The Schwarzschild Event Horizon

The event horizon is the strangest surface in physics: a place where nothing happens locally and everything changes globally. In the Schwarzschild geometry — the unique static, spherically symmetric vacuum solution, singled out by Birkhoff's theorem — it sits at the areal radius $r=2GM$ . There the standard coordinates misbehave: the metric coefficient $g_{tt}$ vanishes, $g_{rr}$ blows up, and light seems to freeze. Yet an astronaut falling through feels nothing special. Resolving that paradox is the whole art of the horizon. The metric's breakdown is a coordinate artefact, removable by a better chart; the causal breakdown — the fact that no signal from inside can ever reach the outside — is absolute and physical. This article separates the two. It shows that the curvature is finite at $r=2GM$ (the Kretschmann invariant is a plain number there), that regular coordinates like Eddington–Finkelstein cross the surface smoothly, and that what really marks the horizon is the tipping of light cones: at $r=2GM$ the outward light ray stands still, and inside, every future points toward $r=0$ . The same surface wears four faces that coincide only because Schwarzschild is static — a null hypersurface, a Killing horizon where $\partial_t$ goes null, an apparent horizon, and a surface of infinite redshift — from which follow surface gravity and the Hawking temperature.

Key takeaways

The Schwarzschild event horizon is the null hypersurface at $r=2GM$ (units $c=1$ ), where $g_{tt}=-(1-2GM/r)$ vanishes and $g_{rr}$ diverges — but the divergence is a coordinate artefact, not curvature.
The Kretschmann scalar $K=48G^2M^2/r^6$ is finite at $r=2GM$ and infinite only at $r=0$ : the true singularity is at the centre, and the horizon is removable by regular coordinates.
It is a Killing horizon — the surface where the timelike Killing vector $\xi=\partial_t$ goes null, $\xi^\mu\xi_\mu=0$ — with surface gravity $\kappa=1/(4GM)$ setting redshift, freezing, and the Hawking temperature.
The horizon is a one-way membrane: light cones tip until, inside $r<2GM$ , the coordinate $r$ becomes timelike and decreasing, so every worldline is forced to $r=0$ .
An infaller crosses in finite proper time, but a distant observer sees the image redshift and dim as $e^{-t/4GM}$ , frozen forever on the surface.
Key idea: $r=2GM$ is where the outward light ray stands still — the metric singularity is removable, but the one-way causal disconnection is real, invariant, and forever.

The core insight

At $r=2GM$ the timelike Killing vector $\partial_t$ goes null and the outgoing light cone tips to vertical, so the surface becomes a one-way null membrane — the coordinate breakdown is removable, but the causal disconnection is absolute and physical.

Where this lives today

The classical event horizon is now something we image and listen to. The Event Horizon Telescope released the first resolved pictures of horizon-scale structure — M87\ in 2019, Sgr A\ in 2022 — where the dark "shadow" bounded by the photon ring at $\sqrt{27}\,GM/c^2$ has angular size $\sim\sqrt{27}\,GM/(c^2D)$ , encoding the mass and matching the Kerr–Schwarzschild prediction. LIGO–Virgo–KAGRA hear horizons form and settle: after a merger the remnant rings down in quasinormal modes whose frequencies and damping times are fixed by the final horizon's mass, surface gravity, and area, testing the no-hair and area-increase theorems directly — GW150914 onward. And black-hole thermodynamics remains the deepest open frontier: the surface gravity $\kappa=1/(4GM)$ and area $A=16\pi G^2M^2/c^4$ give a temperature and a Bekenstein–Hawking entropy $S=k_Bc^3A/(4G\hbar)$ , and reconciling that entropy and the fate of infalling information with unitary quantum mechanics — the information paradox — drives much of modern quantum gravity.

The physical picture

Picture a river flowing toward a waterfall, faster and faster as it nears the edge. A fish can swim at some top speed; past the point where the current outruns that speed, the fish is swept over no matter how hard it kicks. The event horizon is the surface where the inward flow of space reaches the speed of light: light itself, the fastest swimmer there is, can no longer make headway outward. Nothing about the water is torn or infinite there — the fish crossing it notices only that the far bank has become unreachable. That is the horizon exactly: locally unremarkable, globally a point of no return. The first figure makes this causal tipping precise by drawing the light cones themselves.

Figure

Light cones tip across the Schwarzschild horizon: far out they open at ±45°, at r = 2GM the outgoing ray stands still dt/dr → ∞, and inside every future cone points toward r = 0

Two observers describe a crossing in flatly contradictory terms, and both are right. The infaller sails through $r=2GM$ in finite proper time, noticing nothing but the tidal stretch (small for a large black hole). The distant observer never sees the crossing: the image redshifts and dims exponentially, time constant $\sim 4GM/c^3$ , freezing onto the surface and fading below detection. This is not a contradiction but the relativity of simultaneity pushed to its extreme. The horizon is also, remarkably, a global object — to know exactly where it lies in a collapsing star you must know the entire future, since it is the boundary of what can ever reach infinity. The second figure gives the horizon its physical scale.

Figure

Anatomy and real scale of a Schwarzschild black hole: horizon r = 2GM/c² is 2.95 km for one solar mass, 0.08 AU for Sgr A, and 128 AU for M87 — with the photon sphere at 1.5 rs and the ISCO at 3 rs

The core equations

The stage is the Schwarzschild line element, the unique static spherically symmetric vacuum metric. Everything about the horizon is read off from where its coefficients change character. In geometric units the coordinate horizon sits where the factor $1-2GM/r$ passes through zero.

\mathrm{d}s^2=-\left(1-\frac{2GM}{r}\right)\mathrm{d}t^2+\left(1-\frac{2GM}{r}\right)^{-1}\mathrm{d}r^2+r^2\,\mathrm{d}\Omega^2 \qquad (1)

where: $M$ — mass of the source (units $G=c=1$ send $2GM\to 2M$ ); $r$ — areal radius, so spheres of constant $r$ have area $4\pi r^2$ ; $\mathrm{d}\Omega^2=\mathrm{d}\theta^2+\sin^2\theta\,\mathrm{d}\phi^2$ is the unit two-sphere; $r=2GM$ — the coordinate horizon, where $g_{tt}=0$ and $g_{rr}\to\infty$ . Reading: the $tt$ and $rr$ coefficients are reciprocals, so wherever one vanishes the other diverges — that is the whole coordinate drama. The horizon's invariant meaning comes from the timelike Killing vector $\xi=\partial_t$ that encodes staticity. Its squared norm is just $g_{tt}$ , so tracking that one number tells us when the time-translation symmetry stops being timelike. The surface where it goes null is the Killing horizon.

\xi^\mu\xi_\mu=g_{tt}=-\left(1-\frac{2GM}{r}\right)\ \begin{cases}<0 & r>2GM\ \text{(timelike)}\\ =0 & r=2GM\ \text{(null)}\\ >0 & r<2GM\ \text{(spacelike)}\end{cases} \qquad (2)

where: $\xi=\partial_t$ — generator of time translations; $\xi^\mu\xi_\mu$ — its norm, whose sign flips across $r=2GM$ . Reading: the horizon is the exact locus where symmetry under "waiting" turns into symmetry under a spatial shift — the reason a static observer cannot exist inside. To prove the divergence in (1) is not physical, compute a coordinate-invariant curvature scalar. The Kretschmann scalar is the natural choice because it is quadratic in the full Riemann tensor and cannot hide a real singularity.

K=R_{\mu\nu\rho\sigma}R^{\mu\nu\rho\sigma}=\frac{48G^2M^2}{r^6}\ \xrightarrow{\ r=2GM\ }\ \frac{3}{4G^4M^4} \qquad (3)

where: $K$ — quadratic curvature invariant, identical in every coordinate system; $R_{\mu\nu\rho\sigma}$ — Riemann tensor. Reading: $K$ is a finite number at the horizon and blows up only as $r\to0$ — the geometry is perfectly smooth at $r=2GM$ and genuinely singular only at the centre. The cleanest proof that the horizon is regular is to write a metric that is manifestly smooth there. Trading Schwarzschild time $t$ for the advanced null coordinate $v=t+r_*$ , with tortoise coordinate $r_*=r+2GM\ln|r/2GM-1|$ , gives the ingoing Eddington–Finkelstein form.

\mathrm{d}s^2=-\left(1-\frac{2GM}{r}\right)\mathrm{d}v^2+2\,\mathrm{d}v\,\mathrm{d}r+r^2\,\mathrm{d}\Omega^2 \qquad (4)

where: $v$ — advanced time, constant along ingoing radial light rays; the cross term $2\,\mathrm{d}v\,\mathrm{d}r$ keeps the metric determinant $g=-r^4\sin^2\theta$ finite and nonzero at $r=2GM$ . Reading: no coefficient diverges at the horizon, so ingoing light and infalling matter cross it in this chart without any pathology. The horizon's dynamical fingerprint is its surface gravity — the redshifted acceleration of the Killing flow. It is finite even though a hovering observer's local acceleration diverges, and it fixes both the freezing rate and the Hawking temperature.

\kappa=\frac{GM}{r^2}\bigg|_{r=2GM}=\frac{1}{4GM}=\frac{c^4}{4GM},\qquad T_H=\frac{\hbar\kappa}{2\pi k_B c}=\frac{\hbar c^3}{8\pi G M k_B} \qquad (5)

where: $\kappa$ — surface gravity; $T_H$ — Hawking temperature; area $A=4\pi(2GM)^2$ and entropy $S=k_B A c^3/(4G\hbar)$ follow. Reading: bigger black holes have weaker surface gravity and colder temperature — a solar-mass horizon sits at $62\ \mathrm{nK}$ . Finally, the two observer-dependent facts. Light climbing out from radius $r$ is redshifted by the lapse factor, which vanishes at the horizon; yet the infaller's proper time to reach it is a convergent integral.

\frac{\nu_\infty}{\nu_{\rm emit}}=\sqrt{1-\frac{2GM}{r}}\ \xrightarrow{r\to2GM}\ 0,\qquad \Delta\tau=\int_{2GM}^{r_0}\frac{\mathrm{d}r}{\sqrt{\dfrac{2GM}{r}-\dfrac{2GM}{r_0}}}<\infty \qquad (6)

where: $\nu_{\rm emit}$ , $\nu_\infty$ — emitted and received frequencies; $r_0$ — release radius. Reading: the light freezes and reddens without limit for the distant watcher, but the faller crosses in finite time — the horizon is a boundary in causal structure, not a wall in space.

Derivation AWhy the horizon is a coordinate singularity, not a curvature singularity

The goal is to establish, beyond doubt, that the apparent catastrophe at $r=2GM$ in the Schwarzschild line element is an artefact of a bad chart rather than a genuine feature of the geometry, while the singularity at $r=0$ is real. We do this two ways that must agree: by computing a coordinate-independent curvature invariant and showing it stays finite at the horizon, and by exhibiting an explicit coordinate system in which no metric coefficient misbehaves there. Both are needed — a finite invariant rules out a curvature singularity, and a regular chart rules out any subtler pathology.

Step 1. Notice that in Schwarzschild coordinates $g_{rr}=(1-2GM/r)^{-1}$ diverges at $r=2GM$ while $g_{tt}$ vanishes, and the metric determinant $g=-r^4\sin^2\theta$ passes through the horizon unremarkably. Why: a divergent metric component is suggestive but proves nothing — coordinates can be pathological even where the manifold is smooth, exactly as polar coordinates fail at the North Pole of a perfectly round sphere.

Step 2. Compute the Kretschmann scalar $K=R_{\mu\nu\rho\sigma}R^{\mu\nu\rho\sigma}=48G^2M^2/r^6$ and evaluate it at the horizon: $K(2GM)=48G^2M^2/(2GM)^6=3/(4G^4M^4)$ , a finite number. Why: $K$ is a genuine scalar, identical in every coordinate system, so if it were infinite the geometry would be singular in every chart — its finiteness at $r=2GM$ certifies the curvature there is bounded.

Step 3. Track $K$ as $r\to0$ : it grows without bound like $r^{-6}$ , so the centre is a true curvature singularity where tidal forces diverge and the classical theory ends. Why: this pins the only physical singularity to $r=0$ and confirms the horizon at $r=2GM$ is qualitatively different — smooth geometry, not infinite curvature.

Step 4. Exhibit a regular chart by transforming to ingoing Eddington–Finkelstein coordinates $v=t+r_*$ ; the line element becomes $\mathrm{d}s^2=-(1-2GM/r)\mathrm{d}v^2+2\,\mathrm{d}v\,\mathrm{d}r+r^2\mathrm{d}\Omega^2$ , whose coefficients and inverse are all finite at $r=2GM$ . Why: producing even one coordinate system smooth across the surface is a constructive proof that the earlier divergence was removable, and it is the chart in which infalling geodesics visibly sail through. *Together these confirm the key result: $K(2GM)=3/(4G^4M^4)$ is finite, $K\to\infty$ only as $r\to0$ , and the Eddington–Finkelstein determinant $g=-r^4\sin^2\theta$ remains finite at the horizon — the geometry is smooth there and singular only at the centre.*

Go deeper · Derivation B · 2 minThe one-way membrane from radial light cones

Derivation BThe one-way membrane from radial light cones

The aim is to show that the horizon earns its name as a causal boundary purely from the behaviour of light. We set $\mathrm{d}s^2=0$ for radial rays, watch the light cones narrow and then tip over as $r$ crosses $2GM$ , and conclude that inside the horizon both the ingoing and the nominally outgoing ray move to smaller $r$ , so no future-directed causal curve can ever climb back out. This is a statement about the ordering of events, and unlike the redshift or the coordinate blow-up it survives every change of chart.

Step 1. For a radial null ray, $\mathrm{d}\theta=\mathrm{d}\phi=0$ and $\mathrm{d}s^2=0$ , so $0=-(1-2GM/r)\mathrm{d}t^2+(1-2GM/r)^{-1}\mathrm{d}r^2$ , giving the cone slope $\mathrm{d}t/\mathrm{d}r=\pm(1-2GM/r)^{-1}$ . Why: the light-cone slope is the local speed limit drawn in the coordinate plane — every timelike worldline must stay steeper than these lines, so the cones dictate where anything can go.

Step 2. As $r\to2GM^+$ the slope $|\mathrm{d}t/\mathrm{d}r|\to\infty$ : the cones pinch shut to vertical, so an outgoing ray needs infinite coordinate time $t$ merely to leave the horizon. Why: this is the coordinate face of freezing — in Schwarzschild time the surface is infinitely far in the future, which is why the distant observer never records a crossing.

Step 3. Pass to Eddington–Finkelstein coordinates, where the two radial families read $\mathrm{d}v/\mathrm{d}r=0$ (ingoing) and $\mathrm{d}v/\mathrm{d}r=2(1-2GM/r)^{-1}$ (outgoing); the ingoing rays cross $r=2GM$ smoothly. Why: a good chart shows what really happens to light at the surface — nothing dramatic for the ingoing ray — separating physics from the coordinate stall of Step 2.

Step 4. Inside, where $1-2GM/r<0$ , the "outgoing" family also satisfies $\mathrm{d}r<0$ : both future-directed null directions decrease $r$ , so every timelike worldline (which lies inside the cone) is driven to smaller $r$ and ultimately to $r=0$ . Why: with both edges of the light cone pointing inward, escape is not merely hard but geometrically impossible — the horizon is the boundary of the region that can signal infinity, the definition of an event horizon. *These steps confirm the central result: $\mathrm{d}t/\mathrm{d}r=\pm(1-2GM/r)^{-1}$ follows directly from $\mathrm{d}s^2=0$ , and the sign flip of $1-2GM/r$ for $r<2GM$ forces both null directions inward — the causal one-way character of the horizon is a direct consequence of the light-cone geometry.*

Go deeper · Derivation C · 2 minSurface gravity and the exponential freezing of the infalling image

Derivation CSurface gravity and the exponential freezing of the infalling image

The purpose here is to compute the surface gravity $\kappa=1/(4GM)$ from the acceleration of a static observer redshifted to infinity, and then to connect that single number to the two most vivid features of the horizon: the way a distant watcher sees an infalling beacon freeze and redden without limit, and the Hawking temperature. The subtlety is that the local acceleration needed to hover diverges at the horizon, yet $\kappa$ is finite — the redshift factor that suppresses signals from deep in the well is exactly what tames the divergence.

Step 1. A static observer at radius $r$ must fire engines with proper acceleration $a(r)=(GM/r^2)/\sqrt{1-2GM/r}$ , which diverges as $r\to2GM$ . Why: it takes infinite thrust to hover just above the horizon, the local sign that "standing still" is about to become impossible — but a locally divergent quantity is not yet an invariant of the surface.

Step 2. Define surface gravity as the acceleration redshifted to infinity, $\kappa=\lim_{r\to2GM}\sqrt{1-2GM/r}\;a(r)=GM/r^2$ , and evaluate at the horizon: $\kappa=GM/(2GM)^2=1/(4GM)$ . Why: multiplying by the lapse $\sqrt{1-2GM/r}$ removes the divergence, leaving the force per unit mass as felt at infinity — the quantity that appears in the laws of black-hole mechanics and equals $c^4/(4GM)$ in SI units.

Step 3. Keep two cases apart. A source hovering at fixed $r$ is received at the steady redshift $\nu_\infty=\nu_{\rm emit}\sqrt{1-2GM/r}$ , constant in time — no fade. For a source falling through the horizon the static lapse factor no longer suffices: one must add the emitter's four-velocity and follow the outgoing rays, which peel away from the horizon as $r-2GM\propto e^{\kappa t}$ , so the received frequency and photon rate die as $\nu_\infty\propto e^{-\kappa t}=e^{-t/4GM}$ . Why: the redshift and the coordinate stall combine into an exponential fade with e-folding time $4GM/c^3$ , so the image does not blink out but asymptotically freezes and reddens — the origin of the phrase "frozen star".

Step 4. Identify the same $\kappa$ as the temperature of the horizon, $T_H=\hbar\kappa/(2\pi k_B c)=\hbar c^3/(8\pi GMk_B)$ . Why: the classical surface gravity, a purely geometric quantity, fixes a quantum temperature — the bridge from the event horizon to black-hole thermodynamics and Hawking radiation. *These steps confirm the key numbers: the proper acceleration $a(r)=(GM/r^2)/\sqrt{1-2GM/r}$ redshifts to $\kappa=GM/r^2|_{2GM}=1/(4GM)$ ; the exponential freezing has e-folding time $4GM/c^3\approx20\ \mu\mathrm{s}$ for $1\,M_\odot$ ; and the Hawking temperature is $T_H(1\,M_\odot)\approx62\ \mathrm{nK}$ .*

Limiting cases

Case	Result	Meaning
Far field $r\gg2GM$	$g_{tt}\to-\left(1-\frac{2GM}{r}\right)\to-1$	Geometry relaxes to flat Minkowski; the horizon's pull fades as a Newtonian $1/r$ potential and nothing marks the surface's influence.
Newtonian coincidence	$v_{\rm esc}=\sqrt{\frac{2GM}{r}}=c\Rightarrow r=\frac{2GM}{c^2}$	Setting Newtonian escape speed to $c$ reproduces $r=2GM/c^2$ exactly — a numerical accident (Michell, 1783), not a valid derivation.
Large mass $M\to\infty$	$\kappa=\frac{c^4}{4GM}\to0$ , tides $\sim GM/r^3\|_{2GM}\to0$	Supermassive horizons have weak surface gravity and gentle tides at crossing; an astronaut passes through unharmed and unaware.
Adding charge $Q$	$r_\pm=GM\pm\sqrt{(GM)^2-GQ^2}$	Schwarzschild ( $Q=0$ ) has the single horizon $r=2GM$ ; charge (or, similarly, spin) splits it into inner and outer horizons that merge in the extremal limit $\kappa\to0$ .

Common trapthe misconception

The blow-up of $g_{rr}=(1-2GM/r)^{-1}$ at $r=2GM$ does NOT mark a physical singularity — it is a coordinate artefact, erased by Eddington–Finkelstein or Kruskal coordinates, and the invariant curvature $K=48G^2M^2/r^6$ is perfectly finite there. Two corollaries people get wrong: first, a freely falling observer feels nothing special at crossing (no wall, no flash, no infinite force), because locally the geometry is smooth and the tides are finite — indeed small for a large black hole. Second, "frozen star" describes only the distant observer's redshifted image, not the faller's experience: the infaller crosses in finite proper time and reaches $r=0$ shortly after. The one thing that is absolute and physical is the causal one-way character — that survives every coordinate change and cannot be undone by any engine.

Worked examples

1 · Horizon size, surface gravity, and temperature across the mass range (fully worked)

Take three black holes — one solar mass, Sgr A\ ( $4.15\times10^6\,M_\odot$ ), and M87\ ( $6.5\times10^9\,M_\odot$ ). Find each horizon radius $r_s=2GM/c^2$ , and for the solar-mass case the surface gravity and Hawking temperature.

Step 1. For one solar mass, $M_\odot=1.989\times10^{30}\ \mathrm{kg}$ , so $r_s=2GM/c^2$ . Numerator $2GM=2.655\times10^{20}$ , divide by $c^2=8.988\times10^{16}$ : $r_s=\mathbf{2.95\ km}$ . The horizon of a black hole as massive as the Sun would fit inside a small town.

Step 2. Horizon radius scales linearly with mass, $r_s\propto M$ , so scale up. Sgr A\: $r_s=2.95\ \mathrm{km}\times4.15\times10^6=1.23\times10^7\ \mathrm{km}=\mathbf{0.082\ AU}$ (about $18\ R_\odot$ ). M87\: $r_s=2.95\ \mathrm{km}\times6.5\times10^9=1.92\times10^{10}\ \mathrm{km}=\mathbf{128\ AU}$ — larger than the entire orbit of Neptune.

Step 3. Surface gravity for the solar-mass hole, $\kappa=c^4/(4GM)=(2.998\times10^8)^4/[4(6.674\times10^{-11})(1.989\times10^{30})]=8.08\times10^{33}/5.31\times10^{20}=\mathbf{1.5\times10^{13}\ m/s^2}$ . Enormous locally — yet it scales as $1/M$ , so supermassive horizons are far gentler.

Step 4. Hawking temperature, $T_H=\hbar c^3/(8\pi GMk_B)$ . The solar-mass numbers give $T_H=\mathbf{6.2\times10^{-8}\ K}\approx62\ \mathrm{nK}$ — a hundred million times colder than the CMB, which is why astrophysical Hawking radiation is undetectable, and colder still as $T_H\propto1/M$ . rs = 2.95 km, 0.08 AU, 128 AU; κ = 1.5×10¹³ m/s²; T_H = 62 nK for one solar mass.

2 · How long do you have inside the horizon? (fully worked)

Once past $r=2GM$ , the singularity is unavoidable. The maximum proper time to reach it — achieved by falling freely and NOT firing engines — is $\tau_{\max}=\pi GM/c^3$ . Evaluate it for the three black holes above, and draw the counter-intuitive moral.

Step 1. Inside the horizon $r$ is timelike, so reaching $r=0$ is as inevitable as reaching next Tuesday; the only freedom is how much proper time elapses first. Maximizing the interior geodesic gives $\tau_{\max}=\pi GM/c^3$ (equivalently $\pi M$ in geometric units), the longest survival between crossing $r=2GM$ and hitting $r=0$ .

Step 2. Solar mass: $\tau_{\max}=\pi(6.674\times10^{-11})(1.989\times10^{30})/(2.998\times10^8)^3=\pi(1.327\times10^{20})/(2.694\times10^{25})=\mathbf{1.5\times10^{-5}\ s}\approx15\ \mu s$ . A stellar-mass black hole gives you about fifteen microseconds — no time to notice.

Step 3. Scale by mass, $\tau_{\max}\propto M$ . Sgr A\: $15.5\ \mu\mathrm{s}\times4.15\times10^6=\mathbf{64\ s}$ , roughly a minute. M87\: $15.5\ \mu\mathrm{s}\times6.5\times10^9=1.0\times10^5\ \mathrm{s}=\mathbf{28\ hours}$ — more than a day of life inside a supermassive horizon.

Step 4. The moral is sharper than the folklore. The bound follows from $\mathrm{d}\tau^2\le\mathrm{d}r^2/(2GM/r-1)$ and is saturated only by the radial geodesic with $E\to0$ — the one released from rest just outside the horizon. A generic infaller crosses with $E>0$ and has strictly less time left, and for such an observer a burn that kills the inward radial momentum, steering toward the $E\to0$ trajectory, actually buys proper time; only once you are on the maximizing geodesic does any further thrust shorten $\tau$ . τmax = πGM/c³: 15 μs (1 M⊙), 64 s (Sgr A), 28 h (M87); free fall, not thrust, maximizes it.

3 · The redshift ladder outside the horizon

A beacon emits green light, $\nu_{\rm emit}$ , while hovering at three radii — the ISCO ( $6GM$ ), $4GM$ , and the photon sphere ( $3GM$ ). What fraction of its frequency reaches a distant observer, using $\nu_\infty/\nu_{\rm emit}=\sqrt{1-2GM/r}$ ?

Step 1. At the ISCO, $r=6GM$ : factor $=\sqrt{1-2/6}=\sqrt{2/3}=\mathbf{0.816}$ . Green ( $550$ nm) arrives at $674$ nm — mildly reddened.

Step 2. At $r=4GM$ : factor $=\sqrt{1-1/2}=\sqrt{1/2}=\mathbf{0.707}$ . The same green now arrives at $778$ nm, deep red.

Step 3. At the photon sphere, $r=3GM$ : factor $=\sqrt{1-2/3}=\sqrt{1/3}=\mathbf{0.577}$ ; green shifts to $953$ nm, the near-infrared. As $r\to2GM$ the factor $\to0$ — the beacon reddens out of every band and freezes. Factor 0.816 → 0.707 → 0.577 nearing the horizon, then → 0.

Physics of the portrait

The event horizon was hidden in Schwarzschild's 1916 solution for four decades, mistaken for a true singularity by nearly everyone — until a handful of relativists learned to change coordinates and saw a one-way door instead of a wall.

Origins · 1 minDavid Finkelstein and how this was discovered

David Finkelstein

1929–2016 · the horizon as a one-way causal membrane and the first regular coordinates across it

When Karl Schwarzschild wrote down his vacuum solution in 1916, within weeks of Einstein's field equations and while serving on the Russian front, the surface $r=2GM$ looked like a genuine edge of the world; even Einstein and Eddington regarded it as pathological. Arthur Eddington had already, in 1924, written a coordinate change that quietly tamed it, but no one grasped what it meant. The decisive clarification came from David Finkelstein in 1958: working on unified field theory, he introduced the advanced-time coordinate $v=t+r_*$ and showed that $r=2GM$ is not a barrier but "a perfect unidirectional membrane" — a surface that lets causal influences pass inward but never outward. The same geometry, he made vivid, looks like an impassable freeze from outside and a smooth crossing from within. Kruskal and Szekeres completed the picture in 1960 with the maximal analytic extension, and Roger Penrose supplied the global language of conformal infinity and, in 1965, the singularity theorem showing that once a trapped surface forms the spacetime must be null-geodesically incomplete — a singularity is unavoidable, though the theorem alone does not establish that an event horizon forms. John Wheeler popularized the name "black hole" in 1967. Finkelstein's membrane is the conceptual hinge that turned $r=2GM$ from a mathematical embarrassment into the defining structure of a black hole.

History

1916

— Schwarzschild publishes the static spherically symmetric vacuum solution

1924

— Eddington writes a coordinate form regular at r = 2GM, its meaning unnoticed

1958

— Finkelstein identifies r = 2GM as a one-way causal membrane

1960

— Kruskal and Szekeres give the maximal analytic extension

1965

— Penrose proves the singularity theorem for gravitational collapse

Going further

Texts. Misner, Thorne & Wheeler, Gravitation, §31–32 · Wald, General Relativity, §6.1–6.4 and §12.5 · Carroll, Spacetime and Geometry, §5.6–5.7 · Hartle, Gravity, ch. 12 · Schutz, A First Course in General Relativity, ch. 11. The key results collected here are: Kretschmann scalar $K=48G^2M^2/r^6$ ; surface gravity $\kappa=1/(4GM)$ ; maximum interior proper time $\tau_{\max}=\pi GM/c^3$ . Horizon radii: $r_s(1\,M_\odot)=2.95$ km, Sgr A $=0.082$ AU, M87 $=128$ AU. Surface gravity $\kappa(1\,M_\odot)=1.5\times10^{13}\ \mathrm{m/s^2}$ ; Hawking temperature $T_H(1\,M_\odot)=62$ nK; survival times $\tau_{\max}=15\,\mu$ s ( $1\,M_\odot$ ), $64$ s (Sgr A), $28$ h (M87). Photon sphere at $3GM$ , ISCO at $6GM$ , shadow radius $\sqrt{27}\,GM$ .

Prerequisites. The Schwarzschild Metric · Schwarzschild Radius and the Geometric Mass · Killing Vectors and Conserved Quantities in Schwarzschild · Birkhoff's Theorem. Leads to. Gravitational Redshift in Schwarzschild Spacetime · Photon Orbits and the Photon Sphere · Innermost Stable Circular Orbit.

Ask the tutor: Ask me to walk you through why the light cones tip over at $r=2GM$ , or to work the Hawking temperature of any black hole you name.

Is the event horizon a physical place you could paint or touch?

No. It is a surface in causal structure, not a material object. If you fell toward it you would find nothing there — no wall, no membrane, no flash, no marker of any kind. Locally the spacetime at $r=2GM$ is as smooth as anywhere else, and for a large black hole the tidal forces are gentle. What is special about the surface is entirely non-local: it is the last place from which a light ray can still, in principle, escape to infinity. You only discover you have crossed it by later finding that no signal you send ever gets out. The horizon is defined by the global future of light rays, which is why you cannot feel it as you pass.

If the infaller crosses in finite time, why do we say things "freeze" at the horizon?

Because "freeze" describes the image a distant observer receives, not the faller's experience. As the infaller approaches $r=2GM$ , each successive photon they emit takes longer to climb out and arrives more redshifted, so the distant observer sees the clock tick ever slower and the light redden and dim as $e^{-t/4GM}$ . The image asymptotically freezes onto the surface and fades below detection, but it never quite shows the crossing. Meanwhile the faller's own wristwatch runs normally; they cross in finite proper time and reach the singularity shortly after. Both accounts are correct in their own coordinates — this is the relativity of simultaneity taken to its extreme, not a contradiction.

Why can't a powerful enough rocket escape once inside?

Because inside the horizon "outward" is no longer a direction in space — it is a direction in time, into the past. For $r<2GM$ the coordinate $r$ becomes timelike, so decreasing $r$ is as inevitable as the advance of time is for us outside. Reaching $r=0$ is not a matter of overcoming a force; it is a matter of the future arriving. A rocket firing "outward" is like a rocket trying to fly into yesterday: no amount of thrust helps, and in fact accelerating only shortens the proper time you have left. The light cones themselves have tipped so that every future-directed path leads to smaller $r$ .

What exactly is surface gravity, and why is it finite when hovering takes infinite force?

The proper acceleration needed to hover at radius $r$ is $a(r)=(GM/r^2)/\sqrt{1-2GM/r}$ , which diverges at the horizon — you would need infinite thrust to stand still there. Surface gravity $\kappa$ is that acceleration redshifted to infinity: multiply by the lapse $\sqrt{1-2GM/r}$ and the divergence cancels, leaving $\kappa=GM/r^2=1/(4GM)$ at the horizon. Physically it is the force per unit mass felt at the end of a very long massless rope held taut by a distant assistant. It is finite, decreases as $1/M$ for bigger holes, and sets both the freezing rate and the Hawking temperature.

Is the horizon a local surface or a global one?

Fundamentally global. The true event horizon is the boundary of the region that can send signals to infinity, so deciding whether a given light ray ever escapes requires knowing the entire future of the spacetime. In a collapsing star the horizon can form and grow before the infalling matter reaches it, in anticipation of mass not yet arrived. What makes Schwarzschild special is that, because it is static, this global surface coincides exactly with several locally checkable ones — the Killing horizon $\xi^\mu\xi_\mu=0$ , the apparent horizon, and the infinite-redshift surface — all at $r=2GM$ . In dynamical spacetimes those surfaces separate, and only the global definition tracks the genuine event horizon.

How does the classical horizon connect to Hawking radiation and entropy?

Through the surface gravity and area. Classically the horizon obeys laws that mirror thermodynamics: its area never decreases (like entropy), and its surface gravity $\kappa$ is constant over the surface (like temperature in equilibrium). Hawking's 1974 calculation made the analogy literal — quantum fields near the horizon produce thermal radiation at $T_H=\hbar\kappa/(2\pi k_Bc)$ , and the horizon carries a Bekenstein–Hawking entropy $S=k_Bc^3A/(4G\hbar)$ proportional to its area, not its volume. For astrophysical holes $T_H$ is nanokelvin-scale and unobservable, but conceptually this ties a purely geometric classical surface to quantum information, and the tension between horizon entropy and unitary evolution is the modern information paradox.