Negative Energy and Horizon Shrinkage

Positive Hawking luminosity at future null infinity lowers the black hole’s Bondi mass. In a quasistationary description, stress-tensor conservation pairs that loss with negative Killing-energy flux into the near-horizon region. The renormalized quantum stress tensor can violate the pointwise null energy condition assumed by the classical area theorem, so a shrinking horizon is no contradiction. One must still distinguish the locally defined trapping horizon from the teleological event horizon.

Helpful background. Quantum Fields in Curved Spacetime supplies state-dependent renormalized stress tensors; Einstein Equations supplies the flux–geometry relation; Black-Hole Geometry and Mechanics supplies expansions and area laws; Semiclassical Gravity supplies mean backreaction; and Black-Hole Dynamics supplies trapping and event horizons.

Conserved flux does not require positive local energy

The semiclassical equation uses a renormalized expectation value,

$$G_{\mu\nu}+\Lambda g_{\mu\nu} +H_{\mu\nu}^{\mathrm{ren}} =8\pi G\, \langle T_{\mu\nu}\rangle_{\mathrm{ren}}.$$

Diffeomorphism invariance enforces $\nabla^\mu\langle T_{\mu\nu}\rangle_{\mathrm{ren}}=0$ when the gravitational counterterms and matter prescription are treated consistently. Conservation relates fluxes across the boundaries of a spacetime region; it does not impose $T_{\mu\nu}k^\mu k^\nu\geq0$ pointwise.

In the Unruh state, outgoing modes carry positive Killing energy to infinity, while the near-horizon renormalized stress contains an ingoing negative- energy component relative to the asymptotic time generator. This is a global balance statement, not a literal assertion that each escaping particle has a localized negative partner. Auffinger summarizes the state, luminosity, and mass-loss interpretation in Auffinger 2022, §§3.1–3.3, PDF.

Vaidya balance and a shrinking trapping horizon

A simple local model that solves Einstein’s equation with a prescribed null source is the ingoing Vaidya metric

$$ds^2 =-\left(1-\frac{2Gm(v)}{r}\right)dv^2 +2\,dv\,dr+r^2d\Omega^2.$$

Its Einstein equation gives

$$T_{vv}=\frac{1}{4\pi r^2}\frac{dm}{dv}.$$

For evaporation $dm/dv\lt0$, so the ingoing null flux has negative energy. To identify the horizon, set $f=1-2Gm/r$ and choose future null normals $\ell=\partial_v+(f/2)\partial_r$ and $n=-\partial_r$, with $\ell\cdot n=-1$. Their expansions are $\theta_{(\ell)}=f/r$ and $\theta_{(n)}=-2/r$. Thus the outer marginal surface is

$$r_{\mathrm H}(v)=2Gm(v), \qquad A_{\mathrm H}(v)=16\pi G^2m(v)^2,$$

and therefore

$$\frac{dA_{\mathrm H}}{dv} =32\pi G^2m\frac{dm}{dv}\lt0.$$

This calculation makes signs and dimensions explicit. It is an effective near-horizon geometry sourced by a declared stress component; it is not a derivation of the Unruh stress tensor from fixed-background modes. Nor does one ingoing Vaidya patch reproduce both the outgoing asymptotic flux and the full stress tensor everywhere.

The instantaneous area–mass relation also satisfies

$$\frac{dm}{dv} =\frac{\kappa}{8\pi G}\frac{dA_{\mathrm H}}{dv}, \qquad \kappa=\frac{1}{4Gm}.$$

Both sides are negative. This is an exact algebraic check within the Vaidya ansatz, using the instantaneous Schwarzschild value of $\kappa$. There is no exact Killing generator when $dm/dv\ne0$; identifying this relation with the physical-process first law for an event horizon requires an additional adiabatic approximation and matched time normalization.

Why the classical area theorem does not apply

For an affinely parametrized null congruence,

$$\frac{d\theta}{d\lambda} =-\frac12\theta^2-\sigma_{\mu\nu}\sigma^{\mu\nu} -8\pi G\,T_{\mu\nu}k^\mu k^\nu$$

in Einstein gravity. The classical future-event-horizon area theorem uses, among other global assumptions, the null energy condition $T_{\mu\nu}k^\mu k^\nu\geq0$. That sign focuses null generators in the manner needed to exclude future area decrease. A renormalized quantum expectation value can be negative, so a central theorem hypothesis is absent. Wald reviews the classical laws and their semiclassical replacement by the generalized second law in Wald 2001, §§2 and 4.1, PDF.

Area decrease therefore does not show a violation of a valid classical theorem. The semiclassical entropy question is instead whether

$$S_{\mathrm{gen}} =\frac{A}{4G_{\mathrm{ren}}} +S_{\mathrm{out}}^{\mathrm{ren}}$$

is nondecreasing under the relevant GSL hypotheses. The geometric term may fall while exterior entropy rises by a larger amount.

Trapping horizon versus event horizon

The Vaidya surface $r=2Gm(v)$ is defined by a local vanishing expansion. An event horizon is the boundary of the causal past of future infinity and depends on the entire future history. In an adiabatic evaporation model the two can be close but not identical; their separation is of derivative order. Claims about causal accessibility or information cannot be inferred from the trapping-horizon formula alone.

The distinction also controls signs in horizon laws. A future outer trapping horizon can become timelike under negative ingoing flux and shrink locally. An event-horizon generator obeys a teleological boundary condition. Replacing one with the other while keeping the same differential equation silently changes the problem.

The causal character can be checked without a diagram. Along the marginal surface $r_{\mathrm H}=2Gm(v)$, the radial induced line element is $ds^2=4G(dm/dv)\,dv^2$. It is negative during evaporation, so this surface is timelike. An outgoing null generator instead obeys $dr/dv=f/2$. If a global event horizon exists and follows a slowly varying branch, writing $r_{\mathrm E}=2Gm+\delta r$ gives $\delta r\simeq8G^2m\,dm/dv\lt0$ at leading derivative order. Its location is slightly inside the marginal surface in this model. Selecting that particular null solution still requires future boundary data.

Regime and failure cases

The Vaidya calculation assumes spherical symmetry, a slowly varying mass, Einstein gravity at the retained order, and a stress tensor dominated by one null component. A physical four-dimensional Unruh stress has angular pressures, vacuum polarization, greybody-dependent flux, and state-dependent near-horizon structure. Hu and Verdaguer discuss why a full backreaction problem must control the renormalized mean stress and its fluctuations Hu and Verdaguer 2008, §8.

The model fails near rapid transients, extremal throats, strong curvature, or large stress fluctuations. It also cannot be used to infer a nonsingular endpoint. Its value is narrower: it demonstrates explicitly how local conservation, negative quantum energy, mass loss, and horizon shrinkage can be mutually consistent.

Exercise

Let $m(v)=M_0-\mu v$ with constant $\mu\gt0$ over a short adiabatic interval. Compute $T_{vv}$ at the trapping horizon, its area-loss rate, and verify the instantaneous area–mass identity above.

Solution

Since $dm/dv=-\mu$ and $r_{\mathrm H}=2Gm$,

$$T_{vv}\big|_{\mathrm H} =-\frac{\mu}{4\pi(2Gm)^2} =-\frac{\mu}{16\pi G^2m^2}\lt0.$$

The area derivative is

$$\frac{dA_{\mathrm H}}{dv} =32\pi G^2m(-\mu) =-32\pi G^2m\mu.$$

Finally,

$$\frac{\kappa}{8\pi G}\frac{dA_{\mathrm H}}{dv} =\frac{1}{4Gm}\frac{1}{8\pi G} \left(-32\pi G^2m\mu\right) =-\mu =\frac{dm}{dv}.$$

The equality checks the normalization and sign. It does not establish that a real Unruh-state stress tensor has exactly the Vaidya form.

References

• Auffinger, J. (2022), “Primordial Black Hole Constraints with Hawking Radiation—A Review,” arXiv:2206.02672v1. Open PDF.
• Hu, B. L., and E. Verdaguer (2008), “Stochastic Gravity: Theory and Applications,” Living Reviews in Relativity 11, 3. DOI.
• Wald, R. M. (2001), “The Thermodynamics of Black Holes,” Living Reviews in Relativity 4, 6. Open PDF.