Pedestal Ideal in Tokamaks and Photonics

• Pedestal Ideal is a concept describing an idealized edge transport barrier in tokamak plasmas, defined by stability limits and engineered geometries.
• It integrates ideal-MHD thresholds with kinetic and statistical models to refine predictions of pedestal height, width, and overall performance.
• In photonics, pedestal structures are fabricated to enhance third-order nonlinearity through confined acoustic phonons in modified silicon geometries.

Pedestal ideal denotes, in the literature surveyed here, an idealized description of pedestal behavior in which the pedestal is constrained by formal stability limits or deliberately engineered geometry. In tokamak research, the pedestal is the steep-gradient edge transport barrier that forms in H-mode confinement and strongly influences core performance; in that setting, “ideal” most often refers to ideal magnetohydrodynamic ballooning or peeling–ballooning limits, sometimes used as proxies for kinetic ballooning onset or as boundaries in predictive workflows (Dickinson et al., 2011, McClenaghan et al., 24 Jun 2026). The same literature also shows that pedestal behavior is not exhausted by a single fixed ideal threshold: $E\times B$ shear, microturbulence, collisionality, stochastic magnetic transport, and impurity-modified transport all alter the operational meaning of pedestal ideality (Hatch et al., 2017, Miller et al., 2024, Banerjee et al., 1 Jun 2026). In a separate photonics usage, a pedestal structure is a fabricated geometry that improves nonlinear response by altering acoustic confinement rather than plasma stability (Zhang et al., 2016).

1. Pedestal as a technical object in confinement physics

In tokamak plasma physics, the pedestal is the edge transport barrier that forms in H-mode and sets the boundary condition for core confinement. Its experimentally used descriptors are pedestal height, pedestal width, and the local pressure-gradient region. A recent spherical-tokamak machine-learning framework defines the total pedestal width as

$$\Delta_{\mathrm{ped}} \equiv \frac{\Delta_{n_e,\mathrm{ped}}+\Delta_{T_e,\mathrm{ped}}}{2},$$

and characterizes pedestal height through the normalized pedestal pressure

$$\beta_{\theta,\mathrm{ped}} \equiv \frac{2\mu_0 p_{\mathrm{ped}}}{\overline{B}_{\mathrm{pol}}^2}, \qquad p_{\mathrm{ped}} \equiv 2\,p_e(\psi_{\mathrm{ped}}),$$

with $p_e=n_eT_e$ and $\overline{B}_{\mathrm{pol}}=\mu_0 I_p/l$ (Parisi et al., 28 Apr 2025). This is consistent with EPED-style usage, in which pedestal height and width are the central state variables.

The predictive importance of these quantities is underscored by the fact that EPED-like models do not predict the pedestal top electron density $n_{\text{e,ped}}$ internally; instead, additional closure models are required. The JET study on supervised learning for pedestal density frames this as a limitation of conventional EPED/EUROPED workflows, since pedestal top pressure can be physics-constrained while $n_{\text{e,ped}}$ must still be supplied externally (Kit et al., 2022). This suggests that any “ideal” pedestal picture which relies only on a single width–height constraint is structurally incomplete.

A further implication of the recent literature is that pedestal ideality is machine- and regime-dependent. Spherical tokamaks, conventional tokamaks, and high-field reactor-like cases do not exhibit the same width–height parameterization or the same dominant limiting instabilities (Parisi et al., 28 Apr 2025, McClenaghan et al., 24 Jun 2026). The literature therefore suggests that pedestal ideality is better understood as a family of constrained operating states than as a universal edge-state formula.

2. Ideal ballooning, peeling–ballooning, and KBM proxies

The classical ideal-MHD interpretation treats the pedestal as limited by ballooning or peeling–ballooning stability. On MAST, inter-ELM reconstructions show that pedestal height and width increase together while the peak pressure gradient changes only weakly, by less than about $20\%$ over the ELM cycle; simultaneously, the region unstable to $n=\infty$ ideal ballooning modes broadens and moves inward (Dickinson et al., 2011). All five experimental equilibria examined in that study are unstable to infinite-$n$ ballooning modes somewhere in the pedestal, and the radial extent of the unstable region increases during the cycle. The same work reports a good correlation between the region unstable to infinite-$$\Delta_{\mathrm{ped}} \equiv \frac{\Delta_{n_e,\mathrm{ped}}+\Delta_{T_e,\mathrm{ped}}}{2},$$0 ideal ballooning modes and the region in which GS2 identifies KBMs with twisting parity as the dominant microinstabilities (Dickinson et al., 2011).

That result is technically important because it refines, rather than simply confirms, the notion of a gradient-limited pedestal. The MAST analysis does not support a fixed, externally imposed pressure-gradient threshold in the usual sense. Instead, the pedestal broadens and grows in height while the unstable domain broadens with it; the stability boundary effectively moves with the pedestal. The best-performing criterion in that dataset is the unusual requirement that $$\Delta_{\mathrm{ped}} \equiv \frac{\Delta_{n_e,\mathrm{ped}}+\Delta_{T_e,\mathrm{ped}}}{2},$$1 of the pedestal region be unstable to $$\Delta_{\mathrm{ped}} \equiv \frac{\Delta_{n_e,\mathrm{ped}}+\Delta_{T_e,\mathrm{ped}}}{2},$$2 ballooning modes, a criterion that tracks the pedestal pressure height within about $$\Delta_{\mathrm{ped}} \equiv \frac{\Delta_{n_e,\mathrm{ped}}+\Delta_{T_e,\mathrm{ped}}}{2},$$3 but is explicitly noted to lack a firm theoretical basis (Dickinson et al., 2011).

A 2026 integrated pedestal-stability workflow generalizes this picture by combining equilibrium scans with ELITE and GATO ideal-MHD boundaries and CGYRO/QLGYRO gyrokinetic transport maps (McClenaghan et al., 24 Jun 2026). In that framework, the standard EPED width scaling,

$$\Delta_{\mathrm{ped}} \equiv \frac{\Delta_{n_e,\mathrm{ped}}+\Delta_{T_e,\mathrm{ped}}}{2},$$4

is retained as a useful organizing principle, but the accessible pedestal is determined by overlap between transport and stability constraints rather than by a single local boundary. The workflow reproduces KBM first- and second-stability structure, shows that low-$$\Delta_{\mathrm{ped}} \equiv \frac{\Delta_{n_e,\mathrm{ped}}+\Delta_{T_e,\mathrm{ped}}}{2},$$5 peeling physics matters in spherical tokamaks, and finds that SPARC-like pedestals lie in an intermediate regime bounded by local KBM second stability on one side and global finite-$$\Delta_{\mathrm{ped}} \equiv \frac{\Delta_{n_e,\mathrm{ped}}+\Delta_{T_e,\mathrm{ped}}}{2},$$6 ballooning instability on the other (McClenaghan et al., 24 Jun 2026). Taken together, these results suggest that ideal pedestal theory remains central, but only when embedded in a coupled local–global stability picture.

3. $$\Delta_{\mathrm{ped}} \equiv \frac{\Delta_{n_e,\mathrm{ped}}+\Delta_{T_e,\mathrm{ped}}}{2},$$7 shear and the persistence of pedestal turbulence

The strongest first-principles statement on shear suppression in the edge pedestal is provided by the gyrokinetic and analytic study of ITG pedestal turbulence (Hatch et al., 2017). There, pedestal modes are described as “not curvature-driven like the core instabilities,” and their suppression by $$\Delta_{\mathrm{ped}} \equiv \frac{\Delta_{n_e,\mathrm{ped}}+\Delta_{T_e,\mathrm{ped}}}{2},$$8 shear is shown to follow a Zhang–Mahajan-type decorrelation theory. The control parameter is

$$\Delta_{\mathrm{ped}} \equiv \frac{\Delta_{n_e,\mathrm{ped}}+\Delta_{T_e,\mathrm{ped}}}{2},$$9

and the central nonlinear closure is

$$\beta_{\theta,\mathrm{ped}} \equiv \frac{2\mu_0 p_{\mathrm{ped}}}{\overline{B}_{\mathrm{pol}}^2}, \qquad p_{\mathrm{ped}} \equiv 2\,p_e(\psi_{\mathrm{ped}}),$$0

with a high-shear asymptote giving approximately $$\beta_{\theta,\mathrm{ped}} \equiv \frac{2\mu_0 p_{\mathrm{ped}}}{\overline{B}_{\mathrm{pol}}^2}, \qquad p_{\mathrm{ped}} \equiv 2\,p_e(\psi_{\mathrm{ped}}),$$1 for fitted $$\beta_{\theta,\mathrm{ped}} \equiv \frac{2\mu_0 p_{\mathrm{ped}}}{\overline{B}_{\mathrm{pol}}^2}, \qquad p_{\mathrm{ped}} \equiv 2\,p_e(\psi_{\mathrm{ped}}),$$2 near unity (Hatch et al., 2017). Gene simulations with local constant shear, global constant shear, global full shear, and self-consistent $$\beta_{\theta,\mathrm{ped}} \equiv \frac{2\mu_0 p_{\mathrm{ped}}}{\overline{B}_{\mathrm{pol}}^2}, \qquad p_{\mathrm{ped}} \equiv 2\,p_e(\psi_{\mathrm{ped}}),$$3 scans follow this decorrelation prediction closely. The result is a quantitatively validated shear-suppression law for ITG pedestal turbulence, especially relevant to low-shear burning-plasma extrapolation.

At the same time, several studies show that the pedestal is not a turbulence-free barrier. An XGC1 total-$$\beta_{\theta,\mathrm{ped}} \equiv \frac{2\mu_0 p_{\mathrm{ped}}}{\overline{B}_{\mathrm{pol}}^2}, \qquad p_{\mathrm{ped}} \equiv 2\,p_e(\psi_{\mathrm{ped}}),$$4 gyrokinetic simulation in DIII-D-like geometry finds that normalized electron density fluctuations are about $$\beta_{\theta,\mathrm{ped}} \equiv \frac{2\mu_0 p_{\mathrm{ped}}}{\overline{B}_{\mathrm{pol}}^2}, \qquad p_{\mathrm{ped}} \equiv 2\,p_e(\psi_{\mathrm{ped}}),$$5 in the upper pedestal, rise sharply near the steep gradient region around $$\beta_{\theta,\mathrm{ped}} \equiv \frac{2\mu_0 p_{\mathrm{ped}}}{\overline{B}_{\mathrm{pol}}^2}, \qquad p_{\mathrm{ped}} \equiv 2\,p_e(\psi_{\mathrm{ped}}),$$6, and reach $$\beta_{\theta,\mathrm{ped}} \equiv \frac{2\mu_0 p_{\mathrm{ped}}}{\overline{B}_{\mathrm{pol}}^2}, \qquad p_{\mathrm{ped}} \equiv 2\,p_e(\psi_{\mathrm{ped}}),$$7 to $$\beta_{\theta,\mathrm{ped}} \equiv \frac{2\mu_0 p_{\mathrm{ped}}}{\overline{B}_{\mathrm{pol}}^2}, \qquad p_{\mathrm{ped}} \equiv 2\,p_e(\psi_{\mathrm{ped}}),$$8 near and beyond the separatrix (Churchill et al., 2017). The skewness changes from slightly negative in the closed-field-line pedestal, consistent with “holes,” to strongly positive in the SOL, consistent with “blobs,” while blob structures are born around the separatrix and propagate outward with radial velocities generally less than $$\beta_{\theta,\mathrm{ped}} \equiv \frac{2\mu_0 p_{\mathrm{ped}}}{\overline{B}_{\mathrm{pol}}^2}, \qquad p_{\mathrm{ped}} \equiv 2\,p_e(\psi_{\mathrm{ped}}),$$9 and strong poloidal motion near $p_e=n_eT_e$0 (Churchill et al., 2017). This directly contradicts the simplified picture of a quiescent pedestal sharply separated from a turbulent SOL.

The JET hybrid H-mode pedestal study reaches a similar conclusion from global nonlinear electromagnetic GENE simulations. In JET #97781, the pedestal top is dominated by ITG modes, the pedestal center and foot are ETG-dominated, and $p_e=n_eT_e$1 shear strongly reduces absolute turbulence levels (Leppin et al., 2024). The same study reports that impurities reduce main-ion transport, and that increased density can push the pedestal center toward a KBM threshold (Leppin et al., 2024). A plausible implication is that pedestal ideality in modern gyrokinetic practice is not the absence of turbulence but the coexistence of strong profile gradients with regulated, mode-selective turbulent transport.

4. Loss of ideality: collapse, small ELMs, collisionality, and impurity-modified transport

When the pedestal is driven beyond ideal stability thresholds, the nonlinear outcome is not always extreme stochastic transport. In a BOUT++ reduced-MHD edge-pedestal-collapse simulation initialized well above the ideal ballooning threshold, with $p_e=n_eT_e$2 against a critical $p_e=n_eT_e$3, the magnetic Kubo number remains in the range $p_e=n_eT_e$4 during the crash and never exceeds unity (Kim et al., 2018). Even though island overlap and stochastic field lines appear, the paper concludes that quasilinear Gaussian diffusion is sufficient and that percolation theory is not necessary. This is a precise statement about a pedestal in the ideal-unstable regime: ideal ballooning collapse need not imply strongly supercritical magnetic transport.

The EAST ELM-dynamics study shows a complementary route in which pedestal density profile control weakens the dominant instability branch (Li et al., 2022). As separatrix density increases and the pedestal density gradient flattens, the most unstable mode changes from high-$p_e=n_eT_e$5 ideal ballooning to intermediate-$p_e=n_eT_e$6 peeling–ballooning, and then to a pedestal that is peeling–ballooning stable but still susceptible to a local foot instability. In the weak-gradient case, the study identifies a small-ELM collisionality window $p_e=n_eT_e$7, while at very low collisionality near $p_e=n_eT_e$8 the system returns to stronger instability and larger ELMs (Li et al., 2022). Small ELMs are defined there by an energy loss fraction below $p_e=n_eT_e$9 of pedestal stored energy.

On Alcator C-Mod, the regulation is collisional rather than directly ELM-dynamical. Below a critical net power crossing the separatrix,

$\overline{B}_{\mathrm{pol}}=\mu_0 I_p/l$0

the pedestal becomes colder, denser at the separatrix, and more collisional; $\overline{B}_{\mathrm{pol}}=\mu_0 I_p/l$1 saturates near $\overline{B}_{\mathrm{pol}}=\mu_0 I_p/l$2 despite continued increase in ionization, while the effective diffusivity $\overline{B}_{\mathrm{pol}}=\mu_0 I_p/l$3 rises from roughly $\overline{B}_{\mathrm{pol}}=\mu_0 I_p/l$4 to roughly $\overline{B}_{\mathrm{pol}}=\mu_0 I_p/l$5 (Miller et al., 2024). The transport increase correlates better with separatrix collisionality $\overline{B}_{\mathrm{pol}}=\mu_0 I_p/l$6 than with pedestal-top collisionality, and SOLPS-ITER modeling suggests even larger growth in $\overline{B}_{\mathrm{pol}}=\mu_0 I_p/l$7 at the same threshold (Miller et al., 2024). This is a transport-limited pedestal rather than a source-limited one.

A further modification of pedestal stability is reported in DIII-D under boron powder injection (Banerjee et al., 1 Jun 2026). At $\overline{B}_{\mathrm{pol}}=\mu_0 I_p/l$8 B, ELM frequency is reduced by $\overline{B}_{\mathrm{pol}}=\mu_0 I_p/l$9; at $n_{\text{e,ped}}$0, long ELM-free periods of $n_{\text{e,ped}}$1 are achieved; and at $n_{\text{e,ped}}$2, confinement degrades and an H–L back transition occurs (Banerjee et al., 1 Jun 2026). ELITE analysis shows pronounced decoupling of peeling and ballooning boundaries at moderate B injection levels, while fluctuation diagnostics identify an impurity-driven low-frequency turbulence channel that increases inter-ELM particle transport (Banerjee et al., 1 Jun 2026). The paper explicitly links this transport channel to a feedback loop between turbulence, particle transport, and pedestal conditions. This suggests that “ideal” pedestal stability boundaries can themselves be reshaped by transport pathways that are not present in purely ideal-MHD formulations.

5. From ideal scalings to integrated and data-driven pedestal prediction

Recent predictive work replaces one-parameter pedestal ideality with hybrid workflows that combine stability theory, gyrokinetics, and statistical surrogates. In spherical tokamaks, HIPED finds that a simple power-law relation between pedestal width and $n_{\text{e,ped}}$3,

$n_{\text{e,ped}}$4

has very low accuracy, with $n_{\text{e,ped}}$5 in each of the H-mode all, pre-ELM, and ELM-free filter categories (Parisi et al., 28 Apr 2025). For pedestal height, simple regression against $n_{\text{e,ped}}$6 gives $n_{\text{e,ped}}$7, adding elongation $n_{\text{e,ped}}$8 gives $n_{\text{e,ped}}$9, and including more features saturates around $n_{\text{e,ped}}$0 (Parisi et al., 28 Apr 2025). Random Forest models do substantially better, reaching $n_{\text{e,ped}}$1–0.76 for $n_{\text{e,ped}}$2, while a control-room-only model still achieves $n_{\text{e,ped}}$3 (Parisi et al., 28 Apr 2025). The dominant predictors are $n_{\text{e,ped}}$4, $n_{\text{e,ped}}$5, $n_{\text{e,ped}}$6, and $n_{\text{e,ped}}$7. This is a direct rejection of the idea that pedestal width or height in spherical tokamaks is governed by a single EPED-like width–height law.

For EPED-like density closure, supervised learning on the JET pedestal database yields a comparable shift away from reduced-form ideal scalings. Using the same reduced input sets as traditional log-linear models, decision tree ensembles and deep learning improve predictive quality for $n_{\text{e,ped}}$8 by about $n_{\text{e,ped}}$9 relative to log-linear regression, measured by RMSE (Kit et al., 2022). Expanding to the full numerical engineering feature set gives about $20\%$0 further improvement, while adding $20\%$1 and $20\%$2 gives only a few percent more (Kit et al., 2022). XGBoost is the best overall performer, reaching RMSE around $20\%$3 on the ECBZ input space, compared with baseline log-linear RMSE values about $20\%$4 for Urano and $20\%$5 for Frassinetti (Kit et al., 2022). The result is a practical replacement for weak density scalings in EPED-like modeling.

A reduced kinetic-stability surrogate has also been validated for NSTX pedestal conditions. The GFS model, tuned by Bayesian optimization against a CGYRO database of 864 converged cases—706 KBM, 89 MTM, and 69 TEM—achieves growth-rate RMS error $20\%$6, frequency RMS error $20\%$7, and mode mismatch fraction $20\%$8 at the optimized pedestal setting

$20\%$9

(Yang et al., 16 Sep 2025). Accuracy degrades at low magnetic shear $n=\infty$0 and near the separatrix $n=\infty$1, but optimized GFS still outperforms optimized TGLF for this pedestal regime (Yang et al., 16 Sep 2025). In predictive practice, this places a fast kinetic constraint alongside ideal-MHD constraints.

Framework	Primary target	Reported outcome
HIPED (Parisi et al., 28 Apr 2025)	Pedestal height and width in MAST-U	RF gives $n=\infty$2–0.76 for $n=\infty$3
Supervised density closure (Kit et al., 2022)	$n=\infty$4 for EPED-like models	ML improves RMSE by about $n=\infty$5 over log-linear scaling
GFS validation (Yang et al., 16 Sep 2025)	Fast linear gyrokinetic pedestal stability	Growth-rate RMS error $n=\infty$6 vs CGYRO
Integrated ELITE/GATO + CGYRO/QLGYRO (McClenaghan et al., 24 Jun 2026)	Overlap of transport and stability windows	Pedestal predicted from intersection of gyrokinetic and ideal-MHD bounds

These developments indicate that contemporary “pedestal ideal” modeling is no longer a single ideal-MHD threshold calculation. It is an integrated enterprise in which ideal boundaries, kinetic mode structure, and statistical surrogates are coupled explicitly.

6. Distinct non-fusion usage: pedestal structures in silicon photonics

Outside plasma physics, pedestal denotes a structural geometry rather than an edge transport barrier. An air-cladding silicon pedestal waveguide is fabricated by partially removing the buried oxide beneath a silicon core with HF etching, leaving a pedestal-like support with air surrounding most of the waveguide (Zhang et al., 2016). In the reported devices, both the pedestal waveguide and the SOI control used a $n=\infty$7 width, $n=\infty$8 silicon thickness, and $n=\infty$9 length, but only the pedestal device had the BOX undercut (Zhang et al., 2016).

The principal result is a modest but measurable enhancement of third-order optical nonlinearity. Under $n$0 chirped-pulse excitation at $n$1, the pedestal waveguide exhibits stronger self-phase-modulation spectral broadening than the conventional SOI waveguide, and its nonlinear-index coefficient is measured as

$n$2

about $n$3 larger than the SOI reference value $n$4 (Zhang et al., 2016). Because the simulated effective areas are nearly identical,

$n$5

the enhancement is not explained by optical confinement alone (Zhang et al., 2016).

The physical origin is attributed to confined acoustic phonons in the pedestal geometry. Using a Kramers–Kronig relation tied to a Lorentzian Brillouin gain/loss spectrum, together with parameters $n$6, $n$7, and $n$8, the authors estimate an optomechanical contribution $n$9 (Zhang et al., 2016). An independent estimate based on electrostriction and radiation pressure reaches the same value. In that literature, the pedestal structure is “ideal” only in the narrow sense that it preserves the optical nonlinearity of silicon while adding an acoustic-phonon-assisted nonlinear enhancement; it is unrelated to ideal-MHD pedestal theory (Zhang et al., 2016).