Resonant Magnetic Perturbations (RMPs) in Tokamaks

• Resonant Magnetic Perturbations (RMPs) are externally applied, non-axisymmetric magnetic fields designed to modify the edge plasma by resonating with rational-q surfaces for effective ELM control.
• They alter the magnetic topology to create magnetic islands and stochastic regions, thereby enhancing parallel particle and heat transport without disrupting core confinement.
• Experimental studies in devices like DIII-D and MAST demonstrate that optimized RMP application leads to notable ELM suppression and density pump-out, balancing power thresholds and operational windows.

Resonant Magnetic Perturbations (RMPs) are externally applied, non-axisymmetric magnetic fields designed to alter the magnetohydrodynamic (MHD) stability and transport at the edge of toroidal plasma, primarily for the purpose of controlling Edge-Localized Modes (ELMs) in high-confinement (H-mode) tokamak regimes. RMPs modify the edge magnetic topology by resonating with low-order rational-q surfaces, driving magnetic islands and/or creating edge stochasticity, thereby altering particle and heat transport in the plasma boundary region without destroying core confinement. This technique plays a pivotal role in ELM control and has major implications for the power exhaust and operational flexibility of next-generation devices, such as ITER.

1. Physical Principles and Theoretical Framework

Resonant Magnetic Perturbations are implemented by applying small, toroidally non-axisymmetric magnetic fields with a specific helicity that matches rational surfaces of the plasma safety factor profile, $q(r) = m/n$, where $m$ and $n$ are poloidal and toroidal mode numbers. The applied field takes the generic spectral form

$$\delta B_r(r, \theta, \phi) = \sum_{m, n} \delta B_{mn}(r) e^{i(m\theta - n\phi)}$$

with the resonance condition $q(r_s) = m/n$ determining the locus of magnetic island formation or stochastization (Kirk et al., 2013, Hu et al., 2019, Hu et al., 2019).

In the context of resistive two-fluid MHD, as implemented in the TM1 code, the coupled equations for particle, momentum, generalized Ohm's law, induction, and energy include the influence of externally imposed fields and the plasma's response (rotation screening, resistivity, flows). The penetration and efficacy of RMPs is governed by a threshold amplitude required for the resonant field component at the pedestal top to overcome plasma screening: $B_r^{\rm th} \propto \rho_m \sim m_i n_{\rm ped}$ so that higher pedestal density leads to stronger screening and reduced RMP penetration (Hu et al., 2019). The resulting magnetic islands or stochastic layers provide enhanced parallel particle transport and flatten the electron pressure near the rational surface.

2. RMP-Induced Edge Transport: ELM Suppression and Pump-Out

RMPs exert two principal effects:

1. ELM Suppression: At sufficiently low edge density/collisionality, RMP-driven islands at the pedestal top (e.g., $m/n = 8/2$ at $\mathrm{UN}\approx0.93$ in DIII-D) equilibrate to a width $w\sim0.02$ in normalized flux. Parallel heat and particle transport across the island separatrix reduces the local electron pressure by $\Delta p_e/p_e\sim10-20\%$, stabilizing Peeling–Ballooning Modes (PBMs) below the stability threshold obtained from the EPED model:

$p_{\rm ped}^{\rm TM1} < p_{\rm PBM}^{\rm EPED}$

and thus suppressing Type-I ELMs (Hu et al., 2019, Kirk et al., 2013).

2. Density Pump-Out: At the pedestal foot ($q=11/2$, $\mathrm{UN}\approx0.99$), much smaller islands ($w\sim0.01-0.02$) drive enhanced parallel particle transport:

$$\Gamma_{\parallel} \sim \frac{nT}{\eta L_\parallel} \propto \frac{1}{n_{\rm ped}}$$

resulting in a substantial decrease in edge density (up to $-20\%$), with the magnitude of pump-out inversely proportional to pedestal density (Hu et al., 2019, Kirk et al., 2013, Hu et al., 2019).

The combination of these effects forms a narrow “island sandwich” structure—narrow islands are present only at the top and foot of the pedestal, with robust screening preventing stochasticity in the core of the edge transport barrier (ETB), and thus preserving the temperature gradient there ($\Delta T/T < 5\%$) (Hu et al., 2019).

3. Modelling Approaches and Criteria for Resonance

Pedestal and edge response to RMPs are modelled via nonlinear, time-dependent two-fluid MHD (e.g., TM1, coupled with GPEC or EFIT for equilibrium reconstruction), and linear resistive MHD codes such as MARS-F for plasma response and screening studies (Hu et al., 2019, Kirk et al., 2013, Kirk et al., 2013, Xie et al., 2021). The critical determinant for RMP efficacy is the penetration of the resonant field harmonic through the rotation (flow) and resistivity shield at the pedestal top. The width of the resonant magnetic island at $q = m/n$ is given by: $$w \approx 2\sqrt{\frac{r_s |\Delta B_{mn}|}{m B_\theta \, d q/dr}}$$ where $\Delta B_{mn}$ includes both the vacuum perturbation and the plasma response (Hu et al., 2019, Hu et al., 2019).

Field penetration is only achieved above a threshold $B_r^{\rm th}$ that grows with pedestal density. The position of rotation zero-crossing and the magnitude of local $E\times B$ and electron perpendicular flows at the rational surface further control the screening/penetration behavior (Xie et al., 2021).

4. Experimental Observations and Operational Regimes

Experiments in DIII-D and MAST have established the following operational features:

• RMPs suppress Type-I ELMs within discrete or (at lower density) contiguous $q_{95}$ windows. The $q_{95}$-extent of ELM suppression widens as pedestal density decreases, and continuous windows are achieved when the resonance threshold is sufficiently low for multiple islands to coexist near the pedestal top (Hu et al., 2019).
• The power penalty for entering H-mode ($P_{\rm LH}$) increases with RMP strength. Experiments show a $+20\%$ threshold increase for moderate ($n=3,4$) RMPs and up to $+60\%$ for high-$n$ ($n=6$) RMPs, with scaling threshold $b'_{\rm th}\sim(0.6\times 10^{-3})$ and a linear dependence $$P_{\rm LH}^{\rm RMP}(n)=P^0_{\rm LH}[1+\alpha_n(b'_{\rm res}-b'_{\rm th})]$$ beyond threshold (Scannell et al., 2014, Kriete et al., 2020).
• In MAST and DIII-D, type-I ELM frequency typically increases linearly with RMP coil current above the threshold, concomitantly reducing the energy per ELM by up to a factor of 3 or more, holding the product $$f_{\rm ELM}\Delta W_{\rm ELM}\approx \textrm{constant}$$ (Kirk et al., 2013, Kirk et al., 2013, Thornton et al., 2014, Kirk et al., 2013).
• Experiments confirm that complete ELM suppression occurs only over narrow operational windows at low density or collisionality; mitigation is achievable in broader conditions and is robust to plasma fueling and 3D alignment of the applied field spectrum (Kirk et al., 2013, Hu et al., 2019).

5. Magnetic Topology, Screening, and Edge Structure

A key feature of RMP action is the topological alteration of the edge field configuration. Penetration of the resonant component to $q=m/n$ surfaces forms magnetic islands whose width is analytically related to the perturbation amplitude and local shear: $$w_{mn} = \sqrt{ \frac{4\, |b^1_{mn}(s_0)| }{ q'(s_0) m } }$$ with $b^1_{mn}(s_0)$ the spectral component at the resonant surface (Cahyna et al., 2010).

Screening currents (arising from plasma rotation, resistivity, and equilibrium flows) suppress the RMP-induced field except where plasma rotation is sufficiently low and/or resistivity is high at the rational surfaces. Detailed modelling (e.g., with JOREK-CARIDDI, MARS-F) reveals that screening by conducting structures (PSL, wall, coil casing) can further attenuate the applied RMP, with only $\sim28\%$ of the field reaching the plasma edge at high frequency, necessitating increased coil currents to achieve equivalent plasma responses (Slotema et al., 2024).

Diagnostic imaging in MAST has demonstrated X-point lobe or manifold structures as direct, visible evidence of 3D separatrix splitting due to RMPs above threshold current. The number and separation of lobes correspond with the toroidal mode number of the applied field. The appearance and spatial scale of lobes correlate with the increase in ELM frequency and density pump-out, and their quantitative modeling matches field-line tracing predictions once screening is incorporated (Kirk et al., 2012, Thornton et al., 2013).

6. Implications for Reactor Design and Future Directions

Experimental and modeling work has demonstrated:

• The capability of RMPs to provide robust ELM suppression or mitigation relies on achieving field penetration at the pedestal top with minimal core confinement degradation. Wide $q_{95}$ suppression windows at high pedestal pressure can be realized by employing higher-$n$ RMP spectra; DIII-D two-fluid simulations predict $\Delta q_{95}\propto n$ for suppression at fixed edge field (Hu et al., 2019).
• Future reactor applications (e.g., ITER, DEMO) must balance the auxiliary power required for H-mode access ($P_{\rm LH}$) against the amplitude needed for reliable ELM suppression—a consequence of the shear-decorrelation and enhanced particle transport induced by RMPs (Scannell et al., 2014, Kriete et al., 2020).
• Plasma response optimization—minimizing the required RMP field at the edge, maximizing penetration at target surfaces, controlling up–down and toroidal phasing, and tailoring the applied spectral content—will be essential to minimize deleterious effects on core rotation, confinement, and fast-ion populations.
• Advanced modeling, including full two-fluid physics, realistic wall structures, and kinetic effects, is required to capture subtle interplay between screening, field penetration, density/collisionality dependence, and edge turbulence regimes (Slotema et al., 2024, Hu et al., 2019).

7. Summary Table of Key RMP Effects in DIII-D and MAST

Effect	DIII-D (n=2/3)	MAST (n=3/4/6)
ELM suppression window	$\Delta q_{95}<0.1-0.7$ (density-dependent)	Mitigation only
Threshold $n_{e,\rm ped}$	$\lesssim2.5\times10^{19}\mathrm{m}^{-3}$	No full suppression
Pedestal pressure drop at suppression	$10-20\%$	$10-20\%$
Density pump-out magnitude	$−(10-20)\%$	$−(10-20)\%$
Mitigation threshold (coil current)	$I_{\rm coil}\sim4$ kA (n=2)	$4-6$ kAt (n=3,4,6)
Island width $w_{mn}$ (top/foot)	$0.02$ / $0.01-0.02$ (norm. flux)	$0.01-0.03$ (norm. flux)
Scaling: $P_{\rm LH}$ increase	+20–60%, linear vs b$_{\rm res}'$	+20–100% (n=3–6)

RMPs stand as a cornerstone of ELM control physics and technology for next-step fusion reactors, conditional on a precise understanding and control of their field penetration, operational windows, topology modifications, and impact on overall plasma performance (Hu et al., 2019, Kirk et al., 2013, Hu et al., 2019, Kirk et al., 2012, Thornton et al., 2013, Slotema et al., 2024).