Schwarzschild, Extremal, and Near-Horizon Limits

Kerr links several distinct geometries through controlled limits. Zero spin at fixed mass gives Schwarzschild in the same exterior chart. Extremality merges two radial roots while leaving a smooth, nondegenerate four-metric in horizon-penetrating coordinates; it is the Killing horizon that is degenerate, with a double radial root and zero surface gravity. The near-horizon extreme Kerr geometry appears only after zooming into that double root while simultaneously rescaling time and corotating angle; simply substituting $r=r_+$ destroys the radial direction.

Required background. Frames, Coframes, and Anholonomy supplies regular frames for comparing degenerate coordinate limits.

Helpful background. The Einstein Equation supplies the vacuum and electrovac families; The Riemann Curvature Tensor supports invariant convergence checks; Exact Solutions of Einstein’s Equations supplies controlled-limit discipline; Causality and Global Structure supplies horizon qualifications; Killing Fields and Isometry Groups supplies the corotating generator.

Zero spin and asymptotic limits

On the $M\ne0$ Kerr branch, use $m=GM$ and $a=J/M$. At fixed Boyer–Lindquist $(t,r,\theta,\phi)$ with $r>2m$, the limit $a\to0$ gives

$$\Sigma\to r^2, \qquad \Delta\to r(r-2m),$$

and every $dt\,d\phi$ term vanishes. The metric becomes Schwarzschild in its static chart. For every nonzero subextremal spin, $0<|a|<m$, the second root is an inner horizon and

$$r_-=m-\sqrt{m^2-a^2}=\frac{a^2}{2m}+O(a^4/m^3).$$

It approaches zero as $a\to0$, but it does not survive as a horizon at the endpoint: at $a=0$, $r=0$ is the Schwarzschild curvature singularity. The convergence is therefore not uniform near $r_-$, and the nonzero-spin inner-horizon geometry cannot be carried through the Schwarzschild endpoint by merely taking its coordinate radius to zero.

At fixed $m$ and $a$, $r\to+\infty$ gives Minkowski spacetime plus $O(m/r)$ mass and $O(ma/r)$ gravitomagnetic corrections. Sending $m\to0$ while keeping $a$ finite yields flat spacetime in oblate coordinates away from the coordinate branch disk, as the explicit Cartesian transformation shows in Frolov, Krtouš, and Kubizňák 2017, § 3.6, printed pp. 59–60, Eqs. (3.88)–(3.90), Open PDF. Because $a=J/M$ only on the $M\ne0$ Kerr branch, that limiting path is defined in the metric parameter space and does not hold physical $J$ fixed; a limit statement must specify which parameters are independent.

Extremality is a double-root condition

For Kerr, extremality is $|a|=m$. Take $a=m>0$ for definiteness. Then

$$\Delta=(r-m)^2, \qquad \Omega_H=\frac{1}{2m},$$

and the horizon generator is $\chi=\partial_t+(2m)^{-1}\partial_\phi$. Ingoing coordinates show that $r=m$ remains a smooth degenerate Killing horizon, but its spatial throat becomes infinitely long on an exterior stationary slice. The nonextremal bifurcation surface does not persist as an ordinary bifurcation surface of the extremal spacetime.

The degeneracy is captured by

$$\kappa_H=\frac{r_+-r_-}{2(r_+^2+a^2)}=0$$

at $r_+=r_-=m$. This is a statement about the horizon generator and its double root, not about the rank of the regular four-metric.

Taking $a\to m$ at fixed distance from the horizon gives the extremal Kerr exterior. Taking the same limit while following the shrinking region $r-r_+=O(r_+-r_-)$ can produce a throat geometry. These pointed limits retain different regions and therefore need not be isometric.

The near-horizon extreme Kerr scaling

Start from extremal Kerr and, for the exterior Poincaré-type patch, introduce

$$r=m(1+\lambda R), \qquad t=\frac{2m}{\lambda}T, \qquad \phi=\Phi+\frac{T}{\lambda}.$$

The angular shift is corotation by $\Omega_H$: $\phi-\Omega_Ht=\Phi$ remains finite. Taking $\lambda\to0$ with $(T,R,\theta,\Phi)$ fixed and $R>0$ yields

$$ds^2=2m^2\Gamma(\theta) \left[ -R^2dT^2+\frac{dR^2}{R^2}+d\theta^2 +\Lambda(\theta)^2(d\Phi+R\,dT)^2 \right],$$

where

$$\Gamma(\theta)=\frac{1+\cos^2\theta}{2}, \qquad \Lambda(\theta)=\frac{2\sin\theta}{1+\cos^2\theta}.$$

This near-horizon extreme Kerr, or NHEK, metric is a smooth vacuum solution with an enhanced $SL(2,\mathbb R)\times U(1)$ local isometry group. In the displayed coordinates, $R=0$ is the degenerate horizon boundary of the Poincaré-type patch, not a curvature singularity. Global NHEK is a separate extension obtained with global coordinates and a different coordinate range; it should not be identified with the single $R>0$ patch. The scaling, Poincaré analogy, and global form are derived in Bardeen and Horowitz 1999, § 2, printed pp. 3–5, Eqs. (2.1)–(2.10), Open PDF; an independent derivation is given by Frolov, Krtouš, and Kubizňák 2017, § 3.6, printed pp. 61–62, Eqs. (3.99)–(3.107), Open PDF.

The limiting spacetime is not merely the induced metric on the original horizon. It retains a rescaled radial coordinate $R$, a rescaled time $T$, and the corotating fiber. Setting $r=m$ before scaling makes $g_{rr}=\Sigma/\Delta$ undefined and reduces the dimensional rank; it cannot produce a four-dimensional Lorentzian solution.

A specified near-extremal scaling

To follow the shrinking region between two simple horizons rather than set extremality first, fix $\eta>0$ and take

$$\begin{aligned} a(\lambda)&=m\sqrt{1-\lambda^2\eta^2}, &r&=m(1+\lambda R),\\ t&=\frac{2m}{\lambda}T, &\phi&=\Phi+\frac{T}{\lambda}, \end{aligned} \qquad \lambda\to0.$$

Then

$$r_\pm=m(1\pm\lambda\eta), \qquad \Delta=m^2\lambda^2(R^2-\eta^2), \qquad \kappa_H=\frac{\lambda\eta}{2m}+O(\lambda^2).$$

The outer and inner horizons remain at the finite scaled locations $R=\eta$ and $R=-\eta$. The limiting metric is

$$ds^2=2m^2\Gamma(\theta) \left[ -(R^2-\eta^2)dT^2+\frac{dR^2}{R^2-\eta^2}+d\theta^2 +\Lambda(\theta)^2(d\Phi+R\,dT)^2 \right].$$

Thus “near extremal and near horizon” means a double scaling in which $r_+-r_-=O(\lambda)$ and $r-m=O(\lambda)$. Holding $a<m$ fixed while sending $r\to r_+$ instead produces an ordinary nonextremal near-horizon expansion.

Order-of-limits checks

The zero-spin and extremal conditions are incompatible at fixed nonzero mass: $a\to0$ and $|a|=m$ meet only if $m\to0$, where the NHEK scaling itself loses its curvature scale. Likewise, taking a nonextremal near-horizon expansion at fixed surface gravity gives a Rindler-like leading form, whereas taking extremality first gives the $AdS_2$-like NHEK base. The limits do not commute because the radial root changes from simple to double.

Kerr–Newman has the analogous extremal condition $a^2+q^2=m^2$. Its near-horizon scaling retains both rotation and Maxwell flux; the $a\to0$ extremal endpoint is the $\mathrm{AdS}_2\times S^2$ throat of Reissner–Nordström, not NHEK. The charged throat and its $a=0$ endpoint are displayed in Bardeen and Horowitz 1999, § 4, printed p. 12, Eqs. (4.1)–(4.2), Open PDF. This family comparison is an invariant check on charge and angular limits.

Exercise

Explain algebraically why both the time rescaling and angular shift are necessary in the NHEK limit.

Solution

Near the double root, $\Delta=O(\lambda^2)$, so the norm of the horizon generator is $O(\lambda^2)$. Scaling $t\sim\lambda^{-1}$ keeps its time contribution finite. But $\partial_t$ is not the null generator; $\chi=\partial_t+\Omega_H\partial_\phi$ is. Holding the corotating angle $\Phi=\phi-\Omega_Ht$ fixed removes the divergent rigid rotation and retains the finite fiber term $d\Phi+R\,dT$.

References

• Bardeen, James M., and Gary T. Horowitz. “The Extreme Kerr Throat Geometry: A Vacuum Analog of $\mathrm{AdS}_2\times S^2$.” Physical Review D 60, no. 10 (1999): 104030. DOI: 10.1103/PhysRevD.60.104030. Open PDF.
• Frolov, Valeri P., Pavel Krtouš, and David Kubizňák. “Black Holes, Hidden Symmetries, and Complete Integrability.” Living Reviews in Relativity 20 (2017): 6. DOI: 10.1007/s41114-017-0009-9. Open PDF.