General Relativity and Gravitation (2025) 57:87

https://doi.org/10.1007/s10714-025-03417-x

RESEARCH

The Entropy of Black Holes

Robert M. Wald1

Received: 9 November 2024 / Accepted: 29 April 2025

© The Author(s) 2025

Abstract
The remarkable connection between black holes and thermodynamics provides the
most significant clues that we currently possess to the nature of black holes in a
quantum theory of gravity. The key clue is the formula for the entropy of a black
hole. I briefly review some recent work that provides an expression for a dynamical
correction to the entropy of a black hole and briefly discuss some of the implications
of this new formula for entropy.

More than fifty years ago, Bekenstein [1] proposed that a black hole in general relativity
should be assigned an entropy proportional to its area, A. Shortly thereafter, Bardeen,
Carter, and Hawking [2] showed that black holes satisfy laws of “black hole mechanics”
that are precise mathematical analogs of the laws of thermodynamics, with A playing
the role of entropy, S, the mass, M, of the black hole playing the role of energy, E, and
the surface gravity, κ, of the black hole playing the role of temperature, T . However, at
that time it was not consistent to identify the laws of black hole mechanics as physically
corresponding to the laws of thermodynamics because the physical temperature of a
black hole in classical general relativity is absolute zero. Soon thereafter, this situation
changed dramatically: Hawking [3] discovered that when quantum field theory effects
are considered, black holes radiate as black bodies at temperature T = κ/2π . From this
point on, it was clear that the correspondence of the laws of black hole mechanics and
the laws of thermodynamics is truly a physical correspondence, and that the entropy
assigned to a black hole must represent its true thermodynamic entropy. The formula
for the entropy of a black hole thus gives us an important clue to how black holes will
be described in a quantum theory of gravity.
Thirty years ago Iyer and I [4] derived a first law of black hole mechanics for
perturbations of a stationary black hole in an arbitrary theory of gravity obtained
from a diffeomorphism covariant Lagrangian, thereby enabling us to obtain a formula
for black hole entropy in these much more general theories. However, our derivation
required evaluation of the black hole entropy on the bifurcation surface, B, of the
black hole, thus restricting the validity of our formula for entropy to stationary black
holes and their linear perturbations, evaluated at the “time” represented by B. It is of
B Robert M. Wald
rmwa@uchicago.edu

1 Enrico Fermi Institute and Department of Physics, University of Chicago, 933 E. 56th St., Chicago,
IL 60637, USA

0123456789().: V,-vol 123

 87 Page 2 of 5 R. M. Wald

considerable interest to obtain an expression for the entropy of a non-stationary black

hole in a general theory of gravity at a “time” represented by an arbitrary cross-section,
C, since, in particular, this would allow one to investigate whether, classically, black
hole entropy satisfies a second law (i.e., whether it is non-decreasing with time) and
whether, semiclassically, black hole entropy satisfies a generalized second law (i.e.,
whether the sum of black hole entropy and a matter contribution to entropy is non-
decreasing). Iyer and I had proposed a formula for dynamical black hole entropy in a
general theory of gravity, but our formula was not field redefinition invariant, and we
retracted our proposal in a “note in proof” in the published version of our paper [4].
About ten years ago, Dong [5] proposed a formula for dynamical black hole entropy in
theories whose Lagrangian is an arbitrary function of curvature (but not derivatives of
curvature) based on holographic entanglement arguments. Shortly thereafter, Wall [6]
proposed a general prescription for dynamical black hole entropy for perturbations of
a stationary black hole. When evaluated in the case of a Lagrangian that is a function
of curvature, it was found to agree with Dong’s formula [6].
Very recently, Hollands, Zhang, and I [7] have proposed a new definition of dynami-
cal black hole entropy. As the derivation of [4] makes clear, the formula for the entropy
of a black hole evaluated at its bifurcation su4rface B is closely analogous to the for-
mula for ADM mass at spatial infinity. On the other hand, the desired properties of
black hole entropy on an arbitrary cross section C are closely analogous to the desired
properties of Bondi mass at null infinity. Zoupas and I had previously proposed a
strategy for defining Bondi mass and other BMS charges at null infinity [8]. In [7],
Hollands, Zhang, and I applied an analog of this strategy to the definition of dynamical
black hole entropy. Our resulting formula is intended to be applied only at leading non-
trivial order in perturbation theory about a stationary black hole. By its construction,
it is guaranteed to satisfy a “local physical process version” of the first law of black
hole mechanics [7].
We had no intention of modifying the Bekenstein-Hawking formula, SBH = A/4,
for the entropy of a black hole in general relativity. However, in general relativity,
our prescription yields a nontrivial dynamical correction to the Bekenstein-Hawking
entropy formula on an arbitrary cross-section, C, namely [7]

A[C] 1
SC = − Vϑ (1)
4 4 C

where V is an affine parameter of the null generators of the horizon (with V = 0

corresponding to the bifurcation surface B), and ϑ is the expansion of these generators.
For more general theories of gravity, our formula differs from the formula proposed
by Dong and Wall by a similar dynamical correction term.
An important consequence of the fact that our notion of entropy satisfies a local
physical process version of the first law of black hole mechanics is that our entropy
will change only when matter or radiation enters the black hole. This contrasts with the
Bekenstein-Hawking entropy, which increases before matter enters the black, since
the teleological nature of the event horizon requires that it must move outward in
anticipation of matter being thrown in at a later time. However, for the same reason,
the horizon will also start expanding in anticipation of the matter that will later be

123
The Entropy of Black Holes Page 3 of 5 87


thrown in. Our dynamical correction term − 41 C V ϑ exactly compensates for the area
increase (to leading order in perturbations about a stationary black hole), so that SC
does not change until matter or radiation actually enters the black hole. Furthermore,
although eq. (1) is evaluated on a cross-section C of the true event horizon, we have
shown that the entropy SC is, in fact, the area of the apparent horizon corresponding
to “time” C [7] and thus can be determined locally, without knowledge of the future
behavior of the spacetime.
Another important consequence of the fact that our notion of entropy satisfies a
local physical process version of the first law in an arbitrary theory of gravity is that
our entropy satisfies the second law—i.e., it is nondecreasing—when matter satisfying
the null energy condition is thrown into the black hole [7]. However, it will satisfy
the second law when gravitational radiation enters the black hole if and only if the
“modified canonical energy flux” of this radiation is positive [7]. This is the case for
general relativity but would not be expected to hold in more general theories of gravity.
The quantum null energy condition (QNEC) [9] asserts that if C is a cross section of
a null hypersurface with vanishing expansion and shear at C, then the expectation value
of the stress-energy flux of a quantum field through C bounds the second time derivative
of the von Neumann entropy outside of C. We have shown [7] that when applied to
a (perturbed) Killing horizon, QNEC is equivalent to a version of the generalized
second law, with the total entropy taken to SC plus a dynamically modified version
of the von Neumann entropy for the matter outside the black hole. On the other
hand, an integrated form of QNEC with suitable future boundary conditions yields the
generalized second law for the Dong-Wall entropy of the black hole plus the ordinary
von Neumann entropy of the matter outside the black hole.
There are a number of issues raised by our new formula, eq.(1), for black hole
entropy in general relativity and its generalization to arbitrary theories of gravity. First,
the dynamical correction term is well defined only at leading order in perturbations
about a stationary black hole. Even in the case of general relativity, the choice of origin
of V in eq.(1) implicitly depends on a choice of background horizon Killing field and
would not be well defined on an arbitrary, non-stationary horizon. This contrasts with
the Bekenstein-Hawking entropy, which is well defined for an arbitrary black hole.
For more general theories of gravity, the are considerable additional ambiguities in the
definition of entropy beyond leading order in deviations from stationarity [7]. It does
not seem unreasonable that the notion of black hole entropy should be well defined
only “near equilibrium,” but it would be worth considering whether our new notion
of black hole entropy can be generalized to the case of black holes that are far from
equilibrium.
Another issue arises from the fact that, at leading order, SC in general relativity is
equal to the area of the apparent horizon. This suggests that, perhaps, the entropy of a
black hole should be associated with a structure that is defined locally in time, such as
a dynamical horizon [10], rather than the globally defined event horizon. In general,
apparent horizons are slicing dependent and dynamical horizons are similarly non-
unique, so it is hard to see how to view them as fundamental structures underlying
the properties of black holes. However, this non-uniqueness is greatly alleviated at
leading order in perturbations of a stationary black hole, which, as discussed above,

123
 87 Page 4 of 5 R. M. Wald

is the only context in which our entropy has been defined. Thus, the possibility of
attributing black hole entropy to a locally defined structure remains open.
A further issue is the appearance of the von Neumann entropy of matter and/or its
dynamical modification in the generalized second law. In the first place, it is quite
surprising that there is any precise statement and argument/proof of the generalized
second law for black holes and matter, since there is no corresponding precise state-
ment or proof of the ordinary second law in the context of non-general-relativistic
physics. It is even more surprising that the von Neumann entropy and/or its dynamical
modification appears in the generalized second law, since the von Neumann entropy
would be constant in time for any complete system in non-general-relativistic physics.
In addition, it would be interesting to see if further justification can be given for the
dynamical modification term to the von Neumann entropy introduced in [7].
Finally, the issue of the fundamental meaning of black hole entropy remains to be
resolved. In ordinary quantum statistical physics, the entropy of a system at a given
value of energy and other state parameters is the logarithm of the number of states
of that system in a small interval around that energy and other state parameters. It is
normally taken for granted that the same interpretation of entropy applies to a black
hole, so that the value of the entropy of a black hole corresponds to counting states of
the black hole in a quantum gravity description. But where do the degrees of freedom
associated with these states of a black hole reside? Are they in the deep interior of
the black hole? Or, are they located on the event horizon? Or, alternatively, do they
reside in the “thermal atmosphere” surrounding the black hole? There are difficulties
with all of these possible answers. Furthermore, in the context of quantum general
relativity, it should be possible for new effective degrees of freedom to be created
and/or destroyed. For example, in an expanding universe, field modes that would be
counted as ordinary degrees of freedom in the present universe would have been sub-
Planckian in the early universe and presumably would not have been included in the
effective degrees of freedom of the field at early times. Similarly, near the singularity
(or whatever replaces it) inside a black hole, it is possible that effective degrees of
freedom may be destroyed. Thus, it does not seem obvious that black hole entropy
corresponds to counting states in any usual sense.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence,
and indicate if changes were made. The images or other third party material in this article are included
in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If
material is not included in the article’s Creative Commons licence and your intended use is not permitted
by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

References
1. Bekenstein, J.D.: Phys. Rev. D 7, 2333 (1973)
2. Bardeen, J.M., Carter, B., Hawking, S.W.: Commun. Math. Phys. 31, 161 (1973)
3. Hawking, S.W.: Commun. Math. Phys. 43, 199 (1975)

123
The Entropy of Black Holes Page 5 of 5 87

4. Iyer, V., Wald, R.M.: Phys. Rev. D 50, 846 (1994)

5. Dong, X.: JHEP 2014, 44 (2014)
6. Wall, A.C.: Int. J. Mod. Phys. D 24, 1544014 (2015)
7. Hollands, S., Wald, R.M., Zhang, V.G.: Phys. Rev. D 110, 024070 (2024)
8. Wald, R.M., Zoupas, A.: Phys. Rev. D 61, 084027 (2000)
9. Bousso, R., Fisher, Z., Leichenauer, S., Wall, A.C.: Phys. Rev. D 93, 064044 (2016)
10. Ashtekar, A., Krishnan, B.: Phys. Rev. D 69, 104030 (2003)

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.

123