Innermost Stable Circular Orbit (ISCO)

The innermost stable circular orbit is the sharpest place where general relativity parts company with Newton. In Newtonian gravity a body can circle a mass at any radius, and every such circle is stable — nudge the orbit and it merely wobbles about its old track. Schwarzschild geometry forbids this. Below a definite radius, $r_{\mathrm{ISCO}}=6GM/c^2$ , no stable circular orbit exists at all: the smallest inward push sends matter spiralling through the horizon. That radius has no free parameters — for a non-rotating hole it is exactly three Schwarzschild radii, independent of the particle's energy or angular momentum. The result flows from one idea built in the preceding articles: the conserved energy and angular momentum of the timelike and azimuthal Killing vectors collapse the geodesic equation into one-dimensional motion in an effective potential $V_{\mathrm{eff}}(r)$ . Circular orbits sit at its extrema, stable ones in its minima. As the angular momentum is lowered, the minimum (a stable orbit) and the neighbouring maximum (an unstable one) slide together and annihilate at a single inflection point. That coalescence — $V_{\mathrm{eff}}'=V_{\mathrm{eff}}''=0$ — is the ISCO. Its consequences are enormous. A particle lowered quasi-statically from rest at infinity to the ISCO must shed a fixed fraction of its rest energy, $1-2\sqrt2/3\approx5.7\%$ , and this single number caps the efficiency of every thin accretion disc around a Schwarzschild black hole — an order of magnitude beyond nuclear fusion. The ISCO marks the inner edge of that disc, sets the inspiral-to-plunge transition in a merger, and, through its strong dependence on spin, is a principal handle by which astronomers weigh how fast real black holes rotate.

Key takeaways

For a massive test particle in Schwarzschild geometry, stable circular orbits exist only outside $r_{\mathrm{ISCO}}=6GM/c^2=3\,r_s$ — the innermost stable circular orbit.
The ISCO is where the effective potential's stable minimum and unstable maximum merge into an inflection: $V_{\mathrm{eff}}'=V_{\mathrm{eff}}''=0$ at the same radius.
There the specific angular momentum and specific energy are both minimised, $\tilde L=\sqrt{12}\,GM/c$ and $\tilde E=\tfrac{2\sqrt2}{3}c^2\approx0.943\,c^2$ .
The binding energy at the ISCO, $1-\tfrac{2\sqrt2}{3}\approx5.7\%$ of rest mass, sets the maximum radiative efficiency of a thin accretion disc — versus $0.7\%$ for hydrogen fusion.
Inside the ISCO matter plunges on a nearly geodesic path with little further radiation, so the ISCO is the effective inner edge of the disc and a fiducial inspiral-to-merger point for gravitational waves.
For a maximally spinning Kerr hole the prograde ISCO shrinks to $GM/c^2$ (efficiency $\approx42\%$ ) and the retrograde one grows to $9GM/c^2$ ( $\approx3.8\%$ ), which is how the ISCO measures black-hole spin.
Key idea: the ISCO is the radius at which the stable minimum and unstable maximum of the effective potential coalesce — where $V_{\mathrm{eff}}'=V_{\mathrm{eff}}''=0$ — giving the spin-independent Schwarzschild value $r=6GM/c^2$ and a maximum disc efficiency of $\approx5.7\%$ .

The core insight

The ISCO is the radius at which the stable minimum and unstable maximum of the effective potential coalesce — where $V_{\mathrm{eff}}'=V_{\mathrm{eff}}''=0$ — yielding the spin-independent Schwarzschild value $r=6GM/c^2$ and a maximum disc efficiency of $\approx5.7\%$ .

Where this lives today

The ISCO is now a measured quantity, not just a textbook radius. In X-ray binaries and active galactic nuclei two techniques read it directly: continuum-fitting infers the inner-disc radius from the thermal spectrum, and X-ray reflection spectroscopy fits the relativistically broadened, redshifted iron K $\alpha$ line emitted just outside the ISCO. Because the Kerr ISCO ranges from $GM/c^2$ to $9GM/c^2$ , locating it pins the dimensionless spin $a/M$ — this is how holes such as Cygnus X-1 and MCG–6-30-15 are found to spin, some near extremal. In gravitational-wave astronomy the ISCO frequency $f_{\mathrm{GW}}\approx4.4\,\mathrm{kHz}\times(M_\odot/M)$ marks the inspiral-to-plunge transition and anchors the matching of post-Newtonian and numerical-relativity waveforms used to read every LIGO–Virgo–KAGRA detection. And because the ISCO exists only through the relativistic $1/r^3$ term, any departure of its location from the Kerr prediction would flag modified gravity or a horizonless compact object — a strong-field test complementary to the photon ring imaged around M87 and Sgr A by the Event Horizon Telescope.

The physical picture

The Newtonian effective potential is $-GM/r+L^2/2r^2$ : a $-1/r$ well and a $+1/r^2$ centrifugal barrier. However small $L$ is, the barrier always wins at small radius, so the potential turns up steeply and every circular orbit sits in a well — stable forever. General relativity adds a third term, $-GML^2/(c^2r^3)$ , that falls off faster than the barrier and eventually overwhelms it. Lower the angular momentum and the well grows shallow; at a critical value it flattens completely, and below that no well is left to hold an orbit. The first figure shows this family of potentials and the moment the well disappears.

Figure

The effective potential Veff(r) for three values of specific angular momentum L: with L = 4 GM/c a stable minimum sits at r = 12GM/c² behind an unstable barrier at r = 4GM/c², the critical value L = √12 GM/c ≈ 3.46 GM/c merges them into one flat inflection at the ISCO r = 6GM/c², and L = 3 GM/c is below the minimum so no circular orbit exists and the particle plunges

Think of the ISCO as the edge of a cliff for orbiting matter. Just outside it a small inward nudge gives a bounded oscillation about the circular radius — the orbit is stable. Just inside it the same nudge finds no restoring force, because the radial "spring constant" $V_{\mathrm{eff}}''$ has gone negative, and the particle runs away inward. The second figure follows what astronomers actually measure: the specific energy of a circular orbit versus radius. It bottoms out precisely at the ISCO, and the depth of that dip below the rest-mass line is the energy an accreting particle can radiate.

Figure

Specific energy E of a circular orbit versus radius r: the stable branch outside r = 6GM/c² and the unstable branch between 3 and 6 GM/c² meet at the ISCO minimum r = 6GM/c², where E = 0.943c² is least, so the maximum binding energy a thin disc can radiate is 1 − 0.943 = 5.7% of the rest mass, and E diverges at the photon sphere r = 3GM/c²

The core equations

The starting point is the radial equation of motion for a timelike geodesic, obtained by feeding the two conserved quantities into the four-velocity normalisation. It reads as an energy equation for one-dimensional motion in $r$ ; every result below comes from it.

\left(\frac{\mathrm{d}r}{\mathrm{d}\tau}\right)^{2}=\tilde E^{2}-V_{\mathrm{eff}}^{2}(r) \qquad (1)

where: $r$ — Schwarzschild (areal) radial coordinate; $\tau$ — proper time; $\tilde E$ — conserved energy per unit rest mass (units of $c^2$ ); $V_{\mathrm{eff}}$ — effective potential for radial motion. Reading it: a turning point occurs where $\dot r=0$ , i.e. $\tilde E=V_{\mathrm{eff}}$ ; a circular orbit needs that turning point to persist, forcing $V_{\mathrm{eff}}'=0$ . The effective potential is the product of the metric's redshift factor and a centrifugal-plus-rest term. Written as a square it is the cleanest object to differentiate, and its extrema and inflections coincide with those of $V_{\mathrm{eff}}$ since $V_{\mathrm{eff}}>0$ .

V_{\mathrm{eff}}^{2}(r)=\left(1-\frac{2GM}{c^{2}r}\right)\left(c^{2}+\frac{\tilde L^{2}}{r^{2}}\right) \qquad (2)

where: $\tilde L$ — conserved angular momentum per unit rest mass; $M$ — central mass; $G,c$ — Newton's constant and light speed. Reading it: expanding the product yields a term $-GM\tilde L^2/(c^2r^3)$ with no Newtonian counterpart — that single relativistic term destroys the potential well and creates the ISCO. Any circular orbit lives at an extremum; whether it is stable depends on the sign of the second derivative — these two conditions separate stable orbits (minima) from unstable ones (maxima). The ISCO is the borderline case where well and barrier have just annihilated, so both derivatives vanish together, one condition more than an ordinary circular orbit and hence an isolated radius.

V_{\mathrm{eff}}'(r)=0,\ \ V_{\mathrm{eff}}''(r)>0\ \ \text{(stable)};\qquad V_{\mathrm{eff}}'(r)=V_{\mathrm{eff}}''(r)=0\ \ \text{(ISCO)} \qquad (3)

where: $V_{\mathrm{eff}}'$ — first radial derivative, vanishing at every circular orbit; $V_{\mathrm{eff}}''$ — second derivative, positive in a well (restoring force, bounded oscillation), negative on a hilltop (runaway), zero at the marginal ISCO inflection. Reading it: the extra $V_{\mathrm{eff}}''=0$ is exactly what promotes "some circular orbit" into "the innermost stable one". Solving $V_{\mathrm{eff}}'=0$ for the angular momentum shows how much spin a circular orbit needs at each radius; the formula diverges at the photon sphere and bottoms out at the ISCO.

\tilde L^{2}(r)=\frac{GM\,r}{1-3GM/(c^{2}r)} \qquad (4)

where: $\tilde L(r)$ — specific angular momentum required for a circular orbit at radius $r$ . Reading it: as $r\to3GM/c^2$ (the photon sphere) the required $\tilde L\to\infty$ , and the function turns over at $r=6GM/c^2$ , so no circular orbit needs less than $\tilde L_{\min}=\sqrt{12}\,GM/c$ . Imposing the double condition (3) collapses everything to a single number: the ISCO radius is a fixed multiple of the geometric mass, independent of the particle.

r_{\mathrm{ISCO}}=\frac{6GM}{c^{2}}=3\,r_s \qquad (5)

where: $r_{\mathrm{ISCO}}$ — coordinate radius of the innermost stable circular orbit; $r_s=2GM/c^2$ — the Schwarzschild radius. Reading it: the ISCO sits at three Schwarzschild radii, twice the photon-sphere radius, independent of $\tilde L$ and $\tilde E$ . Evaluating the conserved quantities on the ISCO gives their minimal values; the energy fixes the accretion efficiency.

\tilde L_{\mathrm{ISCO}}=\sqrt{12}\,\frac{GM}{c},\qquad \tilde E_{\mathrm{ISCO}}=\frac{2\sqrt2}{3}\,c^{2}\approx0.9428\,c^{2} \qquad (6)

where: $\tilde L_{\mathrm{ISCO}}$ — minimum specific angular momentum of any stable circular orbit; $\tilde E_{\mathrm{ISCO}}$ — specific energy there. Reading it: since $\tilde E_{\mathrm{ISCO}}<c^2$ , a particle from infinity is bound, and the deficit is what it radiated on the way in. That deficit, expressed as a fraction of rest energy, is the maximum radiative efficiency of a thin disc.

\eta=1-\frac{\tilde E_{\mathrm{ISCO}}}{c^{2}}=1-\frac{2\sqrt2}{3}\approx0.0572 \qquad (7)

where: $\eta$ — fraction of rest-mass energy radiated spiralling from rest at infinity to the ISCO. Reading it: $5.7\%$ is a ceiling, not a rate — inside the ISCO the plunging gas carries its remaining $94.3\%$ across the horizon largely unradiated. Finally, the orbital angular velocity keeps the Newtonian Keplerian form even in the strong field, and doubling it gives the dominant gravitational-wave frequency at the ISCO.

\Omega=\sqrt{\frac{GM}{r^{3}}},\qquad f_{\mathrm{GW}}=\frac{\Omega}{\pi}\Big|_{r_{\mathrm{ISCO}}}=\frac{c^{3}}{6^{3/2}\pi\,GM} \qquad (8)

where: $\Omega$ — coordinate angular velocity; $f_{\mathrm{GW}}$ — dominant (quadrupole) wave frequency at the ISCO. Reading it: it scales as $1/M$ , giving the fiducial $f_{\mathrm{GW}}\approx4.4\,\mathrm{kHz}\times(M_\odot/M)$ that marks the end of the slow inspiral.

Derivation ADeriving the effective potential from the geodesic equation

The goal is to show how a geodesic in four-dimensional curved spacetime becomes one-dimensional motion in a potential, since that reduction is the entire engine behind the ISCO. Two symmetries of the metric — time-translation and axial rotation — supply two constants of motion through their Killing vectors, and the four-velocity normalisation supplies a third relation; combining them collapses the geodesic equations into a single first-order equation for $r$ whose potential carries a purely relativistic term absent from Newton.

Step 1. Restrict to the equatorial plane $\theta=\pi/2$ and write the four-velocity normalisation $g_{\mu\nu}u^\mu u^\nu=-c^2$ explicitly: $-(1-2GM/c^2r)c^2\dot t^2+(1-2GM/c^2r)^{-1}\dot r^2+r^2\dot\phi^2=-c^2$ (overdot is $\mathrm{d}/\mathrm{d}\tau$ ). Why: this one constraint contains all the radial dynamics once $t$ and $\phi$ are eliminated.

Step 2. Introduce the conserved quantities of the timelike and azimuthal Killing vectors, $\tilde E=(1-2GM/c^2r)c^2\dot t$ and $\tilde L=r^2\dot\phi$ , each constant because contracting a Killing vector with $u^\mu$ gives a quantity whose proper-time derivative vanishes. Why: these fix $\dot t$ and $\dot\phi$ , leaving $r$ as the only dynamical variable.

Step 3. Substitute $\dot t,\dot\phi$ back and rearrange: $\dot r^2=\tilde E^2/c^2-(1-2GM/c^2r)(c^2+\tilde L^2/r^2)$ . Identifying the last product as $V_{\mathrm{eff}}^2(r)$ gives the energy equation (1). Why: the geodesic is now a particle of energy $\tilde E$ rolling in the potential $V_{\mathrm{eff}}$ — the Newtonian orbit tool made relativistic.

Step 4. Expand the potential: $V_{\mathrm{eff}}^2=c^2-2GMc^2/r+\tilde L^2/r^2-2GM\tilde L^2/(c^2r^3)$ . The first three pieces mirror Newton (rest energy, gravity, centrifugal barrier); the fourth, negative, is new. Why: this $-1/r^3$ term outgrows the $+1/r^2$ barrier at small radius, and is the necessary and sufficient reason an ISCO exists at all. *Verified: expanding $(1-2/r)(1+\tilde L^2/r^2)$ (with $GM/c^2=1$ ) reproduces the four terms; dropping the $1/r^3$ piece recovers a well stable for every $r$ , per Carroll §5.4 and Hartle Ch. 9.*

Go deeper · Derivation B · 1 minLocating the ISCO where the potential's extrema merge

Derivation BLocating the ISCO where the potential's extrema merge

The aim is to turn the geometric statement "the minimum and maximum of the potential coalesce" into an algebraic solution for the radius, and to show the answer is independent of the orbiting particle. Working with $W(r)\equiv V_{\mathrm{eff}}^2(r)$ is legitimate because $V_{\mathrm{eff}}>0$ , so $W$ shares every extremum and inflection with $V_{\mathrm{eff}}$ ; imposing $W'=W''=0$ gives two equations in $r$ and $\tilde L$ , and eliminating $\tilde L$ leaves a single clean root. Throughout, $u\equiv GM/c^2$ .

Step 1. Differentiate $W=(1-2u/r)(c^2+\tilde L^2/r^2)$ once: $W'(r)=2uc^2/r^2-2\tilde L^2/r^3+6u\tilde L^2/r^4$ . Setting $W'=0$ and clearing $r^4/2$ gives the circular-orbit relation $uc^2r^2-\tilde L^2 r+3u\tilde L^2=0$ . Why: this is the general circular-orbit condition; solving it for $\tilde L^2$ yields $\tilde L^2=uc^2r^2/(r-3u)$ , the angular momentum of equation (4).

Step 2. Differentiate again: $W''(r)=-4uc^2/r^3+6\tilde L^2/r^4-24u\tilde L^2/r^5$ . Setting $W''=0$ and clearing $r^5/2$ gives $-2uc^2r^2+3\tilde L^2 r-12u\tilde L^2=0$ . Why: this second relation is the extra condition that marks the boundary of stability — it holds only at the inflection where a minimum and maximum have just met.

Step 3. Eliminate $\tilde L^2$ by substituting $\tilde L^2=uc^2r^2/(r-3u)$ from Step 1 into the Step 2 relation, then divide through by the non-zero factor $uc^2r^2/(r-3u)$ . What remains is $-2(r-3u)+3r-12u=0$ , which simplifies to $r-6u=0$ . Why: the two conserved-quantity–dependent equations combine into one equation for $r$ alone, so the ISCO radius cannot depend on $\tilde L$ .

Step 4. Read off $r_{\mathrm{ISCO}}=6u=6GM/c^2$ and back-substitute: $\tilde L_{\mathrm{ISCO}}^2=uc^2(6u)^2/(6u-3u)=12u^2c^2$ , so $\tilde L_{\mathrm{ISCO}}=\sqrt{12}\,GM/c$ . Why: this is the minimum of $\tilde L(r)$ — no stable circular orbit needs less angular momentum, i.e. none exists below the ISCO. *Verified: substituting $r=6GM/c^2$ , $\tilde L^2=12(GM/c^2)^2c^2$ into $W'=0$ and $W''=0$ satisfies both identically; matches MTW §25.5 and Shapiro & Teukolsky §12.4.*

Go deeper · Derivation C · 1 minThe ISCO energy and the maximum accretion efficiency

Derivation CThe ISCO energy and the maximum accretion efficiency

This derivation extracts the single number that makes the ISCO astrophysically famous: the fraction of rest energy accreting matter can radiate. Because a circular orbit has $\dot r=0$ , equation (1) gives $\tilde E^2=V_{\mathrm{eff}}^2(r)=W(r)$ , so evaluating $W$ at the ISCO returns the specific energy directly. Subtracting it from the rest-mass value $c^2$ gives the binding energy, and since a thin disc radiates exactly this deficit as gas drifts inward, that binding energy is the disc's maximum efficiency.

Step 1. Evaluate the redshift factor at the ISCO: $1-2GM/(c^2r_{\mathrm{ISCO}})=1-2u/6u=2/3$ . Why: this factor is the gravitational time-dilation the orbiting matter sits in, and it carries most of the energy loss.

Step 2. Evaluate the centrifugal bracket with $\tilde L_{\mathrm{ISCO}}^2=12u^2c^2$ and $r_{\mathrm{ISCO}}^2=36u^2$ : $c^2+\tilde L_{\mathrm{ISCO}}^2/r_{\mathrm{ISCO}}^2=c^2(1+1/3)=\tfrac43c^2$ . Why: keeping exact fractions makes the answer a clean closed form.

Step 3. Combine the two factors: $\tilde E_{\mathrm{ISCO}}^2=\tfrac23\cdot\tfrac43c^2\cdot c^2=\tfrac89c^4$ , so $\tilde E_{\mathrm{ISCO}}=\sqrt{8/9}\,c^2=\tfrac{2\sqrt2}{3}c^2\approx0.9428\,c^2$ . Why: this is the energy of a particle sitting on the ISCO, the lowest specific energy of any stable circular orbit.

Step 4. Form the binding energy per unit rest mass, $\eta=(c^2-\tilde E_{\mathrm{ISCO}})/c^2=1-\tfrac{2\sqrt2}{3}\approx0.0572$ . A particle lowered quasi-statically from rest at infinity ( $\tilde E=c^2$ ) to the ISCO must therefore radiate about $5.7\%$ of its rest energy. Why: for a Novikov–Thorne / Shakura–Sunyaev thin disc this deficit is emitted as heat and light, so $\eta$ is the accretion efficiency ceiling for a Schwarzschild hole. *Verified: $2\sqrt2/3=0.942809\ldots$ , $\eta=0.057191\ldots$ , the standard $\approx5.7\%$ of Shapiro & Teukolsky §12.4 and MTW §25.5; the second figure's energy curve bottoms at $0.9428$ exactly at $r=6GM/c^2$ .*

Limiting cases

The ISCO reduces to memorable closed forms in the cases that matter most; the spin cases show that $6GM/c^2$ is only the non-rotating anchor of a wide range.

Case	Result	Meaning
Schwarzschild (non-rotating)	$r_{\mathrm{ISCO}}=6GM/c^2=3\,r_s$	Canonical value, independent of the particle; twice the photon sphere.
Extremal Kerr, prograde	$r_{\mathrm{ISCO}}=GM/c^2=\tfrac12 r_s$	Maximal co-rotation ( $a=GM/c^2$ ) drags the ISCO down to the horizon in Boyer–Lindquist $r$ ; efficiency rises to $1-1/\sqrt3\approx42\%$ .
Extremal Kerr, retrograde	$r_{\mathrm{ISCO}}=9GM/c^2$	Counter-rotating orbits are pushed outward; efficiency falls to $1-\sqrt{25/27}\approx3.8\%$ .
Marginally bound orbit	$r_{\mathrm{mb}}=4GM/c^2,\ \tilde E=c^2$	The circular orbit with exactly escape energy; between it and the ISCO orbits are bound but unstable.
Newtonian limit	$V_{\mathrm{eff}}''>0$ for all $r$	With no $1/r^3$ term every circular orbit is stable — there is no ISCO. The phenomenon is intrinsically relativistic.

Common trapthe ISCO is just where the centrifugal force can no longer balance gravity

The tempting Newtonian story — "at $6GM/c^2$ gravity finally overwhelms the centrifugal barrier" — is wrong twice over. First, a plain force balance is exactly $V_{\mathrm{eff}}'=0$ , satisfied at every circular orbit and never singling out an innermost radius; the ISCO is the second condition $V_{\mathrm{eff}}''=0$ , where the stable minimum and unstable maximum merge. Newtonian gravity has orbits at all radii and no ISCO precisely because it lacks the $-GM\tilde L^2/(c^2r^3)$ term. Second, $r=6GM/c^2$ is a coordinate (areal) radius, not a proper distance, and holds only for non-rotating holes — spin moves it anywhere from $GM/c^2$ to $9GM/c^2$ .

Worked examples

1 · The ISCO of Sagittarius A* (fully worked)

The black hole at the Galactic Centre has $M=4.3\times10^{6}\,M_\odot$ . Find the radius of its ISCO in kilometres, in astronomical units, and in Schwarzschild radii, and the orbital period of matter sitting on it. Use $GM_\odot/c^2=1.477$ km and $GM_\odot/c^3=4.925\times10^{-6}$ s.

Step 1. Geometric mass. Scale the solar value: $GM/c^2=4.3\times10^{6}\times1.477\ \mathrm{km}=6.35\times10^{6}$ km — the natural length unit for every radius here.

Step 2. ISCO radius. Six geometric masses: $r_{\mathrm{ISCO}}=6GM/c^2=\mathbf{3.81\times10^{7}\ km}=\mathbf{0.25\ AU}$ (a quarter of the Earth–Sun distance), and since $r_{\mathrm{ISCO}}=3\,r_s$ always, exactly three Schwarzschild radii — about $55\,R_\odot$ .

Step 3. Angular velocity. With $\Omega=c^3/(6^{3/2}GM)$ , $GM/c^3=21.2$ s and $6^{3/2}=14.70$ : $\Omega=1/(14.70\times21.2\ \mathrm{s})=3.21\times10^{-3}\ \mathrm{rad/s}$ .

Step 4. Orbital period. $T=2\pi/\Omega=\mathbf{1955\ s\approx33\ minutes}$ . Matter on the ISCO of Sgr A* orbits in barely over half an hour — close to the tens-of-minutes timescale of the infrared flares the GRAVITY interferometer has tracked near the hole, a direct look at strong-field motion.

2 · How much a quasar must swallow (fully worked)

A luminous quasar radiates $L=10^{39}\ \mathrm{W}$ . For a thin disc at the Schwarzschild efficiency ceiling, what accretion rate does this need, and how does it compare with fusion?

Step 1. Efficiency relation. A disc radiates at $L=\eta\dot M c^2$ , so $\dot M=L/(\eta c^2)$ with $\eta=1-2\sqrt2/3=0.0572$ .

Step 2. Insert numbers. $\eta c^2=0.0572\times9.0\times10^{16}\ \mathrm{J/kg}=5.1\times10^{15}\ \mathrm{J/kg}$ , so $\dot M=10^{39}/5.1\times10^{15}=\mathbf{1.9\times10^{23}\ kg/s}$ .

Step 3. In solar masses per year. Multiply by $3.156\times10^{7}\ \mathrm{s/yr}$ and divide by $M_\odot=1.99\times10^{30}$ kg: $\dot M=1.9\times10^{23}\times3.156\times10^{7}/1.99\times10^{30}=\mathbf{3.1\ M_\odot/yr}$ . A bright quasar devours a few Suns of gas each year.

Step 4. Contrast with fusion. Fusion liberates only $\approx0.007$ of rest mass, so it would need $\approx\mathbf{25\ M_\odot/yr}$ — eight times the fuel. Accretion down to the ISCO is the most efficient steady power source known, which is why quasars, not fusion, dominate the luminous universe. The same physics gives a fiducial gravitational-wave frequency $f_{\mathrm{GW}}\approx4.4\,\mathrm{kHz}\times(M_\odot/M)$ — about $73$ Hz for a $60\,M_\odot$ binary, in LIGO's band, marking the inspiral-to-plunge handoff.

Physics of the portrait

The ISCO belongs to the era when relativists turned Schwarzschild's and Kerr's geometries into working astrophysics — computing exactly how orbits behave near a black hole and what that means for the light we see.

Origins · 1 minJames Bardeen and how this was discovered

James Bardeen

1939–2022 · the relativistic theory of black-hole orbits and accretion efficiency

James Maxwell Bardeen — son of John Bardeen, the only person to win two Nobel Prizes in physics — became one of the deepest relativists of his generation. In the 1972 paper "Rotating Black Holes: Locally Nonrotating Frames, Energy Extraction, and Scalar Synchrotron Radiation," written with William Press and Saul Teukolsky, he laid out the geodesic structure of the Kerr metric in full: the innermost stable circular orbit as a function of spin, its binding energy, and hence the efficiency of accretion — from $5.7\%$ for a non-rotating hole up to $42\%$ for a maximally rotating one. That calculation turned the ISCO from a curiosity of Schwarzschild geometry into a quantitative diagnostic of black-hole spin. Bardeen's reach went further: with Brandon Carter and Stephen Hawking he formulated the four laws of black-hole mechanics that seeded black-hole thermodynamics, and in cosmology he introduced the gauge-invariant "Bardeen potentials" behind the modern theory of cosmological perturbations. The ISCO efficiencies that now let astronomers weigh a black hole's spin from its X-ray glow are, in large part, his legacy.

History

1916

— Schwarzschild publishes the exterior metric whose geodesics contain the last stable orbit

1963

— Kerr finds the rotating black-hole solution, making the ISCO depend on spin and orbit sense

1969

— Lynden-Bell proposes that accretion discs around massive black holes power quasars

1972

— Bardeen, Press & Teukolsky compute the ISCO and its binding energy across all Kerr spins

1973

— Novikov & Thorne and Shakura & Sunyaev build the thin-disc model whose efficiency is fixed by the ISCO

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973), §25.5 · Wald, General Relativity (1984), §6.3 · Carroll, Spacetime and Geometry (2004), §5.4 · Hartle, Gravity (2003), Ch. 9 · Schutz, A First Course in General Relativity, 2nd ed. (2009), §11.1 · Shapiro & Teukolsky, Black Holes, White Dwarfs and Neutron Stars (1983), §12.4 · Bardeen, Press & Teukolsky, ApJ 178, 347 (1972).

Provenance. The effective potential, ISCO radius, and efficiency are standard Schwarzschild results. A verifier should: (1) reproduce $V_{\mathrm{eff}}^2=(1-2GM/c^2r)(c^2+\tilde L^2/r^2)$ from the timelike-geodesic normalisation and the two Killing conserved quantities (Carroll §5.4, Hartle Ch. 9); (2) confirm $r_{\mathrm{ISCO}}=6GM/c^2$ and $\tilde L_{\mathrm{ISCO}}=\sqrt{12}\,GM/c$ by solving $W'=W''=0$ (MTW §25.5, Wald §6.3); (3) verify $\tilde E_{\mathrm{ISCO}}=\tfrac{2\sqrt2}{3}c^2=0.9428c^2$ and $\eta=1-\tfrac{2\sqrt2}{3}=0.0572$ (Shapiro & Teukolsky §12.4); (4) check the numerical results $r_{\mathrm{ISCO}}(\mathrm{Sgr\,A^*})=3.8\times10^{7}$ km $=0.25$ AU with a $33$ -min period, quasar accretion $3.1\,M_\odot/\mathrm{yr}$ at $10^{39}$ W, $f_{\mathrm{GW}}\approx4.4\,\mathrm{kHz}\times(M_\odot/M)$ ; and (5) the Kerr limits $r_{\mathrm{ISCO}}=GM/c^2$ (prograde, $\eta=1-1/\sqrt3\approx42\%$ ) and $9GM/c^2$ (retrograde, $\eta\approx3.8\%$ ) against Bardeen, Press & Teukolsky (1972). No non-standard values are asserted.

Prerequisites. The Schwarzschild Metric · Effective Potential for Massive-Particle Orbits · Circular Orbits and Their Stability · Killing Vectors and Conserved Quantities in Schwarzschild · Geodesics in Schwarzschild Spacetime. Leads to. Perihelion Precession in Schwarzschild Geometry · Photon Orbits and the Photon Sphere · Gravitational Redshift in Schwarzschild Spacetime.

Ask the tutor: Ask me to walk through any step — from why the $1/r^3$ term creates the ISCO, to how the $5.7\%$ efficiency lights up a quasar, or how black-hole spin slides the ISCO between $GM/c^2$ and $9GM/c^2$ .

Why does the ISCO exist in general relativity but not in Newtonian gravity?

It comes down to one extra term. Newton's potential $-GM/r+L^2/2r^2$ always turns up steeply at small $r$ because the centrifugal barrier $+L^2/2r^2$ diverges faster than gravity, so every circular orbit sits in a well and is stable. General relativity adds an attractive $-GML^2/(c^2r^3)$ piece that dominates at small radius and pulls the potential back down; as the angular momentum is reduced the minimum and neighbouring maximum annihilate at $r=6GM/c^2$ , and below that no stable circular orbit exists — a feature with no Newtonian counterpart.

Is the ISCO the same as the event horizon or the photon sphere?

No — they are three distinct radii, all fixed multiples of $GM/c^2$ . The event horizon sits at $2GM/c^2$ , the unstable photon sphere (last circular orbit for light) at $3GM/c^2$ , and the ISCO (last stable orbit for massive particles) at $6GM/c^2$ — twice the photon sphere, three times the horizon. Between the photon sphere and the ISCO circular orbits for matter still exist but are all unstable; only outside $6GM/c^2$ can matter settle into a lasting circular track, which is why accretion discs end there and not at the horizon.

Where does the 5.7% accretion efficiency actually come from?

It is a binding energy. A particle at rest far from the hole has $\tilde E=c^2$ ; on the ISCO it has $\tilde E_{\mathrm{ISCO}}=\tfrac{2\sqrt2}{3}c^2\approx0.943c^2$ . In a thin disc the gas drifts inward through nearly circular orbits, radiating the energy difference between neighbouring radii until it reaches the ISCO; the total shed is the deficit $1-0.943=0.057$ , or $5.7\%$ of rest energy. Inside the ISCO there is no stable orbit, so the gas plunges across the horizon carrying its remaining $94.3\%$ largely unradiated. The $5.7\%$ is thus a ceiling on how much accreted mass a Schwarzschild disc can turn into light.

How does black-hole spin change the ISCO?

Dramatically, and in a direction that depends on which way you orbit. For a Kerr hole the ISCO depends on the spin $a$ and on whether the orbit is prograde (co-rotating) or retrograde. As the spin rises from zero to the extremal $a=GM/c^2$ , the prograde ISCO shrinks from $6GM/c^2$ down to $GM/c^2$ , while the retrograde ISCO grows to $9GM/c^2$ . Because binding energy tracks the ISCO, the efficiency swings from $5.7\%$ (Schwarzschild) up to $\approx42\%$ (prograde) or down to $\approx3.8\%$ (retrograde). This strong, monotonic dependence is exactly what lets astronomers read a black hole's spin from the inner edge of its disc.

What happens to matter once it crosses inside the ISCO?

It plunges. Just inside $6GM/c^2$ the potential has no minimum, so with no restoring force a gas element follows a nearly radial, nearly geodesic path to the horizon in a short proper time. This plunge is fast compared with the slow viscous inspiral outside and radiates very little further energy — which is why the ISCO is the effective inner edge of the disc and why the disc spectrum cuts off at a temperature set by the ISCO radius. In the standard thin-disc model the plunging region is treated as a torque-free boundary, an idealisation that works because so little happens there.

Is r = 6GM/c² a real distance you could measure with a ruler?

Not directly — it is a coordinate radius, not a proper distance. The Schwarzschild $r$ is defined so a sphere at fixed $r$ has area $4\pi r^2$ (the areal radius), but radial ruler distance is stretched by $(1-2GM/c^2r)^{-1/2}>1$ , so the proper distance from the horizon to the ISCO exceeds the coordinate gap $\Delta r=4GM/c^2$ . What is observable is not ruler distance but quantities like the orbital frequency $\Omega=\sqrt{GM/r^3}$ and the redshift of light from the orbit — which is how the ISCO is actually located in data. " $6GM/c^2$ " is a precise statement about the areal coordinate, not tape-measure distance.