The centre of the Milky Way in near-infrared light from the Hubble Space Telescope, thousands of stars and glowing gas around the position of Sagittarius A*
Unit 12 · Advanced and Advanced + Gravitational Waves · 70 min

Black Holes and Event Horizons

What happens at the Schwarzschild radius, and how we know black holes are real
Image: NASA, ESA, and the Hubble Heritage Team, via NASA (CC BY 2.0)

Separates the coordinate singularity at the Schwarzschild radius from the physical singularity at the centre, follows a probe across the horizon on its own clock and on the clock of a distant observer, and reads the causal structure of a black hole from Eddington-Finkelstein, Kruskal, and Penrose diagrams. Tidal forces, the photon sphere, and the shadow follow, with Penrose's theorem of 1965 on collapsing stars and the observations that made black holes real: the stars orbiting Sagittarius A*, the images of the Event Horizon Telescope, and the area theorem checked on merging black holes.

Preview: the beginning of the unit

Unit 11 stopped at the Schwarzschild radius. Every orbit, every deflected ray, and every clock of that unit lived at \(r > r_{\mathrm{s}}\), outside the Sun, a neutron star, or anything else whose surface lies well beyond \(r_{\mathrm{s}}\); for the Sun that radius is 2.95 kilometres, deep inside a body of radius 696,000 kilometres, and the question of what happens at \(r = r_{\mathrm{s}}\) never arises. This unit asks that question, and the answer took half a century to understand, from Schwarzschild’s letter of 1915 to a three-page paper by Roger Penrose in 1965, and another half century to confirm by observation: stars have been watched orbiting an invisible mass of four million suns at the centre of the Milky Way, and in 2019 a ring of light was photographed around a dark disc 55 million light-years away.

1 Two singularities, one of them harmless

The Schwarzschild line element of Unit 11, with the signature \((-,+,+,+)\) of Unit 9, reads \begin{equation} \dd s^2 = -\left(1 - \frac{r_{\mathrm{s}}}{r}\right)c^2\,\dd t^2 + \left(1 - \frac{r_{\mathrm{s}}}{r}\right)^{-1}\dd r^2 + r^2\,\dd\theta^2 + r^2\sin^2\theta\,\dd\phi^2 . \label{eq:schw} \end{equation} Here \(r_{\mathrm{s}} = 2GM/c^2\) is the Schwarzschild radius, \(t\) is the time of a clock at rest far from the mass, and \(r\) is the circumference of a circle around the mass divided by \(2\pi\), as Unit 11 defined it. Two things go wrong with the coefficients at \(r = r_{\mathrm{s}}\). The factor in front of \(\dd t^2\) vanishes, and the factor in front of \(\dd r^2\) becomes infinite. Below \(r_{\mathrm{s}}\) both change sign. At \(r = 0\) the factor in front of \(\dd t^2\) is infinite as well. Schwarzschild himself, Einstein, and most physicists for the next forty years believed that the first of these places, \(r = r_{\mathrm{s}}\), was a real edge of the geometry, and the name Schwarzschild singularity was in use until the 1960s. It is the wrong name, and seeing why is the first task of this unit.

1.1 What the coefficients say

Consider first an observer who stays at rest at a fixed \(r\) outside \(r_{\mathrm{s}}\), with rockets firing to hold the position. Unit 11 showed that such a clock runs slow by the factor \(\dd\tau/\dd t = \sqrt{1 - r_{\mathrm{s}}/r}\), and that light sent upwards from there arrives far away with its frequency lowered by the same factor, which is the exact form of the gravitational redshift of Unit 8. In the language of a redshift \(z\), which astronomers define by \(1 + z = f_{\mathrm{emitted}}/f_{\mathrm{received}}\), the frequency ratio becomes a formula worth remembering.

At \(r = 2r_{\mathrm{s}}\) the redshift is \(1 + z = \sqrt{2}\), so a wavelength of 500 nanometres arrives at 707 nanometres, already in the deep red, and at \(r = 1.01\,r_{\mathrm{s}}\) the factor is \(10\). A lamp held ever closer to \(r_{\mathrm{s}}\) fades to infinite wavelength, and this is one meaning of the event horizon: it is the surface \(r = r_{\mathrm{s}}\) from which no light, and nothing slower than light, can reach the outside. A ruler fares no better than the clock. Setting \(\dd t = 0\) in Equation \eqref{eq:schw} shows that the radial distance between \(r\) and \(r + \dd r\), as a ruler held at rest would measure it, is \(\dd r/\sqrt{1 - r_{\mathrm{s}}/r}\), which also grows without limit as \(r\) approaches \(r_{\mathrm{s}}\), so the coordinate \(r\) stops measuring distance there. The rockets tell the same story. The acceleration needed to hover at \(r\) is \(GM/(r^2\sqrt{1 - r_{\mathrm{s}}/r})\), a result quoted here without its short derivation, which is the Newtonian value \(GM/r^2\) divided by the same square root. It grows without limit as \(r\) approaches \(r_{\mathrm{s}}\), so no rocket can hover at the horizon; Section 3 shows that only a light ray can stay at \(r = r_{\mathrm{s}}\), so hovering there would mean moving at the speed of light.