rev. Lett. 101, 085004 (2008).
  21. [21]Y. Liu, Q. Zhang, W. Jiang, L. Hou, X. Jiang, W. Lu, and Y. Wang, Phys. Rev. Lett. 107, 055002 (2011).
  22. [22]K. Zhao, D.-Q. Wen, Y.-X. Liu, M. A. Lieberman, D. J. Economou, and Y.-N. Wang, Phys. Rev. Lett. 122, 185002 (2019).
  23. [23]H. Sun, S. Banerjee, S. Sharma, A. T. Powis, A. V. Khrabrov, D. Sydorenko, J. Chen, and I. D. Kaganovich, Physics of Plasmas 30, 103509 (2023).
  24. [24]J.-Y. Sun, F. Gao, F.-J. Zhou, J. Schulze, and Y.-N. Wang, Phys. Rev. Lett. 133, 235302 (2024).
  25. [25]D. Yang, J. P. Verboncoeur, and Y. Fu, Phys. Rev. Lett. 134, 045301 (2025).
  26. [26]M. M. Turner, Phys. Rev. Lett. 71, 1844 (1993).
  27. [27]V. I. Kolobov and W. N. G. Hitchon, Phys. Rev. E 52, 972 (1995).
  28. [28]V. I. Kolobov, G. J. Parker, and W. N. G. Hitchon, Phys. Rev. E 53, 1110 (1996).
  29. [29]U. Kortshagen, I. Pukropski, and L. D. Tsendin, Phys. Rev. E 51, 6063 (1995).
  30. [30]I. D. Kaganovich and O. Polomarov, Phys. Rev. E 68, 026411 (2003).
  31. [31]M. Tuszewski, Phys. Rev. Lett. 77, 1286 (1996).
  32. [32]Y. O. Tyshetskiy, A. I. Smolyakov, and V. A. Godyak, Phys. Rev. Lett. 90, 255002 (2003).
  33. [33]H.-C. Lee, Applied Physics Reviews 5, 011108 (2018).
  34. [34]Y. W. Lee, H. L. Lee, and T. H. Chung, Journal of Applied Physics 109, 113302 (2011).
  35. [35]J. Han, P. Pribyl, W. Gekelman, A. Paterson, S. J. Lanham, C. Qu, and M. J. Kushner, Physics of Plasmas 26, 103503 (2019).
  36. [36]J. Han, P. Pribyl, W. Gekelman, and A. Paterson, Physics of Plasmas 27, 063509 (2020).
  37. [37]T. Piskin, Y. Qian, P. Pribyl, W. Gekelman, and M. J. Kushner, Journal of Applied Physics 133, 173302 (2023).
  38. [38]Lieberman, M. A. and Boswell, R. W., J. Phys. IV France 08, Pr7 (1998).
  39. [39]P. Subramonium and M. J. Kushner, Applied Physics Letters 85, 721 (2004).
  40. [40]C.-C. Hsu, M. A. Nierode, J. W. Coburn, and D. B. Graves, Journal of Physics D: Applied Physics 39, 3272 (2006).
  41. [41]C. Corr, E. Despiau-Pujo, P. Chabert, W. Graham, F. Marro, and D. B. Graves, Journal of Physics D: Applied Physics 41, 185202 (2008).
  42. [42]J. Bukowski, D. Graves, and P. Vitello, Journal of Applied Physics 80, 2614 (1996).
  43. [43]R. G. Chambers and O. R. Frisch, Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences 215, 481 (1952).
  44. [44]V. I. Kolobov and D. J. Economou, Plasma Sources Science and Technology 6, R1 (1997).
  45. [45]V. A. Godyak, R. B. Piejak, B. M. Alexandrovich, and V. I. Kolobov, Phys. Rev. Lett. 80, 3264 (1998).
  46. [46]V. Godyak, B. Alexandrovich, R. Piejak, and A. Smolyakov, Plasma Sources Science and Technology 9, 541 (2000).
  47. [47]V. Godyak, R. Piejak, B. Alexandrovich, and A. Smolyakov, Plasma Sources Science and Technology 10, 459 (2001).
  48. [48]S. Sharma, S. Patil, S. Sengupta, A. Sen, A. Khrabrov, and I. Kaganovich, Physics of Plasmas 29 (2022).
  49. [49]H. Sun, J. Chen, I. D. Kaganovich, A. Khrabrov, and D. Sydorenko, Phys. Rev. E 106, 035203 (2022).
  50. [50]H. Sun, J. Chen, I. D. Kaganovich, A. Khrabrov, and D. Sydorenko, Phys. Rev. Lett. 129, 125001 (2022).
  51. [51]L. Xu, H. Sun, D. Eremin, S. Ganta, I. Kaganovich, K. Bera, S. Rauf, and X. Wu, Plasma Sources Science and Technology 32, 105012 (2023).
  52. [52]O. Polomarov, C. Theodosiou, I. Kaganovich, D. Economou, and B. Ramamurthi, IEEE Transactions on Plasma Science 34, 767 (2006).
  53. [53]H.-C. Lee, Applied Physics Reviews 5, 011108 (2018).
  54. [54]V. A. Godyak, B. M. Alexandrovich, and V. I. Kolobov, Phys. Rev. E 64, 026406 (2001).
  55. [55]F. F. Chen, Physics of Plasmas 8, 3008 (2001).
  56. [56]J. D. Evans and F. F. Chen, Phys. Rev. Lett. 86, 5502 (2001).
  57. [57]A. Smolyakov, V. Godyak, and A. Duffy, Physics of Plasmas 7, 4755 (2000).
  58. [58]Y. O. Tyshetskiy, A. I. Smolyakov, and V. A. Godyak, Plasma Sources Science and Technology 11, 203 (2002).
  59. [59]W. Jiang, H. yu Wang, S. xia Zhao, and Y. nian Wang, Journal of Physics D: Applied Physics 42, 102005 (2009).
  60. [60]H.-C. Lee, D.-H. Kim, and C.-W. Chung, Applied Physics Letters 102, 234104 (2013).
  61. [61]I. M. El-Fayoumi and I. R. Jones, Plasma Sources Science and Technology 7, 162 (1998).
  62. [62]S. Xu, K. N. Ostrikov, Y. Li, E. L. Tsakadze, and I. R. Jones, Physics of Plasmas 8, 2549 (2001).
  63. [63]F. Gao, H. Li, W. Yang, J. Liu, Y.-R. Zhang, and Y.-N. Wang, Physics of Plasmas 25, 013515 (2018).
  64. [64]M. V. Isupov and V. A. Pinaev, Journal of Applied Mechanics and Technical Physics 64, 757 (2023).
  65. [65]S. Ashida, M. R. Shim, and M. A. Lieberman, Journal of Vacuum Science and Technology A 14, 391 (1996).
  66. [66]G. Cunge, B. Crowley, D. Vender, and M. M. Turner, Plasma Sources Science and Technology 8, 576 (1999).
  67. [67]B. Ramamurthi and D. J. Economou, Plasma Sources Science and Technology 11, 324 (2002).
  68. [68]P. Subramonium and M. J. Kushner, Journal of Vacuum Science and Technology A: Vacuum, Surfaces, and Films 20, 325 (2002).
  69. [69]P. Subramonium and M. J. Kushner, Journal of Vacuum Science and Technology A: Vacuum, Surfaces, and Films 20, 313 (2002).
  70. [70]S. Banna, A. Agarwal, G. Cunge, M. Darnon, E. Pargon, and O. Joubert, Journal of Vacuum Science and Technology A 30 (2012).
  71. [71]Y. Qian, W. Gekelman, P. Pribyl, T. Piskin, and A. Paterson, Physics of Plasmas 31, 063507 (2024).
  72. [72]U. Fantz, H. Falter, P. Franzen, D. Wünderlich, M. Berger, A. Lorenz, W. Kraus, P. McNeely, R. Riedl, and E. Speth, Nuclear fusion 46, S297 (2006).
  73. [73]E. Speth, H. Falter, P. Franzen, U. Fantz, M. Bandyopadhyay, S. Christ, A. Encheva, M. Fröschle, D. Holtum, B. Heinemann, et al., Nuclear Fusion 46, S220 (2006).
  74. [74]H. Li, F. Gao, D.-Q. Wen, W. Yang, P.-C. Du, and Y.-N. Wang, Journal of Applied Physics 125, 173303 (2019).
  75. [75]D. Zielke, S. Briefi, S. Lishev, and U. Fantz, Plasma Sources Science and Technology 31, 035019 (2022).
  76. [76]I. Kaganovich, E. Startsev, and G. Shvets, Physics of Plasmas 11, 3328 (2004).
  77. [77]I. D. Kaganovich, O. V. Polomarov, and C. E. Theodosiou, Physics of Plasmas 11, 2399 (2004).
  78. [78]A. I. Smolyakov, V. A. Godyak, and Y. O. Tyshetskiy, Physics of Plasmas 10, 2108 (2003).
  79. [79]H.-C. Lee, S. Oh, and C.-W. Chung, Plasma Sources Science and Technology 21, 035003 (2012).
  80. [80]B. Ramamurthi, D. J. Economou, and I. D. Kaganovich, Plasma Sources Science and Tech-nology 12, 170 (2003).
  81. [81]A. M. Froese, Particle-in-cell simulations of nonlocal and nonlinear effects in inductively coupled plasmas, Ph.D. thesis, University of Saskatchewan (2007).
  82. [82]W. Yang and Y.-N. Wang, Plasma Physics and Controlled Fusion 63, 035031 (2021).
  83. [83]W. Yang, F. Gao, and Y.-N. Wang, Physics of Plasmas 29, 063503 (2022).
  84. [84]W. YANG, F. GAO, and Y. WANG, Plasma Science and Technology 24, 055401 (2022).
  85. [85]W. Yang, F. Gao, and Y.-N. Wang, AIP Advances 12, 055222 (2022).
  86. [86]R. H. Cohen and T. D. Rognlien, Physics of Plasmas 3, 1839 (1996).
  87. [87]R. H. Cohen and T. D. Rognlien, Plasma Sources Science and Technology 5, 442 (1996).
  88. [88]M. R. Gibbons and D. W. Hewett, Journal of Computational Physics 120, 231 (1995).
  89. [89]M. M. Turner, Plasma Sources Science and Technology 5, 159 (1996).
  90. [90]D. Y. Sydorenko, A. I. Smolyakov, Y. O. Tyshetsky, and V. A. Godyak, Physics of Plasmas 12, 033503 (2005).
  91. [91]C. Fu, Y. Dong, Y. Li, W. Wang, Z. Wang, and W. Liu, Journal of Physics D: Applied Physics 57, 135201 (2024).
  92. [92]H. Wen, J. Schulze, Y. Fu, J.-Y. Sun, and Q.-Z. Zhang, Plasma Sources Science and Technology 34, 03LT01 (2025).
  93. [93]D. Sydorenko, I. D. Kaganovich, A. V. Khrabrov, S. A. Ethier, J. Chen, and S. Janhunen, Improved algorithm for a two-dimensional darwin particle-in-cell code (2024), arXiv:2409.19559 [physics.plasm-ph].arxiv.org/abs/2409.19559
  94. [94]T. Charoy, J.-P. Boeuf, A. Bourdon, J. A. Carlsson, P. Chabert, B. Cuenot, D. Eremin, L. Garrigues, K. Hara, I. D. Kaganovich, et al., Plasma Sources Science and Technology 28, 105010 (2019).
  95. [95]B. Jin, J. Chen, A. V. Khrabrov, Z. Wang, and L. Xu, Plasma Sources Science and Technology 31, 115015 (2022).
  96. [96]Q. Cao, J. Chen, H. Sun, G. Sun, S. Liu, C. Tan, and Z. Wang, Physics of Plasmas 30, 103501 (2023).
  97. [97]J. Chen, A. V. Khrabrov, I. D. Kaganovich, and H.-P. Li, Physics of Plasmas 31, 042112 (2024).
  98. [98]B. Jin, J. Chen, G. Sun, Z. Wang, and H. Sun, Plasma Sources Science and Technology 33, 06LT01 (2024).
  99. [99]B. I. Cohen, A. Langdon, and A. Friedman, Journal of Computational Physics 46, 15 (1982).
  100. [100]S. Mattei, K. Nishida, S. Mochizuki, A. Grudiev, J. Lettry, M. Q. Tran, and A. Hatayama, Plasma Sources Science and Technology 25, 065001 (2016).
  101. [101]H. Sun, J. Chen, A. Khrabrov, I. D. Kaganovich, W. Yang, D. Sydorenko, and S. Brunner, Submitted to Physical Review E as a joint submission with this letter (2025).
  102. [102]See Supplemental Material at [URL will be inserted by publisher] for movie of nonlinear particle dynamics (2025).
  103. [103]A. M. Froese, A. I. Smolyakov, and D. Sydorenko, Physics of Plasmas 16, 080704 (2009).
  104. [104]D. Y. Sydorenko, A. I. Smolyakov, Y. O. Tyshetskiy, and V. A. Godyak, Physics of Plasmas 12, 033503 (2005).
  105. [105]H. Qin, W. M. Tang, G. Rewoldt, and W. W. Lee, Physics of Plasmas 7, 991 (2000).
  106. [106]W. W. Lee and H. Qin, Physics of Plasmas 10, 3196 (2003).
  107. [107]J. R. Cary and A. J. Brizard, Rev. Mod. Phys. 81, 693 (2009).
  108. [108]S. Brunner, 3d circle course at Centre de Recherches en Physique des Plasmas, Association Euratom-Confédération Suisse, Ecole Polytechnique Fédérale de Lausanne, Switzerland, https://crppwww. epl. ch/brunner/inhomoplasma. pdf (entered 01.06. 2018) (2014).
  109. [109]R. G. Littlejohn, Journal of Plasma Physics 29, 111–125 (1983).
  110. [110]J. R. Cary and R. G. Littlejohn, Annals of Physics 151, 1 (1983).
  111. [111]T. S. Hahm, Nonlinear gyrokinetic equations for tokamak microturbulence, Tech. Rep. (Princeton Plasma Physics Lab. (PPPL), Princeton, NJ (United States), 1988).
  112. [112]U. Kortshagen, N. D. Gibson, and J. E. Lawler, Journal of Physics D: Applied Physics 29, 1224 (1996).
  113. [113]P. Chabert, A. J. Lichtenberg, M. A. Lieberman, and A. M. Marakhtanov, Journal of Applied Physics 94, 831 (2003).
  114. [114]T. Wegner, C. Küllig, and J. Meichsner, Plasma Sources Science and Technology 26, 025007 (2017).

Paper details

vlnc:W_75utkdjx
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents