betaFeedback

WorksPhysics and Astronomy

Self-organization of magnetic islands and zonal magnetic fields driven by slab microtearing modes in the absence of magnetic shear

Abstract

Self-organization in collisional microtearing turbulence is investigated using gyrokinetic simulations in a slab geometry without equilibrium magnetic shear. Unlike magnetically driven conventional tearing instabilities, microtearing instabilities are driven by the electron temperature gradient. Electron temperature fluctuations are found to contain most of the gyrokinetic free energy. Although magnetic perturbations contribute little to the free energy, the magnetic field structure plays a crucial role in the nonlinear evolution of microtearing turbulence. Zonal magnetic fields emerge spontaneously through the nonlinear interaction between magnetic perturbations and electron temperature fluctuations. These zonal fields provide an effective magnetic shear that enables magnetic islands to form despite the absence of equilibrium magnetic shear. The electron temperature profile is flattened within the islands. This flattening reduces the instability drive and is expected to promote turbulence saturation.

Introduction

Microturbulence driven by electron pressure gradients causes anomalous electron heat transport in magnetic confinement devices. Understanding the heat transport is a crucial issue in the design of future fusion reactors. In conventional low-β\beta devices (β\beta is the ratio of the plasma pressure to the magnetic pressure), electrostatic turbulence driven by the electron temperature gradient (ETG) mode and the trapped electron mode (TEM) is known to contribute to electron heat transport (see, e.g., [1, 2]). In higher-β\beta devices, electromagnetic instabilities such as microtearing and kinetic ballooning modes can also cause anomalous electron heat transport through magnetic fluctuations.

Numerous gyrokinetic simulations have shown that microtearing modes become unstable and contribute to electron heat transport in existing devices, such as spherical tokamaks (STs) [3, 4, 5, 6, 7], conventional tokamaks [8, 9, 10], reversed-field pinches (RFPs) [11, 12], and stellarators [13, 14], as well as in a conceptual ST [15, 16, 17].

The microtearing mode changes the topology of the magnetic field and generates small-scale magnetic islands. The islands may overlap to provide a stochastic magnetic field structure, thereby enhancing the electron heat transport along the field lines [18, 19, 20]. Nonlinear gyrokinetic simulations using experimental parameters of the National Spherical Torus Experiment (NSTX) showed that electron heat transport driven by microtearing turbulence increases with electron collisionality [5], consistent with the observed NSTX confinement scaling [21].

The microtearing mode is an ion-scale tearing instability and was first proposed theoretically by Hazeltine et al. [22] using a kinetic model with the rigorous Fokker–Planck collision operator. They revealed that the velocity dependence of the collision frequency is indispensable for the temperature gradient to drive the microtearing instability. Subsequently, the dependence of the microtearing mode on the electron–ion collision frequency was investigated theoretically [23] and numerically [24, 25]. These studies indicated that the mode is stable in both the collisional and collisionless limits. Based on the analysis of Drake and Lee [23], the destabilizing effect is due to pitch-angle scattering with a velocity-dependent collision frequency in the presence of electron temperature nonuniformity. In the fluid description, this destabilizing effect is represented as the time-dependent thermal force [26]. Through this force, the electron temperature nonuniformity generates a parallel inductive electric field that lags in phase behind the perturbed magnetic field [27]. This phase relation is analogous to the relation between density fluctuations and the electrostatic potential in electrostatic drift waves. The microtearing mode is fundamentally an electromagnetic drift wave and is destabilized solely by the time-dependent thermal force even in the absence of equilibrium magnetic shear [28]. Electromagnetic microinstabilities driven by the electron temperature gradient have been reported even in the collisionless regime, both in slab geometry and in toroidal geometry. These modes have been discussed as a tearing-parity electromagnetic ETG mode [29, 30] and a collisionless microtearing mode [31, 32, 33], respectively. Such collisionless modes are outside the scope of this paper.

The nonlinear saturation of microtearing turbulence has been investigated since the early drift-kinetic analysis of a slab plasma without magnetic shear by Drake et al. [34]. They showed that electron streaming along the perturbed magnetic field B⊥\mathbf{B}_{\perp} transfers energy from shorter- to longer-wavelength modes. Saturation occurs when the stabilizing effect of this nonlinear energy transfer balances the linear growth. The resulting saturation amplitude scales as ∣B⊥∣/B0∼ρe/LT0e|B_{\perp}|/B_{0}\sim\rho_{\mathrm{e}}/L_{T_{0\mathrm{e}}}, where ρe\rho_{\mathrm{e}} and LT0eL_{T_{0\mathrm{e}}} are the electron Larmor radius and temperature gradient scale length, respectively. This scaling was reproduced by nonlinear gyrokinetic simulations [19]. Several numerical studies have reported different saturation mechanisms for microtearing turbulence. One such mechanism is the local flattening of the electron temperature profile at low-order rational surfaces, which has been shown to reduce the linear drive and to play an important role in determining the saturation level of the electron heat flux [35].

The generation of zonal magnetic fields has been proposed as another saturation mechanism for microtearing turbulence. Some gyrokinetic simulations of microtearing turbulence have demonstrated that zonal magnetic fields play an important role in nonlinear saturation, as removing the zonal magnetic field component leads to a significant increase in the electron heat flux [20, 36]. In contrast, other studies using gyrokinetic simulations have reported that zonal fields do not play a significant role in nonlinear saturation [19, 35, 37]. The role of zonal magnetic fields in microtearing turbulence remains controversial. We note that zonal magnetic fields were not explicitly considered in the saturation theory of Drake et al. [34].

We investigate the self-organization mechanism of zonal magnetic fields in microtearing turbulence using the configuration without magnetic shear considered by Drake et al. [34]. This configuration retains the destabilization mechanism of the collisional microtearing mode, i.e., the time-dependent thermal force, since magnetic shear acts only to stabilize the mode [23]. A region of weak or nearly zero magnetic shear is realized in magnetic confinement devices as well, e.g., in stellarators [13, 38, 39] and in advanced tokamaks [40, 41, 42]. The insights obtained in this study are expected to contribute to the understanding of electromagnetic turbulence in weak- or nearly-zero-magnetic-shear regimes.

In this paper, we present numerical analyses of collisional microtearing turbulence in the absence of magnetic shear using a gyrokinetic model. We restrict our analyses to slab geometry as the simplest problem setup, so that toroidal effects such as the magnetic drift and trapped particle effects are excluded. We show that zonal magnetic fields are self-generated by the nonlinear interaction between magnetic perturbations and electron temperature fluctuations. The zonal magnetic fields act as an effective magnetic shear, leading to the formation of magnetic islands even in the absence of equilibrium magnetic shear. We also show that the electron temperature profile is flattened within the self-generated magnetic islands. Regardless of the presence of magnetic shear, this flattening weakens the local temperature-gradient drive of the microtearing instability and is expected to promote turbulence saturation.

The remainder of this paper is organized as follows. In § 2, we describe the gyrokinetic model and the simulation setup, and show that the microtearing modes are unstable in the absence of magnetic shear. In § 3, we present the nonlinear simulation results, including the analysis of the gyrokinetic free energy in terms of fluid moments, the formation of magnetic islands in the absence of magnetic shear, and the generation mechanism of zonal magnetic fields. The flattening of the electron temperature profile within the magnetic islands is also shown in the same section. Finally, we summarize the paper in § 4.

Gyrokinetic framework and linear instability

Model and simulation setup

We solve the electromagnetic gyrokinetic equations of ions and electrons for the non-Boltzmann part hsh_s of the perturbed distribution function δfs\delta f_s, where δfs=−(qsϕ/T0s)f0s+hs\delta f_s = -(q_s\phi/T_{0s})f_{0s} + h_s, and the subscript ss is the species label (s=i,es = \mathrm{i}, \mathrm{e}). The gyrokinetic equation and the definitions of the associated symbols are given in Appendix A.§ 1. The finite Larmor radius (FLR) effects are fully retained for ions. Electron FLR effects are also retained, but they are small for the parameters specified below. The parallel magnetic perturbation δB∥\delta B_{\parallel} is not included in this paper.

We perform gyrokinetic simulations of electromagnetic turbulence driven by collisional microtearing modes in a 2–D slab geometry using the GS2 code [43]. We consider a non-uniform plasma with electron pressure P0e(x)P_{0e}(x) embedded in an equilibrium magnetic field B0=B0z^+Byeq(x)y^\mathbf{B}_0 = B_0\hat{\mathbf{z}} + B_y^{\mathrm{eq}}(x)\hat{\mathbf{y}}. Here, Byeq(x)=B0(x/Ls)B_y^{\mathrm{eq}}(x) = B_0(x/L_s) is the sheared field component with LsL_s being the magnetic shear scale length. Unless otherwise stated, we consider the configuration without magnetic shear, i.e., Ls−1=0L_s^{-1} = 0. The electron pressure gradient is taken in the xx direction and is characterized by LP0e−1=−∂xln⁡P0e(x)L_{P_{0e}}^{-1} = -\partial_x \ln P_{0e}(x). The corresponding diamagnetic drift frequency is given by ω∗,e=ω∗,n0e+ω∗,T0e\omega_{*,e} = \omega_{*,n_{0e}} + \omega_{*,T_{0e}}, where ω∗,n0e\omega_{*,n_{0e}} and ω∗,T0e\omega_{*,T_{0e}} are the diamagnetic drift frequencies associated with the density gradient Ln0e−1L_{n_{0e}}^{-1} and temperature gradient LT0e−1L_{T_{0e}}^{-1}, respectively. The definitions of ω∗,n0e\omega_{*,n_{0e}} and ω∗,T0e\omega_{*,T_{0e}} are given in Appendix A.§ 1. The characteristic time scale of the system is determined by ω∗,e−1\omega_{*,e}^{-1}. In this paper, ω∗,e\omega_{*,e} is evaluated at kyρi=1k_y\rho_i = 1. With the parameters presented later, ω∗,e−1\omega_{*,e}^{-1} is related to the commonly used time scale LT0e/vth,iL_{T_{0e}}/v_{\mathrm{th},i} by ω∗,e(LT0e/vth,i)=0.6\omega_{*,e}(L_{T_{0e}}/v_{\mathrm{th},i}) = 0.6. The time variable is normalized by ω∗,e−1\omega_{*,e}^{-1} and denoted by t^\hat{t} throughout this paper.

The physical parameters are set as follows: the electron plasma beta βe≡n0eT0e/(B02/2μ0)=0.01\beta_e \equiv n_{0e}T_{0e}/(B_0^2/2\mu_0) = 0.01, the temperature ratio T0i/T0e=1T_{0i}/T_{0e} = 1, the density ratio n0i/n0e=1n_{0i}/n_{0e} = 1, the mass ratio me/mi=1/1836m_e/m_i = 1/1836, the normalized temperature gradients Ln0e/LT0e=5L_{n_{0e}}/L_{T_{0e}} = 5 and Ln0i/LT0i=0L_{n_{0i}}/L_{T_{0i}} = 0, and the ion charge number −qi/qe=1-q_i/q_e = 1. GS2 employs the model Fokker–Planck collision operator [44, 45], which includes pitch-angle scattering, energy diffusion, and terms that ensure the conservation of the particle number, momentum, and energy. The electron collisionality is set to νee/ω∗,e=0.15\nu_{ee}/\omega_{*,e} = 0.15. Since we assume a single ion species with −qi/qe=1-q_i/q_e = 1, we have νee=νei\nu_{ee} = \nu_{ei}. We set the ion collision frequency to be much smaller than that for electrons in this paper.

Microtearing instability in the absence of magnetic shear

We perform linear simulations to identify the range of kyρik_y\rho_i over which microtearing modes are unstable. In linear runs, we set the numerical resolution in the velocity space to (Nλ,NE)=(8,16)(N_{\lambda}, N_E) = (8, 16), where NλN_{\lambda} and NEN_E denote the numbers of grid points in the pitch-angle and energy coordinates, respectively [46].

Figure 1 shows the linear growth rate (left) and real frequency (right) of microtearing modes as a function of kyρik_y\rho_i. The eigenvalues are normalized by ω∗,e\omega_{*,e}.

Figure 1. Growth rate (left) and real frequency (right) of microtearing modes as a function of kyρik_y\rho_i.

The solid red lines denote the eigenvalues without magnetic shear. The microtearing modes become unstable even in the absence of magnetic shear (left) and propagate in the direction of the electron diamagnetic drift (right). For comparison, the eigenvalues with finite magnetic shear are also shown by the green circles and orange squares. The magnetic shear becomes stronger as LT0e/LsL_{T_{0e}}/L_s increases at fixed LT0eL_{T_{0e}}. The real frequency is unaffected by magnetic shear, while the growth rate is suppressed by its stabilizing effect, consistent with theoretical [23] and numerical [13, 40] studies. In the finite magnetic shear case, sufficient grid points along the equilibrium magnetic field are used to resolve the fine structure of the mode. We have confirmed the convergence of the results.

Nonlinear evolution and self-organization in microtearing turbulence

In this section, we present the results of nonlinear simulations in the absence of magnetic shear. The simulations are performed with the numerical resolution (Nx,Ny,Nλ,NE)=(64,64,8,16)(N_x, N_y, N_\lambda, N_E) = (64, 64, 8, 16) in the domain of Lx/ρi=Ly/ρi=10πL_x/\rho_i = L_y/\rho_i = 10\pi. Artificial hyperviscosity is included to dissipate the energy at small scales and to achieve a steady state. Its amplitude and cut-off wavenumber are chosen to be Dhyp/ω∗,e≈0.3D_{\mathrm{hyp}}/\omega_{*,e} \approx0.3 and kx,ycutρi≈0.9kx,ymaxρik^{\mathrm{cut}}_{x,y}\rho_i \approx0.9 k^{\mathrm{max}}_{x,y}\rho_i respectively, where kx,ymaxρi≈2π(ρi/Lx,y)(Nx,y/3)k^{\mathrm{max}}_{x,y}\rho_i \approx2\pi(\rho_i/L_{x,y})(N_{x,y}/3) is the maximum wavenumber. As shown in § 3.1, reaching nonlinear saturation requires long simulation times, which makes high-resolution simulations computationally expensive. We therefore use a lower spatial resolution in this paper. We have confirmed that the conclusions of this paper do not change with resolution and hyperviscosity.

Analysis of energy evolution and fluid-moment contributions

We first consider the energetics of the system. The gyrokinetic free energy WW consists of the particle part KsK_s and the magnetic field part M⊥M_\perp, and is given by

W=∑sKs+M⊥=1LxLy∫[∑s∫T0s∣δfs∣22f0s dv+∣∇⊥A∥∣22μ0]dr.(1)W = \sum_s K_s + M_\perp= \frac{1}{L_xL_y} \int\left[\sum_s \int\frac{T_{0s}|\delta f_s|^2}{2f_{0s}}\,\mathrm{d}\mathbf{v} + \frac{|\nabla_\perp A_\parallel|^2}{2\mu_0}\right] \mathrm{d}\mathbf{r}. \tag*{(1)}

The subscripts ∥\parallel and ⊥\perp denote the components parallel and perpendicular to the equilibrium magnetic field, respectively. We decompose the particle energy KsK_s into contributions from low-order fluid moments by expanding δfs\delta f_s in Hermite–Laguerre polynomials

Ks≈Ksn~+Ksu~∥+KsT~+KsT~∥,⊥,(2)K_s \approx K_s^{\tilde n} + K_s^{\tilde u_\parallel} + K_s^{\tilde T} + K_s^{\tilde T_{\parallel,\perp}}, \tag*{(2)}

where the terms on the right-hand side are defined as follows:

Ksn~=1LxLy∫12n0sT0s∣n~sn0s∣2 dr,(3)K_s^{\tilde n} = \frac{1}{L_xL_y} \int\frac{1}{2}n_{0s}T_{0s}\left|\frac{\tilde n_s}{n_{0s}}\right|^2 \,\mathrm{d}\mathbf{r}, \tag*{(3)}
Ksu~∥=1LxLy∫n0sms2∣u~∥,s∣2 dr,(4)K_s^{\tilde u_\parallel} = \frac{1}{L_xL_y} \int\frac{n_{0s}m_s}{2}|\tilde u_{\parallel,s}|^2 \,\mathrm{d}\mathbf{r}, \tag*{(4)}
KsT~=1LxLy∫34n0sT0s∣T~sT0s∣2 dr,(5)K_s^{\tilde T} = \frac{1}{L_xL_y} \int\frac{3}{4}n_{0s}T_{0s}\left|\frac{\tilde T_s}{T_{0s}}\right|^2 \,\mathrm{d}\mathbf{r}, \tag*{(5)}
KsT~∥,⊥=1LxLy∫n0sT0s6∣T~∥,s−T~⊥,s∣2T0s2 dr.(6)K_s^{\tilde T_{\parallel,\perp}} = \frac{1}{L_xL_y} \int\frac{n_{0s}T_{0s}}{6}\frac{|\tilde T_{\parallel,s}-\tilde T_{\perp,s}|^2}{T_{0s}^2}\,\mathrm{d}\mathbf{r}. \tag*{(6)}

The quantities n~s\tilde n_s, u~∥,s\tilde u_{\parallel,s}, T~s\tilde T_s, T~∥,s\tilde T_{\parallel,s}, and T~⊥,s\tilde T_{\perp,s} are the velocity moments of δfs\delta f_s defined in Appendix A.§ 2. For simplicity, (2)–(6) are derived under the drift-kinetic approximation, k⊥ρs→0k_\perp\rho_s \to0.

Figure 2 shows the time evolution of the energies. All energies are normalized by the maximum value of the electron particle energy over time, Kemax⁡=max⁡Ke(t^)K_e^{\max} = \max K_e(\hat t). Unlike conventional tearing instabilities that release energy stored in the equilibrium magnetic field, microtearing instabilities are driven by the electron temperature gradient. As shown in Figure 2, the electron particle energy KeK_e dominates the gyrokinetic free energy, and the magnetic energy M⊥M_\perp remains much smaller. Nevertheless, the magnetic field structure plays a crucial role in the nonlinear evolution of the turbulence, as discussed later. Most of KeK_e is associated with isotropic electron temperature fluctuations (see KeT~K_e^{\tilde T}). The energy associated with the temperature anisotropy, KeT~∥,⊥K_e^{\tilde T_{\parallel,\perp}}, is much smaller than KeT~K_e^{\tilde T} because pitch-angle scattering relaxes the difference between T~∥,e\tilde T_{\parallel,e} and T~⊥,e\tilde T_{\perp,e}. Although not shown in the figure, Ken~K_e^{\tilde n} and Keu~∥K_e^{\tilde u_\parallel} are also much smaller than KeT~K_e^{\tilde T}.

Figure 2. Time evolution of the gyrokinetic free energy and its components.

We divide the time evolution of the energy shown in Figure 2 into four phases. In the linear phase (t^≲50\hat t \lesssim50), the energy grows exponentially, as it is injected from the background temperature gradient through the microtearing instability. In the subsequent early nonlinear phase (50<t^≲60050 < \hat{t} \lesssim600), the energy reaches its peak as the system transitions from linear growth to nonlinear dynamics. This phase consists of two stages. In the zonal-field generation stage (50<t^≲20050 < \hat{t} \lesssim200), coherent zonal magnetic fields, uniform in the yy direction, are self-generated through the nonlinear interaction between magnetic perturbations and electron temperature fluctuations. In the island formation stage (200<t^≲600200 < \hat{t} \lesssim600), these zonal magnetic fields act as an effective magnetic shear, thereby leading to the formation of magnetic islands. In the intermediate nonlinear phase (600<t^≲4200600 < \hat{t} \lesssim4200), the energy gradually decays. Finally, the energy becomes nearly constant in the saturation phase (t^>4200\hat{t} > 4200). This paper focuses on the early nonlinear phase.

Self-organization of magnetic islands and zonal magnetic fields

We next examine the generation of zonal magnetic fields and their role in the formation of magnetic islands.

Figure 3 shows snapshots of the spatial distribution of the power of the electron temperature fluctuation, T~e2\widetilde{T}_{e}^{2}, in the xx–yy plane at t^=150\hat{t} = 150, 480, 2910, and 4800. T~e\widetilde{T}_{e} is normalized by T0eρi/LT0eT_{0e}\rho_i/L_{T_{0e}}. The contour lines of A∥A_{\parallel}, also shown in Figure 3, represent the perturbed magnetic field lines in the xx–yy plane. Figure 3(a) shows that the magnetic fields are in a turbulent state. Subsequently, coherent chains of magnetic islands aligned in the yy direction are formed (Figure 3(b)). Note that these islands are formed even in the absence of magnetic shear. The island width then gradually decreases (Figure 3(c)), and the zonal field structure becomes dominant (Figure 3(d)). Throughout these processes, T~e2\widetilde{T}_{e}^{2} is localized along the magnetic field lines and gradually decays with time. This spatial correspondence indicates a coupling between the magnetic perturbations and the electron temperature fluctuations.

Figure 3. Snapshots of the spatial distribution of the electron temperature fluctuation power T~e2\widetilde{T}_{\mathrm{e}}^{2}. The contour lines indicate A∥A_{\parallel}.

Figure 4 shows the time evolution of the xx-averaged power of A∥A_{\parallel} for the zonal component (kyρi=0k_y\rho_i=0, solid blue line) and the non-zonal components (ky>0k_y>0, dashed red line). A∥A_{\parallel} is normalized by (qevth,e/T0e)−1(ρi/LT0e)(q_{\mathrm{e}}v_{\mathrm{th,e}}/T_{0\mathrm{e}})^{-1}(\rho_i/L_{T_{0\mathrm{e}}}). To focus on the island formation, we show the time evolution for t^≤600\hat{t}\leq600. The non-zonal components grow exponentially in the linear phase (t^≲50\hat{t}\lesssim50), and their growth stops around t^≈100\hat{t}\approx100. The zonal component grows much more rapidly for 50<t^≲10050<\hat{t}\lesssim100 and reaches an amplitude comparable to that of the non-zonal components.

Figure 4. Time evolution of the xx-averaged power of the zonal and non-zonal components of A∥A_\parallel.

We decompose A∥A_{\parallel} in the xx–yy plane at t^=480\hat{t}=480 into its zonal and non-zonal components, shown in Figure 5(a) and Figure 5(b), respectively. The contour lines also denote A∥A_{\parallel}. The chains of magnetic islands aligned in the yy direction are absent when the zonal component is removed (Figure 5(b)), highlighting the role of zonal magnetic fields in their formation. Figure 6 shows the xx profile of the zonal magnetic field ⟨By⟩y=−∂⟨A∥⟩y/∂x\langle B_y\rangle_y=-\partial\langle A_{\parallel}\rangle_y/\partial x at different times, where ⟨⋅⟩y\langle\cdot\rangle_y denotes the yy-average. ByB_y is normalized by B0ρi/LT0eB_0\rho_i/L_{T_{0\mathrm{e}}}. The O-points of the magnetic islands are located near the xx positions where ⟨By⟩y=0\langle B_y\rangle_y=0. The turbulence thus generates its own magnetic shear, allowing the magnetic perturbations to self-organize into island chains even in a shearless equilibrium.

Figure 5. A snapshot of A∥A_\parallel for (a) the zonal component and (b) the non-zonal components at t^=480\hat{t}=480. The contour lines also denote A∥A_\parallel.

Figure 6. Zonal magnetic field profile in the xx direction at different times. ⟨By⟩y=−∂⟨A∥⟩y/∂x\langle B_y\rangle_y=-\partial\langle A_\parallel\rangle_y/\partial x is plotted, where ⟨⋅⟩y\langle\cdot\rangle_y denotes the yy-average.

To identify the source of the zonal magnetic fields, we analyze the power balance of A∥A_{\parallel} using the generalized Ohm’s law. The growth rate of ∣A∥∣|A_{\parallel}| at each kyk_y is decomposed into contributions from the individual terms in the generalized Ohm’s law:

γ^ky=∑jγ^kyj,(7)\hat{\gamma}_{k_y}=\sum_j\hat{\gamma}^{j}_{k_y}, \tag*{(7)}

where the hat on γ\gamma denotes normalization by ω∗,e\omega_{*,e}. The index jj labels the contributions from electron inertia, E×BE \times B advection, nonlinear interactions between magnetic perturbations and density and temperature fluctuations, and the collision term. The detailed derivation of (7) is given in Appendix B.

Figure 7 shows the time evolution of each contribution to the growth rate of the zonal component (kyρi=0k_y\rho_i=0) during the linear phase and the zonal-field generation stage (t^≤200\hat{t}\leq200). The solid black, solid red, and dashed blue lines represent γ^ky\hat{\gamma}_{k_y}, γ^kyA∥,T~\hat{\gamma}^{A_\parallel,\tilde{T}}_{k_y}, and γ^kycoll\hat{\gamma}^{\mathrm{coll}}_{k_y}, respectively, and the dotted gray line represents the sum of the other terms. For t^≲50\hat{t}\lesssim50, γ^ky≈0\hat{\gamma}_{k_y}\approx0 because the zonal component has no linear drive. The zonal component grows rapidly for 50<t^≲10050 < \hat{t} \lesssim100, driven primarily by γ^kyA∥,T~\hat{\gamma}_{k_y}^{A_\parallel,\tilde{T}}. The term γ^kyA∥,T~\hat{\gamma}_{k_y}^{A_\parallel,\tilde{T}} represents the nonlinear interaction between magnetic perturbations and electron temperature fluctuations in the generalized Ohm’s law. Zonal magnetic fields are generated by this nonlinear interaction.

Figure 7. Time evolution of the growth rate of ∣A∥∣|A_\parallel| for the zonal component (kyρi=0k_y\rho_i=0).

Electron temperature profile flattening within magnetic islands

In this section, we analyze the electron temperature fluctuation profile within the self-generated magnetic islands. Figure 8 shows the spatial distribution of T~e\tilde{T}_e at t^=480\hat{t}=480. The contour lines represent A∥A_\parallel. Remarkably, the temperature fluctuations spontaneously organize within the self-generated magnetic islands to produce a gradient that nearly cancels the background temperature gradient. We focus on the magnetic island centered at (x/ρi,y/ρi)≈(2.9,2.5)(x/\rho_i, y/\rho_i) \approx(2.9, 2.5). The island width is w≈9.8ρiw \approx9.8\rho_i. With the normalized temperature fluctuation T^e=T~e/(T0eρi/LT0e)\widehat{T}_e = \widetilde{T}_e/(T_{0e}\rho_i/L_{T_{0e}}) and the normalized coordinate x^=x/ρi\widehat{x} = x/\rho_i, T^e\widehat{T}_e varies from approximately −5-5 to 55 across the island, so that ∂T^e/∂x^=(LT0e/T0e)∂T~e/∂x≈1\partial\widehat{T}_e/\partial\widehat{x} = (L_{T_{0e}}/T_{0e})\partial\widetilde{T}_e/\partial x \approx1. In the same normalization, the background temperature gradient is (LT0e/T0e)∂T0e/∂x=−1(L_{T_{0e}}/T_{0e})\partial T_{0e}/\partial x = -1, where ∂T0e/∂x=−T0e/LT0e\partial T_{0e}/\partial x = -T_{0e}/L_{T_{0e}}. Although the above estimate is based on a single island in the snapshot shown in Figure 8, we have confirmed that the flattening also occurs in other islands and at other times.

Figure 8. Snapshot of the electron temperature fluctuation during the formation of magnetic islands. The contour lines denote A∥A_{\parallel}.

The flattening weakens the local temperature-gradient drive of the microtearing instability and is therefore expected to promote turbulence saturation. The same saturation mechanism may thus operate with and without equilibrium magnetic shear.

Summary

We have investigated the self-organization of magnetic islands and zonal magnetic fields in collisional microtearing turbulence in the absence of magnetic shear, using gyrokinetic simulations with the GS2 code.

We have shown that the gyrokinetic free energy is dominated by the electron particle energy associated with isotropic electron temperature fluctuations. The power balance analysis of A∥A_{\parallel} based on the generalized Ohm’s law has demonstrated that zonal magnetic fields are generated by the nonlinear interaction between magnetic perturbations and electron temperature fluctuations. These self-generated zonal magnetic fields act as an effective magnetic shear, allowing the magnetic perturbations to self-organize into magnetic islands even in the absence of equilibrium magnetic shear. We have also found that, within the magnetic islands, the electron temperature fluctuations produce a gradient that nearly cancels the background temperature gradient, thereby flattening the electron temperature profile. This flattening weakens the instability drive of the microtearing modes and is therefore expected to promote turbulence saturation, suggesting that the same saturation mechanism may operate with and without equilibrium magnetic shear. The overall self-organization process is illustrated schematically in Figure 9. A detailed investigation of the subsequent evolution in the intermediate nonlinear and saturation phases is left for future work.

Figure 9. Schematic diagram of the self-organization process driven by microtearing modes in the absence of magnetic shear.

The results of this paper may have implications for electromagnetic turbulence in low-magnetic-shear regions of magnetic confinement devices. In the Wendelstein 7-X stellarator, microtearing turbulence has recently been reported in a region with nearly zero magnetic shear [13]. Weak or nearly zero magnetic shear is also relevant to advanced tokamak operating scenarios, such as those planned for JT-60SA [41, 42]. In such regions, the self-organization mechanism investigated in this work may play an important role in the nonlinear evolution of microtearing turbulence.

We speculate that the role of zonal magnetic fields in the nonlinear evolution of microtearing turbulence depends on the strength of magnetic shear. In weak-magnetic-shear regimes, self-generated zonal magnetic fields may provide an effective magnetic shear and play an important role in the nonlinear evolution, as observed in this study. In contrast, in regimes with sufficiently strong magnetic shear, such self-organization may be suppressed, and zonal magnetic fields may play a less significant role. A systematic investigation of the relationship between magnetic shear strength and the generation of zonal magnetic fields is therefore an important direction for future work.

Acknowledgements

This work was performed on “Plasma Simulator” of NIFS with the support and under the auspices of the NIFS Collaboration Research program (NIFS25KISM016). This work was supported by a research grant from Wesco Science Promotion and the Notre Dame Seishin University research grant.

Gyrokinetic–Maxwell equations and fluid moments

Gyrokinetic–Maxwell equations

The non-Boltzmann part hsh_s of the perturbed distribution function of species ss (s=i,es = \mathrm{i}, \mathrm{e}) obeys the gyrokinetic equation given by

∂hs∂t+v∥(B0B0+B⊥B0)⋅∇hs+uE⋅∇⊥hs=qsf0sT0s(∂∂t+v∗,sT(v)⋅∇⊥)⟨χ⟩Rs+C(hs),(8)\frac{\partial h_s}{\partial t} + v_{\parallel}\left(\frac{\boldsymbol{B}_0}{B_0}+\frac{\boldsymbol{B}_{\perp}}{B_0}\right)\cdot\nabla h_s + \boldsymbol{u}_{E}\cdot\nabla_{\perp}h_s = \frac{q_s f_{0s}}{T_{0s}}\left(\frac{\partial}{\partial t}+\boldsymbol{v}^{T}_{*,s}(v)\cdot\nabla_{\perp}\right)\langle\chi\rangle_{R_s}+C(h_s), \tag*{(8)}

where C(hs)C(h_s) is the collision term. The E×B\boldsymbol{E}\times\boldsymbol{B} drift velocity is defined by uE=(B0/B0)×∇⊥⟨ϕ⟩Rs/B0\boldsymbol{u}_{E}=(\boldsymbol{B}_0/B_0)\times\nabla_{\perp}\langle\phi\rangle_{R_s}/B_0. B⊥=∇⊥⟨A∥⟩Rs×(B0/B0)\boldsymbol{B}_{\perp}=\nabla_{\perp}\langle A_{\parallel}\rangle_{R_s}\times(\boldsymbol{B}_0/B_0) is the magnetic perturbation associated with the parallel magnetic vector potential A∥A_{\parallel}. The gyrokinetic potential is defined as χ=ϕ−v∥A∥\chi=\phi-v_{\parallel}A_{\parallel}. Here, δB∥\delta B_{\parallel} is neglected. The angle brackets ⟨⋅⟩Rs\langle\cdot\rangle_{R_s} denote the gyroaverage at fixed guiding-center position Rs=r+v×(B0/B0)/Ωcs\boldsymbol{R}_s=\boldsymbol{r}+\boldsymbol{v}\times(\boldsymbol{B}_0/B_0)/\Omega_{cs}, where r\boldsymbol{r} is the particle position and Ωcs=qsB0/ms\Omega_{cs}=q_sB_0/m_s is the cyclotron frequency. The equilibrium distribution function f0sf_{0s} is the local Maxwellian,

f0s=n0sπ3/2vth,s3exp⁡(−v2vth,s2),(9)f_{0s}=\frac{n_{0s}}{\pi^{3/2}v_{\mathrm{th},s}^{3}}\exp\left(-\frac{v^2}{v_{\mathrm{th},s}^{2}}\right), \tag*{(9)}

with vth,s=2T0s/msv_{\mathrm{th},s}=\sqrt{2T_{0s}/m_s} and v=∣v∣v=|\boldsymbol{v}|. The velocity-dependent diamagnetic drift velocity is given by

v∗,sT(v)=−T0sqsB0[1Ln0s+(v2vth,s2−32)1LT0s]y^,(10)\boldsymbol{v}^{T}_{*,s}(v)=-\frac{T_{0s}}{q_sB_0}\left[\frac{1}{L_{n0s}}+\left(\frac{v^2}{v_{\mathrm{th},s}^{2}}-\frac{3}{2}\right)\frac{1}{L_{T0s}}\right]\hat{\boldsymbol{y}}, \tag*{(10)}

where the density and temperature gradients are characterized by Ln0s−1=−∂xln⁡n0sL_{n0s}^{-1}=-\partial_x\ln n_{0s} and LT0s−1=−∂xln⁡T0sL_{T0s}^{-1}=-\partial_x\ln T_{0s}, respectively. These gradients give rise to the diamagnetic drift frequencies, defined as

ω∗,n0s,T0s≡−∣qs∣qskyρs2vth,sLn0s,T0s,(11)\omega_{*,n0s,T0s}\equiv-\frac{|q_s|}{q_s}\frac{k_y\rho_s}{2}\frac{v_{\mathrm{th},s}}{L_{n0s,T0s}}, \tag*{(11)}

where ρs=vth,s/∣Ωcs∣\rho_s=v_{\mathrm{th},s}/|\Omega_{cs}| is the Larmor radius and kyk_y is the wavenumber in the yy direction. Since the pressure gradient is given by LP0s−1=Ln0s−1+LT0s−1L_{P0s}^{-1}=L_{n0s}^{-1}+L_{T0s}^{-1}, the diamagnetic drift frequency associated with LP0s−1L_{P0s}^{-1} is defined as ω∗,s≡ω∗,n0s+ω∗,T0s\omega_{*,s}\equiv\omega_{*,n0s}+\omega_{*,T0s}.

The gyrokinetic equation (A.1) is closed by the quasi-neutrality condition

∑sqs2n0sT0sϕ=∑sqs∫⟨hs⟩r dv,(12)\sum_s \frac{q_s^2 n_{0s}}{T_{0s}} \phi= \sum_s q_s \int\langle h_s \rangle_r \,\mathrm{d}v, \tag*{(12)}

and the parallel component of Ampère’s law

−∇⊥2A∥=μ0∑sqs∫v∥⟨hs⟩r dv,(13)-\nabla_\perp^2 A_\parallel= \mu_0 \sum_s q_s \int v_\parallel\langle h_s \rangle_r \,\mathrm{d}v, \tag*{(13)}

which determine ϕ\phi and A∥A_\parallel, respectively. Here, μ0\mu_0 is the vacuum permeability, and the brackets ⟨⋅⟩r\langle\cdot\rangle_r denote the gyroaverage at fixed particle position rr.

Fluid moments

Some low-order velocity moments of δfs\delta f_s are given by

n~s=∫δfs dv,(14)\widetilde{n}_s = \int\delta f_s \,\mathrm{d}v, \tag*{(14)}
n0su~∥,s=∫v∥δfs dv,(15)n_{0s}\widetilde{u}_{\parallel,s} = \int v_\parallel\delta f_s \,\mathrm{d}v, \tag*{(15)}
p~∥,s=∫msv∥2δfs dv,(16)\widetilde{p}_{\parallel,s} = \int m_s v_\parallel^2 \delta f_s \,\mathrm{d}v, \tag*{(16)}
p~⊥,s=∫msv⊥22δfs dv,(17)\widetilde{p}_{\perp,s} = \int\frac{m_s v_\perp^2}{2} \delta f_s \,\mathrm{d}v, \tag*{(17)}

where n~s\widetilde{n}_s, u~∥,s\widetilde{u}_{\parallel,s}, p~∥,s\widetilde{p}_{\parallel,s}, and p~⊥,s\widetilde{p}_{\perp,s} are the density, parallel flow velocity, and parallel and perpendicular pressure fluctuations, respectively. To first order in the fluctuations, the pressure fluctuations are decomposed as p~∥,s=n~sT0s+n0sT~∥,s\widetilde{p}_{\parallel,s} = \widetilde{n}_s T_{0s} + n_{0s}\widetilde{T}_{\parallel,s} and p~⊥,s=n~sT0s+n0sT~⊥,s\widetilde{p}_{\perp,s} = \widetilde{n}_s T_{0s} + n_{0s}\widetilde{T}_{\perp,s}. The parallel and perpendicular temperature fluctuations are therefore given by

T~∥,s=1n0s∫(msv∥2−T0s)δfs dv,(18)\widetilde{T}_{\parallel,s} = \frac{1}{n_{0s}} \int\left(m_s v_\parallel^2 - T_{0s}\right)\delta f_s \,\mathrm{d}v, \tag*{(18)}
T~⊥,s=1n0s∫(msv⊥22−T0s)δfs dv.(19)\widetilde{T}_{\perp,s} = \frac{1}{n_{0s}} \int\left(\frac{m_s v_\perp^2}{2} - T_{0s}\right)\delta f_s \,\mathrm{d}v. \tag*{(19)}

The isotropic temperature fluctuation is defined as

T~s=T~∥,s+2T~⊥,s3=1n0s∫(msv23−T0s)δfs dv.(20)\widetilde{T}_s = \frac{\widetilde{T}_{\parallel,s} + 2\widetilde{T}_{\perp,s}}{3} = \frac{1}{n_{0s}} \int\left(\frac{m_s v^2}{3} - T_{0s}\right)\delta f_s \,\mathrm{d}v. \tag*{(20)}

Power balance equation for A∥A_\parallel

In this appendix, we derive the power balance equation for A∥A_\parallel used in § 3.2. We consider the electron gyrokinetic equation (A.1) in the drift-kinetic approximation and in the absence of magnetic shear. Taking its v∥v_\parallel moment yields the electron parallel momentum equation. Solving the momentum equation for ∂A∥/∂t\partial A_\parallel/\partial t, we obtain the generalized Ohm’s law,

∂A∥∂t=−meqedu~∥,edt−uE⋅∇⊥A∥−T0eqeB⊥B0⋅∇⊥(n~en0e+T~∥,eT0e)−v∗,n0e(1+ηe)∂A∥∂y−E∥coll,(21)\frac{\partial A_\parallel}{\partial t}=-\frac{m_e}{q_e}\frac{d\widetilde{u}_{\parallel,e}}{dt}-\mathbf{u}_E\cdot\nabla_\perp A_\parallel-\frac{T_{0e}}{q_e}\frac{\mathbf{B}_\perp}{B_0}\cdot\nabla_\perp\left(\frac{\widetilde{n}_e}{n_{0e}}+\frac{\widetilde{T}_{\parallel,e}}{T_{0e}}\right)-v_{*,n_{0e}}(1+\eta_e)\frac{\partial A_\parallel}{\partial y}-E_\parallel^{\mathrm{coll}}, \tag*{(21)}

where d/dt=∂/∂t+uE⋅∇⊥d/dt=\partial/\partial t+\mathbf{u}_E\cdot\nabla_\perp is the convective time derivative, v∗,n0e=ω∗,n0e/kyv_{*,n_{0e}}=\omega_{*,n_{0e}}/k_y is the electron diamagnetic drift velocity, and E∥collE_\parallel^{\mathrm{coll}} represents the contribution of the collision term to the parallel electric field. (B.1) has the dimensions of the parallel electric field, as follows from the parallel component of Faraday’s law, ∂A∥/∂t=−E∥\partial A_\parallel/\partial t=-E_\parallel. The temperature gradient normalized by the density gradient is denoted by ηe≡Ln0e/LT0e\eta_e\equiv L_{n_{0e}}/L_{T_{0e}}. Since the anisotropic temperature fluctuation is negligible compared with the isotropic one (Figure 2), we take T~∥,e=T~e\widetilde{T}_{\parallel,e}=\widetilde{T}_e.

We expand (B.1) in a Fourier series in the yy direction. Multiplying the kyk_y component by the complex conjugate A∥,ky∗(x,t)=A∥∗(x,ky,t)A^*_{\parallel,k_y}(x,t)=A^*_\parallel(x,k_y,t) and taking the real part, we obtain the power balance equation for A∥A_\parallel:

12∂∂t∣A∥,ky(x,t)∣2=∑jℜ[A∥,ky∗(x,t)Nkyj(x,t)],(22)\frac{1}{2}\frac{\partial}{\partial t}\left|A_{\parallel,k_y}(x,t)\right|^2=\sum_j\Re\left[A^*_{\parallel,k_y}(x,t)N^j_{k_y}(x,t)\right], \tag*{(22)}

where jj labels the following terms:

Nkyu~∥,e=−meqe(du~∥,edt)ky,(23)N^{\widetilde{u}_{\parallel,e}}_{k_y}=-\frac{m_e}{q_e}\left(\frac{d\widetilde{u}_{\parallel,e}}{dt}\right)_{k_y}, \tag*{(23)}
Nkyφ,A∥=−(uE⋅∇⊥A∥)ky,(24)N^{\varphi,A_\parallel}_{k_y}=-\left(\mathbf{u}_E\cdot\nabla_\perp A_\parallel\right)_{k_y}, \tag*{(24)}
NkyA∥,n~=−T0eqe(B⊥B0⋅∇⊥n~en0e)ky,(25)N^{A_\parallel,\widetilde{n}}_{k_y}=-\frac{T_{0e}}{q_e}\left(\frac{\mathbf{B}_\perp}{B_0}\cdot\nabla_\perp\frac{\widetilde{n}_e}{n_{0e}}\right)_{k_y}, \tag*{(25)}
NkyA∥,T~=−T0eqe(B⊥B0⋅∇⊥T~eT0e)ky,(26)N^{A_\parallel,\widetilde{T}}_{k_y}=-\frac{T_{0e}}{q_e}\left(\frac{\mathbf{B}_\perp}{B_0}\cdot\nabla_\perp\frac{\widetilde{T}_e}{T_{0e}}\right)_{k_y}, \tag*{(26)}
Nkycoll=−Ekycoll(x,t).(27)N^{\mathrm{coll}}_{k_y}=-E^{\mathrm{coll}}_{k_y}(x,t). \tag*{(27)}

(⋅)ky(\cdot)_{k_y} denotes the Fourier component at kyk_y, which includes the full nonlinear mode coupling. The linear diamagnetic drift term in (B.1) does not contribute to (B.2). This term is proportional to A∥∂A∥/∂y=∂(A∥2/2)/∂yA_\parallel\partial A_\parallel/\partial y=\partial(A_\parallel^2/2)/\partial y. The total yy-derivative vanishes upon averaging over yy, and the diamagnetic drift term only represents the propagation of A∥A_\parallel in the yy direction.

Averaging (B.2) over xx, we obtain

12∂PA∥,ky∂t=∑jFkyj.(28)\frac{1}{2}\frac{\partial P_{A_\parallel,k_y}}{\partial t}=\sum_j F^j_{k_y}. \tag*{(28)}

with

PA∥,ky≡⟨∣A∥,ky(x,t)∣2⟩x,(29)P_{A_{\parallel,k_y}} \equiv\left\langle\left|A_{\parallel,k_y}(x,t)\right|^2 \right\rangle_x, \tag*{(29)}
Fkyj≡⟨Re⁡[A∥,ky∗(x,t)Nkyj(x,t)]⟩x,(30)F^j_{k_y} \equiv\left\langle\operatorname{Re}\left[A^*_{\parallel,k_y}(x,t)N^j_{k_y}(x,t)\right]\right\rangle_x, \tag*{(30)}

where ⟨⋅⟩x\langle\cdot\rangle_x denotes the xx-average. Dividing (B.8) by PA∥,kyP_{A_{\parallel,k_y}} and introducing the growth rate γky≡∂ln⁡PA∥,ky1/2/∂t\gamma_{k_y} \equiv\partial\ln P^{1/2}_{A_{\parallel,k_y}}/\partial t, we obtain

γky=∑jγkyj,(31)\gamma_{k_y} = \sum_j \gamma^j_{k_y}, \tag*{(31)}

where γkyj≡Fkyj/PA∥,ky\gamma^j_{k_y} \equiv F^j_{k_y}/P_{A_{\parallel,k_y}} is the contribution of each term to the growth rate.

References

References

  1. [1]Dorland W, Jenko F, Kotschenreuther M and Rogers B N 2000 Phys. Rev. Lett. 85 5579
  2. [2]Dannert T and Jenko F 2005 Phys. Plasmas 12 072309
  3. [3]Applegate D J, Roach C M, Cowley S C, Dorland W D, Joiner N, Akers R J, Conway N J, Field A R, Patel A, Valovic M and Walsh M J 2004 Phys. Plasmas 11 5085
  4. [4]Wong K L, Kaye S M, Mikkelsen D R, Krommes J A, Hill K, Bell R and LeBlanc B 2007 Phys. Rev. Lett. 99 135003
  5. [5]Gutfetter W, Candy J, Kaye S M, Nevins W M, Wang E, Zhang J and Bell R E 2012 Phys. Plasmas 19 056119
  6. [6]Dickinson D, Roach C M, Saarelma S, Scannell R, Kirk A and Wilson H R 2013 Plasma Phys. Control. Fusion 55 074006
  7. [7]Patel B S, Hardman M R, Kennedy D, Giacomin M, Dickinson D and Roach C M 2025 Nucl. Fusion 65 026063
  8. [8]Doerk H, Jenko F, Görler T, Told D, Pueschel M J and Hatch D R 2012 Phys. Plasmas 19 055907
  9. [9]Jian X, Holland C, Candy J, Belli E, Chan V, Garofalo A M and Ding S 2019 Phys. Rev. Lett. 123 225002
  10. [10]Hatch D R, Kotschenreuther M, Mahajan S M, Pueschel M J, Michoski C, Merlo G, Hassan E, Field A R, Frassinetti L, Giroud C and JET Contributors 2021 Nucl. Fusion 61 036015
  11. [11]Predbeon P and Sattin F 2013 Phys. Plasmas 20 040702
  12. [12]Carmony D, Pueschel M J and Terry P W 2013 Phys. Plasmas 20 052110
  13. [13]Cu-Castillo H, Navarro A B, Merlo G, Reimold F, Romba T, Ford O, Bannmann S, Vanó L, Geiger J, Goodman A, Zocco A, Jenko F and W7-X Team 2026 Phys. Rev. Lett. 137 035102
  14. [14]Merlo G, Navarro A B, Görler T, Jenko F and Wilms F 2026 Nucl. Fusion 66 034003
  15. [15]Patel B S, Dickinson D, Roach C M and Wilson H R 2022 Nucl. Fusion 62 016009
  16. [16]Kennedy D, Giacomin M, Casson F J, Dickinson D, Hornsby W A, Patel B S and Roach C M 2023 Nucl. Fusion 63 126061
  17. [17]Giacomin M, Kennedy D, Casson F J, Ajay C J, Dickinson D, Patel B S and Roach C M 2024 Plasma Phys. Control. Fusion 66 055010
  18. [18]Rechester A B and Rosenbluth M N 1978 Phys. Rev. Lett. 40 38
  19. [19]Doerk H, Jenko F, Pueschel M J and Hatch D R 2011 Phys. Rev. Lett. 106 155003
  20. [20]Giacomin M, Dickinson D, Kennedy D, Patel B S and Roach C M 2023 Plasma Phys. Control. Fusion 65 095019
  21. [21]Kaye S M, Levinton F M, Stutman D, Tritz K, Yuh H, Bell M G, Bell R E, Horton W, Kim J, LeBlanc B P, Maingi R, Mazzucato E, Menard J E, Mikkelsen D, Park H, Rewoldt G, Sabbagh S A, Smith D R, Wang W and the NSTX Group 2007 Nucl. Fusion 47 499
  22. [22]Hazeltine R D, Dobrott D and Wang T S 1975 Phys. Fluids 18 1778
  23. [23]Drake J F and Lee Y C 1977 Phys. Fluids 20 1341
  24. [24]Gladd N T, Drake J F, Chang C L and Liu C S 1980 Phys. Fluids 23 1182
  25. [25]D’Ippolito D A, Drake J F and Lee Y C 1980 Phys. Fluids 23 771
  26. [26]Hassam A B 1980 Phys. Fluids 23 38
  27. [27]Yagyu M and Numata R 2023 Plasma Phys. Control. Fusion 65 065003
  28. [28]Hassam A B 1980 Phys. Fluids 23 2493
  29. [29]Zocco A, Loureiro N F, Dickinson D, Numata R and Roach C M 2015 Plasma Phys. Control. Fusion 57 065008
  30. [30]Geng C, Dickinson D and Wilson H 2020 Plasma Phys. Control. Fusion 62 085009
  31. [31]Catto P J and Rosenbluth M N 1981 Phys. Fluids 24 243
  32. [32]Applegate D J, Roach C M, Connor J W, Cowley S C, Dorland W, Hastie R J and Joiner N 2007 Plasma Phys. Control. Fusion 49 1113
  33. [33]Chandran B D G and Schekochihin A A 2024 J. Plasma Phys. 90 905900204
  34. [34]Drake J F, Gladd N T and Chang C L 1980 Phys. Rev. Lett. 44 994
  35. [35]Ajay C J, McMillan B F and Pueschel M J 2023 Nucl. Fusion 63 066024
  36. [36]Pueschel M J, Hatch D R, Kotschenreuther M, Ishizawa A and Merlo G 2020 Nucl. Fusion 60 124005
  37. [37]Hamed M, Hatch D R, Pueschel M J and Jitsuk T 2025 Phys. Plasmas 32 092303
  38. [38]Faber B J, Pueschel M J, Terry P W, Hegna C C and Roman J E 2018 J. Plasma Phys. 84 905840503
  39. [39]Mulholland P, Aleynikova K, Pueschel M J, Proll J H and Ishizawa A 2026 Nucl. Fusion 66 096008
  40. [40]Jian X, Holland C, Candy J, Ding S, Belli E, Chan V, Staebler G M, Garofalo A M, McClenaghan J and Snyder P 2021 Phys. Plasmas 28 042501
  41. [41]Wakatsuki T, Yoshida M, Narita E, Suzuki T and Hayashi N 2023 Nucl. Fusion 63 076017
  42. [42]Gabriellini S, Zotta V K, Garzotti L, Aiba N, Artaud J F, Giruzzi G, Pucella G, Sozzi C, Taylor D, Wakatsuki T, Buria L, Leoni C, Gatto R and the JT-60SA Integrated Project Team 2025 Nucl. Fusion 65 056041
  43. [43]Barnes M, Dickinson D, Dorland W, Hill P A, Parker J T, Roach C M, Giacomin M, Mandell N, Numata R, Schuett T, Biggs-Fox S, Christen N, Parisi J, Wilkie G, Anton L, Ball J, Baumgaertel J, Colyer G, Hardman M, Hein J, Highcock E, Howes G, Jackson A, Kotschenreuther M T, Lee J, Leggate H, Mauriya A, Patel B, Tatsuno T and Van Wyk F 2024 GS2 version 8.2.1 URL https://gyrokinetics.gitlab.io/gs2/
  44. [44]Abel I G, Barnes M, Cowley S C, Dorland W and Schekochihin A A 2008 Phys. Plasmas 15 122509
  45. [45]Barnes M, Abel I G, Dorland W, Ernst D R, Hammett G W, Ricci P, Rogers B N, Schekochihin A A and Tatsuno T 2009 Phys. Plasmas 16 072107
  46. [46]Barnes M, Dorland W and Tatsuno T 2010 Phys. Plasmas 17 032106

Paper details

vlnc:W_cjfruww7
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