Nondegenerate and Degenerate Horizon Geometry

Nondegenerate and degenerate Killing horizons differ by whether their normalized generator has nonzero or zero surface gravity. That distinction is invariant under every nonzero constant rescaling of the generator, and it changes the transverse order of its norm, affine parametrization, bifurcation structure, proper-distance throat, and available near-horizon scaling. The extremal member of a metric family is therefore not obtained by harmlessly setting $\kappa$ to zero in every nonextremal construction.

Required background. Killing Fields and Isometry Groups supplies normalized symmetry generators and their flows.

Helpful background. Exact Solutions of Einstein’s Equations supplies the Reissner–Nordström extensions; Causality and Global Structure supplies completeness and extension language; The Einstein Equation supplies the classical solution families; Black-Hole Geometry and Mechanics places degeneracy among invariant horizon properties.

A simple zero versus a double zero

Let $\mathcal H$ be a smooth Killing horizon of $\chi^a$, with its scale fixed by a stated asymptotic or boundary time. In Gaussian null coordinates adapted to $\chi^a=(\partial_v)^a$ and $\mathcal H=\{r=0\}$,

$$ds^2 =2\,dv\left(dr+r h_A\,dx^A+\frac12r f\,dv\right) +\gamma_{AB}\,dx^A dx^B,$$

and the norm-gradient identity gives $f(0,x)=-2\kappa(x)$. On a connected horizon for which the relevant zeroth-law hypotheses make $\kappa$ constant, the two cases are:

• nondegenerate: $\kappa\ne0$ and $\chi^2=-2\kappa r+O(r^2)$, a simple transverse zero;
• degenerate: $\kappa=0$ and $\chi^2=r^2F(r,x)$, at least a double transverse zero.

The degenerate Gaussian null form and its smooth near-horizon scaling are derived in Kunduri and Lucietti 2013, §2.1, Eqs. (5)–(7). The statement is geometric; no particular black-hole family is required.

Affine parameters and fixed points

Along a generator with Killing parameter $v$ and affine parameter $\lambda$,

$$\lambda-\lambda_0 =\frac{C}{\kappa}e^{\kappa v} \quad (\kappa\ne0), \qquad \lambda=Cv+\lambda_0 \quad (\kappa=0).$$

For a future nondegenerate branch with $\kappa\gt0$, the complete range $v\in\mathbb R$ reaches a finite past affine value. A regular extension can place a bifurcation surface there. By contrast, a complete degenerate Killing orbit is affine-complete in both directions and has no finite affine endpoint at which an ordinary bifurcation surface could be attached. The parameter relation follows from Gourgoulhon and Jaramillo 2006, arXiv v2, §2.5, Eqs. (2.21)–(2.26), Open PDF, pp. 21–23.

There is also a local fixed-point argument. On a bifurcation surface $\mathcal B$,

$$\chi^a|_{\mathcal B}=0, \qquad \nabla_a\chi_b|_{\mathcal B}=\kappa\epsilon_{ab}.$$

If $\kappa$ also vanished, both a Killing field and its first derivative would vanish at a point. Those data uniquely determine a Killing field, forcing $\chi^a$ to vanish identically in the connected region. Hence a nontrivial smooth bifurcate Killing horizon is necessarily nondegenerate. Rácz and Wald’s extension theorem correspondingly assumes constant nonzero surface gravity Rácz and Wald 1996, §4, Theorem 4.2, pp. 548–550.

Reissner–Nordström as a discriminating family

Use a geometrized charge parameter $q$ of dimension length; its conversion to a physical electric charge depends on the Maxwell action normalization. The four-dimensional Einstein–Maxwell metric function is

$$f(r)=1-\frac{2GM}{r}+\frac{q^2}{r^2} =\frac{(r-r_+)(r-r_-)}{r^2},$$

with

$$r_\pm=GM\pm\sqrt{(GM)^2-q^2}.$$

For $|q|\lt GM$, the outer horizon is nondegenerate and

$$\kappa_+=\frac12f'(r_+) =\frac{r_+-r_-}{2r_+^2}\gt0.$$

As $|q|\to GM$, the two roots coalesce, $f(r)=(1-GM/r)^2$, and $\kappa_+\to0$. This correctly diagnoses degeneracy, but the limiting spatial geometry is discontinuous near the horizon. On a static slice the radial proper distance behaves as

$$\ell=\int\frac{dr}{\sqrt{f(r)}}.$$

For a simple root, $f\simeq2\kappa_+(r-r_+)$ and the integral to the horizon is finite. At the double root, $f\simeq(r-GM)^2/(GM)^2$ and $\ell$ diverges logarithmically. The extremal horizon sits at the end of an infinite static throat even though regular ingoing coordinates cross it smoothly.

This family supplies three independent checks: $q\to0$ recovers $\kappa=1/(4GM)$; $r_+\to r_-$ drives $\kappa$ to zero; and the order of the root agrees with the Gaussian null norm expansion.

Different scaling limits

Near a nondegenerate horizon, the leading two-dimensional geometry is Rindler-like because the Killing norm has a simple zero. Its regular extension uses exponentials of $\kappa v$. That construction becomes singular as $\kappa\to0$ because its boost scale disappears.

For a degenerate horizon, write the Gaussian null metric with $g_{vv}=r^2F$. The diffeomorphism

$$v\mapsto\frac{v}{\varepsilon}, \qquad r\mapsto\varepsilon r$$

has a smooth $\varepsilon\to0$ limit, leaving horizon data $(F,h_A,\gamma_{AB})$ on a cross-section. For extremal Reissner–Nordström this produces an $\mathrm{AdS}_2\times S^2$ throat with equal curvature radii $GM$. The limiting procedure and the original asymptotically flat exterior retain different information: the near-horizon limit discards the asymptotic region.

Taking a near-horizon limit and taking an extremal parameter limit therefore need not commute. This is why an extremal geometry should be constructed and checked in a horizon-regular chart, rather than inferred solely by continuity of thermodynamic-looking formulas.

Terminology and limits of the comparison

“Degenerate” is an intrinsic statement about a chosen Killing horizon: $\kappa=0$. “Extremal” often means that charges saturate a bound within a specified solution family. The two coincide in standard Kerr–Newman black holes, but neither word alone supplies stationarity, field equations, asymptotics, compact topology, or an event-horizon interpretation. A degenerate Killing horizon can occur outside the black-hole setting.

Likewise, nondegeneracy is not sufficient for a global bifurcate extension; the extension theorem requires causal and compactness hypotheses. Degeneracy does not make the horizon locally singular; it removes the ordinary boost fixed-point structure and changes the scaling geometry.

Compare the two orders of vanishing

For the static metric $ds^2=-f(r)dt^2+dr^2/f(r)+r^2d\Omega_2^2$, suppose $f(r)=A(r-r_H)^p+\cdots$ with $A\gt0$ outside. Determine the surface gravity and near-horizon proper-distance behavior for $p=1$ and $p=2$.

Solution

For $p=1$, $\kappa=f'(r_H)/2=A/2\ne0$, and $\int dr/\sqrt{f}\sim2\sqrt{(r-r_H)/A}$ is finite: the horizon is nondegenerate. For $p=2$, $f'(r_H)=0$, so $\kappa=0$, while $\int dr/\sqrt{f}\sim A^{-1/2}\ln|r-r_H|$ diverges: the static slice develops an infinite throat. The divergence does not signal a curvature singularity; one must test curvature and use a regular horizon-crossing chart separately.

Conserved Charges, Smarr Relations, and First Laws next uses the normalized generator and horizon data in global and variational charge relations.

References

• Gourgoulhon, Éric, and José Luis Jaramillo. “A 3+1 Perspective on Null Hypersurfaces and Isolated Horizons.” Physics Reports 423 (2006): 159–294. DOI. Open PDF.
• Kunduri, Hari K., and James Lucietti. “Classification of Near-Horizon Geometries of Extremal Black Holes.” Living Reviews in Relativity 16 (2013): 8. DOI.
• Rácz, István, and Robert M. Wald. “Global Extensions of Spacetimes Describing Asymptotic Final States of Black Holes.” Classical and Quantum Gravity 13 (1996): 539–553. DOI. Open PDF.