Approach to Planckian Curvature

In the specified model of an isolated, neutral, four-dimensional Schwarzschild black hole with adiabatic Hawking mass loss and a small ordinary-species luminosity coefficient, the dimensionless horizon-curvature parameter becomes order one before the temperature and adiabaticity parameters do. All three scale as inverse powers of mass, but their coefficients differ. This ordering is model dependent and does not determine the microscopic endpoint.

Helpful background. Quantum Fields in Curved Spacetime supplies the fixed-background flux; Einstein Equations supplies curvature sourcing; Black-Hole Geometry and Mechanics supplies Schwarzschild invariants; Semiclassical Gravity supplies backreaction and validity tests; Black-Hole Information separates fine-grained recovery from mass loss; and Quantum Gravity receives the endpoint once no controlled mean metric remains.

A declared evaporation model

Use $c=\hbar=k_{\rm B}=1$ with $G$ explicit and define $\ell_{\rm P}=\sqrt G$ and $M_{\rm P}=G^{-1/2}$. For a Schwarzschild mass $M\gg M_{\rm P}$,

$$r_h=2GM, \qquad T_{\rm H}=\frac{1}{8\pi GM}, \qquad \frac{dM}{dt}=-\frac{\alpha}{G^2M^2}.$$

The positive dimensionless coefficient $\alpha$ contains species and greybody factors and is held constant in this scaling model. The ordering claimed on this page assumes the ordinary-species regime $\alpha\ll1$; at sufficiently large species number, adiabaticity can fail first. The emission formula, greybody dependence, and numerical workflow are reviewed in Auffinger 2022, §§ 3.1–3.4, PDF. The state is Unruh-like, the geometry is assumed quasi-stationary, and the flux is inserted into a one-parameter mean metric. Charge, spin, accretion, environmental temperature, and stochastic recoil are excluded from this calculation.

Integrating the mass law gives

$$M(t)^3=M_0^3-\frac{3\alpha}{G^2}t,$$

so a formal endpoint time exists. The formula is trustworthy only while every control parameter below is small; extrapolating it to $M=0$ would assume what must be tested.

Curvature, temperature, and adiabaticity

The Kretschmann scalar is $K=48G^2M^2/r^6$. At the horizon,

$$K_h=\frac{3}{4G^4M^4}, \qquad \epsilon_K\equiv \ell_{\rm P}^2\sqrt{K_h} =\frac{\sqrt3}{2}\frac{M_{\rm P}^2}{M^2}.$$

The temperature measured in Planck units is

$$\epsilon_T\equiv \ell_{\rm P}T_{\rm H} =\frac{1}{8\pi}\frac{M_{\rm P}}{M}.$$

For the surface gravity $\kappa=1/(4GM)$, a local adiabaticity measure is

$$\epsilon_{\rm ad} \equiv\frac{|\dot\kappa|}{\kappa^2} =4\alpha\frac{M_{\rm P}^2}{M^2}.$$

Thus $\epsilon_K=1$ at $M=(3/4)^{1/4}M_{\rm P}\simeq0.93M_{\rm P}$. The formal thresholds $\epsilon_T=1$ and $\epsilon_{\rm ad}=1$ occur at $M=M_{\rm P}/(8\pi)$ and $M=2\sqrt\alpha\,M_{\rm P}$. For the small Hawking luminosity coefficient of ordinary species, both lie below the curvature threshold, where the calculation is already uncontrolled. In this declared model, horizon curvature is therefore the first of these three quantities to become order one.

What can fail earlier

The ordering does not prove that the mean-field approximation survives until $\epsilon_K=1$. A large number of species changes $\alpha$ and can lower the gravitational effective cutoff. Stress fluctuations can induce appreciable fluctuations in the geometry. A wave packet can reach the local EFT frequency cutoff even when $K_h$ is small; merely having frequency above the curvature scale is not a failure and often improves the short-wavelength approximation. Secular errors can accumulate over the lifetime, and the state may cease to be well approximated by an instantaneous Unruh state.

Charge and spin add small denominators near extremality; a long throat or mass gap can invalidate the thermal continuum before four-dimensional curvature is Planckian. Conversely, a regular effective metric may keep scalar curvature small while requiring an assumed source with uncontrolled fluctuations. The first failed condition must therefore be recomputed from the model’s state, geometry, field content, and cutoff, not inherited from Schwarzschild scaling.

A complementary test smears the stress tensor over a spacetime cell of size $L$ and compares its connected variance with its mean,

$$\epsilon_{\rm fluc}(L) =\frac{\sqrt{\langle T_L^2\rangle-\langle T_L\rangle^2}} {\max\!\left(|\langle T_L\rangle|,L^{-4}\right)}.$$

Here $T_L$ is one specified frame component smeared with a normalized smooth spacetime profile; the reference scale in the denominator is a convention, not a universal validity threshold. In particular, vacuum stress can fluctuate while its mean vanishes. The decisive check propagates the noise kernel through the retarded metric response and compares the resulting variance of a relational curvature, expansion, or timing observable with its background scale and the claimed accuracy. An order-one value of the displayed stress ratio alone neither establishes nor excludes mean-field validity. See Validity of Stochastic and Mean-Field Gravity for that distinction. Large-$N$ counting and correlations affect the result but cannot replace the state and smearing specification.

Frequency hierarchy is independent as well. A mode with local frequency $\omega_{\rm loc}$ tests $\epsilon_\omega=\ell_{\rm P}\omega_{\rm loc}$ and the adiabatic ratio $|\dot\omega_{\rm loc}|/\omega_{\rm loc}^2$. Near a horizon, an unphysical static-frame blueshift can grow without indicating an invariant breakdown, so the observer or freely falling frame must be named. A genuine high-frequency excitation or an ultraviolet-sensitive subtraction, by contrast, can invalidate the effective expansion before scalar curvature grows.

Independent checks and boundary data

Dimensional check. $K_h$ has dimension $L^{-4}$, so $\ell_{\rm P}^2\sqrt K$ is dimensionless. Both $\epsilon_K$ and $\epsilon_{\rm ad}$ scale as $M^{-2}$, while $\epsilon_T$ scales as $M^{-1}$.

Timescale check. The crossing time is $t_{\rm cross}\sim GM$ and the fractional mass-loss time is $t_M=M/|\dot M|=G^2M^3/\alpha$. Their ratio is $t_{\rm cross}/t_M=\alpha/(GM^2)$, the same scaling as $\epsilon_{\rm ad}$.

Classical limit. In the units used here, the controlled large-black-hole limit is $M/M_{\rm P}\to\infty$, which sends all three parameters to zero and separates the evaporation time from the light-crossing time. Equivalently, after restoring $\hbar$, one may hold the classical geometry fixed and take $\hbar\to0$; the Hawking flux and the Planck-scale ratios then vanish. Taking $G\to0$ at fixed physical $M$ is not this limit: because $M_{\rm P}=G^{-1/2}$, it sends $M/M_{\rm P}\to0$ and makes the displayed Planck-unit control parameters grow rather than shrink.

Once $\epsilon_K$, a fluctuation measure, or another declared control parameter reaches order one, the robust output is boundary data: remaining charges, flux history, correlations already computed, and the last controlled causal geometry. A microscopic continuation or singularity resolution belongs to Quantum Gravity; fixed-background Hawking derivations remain in QFT in Curved Spacetime.

Exercise

For $\alpha=10^{-4}$, evaluate $\epsilon_K$, $\epsilon_T$, and $\epsilon_{\rm ad}$ at $M=10M_{\rm P}$. Which is largest there? By how much may the mass fall before each formal order-one threshold is reached?

Solution

The values are $\epsilon_K=(\sqrt3/2)10^{-2}\simeq8.66\times10^{-3}$, $\epsilon_T=1/(80\pi)\simeq3.98\times10^{-3}$, and $\epsilon_{\rm ad}=4\times10^{-6}$. All are small, with curvature largest at this mass. The formal threshold masses are $0.931M_{\rm P}$, $0.0398M_{\rm P}$, and $0.0200M_{\rm P}$, respectively: decreases by factors about $10.7$, $251$, and $500$ from $10M_{\rm P}$. The last two thresholds are extrapolations into a regime where the curvature expansion has already failed. Ratios with different mass powers should not be ranked at one mass and assumed to preserve that ordering everywhere.

References

• Auffinger, J. (2022). “Primordial Black Hole Constraints with Hawking Radiation—A Review.” arXiv:2206.02672. Open PDF.