Innermost Stable Circular Orbit

In Newtonian gravity a test particle can circle a point mass at any radius: the centrifugal wall $+L^2/2r^2$ always beats gravity as $r\to0$ , so for every angular momentum there is a stable circular orbit and Keplerian disks extend inward without limit. General relativity breaks this promise. The Schwarzschild geodesic equation carries an extra attractive term falling off as $-1/r^3$ , steeper than the centrifugal barrier, and near a black hole that term wins. Below a finite radius the stable minimum of the potential and the unstable maximum guarding it rush together and annihilate. The radius where they meet — where a circular orbit is marginally stable — is the innermost stable circular orbit, the ISCO. For a non-rotating hole the ISCO sits at $r_{\mathrm{ISCO}}=6GM/c^2$ , exactly three Schwarzschild radii and twice the photon-sphere radius — a distinct surface between the outer world of ordinary orbits and the inner world of inevitable plunge. Inside it, circular orbits still exist, but they are unstable: the slightest nudge sends the particle spiralling to the horizon on a dynamical time. This one radius does an enormous amount of astrophysical work. It sets the inner edge of a thin accretion disk, the deepest gravitational well from which infalling gas can still radiate before it plunges. The binding energy at the ISCO fixes the maximum efficiency of accretion — a few percent for a static hole, up to 42% for a maximal spin — which is why accreting black holes outshine every other persistent engine in the universe, and why measuring the ISCO measures black-hole spin.

Key takeaways

The innermost stable circular orbit (ISCO) is the smallest radius at which a test particle holds a stable circular orbit around a compact object; inside it, circular orbits exist but are unstable and plunge inward.
For a Schwarzschild (non-rotating) black hole the ISCO is at $r_{\mathrm{ISCO}}=6GM/c^2=3\,r_s$ , twice the photon-sphere radius and three times the horizon radius.
It exists because GR adds an attractive $-1/r^3$ term to the effective potential; below the ISCO this term overwhelms the centrifugal barrier and the stable minimum vanishes.
Spin moves it: the prograde ISCO shrinks to $GM/c^2$ for an extremal Kerr hole, the retrograde ISCO grows to $9GM/c^2$ , so it encodes both mass and angular momentum.
The ISCO binding energy fixes the maximum radiative efficiency: $\eta\approx5.7\%$ for Schwarzschild, up to $\approx42\%$ at extremal prograde spin — dwarfing hydrogen fusion's $0.7\%$ .
ISCO radii are read observationally through continuum fitting and broadened Fe Kα lines, turning the accretion-disk inner edge into a spin measurement.
Key idea: in general relativity the effective potential for a massive particle carries an extra $-1/r^3$ term absent in Newtonian gravity; this term overwhelms the centrifugal barrier at small radii, so the stable circular-orbit minimum disappears below a finite radius — the ISCO — inside which every circular orbit is unstable.

The core insight

In general relativity the effective potential for a massive particle includes an extra $-1/r^3$ term absent in Newtonian gravity; this term overwhelms the centrifugal barrier at small radii, so the stable circular-orbit minimum disappears below a finite radius — the ISCO — inside which all circular orbits are unstable.

Where this lives today

The ISCO is now a measured quantity, and measuring it is how black-hole spins are weighed. Two X-ray techniques dominate. Continuum fitting takes the thermal disk spectrum of a stellar-mass hole, extracts the inner-disk radius, identifies it with the ISCO, and — given mass, distance, and inclination — reads off $a_*$ from the Bardeen–Press–Teukolsky curve. Relativistic reflection spectroscopy instead models the broad, skewed Fe Kα line near $6.4\ \mathrm{keV}$ , whose red wing records how deep the disk reaches, for stellar-mass and supermassive holes alike; RXTE, XMM-Newton, NuSTAR, and the polarimeter IXPE have made these routine spin surveys. Meanwhile the Event Horizon Telescope images the shadow of M87\ and Sgr A\ — a different surface, the same strong-field regime — and LIGO–Virgo–KAGRA watch binaries merge, where the remnant's ISCO marks the inspiral-to-plunge transition. Even high-frequency QPOs in some X-ray binaries hover near the predicted ISCO frequency, hinting that we time gas at the innermost stable orbit directly. The one radius GR adds to Newtonian dynamics has become a standard ruler for the strongest gravity we can observe.

The physical picture

Picture an orbit as a marble rolling in a bowl whose profile is the effective potential $V_{\mathrm{eff}}^2(r)$ . In Newtonian gravity the bowl always has a dip — a stable circular orbit — for any angular momentum, because the centrifugal wall curls upward faster than gravity curls down. General relativity bends the inner wall downward through an extra $-1/r^3$ term. For a marble carrying plenty of angular momentum the dip survives, flanked by an outward-facing hump; as the angular momentum falls the dip and hump slide together, the well grows shallow, and at a critical value they merge into a single flat inflection. That inflection is the ISCO. A step further in and the bowl tilts monotonically into the hole — no dip, no barrier, only plunge. The first figure is that family of bowls, from the real Schwarzschild potential.

Figure

The general-relativistic effective potential Veff²(r) for three angular momenta near a Schwarzschild black hole. For large L̃ a stable minimum and an unstable maximum bracket a potential well; as L̃ falls the well shrinks; at the critical value L̃ = 2√3 GM/c the two extrema merge into a flat inflection at r = 6GM/c², the ISCO — the last radius that still holds a stable orbit

Spin changes the story quantitatively. A rotating hole drags spacetime with it: a prograde orbit is carried along and balances closer in before the well collapses, so its ISCO shrinks, while a retrograde orbit fights the dragging and sits farther out. The dependence is sharp — as the spin $a_*=cJ/GM^2$ runs from $-1$ to $+1$ , the prograde ISCO sweeps from $9GM/c^2$ down to $GM/c^2$ . Because the deepest reachable orbit fixes the binding energy gas can shed, that same curve is a curve of maximum efficiency — the second figure is the one an observer inverts to weigh a spin.

Figure

The ISCO radius versus black-hole spin from the Bardeen–Press–Teukolsky formula. The prograde branch falls from 6GM/c² at a = 0 to 1GM/c² at extremal spin a = 1, while the retrograde branch rises to 9GM/c²; the shrinking prograde ISCO deepens the accessible well, so the maximum accretion efficiency climbs from about 5.7% to about 42%

The disappearance of the stable orbit is a genuine bifurcation. A circular orbit needs $V_{\mathrm{eff}}'=0$ and a stable one additionally $V_{\mathrm{eff}}''>0$ ; along the sequence of circular orbits the minimum and maximum coalesce where both vanish at once — a saddle-node bifurcation at exactly one radius, the ISCO.

The core equations

Radial motion on a Schwarzschild geodesic reduces to a one-dimensional energy problem. The conserved energy and angular momentum per unit rest mass, $\tilde E$ and $\tilde L$ , package the two Killing symmetries; what remains is a particle of "energy" $\tilde E^2$ in an effective potential.

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

where: $\tau$ — proper time; $\tilde{E}$ — conserved energy per unit rest mass, in units of $c^2$ ; $\tilde{L}$ — conserved specific angular momentum; $M$ — central mass; $r$ — Schwarzschild radial coordinate. Turning points satisfy $\tilde E^2=V_{\mathrm{eff}}^2$ ; a circular orbit sits at an extremum of $V_{\mathrm{eff}}^2$ . Multiplying out the product exposes the term that makes GR different: the first three are the Newtonian potential plus centrifugal barrier, the fourth has no Newtonian counterpart.

V_{\mathrm{eff}}^2(r)=1-\frac{2GM}{c^2 r}+\frac{\tilde{L}^2}{c^2 r^2}-\frac{2GM\tilde{L}^2}{c^4 r^3} \qquad (2)

where: $-\dfrac{2GM\tilde{L}^2}{c^4 r^3}$ — the purely relativistic attractive term, a cross-product of gravity and angular momentum, that overwhelms the $+\tilde L^2/c^2r^2$ barrier at small $r$ . This single $r^{-3}$ term is the entire reason an ISCO exists; delete it and orbits are stable at every radius, as in Newton. Imposing marginal stability — the double root of the last section — yields a clean closed form for the non-rotating case.

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

where: $r_{\mathrm{ISCO}}$ — innermost stable circular orbit radius; $r_s=2GM/c^2$ — Schwarzschild radius. It sits at three horizon radii and two photon-sphere radii, a fixed multiple of the mass, so it scales linearly with $M$ from stellar-mass to supermassive holes. The energy a particle carries on the ISCO is what it could not radiate away — the residual specific energy that sets the deepest well the disk can tap.

\tilde{E}(r)=\frac{1-2GM/c^2r}{\sqrt{1-3GM/c^2 r}}\;\xrightarrow{\ r=6GM/c^2\ }\;\tilde{E}_{\mathrm{ISCO}}=\sqrt{\frac{8}{9}}\approx 0.9428 \qquad (4)

where: $\tilde{E}(r)$ — specific energy of the circular orbit at radius $r$ , in units of $c^2$ ; $\tilde E_{\mathrm{ISCO}}$ — its ISCO value. A particle at rest at infinity has $\tilde E=1$ , so the deficit $1-\tilde E_{\mathrm{ISCO}}$ is the fraction of rest energy extractable as radiation before the plunge. That deficit is the maximum radiative efficiency of accretion — the number that makes black holes the brightest steady engines in the universe.

\eta = 1-\tilde{E}_{\mathrm{ISCO}} = 1-\sqrt{\tfrac{8}{9}}\approx 0.0572 \qquad (5)

where: $\eta$ — fraction of accreted rest-mass energy $mc^2$ radiated on the way to the ISCO. This $5.7\%$ dwarfs hydrogen fusion's $0.7\%$ , so gravity out-earns the strong force by nearly an order of magnitude. For a rotating hole the ISCO follows the Bardeen–Press–Teukolsky formula, exact for equatorial circular geodesics of the Kerr metric.

r_{\mathrm{ISCO}}=\frac{GM}{c^2}\left[3+Z_2\mp\sqrt{(3-Z_1)(3+Z_1+2Z_2)}\right] \qquad (6)

where: $Z_1=1+(1-a_*^2)^{1/3}\big[(1+a_*)^{1/3}+(1-a_*)^{1/3}\big]$ ; $Z_2=\sqrt{3a_*^2+Z_1^2}$ ; $a_*=cJ/GM^2\in[-1,1]$ ; the lower sign is prograde, the upper retrograde. At $a_*=0$ both give $6GM/c^2$ ; at $a_*=+1$ prograde gives $GM/c^2$ , retrograde $9GM/c^2$ . Finally, the coordinate angular velocity of a circular geodesic — the frequency an outside clock times — keeps its Keplerian form in Schwarzschild coordinates, with a frame-dragging correction for Kerr.

\Omega=\frac{\mathrm{d}\phi}{\mathrm{d}t}=\pm\frac{\sqrt{GM/r^3}}{1\pm a_* (GM/c^2 r)^{3/2}} \qquad (7)

where: $\Omega$ — orbital angular velocity seen at infinity; $\pm$ selects prograde ( $+$ ) or retrograde ( $-$ ). For Schwarzschild ( $a_*=0$ ) this is simply $\Omega=\sqrt{GM/r^3}$ ; at the ISCO it gives the highest coherent orbital frequency a hole of mass $M$ can imprint on its disk.

The core derivations

Derivation ALocating the Schwarzschild ISCO from the effective potential

The goal is to prove that the last stable circular orbit around a non-rotating black hole sits at $r_{\mathrm{ISCO}}=6GM/c^2$ , using only the Schwarzschild effective potential and two conditions: that the orbit be circular and that it be marginally stable — the radius where the stable minimum and unstable maximum of $V_{\mathrm{eff}}^2$ coalesce. We work in geometrized units $G=c=1$ , restoring dimensions at the end; the squared potential is $V^2(r)=(1-2M/r)(1+L^2/r^2)$ with $L$ the specific angular momentum.

Step 1. A circular orbit is an extremum of the potential, so set the first derivative to zero: $\dfrac{\mathrm{d}V^2}{\mathrm{d}r}=\dfrac{2M}{r^2}+L^2\left(-\dfrac{2}{r^3}+\dfrac{6M}{r^4}\right)=0$ . Why: an extremum of $V^2$ is precisely the condition for a circular (constant- $r$ ) orbit.

Step 2. Multiply through by $r^4/2$ and solve for the angular momentum that makes radius $r$ circular: $L^2=\dfrac{Mr^2}{r-3M}$ . Why: this fixes the one free constant; note $L^2>0$ demands $r>3M$ , the photon sphere, below which no massive circular orbit exists.

Step 3. Stability is decided by the second derivative: $\dfrac{\mathrm{d}^2V^2}{\mathrm{d}r^2}=-\dfrac{4M}{r^3}+L^2\left(\dfrac{6}{r^4}-\dfrac{24M}{r^5}\right)$ , positive for a minimum (stable), zero at the marginal case. Why: a minimum restores a nudged particle, an inflection does not — the ISCO is the inflection where the minimum becomes a maximum.

Step 4. Set that second derivative to zero and substitute $L^2=Mr^2/(r-3M)$ from Step 2: $-\dfrac{4M}{r^3}+\dfrac{Mr^2}{r-3M}\left(\dfrac{6}{r^4}-\dfrac{24M}{r^5}\right)=0$ . Why: one $L$ must make the orbit both circular and marginally stable — two conditions, one radius.

Step 5. Clear the common factor $M/r^3$ and combine over $r-3M$ : the bracket collapses to $-4(r-3M)+6r-24M=0$ , i.e. $-4r+12M+6r-24M=0$ . Why: once common factors cancel, the transcendental-looking condition reduces to a single linear equation in $r$ .

Step 6. Solve: $2r-12M=0\Rightarrow r=6M$ . Restoring units gives $r_{\mathrm{ISCO}}=6GM/c^2$ , with $L^2=M(6M)^2/(6M-3M)=12M^2$ , so $\tilde L_{\mathrm{ISCO}}=2\sqrt{3}\,GM/c$ — the critical value at which the well disappears. Why: a unique positive root confirms exactly one marginally stable radius, the boundary between the stable disk outside and the plunge region inside. *The algebra yields a unique positive root $r=6M$ , with $\tilde L_{\mathrm{ISCO}}^2=12$ in $M$ -units — confirming exactly one marginally stable radius.*

Go deeper · Derivation B · 1 minThe radiative efficiency of Schwarzschild accretion

Derivation BThe radiative efficiency of Schwarzschild accretion

The goal is to compute the maximum fraction of an accreted particle's rest-mass energy that can be radiated before it plunges, and to show it equals $\eta=1-\sqrt{8/9}\approx5.7\%$ . In the standard thin disk, gas drifts inward through a near-continuous sequence of circular orbits, radiating the binding-energy difference at each step, until at the ISCO stable orbits end and it falls in almost freely across the horizon. The efficiency is therefore fixed entirely by the specific energy of the ISCO orbit relative to rest at infinity.

Step 1. The specific energy of a circular orbit at radius $r$ follows from inserting $L^2=Mr^2/(r-3M)$ into $\tilde E^2=V^2$ , giving $\tilde E(r)=\dfrac{1-2M/r}{\sqrt{1-3M/r}}$ . Why: on a circular orbit the energy is set by the radius alone, so $\tilde E$ becomes a pure function of $r$ along the disk.

Step 2. Evaluate at the ISCO, $r=6M$ : $\tilde E(6M)=\dfrac{1-2/6}{\sqrt{1-3/6}}=\dfrac{2/3}{\sqrt{1/2}}=\dfrac{2\sqrt2}{3}=\sqrt{\dfrac{8}{9}}\approx0.9428$ . Why: the ISCO is the last orbit that radiates, so its specific energy is the floor of the accessible well — everything deeper is lost.

Step 3. A particle begins at rest at infinity with $\tilde E=1$ , so the energy it sheds to reach the ISCO is $E_{\mathrm{rad}}=(1-\tilde E_{\mathrm{ISCO}})mc^2$ . Why: by conservation, the radiated total is the drop in specific energy times the rest mass.

Step 4. The efficiency is that fraction: $\eta=1-\sqrt{8/9}=1-0.94281\ldots=0.0572$ , i.e. $5.7\%$ . Why: against fusion's $\sim0.7\%$ this shows why accretion powers quasars; the same argument at the extremal-Kerr ISCO, where $\tilde E_{\mathrm{ISCO}}=1/\sqrt3$ , gives $\eta=1-1/\sqrt3\approx42\%$ . *The numbers are $\tilde E(6M)=0.94281$ and $\eta=0.05719$ ; the same argument at the extremal-prograde ISCO gives $1-1/\sqrt3=0.4226\approx42\%$ .*

Go deeper · Derivation C · 1 minThe orbital frequency at the ISCO and the fastest clock a black hole carries

Derivation CThe orbital frequency at the ISCO and the fastest clock a black hole carries

The goal is to turn the ISCO radius into an observable frequency — the rate at which inner-edge gas circles the hole, seen by a distant astronomer — and to show it scales inversely with mass, roughly $2.2\ \mathrm{kHz}$ per solar mass. This is what a high-frequency quasi-periodic oscillation would trace if it marks orbital motion at the ISCO, connecting a geometric radius to a timing signal in an X-ray light curve. We use the Schwarzschild coordinate angular velocity, evaluated at $r=6GM/c^2$ .

Step 1. For Schwarzschild circular geodesics the coordinate angular velocity keeps its Keplerian form exactly, $\Omega=\mathrm{d}\phi/\mathrm{d}t=\sqrt{GM/r^3}$ . Why: despite the curved geometry, $\mathrm{d}\phi/\mathrm{d}t$ collapses to the Newtonian relation in these coordinates — a coincidence special to Schwarzschild, not Kerr.

Step 2. Substitute $r=6GM/c^2$ : $\Omega_{\mathrm{ISCO}}=\sqrt{\dfrac{GM}{(6GM/c^2)^3}}=\dfrac{c^3}{6^{3/2}\,GM}=\dfrac{c^3}{GM\sqrt{216}}$ . Why: the frequency now depends only on $M$ , since $GM/c^3$ is the natural time unit of a black hole.

Step 3. Convert to ordinary frequency $f=\Omega/2\pi$ and insert $GM_\odot/c^3=4.925\times10^{-6}\ \mathrm{s}$ : $f_{\mathrm{ISCO}}=\dfrac{c^3}{2\pi\sqrt{216}\,GM}=\dfrac{2199\ \mathrm{Hz}}{M/M_\odot}$ . Why: the $1/M$ scaling means a $10\,M_\odot$ hole tops out near $220\ \mathrm{Hz}$ and a $4\times10^6\,M_\odot$ hole near $0.5\ \mathrm{mHz}$ — one formula from binaries to galactic nuclei, with spin (eq. 7) lifting the ceiling further. *The result is $f_{\mathrm{ISCO}}=2199\ \mathrm{Hz}\,(M_\odot/M)$ ; for a $10\,M_\odot$ hole this gives $f=220\ \mathrm{Hz}$ and an orbital period $T=4.5\ \mathrm{ms}$ , squarely in the kHz range observed for compact accretors.*

Limiting cases

The ISCO radius depends on the spin parameter $a_*=cJ/GM^2\in[-1,1]$ ; the endpoints and the Newtonian limit bracket the whole physical range.

Case	Result	Meaning
Schwarzschild, a = 0	$r_{\mathrm{ISCO}}=\dfrac{6GM}{c^2}=3\,r_s$	Non-rotating hole, ISCO at three Schwarzschild radii; binding-energy fraction $1-\sqrt{8/9}\approx0.057$ .
Extremal prograde Kerr, a = +1	$r_{\mathrm{ISCO}}=\dfrac{GM}{c^2}=r_g$	Co-rotating orbit hugs the horizon (also at $r_g$ for extremal Kerr); efficiency up to $1-1/\sqrt3\approx0.42$ .
Extremal retrograde Kerr, a = −1	$r_{\mathrm{ISCO}}=\dfrac{9GM}{c^2}$	Counter-rotating orbit pushed out by frame dragging; lowest efficiency $\approx0.038$ .
Thorne limit, a = 0.998	$r_{\mathrm{ISCO}}\approx\dfrac{1.24\,GM}{c^2}$	Maximum spin realistically reached by accretion once photon capture is included; $\eta\approx0.32$ .
Newtonian / large-r	$r_{\mathrm{ISCO}}\to 0$	No ISCO: every radius supports a stable circular orbit, recovering Keplerian disks.

Common trapThe ISCO is the horizon, or the photon sphere, or "the last orbit"

It is none of these. For Schwarzschild the ISCO sits at $6GM/c^2=3r_s$ , well outside both the horizon at $2GM/c^2$ and the photon sphere at $3GM/c^2$ . And "no stable orbit inside the ISCO" does not mean "no orbit at all": unstable circular orbits persist from the ISCO down to the photon sphere, where the innermost circular orbit of any kind — a photon's — finally sits. A massive particle can circle at $4GM/c^2$ ; it simply cannot stay there, since the tiniest perturbation grows and it spirals in. A subtler point: the ISCO is a coordinate radius, and the disk "inner edge" is a physical assumption — plunging gas stops radiating — that observations support but geometry alone does not force.

Worked examples

1 · The ISCO of a stellar-mass black hole (fully worked)

A black hole of mass $M=10\,M_\odot$ is accreting. Treat it as non-rotating. Find the ISCO radius in kilometres, the orbital frequency of gas at the ISCO, and the luminosity produced if it accretes at $\dot M=10^{-8}\,M_\odot\,\mathrm{yr}^{-1}$ .

Step 1. The gravitational radius is $GM/c^2=1.477\ \mathrm{km}\times(M/M_\odot)=14.77\ \mathrm{km}$ . The ISCO is six of these: $r_{\mathrm{ISCO}}=6\times14.77\ \mathrm{km}=\mathbf{88.6\ km}$ , with the horizon at $2\times14.77=29.5\ \mathrm{km}$ — the disk's inner edge stands well outside the hole.

Step 2. Orbital frequency at the ISCO: $f_{\mathrm{ISCO}}=2199\ \mathrm{Hz}\times(M_\odot/M)=2199/10=\mathbf{220\ Hz}$ , period $T=1/f=\mathbf{4.5\ ms}$ — gas whips around two hundred times a second, the timescale of the X-ray's high-frequency variability.

Step 3. The accretion rate in SI is $\dot M=10^{-8}\times1.989\times10^{30}/(3.156\times10^{7})=6.3\times10^{14}\ \mathrm{kg\,s^{-1}}$ . With $\eta=0.057$ the luminosity is $L=\eta\dot Mc^2=0.057\times6.3\times10^{14}\times(3\times10^8)^2$ .

Step 4. Evaluating: $L=0.057\times6.3\times10^{14}\times9\times10^{16}=\mathbf{3.2\times10^{30}\ W}\approx8\times10^{3}\,L_\odot$ . A hundredth of a millionth of a solar mass per year already outshines thousands of Suns, purely on binding energy released down to the ISCO. r_ISCO = 88.6 km, f_ISCO = 220 Hz, L ≈ 3×10³⁰ W.

2 · How fast must a quasar feed? (efficiency in reverse)

A luminous quasar radiates $L=10^{39}\ \mathrm{W}$ ( $10^{46}\ \mathrm{erg\,s^{-1}}$ ). Assuming a moderately spinning hole with efficiency $\eta=0.1$ , find the required mass accretion rate, and compare with what nuclear fusion would demand.

Step 1. Invert $L=\eta\dot Mc^2$ : $\dot M=L/(\eta c^2)=10^{39}/(0.1\times9\times10^{16})=1.1\times10^{23}\ \mathrm{kg\,s^{-1}}$ , the rest-mass supply the ISCO efficiency turns into light.

Step 2. In astronomers' units: $\dot M=1.1\times10^{23}\times(3.156\times10^{7})/(1.989\times10^{30})=\mathbf{1.8\ M_\odot\,yr^{-1}}$ — the quasar swallows roughly two Suns a year.

Step 3. From hydrogen fusion at $\eta_{\mathrm{fus}}=0.007$ the required rate would scale up by $0.1/0.007\approx14$ , to $\approx25\,M_\odot\,\mathrm{yr}^{-1}$ — and, more damningly, fusion cannot make this luminosity from a compact region at all, since it needs stellar interiors, not an accretion flow.

Step 4. The ISCO efficiency is what makes the numbers work: extracting $10\%$ of rest energy against fusion's $0.7\%$ , a black hole powers the brightest objects in the universe on a few solar masses a year. The Soltan argument turns this around — total quasar light divided by the mass in today's black holes implies an average $\eta\approx0.1$ , moderately spinning holes. Ṁ ≈ 1.8 M_⊙/yr; the ISCO efficiency, not fusion, sets the budget.

Physics of the portrait

The ISCO has no single discoverer — it falls out of the Schwarzschild geodesics of 1916 — but its spinning generalization and its promotion into a working tool of astrophysics belong to one careful relativist.

Origins · 1 minJames Bardeen and how this was discovered

James Bardeen

1939–2022 · rotating black holes and relativistic astrophysics

James Maxwell Bardeen — son of the two-time Nobel laureate John Bardeen — spent his career on the exact structure of rotating black holes and the matter around them, and it is largely through his work that the ISCO became a number observers could measure rather than a curiosity of the Schwarzschild solution. The static result $r_{\mathrm{ISCO}}=6GM/c^2$ is implicit in Schwarzschild's 1916 metric and was made explicit when Yusuke Hagihara worked out the full geodesic structure in 1931. The story then stalled for decades, because the astrophysically relevant holes spin and Kerr's rotating solution did not arrive until 1963. In 1972 Bardeen, with William Press and Saul Teukolsky, published the definitive analysis of equatorial circular orbits in Kerr spacetime — the closed-form ISCO of equation (6), its prograde and retrograde branches, and the efficiency curve running to $42\%$ . The next year he joined Novikov and Thorne in the relativistic thin-disk model that made the ISCO the physical inner edge of an accretion flow. Bardeen also gave the field its foundational black-hole thermodynamics (the "four laws," with Carter and Hawking) — but for accretion astrophysics his legacy is that a measured disk inner edge is now read directly as a spin, exactly as his 1972 formulae prescribe.

History

1916

— Schwarzschild publishes the vacuum solution, implicitly fixing the last stable orbit at 6GM/c²

1931

— Hagihara works out the complete Schwarzschild geodesic structure and identifies the marginally stable orbit

1963

— Kerr finds the exact metric of a rotating black hole

1972

— Bardeen, Press and Teukolsky derive the equatorial ISCO of Kerr, its energy, and its efficiency

1973

— Novikov, Thorne, Shakura and Sunyaev make the ISCO the radiating inner edge of the thin disk

Going further

Texts. Carroll & Ostlie, An Introduction to Modern Astrophysics, ch. 17 & 18 (Schwarzschild orbits and accretion) · Misner, Thorne & Wheeler, Gravitation, §25 (orbits in Schwarzschild) · Shapiro & Teukolsky, Black Holes, White Dwarfs, and Neutron Stars, ch. 12 & 14 · Binney & Tremaine, Galactic Dynamics, orbits in spherical potentials · Bardeen, Press & Teukolsky (1972), ApJ 178, 347 (Kerr circular orbits). The key results collected here are: (1) the circular-orbit condition $\tilde L^2=Mr^2/(r-3M)$ and marginal stability together give $r=6M$ uniquely, with $\tilde L_{\mathrm{ISCO}}=2\sqrt3\,GM/c$ ; (2) $\tilde E_{\mathrm{ISCO}}=\sqrt{8/9}\approx0.9428$ and $\eta=1-\sqrt{8/9}\approx0.0572$ ; (3) the Kerr endpoints from the Bardeen–Press–Teukolsky formula are $r=GM/c^2$ (extremal prograde) and $9GM/c^2$ (retrograde), with efficiencies $1-1/\sqrt3\approx0.42$ and $\approx0.038$ ; (4) $f_{\mathrm{ISCO}}=2199\ \mathrm{Hz}\,(M_\odot/M)$ from $\Omega=\sqrt{GM/r^3}$ evaluated at $6GM/c^2$ .

Prerequisites. Schwarzschild Metric · Effective Potential for Orbits · Neutron Stars · Tolman–Oppenheimer–Volkoff Equation. Leads to. Accretion Disks and Radiative Efficiency · Black-Hole Spin Measurement · Pulsars.

Ask the tutor: Ask me to walk through the marginal-stability algebra that pins the ISCO at 6GM/c², or to show how a measured Fe Kα line width becomes a black-hole spin.

Is the ISCO the same thing as the event horizon?

No, and confusing them is the most common error. For a non-rotating hole the horizon sits at $2GM/c^2$ , while the ISCO is three times farther out at $6GM/c^2$ . The horizon is the causal boundary — once inside, no signal escapes — whereas the ISCO is a dynamical boundary for circular orbits: outside it a small nudge leaves a particle on a nearby stable orbit, inside it the nudge grows and the particle spirals in. Gas exists between the two; it simply cannot orbit there stably, so it plunges. The radii coincide only in the extremal-Kerr limit, and even then they are conceptually distinct surfaces.

Do orbits simply not exist inside the ISCO?

Circular orbits do exist inside the ISCO — they are just unstable. Between the ISCO at $6GM/c^2$ and the photon sphere at $3GM/c^2$ a massive particle can travel on a circular path, but it is balanced on a knife edge: any perturbation grows and, usually, sends it plunging to the horizon. Below the photon sphere even that is impossible, because the innermost circular orbit of anything — including a photon — is the photon sphere itself. So the sequence is: stable orbits outside $6GM/c^2$ , unstable ones between $3$ and $6\,GM/c^2$ , none inside $3GM/c^2$ . Instability, not nonexistence, is what changes at the ISCO.

Why does the ISCO set the efficiency of accretion?

Because the disk radiates by letting gas sink deeper into the potential well, and the ISCO is the deepest point from which it can still radiate before falling in. Spiralling inward through stable orbits, gas sheds binding energy as light at each step; at the ISCO the stable orbits run out and the gas plunges nearly freely across the horizon, carrying its leftover energy with it. So the radiated total per unit mass is fixed by the ISCO binding energy, $\tilde E_{\mathrm{ISCO}}$ relative to rest at infinity. A smaller ISCO (higher spin) means a deeper well, which is why efficiency climbs from $5.7\%$ at zero spin to $42\%$ at extremal prograde spin.

How does black-hole spin move the ISCO?

A rotating hole drags spacetime around with it — frame dragging — changing the balance an orbit must strike. A prograde orbit, co-rotating, is carried along, needs less independent support, and survives closer in; its ISCO shrinks to $GM/c^2$ at maximal spin. A retrograde orbit fights the dragging like a headwind and must sit farther out, so its ISCO grows to $9GM/c^2$ . The dependence is steep near extremal spin, which makes the ISCO a sensitive diagnostic: measure the disk's inner radius and, with the mass known, the Bardeen–Press–Teukolsky formula returns $a_*$ .

Why is the orbital frequency at the ISCO the same formula as Kepler's?

It looks like a coincidence and partly is one. In Schwarzschild coordinates the angular velocity of a circular geodesic works out to exactly $\Omega=\sqrt{GM/r^3}$ — the Newtonian Kepler form — even though the geometry is strongly curved. What is not Newtonian is where you may apply it: there are no stable circular orbits inside $6GM/c^2$ , so the frequency has a finite maximum at the ISCO, $f_{\mathrm{ISCO}}\approx2.2\ \mathrm{kHz}\,(M_\odot/M)$ , with nothing faster and stable beyond. For a spinning hole equation (7) breaks the exact Keplerian form and pushes the ceiling higher. This ISCO frequency is a leading candidate for the high-frequency QPOs seen in accreting compact objects.

Do neutron stars have an ISCO too?

Yes, in principle — the ISCO is a property of the external spacetime, not of a horizon, so any object compact enough to sit inside $6GM/c^2$ has an ISCO in the vacuum around it. For a $1.4\,M_\odot$ neutron star the ISCO is near $12\ \mathrm{km}$ , comparable to the stellar radius, so whether it lies outside the surface depends on the equation of state. If it does, the inner disk truncates there just as for a black hole, and kilohertz QPOs may be probing that orbital motion — making the ISCO a potential lever on the dense-matter equation of state, tied to the neutron-star and Tolman–Oppenheimer–Volkoff physics of this article's prerequisites.