If not even light can leave a black hole, how could radiation carry away its mass? The contradiction appears when we try to explain a quantum effect using only classical particle trajectories.
Stephen Hawking’s prediction says that quantum effects make a black hole radiate, and it can lose mass if it releases more energy than it absorbs. This is a well-supported theoretical result. However, as of the source check on 30 October 2026, there is no confirmed direct detection of Hawking radiation from an astrophysical black hole. Hawking’s original 1975 paper, APS: status in October 2026.
The horizon does not become a transparent shell
An event horizon is a boundary from inside which a future light signal cannot reach a distant outside observer. The classical description does not allow a flashlight beam to pass out through it. Hawking radiation does not mean that such a beam has found a gap.
In quantum physics, light and particles are described through fields that exist in spacetime. A quantum vacuum is a field state without particles in the relevant description, rather than an absence of the field itself. The calculation follows a field as a black hole forms: an initial vacuum can correspond to an outgoing particle flux that a distant detector would later register.
This is a semiclassical approach: quantum theory describes the fields while classical gravity describes spacetime geometry. It predicts thermal radiation, modified by how waves propagate through the surrounding geometry. It is not a complete quantum theory of gravity. Bernard Kay: the effect and limits of the particle description, section 7.
The black hole's mass pays for the outgoing energy
In the model, a distant observer sees energy flowing outward. If nothing comes in to replace it, the black hole’s energy decreases, together with its mass. Radiation does not provide free energy from emptiness. Recall how energy accounting works.
A popular picture shows a particle pair near the horizon: one falls in and the other escapes. It can suggest the link between the outward flux and the black hole’s changing energy, but it is a supporting analogy. It should not be imagined as footage of two tiny balls that briefly “borrowed” energy. In particular, antimatter does not automatically have negative energy.
The physical prediction comes from quantum fields in black-hole geometry, rather than literally tearing virtual pairs apart at a surface. Auffinger’s review: the prediction and analogies, section 3.1.3.
A smaller black hole is hotter in this model
For a black hole with no rotation or electric charge, the Hawking temperature is inversely proportional to mass. Double the mass and the temperature halves. This is the radiation temperature defined for a distant observer, rather than a thermometer reading on a solid surface: there is no such surface here.
In a simplified model of isolated evaporation, lifetime instead rises rapidly with initial mass. Compare three hypothetical cases: with a constant loss coefficient, doubling the mass gives about eight times the model lifetime. All figure values are ratios, rather than years or observations.
Model prediction: mass, temperature and time
Case 1
- Initial mass
- M₀
- Temperature
- T₀
- Model lifetime
- τ₀
Case 2
- Initial mass
- 2M₀
- Temperature
- T₀/2
- Model lifetime
- 8τ₀
Case 3
- Initial mass
- 4M₀
- Temperature
- T₀/4
- Model lifetime
- 64τ₀
As it loses mass, such a black hole heats up, strengthening further radiation. But rotation, available particle species and wave propagation alter the loss rate. The simple cubic relationship cannot be extended unchanged to every black hole or the very end of evaporation. Auffinger: temperature and mass evolution, equations 19 and 25–26.
Branch · with a formulaOptional · you can skip this
Calculate the temperature for a mass of 10³¹ kg
For an uncharged, nonrotating black hole:
T_(H) = (ℏ(c)^(3))/(8πGMk_(B)).
Here M is mass in kilograms, G is the gravitational constant, c is vacuum light speed, ℏ is Planck’s constant divided by 2π, and k_(B) is Boltzmann’s constant. The temperature T_(H) is in kelvins. This is a distant-observer quantity, rather than a local measurement close to the horizon. Cambridge lecture notes, formula D.116.
Insert the hypothetical mass M = (10)^(31) kg and recommended NIST constants. The result is approximately 1.23 × 10⁻⁸ K, or 12.3 nanokelvins. One nanokelvin is a billionth of a kelvin. At 2 × (10)^(31) kg, the result halves to about 6.1 nanokelvins.
For comparison, today’s cosmic microwave background temperature is about 2.7 K. Our model radiation is therefore extraordinarily cold against the cosmic light already present. ESA: the background radiation temperature.
Emitting and shrinking are different claims
A black hole can also absorb surrounding radiation and gas. If energy input exceeds the weak losses, its mass grows, even though the model’s quantum emission does not switch off. For stellar-mass black holes, even today’s microwave background is much warmer than their Hawking temperature.
What about the bright disk in black-hole images? It is mostly hot matter outside the horizon, radiating before it falls in. That glow is not a detection of Hawking radiation. NASA: the accretion disk, Cambridge: absorption at the end of section D.5.
Branch · model limitsOptional · you can skip this
Why does lifetime involve mass cubed?
For an isolated, nonrotating black hole, take the approximation:
(dM)/(dt) = − (a)/((M)^(2)).
The coefficient a describes the losses; for mass in kg and time in seconds, its unit is kg³/s. Here we deliberately hold it constant: the set of emitted particles and corresponding corrections do not change. Integration gives a model estimate for initial mass M_(i):
τ = ((M_(i))^(3))/(3a).
Thus masses M_(0), 2M_(0), 4M_(0) give lifetimes in the ratio 1 : 8 : 64, as in the figure. The units kg³/(kg³/s) do indeed yield seconds. Auffinger: the more general equations 25–26.
This integrates a simplified model down to zero mass; it does not prove that nature follows the entire path that way. When large quantum corrections to geometry itself become necessary, the semiclassical description becomes unreliable. This formula does not determine the exact final state.
What has been tested, and what remains a prediction?
Laboratories create horizon analogues for waves, for example in flowing ultracold atoms or an optical fibre. They allow related equations and correlations to be studied. They are not actual gravitational black holes. The primary Bose–Einstein-condensate experiment, the 2026 analogue-gravity review.
For example, an optical paper in July 2026 studied stimulated analogue radiation and its effect on a light pulse. The authors explicitly distinguish it from spontaneous quantum emission from vacuum. The result is not an observation of an astrophysical black hole losing mass. Procopio and colleagues: full text and experimental limits.
Questions remain about the final stage and information: in a complete quantum description, how are the state of the infalling matter and the state of the radiation after evaporation related? The temperature formula alone cannot answer this or establish a final flash or remnant. Kay: the information question, section 10.
Check the reasoning: if two black holes obey our model but one is twice as massive, it is colder and has a longer model lifetime. Yet emission alone does not imply that every black hole is shrinking right now: absorption also belongs in the energy balance.
The quantum prediction changes the picture of a black hole as an eternal trap while leaving the untested endpoint distinct from an observed fact. When a headline says “Hawking confirmed,” establish what was actually measured: glowing matter, a laboratory analogue, or radiation from the astrophysical black hole itself.