Impurity-Driven Turbulence: Mechanisms & Effects

• Impurity-driven turbulence is characterized by impurity-induced modifications to density and temperature gradients that alter fluctuation spectra and transport properties in magnetically confined plasmas and quantum fluids.
• Controlled impurity injection can mitigate edge-localized modes and adjust pedestal stability, as shown by diagnostic thresholds, nonlinear simulations, and gyrokinetic analyses.
• The phenomenon extends beyond fusion, with impurity potentials in rotating Bose–Einstein condensates disrupting vortex lattices and triggering turbulence via melting transitions.

Impurity-driven turbulence denotes a class of regimes in which impurity injection, impurity density and temperature gradients, or impurity potentials modify fluctuation spectra, turbulent transport, and stability boundaries. In magnetically confined fusion plasmas, the supplied literature shows that impurities can either enhance or reduce turbulence, depending on profile alignment, concentration, charge state, and the underlying instability branch. In tokamak pedestals, controlled injection of a low-$Z$ impurity can fundamentally modify pedestal transport and stability, enabling access to long ELM-free periods through impurity-driven turbulence (Banerjee et al., 1 Jun 2026). In stellarators, impurities, depending on the sign of their density gradient, can significantly enhance or reduce turbulent ion heat losses (García-Regaña et al., 2023). In rotating Bose–Einstein condensates, impurity potentials disrupt an ordered vortex lattice, induce melting, and produce signatures of turbulence (Boral et al., 2024).

1. Edge-pedestal regulation and ELM suppression in tokamaks

Banerjee et al. report that real-time injection of low-$Z$ boron powder in the DIII-D tokamak raises the edge $Z_{\rm eff}$ and introduces a low-frequency, ion-diamagnetic-direction (IDD) fluctuation in the pedestal with $f < 10\ {\rm kHz}$ and $k_\perp \rho_i \approx 0.01$–$0.03$. These IDD fluctuations intensify inter-ELM particle transport, preventing the pedestal from over-steepening, which in turn lowers the peeling-ballooning (PB) drive and reduces ELM frequency. At moderate B injection levels, pedestal stability analysis reveals a pronounced decoupling of peeling and ballooning stability boundaries, opening a stability channel toward super-high confinement operation; at higher injection rates, long $(\sim 300\ {\rm ms})$ ELM-free periods are achieved (Banerjee et al., 1 Jun 2026).

The PB analysis is carried out with the ELITE code on a grid of normalized pedestal pressure gradient

$$\alpha \equiv -\frac{a\,\partial p/\partial \psi}{p_0}$$

and normalized edge current density

$j_N \equiv \frac{j_{\rm edge}}{I_p/\pi a^2}.$

The locus $\Gamma(\alpha,j_N)=1$, where $Z$0, defines the PB boundary. In the reported equilibria, boron raises $Z$1 and stiffens the total-pressure gradient; the resulting equilibria migrate deeper into the “second-stability” region, and the ballooning limit moves to larger $Z$2 (Banerjee et al., 1 Jun 2026).

A plausible implication is that impurity actuation in the pedestal need not be viewed only as radiative exhaust or dilution. In the DIII-D case, it acts as a transport-regulation mechanism that modifies the pedestal trajectory in stability space and enables ELM suppression without relying on externally applied perturbations.

2. Threshold behavior, diagnostics, and empirical hysteresis

The tokamak pedestal study is unusually explicit about diagnostics and unusually cautious about analytic closure. Experimentally, once the SPRED-measured B-light intensity exceeds $Z$3, the RMS IDD fluctuation level jumps up; as B levels fall below that threshold, the IDD intensity remains high, revealing a clear hysteresis loop. By power balance, the inferred inter-ELM turbulent power losses $Z$4 rise from $Z$5 with no B to $Z$6 at high B, a factor $Z$7 increase. Intermediate-$Z$8 density fluctuations with $Z$9–$Z_{\rm eff}$0 are measured by Doppler backscattering, with RMS level obtained by integrating the Doppler-shifted peak in the backscattered spectrum, while low-$Z_{\rm eff}$1 density fluctuations with $Z_{\rm eff}$2–$Z_{\rm eff}$3 are measured by BES. At the pedestal top, two concurrent modes are identified: a high-frequency electron-diamagnetic-direction (EDD) mode at $Z_{\rm eff}$4 whose amplitude drops with B, and a low-frequency IDD mode $Z_{\rm eff}$5 whose amplitude rises with B and dominates inter-ELM transport. Cross-power and cross-phase $Z_{\rm eff}$6 between adjacent BES channels are shown in figure form; positive $Z_{\rm eff}$7 corresponds to EDD propagation, and negative $Z_{\rm eff}$8 to IDD (Banerjee et al., 1 Jun 2026).

The same work also states what is not provided. No dispersion relation $Z_{\rm eff}$9 or $f < 10\ {\rm kHz}$0 for an “impurity-driven” mode is derived. No kinetic calculation of real-frequency or growth-rate dependencies on $f < 10\ {\rm kHz}$1, gradient scale lengths, or collisionality is given. No analytic formula is provided for a critical impurity fraction $f < 10\ {\rm kHz}$2 or critical $f < 10\ {\rm kHz}$3 or $f < 10\ {\rm kHz}$4. No local diffusivity $f < 10\ {\rm kHz}$5 or thermal conductivity $f < 10\ {\rm kHz}$6 scaling with impurity concentration is calculated or fitted. No semi-analytic or reduced model equations for the feedback loop are provided, and no analytic fit to the spectra or phase is offered (Banerjee et al., 1 Jun 2026).

This empirical status is important. The reported hysteresis establishes a feedback loop—higher B $f < 10\ {\rm kHz}$7 stronger IDD turbulence $f < 10\ {\rm kHz}$8 increased inter-ELM particle exhaust $f < 10\ {\rm kHz}$9 softer gradients $k_\perp \rho_i \approx 0.01$0 reduced PB drive—but the closure remains phenomenological rather than first-principles analytic.

3. Gyrokinetic descriptions in stellarators: reduction, enhancement, and optimal impurity content

In stellarator turbulence studies, impurities enter through multispecies gyrokinetics and quasineutrality. In the electrostatic, collisionless limit, the lowest-order field constraint is

$k_\perp \rho_i \approx 0.01$1

The turbulent ion heat flux is computed in gyro-Bohm units and the ion heat diffusivity is defined as

$k_\perp \rho_i \approx 0.01$2

Within this framework, impurities can significantly enhance or reduce turbulent ion heat losses, depending on the sign of their density gradient. A peaked impurity profile $k_\perp \rho_i \approx 0.01$3 reduces ITG-driven turbulence, whereas a hollow impurity profile $k_\perp \rho_i \approx 0.01$4 enhances it. Nonlinear scans in $k_\perp \rho_i \approx 0.01$5 with $k_\perp \rho_i \approx 0.01$6 show that $k_\perp \rho_i \approx 0.01$7 first decreases with $k_\perp \rho_i \approx 0.01$8 up to $k_\perp \rho_i \approx 0.01$9–$0.03$0, then increases again; the minimum occurs at $0.03$1–$0.03$2, essentially independent of magnetic geometry. In W7-X at $0.03$3, $0.03$4 and $0.03$5 (García-Regaña et al., 2023).

A later analytic treatment solves the ITG dispersion relation in certain limits and writes the relative growth-rate shift as

$0.03$6

with $0.03$7. The three terms are identified as the impurity density-gradient contribution, the impurity temperature-gradient contribution, and dilution. The density-gradient term is stabilizing when impurity and ion density gradients have the same sign and destabilizing when they have opposite sign; the temperature-gradient term is $0.03$8; the dilution term is always stabilizing. Linear gyrokinetic simulations reproduce the predicted scalings, and a scatter plot of $0.03$9 versus $(\sim 300\ {\rm ms})$0 from $(\sim 300\ {\rm ms})$1 nonlinear runs collapses nearly onto a straight line, with a linear fit giving roughly

$(\sim 300\ {\rm ms})$2

(Calvo et al., 22 Oct 2025).

A common misconception is that impurities are intrinsically stabilizing. The stellarator results do not support such a universal statement. They show instead that impurity effects are sign-sensitive: the same impurity species can reduce or enhance turbulence, depending on whether the impurity profile is peaked or hollow.

4. Edge and scrape-off-layer transport: vorticity, holes, blobs, and inward pinch

In the tokamak edge and scrape-off layer, Raj et al. derive an analytical relation between impurity density, vorticity, sources and sinks, and the mass-to-charge ratio. With plasma vorticity

$(\sim 300\ {\rm ms})$3

and mass-to-charge parameter

$(\sim 300\ {\rm ms})$4

the impurity continuity equation can be rewritten as

$(\sim 300\ {\rm ms})$5

which leads formally to

$(\sim 300\ {\rm ms})$6

In the limit of no sources and sinks, this recovers the Hasegawa–Ishiguro result $(\sim 300\ {\rm ms})$7 (Raj et al., 2023).

BOUT++ simulations of interchange plasma turbulence with $(\sim 300\ {\rm ms})$8, Ne, and Ar seeding show that $(\sim 300\ {\rm ms})$9 moves more strongly inward compared to $$\alpha \equiv -\frac{a\,\partial p/\partial \psi}{p_0}$$0 and $$\alpha \equiv -\frac{a\,\partial p/\partial \psi}{p_0}$$1. In the edge region, $$\alpha \equiv -\frac{a\,\partial p/\partial \psi}{p_0}$$2 is most negative; $$\alpha \equiv -\frac{a\,\partial p/\partial \psi}{p_0}$$3 and $$\alpha \equiv -\frac{a\,\partial p/\partial \psi}{p_0}$$4 are also negative but weaker in magnitude. In the SOL, all three first-ion states show small or positive fluxes. The inward transport is directly associated with monopolar density holes in the presence of the electron temperature gradient, whereas outward transport is associated with plasma blobs. Hole-fraction analysis gives $$\alpha \equiv -\frac{a\,\partial p/\partial \psi}{p_0}$$5, $$\alpha \equiv -\frac{a\,\partial p/\partial \psi}{p_0}$$6, and $$\alpha \equiv -\frac{a\,\partial p/\partial \psi}{p_0}$$7 (Raj et al., 2023).

The physical interpretation follows directly from the analytic form $$\alpha \equiv -\frac{a\,\partial p/\partial \psi}{p_0}$$8. For a given positive $$\alpha \equiv -\frac{a\,\partial p/\partial \psi}{p_0}$$9, heavier $j_N \equiv \frac{j_{\rm edge}}{I_p/\pi a^2}.$0 has a larger amplification than $j_N \equiv \frac{j_{\rm edge}}{I_p/\pi a^2}.$1 or $j_N \equiv \frac{j_{\rm edge}}{I_p/\pi a^2}.$2, so it exhibits the strongest inward pinch. Higher charge states have smaller $j_N \equiv \frac{j_{\rm edge}}{I_p/\pi a^2}.$3, weaker sensitivity to $j_N \equiv \frac{j_{\rm edge}}{I_p/\pi a^2}.$4, and largely follow blob-driven outward convection. This suggests that impurity-driven turbulence in the edge/SOL is inseparable from intermittent coherent structures and from charge-state-dependent trapping.

5. Zonal flows, momentum transport, and impurity peaking

Impurities also modify turbulence indirectly by changing the generation of zonal flows, the residual stress that drives intrinsic rotation, and the impurity peaking factor itself. For collisionless trapped-electron-mode (CTEM) turbulence, Guo et al. derive a zonal-flow growth rate in which the maximal normalized value scales as

$j_N \equiv \frac{j_{\rm edge}}{I_p/\pi a^2}.$5

The denominator $j_N \equiv \frac{j_{\rm edge}}{I_p/\pi a^2}.$6 is the total polarization shielding, while the numerator contains the TEM growth rate and group velocity. Fully ionized non-trace light impurities with relatively flat density profile reduce the maximum normalized ZF growth rate; highly ionized trace tungsten slightly reduces it; fully ionized non-trace light impurities with relatively steep density profile can enhance it. For high-temperature helium from D–T reaction, the effect depends on the temperature ratio: $j_N \equiv \frac{j_{\rm edge}}{I_p/\pi a^2}.$7 is enhanced when $j_N \equiv \frac{j_{\rm edge}}{I_p/\pi a^2}.$8 and reduced when $j_N \equiv \frac{j_{\rm edge}}{I_p/\pi a^2}.$9 (Guo et al., 2017).

In a quasi-linear slab ITG analysis of turbulent momentum transport, impurities enter both the effective inertia and the free-energy drive. The combined center-of-mass residual stress is

$\Gamma(\alpha,j_N)=1$0

An impurity profile aligned with that of main ions causes a considerable reduction of the residual stress, whereas wall-peaked impurities enhance it more weakly. Numerically, for core-peaked impurities, the quasilinear ratio $\Gamma(\alpha,j_N)=1$1 falls by up to $\Gamma(\alpha,j_N)=1$2 for reasonable impurity fractions $\Gamma(\alpha,j_N)=1$3–$\Gamma(\alpha,j_N)=1$4 (Ko et al., 2015).

Gyrokinetic studies of a balanced neutral beam injection deuterium discharge from DIII-D further show that impurities alter the scaling of transport on the charge and mass of the main species and can dramatically change the energy transport even in relatively small quantities. At fixed $\Gamma(\alpha,j_N)=1$5, $\Gamma(\alpha,j_N)=1$6 reduces the ITG growth rate by about $\Gamma(\alpha,j_N)=1$7–$\Gamma(\alpha,j_N)=1$8. In nonlinear runs, carbon at $\Gamma(\alpha,j_N)=1$9 $Z$00 reduces the total $Z$01 by roughly $Z$02 in the hydrogen main-ion case and by about $Z$03 in the electron heat flux $Z$04. The same work gives an approximate expression for the impurity peaking factor and shows that a poloidally varying equilibrium electrostatic potential can lead to a strong reduction or sign change of the impurity peaking factor due to the combined effect of the in-out impurity density asymmetry and the $Z$05 drift of impurities; impurity peaking is not significantly affected by impurity self-collisions (Pusztai et al., 2012).

Taken together, these results show that impurity-driven turbulence is not limited to direct modification of linear growth rates. It also reorganizes turbulence regulation through zonal-flow shielding, momentum-flux symmetry breaking, and equilibrium asymmetry effects that alter impurity accumulation.

6. Quantum-fluid analogues: vortex-lattice melting and turbulence in rotating Bose–Einstein condensates

The supplied literature includes a non-fusion analogue in rotating two-dimensional Bose–Einstein condensates. In the dissipative Gross–Pitaevskii framework,

$Z$06

with a square optical lattice and either static random impurity potentials or a dynamic Gaussian obstacle, impurity forcing destroys long-range vortex order and generates turbulence (Boral et al., 2024).

Without impurities, a rotating condensate with $Z$07, $Z$08, $Z$09, and $Z$10 forms a perfect square Abrikosov-like lattice pinned by the optical lattice. For static impurities, the lattice is declared melted when the structure factor

$Z$11

drops below $Z$12. Numerically, no melting occurs for $Z$13 even up to $Z$14; melting occurs for $Z$15 at $Z$16, for $Z$17 at $Z$18, and for $Z$19 at $Z$20. For a rotating Gaussian obstacle, no melting occurs at $Z$21 even up to $Z$22; at $Z$23, disorder begins around $Z$24; at $Z$25, melting sets in for $Z$26; and at $Z$27, melting occurs for $Z$28. In the melted regime, whether driven by static disorder or stirring, the angular-momentum increment obeys

$Z$29

The turbulence signatures include an incompressible kinetic-energy spectrum

$Z$30

and a compressible spectrum with $Z$31, $Z$32, and $Z$33 ranges, together with strong temporal fluctuations in $Z$34 and drag force, and $Z$35 above melting (Boral et al., 2024).

This quantum-fluid example does not share the gyrokinetic or pedestal language of fusion plasmas, but it preserves the same structural idea: impurities destabilize an ordered state, trigger a threshold-like disordering transition, and release energy into a broad cascade. That commonality helps delimit the term “impurity-driven turbulence” as a family of impurity-mediated pathways to turbulent transport rather than a single universal instability.