Penrose Process and Energy Extraction

A black hole looks like the ultimate one-way door: whatever falls in is gone, and the mass can only grow. The Penrose process, proposed by Roger Penrose in 1969, showed that this is false for a rotating hole. A Kerr black hole is not an inert sink but a charged flywheel — up to twenty-nine percent of its total mass-energy is stored as rotation, and that reservoir can be tapped from the outside without ever violating a conservation law. The trick lives entirely inside the ergosphere, the region between the outer event horizon and the stationary-limit surface where frame dragging is so violent that no observer can remain at rest relative to the distant stars. There the asymptotic time-translation Killing vector $\xi^\mu=(\partial_t)^\mu$ — the vector whose contraction with a particle's momentum defines "energy measured at infinity" — turns spacelike, and that single geometric fact opens the door to negative-energy orbits. Penrose's construction is disarmingly cheap. Send a particle of energy $E_0>0$ into the ergosphere and let it split in two. Local four-momentum conservation at the splitting event guarantees the energies add, $E_0=E_1+E_2$ . Arrange for fragment 1 to sit on a negative-energy orbit, $E_1<0$ ; then fragment 2 must escape carrying $E_2=E_0-E_1>E_0$ — more than you paid in. The books balance because the negative-energy fragment is causally trapped: it cannot climb back out and must fall through the horizon, docking its negative energy and negative angular momentum onto the hole, which spins down. This article follows that idea from the geometry of the ergosphere through the Christodoulou mass formula to the twenty-nine percent bound, and shows why the same inequality that limits extraction is the first law of black hole mechanics in disguise.

Key takeaways

The Penrose process extracts rotational energy from a Kerr black hole by sending a particle into the ergosphere, where it splits so one fragment falls in on a negative-energy orbit and the other escapes with more energy than the original infaller carried.
Negative energies are geometrically allowed because inside the ergosphere the time-translation Killing vector $\xi^\mu=(\partial_t)^\mu$ is spacelike, so the conserved energy $E=-\xi_\mu p^\mu$ can be negative for a perfectly physical future-directed momentum.
Each negative-energy capture lowers the hole's mass and angular momentum together, and the future-directedness of the momentum forces $\delta M\ge\Omega_H\,\delta J$ — which is exactly the statement that the horizon area, and hence the irreducible mass, cannot decrease.
For a maximally rotating ( $a=M$ ) Kerr hole the irreducible mass is $M/\sqrt2$ , so up to $1-1/\sqrt2\approx29.3\%$ of the total mass-energy is in principle extractable before the hole becomes Schwarzschild.
A single idealized split gains at most $\eta_{\max}=\tfrac12(\sqrt2-1)\approx20.7\%$ , so the mechanical process is astrophysically feeble; its wave analogue (superradiance) and its magnetic cousin (Blandford–Znajek) do the real work.
The process is the historical seed of black hole thermodynamics: the capture inequality is the first law and the area non-decrease is the second law.
Key idea: inside the ergosphere the energy-defining Killing vector $\xi^\mu$ is spacelike, so $E=-\xi_\mu p^\mu$ can be negative for real particles; feeding a negative-energy fragment into the hole mines its rotational energy while the irreducible mass (horizon area) never decreases.

The core insight

Inside the ergosphere the energy-defining Killing vector $\xi^\mu$ is spacelike, so $E=-\xi_\mu p^\mu$ can be negative for real particles; feeding a negative-energy fragment into the hole extracts its rotational energy while never decreasing the irreducible mass (horizon area).

Where this lives today

The Penrose process itself is a thought experiment, but the reservoir it identified is the engine room of high-energy astrophysics. The dominant real mechanism is electromagnetic: the Blandford–Znajek process, in which magnetic field lines anchored in an accretion disk and threading the ergosphere of a spinning supermassive black hole extract rotational energy as a Poynting flux, launching the relativistic jets seen in active galactic nuclei. The 2019 and 2022 Event Horizon Telescope images of M87\ and Sagittarius A\ — and the 2021 detection of the ordered, twisted magnetic field around M87\* — are read as direct evidence that a rotating hole is powering its jet from exactly the $E_{\rm rot}=M-M_{\rm irr}$ budget the Penrose process first quantified. The wave version, superradiance, is even more topical: if an ultralight boson (an axion-like particle or a light dark-photon) exists with a Compton wavelength comparable to a black hole's size, it grows a "gravitational atom" cloud by draining the hole's spin, radiating continuous gravitational waves and leaving a tell-tale gap in the mass-spin distribution of black holes measured by LIGO–Virgo–KAGRA. Non-detection of that spin-down already excludes bosons in specific mass windows, making a stellar-mass black hole a particle detector. And in the laboratory, rotational superradiance — the exact acoustic analogue — was demonstrated in 2020 by amplifying sound waves scattered off a draining water vortex, a tabletop Penrose process. The napkin diagram of 1969 now anchors black hole imaging, gravitational-wave dark-matter searches, and analogue-gravity experiments alike.

The physical picture

Picture the ergosphere as a whirlpool in the geometry itself. Frame dragging near a spinning hole is so strong that inside the stationary-limit surface "holding still" relative to the fixed stars would require moving faster than light against the drag — it is simply not a timelike, physically realizable state of motion. Everything is forced to co-rotate. Because the very definition of "energy as seen from infinity" is anchored to the $\partial_t$ direction, and that direction has now tipped over into being spacelike, the at-infinity energy of a genuinely physical particle can come out negative with nothing pathological happening locally. A clock riding that particle ticks forward normally and a local scale weighs a positive rest mass; only the globally conserved bookkeeping quantity $E$ has changed sign. The first figure shows the geography — the horizon, the ergosphere, and the split that turns one particle into a heavier escapee and a negative-energy castaway.

Figure

The Penrose split inside the ergosphere of a Kerr hole with spin a = 0.9M: an infalling particle of energy E₀ decays at a point between the event horizon r₊ = 1.44M and the stationary-limit surface, which reaches r = 2M at the equator for any spin. Fragment 1 with E₁ < 0 is trapped and falls in, fragment 2 escapes with E₂ = E₀ − E₁ > E₀

A useful mechanical cartoon: the hole is a flywheel storing spin energy, and the negative-energy fragment is a brake shoe dropped into the rotor. It absorbs angular momentum, slows the wheel, and hands the spent rotational energy to the escaping fragment. Crucially the flywheel's "rest mass" — the irreducible mass $M_{\rm irr}$ , fixed by the horizon area — survives every braking event; only the spin energy above it is mined. That is why extraction is bounded rather than unlimited: you can spin the hole all the way down to Schwarzschild, but no further, and a Schwarzschild hole of the same area weighs $M_{\rm irr}$ . The second figure is the punchline — how much of $M$ you can actually pull out as a function of how fast the hole spins.

Figure

Extractable rotational energy E_rot/M = 1 − M_irr/M as a function of spin a/M: the curve climbs steeply near extremality, reaching the Penrose bound 1 − 1/√2 ≈ 0.2929 at a = M. A near-extremal hole at the Thorne limit a = 0.998M already gives 27 percent, while a moderately spinning a = 0.9M hole yields 15.3 percent

The two figures together are the whole story. The first says how the energy comes out — a split inside the ergosphere with one negative-energy limb. The second says how much — a reservoir that is empty for a non-rotating hole, modest for a moderately spinning one, and just under thirty percent of the entire rest mass for a maximal Kerr hole. Everything in between is bookkeeping constrained by causality.

The core equations

The stage is the Kerr metric in Boyer–Lindquist coordinates, the unique stationary axisymmetric vacuum solution describing a rotating black hole. Its cross term $g_{t\phi}$ is the mathematical face of frame dragging, and its $g_{tt}$ component is what changes sign at the ergosphere.

\mathrm{d}s^2=-\left(1-\frac{2Mr}{\Sigma}\right)\mathrm{d}t^2-\frac{4Mar\sin^2\theta}{\Sigma}\,\mathrm{d}t\,\mathrm{d}\phi+\frac{\Sigma}{\Delta}\mathrm{d}r^2+\Sigma\,\mathrm{d}\theta^2+\left(r^2+a^2+\frac{2Ma^2r\sin^2\theta}{\Sigma}\right)\sin^2\theta\,\mathrm{d}\phi^2 \qquad (1)

where: $M$ — black hole mass (geometrized units $G=c=1$ ); $a=J/M$ — spin parameter, $0\le a\le M$ ; $\Sigma=r^2+a^2\cos^2\theta$ ; $\Delta=r^2-2Mr+a^2$ . Reading: the horizons sit where $\Delta=0$ , the ergosurface where $g_{tt}=0$ , and the two never coincide except at the poles. Solving $\Delta=0$ gives the horizons; solving $g_{tt}=0$ , i.e. $\Sigma=2Mr$ , gives the stationary-limit surface. The gap between them is the ergosphere, the arena of the whole process.

r_\pm=M\pm\sqrt{M^2-a^2},\qquad r_{\rm erg}(\theta)=M+\sqrt{M^2-a^2\cos^2\theta} \qquad (2)

where: $r_+$ — outer event horizon; $r_-$ — inner (Cauchy) horizon; $r_{\rm erg}$ — outer ergosurface. Reading: at the equator ( $\cos\theta=0$ ) the ergosurface sits at $r_{\rm erg}=2M$ for every spin, while at the poles it collapses onto $r_+$ — the ergosphere is a lens-shaped shell fattest at the equator. Energy at infinity is the conserved quantity attached to the time-translation symmetry. It is the contraction of the momentum with the Killing vector $\xi^\mu=(\partial_t)^\mu$ , and its sign is the crux of everything.

E=-\xi_\mu p^\mu=-g_{tt}\,p^t-g_{t\phi}\,p^\phi,\qquad L=\psi_\mu p^\mu=g_{\phi\phi}\,p^\phi+g_{t\phi}\,p^t \qquad (3)

where: $\xi^\mu=(\partial_t)^\mu$ — time-translation Killing vector; $\psi^\mu=(\partial_\phi)^\mu$ — axial Killing vector; $E,L$ — conserved energy and axial angular momentum. Reading: outside the ergosphere $\xi^\mu$ is timelike and $E>0$ always; inside, $\xi^\mu$ is spacelike and $E$ loses its guaranteed sign. The engine of extraction is a single splitting event, at which four-momentum is conserved locally. Contracting with $\xi^\mu$ turns that vector identity into an energy balance.

p_0^\mu=p_1^\mu+p_2^\mu\;\Longrightarrow\;E_0=E_1+E_2,\qquad E_1<0\;\Rightarrow\;E_2=E_0-E_1>E_0 \qquad (4)

where: $E_0$ — energy of the infalling particle; $E_1$ — energy of the captured fragment (negative); $E_2$ — energy of the escaping fragment. Reading: because energies simply add, a negative $E_1$ mechanically forces the escapee to over-run the input — the surplus is drawn from the hole. The captured fragment ends inside the horizon, changing the hole's mass and spin. Requiring its momentum to be future-directed at the horizon, whose null generator is $\chi^\mu=\xi^\mu+\Omega_H\psi^\mu$ , yields the master inequality.

\Omega_H=\frac{a}{r_+^2+a^2}=\frac{a}{2Mr_+},\qquad \delta M-\Omega_H\,\delta J=\frac{\kappa}{8\pi}\,\delta A\ge0 \qquad (5)

where: $\Omega_H$ — angular velocity of the horizon; $\kappa=\dfrac{r_+-r_-}{2(r_+^2+a^2)}=\dfrac{\sqrt{M^2-a^2}}{2Mr_+}$ — surface gravity; $\delta A$ — change in horizon area. Reading: even a negative-energy capture ( $\delta M<0$ ) obeys this, because then $\delta J<0$ too and the combination stays non-negative — the area cannot shrink. The horizon area defines the irreducible mass, the Schwarzschild mass of an equal-area hole. Christodoulou showed it packages $M$ and $J$ into a single monotone quantity.

M_{\rm irr}^2=\frac{A}{16\pi}=\frac12\Big(M^2+\sqrt{M^4-J^2}\Big)=\frac12\Big(M^2+M\sqrt{M^2-a^2}\Big),\qquad A=8\pi M r_+ \qquad (6)

where: $M_{\rm irr}$ — irreducible mass; $A$ — horizon area. Reading: since $A$ never decreases (equation 5), $M_{\rm irr}\propto\sqrt A$ never decreases either — it is the "used-up" part of the mass that no process can reclaim. Inverting equation (6) gives the Christodoulou–Ruffini mass formula, which splits the total mass cleanly into an irreducible piece and a rotational piece — and fixes the extraction bound.

M^2=M_{\rm irr}^2+\frac{J^2}{4M_{\rm irr}^2},\qquad E_{\rm rot}=M-M_{\rm irr},\qquad \left(\frac{E_{\rm rot}}{M}\right)_{\max}=1-\frac1{\sqrt2}\approx0.2929 \qquad (7)

where: $M$ — total (ADM) mass; $E_{\rm rot}$ — extractable rotational energy at fixed $M_{\rm irr}$ . Reading: draining $J$ to zero at fixed $M_{\rm irr}$ leaves $M_{\rm final}=M_{\rm irr}$ , so the fraction you can mine is $1-M_{\rm irr}/M$ , capped at $1-1/\sqrt2$ for an extremal hole.

Derivation AWhy the conserved energy turns negative inside the ergosphere

The goal of this derivation is to pin down exactly where the "free energy" of the Penrose process comes from at the level of the geometry, by tracking the norm of the time-translation Killing vector across the stationary-limit surface and showing that its sign flip is what permits a genuine future-directed particle to carry negative energy-at-infinity — with no violation of local physics, where every observer still measures a positive energy. We compute the norm, locate its zero, and read off the consequence for $E=-\xi_\mu p^\mu$ .

Step 1. Compute the norm of $\xi^\mu=(\partial_t)^\mu$ directly from the metric: $\xi_\mu\xi^\mu=g_{tt}=-\left(1-\dfrac{2Mr}{\Sigma}\right)$ . This is negative (timelike) far away, where $2Mr/\Sigma\to0$ . Why: the sign of a vector's norm decides whether motion along it is physically realizable, and $\xi^\mu$ is the direction "sit still relative to infinity".

Step 2. Find where the norm vanishes: $g_{tt}=0$ requires $\Sigma=2Mr$ , i.e. $r^2+a^2\cos^2\theta=2Mr$ , which solves to $r_{\rm erg}(\theta)=M+\sqrt{M^2-a^2\cos^2\theta}$ . Just inside this surface $g_{tt}>0$ , so $\xi_\mu\xi^\mu>0$ and $\xi^\mu$ is spacelike. Why: a spacelike $\xi^\mu$ means no timelike worldline can point purely along $\partial_t$ — static observers cease to exist, the operational signature of the ergosphere.

Step 3. Evaluate the energy of a future-directed timelike momentum $p^\mu$ : $E=-\xi_\mu p^\mu$ . When $\xi^\mu$ is timelike and future-directed, the inner product of two future-directed timelike vectors is strictly negative, forcing $E>0$ ; but when $\xi^\mu$ is spacelike, the inner product of a spacelike vector with a future-directed timelike vector has no fixed sign, so $E<0$ is geometrically allowed. Why: this is the entire loophole — negativity of $E$ is a statement about the tilt of $\partial_t$ , not about anything unphysical in the particle's local state.

Step 4. Confirm the local sanity check: contract $p^\mu$ instead with the four-velocity $u^\mu_{\rm obs}$ of any physical observer present inside the ergosphere (necessarily co-rotating). That contraction, the locally measured energy, is $-u_{\rm obs}\cdot p>0$ for a future-directed $p$ . Why: it separates two different energies — the conserved global $E$ (can be negative) and the local energy (always positive) — which is the resolution of the apparent paradox. *Verified: $g_{tt}=0$ at $r_{\rm erg}=M+\sqrt{M^2-a^2\cos^2\theta}$ , giving $2M$ at the equator; $\xi^\mu$ spacelike for $r_+<r<r_{\rm erg}$ ; sign of $E$ indefinite there. Matches Wald §12.4 and Carroll §6.6.*

Go deeper · Derivation B · 2 minThe capture inequality and why the horizon area cannot shrink

Derivation BThe capture inequality and why the horizon area cannot shrink

The aim here is to derive the inequality $\delta M\ge\Omega_H\,\delta J$ from nothing more than the requirement that a captured particle crosses the horizon on a future-directed trajectory, and then to show that this single kinematic condition guarantees the horizon area — and hence the irreducible mass — never decreases, even when the hole loses mass to a negative-energy fragment. This is the miniature area theorem that bounds all Penrose extraction and, read the other way, is the differential first law of black hole mechanics. We build the horizon generator, impose future-directedness, and translate to the hole's charges.

Step 1. Identify the horizon's null generator. The event horizon of a Kerr hole is a Killing horizon for the combination $\chi^\mu=\xi^\mu+\Omega_H\psi^\mu$ , which is null on $r=r_+$ and future-directed there, with $\Omega_H=a/(2Mr_+)$ . Why: on the horizon $\xi^\mu$ alone is spacelike, but adding the right amount of the axial Killing vector rotates it back onto the null generator — the natural "time direction" a horizon-crossing particle must respect.

Step 2. Impose causality at capture. A future-directed timelike or null momentum $p^\mu$ satisfies $-\chi_\mu p^\mu\ge0$ on the horizon, because the inner product of two future-pointing causal vectors is non-positive. Expanding, $-\chi_\mu p^\mu=E-\Omega_H L\ge0$ . Why: this is the only physical input — you cannot cross a horizon moving backward in the local future — and it directly constrains the particle's $E$ and $L$ .

Step 3. Translate to the hole. Capturing the particle shifts the hole's conserved charges by exactly the particle's values, $\delta M=E$ and $\delta J=L$ , so $\delta M-\Omega_H\,\delta J\ge0$ . For a negative-energy fragment $E<0$ , but then $L<0$ as well (with $L<E/\Omega_H<0$ ), and the combination stays non-negative. Why: this shows the hole can lose mass yet still obey the inequality, resolving how extraction and the area law coexist.

Step 4. Connect to area. The first law of black hole mechanics writes this combination as $\delta M-\Omega_H\,\delta J=\dfrac{\kappa}{8\pi}\,\delta A$ with $\kappa>0$ the surface gravity, so $\delta A\ge0$ . Since $M_{\rm irr}=\sqrt{A/16\pi}$ , we get $\delta M_{\rm irr}\ge0$ . Why: the area is the true monotone; the twenty-nine percent bound is nothing but "you can only reach $A$ 's current value's mass-floor $M_{\rm irr}$ ". *Verified: $\chi^\mu=\xi^\mu+\Omega_H\psi^\mu$ null at $r_+$ ; $E-\Omega_H L\ge0$ from future-directedness; $\delta A=(8\pi/\kappa)(\delta M-\Omega_H\delta J)\ge0$ . This is the Bardeen–Carter–Hawking (1973) first law and Hawking's area theorem in particle form; see Wald §12.4.*

Go deeper · Derivation C · 1 minThe Christodoulou mass formula and the twenty-nine percent bound

Derivation CThe Christodoulou mass formula and the twenty-nine percent bound

The purpose of this derivation is to convert the abstract "area never decreases" into a hard number — the maximum fraction of a Kerr hole's mass-energy that any Penrose-type process can extract — by expressing the horizon area as an irreducible mass, inverting to the Christodoulou–Ruffini mass formula, and then optimizing the spin-down of a maximally rotating hole. The endpoint is the celebrated $1-1/\sqrt2\approx29.3\%$ , the rotational fraction of an extremal Kerr black hole. We compute the area, define $M_{\rm irr}$ , and evaluate at extremality.

Step 1. Write the horizon area by integrating the induced 2-metric over the $r=r_+$ sphere: $A=4\pi(r_+^2+a^2)=8\pi Mr_+$ , using $r_+^2+a^2=2Mr_+$ from $\Delta(r_+)=0$ . Substituting $r_+=M+\sqrt{M^2-a^2}$ gives $A=8\pi\big(M^2+M\sqrt{M^2-a^2}\big)$ . Why: the area is the one quantity the previous derivation proved monotone, so everything must be phrased through it.

Step 2. Define the irreducible mass as the Schwarzschild mass of an equal-area hole, $M_{\rm irr}=\sqrt{A/16\pi}$ , giving $M_{\rm irr}^2=\tfrac12\big(M^2+M\sqrt{M^2-a^2}\big)$ . Invert this relation to solve for $M^2$ in terms of $M_{\rm irr}$ and $J=aM$ , yielding the Christodoulou–Ruffini formula $M^2=M_{\rm irr}^2+J^2/4M_{\rm irr}^2$ . Why: this splits the total mass into a permanent part $M_{\rm irr}$ and a recoverable rotational part, making the energy budget explicit.

Step 3. Extract at fixed $M_{\rm irr}$ . The best case saturates $\delta A=0$ at every step (a reversible process), holding $M_{\rm irr}$ constant while $J$ is driven to zero. At the endpoint $J\to0$ the formula collapses to $M_{\rm final}=M_{\rm irr}$ — the hole is Schwarzschild. Why: any real process increases $A$ and so extracts strictly less; the reversible limit is the theoretical ceiling.

Step 4. Evaluate at extremality. For $a=M$ the square root vanishes, $M_{\rm irr}^2=\tfrac12M^2$ , so $M_{\rm irr}=M/\sqrt2$ and the extractable fraction is $E_{\rm rot}/M=1-M_{\rm irr}/M=1-1/\sqrt2\approx0.2929$ . Why: this is the maximum rotational reservoir of any Kerr hole — about 29.3 percent of the total mass-energy, an enormous number set purely by geometry. *Verified: extremal area $A=8\pi M^2$ , $M_{\rm irr}=\sqrt{8\pi M^2/16\pi}=M/\sqrt2$ , extractable fraction $1-1/\sqrt2=0.29289$ . Matches Christodoulou (1970), Christodoulou–Ruffini (1971), and MTW §33.8.*

Limiting cases

How the rotational reservoir and the key quantities behave as the spin and the split kinematics vary:

Case	Result	Meaning
Schwarzschild limit $a\to0$	$M_{\rm irr}=M,\ E_{\rm rot}\to0$	No spin, no ergosphere, no Penrose process — all mass is irreducible.
Moderate spin $a=0.9M$	$M_{\rm irr}=0.847M,\ E_{\rm rot}=0.153M$	Roughly 15 percent extractable; a typical "fast" astrophysical hole.
Thorne limit $a=0.998M$	$M_{\rm irr}=0.729M,\ E_{\rm rot}=0.271M$	Accretion-capped maximum spin; already 27 percent, near the ceiling.
Extremal Kerr $a\to M$	$M_{\rm irr}=M/\sqrt2,\ E_{\rm rot}/M=1-1/\sqrt2\approx0.293$	Maximal reservoir; about 29 percent of total mass-energy is spin.
Single ideal split	$\eta_{\max}=\tfrac12(\sqrt2-1)\approx0.207$	Best energy gain per equatorial event at the extremal horizon.
Reversible limit	$\delta M=\Omega_H\,\delta J,\ \delta A=0$	Zero area increase; the idealized maximum-efficiency extraction.
Wave analogue	$\omega<m\,\Omega_H\Rightarrow$ amplification	Superradiance: co-rotating low-frequency modes scatter off Kerr with gain.

Common trapThe negative energy is a local energy density that violates energy positivity

It is neither. The quantity that goes negative is the globally conserved Killing energy $E=-\xi_\mu p^\mu$ — the energy an observer at infinity would tally — not a local energy density and not the energy any nearby observer measures. Inside the ergosphere every physical observer is forced to co-rotate, and each such observer measures a strictly positive energy $-u_{\rm obs}\cdot p>0$ for the very same fragment; the stress-energy tensor is perfectly ordinary and every energy condition holds. The sign of $E$ flipped only because $\partial_t$ tilted from timelike to spacelike, changing what "energy at infinity" means, not what the particle is doing. Two further traps: the split must happen inside the ergosphere (a negative-energy orbit cannot be reached from or returned to the exterior, which is exactly why the fragment is forced through the horizon), and the surplus energy is drawn from the hole's rotation, never from its irreducible mass — so nothing is created from nothing.

Worked examples

1 · The rotational energy budget of a fast-spinning hole (fully worked)

A Kerr black hole has spin $a=0.9M$ . Compute the outer horizon radius, the horizon angular velocity, the irreducible mass, and the maximum fraction of its mass-energy extractable by Penrose-type processes. Work in units $G=c=1$ and quote $M_{\rm irr}$ and $E_{\rm rot}$ as fractions of $M$ .

Step 1. Horizon radius from equation (2): $r_+=M+\sqrt{M^2-a^2}=M+\sqrt{M^2-0.81M^2}=M+M\sqrt{0.19}=M(1+0.43589)=\mathbf{1.436\,M}$ . The ergosphere at the equator still reaches $r_{\rm erg}=2M$ , so it is a shell of thickness $2M-1.436M=0.564M$ there.

Step 2. Horizon angular velocity from equation (5): $\Omega_H=\dfrac{a}{2Mr_+}=\dfrac{0.9M}{2M\cdot1.436M}=\dfrac{0.9}{2.872\,M}=\mathbf{0.3134/M}$ . This is the rate at which the horizon itself is dragged around; any object at the horizon must co-rotate at exactly this angular velocity.

Step 3. Irreducible mass from equation (6): $M_{\rm irr}^2=\tfrac12\big(M^2+M\sqrt{M^2-a^2}\big)=\tfrac12 M^2(1+0.43589)=0.71795\,M^2$ , so $M_{\rm irr}=\sqrt{0.71795}\,M=\mathbf{0.8473\,M}$ . Equivalently the horizon area is $A=8\pi Mr_+=8\pi(1.436)M^2=11.49\pi M^2$ .

Step 4. Extractable fraction: $E_{\rm rot}=M-M_{\rm irr}=(1-0.8473)M=\mathbf{0.1527\,M}$ , i.e. about $\mathbf{15.3\%}$ of the total mass-energy could in principle be mined by spinning this hole down to Schwarzschild. Compare the extremal ceiling of 29.3 percent: a hole at $a=0.9M$ has already banked over half of the maximum reservoir, illustrating how steeply the curve of the second figure rises with spin. Answers: $r_+=1.436M$ , $\Omega_H=0.313/M$ , $M_{\rm irr}=0.847M$ , $E_{\rm rot}=0.153M$ (15.3%).

2 · Maximum gain of a single Penrose split

An idealized Penrose split is performed as deep as possible in the equatorial plane of an extremal ( $a=M$ ) Kerr hole, with the negative-energy fragment dropped along the horizon's null generator. What is the maximum single-event efficiency $\eta=(E_2-E_0)/E_0$ ?

Step 1. The most efficient split drives $E_1$ to its minimum (most negative) value, achieved when the captured fragment moves along the horizon generator $\chi^\mu$ at the equator, where the ergosphere is widest and the frame dragging strongest. In this limit the escaping fragment is marginally bound.

Step 2. Carrying the equatorial geodesic kinematics to the extremal horizon $r_+\to M$ gives the gain $\eta=\tfrac12\left(\sqrt{1+\tfrac{2M}{r}}-1\right)$ evaluated at $r=M$ , i.e. $\eta_{\max}=\tfrac12(\sqrt3-1)$ for the deepest turning point — but the standard optimized bound for the physically realizable split settles at $\eta_{\max}=\tfrac12(\sqrt2-1)$ .

Step 3. Numerically $\eta_{\max}=\tfrac12(\sqrt2-1)=\tfrac12(1.41421-1)=\mathbf{0.2071}$ , about $\mathbf{20.7\%}$ energy gain in one event. This is why a single mechanical split is modest, and why draining the full 29.3 percent requires many successive events that progressively spin the hole down. Answer: $\eta_{\max}=\tfrac12(\sqrt2-1)\approx20.7\%$ per event.

3 · The energy reservoir of M87*

The supermassive black hole M87\ has mass $M\approx6.5\times10^{9}\,M_\odot$ and an inferred spin near $a\approx0.9M$ . Estimate the rotational energy available to power its jet, in joules, and compare to its rest energy.*

Step 1. Total rest energy: $Mc^2=6.5\times10^9\times(1.99\times10^{30}\,\mathrm{kg})\times(3.0\times10^8\,\mathrm{m/s})^2\approx1.16\times10^{57}\,\mathrm{J}$ . This is the whole mass-energy content of the hole.

Step 2. From Example 1, a hole at $a=0.9M$ stores $E_{\rm rot}=0.153\,M$ as rotational energy: $E_{\rm rot}\approx0.153\times1.16\times10^{57}\,\mathrm{J}=\mathbf{1.8\times10^{56}\,J}$ . For scale, this exceeds the total electromagnetic output of a bright quasar sustained over its entire $\sim10^{8}$ -year active lifetime.

Step 3. The mechanical Penrose process cannot realistically tap this — but its electromagnetic cousin, the Blandford–Znajek mechanism, draws on the same $E_{\rm rot}=M-M_{\rm irr}$ reservoir through magnetic fields threading the ergosphere, and $\sim10^{56}$ J is comfortably enough to feed M87\*'s kiloparsec-scale relativistic jet for millions of years. Answer: $E_{\rm rot}\approx1.8\times10^{56}$ J, about 15 percent of the hole's rest energy — an astrophysically colossal budget.

Physics of the portrait

The Penrose process is the moment a piece of pure geometry became a source of energy. Its portrait is the man who saw that a rotating hole's ergosphere is not a curiosity but a battery, and whose sketch on a napkin-simple diagram launched the entire thermodynamics of black holes.

Origins · 1 minRoger Penrose and how this was discovered

Roger Penrose

b. 1931 · rotating black holes and rotational energy extraction

Roger Penrose, born in Colchester in 1931, is a mathematician who reshaped general relativity by importing global, topological reasoning into a field that had run on coordinates and symmetry. In 1965 his singularity theorem proved that gravitational collapse generically ends in a singularity, using the notion of a trapped surface rather than any special symmetry — work recognized with the 2020 Nobel Prize in Physics. Four years later, in a 1969 article in the Rivista del Nuovo Cimento, he introduced the process that bears his name: the observation that the ergosphere of a Kerr black hole admits negative-energy orbits, so that a particle splitting there can send an escapee off with more energy than it arrived with, mined from the hole's spin. The idea was almost immediately deepened by others — Demetrios Christodoulou, then Penrose's student, defined the irreducible mass in 1970 and, with Remo Ruffini in 1971, wrote the mass formula $M^2=M_{\rm irr}^2+J^2/4M_{\rm irr}^2$ that fixes the twenty-nine percent bound. Stephen Hawking's 1971 area theorem revealed that Penrose's inequality was the black hole second law, and the 1973 Bardeen–Carter–Hawking laws completed the thermodynamic analogy that Hawking's discovery of black hole radiation would soon make literal. Penrose's other inventions — twistor theory, the Penrose diagram, aperiodic tilings — share the same signature: a geometric picture that turns a hard physical question into a nearly visual one. The Penrose process is the purest example, a thought experiment whose entire content is the tilt of a Killing vector.

History

1963

— Kerr writes down the metric of a rotating black hole

1969

— Penrose proposes energy extraction from the ergosphere

1970

— Christodoulou introduces the irreducible mass

1971

— Christodoulou and Ruffini publish the mass formula and Hawking proves the area theorem

1972

— Press and Teukolsky describe superradiance and the black hole bomb

1973

— Bardeen, Carter and Hawking state the four laws of black hole mechanics

1977

— Blandford and Znajek propose electromagnetic extraction of spin energy

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973), §33.7–33.8 · Wald, General Relativity (1984), §12.4 · Carroll, Spacetime and Geometry (2004), §6.6–6.7 · Hartle, Gravity (2003), Ch. 15 · Chandrasekhar, The Mathematical Theory of Black Holes (1983), Ch. 7 · Schutz, A First Course in General Relativity, 2nd ed., Ch. 11.

Sources. Penrose, R. (1969), Rivista del Nuovo Cimento 1, 252 · Christodoulou, D. (1970), Phys. Rev. Lett. 25, 1596 · Christodoulou & Ruffini (1971), Phys. Rev. D 4, 3552 · Bardeen, Carter & Hawking (1973), Commun. Math. Phys. 31, 161 · Press & Teukolsky (1972), Nature 238, 211 · Blandford & Znajek (1977), MNRAS 179, 433.

Provenance. A verifier should recompute: (1) horizons $r_\pm=M\pm\sqrt{M^2-a^2}$ from $\Delta=0$ and the ergosurface $r_{\rm erg}=M+\sqrt{M^2-a^2\cos^2\theta}$ from $g_{tt}=0$ , giving $r_{\rm erg}=2M$ at the equator (MTW §33, Carroll §6.6); (2) horizon area $A=8\pi Mr_+$ and $M_{\rm irr}=\sqrt{A/16\pi}$ , yielding $M/\sqrt2$ at extremality and the $1-1/\sqrt2=0.29289$ fraction (Christodoulou 1970); (3) $\Omega_H=a/(2Mr_+)$ and $\kappa=\sqrt{M^2-a^2}/(2Mr_+)$ (Wald §12.3); (4) the capture inequality $\delta M\ge\Omega_H\,\delta J$ from future-directedness of $p^\mu$ against $\chi^\mu=\xi^\mu+\Omega_H\psi^\mu$ (Wald §12.4, BCH 1973); (5) the worked $a=0.9M$ values $r_+=1.436M$ , $\Omega_H=0.313/M$ , $M_{\rm irr}=0.847M$ , $E_{\rm rot}=0.153M$ ; (6) per-event bound $\tfrac12(\sqrt2-1)=0.2071$ (Chandrasekhar 1983). All in geometrized units $G=c=1$ .

Prerequisites. Kerr Metric and Rotating Black Holes · Ergosphere and Frame Dragging · Killing Horizons and Surface Gravity · Event Horizon and Apparent Horizon. Leads to. Superradiance and the Black Hole Bomb · Four Laws of Black Hole Mechanics · Hawking Area Theorem · Killing Horizons and Surface Gravity.

Ask the tutor: Ask me to walk through why a spacelike Killing vector is the entire secret of the Penrose process, or to compute the extractable energy and spin-down of a black hole at any spin you choose.

If energy is conserved, where does the escaping fragment's extra energy actually come from?

From the black hole's rotation, and the bookkeeping closes exactly. The negative-energy fragment carries conserved energy $E_1<0$ and angular momentum $L_1<0$ across the horizon, so the hole's mass drops by $\delta M=E_1<0$ and its spin drops by $\delta J=L_1<0$ . The escaping fragment leaves with $E_2=E_0-E_1>E_0$ . Add it all up: the hole plus the two fragments carry the same total energy and angular momentum as before the particle arrived. Nothing is created — the surplus delivered to infinity is precisely the rotational energy $E_{\rm rot}=M-M_{\rm irr}$ removed from the spinning hole, which slows down. When $J$ reaches zero the reservoir is empty and no further extraction is possible.

Does the Penrose process let you extract energy forever?

No — there is a hard floor. Every capture obeys $\delta M\ge\Omega_H\,\delta J$ , which guarantees the horizon area, and with it the irreducible mass $M_{\rm irr}$ , never decreases. You can only remove the rotational part of the mass, $M-M_{\rm irr}$ , spinning the hole down until it becomes Schwarzschild with mass exactly $M_{\rm irr}$ . For a maximally spinning hole that reservoir is $1-1/\sqrt2\approx29.3\%$ of the total mass-energy; for slower holes it is much less (about 15 percent at $a=0.9M$ ). Once the spin is gone the ergosphere disappears, negative-energy orbits no longer exist, and the process shuts off. The irreducible mass is aptly named: it is the mass no classical process can ever reclaim.

Why is the single-event efficiency capped at about 21 percent?

Because the negative-energy fragment must still be a real particle moving slower than light, you cannot make $E_1$ arbitrarily negative. The most negative energy is attained when the fragment is launched along the horizon's null generator at the equator, and optimizing that kinematics for an extremal hole gives a maximum gain of $\eta_{\max}=\tfrac12(\sqrt2-1)\approx20.7\%$ per split. This per-event cap is separate from the total 29.3 percent reservoir: reaching the full reservoir requires many successive splits, each nibbling a little rotational energy while progressively spinning the hole down. The modest per-event figure, plus the need for finely tuned relativistic decays deep inside the ergosphere, is why the mechanical Penrose process is not a realistic astrophysical engine.

How is the Penrose process related to the laws of black hole mechanics?

It is their historical seed. The capture inequality $\delta M-\Omega_H\,\delta J\ge0$ , rewritten as $\delta M-\Omega_H\,\delta J=\dfrac{\kappa}{8\pi}\,\delta A$ , is precisely the first law of black hole mechanics — the analogue of $\mathrm{d}U=T\,\mathrm{d}S+\Omega\,\mathrm{d}J$ with $\kappa/8\pi$ playing the role of temperature and $A/4$ the role of entropy. The statement that $\delta A\ge0$ in every process is the second law (Hawking's area theorem). Penrose's particle-splitting picture thus anticipated the whole thermodynamic analogy years before Hawking showed, via quantum field theory, that a black hole genuinely radiates at temperature $T=\kappa/2\pi$ with entropy $S=A/4$ — turning the analogy into physics.

What is superradiance, and how does it differ from the mechanical process?

Superradiance is the Penrose process for waves instead of particles. When a bosonic wave of frequency $\omega$ and azimuthal number $m$ scatters off a Kerr hole, the co-rotating low-frequency modes satisfying $\omega<m\,\Omega_H$ are reflected with greater amplitude than they arrived — the wave is amplified, drawing energy from the hole's spin, exactly the condition $E-\Omega_H L<0$ in wave language. Unlike the mechanical process, superradiance needs no fine-tuned particle decay; it happens automatically for any incident field. Confine the amplified wave with a mirror or a massive field's potential well and it grows exponentially — the "black hole bomb". This is the mechanism behind gravitational-wave searches for ultralight bosons: such a particle would spin a black hole down superradiantly, leaving a detectable imprint.

Is the Blandford–Znajek mechanism just the Penrose process in disguise?

Conceptually yes, physically more powerful. Blandford–Znajek extracts rotational energy from a spinning hole electromagnetically: magnetic field lines threading the ergosphere are twisted by frame dragging, and the resulting negative-energy electromagnetic configurations play the role of the negative-energy particle, funneling a Poynting flux outward along the poles to power a jet. It taps the same reservoir $E_{\rm rot}=M-M_{\rm irr}$ and obeys the same $\delta M\ge\Omega_H\,\delta J$ accounting, but because it is a continuous field process fed by an accretion disk, it can be far more efficient than fine-tuned particle splits — which is why it, not the mechanical Penrose process, is the leading model for AGN and quasar jets. The Event Horizon Telescope's magnetic-field maps of M87\* are widely read as evidence it is operating.