Aretakis Instability in Extremal Black Holes

• Aretakis instability is a phenomenon where, in extremal black holes, the field decays but a key transverse derivative remains non-decaying, leading to polynomial growth over time.
• It arises due to a vanishing surface gravity and loss of the redshift effect, which result in conserved charges along the degenerate horizon that obstruct full decay.
• The instability extends to various settings such as extremal Kerr, Kerr–Newman, and charged scalar fields, with late-time self-similar scaling and derivative blow-up characterizing its dynamics.

Aretakis instability is the horizon instability of linear perturbations on extremal black holes: the field itself may decay on and outside the event horizon, yet a translation-invariant derivative transverse to the horizon generically fails to decay, and sufficiently high transverse derivatives grow polynomially in advanced time. In the modern formulation, the mechanism is tied to vanishing surface gravity, loss of the redshift effect, and conserved quantities intrinsic to degenerate horizons (Aretakis, 2012). Subsequent work has shown that this phenomenon is not confined to extremal Reissner–Nordström, but extends—sometimes in sharpened or modified form—to extremal Kerr, Kerr–Newman, charged scalars, higher dimensions, black branes, and certain nonlinear near-horizon constructions (Lucietti et al., 2012).

1. Extremality, degenerate horizons, and loss of redshift

The geometric prerequisite for Aretakis instability is extremality. In the axisymmetric framework of extremal horizons, the horizon generator $V$ satisfies

$\nabla_V V=0 \qquad \text{on } H,$

which is the vanishing of the surface gravity on the horizon (Aretakis, 2012). In general static, spherically symmetric extremal black hole spacetimes, one may write the near-horizon metric in Gaussian null coordinates $(v,\rho,\theta^A)$ with horizon at $\rho=0$ as

$$ds^2=-\rho^2\bigl(\lambda_0+\delta\lambda(\rho)\bigr)\,dv^2+2\,dv\,d\rho+\bigl(\gamma_0+\delta\gamma(\rho)\bigr)\,d\Omega^2_{n-2},$$

so that the near-horizon scaling limit is $AdS_2\times S^{n-2}$ (Katagiri et al., 2021).

For extremal Kerr–Newman in the slowly rotating or strongly charged regime, extremality is the double-root condition

$a^2+e^2=M^2,\qquad \Delta=r^2-2Mr+a^2+e^2=(r-M)^2,$

with event horizon

$H=\{r=M\},$

and vanishing surface gravity (Fang et al., 29 Jun 2026). In this setting the horizon is naturally treated as a $b$-boundary with defining function $\rho_H=r-M$, and the inverse metric in ingoing Eddington–Finkelstein coordinates takes the $\nabla_V V=0 \qquad \text{on } H,$0-form

$\nabla_V V=0 \qquad \text{on } H,$1

This formulation makes explicit the degeneracy of the extremal horizon (Fang et al., 29 Jun 2026).

The contrast with the subextremal case is structural. In subextremal black holes, positive surface gravity yields a redshift estimate near the horizon; in the extremal case the redshift disappears completely (Fang et al., 29 Jun 2026). This disappearance is the common geometric substrate of the conservation laws and derivative hierarchies that define Aretakis instability.

2. Horizon conservation laws and the original instability mechanism

The foundational mechanism is a conserved quantity on the extremal horizon. In the general axisymmetric setting, there exist smooth bounded functions $\nabla_V V=0 \qquad \text{on } H,$2, invariant under the horizon generator $\nabla_V V=0 \qquad \text{on } H,$3, such that for solutions of $\nabla_V V=0 \qquad \text{on } H,$4,

$\nabla_V V=0 \qquad \text{on } H,$5

is conserved along the horizon (Aretakis, 2012). This conservation law is local to the extremal horizon and does not depend on global asymptotic structure (Aretakis, 2012).

For extremal Reissner–Nordström, the classical $\nabla_V V=0 \qquad \text{on } H,$6 Aretakis constant is

$\nabla_V V=0 \qquad \text{on } H,$7

which is independent of $\nabla_V V=0 \qquad \text{on } H,$8 (Lucietti et al., 2012). For generic initial data, $\nabla_V V=0 \qquad \text{on } H,$9. Since $(v,\rho,\theta^A)$0 decays on the horizon, one obtains

$(v,\rho,\theta^A)$1

so the first transverse derivative does not decay, while

$(v,\rho,\theta^A)$2

More generally,

$(v,\rho,\theta^A)$3

for large $(v,\rho,\theta^A)$4 (Lucietti et al., 2012). For higher spherical harmonics $(v,\rho,\theta^A)$5, the conserved quantity is

$(v,\rho,\theta^A)$6

and the instability shifts to higher derivative order: $(v,\rho,\theta^A)$7 generically does not decay and $(v,\rho,\theta^A)$8 (Lucietti et al., 2012).

Extremal Kerr admits analogous horizon charges. In the axisymmetric analysis, the explicit $(v,\rho,\theta^A)$9 conserved quantity is

$\rho=0$0

and for generic solutions one has

$\rho=0$1

(Aretakis, 2012). For extremal Kerr–Newman, the horizon charge becomes

$\rho=0$2

and its conservation implies non-decay of the first transversal derivative for generic data with $\rho=0$3 (Fang et al., 29 Jun 2026).

The common logic is fixed: the conserved horizon charge obstructs full decay on the degenerate horizon. If the field and tangential derivatives decay, then a transverse derivative must remain nonzero asymptotically; differentiating the field equation once more then forces blow-up at the next transverse order.

3. Late-time tails, scaling exponents, and derivative blow-up

The conservation-law picture admits a complementary asymptotic formulation in terms of late-time tails. For general static, spherically symmetric extremal black holes in arbitrary dimension, a massive Klein–Gordon mode $\rho=0$4 is governed near the horizon by an effective $\rho=0$5 equation

$\rho=0$6

with conformal weight

$\rho=0$7

For special masses satisfying $\rho=0$8, there is an exact Aretakis constant $\rho=0$9; for generic masses there need not be one, but the near-horizon tail still has the universal form

$$ds^2=-\rho^2\bigl(\lambda_0+\delta\lambda(\rho)\bigr)\,dv^2+2\,dv\,d\rho+\bigl(\gamma_0+\delta\gamma(\rho)\bigr)\,d\Omega^2_{n-2},$$0

and the horizon derivatives satisfy

$$ds^2=-\rho^2\bigl(\lambda_0+\delta\lambda(\rho)\bigr)\,dv^2+2\,dv\,d\rho+\bigl(\gamma_0+\delta\gamma(\rho)\bigr)\,d\Omega^2_{n-2},$$1

Thus blow-up begins when $$ds^2=-\rho^2\bigl(\lambda_0+\delta\lambda(\rho)\bigr)\,dv^2+2\,dv\,d\rho+\bigl(\gamma_0+\delta\gamma(\rho)\bigr)\,d\Omega^2_{n-2},$$2 (Katagiri et al., 2021).

In the special quantized cases with an exact horizon charge, the hierarchy is sharper: $$ds^2=-\rho^2\bigl(\lambda_0+\delta\lambda(\rho)\bigr)\,dv^2+2\,dv\,d\rho+\bigl(\gamma_0+\delta\gamma(\rho)\bigr)\,d\Omega^2_{n-2},$$3 The $$ds^2=-\rho^2\bigl(\lambda_0+\delta\lambda(\rho)\bigr)\,dv^2+2\,dv\,d\rho+\bigl(\gamma_0+\delta\gamma(\rho)\bigr)\,d\Omega^2_{n-2},$$4-st derivative is asymptotically constant and the next derivative grows linearly (Katagiri et al., 2021).

For extremal Kerr, a symmetry-based reformulation identifies the instability with asymptotic self-similarity. Writing

$$ds^2=-\rho^2\bigl(\lambda_0+\delta\lambda(\rho)\bigr)\,dv^2+2\,dv\,d\rho+\bigl(\gamma_0+\delta\gamma(\rho)\bigr)\,d\Omega^2_{n-2},$$5

one obtains on the horizon

$$ds^2=-\rho^2\bigl(\lambda_0+\delta\lambda(\rho)\bigr)\,dv^2+2\,dv\,d\rho+\bigl(\gamma_0+\delta\gamma(\rho)\bigr)\,d\Omega^2_{n-2},$$6

The universal weak scaling weight for generic scalar, electromagnetic, and gravitational perturbations is

$$ds^2=-\rho^2\bigl(\lambda_0+\delta\lambda(\rho)\bigr)\,dv^2+2\,dv\,d\rho+\bigl(\gamma_0+\delta\gamma(\rho)\bigr)\,d\Omega^2_{n-2},$$7

so the field decays like $$ds^2=-\rho^2\bigl(\lambda_0+\delta\lambda(\rho)\bigr)\,dv^2+2\,dv\,d\rho+\bigl(\gamma_0+\delta\gamma(\rho)\bigr)\,d\Omega^2_{n-2},$$8 while each transverse derivative raises the power of $$ds^2=-\rho^2\bigl(\lambda_0+\delta\lambda(\rho)\bigr)\,dv^2+2\,dv\,d\rho+\bigl(\gamma_0+\delta\gamma(\rho)\bigr)\,d\Omega^2_{n-2},$$9 by $AdS_2\times S^{n-2}$0 (Gralla et al., 2017). In this formulation, the Aretakis hierarchy is the direct consequence of an emergent near-horizon, late-time scaling symmetry.

A standard misconception is that Aretakis instability is equivalent to scalar-invariant blow-up. The extremal Kerr scaling analysis states the opposite: despite the growth of transverse derivatives, all generally covariant scalar quantities decay to zero (Gralla et al., 2017). The instability is therefore a statement about particular horizon-transverse components or derivatives, not about generic scalar curvature blow-up at the linear level.

4. Kerr, Kerr–Newman, and non-axisymmetric generalizations

The extremal Kerr problem extends the original scalar instability in two distinct directions: to higher-spin perturbations and to genuinely non-axisymmetric scalar dynamics. For linearized gravitational and electromagnetic perturbations of extreme Kerr, the Teukolsky formalism yields horizon conservation laws for gauge-invariant Newman–Penrose quantities. For $AdS_2\times S^{n-2}$1,

$AdS_2\times S^{n-2}$2

is conserved, and the next derivative obeys a linear-in-$AdS_2\times S^{n-2}$3 growth law. The paper states, in particular, that if $AdS_2\times S^{n-2}$4 decays then a transverse derivative of $AdS_2\times S^{n-2}$5 generically does not decay and certain second transverse derivatives blow up, while for $AdS_2\times S^{n-2}$6 the first non-decay and blow-up occur at higher derivative order (Lucietti et al., 2012).

For scalar waves on extremal Kerr without axisymmetry, the instability is stronger than the classical axisymmetric Aretakis picture. For fixed azimuthal mode $AdS_2\times S^{n-2}$7, $AdS_2\times S^{n-2}$8, the sharp asymptotics on the horizon are

$AdS_2\times S^{n-2}$9

and for every $a^2+e^2=M^2,\qquad \Delta=r^2-2Mr+a^2+e^2=(r-M)^2,$0 there exists a sequence $a^2+e^2=M^2,\qquad \Delta=r^2-2Mr+a^2+e^2=(r-M)^2,$1 such that

$a^2+e^2=M^2,\qquad \Delta=r^2-2Mr+a^2+e^2=(r-M)^2,$2

Thus the first transversal derivative already blows up for non-axisymmetric modes, and the mechanism is not a horizon conservation law but a sharp late-time asymptotic profile governed by global coefficients $a^2+e^2=M^2,\qquad \Delta=r^2-2Mr+a^2+e^2=(r-M)^2,$3 (Gajic, 2023).

The most recent rotating result in the supplied literature is the slowly rotating, strongly charged extremal Kerr–Newman case. There one proves uniform energy boundedness, integrated local energy decay, a horizon $a^2+e^2=M^2,\qquad \Delta=r^2-2Mr+a^2+e^2=(r-M)^2,$4-hierarchy, $a^2+e^2=M^2,\qquad \Delta=r^2-2Mr+a^2+e^2=(r-M)^2,$5-estimates near null infinity, and pointwise decay throughout the exterior without any symmetry assumptions (Fang et al., 29 Jun 2026). The pointwise estimate for $a^2+e^2=M^2,\qquad \Delta=r^2-2Mr+a^2+e^2=(r-M)^2,$6 is

$a^2+e^2=M^2,\qquad \Delta=r^2-2Mr+a^2+e^2=(r-M)^2,$7

with decay only at the weak rate roughly $a^2+e^2=M^2,\qquad \Delta=r^2-2Mr+a^2+e^2=(r-M)^2,$8 (Fang et al., 29 Jun 2026). Against that decay background, the conserved charge $a^2+e^2=M^2,\qquad \Delta=r^2-2Mr+a^2+e^2=(r-M)^2,$9 implies

$H=\{r=M\},$0

for generic data with nonzero horizon charge, and there is a second-order derivative of $H=\{r=M\},$1 which blows up asymptotically along $H=\{r=M\},$2 (Fang et al., 29 Jun 2026). This closes a longstanding gap: it is the first boundedness and pointwise decay result for scalar waves on a rotating extremal black hole without symmetry assumptions, and it derives the corresponding Aretakis instability from a full exterior decay theory (Fang et al., 29 Jun 2026).

5. Frequency-domain, geometric, and causal interpretations

Aretakis instability also admits a frequency-domain interpretation. For extremal Kerr, the synchronous frequency

$H=\{r=M\},$3

is the threshold of superradiance. Exact analysis at extremality shows that these synchronous frequencies are not genuine quasinormal modes or normal modes, but scattering modes. For regular synchronous solutions,

$H=\{r=M\},$4

so sufficiently high radial derivatives diverge on the horizon whenever

$H=\{r=M\},$5

The paper’s conclusion is that Aretakis instability is a horizon derivative instability tied to synchronous scattering states and branch points of the Green function, not a standard quasinormal-mode instability (Richartz et al., 2017).

A closely related causal picture appears in the extremal BTZ black hole. With Dirichlet boundary conditions at the AdS boundary, the explicit retarded Green function obtained by the method of images shows that the field decays as

$H=\{r=M\},$6

but on the horizon as

$H=\{r=M\},$7

The mismatch between on- and off-horizon decay forces transverse derivatives to grow along the horizon (Gralla et al., 2019). The same work solves the null geodesic equation in full generality and shows that the instability is associated with a class of null geodesics that orbit near the event horizon arbitrarily many times before falling in (Gralla et al., 2019). This gives a direct geometrical account of why horizon support persists at late times in the extremal case.

A more invariant geometric interpretation arises in $H=\{r=M\},$8. For masses

$H=\{r=M\},$9

the Aretakis constants are

$b$0

and the next derivative grows linearly: $b$1 In a parallelly propagated null frame, these are precisely components of higher covariant derivatives of the scalar field (Katagiri et al., 2021). Because $b$2 is maximally symmetric, the same structure appears on any null hypersurface; the paper therefore interprets the $b$3 Aretakis phenomenon as singular behavior of higher-order covariant derivatives on the whole null boundary rather than on one preferred horizon (Katagiri et al., 2021). The same work further relates the conserved quantities to conformal Killing tensors and to ladder operators constructed from the spacetime conformal symmetry (Katagiri et al., 2021).

An explicitly conjectural causal analogy appears in the higher-dimensional Bañados–Silk–West literature: the paper on high-energy collisions near extremal black holes proposes that the BSW process should be viewed as a particle-level manifestation of the same near-horizon physics that underlies the linear instability of test fields. The claim there is interpretive rather than a derivation of Aretakis conservation laws (Tsukamoto et al., 2013).

6. Charged fields, higher dimensions, black branes, and nonlinear developments

Charged perturbations alter both the exponents and the observables through which the instability appears. For a charged massless scalar on extremal Reissner–Nordström, the instability is tied to the branch point at the superradiant bound

$b$4

For principal-series modes, the late-time extremal law on the horizon is

$b$5

while near-extremal black holes exhibit the corresponding transient law

$b$6

before crossing over to exponential decay (Zimmerman, 2016).

When the Maxwell field is evolved together with a charged scalar on a fixed extremal Reissner–Nordström background, the instability acquires a new manifestation: horizon charge accumulation. The paper finds that the horizon charge density obeys

$b$7

and for sufficiently large charge coupling,

$b$8

one has $b$9 and therefore

$\rho_H=r-M$0

at late times (Gelles et al., 6 Mar 2025). In the same regime the horizon energy density grows polynomially (Gelles et al., 6 Mar 2025). This extends the classical Aretakis picture from a statement about field derivatives to one about a directly physical, gauge-invariant electromagnetic observable on the horizon.

Higher-dimensional generalizations reveal that the instability is controlled by effective near-horizon scaling dimensions rather than by four-dimensional special features. For non-dilatonic extremal black $\rho_H=r-M$1-branes with near-horizon geometry

$\rho_H=r-M$2

the late-time horizon law is

$\rho_H=r-M$3

where $\rho_H=r-M$4 is the $\rho_H=r-M$5 scaling dimension set by the Kaluza–Klein mass on the Freund–Rubin compact factor (Chen et al., 16 Jul 2025). In this sense, determining the severity of the Aretakis instability reduces to computing a Kaluza–Klein spectrum. Because $\rho_H=r-M$6 with $\rho_H=r-M$7, more transverse derivatives are typically needed before non-decay or blow-up appears, so the instability is weaker than for extremal black holes in the precise sense stated in that work (Chen et al., 16 Jul 2025).

Nonlinear developments remain more selective. A recent construction of dynamical extreme Reissner–Nordström black holes gives a closed-form near-horizon description in which the metric tends to static extremal RN while the scalar field displays the linear Aretakis instability ad infinitum in the nonlinear theory (Porfyriadis et al., 29 Dec 2025). In that construction, the late-time horizon expansions remain of Aretakis type: $\rho_H=r-M$8 with the Aretakis constant $\rho_H=r-M$9 (Porfyriadis et al., 29 Dec 2025). By contrast, the small-data semilinear theory on extremal Reissner–Nordström can be proved using weaker estimates that are compatible with, but do not imply, the non-decay and growth hierarchy expected from the Aretakis instability (Angelopoulos et al., 31 Jan 2025). This suggests that nonlinear global existence and Aretakis-type horizon growth need not be analytically incompatible.

A useful corrective to overextension appears in scalarization studies: a large static tachyonic scalar cloud near an extremal horizon is not, by itself, an Aretakis instability. In the qOS-extremal black hole analysis, the tachyonic near-horizon cloud for $\nabla_V V=0 \qquad \text{on } H,$00 is interpreted as the onset of scalarization, whereas genuine Aretakis instability is reserved for a propagating scalar with standard positive mass and polynomial-in-$\nabla_V V=0 \qquad \text{on } H,$01 derivative growth on the horizon (Myung, 28 Sep 2025). This distinction is conceptually important: horizon-enhanced behavior is broader than the Aretakis mechanism.

Taken together, these developments support a precise synthesis. Aretakis instability is a universal horizon effect of extremality, governed by conserved quantities or by near-horizon scaling data, localized in transverse derivatives rather than generic scalar observables, and modulated by field content, symmetry class, and near-horizon representation theory. Its modern formulations range from exact conservation laws and late-time tails to branch-point singularities, self-similar critical exponents, and $\nabla_V V=0 \qquad \text{on } H,$02- or $\nabla_V V=0 \qquad \text{on } H,$03-controlled effective theories, but the defining feature remains unchanged: the degenerate horizon remembers perturbations in derivatives that the non-extremal redshift would otherwise dissipate.