Extremality Criteria and Bounds

An extremal black hole sits at a boundary: its horizon is degenerate, two roots of a stationary metric may coincide, and an allowed mass–spin–charge inequality may be saturated. These descriptions agree for familiar Kerr–Newman black holes, but they are not definitions that can be exchanged without hypotheses. The geometric criterion is the vanishing of the surface gravity of a specified Killing horizon; parameter and area criteria require additional field equations, symmetry, and global information.

Helpful background. Exact Solutions of Einstein’s Equations supplies the Kerr–Newman families used as benchmarks; Causality and Global Structure supplies event-horizon and extension language; The Einstein Equation supplies the dynamics behind mass–charge–angular-momentum bounds; Black-Hole Perturbation Theory treats the dynamical stability questions deliberately left onward; Black-Hole Geometry and Mechanics supplies Killing horizons, surface gravity, and horizon mechanics.

Four criteria and their logical status

Let $K^a$ be a Killing field normal to a smooth null hypersurface $\mathcal H$. Its surface gravity is defined on $\mathcal H$ by

$$K^b\nabla_bK^a=\kappa K^a.$$

The horizon is degenerate when $\kappa=0$. Multiplying $K^a$ by a positive constant multiplies $\kappa$ by that constant, so the numerical value depends on the normalization inherited from the stationary exterior. Its vanishing does not. This is the robust local criterion whenever a Killing horizon is present. The construction and the near-horizon consequence are stated in Kunduri and Lucietti 2013, §§1 and 2.1, pp. 3 and 11–12.

Three other criteria occur frequently:

• A repeated horizon root. In a particular stationary coordinate family, the radial function locating the horizon has a double zero.
• Saturation of a parameter bound. The parameters reach the boundary between black-hole and horizonless members of that family.
• Saturation of a quasi-local inequality. A horizon area, angular momentum, and charge obey an equality case of an inequality proved under stated stability, topology, matter, and symmetry assumptions.

The first two are coordinate-family statements until related to invariant geometry. The third can characterize extremal horizon data, but only inside the theorem’s hypothesis class. For example, the equality case $A=8\pi G|J|$ selects the extreme Kerr horizon geometry for the axisymmetric stable vacuum setting of Dain and Reiris 2011, Theorem 1 and Eq. (10), pp. 1–2. It is not an inequality for arbitrary nonsymmetric surfaces.

There is also a useful hierarchy of information. Degeneracy is intrinsic to the pair $(\mathcal H,K)$ and can be read in an arbitrarily small neighborhood of the horizon. A repeated root refers to a chosen representation of an exact solution. A mass bound requires a definition of charges at an asymptotic end, whereas a quasi-local area inequality can be meaningful without such an end. Consequently, a spacetime may admit a well-defined degenerate isolated or Killing horizon even when no Kerr–Newman-like parameter polynomial has been identified. Conversely, writing down parameters that saturate a formal algebraic relation does not prove that the associated metric is regular or possesses a black-hole exterior. One must first establish existence and the causal role of the candidate horizon.

Why the criteria coincide for Kerr–Newman

Write the Kerr–Newman radial polynomial for $M\gt0$ using a geometrized charge parameter $q$:

$$\Delta(r)=r^2-2GMr+a^2+q^2, \qquad a=\frac{J}{M}.$$

Real algebraic roots exist precisely when

$$\mathcal D=(GM)^2-a^2-q^2\ge 0, \qquad r_\pm=GM\pm\sqrt{\mathcal D}.$$

In the strictly subextremal non-Schwarzschild sector $0\lt a^2+q^2\lt(GM)^2$, these roots are the outer and inner horizon radii. At $a=q=0$, however, $r_-=0$ is the central curvature singularity rather than an inner horizon; at $a^2+q^2=(GM)^2$, the two roots merge into one degenerate horizon.

For the outer Killing horizon generated by $K=\partial_t+\Omega_H\partial_\phi$, normalized by the asymptotic stationary time, direct evaluation gives

$$\kappa_+=\frac{r_+-r_-}{2(r_+^2+a^2)} =\frac{\sqrt{\mathcal D}}{r_+^2+a^2}.$$

Thus four statements are equivalent within this family: $\kappa_+=0$; $r_+=r_-$; $\Delta$ has a double root; and $a^2+q^2=(GM)^2$. The derivation exposes every extra ingredient: a stationary Einstein–Maxwell solution, its asymptotic normalization, and the explicit relation between parameters and horizon roots. The coincidence is not supplied by the definition $\kappa=0$ alone.

Two checks are immediate. On the positive-spin uncharged Kerr branch, extremality gives $a=GM$, $J=GM^2$, and

$$A_H=4\pi(r_+^2+a^2)=8\pi GJ.$$

In the nonrotating charged limit, $q=GM$ and the single radius is $r_+=GM$. Both agree with the corresponding saturated bounds. The classical mass–angular-momentum–charge relations and their equality cases are reviewed, with their different global and quasi-local hypotheses, in Dain and Gabach Clément 2018, §§3.2, 4.2, and 5.

The Gaussian-null test

The same distinction appears without choosing an exact solution. Choose Gaussian-null coordinates $(v,r,x^A)$ with $K=\partial_v$, horizon $r=0$, and $g_{vr}=1$. A smooth Killing horizon has

$$ds^2=2\,dv\!\left(dr+r h_A(r,x)dx^A +\frac12 r f(r,x)dv\right)+\gamma_{AB}(r,x)dx^A dx^B.$$

With this transverse orientation, $\kappa=-\tfrac12 f(0,x)$. Hence degeneracy removes the linear term in $g_{vv}$ and smoothness permits the invariantly useful form

$$ds^2=2\,dv\!\left(dr+r h_A(r,x)dx^A +\frac12 r^2F(r,x)dv\right)+\gamma_{AB}(r,x)dx^A dx^B.$$

This gives a local diagnostic independent of a Kerr–Newman radial polynomial: $g(K,K)$ has at least a quadratic transverse zero. The converse requires the coordinate to be adapted to the normal Killing field; a double zero in an arbitrary metric component proves nothing.

Solved diagnostic: approaching extreme Reissner–Nordström

Set $a=0$ and $q=GM(1-\delta)$ with $0\lt\delta\ll1$. Determine the leading behavior of the outer-horizon surface gravity as $\delta\to0^+$, and decide whether it vanishes linearly in the distance from the extremal parameter bound.

Solution

The discriminant and outer radius are

$$\sqrt{\mathcal D}=GM\sqrt{2\delta-\delta^2}, \qquad r_+=GM\left(1+\sqrt{2\delta-\delta^2}\right),$$

and therefore

$$\kappa_+=\frac{\sqrt{2\delta-\delta^2}} {GM\left(1+\sqrt{2\delta-\delta^2}\right)^2} =\frac{\sqrt{2\delta}}{GM}+O(\delta).$$

The horizon separation and surface gravity vanish as $\sqrt\delta$, not linearly in the parameter distance from the bound. At $\delta=0$, the root is double and $\kappa_+=0$. This square-root approach is a useful warning against treating a coordinate parameter’s distance from extremality as a universal geometric measure.

Temperature language and boundaries of use

In stationary semiclassical black-hole thermodynamics one writes $T_H=\kappa/(2\pi)$ for a horizon generator with the physically chosen normalization. Calling an extremal horizon “zero temperature” is therefore a consequence in that setting, not the primary geometric definition. Euclidean continuations, quantum states, and limiting procedures can introduce further subtleties; they are outside this geometric comparison.

Nor does $\kappa=0$ by itself assert stability, supersymmetry, a regular global black-hole exterior, or the existence of a family whose parameters can be tuned away from extremality. Those are separate questions. The next page develops the local geometry of degenerate Killing horizons; perturbative consequences belong to Black-Hole Perturbation Theory.

References

• Dain, Sergio, and María Eugenia Gabach Clément. “Geometrical Inequalities Bounding Angular Momentum and Charges in General Relativity.” Living Reviews in Relativity 21, 5 (2018). DOI.
• Dain, Sergio, and Martín Reiris. “Area–Angular-Momentum Inequality for Axisymmetric Black Holes.” Physical Review Letters 107, 051101 (2011). DOI. Open PDF.
• Kunduri, Hari K., and James Lucietti. “Classification of Near-Horizon Geometries of Extremal Black Holes.” Living Reviews in Relativity 16, 8 (2013). DOI.