Supermassive Black Holes

A supermassive black hole is one of the most massive macroscopic objects in the universe and among the most consequential astrophysical entities, exerting a profound influence on the dynamics and evolution of its host galaxy. The no-hair theorem promises that an isolated, uncharged, stationary black hole is described completely by the Kerr metric with just two numbers: its mass $M$ — for the supermassive class, $10^6$ to $10^{10}\,M_\odot$ — and its dimensionless spin $a_*=cJ/GM^2\in[0,1]$ . Nothing about the formation history of a black hole is preserved in its external geometry; once formed, a stationary black hole is characterized solely by its mass, spin, and charge. Supermassive black holes are distinguished less by their local geometry than by their astrophysical environment. Like stellar-mass black holes, they are described by the Kerr geometry when rotating, but they reside at galactic centers and grow through gas accretion, tidal disruption, and mergers.

Almost everything we observe from a black hole comes from the material around it, rather than from the black hole itself. Gas spiralling inward settles into an accretion disk, converting gravitational binding energy into radiation with an efficiency far exceeding that of nuclear fusion. This process powers active galactic nuclei—quasars, Seyfert galaxies, blazars, and radio galaxies—which can outshine their entire host galaxies while originating from a region smaller than the Solar System. The Eddington luminosity sets an upper limit on this radiative output, reached when the outward force exerted by radiation on the infalling gas balances the inward pull of gravity. This limit also determines the characteristic timescale for exponential black-hole growth. Over the past three decades, an even deeper connection has emerged: the mass of a supermassive black hole correlates strongly with the properties of its host galaxy, particularly the stellar velocity dispersion through the $M-\sigma$ relation. This remarkably tight correlation points to a self-regulating feedback between a black hole whose direct influence extends over parsec scales and a galaxy spanning kiloparsecs. This article develops that picture from first principles: the two-parameter spacetime, the accretion engine, the growth timescale, and the coevolution that links black holes to galaxies.

Key takeaways

A supermassive black hole is a two-parameter (mass $M$ , dimensionless spin $a_*$ ) Kerr spacetime of $10^6$ – $10^{10}\,M_\odot$ sitting at a galaxy's dynamical center; its gravitational radius $r_g=GM/c^2\approx1.5\times10^{9}\,(M/10^6 M_\odot)$ m is tiny, yet accretion onto it powers the most luminous persistent sources in the universe.
The Eddington luminosity $L_{\rm Edd}=4\pi GMm_pc/\sigma_T\approx1.3\times10^{31}(M/M_\odot)$ W caps quiescent output and sets an exponential growth clock, the Salpeter $e$ -folding time $\sim5\times10^7$ yr.
Radiatively efficient accretion converts $\eta=1-E_{\rm ISCO}$ of rest mass to light: $5.7\%$ for a non-spinning hole (ISCO at $6r_g$ ), up to $42\%$ for maximal prograde Kerr (ISCO at $r_g$ ).
Black-hole mass tracks host-bulge stellar velocity dispersion through the $M$ – $\sigma$ relation, $M_{\rm BH}\propto\sigma^{4\text{–}5}$ , implying coevolution mediated by AGN feedback.
Direct dynamical proof comes from stellar orbits around Sgr A ( $4.2\times10^6 M_\odot$ ) and Event Horizon Telescope shadow images of M87 and Sgr A*.
"Supermassive" means large mass, not large density or curvature: the mean density inside the horizon falls as $M^{-2}$ and drops below water near $M\approx1.4\times10^8 M_\odot$ . The more massive a black hole is, the lower its average horizon-associated density becomes.
Key idea: A supermassive black hole is a two-parameter ( $M$ , $a_*$ ) spacetime object whose observable power comes from converting accreted rest-mass energy at efficiency $\eta\sim0.06$ – $0.42$ ; its mass is self-regulated by feedback, producing tight scaling relations with the host galaxy.

Where this lives today

Supermassive black holes are now a working laboratory across observational astrophysics. The Event Horizon Telescope has resolved the horizon-scale emission of both M87 (2019) and Sgr A (2022), imaging the photon ring and mapping the magnetic fields that launch jets. At the Galactic Center the GRAVITY interferometer has detected the relativistic pericenter precession and gravitational redshift of S2, confirming general relativity in a regime no other experiment reaches. Reverberation mapping weighs holes in thousands of distant galaxies by timing the delay between an AGN's continuum and its broad emission lines, calibrated against the megamaser mass of NGC 4258. And the pulsar-timing arrays (NANOGrav and partners) announced in 2023 evidence for a nanohertz gravitational-wave background — plausibly the collective hum of supermassive binary inspirals, which the space interferometer LISA will resolve individually in the 2030s. The frontier questions are about origins: JWST is finding surprisingly massive holes and "little red dot" AGN at $z>7$ , sharpening the seed problem, while cosmological simulations use AGN feedback as the ingredient that reproduces the galaxy stellar-mass function and color bimodality.

The physical picture

Think of a supermassive black hole as a gravitational drain at a galaxy's center. What we observe is never the hole but the friction-heated gas circling it: like water glowing as it accelerates into a whirlpool, the accretion disk radiates the potential energy released as matter falls toward the innermost stable circular orbit and then plunges. The energy released per gram is fixed by how deep the well can be tapped — and that depth is set by spin. A non-spinning hole has its last stable orbit at six gravitational radii; drag from a maximally spinning hole pulls it in to a single gravitational radius, wringing out seven times more binding energy. The first figure makes this the whole story: spin sets the inner edge, the inner edge sets the efficiency.

Figure

Spin sets the inner edge of the accretion disk and hence the energy released. A non-spinning Schwarzschild hole has its innermost stable circular orbit at 6

r_g

and radiates 5.7% of the accreted rest mass; a maximally spinning Kerr hole pulls the orbit in to

r_g

and reaches 42% — the deeper the well is tapped, the more energy escapes per gram

The horizon is deceptively gentle for large $M$ . Tidal forces scale as $M/r^3$ , so evaluated at the horizon they scale as $M/r_g^3\propto M^{-2}$ : an astronaut is spaghettified at a stellar-mass hole but would cross a $10^9\,M_\odot$ horizon without feeling a thing. By the same scaling the mean density inside the horizon falls as $M^{-2}$ and drops below water for $M\gtrsim1.4\times10^8\,M_\odot$ — "supermassive" is about total mass and horizon size, never local density or curvature. The genuinely astonishing fact is a piece of cosmic bookkeeping: the hole dominates only its sphere of influence, a few parsecs, yet its mass tracks the velocity dispersion of stars kiloparsecs away. Gravity alone cannot link those scales — the correlation is a fossil record of feedback, in which the growing hole injected enough energy to regulate the very gas reservoir that fed it and formed the stars around it. The second figure is that fingerprint: the $M$ – $\sigma$ relation.

Figure

The M–sigma relation. Black-hole mass climbs as roughly the 4.4 power of the host bulge stellar velocity dispersion, M_BH ≈ 2×10⁸ M_⊙ times (σ / 200 km per s) to the 4.4. Sgr A near 4×10⁶ M_⊙, NGC 4258 near 4×10⁷ M_⊙, and M87 near 6.5×10⁹ M_⊙ span three decades of mass along one line — a correlation gravity alone cannot produce

The core equations

The two length scales of any black hole are the gravitational radius and, for the non-spinning case, the Schwarzschild radius, which is twice as large as the gravitational radius and marks the horizon. These fix every other scale in the problem and, for SMBHs, come out astonishingly small compared with the galaxy yet — for the largest holes — comparable to the Solar System.

r_g\equiv\frac{GM}{c^2},\qquad r_s=\frac{2GM}{c^2}\approx2.95\,\mathrm{km}\left(\frac{M}{M_\odot}\right) \qquad (1)

where: $r_g$ — gravitational radius, the length scale of the metric; $r_s$ — Schwarzschild radius, the horizon of a non-spinning hole; $M$ — black-hole mass; $G,c$ — gravitational constant and speed of light. Reading: for $10^6\,M_\odot$ , $r_s\approx0.02$ AU (inside Mercury's orbit); for M87* at $6.5\times10^9\,M_\odot$ , $r_s\approx128$ AU — a Solar-System-sized horizon, which is why the EHT can resolve it at $16.8$ Mpc. Accretion cannot shine arbitrarily brightly. At the Eddington luminosity the outward radiation force on the plasma's electrons balances the inward gravity on its protons, capping the steady luminosity of any spherical accretor and defining the scale against which every AGN is measured.

L_{\rm Edd}=\frac{4\pi GMm_pc}{\sigma_T}\approx1.26\times10^{31}\left(\frac{M}{M_\odot}\right)\,\mathrm{W} \qquad (2)

where: $m_p$ — proton mass; $\sigma_T$ — Thomson cross section, $6.65\times10^{-29}\,\mathrm{m^2}$ ; $L_{\rm Edd}$ — the luminosity at which radiation pressure on electrons balances gravity on protons for ionized hydrogen. Reading: $L_{\rm Edd}$ is linear in $M$ (equivalently $3.3\times10^4\,(M/M_\odot)\,L_\odot$ ), so a $10^9\,M_\odot$ hole can persistently radiate $\sim10^{40}\,$ W — outshining its host galaxy. Because the horizon volume grows as $M^3$ while the mass grows only as $M$ , the mean density inside the horizon falls with mass — the precise statement behind "supermassive is not dense."

\bar\rho=\frac{M}{\tfrac{4}{3}\pi r_s^3}=\frac{3c^6}{32\pi G^3M^2} \qquad (3)

where: $\bar\rho$ — mass divided by the volume of the Schwarzschild sphere. Reading: $\bar\rho\propto M^{-2}$ , so $\bar\rho(10^8\,M_\odot)\approx1.8\times10^3\,\mathrm{kg\,m^{-3}}$ (just under twice water), falling below water ( $10^3\,\mathrm{kg\,m^{-3}}$ ) for $M\gtrsim1.4\times10^8\,M_\odot$ and below air for the largest holes. The observable output of a disk is a fixed fraction of the accreted rest-mass energy. That fraction, the radiative efficiency, is set by the binding energy of the innermost stable circular orbit, from which matter finally plunges through the horizon carrying only its ISCO energy inward.

\eta=1-E_{\rm ISCO},\qquad L=\eta\,\dot Mc^2 \qquad (4)

where: $\eta$ — fraction of rest-mass energy radiated; $E_{\rm ISCO}$ — specific energy (in units of $c^2$ ) at the innermost stable circular orbit; $\dot M$ — accretion rate; $L$ — emitted luminosity. Reading: $E_{\rm ISCO}=\sqrt8/3$ for Schwarzschild gives $\eta=0.057$ ; $E_{\rm ISCO}=1/\sqrt3$ for maximal prograde Kerr gives $\eta=0.42$ — even $5.7\%$ is $\sim8\times$ more efficient than hydrogen fusion ( $0.7\%$ ). Setting $L=L_{\rm Edd}$ makes the accretion rate proportional to the mass, so mass grows exponentially with a characteristic $e$ -folding time — the Salpeter time — that governs how fast a seed can become a monster.

t_{\rm S}=\frac{\eta\,\sigma_Tc}{4\pi Gm_p(1-\eta)}\approx5.0\times10^{7}\left(\frac{\eta/(1-\eta)}{0.11}\right)\,\mathrm{yr} \qquad (5)

where: $t_{\rm S}$ — $e$ -folding time for Eddington-limited growth; the mass obeys $M(t)=M_0\,e^{t/t_{\rm S}}$ . Reading: with $\eta=0.1$ the Eddington time $\sigma_Tc/4\pi Gm_p=0.45$ Gyr is reduced by $\eta/(1-\eta)$ to $t_{\rm S}\approx5\times10^7\,$ yr, so a doubling takes only $\ln2\times t_{\rm S}\approx35\,$ Myr. The central empirical law of black hole–galaxy coevolution ties $M_{\rm BH}$ to the stellar velocity dispersion of the host bulge with strikingly small scatter — the correlation drawn in the second figure.

M_{\rm BH}\approx2\times10^8\,M_\odot\left(\frac{\sigma}{200\,\mathrm{km\,s^{-1}}}\right)^{4.4} \qquad (6)

where: $\sigma$ — line-of-sight stellar velocity dispersion of the bulge; the slope lies in $4$ – $5$ across calibrations. Reading: because $\sigma$ is a global galaxy property, this steep, tight relation is the strongest clue that black-hole and galaxy growth are coupled. The hole dominates the potential only within its sphere of influence, the radius at which its Keplerian speed matches the stellar dispersion. Resolving this radius is what makes a dynamical mass measurement possible in the first place.

r_h=\frac{GM_{\rm BH}}{\sigma^2}\approx10.8\,\mathrm{pc}\left(\frac{M_{\rm BH}}{10^8 M_\odot}\right)\left(\frac{\sigma}{200\,\mathrm{km\,s^{-1}}}\right)^{-2} \qquad (7)

where: $r_h$ — sphere-of-influence radius. Reading: a few parsecs for a $10^8\,M_\odot$ hole — millions of times smaller than the galaxy whose $\sigma$ it correlates with, the whole puzzle in one number. Finally, general relativity predicts a dark shadow of definite angular size, the silhouette of the photon capture cross section, now imaged directly.

\Delta\theta=\frac{2\sqrt{27}\,GM}{c^2D}=\frac{6\sqrt3\,GM}{c^2D} \qquad (8)

where: $\Delta\theta$ — angular diameter of the Schwarzschild shadow; $D$ — distance to the hole. Reading: the critical impact parameter for photon capture is $b_{\rm ph}=3\sqrt3\,r_g$ , so the shadow diameter is $6\sqrt3\,r_g/D$ ; for M87* ( $6.5\times10^9\,M_\odot$ at $16.8$ Mpc) this is $\approx40\,\mu$ as, matching the $42\pm3\,\mu$ as EHT ring.

Derivation AThe Eddington luminosity from radiation-gravity force balance

The goal is to derive the maximum steady luminosity of a spherical accretor by balancing two forces on an ionized-hydrogen plasma: the outward push of radiation, acting mainly on the light electrons through Thomson scattering, against the inward pull of gravity, acting mainly on the heavy protons. Because electrons and protons are electrostatically bound into a neutral plasma they must move as a unit, and the balance point defines a luminosity above which radiation blows the accretion flow apart. Remarkably the radius drops out, so the result is a clean global cap set only by mass.

Step 1. Write the outward radiation force on a single electron at radius $r$ from a source of luminosity $L$ : the photon momentum flux is $L/4\pi r^2c$ , and the electron intercepts it through its Thomson cross section, so $F_{\rm rad}=\sigma_TL/4\pi r^2c$ . Why: Thomson scattering off free electrons is the dominant momentum-coupling channel in an ionized plasma, and momentum flux is energy flux divided by $c$ .

Step 2. Write the inward gravitational force on an electron–proton pair: $F_{\rm grav}=GM(m_p+m_e)/r^2\approx GMm_p/r^2$ . Why: the electron feels the radiation but the proton carries the weight; electrostatic coupling forces them to fall or resist as one pair, and the proton mass dominates by a factor of $1836$ .

Step 3. Set the two equal at the critical luminosity, $\sigma_TL_{\rm Edd}/4\pi r^2c=GMm_p/r^2$ . The $r^{-2}$ on both sides cancels identically. Why: both radiation and gravity fall off as the inverse square of distance, so their ratio is independent of where you evaluate it — the limit is scale-free and applies as a single number for the whole flow.

Step 4. Solve for the luminosity: $L_{\rm Edd}=4\pi GMm_pc/\sigma_T\approx1.26\times10^{31}\,(M/M_\odot)\,$ W. Above this value radiation pressure exceeds gravity and halts steady spherical accretion. Why: this is the physical ceiling that shapes AGN light curves and, combined with the efficiency, sets the growth clock derived below.

Go deeper · Derivation B · 1 minThe innermost stable circular orbit and the maximum accretion efficiency

Derivation BThe innermost stable circular orbit and the maximum accretion efficiency

The goal is to locate the innermost stable circular orbit in the Schwarzschild geometry, because its binding energy is exactly the fraction of rest mass a thin disk can radiate before matter plunges. Unlike in Newtonian gravity, circular orbits in general relativity cease to be stable inside a critical radius, below which the effective potential has no minimum and any perturbation sends the particle inward. Finding that radius and the specific energy there converts a geometric fact about orbits into the single most important number for accretion power, the efficiency $\eta=1-E_{\rm ISCO}$ .

Step 1. Start from the effective potential for a massive test particle in Schwarzschild geometry (per unit mass, $c=1$ ), $V_{\rm eff}^2=\left(1-2GM/rc^2\right)\left(1+\tilde L^2/r^2c^2\right)$ , where $\tilde L$ is the specific angular momentum. Why: geodesic motion reduces to one-dimensional motion in this potential, so orbits and their stability are read off from its extrema, as in the Newtonian Kepler problem but with a GR term that dominates at small $r$ .

Step 2. Impose the circular-orbit condition and substitute the resulting $\tilde L(r)$ to get the specific energy of a circular orbit, $E(r)=\dfrac{1-2GM/rc^2}{\sqrt{1-3GM/rc^2}}$ . Why: this compact expression carries all the physics — it diverges at the photon sphere $r=3GM/c^2$ and has a minimum that marks the most tightly bound stable orbit.

Step 3. Find the ISCO by setting $\mathrm dE/\mathrm dr=0$ , which gives $r_{\rm ISCO}=6GM/c^2=6r_g$ . Inside $6r_g$ no stable circular orbit exists and matter plunges. Why: the marginally stable orbit is where the binding energy is maximal; it is the effective inner edge of a thin disk and the boundary between orbiting and plunging.

Step 4. Evaluate the energy there: $E_{\rm ISCO}=\sqrt8/3\approx0.943$ , so $\eta=1-E_{\rm ISCO}\approx0.057$ . Spin shrinks the ISCO — to $r_g$ for maximal prograde Kerr — lifting $\eta$ to $1-1/\sqrt3\approx0.42$ . Why: the disk radiates the binding energy accumulated down to the ISCO, so a deeper inner edge means more energy released per gram, which is why spin is the master dial on quasar power.

Go deeper · Derivation C · 1 minExponential growth and the Salpeter e-folding time

Derivation CExponential growth and the Salpeter e-folding time

The goal is to show that Eddington-limited accretion makes a black hole grow exponentially, and to compute the doubling clock that decides whether the billion-solar-mass quasars seen in the first gigayear are even reachable. The key link is that at fixed Eddington ratio the accretion rate is proportional to the current mass, turning the growth equation into the equation for compound interest. The resulting $e$ -folding time, the Salpeter time, is short enough to grow monsters — but only barely, given how little time the early universe allows.

Step 1. Fix the luminosity at Eddington, $L=L_{\rm Edd}=4\pi GMm_pc/\sigma_T$ , and relate it to the rate at which rest mass is supplied, $L=\eta\dot M_{\rm acc}c^2$ . Only the non-radiated fraction is actually swallowed, so the hole's mass grows at $\dot M_{\rm BH}=(1-\eta)\dot M_{\rm acc}$ . Why: bookkeeping matters — a fraction $\eta$ of the inflow leaves as light and never crosses the horizon, so the growth rate is reduced by $(1-\eta)$ .

Step 2. Combine to get $\dot M_{\rm BH}=\dfrac{1-\eta}{\eta}\dfrac{L_{\rm Edd}}{c^2}=\dfrac{1-\eta}{\eta}\dfrac{4\pi Gm_p}{\sigma_Tc}M$ , proportional to $M$ . Why: because $L_{\rm Edd}\propto M$ , the growth rate is proportional to the mass itself — the signature of exponential growth.

Step 3. Integrate to $M(t)=M_0\,e^{t/t_{\rm S}}$ with $t_{\rm S}=\dfrac{\eta}{1-\eta}\dfrac{\sigma_Tc}{4\pi Gm_p}\approx5\times10^7\,$ yr for $\eta=0.1$ . Why: the constant $\sigma_Tc/4\pi Gm_p=0.45$ Gyr is the Eddington time; multiplying by $\eta/(1-\eta)$ gives the mass $e$ -folding time, roughly $45$ – $50$ Myr.

Step 4. Count the folds needed to grow a seed: $100\,M_\odot\to10^9\,M_\odot$ is a factor $10^7$ , or $\ln(10^7)\approx16$ $e$ -folds, requiring $\approx0.7$ – $0.8$ Gyr of continuous near-Eddington accretion. Why: the universe is only $\sim0.7$ Gyr old at $z\approx7.5$ , where such quasars are seen, so the clock is almost fully spent — the tension that drives the whole seed-formation problem.

Limiting cases

SMBH phenomenology is organized by two dials: the accretion rate relative to Eddington, $\dot m=\dot M/\dot M_{\rm Edd}$ , and the spin $a_*$ .

Case	Result	Meaning
Radiatively efficient (thin disk)	$0.01\lesssim\dot m\lesssim1$	Geometrically thin, optically thick Shakura–Sunyaev disk; $\eta\approx0.057$ – $0.42$ ; luminous quasars and Seyferts live here
Radiatively inefficient (Advection-Dominated Accretion Flow)	$\dot m\ll0.01$	Advection-dominated hot flow; most energy is swallowed, not radiated. Sgr A* accretes at $\dot m\sim10^{-8}$ , explaining its extreme underluminosity
Super-Eddington (slim disk)	$\dot m\gtrsim1$	Radiation trapping and advection cap the luminosity near $L_{\rm Edd}$ with only logarithmic dependence on $\dot m$ ; may enable rapid seed growth
Maximal Kerr spin	$a_*\to1,\ r_{\rm ISCO}\to r_g$	ISCO reaches $GM/c^2$ ; ISCO binding energy $1-1/\sqrt3\approx0.42$ gives the maximum efficiency and the most powerful Blandford–Znajek jets

Common trapThe M–sigma relation is just gravity, and "supermassive" means "dense".

Both halves are wrong, and they are the two most common misconceptions about SMBHs. First, the black hole gravitationally controls only its sphere of influence — a few parsecs — while $\sigma$ is measured over kiloparsecs, so gravity alone cannot causally link them; the $M$ – $\sigma$ relation is surprising precisely because it demands a non-gravitational messenger, namely AGN feedback coupling accretion energy to the galaxy's gas. Second, "supermassive" is a mass label, not a density label: since $r_s\propto M$ , the mean density inside the horizon falls as $M^{-2}$ and drops below water for $M\gtrsim1.4\times10^8\,M_\odot$ , and tidal forces at the horizon fall as $M^{-2}$ too, so a large hole's horizon is a region of weak local curvature, not a crushing one. Two further traps: the horizon is not a material surface, and the accretion efficiency is a property of the innermost stable orbit, not of the horizon.

Worked examples

1 · The dynamical mass of Sgr A* from the orbit of the star S2 (fully worked)

The star S2 orbits the radio source Sgr A at the Galactic Center on a Keplerian ellipse with period $P=16.05$ yr and angular semimajor axis $a=0.1255''$ ; the distance is $R_0=8.18$ kpc. Find the enclosed mass and argue it must be a single black hole.*

Step 1. Convert the angular semimajor axis to a physical length. At $R_0=8.18$ kpc, $a=0.1255''\times8180\,\mathrm{pc}\approx1027\,\mathrm{AU}\approx1.54\times10^{14}\,$ m. This is the true size of the orbit; the semimajor axis is about 26 times Pluto's orbital radius.*

Step 2. Apply Kepler's third law in the limit $M\gg m_\star$ : $M=\dfrac{4\pi^2a^3}{GP^2}$ . With $P=16.05\,\mathrm{yr}=5.06\times10^8\,$ s, $M=\dfrac{4\pi^2(1.54\times10^{14})^3}{(6.674\times10^{-11})(5.06\times10^8)^2}$ .

Step 3. The arithmetic gives $M\approx8.4\times10^{36}\,\mathrm{kg}\approx\mathbf{4.2\times10^6\,M_\odot}$ — matching the GRAVITY Collaboration value $4.15\times10^6\,M_\odot$ .

Step 4. Compactness clinches it. S2's pericenter is $r_{\rm peri}\approx120$ AU, so $4\times10^6\,M_\odot$ sits within a region smaller than the Solar System, a mean density $\bar\rho\gtrsim 5\times10^{15}\,M_\odot\,\mathrm{pc^{-3}}$ . Such an extreme concentration of dark mass is very difficult to maintain as a stable cluster of stars or stellar remnants. Together with the long-term stability of the observed stellar orbits and the compact radio source Sgr A*, this provides compelling dynamical evidence for a single supermassive black hole at the Galactic Center.

2 · Growing a billion-solar-mass quasar by z ≈ 7

The quasar J0313−1806 hosts a $\sim1.6\times10^9\,M_\odot$ black hole at $z=7.6$ , when the universe was $\sim0.67$ Gyr old. Can a stellar-remnant seed reach this mass, and what does it require?

Step 1. Take a Population III remnant seed of $M_0\approx100\,M_\odot$ and count the $e$ -folds to $1.6\times10^9\,M_\odot$ : $\ln(1.6\times10^9/100)=\ln(1.6\times10^7)\approx16.6$ .

Step 2. At the Salpeter clock $t_{\rm S}\approx5\times10^7\,$ yr (efficiency $\eta=0.1$ ), the required time is $16.6\times5\times10^7\approx\mathbf{0.83\,Gyr}$ of continuous Eddington-limited accretion.

Step 3. Compare with the age available, $\sim0.67$ Gyr. The seed growth overruns the clock, so at least one of three escapes is needed: a heavier seed ( $10^4$ – $10^5\,M_\odot$ direct-collapse), a lower efficiency (smaller $t_{\rm S}$ ), or episodes of super-Eddington accretion. Beyond these three, the tension can also be eased by very high duty cycles and/or additional growth through black-hole mergers. This is the high- $z$ quasar seed problem, one of the sharpest open questions in the field.

Physics of the portrait

The supermassive black hole is where two and a half centuries of physics converge: a Newtonian speculation, a relativistic exact solution, a rotating generalization, an accretion engine, and finally a direct image.

Origins · 1 minsupermassive black hole and how this was discovered

supermassive black hole

the two-parameter Kerr object set by mass M and spin a at a galaxy's center · galactic nuclei, accretion power, and black hole–galaxy coevolution

The idea began as a Newtonian curiosity: in 1783 John Michell reasoned that a star massive enough would have an escape speed exceeding light's, making it a "dark star" — the right conclusion from the wrong physics. It became exact in 1916 when Karl Schwarzschild, solving Einstein's field equations for a point mass while serving on the Russian front, found the metric with its horizon at $r_s=2GM/c^2$ ; for decades it was thought a pathology until the 1960s black-hole renaissance made it physical. Roy Kerr's 1963 rotating solution supplied the second parameter, spin, and the no-hair theorems established that an astrophysical black hole is fixed by $M$ and $a$ alone. The link to galaxies came from accretion: Salpeter and Zel'dovich (1964) argued that infalling gas, not the hole, is the power source, and Donald Lynden-Bell in 1969 proposed that quasars are accretion disks around $\sim10^8\,M_\odot$ holes and that dead ones lurk in ordinary galactic centers — including our own. Sołtan (1982) closed the energetics by matching integrated quasar light to the local black-hole mass density. The final turns were observational: Genzel's and Ghez's teams tracked stars around Sgr A into full orbits (Nobel Prize 2020), the $M$ – $\sigma$ relation revealed coevolution (2000), and in 2019 the Event Horizon Telescope imaged the shadow of M87 — turning Michell's dark star into a photograph.

History

1783

— John Michell conceives "dark stars" whose escape speed exceeds that of light

1916

— Karl Schwarzschild solves Einstein's equations for a point mass, finding the horizon

1963

— Roy Kerr publishes the rotating black-hole metric, supplying the spin parameter

1969

— Donald Lynden-Bell proposes accreting massive black holes power quasars and hide in galactic nuclei

2019

— the Event Horizon Telescope images the shadow of M87*

Going further

Texts. Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., ch. 17 (black holes) and ch. 28 (AGN and quasars) · Binney & Tremaine, Galactic Dynamics, 2nd ed., §1.1.3 and dynamical mass estimation · Mo, van den Bosch & White, Galaxy Formation and Evolution, ch. 14 (AGN and feedback) · Dodelson, Modern Cosmology (structure growth and seeds) · Frank, King & Raine, Accretion Power in Astrophysics, 3rd ed. · Kormendy & Ho 2013, ARA&A 51, 511 (coevolution review). All numerical results use physical constants ( $G,c,m_p,\sigma_T,M_\odot$ ) from CODATA. Key values: $r_s\approx2.95\,(M/M_\odot)\,$ km; $L_{\rm Edd}=1.26\times10^{31}\,(M/M_\odot)\,$ W; $\bar\rho(10^8M_\odot)\approx1.8\times10^3\,\mathrm{kg\,m^{-3}}$ with the water crossover at $M\approx1.4\times10^8\,M_\odot$ ; $\eta_{\rm Schw}=1-\sqrt8/3=0.057$ and $\eta_{\rm Kerr,max}=1-1/\sqrt3=0.42$ ; the Salpeter $e$ -folding time $t_{\rm S}=(\eta/(1-\eta))\,\sigma_Tc/4\pi Gm_p\approx5\times10^7\,$ yr for $\eta=0.1$ — note that $\sigma_Tc/4\pi Gm_p=0.45$ Gyr is the underlying Eddington time, not the mass $e$ -folding time; the Sgr A Kepler mass $\approx4.2\times10^6\,M_\odot$ from $P=16.05$ yr, $a=0.1255''$ , $R_0=8.18$ kpc (GRAVITY 2019); and the M87 shadow $\approx40\,\mu$ as from $6\sqrt3\,GM/c^2D$ (EHT 2019). The $M$ – $\sigma$ normalization and slope $4$ – $5$ follow Kormendy & Ho 2013.

Prerequisites. Eddington Luminosity and Accretion Limits · Thin Accretion Disks and the Shakura–Sunyaev Model · Black Hole Spin and the Ergosphere. Leads to. Active Galactic Nuclei: the Unified Model · The M–sigma Relation and Black Hole–Galaxy Coevolution · AGN Feedback · Quasars and Quasar Evolution.

Ask the tutor: Ask me to walk through why a maximally spinning black hole reaches $42\%$ efficiency while a non-spinning one manages only $5.7\%$ , starting from the ISCO.

Have we actually proven Sgr A* is a black hole and not a dense star cluster?

As close to proven as dynamical astronomy allows. Two decades of infrared monitoring have tracked stars — most famously S2 — through complete orbits around an invisible mass, giving $M\approx4.15\times10^6\,M_\odot$ within S2's pericenter of $\sim120$ AU. That implies a mean density above $10^{16}\,M_\odot\,\mathrm{pc^{-3}}$ , at which any cluster of dark remnants would collapse or evaporate far faster than the galaxy's age. Add the EHT's 2022 horizon-scale ring of the predicted size and GRAVITY's detection of S2's relativistic redshift and orbit precession, and no non-black-hole model survives. The 2020 Nobel Prize to Genzel and Ghez recognized exactly this chain of evidence.

Why is Sgr A* so faint if accretion onto black holes is supposed to be luminous?

Because it is starving. Sgr A* accretes at an Eddington ratio $\dot m\sim10^{-8}$ , deep in the radiatively inefficient regime where the flow is a hot, thick advection-dominated accretion flow (ADAF) rather than a thin, luminous disk. In an ADAF most of the liberated energy is not radiated — it is carried across the horizon as heat — so the efficiency plummets far below the canonical $10\%$ . Our $4\times10^6\,M_\odot$ hole thus shines at only $\sim10^{-9}$ of its Eddington luminosity, whereas the same mass fed near Eddington would outshine the entire Milky Way.

What sets the accretion efficiency — the horizon or something else?

The innermost stable circular orbit, not the horizon. A thin disk radiates the binding energy that gas accumulates as it spirals down to the ISCO; once inside that radius the gas plunges nearly in free fall and carries its remaining energy across the horizon largely unradiated. So the efficiency is $\eta=1-E_{\rm ISCO}$ , fixed by where the last stable orbit sits. For a non-spinning hole the ISCO is at $6r_g$ and $\eta=5.7\%$ ; spin drags the ISCO inward — to $r_g$ for maximal prograde Kerr — raising $\eta$ to $42\%$ . The horizon itself is not a radiating surface; it is the efficiency of the orbits just outside it that governs how bright accretion can be.

How can billion-solar-mass black holes exist when the universe was under a gigayear old?

This is the seed problem, and it is genuinely unresolved. Growing a $10^9\,M_\odot$ hole from a $\sim100\,M_\odot$ remnant needs about $16$ Salpeter $e$ -foldings, roughly $0.7$ – $0.8$ Gyr of uninterrupted Eddington accretion — essentially the entire age of the universe at $z\approx7.5$ , with no slack for the gaps real accretion shows. The proposed escapes are heavier "direct-collapse" seeds of $10^4$ – $10^5\,M_\odot$ from atomic-cooling halos, episodes of super-Eddington accretion that fold faster, or runaway mergers in dense early clusters. JWST's overmassive holes and faint red AGN at $z>7$ have made the problem sharper, not easier.

Is the M–sigma relation a coincidence, or does it mean something deep?

It means the black hole and its host grew together. The relation is far too tight — scatter of only $\sim0.3$ dex over three decades in mass — to be an accident, yet the hole's gravity reaches only parsecs while $\sigma$ is a kiloparsec-scale property, so gravity cannot be the messenger. The standard resolution is AGN feedback: the energy from growing the hole, $\eta M_{\rm BH}c^2$ , exceeds the bulge binding energy $\sim M_{\rm bulge}\sigma^2$ by orders of magnitude, so even a percent-level coupling to the gas can heat or expel the reservoir, shutting off both star formation and further growth. That self-regulation naturally produces the observed slope near $M_{\rm BH}\propto\sigma^4$ – $\sigma^5$ .

Would you be torn apart crossing the horizon of a supermassive black hole?

No — and this surprises people. Tidal forces scale as $M/r^3$ , so at the horizon ( $r\sim r_g\propto M$ ) they scale as $M^{-2}$ : the bigger the hole, the gentler the horizon. At a stellar-mass hole the tidal stretching ("spaghettification") is lethal well outside the horizon, but at a $10^9\,M_\odot$ hole the tidal gradient across a body at the horizon is imperceptible — you would cross it with no local signal that anything special had happened. The horizon is a global feature of the spacetime, a one-way surface, not a wall of force. What is inescapable is the future: once inside, all paths lead to the singularity.