Hawking Area Theorem

Hawking’s area theorem states that later cross-sections of a classical future event horizon cannot have smaller area than earlier cross-sections. The proof combines local null focusing with a global fact: an event horizon is an achronal causal boundary whose generators cannot leave it toward the future. An energy condition controls focusing, while completeness or predictability prevents a generator from ending at an exposed singularity before the focusing argument reaches a contradiction.

Required background. The Einstein Equation converts the null energy condition to null convergence.

Helpful background. Exact Solutions of Einstein’s Equations supplies collapse and stationary examples; Causality and Global Structure supplies achronal boundaries and predictability; Flat-Background Hilbert Stress-Energy from Metric Variation supplies the energy condition; and Black-Hole Geometry and Mechanics supplies event-horizon generators and null expansion.

A precise classical statement

Let $\mathcal H^+$ be a future event horizon with compact spacelike cross-sections $S_1$ and $S_2$, where $S_2$ lies to the future of $S_1$ along the generators. Assume sufficient regularity to define their areas and the generator flow almost everywhere. Assume the null convergence condition

$$R_{ab}k^ak^b\geq0$$

for every null $k^a$ on the relevant horizon portion. In Einstein gravity this is equivalent to $T_{ab}k^ak^b\geq0$, independent of the cosmological term. Finally assume a global condition that rules out premature future endpoints: for example, future completeness of the horizon generators, or an appropriate globally hyperbolic conformal completion with regular future infinity.

Under these conditions,

$$\operatorname{Area}(S_2) \geq \operatorname{Area}(S_1).$$

Modern regularity theorems establish this result without pretending that a merger horizon is everywhere smooth. Chruściel, Delay, Galloway, and Howard state three sufficient global hypothesis packages and prove the low- regularity area result in Chruściel, Delay, Galloway, and Howard 2001, §§1 and 3–6, pp. 109–158. The theorem is about future event horizons; it is not automatically a theorem about apparent, trapping, or timelike marginal tubes.

Raychaudhuri excludes negative expansion

On a smooth horizon portion, affinely parametrize a null generator by $k^a=(d/d\lambda)^a$. The twist vanishes because the congruence is normal to a null hypersurface. In four dimensions, Raychaudhuri’s equation is

$$\frac{d\theta}{d\lambda} =-\frac12\theta^2 -\sigma_{ab}\sigma^{ab} -R_{ab}k^ak^b.$$

The shear term is nonnegative, and null convergence therefore gives

$$\frac{d\theta}{d\lambda}\leq-\frac12\theta^2.$$

Suppose $\theta(\lambda_0)=\theta_0\lt0$. Integrating the inequality shows that $\theta$ diverges to negative infinity within affine parameter no larger than $2/|\theta_0|$. The associated transverse area element collapses and a conjugate point forms.

But a null generator containing a conjugate point cannot continue to generate an achronal boundary: beyond the conjugate point, nearby causal curves provide a timelike connection between points on the would-be boundary. An event horizon is the boundary of a causal past, so it is achronal. If the generator is future complete—or global predictability ensures the relevant continuation— it cannot evade the contradiction by simply ending first. Therefore $\theta\geq0$ wherever the smooth expansion is defined.

This is the proof architecture in Hawking’s original treatment Hawking 1972, §3, pp. 156–160. The censorship-like input enters at the endpoint step, not in Raychaudhuri’s local differential equation.

From generator expansion to area

Along a smooth generator,

$$\frac{d}{d\lambda}\ln dA=\theta.$$

Since $\theta\geq0$, the Jacobian carrying an area element from $S_1$ to $S_2$ cannot decrease. Event-horizon generators may join at past endpoints, creases, or caustics; during a merger these new generators add area to later sections. A generator cannot leave the horizon through a future endpoint in the regular exterior. Combining nonnegative Jacobian with possible new generators gives the integrated area inequality.

The nonsmooth case cannot be justified by writing $\theta$ at a crease where it is undefined. Rigorous proofs use the almost-everywhere second-order structure of horizons and measure-theoretic area comparison. The elementary Raychaudhuri argument remains the correct local mechanism, while the regularity theorem supplies the missing global analysis.

Equality is informative. If a smooth portion has $\theta=0$ throughout, then Raychaudhuri and null convergence imply

$$\sigma_{ab}=0, \qquad R_{ab}k^ak^b=0.$$

Thus no focusing matter flux or gravitational shear crosses that equilibrium portion. The converse requires boundary and regularity control; zero net area change is not permission to ignore all local hypotheses.

Where the theorem can fail

Energy condition. Quantum stress–energy can violate the pointwise null energy condition. A semiclassically evaporating horizon can shrink, so the classical theorem cannot be applied by replacing $T_{ab}$ with an unrestricted renormalized expectation value. Generalized entropy belongs to the semiclassical treatment, not to this area proof.

Global endpoint control. If a generator can end at a naked singularity before the predicted conjugate point, negative expansion need not contradict achronality on the available spacetime. This is why a completeness, predictability, or regular conformal-completion hypothesis is substantive.

Wrong horizon. A timelike marginal tube can be crossed in both directions, and an apparent horizon depends on slicing. Its area behavior is governed by different sign and stability conditions. Conversely, the event horizon can expand through a locally vacuum region in anticipation of later infall; the area theorem is global and does not provide an instantaneous local flux law.

Bardeen, Carter, and Hawking formulate the second law and emphasize its classical energy assumptions in Bardeen, Carter, and Hawking 1973, Open PDF, pp. 168–169.

Check your understanding

At one smooth event-horizon point, suppose $\theta_0=-0.04/L$ for a length scale $L$, and assume null convergence. What does Raychaudhuri predict, and which global hypothesis turns that prediction into a contradiction?

Solution

The focusing inequality predicts a conjugate point within affine distance at most $2/|\theta_0|=50L$. A generator cannot remain on an achronal boundary beyond that point. Future completeness, or a suitable predictability and regular-infinity condition ensuring continuation through that affine range, prevents the generator from ending first. With that global input, the assumed negative expansion is impossible.

Next, Physical-Process First Law linearizes the same focusing equation about equilibrium to relate a small horizon flux to its area change.

References

• Bardeen, J. M., B. Carter, and S. W. Hawking. “The Four Laws of Black Hole Mechanics.” Communications in Mathematical Physics 31 (1973): 161–170. DOI.
• Chruściel, Piotr T., Erwann Delay, Gregory J. Galloway, and Ralph Howard. “Regularity of Horizons and the Area Theorem.” Annales Henri Poincaré 2 (2001): 109–178. arXiv:gr-qc/0001003.
• Hawking, Stephen W. “Black Holes in General Relativity.” Communications in Mathematical Physics 25 (1972): 152–166. DOI.